Spectroscopic Studies of Singlet Fission and Triplet Excited States in Assemblies of Conjugated Organic Molecules

UNIVERSITY OF CALIFONIA, SAN DIEGO

Spectroscopic studies of singlet fission and triplet excited states in assemblies of conjugated organic molecules

A dissertation submitted in partial satisfaction of the requirements for the degree Doctor of Philosophy

Chemistry

Chen Wang

Committee in Charge:

Professor Michael Tauber, Chair Professor Michael Galperin Professor Clifford Kubiak Professor Lu Sham Professor Amitabha Sinha

2014

The dissertation of Chen Wang is approved and it is acceptable in quality and form for publication on microfilm and electronically:

Chair

University of California, San Diego

2014

iii

DEDICATION

This dissertation is dedicated to my parents Xunsheng Wang and Jinghua Chen

TABLE OF CONTENTS

Signature Page ...... iii

Dedication ...... iv

Table of Contents ...... v

List of Figures ...... viii

List of Schemes ...... xiii

List of Tables...... xvi

Acknowledgements ...... xv

Vita ...... xvii

Abstract of the Dissertation...... xix

Chapter 1 Introduction ...... 1

Chapter 2 Characterization of Carotenoid Aggregates by Steady-State Optical Spectroscopy ...... 24

2.1 Introduction ...... 24

2.2 Experimental ...... 26

2.3 Results ...... 33

2.4 Discussion ...... 47

2.5 Conclusions ...... 64

Appendix ...... 75

Chapter 3 High Yield Singlet Fission in a Zeaxanthin Aggregate Observed by Picosecond Time-Resolved Resonance Raman Spectroscopy ...... 92

3.1 Introduction ...... 92

3.2 Result and discussion ...... 93

3.3 Conclusions ...... 101

Appendix ...... 106

Chapter 4 Triplet Exciton of Carotenoids Formed by Singlet Fission in a Membrane ... 123

4.1 Introduction ...... 123

4.2 Experimental ...... 124

4.3 Results and discussion ...... 125

4.4 Conclusions ...... 132

Chapter 5 Singlet fission dynamics of carotenoid aggregates: insights from transient absorption spectroscopy ...... 138

5.1 Introduction ...... 138

5.2 Experimental ...... 139

5.3 Results ...... 144

5.3.1 Fs-ns-TA spectroscopy of zeaxanthin aggregates ...... 144

5.3.2 Ns-TA spectroscopy of zeaxanthin aggregates ...... 150

5.4 Discussion ...... 152

5.5 Conclusions ...... 159

Chapter 6 Singlet Fission Process in Zeaxanthin Aggregates: Insights from Transient Resonance Raman and Absorption Spectroscopy ...... 166

6.1 Introduction ...... 166

6.2 Materials and methods ...... 169

6.2.1 Preparation of samples ...... 169

6.2.2 Picosecond time-resolved Raman spectroscopy (TRRR) ...... 170

6.2.3Transient microscopic resonance Raman spectroscopy ...... 174

6.3 Results ...... 175

6.3.1 The ground state absorption and terminology ...... 175

6.3.2 Transient absorption results and the selection of TRRR probe wavelengths ...... 176

6.3.3 TRRR Stokes spectra ...... 178

6.3.4 Reconstructed excited state absorption (ESA) and Raman excitation profile (REP) ...... 187

6.3.5 TRRR anti-Stokes spectroscopy ...... 190

6.4 Discussion ...... 192

6.4.1 Exciton coupling of ground state transition in aggregates ...... 192

6.4.2 The excited states in aggregates ...... 195

6.4.3 Singlet fission in different aggregates...... 199

6.4.4 The influence of exciton coupling on excited state spectra ...... 201

6.4.5 Vibrational relaxation in singlet fission and TT annihilation ...... 206

Appendix ...... 221

Chapter 7 Resonance Raman Spectroscopy of the Triplet Excited State of Oligothiophenes...... 233

Appendix ...... 252

vii

LIST OF FIGURES

Figure 1.1 The spin diagram of singlet and triplet state ...... 1

Figure 1.2 Electronic state transitions of zeaxanthin monomer ...... 9

Figure 2.1. TEM images of zeaxanthin J1, J2 and H-aggregates ...... 34

Figure 2.2. UV-vis absorption spectra and circular dichroism spectra of zeaxanthin monomer in EtOH, H, J1, and J2-aggregate ...... 35

Figure 2.3. Fits to absorption and emission spectra of the zeaxanthin monomer and aggregates...... 38

Figure 2.4. Emission spectra of monomeric zeaxanthin (1.0 µM in EtOH) and aggregates, with 488.0 nm excitation ...... 41

Figure 2.5. Fluorescence excitation spectra and fitted emission spectra of zeaxanthin monomer and aggregates ...... 44

Figure 2.6. Raman spectra of monomer and J1-aggregate of zeaxanthin with 488.0 nm excitation, and the DFT calculated spectrum of zeaxanthin ...... 46

Figure 2.7. Model for the absorption, emission, and non-radiative relaxation processes of zeaxanthin aggregates ...... 60

Figure A2.1. H-aggregate spectra resulting from varied rates of H2O addition to 25-75 μM solutions of zeaxanthin in ethanol...... 76

Figure A2.2A. Temperature-dependent UV-Vis spectra for the H- and J1-aggregates ... 77

Figure A2.2B. Temperature dependent UV-Vis spectra for the monomer and J2- aggregate ...... 78

Figure A2.3. Isomerization of samples, monitored by the cis-band at ~335 nm ...... 79

Figure A2.4: Emission from zeaxanthin monomer, and J1-, J2-, and H-aggregates...... 85

Figure A2.5. Resonance Raman spectra acquired with 488.0 nm excitation ...... 86

Figure A2.6. Fluorescence decay curves of zeaxanthin monomer and J1-, J2- and H- aggregate at room temperature, as measured by TCSPC ...... 90

viii

Figure 3.1. Molecular structure of 3R,3'R zeaxanthin monomer, with UV-Vis absorption spectra of monomer and J-aggregate ...... 94

Figure 3.2. CW-resonance Raman spectra of zeaxanthin monomer in THF, and J1- aggregate with 458 nm excitation ...... 95

Figure 3.3. TRRR spectra of monomer S1 state, probed at 551 nm, and J1-aggregate probed at various delays with 551 nm and 473 nm probe ...... 96

Figure 3.4. Kinetic fitting for Relative Raman scattering intensity in the ethylenic band region...... 97

Figure 3.5. Percentage of ground state depletion for 415 nm photoexcitation, with varying pump energies at 4 psec time delay ...... 99

Figure A3.1. NMR spectrum of zeaxanthin in CDCl3 ...... 107

Figure A3.2. DLS probability distribution for the zeaxanthin J1-aggregate ...... 108

Figure A3.3. Circular dichroism spectrum of zeaxanthin J1-aggregate ...... 109

Figure A3.4. The cross-correlation of 415 nm pump and 551 nm probe beams...... 111

Figure A3.5 TRRR spectra work up process ...... 114

Figure A3.6. Representative TRRR spectra of zeaxanthin monomer in EtOH ...... 115

Figure A3.7. Convolution of instrument response and kinetics of ground state recovery of monomer and J1-aggregate ...... 121

Figure 4.1. Molecular structure of 3R,3'R-zeaxanthin. Model of zeaxanthin in the gel (T <

Tm) and fluid (T > Tm) phases of the bilayer. UV-Vis absorption and circular dichroism spectra of zeaxanthin in lipid bilayers at room-temperature (RT) or 50 °C ...... 126

Figure 4.2. Steady-state resonance Raman spectra of zeaxanthin in lipid bilayers and in ethanol ...... 127

Figure 4.3. TRRR difference spectra of zeaxanthin at room temperature in the gel phase of DPPC lipid bilayers and in the fluid phase of the POPC lipid bilayers. TRRR spectra of zeaxanthin monomer in cyclohexane are shown for comparison ...... 129

Figure 4.4. Percentage of ground state bleach, as a function of excitation percentage calculated from the sample absorption, pump pulse energy, and beam cross-section .... 131

Figure 5.1. UV-vis absorption spectra of zeaxanthin monomer in EtOH, H-aggregate in

20:80 (v/v) EtOH/H2O, J1-aggregate in 10:90 (v/v) THF/H2O and J2-aggregate in 20:80 (v/v) 1,4 Dioxane/H2O ...... 140

Figure 5.2. Fs-TA spectra at different time delays of zeaxanthin monomer in ethanol, J1- aggregate, J2- and H-aggregate ...... 145

Figure 5.3. TA kinetics at selected wavelengths with 520 nm excitation of the J1- aggregate and the H-aggregate ...... 149

Figure 5.4. Ns-TA spectra for the zeaxanthin J1-aggregate excited at 410 nm (Upper) and the H-aggregate excited at 355 nm (Lower) ...... 151

Figure 5.5. Ns-TA kinetic decays for the zeaxanthin J1and H-aggregates...... 152

Figure 5.6. Summary of zeaxanthin monomer photophysics ...... 153

Figure 5.7. Excited-state dynamics of zeaxanthin aggregates ...... 160

Figure 6.1. UV-vis spectra of zeaxanthin monomer in EtOH, SH-aggregate in 20%

EtOH/H2O (v/v) and WH-aggregate in 10% THF/H2O (v/v) ...... 176

Figure 6.2. Transient absorption spectra of weak (WH) and strong (SH) H-aggregates of zeaxanthin at 3 and 150 ps delay between pump and probe ...... 177

Figure 6.3.TRRR spectra workup of WH-554 ...... 179

Figure 6.4. TRRR spectra of the WH-aggregate, probed with 599 nm ...... 180

Figure 6.5.TRRR spectra workup of WH-497 and SH-413 ...... 182

Figure 6.6. TRRR spectra for WH-aggregates with 413 nm pump, 471 nm probe at different pump/probe time delays...... 184

Figure 6.7. TRRR spectra of zeaxanthin WH and SH aggregates at 150 ps time delay with different probe wavelengths ...... 186

Figure 6.8. Raman intensity analysis and reconstructed excited state absorption spectra of WH SH aggregate at 150 ps...... 188

Figure 6.9. Anti-Stokes TRRR spectra of the WH-aggregate pumped with 418 nm or 397 nm, and probed with 557 nm or 592 nm, respectively ...... 191

Figure 6.10. Exciton coupling model of ground state electronic transitions for WH and SH aggregates, in comparison with the monomer (Mon) ...... 193

Figure 6.11. Exciton coupling model for triplet excited states generated from singlet fission in zeaxanthin SH and WH aggregates ...... 198

Figure 6.12. Exciton coupling between the S0→S2 and T1→Tn transitions ...... 204

Figure A6.1. Stokes and anti-Stokes of spectra of zeaxanthin WH-aggregate probed with 557 nm, 4 ps pulse laser and 568 nm CW laser ...... 221

Figure A6.2A. Data workup for WH-554 experiment...... 223

Figure A6.2B. Data workup for WH-497 experiment ...... 224

Figure A6.2C. Data workup for SH-413 experiment ...... 225

Figure A6.2D. Data workup for WH-471 experiment...... 225

Figure A6.3. TRRR spectra of zeaxanthin WH and SH aggregate with different probe wavelengths at 150 ps time delay. Transient microscopic Raman spectra of the T1 state of zeaxanthin monomer in acetone formed through sensitization of anthracene ...... 227

Figure A6.4. Kinetics for selected bands of the anti-Stokes TRRR spectra of the WH- aggregate ...... 228

Figure A6.5. Stabilization of the S2 state in zeaxanthin aggregate due to exciton coupling and formation of excimer states ...... 229

Figure 7.1. Experimental and calculated Raman spectra of tetrathiophene and hexathiophene in their S0 and T1 states...... 236

Figure 7.2. Vibrational normal modes of the T1 excited state of model molecule (ethyl)2- 4T, from DFT computations ...... 239

Figure 7.3. Representative structures of the T1 state of (Dodec)2-4T and (Oct)4-6T, and dication of (Oct)4-6T...... 241

Figure 7.4. Raman spectra of the T1 and S0 states of (Oct)4-6T aggregate in 5/95 (v/v) THF/water ...... 244

Figure A7.1. Steady state absorption of the S0 state, and transient absorption spectra of the T1 excited-state of (Dodec)2-4T in chlorobenzene and (Oct)4-6T in THF ...... 253

Figure A7.2. Steady state absorption of the S0 state, and transient absorption spectra of the T1 excited-state of (Oct)4-6T monomer in THF and aggregates in 5/95 (v/v) THF/H2O ...... 254

Figure A7.3. Transient Raman spectra data workup for the monomer of (Oct)4-6T ...... 259

Figure A7.4. Final spectrum of the T1 state for tetraoctyl hexathiophene in THF ...... 260

Figure A7.5. Transient resonance Raman spectra of (Oct)4-6T monomer in THF solution with 420 nm pump and 647.1 nm probe at ~5 µs, ~20 µs time delay and ~5 µs time delay with O2 ...... 261

Figure A7.6. Experimental Raman spectra probed at 647.1 nm for ground-state (Dodec)2- 4T in chlorobenzene and (Oct)4-6T in THF, and the DFT calculated spectra (lower spectra) of the ethyl substituted tetramer and hexamer ...... 263

Figure A7.7. Representative vibrational normal modes from DFT calculations of (ethyl)2- 4T and (ethyl)4-6T in the ground state ...... 264

Figure A7.8. DFT calculated Raman spectra for the T1 state of (ethyl)2-4T with B3LYP functional using different basis sets ...... 265

Figure A7.9. DFT calculated Raman spectra for the T1 state of (ethyl)4-6T with B3LYP, CAM-B3LYP and ωB97X-D functionals with 6-31G(d) basis set ...... 266

Figure A7.10. Raman spectra of (ethyl)4-6T in the ground state, calculated with DFT and the B3LYP or PBEPBE functionals ...... 267

Figure A7.11. Raman spectra of (ethyl)4-6T in the triplet (T1) state, calculated with DFT and the B3LYP or PBEPBE functionals ...... 268

Figure A7.12. Comparison between similar T1 vibrational normal modes for (ethyl)2-4T and (ethyl)4-6T from DFT (UB3LYP/6-31G(d)) calculation ...... 269

xii

LIST OF SCHEMES

Scheme 1.1. Molecular structure of 3R,3´R-zeaxanthin and β-carotene ...... 7

Scheme 7.1. Molecular structures of didodecyl quaterthiophene (Dodec)2-4T and tetraoctyl sexithiophene (Oct)4-6T...... 235

xiii

LIST OF TABLES

Table 2.1. Fitted absorbance and emission peaks, and fluorescence quantum yields (QY) of zeaxanthin aggregates ...... 37

Table A2.1. Observed peak positions and amplitudes of UV-Vis absorption and CD spectra ...... 75

Table A2.2. Peak positions of five Gaussian bands, fitted to H aggregate spectra shown in Figure A2.1 ...... 76

Table A2.3. Components for calculations of fluorescence quantum yields, ΦF ...... 84

Table A2.4. Descriptions of selected vibrational modes from DFT calculations ...... 87

Table A3.1. Data for J1-aggregate, supporting Figure 3.3 of maintext ...... 119

Table 5.1. Multiexponential fitting results for kinetics of fs-TA bands of zeaxanthin monomer and aggregates ...... 150

Table 6.1. Laser wavelengths of the pump and probe beams ...... 173

Table 6.2. Loss of Raman scattering caused by pump excitation of WH and SH TRRR spectra measured at various probe wavelengths ...... 183

Table 6.3. Predicted and observed enhancement factors for active Raman modes in AS- TRRR...... 208

Table A6.1. Laser wavelengths of the pump and probe beams and the corresponding filters for TRRR experiments ...... 229

Table A6.2. Pump and probe parameters for TRRR experiments...... 230

Table 7.1. Categorization of Raman modes calculated by DFT for model quaterthiophene and sexithiophenein the triplet (T1) excited state ...... 238

xiv

ACKNOWLEDGEMENTS

My doctoral research at University of California, San Diego relied upon the help and support of many people. Here, I would like express my gratitude to some of them.

Above all, I acknowledge my advisor Prof. Michael Tauber, who brought me to

UCSD and guided my study. I respect his scientific spirit and will always appreciate his direction on many aspects of my development. In the Tauber lab, I had the opportunity to work with several wonderful colleagues, including Ms. Maria Angelella and Ms.

Samantha Doyle. Two other students in the lab, Ms. Brittany Merrill and Mr. Christopher

Berg, assisted my research on the carotenoids.

I thank Prof. Judy Kim for her insights, and for sharing apparatus in her lab.

Without her help, it would have taken a much longer time to complete my work. Several members of the Kim lab, including Dr. Hannah Shafaat, Dr. Diana Schlamadinger, and

Dr. Brian Leigh provided invaluable assistance on topics ranging from the operation of lasers, to handling lipid vesicles. I wish to thank others at UCSD, including Dr. Wei Lin and Dr. Ferran Feixas who helped me to setup DFT calculations, Dr. Miao-ping Chien and Dr. Norm Olson for assistance with TEM, and Dr. Yongxuan Su for his help with

HPLC and mass spectrometry.

The work described in Chapter 7 resulted from our participation in a DOE-funded

Energy Frontier Research Center (EFRC). I thank Prof. Peter Rossky for his ideas and involvement, and Prof. Bradley Holliday’s group for providing an oligothiophene sample for my work.

Last, I would like to express my special gratitude to my family. It was my parents who encouraged and inspired me to pursue a life in science. They also provided me constant support for dozens of years, and the opportunity for a great education. I will also thank the most special lady I have met on the UCSD campus, Miss Yipeng Wu. She has been standing behind me, forgiving my carelessness, bearing my poverty, and sharing my busy days, while offering me comfort and understanding.

Chapter 2, in full, is a reprint of the material as it appears in the Journal of

Physical Chemistry B, 2012, 116(35), pp 10617-10630, by Chen Wang, Christopher J.

Berg, Cheng-Chih Hsu, Brittany A. Merrill, and Michael J. Tauber. The author of this dissertation was one of the two primary experimentalists, and was a coauthor of the paper.

Chapter 3, in full, is a reprint of the material as it appears in the Journal of the

American Chemical Society, 2010, 132(40), pp 13988-13991, by Chen Wang and

Michael J. Tauber. The author of this dissertation conducted the experiments, and was a coauthor of the paper.

Chapter 4, in full, is a reprint of the material as it appears in ChemPhysChem,

2011, 12(16), pp 2891-2894, by Chen Wang, Diana E. Schlamadinger, Varsha Desai, and

Michael J. Tauber. The author of this dissertation conducted most of the experiments, and was a coauthor of the paper.

Chapter 5, in full, is a reprint of the material as it appears in the Proceedings of the SPIE, Physical Chemistry of Interfaces and Nanomaterials XI, 2012, 8459(845905), by Chen Wang, Maria Angelella, Chun-Hung Kuo, and Michael J. Tauber. The author of this dissertation was the primary experimentalist, and was a coauthor of the paper.

xvi

VITA

2004 B.S. in Applied Chemistry, Fudan University, Dept. of Chem. 2007 M.S. in Physical Chemistry, Fudan University, Dept. of Chem. 2007 –2008 Teaching Assistant at Michigan State Univ., Dept. of Chem. 2008 –2014 Research and Teaching Assistant at UC San Diego, Dept. of Chem. & Biochem. 2014 Ph. D., UC San Diego, Dept. of Chem. & Biochem.

PUBLICATIONS 1. M. Angelella, C. Wang, M. J. Tauber. Resonance Raman Spectra of a Perylene Bis(dicarboximide) Chromophore in Ground (S0) and Lowest Triplet (T1) States. J. Phys. Chem. A, 2013, 117, (38), 9196 2. C. Wang, M. Angelella, C. H. Kuo, M. J. Tauber. Singlet Fission in Carotenoid Aggregates: Insights from Transient Absorption Spectroscopy. SPIE Proceeding: NanoSicence Engineering, 2012, 845905 13 3. C. Wang, C. J. Berg, C. C. Hsu, B. A. Merrill, M. J. Tauber. Characterization of Carotenoid Aggregates by Steady-State Optical Spectroscopy. J. Phys. Chem. B, 2012, 116, (35), 10617 4. C. Wang, D. E. Schlamadinger. V. Desai, M. J. Tauber. Triplet Excitons of Carotenoids Formed by Singlet Fission in a Membrane. ChemPhysChem, 2011, 12, 2891. 5. Z. Fei, G. O. Andreev, W. Bao, L. M. Zhang, A. S. McLeod, C. Wang, M. K. Stewart, Z. Zhao, G. Dominguez, M. Thiemens, M. M. Fogler, M. J. Tauber, A. H. Castro- Neto, C. N. Lau, F. Keilmann D. N. Basov. Infrared Nanoscopy of Dirac Plasmons at the Graphene–SiO2 Interface. Nano Lett., 2011, 11, 4701. 6. C. Wang, M. J. Tauber. High-Yield Singlet Fission in a Zeaxanthin Aggregate Observed by Picosecond Resonance Raman Spectroscopy. J. Am. Chem. Soc., 2010, 132, 13988. 7. Y. Li, C. Wang, H. Chen, W. Hua, Y. Yue, Z. Gao. Isomerization of α-Pinene Over Porous Phosphate Heterostructure Materials: Effects of Porosity and Acidity. Catal. Lett., 2009, 131, 560. 8. C. Wang, W. Hua, Y. Yue , Z. Gao. Delaminated Microporous Aluminophosphate- Filled Poly (Vinyl Alcohol) Membrane for Pervaporation of Aqueous Alcohol Solutions. Micro. Meso. Mat., 2007, 105, 149.

xvii

9. C. Wang, Y. Li, W. Hua, Y. Yue , Z. Gao. Delamination and Intercalation of Layered 3- Aluminophosphate with [Al2P3O12] Stoichiometry by a Controlled Two-step Method. Stud. Surf. Sci. Catal., 2007, 165, 143. 10. C. Wang, W. Hua, Y. Yue , Z. Gao. Controlled Delamination and Intercalation of Layered Microporous Aluminophosphate by a Novel Two-step Method. Micro. Meso. Mat., 2005, 84, 297. 11. C. Wang, L. Peng, W Hua, Y. Yue, Z. Gao. Delamination and Aromatic Amine Intercalation of Layered Microporous Alumino-phosphates. Chem. J. Chinese Univ.- Chinese, 2006, 27, 1509. 12. C. Wang, W. Hua, Y. Yue , Z. Gao Delamination and Aromatic Amine Intercalation 3− of Layered Aluminophosphate with [Al3P4O16] Stoichiometry. J. Coll. Interf. Sci., 2005, 285, 731.

xviii

ABSTRACT OF THE DISSERTATION

Spectroscopic studies of singlet fission and triplet excited states in assemblies of conjugated organic molecules

Chen Wang

Doctor of Philosophy in Chemistry

University of California, San Diego, 2014

Professor Michael Tauber, Chair

Triplet states of organic molecules are excited-state species that are important on a fundamental level, as well as for emerging organic electronic devices. The singlet fission mechanism, which generates as many as two triplet excitons per absorbed photon, may improve the efficiency of new OPVs. Singlet fission is a primary focus of this dissertation, particularly the relationship between interchromophore coupling and the yield of triplet excited states. Zeaxanthin, a carotenoid, is a model molecule for this study because it forms aggregates with different interchromophore couplings. Three types of aggregates, as well as zeaxanthin multimers in vesicles, were prepared and probed. The spectroscopic techniques included transient absorption and time-resolved resonance

xix

Raman spectroscopy. The coupling strengths between chromophores spanned the range from weak to strong. High yields of triplets by singlet fission were found in all systems with exciton coupling. The most triplets were produced in assemblies with the greatest coupling strength. The results disprove earlier suggestions in the literature that the singlet fission in carotenoids is intramolecular. Our findings also reveal that an optically-allowed

S2 state is the parent state for fission, rather than the lowest excited singlet state (S1) as concluded in earlier theoretical work.

The first resolved resonance Raman spectra of oligothiophenes in their triplet excited state are also reported and analyzed in this dissertation.

Chapter 1 Introduction

The total spin of two electrons in separate orbitals can either be 0 or 1, as shown with cone diagrams in Figure 1.1. For S=0, there is only one coupling of spin vectors, so it is called the singlet state. The S=1 state involves three different orientations of the spin vectors, hence it is called a triplet.

Figure 1.1. The spin diagram of singlet and triplet state1

The spin wavefunctions define the different properties of the lowest triplet state compared to singlet excited states:

(1) In the absence of spin-orbit coupling, optical or nonradiative transitions between states with different spin multiplicities are not allowed. Spin-allowed transitions such as fluorescence and internal conversion are usually several orders of magnitude faster than spin-forbidden processes, like phosphorescence and intersystem crossing.

Hence, the lowest triplet excited state of a molecule usually has a much longer lifetime compared to its singlet counterpart, in the typical situation when the ground state of the molecule is a singlet state.

(2) The energy level of a triplet state is usually lower than the corresponding singlet counterpart. This ordering is a result of the fundamental property of fermions. The

Pauli exclusion principle requires that the total electronic wavefunction be antisymmetric upon exchange of any two electrons. Therefore, the symmetric spin wavefunctions of the triplet state require that the spatial wavefunction be antisymmetric, and vice versa for the singlet. In the triplet state, the unpaired electrons avoid each other, because the antisymmetric spatial wavefunction vanishes when the two electrons have the same coordinates.2 This phenomenon is known as the Fermi hole.3, 4 In contrast, for the singlet, the same Fermi correlation increases the probability of finding the two coupled electrons at the same position.2, 5 The avoidance of electrons in the case of the antisymmetric spatial wavefunction has been mentioned as the reason why the electron-electron repulsion energy is lower for the triplet state.1, 6 However, other work has shown that a more important reason for energetic stabilization of the triplet is that there is more favorable interaction between electrons of the antisymmetric spatial wavefunction and the nuclei.7-9 With either justification, the triplet state energy is lower than that of the singlet state for a pair of electrons that occupy two different spatial orbitals.

The fundamental research described in this dissertation is mainly connected to organic photovoltaic devices (OPVs). Triplet excited states are relevant for OPVs in two ways, which are associated with the basic triplet properties discussed above. First, the long lifetime of triplet excited states allows much longer diffusion lengths compared with singlet excitons.10-12 Therefore, if an OPV device were to utilize triplet excitons, it could benefit from reduced decay of excitons before undergoing charge separation at an interface. This advantage would require an efficient way to generate triplet excitons in the 3 first place. Second, the low energy triplet state has been shown to be a relaxation pathway of detrimental charge recombination.13-15 Knowledge about the basic properties of the triplet states is crucial for improving the efficiency of organic photovoltaics.

My research has focused primarily on a mechanism that can generate triplet excited states with extraordinarily high yields. The mechanism is singlet fission, in which two triplet excitons are generated simultaneously from one singlet excited state. Singlet fission could improve the energy conversion efficiency of OPVs. In a conventional single bandgap device, only one singlet exciton is formed with a high energy incoming photon.

The excess exciton energy above the conduction band of semiconductor is wasted. The maximum efficiency of a single bandgap device is approximately 30% under standard solar illumination.16 An improved device consists of a bilayer. The upper layer absorbs the relatively high-energy (blue/green) photons, and the lower absorbs the relatively low energy (red) photons. If the blue-green absorbing layer is one where the mechanism of singlet fission can take place, then two excitons can be generated per absorbed photon, and in principle the photocurrent can double as a result of such a layer. An ideal singlet fission material having 200% triplet yield, in combination with an efficient red absorber, is predicted to improve the limiting conversion efficiency by ~50% over the Shockley-

Queisser limit.17-19

Singlet fission is a spin-allowed process which involves transitions between excited states with different spin multiplicities. To describe the mechanism, we first consider a system consisting of two unpaired electrons in one molecule. For the systems with coupled electrons, it is convenient to replace the triplet spin functions of Figure 1.1 with the three eigenstates of the spin-dipolar Hamiltonian in zero field:20 4

Eq. 1.1 (a)

Eq. 1.1 (b)

Eq. 1.1 (c)

Here, the new functions of Equation 1.1 are labeled because they are aligned along the set of principal axis (x, y, z) of a molecule that diagonalize the zero field Hamiltonian.20

Since fission involves two molecules, we expand the system to four electrons located in two molecules (labeled A and B, respectively). The number of spin eigenstates increases to sixteen after diagonalization, including two with S=0, nine with S=1, and five with S=2.21, 22 The two eigenstates with S=0 can be expressed as,

Eq.1.2 (a)

Eq.1.2 (b)

Eq.1.2 (a) represents a pair of singlet states located on each molecule. Eq.1.2 (b) can be rewritten in terms of the zero-field triplet wavefunctions of equation 1.1(a-c) as,

Eq. 1.3

Eq. 1.3 indicates that the state can be interpreted as a superposition of triplet states located on molecule A and B, that are coupled to form an overall singlet state. Therefore,

singlet fission represents a special type of spin-allowed internal conversion, , in which consists of one singlet excited state and one ground state. The fission process can happen efficiently when the triplet energy is equal or lower than half of the energy of the excited singlet state.23 5

The phenomenon of singlet fission was first discovered in crystalline acenes24, 25 and subsequently in other systems including conjugated polymers26-28 and carotenoids.26-

35 The possible application to photovoltaics spurred a worldwide revival of research in singlet fission during the period of my doctoral research. A variety of systems have been explored recently, including acene derivatives,36-41 carotenoid aggregates,42-44 a semiconducting polymer,45 1,3-diphenylisobenzofuran,46 diphenylpolyene,47 and perylene diimide.48 The extraction of charges generated via fission and subsequent charge separation has been reported in several studies.49-54

Most studies of the singlet fission mechanism have focused on molecules whose lowest singlet excited states (S1) are optically allowed and can be described as electron promotion from the highest-occupied molecular orbital (HOMO) to lowest unoccupied molecular orbital (LUMO). These molecules (termed “Class I”)18 include tetracene and

55-60 pentacene. Their S1 energy level is close to twice that of T1. A precursor state that is almost degenerate with S1 and has multiexciton character is proposed to couple with S1 and the two separated triplet states. This coupling can occur either via a direct mechanism,59,60 or an indirect mechanism with the mediation of charge-transfer states.61,62

Class III is another group of fission chromophores. Polyenes are among this class of molecules. The optically allowed transition is not to the lowest excited singlet state, but rather to a higher one that is generally labeled S2. The lowest singlet excited state (S1) has been suggested as the precursor for efficient singlet fission as described later in this introduction.18,63 Energy dissipation processes during the fission of class III may be different from that of class I, because of the large energy gap between the high-lying optically allowed S2 and the low-lying T1 states. Moreover, triplet-triplet (T-T) 6

annihilation through the low-lying T2 state of class III chromophores is predicted to compete with fission.18. There are many fewer studies of singlet fission among class III chromophores, in comparison with class I.

Another comparison between class I and III is related to the role of interchromophore coupling in fission process. The coupling effects on fission yield have been discussed in detail for systems of class I.61, 62, 64-67 However, for chromophores of class III, an intramolecular mechanism was proposed for carotenoids in light-harvesting systems.31 This proposal, if true, would obviate the need for interchromophore coupling in singlet fission of class III, because two triplet states would form and reside on a single molecule. The relationship between basic aspects of interchromophore coupling

(orientation, or strength) and intermolecular singlet fission in class III molecules was unexplored prior to my graduate work, and provided much motivation for the experiments described in this thesis.

An ideal molecule for this study would have two qualities. First, its energy levels should be suitable for efficient singlet fission. Second, the electronic coupling between molecules should be tunable. We selected a carotenoid known as zeaxanthin (Scheme

1.1) for the studies of singlet fission reported in this thesis. Zeaxanthin has eleven conjugated double bonds, and therefore it is a polyene and classified as a class III chromophore. Its properties include: (1) Its low triplet state energy level fulfills the

energy criterion of efficient fission, . (2) It self-assembles to form aggregates with different degrees of coupling among constituent molecules. 7

Scheme 1.1. Molecular structure of 3R,3´R-zeaxanthin and β-carotene

This dissertation is organized as follows: First, the preparation and characterization of three types of zeaxanthin aggregates are described (Chapter 2).

Second, I describe our initial observations of singlet fission in these systems and other assemblies, using transient (pump-probe) spectroscopy (Chapter 3-5). Third, I report our efforts to gain in-depth insights into the mechanism via the combined use of transient absorption (TA) and time-resolved resonance Raman (TRRR) spectroscopy, over times ranging from subpicosecond to microseconds (Chapter 6). Both singlet fission and the triplet-triplet annihilation processes are monitored. The fission yields for systems with different coupling strengths were compared.

In the remainder of this introductory chapter, I will briefly review the electronic structure of carotenoids, their aggregation properties, and main experimental techniques used in my work.

Electronic structure of zeaxanthin. Electronic excited states of carotenoids are defined by the polyene core. A strong optically-allowed transition from the ground state

1 - 1 + (1 Ag ) to the excited singlet state (1 Bu ) dominates the absorption spectrum (Figure

1.2). The final state is commonly denoted S2, which corresponds to the promotion of a

1 - single electron from HOMO to LUMO. The lowest excited singlet state, S1(2 Ag ), has 8

68-75 been characterized both experimentally and theoretically. The S1 state is formed

76-79 through a fast internal conversion from the S2 state, on the time scale of 100-200 fs.

The S1 states of β-carotene and zeaxanthin relax to S0 with a time constant of ~10 ps through internal conversion.73, 79-81 Numerous studies report evidence for at least one

68-72 electronic excited state that is located between the S1 and optically-allowed S2 states, however the exact nature of these states remain controversial.82 Compared to the internal conversion, intersystem crossing is highly improbable for the carotenoids. For example, zeaxanthin and β-carotene have triplet quantum yields of lower than 0.2% or less.83, 84

-1 The energy levels of the T1 state of β-carotene and zeaxanthin are about 7000 cm above

85-87 88, 89 the S0. The lifetime of the T1 state is reported to be 9 to 22 microseconds.

For polyenes, electronic configurations of excited states are not very well

71 described by the picture of molecular orbitals. The S1 state of carotenoids consists of multiple configurations, with the dominant one a double excitation where both electrons

71, 90, 91 of the HOMO are promoted to the LUMO. The T1 state can be primarily

92, 93 represented with a single HOMO→LUMO excitation. Hence, the S1 state can be considered as two local triplet states coupled on the same molecules.63, 69, 71, 94 It has been suggested that the S1 state has the potential to dissociate into two triplet excited states via singlet fission.18, 63 9

Figure 1.2. Electronic state transitions of zeaxanthin monomer. Arrows indicate transitions: dashed lines: internal conversion; dotted lines: intersystem crossing. Most (a) citations for the data can be found in text. Literature 88: 9 μs; literature 89:22 μs.

Zeaxanthin aggregates. Carotenoids are often assembled in natural systems with significant electronic coupling to chromophores of the same or different types. The exciton coupling interaction can have a significant effect on optical transitions.95 The exciton coupling effect between carotenoids was proposed as an explanation for unique colors in flower petals96 and the shells of crustaceans.97, 98 One of the useful properties for carotenoids with hydroxide groups, such as zeaxanthin or lutein, is that they can self- assemble to form aggregates in binary organo-aqueous solutions.99-104 The structures and exciton coupling features of carotenoid aggregates have been discussed in previous research.102, 103, 105-107 In this thesis, three types of zeaxanthin aggregates were prepared.

With multiple steady state characterization methods, we investigated the 10 interchromophore coupling in zeaxanthin aggregates. The knowledge serves as a foundation for understanding the effects of exciton coupling strength on fission efficiency of chromophores of class III.

Most early research on singlet fission relied on indirect observation of fluorescence of the singlet state.25 Mechanistic aspects of singlet fission were revealed via the effects of applied magnetic field on the emission.108, 109 Later time-resolved fluorescence study provided kinetic information about fission process.38, 55, 110, 111

However, the techniques used by earlier researchers were unable to provide direct spectroscopic “snapshots” of the singlet fission dynamics, and in general the samples were solid-state crystals. Our interest has centered on pump-probe techniques to acquire more direct spectroscopic and kinetic insights into the singlet fission process. Two types of spectroscopy that have played a central role in my doctorate research are described briefly below.

Transient absorption spectroscopy. Exceptionally fast time-resolution (even sub-fs)112-114 and broad spectral range make transient absorption spectroscopy a versatile method for investigating ultrafast dynamics of excited states. In principle, excited states with different electronic structure can be distinguished, via distinct absorption spectra.

Kinetics is another characteristic that can be used to identify excited species. The singlet and triplet excited states and other transient products formed in photoexcitation can be distinguished by their different lifetimes. Although fs-TA is a powerful technique, it has some limitations. Most important among them, the electronic absorption bands can be broad and featureless. If the bands are also overlapped, the assignment of different 11 excited species may be difficult or impossible. For that reason, a technique that can probe each transition individually provides a powerful complement to fs-TA spectroscopy.

Time-resolved resonance Raman spectroscopy. Vibrational spectroscopy is a sensitive probe of bonding between molecules. Since bonding patterns and force constants differ for various electronic states of any molecule, time-resolved vibrational spectroscopy is a powerful approach for probing transitions between electronic states.

Resonance Raman spectroscopy is particularly well-suited for our studies of conjugated molecules, because specific electronic transition can be amplified by selection of the excitation wavelength. Time-resolved resonance Raman spectroscopy has been reviewed previously.115, 116 Furthermore, the resonance Raman excitation profile (REP) yields insights into the higher (resonant) electronic states.117 The fastest time-resolution for spontaneous Raman experiment involves a tradeoff with spectral resolution. A spectral resolution of ~10 cm-1 resolution is associated with time resolution on the order of several picoseconds.115 It is possible to achieve time-resolution to the level of fs-TA experiments with femtosecond stimulated Raman spectroscopy (FSRS).118 However, these experiments are significantly more complex than spontaneous TRRR. Additionally, the samples in FSRS are subjected to much higher photon fluxes as required for a stimulated process. In our studies, we aimed to minimize photodamage and photon flux, for example by defocusing beams in our ps-TRRR experiments.

When both fs-TA and ps-TRRR are combined, we gain a comprehensive picture about excited state dynamics of singlet fission and T-T annihilation in zeaxanthin aggregates. The TA experiment is used to monitor the kinetics of excitation process.

Using TA spectra as a roadmap, TRRR experiments with multiple probe wavelengths are 12 employed to identify and characterize various excited species. The Raman intensity profile analysis is employed as the supplementary information for TA spectra to understand the electronic transitions of excited states. Anti-stokes Raman experiments shed light on the energy dissipation processes associated with fission and annihilation.

Aside from their role in singlet fission, the dynamics of triplet excited states in organic materials is also important for improving the efficiency of OPVs.11 For example, recently, researchers found that a comprehensive consideration of energy levels, the wavefunction delocalization and the spin of the triplet and the charge transfer states is necessary to identify the optimal conditions for suppressing charge recombination in bulk heterojunction devices.119 To understand these dynamics, there is a fundamental need to track the behavior of triplet excitons in organic materials with spectroscopic methods.

In the final chapter of this dissertation, I describe our efforts to characterize the T1 state of oligothiophenes using resonance Raman spectroscopy. Polymers and oligomers of thiophene are an important class of organic semiconductor, with emerging applications in OPVs.120 The triplet properties of the oligomer evolve with chain length and can be extrapolated to the polymer.121, 122 Two short chain oligothiophenes, a tetramer and a hexamer, are chosen as model compounds. The triplet states of these oligomers are characterized and analyzed by TRRR for the first time, in both their solvated and aggregated forms.

References

1. Turro, N. J.; Ramamurthy, V.; Scaiano, J. C., Electronic, vibrational, and spin configurations of electronically excited states. In Principles of Molecular Photochemistry: An Introduction, University Science Books: 2009; p 59.

2. McGlynn, S. P.; Azumi, T., Molecular spectroscopy of the triplet state. Prentice- Hall: 1969.

3. Boyd, R. J.; Coulson, C. A., Fermi hole in atoms. Journal of Physics B-Atomic Molecular and Optical Physics 1974, 7, (14), 1805-1816.

4. Buijse, M. A., Exchange correlation and holes. In Electron Correlation: Fermi and Coulomb Holes Dynamical and Nondynamical Correlation, Centrale Huisdrukkerij Vrije Universiteit: 1991; Vol. Ph.D.

5. Kutzelnigg, W., Electron correlation and electron pair theories. In New Concepts I, Springer: 1973; pp 31-73.

6. Liu, R. S. H., "You're repulsive!"-Teaching VSEPR in a not-so-elegant way. Journal of Chemical Education 2005, 82, (4), 558.

7. Shim, I.; Dahl, J., A new interpretation of Hund's first rule. Theoretica chimica acta 1978, 48, (2), 165-174.

8. Boyd, R. J., A quantum mechanical explanation for Hund's multiplicity rule. Nature 1984, 310, (5977), 480-481.

9. Rioux, F., Hund's Multiplicity Rule Revisited. Journal of Chemical Education 2007, 84, (2), 358.

10. Shao, Y.; Yang, Y., Efficient organic heterojunction photovotaic cells based on triplet materials. Advanced Materials 2005, 17, 2841-2844.

11. Köhler, A.; Bässler, H., Triplet states in organic semiconductors. Materials Science and Engineering R 2009, 66, (71-109).

12. Yost, S. R.; Hontz, E.; Yeganeh, S.; Voorhis, T. V., Triplet vs singlet energy transfer in organic semiconductors: The tortoise and the hare. Journal of physical Chemistry C 2012, 116, 17369-17377.

13. Ohkita, H.; Cook, S.; Astuti, Y.; Duffy, W.; Heeney, M.; Tierney, S.; McCulloch, I.; Bradley, D. D. C.; Durrant, J. R., Radical ion pair mediated triplet formation in polymer-fullerene blend films. Chemical Communications 2006, (37), 3939-3941.

14. Westenhoff, S.; Howard, I. A.; Hodgkiss, J. M.; Kirov, K. R.; Bronstein, H. A.; Williams, C. K.; Greenham, N. C.; Friend, R. H., Charge recombination in organic 14 photovoltaic devices with high open-circuit voltages. Journal of the American Chemical Society 2008, 130, (41), 13653-13658.

15. Schlenker, C. W.; Chen, K. S.; Yip, H. L.; Li, C. Z.; Bradshaw, L. R.; Ochsenbein, S. T.; Ding, F. Z.; Li, X. S. S.; Gamelin, D. R.; Jen, A. K. Y.; Ginger, D. S., Polymer triplet energy levels need not limit photocurrent collection in organic solar cells. Journal of the American Chemical Society 2012, 134, (48), 19661-19668.

16. Shockley, W.; Queisser, H. J., Detailed balance limit of efficiency of p-n junction solar cells. Journal of Applied Physics 1961, 32, (3), 510-519.

17. Hanna, M. C.; Nozik, A. J., Solar conversion efficiency of photovoltaic and photoelectrolysis cells with carrier multiplication absorbers. Journal of Applied Physics 2006, 100, (7), 0745101-8.

18. Smith, M. B.; Michl, J., Singlet fission. Chemical Reviews 2010, 110, (11), 6891- 6936.

19. Tayebjee, M. J. Y.; Gray-Weale, A. A.; Schmidt, T. W., Thermodynamic limit of exciton fission solar cell efficiency. Journal of Physical Chemistry Letters 2012, 3, (19), 2749-2754.

20. Carrington, A.; McLachlan, A. D., Introduction to magnetic resonance with applications to chemistry and chemical physics. Harper & Row: 1967.

21. Burdett, J. J.; Bardeen, C. J., The dynamics of singlet fission in crystalline tetracene and covalent analogs. Accounts of Chemical Research 2013, 46, (6), 1312-1320.

22. Burdett, J. J.; Piland, G. B.; Bardeen, C. J., Magnetic field effects and the role of spin states in singlet fission. Chemical Physics Letters 2013, 585, 1-10.

23. Paci, I.; Johnson, J. C.; Chen, X. D.; Rana, G.; Popovic, D.; David, D. E.; Nozik, A. J.; Ratner, M. A.; Michl, J., Singlet fission for dye-sensitized solar cells: Can a suitable sensitizer be found? Journal of the American Chemical Society 2006, 128, (51), 16546-16553.

24. Singh, S.; Jones, W. J.; Siebrand, W.; Stoichef, B. P.; Schneide, W. G., Laser generation of excitons and fluorescence in anthracene crystals. Journal of Chemical Physics 1965, 42, (1), 330-342.

25. Swenberg, C. E.; Geacintov, N. E., Exciton interactions in organic solids. In Organic Molecular Photophysics, Birks, J. B., Ed. John Wiley & Sons: London, 1973; Vol. 1, pp 489-564.

26. Kraabel, B.; Hulin, D.; Aslangul, C.; Lapersonne-Meyer, C.; Schott, M., Triplet exciton generation, transport and relaxation in isolated polydiacetylene chains: Subpicosecond pump-probe experiments. Chemical Physics 1998, 227, (1-2), 83-98. 15

27. Osterbacka, R.; Wohlgenannt, M.; Shkunov, M.; Chinn, D.; Vardeny, Z. V., Excitons, polarons, and laser action in poly(p-phenylene vinylene) films. Journal of Chemical Physics 2003, 118, (19), 8905-8916.

28. Lanzani, G.; Stagira, S.; Cerullo, G.; De Silvestri, S.; Comoretto, D.; Moggio, I.; Cuniberti, C.; Musso, G. F.; Dellepiane, G., Triplet exciton generation and decay in a red polydiacetylene studied by femtosecond spectroscopy. Chemical Physics Letters 1999, 313, (3-4), 525-532.

29. Cogdell, R. J.; Hashimoto, H.; Koyama, Y., Picosecond Raman-spectroscopy of the B830 LH2 complex of Chromatium purpuratum BN-5500. Photosynthesis Research 1993, 35, (1), 79-85.

- 30. Zhang, J. P.; Inaba, T.; Koyama, Y., The role of the newly-found 1Bu state of + - carotenoid in mediating the 1Bu to 2Ag internal conversion and the excited-state dynamics of carotenoid and bacteriochlorophyll in a bacterial antenna complex. Journal of Molecular Structure 2001, 598, (1), 65-78.

31. Gradinaru, C. C.; Kennis, J. T. M.; Papagiannakis, E.; van Stokkum, I. H. M.; Cogdell, R. J.; Fleming, G. R.; Niederman, R. A.; van Grondelle, R., An unusual pathway of excitation energy deactivation in carotenoids: Singlet-to-triplet conversion on an ultrafast timescale in a photosynthetic antenna. Proceedings of the National Academy of Sciences of the United States of America 2001, 98, (5), 2364-2369.

32. Papagiannakis, E.; Kennis, J. T. M.; van Stokkum, I. H. M.; Cogdell, R. J.; van Grondelle, R., An alternative carotenoid-to-bacteriochlorophyll energy transfer pathway in photosynthetic light harvesting. Proceedings of the National Academy of Sciences of the United States of America 2002, 99, (9), 6017-6022.

33. Agostini, G.; Corvaja, C.; Giacometti, G.; Pasimeni, L., Optical, zero-field ODMR abd EPR studies of the triplet states from singlet fission in biphenyl-TCNQ and biphenyl-tetrafluoro-TCNQ charge transfer crystals. Chemical Physics 1993, 173, (2), 177-186.

34. Watanabe, S.; Furube, A.; Katoh, R., Generation and decay dynamics of triplet excitons in Alq3 thin films under high-density excitation conditions. Journal of Physical Chemistry A 2006, 110, (34), 10173-10178.

35. Rademaker, H.; Hoff, A. J.; Van Grondelle, R.; Duysens, L. N. M., Carotenoid triplet yields in normal and deuterated Rhodospirillum rubrum. Biochimica et Biophysica Acta (BBA) - Bioenergetics 1980, 592, (2), 240-257.

36. Roberts, S. T.; McAnally, R. E.; Mastron, J. N.; Webber, D. H.; Whited, M. T.; Brutchey, R. L.; Thompson, M. E.; Bradforth, S. E., Efficient singlet fission discovered in a disordered acene film. Journal of the American Chemical Society 2012, 134, (14), 6388-6400. 16

37. Ramanan, C.; Smeigh, A. L.; Anthony, J. E.; Marks, T. J.; Wasielewski, M. R., Competition between singlet fission and charge separation in solution-processed blend films of 6,13-bis(triisopropylsilylethynyl)pentacene with sterically-encumbered perylene- 3,4:9,10-bis(dicarboximide)s. Journal of the American Chemical Society 2012, 134, (1), 386-397.

38. Piland, G. B.; Burdett, J. J.; Kurunthu, D.; Bardeen, C. J., Magnetic field effects on singlet fission and fluorescence decay dynamics in amorphous rubrene. Journal of Physical Chemistry C 2013, 117, (3), 1224-1236.

39. Ma, L.; Zhang, K. K.; Kloc, C.; Sun, H. D.; Michel-Beyerle, M. E.; Gurzadyan, G. G., Singlet fission in rubrene single crystal: direct observation by femtosecond pump- probe spectroscopy. Physical Chemistry Chemical Physics 2013, 14, (23), 8307-8312.

40. Lee, J.; Bruzek, M. J.; Thompson, N. J.; Sfeir, M. Y.; Anthony, J. E.; Baldo, M. A., Singlet exciton fission in a hexacene derivative. Advanced materials 2013, 25, (10), 1445-1448.

41. Minami, T.; Ito, S.; Nakano, M., Theoretical study of singlet fission in oligorylenes. Journal of Physical Chemistry Letters 2012, 3, (18), 2719-2723.

42. Wang, C.; Tauber, M. J., High-yield singlet fission in a zeaxanthin aggregate observed by picosecond resonance Raman spectroscopy. Journal of the American Chemical Society 2010, 132, (40), 13988-13991.

43. Wang, C.; Angelella, M.; Kuo, C. H.; Tauber, M. J. In Singlet fission in carotenoid aggregates: insights from transient absorption spectroscopy, SPIE NanoScience+ Engineering, 2012; International Society for Optics and Photonics: 2012; pp 845905-845905-13.

44. Wang, C.; Schlamadinger, D. E.; Desai, V.; Tauber, M. J., Triplet excitons of carotenoids formed by singlet fission in a membrane. Chemphyschem 2011, 12, (16), 2891-2894.

45. Musser, A. J.; Al-Hashimi, M.; Maiuri, M.; Brida, D.; Heeney, M.; Cerullo, G.; Friend, R. H.; Clark, J., Activated singlet exciton fission in a semiconducting polymer. Journal of the American Chemical Society 2013, 135, (34), 12747-12754.

46. Johnson, J. C.; Nozik, A. J.; Michl, J., High triplet yield from singlet fission in a thin film of 1,3-diphenylisobenzofuran. Journal of the American Chemical Society 2010, 132, (46), 16302-16303.

47. Dillon, R. J.; Piland, G. B.; Bardeen, C. J., Different rates of singlet fission in monoclinic versus orthorhombic crystal forms of diphenylhexatriene. Journal of the American Chemical Society 2013, 135, 17278-17281. 17

48. Eaton, S. W.; Shoer, L. E.; Karlen, S. D.; Dyar, S. M.; Margulies, E. A.; Veldkamp, B. S.; Ramanan, C.; Hartzler, D. A.; Savikhin, S.; Marks, T. J.; Wasielewski, M. R., Singlet exciton fission in polycrystalline thin films of a slip-stacked perylenediimide. Journal of the American Chemical Society 2013, 135, (39), 14701- 14712.

49. Rao, A.; Wilson, M. W. B.; Hodgkiss, J. M.; Albert-Seifried, S.; Bassler, H.; Friend, R. H., Exciton fission and charge generation via triplet excitons in pentacene/C60 bilayers. Journal of American Chemical Society 2010, 132, (36), 12698-703.

50. Jadhav, P. J.; Mohanty, A.; Sussman, J.; Lee, J.; Baldo, M. A., Singlet exciton fission in nanostructured organic solar cells. Nano Letters 2011, 11, (4), 1495-1498.

51. Jadhav, P. J.; Brown, P. R.; Thompson, N.; Wunsch, B.; Mohanty, A.; Yost, S. R.; Hontz, E.; Van Voorhis, T.; Bawendi, M. G.; Bulovic, V.; Baldo, M. A., Triplet exciton dissociation in singlet exciton fission photovoltaics. Advanced Materials 2012, 24, (46), 6169-6174.

52. Ehrler, B.; Walker, B. J.; Bohm, M. L.; Wilson, M. W. B.; Vaynzof, Y.; Friend, R. H.; Greenham, N. C., In situ measurement of exciton energy in hybrid singlet-fission solar cells. Nature Communications 2012, 3, 6.

53. Thompson, N. J.; Hontz, E.; Congreve, D. N.; Bahlke, M. E.; Reineke, S.; Van Voorhis, T.; Baldo, M. A., Nanostructured singlet fission photovoltaics subject to triplet- charge annihilation. Advanced Materials 2013.

54. Congreve, D. N.; Lee, J.; Thompson, N. J.; Hontz, E.; Yost, S. R.; Reusswig, P. D.; Bahlke, M. E.; Reineke, S.; Van Voorhis, T.; Baldo, M. A., External quantum efficiency above 100% in a singlet-exciton-fission-based organic photovoltaic cell. Science 2013, 340, (6130), 334-337.

55. Burdett, J. J.; Bardeen, C. J., Quantum beats in crystalline tetracene delayed fluorescence due to triplet pair coherences produced by direct singlet fission. Journal of the American Chemical Society 2012, 134, (20), 8597-8607.

56. Burdett, J. J.; Gosztola, D.; Bardeen, C. J., The dependence of singlet exciton relaxation on excitation density and temperature in polycrystalline tetracene thin films: Kinetic evidence for a dark intermediate state and implications for singlet fission. Journal of Chemical Physics 2011, 135, (21), 10.

57. Chan, W. L.; Ligges, M.; Jailaubekov, A.; Kaake, L.; Miaja-Avila, L.; Zhu, X. Y., Observing the multiexciton state in singlet fission and ensuing ultrafast multielectron transfer. Science 2011, 334, (6062), 1541-1545. 18

58. Chan, W. L.; Ligges, M.; Zhu, X. Y., The energy barrier in singlet fission can be overcome through coherent coupling and entropic gain. Nature Chemistry 2012, 4, (10), 840-845.

59. Zimmerman, P. M.; Bell, F.; Casanova, D.; Head-Gordon, M., Mechanism for singlet fission in pentacene and tetracene: from single exciton to two triplets. Journal of the American Chemical Society 2011, 133, (49), 19944-19952.

