1 ELECTRONIC PROCESSES IN POLYMER SOLAR CELLS By SONG CHEN 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 2012
2 2012 Song Chen
3 To my m om
4 ACKNOWLEDGMENTS First, I would appreciate my supervisor Prof. Franky So who has been showing tremendous confidence and support to my research professional working attitude inspired and will always influence my future career. I should say thank you learned and benefited a lot ever student with pure physics background, I appreciate this great experience to collaborate with so many talented chemists. I need to acknowledge my supervisory committee Prof Jiangeng Xue, Prof. Andrew Rinzler Prof David Norton an d Prof. Jacob J. Jones I owe my great PhD experience to many colleagues as well as collaborators. I would like to thank Dr. Jaewon Lee and Dr. Neetu Chopra for their supervisory during my first year of PhD, Dr. Kaushik Roy Choudhury for introducing me to the area of device physics, Dr. Lei Qian and Dr. Chi Hung Cheung for our collaboration and friendship, Dr. Jegadasan Subbiah for your mentoring of thin film processing, Dr. Chad Amb for synthesizing the best photovoltaic polymer in the world, Cephas Small for our shared effort for that record power conversion efficiency, Dr. Sai Wing Tsang for your guidance and our successful collaboration in 2012 That is a period I have my best ever research experience. To all the rest of my solar cell group members Tzung han L ai, Erik Klump, Sujin Beak, Iordania Constantinou, and Michael Hartel, I am happy that you can tolerate leadership in the past few months. Thanks t o the rest members of our huge OEMD lab, it is so nice to communicate with people from different cultural backgrounds. For the research projects I have worked on, I acknowledge the financial support from Office of
5 Naval Research (ONR) Department of Energy (DOE), Air Force Office of Scientific Research, and Fl orida Energy Systems Consortium (FESC). I also thanks to my classmates Weiran Cao Rui Qing, Dr. Yixing Y ang, Dr. Ying Zheng and Renjia Zhou We had a lot of fun during the lunch breaks and weekend basketball time. At last but not the least, I would like to express my deepest gratitude to my this degree in the past four years.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 10 LIST OF ABBREVIATIONS ................................ ................................ ........................... 14 ABSTRACT ................................ ................................ ................................ ................... 17 CHAPTER 1 FUNDAMENTALS OF POLYMER SOLAR CELLS ................................ ................. 19 1.1 Photo Induced Charge Transfer ................................ ................................ ....... 19 1.1.1 Marcus Theory ................................ ................................ ........................ 19 1.1.2 From Excitons to Free Carriers ................................ ............................... 21 1.2 Carrier Transport and Recombination ................................ ............................... 26 1.2.1 Localization ................................ ................................ ............................. 26 1.2.2 Hopping Mechanism ................................ ................................ ................ 27 1.2.3 J V Characteristics Trap Limited and Bulk Limited Transport ............... 28 1.2.4 Bimolecular Recombination ................................ ................................ ..... 30 1.3 Device Architectures ................................ ................................ ......................... 32 1.3.1 Bulk Heterojunction (BHJ) ................................ ................................ ....... 32 1.3.2 Electrode Contact ................................ ................................ .................... 33 2 CHARAC TERIZ ING POLYMER SOLAR CELLS ................................ .................... 34 2.1 Standard Spectra ................................ ................................ .............................. 34 2.2 Transport Measurements ................................ ................................ .................. 35 2.2.1 Space Charge Limited Current ................................ ................................ 35 2.2.2 Carrier Extraction with Linearly Increasing Voltage (CELIV) ................... 37 2.3. Biomolecular Recombination Measurements ................................ ................... 38 2.3.1 Photo Carrier Extraction with Linearly Increasing Voltage (Ph CELIV) ... 38 2.3.2 Transient Photo Voltage (TPV) ................................ ............................... 39 2.4 Probing the Energy Alignment ................................ ................................ .......... 40 2.4.1 Cyclic Voltammetry (CV) ................................ ................................ ......... 40 2.4.2 Electro Absorption (EA) Spectroscopy ................................ .................... 40 3 SPACE CHARGE LIMITED PHOTOCURRENT ................................ ..................... 44 3.1 Abstract ................................ ................................ ................................ ............. 44 3.2 Introduction ................................ ................................ ................................ ....... 44 3.3 Experimental ................................ ................................ ................................ ..... 50
7 3.4 Results and Discussions ................................ ................................ ................... 51 3.4.1 Carrier Mobility Measurements ................................ ................................ 51 3.4.2 Photocurrent J ph (V) Analysis ................................ ................................ ... 55 3.4.3 Photocurrent J ph (G) Analysis ................................ ................................ ... 58 3.4.4 Charge Balanced Systems ................................ ................................ ...... 60 3. 5 Summary ................................ ................................ ................................ .......... 63 4 BIMOLECULAR RECOMBINATION LIMITED PHOTOCURRENT ......................... 65 4.1 Abstract ................................ ................................ ................................ ............. 65 4.2 Introduction ................................ ................................ ................................ ....... 65 4.3 Experimental ................................ ................................ ................................ ..... 68 4.4 Results and Discussions ................................ ................................ ................... 69 4.4.1 The Balance of Charge Extraction ................................ ........................... 69 4.4.2 Recombination at the Maximum Power Condition ................................ ... 72 4.4.3 Recombination at the Open Circuit Condition ................................ .......... 76 4.4.4 Energetic Disorder ................................ ................................ ................... 80 4.5 Summary ................................ ................................ ................................ .......... 84 5 INTERFACIAL ENGINEERING ................................ ................................ .............. 86 5.1 Abstract ................................ ................................ ................................ ............. 86 5.2 Introduction ................................ ................................ ................................ ....... 86 5.2.1 Zinc Oxide (ZnO) ................................ ................................ ..................... 89 5.2.2 Moly bdenum Oxide (MoO 3 ) ................................ ................................ ..... 91 5.2.3 The Inverted Device Geometry ................................ ................................ 93 5.3 Experimental ................................ ................................ ................................ ..... 95 5.4 Results and Discussions ................................ ................................ ................... 96 5.4.1 Inverted Polymer Solar Cells ................................ ................................ ... 96 5.4.2 Inverted Cells with Reduced Interface Recombination .......................... 100 5.4.3 Device Stability ................................ ................................ ...................... 109 5.5 Summary ................................ ................................ ................................ ........ 110 6 OPEN CIRCUIT VOLTAGE LOSS ................................ ................................ ........ 111 6.1 Abstract ................................ ................................ ................................ ........... 111 6.2 Introduction ................................ ................................ ................................ ..... 112 6.2.1 Effective Bandgap ................................ ................................ ................. 112 6.2.2 Recombination and Excitonic Loss ................................ ........................ 116 6.3 Experimental ................................ ................................ ................................ ... 119 6.4 Results and Discussions ................................ ................................ ................. 120 6.4.1 Subgap S ignals in EA Spectra ................................ .............................. 120 6.4.2 Exploring the Excitonic Loss of Open Circuit Voltages .......................... 129 6.5 Summary ................................ ................................ ................................ ........ 136 7 CONCLUSIONS AND FUTURE WORK ................................ ............................... 138 7.1 Conclusions ................................ ................................ ................................ .... 138
8 7.2 Future Work ................................ ................................ ................................ .... 141 7.2.1 The Reduct ion of Energetic Disorder ................................ ..................... 141 7.2.2 Exciton Migration in Donor Acceptor Polymers ................................ ..... 141 APPENDIX A LIST OF MOLECULE STRUCTURES ................................ ................................ .. 145 B LIST OF PUBLICA TI ONS AND CONFERENCE PRESENTATIONS ................... 147 Peer Reviewed Publications ................................ ................................ ................. 147 Oral Presentations in Conferences ................................ ................................ ....... 149 Poster Presentations in Conferences ................................ ................................ .... 151 LIST OF REFERENCES ................................ ................................ ............................. 152 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 165
9 LIST OF TABLES Table page 3 1 Calculated maximum generation rate of electron hole pairs and their separation efficiency under short circuit condition for blends of PG1, PG2, PG3 with PC 61 BM. ................................ ................................ .............................. 57 4 1 Hole and electron mobility in PDTS BTD:PC 71 BM and P3HT:PC 71 BM. .............. 71 5 1 S ummary of device performance ................................ ................................ ...... 105 6 1 Summary of carrier concentrations and recombination induced V OC loss ......... 135
10 LIST OF FIGURES Figure page 1 1 Many dimensional potential profile of reactants (R) and products (P) with respectively with their surrounding miedia.. ................................ ........................ 19 1 2 (color) The illustration of the evolution from an exciton. ................................ ..... 22 1 3 (color) The free energy diagram revealing the process from exciton generation to, charge transfer and charge separation states.. ........................... 23 2 1 Standard solar spectra ................................ ................................ ....................... 35 2 2 The J V curves of two polymer:fullerene systems. ................................ ............ 36 2 3 A sketch showing the physical process of (Ph )CELIV measurement. ............... 37 2 4 The sketch of TPV setup and an example showing the TPV signals of a P3HT:PCBM solar cell. ................................ ................................ ....................... 39 2 5 Energy levels contributing to the Stark effect in a two level dominant system .... 41 3 1 The sketch of space charge accumulation in a system with imbalanced electron hole mobility. ................................ ................................ ......................... 48 3 2 The structures of the polymers to be discussed. ................................ ................ 50 3 3 Current densities as a function of effective electric field for different ho le only BHJ films with optimized donor acceptor compositions for photovoltaic performance. ................................ ................................ ................................ ...... 53 3 4 Current densities as a function of effective electric field for different electron only BHJ films with optimized donor acceptor compositions for photovoltaic performance. ................................ ................................ ................................ ...... 54 3 5 Experimental photocurrent plotted as a function of effective bias. ...................... 56 3 6 Normalized photocurrent plotted as a function of effective bias. ......................... 57 3 7 Photocurrent in PG3: PC 60 BM solar cells as a function of effective applied voltage paramet ric in incident light intensity ................................ ...................... 59 3 8 Photocurrent as a function of incident light intensity at three operating voltages, low, intermediate and saturation.. ................................ ....................... 60 3 9 Saturation voltage as a function of incident light power. ................................ ..... 60
11 3 10 Photocurrent as a function of effective applied voltage for different incident light intensities for PDTS BTD: PC 71 BM. ................................ ............................ 61 3 11 Photocurrent as a function of effective applied voltage for different incident light intensities for PV1000 ink solar cells. ................................ .......................... 62 3 12 Short circuit photocurrent as a function of incident light intensity for the two photovoltaic systems shown in Figure 2 10 and 2 11.. ................................ ....... 63 4 1 (color) J V characteristics of a P3HT:PC 71 BM cell and a PDTS BTD:PC 71 BM cell measured under A.M. 1.5G illumination. ................................ ...................... 66 4 2 (color) J V characteristics of a hole only and an electron only device of PDTS BTD:PC 71 BM.. ................................ ................................ .......................... 70 4 3 (color) J ph plotted as a function of effective voltage. ................................ ........... 72 4 4 (color) Photo CELIV curves of PDTS BTD:PC 71 BM as a function of delay time ................................ ................................ ................................ ..................... 73 4 5 (color) The recomb ination dynamics extracted from the photo CELIV results .... 74 4 6 (color) The recombination dynamics extracted from transient photo voltage measurements with different incident light intensity ................................ ............ 77 4 7 (color) The carrier lifetimes of P3HT:PC 71 BM and PDTS BTD: PC 71 BM systems. ................................ ................................ ................................ ............. 78 4 8 (color) The recombination coefficients of P3HT:PC 71 BM and PDTS BTD:PC 71 BM systems. ................................ ................................ ....................... 79 4 9 (color) CELIV mobility as a function of delay time. ................................ ............. 83 4 10 (color) Zero field mobilities plotted as a function of temperature. ....................... 83 5 1 (color) Energy diagrams of state of art photovoltaic polymers and transition metal oxides that are commonly used for interfacial engineering ....................... 89 5 2 (color) Transmission electron microscopic images of ZnO colloidal nanoparticles with a scale of 5 nm (left) and 50 nm (right). ................................ 90 5 3 (color) X ray diffraction of ZnO nanoparticles. The pattern reveals the wurtzite structure. ................................ ................................ ............................... 91 5 4 (color) The illustration of normal (left) and inverted (right) structure for polymer solar cells. ................................ ................................ ............................. 94 5 5 The monomer structure of PDTG TPD and PDTS TPD. ................................ .... 97
12 5 6 (color) The absorption of PDTG TPD and PDTS TPD measured in solution phase. ................................ ................................ ................................ ................. 98 5 7 (color) The J V characteristics measured under A.M. 1.5G condition. ................ 99 5 8 (color) The external quantum efficiency (EQE) for the same devices in Fi gure 5 7. ................................ ................................ ................................ ................... 100 5 9 (color) The photoluminescence of ZnO nanoparticles films with and without UVO treatment. ................................ ................................ ................................ 102 5 10 (color) The J V (left, A.M. 1.5G) and EQE spectrum of inverted PDTG TPD:PC 71 BM cells. ................................ ................................ ........................... 104 5 11 (color) The J V (left, A.M. 1.5G) and EQE spectrum of inverted PDTS TPD:PC 71 BM cells. ................................ ................................ ........................... 104 5 12 (color) The certification from Newport PV lab. The device was made by Song Chen. ................................ ................................ ................................ ................ 105 5 13 (color) The carrier recombination measured by transient photocurrent (small signal mode). ................................ ................................ ................................ .... 107 5 14 (color) The DC dependent electroabsorption 1 st harmonic signals. .................. 108 5 15 The J V of a PDTG TPD:PC 71 BM cell made in June 6 th 2011 and re measured in August 29 th 2012. ................................ ................................ ........ 110 6 1 (color) The normalized electroabsorption spectra ................................ ............. 122 6 2 (color) Electroabsorption signal. ................................ ................................ ....... 123 6 3 (color) Illustration of CMEAS mechanism ................................ ........................ 124 6 4 (color) CMEAS results of pristine PCDTBT. ................................ .................... 125 6 5 (color) The normalized electroabsorption spectra of pristine PCDTBT under reverse (black) and positive (red) DC voltage biases V DC ............................... 126 6 6 (color) Open circuit voltages plotted as a function of effective bandgap measured by CV, the data is from reference. The das hed line is an indicator of Eq. 6 1. ................................ ................................ ................................ ......... 130 6 7 (color) CMEAS from a pristine PDTG TPD sample, a PDTG TPD:PC 71 BM blend and a linear co mbination of pristine PDTG TPD and PC 71 BM signals. All the spectra are re scaled using the same field condition 10 5 Vcm 1 ............ 131 6 8 (color) Sub bandgap CMEAS signals of a series of photovoltaic polymers. The onset of the sub bandgap feature is defined as the effective bandgap. .... 132
13 6 9 (color) V OC s plotted as a function of measured E eff s measured by CMEAS ...... 133 6 10 (color) plotted as a function of the inverse of dielectric constants. The dashed line is an indicator of eye. ................................ ............ 134 6 11 (color) E CS (the V OC loss during the charge separation process) plotted as a function of the inverse of dielectric constants. ................................ .................. 136 7 1 (color) The absorption spectra of PDTG TPD, PDTG BTD and P(il DTG). The measurements were carried out with solid state polym er films on quartz substrates. ................................ ................................ ................................ ........ 144 7 2 (color) The fitting of absorption spectra and Huang Rhys factor extracted by Gaussian fitting. ................................ ................................ ................................ 144
14 LIST OF ABBREVIATION S AM Air mass BHJ Bulk heterojunction CDM Correlated disorder model CELIV Carrier extracti on with linearly increasing voltag e CMEAS Charger modulated electroabsorption spectroscopy CS Charge separation CT Charger transfer CTC Charge transfer complex CV Cyclic voltammetry D A Donor acceptor DOS Density of states EA Electroabsorption EQE External quantum efficiency ETL Electron transporting layer FCWD Frank C ondon weighted density of states factor FF Fill factor GDM Gaussian disorder model HOMO Highest occupied molecular orbital IC 60 BA I ndene C60 bisadduct IC 70 BA I ndene C 7 0 bisadduct IQE Internal quantum efficiency ITO Indium tin oxide J P H Photo current density J SC Short circuit current density
15 LUMO Lowest unoccupied molecular orbital MDMO PPV Poly(2 methoxy 5 dimethyloctyloxy) 1,4 phenylenevinylene) MoO 2 Molybdenum(IV) oxide MoO 3 Molybdenum(VI) oxide NiO Nickel (II) oxide OPV Organic photovoltaic P3HT P oly (3 hexyl)thiophene PC 61 BM [6,6] phenyl C 6 1 butyric acid methyl ester PC 71 BM [6,6] phenyl C 7 1 butyric acid methyl ester PCDTBT P oly[ [9 (1 octylnonyl) 9H carbazole 2,7 diyl] 2,5 thiophenediyl 2,1,3 benzothiadiazole 4,7 diyl 2,5 thiophenediyl] PCE Power conversion efficiency PDTG TPD Poly[2,6 (4,4' bis(2 ethylhexyl)dithieno[3,2 b:2',3' d]germole) alt 1,3 (5 octyl 4H thieno[3,4 c]pyrrole 4,6 dione)] PDTG BTD Poly[2,6 (4,4' bis(2 ethylhexyl)dithieno[3,2 b:2',3' d] germole ) alt 4,7(2,1,3 benzothiadiazole)] PDTS BTD Poly[2,6 (4,4' bis(2 et hylhexyl)dithieno[3,2 b:2',3' d]silole) alt 4,7(2,1,3 benzothiadiazole)] PDTS TPD Poly[2,6 (4,4' bis(2 ethylhexyl)dithieno[3,2 b:2',3' d]silole) alt 1,3 (5 octyl 4H thieno[3,4 c]pyrrole 4,6 dione)] PEDOT:PSS P oly(3,4 ethylenedioxythiophene):poly(styrenesulfonate) P H CELIV Photo induced carrier extraction with linearly increasing voltage PSC Polymer solar cell PTB7 Poly[[4,8 bis[(2 ethylhexyl)oxy]benzo[1,2 b:4,5 b']dithiophene 2,6 diyl][3 fluoro 2 [(2 ethylhexyl )carbonyl]thieno[3,4 b]thiophenediyl]] P(il DTG) Poly[6,6' ((E) 1,1' bis(2 ethylhexyl) [3,3' biindolinylidene] 2,2' dione) alt 2,6 (4,4' bis(2 ethylhexyl)dithieno[3,2 b:2',3' d] germole )] P (il DTS) Poly[6,6' ((E) 1,1' bis(2 ethylhexyl) [3,3' biindolinylide ne] 2,2' dione) alt 2,6 (4,4' bis(2 ethylhexyl)dithieno[3,2 b:2',3' d]silole)]
16 REDOX Reduction oxidization SCLC Space charge limited current TCLC Trap ped charge limited current TCO Transparent conducting oxide TEM Transmission electron microscopy T I O 2 Titanium ( IV ) oxide TPA Transient photocurrent TPV Transient photovoltage UVO Ultra violet ozone V 2 O 5 Vanadium (V) oxide V OC Open circuit voltage WF Work function WO 3 Tungsten (VI) oxide Z n O Zinc oxide
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 ELECTRONIC PROCESSES IN POLYMER SOLAR CELLS By Song Chen December 2012 Chair: Franky So Major: Materials Science and Engineering Polymer solar cells (PSCs) with power conversion efficiencies (PC E s) exceeding 10 % have been demonstrated in laboratories due to the development of novel materials film proces s ing techniques and device architectures H owever, some critical p hysical properties of the polymer:fullerene bulk heterojunction (BHJ) such as the charge transfer process, energy alignment carrier recombination and transport still remain to be further studied First, we studied the impact of charge transport on the solar cell performance. Space charge limited photocurrent, which is directly caused by imbalanced electron hole mobility, severely reduces the photocurrent extraction efficiency. This model explains the efficiency los s mechanism in certain polymer:fullerene BHJ cells, while the photocurrent loss in the systems with balanced charge transport needs to be studied using detailed recombination measurements. Next, by measuring the carrier lifetime, we concluded that bimolecu lar recombination significantly limits the fill factors and open circuit voltages of polymer solar cells, especially in the systems with large energetic disorder.
18 Following the studies of bulk recombination we also investigated the recombination at the in terface between absorber and electrode. The intrinsic point defects in the transition metal oxide interlayer can reduce the local carrier lifetime and thus the short circuit current of polymer solar cell. Upon the defects reduction using UV ozone treatment we are able to improve the performance of our inverted cell, resulting in a record efficiency of 8.1%. In the final section, we first demonstrate a novel method to probe the energy alignment in the polymer:fullerene BHJ Charge modulated electroabsoprtio n spectroscopy (CMEAS) directly determines the effective bandgap in a solid state BHJ film, which o utplays the previous electro chemical methods that can only be applied to individual pristine materials. Combining the study of carrier recombination with th e result of CMEAS, we successfully demonstrate that the loss of open circuit voltages in excitonic cell s strongly depends on the dielectric constant of the BHJ system.
19 CHAPTER 1 FUNDAMENTALS OF POLY MER SOLAR CELLS 1.1 Photo Induced Charge Transfer As an emerging thin film solar technology, the PCEs of OPV successfully sur passed the 10.0% milestone recently 1 2 Such achievement is realized by the great development of photo ac tive materials, device architecture and the understanding of device physics. To understand the origin of the photovoltaic effect in organic solar cells and then a special case polymer:fullerene BHJ solar cells, we need to first discuss the process of charg e generation photo induced charge transfer. 3 With this discussion, we can further understand the rationale of device architecture and other pieces in the framework of device physics. 1.1.1 Marcus T heory The theory of bimolecular electron transfer reaction in solution was studied since 4 6 A detailed review was presented by Marcus in 1984. 7 Figure 1 1 Many dimensional potential profile of reactants (R) and products (P) with respectively with their surrounding miedia. A and C respectively marks the nuclear coordinates under equilibrium for R and P. B denotes the intersection of the two potential energy surfaces. The dotted line indicates the energy level splitting due to the energy states interaction in R.
20 A c lassic way to desc ribe the electron transfer rate is presented as following. The vibrations within the reactants were treated as harmonic oscillators. The fluctuation of nuclei position affects the potential energy of the outer that the o uter sphere e lectron has the same energy as an outer sphere unoccupied state owned by an adjacent reactant. Such possibility is referred as the probability of reaching the many dimensional intersection region on the coordinate of potential energy. A term d escribing the possibility of crossing the intersection region should be included. Thus, the transition rate is given by: ( 1 1) where or is the transmission coefficient for electron transfer through the intersection region, has the dimension of collision frequency, is the mean separation distance in the t ransition state of the reaction is the free energy of activation that is related to the reorganization parameter With quantum consideration under the approx imation of weak electronic coupling ( ) 8 10 the first order transition rate is given by ( 1 2) where is the matrix elements describing the electronic coupling between a electronic state of reactant and that of a product. is the Franck Condon term which is a sum of products of overlap integrals of the vibrational and solvational wavefunctions of the reactants with those of the products. When all the coordinates (solvent and inner shell) are treated as oscillators in a quantum mechanical way, Eq.1 2 is wr itten as: ( 1 3)
21 where and label the quantum states of reactants and products, is the Franck Cordon term of the given quantum states. is the Boltzman probability to find the system in state Such a treatment is proved to be inappropriate for the reaction experience s a significant change of entropy. H owever, if the quantum mechanics treatment is only applied to the inner shell coordinates, while the solvent is treated classically, Eq. 1 3 is r ewritten as Eq. 1 4 and its high temperature extreme as Eq. 1 5 which is the version widely used in numerical analysis. ( 1 4) ( 1 5) From Eq. 1 1 to Eq.1 5, we can evaluate what parameters can impact the charge transfer in a polymer :fullerene solar cell. First, the electronic coupling is affected by the molecular structure, the interface effective force, the BHJ morphology (the mean spa cing between reactants and products) and the dielectric constant s of the polymer and fullerene derivatives. Second, the energy of initial and final states of the transfer process may be related to the recombination process in the disordered systems. 11 12 Third, the reorganization energy of different polymers is a topic that has not been well studied. 1.1 .2 From E xcitons to F ree C arriers W hat does charge transfer specifically mean in polymer solar cells? Here, as illustrated in Figure 1 2 it is specified as a n effective interface force assisted process in which a photo generated exciton in an electron donating molecule gets separated at the donor acceptor interface resulting in the electron donating mole cule charged with a hole and the adjacent electron accepting molecule charged with an electron. 3 Similarly
22 such a process also appl ies to hole transfer Separated electrons and holes are generated across the donor acceptor interface and transport under the assistance of internal electric field. This is the origin of photovoltaic effect in a polymer:fullerene BHJ excitonic cell. Figu re 1 2 (color) The illustration of the evolution from an exciton A ) photo generated in the donor phase to t he separated electron and hole D ); D denotes the donor phase and A denotes the acceptor phase; B ) stands for the CT stat e. T he solid straight line denotes the donor acceptor (polymer fullerene) interface During charge transfer, t he status wherein an electron hole pair is loosely bonded across the donor acceptor interface is referred as 13 Such a process is ultrafast as transient absorption measurements repeatedly reveal a sub picosecond dynamics 14 15 During this short period of time, excitons experience all the following steps to finish transferring the charge. First, the ph o to generated exciton first relaxes to the one of the many CT manifolds. The role of excess energy during such a process is still under debate. 14 16 18 Many reports claim that, in efficient polymer:fullerene systems, the CT exciton in the higher manifolds (CT n ), which is al are easier to get further
23 separated due to their relatively delocalized wavefunctions. 15 T hose excitons cannot dissociate in such a ultrafast fashion will relax to the ground states CT 0 and then either decay to the ground state (S 0 ) or get enough energy to hop to higher CT manifolds to complete the dissociation. 19 The transition from CT 0 to S 0 belongs to ge minate type of recombination, which is in contrast with the bimolecular recombination between free carrie r s. G e minate recombination is composed of a tiny fraction ( EQE ~10 6 ) of radiative recombination 20 and the rest of non radiative type. Organic p hotovoltaic systems are not good light e mitters; the CT emission is e xtremely inefficient for polymer:fullerene systems, which is in agreement with the discrepancy between typical polymer solar cells performance and limit predicted by Shockley Queisser detailed balance theory 21 Figure 1 3 (color) The free energy diagram revealing the process from exciton generation to, charge transfer and charge separation states. The recombination events are marked as dashed blue arrow; the dashed red arrow denote s the CT relaxation. Figure 1 3 is the so called free energy diagram showing all the above mentioned processed from exciton generation to charge transfer. All the excitation, CT, and recombination pathways are indicated using different arrows. It is worth to notice that,
24 in the energy coordinate due to the small dielectric con stant and an electron hole separation less than 10 nm. Thus free carriers may have lower energ ies than some CT manifolds Charge transfer cannot happen without photo generated excitons. In fact photo excitation is also a charge transfer process that happ ens within a polymer repeating unit. Thus, the exciton binding energy highly depends on the charge distribution in the excited states. Some researchers also use internal dipole moment to describe different exciton binding energies. 22 Usually, a larger dipole moment in the repeating block corresponds to a smaller exciton binding energy and thus a smaller driving force required for charge dissociation. Some earlier thiophene based polyme rs, such as P3HT, 23 have an int ra molecular charge transfer in the scale of a thiophene unit. Therefore, the center of electron and hole wavefunctions are almost overlapped, leaving a very large exciton binding energy of 0.3 eV~0.4 eV. To separat e charges from such a large binding energy, a significant interface electric field (~10 6 V/cm) is required. That leads to a widely accepted but empirical value of 0.3 eV LUMO or HOMO band offset for a minimum requirement to finish the charge dissociation. 24 However, some lately studied donor acceptor polymers show very efficient charge dissociat ion even with a LUMO band offset less than 0.3 eV. To address the difference, we need to first show the basics of donor acceptor polymers. 25 27 Typically, in a r epeating block of such a s bearing quite different electron affinities. This polymer construction strategy allows chemists to create polymers by trying different of donor acceptor combinations, generating pol ymers with low bandgap and deep lying
25 HOMO energies which are required for the harvesting of photocurrent and photovoltage. 28 32 Due to the donor acceptor structure, the exciton excitation is essentially a process of electron transferring from the donor unit to the ac ceptor unit. Since the two unit s are quite separated in the repeating block, the generation of an ex citon caus e s a significant change of int ra molecular dipole moment. 22 The binding energy of the exciton is consequently lower than the traditional single unit polymers. This small binding energy is also consistent with the ultrafast charge transfer that is mentioned in previous text. In sum, the purpose of this section is not only to show the origin of the photovoltaic effect, but more important ly, the difference between organic solar cells and traditional inorganic thin film cells such as amorphous silicon and chalcogenide cells O rganic solar cells genera te charges in two different materials instead of one. As a result, polymer solar cells have to apply the hetrojunction architecture. The driving force of photocurrent is also more than just the equilibrium built in voltage which is the case for inorganic cells. Due to the different direction of carrier concentration gradient, the profile of carrier concentration also contributes to the local fluctuation of quasi Fermi energies, leading to local photocurrent flow. 33 The organic cell s with a two layer structure 24 is very typical in terms of the chemical potential driven photocurrent ; while, in the so called BHJ structure, the carrier concentration is almost uniformly dist ributed in the light absorbing layer, leaving its gradient too small to generate an impact on the quasi Fermi level splitting. But, qualitatively, the photovoltaic effect of a BHJ polymer solar cell is still driven by both the equilibrium built in electri c field and the local fluctuation of chemical potential.
