UNIVERSITY OF CINCINNATI

Transcription

1 UNIVERSITY OF CINCINNATI Date: I,, hereby submit this work as part of the requirements for the degree of: in: It is entitled: This work and its defense approved by: Chair:

2 Optical Properties of Organic Nanostructures Grown By Organic Molecular Beam Deposition A dissertation submitted to the Division of Research and Advanced Studies of the University of Cincinnati in partial fulfillment of the requirements for the degree of DOCTORATE OF PHILOSOPHY (PH.D.) in the Department of Physics of the College of Arts and Sciences 2006 by Landewatte A. Ajith De Silva B.S., University of Ruhuna, Sri Lanka, 1996 M.S., University of Cincinnati, 2001 Committee Chair: Professor Hans-Peter Wagner i

3 Abstract In this work, optical properties of organic nanostructures are investigated. In temperature dependent ( K) absorption studies of PTCDA and PTCDA/Alq 3 multilayers a red shift and a line narrowing of vibrational Frenkel exciton bands is observed when the temperature is decreased. The reduced transition energy is explained by a thermal contraction along the PTCDA molecular stacks that cause an increased inter-molecular overlap between PTCDA molecules leading to an enhanced environmental shift. The reduction of the inhomogeneous broadening of the bands is explained by a reduced population of internal and external vibrational levels of the electronic ground state. The reduced temperature shift in multilayer is attributed to a reduced thermal contraction in the PTCDA crystallites due to adjacent Alq 3 interlayers that possess a smaller thermal contraction than PTCDA. The exciton emission in PTCDA thin films, PTCDA/Alq 3 multilayers and co-deposited PTCDA/Alq 3 layers is studied by temperature dependent ( K) PL measurements. The different recombination channels arising from Frenkel excitons and self-trapped excitons (charge-transfer excitons CT1-nr, CT1, CT2 and excimers) that were observed earlier in PTCDA single crystals also appear in films, multilayers and in co-deposited layers. In PTCDA/Alq 3 multilayers, an unknown low energy line dominates the emission spectrum up to 200 K. In accordance to investigations using X-ray diffraction, FTIR absorption and strain dependent PL measurements the new channel is attributed to a modified CT2 transition due the compressive strain between stacked molecules. Temperature dependent ( K) PL measurements of Alq 3 layers are performed. An exciton trapping model which includes the formation of self-trapped excitons is proposed to explain the observed temperature dependent PL intensity enhancement and the spectral red-shift ii

4 of the PL spectrum (at ~ 180 K). Alq 3 based OLED structures are fabricated and electro-optical measurements are performed. The I-V measurements reveal a trap charge limited current behavior. The EL efficiency of the device shows similar temperature dependence as the PL intensity obtained from the Alq 3 film. Furthermore, both the EL and PL spectra reveal a maximum redshift at 180 K which is tentatively attributed to the formation of self-trapped excitons within the Alq 3 layer. iii

5 iv

6 This thesis is dedicated to my parents my wife Harsha who sacrificed so many things for me. v

7 Acknowledgements It is very difficult for me to find the right expressions of gratitude to thank those who helped me in numerous ways to complete this work. Many opportunities and experiences that I enjoyed were made possible by the support and the efforts of others. The intensive supervision of Professor Hans-Peter Wagner through the course of this thesis led to several publications and the work presented here. His underlying support through the years, academically, professionally, as well as personally, has made it possible for me to achieve my goal. I wish to thank him for giving me the opportunity to work with him. Professor Hans-Peter Wagner was always available for support, encouragement, and stimulating discussions. It has been a pleasure to work under his guidance, and I am indebted to him for his patient help throughout the course of my research work. His professionalism, guidance, energy, humor, thoroughness, dedication and inspiration categorize him as an example of the perfect supervisor. I am so thankful to Professors Rohana Wijewardhana, Leigh Smith and Andrew Steckl for serving in my thesis committee. Professor Wijewardhana will always be remembered a person who extended personal and educational advice on many occasions in addition to being a wonderful teacher. Professor Steckl has been so kind in giving me the opportunity to share UC Nanolab expertise for OLED fabrication and characterization. There are so many people I should thank, however it is very difficult to mention each and every name, but some names stand out naturally. I want to extend special thanks to John Markus, the instrumentation specialist at the department of Physics, for extending me help at many important occasions and for being such good friends. Many thanks to faculty and staff in the Department of Physics, especially to Melody Whitlock (former graduate secretary in Physics), vi

8 Donna Deutenberg, Elle Mengon, Bob Schrott, Mark Ankenbauer, Mark Sabatelli, and John Whitaker. I thank my lab mates, Suvranta Tripathy, Pradep Bajracharya, and Venkat Gangilenka for sharing their time, expertise, and humor with me. A big thank to Robert Jones and Wayne Li in UC Nanolab and to Dr. Klotzkin for all the help with OLED characterization. A special acknowledgement is extended to the following senior research scientists for their contributions: Dr. Thorsten Kampen (Fritz-Haber-Institiut Germany - OMBD), Dr. Heidrun Schmitzer (Xavier University - PL Setup), Dr. Young Kim (UC - FTIR), Dr. Warren D Huff (UC - XRD), Dr. Bernard Weinstein (SUNY Buffalo - pressure measurements), Dr. Andrei Kobitski (University of Ulm, Germany - TDPL) and Dr. Reinhard Scholz (TU Chemnitz, Germany - DFT). Their involvement in my thesis project added completeness of interpretations and results which led to many publications. I would like to extend special thanks to my long time friend, Theja, he is one of the reasons that I came to Cincinnati, also to Prasanna, Luminda, Sumith, Nimal and other Sri Lankan friends for their valuable support which has made my stay at Cincinnati fun and more enjoyable. Words cannot truly express my deepest gratitude and appreciation to my parents, who always gave me their love, blessings, and emotional support. I am forever indebted to my parents and amazed at their generosity. I am also indebted to my bothers and sisters, for emotional support and for looking after my parents back home during my absence. At last but not least, I express my admiration and heartfelt gratitude to my wonderful wife Harsha for coming together with me to the USA, even leaving behind her medical carrier at some points for me and for all the support and love and understanding. Further, thanks for the vii

9 countless sacrifices she has made over the past one and a half years being alone in New York City so that I could achieve my dream. viii

10 Table of Contents Table of Contents 1 List of Figures 5 List of Tables 20 List of Abbreviations 21 1 Introduction Outline of the Thesis 29 2 Absorption and Emission Studies in PTCDA Crystals, Films and PTCDA/Alq 3 Multilayers Sample Fabrication Organic Materials PTCDA Alq Sublimation of the Materials Organic Molecular Beam Deposition and Thin Film Growth Morphology Studies on the Films X-ray Diffraction Fourier Transform Infrared Spectroscopy Exciton Absorption and Emission Process in Organic Molecules and Crystals: Introduction to Theory 48 1

11 2.2.1 Optical Excitation of Isolated Molecules Frenkel Excitons in Organic Crystals Charge Transfer Excitons and Excimer Transitions Experimental Setups Transmission Measurements Photoluminescence Measurements Strain Dependent Measurements Experimental Studies Temperature Dependent Absorption of PTCDA Films Temperature Dependent Photoluminescence PTCDA Thin Films PTCDA/Alq 3 Multilayers PTCDA Crystals, Thin Films and PTCDA/Alq 3 Monolayers PTCDA/Alq 3 Co-deposited Layers Strain Dependence of PTCDA Crystals Strain Dependence of PTCDA Thin Films Characterization of Alq 3 Layers and Alq 3 Based Light Emitting Diodes Fabrication of Organic Light Diodes ITO Substrate and Patterning Organic Materials Organic Layers and Metal Deposition

12 3.2 Electro-Optical Properties of OLEDs Charge Carrier Injection and Transport in OLEDs Energy Level Diagram in OLED Structures The Efficiency of OLEDs Experimental Setups Current-Voltage-Luminescence Measurements Electroluminescence Measurements Experimental Studies Absorption and PL of Alq 3 Single Molecules, Polycrystals and Thin Films Temperature Dependent PL of Alq 3 Thin Films Room Temperature Studies on Alq 3 Based OLEDs Structures Temperature Dependent Current-Voltage Measurements of OLEDs Temperature Dependent Electroluminescence of Alq 3 Based OLEDs Summary and Conclusions Investigations on PTCDA Crystal, Thin Films and PTDA/Alq 3 Layer Structures Temperature Dependent Absorption Temperature Dependent Photoluminescence Strain Dependent Photoluminescence Characterization of Alq 3 Layers and Alq 3 Based OLED Structures 161 3

13 Appendix A 163 Appendix B 164 Bibliography 165 4

14 List of Figures 2.1 The chemical structure of a PTCDA molecule. The perylene derivative consists of four carboxylic groups at each corner and two anhydride groups at both sides. The length of the molecule is 14.2 Å and the width is 9.2 Å Unit cells of α-ptcda (a) and β-ptcda (b), the molecular plane coincides with the (102) plane of the crystal 38. a, b and c represent unit cell parameters while β is the angle between a and c unit vectors Crystal unit cells of two monoclinic phases of PTCDA, in the plane most closely coinciding with the molecular orientation, where c corresponds to the direction of (c-2a) Chemical structure of Tris (8-hydroxyquinoline) aluminum (III) (Alq 3 ) Geometrical molecular structures for meridinal and facial isomers of Alq 3. The ligands are equivalent for facial while ligands are different for meridional, where A, B, and C are different ligands A schematic drawing of α- and β- Alq 3 crystal packing, both phases have triclinic symmetry

15 2.7 A schematic diagram of sublimation system used to purify the organic materials A photograph of the sublimation system. Also shown are micrographs of the PTCDA crystals after sublimation. The stacking direction of PTCDA crystal is along (102) plane of the unit cell which corresponds to the crystal s cutting edge plane. The direction of the unit vector a is indicated A schematic diagram of the OMBD system A photograph of the OMBD system, important parts are labeled Organic thin films and multilayer structures are deposited on substrate such as Pyrex, Si with natural oxides (SiO 2 ), where Organic-1 and Organic-2 represent different organic materials Bragg reflections from parallel planes in a regular lattice array X-ray diffraction pattern from PTCDA powder (Alridch) and sublimated PTCDA crystals The projection of PTCDA crystal unit cell on to the substrate surface. The z-axis is perpendicular to the molecular plane which is parallel to the 102 plane of the crystal. 46 6

16 2.15 Optical cycle including the elongation q 0 of an internal vibrational energy of h ω between the geometry in the electronic ground state and the relaxed excited state Linear Absorption and PL of PTCDA molecules dissolved in dimethyl-sulfoxide. Open circles: experimental spectra 22, discrete vertical lines: multi-poisson distribution of elongated A g modes, solid lines: absorption spectrum and PL spectrum using an effective mode model described in the text Schematic diagram for the linear absorption and photoluminescence in PTCDA crystals and films due to Frenkel excitons including the energy dispersion as described in the text Schematic sketch for (a) a charge transfer exciton (CT) and (b) an excimer transition in a PTCDA crystal Incident light (with wave vector k) enters perpendicular to the substrate surface. The light first passes the Pyrex substrate then the sample. d is sample thickness, and I 0 and I are incident and transmitted intensities Experimental setup for transmission measurements Experimental setup for photoluminescence measurements. 64 7

17 2.22 A sketch of cross section of uniaxial pressure cell A photograph of separate components of the pressure cell Absorption spectrum of a 70 nm thick PTCDA film on Pyrex and PL of a 36 nm PTCDA on Si (100) at room temperature Absorption spectrum of a 70 nm thick PTCDA film on Pyrex and PL of a 36 nm PTCDA on Si (100) at room temperature. Green and Blue arrows indicate the 0-0 transition of PTCDA dissolved in CH 2 Cl Temperature dependent OD spectra of a pure 70 nm thick PTCDA film deposited on Pyrex. The spectra were recorded at temperatures between 10 and 300 K as labeled. The vertical dashed lines indicate the spectral position of the 0-0 and higher Frenkel exciton transitions at 300 K Temperature dependent OD spectra of a 6 [PTCDA 3nm/Alq 3 3nm] multilayer (Multi-6). The temperature for each curve is given. The vertical line indicates the spectral position for 0-0 Frenkel exciton transition at 300 K. The absorption edge for Alq 3 is also shown. 72 8

18 2.28 Comparison of OD spectra of a 70 nm thick PTCDA film and a 6x [PTCDA 3nm/ Alq 3 3nm] multilayer (Multi-6) recorded at 300 K. The vertical dashed line indicates the spectral position of the 0-0 Frenkel exciton transition of PTCDA Comparison of OD spectra of the 70 nm thick PTCDA film and the 6x [PTCDA 3nm/ Alq 3 3nm] multilayer (Multi-6) recorded at 10 K. The vertical dashed line indicates the spectral position of the 0-0 Frenkel exciton transition of PTCDA Comparison of the optical density of PTCDA and multilayer on Pyrex recorded at 300 K with model calculations. The open circles and triangles represent calculated values of a 70 and 18 nm thick PTCDA film, respectively Comparison of the optical density of PTCDA and multilayer on Pyrex recorded at 10 K with model calculations. The open circles and triangles represent calculated values of a 70 and 18 nm thick PTCDA film, respectively X-ray diffraction spectra of a 36 nm thick PTCDA film and of a 6x [PTCDA 3nm/ Alq 3 3nm] multilayer (Multi-6). The structures were grown on Pyrex substrate. The solid lines show Gaussian fits for the α -PTCDA (102) reflex Photoluminescence spectra of a 36 nm thick PTCDA film on Si(001) excited at two different excitation energies (2.33 and 2.84 ev) indicated as vertical arrows 9

19 and absorption spectrum of a 70 nm thick PTCDA film on Pyrex recorded at 10 K. The spectra are offset for better comparison Temperature dependent PL of a 36 nm thick PTCDA film on Si(001) excited at 2.84 ev in the range 10 to 300 K in 10 K steps. The contributing emission channels are labeled PL spectra of a 36 nm thick PTCDA film on Si(001) excited at 2.33 ev obtained at temperatures 20, 80 and 300 K. The experimental data is shown as open circles. Also shown are results of model calculations. The individual emission channels are labeled, the resulting calculated PL spectrum is shown as full line X-ray diffraction spectra (a) on a 70 nm thick PTCDA film and (b) on a 6x [PTCDA 3nm /Alq 3 4nm] multilayer. Both structures were grown on Si(001). The solid and dashed lines show Gaussian fits for α and β-ptcda (102) reflexes, respectively PL spectra of a 6x [PTCDA 3nm /Alq 3 4nm] multilayer grown on Si(001) excited at two different excitation energies (2.33 and 2.84 ev) indicated as vertical arrows and absorption spectrum of the same multilayer grown on Pyrex recorded at 10 K. The spectra are offset for better comparison

20 2.38 Temperature dependent PL of a 6x [PTCDA 3nm /Alq 3 4nm] multilayer on Si(001) excited at 2.84 ev in the range 10 to 300 K in 10 K steps. The contributing emission channels are labeled PL spectra of a 6x [PTCDA 3nm /Alq 3 4nm] multilayer on Si(001) excited at 2.33 ev obtained at temperatures 20, 80 and 300 K. The experimental data is shown as open circles. Also shown are results of model calculations. The individual emission channels are labeled, the resulting calculated PL spectrum is shown as solid line HOMO-LUMO offsets of a PTCDA/Alq 3 interface obtained from XPS measurements. The dash line indicates a possible indirect energy transfer at the interface Normalized PL spectra of a 10 nm PTCDA film on Si(001) (sample A), a 10 nm PTCDA film on a 2 nm Alq 3 layer on Si (001) (sample B) and a 2 nm Alq 3 layer on a 10 nm PTCDA film on Si (001) (sample C) excited at 2.84 ev and recorded at 10 K. The spectra are offset for better comparison Normalized PL spectra of a 36 nm PTCDA film on Si(001) without and with applied uniaxial strains P 1 and P 2 (P 2 > P 1 ). The layer was excited at 2.33 ev and spectra were recorded at 10 K

21 2.43 Temperature dependent PL spectra of a 36 nm PTCDA film on Si(001) with applied uniaxial strain, P 1. The layer was excited at 2.33 ev and the spectra were recorded at 10, 40 and 80 K Normalized PL spectra of α PTCDA single crystals grown by gradient sublimation, a 36 nm thick PTCDA film, a 6x [PTCDA 3 nm /Alq 3 4 nm] multilayer (Multi-6) and a 12x [PTCDA 1.5 nm /Alq 3 2 nm] (Multi-12). The PTCDA film and multilayers were grown by OMBD on oxide covered Si(001) substrates. The samples were excited at 2.33 ev. The PL measurements were performed at 20, 40 and 80 K X-ray diffraction spectra (a) on a 36 nm thick PTCDA film, (b) on a 6x [PTCDA 3 nm /Alq 3 4 nm] multilayer (Multi-6), and (c) on a 12x [PTCDA 1.5 nm /Alq 3 2 nm] multilayer (Multi-12). All films are deposited on oxide covered Si(001). The solid line and dashed lines show Gaussian fits for α and β PTCDA (102) reflexes respectively Optical density as a function of wave-numbers of a 100 nm thick PTCDA film, a 100 nm thick Alq 3 film and of a 6x [PTCDA 3 nm /Alq 3 4 nm] multilayer (Multi- 6) on oxide covered Si(001), obtained from Fourier transform infrared (FTIR) measurements. The dashed-dotted spectrum shows the difference spectrum between the FIR absorption of vibrational modes in the multilayer and in the pure Alq 3 film

22 2.47 Normalized PL spectra of a 22 nm thick PTCDA film on oxide covered Si(001) (sample A), a 20 nm PTCDA film that was deposited on a 2 nm thick Alq 3 layer (sample B), a 22 nm thick co-deposited PTCDA/Alq 3 layer with 10 % Alq 3 content (Co-10) and a 22 nm thick co-deposited film with 50% Alq 3 content (Co- 50). The samples were excited at 2.33 ev. The PL measurements were performed at 20, 40 and 80 K X-ray diffraction spectra (a) on a 88 nm thick co-deposited PTCDA/Alq 3 layer with 10 % Alq 3 content (Co-10), (b) on a 20 nm PTCDA film deposited on top of a 2 nm thick layer of Alq 3 (sample B) and (c) on a 88 nm thick co-deposited film with 50% Alq 3 content (Co-50). All films are deposited on oxide covered Si(001). The solid line and dashed lines show Gaussian fits for α and β PTCDA (102) reflexes, respectively Room temperature PL spectra of crystalline α-ptcda recorded at different pressures up to 8.2 kbar and at two different excitation powers PL spectra of crystalline α-ptcda measured at 11 K for several pressures up to 54.2 kbar. The excitation power was 5 mw. Different exciton emissions are labeled

23 2.51 Uniaxial pressure dependent PL spectra of a 100 nm PTCDA on Si(001) excited at 2.33 ev obtained at 20, 40, and 80 K, the pressure values are given in kbar for different spectrum Uniaxial pressure dependent PL spectra of a 100 nm PTCDA on Si(001) excited at 2.33 ev obtained pressure is changed from 0 to 1.3 kbar. at 300 K. The PL intensity is increased when applied The photographs of patterned ITO substrate for (a) two-layer structure with two possible active devices (produced by our laboratory) and (b) a multilayer structure with four possible active devices (produced by the UC Nanolab) The chemical structures of (a) TPD, (b) α-npb, the hole transport materials, and (c) PEDOT: PSS, a conducting polymer solution Configuration of the first Alq 3 based two-layer OLED structure produced in our laboratory, individual layer thicknesses are given in parenthesis Configuration of the improved Alq 3 based OLED device structure (produced by the UC Nanolab), individual layer thicknesses are given in parenthesis A simple OLED structure, the OLED emits the light under a forward biased condition

24 3.6 Energy level diagrams for a basic organic light emitting diode structure. The radiative recombination takes place at or near the interface of two organic layers The basic operation of the ITO/TPD/Alq 3 /Mg-Ag OLED structure with the carrier injection mechanism for a forward bias, radiative recombination takes place at the Alq 3 /TPD interface HOMO LUMO energy levels for the organic materials and work functions for selected cathode materials and for ITO anode, relative to the vacuum Experimental set-up for I-V characteristics and luminescence measurements Experimental set-up for temperature dependent I-V-L measurements Experimental set-up for electroluminescence measurements PL spectra at 300 and 4.2 K for various Alq 3 system including solution, a thin film (50 nm) onto Si(001) and different polycrystalline samples of the α- and β- Alq 3 and calthrated Alq 3 -(C 6 H 5 Cl) 1/2 and Alq 3 (MeOH). The figure is taken from Brikmann et al. 43, and the scale is converted to energy [ev] (the original scale was in wavenumber [cm -1 ])