60. Zimmerman, P. M.; Musgrave, C. B.; Head-Gordon, M., A correlated electron view of singlet fission. Accounts of chemical research 2013, 46, (6), 1339-1347.

61. Greyson, E. C.; Vura-Weis, J.; Michl, J.; Ratner, M. A., Maximizing singlet fission in organic dimers theoretical investigation of triplet yield in the regime of localized excitation and fast coherent electron transfer. Journal of Physical Chemistry B 2010, 114, (45), 14168-14177.

62. Johnson, J. C.; Nozik, A. J.; Michl, J., The role of chromophore coupling in singlet fission. Accounts of Chemical Research 2013, 46, (6), 1290-1299.

63. Tavan, P.; Schulten, K., Electronic excitations in finite and infinite polyenes. Physical Review B 1987, 36, (8), 4337-4358.

64. Greyson, E. C.; Stepp, B. R.; Chen, X.; Schwerin, A. F.; Paci, I.; Smith, M. B.; Akdag, A.; Johnson, J. C.; Nozik, A. J.; Michl, J.; Ratner, M. A., Singlet exciton fission for solar cell applications: energy aspects of interchromophore coupling. Journal of physical Chemistry B 2009, 114, (45), 14223-14232.

65. Renaud, N.; Sherratt, P. A.; Ratner, M. A., Mapping the relation between stacking geometries and singlet fission yield in a class of organic crystals. Journal of Physical Chemistry Letters 2013, 4, 1065-1069.

66. Vallett, P. J.; Snyder, J. L.; Damrauer, N. H., Tunable electronic coupling and driving force in structurally well-defined tetracene dimers for molecular singlet fission: A computational exploration using density functional theory. Journal of Physical Chemistry A 2013, 117, (42), 10824-10838.

67. Berkelbach, T. C.; Hybertsen, M. S.; Reichman, D. R., Microscopic theory of singlet exciton fission. II. Application to pentacene dimers and the role of superexchange. Journal of Chemical Physics 2013, 138, (11), -.

68. Hudson, B. S.; Kohler, B. E., Low-lying weak transition in polyene α,ω- diphenyloctatetraene. Chemical Physics Letters 1972, 14, (3), 299-304.

69. Schulten, K.; Karplus, M., Origin of a low-lying forbidden transition in polyenes and related molecules. Chemical Physics Letters 1972, 14, (3), 305-309. 19

70. Tavan, P.; Schulten, K., The low-lying elecronic excitations in long polyenes - A PPP-MRD-CL study. Journal of Chemical Physics 1986, 85, (11), 6602-6609.

71. Hudson, B.; Kohler, B., Linear polyene electronic structure and spectroscopy. In Annual review of physical chemistry, Eyring, H., Ed. Annual Reviews: 1974; Vol. 25, pp 437-460.

72. Hudson, B. S.; Kohler, B. E.; Schulten, K., Linear polyene electronic-structure and potential surfaces. Excited States 1982, 6, 1-95.

73. Wasielewski, M. R.; Johnson, D. G.; Bradford, E. G.; Kispert, L. D., Temperature-dependence of the lowest excited singlet state lifetime of all-trans-β- carotene and fully deuterated all-trans-β-carotene. Journal of Chemical Physics 1989, 91, (11), 6691-6697.

74. Hashimoto, H.; Koyama, Y., Time-resolved resonance Raman spectroscopy of triplet β-carotene produced from all-trans, 7-cis, 9-cis, 13-cis, and 15-cis isomers and high-pressure liquid-chromatography analyses of photoisomerization via the triplet state. Journal of Physical Chemistry 1988, 92, (8), 2101-2108.

75. Hashimoto, H.; Koyama, Y., The C=C stretching Raman lines of β-carotene isomers in the S1 state as detected by pump-probe resonance Raman spectroscopy. Chemical Physics Letters 1989, 154, (4), 321-325.

76. Shreve, A. P.; Trautman, J. K.; Owens, T. G.; Albrecht, A. C., Determination of the S2 lifetime of β-carotene. Chemical Physics Letters 1991, 178, (1), 89-96.

77. Ricci, M.; Bradforth, S. E.; Jimenez, R.; Fleming, G. R., Internal conversion and energy transfer dynamics of spheroidene in solution and in the LH-1 and LH-2 light- harvesting complexes. Chemical Physics Letters 1996, 259, (3-4), 381-390.

78. Macpherson, A. N.; Gillbro, T., Solvent dependence of the ultrafast S2-S1 internal conversion rate of β-carotene. Journal of Physical Chemistry A 1998, 102, (26), 5049- 5058.

79. Polivka, T.; Herek, J. L.; Zigmantas, D.; Akerlund, H. E.; Sundstrom, V., Direct observation of the (forbidden) S1 state in carotenoids. Proceedings of the National Academy of Sciences of the United States of America 1999, 96, (9), 4914-4917.

1 + 1 - 1 - 80. Sashima, T.; Koyama, Y.; Yamada, T.; Hashimoto, H., The Bu , Bu , and 2 Ag energies of crystalline lycopene, β-carotene, and mini-9-β-carotene as determined by 1 + resonance-Raman excitation profiles: Dependence of the Bu state energy on the conjugation length. Journal of Physical Chemistry B 2000, 104, (20), 5011-5019.

81. Billsten, H. H.; Pan, J. X.; Sinha, S.; Pascher, T.; Sundstrom, V.; Polivka, T., Excited-state processes in the carotenoid zeaxanthin after excess energy excitation. Journal of Physical Chemistry A 2005, 109, (31), 6852-6859. 20

82. Polivka, T.; Sundstrom, V., Dark excited states of carotenoids: Consensus and controversy. Chemical Physics Letters 2009, 477, (1-3), 1-11.

83. Hashimoto, H.; Koyama, Y.; Hirata, Y.; Mataga, N., S1 and T1 species of β- carotene generated by direct photoexcitation from the all-trans, 9-cis, 13-cis, and 15-cis isomers as revealed by picosecond transient absorption and transient Raman spectroscopies. Journal of Physical Chemistry 1991, 95, (8), 3072-3076.

84. Nielsen, B. R.; Mortensen, A.; Jørgensen, K.; Skibsted, L. H., Singlet versus triplet reactivity in photodegradation of C-40 carotenoids. Journal of Agricultural and Food Chemistry 1996, 44, (8), 2106-2113.

85. Lambert, C.; Redmond, R. W., Triplet energy level of β-carotene. Chemical Physics Letters 1994, 228, (4-5), 495-498.

86. Marston, G.; Truscott, T. G.; Wayne, R. P., Phosphorescence of β-carotene. Journal of the Chemical Society, Faraday Transactions 1995, 91, (22), 4059-4061.

87. Christensen, R. L., The electronics states of carotenoids. In The Photochemistry of the Carotenoids, Frank, H. A.; Young, A. J.; Britton, G.; Cogdell, R. J., Eds. Kluwer Academic Publishers: Dordecht, 1999; Vol. 8, pp 137-157.

88. Nielsen, B. R.; Jorgensen, K.; Skibsted, L. H., Triplet-triplet extinction coefficients, rate constants of triplet decay and rate constants of anthracene triplet sensitization by laser flash photolysis of astaxanthin, β-carotene, canthaxanthin and zeaxanthin in deaerated toluene at 298 K. Journal of Photochemistry and Photobiology a- Chemistry 1998, 112, (2-3), 127-133.

89. Burke, M.; Land, E. J.; McGarvey, D. J.; Truscott, T. G., Carotenoid triplet state lifetimes. Journal of Photochemistry and Photobiology B-Biology 2000, 59, (1-3), 132- 138.

90. Dreuw, A., Influence of geometry relaxation on the energies of the S-1 and S-2 states of violaxanthin, zeaxanthin, and lutein. Journal of Physical Chemistry A 2006, 110, (13), 4592-4599.

91. Kleinschmidt, M.; Marian, C. M.; Waletzke, M.; Grimme, S., Parallel multireference configuration interaction calculations on mini-β-carotenes and β-carotene. Journal of Chemical Physics 2009, 130, (4).

92. Takahashi, O.; Watanabe, M.; Kikuchi, O., Structure of the T-1-state wave function of linear polyenes. International Journal of Quantum Chemistry 1998, 67, (2), 101-106.

93. Takahashi, O.; Watanabe, M.; Kikuchi, O., The 'triplet-excited region' in linear polyenes: Real or artifactual? Journal of Molecular Structure-Theochem 1999, 469, 121- 125. 21

94. Schulten, K.; Ohmine, I.; Karplus, M., Correlation effects in the spectra of polyenes. The Journal of Chemical Physics 1976, 64, (11), 4422-4441.

95. Kasha, M., Energy transfer mechanisms and the molecular exciton model for molecular aggregates. Radiation Research 1963, 20, (1), 55-70.

96. Zsila, F.; Deli, J.; Simonyi, M., Color and chirality: carotenoid self-assemblies in flower petals. Planta 2001, 213, (6), 937-942.

97. Ilagan, R. P.; Christensen, R. L.; Chapp, T. W.; Gibson, G. N.; Pascher, T.; Polívka, T.; Frank, H. A., Femtosecond time-resolved absorption spectroscopy of astaxanthin in solution and in alpha-crustacyanin. Journal of Physical Chemistry A 2005, 109, (14), 3120-3127.

98. Bartalucci, G.; Coppin, J.; Fisher, S.; Hall, G.; Helliwell, J. R.; Helliwell, M.; Liaaen-Jensen, S., Unravelling the chemical basis of the bathochromic shift in the lobster carapace; new crystal structures of unbound astaxanthin, canthaxanthin and zeaxanthin. Acta Crystallographica Section B-Structural Science 2007, 63, 328-337.

99. Salares, V. R.; Young, N. M.; Carey, P. R.; Bernstein, H. J., Excited-state (exciton) interactions in polyene aggregates - Resonance Raman and absorption spectroscopic evidence. Journal of Raman Spectroscopy 1977, 6, (6), 282-288.

100. Takagi, S.; Nakano, S.; Kameyama, K.; Takagi, T., Biochemical studies on carotene. 17. Lutein dispersed in acetone aqueous solution - Spectroscopic analysis and electron-microscope observation. Agricultural and Biological Chemistry 1987, 51, (6), 1561-1566.

101. Ruban, A. V.; Horton, P.; Young, A. J., Aggregation of higher-plant xanthophylls - Differences in absorption-spectra and in the dependency on solvent polarity. Journal of Photochemistry and Photobiology B-Biology 1993, 21, (2-3), 229-234.

102. Zsila, F.; Bikadi, Z.; Deli, J.; Simonyi, M., Chiral detection of carotenoid assemblies. Chirality 2001, 13, (8), 446-453.

103. Zsila, F.; Bikadi, Z.; Keresztes, Z.; Deli, J.; Simonyi, M., Investigation of the self- organization of lutein and lutein diacetate by electronic absorption, circular dichroism spectroscopy, and atomic force microscopy. Journal of Physical Chemistry B 2001, 105, (39), 9413-9421.

104. Billsten, H. H.; Sundstrom, V.; Polivka, T., Self-assembled aggregates of the carotenoid zeaxanthin: time-resolved study of excited states. Journal of Physical Chemistry A 2005, 109, (8), 1521-1529.

105. Simonyi, M.; Bikadi, Z.; Zsila, F.; Deli, J., Supramolecular exciton chirality of carotenoid aggregates. Chirality 2003, 15, (8), 680-698. 22

106. Spano, F. C., Analysis of the UV/Vis and CD spectral line shapes of carotenoid assemblies: spectral signatures of chiral H-aggregates. Journal of the American Chemical Society 2009, 131, (12), 4267-4278.

107. Köhn, S.; Kolbe, H.; Korger, M.; Köpsel, C.; Mayer, B.; Auweter, H.; E. Lüddecke; Bettermann, H.; Martin, H.-D., Aggregation and interface behaviour of carotenoids. In Carotenoids, Britton, G.; Pfander, H., Eds. Birkhäuser Basel: 2008; Vol. 4, pp 53-98.

108. Merrifield, R. E.; Avakian, P.; Groff, R. P., Fission of singlet excitons into pairs of triplet excitons in tetracene crystals. Chemical Physics Letters 1969, 3, (6), 386-388.

109. Klein, G.; Voltz, R.; Schott, M., Magnetic-field effect on prompt fluorescence in anthracene - Evidence for singlet exciton fission. Chemical Physics Letters 1972, 16, (2), 340-344.

110. Muller, A. M.; Avlasevich, Y. S.; Schoeller, W. W.; Mullen, K.; Bardeen, C. J., Exciton fission and fusion in bis(tetracene) molecules with different covalent linker structures. Journal of the American Chemical Society 2007, 129, (46), 14240-14250.

111. Muller, A. M.; Avlasevich, Y. S.; Mullen, K.; Bardeen, C. J., Evidence for exciton fission and fusion in a covalently linked tetracene dimer. Chemical Physics Letters 2006, 421, (4-6), 518-522.

112. Megerle, U.; Pugliesi, I.; Schriever, C.; Sailer, C. F.; Riedle, E., Sub-50 fs broadband absorption spectroscopy with tunable excitation: putting the analysis of ultrafast molecular dynamics on solid ground. Applied Physics B-Lasers and Optics 2009, 96, (2-3), 215-231.

113. Gallmann, L.; Herrmann, J.; Locher, R.; Sabbar, M.; Ludwig, A.; Lucchini, M.; Keller, U., Resolving intra-atomic electron dynamics with attosecond transient absorption spectroscopy. Molecular Physics 2013, 111, (14-15), 2243-2250.

114. Wirth, A. Attosecond transient absorption spectroscopy. Ludwig Maximilians University Munich, Munich, 2011.

115. Hamaguchi, H.; Gustafson, T. L., Ultrafast time-resolved spontaneous and coherent Raman spectroscopy - The structure and dynamics of photogenerated transition species. Annual Review of Physical Chemistry 1994, 45, 593-622.

116. Bell, S. E. J., Time-resolved resonance Raman spectroscopy. Analyst 1996, 121, (11), 107-120.

117. Myers, A. B.; Mathies, R. A., Resonance Raman intensities: A probe of excited- state structure and dynamics. In Biological Applications of Raman Spectroscopy, John Wiley & Sons: 1987; Vol. 2, pp 1-58. 23

118. Kukura, P.; McCamant, D. W.; Mathies, R. A., Femtosecond stimulated Raman spectroscopy. Annual Review of Physical Chemistry 2007, 58, 461-488.

119. Rao, A.; Chow, P. C. Y.; Gélinas, S.; Schlenker, C. W.; Li, C.-Z.; Yip, H.-L.; Jen, A. K.-Y.; Ginger, D. S.; Friend, R. H., The role of spin in the kinetic control of recombination in organic photovoltaics. Nature 2013, 500, 435-439.

120. Perepichka, I. F.; Perepichka, D. F., Hand book of thiophene-based Materials. John Wiley & Sons Ltd.: West Sussex, 2009; Vol. I, II.

121. Beljonne, D.; Cornil, J.; Friend, R. H.; Janssen, R. A. J.; Bredas, J. L., Influence of chain length and derivatization on the lowest singlet and triplet states and intersystem crossing in oligothiophenes. Journal of the American Chemical Society 1996, 118, (27), 6453-6461.

122. Becker, R. S.; deMelo, J. S.; Macanita, A. L.; Elisei, F., Comprehensive evaluation of the absorption, photophysical, energy transfer, structural, and theoretical properties of alpha-oligothiophenes with one to seven rings. Journal of Physical Chemistry 1996, 100, (48), 18683-18695.

Chapter 2 Characterization of Carotenoid Aggregates by Steady-State Optical Spectroscopy

2.1. Introduction

The self-assembly of organic molecular chromophores can lead to new photophysical properties that are not available to the monomer. The photophysics that emerge upon coupling of various chromophores have an impact on applications ranging from photographic science1, 2 to solar energy conversion.3 In extended systems such as aggregates and thin films, it is well-known that subtle changes in preparation conditions can lead to significantly different packing arrangements, and therefore intermolecular interactions, between the molecular constituents. Changes in absorption, emission, and other properties caused by exciton coupling within aggregates have been investigated for numerous chromophores, including cyanines or merocyanines,4-6 conjugated oligomers,7-9 xanthenes,10 acenes,11 rylene diimides,12-16 porphyrins/chlorins,3, 17 and carotenoids.18

Reviews that encompass several classes of dyes, or self-assembled systems with mixed chromophores are available.19, 20

One photophysical mechanism that generally requires the proximity of two chromophores is singlet fission. In this mechanism, a chromophore is photoexcited to its singlet excited state, and subsequently partitions its energy over two neighboring chromophores that are both left in triplet excited states. The field of singlet fission has undergone a revival in recent years, in large part because of potential benefit to solar energy conversion.21, 22 An important requirement for efficient fission is that the combined electronic energy of the two triplet excited states be less than that of the parent singlet excited state.23 As expected, the orientation and coupling strength between

25 chromophores is also important. Theoretical studies have provided some guidance and predictions on the optimal coupling for various classes of chromophores.22, 24 However, there is a shortage of experimental studies that probe the link between fission dynamics and coupling. Ideally, the coupling strength and orientation among a group of chromophores would be varied, and correlated with the yield or rate of triplet excited- state formation. Such an approach has been successfully pursued with covalent dimers of tetracene, where low yields of singlet fission have been found for some configurations. 25,

Our aim has been to correlate the coupling between carotenoid molecules, with the dynamics of singlet fission.27, 28 The chromophore zeaxanthin was selected because its electronic energy levels are favorable for the production of triplet excited states via fission, as well as its propensity to form aggregates. We recently reported a high yield of triplet excited states by singlet fission within an aggregate of zeaxanthin.27 The aggregate of this study was denoted as J1. It has a steady-state absorption spectrum that is not strongly red- or blue-shifted relative to the monomer, and therefore the coupling between chromophores is classified as weak or intermediate.29 We have also prepared and investigated two other aggregates of zeaxanthin which we denote J2 and H. The J2 aggregate has an absorption spectrum similar to that of a zeaxanthin aggregate previously reported and labeled as a J-aggregate because of an overall red-shifted absorption spectrum relative to the monomer.30 Our H-aggregate of zeaxanthin is the same or similar to other strongly coupled blue-shifted aggregates that have been reported previously for zeaxanthin.30, 31

Optical tools are important for the characterization of aggregates because these samples generally are not amenable to direct structural probes such as x-ray crystallography. We report here the steady-state optical spectroscopy of zeaxanthin aggregates as background characterization for time-resolved pump-probe absorption and

Raman experiments. The methods include absorption and temperature-dependent absorption, continuous-wave (CW) Raman, fluorescence, and circular dichroism. Several findings are discussed. First, the absorption spectra are easily distinguished, and indicate that the aggregates are stable at room temperature for at least several hours. The J1 form is convertible to the J2 upon heating. Second, all three aggregates exhibit similar fluorescence spectra that are red-shifted relative to the monomer. Additionally, the fluorescence quantum yields are found to be 5- to 30-fold lower than that of the monomer, in contrast to a prior report.32 Although the H-aggregate is dominated by strongly- coupled chromophores, the similarity in the emission spectra of the aggregates suggests that weakly-coupled chromophores co-exist as a sub-population. Third, our analysis of both the absorption and emission spectra lead to the conclusion that all of the aggregates are best characterized as H-aggregates, and that the distinguishing factor is largely the strength, rather than type, of coupling between the constituents. This conclusion agrees with recent theoretical work on aggregates of a chromophore that is isomeric with zeaxanthin, namely lutein.33 Finally, resonance Raman spectra reveal that the structures of the chromophores are very similar in the aggregates and monomer; however subtle differences in the vibrational spectra reveal slight flattening of the carotenoid in the aggregates.

2.2 . Experimental

Solvents. Organic solvents (Fisher or Sigma-Aldrich) were spectroscopic-grade and free of stabilizers, or denaturants in the case of ethanol. All were used as received, except tetrahydrofuran (THF), which was passed through an alumina column to remove residual peroxides. Water was deionized with resistivity ≥18 MΩ (Barnstead, model

D11911).

Zeaxanthin. All-trans 3R, 3'R-zeaxanthin was obtained from DSM Nutritional

Products as a 20 % (minimum w/w) suspension in corn oil with 1% w/w (±) α-tocopherol.

The sample was stored at -80 °C until use. The raw material was washed over filter paper

(Whatman, Grade 50) with excess hexanes. The solid zeaxanthin was then dissolved in a minimum quantity of acetone at 40 °C, and crystallized at -20 °C over an ~8 hour period.

The dark red crystals were filtered, re-dissolved in acetone and dried by rotary evaporation. The latter step yielded a thin film of zeaxanthin, which facilitated the subsequent dissolution in ethanol (EtOH). The sample was purified by forming aggregates via the reprecipitation method. Water was added 80% (v/v) to the zeaxanthin/EtOH solution, and the zeaxanthin aggregates were collected by passing the solution through a hydrophilic PTFE filter with 0.1 μm pore size (Millipore, Type JVWP).

The sample was redissolved in acetone and dried under flowing N2. The purity of the zeaxanthin was confirmed as described previously.27 The aggregation steps were repeated as needed to obtain consistent preparation of samples for spectroscopic study

(described below).

Preparation of Aggregates. Three kinds of zeaxanthin aggregates were formed by selecting the conditions for a final reprecipitation step. The H-aggregate (abbreviated as H) formed upon adding 8 ml water to 2 ml of 50 μM zeaxanthin in EtOH. The J1-

28 aggregate (J1) resulted from adding 9 ml water to 1 ml of 100 μM zeaxanthin in THF.

The J2-aggregate (J2) formed with slow addition of 8 ml water to 2 ml of 50 μM zeaxanthin in 1,4-dioxane. In all cases, the mixing of water and zeaxanthin/organic solutions was done while sonicating (Branson, model 3510) and agitating the receiving vial. Samples were sonicated for an additional 10 minutes after mixing.

Transmission Electron Microscopy and Dynamic Light Scattering.

Approximately 15 μL of aggregate solution was applied to a copper grid coated with lacey carbon and settled for 4 min at 4 °C and 100% relative humidity. One drop of 1% uranyl acetate solution was applied to the sample for negative staining to enhance the contrast of the image. Transmission electron microscope (TEM) images were acquired with a FEI Tecnai G2 Sphera microscope with acceleration voltage of 200 KV. The particle size (hydrodynamic diameter) of the aggregates was determined with a dynamic light scattering (DLS) analyzer (JY Horiba, LB-550).

UV−vis Absorption and Circular Dichroism. UV−vis absorption spectra of the samples were measured with a scanning spectrometer (Shimadzu, UV-3600). Samples were held in a quartz cuvette with 1 cm path length. The relative extinction coefficient

(on a per-monomer basis) of each aggregate sample was determined by filtering zeaxanthin aggregates from each solution, followed by quantitative analysis of the residue by dissolution in ethanol and UV−vis absorption measurements. The procedure yielded relative extinction coefficients that were repeatable within 10% and avoided systematic errors that could result from a propensity of the aggregates to collect at interfaces or settle out of solution. For temperature-controlled experiments, a circulator

(PolyScience, 9112A) maintained the temperature of the cuvette to within ±1 °C.

Samples for measurements at elevated temperatures were held in a septum-topped cuvette and sparged with N2 prior to heating. After heating and cooling, the sample was dried and redissolved in EtOH, and a UV−vis spectrum was acquired to assess the extent of isomerization. CD spectra were collected with an AVIV model 215 CD spectrometer.

Samples were held at room temperature in a quartz cuvette with 1 mm path length. The raw signal was converted to molar CD as described in the Appendix.

Resonance Raman. The 488.0 nm excitation beam was generated by an

Argon/Krypton ion laser (Laser Innovations, Innova-70C). The sample solution was flowed through a horizontal capillary with inner dimensions 2×2 mm. The excitation power was 0.3 mW. Scattered light was collected at 90 degrees relative to the excitation beam with an F/1.2 camera lens (Canon FD, focal length 85 mm), and focused onto the spectrograph entrance slit with an F/4 achromatic lens. Photons with polarizations parallel and perpendicular to the horizontally-polarized excitation beam were collected.

A 488.0 nm edge filter (Semrock, RazorEdge) was inserted after the collection and focusing lenses to reject Rayleigh scattering. A 320 mm focal length spectrograph equipped with a 2400 groove/mm holographic grating and open-electrode CCD detector

(Horiba Jobin-Yvon, iHR-320 and Synapse) were used to disperse and detect the Raman light. The entrance slit width was 50 μm, and spectral resolution ~1.7 cm-1. Raman shifts were calibrated by acquiring reference spectra of acetonitrile and toluene which have known peak assignments.34 The accuracy of the peak positions was ±1 cm-1.

Solvent peaks were removed from the raw spectra by subtraction of a solvent-only spectrum. Attenuation of Raman scattering caused by absorption of the sample was corrected with an appropriate transmission spectrum. The Raman spectra were also

30 divided by the instrument response of the detection system that was determined by measuring the spectrum of a calibrated white light source (Optronic Laboratory, OL-

220C). In a final step, the Raman spectra were smoothed with a Savitzky-Golay algorithm with five-point window.

Calculations. Geometry optimization and normal modes of vibration were calculated with density functional theory (DFT) using Gaussian 09W software.35 The

B3LYP functional and 6-31G(d) basis set were employed. The initial geometry of the molecule was taken from crystallographic data of bis[menthyl] carbonate zeaxanthin,36 but with one modification. The torsion angles of the ionone rings around the C6-C7 and

C6′-C7′ bonds are different from one another in the crystal, whereas the angles were modified to a single value prior to the calculation to avoid optimization to an asymmetric minimum. The DFT calculation yielded normal modes with real eigenvalues. A scaling factor of 0.961 yielded the best agreement of calculated and experimental Raman frequencies of the strongest bands.

Fluorescence Spectroscopy. The fluorescence from the aggregates was too weak, and the scattering too strong for reliable measurement with a commercial fluorimeter.

Therefore, the excitation/detection setup for Raman spectroscopy was employed, with changes as noted here. Excitation wavelengths included 457.9, 488.0, and 514.5 nm from the argon/krypton ion laser and 404.3 nm from a blue laser pointer. The power at the sample point was <1.0 mW for each excitation wavelength. Samples were held in a 1 × 1 cm cuvette and had an optical density <0.3/cm at the excitation wavelength. No photodegradation was found by UV −vis absorption at the end of the fluorescence measurements. A polarization scrambler (Opto-Sigma) was placed in front of the

31 spectrograph slit for all measurements. A dichroic polarizer mounted in a rotation stage was positioned between the F/4 focusing lens and scrambler for measurements of the depolarization ratio (see below). The 320 mm spectrograph employed a 600 gr/mm holographic grating, and the slits were adjusted for a spectral resolution of 0.5 nm. Given the low fluorescence emission from the samples, thorough checks for background fluorescence from the optics train were performed. The tests included exchange of antireflection coated optics with those that were uncoated, exchange of various edge or long-pass filters, and checks of the background fluorescence from the optical detection system by replacing the aggregate samples with highly scattering samples of silica.

The fluorescence emission spectra were corrected for the instrument response as described above. The effect of the sample absorption was corrected using Beer’s law to account for two sources of attenuation of the fluorescence emission. First, the raw spectra were divided by a constant (less than 1) that compensated for absorption of the excitation beam over the ~2 mm excitation pathlength. Second, the spectra were divided by a spectral transmission function that compensated for absorption of the emission from the sample. The relative amounts of the two types of attenuation depend upon specific excitation and emission pathlengths of the 90-degree scattering setup, and the correction was optimized by monitoring solvent bands of the carotenoid solutions, relative to the same bands of pure solvents. Fluorescence spectra presented on a wavenumber axis were converted from a constant wavelength band-pass to constant wavenumber band-pass by multiplying the emission spectrum by 2. After these corrections, the emission spectra were fitted to multiple Gaussian curves to separate the broad fluorescence emission from the narrow Raman features.

Determination of Fluorescence Quantum Yields. The fluorescence quantum yields ΦF of zeaxanthin monomer and aggregates were determined by comparing the intensities of the fluorescence to Raman bands of an external reference with known

Raman scattering cross-section. The derivation outlined in the Supporting Information is modified from the initial report of this technique.37 We include polarization in our treatment, and derive an equation that is analogous to one reported in prior studies,38, 39

Eq. 2.1

where NAv is Avogadro’s number, ε is the decadic molar absorption coefficient, and

is the absolute differential scattering cross-section of the reference Raman

scatterer. Our reference was EtOH solvent, which was flushed through the same square capillary tubing following measurements of carotenoid monomer in EtOH or carotenoid aggregates in binary water/organic solvent mixtures. The concentrations cR and c refer to the Raman-scattering reference and to the fluorophore, respectively. IF(Ω) is the fluorescence intensity, which is quantified for each recorded emission spectrum by integrating over the broad bands along an energy axis. Similarly, IR(Ω) is the intensity of

Raman scattering from the reference solvent, integrated over the three prominent C−H stretching bands. The Raman and fluorescence light was collected over the same solid angle (Ω) with the same detection system, which allows cancellation of the detection efficiency and other factors. Scattering and fluorescence with polarizations that are both parallel and perpendicular to the excitation were collected, and spectra were corrected for selfabsorption and converted to constant-energy band-pass as described above. The reference value we employ for of ethanol is 35.0 × 10−30 cm2/molecule-sr with

488.0 nm excitation, as described in the Appendix. The final term in Eq. 2.1 accounts for polarization effects, as quantified by the depolarization ratio ρ = I⊥ /I||. The measured depolarization ratios for the aggregates and monomer deviated 15% or less from 0.33, and this value was used in all calculations of the fluorescence QYs. (See the Results section.)

Fluorescence Excitation Spectroscopy. Fluorescence excitation spectra were measured with a xenon arc lamp (Perkin-Elmer Cermax, 175W) in combination with a single-stage monochromator (PTI, 200 mm focal length; with holographic grating from

Edmund Optics). Short-pass filters were placed between the monochromator output and the sample cuvette to avoid artifacts from residual light scattering from the aggregates.

The excitation power ranged from 0.02 to 0.17 mW at a spectral band-pass of 5.0 nm, across the 370−540 nm range. The detection setup was as described above. All signals were corrected for variation in the lamp output, as well as attenuation of excitation power by absorption from zeaxanthin.

2.3. Results

Transmission electron microscopy. TEM images of the three types of zeaxanthin aggregates are shown in Figure 2.1. The aggregates have rod-like shapes with differing dimensions. Representative H-aggregates are ~50 nm in diameter and 200-300 nm in length, whereas J1-aggregates are typically 20–30 nm in diameter and 500-600 nm in length. The J2-aggregates are much larger, with diameters ~500 nm, and lengths that can exceed 2 µm. The average hydrodynamic diameters inferred from DLS measurement are 200, 100, and 2000 nm for H, J1, and J2 respectively, which are qualitatively consistent with the dimensions determined from the TEM images.

Figure 2.1. TEM images of zeaxanthin J1, J2 and H-aggregates.

UV-Vis absorption spectroscopy. The zeaxanthin aggregates are distinguished from the monomer by their absorption spectra at room temperature (Figure 2.2A). The spectra are stable for at least several hours. The aggregate solutions showed no sign of residual monomer, as verified by UV-Vis analysis of the solution passed through a 0.1

µm filter. The peak extinction coefficients follow the ratio monomer(1.00): J1(0.77):

J2(0.65):H(0.79) where all spectra are reported on the basis of equivalent concentration of monomer. The areas under the same spectra, which reflect the oscillator strengths, have a ratio monomer(1.00):: J1(0.94): J2(0.94): H(0.70), when integrated on a wavenumber axis from 29000 to 14300 cm-1. The integrals of the extinction coefficient divided by energy, ε(ω)/ω, which are proportional to the square of the transition dipole moments, follow the ratio monomer(1.00): J1(1.00): J2(0.99): H(0.65), over the same integration limits.

Figure 2.2. (A) UV-vis absorption spectra and (B) circular dichroism spectra of zeaxanthin monomer in EtOH, H-aggregate in 80:20 (v/v) H2O/EtOH, J1-aggregate in 90:10 (v/v) H2O/THF, and J2-aggregate in 80:20 (v/v) H2O/1,4 dioxane.

We now summarize the vibronic features of the zeaxanthin monomer and aggregates. Maxima in the monomer spectrum at 480 and 452 nm are attributed to 0→0 and 0→1 transitions, respectively. These two peaks and two shoulders can be fitted with four Gaussian functions (Figure 2.3). The three lowest energy bands are separated by

~1300 cm-1, and the fourth band assigned to the 0→3 transition is separated 1500 ± 200 cm-1 from the 0→2 band. At higher energy from the visible absorption is a weak band at

~335 nm, which is a signature of cis-isomers. The small magnitude of this band reflects a minor population of cis-isomers in our zeaxanthin samples. The transition in the ultraviolet (S0→Sn) has a maximum at 274 nm.

The spectrum of J1 has three peak maxima located at 513, 474, and 450 nm. The positions are within 1-2 nm of those reported for aggregates of zeaxanthin diacetate in 3:1 water:ethanol solution.40 The three maxima and two shoulders in our spectrum can be fitted with five Gaussian functions. The intervals between centers of neighboring bands, from low to high energy, are 1470, 1260, 1190, and ~1250 cm-1. The Gaussian decomposition reveals that the 0→0 band of the J1 aggregate is red-shifted by 1260 cm-1 relative to the corresponding band of the monomer. The UV transition of J1 is maximal at 288 nm, which is red-shifted 14 nm (~1800 cm-1) relative to that of the monomer.

The J2-aggregate of our study has an absorption spectrum that is similar to a previously classified J-aggregate of zeaxanthin.18, 30 The most prominent maxima of the

J2 spectrum are located at 522, 482, and 457 nm, which are red-shifted several hundred wavenumbers relative to corresponding peaks of J1. Four Gaussian functions and one combined half-Gaussian/half-Lorentzian function yield a good fit to most of the J2 absorption spectrum (Figure 2.3, Table 2.1). The centers of neighboring bands are separated, from low to high energy, by 1590, 1440, 1540, and ~1670 cm-1. These intervals are greater than those of J1. The wing of the low-energy band extends much further to the red compared with J1. The wing was fitted with a half-Lorentzian with half-width 1135 cm-1. The UV transition for J2 is maximal at ~294 nm, which is red- shifted 6 nm (~700 cm-1) relative to that of J1.

The overall spectral profile of the H-aggregate prepared in our work, and maximum at 388 nm, closely match a prior report of zeaxanthin aggregates in 4:1 water:EtOH.30 Our spectrum is well-fitted with five Gaussian functions. The center position of the strongest band is blue-shifted approximately 5000 cm-1 relative to the peak

37 position of the 0→0 absorption band of the monomer. Several other bands in the spectral range 440-520 nm are also evident. The magnitude of these bands increases with slower addition of water to the ethanol solution of zeaxanthin (Figure A2.1 in the Appendix).

The separations between central positions of the fitted bands, from low to high energy, are approximately 1490, 1500, 1580, and 1900 cm-1 (Figure 2.3, Table 2.1). The UV- band for the H-aggregate is peaked at ~280 nm, which is red-shifted ~6 nm (~780 cm-1) relative to the monomer.

Table 2.1. Fitted absorbance and emission peaks, and fluorescence quantum yields (QY) of zeaxanthin aggregates. Absorbance Peak Positions (cm-1) Emission Peak Positions (cm-1) FL. QYb Peak Peak Peak Peak Peak Peak Peak Peak Peak 488.0 nm 5 4 3 2 1 1 2 3 4a excitation Mon N/A 24900c 23410 22080 20780 20230 19040 17840 16700 12 x 10-5 J1 24690c 23440 22250 20990 19520 19020 17540 16350 15460 2.5 x 10-5 J2 25300c 23630 22090 20650 19060 18730 17400 16310 15290 1.9 x 10-5 H 25820 23920 22340 20840 19350 19140 17530 16430 15380 0.4 x 10-5 a The position and existence of Peak 4 in the emission spectra is uncertain. See text for details. b Estimated uncertainties in the quantum yields are 10% for the monomer and 20% for the aggregates. c The position of these peaks could be subject to ± 200 cm-1 variation.

Figure 2.3. Fits to absorption and emission spectra of the zeaxanthin monomer and aggregates. Experimental absorption spectra are solid black lines; fitted Gaussian components are blue dashed lines; and the sum of the components is a blue dotted line. A split half-Gaussian/half-Lorentzian function is used to fit the low-energy wing of J2. Solid red lines depict the broad fluorescence emission profiles, which result from the sum of least-squares Gaussian fits to the experimental 488.0 nm total emission spectra. The Gaussian components are shown as dashed lines; narrow Raman features are separated in this procedure. The vertical black lines mark the estimated position of the electronic origin, E00. The positions of the bands are listed in Table 2.1.

The UV-Vis spectra of H, J1, and J2 aggregates were monitored as the solutions were gradually heated from room temperature to near the solvent boiling point, and cooled back to room temperature (Figure A2A and A2B in the Appendix). The strong H- aggregate absorption band showed only a small ~2 nm (130 cm-1) blue-shift that remained after cooling. The bands on the low-energy side of the main peak became more distinct upon heating/cooling, and exhibited blue-shifts of ~400-600 cm-1. The changes in the UV band were the most significant in terms of energy; an irreversible ~1500 cm-1 hypsochromic shift of the peak position from 278 nm to 267 nm was found.

The bands of the J1-aggregate change significantly upon heating. The three peak maxima undergo a slight blue-shift when first heated to ~60 °C, but the final positions are

200-400 cm-1 red-shifted relative to their initial positions after heating to 90 °C and subsequent cooling. The maxima for the annealed J1-aggregate have final positions at

519, 482, and 457 nm, which are nearly identical to the peak positions for unheated J2.

Furthermore, the spectrum of J1 after heating and cooling resembles that of the J2- aggregate, with the overall maximum at 482 nm and extended red absorption wing. Upon heating the J2-aggregate solution, the net changes in peak position (~3 nm) are smaller than those observed for the J1-aggregate.

When heated, the zeaxanthin within the aggregate is less likely to isomerize than the monomer in solution, as shown in Figure A2.3 in the Appendix. The dissolution of previously heated J1, J2, and H aggregates revealed weaker bands at ~335 nm, and therefore less cis-isomer formation, than the heated monomeric zeaxanthin in ethanol.

Circular dichroism (CD). Strong optical activity was observed upon aggregation as shown by CD spectroscopy in Figure 2.2B. The monomer shows no CD signal in the

350 - 650 nm spectral region (data not shown), whereas the H-aggregate shows a bisignate couplet that is characteristic of negative and positive Cotton effects. Additional structure is seen for the J1 and J2-aggregates. The amplitude of the double Cotton effect decreased nearly 10-fold for the H-aggregate, and 4- to 5-fold for J1 and J2 aggregates, when the solution was aged for several hours in ambient conditions. The UV-Vis spectra remained constant over the same time period. Similar behavior was noted previously.40

Emission spectroscopy. The emission spectra of zeaxanthin aggregates and monomer with 488.0 nm excitation are shown in Figure 2.4. The sharp Raman features of fundamental, combination, and overtone bands of zeaxanthin are prominent in all spectra. The fundamentals are described further below. The fluorescence spectra of the monomer and aggregates were fitted with a minimum number of Gaussian curves (Figure

2.3 and Table 2.1). The emission of the monomer with 488.0 nm excitation has an appearance that resembles a mirror image of the absorption spectrum, in agreement with

41-46 prior S2→S0 emission spectra of monomeric -carotene and related xanthophylls.

The emission maximum of the zeaxanthin spectrum in ethanol is located at ~527 nm.

The main features of the spectrum can be fitted with three Gaussian functions. The dominant band centered at 19040 cm-1 corresponds to the 0→1 vibronic transition, and two bands are located ~1200 cm-1 on either side of the main one. A fourth band at lower energy (~16700 cm-1) is possible, however the lack of an inflection point in the experimental spectrum makes the position and existence of this peak uncertain. The

-1 electronic origin E00 is estimated to be 20500 cm , as derived from the midpoint of the peak maxima for the 0-0 bands of absorption and emission. The E00 value found in our

41 work is similar to 20360 cm-1 and 20000 cm-1 reported for lutein and zeaxanthin, respectively, at 77 K in an EPA glass.46

Figure 2.4. Emission spectra of monomeric zeaxanthin (1.0 µM in EtOH) and aggregates, with 488.0 nm excitation. The experimental spectra are shown as thin solid lines. Sharp features are Raman fundamentals, overtones, and combination bands from zeaxanthin. Spectra of the aggregates are normalized for 1.0 µM equivalent monomer concentration, and scaled as indicated in the figure. The broad fluorescence component of each experimental spectrum is fitted with four Gaussian functions. (See Figure 2.3) The sum of the Gaussian components is shown as a bold solid line. An asterisk (*) denotes residual Raman scattering from water that remains after subtraction of the solvent-only spectrum.

The emission spectra of all three aggregates are similar to one another, when excited at 488.0 nm (Figures 2.3 and 2.4). The spectra all have a maximum in the range

570-585 nm, which is significantly red-shifted relative to the peak of the monomer spectrum. The values measured for ρ at this excitation wavelength were J1(0.36), J2(0.28)

42 and H(0.38). The depolarization ratios of the aggregates were all within 20% of the ρ measured for the monomer (0.34). The values for the aggregates were not checked for repeatability from sample to sample. Therefore, the significance of the variation from the theoretical limit of 0.33 expected for ultrafast emission and parallel transition dipoles for absorption and emission, is not yet established. Given this uncertainty, a value of ρ=0.33 was used for the calculation of quantum yields. The dependence of Equation 1 on ρ indicates that even if the measured variations from 0.33 were repeatable, the use of the experimental ratios would have only a minor effect on the calculated fluorescence quantum yields for H and J2.

The aggregate spectra can be decomposed into four Gaussian components. The bluest emission in the region of ~20,000 cm-1 is not a clear band and is excluded from the fits reported here. Three bands at highest energy can be assigned for each of the aggregates to vibronic transitions 0→0, 0→1, and 0→2. The dominant band is assigned to 0→1, with central position in the range 17400-17540 cm-1. The separation between

0→0 and 0→1 bands varies from 1330 to 1610 cm-1, depending upon the aggregate, but is greater than the monomer in all cases. The separation between 0→1 and 0→2 bands is

1090-1190 cm-1, which is similar or smaller than the corresponding separation for the monomer. All peak positions for J1 and H are within 120 cm-1 of each other. The peak positions of J2 are red-shifted by 40-290 cm-1 relative to those of J1.

The J1 emission spectrum has a mirror-image quality relative to the absorption spectrum. Therefore an E00 value can be estimated from the average of the 0→0 bands of both emission and absorption, as done for the monomer. The value is 19270 cm-1, which is ~1230 cm-1 smaller than for the monomer. The red-shift is nearly the same as the shift

43 of the lowest energy absorption band (1260 cm-1). The emission from the J2 and H aggregates do not have a mirror-image characteristic, thus we do not report E00 values for these aggregates.

Excitation spectra (Figure 2.5A) were acquired to corroborate the source of fluorescence reported in Figure 2.4. The excitation spectra of the monomer and aggregates largely follow their respective absorption profiles, as expected for pure samples. However, we note that the excitation profile has some dependence upon the wavelength that is monitored, particularly for the H-aggregate. For example, when the monitoring wavelength is 580 nm rather than 630 nm, the normalized excitation profile

(not shown) is relatively higher over the range 430-520 nm than predicted from the absorption spectrum. The reason for the variation is that the fluorescence spectrum of the

H-aggregate changes with excitation wavelength (Figure 2.5B). With higher excitation energy, the spectrum broadens and the emission at the ~580 nm maximum diminishes more than the intensity at other wavelengths. Similar wavelength dependence in the emission spectra has been noted for monomeric carotenoids.43, 47-50 The broadened emission with higher energy excitation is caused by unrelaxed emission, and is discussed further below.

The fluorescence quantum yields (QY) for zeaxanthin monomer and aggregates with 488.0 nm excitation are listed in Table 2.1. A yield of 12×10-5 was determined for the monomer in EtOH. Our value is comparable to a previous reported yield of 9.6×10-5 for zeaxanthin in chloroform,43 however it is significantly greater than 4±2×10-5 recorded for zeaxanthin in EPA at room temperature.46 The yields for each aggregate were lower than the monomer: 2.5× 10-5 (J1); 1.9×10-5 (J2); and 0.4×10-5 (H). Variability in the

44 aggregate preparation led to changes as high as 20% in the QY. The numbers reported here correspond to reductions in the emission QYs of J1, J2, and H which are respectively 5-, 6-, and 30-fold decreased relative to the monomer. It was also observed that the fluorescence quantum yield increases proportionally with the amount of J-type features in the H-aggregate.

Figure 2.5. (A) Fluorescence excitation spectra of zeaxanthin monomer and aggregates. The excitation spectra are comprised of a set of points (open circles) that indicate the emission intensity at 540 nm (monomer), 605 nm (J1), 580 nm (J2), and 630 nm (H). Absorption spectra are shown as black solid lines. The set of emission points for each aggregate is scaled by a factor that is determined by a fit (least-squares) to the corresponding absorption profile. (B) Fitted emission spectra of zeaxanthin monomer and aggregates. The curves are the result of Gaussian fits to the experimental emission spectra (Figure 2.4 and Appendix) acquired with four laser excitation wavelengths, 404.3 nm (green), 457.9 nm (blue), 488.0 nm (red), and 514.5 nm (black). Each spectrum is normalized for equal area.

Raman spectroscopy. Resonance Raman spectra of zeaxanthin aggregates reveal that the frequencies of most bands are nearly unchanged relative to the corresponding bands of the monomer (Figure 2.6). For example, there are insignificant shifts in the frequency of the C-C single-bond stretch at 1157 cm-1, and the methyl rock at 1004-1006 cm-1. The structurally-sensitive hydrogen out-of-plane (HOOP) modes at ~870 cm-1 and

~965 cm-1 have similar frequencies for the monomer and the aggregates. Despite the similarities, small changes in the spectra are apparent (Figure 2.6 and A2.5 in the

Appendix). In the ethylenic region, small downshifts are observed upon aggregation, for both the strong band at ~1520 cm-1 and a weak one at ~1590 cm-1. The 1520 cm-1 normal mode consists of in-phase stretches of the C=C bonds in the center of the polyene chain, whereas the 1590 cm-1 mode consists of in-phase stretches of C=C bonds at the ends of the polyene chain and ionone ring. The extent of the downshift of the main band varies with the excitation wavelength. With 457.9 nm excitation the downshifts are 8-9 cm-1 relative to the monomer; with 488.0 nm, 3-5 cm-1; and with 568.2 nm, 4-6 cm-1. The two

J aggregates show greater downshifts for the ethylenic mode than the H aggregate.

Similar shifts for the zeaxanthin H-aggregate have been reported previously.31 The downshift of the weak band at 1590 cm-1 was similar in magnitude to that of the strong ethylenic band. Finally we note that a shoulder at ~1560 cm-1 is apparent for the aggregates, but not for the monomer.

Figure 2.6. Raman spectra with 488.0 nm excitation. (Upper spectrum): zeaxanthin J1- aggregate in 90:10 (v/v) H2O/THF. (Middle spectrum) monomer in EtOH. (Lower spectrum) Calculated vibrations, from DFT optimized structure and after scaling by a single factor to give the best match to the experimental monomer frequencies (see text). The calculated peaks are shown with a 4 cm-1 broadening factor. All spectra were normalized to equal peak heights of the ethylenic mode. The 200–570 and 1550–1650 cm-1 spectral regions are shown magnified for each spectrum. Labels for the experimental spectra refer to actual maxima of peaks, not fitted. The locations of the maxima for peak shoulders are estimated. Dashed lines guide the eye for corresponding modes among the experimental and calculated spectra. Raman spectra of J2- and Haggregates are similar to that of J1 (see Figure A2.5 in the Appendix).

Although the frequencies shifts are best characterized as minimal changes, the intensity changes between the monomer and aggregates are noteworthy. First, a 373 cm-1 band has stronger Raman activity for all three types of aggregates. Second, the intensity of a vibrational band at 1262-1284 cm-1 is significantly smaller for all the aggregates,

47 relative to the monomer. The band corresponds to a mode with calculated frequency

1269 cm-1, and consists of in-plane C-H bond deformations along the polyene chain, in combination with twists of methylene groups of the ionone rings. The intensity of the

~870 cm-1 band diminishes slightly for all aggregates, relative to the methyl rock band at

1004-1006 cm-1.

2.4. Discussion

Zeaxanthin forms three different types of nanoparticles that are stable in solution phase at room temperature. These aggregates allow us to probe the effects of interactions between chromophores on optical properties or excited-state dynamics. The exciton coupling within the nanoparticles can be similar to the interactions within thin films or crystals, which makes it worthwhile to compare the aggregates with extended solid-state systems. However, experimental measurements of nanoparticles have several advantages over those of solid-state samples. For example, the absorption or emission spectra of the nanoparticles are measured as readily as their monomeric counterparts. Moreover, the aggregates can be prepared in large batches and flowed, thus facilitating pump-probe spectroscopic measurements (time-resolved resonance Raman, and transient absorption spectroscopy) that will be reported elsewhere.

The steady-state optical spectroscopy of the three carotenoid aggregates is the focus of the present work. These spectra probe several characteristics of the aggregates.

First, the results from variable-temperature absorption experiments reflect the relative stability of the aggregates. Second, the aggregates have emission quantum yields that are significantly lower than that of the monomer. Our fluorescence quantum yields correct an earlier report that the emission is significantly higher for zeaxanthin in the aggregated

48 form, relative to the monomer.32 Third, the emission spectra are similar for all aggregates, which suggests a common exciton structure for the emitting states. Finally, the Raman vibrational data are consistent with prior work31 showing that the structure of zeaxanthin in the aggregates is similar to that in solution. Nonetheless, frequency or intensity changes for the Raman spectra of the three aggregates compared with the monomer allows some insights into subtle structural changes that occur upon self-assembly.

Formation of zeaxanthin aggregates.

Carotenoids form aggregates with significantly blue-shifted (H) and slightly red- shifted (traditionally called J) absorption bands in binary aqueous/organic solvent systems. The blue-shifted spectra are commonly observed for xanthophylls such as lutein, astaxanthin, and zeaxanthin.18, 31, 51 Prior studies have highlighted the role of hydrogen bonding among xanthophylls, such as zeaxanthin and lutein, as a factor in the formation of carotenoid H-aggregates.18, 30, 51, 52 The fact that esterification of xanthophylls, for which H-bonding is impossible, leads to the formation of the J-aggregate rather than the blue-shifted H-aggregate, is good evidence of the importance of intermolecular H- bonding.51-53 Furthermore, the addition of sodium hydroxide to the aqueous component, to pH = 10, causes exclusive formation of J-aggregates, presumably because of deprotonation of the –OH groups.30 Despite the influence of intermolecular hydrogen- bonding in forming H-aggregates, this kind of bonding is not necessary for H-aggregate formation. Examples of H-aggregate formation where intermolecular hydrogen bonding cannot be a factor have been identified, for example with the carotenes -carotene,54, 55 tetradesmethyl--carotene,56 and lycopene.57

A full explanation for the differentiation of the aggregates into several types in the present study is not clear. The topic is complex in part because nanoprecipitation is not an equilibrium phenomenon. Previous studies have highlighted the importance of non- equilibrium and kinetic effects.58 The precipitation of zeaxanthin in water/1,4 dioxane provides a simple illustration. If the 1,4-dioxane solution of zeaxanthin is slowly added to water to a solution 4:1 (v/v) water:ether, an H-aggregate forms. However, if the water is added to the ether to reach the same solvent ratio, the J2-aggregate forms. Similarly, the formation of J1- and J2-aggregates in 9:1 water:THF was controlled solely by the rate of addition of water into the 100 µM zeaxanthin/THF.

Based on the present studies of zeaxanthin, we suggest another property, aside from intermolecular H-bonding, that correlates with the outcome of the reprecipitation process. That property is the solubility of the chromophore in the organic solvent. A correlation would be reasonable, since solubility is a direct indicator of the strength of chromophore-solvent interactions, relative to chromophore-chromophore interactions. In the reprecipitation process, the replacement of the solvent molecules around each carotenoid by other carotenoids is the first step towards aggregation. We note that zeaxanthin is more soluble in THF than any other organic solvent, with levels ~20 mg/mL measured in our lab. The solubility of zeaxanthin in 1,4 dioxane is ~4 mg/mL, which was the second highest among all water-miscible solvents that we tested. Both organic ethers, and particularly THF, tend to form J-aggregates of zeaxanthin when water is added. On the other hand, we find much lower solubilities of zeaxanthin in ethanol

(~0.1 mg/mL) and acetone (0.5 mg/mL), which are solvents that form H-aggregates. We

50 suggest that the formation of weakly coupled J-aggregates (see below) is correlated with relatively favorable interactions between chromophore and the organic solvent.

Relative stability and surface-to-volume ratios of the zeaxanthin aggregates.

The temperature-dependence study offers insights into the thermodynamic properties of the aggregates. While heating causes the J1 aggregate to change irreversibly to the J2 aggregate, the J2 aggregate exhibits only slight changes that partially recover after cooling. Evidently the J2-aggregate is more stable than the J1-aggregate. The H aggregate shows small irreversible changes upon heating/cooling. The bands at lower energy from the main peak are augmented and slightly blue-shifted (see below). The annealing experiments suggest that the J2 and H aggregate consist of molecules in two stable alignments. We have not found conditions that cause interconversion of H- and

J2-aggregates of zeaxanthin. However, astaxanthin assemblies were reported to show evolution between H and J aggregates with aging or annealing.58, 59

The fraction of zeaxanthin molecules located on the surface of the aggregates can be estimated from the TEM images, and crystallographic data for β-carotene which indicates two molecules in a unit cell of size ~8Å × 9Å × 24Å.60-62 Taking J1-aggregate as an example, if a typical aggregate has a cylindrical shape with length 600 nm, and diameter 30 nm, it contains approximately ~5×105 molecules. The length of a zeaxanthin molecule is ~35 Å. Assuming an average intermolecular separation of ~4 Å between carotenoids, approximately ~4×104 monomers, or 8%, are located at the aggregate surface. Given that J1 has the smallest volume of the three aggregates, the percentage of molecules at the surface is lower for the other aggregates. The estimate indicates that the

51 vast majority of the zeaxanthin molecules are within the interior of the aggregates, and one would expect that this population dominates the absorption, Raman, and CD spectra.

