University of Southern California Dissertations and Theses

Two-photon absorption spectroscopy and excited state photochemistry of small molecules

(USC Thesis Other)

Two-photon absorption spectroscopy and excited state photochemistry of small molecules

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content TWO-PHOTON ABSORPTION SPECTROSCOPY AND EXCITED STATE PHOTOCHEMISTRY OF SMALL MOLECULES by Yuyuan Zhang A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy (CHEMISTRY) May 2012 Copyright 2012 Yuyuan Zhang ii Dedicated to my father Chengda Zhang and my mother Guizhen Rong iii Acknowledgments It was a life-altering experience to work at the Bradforth lab. I thank Prof. Steve Bradforth for giving me such an opportunity, and his patience and guidance throughout the years. Steve is not only a great teacher but also an exceptional mentor – from him I learned a lot of science, and most importantly, I get to learn how to become a good scientist. I am also very grateful to Steve for his role in helping me to secure a post- doctoral position, and I certainly will never forget his help over various other things: some are as important as how to give a better presentation (how to become a good “salesman” using Steve’s term – something I will never fully master), some are as trivial as lending me furniture. I would also like to express my gratitude towards all past and present group members of the Bradforth group. Dr. Askat Jailaubekov is the person who introduced me to the 2PA spectroscopy. Dr. Chris Rivera then helped me take the water 2PA spectrum, from which he also taught me a great deal of laser and optics related skills and knowledge. Later, this 2PA project was further advanced with the participation of Dr. Chris Elles. Dr. Piotr Pieniazek provided a unique aspect to this project due to his role as a theorist. I also enjoyed many conversations with him about science and other topics. I would also like to thank Dr. Diana Suffern for her insightful comments and suggestions, and her career advice. Dr. Xiyi Chen also gave me a lot of suggestions on career path as we shared a common background. Although Dr. Christi Schroeder and I have been working on different projects, but her encouragement throughout the years has kept me going. iv In the past two years, there have been a number of capable individuals joining the Bradforth group. I would like to thank Dr. Sean Roberts for sharing his knowledge and wisdom, not only on science but also on life in general. Saptaparna Das and Anirban Roy played an important role in collecting TCSPC data for thiophenol and phenol. Dr. Robert Seidel is kind enough to proofread my thesis (a fairly large package!). I thank the newest comers, Dr. Elsa Couderc and Konstantin Kudinov for their support and I wish them best of luck! I would like to give special thanks to our instrumentation director, Frank Niertit. He was very patient on numerous occasions when he explained to me some basic concepts on electronics, and he gave us “expedited services” when our computer, power meter, or anything with a power cord wasn’t behaving. Most importantly, he played a central role in designing and implementing the home built multichannel detection system, both Silicon and InGaAs arrays. I also like to thank the personnel at the Chemistry Main Office, Heather Connor, Michele Dea and Katie McKissick for their help in overcoming the bureaucratic barriers. The wonderful years in Los Angeles couldn’t be wonderful without the company of good friends. I would like to thank Arun and Vinita Sharma, Lewis and Liliana Gomez, Clement Do, Wei Wei and Luis Gomez, for all the great times spent together. Thanks to my friends from Seattle University, Huan Luong, Nelson Khov, Fareez Ismail, Alex Donald, Dustin Huard, Sabrina Mak, Ivon Octavia, Kevin Dong, Derek Yu, Ming Tan, v Eric Tan and James Ding – their humor, comfort and encouragement throughout the years have carried me through the rough patches. I very much enjoy the interactions with our collaborators from University of Bristol – Prof. Mike Ashfold and Dr. Tom Oliver. I am proud to be a part of the wonderful collaborative project on photodissociation of heteroaromatic molecules. Special thanks to Tom, for his careful explanation of gas phase behavior, and of course many intellectual contributions to the project. I am also grateful to have him as a friend, who taught me various British English idioms (via a TV show not to be named). I enjoyed working together with Tom while he visited our lab, despite the long hours and difficult experiments, and I certainly will not forget the papers we wrote together (and the ones yet to be written). I could not have survived any of this without the unconditional love and support from my beloved parents Chengda Zhang and Guizhen Rong. I’ve had ups and downs in the last ten years while I am away from home (more than 7000 miles!), but I know the persons who are always there with me are my parents. I cannot express my gratitude enough and dedicating this thesis to them is the least I can do. Lastly, I want to give my special thanks to Ning Huang. I love every moment I spent with her and looking forward to a lot more. Her patience, tolerance, understanding and love have given me strength during difficult times. vi Table of Contents Dedication ii Acknowledgments iii List of Tables ix List of Figures x Abstract xvii Chapter 1 General Introduction 1 1.1 Competing Dissociation and Ionization Dynamics 1 1.2 2-Photon Absorption 7 1.3 Thesis Plan 10 1.4 Chapter 1 References 12 Chapter 2 Polarization-dependent Broadband 2-Photon Absorption Spectroscopy Experimental Technique 15 2.1 Introduction 15 2.2 Optical Setup 18 2.3 Absolute 2PA Cross Section Calculation 23 2.4 Quantifying and Optimizing Experimental Parameters 28 2.4.1 Pump-probe Wavelengths 28 2.4.2 Pump Pulse Intensity 31 2.4.3 Sample Path Length 34 2.5 Instrument Sensitivity 36 2.6 Polarization-dependent 2PA Cross Section 37 2.7 Interfacing Detector and A/D converter for broadband detection 38 2.8 Summary 40 2.9 Chapter 2 References 42 Chapter 3 Two-Photon Absorption Spectroscopy of Alcohols and Alkanes 45 3.1 Introduction 45 3.2 Experimental 47 3.3 Results 49 3.3.1 Water and Alcohols 49 3.3.2 Alkanes 54 3.4 Discussion 55 3.4.1 Water and Alcohols 55 3.4.2 Alkanes 62 3.5 Conclusion 70 3.6 Chapter 3 References 72 vii Chapter 4 Linking Photochemistry in the Gas and Solution Phase: S-H bond Fission in p-Methylthiophenol Following UV Photoexcitation 76 4.1 Introduction 76 4.2 Experimental 84 4.3 Results 87 4.3.1 UV-Visible Spectra 4.3.2 Transient Absorption in Ethanol and Cyclohexane 89 4.3.2.1 p-MePhSH in Ethanol Solution 89 4.3.2.1.1 267 and 271 nm Excitation and Band Assignments 89 4.3.2.1.2 285 and 295 nm Excitation 98 4.3.2.1.3 200 nm Excitation 100 4.3.2.2 p-MePhSH in Cyclohexane Solution 102 4.3.2.2.1 267 nm Excitation 102 4.3.2.2.2 200 nm Excitation 105 4.4 Discussion 106 4.4.1 Gas Phase Behavior 107 4.4.2 Anisotropy 108 4.4.3 Geminate Recombination 110 4.5 Conclusion 115 4.6 Chapter 4 References 120 Chapter 5 Phenol Oxidation in the Gas and Condensed Phase Following UV Photoexcitation 125 5.1 Introduction 5.2 Experimental 133 5.2.1 Absorption Spectroscopy 133 5.2.1.1 Transient Absorption 133 5.2.1.2 UV-Visible Absorption 134 5.2.2 Emission Spectroscopy 135 5.2.2.1 TCSPC 135 5.2.2.2 Steady State Emission 135 5.2.3 Preparation of Deuterated Phenol (PhOD) 136 5.2.4 Theoretical Methods 137 5.3 Results 137 5.3.1 UV-Visible and Fluorescence Spectra 137 5.3.2 Time-Resolved Experiments 139 5.3.2.1 267 nm Excitation of Phenol Solutions 139 5.3.2.1.1 Phenol in Cyclohexane 139 5.3.2.1.2 Phenol in Water 144 5.3.2.1.3 Phenol in Ethanol 152 5.3.2.2 200 nm Excitation of Phenol Solutions 154 5.3.2.2.1 Phenol in Cyclohexane 154 5.3.2.2.2 Phenol in Water 156 viii 5.3.2.2.3 Phenol in Ethanol 159 5.4 Discussion 160 5.4.1 Fluorescence Lifetime 160 5.4.2 TA experiments 162 5.4.2.1 Excitation to S 1 163 5.4.2.1.1 Step-wise Reactions 164 5.4.2.1.2 Triplet State Dissociation 167 5.4.2.1.3 Ground State Dissociation 169 5.4.2.1.4 Tunneling Under the S 1 /S 2 CI 170 5.4.2.1.5 PCET 176 5.4.2.1.6 Summary of 267 nm excitation 183 5.4.2.2 Excitation to Higher Excited State(s) 184 5.5 Conclusion 187 5.6 Chapter 5 References 190 Chapter 6 Photochemistry of H-bonded Phenol Dimers in the Condensed Phase 196 6.1 Introduction 196 6.2 Experimental 196 6.3 Results 197 6.3.1 IR Analysis of O–H Stretch 197 6.3.2 Transient Absorption Experiments 202 6.4 Discussion 208 6.5 Conclusion 211 6.6 Chapter 6 References 212 Comprehensive Bibliography 214 Appendices 228 Appendix A Optical Setup for in-situ Determination of Sample Pathlength by Group Velocity Delay Method 228 Appendix B Math Reference for Calculating 2PA Cross Section 230 Appendix C MATLAB Script for 2PA Spectrum Processing 235 Appendix D Knife-edge Technique for Measuring Spot Sizes 260 Appendix E Origin of the Polarization Dependence 261 Appendix F Silicon Photodiode Array Interface and Parts List 265 Appendix G Schematics and Control Program for Silicon PDA timer 269 Appendix H InGaAs Photodiode Array Interface and Parts List 276 Appendix I Schematics and Control Program for InGaAs PDA timer 285 Appendix J Interface Operators Manual for InGaAs PDA 296 Appendix K 2PA Spectra of Chlorinated Alkanes 305 Appendix L Preliminary Results for 2PA Spectra of Benzene 307 Appendix M Evidence for Excited State Radical and Product Branching Ratio 308 ix List of Tables Table 2.1 Wavelength and energy ranges for various pump-probe continuum combinations 28 Table 3.1 Absolute 2PA cross section at 7.14 eV and 9.4 eV for water, methanol and ethanol 50 Table 3.2 1PA and 2PA absorption thresholds 52 Table 3.3 Linear polarization ratio at 7.92 eV (266 nm + 380 nm) for selected solvents 54 Table 4.1 Calculated CASPT2(9/8)/AVTZ excitation energies and CASSCF(9/8)/AVTZ TDMs for the p-MePhS radical 92 Table 4.2 Comparison of maximum possible (TKER max ) and average measured TKERs of products arising from excited state s−H bond fission in p-MePhSH at selected photolysis wavelengths 112 Table 4.3 Calculated energy of adduct structures at the CCSD(T)-F12/aug-cc-pVTZ level, and vertical transition energy of structures 1 and 2 at the EOM-CCSD/aug-cc-pVTZ level 114 Table 5.1 Quantum yield and rate of various deactivation channels following excitation to phenol S 1 state 161 Table E.1 F, G and H values for some selected polarization combinations 262 Table E.2 σ values and polarization ratios for some special cases 263 Table F.1 Parts list for broadband continuum detection system with Silicon PDA 266 Table H.1 Parts list for broadband continuum detection system with InGaAs PDA 282 x List of Figures Figure 1.1 Excited state molecule undergoes various pathways to “relax” or redistribute excess energy (not all pathways shown). 2 Figure 1.2 1-photon absorption spectra of water vapor (a) and liquid (b). 3 Figure 1.3 Illustration of 2PA “spike” affecting the kinetics of TA at early time. 8 Figure 2.1 Different experimental methods for 2PA measurements 17 Figure 2.2 Optical layout for 200 nm generation. 19 Figure 2.3 Top view of the sample region in a 2PA experiment involving a sample cell 22 Figure 2.4 Two data processing methods to calculate the absolute 2PA cross section spectrum (σ 2PA ). 26 Figure 2.5 2D contour plot around τ d = 0 region. 27 Figure 2.6 2D contour plot of 90 mM phenol in water measured with 400 nm pump/visible probe (a), and 850 mM phenol in water with 675 nm pump/UV probe (b). (c) and (d) are the 2PA spectra obtained from integration. 29 Figure 2.7 Spot size measurements of selected wavelengths within the broadband continuum probe 32 Figure 2.8 Absolute 2PA cross sections calculated with a single value of probe FWHM and with different probe sizes accounted for. 33 Figure 2.9 2PA spectrum of cyclohexane. 34 Figure 2.10 The peak 2PA absorption of C 60 in toluene solution as a function of cell path length. 35 Figure 2.11 Comparison of the 2PA spectrum of ~2.2 mM C 60 in toluene obtained in our study (red) and by Tahara (blue). 37 Figure 3.1 σ 2PA for water, methanol and ethanol at 4.6 eV (267 nm) pump and 6.2 eV (200 nm) pump 49 xi Figure 3.2 2PA spectra of water and selected alcohols and their corresponding polarization ratio 51 Figure 3.3 Relative 2PA spectra for cyclohexane and hexane isomers 55 Figure 3.4 1PA and 2PA spectra for water, methanol and ethanol, and HOMO/LUMO for water, methanol and ethanol 56 Figure 3.5 Simulation of 1PA and 2PA spectra for neat liquid methanol 58 Figure 3.6 Selected HOMOs and LUMOs of chair form of cyclohexane 63 Figure 3.7 Simulation of 1PA and 2PA spectra for neat liquid cyclohexane and molecular orbitals involved in these transitions 65 Figure 3.8 Selected HOMOs and LUMOs of all-trans form of n-hexane 67 Figure 3.9 Simulation of 1PA and 2PA spectra for neat liquid n-hexane and selected molecular orbitals of n-hexane 69 Figure 4.1 (a) CASPT2(10/10)/aug(S)-cc-pVTZ PECs in the S−H bond stretch of p-MePhSH for the ground state (S 0 ), 1 1 ππ* (S 1 ) and 1 πσ* (S 2 ) excited states. (b) side and (c) top projections of the TDMs associated with the first three singlet excitations in p-MePhSH; (d) TDMs for the three highest oscillator strength transitions from the ground state p-MePhS radical 82 Figure 4.2 UV absorption spectra for p-MePhSH in the vapor phase, cyclohexane and ethanol solution between 310 > λ > 210 nm 88 Figure 4.3 (a) TA spectra of 90 mM p-MePhSH in ethanol measured with 267 nm at selected pump-probe delay times. (b) Time profiles of major TA features measured at λ probe = 363, 500 and 600 nm. Inset: kinetics on an expanded scale, focusing on the early time. (c) Anisotropy spectra at selected pump-probe delay times. (d) Anisotropy decay as a function of time. Inset: anisotropy decay on an expanded scale, focusing on the early time 89 Figure 4.4 TA spectra of Figure 90 mM p-MePhSH in ethanol measured with 271 nm 96 xii Figure 4.5 (a) TA spectra of 90 mM p-MePhSH in ethanol measured with 285 nm. (b) Time profiles of major TA features measured at λ probe = 363, 500 and 600 nm. Inset: kinetics on an expanded scale, focusing on the early time 97 Figure 4.6 (a) TA spectra of 90 mM p-MePhSH in ethanol measured with 295 nm. (b) Time profiles of major TA features measured at λ probe = 363, 500 and 600 nm 98 Figure 4.7 (a) Anisotropy spectra at early delay time (200 fs), and (b) Anisotropy decay for p-MePhS( X % ) radical band centered at 500 nm 99 Figure 4.8 (a) TA spectra measured at different pump-probe time delays following excitation at λ pump = 200 nm. (b) Time profiles for major TA features measured at λ probe = 500, 465 and 385 nm 101 Figure 4.9 (a) 90 mM p-MePhSH in cyclohexane TA spectra measured with 267 nm at selected delay times. (b) 45 mM p-MePhSH in cyclohexane TA spectra measured with 200 nm. The calculated transition dipole moment squared (TDM 2 ) are also shown for transitions originating from S 1 (excited state absorption), as well as the p-MePhS radicals. (c, d) Geminate recombination analysis from the radical (red) and adduct (blue) kinetics after 267 nm and 200 nm photodissociation 103 Figure 4.10 Illustration of geminate recombination pathway leading to methyl substituted cyclohexadienone 114 Figure 5.1 “Un-relaxed” potential energy surfaces (PESs) along the O-H bond length coordinate for the ground state (S 0 ) and first four excited states (S 1 – S 4 ) 128 Figure 5.2 Energetic of phenol and its photoproducts for an isolated molecule and molecule in aqueous solution 130 Figure 5.3 UV-visible absorption and fluorescence spectra of phenol in cyclohexane, water and ethanol 138 xiii Figure 5.4 (a) 10 mM phenol in cyclohexane TA spectra at selected delay times, measured with 267 nm pump. (b) 10 mM phenol in cyclohexane TA spectra overlaid with the calculated EOM-CCSD/aug-cc-pVTZ transition dipole moments (TDM) for transitions originated from S 1 . (c) Kinetics at selected wavelengths obtained from the TA experiment and fluorescence life time obtained from TCSPC experiment 141 Figure 5.5 (a) 10 mM phenol in cyclohexane TA spectra at nanosecond delay times, measured with 267 nm pump. (b) Kinetics at selected wavelengths obtained from the TA experiment and fluorescence life time obtained from TCSPC experiment. (c) Back-to-back measurement of 10 mM PhOH and PhOD in cyclohexane obtained at 14 ns delay time 142 Figure 5.6 (a) 90 mM phenol in cyclohexane TA spectra at nanosecond delay times, measured with 267 nm pump. Inset: phenoxyl radical region at a expanded scale. (b) Kinetics at selected wavelengths obtained from the TA experiment. Inset: Comparison of the kinetics at 475 nm and the fluorescence lifetime obtained from TCSPC experiment 144 Figure 5.7 (a) 90 mM phenol in water TA spectra at the first nanosecond measured with 267 nm pump. (b) Kinetics at selected wavelengths obtained from the TA experiment 145 Figure 5.8 Kinetics of the 600 nm feature obtained in 90 mM phenol in water solution, 90 mM phenol + 0.17 M HCl, and 90 mM phenol + 0.50 M HCl, measured with 267 nm pump 146 Figure 5.9 (a) 90 mM phenol and 0.5 M CsCl in water TA spectra, measured with 267 nm pump. (b) Kinetics at selected wavelengths obtained from the TA experiment. (c, d) Comparison of phenol solution with and without CsCl for 475 nm kinetics, and 600 nm kinetics 147 Figure 5.10 Fluorescence lifetime of 0.1 mM phenol in various solvents 148 Figure 5.11 Side-by-side comparison of TA spectra of 90 mM phenol in water and phenol-OD in D 2 O, excite with 267 nm 150 Figure 5.12 Spectra of 90 mM phenol in aqueous solution with and without H + (electron scavenger) at 13 ns delay time 151 xiv Figure 5.13 (a) 90 mM phenol in ethanol TA spectra at selected time delays before 1 ns, measured with 267 nm pump. (b) Kinetics before 1 ns at selected wavelengths. (c) TA spectra at selected time delays from 1 to 14 ns. (d) Kinetics at selected wavelengths obtained from the TA experiment and fluorescence life time obtained from TCSPC experiment 153 Figure 5.14 (a) 10 mM phenol in cyclohexane TA spectra at selected time delays, measured with 200 nm pump. (b) Kinetics at selected wavelengths. (c) TA spectra at selected time delays for pure cyclohexane obtained in the same experimental condition. (d) Kinetics at selected wavelengths for pure cyclohexane 155 Figure 5.15 (a) 18 mM phenol in water TA spectra at selected time delays, measured with 200 nm pump. (b) Kinetics at selected wavelengths. Inset: Kinetics in the first 10 ps. (c) Early time spectra (t ≤ 1 ps) on an expanded scale. (d) Early time kinetics (t ≤ 1 ps) of 400 nm and 425 nm 157 Figure 5.16 18 mM phenol in ethanol TA spectra at selected time delays, measured with 200 nm. (b) Kinetics at selected wavelengths 159 Figure 6.1 IR absorption spectra of phenol in cyclohexane at various concentrations 198 Figure 6.2 Different structures of dimeric and polymeric phenol complexes and the corresponding IR signatures 199 Figure 6.3 (a) IR absorption spectra of phenol in cyclohexane, expanded to the O-H stretch region. Inset: expanded to show the H-bonded O–H stretch. (b) Spectra normalized to the 3617 cm -1 peak to emphasize the change in the shape of spectra as a function of concentration. (c) Absorption peak intensities as a function of phenol concentration. (d) Free OH fraction estimated from the free O–H stretch extinction coefficient 200 Figure 6.4 Reproduced from Figure 5.6: (a) 90 mM phenol in cyclohexane TA spectra at nanosecond delay times, measured with 267 nm pump. Inset: phenoxyl radical region at a expanded scale. (b) Kinetics at selected wavelengths obtained from the TA experiment. Inset: Comparison of the kinetics at 475 nm and the fluorescence lifetime obtained from TCSPC experiment 202 xv Figure 6.5 (a) TA spectra at selected delay times for 4.1 M benzene in cyclohexane, excited with 267 nm and probed with visible continuum. (b) Kinetics of the 510 nm feature for various concentrations of benzene in cyclohexane solutions and pure benzene 204 Figure 6.6 The kinetics of the phenol excimer band at λ probe = 600 nm, after 267 nm excitation of (a) cyclohexane solutions and (b) water solutions 205 Figure 6.7 (a) TA spectra measured at selected time delays following 200 nm photolysis of 45 mM phenol in cyclohexane, with the super continuum probe pulse polarisation aligned parallel to that of the excitation pulse. The thin lines are the TA signal for pure cyclohexane at 500 fs and 800 ps (same colour code as the solution) (b) Kinetics at selected wavelengths 207 Figure A.1 (a) Optical setup for in situ measurement of the thickness of the liquid film formed in the gravity jet. (b) Calibration with a standard 10.000 ± 0.002 mm quartz cell. (c) 267 nm delayed by the liquid sample in the gravity jet 229 Figure C.1 Flowchart for “CalTPACrossSection.m” 256 Figure C.2 Variable integration window defined by the user 257 Figure C.3 Convergence tests at various wavelengths across the probe spectrum 258 Figure C.4 Final output screen of “CalTPACrossSection.m” 259 Figure D.1 Illustration of the knife-edge technique 260 Figure E.1 Two-photon tensor and σ s 262 Figure F.1 Home-built timer for interfacing the 256-channel Silicon photodiode array with A/D converter 265 Figure H.1 Top layer of the home-built timer for InGaAs array 276 Figure H.2 Bottom layer of the home-built timer for InGaAs array 277 Figure H.3 Component layout for the timer board 278 xvi Figure H.4 Top layer of the home-built power supply for InGaAs array detection system 279 Figure H.5 Bottom layer of the home-built power supply for InGaAs array detection system 280 Figure H.6 Component layout for the timer board 281 Figure J.1 Illustration of data transfer process for a 256-channel detection in a 1 kHz system. 301 Figure J.2 Illustration of manipulating the exposure time indirectly by manipulating the data transfer time 302 Figure K.1 2PA spectra for (a) chloroform and (a) carbon tetrachloride (CCl 4 ) 306 Figure L.1 2PA spectra of benzene obtained with 3.1 eV (red curve) and 1.84 eV (blue curve) pump 307 Figure M.1 Illustration of pump-probe combinations for a 2-photon transition to an unidentified excited state (S n ). Spectra at t = 0 fs for p-MePhSH in ethanol, excited with three different wavelengths 309 Figure M.2 TA spectra of p-MePhSH in ethanol and cyclohexane at 0 ≤ t ≤ 200 fs, excited with 267 nm and probed by broadband Continuum 310 xvii Abstract Photochemistry occurring in a solution phase environment is often complicated by the solute-solvent interaction. Depending on how strongly the solvent interacts with the solute, the electronic structure of an excited state solute molecule can be very different from that in an isolated environment. This change in the electronic character of the excited state can turn on various competing deactivation processes which are otherwise absent in the gas phase. To decipher this complex condensed phase behavior, prior knowledge on the excited state electronic character is crucial. Two-photon absorption (2PA) spectra contain rich information on the excited state symmetry of a given molecule. The measurement of a polarization dependent broadband 2PA spectrum is realized by the pump-dispersed-probe technique, from which the zero-delay spectrum is extracted by removing the white light chirp. The experimental technique is described, and the 2PA spectra of several alcohols and liquid alkanes commonly used as solvents are presented for the first time. This provides information on the window of transparency for these liquids, and further, the symmetry of low lying electronically excited states. 2PA spectra are extremely valuable in deciphering broad and overlapping transitions usually observed in the condensed phase spectra. The ubiquitous nature of πσ* excited states has attracted much attention in the recent years. These dissociative states play an important role in the photochemistry of phenols, thiophenols and other heteroaromatic molecules, as they lead to heteroatom–H homolytic bond fission upon UV irradiation. There has been a surge of frequency-resolved studies xviii for this class of molecules, but as much as the nuances of the photodissociation reaction can be extracted from these gas phase experiments, important photochemical reactions of such molecules of biological importance occur in condensed environments, e.g., in water where strong solute-solvent interactions are expected. In a collaborative project, the ultrafast pump-probe technique is used to study the non-radiative deactivation pathways of phenol and thiophenol in solution upon deep-UV excitation. It is found that thiophenol molecules undergo prompt S-H bond fission exclusively in both cyclohexane and ethanol. In the case of phenol, gas phase-like bond fission is also observed in cyclohexane, but competing pathways such as autoionization and proton-coupled electron transfer come into play in aqueous solution. 1 Chapter 1. General Introduction 1.1 Competing Dissociation and Ionization Dynamics When a molecule is electronically excited upon absorption of one or more photon, the excited state molecule undergoes relaxation via various channels. † For the condensed phase this allows the system to dissipate the excess energy whereas in vacuum the molecule is only redistributing energy amongst different modes. The deactivation channels, for an isolated molecule, generally include non-radiative ones such as internal conversion (IC), intersystem crossing (ISC) and bond fission, and radiative ones such as fluorescence and phosphorescence (see Figure 1.1 for illustration). Which pathway is dominant depends on the nature of the excited state (and the photon energy as it determines which excited state is initially prepared), but among the above-mentioned pathways, bond fission is of photochemical importance as it produces new species in the system. In the past decade, the ubiquitous nature of the nσ* and πσ* excited states has been increasingly recognized. 1-3 These excited states are formed via electronic promotion from a non-bonding n orbital, or a bonding π orbital, to a valence anti-bonding σ* orbital. The repulsive character of such excited states thus suggests their role in bond fission mechanisms which eventually leads to photofragmentation. Moreover, the nσ* or πσ* state can interact and pre-dissociate a bound Rydberg state if the orbitals have the same symmetry and their overlap is large. Thus, at the vertical Franck Condon region (vFC), † For the condensed phase this allows the system to dissipate the excess energy whereas in vacuum the molecule is only redistributing energy amongst different modes. 2 the excited state can be spatially extended due to its Rydberg character, but as the bond length increases, it will acquire anti-bonding character and become more localized. 3,4 Figure 1.1 Excited state molecule undergoes various pathways to “relax” or redistribute excess energy (not all pathways shown). One of the well-known examples illustrating the Rydberg-valence (anti-bonding) interaction is the photodissociation of water. The broad and diffuse features at 7.4 and 9.8 eV observed in the 1-photon absorption spectrum (1PA) of water vapor originate from the A % 1 1 B 1 ← X % 1 1 A 1 and B % 2 1 A 1 ← X % 1 1 A 1 transitions, respectively (see Figure 1.2(a)). The lower orbitals are the two-fold oxygen 2p lone pair, but both transitions have a common upper orbital with mixed Rydberg 3s and anti-bonding σ* O-H character. Therefore, excitation to these two nσ* excited states leads to O–H bond fission and 3 eventually the generation of OH radial and H atom. 5-8 Larger molecules such as alcohols also exhibit similar O–H bond fission pathway when the lowest energy excited state is initially populated. 9 Bio-relevant heteroaromatic molecules such as indole, phenol and thiophenol have excited states with πσ* character (now the 2p orbitals of carbon are delocalized over the ring and form a π system), which is responsible for the X–H bond fission (X = N, O, S). 1-3 Figure 1.2 1-photon absorption spectra of water vapor (a) and liquid (b). Molecular orbitals involved in the corresponding transitions are shown to the right. The z-direction is defined along the C 2 axis, and the x-direction is defined perpendicular to the plane of molecule. This figure was taken from Ref 10. In the gas phase environment, the X–H bond fission in heteroaromatic molecules has been predicted by pioneering ab initio calculations 11 and demonstrated by gas phase photodissociation experiments such as H(Rydberg) Atom Photofragment Translational Spectroscopy (HRA-PTS) 2,3,12-15 and velocity map ion-imaging (VMI). 2,16 From these 4 experiments it is found that once the πσ* state is populated, molecules experience a dissociative potential energy surface (PES) resulting in X−H bond cleavage, a large release of kinetic energy mainly partitioned into translational energy of the light H atom, and vibrational excitation of the partner co-fragment. Of course, in the absence of collision with other particles, this information is preserved and enables the detection of total kinetic energy release (TKER). In the condensed phase environment, however, solvent provides several additional dynamical complexities to the collision free dynamics of the gas phase, and the particulars depend strongly on the specifics of the reaction, the solute-solvent coupling and the range of fluctuations in the solvent structure. Weakly interacting solvents are generally considered excellent mimics of a gas phase environment in terms of their effect on the molecular potential energy surfaces. Thus, exploring the photochemistry of solutes in inert solvents such as cyclohexane provides an instructive approach to see how dissociative reaction dynamics map into the condensed phase. However, such solvents still provide some dynamical complexities when compared to the collision-free environment. For example, collision between the nascent reaction products and the solvent molecules can alter the vibrational energy disposal, as solvent molecules provide an effective sink for vibrational energy relaxation (VER). 17-20 Similarly, the translational motion of products will be eventually stopped due to the friction exerted by the solvent. 21-23 Subsequent diffusion of the incompletely separated reaction products in the solvent can lead to geminate recombination. 17,24 Since the separation of the reaction products depends on the total kinetic energy available to the 5 system (after bond fission), a spherically symmetric model 25,26 or full molecular dynamic simulation 27 can connect the condensed phase experimental observables to those of the gas phase experiments. Moreover, the interaction with solvent molecules can change the location of the conical intersections (CIs), and affect, or even tune, how the excited state molecule approaches (i.e., the velocity and the angle) and thus branches through the CIs. 27 In strongly interacting solvents, such as methanol, ethanol and most importantly, water (where biological reactions take place), the location and the shape of the PESs can be dramatically changed from those of the isolated molecule predicted by ab initio calculations. The shift of the location of the PES, most significantly, can turn on new deactivation pathways which compete with the gas phase-like photodissociation. Photoionization is one such pathway – it takes place at much higher energy in the gas phase (typically 8 – 12 eV) as the electron has to be removed from the attractive cationic core, but in polar solvents the solvation of these charged species provides a large stabilization energy, and in turn makes it possible to ionize the solute at much lower excitation energy. For example, the vertical ionization energy (VIE) of liquid water is 11.2 eV, 28 but it is possible to ionize liquid water with excitation energy as little as 6 eV. 29 The photoionization with energy below the VIE involves complex mechanisms, including autoionization and proton-coupled electron transfer (PCET). The former occurs when the bound optically bright excited state (e.g., a Rydberg state) lies energetically close and couples strongly to the solvent continuum. That is to say with the aid of solvent 6 reorganization, it is possible to non-adiabatically eject an election to the solvent continuum even if the initial excited state of the system is bound. Often the radical cation produced on ionization subsequently deprotonates to yield a neutral radical and hydronium ion. PCET, on the other hand, refers to the mechanism in which a proton and an electron are transferred from a donor to an acceptor in a concerted fashion. 30 A part of this thesis focuses on the dynamical complexities of condensed phase heteroaromatic molecules, specifically, the competition between photodissociation and photoionization when the system is excited to above the ionization product asymptote but below the VIE. The gas phase photodissociation dynamics were extensively studied by our collaborator at the University of Bristol; direct measurements of internal and translational energy disposal, afforded by the frequency resolved HRA-PTS experiments, provide a well-defined reference point. Our femtosecond pump-probe experiments in solution phase, on the other hand, provide the time resolution to directly interrogate the appearance time and nascent energy release of the photoproducts (e.g., radicals, solvated electron etc) as well as their subsequent diffusive chemistry. Combining results from these two techniques, we can determine whether the gas phase reaction motifs are transferable to the various solvent environments. If so, we can explore to what extent the dynamic observables such as the translational recoil motion of the H atom, the energy disposal patterns into various electronic and vibrational degrees of freedom in the molecular radicals can be worked out and then connected back to the gas phase. On the 7 other hand, we will examine under what circumstances photoionization can be turned on, and most importantly, explore the mechanism that lead to electron ejection. 1.2 2-Photon Absorption The time resolution of our experiment is largely determined by the pump pulse duration. With 4-wave mixing technique in a hollow core fiber 31,32 and novel Gires-Tournois Interferometer dispersion mirrors, 33 pulse duration as short as 30 fs with very little higher order dispersion can be achieved for deep-UV pulses (267 nm). Although these are important aspects of our pump-probe apparatus which gives us superior instrument response time in the deep-UV regime, we need to stress that whether ultrafast kinetics at the sub-100 fs regime can be cleanly extracted depends on the amplitude of the so-called “coherent artifact”. The latter originates from the simultaneous and coherent absorption of two photons (1 pump + 1 probe) by a molecule, which occurs when the pump and probe pulses are temporally overlapped. Unfortunately, this “coherent artifact” usually gives large change in optical density (O.D.) which results in a “spike” at time t 0 – the width of this spike is determined by the convolution of the pump and probe pulses (thus remedied by the choice of shorter pulses), and the amplitude is determined by the probability that a molecule, either from the solute or solvent, absorbs two photons simultaneously – a quantity defined as the 2-photon absorption (2PA) cross section (σ 2PA ). See Figure 1.3 for illustration. 8 Figure 1.3 Illustration of 2PA “spike” affecting the kinetics of TA at early time. This set of data was obtained with a ~130 fs 200 nm pump pulse. The time points before the 2PA completely decays away do not contain valuable information as they are contaminated by the outsized 2PA signal. The term “coherent artifact” reflects the undesirable nature of such signal in the context of pump-probe experiments, but 2PA, as 1-photon absorption (1PA), is a general property of a molecule and thus cannot be avoided. On the other hand, 2PA provides rich information on excited state characteristics. As it is governed by different selection rules than 1PA, 2PA is able to access excited states which are otherwise “dark” in 1PA – a complementary relation reminiscent to that between IR and Raman spectroscopy for vibrations. Further, analogous to the Raman cross section, σ 2PA is polarization-dependent which can be used to infer the excited state symmetry – a property enunciated in a seminal work by McClain and co-workers. 34-37 Thus, the polarization ratio measurement is a very powerful tool for condensed phase spectroscopy as it can help in deciphering the broad and overlapping transitions. Lastly, the combination of a deep-UV photon and a UV-visible photon in 2PA will allow access to absorption bands in the vacuum-UV 9 region (of the 1PA spectrum), which is a tremendous advantage for condensed phase spectroscopy as the vacuum environment and light sources are not required. This provides a power tool in studying materials with large band gap, e.g., common solvents such as water, alcohols and alkanes. As powerful as 2PA spectroscopy seems, 2PA spectra are not measured routinely and only handful of research groups, e.g., Tahara, 38,39 van Stryland, 40,41 Prasad 42 and Rebane, 43-47 take advantage of the spectral information 2PA provided. The Rebane group use a traditional one-color (for example, focusing an intense 800 nm beam to the sample, thereby 2 x 800 nm to measure the 2PA cross section at 400 nm) scheme; spectra can be obtained by tuning the laser wavelength, but there will be a large independent error on the cross section associated with each wavelength as the apparatus is scanned and the approach is potentially time consuming. The Prasad group uses a novel spatially dispersed intense continuum to drive multiple one-color 2PA measurement simultaneously, but no polarization-dependent experiment can be carried out in this scheme. To the best of our knowledge, the Tahara and van Stryland groups are the only groups utilizing a pump-continuum-probe setup similar to ours to measure a 2PA spectrum. As a concluding remark, and to connect back to the theme of competing dissociation and ionization pathways in neat water, we will comment on the role of broadband 2PA in deciphering overlapping transitions in an otherwise featureless spectrum, and how the 10 electronic character of the optically excited state are intimately connected to the reaction dynamics that ensue. The 2-photon ionization yield of water is determined to be excitation energy-dependent – ionization products are observed at as low as 6 eV excitation energy, but dissociation plays a more important role in the low energy regime. The ionization yield dramatically increases and out-competes dissociation at > 8.5 eV with 2-photon excitation. So which excited state is responsible for the electron ejection? The 1PA spectrum of liquid water (Figure 1.2(b)) only shows one clear feature at ~8.2 eV which is assignable to the dissociative A % 1 1 B 1 state, but the location of B % 2 1 A 1 and the Rydberg states are unclear. Our polarization-dependent broadband 2PA measurements provide complementary information, as it revealed that transitions to three Rydberg states take place within the 9 – 10.5 eV range, which, as discussed before, are spatially extended and thus very likely to couple to the continuum (satisfied the energy requirement as well since they all lie within the conduction band – which is vertically accessed at ~11.2 eV). Thus, electron ejection can originate from autoionization from one or more of these Rydberg states (for details see Ref. 10 and references therein). 1.3 Thesis Plan Chapter 2 of this thesis will describe the experimental setup for the polarization- dependent broadband 2PA measurement. It aims to walk through the important aspects of the experiment in order to obtain a high quality 2PA spectrum. Chapter 3 will be built on the previously published 2PA spectrum of liquid water, and extends our analysis to the liquid alcohols and alkanes. It explores the excited state properties available to us when 11 comparing 1PA and 2PA spectra, combined with other information such as the 2PA polarization ratio and X-ray absorption results. The last two chapters will focus on reaction dynamics instead of spectroscopy, specifically, the competition between photodissociation and photoionization in the condensed phase. Chapter 4 presents the femtosecond pump-probe experiments on para-methylthiophenol, whereas chapter 5 mainly discusses the photochemistry of phenol. In both studies, the results from gas phase HRA-PTS experiments, carried out at the University of Bristol, are used as a reference point. The dissociation dynamics in cyclohexane is next examined to determine to what extent the gas phase reaction dynamics maps to such weakly interacting solvent. Lastly, photochemistry in ethanol (for para methylthiophenol), and water (for phenol) is examined in terms of competing dissociation and ionization channels. 12 1.4 Chapter 1 References 1. A. L. Sobolewski, W. Domcke, C. Dedonder-Lardeux and C. Jouvet, Phys. Chem. Chem. Phys. 2002, 4, 1093-1100. 2. M. N. R. Ashfold, B. Cronin, A. L. Devine, R. N. Dixon and M. G. D. Nix, Science 2006, 312, 1637-1640. 3. M. N. R. Ashfold, G. A. King, D. Murdock, M. G. D. Nix, T. A. A. Oliver and A. G. Sage, Phys. Chem. Chem. Phys. 2010, 12, 1218-1238. 4. H. Reisler and A. I. Krylov, Int. Rev. Phys. Chem. 2009, 28, 267-308. 5. V. Engel, V. Staemmler, R. L. Vanderwal, F. F. Crim, R. J. Sension, B. Hudson, P. Andresen, S. Hennig, K. Weide and R. Schinke, J. Phys. Chem. 1992, 96, 3201-3213. 6. R. Schinke, Photodissociation dynamics : spectroscopy and fragmentation of small polyatomic molecules. Cambridge University Press: Cambridge [England] ; New York, 1993. 7. S. A. Harich, D. W. H. Hwang, X. F. Yang, J. J. Lin, X. M. Yang and R. N. Dixon, J. Chem. Phys. 2000, 113, 10073-10090. 8. J. Underwood and C. Wittig, Chem. Phys. Lett. 2004, 386, 190-195. 9. J. B. Nee, M. Suto and L. C. Lee, Chem. Phys. 1985, 98, 147-155. 10. C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys. 2009, 130, 084501. 11. A. L. Sobolewski and W. Domcke, J. Phys. Chem. A 2001, 105, 9275-9283. 12. M. G. D. Nix, A. L. Devine, B. Cronin and M. N. R. Ashfold, Phys. Chem. Chem. Phys. 2006, 8, 2610-2618. 13. M. G. D. Nix, A. L. Devine, B. Cronin, R. N. Dixon and M. N. R. Ashfold, J. Chem. Phys. 2006, 125, 133318. 14. M. N. R. Ashfold, A. L. Devine, M. G. D. Nix and R. N. Dixon, J. Phys. Chem. A 2008, 112, 9563-9574. 15. A. L. Devine, M. G. D. Nix, R. N. Dixon and M. N. R. Ashfold, J. Phys. Chem. A 2008, 112, 9563-9574. 16. A. G. Sage, M. G. D. Nix and M. N. R. Ashfold, Chem. Phys. 2008, 347, 300-308. 13 17. J. Larsen, D. Madsen, J. A. Poulsen, T. D. Poulsen, S. R. Keiding and J. Thogersen, J. Chem. Phys. 2002, 116, 7997-8005. 18. C. G. Elles, M. J. Cox and F. F. Crim, J. Chem. Phys. 2004, 120, 6973-6979. 19. C. G. Elles and F. F. Crim, Annu. Rev. Phys. Chem. 2006, 57, 273-302. 20. C. Petersen, N. H. Dahl, S. K. Jensen, J. A. Poulsen, J. Thogersen and S. R. Keiding, J. Phys. Chem. A 2008, 112, 3339-3344. 21. C. L. Thomsen, D. Madsen, J. A. Poulsen, J. Thogersen, S. J. K. Jensen and S. R. Keiding, J. Chem. Phys. 2001, 115, 9361-9369. 22. A. Madsen, C. L. Thomsen, J. A. Poulsen, S. J. K. Jensen, J. Thogersen, S. R. Keiding and E. B. Krissinel, J. Phys. Chem. A 2003, 107, 3606-3611. 23. N. Winter and I. Benjamin, J. Chem. Phys. 2004, 121, 2253-2263. 24. C. G. Elles, M. J. Cox, G. L. Barnes and F. F. Crim, J. Phys. Chem. A 2004, 108, 10973-10979. 25. E. B. Krissinel and N. Agmon, J. Comp. Chem. 1996, 17, 1085- 1098. 26. J. A. Kloepfer, V. H. Vilchiz, V. A. Lenchenkov, A. C. Germaine and S. E. Bradforth, J. Chem. Phys. 2000, 113, 6288-6307. 27. C. A. Rivera, N. Winter, R. V. Harper, I. Benjamin and S. E. Bradforth, Phys. Chem. Chem. Phys. 2011, 13, 8269-8283. 28. B. Winter, R. Weber, W. Widdra, M. Dittmar, M. Faubel and I. V. Hertel, J. Phys. Chem. A 2004, 108, 2625-2632. 29. D. M. Bartels and R. A. Crowell, J. Phys. Chem. A 2000, 104, 3349-3355. 30. S. Y. Reece and D. G. Nocera, Annu. Rev. Biochem 2009, 78, 673-699. 31. C. G. Durfee, S. Backus, M. M. Murnane and H. C. Kapteyn, Opt. Lett. 1997, 22, 1565-1567. 32. A. E. Jailaubekov and S. E. Bradforth, Appl. Phys. Lett. 2005, 87, 021107. 33. C. A. Rivera, S. E. Bradforth and G. Tempea, Opt. Express 2010, 18, 18615-18624. 34. P. R. Monson and W. M. McClain, J. Chem. Phys. 1970, 53, 29-37. 35. W. M. McClain, J. Chem. Phys. 1971, 55, 2789-2796. 14 36. P. R. Monson and W. M. McClain, J. Chem. Phys. 1972, 56, 4817-4825. 37. W. M. McClain, Acc. Chem. Res. 1974, 7, 129-135. 38. S. Yamaguchi and T. Tahara, Chem. Phys. Lett. 2003, 376, 237-243. 39. S. Yamaguchi and T. Tahara, Chem. Phys. Lett. 2004, 390, 136-139. 40. R. A. Negres, J. M. Hales, D. J. Hagan and E. W. Van Stryland, IEEE J. Quantum Electron. 2002, 38, 1205-1216. 41. R. A. Negres, J. M. Hales, A. Kobyakov, D. J. Hagan and E. W. Van Stryland, Opt. Lett. 2002, 27, 270-272. 42. G. S. He, T. C. Lin and P. N. Prasad, Opt. Express 2002, 10, 566-574. 43. M. Drobizhev, A. Karotki, M. Kruk and A. Rebane, Chem. Phys. Lett. 2002, 355, 175-182. 44. M. Drobizhev, A. Karotki, M. Kruk, A. Rebane, E. Nickel, C. W. Spangler and N. Mamardashvili, P Soc Photo-Opt Ins 2003, 4797, 152-162. 45. A. Karotki, M. Drobizhev, M. Kruk, C. Spangler, E. Nickel, N. Mamardashvili and A. Rebane, J. Opt. Soc. Am. B: Opt. Phys. 2003, 20, 321-332. 46. H. A. Collins, M. Khurana, E. H. Moriyama, A. Mariampillai, E. Dahlstedt, M. Balaz, M. K. Kuimova, M. Drobizhev, V. X. D. Yang, D. Phillips, A. Rebane, B. C. Wilson and H. L. Anderson, Nat. Photonics 2008, 2, 420-424. 47. A. Rebane, N. S. Makarov, M. Drobizhev, B. Spangler, E. S. Tarter, B. D. Reeves, C. W. Spangler, F. Q. Meng and Z. Y. Suo, J. Phys. Chem. C 2008, 112, 7997-8004. 