25 3.13 Comparison between the absorption spectrum at 300 K and PL spectrum at 4.2 K of polycrystalline α-alq 3. The position and the intensities of the various components of the vibronic structures are also shown. The figure is taken from Brikmann et al. 43, the original scale was in wevenumber [cm -1 ] and is modified to energy scale Absorption spectra of 40 nm thick Alq 3 on Pyrex and PL of 40 nm thick Alq 3 on Si (001) at 10 K (top) and at 300 K (bottom) Temperature dependent ( K) PL of a 40 nm Alq 3 film on Si (001) excited by a laser at 436 nm. The PL intensity in (b) is offset for better comparison Integrated normalized PL intensity (obtained from Fig b) of the Alq 3 film as a function of temperature Peak energy of the PL spectra (obtained from Fig b) of the Alq 3 film as a function of temperature Simulated results of the proposed model. Dashed curve is the PL intensity contribution from trapped excitons, the dotted-dashed curve is the contribution from STEs and the solid line is the total of both recombination channels

26 3.19 Proposed energy level scheme for the trap distribution and the exciton dynamical process for Alq 3 thin films on Si (001) Room temperature J-V characteristics of the first OLED device produced in our lab, and a sketch of the device structure. (b) - A photograph of the operating device emitting bright green light The luminescence and the efficiency of the device as a function of applied voltage at room temperature The efficiency of the device as a function of luminescence at room temperature The EL spectrum of the device (V =12 V; I = 1.2 ma) and the PL spectrum of a 50 nm thick Alq 3 film (λ excitaion = 436 nm; power = 1 mw) on Si(001) at room temperature Configuration of the improved Alq 3 based OLED device structure, individual layer thicknesses are also given in parenthesis The luminescence and current density of the multilayer OLED as a function of the forward biased voltage at room temperature A photograph of the OLED showing bright green light emission

27 3.27 The temperature dependence ( K) of the turn-on voltage of the device I-V characteristics of the device at different temperatures 60, 180 and 300 K as indicated. Experimental current-voltage data (symbol) of the device. The solid line and the dashed line are fitted curves for the equation 3.12 ( I ~ V m+ 1 ) with m ~ 1 and higher values respectively. 147 E 3.29 Variation of m (= t kt ) as a function of 1/temperature. The temperature is ranging from K. The symbols are calculated m values (see Fig. 3.18) and the red line is the fitted curve Temperature dependent ( K) EL of the multilayer OLED structure for a constant forward bias of 8.5 V. The EL spectra in (b) are offset in intensity from each other for better comparison Peak energy of the EL spectra (obtained from Fig b) of the OLED as a function of temperature The normalized EL (light gray) and PL (dark gray) spectra for different temperatures as indicated. Each spectrum is offset in intensity for better comparison

28 3.33 Integrated normalized EL intensity (symbol as obtained from Fig b) of the OLED structure (ITO/PEDOT/TPD/Alq 3 /Al-LiF) as a function of temperature. Also shown is the normalized luminescence of the OLED structure ITO/PEDOT/α-NPB/Alq 3 /Al-LiF as a function of temperature. The luminescence (solid line) was taken from I-V-L measurements Temperature dependence of quantum efficiencies of the PL of the 40 nm thick Alq 3 layer and of the EL of the Alq 3 based multilayer OLED structure at constant forward bias of 8. 5 V

29 List of Tables 2.1 Parameters of PTCDA unit cells. a, b, c and β are three dimensional unit cell parameters 39 as indicated in Fig. 2.2 where d is (102) plane distance Five different recombination channels were detected in α-ptcda single, the mean energy and the exciton life times are also given Parameters for all six emission channels. For each Gaussian function the peak energies hω j in the limit T 0 K, the full width half maximum (FWHM) 8 ln 2hσ j, its relative area a j and temperature blueshift is given A comparison of some device properties of the two-layer OLED and the multilayer OLED at room temperature

30 List of Abbreviations ε Dielectric constant η EL Electroluminescence efficiency η ext External quantum efficiency κ (ω) Extinction coefficient σ Gaussian broadening n 0 STE Number of self-trapped excitons η int Internal quantum efficiency η L Luminescence efficiency η c Out-coupling efficiency R (λ) Photodiode responsivity η PL Photoluminescence efficiency h Plank constant/2π γ rad, X Radiative decay rate for trapped excitons µ Reduced mass R (ω) Reflective index STE Self-trapped energy f Self-trapped exciton formation energy γ (T ) Temperature dependent decay rate for trapped excitons X X Trapped energy 21

31 γ Non-radiative decay rate for self-trapped excitons non rad, STE γ Non-radiative decay rate for trapped excitons non rad, X γ rad,ste Radiative decay rate for self-trapped excitons γ STE (T ) Temperature dependent decay rate for self-trapped excitons Alq 3 c cd CT CT1 CT2 CT2-nr CuPc cw DAC DFT E F E LUMO ETL ev f FIR FTIR FWHM Aluminum-quinoline (Tris(8-hydroxyquinoline)aluminum (iii)) Speed of light [unit] candela Charge transfer Charge transfer 1 transition Charge transfer 2 transition Charge transfer non-relaxed Copper phthalocyanines Continuous wave Diamond anvil cell Density functional theory Fermi energy Energy at lowest unoccupied molecular orbital Electron transport layer [unit] electron volts Fraction of light coupling Far Infrared Fourier transform Infrared Full with at half maximum 22

32 H h HOMO HTL I I det I OLED IR ITO I-V I-V-L J J SCLC J TCLC k k B kbar L LEDs LUMO MO OD OLED Hamiltonian operator Plank constant Highest occupied molecular orbitals Hole transport layer Intensity / Current Photocurrent / Photo detector Organic light emitting diode s current Infrared Indium tin oxide Current-Voltage Current-Voltage-Luminescence Current density Space charge limited current density Trapped charge limited current density Wave vector Boltzmann constant [unit] kilo bar Luminescence Light emitting diodes Lowest unoccupied molecular orbital Molecular orbital Optical density Organic light emitting diode 23

33 OMBD PEDOT PL PMT P OLED PPS ps PTCDA SCLC STEs TCLC TPD TRPL WF XPS XRD Ω/ α α-npb λ Organic molecular beam deposition Poly(3,4-ethylenedioxythiophene) Photoluminescence Photo multiplier tube Power of the OLED Poly(4-styrenesulfonate) [unit] pico second 3,4,9,10-perylene-tetracarboxylic-dianhydride Space charge limited current Self trapped excitons Trapped charge limited current N,N -diphenyl-n,n -bis(3-methylphenyl)1-1 -biphenyl-4-4 -diamine Time resolved photoluminescence Work function X-ray photoelectron spectroscopy X-ray diffraction [unit] Ohms/square sheet resistance Absorption coefficient N,N'-Bis(naphthalen-1-yl)-N,N'-bis(phenyl)benzidine Wavelength µ n Electron mobility ν τ Frequency Decay time 24

34 Chapter 1 1. Introduction During the past years, organic molecular semiconductors have attracted enormous interest because of their advantages of flexibility, lightness, and low cost as they conjugate with the device technology. The promising applications include organic light emitting diodes (OLED s) 1-5, solar cells 6-8, field effect transistors 9-11, photodetectors 6, and organic lasers 12,13. Despite the intensive research activities in this area, the low mobility 10 of organic semiconductor is still a limiting factor for switching to pure molecular electronics 10,14. However, important optoelectronic applications in organic display technology and in solid-state lighting based on OLED s have been demonstrated in recent years. The organic display technology became available in the market a few years ago with Kodak Sanyo, Pioneer and Philips manufactures. Pioneer was the first who came on the market by introducing OLED displays for automotive audio components. Other manufactures have also started producing OLED displays in cell phones, digital cameras and other consumer products. It has already been predicted by leading market analysts that the OLED display technology market such as cell phone displays and flat panel displays would amount $1.5 billion by the year The other major breakthrough in the organic technology has recently achieved by GE Global Research on solid-state lighting by making large area white light devices as an alternative to the conventional lighting solutions 16. Given the pressing need for very low cost circuits used in devices from smart cards to inventory control, the coming decade may indeed become the age of organic optoelectronics. The optical and electronic properties of organic materials are determined by strong intramolecular covalent bonds between atoms inside the molecule and by the much weaker inter- 25

35 molecular van der Waals interactions between different molecular sites 17. There are two types of organic materials that are being used in the organic semiconductor industry, polymers and π- conjugated small molecules. While polymers in general form disordered structures, some of the small molecules tend to form crystalline structures 17. In this work, I will focus on small molecular materials; the heterostructures based on such materials represent the basic building blocks for many optoelectronic device applications. Goal of this work is to explore the quantum mechanical nature of electronic states in organic nanostructures. The first half of my thesis is devoted to an investigation of optical properties of organic thin films and multilayer structures. Temperature dependent absorption and photoluminescence (PL) spectroscopy are used as investigation tools. In addition, X-ray diffraction and far infrared Fourier transform (FTIR) absorption are performed to characterize the structural properties of thin films and mutilayers for a better understanding of the physical nature of these structures. For the fabrication of molecular films and multilayer structures, the versatile and modern technique of organic molecular beam deposition (OMBD) technique is applied. This ultra high vacuum thermal evaporation technique offers the advantages of source material purification as well as monolayer controlled film deposition. Due to the nature of weak van der Waals forces, polycrystalline organic films can be deposited on almost any substrate, disregarding latticematching condition 10, These organic molecular films show significant modification in absorption and emission properties compared to single molecules that are dissolved in a solution 21,22 and also reveal a strong anisotropy in optoelectronic properties 23. Furthermore, as shown by many investigations 10, the fundamental device properties can be strongly influenced by the crystalline nature of the molecular layers. In order to improve the performance of organic 26

36 semiconductor devices, it is very important to understand the microscopic origin of absorption and recombination processes in organic crystals and in thin films and multilayer structures. PTCDA is one of the most intensively used molecules to study the influence of intermolecular interactions on the structural, optical and electronic properties in organic crystals. PTCDA crystallizes in the monoclinic space group C 2h with two nearly coplanar molecules in the unit cell 10,24. Two different crystalline modifications, α- and β-phase, are known 24,25. For both phases the molecules in a PTCDA crystal form layer stacks which are parallel to the surface of the substrate. The optical band gap of PTCDA thin films is defined by the first absorption peak at 2.22 ev 26 that is due to the HOMO (highest occupied molecular orbital) to LUMO (lowest unoccupied molecular orbital) excitation which couples to several molecular vibrations. An interesting organic molecule for multilayer investigations is Alq 3 that has attracted much attention as a superior material for OLED s 1,3. In contrast to planar PTCDA molecules that result in crystalline layers the propeller shaped Alq 3 molecules in general form amorphous films It is therefore of importance to investigate how the crystalline structure of PTCDA films that are embedded in Alq 3 layers changes compared to pure PTCDA films. In particular different exciton emission channels can be strongly affected by structural changes and generated strain fields within the layer. In addition HOMO-LUMO offset measurements using photoelectron spectroscopy 30 indicate an intermolecular charge transfer transition (at 1.85 ev) at the PTCDA/Alq 3 interface. Therefore specific multilayer structures may be used to intentionally modify optical constants in all-optical organic devices and to shift or extend the emission spectrum in organic LED s. Using organic molecular materials for light emitting devices is fascinating due to the easy handling and their vast variety of options available for basic research and developments. In 27

37 1960 s, Pope and coworkers 31 fabricated the first organic electroluminescence (EL) cell in dc mode. The single-crystal antracene based OLEDs were thick and required a several hundreds of bias voltage to get visible light. Therefore the discovery did not spark an immediate research interest. However two decades later, the demonstration of EL from thin sublimed multilayer structures by Tang and VanSlyke 1 has been truly spectacular and has advanced the basic research towards practical OLED technology. In contrast to previous studies, they discovered a low operating voltage EL device with small organic molecules. Many device improvements were made in the last decade with significant performances in device efficiency, lifetime, and remarkable achievements in brightness and in different colors; blue, red, green and white OLEDs. OLED technology has now reached to a point where serious efforts are being undertaken to replace several consumer optoelectronic applications. As mentioned earlier OLED based display devices and the solid-state lighting solutions are already available in the market. Despite intensive research and development in this field, challenges in the low efficiency and the longterm degradation process 32 of OLEDs are still need to be addressed 10,33. Moreover there is a lack of low temperature investigations leading to the carrier injection and the recombination mechanism of OLED structures. The topic for the second half of my thesis is to study Alq 3 thin films and Alq 3 based OLED structures. In particular OLED characterization, temperature dependent current-voltage and electroluminescence measurements are performed and are compared with temperature dependent PL measurements of the Alq 3 thin films. Further, I try to analyze the physics of an operating multilayer molecular device, which is well suited to illustrate the rich variety of physical phenomena that are encountered in the filed of organic LEDs. Therefore, this work shall lead to a better understanding of excitons emission in organic 28

38 multilayer nanostructures and will also help to improve the device performance of optoelectronic applications. 1.1 Outline of the Thesis The thesis is composed of two main parts. The first half includes investigations of the optical properties of PTCDA crystals, films, PTCDA/Alq 3 multilayers and co-deposited layers, which are discussed in Chapter 2. The sample fabrication procedure is explained in Chapter 2.1. Starting with an introduction of the organic materials used in this investigation, it gives a detailed description for the sublimation steps of the organic materials. The organic molecular beam deposition system (OMBD) and related techniques are discussed giving specific procedures for the nanometer scale organic film deposition. The methods of X-ray diffraction and Fourier transform infrared absorption spectroscopy are introduced as an essential tool for analyzing structural investigations. Chapter 2.2 gives a theoretical review of exciton absorption and the emission process in organic molecules and crystals. It starts with a description of the optical excitation of isolated molecules. In organic crystals, optical properties are described by the Frenkel exciton model, which is also presented here. In addition to the mobile Frenkel excitons, self-trapped excitons like charge transfer and excimer also exist in the organic crystals. The importance of these localized excitons in recombination processes is discussed as well. Experimental setups and methods are presented and described in Chapter 2.3. Experimental studies are presented and discussed in Chapter 2.4. Temperature dependent absorption studies on PTCDA thin films and PTCDA/Alq 3 multilayer are discussed and compared with the Frenkel exciton model. Temperature dependent PL on different samples including PTCDA thin films, PTCDA/Alq 3 multilayers and co-deposited layers are extensively 29

39 investigated for a better understanding of the different recombination channels in these samples. The effects of crystalline PTCDA in multilayer structures are analyzed. Furthermore, strain dependent measurements on PTCDA crystals and PTCDA thin films are discussed. The second half of my thesis, Chapter 3 contains the characterization of Alq 3 layers and Alq 3 based OLED structures. In Chapter 3.1, the fabrication procedures of OLEDs, including methods of ITO substrate processing, an introduction of different organic materials used and a description of metal and organic layer deposition are provided. Chapter 3.2 gives an introduction to electro-optical properties of OLEDs including a theoretical background of charge carrier injection and transport mechanism, the energy level diagram for selected materials and definitions of the device properties. Experimental setups for the characterization of OLED s are given in Chapter 3.3. Experimental studies on Alq 3 layers and OLED structures are presented in Chapter 3.4. It begins with a brief survey on absorption and PL of Alq 3 single molecules, polycrystals and thin films. Then temperature dependent PL investigations of Alq 3 thin films are presented along with a proposed model to explain the experimental observations. Different Alq 3 based OLED structures are investigated at room temperature. Furthermore, temperature dependent current-voltage measurements are analyzed for the multilayer device in order to understand the carrier injection, transport and recombination processes of the OLEDs. Temperature dependent electroluminescence studies of the multilayer OLED structure are also presented. Moreover, the EL of the device is compared with the PL of the Alq 3 film in the range from 10 to 300 K. Finally, Chapter 4 summarizes the achievements of this thesis. 30

40 Chapter 2 2. Absorption and Emission Studies in PTCDA Crystals, Films and PTCDA/Alq 3 Multilayers 2.1 Sample Fabrication The ultra high vacuum thermal evaporation system of organic molecular beam deposition (OMBD) is very similar, in many respects, to a conventional molecular beam epitaxial growth system. Typically, deposition takes place at a background vacuum ranging from mbar of highly purified organic materials. There are few other techniques that are used for fabricating organic thin films such as spin coating and Langmuir Blodgett method 34. Unlike these methods, OMBD allows the fabrication of organic layers with only a few nanometer thickness or even molecular monolayers. It also offers a high reproducibility of multilayer structures which is necessary for organic heterostructures such as multiple layer structures and electroluminescent devices. In inorganic semiconductor epitaxy, strong covalent bonding forces require a lattice matching between the inorganic layer material and the substrate. In contrast, organic molecules are bonded by weak van der Walls forces which have the advantage that no lattice matching is required. The low substrate temperature is another advantage of OMBD because it reduces thermal degradation or damage to the device during deposition. Additionally, OMBD enables the deposition of different organic layers of metals and inorganic materials without breaking the vacuum in the system. Combining all of these advantages, OMBD emerges as an ideal research and fabrication technique for organic optoelectronic device structures. In this thesis, I have setup an OMBD system with a background vacuum in the range of 2x10-8 mbar by modifying a home made molecular beam epitaxy system that has been developed 31

41 at the University of Regensburg, Germany. In this system, the flux of the molecular beam is controlled by a combination of the effusion cell temperature as well as by a mechanical shutter allowing the fabrication of monolayer controlled organic nanostructures. Using several Knudsen cells, multilayer structures consisting of alternating layers of different compounds can be grown. A detailed discussion about the OMBD technique and thin film fabrications will be given in the following sections Organic Materials Organic molecular solids are formed by intermolecular forces that are known as van der Waals and London interactions. The weak van der Walls forces predetermine the mechanical and elastic properties of the organic molecular crystals. The lattice binding energy per electron is several orders of magnitude lower for molecular solids than the lattice binding energy per valance electron in inorganic solids. e.g., the lattice binding energy per electron for antracene crystal is ev while the corresponding value for Si is 1.16 ev 35. Accordingly, organic solids in general have a low mechanical strength, low melting points and sublimation temperatures, and high compressibility 36. Furthermore, most of the organic semiconductors are poor electrical conductivity materials. Because of disorders the carrier mobilities in polycrystalline organic films are usually very low. The weak intermolecular interaction forces produce only slight changes on the electronic structure of molecules, and as a result, the individual molecules mostly retain their identity. Correspondingly the optical spectra of an isolated molecule and of a molecular crystal show similar features including their electronic-vibtronic properties. On the other hand, certain new optical and electronic properties like Davydov splitting 37 emerge in the organic crystal due to 32

42 collective molecular interaction 17. The optical properties of organic nanostructures are the main topics in this thesis and they are discussed extensively in the following sections. Due to the low symmetry of organic molecules, the structural investigation of the organic crystal is quite difficult. However, small planar π-conjugated molecules represent a special class of organic materials. Most of them form monoclinic or triclinic crystalline structures. They have more than one molecule in the unit cell and frequently several polymorphs exist. Examples for these types of materials are perylene and pentacence derivative, phthalocyanines and aromatic hydrocarbon groups PTCDA Among many materials studied in past decades, much work has been focused on 3,4,9,10-perylene-tetracarboxylic-dianhydride (PTCDA). This planar aromatic pigment consists of 38 atoms, C 24 O 6 H 8, and the molecule is rather stable in the absence of water. Fig. 2.1 shows the chemical structures of PTCDA. The bond can be formed via reactive dianhydride group or via extended π-orbital which is perpendicular to the molecular plane Å 9.2 Å Figure 2.1: The chemical structure of a PTCDA molecule. The perylene derivative consists of four carboxylic groups at each corner and two anhydride groups at both sides. The length of the molecule is 14.2 Å and the width is 9.2 Å. 33

43 The point group of the isolated molecule is D 2h, the PTCDA molecule has 108 internal vibrational modes. 54 modes are Raman active, 46 are infrared active and the other 8 modes are silent. PTCDA can crystallize in two monoclinic phases, which are known as α-ptcda and β- PTCDA. The crystalline PTCDA belongs to the 5 C2h point group with nearly two coplanar molecules in the unit cell 10,24. Fig. 2.2 shows the unit cells of the two crystalline phases 38. The unit cell parameters 39 are given in Table 2.1. c a b β b a β c Figure 2.2: Unit cells of α-ptcda (a) and β-ptcda (b), the molecular plane coincides with the (102) plane of the crystal 38. a, b and c represent unit cell parameters while β is the angle between a and c unit vectors. 34