Electronic transitions and exciton coupling in zeaxanthin aggregates. The electronic spectra of monomeric carotenoids with more than 9 conjugated double bonds, including β-carotene, lutein, zeaxanthin, and others are dominated by the symmetry

1 - 1 + 63 allowed S0 (1 Ag ) to S2 (1 Bu ) transition. Although the S0→S2 absorption is strongly allowed, zeaxanthin and other carotenoids have very low fluorescence quantum yields on

-4 42, 64, 65 the order of ~1×10 because of ultrafast internal conversion to the dark S1 state.

The S2→S0 emission spectrum has a mirror-image appearance in comparison with the absorption spectra, as expected for molecules emitting from the same electronic state that is excited. In addition to the S2→S0 transition, emission from the forbidden S1→S0 transition is also measurable,46 but not addressed in the present work.

Absorption spectra of H-aggregate. The changes in the zeaxanthin absorption profiles upon aggregation are largely a result of exciton-coupling, as described previously for various xanthophylls in binary water-organic solutions.18, 31, 33, 52, 66-69 The H- aggregates are characterized by a dominant absorption band that is strongly blue-shifted relative to the monomer. The main band is attributed to an allowed transition between the ground state and an in-phase exciton-coupled state. If the exciton-coupling is strong and extends over multiple chromophores, the absorption bands collapse onto a single, purely electronic transition.70 The sharp, blue-shifted absorption band of the H-aggregate indicates that the exciton is more delocalized for this species than for the J1 and J2 aggregates. The fact that the displacement between the allowed band and the 0-0 band of the monomer is ~5000 cm-1 indicates that the exciton-coupling bandwidth (~10,000 cm-1)

52 is considerably greater than the Franck-Condon bandwidth of the monomer. Therefore, spectra like those reported here for the H-aggregate are clearly in the strong-coupling regime.33, 51, 56, 70

The H-aggregate shows substantial decrease (~30%) in the integral of the extinction coefficient per molecule, relative to unassociated monomer. The loss is known as hypochromism.71, 72 In the case of an exciton-coupled system with a single transition, one would expect that the square of the transition dipole moment, which is proportional to the integral of ε(ω)/ω, should be preserved. The finding that the integral is lower for the H-aggregate than the monomer by 35% indicates that the squared transition dipole of this system is diminished. It is likely that coupling with other transitions, combined with the strong interchromophore interaction for the H aggregate are the basis for the lowered extinction coefficient. The theory of hypochromism predicts that the effect should depend the inverse cube of the distance between chromophores,71, 73 and this distance is expected to be smallest for the strongly coupled aggregate.

Although the nature of the dominant absorption of the H-aggregate is clear, the side bands in the range 420-540 nm need further explanation. An important question regarding these bands is whether they have an origin that is related to the main transition

(homogeneous model). Alternatively, the bands could reflect a separate population of molecules (inhomogeneous model). There are several arguments against the homogeneous model. First, none of the bands are reasonably assigned as an out-of-phase complementary band to the strongly allowed transition. If such a transition were observed for a strongly-coupled aggregate, it would be expected to appear much lower in energy, at approximately 14000 cm-1 when one considers a ~1200 cm-1 non-resonant dispersion

53 interaction and symmetric splitting of the allowed and disallowed transitions with respect to the monomer E00. Second, the out-of-phase transitions are unlikely to have significant oscillator strength in the case of strong intermolecular coupling.33 Two pieces of evidence support the inhomogeneous model. First, the fact that the three lowest energy bands of H have peak positions that are within 100-200 cm-1 of those of J1 strongly suggests that the physical origin of the bands is the same in both aggregate types, which therefore excludes their close association with the dominant H-band. Next, we note that variation in the mixing speed can change the relative amplitudes of the side-bands, with almost no shift in the peak position of the main H-band (Figure A2.1 in the appendix).

The independent variation of the side-bands again underscores that these transitions exist as a separate phenomenon from the strongly-allowed transition of the H-aggregate. In short, the evidence suggests that the H-aggregates have a separate population of chromophores that interact in a way that makes them similar to the J1-aggregate. This subset of chromophores could be small inclusions of J1-aggregates embedded within a matrix of H. Another possibility is that grain boundaries between subsets of the strongly interacting chromophores could give rise to the J1 absorption features.

The present finding of inhomogeneity within zeaxanthin aggregates is in agreement with previous reports for a variety of xanthophylls. For example, aggregates of zeaxanthin, violaxanthin, antheroxanthin, lutein, and other xanthophylls in binary solvents of water/ethanol or water/acetone have spectra with a strong blue-shifted absorption band with peak position less than 400 nm.31, 68 Other bands in the range 490-

520 nm are also evident. The relative magnitude of the blue-shifted and red-shifted bands generally depend upon the specific xanthophyll.31, 68 Similary, violaxanthin

54 aggregates with varying absorption spectra were prepared with two different solvent combinations and experimental procedures. 74

Absorption spectra of J1-aggregate. Carotenoid aggregates that have a predominant bathochromic shift have been traditionally termed J-aggregates.18 We have adopted this nomenclature throughout the present study, but it is important to note that the red-shifted absorption of the carotenoids does not resemble the typical narrow exciton-coupled band of other J-aggregates, such as pseudoisocyanine dyes.75, 76 An alternative explanation for the absorption spectrum of the system denoted in our work as

J1 is in terms of a weakly-coupled H-aggregate.33, 51 The computational results attribute

-1 the ~1200 cm red-shift of E00, and the position of the 0→0 band (relative to the monomer) to non-resonant dispersion interactions.33 Furthermore, the vibronic intensity pattern of J1, in which the most red-shifted 0→0 band is significantly attenuated relative to 0→2 and 0→3 transitions, is in line with expectations for an H-aggregate.77 The energy spacings between vibronic transitions of J1 vary from 1470 to 1190 cm-1, and have a different pattern from the 1300 cm-1 spacing of the monomer absorption. Changes of this kind are expected in the regime where coupling strength is intermediate.70, 78

Other evidence in favor of assigning the so-called J1-aggregate to a weakly-bound H- aggregate are described below, in the context of the emission spectroscopy.

Absorption spectra of J2 aggregate. The J2-aggregate has absorption features that are very similar to carotenoids in the crystalline state. The optical absorption of crystalline β-carotene has been reported in several studies.48, 62, 79-81 The absorption maxima for light with polarization along the long (b) axis of the crystal are located at 535,

493, 456, 426 nm.81 Similar peak positions have been noted in other studies of crystals80

55 and small crystals in solution.82 The fitted bands for the J2 spectrum have maxima at 525,

484, 453, and 423 nm. The red-shifted 0→0 band relative to the monomer has been long recognized as a characteristic of the crystalline state.82 To our knowledge, the apportioning of the red-shifted maxima to resonant (exciton coupling) interaction versus non-resonant dispersion interactions has not been reported. The red-shifted position for the low-energy absorption of J2 implies that this aggregate is more crystalline-like than the J1. The separations between adjacent peaks of the crystal are 1590, 1650, and 1540 cm-1. The first and third values of the separations for the crystal correspond exactly to separations determined for fitted bands of J2. Furthermore, the broad absorption that extends to the red of 600 nm is another connection between the spectrum of the J2 aggregate and the crystal. The origin of this absorption has not been explained in the literature, however the existence of absorption oscillator strength thousands of wavenumbers red-shifted from the monomer peak must invoke exciton coupling, rather than non-resonant interactions. In summary, the slow mixing process in dioxane/water, or the annealing of the J1 aggregate, leads to a configuration of zeaxanthin molecules that is evidently similar to the state of crystalline β-carotene.

UV absorption of the aggregates. We address the UV transitions separately from the main band in part because the shifts of this band upon aggregation or change of temperature are sometimes opposite to the shifts in the main band. Specifically, upon formation of the H-aggregate, the UV band red-shifts to ~280 nm from the 274 nm peak position of the monomer in ethanol. The direction is opposite to the blue-shifted transition of the main band. Upon heating the H-aggregate, the UV band blue-shifts to

267 nm, whereas the main absorption band at 388 nm remains nearly unchanged. The

56 fact that the UV transition in the H-aggregate shifts in a manner that contrasts to the main band is consistent with a direction and location of the UV transition dipole that differ from that associated with the main visible transition. It is known that the UV-transition of the carotenoid monomers is largely localized to the extreme ends of the polyene chain, which includes the double bond of the ionone rings in the case of -carotene and its derivatives.83 For this reason, the CD signal of monomeric zeaxanthin is restricted to the

UV region because the chiral centers at the C3 and C3' are proximal to the transitions at these ends.84 Furthermore, a study of the polarized emission from β-carotene85 agreed with the localization of the UV transition to the ends of the molecule, and also found that the transition dipole for the S0→Sn absorption is shifted 45 degrees relative to the main

85 S0→S2 transition. An earlier study of β-carotene had suggested that the UV transition is largely polarized along the short (a) axis of the crystal, which is nearly perpendicular to the long axis of the molecule.79

A possible explanation for the blue-shift of the UV-band to 267 nm is that heating causes slight translation along the long-axis of molecules that are side-by-side in the H- aggregate. This small structural rearrangement would likely have little or no effect on the exciton coupling for the main transition dipole that extends along much of the polyene chain. On the other hand, transition dipoles situated near the ionone rings, and at an angle 45 degrees turned from the polyene chain, would be more sensitive to these specific translations. A blue-shift of the UV band in the case of the heated H-aggregate suggests that this “slipping” gives rise to a stronger H-like exciton coupling of these transition dipoles, which overcomes the factors (e.g. non-dispersive interactions) that lead to the red-shift upon forming the H-aggregate.

CD spectroscopy. Exciton coupling also accounts for the CD signals in the visible region of the aggregate spectra. The sequence of the positive and negative Cotton effects in the spectra is determined by the chirality of the self-assembled structure, and specifically, the direction of the helix formed by chromophore transition dipoles.52, 86, 87

The three aggregates of this study show positive chirality, suggesting that they all form right-handed helical structures. Circular dichroism signals reflect long-range order, and are sensitive to the structure of the aggregate assembly.33 In our experiments, we have found that the intensities of the CD signal decrease over time, whereas the absorption spectra are constant. This observation is consistent with a structural model in which the aggregates are comprised of multiple layers.33, 51, 52 The loss of CD signal suggests that the interaction between layers may become irregular, or decrease over time, whereas intra-layer interaction between repeating zeaxanthin units, and hence absorption profile, is preserved. An alternative explanation for the decrease in CD signal is agglomeration of multiple nanoparticles, whereby their optical activity is cancelled. Consistent with the latter hypothesis, DLS measurements show that the particle size increases with time.

Fluorescence Emission. The aggregate emission spectra are characterized by (1) a significant red-shift in the emission, relative to the monomer; (2) fluorescence quantum yields that are a factor of 5- to 30- lower than the monomer; and (3) a pattern of vibronic intensities that is similar for the three aggregates. The emission spectra of the aggregates at room temperature with 488 nm excitation each have a maximum in the range 570-585 nm, and after fitting, the primary band is found to have a central position in the range of

570-575 nm (17540-17400 cm-1). By comparison, the dominant emission band of the monomer is centered at 525 nm (19040 cm-1). The 1500-1640 cm-1 red-shift in the

58 emission spectra of the aggregates can be attributed largely to non-resonant dispersion interaction (gas-to-crystal shift). The aggregates exhibit similar spectra relative to one another in terms of both the intensity pattern, as well as the separation between vibronic peaks. The ratio of (0→0) / (0→1) band intensities is considerable smaller for the aggregates, relative to the monomer. Furthermore, all of the aggregates show greater separation between the 0→0 and 0→1 transitions (1330-1610 cm-1), versus the separation between the 0→1 and 0→2 transitions (1090-1190 cm-1). By contrast the monomer has nearly identical (1190 and 1200 cm-1) separations between 0→0, 0→1, and 0→2 transitions.

The reduced emission quantum yields of the three aggregates relative to the monomer is an expected feature for H-aggregates, and has a simple explanation in terms of a disallowed transition from the lowest (out-of-phase) exciton-coupled state.70, 88

Furthermore, the changes in the spectral profiles that are described above are characteristic of H-aggregates.33, 77 Recent exciton coupling models have focused on the ratio of the 0→0 and 0→1 transitions in absorption and emission spectroscopy of various organic molecular or oligomer aggregates.77, 89 The 0→0 transition is forbidden in ordered H-aggregates in the theoretical limit of no site disorder; the transition gains intensity as disorder or temperature increases.77 Aggregates or films of numerous molecular or oligomeric systems, such as oligo(phenylenevinylene), oligothiophene, and

MEH-PPV also show diminished 0→0 bands.7, 9, 90-92 In most cases, these systems are considered H-aggregates.

The observation that the fluorescence spectra from H- and J-aggregates are similar to one another is an intriguing result. The approximately mirror-image

59 appearance of the J1 absorption and emission strongly suggests that the emission originates from those exciton-states that are directly excited. As mentioned above, the low-energy absorption bands of the H-aggregates are best described in terms of a subpopulation of molecules that are weakly coupled. It is likely that these same states are also responsible for the fluorescence emission in the H-aggregates. A heuristic model of the emission from the aggregates is illustrated in Figure 2.7. The emissive states are depicted as weakly coupled, and primarily H-like in character, which is consistent with both the emission pattern and the low quantum yield of fluorescence. The fact that the excitation profile of the H-aggregate corresponds reasonably well to the absorption spectrum, when monitored 630 nm, suggests that energy transfer from the strongly coupled states to the mid-gap states is relatively independent of excitation wavelength.

Furthermore, the relaxation and emission process must occur in < 5 ps, as determined by the upper limit from TCSPC measurements (Figure A2.6 in the Appendix). The fast emission additionally provides some evidence that the emission from the aggregates does not simply reflect long-lived trap states that happen to exist in the aggregated form. If these states were to exist, the emission lifetime would be longer.93

Figure 2.7. Model for the absorption, emission, and non-radiative relaxation processes of zeaxanthin aggregates. Where possible, energy levels are located based upon values in Table 2.1. The aggregate states are qualitatively denoted as weakly (W) or strongly (S) coupled. The (+) and (-) label in-phase and out-of-phase exciton states. Excited vibrational states are indicated by thin lines. D′ and D′′ label the shifts in the excited state caused by non-resonant dispersion interactions, or gas-to-crystal shifts, which vary for each aggregate type. Similar interactions for the ground state are not shown.

There are few reports of fluorescence emission from carotenoids in crystalline form.80, 94 Three bands were reported for -carotene crystal at 4 K, at 18100, 16900, and

15770 cm-1 (552, 592, 634 nm).80 The overall mirror-image resemblance of the

80 emission features to the absorption allows assignment to S2→S0 emission. The 16900 cm-1(592 nm) band is dominant, and is presumably the 0→1 band. The highest energy band at 18100 cm-1 (552 nm) has the lowest intensity of the three Gaussians, although self-absorption by the crystal may affect the intensity and position of this peak.

The overall intensity pattern for low-temperature -carotene crystal is similar to that of the aggregates. However the separation between the 0→0 and 0→1 transitions is 1200 cm-1, which is significantly smaller than for the aggregate. However, it is possible that

61 the position of the 0→0 peak for the crystal could be significantly blue-shifted if reabsorption were not a factor, thus increasing 1200 cm-1 to a value that is closer to what we report here for the J2-aggregate.

An additional noteworthy aspect of the H-aggregates in particular is the variation in emission spectrum, as a function of excitation wavelength. This characteristic is certainly not unique to the H-aggregate; analogous character has been described for the β- carotene monomer. 43, 48, 49 The fact that the emission of J1 and J2 is less dependent upon excitation wavelength is reasonable, because in the case of H, additional relaxation processes must occur before the mid-gap states are populated. The fact that the emission shifts significantly to higher energy with bluer excitation indicates that the non-radiative vibrational and/or electronic processes occur at a similar rate as the emission itself.

Similar conclusions were drawn for the monomer, based primarily upon time-resolved fluorescence spectroscopy.

Molecular conformation from Resonance Raman spectra. The assignments of the vibrational peaks of monomeric carotenoids have been the topic of numerous papers.95-98 In large part, the mode frequencies of the aggregates are nearly unchanged from those of the monomer, which strongly suggests little perturbation of the monomer structure upon aggregation. Nonetheless, the small changes are ones that reveal subtle, yet important features, of the aggregates.

One of the most obvious changes is the 3-9 cm-1 downshifts of the main ethylenic band at ~1520 cm-1. The downshift is consistent with prior measurements.31, 48 Previous calculations as well as our own reveal that the strong ethylenic v1 mode consists of two bands with similar frequencies.98 Both are dominated by in-phase stretches of C=C

62 modes in the center of the chain. Slight shifts in the peak position of the monomer have been noted upon changing the laser excitation wavelength, and these have been attributed to Raman excitation profiles (REP) for the two modes that are spectrally shifted relative to one another. However, it is unlikely that the observed differences in ethylenic frequencies for the aggregates can be attributed to shifted REPs for nearby modes, because we see a consistent ordering of downshifts (J2 ≥ J1 > H) for any excitation wavelength in the range 415 to 600 nm. Furthermore, the coincident downshift of the

~1590 cm-1 band, which is associated with a single normal mode, cannot be explained with the REP argument.

It is well known that the C=C bond frequencies downshift as the length of the conjugated region increases. Some researchers have proposed a linear relationship of the main C=C stretching frequency versus 1/(N+1), where N is the number of carbon double bonds.99, 100 Other investigators have suggested empirical downshifts of 3 to 5 cm-1 per double bond for chromophores that contain ~9 double bonds. 101, 102 The effective conjugation length of polyenes is affected by its structure. In one scenario, if the ionone rings lie in the plane of the conjugated chain, the overall extent of conjugation would be enhanced. However, we note that both lutein and zeaxanthin show exactly the same downshift (6 cm-1) when comparing the molecules solvated in acetone versus aggregated in 9:1 water:acetone.31 If the flattening of the rings were the dominant reason for shifts in the ethylenic stretch, one would expect a decreased effect for lutein, since the double bond of one ring cannot be part of the conjugated system. Similarly, astaxanthin and canthaxanthin, which have possible extensions of the conjugation length via the C=O moiety of these molecules, likewise show 5 cm-1 downshifts when comparing the

63 acetone-solvated monomers versus aggregates.31 The conclusion that the downshift in aggregated state is not primarily a consequence of ionone ring rotation agrees with an earlier report.83 An alternate explanation involves flattening of the polyene chain itself, which also enhances delocalization of the π electrons, and thus causes downshifts of both the ~1520 and 1590 cm-1 ethylenic modes. This explanation is consistent with the diminished HOOP intensity of the aggregates at ~870 cm-1, which is a sensitive indicator of polyene chain distortion.103, 104 Based on the observed trends for downshifts, we conclude that the extent of flattening for zeaxanthin molecules in the aggregates is in the order of J2>J1>H.

Additional Raman data suggest that zeaxanthin loses symmetry in the aggregates, primarily because the ionone rings are twisted asymmetrically. The out-of-phase C=C stretching modes at ~1560 cm-1 are not expected to be resonantly enhanced when the molecule has a high degree of symmetry. The optimized structure for an isolated zeaxanthin molecule based on DFT calculations shows a symmetric configuration in which the two ionone rings are twisted opposite to one other with dihedral angles C5-C6-

C7-C8 and C5′-C6′-C7′-C8′ that are ±47 °. In the H- and J-aggregates, this symmetry appears to be broken, probably because of intermolecular interactions that may involve the two chiral centers on the ionone rings. Lowered symmetry could cause the out-of- phase 1560 cm-1 mode to become more intense for the aggregates. Other changes, including the diminished intensity of the 1270 cm-1 (in-phase combination of C-H bending modes), and the increased intensity of the 373 cm-1 mode (likely a skeletal torsion or CCC bend)105, 106 do not have a clear explanation at this point.

2.5 Conclusions

The steady-state electronic and vibrational study has been largely motivated by our interest in characterizing carotenoid aggregates for studies of singlet fission.27 The zeaxanthin aggregates can be reproducibly prepared, and are relatively robust, with absorption spectra that are stable for at least several hours. Aggregation is found to hinder isomerization, as compared with monomers in solution. The absorption and emission spectra of the three aggregates are best explained in terms of H-type exciton coupling, and distinguished by the strength of the interaction between chromophores.

The minor vibrational structural changes that distinguish monomeric versus aggregated zeaxanthin hardly vary from one aggregate to the next, and one can therefore conclude that the individual chromophores have the same structural form in the three aggregate types. Ongoing work is focused upon the correlation of singlet fission dynamics with exciton coupling, and will be reported elsewhere.

Acknowledgement

Chapter 2, in full, is a reprint of the material as it appears in the Journal of

Physical Chemistry B, 2012, 116(35), pp 10617-10630, by Chen Wang, Christopher J.

Berg, Cheng-Chih Hsu, Brittany A. Merrill, and Michael J. Tauber. The author of this dissertation was one of the two primary experimentalists, and was a coauthor of the paper.

References

1. Reich, C.; Pandolfe, W. D.; Bird, G. R., Attachments and spectral shifts of dye aggregates on silver bromide surfaces. PHOTOGRAPHIC SCIENCE AND ENGINEERING 1973, 17, (3), 334-341.

2. Herz, A. H., Dye-dye interactions of cyanines in solution and at AgBr surfaces. Photographic Science and Engineering 1974, 18, (3), 323-335.

3. McHale, J. L., Hierarchal light-harvesting aggregates and their potential for solar energy applications. Journal of Physical Chemistry Letters 2012, 3, (5), 587-597.

4. Herz, A. H., Aggregation of sensitizing dyes in solution and their adsorption onto silver halides. Advances in Colloid and Interface Science 1977, 8, (4), 237-298.

5. Leng, W. N.; Würthner, F.; Kelley, A. M., Resonance Raman intensity analysis of merocyanine dimers in solution. Journal of Physical Chemistry B 2004, 108, (29), 10284- 10294.

6. R sch, U.; ao, S.; Wortmann, R.; W rthner, F., Fluoresceut H-aggregates of merocyanine dyes. Angewandte Chemie-International Edition 2006, 45, (42), 7026-7030.

7. Ostrowski, D. P.; Lytwak, L. A.; Mejia, M. L.; Stevenson, K. J.; Holliday, B. J.; Vanden Bout, D. A., The effects of aggregation on electronic and optical properties of oligothiophene particles. Acs Nano 6, (6), 5507-5513.

8. Peteanu, L. A.; Sherwood, G. A.; Werner, J. H.; Shreve, A. P.; Smith, T. M.; Wildeman, J., Visualizing core-shell structure in substituted PPV oligomer aggregates using fluorescence lifetime imaging microscopy (FLIM). Journal of Physical Chemistry C 2011, 115, (31), 15607-15616.

9. Sherwood, G. A.; Cheng, R.; Smith, T. M.; Werner, J. H.; Shreve, A. P.; Peteanu, L. A.; Wildeman, J., Aggregation effects on the emission spectra and dynamics of model oligomers of MEH-PPV. Journal of Physical Chemistry C 2009, 113, (43), 18851-18862.

10. Chambers, R. W.; Kajiwara, T.; Kearns, D. R., Effect of dimer formation on the electronic absorption and emission spectra of ionic dyes. Rhodamines and other common dyes. Journal of Physical Chemistry 1974, 78, (4), 380-387.

11. Lim, S.-H.; Bjorklund, T. G.; Spano, F. C.; Bardeen, C. J., Exciton delocalization and superradiance in tetracene thin films and nanoaggregates. Physical Review Letters 2004, 92, (10), 107402.

12. Ahrens, M. J.; Sinks, L. E.; Rybtchinski, B.; Liu, W.; Jones, B. A.; Giaimo, J. M.; Gusev, A. V.; Goshe, A. J.; Tiede, D. M.; Wasielewski, M. R., Self-assembly of supramolecular light-harvesting arrays from covalent multi-chromophore perylene- 3,4:9,10-bis(dicarboximide) building blocks. Journal of the American Chemical Society 2004, 126, (26), 8284-8294.

13. Rybtchinski, B.; Sinks, L. E.; Wasielewski, M. R., Combining light-harvesting and charge separation in a self-assembled artificial photosynthetic system based on perylenediimide chromophores. Journal of the American Chemical Society 2004, 126, (39), 12268-12269.

14. Kaiser, T. E.; Wang, H.; Stepanenko, V.; Würthner, F., Supramolecular construction of fluorescent J-aggregates based on hydrogen-bonded perylene dyes. Angewandte Chemie International Edition 2007, 46, (29), 5541-5544.

15. Ghosh, S.; Li, X.-Q.; Stepanenko, V.; Würthner, F., Control of H- and J-type π stacking by peripheral alkyl chains and self-sorting phenomena in perylene bisimide homo- and heteroaggregates. Chemistry – A European Journal 2008, 14, (36), 11343- 11357.

16. Seibt, J.; Dehm, V.; Wurthner, F.; Engel, V., Circular dichroism spectroscopy of small molecular aggregates: Dynamical features and size effects. Journal of Chemical Physics 2008, 128, (20), 7.

17. Akins, D. L.; Zhu, H. R.; Guo, C., Absorption and Raman scattering by aggregated meso-tetrakis(p-sulfonatophenyl)porphine. Journal of Physical Chemistry 1994, 98, (14), 3612-3618.

18. Köhn, S.; Kolbe, H.; Korger, M.; Köpsel, C.; Mayer, B.; Auweter, H.; E. Lüddecke; Bettermann, H.; Martin, H.-D., Aggregation and interface behaviour of carotenoids. In Carotenoids, Britton, G.; Pfander, H., Eds. Birkhäuser Basel: 2008; Vol. 4, pp 53-98.

19. Wasielewski, M. R., Energy, charge, and spin transport in molecules and self- assembled nanostructures inspired by photosynthesis. Journal of Organic Chemistry 2006, 71, (14), 5051-5066.

20. Chen, Z.; Lohr, A.; Saha-Möller, C. R.; Würthner, F., Self-assembled π-stacks of functional dyes in solution: structural and thermodynamic features. Chemical Society Reviews 2009, 38, (2), 564-584.

21. Hanna, M. C.; Nozik, A. J., Solar conversion efficiency of photovoltaic and photoelectrolysis cells with carrier multiplication absorbers. Journal of Applied Physics 2006, 100, (7), 0745101-8.

22. Smith, M. B.; Michl, J., Singlet fission. Chemical Reviews 2010, 110, (11), 6891- 6936.

24. Greyson, E. C.; Vura-Weis, J.; Michl, J.; Ratner, M. A., Maximizing singlet fission in organic dimers theoretical investigation of triplet yield in the regime of localized excitation and fast coherent electron transfer. Journal of Physical Chemistry B 2010, 114, (45), 14168-14177.

25. Muller, A. M.; Avlasevich, Y. S.; Mullen, K.; Bardeen, C. J., Evidence for exciton fission and fusion in a covalently linked tetracene dimer. Chemical Physics Letters 2006, 421, (4-6), 518-522.

26. Muller, A. M.; Avlasevich, Y. S.; Schoeller, W. W.; Mullen, K.; Bardeen, C. J., Exciton fission and fusion in bis(tetracene) molecules with different covalent linker structures. Journal of the American Chemical Society 2007, 129, (46), 14240-14250.

27. Wang, C.; Tauber, M. J., High-yield singlet fission in a zeaxanthin aggregate observed by picosecond resonance Raman spectroscopy. Journal of the American Chemical Society 2010, 132, (40), 13988-13991.

28. Wang, C.; Schlamadinger, D. E.; Desai, V.; Tauber, M. J., Triplet excitons of carotenoids formed by singlet fission in a membrane. Chemphyschem 2011, 12, (16), 2891-2894.

29. Simpson, W. T.; Peterson, D. L., Coupling strength for resonance force transfer of electronic energy in van der Waals solids. The Journal of Chemical Physics 1957, 26, (3), 588-593.

30. Billsten, H. H.; Sundstrom, V.; Polivka, T., Self-assembled aggregates of the carotenoid zeaxanthin: time-resolved study of excited states. Journal of Physical Chemistry A 2005, 109, (8), 1521-1529.

31. Salares, V. R.; Young, N. M.; Carey, P. R.; Bernstein, H. J., Excited-state (exciton) interactions in polyene aggregates - Resonance Raman and absorption spectroscopic evidence. Journal of Raman Spectroscopy 1977, 6, (6), 282-288.

32. Gruszecki, W. I.; Zelent, B.; Leblanc, R. M., Fluorescence of zeaxanthin and violaxanthin in aggregated forms. Chemical Physics Letters 1990, 171, (5-6), 563-568.

33. Spano, F. C., Analysis of the UV/Vis and CD spectral line shapes of carotenoid assemblies: spectral signatures of chiral H-aggregates. Journal of the American Chemical Society 2009, 131, (12), 4267-4278.

34. McCreery, R. L., Raman spectroscopy for chemical analysis. John Wiley & Sons: 2005; Vol. 225.

35. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.;

Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09, Gaussian Inc.: Wallingford, CT, 2009.

36. Linden, A.; Burgi, B.; Eugster, C. H., Confirmation of the structures of lutein and zeaxanthin. Helvetica Chimica Acta 2004, 87, (5), 1254-1269.

37. Champion, P. M.; Lange, R., On the quantitation of light emission from cytochrome c in the low quantum yield limit. The Journal of Chemical Physics 1980, 73, (12), 5947-5957.

38. Kochendoerfer, G. G.; Mathies, R. A., Spontaneous emission study of the femtosecond isomerization dynamics of rhodopsin. Journal of Physical Chemistry 1996, 100, (34), 14526-14532.

39. Tauber, M. J.; Mathies, R. A., Fluorescence and resonance Raman spectra of the aqueous solvated electron. Journal of Physical Chemistry A 2001, 105, (49), 10952- 10960.

40. Zsila, F.; Bikadi, Z.; Deli, J.; Simonyi, M., Chiral detection of carotenoid assemblies. Chirality 2001, 13, (8), 446-453.

41. Bondarev, S. L.; Bachilo, S. M.; Dvornikov, S. S.; Tikhomirov, S. A., S2→S0 fluorescence and transient Sn→S1 absorption of all-trans-β-carotene in solid and liquid solutions. Journal of Photochemistry and Photobiology A: Chemistry 1989, 46, (3), 315- 322.

42. Gillbro, T.; Cogdell, R. J., Carotenoid fluorescence. Chemical Physics Letters 1989, 158, (3-4), 312-316.

43. Jørgensen, K.; Stapelfeldt, H.; Skibsted, L. H., Fluorescence of carotenoids. Effect of oxygenation and cis/trans isomerization. Chemical Physics Letters 1992, 190, (5), 514-519.

44. Kandori, H.; Sasabe, H.; Mimuro, M., Direct determination of a lifetime of the S2 state of β-carotene by femtosecond time-resolved fluorescence spectroscopy. Journal of the American Chemical Society 1994, 116, (6), 2671-2672.

45. Frank, H. A.; Bautista, J. A.; Josue, J. S.; Young, A. J., Mechanism of nonphotochemical quenching in green plants: energies of the lowest excited singlet states of violaxanthin and zeaxanthin. Biochemistry 2000, 39, (11), 2831-2837.

46. Josue, J. S.; Frank, H. A., Direct Determination of the S1 excited-state energies of xanthophylls by low-temperature fluorescence spectroscopy. Journal of Physical Chemistry A 2002, 106, (19), 4815-4824.

47. Akimoto, S.; Yamazaki, I.; Takaichi, S.; Mimuro, M., Excitation relaxation of carotenoids within the S-2 state probed by the femtosecond fluorescence up-conversion method. Chemical Physics Letters 1999, 313, (1-2), 63-68.

48. Mori, Y., Introductory studies on the growth and characterization of carotenoid solids: an approach to carotenoid solid engineering. Journal of Raman Spectroscopy 2001, 32, (6-7), 543-550.

49. Nakamura, R.; Yamamoto, S.; Nakahara, J., Broad Raman scattering and luminescence in β-carotene solution. Journal of Chemical Physics 2002, 117, (1), 238- 247.

50. Nakamura, R.; Wang, P.; Fujii, R.; Koyama, Y.; Hashimoto, H.; Kanematsu, Y., Vibrational relaxation pathways in the electronic excited state of carotenoid. Journal of Luminescence 2006, 119-120, (0), 442-447.

51. Zsila, F.; Bikadi, Z.; Keresztes, Z.; Deli, J.; Simonyi, M., Investigation of the self- organization of lutein and lutein diacetate by electronic absorption, circular dichroism spectroscopy, and atomic force microscopy. Journal of Physical Chemistry B 2001, 105, (39), 9413-9421.

52. Simonyi, M.; Bikadi, Z.; Zsila, F.; Deli, J., Supramolecular exciton chirality of carotenoid aggregates. Chirality 2003, 15, (8), 680-698.

53. Lematre, J.; Maudinas, B.; Ernst, C., Spontaneous organization of two xanthophyll compounds in water-alcohol solutions as shown by circular dichroism. Photochemistry and Photobiology 1980, 31, (3), 201-205.

54. Auweter, H.; Haberkorn, H.; Heckmann, W.; Horn, D.; Luddecke, E.; Rieger, J.; Weiss, H., Supramolecular structure of precipitated nanosize β-carotene particles. Angewandte Chemie-International Edition 1999, 38, (15), 2188-2191.

55. Horn, D.; Rieger, J., Organic nanoparticles in the aqueous phase - Theory, experiment, and use. Angewandte Chemie-International Edition 2001, 40, (23), 4331- 4361.

56. Okamoto, H.; Hamaguchi, H.-O.; Tasumi, M., Resonance Raman studies on tetradesmethyl-β-carotene aggregates. Journal of Raman Spectroscopy 1989, 20, (11), 751-756.

57. Wang, L. X.; Du, Z. L.; Li, R. X.; Wu, D. C., Supramolecular aggregates of lycopene. Dyes and Pigments 2005, 65, (1), 15-19.

58. Giovannetti, R.; Alibabaei, L.; Pucciarelli, F., Kinetic model for astaxanthin aggregation in water-methanol mixtures. Spectrochimica Acta Part a-Molecular and Biomolecular Spectroscopy 2009, 73, (1), 157-162.

59. Mori, Y.; Yamano, K.; Hashimoto, H., Bistable aggregate of all-trans-astaxanthin in an aqueous solution. Chemical Physics Letters 1996, 254, (1-2), 84-88.

60. Sterling, C., Crystal structure of β-carotene. Acta Crystallographica 1964, 17, (10), 1224-1228.

61. Senge, M. O.; Hope, H.; Smith, K. M., Structure and conformation of photosynthetic pigments and related compounds: 3. Crystal structure of β-carotene. Zeitschrift Fur Naturforschung C-a Journal of Biosciences 1992, 47, (5-6), 474-476.

62. Hashimoto, H.; Sawahara, Y.; Okada, Y.; Hattori, K.; Inoue, T.; Matsushima, R., Observation of solitonlike excitations in all-trans-β-carotene single crystals. Japanese Journal of Applied Physics Part 1-Regular Papers Short Notes & Review Papers 1998, 37, (4A), 1911-1918.

63. Christensen, R. L., The electronics states of carotenoids. In The Photochemistry of the Carotenoids, Frank, H. A.; Young, A. J.; Britton, G.; Cogdell, R. J., Eds. Kluwer Academic Publishers: Dordecht, 1999; Vol. 8, pp 137-157.

64. Shreve, A. P.; Trautman, J. K.; Owens, T. G.; Albrecht, A. C., Determination of the S2 lifetime of β-carotene. Chemical Physics Letters 1991, 178, (1), 89-96.

65. Polivka, T.; Sundstrom, V., Ultrafast dynamics of carotenoid excited states - From solution to natural and artificial systems. Chemical Reviews 2004, 104, (4), 2021-2071.

66. Buchwald, M.; Jencks, W. P., Optical properties of astaxanthin solutions and aggregates. Biochemistry 1968, 7, (2), 834-843.

67. Kopsel, C.; Moltgen, H.; Schuch, H.; Auweter, H.; Kleinermanns, K.; Martin, H. D.; Bettermann, H., Structure investigations on assembled astaxanthin molecules. Journal of Molecular Structure 2005, 750, (1-3), 109-115.

68. Ruban, A. V.; Horton, P.; Young, A. J., Aggregation of higher-plant xanthophylls - Differences in absorption-spectra and in the dependency on solvent polarity. Journal of Photochemistry and Photobiology B-Biology 1993, 21, (2-3), 229-234.

69. Alster, J.; Polivka, T.; Arellano, J. B.; Chabera, P.; Vacha, F.; Psencik, J., β- Carotene to bacteriochlorophyll c energy transfer in self-assembled aggregates mimicking chlorosomes. Chemical Physics 2010, 373, (1-2), 90-97.

70. Kasha, M., Energy transfer mechanisms and the molecular exciton model for molecular aggregates. Radiation Research 1963, 20, (1), 55-70.

71. Cantor, C. R.; Schimmel, P. R., Biophysical Chemistry Part II: Techniques for the Study of Biological Structure and Function. W. H. Freeman: San Francisco, 1980.

72. Parson, W. W., Modern Optical Spectroscopy. Springer: Dordecht, The Netherlands, 2007.

73. Tinoco, I., Hypochromism in polynucleotides. Journal of the American Chemical Society 1960, 82, (18), 4785-4790.

74. Gruszecki, W. I., Structural characterization of the aggregated forms of violaxanthin.. Journal of Biological Physics 1991, 18, (2), 99-110.

75. Knoester, J.; Spano, F. C., J-aggregates. In Kobayashi, T., Ed. World Scientific Publishing Co.: Singapore: 1996; pp 111-160.

76. Würthner, F.; Kaiser, T. E.; Saha-Möller, C. R., J-aggregates: From serendipitous discovery to supramolecular engineering of functional dye materials. Angewandte Chemie International Edition 2011, 50, (15), 3376-3410.

77. Spano, F. C., The spectral signatures of frenkel polarons in H- and J-aggregates. Accounts of Chemical Research 2010, 43, (3), 429-439.

78. Fulton, R. L.; Gouterman, M., Vibronic coupling. 2. Spectra of dimers. Journal of Chemical Physics 1964, 41, (8), 2280-2286.

79. Chapman, D.; Cherry, R. J.; Morrison, A., Spectroscopic and electrical studies of all-trans-β-carotene crystals. Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences 1967, 301, (1465), 173-193.

80. Gaier, K.; Angerhofer, A.; Wolf, H. C., The lowest excited electronic singlet states of all-trans-β-carotene single crystals. Chemical Physics Letters 1991, 187, (1-2), 103-109.

81. Hashimoto, H.; Kiyohara, D.; Kamo, Y.; Komuta, H.; Mori, Y., Molecular orientation of all-trans-β-carotene in spin-coated film and in Langmuir-Blodgett film as

72 detected by polarized optical absorption and reflection spectroscopies. Japanese Journal of Applied Physics 1996, 35, (1R), 281.

82. Shibata, K., Absorption spectra of suspensions of carotene crystals. Biochimica et biophysica acta 1956, 22, (2), 398-399.

83. Weesie, R. J.; Merlin, J. C.; Lugtenburg, J.; Britton, G.; Jansen, F.; Cornard, J. P., Semiempirical and Raman spectroscopic studies of carotenoids. Biospectroscopy 1999, 5, (1), 19-33.

84. Buchecker, R.; Noack, K., Circular dichroism. In Carotenoids, Britton, G.; Liaaen-Jensen, S.; Pfander, H., Eds. Birkhauser Verlag: Basel, Switzerland, 1995; Vol. 1B.

85. Bondarev, S. L.; Knyukshto, V. N.; Bachilo, S. M., Polarized fluorescence of β- carotene and related polyenes. Journal of Applied Spectroscopy 2000, 67, (1), 88-94.

86. Kuhn, H., Interaction of chromophores in mono layer assemblies. Pure and Applied Chemistry 1971, 27, (3), 421-438.

87. Berova, N.; Di Bari, L.; Pescitelli, G., Application of electronic circular dichroism in configurational and conformational analysis of organic compounds. Chemical Society Reviews 2007, 36, (6), 914-931.

88. McRae, E. G.; Kasha, M., Enhancement of phosphorescence ability upon aggregation of dye molecules. Journal of Chemical Physics 1958, 28, (4), 721-722.

89. Kistler, K. A.; Pochas, C. M.; Yamagata, H.; Matsika, S.; Spano, F. C., Absorption, circular dichroism, and photoluminescence in perylene diimide bichromophores: Polarization-dependent H- and J-aggregate behavior. Journal of Physical Chemistry B 2012, 116, (1), 77-86.

90. Egelhaaf, H. J.; Gierschner, J.; Oelkrug, D., Characterization of oriented oligo(phenylenevinylene) films and nano-aggregates by UV/Vis-absorption and fluorescence spectroscopy. Synthetic Metals 1996, 83, (3), 221-226.

91. Clark, A. E.; Qin, C. Y.; Li, A. D. Q., Beyond exciton theory: A time-dependent DFT and Franck-Condon study of perylene diimide and its chromophoric dimer. Journal of the American Chemical Society 2007, 129, (24), 7586-7595.

92. Spano, F. C., Absorption in regio-regular poly(3-hexyl)thiophene thin films: Fermi resonances, interband coupling and disorder. Chemical Physics 2006, 325, (1), 22- 35.

93. Ahn, T.-S.; Müller, A. M.; Al-Kaysi, R. O.; Spano, F. C.; Norton, J. E.; Beljonne, D.; Brédas, J.-L.; Bardeen, C. J., Experimental and theoretical study of temperature

73 dependent exciton delocalization and relaxation in anthracene thin films. Journal of Chemical Physics 2008, 128, (5), 0545051-05450511.

94. Włoch, E.; Więckowski, S.; Turek, A. M., Spectroscpic characteristics of the long-wavelength absorbing form of β-carotene. Photosynthetica 1987, 21, (1), 2-8.

95. Saito, S.; Tasumi, M.; Eugster, C. H., Resonance Raman spectra (5800-40 cm-1) of all-trans and 15-cis isomers of β-carotene in the solid-state and in solution - Measurements with various laser lines from ultraviolet to red. Journal of Raman Spectroscopy 1983, 14, (5), 299-309.

96. Okamoto, H.; Saito, S.; Hamaguchi, H.; Tasumi, M.; Eugster, C. H., Resonance Raman spectra and excitation profiles of tetradesmethyl-β-carotene. Journal of Raman Spectroscopy 1984, 15, (5), 331-335.

97. Schlucker, S.; Szeghalmi, A.; Schmitt, M.; Popp, J.; Kiefer, W., Density functional and vibrational spectroscopic analysis of β-carotene. Journal of Raman Spectroscopy 2003, 34, (6), 413-419.

98. Tschirner, N.; Schenderlein, M.; Brose, K.; Schlodder, E.; Mroginski, M. A.; Thomsen, C.; Hildebrandt, P., Resonance Raman spectra of β-carotene in solution and in photosystems revisited: an experimental and theoretical study. Physical Chemistry Chemical Physics 2009, 11, (48), 11471-11478.

99. Merlin, J. C., Resonance Raman analysis of astaxanthin–protein complexes. Journal of Raman Spectroscopy 1987, 18, (7), 519-523.

100. Rimai, L.; Heyde, M. E.; Gill, D., Vibrational spectra of some carotenoids and related linear polyenes - Raman spectroscopic study. Journal of the American Chemical Society 1973, 95, (14), 4493-4501.

101. Koyama, Y.; Jiang, Y. S.; Watanabe, Y., Fluorescence spectroscopy of 13'-cis- spheroidene at 170 K - a reason for the natural-selection of the all-trans configuration for the light-harvesting function. Biospectroscopy 1995, 1, (2), 125-132.

102. Ruban, A. V.; Pascal, A.; Lee, P. J.; Robert, B.; Horton, P., Molecular configuration of xanthophyll cycle carotenoids in photosystem II antenna complexes. Journal of Biological Chemistry 2002, 277, (45), 42937-42942.

103. Eyring, G.; Curry, B.; Broek, A.; Lugtenburg, J.; Mathies, R., Assignment and interpretation of hydrogen out-of-plane vibrations in the resonance Raman spectra of Rhodopsin and bathorodopsin. Biochemistry 1982, 21, (2), 384-393.

104. Curry, B.; Palings, I.; Broek, A. D.; Pardoen, J. A.; Lugtenburg, J.; Mathies, R., Vibrational analysis of the retinal isomers. In In Advances in Infrared and Raman Spectroscopy, Clark, R. J. H.; Hester, R. E., Eds. Heyden: New York, 1984.

105. Lopez-Ramirez, M. R.; Sanchez-Cortes, S.; Perez-Mendez, M.; Blanch, G., Trans-cis isomerisation of the carotenoid lycopene upon complexation with cholesteric polyester carriers investigated by Raman spectroscopy and density functional theory. Journal of Raman Spectroscopy 2010, 41, (10), 1170-1177.

106. Lin, S. W.; Groesbeek, M.; van der Hoef, I.; Verdegem, P.; Lugtenburg, J.; Mathies, R. A., Vibrational assignment of torsional normal modes of rhodopsin: Probing excited-state isomerization dynamics along the reactive C-11=C-12 torsion coordinate. Journal of Physical Chemistry B 1998, 102, (15), 2787-2806.

Appendix of Chapter 2

a Table A2.1. Observed peak positions and amplitudes of UV-Vis absorption and CD spectra Absorption (UV-Vis) Circular dichroism (CD) Peak max. ε b Peak max/min Δεc (nm) (M-1cm-1 x 105) (nm) (M-1cm-1) 274 0.28 ~400 (shoulder) Mon in EtOH ~425 (shoulder) 452 1.40 480 1.24 288 0.27 420 -224 ~400 (shoulder) 482 442 ~420 (shoulder) 516 259 J1 in THF/H O 2 450 1.08 474 1.05 513 0.61 294 0.26 417 -377 ~400 (shoulder) 492 661 ~420 (shoulder) 524 627 J2 in Dioxane/H O 2 457 0.82 482 0.91 522 0.75 280 0.20 379 -460 388 1.11 397 426 H in EtOH/H2O ~445 (shoulder) ~475 (shoulder)

~520 (shoulder) a The peak positions and amplitudes in this table are not from fits; they are the observed maxima or minima. Positions of shoulders are estimated. b Molar extinction coefficient c Molar circular dichroism. The ellipticity, θ (millidegree) is converted to differential absorption (ΔA) for left and right circularly polarized light using the following equation:1

Eq. A2.1

The molar circular dichroism, ∆ε, is computed by division of ΔA by the pathlength, l (cm), and concentration of sample, c (M):

Figure A2.1. H-aggregate spectra resulting from varied rates of H2O addition to 25-75 μM solutions of zeaxanthin in ethanol. The final ratio of solvents is 80:20 (v/v)

H2O/EtOH in all cases. Slower addition enhances the absorption bands in the 420-530 nm region. Each spectrum can be fitted with five Gaussian bands, with center positions listed in Table A2.1.

Table A2.2. Peak positions of five Gaussian bands, fitted to H aggregate spectra shown in Figure A2.1. H- Peak 5 Peak 4 Peak 3 Peak 2 Peak 1 aggregate H-1 25820 23920 22340 20840 19350 H-2 25840 23720 22330 21020 19490 H-3 25840 23890 22280 21040 19510 H-4 25760 23870 22520 21070 19480 Average 25820 23850 22370 20990 19460

Figure A2.2A. Temperature-dependent UV-Vis spectra for the H- and J1-aggregates.

Figure A2.2B. Temperature dependent UV-Vis spectra for the monomer and J2- aggregate.

Figure A2.3. Isomerization of samples, monitored by the cis-band at ~335 nm. The zeaxanthin monomer is shown before (black) and after (dashed) heating to 78 °C and cooling to room temperature. The solutions of J1-, J2-, and H-aggregate were each heated to 90 °C, and cooled to room temperature. The aggregates were then filtered and each was dissolved in ethanol to form a monomeric solution. The set of monomeric spectra are shown above. Very little isomerization to cis species is observed for the heated aggregates.

Determination of fluorescence quantum yields by comparison with Raman scattering from a reference molecule

We derive the ratio of emission from a fluorophore to the Raman scattering from a reference olecule that is contained within the same illuminated volume.2 The Raman scattering molecule can be the fluorophore itself or a different species (for example, a solvent), as long as its concentration and Raman scattering cross-section are known. The derivation is similar to that reported previously.2 but here we address the polarization characteristics of the excitation and emitted light.

The fluorescence dIF(4π) emitted in all directions and with any polarization from chromophores in an illuminated segment dz is equal to the number of photons absorbed in that segment (as determined by Beer’s law), multiplied by the fluorescence quantum yield ΦF:

= Eq. A2.2a

= Eq. A2.2b

In equation (Eq. A2.2a) the intensity of the laser light at any location z along the sample

–αcz is I(z) = Ioe , where Io is the incident laser intensity,  is the Naperian molar absorption coefficient, c is the molar concentration of the fluorescing sample, and z is the distance that the laser travels through the sample. The decadic molar absorption coefficient, , is used in equation (Eq. A2.2b).

The Raman scattering dIR(4π) emitted in all directions and with any polarization from reference molecules in illuminated segment dz is:

Eq. A2.3

In equation (2) the reference molecule has a total absolute Raman scattering cross section

2 3 R (cm /molecule), and number density NR (molecules/cm ). The intensity of the

illumination of any segment dz is .

The total fluorescence emission IF(4π), and the total Raman scattering IR(4π) from the reference molecules are found by integration of dIF(4π) and dIR(4π) along the z-axis, over the illuminated pathlength through the sample. At each point of the integration, both the fluorescence and Raman emission are proportional to the laser intensity at I(z),

81 therefore the ratio of the fluorescence and Raman emission, as determined by equations

1b and 2, is independent of the pathlength:

Eq. A2.4

Eq. A2.4 must be adapted for the typical configuration in which light is collected within a small solid angle, , rather than the full 4π steradians. In general, the fraction of the total Raman scattering or fluorescence intensity within the solid angle, IR( ) or IF( ), depends upon the viewing direction, as well as the polarizations of the excitation beam and of the emitted Raman or fluorescence light. In the specific case of incident radiation with a beam polarized perpendicular to the scattering plane, and a randomly oriented sample, the intensity of Raman or fluorescence emission is constant along any direction in the scattering plane.2, 3 The invariance of the emission holds true for emission that is polarized either parallel or perpendicular to the excitation.2, 3 The relationship between differential and total Raman scattering cross-sections, and the corresponding intensities is,4, 5

Eq. A2.5a

Eq. A2.5b

⊥ where is the Raman depolarization ratio. The term is the total

differential Raman cross-section (units cm2/(molecule-sr)) of the reference, for scattering of both polarizations, measured along any direction that is perpendicular to the polarization of the excitation beam.

The total fluorescence emitted into 4π steradians IF(4π) relates to the fluorescence emitted into a solid angle IF( ) in a way that is analogous to the relationship in Eq.

A2.5b,

Eq. A2.6

where the polarization of the fluorescence is expressed in terms of the fluorescence

⊥ depolarization ratio . We note that the alternative measure of fluorescence

polarization, the anisotropy r, is trivially related to the depolarization ρF:

and Eq A2.7

Thus, Eq. A2.6 can be rewritten in a form that is similar to expressions derived elsewhere

(e.g. equation 21a of Cehelnik et al.3):

Eq. A2.8

Eq. A2.4 is expressed in terms of IR( ) and IF( ), after substitution of the results from Eq. A2.5a, 5b, and 2.6:

Eq. A2.9

Cancellation of like terms and rearrangement yields the equation in the main text,

Eq. A2.10

where NAv is Avagadro’s number, cR and c are respectively the concentrations of the

Raman-scattering reference and fluorophore, in the same units. In our experiments, the

reference is ethanol (EtOH), with cR = 17.1 M. The emission from zeaxanthin is assumed to have ρF = 0.33, as expected for ultrafast fluorescence, where transition dipoles for absorption and emission should be parallel to one another. A value of ρF =0.37 can be derived from emission measurements for excitation into the main band of -carotene,6 which supports the use of the limiting 0.33 value in our calculations.

Experimentally, the primary goal is the quantification of IF( ) and IR( ). The use of the same excitation geometry and detection system for fluorescence and Raman measurements allows for cancellation of detection efficiency and other factors. Light of both polarizations is collected and scrambled before entering the spectrograph. Spectra are corrected for self-absorption as described in the main text. Both fluorescence and

Raman spectra are plotted on an energy (wavenumber) axis, and corrected for a constant-

2 energy band-pass by multiplying by (wavelength) . IF( ) is quantified by integration of the full fluorescence spectrum. IR( ) is obtained by integrating over specific Raman bands of the reference. In the present work, the bands are three C-H stretches of EtOH centered at 2880, 2930, and 2970 cm-1. The sum of total differential scattering cross- sections used in our calculations is 35.0×10-30 cm2/molecule-sr, for excitation with 488.0 nm light. This value is derived from a reported value of 27.6×10-30 cm2/molecule-sr at

7 -1 3 514.5 nm by using a correction of L(L – 2900 cm ) which accounts for the shift in laser excitation energy, L. An alternative value for the differential cross sections is

39.8×10-30 cm2/molecule-sr at 488.0 nm, based upon a sum of total differential scattering cross-sections reported by Colles and Griffiths.8 This value is in good agreement with the extrapolated value from Abe and Ito. However, Griffiths later revised the value to

14.8×10-30 cm2/molecule-sr.9 The revised value does not agree with the work of Abe and

Ito, nor with a cross-section at 488.0 nm that can be extrapolated from a pre-resonance

Raman expression.10 Our questions about the cross-sections reported by Colles and

Griffiths have led us to favor the value of 35.0×10-30 cm2/molecule-sr extrapolated from the measurements of Abe and Ito.

We note that the ethanol Raman scatterer in our experiments is formally an external reference sample, therefore exact cancellation of optical factors is not achieved, as in the case of an internal reference. Nonetheless, there is effective cancellation because the optical setup is identical for sample and reference runs, and because of the self-absorption corrections. An index of refraction correction n2 is also not necessary, since the solvents (water, ethanol) have nearly identical values over the spectral regions considered here.

A table that lists relevant components for use of Eq. A2.10 is below.