15 Chapter 2. Polarization-dependent Broadband Two-Photon Absorption Spectroscopy Experimental Technique 2.1 Introduction Two-photon absorption (2PA) is a completely general property of all molecules, e.g., small molecules, DNA bases, complex proteins and nanoparticles, and is comparable to one-photon absorption that is studied routinely with the aid of any commercial UV- Visible spectrometer. On the other hand, in comparison to linear absorption, the simultaneous absorption of two photons provides extra molecular information and technological advantages. First of all, one can access higher electronically excited states of a molecule, or the first excited state of a material with very large band gap. The advantage of 2PA will be apparent if the energy range of interest is in the vacuum-UV region, where absorption spectroscopy is difficult to carry out due to availability of light sources, detectors and the requirement of vacuum. Secondly, 2PA is governed by different selection rules, which means in highly symmetric molecular systems, an entirely different set of electronically excited states can be accessed and probed. The observation of these one-photon forbidden transitions provides invaluable molecular information complementary to the one-photon spectrum. This is analogous to the relationship between IR and Raman spectroscopy. Lastly, the 2PA cross-section is strongly polarization dependent, i.e., sensitive to the relative alignment of the electric fields of the two incoming photons. This aids spectral assignments for broad, overlapping and featureless bands, such as those usually observed in condensed phase spectroscopy. 16 Two photon absorption also has a wide array of applications in various fields, e.g., optical limited materials for laser safety purposes, 1,2 two-photon fluorescence microscopy, 3 three-dimensional optical data storage system 4-6 and microfabrication (utilizing high peak intensity requirement for two-photon initiated processes and thus achieve excellent spatial resolution which cannot be realized with one-photon process), 7,8 up-converted lasing (e.g., materials absorb at NIR then fluoresce and lase at visible), 9-11 and photodynamic therapy (e.g., a sensitizer producing singlet oxygen and thus becoming toxic to the targeted cell only upon photoexcitation by simultaneous absorbing two photons) 12,13 along with many more. The above applications rely heavily on the design, synthesis and characterization of the two-photon dye of interest. Although ab initio calculations can provide a more complete picture of the two-photon property of a particular molecule, computational limitation dictates that this approach is only suitable for very small molecules, such as water, 14 whereas most molecules with large 2PA cross sections have many more atoms and electrons. Currently, a large fraction of one-photon dyes are used for two-photon applications, 15 which spurs prediction of the 2PA cross section simply based on the linear spectroscopy measurements of the molecule. 16 Experimentally, the conventional methods for determining 2PA cross sections include open aperture z-scan 17-19 and 2-beam degenerate 20 measurement (similar to one-color pump-probe), which both are single-color experiments (see Figure 2.1). Thus, in order to obtain a spectrum from such single color technique, the laser wavelengths must be scanned, which will lead to large error 17 associated with each measurement and potentially very time consuming. To circumvent this problem, a multi-color z-scan method was developed – different wavelengths within an intense broadband continuum are spatially dispersed and focused onto a sample, and the attenuation of each individual color is monitored as a function of focal spot sizes. 21 However, this technique still provides no polarization information due to its 1-beam nature. Figure 2.1 Different experimental methods for 2PA measurements (see text for detail explanation of each method). Our current experimental setup measures 2-beam non-degenerate 2PA, and it is identical to the pump-broadband probe setup. The intense pump pulse can be higher harmonics of 800 nm, such as 400, 267 and 200, or outputs from an Optical Parametric Amplifier (OPA), which covers a broad range of deep-UV, visible and NIR. The weak broadband probe covers 300 – 700 nm (detected by silicon array) or 650 – 1400 nm (detected by InGaAs array). Here, our current experimental setup, including optical layout and timing circuit (synchronizing laser, chopper and diode array timing) design, and post-experiment 18 data analysis for polarized broadband 2PA spectroscopy will be presented. The theoretical background for the polarization dependent cross section will be introduced as well. 2.2 Optical Setup The 800 nm fundamental wavelength was provided by a 1 kHz Ti:Sapphire regenerative amplifier (Spectra Physics Hurricane: 1 W, Coherent Legend Elite: 3.6 W). In the current set up with the Legend system, the generation of the intense pump pulses is presented in Figure 2.2. For 2 nd , 3 rd and 4 th harmonics of 800 nm, ~560 μJ of the fundamental was gently brought to focusing condition by a f = 2 m lens, and a 500 μm type-I β-barium borate (BBO) crystal (Red Optronics) was placed in front of the focus to generate 400 nm. The resulting pulse can be used directly as the pump pulse for the experiment, or used to subsequently generate the 267 nm. A 150 μm type-II BBO crystal (Red Optronics) was placed further downstream, where the 400 nm and part of the residual 800 nm was sum- frequency mixed to give the 267 nm pulse. This pulse has a bandwidth of 2 nm and a maximum pulse energy of ~8 μJ, and can be used either directly as the pump pulse, or generating the 200 nm via sum frequency mixing with the residual 800 nm (~60 μJ) at a 75 μm-thick, type-I BBO crystal (Red Optronics) which is carefully placed at the focus of the 267 and 800 nm for maximum conversion efficiency. 2μJ of the 200 nm with a bandwidth of ~1 nm can be generated and used for the 2PA experiment. The deconvoluted temporal widths of the 267 and 200 nm are ~100 and ~130 fs, respectively, as determined by cross-correlation with the continuum. 19 Figure 2.2 Optical layout for 200 nm generation. BS: beam splitter, HR: high reflector, 800 nm T/400 nm R: dichroic mirror to transmit 800 nm and reflect 400 nm, TS: translation stage to compensate difference in travelling path lengths. Alternatively, the 3 rd harmonic wavelength can be generated via four-wave mixing (4WM) in a hollow core fiber. 22 The resulting pulses had a pulse energy > 2.0 μJ (16 μJ of 400 nm and 33 μJ of 800 nm input to a ~30 cm long fiber with 75 μm opening), and a generated bandwidth of 4-5 nm. This light was subsequently compressed by a pair of Gires-Tournois Interferometer dispersion mirrors, 23 generating a 30 fs (FWHM) pulse with very little higher order dispersion as determined by auto-correlation, or it can be compressed by a conventional prism-pair compressor. The merit of utilizing short pulses in a 2PA experiment is to reduce interference of slowly rising pump-probe (i.e., time- delayed) signal, such as solvated electron absorption that spans the whole probing region. Since the 2PA coherent “spike” in the time domain is the convolution between the two Gaussian pulses, shorter pulses result in a narrow spike which allows clear separation of the solvated electron signal rising on a time scale of ~500 fs. Thus, there are advantages to using short pulses as the “pump”, but the practical issue is that, with our current 4WM 20 setup, its stability (~5% noise) is inferior to UV generated from a crystal and thus results in more noise in the 2PA spectral domain, i.e., the spectra obtained from 4WM was not as smooth as those from crystals. Moreover, shorter pulse is more likely to induce cross phase modulation (XPM), which will skew the shape of 2PA signal in the time domain and thereby complicates the analysis (see Figure 2.5 and its associated text in Section 2.3) Wavelengths other than harmonics of the 800 nm were realized by a two-stage OPA using a type-II phase matching process (OPA-800C, Spectra-Physics). The resulting signal and idler spanned from ~1.1 to ~2.9 μm, with maximum pulse energy of ~ 25 μJ (using 340 μJ of 800 nm input). Although it is entirely possible to use these wavelengths directly as the pump, 2PA experiments with a NIR pump have only recently been demonstrated in our laboratory for diphenyltetracene (data not shown). Such experiments have been reported in the literature by the Tahara group for C 60 24 and retinal 25 . The signal or the idler pulses can be either frequency-doubled to 600 – 1200 nm, or frequency mixed with 800 nm to give 480 – 600 nm. The latter can be further frequency-doubled to the deep-UV region. In the next chapter, 2PA experiment with 675 nm pump wavelength is demonstrated with several DNA bases. Also noteworthy is that the OPA output can be used as the seed of the aforementioned 4WM process to generate deep-UV from ~220 – 245 nm 22 and thus provides extra flexibility to access higher energy 2PA spectral regions. The weak broadband probe, covering wavelengths from 300 to 700 nm, was produced by focusing a small fraction of 800 nm onto a rotating (to avoid burning) 2 mm thick 21 calcium fluoride window (Koch Crystal Finishing). A pair of aluminum-coated off-axis parabolic mirrors (Janos Technology) was used to collimate then focus the continuum, in order to minimize chromatic aberration and temporal chirp induced by transmissive optics. Alternatively, the continuum was generated from a 3 mm thick sapphire window, which spanned further into the NIR region (up to ~1400 nm), at the expense of the short wavelength region. The shortest wavelength produced with sapphire is ~ 400 nm (cf. 300 nm by CaF 2 ). The relative polarization between the pump and probe pulse was controlled by rotating the 800 nm polarization prior to continuum generation with an air-spaced zero-order half waveplate (Karl Lambrecht Corporation). With this method, the need to obtain zero-order half waveplate for every pump wavelength is eliminated. The liquid sample was flowed through by a recirculating wire-guided gravity jet; 26 the 100 μm by 3 mm rectangular nozzle opening produced a liquid film no thicker than 100 μm, which minimizes the group velocity walk-off between the deep-UV pump and the broadband continuum, and eliminates the significant 2PA and cross-phase modulation signals from cell windows. The exact path length of the sample varies – it depends on the distance from the nozzle opening and the liquid property, such as density and viscosity. 26 Due to the fact that the determination of 2PA cross section requires an accurate determination of the sample path length (see section 2.3), it is necessary to measure the path length on a routine basis. Appendix A presents the optical setup for in situ determination of sample path length by group velocity delay method. As a test case, the thickness of the water film was determined to be ~50 ± 10 μm. 22 Alternatively, a conventional static quartz cell (Starna) was used when longer path length was desired. It was found that a 1 mm cell gave the best results. Since the pump and probe beams overlapped at the sample with a small angle (usually ~10˚), a short path length (e.g.,100 μm) caused substantial signal from the cell wall (see Figure 2.3). While a longer path length (e.g., 1 cm) eliminated this problem, the effective path length is difficult to ascertain due to the varying spot size, crossing angle of the two beams, and significant temporal walk-off between the pump and probe. Figure 2.3 Top view of the sample region in a 2PA experiment involving a sample cell. The pump beam is usually more gently focused to a bigger spot, whereas the continuum probe is more tightly focused. The two beams usually intersect one another with crossing angle of ~10˚. If the cell path length is too short, the pump and probe is still tightly focused and interacting at the front and back of the cell wall, and thus produce signals (green line denotes cell wall). On the other hand, if the cell path length is too long, the effective path length is shorter as the pathlength in this case is determined by the interaction volume of the two crossing beams (red line denotes cell wall in this case). The attenuation of each wavelength within the continuum probe was historically presented in the form of transmittance (T), due to the fact that the pump intensity dependent transmission is measured in the z-scan method. In our broadband study, 23 however, the probe attenuation was represented by the change in absorbance (ΔA 2PA ), calculated from the equation below:         = Δ on pump off pump 2 log I I A PA (1) where I pump on and I pump off were the probe intensities when the pump was going through the chopper aperture and when it was being blocked by the chopper blade, respectively. This was realized by the utilization of chopper - every other shot of the pump pulses was blocked by a chopper operating at, and phase synchronized to, half of the laser repetition rate (in our case, chopper frequency was set at 500 Hz as the laser repetition rate was 1 kHz). To monitor the intensity of each probe wavelength, after passing through the sample, the broadband continuum was dispersed and projected onto a 256-channel Silicon (for UV-Visible detection, S3901-256Q sensor mounted on C7884-20 detector head, Hamamatsu) or InGaAs (for NIR detection, G9213-256S sensor mounted on C8061-01 detector head) photodiode array. The latter was synchronized to both the laser and the chopper – see Section 2.7 for detail. 2.3 Absolute 2PA Cross Section Calculation In order to discuss the optimal set of experimental parameters for 2PA measurements, it is essential to first understand the key variables involved in the calculation of 2PA cross section. It is noteworthy that, even in experiments where the absolute cross section is not required, negligence in choosing a correct set of parameters can result in the distortion of 24 the spectrum. The change in absorbance due to 2PA (see equation (1)) can be written as a function of the following variables. ( ) 2 2 2 2 exp 2 1 ) 10 ln( 1 cc d cc P PA L E f A τ τ τ π β − ⋅ ⋅ ⋅ ⋅ ⋅ = Δ (2) where 2PA coefficient β is in cm/W, pump energy E p is in Joules, path length L is in cm, cross correlation width τ cc is in s -1 , τ d is the delay between the pump and probe pulse, and f is a constant factor accounting for the beam size difference between the pump and probe, which is defined as the equation below and has a unit of cm -2 . ( ) ( ) 2 , 2 , 2 , 2 , 2 1 y probe y pump x probe x pump s s s s f + ⋅ + = π (3) where s i,j again is the half width at e -1/2 of the maximum value, in the horizontal (x) or vertical (y) direction, s i,j can be converted to the more commonly used full width at half maximum (FWHM) via s s FWHM ⋅ = ⋅ = 355 . 2 ) 2 ln( 2 2 . The full derivation of equation (2) is shown in Appendix B. From here, there are two approaches to calculate the β value from ΔA 2PA . The more intuitive approach is to take the value at τ d = 0 (refer to as time zero cut method). From equation (1) it is apparent that the maximum attenuation of the probe and thus the maximum value of ΔA 2PA occur at zero time delay between to pump and probe. Since ( ) 1 2 exp 2 2 = − cc d τ τ in this case, 2PA coefficient β is: ) 0 ( 2 ) 10 ln( 2 = Δ ⋅ ⋅ = d PA P cc A L fE τ τ π β (4) 25 Another approach is to integrate over the delay time τ d (referred to as the integration method): ( ) ) 10 ln( 2 2 1 ) 10 ln( 1 2 exp 2 1 ) 10 ln( 1 2 2 2 L E f L E f d L E f d A P cc cc P d cc d cc P d PA β τ π τ π β τ τ τ τ π β τ = ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ = − ⋅ ⋅ ⋅ ⋅ ⋅ = Δ ∫ ∫ +∞ ∞ − +∞ ∞ − After rearranging one can obtain: d PA P d A L fE τ β ∫ +∞ ∞ − Δ ⋅ = 2 ) 10 ln( (5) It is important to point out that mathematically the integration of ΔA 2PA in the delay time domain should be carried out from -∞ to +∞, but this is unnecessary (and unrealistic) in practice due to finite width of the pulses, and their overlap is infinitely small at these limits. It is apparent that the integration method is simpler compared to the time zero cut method, since the former does not require the knowledge of the pulse duration of the pump, which is often not a trivial task to obtain in practice (need to perform autocorrelation). Further, the time zero method requires fitting the 2PA “spike” to a Gaussian function, from which the peak height can be determined. At low signal level, however, the poor fitting quality results in large noise level in the 2PA spectrum (see Figure 2.4). 26 Figure 2.4 Two data processing methods to calculate the absolute 2PA cross section spectrum (σ 2PA ). Left: 2D contour plot for water 2PA around time zero region. The dotted lines represent the lower and upper integration limits in the time domain; the solid line represents zero delay between the pump and probe (τ = 0). Graph reproduced from Ref. 14. Right: water 2PA spectrum (267 nm pump + UV probe) obtained with integrating (red circles) over the limits defined on the left, and taking the maximum value along zero delay (blue line). With the same set of experimental parameters, the two methods produce the same result when the signal-to-noise is high. The time zero method, however, failed at the low signal region (low excitation energy – longer probe wavelength) due to poor fitting quality. A more fundamental reason why the former is preferred is that the integration over the delay time domain eliminates the interfering signal from cross phase modulation (XPM), which arises from change in the probe continuum frequency in the presence of an intense pump. The modification of the ΔA signal due to XPM is thoroughly studied by Kovalenko and co-workers 27,28 and Lorenc and co-workers 29 , both groups demonstrated that the ΔA purely caused by XPM (non-resonant signal) can be integrated to zero in the time domain (see Figure 2.5). Therefore, the integration method (Equation 4) is a more rigorous approach in calculating 2PA coefficients, whereas simply taking the peak value at τ d = 0 (Equation 3) may overestimate the 2PA signal due to the contribution from XPM. 27 Figure 2.5 2D contour plot around τ d = 0 region. Left graph: water 2PA spectrum obtained with ~40 fs duration 267 nm pulse (generated via 4WM); left inset: time domain data for 350 nm, right inset: time domain data for 500 nm. The signature for XPM at short probe wavelength is a spike with two negative wings on the side (the small signal at the earliest time is an experimental artifact), and that for longer probe wavelength is a negative signal followed by a positive signal with equal intensity. Right graph: water 2PA spectrum obtained with ~110 fs duration 267 nm pulse (generated via frequency mixing in BBO); left inset: time domain data for 350 nm, right inset: time domain data for 500 nm. With a longer duration pulse (and thus lower peak intensity), the 2PA “spike” is Gaussian-like, and in the region exhibiting no 2PA, non-resonant signal is not observed. The absolute 2PA cross section σ 2PA , in units of Göppert-Mayer (1 GM = 10 -50 cm 4 s photon -1 molecule -1 ), can be obtained from 2PA coefficient β, in units of cm/W via the following relation: β ω σ ⋅ = n pump PA h 2 (6) where ω pump is the frequency of the pump pulse and n is the number density of the system of interest (related to concentration). It is important to bear in mind that the 2PA coefficient β depends on the number of molecules present in the system, whereas the 2PA cross section does not, which is obvious from its unit. The complete MATLAB script for 28 calculating the absolute 2PA cross section (σ 2PA ) via integration and/or time zero cut methods is presented in Appendix C. 2.4 Quantifying and Optimizing Experimental Parameters From Equations (3) – (6) we can see that in order to calculate the 2PA coefficient and absolute cross section correctly, four parameters are required: pump pulse energy, sample path length, and the pump and probe spot size. In addition, a cross correlation width is also required if the time zero method is used. Table 2.1 Wavelength and energy ranges for various pump-probe continuum combinations. The UV-pumps are harmonics of the 800 nm amplifier fundamental, 1400 nm is the most intense signal pulse generated from the OPA. The broadband UV-Visible probe are generated from a rotating CaF 2 crystal and detected by a Si photodiode array, whereas the NIR probe are generated from a stationary sapphire crystal and detected by an InGaAs diode array. Broadband Probe Region / nm UV-Vis (300 – 700) NIR (700 – 1400) λ pump / nm One-photon Equivalent Wavelength / nm (Total Energy / eV) 200 120 – 156 (10.3 – 8.0) 156 – 175 (8.0 – 7.1) 267 141 – 193 (8.7 – 6.4) 193 – 224 (6.4 – 5.5) 400 171 – 254 (7.2 – 4.9) 254 – 311 (4.9 – 4.0) 800 218 – 373 (5.7 – 3.3) 373 – 509 (3.3 – 2.4) 1400 247 – 467 (2.7 – 5.0) 467 – 700 (2.7 – 1.7) 2.4.1. Pump-probe Wavelengths Since 2PA is a coherent non-linear process, i.e., the absorption of the pump and probe photons is simultaneous, the term “pump-probe” only refers to the relative intensity of the two incoming beam instead of the order of their absorption. For example, either the combination of 267 nm pump + 800 nm probe or 800 nm pump + 267 nm probe can 29 access 2PA at 200 nm. The combinations of the intense pump and the broadband probe for commonly investigated 2P energy ranges are summarized in Table 2.1. Figure 2.6 2D contour plot of 90 mM phenol in water measured with 400 nm pump/visible probe (a), and 850 mM phenol in water with 675 nm pump/UV probe (b). (c) and (d) are the 2PA spectra obtained from integration. The region of interest is in the 1PA equivalent wavelength range from 200 – 290 nm (where there is strong UV absorption in 1PA). The Raman signal in the 400 nm pump experiments resides in the region of interest, whereas the 675 nm pump shifts the Raman peak to a region with no 1PA absorption (1PA shown in orange trace for ease of comparison). From Table 2.1 it is apparent that the broadband UV-Visible probe can access a wider energy region (~2.3 eV) compared to that of the broadband NIR probe (~1.0 eV). 30 Therefore, it is more desirable to use the UV-Visible probe for survey scans, and the NIR probe if higher spectral resolution is needed (typically as 256 element array detector is used for both probe ranges). However, given that most 2PA spectra, at least in polar solvents, are broad (see Chapter 3 for example), the spectral resolution afforded by the UV-Visible probe should be sufficient. Another factor one must consider when choosing the pump-probe combination is the Raman response which also contributes to the coherent signal scattered in the direction of the probe. The discussion of the origin of the Raman signal is outside the scope of this work, but in practice, this additional signal results in spectral peaks with large intensity (and widths determined by the bandwidth of the pump pulse), which can overwhelm the 2PA signal we seek. Further, it is found that when the intense pump pulse is within the wavelength range of the broadband probe, the Raman signal will reside in the probe wavelength range and thereby interfere with the 2PA spectrum. Thus, it is necessary to choose the optimal pump-probe combination such that the Raman signal is outside of the investigated wavelength region. Figure 2.6 illustrate such choice is important in some experiments – for investigating excited states in the deep UV region (200 < λ < 300 nm) where most small heteroaromatic molecules absorbs, we can choose 400 nm pump and the visible part of the CaF 2 probe continuum, or a 675 nm pump (generated from the OPA) and the UV part of the same continuum. The latter is a superior choice in these cases since the Raman peak is out of the region of interest, thereby providing a cleaner 2PA spectrum. From Figure 2.6 it is also apparent that the Raman signal is much stronger 31 than the 2PA signal but its narrow bandwidth nature allows immediate identification of the undesirable contribution. 2.4.2 Pump Pulse Intensity The choice of pump pulse energy is important in order to avoid the generation of the photoproduct. In the case of water and alcohols, the generation of photoproducts, which are usually solvated electrons (plus the ionized solvent cation), are often resulted from absorption of two pump photons (a probable process if tightly focused 200 or 266 nm pump pulses are used). Thus, when choosing the pump pulse energy, its spot size also needs to be considered carefully. In the case where pump pulse energy is abundant, e.g., 800, 400 and 267 nm, the spot size can be as large as possible and the pump pulse energy can be attenuated to slightly below the threshold of generating photoproducts. However, in some cases, the pump energy is not sufficient to produce a good 2PA spectrum with a large spot size. Care must be taken to characterize the spot size of the pump and probe as equation (2) – (6) shows that the 2PA coefficient and cross section are a function of these variables. Equation (2) takes into account the relative beam size of the pump and probe. One of the technical difficulties of the broadband 2PA measurement is the determination of the probe beam size for each individual color. Although a pair of parabolic mirrors is used instead of transmissive lens, chromatic aberration is still difficult to avoid, e.g., different colors being focused to different positions (x, y and z direction). Figure 2.6 presents the 32 spot sizes of the broadband continuum (310 < λ probe < 650 nm) measured with knife edge technique (see Appendix D for detail). From Figure 2.7 (a) we can see that the centers of the selected wavelengths do not always overlap with each other – it can be shifted as much as 100 μm across the probe spectrum. Moreover, from Figure 2.6 (a) it is apparent that the shorter wavelengths have larger spot sizes, and the position of the focus is slightly off in the horizontal direction. Figure 2.7 Spot size measurements of selected wavelengths within the broadband continuum probe. Left: raw data obtained from knife-edge technique (Appendix D). Right: the FWHM of each wavelength obtained by fitting with a complimentary error function (Appendix D). In the case where the FWHM of the pump beam is much larger than that of the probe, the difference between each probe wavelength leads to a very small difference in the constant factor f (equation 3). Thus, using a single FWHM value of the probe will not result in significant error when calculating the 2PA cross section. On the other hand, if the pump size is approximately the same with the probe size (common pump spot size used in a transient absorption experiment is ~200 μm), then the measurement of FWHM of each probe wavelength is required, in order to obtain an accurate 2PA cross section. Figure 2.8 33 shows the 2PA spectrum of water obtained with 267 nm pump (FWHM = 256 μm) and UV probe (FWHM = 53 – 226 μm, shorter wavelengths have larger spot size). If we were to calculate the 2PA cross section with a single probe FWHM value, 53 μm, it can lead to significant error in cross section in the short wavelength range. Figure 2.8 Absolute 2PA cross sections calculated with a single value of probe FWHM (red), and with different probe sizes accounted for (blue). The spectrum was taken for neat water, with 267 nm pump and UV probe. The spot size of the pump, as measured by the knife edge technique, is 256 μm. The spot sizes of the probe vary from 53 μm at the red end, to 226 μm at the blue end. From the above example we can see that it is much more convenient to place the sample at the focus of the probe and use a pump spot size as large as possible, in order to use a single FWHM value of the probe spot size to calculate the 2PA cross section. With this condition, the error in the calculation is negligible without taking into account the spot sizes for all probe wavelengths. It is noteworthy that in many cases where absolute cross sections are not measured, care still must be taken when choosing the pump and probe spot sizes. From numerous measurements of the continuum spot sizes, it is found that the spot sizes at the blue end (close to 300 nm) are generally larger than those at the red end (close to 700 nm). If the sample was not carefully placed at the focusing condition of the 34 probe, it is very likely that the variation in spot sizes between wavelengths is even larger than those shown in Figure 2.7. This will result in spectral distortion which is commonly seen in the short wavelength region of the probe (and thereby the high energy region of the 2PA spectrum), as illustrated in Figure 2.9. Figure 2.9. 2PA spectrum of cyclohexane, measured with 267 nm pump and UV probe. The red curve exhibits a “peak” at ~8.5 eV, which is due to poor pump-probe overlap at the short wavelength region (the probe spot size is larger than the pump), whereas the blue curve shows no such feature. The latter is obtained after increasing the spot size of the pump beam. 2.4.3 Sample Path Length As discussed in the previous section, the path length of the cell has to be carefully selected, since it is necessary to reduce 2PA signal from the cell wall on the one hand, while the effective path length needs to be taken into account on the other hand. The interaction length can be estimated from the overlapping volume between the pump and probe beams. Assuming a constant beam profile across the path length (constant FWHM), the interaction length as a function of crossing angle (θ) can be expressed as ( ) θ sin pump FWHM . Therefore, the full interaction length for a typical pump-probe experiment (FWHM pump = 200 μm, θ = 10˚) is ~1.1 mm, and for a typical 2PA 35 experiment (FWHM pump = 400 μm, θ = 10˚) is ~2.2 mm. The non-linearity in signal with path length because of the effective path is illustrated in Figure 2.10, where the ΔOD 2PA signal of C 60 in toluene solution is plotted as a function of cell path lengths. Figure 2.10 The peak 2PA absorption of C 60 in toluene solution as a function of cell path length. The ΔOD 2PA signals were normalized according to concentration and pump intensity. The peak of the absorption band is at 315 nm, as determined by Tahara. 24 All experiments were performed with 160 μm pump spot size (FWHM). It is noteworthy that temporal walk-off between the ultrashort pump and probe in the sample also reduces the effective interaction length. Thus, even without the geometrical constraint described above, the 2PA signal cannot just be increased by making the sample path longer. Further, the effective interaction length varies for the different colors in the continuum. The walk-off is largest when the pump and probe wavelengths are very different, 25,30 e.g., between a 100 fs 266 nm pump and the reddest colors in the continuum, the effective interaction length is ~200 μm. 26 The variation of effective interaction length as a function of probe wavelengths is avoided by the use of paths shorter than the 36 smallest effective interaction length, thus justifying our normal choice to use a < 100 μm thin film. However, for characterizing electronic spectra with smaller energy gaps, use of 800 nm or NIR pump and the visible/NIR continuum means that at least a 4 times longer path can safely be used as the temporal walk-off between these wavelengths are essentially small. If the spectral chirp is carefully characterized, van Stryland has shown an analytical correction to account for the temporal walk-off between the pump and probe, 30,31 which will not be repeated here. 2.5 Instrument Sensitivity Combining equation (3) – (6) with a minimum detectable ΔA of 0.1 mOD in our current transient absorption apparatus, empirically we have established that to conservatively detect σ 2PA ·[solute] = 2 GM·mol/L, the required combination of pump intensity and path length (I pump ·L) is ~ (10 11 W·cm -2 )(0.1 mm). If temporal walk-off is negligible and thus the use of thicker cell (0.1 – 1 mm) yields correspondingly larger signals, the instrument sensitivity and dynamic range can be increased dramatically. Therefore, the current apparatus in our laboratory for 2PA spectroscopy has sensitivity sufficient to observe weak (σ 2PA < 0.05 GM) 2PA from solvents and 2PA from solutes with σ 2PA ~ 20 GM at 0.1 M concentrations. For the longer λ chromophore-containing molecules (nanocrystals, dyes, nanotubes, DNA bases), path lengths can be closer to 1 mm and similar signal to noise can be expected at mM concentrations. 24 In fact, we have successfully measured 2PA spectrum (with σ 2PA in GM) for mM concentration of C 60 in toluene – both the spectrum and σ 2PA value is in good accord with those obtained by the Tahara group (see 37 Figure 2.11 for detail). 24 Two photon cross sections for a variety of chromophores ranging from GFP to perylene derivatives show peak σ 2PA from 10 – 4000 GM. 24,25,31-33 Figure 2.11 Comparison of the 2PA spectrum of ~2.2 mM C 60 in toluene obtained in our study (red) and by Tahara (blue). Our study: 1.2 μJ 800 nm pump with FWHM = 160 μm, visible probe, 1 mm quartz cell, 2.3 mM C 60 in toluene solution. 2.6 Polarization-dependent 2PA Cross Section Polarization dependent 2PA spectroscopy is a powerful tool to investigate the electronic symmetry of a molecular excited state wavefunction. The origin of the polarization dependence has been discussed in detail by the seminal work from McClain, 34-37 and it is summarized in Appendix E. To obtain such information, the relative polarization between the two incoming photons must be manipulated. It has been theoretically shown 38 and experimentally demonstrated 39 that 2PA measured with circularly polarized light is useful to study molecules with a chiral center. Although in principle circularly polarized light can be easily achieved in our laboratory (by a ¼ - waveplate) and it is of our interest to obtain such broadband 2PA spectra for proteins and other molecules with a chiral center 38 for pharmaceutical purposes, so far we have only measured linearly polarized 2PA spectra for various molecules. The parallel and perpendicular 2PA spectra are typically obtained from back-to-back measurements where the polarization relation between the pump and probe beam is changed by rotating the polarization axis of the 800 nm prior to the continuum generation. Then the linear polarization ratio can be computed from the following equation: ∫ ∫ Δ Δ = = τ τ τ τ σ σ d A d A R perp PA para PA perp para ) ( ) ( , 2 , 2 (7) The polarization purity of the pump and probe beam can be measured by extinction of the light through a calcite polarizer. The purity of the 200 nm, due to its high extinction coefficient through the polarizer, must be measured using a calibrated stack of silica plates at Brewster angle (Brewster Stack). Several examples of the measurement and utility of the linear polarization ratio in 2PA are given in Chapter 3. 2.7 Interfacing Detector and A/D converter for broadband detection The key to broadband 2PA technique is multichannel detection. In our apparatus, a 256- channel photodiode array (PDA) is used, and thus an analog-to-digital (A/D) converter capable of handling 256 “bursts” of analog data in between every laser shot (i.e. 1 ms for a typical 1 kHz system) is required. Further, a dual-array setup (for simultaneous signal and reference in transient absorption experiments, and simultaneous measurement of parallel and perpendicularly polarized 2PA spectra) would require an A/D converter with 39 sampling rate of 512 kHz or more. Our current A/D converter has a sampling rate of 520 kHz, which is capable of handling dual-256-channel detection at 1 kHz laser repetition rate. This is made possible by a standard 500 kHz A/D converter (PD2-MFS-4-500/16DG, UEI) with a chipset upgraded to 520k Sample per second (520 kS/s or 520 kHz, see Appendix F). The following describes the interface between the PDA and the A/D converter in a simplified and most general manner (in a user standpoint, not designer or programmer). First the user sends a signal (“run” / “enable”) indicating the A/D converter is ready for data collection via a home-built timer. The latter provides the necessary signals (clock – “CLK”, start – “STRT”) to the PDA to initiate light collection. These two signals are synchronized with the laser repetition rate signal (“LASER”) such that the PDA knows when to expect the laser pulses. After receiving one broadband continuum pulse but before the next incoming pulse (within 1 ms), the PDA sends signals (end-of- scan – “EOS”, trigger – “TR”) back to the timer, indicating it has completed one cycle of data collection. Upon receiving such signals, the timer will in turn prepare a pulse train consists of 256 pulses (channel list clock – “CL Clock” or “AD”) and send it to the A/D converter (“fast” board). The purpose of this pulse train is to prepare the A/D converter for receiving 256 analog signals from the PDA. These analog signals are then digitized and recorded by the computer. This data collection cycle is then repeated 1000 times within 1 second (for 1 kHz system). Therefore every individual shot is recorded which allows for data filtering in software as well as straightforward signal averaging. Filtering the data (for example, removing outliers from jet fluctuations or white light flicker) can dramatically improve S/N compared to simple averaging. 40 For broadband 2PA spectroscopy, and more generally broadband pump-probe spectroscopy, the change of absorption (ΔA) of the continuum in the presence of the intense pump pulse (pump-on) is recorded. Considering that the shot-to-shot instability of the laser pulses is the major source of noise, ΔA should be calculated from an immediate adjacent pair of data– e.g., absorption of the continuum at the first ms when the pump is turned on, and that at the second ms when the pump is turned off. In our laboratory, average 500 points of such ΔA values (1000 data points, 1 s averaging time for 1 kHz system) can typically give a minimally detectable ΔA = 0.1 mOD. The fast on/off switching of the pump can be realized by a chopper operating at half of the laser repetition rate, e.g., 500 Hz for a 1 kHz system, which blocks alternate pump pulses. Thus, the chopper signal also needs to be synchronized with the system. To implement this, the chopper is triggered by an external 500 Hz source, which is in turn triggered by the 1 kHz laser signal. The former analog signal is then digitized and recorded by a “slow” A/D converter. The detail description of the timer and the required components for implementing broadband detection are presented in Appendices F – I. 2.8 Summary This chapter presented the experimental technique for polarization-dependent broadband 2PA spectroscopy. This technique is based on the traditional pump-supercontinuum- probe experiments. The generation of the intense pump and broadband probe was presented in a brief fashion, but the optimization of experimental conditions was emphasized, i.e. the choice of wavelengths of the pump and probe, the pump pulse 41 duration, the focusing conditions of the pump and probe, and the choice of wire-guided gravity jet vs. conventional quartz cell and the choice of sample pathlength. All these factors will affect the overall quality of 2PA spectra – when chosen poorly, they will result in artifacts. Moreover, the calculation of the absolute 2PA cross section (σ 2PA ) has been as discussed and the origin of the polarization-dependence was summarized. 42 2.9 Chapter 2 References 1. J. E. Ehrlich, X. L. Wu, I. Y. S. Lee, Z. Y. Hu, H. Rockel, S. R. Marder and J. W. Perry, Opt. Lett. 1997, 22, 1843-1845. 2. C. W. Spangler, J. Mater. Chem. 1999, 9, 2013-2020. 3. F. Helmchen and W. Denk, Nature Methods 2005, 2, 932-940. 4. D. A. Parthenopoulos and P. M. Rentzepis, Science 1989, 245, 843-845. 5. S. Kawata and Y. Kawata, Chem. Rev. 2000, 100, 1777-1788. 6. E. Walker, A. Dvornikov, K. Coblentz and P. Rentzepis, Appl. Opt. 2008, 47, 4133- 4139. 7. W. H. Zhou, S. M. Kuebler, K. L. Braun, T. Y. Yu, J. K. Cammack, C. K. Ober, J. W. Perry and S. R. Marder, Science 2002, 296, 1106-1109. 8. C. N. LaFratta, J. T. Fourkas, T. Baldacchini and R. A. Farrer, Angew. Chem. Int. Ed. 2007, 46, 6238-6258. 9. G. S. He, J. D. Bhawalkar, C. F. Zhao and P. N. Prasad, Appl. Phys. Lett. 1995, 67, 2433-2435. 10. G. S. He, G. C. Xu, P. N. Prasad, B. A. Reinhardt, J. C. Bhatt, R. McKellar and A. G. Dillard, Opt. Lett. 1995, 20, 1930-1930. 11. J. D. Bhawalkar, G. S. He, C. K. Park, C. F. Zhao, G. Ruland and P. N. Prasad, Opt. Commun. 1996, 124, 33-37. 12. S. B. Brown, E. A. Brown and I. Walker, Lancet Oncol. 2004, 5, 497-508. 13. H. A. Collins, M. Khurana, E. H. Moriyama, A. Mariampillai, E. Dahlstedt, M. Balaz, M. K. Kuimova, M. Drobizhev, V. X. D. Yang, D. Phillips, A. Rebane, B. C. Wilson and H. L. Anderson, Nat. Photonics 2008, 2, 420-424. 14. C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys. 2009, 130, 084501. 15. N. Lin, Y. Luo, K. Ruud, X. Zhao, F. Santoro and A. Rizzo, Chemphyschem 2011, 12, 3392-3403. 16. A. Rebane, N. S. Makarov, M. Drobizhev, B. Spangler, E. S. Tarter, B. D. Reeves, C. W. Spangler, F. Q. Meng and Z. Y. Suo, J. Phys. Chem. C 2008, 112, 7997-8004. 43 17. D. N. Nikogosyan, A. A. Oraevsky and V. I. Rupasov, Chem. Phys. 1983, 77, 131- 143. 18. S. Pommeret, F. Gobert, A. Mostafavi, I. Lampre and J. C. Mialocq, J. Phys. Chem. A 2001, 105, 11400-11406. 19. A. Dragomir, J. G. McInerney, D. N. Nikogosyan and P. G. Kazansky, Appl. Phys. Lett. 2002, 80, 1114-1116. 20. A. Reuther, D. N. Nikogosyan and A. Laubereau, J. Phys. Chem. 1996, 100, 5570- 5577. 21. G. S. He, T. C. Lin and P. N. Prasad, Opt. Express 2002, 10, 566-574. 22. A. E. Jailaubekov and S. E. Bradforth, Appl. Phys. Lett. 2005, 87, 021107. 23. C. A. Rivera, S. E. Bradforth and G. Tempea, Opt. Express 2010, 18, 18615-18624. 24. S. Yamaguchi and T. Tahara, Chem. Phys. Lett. 2004, 390, 136-139. 25. S. Yamaguchi and T. Tahara, Chem. Phys. Lett. 2003, 376, 237-243. 26. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum. 2003, 74, 4958-4960. 27. S. A. Kovalenko, A. L. Dobryakov, J. Ruthmann and N. P. Ernsting, Phys. Rev. A: At. Mol. Opt. Phys. 1999, 59, 2369-2384. 28. A. L. Dobryakov, S. A. Kovalenko and N. P. Ernsting, J. Chem. Phys. 2003, 119, 988-1002. 29. M. Lorenc, M. Ziolek, R. Naskrecki, J. Karolczak, J. Kubicki and A. Maciejewski, Appl. Phys. B 2002, 74, 19-27. 30. R. A. Negres, J. M. Hales, A. Kobyakov, D. J. Hagan and E. W. Van Stryland, Opt. Lett. 2002, 27, 270-272. 31. R. A. Negres, J. M. Hales, D. J. Hagan and E. W. Van Stryland, IEEE J. Quantum Electron. 2002, 38, 1205-1216. 32. S. L. Oliveira, D. S. Correa, L. Misoguti, C. J. L. Constantino, R. F. Aroca, S. C. Zilio and C. R. Mendonca, Adv. Mater. 2005, 17, 1890-1893. 33. M. Schneider, S. Barozzi, I. Testa, M. Faretta and A. Diaspro, Biophys. J. 2005, 89, 1346-1352. 44 34. P. R. Monson and W. M. McClain, J. Chem. Phys. 1970, 53, 29-37. 35. W. M. McClain, J. Chem. Phys. 1971, 55, 2789-2796. 36. P. R. Monson and W. M. McClain, J. Chem. Phys. 1972, 56, 4817-4825. 37. W. M. McClain, Acc. Chem. Res. 1974, 7, 129-135. 38. D. Wanapun, R. D. Wampler, N. J. Begue and G. J. Simpson, Chem. Phys. Lett. 2008, 455, 6-12. 39. L. De Boni, C. Toro and F. E. Hernandez, Opt. Lett. 2008, 33, 2958-2960. 45 Chapter 3. Two-Photon Absorption Spectroscopy of Alcohols and Alkanes 3.1 Introduction Two-photon absorption (2PA) has been widely used in many condensed phase applications as mentioned in Chapter 2 Introduction, but the 2PA cross sections (σ 2PA ) of common liquids used as solvents are sparse in the literature. 1-5 They are usually obtained at a single wavelength via the traditional z-scan technique, 6-8 but broadband spectra analogous (and complementary) to the conventional linear UV-Visible spectra are only available for handful of much larger systems such as retinal, 9 C 60, 10 porphyrins, 11-14 and ZnS nanocystals. 15,16 This point is most clearly made when it is appreciated that the 2PA spectra of all the commonly used solvents have not yet been reported, with the exception of the previously published water spectrum by our group. 17 Further, prior single point σ 2PA values for water, obtained from different studies, span almost an order of magnitude, 1,2,4-7 which makes it extremely difficult to construct a “spectrum” from a set of single wavelength measurements. Without a full broadband spectrum, it is impossible to analyze the difference between the 1-photon absorption (1PA) and the 2PA spectra. However, much like IR and Raman spectra, the comparison of the 1PA and 2PA spectra would provide a complete picture of the electronic transitions, as these two complementary techniques are governed by different selection rules. This has been shown very clearly in the work on water – our previously published water 2PA spectrum has shown that the entire Ã1 1 B 1 band observed 46 in the 1PA spectrum is missing due to its pure atomic oxygen character. 17 Further, the complete water 2PA spectrum from 6.5 – 10.5 eV, coupled with the linear polarization ratio (see Equation 3), has revealed at least four states in these energy region, which are Ã1 1 B 1 , 1 1 A 2 B ~ , 1 1 A 2 and 3 1 A 1 states. The latter three states lie in higher energy (> 9 eV) and overlaps with each other. Ab initio calculations and gas phase 1PA spectroscopy provide important reference point to the condensed phase analysis. For isolated water, alcohol and alkane molecules, the low-lying excited states are often characterized as mixed Rydberg and valence (anti- bonding) states, 18 which initially possess Rydberg character at the vertical Franck Condon (vFC) region but subsequently acquire anti-bonding character due to same- symmetry avoid-crossing. 