<%BANNER%>

Ultrafast Time Resolved Excitation Dynamics in Conjugated Dendrimers


PAGE 1

ULTRAFAST TIME RESOLVED EXCITATION DYNAMICS IN CONJUGATED DENDRIMERS By EVRIM ATAS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Evrim Atas

PAGE 3

To Selim, Avni, and my parents

PAGE 4

iv ACKNOWLEDGMENTS As I look back upon the years that have led to this dissertation, I have been fortunate to have been surrounded by the he lp and influence of numerous exceptional people. I would like to extend my warmest th anks to my research advisor Professor Valeria D. Kleiman, for her inspiring, pa tient guidance and continuous motivation throughout this enjoyable, yet challenging jour ney. She has served as a great mentor and friend from whom I have learned plenty. It was her genuine understanding, unique enthusiasm and well-placed trust that has led this project to completion. I wish to thank my supervisory committ ee members, Professors Jeff Krause and Kirk S. Schanze for their guidance, fruitful collaborations and valuable discussions throughout my graduate studi es, and Professors Russ Bo wers and David Reitze for accompanying me in the final stage of my grad uate career. I extend my thanks to Prof. Adrian Roitberg for his contribution in the theoretical work and Prof. Nic Omenetto for teaching me the basics of lasers. I also thank Prof. Zhonghua Peng from University of Missouri-Kansas for providing unsymmetrical dendrimers and Dr. Joseph S. Melinger for his contributions to the initi al stage of this project. Many friends as well as coworkers have ha d an important role during my graduate studies in Gainesville with their discu ssions and companionships. The members of Kleiman Group provide a fun work environment. Thanks go to Dr. Jrgen Mller for his ongoing friendship and working closely with me on the polymer project presented in Chapter 6 of this dissertation. I thank Daniel Kuroda for his contag ious energy, personal

PAGE 5

v support and having an answer for my every question. I would like to thank Lindsay Hardison for being such a cool girl, for sh aring both science a nd personal matters, and for her suffering to correct my English in this dissertation. I thank Ch ad Mair for his late night support and chats in the lab, and teachi ng me Matlab for data analysis. The newest member of our group, Aysun Altan, thanks for your hard work and contribution of low temperature data in Chapter 4. I would lik e to thank Dr. Chunyan Tan, my collaborator from Dr. Schanze’s Group, for enjoyable disc ussions. I also thank Dr. Wilfredo Ortiz and Julio Palma for their contributions with the th eoretical aspect of de ndrimers. I especially thank Bob Letiecq from Spectra Physics who provided day and night technical support. My time here would not have been the same without the social diversions provided by all my friends in Gainesville. I am particul arly thankful to Mezi yet-Enes family, Enes alik, Ece nr, Dilber and Yavuz for their continuous friendship. I would like to thank zlem Demir for being a wonderful friend a nd great support since the day we met in Ankara ten years ago. I thank my parents Reize and Akil for allowing me to make my own decisions since I was a little child. They always believed in me and supported me whatever I do. I thank my sisters Zehra, Eda, and Seda for shar ing a very happy childhood together. I thank my Aunt Nazmiye for her directions and support whenever I needed. Finally, I give my special thanks to my husband, Avni, for his true love and understanding me without words. You are my s oul mate and you bring colors to my life.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES.............................................................................................................x ABSTRACT....................................................................................................................... xv CHAPTER 1 INTRODUCTION........................................................................................................1 History of Dendrimers..................................................................................................1 Structural Properties.....................................................................................................4 Light-Harvesting Dendrimers.......................................................................................8 Symmetrical PE Dendrimers...............................................................................16 Unsymmetrical PE Dendrimers...........................................................................23 Excitation Energy Transfer.........................................................................................28 Radiative Energy Transfer...................................................................................29 Non-radiative Energy Transfer............................................................................30 Outline of the Dissertation..........................................................................................42 2 EXPERIMENTAL METHODS.................................................................................45 Chemicals and Materials.............................................................................................45 Steady State Measurements........................................................................................47 Why Time-Resolved Spectroscopy?...........................................................................47 The Laser System.......................................................................................................48 Ultrafast Time-Resolved Emission Spectroscopy......................................................51 Time-Correlated Single Photon Counting...........................................................51 Fluorescence Upconversion Technique...............................................................53 Homemade Upconversion Apparatus..................................................................55 Ultrafast Transient Absorption Spectroscopy.............................................................58 Probe Characteristics and White Light Continuum Generation..........................62 Experimental Setup.............................................................................................63 Continuum generation..................................................................................64 Time resolution of the experiment...............................................................69 Concentration and Pump Pulse Energy Dependence..................................................71

PAGE 7

vii 3 ENERGY TRANSFER IN GENERATION 1 UNSYMMETRICAL PHENYLENE ETHYNYLENE DENDRIMERS......................................................73 Materials and Methods...............................................................................................77 Steady State Spectroscopy..........................................................................................80 Time-Resolved Emission Experiments.......................................................................83 Time-Resolved Broadband Transi ent Absorption Measurements..............................88 Kinetic Model.............................................................................................................93 Energy Transfer via Weak Coupling..........................................................................98 Conclusions...............................................................................................................100 4 ENERGY TRANSFER IN GENERATION 2 UNSYMMETRICAL PHENYLENE ETHYNYLENE DENDRIMERS....................................................101 Materials and Methods.............................................................................................102 Steady State Spectroscopy of Ph enylene Ethynylene Dendrimers...........................104 Time-Resolved Emission Experiments.....................................................................106 Time-Resolved Broadband Transi ent Absorption Measurements............................112 Kinetic Model for Energy Transfer..........................................................................116 Energy Transfer in the Weak-Coupling Limit..........................................................121 Vectorial Energy Transfer in Unsymmetrical PE Dendrimers.................................124 Conclusions...............................................................................................................128 5 ENERGY TRANSFER IN SYMMETRICAL PE DENDRIMER: NANOSTAR...130 Materials and Methods.............................................................................................133 Transient Absorption.........................................................................................134 Time-resolved Emission....................................................................................135 Steady State Spectroscopy........................................................................................136 Transient Absorption Spectroscopy..........................................................................137 Model Compound DPA.....................................................................................139 Model Compound Phenylethynylene Perylene.................................................141 Nanostar.............................................................................................................143 372 nm excitation.......................................................................................143 352 nm excitation.......................................................................................145 310 nm excitation.......................................................................................148 Kinetic Model for Nanostar......................................................................................151 Model for 372 nm Excitation............................................................................152 Model for 352 nm Excitation............................................................................154 Model for 310 nm Excitation............................................................................156 Time-Resolved Emission Experiments.....................................................................158 Energy Transfer........................................................................................................163 Conclusions...............................................................................................................169

PAGE 8

viii 6 THE ROLE OF EXCITON HOPPING A ND DIRECT ENERGY TRANSFER IN THE CONJUGATED PO LYELECTROLYTES.....................................................171 Steady State Experiments.........................................................................................174 Time Resolved Fluorescence Spectroscopy.............................................................176 Comparison of Transient Fluor escence and Absorption Data..................................181 Anisotropy and Ionic Complexes.............................................................................185 Loss of Anisotropy and the Absence of Quenching by Free Dye Molecules...........188 Comparison of TAand PL-upconversion Dynamics..............................................189 Quenching Dynamics and the State Contribution....................................................190 Modeling...................................................................................................................192 Conclusions...............................................................................................................201 7 SUMMARY AND PERSPECTIVE.........................................................................204 APPENDIX A THE FLUORESCENCE UP-CONVERSION TECHNIQUE..................................209 The Excitation and Collection of Fluorescence........................................................209 Alignment of the Gate Beam and Up conversion Crystal Phase Matching...............214 Detection of Upconverted Fluorescence...................................................................216 UV-Light Compression............................................................................................217 B DATA ANALYSIS..................................................................................................219 Analysis of Transient Absorption Changes..............................................................219 2G1-m-OH.................................................................................................................223 2G2-m-OH.................................................................................................................227 Nanostar Excited at 310 nm......................................................................................231 LIST OF REFERENCES.................................................................................................235 BIOGRAPHICAL SKETCH...........................................................................................247

PAGE 9

ix LIST OF TABLES Table page 3-1. Fits for time-resolved fluorescence data....................................................................96 4-1. Lifetime measurements from TCSPC.......................................................................107 4-2. Fits for time-resolved fluorescence data...................................................................119 5-1. Fits for transient absorption data..............................................................................158 6-1. Parameters recovered from kinetic m odeling of PPESO3 fuorescence decays with HMIDC in MeOH..................................................................................................176 6-2. Parameters and variables used in the num erical simulations and fitting of the time-resolved PL-upconversion and the transient absorption data........................194 6-3. Average distance between two acceptor mo lecules in complex with the PPESO3..195

PAGE 10

x LIST OF FIGURES Figure page 1-1 Representation of dendrimer growth by the divergent and convergent methods.......3 1-2 Schematic representation of b acterial light-harvesting complexes............................9 1-3 The dendritic triad....................................................................................................14 1-4 Poly(aryl ether) dendrimer functionalized with dyes...............................................15 1-5 Nanostar dendrimer (a) Chemical structure (b) Energy level diagram....................20 1-6 Chemical structures of unsymmetrical PE dendrimers............................................24 1-7 Absorption and emission spectra of GnOH monodendrons.....................................25 1-8 Steady state spectra of GnPer monodendrons: absorption (a) and emission (b).......26 1-9 Chemical structure of PE didendrons.......................................................................28 1-10 Model picture for energy transfer sh owing resonant transitions of donor and acceptor, and spectral overlap of donor emission and acceptor absorption.............31 1-11 Schematic representation of energy transfer mechanism.........................................32 1-12 Definition of the angles used to calc ulate the orientation factor between the dipoles......................................................................................................................35 1-13 Differences between strong, w eak, and very weak coupling...................................37 2-1 The laser system for the production of tunable femtosecond laser pulses with high energy per pulse...............................................................................................49 2-2 Fluorescence up-conversion techniqu e (a) Illustration of the upconversion principle (b) Up-converted fluorescence signal generated in a nonlinear crystal only while the delayed ga te pulse is present............................................................55 2-3 Fluorescence upconversi on experimental setup.......................................................57 2-4 Basic principle of tran sient absorption experiment..................................................59

PAGE 11

xi 2-5 The theoretical scheme of certain signals observed as transient absorption signals.......................................................................................................................6 1 2-6 Experimental setup for transient ab sorption experiment probing with white light continuum. UV pump pulses are obtained from the OPA........................................63 2-7 Spectrum of the white light continuum generated by CaF2. The probe (blue) and reference (red) beams used for tr ansient absorption experiment.............................65 2-8 The chirp of the white light continuu m determined from the delay between the signals.......................................................................................................................6 8 2-9 Coherent artifact of hexane solven t excited at 310 nm, probed at 330 nm and 380 nm......................................................................................................................70 3-1 Chemical structures of generati on 1 phenylene ethynylene dendrimers: (a) 2G1m-OH (b) 2G1-m-Per................................................................................................77 3-2 Normalized absorption spectra of — 2G1-m-OH, ….. G1-OH, and — G2-OH in dichloromethane. Normalized emission spectrum of — 2G1-m-OH.......................81 3-3 Normalized absorption spectra of -----2G1-m-OH, …EPer, and —2G1-m-Per and fluorescence spectrum of —2 G1-m-Per, excited at 315 nm.............................82 3-4 2G1-m-OH in dichloromethane excited at a) 370 nm b) 315nm..............................85 3-5 Upconversion signal of 2G1-m-Per detected at 485 nm, excited at (a) 370 nm, (b) 315nm.................................................................................................................87 3-6 2G1-m-Per Upconversion Signal. excitation=315 nm, emission = 400 nm...................88 3-7 Transient absorption spectra of 2G1-m-OH molecule at di fferent time delays, excited at 315 nm.....................................................................................................90 3-8 Transient absorption spectra of the 2G1-m-OH molecule at different delay times. Detailed display of the 350 nm-450 nm region........................................................91 3-9 Transient absorption spectra of 2G1-m-Per at different time delays........................92 4-1 Chemical structures of phe nylene ethynylene dendrimers: (a) 2G2-m-OH (b) 2G2-m-Per and (c) 3D model of the 2G2-m-Per from a MD simulation................103 4-2 Normalized absorption spectra of -----2G2-m-OH, …EPer and — 2G2-m-Per and fluorescence spectrum of — 2G2-m-Per excited at 320 nm...................................106 4-3 2G2-m-OH in dichloromethane excited at a) 415 nm b) 372nm and c) 330 nm....108

PAGE 12

xii 4-4 Upconversion signal of 2G2-m-Per detected at 485 nm, excited at a) 465 nm, b) 420 nm, c)380 nm and d) 340 nm..........................................................................110 4-5 2G2-m-Per upconversion signal. excitation=320 nm emission = 435 nm...................112 4-6 Transient absorption spectra of 2G2-m-OH at different delay times......................114 4-7 The excitation spectrum of 2G2-m-OH at 298 K (red) and 77 K (blue), emission detected at 450 nm (top). The excitation anisotropy of 2G2-m-OH at 77 K (bottom)..................................................................................................................116 4-8 Model describing the energy ladder. The intermediate st ate (I) is detected at 400 nm or 435 nm. Emission from trap is at 485 nm....................................................125 4-9 Temporal evolution of the inte rmediate state population followed by fluorescence up-conversion....................................................................................127 5-1 Chemical structure of nanostar dendrimer (2 dimensional sketch.........................134 5-2 Absorption and emission spectrum of nanostar in DCM at room temperature......136 5-3 Normalized absorption spectrum of (a) 2-ring (black), 3-ring (green), and 4-ring (red) PE units (b) Nanostar absorption (red) and sum of ri ngs’ absorption at 298 K (c) nanostar absorption at 298 (blue) and 10 K (red).........................................138 5-4 Transient absorption spectr um of model compound DPA.....................................140 5-5 Transient absorption spectrum of the model compound phenylethynylene perylene..................................................................................................................142 5-6 Transient absorption spectrum of nanostar after excitation at 372 nm..................144 5-7 Transient absorption signal as a function of time record ed at 340 nm (black), and 515 nm (blue) following excitation at 372 nm.......................................................145 5-8 Transient absorption spec trum of nanostar after excitation at 352 nm. (a) Short time window ( t< 450fs). (b) Long time window (0.550-50 ps)...........................147 5-9 Transient absorption signal as a function of time record ed at 360 nm (black), and 520 nm (red) following excitation at 352 nm.........................................................148 5-10 Transient absorption spectrum of nanostar after excitation at 310 nm..................149 5-11 Transient absorption signal as a func tion of time for three different excitation wavelengths............................................................................................................151 5-12 Kinetic model proposed for the dynamics of nanostar...........................................153

PAGE 13

xiii 5-13 Fluorescence upconversion signal detected at 485 nm (blue) and at 380 nm (red) following the 310 nm excitation.............................................................................160 6-1 Conjugated polyelectrolytes PPESO3 (l eft) and cyanine dye HMIDC (right)......174 6-2 Absorption and emission spectra of PPESO3 and HMIDC in methanol...............175 6-3 Time-resolved fluorescence of PPESO3 (34 M) in MeOH for different HMIDC concentrations.........................................................................................................178 6-4 Normalized time-resolved fluorescence of HMIDC in MeOH..............................181 6-5 Transient absorption and up-conve rsion signal of PPESO3 in MeOH for different concentrations of added HMIDC.............................................................183 6-6 PL-upconversion and A of a pure PPESO3 solution...........................................184 6-7 Time dependent loss of anisotropy for pure PPESO3 and PPESO3 with HMIDC quencher.................................................................................................................186 6-8 The photoluminescence yield of samp les containing HMIDC, divided by the photoluminescence of the pure PPESO3 solution..................................................190 6-9 Stern-Volmer plot of the time-in tegrated photoluminescence and the instantaneous photoluminescence from the upconversion experiment..................192 6-10 Average distance between quencher mo lecules complexed with the PPESO3 on a reciprocal scale as a function of the quencher concentration..............................195 6-11 Individual contributions to the integrated energy tran sfer, random walk mediated ( a ) direct Frster-transfer ( b ) versus time after excitation.....................................199 A-1 Diagram of off-axis aluminum parabolic mirrors used to collect and image the fluorescence of the sample.....................................................................................210 A-2 Layout of the alignment beams and the collection by parabolic mirrors...............211 B-1 Singular values.......................................................................................................223 B-2 Transient spectra of SVD output............................................................................223 B-3 Dynamics of SVD output.......................................................................................224 B-4 Model fits for the relevant com ponents with large singular values.......................224 B-5 Reconstructed versus real A as a function of wavelength...................................225 B-6 Reconstructed versus real A as a function of time..............................................226

PAGE 14

xiv B-7 Singular values.......................................................................................................227 B-8 Transient spectra of SVD output............................................................................227 B-9 Dynamics of SVD output.......................................................................................228 B-10 Model fits for the relevant com ponents with large singular values.......................228 B-11 Reconstructed versus real A as a function of wavelength...................................229 B-12 Reconstructed versus real A as a function of time..............................................230 B-13 Singular values.......................................................................................................231 B-14 Transient spectra of SVD output............................................................................231 B-15 Dynamics of SVD output.......................................................................................232 B-16 Model fits for the relevant com ponents with large singular values.......................232 B-17 Reconstructed versus real A as a function of wavelength...................................233 B-18 Reconstructed versus real A as a function of time..............................................234

PAGE 15

xv Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ULTRAFAST TIME RESOLVED EXCITATION DYNAMICS IN CONJUGATED DENDRIMERS. By Evrim Atas May 2006 Chair: Valeria D. Kleiman Major Department: Chemistry Light-matter interactions play an important role in light-harvesting processes such as photosynthesis, which has attracted much attention due to its major impact on the cycle of life. Understanding th e fundamental principles of th is energy transfer process is possible through the study of ar tificial light harvesting system s. Dendrimers are perfectly branched synthetic macromolecules having num erous peripheral ch ain-ends surrounding a single core. Incorporating suitable chromophor e groups into their structure can create very efficient antenna systems. This PhD thesis details the dynamics of intramolecular energy transfer in conjugated phenyle ne ethynylene dendrimers. Built-in energygradients in the dendrimer st ructure enable a unidirectiona l energy transfer from the periphery to the core. Depending on the substitution pattern on the phenyl ring, symmetrical and unsymmetrical architectur es are formed that yield different photophysical properties.

PAGE 16

xvi Ultrafast time-resolved fluorescence and ab sorbance techniques are utilized to study the fast dynamics of energy transfer. Ou r approach is based on a comparative study of symmetrical and unsymmetri cal dendrimers with various -conjugation and sizes. Dendrimer backbones are selectively excited at specific absorption wavelengths and the energy migration toward the acceptor is monitored. Time-resolved fluorescence measurements explore the population of intermed iate states and the final energy acceptor, while broadband transient absorption (300 nm to 600 nm) probes the dynamics from the initially excited state to the final trap. To understand the dynamics and mechanisms of energy transfer we propose kinetic models describing the time-resolved data as a function of dendrimer size, presence or absence of a trap and excitation wavelengt h. For unsymmetrical didendrons, typical energy transfer times are in the range of 200750 fs. While absorption is into delocalized exciton states, emission occurs from localized states. In the presence of attached perylene trap, excitation energy migrates through mu ltiple channels. The calculated interaction energies (75-100 cm-1) indicate that dendrons and pery lene are weakly coupled. The symmetrical phenyl ethynylene dendrimer, how ever, shows energy transfer times from 200 fs to 20 ps, much slower than the unsy mmetrical molecule. Considering the broken -conjugation due to the meta substitution, the subunits of the nanosta r are investigated independently via transient absorption. The ki netic model analysis shows that there are both direct and indirect tran sfer (through the cascade) path ways. The experimental energy transfer rates are discussed within the Frster theory to understand the extent of the electronic coupling. In additi on, an ultrafast study of ex citon transport in a phenyl ethynylene polyelectrolyte is perf ormed through quenching experiments.

PAGE 17

1 CHAPTER 1 INTRODUCTION History of Dendrimers The synthesis of dendrimers is an im portant stage in the evolution of macromolecular chemistry. Dendrimers ar e hyperbranched, well-defined, threedimensional, and perfectly monodisperse macromolecules.1-4 Although Flory theoretically investigated the role of branch ed units in macromolecular architectures half a century ago,5,6 the first successful synthesis of a de ndritic structure did not occur until the late 1970s. The first exam ple of an iterative cascade pr ocedure toward well-defined branched structures, such as low molecular weight branched amines, has been reported by the Vgtle group.7 However, not all regularly bran ched molecules are dendrimers. Important characteristics, which will be expl ained in detail later in this chapter, are reached when globularity is achieved at a certain generation and size threshold. The Vgtle group’s cascade molecules are too sma ll to exhibit the pr operties of dendrimers and are used as branched oligomeric bu ilding blocks in dendrimer construction.7 Optimization of the iterative method with Michael addition enabled the synthesis of the first globular dendrimers called PAMAM (polyamidoamine ) by Tomalia et al. at Dow Chemical Research Laboratories.8-10 PAMAM dendrimers are the first dendrimer family to be commercialized and they have been thor oughly investigated to date. Shortly after, Newkome et al.11 reported the synthesis of arborols, another family of trisbranched polyamide dendrimers, and two research gr oups, Mlhaupt and Meijer, were able to improve the Vgtle’s synthesis approach to enable the produc tion of poly(propylene

PAGE 18

2 imine) dendrimers.12,13 These dendrimers were constructe d divergently, implying that the synthesis starts with a functional core mol ecule and is expanded to the periphery. In 1990, Hawker and Frchet introduced the c onvergent approach to produce aromatic polyether dendrimers.14,15 In contrast to polymers, dendrimers are core-shell structures possessing three basic architectural components: 1) a core, 2) repeati ng units in the interior of shells consisting of branching points (generations), 3) te rminal functional groups (periphery). Two complementary methods, the divergent and convergent synthesis, have been used to construct high-generation dendrimers.3,4,15,16 Both methods consist of a repetition of reaction steps, accounting for the creation of an additi onal generation. Within each of these major approaches there may be variati ons in methodology. The features desired for the target molecule and specific building bl ocks justify the choi ce of the synthetic approach. Divergent approach Based on the work of Tomalia and Newkome, the growth starts at the core and proceeds radi ally outward toward the periphery.10,11 The number of reactions that must be completed at each step of growth in creases exponentially. Therefore, a large excess of reagents is requ ired making it harder to maintain the purity and structural uniformity. However, this me thod is used widely for the preparation of high generation dendrimers and for the synthesi s on large scale. The major drawback is the poor yield of defect-free dendrimers. Convergent approach. This method, first reported by the Frchet group, initiates growth at what will become the periphery of the molecule and proceeds inward towards the focal point.14 This approach is best described as an “organic chemist” approach to

PAGE 19

3 globular macromolecules, since it provides out standing control over growth, structure, and functionality.17 The inward growth allows for the reduction in the amount of synthetic steps and intermediate purification at each step of growth. The yield of defectfree dendrimers is about 80%. Figure 1-1 illu strates the dendrimer growth by both the divergent and the convergent methods. Since the discovery of dendrimers, one of the controversial issues has been the purity of these structures. The quality of the final dendritic product is directly related to the chosen synthetic method. A variety of convergently synthesized dendrimers have been reported in the last decade, and thes e dendrimers have shown that the convergent approach provides greater struct ural control than the diverg ent approach; a llowing purity, structural uniformity, and functional vers atility. Another attrac tive feature of the convergent approach is its ability to selectiv ely modify both the focal point and the chain ends. In addition, functional groups can be precisely placed throughout the structure. Overall, the organic nature of the converge nt method results in defect-free dendrimers with appropriate purification. Figure 1-1. Representation of dendrimer gr owth by the divergent and convergent methods. Figure is adapted from Tomalia et al.18

PAGE 20

4 Structural Properties Structural and conformationa l behavior of dendrimers is discussed in many books and publications. Several intriguing questions arise: Are dendrimers always globular or can their shape be highly dist orted? How rigid are they? Ca n the end groups back-fold? Are there cavities present within dendri mers? How do the physical properties change with generation? What are the sim ilarities with linear analogues? The dendrimer structure can be divided in to three distinct architectural regions: core or focal moiety, branched repeat units and end groups on the outer layer. Their structural precision leads to an exact number of branching points or generations, which differentiates dendrimers from hyperbranched polymers. In contrast to linear polymer analogues, dendrimers have several sharp characteristic features. (i) A dendrimer will have size monodispersity due to its well-defined iterative synthesis, whereas most linear polymers are synthesized composing a range of molecular species differing in size and molecular weight. (ii) While the linear polymers contain on ly two end groups, the number of dendrimer end groups increases exponen tially with generation. As the size of the dendrimer increases, the na ture of end groups will determine important properties such as solubi lity, chemical reactivity, and glass transition temperature. (iii) In theory, polymers can grow as much as their solubility allows them, whereas dendritic growth is math ematically limited. The number of monomer units increases exponentially, but the volume available to the dendrimer grows proportionally to the cube of its radius As a result of this physical limitation, dendrimers develop more globular conformation as the generation increases. In contrast to polymers, the intrinsic viscosity of dendrimers does not increase with molecular weight. More extended arrangements for lower generation dendrimers will gradually transform into compact and globular shapes fo r higher generation dendrimers. In general, this gradual transition in overall shape result s in the deviation in physical behavior of dendrimers from those of linear macromolecules.17

PAGE 21

5 Dendrimers might be flexible or fair ly rigid depending on the actual dendritic structure. Recent calculations and measuremen ts have suggested backfolding of the chain ends. For example, the polyether dendrimers synthesized by the Frchet group have been investigated in detail to verify the possibilities for backfolding.19 The flexible nature of these dendrimers implies that the end gr oups are found throughout the dendrimer volume. However, when the end groups can communi cate with each other with attractive secondary interactions such as interactions, electrosta tic repulsions, and hydrogen bonding interactions, the terminal units will assemble at the periphery precluding back folding.20 One of the most studied rigid dendrimer family is phenylethynylene dendrimers, first synthesized by Moore et al.21 These dendrimers are distinguished from other dendrimers by their rigidity and shape persis tence as confirmed by various experimental measurements.22 Another type of shape persistent dendrimers is based on polyphenylene units. Mllen and coworkers inve stigated these molecules an d found that their rigidity originates from the very dens e packing of benzene rings.23-25 Do cavities exist within the dendrimer? Indeed, unlik e linear polymers, properly designed high generation dendrimers exhibit a distinct interior where molecules have been encapsulated in a noncovalent manner.26-28 The encapsulation does not necessarily indicate the presence of a permanent and ri gid cavity within the dendrimer. Especially, flexible dendrimers can accommodate guest molecules. When solvent molecules that freely penetrate dendrimers are removed, the volume collapses leaving the guest molecules trapped inside the dendrimer. For example, the well-designed and rigidified dendrimer structure called “d endritic box” can encapsula te various small organic

PAGE 22

6 molecules and control their release by modi fying the steric crowding of the dendritic periphery.29-31 The encapsulation of a func tional core moiety creating specific site-isolated nanoenvironments leads to a variety of bi oand nanotechnology applications including light-harvesting, amplification, and drug delivery.32,33 Having full control of the structure and architecture, researchers are able to place active sites that have photophysical, photochemical, electrochemical, or catalytic functional groups at the core of the dendrimers. One of the more elegant works on efficient, unidirec tional energy transfer from a dendritic framework to a single chromophore was reported by Xu and Moore.34,35 A gradient effect was cr eated using a poly(phenyle thynylene) dendrimer. The conjugation length of the repeat units of this dendrimer in creases with generation from the periphery to the core. This is the so cal led “nanostar” molecule later discussed in Chapter 5 of this dissertation. The phenylet hynylene units of Moore dendrimers can be used to create unsymmetrical dendrimer architect ures, which are also investigated in this dissertation (Chapters 3 and 4). Frchet group’s poly(benzylether) dendrim ers, functionalized with different dye chromophores at the periphery and core, are ab le to harvest light a nd transfer the energy efficiently to a chromophore located at the center of the dendrimer structure. It was shown that the core chromophor e emission is significantly amplified compared to the same chromophore without the dendritic fr amework. As the size of the dendrimer increases, so does the number of peripheral units, therefore the energy transferred to the core increases due to a larger absorption cross-section.36-38Another application of the same dendrimer structure is in optical signa l amplification. Luminescent lanthanide ions

PAGE 23

7 are used as signal amplifiers for optical fiber communications. However, their selfquenching in the solid state greatly limits the effectiveness. Kawa et al. encapsulated individual lanthanide ions within poly(benzy lether) dendrons lead ing to site-isolation, thereby decreasing the self-quenc hing effect. This antenna lig ht-harvesting effect results in emission signal amplification.33 In another report, Jiang and Aida used az obenzene containing aryl ether dendrimers to study energy transfer.39 They demonstrated that cis-trans isomerization of azobenzene moiety at the core accelerat es for larger generation (e.g. G-4, G-5) dendrimers. The acceleration was observed via exciting a stretchi ng mode of the aromatic rings with IR irradiation. However, UV excitation of the dendrons did also result in accelerated isomerization. Thus, it was proposed that th e dendritic shell not only insulates the azobenzene core from collisional energy diss ipation, but it acts as a photon harvesting antenna. Mllen and coworkers reported polyphenyl ene dendrimers functionalized with different chromophores at the peripher y and core, while the De Schryver group investigated this family of dendrimers and explored the energy transfer dynamics via time-resolved spectroscopy experiments.40,41 The dendrimer rigid structure decorated with a unique selection and pos itioning of the chromophores allows a systematic study of possible energy transfer mechanisms. A different type of encapsulation involv es the formation of metal nanoparticles within dendrimers and it has been widely us ed to prepare organic-inorganic composite structures useful in catalytic applications.42 Since dendrimers have nanoscopic dimensions and can be dissolved molecularl y, catalytic active site can be placed at a

PAGE 24

8 particular, isolated position resulting in beneficial inte ractions with the substrate.17 Brunner et al.43 introduced the first branched molecules containing internal core catalytic sites, later called “dendrzymes.” In parallel to this work, the first dendritic catalyst with multiple catalytic sites at the periphe ry has been reported by Ford et al.44 Both studies concluded that low-generation dendrimers are better catalytic supports than higher generation dendrimers. More recently, Crooks et al. have shown that substrates can penetrate the dendrimers to access the catalytic sites and undergo simple reactions such as hydrogenations.42 The Crooks group also develope d a system composed of PAMAM molecules covalently attached to a metallic su rface and tested its function as a chemical sensor.45 Recently, a sequence of dendrimers contai ning Zn-porphyrin monomers situated on the surface were investigated by Sundstrm and coworkers.46,47 The goal of this project was to study energy transfer between the in dividual Zn-porphyrin s within a dendrimer, and to measure whether this process become s more efficient with increasing dendrimer size. The molecules based on porphyrin chro mophores are an example of compounds that can be used as a model for the bacteriochlo rophyls (BChls) in the na tural light-harvesting (LH2) complex. Before going through the details of these sp ecific structures, a brief summary of light-harvesting dendrimers and the role of an energy gradie nt within these dendrimers will be presented. Light-Harvesting Dendrimers Photosynthesis is an extremely effective na tural process for harvesting sunlight and converting sunlight into useful chemical energy stored in th e form of ATP. Thus, it has been of vital importance to the evolution of life and it is essential for almost all life-

PAGE 25

9 forms. Therefore, it is inev itable that the photochemistry of photosynthesis be the focus of considerable scientific research. There are two key processes in photosynthesi s. First is the absorption of photons by an antenna system, followed by a rapid and effi cient transfer of excitation energy to the reaction center. Then, a sequence of charge tr ansfer events from the excited state of the reaction center results in storage of chem ical energy. To date, the most studied photosynthetic system is pr obably the purple bacteria.48 The high resolution X-ray crystal structure of this bacteria? reveals a cent ral reaction center surr ounded by light harvesting complexes as shown in Figure 1-2. These chlorophyll containing assemblies are capable of absorbing photons from a broad spectral rang e of sunlight, which makes them a perfect light-harvesting antenna. Figure 1-2. Schematic representation of bacterial light-harvesting complexes.

PAGE 26

10 In the last decade, there has been grea t devotion to the design and synthesis of molecular or supramolecular species that can function as antennas in artificial systems. The first requirements to develop a light ha rvesting system are that its components must absorb in a substantial part of the visible spectral regi on and the light-absorbing units must be chemically and photochemically stable In order to have high light-harvesting efficiency, the excitation energy must be delivered to a common acceptor component. Several groups have tried to develop ar tificial light harvesting systems with custom-designed molecules such as small c ovalent arrays containing photoactive units, polymeric and supramolecular systems with multichromophores.17,20 It has been concluded that linear-chain macromolecules do not have the most ideal architecture for efficient light harvesting.20 First of all, it is difficult to make polymeric systems with an energy gradient, which is shown to be vital for vectorial energy transfer. Secondly, most polymeric chains are flexible enough to form aggregates or excimers which will act as energy traps. In this regard, dendrimers char acterized by their high degree of order and the possibility to contain selected photoactiv e chemical units in predetermined sites are excellent candidates for light harvesting antenna. Proper choice and placement of chromophores enable the investigation of effi cient energy transfer from the periphery to the core of the dendrimer. Balzani and coworkers49 have reported initial studies on multichromophoric dendrimers undergoing intramolecular energy tran sfer. They incorporated different metal and ligand combinations into low-generation dendrimers and found that energy transfer occurs from internal higher energy units to the external lower energy units. The concept of intramolecular energy transfer was clearly illustrated by these initial reports.50-55

PAGE 27

11 However, these structures are not ideal photos ynthetic mimics. As mentioned before, the optimum light harvesting system should have numerous peripheral chromophore channeling the absorbed energy in a unidirect ional manner to a single and central energy acceptor molecule or complex. The first report of an efficient, unidirectional energy transfer from a dendrimer structure to a single core chromophore was published by Xu and Moore.34 These systems and their properties will be explained in details in the symmetrical phenyl ethynylene dendrimers section. The Frchet group has then studied the lanthanide –cored poly(benzylether) dendrimer. In this study, it was shown that the excitation energy was channeled from a dendrimer shell to a single core unit.33 Later, Jiang and Aida observed similar antenna effect utilizing a different lumi nescent core su ch as porphyrin.56 A variety of structures were designed and studied differing in the number of poly(be nzylether) dendrons attached to the central porphyrin as well as in the generation number of dendrons,.56 In a similar manner, Balzani et al reported a polylysin dendrimer with chromophoric dansyl units in the periphery playi ng the role of a ligand for lanthanide ions with efficient conversion of UV light into light of different frequencies in the visible or near infrared region.52,53,57 In another approach, also devel oped by the Frchet group, the flexible poly(benzylether) dendrimer wa s functionalized with dye mol ecules at the periphery that served as the molecular antenna while the core functionalized with a proper dye molecule served as the energy acceptor Both steady state and time-re solved experiments indicated that the energy migration from the periphery to the core wa s extremely efficient, thus most of the absorbed energy is concentrated at a single center.36,58,59

PAGE 28

12 Inspired by natural photosynthetic system s including elegant light-harvesting antenna of photosynthetic bacteria, system s capable of directional energy transfer between several chromophores have gained much attention.60 In the absence of a gradient, the exciton will randomly hop betw een neighboring localized states following the photoexcitation. The hopping probability is based on the separation distance between chromophores. Due to the branched structure, there is an entropic bias increasing the probability of energy dissipation outwards towa rd the molecular periphery rather than inwards toward the core. Recent theoretical investigations of dendrimers without an energy gradient have shown that the efficiency of exciton trapping at the core decreases with an increase in molecular size, even though larger number of absorbing units is present.61,62 On the other hand, the presence of an energy gradient toward the locus will introduce an energetic bias that will overcome the entropic bias. Two different approaches are known to create dendrimers with an energy gradient. In one approach, Moore and coworkers developed a series of dendrimers serving both as the lightabsorbing antenna and as an energy transport medium, which is the case for the dendrimers investigated in this doctoral res earch. In the other approach, both periphery and core are functionalized with donor and acc eptor moieties, respectively. The dendritic framework is photochemically silent and acts as a transparent spacer to separate the donor groups at the periphery from the acceptor groups at the core.40,63 An example of this later approach was reported by Mllen and co-workers64,65 who investigated structurally we ll-defined, conformationally ri gid dendrimers consisting of up to three different chromophores. This “dendr itic triad” includes globular polyphenylene dendrimers bearing a terrylene tetracarboxdi imide (TDI) chromophor e in the center and

PAGE 29

13 perylene dicarboxmonoimide (PMI) as we ll as naphthalene dicarboxmonoimide (NMI) chromophores at the periphery as shown in Figure 1-3. The design of the cascade system places the naphthalene chromophore at the thir d branch point, the perylene chromophore at the second branch point, and the terry lene chromophore at the core representing spatially the desired direction for energy tr ansfer. These rylene chromophores were chosen since they possess excellent photos tability, high extinction coefficients, and fluorescence quantum yield clos e to unity. The triad absorb s over the whole range of visible spectrum and shows well-separated abso rption envelopes. Thus, it is possible to specifically excite distinct chromophores w ithin the dendrimer, which helps with the investigation of vectorial en ergy transfer. Mllen group’s work64-66 clearly indicates the existence of an energy gradient and is consis tent with the stepwise energy transfer from periphery to the center of the molecule. Recently, single molecule fluorescence st udies on the same triad further support the multi-step unidirectional energy transfer, including a component of direct transfer from each donor to the acceptor. Note that using a rigid polyphenylene dendrimer overcomes many possible complications due to conformational mobility. Undesired chromophore interactions such as aggreg ation, excimer formation, and even self quenching of dyes are minimized with a shape-persistent dendrimer. The Mllen group’s work represents the firs t example of a dendritic triad in which energy gradient is induced by different types of chromophores spatially and energetically distributed within the dendrimer and thus independent of the dendrimeric scaffold.

PAGE 30

14 Figure 1-3. The dendritic tria d. Adapted from reference64. Shortly after, Frchet reported the de sign and synthesis of a cascade light harvesting system based on a flexible dendrimer scaffold. In order to obtain the required spatial distribution, coumarin 2 and fluorol dyes were placed at the third and second branch point of a poly(aryl ether) dendrim er, respectively (Figure 1-4). The final energy acceptor consisted of a perylene derivative at the core of the dendrimer. Similar to polyphenylene dendrimers, the dendritic bac kbone does not participate in the energy transfer process. The steady state photophysical analysis suggested en ergy transfer within this system favoring a cascade route, moving from coumarin groups through intermediate fluorol units and into a fi nal acceptor ethynylen eperylene chromophor e. This system demonstrates that with proper chromophore sele ction, vectorial energy transfer process is generated despite the flex ibility of the dendrimer.

PAGE 31

15 Figure 1-4. Poly(aryl ether) dendrimer functiona lized with dyes. Adapted from reference67. The dendrimers mentioned previously point out the role and necessity of an energy gradient for efficient energy transfer process as in the natural photosynthetic systems. However, for these nonconjugated dendrimers, th e role of the dendritic backbone is only structural and not functional (the dendrime r backbone is just a spacer). In conjugated dendrimers, the backbone itself serves as a to ol for the energy transfer. The first built-in multistep energy gradient within dendrimers was reported by Moore and coworkers using a phenylene ethynylene (PE) dendrimer, with a repeat unit conj ugation length that increased with generation from periphery to the core.68-70 As a result, HOMO-LUMO gaps of conjugated repeat units decrease from the exterior to the interior, generating a directional energy flow toward the core. Th ey also investigated the PE dendrimers without a gradient, composed of phenylethynylene chains of identical lengths. These PE dendrimers are characterized as structurally symmetric due to the meta substitution of the benzene ring at each branching point. R ecently, a new type of PE monodendron was

PAGE 32

16 characterized as unsymmetrical since the bran ches extending outward were structurally nonequivalent and linked through meta and para positions on the phenyl rings. The unsymmetrical PE dendrimers also possess an intrinsic energy gradient resulting in efficient energy funneling.71 The work presented in this dissertation mainly comprises the investigation of the photophysical properties and the energy tran sfer processes in several unsymmetrical PE dendrimers and a symmetr ic PE extended dendrimer. The primary photophysical characterization of symmetrical and unsymme trical PE dendrimers is summarized in the next section. Symmetrical PE Dendrimers Theoretical studies showed that ordered “Bethe trees” might be the optimal energy funnels.61,72 With this in mind, Moore and Kopelm an synthesized and investigated the photophysics of a series of phenylethynylene dendrimers.72 Two families of these dendrimers, compact ad extended, are char acterized by symmetrical branching. The branching point is always a meta substitution of the phenyl ring leading to structurally symmetric macromolecules. Branching at para positions would grow linear chains, while branching at ortho positions would terminate the tree-like structure qu ickly due to steric hindrance. Thus, symmetric geometry for a large dendrimer is optimized with meta arrangement, which also permits a large degr ee of orientational flexibility. Deviations from planar configurations overcome the st eric hindrance and enable the synthesis of higher generation dendrimers.22,73 The main difference between compact and extended dendrimers is the number of phenylene ethynylene units between consecutive branching points. In compact ones, each generational unit is composed of identical diphenylacetylene chains The extended ones have diphenylethynylene chains around the periphery, but linear phenyleneethynylene

PAGE 33

17 chains show consecutively increasing length toward the center of the molecule. Even though this seems to be a minor modification to the molecular structure, it introduces significant electronic and energetic characte ristics that greatly enhance the energy funneling abilities of these systems.70 Recently, Ortiz et al. have presented a theore tical investigation of energy transfer in the nanostar molecule.74 Molecular dynamics simulations have been performed to reveal the role of structural cha nges on the dynamics. The energy tr ansfer rates were calculated between 2and 3 –ring chromophores using th e ideal dipole approximation (IDA) and the transition density cube method (TDC). The rapid flipping of the phenyl groups at room temperature resulted in large changes in tr ansition densities. It was shown that the traditional Frster model employing IDA was not able to reveal this dynamical effect on the transfer rates. Also, the accuracy of th e IDA fouls when the size of the chromophores is comparable to the distance between them. However, the rate constants obtained with TDC were extremely sensitive to the phenyl rota tion and therefore expected to yield more accurate energy transfer rates. In addition, Kleiman et al.75 investigated the energy transfer in the nanostar w ith femtosecond degenerate pump-probe spectroscopy. They measured the recovery time of the gr ound state absorption of 2-ring and 3-ring chromophores. The experimental transfer rates were compared with the calculated ones using the Frster model. Even though there wa s a qualitative agreement, the rates were overestimated. Ortiz et al. discussed that the discrepancy between Frster model and experimental results would be improved by th e use of TDC and the data from molecular dynamics.74

PAGE 34

18 Theoretical calculations by Mukamel et al. indicate that meta branching electronically decouples the reso native conjugation among PE units.76-79 As a result, ,the optical excitation is localized on each PE chain in compact and extended dendrimers. Experimental evidence for this excitonic localization can be seen by steady state spectroscopy. The absorption spectrum of a ny compact structure cl osely reproduces the spectrum of an isolated PE unit. The tota l absorption intensity increases monotonically with generation, but exhibits no red-shift. If excitations were delocalized over the entire dendrimer backbone, a red-shift should have been observed. Due to identical chain length of all subunits, compact dendrimers have an en ergetically degenerate nature, allowing the exciton to rapidly hop between ne ighboring localized states. An exciton initia lly localized on a particular PE chain will not encounter any energy gradient towards the locus. On the contrary, an entropic bias is observed whic h increases the probability of hopping toward the periphery. Thus, compact dendrim ers do not act as energy funnels. The extended PE dendrimers also exhibit lo calized electronic excitations due to branching at meta positions of phenyl ri ngs. The difference lies in the HOMO-LUMO energy of these localized excitations. It is known that the HOMO-LUMO excitation energy of a molecule decreases with an increase in the ex tent of conjugation. While the single diphenylacetylene? (DPA) chains around the periphery ha ve the greatest excitation energy, this value decreases monotonically towa rd the center of the mo lecule as the chain length increases. The absorption spectra for th e extended dendrimer series also exhibit a high-energy peak assigned to the shortest DPA chains, but additionally increasing red shifted peaks are observed associated with longer PE units (3and 4-ring).80 As a result, a

PAGE 35

19 built-in energy gradient is observed and intr amolecular energy transfer in the extended dendrimer series is well directed from periphery to the core. The random hopping and funneling characteri stics of both compact and extended PE dendrimers were investigated theoretical ly. Mukamel and coworkers investigated the optical properties of such systems usi ng the Collective Electr onic Oscillator (CEO) approach and the Frenkel-exciton model.76,77 In parallel to the experimental results, theoretical studies have shown that optical ex citations involve no char ge transfer and are completely localized between linear segmen ts. Since the meta branching disrupts the charge transfer between i ndividual PE segments, exciton migration proceeds via Coulombic interaction and these systems can be represented by the Frenkel Exciton Hamiltonian. The linear absorption spectra of these dendrimers were calculated using the CEO approach and showed excellent agreemen t with the experiment. It was concluded that the linear segments can be consider ed as effective chromophores where optical excitations reside. Upon photoexc itation, the electron-hole pair is confined to a single chromophore, whereas its center of mass can move around representing energy migration across the molecule. The extended PE series represent the firs t example of a built-in multi-step energy gradient within dendritic systems and are very efficient photosynthetic mimics. By functionalizing the core of these structures with the lower-bandgap ethynyleneperylene chromophore, an energy “sink” is introduced into the system.81 The most studied extended derivative, both experimentally and theoretically, has 4 generations (2-, 3and 4ring PE units) and is referred as the “na nostar” (Figure 1-5a). Excitation of the nanostar backbone results in emission emanating solely from this ethynyleneperylene

PAGE 36

20 dye. Hence, the PE units act as energy donor s and ethynyleneperylene acts as the central acceptor. The absorption data along with the lifetime data indicate that the ethynyleneperylene unit at the focal point of nanostar has a well localized excited state. Figure 1-5b shows the energy level diagram illustrating the vibrationless electronic excitation energies of each of the localized states in the nanostar.69,82 This graphic representation points ou t the impressive energy funneling ch aracteristics of the nanostar. (a) (b) Figure 1-5. Nanostar dendrimer (a) Chemical structure (b) Energy level diagram. Adapted from reference 80 Theoretically, the Mukamel group intensivel y investigated the nanostar molecule and computed the exciton energies, transition dipole moments, and electrostatic interactions in PE segments using the CEO method.76,78 They computed linear absorption, frequency gated fluorescence spectra, and even frequency-domain pu mp probe signal. In addition, Tada et al. investigated photoex cited states of the nanostar and singlet excitations of linear PE units involved in the dendrimer with time dependent density functional theory and molecular orbital method.83 It was concluded th at the orbitals of nanostar are localized in space as well as in energy. While the steady state comparison

PAGE 37

21 with these calculations implies the model of weakly coupled Frenkel excitons, the dynamics of the excited states in PE dendr imers may lead to a different picture of excitations. This issue is crucial to unders tand the nature of elect ronic excitations and further calculations have been recently carried out by Bardeen and coworkers. They reported both high-level electr onic structure calculations and steady-state experiments based on the smallest building blocks of PE dendrimers.84 Their emission spectral shapes, radiative lifetimes, and anisotropies change dramatically with increasing number of substituents. This yields strongly couple d diphenylethynylene un its and contradicts previous findings. The excite d state electronic structure wa s investigated theoretically using ab initio CASSCF and CASPT2 calculations and the electronic coupling was found to vary with molecular geometry. In partic ular, the presence of la rge electronic coupling in the emitting geometry was not seen for th e absorbing geometry of the same molecule. In order to analyze the variab ility in electronic coupling, th ey extended their ab initio results in terms of the Harcourt model.85 This model was develope d to classify different interactions such as through bond, through space a nd charge transfer interactions between coupled chromophores. The relative roles of these three interac tion terms and their dependence on metaversus parasubstitution were investigated in detail. However, the experiments in larger dendrimers do not show the spectral features (shifts) predicted in the smaller systems.70,86 So, the nature of the excited st ates for PE dendrimers remains an open question. In another study to address the compli cated photoexcitation energy transfer between subunits of a dendritic system, a simple compact model system with one branching center was investig ated by Goodson and coworkers.87,88 In that study, each

PAGE 38

22 component of the branching center was a conj ugated linear segment. The interactions between these segments (chromophores) are strongly influenced by the electronic and structural connectivity of the branching center. Varnasvski et al. have investigated the nature of these interactions using fluorescence anisotropy.87 For example, the very fast depolarization rate in a nitr ogen centered triphenylamine molecule indicates strong electronic coupling between segments. In another publication by the Goodson group, fluorescence anisotropy dynamics of a syst em containing pyridine distyrlbenzene chromophores attached to benzene center was reported.89 Their results confirmed that the benzene branching center acts as a weak coupl er and electronic de localization across the branching center is hindered by meta substitu tion of the chromophores. It is important to recall that phenyl is the br anching center between the in dividual chromophores in PE dendrimers. Therefore, most of the th eoretical and experimental studies are complementary in terms of verifying the loca lized nature of excitations for symmetrical PE dendrimers, with the exception of Bardeen’s results.84,90 Our group has an ongoing collaboration with Prof. Jeff Krause and Prof. Adrian Roitberg to investigate this unique symmetric PE dendrimer nanostar. As discussed in detail in Chapter 5 of this thesis, we performed femtosecond time-resolved experiments on the nanostar. We explored the excited st ate dynamics by measur ing the fluorescence from both the ethynyleneperylene trap (accepto r) and the dendritic backbone (donor). Broadband transient absorption following the excitation at different chromophores was also examined. In addition, room and low te mperature steady-state absorption spectra of the individual PE components and the na nostar were measured and compared to theoretical calculations performed to predict the spectra.

PAGE 39

23 Unsymmetrical PE Dendrimers Most of the dendrimers developed for light harvesting applications have symmetrical structures. As discussed in th e previous section, for symmetrical PE dendrimers, the electronic comm unication between the peripher y and the core has to go through each sub-branch. For unsymmetrical dendrimers, shortcuts exist between the periphery and the core, such that in some cases shorter PE chains are directly attached to the longest chain extending to the core. Th e core of the dendrimer can directly communicate with the periphery. Therefore, it is anticipated that unsymmetrical dendrimers may be better light-harvesting ante nna molecules. To prove this concept, Peng and coworkers91-94 reported a new class of conj ugated PE dendrimers based on unsymmetrical branching which occurs at both ortho and meta positions of the branching benzene rings and leads to nonequivalent branches. Two features are crucial for unsymmetrical PE dendrimers: rapidly incr easing conjugation lengt h results in broad absorption spectra and the conjugation length increases toward the core generating an intrinsic energy gradient. Figure 1-6 shows the structures of unsymmetrical PE monodendrons. These dendritic molecules have both ortho and me ta substitution while the symmetrical ones have only meta linkage. The conj ugation length of the longest chain is significantly larger compared to the meta-linked dendrimer composing of the same number of phenyl ethynylene groups. Therefore, they are expected to have different optical properties.

PAGE 40

24 HO MeOG1OH HO OMe MeOG2OHOMe HO OMe MeO MeO OMe OMe OMe MeOG3OH Figure 1-6. Chemical structures of unsymmetrical PE dendrimers Figure 1-7 shows the absorpti on and emission spectra of GnOH (n=1-4) unsymmetrical PE dendrons. As the generation number increases, the lowest excitation energy (absorption band edge) shifts to longer wavelengths and the molar extinction coefficient increases. Even though the ortho s ubstitution prevents th e phenyl rings from having a planar geometry due to steric hind rance, the effective c onjugation length clearly increases with each generation. Variable c onjugation length throughout the dendrimer accounts for the much broader absorption sp ectrum than that of the meta-linked HO MeO OMe MeO MeO MeO OMe OMe MeO OMe MeO MeO OMe OMe MeO MeOG4OH

PAGE 41

25 symmetric analogs. This broad absorpti on spectrum along with direct electronic communication between periphery and the co re could possibly make these types of dendrimers more efficient energy transfer f unnels. As seen in the emission spectrum, fluorescence is red shifted for higher generation dendrimers. Following the photoexcitation, the energy transfer will pr oceed from shorter conjugation-length segments to the longest conj ugated segment, thus for a gi ven dendrimer, emission is only observed from the longest chain. Th e fluorescent quantum yield of GnOH dendrimers varies from 40% up to 80% (in CH2Cl2). 2503003504004505000.0 0.2 0.4 0.6 0.8 1.0 G1OH G2OH G3OH G4OHAbsorbanceWavelength (nm) 3003504004505005506000.0 0.2 0.4 0.6 0.8 1.0 G1OH G2OH G3OH G4OH Fluorescence Intensity (normalized)Wavelength (nm) Figure 1-7. Absorption and emission spectra of GnOH monodendrons. Adapted from reference71 Since it is hard to differentiate each segment within the unsymmetrical dendrimers, it would be hard to quantita tively evaluate the energy transfer efficiency. However, analogous to the symmetrical PE dendrime rs, an ethynyleneperyl ene unit serving as energy trap has been attached to th e focal point of the dendritic backbone. Ethynyleneperylene, having well -separated absorption from the dendrimer absorption, will help explore the excitation energy transf er. The absorption and emission spectra of GnPer series are shown in Figure 1-8. The abso rption features of pery lene can be clearly distinguished from the PE backbone. The pe rylene bands are approximately 50 nm red-

PAGE 42

26 shifted compared to the free perylene in the CH2Cl2, which indicates th e delocalization of perylene transition dipole over the PE backbone. The fluorescence spectra of GnPer series following 350-nm excitation is also shown in Figure 1-8b. For any given dendrimer, the emission is almost entirely from the perylene trap, which indicates very efficient energy transfer from the dendrimer backbone to th e ethynyleneperylene tr ap. Melinger et al. presented a detailed photophysical characte rization of unsymmetrical PE monodendrons in various solvents.95 They reported steady-stat e absorption and fluorescence measurements along with the time-depende nt fluorescence measurements for PE monodendrons up to fourth generation. The phot ophysical properties of unsymmetrical PE monodendrons were compared to those of symmetrical PE dendrimers. In addition, ultrafast degenerate pump-probe spectroscopy was applied to G3OH and G3Per to explore the excited state dynamics.95 These initial measurements s uggested that energy transfer process into the ethynyleneperylene trap occu rs in a subpicosecond time scale. However, a direct measurement of the trapping time is yet to be ascertained through multi-color pump probe experiments or by measuring the time evolution of the ethynyleneperylene fluorescence with femtosecond resolution. This is one of the goals of this dissertation. (a) (b) Figure 1-8. Steady state spectra of GnPer monodendrons: absorption (a) and emission (b). Adapted from reference96

PAGE 43

27 Recently, Peng and coworkers synthesized PE didendrons along with some tridendrons so that an extensive investigati on is possible to establ ish a structure-property relationship regarding energy transfer and -conjugation.97 In that work, two PE monodendrons were linked by a phenyl ring at me ta positions and these new structures were named as 2GnOH series. For 2GnPer dendrimers, an ethynyleneperylene unit was attached to the central benzene ring in meta position. Due to meta substitution at the core of the dendrimer, the conjugation is expected to be disrupted a nd an extra degree of localization might be provided between the monodendrons and/or between the monodendron and the ethynyleneperylene unit. The chemical structures of didendrons are illustrated in Figure 1-9. Through collabora tion with the Peng group, we obtained the didendrons studied in this dissertation. Mainly, 2GnOH and 2GnPer ( n=1,2) are studied via both steady-state and time-resolved spect roscopic techniques in Chapter 3 and 4. Even though initial studies with unsy mmetrical PE monodendrons reveal some spectroscopic evidence for the highly efficien t and ultrafast energy transfer, questions remain regarding the nature of electronic excitations and related mechanisms for the transfer process. The ultrafast absorp tion and emission experiments designed and performed in this dissertation aim to explore these questions. Unsymmetrical PE dendrimers are an attract ive prospect as they offer a handle to obtain various extends of conjugation and he lp in understanding electronic structureproperty relationship for bette r light-harvesting systems. The processes following the optical excitation in a molecule and basic en ergy transfer mechanisms will be explained briefly in the following section.

PAGE 44

28 OCH3OCH3X2G1X OCH3OCH3OCH3H3CO OCH3H3CO X2G2X X OCH3OCH3OCH3H3CO OCH3OCH3OCH3OCH3H3CO H3CO OCH3OCH3OCH3H3CO2G3X 2GnOH (n=1-3): X=OH 2GnPer (n=1-3): X= Figure 1-9. Chemical stru cture of PE didendrons. Excitation Energy Transfer The motivation behind the work presented in this dissertation is to identify the mechanisms of intramolecular electronic exc itation energy transfer in light-harvesting dendrimers and the structure-function relati onship that make energy transfer very efficient in these systems. Other than self -relaxation processes, the excited states may relax to the ground state via tr ansferring the electronic exci tation to other chromophores present in the system by a bimolecular proc ess. During this process the excited donor chromophore D* returns to its ground state w ith simultaneous transf er of its electronic energy to the acceptor chromophore A: D* + A D +A* Subsequently, the photoexcited chrom ophore A* may proceed either giving a sensitized photochemical reac tion or exhibit sensitized pho toluminescence. Under these conditions D chromophores are termed as sens itizers while A chromophores as activators. There are mainly two conditions required for energy transfer to occur: (i) the energy of

PAGE 45

29 D* should be higher than the energy of A*; (ii) the energy transf er process should be faster than the natural lifetime of D*. The electronic energy transfer can be described further according to the photophysical processes that are involved in it. Radiative Energy Transfer Radiative energy transfer is a two-step process: a photon is first emitted from the excited donor and then it is reab sorbed by the acceptor ground state: D* D + h h + A A* Since the transfer mechanism is based on a radiative step (i.e. photons emitted by one chromophore are absorbed by a second chro mophore), this process is named as the “trivial” case of electronic energy transfer. It does not involve the di rect interaction of donor and acceptor. The most important factor that influences the process is the quantum efficiency of the donor in the spectral re gion where the light-absorbing ability of the acceptor is high. The trivial transfer is favor ed when the following conditions are met: high quantum yield of D*, high concentration and extinction coefficient of A, and good overlap between the emission of D* and abso rption of A. This kind of energy transfer might be the dominant mechanism in dilute solutions since the dependence of energy transfer efficiency on separation distan ce between donor and acceptor chromophores is weak. The viscosity of the solvent does not a ffect the rate of radi ative energy transfer. Radiative transfer re sults in a decrease of the donor fluorescence intensity in the region of spectral overlap. Wh en donor and acceptor chromophores are identical and emission and absorption overlap sufficiently, the observed fluorescence lifetime increases as a result of repeated absorption and emission (radiative trapping).

PAGE 46

30 Non-radiative Energy Transfer Non-radiative energy transfer is a single step process that requires donor-acceptor interaction as a result of spectral overlap between donor’s emission spectrum and acceptor’s absorption spectrum. D* D and A A* transitions are isoenergetic, implying that several vibronic transitions in the donor will have the same energy as the corresponding transitions in the acceptor. Su ch transitions are coupled, and are in resonance as shown in Figure 1-10. For non-radiative energy transfer, the terms resonance energy transfer (RET) and excitati on energy transfer (EET) are often used. If the excited state vibrational relaxation is faster than the energy transfer, and if the energy transfer is a vertical process as implie d by the Franck-Condon principle, the spectral overlap can be evaluated using: 0()()DAJIvvdv (1-1) This integral is proportional to the number of resonant transitions in the emission spectrum of the donor and absorp tion spectrum of the acceptor as illustrated in Figure 110. The spectral distribution of the donor emission and the acceptor absorption are normalized to a unit area on the wave-number scale: 00()()1DA I vdvvdv (1-2) According to this required spectral over lap condition for normalized spectra, it is clear that the magnitude of the spectral overlap is not connected to the absolute values of the oscillator strengths of the transiti ons that are involved in the process. There are two different interaction mech anisms related to non-radiative energy transfer process: coulombic and exchange interactions. The coulombic interactions

PAGE 47

31 consist of long-range dipole-dipole interactio ns and short-range mu ltipolar interactions. The interactions due to intermolecular orbi tal overlap include tw o electron exchange (Dexter mechanism) and charge resonance in teractions, which are effective in short range. The total interaction energy can be written as the sum of a coulomb term, Uc and an exchange term Uex. Considering that only two electr ons are involved in a transition (one on D and one on A), the energy transfer mech anisms are schematically represented in Figure 1-11. In the figure, the empty circles represent the electrons whose interactions with other electrons are assume d to be roughly constant during the energy transfer step. DONOR ACCEPTOR D* A* D A 3211*2*3* E resonant transitions Figure 1-10. Model picture for energy transfer showing resonant transitions of donor and acceptor, and spectral overlap of donor emission and acceptor absorption.