26 1 .2 Carrier T ransport and R ecombination These two electric processes are actually very important topics in organic electronics. They are introduced in the same section because they are quite correlated to each other in the photovoltaic behavior. The carrier transport in the nano scale network is no different from the hopping conduction that people proposed for typical disordered systems; 34 the recombination is much lower than the prediction by Langevin type mechanism, 3 5 whereas, the it scales perfectly with mobility values which can be tailored by measuring the mobility and recombination at different temperatur e s. 36 37 The impact of transport and recombination will be discussed in Chapter 2 and C hapter 3 in details respectively. 1 .2.1 Localization From the classic solid state physics, we know that the defects in crystal structure induce the localization of electron wavefunction and affect the band conduction. The loss of potential periodicity in disordered organic solids determines a different energy level structure to those crystalline solids. Another consequence with the inclusion of disorder is the broadening of density of states (DOS) distribution. 38 In either pro file of DOS tail i.e. exponential or Gaussian, the tail states are more localized than those model, 39 a uniform probability distribution of disorder gives an exponential band tail, 38 which is commonly used to describe the band tail energies in disordered solids. Amorphous silicon and some organic materials are good examples of using an exponential tail distribution. 40 Another commonly used DOS distribution is the Gaussian distribution which is a result of totally randomized energy level occupation. 41 These two DOS profiles are widely used in numerical or analytical models wherein the important factors like tail width usually become fitting
27 parameters. Despite of the difficulties, there are so me very important models, such as Mott transition 42 and Anderson transition 39 derived for the description of metal insulator transit ion. According to the concept of Anderson transition, the weak van der Waals bonding between organic molecules causes the narrowed electronic bandwidth which can be easily overwhelmed by the energy level variation due to energetic disorder. As a result, t he transition from metal to insulator (Anderson transition) happens along with carrier localization. Structural (physical) disorder 43 and polarization 44 are two main factors that contribute to the energetic ( diagonal ) disorder and charge localization 1 .2.2 Hopping M echanism If the energetic disorder and charge localization is mainly determined by the polarization due to the intramolecular structure change, or in other words, if the hopping rate is limited by the molecular reor ganization, polaron models are valid. Marcus theory is usually applied in this scenario since it is similar to the small polaron theory. The hopping activation energy is then in the form of Eq. 1 4 and Eq. 1 5. However, comparing with the hopping mechanism purely based on structural disorder, the polaron model is not very popular in the study of carrier transport. In the model of disorder, it is easy to understand the activation energy only appears in an endothermic process. Such a model was initiated in the topic of impurity conduction at low concentrations. The Miller Abrahams rate 34 is a hopping model built for the impurity states in crystalline materials. Similarly, Ambergaokar r educ ed the hopping model to an equiva lent random resistance network. 45 Following the correlation of localization and mobility gap which was introduced by Anderson and developed by Mott 46 the hopping between two random sites happens if the condition is fulfilled:
28 with as the electronic energy difference between the two sites, U as the binding energy of the localized states (trap) in the mobility gap, as a length describing the special decay of wavefunctions centered at a localized states. F urther assuming makes E q. 1 6 a good approximation o f hopping mechanism : 45 ( 1 6) denotes the transition rate from site i to site j is a constant depending on the electoron phonon coupling strength. is a result of considering the hopping as a tunneling process between two trap states. is introduced due to the requirement to maintain symmetry of net charge flow between i and j Such a model provides insight to explain the temperature dependence in amorphous silicon, and germanium 47 48 1 .2.3 J V C haracteristics Trap Limited and Bulk Limited T ransport The correlation between the hopping mechanism and derivation of J V characteristics is y et clearly demonstrated. Depend ing on the mechanism of disorder and localization, the transport law may vary a lot. Even in different organic materials, contrasting tem perature electric field dependence in J V has been consistently reported. However, there s still a bridge through which the significance of disorder and charge localization is connected by the classical transport parameters such as mobility The effect of hopping can be further reflected by modifying mobility or carrier concentration in J ( n, F ) The trap free space charge limited current (SCLC) is a simple model that makes the connection. Assuming all the carriers contribute to the current flow is provided from
29 injection which means the material has a very low carrier concentration comparing with injection; current is not injection limited, we can combine with ( is the carrier charge, is the carrier concentration, is the local electric field, is the carrier mobility, is the dielectric constant of the sample) and get Mott Gurney law: 49 Due to the presence of traps and disorder, a decent trap free SCLC characteristic can only be found in certain bias condi tions The J V in this range is useful for the estimation carrier mobility. For the J V characteristics other than 50 the in fluence of disorder and localization has to be included. Modifying the SCLC model with a trap distribution, e.g. single level and exponential distribution, gives the so called trap ped charge limited current (TCLC). 49 51 Specifically, an exponential trap distribution helps to explain the slops that are larger than 2 in the plot. 50 The presence of traps change s the carrier concentration n ( E ) since the DOS above trap states obeys a Boltzman distribution of with N T as the trap states density and E T as the width of trap band Some other phenomenological approaches such as including temperature and electric field dependen ce factors are also widely used to describe the J V characteristics Poole Frenkel theory 52 53 is a very useful model to study the electric field dependent mobility. Although the mechanism used to derive the is a special case the carrier dis sociation from a c oulombic potential under the assistance of applied field F, the dependence is very common in reported results. To better describe the mechanism of field and temperature dependence, different disorder model was used. Assuming a Gaussi an distribution of DOS on LUMO or HOMO band with a diagonal
30 disorder and an off diagonal disorder to describe the variation of wavefunction overlap, the field and temperature dependent mobility is described as following: 41 ( 1 7) wherein C is a fitting parameter. W hen is replaced with 1.5 in Eq. 1 7. In contrast to the Gaussian disorder model (GDM) which completely ignore the energetic interaction between different molecule sites, the 3D correlated disorder model (CDM ) 54 assumes a random orientation of molecule dipoles and further calculated the energetic fluctuation due to the dipole dipole interaction with the nearest neighb oring dipoles. The following empirical expression was achieved: ( 1 8) wherein is a fitting parameter to describe the geometry disorder similar in GDM, C 0 is a constant equals 0.78, a is the lattice spacing. Eq. 1 8 is used to extract th e energetic disorder values in C hapter 4. 1 .2.4 Bimolecular R ecombination Bimolecular recombination is a very common topic in all semiconductor related d evices. Since organic materials are intrinsic under equilibrium low in carrier concentration with one sun illumination, there s no need to distinguish majority and minority carrier s Some researchers pointed out OPV (under illumination) is majority carrier cell as they counted both hole and electron are majority carriers. 33 According to different excitation condition, we can distinguish the dynamics based on high or low level excitation. High level excitation means the number photo generated carriers is much larger than that in the background ( ). Using l ight source to excite a cell originally under equilibrium condition is a typical exampl e of high level
31 excitation. In C hapter 4, we will discuss a related experiment. In contrast, low level excitation is fulfilled when Usually, a low level excitat ion can be realized by using an attenuated source to generate a little fraction of carriers in the sample which is simultaneously illuminated by an intensive source The related recombination measurements using low level excita tion will also be discussed i n C hapter 4. The recombination dynamics due to high level excitation is described using exponential decay. The decay rate is usually in the form of: 55 ( 1.9) with is the initial carrier concentration due to the high level excitation, is the time dependent carrier lifetime, is a fitting param eter characterizing the dispersive behavior of the disordered material sample. In contrast, the dynamics of the low level excitation is in the form of single exponential decay: ( 1.10) Since the background carrier concentration is a very large constant, the only parameter that reflects the influence of background carrier is the lifetime constant. Reviewing the origin of photovoltaic effect in OPV, or polymer solar cells specifically we can understand how significantly the bimolecular recombination process impacts the PCE. 56 58 When a BHJ cell is f r ee of any bimolecular recombination, the device performs as following: F irst, the quasi Fermi level of electrons and holes will be infinitely close to the band edge of LUMO and HOMO, resulting in negligible loss of V OC This is why the V OC measured under low temperature is very close to the effective bandgap of the BHJ. Suc h a topic will be discussed in C hapter 6. Second, the increase of carrier concentration at low internal field condition does not change the charge generation and extracti on, which means a FF value close to 100% if there s also no loss
32 d ue to space charge effect (see C hapter 3) and the charge collection at electrodes (see C hapter 5) Third, the short circuit current will be increased slightly because bimolecular recombinati on is limited by the very low concentration ( is the concentration of polymer/fullerene domain in a typical BHJ structure) at the short circuit condition. 1 .3 Device A rchitectures 1 .3.1 Bulk H eterojunction (BHJ) For the light absorbing layer, the idea of BHJ was created to provide more donor acceptor interface to result in more charge dissociation than the traditional two layer structure. 59 By blending the donor and acceptor materials in the solution phase, the resulting solid film is composed of two interpenetration network s of donor and acceptor phases whose dom ains have a dim ension around 1 0 nm. The actual size of polymer and fullerene domain is very critical to the solar cell performance. An exciton generated in a polymer domain need to migrate to the heterojunction interface before decaying to S 0 The charge generation will be greatly decreased when the domain size is too large. Also, the carrier transport in such a network significantly relies on the connection between each domain. Using the terminology of percolation theory, an over mixed solid solution may have a domain si ze that is too small to provide the critical number of off between charge Many processing technique s have been re ported to approach the optimum morphology. Distinguished by the evaporation speed of solvent, there are thermal annealing 60 fast / slow dr ied solvent processes 61 62 Using solvent addit ives is a novel approach to optimum morphology in most polymer:fullerene systems. The merit of solvent additives may rely in its high boil ing point. However, the common additives such
33 as 1,8 diiodooctane along with the chlorinated solvents that are used for dissolving the polymer:fullerene blend are not compatible with environmental benignity This will be a concern for the future development of solar cell printing technology. 1 .3.2 Electrode C ontact As another part of the sandwich structure in polymer solar cells, the electrodes need to for m Ohmic contacts with the active layer for the following reasons. First, it helps to reduce the series resistance as well as the solar cell FF; second, it increases the equilibrium built in potential which is critical for V OC and J SC The traditional approach to form Ohmic contacts is to use conducting materials (such as metals and transparent conducting oxides) with very large or small work function to provide a good Fermi level alignment with the active layer. However, due to the instability of these conducting materials as well as merits of solution processed transition metal oxides, the later has become an effective way to make efficient polymer solar cells. The topic of interface en gineering will be discussed in C hapter 5.
34 CHAPTER 2 CHARACT ERIZ ING POLYMER SOLAR CE LLS 2.1 Standard Spectr a The PCE of a solar cell is defined by the ratio of output electrical power and incident radiative power. While the electrical powe r can be easily calculated through J V curves, the incident power from the so lar simulator needs to be standardized. When the sun and earth is separated by a mean distance ~149,597,890 km, the irradiance of the sun that hit s the outer atmosphere is about 136 mW/cm 2 Such a constant is called the solar constant which is also the ori Its value can be varied due to the change of earth sun distance and activity of the sun. From the outer atmosphere surface to the ground, the irradiation is further re duced depending on the air mass, zenith angle and path length Air mass (AM) 1, 1.5 and 2 are used to characterize the different conditions of the atmosphere effect. The conditions for the AM 1.5 spectra were widely chosen because they are representative of average conditions in region of United States T spectrum. The consideration here is to include the scattering and diffusion effect of atmosphere and their contribution to the irradiation. That is why a direct spectr um has a smaller radiative po wer than the global one since an additional portion is included. AM 1.5G is one of the most commonly used standard spectra for the J V characterization of solar cells. It has the spectra showing in Figure 2 1 and an integrated radiative power of 100mW/cm 2 Depending on the spectrum content, spatial uniformity and temporal stability, solar simulators are graded into different levels. Solar is the flux of electrons that are extracted with a monochrom atic light at the wavelength of is the flux of
35 incident p hoto ns Integrating the measured EQE spectrum with a standard solar spectrum generates a current density that equals to the J SC measured under the same standard spectrum. Figure 2 1 Standard solar spectra 2.2 Transport Measurements 2.2.1 Space C harge L imited C urrent Using the SCLC model to extract the zero field mobility is still the mostly used approach t o determine the electron and hole transport in the BHJ film Trap free SCLC is usually fulfilled when the carrier injection increase to certain level so the quasi Fermi level can surpass the energies of most deep traps. To correctly extract the mobility value, some details need to be considered. First, the injection contact needs to be Ohmic to avoid injection limited current. For hole injection, a top contact of MoO 3 /Ag has the highest injection efficiency comparing with other high work function electrodes that
36 are currently used. LiF/Al is a good choice f or electron injection. Second, the injection of the other type of carrier should be avoided. For example in the hole mobility measurement, the electron injection needs to be blocked at the counter electrode, MoO 3 is again a good candidate for the purpose. The structure of ITO/MoO 3 /BHJ/MoO 3 /Ag is typical for measur ing hole SCLC mobility. Third, the built in potential needs to be subtracted from the applied voltage before the data fitting using the SCLC model. Fourth, the voltage drop due to the series resistance at the contacts needs to be subtracted. A TCO with a sheet resistance of 20 at least consumes 0.1V of voltage when the current is 5mA. Finally, a Poole Fre nckel factor needs to be inclu ded to address the field dependence issue. A typical SCLC current is shown in Figure 2 2. Figure 2 2 The J V curves of two polymer:fullerene systems. The black line indicates the SCLC region. Both devices have the structure of ITO/MoO 3 /BHJ/MoO 3 /Ag. The holes are injected from the silver side.
37 2.2.2 Carrier E xtraction with L inearly I ncreasing V oltage (CELIV) As observed in many polymer:fuller ene systems, the J V characteristics cannot be fitted using trap free SCLC model even with the fitting parameters such as the Poole Frenkel factor. Sometimes the injection limited current makes the mobility underestimated. Figure 2 3 A sketch showing the physical process of (Ph )CELIV measurement. Alternatively, a carrier transient measurement was invented with its advantage of using a linearly increasing extraction field instead of a rectangular wave which is applied in the traditional dark injection transient measurement 63 The on edge of a rectangular wave V (t) has a first derivative closes to infinity, resulting in a huge RC respons e from the sample since the magnitude of a RC oscillat ion is proportional to If carriers are extracted using a linearly increasing voltage which has a constant value, the RC response does not superimposed with the carrier transient s ignal, making it possible
38 to extract mobility values higher than 10 3 cm 2 V 1 s 1 This technique is called carrier extraction using linearly increasing voltage (CELIV) 64 A typical carrier transient signal is shown in Figure 2 3. Assuming a sample is under flat band cond ition, a transient peak due to carrier extraction can be seen on top of a rectangular signal which is a displacement current proportional to the increasing rate ( A ) of the applied reverse triangle wave : The triangle wave is applied starting fro m a flat band condition. Thus, if the carriers to be extracted are photo generated e.g. by a laser pulse the triangle wave should start from a positive value which can compensate the photo bias. The shape of the transient signal is highly related to the t ransport balance between hole and electron. A detailed numerical solution can be found in other references. The t max in Figure 2 3 is called transient maximum time, it characterizes the average mobility of the extracted holes and electrons. Using t max the mobility can be calculated with either of the following: or 65 66 Furthermo re, by changing the delay time t del between the laser pulse and the onset of triangle wave, CELIV is able to provide a relaxation time dependent CELIV mobility, which is a tool to probe the energetic disorder and dispersive transport 55 in different polymer:fullerene systems. The experiment will be shown in C hapter 4. 2.3. Biomolecular Recombination Measurements 2.3.1 Photo Carrier Extraction with Linearly Increasing Voltage (Ph CELIV) As seen in the last section, the extracted carrier concentration can be calculated by integrating the transient signal. If the carriers are photo generated using a short laser pulse prior to application the linearly increasing voltage (by a delay time t del ), photo
39 carrier recombination c an be monitored. By changing t del the plot of carrier concentrations vs. time is generated Also, since Ph CELIV will finally deplete the photo generated carriers, the recombination dynamics is the result of high level excitation. The related experiments and discussions will be presented in C hapter 4. 2.3.2 Transient Photo Voltage (TPV) Different from the recombination measurements based on a high level excitation, the so called TPV technique measures the carrier decay due to a low level excitation i.e. 58 67 is background carrier concentration resulted from a solar simulator. The photo bias under one sun condition is close to the V OC of the sample. Figure 2 4 The sketch of TPV setup and an example showing the TPV signals of a P3HT:PCBM solar cell Figure 2 4 shows a setup of TPV measurement, u nder the dynamic equilibrium condition, e.g. at V OC the bimolecular recombination rate is equal to the charge generatio n rate as a result of continuity equation Simultaneously, an attenuated pulse laser is used to generate The photo voltage signal and the decay of the perturbation are shown on an oscilloscope As long as the excitation is kept at low level ( ) the and share the same lifetime. To de rive this conclusion, we need to get the
40 differential s on both sides the photo voltage expression : T he decay rate of photo voltage can written as Thus the lifetime of photo voltage perturbation is the same as the carrier lifetime which is dominated by the bimolecular process at open circuit condition The results of TPV measures are included in C hapter 4 and C hapter 6. 2.4 Probing the Ene rgy Alignment As will be discussed in C hapter 6, the topic of energy alignment is of unique importance for a better understanding of polymer:full erene BHJs. 2.4.1 Cyclic V oltammetry (CV) The purpose of this section is to highlight the drawback of this traditional electro chemical method. T he cyclic r edox proc ess is realized by scanning cyclic electric potential of the working electrode. The actual potential value is further calibrated by a reference ele ctrode. The HOMO and LUMO levels of pristine materials can be determined by the onset of charging and discharging current s The measured results do not have good repeatability Due to the potential variation of different reference electrode s the error can be as large as 0.3 eV which is not acceptable for the study of energy alignment of BHJ 2.4.2 Electro A bsorption (EA) S pectroscopy As a traditional differential spectroscopy, EA is used to study the fine structure of energy bands in inorganic materials b ecause those transitions cannot be detected by a linear absorption measurement. Due to t he external electric field, certain transitions are altered due to the mixing of eigen states. It was shown that the low field induced linear dielectric function change in germanium is in good agreement with the third derivative of
41 the linear dielectric function. 68 Such a phenomenon is explained as a result of d iscontinuous change of a second order term, i.e. the acceleration of electron s leads to a third order change in the third derivative spectrum. The lineshape of EA spectra is believed to be unchanged and scaled with The analytical derivation of this process is not included here. When the external field is larger than a perturbation, the band structure of the sample will be changed, leading to the modification of energy levels such as Stark shift. Figure 2 5 Energy levels contributing to the Stark effect in a two level dominant system It has been shown that the model of Stark shift is in agreement with the EA response of many conjugated polymers. 69 70 This is probably due to their simpler energy level structure than inorganic semiconducting materials. As shown in Figure 2 5, the optical transition of many conjugated polymers can be illustrated. A dominant excitonic transition from 1A g to 1B u and their vibronic features compose the linear absorption spectra. The transition from 1A g to nA g is forbidden due to the selection rule. With electric fields, d ue to the mixing effect between 1B u and higher forbidden states,
42 the transition from 1A g to 1B u is less allowed and the transition from 1A g to nA g becomes less forbidden. Again, using the time independent perturbation theory here, the mixing effect is shown in Eq. 2 1 is the result o f non vanishing matrix elements. ( 2 1) In the situation shown in Figure 2 5, Eq.2 1 can be reduced to the form of Eq. 2 2. Using the classical description, the oscillating strength is transferred from the allowed transition to the forbidden transition by a tiny fraction. Such a transfer of oscillating strength leads to the subtle change in the linear absorption spectrum. By applying an additional modulation field F ac the s ubtle change in absorption spectra can be detected by locking in the first or second harmonic portion of the EA signal. As shown in Eq.2 4, the DC field only exists in the term of first harmonic response while the second harmonic term is related to the amp litude of the modulating field. Thus, apart from detecting the position excitonic level using Eq. 2 4, EA can also probe the equilibrium built in potential of an organic diode. 71 73 ( 2 2) ( 2 3) ( 2 4) The above conclusions are based on a simplified two level model and the approximation of perturbation however, the experimental results of many conjugated polymers do show the same field dependence. When the transitions involve the higher
43 continuous levels, Franz Keldysh effect needs to be considered, 74 the discussion will not be included here since they are not related to the experiments in C hapter 6. If the sample is composed of more than one m aterial, the detection of charge transfer related energy levels becomes impossible for the traditional electro chemical methods. However, EA is still able to provide clear signals due to its sensitivity to the charge transfer process. As will be seen in C h apter 6, EA is used to study the effective bandgap of polymer:fullerene BHJs. The origin of those subgap feature is totally different from the above mentioned quadratic Stark effect. The subgap pumping induced charge transfer from the HOMO of donor to the LUMO of acceptor gives rise to the population of loosely bounded charge pairs. These dipoles across the polymer fullerene interface can respond the external modulating field. With a higher energy than the CT 0 energy and relaxed polaron, these carriers gene rated by sub bandgap pumping induce additional features in the subgap region. The experimental instruction is detailed in Chapter 6.
44 CHAPTER 3 SPACE CHARGE LIMITED PHOTOCURRENT 3 .1 Abstract The development of photovoltaic polymers has been pushed by the demand of high mobility, wide absorption band, low lying highest occupied molecular orbital (HOMO), better film morphology when blended with an electron accepting fullerene derivative. In this chapter, in order to specify the impact of transport on the solar cell performance, we carried out detailed transport experiments on some typical systems to gain an insight into transport balance and its impact on the parameters such as fill factors and s hort circuit currents. The results show that, with severely unbalanced electron hole mobilities, the solar cells made from a series of dioxythiophene benzothiadiazole (DOT BTD) copolymers (PG1 PG3) and fullerene derivatives show a clear feature space char ge limited photocurrent. In contrast, the systems based on dithienosilole benzothiadiazole (DTS BTD), with fairly balanced transport and an improved solar cell fill factor, exhibits no fingerprint of space charge effect, leaving the loss mechanism to be di scussed in C hapter 4 3 .2 Introduction As t he follow up topic of C hapter 2 it is necessary to reemphasize the well known operating mechanism of a BHJ solar cell: light absorption in polymer:fullerene absorber creates excited excitons that subsequently mi grate to the donor acceptor interface and dissociate via ultrafast charge transfer from the exited donor to the acceptor. 3 Following the photoinduced charge transfer, geminate pairs are formed across the interface, which require further separation into free carriers (or polarons) by the internal electric field. Subsequently, the free holes and electrons need to transport through the
45 respect ive donor and acceptor phases, resulting in an external photocurrent. The process of carrier transport occurs in the interpenetrating network defined by the morphology of polymer and fullerene derivatives. Due to the nature of bulk heterojunection, electro ns and holes still have great chance to meet each other at the polymer fullerene interface to facilitate bimolecular recombination which is a topic will be covered in C hapter 4 The competition between the intensive recombination rate and carrier transport finally determines the quantity of collectable carriers. As a result of the weak van der waals force between the organic molecules, the carrier transport in typical polymer solar cells is based on hopping mechanism, resulting in the a carrier mobility of 10 5 ~10 3 cm 2 V 1 s 1 which is significantly lower than that of inorganic thin film absorbers such as multicrystalline/ amorphous silicon, chalcogenide with an electron and hole mobility >10cm 2 V 1 s 1 It is important to notice that the mobility values of or ganic materials highly depend on the characterization method. A lot of organic molecules are reported showing hole mobilities over 1cm 2 V 1 s 1 when being operated as a thin film transistor (TFT) 75 wherein the transport is greatly facilitated by inte rface states intermolecular hopping With large number of injected carrier s filling the traps in mobility tails and lift ing the Fermi energy over the mobility edge 42 the mobility is significantl y increased and sensitive to carrier concentration 76 77 In the contrasting scenario of a thin film solar cell, the driven field is applied in the vertical direction, allowing the limited amount of carriers to hop in a 3 dimension fashion or to be tr apped As a result, it is usu ally observe d that hole mobility values measured by vertical direction transport can be much lower than those displayed in a lateral transporting thin film transistor.