Table A2.3. Components for calculations of fluorescence quantum yields, ΦF

I ( ) I ( ) Concentration  (488.0 nm) R F (integrated area) (integrated area) Φ (μM) (M-1cm-1) F (counts-cm-1) (counts-cm-1) Mon 3.35 1.04 × 105 7.79 × 106 1.95 × 108 1.2 × 10-4 J1 4.22 7.21 × 104 7.79 × 106 3.56 × 107 2.5 × 10-5 J2 2.48 8.83 × 104 2.51 × 106 6.37 × 106 1.9 × 10-5 H 11.5 1.40 × 104 2.51 × 106 1.03 × 106 4.2 × 10-6

Figure A2.4: Emission from zeaxanthin monomer, and J1-, J2-, and H-aggregates. The excitation wavelengths are 404.3 nm (green), 457.9 nm (blue), 488.0 nm (red), and 514.5 nm (black). The spectra are normalized so that the area under each spectrum is equal. Gaussian fits to the broad fluorescence component of each spectrum are presented in Figure 2.5B.

Figure A2.5. Resonance Raman spectra acquired with 488.0 nm excitation. From top to bottom, the spectra are the H-aggregate in 80:20 (v/v) H2O/EtOH; the J2-aggregate in 80:20 (v/v) H2O/1,4 dioxane; the J1-aggregate in 90:10 (v/v) H2O/THF; and the zeaxanthin monomer in EtOH. The spectra were normalized for equal amplitude of the ethylenic mode. The 200–950 cm-1, 1240–1460 cm-1, and 1550–1650 cm-1 spectral regions are shown magnified for each spectrum. Labels refer to actual maxima of peaks, not fitted. The positions of maxima are estimated in the case of peak shoulders.

Table A2.4. Descriptions of selected vibrational modes from DFT calculations Modes (cm-1) Vibrational vectors

Descriptiona Exp. Calc.

Symmetric stretch C5=C6, C7=C8, 1596 1592 C15=C15′, (Mon) C8′=C7′, C6′=C5′

Asymmetric stretch C9=C10, 1567 C13=C14, C14′=C13′,

1561 C10′=C9′ (J1) Asymmetric stretch C7=C8, 1565 C11=C12 (major), C12′=C11′ (major),

C8’=C7′ Symmetric stretch C11=C12, 1527 C15=C15′ (major), 1525 C12′=C11′ (Mon) Symmetric stretch 1519 C13=C14, C14′=C13′

C-H in plane rock 1270 1269 and ionone ring (Mon) methylene twist

373 CCC bend or 369 (J1) skeletal torsion a Primary internal coordinates are listed. Symmetric stretch refers to an in-phase combination of internal coordinates; asymmetric stretch refers to an out-of-phase combination.

Upper limit on emission lifetime from time-correlated single-photon counting

(TCSPC)

The fluorescence lifetime measurements were performed with a homebuilt

TCSPC system, earlier versions of which have been described previously.11, 12 Briefly, the excitation source is a homebuilt titanium:sapphire oscillator, which produced femtosecond pulses at a repetition rate of 98.40 MHz. The wavelength was tuned to 840 nm, and the second harmonic at 420 nm was employed in the present experiments. The excitation pulse energy was well below 10 pJ. Samples were placed in a 1-cm pathlength luminescence cuvette and emission was monitored at 90o relative to the excitation beam.

A 0.25 m subtractive double monochromator was used to isolate the detected wavelength with very good discrimination against any Rayleigh scattering, and to eliminate pulse broadening due to differences in optical pathlengths.

The instrument response function was assessed by measuring the Rayleigh scattering of the excitation beam from a colloidal suspension, with the monochromator set to 420 nm, and confirmed by measuring Raman scattering at longer wavelengths. The detector was a microchannel plate photomultiplier, for which time dispersion at different wavelengths is very small. The full-width at half-maximum for the instrument response was approximately 50 picoseconds. Based on experience with other short-lived emitters, the shortest decay time that can be measured with the system is 10 ps (with 20-30% uncertainty) after multiple measurements, and careful deconvolution.13 From these tests, one may infer that an upper limit for a lifetime that is deemed immeasurably fast must not exceed approximately 5 ps.

The fluorescence decay signals were monitored at 570 nm for zeaxanthin monomer and the three aggregates. Long-pass filters were used to verify that the emission was at the correct wavelength, and the strength of the signals (monomer > J1 ≈

J2 > H) corresponded to what one would expect from the fluorescence quantum yields.

All samples showed decays that were indistinguishable from the instrument response function (Figure A2.6). No resolvable decay was obtained by deconvolution of the instrument response, which indicates that the fluorescence lifetime of the emission is less than 5 picoseconds in all cases, consistent with luminescence yields well below 10-3.

Similar decays were found at 530 nm (not shown).

Figure A2.6. Fluorescence decay curves of zeaxanthin monomer and J1-, J2- and H- aggregate at room temperature, as measured by TCSPC. The excitation wavelength is 420 nm, and the emission is detected at 570 nm. Data points (open circles) are separated by a time interval of 6.47 ps. Solid lines are the instrument response function, determined from Rayleigh scattering at 420 nm, and adjusted to match the amplitude of the data points.

Reference

1. Bengt Nordén, A. R. a. T. D., Linear Dichroism and Circular Dichroism: A Text Book on Polarized-Light Spectroscopy. Royal Society of Chemistry Cambridge, 2010; p 34.

2. Damen, T.; Leite, R.; Porto, S., Angular dependence of the Raman scattering from benzene excited by the He-Ne cw laser. Physical Review Letters 1965, 14, (1), 9-11.

3. Cehelnik, E. D.; Mielenz, K. D.; Velapoldi, R. A., Polarization effects on fluorescence measurements. Journal of Research of the National Bureau of Standards - A. Physics and Chemistry 1975, 79A, (1), 1.

4. Myers, A. B.; Harris, R. A.; Mathies, R. A., Resonance Raman excitation profiles of bacteriorhodopsin. The Journal of Chemical Physics 1983, 79, (2), 603-613.

5. Trulson, M. O.; Mathies, R. A., Raman cross section measurements in the visible and ultraviolet using an integrating cavity: Application to benzene, cyclohexane, and cacodylate. The Journal of Chemical Physics 1986, 84, (4), 2068-2074.

6. Bondarev, S. L.; Knyukshto, V. N.; Bachilo, S. M., Polarized fluorescence of β- carotene and related polyenes. Journal of Applied Spectroscopy 2000, 67, (1), 88-94.

7. Abe, N.; Ito, M., Effects of hydrogen bonding on the Raman intensities of methanol, ethanol and water. Journal of Raman Spectroscopy 1978, 7, (3), 161-167.

8. Colles, M. J.; Griffiths, J. E., Relative and Absolute Raman Scattering Cross Sections in Liquids. Journal of Chemical Physics 1972, 56, (7), 3384-3391.

9. Griffiths, J. E., Raman scattering cross sections in strongly interacting liquid systems: CH3OH, C2H5OH, i-C3H7OH, (CH3)2CO, H2O, and D2O. Journal of Chemical Physics 1974, 60, (6), 2556-2557.

10. Johnson, A. E.; Myers, A. B., Emission cross sections and line shapes for photodissociating triiodide in ethanol: Experimental and computational studies. Journal of Chemical Physics 1995, 102, (9), 3519-3533.

11. Magde, D.; Rojas, G. E.; Seybold, P. G., Solvent dependence of the fluorescence lifetimes of xanthene dyes. Photochemistry and Photobiology 1999, 70, (5), 737-744.

12. Sohn, H.; Sailor, M. J.; Magde, D.; Trogler, W. C., Detection of nitroaromatic explosives based on photoluminescent polymers containing metalloles. Journal of the American Chemical Society 2003, 125, (13), 3821-3830.

13. Monti, S.; Malatesta, V.; Bortolus, P.; Magde, D., Picosecond fluorescence of merocyanines derived from photochromic spironaphthoxazines. Photochemistry and Photobiology 1996, 64, (1), 87-91.

Chapter 3 High Yield Singlet Fission in a Zeaxanthin Aggregate Observed by Picosecond Time-Resolved Resonance Raman Spectroscopy

3.1 Introduction

The generation of multiple excitons upon absorption of a single photon has attracted attention recently, in part because of potential benefits to solar energy conversion.1, 2 Singlet fission, in which two triplet excited states are created from one singlet excited state, is a process that could significantly enhance the efficiency of organic molecular photovoltaics.3, 4 Fission plays a role in the photophysics of acenes,5-9 polymers,10-12 and other systems.13-15 Correlations between triplet quantum yield (QY) and the relative energies of the lowest singlet and triplet states have been a primary focus.3 However, investigations of the mechanism of intermolecular fission, and the influence of fundamental properties such as the coupling strength and relative orientation of the interacting chromophores are rarely investigated.8 Additionally, there are few reports of absolute triplet quantum yields from fission, and even fewer where triplet signatures are directly monitored as part of the measurement. New approaches for time- resolved measurement of triplets and their yield must be employed to gain an improved understanding of the fission mechanism, and to evaluate possible applications.

Here, we describe a new aggregate of the carotenoid all-trans 3R,3'R-zeaxanthin

(Figure 3.1). Upon photoexcitation, we find direct evidence of intermolecular singlet fission by picosecond time-resolved resonance-Raman spectroscopy (TRRR).

Zeaxanthin was selected principally because its excited states are expected to meet one criterion for efficient singlet fission: the T1 excited state energy should be ~50% or

3, 11 lessthan the energy of the S1 excited state. Valence bond theory predicts that long

16 -1 polyenes meet this criterion; experimentally, the S1 energy of zeaxanthin is 14,500 cm , and the triplet energy is estimated to be 7000 ± 800 cm-1.16, 17 A secnd advantage of zeaxanthin is that it readily self-assembles in binary organic/aqueous solutions.18 The aggregate studied here was prepared by adding water to a 100 μM solution of zeaxanthin in tetrahydrofuran (THF) to create a 90% (v/v) aqueous solution. Transmission electron microscopy (TEM) shows rod-shaped structures that are several hundreds of nanometers long and 20-30 nm wide (Figure 3.1). The TEM result is consistent with a broad probability distribution in hydrodynamic diameter (50-600 nm, with peak at 100 nm;

Figure A3.2) determined by dynamic light scattering (DLS).

3.2 Results and discussion

A UV-Vis absorption spectrum of the aggregate (Figure 3.1) has vibronic bands at

450, 474 and 512 nm which are within 2 nm of those reported for a J-aggregate of all- trans 3R,3'R-zeaxanthin diacetate.18 A recent theoretical study on a closely related lutein diacetate aggregate has addressed the similarity of the monomer and aggregate absorption spectra, and the observed red shift is attributed to non-resonant dispersion interactions rather than predominant J-type head-to-tail exciton coupling.19 Nonetheless we denote the new aggregate J1 in keeping with the earlier literature, and to distinguish it from a more red-shifted J-aggregate that was characterized in a recent ultrafast study.20

Figure 3.1. (Top) Molecular structure of 3R,3'R zeaxanthin monomer, with UV-Vis absorption spectrum in EtOH (black). The spectrum of the J1-aggregate in 90:10

H2O/THF is shown in red. Dashed lines indicate the 415 nm (pump), and 473 nm or 551 nm (probe) TRRR wavelengths. (Inset) TEM image of J1 aggregates.

Continuous-wave (cw) resonance Raman spectroscopy shows nearly identical peak positions for the aggregate vs. the monomer (Figure 3.2). Therefore the primary conclusion is that the molecular constituents in the aggregate are structurally similar to zeaxanthin monomers in solution, as found previously for other aggregates of zeaxanthin.21 The aggregate shows slightly less Raman activity in the structurally- sensitive C-H out-of-plane wagging modes at 874 cm-1 and 967 cm-1 (Figure 3.2, inset).

This finding suggests that the molecules in the aggregate experience less structural distortion than in solution.15, 22 The 9 cm-1 downshift for the ethylenic band of the aggregate is similar to other shifts for carotenoids in the solid state, and have been attributed to crystal packing effects.22

Figure 3.2. CW-resonance Raman spectra of zeaxanthin monomer in THF, and J1- aggregate in 90:10 H2O/THF excited with 458 nm. The spectra were normalized to equal peak heights of the ethylenic mode. (Inset) Magnified view of the fingerprint region. “*” Residual peak from imperfect subtraction of THF solvent.

TRRR was employed to distinguish and quantify excited state species. Samples were photoexcited at 415 nm and probed at 551 or 473 nm. A spectrum of the excited monomer in EtOH (Figure 3.3A) shows a highly upshifted ethylenic band at ~1790 cm-1,

23, 24 which is a signature of the S1 excited state. The S1 decay and ground-state bleach recovery times are consistent with a ~10 psec lifetime for monomer β-carotene and zeaxanthin.25, 26 By contrast, the resonance Raman spectra of the aggregate (Figure 3.3B,

C) do not show S1 features. Difference spectra acquired with a 551 nm probe at various delay times ranging from >20 psec to 3 nsec are dominated by an ethylenic band peaked at 1503 cm-1 and a band in the C-C fingerprint region peaked at 1127 cm-1. Additional bands in the fingerprint region have maxima at 1009, 1192 and 1239 cm-1. Spectra acquired with 473 nm probe have nearly identical peak positions as those found with 551 nm (Figure 3.3C). Furthermore, the positions of the maxima are nearly identical to those

96 of triplet β-carotene monomer.27 The frequencies of the long-lived species do not match those of carotenoid cation or anion radicals, thus charge-separation is not supported.28, 29

Figure 3.3. TRRR spectra of (A) monomer S1 state, probed at 551 nm. (B) J1-aggregate in 90:10 H2O/THF probed at various delays between 415 nm pump and 551 nm probe. “T” indicates bands from the triplet state. (C) Same as (B) but probed with 473 nm.

Significant evolution of the difference spectra is evident at pump-probe delay times from 0 psec to times later than 12 psec. The changes are analyzed by fitting the 4 psec spectra to a set of Lorentzian functions (Figure 3.3B, 3C). Short-lived bands centered at 1512 and ~1154 cm-1 that do not coincide with the triplet bands are also

97 apparent. To gain further insight into the dynamics of the system, we plot the integrated area of the ethylenic band growth and depletion at each time point (Figure 3.4). The points are fit to a function which includes a 5.6 psec Gaussian instrument response and several variable exponential terms. The time constants that are found upon fitting are 5-7 psec, 600-700 psec, and >3 nsec, in the proportions listed in Figure 3.4.

Figure 3.4. Relative Raman scattering intensity in the ethylenic band region. (Upper plot) Kinetics of positive ethylenic band intensity. (Lower plot) Kinetics of bleached ground state intensity centered at 1519 cm-1. The insets are magnifications at early times.

The kinetics and spectral analysis support the following picture: The triplet features appear at the earliest (0 psec) delay of pump and probe, and reach a maximum within 4 psec. The ground state depletion at ~1519 cm-1 reaches a minimum on a similar time-scale. The rapid kinetics are consistent with triplet formation via an allowed singlet fission process. The precursor state is not seen, but could be a hot S1 state that is too short-lived to be detected. The prompt <10 psec decay of positive ethylenic intensity centered at 1503-1512 cm-1 has two likely origins. First, prompt annihilation of triplets

formed by singlet fission that do not diffuse apart can occur via T1+T1 S0 (hot) + T2, in which the T2 state is expected to quickly relax to T1. Second, vibrational cooling of hot

S0 produced by the annihilation can also cause the <10 psec decay. The bands at 1512

-1 -1 cm and 1154 cm are likely signatures of a hot S0 state. The 600-700 psec kinetic component is attributed to triplets that escape immediate recombination, and subsequently annihilate or decay back to the ground state. The existence of a >3 nsec component support the assignment of the long-lived species to a triplet state in the aggregate.

The analyses of triplet QY depend upon quantification of the decreased scattering from the ground-state ethylenic band after photoexcitation. The depletion at 4 psec was measured for the monomer and aggregate samples as a function of percentage excitation

(Figure 3.5). The percentage excitation is determined from Beer’s law, the pump energy, and the laser beam cross-sectional area. The ground-state depletion percentage is determined from the subtraction coefficient of a probe-only spectrum that gives the best removal of ground state from the pump+probe spectra (see the Appendix). Both the monomer and aggregate show linear depletion up to the 20% excitation ratio explored here. It is significant that the extent of depletion for the aggregate at these powers is twice that of the monomer, and within experimental error the slopes of the lines in Figure

3.5 have a ratio of 2:1. There are two possible explanations for this result. First, one would expect the depletion of two chromophores per absorbed photon in the case of highly efficient intermolecular singlet fission. If annihilation occurred after fission, the bleach would still remain, as long as the hot S0 ethylenic frequency is significantly shifted from the parent ground state frequency at 1519 cm-1. Thus the 2-fold greater slope

99 for the J1 aggregate is indirect evidence for a ~200% triplet yield, i.e. quantitative intermolecular singlet fission.

Figure 3.5. Percentage of ground state depletion for 415 nm photoexcitation, with varying pump energies at 4 psec time delay. The excitation % is determined from spot size measurements and laser pulse energy; the depletion is determined from difference spectra.

A second possible origin of a 2-fold greater magnitude for the depletion of the aggregate is that the transition dipole of exciton-coupled systems is theoretically expected

0.5 to scale as (Ncoh) , where Ncoh is the number of coherently-coupled molecules in the exciton state.30 Therefore, the TRRR transient bleach signals of the exciton-coupled zeaxanthin aggregate could be enhanced relative to the monomer signal by a factor of

2 (Ncoh) , since four interactions with the pump and probe fields are involved.

Enhancements of this kind have been discussed for pump-probe transient absorption spectroscopy of strongly coupled J-aggregates of pseudo-isocyanine (PIC) dyes.30

100

However, we consider this possibility unlikely in the case of J1-zeaxanthin for two reasons. First, the cross-section per molecule (Figure 3.1) closely corresponds to the monomer cross-section in most spectral regions, and particularly at the 415 nm pump wavelength; thus the J1-zeaxanthin is qualitatively different from PIC in this regard.

Second, the QY result for J1 is independent of whether the probe is resonant with the 473 nm monomeric region, or significantly red-shifted to 551 nm. Together the two factors suggest that the zeaxanthin J1-aggregate is weakly exciton-coupled19 and can be treated as if Ncoh ≈ 1. In that case, the pump pulse in our experiments is a linear actinic pulse for

< 20% excitation.

A second approach for analyzing the triplet QY incorporates the triplet excited- state band area along with the ground state depletion (See the Appendix). First the ratio of the triplet and ground state resonance Raman cross-sections is found from the percentage of ground state depletion and the ratio of triplet/ground state ethylenic band intensities at 3 nsec. Next, the percentage of molecules in the triplet excited state can be determined at the 4 psec pump-probe delay, assuming that (1) the triplet cross section remains the same at 4 psec as for 3 nsec and (2) the area of the 1503 cm-1 Lorentzian band is a good indicator of triplet population. When the percentage of triplets is divided by the excitation ratio the resulting triplet QY at 4 psec is ~90% per absorbed photon, at the lowest pump energy. Similar results are obtained for either 473 nm or 551 nm probe.

The 90% estimate is obviously lowered by any annihilation that causes loss of scattering intensity at 1503 cm-1 during the 4 psec delay. Without annihilation, the direct triplet yield could reach the ~200% found by examination of ground-state depletion (i.e. slopes of Figure 3.5).

101

Triplet yields as high as 30% by singlet fission have been reported in carotenoids of light harvesting centers.15, 31 However in these systems the absence of proximal carotenoids rules out the possibility of intermolecular homofission. Instead, the recent reports support an intramolecular fission mechanism which is enhanced by distortion of the carotenoid.15, 32, 33 An alternative mechanism is singlet heterofission between a carotenoid and nearby bacteriochlorophyll molecule.31 By contrast, the carotenoid aggregates describe here show no sign of structural distortion, and have higher triplet QY.

3.3 Conclusions

In summary, picosecond TRRR is found to be a powerful tool for identifying and quantifying the triplet yields in a self-assembled carotenoid aggregate. The spectra reveal that J1-zeaxanthin forms triplets in < 4 psec by a highly efficient intermolecular singlet fission process, possibly via a short-lived hot S1 state. The triplet QY of 90-200% derived assuming weak coupling is remarkable in view of the 0.2% triplet QY for monomeric zeaxanthin.34 The influence of aggregate structure on QY, and further investigation of the exciton coherence number are aims of ongoing experiments which will be reported elsewhere.

Acknowledgement

Chapter 3, in full, is a reprint of the material as it appears in the Journal of the

American Chemical Society, 2010, 132(40), pp 13988-13991, by Chen Wang and

Michael J. Tauber. The author of this dissertation conducted the experiments, and was a coauthor of the paper.

102

References

1. Nozik, A. J., Multiple exciton generation in semiconductor quantum dots. Chemical Physics Letters 2008, 457, (1-3), 3-11.

2. McGuire, J. A.; Joo, J.; Pietryga, J. M.; Schaller, R. D.; Klimov, V. I., New aspects of carrier multiplication in semiconductor nanocrystals. Accounts of Chemical Research 2008, 41, (12), 1810-1819.

3. Paci, I.; Johnson, J. C.; Chen, X. D.; Rana, G.; Popović, D.; David, D. E.; Nozik, A. J.; Ratner, M. A.; Michl, J., Singlet fission for dye-sensitized solar cells: Can a suitable sensitizer be found? Journal of the American Chemical Society 2006, 128, (51), 16546-16553.

4. Schwerin, A. F.; Johnson, J. C.; Smith, M. B.; Sreearunothai, P.; Popovic, D.; Cerny, J.; Havlas, Z.; Paci, I.; Akdag, A.; MacLeod, M. K.; Chen, X. D.; David, D. E.; Ratner, M. A.; Miller, J. R.; Nozik, A. J.; Michl, J., Toward designed singlet fission: electronic states and photophysics of 1,3-diphenylisobenzofuran. Journal of Physical Chemistry A 2010, 114, (3), 1457-1473.

5. Singh, S.; Jones, W. J.; Siebrand, W.; Stoicheff, B. P.; Schneider, W. G., Laser generation of excitons and fluorescence in anthracene crystals. Journal of Chemical Physics 1965, 42, (1), 330-342.

6. Swenberg, C. E.; Geacintov, N. E., Exciton interactions in organic solids. In Organic Molecular Photophysics, Birks, J. B., Ed. John Wiley & Sons: London, 1973; Vol. 1, pp 489-564.

7. Jundt, C.; Klein, G.; Sipp, B.; Le Moigne, J.; Joucla, M.; Villaeys, A. A., Exciton dynamics in pentacene thin films studied by pump-probe spectroscopy. Chemical Physics Letters 1995, 241, (1-2), 84-88.

8. Müller, A. M.; Avlasevich, Y. S.; Schoeller, W. W.; Müllen, K.; Bardeen, C. J., Exciton fission and fusion in bis(tetracene) molecules with different covalent linker structures. Journal of the American Chemical Society 2007, 129, (46), 14240-14250.

9. Thørsmolle, V. K.; Averitt, R. D.; Demsar, J.; Smith, D. L.; Tretiak, S.; Martin, R. L.; Chi, X.; Crone, B. K.; Ramirez, A. P.; Taylor, A. J., Morphology effectively controls singlet-triplet exciton relaxation and charge transport in organic semiconductors. Physical Review Letters 2009, 102, (1), 017401.

10. Kraabel, B.; Hulin, D.; Aslangul, C.; Lapersonne-Meyer, C.; Schott, M., Triplet exciton generation, transport and relaxation in isolated polydiacetylene chains: subpicosecond pump-probe experiments. Chemical Physics 1998, 227, (1-2), 83-98.

103

11. Österbacka, R.; Wohlgenannt, M.; Shkunov, M.; Chinn, D.; Vardeny, Z. V., Excitons, polarons, and laser action in poly(p-phenylene vinylene) films. Journal of Chemical Physics 2003, 118, (19), 8905-8916.

12. Lanzani, G.; Stagira, S.; Cerullo, G.; De Silvestri, S.; Comoretto, D.; Moggio, I.; Cuniberti, C.; Musso, G. F.; Dellepiane, G., Triplet exciton generation and decay in a red polydiacetylene studied by femtosecond spectroscopy. Chemical Physics Letters 1999, 313, (3-4), 525-532.

13. Agostini, G.; Corvaja, C.; Giacometti, G.; Pasimeni, L., Optical, zero-field ODMR abd EPR studies of the triplet states from singlet fission in biphenyl-TCNQ and biphenyl-tetrafluoro-TCNQ charge transfer crystals. Chemical Physics 1993, 173, (2), 177-186.

14. Watanabe, S.; Furube, A.; Katoh, R., Generation and decay dynamics of triplet excitons in Alq3 thin films under high-density excitation conditions. Journal of Physical Chemistry A 2006, 110, (34), 10173-10178.

15. Papagiannakis, E.; Das, S. K.; Gall, A.; van Stokkum, I. H. M.; Robert, B.; van Grondelle, R.; Frank, H. A.; Kennis, J. T. M., Light harvesting by carotenoids incorporated into the B850 light-harvesting complex from Rhodobacter sphaeroides R- 26.1: Excited-state relaxation, ultrafast triplet formation, and energy transfer to bacteriochlorophyll. Journal of Physical Chemistry B 2003, 107, (23), 5642-5649.

16. Christensen, R. L., The electronic states of carotenoids. In The Photochemistry of the Carotenoids, Frank, H. A.; Young, A. J.; Britton, G.; Cogdell, R. J., Eds. Kluwer Academic Publishers: Dordecht, 1999; Vol. 8, pp 137-157.

17. Josue, J. S.; Frank, H. A., Direct determination of the S1 excited-state energies of xanthophylls by low-temperature fluorescence spectroscopy. Journal of Physical Chemistry A 2002, 106, (19), 4815-4824.

18. Zsila, F.; Bikádi, Z.; Deli, J.; Simonyi, M., Chiral detection of carotenoid assemblies. Chirality 2001, 13, (8), 446-453.

19. Spano, F. C., Analysis of the UV/Vis and CD spectral line shapes of carotenoid assemblies: Spectral signatures of chiral H-aggregates. Journal of the American Chemical Society 2009, 131, (12), 4267-4278.

20. Billsten, H. H.; Sundström, V.; Polívka, T., Self-assembled aggregates of the carotenoid zeaxanthin: time-resolved study of excited states. Journal of Physical Chemistry A 2005, 109, (8), 1521-1529.

21. Salares, V. R.; Young, N. M.; Carey, P. R.; Bernstein, H. J., Excited-state (exciton) interactions in polyene aggregates - Resonance Raman and absorption spectroscopic evidence. Journal of Raman Spectroscopy 1977, 6, (6), 282-288.

104

22. Saito, S.; Tasumi, M.; Eugster, C. H., Resonance Raman spectra (5800-40 cm-1) of all-trans and 15-cis isomers of β-carotene in the solid-state and in solution - Measurements with various laser lines from ultraviolet to red. Journal of Raman Spectroscopy 1983, 14, (5), 299-309.

23. Orlandi, G.; Zerbetto, F.; Zgierski, M. Z., Theoretical analysis of spectra of short polyenes. Chemical Reviews 1991, 91, (5), 867-891.

24. McCamant, D. W.; Kukura, P.; Mathies, R. A., Femtosecond time-resolved stimulated Raman spectroscopy: Application to the ultrafast internal conversion in beta- carotene. Journal of Physical Chemistry A 2003, 107, (40), 8208-8214.

25. Wasielewski, M. R.; Kispert, L. D., Direct measurement of the lowest excited singlet-state lifetime of all-trans beta-carotene and related carotenoids. Chemical Physics Letters 1986, 128, (3), 238-243.

26. Polívka, T.; Sundström, V., Ultrafast dynamics of carotenoid excited states - From solution to natural and artificial systems. Chemical Reviews 2004, 104, (4), 2021-2071.

27. Hashimoto, H.; Koyama, Y., Time-resolved resonance Raman spectroscopy of triplet β-carotene produced from all-trans, 7-cis, 9-cis, 13-cis, and 15-cis isomers and high-pressure liquid-chromatography analyses of photoisomerization via the triplet state. Journal of Physical Chemistry 1988, 92, (8), 2101-2108.

28. Jeevarajan, A. S.; Kispert, L. D.; Chumanov, G.; Zhou, C.; Cotton, T. M., Resonance Raman study of carotenoid cation radicals. Chemical Physics Letters 1996, 259, (5-6), 515-522.

29. Jin-Yeol, K.; Furukawa, Y.; Tasumi, M., Electronic absorption and Raman studies of the radical anion and dianion of a polyene molecule (19,19',20,20'-telranor-β, β- carotene). Chemical Physics Letters 1997, 276, (5-6), 418-422.

30. Knoester, J.; Spano, F., Theory of pump-probe spectroscopy of molecular J- aggregates. In J-Aggregates, Kobayashi, T., Ed. World Scientific Publishing Co.: Singapore, 1996; pp 111-160.

31. Angerhofer, A., Electron magnetic resonance of carotenoids. In The Photochemistry of the Carotenoids, Frank, H. A.; Young, A. J.; Britton, G.; Cogdell, R. J., Eds. 1999; pp 203-222.

32. Gradinaru, C. C.; Kennis, J. T. M.; Papagiannakis, E.; van Stokkum, I. H. M.; Cogdell, R. J.; Fleming, G. R.; Niederman, R. A.; van Grondelle, R., An unusual pathway of excitation energy deactivation in carotenoids: Singlet-to-triplet conversion on an ultrafast timescale in a photosynthetic antenna. Proceedings of the National Academy of Sciences of the United States of America 2001, 98, (5), 2364-2369.

105

33. Tavan, P.; Schulten, K., Electronic excitations in finite and infinite polyenes. Physical Review B 1987, 36, (8), 4337-4358.

34. Nielsen, B. R.; Mortensen, A.; Jørgensen, K.; Skibsted, L. H., Singlet versus triplet reactivity in photodegradation of C-40 carotenoids. Journal of Agricultural and Food Chemistry 1996, 44, (8), 2106-2113.

106

Appendix of Chapter 3

1. Characterization of monomer zeaxanthin. Unless otherwise noted, all solvents were spectroscopic grade from Fisher or

Sigma-Aldrich chemical, and used as received. All-trans 3R,3'R-Zeaxanthin was purchased from CaroteNature (HPLC 97% pure) and used as received. The purity of the sample was confirmed by several methods. First, UV-Vis absorption spectra (Figure 3.1) were collected with a UV-Vis-NIR spectrometer (Shimadzu UV-3600). The cis-band at

340 nm is very weak, consistent with the presence of minimal cis-isomer. Second, 500

MHz proton NMR spectra of concentrated zeaxanthin samples in CDCl3 (Figure A3.1) showed no sign of resonances characteristic of cis-isomers, or other impurities. Third,

HPLC-MS was performed following an earlier procedure1 with detection by an atmospheric-pressure chemical ionization mass spectrometer.

107

Figure A3.1. NMR spectrum (500 MHz) of zeaxanthin in CDCl3 with concentration of ~ 1mg/ml. The two lower spectra are magnified views of the boxed sections of the upper spectrum.

2. Preparation of the zeaxanthin J1-aggregate and estimation of size by dynamic light scattering (DLS)

Aggregates for spectroscopic measurements were formed by adding water

simultaneously agitating and sonicating (Branson Model 3510) the solution. DLS was measured with a JY Horiba LB-550 particle size analyzer. The solvent for the J1- aggregates is the same 10:90 THF/H2O solution as for all other measurements. The most

108 probable hydrodynamic diameter of the aggregate is 102 nm with a broad distribution extended to 600 nm. The average aggregate size is 120 nm (Figure A3.2).

Figure A3.2. DLS probability distribution for the zeaxanthin J1-aggregate.

3. Circular Dichroism (CD) spectroscopy of the zeaxanthin J1-aggregate.

Circular dichroism spectra were collected with an AVIV model 215 CD spectrometer. The sample was held in a quartz cuvette with 1-mm pathlength. The monomer has no CD signal in the main visible absorption band, since the two chiral centers of zeaxanthin are not part of the conjugated polyene system. However the CD spectrum of the J1-aggregate shown in Figure A3.3 reveals a strong positive Cotton effect

(Δε = 430 OD M-1cm-1 at 481 nm; 259 OD M-1cm-1 at 516 nm) and a broad negative

Cotton effect (Δε = -224 OD M-1cm-1 at 419 nm). The strong CD signals are attributed to exciton coupling within the supramolecular chiral structure of the aggregate.2, 3

109

Figure A3.3. Circular dichroism spectrum of zeaxanthin J1-aggregate in 10:90

THF/H2O solution.

4. Continuous-wave resonance Raman spectroscopy of the zeaxanthin J1-aggregate.

Continuous-wave (CW) resonance Raman spectroscopy was performed with 458 nm excitation from an Argon/Krypton ion laser (Laser Innovations Innova-70C). The sample was flowed through a capillary and illuminated with 3.5 mW of laser power. The

Raman-scattered light was collected using a 90-degree scattering geometry. A 458 nm edge filter (Semrock) was inserted after the collection/focusing lenses to reject Rayleigh scattering. A 320 mm focal length spectrograph (Horiba Jobin-Yvon iHR-320) equipped with a 1800 gr/mm holographic grating was employed to disperse the Raman signal, and signal was detected with an open-electrode CCD detector (Horiba Jobin-Yvon Synapse).

Raman shifts were calibrated by reference to known solvent peak assignments. The spectra of the zeaxanthin J1-aggregate (Figure A3.4) show only slight frequency shifts relative to the monomer. The most pronounced change is a 9 cm-1 downshift of the

110 ethylenic band for the aggregate, relative to the 1525 cm-1 band of the monomer. The shifts are similar to those seen for solid state spectra and have been attributed to crystal packing effects.4 As found earlier for another zeaxanthin aggregate, the ground-state zeaxanthin molecule in the aggregates is structurally similar to the monomer in solution.5

5. Picosecond time-resolved resonance Raman spectroscopy.

The time-resolved resonance-Raman spectrometer is based upon a Ti:sapphire oscillator (Spectra-Physics Mai Tai) which seeds a Ti:sapphire regenerative amplifier

(Spectra-Physics Spitfire XP-Pro). The system generates 120 fsec pulses >3 mJ at a 1 kHz repetition rate. The 830 nm beam is then sent to a specialized doubling & bandwidth-narrowing unit (Light Conversion SHBC). Briefly, the SHBC creates positive and negative chirps along two separate beam paths of the fundamental. The oppositely- chirped beams are combined in a BBO crystal to generate narrow-bandwidth (<10 cm-1), high energy (~0.8 mJ) 415 nm light with pulse duration of ~3 psec. 40% of the beam was

Raman-shifted using a 0.5 m pipe filled with 1000 psi D2 pipe to produce the second-

Stokes 551 nm line for the probe. A small fraction of the 415 nm beam was used as the pump. Waveplates and polarizers were employed to adjust the pump and probe energies.

The Raman spectra were recorded at the lowest possible energy. Spectra shown in Figure

3.2 are collected with a pump energy of 55 nJ/pulse and probe energy of 170 nJ/pulse.

The pump energy for data in Figure 3.3 varied from 11 to 124 nJ/pulse. The probe pulse energy was constant at 170 nJ/pulse.

A cross-correlation measurement was performed to determine the overall instrument response time. The pump and probe beam were focused on a BBO crystal to obtain the sum-frequency generation (SFG) at 237 nm. By monitoring the power of the

111

SFG signal, a Gaussian cross-correlation time of 5.6 psec was determined (Figure A3.4).

This cross-correlation is the dominant component of our instrument response.

Figure A3.4. The cross-correlation of 415 nm pump and 551 nm probe beams.

The pump and probe beams were combined with a 50/50 beamsplitter and directed toward the sample region. The beams were focused at the sample with two cylindrical lenses with focal lengths 150 mm and 75 mm. The spot size was measured multiple times by monitoring the transmitted beam power upon translation of a razor blade through the sample point with a micrometer. The spot size for the pump was 320 +

5 μm along the optic axis, and 56 + 5 μm in the transverse direction. For the probe the respective measurements were 263 + 5 μm and 41 + 5 μm, thus the probe area is 60% smaller than the pump. A 30 mL volume of sample containing 10 µM zeaxanthin

(concentration in units of monomer equivalent) was pumped continuously through a 2 x 2 mm quartz capillary with a syringe pump (KDS Legato 270). The flow speed of 11 cm/s

112 was sufficient to replenish the sample at each laser shot. Raman scattering was collected in a 90-degree geometry with an F/1.2 camera lens (Canon) with focal length 55mm. The

Rayleigh scattering of the probe was attenuated with a 568 nm edge filter (Semrock). The

Raman light was focused onto the entrance slit of a spectrograph (Horiba Jobin Yvon iHR-320) with an F/4, 200 mm focal length achromatic doublet lens. The entrance slit opening was 200 μm (13 cm-1). The detector was a back-illuminated CCD (Andor

Newton DU920N-BU) thermoelectrically cooled to -90 oC.

The spectra were collected as follows: A pump+probe spectrum was collected for

T min, followed by a 2xT min collection of probe-only spectra and finally another T min pump+probe collection. The time T ranged from 10 to 30 minutes, depending on pump and probe powers. A pump-only spectrum was collected once for the whole series since the pump induced background is constant. Different time delays between pump and probe were obtained with an optical delay stage.

113

6. TRRR subtraction procedure.

An example of subtraction procedure for time-resolved Raman spectra analysis is shown in Figure A3.5. All spectra are the sum of 30 acquisitions, each 40 sec. Spectrum

C is the difference (pump+probe) - (pump-only). Spectrum E is obtained by subtracting a fraction q1 of the solvent-only (spectrum D) from spectrum C. The subtraction factor q1 is less than unity because of self-absorption by zeaxanthin. The optimal factor is determined by monitoring the 920 cm-1 peak of the THF solvent. Spectrum E contains both excited- state and ground-state features. Spectrum F is the difference of the probe-only and scaled solvent-only spectrum, (probe-only) - q2 (solvent-only). Pure excited state spectrum G was obtained by the difference (spectrum E) - q3 (Spectrum F). The factor q3 was determined by monitoring the ethylenic peak region, as shown in the inset of Figure A3.5.

Correct subtraction should remove all the features from ground state at 1519 cm-1 without over-subtraction. The best subtraction was obtained with q3 = 0.87. From the subtraction procedure, the ground state depletion can be determined as 1- q3. A baseline H was fitted and subtracted to provide background free excited state spectrum I.

114

800 1000 1200 1400 1600 1800 2000

25 * A 20 B

5 C

CCD Counts (X10000) Counts CCD C = A - B D 0 15031501 10 0.83 0.85 0.870.87

0.89

0.91 8

1350 1400 1450 1500 1550 1600

6 Raman shift (cm -1)

4 E E=C - q x D F 1 2

CCD Conunts (X10000) Conunts CCD G 0 G=E - q x F H 3

2 I I = G - H

0 CCDcounts (X1000) 800 1000 1200 1400 1600 1800 2000 -1 Raman shift (cm )

Figure A3.5. Spectrum A = pump+probe. Spectrum B = pump-only. Spectrum C = the difference A-B. Spectrum D = solvent-only spectrum. Spectrum E is obtained by subtracting a fraction (q1) of the solvent-only spectrum D from C. Spectrum F is probe- only minus a scaled (q2) solvent spectrum. Spectrum G = pure excited state, obtained by removing a scaled (q3) probe-only ground state spectrum F from E. H is the baseline for the excited state spectrum. I is the background-free excited state spectrum. “*” indicates strong solvent peak.

7. TRRR spectra of zeaxanthin monomer

TRRR spectra of zeaxanthin monomer are shown in Figure A3.6. Both S1 and hot S0 features contribute to the spectra. The S1 state has a strongly upshifted ethylenic band at

~1790 cm-1. Note: the instrument response for this experiment is 10 psec instead of 5.6

115

psec; deconvolution (not shown) gives the expected S1 decays of ~10 ps. No excited state signal is evident after 60 psec.

Figure A3.6. Representative TRRR spectra of zeaxanthin monomer in EtOH. Instrument response time is 10 psec for this run. *Indicates region where Raman scattering from solvent is strong.

8. Determination of the excitation ratio

The first step in calculating the triplet quantum yield (QY) is determining the number of pump photons that are absorbed by the sample, within the region that is probed with 551 nm probe light. From the number of absorbed photons, and the concentration of zeaxanthin, one can compute the percent of zeaxanthin monomer molecules that is excited, or “excitation ratio”.

Beer’s law is found to hold true with our very low pump pulse energies, thus the number of absorbed photons Nabs is:

116

E N  (110OD ) (1) Abs hc  where E is the pump energy, ODλ is the optical density of the sample at the excitation wavelength λ, c is the speed of light in vacuum and h is Planck’s constant.

The excitation ratio fexc for zeaxanthin molecules in the probed region can be calculated from:

 E(110OD )  N Abs hc fexc   (2) NTot Al C  N A

In equation 2, the probe beam size is A. This area is determined experimentally by translating a razor blade through the probe beam at the sample focus; the micrometer positions at 10% to 90% ranges of transmitted probe power are used to calculate an elliptical area (π)(d1/2)(d2/2). The energy E′ is the portion of pump energy contained within the same probe area A, determined experimentally by the same razor blade and micrometer positions. NTot is the total number of zeaxanthin molecules in the probed volume, l is the pathlength through the sample, NA is Avogadro’s number and C is the concentration of zeaxanthin.

9. Determination of the relative Raman cross section ratio of S0 and T1 states.

The next step in quantifying the triplet yields is the estimation of the relative

Raman scattering cross-sections of zeaxanthin in the S0 ground state and in the T1 triplet excited state, at the 551 nm Raman probe wavelength. No Raman spectrum of the triplet state in a carotenoid aggregate has been reported before, thus the cross section is unknown. Furthermore, neither absolute nor relative Raman cross sections of the triplet state for zeaxanthin monomer have been reported previously.

117

We estimate the relative Raman cross section of the triplet and singlet by a simple equation that relates the fraction of triplet species f to the integrated band areas and the T1 differential cross-sections, as written in equation 3:

N I  f  T1  T1 S0 (3) T1 N I ( probe only)  Tot S0 T1

In equation 3, N is the number of zeaxanthin molecules in the triplet state, T1

within the probed volume; NTot is the total number of molecules in the probed volume.

I is the Raman intensity of the triplet ethylenic band integrated over the spectral region T1 centered at 1503 cm-1, and I ( probe only) is the Raman intensity of the probe-only S0 spectrum integrated over the spectral region centered at 1519 cm-1. The differential cross- section of the zeaxanthin ethylenic in the S0 ground state is  , and in the lowest T1 S0 triplet excited state it is  . Possible changes in the depolarization ratio of the triplet T1 versus singlet states are not considered since these will cancel in our analysis. The ratio

 / with 551 nm Raman excitation can be found empirically at the 3 nsec time delay, S0 T1 where one can reasonably assume that any excited state species must be in the T1 state.

At that time point, a typical experiment shows 7.5% ground state depletion within the illuminated area of the probe beam, as determined by the subtraction coefficient that is

-1 obtained from monitoring the S0 ethylenic peak at 1519 cm . Thus, we directly estimate f  7.5% within the probe beam area, at a 3 nsec delay. Furthermore, integration of the T1

-1 ethylenic peaks of the T1 state at 1503 cm , relative to the integrated area of the probe- only S0 ethylenic shows that within the illuminated area of the probe-beam, I / I  12% T1 S0

118

thus from equation 3 we find that the ratio of cross-sections  /  0.63 with 551 nm S0 T1 excitation.

We can now estimate the fractional triplet yields at other time delays, again by use of equation 3 and with the ratio of 551 nm Raman differential cross-sections. For example, at 4 ps time delay where the largest triplet signal at 1503 cm-1 is observed,

I / I  28% . Thus we estimate (0.28)(0.63)= ~18% of zeaxanthin molecules in the T1 S0 region illuminated by the probe are in the T1 excited state at 4 psec.

The final step in the determination of yield of ground state depletion, or triplet yields per absorbed photon involves the ratios of the quantities derived above.

Specifically, the percent of ground state depletion divided by the excitation ratio gives the absolute quantum yield (QY) of ground state depletion (per absorbed photon). The fraction of triplet species f divided by the excitation ratio gives the absolute triplet QY T1

(per absorbed photon). A summary of relevant quantities for the J1-aggregate are listed in

Table S1. The extent of ground state depletion (column #3) and the quantum yield of triplets (column #7) are plotted in Figure 3.3 of the main paper.

119

Table A3.1. Data for J1-aggregate, supporting Figure 3.3 of maintext. a b c d e Pump Excitation Ground State QY of ground IT1/IS0 Triplet Triplet Energy Ratio Depletion state depletion Yield QY (nJ/Pulse) (%) (%) (%) (%) (%)

11 3.1 5.8 190 7.7 4.8 150 22 6.2 12 190 13 8.4 140 31 8.7 16 180 18 12 140 38 11 18 160 21 13 120 58 16 27 170 27 17 110 67 19 29 150 30 19 100 96 27 30 110 28 18 70 124 35 33 90 31 20 60 a Excitation ratios from equation 2. Small corrections are included for minor variation in probe illuminated area on different days. b Ground state depletion % from subtraction coefficient (1- q3) in Figure A3.5. c Ground state depletion % divided by excitation ratio. d From equation 3. e Ratio of triplet yield % and excitation ratio %, rounded to nearest 10%.

10. Adjustment of yields for instrument response time We now refine our yields measured at 4 psec in view of the finite instrument response time. During the pump and probe pulses, the decay of the monomer zeaxanthin

(reference) signal is significant, as is the loss of triplet species in the aggregate due to annihilation. Excited states of the aggregate and monomer reference molecule experience different decay kinetics. In the following, correction factors are derived that allow prediction of ground-state depletion at times shorter than the instrument response time.

The correction factors are found from considering “intrinsic” molecular decay curves, where the finite instrument response time is deconvolved from the experimental kinetics.

120

The recovery of monomer ground state depletion is characterized by 1/e time of

~10 psec. The instrumental response is treated as a Gaussian function with FWHM = 5.6 psec. The convolution of this instrumental response with a single exponential function with 10 psec lifetime yields a kinetic trace that has maximum amplitude at 4 ps which is

65% of the initial amplitude of the intrinsic molecular response (Figure A3.7A). The convolved kinetic trace closely tracks the actual experimental data (not shown).

Therefore, our best prediction of the maximum ground state depletion near t = 0 psec is found by dividing the experimental depletion at 4 psec by a factor of 0.65. The slope of

0.7 + 0.1 in Figure 3.3 must also be divided by the same factor of 0.65 which leads to a corrected slope of 1.1 + 0.2. A slope with this value is in full accord with our expectation that every photon creates a single S1 excited state in the monomer.

The correction for the aggregate excited state population near t=0 is not as straightforward as for the monomer, because the decay is not expected to be first-order, and is therefore dependent upon the triplet population and pump power. We leave the triplet yields uncorrected since the triplet decay rates have not been characterized for all pump powers. Nonetheless, an approximate correction to the slope of the ground state depletion 1.6 + 0.1 is reasonable, since this slope is an average best fit to the depletion over a range of pump powers. The experimental kinetics of Figure A3.7 were measured at a representative pump power with excitation ratio of 14%. The portion of the kinetic due to the 5.6 psec Gaussian instrument response is extracted (Figure A3.7B) which results in an “intrinsic” molecular decay. The observed maximum depletion at 4 psec is a factor of

0.79 reduced relative to the predicted value at t = 0. The slope of 1.6 + 0.1 divided by this correction factor gives a slope of 2.0 + 0.1. The corrected value for the slope is in

121 full agreement with expectations based on a singlet fission process, in which each photon creates two triplet states and depletes two ground state molecules.

The main conclusion is that similar incoming excitation depletes twice as many ground state molecules from the J1-aggregate in comparison with the monomer. We cannot correct the triplet yields in the same way because of the lack of kinetic traces at each power. Nonetheless, the lower limits to the triplet quantum yields reported here are already in excess of 100% and also support the conclusion of high-yield singlet fission in the J1-aggregate.

Figure A3.7. Convolution of instrument response and kinetics of ground state recovery of (A) monomer and (B) J1-aggregate

122

References

1. Epler, K. S.; Sander, L. C.; Ziegler, R. G.; Wise, S. A.; Craft, N. E., Evaluation of Reversed-Phase Liquid-Chromatographic Columns for Recovery and Selectivity of Selected Carotenoids. Journal of Chromatography 1992, 595, (1-2), 89-101.

2. Zsila, F.; Bikadi, Z.; Deli, J.; Simonyi, M., Chiral detection of carotenoid assemblies. Chirality 2001, 13, (8), 446-453.

3. Simonyi, M.; Bikadi, Z.; Zsila, F.; Deli, J., Supramolecular exciton chirality of carotenoid aggregates. Chirality 2003, 15, (8), 680-698.

4. Saito, S.; Tasumi, M.; Eugster, C. H., Resonance Raman spectra (5800-40 cm-1) of all-trans and 15-cis isomers of β-carotene in the solid-state and in solution - Measurements with various laser lines from ultraviolet to red. Journal of Raman Spectroscopy 1983, 14, (5), 299-309.

5. Salares, V. R.; Young, N. M.; Carey, P. R.; Bernstein, H. J., Excited-state (exciton) interactions in polyene aggregates - Resonance Raman and absorption spectroscopic evidence. Journal of Raman Spectroscopy 1977, 6, (6), 282-288.

Chapter 4 Triplet Exciton of Carotenoids Formed by Singlet Fission in a Membrane

4.1 Introduction

Singlet fission is a spin-allowed process whereby a singlet excited state forms a pair of triplet excitons. The potential advantage of singlet fission for photovoltaics has stimulated a revival of research on this topic.1-5 However, to our knowledge, a functional role for singlet fission in biology has not been proposed.

Experimental evidence indicates that singlet fission is responsible for triplet yields as high as 30% in light-harvesting complexes (LHCs) of purple bacteria.6-11 Nevertheless, basic aspects of fission in LHCs, notably the identity of the participating chromophores, remain unclear. Intermolecular singlet fission between two carotenoids, or between a carotenoid and bacteriochlorophyll molecule, have both been proposed in LHCs of

Rhodospirillum rubrum and Rhodobacter sphaeroides.6, 8 Later ultrafast spectroscopic studies on these and other bacterial photosynthetic systems led researchers to propose intramolecular singlet fission, in which two triplet excitons are apparently formed on a single carotenoid molecule.10-14 Most LHCs contain multiple carotenoids; thus intermolecular singlet fission would be supported if the interaction between these chromophores were suitable. Model systems, as presented here, are timely for exploring singlet fission in biology. More generally, the question remains: Does singlet fission serve a function in biological systems?

Our interest in exploring the biological relevance of fission prompted us to examine carotenoid assemblies embedded in membranes. Several factors guided our focus on zeaxanthin in synthetic phospholipid bilayers. First, zeaxanthin is a prevalent

123

124 bchromophore in nature. It is considered the most important xanthophyll for nonphotochemical quenching (NPQ) in plants,15, 16 and is a crucial blue-light absorber and antioxidant in the maculae of primates.17 Second, a high yield of triplets via intermolecular singlet fission was found in nanometer-size aggregates of zeaxanthin.5

Thus, it is feasible that the intermolecular process could occur efficiently within much smaller assemblies of the chromophore. Third, when zeaxanthin is embedded within a lipid bilayer in the gel phase, a significant fraction of the carotenoids self-assemble in small aggregates.18-20 It is reasonable to compare these assemblies in liposomes, to carotenoids in biological systems that are in close proximity to one another.21-23 Finally, zeaxanthin is predominantly disaggregated in the fluid phase of the bilayers.18-20 The possibility of controlling intermolecular singlet fission by changing the bilayer phase is advantageous from the standpoint of fundamental studies.

4.2 Experimental

3R,3'R-zeaxanthin was obtained in pure form as described previously.5 Samples were prepared by co-drying zeaxanthin with the appropriate lipid (Avanti Lipids), followed by resuspension in tricine buffer. Small unilamellar vesicles were generated by sonicating the solution of zeaxanthin and lipid at a temperature above Tm, using a probe sonicator for 30 minutes (50% duty cycle). The appropriate samples were collected after passing the solution through a gel filtration column (10DG, Bio-Rad). The initial concentrations were 2 mol% zeaxanthin and 6.7 mg/mL lipid. The concentrations of zeaxanthin incorporated in the membranes were 0.7 mol% (DPPC) and 1.1 mol %

(POPC).

Steady-state resonance Raman spectra are acquired with an excitation wavelength 125 of 488.0 nm, 2 mW average power. The instrument bandpass is ~5 cm-1. The frequency assignments are calibrated with a neon lamp and are accurate to ±1 cm-1.

Time-resolved resonance Raman spectra were acquired with 415 nm pump

(80~110 nJ/pulse), and 551 nm probe (200 nJ/pulse) beams. Difference spectra are obtained from the subtraction of (pump+probe) – (pump-only) – (scaled probe-only).

Details of the TRRR system and analysis procedure are described elsewhere.5

4.3 Results and discussion

Here, we report spectroscopic evidence that singlet fission occurs within zeaxanthin multimers in the gel phase of synthetic bilayers. Liposomes with approximately 1 mol% incorporated zeaxanthin were synthesized by standard protocols.19

UV-Vis absorption spectra (Figure 4.1) of zeaxanthin in the fluid phase of either DPPC

(at 50 °C) or POPC (at room temperature, RT) are similar to that of the monomer in solution. Prior studies have shown that all-trans zeaxanthin spans the width of the lipid bilayer, and is predominantly monomeric in the fluid phase and at the low concentrations of our studies.19, 20, 24 In the gel phase of DPPC (at RT), the higher energy vibronic bands are enhanced relative to the 0-0 transition, which leads to a hypsochromic shift of the absorption maximum. The UV-Vis characteristics are similar to those previously reported for ~1 mol% incorporated zeaxanthin.19 The changes in vibronic intensities in the gel phase are a typical result of exciton coupling within small assemblies of chromophores.25, 26 Furthermore, a circular dichroism (CD) spectrum of zeaxanthin in the gel phase of DPPC reveals a bisignate couplet in the 350-500 nm spectral region. The

CD signal is caused by exciton coupling within small H-aggregates of zeaxanthin in liposomes.24, 27, 28 By contrast, zeaxanthin in the monomeric form (in the fluid phase at 126 low concentration, or in ethanol) does not show a significant CD signal in the 350-500 nm region. The result is expected because of the diminished exciton coupling between chromophores, and the fact that the chiral centers are not part of the polyene chain.