PAGE 48

32 The coulombic term corresponds to the pr ocess in which the initially excited electron on the donor (1) returns to the ground st ate orbital on D, while simultaneously an electron on the acceptor A (2) is promoted to excited state (Figure 1-11, top). The exchange term is represented with an excha nge of two electrons between D and A, which is analogous to a moving part icle transferring its energy to the other particles via collisions. Coulombic resonance interaction o ccurs via electromagnetic field and does not require physical contact of interacting donor and acceptor. The basic mechanism involves the induction of a dipole oscillation in A by D*, thus the Coulombic mechanism can be effective at large distances (up to 80-100 ). As shown in Figure 1-11(bottom), the exchange interaction represents a “double” electron substituti on reaction, i.e., the electron 1 2 electron exchange electron exchange 1 2D*(1) A(2) D*(1) A(2) C o u l o m b i c i n t e r a c t i o n LUMO HOMO HOMO LUMO 1 2 electron exchange electron exchange 2 1D(2) A*(1) D(1) A*(2) C o u l o m b i c i n t e r a c t i o n D A D A HCHE Figure 1-11. Schematic representation of en ergy transfer mechanism. Top: Coulombic mechanism. Bottom: Exchange mechanism98

PAGE 49

33 initially on D* jumps to A simultaneously with the jump of an electron on A to D*. This exchange interaction occurs via overlap of the molecular orbitals requiring physical contact between donor and acceptor. The interact ion is operative only at short-range (1015 ) because the electron density exponentia lly decreases outside the boundaries of molecules. For allowed transitions on D and A (no cha nge in spin), coulombic interaction is the predominant mechanism. Thus, singlet-singlet energy transfer such as 1D* + 1A 1D + 1A* and 1D* + 3A 1D + 3A* are fully allowed. However, for forbidden transitions on D and A: 3D* + 1A 1D + 3A coulombic interaction is negligible and triplet-triplet energy transfer is only due to orbital overlap. It should be noted that for singlet-s inglet energy transfer, both interactions may be involved, but in general coulomb mechanism predominates. The interaction energy describing the coupling between the initial and final states is given by: ifUH (1-3) where H’ contains the electrostatic interactions of all electrons and i and f are the electronic wavefunctions for the initial a nd the final excited state, respectively. Considering that only two electrons are involved in a transition, the antisymmetrized product wave functions of the initial and final state can be written as:

PAGE 50

34 ** **1 (1)(2)(2)(1) 2 1 (1)(2)(2)(1) 2iAA DD fDD AA (1-4) where the numbers 1 and 2 refer to the two electrons involved. The total interaction U can be written as the sum of the Coulomb and the exchange terms: ****'11 (1)(2)(2)(1)(1)(2)(2)(1) 22AADD DDAAUH (1-5) ** **' '(1)(2)(1)(2) (1)(2)(2)(1)CAD DA exAD DAUH UH (1-6) CexUUU (1-7) The Coulomb term,CU represents the classical interaction of the charge distributions and may be expanded into multip le terms: dipole-dipole, dipole-quadruple, etc.,Dipole-dipole inte raction dominates for allowed transitions: 3 01 4 D A Cdd DAUU R (1-8) where D and A denote the transition dipole mome nts of the two molecules (D* D and AA*) and RDA is the distance between the donor and the acceptor. Here, the orientation factor is defined by: .3(.)(.) 2coscossinsincosDADA DADA DADARR (1-9) The vectors, angles and separation betw een dipoles are defined in Figure 1-12. When Udd is expressed in cm-1, the transition moments are in Debye and RDA is in nm. is the angle between tw o transition moments and D and A are the angles between each

PAGE 51

35 transition moment and the vector connecting them. Considering this picture, it is clear that no interaction would be observed between perpendicularly oriented chromophores. Figure 1-12. Definition of the angles used to calculate the orientati on factor between the dipoles.99 The dipole-dipole approximation provides re liable estimation of the electronic coupling between point dipoles, i.e. when the donor acceptor separation is much larger than the molecular size of the donor and acceptor transition dipole moments. At short distances or when the separation is comparab le to molecular dimensions (large dipole moments), the point dipole appr oximation is not valid, and a better description of the shape of the dipoles should also be include d in the calculations. The transition moment magnitude (in units of Debye) is related to the dipole strength of an absorption band, measured in media of refractive index n, according to: 9 24.310 () x f vdv n (1-10) Thus, the Coulomb interaction te rm can be related to experimentally measured quantities. The exchange interaction is a purely quantum mechanical phenomenon and does not depend on the oscillator strengths of the transitions involved. The exchange integral **2 12(1)(2)(2)(1)exAD DAe U r (1-11)

PAGE 52

36 represents the interaction of charge densities separated by a distance r12 Since charge densities depend on the spatial ove rlap of orbitals of D and A, the exchange interaction decreases exponentially with incr easing internuclear distances. The non-radiative transfer rate is ba sically given by Fermi’s Golden Rule. According to the Golden Rule, the rate of tr ansition between two states is related to the magnitude of a perturbation which changes the positions or motions of particles of the initial state and reshapes the initial state so that it looks like the final state. Beyond the Born-Oppenheimer approximation, it is neces sary to include the interaction between different vibrational-el ectronic molecular states in order to describe such transitions. Using the time-dependent perturbati on theory, the rate constant kET is formulated as: 2 2'22ETifkVH (1-12) where is the density of interacting initial a nd final states as determined by FranckCondon factors, and it is rela ted to the spectral overlap be tween the emission of the donor and the absorption of the acceptor in a system without inhomogeneous broadening. Using the relations for the interaction energy given above, Frster and Dexter derived expressions for the rate constant of energy transfer using the Coul omb and the exchange mechanism, respectively.100,101 Dexter’s formulation points out that an exponential depe ndence is expected from the exchange mechanism. The rate cons tant for transfer can be written as: 122 exp(2/)ETkKJrL h (1-13) where J is a spectral overlap in tegral with the normalization condition, L is the average Bohr radius and K is related to specific orbital interactions, not related to any

PAGE 53

37 spectroscopic data. Thus, it is difficult to characterize the exchange mechanism experimentally. While the nature of the interaction, whether Coulombic or exchange, is an important task to investigate, the magnitude of the interaction also needs to be explored. Frster proposed to discriminate between very weak, intermediate (weak), and strong electronic coupling depending on th e relative values of the interaction energy (V), which is the pure electronic energy difference between D* and A*( E), the absorption bandwidth ( w), and the vibronic bandwidth ( ) (Figure 1-13). Strong coupling. In this case, the intermolecular interaction, Vc, is much larger than the width of the individual transitions DD* and AA*. Then, all the vibronic subtransitions in both molecule s are virtually at resonance wi th one another. The transfer of excitation energy is faster than the nucle ar vibrations and vibr ational relaxation. The absorption spectra of strongly coupled systems will be different from those of the individual components. The donor and accepto r electronic states will mix to produce Figure 1-13: Differences between strong, weak, and very weak coupling. D* A* E STRONG COUPLING V >> E V >> w WEAK COUPLING V >> E w >> V >> VERY WEAK COUPLING V << << w V: interaction energy w

PAGE 54

38 new, delocalized states. Thus, the transfer of excitation is a coherent process and the excitation oscillates back and forth between D* and A*. The rate constant is derived as: 4ETV k h (1-14) where V is approximated by dipole-dipole intera ction. The distance dependence of V, and consequently ofETk is r-3. Intermediate (Weak) coupling. This is a particularly challenging case to model RET dynamics. The interaction energy, V, is larg er than the width of an isolated vibronic level but much smaller than the full abso rption bandwidth. Compared to the strong coupling case, the electronic excitation is more localized. On the other hand, the vibronic excitation is still delocalized and the system can be described in terms of stationary vibronic exciton states. The transf er rate is fast compared to vibrational relaxation .It is approximated as: 24vw ETVS k h (1-15) wherevwS is the vibrational overlap integral of the intramolecular transition vw Since vwS <1, the transfer rate would be slower than in the case of strong coupling. Very weak coupling. The interaction energy is much lower than vibronic and absorption bandwidth. The vibrational relaxation occurs before the transfer occurs. The absorption spectra of the components are not altered. The transfer rate is given by: 224()vw ETVS k h (1-16) The characteristic feature of this very weak coupling case is the quadratic dependence of the transfer ra te on the interaction energy, as opposed to the linear

PAGE 55

39 dependence found for the intermediate and stro ng coupling. For dipol e-dipole interaction, the distance dependence is r-6, whereas it is r-3 for the preceding cases. Frster theory for RET is formulated for the very weak coupling limit. It is based on an equilibrium Fermi Golden Rule approach with a second-order perturbation theory treatment of the electroni c coupling between donor and acceptor. The theory was established for coupling of a st ate to a quasi continuum of se condary states. The primary, discrete states are non-stationary and they carry all the oscillator strength. Such model allows rationalizing dynamic processes invol ving decay of a “s tationary” state for radiationless transitions. In other words, th e transfer rate occurs after vibrational relaxation. In this way, Frster model eval uates a Fermi Golden Rule expression for the RET rate, where the matrix element of in teraction between excited state donor and ground state acceptor is purely electronic coupl ing V. Conditions of energy conservation and nuclear overlap factors, separated from the electronic coupling, relate the donor emission and acceptor absorption events: 2 02 ()()(,,)kk ETda klkdPkPlu (1-17) where d k is the energy gap of the donor molecule P(k) is thermal population of mode k in the excited state, and a k is the energy of the acceptor gr ound state. The matrix element of interaction between the excited state donor and ground state acceptor is independent of energy and can be written as: 2 2()()(,,()kk da kluPkPluVJ (1-18) where() J is the spectral overlap between donor emission f( ) and acceptor absorption a( ), and can be written as() J = f( ).a( ) f( ) and a( ) have each been normalized to

PAGE 56

40 unit area on an energy scale: ()()1fdad .The rate constant can be rewritten as: 2 2 2 0021 ()()ETkVdJVdvJv c (1-19) where Vis expressed in units of cm-1, and /2 vc In summary, estimating a rate for RET in the weak coupling limit requires the kn owledge of the electronic coupling, V, and the spectral overlap, J, between donor and acc eptor transitions. The final rate equation is defined as: 2 2 24ETJ kV hc (1-20) It is important to note that J in this equation is defined as 2()()() Jcmfad and has units of cm. By substituting the Coulomb interaction energy term, VC, into the Golden Rule equation, and assuming a dipol e-dipole approximation, Frster derived the final expression for energy transfer rates based on spectroscopically measurable parameters. The Frster equation is: 2 5469000ln(10) 128 D DA ET A DADJ k nNR (1-21) where 2is the orientation factor of the transition dipole moments, D is donor quantum yield, D is the donor lifetime in th e absence of the acceptor, A N is Avogadro’s number, and n is the refractive index of the solvent. Note that in this equation spectral overlap J is defined as:

PAGE 57

41 4 4 00()() ()() ()DA DADAfvv Jfddv v (1-22) ()Df is the fluorescence spectrum of the donor normalized so that 0()1Dfd and ()A is the molar absorption coefficient of the acceptor. Hence, JDA has units of M-1cm3. At a specific distance of RDA, the rate at which D* emits light is equal to the rate at which it transfers its excitation energy to A. At this critical distance R0, called Frster radius, one can write 1ET D k then inserting this into rate equation, and solving for R0 yields: 2 6 0 549000ln(10) 128D D A A R J nN (1-23) R0 is the donor-acceptor distance at which the probability for energy tr ansfer is equal to 0.5. The energy transfer rate can al so be written in the following form: 6 01ET DR k R (1-24) The transfer efficiency is defined as: 1/ETET ET D ETDETkk kkk (1-25) Using the preceding equation, the transfer e fficiency can be related to the ratio R/R0 by: 6 01 1(/)ETRR (1-26) Note that the transfer efficiency is 50% wh en the donor acceptor distance is equal to the Frster critical radius.

PAGE 58

42 The Frster equation for RET is accura te provided that four conditions are satisfied: (i) a dipole-dipole approximation for the electron ic coupling can be utilized appropriately for the donor-acceptor interact ion, (ii) the donor fl uorescence lifetime, emission line shape, acceptor absorption line shape, and oscillator strength are not perturbed because of interactions among donors and acceptors (weak coupling), (iii) inhomogeneous line broadening is absent in the donor and acceptor line shapes, (iv) the energy transfer dynamics is incoherent. In this dissertation, energy transfer pro cesses in various PE dendrimers are studied. In addition to characterizing a kinetic mode l for each molecule, the processes according to the strength of D-A coupling will be clas sified and the validity of previously used Frster approximations is tested. Outline of the Dissertation The main scope of this work is to investigate energy transfer processes in conjugated, symmetrical and unsymmetrical phenyl ethynylene (PE) dendrimers. We have conducted experimental studies to impr ove the understanding of electronic structure of these molecules and its effect on the light -harvesting properties. Chapter 1 is an introduction to dendrimers and a survey of the current scientific literature regarding dendrimer photophysics. This Chapter also includes a brief introduction to energy transfer theory. Chapter 2 summarizes the experimental me thods utilized to study timeresolved emission and absorption characterist ics of PE dendrimers. This chapter is complemented with Appendix A, a brief desc ription for experimental beginners on how to successfully perform these novel experiments.

PAGE 59

43 Unsymmetrical generation one didendrons are investigat ed in Chapter 3. The excited state dynamics is studied using both time-resolved and steady state spectroscopy. In an attempt to more quantitatively anal yze the results, a kinetic model is proposed. Chapter 4 describes the excited state dyna mics of generation two didendrons and we perform a comparative analysis by transi ent absorption and time-resolved emission spectroscopy. To explain the multicomponent ri se and decays of the excited states, the basic kinetic model (described in Chapter 3) is extended to a more complex level. Apart from the initially excited state, the pres ence of a second state is verified via low temperature absorption and excitation anisot ropy measurements. This chapter further evaluates the effect of gene ration on transfer rates. In Chapter 5, we study a symmetrical PE dendrimer, namely “the nanostar” with very detailed experiments and global analysis In contrast to unsymmetrical dendrimers, this molecule can be considered as a co mbination of individual chromophores with variable conjugation length. As referen ce compounds, 2-,3-, and 4-ring phenyl ethynylene chromophores are studied with tran sient absorption spectro scopy. Particular emphasis is given to recent developments in theories and the proposed kinetic model that offer physical understanding of the energy transfer. Chapter 6 describes an independent project that we collaborated with Dr. Schanze’s group at UF. The dynamics of fluorescence que nching of a conjugate d polyelectrolyte by a cyanine dye are investigated by fluores cence up-conversion and pol arization resolved transient absorption. The data are analyzed w ith a model based on the random walk of the exciton within the polymer chain and a l ong-range direct energy transfer between polymer and dye.

PAGE 60

44 Appendix A is provided to supplement expe rimental details and tricks for the upconversion technique. Appendix B gives a br ief introduction to the global analysis method. The Singular Value Decomposition analys is combined with the kinetic model for each molecule is given in detail.

PAGE 61

45 CHAPTER 2 EXPERIMENTAL METHODS The light harvesting properties of dendrim ers lead to the broad investigation of energy transfer processes. It becomes important to evaluate the electronic structure of dendrimers and its effect on energy transfer mechanisms and associated dynamics. This chapter provides background information on the experimental methods employed to explore excited state dynamics of the dendrimers studied thro ughout this dissertation. An overview was given about the relaxation proces ses occurring after the photoexcitation of molecules in chapter 1. Af ter photoexcitation, excited mo lecules in solution undergo various relaxation processes, wh ich can be classified in four major categories: electronic, orientational, vibrational rela xation, and solvent relaxation.102 Here, we are mainly interested in electronic relaxation processes su ch as transferring the excitation energy into a specific trap. Steady state absorption and fluorescence spectroscopy give some insights into the spectral composition of the dendrim ers and energy transfer pathways. For better understanding of the energy tran sfer processes, time reso lved techniques, such as transient absorption and time resolved fluores cence, are extensively used in this work. By means of presented experimental tec hniques, important “u ltrafast” phenomena such as energy transfer and na ture of excitations in conjuga ted systems are studied within this dissertation. Chemicals and Materials Throughout this dissertation, we i nvestigate unsymmetrical conjugated pheneylethynylene (PE) monoand di-dendrons and a unique symmetrical PE structure

PAGE 62

46 named the ”nanostar”. The detailed synthesis and structural characterizations of such conjugated PE dendrimers ar e reported in literature.69,70,81,82,103 These dendrimers are supplied to us by our collabo rators, Prof. Zhonghua Peng from the University of Missouri-Kansas City and Prof. Jeffrey S. Moore from the University of Illinois at Urbana-Champaign. They confirmed the struct ure and purity of our samples by thin-layer chromatography, elemental analysis, 1H and 13C NMR spectroscopy, and matrix-assisted laser desorption/ionization time-of-fli ght (MALDI-TOF) mass spectroscopy.104 These samples are used as received. For stea dy state and time-resolved spectroscopic measurements, the dendrimer samples ar e dissolved in dichloromethane (CH2Cl2). The solvent is purchased from Aldrich and th e purity was UV-spectroscopy grade (>99.9%). It is kept under nitrogen and used without a ny further purification. In order to perform steady state measurements at lo w temperatures, a liquid nitroge n flow cryostat is used to control the temperature in the range from 77 K to 298K. Methylterahydrofurane (MeTHF), purchased from Aldrich, is used as a solvent. To obtain a glassy sample, MeTHF is further purified and distilled to be anhydrous prior to each measurement. 1,2 diphenylacetylene (2-ring, DPA) is pur chased from Aldrich and used as received. 1,4 bis(phenylethynyl) benzene (3-ring, para) was also purchased from Aldrich, but it is purified by recrystallization from tolu ene, yielding analytically pure material as determined by elemental analysis performe d by Joseph Melinger at Naval Research Labarotories (NRL).105 4,4’-bis(phenylethynyl)-tolane is synthesized by Prof. Andrew Beeby’s group at the Department of Chemistr y, University of Durham. The sample is used as received. 1,3 (bisphenylethylnyl) benz ene (3-ring meta) is synthesized in house according to procedures reported elsewhere.35

PAGE 63

47 The conjugated polyelectrolyte poly(phe nylene ethynylene) sulfonate (PPESO3) and the cationic dye molecule (HMIDC) used in the work presented in Chapter 6 are obtained from the group of Prof. Kirk S. Schanze at the University of Florida. The synthesis of PPESO3 has been described recently. 106,107 The average molecular weight, Mn, of the polymer is estimated to be 100 kDa, corresponding to about 200 monomer units. HMIDC is purchased from Aldrich and us ed as received. Solutions with different HMIDC concentrations are prepared and labeled according to their steady state quenching efficiency. Steady State Measurements Steady state absorption spectra of the sample s is recorded with a UV-VIS VarianCary 100 spectrometer. The wavelength range detected is from 190 to 2200 nm with 1 nm spectral resolution. The steady state emi ssion spectra are measured with a Jobin-Yvon instrument (Spex Fluorolog-3) as a functi on of wavelength. For room temperature spectroscopic measurements, all samples are di ssolved in dichloromethane. The optical density (OD) of the samples is approximately 0.2 -0.3 mm-1 at the absorption maximum. Why Time-Resolved Spectroscopy? Time-resolved spectroscopy is defined as “a ny technique that allows to measure the temporal dynamics and the kineti cs of photophysical processes”.108 The development of ultrafast lasers and pulse sh aping techniques, among other in novations, have opened up a wide range of investigations of comple x systems in chemistry, physics and biology. After interacting with a short light pulse (f rom milliseconds to femtoseconds), the sample under investigation will change spectroscopic pr operties such as energy, polarization, or phase. The fate of the ground and excited stat es of the system can be determined by investigating energy and charge transfer processes, coupling of electronic and vibrational

PAGE 64

48 degrees of freedom, vibrational and conformati onal relaxation, isomerization, etc. In the case of light harvesting systems, such as the dendrimers studied throughout this dissertation, ultrafast spectroscopy is used to study energy transfer processes. These processes often take place on a (sub)pico second time scale, which puts strict requirements for the light source to be used in these experiments. This source should provide very short light puls es with appropriate wavelengths within the electronic transitions of the system, and with suitable power and reasonable repetition rate. Mode-locked Ti-Sapphire laser systems, su ch as the oscillator called Tsunami, produce less than 40 fs pulses with a very good stability. However, the pulse energies (nJ regime) is not enough for many spectroscopic measurements and the high repetition rate of 80 MHz is too fast for transient absorption techniques. In our lab, the amplification of these pulses are achieved by means of Ti-Sa Regen Amplifier (Spitfire) pumped with a Q-switched Nd:YLF laser (Evolution X). Theref ore, it is possible to obtain pulses up to 0.90 mJ with 1kHz repetition rate. The output wavelength of the Regen Amplifier is centered at 790 nm, which is not suitable for many natural a nd synthetic light-harvesting systems. This problem is solved by employi ng OPAs combined with second and fourth harmonic and sum frequency techniques, which are tunable within a broad spectral region, from UV to IR (300-5000 nm).109 The Laser System To perform time-resolved experiments with femtosecond time resolution, short and intensive laser pulses with variable photon energies are required. In addition to a commercial laser system, nonlinear elements are used for our applications. Figure 2-1 shows a diagram of the laser system. The f unction of individual components will be discussed briefly.

PAGE 65

49 Figure 2-1. The laser system for the producti on of tunable femtosecond laser pulses with high energy per pulse. (1) Millenia Vs: The Spectra-Physics Millenia Vs uses the output from a diode laser to pump Nd+3 ions doped in yttrium vanadate crystalline matrix (Nd:YVO4). An LBO crystal converts the 1064 nm light from th e laser crystal to the green light, 532 nm, which becomes the output of the laser. Millenia Vs is an all solid-state CW laser, which offers near diffraction limited TEM00 beam quality and ultra-low noise with an output power range from 2W to 10 W. Millenia’s output pumps the mode-locked Ti-Sa Oscillator. (2) Ti-Sa Oscillator Tsunami : This laser, with a titanium sapphire crystal as the laser medium-called Tsunami from Spectra Physics, provid es very short (35 fs), but relatively weak laser pulses w ith 80 MHz repetition rate. The output spectrum is peaked at 790 nm and has ~45 nm bandwidth (FWH M). This output is the seed of our regenerative amplifier. (3) Evolution X: Evolution is a diode pumped (by four AlGaAs laser diode arrays), intracavity doubled Nd:YLF laser capable of pr oducing Q-switched pulses with Ti-Sa (2) Oscillator Ti-Sa (4) Regen Amplifier Nd:YLF Laser ( 3 ) OPA 1 (5) OPA 2 (5) = 800 nm FWHM=35 fs 80 MHz 10 nJ = 800 nm FWHM=50 fs 1 kHz 1 mJ Pulses = 300-900 nm FWHM= 100-200 fs Nd:YVO4 Laser ( 1 )

PAGE 66

50 average powers > 6 W at 527 nm. The laser reso nator is acousto-optically Q-switched at repetition rates of 1 kHz. It offers high e fficiency, low maintenance, and excellent beam quality. It ideally pumps Ti-S apphire ultrafast amplifiers, and has been optimized as a pump source for the Spitfire regenerative amplifier system. (4) Regenerative Amplifier Spitfire: Ti-Sa crystal is the active laser medium, which is optically pumped by an external laser (Evolution) and uses chirp pulse amplification to generate high intensity la ser pulses centered at 790 nm. The repetition rate is set to 1 kHz. The seed pulse coming from the os cillator is first stretched temporally using a grating scheme and then inserted into a cavity using a pockels cell. The laser cavity is built in a Z form sche me. After many round trips, this pulse is amplified and released from the cavity usi ng a second pockels cell and a thin film polarizer. The overall amplif ication is about 3.3x106 yielding a power of 1.25 W at this point. Finally, the pulses are compressed in a similar grating arrangement to the stretcher and routinely produce pulses of ~0.85 mJ cen tered at 790 nm with pulse widths around 50 fs (FWHM). (5) Optical Parametric Amplifier (OPA): The optical parametric amplifier system, OPA-800C, offers broad wavelength coverage from UV to mid IR with near transform limited output pulses and high pulse energies. Th e Spitfire amplifier output is split into two beams (50 %) and used as pumps for two independent OPA sy stems, providing two highly stable, inherently sync hronized outputs with independen t wavelength control. By proper selection of signal or idler beam, polarization direction, phase matching angle/type, and number of harmonic cr ystals, a range from 300 nm up to 5 m can be

PAGE 67

51 covered completely. To obtain this wide range of wavelengths, harmonic generation of the OPA signal or idler is used. Within the capabilities of th e laser system described above, two main time-resolved experimental setups have been developed in our labs to study the excited state dynamics of light-harvesting dendrimers. The first one, Fluorescence Upconversion, is a setup that I designed and built. The second one, Transien t Absorption (Pump-Probe), uses a white light super continuum as the probe. Both of these techniques and the crucial components of the setup will be explained in detail. Ultrafast Time-Resolved Emission Spectroscopy The energy transfer processes within the conjugated dendrimers are investigated with time-resolved emission spectroscopy. For the initial studies we actually used a timecorrelated-single photon counting (TCSPC) in strument, which was available in the Schanze lab (specifically for the molecules st udied in Chapter 3 and 4) and the lifetime measurements for the nanostar with TCSPC were already done by Swallen et al.70This conventional method is widely used for the determination of lifetimes. Time-Correlated Single Photon Counting When an ensemble of fluorophores is excite d with a very short optical pulse, this results in an initial occupa tion of the excited state by N0 fluorophores. The population of the excited state will decay radiatively a nd/or nonradiatively to the ground state according to the following equation: () ()()rnrdNt kkNt dt (2-1) where kr and knr are the radiative and nonradiative d ecay rates, respectively. The decay of the excited state population is e xponential which can be written as:

PAGE 68

52 00()exp(())exp(/)rnrfluoItItkkIt (2-2) where fluo is the relaxation time of the excited state. Even though this equation shows only monoexponential decay, for complex systems, the fluorescence decay becomes multior nonexponential. The basic princi ple of TCSPC experiment is that the probability of detecting a singl e photon at time t after pulse excitation is proportional to the fluorescence intensity at that time.110The time lag between th e excitation pulse and the detected single photon is measured and the decay histogram is reconstructed from individual time lag measurements. Upon arrivi ng a detector (e.g. PMT), a multichannel plate, or an avalanche photodiode, each emitted photon creates a reference electrical pulse that is fed to a constant fraction discriminator that triggers a time-to-amplitude converter (TAC). Meanwhile, the excited samp le emits and when the detector sees the first photon from the sample, it feeds a stop pul se to the TAC. The TAC consists of a highly linear ramp voltage generator that is started by one signal a nd stopped by other, and delivers an output voltage whose amplitude is directly proportional to the time difference between the two signals. This TAC signal is then analyzed by an analogue-todigital converter and one count is stored in a multichannel analyzer (MCA) for each voltage. Excitation and detection events are re peated in this way until the histogram of the number of “counts” against each time window is large enough to give a reliable decay curve of emission. It is important to note th at the emitted fluorescence intensity should be low enough that the probability of detecting one photon per excitati on cycle is less than unity. The main advantages of TCSPC method are its high sensitivity and outstanding dynamic range (signal to noise: 10000/1). Ho wever, due to the electronics and the

PAGE 69

53 detector, the best time resolution of the in strument is low (about 50 ps). Fairly long lifetimes, up to milliseconds, can be measured. This method was initially used to have a general idea for the emission decay rates of the dendrimers that are mainly in the nanosecond time-regime. The energy transfer process in PE dendrimers occurs in the subpicosecond time scale and therefore requires an experiment with a much better time resolution. This can be achieved with th e recently developed Fl uorescence Up-conversion technique. Fluorescence Upconversion Technique The fluorescence up-conversion technique is used to measure time resolved emission dynamics with a time resolution of tens to hundreds of femtoseconds. This method was first applied by Mahr and Hirsch111 and it is based on sum frequency generated by the temporal and spatial overl ap of the incoherent fluorescence and an ultrafast gate pulse on a nonlinear crystal.112 (It is also possible to generate a difference frequency, called down-conversion). This sum frequency is detected as a function of the time delay between the gate pulse and excita tion pulse which induces fluorescence from the sample. The up-conversion technique allo ws the mapping of the temporal evolution of the fluorescence. The up-conversion signal has a photon frequency given by: s umgatefluo (2-3) implying, 111 s umgatefluo (2-4)

PAGE 70

54 As illustrated in Figure 2-2a, when the gate pulse and the emission are overlapped in the nonlinear crystal, frequency mixing o ccurs creating an up-c onverted signal. The up-converted frequency is determined both by the angle between the optical axis of the crystal and the incoming beams, and by the optical frequencies of these beams. The nonlinear crystal behaves as an optical gate which is opened when the gate pulse is present in the crystal. Scanning the delay of the gate pulse relative to the excitation pulse opens this optical gate in different portions of time and the fluorescence signal is mapped out at these different tim e delays (Figure 2-2b). The intensity of the sum-frequency signa l is given by the convolution of the fluorescence intensity and the gate pulse intensity: ()()()sumfluogateIItItdt (2-5) where is the time delay between the gate beam and the fluorescence of the sample. The advantage of using this optical gating technique is that the time resolution is determined by the width of the pulses (pump and gate pulses), not by the time resolution of the detection system.113 The time resolution of the upconversion experiment is determined by the instrument response function (IRF), which is pr oportional to the cros s correlation of the excitation pulse with the gate pulse. Operatio nally, the IRF is measured by angle-tuning the crystal to up-convert transmitted or scat tered pump light. For very short pulses <100 fs (FWHM), crystals much thinner than 1 mm are required. The sum frequency is generated throughout the thickne ss of the crystal as long as the gate pulse and

PAGE 71

55 Figure 2-2. Fluorescence Up-Conversion Tech nique (a) Illustration of the upconversion principle (b) Up-converted fluoresce nce signal generated in a nonlinear crystal only while the delayed gate pulse is present fluorescence are temporally and spatially ove rlapped. However, the group velocity mismatch between the fluorescence and the gate wavelengths can cause broadening of the IRF which needs to be accounted for the de termination of the time resolution of the experiment. The bandwidth of the up-convert ed signal depends on crystal properties and on how tightly the gate pulse and fluorescen ce are focused on the crystal. The optical layout for our upconversion setup is explai ned in detail in the next section. Homemade Upconversion Apparatus The experimental setup is shown schematica lly in Figure 2-3. In this section, we review the components together before they are separately discussed in detail. The Ti:sapphire Regenerative amplifier system provides 50 fs, 840 J pulses at 790 nm with 1 kHz repetition rate. This beam is sp lit into two with 1:1 ratio. Each 420 J beam is independently used to pump an optical parametric amplifie r (OPA 800C, Spectra Physics). The first OPA delivers the pump pulses in the UV and visible region through fourth harmonic of signal and idler, respectively. The OPA ou tput is sent through a prism gate flu sum Luminescence Gate Pulse Up-converted signal Non-linea r Crystal Excitation pulse Luminescence Gate pulse Up-converted signal (a) (b)

PAGE 72

56 compressor to get the shortest pulse possibl e, as required to i nvestigate very fast relaxation dynamics. The polariz ation is controlled with a /2 waveplate, and the beam is focused onto a sample with a lens (f=200 mm, fused silica). The re sidual of the 790 nm pumping the OPA is used as the gate pulse. It passes a delay stage and finally relayed on the nonlinear crystal by a lens (f=300 mm). This gate pulse is not as short as the fundamental pulse since it goe s through all the op tics in the OPA. The autocorrelation measurements proved that the OPA typically delivers a 120 fs gate pulse at 800 nm and its bandwidth is narrower than the fundamental 790 nm beam. The sample is held in a 1mm rotating quartz cell (home made, 1 mm windows) to ensure sample photostability. Th e polarization plane of the exc itation light is set to magic angle with respect to that of the gating pul se in order to examine the population dynamics without the influence of rotational diffusion of the solute molecules on the decay of fluorescence. A pair of off-axis parabolic mirro rs (A8037-207 Aluminum Uncoated Mirror, Janos Technology) collects th e fluorescence and focuses the fluorescence into a nonlinear crystal. A negative focal lengt h lens is used to magnify the fluorescence image in the nonlinear crystal. A type I phase-matching BBO crystal (0.3 mm) is chosen for the wavelength region studied here. The generated sum frequency light is then collimated and focused into the entrance slit of a 250 mm monochromator (S pectraMini). A UG11 UV cutoff filter placed in front of the m onochromator minimizes the 400 nm generated on the crystal by second harmonic generation of the gate pulse. A UV sensitive photomultiplier tube (R7154, Hamamatsu) detect s the signal. This electrical signal is gated by a boxcar averager SR 250, Stanford Research Systems. A personal computer is

PAGE 73

57 connected to the detection system and the tran slational stage to cont rol the experiment. A Labview program was written to control the tran slation stage and da ta acquisition card. Figure 2-3. Fluorescence upconve rsion experimental setup. The upconversion technique is relatively si mple in principle, and has been used widely over the past decade. However, it ha s some crucial components, which happens to be also the most difficult to align. In A ppendix A, I will explain the details of the experiment for the new users. These are mainly suggestions from an experienced graduate student who spent a lot of time designing and optimizing the upconversion setup. BBO PMT exc SAMPLE g ate=800 nm

PAGE 74

58 Ultrafast Transient Absorption Spectroscopy The excited state dynamics of PE dendrim ers are also investigated by ultrafast transient absorption (pump-probe) experiments. The principle of the transient absorption experiments is rather simple. At least two ultrashort laser pulses are needed. The intensive one, “pump” pulse, perturbs the sample at t=0. The probe pulse, which is delayed with respect to pump pulse, crosses the perturbed part of the sample and will probe the action of the pump pulse on the sa mple. The perturbation created by the pump and monitored by probe pulse can be analyzed in two ways: The modifications of the probe pulse characteris tics (intensity, phase, etc.) after passing through the sample can be compared before and after the action of pump pulse. This measurement is then called the transient absorption technique. On the other hand, it is quit e possible to observe the new effects created by the probe pulse itself before and af ter the pump pulse. Raman spectroscopy, Coherent Anti-Stokes Rama n Spectroscopy (CARS), and laser-induced fluorescence are such experiments.108 Here, the changes in the absorption spectru m of a sample after being perturbed by an ultrashort pulse will be observed and measured. The ab sorption of the sample may increase or decrease or new absorption ba nds corresponding to new transitions appearing under perturbation may evolve. By changing th e time delay between the pump and probe beams, it would be possible to do temporal analysis of these changes. For example, the simplest photophysical event when a molecule is interacting with a light pulse is the excitation of molecule from its ground electroni c state to its first excited electronic state, followed by the return of the molecules to the ground state by fluorescence and/or internal conversion. The return of the molecules to the ground state can be monitored by

PAGE 75

59 the change in transmission of a weak probe pulse through the sample as a function of delay between the pump and probe pulses. The principle of this method is shown in Figure 2-4. At time t=0, the pump pulse excites the sample. At time t+ t, the probe pulse passes through the perturbed volume of the sample ( t is tunable by means of an optical dela y line). The probe pulse intensity is measured. The spectral distribution of the probe is also recorded to improve the sensitivity of the measurement simultaneously in the presence and in the absence of the perturbation of the sample at each laser shot.108 As shown in Figure 2-4, the probe beam is split into two equal beams; while the first partial beam crosses the pumped part of the sample, the other beam crosses the unpert urbed (not pumped) part of the sample. Figure 2-4. Basic principle of transient absorption experiment These two beams are detected by a CCD. By doi ng so, shot to shot fluctuations of the laser power are compensated for. Experimentally, the efficiency of the li ght absorption at a wa velength by a medium is characterized by the absorption or the transmission defined as: C C D Sample I t I0 Probe Pulse Pump Pulse R = 50%

PAGE 76

60 0 0() ()loglog() () () () () I AT I I T I (2-6) The detector then measures the probe pulse intensity 0() I (no perturbation by pump) and (,) I t (sample perturbed by pump). Consid ering the Beer-Lambert law, one can write: () 0(,)()10NtlItIx (2-7) Where is the absorption coefficient of the sample at wavelength N( t) is the population absorbing at time t at wavelength and l is the length of the sample excited. In fact, the optical density of the sample is measured: 0() (,)log() (,) I ODtNtl It (2-8) The detected signal measured in the transi ent absorption experiments is actually the change in absorption (transmi ssion). The detector measures the intensity of the probe beam in the presence and absence of th e pump excitation as a function of time. ,, 00 , 01tpumptnopump pumpnopumptpump tnopump nopumptnopumpII TTI II T I TTI I (2-9) Then, the change in the ab sorption is defined as: log(1) T A T (2-10) After excitation of the sample with appropriate wavelength, it is possible to follow the population dynamics at a given wavelength (single color probe is enough for such

PAGE 77

61 measurement) by varying time. For a complex system, it is highly desirable to monitor the entire transient spectrum at any time de lay after the excitation. Measurement of the full transient spectrum is very helpful in assignment of different absorbing units (species). For the interpretation of the transient spect ra and their relative sign and values, it must be recalled that ther e are three origins for the pump-probe signal: ground state bleaching, stimulated emission, and excited st ate absorption. When the sample is pumped within its absorption spectrum, a certain num ber of molecules will be excited (Figure 25). During the probing process if there is popula tion in the excited state, the transmission of the sample increases and ground state bleaching is observed (negative absorption signal). Stimulated emission occurs when th e probe beam stimulates the excited state molecules to return to th e ground state. The detector will see more photons in the emission range of the sample, so the tran smission increases resulting in a negative A signal. Note that the probability of the stim ulated absorption is the same as of the Figure 2-5. The theoretical scheme of certa in signals observed as transient absorption signals. S1 S0 .. t.. Excitation Stimulated Emission A < 0 Photoinduced Absorption A > 0 Sn h 1 h 2 h 2Ground State Bleach A < 0

PAGE 78

62 stimulated emission at the emission wavelengt hs. Excited state abso rption (photoinduced absorption) takes place when the excited mol ecules are excited to even higher electronic states by the probe pulse. This will lead to positive A signals. It is hi ghly probable that these three different signals will spectrally overlap complicating the interpretation of the overall signal. Probe Characteristics and White Light Continuum Generation The appropriate excitation (pump) wave length, which depends on the absorption spectrum of the investigated system, can be generated by an OPA. The probe beam can be monochromatic or it can have a broad sp ectrum. In the simple st case, namely onecolor (degenerate) experiments, the pump and probe pulses are split from the same initial beam and one of them is delayed with respec t to the other. For two-color experiments another light source is needed to generate the required wavelength of the probe pulse. Prior to choosing the probe pulse wavelength, some preliminary studies, such as steady state spectroscopy, should have been used to characterize the sample. The probe wavelength should be in a spectral domain wher e it is expected that some species created by the excitation will be present in the sample at a certain time t, and will have resonant electronic transitions for that specific pr obe wavelength. However, it is highly informative to use a probe spectrally as broa d as possible. The whole transient absorption spectrum can be measured if a multicolor probe, such as white light continuum, is used. Measuring the whole transient spectrum as a function of time will help determine the dynamics of the excited states of the system.114-118 For the complex systems studied here, it was preferred to use a white light continuum as the probe pulse. How we generated this continuum will be explained briefly, but first the experimental setup will be explained.

PAGE 79

63 Experimental Setup The experimental transient absorption setup for probing with a white light continuum is presented in Figure 2-6. The Ti:sapphire Regenerative amplifier system provides 50 fs, 840 J pulses at 800 nm with 1 kHz repetitio n rate. This beam is split into two with 1:1 ratio. Both 420 J beams are used to pump an optical parametric amplifier (OPA 800C, Spectra Physics). The second OPA is used to deliver the pump pulses for transient absorption experiment. The OPA is tunable in the UV and visible region (300900 nm). Since the dendrimer systems under i nvestigation absorb in the UV region (<400 nm), the OPA was aligned to generate UV pul ses. A double-pass prism pair is used to compress these pulses. A motorized transl ation stage (Model No:M-415 D6, Physik Intrumente) is used to vary the time delay between the pump and probe pulse up to 1 ns. A chopper wheel is used to chop the pump be am with 10 Hz frequency in order to compare signal with and without pump. Figure 2-6. Experimental set up for transient absorption e xperiment probing with white light continuum. UV pump pulses are obtained from the OPA. A chopper wheel used to compare the signal with and without pump. The white light continuum is generated in CaF2 window. OPA Spectrograph C C D chopper Delay line CaF2 plate /2 Polarizer Sample Spitfire 800 nm probe reference pump Prism compressor

PAGE 80

64 Continuum generation When a high peak power short pulse is focu sed in a (transparent ) medium such as glass, water, or sapphire, a continuum of light which may even appear as white light, is generated. The origin of th is process is mainly governed by self-phase-modulation and stimulated Raman emission. The directionality of the white-light pulse makes it possible to use it as a spectrally br oad probe and measure the tran sient absorption at different wavelengths simultaneously. A small fraction, 3-4 J, of the fundamental output of the amplified laser pulse is converted to white light by fo cusing it into a 1mm thick CaF2 window (1” diameter, PW1004-CFUV from CVI-laser). The CaF2 is held in a home-made rotating stage to avoid shot to shot intensity fluctuations, temperat ure effects. A 10 mm lens was used, and the size of the focus was about 200 m. The spectrum needed as a probe for the measurements on dendrimers lies in the re gion of 300-600 nm (abs orption and emission of the system). Figure 2-7, the top curve, shows the spectrum of the white light continuum directly (without the optics necessary for the experi ment) sent to the detector (CCD). As shown in Figure 2-7, the probe and reference beams have very similar spectrum. The quantity of photons (signal on th e detector) and the noi se evaluated by the ratio of probe/reference with a nd without the excitation will determine the quality of the probe pulse. Below 365 nm, the intensity of the spectrum decreases rapidly, but for a good signal quality, it is necessary to produce a spectral distribu tion as flat as possible. The super continuum shown in Figure 2-7 is fl attened spectrally by a home made filter. This filter consists of a fused silica cuvette with 1.25 mm thick windows, filled with a mixture of dyes and polymers dissolved in dichloromethane. The total thickness of the

PAGE 81

65 cuvette is 4.5 mm. Our efforts proved that the quality of the supercontinuum, its stability and spectral smoothness, depends critically on the pump pulse energy, pump beam diameter, and focusing parameters of the lens. One has to play with the distance of the lens with respect to CaF2 plate (distances slight ly shorter or larger than focal length) and the pump energy to obtain the optimum, spect rally broad, and smooth supercontinuum. Before using a CaF2 window, we tried a sapphire plate (which is used in OPAs as a supercontinuum medium). Even though the st ability was good enough, spectrally there was no light below 350 nm. Thus, we searched for a medium which will generate enough light to get some signal down to 300 nm. Our investigations suggest that a white light continuum generated in CaF2 provides significantly more seed photons for shorter wavelengths ( < 500 nm) than the white light continuum generate d in sapphire (both pumped at 800 nm). The only disadvan tage is the instability of CaF2 material upon 3003504004505005500 5000 10000 15000 20000 25000 30000 CountsWavelength(nm) reference probewhitelight generation by CaF2 Figure 2-7. Spectrum of the white light continuum generated by CaF2. The probe (blue) and reference (red) beams used fo r transient absorp tion experiment.

PAGE 82

66 focusing a short pump pulse due to its very low damage threshold.119 Fortunately, by rotating the CaF2 window, the problem was solved enab ling the use of such white light continuum as probing beam. The fundamental 800 nm beam used to ge nerate white light continuum is passed through a half wave plate and a thin film polarizer is oriented at 45 degrees with respect to the pump pulse (excitation pulse for the sample) before it reaches to the CaF2 medium. By doing so, the pump energy for the continuum generation is controlled and optimized. At the same time, the CaF2 window is positioned at the fo cus of an off-axis parabolic mirror, which collects and collimates the white light continuum. Th e continuum is split into two equal beams: probe and reference beam*. After passing through the sample, the probe and reference beams are split into two equal portions independently (total four continuum beams). These beams go through four Glan-Taylor polarizers aligned perpendicular and parallel with respect to the pump pulse, allowing simultaneous detection of both polarizations In order to obta in the dynamics free from reorientation effects, the magic angle signal is calculated from parallel and perpendicular signals: //2 3magicAA A (2-11) The temporal evolution of anisotropy can also be obtaine d by evaluation of: // //2 AtAt rt AtAt (2-12) *Many beam splitters are not good enough for UV light down to 300 nm. Since the continuum has much less photons below 350 nm, it is vital to use the best optics for UV. I used reflective neutral density filters with OD 0.5 ( 33.3% absorption, 33.3. % transmission, 33.3 % reflection) to split probe and reference beams.

PAGE 83

67 where //At and A tcorrespond to the transient ab sorption of the polarization oriented parallel or perpendicular to the excitation beam polarization, respectively. The four tracks of white light are dispersed with an imaging grating monochromator (focal length 30.3 cm, 300 lin es/mm). The iStar in tensified CCD camera (iStar 720DH-720-25F-03, Andor Technology) is employed as the detection system. Note that, not all the probe wavelengths propagate through the optical components at the same speed (including the transparent medium used for continuum generation). Since the velocity, () () c v n at which a given frequency will travel depends on the refraction index of that material at that wavelength. When using the white light continuum, it is important to account for the influence of group velocity dispersion e ither experimentally or numerically. If the dispersion for the con tinuum is well characterized, the spectra can be corrected numerically by using a home-written Labview program. With the setup explained previously, data can be acquired in the following two ways: either time resolved measurements ( A versus t plots) at a fixed probe wavelength or spectrally resolved measuremen ts corresponding to a fixed delay between pump and probe ( A versus wavelength region of the probe). During our measurements, the signal appears earlier at shorter wavelengths than at a longer wavelength, which is a consequence of the temporal distribut ion of the different spectral components of the supercontinuum pr obe pulse, called “chirp”. The chirp of the white light continuum is determined from th e delay between the signals probed at 330 nm and 560 nm, for the measurements on the dendrim ers. As shown in Figure 2-8a, the chirp leads to a delay of ~520 fs between the spectral components at 350 and 575 nm. This chirp is corrected for the time-resolved measurements analyzed in the following

PAGE 84

68 chapters. Figure 2-8b also shows the signal fo r the DCM probed at di fferent wavelengths. The delay here is around 570 fs. The FWHM of the Raman-like (or coherent artifacts) signal generated in such solvents (when pumpe d with couple of micr o joules) also gives information about the time resolutio n of the experimental setup. The chirp correction is done after the experimental measurements. Each data set is analyzed to get an initial estimation of the chirp and then using a labview program based on calculating dispersion of light in every refracting medium, th e chirp corrected data sets are produced. This numerical correction employs the Sellmeier equation, which is an empirical relationship between the refractive index n and wavelength for a particular -0.250.000.250.500.751.001.25 0.000 0.005 0.010 0.015 0.020 probe=330 nm ODTime(ps) 520 fs 2G2mOH dendrimerprobe=560 nm (a)-0.250.000.250.500.751.001.25 0.000 0.005 0.010 0.015 probe=330 nm ODTime(ps) 570 fs pure DCMprobe=560 nm (b) Figure 2-8. The chirp of the white light c ontinuum determined from the delay between the signals (a) probed at 330 (square s) and 560 nm (circles) for the measurements of 2G2mOH dendrimer (b) pure dichloromethane. transparent medium. The common form of the equation for glasses is: 2 22 2 3 12 222 123()1 B BB n CCC (2-13) where B1,2,3 and C1,2,3 are experimentally determined Sellmeier coefficients. Different forms of the equation are used for certain type of materials, such as crystals and common

PAGE 85

69 organic solvents.120,121 Chirp correction is necessary a nd should be done very carefully since most of the dynamics in PE dendrimers takes place in a subpicosecond time scale. Time resolution of the experiment In general, to determine the time resolution of an experiment where two laser pulses are used, difference or sum freque ncy mixing between those pulses (pump and probe) is performed at the place of the sample using a very thin (50-300 m thick) type I BBO crystal. On the other hand, in the expe riments where the white light continuum is used as the probe, the time resolution can be determined from the coherent artifact (Raman Scatter) observed in some solvents. Many basic molecular liquids and optical solids are transparent in the visible and nea r-UV spectral range for low intensity radiation (< 1010 W/m2). However, when high power ultrashor t laser pulses are applied, these media can absorb efficiently through a multiphoton absorption mechanism (two photon absorption) and Stimulated Raman Amplification (SRA).122 Furthermore, a spectrally broad probe pulse, similar to white light continuum, favors efficient cross-phase modulation. Each of the signals is produ ced by the simultaneous action of two photons, one from the pump and the other from the probe. These artifact signals will terminate rapidly following excitation, thus their duration is directly related to the temporal width of the pump-probe cross-correlation function. This is verified by comparing duration of the coherent artifact at a specific probe wa velength with the cross-correlation measured using a single color probe mixed with the pump pulse in the BBO crystal. Simultaneous absorption of a pump photon and a probe photon gives rise to two-photon absorption, while the interchange of photons between pump and probe through a material’s vibrational energy level gives ri se to SRA. Moreover, the cr oss-phase modulation leads to

PAGE 86

70 spectral modifications within the probe upon pump-induced temporal changes of the refractive index.122,123 These coherent effects are observed in many common solvents, such as, acetonitrile, methanol, water, ethanol, cycl ohexane, and dichloromethane with different strength. Figure 2-9 presents typical coherent artifact signals in hexane, commonly used to measure the instrument response func tion. They are shown at two different wavelengths of the probe and it is obvious that the coherent artifact is getting smaller with increasing probe wavelength. The tempor al width is also dependent on the probe wavelength, due to group velocity mismatch As the probe wavele ngth is more distant from the pump wavelength, cross-correlation signals gets temporally broader. Such behaviors are in agreement with literature results.122 -0.250.000.250.500.751.00 -0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 probe=380 nm CC=180 fspump=310 nmprobe=330 nm CC= 110 fsODTime(ps) Figure 2-9. Coherent artifact of hexane solvent excited at 310 nm, probed at 330 and 380 nm. Observing such coherent artifacts from the solvents will provoke the question, “do we have solvent contribution to the signal?” It is essential to check the solvent response at the same pump energy where the experiment s are performed. In our experiments, these

PAGE 87

71 measurements confirmed that there is no cohe rent artifact from the solvent at the pump power intensities used for excitation of the real dendrimer samples. Otherwise, the signal should be corrected for solvent contributions. Concentration and Pump Pulse Energy Dependence For both fluorescence upconversion and tr ansient absorption experiments, the optical density of samples is approximately 0.2-0.3 mm-1, yielding a concentration range of 10-510-6 M. All spectroscopic measurements of dendrimers at room temperature are carried out in dichloromethane (DCM). Low temperature experiments were performed in glass forming methyl-tetrahydrofuran (Me-THF). It has been shown that phenyleethynylene dendrimers do not aggreg ate in dichloromethane despite some aggregation in solvents like is opentane (lower dielectric cons tant). The solvent should be anhydrous so that the dendrimer molecules can stay stable in solution for a longer period of time. The polymer experiments are performed in methanol and the samples were prepared at a concentration where stea dy state measurements did not show any aggregation. Same measurements may differ from one a nother due to different conditions of the day that the experiment is carried out. The excitation energy, beam diameter and even the concentration of the sample might be diffe rent on a particular day. For a reliable comparison, it is important to know if and how the pump energy influences the signal. The main criterion is to check the linear region of the pump energy. As long as the molecule is excited with power in the linea r regime, the shape of the signal will not change. The optimum pump energy can be de termined by varying the pump pulse energy until the shape of the signal changes. For each measurement, one should calculate the number of photons absorbed per molecule The most common problem with high energy

PAGE 88

72 pulses is annihilation, which may occur when more than one photon is absorbed by a molecule. In such a case, one excited state can act as a quencher for other excitations, resulting in additional decay components in the process. Thus, to detect the dynamics associated with the studied process, the excitati on intensity must be adjusted in a way that the average number of photons absorbed per molecule should be less than unity. Another scenario is the local heating of the sample leading to sample defects due to high excitation densities which can increase the pr oportion of the radiationless processes. The power densities for each measurement are given in the following chapters. The photostability of the samples is anot her important issue. The dendrimers are not as stable as the polymer samples. Depending on the pump power, they would photobleach much faster. Rotating sample ce lls are used for time-resolved experiments to use the minimum sample for the maximu m scan time. Absorption spectra of the investigated molecules were checked befo re and after each laser measurement which proves the photo stability of all compounds under the experimental conditions used in this work. All time-resolved experiments pres ented in this work are performed at roomtemperature.

PAGE 89

73 CHAPTER 3 ENERGY TRANSFER IN GENERATION 1 UNSYMMETRICAL PHENYLENE ETHYNYLENE DENDRIMERS The search for artificial light-harvesters has led to intense studies of conjugated dendritic macromolecules.36,81,124 Dendrimers have potential applications in photonic devices due to their highly branched architectures and uni que physical properties. With recent advances in synthetic methods,2,4,125 the size, topology, flexibility, and surface chemistry of dendrimers can be controlled at the molecular level with high precision. Accurate positioning of chromophores at the core periphery, focal point, or even at each branching point of the dendrimer can now be achieved.52,59,126 For photonic applications, the dendritic architecture cr eates large transition dipoles due to the high number of chromophore units. Some judiciously designed phenylene ethynylene (PE) based dendrimers show highly efficient and unidirecti onal energy-transfer properties.69,72 Their topology suggests applications as sca ffolds for light-harvesting devi ces. In addition, the large number of chromophore units lead to the formation of excitonic bands. These dendrimers’ photophysical properties cannot be understood as simply additive properties associated with molecular orbitals on singl e chromophore units. In this scenario, it is essential to understand the electronic coupling,84 exciton formation,72,127 and energy transfer in detail. A variety of dendritic arch itectures have now been s ynthesized, leading to unique photophysical properties. PE units coupled exclusively through meta or para substitution

PAGE 90

74 on a phenyl ring lead to either compact or extended dendrimers.72 These two families of dendrimers differ in the number of PE un its between consecutive branching points. Compact dendrimers have a fixed-length linear unit, while extended dendrimers have a variable number of linear PE units dependi ng on the number of branching points between the unit and the core. Both families of dendrimers have been investigated theoretically19,78,127,128 and experimentally.69,75,129 In compact dendrimers, steady state e xperiments performed by Moore and coworkers show that the optical excitation is localized on the PE units.72 The excitonic localization on individual PE units is evident by the m onotonic increase in absorption intensity and the lack of spectral shift with generation number.81 The extended series exhibit exciton localization on PE units of incr easing length (2-,3-, 4ring), which creates an energy funnel yielding multistep energy transfer. Mukamel and co-workers have studied compact and extended dendrimers using a Frenkel Exciton Hamiltonian.127 Applying the collective electronic oscillator model, they concluded that the electronhole pair was confined to the linear segments between branching points. The bound Frenkel excit ons are free to migrate throughout the molecule. Depending on the strength of the coupl ing, the migration leads to coherent or incoherent energy transfer. Absorption spectra ca lculated from this model are in excellent agreement with experiment. Using time correlated single photon counting, Swallen et al130 studied an extended PE dendrimer and found an instrument-lim ited value of about 10 ps for the energy transfer from the lowest energy chromophore in the bac kbone to a phenylene ethynylene perylene trap. Subsequent experiments by Klei man et al investigated the same extended

PAGE 91

75 dendrimer using femtosecond degenerate pump-probe spectroscopy and revealed a stepwise energy transfer from the shor ter PE units to th e longer PE chains.75 These experiments indicated that some of the steps in the energy transfer occur on a subpicosecond time scale. The novel characteristics of PE dendrimers arise from the electr onic properties at the branching points. In both co mpact and extended dendrimers, meta substitutions on the phenyl rings result in broken -electron conjugation in the gr ound electronic state. The situation in the excited state is less clear A recent study based on a di-ethynylene phenyl unit with H or phenyl substituents shows dram atic changes in both the emission spectra and the radiative lifetimes.90 The electronic structure calcula tions indicate that the phenyl ethynylene (H or phenyl substitu ted) units are strongly coupled in a relaxed geometry on the excited state.84 Experiments in larger dendrimers do not show the shifts predicted in the smaller systems (see ref 81, Figure 4). The extent of loca lization in the excited state for sizable dendrimers remains an open question. Unsymmetrical architectures in which coupling among the PE unit occurs through para and ortho substitutions have been synthesized by Peng and co-workers96,97,131 (Figure 3-1). In these dendrimers, the substitutions create combinations of PE units of variable lengths, analogous to thos e encountered in ex tended dendrimers.97 Unsymmetrical branching leads to rapidly gr owing conjugation lengths as the generation number increases, providing a broad abso rption spectral range with large molar absorptivities and high fl uorescent quantum yields.95 Linear conjugated segments connecting the periphery to the core suggest fa ster and more efficient energy transfer to the core.95 Furthermore, the presence of ortho substitutions may allow stronger coupling

PAGE 92

76 of PE units, leading to more delocalized ex citation throughout the en tire molecule. For these architectures, confined Frenkel excitons127 might extend over regions of the molecule that includes substitutions at ortho positions. Symmetric dendrimers’ absorption spectra can be interpreted as the addition of building blocks, defined by the confined excitons.127 The role of ortho substitution and exciton confinement in unsymmetric dendrimers is not as well characterized and there are still unanswered questions: Are the absorption band structures associated with exciton localization and “buildin g blocks”? Does coherent or in coherent energy transfer occur between those lo calized states? The goal of our study is to understand th e exciton size and th e rates of energy migration. We focus here on the characteriza tion of intramolecular interactions and the energy-transfer mechanisms in unsymmetrical PE dendritic molecules. To investigate the extent of delocalization with in the dendritic structure, we consider unsymmetrical monodendrons with multiple ortho and para substitutions. Energy transfer mechanisms are monitored by adding an ethynylene perylene trap (EPer), which acts as reporter for energy transfer. Time-resolved photoluminescence experiment s in the subpicosecond time scale are employed to follow the energy initially depos ited in the dendrimer’s backbone by an ultrafast pulse. A kinetic model is proposed to interpret the rise times of the fluorescence measured in unsymmetrical dendritic structures with and without an energy trap. Finally, we present an analysis of the validity of Frster model by comparing the model predictions with our experimental results.

PAGE 93

77 Materials and Methods The synthesis of unsymmetrical de ndrons is described elsewhere.96 Unsymmetrical monodendrons can be covalently bonded to form larger and more symmetrical macromolecules named as dior tri-dendrons. Here, two G1 (generation 1) monodendrons are coupled to a phenyl ring in the meta positions. When the phenyl ring has an additional OH or ethynylene perylene group in the other meta position, the molecule is named 2G1m-OH or 2G1-m-Per, respectively (Figure 3-1). The ethynylene perylene unit acts as an energy trap and is utilized to probe energy transfer dynamics. As any PE dendrimers, the molecules under investigation are quite ri gid molecules that do not allow for the backfolding of any of the branches. OCH3OCH3OH OCH3OCH3 (a) (b) Figure 3-1. Chemical struct ures of generation 1 phenylen e ethynylene dendrimers: (a) 2G1-m-OH (b) 2G1-m-Per. For spectroscopic measurements, all samples are prepared in dry CH2Cl2 without further purification. Absorption spectr a are recorded on a Varian Cary 100 spectrophotometer. The fluores cence spectra are measured with a Jobin-Yvon instrument (Spex–Fluorolog-3).The optical de nsity of samples used in a ll measurements is about 0.3 mm-1, which provides a concentration less than 10-6 M to avoid any aggregation and excimer formation. All steady state measurem ents and transient absorption experiments

PAGE 94

78 are performed in a 2 mm optical path length quartz cuvvettes. In addition, a homemade rotating cell with optical path of 1 mm (for optimum time-resolution) is used for timeresolved emission experiments. The lase r system, fluorescence upconversion, and transient absorption setup are described in detail in Chapter 2. The upconversion experiment measures the temporal evolution of the fluorescence with subpicosecond resolution. It is based on the sum-frequency mixing of the molecules’ emission with an ultrafast gate pulse in a non linear crystal. Briefly, excitation pulses are derived from an optical parametric amp lifier (OPA), pumped by a commercial TiSapphire laser system consisting of a Ti-Sa oscillator (Tsunami, Spectra-Physics) and subsequent amplifier (Spitfire, Spectra-Physic s) with a repetition rate of 1 kHz. The fourth harmonic of the OPA output signal or id ler is used to generate tunable excitation pulses in the 315 nm to 370 nm spectral region. Pump pulses of ~40 nJ with a beam diameter of 200 m are used to maintain a linear optic al response. After all the optics in the OPA and harmonic generation processes, the UV and visible pulses accumulate group velocity dispersion, yielding longer excita tion pulses and poor experimental timeresolution. To overcome this pulse lengthe ning we use a pair of quartz prisms to compensate the chirp. The homemade rotating cell has a 1 inch diam eter and an optical path length of 1 mm to guarantee excitation of a new sample vol ume with every laser shot with minimum consumption of sample. The photoluminescence is collected by two off-axis parabolic mirrors and the excitation volume is imaged onto a 300 m -BBO crystal. Usually a small portion of the regenera tive amplifier beam (~30 J/pulse, FWHM = 60 fs) is weakly focused (50 cm focal length) and the be am diameter at the crystal is kept larger

PAGE 95

79 than the imaged fluorescence. However, some modifications are necessary for experimentally challenging fluorescence wa velengths. While collecting fluorescence of 400 nm, a lot of background signal is in troduced because of the sum frequency generation of 800 nm (gate beam) and its s pontaneous second harmonic generation (400 nm) at the BBO crystal. With such leve l of background signal, it is impossible to distinguish the upconverted signal originated from the molecules studied here. Therefore, the gate beam has to be different from 800 nm. We modified the experimental setup to overcome this experimental limitation.The second harmonic of the OPA signal (630 nm and 740 nm) is used as the gate beam. Ho wever, it is still difficult to get good signal/noise with these gate beams since the intensity of SHG of the OPA signal is an order of magnitude smaller than the intens ity of the 800 nm beam. The samples were excited at 315 and 370 nm using the fourth ha rmonic of the signal generated in the OPA. Spatially and temporally overlapped ga te and collected fluorescence in the -BBO crystal generates a nonlinear response signal in the UV. Colored f ilters (UG11) are used to remove scattered light from the excitation pu lse and the second harmonic of the gate, which is also generated at the crystal. The upconverted beam is dispersed by a 0.25 m monochromator and detected with a PMT. Boxcar integration and averaging of 104 pulses per time step leads to a sign al to noise ratio of about 50:1. The time resolution of the setup was measur ed by detection of cross-correlation of scattered light from solvent and gate pulse. For UV excitation, the time resolution was determined to be about 225 fs. In the transient absorption experiments as sketched in Chapter 2, a fraction of the Ti:Sa amplifier output is focused on a 1 mm CaF2 plate to generate a white-light

PAGE 96

80 continuum, which is used as probe and refere nce beams. Using a thin-film polarizer, the probe light polarization is or iented at 45 degrees with respect to the pump pulse. After passing through the sample, a Gl an-Taylor polarizer splits the probe beam into its polarization components, parall el and perpendicular with re spect to the pump, allowing for the simultaneous detection of both polari zations. Pump induced absorption changes of both probe polarization components are measured as a function of pump-probe time delay by modulation of the pump beam with a mech anical chopper and dete ction of the probe beams and a reference beam with the pum p on and the pump off (to overcome shot-toshot fluctuations) using a CCD camera e quipped with a 30 cm spectrograph. For all transient absorption measurements performe d here, the excitation wavelength is 315 nm. By measuring the coherent artifacts from the pure solven t (i.e Stimulated Raman Amplification, discussed in Chapter 2), the time resolution of this setup was determined to be about 150 fs. Data analysis involves the convolution of decay and rise time functions with the corresponding experimental IRF for each expe riment. The integrity of the sample was checked before and after each set of measurements. Steady State Spectroscopy The steady state absorption spectra of 2G1-m-OH and related monodendrons are shown in Figure 3-2. Any 2Gndidendron is composed of two Gn monodendrons coupled through the meta positions of a phenyl ring. Th e branching center de termines the strength of interactions among chromophores, which plays a key role in determining the mechanism of energy transfer. In Figu re 3-2, the absorption spectrum of 2G1-m-OH, G1OH, and G2OH are compared.