46 The issue of low carrier mo bility, especially hole mobilities in polymeric materials wa s once considered to be the overwhelming factor limiting the PCE of PSCs. Some level as expected 78 79 Including the consideration of Langevin relat ed recombination some later computational results revealed that very high mobility values do not necessarily benefit the solar cell performance. 80 81 Also, w ith very a high hole mobility in the BHJ device, electrons become the slow carriers that cause imbalanced transport which is to be dis cussed in the following Consequentially, the question comes to how is the J sc, FF affect ed within the hole mobility of 10 5 ~10 3 cm 2 V 1 s 1 ? To answer this, the role of electron transporter, i.e. fullerene derivatives, has to be considered. First, we consider the situation wherein the mean velocity of carriers is too small to let carriers escape the recombination or trapping before reaching the electrode. This is common in low mobility systems or s amples with very large thicknesses. Such phenomenon is called drift length limited photocurrent that the mean electron and hole drift lengths w h(e) = h(e) h(e) E is smaller than the active layer thickness L : with h(e) the hole (electron) mobility, h(e) the carrier lifetime and E the electric field, respectively. If the lifetime is mainly determined by the bimolecular recombination, the extraction of photocurrent is bimoledular recombination limited. Clearly, by tuning the thickness of solar cells, the negat ive effect of drift length photocurrent can be decreased. This partially explains why most of the laboratory made BHJ solar cells have an optimum device thickness ~ 100nm. Second, the mobility of one type of carrier is significantly larger (>10 0 times) than the other. This is quite typical in the earlier stage of the PSC development because the fullerene phase keeps relatively constant electron mobility 10 4 ~10 3 cm 2 V 1 s 1 while
47 the hole mobility in the polymer phase stay at 10 6 ~10 5 cm 2 V 1 s 1 A direct impact of such unbalanced transport is that holes tend to accumulate near the anode and change the distribution of internal electrical field. To meet the requirement of charge neutrality, most of the electric field will eventually drop on the sp ace charge region to increase the extraction of the slow carriers holes and, on the other side, decelerate the extraction of fast carriers electrons by distributing less electrical field out of the space charge region T he band diagram in Figure 3 1 helps to clarify the physical picture: space charge accumulation induces the non uniform field distribution which is manifested by the band bending at the anode site. Since the internal field is redistributed to the level that drift length reaches the length of the space charge region, we have L 1 = 1 1 E 1 = 1 1 V 1 / L 1 wherein L 1 is the thick ness of the space charge region Usi ng the general expression of ge minate pair extr action established by Goodman 82 we have: ( 3 1) q is the carrier charge This equation describes the hole transport in the space charge (hole) region wherein the population of electron is very limited. The amount of accumulated holes will reach a limit that the hole generation in this region is just able to provide the same carrier flux as the space charge li mited current Thus we have: ( 3 2) where J SCL is the space charge limited conduction expression derived by the from the external source such as carrier injection or photo generation. o r is the dielectric permittivit y. By reducing L 1 from the expression, we get the expression of space charge limited (SCL) ph o to current ( ) : 83
48 ( 3 3) Figure 3 1 The sketch of space charge accumulation in a system with imbalanced electron hole mobility. We see that, SCL photocurrent extraction not only shows a dependence which is the same as the drift length limited condition, but displays a behavior which is different from the J ph ~ G dependence shown in Eq. 3 1. Thus, by measuring bias photocurrent ( J ph =J illum J dark ) under different incident power density, one can experimentally determine if the photocurrent space charge limited Experimentally, the photocurrent dependence also relies on the device thickness, because the charge accumulation needs to be supported by t he significant amount of charge generation in the sample. 84 The first typical SCL photocurrent phenomenon was observed in a MDMO PPV:PCBM sample with a thickness close to 300 nm. 83 The model of SCL photocurrent is of significance for profiling the charge accumulation in BHJ layers A direct measurement of internal field distribution and charge accumulation is challenging whi le the measurement of bias/intensity dependent photocurrent enables a very effective tool for the purpose. When the BHJ is composed of a fullerene derivative and a
49 polymer with very low hole mobility, SCL photocurrent should be the dominant loss mechanism of J sc and FF of the solar cell. For the system with balanced electron hole mobility, the J sc and FF loss should be related to either exciton dissociation, bimolecular recombination or poor interlayer engineering which will be mentioned in the next few chapters. In our experiments, some donor acceptor polymers with different hole mobility values will be discussed from the viewpoints of unbalanced charge transport to address the loss mechanis ms of FFs and J sc s in the resultant solar cells. To experimentally probe the impact of SCL photocurrent we compare the performance of our model donor acceptor copolymer (PG1 PG3): PCBM system with that of two other model polymer systems optimized for sola r cells: a low bandgap donor bis(2 ethylhexyl)dithieno[3,2 d]silole) 2,6 diyl alt (2,1,3 benzothiadiazole)4,7 diyl], PDTS BTD 85 86 blended with PC 7 1 BM, with a broad band PV response over a range of 300 to 800 nm leading to PCEs exceeding 5%, 85 and a commercially available poly (3 hexyl)thiophene: PCBM (P3HT: PC 71 BM) blend PV ink system, Plexcore PV1000, with a reported PCE higher than 4%. Among these polymers:fullerene systems, the solar cells made by PG1 PG3 and PDTS BTD all show low FFs below 56% despite their discrepancy in J sc s. Also, we didn t prepare very thick samples because very low FFs are observed even when we controlled the thickness down to 150 nm. Figure 3 2 shows the molecular structure of copolymers PG1, PG2, PG3 and the low bandgap polymer PDTS BTD We examine the photocarrier generation processes and the carrier transport properties in these model systems, and our analysis indicates that charge balance play s an important role in determining J sc FF and thus the solar cell PCE
50 Figure 3 2 The structures of the polymers to be discussed 3 .3 Experimental The series of low bandgap donor acceptor copolymers named PG1, PG2 and PG3 based on dioxythienylenes, unsubstituted thiophene and benzothiodiazole building blocks were synthesized as described elsewhere 87 PDTS BTD was synthesized and purified via a method involving significant modifications to a protocol reported in the literature. 85 The P3HT: PC 71 BM ink was provided by Plextronics Inc. Sputtered I TO on soda lime glass substrates were used as transparent electrode. All the organic layers were spin coated from solutions. Metallic electrodes were deposited in vacuum using thermal evaporation. Hole only devices had the structure of ITO/ poly(3,4 ethyle nedioxythiophene):poly(styren esulfonate) PEDOT:PSS/polymer or polymer:fullerene/Pd, while electron only devices had the structure of ITO/active layer/LiF/Al. As a conducting polymer with a work function ~ 5.2 eV, PEDOT:PSS forms an Ohmic contact for hole i njection, Pd suppresses electron injection into the organic layer due to a large electron injection barrier built up by its work function (~5.2 eV ) 88 and the lowest unoccupied molecular orbital (LUMO) energy of PCBM (~4.3 eV) To make
51 electron only devices, the ITO substrates were cleaned in acetone and isopropanol without subsequent UV ozone treatment to prevent hole injection. Solar cells had a stacking structure similar to the hole only devices with LiF/Al replacing Pd as the low work function cathode Current density versus voltage measurements were carried out in ambient using a Keithley 4 200 semiconductor characterization system. For photocurr ent measurements, AM1.5G 100 mW cm 2 white light from a 150 W ozone free xenon arc lamp (Newport) was used. A set of neutral density filters were used to control the incident power density from 35 mW cm 2 to 130 mW cm 2 for the study of charge generation dependence A Newport 70260 power meter in conjunction with a Newport 70268 detector was used to measure the power densities of the incident white light For photocurrent measurements under reverse bia s, the applied voltage is scanned from +1V to 10V. 3 .4 Results and Discussions 3 .4.1 Carrier Mobility M easurements According to the physical model presented in the introduction section, the polymer:fullerene system with unbalanced hole electron transport show SCL photocurrent extraction which jeopardizes the FF s and J SC s in the resultant solar cells. The first part of the experiment is to measure in the hole mobility of the actual BHJ layer with the optimum solar cell performance. T o achieve the best PCEs in devices made by the three green DA polymers (PG1, PG2, and PG3), a very large proportion of the electron accepting PC 60 BM is required in the BHJ layer Mixing a fullerene derivative into a photovoltaic polymer can significantly alter the hole mobility from that in the pristine materials. 89 90 Here, we use single carrier devices to measure t he charge carrier mobilities, which is done by modeling the current density voltage ( J V ) in the dark with
52 the space charge limited current (SCLC) formalism assuming a field dependent mobility. 89 90 For unipolar transport in a trap free semiconductor with an Ohmic injecting contact, the SCLC is given by the Mott Gurney equation which is the first equation of the Eq. 3 2, where is the charge carrier mobility, V 1 should be switched to V as the applied voltage across the whole active layer, and L 1 is replaced by L as the thickness of the active layer. To determine the ho le mobility, either in pristine polymers or in the polymer phase of a blend with PCBM, we fabricated and characterized hole only devices (details of device fabrication and structures are provided in the Experimental section). From the J E measurements we d etermine a moderate hole mobility in PG1 of 5.410 8 cm 2 V 1 s 1 Similar analyses from the hole only devices yield increasing mobility values of 6.910 7 cm 2 V 1 s 1 and 310 6 cm 2 V 1 s 1 for PG2 and PG3 respectively. These results agree well with the expected increase in backbone planarity of the copolymers (going from P1 to P3) leading to improved hole transport. 91 Next, charge transport in t he hole only devices was characterized for blends of the three copolymers with PC 6 1 BM, with compositions that yielded the best PV performance for each copolymer. Figure 3 3 displays the measured J E characteristics in the hole only devices of the different copolymer: PCBM blends. Interestingly, the hole mobilities in all of the blends of the copolymers PG1 PG3 with the fullerenes of optimum composition are enhanced significantly (to 2 10 6 cm 2 V 1 s 1 5 10 6 cm 2 V 1 s 1 and 2 10 5 cm 2 V 1 s 1 respectively) by one to two orders of magnitude, compared to their corresponding values in the pristine copolymers. These results are in agreement with those observed for the MDMO PPV 89 90 and poly[2 methoxy 5 (30,70 dimethyloctyloxy) p phenylenevinylene] (OC 1 C 10 PPV) 88 blends
53 where the hole mobilities in the polymer blends with PC 6 1 BM were significantly higher than the values in the corresponding pristine polymer. Such compositional dependent hole mobility is an interesting topic related with the formation of percolation netw ork of the polymer phase. The detailed discussion can be found in our ori ginal publication. 92 Figure 3 3 also displays J E characteristics of hole only devices from PDTS BTD: PC 71 BM and from the P3HT: PCBM PV ink system. It is interesting to note that the hole current density in both these devices is significantly larger than in our PG: PC 6 1 BM devices, indicating higher values of hole mobility. Figure 3 3 Current densities as a function of effective electric field for different hole only BHJ films with optimized donor acceptor compositions for photovoltaic performance. Here, we also address the electron transport and the charge balance in the different polymer: PCBM blend systems. In order t o determine the electron mobility in the PCBM phase in a blend with a polymer, we fabricated electron only devices from
54 blends of PG3 and PDTS BTD with PCBM of optimized composition (1:8 and 1:1.5, respectively), and from the P3HT: PCBM blend ink system. Figure 3 4 Current densities as a function of effective electric field for different electron only BHJ films with optimized donor acceptor compositions for photovoltaic performance. The electron mobility in PG3: PC 6 1 BM (1:8) blend is determined to be 1.3 10 3 cm 2 V 1 s 1 The results show that even at its optimum composition, the electron mobility of the PG3: PC 6 1 BM blend is about 2 orders of magnitude higher than the hole mobility (210 5 cm 2 V 1 s 1 ), indicating an imbalance in charge transport. This imbalance in carrier transport in the BHJ PV cells from PG3 is expected to lead to build up of charges and increased recombination events in the blend, resulting in lower efficiencies. The electron mobilities of the PDTS BTD: PC 71 BM (1:1.5) blend and the P 3HT: PCBM ink system are 1.210 3 cm 2 V 1 s 1 and 410 4 cm 2 V 1 s 1 respectively. When compared to the corresponding hole mobilities of 6.7510 4 cm 2 V 1 s 1 and 510 4 cm 2 V 1 s 1 in the PDTS BTD: PC 71 BM blend and the P3HT: PCBM ink system, respectively, it can be
55 concluded that carrier transport is significantly more balanced in these systems, preventing charge build up and reducing recombination. 3 .4.2 Photocurrent J ph (V) A nalysis As mentioned, BHJ solar cells fabricated from the PG: PC 6 1 BM blends exhibit poor performances. In particular, PG3: PC 61 BM shows a FF<45% and peak external quantum efficiencies<50%. 87 In this section, we investigate the origin of the low FF and possible efficiency loss mechanisms in the BHJ solar cells from PG3. Figure 3 5 shows the photocurrent density, J ph under simulated solar white light intensity of 100 mW cm 2 as a function of the effective applied voltage V eff = V 0 V app across the sample for different polymer:PCBM blends used in this study. The photocurrent density J ph is defined as J ph = J illumination J dark where J illumination and J dark are the current den sities under illumination and in the dark, respectively. 82 83 90 93 94 V 0 is the value of applied bias V app where the ph otocurrent in the device is minimum ( J illumination = J dark ), corresponding to a zero driving force condition The J ph V eff plots from the solar cells of all three green polymers (PG1, PG2, PG3) blended with PC 61 BM show three regimes. In the region where t he effective field is low ( V eff <0.1 V), the photocurrent increases linearly with voltage. This is followed by a region where a square root dependence of the photocurrent (as shown by the solid line) as a function of effective voltage is observed. Regardin g the unbalanced transport, such intermediate region indicates that the photocurrent probably be limited by the build up of holes close to the anode of BHJ solar cells, where Eq. 3 3 holds. At last there is a saturation region of J ph at large V eff corresponding to high reverse biases, where all the dissociated electron hole pairs are extracted and collected at the electrodes, without observable bimolecular recombination processes In the saturation region the photocurrent becomes a constant that i s
56 proportional to G max which represents the maximum generation rate of free carriers at a given intensity of illumination. Figure 3 5 Experimental photocurrent plotted as a function of effective bias. It must be noted that at any given electric field a nd temperature, due to the stron g excitonic characteristics, ge minate loss of excitons, unbalanced transport and bimolecular recombination process only a fraction of the photogenerated photo generated excitons will split into free carriers with dissociation efficiency ( E T ). This gives the generation rate at any given electric field and temperature as G ( E T ) = G max ( E T) while J ph = qG max ( E T ) L To compare the efficiency of charge dissociation at the donor acceptor in terface for the different copolymer: PC 6 1 BM solar cells, we normalize the photocurrent by the experimentally observed saturation current density ( J sat ) in Figure 3 6 The short circuit operation points are marked by the dashed lines in the figure. While allows the determination of the maximum rate of photogeneration of electron hole pairs, Figure 3 6 gives a direct measure of the charge dissociation efficiency at different applied biases. These parameters for the different
57 copolymer: PC 6 1 BM combinations are presented in Table 3 1 The highest value of G max obtained for the solar cell fabricated from the copolymer PG3, is 4.710 27 m 3 s 1 This electron hole pair generation rate is 17% and 34% higher than the values in the solar cells f rom PG1 and PG2, respectively. Under short circuit conditions, i.e. at V eff = V 0 Figure 3 6 shows that only 39% and 43% of the total photogenerated electron hole pairs in the solar cells from PG1 and PG2 are successfully extracted and collected at electro des In the devices fabricated from PG3, the short circuit charge extraction efficiency, sc increases to 53%. However, at the maximum power point where the PCE is defined, this value drops to ~41%. Clearly, the FF s and J SC s in these solar cells are limited due to the unbalanced transport and resultant J ph ~ V 1/2 behavior Table 3 1 Calculated maximum generation rate of electron hole pairs and their separation efficiency under short circuit condition for blends of PG1, PG2, PG3 with PC 61 BM. Composition PG1:PC 61BM (1:4) PG2:PC 61 BM (1:5) PG3:PC 61 BM (1:8) G max (max generation rate; m 3 s 1 ) 4. 0 10 27 3.5 10 27 4.710 27 sc (at short circuit condition; %) 39 43 53 Figure 3 6 Normalized photocurrent plotted as a function of effective bias.
58 3 .4.3 Photocurrent J ph (G) Analysis As mentioned in the previous section, the imbalanced carrier mobilities and the J ph ~ V 1/2 dependence is not sufficient to attribute low FF s and J SC s in the PG:PCBM systems to SCL photocurrent In particular i f the mean electron or hole drift length becomes smaller than the active layer thickness, the main photocurrent loss is mainly due to bimolecular recombination. Such recombination limited photocurrent is given by Eq. 3 1 Experimentally, the two cases can be distinguished by performing light intensity dependent measurements. While the SCL photocurrent scales with a 3/4 power dependence on light intensity (Eq 3 3 ), in the scenario of recombination limited photocurrent, J ph in the V 1/2 dependence region is linear with light intensity. Additionally, the transition bias between the J ph ~ V 1/2 dependence and the saturation regime, denoted as V sat should scale with the square root of light intensity in the former case 82 83 while stay constant with different light intensity in the latter. 57 Figure 3 7 shows the results of the J ph ( V eff ) as a function of illumination intensity. Firstly, the photocurrent shows a regime with square root dependence on the applied bias for all light intensities, this regime shifts to higher effective voltages with increasing light intensity, and secondly, V sat shifts with increasing light intensity. In Figure 3 8 we show the intensity dependence of J ph ( V eff ) at three different effective biases, corresponding to small, intermediate, and large internal field regimes. The slope S determined from a linear fit to the measured J ph yields values of 0.75, 0.78 and 0.99 at 0.1V, 1.5V and 10V effective voltages, respectively. The J ph ~ G 3/4 dependence confirms the SCL photocurrent at small and intermediate internal fields.
59 Figure 3 7. Photocurrent in PG3: PC 60 BM solar cells as a function of effective applied voltage parametric in incident light intensity, measured at room temperature. Legends indicate different light intensities used. The arrow is a guide to the eye indicating the variation of the voltage at which the photocurrent enters the saturation regime. The short circuit operating point is marked. Moreover, the fact that J ph exhibits a linear dependence ( S =0.99) on light intensity in the saturation region confirms that the photocurrent extraction neither limited by SCL or bimolecular recombinatio n. In Figure 3 9 V sat is plotted as a function of incident light intensity. A slope of 0.51 extracted from this plot confirms a V sat ~ G 1/2 dependence which is consistent with the prediction. 83 In sum, the light intensity dependent results confirm that SCL photocurrent determines the significant FF and J SC loss in the PG3: PC 61 BM BHJ system. It is interesting to notice that these samples reach the SCL photo current limit at a thickness of ~150 nm (Figure 3 5) which is much smaller than the previous findings in MDMO PPV:PCBM systems. We postulate the different might stem from the higher charge generation rate in a D A polymer The intensive dispersed mobility due to large energetic disorder may also play a role.
60 Figure 3 8 Photocurrent as a function of incident light intensity at three operating voltages, low, intermediate and saturation. Slopes determined from linear fits are shown and indicate space charge limited behavior in the low and intermediate regime. Figure 3 9 Saturation voltage as a function of incident light power. 3 .4.4 Charge Balanced S ystems In the preceding section, we demonstrated that, due to severely unbalanced transport of electrons and holes, the BHJ of PG3:PC 61 BM leads to SCL photocurrents
61 even at the illumin ation intensity as low as 35 mW cm 2 For a charge balanced system where the photocurrent is not space charge limited, it can be expected to vary linearly with intensity. The loss of photocurrent at low extraction field is purely due to the recombination resulted from drifting length limitat ion. In section 3 .3.1, we showed the another low bandgap donor acceptor polymer, PDTS BTD, exhibits an electron mobility larger than the hole mobility by a factor of 2 in the polymer: PC 71 BM (1:1.5) system. In the P3HT:PCBM (PV ink) system, the electron a nd hole transport are even more balanced. These two systems, along with the PG3: PC 6 1 BM blend that was just presented, are surely typical to demonstrate the impact of charge balance on the performance of polymer solar cells. Figure 3 10 Photocurrent a s a function of effective applied voltage for different i ncident light intensities for PDTS BTD : PC 7 1 BM Figures 3 10 and 11 show the J ph ( V eff ) plotted as a function of illumination intensity in the PDTS BTD:PC 71 BM and P3HT:PC 61 BM PV ink systems. In contrast to the PG3: PC 61 BM system, the J ph ( V eff ) curves in both PDTS BTD and P3HT: PCBM ink
62 devices are much steeper than those in the PG3: PC 61 BM blends before they reach the saturation region. Figure 3 11 Photocurrent as a function of effective applied voltage for different incident light intensities for PV1000 ink solar cells. Moreover, the photocurrent for all incident light intensities become saturated at a effective bias as low as 0.4 V and stay almost constant at higher fields, indicating hi gher charge extraction efficiency at low biases. The trend of J ph ~ V 1/2 was not observed in any region of the curves, indicating no SCL effect. The impact of bimolecular recombination, a topic in Chapter 4 is hard to distinguish here because even no J ph ~ V 1/2 dependence was observed, the severe recombination may happen at certain bias condition which directly determines the carrier concentration in the sample. Additionally, a V sat that does not vary with incident light intensity also substantiates this. Th e data also points out another outperforming factor in the PDTS BTD: PC 71 BM and P3HT: PCBM PV ink systems, compared to PG3: PC 61 BM, that the short circuit point locates in the regime where J ph is already saturated. This saturation of J ph results in much hi gher
63 short circuit charge dissociation efficiencies, 70% for the PDTS BTD:PC 71 BM and 83% for the P3HT: PC 61 BM. A combination of the absence of SCL photocurrent and the saturation of J ph at relatively low fields results in higher FF s in these two PV systems Especially for P3HT, FF can reach >65%. Figure 3 12 Short circuit photocurrent as a function of incident light intensity for the two photovoltaic systems shown in Figure 2 10 and 2 11 Slopes determined from linear fits are shown. 3 .5 Summary We have studied charge transport, photocurrent extraction in BHJ solar cells fabricated from three different polymer: fullerene model systems to understand the dependence of PV performance on transport balance. In the series of dioxythienylene benzothiodia zole donor acceptor co polymers blended with PC 61 BM that are targeted for color tunable photovoltaics, hole transport in the polymer phase of the blend is found to be enhanced by one to two orders by loading PC 61 BM. However, even in the optimized compositi on, due to unbalanced mobilities between electrons and holes, space charges (holes) induce redistribution of internal electrical field and eventually
64 reach a steady state wherein the photocurrent output is limited by the slow carrier, leading to a reductio n of FF s in these devices. In contrast, carrier transport in the PDTS BTD: PC 71 BM and the P3HT: PCBM PV ink systems is more balanced, resulting in significantly improved charge separation efficiency and fill factors. These findings reinforce the importance of attaining charge balance in these BHJ devices to avoid space charge effects in donor acceptor blends for PV applications. The difference of recombination loss between PDTS BTD:PCBM and P3HT:PCBM will be discussed in C hapter 4
65 CHAPTER 4 BIMOLECULAR RECOMBIN ATION LIMITED PHOTOC URRENT 4.1 Abstract The model of transport balance is not sufficient to explain the loss of photocurrent in some bulk heterojunction systems. With the development of advanced donor acceptor polymer, high hole mobilit y and low bandgap can be realized. However, many of these solar cells still suffer from low fill factors poor photocurrent extraction at the low extraction field. We detailed the investigation by comparing the recombination dynamics in two model systems wi th fairly balanced transport. The results show bimolecular recombination can be the limiting factor of photocurrent extraction even without the existence of space charge effect. Further experiments also show that larger rate of bimolecular recombination is associated with the large energetic disorder in the blend system. 4.2 Introduction As a bench mark BHJ system, the blend of poly(3 hexylthiophene 2,5 diyl) (P3HT) and [6,6] phenyl C 6 1 (or C 71 ) butyric acid methyl ester (PC 61 BM or PC 71 BM) has been extensively studied to reach a PCE approaching 5 % and FF close to 70%. 95 Optimized P3HT: fullerene solar cells demonstrate almost saturated photocurrent output even at low extraction fields, 92 96 However, the loss of low energy photons in the solar spectrum limits the further enhancement of P3HT system. To incre ase photon harvesting, tremendous effort has been taken to develop low bandgap donor acceptor copolymers with absorption edges extended to the near IR range. 85 97 99 High short circuit currents and high open circuit voltages have been achieved in some low bandgap de ep HOMO polymer:PCBM systems resulting in PCE exceeding 7%. 29 At the same time, many
66 potential polymer:PCBM systems show both balanced transport and low FFs, we therefore think it is necessary to figure out the physical pr ocesses that limit the performance in these low bandgap polymer:fullerene systems. In this chapter we examine the low bandgap donor acceptor copolymer poly ((4, 4 dioctyldithieno (3,2 b:2',3' d) silole) 2,6 diyl alt (2,1,3 benzothiadiazole) 4,7 diyl ) (PDTS BTD which has yielded a PCE over ~6% 100 when blended with [6,6] phenyl C71 butyric acid methyl ester, PC 7 1 BM. However, the maximum FF attained in these devices is only 56%. In contrast, P3HT:PC 7 1 BM cells fabricated in many laboratories including ours regularly attain FFs of 65 70% (Figure 4 1) Higher FF characterizes more efficient photocurrent collection, especially in the low field condition. Therefore, understanding the loss mechanism may lead to the methods for better solar cell performance. Figure 4 1 (color) J V characteristics of a P3HT:PC 71 BM cell and a PDTS BTD:PC 71 BM cell measured under A.M. 1.5G illumination
67 Cons idering the PDTS BTD:PC 71 BM system we find that the mobilites of electron and hole are well balanced. Similar to what has been observed in P3HT:PC 71 BM system, the space charge effect cannot be the limiting factor of solar cell FFs. As mentioned in C hapter 3 both systems don t show a J ph ~ V 1/2 dependence, suggesting no significant recombination limited photocurrent under the medium or large bias condition. However, the carrier dynamics remain unclear at the zero internal field situation which corresponds to a very narrow window in the V eff axis. At such low extraction field, the device has a very high concentration of photo generated carriers and therefore the trend of intensive recomb ination. Here, we examined the carrier recombination rate in both P3HT:PCBM and PDTS BTD:PCBM systems through photo induced carrier extraction with linearly increasing voltage (photo CELIV) 65 and transient photo voltage (TPV) 67 101 102 experiments. Photo CELIV allows one to adjust the internal field such as the field corresponding to the maximum power point, of devices during the recombination measurement In contrast, TPV is carried out by shinning continuous white light on the sample and generatin g high concentration of carriers. Since the sample is measured under the open circuit condition, TPV provides carrier recombination results that are complementary to the photo CELIV measurements. Our photo CELIV data show that, the two polymer:fullerene sy stems both show sufficiently long carrier lifetime to avoid recombination limited photocurrent at the power maximum point. However, with the increase of carrier concentration at lower extraction field, PDTS BTD:PC 71 BM exhibit much shorter carrier lifetime than P3HT:PC 71 BM, leading to significant photocurrent loss at the open circuit condition. Such intensive recombination can also lower the open circuit voltage in PDTS BTD:PC 71 BM.