Figure 4.1. Top: 3R,3'R-zeaxanthin. Left: Model of zeaxanthin (black) in the gel (T <

Tm) and fluid (T > Tm) phases of the bilayer. The membrane structures (orange/grey) are derived from Heller et al.29 Right: UV-Vis absorption (upper panel) and circular dichroism (CD, lower panel) spectra of zeaxanthin in lipid bilayers at room-temperature

(RT) or 50 °C. Lipids are DPPC and POPC, Tm = -2 °C. The percentage of zeaxanthin incorporated is 0.7 mol% (DPPC) and 1.1 mol% (POPC). The monomer absorption and CD spectra at RT in tetrahydrofuran or ethanol solvent, respectively, are shown for comparison.

Steady-state resonance Raman spectra of zeaxanthin embedded in gel or fluid phases are nearly the same as a spectrum acquired with ethanol solvent (Figure 4.2). The most significant difference is a band at ~1135 cm-1 which appears in the gel phase as a 127 shoulder to the main C-C single bond stretch band. A similar downshifted C-C mode is observed in spectra of solid-state all-trans--carotene, and is not an indicator of structural distortion.30 It is important to note that there are no significant changes in the hydrogen out-of-plane (HOOP) wagging bands. These modes are an established indicator of structural distortion in polyenes.12, 30, 31 The HOOP bands of zeaxanthin at ~874 and ~959 cm-1 are nearly identical in the fluid and gel phases, and reveal that the carotenoids are not distorted in the bilayers. The findings support the conclusion that exciton coupling in the gel phase is the cause of the enhanced blue absorption in the UV-Vis spectrum. A similar conclusion was reached in an earlier resonance Raman study of zeaxanthin in

DPPC membranes.32 An alternative explanation that invokes structural distortion and differing Franck-Condon factors in each bilayer phase, is not supported by our spectra.

Figure 4.2. Steady-state resonance Raman spectra of zeaxanthin in lipid bilayers and in ethanol. The excitation wavelength is 488 nm. All spectra are acquired at 25 ºC, where POPC is in the fluid phase and DPPC is in the gel phase. An asterisk (*) marks a residual artifact from subtraction of the solvent-only spectrum.

128

Time-resolved resonance Raman spectroscopy (TRRR) is a powerful tool for probing carotenoid excited state dynamics on the picosecond-to-microsecond timescale.5,

9, 33-36 TRRR difference spectra of zeaxanthin embedded within DPPC in the gel phase

(Figure 4.3A) are characterized by three groups of vibrational features. First, several bands of the triplet are identified by comparison with known monomer triplet spectra.33-35

The strongest modes are the C-C and C=C stretches which have peak intensities at 1127 and 1503 cm-1, respectively. These triplet bands are present at the earliest time delays and persist beyond 150 ps. A second set of features can be assigned to the S1 singlet

36 excited state. The most prominent S1 band is a high-frequency ethylenic mode centered

-1 at ~1785 cm . Kinetic analysis shows that the lifetime of the S1 ethylenic feature is 11 ±

1 ps. A third group of bands are centerd at approximately 1004, 1150, 1273, and 1510

-1 36 cm . These bands can be assigned to hot S0 or other species, and are the topic of an upcoming report. 129

Figure 4.3. TRRR difference spectra of zeaxanthin at room temperature (A) in the gel phase of DPPC lipid bilayers, where zeaxanthin is a mixture of multimers and monomers; and (B) in the fluid phase of the POPC lipid bilayers (black) where zeaxanthin is predominantly monomeric. TRRR spectra of zeaxanthin monomer in cyclohexane are shown for comparison (red). Peak positions of selected vibrational bands that identify T1 triplet and S1 singlet excited states are labelled.

In the fluid phase of the POPC lipid bilayer (Figure 4.3B) only features of S1 and

-1 hot S0 states appear. The characteristic S1 ethylenic band centered at 1785 cm , and its 130

12 ± 1 ps exponential decay are indistinguishable from the S1 characteristics of zeaxanthin in the gel phase of the DPPC bilayer. The kinetics of monomeric zeaxanthin within membranes resemble those found in prior studies of monomeric -carotene or zeaxanthin in organic solvent,37, 38 as well as non-aggregated -carotene in liposomes

39 above the phase transition temperature. The S1 properties observed here are consistent with the known dynamics of xanthophylls related to -carotene: A prompt ~200 fs decay of the optically-pumped S2 state populates an S1 state that persists for approximately 10

40 ps before returning to the ground state. The lack of T1 signatures in monomeric zeaxanthin is expected, because the triplet yield by intersystem crossing is less than 1%.41

Typically a sensitizer or multiple pump-pulses are required to observe carotenoid monomers in the triplet excited state with resonance Raman spectroscopy.9, 33-35

The formation of high-yield triplet excitons in the gel phase, on a timescale that is faster than the ~6 ps response time of our instrument, is consistent with singlet fission.1

Furthermore, the fact that the triplet formation coincides with the appearance of exciton coupling supports an intermolecular mechanism. It is likely that singlet fission occurs in relatively high yield within the subset of zeaxanthin multimers in the gel phase, as in the

WH-aggregates. Additional evidence for intermolecular singlet fission is provided by excitation-dependence data (Figure 4.4). The ground-state depletion reveals that more than one chromophore is bleached per absorbed photon for the lipid bilayer that contains multimers. 131

Figure 4.4. Percentage of ground state bleach, as a function of excitation percentage calculated from the sample absorption, pump pulse energy, and beam cross-section. Pump pulse energies vary from 30 to 100 nJ. A (0, 0) point is included when fitting the data. The depletion of zeaxanthin in DPPC (primarily multimers) follows a line with a slope that is 1.6-fold greater than that for zeaxanthin in POPC (primarily monomers). No correction is made for ground-state recovery during the instrument response time; however the ratio of the slopes is expected to remain the same with this correction.

The vibrational S1 feature in Figure 4.3A is best assigned to a subset of monomeric carotenoids that exists in the gel phase. This explanation is consistent with the finding of identical vibrational frequencies and lifetimes of the signature S1 ethylenic band, in either phase of the lipid bilayer. An inhomogeneous distribution of monomers and small multimers in the gel phase agrees with prior descriptions of these systems.18, 20

An alternative scenario, in which the distribution of chromophores is uniform and the formation of triplet and S1 excited states proceeds on parallel paths, is possible but not well-supported by the data here. 132

A key conclusion from our data is that the phase of the bilayer, and therefore the aggregation state of the carotenoids, strongly impacts the formation of triplet excitons.

The fact that coupling of carotenoids enables the formation of triplet excited states in high yield supports the hypothesis that singlet fission could be a mechanism for dissipating energy in biological systems. The carotenoid absorbs ~2.5 eV/photon upon formation of the S2 excited state. Intermolecular singlet fission would partition a majority of the excited-state energy over two carotenoids, and yield two triplet excited states.

Each carotenoid with nine or more conjugated double bonds is a reservoir of up to ~0.9

42 eV in the T1 excited state. The low excited state energy prevents formation of harmful singlet oxygen. Furthermore, the low triplet energy level and slow return to the ground state could minimize local heating and photodestruction during relaxation. An additional benefit of singlet fission is that the process could be modulated by changes in the relative

1, 3 orientation or spacing of the proximal carotenoids. Our proposal is not intended as an alternative to other mechanisms of photoprotection.43, 44 Instead, the finding here motivates further investigation of biological systems in which membrane-embedded carotenoids are known to exist in close proximity with one another, such as in some types of LHCs,21-23 or in primate maculae.45

4.4 Conclusions

In conclusion, we discover high-yield formation of triplet excited states in zeaxanthin aggregates, and its multimers that exist in the gel phase of a membrane. The triplet formation occurs rapidly, within several picoseconds. The triplet yield in the aggregate is as high as 90 to 200%, but lower in the lipid vesicles. For the lipsome systems, singlet fission only happens for zeaxanthin in the gel phase when there is 133 exciton coupling. No triplet formation was observed by TRRR for monomeric zeaxanthin in either solution phase or in the fluid phase of the bilayer. These observations provide strong support for intermolecular singlet fission in a biomimetic system. In the following chapters, we will focus on investigating the fission process in the self-aseembled aggregates.

Acknowlegement

Chapter 4, in full, is a reprint of the material as it appears in ChemPhysChem,

2011, 12(16), pp 2891-2894, by Chen Wang, Diana E. Schlamadinger, Varsha Desai, and

Michael J. Tauber. The author of this dissertation conducted most of the experiments, and was a coauthor of the paper.

References

1. Smith, M. B.; Michl, J., Singlet fission. Chemical Reviews 2010, 110, (11), 6891- 6936.

2. Paci, I.; Johnson, J. C.; Chen, X. D.; Rana, G.; Popovic, D.; David, D. E.; Nozik, A. J.; Ratner, M. A.; Michl, J., Singlet fission for dye-sensitized solar cells: Can a suitable sensitizer be found? Journal of the American Chemical Society 2006, 128, (51), 16546-16553.

3. Greyson, E. C.; Vura-Weis, J.; Michl, J.; Ratner, M. A., Maximizing singlet fission in organic dimers theoretical investigation of triplet yield in the regime of localized excitation and fast coherent electron transfer. Journal of Physical Chemistry B 2010, 114, (45), 14168-14177.

4. Johnson, J. C.; Nozik, A. J.; Michl, J., High triplet yield from singlet fission in a thin film of 1,3-diphenylisobenzofuran. Journal of the American Chemical Society 2010, 132, (46), 16302-16303.

5. Wang, C.; Tauber, M. J., High-yield singlet fission in a zeaxanthin aggregate observed by picosecond resonance Raman spectroscopy. Journal of the American Chemical Society 2010, 132, (40), 13988-13991. 134

6. Rademaker, H.; Hoff, A. J.; Vangrondelle, R.; Duysens, L. N. M., Carotenoid triplet yields in normal and deuterated Rhodospirillum-rubrum. Biochimica Et Biophysica Acta 1980, 592, (2), 240-257.

7. McGann, W. J.; Frank, H. A., Magnetic field effects on the fluorescence of mutant strains of Rhodopseudomonas capsulata. Biochimica et Biophysica Acta - Bioenergetics 1983, 725, (1), 178-189.

8. Kingma, H.; van Grondelle, R.; Duysens, L. N. M., Magnetic-field effects in photosynthetic bacteria. II. Formation of triplet states in the reaction center and the antenna of Rhodospirillum rubrum and Rhodopseudomonas sphaeroides. Magnetic-field effects. Biochimica et Biophysica Acta - Bioenergetics 1985, 808, (3), 383-399.

9. Hayashi, H.; Noguchi, T.; Tasumi, M.; Atkinson, G. H., Vibratioanl spectroscopy of excited electronic states in carotenoids invivo - Picosecond time-resolved resonance Raman scattering. Biophysical Journal 1991, 60, (1), 252-260.

10. Gradinaru, C. C.; Kennis, J. T. M.; Papagiannakis, E.; van Stokkum, I. H. M.; Cogdell, R. J.; Fleming, G. R.; Niederman, R. A.; van Grondelle, R., An unusual pathway of excitation energy deactivation in carotenoids: Singlet-to-triplet conversion on an ultrafast timescale in a photosynthetic antenna. Proceedings of the National Academy of Sciences of the United States of America 2001, 98, (5), 2364-2369.

11. Papagiannakis, E.; Kennis, J. T. M.; van Stokkum, I. H. M.; Cogdell, R. J.; van Grondelle, R., An alternative carotenoid-to-bacteriochlorophyll energy transfer pathway in photosynthetic light harvesting. Proceedings of the National Academy of Sciences of the United States of America 2002, 99, (9), 6017-6022.

12. Papagiannakis, E.; Das, S. K.; Gall, A.; van Stokkum, I. H. M.; Robert, B.; van Grondelle, R.; Frank, H. A.; Kennis, J. T. M., Light harvesting by carotenoids incorporated into the B850 light-harvesting complex from Rhodobacter sphaeroides R- 26.1: Excited-state relaxation, ultrafast triplet formation, and energy transfer to bacteriochlorophyll. Journal of Physical Chemistry B 2003, 107, (23), 5642-5649.

13. Papagiannakis, E.; van Stokkum, I. H. M.; Vengris, M.; Cogdell, R. J.; van Grondelle, R.; Larsen, D. S., Excited-state dynamics of carotenoids in light-harvesting complexes. 1. Exploring the relationship between the S-1 and S* states. Journal of Physical Chemistry B 2006, 110, (11), 5727-5736.

14. Polivka, T.; Sundstrom, V., Dark excited states of carotenoids: Consensus and controversy. Chemical Physics Letters 2009, 477, (1-3), 1-11.

15. Demmig-Adams, B., Carotenoids and photoprotection in plants: A role for the xanthophyll zeaxanthin. Biochimica et Biophysica Acta (BBA) - Bioenergetics 1990, 1020, (1), 1-24. 135

16. Frank, H. A.; Bautista, J. A.; Josue, J. S.; Young, A. J., Mechanism of nonphotochemical quenching in green plants: energies of the lowest excited singlet states of violaxanthin and zeaxanthin. Biochemistry 2000, 39, (11), 2831-2837.

17. Krinsky, N. I.; Landrum, J. T.; Bone, R. A., Biologic mechanisms of the protective role of lutein and zeaxanthin in the eye. Annual Review of Nutrition 2003, 23, 171-201.

18. Okulski, W.; Sujak, A.; Gruszecki, W. I., Dipalmitoylphosphatidylcholine membranes modified with zeaxanthin: numeric study of membrane organisation. Biochimica Et Biophysica Acta-Biomembranes 2000, 1509, (1-2), 216-228.

19. Sujak, A.; Okulski, W.; Gruszecki, W. I., Organisation of xanthophyll pigments lutein and zeaxanthin in lipid membranes formed with dipalmitoylphosphatidylcholine. Biochimica Et Biophysica Acta-Biomembranes 2000, 1509, (1-2), 255-263.

20. Gruszecki, W. I., In The Photochemistry of Carotenoids, Frank, H. A.; Young, A. J.; Britton, G.; Cogdell, R. J., Eds. Kluwer Academic Publishers: New York, 1999; Vol. 8, p 363.

21. Ferreira, K. N.; Iverson, T. M.; Maghlaoui, K.; Barber, J.; Iwata, S., Architecture of the photosynthetic oxygen-evolving center. Science 2004, 303, (5665), 1831-1838.

22. Loll, B.; Kern, J.; Saenger, W.; Zouni, A.; Biesiadka, J., Towards complete cofactor arrangement in the 3.0 Å resolution structure of photosystem II. Nature 2005, 438, (7070), 1040-1044.

23. Standfuss, J.; Terwisscha van Scheltinga, A. C.; Lamborghini, M.; Kühlbrandt, W., Mechanisms of photoprotection and nonphotochemical quenching in pea light- harvesting complex at 2.5 Å resolution. The EMBO Journal 2005, 24, (5), 919-928.

24. Milon, A.; Wolff, G.; Ourisson, G.; Nakatani, Y., Organization of carotenoid- phospholipid bilayer systems. Incorporation of zeaxanthin, astaxanthin, and their C50 homologues into dimyristoylphosphatidylcholine vesicles. Helvetica Chimica Acta 1986, 69, (1), 12-24.

25. Fulton, R. L.; Gouterman, M., Vibronic coupling. 2. Spectra of dimers. Journal of Chemical Physics 1964, 41, (8), 2280-2286.

26. Clark, A. E.; Qin, C. Y.; Li, A. D. Q., Beyond exciton theory: A time-dependent DFT and Franck-Condon study of perylene diimide and its chromophoric dimer. Journal of the American Chemical Society 2007, 129, (24), 7586-7595.

27. Chéron, M.; Bolard, J., Incorporation of (3R,3'R) zeaxanthin in dipalmitoyl phosphatidylcholine vesicles - Circular dichroism analysis. Comptes Rendus De L Academie Des Sciences Serie Iii-Sciences De La Vie-Life Sciences 1981, 292, (21), 1125- 1128. 136

28. Köhn, S.; Kolbe, H.; Korger, M.; Köpsel, C.; Mayer, B.; Auweter, H.; E. Lüddecke; Bettermann, H.; Martin, H.-D., Aggregation and interface behaviour of carotenoids. In Carotenoids, Britton, G.; Pfander, H., Eds. Birkhäuser Basel: 2008; Vol. 4, pp 53-98.

29. Heller, H.; Schaefer, M.; Schulten, K., Molecular dynamics simulation of a bilayer of 200 lipids in the gel and in the liquid crystal phase. Journal of Physical Chemistry 1993, 97, (31), 8343-8360.

30. Saito, S.; Tasumi, M.; Eugster, C. H., Resonance Raman spectra (5800-40 cm-1) of all-trans and 15-cis isomers of β-carotene in the solid-state and in solution - Measurements with various laser lines from ultraviolet to red. Journal of Raman Spectroscopy 1983, 14, (5), 299-309.

31. Eyring, G.; Curry, B.; Broek, A.; Lugtenburg, J.; Mathies, R., Assignment and interpretation of hydrogen out-of-plane vibrations in the resonance Raman spectra of Rhodopsin and bathorodopsin. Biochemistry 1982, 21, (2), 384-393.

32. Mendelsohn, R.; Vanholten, R. W., Zeaxanthin ([3R,3'R]-β,β-carotene-3,3'diol) as a resonance Raman and visible absorption probe of membrane structure. Biophysical Journal 1979, 27, (2), 221-235.

33. Dallinger, R. F.; Guanci, J. J.; Woodruff, W. H.; Rodgers, M. A. J., Vibrational spectroscopy of the electronically excited-state - pulse radiolysis-time-resolved resonance Raman study of triplet β-carotene. Journal of the American Chemical Society 1979, 101, (5), 1355-1357.

34. Jensen, N. H.; Wilbrandt, R.; Pagsberg, P. B.; Sillesen, A. H.; Hansen, K. B., Time-resolved resonance Raman spectroscopy - The excited triplet state of all-trans-β- carotene. Journal of the American Chemical Society 1980, 102, (25), 7441-7444.

35. Hashimoto, H.; Koyama, Y., Time-resolved resonance Raman spectroscopy of triplet β-carotene produced from all-trans, 7-cis, 9-cis, 13-cis, and 15-cis isomers and high-pressure liquid-chromatography analyses of photoisomerization via the triplet state. Journal of Physical Chemistry 1988, 92, (8), 2101-2108.

36. McCamant, D. W.; Kim, J. E.; Mathies, R. A., Vibrational relaxation in β- carotene probed by picosecond Stokes and anti-Stokes resonance Raman spectroscopy. Journal of Physical Chemistry A 2002, 106, (25), 6030-6038.

37. Wasielewski, M. R.; Gerald, R.; Kispert, L. P., Direct measurement of the lowest excited singlet state lifetime of β-carotene and related carotenoids. Biophysical Journal 1986, 49, (2), A584-A584. 137

38. Niedzwiedzki, D. M.; Sullivan, J. O.; Polivka, T.; Birge, R. R.; Frank, H. A., Femtosecond time-resolved transient absorption spectroscopy of xanthophylls. Journal of Physical Chemistry B 2006, 110, (45), 22872-22885.

39. He, Z.; Kispert, L. D.; Metzger, R. M.; Gosztola, D.; Wasielewski, M. R., Carotenoids in liposomes: Photodegradation, excited state lifetimes, and energy transfer. Journal of Physical Chemistry B 2000, 104, (26), 6302-6307.

40. Polivka, T.; Sundstrom, V., Ultrafast dynamics of carotenoid excited states - From solution to natural and artificial systems. Chemical Reviews 2004, 104, (4), 2021-2071.

41. Nielsen, B. R.; Jorgensen, K.; Skibsted, L. H., Triplet-triplet extinction coefficients, rate constants of triplet decay and rate constants of anthracene triplet sensitization by laser flash photolysis of astaxanthin, β-carotene, canthaxanthin and zeaxanthin in deaerated toluene at 298 K. Journal of Photochemistry and Photobiology a- Chemistry 1998, 112, (2-3), 127-133.

42. Frank, H. A.; Chynwat, V.; Hartwich, G.; Meyer, M.; Katheder, I.; Scheer, H., Carotenoid triplet state formation in Rhodobacter-Sphaeroides R-26 reaction centers exchanged with modified bacteriochlorophyll pigments and reconstituted with spheroidene. Photosynthesis Research 1993, 37, (3), 193-203.

43. Demmig-Adams, B.; Gilmore, A. M.; Adams, W. W., Carotenoids 3: in vivo function of carotenoids in higher plants. The FASEB Journal 1996, 10, (4), 403-412.

44. Pascal, A. A.; Liu, Z.; Broess, K.; van Oort, B.; van Amerongen, H.; Wang, C.; Horton, P.; Robert, B.; Chang, W.; Ruban, A., Molecular basis of photoprotection and control of photosynthetic light-harvesting. Nature 2005, 436, (7047), 134-137.

45. Li, B.; Vachali, P.; Bernstein, P. S., Human ocular carotenoid-binding proteins. Photochemical & Photobiological Sciences 2010, 9, (11), 1418-1425.

Chapter 5 Singlet fission dynamics of carotenoid aggregates: insights from transient absorption spectroscopy

5.1 Introduction

Carotenoids have been the topic of widespread research because of their numerous functions in biology. For example, they serve an important role in harvesting the energy of sunlight1 and for photophotection in plants2, 3 and animals.4 Carotenoids are often assembled in natural systems so that they interact electronically with other chromophores of the same or different types. The electronic interaction between chromophores (exciton coupling) is crucial for energy transfer in photosynthesis.5-9 The exciton coupling between carotenoids has also been proposed as an explanation for unique colors in flower petals10 and the shells of crustaceans.11, 12

One photophysical mechanism that fundamentally depends upon electronic coupling between chromophores, is singlet fission. Singlet fission is a process whereby the energy of a photoexcited singlet state partitions to form two triplet excitons.13Carotenoids in nature have been found to form triplet excited states via fission with substantial efficiency (up to 30%).14-17 The mechanism is possible in large part because of the low-lying triplet excited-state energy levels of carotenoids.16-20 Among the important questions that remain from the work on photosynthetic systems is the identity of the chromophores involved in the singlet fission process.{Zhang, 2001 #15} More generally, researchers focused on singlet fission have endeavored to understand the factors the influence the singlet fission process.13 These efforts have focused on the energy levels of the constituent chromophores21, the number of chromophores involved22,

23, and the strength and orientation of the electronic coupling.13, 22, 24

138

139

We have investigated singlet fission within assemblies of the carotenoid zeaxanthin.19, 25 These studies utilized picosecond-to-nanosecond time-resolved resonance Raman spectroscopy (ps-TRRR). A high yield of triplet excited states, formed via intermolecular singlet fission, was found in a weakly-coupled aggregate.25 More recently, we characterized three zeaxanthin aggregates with steady-state optical spectroscopy, as a foundation for further time-resolved studies.26 One of the aggregates

(H) has a highly blue-shifted absorption spectrum relative to the monomer, and two

(denoted J1 and J2) have slightly red-shifted absorption spectra. The labels H and J are used in our present and prior work, for consistency with the work of others in the community.27 However, results from optical spectroscopy are more consistent with predominantly H-type interactions among the chromophores of all three aggregates, and variation in the coupling strength from weak to strong.26 The conclusions from steady- state spectroscopy are consistent with an earlier theoretical study of lutein aggregates.28

Here, the excited-state dynamics of the three types of zeaxanthin aggregates are probed with femtosecond-tonanosecond transient absorption (fs-TA) and nanosecond-to- microsecond transient absorption (ns-TA) spectroscopy. Triplet excited states form via singlet fission in all three aggregates, and the quantum yields are estimated from the transient spectra. A prominent excited-state absorption band is reported for the strongly- coupled H-aggregate, and an assignment is proposed. Findings from the aggregates can be extended to solid-state systems such as thin films and crystals of polyenes, or to carotenoids in biological systems.

5.2 Experimental 140

Preparation of samples. Three zeaxanthin aggregates in binary solvent systems were formed with procedures described elsewhere.25, 26 The zeaxanthin aggregates are distinguished by their absorption spectra (Figure 5.1). The H-aggregate has a strong blue-shifted absorption band with maximum at 388 nm. The aggregate denoted as J1 has a slightly red-shifted spectrum compared with the monomer, and three distinct maxima in the visible range at 450, 474, and 513 nm. The aggregate denoted as J2 has maxima that are red-shifted 7-9 nm relative to J1, and an extended red absorption wing. The features of J2 are similar to those of a previously reported J-aggregate of zeaxanthin.29

Figure 5.1. UV-vis absorption spectra of zeaxanthin monomer in EtOH, H-aggregate in

20:80 (v/v) EtOH/H2O, J1-aggregate in 10:90 (v/v) THF/H2O and J2-aggregate in 20:80 (v/v) 1,4 Dioxane/H2O.

Femtosecond transient absorption spectroscopy. The focus in the present study is the transient absorption spectroscopy of H and J1. Femtosecond transient absorption

(fs-TA) spectra were acquired with a commercial system (Ultrafast Systems LLC, Helios).

Pump and probe pulses originated from a Ti:sapphire regenerative amplifier (Spectra- 141

Physics, Spitfire XP-Pro) seeded by a Ti:sapphire oscillator (Spectra Physics, Tsunami).

The ~3 mJ 800 nm, 100 fs pulses from the amplifier were attenuated to ~0.8 mJ/pulse.

Most of the attenuated 800 nm beam was directed along an optical path for generating the pump pulses, and chopped at 500 Hz. For experiments with 400 nm excitation, the second-harmonic beam was generated with a type I BBO crystal. Experiments with 520 nm excitation employed an optical parametric amplifier (OPA; Light Conversion,

TOPAS-C). The signal output from the OPA was frequency-doubled with a type I BBO crystal. The probe white light was generated by directing ~3% of the attenuated 1 kHz output from the amplifier along an optical path that included a spherical mirror for focusing the 800 nm beam, and a crystal for generating a white-light continuum. Spectra acquired in the visible 370-750 nm range utilized a 2 mm thick LiF2 crystal. Near- infrared (NIR) 850-1400 nm wavelengths were generated with a 12 mm thick sapphire crystal. A second spherical mirror was employed to collect the white light and focus it to a 1/e2 diameter of ~0.26 mm at the sample point. The pump diameter (1/e2) at the sample ranged from 0.6 to 0.7 mm. The pump and probe beams intersected at an angle of ~10 degrees, and the polarization of the pump pulse was 54.7 degrees (magic angle) relative to the probe to avoid kinetic components due to anisotropy. Spectra were acquired with pump-probe delays ranging from -600 fs (probe before pump) to +2.2 ns, where the arrival time of the pump pulse was controlled by a translation stage. The transmitted probe signal was detected with a CCD detector in the visible spectrum, and an InGaAs array in the near-IR. LabView based software written by Ultrafast Systems was used to control the experiments, and record the data. 142

The instrument response times in the visible window were approximately 150 fs

(520 nm pump) and 320 fs (400 nm pump), as determined from the full-width at half- maximum (FWHM) amplitudes of stimulated Raman signals from the solvent. The instrument response time in the NIR was shorter than 500 fs, as estimated from the decay of the transient absorption signal from the zeaxanthin monomer S2 excited-state. A spectral chirp over the range 350-750 nm was corrected by comparison with the solvent- only response. The kinetic fits to TA signals at selected wavelengths were determined using a least-square iterative fit to a function that included multiple exponential terms convolved with a single Gaussian function. The Gaussian function accounted for the finite instrument response time, and the FWHM of this function was set equal to the measured instrument response time.

A 30 mL volume of sample containing 10 μM zeaxanthin (concentration in units of monomer equivalent) was pumped continuously through a quartz capillary with 2 × 2 mm square cross-section. The maximum optical density (OD) of each sample ranged from 0.22 to 0.28 for a 2 mm path. The percent excitation of the samples ranged from 5 to 8% as estimated from the pump pulse energy, beam size, the extinction coefficient per monomer at the pump wavelength (from Figure 5.1), and the sample OD. Degradation was less than 15% at the end of the experiments for the J1 and H aggregate. The degradation of the J2 aggregate under excitation made collection difficult, and hence only several snapshots of fs-TA spectra are presented here.

Nanosecond Transient absorption spectroscopy. Nanosecond-to-microsecond

TA spectra were acquired with a home-built system. Pump pulses at 355 nm were generated by the third harmonic of a Nd:YAG laser (Quantel, Brilliant). Pump pulses 143 with wavelength 410 were generated by an optical parametric oscillator (OPOTEK,

Magic Prism) pumped with the third harmonic. The 20 Hz pump pulses were ~6 ns in duration. At the sample, the energy per pulse was 0.1-0.2 mJ, and the beam diameter was

4-5 mm. The probe white light was generated from a continuous-wave 75 W Xe arc lamp

(PTI) and passed through appropriate long- and short-pass filters to select the wavelength range incident on the sample. Two spherical mirrors with focal length of 200 mm were used to focus the probe light onto the sample, and collect the transmitted beam. The light was then focused into a 1 mm diameter fiber optic, and the output was relayed to the entrance slit of a 200 mm f.l. spectrometer (PTI). The resolution of the spectrometer was

≤ 10 nm. Signals were detected with a PMT (Hamamatsu R928) with 5 dynodes biased.

The anode current was sent to a ×100 voltage amplifier (Phillips Scientific 6931) or a transimpedance amplifier (Edmund Optics 59-179) prior to being recorded with a digital storage oscilloscope (LeCroy Wavesurfer 452). The bandwidth of the scope was limited to 20 or 200 MHz to improve the signal-to-noise ratio for small (sub-milliOD) transients.

The time response of the system was 20-30 nsec.

The ns-TA system was computer controlled using a custom LabVIEW (Version 7) program. Signals were acquired for 4000 pump pulses at each probe wavelength. Half of the traces were recorded with the pump + probe beams incident on the sample, followed by an equal number with pump-only. The pump-only traces were subtracted from the pump+probe traces to reduce the artifacts caused by laser scatter and fluorescence from the sample. The voltage signal V(t) was converted to a transient absorption (ΔOD) using

, where Vt<0 is the voltage offset caused by the transmitted probe light immediately prior to the pump pulse. 144

Samples for the ns-TA experiments were prepared in a glovebox (MBraun Unilab) with oxygen level < 1 ppm. Solvents were degassed with several freeze-pump-thaw cycles prior to being loaded in the glovebox. The sample solution was loaded in a 1 × 1 cm square cuvette with gas-tight fittings. During the course of the experiment, the samples were constantly stirred with a magnetic stir bar. The OD of the samples was <

0.5 at the pump wavelengths. The percent excitation of the samples was estimated to be

10% for both the H and J1 aggregates. At the end of the nsec-µsec experiments, the degradation was less than 5%.

5.3 Results

5.3.1 Fs-ns-TA spectroscopy of zeaxanthin aggregates.

The fs-TA spectra of zeaxanthin monomer and the H, J1, and J2-aggregates are presented in Figure 5.1 over the visible and near-IR range. No significant differences in the TA spectra were found upon changing the pump wavelength from 520 nm to 400 nm.

Therefore, spectra in Figure 5.1 were selected from runs with either pump wavelength, and illustrate the best quality spectra that we have available. Transient absorption kinetics and fits at selected wavelengths for the J1- and J2-aggregates are shown in Figure 5.2, and numerical values are listed in Table 5.1. 145

Figure 5.2. Fs-TA spectra at different time delays of zeaxanthin monomer (Mon) in ethanol, photoexcited at 400 nm with 0.6 µJ/pulse; J1-aggregate (J1) photoexcited at 520 nm with 0.32 µJ pulse; J2-aggregate (J2) pumped at 400 nm with 0.3 µJ/pulse; and H- aggregate (H) pumped at 520 nm with 5.0 µJ/pulse. The absorption spectra of the ground state are shown as dotted lines, and are inverted and scaled for best comparison with the initial bleach. Solvent signals are indicated by *. Spectra in near-IR region are shown as insets. All near-IR spectra were acquired with 400 nm photoexcitation. The instrument time response is slower in the near-IR window than the visible, therefore spectra illustrating dynamics are selected at different times relative to those in the main panels.

Zeaxanthin monomer. The fs-TA spectra and kinetics of monomeric zeaxanthin are shown in Figure 5.2. The dynamics are known to be similar to those of β-carotene.30

In brief, the initial monomer spectrum has a broad excited-state absorption (ESA) band that spans the range 900−1100 nm. This absorption originates from the S2→Sn transition 146 and decays with a 100−200 fs time constant which is characteristic of the solvent-

31 dependent S2 excited state lifetime. A band with maximum at 550 nm and a broad near-

IR band in the region 1050 to 1200 nm appear within 1 ps. The 550 nm band is

1, 30, 32 characteristic of an S1→Sn’ transition. The absorption in the 900-1300 nm range is

33 attributed to the S1→S2 transition. The 9−12 ps decay times of these features

1, 34, 35 correspond to the known lifetime of the S1 state. Global fitting procedures highlight an additional absorption band on the high-energy side of the 550 nm S1 band, which has been assigned to a state labeled as S*.30, 31, 35-37 No signatures of the triplet excited state were observed in our measurements because the intersystem-crossing yield of zeaxanthin, like other carotenoids, is significantly less than 1% with visible excitation. 38

J1-aggregate. Selected spectra obtained with 520 nm 0.32 µJ excitation at different time delays are shown in Figure 5.2. At the earliest time delays a relatively flat

ESA spans the visible and near-infrared region from 550 nm to wavelengths longer than

1250 nm. The formation time of the ESA at 700 nm and longer wavelengths is within the instrument response time. The majority of the decay at ~700 nm occurs with a time- constant of ~200 fs (Table 1). Two ESA bands centered at 550 nm and 590 nm dominate the TA spectra for delay times longer than 200 fs. The risetime of both bands is 200 fs or less. The transient absorption at 590 nm decays with exponential time constants of 270 fs

(44% amplitude) and 7.8 ps (50%). The absorption is significantly diminished at 15 ps delay. The decay of the 550 nm band can be fitted with three exponential components,

330 fs (45%), 5.1 ps (20%) and 230 ps (20%). In contrast to the 590 nm band, significant

(~15%) amplitude of the 550 nm band remains at a delay time of 2 ns. 147

There are additional ESA bands in the 380-530 nm region that overlap with the transient ground-state bleach. These bands can be discerned as early as 100 fs, based upon the positive differences between the transient ground state bleach (GSB) as compared with the inverted steady-state UV-Vis absorption spectrum. An ESA band at

500 nm competes with the GSB as early as 100 fs, and at ~270 fs and later times, the band appears as a slightly positive feature. ESA features are also apparent at ~435 and

~463 nm for delay times ranging from less than 1 ps to greater than 150 ps.

We note that a simplifying aspect of the transient spectra of WH is the absence of any stimulated emission, which would appear as negative features in the fs-TA spectra.

Consistent with this observation from steady state spectroscopy, the extremely low

2.5×10-5 fluorescence quantum yield of the aggregate makes it very unlikely that a radiative transition to the ground state would be present past 100 fs. The quantum yields of the other aggregates are even smaller, and likewise no stimulated emission is seen for these samples.

We note that a simplifying aspect of the transient spectra of J1 is the absence of any stimulated emission component, which would be in the negative direction if present.

Consistent with this observation, the extremely low 2.5 × 10-5 fluorescence quantum yield of the aggregate26 makes it very unlikely that a radiative transition should be present past 100 fs. The quantum yields of the other aggregates are even smaller, and likewise no stimulated emission is seen for these samples as well.

J2-aggregate. The J2-aggregate has transient absorption spectra that are similar to those of the J1-aggregate, but slightly red-shifted (Figure 5.2). ESA bands are noted at

465, 505, 560, and 595 nm which are at nearly the same positions as for J1. However the 148 relative intensity of the 595 nm band is clearly greater than that of the 560 nm band at the earliest time delay, which is reverse of what is seen for similar bands of J1. As seen with

J1, the ESA band at 595 nm decays faster than the 560 nm band. The 560 nm band remains strong after several hundreds of picoseconds.

H-aggregate. When pumped with 520 nm, 5 μJ pulses, the initial bleach is strongest at ~380 nm (Figure 5.2), which is similar to the position of the maximum ground-state absorption. A broad ESA band is apparent at the earliest times across the visible/near-IR region. Subsequently, three resolved ESA features appear. The dominant transient absorption band of the H-aggregate has a maximum at ~410 nm. Weaker bands have peak positions at ~500 nm and 545−570 nm. The minor bands were previously reported for the same aggregate, however the 410 nm band was outside the spectral window of the prior study.29 The apparent maximum of the dominant band shifts from

420 nm (at 227 fs) to ~400 nm (at 2 ns). The likely reason is that the ESA competes with

GSB. The rise-time of the dominant feature cannot be distinguished from the ~150 fs instrument response time because of spectral overlap with the bleach. Approximately 80% of the decay of the band occurs with 4 and 50 ps time constants. It is difficult to extract a rise time for the ~550 nm band because of competition with the decay of the initial broad

ESA band. Most (70−80%) of the decay of the ~500 and ~550 nm bands occurs within 10 ps.

Power dependence of prominent H and J bands. The kinetics of both 550 nm and 410 nm species were compared under different excitation powers. For the J1- aggregate, we investigated the kinetics of the 550 nm ESA for pulse energies ranging from 0.32 to 0.9 µJ/pulse, corresponding to estimated excitation ratios from 5 to 13 %. 149

When the excitation pulse energy is varied over this ~3-fold range, the amplitude of the

350 fs component remains almost unchanged (48% to 52%), whereas the 6 ps component increases from 19 to 28%. The kinetics of the 550 and 410 nm ESA of H aggregate were studied over excitation ratios of 5 to 11%. The excitation photon density correlates weakly to the kinetics of the 550 nm ESA of H. Only the 750 fs component showed an increase in amplitude from 49% to 57% when the incoming photon density was increased by a factor of 2.3. The 410 nm band also exhibits very weak dependence upon pulse energy.

Figure 5.3. TA kinetics at selected wavelengths with 520 nm exci tation of the J1- aggregate (0.32 µJ/pulse) and the H-aggregate (5.0 µJ/pulse). The kinetics are fitted to a multi-exponential function convolved with a Gaussian component which is used to describe the instrument response function (abbreviate IRF above). 150

Table 5.1 Multiexponential fitting results for kinetics of fs-TA bands of zeaxanthin monomer and aggregates. a Species/ Probe τ0, (amp.) τ1, (amp.) τ2, (amp.) τ3,(amp.) τ > 1 ns Pump (nm) ps, (%) ps, (%) ps, (%) ps, (%) (% amp.) (µJ/pulse) 480 10.2,(100) Monomer/ 550 0.15, (-100) 9.8, (100) 400 nm, 1000 0.19, (100) 0.60 1200 0.21, (77) 12, (23) 451 0.62, (67) 7.5, (20) 152, (8) (5) 500 7.3, (21) 230, (42) (12) J1-agg/ 550 0.10, (-90) 0.33, (45) 5.1, (20) 230, (20) (15) 520 nm, 590 0.20, (-80) 0.27, (44) 7.8, (50) (6) 0.32 709 0.19, (91) 5.7, (9) 1020 0.45, (100) 383 0.85, (45) 12, (34) 290, (14) (7) H-agg/ 418 0.14, (-91) 3.9 (58) 48, (26) 610, (11) (5) 520 nm, 555 0.75, (54) 10, (28) 300, (11) (7) 5.0 702 0.14, (90) 7.2, (10) 1020 0.30, (80) 3, (10) >20, (10) a Numbers in parentheses denote the relative amplitude (%) of the fitting components. For τ0, the growth of a signal is distinguishable from the instrument response at several probe wavelengths; the appearance time is denoted as negative amplitude.

5.3.2 Ns-TA spectroscopy of zeaxanthin aggregates

TA spectra of J1 and H aggregates on the nanosecond-to-microsecond timescale are shown in Figure 5.4. The spectra of the J1-aggregate correspond well to those acquired at the longest (nanosecond) times with the fs-TA system. In particular the most prominent absorption bands are centered at ~550 nm and ~500 nm. The H-aggregate also shows a transient absorption at 550 nm, but the amplitude is much smaller than the same band of J1, for the same (~10%) excitation ratio. The ~410 nm ESA of the H-aggregate is approximately 5-fold stronger than the band at 550 nm. This finding on the nanosecond-to-microsecond timescale is consistent with the result that the 410 nm band is also the strongest band on the femtosecond-picosecond timescale. 151

Kinetic data for the ns-TA spectra of the J1 and H aggregate are shown in Figure

5.5. In general, these kinetics traces can be fit well with double exponential decays. It should be noted that the amplitudes reported in Figure 5.5 indicate the relative importance of the components within the nanosecond-to-microsecond time. These values do not take into account the much greater decay that occurs on the femtosecond-to- picosecond timescale.

Figure 5.4. Ns-TA spectra for the zeaxanthin J1-aggregate excited at 410 nm (Upper) and the H-aggregate excited at 355 nm (Lower). 152

Figure 5.5. (Upper) Ns-TA kinetic decays for the zeaxanthin J1-aggregate, probed at 555 nm, following excitation at 410 nm. (Lower) Kinetic decays for the H-aggregate, probed at 550 and 410 nm, following 355 nm excitation.

5.4 Discussion

The dynamics of the self-assembled systems have some aspects that are closely connected to the established photophysics of the monomer. The photophysics of monomeric b-carotene or zeaxanthin are summarized in Figure 5.6.18 The signatures of the initial S2 excited state for the monomer and aggregates alike are broad ESA bands which span the visible-NIR region. One distinguishing aspect of the monomer is the maximum near 1000 nm, which is attributed to S2→Sn absorption. The comparatively flat absorption bands of the aggregates can be attributed to the wide distribution of energy sublevels for the exciton-coupled S2 state. The lifetimes of the S2 features of aggregates 153 appear to be slightly longer than that of the monomer (<200 fs). However, precise measurements of the S2 lifetime are limited by the instrument response of our system.

Figure 5.6. Summary of zeaxanthin monomer photophysics. The black arrows indicate allowed optical transitions, and the dash arrows indicate internal conversion (IC) and intersystem crossing (ISC) pathways.

Next, it is reasonable to assign the 590−600 nm TA bands of the weakly-coupled aggregates SW and JH to an S1→SN transition. The location of the maximum is approximately 40 nm red-shifted relative to the corresponding band of the monomer in ethanol. Red-shifts in the transient features are expected for the aggregates relative to the solvated monomer, similar to shifted transitions originating from the ground state. The shifts are a consequence of non-resonant dispersion interactions, which are greater for the aggregates than the solvated monomer. The decay of the 590 nm band occurs largely with times similar to 0.27 and 7.8 ps. It is possible that the faster component is a consequence 154

of overlap with the 550 nm band which has a fast decay unrelated to the S1 state (see below). We assign the 7.8 ps component to the intrinsic S1 lifetime of the weakly coupled aggregates. This lifetime is only slightly shorter than the 9 ps measured for the monomer.

We did not observe a 20−30 ps component in the 590 nm decay, thus our results from fs-

29 TA spectroscopy do not agree with the longer S1 lifetime reported previously. No sign of the S1 state was found in our studies of the strongly-coupled SH-aggregate.

The transient absorption spectra of the aggregates, and particularly J1 and J2, show a strong band with maximum at 550 nm which lasts for hundreds of nanoseconds.

These bands do not have a counterpart in the fs-TA spectra of the monomer. Our ps-

TRRR experiments with probe wavelengths 473 and 550 nm strongly support the

25 assignment of both the 465 and 550 nm bands of the J1 aggregate to T1→TN transitions.

Similarly, our TRRR studies of the H-aggregate (to be published) fully support the assignment of the weak 550 nm band to the T1 state, in agreement with the previous

29 assignment of that band from fs-TA spectroscopy. The assignments for the T1→TN bands of the aggregates are consistent with numerous prior studies of monomeric carotenoids that have located the maxima of these transitions in the 450-550 nm region.36,

39-42 In general, the studies of monomers indicate that the lowest energy T1→TN transitions are approximately 900 cm-1 shifted to lower energy of the absorption onset of

S0→S2. Similarly, the T1→TN transitions in the J1-aggregate are also red-shifted relative to transitions from the S0 state.

We now turn our attention to the assignment of the dominant absorption band of

H, which has a maximum at 400-420 nm. The fact that the band is apparent on timescales ranging from femtoseconds to nanoseconds distinguishes it from typical carotenoid 155 excited-state features, and is an important piece of evidence that this absorption, like the

550 nm band, is a signature of a triplet excited state. A crucial aspect of our assignment rests upon the proposal that the T1→TN transition is strongly exciton-coupled to the S0—

S2 transitions of neighboring molecules. The fact that the transient ESA resulting from the proposed (T1→TN + S0→S2) coupling is slightly red-shifted relative to the (S0→S2 +

S0→S2) coupling that gives rise to the 388 nm peak position of the H-aggregate, is consistent with the lower energy of the T1→TN transition relative to S0→S2 of the isolated chromophores.

The (T1→TN + S0→S2) exciton coupling described above does not seem to have been addressed clearly in the literature, to the best of our knowledge. We suggest that it could be a common feature in transient spectroscopy of exciton coupled systems, since it is not unusual for T1→TN transitions of chromophores to have oscillator strength and energies similar to those on the singlet manifold. The similar energies and transition strengths of singlet and triplet transitions in the case of the carotenoids has been well established theoretically43, 44 and experimentally39, 42. We emphasize that the more commonly discussed decoupling of triplet excited states from their surroundings is based upon very low oscillator strengths of S0↔T1 transitions that are typical for organic molecules, and therefore correspondingly weak exciton coupling between the singlet S0 and triplet T1 states. The latter should not be confused with the different type of exciton coupling between triplet and singlet optical transitions, such as T1→TN and S0→S2 transitions that are possible once the T1 states are generated e.g. in pump-probe spectroscopy. 156

Transient absorption spectroscopy alone cannot exclude other possibilities for the assignment of the 410 nm absorption of the H-aggregate. These include excimers with charge-transfer character, and other long lifetime excited states such as a singlet fission

1* 13 precursor (T1T1) . Recently, an ESA feature in the TA spectra of pentacene film has been assigned to such a coupled state, excimer formation was excluded because of the

45 1* crystallographic structure. The (T1T1) state can be considered as a doubly-excited state that is weakly stabilized by van der Waals interactions between the triplet pairs.

Evidence from magnetic field effect experiments suggests that there is significant binding

46 energy between two neighboring T1 excited states of perylene. Further investigations, such as time-resolved vibrational spectroscopy, are needed to confirm the identity of this major relaxation product.

The appearance of a minor 550 nm absorption in the H-aggregate can be explained by inhomogeneity within this aggregate. The absorption bands in the 420- 520 nm range for the H-aggregate are found to correspond to features in the weakly-coupled aggregates. The correspondence as well as other evidence leads to the suggestion that the strongly coupled aggregate has a subpopulation of chromophores that are not strongly coupled to their surroundings.26 It is likely these chromophores are the ones responsible for the triplet absorption band at 550 nm in the TA spectra.

One might suppose that the inhomogeneous model conflicts with the fact that the dynamics of the H-aggregate are indistinguishable with 400 and 520 nm excitation.

Specifically, how is it possible that either pump excitation can give rise to strong ground state bleach at 380 nm? One explanation is that it is possible that the energy transfer between the weakly coupled subpopulation of chromophores and the strongly coupled 157 surroundings is faster than the time resolution of our system. Specifically, the subpopulation that is excited at 520 nm relaxes in <150 fs to the much lower energy (out- of-phase) level of the strongly-coupled exciton band. The occupation of this lower energy level diminishes the population of the strongly coupling chromophores, and hence causes the bleach of its optical allowed transition.

From above analysis, we can summarize four aspects of the relaxation of zeaxanthin aggregates that differ from the monomer:

1. The S1 excited-state is formed in low yield for the aggregates. For the H- aggregate, we see no evidence of the S1 state following photoexcitation. For J1-aggregate, the S1 state yield can be estimated from the TA spectra. The excitation ratios for pump- probe experiments of J1 are estimated to be 5-7%. Based on literature reports on the relative magnitudes of S1 ESA and the GSB for the monomeric carotenoids in the β-

5 5 carotene family, the extinction coefficient of the S1 state range from 1.7×10 to 3.6×10

M-1cm-1. 30, 47 With these values in mind and the percent excitation in mind, the absorption bands of the J1 spectrum lead to quantum yields of the S1 state that range from

10-30%. The quality of the J2 TA spectra prevents us from making a similar estimate for this aggregate.

2. The S1 excited state is not the parent state for the singlet fission process.

The S1 state is clearly not the parent state for singlet fission, because (a) the signatures of the T1 and S1 states form simultaneously and (b) the decay of 590 nm S1 band does not lead to a concurrent rise in the 550 nm T1 band. The fact that the S1 state is not the parent state for singlet fission agrees with a prior study of carotenoids in light-harvesting complexes.48 158

3. The ultrafast formation time and high yield of triplet excited states reveal that singlet fission is an important pathway for all of the photoexcited aggregates.

The absorption bands for the T1 state appear within ~300 fs for both the J- and H- aggregates, which is highly improbable to happen via intersystem crossing. The triplet

5 -1 -1 42 T1→Tn extinction coefficient is 1×10 M cm for the monomer. Assuming the extinction coefficient at 550 nm in the aggregate is equal to the value of the monomer, the triplet quantum yield of the J1-aggregate is estimated to be 60−80% based upon the amplitude of 550 nm band at 300 fs. The spread in the yields is largely dominated by uncertainty in the measurement of the excitation ratio. These values would imply that

30−40% of the relaxation of J1 occurs via singlet fission to form the triplet excited states.

It must be emphasized that the triplet yield here is a lower limit, and does not take into account (a) annihilation that occurs within the instrument response time, and (b) diminishment of the T1→Tn extinction coefficient at 550 nm. The transient spectra clearly show strong oscillator strength for the triplet absorption in other regions, e.g. 500 and 460 nm. If the strength of the transition at 550 nm were lower than the assumed

1×105 M-1cm-1 value, then the calculated triplet yields would increase.

The J2-aggregate has a lower triplet yield in comparison with SH, and the main competitive process for singlet fission is internal conversion to the S1 state. The quality of the fs-TA spectra prevents us from drawing quantitative conclusions for the triplet yield of this aggregate.

For the H-aggregate, there is a small portion of T1 states that have a weakly- coupled T1→Tn transition with the same 550 nm band position as for WH and JH. The estimated yield is 10-20%, following the analysis above. A far greater yield of T1 excited 159 states in the SH-aggregate have an absorption band that reveals strong exciton coupling to the surrounding ground-state chromophores (T1→Tn + S0→Sn), or possibly to one another (T1→Tn + T1→Tn). One can infer, from the absence of any S1 product and the 10-

20% yield of weakly-coupled triplets, that the yield of T1 states with strongly coupled transitions could be 180% or more.

4. The decays of the T1 excited states in the aggregates are much faster than that of the monomer because of triplet-triplet annihilation. A 9 μs lifetime of monomeric zeaxanthin has been reported.42 By comparison, the majority of the triplet excited-state signal in the aggregates decays on a timescale that is less than 10 ps. The short

lifetime is attributed to annihilation. The energy of the T2 state for an N=11

44 polyene is expected to be less than twice of that of T1, thus triplets can efficiently annihilate and form T2 and S0 states. Unlike most bimolecular processes, the decay rates of triplet excited states in zeaxanthin aggregates depend only weakly on the energy of the pump pulses. A likely explanation is that in the case of the singlet fission, two triplets are generated in close proximity to one another. Therefore, prompt geminate annihilation that is independent of excitation density can take place in these systems.

5.5 Conclusions

We summarize the excited state dynamics of the zeaxanthin aggregates in Figure

5.7. Zeaxanthin molecules in the aggregates are excited to the S2 state. The normal

(monomer-like) internal conversion to the S1 state competes with singlet fission in the J1- and J2-aggregates. In the case of the H-aggregate, fission appears to be too fast for the generation of measurable S1 excited-state population. Our spectra suggest that singlet 160

* fission occurs directly from the S2 state. If an intermediate (S ) state with covalent character were the fission precursor,17, 48, 49 then the population or lifetime of this species is insufficient for detection in the experiments described here. The T1 excited states in the aggregates show variation in their absorption bands which can be explained by exciton coupling. The 550 nm ESA is a signature of a T1→Tn transition that is only weakly coupled to transitions of neighboring molecules. The 410 nm ESA is associated with strong H-type coupling between T1→Tn and S0→S2 transitions of surrounding chromophores. Alternatively, the 410 nm ESA could be attributed to strong H-type coupling between two T1→Tn transitions, if a triplet pair state (T1T1) were formed. A majority of the triplet species annihilate within 10 picoseconds. However, a portion of the triplet population survives over many nanoseconds, and approaches the intrinsic triplet- state lifetime of monomeric zeaxanthin.

Figure 5.7. Excited-state dynamics of zeaxanthin aggregates. An aggregate that has weak/intermediate coupling is denoted J1. The strongly-coupled aggregate is denoted H.

Internal conversion is denoted by IC. Tn(delocal.) refers to a high-energy state of the triplet that is delocalized because of exciton coupling. The estimated yields of excited- state species in the J2-aggregate are not listed (see text).

161

Acknowledgement

References

1. Polivka, T.; Sundstrom, V., Ultrafast dynamics of carotenoid excited states - From solution to natural and artificial systems. Chemical Reviews 2004, 104, (4), 2021-2071.

2. Holt, N. E.; Fleming, G. R.; Niyogi, K. K., Toward an understanding of the mechanism of nonphotochemical quenching in green plants. Biochemistry 2004, 43, (26), 8281-8289.

3. Holt, N. E.; Zigmantas, D.; Valkunas, L.; Li, X. P.; Niyogi, K. K.; Fleming, G. R., Carotenoid cation formation and the regulation of photosynthetic light harvesting. Science 2005, 307, (5708), 433-436.

4. Krinsky, N. I.; Landrum, J. T.; Bone, R. A., Biologic mechanisms of the protective role of lutein and zeaxanthin in the eye. Annual Review of Nutrition 2003, 23, 171-201.

5. Jennings, R. C.; Bassi, R.; Zucchelli, G., Antenna structure and energy transfer in higher plant photosystems. In Electron Transfer Ii, Springer-Verlag Berlin: Berlin, 1996; Vol. 177, pp 147-181.

6. Krueger, B. P.; Scholes, G. D.; Jimenez, R.; Fleming, G. R., Electronic excitation transfer from carotenoid to bacteriochlorophyll in the purple bacterium Rhodopseudomonas acidophila. Journal of Physical Chemistry B 1998, 102, (12), 2284- 2292.

7. Krueger, B. P.; Scholes, G. D.; Gould, I. R.; Fleming, G. R., Carotenoid mediated B800-B850 coupling in LH2. Physchemcomm 1999, 2, (8), 7.