PAGE 97

81 250300350400450500 0.0 0.2 0.4 0.6 0.8 1.0 2G1-m-OH G1-OH G2-OHNormalized AbsorbanceWavelength (nm)EMS Figure 3-2. Normalized absorption spectra of — 2G1-m-OH, ….. G1-OH, and — G2-OH in dichloromethane. Normalized emission spectrum of — 2G1-m-OH. The absorption spectrum of 2G1-m-OH shows a 35 nm red shift compared to single G1OH dendron.131 This is due to one additional PE unit, which increases the total conjugation for the system. Th e absorption spectrum of G2OH (Figure 1-6) shows more similar features, but there is a 20 nm re dshift. The longest lin ear PE chain in G2OH has the same number of PE units as th e longest linear chain in the 2G1-m-OH. This red shift is due to more extended conjugation between the two longest linear PE units through the ortho linkage in G2OH. Also note that the red shif t here is much more pronounced compared to 2G2-m-OH versus G3OH dendrimers (Figure 4-2). The trap molecule, Eper, is substituted to the didendron molecule in the meta position with respect to both monodendron co mponents. The absorption spectrum of 2G1m-Per is a superposition of absorption spectra of 2G1-m-OH and EPer (Figure 3-3). The broad absoprtion feature between 300 and 400 nm corresponds to the dendritic backbone

PAGE 98

82 while absoprtion at > 400 nm is only associated with the Eper trap. Since the presence of both donor and acceptor groups within the sa me molecule does not lead to appearance of a new band or to differences in the ground state absorption spectrum, it can be concluded that there are no strong interac tions between donor and acceptor moieties in the ground state. 300350400450500550600 0.00 0.25 0.50 0.75 1.00 Normalized absorbanceWavelength(nm) Figure 3-3. Normalized absorption spectra of -----2G1-m-OH, …EPer, and —2G1-m-Per and fluorescence spectrum of —2G1-m-Per, excited at 315 nm. The fluorescence of PE dendrimers attached to Eper trap is completely quenched compared to the ones without trap In fact, after excitation at = 315 nm, the emission arises entirely from the EPer unit. This is again a strong indication that within these dendrimers, the excitation ener gy is efficiently transferred from the dendrimer backbone to the EPer chromophore. Comparison of ab sorption and excitation spectra indicates ~96% efficiency for the energy transfer process.92 Interestingly, at 400 nm, a small band with intensity contributions from unfunctionalized 2G1-m-OH and possibly residual

PAGE 99

83 backbone emission from 2G1-m-Per is noticed in the emission spectrum. Time-resolved data will clarify its origin. Time-Resolved Emission Experiments The fluorescence decays are measured by TCSPC and 2G1-m-OH decays with a 1.8 ns time scale, whereas 2G1-m-Per decays with the emission lifetime of EPer (2.2 ns). Time resolved fluorescence upconversion tec hnique is applied to measure the rise times associated with these d ecays. First, the dendrimer wit hout EPer trap is studied to understand the extent of intramolecular inte ractions within the didendron backbone. The absorption spectrum of 2G1-m-OH has two distinguishable bands peaked at 305 and 365 nm. Since one of the goals of this study is to explore the possibi lity of assigning the absorption band structure to exciton local ization, it is reasonable to excite 2G1-m-OH at selective wavelengths with significant cont ributions from each band. Therefore, the excitation wavelengths are chosen to be 315 and 370 nm. The emission is detected at backbone fluorescence of 400 nm. Note that the instrument response functions of 180220 fs were used in the analysis of all meas urements for deconvolution of data sets (IRF was routinely recorded during each measurement session). Figure 3-4 shows no detectable exci tation wavelength dependence for the subpicosecond risetime. Convolution of the IRF with the exponential rise function yields a 300 fs time constant for both excitation wa velengths, which suggests the delocalization of the initially excited stat e throughout the monodendrons. This risetime is definitely longer than the experimental time-resolution, meaning that it essentially takes 300 fs for the initially deposited energy to reach the lowest lying emitting state. It is thus suggested that absorbing and emitting states of the dendrimer are two different excited states.132 A second component with a decay time constant of 6 ps is found in addition to the long

PAGE 100

84 decay time of 2G1-m-OH (1.8 ns). This ki netic component can be at tributed to vibrational relaxation in the excited stat e of the monodendrons, which is coupled to relaxation and reorganization of the solva tion shell around the monodendrons.133 The solvent molecules have to accommodate for the newly populated S1 state of the whole molecule. The fluorescence is detected at the blue end of the emission at 400 nm (Figure 3-2). At fluorescence detection wavelengths close to th e excitation, the vibrational relaxation will be observed as a fast decay component, wher eas at longer wavelengths this decay would be seen as a risetime, since the fluorescence is detected from a state that has to be populated with this time constant. Moreover, the relative amplitude a ssociated with 6 ps component will depend on the excitation wavelength. The excitation wavelength dependence of the 6 ps component will be expl ored with the kinetic analysis. If it is attributed to vibrational relaxation, its amplitude will decrease at longer excitation wavelengths. The objective of i nvestigating the 2G1-m-OH molecule was to understand the excitation energy transfer dynamics among th e similar dendrons before the energy is transferred to a core trap. We investigate al so the same dendritic structure with an EPer trap attached in meta position to the core phenyl unit. The risetime of the Eper emission is experimentally measured to follow the ex citation energy migration from the initially excited state on the didendron to the final tr ap. Figure 3-5 shows the temporal evolution of the 2G1-m-Per fluorescence as a function of ex citation wavelengths. Fluorescence is detected at the maximum emission wavelength of Eper (480 nm). Direct excitation of EPer provides the time resolution limit for fluorescence risetimes. Any risetime longer than 150 fs can be assigned to excited state dynamics.

PAGE 101

85 0 1 1 024 0 1 Upconverted Fluorescence(b) 315 nm (a) 370nm Time (ps) Figure 3-4. 2G1-m-OH in dichloromethane excited at a) 370 nm b) 315nm. Upconversion signal of fluorescence is detected at 400 nm, the maximum emission wavelength of the molecule. Fitting pr ocedures include the convolution of exponential functions from the kinetic m odel with the IRF. The best fit is shown as the solid line. The longest IR F function is shown in panel b. The bottom panel shows the super position of the experimental data in panels a and b. In addition, any different dynamics compared to what is measured in Figure 3-4 can be attributed to energy transfer to the EPer tr ap. Figure 3-5a and 35b shows the upconverted emission following excitation at 370 and 320 nm, respectively. At these excitation wavelengths, there is no residual absorpti on from the EPer unit. Indeed, the only mechanism responsible for EPer emission is sensitized excitation via energy transfer from the backbone. The risetimes for em ission show an excitation wavelength dependence. As the excitation wavelength gets shorter (higher energy), the fluorescence risetimes become longer. Excitations at 315 nm (b) and 370 nm (a) clearly show longer risetimes (compared to both IRF and risetimes from 2G1-m-OH emission ) for the

PAGE 102

86 fluorescence arising from the EPer excited state, whereas the dendritic backbone (2G1-mOH) does not display such long risetime with excitation wavelength dependence. As shown in the steady state spectrum, most of the 2G1-m-Per emission is from the EPer trap. However, a small band around is observed 400 nm, the same wavelength region as the emission from 2G1-m-OH. This band has contribution from unfunctionalized dendrimer (2G1-m-OH) and residual emission from the dendritic backbone even in the presence of the trap. To confirm this assignment, the time-resolved fluorescence detected at =400 nm from the 2G1-m-Per molecule has to be examined and the temporal behavior will give direct inform ation about the energy tr ansfer rate. As seen in Figure 3-6, a very fast risetime is followed by a fast decay, which is attributed to energy transfer from the backbone to the EP er trap. The signal does not decay to zero with the same fast time constant. The long time component corresponds to unfunctionalized 2G1-m-OH. Using the kinetic m odel described in the next section, the signal is simulated using the same time c onstants obtained from Figures 3-4 and 3-5 along with the nanosecond com ponent corresponding to 2G1-m-OH emission lifetime. The relative contribution to the steady state emission spectrum from the 2G1-m-OH (as impurity) and the residual backbone emission can be obtained by integrating the timeresolved data. Even though the extent of unfunctionalized 2G1-m-OH of the sample is checked by Thin Layer Chromatography show s no impurity, spectroscopic measurements are more sensitive and can detect 1-2% impurity. A concentration of 1% 2G1-m-OH impurity in a solution of 2G1-m-Per yields at least 99% of the integrated fluorescence of the band peaked at 400 nm. The fast rise and decay component associ ated with the energy transfer can only be dete cted because of the ultrafast time-window probed.

PAGE 103

87 0 1 1 024 0 1 (b) 315 nm Upconverted Fluorescence (a.u.) (a) 370 nm Time (ps) Figure 3-5. Upconversion signal of 2G1-m-Per detected at 485 nm, excited at (a) 370 nm, (b) 315nm. The fittings correspond to th e convolution of the IRF and data. The longest IRF function (at 315 nm) is shown in panel b. The bottom panel shows the superposition of the experi mental data in panels a and b. In summary, the femtosecond fluorescence measurements on generation 1 PE dendrimers exhibit subpicosecond component s along with a few picoseconds (~6 ps) decay component. Without the Eper trap, 300 fs risetime independent of the excitation wavelength indicates the deloca lization of the initially excited state while the emission occurs from a more localized state. The kinetic analysis will reveal the amplitude of each component. The 6 ps time constant and th e excitation wavelengt h dependence of its amplitudes yield a strong argument to attribut e this component to a vibrational/solvent relaxation process going on in the excited S1 state. For the 2G1-m-Per, the time-resolved fluorescence measurements along with the steady state data verify the very efficient and fast energy transfer process. The nanosecond co mponent is straightforwardly attributed to

PAGE 104

88 the fluorescence lifetime decay. The presence of these kinetic processes is also investigated with broadband tr ansient absorption measurements. -1012345678 0.0 0.2 0.4 0.6 0.8 1.0 Upconverted Fluorescence (a.u.)Time (ps) Figure 3-6. 2G1-m-Per Upconversion Signal. excitation=315 nm, emission = 400 nm. Solid line is a simulation of the E fluorescence with fixed time constants from previous fits. See text for the details about the model. Time-Resolved Broadband Transient Absorption Measurements Before the interpretation of the transien t absorption spectra, one should remember that there are three contribu tions to a pump-probe signal. While the excited state absorption leads to increased absorption (positive A values), both stimulated emission and ground state bleaching leads to a decrease in absorption (negative A values). Transient absorption (TA) data for 2G 1-m-OH is shown in Figure 3-7. The excitation wavelength was set to 315 nm for the measurement of transient absorption with a maximum bleach signal of 3 x 10-3 OD. At positive time delays, two different main components are observed. A negative signal from 300 nm to 425 nm and a positive signal beyond 425 nm are observed. Both feat ures can be seen instantaneously after excitation (it takes 300 fs to reach the max values) and they decay on a nanosecond time

PAGE 105

89 scale. The signal in the tran sient absorption spectrum for > 420 nm is positive, so it can definitely be attributed to an excited state absorption. From the steady state studies, it is known that 2G1-m-OH has fluores cence quantum yield of 0.70 and a fluorescent lifetime of 1.8 ns.92 Thus, the excited state absorption observ ed in the measurements reported here can be attributed to S1-Sn absorption within the 3 ri ng phenylethynylene chromophore.134 Since the steady state absorption spectrum ends at 405 nm and the fluorescence signal extents from 365 nm to 600 nm, the ne gative signal in the transient absorption spectrum can not solely be attributed to ground state bleaching. It seems reasonable to say that ground state bleach dominates the signal between 300 and 365 nm. For > 365 nm, both ground state bleaching and stimulated emission are responsible for the negative absorption signal, whereas between 385 nm and 420 nm stimulated emission dominates. However, since the steady state fluorescen ce spectrum extends from 380 nm up to 600 nm with a maximum at 400 nm, a contribution of stimulated emission beyond 420 nm (where the positive photoinduced absorption sign al dominates) is also expected. The net result of transient photoinduced absorption and stimulated em ission gives rise to a large positive signal, implying that in this region th e cross section for excited state absorption is larger than that for the stimulated emission. The assignment of ground state bleaching a nd stimulated emission in the negative region of the transient signal needs furthe r discussion. The maximum of the negative signal is centered at 382 nm. This is in between the maximum of the ground state absorption band (the maximum is around 372 nm at the red edge of the absorption) and the maximum of the steady state fluorescence (400 nm). Therefore, the negative signal at 382 nm is a combination of both bleach and st imulated emission. It is important to note

PAGE 106

90 that the broad bleach signal reaches its maxi mum value within 300 fs while the peak at 382 nm still increases in amplitude. The br oad bleach signal extended over the whole absorption spectrum reveals the delocalizat ion of excitation energy on the initially excited state. The band peaked at 382 nm ri ses with a slower rate and shows a more localized characteristics compared to the broad bleach signal in the blue region. 350400450500550 -5 0 5 A(mOD)Wavelength(nm)t in ps -1.00 1.00 1.15 2.50 7.00 50.00 -5 0 5 A(mOD)t in ps -1.00 0.10 0.25 0.50 0.75 ABSEMS Figure 3-7. Transient ab sorption spectra of 2G1-m-OH molecule at di fferent time delays, excited at 315 nm. Time delays are s hown with different colors. The solid blue and red lines are absorption and emission spectrum, respectively. Top panel shows rising, whereas bottom panel shows decaying components. The transient signals decay mainly on a nanosecond time scale. However, a kinetic analysis of the transient absorption intensitie s as a function of dela y time at different wavelengths reveals an additional picosecond relaxation process. Th is relaxation process can be seen easily in Figure 3-8. The ba nd with a maximum at 382 nm becomes narrower

PAGE 107

91 due to the decrease in the stimulated emission contributi on. The photoinduced absorption is observed instantaneously for the wavelengths > 420 nm, and its intensity decreases along with a red shift. The intensity decr ease of the 382 band and broad photoinduced absorption occurs within the same time scale. If we look at the intensity of the transient absorption signal in function of time at 382 nm and 440 nm, an additional process can be determined with time constant of 6 ps t ogether with the nanos econd fluorescence lifetime decay. The presence of this kinetic proce ss and its wavelength dependence was discussed in the results of upconversion experiment. Th is component is assi gned to vibrational relaxation in the emitting excited state of the molecule. As shown in Appendix B, for the numerical analysis we had to treat this rela xation process in a different way than the raw data set, since SVD is not suitable for spectral shift, broadening or narrowing in time. 375400425450 -5 0 A(mOD)Wavelength(nm)t in ps -1.00 1.00 2.50 4.50 10.00 21.00 50.00 Figure 3-8. Transient abso rption spectra of the 2G1-m-OH molecule at different delay times. Detailed display of the 350 nm-450 nm region. The transient absorption spectra of 2G1-m-Per in CH2Cl2 after the excitation of 315 nm are shown in Figure 3-9. Immediately afte r excitation, the transient absorption signal

PAGE 108

92 starts evolving spectrally in the same way as the 2G1-m-OH molecule. However, this transient spectrum undergoes changes on a ve ry fast time scale of a few hundred femtoseconds. As shown in Figure 3-9, within 1 ps, the transient signal corresponds to that expected for the ethynylene perylene ( acceptor) molecule. This time evolution of the transient spectrum clearly shows the exci tation energy transfer from the dendritic backbone to the trap molecule. A very sma ll backbone bleach signal is observed at early times, but it is rapidly overcome by a ne w photoinduced absorption bad from EPer ( A>0). At longer wavelengths the negative signal corres ponds to the bleach and stimulated emission signal from EPer. 350400450500550 -10 -5 0 A(mOD)Wavelength(nm)t in ps -1.00 1.00 1.15 2.50 7.00 21.00 50.00 -10 -5 0 5 A(mOD)t in ps -1.00 0.00 0.10 0.25 0.50 0.75 1.00 ABSEMS Figure 3-9. Transient ab sorption spectra of 2G1-m-Per at different time delays, excited at 315 nm. Top panel shows the short time scale while the bottom panel shows the long time scale. The subpi cosecond energy transfer is verified with the fast ground state bleaching and fast evoluti on of EPer stimulated emission.

PAGE 109

93 Kinetic Model Both time-resolved experiments reveal that the energy is efficiently transferred from the dendrons to the trap. The goal of these experiments is to understand the mechanism of this process and the extent of interaction (due to electronic coupling) between donor and acceptor moieties. The detailed analysis of both experiments will estimate the rate of energy transfer and will he lp understand the electronic structure of the backbone and its interaction with the trap molecule. The data analysis is performed independe ntly for each molecule. The fluorescence measurements are analyzed globa lly using a nonlinear least squa re fit routine from Origin Software. The broad transient absorption meas urements are analyzed using the Singular Value Decomposition (SVD) algorithm from Matlab. For all molecules, a sum of exponentials will fit the data se ts properly. In fact, we solve the associated differential equations of the proposed kinetic model to describe the populati on of each electronic state. We believe that it is crucial to appl y a kinetic model instead of using a sum of independent exponentials to derive the physic al meaning. Upon combination of these two analysis approaches for two different expe riments, one can confirm the measurements, calculations and fit procedures. For the 2G1-m-OH, the simplest case is consid ered as shown below in Equation 31. D is the ground state of the backbone, D* is the initially excited state, and E* is the emissive state of the backbone. 1 **rad BBk h kDDE (3-1)

PAGE 110

94 After convolution of the data with the IRF, a risetime of 300 fs was obtained from the fit to the emission data. This risetime was not dependent on excitation wavelength, which indicates the initial de localization of excitation with in the backbone. Also this relatively long risetime (not in stantaneous, longer than IRF) suggests that the initially excited state undergoes a conformational relaxa tion before emitting. In addition to a long nanosecond decay of the fluorescence, there is a second decay component with a 6 ps time constant. The partial amplitude of th is component changes as a function of excitation wavelength. When excited at 370 nm, the partial amplitude is smaller compared to the one at 320 nm. Taking into account the amplitude behavior of this kinetic component, this feat ure is attributed to a vi brational relaxation in the electronically excited state of the dendritic backbone. Excitation at the blue end of the spectrum (315 nm) leads to a highly excite d vibrational population. This process is coupled to a relaxation and re organization of the solvent around the chromophore as the solvent molecules have to accommodate for the newly populated S1 state of the dendrimer. At excitation wavelengths close to the blue end of the absorption band, this relaxation will be seen as a decay component with larger amplitude, whereas at longer excitations the amplitude will decrease. In similar investigations, time constant of a few picoseconds were observed and attributed to vibrational relaxati on process of polyphenyl ethynylene dendrimers in organic solvents. Spectral changes are measured as a function of delay time, and it is also possible to get kinetic information by looking into the inte nsity of TA signal as a function of time at particular wavelengths. In general, the transient absorption signal (both positive and negative signals) of this molecule decays with 1.8 ns time constant. There is very fast

PAGE 111

95 risetime around 300 fs for both ground state bl each signal and exci ted state absorption signal. Considering the time dependent deve lopment of transient absorption spectrum at wavelengths between 360 nm and 435 nm, a re laxation process leading to a small red shift of the fluorescence spectru m within 6 picoseconds can be observed. This is in very good agreement with vibrational relaxa tion process observed in the emission experiments. The fluorescence upconversion and tr ansient absorption signals give complementary results. When 2G1-m-OH is excited, it takes ~300 fs for the bleach signal to evolve into the steady state absorption. Th is is an indication of initially delocalized excitation and this is the time scale associat ed with the conformational change. After the initial excitation, the vibrational cooling is observed and finally a 1.8 ns fluorescence lifetime is measured. For the 2G1-m-Per molecule, addition of EPer trap results in a new final excited state P*. Emission is detected from the P* and the backbone state E*. The simplest deactivation pathway, population transfer from the emitting state of the dendrimer backbone (E*) to the emitting state of the EPer (P*), is proposed. According to this model, the excitation energy is flowing to the trap molecule in a stepwise manner. Thus the vectorial nature of the energy transfer is investigated. The analysis simply involves the following scheme: 12***radrad BBEperkk h kkDDEP (3-2) Measurements of the 2G1-m-OH fluorescence probe the E* state directly, whereas experiments with the 2G1-m-Per probe the EPer emission ( P* ) when detection is at =

PAGE 112

96 485 nm and backbone emission ( E* ) from 2G1-m-Per for early-time detection at = 400 nm. By solving the kinetic equations for this model, one can obtain the populations of each excited state with minimum number of parameters in the fitting functions. The backbone excited state p opulation is given by: 121 21*ktktk Etee kk (3-3) The population of the trap excited state (P*) can be written as the following: 12* 12 21121ktktkk ee Pt kkkk (3-4) The fluorescence lifetimes of E* and P* appear only as a constant offset in the model since they are on the order of nanosec onds. The preexponential factors are defined in terms of rates, which will reduce the number of free parameters, and energy transfer channels are given by char acteristic times of (k1)-1 and (k2)-1. The fitting results are summarized in Table 1. Table 3-1. Fits for Time-Resolved Fluorescence Data exc 1 & # (fs) 2 (fs) 3 (ps) 315 nm 300 20 350 40 6 1.25 (11%) 370 nm 300 30 250 50 6 1.25 (17 %) & Time constants (1) for D* E* and (3 ) for vibrational relaxation of D* are obtained from the time-resolved fluorescence measurement of 2G1-m-OH and kept constant when fitting the 2G1-m-Per data. # Errors correspond to 2 Excitation at 315 nm and 372 nm shows differe nt dynamics. It is crucial to mention that at these wavelengths ther e is no direct excitation of EP er, and the emission is solely

PAGE 113

97 due to energy transfer from the backbone to the trap. Excitation at 315 nm yields an energy transfer step with k2 -1= 350 fs, after the initial deloca lization of exc itation within k1 -1=300 fs. At 370 nm, the fit yields a fast er rate for the energy transfer, k2 -1= 250 fs. At this wavelength the excitation is very cl ose to the emitting state of the dendritic backbone. However, the vibrational re laxation process measured from 2G1-m-OH molecule is not obtained when EPer is attached. This is due to very fast energy transfer to EPer. The energy transfer is faster compared to vibrational relaxation; this is already an indication that the coupling between backbone (donor) and EPer (acceptor) is beyond a very weak coupling. When the transient absorption of 2G1-m-Per is analyzed with SVD using the same kinetic model and the very fast risetime fo r the bleach and photoinduced absorption are obtained. If we compare with the dynamics of 2G1-m-OH molecule, the excitation energy is transferred within 350 fs. In addition, the spectral shif t within 6 ps is not observed anymore, suggesting again the energy transf er is much faster than the vibrational relaxation within the backbone. The kinetic model proposed here is fu rthermore confirmed via collecting the emission dynamics of E*(donor emitting state) in the presence of acceptor EPer. In this case the rise and decay times of the fluores cence gives direct meas urement of the energy transfer process. Using the rates obtained from these fits, we predict the temporal behavior of E*(t). Figure 3-6 shows that the experimental data agrees very well with this prediction.

PAGE 114

98 Overall, the agreement between fluorescen ce upconversion and tr ansient signals is excellent. We need to understand the nature of excitations and the extent of coupling between backbone and EPer. Energy Transfer via Weak Coupling As explained in detail in Chapter 1, en ergy transfer can result from different interaction mechanisms. The interactions can be Coulombic and/or due to intermolecular orbital overlap. The magnitude of this interaction is very impo rtant. Briefly, in the strong coupling case, excitation energy transfer is faster than the nuclear vibrations and vibrational relaxation. The excitation ener gy is delocalized over donor and acceptor, the transfer of excitation is a coherent process. In the weak coupling the electronic excitation is more localized than under strong coupling. However, the system can be described in terms of stationary vibronic exciton states (vibronic excitation is delocalized). The transfer rate is faster than vibrational re laxation but slower than nuclear motions (in contrast to strong coupling). In the case of very weak coupl ing, the vibrational relaxation occurs before the energy transfer takes place.135 Here we report that the donor molecule (2G1-m-OH) has vibrational relaxation of 6 ps (verified by both time-resolved emission a nd absorption experiments). However, when the acceptor is covalently attached to this donor, the energy transfer occurs in a time scale of a few hundred femtoseconds, which is much faster than vibrational relaxation of donor. This is and indication that the donor and acceptor molecules are weakly coupled here. The interaction energy must then be smaller than the absorption bandwidth but larger than the width of an isolated vibronic level. In the very weak coupling regime, the ener gy transfer rate can be evaluated using the simplest model, Frster. This model is based on the dipole-dipole approximation

PAGE 115

99 which is valid when the donor-a cceptor separatio n is much larger than the dipole sizes. For the conjugated dendrimers investigated he re, this approximation is not valid, thus attempts to apply Frster theory may not yield good agreement with the experimental results. Using the excitation transfer rate in the very weak coupling limit, we will calculate the interaction energy and compare it to the values usually obtained within the Frster regime. The transfer rate in the very weak coupling limit can be calculated from 2 2 24ETkJV hc (3-5) where V is the coupling and J is the vibronic spectral overl ap integral between donor and acceptor (in units of cm). The spectral overlap is a measure of the density of interacting initial and final states. The accurate calcula tion of the overlap integral employs the homogeneous vibronic bandwidths. Since th e homogeneous bandwidths of the vibronic bands in the di-dendron are unknown, we re ly on absorption and emission spectra to obtain J The spectral density can be evalua ted by using a normalized absorption spectrum -acceptor, a( )-, and normalized emi ssion spectra –donor, f( )-, 27()10f iJcmfanmd (3-6) yielding J = 1.187x10-4 cm. An experimentally measured transfer rate kET =(300 fs)-1 yields an interaction energy V = 155cm-1. The approximation on the J calculation results on an upper limit for J and therefore, for a given e xperimental rate, a lower limit of the Coulombic interaction. The typical transfer rates obtained for the ot her dendrimer structures using Frster model are slower yielding lower interaction energies.132,136,137 Thus, the intera ction energy of 155 cm-1 is not in the very weak coupling regime.

PAGE 116

100 Conclusions Time-resolved fluorescence and absorption dynamics of a generation 1 phenylethynylene didendron with and without en ergy trap are presented in this chapter By globally analyzing the data sets from fluorescence upconversion and transient absorption measurements and comparing the resu lts, we are able to identify and attribute various kinetic components. A 6 ps decay component observed from the 2G1-m-OH molecule is attributed to vibrational relaxa tion in the excited state. Since the emission detected from the longest linear PE chain in the 2G1-m-OH (3-ring PE) shows a risetime of 300 fs independent of the excitation wavelength, deloca lization is noticed in the initially excited state. The presence of ortho and para substitutions in such unsymmetrical structures supports th e initial exciton delo calization (unlike meta substitution). After excitation, a change in the excited state surf ace leads to localization, which is supported with the localized peak ar ound 382 nm. The built-in energy gr adient results in very efficient energy transfer to the EPer trap and yields a cascade mechanism. The energy transfer process occurs in sub-picosec ond (250-350 fs) time scale. However, the interaction energy calculated using very weak coupling limit formulation yields a value of 155 cm-1, which is above the very weak coupli ng limit. Moreover, the transfer rate being faster than vibrational relaxation reduces th e very weak coupling probability, thus questions the validity of F rster mechanism for the systems under investigation. Further investigations of larger generati on dendrimers will complete the map of the energy transfer pathways in the unsymmetrical PE dendrimers, and will help understand the mechanisms responsible for maintain ing the highly efficient light-harvesting properties.

PAGE 117

101 CHAPTER 4 ENERGY TRANSFER IN GENERATION 2 UNSYMMETRICAL PHENYLENE ETHYNYLENE DENDRIMERS As described in Chapter 3, the unique uns ymmetrical branching leads to rapidly increasing conjugation lengths of PE chains as the generation number is increased.131 A perylene unit is attached to the focal point of dendrons to investigate the structure – property relationship regarding th e energy transfer dynamics and -conjugation. In one case, perylene is directly attached to the monodendrons lin early resulting in extended conjugation between the PE dendron and perylene unit.96 The other case, in which two unsymmetrical PE dendrons and an ethynyl perylene trap are linked through the meta positions of a phenyl ring, creates a new family of di-dendrons.97 The meta substitution of the central phenyl ring disrupts the ground state -conjugation, providing a degree of isolation between the monodendron backbone and the trap. To evaluate the effect of -electron delocalization on th e energy transfer dynamics, we can compare the 2Gn-m-Per di-dendrons with Gn+1Per monodendrons. Both molecules have the same total number of PE units and th e same number of PE units in the longest linear chain.95 To determine the influence of the gene ration and size of PE dendrimers on the energy transfer process, and degree of localiz ation, a larger didendron with and without ethynylperylene trap is investig ated. The main focus of this study is to explore the nature of electronic excitations as well as the speed and efficiency of energy transfer. In this chapter, complementary experiments of u ltrafast time resolved emission and broadband

PAGE 118

102 transient absorption experiments are describe d. A kinetic model parallel to the one described in Chapter 3 is applied for both molecules. The energy transfer dynamics is compared to the generation one di-dendron. Th e unidirectional energy transfer is verified via detecting emission from an intermediate st ate. Finally, we present an analysis of the validity of Frster model by comparing the model predictions with our experimental results. Materials and Methods The synthesis of unsymmetrical PE de ndrons has been described elsewhere.97 Unsymmetrical monodendrons can be combined in to larger structures, leading to more symmetric macromolecules. For example, Figure 4-1a shows two G2 (generation 2) dendrons coupled to a phenyl ring in meta positions. The phenyl ring has an additional OH group in the other meta position and is thus named 2G2-m-OH. The addition of an energy trap to a similar structure can be used to quantitatively probe the energy-transfer dynamics. In this case, an ethynyleneperylene unit replaces the OH group, leading to the 2G2-m-Per molecule as shown in Figure 4-1b. Even though the chemical structure seems to describe a planar molecule, the 3D modeling shown in Figure 4-1c clearly demonstrates otherwise. To obtain a reasonable picture of th e room temperature structure of 2G2-m-Per, we look at Molecular Dynamics (MD) simulations performed by Roitberg and Krause.* The system is built in Hyperchem 7.0 and run for 100 ns (T=300 K) using the program TINKER. The MM3 force field is used with the rota tional barrier around the ethynylene triple bond raised to 0.6 kcal/mol to reflect experimental observables. The simulations are performed in vacuum and shoul d be interpreted only as guides to overall Personal communication

PAGE 119

103 conformational shapes. After se veral nanoseconds of the MD run, a snapshot (Figure 41c) was taken showing the globular shape of the dendrimer. The small size of this dendrimer and its rigidity does not allow for the “fold-back” of any of the branches. In addition, the trap is completely exposed, w ith the dendritic structure forming a sort of “bouquet”. H3CO OCH3H3CO OCH3OCH3OCH3OH Figure 4-1. Chemical structures of phenylene ethynylene dendrimers: (a) 2G2-m-OH (b) 2G2-m-Per and (c) 3D model of the 2G2-m-Per from a MD simulation. Steady state characterization of the di-dendrons is performed via UV-Vis absorption in a Varian-Cary 100 and emission in a Jobin-Yvon instrument (Spex Fluorolog-3). All steady state measurements are performed in the same homemade rotating cell (optical path of 1 mm), which is also used for the time-resolved emission experiments. Sample concentrations are kept below 10-6 M to avoid any aggregation138 or excimer formation,139 yielding optical dens ities less than 0.3 mm-1. Femtosecond timeresolved photoluminescence is employed to explore excited state dynamics and energy transfer. The upconversion experiment meas ures the temporal evolution of the fluorescence with subpicosecond re solution. It is explained in detail in Chapter 2 and additionally some brief explanat ion is given in Chapter 3. The homemade rotating cell has a 1 inch diam eter and an optical path length of 1 mm to guarantee excitation of a new sample vol ume with every laser shot with minimum (a) (c) H3CO OCH3H3CO OCH3OCH3OCH3 (b)

PAGE 120

104 consumption of sample. The photoluminescence is collected by two off-axis parabolic mirrors and the excitation volume is imaged onto a 500 m -BBO crystal. A small portion of the regenerative am plifier beam (800 nm, ~30 J/pulse, FWHM = 60 fs) is used as the gate pulse. Excitations pulses at variable wavelengths are obtained from the fourth harmonic generation of the OPA’ s signal output.The Instrument Response Function (IRF) is measured with pump sca ttered light, which is up-converted with the gate pulse yielding cross-correlations FWHM of 175 fs and 220 fs in the visible and UV regions, respectively. The transient absorption measurement is performed using a whitelight continuum as the probe beam. The experimental deta ils are explained in Chapter 2. For the molecules studied in this chapter, the exci tation pulse is set to be 320 nm. The IRF is measured to be ~ 150 fs. Data analysis i nvolves the convolution of decay and rise time functions with the corresponding experimental IRF. The photostability of the sample is checked by steady state UV and emission sp ectroscopy before and after each timeresolved experiment. Steady State Spectroscopy of Phenylene Ethynylene Dendrimers G2-OH is a second-generation unsymmetrical (PE) dendron and 2G2-m-OH is a didendron consisting of two G2 dendrons coupled through the meta positions of a phenyl ring (Figure 4-1a). It has been shown previous ly that the choice of site-substitution on the focal point governs the nature of the optic al excitations for the entire molecule.87,127,140 For the two molecules u nder consideration (2G2-m-OH and 2G2-m-Per), substitution at the focal point is the same, leading to sim ilar steady state characteristics as shown in Figure 4-2a. The absorption spectra of 2G2-m-OH shows a 15 nm red shift compared to

PAGE 121

105 the single G2-OH dendron.141 This shift is due to the additional phenyl ethynylene unit, which increases the conjugation length in ea ch individual dendron. A better comparison can be made with the steady state spectra of a generation 3 monodendron (Figure 4-2b). In the G3OH monodendron, the longest linear PE chai n has the same number of PE units as the longest linear chain in the 2G2-m-OH, and their absorption spectra show similar features (bands, bandwidth, and red-shif t). There is a slight red shift of G3OH over 2G2m-OH due to some extended conjugation betwee n the two longest line ar PE units through the ortho linkage. To probe the dynamics of energy transf er, EPer is added to the di-dendron molecule in the meta position with respect to both monodendron components. The absorption spectrum of 2G2-m-Per resembles the sum of the absorption spectra of 2G2-mOH and EPer obtained independently as show n in Figure 4-2a. This suggests a weak ground-state coupling between the dendtritic backbone and the EPer energy acceptor. From these spectra we conclude that th e broad absorption feature between 300 and 430 nm corresponds to the dendrit ic backbone, and absorption at > 430 nm corresponds to direct excitation of the EPer trap. The phot oluminescence spectrum s hows emission from 2G2-m-Per originating almost entirely from the EPer trap. At excitation wavelengths between 300 nm and 400 nm there is no direct excitation of EPer, and the emission is solely due to energy transfer from the backbone to the trap. This was proven by exciting 2-phenylethynyleneperylene in CH2Cl2 at selected wavelengths and obtaining no emission at 485 nm (max wavelength emissi on of EPer). We conclude that energy transfer from the backbone to the trap is very efficient. Comparison of absorption and excitation spectra indicates ~94% effici ency for the energy transfer process.97

PAGE 122

106 Interestingly, at 435 nm, we notice a sma ll band with intensity contributions from incomplete substitution of the OH group in 2G2-m-OH97 by EPer, and residual backbone emission from 2G2-m-Per. Time-resolved data will clar ify the origin of this backbone fluorescence band. 300350400450500550600 0.00 0.25 0.50 0.75 1.00 Normalized absorbanceWavelength(nm) 300350400450 0.00 0.25 0.50 0.75 1.00 2G2-m-OH G2-OH G3-OHNormalized AbsorbanceWavelength(nm) (a) (b) Figure 4-2. (a) Normalized absorption spectra of -----2G2-m-OH, …EPer and —2G2-m-Per and fluorescence spectrum of —2G2-m-Per excited at 320 nm. (b) Normalized absorption spectra of — 2G2-m-OH, ….. G2-OH and — G3-OH in dichloromethane. Time-Resolved Emission Experiments The fluorescence decays of 2G2-m-OH and 2G2-m-Per occur on a ns time scale and are measured by Time-Correlated Single Photon Counting (TCSPC). The fluorescence decays for these dendrimers are characterized by a single exponential decay as shown in Table 4-1. The decay for the 2G2-m-Per at 500 nm is in good agreement with that of 2phenylethynyleneperylene emission lifetime at the same wavelength. Detection of 2G2m-Per emission at 450 nm yields a comb ination of the emission lifetime of 2G2-m-Per (measured at 500 nm) and unsubstituted 2G2-m-OH whose fluorescence spectrum has a peak at 450 nm.

PAGE 123

107 Table 4-1. Lifetime measurements from TCSPC excitation (nm) detection (nm) (ns) EPer 450 500 2.36 0.01 2G2-m-OH 370 450 1.83 0.01 2G2-m-Per& 370 450 1.83 (52%) 2.34 (48%) 2G2-m-Per 370 500 2.34 0.01 # Errors correspond to 2 he s were fixed and only the re lative contributions were fitted. The rise times associated with thes e fluorescence decays are measured by the femtosecond time-resolved fluorescence upconversion technique. To understand the excitation delocalization and intramolecular interactions within the di-dendron backbone, we first study the dynamics in th e absence of the EPer trap. The absorption spectrum of 2G2-m-OH has three distinguishable bands peaked at 320 nm, 363 nm and 411 nm. Exciting 2G2-m-OH at selective wavelengths (330 nm, 372 nm, and 415 nm), we seek to probe regions with considerable contributions from each band and explore the possibili ty of assigning the absorpti on band structure to exciton localization. After excitation at the selected wavele ngths, emission is detected at 435 nm corresponding to backbone fluorescence (Fig ure 4-2). Convolution of the instrument response function and the exponential rise f unction reveals a 500 fs time constant (see model below). It is importan t to note that the fitting invol ves the convolution of the IRF with the model functions and therefore the experimental time reso lution is c.a. 150 fs (slightly better then the IRF FWHM).

PAGE 124

108 0 1 1 1 0.00.51.01.52.02.5 0 1 (c) 330 nm (b) 372 nm Upconverted Fluorescence (a.u.) (a) 415 nm Time (ps) Figure 4-3. 2G2-m-OH in dichloromethane excited at a) 415 nm b) 372nm and c) 330 nm. Upconversion signal of fluorescence is detected at 435 nm, the maximum emission wavelength of the molecule. Fitting procedures include the convolution of exponential functions from the kinetic model with the IRF. The best fit is shown as the solid line. Th e longest IRF function is shown in panel c. The bottom panel shows the superpos ition of the experimental data in panels a-c. The surprising result is the lack of ri se time dependence on excitation wavelengths: the three plots in Figure 4-3 show similar rise times (510 20 fs at = 330 nm, 540 30 fs at = 372 nm, and 500 35 fs at = 420 nm). We conclude that the initially excited state must be delocalized throughout the m onodendron and it takes about 500 fs to reach the lowest lying state (the emissive state). As mentioned previously, when the EPer trap is attached to the di-dendron, a strong emission from the EPer unit is observed, whic h indicates efficient energy transfer. To

PAGE 125

109 follow the energy migration from the initially excited state on the di -dendron backbone to the EPer trap, we measure the rise time of the EPer emission. Figure 4-4 shows the temporal evolution of the 2G2m-Per fluorescence as a function of excitation wavelengths. The fits correspond to the convol ution of exponential f unctions with the instrument response function (see kinetic mode l below). The excitation wavelengths are shown as arrows in the absorption spectrum in Figure 4-2a. Detection of the fluorescence is at the EPer maximu m emission wavelength ( = 485 nm). Panel a in Figure 4-4 shows the emission response following direct exci tation of the EPer trap at 465 nm. A single exponential function ( 150 fs) yields a good fit to the experimental data, which provides the time-resolution limit for fluorescence rise times. Any rise time longer than that limit can therefore be attributed to excited-stat e dynamics in the backbone and energy transfer to the EPer trap. Panel b shows excitation at = 420 nm. At this ex citation wavelength, in addition to backbone absorption, there is some residual absorption from the EPer unit. Indeed, there are two mechanisms for th e excitation of EPer acceptor at 420 nm: i ) absorption by the EPer moiety and ii ) sensitized excitation via energy transfer from the backbone. To characterize these two compone nts, we measure the photoluminescence dynamics of a solution containing only EPer in CH2Cl2 after direct excitation at = 420 nm. The difference in fluorescence dynamics of 2G2-m-Per and EPer can be attributed to the backbone-to-EPer energy tran sfer. After considering the direct excitation component, we still find that the energy absorbed by the di-dendron backbone is efficiently transferred to the EPer. This component leads to slower dyna mics observed after excitation at 420 nm (panel b ) compared to excitation at 465 nm (panel a ).

PAGE 126

110 0 1 1 1 1 0.00.51.01.52.02.5 0 1 (d) 340 nmUpconverted Fluorescence (a.u.) (c) 380 nm (b) 420 nm (a) 465 nmTime (p s ) Figure 4-4. Upconversion signal of 2G2-m-Per detected at 485 nm, excited at a) 465 nm, b) 420 nm, c)380 nm and d) 340 nm. The fittings correspond to the convolution of the IRF and the decays shown in Table 3-2. The longest IRF function (at 340 nm) is s hown in panel d. The bottom panel shows the superposition of the experiment al data in panels a-d. At shorter excitation wavelengths, the fluor escence rise times become even longer. Excitations at = 380 nm (panel c ) and = 340 nm (panel d ) clearly show longer rise times for the fluorescence arising from the EPer excited state. The longer rise times correspond to energy-transfer since only b ackbone absorption is observed at these excitation wavelengths and 2G2-m-OH does not display any long rise times at highenergy excitation wavelengths (Figure 4-3).

PAGE 127

111 The steady state fluorescence spectrum (F igure 4-2a) shows that most of the 2G2m-Per emission (94%) derives from the EPer trap. On the blue end of the emission spectrum, a small band is centered at 435 nm which is the same wavelength as the emission from 2G2-m-OH. This band has contribu tion from the unsubstituted 2G2-m-OH (as an impurity) and an additional contribution from backbone emission, even in the presence of the trap. This assertion can be confirmed by examining the time-resolved data detected at = 435 nm (Figure 4-5). The temporal behavior of the fluorescence shows a very fast rise time followed by a fast decay corresponding to energy transfer from the backbone to the EPer trap. This decay time re flects the average lifet ime of the excitation energy deposited in the lowest-lying stat e of the donor backbone. The long time component (ns) seen in Figure 4-5 corresponds to unsubstituted 2G2-m-OH. Using the kinetic model described below, we simulate this signal using the time constants obtained from Figures 4-3 and 4-4 (including a ns component corresponding to 2G2-m-OH emission lifetime). Integrating the time-resolved data yields the rela tive contributions to the steady state spectrum from the 2G2-m-OH impurity and the residual backbone emission. Unsubstituted 2G2-m-OH emission accounts for more than 99% of the steady state intensity at 435 nm (impurity concentration of ~1%).† The fast decaying component corresponds to less than 1% of the steady state fluorescence; at this wavelength it can only be detected with time-resolved expe riments probing in a very short time window. † The purity of the sample was checked by Thin Layer Chromatography. In addition, we performed an experiment by titrating the 2G2-m-Per with 2G2-m-OH, comparing absorption and emission spectra as a function of added 2G2m-OH. We conclude that the spectroscopy experiments are more sensitive than TLC and can detect 1-2% impurity. A concentration of 1% 2G2m-OH yields at least 99% of the integrated fluorescence of the band peaked at 435 nm seen in Figure 4-3.

PAGE 128

112 -1012345678 0.0 0.2 0.4 0.6 0.8 1.0 Upconverted Fluorescence (a.u.)Time (ps) Figure 4-5. 2G2-m-Per Upconversion Signal. excitation=320 nm emission = 435 nm. Solid line is a simulation of the E fluorescence (equation 4) with fix time constants from Table 3-2. Inset shows steady st ate emission spectrum in the region of upconversion detection. See text for details. Time-Resolved Broadband Transient Absorption Measurements Transient absorption spectra of 2G2-m-OH are obtained foll owing excitation at 320 nm. The results are shown in Figure 4-6. At positive time delays, the signal is decaying on a nanosecond time scale and it has a nega tive value throughout the ground state absorption range (325 nm-435 nm) along with a positive signal for > 435 nm. Considering the previous steady state and single photon counting experiments,142 it is known that 2G2-m-OH has quantum yield of 0.81 a nd a fluorescence lifetime of 1.9 ns. Since the excited state absorp tion observed at wavelengths > 435 nm decays on a nanosecond time scale, this is attributed to an S1-Sn absorption. When the ground state abso rption spectrum of the 2G2-m-OH is compared with its transient absorption spectrum at 750 fs after th e excitation, a few differences can be seen (Figure 4-6). In the former, mainly three absorption peaks are observed at 320 nm, 370

PAGE 129

113 nm, and 415 nm. In the latter, the bands at 370 nm and 415 nm are 5 nm red shifted and the intensity ratio between these two bands ha s changed significantly. The origin of these differences between the ground state absorp tion spectrum and the transient absorption spectrum can be explained in the following manner. The steady state spectrum of 2G2-mOH (Figure 4-2) shows that the ground absorption ends at 435 nm, whereas the fluorescence spectrum covers the spectral region from 405 nm to around 590 nm. As a consequence, the negative signal up to 405 nm is fully dominated by ground state bleaching. At the red edge of the absorpti on spectrum (420 nm), the intensity of the negative signal is larger compared to the band at 370 nm. However, the ground state absorption has almost two times smaller in tensity at 420 nm. Thus, this peak has contribution from stimulated emission al ong with the bleach signal. Even though the stimulated emission is contributing to the transient signal in the complete fluorescence range, photoinduced absorption signal dominates beyond 435 nm. The most significant feature of the transient spectra of 2G2-m-OH is the broad bleach signal rising within 300 fs and a more localized band at 420 nm rising independently within 750 fs. Such temporal and spectral behavior elucidates the presence of at least two electronic states within the absorption band, such that the emitting state of the molecule is different than the absorbing state. Moreover, this feature is in accordance with the slow (500 fs) risetime of the emi ssion measured at the blue edge of the fluorescence spectrum. Due to the complete de localization of the absorption band, it is reasonable to obtain the same rise time for any excitation wavelength.

PAGE 130

114 325350375400425450475500 -7.5 -5.0 -2.5 0.0 2.5 EMS ABS Wavelength (nm)A(mOD) t in fs -1000 50 150 200 250 300 600 750 Figure 4-6. Transient ab sorption spectra of 2G2-m-OH at different delay times. Steady state absorption and emission are shown in solid blue and magenta colors, respectively. It is important to note that complex signals from multiple electronic states can be mixed in the transient absorption signals. Even though time-resolved experiments give complementary results, one can argue the pres ence of the different electronic states. Are there really two electronic stat es or are we just observi ng some relaxation process in subpicosecond regime? To confirm our explanat ion, low temperature (77 K) steady state experiments are performed. Recently, it has been shown that the phenyl rings are free to rotate around the ethynylene bonds although the overall struct ure of PE dendrimers are quite rigid. At room temper ature, the torsional barrier is of the same order as k T, and thus the rings rotate freely. At low temperature ( 10 K and 77 K), the rings essentially lie in the same plane. The dipole arrangement of the chromophore system is approaching planarity as the temperature is decreased. This is related to the fact that many different conformations and rotations that are possible for the molecu le at room temperature in solution are not accessible at low temperature in glass forming solvents (like MeTHF).

PAGE 131

115 Figure 4-7a shows the ex citation spectrum of 2G2-m-OH at room temperature (red line) and at 77 K (blue line). In the spectrum meas ured at 77 K, new absorption peaks appear that were hidden in the broader homogenous line width at room temperature. The absorption bands are much sharper and approximately 12 nm redshifted for > 350 nm. The red shift indicates the presence of a more planar geometry as the temperature is decreased. This temperature dependent change of geometry has significant effect on the transition density, which directly effects the en ergy transfer process. Moreover, the peak at 425 nm has more intensity and localized char acter at 77 K, which is very similar to the independent rise of the sharp features in the transient absorption data. To further characterize the interactio ns between the chromophores and the excitation energy transfer dynamics in dendr itic systems, fluorescence anisotropy is a powerful method. Anisotropy measures the orie ntational memory of dipoles. Any change in the transition moment during the lifetime of the excited state will cause the anisotropy to decrease. The depolarization of fluorescen ce might be due to transfer of excitation energy to another molecule with different orientation. Thus, using fluorescence depolarization, it is possible to observe the intramolecular excitation transfer accompanied by reorientation of the transiti on dipole. The different orientation of transition dipole moments are investigated with excitation anisotropy, which allows us to distinguish between electron ic states. Figure 4-7b show s the excitation anisotropy measured at 77 K. There are distinctivel y three different regions in the excitation anisotropy spectrum. The zero anisotropy at < 375 nm indicates an energy transfer from the initially excited state to the final emitting step. The nonzero anisotropy value at 425 nm verifies the existence of a second electroni c state other than the absorbing state. We

PAGE 132

116 also suggest that the region between 375 nm and 425 nm has contributions from both states. These steady state experiments fu rther confirm our conclusion suggesting a localization of excitation be fore emission. The emission would occur from localized states while absorption is into delocalized exciton states. Figure 4-7. The excitation spectrum of 2G2-m-OH at 298 K (red) and 77 K (blue), emission detected at 450 nm (top) The excitation anisotropy of 2G2-m-OH at 77 K (bottom). Kinetic Model for Energy Transfer Both steady state and time-resolved experi ments suggest that the energy transfer efficiency from the dendrons to the trap is nearly unity. To understand the mechanism of this process and the effect of the electronic structure of the backbone, we must estimate the rate of energy transfer fo r each individual step. In this section, we propose a model to understand that process through an interpreta tion of our time-resolved emission data. We consider the simplest case in which D is the ground state of the backbone, D* is the initially excited state and E* is the emissive state of the backbone: 300325350375400425 0.0 0.2 0.4 0.0 0.5 1.0 (b)Anisotropy (r) Wavelength (nm) Intensity (a)

PAGE 133

117 1 **rad BBk h kDDE (4-1) According to this model, the 500 40 fs (k1 -1) rise time measured from 2G2-m-OH must be attributed to one of the two possi bilities. Either the em issive state of the backbone is different from the st ate that is excited initially, or the initially excited state undergoes a conformational rearrangement before emitting. The insensitivity of the rise time to pump wavelength indicates delocalization of the optical excitation within the backbone prior to relaxation. Meta substitution prevents the exciton from delocalizing between the backbone and the Eper unit.127 Since the meta substitution in the core ring breaks the -conjugation, we consider that the delocalization occurs within each monodendron. The addition of the EPer trap results in a new final excited state P*. Emission from the 2G2m-Per arises from the P* and the backbone state E* depending on detection wavelength. This leads to two possible deact ivation pathways for the initially excited state: direct, 3**rad BBk h kDDP (4-2) or indirect, 12***radrad BBEperkk h kkDDEP (4-3) Measurements of the 2G2-m-OH fluorescence probe the E* state directly, whereas experiments with the 2G2-m-Per probe either the EPer emission ( P* ) when detection is at

PAGE 134

118 = 485 nm or the backbone emission ( E* ) from 2G2-m-Per for early-time detection at = 435 nm. By solving the kinetic system of equa tions, we can obtain the population of each excited state with only 2 free para meters in the fitting function ( k2 and k3 are free, while k1 is defined by the previous fitting to data in Figure 4-3). The backbone excited state population is given by: 13 21 213*kkt ktk Etee kkk (4-4) The population of the trap excited state ( P* ) includes two components, indirect (energy goes through the E* state) 13 2* 121 21313213 kkt kt Ikkk ee Pt kkkkkkkk (4-5) with ( k1+k3)-1 and ( k2)-1 as characteristics times, and direct 13() 33 1313 kkt Dkk Pte kkkk (4-6) with ( k1+k3)-1 as the characteristic time. The fluorescence lifetimes of E* and P* are on the order of ns (see Table 4-1), appearing only as a constant offset in the model. The pre-exponential factors are defined in terms of the rates (equations 4-5 and 4-6), and the relative amplitudes of the contributions for each transfer channel are given by k1 and k3. The results are shown in Table 4-2. To fit the data for different excitation wavelengths, we consider the relative contribution to the absorption from the EP er unit and/or backbone at, for example = 420 nm. Allowing this relative contribution to become a free fitting parameter, we obtain

PAGE 135

119 from the time-resolved data that ~ 70% contribution is from straight excitation of EPer and only ~ 30% contribution is from the exc itation via energy transfer from backbone dendrons. Table 4-2. Fits for Time-Resolved Fluorescence Data exc 1 & # (fs) 2 (fs) 3 (fs) Direct % Indirect % 340 nm 510 20 755 60 440 45 54 46 380 nm 540 30 855 60 305 20 64 36 420 nm 500 35 550 150 465 nm 150 20 & Time constants (1) for E* P* are obtained from the time-resolved fluorescence measurement of 2G2-m-OH and kept constant when fitting the 2G2-m-Per data. # Errors correspond to 2 Excitation at 340 and 380 nm (panels d and c in Figure 4-4, respectively) shows different dynamics with longer rise times. At these wavelengths, there is no direct excitation of EPer and the emission is solely due to energy transfer from the backbone to the trap through both direct and indirect paths. This was proven by exciting 2phenylethynyleneperylene in CH2Cl2 at the selected wavelengths and obtaining no emission at 485 nm. Excitation at 340 nm yiel ds the one step energy transfer (Dir) from the initially excited state with 1 3k = (440 45) fs, and the indirect energy transfer (Ind) (two steps) with 1 = (510 20) fs (determined from the 2G2-m-OH data) and 2 = (755 60) fs ( 2). The relative contribution from each path can be evaluated as 3 1 1313k k Ind. = and Dir. = kk kk (4-7)

PAGE 136

120 The results of this procedure suggest that 54% of energy accept ed by the EPer is attributed to the direct path, while the cont ribution from the indirect, multistep pathway is 46%. At 380 nm, the fit yields a fast er rate for the direct path, 3 = (305 20) fs. At this pump wavelength, the contributions from th e direct and indirect channel are 64% and 36%, respectively. As the initially excited state becomes closer to resonance with the EPer transition, the contribution from the direct energy transfer pathway is more pronounced. This kinetic model reve als that for the indirect ch annel, the slowest step is the energy transfer from the rela xed excited state of the donor ( E* ) to the acceptor ( P* ). A confirmation of the presence of the indirect channel is shown in Figure 4-5. This figure presents the data collected at 435 nm a frequency at which there is residual emission from the backbone. In this case, the fast rise and decay times of the fluorescence can be predicted from the fixed values of 1 (from Figure 4-3) and 2 and 3, from Figure 4-4 (including a ns component corresponding to 2G2-m-OH emission lifetime). The agreement between predicted and experimentally measured data is excellent. The rise of the signal is due to ( 1 -1+ 3 -1) and its decay is given by 2, thus it is the energy transfer process to the acceptor EPer that controls the fast decay of donor fluorescence. Comparison of absorption with photolum inescence excitation spectrum indicates that the efficiency of energy transfer for the molecule investigated here is close to unity. When the energy transfer efficiency is n ear unity, measurement of the quantum yield comparing steady state excitation and absorption is not as accurate as using time resolved techniques. We compare the energy transfer rate (in this case from the experimental rise

PAGE 137

121 time in the acceptor1 E Tk) with the donor’s fluorescence lif etime in the absence of the acceptor (o) using 11 1ET ETok (4-8) where ET is the energy transfer quantum yield. Considering 2G2-m-OH as the donor and the EPer as the acceptor, we obtain a ET>0.99 which is in reasonable agreement with the value of ET ~0.94 obtained from steady state measurements. Energy Transfer in the Weak-Coupling Limit Mukamel and co-workers’ calculations using the Frenkel exciton model127 assume that meta substitution confines the electron-hole pair (exciton). The possibility of evaluating the coherent coupli ng between monomers in the 2G2-m-Per is somewhat hindered by the delocaliz ation expected through ortho substituted segments. In the unsymmetrical dendrons, the ortho substitution is expected to have qualitatively similar characteristics as para substitution and thus the electron-hole pair is expected to be delocalized within each monodendron. The meta substitution at the core does not allow the exciton delocaliza tion to go beyond the central phe nyl group, and the exciton is confined to individual monodendrons. Once the excitation is created it can migrate via two path ways (direct or indirect) towards the bottom of the energy funnel. Since both donor and acceptor have allowed optical transitions with strong oscillator strength, the larges t contribution to the coupling will be due to Coulombic in teractions. The magnitude of the Coulombic coupling can lead to incoherent exciton migration for weak interactions or coherent exciton transfer for strong interactions.

PAGE 138

122 Energy migration can be interpreted using the simplest model (Frster) that invokes very weak coupling between two point dipoles.143 The “point-dipole/pointdipole” coupling approximation is valid when the donor-acceptor sepa ration (R) is much larger than the dipole sizes. The very weak coupling is valid when the homogeneous vibronic bandwidth is larger than the dipole-dipole coupling. For the dendritic macromolecule under investigation, we evaluate the validity of each of these conditions independently. The excitation transfer rate in the weak-coupling limit can be calculated from144 2 2 24ETkJV hc (4-9) where V is the coupling and J is the vibronic spectral overl ap integral between donor and acceptor (in units of cm). The spectral overlap is a measure of the density of interacting initial and final states. The accurate calcula tion of the overlap integral employs the homogeneous vibronic bandwidths. Since th e homogeneous bandwidths of the vibronic bands in the di-dendron are unknown, we re ly on absorption and emission spectra to obtain J The spectral density can be evalua ted by using a normalized absorption spectrum -acceptor, a( )-, and normalized emi ssion spectra –donor, f( )-, 27()10f iJcmfanmd (4-10) yielding J = 2.02x10-4 cm. An experimentally measured transfer rate kET =(750 fs)-1 yields an interaction energy V = 75 cm-1. The approximation on the J calculation results on an upper limit for J and therefore, for a given e xperimental rate, a lower limit of the Coulombic interaction.

PAGE 139

123 Donor and acceptor spectra in the 350 nm 450 nm (28,600cm-1-22,200 cm-1) range are very broad. Measuring single -molecule fluorescence, Schryver and coworkers145 estimated vibronic bandwidths between 650-850 cm-1. Even though this number might be an overestimation of the tr ue homogenous bandwidth, it is still very large compared to the small Coulombic intera ction energy, endorsing the use of the very weak coupling model. To understand the role of the dipole size, we again assume the Frster model and compare the result of this model with the experimental values. The critical radius R0 (the distance at which emission and energy transf er occur with equal probability) for the energy transfer step between monodendrons excited state E* and phenyl ethynylene perylene excited state P* can be evaluated from 1/6 2 4 4 00.2108D oDARfd n (4-11) where is the orientation factor ( avg = 2/3), D is the fluorescence quantum yield of the donor in the absence of the acceptor, n is the index of refraction of the solvent (1.4 for CH2Cl2), Df is the normalized fluorescence spectrum, and A is the molar extinction coefficient of the acceptor. The overlap integral (in units of cm6/mol) is evaluated by taking the norma lized emission spectrum of 2G2-m-OH and the absorption extinction coefficient of phenyl ethynylene pe rylene. With these parameter values, we find Ro = 44.25. Molecular Dynamic simulations on thes e macromolecules provide information about the 3-D architecture of the di-dendrons (see Figure 4-1c). Despite what a 2D chemical structure sketch might suggest, the unsymmetrical dendrimers are not planar.

