Linear and non-linear optical properties of OMBD grown PTCDA and Alq3 films

Transcription

1

2 Linear and non-linear optical properties of OMBD grown PTCDA and Alq3 films 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 DOCTOR OF PHILOSOPHY (PhD) in the Department of Physics of the College of Arts and Sciences 01 by Ahamed Milhan Ajward B.Sc., University of Ruhuna, Sri Lanka, 00 M.S., University of Cincinnati, 006 Committee Chair: Professor Hans-Peter Wagner ii

3 Abstract In this thesis, linear and non-linear optical properties of perylene-3,4,9,10- tetracarboxylic-3,4,9,10-dianhydride (PTCDA) and of tris(8-hydroxyquinolinato)aluminum (Alq3) films have been investigated using various experimental methods. The films were grown using the technique of organic molecular beam deposition (OMBD). The dispersion of the in-plane and normal refractive index in PTCDA waveguides has been studied using the m-line technique. The PTCDA waveguides were grown on Pyrex substrates. TE and TM mode coupling at excitation wavelengths ranging from 633 to 910 nm has been accomplished using a Rutile prism. The derived refractive index values are in good agreement with existing ellipsometric data, which emphasizes the high structural quality of our waveguides. The nonlinear absorption (two-photon absorption) of PTCDA and Alq3 films has been investigated using the z-scan technique. The films were grown on Pyrex substrates and a high repetition rate (80 MHz) laser was used as excitation source. Various methods have been utilized to minimize laser induced damaging of the soft organic films and to improve the signal-to-noise ratio. In addition, nonlinear fluorescence (two-photon induced fluorescence) measurements have been performed on Alq3 films to further investigate nonlinear absorption processes in this material. The singlet-singlet annihilation (nonlinear bimolecular quenching) of excitons in disordered quasi-amorphous Alq3 films has been investigated using both time-resolved and continuous wave (cw) spectroscopic techniques at temperatures ranging from 15 K to 300 K. A iii

4 significant decrease of the PL efficiency with increasing exciton density (excitation intensity), especially at low temperatures, has been observed which is attributed to bimolecular quenching of excitons that are funneled into low-energy traps. This effect is different from the known diffusion based singlet-singlet annihilation in Alq3. To explain both the intensity and temperature dependence of the quenching phenomena a simple model has been developed and used to analyze and to simulate the complex experimental observations. To investigate the possible influence of the degree of disorder on the bimolecular quenching of trapped excitons, PL efficiency experiments have been performed on Alq3 films that have been grown at different growth conditions. The investigations include films with varying thickness, films that were grown at various growth rates and on different substrates as well as annealed Alq3 films. Changes of bimolecular annihilation effects due to these modifications have been qualitatively explained. iv

5 v

6 To my parents Ikram and Hidaya Ajward vi

7 Acknowledgements First and foremost, I would like to express my sincere gratitude to my advisor Prof. Hans- Peter Wagner for his motivation, enthusiasm and patience throughout my PhD dissertation work. His immense knowledge in the field and superior guidance was extremely helpful throughout my research work and also in writing my thesis dissertation. I like to mention that I could not imagine having a better advisor or mentor than him and he is one of the best advisors a student could have. Next I would like to thank my thesis committee: Prof. Young Kim, Prof. Leigh Smith, and Prof. Rohana Wijewardhana, for serving as members in my thesis committee and also their valuable ideas, and encouragement during and in completing the thesis. Then I m much grateful to Dr. Xiaosheng Wang for his insightful comments and ideas during the research work. Also his skills and suggestions were much helpful in getting my thesis work completed. Next I would like to thank Dr. Heidrun Schmitzer (Xavier University) for providing ideas and support for the z-scan experiments. I would also like to thank John Marcus (Microelectronics Engineer in the physics department) for his ideas and suggestions in solving different technical problems in the laboratory. Also I would like to thank to Dr. Warren D Huff for performing the X-ray diffraction measurements on the Alq3 samples and he was very much helpful in these measurements. vii

8 Next I would like to thank all my fellow research colleagues: Venkateshwar Gangilenka, Amin Kabir, Pradeep Bajracharya, Amit Dongol, Niranjala Wickremasinghe, and Masoud Kaveh for sharing their valuable thoughts, and time with me and also making the research life much more enjoyable. Then I would like to further extend my thanks to faculty and staff in the physics department, especially to Dr. David Groh (current graduate student advisor in the physics department), Diane Willingham, Melody Whitlock (former graduate secretary in Physics), Donna Deutenberg, Elle Mengon, John Whitaker, Bob Schrott, Mark Ankenbauer, and Mark Sabatelli. Last but not the least, I would like to thank my family: my parents, Ikram Ajward and Hidaya Ajward, for giving me the emotional and spiritual support whenever I needed them through the studies and my life. Then I give my thanks to my brother and sister, Roshan Ajward and, Farinza Ajward for their great support and patience especially during hard times. viii

9 Table of contents 1 Introduction... 1 Small molecule organic solids Introduction Molecular orbitals and energy levels Absorption and emission in an organic material Decay and quantum efficiency of the fluorescence in organic materials Selection rules in molecular systems Electron and energy transfer in organic materials Förster resonance energy transfer Dexter energy transfer Exciton migration in organic solids Interactions between excitons Organic film growth and thickness monitor calibration Introduction to used organic materials Introduction to Alq Introduction to PTCDA Intermolecular transitions between molecules Origin of traps in organic semiconductors Organic molecular beam deposition (OMBD) Organic molecular beam deposition technique (OMBD) and film growth Reflection measurements and film thickness calibration Experimental details of the reflection measurements ix

10 4 Refractive index dispersion in polycrystalline PTCDA waveguides Introduction Introduction to prism-coupling and waveguiding Details of the experiment Experimental results and discussion Nonlinear responses of organic materials and z-scan technique Light interaction with matter Light propagation through a nonlinear optical medium The origin of nonlinear optical properties of organic material σ and π bonds: The effects of symmetry Second order nonlinear optical processes Third order nonlinear optical processes Nonlinear refraction Nonlinear absorption z-scan technique Nonlinear refraction and z-scan measurements Theoretical description of nonlinear refraction and z-scan Nonlinear absorption and z-scan measurements Experimental setup Pulse width measurements z-scan measurements on Alq3 films z-scan measurements on polycrystalline PTCDA films Two-photon induced fluorescence Experimental setup for two-photon induced fluorescence measurements x

11 6 Singlet-singlet annihilation due to funneling of excitons to traps in Alq3 films Photoluminescence studies of tris(8-hydroxyquinolinato)aluminum films Singlet-singlet annihilation in Alq3 films Time resolved photoluminescence measurements Modeling of the singlet-singlet annihilation in Alq3 films Simulating the integrated PL efficiency as a function of temperature and intensity Using Gaussian beam profiles for the laser excitation Singlet-singlet annihilation in Alq3 films of different layer thickness and in layers grown at different growth parameters Summary Bibliography xi

12 List of Figures Figure 1.1: Several examples where OLEDs have been used in commercially available devices. Figure 1.: Simple OLED device (Electro-luminescent device).... Figure.1: (a) Example of the conjugated -electron system. The and bonds from organic material, ethene. (b) Energy level diagram of a -conjugated molecule. The energy is lowest between the bonding and anti-bonding -orbitals. (from reference 16 ) Figure.: Comparison of typical optical spectra of organic molecules in different environments Figure.3: Large binding energy of Frenkel excitons when compared to Wannier exciton Figure.4: Diagram for the energy vs. configuration coordinate of the absorption and emission between two electronic states S 0 and S 1 of an organic molecular material. Each electronic state includes several vibronic modes. The arrow Q represents the nonradiative transition from the highest excited vibronic mode to the lowest LUMO level. The Frank-Condon transitions are the vertical arrows E and P Figure.5: Stokes-shift between absorption and emission spectra is shown in the diagram. Each vibronic transition in either case contributes to the spectrum Figure.6: Energy diagram for fluorescence (from singlets) and phosphorescence (from triplet states). The singlet to triplet energy transfer is called the intersystem crossing (ISC) Figure.7: Excitonic energy transfer in molecules: (a) Förster energy transfer (b) Dexter energy transfer. In both cases energy is transferred from singlet exciton D* to molecule A. The final states are shown in (c) xii

13 Figure.8: Förster energy transfer from a donor molecule (singlet) to an acceptor molecule. The vibrational relaxations of both molecules are indicated Figure 3.1: Chemical structure of tris(8-hydroxyquinoline) aluminum(iii) (Alq3) (from 16 )... 3 Figure 3.: Different isomers of Alq3 (from 16 )... 3 Figure 3.3: Crystalline packing in different phases of Alq3: (a) Alq3(C 6 H 5 Cl) 1/ (b) Alq3 α- phase; and (c) Alq3 β-phase (from 88 ) Figure 3.4: (a) single PTCDA molecule (b) HOMO 90 (c) LUMO 90, energy in PTCDA. The blue and red colors represent the polarity of the molecular orbital coefficients Figure 3.5: Sketch of PTCDA crystals (Figure 3.5 (b) from 9 ) Figure 3.6: Different crystalline phases of PTCDA films Figure 3.7: (a) Excimer (b) Charge transfer exciton between two molecules Figure 3.8: Schematic diagram of the OMBD growth chamber (from 13 ) Figure 3.9: Photograph of the OMBD growth chamber in our laboratory Figure 3.10: Experimental setup for F.P. reflection measurements (from 99 ) Figure 3.11: F.P. reflections of an 850 nm thick Alq3 film on glass Figure 3.1: F.P. reflections of an 865 nm thick Alq3 film on Si(100) Figure 3.13: F.P. reflections of a 1835 nm thick Alq3 film on glass Figure 3.14: F.P. reflections of a 1970 nm thick Alq3 film on Si(100) Figure 4.1: Two prism couplers used to feed and extract light at two ends of a planar film waveguide Figure 4.: A prism coupler is used to detect the coupling of propagating modes in a waveguide xiii

14 Figure 4.3: Sketch of a typical signal obtained from the reflected beam intensity in a prism coupling experiment Figure 4.4: Sketch of the rutile prism on the waveguide showing the incident and coupling angles Figure 4.5: Photograph of the adjustable waveguide holder with Rutile prism. Also shown is the microscope objective lens. On the right side a photograph of a m-line in the reflected laser spot is shown Figure 4.6: Effective refractive index values for TE and TM modes. The values are calculated with equation (4.) for different light coupling incident angles and at different wavelengths ranging from 633 to 910 nm. Effective refractive index values which belong to the same mode number are plotted in the same color Figure 4.7: Observed effective refractive index values n eff of TE modes (black circles) at = 66 nm. Red circles show observed n eff values of TM modes. Solid and dashed lines represent calculated values of n eff as a function of the waveguide thickness for different TE and TM modes as labeled Figure 4.8: Inplane n and normal n refractive index values (full and open blue circles, respectively) derived from m-line measurements in the spectral region between = 663 to 910 nm. The circles are connected by the Sellmeier like equation (4.5). For comparison, n and n values as well as refractive index dispersion curves derived by other groups are also given. Stars: Fuchuigami et al. 115, dashed green lines: Friedrich et al. 100, dashed-dotted black line: Djurisic et al. 116, short-dashed black line: Djurisic et al xiv

15 Figure 5.1: Benzene (C 6 H 6 ) resonance structures Figure 5.: (a) Sketch for SHG, (b) Energy-level diagram for explaining the SHG process Figure 5.3: (a) Sketch for sum frequency generation, (b) Energy-level diagram for explaining the sum frequency generation process Figure 5.4: (a) Geometry for difference frequency generation, (b) Energy-level diagram for explaining the difference frequency generation process Figure 5.5: (a) Sketch of THG interaction (b) Energy level diagram of THG interaction Figure 5.6: Sketch of self-focusing in a third order nonlinear optical medium Figure 5.7: Energy level diagram for two photon absorption Figure 5.8: Closed aperture z-scan setup for nonlinear refraction measurements Figure 5.9: Depending on the sign of the change of the nonlinear refractive index n the sample may act as a converging or diverging lens (after 140 ) Figure 5.10: z-scan traces for the transmission through the aperture for negative and positive nonlinear refractive index change (after 140 ) Figure 5.11: When both nonlinear refraction and absorption are present at the same time the division of closed aperture data by the open aperture provides the data purely due to the nonlinear refraction (after 140 ) Figure 5.1: Normalized transmittance of a closed aperture z-scan experiment. The height between peak and valley ΔT is proportional to the phase change due to nonlinear p v refraction and can be used to calculate the nonlinear refractive index of the material xv

16 Figure 5.13: z-scan setup for nonlinear absorption measurements. In order to measure the nonlinear absorption all the transmitted light is collected by a photodiode(after 140 ) Figure 5.14: Illustration of the normalized transmittance of a nonlinear absorption measurement using the z-scan technique Figure 5.15: Experimental setup for the z-scan experiments Figure 5.16: The high repetition rate laser pulse train is chopped using the mechanical shutter at regular time intervals to reduce thermal effects in the thin films. The figure shows the pulse trains transmitted through the shutter at regular intervals Figure 5.17: Autocorrelation setup for pulse width measurements Figure 5.18: Open aperture z-scan trace of a 5 m Alq3 film on glass at low incident average power (~7.5 mw) Figure 5.19: Open aperture z-scan trace of a 5 m Alq3 film on glass at high incident average power (~746 mw) Figure 5.0: Open aperture z-scan trace of a 5 m Alq3 film on glass at 746 mw average power Figure 5.1: Open aperture z-scan trace of a 5 m Alq3 film on glass at 480 mw average power Figure 5.: Open aperture z-scan trace of a m PTCDA on glass at low average power (~8 μw) Figure 5.3: Open aperture z-scan trace of a m PTCDA film on glass at high average power (.8 mw) Figure 5.4: Open aperture z-scan trace of a m PTCDA film on glass xvi

17 Figure 5.5: Images of laser induced melting of holes into the 5 m thick PTCDA film at different magnifications Figure 5.6: Microscopic images focused at different depth levels Figure 5.7: z-scan trace of a laser damaged film Figure 5.8: Illustration of the cross section of the molten area in the film Figure 5.9: Experimental setup for two-photon fluorescence measurements. The reference detector was used to record fluctuations of the laser Figure 5.30: Two-photon induced fluorescence spectrum of a 5 m Alq3 sample at an incident average power of 7 mw Figure 5.31: Quadratic intensity dependence of the two-photon induced fluorescence obtained from a 5 m Alq3 film Figure 6.1: PL spectra of an 80 nm thick Alq3 film deposited on Si (001) as function of temperature ranging from 10 to 300 K in 0 K steps Figure 6.: Due to the energetic disorder in an amorphous material the excitons migrate downhill the density of states (DOS) and are trapped in energetically deeper states. This mechanism also allows more than one exciton to be funneled to two adjacent trapped exciton states, where singlet-singlet annihilation can occur Figure 6.3: Dependence of the PL efficiency of a 10 nm Alq3 film as a function of temperature measured at a wavelength of 534 nm. The incident average laser power was 0.3, 0.03 and 0.01 mw at 407 nm excitation wavelength Figure 6.4: Time-resolved and normalized PL intensity obtained from a10 nm Alq3 film at a detection wavelength of 534 nm at different average power as labeled Figure 6.5: Experimental setup for time-resolved measurements xvii

18 Figure 6.6: Time-resolved and normalized PL intensity obtained from a 10 nm Alq3 film at an average excitation power of 0.3 mw and different temperature as labeled Figure 6.7: Normalized initial PL signal of time-resolved traces as a function of temperature obtained from a10 nm Alq3 film at different average excitation powers as labeled Figure 6.8: Time-resolved and normalized PL intensity (thin black curve) obtained from a10 nm Alq3 film at an average power of 0.01 mw and at 15 K. Also shown (thick red curve) is a two-exponential fit as explained in the text Figure 6.9: Time-resolved and normalized PL intensity (thin black curve) obtained from a 10 nm Alq3 film at an average power of 0.3 mw and at 15 K. Also shown (thick red curve) is a fit that includes bimolecular quenching of S excitons as explained in the text... 1 Figure 6.10: Difference signal between the time-resolved and normalized PL intensities at average powers of 0.01 mw and 0.3 mw at 15 K (thin black curve). Also shown (thick red curve) is the calculated difference signal obtained from eqs. (6.3) and (6.8) using the extracted parameters from Figs 6.7 and Figure 6.11: Time-resolved and normalized PL intensity (thin curves) obtained from a10 nm Alq3 film at an average power of 0.3 mw and different temperatures as labeled. Also shown (thick curves) are calculated PL traces obtained from the model as explained in the text Figure 6.1: Time-resolved and normalized PL intensity (thin curves) obtained from a 10 nm Alq3 film at an average power of 0.03 mw and different temperatures as labeled. xviii