8. Walla, P. J.; Linden, P. A.; Hsu, C. P.; Scholes, G. D.; Fleming, G. R., Femtosecond dynamics of the forbidden carotenoid S1 state in light-harvesting complexes of purple bacteria observed after two-photon excitation. Proceedings of the National Academy of Sciences of the United States of America 2000, 97, (20), 10808-10813. 162

9. Konig, C.; Neugebauer, J., Quantum chemical description of absorption properties and excited-state processes in photosynthetic systems. Chem. Phys. Chem 2012, 13, (2), 386-425.

10. Zsila, F.; Deli, J.; Simonyi, M., Color and chirality: carotenoid self-assemblies in flower petals. Planta 2001, 213, (6), 937-942.

11. Ilagan, R. P.; Christensen, R. L.; Chapp, T. W.; Gibson, G. N.; Pascher, T.; Polívka, T.; Frank, H. A., Femtosecond time-resolved absorption spectroscopy of astaxanthin in solution and in alpha-crustacyanin. Journal of Physical Chemistry A 2005, 109, (14), 3120-3127.

12. Bartalucci, G.; Coppin, J.; Fisher, S.; Hall, G.; Helliwell, J. R.; Helliwell, M.; Liaaen-Jensen, S., Unravelling the chemical basis of the bathochromic shift in the lobster carapace; new crystal structures of unbound astaxanthin, canthaxanthin and zeaxanthin. Acta Crystallographica Section B-Structural Science 2007, 63, 328-337.

13. Smith, M. B.; Michl, J., Singlet fission. Chemical Reviews 2010, 110, (11), 6891- 6936.

14. Cogdell, R. J.; Hashimoto, H.; Koyama, Y., Picosecond Raman-spectroscopy of the B830 LH2 complex of Chromatium purpuratum BN-5500. Photosynthesis Research 1993, 35, (1), 79-85.

- 15. Zhang, J. P.; Inaba, T.; Koyama, Y., The role of the newly-found 1Bu state of + - carotenoid in mediating the 1Bu to 2Ag internal conversion and the excited-state dynamics of carotenoid and bacteriochlorophyll in a bacterial antenna complex. Journal of Molecular Structure 2001, 598, (1), 65-78.

16. Papagiannakis, E.; Das, S. K.; Gall, A.; van Stokkum, I. H. M.; Robert, B.; van Grondelle, R.; Frank, H. A.; Kennis, J. T. M., Light harvesting by carotenoids incorporated into the B850 light-harvesting complex from Rhodobacter sphaeroides R- 26.1: Excited-state relaxation, ultrafast triplet formation, and energy transfer to bacteriochlorophyll. Journal of Physical Chemistry B 2003, 107, (23), 5642-5649.

17. Papagiannakis, E.; Kennis, J. T. M.; van Stokkum, I. H. M.; Cogdell, R. J.; van Grondelle, R., An alternative carotenoid-to-bacteriochlorophyll energy transfer pathway in photosynthetic light harvesting. Proceedings of the National Academy of Sciences of the United States of America 2002, 99, (9), 6017-6022.

18. Christensen, R. L., The electronics states of carotenoids. In The Photochemistry of the Carotenoids, Frank, H. A.; Young, A. J.; Britton, G.; Cogdell, R. J., Eds. Kluwer Academic Publishers: Dordecht, 1999; Vol. 8, pp 137-157. 163

19. Wang, C.; Schlamadinger, D. E.; Desai, V.; Tauber, M. J., Triplet excitons of carotenoids formed by singlet fission in a membrane. ChemPhysChem 2011, 12, (16), 2891-2894.

20. Hayashi, H.; Noguchi, T.; Tasumi, M.; Atkinson, G. H., Vibrational spectroscopy of exited electonic states in carotenoids in vivo - picosecond time-resolved resonance Raman-scattering. Biophys. J. 1991, 60, (1), 252-260.

21. Paci, I.; Johnson, J. C.; Chen, X. D.; Rana, G.; Popovic, D.; David, D. E.; Nozik, A. J.; Ratner, M. A.; Michl, J., Singlet fission for dye-sensitized solar cells: Can a suitable sensitizer be found? Journal of the American Chemical Society 2006, 128, (51), 16546-16553.

22. Muller, A. M.; Avlasevich, Y. S.; Schoeller, W. W.; Mullen, K.; Bardeen, C. J., Exciton fission and fusion in bis(tetracene) molecules with different covalent linker structures. Journal of the American Chemical Society 2007, 129, (46), 14240-14250.

23. Muller, A. M.; Avlasevich, Y. S.; Mullen, K.; Bardeen, C. J., Evidence for exciton fission and fusion in a covalently linked tetracene dimer. Chemical Physics Letters 2006, 421, (4-6), 518-522.

24. Greyson, E. C.; Stepp, B. R.; Chen, X.; Schwerin, A. F.; Paci, I.; Smith, M. B.; Akdag, A.; Johnson, J. C.; Nozik, A. J.; Michl, J.; Ratner, M. A., Singlet exciton fission for solar cell applications: energy aspects of interchromophore coupling. Journal of physical Chemistry B 2009, 114, (45), 14223-14232.

25. Wang, C.; Tauber, M. J., High-yield singlet fission in a zeaxanthin aggregate observed by picosecond resonance Raman spectroscopy. Journal of the American Chemical Society 2010, 132, (40), 13988-13991.

26. Wang, C.; Berg, C. J.; Hsu, C. C.; Merrill, B. A.; Tauber, M. J., Characterization of carotenoid aggregates by steady-state optical spectroscopy. Journal of Physical Chemistry B 2012, 116, (35), 10617-10630.

27. Köhn, S.; Kolbe, H.; Korger, M.; Köpsel, C.; Mayer, B.; Auweter, H.; E. Lüddecke; Bettermann, H.; Martin, H.-D., Aggregation and interface behaviour of carotenoids. In Carotenoids, Britton, G.; Pfander, H., Eds. Birkhäuser Basel: 2008; Vol. 4, pp 53-98.

28. Spano, F. C., Analysis of the UV/Vis and CD spectral line shapes of carotenoid assemblies: spectral signatures of chiral H-aggregates. Journal of the American Chemical Society 2009, 131, (12), 4267-4278.

29. Billsten, H. H.; Sundstrom, V.; Polivka, T., Self-assembled aggregates of the carotenoid zeaxanthin: time-resolved study of excited states. Journal of Physical Chemistry A 2005, 109, (8), 1521-1529. 164

30. Niedzwiedzki, D. M.; Sullivan, J. O.; Polivka, T.; Birge, R. R.; Frank, H. A., Femtosecond time-resolved transient absorption spectroscopy of xanthophylls. Journal of Physical Chemistry B 2006, 110, (45), 22872-22885.

31. Zhang, J. P.; Skibsted, L. H.; Fujii, R.; Koyama, Y., Transient absorption from the + 1Bu state of all-trans-beta-carotene newly identified in the near-infrared region. Photochemistry and Photobiology 2001, 73, (3), 219-222.

32. Billsten, H. H.; Pan, J. X.; Sinha, S.; Pascher, T.; Sundstrom, V.; Polivka, T., Excited-state processes in the carotenoid zeaxanthin after excess energy excitation. Journal of Physical Chemistry A 2005, 109, (31), 6852-6859.

33. Polívka, T.; Herek, J. L.; Zigmantas, D.; Akerlund, H. E.; Sundstrom, V., Direct observation of the (forbidden) S1 state in carotenoids. Proceedings of the National Academy of Sciences of the United States of America 1999, 96, (9), 4914-4917.

34. Wasielewski, M. R.; Gerald, R.; Kispert, L. P., Direct measurement of the lowest excited singlet state lifetime of β-carotene and related carotenoids. Biophysical Journal 1986, 49, (2), A584-A584.

35. Cong, H.; Niedzwiedzki, D. M.; Gibson, G. N.; Frank, H. A., Ultrafast time- resolved spectroscopy of xanthophylls at low temperature. Journal of Physical Chemistry B 2008, 112, (11), 3558-3567.

36. Jailaubekov, A. E.; Vengris, M.; Song, S. H.; Kusumoto, T.; Hashimoto, H.; Larsen, D. S., Deconstructing the excited-state dynamics of β-carotene in solution. Journal of Physical Chemistry A 2011, 115, (16), 3905-3916.

37. Polivka, T.; Sundstrom, V., Dark excited states of carotenoids: Consensus and controversy. Chemical Physics Letters 2009, 477, (1-3), 1-11.

38. Nielsen, B. R.; Mortensen, A.; Jørgensen, K.; Skibsted, L. H., Singlet versus triplet reactivity in photodegradation of C-40 carotenoids. Journal of Agricultural and Food Chemistry 1996, 44, (8), 2106-2113.

39. Land, E. J.; Sykes, A.; Truscott, P. G., In-vitro photochemistry of biological molecules .2. Triplet states of β-carotene and lycopene excited by pulse radiolysis. Photochemistry and Photobiology 1971, 13, (4), 311-320.

40. Hashimoto, H.; Koyama, Y.; Ichimura, K.; Kobayashi, T., Time-resolved absorption spectroscopy of the triplet-state produced from the all-trans, 7-cis, 9-cis, 13- cis, and 15-cis isomers of β-carotene. Chemical Physics Letters 1989, 162, (6), 517-522.

41. Rodgers, M. A. J.; Bates, A. L., Kinetic and spectroscopic features of some carotenoid triplet states - Sensitization by singlet oxygen. Photochemistry and Photobiology 1980, 31, (6), 533-537. 165

42. Nielsen, B. R.; Jorgensen, K.; Skibsted, L. H., Triplet-triplet extinction coefficients, rate constants of triplet decay and rate constants of anthracene triplet sensitization by laser flash photolysis of astaxanthin, β-carotene, canthaxanthin and zeaxanthin in deaerated toluene at 298 K. Journal of Photochemistry and Photobiology a- Chemistry 1998, 112, (2-3), 127-133.

43. Gijzeman, O. L.; Sykes, A., Relationship between singlet and triplet absorption spectra of conjugated polyenes. Photochemistry and Photobiology 1973, 18, (4), 339-341.

44. Kleinschmidt, M.; Marian, C. M.; Waletzke, M.; Grimme, S., Parallel multireference configuration interaction calculations on mini-β-carotenes and β-carotene. Journal of Chemical Physics 2009, 130, (4).

45. Kuhlman, T. S.; Kongsted, J.; Mikkelsen, K. V.; Moller, K. B.; Solling, T. I., Interpretation of the ultrafast photoinduced processes in pentacene thin films. Journal of the American Chemical Society 2010, 132, (10), 3431-3439.

46. Albrecht, W. G.; Michelbeyerle, M. E.; Yakhot, V., Exciton fission in excimer forming crystal - Dynamics of an excimer buildup in α-perylene. Chemical Physics 1978, 35, (1-2), 193-200.

47. Lustres, J. L. P.; Dobryakov, A. L.; Holzwarth, A.; Veiga, M., S2→S1 internal conversion in β-carotene: Strong vibronic coupling from amplitude oscillations of transient absorption bands. Angewandte Chemie-International Edition 2007, 46, (20), 3758-3761.

48. Gradinaru, C. C.; Kennis, J. T. M.; Papagiannakis, E.; van Stokkum, I. H. M.; Cogdell, R. J.; Fleming, G. R.; Niederman, R. A.; van Grondelle, R., An unusual pathway of excitation energy deactivation in carotenoids: Singlet-to-triplet conversion on an ultrafast timescale in a photosynthetic antenna. Proceedings of the National Academy of Sciences of the United States of America 2001, 98, (5), 2364-2369.

49. Tavan, P.; Schulten, K., The low-lying elecronic excitations in long polyenes - A PPP-MRD-CL study. Journal of Chemical Physics 1986, 85, (11), 6602-6609.

Chapter 6 Singlet Fission Process in Zeaxanthin Aggregates: Insights from Transient Resonance Raman and Absorption Spectroscopy

6.1 Introduction

Singlet fission is a photophysical process which splits the energy of one singlet exciton to form a pair of excited triplets. Singlet fission was first discovered in crystalline acenes1, 2 and then found in systems of various chromophores, such as polymers 3-5, carotenoids in light-harvesting antennae,6-9and others. 10, 11

Recently singlet fission was highlighted as a possible mechanism for improving the efficiency of solar energy conversion.12-14 The potential application of singlet fission has spurred both experimental and theoretical studies of this topic. A series of compounds

15 were selected according to the energy criteria: and . Recent reports of efficient singlet fission include derivaties of the acenes,16-21 a semiconductive polymer22, 1,3-diphenylisobenzofuran,23 diphenylpolyene,24 perylenediimide,25 and carotenoid aggregates.26-28 A few experiments have been conducted to extract the charge from fission products.29-33 One device that employs singlet fission property has already been fabricated.34

Most studies of the singlet fission mechanism have focused on molecules whose lowest singlet excited states (S1) are optically allowed and can be described as electron promotion from the highest occupied molecular orbital (HOMO) to lowest unoccupied molecular orbital (LUMO). These molecules (denoted “Class I”)13 include tetracene and

35-40 pentacene. . Their S1 energy level is close to twice of their triplet. A precursor state that is almost degenerate with S1 and carries on multiexciton characters is proposed to

166

167

couple with S1 and the two separated triplet states. This coupling can occur either via a direct mechanism,39, 40 or via an indirect mechanism with the mediation of charge transfer states.41, 42

Class III is another group of fission chromophores differing from class I. Polyenes are among this class of molecules. Their lowest optically allowed transition state is to the

S2 state, which corresponds to a single electron promotion from HOMO to LUMO. The lowest singlet excited state (S1) is a dark state which is characterized mainly by a configuration with both electrons in HOMO promoted to the LUMO.43 The doubly excited S1 state can be described as two local triplet excitations located on the same molecule,44-47 and therefore is suggested as the precursor for singlet fission.13, 47 Energy dissipation processes during the fission of class III may be different from that of class I, because of the large energy gap between the high-lying optically allowed S2 and the low- lying T1 states. Moreover, triplet-triplet (T-T) annihilation through the low-lying T2 state of class III chromophores is predicted to compete with fission.13. Fewer studies have been conducted for singlet fission of chromophores of class III, in comparison with class I.

III thus becomes blurred, because it seems that the two triplet state can be formed on a single molecule.

168

We have chosen zeaxanthin, a type of carotenoids, as the model compounds to shed lights on chromophores of class III. Zeaxanthin can form three types of aggregates in aqueous/organic solvent systems.28 All of them are best characterized as H-type coupling with their ground state transition dipoles aligned side-by-side, whereas the coupling strengths of aggregates are different.28, 52 Based on previous theoretical and experimental insights, we now rename the previous J1-aggregate and H-aggregates as

“WH” and “SH” to denote weak and strong H-aggregates, respectively. These two zeaxanthin aggregates provide a pair of model systems to test the relationship between interchromophore coupling strength and singlet fission yield.

Our earlier studies with time-resolved resonance Raman (TRRR) and transient absorption (TA) spectroscopy have answered several key questions for singlet fission of class III. With TRRR spectroscopy, we confirmed the necessity of interchromophore for

26, 27 the fission of zeaxanthin. The TA kinetics excluded the possibility of the S1 state as the fission precursor for zeaxanthin aggregate.53 It was also found that both the absorption spectra and evolution kinetics of excited states strongly depend on the couplings of aggregates.53 However, the explicit relationship between the coupling strength and the intermolecular singlet fission in class III molecules is still unclear.

One major challenge associated with the investigation of any system with interchromophore coupling is the development of new (or shifted) transitions as a result of the exciton coupling effect. This distortion is not particular for zeaxanthin aggregates, but also affects research on singlet fission in solid state acenes.13, 54 The spectral changes lead to difficulties for assigning different excited states and quantitatively determining their yields based on one spectroscopic techniques.

169

In this work, we present a complete resonance Raman analysis on singlet fission and T-T annihilation processes of zeaxanthin aggregates based on the roadmap from TA spectra. We first try to clarify the ambiguities left in previous TA study. The identities of excited species are determined according to their unique vibrational signatures. With improved knowledge about the excited states, we are able to connect the exciton coupling strengths with the singlet fission yield, and discussed the influence of coupling strength on singlet fission and the parallel internal conversion processes in different aggregates.

Moreover, we reconstruct the excited states absorption spectra, and compare it with the resonance Raman excitation profile (REP). The discussion of exciton coupling in excited state transitions and its influence on TA and TRRR spectra can help improve the methodology for analyzing singlet fission systems. Last, to reveal the energy dissipation process in fission and annihilation, we employ anti-Stokes (AS) TRRR spectroscopy to monitor the vibrationally excited species formed in the photoexcitation of zeaxanthin aggregates.

6.2 Materials and methods

6.2.1 Preparation of samples

Zeaxanthin aggregates were prepared as described previously.26, 28 Briefly, the

WH-aggregate formed upon rapid addition of water to a 100 μM solution of zeaxanthin in tetrahydrofuran (THF) until the water:THF ratio was 9:1 (v/v). Similarly, the SH- aggregate formed upon rapid addition of water to a 50 μM solution of zeaxanthin in ethanol to a solvent ratio 8:2 (v/v). The mixing processes were accelerated by swirling and sonication. The samples were sonicated for another 10 min before spectroscopy was performed. The degradation of samples in all spectroscopic experiments was monitored

170 by UV-vis absorption. For all experiments the degradation was kept less than 10%.

6.2.2 Picosecond time-resolved Raman spectroscopy (TRRR)

Fundamental pulses for the picosecond time-resolved resonance Raman experiment were generated by a Ti:sapphire regenerative amplifier (Spectra-Physics;

Spitfire XP-Pro) seeded by a Ti:sapphire oscillator (Spectra-Physics; Mai Tai). The wavelength of the oscillator and amplifier was tunable from 740 to 840 nm. The repetition rate of the amplified pulse train was 1 KHz. The pulse energy was >3mJ and the pulse duration was 90-120 fs as measured by a single-shot autocorrelator (Minioptic

Technologies; Delta).

The amplified pulses were reduced in diameter by a factor of 0.7 with pair of curved mirrors. The beam was then directed into a frequency-doubling unit that was also designed to narrow the bandwidth for vibrational spectroscopy (Light Conversion;

SHBC). Within the unit, the fundamental was split into two beams of equivalent pulse energies, and a grating was used to apply a positive chirp along one path, and negative along the other. The oppositely-chirped pulses were then combined spatially and temporally in a Type I BBO crystal for frequency doubling. By controlling the chirp along the two optical paths, the duration of the second-harmonic pulses could be varied.

The pulselengths were set to 7 ps for Stokes Raman experiments, and 4 ps for anti-Stokes experiments. In both cases, the FWHM bandwidths of the second harmonic were <10 cm-

1. The center wavelengths of the doubled output could be varied by approximately 5 nm by fine-tuning the relative arrival time of the chirped pulses at the crystal.

The ps-length second harmonic beam was divided into pump and probe paths of the TRRR setup. Wavelengths different from the second harmonic were generated as

171 needed for pump, probe, or both, via stimulated Raman scattering from hydrogen gas.

One or both beams were focused into a pair of pipes (length 47 cm) with windows at both ends, and that were filled with either compressed D2 (1000 psi) or H2 (1200 psi) gas.

Stimulated Raman beams with Stokes or anti-Stokes shifted wavelengths were generated.

A Pellin Broca prism and two irises positioned after each pipe were used to select an appropriate wavelength for the pump or probe. Exact probe wavelengths used in experiments were determined by comparing with standard lines of neon or mercury narrow-line emission lamps (Newport; Ne 6032, Hg(Ar) 6035). All pump and probe wavelengths used in this work are listed in Table 6.1. The probe wavelengths are illustrated in Figure 6.2.

The instrument response times were 10 ps for the Stokes experiments and 5.6 ps for the anti-Stokes experiments, as determined by cross-correlation of the pump and probe pulses. Different time delays between pump and probe were achieved by controlling the arrival time of the pump pulses with a translation stage (Newport,

IMS600CCHA) and broadband hollow retroreflector (Newport, UBBR2.5-1UV). The zero picosecond delay point was determined where the cross-correlation signals of pump and probe achieves its maximum. The pump and probe beams were combined with a dichroic beamsplitter (Thorlabs, DMLP425) and directed collinearly to the sample. The beams are focused at the sample with a pair of cylindrical lenses with focal lengths of 150 and 75 mm. The cylindrical lenses were set orthogonal to each other allowing separate control of the focus along two dimensions. Beams were loosely focused along optical axis of the collection lens, and tightly focused in the transverse direction. This focal condition utilized the depth of field of the collection lens to provide efficient collection

172 for Raman signals, while it kept probe photon densities sufficiently low to prevent probe self-excitation. The beam cross-section at the sample point was determined by monitoring the transmitted beam power upon orthogonal translations of a razor blade through the sample point with a micrometer. The area was on the order of 10-3 cm2, and the pump area was always larger than the probe. Exact pump and probe spot sizes can be found in the Appendix.

The fraction of molecules excited by the laser (excitation ratio) can be estimated from beam parameters. The pump excitation ratios were typically 12-18%. The excitation ratio for the probe was selected to be approximately 3%. This level was determined by comparing the spectra to ones acquired with extremely low (0.3%) probe excitation, and verifying that no broadening for either ground- or excited-state bands were observed.

For anti-Stokes experiments, the choice of probe power was particularly important. The ratio between anti-Stokes and Stokes signal (AS/S) for the 1520 cm-1 mode is very sensitive to self-heating and was monitored to select a safe pulse energy (or excitation ratio). The AS/S ratio measured with CW laser was considered to represent the population of vibrationally hot state under thermal equilibrium at room temperature. The probe excitation ratio of 557 nm experiment needed to be less than 0.5% to obtain similar result as the CW experiment; a value of 0.3% was selected (see the Appendix). For 592 nm probe, there was no absorption for the sample. Therefore, the probe power density could be raised to improve the signal strength.

During an experiment, the solution of aggregate (~20 mL volume; 15 µM in zeaxanthin) was cycled by a syringe pump (KDS, Legato 270) through a quartz capillary with 2×2 mm interior dimensions. Raman scattering was collected at 90 degrees relative to

173 the excitation beam with an F/1.2 camera lens (Canon, FD; focal length 55 mm). The scattered light was focused onto the entrance slit of a spectrograph (Horiba Jobin Yvon, iHR-320) with an F/4, 200 mm focal length achromatic doublet lens. The Rayleigh scattering of the probe was blocked by edge filters or notch filters (Table A6.1). The spectra were acquired with a back-illuminated CCD (Andor, Newton DU920N-BU) thermoelectrically cooled to -90 oC.

Table 6.1 Laser wavelengths of the pump and probe beams Pump /nm c Probe /nm c Samples/Expt. a 471 (1 S D2) 413 WH and SH/Stokes a 413 471 (1 S D2) WH/Stokes a 413 497 (1 S H2) WH/Stokes a 416 554 (2 S D2) WH and SH/Stokes a 418 557 (2 S D2) WH/As-Stokes b 401 527 (2 S D2) WH/Stokes b 400 599 (2 S H2) WH/Stokes b 397 592 (2 S H2) WH/AS-Stokes a Second harmonic wavelength; fundamental (regenerative amplifier) centered at 830 nm. b Second harmonic wavelength; fundamental (regenerative amplifier) centered at 800 nm. c Parenthetical notes source of beam: e.g. “1 S D2” represents the 1st Stokes- shifted stimulated Raman line from D2 gas.

For Stokes Raman spectra, a 2400 groove/mm holographic grating was used for experiments with 413, 471, and 497 nm probe excitation wavelengths, and a 1200 groove/mm grating for all other experiments. The width of the entrance slit was 0.15 mm.

The spectral bandwidth ranged from 4 to 8 cm-1 for the Stokes experiments, depending upon the grating and wavelength region. For anti-Stokes experiment, a 1200 groove/mm grating was used. The width of the entrance slit was 0.20 mm, and the overall spectral resolution was approximately 15 cm-1 for the 557 nm and 592 nm excitation wavelengths.

Multiple (pump+probe), (probe only) and (pump only) spectra were acquired in an

174 interleaved manner for each experiment in order to cancel changes caused by slight degradation of the sample, or drift in pulse energies. The total integration time for each raw spectrum was between 40 and 120 min. A probe-only spectrum of the pure solvent was also collected. The Raman shifts was determined by calibrating with an acetonitrile/toluene (1:1, v/v) standard solution. All spectra were corrected for self- absorption by dividing by the transmittance spectrum of the sample for a 1 mm pathlength.

6.2.3Transient microscopic resonance Raman spectroscopy

Resonance Raman spectra of triplet excited state of monomeric zeaxanthin were collected with a modified optical microscope which is described in detail in chapter 7.

The triplet state of zeaxanthin was generated through triplet energy transfer from anthracene. ~2 mM of anthracene and ~120 µM of zeaxanthin was dissolved in 22 ml of acetone, and contained in a brass cylinder pressurized by high purity N2. The pump wavelength was 375 nm, and the power at the sample was about 1.2 mW. The probe wavelengths 457.9 nm, 488.0 nm and 520.8 nm were selected to be resonant with the absorption bands of the lowest triplet (T1) state of monomeric zeaxanthin. The probe power varied from 0.6 to 1.5 mW. The pump and probe beams were focused with an apochromatic microscope objective (Seiwa, UV PlanAPO 20, NA = 0.5, focal length 10 mm) with their foci separated by about 10 µm. The sample solution was pressure-driven through a 50 µm diameter capillary that intersected the pump and probe beams with high- purity N2. The pressure of N2 was set at 1.4 MPa (200 psi), so that the velocity of the solution in the center part of capillary is about 5 m/s. The setup provides a time delay of 2

µs between pump and probe. The Raman signal from the probe was collected by the same

175 objective and dispersed in the spectrometer (Horiba Jobin-Yvon, iHR-320) using a 1200 groove/mm holographic grating. The spectra were acquired with an open-electrode CCD detector (Synapse). Four raw spectral components were acquired, namely, a

(pump+probe), a (probe only), a (pump only) spectra and a (solvent only) spectra, each for 20 min integration time.

6.3 Results

6.3.1 The ground state absorption and terminology

The two aggregates have distinct absorption spectra, as shown in Figure 6.1. The aggregate formed in EtOH/water solution has a highly blue-shifted absorption band, with peak maximum at 388 nm. The aggregate formed in THF/water solution has a slightly red-shifted absorption, and three distinct maxima in the visible range at 450, 474, and 513 nm. Theoretical and experimental studies indicate that the absorption spectra of the two aggregate are both consistent with H-type exciton coupling, and that the differences between them reflect the differing strength of the interaction among the carotenoids.28, 52

Therefore, the J-aggregate label, which we and others 26, 28, 55, 56used previously for aggregates with redshifted spectra, is best avoided. Here, we replace the previous H- aggregate and J1 aggregate names with SH and WH to denote strongly- and weakly- coupled H aggregates, respectively.

176

Figure 6.1. UV-vis spectra of zeaxanthin monomer in EtOH, SH-aggregate in 20%

EtOH/H2O (v/v) and WH-aggregate in 10% THF/H2O (v/v).

6.3.2 Transient absorption results and the selection of TRRR probe wavelengths

The probe wavelengths, and pump-probe delay times selected for TRRR experiments were guided by results from fs-TA spectroscopy.53 Four absorption bands

(Figure 6.2) were observed in the fs-TA spectroscopy of the WH-aggregate, with maxima at 465, 500, 550 and 590 nm.53 The Raman probe wavelengths 471, 497, 554, 599 nm were selected to be near-coincident with these maxima, and others (413 or 527 nm) were employed as well. The 590 nm band decayed on a 7.8 ps timescale by transient absorption spectroscopy; its position and lifetime were consistent with assignment to the

53 S1 excited state. In the present TRRR experiments, the 590 nm band was probed at time delays less than 15 ps, and at 150 ps, after the pump pulse. The three other transient absorption bands decayed by a factor of 3 within 100 ps, but about 15% of these three transients were observable at ~3 ns delay and the signals can last for hundreds of nanoseconds. These bands were probed at 150 ps or later time delays, to eliminate possible contribution of hot ground state.

177

Figure 6.2. Transient absorption spectra of weak (WH) and strong (SH) H-aggregates of zeaxanthin at 3 and 150 ps delay between pump and probe. The dash-dotted lines indicate the probe wavelengths for Stokes (black) and anti-Stokes (red) TRRR experiments.

The position of the 550 nm band in fs-TA experiments, 53 as well as prior TRRR

26 spectra strongly support its assignment to a T1 excited state. The 500 and 465 nm bands decayed with similar kinetics as the 550 nm band, but their identification was less certain.

Femtosecond TA spectra of the SH-aggregate consist of a dominant band at 410 nm and a weak absorption at 550 nm, both of which remain for hundreds of nanoseconds.53 These bands were probed with Raman excitations wavelengths of 413 and

554 nm, respectively. The Raman signals from the SH aggregate acquired with other excitation wavelengths as used for the WH-aggregate were too weak for analysis.

The choice of anti-Stokes Raman probe wavelengths is described in section 6.3.5

178

6.3.3 TRRR Stokes spectra

The TRRR spectrum of the WH aggregate probed at 554 nm at 150 ps delay

(WH-554) exemplifies our approach with the data workup. A TRRR difference spectrum

[(pump+probe) – (probe only)] of the WH-554 is shown in Figure 6.3. This spectrum can be considered an analog to difference spectra in transient absorption spectroscopy. A spectrum that highlights the excited state signals is obtained by adding a small fraction of the probe-only spectrum to the 1:1 difference spectrum. The smallest fraction that yields a spectrum that is fully above baseline, and without a negative dip in the ethylenic region, is considered optimal. In the example of Figure 6.3, 11% of the probe is added to the 1:1 difference. Raman bands of the resultant excited-state spectrum are centered at 1499,

1276, 1242, 1127 and 1015 cm-1. The excited-state spectrum from TRRR resembles the resonance Raman spectrum of the triplet-sensitized zeaxanthin monomer probed at 520.8 nm. The main triplet Raman bands are within 3 cm-1 of those reported for monomeric zeaxanthin and β-carotene.57-60

179

Figure 6.3. (Middle panel) TRRR spectra of the WH aggregate, acquired at 150 ps with 554 nm probe. A difference spectrum (see text) is shown in blue. Negative features reflect loss of the ground state. The transient spectrum (black) results from the re- addition of a scaled probe-only spectrum (lower panel) to the difference. (Upper panel) Raman spectrum of the triplet-sensitized zeaxanthin monomer acquired with a cw-laser at

520.8 nm. Dotted lines indicate the correspondence of the bands of the monomer T1 excited state, with those of the transient TRRR spectrum of the WH aggregate.

The TRRR spectra of the WH-aggregate acquired with 599 nm probe (WH-599) and at early time delay are stand out because of the appearance of a high frequency band at 1770 cm-1 (Figure 6.4). The upshifted ethylenic mode and the short lifetime are

61-63 established signatures of the S1 state. The TRRR experiments at time delays from 0 ps to 12 ps reveal that the 1770 cm-1 ethylenic decays with a time constant of 6.6 ps, in good agreement with the 7.8 ps for the lifetime of the 590 nm band as found by fs-TA.

Changes of lineshape of the vibrational band from 0 to 4 ps reflect vibrational cooling: a shoulder on the low frequency side disappears within 4 ps, and the overall band narrows by 20-30 cm-1. In the 1000-1600 cm-1 region, bands similar to those found with 554 nm

180 are apparent, and can be assigned to the T1 excited state. At later time delays (>150 ps), only these bands remain.

Figure 6.4. TRRR spectra of the WH-aggregate, probed with 599 nm. A monomer TRRR

S1 state spectrum (4 ps time delay) probed at 550 nm is shown in red for comparison.

The spectrum of the WH-aggregate acquired with 497 nm excitation (WH-497,

Figure 6.5 A) is more complex than spectra acquired with 554 or 599 nm. Similar to WH-

554, the 1-to-1 difference spectrum of WH-497 shows positive and negative features. In order to remove all negative features across the full Raman spectrum, we must add 7% of the ground state (probe only) at minimum. Smaller addition factors yield a negative feature at 1010 cm-1. With the selection of 7% addition, small triplet bands e.g. at 1127 cm-1 are observed. However, the bands of the T1 state are overshadowed by others centered at 1157, 1212 and 1522 cm-1 that coincide with the ground state bands. The appearance of these bands implies that the simple scalar addition of the probe-only spectrum has some limitations. The pump excitation not only generates triplets, but also induces changes in relative band intensities of the ground-state. The changes clearly

181 depend upon the probe wavelength, and it is reasonable to attribute them to the alterations in exciton coupling caused by the pump. The probe wavelengths 413 and 471 nm also yield transient spectra that are necessarily a complex mixture of T1 and ground state bands.

182

Figure 6.5. (Middle panel) TRRR spectra of the (A) WH aggregate with 497 nm probe, and (B) SH aggregate with 413 nm acquired at 150 ps. Difference spectra (see text) are shown in blue. Negative features reflect loss of the ground state. The transient spectra (black) result from the re-addition of a scaled probe-only spectrum (lower panel) to the difference. (Upper panel) Raman spectrum of the triplet-sensitized zeaxanthin monomer acquired with a cw-laser at (A) 488.0 nm and (B) 457.9 nm. Dotted lines indicate the correspondence of the bands of the monomer T1 excited state, with those of the transient TRRR spectra of aggregates.

183

The magnitude of ground state depletion was also found to vary considerably for different probe wavelengths, despite similar pump excitation conditions (Table 6.2). We emphasize that the observed magnitude of ground state depletion is determined by the percentage of the probe-only spectrum that is necessary to achieve a fully positive transient Raman spectrum. The observed loss of GS tends to be smallest for probe wavelengths that are resonant with the blue side of excited-state absorption. At 413 nm probe of the SH aggregate, the pump appears to cause changes in exciton coupling so that no loss of ground state signal is observed.

Table 6.2. Loss of Raman scattering caused by pump excitation of WH and SH TRRR spectra measured at various probe wavelengths. Percentage Observed Loss ratioc Probe Aggregate Excitation depletion of (%) (nm) (%)a GSb (%) WH 413 14 3.7 26 471 15 8.5 57 497 15 7.0 47 527 15 11 73 554 15 11 73 72 599 18 13 SH 413 15 0d 0 554 14 11 79 a The percentage of molecules excited in the aggregate was estimatedusing the measurements of the beam, and with Beer’s Law. (See the Appendix) b The observed depletion of the ground state is the minimum percentage of the probe-only spectrum that must be added to the difference spectrum for a fully positive result (See Figure 6.2, 6.3 and the Appendix) c The ratios between observed bleach factors and the estimated excitations. d No loss in ground state was observed, as shown in Figure 6.5 (B).

The temporal evolution of the T1 and the ground-state-like bands is subtle, as illustrated by sequential spectra with 471 nm probe (Figure 6.6). In the fingerprint region all spectral features are conserved from 10 ps to 2 ns time delay, except a 5 cm-1 shift of

184 the peak position are observed for the ~1155 cm-1 band (from 1152 cm-1 at 10 ps to 1157 cm-1 after 150 ps). In the ethylenic region, the overall spectral profile shows broadening in the same period of time, and the peak maximum shifts from 1513 cm-1 to 1520 cm-1.

These changes in the first 150 ps can be attributed to signals from hot ground state.

Raman bands that have been previously associated with the hot ground state have maxima at 1510 and 1150 cm-1.26, 63 When the hot ground state signal has decayed (e.g at

150 ps), the spectral profile and the relative intensities of T1 bands became constant.

Figure 6.6. TRRR spectra for WH-aggregates with 413 nm pump, 471 nm probe at different pump/probe time delays.

185

The SH-aggregate was probed with similar pump/probe conditions as the WH, but only the spectra excited with 554 and 410 nm had signal-to-noise ratio that was adequate for further analysis. The SH-554 experiment yielded unambiguous triplet features that similar to the monomer T1 spectrum probed at 520.8 nm. The spectral profile of SH-413 is similar to the WH-497, consisting of some triplet features and strong ground state like

Raman bands. The SH-413 experiment is unique because the different spectrum shows only positive signals and does not require the addition of any probe-only spectrum to compensate for the loss of ground state. No S1 state signal was observed for SH-aggregate.

For quantitative analysis of the T1 Raman intensities, the fingerprint region of

TRRR spectra were fitted with a set of Lorentzian functions (Figure 6.7). The extended wings of the Raman bands observed in our TRRR spectra make Lorentzian functions a better choice than Gaussian functions. These wings are also apparent in the probe excitation.

186

Figure 6.7. TRRR spectra of zeaxanthin WH and SH aggregates at 150 ps time delay with different probe wavelengths. Vertical dotted lines are drawn to guide the eye. The green lines indicate the T1 band positions, while the red lines indicate the peak positions of ground state signals. Band decomposition with Lorentzian functions is carried out for the figure printed; fitted peaks are confined within an error of ±3 cm-1. Green curves represent signal from the T1 state. Red curves represent the unidentified signals.

The fits included triplet bands that were restricted to positions within ±3 cm-1 of the monomer (1127, 1192, 1242, 1276 cm-1). Extra bands were included in the fitting, as

187 needed A Raman band centered at 1155±5 cm-1 was typically necessary to fit the profile.

The intensity of the 1155 cm-1 band varies with the probe wavelengths. For spectra acquired with bluer probe wavelengths, a band located at 1212 cm-1 is also necessary to include in to the fit. The intensities of the ~1155 and ~1212 cm-1 bands correlate with a relatively upshifted ethylenic band in comparison with that of the triplet. Broadened bands in the ethylenic region prevented a precise determination of this enhanced signal, but it appears to be nearly coincident with ground state ethylenic (~1520 cm-1).

6.3.4 Reconstructed excited state absorption (ESA) and Raman excitation profile

(REP)

The ESA spectra are reconstructed most straightforwardly by adding fractions of the ground state absorption (Figure 6.8 lower panels). The fraction is determined by (1) the need to have a positive absorption coefficient across the spectrum, (2) compensating for the spectral dip from the ground state bleach in the low wavelength region. Three reconstructed spectra are shown for both aggregate types, with different addition factors to illustrate the effects of variation. The simulation assumes that the ground state bleach responds linearly to the pump-probe combination. Non-linear effects are likely important for the aggregates, in analogy to strongly coupled J-aggregates such as pseudo-isocyanine

(PIC) dyes.64 Nevertheless, the simulated spectra in Figure 6.8 should provide some indication of the overall ESA for both aggregates.

188

Figure 6.8. Raman intensity analysis (upper panels) and reconstructed excited state absorption spectra (lower panels) of (A) WH (B) SH aggregate at 150 ps. Relative Raman intensities of the triplet state 1128 cm-1 band probed at different wavelengths are normalized against toluene 1003 cm-1 band. The ESA spectra are reconstructed by addition of scaled ground state absorption (dotted line, inversed) to make up the loss of ground state absorption in the difference spectra. (See text) Three spectra with addition of different portion of ground state absorption are shown for each aggregate.

The reconstructed ESA of the aggregates significantly differ from the monomer

T1 absorption. For the WH -aggregate, the lowest absorption band (550 nm) of T1 shows

189 redshift of 1050 cm-1 compared to the monomer. This spectral shift is close to the spectral shift observed for the electronic transitions of the S0 and S1 state. The triplet ESA bands of aggregates also appear with large energy separations compared to the vibronic progression of monomer triplet state.65 The energy differences between 0-0, 0-1 and 0-2 bands of the triplet spectra are 1820 and 1500 cm-1. For SH-aggregate, the main triplet absorption appears as a single band with a 5160 cm-1 blueshift. The 550 nm absorption band in the difference spectra blurs in the tail extending to the red. In general, spectral changes of the ESA spectra resemble the changes of the ground state absorption from monomer to the aggregates.

The intensity profile of resonance Raman signals can be correlated with the electronic absorption, as shown in Figure 6.8 upper panel. Triplet Raman signals are represented by the 1127 cm-1 band, which is well separated from the Raman featured at

-1 ~1155 cm . The band intensities of T1 obtained from band decomposition are normalized against the solvent standard (toluene 1003 cm-1) acquired under the same collection condition. For the WH-aggregate, the Raman excitation profile yields two triplet electronic transitions centered around 471 and 554 nm. The REP corresponds to the two reconstructed ESA bands with the peak positions at 465 and 550 nm. However, the ESA band with peak maximum at 500 nm coincides with a local minimum of the triplet

Raman signals. For SH-aggregate, the two centers of triplet electronic transitions located at 410 and 550 nm, and are separated by 6210 cm-1. The strongest Raman signals of the

T1 state for the SH is observed when resonant with 550 nm absorption, despite the much greater strength of the 410 nm transition in the reconstructed absorption spectrum.

190

6.3.5 TRRR anti-Stokes spectroscopy

Anti-Stokes Raman spectra were acquired for the WH aggregate to monitor species which are vibrationally excited. Raman excitation profiles of hot states generally shift to the red side of the ground state spectra by one vibrational quantum for the REP under consideration.66 Therefore, the 557 nm (AS-557) and 592 nm (AS-592) probes were selected in anti-Stokes TRRR experiments to enhance the anti-Stokes scattering from hot ground and triplet excited states, respectively. Raman excitation wavelengths that were most resonant with the S0S2 ground state transition were avoided to minimize self-heating caused by the probe. Anti-stokes experiments for SH-aggregate were attempted, but poor signal-to-noise ratio of these spectra prevented their analysis.

The hot ground state signals are illustrated with AS-557 experiment and shown in

Figure 6.9A. The pump-induced signals show rapid growth within the pulse duration. In particular the 1520 cm-1 ethylenic shows an enhancement of 6.7. The three major Raman bands at 1520, 1157 cm-1, and 1007 cm-1 show no spectral shift or broadening compared to the probe only spectrum, indicating no hot ground state with higher vibrational energy level is observed. Two low frequency modes (517 and ~370 cm-1) are also very active under pump excitation, and show increases of 5.1 and 3.7, respectively. The ~370 cm-1 mode shows a bathochromic spectral shift from 361 to 371 cm-1 through the decay indicating vibrational cooling.

191

Figure 6.9. Anti-Stokes TRRR spectra of the WH-aggregate pumped with 418 nm or 397 nm, and probed with 557 nm or 592 nm, respectively. (Probe only) spectra (red) are overlaid with each set of spectra for comparison. For 557 nm panel, the (probe only) spectra are shown as it is for comparison of the enhancement due to pump excitation. For 592 nm panel, (probe only) signals of ethylenic and finger print regions are normalized to the band intensities of each (pump+probe) spectra respectively to indicate changes of band shapes.

The fast initial growth and decay of anti-Stokes signals are beyond our instrument response, and hence cannot be analyzed by more sophisticated modeling. The decay kinetics are fitted with double exponential. The majority of pump-induced signals recover

192 within 0.3~0.5 ps. The slow components of the decay have time constants of tens of picoseconds. At 150 ps time delay, the pump-induced signals proximate to the level of

“probe only” spectrum, but do not fully recover.

For the AS-592 (Figure 6.9B), the most important distinction compared to AS-

557 is the spectral broadening on the low frequency side of ethylenic and fingerprint regions. The new spectral feature in the fingerprint region is prominent, and appears close to the triplet band at 1127 cm-1. The broadening signals disappear after 12 ps. Other major ground state Raman signals show similar spectral and kinetic features to WH-557.

6.4 Discussion

6.4.1 Exciton coupling of ground state transition in aggregates

To illustrate the effect of exciton coupling on singlet fission, it is important to determine the coupling strengths of the ground state transition in the aggregates. The coupling strength can be evaluated by 2J/Δ, where J is the exciton bandwidth which represents the interaction energy between the transition dipoles, and Δ is the Franck-

Condon bandwidth.67, 68 Coupling strengths of carotenoid aggregates have been measured previously by Zsila et al.55, and recently discussed by Spano with a more sophisticated theoretical treatment.52 Quantitative analysis of coupling strengths based on UV-vis spectra of our aggregate are presented in below, in the manner of Spano’s approach.52

The exciton bandwidth for the WH-aggregate (JW) can be determined from the spectral details of vibronic progression of the 0-0 and 0-1 absorption bands.52, 69 The first method relates the bandwidth to the relative line strengths between the 0-0 and 0-1 bands, according to52

Eq. 6.1

193

-1 -1 where the ratio of oscillator strength I00/I01=0.55, and =1400 cm , so JW = 580 cm . A second method to determine the bandwidth relies on the spectral separation between 0-0

-1 28 52 and 0-1 bands (Δ01=1470 cm , literature ),

Eq. 6.2

-1 Following the second method, JW = 700 cm . The results of both methods fall in the weak or intermediate coupling regime, with 2J/Δ < 1.

The absorption spectral shift of the weakly coupled aggregate is dominated by nonresonant dispersive shifts (D),52, 69 which can be read from Figure 6.1 as -1260 cm-1.

Spectral splitting and the vibronic levels of the WH-aggregate are illustrated in the left part of Figure 6.10.

Figure 6.10. Exciton coupling model of ground state electronic transitions for WH and SH aggregates, in comparison with the monomer (Mon). Superscripts W and S denote electronic states that are weakly and strongly coupled in the transitions. The (+) and (-) label the in-phase and out-of-phase coupling of transition dipoles, respectively. Accessible vibronic states are illustrated with dashed lines. Arrows represent radiative (solid line) and non-radiative (dashed line) transitions. Weak optical transitions are denoted with arrows with gray lines. Parameters: nonresonant shift D=-1260 cm-1; -1 -1 -1 vibrational quantum ω0=1400 cm ; exciton coupling JW =580 cm and JS = 4900 cm .

194

The exciton bandwidth of the SH-aggregate is measured directly from the UV-vis spectra. The position of the strong absorption band located at 388 nm reflects spectral shift due to both the resonant exciton coupling and the nonresonant dispersive shift,

ω Eq. 6.3

-1 2 where the monomer 0-0 band energy is = 20730 cm , is the Huang-Rhys factor and equals 1.1,52 and D is assumed to be the same value as in WH-aggregate. From Eq.

-1 6.3, we calculated the exciton bandwidth for the SH-aggregate, JS = 4900 cm (Figure

6.10, right). Minor absorption bands of the SH aggregate that coincide with major absorption bands of the WH aggregate, as well as their similar emission profile, strongly suggest that there is a subset of weakly coupled chromophores in the SH aggregate.28

In addition to the shifts of the electronic transition, exciton coupling can also influence the intensity of resonance Raman scattering. We observed a general decline of

Raman intensities for the aggregates. Raman cross-sections decrease by factors of 2 or 3 for the WH-aggregate, and 3 to 50 for the SH-aggregate compared to the monomer depending on probe wavelengths.70 The decrease of Resonance Raman cross-section can be attributed to exciton coupling in the following three ways. (1) Exciton coupling can cause mixing of electronic transitions, which decrease the transition dipole of strongly allowed transitions, known as hypochromism.71, 72 Resonance Raman cross-sections have a strong dependence (fourth power) on the transition dipole moment.73, 74 (2) The resonant coupling effect causes exciton delocalization over several chromophores, which decreases the Frank-Condon displacements.75 (3) The possible existence of a dephasing mechanism for the excited state in the coupled system, which can truncate the resonance

Raman cross-section because of the increase in the homogeneous broadening factor.74.

195

Considerations of effects of exciton coupling on the Raman cross-sections are essential for the spectral analysis in this work.

6.4.2 The excited states in aggregates

The singlet states. The singlet excited states are the major excitation and relaxation pathways for monomeric carotenoids. The strong optically-allowed transition

1 - 1 + 1 + from S0 (1 Ag ) to S2 (1 Bu ) dominates the ground state absorption spectra. The S2 (1 Bu )

1 - 76-79 state relaxes through internal conversion to the S1 state (2 Ag ) in 100~200 fs. For

79, 80 zeaxanthin, the S1→S0 internal conversion has time constant of ~9 ps. The existence

* 1 - of other intermediate dark singlet states such as S (1 Bu ) between S1 and S2 is still a controversial topic.81-84

In the aggregates, the strongly allowed transition remains S0→S2. The formation of the S2 states gives rise to a broad transient absorption in the NIR region on the sub-

53 picosecond timescale. The lifetimes of S2 signals in the aggregates (SH: 300 fs and WH:

450 fs) are slightly longer than that of the monomer,53 but are still too short to characterize with our TRRR experiments. Stimulated Raman spectroscopy with femtosecond time-resolution has recorded the Raman features of this excited state.85

The behavior of S1 state in aggregates significantly differs from the monomer. The

S1 is only observed for WH-aggregate with a quantum yield of 10-30% determined by

53 S1→Sn absorption signal, and is entirely absent from in the SH-aggregate spectra.

Therefore, we conclude that S1 state is not the primary relaxation pathway for either aggregate. In WH-aggregate, the S1→Sn absorption maximum is shifted from 550 nm of the monomer to 590 nm. The 1230 cm-1 bathochromic shift is caused by a combined effect of exciton coupling between S1→Sn and nearby transition dipoles and nonresonant

196

53 dispersive shift. The lifetime of the S1 state in aggregate is slightly shorter than in the

56 monomer. TRRR spectra obtained with probe resonant to the S1 absorption (WH-599) featured a 1770 cm-1 band. Compared to the well known TRRR spectra of monomer carotenoids,61-63 this signature ethylenic band is downshifted by 25 cm-1. This downshift can be explained by the lowered S2 energy level in the WH-aggregate, which enhances vibronic coupling between S1 and S2 states. A similar trend has been observed for the

C=C stretch frequency variations of carotenoids in different solvents.62

The S* state has been proposed as the precursor of the triplet state.8, 9, 82 Raman experiments have reported that the S* state shows several vibrational bands in the

1700~1850 cm-1 region,86 which do not appear in any of our TRRR spectra probed at different wavelengths. However, we cannot exclude the possibility of S* as a relaxation intermediate. Certain processes, such as singlet fission, can possibly facilitate the decay of the S* state so that its lifetime is significantly shorter than the 5.6 ps time-resolution of our TRRR setup.

The T1 state. The triplet state is not a primary relaxation pathway for monomeric carotenoid molecules due to much slower intersystem crossing rate compared to the internal conversion. For zeaxanthin monomer in solution phase, the triplet quantum yield

(Q.Y.) is ≤0.2%.87 In aggregates, the high yield of triplet state is due to ultrafast singlet fission that occurs on a subpicosecond time scale.26, 53

The T1→Tn electronic transition is probed by TRRR at different wavelengths.

First, for both WH and SH-aggregates, the 554 nm probe yields Raman spectra that are identical to the sensitized T1 state of zeaxanthin monomer. This observation implies that the 550 nm excited state absorption is attributed to pure triplet absorption. For WH-

197 aggregate, another electronic transition of the triplet state is centered at 465 nm, which shows strong triplet resonance Raman signals when probed at 471 nm. The 465 nm band, together with the 500 and 550 nm absorption bands are most likely the vibronic progression for the T1→Tn transition coupled with nearby transitions, as illustrated in right panel of Figure 6.11.

The maximum of triplet absorption of the SH-aggregate is located at 410 nm,

-1 which is separated from the weak absorption band of pure T1 (550 nm) by 6200 cm . The large energy separation is beyond any possible vibronic progression and cannot be attributed to two different electronic transitions, because the computational work has

43 determined that there is only one prominent triplet electronic transition, T1→T4. It is interesting to notice that the T1 Raman signals probed at 413 nm are weaker than those probed resonant with the 550 nm band, despite the 10 times stronger absorption at 410 nm. The largely blueshifted absorption and much smaller T1 Raman cross-section at 410 nm region are both consistent with assigning this band to a T1→Tn transition that is strongly exciton coupled to its surroundings, whereas the 550 nm band is a signature of the weakly coupled triplet state which is similar to the WH-aggregate (Figure 6.11).

198

Figure 6.11. Exciton coupling model for triplet excited states generated from singlet fission in zeaxanthin SH and WH aggregates. Superscripts W and S denote electronic states that are weakly and strongly coupled in the transitions. The (+) and (-) label represent in-phase and out-of-phase coupling of transition dipoles, respectively. Accessible vibronic states are illustrated with dashed lines. Arrows represent radiative (solid line) and non-radiative (dashed line) transitions. Weak optical transitions are denoted with gray lines.

Observation of ground state signals in TRRR spectra. Besides the triplet state features, TRRR experiments with probe wavelengths bluer than the 554 nm include

Raman signals with bands positioned around 1520, 1212, and 1155 cm-1. Both the frequencies and the bandwidths of these signals are coincident with the (probe only) spectra. The lifetime of these signals is almost the same as the T1 features excluding the assignment to hot ground state because the majority of vibrational cooling of the S0 completes within 20 ps according to the anti-Stokes experiment.

We have also considered that the signals belong to an excimer state with charge transfer (CT) characteristics. This consideration is because carotenoid molecules in crystal structure can be as close as <4 Å to each other.88 According to a theory proposed by Lin et al., for polyenes with conjugation number N=11, the contribution of CT state becomes significant when the interchain distance is less than 4.5 Å.89 However, the

199

Raman intensity profiles reveal that the electronic transitions corresponding to these signals appear at 500 nm for WH, and 410 nm for SH. In contrast, the strong excimer→ion-pair transition usually shows in the near-IR region.90, 91 A detailed energy analysis using our data following the method of Katoh et. al90 (shown in the Appendix) requires unreasonably high excimer stabilization energies. Therefore, the assignment to an excimer is not valid.

After excluding possible excited species, the close Raman band positions of these signals to the S0 state make us to consider assigning these species to the ground state. The ground state signal appears due to artifacts in the data workup process, which will be explained in the later part of this discussion.

6.4.3 Singlet fission in different aggregates

The singlet fission yield can be estimated through different methods from TA or

TRRR data. Previously, we have determined a triplet quantum yield (QY) of 90-200% for the WH-aggregate with TRRR experiment by comparing the loss of ground state signals with the number of absorbed photons..28 Later, the triplet QY was revised to 60-80%,53 based on TA results using the extinction coefficient (1×105 M-1cm-1) of the monomer triplet.65 The TRRR experiments in the present work identify no excited-state species other than T1 and S1 states.. If the S1 yield in the WH-aggregate is 30% as suggested from

53 the transient absorption spectra, then the T1 yield from singlet fission may be as high as

2  70% = 140%. For the SH-aggregate, the T1 yield can be close to 200%, since no S1 state is observed.

However, with a comprehensive review of TA and TRRR data, we found that all estimations described above have weaknesses. For TRRR experiments, we find strong

200 variation of the ground state loss as a function of Raman probe wavelengths (Table 6.2).