44 a[å] b[å] c[å] β [deg] d(102)[å] α-ptcda β-ptcda Table 2.1: Parameters of PTCDA unit cells. a, b, c and β are three dimensional unit cell parameters 39 as indicated in Fig. 2.2 where d is (102) plane distance. Two molecules are positioned at (000) and (½ ½ ½) in the both of the monoclinic phases. The different distinguish is from the stacking direction of the unit cell in the rectangular plane. Fig. 2.3 shows the unit cell projection of PTCDA which is most closely coinciding with the molecular plane. y α-ptcda c =19.89 Å b =11.96 Å y β-ptcda b =19.30 Å x c =12.45 Å x Figure 2.3: Crystal unit cells of two monoclinic phases of PTCDA, in the plane most closely coinciding with the molecular orientation, where c corresponds to the direction of (c-2a). 35

45 Alq 3 Tris (8-hydroxyquinoline) aluminum (Alq 3 ) has attracted much attention as a superior material for organic LEDs 1-3. The chemical structure of Alq 3 is shown in Fig This quinoline group molecule consists of 52 atoms, (C 9 H 6 NO) 3 -Al. Alq 3 has been widely studied because of its technological importance. Films of Alq 3 are used for both light emitting and electron transport layer in OLEDs. Despite intensive research and development of Alq 3 based OLEDs, structural, electronic and optical properties 40,41 42 of this material have not been clearly understood yet Figure 2.4: Chemical structure of Tris (8-hydroxyquinoline) aluminum (III) (Alq 3 ). There are two types of octahedral modifications, they are known as facial and meridional 40,46 and are shown in Fig 2.5. In contrast to planar PTCDA molecules that result in highly ordered polycrystalline layers the propeller shaped Alq 3 molecules in general form amorphous films 10,25,40. Meridional is usually predominant in amorphous films 40. It has been demonstrated that Alq 3 can crystallize in four different phases (α, β, γ and δ) 40,47,48. The Alq 3 crystals have a triclinic structure with two molecules per unit cell. The crystals belong to P-1 space group 40,46. As examples, schematic drawings of the α and β Alq 3 crystal 36

46 packing are given in Fig In Alq 3 solids and thin films, a mixture of the different phases can exist with small grain sizes, and they generally are regarded as amorphous. The optical properties such as absorption and photoluminescence of Alq 3 and electroluminescence of Alq 3 based OLEDs will be discussed in Chapter 3. Figure 2.5: Geometrical molecular structures for meridinal and facial isomers of Alq 3. The ligands are equivalent for facial while ligands are different for meridional, where A, B, and C are different ligands 40. In this chapter, amorphous Alq 3 films are used as inter-layers between crystalline PTCDA films in order to investigate how the crystalline structure of PTCDA films in multilayer structures changes compared to pure PTCDA films. In particular different exciton emission channels can be strongly affected by structural changes and generated strain fields. The experimental observations are presented and discussed in section

47 α-alq 3 β-alq 3 Figure 2.6: A schematic drawing of α- and β- Alq 3 crystal packing, both phases have triclinic symmetry

48 2.1.2 Sublimation of the Organic Materials In general, the purity of the organic materials is lower than the elements used for inorganic semiconductors. Therefore, prior to OMBD growth the organic source materials were purified by the thermal gradient sublimation method. A schematic diagram of the apparatus used for the sublimation process is shown in Fig Pump Glass wools Purified materials Volatile impurity residue Furnace Impurity residue Figure 2.7: A schematic diagram of sublimation system used to purify the organic materials. 39

49 A glass sublimer containing a specific organic material (PTCDA or Alq 3 ) is evacuated to the range of 1x10-6 mbar using a pump station. The pumping station consists of a turbomolecular and a dry pump. A glass sleeve with glass wools was placed at the open end of the sublimation chamber to prevent any contamination from the pumping station. The glass chamber is heated up gradually to the sublimation temperature of the organic material by the furnace. The sublimation temperatures for PTCDA and Alq 3 (Sigma Aldrich) are and C respectively. After several days, the source material is reasonably free of impurities and is then filled into the effusion cell immediately. The PTCDA samples are in the form of needles, typically 50 x 50 µm 2 in cross section and 2 mm in length. A photograph of the sublimation set up and PTCDA crystals extracted from sublimation are shown in Fig µm a 300 µm Figure 2.8: A photograph of the sublimation system. Also shown are micrographs of the PTCDA crystals after sublimation. The stacking direction of PTCDA crystal is along (102) plane of the unit cell which corresponds to the crystal s cutting edge plane. The direction of the unit vector a is indicated. 40

50 2.1.3 Organic Molecular Beam Deposition and Thin Film Growth A schematic diagram of the organic molecular deposition (OMBD) system is shown in Fig The set up is very similar to a conventional molecular beam epitaxial system. The OMBD system has a background vacuum in the range of 2x10-8 mbar. Quartz thickness monitor Turbo-molecular pump x θ z y Sample holder Manipulator Sample holder & heater Substrate Load-lock Transferring rod Shutter Turbo-molecular pump Knudsen cell Figure 2.9: A schematic diagram of the OMBD system. 41

51 A photograph of the OMBD system is shown in Fig The system consists of the main growth chamber and the load-lock. Two chambers are connected via a gate valve and a sample can be transferred between them while keeping the vacuum in the main chamber. The base pressure of the load-lock is in the range of 5x10-7 mbar. The vacuum of the each chamber is maintained by a combination of turbo-molecular pump (Pfeiffer vacuum) and an oil free membrane pump. The growth chamber is equipped with five Knudsen cells. Each cell has a computer controlled mechanical shutter. Four cells are filled with organic materials, namely PTCDA, Alq 3, TPD and CuPc. The remaining fifth cell is filled with an Ag-Mg alloy. The cell temperatures are often maintained at C in order to outgas the source materials. Each Knudsen cell has a water circulation system to keep its temperature constant. The substrate holder is about 10 cm away from the Knudsen cells. It is also equipped with a heater, therefore the substrate temperature can be set between 27 0 C and C during the growth. The substrate holder is attached to a manipulator where it can be moved along x, y and z directions and rotated around z- axis. The growth rate and the film thickness are monitored (MAXTEK TM-100) using a quartz crystal oscillator, which is located close to the substrate. The flux of the molecular beam is controlled by the effusion cell temperature as well as by a mechanical shutter allowing a monolayer controlled growth of organic thin films. Typical deposition rates from the effusion cells are 0.01 nm/s at temperatures of C, C and C for the PTCDA, Alq 3, and CuPc respectively. TPD and Mg-Ag were mainly used for fabrication of OLED structures (see Chapter 3). The deposition rates for TPD are 0.4 nm/s and 0.5 nm/s for Ag-Mg alloy and evaporation temperatures are C and C respectively. A sixth Knudsen will be installed soon in order to facilitate LiF which is being considered as a useful material for OLED structures. Additionally Mg-Ag alloy will be replaced 42

52 by Al metal, where the evaporation temperature is ~ C. The modified OMBD system is then capable of fabricating monolayer controlled organic thin films, multilayer structure, waveguides and organic light emitting devices. Sample holder manipulator Main chamber Load-lock Turbo-molecular pump Sample transferring rod Knudsen cells Figure 2.10: A photograph of the OMBD system, important parts are labeled. 43

53 2.1.4 Morphology Studies on the Films The morphology and the crystalline structure of the organic films are determined by several parameters such as the growth temperature, the surface roughness of the substrate, the deposition rate and the first monolayer geometry on the substrate. Many organic films exhibit significant changes of their optical and electrical properties depending on their morphology 14. Figure 2.11 shows schematic diagrams that represent an organic thin film and a multilayer structure deposited on substrates like Pyrex and Si with natural oxides. Organic material 1, (Organic-1) and organic material 2, (Organic-2) represent different organic materials such as PTCDA and Alq 3. Due to their different thermal expansion coefficients and structural properties, the optical and electrical properties are expected to change in these samples. It is therefore important to study the structural properties and changes of pure organic layers or multilayer films that are grown on different substrate. The following methods were used to determine the morphology of samples that were studied in this work. Organic-1 Organic-2 Organic-1 Substrate Substrate Figure 2.11: Organic thin films and multilayer structures are deposited on substrate such as Pyrex, Si with natural oxides (SiO 2 ), where Organic-1 and Organic-2 represent different organic materials. 44

54 X-ray Diffraction In general, X-ray diffraction (XRD) is one of the most common experimental methods of obtaining details about the crystallinity of organic crystal and films. It is a useful tool to determine lattice constant, stresses and inhomoginities in θ d organic materials. A crystal can be regarded as a three dimensional diffraction grating for electromagnetic waves, where the wavelength is the same order of magnitude as the lattice constant (see Fig 2.12). An incident wavelength λ of Figure 2.12: Bragg reflections from parallel planes in a regular lattice array. the X-ray beam is diffracted by successive parallel planes of atoms according to the Bragg s condition, λ = 2d sinθ 2.1 where, d is the distance of crystal planes and θ is the angle between the incident rays and the plane as indicated in Fig The X-ray diffraction pattern for PTCDA powder (Aldrich), for PTCDA crystals (sublimation), PTCDA thin films and PTCDA/Alq 3 layers were recorded using a Siemens D500 diffractometer with CuK ( Å) radiation. The XRD measurements were performed in the Department of Geology University of Cincinnati with collaboration of Dr. W. Huff. The data were collected in θ - 2θ scan mode at room temperature. The θ - 2θ scan was performed in the range degrees with a step width of 0.02 degree and 10 seconds dwell time. The diffraction intensity [au] (102) PTCDA crystal Aldrich diffraction angle (2θ) Figure 2.13: X-ray diffraction pattern from PTCDA powder (Alridch) and sublimated PTCDA crystals. 45

55 diffraction patterns for PTCDA powder and sublimated crystals are shown in Fig Different Bragg reflexes correspond to particular planes in the crystal according to equation 2.1. Fig shows α-ptcda (102) reflexes occurring at 2θ = 27.8 o for PTCDA powder and for crystals. PTCDA crystals show a clear resolved peak compared to PTCDA powder indicating better structural quality and larger crystal size of the sublimated PTCDA crystals than commercially available PTCDA powder. X-ray diffraction patterns for PTCDA thin films and multilayers are extremely useful to support and explain the interpretations of experimental observations as discussed in section Fourier Transform Infrared Spectroscopy The photon energies associated with the Infrared (IR) spectrum are not large enough to excite electronic transitions but they can excite vibronic modes in a molecule. The energy of these absorptions depends on mass of the atoms, length of the bond and its strength. Fourier hν (102) plane Substrate Figure 2.14: The projection of PTCDA crystal unit cell on to the substrate surface. The z-axis is perpendicular to the molecular plane which is parallel to the 102 plane of the crystal. 46

56 Transform IR spectroscopy can be used to investigate dipole allowed vibronic modes in molecules or solids. An isolated PTCDA molecule has the point group D 2h resulting in 108 internal vibronic modes, 46 of those are IR active modes of B u symmetry. Considering a 3- dimensional PTCDA unit cell geometry given in Fig. 2.14, 36 of these vibrations are in the molecular plane. The remaining, IR active 10 vibronic modes are along the z-axis, which is perpendicular to the molecular plane. As discussed in section 2.2, in both crystalline phases, the unit cell contains two molecules that are nearly coplanar to each other. For weakly interacting substrates like Si with natural oxide surface or Pyrex, PTCDA molecules are deposited with the first monolayer lying flat on the substrate surface. However, the presence of a rough substrate surfaces influences the preferential orientation of the grown layers. The inset of Fig shows the k vector (red) and polarization vector (black) of incident IR light with the direction of the wave vector perpendicular to the surface of the substrate. In the presence of slightly tilted crystallites due to an uneven substrate surface, the IR absorption spectrum shows a higher fraction of out-of-plane modes with respect to the in plane modes. Accordingly the film morphology and the average orientation of the crystallites relative to the substrate surface can be determined 14,35. In this work, far-infrared Fourier transform measurements (FTIR) were performed ex situ using a Bruker FTIR spectrometer 133v by Dr. Young Kim. An unpolarized Globar emission was directed perpendicular to the organic film surfaces. Transmission spectra for Si substrate, PTCDA and Alq 3 thin films as well as PTCDA/Alq 3 multilayers were recorded in the spectral range from cm -1 at room temperature. The experimental observations are discussed in section

57 2.2 Exciton Absorption and Emission Processes in Organic Molecules and Crystals: Introduction to Theory Organic molecules are the structural units of organic crystals and organic films. The atoms within the molecules are bound by covalent bonds. In contrast molecular solids are formed by much weaker intermolecular interactions, e.g., by van der Waals interactions for non-polar organic molecules 17. Therefore organic solids show the properties of their individual molecules to a large extend. The optoelectronic properties of organic molecules are strongly influenced by the π-conjugated electron system which is formed by the overlap of carbon p z orbitals. The nature of π bonding in organic molecular crystals can be described by the orbital hybridization method 17. The highest occupied molecular orbitals (HOMO) as well as the lowest unoccupied molecular orbitals (LUMO) consist of the π electrons; the electronic excitation from HOMO to LUMO is therefore designated as π-π* transition Optical Excitation of Isolated Molecules follows, If the wave function of the system is ψ, the Schrödinger equation can be written as H ψ = Eψ 2.2 where H is the total Hamiltonian operator and ψ consists of electronic states and a nuclear part. According to Born-Oppenheimer approximation 17,36, the change in the distance between nuclei is almost negligible during the electronic transition. Therefore the electronic motion and the nuclear motion of a molecule can be treated separately. Considering Perimeter free electron theory 17 the electronic wave function can be further simplified and obtained the discrete energy 48

58 levels of molecules associated with molecular orbitals (MO). These MOs are a linear combination of atomic orbitals 17,36. In the molecular system, the MOs are filled with electrons considering Pauli s exclusive principle. The lowest energy state (the electronic ground state) can be obtained by adding electron pairs with different spins to the all states below the Fermi level. Due to the orbital overlap the p z orbitals release their degeneracy and form bonding (π) and antibonding orbitals (π*). The lowest optical transition energy is the energy difference between the lowest antibonding molecular orbital (LUMO) and highest bonding molecular orbital (HOMO). The total wave function of the molecules can be written as a product of electronic, vibronic and rotational wave functions, ψ t = φel χvϕr 2.3 with the total energy equal to the sum of the different energies, E t = Eel + Ev + Er 2.4 where, el, v and r are electronic, vibronic and rotational quantum numbers and for a given electronic state. The different vibrational eigenmodes of a molecule can be considered as the motion of single particle harmonic oscillators with an effective mass of µ. Using normal coordinates q and a vibronic coupling constant K, the Schrödinger equation for an electron is given by, 2 2 χν 8π µ E Kq χ = ν q h The solution for this equation gives the vibronic energy levels of the eigenmode as follows, 1 E(ν) = h ω [ ν + ]

59 where, ω is the angular frequency and ν is the vibronic quantum number. Due to these vibronic eigenenergies the optical absorption and emission spectra of a single molecule show a pronounced vibronic progression. In the simplified case of a diatomic molecule (one vibronic eigenmode) the absorption and recombination process can be explained by displaced harmonic oscillators as shown in Fig The minimum of the excited state potential is displaced relative to the ground state potential by the normal coordinate q 0 due to the molecule s different equilibrium elongation. For a dipole allowed HOMO-LOMO transition and assuming that the thermal energy is much lower than the vibronic energy, the optical absorption starts from the lowest vibronic level in electronic ground state 0 g. The transition to different vibronic levels ν e (with ν e 0) of the excited state potential has to be weighted with Franck- Condon factors 14,17 expressed by a Poisson distribution P v over the vibronic levels, S 2v 2 v = 2 2 g g 2 = ν e 0 g = e Pv ( g ) 2.7 ν! In Eq. 2.7 the argument 2 g is the vibronic coupling constant. The moments of the Poisson distributions (Eq. 2.7) have a physical interpretation. The lowest moment gives the reorganizing energy λ, coinciding with the energies measured with respect to the classical potential minima 14 as shown in Fig E 0 = E0 0 + g h ω = E0 + λ 2.8 g The second moment can be related to the line-width of the electronic transition 14,49,50. e g e 2 ( E E ) ghω E = = 2.9 A PTCDA molecule (D 2h point group) possesses 108 vibrational modes, 19 modes have symmetry A g (Raman active modes) 35,51. These symmetric A g breathing modes are invariant with 50

60 respect to D 2h symmetry operations in order to obey the optical dipole selection rules for HOMO-LUMO transitions. Therefore the absorption spectrum of a PTCDA molecule consists of different discrete energy ladders between states 0 gi and ν ei with i = 1, 2, 19. The individual vibronic progression has to be weighted with different vibronic coupling constants 2 g i. E λ = g 2 hω abs PL λ = g 2 hω 0 q 0 q Figure 2.15: Optical cycle including the elongation q 0 of an internal vibrational energy of h ω between the geometry in the electronic ground state and the relaxed excited state. 51

61 In addition sum frequencies between different modes become possible with increasing excitation energy leading to the occurrence of further dipole allowed experiment simulation absorption, PL energy [ev] Figure 2.16: Linear Absorption and PL of PTCDA molecules dissolved in dimethyl-sulfoxide. Open circles: experimental spectra 22, discrete vertical lines: multi-poisson distribution of elongated A g modes, solid lines: absorption spectrum and PL spectrum using an effective mode model described in the text. transitions. Fig shows the linear absorption and PL of PTCDA molecules dissolved in dimethyl-sulfoxide (open circles) 21,22. Also shown is the calculated discrete multi- Poisson distribution of elongated A g modes (vertical lines) using the empirically introduced ground state ν = 0 to excited state µ = 0 transition energy E g 0 e 0 = ev. The calculations were performed using density functional theory (DFT) 14,52. In addition screening effects induced by the high dielectric constant of the solvent were (empirically) considered 14,51. Due to fluctuations of the solvent environment as well as by excitation of rotational levels the vibrationic sub-structure cannot be resolved. After a suitable 52

62 Gaussian broadening of a full width at half maximum (FWHM) = 90 mev the different discrete levels merge into large bands 14 shown in Fig ,50,53. The DFT calculations reveal that the most strongly coupled modes are nearly degenerate and can be replaced by a single internal effective mode 14 with a vibronic energy of ~ 170 mev. Accordingly the PTCDA spectrum can be approximately treated as diatomic molecule with a vibronic coupling constant 2 g = 0.77, a vibronic energy of h ω 0 = 0.17 ev and an energy ' E 0 g 0e = 2.38 ev. The empirically introduced blueshift of Elow = 0.15 ev compared to energy E 0 0 that was used in the DFT calculations is caused by low frequency modes that are not considered by this model 14,50,52,53. The calculated result 51 using this effective model and a Gaussian broadening with FWHM = 90 mev is shown in Fig as a full line. g e The PL spectrum of a dissolved molecule is (up to factor ω 2 ) a mirror image of the absorption spectrum. Using the effective mode model the PL spectrum is given by the expression, 3 I( ω) ω v ( hω ( E ν hω ) 2 P v ( g ) exp σ 2π 2 σ where P ν ( g ) are components of the Poisson distribution, h ω0 is the effective mode frequency, σ is the broadening parameter of normalized Gaussians. The transition energy is '' E 0 0 g e = 2.36 ev, the introduced redshift of Elow = 0.15 ev compared to the DFT calculations is again caused by low frequency modes that are not considered in this 3 effective mode model. The multiplicative factor ω is related to the density of states of the absorbed photons. 53

63 2.2.2 Frenkel Excitons in Organic Crystals Since the molecules in a molecular crystal retain their identity to a large extend, the description of the electronic states of the organic crystal is quite different from that of metals or covalently bounded solids. The basic theoretical concepts were formulated by Frenkel 10,17,50,53 and Davydov 10,37. The basic Ansatz for a crystalline ensemble of N molecules is given by a Hamiltonian H that contains the sum of molecular energies H n and interactions V nm between nuclei and electrons from different molecules: H = H n + Vnm n n, m>n 2.11 Since in molecular crystals the overlap of wave functions from adjacent molecules can be (to good approximation) neglected the ground electronic state of the crystalline ensemble can be described as a product of ground-state wave functions, one for each molecule: N ( g) ( g) ψ = φi 2.12 i= 1 The first excited electronic states will be made by exciting a single molecule to an excited state; however, this is N-fold degenerate in the crystal: ( e) ( e) ( g ) ψ i = φi φ j, i = 1, 2, 3... N 2.13 j i (g) (e) (It should be noted that the molecule wave functions φ i and φ i in the formed crystal are slightly different from the wave functions of the isolated molecules as a result of the mutual polarization of the molecules in the crystal lattice.) The intermolecular interactions split this degeneracy of (e) ψ i and forms N non-degenerate states. The influence of the interaction potential can be described by perturbation theory in first 54

64 55 order. The energies of these electronic states can be found by diagonalizing the electronic Hamiltonian in the basis of these states. The matrix elements read 14,53 ; g g g m g n m n nm g m g n g g g D E V E H + = + = > ) ( ) ( ) ( ) ( ) ( ) ( φ φ φ φ ψ ψ = + + = > n m nm e e g m e n n m nm e m g n g m e n n m nm g m e n e e e M D E V V E H ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( φ φ φ φ φ φ φ φ ψ ψ 2.15 g D and e D describe the environmental shift due to the molecular lattice. The matrix elements M nm describe the energy transfer from excited molecule n to the unexcited molecule m. For a periodic crystal lattice this Ansatz is solved using Bloch waves leading to the occurrence of an energy band as a function of the wave vector k. In the case of a molecular crystal with two different oriented molecules per unit cell, as for PTCDA crystals, the transition energy band is given as follows: ) ( ) ( ) ( ) ( ) ( ) ( ) ( k k k k k T T D E T T D D E E E e g g e e g ± + + = ± + + = ± 2.16 with ) exp( ) (, β β α αβ m n m n M n ikr T = k 2.17 where ) ( 11 k T is the interaction with molecules that are invariant with respect to translations and ) ( 12 k T is the interaction with not translational invariant molecules. The occurrence of two bands with splitting energy of ) ( 2 12 k T is called Davydov splitting. For direct optical transitions the wave vector is k = 0.