68 Further, we correlate the recombination rates of the two blend systems with t he intrinsic property of the polymer donor. By considering carrier recombination as a process of charge transfer described by the small polaron model, 4 103 the short photo carrier lifetime in the PDTS BTD:PC 71 BM system can be attributed to the higher degree of diagonal disorder in this polymer. 4.3 Experimental PDTS BTD was synthesized and purified via a method involving significant modifications to a protocol reported in the literature 85 Details of the fabrication of single carrier devices and PV devices are also provided elsewhere 87 100 104 Hole only devices had the structure of ITO/ MoO 3 / active layer/ Au, electron only devices had the structure of ITO/ active layer/ LiF/Al. MoO 3 was used as hole injection layer because it provides better hole injection than poly(3,4 ethylenedioxythiophe ne): poly(styrenesulfonate) PEDOT:PSS. ITO without UV ozone treatment matches the LUMO level of PCBM. Current density versus voltage measurements were carried out in ambient using a Keithley 4200 semiconductor characterization system. For temperature depen dent measurements, a LakeShore 321 temperature controller and a Janis VPF 100 cryostat were employed. The photo CELIV setup consists of a LASER EXPORT DTL 319QT (527 nm, 8 ns time width 60 function/arbitrary waveform generator, a Tekxtronix TPS2024 oscilloscope and a SRS DG535 delay/pulse generator. The laser was used to generate photo carriers for extraction. The delay generator was used as a triggering source as well as delay time controller of the function generator and the laser pulse. The photo generated carriers are extracted by the triangle wave applied by the function generator. Detailed description of the experimental setup can be found in previous publications. 64 For
69 transient photo voltage experiment, the white li ght for photo biasing the sample was obtained from a 150 W ozone free xenon arc lamp (Newport). Different intensities were achieved by using a set of neutral density filters, and were calibrated by a Newport 70260 radiant power meter in conjunction with a Newport 70268 photodiode. Photo voltage decay was obtained using a Tekxtronix TPS2024 oscilloscope. The fundamentals of the optical perturbation as well as the calculation of charge density were reported elsewhere. 105 106 4.4 Results and Discussions 4.4.1 The Balance of Charge E xtraction In the BHJ polymer: fullerene solar cells, the transport of holes and electrons happens through different percolation pathways. The nature of hole transport of the blend is thus quite different from that of the neat polymer. The hole mobility in this situa tion is not only determined by the donor polymer, but is also affected by the bicontinuous nanostructure formed together with PCBM. As a result, the transport of both electrons and holes is influenced by the composition and morphology of the polymer: fulle rene system. 89 90 107 Since mobility in these composite systems is highly sensitive to the material compositions and processing conditions, electrons and holes might have different transport properties. A str ong imbalance in the mobility of the two types of carriers results in inefficient extraction of the faster carriers to maintain charge neutrality in the device. This leads to accumulation of space charges in the device and reduction of photocurrent extraction Moreover, the accumulated space charges promote local increase of recombination which is an additional source of loss mechanism caused by unbalanced transport. 90
70 Figure 4 2 (color) J V characteristics of a hole only and an electron only device of PDTS BTD:PC 71 BM. The SCLC fitting is shown. The balance of carrier transport in the PDTS BTD:PC 71 BM blend was determined by measuring electron and hole mobility from single carrier devices. Figure 4 2 shows the electron and hole only current of PDTS BTD:PC 71 BM blends. The current density voltage (J V) data was fitted by a ssuming the SCLC formalism and a field dependent mobility. For unipolar transport in a trap free semiconductor with an Ohmic injecting contact, the SCLC is given by the Mott Gurney equation, ( 4 1) where is the charge carrier mobility, V is the applied voltage, and L is the thickness of the active layer. As summarized in Table 1, zero field mobilities of 9.610 4 cm 2 Vs 1 and 5.210 4 cm 2 Vs 1 are calculated by taking into the account of the device series r esistance, respectively for electrons and holes. As revealed by the mobility values, the transport of electrons and holes in PDTS BTD:PC 71 BM is well balanced. Compared to
71 the mobilities in P3HT: PC 71 BM, which are also included in Table 1, PDTS BTD:PCBM has even higher mobility for both carriers. Since the balanced transport in P3HT:PCBM has been reported 92 as the key reason for its high FF, this leads us to believe that a similar situation, if not a be tter one, prevails in the PDTS BTD:PCBM system as well. Table 4 1 Hole and electron mobility in PDTS BTD:PC 71 BM and P3HT :PC 71 BM. h e PDTS BTD:PC 71 BM 5.2 10 4 cm 2 V 1 s 1 9.6 10 4 cm 2 V 1 s 1 P3HT BTD:PC 71 BM 2.9 10 4 cm 2 V 1 s 1 4.0 10 4 cm 2 V 1 s 1 However, from Figure 4 3 showing the photocurrent as a function of extraction field, it is clear that despite the balanced transport properties of the two material systems, the loss of photocurrent behaves quite differently. For P3HT, the photocurrent gets nearly saturated even at low extraction field and reaches 82% of its maximum under short circuit condition, while the photo current extraction in PDTS BTD:PC 71 BM is far from saturation and only reaches 68% of it s maximum under short circuit condition. The external quantum efficiency reported for the two systems 85 95 supports this difference in extraction efficiency. At lower extraction field, P3HT:PC 71 BM maintains its photo current before the sharp drop around 0.4 V. In contrast, PDTS BTD:PC 71 BM has a much stronger field dependent photo current over the entire range of cell operation which leads to lower FF in the solar cells. As shown by the carrier mo bility values, the imbalance of carrier transport leading to build up of space charges is not the factor limiting photocurrent extraction. The difference in the extraction of photocurrents, therefore, must have its origin in a different limiting mechanism. The following sections will be deal ing with another possible limiting factor bimolecular recombination.
72 Figure 4 3 (color) J ph plotted as a function of effective voltage. 4.4.2 Recombination at the Maximum Power C ondition According to the J V characteristics of the devices under illumination, the loss of photocurrent happens both at low extraction fields and under open circuit condition. In this study, we employ the technique of photo CELIV 65 108 to study bimolecular recombination at power maximum condition, which starts with a low extraction field across the cell. The low extraction field is simulated by biasing the solar ce ll devices to voltages approaching the open circuit voltage (V oc ) which is slightly lower than the photo bias provided by a 527nm pulse laser with a maximum power of 60 offset voltage is maintained just below the dark current turn on voltage (power maximum point) to ensure that the number of injected carriers from the contacts is negligible compared to the population of photo carriers generated by the laser pulse. Figure 3.3(a) shows the photo CELIV transients from the PDTS BTD samples. By in troducing a variable delay between the exciting laser pulse and the onset of carrier
73 extraction, we can introduce different recombination times in the system under investigation. Figure 4 4 (color) Photo CELIV curves of PDTS BTD:PC 71 BM as a function of delay time Figure 4 5 plots the decay of photocarrier concentration as a function of time in the two systems, which is obtained by integrating the photo CELIV transients. Here the time scale is composed of the delay time and transient maximum time. The delay time, as mentioned earlier, is the time between the application of the laser pulse and the extraction voltage ramp, while the transient maximum is the time when the transient current reaches its peak value. In the pho to CELIV experiment, the samples are initially in the dark before photocarriers are generated by the laser. In this case the excit ation cannot be considered as a happens after a strong excitation. Ac cording to Figure 4 5 the concentration of photo 55 ( 4 2) ( 4 3)
74 ( 4 4) Figure 4 5 (color) The recombination dynamics extracted from the photo CELIV results where n and p stands for electron and hole concentrations respectively, is temperature dependent dispersive parameter, k 0 the recombination coefficient at time zero. To fit the data, was measured to be 0.98 and 0.90 respectively for P3HT:PC 71 BM and PDTS BTD:PC 71 BM P3HT:PC 71 BTD:PC 71 BM. Thus, according to the lifetime results, P3HT:PC 71 BM and PDTS BTD:PC 71 BM show similar carriers decay rates at low extraction fields. The carrier lifetime data can furt her be used to calculate the bimolecular recombination rate by assuming a Langevin type recombination for low mobility materials where the recombination rate is ( k L Langevin recombination coefficient) ( 4 5) Calculated re sults show that the ratio of the recombination rates k/k L is about 10 3 The resultant recombination induced photo current loss ~10 2 mA/cm 2 which is similar to
75 other results found under short circuit condition 35 This negligible bimolecular recombination photocurrent at low extraction fields can be rationalized as follows. Before current injection turns on in the device, the drift current determines the Due to the extraction field, the photo generated carriers are depleted, resulting in lower carrier concentration comparing with the open circuit condition. Since the recombination coefficient and the recombination rate are both sensitive to carrier concentration, the photo current at or c lose to short circuit condition is not affected significantly by recombination. Even at a bias just below the dark current turn on voltage, recombination is still limited as shown by the photo CELIV results. The nano morphology of the bicontinuous interpen etrating network in both systems 95 100 also explains why the recombination is reduced at extraction field. The bulk heterojunction enable s the formation of electron and hole percolation pathways and improve the transpo rt in both phases. At a carrier concentration (~10 16 cm 3 ) lower than the domain concentration (~10 1 7 cm 3 ), the probability of holes and electrons to meet and recombine is li mited. 35 In such a situation, mono molecular 109 110 (ge minate recombination) may overwhelm the bimolecular process and become the dominant recombination mechanism. A few points regarding the technique of photo CELIV and its relevance and shortcomings in probing recombination in these BHJ solar cells need to be pointed out. First, photo CELIV does not provide true operating conditions necessary for inducing strong carrier recombination since it does not continuously flood the device with high carrier concentration. Second, while both the polymer:fuller ene systems in our study show similar lifetimes, their results are extracted from different carrier concentration
76 levels. The carrier concentration is also relatively low compared to that under open circuit condition. The lower carrier concentration can be partially explained in Figure 4 4 Because solar cells do not have blocking layer in terms of extraction, the application of laser pulse can collect the carriers at contact region under low extraction field thus generates some instantaneous current signal s which correspond to carrier concentration ~ 10 16 cm 3 Finally, photo CELIV does provide low extraction field but not true open circuit condition. Even through a bias can be tuned for the purpose, the photo bias is very sensitive to laser pulse and makes it hard to fulfill the true open circuit condition Due to all these it is necessary to resort to a different technique to probe further the recombination properties in these material systems. 4.4.3 Recombination at the Open Circuit C ondition The recombination in BHJ solar cells at open circuit condition is a topic that has has been a topic of discussion over recent years. 101 111 112 Despite the existence of a diffusion current, due to the ab sence of internal electric field. Since carrier recombination is a function of background charge density, the magnitude of extraction field is expected to play a more significant role in this case, and the results are anticipated to be different from those acquired through the photo CELIV experiment which only measures the recombinatio n under power maximum condition. Also, as recombination directly affects the carrier concentration in the sample, such a value can also be easily extracted using a lifetime me asurement. The obtained charge carrier population can not only be used to analyze the photocurrent loss and FFs, but also the splitting of quasi Fermi levels and V OC s of polymer:fullerene systems.
77 Figure 4 6 (color) The recombination dynamics extracted from transient photo voltage measurements with different incident light intensity Transient photo voltage (TPV) is a technique that, in contrast to photo CELIV, provides continuous solar illumination and true open circuit c ondition. Under illumination with a solar simulator the device is under a light bias, and subsequently, a weak laser pulse with a typical energy ~1 nJ/pulse impinges on the photo biased device, creating a weak perturbation (Figure 2 4) Following this, the temporal decay of photo carriers can be monitored by probing the decay of perturbation (<5mV) created by the laser pulse 67 102 In this study, we applied TPV to study the recombination at open circuit condition. The P3HT:PC 71 BM and PDTS BTD:PC 71 BM solar cell devices used for these measurements have PCEs of 4% and 5.5% respectively. The intensity of the solar simulator is varied to change the bac kground charge density in the devices. As is evident from the photo voltage transients in Figure 4 6 the photo carrier lifetime varies inversely with the charge density in both systems. However, at the maximum charge
78 density (100 mW/cm 2 in this experiment), photocarriers in the P3HT:PC 71 BM cell have a BTD:PC 71 BM ~ 400 ns (Figure 4 7 ). Additionally, P3HT:PC 71 BM cells have a larger charge density than tha t of the PDTS BTD:PC 71 BM cells when the dynamic equilibrium between the photo carrier generation and recombination, given by the equation, ( 4 6) is reached. Here, G max is the internal carrier pairs generation rate and P is the proportion of exciton dissociation. Since the effective generation rate of photo carriers ( PG max ) is roughly of the same order in the two polymer: fullerene systems, the longer carrier lifetime in P3HT:PC 71 BM results in a larger carrier density. Figure 4 7 (color) The carrier lifetimes of P3HT:PC 71 BM and PDTS BTD:PC 71 BM systems. Additionally, PDTS BTD:PC 71 BM cells show a stronger dependence of carrier lifetime on charge density (in the form of photo bias in Figure 4 7 ). Hence, th e difference of lifetime between the two systems becomes smaller when carrier concentration
79 decreases, allowing for both PDTS BTD:PCBM ( n ~10 15 cm 3 ) and P3HT:PCBM ( n ~10 16 cm 3 ) to show very similar lifetimes ~20 from photo CELIV experiments, as highlighted in Figure 4 7 In the two measurements, although recombination happens under different extraction fields, the lifetime is mainly determined by carrier concentrations Figure 4 8 (c olor) The recombination coefficients of P3HT:PC 71 BM and PDTS BTD:PC 71 BM systems. Another important parameter determining photocurrent loss is the strength of the recombination rate compared to the Langevin mode rate which is shown in Figure 4 8 In PDTS BTD cells ratio of k/k L ~10 1 is significantly high, with a stronger dependence on charge carrier density. In contrast, in P3HT:PC 71 BM, this ratio is ~10 3 has and shows a very weak dependence on charge density. As a consequence, at the extractio n field lower than the maximum power point, the photo current loss due to bimolecular recombination is as high as ~1mA/cm 2 in PDTS BTD:PC 71 BM devices while it is ~0.1
80 mA/cm 2 in P3HT:PC 71 BM cells. Although the loss of photo current reduces when the extracti on field grows, it still explains why PDTS BTD:PC 71 BM cells show low FF. At zero extraction fields, the dynamic equilibrium is estabilished. Thus recombination loss equals the charge generation rate. When the internal field emerges, the short (400ns) carri er lifetime leads to a photo current loss > 1 mA/cm 2 thus lowering the extraction and pushing the maximum output power to a lower bias. At the power maximum point, recombination rate decreases with decreased carrier concentration. Finally, a t even high er e xtraction fields (at and beyond short circuit condition), the dynamic equilibrium between generation and recombination is completely broken. The photo generated charges almost contribute to the output current, while recombination loss is negligible. Moreov er, since most of the generated carriers are pulled out of the device, the carrier lifetime in different material systems should be similar. Even at extraction fields lower than the short circuit condition, this similarity of lifetimes prevails until the c harge injection turns on. This fact further stresses the difference between photo CELIV and TPV techniques in probing carrier recombination. The experiments also show that although recombination is less likely to be a universal loss mechanism over the enti re bias range, it can rationalize the FF difference between two efficient polymer: fullerene systems. 4.4.4 Energet ic D isorder Once we have established that carrier recombination plays a critical role in limiting the photo current in the PDTS BTD based cel ls, it is important to f igure out the physical reason for the stronger recombination compared with the P3HT based cells. In disordered material systems, the electric and optical properties are often correlated with the degree of disorder.
81 In amorphous silicon solar cells with a strong energetic disorder, the main source of photo current loss was identified as the recombination of photocarriers in band tail states. In disordered materials, tail states exist in the bandgap. Photogeneration of carrier pair s is followed by energetic relaxation to attain dynamic equilibrium state. The photogenerated carriers are likely to execute a random walk, basically spatial tunneling, between neighboring energy states before relaxing into tail states. The overall shift o f mean energy caused by relaxation is determined by the energetic disorder and temperature. The expression for this energy shift is given as 41 113 ( 4 7) where is the energetic disorder and, k is Boltzman constant, T is temperature in Kelvin. As a result, the photo carriers finally get distributed over these large number of tail states. Carrier transport is significantly slower through these tail states than the transport through mean states. Energy relaxation is further enhanced when the material is cooled to low temperatures, because spatial tunneling dominates over thermally activated transport. This is alternately described as the occurrence of a larger effec tive energetic disorder at lower temperature. Photo CELIV can provide information about this energetic disorder. The delay time dependent measurement using photo CELIV allows us to probe the energy relaxation in different polymer:fullerene systems. 55 108 In Figure 4 4 we calculate the CELIV mobilities from the expression ( 4 8) where d is the device thickness, A is the voltage ramping rate in Vs 1 t max is the j and j (0) are respectively the magnitude of pure transient current and displacement current. 114 The transport of PDTS BTD:PCBM shows a strong
82 dependence on the variable delay time by demonstrating a longer transient time when delay time increases. In addition to this, PDTS BTD:PCBM shows a clear reduction of mobility with the growth of delay time in Figure 4 9 As explained above, with longer delay times, more photo carriers get a chance to relax into tail states with the transport getting slower when the electric field is applied. Thus, this type of mobility degradation is a proof of energetic disor der at a certain temperature. In contrast, the mobility of carriers in the P3HT:PC 71 BM system does not show any signs of reduction (Figure 4 9 ) at the same temperature (RT in this case). However, by taking advantage of the above mentioned relationship betw een temperature and effective energetic disorder, this decay of CELIV mobility should be detectable at low temperatures. This data depicting the decay of mobility at 100K in the P3HT:PC 71 BM system is also shown in Figure 4 9. From these results it can be c oncluded that the P3HT:PC 71 BM system has lower energetic disorder than the PDTS BTD:PC 71 BM system at room temperature(RT). This disorder cannot be detected using photo CELIV experiment at RT. But, by going to temperatures as low as 100K, thermally activate d transport is greatly suppressed, and the large population of carriers in tail states then allows photo CELIV to detect the disorder in the P3HT:PCBM system. Energetic disorder, which is the origin of delay time dependent mobilities, can be quantitatively estimated with the assumption of a specific DOS model Upon photo excitations, the DOS of the polymer relax on the coordinate of energy. In other words, the carriers relax into tail states between which the hopping process is quite limited comparing with the states above the mobility edge. Several models have been formulated to describe the o rigin of disorder in materials.
83 Figure 4 9 (color) CELIV mobility as a function of delay time. Figure 4 10 (color) Zero field mobilities plotted as a function of temperature. In this study, we employ the 3D correlated disorder model 54 to model the energetic disorder and its nature in these composite systems. This model assumes a random 3 dimentional distribution of dipoles in the material. The disorder in molecular site energy
84 stems from the interaction between these dipoles. The carrier mobility is calculated and exp erimentally proven to have the following field and temperature dependence in the form of Eq. 1 8 Using th is expression, the effective energetic disorder in the system can be extracted by plotting the zero field mobility as a function of temperature. In Fi gure 4 10 energetic disorder is extracted to be 84meV for the PDTS BTD:PC 71 BM system compared to 61meV for the P3HT:PC 71 BM system. The increased disorder in the low bandgap polymer:fullerene system is consistent with the earlier findings from photo CELIV. PDTS BTD:PCBM indeed has a broader tail band distribution of molecular energies than P3HT:PCBM, leading to the higher deg ree of disorder in the material 4.5 Summary In this study we investigated the physical processes limiting the FF in a low bandgap pol ymer:fullerene system. We studied the transport and recombination of photo carriers in the PDTS BTD:PC 71 BM BHJ system and compare the results with the highly optimized blend of P3HT:PC 71 BM. Mobility measurements shows well balanced transport of electrons and holes in PDTS BTD:PC 71 BM cells; this rules out the possibility that photo current loss is due to unbalanced transport. Two different techniques are used to study photo carrier recombination under different extraction fields. Results from the photo CELI V measurements show that the recombination of the two systems at P max condition does not limit the photo current. In contrast, carrier pairs experience strong recombination when the PDTS BTD solar cells are operated at open circuit condition, leading to a significant loss of photo current. The differences in lifetime are found to be determined by the carrier concentration at corresponding extraction fields. Finally, we show a correlation between the strong recombination of photo carriers in PDTS BTD:PC 71 BM and the considerable energetic disorder present in the material.
85 This is established by the delay time dependent photo CELIV mobility in DTS BTD:PC 71 BM and the larger value of the disorder parameter compared with P3HT:PC 71 BM.
86 CHAPTER 5 INTERFACIAL ENGINEERING 5.1 Abstract As a typical excitonic cell, bulk heterojunction polymer solar cells are able display photovoltaic effect even without built in potential. The ultra fast charge generation across the donor acceptor interface enables the splitting of quasi fermi level of electron and holes entirely upon chemical potential gradient. This unique photovoltaic mechanism is quite different from those in traditional inorganic cells wherein charge generation occurs in the same material phase. However, sinc e the charge generation and recombination both happen at the polymer f ullerene interface, they become the determining factors to the performance in polymer solar cells. In addition efficient charge extraction and collection rely on maximized built in potential which is realized by making Ohmic contact with the polymer and fullerene phase s As a result, significant progress in power conversion efficiencies and stabilities of polymer solar cells has been achieved using semiconduct ing metal oxid es as charge extraction interlayer s. Both n and p type t ransition metal oxides with good transparency in the visible as well as infrared region make good Ohmic contacts to both donors and acceptors in polymer bulk heterojunction solar cells. Their compati bility with roll to roll processing makes them attractive for low cost manufacturing of polymer solar cells. In this chapter, we will present the recent results on some of these metal oxides along with the device performance of the solar cells using metal oxides as charge collection i nterlayers. 5.2 Introduction T he power conversion efficiency of PSCs has been driven by the development of photoactive materials 28 115 and device architectures 59 116 118 In particular, variant
87 polymers with donor acceptor building block ha ve been synthesized to control the highest molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels. Such bandgap engineering allows many low bandgap polymers to reach low lying HOMO energies resulting in enhancements in short circuit current s (J sc ) as well as open circuit voltage s (V oc ) which are the two key parameters for photovoltaic cells. Using these polymers PCEs of 7~9% have already been demonstrated along with the interlayer materials, 29 31 32 119 122 among which transition metal oxides become attractive because of their capability of efficient charge extraction and the ir compatibility with roll to roll (R2R) processing 123 124 While s olution processed electron extraction materials such as zinc oxide have been successfully used in both small area device s or scaled up module s further development of solution processed high work funct ion metal oxides is still needed. Most PSCs apply the bulk heterojunction s (BHJ s ) structure 125 wherein an electron donating polymer and an electron accepting fullerene derivative form nano scaled interpenetrating networks allow ing efficient exciton dissociation and carrier transport. Although the photovoltaic effect of an excitonic ce ll does not entirely rely on the built in potential established under thermal equilibrium, a good solar cell performance still requires the Ohmic contact between the absorber and electrodes. U nlike inorganic solar cells, where Ohmic contacts can be formed via surface doping, PCSs need alternative methods to reduce carrier injection/extraction barrier In general, poor Ohmic contact s between the organic layer and electrodes are due to the mismatch of work function, 126 127 the interfacial dipole s 128 130 and a large number of trap states 131 132 Among the electrode interlayer materia ls used in high efficiency PSCs transition metal oxides are
88 very successful owing to their better environmental stability higher optical transparency and easier synthesis routes with compar ed with alkali metal compounds, 133 136 aqueous conducting polymers 137 140 and conjugated polyelectrolytes 120 141 The PCE of PSC s is a product of J sc V oc and fill factor (FF) J sc is determined by the light absorption and the charge collection under large extraction fields V oc is limited by intensive recombination as well as the energy band alignment between the photoactive polymer donor and the fullerene acceptor FF is affected by the series/ shunt resistance and the charge recombination/ extraction rate under low internal fields. The n and p type semiconductivities in m etal oxides are in general due to the intrinsic point defects such as atom ic vacancies present in the oxides Figure 5 1 shows the HOMO as well as LUMO energy levels of some of the state of the art photovoltaic polymers, fullerene derivatives and the electron affinities and ionization potentials of high/low work function metal oxides 142 These metal oxides form Ohmic contact s to photovoltaic polymers and fullerenes through favor able vacuum level shift, energy level bending and Fermi level pinning at th h e polymer/electrode interface s In addition the use of oxide interlayers circumvents the direct contact between photoactive polymer and electrodes where high densit ies of carrier traps or unfavorable interface dipoles hinder efficient charge collection. Making Ohmic contact s is equally important to V oc because the reduction of built in potential by direct contact between the photoactive polymer and the electrodes leads to a decrease of quasi fermi level splitting as well as an increase of carrier recombination.
89 Figure 5 1 (color) Energy diagrams of state of art photovoltaic polymers and transition metal oxides that are commonly used for interfacial engineering 5.2.1 Zinc Oxide (ZnO) ZnO is superior as an electron transporting layer (ETL) materials in PSC s. The Fermi level at 3.7 ~4. 3 eV 119 143 makes it a perfect match with TCO and the lowe st unoccupied molecular orbital (LUMO) of [6,6] phenyl C 61(/ 71 ) butyric acid methyl ester (PC 61/ 71 BM) so far the most effective electron accepting materials when blended with photovoltaic polymers. The ionization potential of ZnO (~7 eV) is enough to serve as an excellent hole blocker to suppress hole transport across the cathode region and therefore to increase the shunt of the BHJ diode. Optically the absorption band of fullerenes and photoactive polymers usually ranging from blue to near infra red region, are just able to locate within the long pass window defined by the band gap of ZnO and thus get protected from ultraviolet (UV) illumination. In contrast to the deposition techniques used in other thin film solar technologies, the ZnO used for PSC s is aimed to
90 solution process at low temperature and ambient condition. With such requirement s some earlier developed ZnO nanomateirals such as colloidal nanoparticles and sol gel ZnO films were consequentially incorporated in PSC fabrication For pra ctical purpose, the most attractive feature of NPs is that it allows the material to be utilized in crystal line phase, although at nano scale, without the requirements of vacuum, high temperature annealing or prolonged fabrication time. ZnO NPs have such a dvantages to fit roll to roll (R2R) technologies as it has already been demonstrated by modified slot die coating 124 for large area BHJ cells fabrication For the lab scale device fabrication ZnO NPs are usually deposited through spin coating as a bottom or top interlayer depending on the device geometry. Figure 5 2 (color) Transmission electron microscopic images of ZnO colloidal nanoparticles with a scale of A) 5 nm and B) 50 nm ZnO NPs are readily synthesized from zinc acetate dihydrate in alcoholic solvent with either of the following strong base sodium hydroxide (NaOH) 144 potassium hydroxide (KOH) 145 147 or tetramethylammonium hydroxide (TMAH ) 119 148 The synthetic wurtzite ty pe ZnO NPs have diameter s of 3 ~ 6 nm, band gap 3.3~3.6eV and form
91 transparent colloid s in polar or non polar solvent 119 144 148 Figure 5 2 (a) and (b) show the TEM images with the scale bar of 5nm and 50 nm repectively The XRD feat ures are also shown in Figure 5 3. Figure 5 3 (color) X ray diffraction of ZnO nanoparticles. The pattern reveals the wurtzite structure. 5.2.2 Molybdenum Oxide (MoO 3 ) The application of MoO 3 started with the exploration of efficient hole injecting material s for light emit ting organic semiconductors with deep HOMO levels 149 Owning to the low melting poin t temperature (795 o C) molybdenum oxide can be deposit ed through thermal evaporation which allows accurate thickness control at nanometer scale Due to the well known hole injecting capabilities MoO 3 was once misidentified as a p type semiconductor until the n type characteristics were directly proved by ultraviolet / inverse photoelectron spectroscopy (UPS /IPES ) 150 Similar to TiO x the d band of MoO 3 remains unoccupied and cont ributes to the conduction band of the oxide.