PAGE 140

124 The 3D conformation has a somewhat rigi d Y shape with only the peripheral phenyl groups rotating almost freely around the ethyny lene bonds. These MD simulations allow the evaluation of the moieties’ sizes, yi elding ~15.5 for the monodendron and ~13.5 for the phenyl ethynylene perylene. Th e distance between the center of each monodendron and the center of the phenylethyny lene perylene unit is ~17.2 . If we consider that the size of monodendrons will be on the order of the dipole size, we would expect an energy transfer ra te constant of (6.5 ps)-1. Experimentally, the energy transfer rate constant is (750 fs)-1, clearly indicating that the Frs ter model cannot account for this fast process. The disparity in the calculated versus measured rates is due to the sizes of the chromophores, which are very similar to the dipole-dipole distances. In a point-dipole description, the dipole is assumed to be loca ted at the “spatial” center of each dendron. A point-dipole approximation does not account for the transition dipole density distribution which becomes an important factor wh en dipoles are in close proximity.128 A better treatment of the Coulombic coupli ng must include at least the transition dipole shape and distribution, and can be pe rformed using the Transition Density Cube method,19,146 which provides a numerical approximation to the complete Coulombic coupling. Vectorial Energy Transfer in Unsymmetrical PE Dendrimers The cascade energy transfer is proposed to explain trap fluorescence spectra following excitation of differen t chromophores embedded in dendr imers. To the best of our knowledge, all evidence of the existe nce of energy cascade s are steady state measurements providing only indi rect support for the presence of the cascade. In this chapter and Chapter 3, we present a dire ct experimental measurement of energy migration as it goes through an intermediate st ate in a funnel-type dendritic structure.

PAGE 141

125 Direct confirmation of the presence of the energy gradient can be obtained by following the energy as it migrates through th e individual energy step s down the gradient. The process of rise and decay of the in termediate state population is followed by measuring the temporal evolution of the fl uorescence at the emission wavelength of the intermediate state. This method provides an une quivocal proof of the vectorial nature of the energy transfer in a ladder-type structure. Time-resolved fluorescence measurements of the dendrimer with and without the final energy sink led us to the 3 state (l adder) kinetic model shown in Figure 4-8.132 backbone 1 3 trap I485 400/435 320 Figure 4-8. Model describing the energy ladder. The intermediate state (I) is detected at 400 nm or 435 nm. Emission from trap is at 485 nm. The two time constants involved in the cascade process are =300fs and 2 =350 fs for the 2G1-mPer and =510 fs and 2 = 755 fs for 2G2-mPer. While the model for 2G1mPer contains only the ladder mechanism, the 2G2-mPer data shows that half of the energy transfer process occurs through the la dder process, and the other half proceeds through a direct process ( 3~440 fs). The relative contributions of these two pathways have a slight dependence on excitation wavelength.132 Noteworthy in the model is the prediction of an intermediate state (I in Figure 4-8) consisting of a low-energy state of the bac kbone. The presence of this state leads to

PAGE 142

126 efficient energy transfer afte r excitation at 320 nm by creatin g a ladder-type electronic structure which causes the vectorial energy transfer. The model predicts that the intermediate step in the ladder-type system corresponds to the lowest-lying excited st ate of the backbone. Detecti on of fluorescence at this wavelength (400 nm and 435 nm for the 2G1-mPer and 2G2-mPer, respectively) provides direct confirmation of the cascade mechanism. As predicted, Figur e 4-9 (circles) shows the data obtained after excita tion at 320 nm. For both mol ecules, the fluorescence shows a fast rise time (though slower than our in strument response function), and a slower decay. If the model is correct, one should be able to simulate the data directly without requiring additional fitting parameters. Using the model obtained in the independent measurements of the trap fluorescence, we simulate the population transfer through this intermediate state. The convol ution of the intermediate st ate populations with the IRF yields the solid curv es in Figure 4-9. The simulation shows excellent agreement with the experimental data obtained from the intermediate state. Note that the on ly variable in this simulated curve is the normalization constant (Fmax). Both the decay times and the pre exponential factors are modeled as 13 21 max2131 constant ,kkt ktk Ftee Fkkk (4-12) where k1 = ( 1)-1 k2 = ( 2)-1, and k3 = ( 3)-1 are extracted from the model and the constant corresponds to the 1.8 ns emi ssion lifetime of 2Gn-mOH pres ent in the sample as an impurity. Since the quantum yield for energy transfer is > 0.96, the emission from the

PAGE 143

127 backbone in the presence of the sink is not observed in the steady state emission spectra, as it can only be detected only by using an extremely fast (<6 ps) time window.‡ -10123456 0.0 0.5 1.0 time (ps) Upconverted Fluorescence (a.u.)2G2-m-Per0.5 1.0 2G1-m-Per Figure 4-9. Temporal evolution of the intermediate state population followed by fluorescence up-conversion. The solid lin e corresponds to a simulation of the population of the excited state. The IRF is also plotted. According to the experimental data a nd modeling, the cascade mechanism is present for both generations whereas the dir ect path is available for the generation 2 dendrimer. The energy transfer through a cas cade manner and direct channel will be competing processes, but they also create a complete funnel in term s of transferring the excitation energy to a single accep tor. Despite the fact that 2G1-m-Per molecule is a less complete funnel, it transfers the energy faster than the 2G2-m-Per molecule. This is due to the relatively shorter distances between the donor (longest chain 3-ri ng PE unit) and EPer acceptor. ‡ The purity of the sample is checked by TLC yiel ding better than 98% purity. Our spectroscopic methods are more sensitive than that. The limit of our emission spectrum sensitivity is ~0.5% for the concentration of 2G2-m-OH in a 2G2-m-per solution with the O.D. used in the experiment.

PAGE 144

128 Conclusions We measured the fluorescence dynamics of a phenylethynylene di-dendron with and without an energy trap. The fast transf er dynamics yields a highly efficient lightharvester, with a subpicosecond time scale for energy transfer. Based on steady state spectroscopy, initial studies of unsymmetr ical architectures suggested that the electronic st ructure could be interpreted as the addition of individual building blocks. The time-resolved data presen ted in this work sugge st that the building blocks encompass each monodendron with two po ssible pathways for exciton migration. The presence of a combination of ortho and para substitution leads to initial exciton delocalization within each monodendrons and the meta substitutions confine them there. The electronic structures of the monodendron and the trap are weakly coupled, though a more complete characterization of dipole size an d shape is required to accurately simulate the transfer rates. Linear symmetric dendrimers with phenyl ethynylene units of variable length69 show energy gradients due to the localization of excitons on the different length PE units. For unsymmetrical dendrimers, the ortho and para substitutions suggest a complete initial delocalization within the monodendrons. In the 2G2-m-OH system, delocalization is found in the initially excited state as all results are independent of excitation wavelength. After excitation, a change in th e excited state surface leads to localization and the formation of an energy gradient. Th is excited state ener gy gradient yields efficient multistep energy transfer in the 2G2-m-Per macromolecule. Roughly 50% of the energy transfer occurs through a multistep pathwa y, but the process is still completed in a subpicosecond timescale. We present here the first direct measurem ent of cascade energy transfer in phenyl ethynylen e dendrimers. The presence of gradients makes it possible to

PAGE 145

129 use larger dendrimers for light-harvesting wh ere the intermediate states in the cascade process allow for efficient, unidirectional energy transfer.

PAGE 146

130 CHAPTER 5 ENERGY TRANSFER IN SYMMETRICAL PE DENDRIMER: NANOSTAR Some judiciously designed phenylene ethynylene (PE) dendrimers demonstrate highly efficient and unidirecti onal energy transf er properties.69,95 In chapters 3 and 4, the unsymmetrical PE dendrimers which include variable length of PE segments were investigated to explore the excitation energy dynamics. It was shown that the coupling among PE units occurs through para and ortho substitutions, and this leads to rapidly growing conjugation lengths as the generati on number increases. Therefore, the most novel characteristic of PE dendrimers aris es from the branching pattern at each generational node. As opposed to unsymmetrical dendrimers, subsequent generations are bonded at the phenyl group in meta positions in symmetrical dendrimers. This arrangement plays a crucial role in the excited state electronic structure of the dendrimers. Two series of symmetrical dendr imers have been synthesized in Moore’s group. They differ in the number of PE un its between consecutive branching points and were described as “compact” and “extended”. In compact ones, each generational unit is composed of identical diphenylacetylene (D PA) chains. Compact dendrimers exhibit disruption of conjugation because of meta branch ing and they do not exhibit energy funneling due to the equal ch ain lengths. In the extended dendrimers, the localized excitations between meta branching points are comprised of varying excitation energies which correlate inversely with chain length. Thus, extended dendrimers serve as energy funnels directing energy toward s the dendrimer focal sites.

PAGE 147

131 As shown earlier, the most novel characteris tic of these PE dendrimers arises from the branching pattern. While branching at pa ra positions grows linear chains, branching only at ortho positions terminate the tree structure due to steric hindrance. The meta substitution allows a large de gree of orientational flexibil ity and minimizes the steric hindrance between neighboring units. More impor tantly, the meta subs titution plays a big role in the excited state electroni c structure of the PE dendrimers. The work described in this chapter is based on an extended phenylene ethynylene dendrimer known as the “nanostar” where an et hynylene perylene trap acts as an exciton trap, collecting the photoexcitat ions initially deposited anyw here in the dendrimer. The nanostar has a built-in energy gradient which is engineer ed by linking chromophore units of increasing length (2 -, 3-, and 4-ring) towards the ethynylene perylene trap. This gradient provides a way of overcoming entrop ic effects and decrease s the probability of energy dissipation to the environment. The early experiments on the nanostar show ed that the efficiency of excitation energy transfer from the peripheral groups to the core, monitored via fluorescence measurements, was nearly 100 % efficient. The energy transfer was monitored via fluorescence measurements. Using time-correlated single photon counting, Swallen et al. determined an upper limit of about 10 ps for the energy transfer from the lowest energy chromophore (4-ring) to phenyl ethynylene perylene trap.130 For excitation at the periphery chromophores (2-ring), an upper limit of 270 ps trapping was estimated. Further experiments by Kleiman et al. inves tigated this energy f unneling process in the nanostar for the first time with femtosecond time-resolution.75 Degenerate pump-probe spectroscopy was used to preferentially meas ure the transients of the 2-ring and 3-ring

PAGE 148

132 chromophores. A stepwise energy transfer was revealed from shorter to longer PE units. It was concluded that some of the steps in the energy tran sfer occurs on a subpicosecond time scale. The goal of their study was to explor e the nature of excitations and degree of localization/delocalization as well as the m echanism of energy transfer. Key points are the experimental energy transf er rates and the understanding of the conjugation pattern at the branching point of the dendrimer. Both theoretical and experimental work done so far aimed to clarify the nature of excitations and extent of coupling associat ed with energy transfer mechanism within these dendrimers. However, to our knowledge, the experimental invest igations so far did not answer the question of whether the exci tations at the periphery migrate to the ethynylene perylene through intermediate stat es or there are direct jumps from the periphery to the trap. In other words, are th ere direct and indirect energy transfer paths between the initially excited chrom ophores and ethynylene perylene trap? In this chapter, we examine the nanost ar molecule extensively using femtosecond time-resolved experiments. We present time-resolved fluorescence and broadband absorption studies. Our goal is to record the energy transfer process by varying the excitation energy and monitoring absorpti on transients and emission dynamics. The results indicate the presence of intermed iate steps through the funneling process and direct paths from the initially excited states to the emitting step. Experiments show a slow trapping time for the excitation at the pe ripheral chromophores, while much faster transfer times are measured for 3 and/ or 4 ring chromophores. The time-resolved emission detected both at the trap fluore scence and donor fluorescence wavelengths give consistent results and prove the presen ce of direct jumps within the cascade.

PAGE 149

133 Another challenging i ssue with dendrimers is determin ation of the geometry of the molecule that can cause large variat ions in energy transfer properties.19,81 Our collaborators Krause and Roitberg groups at UF have recently performed large scale molecular dynamics (MD) simulations for the nanostar to obtain structural information.74 The interparticle interactions were modeled with an MM3147-149 force field (TINKER, version 3.9) for the minimization and subse quent dynamics simulation. They found that while the overall structure of the nanostar is rigid, the phenyl rings not involved in branching are free to rota te around the ethynylene bonds. Th e distribution of torsion angles for the center ring of the three-ring chromophores was calcula ted at two different temperatures, 10 K and 300 K. The results indica ted that the rings li e essentially in one plane at low temperature whereas they rotate freely at room temperature. This change in geometry was proven to have a significant effect on the transition density, and hence energy transfer rates.74 Materials and Methods The synthesis of the nanostar has been described elsewhere34,35 and its chemical structure is shown in Figure 5-1. For the spectroscopic measurements, the sample is dissolved in dichloromethane where there is no excimer formation. The optical density used for transient absorption and fluo rescence measurements is about 0.3 mm-1. The steady state absorption of the sample was compared before and after measurement. Steady state characterization is performed via UV-Vis abso rption (Varian-Cary 100) and emission in a Jobin-Yvon instrument (Spex Fluorolog-3). The laser system, fluorescence upconversion and transient absorp tion setup were described in detail in Chapter 2. Here we give a brief description of experiments performed.

PAGE 150

134 Transient Absorption Absorption difference spectra are reco rded with a homemade pump-probe apparatus, described in detail in Chapte r 2. In brief, the output of Ti-Sapphire Regenerative Amplifier (1kHz, 800 nm, and 60 fs pulses) is used to pump an optical parametric amplifier (OPA, Spectra Physic s). Pump beams at UV region with tunable central frequency in the UV region are pr oduced in an OPA via fourth harmonic Figure 5-1. Chemical structure of nanos tar dendrimer (2 dimensional sketch). generation of the signal. Prism compression is used to reduce the pulsewidth of these UV-pulses. The excitation beam is focused on a spot with 300 m diameter. The energy of excitation is 30 nJ/pulse upon UV excita tions. Broadband probe and reference beams are generated via focusing a small fract ion of amplifier ou tput on a 1mm CaF2 plate Using a thin-film polarizer, the probe light polarization is orient ed 45 degrees with respect to the pump pulse. Af ter passing through the sample, a Glan-Taylor polarizer splits the probe beam into its polarization co mponents, parallel and perpendicular with respect to the pump, allowing for the simulta neous detection of bot h polarizations. Pump

PAGE 151

135 induced absorption changes of both probe pol arization components are measured as a function of pump-probe time delay by modulat ion of the pump beam with a mechanical chopper and detection of the probe and a re ference beams using a CCD camera equipped with a 30 cm spectrograph. The time resolution is typically 150 fs (for 310 nm, the time resolution is around 200 fs). The data are corrected for white light gr oup velocity dispersion and fitted globally including the convolutio n with the instrument response function. The optical path of samples is 2 mm and the sample is stirred with a magnet in order to refresh the sample from shot to shot. The integrity of the samp le was checked before and after each set of measurements. Time-resolved Emission Time-resolved fluorescence emission spectra are recorded with a home-built Fluorescence Upconversion setup described in detail in Chapter 2. Briefly, the pump pulses are generated within the OPA. The excitation wavelength is 310 nm, and the emission from the intermediate state of th e donor and the final acceptor are detected at 380 nm and 485 nm, respectively. The gate b eam is a residual 800 nm beam from the OPA. The excitation energy is 30 nj/pulse a nd the excitation beam is focused on a spot with 200 m diameter. Fluorescence is collected at magic angle with respect to the polarization of the excitation b eam. The sample is placed in a spinning cell with optical path of 1 mm. The time resolution of the sy stem is measured by detection of crosscorrelation of scattered light from solvent and gate pulse. It is about 250 fs for the 310 nm excitation pump pulse (even t hough prism compression is used it is hard to compress UV pulses due to much narrower bandwidth).

PAGE 152

136 Steady State Spectroscopy The absorption and emission spectra of th e nanostar in dichloromethane at room temperature is shown in Figure 5-2. The emi ssion is almost entirely from the ethynylene perylene trap, indicating 98 % effi cient excitation energy transfer.86 Theoretical and experimental studies assign the peaks/shoul ders near 310 nm, 352 nm, and 372 nm to the vibrationless excitonic bands of 2-, 3-, and 4ring linear chromophores, respectively.72,75,76,86,150 300400500600700 0.00 0.25 0.50 0.75 1.00 Fluorescence Absorption(Norm)Wavelength (nm) Figure 5-2. Absorption and emission spectrum of nanostar in DCM at room temperature. Arrows represent the peaks/shoulders at 310 nm, 352 nm and 372 nm. The absorption data for all ethynylene perylene substituted dendrimers are spectrally identical to their unsubstituted dendrimer pa rents, with an additional absorption peak around 450 nm due to perylene moiety. He re, this indicates that phenylethynylene perylene unit has a well localized excited state, and does not perturb the electronic structure of the dendritic backbone. The high energy peak in the absorption spectrum of the nanostar at 310 nm is due to DPA chains found at the periphery of the molecule. The

PAGE 153

137 longer PE chains in the interior of the molecule give rise to well-defined lower energy peaks. In order to investigat e the idea of electronically decoupled PE units, due to meta conjugation, the absorption spectrum of each PE unit (2-, 3-, 4ring) are measured and its sum is compared with the nanostar abso rption. Figure 5-3a shows the absorption spectrum of independent free 2-, 3-, 4ri ng PE units. The addition of these spectra, normalized to their relative number present in the nanostar, is shown in Figure 5-3b together with the experimental nanostar spectrum. These two spectra are remarkably similar. The only difference is a12 nm overall sh ift, probably due to a local change in the solvent’s dielectric constant. By comparing the features of individual PE rings’ spectra to the features in the sum, it is reasonable to assign the region peaked at 310 nm to 2-ring absorption, the 352 nm shoulder to mostly 3-ring absorption, and 372 nm shoulder to 4ring absorption. This can be confirmed by measuring the low temperature (10K) absorption spectrum of the nanostar.* As shown in Figure 5-3c, the low temperature absorption spectrum of the nanostar has sh arp bands correspondi ng to the planar configurations of the 2-,3-, 4-ring units. A ltogether, these spectra reveal that in the ground state the 2-, 3-, 4-ring units are electronically decoupled. Transient Absorption Spectroscopy Time-resolved transient difference absorpti on spectra of nanostar and its individual components are measured. Our aim is to compare the dynamics of individual components with the vectorial energy transfer dynamics in the nanostar. The broadband transients of the nanostar are measured after excitati on at 310, 352, and 372 nm, as indicated by arrows in Figure 5-2. These wavelengths are chos en to excite preferentially the 2-ring 3* Personal communication from Joseph S. Melinger from Naval Research Labs

PAGE 154

138 300325350375400 0.0 0.2 0.4 0.6 0.8 1.0 Normalized AbsorptionWavelength(nm)(a)300325350375400 0.0 0.2 0.4 0.6 0.8 1.0 Normalized AbsorptionWavelength(nm)(b)12 nm red-shifted 300350400450500 0.0 0.2 0.4 0.6 0.8 1.0 Normalized AbsorptionWavelength(nm)(c) Figure 5-3. Normalized absorption spectrum of (a) 2-ring (black), 3-ring (green), and 4ring (red) PE units (b) Nanostar absorp tion (red) and sum of rings’ absorption at 298 K (c) nanostar absorption at 298 (blue) and 10 K (red).

PAGE 155

139 ring, and 4-ring chromophores. However, the ex citations are not highl y selective due to overlap of vibronic transitions associated with the next longest chromophore species as shown in Figure 5-3. Results obtained from the nanostar are compared with those of model compounds: DPA excited at 305 nm a nd phenylethynylene perylene excited at 400 nm. Model Compound DPA The highest peak in the absorption spectrum of nanostar at 310 nm is mostly due to absorption of peripheral 2-ri ng chromophores. To discuss th e dynamics of the nanostar when excited at 310 nm, it is extremely in formative to look into the excited state dynamics of diphenylacetylene (DPA) mono mers. Hirata et al. investigated photophysical properties of DPA in vari ous solvents by pico second time-resolved absorption measurements.151 When DPA in n-hexane was ex cited at 295 nm, a short-lived absorption band at 500 nm was attributed to the higher excited singlet state S2. It was concluded that the emitting state of DPA was not the lowest excited singlet state but the short-lived S2 state. The lifetime of S2 state was estimated to be about 8 ps. Thus, S2 state is initially populated and it can return to the ground st ate through mainly nonradiative relaxation, or undergo internal conversion to S1 state. Then a triplet state T1 is formed via the intersystem crossing from S1 with a time constant of about 200 ps. In this chapter, we present the spectral and temporal evolution of DPA in dichloromethane after excitation at 305 nm (Figure 5-4). The steady state fluorescen ce spectrum peaked at 325 nm is also shown for comparison. Figure 5-4 shows very little stimulated emission starting at 320 nm and a very broad phototoinduced ab sorption band around 410-450 nm. This band can be assigned to a Sn S1 transition of DPA with a lifetim e of 200 ps. At 500 nm, there is

PAGE 156

140 a well distinguished photoinduced absorption band decaying with an 8 ps lifetime. This is in well agreement with the previ ously measured lifetime of S2 state shown by Kleiman et al.75 In Figure 5-4, bottom panel, we pres ent the dynamics at three particular wavelengths to understand better the DPA data In the 320-330 nm region, the initial stimulated emission Figure 5-4. Transient absorption spectrum of model compound DPA(top), transient absorption signal as a function of time r ecorded at 3 different wavelengths: 315 nm (red), 473 nm (blue),500 nm (black) (bottom). (decreased absorption) is observed. In the 432-442 nm region, we observe the photoinduced absorption (PIA) which has a very long (hundreds of picoseconds) decay time, and finally the averaged signal betw een 495 and 503 nm shows the photoinduced 05101520 -1 0 1 2 3 4 325350375400425450475500525550575 -1 0 1 2 3 4 500 nm 315 nm 437 nm A (mOD)Time(ps)A (mOD)t in ps 0.00 0.45 0.75 1.50 3.50 13.5

PAGE 157

141 absorption of the S2 state with an 8 ps decay time. We will try to compare the dynamics of DPA with the nanostar dynamics when ex cited at 310 nm where 2-ring chromophores are excited initially. Model Compound Phenylethynylene Perylene It was proven that the energy gradient i nherent in the structure of nanostar is sufficient to supersede the entropic bias to wards the periphery and absorbed light-energy is efficiently funneled to the perylene moiety.70 When the excited state lifetime of phenylethynylene perylene is measured followi ng direct excitation it shows a precise agreement with the nanostar trap moiety. Both nanostar and isol ated phenylethynylene perylene decay with a lifetime of 2.2 ns Thus, the electronic conjugation between the dendrimer backbone and the EPer trap is broken by the meta bonding resulting in localized excitations. The knowledge of excited state dynamics of the isolated phenylethynylene perylene will be a reference po int to discuss the energy transfer arising from excitation into dendritic states. We measure the transient absorption spectra of phe nylethynylene perylene in dichloromethane .The excitation wavelength wa s set to 400 nm and the resulting spectra are shown in Figure 5-5. At positive times, two different components can be observed. A broad positive signal extends from 325 nm to 415 nm. This feature is seen instantaneously after excitati on and it decays with a nanosecond time scale. Since this broad signal is positive, it can be attributed to excited state absorption. Steady state absorption shows that the S2 state of the phenylethynylene perylene absorbs below 350 nm, we attribute this broad PIA to the Sn S2 absorption.

PAGE 158

142 425450475500525550575 -5 -4 -3 -2 -1 0 1 2 Wavelength (nm) A(mOD)t in ps 0 0.5 1.5 5.0 7.5Absoprtion Emission Figure 5-5. Transient absorption spectr um of the model compound phenylethynylene perylene. The steady state absorption and emission spectra are also shown. A negative signal starting at 415 nm can be also seen following excitation. The steady state absorption spectrum ends at 500 nm and the fluorescence spectrum extends from 450 nm to 750 nm. Thus, the negative sign al in the transient absorption spectrum can not solely be associated with ground st ate bleaching. The negative peak centered at 440 nm corresponds to the bleach signal of the 1 0 vibronic transition of the ground state. The maximum of the negative si gnal is centered around 475 nm. Since this wavelength is neither at the maximum of the ground state absorption band nor at the maximum of the steady state fluorescence, we assign this peak to the sum of ground state bleaching and stimulated emission. As st ated before, the transient signal from phenylethynylene perylene decays mainly on a nanosecond time scale. However, a kinetic analysis of the transient absorption intensities as a function of delay time for wavelengths around 475 nm reveals an additi onal picosecond relaxation process. The intensity at 465 nm decays with a 2-3 ps time constant as the intensity at 480 nm rises with the same constant. The peak at 510 nm and the shoulder around 550 nm are only due

PAGE 159

143 to stimulated emission since they are furthe r red compared to any ground state absorption band. Nanostar To follow the energy migration from the initially excited state on the nanostar backbone to the perylene trap, transient absorption spectra following excitation at different wavelengths are monitored. The exci tation wavelengths are selected to be 372 nm (4-ring PE), 352 nm (3-ring PE), and 310 nm (2-ring PE) and are shown as arrows in the absorption spectrum in Figure 5-2. 372 nm excitation. The nanostar in dichloromethan e is excited at 372 nm with a 100 fs pulse (FWHM). The transient spectra in the 301-582 nm spect ral range are shown in Figure 5-6 for several pump-probe de lays. The steady state absorption and fluorescence spectra are also shown in the fi gure. Looking into the absorption spectrum of the 4-ring PE unit and its contribution in the whole nanostar molecule, excitation at 372 nm is mostly resonant with the 4-ring ch romophores, the closest chains to the trap. Thus, dynamics associated with this excitation can be directly attributed to the energy transfer from the lowest state of th e dendrimer (4-ring ch romophore) to the ethnyleneperylene trap. Looking at the time-resolved data, there ar e two distinct tempor al windows (Figure 5-6). Following the excitation pulse at 372 nm and for time delays shorter than 350 fs, the differential optical density ( OD) is negative in the whol e spectral range (Figure 5-6a). At wavelengths shorter than 400 nm, the sp ectra are dominated by the bleaching of the dendrimer absorption band peaked around 375 nm For wavelengths longer than 400 nm, the negative OD values are due to bleaching signal from the ethynylene perylene trap.

PAGE 160

144 350400450500550 -4 -3 -2 -1 0 1 (b) A(mOD)Wavelength(nm)t (ps) -1.000 0.450 0.550 1.250 10.00 50.00 -4 -3 -2 -1 0 1 2 EMS A(mOD)(a)t (ps) -1.000 0.050 0.150 0.250 0.350ABS Figure 5-6. (a) Transient absorption spectrum of nanostar within 400 fs after excitation at 372 nm. (b) Long time window (0.450-50 ps). Scattered light from the excitation beam is seen at 372 nm. At this wavelength, there is some direct exc itation of the EPer, and thus the bleach signal between 400 nm and 475 nm is observed. For time delays between 450 fs and 5 ps, a fast decrease of the bleach signal for >375 nm occurs simultaneously with the rise of a photoinduced absorption band around 385 nm. At longer wavelengths, we can see that the negative signal is a combin ation of bleach and stimulated emission. At even longer delay times ( t>5 ps), the red side of the spectrum ( >420 nm) evolves to resemble to that of phenylethnylene perylene. Between 420 and 475 nm, the bleaching of perylene absorption is dominant whereas above 475 nm, stimulated emission is the only

PAGE 161

145 contribution. As discussed before, the peak around 480 nm is a combination of both bleach and stimulated emission signals. In order to determine the kinetics follow ing the 372 nm excitation, the transient signals at two different wave lengths are shown in Figure 5-7. At 340 nm, we are solely probing the ground state bleach. This signal ap pears within the instrument response function and its subpicosecond decay represen ts the recovery time of the ground state population in the 4-ring component of the dendrimer. At 515 nm, only the stimulated emission from the ethynylene pe rylene trap is probed. This signal rises with the same time constant observed for the decay of the dendrimer backbone bleach. This shows that the energy transfer from the 4-ring PE segmen t to the perylene trap is very fast and efficient as the 4-ring PE is the closest to the perylene unit both spatially and energetically. -101234 -2.0 -1.5 -1.0 -0.5 0.0 0.5 A (mOD)Time(ps) 340 nm 515 nm Figure 5-7. Transient absorption signal as a function of time recorded at 340 nm (black), and 515 nm (blue) following excitation at 372 nm. 352 nm excitation. Although the 352 nm excitation is initially considered to be specific to the 3-ring PE, the absorption studies illustrated in Figur e 5-3a and 5-3b show

PAGE 162

146 that the 4-ring PE also cont ributes to the tota l absorption. The 352 nm excitation is therefore resonant with both the 3-ring a nd 4-ring PE chromophores. Figure 5-8 shows the pump-probe spectra after excitation at 352 nm which has similar characteristics to the 372 nm excitation spectra. The lack of bleach signal in the region of EPer absorption band (400-475 nm) demonstrates that at th is excitation wavelength only the dendrimer backbone is excited, without contribu tion from direct excitation of EPer. At time zero, the 352 nm excitation pulse promotes a 3-ring chromophore within the nanostar from its ground state to the S1 state. Therefore, imme diately after excitation the transient spectrum of the nanostar resemb les the sum of absorption spectrum of 3-ring and 4-ring PE chromophores. The excited 3-ri ng (or 4-ring) PE chromophore transfers its energy into the nearest 4-ring and then to th e ethynylene perylene trap. At very early times ( t<350 fs), the OD is negative for wavelengths shorter than 480 nm. For > 480 nm, there is a positive signal which is attrib uted to the photoinduced absorption of the 3ring. This broad PIA is consistent with Beeby et al.’s experiments on para phenylene ethynlene (3-ring namely 1,4-bis(phenylethynyl) benzene) and with our results from the transient absorption of 3-ring and 4-ring PE units.152 Such a positive signal is absent (for 300 nm < >575 nm) when the molecule was excited at 372 nm. When the free 4-ring molecule is excited, there is no photoinduced absorption for < 575 nm, which does not exclude the possibility of PIA signal being red-shifted. At longer times, this signal is masked by the stimulated emission from the et hynylene perylene trap and thus the overall signal in this region be comes negative. After a few picoseconds, the transient spectrum of the nanostar resembles that of free ethynylene perylene. Figu re 5-9 shows the temporal

PAGE 163

147 350400450500550 -4 -2 0 2 (b) A(mOD)Wavelength(nm)t (ps) -1.000 0.550 0.650 1.000 1.400 3.000 50.00-6 -4 -2 0 2 EMS A(mOD)(a)t (ps) -1.000 0.000 0.050 0.150 0.250 0.400 ABS Figure 5-8. Transient absorption spectrum of nanostar after excitation at 352 nm. (a) Short time window ( t< 450fs). (b) Long time window (0.550-50 ps). The scattered light from the pump beam is observed at 352 nm. behavior in two wavelength regions: bleachi ng signal of the dendrimer ground state (360 nm) and stimulated emission signal from th e ethynylene perylene acceptor (520 nm). The bleach signal rises instantaneously with the excitation, and decays within a couple of picoseconds. The stimulated emission signal ri ses with the same time constant as the bleach decays. A good correlation is observed between these two signals, indicating the vectorial energy transfer within the nanos tar molecule. The energy transfer to the perylene trap at 352 nm excitation wavelength also proves to be an efficient and fast process.

PAGE 164

148 -10123456 -6 -4 -2 0 A (mOD)Time (ps) 360 nm 520 nm Figure 5-9. Transient absorption signal as a function of time recorded at 360 nm (black), and 520 nm (red) following excitation at 352 nm. 310 nm excitation. With a pump pulse spectrum cen tered at 310 nm, excitation energy is deposited only on the 2-ring units, wh ich also results in energy transfer to the EPer trap. Figure 5-10a shows the pump-pr obe spectra along with the steady-state absorption and emission spectra. In Figure 5-10a, within the first 550 fs, the evolution of the bleach signal between 300 nm and 380 nm ha s better defined features resembling the steady-state absorption. Even though the initia l excitation is deposited on the periphery, the absorption peaks associated with 3-ri ng and 4-ring PE units are observed in the transient spectrum proving the presence of cas cade energy transfer. In this time window, an instantaneous photoinduced absorption signal peaked at = 380 nm is observed. It is reasonable to compare this signal with the absorption of the DPA molecule shown in Figure 5-4a. DPA has a broad photoinduced abso rption signal peaked at 500 nm with a lifetime of 8 ps. The signal shown in Figure 5-10a matches well with the signal obtained from the standalone DPA molecule. The instantaneous photoinduced absorption for > 425 nm resembles the

PAGE 165

149 -20246810 0 2 (b) A(mOD)Time (ps) 352 nm 500 nm300350400450500550 -4 -2 0 2 EMS A(mOD)(a)t (ps) -1.000 0.050 0.250 0.550 1.000 3.000 50.00 ABS Figure 5-10. (a) Transient ab sorption spectrum of nanostar after excitation at 310 nm. There is scattered light from 310 pump beam (b) Transient absorption spectra as function of time recorded at 352 nm and 500 nm. excited state absorption of DPA. This indicates that the exc itation at 310 nm is localized on the peripheral 2-ring units. The bleach signal between 300 nm and 380 nm corresponding to the dendrimer backbone absorption decays within 20 ps which is considerably slower than the previous two bleaching signals of the 4-ring and the 3-ring excitations at 372 nm and 352 nm, respectively. Simultaneously, the stimulated emission signal from the perylene trap peaked at 485 nm rises to its maximum value. At t= 50 ps, the transient spectrum has almost evolved into the spectrum of phenylet hynylene perylene. Here, it is obvious that

PAGE 166

150 depending on the initially excited state, en ergy transfer can progress from a few hundreds of femtoseconds to tens of picoseconds. Figure 5-10b shows the decay of the bleach signal detected at 352 nm and the rise of the stimulated emission sign al detected at 500 nm. The results obtained with broadband tran sient experiments are quantitatively in good agreement with the previous degenerate pump-probe investigation by Kleiman et al.75 One way to directly compare the rates of energy transfer is by looking at the rise time of the ethynylene perylene’s stimulated emission. Figure 5-11 shows the kinetics of stimulated emission signal detected at 485 nm (maximum emission wavelength of perylene) following different excitation wavele ngths. Since each of this excitation is proven to electronically access a specific chromophore in the nanostar, different dynamics are expected, specifically slower ki netics for periphery excitation and faster dynamics for longer segment excitation. Th e transient stimulated emission signal following excitation at 352 nm and 372 nm have similar time scales for the energy transfer (the risetime of 372 nm excitation is slightly fast er than the 352 nm excitation, but both are in subpicoseconds regime). Howe ver, the risetime of the emission following the 310 nm excitation is consid erably slower, indicating a sl ow energy transfer from the periphery to the core. Another important poi nt we can derive from Figure 5-11 is the initial positive signal observed only for the 310 nm excitation. The transient of the model compound DPA showed a significant photoinduced absorption signal peaked at 500 nm. Thus, this initial positive signal is due to th e localization of excitation on the 2-ring PE units.

PAGE 167

151 05101520 -1.00 -0.75 -0.50 -0.25 0.00 0.25 normalized)Time(ps) 310 nm 352 nm 372 nm Figure 5-11. Transient absorption signal as a function of time for three different excitation wavelengths. Detection is at 485 nm, the maximum stimulated emission from the perylene trap. Kinetic Model for Nanostar Both steady state and transient absorpti on experiments suggest that in the ground state there is no coupling between the PE se gments. The meta branching results in a nearly complete loss of conjugative deloca lization between the neighboring PE units. The photoinduced excitons are init ially localized on a single ch romophore unit. Following the excitation of an individual segment, the ex citon will funnel thr ough the accessible energy states and radiatively decay from the ethynyl ene perylene trap (core). To understand the kinetic phenomena associated with this process, we need to determine the rate of energy transfer for each individual step. Here, we propose a kinetic model to quantify and account for the differences in the transfer ra tes at three excitation wavelengths, namely 372 nm, 352 nm, and 310 nm. Figure 5-12 illu strates all possible paths yielding the vectorial transfer to the core of the molecule A, B, and C represent the excited states 2ring, 3-ring, and 4-ring units, respectively. P st ands for the excited st ate of the ethynylene

PAGE 168

152 perylene trap. When the kinetic equation repr esenting the population of each excited state is solved, there are up to fi ve exponential functions. It w ould be too complicated and impractical to comprehend the physical meani ng of these exponential fits for the timeresolved data. Therefore, we will start with the simplest case of kinetic process, obtain the fitting parameters, and iteratively use th ese as constraints to analyze more complex transfer processes. For instance, the simplest process is the one step energy transfer from the 4-ring PE unit into the perylene trap when the nanostar is excited at 372 nm (k3). While analyzing the time-resolved data for higher energy excitation, the rates obtained from the fit to the lower energy excitations will be incorporated. By doing so, we reduce the number of variables and the data fitting becomes more informative. Model for 372 nm Excitation When the nanostar is excited at 372 nm, both steady state and transient absorption data revealed that only 4-ring units are init ially excited. In Figure 5-12, D is the ground state of the dendrimer backbone, C is the excite d state of 4-ring unit, and P* is the excited state (emitting state) of th e ethynylene perylene trap. 3 *radk h kC DP (5-1) According to this model, k3 is the energy transfer rate from the dendrimer to the ethynylene perylene trap. The ethynylene pe rylene trap’s radiative decay rate krad was measured via time resolved emission experiments and found to be (2.2 ns)-1. While analyzing the broadband tran sient absorption spectra, on e needs to know the population

PAGE 169

153 Figure 5-12. Kinetic Model proposed for the dynamics of nanostar. of each excited state. By solving the kinetic equations for this model, the excited state population of 4-ring unit is given by: 3ktCte (5-2) Then, the excited state popul ation of the ethynylene pery lene trap follows as: 3* 3 3()()radktkt radk Ptee kk (5-3) The data measured here extend over a broa d spectral range and over several orders of magnitude in time. The data sets are evaluated through Single Value Decomposition (SVD) method as explained in details in Appe ndix B, and the kinetic model is applied to fit both spectral and temporal evolution of th e energy transfer process. The SVD analysis, reconstruction of the data in spectrum, and ki netic fits are shown in Appendix B. Since the krad -1 is directly measured to be 2.2 ns, the only free parameter to be determined is k3. Using equations 2 and 3, the energy transfer time from the lowest excited state of the Ethynylene Perylene Dendrimer backbone S0 Sn P* A D B C P k1k2k3k4krad k5

PAGE 170

154 dendritic backbone (C) to the ethynylene peryle ne core (P*) is found to be very fast (k3 -1= 290 fs). This is an excellent agreement with the value obtained by Kleiman et al. from degenerate pump-probe experiments.75 Model for 352 nm Excitation Exciting the nanostar at 352 nm leads to initial excitati on of 3-ring units. As shown in Figure 5-12, B represents the excited st ate of 3-ring units. There are two possible pathways that will transfer the energy from this initial state into the ethynylene perylene: direct, 4 *radk h kDBP (5-4) or indirect, 3 2 *radk k h kC DBP (5-5) As shown in this model, k2 is the rate of interchrom ophore energy transfer within the cascade (from 3ring to 4-ring) while k4 is the direct transfer rate. The excited state population of 3-ring units is derived as: 24()()kktBte (5-6) The population of the intermediate step C (e xcited state of 4-ring unit) is given by: 24 32 324 kkt ktk Ctee kkk (5-7) Then, the population of trap excited state (P*) includes tw o components, indirect where energy goes through the st ate C (cascade step)

PAGE 171

155 24 3* 23 2 32424324 kkt kt Ikk k ee Pt kkkkkkkk (5-8) with (k2+ k4)-1 and (k3)-1 as characteristic times, and direct 24() 44 2424 kkt Dkk Pte kkkk (5-9) with (k2+ k4)-1 as the characteristic time. Since the fluorescence lifetimes of B, C, and P* are on the order of ns (obtained from the free molecules), they are included as constant offset terms in the model. Defining the pre-exponential factors in terms of the rates, and using the k3 obtained from the 372 nm excitation analysis leads to k2 and k4 as the only free parameters. Using equations 5-9, fitting of the data sets yields direct transfer time of (k4)-1 = 4 ps, and the indirect transfer with (k2)-1 = 500 fs. The relative contribution from each path can be evaluated as: 24 2424kk Indirect = and Direct = kk kk According to this model, 89% of energy tr ansferred to ethynylene perylene trap is attributed to indirect, multistep path (fro m 3-ring to 4 ring and then to the ethynylene perylene), while the direct path contributes only 11% As a result, the proposed model here implies that most of the energy is flowing through the cascade which includes step by step interchromophore energy transfer. Even though, only a small percentage of the initially deposited energy is flowing thr ough the direct path, we can not neglect its contribution. The rate constants obtained from this anal ysis will help to understand the degree of coupling between different size PE chromophores and the mechanism of the energy

PAGE 172

156 transfer. For instance, confirmation of the cas cade transfer when nanostar is excited at 352 nm, definitely eliminates the strong coupling of 3 and 4 –ring units. If they were coupled it would have been possible to fit th is data set with a si ngle step mechanism as in the case of 372 nm excitation described in the previous section. The extent of coupling/localization and energy transfer mechan ism for the nanostar will be discussed in more details in the upcoming sections of this chapter. Model for 310 nm Excitation 2-ring chromophores in the na nostar have the highest exc itation energy and they are electronically accessible via 310 nm excitation. All the possible paths shown in Figure 512 will be available upon this excitation. Solving the equations for this model and implementing them into the fitting procedur e will be quite complicated. From the previous analysis performed for 372 and 352 excitations, the populat ion dynamics of 3ring (state B) and 4-ring excited states (state C) are explored in detail. Here, we need to concentrate on the population dynamics of 2-ring excited state A. In the same manner, following the excitation at 310 nm, the populat ion on state A can flow through a cascade or can transfer directly to the ethynylene perylene trap. For the simplicity of the fitting procedure, the overall model can be reduced to a direct path: 5 *radk h kDAP (5-10) and the cascade path: 3 12 *radk kk h kC DABP (5-11)

PAGE 173

157 Since 2k k3 and krad are already known, only two parameters, k1 and k5 will need to be determined. We will use an effective 2k value composed of k2 and k4 measured in the previous subsection. According to this model, the excited state populations are given by: 15()()kktAte (5-12) 15 21 215 kkt ktk Btee kkk (5-13) 15 2 3() ** 1212 *** 315 2153231532() ()()kkt kt ktkkkk ee Cte kkk kkkkkkkkkk (5-14) and 15 2 3* () 123 12 **** 15315 21523231532* ()(()) ()()()kkt kt ktkkk kk ee Pt e kkkkk kkkkkkkkkkk (5-15) Implementing these equations into the fit ting program we obtain transfer times of (k1)-1= 10 ps and (k5)-1= 20 ps. These energy transfer tim es suggest that 66% of the energy is transferred through the multistep mechanism while 33.3% of the energy can directly go to the ethynylene pery lene trap. At this point it is important to r ecall that this model is a simplified version of what we really proposed. The 33.3 % of the energy transfer actually represents all the paths that are not included in the cascade mechanism. It is possible that 2-ring units might inte ract with both 4-ring units and ethynylene perylene trap other than just with 3ring as the cascade mechanism suggests The overall results of the kinetic model an alysis are summarized in Table 5-1. The most crucial conclusion is the slow energy tran sfer time when nanostar is excited at 310

PAGE 174

158 nm. Even though energy transfer times are in subpicosecond regime from the intermediate chromophores, it takes 10 ps from the peripheral 2-ring PE units. Our results are in accordance with the relaxation dynamics of the same molecule measured recently via degenerate pump-probe experiment.75 To our knowledge, this is the most extensive experimental investigation of the energy f unneling process in the nanostar providing the time scale for each of the individual steps through a multi-step mechanism and direct jumps from the periphery to the ethynylene perylene trap with femtosecond temporal resolution. For a complete picture of this vectorial energy transfer we also performed time-resolved emission experiments. Table 5-1. Fits for Transient Absorption Data exc 1 (ps) 2 (fs) 3 (fs) 4 (ps) 5 (ps) Direct % Indirect % 372 nm 290 20 100 352 nm 500 40 290 4 0.2 11 89 310 nm 10 1 550 (1/* 2k ) 290 4 20 3 33.3 66.6 # Errors correspond to 2 Time-Resolved Emission Experiments The steady state photoluminescence spectrum shows emission from the nanostar originating almost entirely from the ethynylen e perylene trap, and its efficiency is independent of the excitation wavelength. The energy transfer quantum yield can be estimated by comparing the absorption sp ectrum with the excitation spectrum while monitoring emission from the acceptor. In fact, the nanostar’s normalized excitation spectrum detected at 515 nm (perylenic emission region) and absorption spectrum are compared .The efficiency of energy transf er is quantified to be 98% within this molecule.70

PAGE 175

159 If we take a closer look into the emission region around 380 nm, a small band consisting of emission from un-functionali zed (without ethynylen e perylene trap) dendrimer backbone and some residual dendrit ic emission (even in the presence of the trap) from nanostar is noticed. The main emi ssion band peaked at 485 nm is solely from the ethynylene perylene trap. We detect th e fluorescence both at 380 nm and 485 nm. The time-resolved emission data will clarify the origin and dynamics of each signal. The natural electronic excited state lifetime of the nanostar is 2.2 ns, same as the lifetime of phenylethynylene perylene.70The nanostar’s backbone (without the Eper trap) was measured to have a 270 ps lifetime in DCM. We performed femtosecond timeresolved fluorescence upconversion experiment s to monitor the arrival time of the energy into the ethynylene perylene tr ap. After excitation at 310 nm, emission is detected at 485 nm (emission from P* in Figure 5-12) and 380 nm (emission from C in Figure 5-12) corresponding to ethynylene perylene tr ap fluorescence and dendrimer backbone fluorescence, respectively. The dynamics obtained from these two data sets are complementary and allow us to follow the en ergy migration from the initially excited state on the nanostar’s backbone to the ethynyl ene perylene trap. Figure 5-13 shows the temporal evolution of the nanostar emission at 485 nm (blue line) and 380 nm (red line). The temporal behavior of the fluoresce nce detected at the acceptor’s emission wavelength (485 nm) shows a slow rise time ( 10-20 ps, the slowest rise that is measured for any PE dendrimers so far) and a nanosecond decay. On the other hand, the temporal behavior of fluorescence dete cted at 380 nm exhibits a subpicosecond rise and two decays one in the order of tens of pico seconds and the other on the hundreds of picoseconds. The ultrafas rise time and fast decay time are associated with the excitation

PAGE 176

160 energy transfer, whereas the slower decay time corresponds to unsubstituted nanostar backbone emission. This unsubstituted backbone emission accounts for most of the steady state emission intensity at 380 nm. Th is emission band coincides very well with the emission from the longest chain of nanos tar dendrimer, which is the 4-ring PE unit. 01020304050 0.0 0.5 1.0 det= 380 nm upconverted fluorescencetime(ps)det= 485 nm Figure 5-13. Fluorescence Upconversion signal de tected at 485 nm (blue) and at 380 nm (red) following the 310 nm excitation. The solid lines in Figure 5-13 correspond to the fits via the kinetic model solutions derived in previous section. It is importan t to note that what we measure here is the population dynamics of excited states C (red ) and P* (blue) given by equations 5-14 and 5-15. The fits are calculated from the convolut ion of these functions with the instrument response function. Exciting the nanostar at 310 nm as discussed earlier, directly deposits the excitation energy into the 2-ring units highest-lying st ate of the dendrimer backbone. Considering the transient abso rption results obtained at 310 nm excitation wavelength, a stepwise energy transfer, from 2-ring to 3-ring, 4-ring, and finally to the Eper trap is expected. The fitting procedure of the emission detected at 485 nm yields mainly two risetimes, a fast component (average of 2k-1 and k3 -1 ) in the subpicosecond time scale, and a very slow component 15 = 1/(k1+k5). The fast component rises with 23= 700 fs,

PAGE 177

161 while the slow component rises with 15 = 7 ps time constant. These time constants are in good agreement with the time constants obtaine d in transient absorp tion analysis when excited at the same wavelengt h. This kinetic model reveal s the presence of both direct and indirect pathway, and for the indirect (mul tistep) channel, the rate limiting step is the energy transfer from the peripheral 2-ri ng chromophores to the adjacent 3-ring chromophores. A confirmation of the cascade mechanism conc urrently with direct channels is also presented in Figure 5-13 (red line), where the emission from the intermediate state (backbone of the nanostar) at 380 nm is monito red. In this case, a very fast rise is followed by a slow decay. According to this model, the slow decay corresponds to population arriving at the emi ssive state C and the fast rise corresponds to the depopulation of the same state. The fit of the experimental da ta with equation 5-14 yields the following time constants: 1 =7.5 ps 2 =470 fs, 3 =300 fs and 5 =21 ps. These values coincide well with the previous resu lts obtained for ethynylene perylene emission at 485 nm. The time constants are very si milar in both fits verifying the wellestablishment of the predicted kinetic mode l. If the energy tran sfer was only via the cascade process, the temporal behavior of the final emitting state of the nanostar “P*” should be exactly same as the temporal behavior of the lowest lying state of the backbone “C”. However, due to direct channels from th e initially excited state of the nanostar to the final trap, the data represen ting the population dynamics of P* and C do not overlap completely. In addition, the long decay obser ved for the emission detected at 380 nm has a significant contribution from the 270 ps lifetime of the un-functionalized nanostar backbone.

PAGE 178

162 We are able to fit the transient absorp tion data, for three different excitation wavelengths, and the fluorescence data with the same kinetic model. Thus, even though there might be some approximations for the simplicity of the fitting procedure, the proposed model provides a clear picture of the energy transfer process in this macromolecule. Since the nanostar is a system with se veral chromophores absorbing light in different spectral regions and the excitation energy is localized in each chromophore, we label the electronic states independently. Un idirectional energy transf er takes place from the chromophore with the energetically highest S0-S1 transition (2-ring peripheral groups) to the chromophore with th e energetically lowest S0-S1 transition (et hynylene perylene core). Considering the S0-S1 energetics, this process involves two more chromophore intermediates. If the emi ssion spectrum of the chro mophore with the highest S0-S1 transition energy overlaps with a S0-Sn absorption band of the chromophore with the lowest S0-S1 transition energy, and if both chrom ophores are at favorable distance and orientation, the presence of intermedia te chromophores is not required for 2-ring EPer energy transfer. If so, distinct direct energy transfer from the peri phery to the core does occur through a direct pathway. The fluoresce nce of 2-ring unit of nanostar overlaps with both S0S1 and S0 –S2 absorption band of ethynylene perylene, and molecular geometry optimization indicates that interchromophoric distance between these two chromophores are short enough for efficient Frs ter energy transfer. Since the S0S1 and S0S2 dipole moments of EPer are perpendicular to each ot her, we expect that 2-rings oriented at perpendicular position contribute to the different pathways. In addition, the intermediate 3-ring chromophore’s emission has significant overlap with the S0S1 absorption band of

PAGE 179

163 ethynylene perylene. Thus, unidi rectional energy transfer from 2and 3ring PE chromophores to the core EPer is not to be excluded and will compete with the cascade energy transfer. Considering the possibility of the direct transfer via the existence of spectral overlap, we can now assert the accu racy of the propos ed kinetic model. Energy Transfer The goal of the study presented in this ch apter is to explor e the interchromophore energy transfer process, which can result fr om different interaction mechanisms. The through-bond mechanism requires spatial orb ital overlap, and is effective at short distances between donor and acceptor. The me ta substitution between the chromophores in the nanostar disrupts the -electron conjugation, thus suggesting the dendrimer molecule to be an ensemble of linear ch romophores with no (weak) charge transfer between them. If such a charge transfer is completely eliminated, and linear segments are well separated, through bond mechanism is expected to be suppressed. At donor acceptor separations beyond their van der Waals radii, the coupling is described primarily by through-space, Coul omb interaction. Both donor and acceptor chromophores in the nanostar have elect ronically allowed emission and absorption transitions, so the largest contribution to the coupling will be due to Coulombic interaction. Once the exciton is created, it is shown that it can migrate through a direct path and multistep path towards the acceptor molecule at the core. The strength of the Coulombic coupling will lead to cohere nt or incoherent exciton migration. In the case of very weak coupling, th e energy transfer rate is described by: 2 2 24ETkJV hc (5-16)

PAGE 180

164 where V is the Coulombic coupling and J is th e spectral overlap integral in units of cm. This overlap term is calculated by using the normalized absorption spectrum of the acceptor –a( )-, and the normalized emi ssion spectrum of donor f( ). The resonance interaction strength, V, is cal culated using the experimental transfer rates and overlap integral. Minami et al. theoretically estimated the interaction strength between the segments connected at meta positions using the Collective Electronic Oscillator approach.78 The Frenkel exciton Hamiltonian was us ed to represent the nearest neighbor couplings, i.e. between linear segments connect ed by a benzene ring at the meta position. Using equation 5-16, the interaction streng th between linear segments along with the interaction of the longest segment with the perylene trap is calculated. When the longest linear segments, 4-ring units ar e preferentially excited at 372 nm, k1 =1/290 fs-1 is obtained as the transfer rate (kET). The overlap integral between the 4-ring and the ethynylene perylene is utilized as J, which is calculated to be 9.423 x10-5 cm. In order to calculate the overlap integral, the free 4-ring chromophore is used as model for the donor, with emission spectrum red-shifted 12 nm to acc ount for the red shift observed when the chromophore is part of the nanostar. The in teraction energy between the 4-ring and the Eper is estimated to be V4-pery 176 cm-1. Minami et al. estimated a theoretical value of 302 cm-1 for the interaction between the same pair of donor and acceptor. Due to the stepwise (interchromophore) en ergy transfer, the excitation of the 3-ring chromophore at 352 nm results in energy transf er to the 4-ring chromophore. As noted in the analysis of transient absorption data, this transfer rate constant is determined to be kET (k2) =1/500 fs1. The overlap integral between the 3ring chromophore’s emission and the 4-ring chromophore’s absorption is evaluated to be J = 1.065 x 10-4 cm. Then, the interaction

PAGE 181

165 strength is estimated to be V3-4 126 cm-1. The same interaction was theoretically calculated to be about 325 cm-1. If the peripheral 2-ring units (highest lying state of the nanostar) are excited at 310 nm, the excita tion energy will be transferred to the neighboring 3-ring chromophores as the first step of the cascade mechanism .The analysis of both transient absorption and fluor escence upconversion data yields a transfer rate of kET (k3) =1/10 ps-1. Since the 2-ring chromophore assembly resembles a G2 phenyl-terminated monodendron, it is reasona ble to calculate the spectral overlap by using the emission spectrum of the G2 phenyl terminated monodendron (structure 9 of Devadoss et al.81) and the absorption spectrum of a free 3-ring chromophore. Based on this remark, the calculations yield a value of J = 4.718 x10-5 cm, and the interaction strength is estimated to be V2-3 45 cm-1. Theoretically, the interaction energy between the 2-ring and 3-ring units wa s calculated to be 158 cm-1. In the calculations reported here, the donor and acceptor chromophores’ sp ectra are always red-shifted ~12 nm to account for the red shift observed when thes e chromophores are part of the nanostar. However, no change is expected for the fl uorescence emission of donor G2 since it has the same periphery composition as the nanostar. In general, the values of interaction strength calculated using the experimental data are in qua litative agreement with the theoretically estimated values even though the experiments yields 2-3 times weaker than those theoretically estimated. In a series of recent studies, Mukamel’s group have computed the electronic excitations, lin ear absorption, pump-probe spectra, and time/frequency –gated fluorescen ce signals for the nanostar.76-78,150,153 Redfield theory was used to describe the exciton trans port. The effects of nuclear motion were incorporated through relaxation superoperators and calculated perturbatively in exciton-

PAGE 182

166 phonon coupling. The fluorescence signal was computed using the Doorway-Window representation and it provided a direct probe for exciton dynamics. The frequency domain pump-probe signal is simulated as well, s howing the effects of exciton coupling and excited-state absorption in the nanostar. Calculations were performed at room temperature T=300 K, and the excitation pulse was tuned to coincide with the highest periphery exciton. Both simulated time-r esolved fluorescence spectra and pump-probe signals for different delays i ndicated that within 100 ps, mo st of the excitation energy had reached the trap. Thus, the theoretical upper li mit for energy transfer time in the nanostar is about 100 ps. The calculated rates are cl early much slower than our experimental results and the interaction energies between any chromophores are la rger than the ones obtained by using our experiment al findings. This difference mu st be related to Redfield treatment. Redfield theory is based on the assumption that th e electron-phonon coupling is weaker than Coulomb c oupling between the chromophores and can be applied when the relaxation of the excitonica lly coupled molecules is slow compared to the relaxation of surroundings. In order to test the limitations of Frster and Redfield theories, Yang et al.154 used a simple system of two interacting molecules and compared the energy transfer rates obtained via both Frster and Redfield theories. It was concluded that standard Redfield theory only works well for a small energy gap between the interacting states. When the energy gap is large, the exciton st ates are localized on each molecule and the exciton relaxation rate reduces to the hopping rate in Frster regime. For example, in the case of relatively weakly coupled pigments (coupling strength = 20 cm-1) having similar site energies (energy gap = 0-100 cm-1), the Redfield rates can be up to 10-100 times larger than the Frster rates. Whereas, for larger energy gaps (200-1000 cm-1), the

PAGE 183

167 Redfield theory gives significantly slower rates than Frster theory. Therefore, it is expected to get faster transfer rates from our experiments compared to the simulation of the same experiments described by Redfie ld theory. It was emphasized that the simulations of faster relaxation lifetimes ( < 100 ps) would questi on the validity of the Redfield equations and require parameters of different level of modeling (modified Redfield theory). In summary, the relaxation times observed for excitation at 372 nm and 352 nm are considerably faster than for exci tation at 310 nm. The subpicosecond ( 1 =290 fs and 2 =500 fs) relaxations are attributed to larg er electronic coupling of the corresponding chromophores. The coupling of the 4-ring ch romophore and ethynylene perylene trap as well as the coupling of the 3-ring and 4-ring chromophores is much larger than any interaction of the 2-ring peri pheral groups. It was discussed that the faster relaxation could be an indication of a better overlap between the emission spectrum of the donor and the absorption spectrum of the acceptor.75 However, the spectral overlap obtained using the measured spectrum of free of 2-, 3-, and 4-ring chromophores (red-shifted to account for the red-shift observed when th e chromophore is part of the nanostar molecule) yields very similar values. The stre ngth of the coupling de termines the rate of the transfer. In the simplest form of the F rster mechanism, the Coulomb interaction is approximated by a dipole-dipole interaction, and th e density of states is expressed in terms of the shape of the donor lumines cence and acceptor absorption spectrum. The energy transfer rate constant is then e xpressed with the equation 1-21. Using this equation, Kleiman et al. estimated kET by considering coupling of the 2-ring chromophores connected to a single 3ring chromophore. The R values and the

PAGE 184

168 orientation factors were estimated for a planar configuration. The se paration, R, between the 2-ring and 3-ring chromophores ranges from 9.1 to 15.1 . The spectral overlap was estimated to be 2.8 x10-14 cm6/mol. It was found that kET -1 ranges between 0.83 and 20.0 ps, depending on the position of the 2-ring chromophore at the periphery. Our experimental findings for the 310 nm exci tation along with preceding results from degenerate pump-probe experiments show rough qualitative agreement with the estimated Frster rates. The uncertainties in the orientation factor s and interchromophore distances has to be re-evaluated consider ing the globular shape of the PE dendrimers and the increase in the rotational flexibility near the dendrimer periphery. Moreover, the Frster’s formulation based on the dipole-di pole approximation is questionable when the donor and acceptor molecules are closely prox imate relative to molecular dimensions. More accurate expansion of the Coulomb interaction beyond the dipole-dipole level is needed for such systems. The Transition Density Cube (TDC) method, which takes advantage of modern quantum chemical calcu lations of excited-st ate wavefunctions, was developed to obtain the local interacti ons between the donor and acceptor transition densities. Recently, Ortiz et al. has mode led the Coulombic cont ribution to the energy transfer between the 2-ring and 3-ring ch romophores of the nanostar using TDC method.74 The MD simulations of the nanostar s howed that the phenyl rings are free to rotate around the ethynylene bonds. The transi tion densities were calculated for both planar geometry and the off-planar geomet ry. While the rotation around the central angle effects the magnitude of th e electronic transition dipole moment for the 3-ring chromophore, the rotation of the outer ri ng in the 2-ring chrom ophore was hindered due to bulky groups at the periphery. The energy transfer rates between the 2-ring and 3-ring

PAGE 185

169 chromophores were calculated as a function of the torsion angle of the central phenyl ring of the 3-ring chromophore. The Frster formul a does not explicitly depend on the torsion angle (implicit dependence through parameter J) whereas the TDC has shown a strong dependence on torsion angle. The rate consta nts were almost 3 orders of magnitude different with a torsion angle of 900. It was concluded that the ri ng rotation is a factor that must be included when modeling energy tran sfer in the nanostar and more accurate coupling can be calculated by reconsideri ng the effect of torsional angle on 2 and the spectral overlap. Conclusions We employed broadband transient abso rption and time-resolved fluorescence experiments to investigate the complete map of the energy pathways in the nanostar. The spatial positioning of different size PE chro mophores and EPer chromophore within the nanostar and their respective spectral properties make this system an efficient lightharvester, which is able to capture light over a broad spectral range and transfer it directly and in a cascade fashion to the core EPer. Using the 372 nm excitation, a highly efficient, subpicosecond unidirectional energy transfer from the 4-ring units to th e EPer core is evident. Upon 352 nm and 310 nm excitation, 3-ring and 2 ring chromophores can transfer their ex citation energy either directly or in a cascade fashion. Within the nanostar, both processes are present and will compete with each other, because the emissi on of 2and 3-ring chromophores overlap considerably with both the S0-S1 and S0-S2 absoprtion band of EPer. Based on the spectral data and our kinetic modeling, direct energy transfer take s place with 33 % probability, while the cascade mechanism accounts for 66 % of the energy reaching the final trap.