19 Also shown (thick curve) are calculated PL traces obtained from the model as explained in the text Figure 6.13: Time-resolved and normalized PL intensity (thin curves) obtained from a 10 nm Alq3 film at an average power of 0.01 mw and different temperatures as labeled. Also shown (thick curve) are calculated PL traces obtained from the model as explained in the text Figure 6.14: Calculated total time integrated PL efficiency as a function of temperature for 0.3, 0.03 and 0.01 mw excitation power, respectively, as well as individual contributions of light emission from states S 1, S and CT states at 0.3 mw Figure 6.15: Spectrally integrated PL efficiency of a 10 nm thick Alq3 film as a function of temperature and laser intensity as labeled. In these measurements a cw GaN laser at a wavelength of 407 nm and a photodiode for detection of the PL were used. (These measurements were performed by Dr. Xiaosheng Wang) Figure 6.16: Time-resolved and normalized PL intensity (thin black curve) obtained from a 10 nm Alq3 film at an average power of 0.3 mw and 15 K. Also shown (thick red curve) is a fit that includes bimolecular quenching of S excitons considering a Gaussian beam profile as explained in the text Figure 6.17: Spectrally integrated PL efficiency of Alq3 films grown on glass with different layer thickness as labeled Figure 6.18: PL spectra of an 8, 0 and 10 nm thick Alq3 film at 15 K Figure 6.19: Spectrally integrated PL efficiency of 10 nm thick Alq3 films as a function of growth rate as labeled xix

20 Figure 6.0: Spectrally integrated PL efficiency of 10 nm thick Alq3 films as a function of annealing temperature as labeled Figure 6.1: Spectrally integrated PL efficiency of 10 nm thick Alq3 films grown on different substrates as labeled xx

21 List of Tables Table.1: Comparison of exciton diffusion properties of some known organic solids Table 4.1: Different lasers used in the m-line experiments Table 4.: The parameters A, B and C in Sellmeier equation ( in m) to calculate the in-plane n and normal n refractive index dispersion in PTCDA polycrystalline films Table 6.1: Parameters used to simulate the dynamical behavior of bimolecular annihilation of trapped excitons and conversion to charge transfer excitons in Alq3 films. trapping rate from free excitons, radf, 1,, CT f is the, are the radiative rates of free excitons and of states S 1, S and CT states, respectively. 3 and bq are the transition rate from state S to CT excitons and the bimolecular quenching rate of state S, respectively. 1,, CT are localization energies of states S 1, S, and CT. Energies c and b denote the thermal activation and binding energies of the CT state, respectively xxi

22 Chapter 1 1 Introduction Since the development of the first organic electroluminescent diode by Tang et al. in luminescent devices based on organic materials have become one of the most promising technologies which might eventually replace their inorganic counterparts (LCD). OLED displays are very attractive because they are slimmer, faster, energy efficient, flexible, durable, and can be fabricated at cheaper production costs when compared with other display technologies. In addition OLED displays have a larger viewing angle and put less stress on the eyes of the viewer because of the better contrast, color and brightness. While there are so many advantages of this technology, short life span 3, and highly susceptible to water damage 4 are some of the major disadvantages. These days OLEDs based on small molecules are used in all sorts of commercial display applications and different thermal deposition techniques are used to fabricate these devices 5. The first monochrome OLED display was produced by Pioneer cooperation in In 003 both Sanyo and Kodak introduced the first color Active-Matrix OLED (AMOLED) display products 7. Also in 008 Sony produces the first color AMOLED TV 8. Although the OLED display technologies have evolved quickly the quantum efficiencies of the OLED devices are relatively small namely about 1.4% 9. Since the photoluminescence (PL) efficiencies are around 30% 9, there is huge possibility to improve the electroluminescence (EL) efficiencies of these devices. In order to improve both EL and PL efficiencies of these OLED devices, the dynamics of the luminescence emission has been studied for past few decades. The EL efficiency of an OLED (see Figure 1.) is spin dependent because the electron and hole may recombine as a singlet or as a triplet. Statistically 5% of the excitons are singlets and 75% are 1

23 triplets. Since the triplets recombine non-radiatively the maximum possible theoretical efficiency of the OLED device is 5%. = (a) Transparent OLED Laptop Display form Samsung 10 (b) LG OLED TV 11 (c) OLED camera from Samsung 1 Figure 1.1: Several examples where OLEDs have been used in commercially available devices. Figure 1.: Simple OLED device (Electro-luminescent device). Despite the enormous research effort by researchers all over the world to improve OLED efficiencies the complex dynamics of electronic excitations, their migration and relaxation in organic materials is still not completely understood which often limits the improvement of organic devices. In addition the contribution of intermolecular interactions in organic solids is

24 often not well known but is of importance to understand the linear and nonlinear properties in crystalline and amorphous organic solids. In this thesis I have used different optical characterization techniques to study the linear and non-linear optical properties of polycrystalline 3,4,9,10 perylene tetracarboxylic dianhydride (PTCDA) and of quasi-amorphous tris (8-hydroxy) quinoline aluminum (Alq3) organic films. The high quality films used in this thesis were grown using the modern nearly ultra-high vacuum growth technique of organic molecular beam deposition (OMBD). The OMBD technique is capable of fabricating high purity organic thin films, multilayers and optoelectronic devices at extremely clean environmental conditions and with a monolayer thickness control. Because of the rapid development of the optical integrated circuits (OICs) it has become very important to study and to investigate the properties of organic waveguide and their ability to couple transverse electric (TE) and transverse magnetic (TM) optical modes. The high anisotropy in the refractive index in poly-crystalline PTCDA films makes them strong candidates for waveguiding applications and for pressure sensitive devices. In this thesis I investigate the dispersion of the in-plane and normal refractive index of OMBD grown PTCDA films using the m-line technique. The experimental results are compared with ellipsometric data obtained from literature. Also nonlinear optical properties of organic materials have become of great interest during the past decades because of their large nonlinear response in organic materials compared with their inorganic counterparts. With modern growth methods like OMBD a variety of few µm thick organic films as well as multilayers that are composed of different organic materials can be fabricated. An interesting molecular system with promising high nonlinear optical properties are polycrystalline films of planar PTCDA (3,4,9,10-perylene-tetracarboxylic-dianhydride) 3

25 molecules that form stacks along the growth direction. Another interesting organic material system are tris-(8-hydroxy)-quinoline-aluminum (Alq3) films. In this thesis the optical nonlinearities of these organic films are investigated by the z-scan technique. While the determination of the nonlinear absorption and of the nonlinear refractive index in inorganic crystals of several mm size is standardized the investigation of a few µm thick organic films using the z-scan method is challenging. Purpose of this work is therefore to study the nonlinear absorption of large area (~ 1 cm ) but only few µm thin organic films by constructing a sensitivez-scan set-up which minimizes thermal effects caused by the incident light beam. In the last chapter of my thesis I investigate the light emission of optically excited Alq 3 films deposited on Si (001) by spectrally resolved and temperature dependent time integrated as well as time-resolved photoluminescence. Goal of these investigations is to identify the contribution of different exciton states to the total light emission in these technological important films. The experimental results indicate that emission at low temperature is predominantly caused by trapped excitons. With increasing exciton density (excitation intensity) a significant decrease of the PL efficiency has been observed which is attributed to singlet-singlet annihilation of excitons that are funneled into low-energy traps. The trapping process significantly enhances the local density of excitons that is necessary to generate bimolecular annihilation. Therefore this quenching process is observable at much lower excitation levels as singlet-singlet annihilation processes that are based on exciton diffusion. With increasing temperature the effect of bimolecular quenching nearly vanishes which is attributed to the formation of thermally activated two-molecule states between adjacent molecules as for instance charge-transfer or excimer-like excitons. Our interpretations are supported by a simplified coupled rate equation 4

26 model and might be useful to explain the temperature dependent behavior of the EL efficiency of OLED devices 13. 5

27 Chapter Small molecule organic solids.1 Introduction All organic semiconductors are made of carbon based compounds and these materials are divided into two major categories as, low molecular weight materials and polymers 16,17. In both of these types of organic materials so called (sigma) and (pi) bonds form the molecule. The and bonds and its electron systems are formed because of sp and sp 3 hybridization in the carbon atoms (see Figure.1). The bond is much weaker when compared with the bonds and the lowest energy transitions occur between * energy levels in conjugated molecules 16,18. is called the bonding orbital and molecules. In general this gap between spectrum 16. * is called the anti-bonding orbital of the * is around 1.5 and 3 ev and is in the visible The two categories of the organic solids are processed in two different methods. While small molecular materials are grown using sublimation (or evaporation) at very high vacuum, conjugated polymers cannot be deposited using this method because of their weight 16,19. For conjugated polymers solution based deposition techniques, such as spin-coating is used 0. Organic solids are different from their inorganic counterparts because of the differences in the binding strength. While in inorganic solids strong covalent or ionic bonds are present, in organic solids the molecules are bound by van der Waals bonds and the intermolecular interactions are very weak 1,. These weak interactions have an impact on the optical and charge carrier transport properties 3,4 and on the structural properties of organic solids. For instance 6

28 organic materials are relatively soft and have a lower melting point compared to inorganic solids because of the weak interactions among the molecules 5,6. Figure.1: (a) Example of the conjugated -electron system. The and bonds from organic material, ethene. (b) Energy level diagram of a -conjugated molecule. The energy is lowest between the bonding and anti-bonding -orbitals. (from reference 16 ). Figure.: Comparison of typical optical spectra of organic molecules in different environments. 7

29 Due to the weak intermolecular interactions between molecules the optical absorption and photoluminescence spectra of organic solids are similar to the spectra of the molecules in gas phase or in solution, besides the observed solvent shift 7,8 (see Figure.). In organic solids the intermolecular vibrations play an important part in the spectra and can be observed in the absorption or luminescence spectra 6,9. Because of this in some cases molecular crystals are treated as an oriented gas 6,30. However, the spectra of organic solids can be different from a gas with respect to the selection rules, energetic positions and oscillator strengths, because of their crystal structure or packing 16. Figure.3: Large binding energy of Frenkel excitons when compared to Wannier exciton. An excited electron in the conduction band and the resulting hole in the valence band in an inorganic semiconductor crystal are weakly bound by Coulomb interaction resulting in an exciton in which the electron-hole separation is large compared to the lattice constant of the semiconductor crystal. Such excitations are called Mott-Wannier excitons (see Figure.3). Such excitons can be thought to be moving in a uniform crystal with dielectric constant 31,3 r. The electron-hole pair can be thought of a hydrogen-like system with a reduced effective mass ( 8

30 , where m e and mh * * 1 1 me 1 mh are the effective masses of the electron and hole respectively) and the dielectric constant r. The resulting binding energy of a Wannier exciton is in the order of a few 10 mev. In contrast to inorganic semiconductors, excitons in organic semiconductors are generally localized to one molecule and this type of exciton is called a Frenkel exciton. The binding energy of a Frenkel exciton is in the order of 0.5 to 1 ev. Since the exciton is localized to a single molecule a continuous medium with dielectric constant r cannot be considered like in the case of Mott-Wannier excitons 31,3. For Frenkel excitons any interaction with other molecules play a secondary role and the wavefunction of the exciton can be generated from the wavefunctions of isolated molecules. In contrast to the Wannier-Mott excitons Frenkel excitons have a mean electron-hole distance that is much smaller than the crystal lattice constant. The motion of an exciton can be either characterized by wave packet like motion or hopping motion depending on the nature of the exciton. If the exciton motion is wave packet like as in inorganic semiconductors then the excitionic properties can be described using a wave vector k and the group velocity of the propagation of the excition is given by3, where E s (k) is the exciton energy. 1 (.1) v E ( k) s But when the wavevector uncertainty k is in the order of k then the excition can no longer be characterized by means of a wavevector and the exciton is said to be localized. Further if k is in the order of inverse of the lattice constant the exciton is strongly localized in 9

31 one or in two molecules and will propagate by hoping from one molecule to the other. A localized exciton cannot be defined by a wave vector k and it is defined by the vector that is associated with the crystal cell where it is localized 3. In general an exciton that propagates through a crystal in a wave like manner and if the temperature of the solid is increased will get scattered by phonons, defects and impurities. This will result a decrease in the mean free path due to the increased scattering with increasing temperature. In contrast, if an exciton is localized and migrates from one molecule to another molecule through hopping motion, then the mean free path will increase with increasing temperature. This is because excitons are getting detrapped by overcoming the localization energies with increasing temperature.. Molecular orbitals and energy levels Because of the complex nature of the simplest organic molecules, the so called Born- Oppenheimer approximation is used to analyze the physical properties of molecules In this approximation the electronic and nuclear motions are treated separately. Also the spin of the electrons can be treated separate from both the motion of the electrons and the motion of the nuclei in space. This is because the electronic spin is associated with the magnetic field 33,36 and the magnetic field is weakly related to the electronic interaction in molecules 37,38. Furthermore, molecular orbitals (MO) can be approximated as a sum of atomic orbitals (AO) 39,40. Here the electrons in the MOs are thought to belong to the whole molecule, rather than to a particular molecular nuclei. In such instances the MOs are approximated as a linear combination of atomic orbitals (LCAO) of the constituent atoms, and the Schrodinger equation is solved in order to obtain mathematical forms of the MOs. The electrons are placed into these MOs according to the Pauli and Hund rules 41,4, in the same way electrons are placed into an atom. The electronic 10

32 configuration generated by placing electron pairs in all the orbitals below the Fermi energy is called the electronic ground state of the molecule 43,44. When excited electrons are removed from the doubly occupied orbitals and are added to the anti-bonding orbitals..3 Absorption and emission in an organic material Absorption and emission processes in an organic material are similar to that in an inorganic material. When an organic material is excited with a photon, an electron is excited from the ground state to a higher excited state. After some time the excited electron will relax back to its initial ground state while emitting a photon. The energy levels in a typical organic material are shown in Figure.4. The electron excitation occurs in the time scale of ~10-15 s and molecular (nuclear) responses occur in a larger time scale (e.g. ~10-13 s) 38,45. Because of this fact, the molecular nuclei have not enough time to change their spatial configuration. Therefore the electronic transition occurs vertically in a configuration diagram (see Figure.4). These transitions are explained by the Frank-Condon principle 38,45 which are also equivalent to the Born-Oppenheimer approximation 46,47. The electronic transition occurs from the lowest vibrational state of the electronic ground state to an upper vibrational state of an excited electronic state which has the most overlap of the wavefunctions to the lowest ground vibrational state (see Figure.4). Shortly after the initial electronic excitation the nuclear coordinates of the molecules will shift into their equilibrium position and the electron will radiation less decay into the lowest vibrational state. Then the electron can relax to the ground state. This decay may be radiative or non-radative. The energy of the emitted photon is higher than the energy of the absorbed photon (Stokes-shifted) because of the non-radiatve relaxation of the excited electron. 11

33 Figure.4: Diagram for the energy vs. configuration coordinate of the absorption and emission between two electronic states S 0 and S 1 of an organic molecular material. Each electronic state includes several vibronic modes. The arrow Q represents the non-radiative transition from the highest excited vibronic mode to the lowest LUMO level. The Frank-Condon transitions are the vertical arrows E and P. In general, excited vibrational states are not fully overlapping with the initial ground state vibrational wavefunction 46,48. The multiple transitions that are allowed will create a series of peaks in the fluorescence spectrum. These peaks are often not resolvable due to weaker, rotational (in solution) or vibronic modes. According to the Condon approximation the relative transition intensities of the peaks are proportional to the magnitude of the electronic transition dipole moment among the initial and final electronic states ( ' ) and the overlapping of the 1