92 E vident by XPS and UPS measurement s oxygen deficiency in e MoO 3 gives rise to the defect band which lifts the Fermi energy closer to the conduction band 151 152 Like other n type oxides, e MoO 3 forms Ohmic contact with ITO through efficie nt electron transfer with the substrate. According to the UPS spectra upon the deposition of MoO 3 onto the ITO substrate, the interface experiences a significant vacuum level shift of 2 eV which aligns the Fermi level of MoO 3 and ITO. Summarized from few independent in itu measurements carried out under ultra high vacuum (UHV, <10 10 Torr), the work function of evaporated MoO 3 locates at 6.7 0.2 eV below the vacuum level with a separation of 0.5 0.2 eV and 2.7 0.2 eV to its co nduction band and valance band respectively 150 152 155 Some further study reveals that the work function of MoO 3 exhibits strong depend ence on the stoichiometry and surface contamination W ith oxygen or air exposure, the work function decreases dramatically to 5. 3~5.7 eV 154 156 which is still enough to keep the hole injection capability uncompromised 154 Further reduction of the suboxide contributes to the significant extension gap states towards the Fermi energy eventually resulting in the metallic behavior of MoO 2 157 The mechanism of Fermi level pinning between a high work function oxide, such as MoO 3 and organics has been intensively studied using UPS It was reported that the HOMO level of aluminum phthalocyanine chloride (AlPcCl) shifts by 1.6 eV within a thickness of few nanometers, leading to an energy level bending on the organic side of the interface. S imilar HOMO level change is also directly confirmed by studying the diphenyl bis ( 1 naphthyl ) b iphenyl 4,4 diamine ( NPD)/ MoO 3 interface 152 158 Such energy level bending is caused by strong electron transfer from the HOMO levels of organic materials to the 4d band of Mo, which result s in a Fermi
93 level pinning within a depletion region as thin as few nanometers The guiding rule of forming Fermi level pinning and Ohmic contact can be well explained by thermodynamics that, regardless to n type or p type oxides, given their work functions are higher than HOMO levels of adsorbed organic materials, the HOMO offset, which is defined as the difference between the HOMO levels and Fermi energy after equilibrium, will approach to a fixed minimum value which indicates the formation of Fermi level pinning 159 5.2.3 The Inverted Device Geometry As mentioned earlier, in organic electronic devices, Ohmic contacts are made by matching the electrode Fermi level to the HOMO or LUMO levels of the organic layer. Such requirement does not affect the selection of anode materials since the HOMOs of most common organic electronic materials have a larger any than the (5.2 eV) by the oxygen in amb ient, which means anode materials are generally stable in the air. However, the cathode materials are prone to oxidization because of their low work function. In particular, alkali metals based top electrodes are known for very fast degradation in ambient. Even with a work function larger than 4.2 eV, aluminum still cannot be used as a top electrode in the roll to roll printing process which cannot be conducted in vacuum. Silver paste is a good candidate for the top electrode materials, but its work functio n is too low to serve as a cathode. Those TCO with relatively low work functions cannot be directly sputtered organic materials due to the softness of the functional organic layer as well as the requirement of excluding vacuum deposition. As a result, ins tead of using a top cathode bottom anode structure which is referred as the standard or normal device structure, an inverted structure with a top anode and a bottom cathode has become s prototypical for not only the small area
94 device demonstration but also the module production in industry. As shown in Fig ure 5 4, the top electrode of an inverted cell can be deposited by printing silver paste. The bottom TCO, with a relatively low work function, will not affect the roll to roll process in ambient because it is sputtered or coated before the coating the whole organic stack. Some nanocrystalline, solution processed metal oxides can further more modify the bottom cathode work function, resulting in better solar cell performance Figure 5 4 (color) The illustration of normal A ) and inverted B ) structure for polymer solar cells. The HOMO levels of high efficiency photovoltaic polymers generally resides between 5.2 eV and 5.6 eV by considering the trade off relati onship between Voc and optical absorption band 160 According to the criteria of forming Ohmic contact between metal oxides and organic materials, as summarized in Table 5 1, MoO 3 can be applied to almost all the advanced photovoltaic polymers that have shown promising PCE values. Furthermore some injection studies revealed that using sequentially evaporated MoO 3 and silver as a top electrode exhibits even better Ohmic contact tha n that directly deposited on the bottom transparent conducting substrate 161 This makes the inverted device architecture, wherein MoO3 as a top anode interlayer, one of the prototypical structures for the lab scale device demonstration. The P3HT:PC 71 BM cells fabricated
95 using such inverted structure are reporte d with PCEs 4.5~5% 58 162 consistently higher than the traditional structure using PEDOT:PSS and low function metals as electrodes. The results of higher efficien cy cells will be presented in the section of result discussions. In this chapter I will discuss the synthesis and characterization of some of the vacuum deposited and solution processed metal oxides including zinc oxide, molybdenum oxide for interlayers in PSCs I will also describe how their intrinsic defects affect the performance of some of the state of the art PSCs. 5.3 Experimental ZnO NPs were synthesized using the zinc acetate dehydrate and tetramethylammonium hydroxide (TMAH) as precursors. They were used as received from Sigma Aldrich. The details of synthesis and microscopic characterization were also included in earlier publications 163 164 Before devices fabrication, the ITO coated glass substrates were cleaned by acetone, isopropyl alcohol and DI water. ZnO NPs were sp i n coated onto cleaned ITO s ubstrates at 8000 rpm to reach a 25 3 nm film The ZnO film was placed under UV lamp which emits at 254nm for 10 min. The P DTG TPD: PC 71 BM (1:1.5) blend was dissolved in chlorobenzene (CB) with 5% 1,8 diiodoctane (DIO) in volume. The P DTG TPD: PC 71 BM solution has a conc entration of 20mg/ml and was sp i n coated onto ZnO film to reach an optimum thickness around 105 nm. Then the film was annealed at 80 o C for 30 min. 5 nm thick Molybdenum Oxide (MoO 3 ) and 100nm silver was thermal ly evaporated under the va cuum of 1 10 6 Torr. Pe r kin Elmer LS55 fluorescence spectrometer and Lambda 750 UV/VIS spectrometer was used for photoluminescence and abso rp tion experiment s. The transient photo current setup was
96 composed of a Tekstronix T P S2 024 oscilloscope a DTL 319QT p ulse laser source from LASER EXPORT Co. Ltd (532nm, 8ns time width), a Stanford research system (SRS) DG535 delay generator and a n ORIEL Xe DC arc lamp for light biasing Similar experiment has been published with detailed experiment process 165 Devices were encapsulated before measurements. Photo current voltage (J V) characterization was done with a Keithley 4200 semiconductor characterization system an d a Newport Thermal Oriel 94021 1000W solar simulator ( 4 in. by 4 in. beam size). The light intensity was determined by an ORIEL 91150V monosilicon reference cell calibrated by Newport Corporation. EQE measurement was conducted using a Xe DC arc lamp as light source, a n ORIEL 74125 monochromator, a Keithley 428 current amplifier, a n SR 540 chopper system and a n SR 830 DSP lock in amplifier from SRS The generated EQE spectrum was integrated with A.M. 1.5 G spectrum to compare with the measured sho rt circuit current value. 5.4 Results and Discussions 5.4.1 Inverted Polymer Solar Cells With the demand of demonstrating PSCs with higher efficiencies as well as the competitive strength as a candidate for the next generation thin film solar cells, recen tly reported advanced photovoltaic polymers not only feature the extension of absorption band, but also maximization open circuit voltages by tuning the energy levels of HOMO. Here, we first introduce a novel donor acceptor block copolymer containing dithi enogermole (DTG) and 1,3 dibromo N octyl thienoporrolodione (TPD). The polymer was synthesized and characterized by the SoRey joint group.
97 Figure 5 5 The monomer structure of PDTG TPD and PDTS TPD. As shown in Figure 5 5, the dithienogermole unit has a germanium atom in the bridging position as in contrast with the previously reported dithienosilole (DTS) unit which referred in the following part of the section. The thienoporrolodione unit was chosen to couple with DTG b ecause several groups reported that this strong electron accepting unit can result in high open circuit voltages. The comonomers were polymerized using the Pd2dba3/P(o tol)3 catalyst system, and resulted in number a verage (Mn) molecular weights of 31kDa (P DI 1.7) for P DTS TPD and 48kDa (PDI 1.7) for P DTG TPD. Figure 5 6 illustrate the UV visible spectra of pristine polymers films coated on quartz substrates. P DTS TPD shows an excitonic absorption peak at 670 nm, corresponding to a n optical bandgap around 1.73eV, consistent with previous reports. Upon substitution of the silicon atom for germanium in P DTG TPD a red shifted absorption spectrum was observed with respect to P DTS TPD with the excitonic peak at 679 nm and a n optical bandgap of 1.69 eV. Electri cal chemical measurements reveal the HOMO level at 5.60 eV for PDTG TPD, 50 meV smaller than the silicone counterpart.
98 Figure 5 6 (color) The absorption of PDTG TPD and PDTS TPD measured in solution phase. Inverted solar cells were fabricated using P DTS TPD: PC 71 BM and P DTG TPD :PC 71 BM as active layers in inverted device architectures : ITO/ZnO (NPs) / p olymer:PC 71 BM/MoO 3 /Ag. Figure 5 7 shows J V curves of both polymers under A.M 1.5 G illumination with and without diiodooctane (DIO) as a processing additive. T he short circuit currents and fill factors of both cells are greatly enhanced by adding 5 vol % DIO into the polymer:PC 71 BM solution Upon optimization, PDTS TPD reached an average short circuit current density of 11.5mA/cm 2 an open circuit voltage of 0.89V, and a ll factor of 65%, resulting in an average power conversion efficiency of 6.6%. The performance is slightly lower than the earlier reported devices made from a regular structure. The DTG containing polymer P DTG TPD displayed a higher short circuit current density (12.6mA/cm 2 ) and fill factor (68%) than P DTS TPD despite of a lower
99 Voc of 0.85V, for an average PCE of 7.3%. The lower Voc of 40mVfor P DTG TPD is in good agreement with the above mentioned electrical chemical measurements Figure 5 7 (color) The J V characteristics measured under A.M. 1.5G condition. The external quantum efficiency (EQE) spectra of the devices are shown in Figure 5 8 and it can be seen that without DIO both devices display fairly low quantum efficiency With DIO as a solvent additive, the photo response of both PSCs increase s significantly The EQE values of P DTG TPD ranges from 55 to 65% within the absorption band of the polymer while that of P DTS TPD ranges from 50 to 56%, with P DTG TPD showing a n onset at a longer wavelength. This is consistent with the slightly red shifted absorption spectrum of P DTG TPD with respect to P DTS TPD The 7.3% PCE demonstrated here was the highest record value of the PCSs using any inverted device structure and is still among the highest values for the single junction PSCs ever reported. Some better device performance due to further optimization of Zn O layer will be shown in the next section.
100 Figure 5 8 (color) The external quantum efficiency (EQE) for the same devices in Figure 5 7. 5.4.2 Inverted Cells with Reduced Interface Recombination In inverted polymer solar cells, solution processed ZnO is widely used in the form of nanoparticles 123 146 162 166 ZnO NPs usually have diameters less than 10nm, are known to have a large density of point defects I t has been shown that almost 30% of the atomic bonds in one ZnO nanoparticle are dangling bonds 167 When using ZnO to make so lar cells, a common and easy way to address defects issue is to light soak 168 the device for a period of time after the device fabrication During this process, incident photons excite photo carriers to fill the mid gap states of ZnO NPs and electrons in the conduction band will be able to transport through the ZnO NPs without getting recombined in the mid gap states which explains why J sc V oc and FF increase upon soaking. For inorganic thin film solar cells, it was also found that when ZnO serves as buffer layer, strong light soaking e ffect was observed 169 170 However, these defects are not eliminated during illumination because the photo carriers that occupy the mid gap
101 states will finally recombine or relax into valance band leaving mid gap states once again unoccupied Knowing that light soaking cannot completely passivate the defect states, UV ozone treatment was done on the ZnO NPs films. ZnO NPs were sp i n coated on ITO substrates, subsequently annealed to remove the solvent from the film s. ZnO NPs films were then exposed to a UV lamp (254 nm emission, 50W) before the deposition of the polymer: PC 71 BM film. In order to confirm the passivation effect by UVO treatment photoluminescence experiments were carried out to characterize the ZnO NPs film s As shown in Figure 5 9 the as ann ealed film shows an emission band peaked at 3 72 nm which is indicative of band edge emission. Upon UVO treatment this band edge emission increases slightly. In addition to band edge emission, we observed a strong broadband emission with a maximum at 5 19 nm for untreated films This broadband emission has been reported as an evidence of the presence of defect states due to oxygen vacancies 171 172 zinc vacancies 172 173 and interstitials. Estimated from the spectra integration, photo excited elec trons have more chance to fall into mid gap states than a direct band to band transition Upon a UVO treatment, the defects emission almost disappears while the band edge emission is enhanced. When the defect population decreases, the band to band transiti on becomes dominant. It should be noted that the UVO treatment is no t just a surface treatment it also passivates the defects in the bulk as PL emission is a bulk effect Moreover the PL data helps to distinguish the defects passivation effect from previ ously discussed light soaking. In the PL measurements, the excitation source actually soak s the ZnO films with UV during the 2 minutes scanning time which is much longer than the 1~2 seconds required for
102 midgap states filling. According to the resultant PL spectrum, however, these still show significant defects emission indicating light soaking cannot completely fill the defect states. In contrast, the defect passivation by UVO treatment results in a change in material structure and energ y band structures instead of a temporal defect states filling. Based on the PL study, we propose the defects passivation in ZnO NPs films will help to reduce interface recombination between the ZnO and photoactive layer which contributes to a photo current loss in the inverted PSC we shown in the last section. The correlation between interface recombination and defects is well known in the case of inorganic thin film photovoltaics. For c opper indium gallium (di)selenide (CIGS) solar cells, the recombination at the ZnO surface can be greatly reduced by inserting a CdS buffer layer between the CIGS and the ZnO layers 174 Figure 5 9 (color) The photoluminescence of ZnO nanoparticles films with and without UVO treatment. To finish the fabricat ion of inverted solar cells, both P DTG TPD and P DTS TPD were mixed with PC 71 BM as the photoactive layer. Here, the optimal blending ratio s for both polymer: PC 71 BM blends are 1:1.5 in weight. The optimu m composition ratio
103 principally give s a balanced transport of electrons and holes 92 In addition, similar to the presentation in the last section, 5 % (volume) solvent additive 1,8 diiodoctane (DIO) is required to fully dissolve the PC 71 BM in chlorobenzene (CB) in the presence of P DTG TPD or PDTS TPD Fully dispersed PC 71 BM is of great importance to the blend morphology since aggregates of PC 71 BM generates over sized domains which inhibit exciton diss ociation and carrier transport 175 The current density voltage (J V) curves of P DTG TPD: PC 7 1 BM and P DTS TPD: PC 71 BM solar cells J V curves are shown in Figure 5 10 and 5 11 Upon treating the ZnO layer with UVO, both solar cells demonstrate a 10% enhancement of J sc The device characteristics of these cells are shown in Table 5 1 As shown in th e table, the controlled P DTG TPD: PC 71 BM cells gives a short circuit current of 12.9 mA/cm 2 while the devices with UVO treated ZnO NPs film have an average value of 14.1 mA/cm 2 The FFs are increased in both P DTG TPD and P DTS TPD devices, indicat ing the pa ssivation of ZnO NPs enhances charge collection even at very low extraction fields for both material systems. V OC was observed as unchanged indicating that there is no change in interface energetics. The EQE data in Figure 5 10 confirms the enhancement of photo current. As shown the data, a maximum EQE o f 7 2 % percent is very high for inverted polymer solar cells. Integrating the EQE spectrum with A.M. 1.5G spectrum gives a current of 14.0 mA/cm 2 for the P DTG TPD: PC 71 BM cell. Th e integrated photocurrent is within 2 % of the direct measured value of J sc In addition, the P DTG TPD is the first reported polymer giving a PCE over 8%. For P DTS TPD :PC 71 BM cells we obtained an average 7.8% PCE after UVO treatment of ZnO NPs films, this is even higher than previously reported best devices in conventional structure 31
104 Figure 5 10 (color) The J V ( A.M. 1.5G) and EQE spectrum of inverted PDTG TPD:PC 71 BM cells. Figure 5 11 (color) The J V (A.M. 1.5G) and EQE spectrum of inverted PDTS TPD:PC 71 BM cells. As an important part of the work in this chapter, I made PDTG TPD:PC 71 BM devices for the power conversion efficiency certification (Newport Calibr ation Cert. #0341) carried out in the TAC PV lab of Newport Inc. Due to some encapsulation problems, the certification results averaged at 7.4%, which was lower than the laboratory measured values. When this chapter is being drafted however, the 7.4% effi ciency is still recognized as the highest value of inverted polymer solar cells ever reported on an academic journal.
105 Table 5 1 S ummary of device performance Polymer J sc (mA/cm 2 ) V oc (V) FF (%) Average PCE (%) PDTG TPD 12.8/14.1 0.86/0.86 66.8/67.3 7.4/8.1 PDTS TPD 11.4/13.1 0.90/0.90 64.8/66.5 6.6/7.8 Figure 5 12 (color) The certification from Newport PV lab. The device was made by Song Chen. It can be seen from the EQE spectrum and J V curves, without the defect passivation on ZnO, the devices exhibit a photo current loss about 10% of the J sc value ~ 1.3mA/cm 2 Such a difference should be reflected in the photo carrier lifetime measured under short circuit condition 57 In order to see the difference of recombination, transient photo current measurements were done on devices with and without UVO treatment. To carry out these measurements, the active layer of the device is excited with a pulsed laser with an emission wa velength at 527 nm with an intensity of ~1 nJ/pulse while the sample is also under light biased from a solar simulator It should be noted that the 527 nm excitation does not excite the ZnO NPs and the
106 photoexcitation is primarily due to the polymer blends The photocurrent transient was measured with a 3 0 resistor connected in series with the solar cell simulating the short circuit condition. The single exponential decay of the transient photo current as shown in Figure 5 13 is the result of laser pulse generated carriers recombining either in the bulk or at the interfaces. According to the decay of photocurrent perturbation the photo carriers lifetime is around 130ns and 2 1 whic h is much shorter than the results measured with high level excitation by which the photocurrent decay mainly characterizes the carrier extraction processes and thus deviates from the single exponential dependence Considering the difference of carrier lif etime, we noticed all the devices have the same photo active layer and p contact. Therefore, the difference in carrier lifetime can only be attributed to the interface recombination at the ZnO/photo active layer. Although recombination also happens in the bulk materials, it is unlikely to overwhelm since the carriers concentration is much lower at short circuit condition than open circuit condition 35 Thus, the carrier concentration decay due to bulk recombination is supposed to exhibit a much longer li fetime. The physic al picture of defects induced loss mechanism can be explained as: one side, the electron hole pairs dissociated at the polymer fullerene interfaces experience bimolecular recombination when they are in contact with ZnO; the other side, th e het e rojunction formed between polymers and ZnO has poorer exciton dissociation with the existence of a high defect density. With the transient photocurrent measurement, we conclude that defects are passivated by UVO treatment resulting in an increase in carrier lifetime.
107 Figure 5 13 (color) The carrier recombination measured by transient photocurrent (s mall signal mode). In order to confirm the enhancement of photocurrent is not caused by the change in work function of the ZnO NPs layer after UVO treatment, we studied the change in build in potential by electroabsorption (EA) 73 176 on ZnO NPs/ polymer/ Al devices with and without UVO treatment The purpose of using this device structure instead of the actual PV devices for the EA measurements is to suppress carrier injection under flat band condition thus maintaining a uniform field distribution across the polymer layer during the EA measurements. Figure 5 14 shows that the first harmonic EA signals vary linearly with t he applied voltage bias until the internal field reaches zero. The flat band condition is reached when the DC bias compensat e s the built in potential (V bi ) which is determined by the work function difference between the ZnO NPs layer and the Al electrode With UVO treatment on the ZnO NPs, as shown in Figure 5 14 the built in potential of the ZnO NPs/ polymer/ Al device only changes from 0.55 eV to 0.51 eV, indicating a 0.04 eV increase of work function on the ZnO NPs. This small change of
108 work function i s not favorable for electron extraction and thus cannot be the reason for the J sc enhancement. Figure 5 14 (color) The DC dependent electroabsorption 1st harmonic signals. As already shown, the effect of UV treatment works on the two deep HOMO polymer PDTG TPD and PDTS TPD. In order to better understand the physic al process es due to the UV treatment we applied the same UVO treatments to inverted P3HT:PC 71 BM cells and MDMO PPV: PC 71 BM cells Unlike the positive effect when usi ng P DTG TPD and P DTS TPD, we see significant reduction in V OC and FF, suggesting significant surface properties change on polymers after UV O treatment s 177 For polymers with shallow HOMO energies like P3HT and MDMO PPV, the excess oxygen present at the ZnO surface due to the UVO tre atments not only helps to repair the surface defects but also tends to oxid ize the polymers. But extra oxygen may also be adsorb ed on the ZnO NPs surface s when they make the contact with polymer films,
109 surface oxidization is possible. Therefore, we see UVO treatment significantly reduces the V OC and FF. According to the HOMO levels of these polymers, the threshold of oxidization level should be between 5.4eV and 5.6eV. To sum up this section, we presented a simple process ing method to enhance the PDTG TPD:PC 71 BM cells from 7.4% to 8.1%. The defects passivation mechanism is confirmed by PL spectrum and transient photo current decay. The same effect was seen on PDTS TPD:PC 71 BM, but not on shallow HOMO polymer :PC 71 BM system such as P3HT and MDMO PPV. The EQE for the optimum cell is as high as 72%, indicating the PDTG TPD is a promising polymer for photovoltaic application and UVO treatment on ZnO NPs films can significantly enhance the device performance. 5.4.3 Device Stability In the previous section, we discussed the mechanisms of PCE improvement in the inverted PDTG TPD:PC 71 BM and PDTS TPD:PC 71 BM cells. The efficiency values are very encouraging for such a device structure that is compatible with large scale printing process. However, the other side of the device performance stability is yet characterized. It would be interesting to learn if the organic photo active absorber, which is sandwiched between two metal oxide layers, can perform still perform well after a ve ry long storage time. Figure 5 15 shows a J V curve of a 15 months aged PDTG TPD cell which owned a PCE of 8.1% right after device fabrication. As it can be seen, with a simple encapsulation using cavity glass, getter and epoxy, the PCE maintains 89% of th e original efficiency after more than one year. T his is strong evidence showing the great device stability of our inverted polymer solar cells.
110 Figure 5 15 The J V of a PDTG TPD:PC 71 BM cell made in June 6 th 2011 and re measured in August 29 th 2012. The original PCE is 8.1%. 5.5 Summary In this chapter, we first review the importance of interfacial engineering of polymer solar cells from the perspectives of device physics and materials science. These metal o xides have unique advantages when serving as charge extraction interlayers in PSCs. The combination of ZnO, MoO 3 is a good example that both anode and cathode interface benefit from the efficient charge collection through these metal oxides. Using such a s trategy, we fabricated inverted PSCs using PDTG TPD a novel dithienogermole containing D A polymer. Devices show a preliminary PCE of 7.3%. With further improvement of ZnO layer, we are able to reduce the recombination at the cathode interlayer region and thus enable an additional efficiency enhancement. The optimum performance is up to 8.1%, which is not only a record value for inverted polymer solar cells, but also remains a value of 7.2% after a storage time of 15 months.
111 CHAPTER 6 O PEN CIRCUIT VOLTAGE LOSS 6.1 Abstract In polymer bulk heterojunction (BHJ) photovoltaic (PV) cells, it is generally found that the open circuit voltages ( V OC s) are strongly depend ent on the energetic alignment at the junction of the electron donors and electron acceptors. However, it has been a challenge to directly measure the energetic alignment at the bulk heterojunction of an actual device and that severely limits our understandin g of the origin of the V OC loss. In this work, we demonstrate that the bulk heterojunction alignment can be directly measured using charge modulated electroabsorption spectroscopy (CMEAS) technique. Above a sub bandgap optical pumping energy, organic molec ules are excited to the higher level s of charge transfer (CT) states through which ultra fast exciton dissociation occurs subsequently at the polymer fullerene interface. These separated electron hole pairs are also active enough to couple with the modulat ion electrical field and introduce subtle changes in the in the sub bandgap region of the absorption spectrum Using CMEAS, we directly measured the effective bandgaps of a series of high efficiency polymer:fullerene systems. We also sought to investigate the loss of open circuit voltages by combining the effective bandgap with recombination study. In contrast to the previous findings that the V OC varies linearly with the polymer effective bandgap, we found that the deviation from linearity is due to the co ulombic nature of the Frenkel excitons in the bulk heterojunction system with lower dielectric constants The coulombic loss is further confirmed by the temperature dependent measurement of open circuit voltages.