The reason for this variation will be discussed in a later section of this discussion. Triplet yield estimated with TA results relies on two assumptions: (1) Extinction coefficients of the excited states remain the same as the monomer. (2) Each band in TA spectra consists of oscillator strength from only one transition. The first assumption may be not valid because electronic transitions are strongly affected by interchromophore coupling. The comparison between Raman intensity profiles and reconstructed excited state absorption spectra shown in Figure 6.8 reveals the problem of the second assumption. Bands in TA spectra may have contribution other than the T1 transition. Therefore, to determine the exact singlet fission yield is difficult. Nevertheless, qualitative conclusions can be drawn from our results: (1) Even with weak coupling, singlet fission can still be significant, and account for a T1 yield close to 100%. (2) The T1 yield increases with stronger interchromophore coupling.

Singlet fission and the parallel reaction, S2→S1 internal conversion both take place from the lowest exciton coupling state correspond to the out-of-phase coupling of the H-type aggregates (Figure 6.11). The high-energy-level S2 state of zeaxanthin excludes the dilemma of the optical coupling strength faced by class I chromophores, which is on one thing the fission rate increases with the coupling, on the other the energy

criteria may not hold because the strong exciton coupling lower the singlet energy.48 Even for the SH-aggregate, the out-of-phase level is located at 14680 cm-1, and is sufficient to split into two triplet states. The internal conversion pathway in SH is completely shut down. This phenomenon could be related to fast delocalization of S2 over multiple chromophores. On one hand the delocalization decreases the Frank-Condon

201

overlap between the S1 and S2 potential surfaces, and hence slows down the rate of internal conversion. On the other hand, the delocalization facilitates the singlet fission process as suggested by a previous theoretical study.48

A Jablønski diagram for the relaxation of SH-aggregate is shown in the left panel of Figure 6.11. SH-aggregate consists of chromophores with both weak and strong coupling.28 The weakly coupled species shows similar behavior to WH aggregate and forms weakly coupled T1 absorbing at 550 nm. The major species with strongly coupled chromophores generates strongly coupled T1 absorbing at 410 nm. It is worthwhile to point out that this heterogeneous model for SH-aggregate is not consistent with the homogeneous bleach observed in the TA experiments with different pump excitation wavelengths.53 This independence of ground state bleach with respect to excitation can be best explained with a ultrafast energy transfer between the weakly and strongly coupled species with a rate that is faster than the resolution of these experiments.

6.4.4 The influence of exciton coupling on excited state spectra

The exciton coupling effect in aggregates becomes more complicated when considering pump/probe spectroscopy. The pump excitation removes ground state population and disturbs the original exciton coupling among the S0→S2 transitions. The formation of excited states introduces new electronic transitions, which can couple with the S0→S2 transitions. It is difficult to distinguish different transitions with TA spectroscopy because of the absorption bands tend to be broad and featureless and transitions from different excited states overlap with each other. Unlike the TA spectroscopy, resonance Raman spectroscopy allows us to determine the extent to which an “elemental” electronic transition (S0→S2 or T1→Tn) contributes to a complex

202 transition, at each excitation wavelength. For example, TRRR experiments find ground state contribution to the spectra of WH-471, 497 and SH-413, which will be discussed in detail later. Therefore, we present a comprehensive analysis of exciton coupling effect on both spectroscopies.

In zeaxanthin aggregates, there are two possible new couplings induced by pump excitation, the heterocoupling T1→Tn/S0→S2 and the homocoupling, T1→Tn/T1→Tn.

Even with the high pump excitation percentages that is typical of our TRRR experiments,

T1 states are statistically more likely to be surrounded by ground state molecules, rather than other T1 excitons. The triplet state electronic transition has a transition dipole that is similar to the ground state. Both T1→Tn and S0→S2 transition dipoles are oriented along the long axis of the zeaxanthin molecule. The reconstructed absorption spectra of β- carotene and zeaxanthin in solution phase suggest that the T1→Tn and S0→S2 absorption have similar vibronic progression and strongly overlap in the 400-550 nm region.65, 92

These findings are consistent with computational predictions.43, 93 Therefore, cross- coupling of these transitions is expected to be significant.

Exciton coupling between two T1→Tn transitions also needs to be considered. The existence of the homocoupling depends on whether the two triplet state generated from singlet fission can stay close to each other. It has been suggested that after fission the T1 states depart from each other.35 Furthermore, triplet states that do not diffuse away are expected to decay through fast geminate annihilation.53 However, triplet excited states were also reported to form stabilized pairs in systems like pyrene, perylene, anthracene.94-

96 In singlet fission, the stabilized triplet pair can appear as the relaxation product of the

1 fission precursor, (T1T1) . A recent computational work has assigned an absorption

203 feature in the TA spectra of pentacene film to a doubly excited exciton associated with two individual triplets.54 The homocoupling can be a significant contribution to the spectral shifts because of a good match of transition dipoles.

The coupling strengths of the excited states transitions can be calculated using similar methods as we presented in section 4.1. It is necessary to point out that since the heterocoupling states consist of components from both the S0 and T1, we need to use the

-1 -1 average energy level of S0→S2 (20780 cm ) and T1→Tn (19230 cm ) transitions at

-1 20010 cm as the “E00” of the monomer. For SH-aggregate, assuming that all other factors remains the same for the T1 transition we obtained an exciton bandwidth JST =

4100 cm-1 using Eq. 6.3. For WH-aggregate, the separation between the 0-0 and 0-1 transition is quite large, 1820 cm-1, which yield an exciton bandwidth of 1900 cm-1 according to Eq. 6.1.

Above estimations of exciton bandwidths suggest that for both aggregates, the coupling strengths of the excited state are comparable to the ground state coupling. The results imply the importance of homocoupling, because of the better match of transition dipoles compared to heterocoupling can help to explain the strong coupling effect.

However, the result for the WH-aggregate is still unexpected because the estimated bandwidth is even larger than the case of ground state. With TRRR experiment, it is found that the large separation between 0-0 and 0-1 bands reflects the energy difference between two transitions: the lowest transition at 550 nm is predominately T1→Tn, while the 500 nm 0-1 band consists of large portion of S0→S2, as illustrated in Figure 6.12.

204

Figure 6.12. Exciton coupling between the S0→S2 and T1→Tn transitions.

The heterocoupling of the excited state transition can still be considered analogous to ground coupling except for two noteworthy effects related to the energy difference between T1→Tn and S0→S2 transitions. (1) The unbalanced energy distribution causes that greater T1→Tn character appears on the low energy side of the spectra, while

S0→S2 shows on the high energy side. This effect is observed as strong ground state signals in the TRRR spectra probed in the blue side of the spectra. (2) The coupling between S0→S2 and T1→Tn mixes the two transitions, so that lower energy T1→Tn transition pulls the S0→S2 to the red side, and vice versa. The mixing effect evolves with the decay of T1 population. The evidence of this effect can be found from the TA spectra of SH-aggregate. With the increasing share of S0→S2 transition coupling because of the

53 decay of T1, the mixing coupling band at 410 nm shows a constant blueshift. This evolution of herterocoupling also explain the absence of isosbestic points in TA spectra,

53 despite there is only one excited species (T1) at late time delay (>150 ps).

This work employs TRRR spectroscopy to distinguish complex transitions resulting from exciton coupling of T1→Tn and S0→S2 transitions. However, it is found

205 that exciton coupling is also a challenge for TRRR experiment, because as a vibronic technique, resonance Raman intensities are affected by the coupling. A key problem is that the workup process of TRRR spectra described in the result section is completely based on a linear assumption that the formation of excited state does not disturb the surrounding ground state. In reality, upon excitation, decoupling of ground state molecules from their surroundings can cause immediate enhancement of S0 signal, to an extent that is difficult to assess. Additionally, the coupling between T1→Tn and S0→S2 transitions shifts the ground state Raman excitation profile (REP). Therefore the assumption that the loss of probe signal is linear with pump excitation is not true for a coupled system.

The breakdown of the linear assumption can qualitatively explain the artifact which brings ground state signals in the reconstructed T1 spectra. The increase of S0 signal due to excitation results in a decrease of observed ground state loss as shown in

Table 6.2. An extreme case is the SH-413 experiment, in which the shift of ground state

REP due to coupling with T1→Tn transition leads to an overall increase of S0 signal with pump excitation. Moreover, the enhancement of ground state signals is mode-specific, because the REPs do not track in lockstep with wavelengths.70 The workup is to remove the largest negative signals appear in the difference spectra. The mismatch of the ground state REP requires over-addition of the ground state signal for some modes, and leads to appearance of the positive ground state signals in the other region of final spectra. Our discussion about exciton coupling does not include the consideration of the coupling of the T1, in other words, the delocalization of T1 exciton. In most cases, the lack of S0→T1 transition dipole makes the triplet exciton more confined respect to the singlet. Studies on

206 other conjugated systems found that the lowest triplet excitons are localized.97-100

However, a recent theoretical work reported a exciton bandwidth > 400 cm-1 for the triplet of acene derivatives when wavefunctions of neighboring molecules strongly

101 overlaps. . Therefore, the T1 exciton coupling may play a role especially in SH- aggregate in which molecules are aligned almost parallel and closely to each other.52, 55

Nevertheless, the T1 coupling is expected to be a order of magnitude smaller than spectral shifts we observed. The analysis of the effects of T1 exciton coupling is beyond the scope of this work.

In conclusion, exciton coupling strongly affects excited state spectra even in resonance Raman spectroscopy which is known to be specific for each electronic state.

To interpret the excited state spectra of aggregates, especially to extract quantitative results such as the triplet yield, exciton coupling effects need to be considered carefully.

One approach to simplify the complexity is to study simple sysems such as dimers in which the exciton coupling effect is easier to model. Attempts have been made for tetracene and related compounds.50, 102, 103 Similar work for carotenoid dimers have been carrying on in our lab.

6.4.5 Vibrational relaxation in singlet fission and TT annihilation

Anti-stokes TRRR spectroscopy focuses on the energy dissipation from electronic states to vibrational states and the following vibrational cooling processes. The AS-592 experiment monitors the hot T1 state, and provides information about energy dissipation associated to singlet fission, while the AS-557 experiment monitors the hot ground state formed through TT annihilation.

207

The AS-592 experiment for WH-aggregate observes hot triplet state with its C-C modes at 1128 cm-1 excited. No other anti-Stokes triplet signals are detectable, which may suggest the fission is strongly coupled with the C-C stretch mode. Previous studies have discussed the importance of certain intermolecular or intramolecular vibrations for singlet fission.35, 40, 104, 105 Our observation suggests the C-C stretch mode may play a role in the energy dissipation during fission process.

The lower limit of energy level for the fission precursor can be inferred from the hot triplet state Raman spectra. Each hot triplet state formed from fission holds at least

-1 -1 87 1128 cm of vibrational energy in addition of the T1 electronic energy (7000 cm ).

The total energy of the fission precursor should be no less than 2×8128=16260 cm-1. This

-1 energy is clearly higher than the S1 state with energy level of 14000 cm . The observation excludes the possibility that the S1 state is the fission precursor for weakly coupled aggregates.

Quantitative analysis of AS-557 signal provides mode specific information for the

T1 relaxation process. The occupation number of the nth vibrational energy level of a certain mode is defined by Boltzmann distribution:

Eq. 6.4

where vi represents the harmonic frequency in Hz of the ith vibrational mode, n is the vibrational quantum number starting from 0, h is the Plank constant, kB is the Boltzmann constant and T is the temperature in Kelvin. For modes with frequency larger than 350 cm-1, the population of the lowest two vibrational levels (n = 0, 1) represents most molecules in the system. Therefore, when discussing the occupation for the five most

208 active AS-Raman modes (1520, 1157, 1007, 517 and 371 cm-1), only their first two vibrational excited levels need to be considered.

The total equilibrium vibrational energy stored by a zeaxanthin molecule at a given temperature is calculated by summing the energy of all 288 modes:

Eq. 6.5

For different vi, we use the vibrational frequencies obtained from DFT calculation

(B3LYP/6-31G*) and treat all modes as harmonic oscillators.28 To quantify the total vibrational energy, we consider the first nine energy levels of each mode. At 298K, the total vibrational energy is 8535 cm-1. If a molecule absorbs a pump photon at 418 nm, the total energy of the system becomes 32458 cm-1, which corresponds to T = 619 K after reaching thermal equilibrium. The occupation numbers after energy redistribution to each vibrational level can be calculated. Given the estimated excitation ratio is 12% for the anti-Stokes experiment, the population of the first excited vibrational energy levels for the five Raman modes are calculated and listed in Table 6.3.

Table 6.3. Predicted and observed enhancement factors for active Raman modes in AS- TRRR. Predicted Mode N a N b Predicted Observed 298K ex occupationc (cm-1) (×10-3) (×10-3) enhancement enhancementd (×10-3) 1520 0.645 28.3 3.96 6.1 6.7 1157 3.72 63.2 10.9 2.9 2.5 1007 7.64 86.9 17.2 2.2 1.9 517 74.2 210 90.5 1.2 5.1 371 139 244 151 1.1 3.7 a Occupation of the first vibrational excited state. b The occupation number after pump photon excitation and following thermalization is equal to the equilibrium value at 619 K. c 12%Nex+88% N298K. d The relative intensity ratio between the “pump+probe” maxima signals and the “probe- only” signals.

209

The predicted population enhancements seem to match with experimental observation for 1520, 1157 and 1007 cm-1 modes. However, the fast decay kinetics of hot ground state signals suggests that the actual occupation at the initial time should be much higher than the thermal equilibrium values. The results imply that relaxation from the T1 state to the ground state is specific to these strong Raman modes. The strong Raman active modes, specifically the C=C stretch vibration, act as the primary energy acceptors in the relaxation. This finding is similar to the cooling process in the S1→S0 internal conversion.63, 106-108

For the two active low frequency modes at 517 and 371 cm-1, the observed enhancement factors exceed the possible limit (2.61 and 3.27 for the 517 and 371 cm-1 modes, respectively, when T→∞). The abnormal enhancement may have three possible reasons. (1) The enhancement may indicate specific energy dissipation routes through these two vibrational modes. (2) It may imply changes of Raman cross-sections because of geometry changes of the annihilation products. These two vibrational modes are associated with skeletal torsion or CCC bend vibrations.109, 110 The triplet state of conjugated polyene chain has an alternate C-C and C=C bonding pattern compared to the

111, 112 ground state. It is possible that the bond order change from T1 to S0 state induces a structural disturbance on the structure of polyene backbones, and leads to special resonance enhancements. (3) The frequencies of these two bands are close to the difference frequencies of the strong resonance Raman modes (371≈1520-1157 cm-1 and

517≈1520-1008 cm-1). It is possible the intensities increase is from the difference bands, and do not related to any specific vibrational normal modes.

210

References

1. Singh, S.; Jones, W. J.; Siebrand, W.; Stoichef, B. P.; Schneide, W. G., Laser generation of excitons and fluorescence in anthracene crystals. Journal of Chemical Physics 1965, 42, (1), 330-342.

2. Swenberg, C. E.; Geacintov, N. E., Exciton interactions in organic solids. In Organic Molecular Photophysics, Birks, J. B., Ed. John Wiley & Sons: London, 1973; Vol. 1, pp 489-564.

3. Kraabel, B.; Hulin, D.; Aslangul, C.; Lapersonne-Meyer, C.; Schott, M., Triplet exciton generation, transport and relaxation in isolated polydiacetylene chains: Subpicosecond pump-probe experiments. Chemical Physics 1998, 227, (1-2), 83-98.

4. Osterbacka, R.; Wohlgenannt, M.; Shkunov, M.; Chinn, D.; Vardeny, Z. V., Excitons, polarons, and laser action in poly(p-phenylene vinylene) films. Journal of Chemical Physics 2003, 118, (19), 8905-8916.

5. Lanzani, G.; Stagira, S.; Cerullo, G.; De Silvestri, S.; Comoretto, D.; Moggio, I.; Cuniberti, C.; Musso, G. F.; Dellepiane, G., Triplet exciton generation and decay in a red polydiacetylene studied by femtosecond spectroscopy. Chemical Physics Letters 1999, 313, (3-4), 525-532.

6. Cogdell, R. J.; Hashimoto, H.; Koyama, Y., Picosecond Raman-spectroscopy of the B830 LH2 complex of Chromatium purpuratum BN-5500. Photosynthesis Research 1993, 35, (1), 79-85.

- 7. Zhang, J. P.; Inaba, T.; Koyama, Y., The role of the newly-found 1Bu state of + - carotenoid in mediating the 1Bu to 2Ag internal conversion and the excited-state dynamics of carotenoid and bacteriochlorophyll in a bacterial antenna complex. Journal of Molecular Structure 2001, 598, (1), 65-78.

8. Gradinaru, C. C.; Kennis, J. T. M.; Papagiannakis, E.; van Stokkum, I. H. M.; Cogdell, R. J.; Fleming, G. R.; Niederman, R. A.; van Grondelle, R., An unusual pathway of excitation energy deactivation in carotenoids: Singlet-to-triplet conversion on an ultrafast timescale in a photosynthetic antenna. Proceedings of the National Academy of Sciences of the United States of America 2001, 98, (5), 2364-2369.

9. Papagiannakis, E.; Kennis, J. T. M.; van Stokkum, I. H. M.; Cogdell, R. J.; van Grondelle, R., An alternative carotenoid-to-bacteriochlorophyll energy transfer pathway in photosynthetic light harvesting. Proceedings of the National Academy of Sciences of the United States of America 2002, 99, (9), 6017-6022.

10. Agostini, G.; Corvaja, C.; Giacometti, G.; Pasimeni, L., Optical, zero-field ODMR abd EPR studies of the triplet states from singlet fission in biphenyl-TCNQ and

211 biphenyl-tetrafluoro-TCNQ charge transfer crystals. Chemical Physics 1993, 173, (2), 177-186.

11. Watanabe, S.; Furube, A.; Katoh, R., Generation and decay dynamics of triplet excitons in Alq3 thin films under high-density excitation conditions. Journal of Physical Chemistry A 2006, 110, (34), 10173-10178.

12. Hanna, M. C.; Nozik, A. J., Solar conversion efficiency of photovoltaic and photoelectrolysis cells with carrier multiplication absorbers. Journal of Applied Physics 2006, 100, (7), 0745101-8.

13. Smith, M. B.; Michl, J., Singlet fission. Chemical Reviews 2010, 110, (11), 6891- 6936.

14. Tayebjee, M. J. Y.; Gray-Weale, A. A.; Schmidt, T. W., Thermodynamic limit of exciton fission solar cell efficiency. Journal of Physical Chemistry Letters 2012, 3, (19), 2749-2754.

15. Paci, I.; Johnson, J. C.; Chen, X. D.; Rana, G.; Popovic, D.; David, D. E.; Nozik, A. J.; Ratner, M. A.; Michl, J., Singlet fission for dye-sensitized solar cells: Can a suitable sensitizer be found? Journal of the American Chemical Society 2006, 128, (51), 16546-16553.

16. Roberts, S. T.; McAnally, R. E.; Mastron, J. N.; Webber, D. H.; Whited, M. T.; Brutchey, R. L.; Thompson, M. E.; Bradforth, S. E., Efficient singlet fission discovered in a disordered acene film. Journal of the American Chemical Society 2012, 134, (14), 6388-6400.

17. Ramanan, C.; Smeigh, A. L.; Anthony, J. E.; Marks, T. J.; Wasielewski, M. R., Competition between singlet fission and charge separation in solution-processed blend films of 6,13-bis(triisopropylsilylethynyl)pentacene with sterically-encumbered perylene- 3,4:9,10-bis(dicarboximide)s. Journal of the American Chemical Society 2012, 134, (1), 386-397.

18. Piland, G. B.; Burdett, J. J.; Kurunthu, D.; Bardeen, C. J., Magnetic field effects on singlet fission and fluorescence decay dynamics in amorphous rubrene. Journal of Physical Chemistry C 2013, 117, (3), 1224-1236.

19. Ma, L.; Zhang, K. K.; Kloc, C.; Sun, H. D.; Michel-Beyerle, M. E.; Gurzadyan, G. G., Singlet fission in rubrene single crystal: direct observation by femtosecond pump- probe spectroscopy. Physical Chemistry Chemical Physics 2013, 14, (23), 8307-8312.

20. Lee, J.; Bruzek, M. J.; Thompson, N. J.; Sfeir, M. Y.; Anthony, J. E.; Baldo, M. A., Singlet exciton fission in a hexacene derivative. Advanced materials 2013, 25, (10), 1445-1448.

212

21. Minami, T.; Ito, S.; Nakano, M., Theoretical study of singlet fission in oligorylenes. Journal of Physical Chemistry Letters 2012, 3, (18), 2719-2723.

22. Musser, A. J.; Al-Hashimi, M.; Maiuri, M.; Brida, D.; Heeney, M.; Cerullo, G.; Friend, R. H.; Clark, J., Activated singlet exciton fission in a semiconducting polymer. Journal of the American Chemical Society 2013, 135, (34), 12747-12754.

23. Johnson, J. C.; Nozik, A. J.; Michl, J., High triplet yield from singlet fission in a thin film of 1,3-diphenylisobenzofuran. Journal of the American Chemical Society 2010, 132, (46), 16302-16303.

24. Dillon, R. J.; Piland, G. B.; Bardeen, C. J., Different rates of singlet fission in monoclinic versus orthorhombic crystal forms of diphenylhexatriene. Journal of the American Chemical Society 2013, 135, 17278-17281.

25. Eaton, S. W.; Shoer, L. E.; Karlen, S. D.; Dyar, S. M.; Margulies, E. A.; Veldkamp, B. S.; Ramanan, C.; Hartzler, D. A.; Savikhin, S.; Marks, T. J.; Wasielewski, M. R., Singlet exciton fission in polycrystalline thin films of a slip-stacked perylenediimide. Journal of the American Chemical Society 2013, 135, (39), 14701- 14712.

26. Wang, C.; Tauber, M. J., High-yield singlet fission in a zeaxanthin aggregate observed by picosecond resonance Raman spectroscopy. Journal of the American Chemical Society 2010, 132, (40), 13988-13991.

27. Wang, C.; Schlamadinger, D. E.; Desai, V.; Tauber, M. J., Triplet excitons of carotenoids formed by singlet fission in a membrane. Chemphyschem 2011, 12, (16), 2891-2894.

28. Wang, C.; Berg, C. J.; Hsu, C. C.; Merrill, B. A.; Tauber, M. J., Characterization of carotenoid aggregates by steady-state optical spectroscopy. Journal of Physical Chemistry B 2012, 116, (35), 10617-10630.

29. Rao, A.; Wilson, M. W. B.; Hodgkiss, J. M.; Albert-Seifried, S.; Bassler, H.; Friend, R. H., Exciton fission and charge generation via triplet excitons in pentacene/C60 bilayers. Journal of American Chemical Society 2010, 132, (36), 12698-703.

30. Jadhav, P. J.; Mohanty, A.; Sussman, J.; Lee, J.; Baldo, M. A., Singlet exciton fission in nanostructured organic solar cells. Nano Letters 2011, 11, (4), 1495-1498.

31. Jadhav, P. J.; Brown, P. R.; Thompson, N.; Wunsch, B.; Mohanty, A.; Yost, S. R.; Hontz, E.; Van Voorhis, T.; Bawendi, M. G.; Bulovic, V.; Baldo, M. A., Triplet exciton dissociation in singlet exciton fission photovoltaics. Advanced Materials 2012, 24, (46), 6169-6174.

213

32. Ehrler, B.; Walker, B. J.; Bohm, M. L.; Wilson, M. W. B.; Vaynzof, Y.; Friend, R. H.; Greenham, N. C., In situ measurement of exciton energy in hybrid singlet-fission solar cells. Nature Communications 2012, 3, 6.

33. Thompson, N. J.; Hontz, E.; Congreve, D. N.; Bahlke, M. E.; Reineke, S.; Van Voorhis, T.; Baldo, M. A., Nanostructured singlet fission photovoltaics subject to triplet- charge annihilation. Advanced Materials 2013.

34. Congreve, D. N.; Lee, J.; Thompson, N. J.; Hontz, E.; Yost, S. R.; Reusswig, P. D.; Bahlke, M. E.; Reineke, S.; Van Voorhis, T.; Baldo, M. A., External quantum efficiency above 100% in a singlet-exciton-fission-based organic photovoltaic cell. Science 2013, 340, (6130), 334-337.

35. Burdett, J. J.; Bardeen, C. J., Quantum beats in crystalline tetracene delayed fluorescence due to triplet pair coherences produced by direct singlet fission. Journal of the American Chemical Society 2012, 134, (20), 8597-8607.

36. Burdett, J. J.; Gosztola, D.; Bardeen, C. J., The dependence of singlet exciton relaxation on excitation density and temperature in polycrystalline tetracene thin films: Kinetic evidence for a dark intermediate state and implications for singlet fission. Journal of Chemical Physics 2011, 135, (21), 10.

37. Chan, W. L.; Ligges, M.; Jailaubekov, A.; Kaake, L.; Miaja-Avila, L.; Zhu, X. Y., Observing the multiexciton state in singlet fission and ensuing ultrafast multielectron transfer. Science 2011, 334, (6062), 1541-1545.

38. Chan, W. L.; Ligges, M.; Zhu, X. Y., The energy barrier in singlet fission can be overcome through coherent coupling and entropic gain. Nature Chemistry 2012, 4, (10), 840-845.

39. Zimmerman, P. M.; Bell, F.; Casanova, D.; Head-Gordon, M., Mechanism for singlet fission in pentacene and tetracene: from single exciton to two triplets. Journal of the American Chemical Society 2011, 133, (49), 19944-19952.

40. Zimmerman, P. M.; Musgrave, C. B.; Head-Gordon, M., A correlated electron view of singlet fission. Accounts of chemical research 2013, 46, (6), 1339-1347.

41. Greyson, E. C.; Vura-Weis, J.; Michl, J.; Ratner, M. A., Maximizing singlet fission in organic dimers theoretical investigation of triplet yield in the regime of localized excitation and fast coherent electron transfer. Journal of Physical Chemistry B 2010, 114, (45), 14168-14177.

42. Johnson, J. C.; Nozik, A. J.; Michl, J., The role of chromophore coupling in singlet fission. Accounts of Chemical Research 2013, 46, (6), 1290-1299.

214

43. Kleinschmidt, M.; Marian, C. M.; Waletzke, M.; Grimme, S., Parallel multireference configuration interaction calculations on mini-β-carotenes and β-carotene. Journal of Chemical Physics 2009, 130, (4).

44. Schulten, K.; Karplus, M., Origin of a low-lying forbidden transition in polyenes and related molecules. Chemical Physics Letters 1972, 14, (3), 305-309.

45. Hudson, B.; Kohler, B., Linear polyene electronic structure and spectroscopy. In Annual review of physical chemistry, Eyring, H., Ed. Annual Reviews: 1974; Vol. 25, pp 437-460.

46. Schulten, K.; Ohmine, I.; Karplus, M., Correlation effects in the spectra of polyenes. The Journal of Chemical Physics 1976, 64, (11), 4422-4441.

47. Tavan, P.; Schulten, K., Electronic excitations in finite and infinite polyenes. Physical Review B 1987, 36, (8), 4337-4358.

48. Greyson, E. C.; Stepp, B. R.; Chen, X.; Schwerin, A. F.; Paci, I.; Smith, M. B.; Akdag, A.; Johnson, J. C.; Nozik, A. J.; Michl, J.; Ratner, M. A., Singlet exciton fission for solar cell applications: energy aspects of interchromophore coupling. Journal of physical Chemistry B 2009, 114, (45), 14223-14232.

49. Renaud, N.; Sherratt, P. A.; Ratner, M. A., Mapping the relation between stacking geometries and singlet fission yield in a class of organic crystals. Journal of Physical Chemistry Letters 2013, 4, 1065-1069.

50. Vallett, P. J.; Snyder, J. L.; Damrauer, N. H., Tunable electronic coupling and driving force in structurally well-defined tetracene dimers for molecular singlet fission: A computational exploration using density functional theory. Journal of Physical Chemistry A 2013, 117, (42), 10824-10838.

51. Berkelbach, T. C.; Hybertsen, M. S.; Reichman, D. R., Microscopic theory of singlet exciton fission. II. Application to pentacene dimers and the role of superexchange. Journal of Chemical Physics 2013, 138, (11), 114103-1-114103-12.

52. Spano, F. C., Analysis of the UV/Vis and CD spectral line shapes of carotenoid assemblies: spectral signatures of chiral H-aggregates. Journal of the American Chemical Society 2009, 131, (12), 4267-4278.

53. Wang, C.; Angelella, M.; Kuo, C. H.; Tauber, M. J. In Singlet fission in carotenoid aggregates: insights from transient absorption spectroscopy, SPIE NanoScience+ Engineering, 2012; International Society for Optics and Photonics: 2012; pp 845905-845905-13.

215

54. Kuhlman, T. S.; Kongsted, J.; Mikkelsen, K. V.; Moller, K. B.; Solling, T. I., Interpretation of the ultrafast photoinduced processes in pentacene thin films. Journal of the American Chemical Society 2010, 132, (10), 3431-3439.

55. Zsila, F.; Bikadi, Z.; Keresztes, Z.; Deli, J.; Simonyi, M., Investigation of the self- organization of lutein and lutein diacetate by electronic absorption, circular dichroism spectroscopy, and atomic force microscopy. Journal of Physical Chemistry B 2001, 105, (39), 9413-9421.

56. Billsten, H. H.; Sundstrom, V.; Polivka, T., Self-assembled aggregates of the carotenoid zeaxanthin: time-resolved study of excited states. Journal of Physical Chemistry A 2005, 109, (8), 1521-1529.

57. Dallinger, R. F.; Guanci, J. J.; Woodruff, W. H.; Rodgers, M. A. J., Vibrational spectroscopy of the electronically excited-state - pulse radiolysis-time-resolved resonance Raman study of triplet β-carotene. Journal of the American Chemical Society 1979, 101, (5), 1355-1357.

58. Jensen, N. H.; Wilbrandt, R.; Pagsberg, P. B.; Sillesen, A. H.; Hansen, K. B., Time-resolved resonance Raman spectroscopy - The excited triplet state of all-trans-β- carotene. Journal of the American Chemical Society 1980, 102, (25), 7441-7444.

59. Dallinger, R. F.; Farquharson, S.; Woodruff, W. H.; Rodgers, M. A. J., Vibrational Spectroscopy of the Electronically Excited-State .4. Nanosecond and Picosecond Time-Resolved Resonance Raman-Spectroscopy of Carotenoid Excited- States. Journal of the American Chemical Society 1981, 103, (25), 7433-7440.

60. Hashimoto, H.; Koyama, Y., Time-resolved resonance Raman spectroscopy of triplet β-carotene produced from all-trans, 7-cis, 9-cis, 13-cis, and 15-cis isomers and high-pressure liquid-chromatography analyses of photoisomerization via the triplet state. Journal of Physical Chemistry 1988, 92, (8), 2101-2108.

61. Hashimoto, H.; Koyama, Y., Raman spectra of all-trans-β-carotene in the S1 state and T1 state produced by direct photoexcitation. Chemical Physics Letters 1989, 163, (2- 3), 251-256.

62. Noguchi, T.; Hayashi, H.; Tasumi, M.; Atkinson, G. H., Solvent Effects on the Ag 1 - C=C Stretching Mode in the 2 Ag Excited-State of β-Carotene and 2 Derivatives - Picosecond Time-Resolved Resonance Raman-Spectroscopy. Journal of Physical Chemistry 1991, 95, (8), 3167-3172.

63. McCamant, D. W.; Kim, J. E.; Mathies, R. A., Vibrational relaxation in β- carotene probed by picosecond Stokes and anti-Stokes resonance Raman spectroscopy. Journal of Physical Chemistry A 2002, 106, (25), 6030-6038.

216

64. Knoester, J.; Spano, F. C., J-aggregates. In Kobayashi, T., Ed. World Scientific Publishing Co.: Singapore: 1996; pp 111-160.

65. Nielsen, B. R.; Jorgensen, K.; Skibsted, L. H., Triplet-triplet extinction coefficients, rate constants of triplet decay and rate constants of anthracene triplet sensitization by laser flash photolysis of astaxanthin, β-carotene, canthaxanthin and zeaxanthin in deaerated toluene at 298 K. Journal of Photochemistry and Photobiology a- Chemistry 1998, 112, (2-3), 127-133.

66. Okamoto, H.; Nakabayashi, T.; Tasumi, M., Analysis of anti-Stokes resonance Raman excitation profiles as a method for studying vibrationally excited molecules. Journal of Physical Chemistry A 1997, 101, (19), 3488-3493.

67. Simpson, W. T.; Peterson, D. L., Coupling strength for resonance force transfer of electronic energy in van der Waals solids. The Journal of Chemical Physics 1957, 26, (3), 588-593.

68. Kasha, M., Energy transfer mechanisms and the molecular exciton model for molecular aggregates. Radiation Research 1963, 20, (1), 55-70.

69. Spano, F. C., The spectral signatures of frenkel polarons in H- and J-aggregates. Accounts of Chemical Research 2010, 43, (3), 429-439.

70. Berg, C. Master thesis. University of California, San Diego, La Jolla, 2013.

71. Cantor, C. R.; Schimmel, P. R., Biophysical Chemistry Part II: Techniques for the Study of Biological Structure and Function. W. H. Freeman: San Francisco, 1980.

72. Parson, W. W., Modern Optical Spectroscopy. Springer: Dordecht, The Netherlands, 2007.

73. Albrecht, A. C.; Hutley, M. C., On the Dependence of Vibrational Raman Intensity on the Wavelength of Incident Light. Journal of Chemical Physics 1971, 55, (9), 4438-4443.

74. Myers, A. B.; Mathies, R. A., Resonance Raman intensities: A probe of excited- state structure and dynamics. In Biological Applications of Raman Spectroscopy, John Wiley & Sons: 1987; Vol. 2, pp 1-58.

75. Leng, W. N.; Würthner, F.; Kelley, A. M., Resonance Raman intensity analysis of merocyanine dimers in solution. Journal of Physical Chemistry B 2004, 108, (29), 10284- 10294.

76. Shreve, A. P.; Trautman, J. K.; Owens, T. G.; Albrecht, A. C., Determination of the S2 lifetime of β-carotene. Chemical Physics Letters 1991, 178, (1), 89-96.

217

78. Macpherson, A. N.; Gillbro, T., Solvent dependence of the ultrafast S2-S1 internal conversion rate of β-carotene. Journal of Physical Chemistry A 1998, 102, (26), 5049- 5058.

80. Billsten, H. H.; Pan, J. X.; Sinha, S.; Pascher, T.; Sundstrom, V.; Polivka, T., Excited-state processes in the carotenoid zeaxanthin after excess energy excitation. Journal of Physical Chemistry A 2005, 109, (31), 6852-6859.

81. Cong, H.; Niedzwiedzki, D. M.; Gibson, G. N.; Frank, H. A., Ultrafast time- resolved spectroscopy of xanthophylls at low temperature. Journal of Physical Chemistry B 2008, 112, (11), 3558-3567.

82. Polivka, T.; Sundstrom, V., Dark excited states of carotenoids: Consensus and controversy. Chemical Physics Letters 2009, 477, (1-3), 1-11.

83. Jailaubekov, A. E.; Vengris, M.; Song, S. H.; Kusumoto, T.; Hashimoto, H.; Larsen, D. S., Deconstructing the excited-state dynamics of β-carotene in solution. Journal of Physical Chemistry A 2011, 115, (16), 3905-3916.

84. Ostroumov, E. E.; Reus, M. G. M. M.; Holzwarth, A. R., On the nature of the "dark S*" excited state of β-carotene. The Journal of Physical Chemistry A 115, (16), 3698-3712.

85. Kukura, P.; McCamant, D. W.; Mathies, R. A., Femtosecond time-resolved 1 + stimulated Raman spectroscopy of the S2 ( Bu ) excited state of β-carotene. Journal of Physical Chemistry A 2004, 108, (28), 5921-5925.

86. Rondonuwu, F. S.; Watanabe, Y.; Zhang, J. P.; Furuichi, K.; Koyama, Y., + - - Internal-conversion and radiative-transition processes among the 1Bu , 1Bu and 2Ag states of all-trans-neurosporene as revealed by subpicosecond time-resolved Raman spectroscopy. Chemical Physics Letters 2002, 357, (5-6), 376-384.

218

88. Sterling, C., Crystal structure of β-carotene. Acta Crystallographica 1964, 17, (10), 1224-1228.

89. Lin, H. C.; Jin, B. Y., Interchain Interactions in Organic Conjugated Dimers: A Composite-Molecule Approach. Journal of Physical Chemistry A 2010, 114, (8), 2885- 2892.

90. Katoh, R.; Sinha, S.; Murata, S.; Tachiya, M., Origin of the stabilization energy of perylene excimer as studied by fluorescence and near-IR transient absorption spectroscopy. Journal of Photochemistry and Photobiology a-Chemistry 2001, 145, (1-2), 23-34.

91. Furube, A.; Murai, M.; Tamaki, Y.; Watanabe, S.; Katoh, R., Effect of aggregation on the excited-state electronic structure of perylene studied by transient absorption spectroscopy. Journal of Physical Chemistry A 2006, 110, (20), 6465-6471.

92. Land, E. J.; Sykes, A.; Truscott, P. G., In-vitro photochemistry of biological molecules .2. Triplet states of β-carotene and lycopene excited by pulse radiolysis. Photochemistry and Photobiology 1971, 13, (4), 311-320.

93. Gijzeman, O. L.; Sykes, A., Relationship between singlet and triplet absorption spectra of conjugated polyenes. Photochemistry and Photobiology 1973, 18, (4), 339-341.

94. Birks, J. B., Quintet state of pyrene excimer. Physics Letters A 1967, A 24, (9), 479-480.

95. Saltiel, J.; Marchand, G. R.; Smothers, W. K.; Stout, S. A.; Charlton, J. L., Concerning the spin-statistical factor in the triplet-triplet annihilation of anthracene triplets. Journal of the American Chemical Society 1981, 103, (24), 7159-7164.

96. Albrecht, W. G.; Michel-Beyerle, M. E.; Yakhot, V., Exciton fission in excimer forming crystal. Dynamics of an excimer build-up in α-perylene. Chemical Physics 1978, 35, (1-2), 193-200.

97. Furukawa, Y., Electronic absorption and vibrational spectroscopies of conjugated conducting polymers. Journal of Physical Chemistry 1996, 100, (39), 15644-15653.

98. Tiago, M. L. N. J. E. L., S.G, Ab initio calculation of the electronic and optical properties of solid pentacene. Physical Chemistry Chemical Physics 2003, 67 115212

99. Cudazzo, P.; Gatti, M.; Rubio, A., Excitons in molecular crystals from first- principles many-body perturbation theory: Picene versus pentacene. Physical Review B 86, (19), 195307.

219

100. Weiss, V.; Port, H.; Wolf, H. C., Excitonic and molecular properties of the triplet T-1-state in diphenylpolyene single crystals. Molecular Crystals and Liquid Crystals A 1997, 308, 147-178.

101. Quarti, C.; Fazzi, D.; Del Zoppo, M., A computational investigation on singlet and triplet exciton couplings in acene molecular crystals. Physical Chemistry Chemical Physics 2011, 13, (41), 18615-18625.

102. Muller, A. M.; Avlasevich, Y. S.; Mullen, K.; Bardeen, C. J., Evidence for exciton fission and fusion in a covalently linked tetracene dimer. Chemical Physics Letters 2006, 421, (4-6), 518-522.

103. Muller, A. M.; Avlasevich, Y. S.; Schoeller, W. W.; Mullen, K.; Bardeen, C. J., Exciton fission and fusion in bis(tetracene) molecules with different covalent linker structures. Journal of the American Chemical Society 2007, 129, (46), 14240-14250.

104. Grumstrup, E. M.; Johnson, J. C.; Damrauer, N. H., Enhanced Triplet Formation in Polycrystalline Tetracene Films by Femtosecond Optical-Pulse Shaping. Physical Review Letters 2010, 105, (25), 2574031-4.

105. Casanova, D., Electronic structure study of singlet-fission in tetracene derivatives. Journal of Chemical Theory and Computation 2013, 10, (1), 324-334.

106. Negri, F.; Orlandi, G.; Zerbetto, F.; Zgierski, M. Z., Theoretical-study of the force-field of the lowest singlet electronic states of long polyenes. Journal of Chemical Physics 1989, 91, (10), 6215-6224.

107. Wasielewski, M. R.; Johnson, D. G.; Bradford, E. G.; Kispert, L. D., Temperature-dependence of the lowest excited singlet state lifetime of all-trans-β- carotene and fully deuterated all-trans-β-carotene. Journal of Chemical Physics 1989, 91, (11), 6691-6697.

108. Nagae, H.; Kuki, M.; Zhang, J. P.; Sashima, T.; Mukai, Y.; Koyama, Y., Vibronic - coupling through the in-phase, C=C stretching mode plays a major role in the 2Ag to - 1Ag internal conversion of all-trans-β-carotene. Journal of Physical Chemistry A 2000, 104, (18), 4155-4166.

109. Lopez-Ramirez, M. R.; Sanchez-Cortes, S.; Perez-Mendez, M.; Blanch, G., Trans-cis isomerisation of the carotenoid lycopene upon complexation with cholesteric polyester carriers investigated by Raman spectroscopy and density functional theory. Journal of Raman Spectroscopy 2010, 41, (10), 1170-1177.

110. Lin, S. W.; Groesbeek, M.; van der Hoef, I.; Verdegem, P.; Lugtenburg, J.; Mathies, R. A., Vibrational assignment of torsional normal modes of rhodopsin: Probing

220 excited-state isomerization dynamics along the reactive C-11=C-12 torsion coordinate. Journal of Physical Chemistry B 1998, 102, (15), 2787-2806.

111. Takahashi, O.; Watanabe, M.; Kikuchi, O., The 'triplet-excited region' in linear polyenes: Real or artifactual? Journal of Molecular Structure-Theochem 1999, 469, 121- 125.

112. Ma, H. B.; Liu, C. G.; Jiang, Y. S., "Triplet-excited region" in polyene oligomers revisited: Pariser-Parr-Pople model studied with the density matrix renormalization group method. Journal of Chemical Physics 2004, 120, (19), 9316-9320.

221

Appendix of Chapter 6

Determine the probe energy for AS-592

Figure A6.1. Stokes and anti-Stokes of spectra of zeaxanthin WH-aggregate probed with (A) 557 nm, 4 ps pulse laser and (B) 568 nm CW laser. Anti-Stokes spectra is normalized against the peak maxima of the Stokes and magnified by a factor of 25. In both cases, the ratio between integrated intensities of AS and S 1520 cm-1 bands are about 0.014.

222

TRRR Data analysis procedure:

The data workup procedures are listed as bellow. Experiments with 554 nm and

413 nm probe for the WH-aggregate and 413 nm probe for the SH-aggregate are chosen as examples to illustrate the procedure as shown in Figure A6.2:

(1) The ground state spectra were obtained by removing the solvent signals from

“probe only” spectra.

(2) The pump background was removed with subtraction (pump+probe)-k*(pump only). The scaling factor k is optimized to yield the flattest baseline background that has close curvature as the (probe only) spectra. Results from step (2) were shown in the upper panels of Figure A6.2.

(3) Raman bands from the solvent and ground state were removed by subtracting a scaled (probe only) spectrum from the result in step (2). The scaling factor is optimized to eliminate the strongest solvent bands. The resulted spectra are shown in the second panels of Figure A6.2.

(4) A small portion of ground state signals (obtained from step (1)) were added back to different spectra from step (3) to cancel the ground state bleach and result the

“pure” excited state spectra. The smallest fraction yielding a reconstructed spectrum that was (a) fully above baseline, (b) free of a negative dip in the ethylenic region and (c) representative to a reasonable feature in C=C and C-C stretching regions, was considered optimal. This process is also shown in the second panels of Figure A6.2.

(5) The final spectrum was obtained by removing a baseline, generated by a cubic

Bezier spline with 25 sampling points, from the spectra obtained in step (3).

223

Figure A6.2A. Data workup for WH-554 experiment. The straight line notated the spectral features which were monitored to determine the best addition.

224

Figure A6.2B. Data workup for WH-497 experiment. The straight line notated the spectral feature which was monitored to determine the best addition.

In principle, the loss of ground state because of pump excitation can be determined from step (4). However, at several wavelengths the observed excitation ratios were much smaller than expected values because the induced exciton coupling between

T1 → Tn and S0 → S2 transitions may increase the Raman cross-section of ground state.

The WH-497 is an example where the observed ground state loss is significantly smaller than that observed in WH-554.

225

Figure A6.2C. Data workup for SH-413 experiment.

SH-413 is an extreme case where no prominent negative feature was observed, and no addition of ground state signals was performed.

226

Figure A6.2D. Data workup for WH-471 experiment. The straight line notated the spectral features which were monitored to determine the best addition.

WH-471 showed different addition in different spectral regions. To make up the negative at the finger print region, 3.8% of ground state signal were needed. However, the negative signal at the ethylenic region was still prominent. To mostly remove the negative feature in the whole spectral window, the addition needed to be 8.5%. A small negative feature was still observed for the ethylenic with 8.5% addition, but further addition leads to a sharp peak at the ethylenic band, and the ground state like feature in

227 the finger print region became unrealistically high. To balance the spectral profile, we stopped at 8.5% addition.

Figure A6.3. TRRR spectra of zeaxanthin (A) WH and (B) SH aggregate with different probe wavelengths at 150 ps time delay. (C) Transient microscopic spectra of the T1 state of zeaxanthin monomer in acetone formed through sensitization of anthracene. All Raman spectra were normalized against their peak maxima.

228

Figure A6.4. Kinetics for selected bands of the anti-Stokes TRRR spectra of the WH- aggregate. Multiexponential fits are shown, and two decay times are reported.

229

Figure A6.5. Stabilization of the S2 state in zeaxanthin aggregate due to exciton coupling and formation of excimer states.

The transition energy values are referred from literature and our previous

1-3 publication. Assuming the excited state transition S2→SN remains as the intrinsic value,

9700 cm-1 after forming the excimer, for SH-aggregate, the ESA of the X state is at 410

-1 -1 nm, which blueshifted for 14700 cm from the original S2→SN. Given a ~5000 cm stabilization gained from exciton coupling effect, the formation of the excimer provides the rest stabilization energy, which is up to 9700 cm-1. For WH-aggregate, similar analysis yields 9200 cm-1 stabilization energy.

Table A6.1. Laser wavelengths of the pump and probe beams and the corresponding filters for TRRR experiments. Fundamental SHG Pump /nm Probe /nm Samples/Experiments Filters (Seed) /nm /nm 1 2 830 (835) 413 471 (1 S D2) 413 WH and SH/Stokes Asahi 422 LP 2 413 413 471 (1 S D2) WH/Stokes Semrock 488 E 2 413 413 497 (1 S H2) WH/Stokes Iridian 496 N 416 416 554 (2 S D2) WH and SH/Stokes Semrock 568 E 418 418 557 (2 S D2) WH/As-Stokes Semrock 561 N 800 (800) 401 401 527 (2 S D2) WH/Stokes Semrock 532 E 400 400 599 (2 S H2) WH/Stokes Semrock 594 N 397 397 592 (2 S H2) WH/AS-Stokes Semrock 594 N 1 Sources of the laser beams: e.g. 1 S D2 represent the 1st Stokes-shifted stimulated Raman of D2 gas. 2 Types of filters: LP: longpass; E: edge; N: notch.

230

Table A6.2. Pump and probe parameters for TRRR experiments. Pump

Wave- εpump Pulse Beam Pump Pump Pump Peak a a b length GS Energy Size Intensity nm M-1cm-1 nJ μm2 GS% ES% GW/cm2 413 59,000 812 900×220 13.8 29.1 0.051 413 59,000 660 950×240 14.0 29.2 0.052 416 64,000 340 510×160 15.3 29.7 0.053 WH 471 113,000 210 780×170 14.4 18.3 0.029 401 43,000 650 820×210 14.9 39.8 0.074 400 43,000 340 490×140 18.2 48.5 0.21 416 48,000 280 510×160 13.8 34.6 0.063 SH 471 16,000 1160 780×170 15.3 101 0.16 WH 418 68,000 250 800×150 15.3 27.6 0.027 AS 397 40,000 370 550×180 11.5 28.8 0.28 Probe

Wave- εprobe Pulse Beam Probe Probe Probe Peak a a b b length GS Energy Size Intensity nm M-1cm-1 nJ μm2 GS% ES% GW/cm2 471 113,000 43 820×190 1.7 2.0 0.0034 WH 497 60,000 110 760×190 3.2 6.5 0.0094 554 4,700 130 500×140 0.8 17.2 0.023 413 63,000 40 700×150 1.4 2.7 0.0051 527 41,000 49 810×160 1.1 3.0 0.0051 599 N/A 400 280×60 N/A 240 0.702 554 2,000 170 500×140 0.7 23.4 0.032 SH 413 48,000 40 700×150 1.1 2.7 0.0051 WH 557 3,800 100 780×120 0.3 10.3 0.014 AS 592 N/A 250 500×140 N/A 37.6 0.094 a Beam size and pulse energy are measured at the sample point. b Peak intensity of pump or probe, estimated from 0.7×(pulse energy), divided by (area of ellipse for 10%-90% measurements) and pulse duration (3 ps for the 599 nm probe of WH-aggregate, and 7ps for the rest of experiments).

231

The fraction of zeaxanthin molecules excited (fexc) is first estimated by comparing the number of photons absorbed by the sample (Nabs) versus the number of zeaxanthin molecules in the illuminated region (Ntot). The ratio can be determined from the laser beam area and pulse energy, as well as the optical density and concentration of the sample. Assuming Beer’s law holds when low pump pulse energies are employed, fexc is given by Eq. A6.1.

Eq. A6.1

is the cross-sectional area of the beam, which is an ellipse with axis lengths

defined by the positions for 10% and 90% attenuation. is the number of pump photons that impinge on the area , and is therefore estimated to be 70% of total number of photons per pulse assuming the electromagnetic field intensity of the pulses following a Gaussian distribution. OD is the optical density of the sample at the excitation wavelength; l(cm) is the pathlength through the sample (0.2 cm); C(mol/L) is the monomer-equivalent concentration of zeaxanthin (10 µM for the 599 nm probe WH

-1 aggregate experiment, and 15 µM for all the other experiments); and NA(mol ) is

Avogadro’s number.

Estimated fractional excitation of an excited state (ES) formed by either pump or probe pulses. The concentration of excited state species is unknown, thus fexc is estimated from a Taylor expansion of the above equation in the limit of small OD which leads to cancellation of concentration terms, and an expression that depends on the decadic molar extinction coefficient (cm-1 M-1) of the excited state species or alternatively the

2 -1 absorption cross-section (cm molecule ) :

232

The expansion is certainly valid in the limit of small (less than 0.1) transient OD which is caused by action of the pump or probe. The extinction coefficient of the excited state species is arbitrarily assumed to be 100,000 M-1cm-1 at all wavelengths. The extinction coefficient, and thus excitation percentage of the excited state species, is likely to be much smaller at some wavelengths, e.g. 599 nm.

References

1. Polivka, T.; Herek, J. L.; Zigmantas, D.; Akerlund, H. E.; Sundstrom, V., Direct observation of the (forbidden) S1 state in carotenoids. Proceedings of the National Academy of Sciences of the United States of America 1999, 96, (9), 4914-4917.

2. Christensen, R. L., The electronics states of carotenoids. In The Photochemistry of the Carotenoids, Frank, H. A.; Young, A. J.; Britton, G.; Cogdell, R. J., Eds. Kluwer Academic Publishers: Dordecht, 1999; Vol. 8, pp 137-157.

3. Wang, C.; Angelella, M.; Kuo, C. H.; Tauber, M. J. In Singlet fission in carotenoid aggregates: insights from transient absorption spectroscopy, SPIE NanoScience+ Engineering, 2012; International Society for Optics and Photonics: 2012; pp 845905-845905-13.

Chapter 7 Resonance Raman Spectroscopy of the Triplet Excited State of Oligothiophenes

The triplet excited states of polythiophenes and oligothiophenes are important for many emerging applications in organic devices. For organic light-emitting diodes

(OLEDs), triplet excited states (or triplets) have long been the focus of both basic and applied research partly because recombination to the T1 state can in principle lead to 75% losses in emission.1-4 For organic photovoltaics (OPVs), the role of triplets has only recently gained attention. For example, triplets appear to be closely linked to detrimental charge recombination within bulk heterojunction OPVs that include a polythiophene donor material.5-7 On the other hand, triplet excited states could have significant benefits for advanced OPVs. Their relatively long lifetimes and transport distances have been highlighted as advantages for OPVs.3, 8, 9 Triplets are also the products of singlet fission, and photovoltaics incorporating this mechanism would be highly desirable.10-12

Fundamental insights into polythiophene and oligothiophenes excited states have frequently relied upon the measurement of properties as a function of oligomer length.13-

21 One of the most important characteristics is the energies of singlet and triplet excited states, which are well known to depend upon chain length. The relative energy levels of

18, S1, T1, and T2 excited states have evolved with refinements in experiment and theory.

20-25 The S1 state is now considered to have higher energy than the T2 state, and this ordering is consistent with efficient ISC in short oligothiophenes.20, 24, 25 The extension of triplet excitation along the conjugated chain has also been a matter of debate. Again, length-dependent studies have played a key role.15, 19, 26 Early theoretical work and zero- field splittings from magnetic resonance spectroscopy were initially interpreted in terms

233

234

15, 27-29 of a highly localized T1 excitation that extended over one or two thiophene rings.

However, a reinterpretation of the zero-field splittings and later theoretical studies19, 26 suggested that the triplet excitation extends to at least four thiophene units when the chain length is sufficiently long. The size of the T1 excitation is an important fundamental insight, yet it remains uncertain at present.

Vibrational spectroscopy can provide profound insights into electronic structures of excited states because normal modes are sensitive to bonding patterns and structure.

Although vibrational spectroscopy has been widely employed for characterization of the ground state neutral and cationic thiophene oligomers and polymers,30-41 few vibrational spectra have been reported for their excited states.39, 42-46 Here, we report well-resolved resonance Raman spectra of two short-chain oligothiophenes in their lowest triplet (T1) excited state. The spectra are very different from those of the ground state, thus resonance

Raman spectroscopy emerges as an excellent probe of these excited-state species.