19,20 In light of this, the upper orbital of the water Ã1 1 B 1 state was described as Rydberg-type 21-24 in some literature and anti-bonding type 25-29 in others, but it is important to bear in mind that it has mixed Rydberg 3s and anti-bonding σ* O-H characters. Thus, the dissociative nature of the Ã1 1 B 1 state explains the broad and diffuse band at 7.4 eV observed in the gas phase 1PA spectrum; similarly, 1 1 A 2 B ~ band observed at ~9.6 eV is also dissociative as both these excited states have a common upper orbital. The upper orbitals of the 1 1 A 2 , 2 1 B 1 and 3 1 A 1 states, on the other hand, have pure Rydberg characters and thus are more spatially extended. The latter two excited states give rise to the sharp resonance in the gas phase 1PA spectrum but the 1 1 A 2 state is symmetry forbidden in the C 2v framework. 47 In the condensed phase, since the Rydberg orbitals are more spatially extended and thereby more likely to mix with the solvent continuum, electronic promotion to a pure Rydberg orbitals is thought to relate to the ionization of the molecule. Excitation to a Rydberg orbital which is pre-dissociated by an anti-bonding orbital, on the other hand, intuitively leads to photofragmentation. The characterization of these orbitals below the vertical ionization energy (VIE) thus provides invaluable information on the competing ionization and dissociation deactivation channels upon excitation to the vacuum-UV (VUV) range. The Ã1 1 B 1 band is the only clearly discernable band in the liquid water 1PA spectrum, but our polarization-dependent 2PA spectrum gives a fuller picture of the excited states just below the VIE, and thus provides more insight to the water ionization dynamics. 17 In this study, we will report for the first time the 2PA spectra of some commonly used solvents, and extend the analysis of water to alcohols and alkanes, which aims to decipher the broad and overlapping transitions in the VUV region. We will mainly focus on comparing the 1PA and 2PA spectra and extract the excited state characters with the aid of symmetry arguments. 3.2 Experimental The VUV energy region is accessible by simultaneous absorption of one deep-UV photon and one UV-Visible photon. The generation of these two pulses was described in Chapter 2 and will not be repeated in here. The 4.6 eV (267 nm) pump pulse was attenuated to 6 μJ and the 6.2 eV (200 nm) pump pulse was 1.5 μJ at the sample. Their spatial full-width- at-half-maximum (FWHM) was in the range of 400 – 500 μm, as determined by the 48 knife-edge technique. The sample was placed at the focal point of the probe continuum (typically 310 < λ probe < 650 nm), where the spatial FWHMs were 60 – 150 μm across the continuum spectrum. The spatial overlap between the pump and probe were carefully characterized and wavelength dependence of the spot sizes was mathematically compensated. As delivered by the wire-guided gravity jet, 30 the thickness of the water, methanol and ethanol thin film varies from 40 to 50 μm, with an estimated 20% error, as measured by group velocity delay method. Methanol, ethanol and cyclohexane samples were obtained from VWR; n-propanol and n-butanol sample were obtained from Sigma- Aldrich; all of the above were spectroscopic grade. The 95% n-hexane was obtained from VWR; the impurities were branched and cyclic alkanes with the same number of carbon atoms. All samples were directly used without further purification. The polarization purity of the continuum was wavelength dependent but better than 200:1 across the spectral region. The purity of the 266 nm pump was better than 70:1, as measured by extinction of the light through a calcite polarizer. The purity of the 6.2 eV pump light was at least 40:1, measured using a calibrated stack of 9 silica plates at Brewster angle. The molecular orbitals calculations for n-hexane and cyclohexane were performed by Tom Oliver from the University of Bristol. The geometries were optimized with DFT/B3LYP and 6-31+G* basis set. The first eight excited states were calculated with time-dependent DFT (TD-DFT) with the same B3LYP function and 6-31+G* basis set. 49 3.3 Results 3.3.1 Water and Alcohols The parallel-polarized 2PA spectra of liquid water, methanol and ethanol are presented in Figure 3.1, where the σ 2PA values are plotted against the total 2PA energy, i.e., the sum of the pump and probe photon energies. The 4.6 eV pump covers the total energy region of 6.7 – 8.5 eV, which provides information on the two-photon absorption threshold, whereas the 6.2 eV pump extends the spectra to a higher energy region up to 10.1 eV. It is apparent that at any given 2PA energy, the larger molecule has a larger 2PA cross section. To the best of our knowledge, the broadband 2PA spectra (with absolute σ 2PA values in GM) in the vacuum ultraviolet region for the two alcohols are measured for the first time. Figure 3.1. σ 2PA for water (red), methanol (green) and ethanol (blue) at 4.6 eV (267 nm) pump and 6.2 eV (200 nm) pump. At each pump wavelength, the spectra for all three solvents were obtained under the same experimental condition (same pump power, spot size, pulse duration) except for the jet thickness. In all cases, the relative polarization of the pump and probe are the same (parallel-polarized). 50 Table 3.1. Absolute 2PA cross section at 7.14 eV and 9.4 eV for water, methanol and ethanol (σ para , parallel polarization, in GM). 2PA energy (photon combination) 7.14 eV (266/500) a 7.14 eV (347/347) b 9.4 eV (200/387) c 9.4 eV (264/264) d Water ~0 / 3.7 ± 1.5 1.1 ± 0.1 Methanol 0.15 ± 0.06 0.32 5.1 ± 2.0 1.7 ± 0.2 Ethanol 0.30 ± 0.12 0.45 7.6 ± 3.0 3.1 ± 0.3 a This study: 266 nm pump and 500 nm probe (selected from the continuum); water 2PA is negligible at this 2PA energy. 40% error was estimated for our measurement. b 2PA coefficient (β, in cm/GW) taken from Ref 3; converted to 2PA cross section (σ, in GM) using the following parameters: M water = 18 g/mol; ρ water = 0.998 g/cm 3 ; M methanol = 32 g/mol; ρ methanol = 0.791 g/cm 3 ; M ethanol = 46 g/mol; ρ ethanol = 0.785 g/cm 3 . The error was not reported by the authors. c This study: 200 nm pump and 387 nm probe (selected from the continuum). d 2PA coefficient (β, in cm/GW) taken from Refs 4,5; converted to 2PA cross section (σ, in GM) using the parameters shown in footnote b. 10% error was reported by the authors for all three solvents. The comparison of the σ 2PA values between this study and the literature is summarized in Table 3.1, from which it can be seen that the two sets of values are in good agreement, considering the fact that several measurements of the σ 2PA value for water in the literature span almost an order of magnitude at a given 2PA energy. 1,2,4-7 Further, it is important to point out that the 2PA cross section is pump wavelength dependent – a possible contribution to the difference between the values obtained from this study and those from the literature. This effect is best illustrated in the 8.3 – 8.5 eV region accessible by both 4.6 and 6.2 eV pump (see Figure 3.1). It is apparent that the σ 2PA values measured with the higher energy pump (6.2 eV) were larger. † The larger cross section obtained with higher energy pump photon is a general property of 2PA measurements and is known as resonance enhancement, in which one of the two photon energy is close to the energy of a 1-photon allowed transition (near resonance) and thereby facilitates the 2-photon † Although when comparing the absolute cross section between different solvents, the uncertainty could be as large as 40%, which is mainly originated from the measurement of the jet thickness and the spot size of the pump and probe beam at the sample, the uncertainty of absolute cross section for the same solvent is slightly smaller (~30%) since the uncertainty for the jet thickness did not need to be accounted for. 51 transition. 17,31-34 This can in turn explain that the σ 2PA values obtained with 6.2 eV pump in this study were consistently higher than those measured with 4.7 eV pump. Figure 3.2. 2PA spectra of water and selected alcohols (a-d) and their corresponding polarization ratio (e-h). Spectra were normalized at the region from 8.3 eV to 8.5 eV and reported as relative cross section in arbitrary units. All 2PA spectra are obtained with parallel polarization. Polarization ratios for water shown in panel (e) were reproduced from Ref 17; it was averaged over several independent measurements. Alcohol ratios were cut off at 7.0 eV due to the low signal-to-noise ratio at energies lower than 7.0 eV. The collection of relative 2PA spectra of water and three alcohols are presented in Figure 3.2(a) – (d). For each solvent, in order to facilitate comparison, the two sets of spectra obtained with two different pump wavelengths are normalized at their overlapped region 52 of 8.3 – 8.5 eV, despite the consistently larger cross sections obtained with the 6.2 eV pump. From the 2PA spectra we can see a clear trend upon substituting an H atom in water with an alkyl group. Water molecules exhibit very weak 2PA cross section at low energy region (6.75 < E 2PA < 8.5 eV). Substituting one H atom with a methyl group (to make methanol), however, has moderately increased the 2PA cross section in this region. Increasing the length of the alkyl chain has further increased the 2PA cross section as seen in ethanol, but alcohols with longer alkyl chain such as n-propanol and n-butanol (data not shown) do not show further spectral change in the low energy region. As a result, all three alcohols have similar 2PA threshold at ~7.2 eV, which is much lower than that for water – see Table 3.2 for summary. The table also shows that all the solvents of interest have a wider transparent window for 2PA. The high 2PA threshold of water and alkanes made them ideal solvents for condensed phase 2PA applications. Table 3.2. 1PA and 2PA absorption thresholds (in eV) Water Methanol Ethanol n-Hexane Cyclohexane 1PA (99% T) ~6.4 a 5.6 b 5.6 b 5.9 b 5.4 b 2PA (99% T) c 7.8 6.9 6.9 --- --- 2PA (90% T) d 8.7 8.5 8.4 --- --- 2PA e 8.2 7.2 7.2 7.6 7.3 a Data taken from Ref 35. The extinction coefficient at 6.4 eV is ~2.25×10 -4 M -1 cm -1 , corresponding to a transmission slightly larger than 99% in a 0.1 cm cell. b Data taken from Ref 36. The thresholds were the energies where the transmission of the liquid was ~99% in a 0.1 cm cell. c This study. The 2PA thresholds were the energies where the transmission of the probe was ~99% (equivalent to absorbance of ~4 mOD) in a 0.1 cm pathlength, and at peak irradiance of the 267 nm pump pulse = ~100 GW/cm 2 . This is estimated based on the energies where the absorbance was ~0.2 mOD with a ~50 μm pathlength. The numbers for alkanes were unknown since the pathlengths were not measured. d This study. The 2PA thresholds were the energies where the transmission of the probe was ~90% (equivalent to absorbance of ~50 mOD) in a 0.1 cm pathlength, and at peak irradiance of the 200 nm pump pulse = ~100 GW/cm 2 . This is estimated based on the energies where the absorbance was ~0.5 mOD with a ~50 μm pathlength and a peak irradiance of the 200 nm pump = 20 GW/cm 2 . e This study. The 2PA thresholds were the energies where the absorption intensity reaches 10% of the peak value within our spectral range. 53 Figure 3.2(e) – (h) present the broadband spectra of the polarization ratios for water and alcohols. ‡ It is apparent that for all three alcohols, the polarization ratios are larger than 1 across the whole spectral region, indicating that 2PA cross sections are larger when the relative polarization of the pump and the probe are the same. Moreover, the polarization ratio of water shows a steep rise from 7.0 to 8.0 eV, whereas this increase is moderate in methanol and becomes relatively constant for alcohols with longer alkyl chains. This trend is reminiscent to the increasing 2PA cross section from water to n-propanol. It is noteworthy that the ratio for methanol is ~3.0 at 10.0 eV but slowly decreasing to ~2.0 at 8.3 eV, and the trend is preserved in the lower energy region covered by the 4.6 eV pump. Neither water nor other alcohols exhibits such a clear trend in the polarization ratio. The polarization ratio at 7.92 eV for methanol and ethanol, along with other solvents, are listed in Table 3.3 and compared to the literature values. § Our experimental values were close but consistently higher than those measured by Rasmusson et al. 37 The latter has considerable amount of transient absorption signals, mostly likely due to 2- (pump-) photon ionization of the solvent, present in every 2PA spectrum. Since the absorption due to photoionization photoproducts, e.g., solvated electron, is isotropic, i.e., R=1, the ‡ Careful examination of the polarization ratio reveals that the ratios measured with the 4.6 eV pump are generally larger than those measured with the 6.2 eV pump. This dependence in pump energy is similar to the pump energy dependent 2PA cross section, which also arises from symmetry considerations, and it was discussed for water elsewhere (see C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys. 2009, 130, 084501, and references therein). § The literature values were measured with 266 nm pump and 380 nm probe, see M. Rasmusson, A. N. Tarnovsky, E. Akesson and V. Sundstrom, Chem. Phys. Lett. 2001, 335, 201-208. Since our work utilized the same pump photon energy, it allows direct comparison without concerning the pump energy dependence. 54 presence of this type of signal will reduce the 2PA polarization ratio. In light of this, it is worthwhile to stress that using appropriate pump irradiance to eliminate transient signal is essential for accurately measuring 2PA polarization ratios. ** Table 3.3. Linear polarization ratio at 7.92 eV (266 nm + 380 nm) for selected solvents. Methanol Ethanol Chloroform Carbon Tetrachloride Hexane * This work a 1.9 3.2 2.6 2.75 2.6 Rasmusson b 1.5 3.1 2.0 2.2 2.2 * This work: hexane isomers; Rasmusson: n-hexane. a 266 nm pump and 380 nm probe (selected from the continuum). b Data taken from Ref 37. 3.3.2 Alkanes The polarized 2PA spectra for cyclohexane and n-hexane (and its isomers), measured with 4.6 eV pump, are presented in Figure 3.3(a) and (b), respectively. Similar to water and alcohols, the 2PA spectra of alkanes are monotonically increasing as 2PA energy increases. For n-hexane, there is no discernable feature in the spectral region. However, there was a small shoulder which is indicative of a lower energy transition centered at ~7.5 eV. As a result, the two-photon absorption threshold for cyclohexane is lower than hexane isomers, but both alkanes have higher 2PA thresholds compared to the alcohols. The cyclohexane polarization ratios change dramatically across the spectral range, whereas hexane isomers have a more constant ratio (~2.5 to 3.0) from 7.6 to 8.5 eV. The decrease of polarization ratio towards lower energy has some contribution from the trace amount of photonionization products, 38-40 which, in the case of alkanes, are difficult to ** Although the contribution from the transient is minimized by taking the peak height at time zero, as done by Rasmusson et al., this method is sensitive to the accuracy of determining the time zero position; as the 2PA contribution decays and transient signal grows, the effective ratio changes dramatically after time zero. 55 avoid as the cross sections for absorbing two 267 nm pump photon should be larger than water and alcohols. Figure 3.3 Relative 2PA spectra for (a) cyclohexane and (b) hexane isomers. Both parallel and perpendicular spectra, as well as the polarization ratios, were shown. Due to low signal-to-noise ratio at low energies, the 2PA ratio for cyclohexane was cut off at 7.3 eV and that for n-hexane was cut off at 7.4 eV. 3.4 Discussion 3.4.1 Water and Alcohols Although the 2PA spectra presented above are generally broad and featureless, they do have subtle differences especially in the lower energy region, which appear as the various different thresholds for different solvents. The trend in the overall 2PA spectral shape and polarization ratios, from water to alcohol to alkanes, provides a window to explore the overlapping transitions in these broad spectra. It is apparent that upon substituting an H atom with an alkyl group, the 2-photon absorption threshold becomes much lower, and the absorption around 8.5 eV gains intensity as the length of alkyl chain increases. These are indicators for new absorption bands developing, which originate from a) transitions 56 involving extra orbitals of the carbon atom, and/or b) symmetry lowering due to the addition of the alkyl chain. Figure 3.4. 1PA and 2PA spectra for water, methanol and ethanol, and HOMO/LUMO for water, methanol and ethanol. Water vapor spectrum (green line) was reproduced from Ref 41, and condensed phase spectrum (neat water, blue line) was re-plotted and smoothed from Ref 42; Alcohol vapor spectra (green line) were from Ref 26, and condensed phase (neat methanol and ethanol, blue line) spectra were from Ref 23. Molecular orbitals were calculated on QChem electronic structure package with HF/6-31G* basis set. Water molecule, where transitions originating from the oxygen play the most important role, provides a starting point for our analysis. The lowest energy transition originates 57 from promoting an electron from the out-of-plane 2p x /1b 1 orbital (HOMO) to the 3s/4a 1 orbital (LUMO). 17,42 This dissociative Ã1 1 B 1 state has a transition energy of ~7.3 eV in an isolated water molecule where H-bonding is absent, but blue-shifted significantly to ~8.2 eV in the condensed phase (see Figure 3.4). It can be immediately seen that this lowest energy transition band is missing in the condensed phase 2PA spectrum. Although the 1 B 1 state is both 1- and 2-photon allowed in the C 2v framework, s←p transitions are 2- photon forbidden by the atomic propensity rule. Since the upper and lower state of this transition can be described by a single oxygen atom, the non-observation of the water Ã state is not wholly surprising. 17 The 1PA spectra of methanol shows that upon substituting an H atom with a methyl group, the lowest energy band originating from the oxygen 3s←2p x orbital promotion (x- axis is defined as out-of-plane) is much weaker but still observable at ~6.8 eV for an isolated methanol molecule. The location of this band, however, is undetermined in either liquid or solid phase methanol. 43,44 Giving the small oscillator strength observed in the gas phase, this feature may be too weak to be detected in the condensed phase. Another possibility is that it can shift significantly to underlie stronger features – this has some ground considering the 3s←2p x transition band has blue-shifted for 0.9 eV upon condensation. Perusing the solid methanol 1PA spectrum by Kuo et al. reveals a subtle shoulder at ~7.4 eV, but the authors did not comment on its possibility of being the 3s←2p x transition. 44 The two clusters of sharp bands in the region of 7.7 – 9.0 eV are transitions to the two-fold 3p Rydberg orbitals. 45 These lie much lower in energy as 58 compared to those for water (sharp features at > 10 eV), and give rise to the broad absorption band at ~8.1 eV in the condensed phase. 43 This assignment is reassured by the absorption spectrum of solid methanol – the two spectra show remarkable resemblance, and a broad feature assignable to 3p←2p x is observed at ~8.4 eV. 24,44 Kuo et al. further commented on that it is not conclusive if this is a single transition or a convolution of transitions to the two-fold 3p Rydberg orbitals. The assignment of the higher energy transitions at ~10 eV is also inconclusive – either transition to the out-of-plane 3p orbital, or to the 3d orbitals. 24,44 Figure 3.5. Simulation of 1PA and 2PA spectra for neat liquid methanol. Red line: 3p y ←2p x ; green line: an unidentified high energy transition. The position and width of these two features were fixed in the simulations of the 1PA and 2PA spectra, but the relative heights were allowed to float. Blue line: 3s←2p x ; dark cyan line: 3p x ←2p x . Every parameter in these Gaussians was allowed to float in order to obtain the best fit. Blue circles: experimental 1PA spectrum (reproduced from Ref 23). Gold line: simulation of 1PA spectrum (sum of red and green lines). Magenta circles: experimental 2PA spectrum (as shown in Figure 3.4). Black line: simulation of 2PA spectrum (sum of four lines). 59 Our 2PA spectrum offers some hints as to where the lowest energy transition band lies in liquid methanol. From Figure 3.5 we can see that the 1PA spectrum can be described by a minimum of two Gaussians (as discussed previously). With the positions (and FWHMs) of these two Gaussian fixed, two additional Gaussians are required to completely simulate the 2PA spectrum. We recognized that this is not a unique fit but it provides visual aid to the subtle features in the spectra. Most significantly, a low lying state at ~7.5 eV (shown in blue line) is required to best reproduce the subtle feature at the low 2PA energy region. The location of this band is in excellent accord with the shoulder observed in the solid methanol, 44 and thus most likely arises from the 3s←2p x transition. Its strong blue-shift from gas to condensed phase (6.8 to 7.5 eV) is reminiscent to water (7.3 to 8.2 eV). More interestingly, this transition is forbidden in water but gains intensity in methanol. This can be reconciled by considering the multi-atom contribution of molecular orbitals in methanol – Figure 3.4 shows that the HOMO has significant carbon 2p x character. As a result, the rigorous atomic selection rule governing the electronic transition of water was in turn relaxed, and the molecular selection rule in the C s framework starts to play an important role. In this low symmetry point group, the LUMO←HOMO(1 1 A'') transition is allowed, which explains the similar 2PA onset in all alcohols (see Figure 3.2). The multi-atom contribution of the HOMO is supported by numerous experimental and theoretical studies. The character of the occupied orbitals can be examined by X-ray emission spectroscopy (XES), where the character of the occupied orbitals is revealed by 60 promoting an oxygen core 1s electron to an unoccupied orbital and monitoring the relaxation of another electron from one of the occupied orbitals to the half-filled oxygen 1s orbital. 46 Oxygen K-edge X-ray emission spectrum for liquid methanol has indeed shown a highly hybridized oxygen and carbon p orbitals for the HOMO. 47 This result is in line with the theoretical prediction that the methanol HOMO has 70% oxygen p character and 30% carbon p and hydrogen s characters. 25 Further, the LUMOs of the alcohols also have contributions from both oxygen and carbon. The X-ray absorption spectroscopy (XAS) reveals the character of an unoccupied orbital by promoting a core 1s electron to it. The oxygen and carbon K-edge XAS for liquid methanol showed that electronic promotion to the LUMO gave rise to the pre-edges of both the oxygen and carbon K-edge spectra, indicating that the LUMO has contribution from both atoms. 29 Vacuum-UV study of alcohols also reveal similar findings. 24,26 We can go some way to explain why the 3s←2p x transition at ~7.5 eV is stronger in methanol 2PA than 1PA. One possibility is that the transition strength associated with the 3p←2p x electronic promotion at ~8.1 eV becomes lower in 2PA, and in turn reveals the weak band around the onset of the spectrum. This argument has some ground based on the observation in water – that the 2PA cross section for 2 1 B 1 (3p z ←2p x ) state is only ~10x larger than the forbidden Ã 1 B 1 (3s←2p x ) state. 17 Along this line, it is interesting to find that the totally symmetric transition 3 1 A 1 (3p x ←2p x ) has a large 2PA cross section as predicted by theory. Bearing both in mind, deconvolution of methanol 2PA spectrum revealed a prominent band centered at ~9 eV which is otherwise missing in the 1PA 61 spectrum. Further, the relative amplitudes between the p z -p x and p x -p x transitions, resulted from spectra deconvolution, are similar to those predicted for an isolated water molecule. Thus, we can tentatively assigned the band centered at ~8.1 eV to 1 A'' (3p z ←2p x ) transition, and that at ~9.0 eV to the totally symmetric 1 A' (3p x ←2p x ) transition. This assignment is also consistent with the polarization ratio observed for methanol. As first enunciated by Monson and McClain 31-34 and later demonstrated for water, 17 the linear polarization ratio (σ para / σ perp ) is expected to be high (≥ 4/3) for totally symmetric transition. If the p x -p x transition is centered at ~9.0 eV, the polarization ratio would be higher at the region of E 2PA > 8.5 eV – this is indeed consistent with our polarization- dependent experiments. The decreasing polarization ratio towards the low energy is thus due to the decreasing contribution from this totally symmetric transition and increasing contribution from the non-totally symmetric transition such as the 1 A'' (3p z ←2p x ) and 1 A'' (3s←2p x ). The equivalent states in water are predicted to have low polarization ratios, 17 which, under the small perturbation of the methyl group, can be preserved in the case of methanol. This is why the overall trend of methanol polarization bears more resemblance to water than other alcohols. As the length of the alkyl chain increases and the overall symmetry of the molecule decreases, the cross section of these two low-lying transitions increases on one hand, the polarization ratio show less variation on the other – as can be seen from Figure 3.2, the overall polarization ratio in our spectral domain is relatively flat for ethanol and n-propanol. 62 Thus far we have ignored the carbon 3p orbital in the y-direction, which is along the molecular backbone and manifests as the σ bond (C–H and C–O). The electronic promotion to the anti-bonding σ* C-O results in C–O bond fission which adds extra complexity in the photochemistry of alcohols. In the gas phase methanol, O–H bond fission is the dominant dissociation pathway at low excitation energy, but C–O bond fission starts to gain importance at ~8.0 eV excitation. 20 The XAS experiment has shown that for liquid methanol, the LUMO+1 (σ* C-O ) and LUMO+2 (σ* C-H ) are only 0.3 and 1.1 eV above the LUMO. 29 This would place the optical transitions of σ* C-O ← HOMO and σ* C-H ← HOMO at around 8.5 eV and 9.4 eV, respectively, which are certainly within our spectroscopic range. Our data cannot definitively confirm or rule out the contribution of the former to the 2PA spectra, but this transition alone cannot explain the spectral difference between methanol and ethanol, since both molecules have a C-O bond. The latter, on the other hand, is not likely to have significant contribution to the E 2PA < 7.5 eV region and thus have no effect on the onsets of the 2PA spectra of alcohol (see next section). 3.4.2. Alkanes Since n-hexane (and its isomers) and cyclohexane do not possess the O–H moiety which gives rise to the low energy transition in alcohols, it is expected that its 2PA threshold lies higher than those of alcohols. Our 2PA spectra of these molecules indeed meet this expectation (see Table 3.2). Further, the lowest energy conformers of both molecules possess a center of inversion (C 2h for n-hexane all-trans conformer, and D 3d for 63 cyclohexane chair conformer), and thus symmetry argument dictates that the 1-photon allowed transitions must be 2-photon forbidden and vice versa. For cyclohexane, results from neutron diffraction experiments were inconclusive in determining which conformer, chair or boat, is the dominant species in liquid, 48 but molecular dynamic (MD) simulations predicted that the liquid phase molecules are mainly present in the chair form. 49 Figure 3.6. Selected HOMOs and LUMOs of the chair form of cyclohexane, calculated by TD- DFT with B3LYP functional and 6-31+G* basis set, and possible transitions (and their corresponding symmetries) involving promotions between these orbitals. Selected HOMOs and LUMOs of the chair form of cyclohexane (D 3d ), calculated by TD- DFT with B3LYP functional and 6-31+G* basis set, were shown in Figure 3.6, along with possible transitions involving these orbitals and the corresponding transition 64 symmetries. Note that the purpose of this relatively low-level calculation is to illustrate the character and symmetry of the molecular orbitals, and thus the orbital numbers shown in Figure 3.6 are for labeling purpose only – they do not imply the energy ordering of the orbitals. Nonetheless, it is apparent that orbital 2 and 3 are two-fold degenerate (and they lie much higher in energy in comparison to orbital 1, according to the current simple calculation), and orbital 4 is a Rydberg 3s-like orbital and thus expected to lie lower in energy than orbitals 5 – 7, which are Rydberg 3p-like orbitals. Thus, the lowest energy transition is 4←2(3) (a 1g /3s←e g ) which has an overall E g symmetry. In the gas phase 1PA spectrum, this forbidden transition was observed at ~7.0 eV via vibronic coupling (and thus low transition strength). 50 The identity of this band was further confirmed by multiphoton ionization experiments. 51 For pure cyclohexane liquid, broad and featureless spectra were observed for both 1PA and 2PA spectra – see Figure 3.7, but their corresponding derivative spectra provides powerful visualization of subtle differences between the two spectra. The 2PA derivative spectrum shows a plateau at 7.5 eV, whereas the 1PA derivative spectrum exhibits one at ~7.8 eV. The plateau in the derivative spectrum gives some hints on the center position of an absorption band, and in turn suggests the first “bright” transition is different in the two spectra. Indeed, simulation of the 2PA spectrum requires a Gaussian band centered at ~7.5 eV to best reproduce the early onset, but this band is not needed to simulate the 1PA spectrum (see Figure 3.7). As the first absorption band in the 2PA spectrum lies lower in energy when comparing to that of the 1PA spectrum, it is trivial to assign this feature to 65 the lowest energy transition, which originates from 4←2(3) (a 1g /3s←e g ) orbital promotion. This transition is blue-shifted for ~0.5 eV in comparison to that of the gas phase, a behavior similar to the lowest energy transition of water and alcohols as discussed before. Moreover, apparently this feature is 2PA allowed but 1PA forbidden – an observation entirely consistently with the expectation of an E g transition within the D 3d framework (see Figure 3.7 blue Gaussian). Figure 3.7. Simulation of 1PA and 2PA spectra for neat liquid cyclohexane (left) and molecular orbitals involved in these transitions (right). Blue circles: experimental 1PA spectrum (reproduced from Ref 52). Gold line: simulation of 1PA spectrum (sum of red and green lines). Magenta circles: experimental 2PA spectrum (as shown in Figure 3.4). Black line: simulation of 2PA spectrum (sum of blue and dark cyan lines). The dash arrows represent 2-photon allowed transitions, and the solid arrow represents 1-photon allowed transition. Interestingly, the 2PA spectrum shows a slight “dip” at ~7.8 eV – a place where the first absorption band in 1PA is expected. This observation then strongly suggests the 7.8 eV absorption band is 1PA allowed but 2PA forbidden. Symmetry argument dictates that this transition must have parity change (u-g). Among the low-lying transitions shown in 66 Figure 3.6, only transitions of overall E u or A 2u symmetry can fulfill this requirement. There were a number of possible transitions, but those originated from orbital 1 should have much higher energy and thus can be excluded. The only remaining choice is 7←2(3) (a 2u /3p←e g ) orbital promotion (overall E u ) – see Figure 3.7 red line. The assignment of higher energy transitions at E 2PA > 8 eV is more complicated, as there likely are a number of broad absorption bands. In order to best reproduce the shape of both 1- and 2PA spectra, at least one broad Gaussian centered at ~8.5 eV is needed (center position of this Gaussian function is slightly different in the two spectra), but our spectral range can only cover a partial band. Despite these difficulties, we can at least comment on one particular transition in this energy region, owing to the extra information provided by our polarization-dependent studies. Recall that the linear polarization ratio of cyclohexane starts at 1.5 in the low energy region but dramatically increases to ~4 at ~8.5 eV. The high polarization ratio is a clear indication of a totally symmetric transition 31-34 (see Appendix E for summary) and the only candidate within these low-lying transitions is the 4←1 (a 1g /3s←a 1g ) transition which is overall A 1g symmetry. Thus, the high energy absorption band in the 2PA spectrum can be assigned to this transition; this is also possible that this transition lies even higher in energy but the broad peak extends to E 2PA < 8.5 eV (see Figure 3.7 dark cyan line). Further, since the electronic promotion from orbital 1 to orbital 4 (Rydberg 3s) lies in E 2PA > 8.5 eV, the transitions from orbital 1 to orbital 5 – 7 (Rydberg 3p) are therefore expected to lie in even higher energy. This 67 reinforces that the 7.9 eV band being a 7←3 (E u ) transition, since any other transitions with change of parity (u-g) originate from orbital 1. Figure 3.8. Selected HOMOs and LUMOs of all-trans form of n-hexane, calculated by TD-DFT with B3LYP functional and 6-31+G* basis set, and possible transitions (and corresponding symmetries) involving these orbitals. Differing from the case of cyclohexane, the structure of liquid n-hexane is far less clear. Neutron diffraction experiments and MD simulations of deuterated liquid n-hexane has revealed an equal distribution of all-trans (C 2h ) and all-gauche conformers at room temperature, 53 but low-frequency isotropic Raman spectra of liquid n-hexane suggests only ~20% of the molecules present as all-trans conformer, whereas ~50% have one gauche C-C bond. 54 At this stage, we can expect that the high symmetry C 2h conformer contributes no more than 50% and there will be a broad distribution of lower symmetry 68 conformers (those with at least one gauche C-C bond). It is also important to point out our sample is only 95% n-hexane. The 5% impurity includes the six-carbon branched and cyclic alkanes – the former is expected to have lower symmetry. The lowest energy transitions of n-hexane, which are responsible for the shoulder observed at ~7.6 eV in the liquid 1PA spectra, are determined to be an electronic promotion to orbital 4 (LUMO), which has mixed Rydberg 3s and valence anti-bonding σ* C-H characters (different from the 3s/σ* O-H orbitals in water and alcohols), see Figure 3.8 for illustration. 55,56 Since n-hexane liquid contains mixture of high and low symmetry conformers, the overall spectral difference between 1- and 2PA is not expected to be as significant as the cyclohexane case. Indeed, the same set of Gaussian functions can be used to simulate both the 1- and 2PA spectra. However, the relative intensity of the first two absorption bands is dramatically different in the two spectra (red and dark cyan lines in Figure 3.9), which could suggest that the high symmetry all-trans conformer (C 2h ) still plays an important role. For an all-trans n-hexane molecule, all three HOMOs have a gerade symmetry (orbital 1 – 3 in Figure 3.8), and therefore the transitions from these orbitals to the LUMO (orbital 4, a g , Rydberg 3s) would be 1-photon forbidden but 2- photon allowed. This is in line with the stronger 2PA band at ~7.95 eV, which results in the different overall band shape when compare to the 1PA spectrum. 69 Figure 3.9. Simulation of 1PA and 2PA spectra for neat liquid n-hexane (left) and selected molecular orbitals of n-hexane (right). Blue circles: experimental 1PA spectrum (reproduced from Ref 57). Gold line: simulation of 1PA spectrum. Magenta circles: experimental 2PA spectrum (as shown in Figure 3.4). Black line: simulation of 2PA spectrum. The molecular orbitals were calculated at the same level as cyclohexane, and the order of these orbitals should again be considered approximate due to the limited accuracy of the low level electronic structure calculation. The dashed arrows represent 2-photon allowed transitions, and the solid arrow represents a 1-photon allowed transition. The second lowest energy transition, on the other hand, is much stronger in 1PA, which is indicative of a 1PA allowed u-g transition. From Figure 3.8 it is apparent that all Rydberg 3p orbitals have ungerade symmetry and thus we can conclude the upper orbital(s) must have p character. The dark cyan arrow represents the most probable origin of the ~8.2 eV band. Lastly, a relatively strong feature is observed at E 2PA > 8.5 eV. This must be a transition with gerade symmetry, and considering the fact that the polarization ratio of this band is not too high (~3), we can tentatively rule out the totally symmetric transition A 1g . Thus, the 4←2 (B g ) transition is the only remaining candidate. Further, the 70 predominant contribution of this band at the high E 2PA region is consistent with the polarization ratio measurements – recalled that Fig 3.3 (b) shows a relative constant polarization ratio at 8 < E 2PA < 8.5 eV, which is an indication of single absorption band. 3.5 Conclusion The polarization dependent broadband 2PA spectra for some commonly used alcohols and alkanes were reported for the first time. The broadband 2PA spectrum of water provides a starting point to understand the alcohol spectra. The electronic transitions of short chain alcohols (# of C = 1 - 4) are very similar to those determined for water. Their HOMOs, however, have increasing carbon 2p x contributions, which subsequently relaxes the atomic propensity rule and give intensity to the 3s/σ* O-H ← 2p x electronic promotion. The direct consequence is the lower energy 2PA onset of the alcohols, which narrows the transparent window when the alcohols are used in any 2PA experiments. The transitions to the pure Rydberg orbitals (3p x and 3p z ) have intermediate energies (8 – 9 eV), with the totally symmetric transition to the 3p x orbital gaining intensity in 2PA. The two alkanes presented in this work, n-hexane and cyclohexane, demonstrate the different selection rules governing 1-photon and 2-photon transitions. Cyclohexane molecules are present predominantly in the chair conformer in the liquid phase, the high symmetry with a center of inversion (D 3d ) provides excellent demonstration of the complementing 1PA and 2PA techniques. The lowest energy transition of E g symmetry is directly observable in the 2PA spectrum, as it is fully 2-photon allowed. Further, from the 71 polarization ratio measurement a totally symmetric transition (A 1g ) at E 2PA > 8.5 eV is revealed. In the case of n-hexane, independent studies such as neutron diffraction, Raman spectroscopy and MD simulations suggest a broad distribution of trans and gauche conformers in the liquid. Perusing the 1- and 2PA spectra, however, reveals subtle spectral difference, which indicates that the high symmetry all-trans conformer still affect the broad spectral properties. For the alcohols and alkanes discussed in this work, by comparing 1- and 2PA spectra, coupled with the polarization ratio measurements, we can study the broad and featureless condensed phase absorption spectra in far more detail. New transitions which are otherwise absent in the 1PA spectrum can now be observed. 72 3.6 Chapter 3 References 1. A. Reuther, D. N. Nikogosyan and A. Laubereau, J. Phys. Chem. 1996, 100, 5570- 5577. 2. A. Reuther, A. Laubereau and D. N. Nikogosyan, Opt. Commun. 1997, 141, 180-184. 3. S. H. Gong and A. Penzkofer, Opt. Quantum Electron. 1999, 31, 269-290. 4. A. Dragomir, J. G. McInerney and D. N. Nikogosyan, Appl. Opt. 2002, 41, 4365- 4376. 5. A. Dragomir, J. G. McInerney, D. N. Nikogosyan and A. A. Ruth, IEEE J. Quantum Electron. 2002, 38, 31-36. 6. D. N. Nikogosyan, A. A. Oraevsky and V. I. Rupasov, Chem. Phys. 1983, 77, 131- 143. 7. S. Pommeret, F. Gobert, A. Mostafavi, I. Lampre and J. C. Mialocq, J. Phys. Chem. A 2001, 105, 11400-11406. 8. A. Dragomir, J. G. McInerney, D. N. Nikogosyan and P. G. Kazansky, Appl. Phys. Lett. 2002, 80, 1114-1116. 9. S. Yamaguchi and T. Tahara, Chem. Phys. Lett. 2003, 376, 237-243. 10. S. Yamaguchi and T. Tahara, Chem. Phys. Lett. 2004, 390, 136-139. 11. M. Drobizhev, A. Karotki, M. Kruk and A. Rebane, Chem. Phys. Lett. 2002, 355, 175-182. 12. M. Drobizhev, A. Karotki, M. Kruk, A. Rebane, E. Nickel, C. W. Spangler and N. Mamardashvili, P Soc Photo-Opt Ins 2003, 4797, 152-162. 13. A. Karotki, M. Drobizhev, M. Kruk, C. Spangler, E. Nickel, N. Mamardashvili and A. Rebane, J. Opt. Soc. Am. B: Opt. Phys. 2003, 20, 321-332. 14. A. Rebane, N. S. Makarov, M. Drobizhev, B. Spangler, E. S. Tarter, B. D. Reeves, C. W. Spangler, F. Q. Meng and Z. Y. Suo, J. Phys. Chem. C 2008, 112, 7997-8004. 15. R. A. Negres, J. M. Hales, D. J. Hagan and E. W. Van Stryland, IEEE J. Quantum Electron. 2002, 38, 1205-1216. 16. R. A. Negres, J. M. Hales, A. Kobyakov, D. J. Hagan and E. W. Van Stryland, Opt. Lett. 2002, 27, 270-272. 73 17. C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys. 2009, 130, 084501. 18. R. S. Mulliken, Chem. Phys. Lett. 1977, 46, 197-200. 19. H. Reisler and A. I. Krylov, Int. Rev. Phys. Chem. 2009, 28, 267-308. 20. M. N. R. Ashfold, G. A. King, D. Murdock, M. G. D. Nix, T. A. A. Oliver and A. G. Sage, PCCP 2010, 12, 1218-1238. 21. T. K. Sham, B. X. Yang, J. Kirz and J. S. Tse, Physical Review A 1989, 40, 652-669. 22. M. B. Robin, Higher excited states of polyatomic molecules. Academic Press: New York, 1974; p 3 v. 23. J. M. Jung and H. Gress, Chem. Phys. Lett. 2002, 359, 153-157. 24. B. M. Cheng, M. Bahou, W. C. Chen, C. H. Yui, Y. P. Lee and L. C. Lee, J. Chem. Phys. 2002, 117, 1633-1640. 25. F. P. Larkins and A. J. Seen, Phys. Scr. 1990, 41, 827-832. 26. D. R. Salahub and C. Sandorfy, Chem. Phys. Lett. 1971, 8, 71-74. 27. A. J. Seen and F. P. Larkins, Journal of Physics B-Atomic Molecular and Optical Physics 1992, 25, 4811-4822. 28. P. Vaterlein, R. Fink, E. Umbach and W. Wurth, J. Chem. Phys. 1998, 108, 3313- 3320. 29. K. R. Wilson, M. Cavalleri, B. S. Rude, R. D. Schaller, T. Catalano, A. Nilsson, R. J. Saykally and L. G. M. Pettersson, J. Phys. Chem. B 2005, 109, 10194-10203. 30. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum. 2003, 74, 4958-4960. 31. P. R. Monson and W. M. McClain, J. Chem. Phys. 1970, 53, 29-37. 32. W. M. McClain, J. Chem. Phys. 1971, 55, 2789-2796. 33. P. R. Monson and W. M. McClain, J. Chem. Phys. 1972, 56, 4817-4825. 34. W. M. McClain, Acc. Chem. Res. 1974, 7, 129-135. 35. T. I. Quickenden and J. A. Irvin, J. Chem. Phys. 1980, 72, 4416-4428. 74 36. UV Atlas of Organic Compounds. Plenum Press: New York, 1966. 37. M. Rasmusson, A. N. Tarnovsky, E. Akesson and V. Sundstrom, Chem. Phys. Lett. 2001, 335, 201-208. 38. T. Shida and Y. Takemura, Radiat. Phys. Chem. 1983, 21, 157-166. 39. S. Tagawa, N. Hayashi, Y. Yoshida, M. Washio and Y. Tabata, Radiat. Phys. Chem. 1989, 34, 503-511. 40. L. D. A. Siebbeles, U. Emmerichs, A. Hummel and H. J. Bakker, J. Chem. Phys. 1997, 107, 9339-9347. 41. R. Mota, R. Parafita, A. Giuliani, M. J. Hubin-Franskin, J. M. C. Lourenco, G. Garcia, S. V. Hoffmann, N. J. Mason, P. A. Ribeiro, M. Raposo and P. Limao-Vieira, Chem. Phys. Lett. 2005, 416, 152-159. 42. J. M. Heller, R. N. Hamm, R. D. Birkhoff and L. R. Painter, J. Chem. Phys. 1974, 60, 3483-3486. 43. J. M. Jung, Chem. Phys. Lett. 2002, 366, 67-72. 44. Y. P. Kuo, H. C. Lu, Y. J. Wu, B. M. Cheng and J. F. Ogilvie, Chem. Phys. Lett. 2007, 447, 168-174. 45. J. B. Nee, M. Suto and L. C. Lee, Chem. Phys. 1985, 98, 147-155. 46. T. Tokushima, Y. Harada, O. Takahashi, Y. Senba, H. Ohashi, L. G. M. Pettersson, A. Nilsson and S. Shin, Chem. Phys. Lett. 2008, 460, 387-400. 47. S. Kashtanov, A. Augustson, J. E. Rubensson, J. Nordgren, H. Agren, J. H. Guo and Y. Luo, Physical Review B 2005, 71. 48. H. Farman, J. C. Dore, M. C. Bellisentfunel and D. G. Montague, Mol. Phys. 1987, 61, 583-596. 49. A. V. Anikeenko, A. V. Kim and N. N. Medvedev, J. Struct. Chem. 2010, 51, 1090- 1096. 50. J. W. Raymonda, J. Chem. Phys. 1972, 56, 3912-&. 51. B. A. Heath, N. A. Kuebler and M. B. Robin, J. Chem. Phys. 1979, 70, 3362-3368. 52. J. M. Jung and H. Gress, Chem. Phys. Lett. 2003, 377, 495-500. 75 53. G. Venturi, F. Formisano, G. J. Cuello, M. R. Johnson, E. Pellegrini, U. Bafile and E. Guarini, J. Chem. Phys. 2009, 131. 54. R. G. Snyder and Y. Kim, J. Phys. Chem. 1991, 95, 602-610. 55. A. P. Hitchcock, D. C. Newbury, I. Ishii, J. Stohr, J. A. Horsley, R. D. Redwing, A. L. Johnson and F. Sette, J. Chem. Phys. 1986, 85, 4849-4862. 56. A. P. Hitchcock and I. Ishii, J. Electron. Spectrosc. Relat. Phenom. 1987, 42, 11-26. 57. S. Tachibana, Y. Morisawa, A. Ikehata, H. Sato, N. Higashi and Y. Ozaki, Appl. Spectrosc. 2011, 65, 221-226. 76 Chapter 4. Linking Photochemistry in the Gas and Solution Phase: S-H bond Fission in p-Methylthiophenol Following UV Photoexcitation 1 4.1 Introduction The link between gas and solution phase excited state reaction dynamics is becoming much better understood now that measurements in liquids are being achieved with time resolution as fast as the intrinsic solvent motions. Such experiments capture many aspects of the initial reaction dynamics of gas phase prototype reaction systems. 2-10 For I 3 − , excitation directly to the dissociative potential energy surface (PES), in one channel, leads to ultrafast bond cleavage and formation of vibrationally excited I 2 − ( 2 Σ u + ) photoproducts plus I( 2 P 3/2 ) atoms in both the gas and solution phases. 