112 6.2 Introduction 6.2 .1 Effective B andgap P olymer solar cells with power conversion efficiencies (PC E s) over 8% have been demonstrated in laboratories with the advance of novel materials film processes, and device architectures 119 120 122 178 however, many arguments regarding the polymer:fullerene bulk heterojunction (BHJ) such as the charge transfer process and the energy alignment still remain to be further studied Typically, a BHJ photoactive layer is formed by blending an electron donat ing polymer and an electron accepting fullerene derivative the energy alignment between th e s e two phases directly impacts the performance of polymer solar cells. For example the open circuit voltage ( V OC ) in organic PV cells is related to the energy level difference of the highest occupied molecular orbital (HOMO) of the donor and the lowest unoccupied molecular orbital (LUMO) of the acceptor 160 179 Therefore, making polymers with low lying HOMO energies is a commo n strategy to reach large V OC s In addition, the LUMO energy offset between polymer and fullerene is referred as the interfacial driving for ce to split the excitons generated in the polymer. A minimum value such offset is still under debate. Likewise, the HOMO offset helps to separate the excitons formed in the fullerene domains. Based on these half empirical, half experimental conclusions, an optimum energy alignment was tentatively proposed for the ideal polymer to form the best matchup with fullerenes. As a result controlling such energy level alignment has been the rule of thumb to manipulate the photovoltaic properties of a BHJ cell Meanwhile, in contrast to its significance to the framework of de vice physics, the understanding the energy level alignment is limited by the lack of experimental techniques that can directly measure the energy levels of the donor and acceptor phases in a BHJ PV device
113 Previous attempts to determine the energy level alignments in polymer BHJ systems preliminary relied on the measurements of the pristine materials However, the HOMO or LUMO levels obtained by electrical chemistry methods were lack of accur acy and repeatability, resulting in a very rough and inconsistent estimation of E eff s in the BHJ films For a better study of energy alignments, the impact of polymer fullerene interaction has to be considered. First, the effective bandgap should be determ ined by the sub bandgap photo excitation which includes the participation of those meta stable charge transfer states. In addition, upon the formation of polymer:fullerene interpenetrating network, the electrostatic environment of either material is readil y altered by the modification of dielectric properties Also due to the difference of electron affinities and intrinsic electron densities, the charge transfer between the polymer and fullerene derivatives forms interface dipoles and results in further en ergy level shifts. Last but not the least, a lot of polymers were reported with variant HOMO levels because different reference electrodes were used in the electrical chemistry measurements. Based on these electrical chemistry measurements, it has been sho wn that the origin of V OC loss in polymer BHJ systems has been attributed to excitonic effect 179 180 polaronic effect 181 and the nonradiative recombination from charge transfer (CT) states which can be further incorporated in the classical relationship between the V OC and reverse saturat ed dark current ( J 0 ) of a diode 182 As an earlier attempt, Scharber et al reported that the V OC s of a series of BHJ PV devices followed a linear relationship with the energy level difference of the pol ymer HOMO and fullerene LUMO levels 160 i.e.: ( 6 1)
114 Here, | E D HOMO E A LUMO | was referred as E eff of the donor/acceptor blend material and 0.3 V is a n arbitrary empirical constant value related to the V OC loss compensat ing the forward biased dark current. Similarly, the interface driving force of exciton dissociation E dis is also estimated in such a way. Considering the success of the first organic hetrojunction solar cell, people predicted a significant LUMO offset e.g. 0. 3 eV is required to counteract the exciton binding energy While some later arguments pointed out some efficient advanced donor effective force since the electron hole wave functions are already fairly separated upo n the exciton generation in a repeating block. To address such a problem, an accurate measurement of the energy alignment needs to be carried out. I t is quite possible that the polymer fullerene interaction changes the energy alignment resulting in a larg er or smaller effective force than what is predicted by the electrical chemi stry methods. Meanwhile, some recent efforts aiming at better probing technologies worth being recognized. To date, t he most direct way to measure the effective bandgap energy is ultra violet and inverse photoelectron spectroscopy (UPS /IPES ). For example, Guan et al obtained an E eff value of 1.4 eV for P3HT:PC 61 BM by UPS /IPES measurements 183 Ratcliff et al also reported the PCDTBT:PC 71 BM syste m has an E eff of 1.5 eV 184 UPS /IPES is an excellent tool to study small molecule systems where in situ measurements can be performed during the evolution of the bulk heterojunction, however, it does not necessar ily provide the accurate energetic information in the solution processed films The fact that the above cited E eff values are much larger than those estimated from electrical chemistry methods also raises questions whether the inverse electron technology is sensitive enough to the meta stable charge transfer (CT)
115 states and the tail s tates of LUMO levels. As can be seen, the physical process during a UPS/IPES measurement is different with the optical pumping from a polymer s HOMO levels to PCBM s LUMO band. In addition, as a typical surface sensitive technique, UPS/IPES doesn t probe t he energy alignment in the bulk materials. In addition to UPS /IPES several other techniques such as charge transfer state photoluminescence 180 and electroluminescenc e 185 186 have been attempted to determine the ground CT states (CT 0 ) In these measurements before the emission occurs, excitons generally experience significant energy loss during the relaxation from hot CT states to CT 0 Thus, the sub bandgap energies determined from CT 0 emission are supposed to be lower than the excitation energy from the polymer to the fullerene, which is defined as E eff Here we also point out that the low radiative signal from photovoltaic materials may limit the applicability of this technique to donor acceptor systems. In this work we demonstrate a simple method charge modulated electroabsorption spectroscopy (CMEAS) technique to directly determine the actual energy alignment in a polymer:fullerene BHJ blend. By combining the electroabsorption (EA) technique and the charge modulate d (CM) optical absorption of polymers, we are able to observe a clear sub bandgap signal through direct excitation of electrons from the HOMO level of the polymer donor to the LUMO level of the fullerene acceptor. Above a sub bandgap optical pumping energy organic molecules are excited to the higher level s of charge transfer states through which ultra fast exciton dissociation subsequently occurs at the polymer fullerene interface. These photo generated c harges are able to couple with the modulation electrical field and introduce subtle changes in
116 the sub bandgap region of the absorption spectrum Such a differential spectrum technique has a much higher detectivity than a direct absorption measurement gen erating more accurate results. Using CMEAS, we are able to directly probe the actual effective bandgap and interface effective force of polymer:fullerene systems. 6.2 .2 Recombination and E xcitonic L oss Most p revious attempts to study the V OC loss simply a ttribute the to recombination or polaron self trapping without directly measuring the actual recombination or polaronic loss energy. Using the derivation from a typical J V characteristics the reverse saturation dark current becomes another parameter that can be experimentally determined. In that equation and are the photocurrent and dark current at the open circuit condition. Nevertheless the direct measurement of this small current is highly affected by the leakage current and t he ideality factor larger than unity Fortunately r ecombination and reverse saturation dark current can be connected using the Shockley Queisser balance theory 21 which demonstrated the radiative recombination limit. Some further modification was also presen ted 17 9 187 189 As a progressive step Vandewal et al. achieved more accurate estimations of J 0 and V OC by directly measuring the quantum yield of radiative recombination from CT 0 states. 182 Fr om section 6.1.1, we know an accurate measurement of effective bandgap is important and challenging. With the CMEAS technique, it is possible to probe the effective bandgap of variant polymer:fullerene systems by taking care of the contribution of polymer fullerene interaction and energetic disorder The sub bandgap signals in CMEAS probe the onset of hot CT states. Although the signal is induced by the sub
117 bandgap optical pumping, the energy value extracted from CMEAS spectra is actually electrical gap because the charges are separated through the higher manifolds of the CT states. In addition, since E eff is defined at the onset edge of the spectra, it already includes the effect of energetic disorder. Thus we can tentatively express V OC in the followin g way: ( 6 2) wherein is the hot CT energy without the consideration of energetic disorder which is marked as with a negative value, characterizes the loss of quasi Fermi level splitting due to intensive bimolecular recombination. is the additional energy loss happens after the charge transfer and until the so called fully separated state Bimolecular recombination across the donor acceptor interface determines the carrier lifetime and carrier concentration (see Eq. 4 6 ) at open circuit conditions. 56 It is easy to estimate the voltage loss due to the recombination by assuming t he carrier concentration ~ 10 16 cm 3 N c = N v ~ 10 19 cm 3 and 10 18 cm 3 (roughly corresponding to 10 20 cm 3 eV 1 and 10 19 cm 3 eV 1 ) at the peak and tail position of the DOS respectively. 77 190 Then we get the value of 0.36 eV and 0.24 eV which respectively shows how much quasi Fermi level splittin g decrease if we use the DOS peak or tail as the origin of the energy c oordinate. The ir difference of 0. 12 eV is close to the value of energetic disorder in most photovoltaic polymers 58 191 As we mentioned in C hapter 4, the carrier concentration has the maximum value under open circuit condition. Even this value is no more than the concentration of polymer domains, which means recombination is so intensive that there s very little chance to see two carriers exist simultaneously in one
118 domain. 56 The photo voltage loss due to re combination contributes significantly to the right side Eq 6 2. The voltage loss due to E cs is less mentioned in previous reports. According to the charge generation mechanism, the hole electron across the donor acceptor interface is still loosely bounde d although they are considered to be separated Therefore their energy difference keeps decreasing due to the remained coulombic interaction until the geminate pair is fully separated. 179 Thus, the BHJ systems with lower dielectric constants should show more pronounced excitonic/coulombic effect Moreover, the charge transfer process may play a role to the possible energy loss in this process For example, in a polymer:fullerene system wherein most charge transfer and separation happen through the higher CT manifolds, which is also called hot CT states, the separated charge has more chance to experience energy loss since their wavefunctions are too delocalized for hopping transport. For typical inorganic solar cells, such as amorphous silicon and chalcogenide cells, charge generation occurs withou t the formation of exciton or charge transfer states. Thus the loss of V OC can be fully explained by Shockley Queisser limit plus surface recombination effect 33 192 However, due to the different charge generation process in excitonic cells, 3 the loss mechanism of V OC is expected to be different. The terms on right side of Eq. 6 2 can either be directly measured or calculated since V OC is measurable. Using CMEAS, E eff can be accurately determined with the consideration of energetic disorder ( ). TPV helps to extract the carrier concentration under open circuit condition, which is used to further calculate the recombination loss. Thus, if the value of is
119 significantly larger than the measurement error we can conclude the observation of photo voltage loss due to coulombic interaction. We also expect to see the energy loss is dependent on the dielectric. Also, the magnitude of can be confirmed by doing low temperature measurement of V OC because recombination loss can be minimized at low temperature. Measuring V OC with different light intentisy helps to determine the saturation value of V OC on the V OC vs. temperature plot. 6. 3 Experimental Materials and device fabrication PCDTBT and PC 71 BM were purchased from 1 material and nano c respectively. The PCDTBT:PC 71 BM solution has a concentration of 8mg ml 1 with a polymer fullerene weight ratio of 1:4. 1,8 diiodooctane was us ed as a processing additive into clorobenzene with a volume ratio of 3%. The sample has a structure of ITO/ PCDTBT:PC 71 BM/ Al. The ITO substrate was cleaned using actone and isoproponal followed by a 10 min UV ozone treatement. The solution processed film has a thickness ~ 200nm. 200nm aluminum was deposited as a top electrode using thermal evaporation. The device has a size of 2 mm by 2.3 mm. Electroabsorption measurement The samples were kept in a cryostat with a pressure of 10 3 Torr. The measurements were carried out at room temperature. A monochromatic parrallel beam probes the sample through the ITO side with an incident angle of 45 o and gets reflected by the top electrode. Comparing with the absorption, the reflection on the substrate surface is too low to be concerned. The back reflected signal got further absorbed by the organic layer before being collected by a calibrated silicone detector which is further connected to a current amplifier and lock in amplifier. A calibrated germianium diode was also used to colllect the spectra from 1100 nm to 1500 nm, the data was not shown becasue no signal emerged. A DC voltage was used to
120 bias the solar cell sample, a AC field was applied to modulate the lock in amplifier to allow a low noise measurement.The final signal T/T was simply the ratio of the signals with and without AC field modulation. Carrier concentration and dielectric constant measurement The carrier concentration is estimated using balance equation between carrier lif etime, generation rate and carrier concentration. The carrier liftetime is extracted by TPV measurement. The charge generation rate is obtained according to the photocurrent values obtained at a large reverse bias. The dielectric constant of the sample is determeind by the measurement of geometric capactance measrued under reverse bias. 6. 4 Results and Discussions 6. 4 .1 Subgap S ignal s in EA S pectra To demonstrate the measurement of E eff we pick a typical donor acceptor polymer p oly[ [9 (1 octylnonyl) 9H carbazole 2,7 diyl] 2,5 thiophenediyl 2,1,3 benzothiadiazole 4,7 diyl 2,5 thiophenediyl] (PCDTBT). P henyl C 71 butyric acid methyl ester (PC 71 BM) was used to form BHJs with PCDTBT The d etails of device fabrication are shown in the Expe rimental. EA measurements aim to probe the subtle change of absorption coefficient with an electric field across the sample. A 1000 Hz AC modulation field is used trigger the lock in amplifier. Figure 6 1 shows the normalized EA spectra of (a) pristine PCD TBT, (b) pristine PC 71 BM, and (c) the blend of PCDTBT:PC 71 BM. The devices were under reverse bias to prevent carrier injection during measurements. As shown in Fig. 1 (a), the EA spectrum of the pristine PCDTBT film is dominated by the features ranging fro m 1.8ev to 2.5 eV where the exciton responds to electric fields with a quadratic Stark effect due to the coupling between the exciton ic level and higher forbidden states 69 193 194 The resulting line shape similar but not identical to the first
121 derivative of the unperturbed absorption spectrum as some structural ordering still remains in the pristine PCDTBT film 70 195 Similarly, as shown in Fig. 1 (b), the excitonic and vibronic signals of PC 71 BM are located between 1. 7 eV and 2. 4 eV. Nevertheless, t he signal amplitude of the PC 71 BM sample is ov er 10 times smaller than that from the pristine PCDTBT sample when measured under the same field condition, which explains why the EA spectrum of a PCDTBT:PC 71 BM sample in Fig. 1 (c) shows a very similar excitonic feature to Fig. 1 (a) except a little blue shift and broadening of the main peak The broadening is related to the increased disorder with the increase of PC 71 BM concentration 195 I n addition to the EA response from the pristine materials, we also observed a signal emerging at an energy significantly lower than the bandgap energies of both materials. The sub bandgap signal in Fig. 1 (c) has an onset at 1.40 eV, while the EA spec tra onset of pristine PCDTBT and PC 71 BM is respectively 1. 78 eV and 1. 68 eV. Clearly, this is due to the interaction of the donor and acceptor materials. Comparing to sub bandgap features of PCDTBT: PC 71 BM observed by photo induced absorption (PIA) 14 196 and the CT electroluminescence 186 measurements, the EA onset (~1.40 eV) is energetically higher than b oth polaron features and CT 0 The polaron signals are broadly distributed from 0.8 eV to 1.5 eV with a maxima value continuously relaxing from 1.2 eV to lower energies after photo excitation 14 According to the electroluminescence spectra, the CT 0 energy was determined at 1.15eV~1.20 eV 186 Therefore, it is reasonable to assume the sub bandgap onset of EA spectrum characterizes some higher energy levels d ue to the polymer fullerene interaction i.e. hot CT states (CT n ) which is recently attributed to the gateway for ultra fast exciton separation due to their relatively delocalized electron wavefunctions 19
122 Figure 6 1 (color) The normalize d electroabsorption spectra of A ) pristine PCDTBT, B ) pristine PC 71 BM and C ) PCDTBT:PC 71 BM devices. The dashed lines show the corresponding absorption spectra of the films studied. We sought to investigate the origin of the sub bandgap signals by comparing the characteristics to the excitonic transition signal s Fig. 3 shows the (a) modulating frequency and (b) DC voltage bias dependences of the PCDTBT:PC 71 BM blend device
123 with light biases at the sub bandgap regime at 750 nm (1.65 eV) and the excitonic regime at 620 nm (1.98 eV). Changing the modulation frequency, the excitonic transition signal at 620 nm remains the same level with a small fluctuation due to the measurement noise wh ich is consistent with fact that the Stark effect occurs is an instantaneous response to the modulation of internal electrical field. On the other hand, the signal at 750 nm decreases with increasing the modulation frequency and finally changes the phase a t a frequency below 10 4 Hz suggesting the sub bandgap signal has much slower respon se than the excitonic transition signal. Although the details of frequency response remain further studies, at this stage, we correlate it with the coupling between electri cal field and CT states which can be physically described as a response of electric dipoles. Figure 6 2 (color) Electroabsorption signal of A ) frequency B ) DC voltage dependences of the PCDTBT:PC 71 BM device with optical excitation ex at 750 nm and 620 nm. The solid line in B ) is a linear fit to the data. C ) The AC voltage dependence of the EA signal with optical excitation at 750 nm.
124 Figure 6 3 (color) Illustratio n of CMEAS mechanism. A ) The energy diagram and the illustration of the sub bandgap excitation. B ) The free energy diagram of CMEAS measurement. From cooled CT states to hot CT states, the electron wavefunction gradually turns delocalized, allowing ultra fast exciton dissociation to occur. Sub bandgap CMEAS signal detects the carriers dissociated through hot CT states, the onset of CMEAS signal defines the transition from cooled to hot states, as well as the effective bandgap. Furthermore, by varying the DC voltage bias, the excitonic signal shows the typical first harmonic response of the Stark effect with the magnitude of the EA signal varies linearly with the DC applied electric fie ld 197 resulting in a buil t in potential of 0.9 V. In contrast the sub bandga p signal is significantly less sensitive to the DC electric field indicating its origin is not from the Stark effect Based on the above results, we
125 postulate that the sub bandgap EA signal comes from direct excitation of electrons from the HOMO level of PCDTBT to the LUMO level of PC 71 BM as illustrated in Fig. 6 3 ( a ). Using the free energy state diagram of the BHJ system as shown in Fig. 6 3 ( b ), the excitation dynamic process in PCDTBT:PC 71 BM can be described as following : photons wit h sub bandgap energy directly pump the molecules to the hot CT states through Figure 6 4 (color) CMEAS results of pristine PCDTBT. A ) The current voltage characteristics of the pristine PCDTBT device, the inset show s the d ata in logarithm scale. B ) The voltage dependence of in phase and quadrature signals of the same devices with optical excitation at ex = 1 m. which excitons can dissociate into separated charges at sub picosecond range before relaxing to the lowest lying CT 0 14 198 T he se charged molecules form dipoles across the polymer fullerene interfaces and modulate the absorption coefficient of the blend as a
126 response to the oscillating electric field Such a phenomenon is similar to what has been observed in charge modulation spectroscopy (CMS) which is, however, an electrical pumping process by charge injection from external circuit. Burroughes et al observed the injected charges from external circuit would distort the polymer chain with the addition charged soliton states which alter a wide absorption range of the polymer down to 0.5 eV 199 Figure 6 5 (color) The normalized electroabsorption spectra of p ristine PCDTBT under reverse (black) and positive (red) DC voltage biases V DC Based on the above postulation, we expect injecting holes from external circuit into pristine PCDTBT should also give rise to some additional EA features in the sub bandgap region. As shown in Fig 4, we probed the EA signal of a pristine P CD TBT film with a sub bandgap probing source fixed at 1000 nm (1.24 eV) and scanned the applied bias from 3 V to 5 V. Under reverse bias with the absence of PC 71 BM, there is no EA signal in the sub bandgap region due to either charge injection or optical transition With
127 a bias larger than the built in potential, h owever, the sub bandgap signal emerges and scale with the injection current through the PCDTBT film, as the threshold voltage (1.1 V) of the sub bandgap EA signal is in good agreement with the turn on voltage in the J V characteristics. Such a phenomenon was previously reported when voltage mod ulated transmission (VMT) and capacitance show very similar bias dependence 199 We also notice that, the quadrature signal turns to negative at the same 1.1 V, which is consistent with the earlier conclusion that the sub bandgap signal has a slow response to the modulation field. The EA spe ctra are shown in Fig. 5. Two different biases were used to illustrate the sub bandgap feature under reverse and forward biases. D ue to carrier injection at a bias of 3 V there is an additional band of EA signal ranging from 0.8 eV to 1.5 eV which is not found when the device reverse ly biased. Comparing the signal with the results in Fig. 1(c), the carrier injection induced signal has a much broadened distribution and extends to as low as 0.8 eV (Fig. 6 5 inset). Such a phenomenon is very close to the prev iously published PIA results because relaxed polarons should have a lower energy than those hot CT states. The positive or negative values of the signals are related to their different frequency response. In addition, we noticed that the signal between 1.5 eV and 1.7 eV, which was found in the charge transfer induced EA spectra (Fig. 5 1(c)), does n ot appear in Fig. 5. Because there s no CT states in the pristine PCDTBT, electrical pumping can only charge the molecule to singlet, triplet and some trap states, the signal of CT n is expected to be missing. These results confirm that both the carrier in jection in the pristine PCDTBT and the charge transfer in PCDTBT:PC 71 BM both give rise to subtle changes but occupies different region in the sub bandgap region of the absorption spectrum It is
128 worth to note in our approach to measure the sub bandgap res ponse in the PCDTBT:PC 71 BM blend, the device w as under reverse biased to eliminate charge injection. Thus the sub bandgap signal in Fig. 1(c) is only due to CT n modulation resulted from optical pumping Since t his minimum pumping energy marks the graduate transition from cooled CT states to hot CT states, we refer the CT induced EA onset as the effective bandgap of the polymer :fullerene BHJ cells. We propose the transition energy between cooled and hot CT states is more suitable for the definition o f effective bandgap E eff Because in efficient polymer:fullerene systems, like PCDTBT:PC 71 BM, a larger portion of excitons dissociate through such band states instead of localized cooled CT states 15 200 In contrast, the ground CT states, as usually observed in CT emission measurements is not the destination of polymer to fullerene charge transfer because additional energy driven hopping or additional IR push 19 is required to complete charge separation We call the technique as charge modulated electroabsorption spectroscopy (CMEAS) to distinguish it from the traditional EA spectra. The E eff of 1.4 eV of the PCDTBT:PC 71 BM blend measured by CMEA S is in slightly lower than the recently reported value, 1. 5 eV, measured by photoemission spectroscopy 184 We attribute the difference to the inability to detect charge transfer process using IPES. The effective optical bandgap is not the only information we can extract from the CMEAS spectra. It also provides insight into the exciton dissociation dynamics As we already known that the sub while the excitonic feature characterizes the optical gap of the polymer phase in the blend. The difference of these two values is thus a good approximation of the interfac e
129 effective force for the exciton dissociation. As obtained from Fig. 6 1 (c), the difference is found to be 0.3~0.4 eV a reasonable value to dissociate excitons 24 However, such minimum driving force seems to be unnecessary for the exciton dissociation through higher energy CT states. It can be seen the sub bandgap feature of PCDTBT:PC 71 BM is a continuous band ranging from 1.4 eV to the emergence of pristine PC 71 s excitonic response at 1.7 eV. Therefore all the excitons generated with an optical pumping energy between 1.4 eV and 1.8 eV can dissociate into separated carriers, suggesting a ca rrier splitting process requiring negligible repulsing forces. Applying the same analysis, the excitons generated in the fullerene phase have a maximum interface effective force ~ 0.2 eV. The delocalized electron wavefunction in the fullerene structure all ows efficient charge transfer from PC 71 BM to PCDTBT. T he CMEAS result also provides a good profile of the broadly distributed CT n state energies, these charge dissociation gateways distribute at least 0. 3 eV on the energy coordinate 6.4.2 Exploring the Excitonic Loss of Open Circuit Voltages To carry out a comprehensive study of V OC loss, we selected 7 photovoltaic polymers showing V OC values from 0.58 V to 0.9 0 V when blended with PC 71 BM. These V OC values are summarized from the best performance report since we want to exclude the V OC loss due to non use some very high V OC values achieved with the assistance of an equilibrium built in potential larger than the effective bandgap, because th ese large values are generated by compensating the extra built in potential due to large interfacial dipoles at electrodes. 120 121 Three of the seven polymers ( PCDTBT 201 PDTG TPD, and PDTS TPD 119 ) have been reported with PCEs over 7% and V OC larger than 0.85 V. Both PDTG BTD and P3HT show low V OC values around 0.6 V. P(il DTS) is a Isoindigo
130 dithienosilole based donor acceptor polymer s howing a medium V OC value (0.78 V) in the P(il DTS):PC 71 BM structure. 202 PTB7:PC 71 BM solar cell also ha s a medium V OC (0.75 V) if poly electrolyte is not used to increase the built in potential. 99 Figure 6 6 (color) Open circuit voltages plotted as a function of effective bandgap measured by CV, the data is from reference. The dashed line is an indicator of Eq. 6 1. Previous attempt to correlate the E eff and V OC of polymer PV cells showed a linear relationship with a voltage loss of 0.3 V as depicted in Eq. 6 1 160 Here, we plot the V OC versus the E eff obtained from cyclic voltammetry measurements (data summarized from references 28 32 160 202 203 ) for a series of d onor acceptor systems and the results are shown in Fig. 6. While the V OC is generally higher for BHJ systems with a larger E eff there is a large discrepancy to Eq. 6 1. For example, P3HT, PTB7 and P( il DTS) deviate from the empirical prediction by about 0.2 V on the V OC coordinate. The key issue is the energy levels measured in pristine materials ruling out the possible energy level shifting due to interface dipole in the BHJ systems. Also, the V OC lo ss does not seem to be a constant in different polymer:fullerene systems.
131 In the last section, we introduced a novel approach to determine the effective bandgap of polymer:fullerene heterojunctions. The sub bandgap signal in CMEAS signifies the onset of h ot CT band which is better definition of effective bandgap because it is an electrical gap characterizing the direct charge transfer. In addition to the CMEAS signals we showed in section 6.4.1, a series of spectra is also shown for pristine PDTG TPD and PDTG TPD:PC 71 BM blend. It can be seen in Figure 6 7, Figure 6 7 (color) CMEAS from a pristine PDTG TPD sample, a PDTG TPD:PC 71 BM blend and a linear combination of pristine PDTG TPD and PC 71 BM signals. All the spectra are re scaled using the same field condition 10 5 Vcm 1 the excitonic feature is red shifted a little by mixing PC 71 BM into PDTG TPD. Moreover, in the sub bandgap region, we observed an additional feature from 1.1 eV to 1.6 eV. This feature does not belong to any pristine material. The linear combination of PDTG TPD and PC 71 BM spectra (blue dashed line) is almost the same as the spectrum of pristine PDTG TPD. The sub bandgap signal show different electric field dependence (DC amplitude and modulation frequency ) fro m the excitonic signal 1.7 eV to 2.0 eV.