34 vibronic modes of these two electronic states ( ( ', ) ). ( ', ) is called the Franck-Condon factor 48,49. The emission process is similar to the excitation process, but the transition occurs from the upper vibrational state to the lover vibrational (ground) state. Stokes shift absorption PL energy [ev] Figure.5: Stokes-shift between absorption and emission spectra is shown in the diagram. Each vibronic transition in either case contributes to the spectrum Decay and quantum efficiency of the fluorescence in organic materials If we ignore the other effects like intersystem crossing then the rate of decay of emitted photoluminescence is generally given by 51, 13

35 dn dt e ( ) N r nr e (.) where r and nr are the radiative and non-radiative decay rates and N e is the number of excited states (i.e. excited molecules) present at time t. Then the decay time is given by, r 1 nr r nr r nr (.3) The photon emission rate is given by, dn dt (.4) r N e where N is the number of photons emitted at time t. The quantum yield (i.e. fluorescence yield) of the emission is given by, where r r (.5) nr Φ r nr r r nr 1 is the life time of the luminescence if there were no non-radative processes. r.5 Selection rules in molecular systems Even though the electronic transition probabilities are proportional to the electronic dipole moment among the initial and final states of the molecular system, not all such transitions are allowed and the electronic transitions are governed by certain selection rules 35,5,53. er e 1 r In a two electron system the electric dipole moment operator could be written as. Then the dipole moment for the transitions among states of different permutation symmetry could be calculated using 54, * e r1, r )( r1 r ) ( r1, r ( ) d d 1 (.6) 14

36 In equation (.6) if the labels were permutated the sign of would change. Because the value of should not depend on how we label the states, the only possible value for should be zero. This suggests that transitions between symmetric and between anti-symmetric states are forbidden 33,54. Now if we consider the spin of a two-electron system, since the electron is a spin ½ particle the spin of the system may be either 0 or 1. Also the corresponding state of the spin 0 (S = 0) may be written as 54, 1 ( 1,) u (1) d() d(1) u() (.7) where u denotes the spin up state and d denotes the spin down state of the electron in the two electron system. We see that under particle exchange the is anti-symmetric, The spin 1 (S=1) state can be written as 54, 1,) (,1) (.8) ( ( 1) 1 (1,) u (1) u() (.9) (0) 1 (1,) u (1) d() d(1) u() (.10) ( 1) 1 (1,) d (1) d() (.11) All of these states (i.e. (.9)..(.11)) are symmetric under particle exchange (i.e. triplet). According to the Pauli principle the total wave function should be anti-symmetric under particle exchange 33,55. The total wave function consists of both the spatial ( ) and spin ( ) components. In order to conserve Pauli principle the spatial and spin components should have opposite symmetry. Therefore the initial and final spin components should have the same symmetry

37 Therefore the allowed transitions are from singlets to singlets and triplets to triplets. The transitions from singlets to triplets and from triplets to singlets are forbidden. Even though the molecular states are expected to follow this selection rule, it is not always competently true. The reasons for this (i.e. effects of spin-orbit coupling) will be discussed in the following section. The ground state wave-functions of most molecules are spatially symmetric under particle exchange 56. This is because the in these highest occupied molecular orbitals are completely full. Therefore the molecular ground states of most stable molecules are singlets 57. Because of this fact excited singlet states are allowed to radiatively decay but excited triplet states are forbidden to radiatively decay to the ground state. The radiative decay of the singlet states and resulting emission of photons is called florescence. Even though the radiative decay of triplet states is forbidden they show a weak radiative decay due to a mechanism called intersystem crossing. The decay of a triplet state is slow (in the ms time range) compared to the singlet decay (in the ns time range). The triplet emission is called phosphorescence. Figure.6: Energy diagram for fluorescence (from singlets) and phosphorescence (from triplet states). The singlet to triplet energy transfer is called the intersystem crossing (ISC). 16

38 The change of total energy needed for spatially symmetric transitions (singlet) is given by 54, E e 1 d 1d 4 0 r (.1) 1 The change of total energy needed for spatially anti-symmetric transitions (triplet) is given by 54, E e 1 d 1d 4 0 r (.13) 1 Equation (.1) and can be rewritten as, E e r 1 a (1) b() b(1) a() { a(1) b() b(1) a()} d 1d (.14) E J K (.15) where a and b are the individual wave function of each electron of the two electron system and J and K are the coulomb and the exchange integrals defined by, J e 1 a(1) b() a(1) b() d 1d 4 0 r (.16) 1 K e 1 a(1) b() a() b(1) d 1d 4 0 r (.17) 1 In a similar fashion, equation (.13) and can be rewritten as, E e r 1 a (1) b() b(1) a() { a(1) b() b(1) a()} d 1d (.18) E J K (.19) 17

39 According to equation (.19) the triplet state (which has a spatial wavefunction that is anti-symmetric under electron exchange) has lover energy (see Figure.6), when compared to the singlet state energy in (.15). Because of this it makes is possible to have an energy transfer from singlets to triplets and this type of transfer is called Inter-System Crossing (ISC) 58,59. The energy difference between the triplet and singlet states is referred to as exchange energy When the electrons are transferred from the signet to the triplet there is a flipping of the spin. Most of the instances this is caused by the process called the spin-orbit coupling 16. Spin-orbit coupling causes a radiation less transfer of a singlet to a triplet state. Because of this the multiplicity of the state is changed. The intersystem crossing obeys the Frank-Condon principle and occurs at the intersection of the potential energy surfaces of the two electronic states of the molecules 54. At this intersection the vibronic wave-functions of the two electronic states will match to each other 54. Inter-system crossing occurs strongly when heavy atoms are present in the molecule due to the large spin-orbit coupling in them 54,59. Because of the intersystem crossing the electron will continue to thermally degrade its energy and once it is lowered down to the triplet s vibrational ground state of the molecule it will become trapped. Since the triplet to singlet transition is forbidden the molecule is unable to decay to the ground state. Anyway this statement, that molecule cannot decay to ground state is not completely true and indeed the molecule will decay slowly by the intersystem crossing mechanism. In other words the transition is weakly allowed and the state weakly radiates its excess energy due to spin orbit coupling enabling this weak transition. The photons emitted by this kind of transition are labeled as phosphorescence. Since the triplet decay is inefficient most of the triplets decay nonradiatively. The triplets that decay radiatively will contribute to the phosphorescence. 18

40 .6 Electron and energy transfer in organic materials Excitons found in organic materials, especially in amorphous materials are small radius Frenkel excitons (see previous section for types of excitons). Frenkel excitons are in general localized on a single molecule 6,6. The Frenkel excitons that are created through optical excitation of an organic material can migrate from one molecule to the neighboring molecules using different energy transfer mechanisms. The electron and the hole in the excited molecule can hop to another neighboring molecule if an overlap of the molecular orbitals between the molecules exists or energy can be transferred through long-range resonant energy transfer mechanism called the Förster energy transfer 5,46. The energy transfer due to the simultaneous hopping of the electron and the hole to an adjacent molecule is described by the Dexter energy transfer 63,64 (see Figure.7). The transfer mechanisms will be further explained in the following chapters. Figure.7: Excitonic energy transfer in molecules: (a) Förster energy transfer (b) Dexter energy transfer. In both cases energy is transferred from singlet exciton D* to molecule A. The final states are shown in (c). 19

41 .6.1 Förster resonance energy transfer Förster resonance energy transfer (FRET) describes the non-radiative energy transfer from the donor molecule to the acceptor molecule as 58,65, D * * A D A (.0) Because of the dipole-dipole interaction between the electronic states of the donor and acceptor molecules there will be an electronic energy transfer from one molecule to the other. This type of energy transfer was first described by Förster 58,66. The energy transfer process does not involve any actual emission or re-absorption of a photon the optically induced electronic coherence on the donor molecule is resonant with the electronic transition energy of the acceptor molecule 67,68. The magnitude of the transition dipole interaction determines the strength of the FRET energy transfer. The dipole strength depends on the distance between the dipoles, the orientation of the dipoles, and the magnitudes of the donor and acceptor transition matrix elements. The separation of the donor and acceptor molecules can be determined by using a 6 1/ r dependence of the distance between the dipoles. To explain the FRET energy transfer the electronic ground and excited states of the acceptor and donor molecules should be considered. The absorbed photon energy by the donor is rapidly relaxed non-radiatively through vibrational modes to the lowest vibational state usually within few picoseconds (see Figure.8). For the resonant energy transfer to occur it is required that the emission spectrum of the donor molecule should significantly overlap with the absorption spectrum of the acceptor molecule. Also it is necessary that the two molecules are located at a close proximity (i.e nm 69 ). Once the energy is transferred to the acceptor it will emit spectrally red shifted fluorescence after the non-radiative vibronic relaxations. 0

42 Figure.8: Förster energy transfer from a donor molecule (singlet) to an acceptor molecule. The vibrational relaxations of both molecules are indicated. The interaction energy between the two dipoles of the molecules can be written as 46, 1 * * kl R 3[ ][ ] kl M 5 k.ml Mk.Rkl M l.rkl (.1) n R where n is the refractive index of the medium, kl R kl is the vector between the molecular centers of gravity of molecule k and l. The asterisk denotes the complex conjugate of the quantity. The matrix element 46 is defined by Förster as, M k e * φ (r ) r φ (r )dr k k k ' k k k (.) where the center of gravity is selected as the origin of the electrons position vector r k and φ k is the total wavefunction of the electron. 1

43 when the energy transfer occurs slower as compared to the rotational molecular motion then the rate of energy transfer should be averaged over all orientations of both molecules. For this reason, Förster introduced the average rate of energy transfer given by 46, W MkM 4 l (.3) 3n R 6 kl The index k and l can be neglected because the molecules are identical. Assuming a single donor and single acceptor separated by a distance r, the rate of the resonance energy transfer is given by, W MkM 4 l (.4) 3n R 6 kl The R -6 dependence of the rate of energy transfer and the square of the matrix elements in the expression were obtained by Förster using Fermi s Golden rule 46. Furthermore Förster used the Dirac s quantum mechanics to change the expression to form of experimentally accessible quantities. Förster defined the critical molecular separation as R 0. Instead of using Förster s original equations, using modern variables the resonance energy transfer rate between a single acceptor and donor that is separated by a distance r is given by 70, k T QDk 9000(ln10) ( r) F r 18 Nn D 0 D 4 ( ) ( ) d A (.5) where Q D is the quantum yield of the donor in the absence of the acceptor molecule, N is the Avogadro s number, D is the lifetime of the donor molecule when the acceptor is absent. The

44 F D () term in the integral is the corrected florescence intensity of the donor molecule in the range of wavelength to, when the total intensity is normalized to unity. A () is defined as the extinction coefficient of the acceptor at wavelength, the k is the orientation factor which describes the orientation of the donor molecule with respect to the acceptor molecule, n is the refractive index of the medium, r is the distance between the donor and acceptor molecules. The value of orientation factor k is usually determined by experimental conditions and the value introduces an error in the calculated Förster distance R 0, if not properly determined. The spectral overlap between the acceptor and donor molecule is given by 46, The Förster distance R 0 is given by, J ( ) 0 4 FD ( ) A ( ) d (.6) 0 F( ) d R (ln 10) Q Nn D 0 F D 4 ( ) ( ) d 0 F D A ( ) d (.7) at r = R0 the rate of resonance energy transfer is, 1 k( r) k( R0 ) (.8) D where 1 is the rate of decay of the donor molecule when the acceptor is absent. D 3

45 The simplified expression for the rate of Förster energy transfer could be written as, k 6 1 R0 ( r) D r (.9).6. Dexter energy transfer In Dexter energy transfer theory, electrons are exchanged between the donor molecule and acceptor molecule through a non-radiative process 5,6. In such cases a double electron transfer occurs between the donor and acceptor molecules. The resonant energy transfer can occur between an allowed transition in the donor molecule and a forbidden transition in the acceptor molecule 71. In such event an acceptor molecule that will not respond to a provided radiation spectrum will be luminescent after the transfer from the donor. The Dexter transfer process requires that the wavefunctions of the acceptor and donor molecules to overlap, it is a very short range process (< 1 nm). The rate of energy transfer (which relies on electron exchange) depends exponentially with the separation on the donor and acceptor molecules 7, k D RDA L KJe (.30) where k D is the rate of Dexter energy transfer between the molecules, J is the normalized spectral overlap integral (see equation (.31), R DA is the separation of the donor and acceptor molecules relative to their Van dar Waals radius or the Bohr radius L. According to this definition RDA refers to the edge-to-edge separation of the molecules. 4

46 J I 0 D A d (.31) where 1 the frequency, I D is the emitted light intensity and A is the extinction coefficient of the acceptor. The temperature dependent electron energy transfer rate is given by 64, k e H AD ( G ) 4k T 4k T b e b 0 (.3) where H AD is the electronic coupling matrix between the initial and final states, is the reorganization energy, absolute temperature. 0 G is the Gibbs free energy, is the k b Boltzmann constant and T is the The temperature dependent Dexter energy transfer rate between the acceptor and donor molecules can be written as 64, k D H AD 4k T b e G kbt (.33) where H AD is the electronic coupling matrix between the donor and acceptor states and 0 ( G ) G. 4 Dexter energy transfer is very important in explaining the energy transfer to triplet excited states. When the donor and acceptor molecules are very close to each other and when the transitions are spin-forbidden the exchange interaction allows triplet energy transfer from the 5

47 donor to the acceptor molecule (i.e. ISC). Because of this there are four possible energy transfer processes from an excited singlet or triplet donor molecule to an unexcited acceptor molecule as indicated in equations (.34) to (.37), where S and T stands for singlet and triplet states respectively. The ( * ) resembles an excited state. S S * d * d S S S (.34) a a d d * a S S T (.35) * a T T * d * d S S T (.36) a a d d * a S S S (.37) * a.7 Exciton migration in organic solids Usually in organic solids exciton migration is described by a random walk and when many steps are involved in the walk, the diffusion formalism is used 73. The size of the mean free path determines the form of the diffusion coefficient. The exciton will hop incoherently between the molecules if the mean free path is close to the intermolecular distance of the neighboring molecule 73,74. Then the diffusion coefficient will be derived in terms of molecular spacing of the nearest neighbor and the hopping time of the exciton. If the nearest neighbor distance is smaller than the mean free path of the exciton then the exciton motion is coherent and the exciton will move many molecular spacings before getting scattered and the exciton band model is more applicable in this situation 75. Then the exciton diffusion coefficient can be derived in terms of the free path length and the exciton velocity. In general for organic solids at room temperature the singlet excitons are thought to be incoherent hopping exciton. In ultrapure crystals with no 6

48 impurities and at very low temperature so that no phonons are activated coherent exciton motion is possible 75. The exciton diffusion equation is given by 3,73, where n n t I( t) n D n (.38) is the exciton concentration at time t, I (t ) is the pumping light intensity on the sample, is the linear absorption coefficient, is the mono-molecular decay rate (both radiative and non-radiative) and D is the exciton diffusion coefficient. The time-dependent energy transfer rate from a donor to an acceptor is given by 76, k( t) 4 DrN [1 r /( Dt) 1/ ] (.39) where N is the molecular concentration and r is the radius of a molecule. The time dependent rate k(t) has been experimentally determined by high-resolution time-resolved studies of sensitized fluorescence 73. The time dependence of the energy transfer rate is related to time dependence of the exciton concentration in the diffusion equation (.38). For time resolutions greater than 10 ps, the experimental data could be explained using the timeindependent energy transfer rate 73. If the mean free path (or the scattering length ) of the exciton is greater than trapping radius then equation (.39) is not valid and transfer rate constant is given by 77, k d ( t) N (.40) where is the velocity of the excitons and is the trapping cross-section. 7