65 The splitting energies are frequently determined by using the point-dipole approximation with 35,37, 2 ( µ µ ) R 3( µ R µ R ) PD 1 ( k) = nge mge nm nge nm)( 5 4πε 0 εrrnm M nm mge nm 2.18 This approximation is only reasonable if the distance R nm between the center mass of molecules is large compared to the dipole extension which means for distances much larger than the spatial extension of a molecule. For modeling the experimentally observed optical density in polycrystalline PTCDA films and PTCDA/Alq 3 multilayers we applied a recent theoretical approach developed by Scholz et al. 14,49,50,54. In this model the delocalized nature of the transition dipole was considered by expressing the HOMO and LOMO π wave functions as sums of localized atomic-like states. In this approximation the interaction of two extended molecular transition dipoles can be calculated from the interaction between two distributions of point charges on each molecule. The distribution of the transition charge density has been calculated using density functional tight-binding techniques. Also included in this model are excitations into higher vibronic transitions 0 g ne (n e = 0, 1,.5) using an effective vibronic mode of h ω 0 = 0.17 ev. The calculated Frenkel exciton energies reveal a distinct minimum at the Brillouin edge (k z = π/d) along the stacking direction of the monoclinic crystal with d being the distance between stacked PTCDA molecules. E 1 ( k) = E0 0 + nh 0 + D + T11( n, k) T12 ( n, k) 2.19 ± ω ± n g e 56

66 For linear absorption calculations (k = 0) the discrete transition energies with the corresponding oscillator strengths were replaced by Gaussian line profiles in the imaginary part of the dielectric tensor 14,53. For the first six vibronic sublevel imaginary part ε '' xx ( ω) reads, ε '' xx ( ω ω' ω 0 ± gνe e 2 5 h F0 gν e ± 1 2σ ± ) = e e ν mε V E 2πσ 0 ν e 0 gν e ± ν e ± where E0 ν ± g e ω0 ± = gν are the calculated transition frequencies according to Eq. 2.20, e h σ ± are empirically introduced full width at half maximum (FWHM) broadenings and ν e F 0 ν ± are oscillator strengths to individual vibronic transitions. Accordingly the g e low-frequency Low-frequency modes: Modes: mev mev dispersion: 120 mev lack of final state: 170 mev Figure 2.17: Schematic diagram for the linear absorption and photoluminescence in PTCDA crystals and films due to Frenkel excitons including the energy dispersion as described in the text. 57

67 remaining tensor elements can be computed. A more detailed description can be found in Refs. 14,53,55. The absorption process and the energy dispersion of Frenkel exciton are visualized in Fig Due to the energy dispersion (Eq. 2.19) the Frenkel exciton energy at k z = π/d is lowered by 0.12 ev with respect to energy the Γ point which significantly affects the radiative recombination: Generated Frenkel excitons with vanishing k vector relax via vibronic excitations of suitable energy and wave vector into the energy minimum of the lowest dispersion branch (n e = 0, k z = π/d) (see Fig. 2.17). This thermalization process happens in the order of several tens of picoseconds 50. As the electronic and vibronic ground state ψ g has vanishing total momentum it is not available as a final state after vertical radiative recombination. Therefore, the lowest accessible states after recombination are vibronic excitations propagating with the same wave vector as the Frenkel exciton before the recombination. With this knowledge and using the effective vibronic mode of h ω 0 = ev a Hamiltonian for the final states can be defined in analogy to Eq ,52,53. Under certain approximations 14,50,52,53 this Hamiltonian leads to a similar expression as given in Eq and has been used in this PhD thesis to describe the photoluminescence properties of PTCDA films and PTCDA/Alq 3 films. A detailed explanation will be given in section 2.4. The complete optical cycle including absorption at k = 0 and Frenkel exciton emission at k z = π/d is summarized in Figure The recombination of Frenkel excitons starts from an energy that is by the dispersion energy Edisp = 0.12 ev lower than the 0g 0 e absorption energy. The first accessible final state for recombining excitons 58

68 is the n g = 1 level which lowers the observed transition energy by h ω = ev. 0 Furthermore, as already mentioned in paragraph 2.2.1, the effective mode model does not consider Stokes shifts that are caused by low frequency modes. Besides the internal low frequency A g modes in PTCDA molecules, low frequency external vibronic modes (soft lattice vibrations or phonons) become available. Both internal and external modes contribute 14,52,53 to this Stokes shift that is supplementary introduced into the effective mode model. Detailed investigations using Raman measurements show that these lowfrequency modes lead to a further reduction of the PL emission energy of E 0.11 low ev (including small correlation effects 14 ). The Frenkel exciton emission line from PTCDA crystals and films is therefore expected at an energy that is shifted by ~ 0.4 ev with respect to the lowest exciton absorption energy which is in good approximation to experimentally observed PL spectra as demonstrated in section Charge Transfer Excitons and Excimer Transitions Besides the delocalized Frenkel excitons (Bloch-waves) where the molecule location (molecular site) of the optical excitation is undetermined, there are also localized excitations possible in organic crystals. These localized excitations occur when the interaction with lattice phonons that causes a self-made local lattice distortion becomes more important than the interaction of the excitation and the entire crystal which is responsible for delocalization. Since these excitations become immobile by their own generated relaxed lattice deformation (which causes a negative trapping energy) they are also called self-trapped excitons. One type of self-trapped excitons are charge transfer excitons (CT) where the center of mass electron and hole (positive) charge distribution 59

69 lies on two different molecules (see Fig (a)). Another possible excitation is the excimer that schematically sketched in Fig (b). In contrast to the CT exciton the (a) + (b) - Figure 2.18: Schematic sketch for (a) a charge transfer exciton (CT) and (b) an excimer transition in a PTCDA crystal. electron and hole distribution in excimers is extended over two (or several) molecules without different net charge on the two molecules. Self-trapped exciton transitions are in general not important in the optical absorption process due to their low oscillator strength 14,49,50,52,53 but they are of significance for the emission properties since they can be generated during the relaxation of Frenkel excitons. In contrast to free Frenkel excitons where electron and hole wave functions are predominantly extending over one molecular site self-trapped excitons possess relatively long lifetimes because of the larger extended wave function. Recent investigations on PTCDA single crystals 52,56-58 using time-resolved photoluminescence (TRPL) measurements 59 at temperatures between 10 and 300 K have shown, that the emission spectrum at low temperature consists of a Frenkel exciton emission as well as of two different charge transfer excitons (CT1 and CT2). Charge transfer exciton CT1 is between molecules within the same unit cell and CT2 is between 60

70 stacked molecules in different unit cells. At higher temperatures an excimer transition becomes the dominant emission channel. The individual channels have been identified according to their different recombination times. Table 2.2 summarizes the individual emission channels with their mean energies and radiative decay times. E /ev τ /ns Frenkel CT CT Excimer Table 2.2: Four different recombination channels that were detected in α-ptcda single, the mean energy and the exciton life times are also given at 10 K. A detailed microscopic description of the formation and recombination of relaxed excimer and CT transitions for α- PTCDA, presently under investigation in the group of Dr. Scholz 52-54,57, is beyond the scope of this thesis. To describe the CT and excimer transition that are observed in our PTCDA films and PTCDA/Alq 3 multilayers we used the expression given in Eq in effective mode approximation with empirically introduced transition energies, Gaussian broadening and phonon coupling constants. A more detailed description of the PL line shape analysis is given in section

71 2.3 Experimental Setups for Absorption and PL Investigations Transmission Measurements A 50 Watt tungsten halogen lamp was used as light source for transmission measurements of films and multilayers on Pyrex substrates. The lamp emits lights in the range from 300 nm to 2000 nm with the peak wavelength ~ 900 nm with luminous intensity 17.7 cd. In the transmission measurements the light first passes the Pyrex d x I 0 k I y z Figure 2.19: Incident light (with wave vector k) enters perpendicular to the substrate surface. The light first passes the Pyrex substrate then the sample. d is sample thickness, and I 0 and I are incident and transmitted intensities. substrate and subsequently the organic layer as shown in Fig The wave vector k of the incident light fields was oriented perpendicular to the sample surface defining the x-y plane that is equal to the (102) PTCDA crystal lattice plane. The spectrally resolved transmitted intensities for the sample (I) and for the Pyrex alone (I 0 ) are analyzed using a grating monochromator and a GaAs photomultiplier (PMT) enabling a spectral range from 1.5 to 3.5 ev. Then, the optical density OD ( I ) = ln I 0 of the films is obtained. 62

72 For variable temperature measurements between 10 and 300 K a closed-cycle He cryostat (CTI-Cryogenics) was used. Figure 2.20 shows a sketch of the transmission setup. I 0 Sample in the cryostat I Monochromator Halogen lamp x y z GaAs PMT Computer Figure 2.20: Experimental setup for transmission measurements Photoluminescence Measurements In the photoluminescence measurements the films were either excited by frequency doubled 100 fs Ti-sapphire laser pulses at 2.84 ev (λ = 436 nm) or by a cw solid-state laser at 2.33 ev (λ = 532 nm). The laser power was less than 1 mw while the diameter of the laser spot on the sample was around 1 mm 2. The PL spectra were analyzed by a grating monochromator (HR 320) and a GaAs photomultiplier (PMT). Figure 2.21 shows a sketch of the PL setup. The spectral response of the entire detection 63

73 system was determined with blackbody radiation using a tungsten halogen lamp with given emission spectra. Because of the flatness of the spectral response in the PTCDA emission range (1.5 to 2.2 ev) no spectral correction was applied to the PL measurements. For variable temperature measurements between 10 and 300 K a closedcycle He cryostat (CTI-Cryogenics) was used. Solid state laser λ = 532 nm Millennia X Ti: sapphire laser Frequency Doubling λ = 436 nm Sample in the cryostat Monochromator x y z GaAs PMT Computer Figure 2.21: Experimental setup for photoluminescence measurements. 64

74 2.3.3 Strain Dependent Measurements Strain dependent measurements were carried out using a home made uniaxial pressure cell. A 90 nm PTCDA film on Si (001) was placed inside the pressure cell with the sample-facing the sapphire window (see Fig. 2.22). The diameter of the optical window is around 5 mm. The sample can be pressed against the sapphire window using a metal piston with a back screw as shown in Fig To reduce shear forces the screw was equipped with a ball bearing at the top. A photograph of the separate parts of the pressure cell is also shown in Fig This simple pressure cell can be used to apply a uniaxial pressure on the sample up to 1 kbar. The different pressure values were estimated using a reference PL spectra of a PTCDA crystal which were performed in a ruby-calibrated diamond-anvil cell 60. The PTCDA film was photoexcited with cw solid state laser at 2.33 ev (λ= 532 nm) through the optical window (see fig. 2.23). The laser beam was directed at a small angle and back scattered light was collected using a lens and the PL setup which is similar to Fig For variable temperature measurements (between 10 and 80 K) the pressure cell was placed inside a closed-cycle He cryostat (CTI-Cryogenics) Screw Piston Sample Main cell Piston holder Ball bearing Si Substrate Sapphire window Figure 2.22: A sketch of the cross section of uniaxial pressure cell. 65

75 Sapphire window Piston holder/screw Mount stage Main cell Piston Optical window Figure 2.23: A photograph of separate components of the pressure cell. 66

76 2.4 Experimental Studies While the absorption and emission spectra of PTCDA single molecules reveal a well resolved vibronic progression (see Fig. 2.16) no clear vibronic sub-bands are observed in PTCDA thin films. Figures 2.23 and 2.25 show the absorption spectrum of a wavelength [nm] e - 1 g 0.4 ev 10 K PL intensity 0 g - 0 e absorbance energy [ev] Figure 2.24: Absorption spectrum of a 70 nm thick PTCDA film on Pyrex and PL of a 36 nm PTCDA on Si (100) at room temperature. 70 nm thick PTCDA film on Pyrex and the PL spectrum of a 36 nm thin of PTCDA film on Si (100) recorded at 10 K and at room temperature, respectively. The green and blue arrows in Fig indicate the 0-0 transition for absorption and PL of dissolved PTCDA at 300 K for comparison. The absorption and PL spectra of thin films are significantly broader and the Stokes shift between the lowest absorption and the highest PL is larger 67

77 than for dissolved PTCDA molecules. As demonstrated in the following sections the absorption and the PL at 10 K is dominated by transitions of Frenkel excitons. The large Stokes shift at 10 K is explained by the radiative 0e 1g transition of relaxed indirect Frenkel excitons (see section 2.2). The mandatory presence of an effective vibron wavelength [nm] ev excimer 300 K PL intensity 0 g - 0 e absorbance energy [ev] Figure 2.25: Absorption spectrum of 70 nm thick PTCDA film on Pyrex and PL of a 36 nm PTCDA on Si (100) at room temperature. Green and Blue arrows indicate the 0-0 transition of PTCDA dissolved in CH 2 Cl 2. with energy hω = ev lowers the transition energy. The Frenkel exciton dispersion energy of of = 0.12 ev as well as a low frequency mode shift of E 0.11 Edisp low ev finally lead to a total redshift of ~ 0.4 ev of the PL with respect to the 0 g -0 e absorption peak. 68

78 As shown in the following sections the excimer transition is the dominating transition at room temperature. Due to its self-trapping energy the emission energy is further lowered by ~ 0.03 with respect to the 0g 0e Frenkel exciton transition energy. More detailed temperature dependent absorption and photoluminescence studies of PTCDA thin films and PTCDA/Alq 3 multilayers are presented in section to The experimental observed energy shifts are explained using theoretical models 14,49,50,53. The transmission and PL investigations are accompanied by structural investigation such as XRD and FTIR. It will be shown that the emission and absorption is strongly affected by internal strain within the layers. We therefore also performed pressure dependent measurements on PTCDA crystals and PTCDA thin films that are presented in section and

79 2.4.1 Temperature Dependent Absorption of PTCDA Films In this section, the investigations of the exciton absorption in PTCDA films and in PTCDA/Alq 3 multilayers are presented. The measurements are performed at various temperatures ranging from 10 to 300 K. Changes of the line-shape and of the energetic position of the zero-vibron π-π * Frenkel exciton transition as well as of the higher vibronic absorption band provide information about the microscopic interaction between molecules and the coupling of excitons to various vibronic states. The influence of strain, of dielectric shifts and of surface excitons on the absorption properties are discussed in order to explain the observed experimental effects. The absorption measurements are optical density T [K] energy [ev] Figure 2.26: Temperature dependent OD spectra of a pure 70 nm thick PTCDA film deposited on Pyrex. The spectra were recorded at temperatures between 10 and 300 K as labeled. The vertical dashed lines indicate the spectral position of the 0-0 and higher Frenkel exciton transitions at 300 K. 70

80 accompanied by structural investigations using X-ray diffraction and the interpretations are supported by model calculations that were performed by Venkateshwar Gagilenka 55. Figure 2.26 shows the temperature dependent optical density OD ( I ) = ln I 0 of a 70 nm thick PTCDA film at various temperatures ranging from 10 to 300 K in temperature steps as labeled. The ratio I/I 0 is the intensity of the transmitted light passing through the sample divided by the emission spectrum of the lamp including the wavelength dependence of the detection system. At room temperature (300 K), the OD spectra reveal a narrow absorption band at ev (indicated by a vertical dashed line, 0-0) that is attributed to a zero-vibron π-π * Frenkel exciton transition 49,50,53. This 0-0 absorption line is followed by a broad and slightly structured band centered at 2.53 ev that is attributed to Frenkel exciton transitions into higher vibronic subbands as labeled by the vertical dash lines 50,53. The 0-0 line as well as the vibronic structure in the broad absorption band above 2.4 ev becomes more distinct at low temperatures. With decreasing temperature the 0-0 transition as well as the broad band shift by about 20 mev towards lower energies. Similar measurements were performed on a 6 [PTCDA 3nm /Alq 3 3nm] multilayer (Multi-6) sample. Figure 2.27 shows temperature dependent optical density of Multi-6, the 0-0 transition shifts toward the higher energy and has a less relative height compared to PTCDA. The absorption spectrum has a higher tale around 2.8 ev is due to the absorption of Alq 3. The broad band is again attributed to Frenkel exciton transitions into higher vibronic subbands. 71

81 optical density Alq energy [ev] T [K] Figure 2.27: Temperature dependent OD spectra of a 6 [PTCDA 3nm/Alq 3 3nm] multilayer (Multi-6). The temperature for each curve is given. The vertical line indicates the spectral position for 0-0 Frenkel exciton transition at 300 K. The absorption edge for Alq 3 is also shown. Fig summarizes the optical densities obtained from PTCDA and Multi-6 at 300 K. In comparison to the pure PTCDA film, the absorption of multilayers exhibits a spectral shift by ~25 mev to higher energy. Furthermore the 0-0 absorption peak decreases relative to the broad absorption band in the multilayers. The relatively higher absorption in Multi-6 above 2.8 ev is caused by the Alq 3 absorption in the samples. Fig shows the OD spectra of PTCDA and Multi-6 at 10 K. The absorption of pure PTCDA reveals a red shift of the OD spectrum of 20 mev compared to the spectra taken at 300 K. In contrast, the energetic shift of multilayer is significantly weaker and amounts to ~10 mev. Again the 0-0 absorption peak is reduced in intensity relative to the broad 72

82 absorption band in the multilayers with respect to the pure PTCDA film. The slightly higher absorption in Muli-6 above 2.8 ev is again caused by the Alq 3 absorption. optical density wavelength [nm] K 0-0 PTCDA Multi - 6 optical density K 0-0 wavelength [nm] PTCDA Multi energy [ev] Figure 2.28: Comparison of OD spectra of a 70 nm thick PTCDA film and a 6x [PTCDA 3nm/ Alq 3 3nm] multilayer (Multi-6) recorded at 300 K. The vertical dashed line indicates the spectral position of the 0-0 Frenkel exciton transition of energy [ev] Figure 2.29: Comparison of OD spectra of the 70 nm thick PTCDA film and the 6x [PTCDA 3nm/ Alq 3 3nm] multilayer (Multi-6) recorded at 10 K. The vertical dashed line indicates the spectral position of the 0-0 Frenkel exciton transition of PTCDA. PTCDA. For a qualitative explanation the measured OD spectra are compared with an existing model for the dielectric properties that was developed for α-ptcda films and has been introduced by Vragvić et al. 14,50,53,54. As already mentioned the discrete transition energies with the corresponding oscillator strengths were replaced by Gaussian line profiles in the imaginary part of the dielectric tensor, and the real part is obtained from a Kramers-Kronig transform. In a poly-crystalline film with random azimuthal orientation of the crystalline grains the two tensor elements ε xx (ω) and ε yy (ω) have to be 73