132 Therefore, the effective bandgap determined by the onset of hot CT band is 1.10 eV, 0.24 eV higher than the reported V OC values. Using the same experimental approach that carried out for PDTG TPD and PCDTBT we co ntinued the measurement on the other polymer:PC 71 BM BHJ systems. In order to reach the same V OC values in the references, the processing condition was optimized for each polymer:PC 71 BM system. The CMEAS measurements were then carried out in the same condit ion. The results in the sub bandgap range are shown in Figure 6 8. Figure 6 8 (color) Sub bandgap CMEAS signals of a series of photovoltaic polymers The onset of the sub bandgap feature is defined as the effective bandgap. In the sequence of onset energy of sub bandgap signals, seven BHJ systems show effective bandgaps from 0.85 eV to 1.40 eV. We also notice that, the E eff values
133 determined by CMEAS are quite different from those by electrochemical methods. PCDTBT:PC 71 BM sh ows a medium |E D HOMO E A LUMO | value (1.15 eV) in Figure 6 6, but it actually has the largest E eff (1.40 eV) and V OC (0.90 V) values among all the seven systems. P(il DTS) :PC 71 BM shows a |E D HOMO E A LUMO | as large as 1.25 eV, its E eff is in fact only 0.86 eV according to the spectrum. However, low V OC systems, such as PDTG BTD and P3HT generally show lower E eff values than the systems showing higher V OC s. These sub bandgap signals all show very slow response to the external modulation field, the positive or negative sub bangap signal values that are measured using the fixed modulation frequency (1K Hz) can be related to the different frequency response due to variant re organization energies in these systems. Plotting the V OC s as a funct ion of measured E eff s, we get the trend in Figure 6 9. Figure 6 9 (color) VOCs plotted as a function of measured Eeffs measured by CMEAS It can be seen that, although we get a much better trend showing V OC s increases as a function of measured E eff s, the voltage loss is definitely not a universal constant throughout the 7 samples. As a next step we sought to figure out the mechanisms that
134 causes materials dependent V OC loss which is expressed as According to Eq. 6 1, should be composed of a loss due to bimolecular recombination and a further energy loss during the charge separation process. Regarding both of these two mechanisms are highly related to the dielectric properties of the sam ple, we decided to probe as a function of dielectric constant which ca be determined by a measurement of geometric capacitance. The resu lts (Figure 6 10) show that these results are highly dependent on their dielectric constants. We can see that these samples show a decreased V OC loss with their increased dielectric constants, Figure 6 10 (color) plotted as a function of the inverse of dielectric constants. The dashed line is an indicator of eye. as shown by the dashed line. In particular, with the smallest dielectric constant among all the samples, PCDTBT:PC 71 BM shows a V OC larger than 0.5 V which is unreasonable if it is entirely due to bimolecular recombination. Since the carrier concentration at open circuit condition cannot drop to 10 1 3 cm 3 we suspect there s a contribution from the term E CS in Eq. 6 2. For the samples with a V OC loss around 0.3 V, E CS may still be
135 measurable because the E eff s obtained using CMEAS already consider the DOS disorder which makes lower than 0.3 eV. For the samples with a V OC loss lower than 0.2 V, the E CS values are expected to be negligible due to their large dielectric constants. Here we use d TPV to extract the carrier lifetime under open circuit conditions. The decay of the photo voltage perturbation follows the single exponential decay as the recombination is a result of low level excitation. The lifetime values can be further used to calcu late the carrier concentration using All the measurements are carried out under one sun condition. We summarized the results in Table 6 1. During the Table 6 1 Summary of carrier concentrations and recombination induced V OC loss Polymers r ( s) n (10 16 cm 3 ) kTln(np/N c N v ) E CS (eV) PCDTBT 3.36 0.24 2.05 1. 69 0.1 6 0.3 4 P3HT 4.20 0.20 8.11 4.7 4 0.1 0 0.1 9 PDTS TPD 4.17 0.22 0.64 0.67 0.2 0 0. 10 PDTG TPD 4.27 0.21 0.98 1. 20 0.1 7 0. 05 PDTG BTD 4.41 0.14 0.75 0.82 0. 19 0.0 8 PTB7 4.46 0.23 0.71 0.89 0. 19 0.0 4 P(il DTS) 4.95 0.14 1.50 1. 50 0.16 < 0.01 calculation of the number of available states at the band tail is estimated as following: N c =5 10 17 cm 3 N v =1 10 19 cm 3 Because the E eff values already subtracted the degree of disorder, we didn t use a DOS peak value (e.g. N v =1 10 20 cm 3 eV 1 ) in the calculation. The last two columns in Table 6 1 are plotted in 6 11 A clear trend of dielectric dependence can be seen as the energy loss during the charge separation process should be related to the weak but existing coulombic interaction of original geminate pairs. This result is the fi rst experimental demonstration of excitonic induced V OC loss. In particular, such an effect is very pronounced in the systems with
136 lower dielectric constants. In the systems with higher dielectric constants close to 5, the V OC loss is almost entirely due t o bimolecular recombination. Figure 6 11 (color) E CS (the V OC loss during the charge separation process) plotted as a function of the inverse of dielectric constants. 6.5 Summary In summary, we demonstrated a general approach to directly measure the effective bandgap, E eff of a polymer bulk heterojunction photovoltaic cell using charge modulated electroabsorption spectroscopy (CMEAS) Using CMEAS, we investigated the open circuit voltages of high efficiency BHJ devices based on several photovoltaic polymers and found that the open circuit voltage values of these devices have a non linear dependence on the effective bandgap energies. W e have also shown that the deviation from the no n linearity can be explained by taking into account the coulombic interaction during the charge separation stage This result is the first experimental demonstration of this V OC loss mechanism. Our experimental technique as well as the experimental results not only faci litate the detail study of energy level alignment at
137 organic heterojunction s but also provide insight to further improving the PV device performance.
138 CHAPTER 7 CONCLUSIONS AND FUTU RE WORK 7.1 Conclusions In this dissertation, electro nic processes of polymer:fullerene bulk heterojunction solar cells are discussed from the aspects of charge generation, transport, recombination, interfacial engineering and their impact on solar cell performance. Different experimental techniques were use d to probe the carrier mobility, lifetime and the energy alignment in the BHJ structure. The charge generation mechanism of polymer solar cells is in contrast with inorganic cells. Electrons and holes are generated in different materials as almost an inst antaneous response to the photo induced exciton formation. Since the charge generation and recombination both happens at the donor acceptor interface, carriers concentration is severely limited by the bimolecular recombination process. As shown in C hapter 6, we studied the carrier concentration in different polymer:fullerene systems under open circuit conditions. It is universal that bimolecular recombination directly causes a significant loss of quasi Fermi level splitting and thus V OC of the solar cell. T he charge generation process and the photovoltage output are also correlated by the following facts. The nature of excitonic cell determines that the carriers are affected by the coulombic interaction until the geminate pair is completely separated. We fi rst used CMEAS as an innovative tool to probe the effective bandgap of different polymer:fullerene BHJs. Then we extract the voltage loss due to recombination by carrying out TPV measurements. As another important result of C hapter 6, some BHJ systems with lowest dielectric constants show an extra V OC loss during the charge separation process. The CMEAS technique candidate for the definition of effective
139 bandgap for BHJ because it probes the onset of higher energy CT manifolds states). In the measurement, a sub bandgap pumping can directly generate hot CT states across the donor acceptor interfaces. These loosely bounded electron s and hole s can respond to the external oscillating field and result in subtle changes in absorption spectra. Carrier transport in the donor acceptor interpenetration network can be described by hopping mechanism. J V characteristics follow the trend of space charge limited current and trap charge limited current behaviors under different inte rnal fields. Correlated disorder model is a good approximation to explain the electric field and temperature dependent carrier mobility. Despite the low carrier mobility in organic materials, high FFs can still be achieved. From the perspective of transpor t, a balanced hole electron conduction is more important than simply pursuing higher hole mobilities in polymers. In C hapter 3, through a detailed photocurrent analysis, we showed that a severely imbalanced transport can redistribute the internal field dis tribution and cause space charge accumulation. Thus the output photocurrent is space charge limited with its value determined by the slow carrier mobility, charge generation rate and applied bias. Space charge limited photocurrent significantly reduces th e FF of polymer solar cells. A balanced transport usually gives rise to a much higher FF as well as charge extraction efficiency. In C hapter 4, we discuss the role of bimolecular recombination with the absence of charge imbalance issue. Our TPV measurement s revealed that recombination is very intensive for polymer:fullerene solar cells, especially for the systems with larger energetic disorder. We use the small polaron model to correlate the larger disorder and
140 larger recombination rate. This model is appli cable to explain the significant FF drop due to recombination limited photocurrent. In addition, a s shown by Ph CELIV results, the recombination rate is significantly reduced when the internal field increases to a corresponding bias of the maximum power po int. Thus, by experimentally comparing recombination process in BHJ systems with different degree of energetic disorder, we conclude recombination limited photocurrent should occur in the systems with larger energetic disorder despite the balanced electron hole transport. Charge collection and interfacial engineering is the topic we covered in C hapter 5. Transition metal oxide such as ZnO is good candidate to make Ohmic contact between fullerene and electrodes like TCO and aluminum. MoO 3 is a high work fun ction oxide to provide an optimum Fermi level alignment with the HOMO level of most photovoltaic polymers. The inverted structure using both ZnO and MoO 3 is a structure compatible with the future large scale printing process and requirement for better devi ce stability. Using a novel donor acceptor polymer PDTG TPD, we demonstrate a preliminary inverte d cell with an efficiency of 7. 3 %. As the second part of C hapter 5 we sought to reduce the defects concentration in the ZnO nanoparticle interlayer since these defects are proved to reduce the carrier lifetime at the ZnO/polymer interface. Through optimized UV ozone treatment condition, we demonstrated the defects reduc tion in ZnO nanoparticle layer as well as the improvement of solar cell performance. The inverted reached a lab measured efficiency of 8.1% and a certified efficiency of 7.4%, which was the world record for the inverted structure until this September The device with encapsulation still showed an efficiency of 7.2% after a storage time of 14 months.
141 7.2 Future Work 7.2.1 The Reduction of Energetic Disorder The work in C hapter 3 and 4 both deal with the photocurrent loss at low extraction field condition. I t is clear that a high mobility is not sufficient to generate efficient charge extraction. Recombination limited photocurrent occurs even without space charge accumulation. According to the small polaron model, the charge transfer process during bimolecula r recombination is easier to happen when there is larger energetic disorder. However, a direct correlation between the polymer structure and disorder may be more helpful for chemists because better materials can be created by lowering the degree of disorde r. The future work may first rely on the quantum chemistry calculation which can determine which structure generates smaller local potential fluctuation. The energetic disorder in newly synthesized materials can be characterized by mobility relaxation and temperature dependent zero field mobility. Carrier lifetime can be proved by TPV. The dispersive transport and recombination characteristics eventually determine the field dependence of drift length of photo carriers. 7.2.2 Exciton Migration in Donor Acceptor Polymers During the charge generation process of polymer:fullerene solar cells, photo generated exciton need to migrate to the donor acceptor interface for charge transfer. However, the process of exciton migration is less studied when compa ring with the charge transfer process. As a result, the motion of excitons are usually described as diffusion and assumed to be very similar across all the polymer materials. In fact, the migration process is not only different from a classical diffusion mechanism, but also directly affects the photovoltaic performance because the intermolecular exciton transfer
142 rate directly determines number of excitons that can hit the donor ac ceptor interface in a unit time which is a upper limit of charge generation rate for a polymer:fullerene solar cell. 204 Since all the molecules in the polymer domains are bonded with Van de Waals force, the exciton transfer can be modeled in th e approximation of the weak coupling limit and thus a Foster like approach. In this model, the energy transfer rate between an excited donating molecule and a ground state accepting molecule, k DA can be derived from the time dependent perturbation theory and the Fermi golden rule as: 205 ( 7 1) wherein H DA is the matrix element characterizing electronic coupling between donor and acceptor molecules; FCWD stands for Franck Condon weighted density of states factor, which is a product of the density of vibrational states (including 0 th state) in the initial (D* and A) and final states (D and A*) and their spectral overlap. Under the Condon approximation the electronic coupling is independent of nuclear arrangement, the energy transition rate can be derived as: 206 ( 7 2) wherein is the fluorescent emission spectrum of the donor and is the absorption spectrum of the acceptor. From Eq. 7 2, it is clear that both electronic coupling and the spectral overlap factor are critical to the exciton transfer rate. The electronic interaction is mainly determined by the dipole moment and spatial separation of the molecules. The spectral overlap factor is actually can be experimentally determined Some features in the emission and absorption spectra such as Stokes shift can be used to explain the different transition rate in different polymer systems.
143 In the assumed quadratic potential energy surface on one single normal mode of vibration, the lineshape of a vibration peak should f ollow the Poisson distribution: 207 208 where is the Franck Condon factor for normal mode n S is the Huang Rhys factor of the n th mode, defined as the ratio of nuclear reorganization energy and the vibrational energy. It is clear that a larger Huang Rhys factor gives a larger transition rate of that vibration mode. Also, in the simplified quadratic potential model, Huang Rhys factor is proportional to the Stokes shift of the polymer, which is defined the coordinate diff erence between the 0 0 transition in the emission and absorption spectra. 209 Therefore, through the measurement of fluorescence emission and absorption of certain pristine polymers and the analysis of the lineshape, we can build a correlation between the exciton transition rate and charge generation rate. Some polymers with the same donor but different acceptor units could be a good group of systems for the studies. Here, we have thr ee polymers: PDTG TPD, PDTG BTD and P(il DTG), with all the three sharing the same DTG unit as the donor Thus, their absorption lineshape are similar apart from the different weighted contribution from each vibrational peaks (Figure 7 1) As expected, the three polymers show very different Huang Rhys factors according to the Gaussian fitting of their absorpt ion spectra. On the other side when these polymers are individually blended with fullerene derivatives in BHJ solar cells, the external quantum effici encies show significant difference with PDTG TPD over 70%, PDTG BTD ~55% and P(il DTG) ~40%. As shown in Figure 7 2, these three polymer s also show Huang Rhys factor values in the same trend It is thus very interesting to correlate the charge transfer rate with the exciton migration rate by continuing the study with emission measurements and quantum
144 mechanical modeling. If the model is correct, a low Huang Rhys factor should be a thumb up rule for the de sign of efficient photovoltaic polymers. Figure 7 1 (color) The absorption spectra of PDTG TPD, PDTG BTD and P(il DTG) The measurements were carried out with solid state polymer films on quartz substrates. Figure 7 2 (color) The fitting of absorption spectra and Huang Rhys factor extracted by Gaussian fitting
145 APPENDIX A LIST OF MOLECULE STRUCTURES PC 61 BM PC 71 BM IC 60 BA IC 70 BA P3 HT PG1 PG2 PG3
146 PCDTBT PDTG BTD PDTG TPD PDTS BTD PDTS TPD PTB7 P(il DTS) PEDOT:PSS
147 APPENDIX B LIST OF PUBLICA TI ONS AND CONFERENCE PRESENTATIONS Peer Reviewed P ublications 1. Song Chen *, Sai Wing Tsang* (* equal contribution), Tzunghan Lai John R. Reynolds and Franky So, Open Circuit Voltage Loss in Excitonic Solar Cells (in preparation). 2. Sai Wing Tsang*, Song Chen *(* equal contribution) and Franky So, A Direct Measurement of Energy Alignment in Polymer:Fullerene Solar Cells (in preparation). 3. Cephas E. Small, Sai Wing Tsang, Song Chen Sujin Baek, Chad M. Amb, Thickness Dependence of Space Charge Effects in Low Bandgap Polymer Solar Cells 4. Jesse R. Manders, Sai Wing Tsang, Michael J. Hartel, Tzung han Lai, Song Chen Chad M. Amb, John R. Reynolds and Franky So, Solution Processed Nickel Oxide Hole Transport Layer in High Efficiency Polymer Photovoltaic Cells Advanced Functional Materials (submitted) 5. Song Chen Jesse M. Manders, Sai Metal Oxides for Interface Engineering in Polymer Solar Cells Journal of chemistr y materials 2 2, 24202 ( 2012 ) ( I nvited R eview ) 6. Chaoyu Xiang, Wanhoe Koo Song Chen Franky So Xiong Liu, Xiangxing Kong and Yunjun Wang Solution Processed Multilayer Cadmium Free Blue/violet Emitting Quantum Dot Light Emitting Diodes Applied Physics Letters 101, 053303 (2012)
148 7. Song Chen Cephas E. Small, Jegadesan Subbiah, Chad M. Amb, Sai Wing Tsang, Tzung Han Lai, Jesse M. Manders, John R. Reynolds and Franky So Inverted solar cells with reduced interface recombination Advanced Energy Materials 2, 1333 (2012) ( Journal Cover) 8. Song Chen Sai Wing Tsang, Cephas E. Small, John R. Reynolds and Franky Breakthroughs in Photonics: IEEE Photonics Journal 4, 625 (2012) 9. Cephas E. Small*, Song Chen *(* equal contribution) Jegadesan Subbiah, Chad M. Amb, Sai Wing Tsang, Tzung Han Lai, John R. Reynolds and Franky High efficiency inverted dithienogermole thienopyrrolodione based polymer solar cells Nature Photonics 5, 115 (2012) 10. Song Chen Kaushik Roy Choudhury, Jegadeson Subbiah, Chad M. Amb, John Photo Carrier Recombination in Polymer Solar Cells Based on P3HT and Silole Based Copolymer Advanced Energy Materials 1, 963 (2011) 11. Chad M. Amb, Song Chen Kenneth R Graham, Jegadeson Subbiah, Cephas Dithienogermole As a Fused Electron Donnor in Bulk Heterojunction Solar Cells Journal of American Chemistry Society 133, 10062 (2011) (Cited more than 100 times) 12. Kaushik Roy Choudhury, Jegadeson Subbiah, Song Chen Pierre M. Beaujuge, Understanding the performance and loss mechanisms in donor acceptor polymer based solar cells:
149 Photocurrent generation, charge separation and carrier transport Solar Energy Materials and Solar Cells 95, 2502 (2011) 13. Paul A. Lane, Song Chen Electromodulated doping of the hole transport layer in a small molecule light emitting diode Journal of Photonics for Energy 1, 011020 (2011) Or al Presentations in Conferences 1. Je s se R. Manders, Sai Wing Tsang, Song Chen Tzung Han Lai, Michael J. Hartel, Chad M. Amb, Kyukwan Zong, James J. Deininger, John R. Reynolds and Frank So, From Conception to Solar Cells: Solution Procssed Nickel Oxide Hole Transport Layers in High Efficnecy Organic Photovoltaics MRS 2012 spring meeting, San Francisco, J15.3 2. Sai Wing Tsang Cephas E. Small, Song Chen Jegadesan Subbiah, Chad M. Amb, Tzung Han Lai, John R. Reynolds and Franky So, 8% Power Conversion Efficiency Polymer Photovoltaic Cells with a Thick Layer of Dithienogemole Thienopyrrolodione:Fullerene MRS 2012 spring meeting, Boston V6.5 3. Cephas E. Small Song Chen Jegadesan Subbiah, Chad M. Amb, Sai Wing Tsang Tzung Han Lai, John R. Reynolds and F ranky So, High Efficiency Inverted Dithienogermole Thienopyrrolodione Based Polymer Solar Cells MRS 2012 spring meeting, Boston V6.8 4. Cephas E. Small Chad M. Amb, Song Chen Kenneth R. Grah am, Jegadeson Subbiah Dithienogermole As a Fused Electron Donnor in Bulk Heterojunction Solar Cells MRS 2011 fall meeting, Boston, E3 10
150 5. Song Chen Cephas E. Small, Chad M. Amb, Tzung Han Lai, Kenneth R. Achieving Higher than 8% Power Conversion Efficiency by Modified Electron Extraction Layer in Polymer Bulk Heterojunction Solar Cells MRS 2011 fall meeting, Boston, H8.8. 6. Song Chen Kaushik Roy Choudhury, Frederick P. Steffy, Jegadeson Subbiah, Chad M. Amb, John R. Reynolds and Frank Investigating the loss mechanism of photo current in low bandgap donor acceptor copolymer based solar cells: photo carriers recombination SPIE Optics and Photonics 2011, San Diego, 8116 51. 7. Song Chen Jegadesan Subbiah, Chad M. Amb, John R. Reynol ds and Franky Anode interlayer effect and photo current loss mechanism of copolymer: fullerene solar cells ICNP ) 2011, Shanghai, O11. 8. Song Chen Frederick P. Steffy, Kaushik Roy Choudhury, Chad Amb, Joh n R. Studying Charge Transport and Recombination in Organic Solar Cells Using Photo induced Carrier Extraction with Linearly Increasing Voltage MRS 2010 fall meeting, Boston, E7.9. 9. Kaushik Roy Choudhury Song Chen Jegadesan Subbiah, Pierre Beaujuge, Understanding the performance and loss mechanisms in bulk heterojunction solar cells based on a low bandgap green polymer: photocurrent generation, charge separation and carrier transport MRS 200 9 fall meeting, Boston, R12.7.
151 Poster Presentations in Conferences 1. Song Chen Cephas E. Small, Chad M. Amb, Tzung Han Lai, Kenneth R. Achieving Higher than 8% Power Conversion Efficiency by Modified Electron Extrac tion Layer in Polymer Bulk Heterojunction Solar Cells ICSM 2012, Atlanta, PI 24. 2. Song Chen Cephas E. Small, Jegadesan Subbiah, Chad M. Amb, Sai Wing Tsang, Tzung Han Lai, Jesse M. Manders, John R. Reynolds and Franky So Inverted solar cells with reduced interface recombination MRS 2012 spring meeting, San Francisco, Z 7.26 3. Jese R. Manders, Sai Wing Tsang, Song Chen Tzung Han Lai, Michael J. Hartel, Chad M. Amb, Kyukwan Zong, James J. Deininger, John R. Reynolds and Frank So Low Temperature So lution processed Nickel Oxide Hole Transort Layers in High Efficiency Organic Photovoltaics MRS 2012 spring meeting, San Francisco, W4.26. 4. Song Chen Jegadesan Subbiah, Chad M. Amb, John R. Reynolds and Franky Bimolecular Recombination and Energetic Disorder in Low gap Polymer MRS 2011 spring meeting, San Francisco, OO10.3. 5. Song Chen Kaushik Roy Choudhury Jegadesan Subbiah, Pierre M. Beaujuge, Charge Transport Study in Organic Semiconductors via Carrie r Extraction with Linearly Increasing Voltage MRS 2009 fall meeting, Boston, S8.1. 6. Song Chen Improving operating lifetime by enhancing hole injection in organic light emitting diodes S3.5.
15 2 LIST OF REFERENCES 1. M. A. Green, K. Emery, Y. Hishikawa, W. Warta and E. D. Dunlop, Prog. Photovolt. 2012, 20 606 614. 2. R. F. Service, Science 2011, 332 293 293. 3. N. S. Sariciftci, L. Smilowitz, A. J. Heeger and F. Wudl, Science 1992, 258 1474 1476. 4. R. A. Marcus, J. Chem. Phys. 1956, 24 966 978. 5. R. A. Marcus, J. Chem. Phys. 1957, 26 867 871. 6. R. A. Marcus and P. Siders, J. Phys. Chem. 1982, 86 622 630. 7. R. A. Marcus and N. Sutin, Biochimica Et Biophysica Act a 1985, 811 265 322. 8. Dogonadz.Rr, Kuznetso.Am and Vorotynt.Ma, Phys. Status Solidi B. 1972, 54 125 134. 9. Dogonadz.Rr, Kuznetso.Am and Vorotynt.Ma, Phys. Status Solidi B. 1972, 54 425 433. 10. P. Siders and R. A. Marcus, J. Am. Chem. Soc. 1981, 103 741 747. 11. G. Garcia Belmonte and J. Bisquert, Appl. Phys. Lett. 2010, 96 113301. 12. T. Kirchartz, B. E. Pieters, J. Kirkpatrick, U. Rau and J. Nelson, Phys. Rev. B 2011, 83 115209. 13. J. J. Benson Smith, L. Goris, K. Vandewal, K. H aenen, J. V. Manca, D. Vanderzande, D. D. C. Bradley and J. Nelson, Adv. Funct. Mater. 2007, 17 451 457. 14. F. Etzold, I. A. Howard, R. Mauer, M. Meister, T. D. Kim, K. S. Lee, N. S. Baek and F. Laquai, J. Am. Chem. Soc. 2011, 133 9469 9479. 15. B. S. Rolczynski, J. M. Szarko, H. J. Son, Y. Y. Liang, L. P. Yu and L. X. Chen, J. Am. Chem. Soc. 2012, 134 4142 4152. 16. V. I. Arkhipov, P. Heremans and H. Bassler, Appl. Phys. Lett. 2003, 82 4605 4607. 17. D. Veldman, S. C. J. Meskers and R. A. J Janssen, Adv. Funct. Mater. 2009, 19 1939 1948.
153 18. R. A. Marsh, J. M. Hodgkiss and R. H. Friend, Adv. Mater. 2010, 22 3672 +. 19. A. A. Bakulin, A. Rao, V. G. Pavelyev, P. H. M. van Loosdrecht, M. S. Pshenichnikov, D. Niedzialek, J. Cornil, D. Belj onne and R. H. Friend, Science 2012, 335 1340 1344. 20. K. Vandewal, K. Tvingstedt, J. V. Manca and O. Inganas, IEEE J. Sel. Top. Quant 2010, 16 1676 1684. 21. W. Shockley and H. J. Queisser, J. Appl. Phys. 1961, 32 510 &. 22. B. Carsten, J. M. Szarko, H. J. Son, W. Wang, L. Y. Lu, F. He, B. S. Rolczynski, S. J. Lou, L. X. Chen and L. P. Yu, J. Am. Chem. Soc. 2011, 133 20468 20475. 23. D. Spiegel and A. J. Heeger, Polym. Commun. 1988, 29 266 268. 24. C. W. Tang, Appl. Phys Lett. 1986, 48 183 185. 25. H. X. Zhou, L. Q. Yang and W. You, Macromolecules 2012, 45 607 632. 26. Z. G. Zhang and J. Z. Wang, J. Mater. Chem. 2012, 22 4178 4187. 27. C. H. Duan, F. Huang and Y. Cao, J. Mater. Chem. 2012, 22 10416 10434. 28. N. Blouin, A. Michaud and M. Leclerc, Adv. Mater. 2007, 19 2295 +. 29. Y. Y. Liang, Z. Xu, J. B. Xia, S. T. Tsai, Y. Wu, G. Li, C. Ray and L. P. Yu, Adv. Mater. 2010, 22 E135 +. 30. S. C. Price, A. C. Stuart, L. Q. Yang, H. X. Zhou and W. You, J. Am. Chem. Soc. 2011, 133 4625 4631. 31. T. Y. Chu, J. Lu, S. Beaupre, Y. Zhang, J. R. Pouliot, S. Wakim, J. Zhou, M. Leclerc, Z. Li, J. Ding and Y. Tao, J. Am. Chem. Soc. 2011, 133 4250. 32. C. M. Amb, S. Chen, K. R. Graham, J. Subbiah, C. E. Small, F. So and J. R. Reynolds, J. Am. Chem. Soc. 2011, 133 10062. 33. B. A. Gregg and M. C. Hanna, J. Appl. Phys. 2003, 93 3605 3614. 34. A. Miller and E. Abrahams, Phys. Rev. 1960, 120 745 755. 35. A. Pivrikas, G. Juska, A. J. Mozer, M. Scharber, K. Arlauskas, N. S. Sariciftci, H. Stubb and R. Osterbacka, Phys. Rev. Lett. 2005, 94 176806.