2,11 Photoexcitation of ICN in the gas phase cleaves the I−C bond and, due to the 3 Π 0+ / 1 Π 1 conical intersection (CI) (at bent geometries on the excited PES), a large torque is imparted into the I−CN bending motion, which maps through into highly rotationally excited CN fragments plus I( 2 P 3/2 ) products. 12,13 Ultrafast pump-probe studies of ICN photolysis in bulk water or ethanol reveal that the reaction still produces freely rotating CN fragments, with a similar nascent energy distribution. Further, and contrary to statistical-mechanical predictions based on linear response, this product rotational excitation is shown to survive for up to ~10 ps before the solvent structure is restored and is able to dissipate the energy released by the reaction. 8,9 The solvent provides several additional dynamical complexities to the collision free dynamics of the gas phase, and the particulars depend strongly on the specifics of the reaction, the solute-solvent coupling and the range of fluctuations in the solvent structure. The solvent may develop friction to the translational 77 recoil of products exiting the transition state, 14 change the locations and angles of approach to CIs, as well as induce more intuitively understandable effects such as solvent caging of recoiling reaction products or providing a sink to effect vibrational energy relaxation (VER). 15-18 Finally, at longer times, diffusion of the nascent photoproducts can lead to geminate recombination. 15,19 For example, the quantum yield for O−Cl bond cleavage in the photodissociation of CH 3 OCl is reduced from near unity in vacuum 20 to 0.3-0.7 in chlorinated solvents as a result of diffusive geminate recombination. 19 The nascent Cl atom also forms complexes with the solvent, reacts with other CH 3 OCl molecules and abstracts an H atom from the solvent. 19 The last decade has seen a surge in experimental and theoretical gas-phase studies of the radiative and non-radiative relaxation pathways of prototypical heteroaromatic molecules. In the gas phase, photoexcited molecules can fluoresce back to the ground state (S 0 ), or undergo a range of non-radiative decay processes including: internal conversion (IC) to S 0 , intersystem crossing (ISC) to a lower lying triplet state (which might subsequently phosphoresce to S 0 ) or dissociate into radical products. Even if flux manages to couple efficiently to a repulsive excited PES, it can be “de-activated” at CIs en route to dissociation and transfer to S 0 . 21 One research vein has concentrated specifically on photodissociation pathways of such heteroaromatic molecules and the role of the 1 πσ* excited states associated with the X−H (X = O, N, S) bond fission. 22-25 Once the 1 πσ* state is populated, molecules experience a dissociative PES resulting in X−H bond cleavage, a large release of kinetic energy mainly partitioned into translational energy of 78 the light H atom, and vibrational excitation of the partner co-fragment (see Ref. 25 and references therein). In the majority of heteroaromatic molecules, absorption to 1 πσ* states are typically weak, and the UV absorption spectrum of these molecules is dominated by optically “bright” 1 ππ* states, which are diabatically bound within all co-ordinates. UV excitation will thus most likely result in population of 1 ππ* states and, depending upon the rates of competing excited state deactivation processes, can couple on an ultrafast timescale via CIs to the 1 πσ* state. Such couplings between PESs are often mediated by out of ring plane nuclear motions, 26 which can map into vibrational excitation of the radical co- fragment. 27 Measurements of the nascent H atom photoproducts with the high resolution afforded by the H(Rydberg) Atom Photofragment Translational Spectroscopy (HRA-PTS) technique reveal that the vibrational excitation of the radical co-fragment and thereby provide insight into the dynamics that occur on the excited PES of the parent molecule. However, the HRA-PTS technique provides no information about the absolute quantum yield for dissociation (cf. other non-radiative processes). Most condensed phase studies of the heteroaromatic nucleobases and other molecules of biological significance have sought to understand the short fluorescence lifetimes in terms of competing non-radiative decay pathways. 28,29 These ultrafast studies excite molecules into a 1 ππ* state and typically monitor either the fluorescence lifetimes, or probe the transient absorption at fixed wavelengths: measuring the excited state decay 79 rate, or the rate of VER after IC to the S 0 PES. 30 Complementary computational studies confirm the experimental findings of IC to S 0 , that is again driven by out of ring plane motions that move flux into parts of phase space where CIs link the 1 ππ* PES to S 0 . 31 In the case of thymine and uracil for example, the measured biexponential decay in solution has been explained in terms of initial population of a 1 ππ* state from which the flux bifurcates: either passing directly to S 0 via a 1 ππ*/S 0 CI, or reaching S 0 via successive 1 ππ* 1 nπ* S 0 radiationless transitions. 32 This branching behavior is similar to that discerned in gas-phase studies; 33 with the major difference that the latter environment does not provide any means for VER. 34 Fully probe-dispersed transient absorption offers a more complete picture of the early time dynamics leading to the CI in solution. 35 No experiment to date has directly demonstrated ultrafast X−H bond fission in a heteroaromatic molecule in a solution phase environment. The vast majority of ultrafast solution phase studies use pump wavelengths (λ pump = 267 nm) for which the photon energy is below the likely energetic threshold for accessing the 1 πσ* state (which in nucleobases such as adenine will typically be accessible for λ pump < 240 nm). 36,37 Thus the non-observation of excited state X−H bond fission in solution phase studies is not wholly surprising. However, the photodissociation channel of such large molecules may not be dominant in the condensed phase; the presence of solvent offers additional competing deactivation pathways that are not available in the gas phase. For example, the excited molecular electronic wavefunction can couple efficiently to the solvent continuum and, as a result, heteroaromatic molecules can ‘autoionize’ at energies well 80 below the molecular ionization potential, ejecting an electron into solution that subsequently becomes solvated. 38-41 Another excited state deactivation pathway available in the solution phase environment is proton-coupled electron transfer (PCET). Here, a proton and electron are transferred in a concerted reaction process. 42,43 Extensive theoretical calculations and experimental studies of X−H bond fission in clusters of heteroaromatic molecules with water or ammonia, provide a possible bridge between the gas and condensed phases (see Ref. 10 and references therein). Theoretical studies of H atom transfer in phenol-H 2 O and phenol-NH 3 complexes have shown a strong coupling between the phenol and the ‘solvent’ H 2 O and NH 3 molecules. In the bare molecule, the excited 1 πσ* state wavefunction is accurately described as a σ*←π excitation, where both orbitals are localized on the molecule. In the phenol-‘solvent’ complexes, this becomes a solvent ← molecule charge transfer excitation, with the σ* orbital localized on the solvent molecule. Following the minimum energy reaction co- ordinate along the 1 πσ* state PES, ab initio calculations show that the electron and proton are transferred to the ‘solvent’ in a concerted PCET process, which is endothermic in phenol-(H 2 O) n (n=1-3) but exothermic in phenol-(NH 3 ) n (n=1-3) clusters. 44 One of the most detailed experimental cluster studies to date involves UV-IR dip time-resolved picosecond spectroscopy of phenol-(NH 3 ) 3 . Exciting an S 1 level (that in bare phenol is known to predissociate to yield H atoms with high kinetic energy 27 ) in the cluster initiates O−H bond fission. The dissociating H atom forms a σ bond with the (NH 3 ) 3 cluster, which then breaks away, leaving the phenoxyl radical plus NH 4 NH 3 NH 3. The latter 81 radical subsequently rearranges to the more thermodynamically stable NH 3 NH 4 NH 3 product. 45,46 This chapter focuses on the photochemistry of p-methylthiophenol (p-MePhSH, alternatively named p-thiocresol or p-toluenethiol) following UV photoexcitation in the cyclohexane and ethanol solutions. The frequency-resolved experiments provide sufficient details of the S–H bond fission mechanisms in the gas phase. Specifically, HRA-PTS studies at photolysis wavelengths 240 ≤ λ phot ≤ 295 nm have revealed the operation of two excited state S–H bond fission mechanisms. It was discerned that between 295−271 nm, vibrational levels of the 1 1 ππ*(S 1 ) state (formed by π*←π/n electron promotion) were initially photo-prepared, dissociating rapidly via H atom tunneling under the area associated with the 1 1 ππ*(S 1 )/1 1 πσ*(S 2 ) CI 47 (see Figure 4.1(a)). At wavelength between 271 and ~240 nm, the S 2 is directly populated. At even shorter wavelength, excitation to the 2 1 ππ* state is deduced. The 1 πσ* state in the vertical Frank- Condon (vFC) region is formed initially by the 4s←π/n excitation, which subsequently evolves, upon extension of the S−H bond, into a σ*←π/n orbital promotion. 82 Figure 4.1. (a) Calculated CASPT2(10/10)/aug(S)-cc-pVTZ potential energy cuts through the S 0 , 1 1 ππ*, 1 πσ*, 2 1 ππ* states in the S−H stretch dimension (reproduced from Ref. 48), skeletal structures in (b) and (c) display transition dipole moments for p-MePhSH molecule, and in (d) for the p-MePhS radical. (e) UV absorption spectra for p-MePhSH in the vapor phase (black line), cyclohexane solution (blue line) and ethanol solution (orange line) between 310 > λ > 230 nm. The absorption intensities were normalized to the maxima at ~240 nm. At all wavelengths studied, excited state S−H bond dissociation occurs via the 1 πσ* PES with branching between the H + p-MePhS( X % ) and H + p-MePhS( A % ) product channels determined to occur at the 1 πσ*/S 0 (S 2 /S 0 ) CI. 47,49,50 The branching into the two product channels was found to be very dependent upon the pump excitation wavelength 47 and the para- chemical substituent. 47,49 Generally, out-of-plane motions facilitate the flux transfer between the S 2 and S 0 surfaces, and thus encourage the molecule to follow the adiabat upon S–H bond extension. In this case, p-MePhS radicals form in the electronically excited state ( A % ). An important out-of-plane motion which modulates the A % / X % state branching is the torsional motion of the S–H bond. 47 A % state radical is produced when the 83 S–H bond lies out-of-plane as the molecule approaches the S 2 /S 0 CI, but if the S–H bond is parallel to the benzene ring plane, the molecule will follow the diabat and dissociate to the ground state ( X % ) radical. Riyad et al. 51 reported absorption spectra of a range of substituted PhS radical products following flash photolysis of the corresponding thiophenols in three polar solvents. There are several possible mechanisms for production of p-MePhS radicals following UV irradiation of p-MePhSH in solution, however. These include autoionization, PCET and S–H bond fission. Each process will have different associated timescales and spectral signatures. Electrons generated in ethanol by ionization of p-MePhSH, for example, would take up to 20 ps to become fully solvated and then live for ~ 1 μs. Such timescales are a considerably longer than that deduced for direct S–H bond fission on the 1 πσ* of gas phase PhSH molecules. 50 All such eventualities would not have been distinguishable with the limited (μsec) time resolution employed in the earlier solution phase study from Riyad et al., 51 but should be readily apparent with the UV pump-dispersed broadband probe time-resolved transient absorption (TA) capability available in the present study. This method offers an instrument response time 46 -180 fs (at λ pump = 295, 285, 271, 267 and 200 nm), and broadband detection in the wavelength range of typically 320-650 nm. This allows us to follow directly the time evolution of both the parent excited state population and the resulting products, and thus provides an excellent probe to unravel the relative importance of competing dissociation and autoionization processes in solution. Further, the current pump-probe study provides a means to directly probe the absorption 84 of an excited state radical. This cannot be achieved in the flash photolysis and pulse radiolysis experiments due to their inferior overall instrument response time. Thus, by monitoring the A % / X % state branching one can infer how the p-MePhSH molecule approaches the S 1 /S 2 CI with the presence of surrounding solvent molecules. Lastly, geminate recombination of the photofragments in the liquid environment provides information on the initial separation of the partners, which is a strong function of the total kinetic energy release (TKER). Thus, using the gas phase HRA-PTS results as a starting point, the condensed phase photochemistry of p-MePhSH can be study in far greater detail – we can not only discern the competing pathways (if any) being turned on in the solution, but also examine if the solvent molecules changes the location of the CIs, and how the excited molecule approaches the CIs. 4.2. Experimental In the TA experiments, the 800 nm fundamental (1 kHz, Coherent Legend) was converted to 271 nm by four-wave mixing in a hollow core fiber. 52 The resulting pulses had a bandwidth of 4 nm, and were subsequently compressed by a pair of Gires-Tournois Interferometer dispersion mirrors, 53 generating a 30 fs (FWHM) pulse with very little higher order dispersion, as determined by auto-correlation. Photolysis was also carried out at 295, 285, 267 and 200 nm. The 285 and 295 nm pump pulses were obtained by sum-frequency mixing and subsequent frequency doubling of the output from a two-stage optical parametric amplifier using a type-II phase matching process (OPA-800C, Spectra- Physics). The pre-requisite 570 nm and 590 nm pulses were generated by mixing the 800 85 nm fundamental with the idler pulse from the OPA in a type-I BBO crystal. The visible output was subsequently frequency-doubled to obtain the desired UV pulse, which in both cases had a bandwidth of ~3 nm and duration of ~100 fs. 267 and 200 nm pulses were obtained by 3 rd and 4 th harmonic generation by sum-frequency mixing in type I BBO crystals and had bandwidths = 2 and 1 nm, and pulse widths of 80 fs and 130 fs, respectively, as determined by cross-correlation. All experiments employed average pulse energies < 1.2 μJ and a spot size of 100 – 250 μm. The signal from neat ethanol was negligible under these experimental conditions. The transient signal of p-MePhSH was linearly dependent on the pump fluence, as determined by a power dependence study in which the 267 nm pump pulse energy was varied across the range 0.8-8 μJ. Alternate pump pulses were blocked by a chopper operating at 500 Hz. The weak super-continuum probe (typically 310 ≥ λ probe ≥ 650 nm) was generated by focusing the 800 nm fundamental into a rotating CaF 2 disc. A pair of off-axis aluminum- coated parabolic mirrors was used to collimate and then focus the super continuum into the sample. The super continuum was dispersed onto a 256-pixel silicon diode array, where signals with and without the pump beam present were recorded for a transient absorption to be obtained. The relative polarization of the pump and probe was controlled by rotating a zero-order half wave plate in the 800 nm beam driving the continuum generation. The polarization purity of the continuum was determined to be >150:1 across the whole probe range, and that of the pump pulses >100:1, as measured by the extinction of the light through a calcite polarizer. Anisotropy spectra were constructed from TA 86 spectra recorded sequentially with parallel and perpendicular polarizations; analysis of such spectra recorded at different time delays provides one route to investigating the time dependence of the product anisotropy. The decay of product anisotropy at selected wavelengths was also determined from simultaneously acquired parallel and perpendicular TA signals. This was achieved by rotating the super-continuum probe polarization by 45˚ with respect to that of the UV pump pulse, then using an interference filter after the pulses have interacted with the sample to select the desired wavelengths (typically with a bandwidth of 10 nm), with the s- and p- components spatially separated by a Wollaston prism and focused onto a 256-pixel photodiode array. The anisotropy (R(t)) was calculated from eqn. (1) || || ( ) 2 A A R t A A ⊥ ⊥ Δ − Δ = Δ + Δ , (1) where || A Δ and ⊥ ΔA represent TA signals recorded at time t with the ε ε ε ε vectors of the pump and probe pulses aligned, respectively, parallel and perpendicular to one another. For broadband measurements, the temporal chirp of the continuum was corrected mathematically by setting the two-photon absorption peak at each probe wavelength to time zero. 54 The sample was delivered by a wire guided gravity jet, forming a thin film of thickness 50-70 μm at the detection region. The flow rate of the liquid ensures that a fresh sample was interrogated every pump pulse. 55 Various concentrations (10 mM, 45mM and 90 mM) of p-MePhSH (97%, TCI America) were used in ethanol (200 Proof, Koptec) and cyclohexane (>99.99%, spectrophotometric grade, EMD) without further purification. 87 Steady state absorption spectra of these solutions were measured in a Cary 50 UV-Visible spectrophotometer (Varian). 4.3. Results 4.3.1 UV-Visible Spectra Static room temperature UV absorption spectra of p-MePhSH in the gas phase, and in solution in cyclohexane and ethanol are displayed in Figure 4.2(a). Figure 4.2(b) shows the 255 – 310 nm region of interest on an expanded scale, along with illustrative depictions of the center wavelengths and bandwidths of the pump pulses used in the present work. The vapor phase spectrum rises gently once λ < 300 nm, and shows some diffuse structure at long wavelengths which is attributed to 1 1 ππ*←S 0 excitation. 47 The absorption cross-section starts to rise more steeply at λ~260 nm and λ~245 nm. Guided by the results of electronic structure calculations and of H (Rydberg) atom photofragment translational spectroscopy studies of this same molecule in the gas phase it is reasonable to associate these stronger absorptions with, respectively, the 1 1 πσ*←S 0 and 2 1 ππ*←S 0 excitations. 47 The longer wavelength 1 ππ*←S 0 absorption features in the UV absorption spectra of p-MePhSH in cyclohexane and ethanol solution both appear red-shifted for 2 – 3 nm relative to the vapor phase spectrum. Remnants of the vibrational structure associated with the 1 1 ππ*←S 0 transition are evident in cyclohexane and, to a lesser extent, in ethanol. The p-MePhSH in cyclohexane spectrum also shows a (red-shifted) absorption shoulder reminiscent of that observed (in the gas phase) at ~255 nm, but there is no evidence for an obvious feature in the p-MePhSH in ethanol spectrum – hinting that the 88 polar solvent has a differential effect on the oscillator strength and/or center wavelength of the 1 πσ*←S 0 transition. The peak of the absorption band originated from the 2 1 ππ*←S 0 transition is red-shifted by 7 nm from gas to solution phase. Figure 4.2 (a) UV absorption spectra for p-MePhSH in the vapor phase (blue line), cyclohexane solution (orange line) and ethanol solution (green line) between 310 > λ > 210 nm. The absorption intensities were normalized to the maxima at ~240 nm. (b) Zoomed version of (a) between 250 > λ > 310 nm with the four Gaussian functions underneath the absorption spectra representing the wavelengths of the excitation pulses and their corresponding bandwidth. 89 Figure 4.3 (a) TA spectra of 90 mM p-MePhSH in ethanol measured with 267 nm at selected pump-probe delay times, with pump-probe polarization set to magic angle. (b) Time profiles of major TA features measured at λ probe = 363, 500 and 600 nm. Inset: kinetics on an expanded scale, focusing on the early time. (c) Anisotropy spectra at selected pump-probe delay times, obtained from sequential parallel and perpendicular polarization experiments. (d) Anisotropy decay as a function of time obtained from sequential parallel and perpendicular polarization experiments. Inset: anisotropy decay on an expanded scale, focusing on the early time. 4.3.2 Transient UV-Visible Absorption in Ethanol and Cyclohexane 4.3.2.1 p-MePhSH in Ethanol Solution 4.3.2.1.1 267 and 271 nm Excitation and Band Assignments Transient absorption datasets were acquired at total five pump wavelengths (λ pump = 295, 285, 271, 267 and 200 nm) and probed with a white light continuum pulse (typically 310 90 < λ probe < 650 nm). Figure 4.3(a) displays TA data taken at λ pump = 267 nm, at a range of pump-probe time delays, t. Within the instrument response time (~150 fs) the TA contains three prominent absorption features: a double peaked feature across the range 425 < λ probe < 550 nm (with centers at λ probe = 465 and 500 nm), a feature that maximizes outside the probe wavelength range, but is clearly visible at λ probe ~320 nm, and an underlying absorption that spans the whole probe wavelength range monitored. As evident in Figure 4.3(b), the 465 and 500 nm (and ~320 nm) features decay within t ~60 ps. A new feature centered at ~385 nm appears at t > 5 ps, maximizing on a similar ~60 ps timescale. The relative intensities of these absorption features were found to be independent of sample concentration. The 465 and 500 nm features match well with those reported for the p-MePhS( X ~ ) radical in earlier pulsed radiolysis studies of p-MePhSH in ethanol, 56 as does the feature at ~320 nm (the majority of this band is below the short wavelength detection limit of the present experiment). These absorptions are thus taken to indicate p-MePhS( X ~ ) radical formation. It is unlikely that any electronically excited p-MePhS( A ~ ) products will be observed in pulsed radiolysis experiments, and thus the absorption spectrum for this species is unknown. Careful consideration must be taken in discerning the mechanism for the generation of p-MePhS radicals. One obvious possibility is that, as in the gas phase at λ pump = 267 nm, the S−H bond fission on the 1 πσ* state occurs, producing the geminate partners: p-MePhS radical plus an H atom. Other possible routes are autoionization to form p-MePhSH + and a solvated electron (potentially followed by delayed proton transfer 91 to the solvent), or PCET to the solvent. Either process should generate solvated electrons which have a broad depolarized absorption spectrum at room temperature, rising across the wavelength range 400 < λ probe < 650 nm to a peak at ~690 nm. 57 Whilst autoionization should occur on a femtosecond timescale in ethanol, the electron typically takes t > 10 ps to become solvated. 58 However, the TA signal between 575 < λ probe < 650 nm is flat at all pump-probe time delays; comparison of parallel and perpendicular data showed a positive anisotropy (Figure 4.3(c) and (d)). The source of this TA is discussed later, but neither its rise time, nor its polarization, nor its subsequent spectral evolution is consistent with solvated electron production. Moreover, the ionization product p- MePhSH + has a distinct absorption feature centered at 450 nm with an estimated maximum extinction coefficient of ~1800 M -1 cm -1 , † and a lifetime of at least hundred of ns in cyclohexane, as observed in pulse radiolysis experiments 56 − and no such feature is observed in our TA spectra at any delay time, indicating that autoionization does not occur to any significant extent in these experiments. Given the absence of absorption attributable to solvated electrons and p-MePhSH + , we rule out any photoionization mechanisms, and conclude that the 425 < λ probe < 550 nm TA feature is the signature of p-MePhS radicals produced by S−H bond fission. The widths of the TAs assigned to the p-MePhS radical between 425 < λ probe < 550 nm and ~320 nm contract over the first few picoseconds. This can be attributed to vibrational cooling of the hot nascent radical (cf. the 267 nm gas phase data) with intermolecular † Value taken from phenol cation in water, see T. N. Das, J. Phys. Chem. A 2005, 109, 3344-3351. 92 solute-solvent H-bonds providing an efficient vibrational energy transfer pathway. 30,59 The ~320 nm absorption appears broader (at t < 5 ps) than the longer wavelength features but, since only part of the UV feature lies within our λ probe window, it is difficult to quantify the full extent of this. Vibrationally excited ground state parent molecules formed by IC from the photoexcited states to S 0 may also contribute to TA spectra at λ probe < 340 nm. By analogy with previous studies of DNA bases, 30 hot ground state molecules are likely to absorb at slightly longer wavelengths than the onset for parent absorption (~290 nm in p-MePhSH) and to decay by VER on a picosecond timescale. Table 4.1. Calculated CASPT2(9/8)/AVTZ excitation energies and CASSCF(9/8)/AVTZ TDMs for the p-MePhS radical. Transition Orbital Promotion Excitation energy/ eV * Excitation wavelength/ nm * Transition Dipole Moment/ Debye # A ~ ← X ~ b 1 (π)←b 2 (n) 0.42 2992 0.000 B ~ ← X ~ b 1 (π)←b 1 (π/n) 2.74 (2.57) 452 (483) 1.05 C ~ ← X ~ b 1 (π)←a 2 (π) 2.77 447 0.29 D ~ ← X ~ b 1 (π ∗ )←b 2 (n) 4.05 306 0.001 E ~ ← X ~ a 2 (π ∗ )←a 2 (π) 4.39 (3.91) 283 (317) 0.66 F ~ ← X ~ a 2 (π ∗ )←b 2 (n) 4.60 270 0.01 B ~ ← A ~ b 2 (n)←b 1 (π/n) 2.32 543 0.001 C ~ ← A ~ b 2 (n)←a 2 (π) 2.35 527 0.007 D ~ ← A ~ a 2 (π ∗ )←b 1 (π) 3.63 (3.41) 341 (363) 0.112 F ~ ← A ~ a 2 (π ∗ )←b 1 (π) 4.18 297 0.341 *Numbers in parenthesis represent the optimized adiabatic excitation energy for a given transition # The CASSCF/CASPT2 calculations employed here use the C s point group, the transition labels assume free-rotation of the methyl group and thus C 2V symmetry. 93 The ab initio calculations, carried out by our collaborator Tom Oliver at the University of Bristol, provide further confirmation that the prominent features centered at λ probe = 500, 465 and ~320 nm are due to p-MePhS( X ~ ) radical absorptions. As Table 4.1 shows, the CASPT2(9/8)/AVTZ calculations predict electronic transitions of isolated p-MePhS( X ~ ) radicals with significant oscillator strength at 483, 447 and 320 nm, corresponding to the B ~ ← X ~ (adiabatic), C ~ ← X ~ (vertical) and E ~ ← X ~ (adiabatic) excitations respectively. All three match (within 0.3 eV) the maxima of the strongest features observed in the early time solution phase TA spectra recorded at λ pump = 267 nm. CASSCF geometry optimizations of the p-MePhS( E ~ ) state present one possible reason for the observed greater breadth of the E ~ ← X ~ transition: the calculated geometry change associated with this π*←π transition is greater than that accompanying the C ~ ← X ~ (π←π) or B ~ ← X ~ (π←π/n) excitations. However, the calculated TDMs do not match the relative intensities of the TA features. The B ~ ← X ~ transition is calculated to be ~10× stronger than the C ~ ← X ~ transition – a much greater difference than observed experimentally. This can be reconciled by assuming vibronic mixing between the energetically close B ~ and C ~ states of the p- MePhS radical, with the result that the weaker C ~ ← X ~ transition gains intensity from the more intense B ~ ← X ~ excitation. On the other hand, the discrepancy between our experimental spectrum and the calculated TDM could be suggestive of other transient species contributing to the absorption in this region. This has some ground as the 94 spectrum of the nascent p-MePhS( X ~ ) radical is slightly different from those at longer delay time (although the small spectral difference can be explained by vibrational cooling as discussed above) – we will return to discuss this point with p-MePhSH in cyclohexane. Our interpretation of the TA measurements thus far has ignored the possible presence of any nascent p-MePhS( A ~ ) radical products which, if the gas phase behavior was to map into the solution phase, would be expected to be produced in a similar yield to that of p- MePhS( X ~ ) at λ pump = 267 nm. The calculated TDMs for absorption originating in the p- MePhS( A ~ ) radical are much weaker than those for p-MePhS( X ~ ) – see Table 4.1. Within our detection probe wavelength range, only the weak D ~ ← A ~ absorption is expected at ~363 nm. Such feature is not clearly discernable in the current λ pump = 267 nm data, possibly due to the fast electronic quenching of the A ~ state radical. However, with a superior instrument response time afforded by the 271 nm pulse via 4WM, it might be possible to resolve this fast decaying feature – see Figure 4.4(a) and corresponding discussions. Having established that the TA features centered at λ probe = 465 and 500 nm, and the partial peak at λ probe ~320 nm, originate from p-MePhS( X ~ ) radicals, only the 350 < λ probe < 425 nm absorption centered at λ probe = 385 nm remains to be assigned. This feature is only evident at t > 5 ps, and thereafter gains in intensity and is most pronounced at t = 60 ps (see Figure 4.3(b)). This tens of picosecond rise time coincides with the decay of the 95 465 nm and 500 nm absorptions of the p-MePhS( X ~ ) radical, which are thus deduced to decay due to geminate recombination of H atoms with the p-MePhS fragment. If the dissociated S−H bond reforms, thereby recreating the parent molecule, this would refill the bleach at λ probe < 300 nm (outside of the probe range of this experiment). Alternatively, the H atom could recombine with a different part of the radical, e.g. on the benzene ring via a weak van der Waals interaction or by formation of a σ bond, which in either case, we henceforth denote p-MePh(H)S. Riyad et al. observed a feature centered at λ probe ~350 nm in pulsed radiolysis studies of PhSH in acidified (pH = 1.5) aqueous solution. This was assigned to an H atom (or an OH radical, required in the pulsed radiolysis to create the radical) bound to the parent molecule p-MePhSH ring. 51 As noted previously, the rise of this feature coincides with the decay of the p-MePhS( X ~ ) radical at λ probe = 500 and 465 nm – encouraging tentative assignment of this feature to a p- MePh(H)S adduct. We have also considered the possibility that the ~385 nm TA is the result of a nascent H atom abstracting an H atom from an ethanol solvent molecule (the corresponding process in water has a rate constant of 2×10 7 M -1 s -1 (Ref. 60)). A slightly smaller rate constant is expected for this reaction in ethanol, due to more favorable H atom solvation. Such an abstraction process would thus not be expected to occur on timescales faster than ~3 ns, and thus we rule out assigning any TA within our pump- probe time delay window to H abstraction from solvent. From Figure 4.3(c) we can see that the anisotropy spectra at selected delay times display a positive anisotropy across the whole probe wavelengths region, with a non-limiting 96 value of ~+0.17 (the limiting values are 0.4 and -0.2 for parallel and perpendicularly transitions, respectively). Moreover, a slightly more positive initial anisotropy is observed over the p-MePhS( X ~ ) radical absorption region 425 < λ probe < 550 nm. Both of these factors indicate that the initial excitation has a TDM parallel (or pseudo parallel) to the intense B ~ ← X ~ excitation of the radical TDM. Inspection of Figure 4.1 (b-d) shows that only the parent S 3 can be responsible for such a positive anisotropy, and thus implies, relative to S 2 excitation inferred in gas phase experiments, dominant S 3 excitation in ethanol solution. Figure 4.4 (a) TA spectra of Figure 90 mM p-MePhSH in ethanol measured with 271 nm at selected pump-probe delay times, with pump-probe polarization set to magic angle. (b) Time profiles of major TA features measured at λ probe = 363, 500 and 600 nm. Inset: kinetics on an expanded scale, focusing on the early time. In order to explore the early time dynamics, further experiments with a short pulse hollow core fiber source were performed, providing a significantly faster instrument response time (43 fs, λ pump = 271 nm). From Figure 4.4(a), it is clear that the ground state p-MePhS radical is observable at t = 100 fs, and that efforts to determine the radical 97 formation and decay kinetics by following the TA at any given λ probe (Figure 4.4(b)) will be compromised by the evolving underlying absorption. However, a feature centered at ~363 nm is clearly discernable with the aid of the faster instrument response time. This feature coincides well with the calculated D ~ ← A ~ transition energy but decays on a time scale of sub picoseconds. Thus, it is very tempting to assign this feature to the excited state radical absorption. However, since the decay time is very close to the instrument response time, it is very challenging to exclude the contribution from 2-photon absorption (2PA) signals. Appendix M presents evidence which supports our initial assignment of A ~ radical absorption. Figure 4.5 (a) TA spectra of 90 mM p-MePhSH in ethanol measured with 285 nm at selected pump-probe delay times, with pump-probe polarization set to magic angle. (b) Time profiles of major TA features measured at λ probe = 363, 500 and 600 nm. Inset: kinetics on an expanded scale, focusing on the early time. 98 Figure 4.6 (a) TA spectra of 90 mM p-MePhSH in ethanol measured with 295 nm at selected pump-probe delay times, with pump-probe polarization set to magic angle. The spectra were averaged over five delay time points to increase S/N. (b) Time profiles of major TA features measured at λ probe = 363, 500 and 600 nm. Inset: kinetics on an expanded scale, focusing on the early time. 4.3.2.1.2 285 and 295 nm Excitation Figures 4.5(a) and 4.6(a) present the corresponding TA spectra measured in ethanol, at five different pump-probe time delays, at λ pump = 285 and 295 nm, respectively. As at the shorter pump wavelengths, these TA spectra show formation of p-MePhS( A % ) and p- MePhS( X % ) radicals, and ESA from S 1 , within the instrument response time (~150 fs at these wavelengths). It is noteworthy that adduct band centered at ~385 nm seems to be more intense (when compare to the p-MePhS( X % ) radical band centered at 500 nm) than that in shorter pump wavelength. This implies that there is more geminate recombination at λ pump = 295 nm. As a result, more adduct and less p-MePhS( X % ) radical is observed. Figures 4.5(b) and 4.6(b) shows that similar to shorter pump wavelengths, the p- MePhS( A % ) absorption feature and the ESA signal at λ probe ~600 nm decline rapidly with 99 time, and (apart from the small long lived (ns lifetime) component within the latter) have essentially disappeared within ~2 ps. Figure 4.7 (a) Anisotropy spectra at early delay time (200 fs), obtained from sequential parallel and perpendicular polarization experiments. (b) Anisotropy decay for p-MePhS( X % ) radical band centered at 500 nm, obtained from simultaneous parallel and perpendicular polarization experiments utilizing the Wollaston prism. The traces with 271 nm excitation were not shown since it is identical to the 267 nm excitation within the experimental error. Figure 4.7 (a) shows the anisotropy spectra at 200 fs for λ pump = 267, 285 and 295 nm (200 fs is the earliest time without 2PA contribution), calculated using eqn. (1). At the shortest excitation wavelengths, the early time anisotropy is positive across the entire probe wavelength range. As at 267 nm, the anisotropy at λ pump = 271 nm is positive at all probe wavelengths investigated (data not shown). Anisotropy spectra recorded at λ pump = 285 nm return R(t = 200 fs) ~ 0 at all probe wavelengths. The corresponding data measured at λ pump = 295 nm indicates that the product anisotropy is negative (R(t = 200 fs) ~ –0.05) at all probe wavelengths. Fig. 4.7 (b) shows the anisotropy decay for the p- 100 MePhS( X % ) radical band centered at 500 nm as a function of delay time. At λ pump = 267 nm, the starting value of anisotropy is ~+0.17 and it decays to 0 by 30 ps. Consistent with the anisotropy spectra, the starting value decreases with the decreasing pump photon energy, which is suggestive of decreasing contribution from the S 3 state. 4.3.2.1.3 200 nm Excitation Figure 4.8 (a) displays TA spectra measured at several different pump-probe time delays following excitation of p-MePhSΗ in ethanol at λ pump = 200 nm (instrument response time ~180 fs). These spectra contain the same absorption features as observed at λ pump = 267 and 271 nm. The most obvious difference is that the radical absorption bands are all far broader at early time (t <5 ps). The B ~ ← X ~ and C ~ ← X ~ bands now span the range 375 < λ probe < 675 nm and the E ~ ← X ~ band encroaches on the C ~ ← X ~ absorption. The much increased width of the radical absorptions is consistent with the gas phase measurements at a similar short wavelength (λ pump = 193 nm), which revealed a much greater partitioning of available energy into product vibration. Also noteworthy is the time evolution of TAs at λ probe = 500 and 465 nm (the B ~ ← X ~ and C ~ ← X ~ radical bands respectively) which appear to display different kinetics (c.f. λ pump = 267 and 271 nm), as shown in Figure 4.8 (b). This is not surprising given the TA spectra involve several overlapping absorptions. For example, the 465 nm TA measured at early times in Figure 4.8(b) is a superposition of the underlying parent ESA, the C ~ ← X ~ absorption of the radical and also (due to the high degree of product vibrational excitation) overlapping 101 contributions from the radical B ~ ← X ~ and E ~ ← X ~ absorptions. At t > ~5 ps, the parent ESA decays to zero, and the radical absorptions narrow spectrally due to VER, and the p- MePhS absorption decreases as a result of geminate recombination. TA spectra acquired at λ pump = 200 nm were also checked carefully for evidence of solvated electrons at 400 < λ probe < 650 nm. As at λ pump = 267 nm, no absorption at any delay attributable to solvated electrons was discerned. In the case of p-MePhSH in ethanol, therefore, even at photoexcitation energies approaching the ionization threshold (which we can predict to be somewhat below the threshold gas phase value for PhSH (8.3 eV) 61 ), photodissociation surprisingly still dominates over autoionization into the solvent continuum. Figure 4.8 (a) TA spectra measured at different pump-probe time delays following excitation at λ pump = 200 nm (with the pump and probe laser polarizations aligned perpendicular to each other); (b) time profiles for major TA features measured at λ probe = 500, 465 and 385 nm. 102 4.3.2.2 p-MePhSH in Cyclohexane Solution 4.3.2.2.1 271 nm Excitation Fig. 4.9(a) displays representative TA spectra of p-MePhSH in cyclohexane solution with 70 fs 267 nm pump pulse (generated from the hollow core fiber, but the center wavelength is slightly blue shifted, and the pulse duration is longer, possible due to a poorer compression). Features attributable to p-MePhS( X % ) radicals, centered at λ probe = 500, 450 and < 320 nm, 51,62 are clearly evident immediately within the overall instrument response time (~100 fs) consistent with prompt dissociation. The band assignments and amplitudes are in accord with transition wavelengths computed by CASPT2 energies and CASSCF transition dipole moments (see Table 4.1) for the gas phase radical (purple bars shown at base of Figure 3). 47 The radical electronic absorption bands measured at any given pump-probe time delay in cyclohexane are sharper than in the equivalent TA spectrum in ethanol measured previously. 48,62 Such behavior reflects the weaker solute- solvent interactions in the non-polar cyclohexane. 63,64 103 Figure 4.9 (a) TA spectra measured at selected delay times following 267 nm photolysis of p- MePhSH in cyclohexane, with the pump-probe polarization set to magic angle. (b) TA spectra measured following 200 nm photolysis of p-MePhSH in cyclohexane (magic angle polarization). The squares of the calculated transition dipole moments (TDM 2 ) for transitions originating from S 1 (ESA) and from p-MePhS radicals are also shown (as bars in each panel). The radical transitions are from CASPT2/ CASSCF calculations of Ref. 62 and the S 1 band predictions are from the current EOM-CCSD calculations. (c, d) Geminate recombination analysis from the radical (red) and adduct (blue) kinetics after 267 nm and 200 nm photodissociation, respectively. The lines represent the survival probabilities, Ω(t), of the radical and the population rise of the adduct, as predicted by a radical-radical geminate recombination model and using physical constants described in Ref. 62. From this analysis, the average ejection lengths 〈r 0 〉 are shown. Over longer time delays (t > 5 ps), the absorption assigned to the p-MePhS( X % ) radical decays and a transient centered at ~380 nm appears which, similar to ethanol, originates from adduct formation. 48,62 The appearance kinetics of this feature, as well as the decay kinetics at 500 nm, are insensitive to the p-MePhSH concentration (data not shown) − 104 consistent with a product formed from geminate recombination of the primary radical pair. We have previously assigned the 450 nm shoulder to p-MePhS radical C % ← X % transition, which is predicted to have weak absorption strength by CASPT2(9/8)/AVTZ calculations, arguing it gained intensity by vibronic mixing with the stronger B % ← X % transition centered at ~500 nm. 47 In cyclohexane, unlike in ethanol, it is apparent that the 450 nm side of the band decays to about 50% of the height of the 500 nm band by about 50 ps. Thus, the different kinetics at the two marker wavelengths suggests they cannot trace the same population and there must be a contribution of another short-lived species at 450 nm. The electronic structure calculations presented here are helpful in understanding this aspect of the spectral evolution. The EOM-CCSD/aug-cc-pVTZ calculations suggest a series of very bright transitions originating in the excited S 1 state within the broadband continuum range, consistent with the idea of a “broad underlying absorption” assigned to S 1 ESA earlier, 47 but highlight an S 1 transition centered at ~430 nm with the most prominent TDM. This transition is close to the 450 nm shoulder observed in our TA experiments, and thus the latter most likely has contribution from S 1 ESA (as well as p- MePhS C % ← X % ) and explains the complicated evolution of the spectrum in this region. The only part of the transient spectrum that perhaps can be cleanly assigned to excited state absorption is at wavelengths near 600 nm and longer. Most of this signal decays rapidly within 1 ps; signal to noise precludes assessing whether any small part of this absorption is long-lived (hundreds of picoseconds) as seen in our earlier studies in 105 ethanol. The order of magnitude difference in the calculated TDM strengths for ESA compared to product absorption suggest that very little actual population is created or gets trapped on S 1 , its existence is merely amplified by its much larger oscillator strength compared to the p-MePhS radical absorptions. 4.3.2.2.2 200 nm Excitation Figure 4.9 (b) displays representative TA spectra when p-MePhSH is photodissociated at 200 nm. The spectral signatures for the p-MePhS( X % ) radical at 425–500 nm are readily observed within the 180 fs instrument response time, suggesting again ultrafast S–H bond fission. The shape of the strongest of the ground state radical bands rising below 320 nm 51,62 is better seen in this dataset. We have previously assigned this as E % ← X % band based on CASPT2 calculations. 62 Moreover, the initial contour between λ = 425–500 nm is much broader compared to 267 nm excitation, which strongly indicates that the p- MePhS( X % ) radicals are formed with significant vibrational excitation, and subsequently cool down on a ~5 ps time scale. The TA signal intensity ratio of the p-MePhS( X % ) radical absorption at 500 nm to the adduct absorption previously noted at ~380 nm is larger with 200 nm excitation, (e.g., compare the 50 ps TA spectra in figure 4.9 (a) and (b)). This suggests that the geminate recombination noted at 267 nm is much less efficient in the case of higher energy excitation. A more quantitative analysis of geminate recombination is presented in figure 4.9 (c) and (d). The integrated area of the p-MePhS B % ← X % absorption band (integrating 106 over its half peak width 500 ≤ λ probe ≤ 550 nm to account for vibrational cooling) is plotted against the delay time. It is clear that, consistent with the initial observation, the survival probability of the p-MePhS radical is ~60% at 200 ps upon 267 nm photolysis, whereas a 70% survival probability can be obtained for 200 nm photolysis. The radical decay kinetics is in good accord with a full time-dependent diffusion recombination model (see Ref. 62 and references there in for equations and details), where r xn and 〈r 0 〉 are the only adjustable parameters. The former is the reaction radius – the distance between geminate partners where recombination is assumed to occur instantaneously. The latter is the average separation distance of the photofragments. Fitting the experimental data in figure 4.9 (solid lines) returns average ejection lengths of 9.4 and 12.2 Å at 267 and 200 nm photolysis, respectively, and a reaction radius r xn = 4.2 Å. (The diffusion coefficients required in this model, D H and D p-MePhS in cyclohexane, are estimated to be 7.67 × 10 -4 Å 2 fs -1 , using D H in water 65 and scaling for the liquid viscosities via the Stokes-Einstein equation, and D p-MePhS ~ D benzene = 2.26 × 10 -4 Å 2 fs -1 , which is estimated from D benzene in ethanol 66 scaled by the appropriate liquid viscosity difference). 