83 averaged, resulting in an effective dielectric function ε(ω)= 2 1 [εxx (ω)+ε yy (ω)] in the (102) 2 2 plane from which the refractive index n ( ω ) = 1 [ ε ' + ( ε ') + ( ε ") ], the extinction coefficient ( ) 1 ω κ ω = ε"( ω) and the absorption coefficient α( ω) = 2 κ( ω) can be 2n( ω) c calculated in the usual way. Considering the reflections from the surfaces of the organic thin films (thus approximating the Pyrex substrate/organic layer system to a freestanding organic layer) and using Lamberts-Beer Law, the optical density is given by, 2 [ ω ] + α( d ( I ) = ln ( 1 R( )) 2 ln I 0 ω). (2.21) In Eq. (2.21), α(ω) is the absorption coefficient, d is the thickness of the organic thin film, and R(ω) the reflection coefficient, given by 2 2 ( n( ω) 1) + κ ( ω) = 2 2 ( n ( ω) + 1) + κ ( ω) R ( ω) (2.22) at normal incidence. In these calculations interference effects within the organic film are not considered. The small film thickness and inplane refractive index of n < 2.5 in the transparent region generates only faint and spectrally broad interference pattern 55. Figures 2.30 and 2.31 show the comparison between experimentally obtained optical densities and model calculations for the 70 nm thick α-ptcda film (open circles) at 300 and 10 K, respectively. For the model calculations at 300 K parameters were taken from the table of Ref. 49. For optimum agreement the film thickness d in the calculation was adjusted to 68 nm. The small deviation from the measured thickness of 70 nm is attributed to experimental uncertainties in the thickness determination. The agreement 74

84 between the experimental results obtained from the PTCDA thin film and the model calculations are very good demonstrating the high structural and optical quality of the PTCDA films. The structural quality is also supported by X-ray measurements depicted in Figure 2.32 that show a completely symmetric α-ptcda (102) reflex occurring at 2θ = o. However, the full width at half maximum (FWHM) of 0.37 o indicates the presence of strain caused by tilted crystallites with respect to the substrate normal, which is introduced predominantly by the surface roughness of the Pyrex substrate. 3 wavelength [nm] K PTCDA Multi-6 3 wavelength [nm] K PTCDA Multi-6 optical density 2 1 optical density energy [ev] energy [ev] Figure 2.30: Comparison of the optical density of PTCDA and multilayer on Pyrex recorded at 300 K with the model calculation. The open circles and triangles represent calculated values of a 70 and 18 nm thick PTCDA film, respectively. Figure 2.31: Comparison of the optical density of PTCDA and multilayer on Pyrex recorded at 10 K with the model calculations. The open circles and triangles represent calculated values of a 70 and 18 nm thick PTCDA film, respectively. 75

85 For the calculations at 10 K (see Fig. 2.31) the energies of all vibronic sublevels given in Ref. 49 were shifted by 20 mev to lower energies to match the model calculation with the experimental data, and the Kramers-Kronig transform was defined starting from the modified positions of the excitonic resonances. Furthermore the FWHM of the Gaussian 0-0 transition was reduced by a factor of More resolved structures within the high energy broad band also indicate a decrease of the broadenings of the higher vibronic transitions with v 2 but for simplicity the FWHM values for the higher e vibronic levels were kept constant. The reduction of the inhomogeneous broadening of the Gaussian lines is assigned to a reduced population of excited low-frequency internal vibrations and external phonon modes in the electronic ground state. The thermal contraction of the Pyrex substrate is transferred to the in-plane PTCDA molecules, so that the red shift D bulk induced by neighbouring molecules in the (102) plane becomes more pronounced 54. In addition, the film experiences thermal contraction of the PTCDA film along the stack direction. Due to the high expansion coefficient of crystalline PTCDA along this direction 38,61-63 this effect is by one order of magnitude larger than the thermal contraction of Pyrex. This strong thermal contraction along the PTCDA stacks causes an increased intermolecular overlap between PTCDA molecules leading in turn to an enhanced environmental shift D bulk. As the distance dependence of the exciton transfer has only a weak influence on the maximum of the exciton dispersion at Г point, it can be concluded that the observed red shift at low temperatures is mainly induced by the red shift of the molecular HOMO-LUMO transitions determined by the reduced distance to the stack neighbours 60, Figure 2.30 and 2.31 also show calculations for a 18 nm thick PTCDA film (open triangles) at 300 K and 10 K, respectively. The chosen thickness 76

86 in the calculations is equal to the nominal PTCDA net thickness of Multi-6. For optimum agreement between the experimental OD values of Multi-6 and calculated value the energies of all vibronic sublevels have been shifted by 25 mev to higher energy compared to PTCDA film energies 49. This energetic shift in multilayers can have different reasons: One possibility is the reduction of internal compressive strain within the PTCDA layers since the tilted PTCDA crystallites in multilayers are too short to produce significant stress induced by the grain boundaries between different M u lti-6 crystallites. This interpretation is supported by X-ray diffraction measurements (see Fig. 2.32) that show a decreasing FWHM of the diffracted α-ptcda (102) reflex with decreasing PTCDA layer thickness. diffraction intensity P T C D A For the multilayer the FWHM amounts to 0.35 o. Still, there is a significant discrepancy in the relative intensities between the 0-0 transition and the higher vibronic transitions. A detailed analysis of the influence of exciton transfer between equivalent basis molecules on the imaginary part of the dielectric tensor components shows 50, d iffra c tio n a n g le 2 θ Figure 2.32: X-ray diffraction spectra of a 36 nm thick PTCDA film and of a 6x [PTCDA 3nm/ Alq 3 3nm] multilayer (Multi-6). The structures were grown on Pyrex substrate. The solid lines show Gaussian fits for the α - PTCDA (102) reflex. 77

87 that an increasing transfer redistributes the spectral weight of the absorption curve to higher vibronic transitions due to the positive sum of the hopping integrals. It also shifts the lower lying vibronic levels to slightly higher energies. A possible increase of the exciton transfer in the multilayer structure is attributed to a reduced dielectric screening within the PTCDA film due to inter-diffusion of Alq 3 molecules into the PTCDA layer, most probably into areas between PTCDA crystallite grain boundaries. Furthermore the PTCDA molecules at PTCDA/Alq 3 layer interfaces lack a PTCDA stack neighbor and therefore interact with the adjacent Alq 3 molecule. This reduces the environmental shift from D bulk to a value D film 50,53 causing a blue shift of the absorption curve. Finally, the observed red shift of the OD spectra as well as the increased strength of the 0-0 transition with decreasing temperature in Multi-6 is interpreted in a similar way as for the pure PTCDA layer. However, the Alq 3 layers and inter-diffused Alq 3 molecules seem to reduce the thermal contraction within the PTCDA crystallites explaining the reduced red shift of 10 mev compared to the energetic shift of 20 mev observed in pure PTCDA. An extended discussion regarding absorption of PTCDA and multilayers along with optical density measurements of PTCDA/ Alq 3 co-deposited films is given else where

88 2.4.2 Temperature Dependent Photoluminescence In this section, the exciton emission of OMBD grown PTCDA films, PTCDA/Alq 3 multilayers and PTCDA/Alq 3 co-deposited films are investigated by temperature dependent PL spectroscopy. In the photoluminescence measurements the films were either excited by frequency doubled 100 fs Ti-sapphire laser pulses at 2.84 ev (λ = 436 nm) or by a cw solid-state laser at 2.33 ev (λ = 532 nm). To avoid any structural damage in the organic films the average laser power was set below 0.1 mw in all PL measurements. The observed emission bands from films and multilayers are compared to recombination channels obtained from single PTCDA crystals 52,56,58. The PL measurements are accompanied by structural investigations using X-ray diffraction and Fourier transform infrared (FTIR) spectroscopy measurements. Since these investigations are focused on intrinsic film properties the films are prepared on weakly interacting substrates like Si with natural oxide surface or Pyrex PTCDA Thin Films Fig shows the PL spectra at 10 K of a 36 nm thick PTCDA film grown on Si (001) for two different excitation energies (2.84 ev and 2.33 ev) as indicated by vertical arrows. In addition the absorption spectrum of a 70 nm PTCDA film on Pyrex is displayed revealing a narrow absorption band at 2.22 ev that is attributed to a zerovibron π-π * Frenkel exciton transition 49,53. This 0-0 absorption line is followed by a broad and slightly structured band centered at 2.53 ev which is attributed to Frenkel exciton transitions into higher vibronic subbands. No significant changes in the PL spectrum are observed as the excitation energy is changed from 2.84 ev on the high 79

89 energy side of the π-π * absorption to 2.33 ev closer to the narrow 0-0 exciton absorption band. wavelength [nm] K PL intensity absorbance energy [ev] Figure 2.33: Photoluminescence spectra of a 36 nm thick PTCDA film on Si(001) excited at two different excitation energies (2.33 and 2.84 ev) indicated as vertical arrows and absorption spectrum of a 70 nm thick PTCDA film on Pyrex recorded at 10 K. The spectra are offset for better comparison. Compared to the PL obtained from single PTCDA crystals 26,52,56,58 the PTCDA film shows a very similar PL structure, however, the whole emission is shifted to lower energies by approximately 20 mev. This redshift is attributed to weak biaxial compressive strain within the layer that is caused by different expansion coefficients of the Si substrate and the crystalline film. Furthermore the PTCDA film reveals a more pronounced high energy band at 1.95 ev compared to the PTCDA crystals. From 80

90 energetic reasons this band can not be attributed to a relaxed monomer transition. Recent calculations 67,68 show that this transition might be assigned to a non-relaxed charge transfer exciton transition between stacked molecules in different unit cells, denoted as CT2-nr in what follows. Frenkel CT2 CT1 CT2-nr 10 PL intensity temperature [K] energy [ev] 300 Figure 2.34: Temperature dependent PL of a 36 nm thick PTCDA film on Si(001) excited at 2.84 ev in the range 10 to 300 K in 10 K steps. The contributing emission channels are labeled. Fig demonstrates temperature dependent PL spectra of the same PTCDA film in 10 K steps. Again the PL of the PTCDA film shows very similar temperature dependence as observed in single crystals 52,56,58. With increasing temperature the high 81

91 energy CT2-nr band and the dominating Frenkel exciton emission strongly decrease. Above 50 K the relaxed charge transfer transitions CT1 between molecules within the same unit cell and relaxed CT2 between stacked molecules in different unit cells become the dominating emission bands. Their temperature decrease is weaker compared to the CT2-nr and the Frenkel exciton luminescence. A more detailed analysis of these bands is shown in Fig for 20, 80 and 300 K where I applied model calculations that were developed for PTCDA single crystals. The asymmetric Frenkel exciton emission including vibronic progressions of external and internal vibrational modes is reconstructed by six normalized Gaussian functions 52,56,58 : (measurements were taken in wavelength scale and converted to the energy scale, see Appendix A for details) I PL 2 3 a j 1 ω ω j ( h ω) = ω exp (2.23) j σ j 2π 2 σ j Compared to the PL investigations on bulk PTCDA crystals the peak energies h ω j of the Gaussian lines in PTCDA films have been shifted by 26 mev to lower energies while the full width half maxima 8ln 2hσ j as well as the areas of the Gaussian functions a j were not changed. The temperature dependence of the Gaussian broadening σ j was considered using the relation 52,56,58, σ hω h h (2.24) 2 ( ) int ( ) ext ωint coth αext hωext coth 2kBT 2kBT j = σ 0 j + α ω int 82

92 where the different FWHM of the Gaussian functions σ 0 j are given by 69, 92, 104, 123, 158 mev 69. The values of the reorganization energy and frequency of internal modes 2 α int = 0.29, ω int = 233 cm -1 were taken from Refs. 14,52,57,58 and values of the external wavelength [nm] K CT2-nr Frenkel CT1 CT2 excimer PL intensity [normalized] 80 K 300 K monomer energy [ev] Figure 2.35: PL spectra of a 36 nm thick PTCDA film on Si(001) excited at 2.33 ev obtained at temperatures 20, 80 and 300 K. The experimental data is shown as open circles. Also shown are results of model calculations. The individual emission channels are labeled, the resulting calculated PL spectrum is shown as full line. 83

93 2 modes α ext = 7.5 and ω ext = 50 cm -1 were taken from Ref. 22,68. The used parameters for the Frenkel exciton emission are summarized in Table 2.3. Emission Channels j Peak energy [ev] FWHM [mev] Area Blueshift [mev/k] Frenkel Exciton x CT x CT x Excimer CT2-nr x10-2 Monomer Table 2.3: Parameters for all six emission channels. For each Gaussian function the peak energies hω j in the limit T 0 K, the full width half maximum (FWHM) 8 ln 2hσ j, its relative area a j and temperature blueshift is given. 84

94 Each of the PL bands related to self-trapped excitons (CT1, CT2, and excimer) is modeled as a sum of normalized Gaussians with positions areas ω j, broadenings σ j, and a j as given in equation (2.23). However, the spacing between subsequent positions ω j now corresponds to one effective vibrational mode energy ω eff h 160 mev and the ratio of the areas a j of the vibronic subbands are described according to Poisson distributions 14,57,58. As for the Frenkel exciton emission the Gaussian center energies h ω j have been shifted to lower energies by 26 mev while the other parameters σ j and a j were kept same as for the PTCDA single crystals. The temperature dependent broadenings σ j (T ) for the self trapped excitons are free fitting parameters that allow interpolating the PL line shape between different temperatures. Due to thermal expansion and an increased occupation of higher vibrational subbands all bands shift to higher energies with rising temperature except for the excimer transition that shows no noticeable temperature shift. In these calculations I used the same blueshifts for the Frenkel exciton and CT2 emission as given in Ref. 22 but doubled the blueshift of the CT1 emission in order to achieve a better fitting with the experimental data. Finally the high energy CT2-nr transition and an additional weak high energy band that occurs at high temperatures and is attributed to a relaxed monomer transition were considered by a single Gaussian function with transition energies h = and 2.05 ev. The broadenings were set to 8ln 2 hσ j = 100 mev for both lines. A blueshift ω j of 6 x 10-2 mev/k is introduced to the ev band. All used parameters for the self 85

95 trapped exciton emissions and high energy transitions are summarized in Table 2.3. The agreement between the experimental results obtained from the PTCDA thin film and the model calculations that are based on earlier investigations on PTCDA crystals are very good demonstrating the high structural and optical quality of these PTCDA films. The good structural quality is also supported by X-ray measurements depicted in Fig (a) that show a completely symmetric α-ptcda (102) reflex with an α/β-ptcda ratio of better 40:1 and a FWHM of 0.29 o for the α-ptcda reflex occurring at 2θ = o a) diffraction intensity b) diffraction angle 2θ Fig. 2.36: X-ray diffraction spectra a) on a 70 nm thick PTCDA film and b) on a 6x [PTCDA 3nm /Alq 3 4nm] multilayer. Both structures were grown on Si(001). The solid and dashed lines show Gaussian fits for α and β-ptcda (102) reflexes, respectively. 86

96 PTCDA/Alq 3 Multilayers Fig shows the PL spectra of a PTCDA/Alq 3 multilayer structure grown on Si (001) comprising six alternating PTCDA and Alq 3 layers of 3 nm and 4 nm thicknesses, respectively. The two PL spectra were recorded at 10 K at different excitation energies (2.84 ev and 2.33 ev) as indicated by vertical arrows. The absorption spectrum of the wavelength [nm] K PL intensity absorbance energy [ev] Figure 2.37: PL spectra of a 6x [PTCDA 3nm /Alq 3 4nm] multilayer grown on Si(001) excited at two different excitation energies (2.33 and 2.84 ev) indicated as vertical arrows and absorption spectrum of the same multilayer grown on Pyrex recorded at 10 K. The spectra are offset for better comparison. multilayer on Pyrex is also shown. In comparison to the absorption spectrum on pure PTCDA films the absorption above 2.8 ev is slightly enhanced due to the onset of the Alq 3 absorption in these films. Correspondingly the PL spectrum excited at 2.84 ev shows the Alq 3 emission 25,70 peaking at 2.37 ev. The Alq 3 emission is absent when the 87

97 multilayer is excited at 2.33 ev where Alq 3 is transparent. Besides the appearance of the Alq 3 luminescence the PTCDA emission is not affected by changing the laser excitation energies. As for the PTCDA thin film the PL spectra and the absorption curve are offset to each other for better comparison. While the CT2-nr and Frenkel exciton emission lines are not energetically shifted and also have similar PL intensities as compared to the PTCDA film a closer look at the low energy emission shows a more pronounced and red shifted band. This observation becomes more obvious in temperature dependent PL measurements (excited at 2.84 ev) as demonstrated in Fig Above 40 K the low energy peak becomes the governing emission and remains dominant up to 200 K. Frenkel CT1 Alq 3 CT2-nr 10 temperature [K] PL intensity energy [ev] Figure 2.38: Temperature dependent PL of a 6x [PTCDA 3nm /Alq 3 4nm] multilayer on Si(001) excited at 2.84 ev in the range 10 to 300 K in 10 K steps. The contributing emission channels are labeled. 88

98 As for the PTCDA film I applied model calculations on the PTCDA/Alq 3 multilayers for 20, 80 and 300 K (see Fig. 2.39) where the parameters given in Table 2.3. To achieve a satisfying agreement of the simulated data with the experimental data an additional Gaussian line at h = ev, with a FWHM of 100 mev and a blueshift of ω j 4.0x10-2 mev/k had to be introduced. While the relative weight of the CT2 emission channel is suppressed by the introduction of the new line the relative ratios and the overall temperature dependence of the Frenkel exciton and self-trapped exciton emission bands is very similar compared to the pure PTCDA film. wavelength [nm] K CT2-nr Frenkel CT1 CT2 new line PL intensity [normalized] 80 K excimer 300 K monomer energy [ev] Figure 2.39: PL spectra of a 6x [PTCDA 3nm /Alq 3 4nm] multilayer on Si(001) excited at 2.33 ev obtained at temperatures 20, 80 and 300 K. The experimental data is shown as open circles. Also shown are results of model calculations. The individual emission channels are labeled, the resulting calculated PL spectrum is shown as solid line. 89

99 The appearance of the new channel in the PTCDA/Alq 3 multilayers can have different reasons: (a) The low energy emission could be due to an interface transition at the multilayer interface as indicated by XPS measurements 30 that predict an indirect LUMO- HOMO transition at 1.85 ev (see Fig. 2.40). Additional exciton binding energies and/or self-trapping energies could lead to a transition around 1.85 ev. PTCDA Alq ev LUMO 2.2 ev 0.35 ev 1.85 ev 2.7 ev HOMO Figure 2.40: HOMO-LUMO offsets of a PTCDA/Alq 3 interface obtained from XPS measurements. The dash line indicates a possible indirect energy transfer at the interface. (b) The emission could be due to the reabsorption of the Alq 3 emission at the lowest PTCDA absorption subband according to observations of a low energy Y-band where a PTCDA film was excited at 2.2 ev 67,68. (c) The low energy emission could be explained by structural changes and strain effects within the crystalline PTCDA layers that are grown on amorphous Alq 3 layers. 90

100 To decide which one of the above explanations is appropriate additional PL measurements are performed on three different samples that were grown on Si (001). The proposed sample structures are as follows. The first sample (A) is a pure PTCDA film of 10 nm thickness, the second (B) is a 10 nm PTCDA thin film with a 2 nm thick Alq 3 layer on top and the third (C) is a 10 nm thick PTCDA film which was grown on a 2 nm thick Alq 3 layer. The obtained normalized PL spectra performed at 10 K are shown in Fig While there is no difference in the emission from sample A and B, it shows a clear appearance of a new line in sample C. As observed in the multilayer structures this new line becomes the dominant emission band above 40 K (not shown here) while the temperature dependence of sample B is equal to the pure PTCDA film (sample A). Since we have introduced a PTCDA/Alq 3 interface in samples B and C a PTCDA/Alq 3 wavelength [nm] PL intensity [normalized] sample A sample B sample C 10 K energy [ev] Figure 2.41: Normalized PL spectra of a 10 nm PTCDA film on Si(001) (sample A), a 2 nm Alq 3 layer on a 10 nm PTCDA film on Si (001) (sample B) and a 10 nm PTCDA film on a 2 nm Alq 3 layer on Si (001) (sample C) and excited at 2.84 ev and recorded at 10 K. The spectra are offset for better comparison. 91