154 36. P. W. M. Blom, M. J. M. deJong and S. Breedijk, Appl. Phys. Lett. 1997, 71 930 932. 37. A. Foerti g, A. Baumann, D. Rauh, V. Dyakonov and C. Deibel, Appl. Phys. Lett. 2009, 95 052104. 38. B. Kramer and A. Mackinnon, Rep. Prog. Phys. 1993, 56 1469 1564. 39. P. W. Anderson, Phys. Rev. 1958, 109 1492 1505. 40. T. Tiedje, J. M. Cebulka, D. L. More l and B. Abeles, Phys. Rev. Lett. 1981, 46 1425 1428. 41. H. Bassler, Phys. Status Solidi B. 1993, 175 15 56. 42. N. F. Mott, J. Non Cryst. Solids 1968, 1 1 17. 43. R. H. Young, Philos. Mag. B 1995, 72 435 457. 44. M. Pope and C. E. Swenberg, Electronic processes in organic crystals and polymers Oxford University Press, New York, 1999. 45. V. Ambegaokar, B. I. Halperin and J. S. Langer, Phys. Rev. B:Solid St. 1971, 4 2612 +. 46. N. F. Mott, Adv. Phys. 1967, 16 49 &. 47. P. A. Walley, Th in Solid Films 1968, 2 327 &. 48. M. Morgan and P. A. Walley, Philos. Mag. 1971, 23 661 &. 49. N. F. Mott and R. W. Gurney, Oxford University Press, London, 1940. 50. P. E. Burrows, Z. Shen, V. Bulovic, D. M. McCarty, S. R. Forrest, J. A. Cronin and M. E. Thompson, J. Appl. Phys. 1996, 79 7991 8006. 51. M. A. Lampert, Phys. Rev. 1956, 103 1648 1656. 52. Murgatro.Pn, J. Phys. D: Appl. Phys. 1970, 3 151 &. 53. D. M. Pai, J. Chem. Phys. 1970, 52 2285 &. 54. S. V. Novikov, D. H. Dunlap, V. M. Kenkre, P. E. Parris and A. V. Vannikov, Phys. Rev. Lett. 1998, 81 4472 4475.
155 55. A. J. Mozer, G. Dennler, N. S. Sariciftci, M. Westerling, A. Pivrikas, R. Osterbacka and G. Juska, Phys. Rev. B 2005, 72 035217. 56. S. R. Cowan, A. Roy and A. J. Heeger, Phys. Rev. B 2010, 82 245207. 57. M. Lenes, M. Morana, C. J. Brabec and P. W. M. Blom, Adv. Funct. Mater. 2009, 19 1106 1111. 58. S. Chen, K. R. Choudhury, J. Subbiah, C. M. Amb, J. R. Reynolds and F. So, Adv. En ergy Mater. 2011, 1 963 969. 59. G. Yu, J. Gao, J. C. Hummelen, F. Wudl and A. J. Heeger, Science 1995, 270 1789 1791. 60. G. Li, V. Shrotriya, J. S. Huang, Y. Yao, T. Moriarty, K. Emery and Y. Yang, Nat. Mater. 2005, 4 864 868. 61. G. Li, Y. Yao, H. Yang, V. Shrotriya, G. Yang and Y. Yang, Adv. Funct. Mater. 2007, 17 1636 1644. 62. L. M. Chen, Z. R. Hong, G. Li and Y. Yang, Adv. Mater. 2009, 21 1434 1449. 63. K. R. Choudhury, J. W. Lee, N. Chopra, A. Gupta, X. Z. Jiang, F. Amy and F. So, Adv. Funct. Mater. 2009, 19 491 496. 64. G. Juska, K. Arlauskas, M. Viliunas and J. Kocka, Phys. Rev. Lett. 2000, 84 4946 4949. 65. G. Dennler, A. J. Mozer, G. Juska, A. Pivrikas, R. Osterbacka, A. Fuchsbauer and N. S. Sariciftci, Org. Electron. 200 6, 7 229 234. 66. J. Lorrmann, B. H. Badada, O. Inganas, V. Dyakonov and C. Deibel, J. Appl. Phys. 2010, 108 67. C. G. Shuttle, B. O'Regan, A. M. Ballantyne, J. Nelson, D. D. C. Bradley, J. de Mello and J. R. Durrant, Appl. Phys. Lett. 2008, 92 0933 11. 68. D. E. Aspnes, Surf. Sci. 1973, 37 418 442. 69. G. Weiser, Phys. Rev. B 1992, 45 14076 14085. 70. T. W. Hagler, K. Pakbaz and A. J. Heeger, Phys. Rev. B 1994, 49 10968 10975. 71. P. J. Brewer, P. A. Lane, A. J. deMello, D. D. C. Bradley and J. C. deMello, Adv. Funct. Mater. 2004, 14 562 570.
156 72. V. Bodrozic, T. M. Brown, S. Mian, D. Caruana, M. Roberts, N. Phillips, J. J. Halls, I. Grizzi, J. H. Burroughes and F. Cacialli, Adv. M ater. 2008, 20 2410 +. 73. T. M. Brown, J. S. Kim, R. H. Friend, F. Cacialli, R. Daik and W. J. Feast, Appl. Phys. Lett. 1999, 75 1679 1681. 74. D. E. Aspnes, Phys. Rev. 1966, 147 554 &. 75. C. D. Dimitrakopoulos and P. R. L. Malenfant, Adv. Mater. 2002, 14 99 +. 76. R. Coehoorn, W. F. Pasveer, P. A. Bobbert and M. A. J. Michels, Phys. Rev. B 2005, 72 155206. 77. C. Tanase, E. J. Meijer, P. W. M. Blom and D. M. de Leeuw, Phys. Rev. Lett. 2003, 91 216601. 78. Y. P. Zou, D. Gendron, R. Badrou Aich, A. Najari, Y. Tao and M. Leclerc, Macromolecules 2009, 42 2891 2894. 79. R. S. Ashraf, Z. Y. Chen, D. S. Leem, H. Bronstein, W. M. Zhang, B. Schroeder, Y. Geerts, J. Smith, S. Watkins, T. D. Anthopoulos, H. Sirringhaus, J. C. de Mello, M. H eeney and I. McCulloch, Chem. Mater. 2011, 23 768 770. 80. M. M. Mandoc, L. J. A. Koster and P. W. M. Blom, Appl. Phys. Lett. 2007, 90 133504. 81. C. Deibel, A. Wagenpfahl and V. Dyakonov, Physica Status Solidi Rapid Research Letters 2008, 2 175 17 7. 82. A. M. Goodman and A. Rose, J. Appl. Phys. 1971, 42 2823 &. 83. V. D. Mihailetchi, J. Wildeman and P. W. M. Blom, Phys. Rev. Lett. 2005, 94 126602. 84. M. Lenes, L. J. A. Koster, V. D. Mihailetchi and P. W. M. Blom, Appl. Phys. Lett. 2006, 88 243502 243503. 85. J. H. Hou, H. Y. Chen, S. Q. Zhang, R. I. Chen, Y. Yang, Y. Wu and G. Li, J. Am. Chem. Soc. 2009, 131 15586 +. 86. M. C. Scharber, M. Koppe, J. Gao, F. Cordella, M. A. Loi, P. Denk, M. Morana, H. J. Egelhaaf, K. Forberich, G. Dennl er, R. Gaudiana, D. Waller, Z. G. Zhu, X. B. Shi and C. J. Brabec, Adv. Mater. 2010, 22 367 +. 87. J. Subbiah, P. M. Beaujuge, K. R. Choudhury, S. Ellinger, J. R. Reynolds and F. So, ACS Appl. Mater. Interfaces 2009, 1 1154 1158.
157 88. V. D. Mihailetchi L. J. A. Koster, P. W. M. Blom, C. Melzer, B. de Boer, J. K. J. van Duren and R. A. J. Janssen, Adv. Funct. Mater. 2005, 15 795 801. 89. C. Melzer, E. J. Koop, V. D. Mihailetchi and P. W. M. Blom, Adv. Funct. Mater. 2004, 14 865 870. 90. P. W. M. B lom, V. D. Mihailetchi, L. J. A. Koster and D. E. Markov, Adv. Mater. 2007, 19 1551 1566. 91. P. M. Beaujuge, J. Subbiah, K. R. Choudhury, S. Ellinger, T. D. McCarley, F. So and J. R. Reynolds, Chem. Mater. 2010, 22 2093 2106. 92. K. Roy Choudhury, J Subbiah, S. Chen, P. M. Beaujuge, C. M. Amb, J. R. Reynolds and F. So, Sol. Energy Mater. Sol. Cells 2011, 95 2502. 93. L. J. A. Koster, E. C. P. Smits, V. D. Mihailetchi and P. W. M. Blom, Phys. Rev. B 2005, 72 085205. 94. R. Sokel and R. C. Hughes, J. Appl. Phys. 1982, 53 7414 7424. 95. W. L. Ma, C. Y. Yang, X. Gong, K. Lee and A. J. Heeger, Adv. Funct. Mater. 2005, 15 1617 1622. 96. V. D. Mihailetchi, H. X. Xie, B. de Boer, L. J. A. Koster and P. W. M. Blom, Adv. Fun ct. Mater. 2006, 16 699 708. 97. D. Muhlbacher, M. Scharber, M. Morana, Z. G. Zhu, D. Waller, R. Gaudiana and C. Brabec, Adv. Mater. 2006, 18 2884 +. 98. S. H. Park, A. Roy, S. Beaupre, S. Cho, N. Coates, J. S. Moon, D. Moses, M. Leclerc, K. Lee and A. J. Heeger, Nat. Photon. 2009, 3 297 U295. 99. H. Y. Chen, J. H. Hou, S. Q. Zhang, Y. Y. Liang, G. W. Yang, Y. Yang, L. P. Yu, Y. Wu and G. Li, Nat. Photon 2009, 3 649 653. 100. J. Subbiah, C. M. Amb, I. Irfan, Y. Gao, J. R. Reynolds and F. So, Sol. Energy Mater. Sol. Cells 2012, 97 97. 101. C. G. Shuttle, B. O'Regan, A. M. Ballantyne, J. Nelson, D. D. C. Bradley and J. R. Durrant, Phys. Rev. B 2008, 78 113201. 102. R. Hamilton, C. G. Shuttle, B. O'Regan, T. C. Hammant, J. Nelson and J. R. Durrant, Journal of Physical Chemistry Letters 2010, 1 1432 1436. 103. I. I. Fishchuk, V. I. Arkhipov, A. Kadashchuk, P. Heremans and H. Bassler, Phys. Rev. B 2007, 76 045210.
158 104. J. Subbiah, P. M. Beaujuge, K. R. Choudhury, S. Ellinger, J. R. Reynolds and F. So, Org. Electron. 2010, 11 955 958. 105. B. C. O'Regan and J. R. Durrant, J. Phys. Chem. B 2006, 110 8544 8547. 106. B. C. O'Regan, S. Scully, A C. Mayer, E. Palomares and J. Durrant, J. Phys. Chem. B 2005, 109 4616 4623. 107. F. L. Zhang, W. Mammo, L. M. Andersson, S. Admassie, M. R. Andersson, L. Inganas, S. Admassie, M. R. Andersson and O. Ingands, Adv. Mater. 2006, 18 2169 +. 108. A. J. Mozer, N. S. Sariciftci, L. Lutsen, D. Vanderzande, R. Osterbacka, M. Westerling and G. Juska, Appl. Phys. Lett. 2005, 86 112104. 109. A. Liu, S. Zhao, S. B. Rim, J. Wu, M. Konemann, P. Erk and P. Peumans, Adv. Mater. 2008, 20 1065 +. 110. V. D. Mih ailetchi, L. J. A. Koster, J. C. Hummelen and P. W. M. Blom, Phys. Rev. Lett. 2004, 93 216601. 111. K. Maturova, S. S. van Bavel, M. M. Wienk, R. A. J. Janssen and M. Kemerink, Nano Letters 2009, 9 3032 3037. 112. C. G. Shuttle, A. Maurano, R. Hamilton, B. O'Regan, J. C. de Mello and J. R. Durrant, Appl. Phys. Lett. 2008, 93 183501. 113. H. Bassler, G. Schonherr, M. Abkowitz and D. M. Pai, Phys. Rev. B 1982, 26 3105 3113. 114. G. Juska, K. Arlauskas, M. Viliunas, K. Genevicius, R. Osterbacka and H. Stubb, Phys. Rev. B 2000, 62 R16235 R16238. 115. S. E. Shaheen, C. J. Brabec, N. S. Sariciftci, F. Padinger, T. Fromherz and J. C. Hummelen, Appl. Phys. Lett. 2001, 78 841 843. 116. M. Granstrom, K. Petritsch, A. C. Arias, A. Lux, M. R. Andersson and R. H. Friend, Nature 1998, 395 257 260. 117. J. Y. Kim, S. H. Kim, H. H. Lee, K. Lee, W. L. Ma, X. Gong and A. J. Heeger, Adv. Mater. 2006, 18 572 +. 118. G. Li, C. W. Chu, V. Shrotriya, J. Huang and Y. Yang, Appl. Phys. Lett. 2006, 88 253503.
159 119. S. Chen, C. E. Small, C. M. Amb, J. Subbiah, T. H. Lai, S. W. Tsang, J. R. Reynolds and F. So, Adv. Energy Mater. 2012, 2 1333 1337 120. Z. C. He, C. M. Zhong, X Huang, W. Y. Wong, H. B. Wu, L. W. Chen, S. J. Su and Y. Cao, Adv. Mater. 2011, 23 4636 +. 121. Z. He, C. Zhong, S. Su, M. Xu, H. Wu and Y. Cao, Nat Photon 2012, 6 593 597. 122. L. T. Dou, J. B. You, J. Yang, C. C. Chen, Y. J. He, S. Murase, T. Mor iarty, K. Emery, G. Li and Y. Yang, Nat. Photon. 2012, 6 180 185. 123. F. C. Krebs, Sol. Energy Mater. Sol. Cells 2009, 93 1636 1641. 124. F. C. Krebs, Sol. Energy Mater. Sol. Cells 2009, 93 465 475. 125. H. Hoppe and N. S. Sariciftci, J. Mater. Chem. 2006, 16 45 61. 126. Y. Park, V. Choong, Y. Gao, B. R. Hsieh and C. W. Tang, Appl. Phys. Lett. 1996, 68 2699 2701. 127. K. Sugiyama, H. Ishii, Y. Ouchi and K. Seki, J. Appl. Phys. 2000, 87 295 298. 128. H. Ishii, K. Sugiyama, D. Yoshimura, E Ito, Y. Ouchi and K. Seki, IEEE J. Sel. Top. Quant 1998, 4 24 33. 129. J. Hwang, A. Wan and A. Kahn, Mater. Sci. Eng. R. Rep. 2009, 64 1 31. 130. I. G. Hill, D. Milliron, J. Schwartz and A. Kahn, Appl. Surf. Sci. 2000, 166 354 362. 131. M. A. Ba ldo and S. R. Forrest, Phys. Rev. B 2001, 64 132. B. N. Limketkai and M. A. Baldo, Phys. Rev. B 2005, 71 133. C. W. Tang and S. A. Vanslyke, Appl. Phys. Lett. 1987, 51 913 915. 134. E. I. Haskal, A. Curioni, P. F. Seidler and W. Andreoni, Appl. Phys. Lett. 1997, 71 1151 1153. 135. L. S. Hung, C. W. Tang and M. G. Mason, Appl. Phys. Lett. 1997, 70 152 154. 136. C. I. Wu, C. T. Lin, Y. H. Chen, M. H. Chen, Y. J. Lu and C. C. Wu, Appl. Phys. Lett. 2006, 88 152104. 137. G. Gustafsson, Y. Cao G. M. Treacy, F. Klavetter, N. Colaneri and A. J. Heeger, Nature 1992, 357 477 479.
160 138. G. Heywang and F. Jonas, Adv. Mater. 1992, 4 116 118. 139. Y. Yang and A. J. Heeger, Appl. Phys. Lett. 1994, 64 1245 1247. 140. S. A. Carter, M. Angelopoulos, S. Karg, P. J. Brock and J. C. Scott, Appl. Phys. Lett. 1997, 70 2067 2069. 141. J. H. Seo, A. Gutacker, Y. M. Sun, H. B. Wu, F. Huang, Y. Cao, U. Scherf, A. J. Heeger and G. C. Bazan, J. Am. Chem. Soc. 2011, 133 8416 84 19. 142. S. Chen, J. R. Manders, S. W. Tsang and F. So, J. Mater. Chem. 2012, 2 2 24202 24212 143. H. Moormann, D. Kohl and G. Heiland, Surf. Sci. 1980, 100 302 314. 144. D. W. Bahnemann, C. Kormann and M. R. Hoffmann, J. Phys. Chem. 1987, 91 3789 3798. 145. C. Pacholski, A. Kornowski and H. Weller, Angew. Chem. Int. Ed. 2002, 41 1188 +. 146. W. J. E. Beek, M. M. Wienk, M. Kemerink, X. N. Yang and R. A. J. Janssen, J. Phys. Chem. B 2005, 109 9505 9516. 147. T. Y. Chu, S. W. Tsang, J. Y. Zhou, P. G. Verly, J. P. Lu, S. Beaupre, M. Leclerc and Y. Tao, Sol. Energy Mater. Sol. Cells 2012, 96 155 159. 148. L. Qian, Y. Zheng, J. G. Xue and P. H. Holloway, Nat. Photon. 2011, 5 543 548. 149. S. Tokito, K. Noda and Y. Taga, J. Phys. D: A ppl. Phys. 1996, 29 2750 2753. 150. D. Y. Kim, J. Subbiah, G. Sarasqueta, F. So, H. J. Ding, Irfan and Y. L. Gao, Appl. Phys. Lett. 2009, 95 093304. 151. T. S. Sian and G. B. Reddy, Sol. Energy Mater. Sol. Cells 2004, 82 375 386. 152. K. Kanai, K. Koizumi, S. Ouchi, Y. Tsukamoto, K. Sakanoue, Y. Ouchi and K. Seki, Org. Electron. 2010, 11 188 194. 153. M. Kroger, S. Hamwi, J. Meyer, T. Riedl, W. Kowalsky and A. Kahn, Org. Electron. 2009, 10 932 938. 154. J. Meyer, A. Shu, M. Kroger and A. Kahn, Appl. Phys. Lett. 2010, 96 133308. 155. Irfan, H. J. Ding, Y. L. Gao, C. Small, D. Y. Kim, J. Subbiah and F. So, Appl. Phys. Lett. 2010, 96 243307.
161 156. J. Meyer, R. Khalandovsky, P. Gorrn and A. Kahn, Adv. Mat er. 2011, 23 70 +. 157. M. T. Greiner, M. G. Helander, Z. B. Wang, W. M. Tang, J. Qiu and Z. H. Lu, Appl. Phys. Lett. 2010, 96 213302. 158. M. Kroger, S. Hamwi, J. Meyer, T. Riedl, W. Kowalsky and A. Kahn, Appl. Phys. Lett. 2009, 95 123301. 159. M T. Greiner, M. G. Helander, W. M. Tang, Z. B. Wang, J. Qiu and Z. H. Lu, Nat. Mater. 2012, 11 76 81. 160. M. C. Scharber, D. Wuhlbacher, M. Koppe, P. Denk, C. Waldauf, A. J. Heeger and C. L. Brabec, Adv. Mater. 2006, 18 789 +. 161. C. E. Small, S. W. Tsang, J. Kido, S. K. So and F. So, Adv. Funct. Mater. 2012, 22 162. S. K. Hau, H. L. Yip, H. Ma and A. K. Y. Jen, Appl. Phys. Lett. 2008, 93 233304. 163. L. Qian, Y. Zheng, K. R. Choudhury, D. Bera, F. So, J. G. Xue and P. H. Holloway, Nano Today 2010, 5 384 389. 164. D. Bera, L. Qian, S. Sabui, S. Santra and P. H. Holloway, Optical Materials 2008, 30 1233 1239. 165. B. C. O'Regan and F. Lenzmann, J. Phys. Chem. B 2004, 108 4342 4350. 166. S. D. Oosterhout, M. M. Wienk, S. S. van Bavel, R. Thiedmann, L. J. A. Koster, J. Gilot, J. Loos, V. Schmidt and R. A. J. Janssen, Nat. Mater. 2009, 8 818 824. 167. V. A. Fonoberov, K. A. Alim, A. A. Balandin, F. X. Xiu and J. L. Liu, Phys. Rev. B 2006 73 165317. 168. S. Lany and A. Zunger, J. Appl. Phys. 2006, 100 169. F. J. Haug, D. Rudmann, H. Zogg and A. N. Tiwari, Thin Solid Films 2003, 431 431 435. 170. M. R. Balboul, A. Jasenek, O. Chernykh, U. Rau and H. W. Schock, Thin Solid Films 2001, 387 74 76. 171. V. Ischenko, S. Polarz, D. Grote, V. Stavarache, K. Fink and M. Driess, Adv. Funct. Mater. 2005, 15 1945 1954. 172. Y. W. Heo, D. P. Norton and S. J. Pearton, J. Appl. Phys. 2005, 98
162 173. A. Janotti and C. G. Van de Walle, Phys. Rev. B 2007, 76 165202. 174. U. Malm, J. Malmstrom, C. Platzer Bjorkman and L. Stolt, Thin Solid Films 2005, 480 208 212. 175. J. K. Lee, W. L. Ma, C. J. Brabec, J. Yuen, J. S. Moon, J. Y. Kim, K. Lee, G. C. Bazan and A. J. Heeger, J. Am. Chem. Soc. 2008, 130 3619 3623. 176. P. A. Lane, S. Chen and F. So, Journal of Photonics for Energy 2011, 1 011020. 177. D. C. Olson, Y. J. Lee, M. S. White, N. Kopidakis, S. E. Shaheen, D. S. Ginley, J. A. Voigt and J. W. P. Hsu, J. Phys. Chem. C 2008, 112 9544 9547. 178. C. E. Small, S. Chen, J. Subbiah, C. M. Amb, S. W. Tsang, T. H. Lai, J. R. Reynolds and F. So, Nat. Photon. 2012, 6 115 120. 179. B. P. Rand, D. P. Burk and S. R. Forrest, Phys. Rev. B 2007, 75 115327. 180. D. Veldman, O. Ipek, S. C. J. Meskers, J. Sweelssen, M. M. Koetse, S. C. Veenstra, J. M. Kroon, S. S. van Bavel, J. Loos and R. A. J. Janssen, J. Am. Chem. Soc. 2008, 130 7721 7735. 181. A. Cravino, Appl. Phys. Lett. 2007, 91 243502. 182. K. Vandewal K. Tvingstedt, A. Gadisa, O. Inganas and J. V. Manca, Nat. Mater. 2009, 8 904 909. 183. Z. L. Guan, J. B. Kim, H. Wang, C. Jaye, D. A. Fischer, Y. L. Loo and A. Kahn, Org. Electron. 2010, 11 1779 1785. 184. E. L. Ratcliff, J. Meyer, K. X. Steirer, N. R. Armstrong, D. Olson and A. Kahn, Org. Electron. 2012, 13 744 749. 185. K. Tvingstedt, K. Vandewal, A. Gadisa, F. L. Zhang, J. Manca and O. Inganas, J. Am. Chem. Soc. 2009, 131 11819 11824. 186. W. Gong, M. A. Faist, N. J. Ekins Daukes, Z. Xu, D. D. C. Bradley, J. Nelson and T. Kirchartz, Phys. Rev. B 2012, 86 024201. 187. U. Rau, Phys. Rev. B 2007, 76 085303. 188. J. Nelson, J. Kirkpatrick and P. Ravirajan, Phys. Rev. B 2004, 69 035337. 189. M. G ruber, J. Wagner, K. Klein, U. Hrmann, A. Opitz, M. Stutzmann and W. Brtting, Adv. Energy Mater. 2012, 2 1100 1108.
163 190. G. Garcia Belmonte, P. P. Boix, J. Bisquert, M. Sessolo and H. J. Bolink, Sol. Energy Mater. Sol. Cells 2010, 94 366 375. 191. K K. H. Chan, S. W. Tsang, H. K. H. Lee, F. So and S. K. So, Org. Electron. 2012, 13 850 855. 192. R. Klenk, Thin Solid Films 2001, 387 135 140. 193. A. Horvath, H. Bassler and G. Weiser, Phys. Status Solidi B. 1992, 173 755 764. 194. Z. Shen, P. E. Burrows, S. R. Forrest, M. Ziari and W. H. Steier, Chem. Phys. Lett. 1995, 236 129 134. 195. A. Horvath, G. Weiser, G. L. Baker and S. Etemad, Phys. Rev. B 1995, 51 2751 2758. 196. M. H. Tong, N. E. Coates, D. Moses, A. J. Heeger, S. B eaupre and M. Leclerc, Phys. Rev. B 2010, 81 125210. 197. I. H. Campbell, T. W. Hagler, D. L. Smith and J. P. Ferraris, Phys. Rev. Lett. 1996, 76 1900 1903. 198. A. A. Bakulin, D. Martyanov, D. Y. Paraschuk, P. H. M. v. Loosdrecht and M. S. Pshenichn ikov, Chem. Phys. Lett. 2009, 482 99 104. 199. J. H. Burroughes, C. A. Jones and R. H. Friend, Nature 1988, 335 137 141. 200. F. Etzold, I. A. Howard, N. Forler, D. M. Cho, M. Meister, H. Mangold, J. Shu, M. R. Hansen, K. Mullen and F. Laquai, J. Am. Chem. Soc. 2012, 134 10569 10583. 201. Y. M. Sun, C. J. Takacs, S. R. Cowan, J. H. Seo, X. Gong, A. Roy and A. J. Heeger, Adv. Mater. 2011, 23 2226 +. 202. R. Stalder, C. Grand, J. Subbiah, F. So and J. R. Reynolds, Polym Chem Uk 2012, 3 89 92. 2 03. D. Gendron, P. O. Morin, P. Berrouard, N. Allard, B. R. Aich, C. N. Garon, Y. Tao and M. Leclerc, Macromolecules 2011, 44 7188 7193. 204. S. Westenhoff, I. A. Howard and R. H. Friend, Phys. Rev. Lett. 2008, 101 016102. 205. V. May and O. Kuhn, eds., Charge and Energy Transfer Dynamics in Molecular Systems Wiley, Berlin, 2000.
164 206. E. Hennebicq, G. Pourtois, G. D. Scholes, L. M. Herz, D. M. Russell, C. Silva, S. Setayesh, A. C. Grimsdale, K. Mullen, J. L. Bredas and D. Beljon ne, J. Am. Chem. Soc. 2005, 127 4744 4762. 207. M. Lax, J. Chem. Phys. 1952, 20 1752 1760. 208. T. H. Keil, Phys. Rev. 1965, 140 A601. 209. J. L. Bredas, J. Cornil and A. J. Heeger, Adv. Mater. 1996, 8 447.
165 BIOGRAPHICAL SKETCH Song Chen was born in Suzhou, a beautiful city in eastern China in 1985. In 2004, he finished his high school study in Suzhou Middle School and entered one of the top universit ies in China Nanjing University (NJU) for his college study. In 2008, he graduated with a Bachelor of Science degree from the Department of Mat erials Science and Engineering with an honor of Dean Scholarship in 2007 due to his outstanding academic performance. In the fall of the same year, he continue d his graduate study in Uni ted States. Under the supervision of Prof. Franky So, he chose the topic of device physics of polymer solar cell s During the 4 year PhD career, he studied carrier transport, bimolecular recombination, interfacial engineering and energy alignment of polyme r solar cells. M ore than 10 peer reviewed papers were published on high impact journals such as Nature Photonics, Journal of American Chemical Society and Advanced Energy Materials. 10 presentations including 5 oral presentations were given in the confer ences of Material Research Society (MRS), SPIE and International Conference on Science and Technology of Synthetic Metals (ICSM). In December 2012, he received his degree of Doctor of Philosophy in the Department of Materials Science and Engineering at the University of Florida.