PAGE 186

170 Resonant excitations with 3-ring and 4-ri ng chromophores result in subpicosecond energy transfer times, while the excitation of 2-ri ng peripheral chromophores yields transfer times of tens of picoseconds. For the first time, independently estimated theoretical and experimental interaction energies are compared. The quali tative aggreement could be improved by full treatment of Coulombic inte raction using TDC and molecular dynamics. Since we are able to probe the population dynamics of inte rmediate states, the major outcome of the experiments presented is proving the existen ce of direct transfer competing with the cascade type transfer.

PAGE 187

171 CHAPTER 6 THE ROLE OF EXCITON HOPPING AND DIRECT ENERGY TRANSFER IN THE CONJUGATED POLYELECTROLYTES Organic semiconductors and in particular -conjugated polymers have entered numerous fields of applications such as light-emitting diodes, lasers, solar cells and, recently also the important field of ch emical and biological sensing. Here, -conjugated polymers have proven to increase the sensit ivity of fluorescence-based sensors by several orders of magnitude.155-157 For biological applications, the -conjugated polymers need to be soluble in aqueous environments. Recently, amplified fluorescence quenching of a water-soluble -conjugated poly(phenylene viny lene) (PPV) based anionic polyelectrolyte has been demonstrated.156,157 Chen et al. showed that quenching of a polyanionic conjugated polymer by cationic electron acceptors can be a million-fold more sensitive than the corresponding que nching of small molecules of similar structure.156 The amplification of fluorescence quen ching relies on the high mobility of the photoexcitations on the conjugated polymer chain, leading to the quenching of many chromophores upon the binding of a single quencher molecule155,156 However, a variety of effects enhance the fluorescence quenching of conjugated polyelectrolytes (CPE’s). In addition to the importance of the photoexcit ations mobility along the polymer chain, quencher-induced aggregation of the polymer chain increases the number of chromophores in the direct vicinity of the quencher a nd enables interchain energy migration, thereby enhancing the quenching efficiency.156,158,159 Furthermore, the large charge of the conjugated polyele ctrolyte can lead to a local concentration enhancement of

PAGE 188

172 the quencher by more than an order of magnitude160. The influence of these rather extrinsic effects raises the question, to wh at extent does the exciton mobility on the polymer chain contributes to the fluorescence quenching? The rate and efficiency of intramolecular energy transfer in -conjugated polymers is currently a subject of in tense discussion. Recent i nvestigations of a PPV-type conjugated polymer found the intrachain ener gy transfer to be slow and short-range.161 This is supported by measurements and M onte-Carlo modeling of the photoluminescence anisotropy decay of MEH-PPV162 and polythiophene163 in dilute solution, giving a range of the energy transfer of only 6-7conjuga ted segments, corresponding to about 20-30 nm. On the other hand, the enormous amplific ation of luminescence quenching in polymerbased sensors indicates a very efficient intramolecular energy transfer in a water-soluble PPV, extending over about 1000 monomer units.156 Efficient energy transfer on conjugated polymer chains has also b een found in single-molecule studies,164-166 where it was suggested that the energy transfer effi ciency depends strongly on the polymer chain conformation, and quenching of the comple te chain is achieved only for dense conformations forming aggregated structures. In these studies,165 as well as in investigations in La ngmuir-Blodgett films,167 a three-dimensional exciton migration was proposed to explain the high quenching ef ficiency. Altogether, many different mechanisms can improve or limit the quenc hing efficiency of fluorescence from a conjugated polyelectrolyte. A fundamental understanding of the energy transfer mechanisms in conjugated polyelectrolytes is therefore crucial for the development of reliable and sensitive polymer-based sensors.

PAGE 189

173 In general, energy transfer between an excited chromophore on a polyelectrolyte chain and an acceptor molecule can occur i) by direct energy transfer (e.g. by a longrange Frster type dipole-dipole coupling) and ii) in a multi-step mechanism consisting of intrachain energy migration on the polymer chain to a site adjacent to the acceptor, followed by short-range energy transfer to the acceptor.156 Furthermore, for samples in solution one has to distinguish between quenching by acceptors that are in an ionic complex with the CPE and diffusional quenching by free acceptors in the solution, where the latter becomes important at high quencher concentrations.160 Wang et al. described the fluorescence que nching of a PPV-type CPE by methyl viologen with a modified Stern-Volmer equa tion that accounts for static quenching by both acceptor molecules in an ionic comple x with the donor and free acceptors in the solution.160 The superposition of both contributions leads to an expone ntial rise of the quenching efficiency according to I0/I =(1+ KSV [Q]) exp( V ), were I0/I is given by the ratio of donor emission without a nd with the quencher present, KSV represents the association equilibrium constant for complex formation, and V is the volume of the quenching sphere. The factor accounts for an enhanced que ncher concentration in the vicinity of the CPE. The formula is deri ved on the assumption that the quenching is entirely static i.e. much faster than the excited state lifetime of the (unquenched) donor. There is ample evidence, however, that ex citon migration in conjugated polymers is active on a timescale of tens of picoseconds163,168 (i.e. comparable to the excited state lifetime of most polymers). T hus, the quenching should be a dynamic process where the rate of energy migration on the CPE chai n significantly influences the quenching efficiency.

PAGE 190

174 In this work, we inevstigate the energy transfer dynamics from an anionic conjugated poly(phenylene ethyny lene) sulfonate (PPESO3) to a cationic dye molecule (HMIDC). The chemical structures are s hown in Figure 6-1. St eady state fluorescence spectroscopy reveals that formation of an ionic complex between the polymer and dye leads to efficient polymer dye energy transfer.169 In order to monitor the energy transfer dynamics, femtosecond time-resolved fluorescence up-conversion is employed along with transient absorption and polarization anisotropy studies on solutions with systematically varied dye concentration. Figure 6-1. Conjugated polyele ctrolytes PPESO3 (left) and cyanine dye HMIDC (right). In addition, a numerical m odeling is presented using th e analytical solution of a random walk process between energetically identical and equidist ant chromophores. The effects induced by energetic and conformati onal disorder are accounted for by a timedependent hopping rate. Modeling of the dynamics allows us to determine the density of complexes on the polymer chain and the number of chromophores quenched by each complex formed. The individual contributions of intrachain hopping to wards the acceptor and direct long-range Frster-t ype transfer are quantified, allowing design rules for systems with improved sensor activity to be provided. Steady State Experiments The synthesis of PPESO3 has been described recently.158 The molecular weight of the polymer is estimated to be Mn=100 kDa, corresponding to about 200 monomer units (see Figure 6-1 for monomer repeat unit stru cture). HMIDC was purchased from Aldrich n+Na -O3S(H2C)3O O(CH2)3SO3 -Na+ N CH3CH3CH3(HCCH)2CH2N CH CH3H3C H3C

PAGE 191

175 and used as received. Solutions with differ ent HMIDC concentrations were prepared and labeled according to their stea dy state quenching efficiency. Steady state absorption and emission spectra of both materials in methanol solution are shown in Figure 6-2a. The emission of PPESO3 partially overlaps with the HMIDC absorption spectrum, thus enabling singlet ener gy transfer from the polymer to the dye. The large separation of the emission bands of the polymer and dye allows for independent measurement of donor emission and acceptor emission in time-resolved fluorescence experiments. A concentration of 34 M PPESO3 in CH3OH (in polymer repeat units, PRU) was used for the time-resolved fluorescence and transient absorption experiments. 450500550600650700750 0 1 2 3 4 5 01234567 0 20 40 60 80 100 0 1 2 3 4 5 6 300400500600700 0.0 0.1 0.2 0.3 Absorption (arb. units)Wavelength (nm)a )PPESO3 HMIDC Photoluminescence (arb. units) Wavelength (nm)b )Efficiency (%) Quencher conc. (M) Emission (arb. units) (a) (b) Figure 6-2. (a) Absorption () and emission spectra () of PPESO3 and HMIDC in methanol. (b) Fluorescence spectra of 10 M PPESO3 with added HMIDC (0 to 2.5 M), excitation= 400 nm. Inset: efficiency of polymer (34 M) quenching versus dye concentration. Figure 6-2b shows the steady state emissi on spectra of PPESO3 with different concentrations of added dye The excitation source at 400 nm exclusively excites the PPESO3 polymer. As the dye concentrati on is increased, the emission of PPESO3 decreases while the dye emission increases, re vealing an efficient energy transfer from the photo-excited polymer to the cyanine dye. This efficient transf er process can be

PAGE 192

176 attributed to a complex formation between the anionic polymer-donor and the cationic dye-acceptor.169 The inset of Figure 6-2b show s the efficiency of the PPESO3 fluorescence quenching as a func tion of dye concentration. Time Resolved Fluorescence Spectroscopy To further investigate the mechanism of PPESO3-to-cyanine energy transfer we carried out picosecond time-resolved fluor escence experiments by using fluorescence upconversion. These experiments were carried out at a fixed PPESO3 concentration (34 M). HMIDC was selected as the acceptor cy anine dye, and its concentration was varied from 0 – 5 M. Concomitant with the time resolv ed experiments, pa rallel steady-state quenching measurements were carried out, and the fraction of the PPESO3 unquenched fluorescence (Fq = I/Io) for the PPESO3/HMIDC solutions are listed together with the time resolved data in Table 6-1. All of the time resolved experiments were carried out in methanol solution to minimize the influen ce of polymer aggregation on the observed fluorescence dynamics. Table 6-1. Parameters Recovered from Kinetic Modeling of PPESO3 Fluorescence Decays with HMIDC in MeOH [HMIDC] / ( M) [PRU] : [Dye] Fq Ar o / ps o 2 / ps 2 0 --1 1 150 0.6 --0.4 85:1 0.5 0.2 150 0.6 > 1500 0.25 1.2 28:1 0.25 0.41 150 0.6 760 0.25 5.0 6.8:1 0.07 0.64 150 0.6 20 0.2 [PPESO3] = 34 uM in MeOH. Fracti on of total PPESO3 fluorescence unquenched (Fq = I/Io, where I and Io are, respectively, the PPESO3 fluorescence intensity with and without added HMIDC quencher). By using a laser excitation wavelength at 425 nm (the pump wavelength) it is possible to directly excite PPESO3. HM IDC was selected as the energy acceptor for these studies because it absorbs only very w eakly at 425 nm, and consequently when the

PAGE 193

177 425 nm pump is used to excite the PPESO3 /HMIDC mixtures, excitations localized on HMIDC are created almost exclusively by energy transfer from the polymer. In addition, owing to the large wavelength separati on between the fluorescence of PPESO3 ( 450 nm) and HMIDC ( 685 nm) it is possible to sele ctively observe the fluorescence dynamics for PPESO3 and HMIDC (the en ergy donor and acceptor, respectively). Figure 6-3 shows fluorescence decays dete cted at 450 nm with 425 nm excitation for a series of solutions that contain PPESO 3 and HMIDC at concentrations ranging from 0 – 5 M. The decay curves that are plotted in Figure 6-3 represent absolute fluorescence intensities, since were obtained under identical conditions (matched solution concentrations of PPESO3, identical pump powe r, and detection sett ings); therefore the amplitudes of the decays reflect the true re lative instantaneous intensity of the timeresolved fluorescence signal from the different samples. Upon inspection of the data, two features are apparent. First, the initial am plitude of the polymer’s fluorescence decay is reduced by the addition of HMIDC. Second, with increasing HMIDC concentration, the fluorescence decays more rapidly, which can be better seen in the inset of Figure 6-3, where normalized data is plotted. Taken t ogether, these observati ons indicate that the PPESO3-to-HMIDC energy transfer takes place on two distinct timescales. A significant component of the transfer occurs by a pathway that is so fast that it cannot be resolved within the instrument response (~ 4 ps).

PAGE 194

178 0150300450600 0.0 0.5 1.0 0150300450600 0.00 0.02 0.04 0.06 0.08 Relative Fluorescence Signal Normalized Signal time (Ps) Figure 6-3. Time-resolved fluorescence of PPESO3 (34 M) in MeOH for different HMIDC concentrations: 0 M, 0.4 M 1.2 M 5.0 M Excitation at = 425 nm, detection at = 450 nm See text for details. Inset: normalized fluorescence. In addition, the clearly noticeable increas e in the rate of the fluorescence decay with increasing HMIDC concentration (see inset, Figure 6-3) indicates that there is also a slow energy transfer process which may repr esent diffusion of the exciton to the HMIDC acceptor. In order to model the d ecay kinetics of the PPESO3 fluorescence, both in the absence and presence of the HMIDC quenche r, a stretched exponen tial function of the form in equation 6-1 was used: 22 rexp exp A 1 t t t Ioo (6-1) where I(t) represents the fluorescence intensity at time t, the parameters o and o model the natural decay kinetics of the polymer (i.e., in the absence of quencher), and 2 and 2

PAGE 195

179 model the dynamics of the slow energy transfer pathway. The terms provide a measure of the width of the lifetime distribution ( < < 1, where the width of the distribution increases as decreases). The term appearing in the pre-exponential, Ar, accounts for the reduction in the initial amplitude of the fluorescence, and it is attributed to a prompt quenching pathway (lifetime 1) which is so rapid that it cannot be resolved within the instrument response (i.e., 1 < 4 ps). The fluorescence decay of PPE-SO3 in the absence of HMIDC is fitted by equation 1 with the decay time o = 150 ps and o.= 0.6 ( 2 = ) The observation of a stretched exponential decay for the polymer’s fluorescence is not surprising in view of the fact that there is heterogeneity in the chromophores w ithin the polymer chains (for example, due to a distribution of conjugation lengt hs caused by rotation around the Ph-C C bonds). In addition, studies of the fluorescence decay characteristics of phenylene ethynylene oligomers reveal complex dynamics on the 0 – 100 ps timescale associated with to solvation and conformational relaxation of the initially produced singlet exciton.170 It is likely that similar processes occur in the polymer. The decays obtained for solutions that contain HMIDC were fitted also using equation 1, holding o and o constant, and varying Ar (the amplitude of the prompt quenching pathway), 2 and 2 (the lifetime and width of the slow energy transfer pathway). The parameters recovered from the fits are collected in Table 6-1, and several trends are evident from this data. Firs t, the amplitude of the prompt process (Ar) increases significantly with increasing HMIDC concentration. Comparison of the value of Ar with the fraction of total emission quenched (Fq) reveals that, on average, the prompt energy transfer process can account for more than one-half of the total

PAGE 196

180 fluorescence quenching. This cl early demonstrates that a su bstantial component of the PPESO3-to-HMIDC energy transfer takes place on an ultrafast timescale (i.e., < 4 ps). In addition to the ultrafast quenching pathway, the fits reveal that a second, slower energy transfer pathway is operative ( 2 and 2, Table 6-1). The time constant of this slow pathway decreases substantially with increasing HMIDC concentration, yet in all cases the lifetime distribution is broad as reflected by the fact that 2 is low (0.2-0.5). The notion of energy transfer being active on two distinct timescales is confirmed by the temporal behavior of the HMIDC fluorescence. Figure 6-4 compares the decay kinetics of the HMIDC fluorescence for three different samples. First, excitation of a MeOH solution of pure HMIDC, with exc = 636 nm and det = 663 nm (at the maximum of the pure-dye emission), produces a fluorescen ce signal that rises within the instrument response and decays with = 285 ps (single exponential). This is the natural lifetime of the HMIDC dye in MeOH solution. Next, d ecays were obtained for mixtures of PPESO3 (c = 34 M) and HMIDC (1.2 and 5.0 M). In this case the solutions were excited at 425 nm (PPESO3 absorption) with the HMIDC fluor escence detected at 685 nm. It is evident that for both solutions containing PPESO3, the decay rate of the HMIDC fluorescence is slower compared to the pure HMIDC solution ( = 1.2 ns). The slower decay rate is believed to arise from a stabilization of the excited state of the dye upon complex formation with the PPESO3 chains. Of more in terest is the fact that for both of the solutions containing PPESO3 and HMIDC, the ri se of the dye’s fluorescence features two distinct components. The majority of the HMIDC fluorescence signal rises with an instrument-limited response ( 1 < 4 ps), confirming that a significant component of the PPESO3 to HMIDC energy transfer occurs on an ultrafast timescale. However, in

PAGE 197

181 addition to the prompt rise component, both solu tions exhibit a clearly resolved slow risetime component, which corresponds to the slow energy transfer process resolved in the PPESO3 fluorescence decay experiments described earlier (characterized by 2 and 2). The solid lines shown in Figure 6-4 were gene rated by using the same kinetic parameters used to fit the PPESO3 decays (Table 6-1), wh ich reinforces the hypothesis that the effect of HMIDC on PPESO3 fluorescence decay dynamics corresponds to an energy transfer process. 0200400600 0.0 0.2 0.4 0.6 0.8 1.0 Normalized Signal Figure 6-4. Normalized time-resolved fluorescence of HMIDC in MeOH. pure dye ( exc = 636 nm, det = 663 nm) HMIDC (1.2 M) + PPESO3 (34 M) HMIDC (5 M) + PPESO3 (34 M) .For the mixtures, exc = 425 nm, det = 685 nm. Comparison of Transient Fluorescence and Absorption Data The fluorescence up-conversion experiments allow the study of the energy transfer dynamics from the photo-excited PPESO3 to the cyanine dye. Exciting the polymer at

PAGE 198

182 425 nm and detecting the fluorescence of PPE SO3 at 450 nm, allow us to follow the temporal response of the polymer’s emission (Figure 6-5b). The topmost curve (0% quenching) corresponds to emissi on from the pure polymer in CH3OH. The other curves show the emission from the polymer in the presence of the dye for different dye concentrations. With increasing dye con centration the initial amplitude of the fluorescence decreases considerably. This amplitude reduction is assigned to energy transfer from the photoexcited PPESO3 to th e dye, occurring on a timescale faster than the experimental time-resoluti on of about 200 fs. Additionally, at high dye concentrations leading to a steady state quenching of more th an 50% (3 lower curves), the decay of the PPESO3 fluorescence becomes significantly fast er, being reduced from about 250 ps for the pure polymer to 50 ps decay when 6 M of the cyanine dye are added (corresponding to a steady state quenching of ~80 %). In th e following, we will refer to those quenching processes occurring much faster than the excited state lifetime of the PPESO3 as prompt quenching, whereas quenching taking place on th e timescale of the excited state lifetime will be referred to as gradual quenching. Contributions from both prompt and gr adual quenching are also found in the transient absorption data, which allows monitoring the excite d state population of PPESO3 and HMIDC simultaneously. The transi ent absorption of PPESO3 in methanol for different concentrations of cyanine dye is shown in Figure 6-5a (TA experiments are performed by Dr. Mller). Again the topmos t curve corresponds to pure polymer while the other curves correspond to the TA signal in the presence of the quencher. Probing at =680 nm (maximum emission of the dye), the pure polymer solu tion shows a positive transient absorption (an increase in th e absorption of the sample).

PAGE 199

183 0100200300400500 0 1 2 3 t (ps)0 M 6 M PL (arb. units)6 M 0 Ma )-5 0 5 Abs (mOD)b ) Figure 6-5. PPESO3 in MeOH for different concentrations of added HMIDC. The excitation wavelength is 425 nm. a) Tran sient absorption detected at a probe wavelength of 680 nm with [HMIDC]=0, 0.65, 1.35, 3, and 6 M, and b) Upconversion signal of fluorescence detect ed at 450 nm for [HMIDC]=0, 0.65, 1.35, 3, and 6 M. Both plots present signal at magic angle. The results from the numerical calculations are shown as solid lines. Figure 6-6 compares the time evolution of this photo-induced absorption (PIA) signal with the PPESO3 fluorescence decay. The si milarity of these two curves allows us to attribute the transient absorption signal at 680 nm to absorption by the singlet exciton of PPESO3. In agreement w ith the fluorescence up-convers ion data, the addition of HMIDC leads to a reduction of the PIA amp litude and decay time. Furthermore, in the presence of the dye the transient absorpti on data shows a crossover from PIA to a

PAGE 200

184 negative signal (increased transmission) i ndicating either ground state bleaching or stimulated emission. At this probe wavele ngth, neither PPESO3 nor HMIDC absorb, and the polymer does not fluoresce; therefore the negative transient absorption is assigned to stimulated emission (SE) from excited HMIDC. Since direct excitation of dye molecules by the pump pulse at pump = 425 nm does not occur, the in crease of SE with delay time reflects the energy transfer to the dye. Exc itation energy leaving the polymer reduces the PIA from the PPESO3 (decrease of A) while energy arriving on the dye enhances SE, again decreasing the measured signal A. The combination of both effects induces a strong change in the transient absorption sign al and results in a hi gh sensitivity to the energy transfer dynamics. 0100200300400500 0.0 0.5 1.0 PL-upconversion A t (ps)A and PL (normalized) Figure 6-6. PL-upconversion () and A() of a pure PPESO3 solution. The signals have identical decay dynamics, showing that (i) both signals give the population of the first excited state of PPESO3, (ii) no significant energy transfer to aggregates sites with a lower radiative decay rate occurs.

PAGE 201

185 Anisotropy and Ionic Complexes Figure 6-7a shows the anisot ropy evaluated from the tran sient absorption data. The temporal evolution of anisotropy is obtained by evaluation of 2PS PSAtAt rt AtAt where P A t and S A t correspond to the transient ab sorption of the polarization oriented parallel or perpendicular to the excitation beam polarization, respectively. The pure polymer sample f eatures an anisotropy of r0 = 0.3 at t= 0, which is slightly lower than the theoretical value of r0 = 0.4 expected for a three-dimensional, randomly oriented ensemble of pe rfectly linear transition dipoles.144 The reason for this lower starting value is probably an ultrafast energy transfer process occurring within the experimental time-resolution of ca 150 fs. After this initial decay, the further loss of anisotropy with time is a direct measur e of the exciton hopping between conjugated segments of different orientation.163,171 The large molecular weight of the polymer chain slows down the rotational diffusion, leading to an almost static orientation of the molecule on the sub-ns ti mescales discussed here. The anisotropy decays within about 150 ps. A complete randomization of dipole orientations is expected to lead to r = 0 which is in contrast to the residual anisotropy of r = 0.05 observed in the data. On the basi s of the long-lived anisotropy decay we conclude that intrachain energy transfer on the polymer is active during the entire lifetime (250 ps) of the photoexcitations. The crossover from a fast initial d ecay to an extremely slow decay at larger time delays, howeve r, indicates a significant slowdown of the hopping process with time due to energetic diso rder within the CPEs leading to trapping of the excitations.172

PAGE 202

186 0100200300400500 0.0 0.1 0.2 0.3 r (t) Exc=640 nm Exc=425 nm t (ps) r (t)0.0 0.1 0.2 0.3 0.4 [HMIDC]=6 MExc=425 nmb ) a )[HMIDC]=0 Figure 6-7. a) Time dependent lo ss of anisotropy for PPESO3 in CH3OH. Excitation wavelength is 425 nm (polymer absorption maximum); b) Time dependent loss of anisotropy for a solution of PPESO3 with 6 M of quencher, HMIDC. Excitation wavelength is 640 nm () or 425 nm (). Detection wavelength is 680 nm (measured from transient absorption) for all three curves. Transient absorption anisotr opy changes in the presence of HMIDC were examined to understand the role of HMIDC molecule s complexed with the polymer vs. free dye molecules on the energy tran sfer. In the presence of 6 M of HMIDC (corresponding to ~ 80 % quenching of the steady state polymer emission) the anisotropy dynamics show a drastic change. Figure 6-7b shows the anisot ropy detected after excitation with two different pump frequencies, leading either to direct or sensitized dye emission. While probing at 680 nm, which corresponds to the em ission peak of the HMIDC we first excite the sample at 640 nm leading to direct excita tion of the HMIDC (at this wavelength, there is no absorption by PPESO3). Consequently under these conditions the transient

PAGE 203

187 absorption signal is given exclusively by th e stimulated emission of the dye, and the anisotropy calculated there from is the anisotropy of the HMIDC emission. The corresponding transient anisotropy plot in Figure 6-7b shows an initial anisotropy of 0.35or close to the theoretical e xpectation. The decay of the value of the anisotropy to 0 r shows a time constant of about 100 ps. While still probing the dye emission ( = 680 nm), it is possible to investigate the anisotropy behavior after en ergy transfer by exciting the dye exclusively via energy transfer from the PPESO3. This is accomp lished by tuning the excitation wavelength to 425 nm, where the PPESO3 absorption is high and the HMIDC absorption is negligible. Under these conditions, the time-dependent an isotropy shows an initial fast decay and then, for t >50 ps, the anisotropy of the sensitiz ed dye emission remains about constant at a level of 0.1 r To understand the quenching mechanisms it is fundamental to answer the question: do free dye molecules in solution contribute to the quenching? We will first answer this question and then describe the PL-upc onversion and transient absorption data quantitatively by comparing it to the results of the numerical simulation of the excitation hopping on PPESO3 chains with complexed HM IDC. An important outcome of the numerical simulations is that they provide the density of complexes along the polymer chain for the samples with different HMIDC c oncentrations. As we will discuss below, the computed complex density deviates si gnificantly from expectations based on a constant complex association constant.

PAGE 204

188 Loss of Anisotropy and the Absence of Quenching by Free Dye Molecules Figure 6-7b shows the anisotropy decay of the transient absorp tion signal measured on the PPESO3-HMIDC mixture. Due to the fast energy transfer to the dye, the transient absorption signal at t>50 ps is dominated by stimulated emission from dye molecules; the observed anisotropy therefore corresponds to the emission anisotropy of the HMIDC. Direct excitation of the dye leads to a high initial anisot ropy of the dye emission of 0.35or decaying to 0 r with a time constant of about 100 ps. Any dye molecules in complex with the PPESO3 chain are expect ed to maintain their orientation during the timescales of interest here. Consequently, we attribute the anisotropy decay down to r ~0 to the rotation of free dye molecules in solution. Figure 6-5a shows that for the soluti on with highest dye concentration (6 M) the energy transfer occurs very rapidly, and at t>50 ps the transfer from PPESO3 to the dye is almost complete. After completion of the energy transfer, the anisotropy of the sensitized dye emission shown in Figure 6-7b remains almost constant at 0.1 r We conclude that the energy is exclusively tran sferred to dye molecules that are bound to the PPESO3 in an ionic association and cannot free ly rotate, which justifies the neglect of quenching by free dye in solution. Note that this result is in contrast to previous studies on the quenching mechanism of a wate r-soluble poly(phenylenevinylene).160 The relatively high value of the residual anisotropy (0.1 r ) obtained for sensitized dye molecules suggests that in the donor-acceptor co mplex the transition dipole moments of HMIDC molecules are alig ned approximately parallel to the PPESO3 backbone. This is in agreement with the slight red-shif t reported earlier 159in the emission from the dye when PPESO3 is adde d the solution, indicating that occurs stacking

PAGE 205

189 between the PPESO3 and HMIDC. The consta nt anisotropy of the sensitized emission from the complexed dye also shows that th e polyelectrolyte chai n does not significantly alter its orientation (through ro tation) during the excited-state lifetime. This confirms our previous assumption that rotational diffusion of the polymer chain is much slower than the excited state lifetime and does not contribute to the anisotropy decay of photoexcitations on the polymer chain. It fu rther implies that the sample volume in solution probed by the polyelectrolyte is much smaller than its radius of gyration, which is in contrast to previous reports.160 Comparison of TAand PL-upconversion Dynamics The excitation hopping dynamics and hence the anisotropy decay can be strongly affected by the presence of energetic traps formed by in terchain interactions in aggregates.173 However, steady state spectrosc opy provides no evidence for PPESO3 aggregates in dilute solutions in CH3OH.169 This finding is corroborated by the comparison of the fluorescence-upconversion and transient absorption dynamics shown in Figure 6-6. In polar solvents, conjugated polymers can form aggregates with radiative decay rates that differ from those of their unaggregated structures.174 If aggregates are formed under the experimental conditions used in the pr esent investigation, th e energy transfer to the aggregates would be expected to change the radiative decay rate. The photoluminescence upconversion signal is given by 1~upradPLNSk where 1NSis the population number of ex citations of PPESO3 and krad is the radiative decay rate. Assuming a time-independent cross section for the S1 Sn>1 transition, the photoinduced absorption measured in transien t absorption is simply given by 1~ A NS Thus by

PAGE 206

190 comparing fluorescence up-conversion and transient absorption dynamics makes it possible to determine if there is a temporal evolution in the radiat ive decay rate of the polymer. Since both PLUP and A show exactly the same dynamics (Figure 6-6), the radiative decay rate must be constant in time. We conclude that energy transfer to PPESO3-aggregates with a lower radiat ive decay rate does not occur. Quenching Dynamics and the State Contribution The fluorescence upconversion data shown in Figure 6-5b can be used to estimate the relative contributions of static and dynamic quenching. The temporal evolution of the fluorescence of PPESO3 in the presence of HMIDC normalized to the fluorescence of the pure CPE sample is shown in Figure 6-8. Th is emission ratio represents the amount of quenching at a given delay time. A ratio of 1 indicates that no quenching has occurred up to the given delay time. 0100200300400500 0.0 0.2 0.4 0.6 0.8 1.0 3.0 M 1.35 M t (ps)PLquenched/PL00.65 M 6.0 M Figure 6-8. The photoluminescence yield of sample s containing HMIDC, divided by the photoluminescence of the pure PPESO3 solu tion. The ratio is calculated from the upconversion data shown in Figure 65b and provides a di rect description of the dynamic processes leading to quenching.

PAGE 207

191 The lowest quencher concentration of 0.65 M corresponds to 25 % of the steady state (integrated) polymer emission being que nched by the dye acceptor. For this sample, at ~0 t we observe a fluorescence ratio of 0.8, indicating that 20% of the emission is quenched within the experimental time reso lution, therefore being regarded as static quenching. At increased dela y time the fluorescence rati o appears nearly unchanged, indicating the lack of additional dynamic quenching. At increased quencher concentrations the fluorescence ratio begins to decay with time, revealing additional dynamic quenching. At the highest quencher concentration of 6 M about half of the energy transfer is completed within the expe rimental time-resolution (static component), whereas the other half of the energy tran sfer occurs on a 50 ps timescale (dynamic component). At increased quencher concentrations, th e simultaneous presence of static and dynamic is expected to lead to a superlinear Stern-Volmer behavior.135 Figure 6-9 shows the fluorescence signal taken fr om Figure 6-5b, integrated ov er time, and plotted versus dye concentration. This figure. reveals a linear dependence of I0/I as a function of dye concentration. Considering a purely static quenching in a 34 M solution of the polymer, a fit to the Stern-Volmer equation suggests an association constant of ~7.2 x105 M-1. The timeresolved data allow us to draw an equi valent Stern-Volmer plot for the static component only. It is obtained by plotting the rati o of the fluorescence of the PPESO3 in the presence and the absence of dye at t = 0 also shown in Figure 6-9.

PAGE 208

192 0123456 1 2 3 4 5 6 [HMIDC](M)I0/Iat t=0 timeintegrated Figure 6-9. Stern-Volmer plot of th e time-integrated photoluminescence () and the instantaneous photoluminescence ( t=0, ) from the upconversion experiment. The instantaneous photol uminescence shows a strong saturation at quencher concentrations above 1.2 M. Above a quencher concentration of ~2 M, the quenching does not increase any further with increased HMIDC concentration. In the time-integrated Stern-Volmer plot, the saturation of the static quenching com ponent “cancels out” the superlinear response expected from simultaneous static and dynamic quenching, and leads to an appaent linear increase of the steady-state (time-integrat ed) quenching with quenc her concentration. Modeling The quenching dynamics are affected both by the average distance between quencher complexes along the PPESO3 chai n, and by the excitation hopping dynamics. In order to obtain information about the complex density on the CPE and the number of

PAGE 209

193 monomers quenched by a single acceptor, num erical simulations of the intrachain excitation migration and quenching were performed by Dr. Jrgen Mller. The quenching of mobile excitations on a one -dimensional chain with trap sites has been treated intensively experimentally a nd in analytical and numerical calculations.175181 Monte-Carlo simulations of the ener gy transfer in polythiophene and MEH-PPV found that the energy migration covers on average a region of 6 to 8 hopping steps around the originally excited conjugated segm ent, corresponding to a distance of about 30 nm.163,173 The Frster radius R0 for energy transfer from PPESO3 to HMIDC is 49 .169 This distance is smaller than but comparable to typi cal random walk distances (30 nm) indicating that direct Frster transfer can play a significant role in the energy transfer. Hence, in our numerical mode l, the quenching of PPESO3 photoluminescence by the complexed acceptor dye is assumed to occur i) due to the random walk of excitations on the polymer to the complex site, and ii) by a direct long range Frster-type transfer to the complexed acceptor molecules. Energy transfer to free dye molecules in solution can be excluded on the basis of anis otropy measurements as described in the results section. The time-resolved PL-upconversion and tran sient absorption data are simulated. The free variables are the disorder parameter b determining th e excitation hopping dynamics, the average complex distance acomplex, and the scaling factors PL for the fitting of the photoluminescence -upconversion data, and PPESO3, and HMIDC for the fitting of the transient absorption data. Using a single set of parameters it is possible to fit the five independent PL-upconversion and tr ansient absorption datasets.

PAGE 210

194 The overall fits to all the samples with a unique set of parameters are shown as solid lines in Figure 6-5. Good agreement be tween the simulations and the experimental data was achieved for both the upconversion and transient absorption at all quencher concentrations. Table 6-2 summarizes the parameters used for all samples. Table 6-2. Parameters and variables used in the numerical simulations and fitting of the time-resolved PL-upconversion and the transient absorption data.a Fixed parameters Variables 1/kPPESO3 (b) 250 ps b/kT 0.8 1/kHMIDC (c) 1700 ps PL0.33 0.6 PPESO3 1.47 hop,0 (d) 1 ps HMIDC 1.25 cl(e) 6.1 nm a All parameters are identical for the different samples.b Obtained from the PLupconversion of pure PPESO3 in CH3OH, c taken from ref 169, d taken from ref. 182,e taken from ref 183 The distance between complexes formed along the polymer chain (acomplex) obtained from the numeric modeling of the time-resolved quenching dynamics allows one to determine how many polymer repeat units are actually quenched by a single PPESO3/dye complex. The lowe st dye concentration of [Q0] = 0.65 M leads to an integrated quenching of 25% of the polymer repeat units by forming one complex per 104 PRU (see Table 6-3). This corresponds to a bout 26 PRU (35.4 nm) quenched by a single complex. Figure 6-10 shows the most important result of the simulations: the average distance between complexes on the PPESO3 ch ain. The left axis shows the complex distance in hopping steps ( acomplex, in units of cl ) which is the natural unit used in the random walk simulations. By assuming a length of the conjugated segments ( cl ) of about 8-9 phenylene ethynylenes,183 which corresponds to 4.5 polym er repeat units (PRU), the complex distance can be expressed in units of PRU (right axis in Figure 6-10). The

PAGE 211

195 results of the numerical simulation range fr om about 23 segments (104 PRU) in the case of the sample showing the smallest overall quenching (dye concentration 0.65Q M down to 6 segments (27 PRU) for th e sample with the highest quenching concentration 6 M Q Table 6-3. Average distance between two acceptor molecules in complex with the PPESO3. [ Q ] ( M) acomplex (in segments) acomplex (in PRU) 1 segment = 4.5 PRU(a) 0.65 23 104 1.5 11 49 3 8 36 6 6 27 a These values are used to generate the cu rves shown in Figure 6-5. Taken from Ref. 183 0123456 6 12 18 24 30 25 50 75 100 125 acomplex (n)Quencher concentration (M) Numerical Simulation P0= 1.65 M Ka=7.2x105 P0= 34 M Ka=3 x104acomplex (PRU) Figure 6-10. Average distance between quenc her molecules complexed with the PPESO3 on a reciprocal scale as a function of the quencher concentration. The results from the numerical model ar e shown as solid circles (). The linear regime () yields a relatively lower associ ation constant. A good match between theory and the numerical results is achieved by using a reduced conjugated polymer concentration of [ P0]=1.65 M ().

PAGE 212

196 If we assume purely static quenchi ng and a 1:1 PPESO3:HMIDC complex169, the distance between complex sites can also be ev aluated analytically. Using the association equilibrium constant Ka, and the initial CPE and dye concentrations, we obtain the concentration of co mplexes in solution 2 11 0000 0022aaPQKPQK PQPQ (6-2) where P and Q are the corresponding concentrations of uncomplexed molecules and [ PQ ] is the concentration of PPESO3:HMIDC complexes. This equation is derived assuming that all PRU are accessible to fo rm PPESO3:HMIDC complexes. As shown below, this is apparently not the case for solutions with high dye concentration. The ratio of the initial polymer concentration with the concentration of complexes, 0PPQ, gives the average distance between the PPESO3/HMIDC complexes along the polymer chain in PRU, and is direct ly comparable to the complex distance acomplex obtained from the numerical model. Figur e 6-10 compares the numerical simulation results and the ratio 0PQP obtained from the analytical expression for the association equilibrium. At dye concentrations below 1.2 M, the equilibrium equation from an assumed 1:1 PPESO3/HMIDC complex ratio leads to an approximately linear decrease of the inverse complex distance with increasing dye concentration. Using a total polymer concentration of 34 M, equation 6-2 behaves linearly for Ka ~ 3 x 104 M-1. This association constant is consid erably smaller than the one obtained from the Stern-Volmer plot (Figure 69). At dye concentrations above 1.2 M the numerical results deviate si gnificantly from the linear behavior predicted by a 1:1

PAGE 213

197 PPESO3/HMIDC complex. The reason for this early saturation of complex formation is uncertain. It is clear from Figure 6-10 that, at high HMIDC concentrations, the number of complex formed is smaller than predicted by e quation 6-2. One possibili ty is that at these high concentrations, more than one HMIDC mol ecule interacts with a given complex site. This would effectively increase the distance be tween complex sites. Attempts to fit the complex distance by considering complexes with different PPSO3/HMIDC ratios were unsuccessful though the simultaneous presence of complexes with one, two, or more HMIDC molecules cannot be completely ruled out. The onset of the saturation corresponds to an average complex distance of ~30 PRU. Given this large distance it seems unr easonable to attribute the effect to mutual electrostatic repulsion between comp lex sites. It is more likely that the effect arises due to formation of a loose aggregate of the PPE SO3 chain, which effectively reduces the polymer surface area available to the dye in solution. Loose aggregates are defined as aggregates that contain both the polyelectrol yte and the counterions, as opposed to dense aggregates where different branches of the polymer are in Van der Waals contact.184 Within these loose polyelectrolyte aggregates the electrostatic inte ractions between the quencher molecules might be changed. Note that the PPESO3 photoluminescence spectra clearly exclude significant electronic interchain interactions such as stacking of PPESO3 in dilute CH3OH solutions.169 A polymer surface reduced due to the forma tion of loose aggregates corresponds to an effective reduction of the polymer concentr ation available for complex formation with the dye. If we consider that the available c oncentration of PPESO3 to form complexes is smaller than the initial concentration [ P0], we find a reasonable agreement between

PAGE 214

198 equation 6-2 and the numerical modeli ng using an association value of 5-1~7.210 MaK and a polymer concentration of 0~1.65 MP. Apparently the tendency of the polymer chain to form loose aggregates lowers the quenching efficiency. In our previous studies169 on the quenching of PPESO3 by cationic quenchers, we found that quencher induced aggregation of the conjugated polyelectrolytes increases the quenching efficiency significantly. In that case, it is the facilitation of the exciton dynamics through additional interchain hopping of the photoexcitations that enhances the quenching. The modeling of the quenching dynamics allows the evaluation of the individual contributions to the energy transfer given by i) the random-walk of the excitation to the complex site and ii) direct (long-range) Frster-type energy transfer to the complex site. Figure 6-11 shows the time evolution of the in tegrated quenching for each mechanism for the samples quenched by 0.65 and 6 M HMIDC, respectively. For both samples, the random walk gives by far the la rgest contribution to the total energy transfer. Indeed, the random walk seems to be almost the exclus ive pathway. The two contributions have distinctly different dynamics: the Frster pa rt of the energy transfer rate is timeindependent; it depends solely on the Frster radius and the complex concentration. In contrast, the random-walk mechanism is intrin sically time-dependent: the probability of an exciton to hop to a new, distinct segment, i.e. the probability of finding the quencher, decreases with time. This behavior is recogni zed in as a steady increase in the Frster contribution while an initial contribution fo llowed by flat time dependence is observed for the exciton hopping mechanism.

PAGE 215

199 The time dependence found in the random wa lk is a consequence of the quickly decreasing probability of an excitation to hop to a new, so far unvisited site. Even in the limit of a large separation of que ncher sites, the probability to find a complex site decays by 75 % within only 10 hopping steps. Moreover, the decrease of th e average excitation energy due to site disorder leads to a slow down of the hopping rate by a factor of five within 10 ps. This extremely fast decay of th e random-walk driven en ergy transfer leads to a large fraction of the en ergy being transferred on a time scale much faster than the 0 20 40 60 80 a ) 6 M 0.65 M0100200300400500 0.0 0.1 0.2 0.3 0.4 b ) Integrated Energy Transfer (%)t (ps) 6 M 0.65 M Figure 6-11. Individual contri butions to the integrated energy transfer, random walk mediated ( a ) and direct Frster-transfer ( b ), versus time after excitation. For both samples (containing an acceptor concentration of [HMIDC]=0.6 M, and for [HMIDC]=6 M), the random walk process dominates the energy transfer. The random walk driven energy tran sfer can be distinguished by the extremely fast decay of its transfer efficiency, leading to about 50% of the total quenching being completed within less than 1 ps.

PAGE 216

200 intrinsic excited state lifetim e of the donor. Accordingly, the fast excitation hopping is responsible for the dominating prompt quenc hing mechanism found in this material system. For example, for the high quencher concentration sample, at t = 5 ps a 52 % (integrated) quenching has alr eady occurred through the random walk of the excitations, while for the low quencher concentration sample the quenching by random walk has already reached 12%. For both samples, the co ntribution from the Frster mechanism is negligible on this time-scale. The reason for the small contribution by the Frster long-range mechanism is the size of the exciton deloca lization (conjugation length cl ) and its lack of time dependence. Due to the r6 dependence of the Frster mechanism, using a smaller cl parameter leads to a larger Frster contribution to the overall quenching. For example, for a cl corresponding to 4.5 PRU (8-9 rings) the num erical simulations show that at low quencher concentrations, the Frster contri bution accounts for only 1.5% of the total energy transferred to the quenc her. If instead we consider 3.5 PRU per conjugation length (7-8 rings), the Frster contribution rises to 25 % of the overall quenching. In the calculations presented here in we follow Kukula et al183 and use 8-9 rings as the conjugation length (4.5 PRU). Though the Frster component contribution to the overall quenching is <2%, its time-independent characteris tic implies that at long time-scales this mechanism is the major contributor to th e quenching. However, on long time-scales Frster quenching must compete with th e natural decay of the excitation, and consequenctly this transfer mechanism remain s ineffective and it still does not contribute significantly. In conclusion, even if the cl is smaller than the 4.5 PRU used here, the

PAGE 217

201 dominant mechanism responsible for the fast and effective quenching is the random walk of the excitations within the polymer chain. As shown above, the amplification of fluorescence quenching in the PPESO3/HMIDC arises mainly from the rapi d intrachain migration of excitons on the conjugated polyelectrolyte. A reduc tion of the site disorder woul d be expected to lead to significant reduction of excitati on trapping, and thereby furt her enhance the fluorescence quenching efficiency. In our numerical model a vanishing disorder leads to a constant hopping rate. In the limiting case where a poly mer could be prepared with no energetic disorder, the lowest dye concentration of 00.65 MQ would yield a fluorescence quenching efficiency of 97% instead of the 25% found experimentally. The quenchercomplex concentration needed for 50% que nching decreases by a factor of five, corresponding to a dye concentration of only 00.25 MQ Conclusions The fluorescence quenching dynamics in a solution-based system have been investigated using the conj ugated polyelectrolyte PPESO3 as energy-donor and the dye HMIDC as acceptor. Numerical modeling of ps time-resolved fluorescence upconversion and transient absorption measur ements was carried out on a number of samples with systematically varied quencher concentration. In general, the dynamics show that hopping of the photoexcitations to the complex sites is the dominating pathway for fluorescence quenching. Direct Frster tran sfer to the acceptor contributes only about 1.5 % to the total energy transfer at lo w quencher concentrations, and it becomes negligible at elevated quencher concentra tions. Anisotropy measurements lead to the conclusion that all of the quenching occurs with dye molecules bound to the polymer

PAGE 218

202 through ionic association. Simulations of th e intrachain excitation migration allowed us to calculate the actual density of comple xes formed between the ionic donorand acceptor molecules. We found that at low quencher concentrations (below 01.2 QM ) the polyelectrolyte-dye complex density scales approximately linearly with dye concentration, in accordance with the equilibrium equation for a 1:1 complex formation. At higher dye concentrations, how ever, a saturation of the complex density sets in. From the equilibrium equation this behavior would be expected at much higher concentrations, when the dye concentrat ion becomes comparable to the monomer concentration of the polyelectro lyte. We suggest that the form ation of loose aggregates of the polyelectrolyte leads to a strong decrease of the numbe r of polymer sites that are available to form a complex with the dye Since no spectroscopic signature of polyelectrolyte aggregate form ation has been found, here th e term ‘loose’ corresponds to aggregates with an interchain distance bei ng too large to enable electronic interchain interactions (e.g. stacking), but small enough so as to prevent a steric barrier to complex formation. Satisfactory agreement of the complex densities predicted by the equilibrium equation and those calculated from the numerical simulations of timeresolved data is achieved, if the PPESO3 c oncentration used in the equilibrium equation is reduced by a factor of 20. The decr ease of quenching efficiency induced by the formation of loose aggregates is in contra st to the increase of quenching efficiency usually found in case of the formation of de nse aggregates in strongly polar solvents, where interchain migration of excitations is enabled. The results also show that the factor li miting the fluorescence quenching efficiency is trapping of the photoexcitations on th e PPESO3 conjugated polyelectrolyte. The

PAGE 219

203 simulations suggest that a decr ease of site disorder within the conjugated polymer chain well below kBT should lead to an increase of quenc hing efficiency by a factor of 5.

PAGE 220

204 CHAPTER 7 SUMMARY AND PERSPECTIVE Since the discovery of dendrimers first reported in 1985, there has been an immense set of developments in their synt hesis, properties and applications. They represent a key stage in the ongoing evolut ion of macromolecular chemistry. Their success has been possible due to unique arch itectures with accurate positioning of chromophores. Some of the structures incl ude numerous light-collecting chromophores that transfer their energy to a single energy “sink”. In that sense, a dendrimer is reminiscent of the architecture of natural light-harvesting complexes. To maximize the favorable energy transfer inter actions, the selectio n of chromophores is as important as the design of each successive layer. Conjugated dendrimers with a built-in ener gy gradient were studied throughout this dissertation. Our systems consist of symm etrical and unsymmetri cal phenyl ethynylene (PE) dendrimers depending on the substitution at the branching point. We presented a detailed photophysical characterization of these dendrimers by both steady state and time-resolved spectroscopy. The goal of this study was to answer some fundamental questions regarding the time scale and mech anism for the energy transfer process. For example, can we attribute the absorption ba nd structures to speci fic building blocks? What is the role of substitution at the bran ching point? Do excitations at the periphery migrate to the trap in a cascade manner, or are there direct jumps from the periphery to the trap? How can the intermediate steps be probed experimentally? What is the time scale and mechanism of the energy transfer?

PAGE 221

205 Initial studies with a generation one dide ndron described in Chapter 3 showed that the presence of ortho and para substitutions in such unsymmetrical structures supports the initial exciton delocalization (unlike meta subs titution). After excitati on, a change in the excited state surface leads to localization, wh ich is verified with the localized peak around emission wavelength. The bu ilt-in energy gradient results in very efficient energy transfer to the trap and yields a cascade mechanism occurring in a sub-picosecond (250350 fs) time scale. Chapter 4 extended the inve stigation to include larger, generation two didendron which also revealed de localization in the initially excited st ate. We found that while absorption is into delocalized exciton st ates, emission occurs from localized states. The excitation energy migrates to the trap through direct and i ndirect (multi-step) channels. Based on the kinetic model, almost 50% of the energy transfer occurs through the multi-step pathway, but the process is still completed in a sub-picosecond time scale. To our knowledge, we presented the first di rect experimental measurement of energy migration as it goes through an in termediate state in a funnel-t ype dendritic structure. The process of rise and decay of the intermed iate state population was observed by measuring the temporal evolution of the fluorescence at the emission wavelength of the intermediate state. This method provided an unequivocal pr oof of the vectorial na ture of the energy transfer in a ladder-type structure. In a ddition, it was concluded that increasing the generation size for unsymmetrical PE dendrimers will increase and broaden the absorption while maintaining the highly efficient light-harvesting. The symmetrical PE dendrimer was invest igated with time-re solved spectroscopy as complementary to the theoretical and steady state measurements reported in the literature, and the results were described in Chapter 5. In contrast to the unsymmetrical

PAGE 222

206 structures, due to meta branching, exci tations are initially localized on single chromophore units, composed of 2-,3-,4ring PE units. The transient absorption spectra following the selective excitations of these chromophores as part of the nanostar were compared with that of free PE units. Our da ta revealed energy transfer times from hundreds of femtoseconds to tens of picosec onds. Based on the developed kinetic model, we conclude that direct energy transfer takes place with 33.33% chance, while the cascade mechanism has 66.66% probability. We also compared our results, based on the spectral data, with theo retical calculations fr om the literature. We found a factor of 2-3 difference between the measured and the calculated values for the coupling strengths. This disagreement could be improved by a fu ll treatment of Coulombic interaction using TDC and molecular dynamics. In Chapter 6, an independent project is presented. We investigated the role of exciton hopping versus direct energy transfer mechanism within a system of conjugated polyelectrolyte and a cyanin e dye. It was found that rapi d intrachain energy migration towards the complex sites leads to highl y efficient energy transfer, whereas the contribution from direct, long-rang e energy transfer is negligible. As described in this dissertation, having an energy gradient is the key to create efficient light harvesters. For conjugated systems such as PE dendrimers, the substitution at the branching points, whic h actually defines the symmetry of the molecule, produce significantly different optical a nd photophysical properties. We showed that the extent of localization/delocalization of initial excitations depends on the conjugation, and the energy transfer times reveal the strength of the coupling between the chromophores.

PAGE 223

207 The symmetrical dendrimer was proved to be a linear combination of individual PE components. Thus, comparing its dynamics with PE units of different size significantly helped to follow the energy migration. In order to identify the individual chromophores, if they exist, in the unsymmetrical structures it is necessary to compare them with their corresponding analogs. Even though we did not have those analogs our work based on time-resolved spectroscopy and careful model analysis explored the time scale and mechanism of the energy transfer. One of the experiments that could enhance our understanding of these systems would be fluorescence anisotropy measurements. The fluorescence anisotropy dynamics combined with isotropic time resolved and steady state spectroscopy may yield information about the interchromophore energy transfer character. The time scale of de polarization can be correlated with the excitation energy redistribution rate between the chromophores as this process is accompanied by the reorientation of the transition dipole moment. For example fo r delocalized excited states the energy transfer may be inferred as a coherent process and can be related to depolarization rate. Molecular dynamic simulations performed by Krause and Roitberg groups at QTP showed that at room temperat ures phenyl rings are free to rotate around the triple bonds. However, the rings lie essentially in the sa me plane at low temperature (77 K). This change in geometry has a significant effect on the transition density. Therefore, one of the future directions in our lab is to extend th e time-resolved studies to low temperature. Another future enhancement is to use Tran sition Density Cube method for any dendrimer to obtain exact Coulombic in teraction. Especially for unsymmetrical dendrimers, interactions among donor molecules might lead to similar characteri stics as aggregated

PAGE 224

208 molecular assemblies. In this case, the princi ples involved in the op timization of energy transfer are not revealed in a simple way by the absorption and emission spectra. Recently, a generalized Frster theory was prop osed in order to calculate rates of energy transfer in disordered molecular aggregates. It the future, modified forms of Frster can be adapted for modeling dynamics in multichromophoric dendrimers. At the present time, general design principles for efficient light ha rvesting structures can only be revealed by a combination of experiments and theory. Fina lly we hope that our discoveries presented in this dissertation open new possibilities in the design of synthetic light harvesters.

PAGE 225

209 APPENDIX A THE FLUORESCENCE UP-CONVERSION TECHNIQUE The Excitation and Collection of Fluorescence The excitation beam is focused in one surface of the sample cell while the emission is collected from the other surface. As seen in the experimental layout (Figure 2-1), the sample is excited on the back surface. Th e excitation beam is aligned with a 250 with respect to the central axis of the off-axis parabolic mirrors so th at it does not hit the parabolic mirror. First of all, this reduces the possibility of co llecting and upconverting the excitation beam along with the fluorescence. Second, th e time resolution is much better compared to same face excitati on and fluorescence collection, since for fluorescence generated at different positions of the excitation beam, the accumulated changes in group velocity delay is compensated. The fluorescence is collected by a pair of off-axis parabolic mirrors. The first parabolic mirror collects the fluorescence and collimates it, and the second parabolic reflector focuses the fluorescence into the crys tal. The mirrors are 2” in the long axis and purchased from Janos Technology (part number, surface quality). The sample should be in the foci of the first parabolic mirror, while the BBO crystal is designed to be positioned at the foci of the second mirror. We put a pinhole where the sample is positioned and scattered pump beam was co llected to check the alignment of the parabolic mirrors. The initial ca reful efforts to align these mirrors suggested that day to day alignment would be very hard, which will then question the reproducibility of the experiment. These two mirrors are 900, meaning that the optical axis of the mirror and the

PAGE 226

210 position axis of the foci are perpendicular to each other. As shown in Figure A-1, =900, and the optical axis of the mirrors should be parallel to each other to image the fluorescence on the foci of the second mirror. The reflected effective focal length is 152.40 mm for both of the mirrors. Figure A-1. Diagram of off-axis aluminum pa rabolic mirrors used to collect and image the fluorescence of the sample. The position of mirrors with respect to each other is very crucial as well as their position with respect to sample and crystal. To circumvent this painful alignment, we contacted the Astronomy Department at UF, who uses very large off-axis parabolic mirrors for their research. Their expertise w ith such mirrors helped us to design the collection part of the experiment separate ly. The mirrors were pinned to homemade mounts and these mounts were pinned to an extr emely flat tool plate after very careful Central Ray Axis fReflected Effective Focal Length Focal Point D

PAGE 227

211 calculations of positioning with 0.01 mm accu racy. Once the mirrors are placed on the tool plate, they are never to be adjusted. Th is way, the mirrors are always aligned with respect to each other, and a lot of painful a lignment time is saved. When aligning from scratch, the first step is to define the positi on of the sample cell. To begin, the pump beam is needed to define the location of the foci of the parabolic mirrors. It is absolutely crucial that the position of the pump beam is repr oducible from one day to another. The pump beam is defined by several irises with fixed height, and they are the steady components of the setup without further adjustment. As show n in Figure A-2, three irises, separated by as large distance as possible, define the pa th of the pump beam. (Between the first and second irises, a waveplate and polarizer are pl aced). The first step each day should be aligning the pump beam through these 3 irises. Figure A-2. Layout of the alignment beam s and the collection by parabolic mirrors. A steady mount holding a pinhole is restrained (with dowel l pins) on the tool plate in a way that the pinhole would be exactly at the focus of the parabolic mirror 1 (f=152 mm). At the focus of the second parabolic mirro r, another pinhole in a steady (no x,y or z BBO Crystal Focusing Lens Sample IRIS 1 IRIS IRIS 2 IRIS 3 Lens Gate Pump x zFocal Length = 152.4 nm XYZ stage

PAGE 228

212 adjustment) holder is placed. At this point, a quartz plano-convex lens with a focal length of 200 mm is used to pass the pump beam th rough the pinhole 1. The lens is mounted on a translation stage (only in z direction) and positioned to ensure that the pump beam is focused in pinhole 1, which represents the fo cus of the parabolic mirror 1. Using this pinhole is invaluable especia lly when trying to find time zero for the first time. Roughly 90% of the pump beam passes through a 100 m pinhole. There will be some scattered pump light at the pinhole, and the first para bolic mirror will collect and collimate it. Then the second parabolic mirror will focus the sc attered light into the pinhole 2 (when you do this, please note that the collecting negative lens in the experimental setup is not in use). Basically, whatever light source is at the focu s of the first mirror is imaged on the focus of the second mirror. The pump beam here does not hit the parabolic mirror and this significantly reduces the possibi lity of up-converting excitation beam on top of real fluorescence signal. By doing the alignment descri bed to this point, it is ensured that the pump beam is defined and the off-axis pa rabolic mirrors are positioned accurately. The next step is putting the xyz translation stage holding a mount compatible for both a pinhole and sample cell on top of the t ool plate. Before doing so, the steady mount holding the pinhole at the focus of first parabolic mirror s hould be removed. Without changing or replacing anything else, the pinhole in the tran slation stage should replace the position of the previous pinhole. The xyz stage will be aligned so that the scattered light is still collected and focused on the second pinhole. The mount in the stage has been designed in a manner that the pinhole can be re moved and the sample cell is easily placed in exactly the same position. Alignment of the pinhole with the translation stage guarantees that the center of the sample cell is at the focus of the mirror.

PAGE 229

213 Even though the original idea was to put th e nonlinear crystal at the focus of second mirror, our investigations showed that fo r a better phase matching the solid angle between the fluorescence and the gate beam s hould be smaller. Thus, a negative lens (f=-200 mm) was used between the second para bolic mirror and the nonlinear crystal. The lens was mounted to a xyz stage placed on the tool plate. The upconversion crystal was placed about 350 mm apart from the mirror. This crystal can be easily replaced with a pinhole for alignment purposes, especially wh en overlapping the fluorescence and gate beams spatially. One should try to find time zero for th e sum frequency generation with the scattered light from the pinhole where the samp le is positioned. However, in order to fine-tune the negative lens’ posi tion, a very concentrated solution of a dye molecule can be used. We usually used perylene or a very fluorescent polymer (PPE-SO3). The fluorescence that is collected and focused should be easily visible when the room lights are off. The fine-positioning of the sample ce ll can be adjusted so that the pump beam focuses approximately in the center of the cell. If not, there might be some scattered light from the walls of the cell, even some white light generation if the focus is really tight. Since a small volume of the sample is always excited, the fluorescence can not be accounted as a point source. When focused, th e shape of the fluorescence is always a bit unclear, but there is a clear region where the fluorescence is accounted as tightly focused as possible. So far, the pump beam and fluorescence from a real sample are aligned. The sample cell position should not be touched anymore, mo st critically in the z direction. Later on

PAGE 230

214 you can move in x-y direction to get a better spot on the cell, or sometimes an air bubble will appear in the solution and it is best to avoid this bubble during long scans. Alignment of the Gate Beam and Up conversion Crystal Phase Matching The alignment of the gate beam is more straightforward. The residual 800 nm beam of an OPA is used as gate (except SHG of signal used for the experiments explained in Chapter 3). It passes through a delay stage (Physik Instrumente, M-531-DD, Resolution 0.1 m ), a waveplate, and a polarizer. All gol d mirrors are used for optimum spectrum, power and reflectivity. The gate lens (f=300 mm ) is placed on the tool plate. The shorter focal length gives a tighter focus, hence a greater upconversion efficiency. However, the focal areathe gate beam should not be smalle r than the focal volume of the fluorescence on the crystal. We obtained higher upconversio n efficiency when the size of the gate beam was similar (slightly larger) to the si ze of fluorescence. The gate lens, also on the translation stage, should be positioned to optimize the upconversion efficiency, being careful not to focus the gate beam on the BBO crystal (if focused too tight, it will burn the crystal). As mentioned before, the negative lens used to focus the fluorescence allows the solid angle between the gate and fluorescence to be minimized, providing the best phasematching condition in the nonlinear crystal. Ne vertheless, this angl e should not be too small since this will make the spatial separa tion of the up-converted fluorescence and the 400 nm scattered light (from the SHG of the ga te beam in the upconversion crystal) more difficult. In addition, there might be cro ss-correlation signals, especially when the measured fluorescence wavelength is very close to the excitation wavelength.