4.4 Discussion The principal observations of this work are as follows: TA attributable to p-MePhS( X % ) and excited parent molecules both appear promptly (i.e. within the 52 fs experimental time resolution achieved by the 271 nm pulses) following photolysis of p-MePhSH in ethanol at 200, 267, 271, 285 and 295 nm, and in cyclohexane at 200 and 271 nm. There 107 is also some evidence for the generation of the excited state radical p-MePhS(Ã) – this will be discussed in Appendix M. Photolysis carried out with 200 nm, however, does not show spectral evidence supporting the formation of p-MePhS(Ã) radical. The anisotropy measurements show that R(t) at any given λ pump and time t is relatively flat across the range 310 ≤ λ probe ≤ 650 nm, increases as λ pump is tuned from 295 nm (R(t = 200) ~ –0.05) to 267 nm (R(t = 200) ~ +0.17), and tends to zero at t > 50 ps due to rotational dephasing of the radical fragment in bulk solution. The TA attributable to p-MePhS( X % ) radical products exhibits an early (ps timescale) decay attributable to radical-radical geminate recombination, but some 60% of the photo-produced radicals are still evident in the TA spectra recorded at t = 900 ps. A TA attributed to adduct formation grows in (at λ probe ~385 nm) on a timescale commensurate with that associated with p-MePhS( X % ) radical loss by geminate recombination. 4.4.1 Gas Phase Behavior A consistent rationale for these observations can be developed by first reprising the fragmentation dynamics of gas phase p-MePhSH as established by H (Rydberg) atom photofragment translational spectroscopy and complementary electronic structure calculations, 47,62 and then progressively considering the influence of the increasingly interacting solvents used in the present study on both the short and the longer time dynamics. The gas phase studies identify a dominant role for the 1 1 ππ*←S 0 excitation at the longest excitation wavelengths. The photo-excited 1 1 ππ* molecules dissociate by tunneling through the small barrier under the conical intersection (CI) between the 1 1 ππ* 108 and 1 1 πσ* PESs in the R S–H stretch coordinate – as illustrated in Figure 4.1 (a). This barrier provides little impediment to dissociation: the resulting H atoms display anisotropic recoil velocity distributions and the partner p-MePhS radicals are formed in both their X % and Ã states. The electronic branching in the radical can be traced to the parent geometry as the dissociating molecule passes through the region of the 1 1 πσ*/S 0 CI at longer R S–H (Figure 4.1 (a)); molecules with the S–H bond in the ring plane follow the diabatic path to ground state radical products. The vibrational energy disposal in the radical evolves upon decreasing λ pump, as does the pattern of H atom recoil velocities. This evolution can be rationalized by assuming that direct excitation to and dissociation on the 1 1 πσ* PES makes an increasing contribution to the product yield. As Figure 4.2 shows, the UV absorption cross-section increases greatly at yet shorter λ pump : the dominant excitation at these wavelengths is to the 2 1 ππ* state, and again results in prompt dissociation – presumably via efficient radiationless coupling to the 1 1 πσ* potential. 4.4.2 Anisotropy The TA features attributed to p-MePhS radicals in both cyclohexane and ethanol appear within the instrument response time, consistent with fast S−H bond fission of the parent p-MePhSH molecule. The solution phase absorption spectra of p-MePhSH show subtle differences from that measured in the gas phase, but do not suggest that solvation has any dramatic effect on the pattern of excited electronic states. The anisotropy measurements show R(t = 200 fs) evolving from slightly negative (~ –0.05) to positive (~0.15) as λ pump 109 is reduced from 295 nm to 267 nm. The R(t = 200 fs) value measured at any given λ pump value is relatively constant across the range of probe wavelengths investigated, though is typically more positive at the λ pump values assigned to ground state radicals. In the limit that the TDMs of the pumped and probed species are rigorously parallel (perpendicular) to each other, R(t = 0) takes the limiting values of +0.4( −0.2). Thus at no λ pump wavelength is limiting anisotropy observed. Electronic structure calculations 47,62 show that the p-MePhS( X % ) radical absorption in the range 425-550 nm encompasses the (strong) B % ← X % and (weaker) C % ← X % transitions. At any given λ pump , we see no change in R(t = 200 fs) across the TA envelope assigned to this radical, which likely indicates substantial vibronic mixing between the B % and C % states and that the effective TDM is dominated by the stronger B % ← X % transition, i.e. lies parallel to the C–S bond. The radical E ~ ← X % transition (centered at λ probe ~320 nm) is predicted to be similarly z- polarized. The parent 1 1 ππ*−S 0 and 1 1 πσ*−S 0 TDMs are both calculated to lie perpendicular to the C–S bond, while that for the 2 1 ππ*−S 0 transition is aligned along the C–S bond. Therefore, the anisotropy of the TA signals attributed to the p-MePhS( X % ) radicals qualitatively accord with expectation, becoming progressively more positive at shorter λ pump as the partial cross-section for 2 1 ππ*−S 0 absorption increases. The R(t = 200 fs) values for TA attributed to ESA are very similar to that associated with the p- MePhS( X % ) radicals, and show very similar trends with changing λ pump . This implies that the TDM for the ESA is also aligned along the C–S bond, in accord with the results of electronic structure calculations which place the strong 2 1 ππ*←S 1 (1 1 ππ*) parent 110 transition in just the right energy region. The lifetime of the excited state is sufficiently short to cast doubt on reorientation of the S 1 molecule (and the p-MePhS( X % ) radical) as an explanation for the observed non-limiting anisotropy values; increased overlap of the different p-MePhSH absorptions in ethanol (relative to the gas phase) seems a more plausible explanation for this observation. It is noteworthy that calculations showed a small barrier for the rotation of the S–H bond around the C–S axis. Thus, instead of a well defined direction, the TDM will exhibit a distribution slightly off-axis compared to those shown in Figure 4.1 (b) – (d), which is calculated from the lowest energy configuration of the parent molecule p-MePhSH. This observation is also likely to contribute to the non-limiting anisotropy. 4.4.3 Geminate Recombination After S–H bond fission, H atoms depart with substantial kinetic energy. Of course, under a collision free environment, this information is preserved and enables detection of the product total kinetic energy release (TKER). In the condensed phase, the high recoil velocity and small size of the H atom means that most can be expected to require several collisions with solvent molecules prior to translational thermalization within the liquid. This stopping of the ballistically ejected H atoms is expected to be very fast and is not time-resolved in the present experiments, but a measure of the stopping distance (usually called the H atom ejection length) can be determined in the longer timescale diffusive geminate recombination of the radical pair. 111 In the spherically symmetric limit and under field-free diffusion of the geminate pair, 67,68 the long time survival probability, Ω(∞), of the radical pair is a simple function of ejection length (r 0 ) and an encounter radius to which the two partners must return in order to recombine, namely Ω(∞) = (1 - r xn /r 0 ). For a given radical partner, the encounter radius r xn should not vary with photolysis wavelength; the latter simply determines the ejection length via how much translational energy the H atom receives. In principle, with different photoexcitation energies we should be able to infer the stopping power of cyclohexane to H atoms of several different kinetic energies. 69 We note that, unlike photodissociation reactions producing slower, heavier fragments, we see no evidence for fast cage recombination 14,70 in the time dynamics of the X % state p-MePhS radical. The radical survival probabilities after S−H bond cleavage of p-MePhSH at 267 and 200 nm and the ejection lengths obtained from a full time-dependent geminate recombination model 62,67 are in qualitative agreement with the increase in the average TKER determined in the gas phase experiments. The increase in average TKER from ~8,500 to ~14,000 cm -1 determined in the gas phase studies (Table 4.2) correlates with an increase in average ejection length from 9.4 to 12.2 Å (Figure 4.9), if we assume that the energy partitioning established in the gas phase maps directly into the cyclohexane solution study. Moreover, the kinetics of the p-MePhS( X % ) radical and the adduct as deduced by single wavelength analysis at 380 nm show a decay and rise commensurate with one another. The kinetics of this feature suggest assignment to a recombination adduct. There are two geminate recombination pathways. One is trivial: H + p-MePhS( X % ) 112 recombination on the adiabatic ground state PES, reforming S 0 parent molecules. Observing a diminution of the parent bleach signal with increasing time would provide evidence for this process; unfortunately the parent bleach signal would be too deep in the UV for observation in the present experiments. Table 4.2. Comparison of maximum possible (TKER max ) and average measured TKERs of products arising from excited state s−H bond fission in p-MePhSH at selected photolysis wavelengths ‡ p-MePhSH # λ phot / nm 193 216 266 Excitation Energy / cm -1 51 813 46 296 37 594 TKER max ( X % ) / cm -1 24 380 18 870 10 160 TKER max ( A % ) / cm -1 21 060 15 550 6 840 Average TKER( X % ) / cm -1 ~16 500 a ~16 000 b ~10 000 a Average TKER( A % ) / cm -1 ~13 200 a ~12 700 b ~6 700 a # D 0 (p-MePhS−H) = 27430 ± 50 cm -1 , ΔE( A % − X % ) = 3320 ± 50 cm -1 from Refs 47,62 a Refs 47,62 b Ref. 47 The p-MePhSH structure is one of several isomers with chemical formula C 7 H 8 S, but any plausible geminate recombination product must involve an essentially barrierless formation pathway. The p-methyl substituted sulfur analog of cyclohexadienone shown in Figure 4.10 (structure 2) is thus a contender. Ab initio calculations at the CCSD(T)- F12/aug-cc-pVTZ level (carried out by Tom Oliver from U of Bristol) starting with an H atom positioned symmetrically above the benzene ring of the radical rapidly converge to this closed shell structure (with H atom added to the ortho- position), with a calculated minimum energy 2.3 eV below the H + p-MePhS asymptote (1.0 eV above the global S 0 ‡ Adapted from: Y. Zhang, T. A. A. Oliver, M. N. R. Ashfold and S. E. Bradforth, “Contrasting the excited state reaction pathways of phenol and para-methylthiophenol in the gas and liquid phases”, Faraday Discuss., DOI:10.1039/C2FD20043K 113 minimum energy of p-MePhSH). Note that the minimum energy of the para- position adduct lies slightly lower than that of the ortho- position adduct and thus is more thermodynamically stable, as shown in Table 4.3 and illustrated in Figure 4.10. However, diffusional proximity dictates that the ortho- position is more favorable compared to any other recombination positions on the ring. Statistical argument is also in favor of forming an ortho- position adduct instead of para- position (2:1). Moreover, there is some evidence supporting the enhanced reactivity at the ortho- position for para- substituted heteroaromatics, e.g., the primary photoproduct of tyrosine, a para-substituted phenol, is shown to be two tyrosine units linked together via a sigma bond formed at the ortho- position, whereas no such photoproduct was found for the unsubstituted phenol. 71 These three aspects were reinforced by the observation of the TA signal from the ortho- position adduct – the 385 nm feature coincides relatively well with the predicted S 2 -S 0 vertical transition energy of the ortho- position adduct, and the strong oscillator strength of this transition suggests that it is a plausible candidate for the adduct structure. 114 Figure 4.10 Illustration of geminate recombination pathway leading to methyl substituted cyclohexadienone. Structure 1: lowest energy structure, para- substituted; structure 2: ortho- substituted; structure 3: highest energy structure, meta- substituted. The S 0 and S 2 PESs are identical to those shown in Figure 1(a). The green, red and blue curves are for illustration purpose only, and the relative energies between structures are not to scale – refer to Table 4.3 for details. Table 4.3 Calculated energies of the various potential p-MePh(H)S adduct structures relative to the S 0 minimum energy of the p-MePhSH molecule at the CCSD(T)-F12/aug-cc-pVTZ level. Calculated EOM-CCSD/aug-cc-pVTZ vertical excitation energies / eV (nm in brackets) and associated oscillator strengths (f) for the S n ←S 0 transitions for the p- and o- isomers of the p- MePhS(H) adduct. Structure Relative Energy (to 1) in eV S 1 -S 0 eV (nm) S 2 -S 0 eV (nm) S 3 -S 0 eV (nm) 1 (p-) 0 † 2.28 (543) f = 6.3× 10 -7 4.57 (272) f = 0.614 5.19 (238) f = 0.0395 2 (o-) 0.06 2.25 (551) f = 1.01× 10 -4 3.71 (334) 0.292 5.29 (234) f = 1.47× 10 -3 3 (m-) 1.6 – – – † Adduct structure 1 lies 0.96 eV above the global S 0 minimum energy of p-MePhSH. We cannot say what fraction recombines at the S atom compared to the ring, but the relatively large r xn value we derive could indicate that the ring is a major site for 115 recombination. As can be seen in the fit, the early time kinetics of both features are not described perfectly, because absorption at these probe wavelengths is contaminated by S 1 ESA. Fortunately, the majority of the geminate recombination kinetics is not affected because of the very short S 1 state lifetime of p-MePhSH, thus allowing extraction of ejection lengths. The kinetic energy dependence of the H atom ejection length from p-MePhSH photolysis can be compared to that observed for the Cl and OH radicals produced by photolysis of aqueous HOCl. For this photoreaction, the majority of the excess energy after O-Cl bond fission is released as product translation, and very little O–H vibration is observed. 72-75 Madsen et al. have further established that the ejection length (r Cl-OH ), which ranges from 4.4 to 6.0 Å, scales linearly with initial translational energy over the range 1.6–2.8 eV (12,900–22,600 cm -1 ). 69 This result is partially supported by MD simulations. The ejection length would be expected to be a linear function of kinetic energy if the friction, in this case from water as the environment, is velocity independent. The present results suggest that cyclohexane is much less efficient at stopping an H atom than water is in arresting heavier radicals. 4.5 Conclusion The TA studies show that p-MePhS( X ~ ) radicals are formed promptly in both cyclohexane and ethanol solutions when p-MePhSH is excited with 267 < λ pump < 295 nm (above the S 1 origin). Similarly, radical appearance occurs well within the instrument 116 response time at a much shorter wavelength (λ pump = 200 nm) in both solvents. The longer pump wavelength behavior is reminiscent of the gas phase photodissociation that occurs via tunneling through a negligible barrier below the S 1 /S 2 CI, suggesting that the presence of the solvent (even ethanol) does not significantly shift this CI to higher energy. Otherwise the time scale for the tunneling process would have been dramatically increased. However, there is some tentative evidence for a long-lived S 1 state. The electronic absorption of this excited state spans the whole probe wavelength region but peaks at ~450 nm; the oscillator strength of this transition is computed to be far larger than any radical transition and thus “exaggerates” the S 1 contribution. Although this suggests that there is only a small portion of long-lived S 1 p-MePhSH, it is curious why there is any at all since the dissociation barrier is negligibly small and S–H bond fission is much faster. The inconsistency between the negligible barrier and long-lived S 1 state remains to be resolved. It is possible that a very small amount of S 1 is trapped before dissociation via as yet unidentified pathways, but the PES shown in Figure 4.1 (a) does not provide any information to support such notion. It is also possible that the band assignment of the long-lived underlying feature is incorrect – this argument is pending support from the ab initio calculation of the triplet absorption spectrum. Unfortunately, extensive steady state and time-resolved fluorescence experiments in our lab failed to provide evidence to support a long lived S 1 sub-population. In fluorescence experiments, there is significant evidence suggesting that a photoproduct formed even at lamp excitation intensities is a 117 much better fluorophore than the parent p-MePhSH molecule. This hinders our ability to examine the fluorescence lifetime of the parent MePhSH, and we have been unable so far to identify the photoproduct or determine how it is generated. It does however suggest that we can discount the lifetimes reported by Brede et al. 51 The present TA studies also show tantalizing evidence for the generation of p-MePhS( A ~ ) radical products in the solution phase UV irradiation of p-MePhSH. This is reminiscent of the gas phase photochemistry where both X ~ and A ~ radicals were observed. It is difficult to determine the branching ratio in the gas phase HRA-PTS experiments, but in principle, this information can be extracted in our TA experiments if the extinction coefficients of both radicals were known. The reader is referred to Appendix M for detailed discussion on the evidence for the A ~ state radical and our attempt to determine the branching ratio. The radical survival probability can be described by the full time- dependent geminate recombination model. Interestingly, a larger reaction radius provides a better fit to the radical kinetics, strongly suggesting a different recombination site than the sulfur atom (to give the parent p-MePhSH molecule). Indeed, we have observed an additional feature centered at λ probe = 385 nm. The rise of feature, which is first evident at t > 5 ps and becomes most pronounced at t = 60 ps, is commensurate with the decay of the radical absorption. Therefore, it is attributed to geminate recombination of the H atom with the p-MePhS, not to reform the ground state molecule but instead an adduct p- MePh(H)S – with the H atom attacking the benzene ring at the ortho- position. The average ejection length, which is another important parameter used in the geminate 118 recombination model, is in qualitative agreement with the expected total kinetic energy release as determined in the gas phase – the ejection length is longer in 200 nm excitation as compared to 267 nm, since there is more energy available to product translation. No absorption attributable to the solvated electron is observed at any pump wavelengths, implying that autoionization into the solvent continuum is not competitive with S–H bond fission even in a strongly interacting ethanol solvent. The vertical ionization potential of p-MePhSH in ethanol is not currently known but, given the strong binding energy of the solvated electron, we anticipate that 200 nm excitation would raise the molecule to energies well above the p-MePhS + e - (solv) asymptote. Given that autoionization occurs on a timescale often < 10 fs, it is interesting to consider why autoionization is not competitive. Direct dissociation at 200 nm could be taking place on the ~5 fs timescale and poor overlap of the excited molecular wavefunction and the solvent continuum provides another one possible explanation. It thus appears that both cyclohexane and ethanol provides few changes to the earliest epoch of the reaction dynamics; these are determined by the same forces as in the gas phase. We do not see new pathways associated with the presence of solvent such as electron transfer or PCET, though we do discern geminate recombination, adduct formation, and VER of the “hot” nascent p-MePhS( X ~ ) radicals: processes that are unique to the condensed phase. The present results thus adds to the growing consensus that the patterns of energy disposal identified for exoergic photodissociation reactions in 119 the gas phase do provide a useful starting point for understanding the earliest events when the same photodissociation pathway occurs in the liquid phase. 9 For reactions releasing small particles (of which the H atom represents one extreme example), the solvent cannot respond sufficiently rapidly to significantly affect the earliest time reaction dynamics, even though it may well influence the overall dynamics by, for example, shifting the location of (or transmission through) relevant CIs. 70 120 4.6 Chapter 4 References 1. This chapter is based upon and expanded from the previous published work: T. A. A. Oliver, Y. Zhang, M. N. R. Ashfold and S. E. Bradforth, Faraday Discuss. 2011, 150, 439-458, and two manuscripts: 1) Y. Zhang, T. A. A. Oliver, M. N. R. Ashfold and S. E. Bradforth, “Exploring the Excited State Deactivation Pathways of para-methylthiophenol in Ethanol” (in preparation); 2) Y. Zhang, T. A. A. Oliver, M. N. R. Ashfold and S. E. Bradforth, “Contrasting the excited state reaction pathways of phenol and para- methylthiophenol in the gas and liquid phases”, Faraday Discuss., DOI:10.1039/C2FD20043K 2. U. Banin and S. Ruhman, J. Chem. Phys. 1993, 98, 4391-4403. 3. D. A. V. Kliner, J. C. Alfano and P. F. Barbara, J. Chem. Phys. 1993, 98, 5375-5389. 4. N. Pugliano, S. Gnanakaran and R. M. Hochstrasser, Journal of Photochemistry and Photobiology a-Chemistry 1996, 102, 21-28. 5. H. Bursing, J. Lindner, S. Hess and P. Vohringer, Applied Physics B-Lasers and Optics 2000, 71, 411-417. 6. H. Bursing and P. Vohringer, Phys. Chem. Chem. Phys. 2000, 2, 73-82. 7. P. J. Reid, Acc. Chem. Res. 2001, 34, 691-698. 8. A. C. Moskun and S. E. Bradforth, J. Chem. Phys. 2003, 119, 4500-4515. 9. A. C. Moskun, A. E. Jailaubekov, S. E. Bradforth, G. H. Tao and R. M. Stratt, Science 2006, 311, 1907-1911. 10. P. Z. El-Khoury and A. N. Tarnovsky, Chem. Phys. Lett. 2008, 453, 160-166. 11. M. T. Zanni, B. J. Greenblatt, A. V. Davis and D. M. Neumark, J. Chem. Phys. 1999, 111, 2991-3003. 12. I. Nadler, D. Mahgerefteh, H. Reisler and C. Wittig, J. Chem. Phys. 1985, 82, 3885- 3893. 13. J. W. Qian, D. J. Tannor, Y. Amatatsu and K. Morokuma, J. Chem. Phys. 1994, 101, 9597-9609. 14. N. Winter and I. Benjamin, J. Chem. Phys. 2004, 121, 2253-2263. 15. J. Larsen, D. Madsen, J. A. Poulsen, T. D. Poulsen, S. R. Keiding and J. Thogersen, J. Chem. Phys. 2002, 116, 7997-8005. 121 16. C. G. Elles, M. J. Cox and F. F. Crim, J. Chem. Phys. 2004, 120, 6973-6979. 17. C. G. Elles and F. F. Crim, Annu. Rev. Phys. Chem. 2006, 57, 273-302. 18. C. Petersen, N. H. Dahl, S. K. Jensen, J. A. Poulsen, J. Thogersen and S. R. Keiding, J. Phys. Chem. A 2008, 112, 3339-3344. 19. C. G. Elles, M. J. Cox, G. L. Barnes and F. F. Crim, J. Phys. Chem. A 2004, 108, 10973-10979. 20. M. J. Krisch, L. R. McCunn, K. Takematsu, L. J. Butler, F. R. Blase and J. Shu, J. Phys. Chem. A 2004, 108, 1650-1656. 21. A. Sinicropi, W. M. Nau and M. Olivucci, Photochemical & Photobiological Sciences 2002, 1, 537-546. 22. A. L. Sobolewski and W. Domcke, Chem. Phys. 2000, 259, 181-191. 23. A. L. Sobolewski, W. Domcke, C. Dedonder-Lardeux and C. Jouvet, Phys. Chem. Chem. Phys. 2002, 4, 1093-1100. 24. M. N. R. Ashfold, B. Cronin, A. L. Devine, R. N. Dixon and M. G. D. Nix, Science 2006, 312, 1637-1640. 25. M. N. R. Ashfold, G. A. King, D. Murdock, M. G. D. Nix, T. A. A. Oliver and A. G. Sage, Phys. Chem. Chem. Phys. 2010, 12, 1218-1238. 26. M. Barbatti, M. Vazdar, A. J. A. Aquino, M. Eckert-Maksic and H. Lischka, J. Chem. Phys. 2006, 125, 164323. 27. M. N. R. Ashfold, A. L. Devine, R. N. Dixon, G. A. King, M. G. D. Nix and T. A. A. Oliver, Proc Nat Acad Sci USA 2008, 105, 12701-12706. 28. C. E. Crespo-Hernandez, B. Cohen, P. M. Hare and B. Kohler, Chem. Rev. 2004, 104, 1977-2019, and references therein. 29. B. Kohler, J. Phys. Chem. Lett. 2010, 1, 2047-2053, and reference therein. 30. J. M. L. Pecourt, J. Peon and B. Kohler, J. Am. Chem. Soc. 2001, 123, 10370-10378. 31. L. Serrano-Andres and M. Merchan, Journal of Photochemistry and Photobiology C-Photochemistry Reviews 2009, 10, 21-32. 32. P. M. Hare, C. E. Crespo-Hernandez and B. Kohler, Proc Nat Acad Sci USA 2007, 104, 435-440. 122 33. H. R. Hudock, B. G. Levine, A. L. Thompson, H. Satzger, D. Townsend, N. Gador, S. Ullrich, A. Stolow and T. J. Martinez, J. Phys. Chem. A 2007, 111, 8500-8508. 34. E. Samoylova, H. Lippert, S. Ullrich, I. V. Hertel, W. Radloff and T. Schultz, J. Am. Chem. Soc. 2005, 127, 1782-1786. 35. A. E. Jailaubekov. PhD Thesis. University of Southern California, 2007. 36. M. G. D. Nix, A. L. Devine, B. Cronin and M. N. R. Ashfold, J. Chem. Phys. 2007, 126. 37. I. Conti, M. Garavelli and G. Orlandi, J. Am. Chem. Soc. 2009, 131, 16108-16118. 38. L. I. Grossweiner, G. W. Swenson and E. F. Zwicker, Science 1963, 141, 805. 39. J. Peon, G. C. Hess, J. M. L. Pecourt, T. Yuzawa and B. Kohler, J. Phys. Chem. A 1999, 103, 2460-2466. 40. D. Sharma, J. Leonard and S. Haacke, Chem. Phys. Lett. 2010, 489, 99-102. 41. X. Y. Chen, D. S. Larsen, S. E. Bradforth and I. H. M. van Stokkum, J. Phys. Chem. A 2011, 115, 3807-3819. 42. R. I. Cukier and D. G. Nocera, Annu. Rev. Phys. Chem. 1998, 49, 337-369. 43. S. Hammes-Schiffer, Acc. Chem. Res. 2001, 34, 273-281. 44. A. L. Sobolewski and W. Domcke, J. Phys. Chem. A 2001, 105, 9275-9283. 45. S. Ishiuchi, M. Saeki, M. Sakai and M. Fujii, Chem. Phys. Lett. 2000, 322, 27-32. 46. S. Ishiuchi, M. Sakai, K. Daigoku, K. Hashimoto and M. Fujii, J. Chem. Phys. 2007, 127, 23404. 47. T. A. A. Oliver, G. A. King, D. P. Tew, R. N. Dixon and M. N. R. Ashfold, in preparation. 48. Y. Zhang, T. A. A. Oliver, M. N. R. Ashfold and S. E. Bradforth, Faraday Discuss. DOI:10.1039/C2FD20043K. 49. I. S. Lim, J. S. Lim, Y. S. Lee and S. K. Kim, J. Chem. Phys. 2007, 126, 034306- 034310. 50. A. L. Devine, M. G. D. Nix, R. N. Dixon and M. N. R. Ashfold, J. Phys. Chem. A 2008, 112, 9563-9574. 123 51. Y. M. Riyad, S. Naumov, R. Hermann and O. Brede, Phys. Chem. Chem. Phys. 2006, 8, 1697-1706. 52. A. E. Jailaubekov and S. E. Bradforth, Appl. Phys. Lett. 2005, 87. 53. C. A. Rivera, S. E. Bradforth and G. Tempea, Opt. Express 2010, 18, 18615-18624. 54. C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys. 2009, 130. 55. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum. 2003, 74, 4958-4960. 56. R. Hermann, G. R. Dey, S. Naumov and O. Brede, Phys. Chem. Chem. Phys. 2000, 2, 1213-1220. 57. K. N. Jha, G. L. Bolton and G. R. Freeman, J. Phys. Chem. 1972, 76, 3876-&. 58. J.-P. Jay-Gerin, Can. J. Chem. 1997, 75, 1310 - 1314. 59. C. T. Middleton, B. Cohen and B. Kohler, J. Phys. Chem. A 2007, 111, 10460- 10467. 60. A. M. Lossack, E. Roduner and D. M. Bartels, J. Phys. Chem. A 1998, 102, 7462- 7469. 61. J. D. Faulk, R. C. Dunbar and C. Lifshitz, J. Am. Chem. Soc. 1990, 112, 7893-7899. 62. T. A. A. Oliver, Y. Zhang, M. N. R. Ashfold and S. E. Bradforth, Faraday Discuss. 2011, 150, 439-458. 63. D. L. Gerrard and W. F. Maddams, Spectrochim. Acta, Part A 1978, 34, 1205-1211. 64. D. L. Gerrard and W. F. Maddams, Spectrochim. Acta, Part A 1978, 34, 1219-1223. 65. T. Ichino. PhD thesis, University of Notre Dame, 2001. 66. CRC Handbook of Chemistry and Physics. 91st ed.; Taylor and Francis: 2009. 67. J. A. Kloepfer, V. H. Vilchiz, V. A. Lenchenkov, A. C. Germaine and S. E. Bradforth, J. Chem. Phys. 2000, 113, 6288-6307. 68. E. B. Krissinel and N. Agmon, J. Comp. Chem. 1996, 17, 1085- 1098. 69. A. Madsen, C. L. Thomsen, J. A. Poulsen, S. J. K. Jensen, J. Thogersen, S. R. Keiding and E. B. Krissinel, J. Phys. Chem. A 2003, 107, 3606-3611. 124 70. C. A. Rivera, N. Winter, R. V. Harper, I. Benjamin and S. E. Bradforth, Phys. Chem. Chem. Phys. 2011, 13, 8269-8283. 71. S. S. Lehrer and G. D. Fasman, Biochemistry 1967, 6, 757-&. 72. A. J. Bell, P. R. Pardon, C. G. Hickman and J. G. Frey, J. Chem. Soc. Faraday T. 1990, 86, 3831-3836. 73. C. G. Hickman, A. Brickell and J. G. Frey, Chem. Phys. Lett. 1991, 185, 101-104. 74. C. G. Hickman, N. Shaw, M. J. Crawford, A. J. Bell and J. G. Frey, J. Chem. Soc. Faraday T. 1993, 89, 1623-1630. 75. C. L. Thomsen, D. Madsen, J. A. Poulsen, J. Thogersen, S. J. K. Jensen and S. R. Keiding, J. Chem. Phys. 2001, 115, 9361-9369. 125 Chapter 5. Phenol Oxidation in the Gas and Condensed Phase Following UV Photoexcitation 1 5.1. Introduction The competition between phenol dissociation and ionization is of fundamental importance in radiation and biological chemistry, as the oxidation of phenolic compounds to their radical forms is involved in many crucial biological and industrial processes. For example, at combustion temperature, phenol and its oxidative product phenoxyl radical are identified to be the most important products in the early stage oxidation of benzene, a common aromatic additive for unleaded fuel. 2,3 As for a biological example, the anti- oxidative property of α-tocopherol, a component of vitamin E, is essentially originated from the ability of the phenol functional group to rapidly scavenge peroxy radicals (ROO·). By transferring an H-atom or an electron, phenol reacts with the highly oxidative radicals before they attack other biologically important molecules. Moreover, phenol is the long-wavelength chromophore and fluorophore of tyrosine, one of the two essential amino acids with a heteroaromatic moiety (the other one is tryptophan). In photosystem II (PSII) where water is oxidized to oxygen upon absorption of sun light by plants, the oxidation of tyrosine (Tyr Z ) by the chlorophyll cation (P 680 + ) is believed to be the key step in the water oxidation sequence (Kok Cycle 4 ) – the long range transfer of electrons from Tyr Z to P 680 + is concerted with a proton transfer from Tyr Z to an adjacent histidine residue. 5-7 The resulting tyrosyl radical is reduced by the manganese ions in the oxygen evolving complex (OEC) at the last step of the Kok cycle, in which oxygen is evolving from oxidation of water. 7-10 126 It is difficult, however, to elucidate the mechanisms underlying the redox reaction of phenol due to the complexity in the biological environment. Even for the phenol aqueous solution, which is the simplest model system, the strong coupling between the solute and solvent due to H-bonding interaction can change the ordering and/or the shape of the excited state potential energy surfaces (PES) along the reaction coordinate, thereby alter the reaction dynamics upon photoexcitation. Note that the ability of phenol to form strong H-bonding with protic solvents is fundamentally different from the p-methylthiophenol (p-MePhSH) discussed in the previous chapter. The latter does not form strong H-bond with protic solvents due to the weak electronegativity of the sulfur atom. As a result, the gas phase reaction dynamics of p-MePhSH does seem to map to the condensed phase, even in ethanol. 11-13 This chapter aims to unravel the competing dissociation and ionization pathways of phenol in various conditions. First, we will briefly reprise its photodissociation mechanisms in the collision-free environment. This will cover the necessary bases to examine whether the gas phase behavior maps to a weakly interacting solvent. Lastly, we will answer the question if ionization pathways can be turned on in protic solvents, and if so, what is the exact mechanism leading to the production of the ionized products. The photodissociation of a phenol molecule with excitation wavelength (λ phot ) ≥ 193 nm, leading to the generation of a phenoxyl radical and an H-atom, is determined to be an important non-radiative deactivation pathway in the gas phase. 14 In this deep-UV wavelength regime, photoionization does not play an important role as the vertical 127 ionization potential (VIP) of the gas phase phenol molecule is 8.5 eV (~145 nm). 15 The low-lying 1 1 πσ* state is responsible for the H-atom loss upon UV excitation, as predicted by ab initio calculation 16 and suggested by gas phase photodissociation experiments, such as H(Rydberg) Atom Photofragment Translational Spectroscopy (HRA-PTS). 17,18 The diabatic 1 πσ*(S 2 ) curve is repulsive with respect to the O–H bond extension, but it intersects with the diabatically bound 1 ππ*(S 1 ) curve and forms the 1 1 πσ*(S 2 )/ 1 1 ππ*(S 1 ) conical intersection (CI), as shown in Figure 5.1. As a result, in the case of phenol, the optically “dark” 1 πσ*(S 2 ) state can be populated indirectly by non-radiative transfer from the 1 1 ππ*(S 1 ) state via vibronic coupling at the CI. More interestingly, O–H fission occurs even when the molecule is excited to the S 1 region below the 1 1 πσ*(S 2 )/ 1 1 ππ*(S 1 ) CI (correspond to λ phot > 248 nm). This observation was first attributed to dissociation from the highly vibrationally excited ground state, 14,17,18 but later explained by tunneling under the 1 1 πσ*(S 2 )/ 1 1 ππ*(S 1 ) CI from a more careful symmetry consideration. 19,20 The slow tunneling time scale is signified by the isotropic distribution of the recoiling H atom, with a kinetic energy of ~6500 cm -1 observed in the HRA-PTS experiments. 17 Its geminate partner, the ground state phenoxyl radical, is found in a very limited subset of vibrational states, among which the v 16a mode is believed to be the orthogonal mode to the O–H extension, and the coupling mode facilitating the non-radiative S 2 ← S 1 transfer. 20 128 Figure 5.1. “Un-relaxed” potential energy surfaces (PESs) along the O-H bond length coordinate for the ground state (S 0 ) and first four excited states (S 1 – S 4 ), calculated at the CASPT2(10/10)/aug(O)-AVTZ level. The transition dipole moments (TDMs) for the first two transitions, calculated by EOM-CCSD, are also shown in the figure. Adapted from Ref 20. With λ phot < 248 nm, the transition to 1 1 πσ*(S 2 ) gains oscillator strength owing to intensity borrowing from the S 3 ← S 0 transition. Thus, the dissociative state can be directly populated and subsequently results in an H atom loss on a time scale faster than the molecular rotation, which is signified by the observation of recoil anisotropy of the H atom. 17,18 This is distinct from the long wavelength excitation where H atom tunneling plays an important role in dissociation. The observed positive recoil anisotropy has ruled out any contribution from the S 1 state, which is also evident by examining the UV- Visible spectrum. However, the exact initially populated state is undetermined at 200 nm 129 excitation, due to the high density of states in this region. The threshold for excitation to the dissociative 2 1 πσ*(S 4 ) is 193 nm, resulting in the generation of the B % state radical. 21 More than half of the excess energy provided by the photolysis photon is partitioned to the total kinetic energy (~12000 cm -1 on average) at λ phot = 193 nm, the remaining excess energy channels to v 16a ring torsion and extension mode – a direct consequence of intensity borrowing from the S 3 state. 20 Weakly interacting solvents, like rare gas matrices, are generally considered excellent mimics of a gas phase environment in terms of their effect on the shapes of molecular potential energy surfaces (PESs). 22 Thus exploring the photochemistry of solutes in inert solvents such as cyclohexane provides an instructive approach to see how dissociative reaction dynamics map into the condensed phase. However, solvents of this type still provide some dynamically complexities when comparing to the collision-free environment. For example, collision between the nascent reaction products and the solvent molecules can alter the vibrational energy disposal, as solvent molecules provide an effective sink for vibrational energy relaxation (VER). 22-25 Similarly, the translational motion of products will be eventually stopped due to the friction exerted by the solvent. 26 Subsequent diffusion of the incompletely separated reaction products in the solvent can lead to geminate recombination. 23,27 Although interaction with solvent molecules have modest changes on the PES landscape, differential solvation of electronic states can lead to larger changes in the location of the conical intersections (CIs), and affect how the 130 excited state molecules approach (i.e. the velocity and the angle) and thus branch through CIs. Figure 5.2. Energetic of phenol and its photoproducts: (a) isolated molecule, (b) aqueous solution. The energy (in eV) of each state is labeled on top of the horizontal line. a energy required to excite gas phase phenol to S 1 origin (275.113nm) 17,18,20 b vertical ionization potential for gas phase phenol 15 c asymptotic value for producing ground state PhO· and H atom in the gas phase 28 d the origin for ground state phenol in water is 0.6 eV lower than that in the gas phase, due to the solvation enthalpy (ΔH solv ) 29 e the estimated energy required to excite aqueous phenol to S 1 origin (from the shift in UV-Visible spectra between cyclohexane and water solutions) f vertical ionization potential for aqueous phenol solution 30 g asymptotic value for producing solvated PhO· radical and H atom, estimated from the solvation enthalpy of PhO· (ΔH solv = -0.4 eV) 29 and solvation free energy of H atom (ΔG solv = -0.1 eV) 31 . h the onset of the photoelectron spectrum for phenol aqueous solution. 30 Note that the free energy release upon proton transfer from phenol cation to surrounding water is small compared to the uncertainty of the position of the onset (ΔG = -0.15 eV) 32 , and thus not included in the estimate. i asymptotic value for producing fully solvated PhO· radical, proton and electron, estimated from the electron attachment energy in water (ΔG = -1.54 eV) 33 j asymptotic value for producing fully solvated PhO· radical, proton and electron, estimated from the proton transfer from the H atom to water: H (aq) + H 2 O ↔ H 3 O + (aq) + e - (aq) (ΔG = +0.5 eV) 34 131 In polar solvents, the instantaneous electronic polarization and the solvation can stabilize charged species (especially the solvation of electrons) and thus lower the thermodynamic asymptote for producing such products. This energy threshold for ejecting an electron into the solvent continuum depends on how strongly the solvent interacts with the solute. In non-polar solvents such as n-hexane and cyclohexane, the solvent molecules cannot provide any favorable solvation to the charged species, thus the threshold for electron ejection is expected to be very close to the gas phase vertical ionization threshold (VIP). In protic solvents such as water and alcohols, however, photoionization can occur well below the VIP and thereby start to compete with photodissociation. What is the threshold for such ionization mechanisms to take place? Our group has done an independent measurement of the vertical ionization energy of phenol aqueous solution, 30 which also provides direct information on the energy threshold to produce fully solvated reaction products. The results obtained from this study, along with many important thermodynamic parameters of isolated phenol and phenol aqueous solution, are summarized in Figure 5.2. As discussed in Chapter 1, the mechanisms lead to electron ejection at such low energy regime include autoionization and proton-coupled electron transfer (PCET). In autoionization, a molecule is excited to a bound state but with the help of solvent motion, the system can non-adiabatically coupled to a PES which eventually leads to the fully solvated ionization product asymptote (thermodynamic threshold for producing radical (aq) + H + (aq) + e - (aq) ). Within the PCET framework, nuclear motion (solute and/or solvent) 132 after initial excitation provides a favorable geometry for transferring an electron from the solute to the solvent trap state, and thus provides an explanation for ionization taking place at energies well below the VIP. The additional complexity is the coupled proton motion. Together with the electron motion, it defines the degree of charge polarization along the reaction coordinate and ultimately the nature of the PCET reaction. In a medium where charged species are not thermodynamically favorable, the PCET reaction is likely to proceed with minimum charge polarization, which results in homolytic bond fission and thereby produces an H-atom instead of a proton and an electron. This special case of concerted PCET, called H-atom transfer (HAT), is identical to the gas phase photodissociation process. With increasing solvent polarity, on the other hand, the reaction can proceed via a charged transition state and subsequently produce charged products. It is important to point out that the current PCET definition contains a broad range of reactions, but generally they are concerted reactions which do not produce stable charged intermediate(s). 35 Reaction intermediates can be produced in step-wise reactions (not PCET), but only when the available energy is high enough to directly produce a stable radical cation via electron transfer (ET) or an anion via proton transfer (PT) in the rate- limiting step. These step-wise reactions will then proceed with a subsequent deprotonation or electron detachment, respectively. The electron transfer followed by proton transfer (ETPT) is identical to the direct ionization or autoionization discussed above. One the other hand, the proton transfer followed by electron transfer (PTET) is a 133 family of reaction unique to photoacids, in which the excited state acidities are several orders of magnitude higher than those in the ground state and thus transferring a proton to the solvent is favored. In light of the rate-limiting proton transfer step, the PTET reactions occurred on the excited state are alternatively named excited state proton transfer in some literature. Examples of strong photoacids include 1-napthol (pKa = 9.4, pKa* = -0.2), 36 o-cyanophenol (pKa = 7, pKa* = 0.7), 37 and 8-hydroxypyrene 1,3,6- trisulfonate (HPTS, pKa = 8.0, pKa* = 1.4). 38 The remaining chapter is divided as follows. First, a series of experimental methods used in our comprehensive study will be introduced. Next, the representative results obtained from these experiments will be presented, followed by a discussion, where we attempt to decipher the exact pathway(s) leading to the oxidation of phenol in various solvent environments, by comparing and contrasting the condensed environment reaction dynamics to that in the gas phase. 5.2. Experimental 5.2.1 Absorption Spectroscopy 5.2.1.1 Transient Absorption The TA experiments were achieved by exciting the phenol solution with deep UV pulses (λ pump = 271, 267 and 200 nm) and probing the transient species with a broadband super continuum pulse (typically 310 ≤ λ probe ≤ 650 nm). The generation of the pump and probe pulses is described in Chapter 4. All experiments employed pump fluence from 32 – 96 134 J/m 2 . The transient signal of phenol was linearly dependent on the pump fluence, as determined by a power dependence study in which the 267 nm pump fluence was varied across the range 4 – 136 J/m 2 . Alternate pump pulses were blocked by a chopper operating at 500 Hz. The temporal chirp of the continuum was corrected mathematically by setting the two-photon absorption peak at each probe wavelength to time zero. 39 The sample was delivered by a wire guided gravity jet, forming a thin film of thickness 50 – 80 μm at the detection region. The flow rate of the liquid ensures that a fresh sample was interrogated every pump pulse. 40 For the TA experiments, various concentrations (10, 18, 45 and 90 mM) of phenol (99.5%, Avocado) were used in water (purified by Millipore Milli-Q system), ethanol (200 Proof, Koptec) and cyclohexane (>99.0%, HPLC and UV- Spectrophotometric grade, EMD or Mallinckrodt) without further purification. For quenching experiments in aqueous phenol solution, 0.5 M HCl (36.5 – 38.0%, EMD) was used as an electron scavenger (diffusive electron quencher), and 0.5 M CsCl (99%, Alfa Aesar) was used to induce the triplet formation. 5.2.1.2 UV-Visible Absorption The absorption spectra in the UV-Visible range (190 < λ < 800 nm) of phenol in water, ethanol and cyclohexane solutions were measured using a Cary 50 Conc UV-Visible spectrometer (Varian) in a 1 cm quartz cell (Starna Cells), with the instrument resolution set to 1 nm. 0.1 – 0.2 mM of phenol solutions were used in order to produce a low optical density (< O.D. 0.5). 135 5.2.2 Emission Spectroscopy 5.2.2.1 Time-Correlated Single Photon Counting (TCSPC) The time-resolved emission was obtained by TCSPC in a 250 kHz regenerative amplified Ti:Sapphire laser system (Spectra-Physics RegA). A portion of fundamental 800 nm was used to generate 532 nm from a two-stage optical parametric amplifier with type-I phase matching process (OPA 9400, Coherent). This OPA output was frequency-doubled to 267 nm, which was subsequently rotated to s-polarization (with purity better than 120:1), attenuated to < 1 nJ and focused to the sample by an f = 25 cm lens. The fluorescence of the phenol solution was collected at 90˚ to the excitation pulse. A calcite polarizer was set to 54. 7˚ (magic angle) with respect to the 267 nm polarization to ensure isotropic detection along the fluorescence traveling path. The monochromator was set to detect 300 nm photons (peak of the fluorescence as determined by steady state emission). 