49 Organic Solid Multiplicity (of the exciton spin) Lifetime Intermolecular Distance (nm) Diffusion coefficient (cm /s) D Alq3 film Singlet 16ns [78] [78] 4 x 10 Alq3 film Triplet 5 s [79] 8 x 10 TPD film Singlet 1.89 ns [ x 10 Anthracence Singlet 10 ns [80] 0.5 x crystal Anthracence crystal Tetracene crystal Tetracene crystal Triplet 10 ms [75] -3 [75] x 10 Singlet 300 ps [81] -3 [81] 0.5 x 10 Triplet -3 [78] 4.6 s [8] -4 [8] x 10 Table.1: Comparison of exciton diffusion properties of some known organic solids. The diffusion parameters of several crystalline and amorphous organic solids (including Alq3) are listed in Table.1 for comparison. The exciton could make a larger number of hops before it finally relaxes to the ground state. We will show in Chapter 5 that the above mentioned diffusion theories found in literature are not directly applicable for our Alq3 amorphous films. This is due to the directional nature of the exciton migration in our Alq3 film samples. The above description was provided as a literature review of the current understanding of the exciton dynamics organic solids. When the long-range energy transfer between a donor and acceptor molecules dominates then the intermolecular electronic coupling becomes very weak. This type of long range transfer is referred by many names such as Förster energy transfer, mechanical resonance transfer or inductive resonance transfer 73,83. This Förster type energy transfer has been explained in a previous section. 8

50 .8 Interactions between excitons As explained in the previous sections the exciton becomes mobile due to interactions among the molecules. Because of this the excitons may collide with different particles during its life time. The collisions may occur with same kind of species or other different kinds particles like free charge carriers. When such collisions occur it will either cause an annihilation of the energy of the exciton or transfer of energy to the colliding particle. These interactions will result in a change in the optical and electronic properties of the materials that can be observed experimentally. The interaction between two singlet excitons is usually called singlet exciton fusion, annihilation or bimolecular quenching. In an exciton fusion process first a highly excited (hot) singlet state is produced ( S * n ). This interaction is represented by the following equation, S * 1 S1 S0 S n ss (.41) This highly excited singlet state can decompose using several pathways as indicated below, * S n S1 heat S0 h (.4) * S n e h (.43) * S n * 1 * 1 T T (.44) In reaction (.4), the highly excited state will non-radiatively relax into an excited singlet state (vibrationally relaxed singlet state) and then this regular singlet exciton can recombine radiatively (emitting a photon). If the highly excited state is excited above the ionization energy of the molecule then an electron hole pair (i.e. charge carriers) will be created 9

51 (i.e. auto-ionize) as indicated in equation (.43). In addition to these two methods of relaxation the highly excited singlet, it may break (i.e. fission) into two triplet states 6,84 as indicated in equation (.44). The kinetic equation for exciton-exciton interaction (fusion) in the case of cw excitation is given by 85,86, ds dt 1 I k S f S 1 b S 1 (.45) where k s is the intrinsic inverse lifetime of the singlet exciton, b is the singlet exciton annihilation rate (bi-molecular fusion) constant, I is the light intensity of the exciting laser light, is the S 1 exciton concentration at time t, and is the linear absorption coefficient of the organic material. The pre-factor f takes into account auto-ionization and complete non radiative decay of the highly excited excition. 30

52 Chapter 3 3 Organic film growth and thickness monitor calibration 3.1 Introduction to used organic materials Introduction to Alq3 Tris(8-hydroxyquinoline)aluminum(III) (Alq3) is used as an efficient emissive layer in Organic Light Emitting Devices (OLEDs). The chemical structure of Alq3 is shown in Figure 3.1. These OLEDs are used in modern flat panel displays. Despite extensive research in the past few decades on OLEDs, Alq3 is yet the most effective material in the low molecular weight category for these devices. In addition to being a green emissive layer, it also is a good electron transport layer in OLEDs. Alq3 is also is used as the host material for producing materials which emit different colors from green to red. Most of the studies carried out on Alq3 are related to increasing the device efficiency, studying the charge transport properties, or increasing the stability and lifetime of the devices. According to these investigations the charge transport properties of Alq3 films were explained using a hopping model of the charge carriers and the temperature and field dependence of the mobility was explained using Frenkel-Poole like models 87. Also the different effects induced by the distributed trapped states, especially at low fields have been investigated. These traps are formed due to the presence of a mixture of isomers (see Figure 3.) with different energy levels of the Alq3 molecule, self-trapping, and impurities. In general, even though Alq3 films used in OLEDs are considered amorphous it is supposed that the material has nanocrystalline domains within the film structure and referred to as quasi-

53 amorphous. Different morphologies (i.e. α,β, etc., see Figure 3.3) within the Alq3 film will affect the optical and electronic properties and also the charge transport properties of the film. Figure 3.1: Chemical structure of tris(8-hydroxyquinoline) aluminum(iii) (Alq3) (from 16 ) Figure 3.: Different isomers of Alq3 (from 16 ) 3

54 Figure 3.3: Crystalline packing in different phases of Alq3: (a) Alq3(C 6 H 5 Cl) 1/ (b) Alq3 α- phase; and (c) Alq3 β-phase (from 88 ). The two isomers in Alq3 are named facial and meridonal isomers as shown in Figure 3.. The different isomers have different dipole moments being higher in the facial isomer compared to the meridonal isomer. Because of this larger dipole moment the morphology and charge carrier injection is affected by the facial isomer. Due to the difference in their HOMO and LUMO band gap in these isomers it is believed that facial isomers can act as traps to the charge carriers. Because currently there is no experimental proof of the existence of facial isomers it is assumed that the meridonal isomer is dominant in both Alq3 amorphous films and in crystals. But it is under discussion that the different isomers in an Alq3 film will affect the trapping density and charge carrier transport

55 3.1. Introduction to PTCDA PTCDA is a perylene derivative, which is a pigment commercially available. The PTCDA molecules are shown in Figure 3.4-(a). In Figure 5-(b) and (c) the electron density distribution of the HOMO and LUMO level respectively are shown. The blue and red colors represent the polarity of the molecular orbital coefficients and the interactions occur with same colors only. (a) (b) (c) Figure 3.4: (a) single PTCDA molecule (b) HOMO 90 (c) LUMO 90, energy in PTCDA. The blue and red colors represent the polarity of the molecular orbital coefficients. PTCDA crystals are formed by the placement of individual PTCDA molecules in a specific order (see Figure 3.5). The monoclinic crystalline structures of PTCDA have two different crystalline phases called α and β 91 (see Figure 3.6). The PTCDA molecules are stacked along the a-direction during growth. The molecular stacks in PTCDA crystals form a layered structure and the molecules of this structure are parallel to the (10) lattice plane. The molecules are slightly tiled by an angle of 34

56 11 0 with respect to the symmetry axis of the structure. The distance between the (10) planes in the PTCDA crystals are different than the distance between the molecules in the (10) plane. (a) (b) Figure 3.5: Sketch of PTCDA crystals (Figure 3.5 (b) from 9 ) α-ptcda (a) (b) Figure 3.6: Different crystalline phases of PTCDA films. β-ptcda (c) Because of this the interaction overlap of the π-orbitals in the highest occupied molecular orbitals are larger in PTCDA crystals. This causes a large anisotropy in the PTCDA films. The α 35

57 and β phases of PTCDA molecules have slight differences in there crystalline parameters (see Figure 3.6). 3. Intermolecular transitions between molecules In organic solids there are interactions with the neighboring molecules because of their close proximity with each other. These interactions modify the optical and electronic properties of the organic solid significantly, when compared to a single molecule and the energy levels may get split due to such interactions. Due to these kinds of intermolecular interactions so called excimers, and charge transfer excitons are formed (see Figure 3.7). It should be noted that bonding between the two molecules are not a strong chemical bonds. (a) (b) Figure 3.7: (a) Excimer (b) Charge transfer exciton between two molecules. A charge transfer (CT) exaction could be thought of as a cation-anion (C+C-) dimer. In an organic solid a CT excition is formed if the time that the exaction takes to transfer its energy to the neighboring molecule is larger than the time of the lattice deformation and the exciton gets self-trapped in this deformation. An excimer is an excited dimer and is formed from of two molecules where the electrons and holes are shared by the two adjacent molecules. In an excimer the two molecules normally 36

58 would not be bind if they were in their ground state. The molecules in Pyrene form excimers and therefore Pyrene is also a good example for an organic material that forms excimers. An excimer can also be considered a small radius self-trapped exciton state. 3.3 Origin of traps in organic semiconductors Excitons in organic semiconductors get localized at so called traps. Traps in organic material may be formed because of the following reasons. (1) Defects in the structure: The HOMO/LUMO energy levels of a molecule are not only determined by the molecules chemical structure itself, but also the electronic polarization of the surrounding molecules. Structural imperfections in films will cause local changes in the polarization which cause the excitons to get trapped at density-of-states (DOS) energy tails. () Impurities: If the HOMO-LUMO energy level of an impurity molecule is in the gap of a host molecule then the host molecule will form a trap state. (3) Self-Trapping of charge carriers and excitons: Usually the presence of a charge carrier in an organic molecule leads to the deformation of the molecule. This causes a lowering of the energy of this excess charge carrier. This deformation and the charge carrier are usually considered a quasi-particle which is called a polaron. A polaron is usually not a fully trapped state. But the mobility of a polaron is usually several orders less than a free carrier state and will affect the performance of the devices. 37

59 (4) Self-trapping of excitons due to lattice deformation: Self-trapping of an exciton may occur due to a self-induced relaxed lattice deformation in an organic crystal causing a negative trapping energy. 3.4 Organic molecular beam deposition (OMBD) The OMBD technique is method used to grow high quality and purity organic films. When there should be an epitaxial matching between the substrate and film the term OMBE 93,94 is used and for general deposition where no such matching is needed the term OMBD is used. This mechanism is similar to the MBE method used to fabricate inorganic films. Compared with other growth mechanisms OMBD has many advantages like: Ability in growing films with high purity Ability in controlling film thickness to very high precession Superior structural quality of films (crystal ordering) High vacuum makes it is possible to grow high purity films with OMBD because it provides an extremely clean environment during film growth. The mobility of charge carriers are ultimately limited by scattering and trapping which is increased by impurities and defects in the crystals 17. Therefore high purity and superior structural quality films are highly desirable in optoelectronic devices and other fundamental research applications. In order to observe the epitaxial crystal formation during OMBD growth, reflection highenergy electron diffraction (RHEED) method has been used 95. The ability to control the layer thickness is a very important aspect of OMBD and it is even possible to grow a monolayer of the organic film with this sophisticated method 96. In OMBD the source material is sublimated in a high vacuum chamber so that it will be deposited on a substrate. During the film growth the 38

60 molecules interact with the substrate and this interaction is usually the van der Waals type. Due to this interaction the final morphology of the film depends on the substrate and usually several morphologies are found in the same film (polycrystalline film). Even though the film structure is predominantly determined by the growth condition like temperature, pressure and growth rate the quality of the substrate surface plays a large role for the film morphology. 3.5 Organic molecular beam deposition technique (OMBD) and film growth The PTCDA and Alq3 films that were used in our experiments were grown using the OMBD method. The schematic diagram of the growth chamber is shown in Figure 3.8. In OMBD ultra (or very) high vacuum (UHV) is used to achieve films of different thicknesses including monolayer films. Before OMBD methods were developed, monolayers were achieved by the Langmuri-Blodgett method 97. In our OMBD growth the Knudsen cells of the chamber were filled with PTCDA and Alq3 powder purchased from Sigma-Aldrich Company. Metal gaskets were used to secure the cells into the chamber and backup pumps were used to decrease to a pressure level so that the turbo-molecular pumps (Pfeiffer vacuum) were able to operate properly. The chamber was then covered with an aluminum-foil and wrapped with a thermal strip to heat it to desired temperature ~ 95 o C for a desired period. This initial heating was done to outgas the chamber to reduce the partial pressure of impurities in the chamber. After this we are able to grow films until the materials in the Knudsen cells are depleted. The base pressure of the OMBD main growth chamber was maintained at ~ x10-8 mbar. 39

61 Figure 3.8: Schematic diagram of the OMBD growth chamber (from 13 ) The Pyrex glass or Si (100) substrates are first chemically cleaned using acetone, methanol, and ultra-pure water in an ultra-sonic bath. Then they are fixed onto a substrate holder and secured into the load-lock. After vacuuming the load-lock to a pressure of ~4x10-7 mbar, the substrates are transferred into the main growth chamber using the transferring rod and attached to the sample holder in the chamber. Then the x, y and z directional controls in the sample holder manipulator were used to bring the substrate to the appropriate growth position inside the chamber. After the pressure of the chamber has reached the necessary value, the temperature in the Knudsen cell is set to the appropriate sublimation temperature of the growth material using the temperature controller. For Alq3 this temperature is ~ 50 0 C and for PTCDA it is ~ C. Once the required temperature is reached a computer program was used to open the shutters in the Knudsen cells. The growth rates were kept to ~0.01 nm/s for both PTCDA and Alq3 films. During the film growth a quartz-crystal thickness monitor (MAXTEK TM-100) was used to 40

62 monitor and record the growth rate of the films as necessary. The OMBD growth chamber in our research laboratory is shown in Figure 3.9. Figure 3.9: Photograph of the OMBD growth chamber in our laboratory. 3.6 Reflection measurements and film thickness calibration Reflection measurements (Fabry-Perot oscillations) were used to determine the thickness of the OMBD grown Alq3 films and to calibrate the quartz-crystal thickness monitor in the OMBD system. After the reflection of the light by a film that is surrounded by air on both sides the intensity of the light is given by 98, 41

63 I F sin / (3.1) 1 F sin / r I i where 4 n I i and I r are the intensity of the incident and reflected light, t, and F 4R 1 R (3.) here R n n na n a is the reflectivity at the film interface for normal incidence. The absorption at the transparent region has been neglected, t is the thickness of the film layer, and n () is the in-plane refractive index at the wavelength. n a represents the wavelength of the surrounding air. If we consider the refractive index ( n (3.1) can be written as, s ) of the substrate a more detailed version of equation I F F Sin 0 1 r I i 1 F1 Sin / / (3.3) with 4 n I i and I r are the intensity of the incident and reflected light, t, and F 1 4 R 1/ R R 1/ 1 1 R 1 (3.4) and 4

64 F 0 R 1 1 1/ R 1/ R R 1/ 1 (3.5) where R 1 n n na n a is the reflectivity at the film/air interface for normal incidence and R n ns n n s is the reflectivity at the film/substrate interface. Any absorption in the transparent region has been neglected, t is the thickness of the film layer and n () is the inplane refractive index at the wavelength. n is the refractive index of air and a n is the s refractive index of the substrate (i.e. Pyrex glass). Also for Si (100) substrates the F.P. equations can be modified (considering a phase change by π at the Alq3/Si interface compared to Alq3/Pyrex glass) as, I F F Sin 0 1 r I i 1 F1 Sin / / (3.6) with and F F / R1 R R R 1/ 1 1 R 1 1 1/ R 1/ R R 1/ 1 (3.7) (3.8) 43

65 3.7 Experimental details of the reflection measurements For the reflection measurements experiment, a 50 Watt Halogen lamp was used as the light source and a light was focused on to the Alq3 film sample using several lenses as shown in Figure A fiber-optic cable was used to transfer the reflected light to the monochromater/photomultiplier combination. A personal computer (PC) with LabView was used to record the reflected signal from the sample. Figure 3.10: Experimental setup for F.P. reflection measurements (from 99 ). The experimental reflection data (red dotted line) obtained from an Alq3 sample on glass with a nominal thickness of 300 nm is shown in Figure The nominal thickness was derived using a calibration value that has been used before the quartz-crystal in the thickness monitor has been exchanged. Also the theoretical fitting for this experimental data (using equation (3.3)) is 44

66 shown by the solid blue line. The refractive index of Pyrex glass and air used were and 1 respectively. The refractive index values of Alq3 were taken from ellipsometric data that were available from the literature 100. The thickness from the calculations was determined to be ~850 nm. Therefore the thickness correction factor found for the nominal 300 nm thick Alq3 film is ~ nm Alq3 film on glass theory experiment intensity [arb. units] wavelength [nm] Figure 3.11: F.P. reflections of an 850 nm thick Alq3 film on glass. 45