PAGE 231

215 The pinhole is carefully placed in exactly the position as the upconversion crystal and the gate beam is aligned and focused thr ough it. As mentioned previously, it is wise to use the pump scatter to find time zero. When the pump and gate beam foci are ensured to overlap spatially, both of them will pass through that pinhole. Then the upconversion crystal is placed (replacing the pinhole) and with a translation stage you can move it along the z axis to exactly obtain the same sp atial overlap. The gate beam is vertically polarized and it controls the polarization angl e of the crystal. The phase-matching angle will be optimized when the crystal is rotated around the vertical axis. Usually the sum frequency of the pump a nd gate beam is visible to the eye. However, if both the crystal angle and the time zero are not known, one can approximately find the crystal angles by finding the second harmonic generation (SHG) of the pump and gate beam independently. Th ese SHG signals should appear easily as long as they are not in the deep UV region. The phase-matching angle for sum frequency mixing is approximately in the midway between the angles for doubling of each beam. For the experiments performed for this diss ertation, the pump beam was usually below 400 nm, where doubling with type I BBO crystal was not possibl e. Instead, when we did not know either phase-matching angle or time zero, we set the OPA output to be the same wavelength as the fluorescence and obtained SHG of that beam. This will be very useful when trying to get real signal from the sample. After the ti me zero is set, the OPA is aligned to produce the pump wavelength (usually only angle tuning of the crystals in the OPA is enough to switch between real pum p and fluorescence wavelength). The only variable is then the phase matching angle fo r the pump beam. Once the mixing signal is seen or detected via the PM T, the gate beam can be tw eaked, and the position of the

PAGE 232

216 crystal and the delay stage can be adjusted to optimize the signal. It is important to recall that once the pump beam is ali gned, it should not be adjusted. We can summarize what has been done so far if one follows the procedure described here: First, the pump beam is al igned through the irises and the pinhole where the sample cell will also be positioned. The pa rabolic mirrors are never to be moved. The gate beam is tightly focused 2-3 cm after the crystal position with the pump beam and carefully overlapped with the pump using a pinh ole. This pinhole is then replaced with the upconversion crystal and time zero is found s canning the delay stage. In any case, for every experiment, the phase matching angle, the spatial and temporal overlap of the beams would require a little bit of tweaking. Detection of Upconverted Fluorescence After the upconversion crystal, there is an iris rejecti ng the cross-correlation and any other scattered light, such as the SHG of 800 nm. Then, a quartz lens is positioned to collimate the signal, which is always in the UV. Two UV enhanced aluminum mirrors direct the signal to the monoc hromator. After the crystal, one can use any filters and dielectric optics to get as many photons as possible to the detector since the time resolution is not eff ected by the optics after the crystal. The time resolution is only determined by the mixing process in the crysta l as well as the optics before the crystal. Before the monochromator there is another lens that will focus the signal onto the entrance slit. One issue is that as the signal ge ts closer to the 400 nm, there will be more background due to 400 nm generated at th e upconversion crystal. Even though the photons are not traveling in the direction of the signal, there will always be enough 400 nm photons reaching the detector. To minimi ze this contribution, we use UG-11 filters and a solar blind PMT that has on ly 0.01% efficiency around 400 nm.

PAGE 233

217 UV-Light Compression For most of the experiment s, the pump pulse was generated in the UV region from the fourth harmonic generation of either th e signal or idler output of an OPA. The measured cross-correlation showed that the time resolution of the upconversion experiment is around 400-500 fs without a ny pulse compression. Despite the broad tunability of the OPA, it generates very l ong UV pulses with very narrow bandwidth. The width of the UV spectrum is limited by the th ickness of the BBO crystal (second BBO in the OPA, 0.5 mm) due to phase matching. Having a 2-3 nm bandwidth makes it even harder to compress these UV beams, but in or der to obtain an adequate amount of pump energy it is necessary to use a crystal of such a thickness. The compression of the pulses to a minimum duration is performed via UV gr ade isosceles fused silica prism pair (CVI, catalog Number: IB-12.4-69.1-UV). The apex angle of th e prism is cut such that the angle of incidence is the Brewster angle at the in cident wavelength. Thus reflection losses for p-polarized beam are extremely small and transmission efficiency is 98 %. The OPA output between 300 and 400 nm is horizontally po larized (FHG of signal), so the loss due to prisms is minimal. This prism compre ssor transmits 92% of the incoming UV beam. The compressed UV pulses are used as the pum p beam and the typical cross-correlation (CC) signals are measured as the sum freque ncy generation of pump pulse with the gate pulse. The pulse width of the gate beam is measured via autocorrelation. The values for the pulse length of the UV pump pulses can be calculated from the width of the CC signals. The value of the meas ured CC signal is important because it presents the time resolution of the experiment. It is possible to achieve 225 fs FWHM of CC signal when the pump beam is 310 nm. Using a prism compressor for UV pulses definitely helps with improving the time resolution of the upconversion setup.

PAGE 234

218 The time-resolution of the setup is basica lly limited by the temporal width of the optical gating and the excitation pulses. However, for subpicosecond pulses in our system, the group velocity dispersion ( GVD) induced by the nonlinear upconversion crystal and by the various optics will most probably alter the time re solution. Therefore, we try to use a thin nonlinear crystal (0.3 mm thick), along with the parabolic mirrors (instead of lenses, or any other transmitting optics) for the collection of luminescence.

PAGE 235

219 APPENDIX B DATA ANALYSIS Analysis of Transient Absorption Changes We use broadband transient absorption spect roscopy to monitor the energy transfer towards a particular trap. Absorption change A induced by a first exci ting light pulse is recorded as a function of wavelength by a second delayed probe pulse at time t. Recording the difference spectra over a long observation time (femtoseconds to milliseconds) results in a two dimensional ar ray with many individual data points. In order to deduce reasonable information out of the raw data, models has to be used to describe the photophysical event. The model parameters such as time constants are determined by fitting the model to the experimental data points. In many cases, the reaction schemes can be simulated by a sequence of intermediate states with well-defined absorption spectra and time components. In this appendix we will discuss a numerical procedure called Singular Value Decomposition (SVD) method used for data analysis. The absorption change defined as A( ,t) = A( ,t) – A( ,0) is a continous function of the probe wavelength and the delay time t. By changing the settings of spectrometer/detector, and time delay betw een excitation and probing pulses, this continuous function becomes a matrix (Aij), where the absorption changes at fixed probe wavelengths i form the columns and those at fixed delay times tj form the rows: A( i,tj) = (Aij) In a typical time-resolved absorption e xperiment, the matrix A will be built up of N probe wavelengths and Nt delay times. The wavelength ra nge used in the experiment

PAGE 236

220 is from ~ 300 nm to ~ 580 nm with spectral resolution of 0.275 nm. In order to consider kinetics extending over several orders of magnitude we use logarithmically-spaced time delays from ~30 fs to 20 ps. We first plot ce rtain rows or columns of this data matrix by a Labview program developed in our lab (by J.Mueller). Comparing different rows, one can see how the spectrum relaxes in time. Th e absorption difference as a function of time at a fixed wavelength can be observed by pl otting a certain column of A. These two visual examinations of the data (espec ially with the mentioned Labview program) provide the experimentalist with a first gue ss on the involved pr ocesses, i.e. time constants, spectral characteristics. These vi sually obtained results are then supported and expanded by SVD. SVD is a powerful matrix technique that filters out experimental noise and identifies the independently evolving transi ent species from the difference spectra. We use the SVD algorithm in MATLAB. The data matrix (Aij) = A( i,tj) is decomposed into the product of three matrices A=U.S.VT Where U and V are orthogonal matrices of dimensions N x Nt and Nt x Nt respectively. S is an Nt x Nt square diagonal matrix, containing the singular values. Each of these values determine s how much the corresponding colu mns of U and V contribute to the reconstruction of A. For an intuitiv e understanding of SVD, note that U and V are the matrix eigenvectors for AAT and ATA, respectively. The singular value of A are the square roots of shared eigenvalues E of AAT and ATA, which are the diagonal elements of S. The mathematical relationships follow as: AAT = U.E.UT

PAGE 237

221 ATA= V.E.VT, and S= E1/2. N x N matrix AAT contains overlap of kine tic vectors for all pairs of wavelengths, while Nt x Nt matrix ATA contains time-pairwise overlap of spectral vectors. Thus, the columns of U contain th e minimal set of orthonormal basis for the row space of A. Likewise, the columns of V contain the orthonormal basis for the column space of A. hence, SVD identifies the ma thematically independent bases of data, spectrally in U( ) and temporally V(t). We select the usable components of U( ) and V(t) and determine the number of components contributing to the set of spectra from the magnitudes of the singular values of A ( dia gonal elements of S) In the absence of measurement noise, the number of independe nt components would equal the number of non zero singular values. However, all si ngular values are non-zero in the real experimental data. In most cases the magnit ude of singular values, combined with the evaluation of shapes of the corresponding column vectors of U and V will provide enough information to find the number of components. Of this is not the case, certain statistical tests cam be applied to find the number of components. It has to be pointed out that SVD is a purely mathematical method for determining the number of independent vectors in a matrix. As a result, chemically different species that have either equal spectra or equal ki netics will show up as one component. Another remark that should be made is that SV is in principle not suited for components whose spectra shift and/or broaden in time. We first identify the dominant singular values, which could be an extremely difficult decision. The number of dominant singular values determines the number of

PAGE 238

222 time constants used to describe the data. For example, during the SVD analysis of 2G1-mOH molecule, it was hard to decide the number of singular va lues. The initial visualization of the data with Labview, and a simple, yet very useful technique reported for the inspection of two dimensional data sets from transient absorption spectroscopy helped to obtain reasonable starting inform ation for further numerical treatment. We observed a 5-6 ps vibrational relaxation com ponent resulting in a spectral shift. This component is removed from the data set When SVD is applied again on the new data set, it reveals reasonable nu mber of singular values. Thus we only have two relevant SVD components for 2G1-m-OH data: column vectors U1 and V1 associated with singular value S11, and column vectors U2 and V2 associated with singular value S22. Thus, we can reduce the A matrix and e xpress it as the following: A( ,t)= U1 V1 T + (S22/S11) U2 V2 T We apply a kinetic model, which has well-defined population dynamics for each state. The time constants are obtained a nd we can reconstruct the noise-filtered absorption spectra at any delay time. We try to minimize the difference between the reconstructed data via the model, and the actual data in two dimensions. The following section shows the SVD analysis of the molecules investigated through this dissertation and the difference absorption spectra at any delay time is reconstructed using the kinetic models propos ed in Chapters 3,4 and 5 for the relevant molecules.

PAGE 239

223 2G1-m-OH Figure B-1. Singular values Figure B-2. Transient spectra of SVD output. 350 400 450 500 550 -0.04 -0.02 0 0.02 (nm)U1Weighted U1 S = 10.8664 27-Sep-2005 350 400 450 500 550 -0.02 0 0.02 0.04 0.06 0.08 (nm)U2Weighted U2 S = 0.59609 27-Sep-2005 350 400 450 500 550 -0.1 -0.05 0 0.05 (nm)U3Weighted U3 S = 0.19692 27-Sep-2005 350 400 450 500 550 -0.15 -0.1 -0.05 0 0.05 (nm)U4Weighted U4 S = 0.14147 27-Sep-2005 10 20 30 40 50 60 70 10-3 10-2 10-1 100 Component No.S/S1

PAGE 240

224 Figure B-3. Dynamics of SVD output. Figure B-4. Model fits for the relevant components with large singular values. -20 -10 0 10 20 30 40 50 -0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 timeIntensity k = 3.8127 0.0012396 -0.24252 data col 2 27-Sep-2005 -20 -10 0 10 20 30 40 50 -1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 timeIntensity k = 3.8127 0.0012396 -0.24252 data col 1 27-Sep-2005 0 20 40 -1.5 -1 -0.5 0 Delay Time (ps)s*vSV1 27-Sep-2005 0 20 40 -0.2 -0.1 0 Delay Time (ps)s*vSV2 27-Sep-2005 0 20 40 -0.1 -0.05 0 0.05 Delay Time (ps)s*vSV3 27-Sep-2005 0 20 40 -0.02 0 0.02 0.04 0.06 Delay Time (ps)s*vSV4 27-Sep-2005

PAGE 241

225 Figure B-5. Reconstr ucted versus real A as a function of wavelength. 350 400 450 500 550 -10 -5 0 5 x 10-3 t = -0.25 ps 350 400 450 500 550 0 10 20 x 10-3 t = -0.15 ps 350 400 450 500 550 -0.02 -0.01 0 0.01 t = -0.05 ps 350 400 450 500 550 -0.04 -0.02 0 0.02 t = 0.05 ps 350 400 450 500 550 -0.04 -0.02 0 0.02 t = 0.1 ps 350 400 450 500 550 -0.04 -0.02 0 0.02 t = 0.15 ps 350 400 450 500 550 -0.04 -0.02 0 0.02 0.04 t = 0.25 ps 350 400 450 500 550 -0.05 0 0.05 t = 0.45 ps 350 400 450 500 550 -0.05 0 0.05 t = 8 ps 350 400 450 500 550 -0.05 0 0.05 t = 13 ps 350 400 450 500 550 -0.05 0 0.05 t = 15 ps 350 400 450 500 550 -0.05 0 0.05 t = 18 ps

PAGE 242

226 Figure B-6. Reconstr ucted versus real A as a function of time. 0 20 40 0 0.02 0.04 0.06 = 482.9846 nm 0 20 40 0 0.02 0.04 0.06 = 496.7461 nm 0 20 40 0 0.02 0.04 0.06 = 524.2691 nm 0 20 40 0 0.02 0.04 0.06 = 551.7921 nm 0 20 40 0 5 10 x 10-3 = 427.9387 nm 0 20 40 0 0.01 0.02 0.03 = 441.7002 nm 0 20 40 0 0.01 0.02 0.03 0.04 = 455.4617 nm 0 20 40 0 0.02 0.04 = 469.2232 nm 0 20 40 -0.02 0 0.02 = 331.6083 nm 0 20 40 -0.03 -0.02 -0.01 0 = 338.489 nm 0 20 40 -0.02 -0.01 0 = 352.2505 nm 0 20 40 -0.04 -0.02 0 = 386.6543 nm

PAGE 243

227 2G2-m-OH Figure B-7. Singular values Figure B-8. Transient spectra of SVD output. 350 400 450 500 550 -0.06 -0.04 -0.02 0 0.02 (nm)U1Weighted U1 S = 18.2232 19-Sep-2005 350 400 450 500 550 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 (nm)U2Weighted U2 S = 0.76453 19-Sep-2005 350 400 450 500 550 0 0.02 0.04 0.06 (nm)U3Weighted U3 S = 0.30144 19-Sep-2005 350 400 450 500 550 -0.05 0 0.05 (nm)U4Weighted U4 S = 0.22225 19-Sep-2005 10 20 30 40 50 60 70 10-3 10-2 10-1 100 Component No.S/S1

PAGE 244

228 Figure B-9. Dynamics of SVD output. Figure B-10. Model fits for the relevant components with large singular values. -20 -10 0 10 20 30 40 50 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 timeIntensity k = 4.2866 0.00046061 0.15328 data col 2 19-Sep-2005 -20 -10 0 10 20 30 40 50 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 timeIntensity k = 4.2866 0.00046061 0.15328 data col 1 19-Sep-2005 0 20 40 -2 -1 0 Delay Time (ps)s*vSV1 19-Sep-2005 0 20 40 0 0.1 0.2 Delay Time (ps)s*vSV2 19-Sep-2005 0 20 40 -0.05 0 0.05 Delay Time (ps)s*vSV3 19-Sep-2005 0 20 40 -0.05 0 0.05 Delay Time (ps)s*vSV4 19-Sep-2005

PAGE 245

229 Figure B-11. Reconstructed versus real A as a function of wavelength. 350 400 450 500 550 -0.05 0 0.05 0.1 0.15 t = 8 ps 350 400 450 500 550 -0.05 0 0.05 0.1 0.15 t = 13 ps 350 400 450 500 550 -0.05 0 0.05 0.1 0.15 t = 25 ps 350 400 450 500 550 -0.05 0 0.05 0.1 0.15 t = 40 ps 350 400 450 500 550 -0.04 -0.02 0 0.02 0.04 0.06 0.08 t = 0.05 ps 350 400 450 500 550 -0.05 0 0.05 0.1 t = 0.15 ps 350 400 450 500 550 -0.05 0 0.05 0.1 0.15 t = 0.352 ps 350 400 450 500 550 -0.05 0 0.05 0.1 0.15 t = 0.55 ps 350 400 450 500 550 -10 -5 0 5 x 10-3 t = -0.35 ps 350 400 450 500 550 -5 0 5 x 10-3 t = -0.25 ps 350 400 450 500 550 -0.01 0 0.01 t = -0.15 ps 350 400 450 500 550 -0.02 0 0.02 0.04 t = -0.05 ps

PAGE 246

230 Figure B-12. Reconstructed versus real A as a function of time. 0 2 4 0 0.05 0.1 = 530.5994 nm 0 2 4 0 0.05 0.1 0.15 = 544.3609 nm 0 2 4 0 0.05 0.1 0.15 = 558.1224 nm 0 2 4 0 0.05 0.1 0.15 = 571.8839 nm 0 2 4 0 0.01 0.02 = 461.792 nm 0 2 4 0 0.01 0.02 0.03 0.04 = 475.5534 nm 0 2 4 0 0.02 0.04 0.06 = 489.3149 nm 0 2 4 0 0.05 0.1 = 516.8379 nm 0 2 4 -0.04 -0.03 -0.02 -0.01 0 = 351.7001 nm 0 2 4 -0.04 -0.02 0 = 379.2231 nm 0 2 4 -0.04 -0.03 -0.02 -0.01 0 = 406.746 nm 0 2 4 -15 -10 -5 0 5 x 10-3 = 437.0213 nm

PAGE 247

231 Nanostar Excited at 310 nm Figure B-13. Singular values Figure B-14. Transient spectra of SVD output. 350 400 450 500 550 -0.06 -0.04 -0.02 0 0.02 0.04 (nm)U1Weighted U1 S = 1.9739 05-Jan-2006 350 400 450 500 550 -0.02 0 0.02 0.04 0.06 0.08 (nm)U2Weighted U2 S = 1.7966 05-Jan-2006 350 400 450 500 550 -0.02 0 0.02 0.04 0.06 0.08 (nm)U3Weighted U3 S = 0.24303 05-Jan-2006 350 400 450 500 550 -0.05 0 0.05 0.1 0.15 (nm)U4Weighted U4 S = 0.10734 05-Jan-2006 10 20 30 40 50 60 70 10-2 10-1 100 Component No.S/S1

PAGE 248

232 Figure B-15. Dynamics of SVD output. Figure B-16. Model fits for the relevant components with large singular values. 0 5 10 15 20 25 -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 timeIntensity k = 0.10246 2 0.056345 3.43 -0.47486 data col 4 05-Jan-2006 0 5 10 15 20 25 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 timeIntensity k = 0.10246 2 0.056345 3.43 -0.47486 data col 3 05-Jan-2006 0 5 10 15 20 25 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 timeIntensity k = 0.10246 2 0.056345 3.43 -0.47486 data col 2 05-Jan-2006 0 5 10 15 20 25 -0.4 -0.35 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 timeIntensity k = 0.10246 2 0.056345 3.43 -0.47486 data col 1 05-Jan-2006 0 20 40 -0.3 -0.2 -0.1 0 Delay Time (ps)s*vSV1 05-Jan-2006 0 20 40 -0.2 0 0.2 Delay Time (ps)s*vSV2 05-Jan-2006 0 20 40 -0.06 -0.04 -0.02 0 0.02 Delay Time (ps)s*vSV3 05-Jan-2006 0 20 40 0 0.02 0.04 0.06 Delay Time (ps)s*vSV4 05-Jan-2006

PAGE 249

233 Figure B-17. Reconstructed versus real A as a function of wavelength. 350 400 450 500 550 -0.03 -0.02 -0.01 0 0.01 t = 8.4749 ps 350 400 450 500 550 -0.03 -0.02 -0.01 0 0.01 t = 13.4749 ps 350 400 450 500 550 -0.03 -0.02 -0.01 0 0.01 t = 15.4749 ps 350 400 450 500 550 -0.03 -0.02 -0.01 0 0.01 t = 18.4749 ps 350 400 450 500 550 -0.01 0 0.01 0.02 t = 0.57486 ps 350 400 450 500 550 -0.01 0 0.01 0.02 0.03 t = 0.62486 ps 350 400 450 500 550 -0.01 0 0.01 0.02 t = 0.72486 ps 350 400 450 500 550 -0.01 0 0.01 0.02 t = 0.92486 ps 350 400 450 500 550 -4 -2 0 2 4 x 10-3 t = -0.22514 ps 350 400 450 500 550 0 5 10 15 x 10-3 t = -0.025143 ps 350 400 450 500 550 -0.01 -0.005 0 0.005 0.01 t = 0.12486 ps 350 400 450 500 550 -0.02 -0.01 0 0.01 0.02 t = 0.27486 ps

PAGE 250

234 Figure B-18. Reconstructed versus real A as a function of time. 0 10 20 30 -0.04 -0.02 0 = 483.8103 nm 0 10 20 30 -0.01 0 0.01 0.02 = 497.5718 nm 0 10 20 30 -0.01 0 0.01 0.02 = 525.0948 nm 0 10 20 30 0 0.01 0.02 = 552.6178 nm 0 10 20 30 0 2 4 6 8 x 10-3 = 428.7644 nm 0 10 20 30 -10 -5 0 5 x 10-3 = 442.5259 nm 0 10 20 30 -5 0 5 x 10-3 = 456.2874 nm 0 10 20 30 -20 -10 0 x 10-3 = 470.0489 nm 0 10 20 30 -10 -5 0 x 10-3 = 346.1955 nm 0 10 20 30 -4 -2 0 2 4 6 8 x 10-3 = 379.2231 nm 0 10 20 30 0 5 10 x 10-3 = 395.7368 nm 0 10 20 30 0 2 4 6 8 x 10-3 = 415.0029 nm

PAGE 251

235 LIST OF REFERENCES (1) Tomalia, D. A.; Naylor, A. M.; Goddard, W. A., III. Angewandte Chemie 1990, 102 119. (2) Frechet, J. M. J. Science 1994, 263 1710. (3) Newkome, G. R.; Moorefield, C. N. Comprehensive Supramolecular Chemistry 1996, 10 777. (4) Newkome, G. R.; Moorefield, C. N.; Vogtle, F. Dendrimers and Dendrons: Concepts, Syntheses, Applications 2nd ed.; Wiley-VCH: New York, 2001. (5) Flory, P. J. Journal of the Americ an Chemical Society 1952, 74 2718. (6) Flory, P. J. Principles of Polymer Chemistry Cornell University: Ithaca, 1953. (7) Buhleier, E.; Wehner, W.; Voegtle, F. Synthesis 1978, 155. (8) Tomalia, D. A.; Dewald, J. R. Dens e star polymers having two dimensional molecular diameter; United States patent application 85-757546, 1986 (9) Tomalia, D. A.; Baker, H.; Dewald, J.; Hall, M.; Kallos, G.; Martin, S.; Roeck, J.; Ryder, J.; Smith, P. Macromolecules 1986, 19 2466. (10) Tomalia, D. A.; Baker, H.; Dewald, J.; Hall, M.; Kallos, G.; Martin, S.; Roeck, J.; Ryder, J.; Smith, P. Polymer Journal 1985, 17 117. (11) Newkome, G. R.; Yao, Z.; Baker, G. R.; Gupta, V. K. Journal of Organic Chemistry 1985, 50 2003. (12) Woerner, C.; Muelhaupt, R. Angewandte Chemie 1993, 105 1367. (13) de Brabander-van den Berg E. M. M.; Meijer, E. W. Angewandte Chemie 1993, 105 1370. (14) Hawker, C.; Frechet, M. J. Journal of the Chemical Society, Chemical Communications 1990, 1010. (15) Hawker, C. J.; Frechet, J. M. J. Journal of the American Chemical Society 1990, 112 7638.

PAGE 252

236 (16) Newkome, G. R.; Moorefield, C. N.; Vogtle, F. Dendrimers Wiley-WCH: New York, 2001. (17) Bosman, A. W.; Janssen, H. M.; Meijer, E. W. Chemical Reviews 1999, 99 1665. (18) Tomalia, D. A.; Baker, H.; Dewald, J.; Hall, M.; Kallos, G.; Martin, S.; Roeck, J.; Ryder, J.; Smith, P. Polymer Journal 2002, 34 132. (19) Ortiz, W.; Roitberg, A. E.; Krause, J. L. Journal of Physical Chemistry B 2004, 108 8218. (20) Grayson, S. M.; Frechet, J. M. J. Chemical Reviews 2001, 101 3819. (21) Moore, J. S.; Xu, Z. F. Macromolecules 1991, 24 5893. (22) Moore, J. S. Accounts of Chemical Research 1997, 30 402. (23) Morgenroth, F.; Berresheim, A. J.; Wagner, M.; Mullen, K. Chemical Communications 1998, 1139. (24) Morgenroth, F.; Kuebel, C.; Muellen, K. Journal of Materials Chemistry 1997, 7 1207. (25) Morgenroth, F.; Reuther, E.; Muellen, K. Angewandte Chemie, International Edition1997, 36 631. (26) Hawker, C. J.; Wooley, K. L.; Frechet, J. M. J. Journal of the Chemical Society, Perkin Transactions 1: Organic and Bio-Organic Chemistry (1972 1999) 1993, 1287. (27) Mattei, S.; Seiler, P.; Diederich, F.; Gramlich, V. Helvetica Chimica Acta 1995, 78 1904. (28) Stevelmans, S.; Hest, J. C. M. v.; Jansen J. F. G. A.; Van Boxtel, D. A. F. J.; de Berg, E. M. M.; Meijer, E. W. Journal of the American Chemical Society 1996, 118 7398. (29) Jansen, J. F. G. A.; Meijer, E. W. ; de Brabander-van den Berg, E. M. M. Journal of the American Chemical Society 1995, 117 4417. (30) Jansen, J. F. G. A.; de Brabander van den Berg, E. M. M.; Meijer, E. W. Science (Washington, D. C.) 1994, 266 1226. (31) Jansen, J. F. G. A.; de Brabandervan den Berg, E. M. M.; Meijer, E. W. Recueil des Travaux Chimiques des Pays-Bas 1995, 114 225. (32) Naylor, A. M.; Goddard, W. A., III; Kiefer, G. E.; Tomalia, D. A. Journal of the American Chemical Society 1989, 111 2339.

PAGE 253

237 (33) Kawa, M.; Frechet, J. M. J. Chemistry of Materials 1998, 10 286. (34) Xu, Z.; Moore, J. S. Acta Polymerica 1994, 45 83. (35) Xu, Z. F.; Kahr, M.; Walker, K. L.; Wilkins, C. L.; Moore, J. S. Journal of the American Chemical Society 1994, 116 4537. (36) Adronov, A.; Frechet, J. M. J. Chemical Communications 2000, 1701. (37) Frechet, J. M. J. Macromolecular Symposia 2003, 201 11. (38) Frechet, J. M. J. Journal of Polymer Science, Part A: Polymer Chemistry 2003, 41 3713. (39) Aida, T.; Jiang, D. L.; Yashima, E.; Okamoto, Y. Thin Solid Films 1998, 331 254. (40) Hofkens, J.; Latterini, L.; De Belder, G.; Gensch, T.; Maus, M.; Vosch, T.; Karni, Y.; Schweitzer, G.; De Schryver, F. C.; Hermann, A.; Mullen, K. Chemical Physics Letters 1999, 304 1. (41) Weil, T.; Wiesler, U. M.; Herrmann, A.; Bauer, R.; Hofkens, J.; De Schryver, F. C.; Mullen, K. Journal of the American Chemical Society 2001, 123 8101. (42) Crooks, R. M.; Zhao, M.; Sun, L.; Chechik, V.; Yeung, L. K. Accounts of Chemical Research 2001, 34 181. (43) Brunner, H. Journal of Organometallic Chemistry 1995, 500 39. (44) Lee, J.-J.; Ford, W. T. Journal of the Americ an Chemical Society 1994, 116 3753. (45) Crooks, R. M.; Ricco, A. J. Accounts of Chemical Research 1998, 31 219. (46) Larsen, J.; Brueggemann, B.; Polivka, T. ; Sundstroem, V.; Aakesson, E.; Sly, J.; Crossley, M. J. Journal of Physical Chemistry A 2005, 109 10654. (47) Larsen, J.; Andersson, J.; Polivka, T.; Sly, J.; Crossley, M. J.; Sundstroem, V.; Akesson, E. Chemical Physics Letters 2005, 403 205. (48) Hu, X.; Damjanovic, A.; Ritz, T.; Schulten, K. Proceedings of the National Academy of Sciences of th e United States of America 1998, 95 5935. (49) Denti, G.; Campagna, S.; Serroni, S.; Ciano, M.; Balzani, V. Journal of the American Chemical Society 1992, 114 2944. (50) Balzani, V.; Campagna, S.; Denti, G.; Juris, A.; Serroni, S.; Venturi, M. Accounts of Chemical Research 1998, 31 26.

PAGE 254

238 (51) Balzani, V.; Ceroni, P.; Gestermann, S.; Gorka, M.; Kauffmann, C.; Maestri, M.; Vogtle, F. Chemphyschem 2000, 1 224. (52) Balzani, V.; Ceroni, P.; Juris, A.; Venturi, M.; Campagna, S.; Puntoriero, F.; Serroni, S. Coordination Chemistry Reviews 2001, 219 545. (53) Balzani, V.; Vogtle, F. Comptes Rendus Chimie 2003, 6 867. (54) Balzani, V.; Ceroni, P.; Maestri, M.; Vicinelli, V. Current Opinion in Chemical Biology 2003, 7 657. (55) Balzani, V.; Ceroni, P.; Maestri, M. ; Saudan, C.; Vicinelli, V. Luminescent dendrimers. Recent advances. In Dendrimers V: Functional and Hyperbranched Building Blocks, Photophysical Properties, Applica tions in Materials and Life Sciences 2003; Vol. 228; pp 159. (56) Jiang, D.-L.; Aida, T. Journal of the American Chemical Society 1998, 120 10895. (57) Vogtle, F.; Gorka, M.; Hesse, R.; Ceroni, P.; Maestri, M.; Balzani, V. Photochemical & Photobiological Sciences 2002, 1 45. (58) Gilat, S. L.; Adronov, A.; Frechet, J. M. J. Abstracts of Papers of the American Chemical Society 1997, 214 42. (59) Gilat, S. L.; Adronov, A.; Frechet, J. M. J. Angewandte Chemie-International Edition 1999, 38 1422. (60) Serin Jason, M.; Brousmiche Darryl, W.; Frechet Jean, M. J. Chemical communications 2002, 2605. (61) Bar-Haim, A.; Klafter, J.; Kopelman, R. Journal of the American Chemical Society 1997, 119 6197. (62) Bar-Haim, A.; Klafter, J. Journal of Luminescence 1998, 76-7 197. (63) Adronov, A.; Gilat, S. L.; Frechet, J. M. J.; Ohta, K.; Neuwahl, F. V. R.; Fleming, G. R. Journal of the American Chemical Society 2000, 122 1175. (64) Weil, T.; Reuther, E.; Mullen, K. Angewandte Chemie, In ternational Edition 2002, 41 1900. (65) Weil, T.; Reuther, E.; Beer, C.; Muellen, K. Chemistry--A European Journal 2004, 10 1398. (66) Cotlet, M.; Vosch, T.; Habuchi, S.; Weil, T.; Muellen, K.; Hofkens, J.; De Schryver, F. Journal of the American Chemical Society 2005, 127 9760.

PAGE 255

239 (67) Serin, J. M.; Brousmiche, D. W.; Frechet, J. M. J. Chemical Communications (Cambridge, United Kingdom) 2002, 2605. (68) Swallen, S. F.; Shi, Z. Y.; Tan, W. H.; Xu, Z. F.; Moore, J. S.; Kopelman, R. Journal of Luminescence 1998, 76-7 193. (69) Shortreed, M. R.; Swallen, S. F.; Shi, Z. Y.; Tan, W. H.; Xu, Z. F.; Devadoss, C.; Moore, J. S.; Kopelman, R. Journal of Physical Chemistry B 1997, 101 6318. (70) Swallen, S. F.; Kopelman, R.; Moore, J. S.; Devadoss, C. Journal of Molecular Structure 1999, 485-486 585. (71) Melinger, J. S.; Pan, Y.; Kleiman, V. D.; Peng, Z.; Davis, B. L.; McMorrow, D.; Lu, M. Journal of the American Chemical Society 2002, 124 12002. (72) Kopelman, R.; Shortreed, M.; Shi, Z. Y. ; Tan, W. H.; Xu, Z. F.; Moore, J. S.; BarHaim, A.; Klafter, J. Physical Review Letters 1997, 78 1239. (73) Moore, J. S.; Xu, Z. Macromolecules 1991, 24 5893. (74) Ortiz, W.; Krueger, B. P.; Kleiman, V. D.; Krause, J. L.; Roitberg, A. E. Journal of Physical Chemistry B 2005, 109 11512. (75) Kleiman, V. D.; Melinger, J. S.; McMorrow, D. Journal of Physical Chemistry B 2001, 105 5595. (76) Tretiak, S.; Chernyak, V.; Mukamel, S. Journal of Physical Chemistry B 1998, 102 3310. (77) Chernyak, V.; Poliakov, E. Y.; Tretiak, S.; Mukamel, S. Journal of Chemical Physics 1999, 111 4158. (78) Minami, T.; Tretiak, S.; Chernyak, V.; Mukamel, S. Journal of Luminescence 2000, 87-9 115. (79) Mukamel, S.; Tretiak, S.; Wagersreiter, T.; Chernyak, V. Science 1997, 277 781. (80) Kopelman, R.; Shortreed, M.; Shi, Z.-Y.; Tan, W.; Xu, Z.; Moore, J. S.; BarHaim, A.; Klafter, J. Physical Review Letters 1997, 78 1239. (81) Devadoss, C.; Bharathi, P.; Moore, J. S. Journal of the American Chemical Society 1996, 118 9635. (82) Shortreed, M. R.; Shi, Z. Y.; Kopelman, R. Molecular Crystals and Liquid Crystals Science and Technology Section aMolecular Crystals and Liquid Crystals 1996, 283 95. (83) Tada, T.; Nozaki, D.; Kondo, M.; Yoshizawa, K. Journal of Physical Chemistry B 2003, 107 14204.

PAGE 256

240 (84) Thompson, A. L.; Gaab, K. M.; Xu, J. J.; Bardeen, C. J.; Martinez, T. J. Journal of Physical Chemistry A 2004, 108 671. (85) Harcourt, R. D.; Scholes, G. D.; Ghiggino, K. P. Journal of Chemical Physics 1994, 101 10521. (86) Shortreed, M. R.; Swallen, S. F.; Shi, Z.-Y.; Tan, W.; Xu, Z.; Devadoss, C.; Moore, J. S.; Kopelman, R. Journal of Physical Chemistry B 1997, 101 6318. (87) Varnavski, O. P.; Ostrowski, J. C.; Sukhomlinova, L.; Twieg, R. J.; Bazan, G. C.; Goodson, T. Journal of the American Chemical Society 2002, 124 1736. (88) Varnavski, O.; Samuel, I. D. W.; Pal sson, L. O.; Beavington, R.; Burn, P. L.; Goodson, T. Journal of Chemical Physics 2002, 116 8893. (89) Ranasinghe, M. I.; Murphy, P.; Lu, Z. K.; Huang, S. P. D.; Twieg, R. J.; Goodson, T. Chemical Physics Letters 2004, 383 411. (90) Gaab, K. M.; Thompson, A. L.; Xu, J. J.; Martinez, T. J.; Bardeen, C. J. Journal of the American Chemical Society 2003, 125 9288. (91) Pan, Y. Synthesis and optical propert ies of unsymmetrical conjugated dendrimers, Ph.D. Dissertation, University of Missouri, 2002. (92) Pan, Y.; Lu, M.; Peng, Z.; Melinger, J. S. Journal of Organic Chemistry 2003, 68 6952. (93) Pan, Y.; Peng, Z.; Melinger, J. S. Tetrahedron 2003, 59 5495. (94) Xu, B.; Lu, M.; Peng, Z. Abstracts of Papers, 221st ACS National Meeting, San Diego, CA, United States, April 1-5, 2001 2001, PMSE. (95) Melinger, J. S.; Pan, Y. C.; Kleiman, V. D.; Peng, Z. H.; Davis, B. L.; McMorrow, D.; Lu, M. Journal of the Americ an Chemical Society 2002, 124 12002. (96) Pan, Y. C.; Peng, Z. H.; Melinger, J. S. Tetrahedron 2003, 59 5495. (97) Pan, Y. C.; Lu, M.; Peng, Z. H.; Melinger, J. S. Journal of Organic Chemistry 2003, 68 6952. (98) Turro, N. J. Modern Molecular Photochemistry 1978. (99) Scholes, G. D. Annual Review of Physical Chemistry 2003, 54 57. (100) Forster, T. Ann. Physik 1948, 2 55. (101) Dexter, D. L. Journal of Chemical Physics 1953, 21 836.

PAGE 257

241 (102) Fleming, G. R. International Series of Monogr aphs on Chemistry, Vol. 13: Chemical Applications of Ultrafast Spectroscopy Oxford University: Oxford, 1986. (103) Pan, Y. C.; Lu, M.; Peng, Z. H.; Melinger, J. S. Organic & Biomolecular Chemistry 2003, 1 4465. (104) Peng, Z.; Pan, Y.; Xu, B.; Zhang, J. Journal of the American Chemical Society 2000, 122 6619. (105) Beeby, A.; Findlay, K.; Low, P. J.; Marder, T. B. Journal of the American Chemical Society 2002, 124 8280. (106) Pinto, M. R.; Kristal, B. M.; Schanze, K. S. Langmuir 2003, 19 6523. (107) Tan, C.; Atas, E.; Mueller, J. G.; Pinto, M. R.; Kleiman, V. D.; Schanze, K. S. Journal of the American Chemical Society 2004, 126 13685. (108) Rulliere, C. Femtosecond Laser Pulses: Principles and Experiments, 2nd Edition Springer: Verlag, 2004. (109) Schweitzer, G.; Xu, L.; Craig, B.; DeSchryver, F. C. Optics Communications 1997, 142 283. (110) O'Conner, D. V.; Phillips, D. Time-Correlated Single-Photon Counting 1988. (111) Hirsch, M. D.; Marcus, M. A. ; Lewis, A.; Mahr, H.; Frigo, N. Biophysical journal 1976, 16 1399. (112) Kahlow, M. A.; Jarzeba, W.; DuBruil, T. P.; Barbara, P. F. Review of Scientific Instruments 1988, 59 1098. (113) Barbara, P. F.; Kang, T. J.; Jarzeba, W.; Kahlow, M. Proceedings of SPIE-The International Society for Optical Engineering 1988, 910 124. (114) Kovalenko, S. A.; Dobryakov, A. L.; Ruthmann, J.; Ernsting, N. P. Physical Review A: Atomic, Molecu lar, and Optical Physics 1999, 59 2369. (115) Kovalenko, S. A.; Ernsting, N. P.; Ruthmann, J. Chemical Physics Letters 1996, 258 445. (116) Kovalenko, S. A.; Ruthmann, J.; Ernsting, N. P. Journal of Chemical Physics 1998, 109 1894. (117) Kovalenko, S. A.; Schanz, R.; Farztdinov, V. M.; Hennig, H.; Ernsting, N. P. Chemical Physics Letters 2000, 323 312. (118) Ernsting, N. P.; Kovalenko, S. A. ; Senyushkina, T.; Saam, J.; Farztdinov, V. Journal of Physical Chemistry A 2001, 105 3443.

PAGE 258

242 (119) Huber, R.; Satzger, H. ; Zinth, W.; Wachtveitl, J. Optics Communications 2001, 194 443. (120) Lozhkarev, V. V.; Freidman, G. I.; Ginzburg, V. N.; Khazanov, E. A.; Palashov, O. V.; Sergeev, A. M.; Yakovlev, I. V. Laser Physics 2005, 15 1319. (121) Ghosh, G. Applied Optics 1997, 36 1540. (122) Lorenc, M.; Ziolek, M.; Naskrecki, R.; Karolczak, J.; Kubick i, J.; Maciejewski, A. Applied Physics B: Lasers and Optics 2002, 74 19. (123) Ekvall, K.; van der Meulen, P.; Dhollande, C.; Berg, L. E.; Pommeret, S.; Naskrecki, R.; Mialocq, J. C. Journal of Applied Physics 2000, 87 2340. (124) Adronov, A.; Robello, D. R.; Frechet, J. M. J. Journal of Polymer Science Part aPolymer Chemistry 2001, 39 1366. (125) Tomalia, D. A.; Naylor, A. M.; Goddard, W. A. Angewandte ChemieInternational Edition 1990, 29 138. (126) McElhanon, J. R.; McGrath, D. V. Journal of the American Chemical Society 1998, 120 1647. (127) Tretiak, S.; Chernyak, V.; Mukamel, S. Journal of Physical Chemistry B 1998, 102 3310. (128) Ortiz, W.; Krueger, B. P.; Kleiman, V. D.; Krause, J. L.; Roitberg, A. E. Journal Of Physical Chemistry B, 2006 in press. (129) Ranasinghe, M. I.; Hager, M. W.; Gorman, C. B.; Goodson, T. Journal of Physical Chemistry B 2004, 108 8543. (130) Swallen, S. F.; Zhu, Z. G.; Moore, J. S.; Kopelman, R. Journal of Physical Chemistry B 2000, 104 3988. (131) Peng, Z. H.; Pan, Y. C.; Xu, B. B.; Zhang, J. H. Journal of the American Chemical Society 2000, 122 6619. (132) Atas, E.; Peng, Z.; Kleiman, V. D. Journal of Physical Chemistry B 2005, 109 13553. (133) De Belder, G.; Jordens, S.; Lor, M.; Schweitzer, G.; De, R.; Weil, T.; Herrmann, A.; Wiesler, U. K.; Mullen, K.; De Schryver, F. C. Journal of Photochemistry and Photobiology a-Chemistry 2001, 145 61. (134) De Schryver, F. C.; Hofkens, J. Abstracts of Papers of the American Chemical Society 2001, 221 U288.

PAGE 259

243 (135) Lakowicz, J. R. Principles of Luorescence Spectrosocopy 2nd ed.; Kluwer Academic/Plenum Publishers: New York, 1999. (136) Neuwahl, F. V. R.; Righini, R.; Adronov, A.; Malenfant, P. R. L.; Frechet, J. M. J. Journal of Physical Chemistry B 2001, 105 1307. (137) Schweitzer, G.; Gronheid, R.; Jordens, S.; Lor, M.; De Belder, G.; Weil, T.; Reuther, E.; Muellen, K.; De Schryver, F. C. Journal of Physical Chemistry A 2003, 107 3199. (138) Davis, B. L.; Melinger, J. S.; Mc Morrow, D.; Peng, Z. H.; Pan, C. Y. Journal of Luminescence 2004, 106 301. (139) Swallen, S. F.; Kopelman, R.; Moore, J. S.; Devadoss, C. Journal of Molecular Structure 1999, 486 585. (140) Varnavski, O.; Menkir, G.; Goodson, T.; Burn, P. L. Applied Physics Letters 2000, 77 1120. (141) Melinger, J. S.; Davis, B. L.; McMorrow, D.; Pan, C. Y.; Peng, Z. H. Journal of Fluorescence 2004, 14 105. (142) Pan, Y.; Lu, M.; Peng, Z.; Melinger Joseph, S. The Journal of organic chemistry 2003, 68 6952. (143) Frster, T. Modern Quantum Chemistry ; Academic Press: New York, 1965; Vol. 3. (144) Valeur, B. Molecular Fluorescence: Principles and Applications First ed.; Wiley-VCH: Weinheim, 2002. (145) Hofkens, J.; Cotlet, M.; Vosch, T. ; Tinnefeld, P.; Weston, K. D.; Ego, C.; Grimsdale, A.; Mullen, K.; Beljonne, D.; Bredas, J. L.; Jordens, S.; Schweitzer, G.; Sauer, M.; De Schryver, F. Proceedings of the National Academy of Sciences of the United States of America 2003, 100 13146. (146) Krueger, B. P.; Scholes, G. D.; Jimenez, R.; Fleming, G. R. Journal of Physical Chemistry B 1998, 102 2284. (147) Lii, J. H.; Allinger, N. L. Journal of the American Chemical Society 1989, 111 8566. (148) Lii, J. H.; Allinger, N. L. Journal of the American Chemical Society 1989, 111 8576. (149) Allinger, N. L.; Yuh, Y. H.; Lii, J. H. Journal of the American Chemical Society 1989, 111 8551.

PAGE 260

244 (150) Tortschanoff, A.; Mukamel, S. Journal of Physical Chemistry A 2002, 106 7521. (151) Hirata, Y.; Okada, T.; Mataga, N.; Nomoto, T. Journal of Physical Chemistry 1992, 96 6559. (152) Beeby, A.; Findlay, K. S.; Low, P. J.; Marder, T. B.; Matousek, P.; Parker, A. W.; Rutter, S. R.; Towrie, M. Chemical Communications (Cambridge, United Kingdom) 2003, 2406. (153) Kirkwood, J. C.; Scheurer, C.; Chernyak, V.; Mukamel, S. Journal of Chemical Physics 2001, 114 2419. (154) Yang, M.; Fleming, G. R. Chemical Physics 2002, 282 163. (155) Zhou, Q.; Swager, T. M. Journal of the Americ an Chemical Society 1995, 117 12593. (156) Chen, L. H.; McBranch, D. W.; Wang, H. L.; Helgeson, R.; Wudl, F.; Whitten, D. G. Proceedings of the National Academy of Sc iences of the United States of America 1999, 96 12287. (157) Wang, D. L.; Gong, X.; Heeger, P. S.; Rininsland, F.; Bazan, G. C.; Heeger, A. J. Proceedings of the National Academy of Sc iences of the United States of America 2002, 99 49. (158) Tan, C. Y.; Pinto, M. R.; Schanze, K. S. Chemical Communications 2002, 446. (159) Tan, C. Y.; Atas, E.; Muller, J. G.; Pint o, M. R.; Kleiman, V. D.; Schanze, K. S. Journal of the American Chemical Society 2004, 126 13685. (160) Wang, J.; Wang, D. L.; Miller, E. K.; Moses, D.; Bazan, G. C.; Heeger, A. J. Macromolecules 2000, 33 5153. (161) Nguyen, T. Q.; Junjun, W.; Doan, V.; Schwartz, B. J.; Tolbert, S. H. Science 2000, 288 652. (162) Grage, M. M. L.; Zaushitsyn, Y.; Ya rtsev, A.; Chachisvilis, M.; Sundstrom, V.; Pullerits, T. Physical Review B 2003, 67 (163) Grage, M. M.-L.; Pullerits, T.; Ruseckas, A.; Theander, M.; Inganas, O.; Sundstrom, V. Chemical Physics Letters 2001, 339 96. (164) VandenBout, D. A.; Yip, W. T.; Hu, D. H.; Fu, D. K.; Swager, T. M.; Barbara, P. F. Science 1997, 277 1074. (165) Huser, T.; Yan, M.; Rothberg, L. J. Proceedings of the National Academy of Sciences of the United States of America 2000, 97 11187.

PAGE 261

245 (166) Muller, J. G.; Lemmer, U.; Raschke, G.; Anni, M.; Scherf, U.; Lupton, J. M.; Feldmann, J. Physical Review Letters 2003, 91 (167) Levitsky, I. A.; Kim, J. S.; Swager, T. M. Journal of the American Chemical Society 1999, 121 1466. (168) Herz, L. M.; Silva, C.; Friend, R. H.; Phillips, R. T.; Setayesh, S.; Becker, S.; Marsitsky, D.; Mullen, K. Physical Review B 2001, 64 (169) Tan, C. Y.; Alas, E.; Muller, J. G.; Pinto, M. R.; Kleiman, V. D.; Schanze, K. S. Journal of the American Chemical Society 2004, 126 13685. (170) Sluch, M. I.; Godt, A.; Bunz, U. H. F.; Berg, M. A. Journal of the American Chemical Society 2001, 123 6447. (171) Gaab, K. M.; Bardeen, C. J. Journal of Physical Chemistry B 2004, 108 4619. (172) Scheidler, M.; Cleve, B.; Bassler, H.; Thomas, P. Chemical Physics Letters 1994, 225 431. (173) Grage, M. M. L.; Wood, P. W.; Ruseck as, A.; Pullerits, T.; Mitchell, W.; Burn, P. L.; Samuel, I. D. W.; Sundstrom, V. Journal of Chemical Physics 2003, 118 7644. (174) Kamalov, V. F.; Struganova, I. A.; Yoshihara, K. Journal of Physical Chemistry 1996, 100 8640. (175) Rosenstock, H. B. Journal of the Society for Indu strial and Applied Mathematics 1961, 9 169. (176) Montroll, E. W. Journal of Mathematical Physics 1969, 10 753. (177) Fitzgibbon, P. D.; Frank, C. W. Macromolecules 1982, 15 733. (178) Klafter, J.; Blumen, A.; Zumofen, G. Journal of Statistical Physics 1984, 36 561. (179) Palszegi, T.; Mollay, B.; Kauffmann, H. F. Journal of Chemical Physics 1998, 108 7023. (180) Scheidler, M.; Lemmer, U.; Kersting, R. ; Karg, S.; Riess, W.; Cleve, B.; Mahrt, R. F.; Kurz, H.; Bassler, H.; Gobel, E. O.; Thomas, P. Physical Review B 1996, 54 5536. (181) Kauffmann, H. F.; Molla y, B.; Weixelbaumer, W. D.; Burbaumer, J.; Riegler, M.; Meisterhofer, E.; Aussenegg, F. R. Journal of Chemical Physics 1986, 85 3566. (182) Beljonne, D.; Pourtois, G.; Silva, C.; Hennebicq, E.; Herz, L. M.; Friend, R. H.; Scholes, G. D.; Setayesh, S.; Mullen, K.; Bredas, J. L. Proceedings of the National Academy of Sciences of th e United States of America 2002, 99 10982. (183) Kukula, H.; Veit, S.; Godt, A. European Journal of Organic Chemistry 1999, 277.

PAGE 262

246 (184) Nomula, S.; Cooper, S. L. Macromolecules 2000, 33 6402.

PAGE 263

247 BIOGRAPHICAL SKETCH Evrim Atas was born on October 2, 1978 in Igdir, Turkey, where she spent her childhood until she started elementary school. After attending Haydarpasa Anatolian High School for 4 years, she began her unde rgraduate studies in the Fall of 1996 at Bilkent University, Department of Chemis try in Ankara, Turkey. With an intense physical chemistry education and a special in terest in spectroscopy, she came to the University of Florida, Department of Chem istry in the Fall of 2000 to begin doctoral studies under the supervision of Professor Va leria Kleiman in the ar ea of ultrafast laser spectroscopy of dendrimers and conjugated po lymers. Her professional career as a Ph.D. will begin as a post-doctoral fellow with Professor Amit Meller at the Rowland Institute, Harvard University. She will then move to th e Department of Biomedical Engineering, Boston University.


Permanent Link: http://ufdc.ufl.edu/UFE0013101/00001

Material Information

Title: Ultrafast Time Resolved Excitation Dynamics in Conjugated Dendrimers
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013101:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013101/00001

Material Information

Title: Ultrafast Time Resolved Excitation Dynamics in Conjugated Dendrimers
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013101:00001


This item has the following downloads:


Full Text












ULTRAFAST TIME RESOLVED EXCITATION DYNAMICS IN
CONJUGATED DENDRIMERS
















By

EVRIM ATAS


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2006

































Copyright 2006

by

Evrim Atas

































To Selim, Avni, and my parents















ACKNOWLEDGMENTS

As I look back upon the years that have led to this dissertation, I have been

fortunate to have been surrounded by the help and influence of numerous exceptional

people. I would like to extend my warmest thanks to my research advisor Professor

Valeria D. Kleiman, for her inspiring, patient guidance and continuous motivation

throughout this enjoyable, yet challenging journey. She has served as a great mentor and

friend from whom I have learned plenty. It was her genuine understanding, unique

enthusiasm and well-placed trust that has led this project to completion.

I wish to thank my supervisory committee members, Professors Jeff Krause and

Kirk S. Schanze for their guidance, fruitful collaborations and valuable discussions

throughout my graduate studies, and Professors Russ Bowers and David Reitze for

accompanying me in the final stage of my graduate career. I extend my thanks to Prof

Adrian Roitberg for his contribution in the theoretical work and Prof. Nic6 Omenetto for

teaching me the basics of lasers. I also thank Prof. Zhonghua Peng from University of

Missouri-Kansas for providing unsymmetrical dendrimers and Dr. Joseph S. Melinger for

his contributions to the initial stage of this project.

Many friends as well as coworkers have had an important role during my graduate

studies in Gainesville with their discussions and companionships. The members of

Kleiman Group provide a fun work environment. Thanks go to Dr. Jurgen Muller for his

ongoing friendship and working closely with me on the polymer project presented in

Chapter 6 of this dissertation. I thank Daniel Kuroda for his contagious energy, personal









support and having an answer for my every question. I would like to thank Lindsay

Hardison for being such a cool girl, for sharing both science and personal matters, and

for her suffering to correct my English in this dissertation. I thank Chad Mair for his late

night support and chats in the lab, and teaching me Matlab for data analysis. The newest

member of our group, Aysun Altan, thanks for your hard work and contribution of low

temperature data in Chapter 4. I would like to thank Dr. Chunyan Tan, my collaborator

from Dr. Schanze's Group, for enjoyable discussions. I also thank Dr. Wilfredo Ortiz and

Julio Palma for their contributions with the theoretical aspect of dendrimers. I especially

thank Bob Letiecq from Spectra Physics who provided day and night technical support.

My time here would not have been the same without the social diversions provided

by all my friends in Gainesville. I am particularly thankful to Meziyet-Enes family, Enes

Calik, Ece Unur, Dilber and Yavuz for their continuous friendship. I would like to thank

Ozlem Demir for being a wonderful friend and great support since the day we met in

Ankara ten years ago.

I thank my parents Reize and Akil for allowing me to make my own decisions since

I was a little child. They always believed in me and supported me whatever I do. I thank

my sisters Zehra, Eda, and Seda for sharing a very happy childhood together. I thank my

Aunt Nazmiye for her directions and support whenever I needed.