0.1 mM phenol solutions in water, ethanol and cyclohexane were used to produce O.D. = ~0.1 in the 1 cm quartz cell with a magnetic stir bar to increase liquid circulation. Aqueous phenol solutions with 0.5 M HCl and 0.5 M CsCl (same concentration as in the TA experiments) were used to study the electron quenching and heavy atom effect. To compare the results from the above quenching experiments, an aqueous phenol solution with similar ionic strength was made from 0.5 M NaCl (> 99.0%, Mallinckrodt). 5.2.2.2 Steady State Emission The static emission spectra of various phenol solutions were measured in a Fluoromax-3 spectrofluorometer (Horiba Scientific) in the wavelength range of 290 ≤ λ fl ≤ 600 nm, 136 with the instrument resolution of 2 nm. The phenol concentration and the sample cell was identical to those used in TCSPC. 5.2.3 Preparation of Deuterated Phenol (PhOD) For 90 mM PhOD in D 2 O, 0.95 g of the unlabeled phenol (PhOH) was dissolved in 100 mL D 2 O (99.9% D, Cambridge Isotope Laboratories) and gently heated overnight to allow complete exchange. The purity of PhOD in D 2 O solution was determined from its NMR spectrum recorded on a Mercury 500 MHz spectrometer (Varian) at room temperature, by monitoring the disappearance of the broad peak at 4.6 – 7 ppm due to OH proton. 41 For 10 mM PhOD in cyclohexane, ~2 mL of D 2 O (~0.1 mol) was added to 0.95 g (~0.01 mol) of the unlabeled phenol, and the mixture was gently heated overnight, note that the D 2 O is in excess in order to drive the reaction to complete. Cyclohexane was then added to the mixture to dissolve the extra PhOD. The exact concentration of PhOD in cyclohexane was determined by the absorption intensity in the UV-Visible spectrum, from the known extinction coefficient of the unlabeled phenol solution, assuming that the electronic absorption of PhOH and PhOD is identical in cyclohexane. Considering the fact that the OD is labile and thus very susceptible to any water molecules, the major impurity of PhOD in cyclohexane should be PhOH, and the concentration of the latter was determined by the absorption intensity of the O-H stretch in the FT-IR spectrum. 137 5.2.4 Theoretical Methods The calculated excited state absorption spectrum arising from phenol S 1 state was carried out by our collaborator Tom Oliver from the University of Bristol. The excited state energy, relative to S 1, and its transition dipole moment (TDM), were obtained at the EOM-CCSD/aug-cc-pVTZ level. 5.3 Results 5.3.1 UV-Visible and Fluorescence Spectra The UV-visible and fluorescence spectra of phenol in water, ethanol and cyclohexane were presented in Figure 5.3. The vibrational structure of phenol upon excitation to the 1 1 ππ*(S 1 ) manifold is clearly visible in the cyclohexane solution (250 < λ < 290 nm). The negligible absorbance from 250 to 230 nm corresponds to the optically dark 1 πσ*(S 2 ) state, whereas the high extinction coefficient at λ < 230 nm is originated from 2 1 ππ*(S 3 ) and other higher excited states. The spectrum for aqueous phenol solution has a much less pronounced vibrational progression in the 250 < λ < 290 nm region, which is due to H-bonding with the solvent. The spectrum for the ethanol solution is similar to water, but with ~ 3 nm red-shift for the 1 1 ππ*(S 1 ) manifold and ~ 8 nm for the 2 1 ππ*(S 3 ). The loss of the vibrational progression in the S 1 manifold due to H-bonding with protic solvent molecules is consistent with previous findings. 42-46 Moreover, the red shift observed in the ethanol solution is consistent with phenol being a proton donor to ethanol in the H- bonded complex. 43,45 138 Figure 5.3. UV-visible absorption and fluorescence spectra of phenol in cyclohexane (blue), water (green) and ethanol (red). The absorption extinction coefficients were calculated from the known path length of the quartz cell and the phenol concentration. The absolute fluorescence counts were normalized but the relative counts between the three phenol solutions are preserved (with the concentration difference accounted for). The fluorescence spectra of all three solvents, obtained with 267 nm excitation, are characterized by a featureless broad peak centered at ~300 nm. Although is absolute quantum yields of phenol in the three solvents were not measured in our current study, from Figure 5.3 we can see that the ethanol solution has the highest quantum yield, while the fluorescence yield in cyclohexane is only about a third of that in ethanol. The strong solvent dependence of the fluorescence quantum yield, with the highest quantum yield in alcohols and lowest yield in non-polar solvents, is consistent with the literature. 44,47 This is resulted from different excited state deactivation pathways being dominant over others, which will be discussed in the next section. 139 5.3.2 Time-Resolved Experiments 5.3.2.1 267 nm Excitation of Phenol Solutions 5.3.2.1.1 Phenol in Cyclohexane Cyclohexane is a non-polar, non-interacting and low polarizability solvent which provides environment closest to the gas phase. Due to the self association of phenol molecules via H-bonding, low concentration must be used to gives the best mimic to the bare phenol in the gas phase. It can be determined from IR measurements that 10 mM phenol is the maximum concentration in which self association does not occur – for details see Chapter 6 where the photochemistry of phenol dimers and “polymers” are discussed. From the UV-visible spectrum we can see that λ pump = 267 nm will excite the phenol molecule to the S 1 manifold. The TA spectra of phenol in cyclohexane solution are shown in Figure 5.4, from which it can be concluded that there is no phenoxyl radical signature in the first nanosecond, indicating that there is no appreciable photodissociation and/or photoionization. Note that at early time the cyclohexane TA signal also contributes to the total TA signal obtained from the phenol solution but it is negligible after 800 ps. Figure 5.4 (a) shows the cyclohexane signal is ~20% of that from the solution. This number represents the upper bound of cyclohexane contribution since the TA signal originates from two-photon excitation and/or ionization of pure cyclohexane. † The majority of this signal is due to the electronic absorption of cyclohexane excited † From the extinction coefficient (Figure 5.1) of phenol in cyclohexane at 267 nm, the optical density of 10 mM phenol in a 0.08 mm film is ~0.1, which means the effective peak irradiance to drive two-photon excitation and/or ionization of cyclohexane in phenol solution is ~80% of that in pure solvent. Therefore, the actual contribution of the cyclohexane signal is ~80% less than what is shown in Figure 5.4(a). For the concentrated phenol solution at 90 mM, the optical density is close to 1. Due to the strong attenuation of the pump intensity, it is not expected to two-photon excite and/or ionize the solvent. 140 states and radical cation. 48,49 Although the solvated electron in cyclohexane is thought to peak at the near-IR, 50 we cannot rule out the possibility of broad absorption band which extends to the visible range. The TA spectrum at this time range is dominated by the excited state absorption (ESA) of the 1 1 ππ*(S 1 ) state. The S 1 spectrum obtained in our experiment is in full accordance with the theoretical ESA spectrum calculated at the EOM-CCSD/aug-cc-pVTZ level by our collaborator Tom Oliver (see Figure 5.4 (b)), and it has similar shape as that obtained by Hermann et al. 51 The kinetics for the center of the three major features (380, 475 and 600 nm) are similar to one another, indicating that they originate from the same state. Moreover, all three kinetics traces are in good agreement with the 2.1 ns lifetime, as obtained in our TCSPC experiments, which also agrees with the fluorescence lifetime measured by Berlman 44 and Bent et al. 52 This is a further confirmation that the spectral signature observed in 10 mM phenol in cyclohexane is indeed originated from the S 1 ESA. 141 Figure 5.4. (a) 10 mM phenol in cyclohexane TA spectra at selected delay times, measured with 267 nm pump fluence = 32 J/m 2 (1.1 μJ pulse energy and 210 μm spot size). The polarization of the super continuum probe pulse is set to magic angle (54.7˚) with respect to that of the excitation pulse. Thick line: TA signal from the solution; thin line: TA signal from pure cyclohexane. (b) 10 mM phenol in cyclohexane TA spectra overlaid with the calculated EOM-CCSD/aug-cc-pVTZ transition dipole moments (TDM) for transitions originated from S 1 . (c) Kinetics at selected wavelengths obtained from the TA experiment (lines) and fluorescence life time obtained from TCSPC experiment (dots). 142 Figure 5.5. (a) 10 mM phenol in cyclohexane TA spectra at nanosecond delay times, measured with 267 nm pump fluence = 80 J/m 2 (2.8 μJ pulse energy and 210 μm spot size). The polarization of the super continuum probe pulse is set to magic angle (54.7˚) with respect to that of the excitation pulse. (b) Kinetics at selected wavelengths obtained from the TA experiment (lines) and fluorescence life time obtained from TCSPC experiment (dots, same data as shown in Figure 5.4(c) but difference normalization factor). (c) Back-to-back measurement of 10 mM PhOH and PhOD in cyclohexane, obtained at 14 ns delay time. With the best possible precaution to exclude water in our current pump-probe experiment, for the PhOD in cyclohexane experiment, there are ~20% of PhOH initially, which increased to ~50% after a 20 minute scan, as determined by the FT-IR spectrum monitoring the O–H stretch intensity. A surprising result is the observation of phenoxyl radical at the nanosecond scale. Figure 5.5 (a) presents the TA spectra from 1.9 – 14 ns, from which we can see the clear spectral signature for phenoxyl radical. The vibrational progression at ~ 380 and ~ 400 nm is consistent with the phenoxyl radical spectrum in argon matrix obtained by vacuum-UV 143 photolysis of phenol, 53 and that obtained by photodetachment of phenolate in aqueous solution. 54 Figure 5.5 (b) shows the kinetics of 475 and 600 nm are consistent with the fluorescence lifetime at the extended time scale (1 – 14 ns). The kinetics at 380 nm, however, deviates from the S 1 lifetime due to the generation of the phenoxyl radical. From the known extinction coefficient of the phenoxyl radical (3000 M -1 cm -1 ), 54 it is estimated that the radical yield in cyclohexane is ~5% at 14 ns (with 20% error). An experiment investigating the kinetic isotope effect was also carried out, but no significant difference was observed for PhOH vs. PhOD in cyclohexane within the 20% error, as shown in Figure 5.5 (c). Given the large uncertainty associate with the current experimental method and the fact that there is 50% of PhOH present in the PhOD solutioon, the H/D value could be as large as 3 while there is still no significant difference in phenoxyl radical yield between PhOH and PhOD observed in the TA experiments. Similar experiments were also carried out for the concentrated 90 mM phenol in cyclohexane solution, where the 70% of the phenol molecules still exist in monomer form (see Chapter 6 for detail). From Figure 5.6 we can see that qualitatively, the result is the same as that of 10 mM – there is no appreciable amount of phenoxyl radical observed before 1 ns, but it is again clearly observable after a nanosecond (4% at 14 ns, similar to 10 mM solution). The kinetics of the 475 nm feature is in accord with the fluorescence lifetime, but a new pronounced feature is observed at 600 nm, which clearly has a different kinetics compared to other spectral features. We will return to discuss the nature 144 of this feature in the section where the data for aqueous phenol solution is presented, since it is possible to prepare high concentration of quenchers in water (HCl as electron scavenger, CsCl as singlet quencher and triplet enhancer). Figure 5.6. (a) 90 mM phenol in cyclohexane TA spectra at nanosecond delay times, measured with 267 nm pump fluence = 80 J/m 2 (2.8 μJ pulse energy and 210 μm spot size). The polarization of the super continuum probe pulse is set to magic angle (54.7˚) with respect to that of the excitation pulse. Inset: phenoxyl radical region at a expanded scale. (b) Kinetics at selected wavelengths obtained from the TA experiment (lines). Inset: Comparison of the kinetics at 475 nm and the fluorescence lifetime obtained from TCSPC experiment (dots, same data as shown in Figure 5.5(b) but difference normalization factor). 5.3.2.1.2 Phenol in Water Experiments were carried out for 90 mM phenol aqueous solution, as to compare to the gas-phase-like environment provided by cyclohexane. From Figure 5.7 we can see that the distinct absorption bands at 380 and 475 nm observed in cyclohexane, which originate from the S 1 excited state absorption, are much less pronounced in water, and their kinetics does not show a significant decay over the first nanosecond. The most pronounced feature at 600 nm is still observed in water, and the rise of this TA signal lasts almost 900 ps. 145 Figure 5.7. (a) 90 mM phenol in water TA spectra at the first nanosecond, measured with 267 nm pump fluence = 35 J/m 2 (0.93 μJ pulse energy and 185 μm spot size). The polarization of the super continuum probe pulse is set to perpendicular with respect to that of the excitation pulse. (b) Kinetics at selected wavelengths obtained from the TA experiment. Mialocq et al claimed a high yield of solvated electron with a rise time of ~ 50 ps by monitoring the kinetics at 630 nm after exciting with 27 ps duration 265 nm pulse, 55 but our spectral signature does not seem to be consistent with that of the solvated electron. Thus, it is important to examine the nature of this band by using H + as an electron scavenger. Figure 5.8 shows no change in the kinetics as well as the absolute TA signal, indicating that there is no significant quenching effect and therefore no appreciable amount of electron is being produced within 900 ps upon 267 nm excitation. In light of this, it can be concluded that the previous result obtained by Mialocq et al. must be originated from 2-photon ionization of phenol. The comparison between our results and those by Mialocq et al. has demonstrated that the broadband probing is a powerful tool in pump-probe experiments, as it provides a whole spectrum instead of single point. 146 Figure 5.8. kinetics of the 600 nm feature obtained in 90 mM phenol in water solution (black), 90 mM phenol + 0.17 M HCl (red), and 90 mM phenol + 0.50 M HCl (green), with estimated 267 nm pump fluence = 17 J/m 2 . The relative polarization between pump and probe is parallel for the “phenol only” trace, and is perpendicular for the two H + quenching traces. The traces are displaced vertically for ease of comparison. Inset: raw data without vertical displacement. It is thus established that the broad 600 nm feature is not due to the absorption of solvated electron in water. Another possible assignment is the electronic absorption of the triplet (solely from the consideration of the very slow rising time scale). To confirm or rule out this possibility, CsCl salt was added to the aqueous phenol solution, in order to introduce heavy atom effect and thus increase the intersystem crossing (ISC) rate. 56 Figure 5.9 presents the spectra and kinetics of 90 mM phenol in water with 0.5 M Cs + salt added to the system. Qualitatively, heavy atom effect is indeed observed – the 475 nm feature has a much faster decay time, and the 600 nm seems to reach its maximum much faster but the overall TA signal is greatly reduced. 147 Figure 5.9. (a) 90 mM phenol and 0.5 M CsCl in water TA spectra, measured with 267 nm pump fluence = 35 J/m 2 (0.93 μJ pulse energy and 185 μm spot size, measured back-to-back with data shown in Figure 5.7). (b) Kinetics at selected wavelengths obtained from the TA experiment. (c) and (d) Comparison of phenol solution with and without CsCl for 475 nm kinetics (c), and 600 nm kinetics (d). The “simulate” trace (green) is obtained from multiplying “phenol only” trace (black) by the following exponential functions: 0.62 × exp(t / 850ps) + 0.31 for 475 nm, 0.73 × exp (t / 850ps) + 0.16 for 600 nm. In order to understand the kinetics observed in our TA experiment, it is essential to consider the heavy atom effect on fluorescence quantum first to avoid confusion. This is due to the fact that Cs + will increase the ISC rate between S 1 and T n as well as that between T n and S 0 , which inevitably makes it impossible to predict the T n population 148 without knowing the exact rates for these two processes. However, Zechner et al had studied the phenol fluorescence quantum yield as a function of Cs + concentration and found that they strictly follow Stern-Volmer relationship, 56 which means that the Cs + salt can be consider a singlet quencher that follows a simple bimolecular quenching process (pseudo first order). We indeed observed this quenching effect in our TCSPC experiments upon addition of Cs + , which produce a single exponential decay with an 800 ps decay constant (see Figure 5.10). This is to compare with the fluorescence lifetime in the unquenched condition, where a 3.3 ns single exponential decay was observed. This lifetime in water is slightly longer than that observed in cyclohexane. Figure 5.10. Fluorescence lifetime of 0.1 mM phenol in various solvents (In D 2 O the lifetime measured is for PhOD). The sample was excited with 266 nm and the fluorescence was detected at 300 nm at magic angle. All three traces were fitted to a single exponential decay and the resulted time constants are shown in the graph. The heavy atom effect of CsCl is confirmed by using the same concentration of NaCl. Compared to the fluorescence lifetime, the TA experiments produce a similar result upon the addition of Cs + . Figure 5.9 (c) and (d) show that the kinetics at both 475 nm and 600 149 nm exhibit a decay which can be described by a single exponential function with 850 ps time constant (green traces). This means that although the kinetics in the unquenched condition is very different for the two wavelengths, both of them are of S 1 ESA nature. The pronounced 600 nm band is not likely to originate from any triplet, since the increased ISC rates for populating as well as depopulating the triplet cannot result in a kinetics which can be described by a single exponential function. Note that the single exponential functions employed have a constant offset which implies the existence of a longer-lived state as an underlying feature. This is based on that fact that a simple single wavelength analysis at 380 and 475 nm in the TA spectrum failed to produce good agreement to the fluorescence lifetime – the kinetics obtained from TA experiments is too long compared to 3.3 ns, which suggests a slow rise of an underlying feature. From the offsets obtained from simulating the kinetics at 600 and 475 nm, we can imagine that the underlying feature has high absorption at the blue region. Moreover, it is worthy noting that the Cs + does not affect the kinetics at 385 nm as much as that at 475 nm, both of which were determined to be S 1 ESA in cyclohexane. This suggests the triplet contribution increases towards the blue region, which is in line with the triplet spectrum obtained by Bent and Hayon – the triplet ESA peaks at ~250 nm and extends to at least 450 nm in the visible region. 52 To summarize the results for phenol in water described above, neither phenoxyl radical nor solvated electron is observed in aqueous phenol solution within the first nanosecond. 150 The absence of solvated electron is confirmed by the H + quenching experiment. The TA signal in the UV-visible region is dominated by S 1 ESA, with some contribution of triplet as a weak underlying feature. The strong absorption at 600 nm is not originated from any triplet absorption, as confirmed by the Cs + quenching experiment. The kinetics and the origin of this strong feature will be discussed in Chapter 6. Figure 5.11. Side-by-side comparison of TA spectra of 90 mM phenol and phenol-OD in water (left column) and D 2 O (right column), respectively. Data were obtained with 267 nm pump fluence of 55 J/m 2 (1.9 μJ pump energy and 210 μm spot size). 151 Spectra of 90 mM phenol in water at time delay 1 ≤ t ≤ 14 ns were also obtained. Similar to cyclohexane solutions, phenoxyl radical is clearly observed in this time scale (Figure 5.11 (a) and (c)). Unlike cyclohexane, however, after the 600 nm band has decay away, the spectral signature for solvated electron become very clear and it is confirmed by H + quenching experiments (Figure 5.12). From the known spectrum and extinction coefficient of solvated electron in water and those for phenoxyl radical, it is estimate that the concentration ratio of solvated electron vs. phenoxyl radical is ~1 : 1 within our error, and the solvated electron yield is ~ 7% (at 13 ns). Figure 5.12. Spectra of 90 mM phenol in aqueous solution with and without H + (electron scavenger) at 13 ns delay time. Experiments were obtained with 267 nm pump fluence of 35 J/m 2 (3.6 μJ pump pulse energy and 360 μm spot size). The polarization of pump and probe was aligned at magic angle (54.7˚). Deuterium substituted phenol at the hydroxyl position was used to study the kinetic isotope effect. Figure 5.11 (b) and (d) shows the spectra at selected delay time and the 152 kinetics of selected wavelengths for 90 mM PhOD in D 2 O, from which we can see that phenoxyl radical was generated with approximately the same time scale and yield, indicating that the kinetic isotope effect is negligible. 1H-NMR spectra taken before and after a 2-hour experiment do not show any proton peak due to the hydroxyl group, indicating the absence of Phenol-OH in the D 2 O solution. Note that this is distinct from the PhOD in cyclohexane experiment, since the excessive solvent D 2 O prevents the formation of phenol-OH). 5.3.2.1.3 Phenol in Ethanol Ethanol is another protic solvent which is capable of forming H-bond with phenol. From the UV-visible spectra we can see that although the loss of vibrational structure is common between the two solvent, both S 1 and S 2 bands are red-shifted in ethanol, indicating different H-bonding configurations. Note that λ pump = 267 nm still excited phenol molecules to the S 1 excited state in spite of the red-shift. The TA spectra of 90 mM phenol in ethanol, as shown in Figure 5.13, bear remarkable resemblance to those of 10 mM phenol in cyclohexane, in which three distinct features are observed at 380, 475 and 600 nm. Note that the positions of these bands are similar in the two very different solvents. The 600 nm band, however, exhibits a slow rise before 100 ps, then a slight decay that matches those of 380 and 470 nm. 153 Figure 5.13. (a) 90 mM phenol in ethanol TA spectra at selected time delays before 1 ns, measured with 267 nm pump fluence = 80 J/m 2 (2.8 μJ pulse energy and 210 μm spot size). The polarization of the super continuum probe pulse is set to magic angle (54.7˚) with respect to that of the excitation pulse. (b) Kinetics before 1 ns at selected wavelengths. (c) TA spectra at selected time delays from 1 to 14 ns. (d) Kinetics at selected wavelengths obtained from the TA experiment (thick lines) and fluorescence life time obtained from TCSPC experiment (thin line). The spectral signature for phenoxyl radicals is not observed before 800 ps but becomes apparent after 1 ns. At 14 ns, the phenoxyl radical band at 400 nm is most pronounced, and an underlying offset across the whole probe wavelength region is revealed after the decay of the S 1 bands. From the extinction coefficient of phenoxyl radical at 400 nm, it is estimated that the radical yield is ~3% at 14 ns (contribution from underlying feature 154 subtracted). It is important to point out that the appearance of the phenoxyl radical is not accompanied by the observation of solvated electron in ethanol, which has a broad spectrum centered at 690 nm and a lifetime of hundreds of microsecond at room temperature. 57 The absence of solvated electron at all delay times strongly suggests that the phenoxyl radicals are not generated via any ionization mechanisms. The fluorescence lifetime of phenol in ethanol is 4.3 ns (c. f. 3.3 ns in water), which matches well with the TA kinetics at 475 nm (4.3 ns decay plus an offset). 5.3.2.2 200 nm Excitation of Phenol Solutions 5.3.2.2.1 Phenol in Cyclohexane Calculation shows that the second ππ* state (2 1 ππ*(S 3 )) is accessible via 200 nm excitation, and the second πσ* state lies in close proximity. 20 In solution phase, however, the density of states is high in this region and thus it is difficult to ascertain exactly which states are initially excited. Nevertheless, 200 nm photon (6.2 eV) provides energy much closer to the vertical ionization threshold in the phenol solution, e.g., the ionization potentials is 7.9 eV in aqueous solution. 58 155 Figure 5.14. (a) 10 mM phenol in cyclohexane TA spectra at selected time delays, measured with 200 nm pump fluence = 29 J/m 2 (0.9 μJ pulse energy and 200 μm spot size). The polarization of the super continuum probe pulse is set to perpendicular with respect to that of the excitation pulse. (b) Kinetics at selected wavelengths. (c) TA spectra at selected time delays for pure cyclohexane obtained in the same experimental condition. (d) Kinetics at selected wavelengths for pure cyclohexane. In order to compare and contrast the photochemistry in the gas phase and that observed upon S 1 excitation, 10 mM phenol solution in cyclohexane was first examined with 200 nm excitation. Although cyclohexane produces large amount of TA signal in the long- wavelengths region (as discussed in section 5.3.2.1.1), it does not contribute significantly in the short-wavelength region where the phenoxyl radical absorbs. This is especially true 156 for spectra at t > 1 ps, when a large portion of the cyclohexane signal has decayed. Note that the cyclohexane background TA signal is not subtracted, because there is not enough data point in the power dependence study to confirm the quadratic nature of the TA signal as a function of power. Figure 5.14 shows that, distinct from the abovementioned 267 nm result, phenoxyl radicals were generated within the instrument response time (~180 fs). Moreover, vibrationally excited radicals were generated, signified by the broad and structureless band centered at ~390 nm. The contour of this band contracts in a time scale of ~5 ps, after which the spectral signature of the vibrationally cold phenoxyl radical is observed. 5.3.3.2.2 Phenol in Water 18 mM phenol aqueous solution instead of 90 mM was used due to the high extinction coefficient at 200 nm. Note that this concentration is slightly higher than those used in cyclohexane, and phenol dimers were observed at this concentration in cyclohexane solution (see Chapter 6), but the majority of the phenol molecules still exist in the monomer form. It is unclear the extent of phenol dimer formation in water. Upon 200 nm excitation, the vibrational progression originated from the phenoxyl radical absorption at ~400 nm was observed within the instrument response time in aqueous solutions (Figure 5.15), indicating that vibrationally cold radicals were produced in an ultrafast time scale. Also observed is the well known spectral signature for solvated electron in water – its identity is once again confirmed by H + quenching experiments (data not shown). The rise of TA signal within the first 2 ps is due to the solvation of electron in water, consistent 157 with the model proposed by Jay-Gerin. 59 These observations are to compare with those in cyclohexane, in which vibtrationally hot radicals were initially observed and subsequently cool down. Isotopic substitution on the hydroxyl group was carried out in D 2 O. TA experiments performed on PhOD/D 2 O yielded no significant difference in kinetics of the phenoxyl radical and solvated electron generation (data not shown), as compared to PhOH/H 2 O shown in Figure 5.15. Figure 5.15. (a) 18 mM phenol in water TA spectra at selected time delays, measured with 200 nm pump fluence = 75 J/m 2 (0.85 μJ pulse energy and 120 μm spot size). The polarization of the super continuum probe pulse and the excitation pulse is the same (parallel). (b) Kinetics at selected wavelengths. Inset: Kinetics in the first 10 ps. (c) Early time spectra (t ≤ 1 ps) on an expanded scale. (d) Early time kinetics (t ≤ 1 ps) of 400 nm (red) and 425 nm (green). The large signal at t = 0 ps is due to the two-photon absorption of the solute and/or solvent. 158 Further, Figure 5.15 (a) shows a subtle feature around 410 – 400 nm at 200 fs delay time – the contour of the spectrum is significantly different from the later traces. The TA spectra before 1 ps (Figure 5.15 (c)) reveal that this feature is most pronounced at the earliest delay time, and subsequently decays on a sub-picosecond time scale. From the single wavelength analysis at the band center 425 nm we can see that the 200 fs kinetics is completely from of any two-photon absorption (instrument response), which confirms that this feature is indeed originated from a very short-lived transient species. It is unfortunate, however, that the 425 nm kinetics is affected by the solvation of the electron, which makes it difficult to extract the decay of the short-lived component. The position of this band matches well with that of phenol cation (PhOH + ) measured in solid argon (410 nm) 60 , in organic solvents such as n-butyl chloride (440 nm) 61,62 and cyclohexane (430 nm), 62 and in aqueous solution with high acid concentration (420 nm). 32 Therefore, the subtle feature centered at 425 nm can be assigned to PhOH + absorption. It is important to point out that the lifetime of PhOH + varies dramatically in different solvents. In non-polar organic solvents, its lifetime is on the scale of hundreds of nanoseconds, 62 but PhOH + can be quenched by ethanol via transferring a proton to ethanol to produce phenoxyl radical and C 2 H 5 OH 2 + , with a rate of 6 × 10 8 M -1 s -1 , 62,63 which dramatically shortens the lifetime of PhOH + in ethanol solution to ~80 ps. In water, PhOH + is observed in the pulse radiolysis experiment only with the presence of high concentration of acid, which presumably shifts the acid-base equilibrium to the reactant side and allows the detection of the cation (the estimated pKa of phenol cation is -2.75). 32 159 This implies that PhOH + is extremely short-lived due to its reaction with water to produce phenoxyl radical and H 3 O + . It is worth mentioning that the short lifetime of PhOH + is analogous to H 2 O + , which reacts immediately with the surrounding water and produce OH radical and H 3 O + via proton transfer. 64,65 5.3.2.2.3 Phenol in Ethanol 18 mM phenol in ethanol solution was used in the 200 nm pump experiment. A surprising result is that even at such high photon energy, phenoxyl radical and solvated electron are not observed. The TA spectra at 200 fs ≤ t ≤ 900 ps were dominated by excited state absorption that are similar to that observed in 267 nm experiments. This is distinct from cyclohexane and even water, which suggests that the H-bonding between phenol and the protic solvent plays an important role in the photochemistry even at an excitation energy close to the vertical ionization threshold. Figure 5.16. 18 mM phenol in ethanol TA spectra at selected time delays, measured with 200 nm. (b) Kinetics at selected wavelengths. 160 5.4 Discussion 5.4.1 Fluorescence Lifetime The fluorescence lifetime represent the overall S 1 lifetime of phenol. The S 1 lifetime of an isolated phenol molecule is determined to be ~2 ns by picosecond 19 and nanosecond 66,67 pump-probe photoionization experiments carried out in supersonic free jet. Interestingly, For phenol clusters where a phenol molecule forms a H-bonded complex with one water molecule or another phenol molecule, the S 1 lifetime is significantly increased from ~2 ns to ~18 and ~16 ns, respectively. 66,67 This observation is ascribed to the elimination of the otherwise important internal conversion (IC) back to the S 0 . The H-bond lowers the vibrational frequency of the O–H stretch, and in turn makes the latter an inefficient acceptor mode (requires more quanta) for the IC by dramatically decreasing the IC rate. 66,67 More interestingly, the cluster behavior failed to transfer to the condensed phase. The most direct comparison is the fluorescence lifetime of phenol in water, excited with 267 nm, is only 3.3 ns as opposed to 18 ns observed in phenol-water complex. The small (but statistically significant) difference between the S 1 lifetime of phenol aqueous solution and that of the cyclohexane solution suggests the following. (a) The IC rate becomes faster (compared to that in phenol-water complex) in the condensed phase, (b) the rate(s) of other competitive pathway(s), such as ISC, become faster, (c) new competitive channel, which is inactive in the phenol-water complex, is turned on in the condensed phase, or (d) combinations of two or more abovementioned possibilities. 161 Table 5.1. Quantum yield and rate of various deactivation channels following excitation to phenol S 1 state. Q F Q ISC Q IC Q diss Q e k ISC 10 7 s -1 k IC 10 7 s -1 PhOH (g) – 0.38 a / 0.1 b 0.54 a / 0.84 b – – 19 a / 5 b 27 a / 42 b PhOH·H 2 O (g) – 0.86 a / 0.35 b ~0 a / 0.09 b – – 4.8 a / 2.3 b 0.6 b (PhOH) 2 (g) – 0.59 a ~0 a – – 4.3 a – PhOH in C 6 H 12 /C 6 H 14 0.075 c / 0.083 d 0.24 c / 0.27 d 0.56 c / 0.58 d 0.13 c / 0.07 d < 10 -3 e 13 d 26 d PhOH in H 2 O 0.14 e / 0.12 f,g – – 0.002 e 0.021 e / 0.03 g – – PhOH in CH 3 OH/C 2 H 5 OH 0.19 d / 0.21 f 0.67 d 0.07 d 0.07 d – 14 d 1.5 d a Ref. 66 b Ref. 67 c Ref. 68 d Ref. 51 e Ref. 69 f Ref. 70 g Ref. 71 Table 1 summarizes the quantum yield of fluorescence (Q F ), triplet (Q ISC ), internal conversion (Q IC ), dissociation (Q diss ) and ionization (Q e ) upon excitation to the S 1 state, as well as the rate of ISC (k ISC ) and IC (k IC ) following the excitation, from which we can see that the rates of IC and ISC in cyclohexane measured by Hermann et al. 51 is very similar to those in the gas phase measured by Sur et al., 66 and in both gas- and condensed phase, the rate of IC decreases dramatically upon H-bonding formation with water or ethanol. As the IC slows down, the ISC becomes the fastest deactivation channels and thus has the largest quantum yield, even though the ISC rate itself is not significantly affected by the solvents. Another important deactivation channel, radiative decay, is also not dramatically altered by the solvent environment as determined by Berlman. 44 The generation of H atom and/or electron is much slower compared to the IC, ISC and 162 radiative decay, which results in lower quantum yield and small effect on the overall S 1 lifetime as observed in the TCSPC experiments. Therefore, the decrease of IC rate is entirely responsible for the S 1 lifetime difference between nonpolar and polar protic solutions. The different lifetime between the PhOH·H 2 O in the gas phase and the aqueous phenol solution (i.e., 18 ns vs 3.4 ns) is due to the fact that the ISC is reduced in the gas phase upon complexation, but not in the condensed phase. 5.4.2 TA experiments The principal observations of the TA study are as follows. Upon excitation to the S 1 manifold, the TA spectra of phenol in cyclohexane from 150 fs ≤ t ≤ 900 ps are dominated by the excited state absorption of S 1 , which is present within the instrument resolution (150 fs) and has kinetics matched the fluorescence lifetime as measured in TCSPC experiments. Moreover, the spectrum at 100 ps matches well with that measured by Hermann et al 51 and in full accord with our excited state calculation obtained at EOM- CCSD/aug-cc-pVTZ level. Spectral signature of phenoxyl radical is not observed within the first nanosecond but becomes much more apparent afterwards, suggesting a generation of radical in a nanosecond time scale. Similar results are obtained in water – phenoxyl radicals are observed in a very slow time scale comparable to the cyclohexane solution. On the other hand, solvated electrons are observed alongside with phenoxyl radical, as confirmed by H + quenching experiment. In ethanol, the time scale of the phenoxyl radical generation is similar to that in cyclohexane and water, but spectral signature of solvated electrons is absent. 163 Upon 200 nm excitation of phenol in cyclohexane solution, vibrationally excited phenoxyl radicals are observed within the instrument resolution (~180 fs) and subsequently cool down within 5 ps. In aqueous phenol solutions, however, vibrationally cold phenoxyl radicals are generated within the instrument resolution, together with solvated electron. Lastly, neither phenoxyl radical nor solvated electron is observed in ethanol solutions when exciting with 200 nm (within 1 ns); the TA signal in ethanol was dominated by excited state absorption. 5.4.2.1 Excitation to S 1 The most surprising result is the observation of the phenoxyl radical generated in a nanosecond time scale when the bound 1 1 ππ*(S 1 ) is excited. In the gas phase, H atom is observed following excitation to the bound S 1 state, with no recoil anisotropy observed, indicating that the bound 1 1 ππ*(S 1 ) state is coupled to the dissociative 1 1 πσ*(S 2 ) state, and photofragmentation occurs in a time scale slower than the molecular rotation. 17 This observation is interpreted by H atom tunneling underneath the S 1 /S 2 conical intersection. 19,20,72 In the solution phase, H atom is observed by steady state scavenging method (monitoring the H 2 yield), 47,71 and by flash photolysis (monitoring the absorption intensity of the phenoxyl radical at 400 nm). 51,68 A ~10% yield (upper limit) of H atom can be concluded from these studies, but no time scale of the photolysis is giving due to lack of time resolution. Our TA experiments at 267 nm excitation clearly reveal the slow generation of phenoxyl radical (and solvated electron in the case of aqueous solution). This immediately excludes the possibility of dissociation by direct excitation to the πσ* 164 state, since the O–H bond fission following direct excitation to this dissociative state would occur on femtosecond time scale, similar to the S–H bond fission observed in p- methylthiophenol. 13 Solely based on the time scale consideration, several possible mechanisms can be used to explain the very slow generation of phenoxyl radical, which are: a) step-wise reactions in which charged intermediates are produced, i.e., generation of phenol cation followed by deprotonation, or generation of excited state phenolate via excited state proton transfer, followed by electron transfer, b) dissociation on the triplet state PES, c) dissociation on the ground state PES, d) tunneling underneath the S 2 /S 1 conical intersection (analogous to the gas phase interpretation), e) concerted proton- coupled electron transfer in which charged intermediates are not observed, but depending on the solvent polarity, the final product can be either neutral radicals or ions and thus may not be distinguishable from dissociation via tunneling. These candidates will be carefully examined in the following discussion. 5.4.2.1.1 Step-wise Reactions First we consider the ionization mechanism, in which the 267 nm photon ionizes phenol and results in phenol cation (PhOH + ) and solvated electron. PhOH + subsequently deprotonates and generates phenoxyl radical. Thus, the transient species present in the system according to this picture would be the simultaneous production of a solvated electron and the phenol cation, with a subsequent rise of the phenoxyl radicals upon deprotonation. In the case of cyclohexane, the solvent cannot provide favorable solvation to the charged reaction intermediates and products. The thermodynamic threshold to 165 produce fully solvated charged products is thus expected to be very close to the VIP. Therefore, vertical ionization or autoionization is not likely to play a significant role in the deactivation process. Although the spectral signature of the solvated electron is somewhat elusive (From the work of Siebbeles et al 50 , one can infer that the solvated electron may peak at 1000 – 1550 nm), its geminate partner phenol cation has a distinct absorption feature centered at 430 nm with a lifetime of 270 ns in cyclohexane. 62,63 Our TA spectra do not show any distinct features at 430 nm at any delay time. Thus, it is safe to conclude that the ionization mechanism does not play a role in cyclohexane. The same is true for the ethanol solution due to the non-observation of the solvated electron at all delay times. The results for aqueous solution, however, are far less conclusive. First of all, the thermodynamic threshold for producing fully solvated products are significantly lower as compared to the VIP. As Figure 5.2 shows, this threshold is ~4.5 eV – slightly below the S 1 origin. This is means that 267 nm excites the phenol/water system above the threshold but well below the VIP (7.9 eV as shown in Figure 5.2), thus it is energetically feasible to produce solvated electrons. In our 267 nm experiments, solvated electron is indeed observed. This immediately excludes homolytic bond fission and strongly suggests ionization. Although the spectral signature of PhOH + is not observed, our 200 nm data have shown that it has a lifetime of sub-picosecond before it transfers the proton to the solvent and generates phenoxyl radical. Therefore, the non-observation of PhOH + cannot exclude the possibility of ionization if the electron transfer occurs on a nanosecond time 166 scale (rate limiting step). But what is the ionization mechanism? In here we will first discuss autoionization. From our 200 nm experiment it can be determined that autoionization occurs within the 180 fs instrument response time, but its rate is expected to decrease exponentially as the excitation photon energy decreases. This is due to the decrease in density of states which couples the phenol PES to the solvent continuum. Thus, it is entirely possible for the autoionization time scale increase from less than 180 fs to more than 1 ns, when the excitation energy decreased by 1.55 eV. In light of this, we cannot exclude the possibility of autoionization. Next we will consider the possibility of excited state proton transfer. It is well known that upon excitation to S 1 , the pKa of phenol is decreased from ~9.8 to ~4. 73 Due to the dramatic increase of the acidity, in a dilute aqueous solution of 10 mM phenol, the excited state molecules in S 1 are prone to transfer the proton to the nearby water molecules. This would produce an excited state phenolate (PhO - * ) which has a lifetime of ~22 ps and readily ejects an electron. 54 This suggests that the deprotonation must be the rate limiting step (> 1 ns) in order to explain the nanosecond generation of the phenxyl radical and solvated electron. If this is indeed the case for aqueous phenol solution, an H/D effect is expected. For example, the deprotonation time for excited state 1-naphthol is 36 ps, but it is dramatically increased to 133 ps for the isotopically labeled 1-naphthol- OD. 74 Therefore, our PhOH and PhOD experiments exclude the excited state proton transfer process as no significant KIE was observed (see Figure 5.11). Further, excited state proton transfer is strongly pH dependent, i.e., deprotonation will not be favored 167 under high acid concentration. However, our observation shows that in the aqueous phenol solution with high acid concentration (as in the electron quenching experiments, Figure 5.12), phenoxyl radical is still observed. In light of the lack of KIE and pH dependence, it can be concluded that excited state proton transfer does not involve the slow generation of phenoxyl radical. This is consistent with the kinetic evaluation by Zechner et al, in which the authors concluded that less than 3% of the phenol molecules would undergo dissociation within the lifetime of the S 1 , based on the excited state acidity pKa* = 4. 56 This is also consistent with the analysis by Pines, 36 Hasselbacher et al. 75 and Laws et al., 76 in which it is concluded that proton dissociation is not observed for aqueous phenol solution, due to the low excited state acidity compared to the strong photoacids. To summarize the above discussion, the two step-wise reactions, in which either a cationic (PhOH + ) or an anionic (PhO - * ) reaction intermediate is first created in a rate limiting step, are determined to be irrelevant in the slow phenoxyl radical generation upon exciting phenol to S 1 . 5.4.2.1.2 Triplet State Dissociation Prior the enunciation of the dissociative singlet 1 1 πσ* (S 2 ) state, Dellonte et al. attempted to rationalize the photodissociation of phenol by invoking a dissociative triplet, which is populated via ISC from the initially prepared singlet state. 77 However, this mechanism is in direct contradiction with the analysis by Zechner et al. – in aqueous phenol solution 168 with Cs + ion which induces the heavy atom effect to promote ISC, the relationship between the concentration of electron (and hence the phenoxyl radical) and the Cs + concentration follows the Stern-Volmer relationship strictly, indicating that the electron and the phenoxyl radical are generated from a singlet state instead of triplet state. 