67 For a similar 300 nm Alq3 sample grown on Si(100) the obtained experimental reflection data (dotted red line) is shown in Figure 3.1. The theoretical fit (solid blue line) obtained using equation (3.6) is shown as blue curve. The refractive indexes used for Si(100) and air were 4 and 1 respectively. This fitting resulted in a thickness of the Alq3 film of ~865 nm. The resulting thickness correction factor for the quartz crystal calibration in this case is ~ nm Alq3 on Si(100) theory experiment intensity [arb. units] wavelength [nm] Figure 3.1: F.P. reflections of an 865 nm thick Alq3 film on Si(100). In a similar way a nominal 800 nm thick Alq3 sample both on Pyrex glass and Si(100) has been investigated by reflection measurements and the obtained experimental and theoretical 46

68 results are presented in Figure 3.13 and Figure 3.14, respectively. The refractive indices used for the theoretical simulations were the same as for the previous Alq3 samples. The obtained real thicknesses using the theoretical fitting are 1835 nm and 1970 nm for Pyrex glass and Si(100) substrates respectively. Accordingly, the correction ratios for the thicknesses calibration were found to be.9 and.46, respectively. intensity [arb. units] nm Alq3 on glass theory experiment wavelength [nm] Figure 3.13: F.P. reflections of a 1835 nm thick Alq3 film on glass. 47

69 nm Alq3 on Si(100) theory experiment intensity [arb. units] wavelength[nm] Figure 3.14: F.P. reflections of a 1970 nm thick Alq3 film on Si(100). Using the reflection measurements the quartz-crystal thickness calibration was corrected by a factor of.6. The proper calibration of the thickness monitor and the exact knowledge of the thickness of Alq3 films are important in our optical characterization experiments, since incorrect Alq3 film thicknesses may lead to incorrect and misleading interpretations and conclusions. 48

70 Chapter 4 4 Refractive index dispersion in polycrystalline PTCDA waveguides 4.1 Introduction According to the Moore s law the speed of the microprocessors fabricated using electronic components will double every 1.5 years under current technology 101. Butter's law of photonics states that the speeds of devices based on fiber optics components wills double every ~ 9 months. Because of this rapid development integrated optical counterparts for integrated electronic circuits (IC s) and devices have become very important. Integrated optical circuits (IOC s) are similar to electronic integrated circuits (IC), but instead of using electronic components, optical components are used in these miniature devices. Miniature optical components of lenses, beam splitters, lasers, and waveguides are used in these IOC s. Even though there are no truly monolithic IOC s available yet, it has been demonstrated that hybrid versions of optical components and electronic circuits are possible to fabricate. In IOCs the light is propagated from one point to another using for instance planar dielectric waveguides. In this type of film geometry, a rectangular region with a higher refractive index is sandwiched between a material of lower refractive index and monochromatic light is propagated through the central region. Current problems occurring in IOCs are large propagation losses in the waveguide, coupling losses (between optic waveguide and optical fiber), and propagating the light at the sharp bends of the miniature film waveguides. These problems might be overcome using organic semiconductors or hybrid metal/organic films as waveguiding materials. One potential organic material is PTCDA which forms highly anisotropic polycrystalline films with strongly different refractive index values parallel and perpendicular to the molecular 49

71 stacking direction. In this chapter I investigate the dispersion of the in-plane and normal refractive index of high-quality vacuum deposited PTCDA waveguides. As in an earlier work 99 light coupling will be achieved using the prism coupling method and the effective refractive index values will be determined with the correlated m-line technique. These investigations will confirm the high-quality of our OMBD grown PTCDA polycrystalline films and demonstrate the effectiveness of the m-line method which allows both the evaluation of effective refractive index values for TE and TM modes and the determination of the thickness of the organic waveguide. At the end of this chapter I will compare our obtained values with existing ellipsometric data and with values obtained from different waveguide studies. 4. Introduction to prism-coupling and waveguiding Guided modes occur when the bulk refractive index n of the film is higher than both the refractive indices of the substrate n s and cover layer n a resulting in a total internal reflection at the top and bottom interfaces of the waveguide (see Figure 4.1). Both transverse electric (TE) and transverse magnetic (TM) modes exist propagating energy through the planar film waveguides. The m-line (dark-line) technique is one of the commonly used and versatile method to analyze the modes in a planar waveguide. In this method the light is coupled into the planar film waveguides using a prism coupler. A detailed description of this method and of waveguide theory can be found in Here I will only summarize the most important outcomes. The prism is a high refractive index material and the light is coupled into the waveguide with the help of evanescent waves (See Figure 4.1). 50

72 Figure 4.1: Two prism couplers used to feed and extract light at two ends of a planar film waveguide. In Figure 4., the prism coupler is used to detect guided modes in the waveguide. A detection screen (or a photo detector) is used to detect the laser beam that has been totally reflected from the prism base. A typical signal at the detector of the reflected light intensity is shown in Figure 4.3 as a function of the incident angle of the beam on the film. When light is coupling into the waveguide the intensity of the reflected beam is reduced. In Figure 4.3 four propagated modes are schematically shown. Coupling only occurs when the incident angles reach specific values (also called mode angles) when the phase velocity of the light beam in the prism along the interface is equal to the phase velocity in the film-waveguide. This is known as the phase synchronism and is expressed by the following equation: h k n sin( ) (4.1) 0 p γ where h is the longitudinal propagation constant of the waveguide mode, n p is the refractive index of the prism material, and is the incident angle of radiation on the prism base. k 0, where is the wavelength of the light. At this incident angle photons from the laser beam tunnel through the small air gap into the film. The significant drop in the reflected intensity sketched in Figure 4.3 is a result of this mode coupling and thus this method is called the dark-line (or m- line) technique. 51

73 Figure 4.: A prism coupler is used to detect the coupling of propagating modes in a waveguide. In the film the light is guided by optical propagating modes. Using the Maxwell s equations and boundary conditions, the equations for TE and TM modes can be written as 105,106, t n eff tan neff n a 1 neff n s 1 tan m n n eff n n (4.) eff k 1 n n eff where t is the thickness of the sample, 1 and are both 1 for TE modes and for TM modes n they are n a n and n s respectively. n a is the refractive index of air and n and n s are the refractive index of the film (waveguide) and the refractive index of the substrate, respectively. Also k, where is the wavelength of the incident light. m = 0, 1,, is the mode number which represents the number of nodes of the standing electric or magnetic field normal to the waveguide surfaces

74 Figure 4.3: Sketch of a typical signal obtained from the reflected beam intensity in a prism coupling experiment. The synchronous angles m as a function of the mode numbers m can be derived from the incident angles i (see Figure 4.5) at the face of the prism with angle α (=45 in our case) from the relation r m and using Snell s law of refraction, m sin i sin 1 (4.3) p n Hence, the effective refractive indices n eff of various modes m can be obtained by measuring the incident angle i of each dark m-line that appears in the experiment using 53 p 1 p n n sin sin sin / n (4.4) eff i

75 The refractive index p n of our used birefringent Rutile prism has two values. If the electric field is oscillating parallel to the c-axis (z-axis in Figure 4.5) of the prism TE modes TE TM θ i <0 θ i >0 n eff Thin film n p θ θ m r βm y x α are excited and the extraordinary refractive index p ne of Rutile is used to Figure 4.4: Sketch of the rutile prism on the waveguide showing the incident and coupling angles determine the effective refractive indices n eff. If the electric field is oscillating perpendicular to the c-axis of the prism TM modes are excited and the ordinary refractive index n o p of Rutile is used to derive the effective refractive indices n eff. 4.3 Details of the experiment The OMBD method that was described in the Chapter 3, was used to fabricate the polycrystalline PTCDA waveguide on Pyrex glass substrates. The calibrated quartz crystal thickness monitor was used to deposit the necessary thicknesses of the films during the very high vacuum growth. The investigated PTCDA waveguide has a nominal thickness of 880 nm. 54

76 For the m-line experiments a He-Ne laser (at = 633 nm), GaAs based semiconductor lasers (at = 66 and 693 nm) and a tunable Ti: sapphire laser operated in cw mode (at wavelengths = 770, 810, 840, 870 and 910 nm) were used as light sources (see table 1). The light power was kept below 1 mw using neutral density (ND) Prism Thin film filters to avoid light induced heating effects and damage of the waveguides. The exciting laser beam was sent through pin holes in Figure 4.5: Photograph of the adjustable waveguide holder with Rutile prism. Also shown is the microscope objective lens. On the right side a photograph of a m-line in the reflected laser spot is shown. order to simplify the alignment and to shape the spatial laser beam profiles. The unpolarized He- Ne laser and semiconductor lasers were polarized with a sheet polarizer to selectively excite TE or TM modes in the PTCDA waveguide. The polarization of the Ti:Sapphire laser beam was rotated using a / plate. The laser beam was focused by a 50 objective lens (N.A. = 0.4) onto the Rutile prism. The prism and the PTCDA waveguide were mounted on a rotation stage with an angle resolution of ~30 arc sec (see Figure 4.5). The reflected beam from the prism base was monitored on a screen. The photograph in Figure 4.1 shows an m-line at a certain coupling angle. 55

77 Wavelength [nm] Used laser 633 He-Ne laser 66 and 693 GaAs based semiconductor lasers 770, 810, 840, 870, and 910 tunable Ti: sapphire laser operated in cw mode Table 4.1: Different lasers used in the m-line experiments. effective refractive index n eff TE 0 TE 1 TE TE 3 TE 4 TM 0 TM incident angle i Figure 4.6: Effective refractive index values for TE and TM modes. The values are calculated with equation (4.) for different light coupling incident angles and at different wavelengths ranging from 633 to 910 nm. Effective refractive index values which belong to the same mode number are plotted in the same color. 56

78 4.4 Experimental results and discussion Using the m-line technique described in the above paragraphs the incident angle i associated with each mode m was measured. The effective refractive index values ( n () ) as a function of i for TE and TM modes were evaluated for each wavelength using eq. (4.4) and graphed in Figure 4.6. Effective refractive index values which belong to the same order (as labeled) are given in the same color. The extraordinary and ordinary refractive index values of the prism n p in equation (4.) were determined from the given dispersion in 108,109. Also shown eff in Figure 4.7 are the n eff curves for TE and TM modes resulting from eq. (4.4). The first curve of each curve set corresponds to a wavelength of 633 nm and the following curves are for 66, 693, 770, 810, 840, 890 nm in descending order. In these measurements the uncertainty of i was better than 1º and thus the uncertainty of n () was less than eff By inserting the calculated effective refractive index n () values from eq. (4.4) in eq. (4.), the thickness of the sample (i.e. of the film-waveguide) and the bulk refractive index n () were determined using a graphical fitting method as shown for = 66 nm in Figure 4.7. The TE modes are represented by solid black circles and the TM modes are represented by solid red circles. The calculated curves (black lines) for various m-numbers in Figure 4.7 were obtained from eq. (4.3) by varying the bulk refractive index until the best agreement for the TE and TM eff modes with the set of curves has been achieved. The refractive indexes, n 1 a (for air) and n s 1.47 (for the Pyrex substrate) were used to calculate the curves with eq. (4.3). Using these procedure, the bulk in-plane refractive index at a wavelength of 66 nm is found to be n =.3, and the normal refractive index (i.e. n of the TM modes) is found to be n = Also the 57

79 thickness t of this wavelength was determined to be 810 ± 5 nm. The obtained value of n is significantly lower than the n (in-plane refractive index), because of the larger stacking distance of molecules along the 10 growth direction of the PTCDA films. The in-plane refractive index ( n =.3) value is in reasonably good agreement with the values n. 110 and obtained from polycrystalline PTCDA films deposited on glass and n. 39 and.4 11 from PTCDA films deposited on Si and GaAs substrates, respectively. Also the obtained n value is in reasonable agreement with ellipsometric investigations with n = A comparison of these values is shown in Figure 4.8. Likewise, we performed m-line measurements at excitation wavelengths = 633, 693, 770, 810, 870 and 910 nm and evaluated the inplane n and the perpendicular refractive index n as described above. The average thickness of the PTCDA waveguide obtained from all measurements resulted in a value of 805 ± 10 nm. For wavelength above 66 nm only one TM mode has been detected. The perpendicular bulk refractive index has been determined by using the average waveguide thickness (805 nm) in these cases. Figure 4.8 shows the derived refractive indices as full ( n ) and open ( n ) circles. The circles are connected by a fitted curve which has been calculated using a Sellmeier like equation, n C A B /( ) (4.5) with parameters A, B and C ( in m) being summarized in table Table 4.. For comparison the n and n values and dispersion curves derived by other groups are also presented in Figure

80 effective refractive index n eff = 66 nm thickness [m] 0 TE modes TM modes Figure 4.7: Observed effective refractive index values n eff of TE modes (black circles) at = 66 nm. Red circles show observed n eff values of TM modes. Solid and dashed lines represent calculated values of n eff as a function of the waveguide thickness for different TE and TM modes as labeled. A B C n n Table 4.: The parameters A, B and C in Sellmeier equation ( in m) to calculate the in-plane n and normal n refractive index dispersion in PTCDA polycrystalline films. Our in-plane n refractive indices are in good agreement with ellipsometric investigations on PTCDA films of Refs. 11 (dashed green line) and 111 (dashed-dotted black line) 59

81 but show a deviation compared to waveguide investigations at = 830 nm reported in Ref. 114 (black star). We also find deviations compared to the n dispersion obtained from ellipsometric investigations on PTCDA films on glass 110 (short-dashed black line). Our slightly reduced n values and weaker dispersion above = 700 nm as compared to data reported by Friedrich et al. 11 is attributed to slightly higher structural disorder in our PTCDA waveguides grown on Pyrex substrate compared to thin (~10 nm) PTCDA films deposited on Si. The increased structural disorder might be induced by higher surface roughness of used Pyrex substrates and by an increased number of grain boundaries in our ~1 m thick PTCDA waveguides. Both effects lead to an increased density of vacancies and voids which cause decreased n values and a weaker refractive index dispersion at wavelengths above = 700 nm. The higher n values found by Djurisic et al. 110 (short-dashed black line) might be caused by the same reason. Likewise, our higher n values compared to values found by Refs. 115 and 116 are explained by a lower film density and structural order of the used PTCDA films on glass and on Au-coated glass compared to our PTCDA waveguides. Our derived normal refractive index values n are also in good agreement with ellipsometric data obtained by Friedrich et al. 11 (dashed green line). Below = 700 nm, however, the n dispersion in our PTCDA waveguides is significantly weaker than the dispersion found in the ellipsometric investigations 11. This result indicates a reduced optical absorption along the PTCDA molecule stacking direction which might be again explained by the increased number of vacancies and voids at the grain boundaries of PTCDA crystallites in our thick waveguide structure. 60

82 refractive index n.5.0 n n wavelength [nm] Figure 4.8: Inplane n and normal n refractive index values (full and open blue circles, respectively) derived from m-line measurements in the spectral region between = 663 to 910 nm. The circles are connected by the Sellmeier like equation (4.5). For comparison, n and n values as well as refractive index dispersion curves derived by other groups are also given. Stars: Fuchuigami et al. 115, dashed green lines: Friedrich et al. 100, dashed-dotted black line: Djurisic et al. 116, short-dashed black line: Djurisic et al

83 Chapter 5 5 Nonlinear responses of organic materials and z-scan technique 5.1 Light interaction with matter Materials that are composed of organic molecules in general have very low conductivity 10,11 and also are non-magnetic 1,13. Organic materials have their electrons strongly bound to the atoms 14 constituting the molecules. When an electromagnetic radiation field is applied on such a medium the molecules become polarized in the medium and the linear polarization P is related to the susceptibility (1) χ by, P ε (1) 0 χ E (5.1) where E is the electric field and ε0 is the electric permittivity in free space. When the electric field is relatively small the polarization in the material increases linearly with increasing electric field according to equation (.). The polarization of the medium is related to the dielectric displacement D by, D 0 ε E P (5.) The dielectric displacement describes the effects of the electric field on the bound charges of the dielectric medium and is therefore used to explain the optical effects in organic media. The propagation and interaction of electromagnetic radiation through the medium is described by the Maxwell s equations as following, D ρ (5.3) 6