Finally, I give my special thanks to my husband, Avni, for his true love and

understanding me without words. You are my soul mate and you bring colors to my life.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ................................................................................................. iv

L IST O F T A B L E S ......... ............................... ......... ... ....... ....... ix

LIST OF FIGURES ............................... .... ...... ... ................. .x

A B S T R A C T .......................................... ..................................................x v

CHAPTER

1 IN TR OD U CTION ............................................... .. ......................... ..

H history of D endrim ers .................. ............................ ...... ..... .............. ..
Structural Properties .................................. .. .. ... ....... ......... .. .. .4
Light-H arvesting D endrim ers ............................................. ..... ....................... 8
Sym m etrical PE D endrim ers .................................... .......................... .. ........ 16
Unsymmetrical PE Dendrimers................................................. ............... 23
E xcitation E energy T ransfer.............................................................. .....................28
R adiative Energy Transfer........................................................ ............... 29
N on-radiative Energy Transfer....................................... ......................... 30
Outline of the D issertation......................................................... ............... 42

2 EXPERIM ENTAL M ETHODS ........................................ ........................... 45

C hem icals and M aterials................................................. ................................ 45
Steady State M easurem ents ............................................... ............................. 47
Why Time-Resolved Spectroscopy?................................................................ 47
The L aser System ......................................... ........ ............ ............ 48
Ultrafast Time-Resolved Emission Spectroscopy ............................................... 51
Time-Correlated Single Photon Counting ......................................................... 51
Fluorescence U conversion Technique ................................ .... ....................53
Homemade Upconversion Apparatus......... ....................................... 55
Ultrafast Transient Absorption Spectroscopy........................................ ..................58
Probe Characteristics and White Light Continuum Generation........................62
E xperim ental Setup ........................... ........ ............................ ...... ............63
Continuum generation .................................. .. ....................................64
Time resolution of the experiment ............. ............ .. ..... ....... ........69
Concentration and Pump Pulse Energy Dependence......................................71









3 ENERGY TRANSFER IN GENERATION 1 UNSYMMETRICAL
PHENYLENE ETHYNYLENE DENDRIMERS ............... ................... ........... 73

M materials an d M eth od s .................................................................... .....................77
Steady State Spectroscopy .............................................. .............................. 80
Time-Resolved Emission Experiments............................................... ...............83
Time-Resolved Broadband Transient Absorption Measurements ..............................88
K inetic M odel ..........................................................................93
Energy Transfer via W eak Coupling ........................................ ...................... 98
C o n clu sio n s.................................................... ................ 10 0

4 ENERGY TRANSFER IN GENERATION 2 UNSYMMETRICAL
PHENYLENE ETHYNYLENE DENDRIMERS ............................... ..................101

M materials and M methods ...................... ............................ ............................ 102
Steady State Spectroscopy of Phenylene Ethynylene Dendrimers........................... 104
Tim e-Resolved Em mission Experim ents ................... ...... ................................. 106
Time-Resolved Broadband Transient Absorption Measurements..........................112
K inetic M odel for Energy Transfer .................................. ..................................... 116
Energy Transfer in the W eak-Coupling Lim it........................................................ 121
Vectorial Energy Transfer in Unsymmetrical PE Dendrimers .......................... 124
C o n clu sio n s.................................................... ................ 12 8

5 ENERGY TRANSFER IN SYMMETRICAL PE DENDRIMER: NANOSTAR... 130

M materials an d M eth od s .................................................................. ..................... 133
Transient A bsorption ............................. ............................................ 134
T im e-resolved E m mission ........................................................... ....... ............ 135
Steady State Spectroscopy ............................ ................... ............................. 136
Transient Absorption Spectroscopy..... .................... ...............137
M odel Compound DPA .............................................. ............... 139
Model Compound Phenylethynylene Perylene ...........................................141
N an o star ...................................... .............................. ................ 14 3
372 nm excitation .............................. .......... .................... ..............143
352 nm excitation .............................. .......... .................... ..............145
3 10 nm ex citation .............................. .......... .................... .............. 14 8
K inetic M odel for N anostar ......................................................... .............. 151
M odel for 372 nm Excitation ........................................ ........................ 152
M odel for 352 nm Excitation ........................................ ........................ 154
M odel for 310 nm Excitation ..................................... ........................ ......... 156
Time-Resolved Emission Experiments ............... ............................................. 158
Energy Transfer ........................................................... ................. 163
C on clu sion s .............................................. ......... ............... 16 9









6 THE ROLE OF EXCITON HOPPING AND DIRECT ENERGY TRANSFER IN
THE CONJUGATED POLYELECTROLYTES ............... .................... ..........171

Steady State E xperim ents ..................... ......... ..................................................... 174
Time Resolved Fluorescence Spectroscopy .................................... ............... 176
Comparison of Transient Fluorescence and Absorption Data ..............................181
A nisotropy and Ionic Com plexes .................. .. .... ................................... .... 185
Loss of Anisotropy and the Absence of Quenching by Free Dye Molecules..........188
Comparison of TA- and PL-upconversion Dynamics ................... ............... 189
Quenching Dynamics and the State Contribution .......................................... 190
M modeling .............................................................. .... ..... ......... 192
C o n clu sio n s.................................................... ................ 2 0 1

7 SUMMARY AND PERSPECTIVE ................................................. ..............204

APPENDIX

A THE FLUORESCENCE UP-CONVERSION TECHNIQUE...............................209

The Excitation and Collection of Fluorescence ...................................................... 209
Alignment of the Gate Beam and Upconversion Crystal Phase Matching...............214
D election of Upconverted Fluorescence................................................................ 216
U V -Light C om pression ................................................. ............................... 217

B D A T A A N A L Y SIS ......................................................................... ...................2 19

Analysis of Transient Absorption Changes ................................... .................219
2G -m -O H ............................................................................................... ....... 223
2G 2-m -O H .................... ................................................ ..................................... 227
N anostar Excited at 310 nm ...................................................... ....................231

L IST O F R EFER EN CE S ........................................................................... .............235

B IO G R A PH IC A L SK E TCH ........................................ ............................................247
















LIST OF TABLES


Table page

3-1. Fits for tim e-resolved fluorescence data......................................... ............... 96

4-1. Lifetime measurements from TCSPC.................................................................... 107

4-2. Fits for time-resolved fluorescence data .... ........... ...................................... 119

5-1. Fits for transient absorption data ..................................................... ............ 158

6-1. Parameters recovered from kinetic modeling of PPESO3 fuorescence decays with
H M ID C in M eO H ................................................................... .. .... .. 176

6-2. Parameters and variables used in the numerical simulations and fitting of the
time-resolved PL-upconversion and the transient absorption data ......................194

6-3. Average distance between two acceptor molecules in complex with the PPESO3..195
















LIST OF FIGURES


Figure p

1-1 Representation of dendrimer growth by the divergent and convergent methods.......3

1-2 Schematic representation of bacterial light-harvesting complexes........................9

1-3 T he dendritic triad.. ...................... .................... ..................... .. ..... 14

1-4 Poly(aryl ether) dendrimer functionalized with dyes....................... ...............15

1-5 Nanostar dendrimer (a) Chemical structure (b) Energy level diagram ..................20

1-6 Chemical structures of unsymmetrical PE dendrimers ......................................24

1-7 Absorption and emission spectra of GnOH monodendrons. ...................................25

1-8 Steady state spectra of GnPer monodendrons: absorption (a) and emission (b).......26

1-9 Chemical structure of PE didendrons............................................ ...............28

1-10 Model picture for energy transfer showing resonant transitions of donor and
acceptor, and spectral overlap of donor emission and acceptor absorption ............31

1-11 Schematic representation of energy transfer mechanism............... ...................32

1-12 Definition of the angles used to calculate the orientation factor between the
d ip o le s. ........................................................... ................ 3 5

1-13 Differences between strong, weak, and very weak coupling ................................37

2-1 The laser system for the production of tunable femtosecond laser pulses with
high energy per pulse. .......................... ...................... ............... .... ...... ...... 49

2-2 Fluorescence up-conversion technique (a) Illustration of the upconversion
principle (b) Up-converted fluorescence signal generated in a nonlinear crystal
only while the delayed gate pulse is present .................................... ............... 55

2-3 Fluorescence upconversion experimental setup ........................................... 57

2-4 Basic principle of transient absorption experiment...................................59









2-5 The theoretical scheme of certain signals observed as transient absorption
sig n a ls ..................................... .................................................. .. 6 1

2-6 Experimental setup for transient absorption experiment probing with white light
continuum. UV pump pulses are obtained from the OPA ...................................... 63

2-7 Spectrum of the white light continuum generated by CaF2. The probe (blue) and
reference (red) beams used for transient absorption experiment. ..........................65

2-8 The chirp of the white light continuum determined from the delay between the
sig n a ls ........................................................................ 6 8

2-9 Coherent artifact of hexane solvent excited at 310 nm, probed at 330 nm and
3 8 0 n m ...................................... .....................................................7 0

3-1 Chemical structures of generation 1 phenylene ethynylene dendrimers: (a) 2G1-
m -OH (b) 2G1-m -Per .................. ............. ............. .......... ............. 77

3-2 Normalized absorption spectra of- 2G1-m-OH, **- G1-OH, and G2-OH in
dichloromethane. Normalized emission spectrum of- 2G1-m-OH. ....................81

3-3 Normalized absorption spectra of-----2GI-m-OH, ...EPer, and -2Gl-m-Per
and fluorescence spectrum of -2Gl-m-Per, excited at 315 nm. ............................82

3-4 2Gi-m-OH in dichloromethane excited at a) 370 nm b) 315nm............................85

3-5 Upconversion signal of 2G1-m-Per detected at 485 nm, excited at (a) 370 nm,
(b ) 3 15n m .. ...............................................................................8 7

3-6 2G1-m-Per Upconversion Signal. Xexcitation=315 nm, emission = 400 nm...................88

3-7 Transient absorption spectra of 2G1-m-OH molecule at different time delays,
excited at 315 nm .. ....................................................... ................ 90

3-8 Transient absorption spectra of the 2G1-m-OH molecule at different delay times.
Detailed display of the 350 nm-450 nm region...... .... ....................................... 91

3-9 Transient absorption spectra of 2G1-m-Per at different time delays........................92

4-1 Chemical structures of phenylene ethynylene dendrimers: (a) 2G2-m-OH (b)
2G2-m-Per and (c) 3D model of the 2G2-m-Per from a MD simulation................103

4-2 Normalized absorption spectra of -----2G2-m-OH, EPer and -2G2-m-Per and
fluorescence spectrum of-2G2-m-Per, excited at 320 nm ............ ................106

4-3 2G2-m-OH in dichloromethane excited at a) 415 nm b) 372nm and c) 330 nm....108









4-4 Upconversion signal of 2G2-m-Per detected at 485 nm, excited at a) 465 nm, b)
420 nm c)380 nm and d) 340 nm ......................................................... ........... 110

4-5 2G2-m-Per upconversion signal. Xexcitation=320 nm emission = 435 nm. ..................112

4-6 Transient absorption spectra of 2G2-m-OH at different delay times................. 114

4-7 The excitation spectrum of 2G2-m-OH at 298 K (red) and 77 K (blue), emission
detected at 450 nm (top). The excitation anisotropy of 2G2-m-OH at 77 K
(b bottom ).............................................................................................. 116

4-8 Model describing the energy ladder. The intermediate state (I) is detected at 400
nm or 435 nm. Emission from trap is at 485 nm............................125

4-9 Temporal evolution of the intermediate state population followed by
fluorescence up-conversion ......................................................... ............. ..127

5-1 Chemical structure of nanostar dendrimer (2 dimensional sketch .........................134

5-2 Absorption and emission spectrum of nanostar in DCM at room temperature...... 136

5-3 Normalized absorption spectrum of (a) 2-ring (black), 3-ring (green), and 4-ring
(red) PE units (b) Nanostar absorption (red) and sum of rings' absorption at 298
K (c) nanostar absorption at 298 (blue) and 10 K (red). ......................................138

5-4 Transient absorption spectrum of model compound DPA.............. ............. 140

5-5 Transient absorption spectrum of the model compound phenylethynylene
perylene. ................ ... ................................. 142

5-6 Transient absorption spectrum of nanostar after excitation at 372 nm ................144

5-7 Transient absorption signal as a function of time recorded at 340 nm (black), and
515 nm (blue) following excitation at 372 nm ................... .............................. 145

5-8 Transient absorption spectrum of nanostar after excitation at 352 nm. (a) Short
time window (At< 450fs). (b) Long time window (0.550-50 ps)...........................147

5-9 Transient absorption signal as a function of time recorded at 360 nm (black), and
520 nm (red) following excitation at 352 nm ........................................................ 148

5-10 Transient absorption spectrum of nanostar after excitation at 310 nm ................149

5-11 Transient absorption signal as a function of time for three different excitation
w avelengths.............................................................................................151

5-12 Kinetic model proposed for the dynamics of nanostar................. .....................153









5-13 Fluorescence upconversion signal detected at 485 nm (blue) and at 380 nm (red)
follow ing the 310 nm excitation................................ ........................ ......... 160

6-1 Conjugated polyelectrolytes PPESO3 (left) and cyanine dye HMIDC (right). .....174

6-2 Absorption and emission spectra of PPESO3 and HMIDC in methanol .............175

6-3 Time-resolved fluorescence of PPESO3 (34 [tM) in MeOH for different HMIDC
concentrations.................................................................... ......... 178

6-4 Normalized time-resolved fluorescence of HMIDC in MeOH..............................181

6-5 Transient absorption and up-conversion signal of PPESO3 in MeOH for
different concentrations of added HMIDC................................ ............... 183

6-6 PL-upconversion and AA of a pure PPESO3 solution .....................................1.....184

6-7 Time dependent loss of anisotropy for pure PPES03 and PPES03 with HMIDC
qu en ch er. ........................................................................ 186

6-8 The photoluminescence yield of samples containing HMIDC, divided by the
photoluminescence of the pure PPESO3 solution................................................. 190

6-9 Stern-Volmer plot of the time-integrated photoluminescence and the
instantaneous photoluminescence from the upconversion experiment................92

6-10 Average distance between quencher molecules completed with the PPESO3 on
a reciprocal scale as a function of the quencher concentration ............................195

6-11 Individual contributions to the integrated energy transfer, random walk mediated
(a) direct Forster-transfer (b) versus time after excitation. ...................................199

A-i Diagram of off-axis aluminum parabolic mirrors used to collect and image the
fluorescence of the sam ple. ............................................. ........................... 210

A-2 Layout of the alignment beams and the collection by parabolic mirrors...............211

B -l Singular values ................... .... .......................... .. .. ...... .. ............. 223

B-2 Transient spectra of SVD output ........................................ ....................... 223

B -3 D ynam ics of SV D output. ........................................................... .....................224

B-4 Model fits for the relevant components with large singular values. ....................224

B-5 Reconstructed versus real AA as a function of wavelength. ............................225

B-6 Reconstructed versus real AA as a function of time. ................ .....................226









B -7 Singular values ................... .... ............................ .. ...... .. ............. 227

B-8 Transient spectra of SVD output ........................................ ....................... 227

B-9 Dynam ics of SVD output. .............................................. ............................ 228

B-10 Model fits for the relevant components with large singular values. ....................228

B-11 Reconstructed versus real AA as a function of wavelength ................................229

B-12 Reconstructed versus real AA as a function of time. .............................................230

B -13 Singular values ............ .... .......................... ........... ...... ..... .. 231

B-14 Transient spectra of SVD output ................................ ....................231

B -15 D ynam ics of SV D output. ........................................ .......................................232

B-16 Model fits for the relevant components with large singular values. ....................232

B-17 Reconstructed versus real AA as a function of wavelength ................................233

B-18 Reconstructed versus real AA as a function of time. .............................................234















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

ULTRAFAST TIME RESOLVED EXCITATION DYNAMICS IN
CONJUGATED DENDRIMERS.


By

Evrim Atas

May 2006

Chair: Valeria D. Kleiman
Major Department: Chemistry

Light-matter interactions play an important role in light-harvesting processes such

as photosynthesis, which has attracted much attention due to its major impact on the

cycle of life. Understanding the fundamental principles of this energy transfer process is

possible through the study of artificial light harvesting systems. Dendrimers are perfectly

branched synthetic macromolecules having numerous peripheral chain-ends surrounding

a single core. Incorporating suitable chromophore groups into their structure can create

very efficient antenna systems. This PhD thesis details the dynamics of intramolecular

energy transfer in conjugated phenylene ethynylene dendrimers. Built-in energy-

gradients in the dendrimer structure enable a unidirectional energy transfer from the

periphery to the core. Depending on the substitution pattern on the phenyl ring,

symmetrical and unsymmetrical architectures are formed that yield different

photophysical properties.









Ultrafast time-resolved fluorescence and absorbance techniques are utilized to

study the fast dynamics of energy transfer. Our approach is based on a comparative study

of symmetrical and unsymmetrical dendrimers with various nt-conjugation and sizes.

Dendrimer backbones are selectively excited at specific absorption wavelengths and the

energy migration toward the acceptor is monitored. Time-resolved fluorescence

measurements explore the population of intermediate states and the final energy acceptor,

while broadband transient absorption (300 nm to 600 nm) probes the dynamics from the

initially excited state to the final trap.

To understand the dynamics and mechanisms of energy transfer we propose kinetic

models describing the time-resolved data as a function of dendrimer size, presence or

absence of a trap and excitation wavelength. For unsymmetrical didendrons, typical

energy transfer times are in the range of 200-750 fs. While absorption is into delocalized

exciton states, emission occurs from localized states. In the presence of attached perylene

trap, excitation energy migrates through multiple channels. The calculated interaction

energies (75-100 cm-) indicate that dendrons and perylene are weakly coupled. The

symmetrical phenyl ethynylene dendrimer, however, shows energy transfer times from

200 fs to 20 ps, much slower than the unsymmetrical molecule. Considering the broken

n-conjugation due to the meta substitution, the subunits of the nanostar are investigated

independently via transient absorption. The kinetic model analysis shows that there are

both direct and indirect transfer (through the cascade) pathways. The experimental energy

transfer rates are discussed within the Forster theory to understand the extent of the

electronic coupling. In addition, an ultrafast study of exciton transport in a phenyl

ethynylene polyelectrolyte is performed through quenching experiments.














CHAPTER 1
INTRODUCTION

History of Dendrimers

The synthesis of dendrimers is an important stage in the evolution of

macromolecular chemistry. Dendrimers are hyperbranched, well-defined, three-

dimensional, and perfectly monodisperse macromolecules.1-4 Although Flory

theoretically investigated the role of branched units in macromolecular architectures half

a century ago,56 the first successful synthesis of a dendritic structure did not occur until

the late 1970s. The first example of an iterative cascade procedure toward well-defined

branched structures, such as low molecular weight branched amines, has been reported

by the Vogtle group.7 However, not all regularly branched molecules are dendrimers.

Important characteristics, which will be explained in detail later in this chapter, are

reached when globularity is achieved at a certain generation and size threshold. The

Vogtle group's cascade molecules are too small to exhibit the properties of dendrimers

and are used as branched oligomeric building blocks in dendrimer construction.7

Optimization of the iterative method with Michael addition enabled the synthesis of

the first globular dendrimers called PAMAM (polyamidoamine) by Tomalia et al. at Dow

Chemical Research Laboratories.8-10 PAMAM dendrimers are the first dendrimer family

to be commercialized and they have been thoroughly investigated to date. Shortly after,

Newkome et al."1 reported the synthesis of arborols, another family of trisbranched

polyamide dendrimers, and two research groups, Milhaupt and Meijer, were able to

improve the Vogtle's synthesis approach to enable the production of poly(propylene









imine) dendrimers.12,13 These dendrimers were constructed divergently, implying that the

synthesis starts with a functional core molecule and is expanded to the periphery. In

1990, Hawker and Frechet introduced the convergent approach to produce aromatic

polyether dendrimers.14,15

In contrast to polymers, dendrimers are core-shell structures possessing three basic

architectural components: 1) a core, 2) repeating units in the interior of shells consisting

of branching points (generations), 3) terminal functional groups (periphery).

Two complementary methods, the divergent and convergent synthesis, have been

used to construct high-generation dendrimers.3'4'15'16 Both methods consist of a repetition

of reaction steps, accounting for the creation of an additional generation. Within each of

these major approaches there may be variations in methodology. The features desired for

the target molecule and specific building blocks justify the choice of the synthetic

approach.

Divergent approach. Based on the work of Tomalia and Newkome, the growth

starts at the core and proceeds radially outward toward the periphery.10'11 The number of

reactions that must be completed at each step of growth increases exponentially.

Therefore, a large excess of reagents is required making it harder to maintain the purity

and structural uniformity. However, this method is used widely for the preparation of

high generation dendrimers and for the synthesis on large scale. The major drawback is

the poor yield of defect-free dendrimers.

Convergent approach. This method, first reported by the Frechet group, initiates

growth at what will become the periphery of the molecule and proceeds inward towards

the focal point.14 This approach is best described as an "organic chemist" approach to










globular macromolecules, since it provides outstanding control over growth, structure,

and functionality.17 The inward growth allows for the reduction in the amount of

synthetic steps and intermediate purification at each step of growth. The yield of defect-

free dendrimers is about 80%. Figure 1-1 illustrates the dendrimer growth by both the

divergent and the convergent methods.

Since the discovery of dendrimers, one of the controversial issues has been the

purity of these structures. The quality of the final dendritic product is directly related to

the chosen synthetic method. A variety of convergently synthesized dendrimers have

been reported in the last decade, and these dendrimers have shown that the convergent

approach provides greater structural control than the divergent approach; allowing purity,

structural uniformity, and functional versatility. Another attractive feature of the

convergent approach is its ability to selectively modify both the focal point and the chain

ends. In addition, functional groups can be precisely placed throughout the structure.

Overall, the organic nature of the convergent method results in defect-free dendrimers

with appropriate purification.

core chain end

A 9






DIVERGENT: CONVERGENT:
O # of reactions per growth # of reactions per growth
step increases step remains constant
slight structural defects f* defect structures more
occur in large dendrimers readily separated
P net mass increase net mass decrease



Figure 1-1. Representation of dendrimer growth by the divergent and convergent
methods. Figure is adapted from Tomalia et al.18









Structural Properties

Structural and conformational behavior of dendrimers is discussed in many books

and publications. Several intriguing questions arise: Are dendrimers always globular or

can their shape be highly distorted? How rigid are they? Can the end groups back-fold?

Are there cavities present within dendrimers? How do the physical properties change

with generation? What are the similarities with linear analogues?

The dendrimer structure can be divided into three distinct architectural regions:

core or focal moiety, branched repeat units, and end groups on the outer layer. Their

structural precision leads to an exact number of branching points or generations, which

differentiates dendrimers from hyperbranched polymers. In contrast to linear polymer

analogues, dendrimers have several sharp characteristic features.

(i) A dendrimer will have size monodispersity due to its well-defined iterative
synthesis, whereas most linear polymers are synthesized composing a
range of molecular species differing in size and molecular weight.

(ii) While the linear polymers contain only two end groups, the number of
dendrimer end groups increases exponentially with generation. As the size
of the dendrimer increases, the nature of end groups will determine
important properties such as solubility, chemical reactivity, and glass
transition temperature.

(iii) In theory, polymers can grow as much as their solubility allows them,
whereas dendritic growth is mathematically limited. The number of
monomer units increases exponentially, but the volume available to the
dendrimer grows proportionally to the cube of its radius. As a result of this
physical limitation, dendrimers develop more globular conformation as the
generation increases. In contrast to polymers, the intrinsic viscosity of
dendrimers does not increase with molecular weight.

More extended arrangements for lower generation dendrimers will gradually

transform into compact and globular shapes for higher generation dendrimers. In general,

this gradual transition in overall shape results in the deviation in physical behavior of

dendrimers from those of linear macromolecules.17









Dendrimers might be flexible or fairly rigid depending on the actual dendritic

structure. Recent calculations and measurements have suggested backfolding of the chain

ends. For example, the polyether dendrimers synthesized by the Frechet group have been

investigated in detail to verify the possibilities for backfolding.19 The flexible nature of

these dendrimers implies that the end groups are found throughout the dendrimer volume.

However, when the end groups can communicate with each other with attractive

secondary interactions such as 7t- 7t interactions, electrostatic repulsions, and hydrogen

bonding interactions, the terminal units will assemble at the periphery precluding back

folding.20

One of the most studied rigid dendrimer family is phenylethynylene dendrimers,

first synthesized by Moore et al.21 These dendrimers are distinguished from other

dendrimers by their rigidity and shape persistence as confirmed by various experimental

measurements.22 Another type of shape persistent dendrimers is based on polyphenylene

units. Miullen and coworkers investigated these molecules and found that their rigidity

originates from the very dense packing of benzene rings.23-25

Do cavities exist within the dendrimer? Indeed, unlike linear polymers, properly

designed high generation dendrimers exhibit a distinct interior where molecules have

been encapsulated in a noncovalent manner.26-28 The encapsulation does not necessarily

indicate the presence of a permanent and rigid cavity within the dendrimer. Especially,

flexible dendrimers can accommodate guest molecules. When solvent molecules that

freely penetrate dendrimers are removed, the volume collapses leaving the guest

molecules trapped inside the dendrimer. For example, the well-designed and rigidified

dendrimer structure called dendriticc box" can encapsulate various small organic









molecules and control their release by modifying the steric crowding of the dendritic

periphery.2931

The encapsulation of a functional core moiety creating specific site-isolated

nanoenvironments leads to a variety of bio- and nanotechnology applications including

light-harvesting, amplification, and drug delivery.32,33 Having full control of the structure

and architecture, researchers are able to place active sites that have photophysical,

photochemical, electrochemical, or catalytic functional groups at the core of the

dendrimers. One of the more elegant works on efficient, unidirectional energy transfer

from a dendritic framework to a single chromophore was reported by Xu and Moore.34'35

A gradient effect was created using a poly(phenylethynylene) dendrimer. The

conjugation length of the repeat units of this dendrimer increases with generation from

the periphery to the core. This is the so called "nanostar" molecule later discussed in

Chapter 5 of this dissertation. The phenylethynylene units of Moore dendrimers can be

used to create unsymmetrical dendrimer architectures, which are also investigated in this

dissertation (Chapters 3 and 4).

Frechet group's poly(benzylether) dendrimers, functionalized with different dye

chromophores at the periphery and core, are able to harvest light and transfer the energy

efficiently to a chromophore located at the center of the dendrimer structure. It was

shown that the core chromophore emission is significantly amplified compared to the

same chromophore without the dendritic framework. As the size of the dendrimer

increases, so does the number of peripheral units, therefore the energy transferred to the

core increases due to a larger absorption cross-section.3638Another application of the

same dendrimer structure is in optical signal amplification. Luminescent lanthanide ions









are used as signal amplifiers for optical fiber communications. However, their self-

quenching in the solid state greatly limits the effectiveness. Kawa et al. encapsulated

individual lanthanide ions within poly(benzylether) dendrons leading to site-isolation,

thereby decreasing the self-quenching effect. This antenna light-harvesting effect results

in emission signal amplification.33

In another report, Jiang and Aida used azobenzene containing aryl ether dendrimers

to study energy transfer.39 They demonstrated that cis-trans isomerization of azobenzene

moiety at the core accelerates for larger generation (e.g. G-4, G-5) dendrimers. The

acceleration was observed via exciting a stretching mode of the aromatic rings with IR

irradiation. However, UV excitation of the dendrons did also result in accelerated

isomerization. Thus, it was proposed that the dendritic shell not only insulates the

azobenzene core from collisional energy dissipation, but it acts as a photon harvesting

antenna.

Miullen and coworkers reported polyphenylene dendrimers functionalized with

different chromophores at the periphery and core, while the De Schryver group

investigated this family of dendrimers and explored the energy transfer dynamics via

time-resolved spectroscopy experiments.40'41 The dendrimer rigid structure decorated

with a unique selection and positioning of the chromophores allows a systematic study of

possible energy transfer mechanisms.

A different type of encapsulation involves the formation of metal nanoparticles

within dendrimers and it has been widely used to prepare organic-inorganic composite

structures useful in catalytic applications.42 Since dendrimers have nanoscopic

dimensions and can be dissolved molecularly, catalytic active site can be placed at a









particular, isolated position resulting in beneficial interactions with the substrate.17

Brunner et al.43 introduced the first branched molecules containing internal core catalytic

sites, later called "dendrzymes." In parallel to this work, the first dendritic catalyst with

multiple catalytic sites at the periphery has been reported by Ford et al.44 Both studies

concluded that low-generation dendrimers are better catalytic supports than higher

generation dendrimers. More recently, Crooks et al. have shown that substrates can

penetrate the dendrimers to access the catalytic sites and undergo simple reactions such

as hydrogenations.42 The Crooks group also developed a system composed of PAMAM

molecules covalently attached to a metallic surface and tested its function as a chemical

45
sensor.

Recently, a sequence of dendrimers containing Zn-porphyrin monomers situated on

the surface were investigated by Sundstr6m and coworkers.46'47 The goal of this project

was to study energy transfer between the individual Zn-porphyrins within a dendrimer,

and to measure whether this process becomes more efficient with increasing dendrimer

size. The molecules based on porphyrin chromophores are an example of compounds that

can be used as a model for the bacteriochlorophyls (BChls) in the natural light-harvesting

(LH2) complex.

Before going through the details of these specific structures, a brief summary of

light-harvesting dendrimers and the role of an energy gradient within these dendrimers

will be presented.

Light-Harvesting Dendrimers

Photosynthesis is an extremely effective natural process for harvesting sunlight and

converting sunlight into useful chemical energy stored in the form of ATP. Thus, it has

been of vital importance to the evolution of life and it is essential for almost all life-









forms. Therefore, it is inevitable that the photochemistry of photosynthesis be the focus

of considerable scientific research.

There are two key processes in photosynthesis. First is the absorption of photons by

an antenna system, followed by a rapid and efficient transfer of excitation energy to the

reaction center. Then, a sequence of charge transfer events from the excited state of the

reaction center results in storage of chemical energy. To date, the most studied

photosynthetic system is probably the purple bacteria.48 The high resolution X-ray crystal

structure of this bacteria? reveals a central reaction center surrounded by light harvesting

complexes as shown in Figure 1-2. These chlorophyll containing assemblies are capable

of absorbing photons from a broad spectral range of sunlight, which makes them a perfect

light-harvesting antenna.





ET
c: LH2








LH 1
Figure 1-2. Schematic representation of baC light-harvesting complexes.

.. .. *t ; -
*-: ^ ET .

LH1 L- 1

-- ; i-



Figure 1-2. Schematic representation of bacterial light-harvesting complexes.









In the last decade, there has been great devotion to the design and synthesis of

molecular or supramolecular species that can function as antennas in artificial systems.

The first requirements to develop a light harvesting system are that its components must

absorb in a substantial part of the visible spectral region and the light-absorbing units

must be chemically and photochemically stable. In order to have high light-harvesting

efficiency, the excitation energy must be delivered to a common acceptor component.

Several groups have tried to develop artificial light harvesting systems with

custom-designed molecules such as small covalent arrays containing photoactive units,

polymeric and supramolecular systems with multichromophores.17'20 It has been

concluded that linear-chain macromolecules do not have the most ideal architecture for

efficient light harvesting.20 First of all, it is difficult to make polymeric systems with an

energy gradient, which is shown to be vital for vectorial energy transfer. Secondly, most

polymeric chains are flexible enough to form aggregates or excimers which will act as

energy traps. In this regard, dendrimers characterized by their high degree of order and

the possibility to contain selected photoactive chemical units in predetermined sites are

excellent candidates for light harvesting antenna. Proper choice and placement of

chromophores enable the investigation of efficient energy transfer from the periphery to

the core of the dendrimer.

Balzani and coworkers49 have reported initial studies on multichromophoric

dendrimers undergoing intramolecular energy transfer. They incorporated different metal

and ligand combinations into low-generation dendrimers and found that energy transfer

occurs from internal higher energy units to the external lower energy units. The concept

of intramolecular energy transfer was clearly illustrated by these initial reports.5055









However, these structures are not ideal photosynthetic mimics. As mentioned before, the

optimum light harvesting system should have numerous peripheral chromophore

channeling the absorbed energy in a unidirectional manner to a single and central energy

acceptor molecule or complex. The first report of an efficient, unidirectional energy

transfer from a dendrimer structure to a single core chromophore was published by Xu

and Moore.34 These systems and their properties will be explained in details in the

symmetrical phenyl ethynylene dendrimers section.

The Frechet group has then studied the lanthanide -cored poly(benzylether)

dendrimer. In this study, it was shown that the excitation energy was channeled from a

dendrimer shell to a single core unit.33 Later, Jiang and Aida observed similar antenna

effect utilizing a different luminescent core such as porphyrin.56 A variety of structures

were designed and studied differing in the number of poly(benzylether) dendrons

attached to the central porphyrin as well as in the generation number of dendrons,.56 In a

similar manner, Balzani et al. reported a polylysin dendrimer with chromophoric dansyl

units in the periphery playing the role of a ligand for lanthanide ions with efficient

conversion of UV light into light of different frequencies in the visible or near infrared

region.52'53'57 In another approach, also developed by the Frechet group, the flexible

poly(benzylether) dendrimer was functionalized with dye molecules at the periphery that

served as the molecular antenna while the core functionalized with a proper dye molecule

served as the energy acceptor. Both steady state and time-resolved experiments indicated

that the energy migration from the periphery to the core was extremely efficient, thus

most of the absorbed energy is concentrated at a single center.36'58'59









Inspired by natural photosynthetic systems including elegant light-harvesting

antenna of photosynthetic bacteria, systems capable of directional energy transfer

between several chromophores have gained much attention.60 In the absence of a

gradient, the exciton will randomly hop between neighboring localized states following

the photoexcitation. The hopping probability is based on the separation distance between

chromophores. Due to the branched structure, there is an entropic bias increasing the

probability of energy dissipation outwards toward the molecular periphery rather than

inwards toward the core. Recent theoretical investigations of dendrimers without an

energy gradient have shown that the efficiency of exciton trapping at the core decreases

with an increase in molecular size, even though larger number of absorbing units is

present.61'62 On the other hand, the presence of an energy gradient toward the locus will

introduce an energetic bias that will overcome the entropic bias. Two different

approaches are known to create dendrimers with an energy gradient. In one approach,

Moore and coworkers developed a series of dendrimers serving both as the light-

absorbing antenna and as an energy transport medium, which is the case for the

dendrimers investigated in this doctoral research. In the other approach, both periphery

and core are functionalized with donor and acceptor moieties, respectively. The dendritic

framework is photochemically silent and acts as a transparent spacer to separate the

donor groups at the periphery from the acceptor groups at the core.40,63

An example of this later approach was reported by Millen and co-workers64'65 who

investigated structurally well-defined, conformationally rigid dendrimers consisting of up

to three different chromophores. This dendriticc triad" includes globular polyphenylene

dendrimers bearing a terrylene tetracarboxdiimide (TDI) chromophore in the center and









perylene dicarboxmonoimide (PMI) as well as naphthalene dicarboxmonoimide (NMI)

chromophores at the periphery as shown in Figure 1-3. The design of the cascade system

places the naphthalene chromophore at the third branch point, the perylene chromophore

at the second branch point, and the terrylene chromophore at the core representing

spatially the desired direction for energy transfer. These rylene chromophores were

chosen since they possess excellent photostability, high extinction coefficients, and

fluorescence quantum yield close to unity. The triad absorbs over the whole range of

visible spectrum and shows well-separated absorption envelopes. Thus, it is possible to

specifically excite distinct chromophores within the dendrimer, which helps with the

investigation of vectorial energy transfer. Mullen group's work64-66 clearly indicates the

existence of an energy gradient and is consistent with the stepwise energy transfer from

periphery to the center of the molecule.

Recently, single molecule fluorescence studies on the same triad further support

the multi-step unidirectional energy transfer, including a component of direct transfer

from each donor to the acceptor. Note that using a rigid polyphenylene dendrimer

overcomes many possible complications due to conformational mobility. Undesired

chromophore interactions such as aggregation, excimer formation, and even self

quenching of dyes are minimized with a shape-persistent dendrimer.

The Mullen group's work represents the first example of a dendritic triad in which

energy gradient is induced by different types of chromophores spatially and energetically

distributed within the dendrimer and thus independent of the dendrimeric scaffold.






























Figure 1-3. The dendritic triad. Adapted from reference64

Shortly after, Frechet reported the design and synthesis of a cascade light

harvesting system based on a flexible dendrimer scaffold. In order to obtain the required

spatial distribution, coumarin 2 and fluorol dyes were placed at the third and second

branch point of a poly(aryl ether) dendrimer, respectively (Figure 1-4). The final energy

acceptor consisted of a perylene derivative at the core of the dendrimer. Similar to

polyphenylene dendrimers, the dendritic backbone does not participate in the energy

transfer process. The steady state photophysical analysis suggested energy transfer within

this system favoring a cascade route, moving from coumarin groups through intermediate

fluorol units and into a final acceptor ethynyleneperylene chromophore. This system

demonstrates that with proper chromophore selection, vectorial energy transfer process is

generated despite the flexibility of the dendrimer.





























Figure 1-4. Poly(aryl ether) dendrimer functionalized with dyes. Adapted from
reference67

The dendrimers mentioned previously point out the role and necessity of an energy

gradient for efficient energy transfer process as in the natural photosynthetic systems.

However, for these nonconjugated dendrimers, the role of the dendritic backbone is only

structural and not functional (the dendrimer backbone is just a spacer). In conjugated

dendrimers, the backbone itself serves as a tool for the energy transfer. The first built-in

multistep energy gradient within dendrimers was reported by Moore and coworkers using

a phenylene ethynylene (PE) dendrimer, with a repeat unit conjugation length that

increased with generation from periphery to the core.68-70 As a result, HOMO-LUMO

gaps of conjugated repeat units decrease from the exterior to the interior, generating a

directional energy flow toward the core. They also investigated the PE dendrimers

without a gradient, composed of phenylethynylene chains of identical lengths. These PE

dendrimers are characterized as structurally symmetric due to the meta substitution of the

benzene ring at each branching point. Recently, a new type of PE monodendron was









characterized as unsymmetrical since the branches extending outward were structurally

nonequivalent and linked through meta and para positions on the phenyl rings. The

unsymmetrical PE dendrimers also possess an intrinsic energy gradient resulting in

efficient energy funneling.71 The work presented in this dissertation mainly comprises the

investigation of the photophysical properties and the energy transfer processes in several

unsymmetrical PE dendrimers and a symmetric PE extended dendrimer. The primary

photophysical characterization of symmetrical and unsymmetrical PE dendrimers is

summarized in the next section.

Symmetrical PE Dendrimers

Theoretical studies showed that ordered "Bethe trees" might be the optimal energy

funnels.61'72 With this in mind, Moore and Kopelman synthesized and investigated the

photophysics of a series of phenylethynylene dendrimers.72 Two families of these

dendrimers, compact ad extended, are characterized by symmetrical branching. The

branching point is always a meta substitution of the phenyl ring leading to structurally

symmetric macromolecules. Branching at para positions would grow linear chains, while

branching at ortho positions would terminate the tree-like structure quickly due to steric

hindrance. Thus, symmetric geometry for a large dendrimer is optimized with meta

arrangement, which also permits a large degree of orientational flexibility. Deviations

from planar configurations overcome the steric hindrance and enable the synthesis of

higher generation dendrimers.22'73

The main difference between compact and extended dendrimers is the number of

phenylene ethynylene units between consecutive branching points. In compact ones, each

generational unit is composed of identical diphenylacetylene chains. The extended ones

have diphenylethynylene chains around the periphery, but linear phenyleneethynylene









chains show consecutively increasing length toward the center of the molecule. Even

though this seems to be a minor modification to the molecular structure, it introduces

significant electronic and energetic characteristics that greatly enhance the energy

funneling abilities of these systems.70

Recently, Ortiz et al. have presented a theoretical investigation of energy transfer in

the nanostar molecule.74 Molecular dynamics simulations have been performed to reveal

the role of structural changes on the dynamics. The energy transfer rates were calculated

between 2- and 3 -ring chromophores using the ideal dipole approximation (IDA) and the

transition density cube method (TDC). The rapid flipping of the phenyl groups at room

temperature resulted in large changes in transition densities. It was shown that the

traditional Forster model employing IDA was not able to reveal this dynamical effect on

the transfer rates. Also, the accuracy of the IDA fouls when the size of the chromophores

is comparable to the distance between them. However, the rate constants obtained with

TDC were extremely sensitive to the phenyl rotation and therefore expected to yield more

accurate energy transfer rates. In addition, Kleiman et al.75 investigated the energy

transfer in the nanostar with femtosecond degenerate pump-probe spectroscopy. They

measured the recovery time of the ground state absorption of 2-ring and 3-ring

chromophores. The experimental transfer rates were compared with the calculated ones

using the Forster model. Even though there was a qualitative agreement, the rates were

overestimated. Ortiz et al. discussed that the discrepancy between Forster model and

experimental results would be improved by the use of TDC and the data from molecular

dynamics.74









Theoretical calculations by Mukamel et al. indicate that meta branching

electronically decouples the resonative conjugation among PE units.76-79 As a result, ,the

optical excitation is localized on each PE chain in compact and extended dendrimers.

Experimental evidence for this excitonic localization can be seen by steady state

spectroscopy. The absorption spectrum of any compact structure closely reproduces the

spectrum of an isolated PE unit. The total absorption intensity increases monotonically

with generation, but exhibits no red-shift. If excitations were delocalized over the entire

dendrimer backbone, a red-shift should have been observed. Due to identical chain length

of all subunits, compact dendrimers have an energetically degenerate nature, allowing the

exciton to rapidly hop between neighboring localized states. An exciton initially localized

on a particular PE chain will not encounter any energy gradient towards the locus. On the

contrary, an entropic bias is observed which increases the probability of hopping toward

the periphery. Thus, compact dendrimers do not act as energy funnels.

The extended PE dendrimers also exhibit localized electronic excitations due to

branching at meta positions of phenyl rings. The difference lies in the HOMO-LUMO

energy of these localized excitations. It is known that the HOMO-LUMO excitation

energy of a molecule decreases with an increase in the extent of conjugation. While the

single diphenylacetylene? (DPA) chains around the periphery have the greatest excitation

energy, this value decreases monotonically toward the center of the molecule as the chain

length increases. The absorption spectra for the extended dendrimer series also exhibit a

high-energy peak assigned to the shortest DPA chains, but additionally increasing red

shifted peaks are observed associated with longer PE units (3- and 4-ring).80 As a result, a









built-in energy gradient is observed and intramolecular energy transfer in the extended

dendrimer series is well directed from periphery to the core.

The random hopping and funneling characteristics of both compact and extended

PE dendrimers were investigated theoretically. Mukamel and coworkers investigated the

optical properties of such systems using the Collective Electronic Oscillator (CEO)

approach and the Frenkel-exciton model.76'77 In parallel to the experimental results,

theoretical studies have shown that optical excitations involve no charge transfer and are

completely localized between linear segments. Since the meta branching disrupts the

charge transfer between individual PE segments, exciton migration proceeds via

Coulombic interaction and these systems can be represented by the Frenkel Exciton

Hamiltonian. The linear absorption spectra of these dendrimers were calculated using the

CEO approach and showed excellent agreement with the experiment. It was concluded

that the linear segments can be considered as effective chromophores where optical

excitations reside. Upon photoexcitation, the electron-hole pair is confined to a single

chromophore, whereas its center of mass can move around representing energy migration

across the molecule.

The extended PE series represent the first example of a built-in multi-step energy

gradient within dendritic systems and are very efficient photosynthetic mimics. By

functionalizing the core of these structures with the lower-bandgap ethynyleneperylene

chromophore, an energy "sink" is introduced into the system.81 The most studied

extended derivative, both experimentally and theoretically, has 4 generations (2-, 3- and

4- ring PE units) and is referred as the "nanostar" (Figure 1-5a). Excitation of the

nanostar backbone results in emission emanating solely from this ethynyleneperylene









dye. Hence, the PE units act as energy donors and ethynyleneperylene acts as the central

acceptor. The absorption data along with the lifetime data indicate that the

ethynyleneperylene unit at the focal point of nanostar has a well localized excited state.

Figure 1-5b shows the energy level diagram illustrating the vibrationless electronic

excitation energies of each of the localized states in the nanostar.69'82 This graphic

representation points out the impressive energy funneling characteristics of the nanostar.







I 25I-





20
Sitze (schematic)
(a) (b)

Figure 1-5. Nanostar dendrimer (a) Chemical structure (b) Energy level diagram.
Adapted from reference so

Theoretically, the Mukamel group intensively investigated the nanostar molecule

and computed the exciton energies, transition dipole moments, and electrostatic

interactions in PE segments using the CEO method.76,78 They computed linear absorption,

frequency gated fluorescence spectra, and even frequency-domain pump probe signal. In

addition, Tada et al. investigated photoexcited states of the nanostar and singlet

excitations of linear PE units involved in the dendrimer with time dependent density

functional theory and molecular orbital method.83 It was concluded that the orbitals of


nanostar are localized in space as well as in energy. While the steady state comparison









with these calculations implies the model of weakly coupled Frenkel excitons, the

dynamics of the excited states in PE dendrimers may lead to a different picture of

excitations. This issue is crucial to understand the nature of electronic excitations and

further calculations have been recently carried out by Bardeen and coworkers. They

reported both high-level electronic structure calculations and steady-state experiments

based on the smallest building blocks of PE dendrimers.84 Their emission spectral shapes,

radiative lifetimes, and anisotropies change dramatically with increasing number of

substituents. This yields strongly coupled diphenylethynylene units and contradicts

previous findings. The excited state electronic structure was investigated theoretically

using ab initio CASSCF and CASPT2 calculations and the electronic coupling was found

to vary with molecular geometry. In particular, the presence of large electronic coupling

in the emitting geometry was not seen for the absorbing geometry of the same molecule.

In order to analyze the variability in electronic coupling, they extended their ab initio

results in terms of the Harcourt model.85 This model was developed to classify different

interactions such as through bond, through space and charge transfer interactions between

coupled chromophores. The relative roles of these three interaction terms and their

dependence on meta- versus para- substitution were investigated in detail. However, the

experiments in larger dendrimers do not show the spectral features (shifts) predicted in

the smaller systems.7086 So, the nature of the excited states for PE dendrimers remains an

open question.

In another study to address the complicated photoexcitation energy transfer

between subunits of a dendritic system, a simple compact model system with one

branching center was investigated by Goodson and coworkers.87'88 In that study, each









component of the branching center was a conjugated linear segment. The interactions

between these segments (chromophores) are strongly influenced by the electronic and

structural connectivity of the branching center. Vamasvski et al. have investigated the

nature of these interactions using fluorescence anisotropy.87 For example, the very fast

depolarization rate in a nitrogen centered triphenylamine molecule indicates strong

electronic coupling between segments. In another publication by the Goodson group,

fluorescence anisotropy dynamics of a system containing pyridine distyrlbenzene

chromophores attached to benzene center was reported.89 Their results confirmed that the

benzene branching center acts as a weak coupler and electronic delocalization across the

branching center is hindered by meta substitution of the chromophores. It is important to

recall that phenyl is the branching center between the individual chromophores in PE

dendrimers. Therefore, most of the theoretical and experimental studies are

complementary in terms of verifying the localized nature of excitations for symmetrical

PE dendrimers, with the exception of Bardeen's results.84'90

Our group has an ongoing collaboration with Prof. Jeff Krause and Prof. Adrian

Roitberg to investigate this unique symmetric PE dendrimer nanostar. As discussed in

detail in Chapter 5 of this thesis, we performed femtosecond time-resolved experiments

on the nanostar. We explored the excited state dynamics by measuring the fluorescence

from both the ethynyleneperylene trap (acceptor) and the dendritic backbone (donor).

Broadband transient absorption following the excitation at different chromophores was

also examined. In addition, room and low temperature steady-state absorption spectra of

the individual PE components and the nanostar were measured and compared to

theoretical calculations performed to predict the spectra.









Unsymmetrical PE Dendrimers

Most of the dendrimers developed for light harvesting applications have

symmetrical structures. As discussed in the previous section, for symmetrical PE

dendrimers, the electronic communication between the periphery and the core has to go

through each sub-branch. For unsymmetrical dendrimers, shortcuts exist between the

periphery and the core, such that in some cases shorter PE chains are directly attached to

the longest chain extending to the core. The core of the dendrimer can directly

communicate with the periphery. Therefore, it is anticipated that unsymmetrical

dendrimers may be better light-harvesting antenna molecules. To prove this concept,

Peng and coworkers91-94 reported a new class of conjugated PE dendrimers based on

unsymmetrical branching which occurs at both ortho and meta positions of the branching

benzene rings and leads to nonequivalent branches. Two features are crucial for

unsymmetrical PE dendrimers: rapidly increasing conjugation length results in broad

absorption spectra and the conjugation length increases toward the core generating an

intrinsic energy gradient.

Figure 1-6 shows the structures of unsymmetrical PE monodendrons. These

dendritic molecules have both ortho and meta substitution while the symmetrical ones

have only meta linkage. The conjugation length of the longest chain is significantly larger

compared to the meta-linked dendrimer composing of the same number of phenyl

ethynylene groups. Therefore, they are expected to have different optical properties.











































Figure 1-6. Chemical structures of unsymmetrical PE dendrimers

Figure 1-7 shows the absorption and emission spectra of GnOH (n=1-4)

unsymmetrical PE dendrons. As the generation number increases, the lowest excitation

energy (absorption band edge) shifts to longer wavelengths and the molar extinction

coefficient increases. Even though the ortho substitution prevents the phenyl rings from

having a planar geometry due to steric hindrance, the effective conjugation length clearly

increases with each generation. Variable conjugation length throughout the dendrimer

accounts for the much broader absorption spectrum than that of the meta-linked











symmetric analogs. This broad absorption spectrum along with direct electronic

communication between periphery and the core could possibly make these types of

dendrimers more efficient energy transfer funnels. As seen in the emission spectrum,

fluorescence is red shifted for higher generation dendrimers. Following the

photoexcitation, the energy transfer will proceed from shorter conjugation-length

segments to the longest conjugated segment, thus for a given dendrimer, emission is only

observed from the longest chain. The fluorescent quantum yield of GnOH dendrimers

varies from 40% up to 80% (in CH2C2).



10- --GOH 10 GOH
-G OH -G OH
08 GOH 08- G3OH
H OH
-G OH \\ GOH
06
0 06 -

-04 04
< 8

02 0 -
022
002
00 00
250 300 350 400 450 500 300 350 400 450 500 550 600
Wavelength (nm) Wavelength (nm)

Figure 1-7. Absorption and emission spectra of GnOH monodendrons. Adapted from
reference71

Since it is hard to differentiate each segment within the unsymmetrical dendrimers,

it would be hard to quantitatively evaluate the energy transfer efficiency. However,

analogous to the symmetrical PE dendrimers, an ethynyleneperylene unit serving as

energy trap has been attached to the focal point of the dendritic backbone.

Ethynyleneperylene, having well-separated absorption from the dendrimer absorption,

will help explore the excitation energy transfer. The absorption and emission spectra of

GnPer series are shown in Figure 1-8. The absorption features of perylene can be clearly

distinguished from the PE backbone. The perylene bands are approximately 50 nm red-









shifted compared to the free perylene in the CH2C12, which indicates the delocalization of

perylene transition dipole over the PE backbone. The fluorescence spectra of GnPer series

following 350-nm excitation is also shown in Figure 1-8b. For any given dendrimer, the

emission is almost entirely from the perylene trap, which indicates very efficient energy

transfer from the dendrimer backbone to the ethynyleneperylene trap. Melinger et al.

presented a detailed photophysical characterization of unsymmetrical PE monodendrons

in various solvents.95 They reported steady-state absorption and fluorescence

measurements along with the time-dependent fluorescence measurements for PE

monodendrons up to fourth generation. The photophysical properties of unsymmetrical

PE monodendrons were compared to those of symmetrical PE dendrimers. In addition,

ultrafast degenerate pump-probe spectroscopy was applied to G3OH and G3Per to explore

the excited state dynamics.95 These initial measurements suggested that energy transfer

process into the ethynyleneperylene trap occurs in a subpicosecond time scale. However,

a direct measurement of the trapping time is yet to be ascertained through multi-color

pump probe experiments or by measuring the time evolution of the ethynyleneperylene

fluorescence with femtosecond resolution. This is one of the goals of this dissertation.

50, 1.0 ,
/-'. --GiPer b 0.9
/ G3Per
C --"r -'" -- G4Per o .7 ao io (


m -Aap f'ro.. 8 ne 0.3,

1.0

0.0 ________,, o0.0 ...
250 300 350 400 450 500 550 400 450 500 550 600 650 700
Wavelength (nm) Wavelength (nm)
(a) (b)
Figure 1-8. Steady state spectra of GnPer monodendrons: absorption (a) and emission (b).
Adapted from reference96









Recently, Peng and coworkers synthesized PE didendrons along with some

tridendrons so that an extensive investigation is possible to establish a structure-property

relationship regarding energy transfer and x-conjugation.97 In that work, two PE

monodendrons were linked by a phenyl ring at meta positions and these new structures

were named as 2GnOH series. For 2GnPer dendrimers, an ethynyleneperylene unit was

attached to the central benzene ring in meta position. Due to meta substitution at the core

of the dendrimer, the conjugation is expected to be disrupted and an extra degree of

localization might be provided between the monodendrons and/or between the

monodendron and the ethynyleneperylene unit. The chemical structures of didendrons are

illustrated in Figure 1-9. Through collaboration with the Peng group, we obtained the

didendrons studied in this dissertation. Mainly, 2GnOH and 2GnPer ( n=1,2) are studied

via both steady-state and time-resolved spectroscopic techniques in Chapter 3 and 4.

Even though initial studies with unsymmetrical PE monodendrons reveal some

spectroscopic evidence for the highly efficient and ultrafast energy transfer, questions

remain regarding the nature of electronic excitations and related mechanisms for the

transfer process. The ultrafast absorption and emission experiments designed and

performed in this dissertation aim to explore these questions.

Unsymmetrical PE dendrimers are an attractive prospect as they offer a handle to

obtain various extends of conjugation and help in understanding electronic structure-

property relationship for better light-harvesting systems. The processes following the

optical excitation in a molecule and basic energy transfer mechanisms will be explained

briefly in the following section.
















2G1X 2G2X







2GnOH (n=1-3): X=OH

2GnPer(n=1-3): X=
2G3X
Figure 1-9. Chemical structure of PE didendrons.

Excitation Energy Transfer

The motivation behind the work presented in this dissertation is to identify the

mechanisms of intramolecular electronic excitation energy transfer in light-harvesting

dendrimers and the structure-function relationship that make energy transfer very

efficient in these systems. Other than self-relaxation processes, the excited states may

relax to the ground state via transferring the electronic excitation to other chromophores

present in the system by a bimolecular process. During this process the excited donor

chromophore D* returns to its ground state with simultaneous transfer of its electronic

energy to the acceptor chromophore A:

D* + A D +A*

Subsequently, the photoexcited chromophore A* may proceed either giving a

sensitized photochemical reaction or exhibit sensitized photoluminescence. Under these

conditions D chromophores are termed as sensitizers while A chromophores as activators.

There are mainly two conditions required for energy transfer to occur: (i) the energy of









D* should be higher than the energy of A*; (ii) the energy transfer process should be

faster than the natural lifetime of D*. The electronic energy transfer can be described

further according to the photophysical processes that are involved in it.

Radiative Energy Transfer

Radiative energy transfer is a two-step process: a photon is first emitted from the

excited donor and then it is reabsorbed by the acceptor ground state:

D* D + hu

hu + A A*

Since the transfer mechanism is based on a radiative step (i.e. photons emitted by

one chromophore are absorbed by a second chromophore), this process is named as the

"trivial" case of electronic energy transfer. It does not involve the direct interaction of

donor and acceptor. The most important factor that influences the process is the quantum

efficiency of the donor in the spectral region where the light-absorbing ability of the

acceptor is high. The trivial transfer is favored when the following conditions are met:

high quantum yield of D*, high concentration and extinction coefficient of A, and good

overlap between the emission of D* and absorption of A. This kind of energy transfer

might be the dominant mechanism in dilute solutions since the dependence of energy

transfer efficiency on separation distance between donor and acceptor chromophores is

weak. The viscosity of the solvent does not affect the rate of radiative energy transfer.

Radiative transfer results in a decrease of the donor fluorescence intensity in the

region of spectral overlap. When donor and acceptor chromophores are identical and

emission and absorption overlap sufficiently, the observed fluorescence lifetime increases

as a result of repeated absorption and emission radiativee trapping).









Non-radiative Energy Transfer

Non-radiative energy transfer is a single step process that requires donor-acceptor

interaction as a result of spectral overlap between donor's emission spectrum and

acceptor's absorption spectrum. D*- D and A- A* transitions are isoenergetic,

implying that several vibronic transitions in the donor will have the same energy as the

corresponding transitions in the acceptor. Such transitions are coupled, and are in

resonance as shown in Figure 1-10. For non-radiative energy transfer, the terms

resonance energy transfer (RET) and excitation energy transfer (EET) are often used. If

the excited state vibrational relaxation is faster than the energy transfer, and if the energy

transfer is a vertical process as implied by the Franck-Condon principle, the spectral

overlap can be evaluated using:

00
J= IjD( V )dv
(1-1)

This integral is proportional to the number of resonant transitions in the emission

spectrum of the donor and absorption spectrum of the acceptor as illustrated in Figure 1-

10. The spectral distribution of the donor emission and the acceptor absorption are

normalized to a unit area on the wave-number scale:


ID(()dv =f ()d = 1 (1-2)
0 0

According to this required spectral overlap condition for normalized spectra, it is

clear that the magnitude of the spectral overlap is not connected to the absolute values of

the oscillator strengths of the transitions that are involved in the process.

There are two different interaction mechanisms related to non-radiative energy

transfer process: coulombic and exchange interactions. The coulombic interactions






31


consist of long-range dipole-dipole interactions and short-range multipolar interactions.

The interactions due to intermolecular orbital overlap include two electron exchange

(Dexter mechanism) and charge resonance interactions, which are effective in short

range. The total interaction energy can be written as the sum of a coulomb term, Uc and

an exchange term Uex. Considering that only two electrons are involved in a transition

(one

on D and one on A), the energy transfer mechanisms are schematically represented in

Figure 1-11. In the figure, the empty circles represent the electrons whose interactions

with other electrons are assumed to be roughly constant during the energy transfer step.

DONOR ACCEPTOR

Emission Emission
Absorption Absorption









D*
.AE A*


3 2 1 1* 2* 3*



D A A
S i -------------------------------

resonant transitions

Figure 1-10. Model picture for energy transfer showing resonant transitions of donor and
acceptor, and spectral overlap of donor emission and acceptor absorption.









The coulombic term corresponds to the process in which the initially excited

electron on the donor (1) returns to the ground state orbital on D, while simultaneously an

electron on the acceptor A (2) is promoted to excited state (Figure 1-11, top). The

exchange term is represented with an exchange of two electrons between D and A, which

is analogous to a moving particle transferring its energy to the other particles via

collisions. Coulombic resonance interaction occurs via electromagnetic field and does not

require physical contact of interacting donor and acceptor. The basic mechanism involves

the induction of a dipole oscillation in A by D*, thus the Coulombic mechanism can be

effective at large distances (up to 80-100 A). As shown in Figure 1-1 l(bottom), the

exchange interaction represents a "double" electron substitution reaction, i.e., the electron





LUMO

'% H
/%;/O%,H0 I,---

HOMO 2_ _

D* (1) TA (2) D (1) TA* (2)

electron electron
Exchange n
LUMO

electron HE electron
Jx c han ge < change e
HOMO _

D* (1) TA (2) YD (2) TA*(1)

Figure 1-11. Schematic representation of energy transfer mechanism. Top: Coulombic
mechanism. Bottom: Exchange mechanism98









initially on D* jumps to A simultaneously with the jump of an electron on A to D*. This

exchange interaction occurs via overlap of the molecular orbitals requiring physical

contact between donor and acceptor. The interaction is operative only at short-range (10-

15 A) because the electron density exponentially decreases outside the boundaries of

molecules.

For allowed transitions on D and A (no change in spin), coulombic interaction is

the predominant mechanism. Thus, singlet-singlet energy transfer such as

1D* + 1A 1D + 1A* and

D* + 3A D + 3A* are fully allowed.

However, for forbidden transitions on D and A:

3D* +1A 1D+3A

coulombic interaction is negligible and triplet-triplet energy transfer is only due to orbital

overlap. It should be noted that for singlet-singlet energy transfer, both interactions may

be involved, but in general coulomb mechanism predominates.

The interaction energy describing the coupling between the initial and final states is

given by:

U= ( Y H ) (1-3)

where H contains the electrostatic interactions of all electrons and rV and Vf are the

electronic wavefunctions for the initial and the final excited state, respectively.

Considering that only two electrons are involved in a transition, the antisymmetrized

product wave functions of the initial and final state can be written as:









S=-- ( D. (1) (2)- Df. (2)7 (1))
(1-4)
Vf = -- (v (1)v4. (2)- vD (2)v. (1))


where the numbers 1 and 2 refer to the two electrons involved. The total interaction U

can be written as the sum of the Coulomb and the exchange terms:

U=( -(D. (1), (2) .(2)Vf(1)) H' (VD(1)V. (2) VD(2)V (1))) (1-5)


U = D(1). (2)| H' VD(1)V.(2))
(1-6)
Ue = D ((1)V/ (2) H yD (2)A. (1))

U = U U (1-7)

The Coulomb term, Uc, represents the classical interaction of the charge

distributions and may be expanded into multiple terms: dipole-dipole, dipole-quadruple,

etc.,Dipole-dipole interaction dominates for allowed transitions:

Uc U7 = 1 (1-8)
40E0 RDA

where |tD and [tA denote the transition dipole moments of the two molecules (D*-- D and

A-A*) and RDA is the distance between the donor and the acceptor. Here, the orientation

factor K is defined by:

K = p/D ./ 3(D .RDA )(A RDA (1-9)
K = 2 cos OD cos OA + sin 0D sin 0A cos (p

The vectors, angles and separation between dipoles are defined in Figure 1-12.

When Udd is expressed in cm-1, the transition moments are in Debye and RDA is in nm. (p

is the angle between two transition moments and OD and OA are the angles between each









transition moment and the vector connecting them. Considering this picture, it is clear

that no interaction would be observed between perpendicularly oriented chromophores.



PD D A 9A




Figure 1-12. Definition of the angles used to calculate the orientation factor between the
dipoles.99

The dipole-dipole approximation provides reliable estimation of the electronic

coupling between point dipoles, i.e. when the donor acceptor separation is much larger

than the molecular size of the donor and acceptor transition dipole moments. At short

distances or when the separation is comparable to molecular dimensions (large dipole

moments), the point dipole approximation is not valid, and a better description of the

shape of the dipoles should also be included in the calculations. The transition moment

magnitude [t (in units of Debye) is related to the dipole strength of an absorption band,

measured in media of refractive index n, according to:

4.3x109
/2 f= 4.3x (T-)dv (1-10)
n

Thus, the Coulomb interaction term can be related to experimentally measured quantities.

The exchange interaction is a purely quantum mechanical phenomenon and does

not depend on the oscillator strengths of the transitions involved. The exchange integral

2
Uex = (1)VD A (2) D ,(2)V.A(1)) (1-11)
r12









represents the interaction of charge densities separated by a distance r12 Since charge

densities depend on the spatial overlap of orbitals of D and A, the exchange interaction

decreases exponentially with increasing internuclear distances.

The non-radiative transfer rate is basically given by Fermi's Golden Rule.

According to the Golden Rule, the rate of transition between two states is related to the

magnitude of a perturbation which changes the positions or motions of particles of the

initial state and reshapes the initial state so that it looks like the final state. Beyond the

Born-Oppenheimer approximation, it is necessary to include the interaction between

different vibrational-electronic molecular states in order to describe such transitions.

Using the time-dependent perturbation theory, the rate constant kET is formulated as:

k = V2p= -( H' )2 p (1-12)


where p is the density of interacting initial and final states as determined by Franck-

Condon factors, and it is related to the spectral overlap between the emission of the donor

and the absorption of the acceptor in a system without inhomogeneous broadening. Using

the relations for the interaction energy given above, Forster and Dexter derived

expressions for the rate constant of energy transfer using the Coulomb and the exchange

mechanism, respectively.100'101

Dexter's formulation points out that an exponential dependence is expected from

the exchange mechanism. The rate constant for transfer can be written as:

k = 2KJexp(-2r / L) (1-13)
h

where J is a spectral overlap integral with the normalization condition, L is the average

Bohr radius and K is related to specific orbital interactions, not related to any









spectroscopic data. Thus, it is difficult to characterize the exchange mechanism

experimentally.

While the nature of the interaction, whether Coulombic or exchange, is an

important task to investigate, the magnitude of the interaction also needs to be explored.

Forster proposed to discriminate between very weak, intermediate (weak), and strong

electronic coupling depending on the relative values of the interaction energy (V), which

is the pure electronic energy difference between D* and A*(AE), the absorption

bandwidth (Aw), and the vibronic bandwidth (As) (Figure 1-13).

Strong coupling. In this case, the intermolecular interaction, Vo, is much larger

than the width of the individual transitions D-D* and A-A*. Then, all the vibronic

subtransitions in both molecules are virtually at resonance with one another. The transfer

of excitation energy is faster than the nuclear vibrations and vibrational relaxation. The

absorption spectra of strongly coupled systems will be different from those of the

individual components. The donor and acceptor electronic states will mix to produce


D* A*


Aw$ As




AE Y////////////



STRONG COUPLING V >> AE V >> Aw, As
WEAK COUPLING V >> AE Aw >> V >> As
VERY WEAK COUPLING V << As <
V: interaction energy

Figure 1-13: Differences between strong, weak, and very weak coupling.









new, delocalized states. Thus, the transfer of excitation is a coherent process and the

excitation oscillates back and forth between D* and A*. The rate constant is derived as:

4\V
k 4V (1-14)
h

where V is approximated by dipole-dipole interaction. The distance dependence of V, and

consequently ofk is r3.

Intermediate (Weak) coupling. This is a particularly challenging case to model

RET dynamics. The interaction energy, V, is larger than the width of an isolated vibronic

level but much smaller than the full absorption bandwidth. Compared to the strong

coupling case, the electronic excitation is more localized. On the other hand, the vibronic

excitation is still delocalized and the system can be described in terms of stationary

vibronic exciton states. The transfer rate is fast compared to vibrational relaxation .It is

approximated as:

4 VS 2
kT & (1-15)
h

where S,, is the vibrational overlap integral of the intramolecular transition v <- w.

Since S,,<1, the transfer rate would be slower than in the case of strong coupling.

Very weak coupling. The interaction energy is much lower than vibronic and

absorption bandwidth. The vibrational relaxation occurs before the transfer occurs. The

absorption spectra of the components are not altered. The transfer rate is given by:

4 2 (rVS, )2
k 4 (1-16)
hAs

The characteristic feature of this very weak coupling case is the quadratic

dependence of the transfer rate on the interaction energy, as opposed to the linear









dependence found for the intermediate and strong coupling. For dipole-dipole interaction,

the distance dependence is r6, whereas it is r-3 for the preceding cases.