Furthermore, an analysis of the bands provides insights into the triplet excited-state structure that can be extrapolated to the polymer. We also acquired Raman spectra of a triplet oligothiophene in aggregated state, which represents an important step towards probing the T1 excited states in solid samples (e.g. thin film or bulk heterojunction).

Molecular structures of the oligothiophenes are shown in Scheme 7.1. The tetramer with dodecyl substituents (denoted (Dodec)2-4T, or 1) was obtained from

Sigma-Aldrich and used as received. The hexathiophene with octyl substituents ((Oct)4-

47 6T, or 2) was synthesized as reported previously. The aggregates of (Oct)4-6T were prepared by adding a THF solution of 2.6 mM (Oct)4-6T dropwise to swirling water until 235

the THF:water volume ratio was 5:95. The 100 μM solution of (Oct)4-6T in aggregated form was diluted with additional 5:95 THF/water as needed for each spectroscopic measurement. Solutions for Raman spectroscopy were prepared in an argon-filled glovebox.

Scheme 7.1. Molecular structures of 3,3′′′-didodecyl-2,2′:5′,2′′:5′′,2′′′-tetrathiophene

((Dodec)2-4T, or 1) and tetraoctyl hexathiophene ((Oct)4-6T, or 2)

Resonance Raman spectra of transient species were collected with a modified optical microscope, with two beams impinging upon a fast-flowing solution in a capillary.

Some elements of our system were similar to a previous design.48 A quasi continuous- wave (cw) pump beam with wavelength 420 nm was used to photoexcite the sample. A cw probe with wavelength 647.09 nm was selected for resonance with the electronic transitions of the triplet state. Sample solutions were pressure-driven through a quartz capillary that intersected the pump and probe beams. By controlling the flow speed and the micrometer-length separation between pump and probe foci at the capillary, the time delay could be varied between 1 and 20 µs. The high (>50%) yields of intersystem

14 crossing and >15 µs triplet lifetimes of tetra- and hexathiophenes led to substantial T1 236

population at a 5 µs delay time for solution phase experiment. T1 spectra of the aggregate were acquired at a delay time of 1 µs to compensate for a faster rate of decay. Back- scattered light from the sample was collected and collimated with the same near-UV apochromatic objective lens that focused the excitation beams. A notch filter was used to filter the Rayleigh scattering, and the Raman signal was dispersed with a single-grating spectrograph and detected with a CCD. A detailed description can be found in the

Appendix.

Figure 7.1. Experimental Raman spectra of 1 in chlorobenzene (black traces, left), and

2 in THF (black traces, right). The spectra of the T1 excited state (upper) and S0 ground state (lower) molecules were acquired with 647.09 nm probe, which was near- resonant with the T1 absorption and off-resonant with the S0 absorption. Calculated spectra of model (ethyl)2-4T and (ethyl)4-6T molecules are shown as red dashed traces. The DFT calculations utilized the B3LYP functional and 6-31G(d) basis set. The spin was unrestricted for the triplet state calculations. The normal mode frequencies of both the singlet and triplet state calculations were scaled by 0.9614.

237

The experimental Raman spectra of 1 and 2 in the lowest triplet excited state (31*

3 and 2*) are shown in Figure 7.1, along with corresponding S0 spectra. The triplet spectra are distinguished from the ground state by a much broader distribution of bands in the

1300 – 1550 cm-1 region. Also, prominent bands in the triplet spectra are apparent in the

1050 – 1200 cm-1 and 550 – 800 cm-1 regions, where the ground state has only weak bands. It is significant that the triplet spectra of the tetramer and hexamer are clearly different from one another in all spectral regions, whereas the ground-state spectra are nearly identical. The principle differences between 31* and 32* are twofold. First, the most prominent high-frequency band is downshifted 50 cm-1 for the hexamer relative to the tetramer. Second, the hexamer has more Raman bands in the low frequency region

(550-740 cm-1). The origin of these distinctions between tetramer and hexamer is explored below. A minor difference is the greater width of vibrational bands of the hexamer, which may be associated with a greater number of conformers for the longer oligothiophene. However, the exploration of this point is beyond the scope of the present work.

Band and normal mode assignments for the triplet excited state have not been previously reported for oligothiophenes. While spectra of the ground state computed by

DFT correspond well (after scaling) to experimental ground state spectra, the triplet excited state spectra are not a sufficient match for reliable assignment of individual bands. Nevertheless, similar patterns for experimental and DFT-calculated spectra suggest that bands in various spectral regions can be categorized in terms of their dominant local modes. The categories are listed in Table 7.1, and representative normal modes are shown in Figure 7.2. Normal modes with frequencies 1550 – 1100 cm-1 consist 238 of CC stretches coupled with C-H in-plane (IP) bends. Within that spectral range, the specific stretches involved are C=C double bonds (1550 – 1400 cm-1); C-C single bonds

(1100 – 1250 cm-1); and CC bonds with mixed bond orders (1300 – 1400 cm-1). Bands in the 1000 – 1100 cm-1 region are associated with normal modes that are predominantly C-

H IP bends. Strong vibrational bands in the 550-800 cm-1 window are assigned either to ring deformation/breathing in combination with C-S stretches, or skeletal out-of-plane

(OOP) bends.

Table 7.1. Categories of Raman modes calculated by DFT for model tetrathiophene and hexathiophene molecules in their lowest triplet (T1) excited state.

-1 a -1 a Description (Ethyl)2-4T (cm ) (Ethyl)4-6T (cm ) C=C str. and C-H IP bends 1526(s), 1490(w), 1471(m) 1509(m), 1500(s), 1487(s)

1452(s), 1418(m), 1411(m), CC str. with varied bond 1387(m), 1367(s), 1337(s) 1357(m), 1335(s), 1328(m), order, and C-H IP bends 1190(s)

C-C str. and C-H IP bends 1182 (m), 1140(w) 1185(s), 1150(w) C-H IP bends 1207(m) 1079(w), 1068(w) 1072(w), 1047(w) 711(w), 670(w), 649(m), C-S str., ring def. 821(w), 667(w), 664(w), 622(m) 605(w), 589(w), 569(w), CC and C-H OOP bends 631(m), 621(m) 557(w) a The frequencies are scaled relative to the calculation output by a factor of 0.9614. Modes are selected based upon their (1) proximity to experimental bands, and (2) likely resonance Raman enhancement via symmetric motion of the conjugated core. The descriptions (s), (m), (w) refer to strong, medium, and weak calculated band intensities. 239

Figure 7.2. Vibrational normal modes of the T1 excited state of model molecule (ethyl)2- 4T, from DFT computations. Normal modes of (ethyl)4-6T that are analogous to those of (ethyl)2-4T are shown in the Appendix; two modes that are distinct are shown above.

The relatively complex spectra of 31* and 32* compared with those of 1 and 2 suggest that the structures of the T1 excited state are also more complicated. The ground state molecules have few prominent modes; in general, changes in oligomer length cause relatively small changes in their frequencies. The reason is that the thiophene units in the ground state are similar to one another, and the modes reflect regular patterns of movement.49, 50However, the broad frequency distribution of the CC stretching modes in the triplet state reveals that the CC bonds do not have a simple pattern along the 240 thiophene backbone. The loss of a simple bond alternation is a consequence of the changed electronic structure upon excitation, which alters the force constants of various bonds and the composition of normal modes. For many linked aromatic systems including poly- or oligothiophenes, both experimental14, 51 and theoretical15, 52-54 studies indicate that low-lying electronic excitations have bond orders that are switched relative to those of the ground state. The Cα=Cβ double bonds in the ground state switch to Cα=Cα and Cβ=Cβ to form the quinoidal structure. Our vibrational spectra of the T1 state share similar spectral features with oligothiophenes that have been previously described as having a quinoidal ground state structure.55-57 The similarities include numerous bands in the region 1300~1500 cm-1, and signals ~1200 cm-1 that are significantly enhanced compared to ground-state oligothiophenes that are not quinoidal.

The spectral differences between the tetramer and hexamer exclude two possible triplet structures. First, the triplet state cannot be a fully extended quinoidal structure. If it were extended, the regularity would cause little differences between tetramer and hexamer spectra. Second, the view that the triplet excited state is confined to one or two thiophene rings15 is also inconsistent with the significant changes in the hexamer relative to the tetramer.

The Raman spectra of the tetramer and hexamer indicate that the fully developed

T1 excited state region of oligothiophenes extends beyond four thiophene rings. DFT calculations suggest that the quinoidal region is confined to the two central rings, as depicted in Figure 7.3. The unpaired electrons that define the triplet excited state are centered primarily on carbon atoms at the fringe of the quinoidal region. The outer part of the chain consists of thiophene rings that are aromatic, similar to the ground-state. A 241 limited quinoidal region is consistent with an energy penalty caused by loss of aromaticity. Previous studies based on extended Hubbard26 and DFT19 computations provided a similar picture as our calculations.

Figure 7.3. Lewis structures of the T1 state of (A) 1 and (B) 2 from DFT calculation, and (C) a possible structure of a dication of 2. The dashed lines indicate regions with bond orders intermediate between single and double. The spin density centered at each carbon is indicated as a number, and the radical symbols (black dots) are located at the maximum

value. Electronic configurations of the dication and triplet state are shown on the right.

The structures depicted in Figure 7.3 A and B help to explain some aspects of the triplet resonance Raman spectra. A triplet excited state of a thiophene chain has several regions that span a range of CC bond orders, from double to single. The spectral changes from tetramer to hexamer can be attributed to the addition of two aromatic thiophene rings which increases the structural complexity.

We note a striking resemblance between resonance Raman spectra of the triplet and dication states of oligothiophenes.30, 31, 58 The dications for short oligomers are 242 described in terms of a bipolaron structure.40, 59 The two positive charge centers separate the quinoidal and aromatic rings and form a structure similar to the T1 state, as illustrated in Figure 7.3C.30, 40, 59, 60 Furthermore, the bond orders for the two species are matched, because the unoccupied π orbital of the dication is equivalent to the two singly-occupied molecular orbitals (SOMOs) of the triplet excited state with π* and π character, respectively.

The calculated structures depicted in Figure 7.3 A and B provide some rational for the relatively complex experimental Raman spectra and the strong dependence upon oligomer length. However, there are obvious mismatches between the calculated and experimental spectra that indicate the shortcomings of DFT in treating these systems. The band positions of the highest frequency C=C stretch calculated by DFT exceed the experimental positions of the tetramer and hexamer by 45 and 84 cm-1, respectively.

Moreover, the calculated spectra give only a 26 cm-1 downshift for the hexamer, in comparison with a 50 cm-1 downshift in the experiment. Different functionals and basis sets did not improve the simulation of the T1 Raman spectra.

The mismatches between calculated and experimental spectra indicate that DFT does not treat the biradical structure of the T1 excited state in an accurate way. First, the overestimation of the quinoidal Cα=Cα and Cβ=Cβ stretch frequencies by DFT implies that the calculation underestimates the extent of overlap between the biradical with the central quinoidal region. The center site of the chain should consist of more radical-like character than predicted by DFT (Figure 7.3 A and B), and therefore the vibrational frequencies associated with C=C motion in the quinoidal region should be lower than

-1 calculated here. Second, the 50 cm downshift for the central Cα=Cα stretch modes 243

indicates that the triplet excitation extends more than predicted by DFT. The spectral shift is likely the result of the extension of the biradical into the outer ground-state rings which decreases electron density in the chain center. Consistent with our findings, we note that the zero-field splitting (D value) also decreases slightly for a hexamer relative to tetramer.26, 29 In short, the comparison of experimental and the DFT-calculated Raman spectra data implies that the biradical region of oligothiophenes extends more into the quinoidal and aromatic regions than predicted by DFT.

Higher level computations are necessary to fully model the triplet electronic structure. A recent study demonstrated that the CC2 method (approximate coupled cluster with single and double excitations) excelled in simulating the phosphorescence spectra of the thiophene dimer.61 Coupled cluster methods and advances in parallel algorithms now allow the calculation of triplet state properties for large molecules (e.g. β-carotene).62

These methods are likely to yield improved simulations of triplet Raman vibrational spectra.

It is necessary to point out that the resonance Raman enhancement effect was not considered in the above discussion. Though DFT has shown a surprisingly good match to ground state resonance Raman intensities of several chromophores, including

63 64 carotenoids, perylene diimide and oligothiophenes, T1→Tn transition may be different.

The higher triplet state Tn of oligothiophene is likely more delocalized along the conjugated chain as found in other molecules.65, 66 Therefore, the vibronic coupling of the

T1→Tn transition may change with the chain length because of the dependence of Tn wavefunction on molecular size. 244

The triplet excited states were also probed in a self-assembled aggregate of 2.

These aggregates are meant to mimic solid state samples such as thin films of OPVs. The aggregate ground state Raman spectrum is nearly identical with that in the solution phase

(Figure 7.4). The small variations in the frequencies of ground state Raman bands are possibly related to changes in molecular geometry.

Figure 7.4. Raman spectra of the T1 and S0 states of (Oct)4-6T aggregate in 5/95 (v/v) THF/water (black) and monomer in THF (thin, red). The T1 experiments employed a 420 nm pump and 647.09 nm probe, and were acquired with a ~1 µs time delay. The ground state spectra were acquired with 647.09 nm probe.

The triplet Raman spectra of solvated and aggregated 2 have band positions that are close in frequency, but their intensity profiles differ substantially (Figure 7.4). Raman bands of the aggregate show a consistent decrease in intensities towards lower frequencies, relative to the solvated triplet monomer. In particular, the intense Raman bands around 550~750 cm-1 in the solution phase spectrum are nearly absent from the 245 aggregate spectrum. The decreasing trend for intensities of lower frequency bands is possibly related to an increased homogeneous broadening (dephasing) factor for the aggregated state, which would tend to attenuate low-frequency modes in particular.67

This large dephasing factor may be associated with exciton interactions that are specific to the aggregate. For example, the higher Tn state can be delocalized and coupled with the surrounding electronic states as found in earlier research of oligothiophene aggregates.68

Another possible process may be coupling between T1 and a charge transfer state. Given that the triplet and charged states have similar configurations, the reorganization energy between the two states can be very small, facilitating the interaction.

The results described above comprise the first report and analysis of resolved vibrational spectra of oligothiophenes in the triplet excited state. Their Raman spectra differ greatly from the ground state, but are similar to spectra of dications. Vibrational mode analysis based on DFT calculations suggests that the triplet state consists of a quinoidal and biradical regions that are mainly located on four thiophene rings.

Additional rings beyond these four are likely aromatic, and they have a significant effect on the triplet excited state. Computation methods beyond DFT are necessary to fully understand the vibronic structure of the T1 states. The T1 spectral features are largely maintained in an aggregate, and suggest that spectra recorded for solvated species will be relevant for studies of the T1 state in oligo-/polythiophene films or bulk heterojunctions.

References

1. Friend, R. H.; Gymer, R. W.; Holmes, A. B.; Burroughes, J. H.; Marks, R. N.; Taliani, C.; Bradley, D. D. C.; Santos, D. A. D.; Bredas, J. L.; Logdlund, M.; Salaneck, W. R., Electroluminescence in conjugated polymers. Nature 1999, 397, (6715), 121-128. 246

2. Köhler, A.; Wilson, J. S.; Friend, R. H., Fluorescence and Phosphorescence in Organic Materials. Advanced Materials 2002, 14, (10), 701-707.

3. Köhler, A.; Bässler, H., Triplet states in organic semiconductors. Materials Science and Engineering R 2009, 66, (71-109).

4. Yersin, H.; Rausch, A. F.; Czerwieniec, R., Organometallic emitters for OLEDs: Triplet harvesting, singlet harvesting, case structures, and trends. In Physics of Organic Semiconductors, Wiley-VCH Verlag GmbH & Co. KGaA: 2012; pp 371-424.

5. Ohkita, H.; Cook, S.; Astuti, Y.; Duffy, W.; Heeney, M.; Tierney, S.; McCulloch, I.; Bradley, D. D. C.; Durrant, J. R., Radical ion pair mediated triplet formation in polymer-fullerene blend films. Chemical Communications 2006, (37), 3939-3941.

6. Westenhoff, S.; Howard, I. A.; Hodgkiss, J. M.; Kirov, K. R.; Bronstein, H. A.; Williams, C. K.; Greenham, N. C.; Friend, R. H., Charge recombination in organic photovoltaic devices with high open-circuit voltages. Journal of the American Chemical Society 2008, 130, (41), 13653-13658.

7. Schlenker, C. W.; Chen, K. S.; Yip, H. L.; Li, C. Z.; Bradshaw, L. R.; Ochsenbein, S. T.; Ding, F. Z.; Li, X. S. S.; Gamelin, D. R.; Jen, A. K. Y.; Ginger, D. S., Polymer triplet energy levels need not limit photocurrent collection in organic solar cells. Journal of the American Chemical Society 2012, 134, (48), 19661-19668.

8. Shao, Y.; Yang, Y., Efficient organic heterojunction photovotaic cells based on triplet materials. Advanced Materials 2005, 17, 2841-2844.

9. Yost, S. R.; Hontz, E.; Yeganeh, S.; Voorhis, T. V., Triplet vs singlet energy transfer in organic semiconductors: The tortoise and the hare. Journal of physical Chemistry C 2012, 116, 17369-17377.

10. Hanna, M. C.; Nozik, A. J., Solar conversion efficiency of photovoltaic and photoelectrolysis cells with carrier multiplication absorbers. Journal of Applied Physics 2006, 100, (7), 0745101-8.

11. Smith, M. B.; Michl, J., Singlet fission. Chemical Reviews 2010, 110, (11), 6891- 6936.

12. Tayebjee, M. J. Y.; Gray-Weale, A. A.; Schmidt, T. W., Thermodynamic limit of exciton fission solar cell efficiency. Journal of Physical Chemistry Letters 2012, 3, (19), 2749-2754.

13. Bäuerle, P.; Becher, J.; Lau, J.; Mark, P., Sulfur-containing oligomers. In Electronic Materials: The Oligomer Approach, Wiley-VCH Verlag GmbH: 2007; pp 105-233. 247

14. Becker, R. S.; deMelo, J. S.; Macanita, A. L.; Elisei, F., Comprehensive evaluation of the absorption, photophysical, energy transfer, structural, and theoretical properties of alpha-oligothiophenes with one to seven rings. Journal of Physical Chemistry 1996, 100, (48), 18683-18695.

15. Beljonne, D.; Cornil, J.; Friend, R. H.; Janssen, R. A. J.; Bredas, J. L., Influence of chain length and derivatization on the lowest singlet and triplet states and intersystem crossing in oligothiophenes. Journal of the American Chemical Society 1996, 118, (27), 6453-6461.

16. de Melo, J. S.; Silva, L. s. M.; Arnaut, L. s. G.; Becker, R. S., Singlet and triplet energies of α-oligothiophenes: A spectroscopic, theoretical, and photoacoustic study: Extrapolation to polythiophene. Journal of Chemical Physics 1999, 111, (12), 5427-5433.

17. Landwehr, P.; Port, H.; Wolf, H. C., Optical detection of the lowest triplet excited state of oligothiophenes and anthryloligothiophenes. Chemical Physics Letters 1996, 260, (1-2), 125-129.

18. Rentsch, S.; P. Yang, J.; Paa, W.; Birckner, E.; Schiedt, J.; Weinkauf, R., Size dependence of triplet and singlet states of α-oligothiophenes. Physical Chemistry Chemical Physics 1999, 1, (8), 1707-1714.

19. Wasserberg, D.; Marsal, P.; Meskers, S. C. J.; Janssen, R. A. J.; Beljonne, D., Phosphorescence and triplet state energies of oligothiophenes. Journal of Physical Chemistry B 2005, 109, (10), 4410-4415.

20. Fabiano, E.; Sala, F. D.; Cingolani, R.; Weimer, M.; Görling, A., Theoretical study of singlet and triplet excitation energies in oligothiophenes. Journal of Physical Chemistry A 2005, 109, (13), 3078-3085.

21. Siegert, S.; Vogeler, F.; Marian, C. M.; Weinkauf, R., Throwing light on dark states of alpha-oligothiophenes of chain lengths 2 to 6: Radical anion photoelectron spectroscopy and excited-state theory. Physical Chemistry Chemical Physics 2011, 13, (21), 10350-63.

22. Della Sala, F.; Heinze, H. H.; Görling, A., Excitation energies of terthiophene and its dioxide derivative: A first-principles study. Chemical Physics Letters 2001, 339, 343- 350.

23. Oseki, Y.; Fujitsuka, M.; Sakamoto, M.; Cai, X. C.; Majima, T., Energy levels of oligothiophenes in the higher excited triplet states. Journal of Physical Chemistry C 2007, 111, (2), 1024-1028.

24. Rubio, M.; Merchán, M.; Pou-Amérigo, R.; Ortí, E., The low-lying excited states of 2,2′-bithiophene: A theoretical analysis. ChemPhysChem 2003, 4, (12), 1308-1315. 248

25. Rubio, M.; Merchán, M.; Ortí, E., A theoretical study on the low-lying excited states of 2,2′:5′,2′′-terthiophene and 2,2′:5′,2′′:5′′,2′′′-quaterthiophene. ChemPhysChem 2005, 6, (7), 1357-1368.

26. Bennati, M.; Nemeth, K.; Surjan, P. R.; Mehring, M., Zero-field-splitting and pi- electron spin densities in the lowest excited triplet state of oligothiophenes. Journal of Chemical Physics 1996, 105, (11), 4441-4447.

27. Swanson, L. S.; Shinar, J.; Yoshino, K., Optically detected magnetic resonance study of polaron and triplet-exciton dynamics in poly(3-hexylthiophene) and poly(3- dodecylthiophene) films and solutions. Physical Review Letters 1990, 65, (9), 1140-1143.

28. Bennati, M.; Grupp, A.; Bäuerle, P.; Mehring, M., Triplet states of oligothiophenes and electron transfer to C60: A pulsed EPR investigation. Molecular Crystals and Liquid Crystals Science and Technology. Section A. Molecular Crystals and Liquid Crystals 1994, 256, (1), 751-756.

29. Bennati, M.; Grupp, A.; Mehring, M.; Bauerle, P., Pulsed EPR spectroscopy of the photoexcited triplet states of thiophene oligomers in frozen solution. Journal of Physical Chemistry 1996, 100, (8), 2849-2853.

30. Furukawa, Y., Electronic absorption and vibrational spectroscopies of conjugated conducting polymers. Journal of Physical Chemistry 1996, 100, (39), 15644-15653.

31. Louarn, G.; Buisson, J. P.; Lefrant, S.; Fichou, D., Vibrational studies of a series of α-oligothiophenes as model systems of polythiophene. Journal of Physical Chemistry 1995, 99, (29), 11399-11404.

32. Degli Esposti, A.; Moze, O.; Taliani, C.; Tomkinson, J. T.; Zamboni, R.; Zerbetto, F., The intramolecular vibrations of prototypical polythiophenes. Journal of Chemical Physics 1996, 104, (24), 9704-9718.

33. Degli Esposti, A.; Fanti, M.; Muccini, M.; Taliani, C.; Ruani, G., The polarized infrared and Raman spectra of alpha-T6 single crystal: An experimental and theoretical study. Journal of Chemical Physics 2000, 112, (13), 5957-5969.

34. Pop, S. C.; Beu, T. A., Vibrations of polythiophenes. Computational and Theoretical Chemistry 2012, 995, 66-74.

35. Akimoto, M.; Furukawa, Y.; Takeuchi, H.; Harada, I.; Soma, Y.; Soma, M., Correlation between vibrational spectra and electrical-conductivity of polythiophene. Synthetic Metals 1986, 15, (4), 353-360.

36. Furukawa, Y.; Akimoto, M.; Harada, I., Vibrational key bands and electrical- conductivity of polythiophene. Synthetic Metals 1987, 18, (1-3), 151-156. 249

37. Sauvajol, J. L.; Chenouni, D.; Lereporte, J. P.; Chorro, C.; Moukala, B.; Petrissans, J., Resonant Raman spectra and photoluminescence in polythiophene. Synthetic Metals 1990, 38, (1), 1-12.

38. Weinberg-Wolf, J. R.; McNeil, L. E., Resonant Raman spectroscopy on alpha- hexathiophene single crystals. Physical Review B 2004, 69, (12), 4.

39. Clarke, T. M.; Gordon, K. C.; Chan, W. S.; Phillips, D. L.; Wagner, P.; Officer, D. L., Raman spectroscopy of short-lived terthiophene radical cations generated by photochemical and chemical oxidation. Chemphyschem 2006, 7, (6), 1276-1285.

40. Gonzalez, S. R.; Ie, Y.; Aso, Y.; Navarrete, J. T. L.; Casado, J., The frontiers of quinoidal stability in long oligothiophenes: Raman spectra of dicationic polaron pairs. Journal of the American Chemical Society 2011, 133, (41), 16350-16353.

41. Heckel, J. C.; Weisman, A. L.; Schneebeli, S. T.; Hall, M. L.; Sherry, L. J.; Stranahan, S. M.; Dubay, K. H.; Friesner, R. A.; Willets, K. A., Polarized Raman spectroscopy of oligothiophene crystals to determine unit cell orientation. Journal of Physical Chemistry A 2012, 116, (25), 6804-6816.

42. Schaffer, H. E.; Heeger, A. J., Infrared activity of photoexcitations in polythiophene. Solid State Communications 1986, 59, (7), 415-421.

43. Yamaguchi, S.; Hamaguchi, H., Ultrafast vibrational-relaxation in photogenerated S1 α-terthiophene in solution by femtosecond time-resolved absorption emission and picosecond time-resolved Raman spectroscopy. Chemical Physics Letters 1994, 227, (3), 255-260.

44. Yu, W. J.; Zhou, J. W.; Bragg, A. E., Exciton conformational dynamics of poly(3- hexylthiophene) (P3HT) in solution from time-resolved resonant Raman spectroscopy. Journal of Physical Chemistry Letters 2012, 3, (10), 1321-1328.

45. Osterbacka, R.; An, C. P.; Jiang, X. M.; Vardeny, Z. V., Delocalized polarons in self-assembled poly(3-hexyl thiophene) nanocrystals. Synthetic Metals 2001, 116, (1-3), 317-320.

46. Jiang, X. M.; Osterbacka, R.; Korovyanko, O.; An, C. P.; Horovitz, B.; Janssen, R. A. J.; Vardeny, Z. V., Spectroscopic studies of photoexcitations in regioregular and regiorandom polythiophene films. Advanced Functional Materials 2002, 12, (9), 587- 597.

47. Ostrowski, D. P.; Lytwak, L. A.; Mejia, M. L.; Stevenson, K. J.; Holliday, B. J.; Vanden Bout, D. A., The effects of aggregation on electronic and optical properties of oligothiophene particles. Acs Nano 6, (6), 5507-5513. 250

48. Pan, D. H.; Mathies, R. A., Chromophore structure in lumirhodopsin and metarhodopsin I by time-resolved resonance Raman microchip spectroscopy. Biochemistry 2001, 40, (26), 7929-7936.

49. Zerbi, G.; Chierichetti, B.; Ingänas, O., Vibrational spectra of oligothiophenes as model of polythiophenes. The Journal of Chemical Physics 1991, 94, (6), 4637-4645.

50. Zerbi, G.; Castiglioni, C.; Del Zoppo, M., Structure and Optical Properties of Conjugated Oligomers from their Vibrational Spectra. In Electronic Materials: The Oligomer Approach, Wiley-VCH Verlag GmbH: 2007; pp 345-402.

51. Takayanagi, M.; Gejo, T.; Hanazaki, I., Geometry and torsional potential of 2,2'- bithiophene in a supersonic jet. Journal of Physical Chemistry 1994, 98, (49), 12893- 12898.

52. Bredas, J. L.; Themans, B.; Fripiat, J. G.; Andre, J. M.; Chance, R. R., Highly conducting polyparaphenylene, polypyrrole, and polythiophene chains - An abinitio study of the geometry and electronic structure modification upon doping. Physical Review B 1984, 29, (12), 6761-6773.

53. Bredas, J. L.; Street, G. B., Polarons, bipolarons, and solitons in conducting polymers. Accounts of Chemical Research 1985, 18, (10), 309-315.

54. Beljonne, D.; Shuai, Z.; Bredas, J. L., Theoretical study of thiophene oligomers - Electronic excitations, relaxation energies and nonlinear-optical properties. Journal of Chemical Physics 1993, 98, (11), 8819-8828.

55. Ortiz, R. P.; Casado, J.; Hernandez, V.; Navarrete, J. T. L.; Orti, E.; Viruela, P. M.; Milian, B.; Hotta, S.; Zotti, G.; Zecchin, S.; Vercelli, B., Magnetic properties of quinoidal oligothiophenes: More than good candidates for ambipolar organic semiconductors? Advanced Functional Materials 2006, 16, (4), 531-536.

56. Ortiz, R. P.; Casado, J.; Hernandez, V.; Navarrete, J. T. L.; Viruela, P. M.; Orti, E.; Takimiya, K.; Otsubo, T., On the biradicaloid nature of long quinoidal oligothiophenes: Experimental evidence guided by theoretical studies. Angewandte Chemie-International Edition 2007, 46, (47), 9057-9061.

57. Casado, J.; Ortiz, R. P.; Navarrete, J. T. L., Quinoidal oligothiophenes: New properties behind an unconventional electronic structure. Chemical Society Reviews 2012, 41, (17), 5672-5686.

58. Yokonuma, N.; Furukawa, Y.; Tasumi, M.; Kuroda, M.; Nakayama, J., Electronic absorption and Raman studies of BF4--doped polythiophene based on the spectra of the radical cation and dication of alpha-sexithiophene. Chemical Physics Letters 1996, 255, (4-6), 431-436. 251

59. Zade, S. S.; Bendikov, M., Theoretical study of long oligothiophene dications: Bipolaron vs polaron pair vs triplet state. Journal of Physical Chemistry B 2006, 110, (32), 15839-15846.

60. Heeger, A. J.; Kivelson, S.; Schrieffer, J. R.; Su, W. P., Solitons in conducting polymers. Reviews of Modern Physics 1988, 60, (3), 781-850.

61. Andrzejak, M.; Orzel, L., Joint theoretical and experimental study on the phosphorescence of 2,2'-bithiophene. Physical Chemistry Chemical Physics 2014, 16, (12), 5605-5612.

62. Hu, H.-S.; Bhaskaran Nair, K.; Aprà, E.; Govind, N.; Kowalski, K., Towards enabling large-scale open-shell equation-of-motion coupled cluster calculations: Triplet States of β-carotene. Journal of Physical Chemistry A 2014.

63. Wang, C.; Berg, C. J.; Hsu, C. C.; Merrill, B. A.; Tauber, M. J., Characterization of carotenoid aggregates by steady-state optical spectroscopy. Journal of Physical Chemistry B 2012, 116, (35), 10617-10630.

64. Angelella, M.; Wang, C.; Tauber, M. J., Resonance Raman spectra of a perylene bis (dicarboximide) chromophore in ground and lowest triplet states. The Journal of Physical Chemistry A 2013, 117, (38), 9196-9204.

65. Wasserberg, D.; Dudek, S. P.; Meskers, S. C. J.; Janssen, R. A. J., Comparison of the chain length dependence of the singlet- and triplet-excited states of oligofluorenes. Chemical Physics Letters 2005, 411, 273-277.

66. Peeters, E.; Ramos, A. M.; Meskers, S. C. J.; Janssen, R. A. J., Singlet and triplet excitations of chiral dialkoxy-p-phenylene vinylene oligomers. Journal of Chemical Physics 2000, 112, (21), 9445-9454.

67. Myers, A. B.; Mathies, R. A., Resonance Raman intensities: A probe of excited- state structure and dynamics. In Biological Applications of Raman Spectroscopy, John Wiley & Sons: 1987; Vol. 2, pp 1-58.

68. Apperloo, J. J.; Janssen, R. A. J.; Malenfant, P. R. L.; Frechet, J. M. J., Interchain delocalization of photoinduced neutral and charged states in nanoaggregates of lengthy oligothiophenes. Journal of the American Chemical Society 2001, 123, (28), 6916-6924.

252

Appendix of Chapter 7

Steady-state absorption spectroscopy. Absorption spectra were acquired with a

UV-Vis spectrometer (Shimadzu, UV-3600) with bandwidth set to 1 nm. Samples were contained in quartz cuvettes with pathlength 0.2 or 1.0 cm. Absorption spectra of the aggregated sample were also acquired via measurement of diffuse transmittance using an integrating sphere attachment (Shimadzu, ISR-3100).

Nanosecond transient absorption (ns-TA) spectroscopy. Oligothiophene solutions for TA spectroscopy were prepared in an argon-filled glovebox (MBraun

Unilab). The level of O2 was less than 1 ppm. The concentrations of didocyl tetrathiophene ((Dodec)2-4T, or 1) in chlorobenzene, and the tetraoctyl hexathiophene

((Oct)4-6T, or 2) in THF were 40 and 10 µM, respectively. The optical absorption at the pump wavelength was ~0.3/cm for both solutions. Samples were contained in an airtight cuvette with 1 cm pathlength, and continuously agitated with a magnetic stirrer during the pump-probe spectroscopy.

Nanosecond transient absorption spectra were acquired with a system described previously.1 The pump wavelength was 430 nm, and the pulses were ~6 ns in duration at a repetition rate of 20 Hz. The excitation energy was ~0.2 mJ/pulse and beam diameter was 6 mm at the sample.

Results from steady-state and transient absorption spectroscopy. Absorption spectra of 1 and 2 in their ground states and first excited triplet states 31* and 32* are shown in Figure A7.1. The ground state absorption maxima are located at 382 and 410 nm, respectively, both of which are blue-shifted compared to their unsubstituted 253 counterparts (α-4T and α-6T).2 The triplet state absorptions have vibronic structure, with maxima at 585 and 615 nm for 1 and 640 and 685 nm for 2. The band positions of 31* and 32*are similar to the triplet state absorption bands of unsubstituted oligomers.2

Figure A7.1. Steady state absorption of the S0 state, and transient absorption spectra of the T1 excited-state of 1 (in chlorobenzene) and 2 (in THF). All spectra are normalized to their absorption maxima. The transient absorption spectra are obtained with a 430 nm pump wavelength. The red arrow indicates the wavelength position of the probe for the Raman experiments.

The ground-state absorption spectrum of aggregated 2 has a peak position ~600 cm-1 red-shifted compared with solvated 2 (Figure A7.2). The spectrum of the aggregate also has enhanced absorption on the low energy side of the main band, a characteristic that is only slightly diminished when the spectrum is acquired with an integrating sphere

(diffuse transmittance). The latter measurement verifies that scattering from aggregates has only a minor contribution to the overall attenuation. 254

The triplet state absorption maximum of aggregated 2 redshifts 600 cm-1 compared with that of the monomer. The 647.09 nm Raman excitation wavelength is less resonant with the T1 absorption in the aggregate than with the monomer.

Figure A7.2. Steady state absorption of the S0 state, and transient absorption spectra of the T1 excited-state of monomeric 2 in THF, and aggregated 2 in 5/95 (v/v) THF/H2O. All spectra are normalized to their maxima. The transient absorption spectra are obtained with a 430 nm pump. The red arrow indicates the wavelength of the probe for the Raman experiments.

Transient resonance Raman spectroscopy. Resonance Raman spectra of the T1 state were acquired with a pump-probe setup based on a modified optical microscope

(Zeiss, Axio Imager) and a fast-flowing solution in a capillary. The principle of our setup is similar to a previous design.3 The pump beam was produced by second-harmonic generation of the 840 nm laser pulses from a Ti:Sapphire oscillator (Spectra-Physics,

Mai Tai). The high 78.8 MHz repetition rate makes this source effectively a continuous- wave (cw) excitation. The probe beam was produced by an argon/krypton ion laser (Laser 255

Innovations Innova-70C). The wavelength 647.09 nm was selected because of near- resonance with strong absorption bands of the lowest triplet state of 1 and 2 as shown in

Figure A7.1. The pump and probe beams were combined with a dichroic beamsplitter

(Thorlabs, DMLP425) that transmitted the probe and reflected the pump. By adjusting the angle of the beamsplitter, the pump and probe beams entered the microscope objective at slightly different angles, so that their foci at the solution within a capillary (see below) were separated by several tens of micrometers along the direction of the flow. A 50R/50T beamsplitter sent the beams through an apochromatic near-UV objective (Seiwa, UV

PlanAPO 20X). The numerical aperture (NA) was 0.5, and magnification was nominally

20× with focal length 10 mm. The pump beam was elliptical in shape, with axis lengths

4–5 µm parallel to the flow and 1–3 µm perpendicular to the flow.

The average power of the pump beam at the sample point was 1.2 mW. The power of the probe beam was 15 mW at the sample, and the spot was circular with a diameter of 1 to 3 µm. The power density for the pump and probe were high in our experiments. The pump is expected to photoexcite all oligomers in the beampath, as estimated from the power, beam size, sample flow velocity and molar absorptivity.4

Similarly, once the triplet state formed, the focused probe is expected to cause several excitations to higher-lying triplet excited states during the transit of the molecule through the focal point. Additional experiments run with probe photon densities ~100 times lower showed no significant differences from the T1 spectra acquired in the present work. A likely reason why the high probe excitation had no detrimental effect is that the inherent

Tn T1 relaxation rate is much faster than the rate of T1Tn photoexcitation. 256

Solutions of oligothiophenes for Raman spectroscopy were prepared in the argon- filled glovebox, and filtered with a PTFE filter with 5 µm pore size (Millipore, Type

JVWP). The concentrations of the 1 and 2 monomer solutions were ~150 µM. The starting concentration for the sample of aggregated 2 was ~120 µM , but decreased to ~50

µM after filtration. Samples were loaded into a 25 mL glass vial and inserted into a custom brass vessel that was built to withstand internal pressures exceeding 6.9 MPa

(1000 psi). The outlet for the solution was via narrow Teflon tubing that penetrated the cap of the pressure vessel with a Swagelok fitting. The pressure vessel, tubing, and capillary was assembled and sealed in the glovebox.

At the start of an experimental run, the solution was pressure-driven through the quartz capillary (Polymicro, TSP050150). The capillary had inner diameter of 50 µm, length 7.5 cm, and the polyimide coating was removed over a ~1 cm span near the foci of the laser beams. The capillary was connected to the outer end of the Teflon tubing with a

PEEK connector (IDEX). Pressure was provided by high-purity N2 connected via a side- port to the brass vessel. By controlling the pressure of N2 in the range of 0.69 to 3.1 MPa

(100 to 450 psi), the average flow velocity of THF solutions varied from 1 to 5 m/s. The flow rates determined by volumetric measurements compared well with values of 1.3 to

6.5 m/s derived from Poiseuille’s equation.

The time delay was determined by the velocity of the solution and the spatial separation of pump and probe foci. The strongest Raman signal is expected to originate from the center of the capillary, where the flow velocity is twice the average value (2 to

10 m/s). For experiments with the monomer, the probe beam was positioned 45 µm downstream from the pump, therefore the time delay was 5 to 20 µs, depending upon the 257 driving pressure. For the experiment with the aggregate, the time delay was ~1 μs with pump/probe separation of 10 µm and driving pressure of 3.1 MPa.

Back-scattered light from the sample was collected and collimated with the same objective lens used to focus the excitation beams. The scattered light was focused at the entrance slit of a spectrograph with a plano-convex lens with a focal length of 200 mm.

The pump-induced fluorescence was largely blocked by the narrow (0.03 mm) opening of the slit. The Rayleigh scattering from the probe was blocked by an edge filter (Semrock,

647 nm RazorEdge). The Raman signal was dispersed in the spectrograph (Horiba Jobin-

Yvon, iHR-320) with a 1200 groove/mm holographic grating, and detected with an open- electrode CCD detector (Horiba Jobin-Yvon, Synapse). The spectral resolution was 3 cm-1.

Analysis of transient Raman spectra. The workup of the monomeric 2 spectra is presented here as an example. The spectra of oligothiophenes were acquired with three illumination conditions: pump+probe, probe_only, and pump_only. Although the pump- induced fluorescence was attenuated by spatially separating the pump and probe at the sample plane, there was still a substantial pump background that could be caused by pump scatter, or delayed fluorescence from the S1 state formed by T1-T1 annihilation. A probe-only spectrum of the solvent was also acquired. The four raw spectra were acquired for 50 minutes each, and are shown in Figure A7.3A. The Raman shifts were calibrated with a spectrum of a acetonitrile/toluene (1:1, v/v) solution with known peak assignments.

The final triplet state spectrum was obtained via the following the steps: 258

(1) The pump background was removed with the subtraction (pump+probe) – k*(pump_only). The scaling factor k (in this example, k = 1.23) was optimized to yield the flattest baseline as shown in Figure A7.3B.

(2) Raman bands from the solvent and ground state 2 were removed by subtracting a scaled (probe_only) spectrum from the result in step (2). The scaling factor

(in this case, 1.19) was optimized to eliminate the THF band at 913 cm-1 (Figure A7.3B).

(3) The final spectrum (Figure A7.3C) was obtained by removing a baseline, generated by a cubic Bézier spline with 25 sampling points, from the spectrum obtained in step (3).

We note that the procedure above yields the spectrum of the T1 excited state without negative features that may be expected from pump-induced attenuation of ground-state Raman bands, particularly with the high degree of estimated pump photoexcitation. The reason is that the scattering cross sections of the ground state are much lower than those of the triplet excited state at the 647.09 nm probe wavelength

(Figure A7.4). 259

Figure A7.3. Transient Raman spectra data workup of 2 in THF. The intensities of all spectra are shown in CCD counts per second. (A) Raw data. Black arrows indicate signals from the T1 state generated by pump excitation. (B) Difference spectra obtained by subtracting the pump background and the probe_only signal. A probe_only spectrum is shown for comparison. 260

Figure A7.4 Final spectrum of the T1 state for tetraoctyl hexathiophene (2) in THF. The intensities of all spectra are shown in CCD counts per second. An unscaled ground state spectrum is shown to allow comparison of the T1 and S0 Raman cross sections.

Confirmation of T1 Assignment. In the case of monomeric (Oct)4-6T, the assignment of the observed transient species to a triplet excited state is consistent with two other findings shown in Figure A7.5. First, the Raman signals decrease significantly when the pump-probe delay is increased from ~5 to ~20 µs. The finding is consistent with the ~20 µs lifetime for unsubstituted α-sexithiophene.2 Second, the transient signal nearly disappears when the sample is saturated with O2. This observation can be explained by the O2 molecules quenching the triplet excited states of 2 and forming singlet oxygen. An alternative assignment of the transient to a cation is not supported by this test, because its formation would likely be enhanced in the presence of O2.

261

Figure A7.5. Transient resonance Raman spectra of 2 in THF with 420 nm pump and

647.09 nm probe at (a) ~5 µs, (b) ~20 µs time delay and (c) ~5 µs time delay with O2.

DFT Calculations. Electronic structure and normal mode calculations of oligothiophenes were run with Gaussian 09W software.5 The dodecyl and octyl substituents of 1 and 2 were replaced with ethyl groups to reduce computational time.

The ground state molecular structures were first optimized with the Hartree-Fock method and 3-21G basis set, followed by density functional theory (DFT) using the B3LYP functional and 6-31G(d) basis set with tight convergence criterion. The optimization of the lowest triplet state began with the optimized singlet ground-state geometry, and was carried out with the same functional and basis set, but with the spin set to unrestricted.

Raman spectra and normal modes of vibration for the S0 and T1 states were calculated at the same computational level as the geometry optimization. A scaling factor of 0.9614 was applied to calculated frequencies for the ground-state spectra.6 The same factor was 262

7 used to scale the T1 spectra. Normal mode diagrams were prepared using Jmol. We also conducted calculations with other functionals and higher level basis sets for the triplet state, which will be discussed in a later part of this appendix.

Vibrational mode assignments for ground state oligothiophenes. Assignments of the vibrational bands were made by comparison with results from DFT calculations

(Figure A7.6) and prior work. Vector diagrams of representative modes of the tetramer are shown in Figure A7.7. The high frequency region of the experimental spectra spanning 1400 to 1550 cm-1 consists of two strong Raman bands, located at 1467 and

-1 -1 1531 cm for 1 and 1471 and 1515 cm for 2. These Raman bands are assigned to in- phase combinations of Cα=Cβ stretching vibrations. These modes are particularly strong because the polarizability changes are large for in-phase stretching motions of a series of double bonds. The medium-amplitude Raman bands appearing around ~1050 cm-1 can be attributed to several C-H in-plane bending modes. C-C single bond stretching modes are assigned to low-intensity bands centered at 1177 and 1168 cm-1 for 1 and 2, respectively.

In the low frequency region (650~850 cm-1), typical Raman bands for the unsubstituted oligomers are observed at 690~700 and ~740 cm-1. These bands have been previously assigned to ring deformations that include C-S bond stretches.8, 9 Vibrational modes which have frequencies <650 cm-1 are attributed to out-of-plane skeleton bends.

263

Figure A7.6. (Upper panels) Experimental Raman spectra probed at 647.09 nm for ground-state 1 (in chlorobenzene), and 2 (in THF). As asterisk (*) marks a residual Raman feature from subtraction of the solvent. (Lower panels) DFT calculated spectra (lower spectra) of the ethyl substituted tetra- and hexathiophene. Calculated spectra have frequencies scaled by a factor of 0.9614. Bands that do not have close experimental counterparts or those associated with local vibrations on the ethyl groups are not labeled in the calculated spectra.

264

1531 (1530) cm-1 1511 (1515) cm-1

1460 (1468) cm-1 1454 (1470) cm-1

1368 (1368) cm-1 1368 (1390) cm-1

1182 (1177) cm-1 1177 (1168) cm-1

1043 (1049) cm-1 1048 (1053) cm-1

676 (695) cm-1 668 (687) cm-1

575 (581) cm-1 575 (584) cm-1

Figure A7.7. Representative vibrational normal modes from DFT calculations of

(ethyl)2-4T and (ethyl)4-6T in the ground state. Complementary modes are listed side-by- side for comparison. The normal modes are labeled with the calculated values, scaled by 0.9614. In parentheses are experimental frequencies of bands that can be possibly assigned to these modes. 265

Exploration of DFT methods for calculating the triplet Raman spectra. We experimented with various basis sets and functionals in an attempt to improve the agreement between DFT-calculated spectra and experimental results.

(1) Increasing the size of basis set from 6-31G(d) to 6-311G(d) or 6-311G(d,p) did not significantly change the computation result for the triplet state (Figure A7.8).

Figure A7.8. DFT calculated Raman spectra for the T1 state of (ethyl)2-4T with B3LYP functional using different basis sets. The scaling factors are 0.9614, 0.9739 and 0.9890 for 6-31G(d), 6-311G(d) and 6-311G(d,p) basis sets, respectively. These factors are optimized for scaling the ground state frequencies6, but they were used for triplet spectra here for consistency.

(2) To address possible effects of the self-interaction error,10, 11 we explored functionals with long-range corrections such as CAM-B3LYP12 and ωB97X-D13 as reported for studies of other conjugated molecules in their ground state.14-17 The triplet

Raman spectra calculated with these functionals show similar patterns in the region 266

<1400 cm-1 as the spectrum calculated with B3LYP. However, both CAM-B3LYP and

ωB97X-D significantly overestimate frequencies for the C=C stretch modes.(Figure A7.9)

Figure A7.9. DFT calculated Raman spectra for the T1 state of (ethyl)4-6T with B3LYP, CAM-B3LYP and ωB97X-D functionals with a 6-31G(d) basis set. No scaling factors were applied.

(3) Functionals utilizing generalized gradient approximation, such as PBEPBE, have also been used to calculate the triplet state of organic chromophores in previous studies.18, 19 We attempted to calculate the Raman spectra with PBEPBE functional. The ground state spectrum calculated with the PBEPBE functional generally agrees with the experiment but not as well as the B3LYP result, as shown in Figure A7.10. For the triplet state, the PBEPBE functional does not show significant improvement compared with

B3LYP. As shown in Figure A7.11, assignment of the experimental bands to PBEPBE- calculated vibrational modes suffers from similar problems as the B3LYP calculation. 267

The bands in the spectral region 1300-1600 cm-1 are generally a poor match to the experimental data. In the spectral region <1200 cm-1, the difficulty of assignment is further compounded by a complete mismatch of the weak calculated intensities, relative the strong experimental bands.

Figure A7.10. Raman spectra of (ethyl)4-6T in the ground state, calculated with DFT and the B3LYP or PBEPBE functionals. The 6-31G(d) basis set was used for both calculations. The scaling factors were 0.96146 and 0.987520 for B3LYP and PBEPBE functionals, respectively. The experimental spectrum of 2 in THF is shown as dotted lines.

268

Figure A7.11. Raman spectra of (ethyl)4-6T in the triplet (T1) state, calculated with DFT and the B3LYP or PBEPBE functionals. The 6-31G(d) basis set was used for both calculations. The same scaling factors were used as for the S0 calculations. The experimental spectrum of 2 in THF is shown as dotted lines.

To summarize, our exploration of functionals and basis sets provided no significant improvement beyond the results obtained with B3LYP and the 6-31G(d) basis set.

269

T1 normal mode diagrams

1526 cm-1 1500 cm-1

1490 cm-1 1487 cm-1

-1 -1 1337 cm 1328 cm

1182 cm-1 1185 cm-1

1079 cm-1 1072 cm-1

667 cm-1 670 cm-1

-1 -1 621 cm 649 cm

569 cm-1

Figure A7.12. Comparison between similar T1 vibrational normal modes for (ethyl)2-4T and (ethyl)4-6T from DFT (UB3LYP/6-31G(d)) calculation. DFT calculated frequencies were scaled by 0.9614.

270

References

1. Wang, C.; Angelella, M.; Kuo, C. H.; Tauber, M. J. In Singlet fission in carotenoid aggregates: insights from transient absorption spectroscopy, SPIE NanoScience+ Engineering, 2012; International Society for Optics and Photonics: 2012; pp 845905-845905-13.

2. Becker, R. S.; deMelo, J. S.; Macanita, A. L.; Elisei, F., Comprehensive evaluation of the absorption, photophysical, energy transfer, structural, and theoretical properties of alpha-oligothiophenes with one to seven rings. Journal of Physical Chemistry 1996, 100, (48), 18683-18695.

3. Pan, D. H.; Mathies, R. A., Chromophore structure in lumirhodopsin and metarhodopsin I by time-resolved resonance Raman microchip spectroscopy. Biochemistry 2001, 40, (26), 7929-7936.

4. Mathies, R.; Oseroff, A. R.; Stryer, L., Rapid-flow Resonance Raman spectroscopy of photolabile molecules - Rhodopsin and isorhodopsin. Proceedings of the National Academy of Sciences of the United States of America 1976, 73, (1), 1-5.

5. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09, Gaussian Inc.: Wallingford, CT, 2009. 271

6. Cramer, C. J., Essential of Computational Chemistry. John Wiley & Sons, LTD: The Atrium, 2002.

7. Jmol, Jmol: an open-source Java viewer for chemical structures in 3D. http://www.jmol.org/

8. Louarn, G.; Buisson, J. P.; Lefrant, S.; Fichou, D., Vibrational studies of a series of α-oligothiophenes as model systems of polythiophene. Journal of Physical Chemistry 1995, 99, (29), 11399-11404.

9. Degli Esposti, A.; Moze, O.; Taliani, C.; Tomkinson, J. T.; Zamboni, R.; Zerbetto, F., The intramolecular vibrations of prototypical polythiophenes. Journal of Chemical Physics 1996, 104, (24), 9704-9718.

10. Ciofini, I.; Adamo, C.; Chermette, H., Effect of self-interaction error in the evaluation of the bond length alternation in trans-polyacetylene using density-functional theory. Journal of Chemical Physics 2005, 123, (12), 4.

11. Korzdorfer, T.; Parrish, R. M.; Sears, J. S.; Sherrill, C. D.; Bredas, J. L., On the relationship between bond-length alternation and many-electron self-interaction error. Journal of Chemical Physics 2012, 137, (12), 8.

12. Yanai, T.; Tew, D. P.; Handy, N. C., A new hybrid exchange-correlation functional using the Coulomb-attenuating method (CAM-B3LYP). Chemical Physics Letters 2004, 393, (1-3), 51-57.

13. Chai, J. D.; Head-Gordon, M., Long-range corrected hybrid density functionals with damped atom-atom dispersion corrections. Physical Chemistry Chemical Physics 2008, 10, (44), 6615-6620.

14. Jacquemin, D.; Adamo, C., Bond Length Alternation of Conjugated Oligomers: Wave Function and DFT Benchmarks. Journal of Chemical Theory and Computation 7, (2), 369-376.

15. McCormick, T. M.; Bridges, C. R.; Carrera, E. I.; Di Carmine, P. M.; Gibson, G. L.; Hollinger, J.; Kozycz, L. M.; Seferos, D. S., Conjugated Polymers: Evaluating DFT Methods for More Accurate Orbital Energy Modeling. Macromolecules 46, (10), 3879- 3886.

16. Nayyar, I. H.; Batista, E. R.; Tretiak, S.; Saxena, A.; Smith, D. L.; Martin, R. L., Role of geometric distortion and polarization in localizing electronic excitations in conjugated polymers. Journal of Chemical Theory and Computation 9, (2), 1144-1154.

17. Nayyar, I. H.; Batista, E. R.; Tretiak, S.; Saxena, A.; Smith, D. L.; Martin, R. L., Localization of electronic excitations in conjugated polymers studied by DFT. Journal of Physical Chemistry Letters 2, (6), 566-571. 272

18. Clemens, O.; Basters, M.; Wild, M.; Wilbrand, S.; Reichert, C.; Bauer, M.; Springborg, M.; Jung, G., Solvent effects on the absorption/emission spectra of an organic chromophore: A theoretical study. Journal of Molecular Structure: THEOCHEM 2008, 866, (1-3), 15-20.

19. Quarti, C.; Fazzi, D.; Del Zoppo, M., A computational investigation on singlet and triplet exciton couplings in acene molecular crystals. Physical Chemistry Chemical Physics 2011, 13, (41), 18615-18625.

20. Merrick, J. P.; Moran, D.; Radom, L., An evaluation of harmonic vibrational frequency scale factors. Journal of Physical Chemistry A 2007, 111, (45), 11683-11700.