101 interface transition can not be the reason for the low energy emission line. Furthermore, since reabsorption from the Alq 3 film occurs in sample B and C this process is also not responsible for the occurrence of the low energy emission. Therefore it can be concluded that structural changes within the PTCDA films are the microscopic origins for the appearance of the new emission band at 1.63 ev. For a further confirmation of this interpretation we performed strain dependent PL measurements on a 36 nm thick PTCDA film at 10 K using a pressure cell that produces a uniaxial strain up to about 1 kbar along the film growth direction. The applied strain has been estimated from PL experiments on sublimed PTCDA single crystals using a cryogenic diamond-anvil cell 60. As demonstrated in Fig the low energy CT2 PL intensity [normalized] wavelength [nm] no pressure P 1 P 2 10 K energy [ev] Figure 2.42: Normalized PL spectra of a 36 nm PTCDA film on Si(001) without and with applied uniaxial strains P 1 and P 2 (P 2 > P 1 ). The layer was excited at 2.33 ev and spectra were recorded at 10 K. 92

102 emission slightly shifts to lower energy and increases its intensity when uniaxial strain P 1 (about 0.5 kbar) and higher strain P 2 (about 1kbar) is applied to the PTCDA film. The weaker redshift of the CT2 emission compared to the emission in PTCDA/Alq 3 multilayers is attributed to strain inhomogeneities and missing inplane strain components in these measurements. Finally, Fig shows the PL spectra of the 32 nm PTCDA film at 10, 40 and 80 K that were recorded at strain P 1. As in the PTCDA/Alq 3 multilayers the low energy emission gains intensity relative to the Frenkel exciton emission when the temperature is raised and becomes the dominating band above 40 K, confirming the interpretation of a strain modified CT2 emission. More detailed pressure dependent measurement on PTCDA films will be presented in section wavelength [nm] K 40 K 80 K pressure P 1 PL intensity energy [ev] Figure 2.43: Temperature dependent PL spectra of a 36 nm PTCDA film on Si(001) with applied uniaxial strain, P 1. The layer was excited at 2.33 ev and the spectra were recorded at 10, 40 and 80 K. 93

103 2.4.3 PTCDA Crystals, Thin Films and PTCDA/Alq 3 Monolayers In a further investigation 66, the studies were extended to even thinner PTCDA/Alq 3 multilayers and to co-deposited films using temperature dependent PL spectroscopy, X-ray diffraction and Fourier transform infrared (FTIR) spectroscopy. Figure 2.44 shows the PL spectra at 20, 40 and 80 K of PTCDA single crystals, of a 36 wavelength [nm] K Multi-12 Multi-6 PL intensity [normalized] 40K 20K CT2 Frenkel CT1 36nm film crystal CT2-nr Figure 2.44: Normalized PL spectra of α PTCDA single crystals grown by gradient sublimation, a 36 nm thick PTCDA film, a 6x [PTCDA 3 nm /Alq 3 4 nm] multilayer (Multi-6) and a 12x [PTCDA 1.5 nm /Alq 3 2 nm] (Multi-12). The PTCDA film and multilayers were grown by OMBD on oxide covered Si(001) substrates. The samples were excited at 2.33 ev. The PL measurements were performed at 20, 40 and 80 K energy [ev] 94

104 nm thick PTCDA film and of two different multilayers grown on Si. Multi-6 denotes a multilayer structure composed of 6 alternating layers of 3 nm PTCDA and of 4 nm Alq 3 thin films. Multi-12 is a structure composed of 12 alternating layers of PTCDA with 1.5 nm thickness and of 2 nm thick Alq 3 layers. The PL spectra are normalized with respect to the Frenkel exciton emission. The overall thickness of the multilayers as well as the PTCDA and Alq 3 contribution is kept constant. Changing the individual PTCDA and Alq 3 layer thickness composed of 6 alternating layers of 3 nm PTCDA and of 4 nm Alq 3 thin films. Multi-12 is a structure composed of 12 alternating layers of PTCDA with 1.5 nm thickness and of 2 nm thick Alq 3 layers. The PL spectra are normalized with respect to the Frenkel exciton emission. The overall thickness of the multilayers as well as the PTCDA and Alq 3 contribution is kept constant. Changing the individual PTCDA and Alq 3 layer thickness therefore changes the number of interfaces. As identified in earlier investigations on PTCDA crystals 52,56-58,60,65 and crystalline PTCDA films 65 the most prominent band at 1.82 ev is attributed to the emission of an indirect Frenkel exciton. In addition the charge transfer excitons CT1 and CT2 as well as the non-relaxed CT2 are visible. The PTCDA single crystal and crystalline PTCDA film show very similar temperature dependence: With increasing temperature the Frenkel exciton emission slightly shifts to higher energy which is attributed to a thermal population around the indirect minimum of their k-dispersion 14,57,58,65. In addition the high energy CT2-nr band and the dominating Frenkel exciton emission strongly decrease when the temperature is raised. At 80 K the relaxed charge transfer transitions of CT1 and CT2 become the dominant emission bands. The PTCDA/Alq 3 multilayers show a different behavior: 95

105 While the CT2-nr and Frenkel exciton emission lines have similar PL intensities as compared to the PTCDA film, the low energy band is red-shifted and becomes the dominant emission above 40 K. The red-shift and the dominance of the low energy peak increases with decreasing PTCDA layer thickness. As evidenced by pressure dependent 36 nm film (a) diffraction intensity Multi-6 (b) Multi-12 (c) diffraction angle 2θ Figure 2.45: X-ray diffraction spectra (a) on a 36 nm thick PTCDA film, (b) on a 6x [PTCDA 3 nm /Alq 3 4 nm] multilayer (Multi-6), and (c) on a 12x [PTCDA 1.5 nm /Alq 3 2 nm] multilayer (Multi-12). All films are deposited on oxide covered Si(001). The solid line and dashed lines show Gaussian fits for α and β PTCDA (102) reflexes respectively. 96

106 PL measurements on PTCDA crystals 60 and on crystalline PTCDA layers 65,66 (mentioned above) the low energy peak can be attributed to a strain modified CT2 transition. Compressive strain along the π-orbitals of stacked molecules causes an enhancement of the CT2 exciton binding energy and an increased exciton formation probability explaining the red-shifted and stronger pronounced CT2 emission in the PL spectrum. Accordingly we attribute the observed CT2 emission changes in sample Multi- 6 and Multi-12 to compressive strain fields inside the PTCDA layers. The higher internal strain in Multi-12 compared to sample Multi-6 is attributed to an increase of structural inhomogeneities and higher tilt angles between the PTCDA crystallites within the multilayers. Both are due to the fact that in sample Multi-12 the number of interfaces is larger and the individual PTCDA layer thickness is smaller, making the relaxation of molecules in a planar orientation less probable. This interpretation is again supported by X-ray diffraction measurements on the crystalline PTCDA film and on multilayers. Figure 2.45 (a) shows a completely symmetric α-ptcda (102) reflex with an α/β-ptcda ratio of better than 40:1 and a FWHM of 0.29 o for the α-ptcda reflex occurring at 2θ = 27.8 o. These results demonstrate the high structural quality of this sample. In samples Multi-6 and Multi-12 the FWHM of the α-ptcda (102) reflex is increased to 0.35 o and 0.38 o, respectively. The broadening of the (102) reflex indicates inhomogeneous strain inside the PTCDA crystallites that increases with decreasing PTCDA layer thickness in the multilayer structure. In addition we find an increasing (yet small) contribution of β-ptcda in the multilayer structures with an α/β- ratio of 18:1 and 14:1 in Multi-6 and Multi-12, respectively. 97

107 To investigate the out of plane disorder of PTCDA molecules 14,51,71 in different samples far infrared (FIR) Fourier transform measurements were performed on a 100 nm thick pure PTCDA film, on sample Multi-6 and on a 100 nm thick Alq 3 layer deposited on Si. The FIR absorption spectra of some vibrational modes in these samples are demonstrated in Figure In addition the difference spectrum between the FIR absorption of vibrational modes in Multi-6 and in pure Alq 3 is displayed in Figure This spectrum shows slightly more pronounced out-of-plane vibrational modes at 733 and 809 cm -1 than in the pure PTCDA film indicating an increasing tilt of PTCDA molecules with respect to the (102) plane in Multi-6. 98

108 optical density optical density * PTCDA/Alq 3 minus Alq 3 * PTCDA film PTCDA/Alq 3 Alq wavenumbers [cm -1 ] Figure 2.46: Optical density as a function of wave-numbers of a 100 nm thick PTCDA film, a 100 nm thick Alq 3 film and of a 6x [PTCDA 3 nm /Alq 3 4 nm] multilayer (Multi-6) on oxide covered Si(001), obtained from Fourier transform infrared (FTIR) measurements. The dasheddotted spectrum shows the difference spectrum between the FIR absorption of vibrational modes in the multilayer and in the pure Alq 3 film. 99

109 2.4.4 PTCDA/Alq 3 Co-deposited Layers The modification of emission CT2 due to internal strain and lattice distortion is further observed in the PL spectra of co-deposited PTCDA/Alq 3 layers as displayed in Fig The spectra of a 22 nm thick PTCDA layer (sample A) and of a 20 nm PTCDA film deposited on top of a 2 nm thick layer of Alq 3 (sample B) are compared with 22 nm wavelength [nm] K CO-50 B PL intensity [normalized] 40K 20K CO-10 A energy [ev] Figure 2.47: Normalized PL spectra of a 22 nm thick PTCDA film on oxide covered Si(001) (sample A), a 20 nm PTCDA film that was deposited on a 2 nm thick Alq 3 layer (sample B), a 22 nm thick co-deposited PTCDA/Alq 3 layer with 10 % Alq 3 content (Co-10) and a 22 nm thick co-deposited film with 50% Alq 3 content (Co-50). The samples were excited at 2.33 ev. The PL measurements were performed at 20, 40 and 80 K. 100

110 thick co-deposited samples Co-10 and Co-50 with an Alq 3 content of 10% and 50%, respectively. In samples Co-10, sample B and Co-50 again find a red shift and an enhancement of the CT2 band with respect to the pure PTCDA film. The observed changes of the CT2 emission become more pronounced at elevated temperatures. Interestingly the increase of the CT2 emission is higher in sample B as compared to sample Co-10 although the PTCDA content is the same. Obviously the rough amorphous Alq 3 surface in sample B disturbs the structural quality more effectively than the presence of single Alq 3 molecules that are incorporated into the PTCDA lattice during the growth. The PTCDA crystals in Co-10 that are grown on a smooth SiO 2 surface seem to accept defects due to incorporated Alq 3 molecules quite effectively therefore avoiding significant stress between PTCDA crystallites. In contrast the uneven Alq 3 surface in sample B cause significant structural changes between adjacent PTCDA crystallites leading to compressive strain between stacked PTCDA molecules. However, in sample Co-50 the disturbance of the PTCDA crystalline structure increases. At such high Alq 3 concentration the formation of Alq 3 clusters becomes likely, therefore leading to a reduced PTCDA crystal quality, smaller crystal sizes and higher tilt angles between crystallites. Accordingly the red-shift and peak intensity of the CT2 band gets larger compared to the other samples. However, the PL spectrum of the co-deposited sample at 20 K reveals all characteristic features of a Frenkel exciton and of different charge transfer excitons. This result clearly demonstrates that the PTCDA molecules are not randomly distributed in this layer but tend to form crystallites that are embedded in an Alq 3 matrix. 101

111 Again these interpretations are supported by X-ray diffraction measurements shown in Fig that displays the α-ptcda (102) reflex of Co-10 with an α/β-ptcda ratio of better than 20:1 and a FWHM of 0.31 o. The (102) reflex is only slightly broader than in the pure PTCDA demonstrating the high structural quality of sample Co-10 which is in agreement to the suggestion that diffraction intensity Co-10 (a) B (b) Co-50 incorporated Alq 3 molecules do not significantly disturb the PTCDA crystal lattice. In contrast sample-b shows a significantly broader (102) reflex with 0.36 o FWHM and an α/β-ptcda ratio of 14:1 indicating an increase of structural disorder and inhomogeneous strain inside the PTCDA crystallites. Finally in sample Co-50 the FWHM of the α-ptcda (102) reflex is further increased to 0.39 o. The observation of the (102) reflex in co-deposited Co-50 clearly indicates the presence of PTCDA crystallites that are oriented perpendicular the (102) plane (c) diffaction angle 2θ Figure 2.48: X-ray diffraction spectra (a) on a 88 nm thick co-deposited PTCDA/Alq 3 layer with 10 % Alq 3 content (Co-10), (b) on a 20 nm PTCDA film deposited on top of a 2 nm thick layer of Alq 3 (sample B) and (c) on a 88 nm thick co-deposited film with 50% Alq 3 content (Co-50). All films are deposited on oxide covered Si(001). The solid line and dashed lines show Gaussian fits for α and β PTCDA (102) reflexes, respectively. 102

112 in agreement to the observation of different exciton bands in the PL spectrum. However, the broader (102) reflex indicates a further increase of structural disorder and strain inside the PTCDA crystallites Strain Dependence of PTCDA Crystals The application of high pressure to organic solids is a powerful method for changing their physical properties. In particular, high pressure optical spectroscopy PL Intensity (10 3 cts/sec) Energy (ev) α-ptcda PL 300K x 1/4 8.2kbar 50mW 8.2kbar 5mW 1.3kbar 5mW 1 atm λ(nm ) Figure 2.49: Room temperature PL spectra of crystalline α-ptcda recorded at different pressures up to 8.2 kbar and at two different excitation powers 59. provides a useful tool to examine the changes in intermolecular interactions and dynamics in molecular solids. Due to the relative weakness of their intermolecular forces modest pressures can produce large changes in the electronic and crystal structure of 103

113 these solids. Therefore strain effects in soft organic materials can have great potential to impact novel strain dependent devices. Despite active research on the optical and electronic properties of organic materials the effects of strain and their microscopic mechanisms are in many cases not well investigated or understood. Pressure dependent measurements are also an important tool to support and prove the emission channel assignments that were discussed in earlier sections. We therefore started to investigate the pressure dependence of the PL in PTCDA crystals and thin films (see section 2.4.6). The pressure dependent PL experiments on PTCDA crystals are performed using a ruby-calibrated diamond-anvil cell (DAC) mounted in a liquid-he cryostat; the apparatus allows changing pressure and temperature in the ranges kbar and K. This experiment is carried out at 11 K and 300 K for applied pressures up to 54 kbar. All the measurements were performed at the University of Buffalo, NY in Dr. Weinstein s group. Fig shows PL spectra for PTCDA single crystal at room temperature under different applied pressure as indicated. There is a significant redshift of the low energy PL emission band with increasing applied pressure. As explained in earlier sections, this band is assigned to the excimer transition. At 8.2 kbar a high energy shoulder becomes visible, emerging in the ev region as the main peak shifts more rapidly out of this region. The most likely assignment for this shoulder is the isolated monomer transition observed as a fast decaying (~3 ns) component contributing to the high-energy wing of the 1 atm emission spectrum 58,59,65,71. The monomer transition originates mainly from a PTCDA molecule within a randomized surroundings as e.g. near grain boundaries. Excitons trapped at such sites are highly localized, and, consequently, should be relatively insensitive to compression. This is consistent with the weak pressure 104

114 shift of the shoulder compared to the excimer transition that dominates the main peak. When the exciting laser power is increased from 5 to 50 mw at 8.2 kbar the main emission band saturates relative to the high energy shoulder. Since the ~3 ns recombination lifetime of the monomer transition is much shorter than the lifetime of the excimer transition, which is found to be ~20 ns 52,58 a saturation effect is more likely for the latter exciton transition. This provides further support for our assignments of the α- PTCDA emission bands at room temperature. The effect on the applied pressure of PTCDA at low temperature is shown in Fig As already discussed in earlier sections at 1 atm, the most prominent band (1.82 ev) is attributed to the emission of indirect Frenkel exciton. There is another unresolved 5 Energy(eV) x 4 α - P T C D A P L T = 1 1 K k b a r PL Intensity (cts/sec x 10 4 ) x 2 x 4 x k b a r k b a r k b a r k b a r 0 CT2-nr CT1 Frenkel CT2 2.7 k b a r 1 atm λ (n m ) Figure 2.50: PL spectra of crystalline α-ptcda measured at 11 K for several pressures up to 54.2 kbar. The excitation power was 5 mw. Different exciton emissions are labelled

115 exciton component associated with the band at 1.82 ev, is called a relaxed charge transfer exciton (CT1) transition between molecules within the same unit cell. The band at 1.68 ev is assigned as CT2, a relaxed charge transfer exciton transition between stacked molecules in different unit cells in PTCDA crystal. Finally the high energy band at 1.95 ev was assigned to non relaxed CT2 exciton transition. Each band of the spectrum undergoes a redshift with increasing applied pressure at 11 K as shown in Fig Furthermore, each band becomes reasonably boarder and the CT2 transition shows significant enhancement of the intensity compared to the Frenkel exciton with increasing applied pressure. The redshift is caused by the broadening of the π π * transition band due to increased intermolecular overlap. A detailed evaluation of the redshift of Frenkel, CT2 and CT2-nr bands is given elsewhere 60. From these evaluations it is worth noting that the three PL transitions in Fig exhibit considerable differences in the quadratic component of the pressure shift that occurs due to lattice stiffening effects 60,64. This component is largest for the CT2-nr transition, smaller for the Frenkel exciton, and not measurable within uncertainty for the CT2 transition. The absence of the quadratic pressure shift in the case of the CT2 transition is attributed to a shift in the relaxed configuration under applied pressure that compensates for the lattice stiffening effect. The differences in the quadratic components demonstrate the occurrence of different emission channels which is in agreement to time resolved PL measurements 60. Additional experiments, particularly time-resolved PL studies under applied pressure, and theoretical work are needed to clarify these issues. 106

116 2.4.6 Strain Dependence of PTCDA Thin Films A study of uniaxial pressure dependent photoluminescence of PTCDA thin film on Si(001) along the stacking direction in the temperature range from 20 to 80 K is presented in this section. The pressure measurements were carried out using a home made uniaxial pressure cell (see Fig. 2.23). wavelength [nm] T= 80K PL Intensity T = 40K 0 kbar 0.3 kbar 0.5 kbar 0.7 kbar 1 kbar T=20K energy [ev] Figure 2.51: Uniaxial pressure dependent PL spectra of a 90 nm PTCDA on Si(001) excited at 2.33 ev obtained at 20, 40, and 80 K, the pressure values are given in kbar for different spectrum. 107

117 The pressure dependent PL spectra of the 90 nm thick PTCDA film at temperatures 20, 40 and 80 K are displayed in Fig At all temperatures we find a general quenching up to 25 % of the PL intensity with increasing pressure. The reduction in PL intensity is partly attributed to the creation of defects using the pressure cell. Another contribution is due to emission band broadening effects caused by strain inhomogeneity in the PTCDA film accordingly the surface roughness of the sapphire window and the pressing piston. However, at 20 K the CT2 band intensity only decreases by a factor of 0.9 at the highest applied pressure of ~1 kbar while the Frenkel exciton emission is quenched by 30%. In addition the CT2 transition and the Frenkel exciton emission reveal a slight shift to lower energies by a few mev. Also the non-relaxed CT1 transition shows a weak strain dependence. Fig also shows the pressure dependent PTCDA emission at 40 K. At this temperature the Frenkel exciton emission is already decreased due to thermal activation into self-trapped exciton states. Correspondingly, the CT2 becomes the dominant emission channel above a pressure of 0.3 kbar. Again there is an overall PL intensity reduction with increasing strain but the CT2 emission is much less affected than the Frenkel exciton emission. The CT2 emission is redshifted by ~5 mev at the highest applied pressure. At 80 K the CT2 transition is the dominant emission band at atmospheric pressure. Since the Frenkel exciton emission is already strongly reduced at this temperature 52,58,65,66 the high energy band is now predominantly attributed to a CT1 transition 52, Both CT bands exhibit decrease in the emission intensity, however, the relative ratio of the pressure dependent quenching is less than between the CT2 and Frenkel exciton emission at lower temperature. The CT1 and CT2 transition a clearly 108

118 resolved redshift of ~5 mev at the highest applied pressure compared to atmospheric pressure. The predominant emission channel in PTCDA thin films at a temperature higher than 200 K 65,66 is the excimer transition. Figure 2.52 shows the pressure dependent PL of the sample at 300 K. The PL spectrum at atmospheric pressure is significantly broader revealing peak energy at ev. As in the low-temperature experiments the PL intensity of the excimer transition decreases with increasing uniaxial pressure. When the wavelength [nm] PL intensity T = 300K 0 kbar 0.3 kbar 0.5 kbar 0.7 kbar 1 kbar 1.3 kbar 0 kbar energy [ev] Figure 2.52: Uniaxial pressure dependent PL spectra of a 90 nm PTCDA on Si(001) excited at 2.33 ev obtained at 300 K. The PL intensity is increased when applied pressure is changed from 0 to 1.3 kbar. applied pressure reaches ~1 kbar, the PL intensity is about a factor 0.7 lower than without applied pressure. The pressure dependent shift of the peak energy to lower energy is little stronger for the excimer transition as compared to the CT transitions. At the largest applied pressure of ~1.3 kbar the peak energy is ev which corresponds to a ~7 mev shift to the lower energy compared to the maximum emission at atmospheric 109