84 B 0 (5.4) B (5.5) E t D (5.6) H J t where B is the magnetic flux density, H is the magnetic field strength and J is the free current density. (1) Also the relative permittivity ε r of the medium is related to the susceptibility χ by 15, ε r (1) 1 χ (5.7) The optical response of an isotropic medium is represented by its complex refractive index n c (ω) 16, (1) n ( ω) ε ( ω) 1 χ ( ω) (5.8) c r The complex refractive index nc is given by, n c ( ω) n( ω) i( ω) (5.9) The real part n(ω) of n c (ω) represents the optical refraction and the imaginary part (ω) represents the optical absorption in the medium. 63

85 5. Light propagation through a nonlinear optical medium When the intensity of the incident radiation is very high as generated by laser radiation then equation (.) is not sufficient to explain the polarization of the system and we have to consider higher orders 15, (1) 1 () (3) 3 P ε0 χ E χ E χ E... (5.10) where () χ and (3) χ are called the second order and third order nonlinear susceptibilities, respectively. Even though it is possible to have even higher order terms, it is usually very difficult to observe their contribution because it will require extremely large electrical fields. 5.3 The origin of nonlinear optical properties of organic material The origins of nonlinear optical properties in organic materials are rather different from the origins in inorganic bulk materials and compounds. While in inorganic materials the optical nonlinearities are a characteristic of bulk electronic properties, in molecular materials they are predominantly determined by the individual molecular structure 16. In general organic molecular systems have weak van-der-waals interaction among each other 17,18 in contrast to the strong ionic or covalent interactions found in inorganic systems. Therefore usually an oriented-gas model is used to explain the linear and non-linear optical molecular properties of an organic system 16,19. In this model the nonlinear optical properties are determined by the nonlinear optical properties of a single molecule or the molecular unit. Therefore it is important to study and understand the electronic origins of the microscopic nonlinearity of these individual molecular units

86 When an organic material is exposed to electromagnetic radiation the molecules are polarized by the radiation field and are usually explained as a creation of a dipole. When the polarization is weak it is possible to use the power expansion of the electric field components in order to explain the dipole interaction with radiation field 16, P αe β E γe... (5.11)! 3! P is the induced polarization of the organic molecule. The polarizability of the molecule is α. The linear absorption and refraction are determined by α. The constants β and γ are called the first and second order hyperpolarizabilities, respectively, and these are responsible for the second and third order nonlinear optical responses of an organic molecule. If the molecule is center symmetric constant β will disappear. But the constants α and γ will never disappear for any type of molecule σ and π bonds: This section is a brief review of σ and π bonds found in organic molecules. In all organic molecules the element carbon is found. Carbon makes four covalent bonds using its s and p orbitals. When the orbitals are mixed they will form four sp 3 hybrid orbitals resulting in a tetrahedral configuration and this is called sp 3 hybridization. Another carbon or different atom can bind to the sp 3 hybrid orbital by the so called σ bond and their elections are called σ electrons. If the s orbital mixes only with two of the p orbitals, three sp orbitals are created leaving one p orbital unbound. This is called sp hybridization and another carbon or different atom can bind to these three sp hybrid orbitals by a σ bond. The remaining p orbital can bind 65

87 with the p orbital of the other carbon atom, creating a double bond between these atoms. The p-p bond between these two atoms is called a π bond and the electrons in this orbital are called π electrons. The π bond is perpendicular to the molecular plane. When an organic compound is binding in the form of a ring of carbon atoms of alternating single and multiple bonds its π electrons can move along the entire molecular structure. For example in benzene there are two indistinguishable or resonance structures. Figure 5.1: Benzene (C 6 H 6 ) resonance structures. The electrons in this kind of structures have a higher kinetic energy and lower potential energy compared to non-resonance structures and called aromatic molecules. The π electrons in these molecules are delocalized and highly deformable. This delocalization of the conjugated electronic system causes the optical nonlinearities in these molecules 16. Even if the excitation frequencies are different from the electronic resonance frequency the nonlinear polarization of an organic system is usually significantly larger compared to the nonlinear response in inorganic materials. 5.5 The effects of symmetry For molecules with center symmetry (like benzene) the polarization and the potential energy of the molecules are given by: 16 66

88 3 P - α E γ E... (5.1) Form equation (5.13) we see that: V( E ) V( -E) This implies that we cannot include the V( E) - α E γ E... (5.13) 4 β E term in equation (5.13) because then in equation (5.14) yields V( E ) V( -E), which would imply that the molecule has no center symmetry. 5.6 Second order nonlinear optical processes 1. Second harmonic generation (SHG) (also called frequency doubling) (a) (b) Figure 5.: (a) Sketch for SHG, (b) Energy-level diagram for explaining the SHG process When electromagnetic radiation of frequency ω is incident on a non-center symmetric ordered (crystalline) molecular material, the polarization in the system has a frequency component of ω due to the second-order hyperpolarizability β (sketched in Figure 5. (a) and (b)). This phenomena is called second harmonic generation (SHG) or frequency doubling. 67

89 . Sum frequency and difference frequency generation (a) (b) Figure 5.3: (a) Sketch for sum frequency generation, (b) Energy-level diagram for explaining the sum frequency generation process When electromagnetic radiation of frequencies ω 1 and ω are incident on a non-center symmetric molecular crystal, the polarization in the system may have a sum or difference frequency component of ω1 ω or ω1 ω due to the second-order hyperpolarizability β. (See Figure 5.3 and Figure 5.4). (a) (b) Figure 5.4: (a) Geometry for difference frequency generation, (b) Energy-level diagram for explaining the difference frequency generation process 68

90 5.7 Third order nonlinear optical processes 1. Third harmonic generation (THG): (a) (b) Figure 5.5: (a) Sketch of THG interaction (b) Energy level diagram of THG interaction When electromagnetic radiation of frequency ω is incident on a non-center symmetric or center symmetric molecular material, the polarization in the system has a frequency component of3ω due to the third-order hyperpolarizability γ. This phenomena is called third harmonic generation (THG). (See sketch in Figure 5.5). Intensity dependence of the refractive index (and self-focusing): Figure 5.6: Sketch of self-focusing in a third order nonlinear optical medium. When electromagnetic radiation is incident on an organic material the beam gets focused (or defocused) as the beam propagates through the medium because the material acting as a 69

91 virtual lens as sketched in Figure 5.6. This is because the refractive index of the medium is intensity dependent, due to the third-order hyperpolarizability γ. 3. Two photon absorption (intensity dependent absorption coefficient ) Figure 5.7: Energy level diagram for two photon absorption. When an organic molecule or material is excited it may simultaneously absorb two photons of the same frequency or different frequencies. This process is called two photon absorption (TPA). 5.8 Nonlinear refraction In this section I will briefly introduce the theory of nonlinear refraction due to the third order nonlinear susceptibility in an organic material. Since the molecular materials in our investigation have inversion symmetry no second order nonlinear susceptibilities need to be considered in equation (5.10) and the polarization including the self-focusing effect (which is a third order effect) is given by : (1) 1 (3) 3 P ε0 χ E χ E... (5.14) 70

92 The same may be rewritten as 133 : If we neglect (1) (3) P ε0 χ χ E... E (5.15) (4) χ and higher order terms we can write: ( ) (1) (3) where χ NL χ χ E ( NL) P ε χ E (5.16) 0 Equation (5.16) is analogous to the linear polarization equation (.) and can be rewritten: n where n is the complex refractive index of the material. ( NL) 1 χ (5.17) which is equivalent to: (1) (3) 1 1 / χ χ E n (5.18) n (3) 1 / n 0 χ E (5.19) with n0 1 χ (1) being the linear refractive index. We write equation (5.19) as: n (3) χ n 0 1 n0 1/ E (5.0) n χ n 1 (3) E 0 n0 (5.1) n E n n0 (5.) 71

93 ( 3 ) χ where n is called the nonlinear refractive index of the material. This is equivalent to: n 0 where Δn n E n n 0 Δn (5.3) Therefore in a nonlinear medium the change of the refractive index is proportional to the intensity of the incident radiation: Δn E I (5.4) 5.9 Nonlinear absorption In this section I will introduce the nonlinear absorption in an organic medium. When light propagates through a material the intensity at a depth z is given by Beer s law 136 as: I α z ( z) I 0 e (5.5) where I 0 the intensity of the incident is beam and is the linear absorption coefficient. Also the same law can be written as, I αi (5.6) z But when multi photon absorption is present the above equation needs to be be extended to 137, I (5.7) αi βi γi 3... z where β is the two-photon absorption (PA) coefficient γ is the three-photon absorption (3PA) coefficient and the higher order terms are neglected. For a material that has negligible single photon absorption, yet has significant two-photon absorption and if we neglect the higher order terms and only consider the two-photon absorption, 7

94 I (5.8) -βi z The solution of equation (5.8) is given by, I 0 I( z) 1 βi where I 0 is the incident intensity of the incident beam. 0 z (5.9) Also if we consider both single photon and two photon absorption, but neglect higher orders, I (5.30) αi -βi z The solution to the equation (5.30) is given by 138, I ( z) 1 α βi α β 0 e αz 1 (5.31) 5.10 z-scan technique The z-scan technique was developed by Mansoor Sheik-Bahae in order to measure the third order nonlinear refraction and absorption 139. In this technique a sample is moved through an intense focused Gaussian laser beam along the beam propagation direction. The change of the nonlinear refractive index in the sample transverse to the beam propagation direction generates a lensing effect and the beam will be self-focused or defocused. The resulting refraction and nonlinear absorption is related to the real and imaginary parts of the third order susceptibility (3) χ, respectively. An aperture is placed in front of the light collection photodiode to measure the nonlinear refraction ( closed aperture z-scan measurements) and to determine the nonlinear 73

95 absorption in the medium the z-scan experiment is performed without the aperture ( open aperture measurements) Nonlinear refraction and z-scan measurements Figure 5.8: Closed aperture z-scan setup for nonlinear refraction measurements. In order to measure the nonlinear refraction or the real part of (3) χ a sample is moved using a translation stage along the propagation direction of the tightly focused beam and the transmitted light is collected through an aperture placed in front of a photodiode as shown in Figure 5.8. Figure 5.9: Depending on the sign of the change of the nonlinear refractive index n the sample may act as a converging or diverging lens (after 140 ). When the Gaussian beam propagates through the sample, due to the nonlinear refractive index change in the transverse direction the sample acts as converging or diverging lens as displayed in 74

96 Figure 5.9. For a positive refractive index change it would act as converging lens and for a negative refractive index change it would act as a diverging lens. The normalized transmittance through the aperture is shown in Figure 5.10 for both positive and negative refractive index changes. In the presence of nonlinear absorption the nonlinear refraction trace can be finding the ratio between the transmittance with the aperture and without the aperture as shown in Figure Figure 5.10: z-scan traces for the transmission through the aperture for negative and positive nonlinear refractive index change (after 140 ). Figure 5.11: When both nonlinear refraction and absorption are present at the same time the division of closed aperture data by the open aperture provides the data purely due to the nonlinear refraction (after 140 ). 75

97 5.1 Theoretical description of nonlinear refraction and z-scan The electric field of a Gaussian beam can be written as 139,141 : ω 0 r ikr E ( r,t,z) E 0 ( t). exp iφ( z,t). ω( z) ω ( z) R( z) (5.3) where ω 0 is the beam waist, z is the propagation distance from the beam waist, w(z) is radius of the beam at position z, E ( ) is the electric field at the beam waist, r is the radial distance 0 t from the center axis, z 0 R ( z) z 1 is the radius of curvature of the beam s wave fronts, z π k is the wave number, φ( z) is the phase shift. λ The beam radius at position z is given by, ω ( z) z z ω0 1 0 (5.33) and z 0 πω0 is the diffraction length or Rayleigh length, and λ is the laser wavelength. λ The problem is considerably simplified by the so called slowly varying envelope approximation (SVEA) due to the self action limit 140. Therefore the following relation is true for linear absorption (i.e. for negligible nonlinear absorption), di dz' α I with z' being the propagation depth in the sample and α being the absorption coefficient. (5.34) 76

98 Also by using the same SEVA approximation, the change of phase Δφ due to the change of the refractive index Δn holds the following relation 140, dδφ( z,r,t,l) Δn( I) k dz' Equations (5.35) and (5.36)can be solved, giving 139, (5.35) Δφ0 ( t,l) r (5.36) Δ φ( z,r,t,l) exp z w ( z) 1 z0 Δφ(z,r,t,L) represents the phase change at the exit surface and Δφ 0 ( t,l) k Δn( t) Leff is called the on-axis phase shift and L eff 1 e α αl, where L is the sample thickness and α is the linear absorption coefficient. If we include this phase shift in the Gaussian beam profile we get the electric field exciting the sample with the nonlinear phase distortion as 139, αl/ iδφ( r,t,z,l) E e ( r,t,z,l) E( r,t,z) e e (5.37) Using the so called Gaussian decomposition (GD) method given by Weaire et al. 14, the nonlinear phase term can be decomposed into a summation of Gaussian Beams through a Taylor series expansion 143 and it can be shown 139 that the reconstructed beams at the aperture plane lead to a transmittance change of ΔT 0. 5 p v (1 S) Δφ (5.38) 0 77

99 for a specific aperture size.in equation (5.38), S 1 - exp( - r /w a a ) is the aperture linear transmittance, ΔTp v is the peak to valley transmittance change and also Δφ0 π., The change of the transmission is sketched in Figure 5.1. Figure 5.1: Normalized transmittance of a closed aperture z-scan experiment. The height between peak and valley ΔT is proportional to the phase change due to nonlinear refraction p v and can be used to calculate the nonlinear refractive index of the material Nonlinear absorption and z-scan measurements In order to determine the nonlinear absorption of a material the aperture in front of the light collection photodiode the aperture is fully opened or removed as shown in Figure When the normalized transmittance vs. the sample position is recorded then it will be a symmetric curve around the focus as shown in Figure

100 Figure 5.13: z-scan setup for nonlinear absorption measurements. In order to measure the nonlinear absorption all the transmitted light is collected by a photodiode(after 140 ). Figure 5.14: Illustration of the normalized transmittance of a nonlinear absorption measurement using the z-scan technique. The complex third-order nonlinear susceptibility of a material can be written as, χ (3) χ i (5.39) (3) (3) R χ I where subscript R represents the real part and subscript I represents the complex part. The imaginary part is related to nonlinear absorption (PA) coefficient β by, β ω n ε c 0 0 (3) χ I (5.40) 79

101 where ω is the frequency of the optical source. The real part is related to the nonlinear refractive n index by, n 1 n ε c 0 0 (3) χ R (5.41) The intensity dependent absorption coefficient can be written as, α( I) α βi (5.4) and di dz' α( I) I (5.43) where z' the propagation depth in the sample and α is the linear absorption coefficient. The phase change Δφ due to the change of the refractive index Δn holds the following relation (see eq. (5.35)), dδφ( z,r,t,l) Δn( I) k dz' (5.44) Equations (5.43) to (5.45) are used to get the following equations 139, I ( z,r,t) e αl I( z,r,t) e (5.45) 1 q( z,r,t) kγ Δφ( z,r,t) ln1 q(z,r,t) β (5.46) where I e (z,r,t) is the intensity at the exit surface of the sample, q(z,r,t) βi(z,r,t)leff and L eff 1 e α αl. L is the actual length of the sample. 80