56 Our TA experiments also argue against the triplet mechanism. First of all, there is no significant enhancement in phenoxyl radical yield when Cs + is added to the aqueous phenol solution, as observed in the experiments with 267 nm excitation. Moreover, there is no clear correlation between the triplet and phenoxyl radical yield – Table 5.1 shows that the triplet yield in ethanol is almost 2.5x higher than that in cyclohexane, but the radical yield is not significantly enhanced in the former. This is also consistent with our TA measurements where the phenoxyl radical and triplet spectra can be directly monitored. The underlying feature across our probe wavelength range is weak in cyclohexane but quite pronounced in ethanol, indicating that the ISC is much more efficient in the latter (e.g., compare the 14 ns spectrum in Figure 5.6 (a) and Figure 5.13 (c)). However, phenoxyl radical yields in these two solvents are very similar to each other (5±1% in cyclohexane, 3±1% in ethanol). Therefore, it can be safely concluded that the lack of clear correlation between the triplet and radical yield argues against the triplet dissociation mechanism. 169 5.4.2.1.3 Ground State Dissociation The S 2 ←S 0 oscillator strength for an isolated PhOH is small, 14,20 and there is no evidence supporting a dramatic increase in the solution phase – for example, the absorption cross section at ~240 nm is negligible in cyclohexane (Figure 5.3). Therefore, upon 267 nm excitation, the initially populated excited state of phenol is unambiguously S 1 , as predicted by ab initio calculations, indicated in the steady state absorption spectrum, and clearly observed (both spectrally and kinetically) in our TA spectrum. The photon energy provided by 267 nm is higher than the S 1(v=0) ← S 0 transition energy, but is lower than the location of the S 1 /S 2 CI. Intuitively, the deactivation pathways of excited state PhOH in this region include transfer to S 0 via internal conversion (IC) or radiative emission, and to T n via intersystem crossing (ISC). Early studies 66,67 suggest that these three channels are responsible for >80% of the excited state deactivation for isolated PhOH molecules, with IC being the major channel. However, O–H bond fission still occurs at energies beneath the S 1 /S 2 CI, on a ns timescale, and with a mean TKER of ~5600 cm -1 observed in the HRA-PTS experiments. 17 In light of the large predicted IC yield and slow dissociation time scale, the O–H bond fission process was first (incorrectly) thought to occur via coupling at the S 0 /S 2 CI after population (by IC) of high vibrational levels of S 0 with substantial O–H stretch character (see Figure 5.1). 14,17,18,66 This hypothesis has since been reanalyzed and refuted by new experiments (see next section), and the present condensed phase experiment provide further evidence for rejecting this mechanism. The current results show a pattern of 170 reactivity in liquid cyclohexane similar to that observed in the gas phase, namely long- lived S 1 and appearance of PhO( X % ) radicals but only on a nanosecond timescale. Since the solvent provides an effective “sink” to dissipate excess vibrational energy, 22-25 a highly excited O–H vibration could not survive on S 0 without vibrational energy transfer to adjacent cyclohexane molecules. For example, for dilute methanol in carbon tetrachloride solution where H-bonding is absent, the relaxation time of the O–H(v=1) stretch level was determined to be <10 ps. 78 Therefore, the phenoxyl radicals produced on such a nanosecond time scale are unlikely attributable to this ground state dissociation mechanism. 5.4.2.1.4 Tunneling Under the S 1 /S 2 CI So what accounts for the slow dissociation of phenol? The gas phase data has recently been re-interpreted in terms of H atom tunneling from the S 1 state under the lower diabats of the S 1 /S 2 CI, 19,20 as originally proposed by Sobolewski, Domcke and co-workers. 72 Evidence for this idea comes from HRA-PTS experiments which revealed that the geminate partner of the H atom, the phenoxyl( X % 2 B 1 ) radical, is formed in a very limited subset of the available vibrational state density, predominantly an odd progression of v 16a , the nuclear motion advanced as the coupling mode facilitating the non-radiative S 1 →S 2 transfer beneath the S 1 /S 2 CI. 20 Pino et al. measured an S 1 state lifetime of ~2.4 ns using pump-probe ion time of flight measurements 19 and, very recently, Roberts et al. have observed the H atom elimination process directly by time-resolved VMI and deduced an appearance time ≥ 1.2 ns (limited by the maximum delay measurable in their apparatus) 171 for translationally fast H atoms. 79 Both authors appeal to a simple 1D tunneling model in just the O−H stretch dimension to justify a nanosecond tunneling lifetime. This tunneling would only become more significant with a 2D model because of the lower and narrower barrier due to the coupling mode (ν 16a ) in the branching plane; Dixon et al. estimated a 40% greater tunneling using a 2D wavepacket model compared to 1D picture. 20 Roberts et al. 79 found that the measured H atom appearance timescale was seemingly invariant as the excitation wavelength was tuned between 275 and 255 nm, and argued that this was strong evidence for a tunneling mechanism, since IC to S 0 would be expected to show a strong excitation energy dependence on account of the increasing density of vibrational states. (Frequency resolved HRA-PTS experiments had previously shown that excited state O−H bond fission occurred irrespective of the initially photoprepared S 1 vibrational state. 17,18,20 The product TKER spectra reveal a marked propensity for S 1 parent vibration to carry through into the phenoxyl radical products − i.e. to act as a “spectator” to dissociation. Thus the departing H atom experiences essentially the same tunneling barrier, irrespective of the particular S 1 (v) state excited.) However, the parent lifetimes measured by Roberts et al. do decrease with increasing photon energy, which is hard to bring into accord with an explanation based on tunneling in competition with IC. Simple branched kinetics dictates that if parallel processes deactivate the parent, even if the rate linking parent to product is constant, the appearance time of the product must match the decay of the parent. Thus, while the gas phase ultrafast VMI results 172 provide some support for the tunneling model, they are not definitive and could also be rationalized with the earlier ground state dissociation mechanism. 17,18 It appears that the surrounding cyclohexane environment provides little change to the S 1 PES at small O–H bond extensions, which preserves the dynamical behavior of gas phase PhOH. The dense solvent environment might be expected to push up the Rydberg states of a molecule, however, due to Pauli repulsion of solvent electron density with the diffuse solute Rydberg orbital, thereby changing the shape of the S 2 PES – particularly at short R O–H where the Rydberg character is higher. Given such notions, we might expect some possible change in the shape of the tunneling barrier, but the current experiments suggest that the tunneling rate cannot be changed substantially from the gas phase otherwise either the S 1 lifetime would be reduced in solution or a vanishingly small phenoxyl radical yield would be difficult to detect. The fate of the H atom generated from the homolytic O–H bond fission can be related to the generation of the solvated electron. In this case, rather than being a competing process, dissociation can play an important role in ionization, providing that the H atom possesses sufficient energy to react with a neighboring water molecule via the following reaction: H + X (l) ↔ HX + (aq) + e - (aq) (X = C 6 H 12 , CH 3 CH 2 OH and H 2 O) (1) This equilibrium is observed in high temperature reaction of H atom with water from 100 to 250 ˚C in pulse radiolysis reactions. 80 The thermal activation energy of the forward reaction is ~0.7 eV. 65,80,81 From Figure 5.2 we can see that the generation of PhO· (aq) + 173 H 3 O + + e - (aq) is energetically accessible from the bottom of the phenol S 1 state. Using the excitation photon energy of 4.65 eV (267 nm) and the gas phase bond dissociation energy of 3.7 eV, 17 the upper limit of the total available energy to the dissociated fragments (phenoxyl radical + H atom) is 0.95 eV, among which ~0.75 eV is partitioned to the kinetic energy. 17 Thus, the reaction of translationally hot H atom with water does seem energetically feasible if the phenol-water H-bond and the solvent environment does not significantly alter the energy disposal in the condensed phase. This picture is analogous to that of the low energy ionization of liquid water. Upon 8.3 eV excitation of water to the adiabatic conduction band, a very energetic H atom is ejected and subsequently transfers a proton to the adjacent water molecule, thereby generates a solvated electron. 64,65,81 This mechanism is pleasing because it suggests that the phenoxyl radicals are generated via the same pathway in three different solvents and thus inline with their similar yields. The observation of electron in water is simply the consequence of the ejected H atom reacting with water. The premise of this reaction is that the H atom possesses enough energy to overcome the 0.7 eV barrier. In ethanol, the activation energy of an H atom reacting with the solvent to produce an electron and C 2 H 5 OH 2 + could be higher than the total energy available to the system after dissociation. This hypothesis has some ground based on the photodissociation of p-MePhSH in ethanol, as discussed in Chapter 4. It is established that upon 271 nm excitation in the gas phase, p-MePhSH dissociates via direct excitation or efficient coupling to the dissociation state 1 1 πσ* (S 2 ) state, and 174 produces an H atom with at least 1.1 eV of translation energy (corresponds to ground state p-MePhS radical). In ethanol solutions, the p-MePhS radical is observed within our best instrument response time (~50 fs) upon 271 nm excitation, but no solvated electron is observed at any delay time. Even with 200 nm excitation, in which very energetic H atom is produced in the gas phase (~2 eV of translational energy), solvated electron is still not observed. 13 The non-observation of solvated electron in ethanol implies the high barrier of the proton transfer reaction from H atom to ethanol. In comparison, a more possible reaction of H atom with ethanol is the H atom abstraction of the solvent, which has an activation energy of 0.25 eV, 82 much lower than the proton transfer reaction. The abstraction reaction would produce two hydroxyethyl radical isomers and H 2 . One of the isomers, CH 3 CHOH, has absorption onset at 300 nm and peaks at ~250 nm, 83 which is outside of our super continuum probe range. Proton transfer from H atom to cyclohexane is unlikely due to the unfavorable interaction between the non-polar solvent and the charged species (electron and C 6 H 13 + ). Therefore, H atom is more prone to react with cyclohexane via H atom abstraction to produce H 2 and cyclohexyl radical, whose electronic absorption spectrum is also in the deep-UV region (peaks at ~250 nm). 84 From the above discussion we can see that dissociation via tunneling provides a possible explanation to the observation of the phenoxyl radical in cyclohexane and ethanol, and the lack of electron is a direct result of the thermodynamically unfavorable proton transfer reaction from the H atom to the solvent molecule. In the case of aqueous phenol solution, although it is energetically feasible for reaction (1) to take place, the 175 completeness of the reaction deserves some discussion. Shiraishi et al measured the equilibrium constant and rate constants of the forward and backward reaction from pulse radiolysis of water at various temperatures. It was concluded that the equilibrium constant for reaction (1) as written is < 10 -6 at 250 ˚C, with a forward rate on the order of 10 5 s -1 and backward rate of 10 11 M -1 s -1 . 80 This suggests a very small amount of the solvated electron present in the system with majority of the H atom intact at the given temperature. This is to be compared with our observed 1:1 ratio (with 20% error) of phenoxyl radical and solvated electron at room temperature, which suggests that all H atoms have reacted with water and produced solvated electrons. The very energetic (translationally hot) H atom increases the forward rate much more than the backward rate for reaction (1) and thus dramatically increases the equilibrium constant, thereby drives the reaction to a much more complete extent. The increasing of the forward rate seems to be consistent with the abovementioned low energy water dissociation/ionization, where the solvation of electron (< 1 ps) is the slow step and no delayed generation of electron is observed. 64,65,81 However, it is very unlikely that the increase of the forward rate will be enough to result in the equilibrium shifted dramatically to completion. Considering the excitation of liquid water with 8.3 eV photon energy, where a ratio of OH radical to solvated electron >8 is observed. The available energy partitioned to the translation of the H atom is determined to be as high as ~3 eV from the gas phase bond dissociation energy. 65 This is to be compared with the 0.75 eV available energy in the case of H atom generated from a phenol molecule via tunneling under the S 2 /S 1 conical intersection. 176 Therefore, it can be concluded that the 1:1 ratio of phenoxyl radical and solvated electron cannot be resulted from O–H bond fission followed by hot H atom chemistry. 5.4.2.1.5 PCET The experimental observables discussed so far are in line with the hypothesis that phenol exhibits homolytic O–H bond fission via tunneling under the S 2 /S 1 conical intersection in cyclohexane and ethanol. However, the significant charge separation in aqueous phenol solution, i.e., observation of the solvated electron, cannot be explained by the homolytic O–H bond fission followed by hot H-atom dissociation (proton transfer to solvent and produce an electron). Thus, the photochemistry of excited state phenol in water solution must be governed by a different mechanism. We have previously discussed the possibility of autoionization. This section presents another candidate: PCET. Analogous to the PCET reactions in enzymes where specific sites exist to accept electrons and protons, it is possible that phenol produces a proton which binds to an adjacent water molecule to make a H 3 O + and an electron which is subsequently solvated by the surrounding water molecules. This would correspond to a bidirectional PCET where the proton transfer and the electron transfer have different accepting sites but no stable charged intermediate is observed. 35 The question remained, however, is how and when the charge separation and redistribution happened. On the one hand, it is possible that the O–H bond fission starts homolytically and strictly along the O–H distance coordinate in which the tunneling mechanism discussed above still play an important role, but when the distance between the O and H atom is large, charge separation occurs, presumably with 177 the aid of the polar water molecules. On the other hand, it is also possible that the charge separation occurs instantaneously upon the optical excitation and the proton tunnels along the reaction coordinate in which both O–H distance (and its relevant orthogonal mode(s)) and solvent motions are accounted for. The problem associated with the first possibility, in which H atom tunneling mechanism is still involved, is the lack of kinetic isotope effect (KIE) experimentally, although the magnitude of the KIE deserves detailed discussion. The evolving underlying background (in the case of aqueous phenol solution) interferes with the kinetics of the phenoxyl radical, thereby prevents a clean extraction of the rise time of the radical and consequently the O–H(D) bond fission time scale for PhOH and PhOD. On the other hand, the tunneling rate determines the radical yield, e.g., when comparing TA spectra of PhOH and PhOD, no phenoxyl radical would be observed in the latter if a very large KIE is present. However, Figure 5.11 shows that there is no significant difference between PhOH/H 2 O and PhOD/D 2 O – phenoxyl radical and solvated electron are still observed and the yields of these products are approximately the same as those for PhOH (with 20% error), indicating that there is no KIE or it is too small to be observed giving our experimental method. There are several models can be used to estimate the KIE of H atom tunneling. One of the approaches was Bell’s tunneling correction of Arrhenius equation. In this semi-classical approach based on transition state theory, tunneling of a light particle through an inverse 178 parabola potential barrier is explicitly accounted for, and the correction factor, Q, to the rate of reaction can be expressed as the follows: 85 ( )         − − + − − − ⋅ ⋅ − = L u y u y u y e u u u Q kT E π π π 6 4 2 2 / sin 2 / 3 2 (1) kT hv u = (2) hv E e y π 2 − = (3) and m E a v 2 1 ⋅ = π (4) where E and a are the barrier height and half width of the barrier width along the reaction coordinate (in the case of phenol, it is the distance between the O and H), respectively, m is the mass of the tunneling particle, T is the reaction temperature and k is the Boltzmann constant. Assuming that the zero point energies between PhOH and PhOD have negligible effect on the barrier height, the only difference lies in the different masses of the H and D atom, and the KIE can thus be expressed in Q H /Q D . Using the barrier height of 5400 cm -1 and full barrier width of 0.61 Å, obtained from Dixon et al, 20 the Q H /Q D is estimated to be 290, which indicates that the tunneling rate for PhOD is 290 times slower than that of PhOH. The estimated KIE from the tunneling mechanism based on Bell’s model is too large compared to the observed KIE, which is qualitatively inferred from the phenoxyl radical yield. However, the applicability of the semi-classical treatment has brought to question, 179 mainly because this is a one-dimensional model which only focuses on the hydrogen reaction coordinate (O–H stretch) in the case of bare phenol) and ignores the solvent environment. The one-dimensional issue can be addressed by considering the vibrational modes orthogonal to the O–H stretch coordinate and their effect on barrier height and width. Dixon et al. has determined that the ν 16a (a 2 ), the out-of-plane ring torsion mode which mediates the flux transfer between the S 1 (B 2 ) and S 2 (B 1 ) states, effectively lowers the tunneling barrier height from 5400 cm -1 to 5100 cm -1 and the base width from 0.61 Å to 0.52 Å. 20 As a result, the addition of this coupling mode reduces the KIE from 290 to 160. Therefore, the gas phase analogy of the H atom tunneling underneath the S 1 /S 2 conical intersection, using both the one-dimensional (O–H stretch coordinate) and the two-dimensional (O–H stretch and ring torsion coordinate) model, cannot explain the very small H/D value observed in our experiment. In the framework of proton-coupled electron transfer (PCET), there are numbers of theoretical methods and models to treat H atom (H·), proton (H + ) and hydride (H - ) transfer as a multi-dimensional process and take into account the solvent environments. Their derivation and discussion are beyond the scope of this work, but all of them result from modification of the Marcus theory of electron transfer. A general mathematical description of the rate constant can thus be expressed as the following equation: 86 ( ) ∫ − + Δ − ⋅ ⋅ = 1 0 ) ( 0 / ) ( 4 r r T k E m F RT G dr e e e C k b m F λπ λ (5) 180 where the first exponential is the Marcus term, which is related to the reorganization energy (λ) and insensitive to isotopic substitution (of the solute); the second exponential is the Franck-Condon term, which accounts for the overlap of the vibration wavefunction of the reactant and the product along the reaction coordinate, and it is sensitive to isotopic substitution; the third exponential describes the donor-acceptor distance fluctuation, which depends on both temperature and isotopic substitution. Note that the last two exponential functions are integrated over all donor-acceptor distance (r). The extent of the solvent molecule reorganization is given in the first exponential function. Specifically, it describes the change of solvent configuration in order to obtain the optimal geometry for H·/H + /H - tunneling. If this solvent rearrangement is the rate determining step, i.e., slower than electron transfer (solvent-controlled limit), the PCET process is said to be electronically adiabatic. On the other hand, in the case of fast solvent relaxation (golden rule limit), the PCET is classified as electronically nonadiabatic. 87-89 Different H/D values can be expected for these two limiting cases because different dominating terms in equation (5) exhibit different isotopic dependence. Solvent- controlled PCET reactions are characterized by large solvent reorganization energy and strong electronic coupling. In this limit, the vibronic coupling is insensitive to isotopic substitutions, which dictates that the H/D effect results merely from the zero point energy difference between the unlabeled and deuterium-labeled compounds and thus expected to be small. 87,90-92 For example, Costentin et al. used cyclic voltammetry to study the ground state PCET reaction rate for PhOH in water and PhOD in D 2 O. The H/D effect obtained 181 from these experiments is only 2.5 and the authors classified the reaction as quasi- adiabatic. 92 As the nonadiabaticity increases (solvent motion becomes faster with respect to the electron transfer), the contribution of the isotopic-independent solvent reorganization term to the overall reaction rate becomes smaller. The nonadiabatic PCET reactions are characterized by weak electronic coupling, and its vibronic coupling is very sensitive to the overlap of the vibration wavelength functions between the reactant and product. A larger H/D value is expected due to the deuterium-labeled compounds generally exhibits smaller vibration overlap. 87,90-92 For example, the intramolecular PCET of aminophenol (ground state reaction) bears H/D value of 3.5, which is in the nonadiabatic limit. 92 However, it is worth pointing out that a low H/D value does not necessarily exclude nonadiabatic PCET. Peters et al. measured the reaction rates of H + /D + -coupled electron transfer for the benzophenones/N,N-Dimethylanile contact radical ion pair, 93-95 and later the benzophenone/N-methylacridan contact pair. 96 It was determined that the H/D value is 2 – 2.4 for the former and 1.1 – 1.8 for the latter. In both cases, however, the Arrhenius pre-factors obtained in these experiments are smaller than that expected for adiabatic PCET reactions, and the observed kinetic behaviors suggest the existence of the inverted region. These are characteristics of a nonadiabatic PCET reaction. It is also worth mentioning that the current PCET formalisms mainly focus on the proton tunneling occurring on the molecular ground state. Since the electronic wavefunction of an excited state molecule is more diffusive, the orbital overlap between the reactant and 182 product is more extensive than that in the ground state. 88 As a result, the KIE is expected to be even smaller for the excited state PCET reactions. Given the observation of solvated electron, the phenoxyl radical is very likely to be generated via a PCET mechanism in which the water molecules facilitate the charge polarization such that the reaction does not follow the strict one-dimensional O–H stretch coordinate. Further, in view of the lack of KIE, the charge separation is likely to occur at the early stage of the reaction, thereby opening a new reaction pathway with much smaller barrier in comparison to the aforementioned 1D and 2D tunneling model for H atom. The proton, instead of H atom, thus tunnels across the barrier and forms H 3 O + , and the electron is solvated by its surrounding water molecules. The slow step of this reaction is the solvent reorganization in order to obtain the optimal configuration for the proton tunneling. Note that solvent reorganization in this case may have experienced a large barrier considering the very slow rate observed in our experiments. This is similar to the case of intramolecular PCET reaction of 7-azaindole and its cyano derivatives in methanol solutions. Their proton transfer rates are demonstrated to correlate with the solvent reorganization barrier height, which various from hundreds of picoseconds to several nanoseconds. 97-100 The PCET mechanism for the aqueous solution discussed above raises the question whether the concerted PCET is also applicable to cyclohexane and ethanol solutions. Since the non-polar cyclohexane is not likely to support the increasing charge polarization if H + and e - is produced, PCET is not likely to be involved in the cyclohexane solution. Instead, the phenoxyl radical is likely to be produced by homolytic 183 O–H bond fission via H-atom tunneling under the S 1 /S 2 conical intersection, which is in line with the intuition that cyclohexane provides the best mimic of gas phase like environment. In the case of ethanol solution, since the spectral signature of the solvated electron is not observed at any delay time, it can be concluded that charge separation does not occur during the reaction, and phenol undergo homolytic bond fission to produce a phenoxyl radical and an H atom. 5.4.2.1.6 Summary of 267 nm excitation The findings for the photochemistry following 267 nm excitation of phenol to its 1 1 ππ* (S 1 ) state are summarized as follows. Phenoxyl radicals are generated in a nanosecond time scale in three solvents with different polarities. In cyclohexane and ethanol, the radical is generated via an H-atom tunneling under the S 1 /S 2 conical intersection, a deactivation pathway mirroring the gas phase behavior. In both solvents, the O–H bond fission is homolytic throughout the course of the reaction, as charged intermediates and products are not observed at any delay time. In water, charge separation is evident from the observation of the solvated electron. Thus, considering the observed ~1:1 ratio of phenoxyl radical vs. solvated electron, and the lack of significant KIE, autoionization and bidirectional PCET are two possible reaction mechanisms leading to generation of the charged products. Unfortunately, our current experimental data do not definitively confirm or rule out either mechanism. 184 5.4.2.2 Excitation to Higher Excited State(s) Upon 200 nm excitation of phenol in cyclohexane, vibrationally excited phenoxyl radicals are generated within the instrument response time (~180 fs). This much faster H- atom loss (compared to 267 nm excitation) is consistent with the gas phase experiments, where H atom (with vibrationally excited phenoxyl radicals) is observed in a time scale faster than the molecular rotation period when a phenol molecule is excited with a wavelength shorter than 240 nm. This begs the question of how the radicals are generated in comparison to the 267 nm photolysis. As discussed above, H-atom tunneling and PCET (which involves proton tunneling) occurs on a time scale of nanoseconds, which is much too long given the prompt observation of phenoxyl radical. This leaves two possible candidates – direct excitation or ultrafast coupling to a dissociative state (e.g., S 2 or S 4 ), or autoionization. For cyclohexane solutions, we reiterate that the non-polar medium does not support any charged species, and thus autoionization is not likely to take place at excitation energies lower than the VIP. Spectroscopic evidence from our experiment is in line with this intuition. Although the spectral signature and the extinction coefficient of solvated electron in cyclohexane are unknown at this point, should ionization occur, spectral signature of PhOH + in cyclohexane is well-defined and certainly within our probing range. However, no distinct feature at 430 nm can be observed at all delays times. More importantly, the deprotonation of PhOH + in cyclohexane is on the scale of hundreds of nanoseconds, 62 much too long to explain the ultrafast generation of the phenoxyl radical. 185 Thus, it can be safely concluded that ionization of phenol does not occur in cyclohexane and the generation of phenoxyl radical must be produced via a direct dissociation mechanism. In fact, the observation of vibrationally hot phenoxyl radical in the first 5 ps is consistent within a dissociation mechanism. It is difficult, however, to ascertain which dissociative state is responsible for the O–H bond fission. It is possible that the 1 1 πσ* (S 2 ) state is directly populated by intensity borrowing from S 3 , or higher excited states are initially populated and subsequently deactivate via non-radiative population transfer to the dissociative S 2 . Aqueous phenol solution, on the other hand, does not produce vibrationally hot phenoxyl radical upon 200 nm excitation. Phenoxyl radicals at electronic and vibronic ground state are observed within the instrument response time (~180 fs). Moreover, a small amount of phenol cation is observed in the first ~500 fs, indicating the presence of the ionization pathway. Although the conduction band of phenol is not directly accessible with 6.2 eV photon energy, ionization is still possible when the nuclear rearrangement of the excited state molecule and the solvent molecules provides a favorable geometry for the electron ejection. Autoionization of phenol by 6.2 eV photon is energetically feasible, since the thermodynamic threshold for producing charged products is only 4.5 eV, as shown in Figure 5.2. To the best of our knowledge, the spectral signature and the lifetime of phenol cation in a neutral (or slightly acidic) solution is measured for the first time. The difficulty lies in the extremely short lifetime of PhOH + , which deprotonates to phenoxyl radical within 500 fs. The accurate determination of PhOH + lifetime is not possible 186 giving the current 180 fs instrument response time; a qualitative statement, consistent with our experimental observables, is that the majority of PhOH + deprotonates within 180 fs (and generates phenoxyl radical within this time), and its spectral signature completely disappears after 500 fs. Upon 200 nm excitation, the generation of phenoxyl radical in cyclohexane and water is governed by dissociation and autoionization, respectively. This highlights the important role of the solute-solvent interaction on reaction dynamics. In the cyclohexane environment, the solvent molecules do not interact with the phenol molecule and thus cannot provide any favorable solute-solvent configuration for election ejection. In this case, the ionization threshold is expected to be close to the gas phase value and the reorganization energy would be small, as it is dominated by solute-reorganization alone. As a result, the excited state phenol molecules deactivates via O–H bond fission along the dissociative PES, which is signified by the observation of the vibrationally hot phenoxyl radical. In the aqueous environment where water molecules can act as either H bond donor or acceptor to the phenol molecule, ejection of electron is more favorable with the aid of solvent rearrangement. The large reorganization energy significantly lowers the thermodynamic threshold for producing charged products and thus opens the autoionization channel, which is signified by the observation of the short-lived phenol cation (PhOH + ). 187 5.5 Conclusion The photochemistry of phenol upon 267 nm and 200 nm excitation is studied in cyclohexane, ethanol and water solutions. In cyclohexane, the deactivation channel of excited state phenol is similar to that in the gas phase. Upon 267 nm excitation, the S 1 manifold is initially prepared, from which the H atom tunnels under the S 1 /S 2 CI and leads to the O–H bond fission in a nanosecond time scale. This is similar to the photodissociation of p-MePhSH, in which the H atom tunnels under the S 1 /S 2 CI with a much smaller barrier. Dissociation on the ground state can be ruled out due to the fact that the highly vibrationally excited ground is unlikely to survive for more than tens of picoseconds. Moreover, dissociation from the triplet state can also be excluded due to the lack of correlation between the triplet yield and radical yield. Upon 200 nm excitation, on the other hand, the optically dark S 2 can be populated directly via intensity borrowing for the S 3 state and/or indirectly via efficient coupling to the S 2 surface. This subsequently leads to the ultrafast O–H bond fission with the generation of vibrationally excited phenoxyl radicals. In either excitation wavelength, autoionization or PCET does not play in significant role in the generation of the phenoxyl radical. In protic solvents such as ethanol and water, the situation is further complicated by the possible competitions from autoionization and PCET, which originate from the solute- solvent interaction via H-bonding. In ethanol solution, phenoxyl radical is observed in a nanosecond time scale upon 267 nm excitation, similar to the observation in cyclohexane. 188 Solvated electron is not observed, which strongly suggests photodissociation instead of autoionization and PCET. In water, both the phenoxyl radical and the solvated electron are observed on the time scale of nanosecond upon 267 nm excitation. Photodissociation of phenol, followed by subsequent proton transfer of the translationally hot H atom, can be ruled out as the mechanism for solvated electron generation, considering that the ~1:1 ratio of phenoxyl radical and solvated electron. This is due to the fact that the proton transfer from a hot H atom to the solvent is not a favorable equilibrium. Also, excited state proton transfer followed by electron transfer is not favorable, due to the lack of KIE and the fact that the proton transfer rate is too slow compared to other deactivation processes. The two possible reaction mechanisms are 1) a bidirectional PCET, where the proton and electron has different “acceptor sites”, and 2) autoionization in which the rate of electron transfer is much slower compared to the case of 200 nm excitation. Nonetheless, it can be determined that the charge separation occurs at the very early stage of the reaction, i.e., the H-atom does not following the O–H stretching coordinate to tunnel – a notion consistent with the lack of (or very small) KIE. In the case of 200 nm excitation of phenol, the extremely short-lived phenol cation is observed, and the phenoxyl radical is observed in electronic and vibronic ground state. This clearly pinpoints the autoionization mechanism. Although the excitation energy is less than the VIE, it is enough to access the conduction band of phenol. The cold 189 phenoxyl radicals generated from autoionization in aqueous solution are distinct from the hot radicals generated from photodissociation in cyclohexane solution, which highlights the solvent effect on excited state deactivation pathways. In conclusion, the mechanism governing the generation of phenoxyl radical is highly solvent and excitation energy dependent. This is in contrast with the previously discussed p-MePhSH, which has a negligible barrier under the S 2 /S 1 CI and very weak interaction with the solvent. The weakly interacting cyclohexane molecules provide a gas phase-like environment, in which excited state phenol deactivates via the dissociative S 2 state. However, in aqueous solution where phenol molecules can form H-bonded complexes with the surrounding water molecules, PCET or autoionization is responsible for the nanosecond generation of phenoxyl radical and solvated electron upon 267 nm excitation, and autoionization is responsible for the prompt generation of phenol cation (which readily transfers a proton to water and produce phenoxyl radical) and solvated electron upon 200 nm excitation. 190 5.6 Chapter 5 References 1. Part of this chapter is based on a submitted manuscript: Y. Zhang, T. A. A. Oliver, M. N. R. Ashfold and S. E. Bradforth, Faraday Discuss. DOI:10.1039/C2FD20043K. 2. C. Y. Lin and M. C. Lin, Int. J. Chem. Kinet. 1985, 17, 1025-1028. 3. C. Y. Lin and M. C. Lin, J. Phys. Chem. 1986, 90, 425-431. 4. B. Kok, B. Forbush and M. Mcgloin, Photochem. Photobiol. 1970, 11, 457-&. 5. B. A. Diner, D. A. Force, D. W. Randall and R. D. Britt, Biochemistry 1998, 37, 17931-17943. 6. T. J. Meyer, M. H. V. Huynh and H. H. Thorp, Angew. Chem. Int. Ed. 2007, 46, 5284-5304. 7. J. Bonin and M. Robert, Photochem. Photobiol. 2011, 87, 1190-1203. 8. R. J. Debus, B. A. Barry, I. Sithole, G. T. Babcock and L. Mcintosh, Biochemistry 1988, 27, 9071-9074. 9. B. A. Barry, Photochem. Photobiol. 1993, 57, 179-188. 10. R. J. Boerner and B. A. Barry, J. Biol. Chem. 1993, 268, 17151-17154. 11. T. A. A. Oliver, G. A. King, D. P. Tew, R. N. Dixon and M. N. R. Ashfold, in preparation. 12. Y. Zhang, T. A. A. Oliver, M. N. R. Ashfold and S. E. Bradforth, In preparation. 13. T. A. A. Oliver, Y. Zhang, M. N. R. Ashfold and S. E. Bradforth, Faraday Discuss. 2011, 150, 439-458. 14. M. N. R. Ashfold, B. Cronin, A. L. Devine, R. N. Dixon and M. G. D. Nix, Science 2006, 312, 1637-1640. 15. R. J. Lipert and S. D. Colson, J. Chem. Phys. 1990, 92, 3240-3241. 16. A. L. Sobolewski and W. Domcke, J. Phys. Chem. A 2001, 105, 9275-9283. 17. M. G. D. Nix, A. L. Devine, B. Cronin, R. N. Dixon and M. N. R. Ashfold, J. Chem. Phys. 2006, 125, 133318. 18. M. N. R. Ashfold, A. L. Devine, R. N. Dixon, G. A. King, M. G. D. Nix and T. A. A. Oliver, Proc Nat Acad Sci USA 2008, 105, 12701-12706. 191 19. G. A. Pino, A. N. Oldani, E. Marceca, M. Fujii, S. I. Ishiuchi, M. Miyazaki, M. Broquier, C. Dedonder and C. Jouvet, J. Chem. Phys. 2010, 133. 20. R. N. Dixon, T. A. A. Oliver and M. N. R. Ashfold, J. Chem. Phys. 2011, 134. 21. A. Iqbal, M. S. Y. Cheung, M. G. D. Nix and V. G. Stavros, J. Phys. Chem. A 2009, 113, 8157-8163. 22. C. G. Elles and F. F. Crim, Annu. Rev. Phys. Chem. 2006, 57, 273-302. 23. J. Larsen, D. Madsen, J. A. Poulsen, T. D. Poulsen, S. R. Keiding and J. Thogersen, J. Chem. Phys. 2002, 116, 7997-8005. 24. C. G. Elles, M. J. Cox and F. F. Crim, J. Chem. Phys. 2004, 120, 6973-6979. 25. C. Petersen, N. H. Dahl, S. K. Jensen, J. A. Poulsen, J. Thogersen and S. R. Keiding, J. Phys. Chem. A 2008, 112, 3339-3344. 26. N. Winter and I. Benjamin, J. Chem. Phys. 2004, 121, 2253-2263. 27. C. G. Elles, M. J. Cox, G. L. Barnes and F. F. Crim, J. Phys. Chem. A 2004, 108, 10973-10979. 28. G. A. King, T. A. A. Oliver, M. G. D. Nix and M. N. R. Ashfold, J. Phys. Chem. A 2009, 113, 7984-7993. 29. R. C. Guedes, K. Coutinho, B. J. C. Cabral and S. Canuto, J. Phys. Chem. B 2003, 107, 4304-4310. 30. D. Ghosh, A. Roy, R. Seidel, B. Winter, S. E. Bradforth and A. I. Krylov, Submitted. 31. H. D. Gai and B. C. Garrett, J. Phys. Chem. 1994, 98, 9642-9648. 32. T. N. Das, J. Phys. Chem. A 2005, 109, 3344-3351. 33. C. G. Zhan and D. A. Dixon, J. Phys. Chem. B 2003, 107, 4403-4417. 34. O. Marsalek, T. Frigato, J. VandeVondele, S. E. Bradforth, B. Schmidt, C. Schutte and P. Jungwirth, J. Phys. Chem. B 2010, 114, 915-920. 35. S. Y. Reece and D. G. Nocera, Annu. Rev. Biochem 2009, 78, 673-699. 36. Z. Rappoport, The chemistry of phenols. Wiley: Chichester, England ; Hoboken, NJ, 2003. 192 37. G. Granucci, J. T. Hynes, P. Millie and T. H. Tran-Thi, J. Am. Chem. Soc. 2000, 122, 12243-12253. 38. E. Pines, D. Huppert and N. Agmon, J. Chem. Phys. 1988, 88, 5620-5630. 39. C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys. 2009, 130. 40. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum. 2003, 74, 4958-4960. 41. R. J. Abraham and M. Mobli, Magn. Reson. Chem. 2007, 45, 865-877. 42. M. Ito, J. Mol. Spectrosc. 1960, 4, 125-143. 43. M. Ito, J. Mol. Spectrosc. 1960, 4, 106-124. 44. I. B. Berlman, Handbook of Fluorescence Spectra of Aromatic Molecules. Academic Press: New York, 1965. 45. D. L. Gerrard and W. F. Maddams, Spectrochim. Acta, Part A 1978, 34, 1205-1211. 46. D. L. Gerrard and W. F. Maddams, Spectrochim. Acta, Part A 1978, 34, 1219-1223. 47. G. Kohler and N. Getoff, J. Chem. Soc., Faraday Trans. 1976, 72, 2101-2107. 48. T. Shida and Y. Takemura, Radiat. Phys. Chem. 1983, 21, 157-166. 49. S. Tagawa, N. Hayashi, Y. Yoshida, M. Washio and Y. Tabata, Radiat. Phys. Chem. 1989, 34, 503-511. 50. L. D. A. Siebbeles, U. Emmerichs, A. Hummel and H. J. Bakker, J. Chem. Phys. 1997, 107, 9339-9347. 51. R. Hermann, G. R. Mahalaxmi, T. Jochum, S. Naumov and O. Brede, J. Phys. Chem. A 2002, 106, 2379-2389. 52. D. V. Bent and E. Hayon, J. Am. Chem. Soc. 1975, 97, 2599-2606. 53. D. Pullin and L. Andrews, J. Mol. Struct. 1982, 95, 181-185. 54. X. Y. Chen, D. S. Larsen, S. E. Bradforth and I. H. M. van Stokkum, J. Phys. Chem. A 2011, 115, 3807-3819. 55. J. C. Mialocq, J. Sutton and P. Goujon, J. Chem. Phys. 1980, 72, 6338-6345. 193 56. J. Zechner, G. Kohler, G. Grabner and N. Getoff, Chem. Phys. Lett. 1976, 37, 297- 300. 57. K. N. Jha, G. L. Bolton and G. R. Freeman, J. Phys. Chem. 1972, 76, 3876-&. 58. X. Y. Chen, R. Seidel, B. Winter, P. Slavicek and S. E. Bradforth, in preparation. 59. J.-P. Jay-Gerin, Can. J. Chem. 1997, 75, 1310 - 1314. 60. K. Kesper, F. Diehl, J. G. G. Simon, H. Specht and A. Schweig, Chem. Phys. 1991, 153, 511-517. 61. O. Brede, H. Orthner, V. Zubarev and R. Hermann, J. Phys. Chem. 1996, 100, 7097-7105. 62. M. R. Ganapathi, R. Hermann, S. Naumov and O. Brede, Phys. Chem. Chem. Phys. 2000, 2, 4947-4955. 63. R. Hermann, S. Naumov, G. R. Mahalaxmi and O. Brede, Chem. Phys. Lett. 2000, 324, 265-272. 64. C. G. Elles, A. E. Jailaubekov, R. A. Crowell and S. E. Bradforth, J. Chem. Phys. 2006, 125. 65. C. G. Elles, I. A. Shkrob, R. A. Crowell and S. E. Bradforth, J. Chem. Phys. 2007, 126. 66. A. Sur and P. M. Johnson, J. Chem. Phys. 1986, 84, 1206-1209. 67. R. J. Lipert, G. Bermudez and S. D. Colson, J. Phys. Chem. 1988, 92, 3801-3805. 68. G. Grabner, G. Kohler, G. Marconi, S. Monti and E. Venuti, J. Phys. Chem. 1990, 94, 3609-3613. 69. J. Zechner, G. Kohler, G. Grabner and N. Getoff, Canadian Journal of Chemistry- Revue Canadienne De Chimie 1980, 58, 2006-2010. 70. G. Kohler and N. Getoff, Chem. Phys. Lett. 1974, 26, 525-528. 71. N. Getoff, Radiat. Phys. Chem. 1989, 34, 711-719. 72. A. L. Sobolewski, W. Domcke, C. Dedonder-Lardeux and C. Jouvet, Phys. Chem. Chem. Phys. 2002, 4, 1093-1100. 73. E. L. Wehry and L. B. Rogers, J. Am. Chem. Soc. 1965, 87, 4234-&. 194 74. E. Pines and G. R. Fleming, Chem. Phys. 1994, 183, 393-402. 75. C. A. Hasselbacher, E. Waxman, L. T. Galati, P. B. Contino, J. B. A. Ross and W. R. Laws, J. Phys. Chem. 1991, 95, 2995-3005. 76. W. R. Laws, J. B. A. Ross, H. R. Wyssbrod, J. M. Beechem, L. Brand and J. C. Sutherland, Biochemistry 1986, 25, 599-607. 77. S. Dellonte, G. Marconi and S. Monti, Journal of Photochemistry 1987, 39, 33-40. 78. E. J. Heilweil, M. P. Casassa, R. R. Cavanagh and J. C. Stephenson, J. Chem. Phys. 1986, 85, 5004-5018. 79. G. M. Roberts, A. S. Chatterley, J. D. Young and V. G. Stavros, J. Phys. Chem. Lett. 2012, 3, 348-352. 80. H. Shiraishi, G. R. Sunaryo and K. Ishigure, J. Phys. Chem. 1994, 98, 5164-5173. 81. R. A. Crowell and D. M. Bartels, J. Phys. Chem. 1996, 100, 17940-17949. 82. A. M. Lossack, E. Roduner and D. M. Bartels, J. Phys. Chem. A 1998, 102, 7462- 7469. 83. C. Anastasi, V. Simpson, J. Munk and P. Pagsberg, Chem. Phys. Lett. 1989, 164, 18-22. 84. R. H. Schuler and L. K. Patterson, Chem. Phys. Lett. 1974, 27, 369-372. 85. R. P. Bell, The proton in chemistry. 2nd ed.; Cornell University Press: Ithaca, N.Y., 1973. 86. V. Stojkovic and A. Kohen, Isr. J. Chem. 2009, 49, 163-173. 87. S. Hammes-Schiffer and A. V. Soudackov, J. Phys. Chem. B 2008, 112, 14108- 14123. 88. S. Hammes-Schiffer, Acc. Chem. Res. 2009, 42, 1881-1889. 89. I. Navrotskaya and S. Hammes-Schiffer, J. Chem. Phys. 2009, 131. 90. Y. Georgievskii and A. A. Stuchebrukhov, J. Chem. Phys. 2000, 113, 10438-10450. 91. C. Costentin, M. Robert and J. M. Saveant, J. Am. Chem. Soc. 2007, 129, 9953- 9963. 195 92. C. Costentin, M. Robert and J. M. Saveant, Phys. Chem. Chem. Phys. 2010, 12, 11179-11190. 93. K. S. Peters and A. Cashin, J. Phys. Chem. A 2000, 104, 4833-4838. 94. K. S. Peters, A. Cashin and P. Timbers, J. Am. Chem. Soc. 2000, 122, 107-113. 95. K. S. Peters and G. Kim, J. Phys. Chem. A 2001, 105, 4177-4181. 96. K. S. Peters and G. Kim, J. Phys. Org. Chem. 2005, 18, 1-8. 97. M. Negrerie, F. Gai, S. M. Bellefeuille and J. W. Petrich, J. Phys. Chem. 1991, 95, 8663-8670. 98. P. T. Chou, W. S. Yu, C. Y. Wei, Y. M. Cheng and C. Y. Yang, J. Am. Chem. Soc. 2001, 123, 3599-3600. 99. C. C. Hsieh, K. Y. Chen, W. T. Hsieh, C. H. Lai, J. Y. Shen, C. M. Jiang, H. S. Duan and P. T. Chou, Chemphyschem 2008, 9, 2221-2229. 100. C. C. Hsieh, C. M. Jiang and P. T. Chou, Acc. Chem. Res. 2010, 43, 1364-1374. 196 Chapter 6. Photochemistry of H-bonded Phenol Dimers in the Condensed Phase 1 6.1 Introduction Phenol molecules tend to “self-associate” to form dimeric or even polymeric H-bonded complexes in non-polar organic solvents. The effect of H-bonding of two phenol molecules on reaction dynamics was extensively discussed in the gas phase clusters studied by Sur et al., 2 and Lipert et al, 3 both studies show that the excited state S 1 lifetime is significantly increased upon dimerization. Such observation was rationalized by invoking that the decreasing in O–H stretch frequency made it a less favorable acceptor mode for internal conversion (IC), and thereby increase the IC rate. We will explore the H-bonding effect (between two phenol molecules) on a different reaction pathway – photodissociation. In the previous chapter, we have discussed the O–H bond fission for a phenol monomer in cyclohexane. This weakly interacting solvent provides an environment where hydrogen-bonded solute dimers can be studied and we provide a first exploration of this topic by carrying out photolysis experiments on phenol clusters. 