119 pressure. Furthermore, these features are reversible to a reasonable amount indicating that the original PL spectrum is recovered up to 90% from the original intensity when the applied pressure is released. The blue solid line of Fig shows the PL spectrum of the PTCDA thin film after releasing the applied pressure. Both the enhancement of the CT exciton and excimer emission and the redshift of the CT and excimer bands are qualitatively attributed to an increased exciton trapping probability and by a higher exciton binding energy due to the reduced intermolecular distance along the 102 direction. In order to connect the experimental observations with recent theoretical models for excimer and CT exciton transitions 54,59, and for future exploitation of strain dependent effects in technological aspects, it is crucial to know the compliance tensor elements for PTCDA films along the soft a - axis. The compliance tensor relates the applied stress to macroscopic strain, which, in turn, corresponds to length and/or angle changes within the unit cell. In collaboration with Dr. Weinstein, we started to investigate pressure dependent changes in the dimensions of a single PTCDA crystal under hydrostatic pressure in a diamond anvil cell using microphotography. An evaluation of these measurements and comparison with recent calculation by Scholz et al. 72 is presently in preparation. 110

120 Chapter 3 3. Characterizations of Alq 3 Layers and Alq 3 Based Light Emitting Diodes Organic light emitting diodes (OLEDs) have received much interest due to their potential applications in display technology 1,73-78 and solid state lighting 16. OLEDs have been constructed using a variety of active electroluminescent materials like low molecular weight organic materials and polymers. Recently, considerable effort has been undertaken to develop more power efficient devices based on π-conjugated small organic materials which can be driven by a low dc voltage. A better device performance can be achieved by improving the carrier injection efficiency from the electrodes and by obtaining balanced electron-hole recombination within the recombination zone 1,76,78. The charge carrier injection characteristics of organic molecular LEDs have been explained using a band-structure model 76. The operation of OLEDs is based on the injection of negative charge carriers (electrons) and positive charge carriers (holes) from negative and positive electrodes, respectively. The charge injection from the electrodes requires that the charge carries surmount or tunnel through the barriers at the interfaces of the cathode and the anode contacts 79. Tris (8-hydroxy) quinoline aluminum (Alq 3 ) has attracted much attention as active medium in OLED s 1,3,10,73-79 which is used as both electron transport layer (ETL) and emissive layer. Up to now, the reported photoluminescence (PL) efficiency ( η PL ) of Alq 3 thin films is around 30 % 5, According to the spin multiplicity 33, for light weight elements like in Alq 3 molecule, generally possible electroluminescence external quantum 111

121 efficiency η EL is 25% 10,33. Since, the reported value is still much less ( η EL ~ 1%) 84,85 for Alq 3 based OLEDs, there is a large opportunity for further improvement. It has been shown recently that the current-voltage (I-V) characteristics and the electroluminescence (EL) of OLEDs can be explained by the injection of charge carriers into the emissive layer that possesses a large density of traps above the HOMO level 79, However, there is a lack of quantitative discussion regarding the temperature dependent device performance of OLEDs 79,88,89. In this work, I present temperature dependent I-V and EL characteristics in comparison with temperature dependent PL of Alq 3 thin films. Furthermore, I try to describe the physics of an operating multilayer molecular device, which is well suited to illustrate the rich variety of physical phenomena that occur in the organic LEDs. 112

122 3.1 Fabrication of Organic Light Emitting Diodes This section describes different procedures and materials that are involved in the fabrication of organic light emitting diodes (OLEDs), in particular a two-layer OLED structure which can be driven at a voltage below 10 V and a second multilayer structure device with various materials chosen to improve the charge carrier injection ITO Substrate and Pattering A glass substrate (Corning 1737) 2 diameter was cleaned in an ultrasonic bath first using acetone then methanol and subsequently the substrate was rinsed in de-ionized water. Then a thin indium tin oxide (ITO) layer was deposited ( Å) onto the glass substrate using a sputtering system at UC nanolab. The sheet resistance of the sputtered ITO substrate is 60 Ω/. To pattern the ITO, a negative photo-resist (Think & Tinker Ltd MI151 dry-resist) was applied to the substrate and then exposed to UV light for 2-3 minutes through a Cr coated mask. The UV ITO ITO Active area Glass Glass (a) (b) Figure 3.1: The photographs of patterned ITO substrate for (a) a two-layer structure with two possible active devices (produced by our laboratory) and (b) a multilayer structure with four possible active devices (produced by the UC Nanolab). 113

123 exposed resist was then developed using a 1% K 2 CO 3 solution for 1-2 minutes. It was rinsed in de-ionized water, and the ITO was etched using a 25% HCl bath at a rate of 8 Å/s. A NaOH stripper was used to take off the photo-resist from the substrate. Finally, the substrate was rinsed in de-ionized water and the initial cleaning steps were repeated. The described procedure was developed in Dr. Steckl s group at the UC Nanolab. Fig. 3.1 shows the photographs of the patterned ITO substrate for a two-layer (a) and a multilayer (b) structure. The patterned substrate for the two-layer structure is designed (our laboratory) to fabricate two possible active devices, while the substrate for the multilayer structure is engineered (the UC Nanolab) to produce four possible active devices Organic Materials Alq 3 is used as emissive and electron transport layer for all OLED structures investigated in this work. The chemical structure of Alq 3 and the structural properties are discussed in Chapter 2. The chemical structures of other organic materials are given in Fig TPD (N,N - diphenyl-n,n -bis(3-methylphenyl)1-1 -biphenyl-4-4 -diamine) and α-npb (N,N'- Bis(naphthalen-1-yl)-N,N'-bis(phenyl)benzidine) are used as hole transport materials. The chemical structures of TPD and α-npb are shown in Fig. 3.2 (a) and (b) respectively. PEDOT (Poly-3,4-Ethylenedioxythiophene), a conducting polymer doped with PPS (poly(styrenesulfonate)) (Baytron-P from H. C. Stark), is used to improve the hole injection into the hole transport material. The chemical structure of PEDOT: PSS is given in Fig. 3.2 (c). 114

124 (a) TPD (b) α-npb (c) PEDOT:PSS Figure 3.2: The chemical structures of (a) TPD, (b) α-npb, the hole transport materials, and (c) PEDOT: PSS, a conducting polymer solution. 115

125 3.1.3 Organic Layers and Metal Deposition The chemically cleaned patterned ITO substrate was transferred into the OMBD system at a base pressure of 5x10-8 mbar. First TPD and then Alq 3 were thermally evaporated with deposition rates of 0.4 nm/s and 0.5 nm/s, respectively. The two organic layer deposited ITO substrate is then briefly removed to apply a shadow mask (aluminum) which defines active device areas of 3 3 mm 2 each. Finally, the substrate is reinserted into the OMBD system and Mg-Ag (9:1) was evaporated at 0.7 nm/s onto the substrate. The device structure with layer dimensions is illustrated in Fig The electrical and optical properties of the device are presented in section Mg-Ag (100 nm) Alq 3 (40 nm) TPD (60 nm) ITO (60 nm) Glass Substrate (1 mm) Figure 3.3: Configuration of the first Alq 3 based two-layer OLED structure produced in our laboratory, individual layer thicknesses are given in parenthesis. According to the HOMO - LUMO energy levels (see Fig. 3.8), an Al/LiF cathode is a better choice than the Mg-Ag alloy because of its lower work function. Furthermore a PEDOT conducting layer has been introduced on top of the ITO substrate. The procedure for the improved multilayer OLED structure that was fabricated in Dr. Steckl s group at the UC Nanolab is given below and the device structure is shown in Fig The chemically cleaned substrate (using acetone and methanol in an ultrasonic bath for 30 minutes each) is rinsed in de- 116

126 ionized water and dried in an oven. Next, a buffer layer of PEDOT:PSS is spin-coated onto the patterned ITO glass substrate. The PEDOT:PSS solution is diluted, filtered and spin-coated at a speed of 2000 rpm. Afterward, the film is hard baked at 150 C for 15 minutes, resulting in a film thickness of ~ 30 nm. The spin-coated wafer is then rapidly (to avoid oxidation) transferred into the OMBD system with a base pressure of 10-8 mbar. TPD and Alq 3 are then sequentially thermally evaporated with a deposition rate of 0.4 nm/s and 0.5 nm/s, respectively. LiF is evaporated next, at a rate of 0.1 nm/s. The wafer with organic layers is then briefly removed to apply the Al cathode shadow mask which results in active device areas of 2 2 mm 2 each. Finally, the substrate is reinserted into the OMBD system and Al is evaporated at 0.7 nm/s. The completed Alq 3 device has the final structure: ITO (60nm) / PEDOT:PSS (30nm) /TPD (60nm) /Alq 3 (40nm) /LiF (0.6nm) /Al (150nm). The devices were transported to our optical lab in a nitrogen (to avoid ambient exposure) gas container to perform temperature dependent I-V and temperature dependent EL measurements. The electro-optical results are presented and discussed in section and LiF (0.6 nm)/ Al (150 nm) Alq 3 (40 nm) TPD (60 nm) PEDOT (30 nm) ITO (60 nm) Glass Substrate (1 mm) Figure 3.4: Configuration of the improved Alq 3 based OLED device structure (produced by the UC Nanolab), individual layer thicknesses are given in parenthesis. 117

127 3. 2 Electro-Optical Properties of OLEDs Electroluminescence (EL) and photoluminescence (PL) are related physical phenomena. An OLED emits light as EL from an electroluminescent emissive material under the action of electric current passing through the structure. The light is produced by recombination of electrons and holes that have been injected at negative and positive electrodes respectively. The carrier injection and the transport mechanisms are very important for the characterization of the electro-optical properties of OLEDs. The carrier injection and the transport are described by inter-site hopping of the electrons and the holes between localized states inside the organic layers of the OLEDs Cathode ETL/EL HTL Anode Light Figure 3.5: A simple OLED structure, the OLED emits the light under a forward biased condition. Fig. 3.5 shows the basic functions of two layer OLED structure. Typically an indium tin oxide (ITO) layer on a glass substrate is used as a transparent conducting anode. The cathode is a low work function (WF) metal such as Ca, Mg/Ag, Al or Al/LiF. The organic material deposited on the anode should be a good hole transport layer (HTL). Similarly, the organic material that is 118

128 in contact with the cathode should be a good quality electron transport layer (ETL). Due to the disorder of organic solids, electron and hole mobility are several orders of magnitude lower than those in inorganic semiconductors. Furthermore in OLEDs, the hole mobility in the HTL is usually higher compared to the electron mobility in the ETL. Therefore, electron and hole recombination often takes place inside or near the ETL. In samples that are investigated in this work, Alq 3 serves both as the emissive layer (EL) and as ETL. The organic materials TPD (N,N -diphenyl-n,n -bis(3-methylphenyl)1-1 - biphenyl-4-4 -diamine) or α-npb (N,N -Bis(napthalen-1-yl)-N,N -bis(phenyl)benzidine) were used as the HTL. As cathode materials, a Mg-Ag alloy and an Al/LiF bilayer were tested in two different OLED structures Charge Carrier Injection and Transport in OLEDs The molecular materials used for OLEDs behave as an electrical insulator under the action of low electric field (<10 4 V/cm). The typical resistivity of vacuum sublimed organic films is in the order of Ωcm indicating that purified organic materials have intrinsically no net charge carriers without chemical doping of donor or acceptor molecules 33. Despite this fact, in OLEDs a high current density (1 A/cm 2 ) can be achieved 33. Therefore it is important to understand the mechanism of charge carrier injections and principles of current flow in organic semiconductors. The charge carrier injection from electrodes can be explained by Schottky thermal emission and Fowler-Nordheim tunneling 33,87 which is a quantum mechanical process. The current flow in OLEDs can be described by space charge limited current and (SCLC) 86 and trap charge limited current (TCLC) 86,

129 There is a triangle like barrier 33 for both positive carriers (holes) and electrons as illustrated in Fig Furthermore, Fig. 3.7 shows the basic carrier injection and transport LUMO e - Metal Cathode E F HTL ETL E F ITO Anode h + HOMO Figure 3.6: Energy level diagrams for a basic organic light emitting diode structure. The radiative recombination takes place at or near the interface of two organic layers. mechanism in the ITO/TPD/Alq 3 /Mg-Ag OLED structure. The organic materials act as carrier transport layers for both electrons and holes. At a typical Alq 3 /Mg-Ag interface, the energy barrier is around 1 ev which is considerably higher compared to the activation energy (< 0.1 ev) in typical OLEDs. To achieve a large current density as mentioned earlier, Schottky-type carrier injection is possible with thermal assistance via localized levels induced by structural disorder or impurities. The other possible charge carrier injection is by Fowler Nordheim tunneling with the assistance of a local high electric field ( V/cm) at the triangle energy barrier (see Fig. 3.6). 120

130 Glass Anode (ITO) HTL (TPD) LUMO ETL/EL (Alq 3 ) e - e - Cathode (Mg-Ag) LUMO Light HOMO h + h + h + HOMO WF Figure 3.7: The basic operation of the ITO/TPD/Alq 3 /Mg-Ag OLED structure with the carrier injection mechanism for a forward bias, radiative recombination takes place at the Alq 3 /TPD interface. In the regime of low carrier injection (small applied electric field), the current is determined by Ohm s Law. When the injected current density becomes higher than the intrinsic current density, the injected charge carriers form space charges near the organic/electrode interfaces due to the low carrier mobility. An additional electric field inside the organic layers is induced by these excess charges. This leads to an enhanced total internal electric field and the result is a high current density. This is known as space charge limited current (SCLC) and the current density J can be described by (without traps) 86, 2 9εµV J SCLC = d where, ε is the permittivity and d is the thickness of the organic layer and µ is the charge carrier mobility. The steady state current flow through organic materials is further limited by the space trap distribution 79, As the voltage is further increased, the quasi-fermi level in Alq 3 is pulled above the energy of trap levels in the band gap towards the LUMO, resulting in increased 121

131 charge injection. The current density is now modified from the SCLC to the trap charge limited current (TCLC). Assume a continuous exponential energy distribution of traps N t (E) 88, inside the energy gap as (bellow the LUMO), t ( N / kt ) exp[ ( E E ) kt ] N ( E) = / 3.2 t t LUMO Here, N t is the total trap density, E LUMO is the Alq 3 LUMO energy, andt t = E k, where t t / E t is the characteristic trap energy and k is the Boltzmann constant. The current density is then given by 79,86-89, J TCLC m ( m+ 1) ( m+ 1) (1 m) ε m 2m + 1 V = N LUMO q (2m+ 1) N t ( m + 1) m + 1 d µ 3.3 where, m = E / t kt, T is the temperature, and N LUMO is the density of states at the LUMO. The validity of this equation has been demonstrated in the literature for organic thin films 79, Energy Level Diagrams in Various OLED Structures The charge carrier injection characteristics and transport mechanism of organic molecule based LEDs can be explained using a band-structure model 76. Therefore, the HOMO-LUMO energy level position of the organic layers and work function (WF) of the cathode are important parameters of realizing efficient OLEDs. Fig. 3.8 shows the energy level diagram for selected organic materials and metallic cathodes. HOMO-LUMO values are taken from the references given here (ITO 90,91, TPD 91-93, α-npb 90, CuPc 92, PTCDA 30, Alq , Ca 94, LiF-Al 90-92,95, Al 94, Ag 96, Ma- Ag 90,94 and Cu 94 ). HOMO levels have been determined by ultraviolet photoemission spectroscopy (UPS) or by X-ray photoelectron spectroscopy (XPS) measurements. Then LUMO levels were determined according to the optical absorption band measurements. 122

132 2.2 ev 2.6 ev 3.1 ev 3.1 ev Ca 2.9 ev LiF/Al 3.1 ev TPD Mg:Ag 3.7 ev 4.7 ev α-npb CuPc 4.0 ev Alq 3 Al 4.2 ev Ag 4.3 ev Cu 4.6 ev ITO 4.8 ev PTCDA 5.4 ev 5.7 ev 6.1 ev 5.7 ev Figure 3.8: HOMO LUMO energy levels for the organic materials and work functions for selected cathode materials and for ITO anode, relative to the vacuum. 123

133 3.2.3 The Efficiency of OLEDs The device performance is typical characterized by the external quantum efficiency of the device however the quantity of practical significance is luminescence efficiency. The used parameters are defined as follows: The external quantum efficiency ( η ext ) of the device is defined as the ratio number of photons emitted by the OLED to the number of electrons injected. The internal quantum efficiency ( η int ) is defined by the number of photons produced within a device divided by the number of injected electrons. η ext can be expressed as 33, η = η 3.4 ext intη c where ηc is the out-coupling efficiency 33 and related to the refractive index (n) of the emissive layer as 33, η = 1 c 2 2n 3.5 The photocurrent of the OLED is measured using the photodetector. If I ( λ) is the incremental (between wavelength λ and λ+dλ) photocurrent generated in the photodetector by the OLED power, P (λ OLED ) emitted at the center wavelength λ, I det ( λ) can be written as 97, I ( λ) = R( λ) fp ( λ) 3.6 det Where, R (λ) is the incremental photodiode responsivity and f < 1, is the fraction of light emitted that couples into the detector. Then the external quantum efficiency can be written as, OLED det η ext = q hcfi λi OLED det ( λ) dλ R( λ) dλ

134 where, I OLED is the OLED current, q is the electronic charge, h is Plank s constant and c is speed of the light in vacuum. The luminous efficiency, η L (cd/a) which is practically useful in display applications, is defined as 33,77,97, AL η L = 3.8 I OLED where, L is the luminance (brightness) of the OLED in cd/m 2 and A is the active device area. The calculation of η ext and related information for the set-up used are given in Appendix B. Furthermore, a detailed discussion regarding efficiency of OLED can be found in Forrest et al

135 3.3 Experimental Setups Current-Voltage-Luminescence Measurements For the room temperature measurements, the forward biased current-voltageluminescence (I-V-L) characteristics for the Alq 3 device are obtained with an HP-6634B DC power source controlled by a LabView program. The spectrally integrated electroluminescence is obtained with a Minolta CS-100 luminosity meter in the forward direction through the transparent glass substrate as shown in Fig HP-6634B DC power source Interface Computer - + GPIB Minolta CS-100 Anode Organics Cathode Figure 3.9: Experimental setup for I-V characteristics and luminescence measurements. 126

136 For variable temperature measurements between 10 and 320 K a closed-cycle He cryostat (CTI-Cryogenics) and a Lakeshore temperature controller were used. Also a photodiode was installed inside the cryostat as shown Fig Temperature dependent I-V-L measurements were controlled and recorded by using a Lab View program. This experimental setup was used in collaboration with Dr. Klotzkin s group at the department of ECECS in UC. Computer KEITHLEY Source-meter - + Interface GPIB Multimeter 1 kω Voltage source 5V Photo detector Cryostat OLED Figure 3.10: Experimental setup for temperature dependent I-V-L measurements. 127

137 3.3.2 Electroluminescence Measurements In the electroluminescence measurements, the OLED structure was biased by a constant forward voltage of 8.5 V using a DC power supply. The EL spectra were analyzed by a grating monochromator (HR 320, ISA) and a GaAs photomultiplier tube (PMT). For variable temperature measurements between 10 and 300 K a closed-cycle He cryostat (CTI-Cryogenics) and a Lakeshore temperature controller were used. The emitted light from the OLED passes through the transparent ITO substrate and is collected by lenses as illustrated in the Fig DC Power supply - + Sample in the cryostat Anode Monochromator Cathode Organics GaAs PMT Computer Figure 3.11: Experimental setup for electroluminescence measurements. 128

138 3.4 Experimental Studies Absorption and PL of Alq 3 Single Molecules, Polycrystals and Thin Films As demonstrated in Chapter 2, the absorption and emission spectra of PTCDA films show vibronic progressions, which are in particular well resolved in PTCDA single molecules dissolved in CH 2 Cl 21 2 or dimethyl sulfoxide (DMSO) 22. No vibronic sub-bands are observed in the Alq 3 molecules dissolved in CH 2 Cl 2 neither in the PL (see Fig. 3.12) nor in optical absorption wavelength [nm] normalized PL intensity energy [ev] Figure 3.12: PL spectra at 300 and 4.2 K for various Alq 3 system including solution, a thin film (50 nm) onto Si(001) and different polycrystalline samples of the α- and β-alq 3 and calthrated Alq 3 - (C 6 H 5 Cl) 1/2 and Alq 3 (MeOH). The figure is taken from Brikmann et al 43, and the scale is converted to energy [ev] (the original scale was in wavenumber [cm -1 ]). 129