102 Decomposition of the non-linear phase term into a summation of Gaussian Beams similar as for the nonlinear refractive index and applying a zeroth-order Hankel transform 139 the normalized energy transmittance can be written as, where P i t) πw I ( t) and ( 0 0 / q ( z) 0 βi 0 L z 1 z eff 0 By time integrating equation (5.47) the normalized energy transmittance can be written as, 1 r T( z) ln[1 q0 ( z) e ] dr πq ( z) 0 (5.47) Where P i t) πw I ( t) and ( 0 0 / q ( z) 0 βi 0 L z 1 z eff 0. The same transmittance in numerical (for q 0 1) form can be written as a summation as, T ( z) m0 [ q 0 ( z)] ( m 1) m 3/ (5.48) If we consider only up to m = 1 (since q 0 1, higher order terms may be neglected) then than equation (5.48) simplifies to, which is the same as, q ( ) ( ) 1 0 z T z (5.49) Q T( z) [1 z /z 0 ] (5.50) where βi L and also Q 0 1. Q0 0 eff 81

103 Equation (5.50) is used to fit the experimental data obtained from the normalized transmittance as a function of the sample position in a z-scan experiment Experimental setup The Millennia laser operating at 53 nm (CW) was used to pump the Ti-Sapphire Tsunami laser which generates ultrashort pulses with a temporal pulse width of 100 fs at a repetition rate of 80 MHz. High numerical aperture microscope objective lenses (100x with N.A. = 0.55 and 10x with N.A. 0.8) were used to produce a tightly focused beam. The organic film was mounted on a P-50 translation stage which was connected to a C-866 stage controller from PI instruments. High quality silver and gold mirrors were used for minimal energy loss in the setup. Neutral density (N.D.) filters were used for adjusting the laser power levels. Two high speed Agilent digital multimeters (DMM) were connected to photodiodes D1 and D. Diode D was used as the detector for the transmitted light through the sample and diode D1 was used as the reference photo-detector. A third detector D3 was implemented for nonlinear refraction measurements in combination with an optical aperture in front of it. The three photo-detectors are composed of hi-speed, large-area photodiodes purchased from Hamamatsu Corporation. The controllers and DMMs were attached to the computer (PC) using high speed Universal Serial Bus and General Purpose Interface Bus (GPIB) connections. Lab View from National Instruments (NI) was used as the primary programming language for instrument control and data acquisition. The instrument software drivers (NI VISA) and Lab View programs that were obtained from the instrument manufactures were modified as necessary. Also some of the drivers 8

104 were downloaded from NI directly. A schematic diagram of the experimental setup is shown in Figure Figure 5.15: Experimental setup for the z-scan experiments. Multiple reflections in the films were noticed by large oscillations that were present in the z-scan traces. The sample was then slightly tilted with respect to the beam propagation direction to minimize the effects of multiple reflections in the thin films. A fast mechanical shutter was used to excite the organic film at desired intervals, which was controlled by a home built controller using an integrated circuit timer. The chopping of the laser pulse train at given intervals (shown in Figure 5.16,) reduced undesirable thermal effects on the film due to the high repetition rate of the pulsed laser and minimized the possibility of film ablation or melting. The 83

105 digital multimeters (DMM) were synchronized with the shutter using a triggering mechanism (See Figure 5.16). The voltage pulse width to open the shutter open is approximately ~ 0.7 ms. Considering the inertia and friction of the mechanical parts of the shutter this results in an equivalent open time of ~ 3 ms and a close time of ~ 1400 ms. The DMMs start taking measurements when the shutter gate is open and stop when the gate is closed. The data sampling rate of the DMM was set to cps (counts per second). Figure 5.16: The high repetition rate laser pulse train is chopped using the mechanical shutter at regular time intervals to reduce thermal effects in the thin films. The figure shows the pulse trains transmitted through the shutter at regular intervals. At 80 MHz repetition rate the number of pulses passing through the chopper gate is per open cycle. The number of pulses blocked in the closed cycle is The ratio of chopper open time to closed time (cool down time) is ~ 1/466. The shutter frequency is ~ 0.7 Hz. 84

106 5.15 Pulse width measurements The ultra-short 100 fs laser pulses temporally broaden when they pass the (thick) glass elements of the microscope objective which affects the peak power of the pulses. The pulse width of the pulses after passing the microscope objective lens has been measured using the autocorrelation technique as shown in Figure In this setup a modified version of a Michelson interferometer was used. The pulsed laser beam that passed through the microscope objective is split into two beams of equal intensity using a beam splitter such that it passes through both arms of the interferometer. One interferometer arm contains a retro-reflector and the other arm contains a high-reflecting (HR) mirror. The retro-reflector can be scanned along the beam propagation direction. The motion of the retro-reflector controls the time-delay between the pulses that passes through both arms of the interferometer. A lens focuses the two parallel beams at the beam splitter on to a nonlinear crystal which produces a second harmonic (SHG) light beam as shown in Figure

107 Figure 5.17: Autocorrelation setup for pulse width measurements. A photo-detector collects the second harmonic signal and an oscilloscope enables to observe the background-free intensity autocorrelation function of the optical pulses. A plane parallel quartz plate was placed into one of the arms to change the optical path by a known length and the resulting change on the time-scale of the oscilloscope was used for calculating the temporal pulse width of the input pulse. The temporal-pulse width for 5x, 10x, and 100x lenses were ~ 160, ~ 170 and ~ 60 fs (femtoseconds), respectively. Data averaging, time integrating, and noise removal In the z-scan experiments the measured changes in transmission due to nonlinear absorption or refraction is very small compared to the input irradiance. Therefore in most cases the quality of the data obtained by a single scan was not sufficient to generate a reliable plot. Accordingly multiple scans were necessary and the averaged results were used for data analysis. The data averaging was achieved using programming techniques in LabView. The time integration functionality of the digital multimeter (DMM) was used to time integrate the data 86

108 samples that were obtained at cps. Although better signal-to-noise ratio was obtained through longer integration times the scans become slower which created additional problems (e.g. laser stability or sample damage with time). Therefore an appropriate integration time was chosen as necessary. In addition, a lock-in amplifier in combination with an optical chopper wheel was used to test the z-scan setup z-scan measurements on Alq3 films The nonlinear optical properties of organic thin films have been of great interest to many researchers during the past decades because of their large values compared to inorganic materials and due to their potential applications for optical power limiting 144,145 all-optical switching 146,147, and flexible device design 148,149. Also lower processing cost, lower driving voltage, higher bandwidth and fast response have become the driving force behind the recent developments in nonlinear optical (NLO) materials with π electron conjugated systems Despite the fact that tris(8-hydroxyquinolinato)aluminum (Alq3) has been widely used in Organic Light Emitting Device (OLED) applications the ultrafast third order NLO properties of the material has not yet been investigated. Here I have used the z-scan technique to study the third-order nonlinear absorption properties of Alq3. The Alq3 films were grown using the Organic Molecular Beam Deposition (OMBD) technique explained in Chapter 3. For the z-scan experiments a 10x (N.A. = 0.8) microscope objective was used in combination with the optical shutter system previously described. The Tsunami (Ti-sapphire) laser was tuned to 814 nm and first was scanned at a very low intensity (attenuated by OD) in order to get the background as shown in Figure Then a high intensity scan was performed as shown in Figure In order to determine the background free 87

109 z-scan trace the high intensity trace was divided by the very low intensity trace. The lab-view program was used to take the same measurement five times and calculate the average result. The resulting open aperture z-scan trace of a 5 m film at an average laser input power of 746 mw is shown in Figure 5.0. The theoretical fitting to the equation (5.50) is also shown in Figure 5.0 (solid blue line). An absorption coefficient of 355 cm -1 has been used 100 which led to an effective sample length of L eff = 4.6 μm. The electric field beam waist from the z-scan fit resulted in 6 0 μm. This value agrees with the intensity beam waist of 4.5 μm if 0 equation (5.51) and a numerical aperture of ~ is used. This numerical aperture was derived by considering that the laser beam diameter is ~.5 mm and the opening diameter of the microscope objective is 13 mm which reduces the N.A. of the 10 x objective lens by a factor of ~ NA NA (5.51) Z (5.5)

110 1.004 transmittance [a. u.] z/z 0 Figure 5.18: Open aperture z-scan trace of a 5 m Alq3 film on glass at low incident average power (~7.5 mw). The diffraction length was calculated using equation (5.5) to be Z 140μm and this 0 satisfies the necessary condition ( L Z0 ) for thin samples to be used in z-scan experiments. The input irradiance was calculated to be 97.6 GW/cm, according to I P /( π( / )) 0 i w 0 and P i P /( t )) rep where P i is the incident pulse peak power, Δt is the temporal pulse width and a γ rep is the pulse repetition rate. The nonlinear absorption coefficient was calculated to be ~

111 cm/gw transmittance [a. u.] z/z 0 Figure 5.19: Open aperture z-scan trace of a 5 m Alq3 film on glass at high incident average power (~746 mw). 90

112 1.00 normalized transmittance T(z) experimental theoritical z/z0 Figure 5.0: Open aperture z-scan trace of a 5 m Alq3 film on glass at 746 mw average power. The experiment was repeated at 480 mw average input power. Using the same calculations described above for 746 mw input power the obtained z-scan trace for 480 mw power is shown in Figure 5.1. The nonlinear absorption coefficient was calculated to be ~ 0.34 cm/gw. In recent z-scan measurements (performed by Dr. Xiaosheng Wang) a pulse selector was used to reduce the repetition rate by a factor 100 to 400. The reduced repetition rate assures that the two-photon excited state completely relaxes back to the ground state before the second pulse arrives thus minimizing saturation effect and also reducing the accumulation of heat in the 91

113 sample. In these experiments an ~3 times lower nonlinear absorption coefficient has been measured indicating that measurements using the shutter but a high 80 MHz repetition rate might still be affected by sample heating. This heating could cause a divergent thermal lens effect which might explain the deviations between the derived nonlinear coefficient values. normalized transmittance T(z) Experimental Theoritical z/z0 Figure 5.1: Open aperture z-scan trace of a 5 m Alq3 film on glass at 480 mw average power. 9

114 5.17 z-scan measurements on polycrystalline PTCDA films Perylene-3,4,9,10-tetracarboxylic-3,4,9,10-dianhydride (PTCDA) films were grown using the OMBD technique on glass substrate using a similar method described for the Alq3 films (see Chapter ). A wavelength of 880 nm was used and the open aperture z-scan trace of a m thick PTCDA film at very low intensity was obtained and is shown in Figure 5.. Then high intensity z-scan was performed at ~.8 mw and the trace is shown in Figure 5.3. These experiments were performed with a 100 x microscope objective and without using a shutter, because this device was not available at the time of this experiment. The background subtracted z-scan trace was obtained by dividing the high intensity trace by the very low intensity trace and is shown in Figure 5.4. Since the 100 x objective was very sensitive to the optical alignment and due to possibly caused laser induced thermal effects (described below) the measurement results were rarely reproducible for subsequent multiple scans. The two-photon absorption coefficient was calculated to be = ~11 cm/gw (using equation (5.50)) also revealing a beam waist of ~1 μm and an input irradiance of 86 GW/cm. For these calculations an absorption coefficient of 15 cm -1 has been used 153. The derived two-photon absorption coefficient can be converted into the two-photon absorption cross section using the relation n /( ) with n being the density of molecules and being the single photon energy at angular TPA TPA frequency. Using a density of 1 n.6310 cm 3 for α PTCDA 97 a two-photon absorption cross section of cm s 95 GM. This value is by a factor of ~.5 TPA higher as compared to the two-photon absorption cross section 40 GM obtained from z- scan measurements on PTCDA nanoparticles at a wavelength of 815 nm 154. More recent z-scan measurements using a pulse selector and a 10 x objective (performed by Niranjala 93 TPA

115 Wickramsinghe) reveal a ~ times lower nonlinear absorption coefficient at a wavelength of 815 nm in very good agreement to the value found by 154. As for the Alq3 films this might be attributed to a reduction of saturation and laser induced heating in the PTCDA film when a reduced laser repetition rate is used. Further investigations are in progress. 1.0 transmittance [a.u.] z/z 0 Figure 5.: Open aperture z-scan trace of a m PTCDA on glass at low average power (~8 μw). In a different measurement the z-scan experiment has been repeated using a thick PTCDA film of ~ 5 m nominal thickness and a 10 x microscope objective lens (focus diameter 94

116 of 1.5 m) and the shutter was added to the setup to reduce thermal effects. Despite the application of the shutter the 5 m film was significantly damaged by laser induced ablation or melting of the material. The laser induced damaging of the PTCDA film and the creation of crater holes were verified using microscope images at different magnifications shown in Figure transmittance [a.u.] z/z 0 Figure 5.3: Open aperture z-scan trace of a m PTCDA film on glass at high average power (.8 mw). 95

117 The circular shaped rings indicate permanent damage to the PTCDA films. In Figure 5.6 the film images were obtained by focusing the microscope to the top, surface and bottom of the PTCDA film using the 100x magnification. A side-view sketch of the cross section of laser damaged area of the film is illustrated in Figure normalized transmittance T(z) z/z 0 Figure 5.4: Open aperture z-scan trace of a m PTCDA film on glass. 96

118 5 x 10 x 0 x 50 x 100 x Figure 5.5: Images of laser induced melting of holes into the 5 m thick PTCDA film at different magnifications. 97

119 100 x_top 100 x_surface 100 x_bottom Figure 5.6: Microscopic images focused at different depth levels. normalized transmittance T(z) z/z 0 Figure 5.7: z-scan trace of a laser damaged film. 98

120 Surface Top Bottom PTCDA Substrate Figure 5.8: Illustration of the cross section of the molten area in the film. When thermal effects induced by the laser irradiation damage the film a z-scan trace as shown in Figure 5.7 is generated. Once the film melts (at higher intensities) the trace is similar to a z-scan trace produced due to nonlinear absorption but the transmission decrease at z 0 is unusually large (down to ~10 %). Another sign of film melting is that the dip in the transmission after a high intensity scan does not vanish even at very low intensity where non-linear absorption would not be expected. From these observations we conclude that the laser induced craters in the film act like a divergent lens (see Figure 5.8) and therefore causes a reduction of the transmitted intensity as shown in Figure Two-photon induced fluorescence Two-photon induced fluorescence is an alternative method to study the two-photon absorption properties of organic materials and films. The two-photon fluorescence data is usually more difficult to interpret, yet gives more details that are usually not available in two-photon absorption measurements. It is possible to have a two-photon excited state by either a twophoton state which is closely coupled to a one-photon state or a two-photon state alone 133. Depending on the change of parity of the states due to the two-photon absorption and the nature of the relaxed state, the final spectrum could be similar to the one-photon absorption spectrum or 99

121 different than the one-photon absorption spectrum. It gives a similar spectrum if the two-photon state relaxes to the lowest vibrational level of a one photon state. If the two-photon state occurs through an intermediate state the two-photon fluorescence spectrum will differ from the onephoton spectrum. Also it is possible to obtain information about vibrational spectra, dipole moments of transition and photo-induced chemical reactions using the two-photon florescence data 133. In order to explain the two-photon induced fluorescence of excited molecules I will use a simple three level model where two-photon absorption occurs from ground state (with n 0 being the density of molecules in the ground state) to the two-photon excited state ( n ) of the molecule which radiationless relaxes (thermalizes) to the lowest vibrational level of the excited state (with n 1 being the density of molecules in the lowest vibrational level of the excited state). Also I assume that there is no linear absorption and stimulated emission and that the excited states n 1 relax completely to the ground state n 0 after the excitation by one pulse (meaning the pulse repetition rate is much longer than the sum of radiative and non-raditive relaxation rates k 10 of excited states n 1 back to states n 0 ) to avoid saturation effects. The coupled rate equations of the three level system is given by: dn dt σ I (5.53) TPA n0 k1n ω0 dn dt 1 k1n n k10 1 (5.54) 100

122 dn0 dn1 dn dt dt dt (5.55) where σ TPA is the two-photon absorption cross section and k ij is the relaxation rate from the i th state to the j th state. For very short pulses with temporal pulse width p k1, k10 the value of n 1 1 can be given as 133, σ n TPA T / n (0) I ( t ) ω 0 dt / (5.56) where n T n0 n n1 is the total density of two-photon absorbers. For a Gaussian pulse equation (5.56) can be written as, σ n (0) TPA n T τ ω 0 p I 0 π (5.57) where τ p is the pulse width, I 0 is the peak intensity of the exciting laser pulse. If we assume that the density of excited exciton n is very small compared to the density of excitons in the ground state ( n T n0 ) and that two-photon excited states n very quickly relax ( k1 k 10 ) to the lowest excited state n 1 eqs. (5.53) - (5.57) simplify to, dn dt 1 k 10 n 1 (5.58) with n (0) n 1 σ (0) TPA n 0 τ ω 0 p I 0 π (5.59) 101