Forster theory for RET is formulated for the very weak coupling limit. It is based

on an equilibrium Fermi Golden Rule approach with a second-order perturbation theory

treatment of the electronic coupling between donor and acceptor. The theory was

established for coupling of a state to a quasi continuum of secondary states. The primary,

discrete states are non-stationary and they carry all the oscillator strength. Such model

allows rationalizing dynamic processes involving decay of a "stationary" state for

radiationless transitions. In other words, the transfer rate occurs after vibrational

relaxation. In this way, Forster model evaluates a Fermi Golden Rule expression for the

RET rate, where the matrix element of interaction between excited state donor and

ground state acceptor is purely electronic coupling V. Conditions of energy conservation

and nuclear overlap factors, separated from the electronic coupling, relate the donor

emission and acceptor absorption events:


kT =- ds f P(k)P(l) u(edE ) 2 (1-17)
0 k I

where s is the energy gap of the donor molecule, P(k) is thermal population of mode k

in the excited state, and Sak is the energy of the acceptor ground state. The matrix element

of interaction between the excited state donor and ground state acceptor is independent of

energy and can be written as:

u = IIP(k)P(l) u(Ed kk, =VI2 J(E) (1-18)
k I

where J(E) is the spectral overlap between donor emission f(s) and acceptor absorption

a(s), and can be written as J(s) = f(s).a(s) f(s) and a(s) have each been normalized to









unit area on an energy scale: f(s)dE = a(s)de = 1 .The rate constant can be rewritten

as:


kET=-V2 dEJ()= 1 dvJ(v) (1-19)
0 0

where is expressed in units of cm1, and v = e / 2;ihc. In summary, estimating a rate for

RET in the weak coupling limit requires the knowledge of the electronic coupling, V, and

the spectral overlap, J, between donor and acceptor transitions. The final rate equation is

defined as:

42-2J
k h2c V12 (1-20)


It is important to note that J in this equation is defined as J(cm)= If(A)a(A)A2dA and

has units of cm.

By substituting the Coulomb interaction energy term, Vc, into the Golden Rule

equation, and assuming a dipole-dipole approximation, Forster derived the final

expression for energy transfer rates based on spectroscopically measurable parameters.

The Forster equation is:

k 90001n(10)2DD" (1-21)
S 128;r'n4NARDA6D(

where K2 is the orientation factor of the transition dipole moments, ,D is donor quantum

yield, TD is the donor lifetime in the absence of the acceptor, NA is Avogadro's number,

and n is the refractive index of the solvent. Note that in this equation spectral overlap J is

defined as:










jA = JfD (A)(A)A4 = JD v j dv (1-22)
0 0 (V)4


fD (A) is the fluorescence spectrum of the donor normalized so that [fD (A)dA = 1 and
0

EA (A) is the molar absorption coefficient of the acceptor. Hence, JDA has units of

M cm3

At a specific distance of RDA, the rate at which D* emits light is equal to the rate at

which it transfers its excitation energy to A. At this critical distance Ro, called Forster


radius, one can write k = -, then inserting this into rate equation, and solving for Ro
TD

yields:

6 90001n(10) CD (1-
S 1285n4NA DA (1-23)


Ro is the donor-acceptor distance at which the probability for energy transfer is equal to

0.5. The energy transfer rate can also be written in the following form:


kET=- (1-24)


The transfer efficiency is defined as:


ET k= kE k (1-25)
kD +kET i/ T+ kET

Using the preceding equation, the transfer efficiency can be related to the ratio R/Ro by:

1
T =- (1-26)
1+(R/R0)6

Note that the transfer efficiency is 50% when the donor acceptor distance is equal to the

Forster critical radius.









The Forster equation for RET is accurate provided that four conditions are

satisfied: (i) a dipole-dipole approximation for the electronic coupling can be utilized

appropriately for the donor-acceptor interaction, (ii) the donor fluorescence lifetime,

emission line shape, acceptor absorption line shape, and oscillator strength are not

perturbed because of interactions among donors and acceptors (weak coupling), (iii)

inhomogeneous line broadening is absent in the donor and acceptor line shapes, (iv) the

energy transfer dynamics is incoherent.

In this dissertation, energy transfer processes in various PE dendrimers are studied.

In addition to characterizing a kinetic model for each molecule, the processes according

to the strength of D-A coupling will be classified and the validity of previously used

Forster approximations is tested.

Outline of the Dissertation

The main scope of this work is to investigate energy transfer processes in

conjugated, symmetrical and unsymmetrical phenyl ethynylene (PE) dendrimers. We

have conducted experimental studies to improve the understanding of electronic structure

of these molecules and its effect on the light-harvesting properties. Chapter 1 is an

introduction to dendrimers and a survey of the current scientific literature regarding

dendrimer photophysics. This Chapter also includes a brief introduction to energy

transfer theory. Chapter 2 summarizes the experimental methods utilized to study time-

resolved emission and absorption characteristics of PE dendrimers. This chapter is

complemented with Appendix A, a brief description for experimental beginners on how

to successfully perform these novel experiments.









Unsymmetrical generation one didendrons are investigated in Chapter 3. The

excited state dynamics is studied using both time-resolved and steady state spectroscopy.

In an attempt to more quantitatively analyze the results, a kinetic model is proposed.

Chapter 4 describes the excited state dynamics of generation two didendrons and

we perform a comparative analysis by transient absorption and time-resolved emission

spectroscopy. To explain the multicomponent rise and decays of the excited states, the

basic kinetic model (described in Chapter 3) is extended to a more complex level. Apart

from the initially excited state, the presence of a second state is verified via low

temperature absorption and excitation anisotropy measurements. This chapter further

evaluates the effect of generation on transfer rates.

In Chapter 5, we study a symmetrical PE dendrimer, namely "the nanostar" with

very detailed experiments and global analysis. In contrast to unsymmetrical dendrimers,

this molecule can be considered as a combination of individual chromophores with

variable conjugation length. As reference compounds, 2-,3-, and 4-ring phenyl

ethynylene chromophores are studied with transient absorption spectroscopy. Particular

emphasis is given to recent developments in theories and the proposed kinetic model that

offer physical understanding of the energy transfer.

Chapter 6 describes an independent project that we collaborated with Dr. Schanze's

group at UF. The dynamics of fluorescence quenching of a conjugated polyelectrolyte by

a cyanine dye are investigated by fluorescence up-conversion and polarization resolved

transient absorption. The data are analyzed with a model based on the random walk of the

exciton within the polymer chain and a long-range direct energy transfer between

polymer and dye.






44


Appendix A is provided to supplement experimental details and tricks for the

upconversion technique. Appendix B gives a brief introduction to the global analysis

method. The Singular Value Decomposition analysis combined with the kinetic model for

each molecule is given in detail.














CHAPTER 2
EXPERIMENTAL METHODS

The light harvesting properties of dendrimers lead to the broad investigation of

energy transfer processes. It becomes important to evaluate the electronic structure of

dendrimers and its effect on energy transfer mechanisms and associated dynamics. This

chapter provides background information on the experimental methods employed to

explore excited state dynamics of the dendrimers studied throughout this dissertation. An

overview was given about the relaxation processes occurring after the photoexcitation of

molecules in chapter 1. After photoexcitation, excited molecules in solution undergo

various relaxation processes, which can be classified in four major categories: electronic,

orientational, vibrational relaxation, and solvent relaxation.102 Here, we are mainly

interested in electronic relaxation processes such as transferring the excitation energy into

a specific trap. Steady state absorption and fluorescence spectroscopy give some insights

into the spectral composition of the dendrimers and energy transfer pathways. For better

understanding of the energy transfer processes, time resolved techniques, such as

transient absorption and time resolved fluorescence, are extensively used in this work.

By means of presented experimental techniques, important "ultrafast" phenomena

such as energy transfer and nature of excitations in conjugated systems are studied within

this dissertation.

Chemicals and Materials

Throughout this dissertation, we investigate unsymmetrical conjugated

pheneylethynylene (PE) mono- and di-dendrons and a unique symmetrical PE structure









named the "nanostar". The detailed synthesis and structural characterizations of such

conjugated PE dendrimers are reported in literature.69'70'81'82'103 These dendrimers are

supplied to us by our collaborators, Prof. Zhonghua Peng from the University of

Missouri-Kansas City and Prof. Jeffrey S. Moore from the University of Illinois at

Urbana-Champaign. They confirmed the structure and purity of our samples by thin-layer

chromatography, elemental analysis, 1H and 13C NMR spectroscopy, and matrix-assisted

laser desorption/ionization time-of-flight (MALDI-TOF) mass spectroscopy. 104 These

samples are used as received. For steady state and time-resolved spectroscopic

measurements, the dendrimer samples are dissolved in dichloromethane (CH2C12). The

solvent is purchased from Aldrich and the purity was UV-spectroscopy grade (>99.9%).

It is kept under nitrogen and used without any further purification. In order to perform

steady state measurements at low temperatures, a liquid nitrogen flow cryostat is used to

control the temperature in the range from 77 K to 298K. Methylterahydrofurane

(MeTHF), purchased from Aldrich, is used as a solvent. To obtain a glassy sample,

MeTHF is further purified and distilled to be anhydrous prior to each measurement.

1,2 diphenylacetylene (2-ring, DPA) is purchased from Aldrich and used as

received. 1,4 bis(phenylethynyl) benzene (3-ring, para) was also purchased from Aldrich,

but it is purified by recrystallization from toluene, yielding analytically pure material as

determined by elemental analysis performed by Joseph Melinger at Naval Research

Labarotories (NRL).105 4,4'-bis(phenylethynyl)-tolane is synthesized by Prof. Andrew

Beeby's group at the Department of Chemistry, University of Durham. The sample is

used as received. 1,3 (bisphenylethylnyl) benzene (3-ring, meta) is synthesized in house

according to procedures reported elsewhere.35









The conjugated polyelectrolyte poly(phenylene ethynylene) sulfonate (PPESO3)

and the cationic dye molecule (HMIDC) used in the work presented in Chapter 6 are

obtained from the group of Prof. Kirk S. Schanze at the University of Florida. The

synthesis of PPESO3 has been described recently. 106,107 The average molecular weight,

Mn, of the polymer is estimated to be 100 kDa, corresponding to about 200 monomer

units. HMIDC is purchased from Aldrich and used as received. Solutions with different

HMIDC concentrations are prepared and labeled according to their steady state

quenching efficiency.

Steady State Measurements

Steady state absorption spectra of the samples is recorded with a UV-VIS Varian-

Cary 100 spectrometer. The wavelength range detected is from 190 to 2200 nm with 1

nm spectral resolution. The steady state emission spectra are measured with a Jobin-Yvon

instrument (Spex Fluorolog-3) as a function of wavelength. For room temperature

spectroscopic measurements, all samples are dissolved in dichloromethane. The optical

density (OD) of the samples is approximately 0.2 -0.3 mm-1 at the absorption maximum.

Why Time-Resolved Spectroscopy?

Time-resolved spectroscopy is defined as "any technique that allows to measure the

temporal dynamics and the kinetics of photophysical processes".108 The development of

ultrafast lasers and pulse shaping techniques, among other innovations, have opened up a

wide range of investigations of complex systems in chemistry, physics and biology.

After interacting with a short light pulse (from milliseconds to femtoseconds), the sample

under investigation will change spectroscopic properties such as energy, polarization, or

phase. The fate of the ground and excited states of the system can be determined by

investigating energy and charge transfer processes, coupling of electronic and vibrational









degrees of freedom, vibrational and conformational relaxation, isomerization, etc. In the

case of light harvesting systems, such as the dendrimers studied throughout this

dissertation, ultrafast spectroscopy is used to study energy transfer processes. These

processes often take place on a (sub)picosecond time scale, which puts strict

requirements for the light source to be used in these experiments. This source should

provide very short light pulses with appropriate wavelengths within the electronic

transitions of the system, and with suitable power and reasonable repetition rate.

Mode-locked Ti-Sapphire laser systems, such as the oscillator called Tsunami,

produce less than 40 fs pulses with a very good stability. However, the pulse energies (nJ

regime) is not enough for many spectroscopic measurements and the high repetition rate

of 80 MHz is too fast for transient absorption techniques. In our lab, the amplification of

these pulses are achieved by means of Ti-Sa Regen Amplifier (Spitfire) pumped with a

Q-switched Nd:YLF laser (Evolution X). Therefore, it is possible to obtain pulses up to

0.90 mJ with 1kHz repetition rate. The output wavelength of the Regen Amplifier is

centered at 790 nm, which is not suitable for many natural and synthetic light-harvesting

systems. This problem is solved by employing OPAs combined with second and fourth

harmonic and sum frequency techniques, which are tunable within a broad spectral region,

from UV to IR (300-5000 nm).109

The Laser System

To perform time-resolved experiments with femtosecond time resolution, short and

intensive laser pulses with variable photon energies are required. In addition to a

commercial laser system, nonlinear elements are used for our applications. Figure 2-1

shows a diagram of the laser system. The function of individual components will be

discussed briefly.












Nd:YVO4 Nd:YLF OPA 1
Laser (1) Laser (3) 1 (5)
Pulses
1= 300-900 nm
FWHM= 100-200 fs
Ti-Sa (2) A Ti-Sa (4) 1JkM 100-00
Oscillator Regen Amplifier 4 (5
X= 800 nm = 800 nm
FWHM=35 fs FWHM=50 fs
80 MHz 1 kHz
10 nJ ImJ


Figure 2-1. The laser system for the production of tunable femtosecond laser pulses with
high energy per pulse.

(1) Millenia Vs: The Spectra-Physics Millenia Vs uses the output from a diode

laser to pump Nd+3 ions doped in yttrium vanadate crystalline matrix (Nd:YVO4). An

LBO crystal converts the 1064 nm light from the laser crystal to the green light, 532 nm,

which becomes the output of the laser. Millenia Vs is an all solid-state CW laser, which

offers near diffraction limited TEMoo beam quality and ultra-low noise with an output

power range from 2W to 10 W. Millenia's output pumps the mode-locked Ti-Sa

Oscillator.

(2) Ti-Sa Oscillator Tsunami ": This laser, with a titanium sapphire crystal as the

laser medium-called Tsunami from Spectra Physics, provides very short (35 fs), but

relatively weak laser pulses with 80 MHz repetition rate. The output spectrum is peaked

at 790 nm and has -45 nm bandwidth (FWHM). This output is the seed of our

regenerative amplifier.

(3) Evolution X: Evolution is a diode pumped (by four AlGaAs laser diode

arrays), intracavity doubled Nd:YLF laser capable of producing Q-switched pulses with









average powers > 6 W at 527 nm. The laser resonator is acousto-optically Q-switched at

repetition rates of 1 kHz. It offers high efficiency, low maintenance, and excellent beam

quality. It ideally pumps Ti-Sapphire ultrafast amplifiers, and has been optimized as a

pump source for the Spitfire regenerative amplifier system.

(4) Regenerative Amplifier Spitfire: Ti-Sa crystal is the active laser medium,

which is optically pumped by an external laser (Evolution) and uses chirp pulse

amplification to generate high intensity laser pulses centered at 790 nm. The repetition

rate is set to 1 kHz. The seed pulse coming from the oscillator is first stretched

temporally using a grating scheme and then inserted into a cavity using a pockels cell.

The laser cavity is built in a Z form scheme. After many round trips, this pulse is

amplified and released from the cavity using a second pockels cell and a thin film

polarizer. The overall amplification is about 3.3x106 yielding a power of 1.25 W at this

point. Finally, the pulses are compressed in a similar grating arrangement to the stretcher

and routinely produce pulses of -0.85 mJ centered at 790 nm with pulse widths around

50 fs (FWHM).

(5) Optical Parametric Amplifier (OPA): The optical parametric amplifier system,

OPA-800C, offers broad wavelength coverage from UV to mid IR with near transform

limited output pulses and high pulse energies. The Spitfire amplifier output is split into

two beams (50 %) and used as pumps for two independent OPA systems, providing two

highly stable, inherently synchronized outputs with independent wavelength control. By

proper selection of signal or idler beam, polarization direction, phase matching

angle/type, and number of harmonic crystals, a range from 300 nm up to 5 |tm can be









covered completely. To obtain this wide range of wavelengths, harmonic generation of

the OPA signal or idler is used.

Within the capabilities of the laser system described above, two main time-resolved

experimental setups have been developed in our labs to study the excited state dynamics

of light-harvesting dendrimers. The first one, Fluorescence Upconversion, is a setup that I

designed and built. The second one, Transient Absorption (Pump-Probe), uses a white

light super continuum as the probe. Both of these techniques and the crucial components

of the setup will be explained in detail.

Ultrafast Time-Resolved Emission Spectroscopy

The energy transfer processes within the conjugated dendrimers are investigated

with time-resolved emission spectroscopy. For the initial studies we actually used a time-

correlated-single photon counting (TCSPC) instrument, which was available in the

Schanze lab (specifically for the molecules studied in Chapter 3 and 4) and the lifetime

measurements for the nanostar with TCSPC were already done by Swallen et al.70This

conventional method is widely used for the determination of lifetimes.

Time-Correlated Single Photon Counting

When an ensemble of fluorophores is excited with a very short optical pulse, this

results in an initial occupation of the excited state by No fluorophores. The population of

the excited state will decay radiatively and/or nonradiatively to the ground state

according to the following equation:

dN(t)
=-) (k, + kn)N(t) (2-1)
dt

where kr and knr are the radiative and nonradiative decay rates, respectively. The decay

of the excited state population is exponential which can be written as:









I(t) = I, exp(-t(k, + k,)) = I, exp(-t / fJo) (2-2)

where inluo is the relaxation time of the excited state. Even though this equation shows

only monoexponential decay, for complex systems, the fluorescence decay becomes

multi- or nonexponential. The basic principle of TCSPC experiment is that the

probability of detecting a single photon at time t after pulse excitation is proportional to

the fluorescence intensity at that time.11The time lag between the excitation pulse and

the detected single photon is measured and the decay histogram is reconstructed from

individual time lag measurements. Upon arriving a detector (e.g. PMT), a multichannel

plate, or an avalanche photodiode, each emitted photon creates a reference electrical

pulse that is fed to a constant fraction discriminator that triggers a time-to-amplitude

converter (TAC). Meanwhile, the excited sample emits and when the detector sees the

first photon from the sample, it feeds a stop pulse to the TAC. The TAC consists of a

highly linear ramp voltage generator that is started by one signal and stopped by other,

and delivers an output voltage whose amplitude is directly proportional to the time

difference between the two signals. This TAC signal is then analyzed by an analogue-to-

digital converter and one count is stored in a multichannel analyzer (MCA) for each

voltage. Excitation and detection events are repeated in this way until the histogram of

the number of "counts" against each time window is large enough to give a reliable decay

curve of emission. It is important to note that the emitted fluorescence intensity should be

low enough that the probability of detecting one photon per excitation cycle is less than

unity.

The main advantages of TCSPC method are its high sensitivity and outstanding

dynamic range (signal to noise: 10000/1). However, due to the electronics and the









detector, the best time resolution of the instrument is low (about 50 ps). Fairly long

lifetimes, up to milliseconds, can be measured. This method was initially used to have a

general idea for the emission decay rates of the dendrimers that are mainly in the

nanosecond time-regime. The energy transfer process in PE dendrimers occurs in the

subpicosecond time scale and therefore requires an experiment with a much better time

resolution. This can be achieved with the recently developed Fluorescence Up-conversion

technique.

Fluorescence Upconversion Technique

The fluorescence up-conversion technique is used to measure time resolved

emission dynamics with a time resolution of tens to hundreds of femtoseconds. This

method was first applied by Mahr and Hirsch111 and it is based on sum frequency

generated by the temporal and spatial overlap of the incoherent fluorescence and an

ultrafast gate pulse on a nonlinear crystal.112 (It is also possible to generate a difference

frequency, called down-conversion). This sum frequency is detected as a function of the

time delay between the gate pulse and excitation pulse which induces fluorescence from

the sample. The up-conversion technique allows the mapping of the temporal evolution

of the fluorescence.

The up-conversion signal has a photon frequency given by:

Osum = )gate + Ofliuo (2-3)

implying,


S +- (2-4)
sum gate fluo









As illustrated in Figure 2-2a, when the gate pulse and the emission are overlapped

in the nonlinear crystal, frequency mixing occurs creating an up-converted signal. The

up-converted frequency is determined both by the angle between the optical axis of the

crystal and the incoming beams, and by the optical frequencies of these beams. The

nonlinear crystal behaves as an optical gate which is opened when the gate pulse is

present in the crystal. Scanning the delay of the gate pulse relative to the excitation pulse

opens this optical gate in different portions of time and the fluorescence signal is mapped

out at these different time delays (Figure 2-2b).

The intensity of the sum-frequency signal is given by the convolution of the

fluorescence intensity and the gate pulse intensity:

00
(T) = 0 f If (t)Igate(t -)dt, (2-5)
-00

where c is the time delay between the gate beam and the fluorescence of the sample. The

advantage of using this optical gating technique is that the time resolution is determined

by the width of the pulses (pump and gate pulses), not by the time resolution of the

detection system.113

The time resolution of the upconversion experiment is determined by the

instrument response function (IRF), which is proportional to the cross correlation of the

excitation pulse with the gate pulse. Operationally, the IRF is measured by angle-tuning

the crystal to up-convert transmitted or scattered pump light. For very short pulses <100

fs (FWHM), crystals much thinner than 1 mm are required. The sum frequency is

generated throughout the thickness of the crystal as long as the gate pulse and










(a) (b)
Excitation pulse

flu Luminescence Up-converted
^ .7 signal
Luminescence .. ...."'
...** Csum
Up-converted signal
Gate Pulse
Non-linear
Ogate Crystal Gate pulse





Figure 2-2. Fluorescence Up-Conversion Technique (a) Illustration of the upconversion
principle (b) Up-converted fluorescence signal generated in a nonlinear
crystal only while the delayed gate pulse is present

fluorescence are temporally and spatially overlapped. However, the group velocity

mismatch between the fluorescence and the gate wavelengths can cause broadening of the

IRF which needs to be accounted for the determination of the time resolution of the

experiment. The bandwidth of the up-converted signal depends on crystal properties and

on how tightly the gate pulse and fluorescence are focused on the crystal. The optical

layout for our upconversion setup is explained in detail in the next section.

Homemade Upconversion Apparatus

The experimental setup is shown schematically in Figure 2-3. In this section, we

review the components together before they are separately discussed in detail. The

Ti:sapphire Regenerative amplifier system provides 50 fs, 840 pJ pulses at 790 nm with 1

kHz repetition rate. This beam is split into two with 1:1 ratio. Each 420 pJ beam is

independently used to pump an optical parametric amplifier (OPA 800C, Spectra

Physics). The first OPA delivers the pump pulses in the UV and visible region through

fourth harmonic of signal and idler, respectively. The OPA output is sent through a prism









compressor to get the shortest pulse possible, as required to investigate very fast

relaxation dynamics. The polarization is controlled with a X/2 waveplate, and the beam is

focused onto a sample with a lens (f=200 mm, fused silica). The residual of the 790 nm

pumping the OPA is used as the gate pulse. It passes a delay stage and finally relayed on

the nonlinear crystal by a lens (f=300 mm). This gate pulse is not as short as the

fundamental pulse since it goes through all the optics in the OPA. The autocorrelation

measurements proved that the OPA typically delivers a 120 fs gate pulse at 800 nm and

its bandwidth is narrower than the fundamental 790 nm beam.

The sample is held in a 1mm rotating quartz cell (home made, 1 mm windows) to

ensure sample photostability. The polarization plane of the excitation light is set to magic

angle with respect to that of the gating pulse in order to examine the population dynamics

without the influence of rotational diffusion of the solute molecules on the decay of

fluorescence.

A pair of off-axis parabolic mirrors (A8037-207 Aluminum Uncoated Mirror,

Janos Technology) collects the fluorescence and focuses the fluorescence into a nonlinear

crystal. A negative focal length lens is used to magnify the fluorescence image in the

nonlinear crystal. A type I phase-matching BBO crystal (0.3 mm) is chosen for the

wavelength region studied here. The generated sum frequency light is then collimated

and focused into the entrance slit of a 250 mm monochromator (SpectraMini). A UG11

UV cutoff filter placed in front of the monochromator minimizes the 400 nm generated

on the crystal by second harmonic generation of the gate pulse. A UV sensitive

photomultiplier tube (R7154, Hamamatsu) detects the signal. This electrical signal is

gated by a boxcar average SR 250, Stanford Research Systems. A personal computer is








connected to the detection system and the translational stage to control the experiment. A

Labview program was written to control the translation stage and data acquisition card.


Xgate=800 nm


Xexc





t BO SAMPLE


i PMT 4

Figure 2-3. Fluorescence upconversion experimental setup.

The upconversion technique is relatively simple in principle, and has been used
widely over the past decade. However, it has some crucial components, which happens to

be also the most difficult to align. In Appendix A, I will explain the details of the

experiment for the new users. These are mainly suggestions from an experienced

graduate student who spent a lot of time designing and optimizing the upconversion

setup.









Ultrafast Transient Absorption Spectroscopy

The excited state dynamics of PE dendrimers are also investigated by ultrafast

transient absorption (pump-probe) experiments. The principle of the transient absorption

experiments is rather simple. At least two ultrashort laser pulses are needed. The

intensive one, "pump" pulse, perturbs the sample at t=0. The probe pulse, which is

delayed with respect to pump pulse, crosses the perturbed part of the sample and will

probe the action of the pump pulse on the sample. The perturbation created by the pump

and monitored by probe pulse can be analyzed in two ways: The modifications of the

probe pulse characteristics (intensity, phase, etc.) after passing through the sample can be

compared before and after the action of pump pulse. This measurement is then called the

transient absorption technique. On the other hand, it is quite possible to observe the new

effects created by the probe pulse itself before and after the pump pulse. Raman

spectroscopy, Coherent Anti-Stokes Raman Spectroscopy (CARS), and laser-induced

fluorescence are such experiments.108

Here, the changes in the absorption spectrum of a sample after being perturbed by

an ultrashort pulse will be observed and measured. The absorption of the sample may

increase or decrease or new absorption bands corresponding to new transitions appearing

under perturbation may evolve. By changing the time delay between the pump and probe

beams, it would be possible to do temporal analysis of these changes. For example, the

simplest photophysical event when a molecule is interacting with a light pulse is the

excitation of molecule from its ground electronic state to its first excited electronic state,

followed by the return of the molecules to the ground state by fluorescence and/or

internal conversion. The return of the molecules to the ground state can be monitored by









the change in transmission of a weak probe pulse through the sample as a function of

delay between the pump and probe pulses.

The principle of this method is shown in Figure 2-4. At time t=0, the pump pulse

excites the sample. At time t+At, the probe pulse passes through the perturbed volume of

the sample (At is tunable by means of an optical delay line). The probe pulse intensity is

measured. The spectral distribution of the probe is also recorded to improve the

sensitivity of the measurement simultaneously in the presence and in the absence of the

perturbation of the sample at each laser shot.108 As shown in Figure 2-4, the probe beam

is split into two equal beams; while the first partial beam crosses the pumped part of the

sample, the other beam crosses the unperturbed (not pumped) part of the sample.

Pump Pulse



R = 50%
MI (X, At) c

Probe Pulse D

10 W
Sample

Figure 2-4. Basic principle of transient absorption experiment

These two beams are detected by a CCD. By doing so, shot to shot fluctuations of the

laser power are compensated for.

Experimentally, the efficiency of the light absorption at a wavelength by a medium

is characterized by the absorption or the transmission defined as:









A() = log() = -log T(A)
I() (2-6)
T(A() =
10,(A)


The detector then measures the probe pulse intensity I, (A) (no perturbation by

pump) and I(A, At) (sample perturbed by pump). Considering the Beer-Lambert law, one

can write:


I(A, At) = I, ()x10 -N(At)l (2-7)


Where E; is the absorption coefficient of the sample at wavelength X, N(At) is the

population absorbing at time At at wavelength X, and / is the length of the sample excited.

In fact, the optical density of the sample is measured:

OD(A, At) = log = EN(At)l (2-8)
I(A, At)

The detected signal measured in the transient absorption experiments is actually the

change in absorption (transmission). The detector measures the intensity of the probe

beam in the presence and absence of the pump excitation as a function of time.

t,pump t,nopump
AT T -T 10 1 I
A___ pump nopump 1o 1o t,pump I
T P MT 0 I t, (2 -9)
T )nopump ]t,nopump It,nopump
Io


AT
Then, the change in the absorption is defined as: AA = -log(- +1). (2-10)
T

After excitation of the sample with appropriate wavelength, it is possible to follow

the population dynamics at a given wavelength (single color probe is enough for such






61


measurement) by varying time. For a complex system, it is highly desirable to monitor

the entire transient spectrum at any time delay after the excitation. Measurement of the

full transient spectrum is very helpful in assignment of different absorbing units

(species).

For the interpretation of the transient spectra and their relative sign and values, it

must be recalled that there are three origins for the pump-probe signal: ground state

bleaching, stimulated emission, and excited state absorption. When the sample is pumped

within its absorption spectrum, a certain number of molecules will be excited (Figure 2-

5). During the probing process if there is population in the excited state, the transmission

of the sample increases and ground state bleaching is observed (negative absorption

signal). Stimulated emission occurs when the probe beam stimulates the excited state

molecules to return to the ground state. The detector will see more photons in the

emission range of the sample, so the transmission increases resulting in a negative AA

signal. Note that the probability of the stimulated absorption is the same as of the


Sn




S ..At..
hvi hv2 hv2

So

Excitation Ground State Stimulated Photoinduced
Bleach Emission Absorption
AA < 0 AA < 0 AA > 0

Figure 2-5. The theoretical scheme of certain signals observed as transient absorption
signals.









stimulated emission at the emission wavelengths. Excited state absorption (photoinduced

absorption) takes place when the excited molecules are excited to even higher electronic

states by the probe pulse. This will lead to positive AA signals. It is highly probable that

these three different signals will spectrally overlap complicating the interpretation of the

overall signal.

Probe Characteristics and White Light Continuum Generation

The appropriate excitation (pump) wavelength, which depends on the absorption

spectrum of the investigated system, can be generated by an OPA. The probe beam can

be monochromatic or it can have a broad spectrum. In the simplest case, namely one-

color (degenerate) experiments, the pump and probe pulses are split from the same initial

beam and one of them is delayed with respect to the other. For two-color experiments

another light source is needed to generate the required wavelength of the probe pulse.

Prior to choosing the probe pulse wavelength, some preliminary studies, such as steady

state spectroscopy, should have been used to characterize the sample. The probe

wavelength should be in a spectral domain where it is expected that some species created

by the excitation will be present in the sample at a certain time At, and will have resonant

electronic transitions for that specific probe wavelength. However, it is highly

informative to use a probe spectrally as broad as possible. The whole transient absorption

spectrum can be measured if a multicolor probe, such as white light continuum, is used.

Measuring the whole transient spectrum as a function of time will help determine the

dynamics of the excited states of the system. 114118 For the complex systems studied here,

it was preferred to use a white light continuum as the probe pulse. How we generated this

continuum will be explained briefly, but first the experimental setup will be explained.









Experimental Setup

The experimental transient absorption setup for probing with a white light

continuum is presented in Figure 2-6. The Ti:sapphire Regenerative amplifier system

provides 50 fs, 840 ptJ pulses at 800 nm with 1 kHz repetition rate. This beam is split into

two with 1:1 ratio. Both 420 ptJ beams are used to pump an optical parametric amplifier

(OPA 800C, Spectra Physics). The second OPA is used to deliver the pump pulses for

transient absorption experiment. The OPA is tunable in the UV and visible region (300-

900 nm). Since the dendrimer systems under investigation absorb in the UV region (<400

nm), the OPA was aligned to generate UV pulses. A double-pass prism pair is used to

compress these pulses. A motorized translation stage (Model No:M-415 D6, Physik

Intrumente) is used to vary the time delay between the pump and probe pulse up to 1 ns.

A chopper wheel is used to chop the pump beam with 10 Hz frequency in order to

compare signal with and without pump.

Delay line
pump



Shopper
Spitfire
800 nm Prism compressor


n nu is gPolarizer C
OPA
probe
CaF2 plate %
Sample


Figure 2-6. Experimental setup for transient absorption experiment probing with white
light continuum. UV pump pulses are obtained from the OPA. A chopper
wheel used to compare the signal with and without pump. The white light
continuum is generated in CaF2 window.









Continuum generation

When a high peak power short pulse is focused in a (transparent) medium such as

glass, water, or sapphire, a continuum of light, which may even appear as white light, is

generated. The origin of this process is mainly governed by self-phase-modulation and

stimulated Raman emission. The directionality of the white-light pulse makes it possible

to use it as a spectrally broad probe and measure the transient absorption at different

wavelengths simultaneously.

A small fraction, 3-4 jiJ, of the fundamental output of the amplified laser pulse is

converted to white light by focusing it into a 1mm thick CaF2 window (1" diameter, PW-

1004-CFUV from CVI-laser). The CaF2 is held in a home-made rotating stage to avoid

shot to shot intensity fluctuations, temperature effects. A 10 mm lens was used, and the

size of the focus was about 200 |tm. The spectrum needed as a probe for the

measurements on dendrimers lies in the region of 300-600 nm (absorption and emission

of the system). Figure 2-7, the top curve, shows the spectrum of the white light

continuum directly (without the optics necessary for the experiment) sent to the detector

(CCD). As shown in Figure 2-7, the probe and reference beams have very similar

spectrum. The quantity of photons (signal on the detector) and the noise evaluated by the

ratio of probe/reference with and without the excitation will determine the quality of the

probe pulse. Below 365 nm, the intensity of the spectrum decreases rapidly, but for a

good signal quality, it is necessary to produce a spectral distribution as flat as possible.

The super continuum shown in Figure 2-7 is flattened spectrally by a home made filter.

This filter consists of a fused silica cuvette with 1.25 mm thick windows, filled with a

mixture of dyes and polymers dissolved in dichloromethane. The total thickness of the










cuvette is 4.5 mm. Our efforts proved that the quality of the supercontinuum, its stability

and spectral smoothness, depends critically on the pump pulse energy, pump beam

diameter, and focusing parameters of the lens. One has to play with the distance of the

lens with respect to CaF2 plate (distances slightly shorter or larger than focal length) and

the pump energy to obtain the optimum, spectrally broad, and smooth supercontinuum.

Before using a CaF2 window, we tried a sapphire plate (which is used in OPAs as a

supercontinuum medium). Even though the stability was good enough, spectrally there

was no light below 350 nm. Thus, we searched for a medium which will generate enough

light to get some signal down to 300 nm. Our investigations suggest that a white light

continuum generated in CaF2 provides significantly more seed photons for shorter

wavelengths (X< 500 nm) than the white light continuum generated in sapphire (both

pumped at 800 nm). The only disadvantage is the instability of CaF2 material upon




25000

20000

t 15000
0 [reference
probe
10000

5000

0

300 350 400 450 500 550
Wavelength(nm)


Figure 2-7. Spectrum of the white light continuum generated by CaF2. The probe (blue)
and reference (red) beams used for transient absorption experiment.









focusing a short pump pulse due to its very low damage threshold.119 Fortunately, by

rotating the CaF2 window, the problem was solved enabling the use of such white light

continuum as probing beam.

The fundamental 800 nm beam used to generate white light continuum is passed

through a half wave plate and a thin film polarizer is oriented at 45 degrees with respect

to the pump pulse (excitation pulse for the sample) before it reaches to the CaF2 medium.

By doing so, the pump energy for the continuum generation is controlled and optimized.

At the same time, the CaF2 window is positioned at the focus of an off-axis parabolic

mirror, which collects and collimates the white light continuum. The continuum is split

into two equal beams: probe and reference beam*. After passing through the sample, the

probe and reference beams are split into two equal portions independently (total four

continuum beams). These beams go through four Glan-Taylor polarizers aligned

perpendicular and parallel with respect to the pump pulse, allowing simultaneous

detection of both polarizations. In order to obtain the dynamics free from reorientation

effects, the magic angle signal is calculated from parallel and perpendicular signals:

A, +2A
A magc (2-11)
3

The temporal evolution of anisotropy can also be obtained by evaluation of:


r(t) =--A (t)-AA(t) (2-12)
AA// (t) + 2AA (t)





Many beam splitters are not good enough for UV light down to 300 nm. Since the continuum has much
less photons below 350 nm, it is vital to use the best optics for UV. I used reflective neutral density filters
with OD 0.5 ( 33.3% absorption, 33.3. % transmission, 33.3 % reflection) to split probe and reference
beams.









where AA// (t) and AA (t) correspond to the transient absorption of the polarization

oriented parallel or perpendicular to the excitation beam polarization, respectively.

The four tracks of white light are dispersed with an imaging grating

monochromator (focal length 30.3 cm, 300 lines/mm). The iStar intensified CCD camera

(iStar 720DH-720-25F-03, Andor Technology) is employed as the detection system. Note

that, not all the probe wavelengths propagate through the optical components at the same

speed (including the transparent medium used for continuum generation). Since the

c
velocity, v(A) = at which a given frequency will travel depends on the refraction
n(A)

index of that material at that wavelength. When using the white light continuum, it is

important to account for the influence of group velocity dispersion either experimentally

or numerically. If the dispersion for the continuum is well characterized, the spectra can

be corrected numerically by using a home-written Labview program.

With the setup explained previously, data can be acquired in the following two

ways: either time resolved measurements (AA versus At plots) at a fixed probe

wavelength or spectrally resolved measurements corresponding to a fixed delay between

pump and probe (AA versus wavelength region of the probe).

During our measurements, the signal appears earlier at shorter wavelengths than at

a longer wavelength, which is a consequence of the temporal distribution of the different

spectral components of the supercontinuum probe pulse, called "chirp". The chirp of the

white light continuum is determined from the delay between the signals probed at 330 nm

and 560 nm, for the measurements on the dendrimers. As shown in Figure 2-8a, the chirp

leads to a delay of -520 fs between the spectral components at 350 and 575 nm. This

chirp is corrected for the time-resolved measurements analyzed in the following










chapters. Figure 2-8b also shows the signal for the DCM probed at different wavelengths.

The delay here is around 570 fs. The FWHM of the Raman-like (or coherent artifacts)

signal generated in such solvents (when pumped with couple of micro joules) also gives

information about the time resolution of the experimental setup.

The chirp correction is done after the experimental measurements. Each data set is

analyzed to get an initial estimation of the chirp and then using a labview program based

on calculating dispersion of light in every refracting medium, the chirp corrected data sets

are produced. This numerical correction employs the Sellmeier equation, which is an

empirical relationship between the refractive index n and wavelength X for a particular

0.020 .. ........I ''


(a) 2G2mOH dendrimer
po560 nm
0probe=560 nm
0.015

520 fs
0.010


0.005


00000


0.015 -(b) pure DCM
probe=330 nm

0.010 probe560 nm

0
0.005 -



0.000


-0.25 0.00 0.25 0.50 0.75 1.00 1.25 -0.25 0.00 0.25 0.50 0.75 1.00 1.25
Time(ps) Time(ps)

Figure 2-8. The chirp of the white light continuum determined from the delay between
the signals (a) probed at 330 (squares) and 560 nm (circles) for the
measurements of 2G2mOH dendrimer (b) pure dichloromethane.

transparent medium. The common form of the equation for glasses is:


2(,) 1 B,22 B2A2 B3A
n2( Z2= + + + (2-13
A2 C 2 C2 2 C3


where B1,2,3 and C1,2,3 are experimentally determined Sellmeier coefficients. Different

forms of the equation are used for certain type of materials, such as crystals and common


)









organic solvents.120'121 Chirp correction is necessary and should be done very carefully

since most of the dynamics in PE dendrimers takes place in a subpicosecond time scale.

Time resolution of the experiment

In general, to determine the time resolution of an experiment where two laser

pulses are used, difference or sum frequency mixing between those pulses (pump and

probe) is performed at the place of the sample using a very thin (50-300 utm thick) type I

BBO crystal. On the other hand, in the experiments where the white light continuum is

used as the probe, the time resolution can be determined from the coherent artifact

(Raman Scatter) observed in some solvents. Many basic molecular liquids and optical

solids are transparent in the visible and near-UV spectral range for low intensity radiation

(< 1010 W/m2). However, when high power ultrashort laser pulses are applied, these

media can absorb efficiently through a multiphoton absorption mechanism (two photon

absorption) and Stimulated Raman Amplification (SRA).122 Furthermore, a spectrally

broad probe pulse, similar to white light continuum, favors efficient cross-phase

modulation. Each of the signals is produced by the simultaneous action of two photons,

one from the pump and the other from the probe. These artifact signals will terminate

rapidly following excitation, thus their duration is directly related to the temporal width

of the pump-probe cross-correlation function. This is verified by comparing duration of

the coherent artifact at a specific probe wavelength with the cross-correlation measured

using a single color probe mixed with the pump pulse in the BBO crystal. Simultaneous

absorption of a pump photon and a probe photon gives rise to two-photon absorption,

while the interchange of photons between pump and probe through a material's

vibrational energy level gives rise to SRA. Moreover, the cross-phase modulation leads to










spectral modifications within the probe upon pump-induced temporal changes of the

refractive index.122,123

These coherent effects are observed in many common solvents, such as,

acetonitrile, methanol, water, ethanol, cyclohexane, and dichloromethane with different

strength. Figure 2-9 presents typical coherent artifact signals in hexane, commonly used

to measure the instrument response function. They are shown at two different

wavelengths of the probe and it is obvious that the coherent artifact is getting smaller

with increasing probe wavelength. The temporal width is also dependent on the probe

wavelength, due to group velocity mismatch. As the probe wavelength is more distant

from the pump wavelength, cross-correlation signals gets temporally broader. Such

behaviors are in agreement with literature results.122


0.14
X =310 nm
pump
0.12
0.12 e =330 nm
probe
0.10 CC= 110 fs

0.08

0 0.06

0.04 k- jprobe=380 nm

0.02 CC=180 fs

0.00

-0.02 I I I I I *
-0.25 0.00 0.25 0.50 0.75 1.00
Time(ps)
Figure 2-9. Coherent artifact of hexane solvent excited at 310 nm, probed at 330 and 380
nm.

Observing such coherent artifacts from the solvents will provoke the question, "do

we have solvent contribution to the signal?" It is essential to check the solvent response

at the same pump energy where the experiments are performed. In our experiments, these









measurements confirmed that there is no coherent artifact from the solvent at the pump

power intensities used for excitation of the real dendrimer samples. Otherwise, the signal

should be corrected for solvent contributions.

Concentration and Pump Pulse Energy Dependence

For both fluorescence upconversion and transient absorption experiments, the

optical density of samples is approximately 0.2-0.3 mm -, yielding a concentration range

of 10-5- 10-6 M. All spectroscopic measurements of dendrimers at room temperature are

carried out in dichloromethane (DCM). Low temperature experiments were performed in

glass forming methyl-tetrahydrofuran (Me-THF). It has been shown that

phenyleethynylene dendrimers do not aggregate in dichloromethane despite some

aggregation in solvents like isopentane (lower dielectric constant). The solvent should be

anhydrous so that the dendrimer molecules can stay stable in solution for a longer period

of time. The polymer experiments are performed in methanol and the samples were

prepared at a concentration where steady state measurements did not show any

aggregation.

Same measurements may differ from one another due to different conditions of the

day that the experiment is carried out. The excitation energy, beam diameter and even the

concentration of the sample might be different on a particular day. For a reliable

comparison, it is important to know if and how the pump energy influences the signal.

The main criterion is to check the linear region of the pump energy. As long as the

molecule is excited with power in the linear regime, the shape of the signal will not

change. The optimum pump energy can be determined by varying the pump pulse energy

until the shape of the signal changes. For each measurement, one should calculate the

number of photons absorbed per molecule. The most common problem with high energy









pulses is annihilation, which may occur when more than one photon is absorbed by a

molecule. In such a case, one excited state can act as a quencher for other excitations,

resulting in additional decay components in the process. Thus, to detect the dynamics

associated with the studied process, the excitation intensity must be adjusted in a way that

the average number of photons absorbed per molecule should be less than unity. Another

scenario is the local heating of the sample leading to sample defects due to high

excitation densities which can increase the proportion of the radiationless processes. The

power densities for each measurement are given in the following chapters.

The photostability of the samples is another important issue. The dendrimers are

not as stable as the polymer samples. Depending on the pump power, they would

photobleach much faster. Rotating sample cells are used for time-resolved experiments

to use the minimum sample for the maximum scan time. Absorption spectra of the

investigated molecules were checked before and after each laser measurement which

proves the photo stability of all compounds under the experimental conditions used in

this work. All time-resolved experiments presented in this work are performed at room-

temperature.














CHAPTER 3
ENERGY TRANSFER IN GENERATION 1 UNSYMMETRICAL PHENYLENE
ETHYNYLENE DENDRIMERS

The search for artificial light-harvesters has led to intense studies of conjugated

dendritic macromolecules.36'81'124 Dendrimers have potential applications in photonic

devices due to their highly branched architectures and unique physical properties. With

recent advances in synthetic methods,2'4'125 the size, topology, flexibility, and surface

chemistry of dendrimers can be controlled at the molecular level with high precision.

Accurate positioning of chromophores at the core, periphery, focal point, or even at each

branching point of the dendrimer can now be achieved.52'59'126 For photonic applications,

the dendritic architecture creates large transition dipoles due to the high number of

chromophore units.

Some judiciously designed phenylene ethynylene (PE) based dendrimers show

highly efficient and unidirectional energy-transfer properties.69'72 Their topology

suggests applications as scaffolds for light-harvesting devices. In addition, the large

number of chromophore units lead to the formation of excitonic bands. These

dendrimers' photophysical properties cannot be understood as simply additive properties

associated with molecular orbitals on single chromophore units. In this scenario, it is

essential to understand the electronic coupling,84 exciton formation,72'127 and energy

transfer in detail.

A variety of dendritic architectures have now been synthesized, leading to unique

photophysical properties. PE units coupled exclusively through meta orpara substitution









on a phenyl ring lead to either compact or extended dendrimers.72 These two families of

dendrimers differ in the number of PE units between consecutive branching points.

Compact dendrimers have a fixed-length linear unit, while extended dendrimers have a

variable number of linear PE units depending on the number of branching points between

the unit and the core. Both families of dendrimers have been investigated

theoretically19'78'127'128 and experimentally.69,75,129

In compact dendrimers, steady state experiments performed by Moore and co-

workers show that the optical excitation is localized on the PE units.72 The excitonic

localization on individual PE units is evident by the monotonic increase in absorption

intensity and the lack of spectral shift with generation number.81 The extended series

exhibit exciton localization on PE units of increasing length (2-,3-, 4- ring), which creates

an energy funnel yielding multistep energy transfer.

Mukamel and co-workers have studied compact and extended dendrimers using a

Frenkel Exciton Hamiltonian.127 Applying the collective electronic oscillator model, they

concluded that the electron-hole pair was confined to the linear segments between

branching points. The bound Frenkel excitons are free to migrate throughout the

molecule. Depending on the strength of the coupling, the migration leads to coherent or

incoherent energy transfer. Absorption spectra calculated from this model are in excellent

agreement with experiment.

Using time correlated single photon counting, Swallen et all1130 studied an extended

PE dendrimer and found an instrument-limited value of about 10 ps for the energy

transfer from the lowest energy chromophore in the backbone to a phenylene ethynylene

perylene trap. Subsequent experiments by Kleiman et al investigated the same extended









dendrimer using femtosecond degenerate pump-probe spectroscopy and revealed a

stepwise energy transfer from the shorter PE units to the longer PE chains.75 These

experiments indicated that some of the steps in the energy transfer occur on a

subpicosecond time scale.

The novel characteristics of PE dendrimers arise from the electronic properties at

the branching points. In both compact and extended dendrimers, meta substitutions on the

phenyl rings result in broken 7t-electron conjugation in the ground electronic state. The

situation in the excited state is less clear. A recent study based on a di-ethynylene phenyl

unit with H or phenyl substituents shows dramatic changes in both the emission spectra

and the radiative lifetimes.90 The electronic structure calculations indicate that the phenyl

ethynylene (H or phenyl substituted) units are strongly coupled in a relaxed geometry on

the excited state.84 Experiments in larger dendrimers do not show the shifts predicted in

the smaller systems (see ref 81, Figure 4). The extent of localization in the excited state

for sizable dendrimers remains an open question.

Unsymmetrical architectures in which coupling among the PE unit occurs through

para and ortho substitutions have been synthesized by Peng and co-workers96'97'131

(Figure 3-1). In these dendrimers, the substitutions create combinations of PE units of

variable lengths, analogous to those encountered in extended dendrimers.97

Unsymmetrical branching leads to rapidly growing conjugation lengths as the generation

number increases, providing a broad absorption spectral range with large molar

absorptivities and high fluorescent quantum yields.95 Linear conjugated segments

connecting the periphery to the core suggest faster and more efficient energy transfer to

the core.95 Furthermore, the presence of ortho substitutions may allow stronger coupling









of PE units, leading to more delocalized excitation throughout the entire molecule. For

these architectures, confined Frenkel excitons127 might extend over regions of the

molecule that includes substitutions at ortho positions.

Symmetric dendrimers' absorption spectra can be interpreted as the addition of

building blocks, defined by the confined excitons.127 The role of ortho substitution and

exciton confinement in unsymmetric dendrimers is not as well characterized and there are

still unanswered questions: Are the absorption band structures associated with exciton

localization and "building blocks"? Does coherent or incoherent energy transfer occur

between those localized states?

The goal of our study is to understand the exciton size and the rates of energy

migration. We focus here on the characterization of intramolecular interactions and the

energy-transfer mechanisms in unsymmetrical PE dendritic molecules. To investigate the

extent of delocalization within the dendritic structure, we consider unsymmetrical

monodendrons with multiple ortho and para substitutions. Energy transfer mechanisms

are monitored by adding an ethynylene perylene trap (EPer), which acts as reporter for

energy transfer.

Time-resolved photoluminescence experiments in the subpicosecond time scale are

employed to follow the energy initially deposited in the dendrimer's backbone by an

ultrafast pulse. A kinetic model is proposed to interpret the rise times of the fluorescence

measured in unsymmetrical dendritic structures with and without an energy trap. Finally,

we present an analysis of the validity of Forster model by comparing the model

predictions with our experimental results.









Materials and Methods

The synthesis of unsymmetrical dendrons is described elsewhere.96 Unsymmetrical

monodendrons can be covalently bonded to form larger and more symmetrical

macromolecules named as di- or tri-dendrons. Here, two G1 (generation 1) monodendrons

are coupled to a phenyl ring in the meta positions. When the phenyl ring has an additional

OH or ethynylene perylene group in the other meta position, the molecule is named 2G1-

m-OH or 2G1-m-Per, respectively (Figure 3-1). The ethynylene perylene unit acts as an

energy trap and is utilized to probe energy transfer dynamics. As any PE dendrimers, the

molecules under investigation are quite rigid molecules that do not allow for the

backfolding of any of the branches.






OHOCH OCHH




OH



Figure 3-1. Chemical structures of generation 1 phenylene ethynylene dendrimers: (a)
2Gi-m-OH (b) 2G1-m-Per.

For spectroscopic measurements, all samples are prepared in dry CH2C12 without

further purification. Absorption spectra are recorded on a Varian Cary 100

spectrophotometer. The fluorescence spectra are measured with a Jobin-Yvon instrument

(Spex-Fluorolog-3).The optical density of samples used in all measurements is about 0.3

mm- which provides a concentration less than 10-6 M to avoid any aggregation and

excimer formation. All steady state measurements and transient absorption experiments









are performed in a 2 mm optical path length quartz cuvvettes. In addition, a homemade

rotating cell with optical path of 1 mm (for optimum time-resolution) is used for time-

resolved emission experiments. The laser system, fluorescence upconversion, and

transient absorption setup are described in detail in Chapter 2.

The upconversion experiment measures the temporal evolution of the fluorescence

with subpicosecond resolution. It is based on the sum-frequency mixing of the molecules'

emission with an ultrafast gate pulse in a nonlinear crystal. Briefly, excitation pulses are

derived from an optical parametric amplifier (OPA), pumped by a commercial Ti-

Sapphire laser system consisting of a Ti-Sa oscillator (Tsunami, Spectra-Physics) and

subsequent amplifier (Spitfire, Spectra-Physics) with a repetition rate of 1 kHz. The

fourth harmonic of the OPA output signal or idler is used to generate tunable excitation

pulses in the 315 nm to 370 nm spectral region. Pump pulses of -40 nJ with a beam

diameter of 200 |tm are used to maintain a linear optical response. After all the optics in

the OPA and harmonic generation processes, the UV and visible pulses accumulate group

velocity dispersion, yielding longer excitation pulses and poor experimental time-

resolution. To overcome this pulse lengthening we use a pair of quartz prisms to

compensate the chirp.

The homemade rotating cell has a 1 inch diameter and an optical path length of 1

mm to guarantee excitation of a new sample volume with every laser shot with minimum

consumption of sample. The photoluminescence is collected by two off-axis parabolic

mirrors and the excitation volume is imaged onto a 300 |tm P3-BBO crystal. Usually a

small portion of the regenerative amplifier beam (-30 [PJ/pulse, FWHM = 60 fs) is

weakly focused (50 cm focal length) and the beam diameter at the crystal is kept larger









than the imaged fluorescence. However, some modifications are necessary for

experimentally challenging fluorescence wavelengths. While collecting fluorescence of

400 nm, a lot of background signal is introduced because of the sum frequency

generation of 800 nm (gate beam) and its spontaneous second harmonic generation (400

nm) at the BBO crystal. With such level of background signal, it is impossible to

distinguish the upconverted signal originated from the molecules studied here. Therefore,

the gate beam has to be different from 800 nm. We modified the experimental setup to

overcome this experimental limitation.The second harmonic of the OPA signal (630 nm

and 740 nm) is used as the gate beam. However, it is still difficult to get good

signal/noise with these gate beams since the intensity of SHG of the OPA signal is an

order of magnitude smaller than the intensity of the 800 nm beam. The samples were

excited at 315 and 370 nm using the fourth harmonic of the signal generated in the OPA.

Spatially and temporally overlapped gate and collected fluorescence in the 13-BBO crystal

generates a nonlinear response signal in the UV. Colored filters (UG11) are used to

remove scattered light from the excitation pulse and the second harmonic of the gate,

which is also generated at the crystal. The upconverted beam is dispersed by a 0.25 m

monochromator and detected with a PMT. Boxcar integration and averaging of 104

pulses per time step leads to a signal to noise ratio of about 50:1.

The time resolution of the setup was measured by detection of cross-correlation of

scattered light from solvent and gate pulse. For UV excitation, the time resolution was

determined to be about 225 fs.

In the transient absorption experiments as sketched in Chapter 2, a fraction of the

Ti:Sa amplifier output is focused on a 1 mm CaF2 plate to generate a white-light









continuum, which is used as probe and reference beams. Using a thin-film polarizer, the

probe light polarization is oriented at 45 degrees with respect to the pump pulse. After

passing through the sample, a Glan-Taylor polarizer splits the probe beam into its

polarization components, parallel and perpendicular with respect to the pump, allowing

for the simultaneous detection of both polarizations. Pump induced absorption changes of

both probe polarization components are measured as a function of pump-probe time delay

by modulation of the pump beam with a mechanical chopper and detection of the probe

beams and a reference beam with the pump on and the pump off (to overcome shot-to-

shot fluctuations) using a CCD camera equipped with a 30 cm spectrograph. For all

transient absorption measurements performed here, the excitation wavelength is 315 nm.

By measuring the coherent artifacts from the pure solvent (i.e Stimulated Raman

Amplification, discussed in Chapter 2), the time resolution of this setup was determined

to be about 150 fs.

Data analysis involves the convolution of decay and rise time functions with the

corresponding experimental IRF for each experiment. The integrity of the sample was

checked before and after each set of measurements.

Steady State Spectroscopy

The steady state absorption spectra of 2Gi-m-OH and related monodendrons are

shown in Figure 3-2. Any 2Gn- didendron is composed of two Gn monodendrons coupled

through the meta positions of a phenyl ring. The branching center determines the strength

of interactions among chromophores, which plays a key role in determining the

mechanism of energy transfer. In Figure 3-2, the absorption spectrum of 2Gi-m-OH,

G1OH, and G20H are compared.











1.0 -----G-OH
EMS G2-OH
D 0.8 -
0

o 0.6 ,---,


0.4 ,,


z0.2


0.0
250 300 350 400 450 500
Wavelength (nm)

Figure 3-2. Normalized absorption spectra of 2G1-m-OH, G1-OH, and G2-OH
in dichloromethane. Normalized emission spectrum of- 2Gi-m-OH.

The absorption spectrum of 2Gi-m-OH shows a 35 nm red shift compared to single

G1OH dendron.131 This is due to one additional PE unit, which increases the total

conjugation for the system. The absorption spectrum of G20H (Figure 1-6) shows more

similar features, but there is a 20 nm redshift. The longest linear PE chain in G20H has

the same number of PE units as the longest linear chain in the 2G1-m-OH. This red shift

is due to more extended conjugation between the two longest linear PE units through the

ortho linkage in G20H. Also note that the red shift here is much more pronounced

compared to 2G2-m-OH versus G30H dendrimers (Figure 4-2).

The trap molecule, Eper, is substituted to the didendron molecule in the meta

position with respect to both monodendron components. The absorption spectrum of 2G1-

m-Per is a superposition of absorption spectra of 2Gi-m-OH and EPer (Figure 3-3). The

broad absoprtion feature between 300 and 400 nm corresponds to the dendritic backbone









while absoprtion at X> 400 nm is only associated with the Eper trap. Since the presence

of both donor and acceptor groups within the same molecule does not lead to appearance

of a new band or to differences in the ground state absorption spectrum, it can be

concluded that there are no strong interactions between donor and acceptor moieties in

the ground state.




1.00-



c 0.75 -

0

S0.50



z 0.25-



0.00
300 350 400 450 500 550 600
Wavelength(nm)

Figure 3-3. Normalized absorption spectra of-----2Gl-m-OH, ...EPer, and -2Gl-m-Per
and fluorescence spectrum of -2Gl-m-Per, excited at 315 nm.

The fluorescence of PE dendrimers attached to Eper trap is completely quenched

compared to the ones without trap. In fact, after excitation at X= 315 nm, the emission

arises entirely from the EPer unit. This is again a strong indication that within these

dendrimers, the excitation energy is efficiently transferred from the dendrimer backbone

to the EPer chromophore. Comparison of absorption and excitation spectra indicates

-96% efficiency for the energy transfer process.92 Interestingly, at 400 nm, a small band

with intensity contributions from unfunctionalized 2Gi-m-OH and possibly residual









backbone emission from 2G1-m-Per is noticed in the emission spectrum. Time-resolved

data will clarify its origin.

Time-Resolved Emission Experiments

The fluorescence decays are measured by TCSPC and 2Gi-m-OH decays with a 1.8

ns time scale, whereas 2G1-m-Per decays with the emission lifetime of EPer (2.2 ns).

Time resolved fluorescence upconversion technique is applied to measure the rise

times associated with these decays. First, the dendrimer without EPer trap is studied to

understand the extent of intramolecular interactions within the didendron backbone. The

absorption spectrum of 2Gi-m-OH has two distinguishable bands peaked at 305 and 365

nm. Since one of the goals of this study is to explore the possibility of assigning the

absorption band structure to exciton localization, it is reasonable to excite 2Gi-m-OH at

selective wavelengths with significant contributions from each band. Therefore, the

excitation wavelengths are chosen to be 315 and 370 nm. The emission is detected at

backbone fluorescence of 400 nm. Note that the instrument response functions of 180-

220 fs were used in the analysis of all measurements for deconvolution of data sets (IRF

was routinely recorded during each measurement session).

Figure 3-4 shows no detectable excitation wavelength dependence for the

subpicosecond risetime. Convolution of the IRF with the exponential rise function yields

a 300 fs time constant for both excitation wavelengths, which suggests the delocalization

of the initially excited state throughout the monodendrons. This risetime is definitely

longer than the experimental time-resolution, meaning that it essentially takes 300 fs for

the initially deposited energy to reach the lowest lying emitting state. It is thus suggested

that absorbing and emitting states of the dendrimer are two different excited states.132 A

second component with a decay time constant of 6 ps is found in addition to the long









decay time of 2Gi-m-OH (1.8 ns). This kinetic component can be attributed to vibrational

relaxation in the excited state of the monodendrons, which is coupled to relaxation and

reorganization of the solvation shell around the monodendrons.133 The solvent molecules

have to accommodate for the newly populated Si state of the whole molecule. The

fluorescence is detected at the blue end of the emission at 400 nm (Figure 3-2). At

fluorescence detection wavelengths close to the excitation, the vibrational relaxation will

be observed as a fast decay component, whereas at longer wavelengths this decay would

be seen as a risetime, since the fluorescence is detected from a state that has to be

populated with this time constant. Moreover, the relative amplitude associated with 6 ps

component will depend on the excitation wavelength. The excitation wavelength

dependence of the 6 ps component will be explored with the kinetic analysis. If it is

attributed to vibrational relaxation, its amplitude will decrease at longer excitation

wavelengths.

The objective of investigating the 2Gi-m-OH molecule was to understand the

excitation energy transfer dynamics among the similar dendrons before the energy is

transferred to a core trap. We investigate also the same dendritic structure with an EPer

trap attached in meta position to the core phenyl unit. The risetime of the Eper emission

is experimentally measured to follow the excitation energy migration from the initially

excited state on the didendron to the final trap. Figure 3-5 shows the temporal evolution

of the 2G1-m-Per fluorescence as a function of excitation wavelengths. Fluorescence is

detected at the maximum emission wavelength of Eper (480 nm). Direct excitation of

EPer provides the time resolution limit for fluorescence risetimes. Any risetime longer

than 150 fs can be assigned to excited state dynamics.