139 43 as demonstrated by Brinkmann et al. 43,44. The same authors studied the PL of clathrated Alq 3 - (C 6 H 5 Cl) 1/2 and Al 3 (MeOH), α- and β- Alq 3 polycrystalline samples as well as a 50 nm Alq 3 film deposited on Si(001). While no vibronic progressions were found at room temperature they observed distinct vibronic sub-bands in α- and in β- Alq 3 crystals and weaker vibronic progressions in the clathrated samples at 4.2 K, shown in Fig No vibronic progression was found in CH 2 Cl 2 dissolved Alq 3 and in the Alq 3 thin film. In all cases the full width at half maximum of the broad PL band decreases and the PL maximum shifts toward higher energies with decreasing temperature. To analyze the vibronic structure in the polycrystalline samples absorption PL intensity energy [ev] Figure 3.13: Comparison between the absorption spectrum at 300 K and PL spectrum at 4.2 K of polycrystalline α-alq 3. The position and the intensities of the various components of the vibronic structures are also shown. The figure is taken from Brikmann et al 43, the original scale was in wevenumber [cm -1 ] and is modified to energy scale. Brinkmann et al. 43 fitted the PL band with 6 Gaussians at the most apparent features (see Fig. 3.13). Additional 1-2 broad Gaussians were used for the structureless tail in the spectra. The fit was performed in accordance to a Poisson distribution with a vibron coupling constant g 2 = 2.6. The separation of vibronic modes by 64 mev could be attributed to three intense skeletal in- 130

140 plane bending modes of the ligand around 64 mev using Raman spectroscopy. The involvement of ligand modes is in agreement to DFT calculations 98 showing that the lowest singlet electronic transition is primarily localized on one of the quinolate ligands. No detailed spectrally resolved and time resolved PL measurements on Alq 3 polycrystalline samples are available to date; also no calculations on Frenkel excitons in Alq 3 crystals have been performed so far. In accordance to our investigations on PTCDA crystalline samples the Alq 3 fluorescence that exhibits vibronic progressions is attributed to a Frenkel exciton transition. A comparison with crystalline structures of α- and in β- Alq 3 further reveals the existence of close contacts between pairs of quinoxaline ligands belonging to neighboring molecules with interligand spacing in the range 0.39 to 0.35 nm. Therefore the formation of charge-transfer or excimer excitons may become likely in Alq 3 crystals. Accordingly the broad low-energy band (centered at 2.32 ev) in Fig is tentatively attributed to self-trapped excitons. In addition the broad band consists of excitons that are trapped in local energy minima due to crystal imperfection or defects at crystal grain boundaries. As mentioned earlier there are no vibronic progressions in the PL spectrum obtained from the Alq 3 thin film. This observation is in agreement with our PL investigations on Alq 3 films that were grown by OMBD. Figure 3.14 shows the PL spectrum of a 40 nm thick Alq 3 film on Si (001) recorded at 10 K and 300 K, respectively. As in the investigations of Brinkmann et al. 43 we find a decrease of the FWHM of the PL band and a red-shift of the PL maximum (2.385 ev at 300 K and 2.36 ev at 10 K, respectively) when the temperature is decreased. Also shown in figure 3.14 is the first absorption band at ~3.10 ev which originates from a L type exciton transition 43,70,44,

141 The missing vibronic progression in the PL spectrum at 10 K is explained by the polymorphous structure of the film that contains all possible crystalline modifications with very small (several ten nanometers) crystal sizes. This disorder leads to a huge amount of grain boundaries that contain crystal imperfections, defects and vacancies, hence leading to a vast distribution of molecular environments resulting in large inhomogeneous broadening of the vibronic sub-bands. wavelength [nm] K PL intensity 300 K absorbance energy [ev] Figure 3.14: Absorption spectra of 40 nm thick Alq 3 on Pyrex and PL of 40 nm thick Alq 3 on Si (001) at 10 K (top) and at 300 K (bottom). Moreover, crystalline defects lead to potential fluctuations with trapping potentials lower than the Frenkel exciton LUMO energy. Optically excited and mobile Frenkel excitons are trapped within very short (~ ps) time in such potentials. Correspondingly, the fluorescence from thin films is predominantly caused by the emission of those trapped excitons. The energy redshift of the thin film PL compared to the crystalline films supports this interpretation. As will be shown in section the I-V characteristics in Alq 3 based OLEDs reveals a behavior that is 132

142 described by a trapped charged limited current (TCLC) that also supports the presence of trapping potentials (for electrons) below the LUMO level. As in crystalline Alq 3 samples the fluorescence of charge-transfer excitons or excimers might in addition contribute to the PL. Since the self-trapping energy in the same energy range as the average trapping energy caused by dislocations or crystal imperfections in the film it is very difficult to distinguish different emission channels in the cw - PL spectrum. (A first hint on the presence of different emission channels is given by temperature dependent PL studies that will be discussed in the next section.) As in investigations on PTCDA, spectrally and time resolved PL measurements are appropriate tools to investigate the presence of different emission channels. However, there are no extensive studies available so far. Walser et al report on spectrally integrated time resolved measurements on 1-2 µm thick Alq 3 films on quartz with PL decay times ranging from 20 ns at 70 K to 16 ns at 300 K. The decay traces (recorded up to 50 ns) are not single exponential at higher excitation energy and may contain a further slower emission channel (with decay times greater than 20 ns). Humbs at al. 102,103 report on a multi-exponential PL decay in a 150 nm Alq 3 film on quartz within the first 100 ps at 300 K. The PL also shows a slow decaying component of 10 ns which is attributed to the lifetime of the fluorescent state. Humbs et al. 102,103 performed spectrally dependent measurements on four different detection energies that span over the PL band. By fitting their data up to 60 ps decay time they found an increasing contribution of the long living component with decreasing detection energy. However, the slow decaying component was not plotted in their paper. Additional spectrally and time-resolved measurements are therefore necessary (and are planned in the group of Dr. Wagner) to support the presence of self-trapped excitons in the PL of Alq 3 films with decay times longer than 20 ns. 133

143 3.4.2 Temperature Dependent PL of Alq 3 Thin Films Figures 3.15 (a) and (b) show the PL spectra of the Alq 3 film from 10 to 300 K in 10 K steps. At 10 K the peak position (2.385 ev) of the Alq 3 PL spectrum of the sample is shifted by ~ 100 mev to lower energies compared to the PL peak position obtained from α- and β- Alq 3 polycrystalline samples 43. Up to 180 K the PL intensity increases with increasing temperature (also see Fig. 3.16), while the intensity starts to decrease at temperatures above 180 K. In addition the PL spectrum is gradually red-shifted (see Fig (b) and Fig. 3.17) as temperature wavelength [nm] K PL intensity wavelength [nm] 10 K 300 K temperature [K] PL intensity energy [ev] temperature [K] 10 K (a) (b) Figure 3.15: Temperature dependent ( K) PL of a 40 nm Alq 3 film on Si (001) excited by a laser at 436 nm. The PL intensity in (b) is offset for better comparison. is increased from 10 to around 180 K. At higher temperatures the PL band slightly shifts to higher energies again. The observed PL maximum is in contrast to temperature dependent investigations (between 77 and 300 K) of Walser et al. 100,101 who found a monotonously decreasing PL intensity with increasing temperature. 134

144 integrated normolized PL intensity PL peak energy temperature [K] temperature [K] Figure 3.16: Integrated normalized PL intensity (obtained from Fig b) of the Alq 3 film as a function of temperature. Figure 3.17: Peak energy of the PL spectra (obtained from Fig b) of the Alq 3 film as a function of temperature. Presently the PL band is described by the singlet state radiative recombination ( S 1 S 0 + hν ) of Frenkel excitons that are trapped in the neighborhood of imperfections at the grain boundaries of the (several 10 nanometer) small crystals in the quasi-amorphous film. The observed red-shift compared to crystalline Alq 3 samples 43 is attributed to the trapping energy of Frenkel excitons in the disordered Alq 3 thin film. However, this picture alone can not describe the observed PL intensity maximum at 180 K (Fig. 3.16) or the redshift of the PL band up to 180 K (Fig. 3.17) while it slightly shifts back to higher energies at higher temperatures. In analogy to previous investigations on PTCDA films we tentatively propose the following model: In addition to the PL from trapped Frenkel excitons at crystal imperfections a second emission channel may occur originating from self-trapped excitons (excimers and/or charge transfer excitons). These self-trapped excitons (STEs) are in addition to Frenkel excitons formed within the small Alq 3 crystallites. The mobile Frenkel excitons are able to migrate to the grain boundaries leading to trapped excitons that have a high probability of non-radiative decay at 135

145 these sites. In contrast STEs are immobile (as long as they are not thermally released), giving them a high probability for radiative recombination. While the Frenkel exciton is generated during light absorption the creation of the STE may require a formation energy as observed in temperature dependent PL studies on PTCDA crystals 56. This formation energy might be caused by a thermally activated optimum overlap between pairs of quinoxaline ligands belonging to neighboring molecules. The presence of a STE formation energy could explain the observed PL maximum at 180 K. With increasing temperature the number of STEs increases leading to an increasing number of emitted photons from this excitation in addition to the trapped exciton PL from the grain boundaries. At temperatures higher than 180 K both trapped excitons and STEs are thermally activated from their trapping potential giving them the probability to find a deep trap that is a non-radiative center. Accordingly, the PL intensity decreases with increasing temperature (> 180 K). This model also supports the observed redshift of the PL band up to 180 K. If we assume that the generated STEs have a slightly higher binding energy than the trapped excitons at crystal imperfections, the increasing number of STEs leads to a red-shift of the PL maximum. The slight blue-shift of the PL spectrum (see Fig. 3.17) at high temperature is attributed to the thermal population of vibronic states of Alq 3. According to these assumptions the spectrally integrated PL intensity as a function of temperature has approximately been modeled by, nph γ rad, X = n0 X γ X ( T ) γ rad, STE + n0ste exp( f γ STE ( T ) / kt ) 3.9 where n ph is the number of emitted photons, γ rad, X and γ rad, STE are the radiative decay rates of trapped excitons and STEs, respectively, γ X (T ), and γ STE (T ) also include the thermally activated non-radiative rates γ non rad, X and γ non rad, STE, respectively, 136

146 γ X ( T ) = γ rad, X + γ non rad, X exp( X / kt ) 3.10 γ STE( T ) = γ rad, STE + γ non rad, STE exp( STE / kt ) 3.11 with trapping energies X and STE. n 0 X and n 0 STE are numbers of trapped excitons and STEs, which strongly depend on the film quality and size of crystallites in the quasiamorphous films. In particular we expect a significant decrease of n 0 STE if the crystal size is decreased (or the disorder in the films is increased.) This might explain why no STE emission i.e. no PL maximum at 180 K was found in studies of Walser et.al 100,101. For the creation of the STEs a formation energy f is required. Equation (3.9) neglects any photons form recombination of mobile Frenkel excitons since it assumes a trapping on a ps time scale. Furthermore for n 0 X we neglect that the formation of STE with increasing temperatures reduces the number of trapped excitons (no coupling between trapped excitons and STEs limit of low quantum efficiency). We further neglect any spectral dependence of the parameters within the trapped exciton and STE band. The PL intensity given by equation (3.9) is simulated using the following parameters: From the red-shift of the PL maximum in Alq 3 films compared to polycrystalline samples 43 of ~100 mev it assumes an average trapping energy of X = 100 mev. Due to the lack of experimental data we used slightly higher value STE = 120 mev for the self-trapping energy. From low temperature studies of Walser et al. 100,101 we take the radiative rate of trapped excitons γ rad,x = 0.05 ns -1 and use a slightly lower radiative rate for the STEs γ rad, STE = 0.04 ns -1. From the temperature dependent decay time also measured by Walser et al. 100,101 and using = 100 mev and eq. (3.10) we extract a value of γ non rad, X = 0.8 ns -1 and take the same value X 137

147 for the non-radiative rate of STEs γ non rad, STE = 0.8 ns -1. Furthermore the STE formation energy f = 10 mev has been used. Figure 3.18 shows the result of this simulation where the dashed curve gives the PL generated by trapped excitons, the dotted-dashed curve shows the luminescence from STEs and the full line shows the total PL from both recombination channels. Despite the lack of accurate data and the deficiencies of this simple model, the curve shows the essential features of the observed temperature dependence of the PL (see fig 3.16). The discrepancy of the PL intensity in trapped excitons STEs PL intensity temperature [K] Figure 3.18: Simulated results of the proposed model. Dashed curve is the PL intensity contribution from trapped excitons, the dotted-dashed curve is the contribution from STEs and the solid line is the total of both recombination channels. the simulation at higher temperature is attributed to the overestimated number of trapped excitons that is reduced by the generation of STEs which is not considered in the present model. Furthermore the scheme of the proposed exciton trapping model is summarized in Fig as energy level distribution and the exciton dynamical process in Alq 3 thin films. A systematic 138

148 investigation of the exciton lifetime using time resolved and spectrally resolved PL studies are planned in the group of Dr. Wagner to support and improve the present model.. γ non-rad, STE e STE / kt relaxation γ non-rad, X e X / kt f n 0 STE n 0 X LUMO STE X non-radiative traps self-trapped exciton γ rad, STE excitation process trapped exciton at defects γ rad, X non-radiative traps HOMO Figure 3.19: Proposed energy level scheme for the trap distribution and the exciton dynamical process for Alq 3 thin films on Si (001). 139

149 3.4.3 Room Temperature Studies on Alq 3 Based OLED Structures Organic light emitting diodes (OLEDs) are fascinating mainly for two reasons: Their remarkable applications such as flat panel displays, solid state lighting and possible flexible electronics. The other is the vast variety of options available to tune their properties for basic research and development. However, the production of OLEDs is challenging due to the different fabrication steps and various materials involved in the process. An OLED consists of organic materials, inorganic materials and metals as described in section 3.1. My first goal was to produce a two layer OLED structure that emits bright green light under an operating voltage of 10 V. current density [ma/cm 2 ] nm Mg -Ag (9:1) 50nm Alq 3 50nm TPD 60nm ITO Glass voltage [V] (a) (b) Figure 3.20: (a) - Room temperature J-V characteristics of the first OLED device produced in our lab, and a sketch of the device structure. (b) - A photograph of the operating device emitting bright green light. The structure of the device including the materials and the dimensions are shown in Fig (a). Alq 3 serves as both the emissive layer and the ETL while TPD is used as the HTL. A photograph of an operating device is shown in Fig (b). This was taken at an applied voltage of 15 V. The current density-voltage (J-V) characteristics of the device at room temperature is also shown in Fig (a). The observed turn on voltage at room temperature is around 5.5 V 140

150 with a current density of 300 µa/cm 2. A rapid increase of the current starts around 12 V and at 15 V, the current density is reached to 220 ma/cm 2. This is the usual rectification behavior expected for a diode. However, the current density of the device is significantly higher than that of a conventional diode 33. The high current density is typical for OLED structures 33,77. It can be explained by the model of space charge limited current (SCLC) and trap charge limited current (TCLC). A detailed analysis and discussion about temperature dependent current voltage measurement for an improved multilayer OLED structure is presented in section efficiency [cd/a] voltage [V] luminescence [cd/m 2 ] efficiency [cd/a] luminescence [cd/m 2 ] Figure 3.21: The luminescence and the efficiency of the device as a function of applied voltage at room temperature. Figure 3.22: The efficiency of the device as a function of luminescence at room temperature. The brightness and the EL efficiency (calculated by using Eq. 3.8) of the device are shown in Fig as a function of the applied voltage. The device turn on is clearly seen at 5.5 V while the maximum brightness of 900 cd/m 2 is observed around 14 V. As shown in Fig. 3.21, the EL efficiency of the device gradually increases until ~ 11 V after which it decreases with the voltage. The maximum efficiency of 2.2 cd/a is reached at 11 V. The, efficiency dependence as a function of brightness is shown in Fig The maximum EL efficiency (2.2 cd/a) at room temperature is observed at a brightness of ~ 150 cd/m 2. The highest value saturates when the 141

151 brightness further increases. The OLED s properties depend on the quality of the materials and on the fabrication procedures. Our first OLED structure has already shown reasonable performances indicating good quality devices and promising further developments. wavelength [nm] EL intensity [a. u.] 300K EL PL PL intensity [a.u.] energy [ev] Figure 3.23: The EL spectrum of the device (V =12 V; I = 1.2 ma) and the PL spectrum of a 50 nm thick Alq 3 film (λ excitaion = 436 nm; power = 1 mw) on Si(001) at room temperature. Furthermore, the room temperature EL spectrum of the OLED was recorded (V=12 V; I= 2.2 ma) and is shown in Fig together with the PL spectrum of a 40 nm Alq 3 on Si (001) sample. The Alq 3 film is photo-excited by a pulsed laser (power = 1 mw) with 436 nm at room temperature. These spectra raise two questions in relation to the carrier injection and the recombination process. As seen in Fig. 3.23, the EL spectrum is redshifted by ~ 100 mev compared to the PL spectrum. The other is that the PL spectrum has a larger contribution at higher energies while the EL has a higher tale at lower energies. This could be due to the following reasons within the Alq 3 layer of OLEDs: i. A triplet state radiative recombination 47,

152 ii. Strain effects applied to the Alq 3 layer by the other layers 41,60,65,66 iii. Interference effects from different layers of the OLED 107,108 iv. The formation of self-trapped excitons To address these issues, it is necessary to perform a systematic study on OLEDs. The temperature dependent electro-optical properties of an improved multilayer OLED structure have been investigated and the experimental observation is presented in the following sections. The carrier injection mechanism of an OLED can be explained using a band structure model. The energy level diagrams for selected organic materials and work functions for some metals are given in section (see Fig. 3.8). Accordingly, the low work function bi-layer of LiF-Al is a better choice 109 for the cathode material. On the other hand, it has been demonstrated LiF (0.6 nm)/ Al (150 nm) Alq 3 (40 nm) α-npb (60 nm) PEDOT (30 nm) ITO (60 nm) Glass Substrate (1 mm) Figure 3.24: Configuration of the improved Alq 3 based OLED device structure, individual layer thicknesses are also given in parenthesis. that direct ITO/TPD is a poor contact interface 76,110. To improve the contact between the ITO and the HTL, a conducting polymer layer of PEDOT has been introduced on the top of the ITO. This also reduces the barrier level at the anode and facilitates improved hole injection 85. In some OLEDs structures the HTL of TPD is replaced by α-npb concerning the stability of the organic 143

153 materials as α-npb has higher glass transition temperature 111,112 than TPD. These improved multilayer structures (see Fig. 3.24) are fabricated in collaboration with Dr. Steckl s group at the UC Nanolab and transported to our lab in a nitrogen container for the temperature dependent investigations. The experimental studies of this device structure are presented and briefly discussed in what follows. luminescence [cd/m 2 ] K voltage [V] current density [ma/cm 2 ] Figure 3.25: The luminescence and current density of the multilayer OLED as a function of the forward biased voltage at room temperature. The structure of the multilayer device is shown in Fig. 3.24, individual layer thickness are also given in the parenthesis. The luminescence-current density-voltage (L-J-V) characteristics of the device at room temperature are shown in Fig The device turns on at a bias of 2.8 V, and reaches 460 cd/m 2 at 4 V. The brightness saturates at ~ 10,000 cd/m 2 for operating voltages of ~ 15 V (not shown here). The I-V dependence exhibits a typical forward biased diode characteristics. A photograph of an operating device is also shown in Fig for a forward bias of 10 V. 144

154 Figure 3.26: A photograph of the OLED showing bright green light emission. Some of the device properties of the two-layer OLED and the multilayer OLED are summarized at room temperature in Table 3.1. The turn on voltage and brightness are significantly improved in the multilayer OLED. The temperature dependent multilayer device properties are presented and discussed in the following sections. Device property Two layer OLED Multilayer OLED Turn on voltage [V] Maximum luminescence [cd/m 2 ] Voltage at which maximum luminescence [V] Maximum efficiency [cd/a] Voltage at which maximum efficiency [V] 11 5 Table 3.1: A comparison of some device properties of the two-layer OLED and the multilayer OLED at room temperature. 145