123 The density of emitted photons is given by, dn dt ph k 10 rad n 1 (5.60) with k 10 rad being the radiative recombination rate of states n 1. By integrating (5.59) and (5.60) over time the energy emitted in the fluorescence band can be found to be 133, ε pulse ω g ω) n (0) V dω (5.61) ( where ω is the angular frequency, g(ω ) represents the shape of the fluorescence band, V ca is the volume of the radiated energy flux, with A being the area of the detector, c k 10rad k rad 10 being the speed of light and η being the fluorescence yield given by. k The total integrated signal can be written as, 10 S Θε Z (5.6) pulse rep where Θ is the responsively of the detection system, rep the impedance of the detection system. is the repetition rate of the laser, Z is According to equation (5.6) the integrated florescence has a quadratic dependence with the input intensity. For simplicity the same equation may be written as, S B I 0 (5.63) 10

124 5.19 Experimental setup for two-photon induced fluorescence measurements The 5 m Alq3 sample (that was also used in z-scan experiments) was fixed on a sample holder using a thermal paint (silver paint). The Tsunami (Ti:Sapphire) laser was pumped using the Millennia CW laser operating at 53 nm and 10 Watt power. The Tsunami was tuned to 814 nm and the light was directed onto the Alq3 sample using a lens of focal length f 15cm. The repetition rate of the laser was 80 MHz and the pulse width was 100 fs. The emitted fluorescence was collected using a regular lens and directed into the monochromator as shown in Figure 5.9. The photomultiplier output was connected to the SR400 photon counter and the SR400 was connected to the computer using GPIB interface. Figure 5.9: Experimental setup for two-photon fluorescence measurements. The reference detector was used to record fluctuations of the laser. 103

125 Two-photon fluorescence measurements with Alq3 films Even though the single photon fluorescence of Alq3 films and molecules has been deeply studied by many research groups no two-photon fluorescence measurements studies have been reported on this material thus far. The experimental setup described in Figure 5.9 was used providing a maximum excitation power of 343 mw, for these measurements. The power levels on the Alq3 sample have been adjusted using neutral density filters (OD s) photon counts [a. u.] wavelength [nm] Figure 5.30: Two-photon induced fluorescence spectrum of a 5 m Alq3 sample at an incident average power of 7 mw. 104

126 A spectrum of the two-photon induced fluorescence is shown in Figure The spectrum is very similar to a one photon absorption spectrum excited at 407 nm which is shown in Figure 5.1 in Chapter 5. (The spurious oscillations in the spectrum are F.P. oscillations). We therefore conclude that the excited two-photon state is closely coupled to a one-photon state via fast non-radiative relaxation as it has been assumed in the three level model described in the previous section. The spectrally integrated fluorescence versus excitation intensity is displayed in Figure integrated PL [a. u.] S = (6.15) I 0 Experiment Theory 0.0x10 4.0x10 6.0x10 8.0x10 intensity [MW/cm ] Figure 5.31: Quadratic intensity dependence of the two-photon induced fluorescence obtained from a 5 m Alq3 film. The experimental data has been fit using the theoretical equation (5.63). 105

127 with B ~ 6.15 cm s/j. Our experimental data of the two-photon induced fluorescence in Figure 5.31 reveals a quadratic intensity dependence as predicted by theory. However, the value of parameter B in eq. (5.63) cannot directly be used to calculate the two-photon absorption coefficient, because parameters like k 10 and Z in eq. (5.6) are unknown. Further measurements at a reduced repetition rate avoiding possible saturation effects as well as an estimation of parameters k 10 and a calibration of the used detector system will enable to determine the two-photon absorption coefficient and to compare with values that were derived from the z-scan measurements. 106

128 Chapter 6 6 Singlet-singlet annihilation due to funneling of excitons to traps in Alq3 films 6.1 Photoluminescence studies of tris(8-hydroxyquinolinato)aluminum films As described in the previous chapter tris(8-hydroxyquinolinato)aluminum (or Alq3) is a quasi-amorphous material. Since its discovery by the Kodak Company 1 in 1987 it is widely used in organic light emitting diodes (OLEDs) as efficient light emitting layer 155,156. Even though extensive research has been performed on this material for several decades, the optical and electrical properties of Alq3 films are still rather unexplained and not deeply understood. The current external quantum efficiency EL of Alq3 based OLEDs reaches only ~8%157. Researchers all around the world aim to exceed this value by using different OLED architecture as well as modified carrier injection Thermal annealing of OLED structures as well as Alq3 film deposition at different growth conditions are further attempts 16,163 to enhance the performance of Alq3 based OLEDs. However, the influence of crystalline order and the contribution of interacting molecules on the photoemission due to intermolecular transitions are still rather unexplored. Alq3 thin films 88,164,165 contain a vast distribution of various crystalline modifications on a very small (few 10 nanometer sized) length scale. This distribution causes potential fluctuations in particularly at the grain boundaries of nanocrystals with energies below the free Frenkel exciton energies 166,167. Frenkel excitons below the mobility edge are trapped in these potentials within a very short time resulting in an inhomogeneously broadened and red-shifted photoemission (by ~150 mev) compared to bulk Alq3 crystals. The small inter-ligand separation 107

129 of 0.35 to 0.39 nm in and Alq3 nano-crystallites 168 in the quasi-amorphous layer would further allow the formation of charge transfer (CT) excitons or neutral excimers (excited dimers) where the electronic excitation is extended over two adjacent Alq3 molecules. integrated PL intensity temperature [K] PL intensity [a. u.] wavelength [nm] 10 K 300k Figure 6.1: PL spectra of an 80 nm thick Alq3 film deposited on Si (001) as function of temperature ranging from 10 to 300 K in 0 K steps. Inset: Spectrally integrated and normalized PL intensity obtained from Figure 6.1of the 80 nm Alq3 film as a function of temperature. 108

130 Recent temperature dependent PL investigations on a 40 nm thick Alq3 films reveal a change of the PL efficiency by a factor of ~ when the temperature is increased from 10 to ~190 K 13.In order to study the dependence of this effect as a function of the film thickness I have repeated the PL measurements on an 80 nm thick Alq3 layer. The organic film was grown on chemically clean Si (001) substrate using the OMBD method described in a Chapter 1, with a deposition rate of 0.01 nm/s at the growth temperature of 30 0 C. The temperature dependent PL experiments were performed using a continuous wave (cw) GaN semiconductor laser at an emission wavelength of 407 nm and with 0.3 mw output power as excitation source. The laser beam diameter was ~ mm, the focal length of the converging lens was 150 mm. Figure 6.1 shows the PL spectra at temperatures ranging from 10 to 300 K in 0 K steps. As observed in investigations by other groups 88,164 the PL spectra do not show a vibronic progression. Instead, the PL reveals a broad emission band which is attributed to the polymorphous growth of an Alq 3 films containing all possible crystalline modifications 88,164,165, leading to a vast distribution of excitonic state energies as already mentioned. At 10 K the peak position of the PL spectrum (~.36 ev) is shifted by ~100 to 150 mev to lower energies compared to the PL peak position obtained from - and - Alq 3 polycrystalline samples. The observed red-shift is attributed to the trapping of excitons at potential minima in disordered Alq 3 thin film. In the temperature range from 10 to ~190 K the PL intensity increases with increasing temperature and the emission maximum shifts by ~50 mev to lower energies. The inset of Figure 6.1 shows the spectrally integrated PL (which is proportional to the PL quantum efficiency) as a function of temperature for the 80 nm thick Alq3 film. In this plot the PL intensity at 300 K was calibrated to an efficiency value of η ~

131 obtained by other groups As for the 40 nm Alq3 layer 13 the PL efficiency peaks at ~190 K showing an enhancement of the PL efficiency by a factor of ~1.7 when the temperature is increased from 10 to ~190 K. At higher temperatures the PL intensity decreases and the PL band maximum slightly shifts back to higher energies. Supported by time-resolved measurements performed in this work the observed temperature dependence of the PL is attributed to a thermally activated formation of charge transfer (CT)-like or excimer-like excitons. It will be shown that the creation of such twomolecule exciton complexes reduces the bimolecular annihilation of trapped excitons leading to an enhancing PL efficiency when the temperature is risen from 15 to 190 K. This interpretation is further supported by intensity dependent cw and time-resolved PL measurements 99. For the time-resolved measurements a frequency doubled Ti: Sapphire laser providing 100 fs pulses at a repetition rate of 4 MHz has been used as excitation light source. The time-resolved measurements reveal a slight increase of the exciton lifetime from ~3 to 5 ns when the temperature is risen from 10 to 190 K. Above 190 K the PL efficiency and decay time decreases which is attributed to a thermal destabilization of the CT complexes 99 and de-trapping of radiative exciton states with subsequent filling of non-radiative centers. Specific goal of these investigations is to investigate possible contributions of intermolecular transitions to the light emission in quasi-amorphous Alq3 films. Earlier experiments suggested that crystalline order in quasi-amorphous films could lead to an enhancement of the luminescence efficiency due to the generation of self-trapped excitons 13,99. Other investigations 175 suggested that the observed temperature dependence of the PL efficiency is caused by the thermal activation of intermolecular delocalized exciton states with longer 110

132 radiative lifetime. In contrast, our work demonstrates that below ~170 K the PL efficiency of Alq3 films is reduced by singlet-singlet annihilation processes (also known as bimolecular quenching processes). 6. Singlet-singlet annihilation in Alq3 films The singlet-singlet annihilation in amorphous Alq3 films have been thoroughly studied by several research groups In 1996, Sokolik et al. 178 reported that the decrease of the quantum efficiency at high excitation intensities in Alq3 can be explained by the mutual annihilation of singlet excitons. This type of exciton-exciton annihilation has been described in 11 3 Chapter.The reported annihilation rate constant was (3.5.5) 10 cm /s and the diffusion 5 coefficient of excitons was determined to be (1. 0.8) 10 cm /s. A pre-factor f (f = 1/) has been used in the annihilation rate which neglect losses due to auto-ionization of the excitons and possible direct non-radiative relaxation of the super excited state. A relation between the diffusion coefficient and annihilation rate has been described in Chapter. In a more recent study by Mezyk et al. 177 on vacuum-evaporated Alq3 films, the measured singlet-singlet 10 3 annihilation rate constant is ( ) 10 cm /s (using f = 1/) and an average singlet 5 exciton diffusivity of (6 ) 10 cm /s has been reported. In this study it was suggested that singlet-singlet annihilation is not an effective mechanism to reduce the EL quantum efficiency of Alq3 based OLEDs that are operated under steady-state conditions. A temperature dependent lifetime study of the singlet excited states in Alq3 films has been performed by Priestley et al It was shown that the exciton lifetime in Alq3 increases with decreasing temperature which has been attributed to the reduced migration of the excitons to defects with decreasing temperature. The exciton dynamics in prototypic polycarbazole-based photovoltaic donor 111

133 polymer, namely poly[n-1100-henicosanyl-,7-carbazole-alt-5,5-(40,70-di--thienyl-0,10,30- benzothiadiazole)] (Pristline-PCDTBT) films have been studied by Etzold et al. 179 and it has been demonstrated that very low excitation densities are needed in these films to completely stop the bimolecular quenching effects. It has been suggested that the disorder effectively funnels the excitons towards lower energy trapping sites which causes an increase in the local population density at these low-energy areas. In general the exciton migration parameters which are based on random walk theory show significant deviations when disorder is involved in the material. In the presence of disorder the migration is not dominated by random walk but the migration of excitons occurs downhill an energy gradient towards the lower energy sites. In most materials the excition migration during their lifetime is reduced because of disorder. Disorder in molecular systems can have an energetically or spatial nature. Figure 6.: Due to the energetic disorder in an amorphous material the excitons migrate downhill the density of states (DOS) and are trapped in energetically deeper states. This mechanism also allows more than one exciton to be funneled to two adjacent trapped exciton states, where singlet-singlet annihilation can occur. 11

134 The migration distance during the exciton lifetime is related to the short range order in the solid. In a well ordered (crystalline) solid the excitons travel further than in a highly disordered (amorphous) solid. Due to disorder the exciton gets more localized with time because it migrates towards the low-energy tail of the excitonic DOS distribution while lowering its energy (see Figure 6.). Therefore, the migration of the exciton depends on the energy distribution of the excitonic states created by disorder. Depending on the nature of the trapping site more than one exciton can be funneled into the same low-energy DOS area causing bimolecular annihilation even at very low excitation intensities. 6.3 Time resolved photoluminescence measurements The temperature dependence of the PL (and EL) efficiencies of Alq3 films have been investigated by several research groups 166,176 and the results exhibit an increasing efficiency of the PL efficiency with decreasing temperature from ~300 K to ~170 K. Besides only one recent investigation from 175 the PL efficiencies change only slightly when the temperature is further decreased below~ 170 K. Figure 6.3shows the temperature dependence of the time-integrated PL (which is proportional to the PL efficiency)detected at a wavelength of 534 nm when a 10 nm thick Alq3 film has been optically excited with 100 fs pulses at a wavelength of 407 nm ( ~3.05 ev). For these experiments the pulse repetition rate of 80 MHz was reduced to 4 MHz using a pulse picker (see also Figure 6.5).In these measurements the temperature has been varied between 15 and 300 K. As in inset of Figure 6.1the PL efficiency has been calibrated to an efficiency value of η ~ 0.5 at 300 K. At low excitation densities (at an average power of 0.01 mw) the PL efficiency shows nearly no change when the temperature is lowered from 170 to

135 K which is consistent to the result reported by 166,176. However, when the excitation power is increased to 0.3 mw the luminescence efficiency below ~170 K significantly starts to drop as shown in Figure mw 0.03 mw 0.3 mw PL quantum yield temperature [K] Figure 6.3: Dependence of the PL efficiency of a 10 nm Alq3 film as a function of temperature measured at a wavelength of 534 nm. The incident average laser power was 0.3, 0.03 and 0.01 mw at 407 nm excitation wavelength. The significant intensity dependent change of the photoluminescence efficiency in Alq3 films at different temperatures has been further studied using time-resolved photoluminescence experiments. The time-resolved photoluminescence traces of a 10 nm thick Alq3 film at 15 K and at three different average powers (0.3, 0.03 and 0.01 mw) are shown in Figure

136 1 15 K 0.3 mw 0.03 mw 0.01 mw PL [normalized] E decay time [ns] Figure 6.4: Time-resolved and normalized PL intensity obtained from a10 nm Alq3 film at a detection wavelength of 534 nm at different average power as labeled. For these experiments a frequency-doubled Ti-sapphire laser (Tsunami) tuned to ~407 nm was used as the excitation source and the pulse repetition rate of 80 MHz was reduced to 4 MHz using a pulse picker (see Figure 6.5).Also in order to measure the time-resolved traces a Pico-Harp (Picoquant 300) device was connected to the PL set-up consisting of a monochromator (model SR3), single photon counter (model SR400) and a fast GaAs photomultiplier with ~.5 ns response time. Both the fast General Purpose Interface Bus (GPIB port ) and Universal Serial Bus (USB version.0) was used as a high-speed communication 115

137 interface between the computer (PC) and the instruments described above. A cryogenic cooling system (Cryogenics) and the temperature controller were used for cooling the sample to the desired temperatures ranging from ~ 15 K to 300K. Figure 6.5: Experimental setup for time-resolved measurements. The time resolved experiments clearly indicate that an increasing excitation intensity causes a decrease of the PL decay time, which explains the decrease of the PL efficiency observed in the intensity dependent PL measurements shown in Figure 6.1 and Figure 6.3. Furthermore the time-resolved traces indicate that an increasing intensity induces a strong nonexponential decay indicating a non-linear bimolecular annihilation of the singlet excitons due to exciton funneling into traps as described in the previous section. 116