6.2 Experimental The transient absorption experiments and materials used were described extensively in Chapter 5 and thus will not be repeated here. The phenol concentrations used in current experiments to investigate H-bonded dimers were 45 and 90 mM. It is important to quantify the fraction of monomer at the concentration range in which the TA experiment is carried out. UV-Visible spectra of phenol solution with various concentrations can 197 provide some information, but IR spectra monitoring the O–H stretch mode provide a direct observation. The absorption in the IR range (1000 < ν < 4000 cm -1 ) of phenol in cyclohexane solution was determined by a Vertex 80v FT-IR spectrometer (Bruker). The spectra for various phenol concentrations (5.0, 9.7, 20, 49, 66, 96 mM) were taken in a standard IR cell with 1 mm path length and BaF 2 windows (International Crystal Laboratories). The instrument resolution was 1 cm -1 and the spectra were averaged over 16 scans. Both the optical bench and the sample compartment were evacuated to < 300 Pa in order to avoid absorption from carbon dioxide and water vapor. The absorption of the BaF 2 and solvent cyclohexane were measured under the same experimental parameters and subsequently subtracted from the spectra of the above phenol solutions. 6.3 Results 6.3.1 IR Analysis of O–H stretch in Phenol The IR spectra of phenol in cyclohexane at various concentrations are shown in Figure 6.1. There are several features whose intensities scale as a function of phenol concentration and thus indicating that they are phenol vibrational modes. Most low frequency modes (< 1500 cm -1 ) have O–H bend character. The two intense peaks at ~1500 and 1600 cm -1 are the ring deformation modes. The peaks with modest intensity at 3000 – 3100 cm -1 are signatures of C–H stretch from the phenol ring. Most importantly, the broad peaks centered at ~3340 and 3500 cm -1 , and the sharp peak at 3617 cm -1 originates from O–H stretching. These vibrational mode assignments are in full agreement to the theoretical, gas phase and CCl 4 solution studies. 4 198 Figure 6.1. IR absorption spectra of phenol in cyclohexane at various concentrations. The cyclohexane bands at ~1300, ~1400, 2500 – 3000 cm -1 were removed from the spectra for ease of visualization. Much literature has discussed the origin of the three O–H stretch bands, and the consensus is that, in analogy to aliphatic alcohols in non-polar organic solvents, 5-8 the sharp peak at the highest frequency (3617 cm -1 ) originates from free O–H stretching, the relatively broader band centered at ~ 3500 cm -1 is due to the O–H stretch from the phenol being a proton donor in a H-bonded complex, and the very broad structured centered at 3340 cm -1 is caused by the OH group being simultaneously a proton donor and acceptor, which is usually seen in the H-bonded polymeric long chain. 4,9-11 Although there is some debate on distinguishing the free O–H stretch of the monomer from the dimer, and the structure of the phenol dimer (see Figure 6.2), 9,10 the appearance of the dimeric and polymeric bands provides direct evidence of self association, and the fraction of free OH 199 groups at different concentrations can be deduced from the absorption strength of the 3617 cm -1 band. † Figure 6.2. Different structures of dimeric and polymeric phenol complexes and the corresponding IR signatures. Figure 6.3 (a) and (b) show the O–H stretch region on an expanded scale, from which we can see that at very dilute phenol concentration, i.e., 5 and 10 mM, the dimeric and polymeric bands are not yet developed (the negative absorption is due to the imperfect subtraction of the cyclohexane background signal), indicating that the phenol molecules do not self-associate at this concentration regime. The dimeric band at ~3500 cm -1 † There is a subtle effect of relating the number of free O-H to the number of monomers in the solution. As discussed above, the O-H stretch from the terminal of the open chain dimer and polymer may not be distinguishable from the O-H stretch of the true monomer. Therefore, the IR spectra give information on the number of free OH groups, which is the upper limit of phenol monomer. However, the fact that they have similar spectroscopic properties means that they have similar electronic structures and thus are likely to exhibit similar photochemical behaviors. 200 becomes more pronounced at 20 mM solution but no polymeric band is visible. For higher phenol concentrations (49, 66, 96 mM), all three bands are visible, which indicates the coexistence of free molecule, dimer and polymeric long chain. Figure 6.3. (a) IR absorption spectra of phenol in cyclohexane, expanded to the O-H stretch region. Inset: expanded to show the H-bonded O–H stretch. (b) Spectra normalized to the 3617 cm -1 peak to emphasize the change in the shape of spectra as a function of concentration. (c) absorption peak intensities as a function of phenol concentration: ring deformation (blue), O-H stretch (black), O–H bend (red). (d) monomer fraction estimated from the free O–H stretch extinction coefficient (calculated from the 5 and 10 mM solution, assuming all phenol molecules exist in monomer form). To quantify the fraction of free OH in cyclohexane solution, the extinction coefficient of the free O–H stretch is estimated to 270 M -1 cm -1 , which is calculated from the optical 201 density of the 5 and 10 mM solution (average of the two). Note that this number is five times higher than that estimated from HOD and ten times higher than water in n-octane (although in this case there are both symmetric and asymmetric stretch), 12,13 but it is very close to the extinction coefficient of O–H stretch measured from 4-chlorophenol in carbon tetrachloride solution (220 M -1 cm -1 ). 14 Figure 6.3 (c) and (d) present the deviation of O–H stretch optical density from the “expected value” calculated from the extinction coefficient. Figure 6.3 (c) shows the strongest feature in the IR spectrum, the ring deformation mode, has optical density which scales linearly as the phenol concentration (blue circle and solid line). This shows the upper limit of the dynamic range of the instrument is at least O.D. 2.6 and thus firmly establishes that the deviation from linearity of the O–H modes is solely due to the H-bonded complex formation. It is also shown that the O–H bending mode at 1174.6 cm -1 exhibits similar behavior as O–H stretch at 3617.3 cm -1 . Figure 6.3 (d) presents the calculated monomer fraction at our concentration range (blue solid line). Our estimate is fully in accordance with the ultraviolet absorption 15,16 and other IR experiments 17 in carbon tetrachloride. At the highest concentration employed in the TA experiments (90 mM), it is estimated the solution is 70% monomers. 202 6.3.2 Transient Absorption Experiments Figure 6.4 (reproduced from Figure 5.6 in Chapter 5 for ease of comparison) (a) TA spectra measured at t < 5 ps following 267 nm photoexcitation of a 90 mM phenol in cyclohexane solution, with the polarization of the super continuum probe pulse set to magic angle (54.7˚) with respect to that of the excitation pulse. (b) Kinetics of selected wavelengths obtained from the TA experiment at t < 5 ps. Inset: t < 900 ps. (c) TA spectra measured at delay times in the range 100 ps to 14 ns following 267 nm photexcitation of 90 mM phenol in cyclohexane. Inset: phenoxyl radical region displayed on an expanded scale. (d) Kinetics of selected wavelengths obtained from the TA experiment (lines). Inset: Comparison of the kinetics at 475 nm and the fluorescence lifetime obtained from the TCSPC experiment (dots) Figure 6.4 presents the result of femtosecond TA experiments following 267 nm excitation of a 90 mM PhOH solution. The transient spectra observed at early delay times (t < 2 ps) display three distinct features at 380, 475 and 600 nm, with approximately equal intensities, reminiscent to those observed at low concentrations where clustering is 203 relatively much less important. Further, the kinetics of the λ = 475 nm feature is again in good accord with the 2.1 ns fluorescence lifetime. Because these are identical to those observed in 10 mM phenol solutions, and are consistent with the computed ESA spectrum for the gas phase monomer, we assign these ESA signatures to monomer (or ‘monomer-like’) S 1 excited phenol. However, the 600 nm peak behaves differently in these high concentration phenol solutions, growing strongly at later times. The original ‘monomer-like’ 600 nm ESA peak is clearly supplemented by a new transient, such that it ends up with a different set of associated kinetics cf. the monomer kinetics at 380 and 475 nm after t > 2 ps (see Figure 6.4 (b) and (c)). The appearance of a new transient, at concentrations where ground state phenol clustering takes off, hints that the new feature may be attributable to phenol excimers. In fact, this broad feature is very similar to the known spectrum of the benzene excimer, 18 where there is an excited state attractive interaction between two stacked π systems. We have also reproduced the literature benzene excimer spectrum under our experimental condition (see Figure 6.5). For phenol, however, the ~ 100 ps appearance time is much faster than the diffusion limit (e.g., benzene excimer formation in cyclohexane has a diffusion-limited rate of (2 ± 1) × 10 10 M -1 s -1 18 , this value is in good accord with our results for benzene solutions as well). Phenol is different from benzene in that we have seen that it can form a stable dimer in the ground state, which already places two or more aromatic rings primed in close proximity, and the rise time seen here most likely represents the conformational rearrangement of the cluster instead of a diffusional encounter time. Thus, the term excimer in here is used more loosely than intended by Förster’s original definition, 204 wherein aromatic dimers are dissociative in the ground state but associative in the excited state due to a strong π-π interaction. 19 Figure 6.5 (a) TA spectra at selected delay times for 4.1 M benzene in cyclohexane, excited with 267 nm and probed with visible continuum. (b) Kinetics of the 510 nm feature for various concentrations of benzene in cyclohexane solutions and pure benzene, obtained with the same experimental condition as (a). The circles are experimental data and the lines are single exponential rise with the following time constant: 14 ps (pure benzene), 25 ps (4.1 M), 30 ps (3.2 M) and 41 ps (2.1 M). The average rate constant for the 510 nm rise obtained from these measurements is 1 × 10 10 M -1 s -1 . A second piece of information supporting the conformational rearrangement instead of diffusive encountering is the concentration dependence (or otherwise) of the rise time of the excimer feature. Figure 6.6 (a) shows that the rise time of the 600 nm feature is identical in 45 and 90 mM phenol in cyclohexane solutions. The rise time for a diffusional process, such as the excimer formation of benzene in cyclohexane solutions, would exhibit concentration dependence. Specifically, the rise time of the 45 mM solution would be slower than that of the 90 mM. Such concentration dependence is not observed in our cyclohexane experiments. More interestingly, the excimer rise time in 205 water is somewhat dependent on the phenol concentration (both are slower than the cyclohexane solutions), but a more dilute solution exhibit a faster rise time – an observation completely contradicts with any expectation for a diffusional process. We can go some way to reconcile this, but one possible explanation is the formation of phenol polymeric chain, which gives rise to the broad peak at ~3340 cm -1 in the FT-IR spectrum (see Figure 6.2 for structure). In cyclohexane, the concentration of this long chain structure starts to take off at ~50 mM (Figure 6.3). If the formation of the polymeric chain is somehow encouraged in water, the longer rise time at higher concentration could due to the rearrangement of multiple benzene rings, instead of only two in the case of dimer, to achieve the optimal configuration of excimer. Figure 6.6 The kinetics of the phenol excimer band at λ probe = 600 nm, after 267 nm excitation of (a) cyclohexane solutions and (b) water solutions. All traces are normalized to facilitate comparison. The rise time is also obtained from fitting the traces to a single exponential function (actual fitting not shown). 206 From above we can clearly see that the phenol excimer is formed in cyclohexane solution, so does it preserve the reaction dynamics of a phenol monomer? We can start by considering what happens when phenol is excited well below the S 1 /S 2 CI. Comparing Figure 6.4 (a) and (c), we conclude that there is no appreciable amount of phenoxyl radical generated within 1 ns, but that a clearly measurable amount is produced on a longer (ns) time scale. These findings are qualitatively consistent with the observations in the 10 mM PhOH solution, but since there is a large error associated with the PhO radical yield measurement it is difficult to determine whether the radical is generated exclusively from the 70% monomer, or if it contains some contribution from the self associated form as well. We are now poised to explore whether for phenol molecules tied up in clusters, prompt dissociation induced by exciting above the S 1 /S 2 CI will turn off dimer conformational rearrangement – thereby eliminating the spectral signature of excimers. Such experiments, using 200 nm excitation, are presented in Figure 6.6. We note that Figure 5.14 in Chapter 5, a “control” experiment for this section which should involve negligible clustering, shows no appreciable signal attributable to PhOH S 1 ESA at longer delay times, the signature for which is displayed in Figure 6.7. Although we cannot quantify yields accurately, this observation suggests that most monomeric phenol dissociates at 200 nm, or at least that very little undergoes internal conversion to get trapped on S 1 . 207 Figure 6.7 (a) TA spectra measured at selected time delays following 200 nm photolysis of 45 mM phenol in cyclohexane, with the super continuum probe pulse polarisation aligned parallel to that of the excitation pulse. The thin lines are the TA signal for pure cyclohexane at 500 fs and 800 ps (same colour code as the solution) (b) Kinetics at selected wavelengths. Due to the large extinction coefficient of phenol at 200 nm, 45 mM phenol solution was used instead of 90 mM to maintain reasonable optical density samples. This diminishes the cluster fraction to ~15%. For the isolated phenol molecules that make up the majority of the solution, vibrationally excited radicals are formed within the instrument response time just as in Figure 5.14 in Chapter 5, indicating direct O–H bond fission as before. Vibrational cooling occurs on a similar timescale to that observed in the low concentration experiment. However, the peak we have assigned to excimer absorption is still observed, suggesting that a significant fraction of the phenol clusters present do not directly dissociate upon 200 nm excitation. Interestingly, an underlying absorption spreading across the whole spectral window is more pronounced compared to the 10 mM solution (cf. Figure 5.14 in Chapter 5), probably due to a lesser relative contribution from the solvent TA signal. Single wavelength analysis of 475 nm, where the monomer S 1 208 ESA exhibits a weak peak (cf. Figure 5.4 in Chapter 5) but is otherwise free of contributions from PhO radical and excimer absorption, shows kinetics very similar to that of the 10 mM solution at 267 nm excitation – a single exponential decay which can be best described by a time constant of 2.1 ns. This could imply that those phenol clusters that are unable to dissociate undergo radiationless decay to S 1 , where the monomer and excimer forms are in equilibrium with the dimer exploring conformational space. 6.4 Discussion In phenol dimer, hydrogen-bonding is the dominant ground state interaction. Upon excitation to the S 1 state, however, π-π interactions becomes much stronger, and consequentially the aromatic rings move into closer proximity with each other and become aligned in a pseudo-parallel configuration. 20 The precise geometry of the excimer is determined by competing H-bonding and π-stacking interactions, as previously illustrated by rotationally resolved electronic spectroscopy. 21-23 The rise in the 600 nm kinetics in our 267 nm data is attributed to such conformational changes of dimeric and polymeric phenol upon excitation. It is noteworthy that the fluorescence of phenol excimer, which is red shifted to ~345 nm as compared to the monomer fluorescence peak at ~300 nm, is only observed at much higher concentration, e.g., molten phenol. 24 This suggests that TA experiments are much more sensitive than fluorescence experiments in detecting excimers – an observation entirely consistent with benzene excimers. 25 209 The effect of H-bonding on the 1 1 πσ* state and potential photodissociation has not been demonstrated in previous condensed phase studies. Intuitively, the collinear O–H···O geometry for the H-bonded phenol dimer should significantly change the electronic structure of the 1 1 πσ* state at larger R O–H distances− becoming bound rather than repulsive. Such a situation has been described by Chipman for the excited state of the water dimer. 26 Consistent with this intuition, considering the observation of the excimer band at 600 nm, the spectra of 45 mM phenol obtained at λ pump = 200 nm provide tentative evidence that at least some of the clusters do not dissociate upon photoexcitation. This 600 nm band would not appear if all photoexcited phenol molecules dissociated. Although there have been numerous gas phase phenol dimer spectroscopy studies, the majority have focused on the S 1 state. The key observation is that the S 1 state of the phenol dimer has a notably extended lifetime (16 ns, cf. 2.1 ns for phenol monomer). 2 This has been ascribed to elimination of the otherwise important IC channel back to S 0 as the H-bonding between the two phenol molecules lowers the vibrational frequency of the O–H stretch, and in turn makes the latter a less efficient acceptor mode. 2 Such an argument implies that the electronic excitation is localized on the H-bond donor phenol which is consistent with theoretical calculations on the dimer. 22 The excited state lifetime is similarly extended in gas phase phenol/water dimers, 2,3,27 where the phenol O−H bond also acts as the H-bond donor. 27 However, the possible effects of clustering on the tunneling dissociation under the S 1 /S 2 CI on the excited state lifetime were not considered, and this process should be strongly affected. In fact, Brause et al. suggest that in the 210 dimer, the donor O−H bond length is shorter in S 1 compared to monomeric phenol 22 which would result in making the tunneling barrier higher and wider (see Figure 5.1 in Chapter 5). As mentioned earlier, the fact that the 1 1 πσ* surface may become bounded at longer R O−H will further impact the likelihood of dissociation. 26 Combined, this suggests that tunneling dissociation will be a less competitive non-radiative relaxation pathway in the dimer in comparison to the monomer. In our experimental TA data, it is difficult to quantitate a reduced phenoxyl yield because of the coexistence of monomer and clustered species in the solutions; the latter being a minor component. However, it is clear that the excimer absorption band at 600 nm has a lifetime longer than 2.1 ns (see Figure 6.3). Additionally, the monomer S 1 absorption at 475 nm can be fitted with a bi-exponential decay function – at t < 4 ns, the decay is in excellent agreement with the single exponential exhibited by low concentration phenol (τ = 2.1 ns), but deviates from a mono-exponential decay at longer time delays. Given that the π stacked excimer is in conformational equilibrium with an open excited state dimer structure that has little π stacking, the observation of a longer tail to the “monomer” ESA band could then also be explained. These preliminary experiments on the photochemistry of hydrogen bonded cluster species within inert environments like cyclohexane demonstrate that this approach could provide a rich new avenue of research in comparing reaction dynamics in the two phases. 211 6.5 Conclusion The use of cyclohexane as medium makes possible the measurement of excited state reaction dynamics of hydrogen bonded clusters within a liquid environment that can be explicitly compared with gas phase clusters. Our current experiments give tantalizing evidence for the shutdown of the ballistic scission of the PhO–H bond on S 2 when the excited state phenol acts as hydrogen-bond donor. This is fertile ground for dynamical simulations and new experiments that aim to explore the full nature of the solvent influence in guiding chemical dynamics. 212 6.6 Chapter 6 References 1. Appeared as a section of: Y. Zhang, T. A. A. Oliver, M. N. R. Ashfold and S. E. Bradforth, Faraday Discuss. DOI:10.1039/C2FD20043K. 2. A. Sur and P. M. Johnson, J. Chem. Phys. 1986, 84, 1206-1209. 3. R. J. Lipert, G. Bermudez and S. D. Colson, J. Phys. Chem. 1988, 92, 3801-3805. 4. G. Keresztury, F. Billes, M. Kubinyi and T. Sundius, J. Phys. Chem. A 1998, 102, 1371-1380, and references therein. 5. E. E. Tucker, S. B. Farnham and S. D. Christian, J. Phys. Chem. 1969, 73, 3820- 3829. 6. E. E. Tucker and E. D. Becker, J. Phys. Chem. 1973, 77, 1783-1795. 7. E. E. Tucker, S. D. Christian and L. N. Lin, J. Phys. Chem. 1974, 78, 1443-1445. 8. E. E. Tucker and S. D. Christian, J. Phys. Chem. 1977, 81, 1295-1299. 9. M. M. Maguire and R. West, Spectrochim. Acta 1961, 17, 369-378. 10. D. L. Gerrard and W. F. Maddams, Spectrochim. Acta, Part A 1978, 34, 1205-1211. 11. R. J. Abraham and M. Mobli, Magn. Reson. Chem. 2007, 45, 865-877. 12. M. P. Conrad and H. L. Strauss, Biophys. J. 1985, 48, 117-124. 13. M. P. Conrad and H. L. Strauss, J. Phys. Chem. 1987, 91, 1668-1673. 14. D. Michalska, M. Rospenk, B. Czarnik-Matusewicz and T. Zeegers-Huyskens, J. Mol. Spectrosc. 2002, 212, 32-40. 15. M. Ito, J. Mol. Spectrosc. 1960, 4, 125-143. 16. M. Ito, J. Mol. Spectrosc. 1960, 4, 106-124. 17. R. Mecke, Discussions of the Faraday Society 1950, 161-177. 18. H. Miyasaka, H. Masuhara and N. Mataga, J. Phys. Chem. 1985, 89, 1631-1636. 19. T. Forster, Angew. Chem. Int. Ed. 1969, 8, 333-&. 20. A. Weichert, C. Riehn and B. Brutschy, J. Phys. Chem. A 2001, 105, 5679-5691. 213 21. M. Schmitt, M. Boehm, C. Ratzer, D. Kruegler, K. Kleinermanns, I. Kalkman, G. Berden and W. L. Meerts, Chemphyschem 2006, 7, 1241-1249. 22. R. Brause, M. Santa, M. Schmitt and K. Kleinermanns, Chemphyschem 2007, 8, 1394-1401. 23. M. Kolar and P. Hobza, J. Phys. Chem. A 2007, 111, 5851-5854. 24. S. S. Lehrer and G. D. Fasman, J. Am. Chem. Soc. 1965, 87, 4687-&. 25. K. Diri and A. I. Krylov, J. Phys. Chem. A 2012, 116, 653-662. 26. D. M. Chipman, J. Chem. Phys. 2006, 124. 27. G. Berden, W. L. Meerts, M. Schmitt and K. Kleinermanns, J. Chem. Phys. 1996, 104, 972-982. 214 Comprehensive Bibliography R. J. Abraham and M. Mobli, Magn. Reson. Chem. 2007, 45, 865-877. C. Anastasi, V. Simpson, J. Munk and P. Pagsberg, Chem. Phys. Lett. 1989, 164, 18-22. A. V. Anikeenko, A. V. Kim and N. N. Medvedev, J. Struct. Chem. 2010, 51, 1090-1096. M. N. R. Ashfold, B. Cronin, A. L. Devine, R. N. Dixon and M. G. D. Nix, Science 2006, 312, 1637-1640. M. N. R. Ashfold, A. L. Devine, R. N. Dixon, G. A. King, M. G. D. Nix and T. A. A. Oliver, Proc Nat Acad Sci USA 2008, 105, 12701-12706. M. N. R. Ashfold, A. L. Devine, M. G. D. Nix and R. N. Dixon, J. Phys. Chem. A 2008, 112, 9563-9574. M. N. R. Ashfold, G. A. King, D. Murdock, M. G. D. Nix, T. A. A. Oliver and A. G. Sage, Phys. Chem. Chem. Phys. 2010, 12, 1218-1238. U. Banin and S. Ruhman, J. Chem. Phys. 1993, 98, 4391-4403. M. Barbatti, M. Vazdar, A. J. A. Aquino, M. Eckert-Maksic and H. Lischka, J. Chem. Phys. 2006, 125, 164323. B. A. Barry, Photochem. Photobiol. 1993, 57, 179-188. D. M. Bartels and R. A. Crowell, J. Phys. Chem. A 2000, 104, 3349-3355. A. J. Bell, P. R. Pardon, C. G. Hickman and J. G. Frey, J. Chem. Soc. Faraday T. 1990, 86, 3831-3836. R. P. Bell, The proton in chemistry. 2nd ed.; Cornell University Press: Ithaca, N.Y., 1973. D. V. Bent and E. Hayon, J. Am. Chem. Soc. 1975, 97, 2599-2606. G. Berden, W. L. Meerts, M. Schmitt and K. Kleinermanns, J. Chem. Phys. 1996, 104, 972-982. I. B. Berlman, Handbook of Fluorescence Spectra of Aromatic Molecules. Academic Press: New York, 1965. J. D. Bhawalkar, G. S. He, C. K. Park, C. F. Zhao, G. Ruland and P. N. Prasad, Opt. Commun. 1996, 124, 33-37. R. J. Boerner and B. A. Barry, J. Biol. Chem. 1993, 268, 17151-17154. 215 J. Bonin and M. Robert, Photochem. Photobiol. 2011, 87, 1190-1203. R. Brause, M. Santa, M. Schmitt and K. Kleinermanns, Chemphyschem 2007, 8, 1394- 1401. O. Brede, H. Orthner, V. Zubarev and R. Hermann, J. Phys. Chem. 1996, 100, 7097-7105. S. B. Brown, E. A. Brown and I. Walker, Lancet Oncol. 2004, 5, 497-508. H. Bursing, J. Lindner, S. Hess and P. Vohringer, Applied Physics B-Lasers and Optics 2000, 71, 411-417. H. Bursing and P. Vohringer, Phys. Chem. Chem. Phys. 2000, 2, 73-82. CRC Handbook of Chemistry and Physics. 91st ed.; Taylor and Francis: 2009. X. Y. Chen, D. S. Larsen, S. E. Bradforth and I. H. M. van Stokkum, J. Phys. Chem. A 2011, 115, 3807-3819. B. M. Cheng, M. Bahou, W. C. Chen, C. H. Yui, Y. P. Lee and L. C. Lee, J. Chem. Phys. 2002, 117, 1633-1640. D. M. Chipman, J. Chem. Phys. 2006, 124, 044305 P. T. Chou, W. S. Yu, C. Y. Wei, Y. M. Cheng and C. Y. Yang, J. Am. Chem. Soc. 2001, 123, 3599-3600. H. A. Collins, M. Khurana, E. H. Moriyama, A. Mariampillai, E. Dahlstedt, M. Balaz, M. K. Kuimova, M. Drobizhev, V. X. D. Yang, D. Phillips, A. Rebane, B. C. Wilson and H. L. Anderson, Nat. Photonics 2008, 2, 420-424. M. P. Conrad and H. L. Strauss, J. Phys. Chem. 1987, 91, 1668-1673. M. P. Conrad and H. L. Strauss, Biophys. J. 1985, 48, 117-124. I. Conti, M. Garavelli and G. Orlandi, J. Am. Chem. Soc. 2009, 131, 16108-16118. C. Costentin, M. Robert and J. M. Saveant, J. Am. Chem. Soc. 2007, 129, 9953-9963. C. Costentin, M. Robert and J. M. Saveant, Phys. Chem. Chem. Phys. 2010, 12, 11179- 11190. C. E. Crespo-Hernandez, B. Cohen, P. M. Hare and B. Kohler, Chem. Rev. 2004, 104, 1977-2019 R. A. Crowell and D. M. Bartels, J. Phys. Chem. 1996, 100, 17940-17949. 216 R. I. Cukier and D. G. Nocera, Annu. Rev. Phys. Chem. 1998, 49, 337-369. T. N. Das, J. Phys. Chem. A 2005, 109, 3344-3351. L. De Boni, C. Toro and F. E. Hernandez, Opt. Lett. 2008, 33, 2958-2960. R. J. Debus, B. A. Barry, I. Sithole, G. T. Babcock and L. Mcintosh, Biochemistry 1988, 27, 9071-9074. S. Dellonte, G. Marconi and S. Monti, Journal of Photochemistry 1987, 39, 33-40. A. L. Devine, M. G. D. Nix, R. N. Dixon and M. N. R. Ashfold, J. Phys. Chem. A 2008, 112, 9563-9574. B. A. Diner, D. A. Force, D. W. Randall and R. D. Britt, Biochemistry 1998, 37, 17931- 17943. K. Diri and A. I. Krylov, J. Phys. Chem. A 2012, 116, 653-662. R. N. Dixon, T. A. A. Oliver and M. N. R. Ashfold, J. Chem. Phys. 2011, 134, 194303 A. L. Dobryakov, S. A. Kovalenko and N. P. Ernsting, J. Chem. Phys. 2003, 119, 988- 1002. A. Dragomir, J. G. McInerney and D. N. Nikogosyan, Appl. Opt. 2002, 41, 4365-4376. A. Dragomir, J. G. McInerney, D. N. Nikogosyan and P. G. Kazansky, Appl. Phys. Lett. 2002, 80, 1114-1116. A. Dragomir, J. G. McInerney, D. N. Nikogosyan and A. A. Ruth, IEEE J. Quantum Electron. 2002, 38, 31-36. M. Drobizhev, A. Karotki, M. Kruk and A. Rebane, Chem. Phys. Lett. 2002, 355, 175- 182. M. Drobizhev, A. Karotki, M. Kruk, A. Rebane, E. Nickel, C. W. Spangler and N. Mamardashvili, P Soc Photo-Opt Ins 2003, 4797, 152-162. C. G. Durfee, S. Backus, M. M. Murnane and H. C. Kapteyn, Opt. Lett. 1997, 22, 1565- 1567. J. E. Ehrlich, X. L. Wu, I. Y. S. Lee, Z. Y. Hu, H. Rockel, S. R. Marder and J. W. Perry, Opt. Lett. 1997, 22, 1843-1845. P. Z. El-Khoury and A. N. Tarnovsky, Chem. Phys. Lett. 2008, 453, 160-166. 217 C. G. Elles, M. J. Cox, G. L. Barnes and F. F. Crim, J. Phys. Chem. A 2004, 108, 10973- 10979. C. G. Elles, M. J. Cox and F. F. Crim, J. Chem. Phys. 2004, 120, 6973-6979. C. G. Elles and F. F. Crim, Annu. Rev. Phys. Chem. 2006, 57, 273-302. C. G. Elles, A. E. Jailaubekov, R. A. Crowell and S. E. Bradforth, J. Chem. Phys. 2006, 125, 044515 C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys. 2009, 130, 084501. C. G. Elles, I. A. Shkrob, R. A. Crowell and S. E. Bradforth, J. Chem. Phys. 2007, 126. V. Engel, V. Staemmler, R. L. Vanderwal, F. F. Crim, R. J. Sension, B. Hudson, P. Andresen, S. Hennig, K. Weide and R. Schinke, J. Phys. Chem. 1992, 96, 3201-3213. H. Farman, J. C. Dore, M. C. Bellisentfunel and D. G. Montague, Mol. Phys. 1987, 61, 583-596. J. D. Faulk, R. C. Dunbar and C. Lifshitz, J. Am. Chem. Soc. 1990, 112, 7893-7899. T. Forster, Angew. Chem. Int. Ed. 1969, 8, 333-343. H. D. Gai and B. C. Garrett, J. Phys. Chem. 1994, 98, 9642-9648. M. R. Ganapathi, R. Hermann, S. Naumov and O. Brede, Phys. Chem. Chem. Phys. 2000, 2, 4947-4955. Y. Georgievskii and A. A. Stuchebrukhov, J. Chem. Phys. 2000, 113, 10438-10450. D. L. Gerrard and W. F. Maddams, Spectrochim. Acta, Part A 1978, 34, 1205-1211. D. L. Gerrard and W. F. Maddams, Spectrochim. Acta, Part A 1978, 34, 1219-1223. N. Getoff, Radiat. Phys. Chem. 1989, 34, 711-719. D. Ghosh, A. Roy, R. Seidel, B. Winter, S. E. Bradforth and A. I. Krylov, Submitted. S. H. Gong and A. Penzkofer, Opt. Quantum Electron. 1999, 31, 269-290. G. Grabner, G. Kohler, G. Marconi, S. Monti and E. Venuti, J. Phys. Chem. 1990, 94, 3609-3613. 218 G. Granucci, J. T. Hynes, P. Millie and T. H. Tran-Thi, J. Am. Chem. Soc. 2000, 122, 12243-12253. L. I. Grossweiner, G. W. Swenson and E. F. Zwicker, Science 1963, 141, 805. R. C. Guedes, K. Coutinho, B. J. C. Cabral and S. Canuto, J. Phys. Chem. B 2003, 107, 4304-4310. S. Hammes-Schiffer, Acc. Chem. Res. 2001, 34, 273-281. S. Hammes-Schiffer, Acc. Chem. Res. 2009, 42, 1881-1889. S. Hammes-Schiffer and A. V. Soudackov, J. Phys. Chem. B 2008, 112, 14108-14123. P. M. Hare, C. E. Crespo-Hernandez and B. Kohler, Proc Nat Acad Sci USA 2007, 104, 435-440. S. A. Harich, D. W. H. Hwang, X. F. Yang, J. J. Lin, X. M. Yang and R. N. Dixon, J. Chem. Phys. 2000, 113, 10073-10090. C. A. Hasselbacher, E. Waxman, L. T. Galati, P. B. Contino, J. B. A. Ross and W. R. Laws, J. Phys. Chem. 1991, 95, 2995-3005. G. S. He, J. D. Bhawalkar, C. F. Zhao and P. N. Prasad, Appl. Phys. Lett. 1995, 67, 2433- 2435. G. S. He, T. C. Lin and P. N. Prasad, Opt. Express 2002, 10, 566-574. G. S. He, G. C. Xu, P. N. Prasad, B. A. Reinhardt, J. C. Bhatt, R. McKellar and A. G. Dillard, Opt. Lett. 1995, 20, 1930-1930. B. A. Heath, N. A. Kuebler and M. B. Robin, J. Chem. Phys. 1979, 70, 3362-3368. E. J. Heilweil, M. P. Casassa, R. R. Cavanagh and J. C. Stephenson, J. Chem. Phys. 1986, 85, 5004-5018. J. M. Heller, R. N. Hamm, R. D. Birkhoff and L. R. Painter, J. Chem. Phys. 1974, 60, 3483-3486. F. Helmchen and W. Denk, Nature Methods 2005, 2, 932-940. R. Hermann, G. R. Dey, S. Naumov and O. Brede, Phys. Chem. Chem. Phys. 2000, 2, 1213-1220. R. Hermann, G. R. Mahalaxmi, T. Jochum, S. Naumov and O. Brede, J. Phys. Chem. A 2002, 106, 2379-2389. 219 R. Hermann, S. Naumov, G. R. Mahalaxmi and O. Brede, Chem. Phys. Lett. 2000, 324, 265-272. C. G. Hickman, A. Brickell and J. G. Frey, Chem. Phys. Lett. 1991, 185, 101-104. C. G. Hickman, N. Shaw, M. J. Crawford, A. J. Bell and J. G. Frey, J. Chem. Soc. Faraday T. 1993, 89, 1623-1630. A. P. Hitchcock and I. Ishii, J. Electron. Spectrosc. Relat. Phenom. 1987, 42, 11-26. A. P. Hitchcock, D. C. Newbury, I. Ishii, J. Stohr, J. A. Horsley, R. D. Redwing, A. L. Johnson and F. Sette, J. Chem. Phys. 1986, 85, 4849-4862. C. C. Hsieh, K. Y. Chen, W. T. Hsieh, C. H. Lai, J. Y. Shen, C. M. Jiang, H. S. Duan and P. T. Chou, Chemphyschem 2008, 9, 2221-2229. C. C. Hsieh, C. M. Jiang and P. T. Chou, Acc. Chem. Res. 2010, 43, 1364-1374. H. R. Hudock, B. G. Levine, A. L. Thompson, H. Satzger, D. Townsend, N. Gador, S. Ullrich, A. Stolow and T. J. Martinez, J. Phys. Chem. A 2007, 111, 8500-8508. T. Ichino. PhD thesis, University of Notre Dame, 2001. A. Iqbal, M. S. Y. Cheung, M. G. D. Nix and V. G. Stavros, J. Phys. Chem. A 2009, 113, 8157-8163. S. Ishiuchi, M. Saeki, M. Sakai and M. Fujii, Chem. Phys. Lett. 2000, 322, 27-32. S. Ishiuchi, M. Sakai, K. Daigoku, K. Hashimoto and M. Fujii, J. Chem. Phys. 2007, 127, 23404. M. Ito, J. Mol. Spectrosc. 1960, 4, 106-124. M. Ito, J. Mol. Spectrosc. 1960, 4, 125-143. A. E. Jailaubekov. PhD Thesis. University of Southern California, 2007. A. E. Jailaubekov and S. E. Bradforth, Appl. Phys. Lett. 2005, 87, 021107. J.-P. Jay-Gerin, Can. J. Chem. 1997, 75, 1310 - 1314. K. N. Jha, G. L. Bolton and G. R. Freeman, J. Phys. Chem. 1972, 76, 3876-3883. J. M. Jung, Chem. Phys. Lett. 2002, 366, 67-72. J. M. Jung and H. Gress, Chem. Phys. Lett. 2003, 377, 495-500. 220 J. M. Jung and H. Gress, Chem. Phys. Lett. 2002, 359, 153-157. A. Karotki, M. Drobizhev, M. Kruk, C. Spangler, E. Nickel, N. Mamardashvili and A. Rebane, J. Opt. Soc. Am. B: Opt. Phys. 2003, 20, 321-332. S. Kashtanov, A. Augustson, J. E. Rubensson, J. Nordgren, H. Agren, J. H. Guo and Y. Luo, Physical Review B 2005, 71. S. Kawata and Y. Kawata, Chem. Rev. 2000, 100, 1777-1788. G. Keresztury, F. Billes, M. Kubinyi and T. Sundius, J. Phys. Chem. A 1998, 102, 1371- 1380. K. Kesper, F. Diehl, J. G. G. Simon, H. Specht and A. Schweig, Chem. Phys. 1991, 153, 511-517. G. A. King, T. A. A. Oliver, M. G. D. Nix and M. N. R. Ashfold, J. Phys. Chem. A 2009, 113, 7984-7993. D. A. V. Kliner, J. C. Alfano and P. F. Barbara, J. Chem. Phys. 1993, 98, 5375-5389. J. A. Kloepfer, V. H. Vilchiz, V. A. Lenchenkov, A. C. Germaine and S. E. Bradforth, J. Chem. Phys. 2000, 113, 6288-6307. B. Kohler, J. Phys. Chem. Lett. 2010, 1, 2047-2053. G. Kohler and N. Getoff, Chem. Phys. Lett. 1974, 26, 525-528. G. Kohler and N. Getoff, J. Chem. Soc., Faraday Trans. 1976, 72, 2101-2107. B. Kok, B. Forbush and M. Mcgloin, Photochem. Photobiol. 1970, 11, 457-475. M. Kolar and P. Hobza, J. Phys. Chem. A 2007, 111, 5851-5854. S. A. Kovalenko, A. L. Dobryakov, J. Ruthmann and N. P. Ernsting, Phys. Rev. A: At. Mol. Opt. Phys. 1999, 59, 2369-2384. M. J. Krisch, L. R. McCunn, K. Takematsu, L. J. Butler, F. R. Blase and J. Shu, J. Phys. Chem. A 2004, 108, 1650-1656. E. B. Krissinel and N. Agmon, J. Comp. Chem. 1996, 17, 1085- 1098. Y. P. Kuo, H. C. Lu, Y. J. Wu, B. M. Cheng and J. F. Ogilvie, Chem. Phys. Lett. 2007, 447, 168-174. 221 C. N. LaFratta, J. T. Fourkas, T. Baldacchini and R. A. Farrer, Angew. Chem. Int. Ed. 2007, 46, 6238-6258. F. P. Larkins and A. J. Seen, Phys. Scr. 1990, 41, 827-832. J. Larsen, D. Madsen, J. A. Poulsen, T. D. Poulsen, S. R. Keiding and J. Thogersen, J. Chem. Phys. 2002, 116, 7997-8005. W. R. Laws, J. B. A. Ross, H. R. Wyssbrod, J. M. Beechem, L. Brand and J. C. Sutherland, Biochemistry 1986, 25, 599-607. S. S. Lehrer and G. D. Fasman, J. Am. Chem. Soc. 1965, 87, 4687-4691. S. S. Lehrer and G. D. Fasman, Biochemistry 1967, 6, 757-767. I. S. Lim, J. S. Lim, Y. S. Lee and S. K. Kim, J. Chem. Phys. 2007, 126, 034306-034310. C. Y. Lin and M. C. Lin, J. Phys. Chem. 1986, 90, 425-431. C. Y. Lin and M. C. Lin, Int. J. Chem. Kinet. 1985, 17, 1025-1028. N. Lin, Y. Luo, K. Ruud, X. Zhao, F. Santoro and A. Rizzo, Chemphyschem 2011, 12, 3392-3403. R. J. Lipert, G. Bermudez and S. D. Colson, J. Phys. Chem. 1988, 92, 3801-3805. R. J. Lipert and S. D. Colson, J. Chem. Phys. 1990, 92, 3240-3241. M. Lorenc, M. Ziolek, R. Naskrecki, J. Karolczak, J. Kubicki and A. Maciejewski, Appl. Phys. B 2002, 74, 19-27. A. M. Lossack, E. Roduner and D. M. Bartels, J. Phys. Chem. A 1998, 102, 7462-7469. A. Madsen, C. L. Thomsen, J. A. Poulsen, S. J. K. Jensen, J. Thogersen, S. R. Keiding and E. B. Krissinel, J. Phys. Chem. A 2003, 107, 3606-3611. M. M. Maguire and R. West, Spectrochim. Acta 1961, 17, 369-378. O. Marsalek, T. Frigato, J. VandeVondele, S. E. Bradforth, B. Schmidt, C. Schutte and P. Jungwirth, J. Phys. Chem. B 2010, 114, 915-920. W. M. McClain, Acc. Chem. Res. 1974, 7, 129-135. W. M. McClain, J. Chem. Phys. 1971, 55, 2789-2796. R. Mecke, Discussions of the Faraday Society 1950, 161-177. 222 T. J. Meyer, M. H. V. Huynh and H. H. Thorp, Angew. Chem. Int. Ed. 2007, 46, 5284- 5304. J. C. Mialocq, J. Sutton and P. Goujon, J. Chem. Phys. 1980, 72, 6338-6345. D. Michalska, M. Rospenk, B. Czarnik-Matusewicz and T. Zeegers-Huyskens, J. Mol. Spectrosc. 2002, 212, 32-40. C. T. Middleton, B. Cohen and B. Kohler, J. Phys. Chem. A 2007, 111, 10460-10467. H. Miyasaka, H. Masuhara and N. Mataga, J. Phys. Chem. 1985, 89, 1631-1636. P. R. Monson and W. M. McClain, J. Chem. Phys. 1972, 56, 4817-4825. P. R. Monson and W. M. McClain, J. Chem. Phys. 1970, 53, 29-37. A. C. Moskun and S. E. Bradforth, J. Chem. Phys. 2003, 119, 4500-4515. A. C. Moskun, A. E. Jailaubekov, S. E. Bradforth, G. H. Tao and R. M. Stratt, Science 2006, 311, 1907-1911. R. Mota, R. Parafita, A. Giuliani, M. J. Hubin-Franskin, J. M. C. Lourenco, G. Garcia, S. V. Hoffmann, N. J. Mason, P. A. Ribeiro, M. Raposo and P. Limao-Vieira, Chem. Phys. Lett. 2005, 416, 152-159. R. S. Mulliken, Chem. Phys. Lett. 1977, 46, 197-200. I. Nadler, D. Mahgerefteh, H. Reisler and C. Wittig, J. Chem. Phys. 1985, 82, 3885-3893. I. Navrotskaya and S. Hammes-Schiffer, J. Chem. Phys. 2009, 131. J. B. Nee, M. Suto and L. C. Lee, Chem. Phys. 1985, 98, 147-155. M. Negrerie, F. Gai, S. M. Bellefeuille and J. W. Petrich, J. Phys. Chem. 1991, 95, 8663- 8670. R. A. Negres, J. M. Hales, D. J. Hagan and E. W. Van Stryland, IEEE J. Quantum Electron. 2002, 38, 1205-1216. R. A. Negres, J. M. Hales, A. Kobyakov, D. J. Hagan and E. W. Van Stryland, Opt. Lett. 2002, 27, 270-272. D. N. Nikogosyan, A. A. Oraevsky and V. I. Rupasov, Chem. Phys. 1983, 77, 131-143. M. G. D. Nix, A. L. Devine, B. Cronin and M. N. R. Ashfold, Phys. Chem. Chem. Phys. 2006, 8, 2610-2618. 223 M. G. D. Nix, A. L. Devine, B. Cronin and M. N. R. Ashfold, J. Chem. Phys. 2007, 126. M. G. D. Nix, A. L. Devine, B. Cronin, R. N. Dixon and M. N. R. Ashfold, J. Chem. Phys. 2006, 125, 133318. S. L. Oliveira, D. S. Correa, L. Misoguti, C. J. L. Constantino, R. F. Aroca, S. C. Zilio and C. R. Mendonca, Adv. Mater. 2005, 17, 1890-1893. T. A. A. Oliver, G. A. King, D. P. Tew, R. N. Dixon and M. N. R. Ashfold, in preparation. T. A. A. Oliver, Y. Zhang, M. N. R. Ashfold and S. E. Bradforth, Faraday Discuss. 2011, 150, 439-458. D. A. Parthenopoulos and P. M. Rentzepis, Science 1989, 245, 843-845. J. M. L. Pecourt, J. Peon and B. Kohler, J. Am. Chem. Soc. 2001, 123, 10370-10378. J. Peon, G. C. Hess, J. M. L. Pecourt, T. Yuzawa and B. Kohler, J. Phys. Chem. A 1999, 103, 2460-2466. K. S. Peters and A. Cashin, J. Phys. Chem. A 2000, 104, 4833-4838. K. S. Peters, A. Cashin and P. Timbers, J. Am. Chem. Soc. 2000, 122, 107-113. K. S. Peters and G. Kim, J. Phys. Org. Chem. 2005, 18, 1-8. K. S. Peters and G. Kim, J. Phys. Chem. A 2001, 105, 4177-4181. C. Petersen, N. H. Dahl, S. K. Jensen, J. A. Poulsen, J. Thogersen and S. R. Keiding, J. Phys. Chem. A 2008, 112, 3339-3344. E. Pines and G. R. Fleming, Chem. Phys. 1994, 183, 393-402. E. Pines, D. Huppert and N. Agmon, J. Chem. Phys. 1988, 88, 5620-5630. G. A. Pino, A. N. Oldani, E. Marceca, M. Fujii, S. I. Ishiuchi, M. Miyazaki, M. Broquier, C. Dedonder and C. Jouvet, J. Chem. Phys. 2010, 133. S. Pommeret, F. Gobert, A. Mostafavi, I. Lampre and J. C. Mialocq, J. Phys. Chem. A 2001, 105, 11400-11406. N. Pugliano, S. Gnanakaran and R. M. Hochstrasser, Journal of Photochemistry and Photobiology a-Chemistry 1996, 102, 21-28. D. Pullin and L. Andrews, J. Mol. Struct. 1982, 95, 181-185. 224 J. W. Qian, D. J. Tannor, Y. Amatatsu and K. Morokuma, J. Chem. Phys. 1994, 101, 9597-9609. T. I. Quickenden and J. A. Irvin, J. Chem. Phys. 1980, 72, 4416-4428. Z. Rappoport, The chemistry of phenols. Wiley: Chichester, England ; Hoboken, NJ, 2003. M. Rasmusson, A. N. Tarnovsky, E. Akesson and V. Sundstrom, Chem. Phys. Lett. 2001, 335, 201-208. J. W. Raymonda, J. Chem. Phys. 1972, 56, 3912 A. Rebane, N. S. Makarov, M. Drobizhev, B. Spangler, E. S. Tarter, B. D. Reeves, C. W. Spangler, F. Q. Meng and Z. Y. Suo, J. Phys. Chem. C 2008, 112, 7997-8004. S. Y. Reece and D. G. Nocera, Annu. Rev. Biochem 2009, 78, 673-699. P. J. Reid, Acc. Chem. Res. 2001, 34, 691-698. H. Reisler and A. I. Krylov, Int. Rev. Phys. Chem. 2009, 28, 267-308. A. Reuther, A. Laubereau and D. N. Nikogosyan, Opt. Commun. 1997, 141, 180-184. A. Reuther, D. N. Nikogosyan and A. Laubereau, J. Phys. Chem. 1996, 100, 5570-5577. C. A. Rivera, S. E. Bradforth and G. Tempea, Opt. Express 2010, 18, 18615-18624. C. A. Rivera, N. Winter, R. V. Harper, I. Benjamin and S. E. Bradforth, Phys. Chem. Chem. Phys. 2011, 13, 8269-8283. Y. M. Riyad, S. Naumov, R. Hermann and O. Brede, Phys. Chem. Chem. Phys. 2006, 8, 1697-1706. G. M. Roberts, A. S. Chatterley, J. D. Young and V. G. Stavros, J. Phys. Chem. Lett. 2012, 3, 348-352. M. B. Robin, Higher excited states of polyatomic molecules. Academic Press: New York, 1974 A. G. Sage, M. G. D. Nix and M. N. R. Ashfold, Chem. Phys. 2008, 347, 300-308. D. R. Salahub and C. Sandorfy, Chem. Phys. Lett. 1971, 8, 71-74. E. Samoylova, H. Lippert, S. Ullrich, I. V. Hertel, W. Radloff and T. Schultz, J. Am. Chem. Soc. 2005, 127, 1782-1786. 225 R. Schinke, Photodissociation dynamics : spectroscopy and fragmentation of small polyatomic molecules. Cambridge University Press: Cambridge [England] ; New York, 1993. M. Schmitt, M. Boehm, C. Ratzer, D. Kruegler, K. Kleinermanns, I. Kalkman, G. Berden and W. L. Meerts, Chemphyschem 2006, 7, 1241-1249. M. Schneider, S. Barozzi, I. Testa, M. Faretta and A. Diaspro, Biophys. J. 2005, 89, 1346-1352. R. H. Schuler and L. K. Patterson, Chem. Phys. Lett. 1974, 27, 369-372. A. J. Seen and F. P. Larkins, Journal of Physics B-Atomic Molecular and Optical Physics 1992, 25, 4811-4822. L. Serrano-Andres and M. Merchan, Journal of Photochemistry and Photobiology C- Photochemistry Reviews 2009, 10, 21-32. T. K. Sham, B. X. Yang, J. Kirz and J. S. Tse, Physical Review A 1989, 40, 652-669. D. Sharma, J. Leonard and S. Haacke, Chem. Phys. Lett. 2010, 489, 99-102. T. Shida and Y. Takemura, Radiat. Phys. Chem. 1983, 21, 157-166. H. Shiraishi, G. R. Sunaryo and K. Ishigure, J. Phys. Chem. 1994, 98, 5164-5173. L. D. A. Siebbeles, U. Emmerichs, A. Hummel and H. J. Bakker, J. Chem. Phys. 1997, 107, 9339-9347. A. Sinicropi, W. M. Nau and M. Olivucci, Photochemical & Photobiological Sciences 2002, 1, 537-546. R. G. Snyder and Y. Kim, J. Phys. Chem. 1991, 95, 602-610. A. L. Sobolewski and W. Domcke, Chem. Phys. 2000, 259, 181-191. A. L. Sobolewski and W. Domcke, J. Phys. Chem. A 2001, 105, 9275-9283. A. L. Sobolewski, W. Domcke, C. Dedonder-Lardeux and C. Jouvet, Phys. Chem. Chem. Phys. 2002, 4, 1093-1100. C. W. Spangler, J. Mater. Chem. 1999, 9, 2013-2020. V. Stojkovic and A. Kohen, Isr. J. Chem. 2009, 49, 163-173. A. Sur and P. M. Johnson, J. Chem. Phys. 1986, 84, 1206-1209. 226 S. Tachibana, Y. Morisawa, A. Ikehata, H. Sato, N. Higashi and Y. Ozaki, Appl. Spectrosc. 2011, 65, 221-226. S. Tagawa, N. Hayashi, Y. Yoshida, M. Washio and Y. Tabata, Radiat. Phys. Chem. 1989, 34, 503-511. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum. 2003, 74, 4958-4960. C. L. Thomsen, D. Madsen, J. A. Poulsen, J. Thogersen, S. J. K. Jensen and S. R. Keiding, J. Chem. Phys. 2001, 115, 9361-9369. T. Tokushima, Y. Harada, O. Takahashi, Y. Senba, H. Ohashi, L. G. M. Pettersson, A. Nilsson and S. Shin, Chem. Phys. Lett. 2008, 460, 387-400. E. E. Tucker and E. D. Becker, J. Phys. Chem. 1973, 77, 1783-1795. E. E. Tucker and S. D. Christian, J. Phys. Chem. 1977, 81, 1295-1299. E. E. Tucker, S. D. Christian and L. N. Lin, J. Phys. Chem. 1974, 78, 1443-1445. E. E. Tucker, S. B. Farnham and S. D. Christian, J. Phys. Chem. 1969, 73, 3820-3829. UV Atlas of Organic Compounds. Plenum Press: New York, 1966. J. Underwood and C. Wittig, Chem. Phys. Lett. 2004, 386, 190-195. P. Vaterlein, R. Fink, E. Umbach and W. Wurth, J. Chem. Phys. 1998, 108, 3313-3320. G. Venturi, F. Formisano, G. J. Cuello, M. R. Johnson, E. Pellegrini, U. Bafile and E. Guarini, J. Chem. Phys. 2009, 131. E. Walker, A. Dvornikov, K. Coblentz and P. Rentzepis, Appl. Opt. 2008, 47, 4133-4139. D. Wanapun, R. D. Wampler, N. J. Begue and G. J. Simpson, Chem. Phys. Lett. 2008, 455, 6-12. E. L. Wehry and L. B. Rogers, J. Am. Chem. Soc. 1965, 87, 4234-4238. A. Weichert, C. Riehn and B. Brutschy, J. Phys. Chem. A 2001, 105, 5679-5691. K. R. Wilson, M. Cavalleri, B. S. Rude, R. D. Schaller, T. Catalano, A. Nilsson, R. J. Saykally and L. G. M. Pettersson, J. Phys. Chem. B 2005, 109, 10194-10203. B. Winter, R. Weber, W. Widdra, M. Dittmar, M. Faubel and I. V. Hertel, J. Phys. Chem. A 2004, 108, 2625-2632. 227 N. Winter and I. Benjamin, J. Chem. Phys. 2004, 121, 2253-2263. S. Yamaguchi and T. Tahara, Chem. Phys. Lett. 2004, 390, 136-139. S. Yamaguchi and T. Tahara, Chem. Phys. Lett. 2003, 376, 237-243. M. T. Zanni, B. J. Greenblatt, A. V. Davis and D. M. Neumark, J. Chem. Phys. 1999, 111, 2991-3003. J. Zechner, G. Kohler, G. Grabner and N. Getoff, Chem. Phys. Lett. 1976, 37, 297-300. J. Zechner, G. Kohler, G. Grabner and N. Getoff, Canadian Journal of Chemistry-Revue Canadienne De Chimie 1980, 58, 2006-2010. C. G. Zhan and D. A. Dixon, J. Phys. Chem. B 2003, 107, 4403-4417. Y. Zhang, T. A. A. Oliver, M. N. R. Ashfold and S. E. Bradforth, Faraday Discuss. DOI:10.1039/C2FD20043K. Y. Zhang, T. A. A. Oliver, M. N. R. Ashfold and S. E. Bradforth, In preparation. W. H. Zhou, S. M. Kuebler, K. L. Braun, T. Y. Yu, J. K. Cammack, C. K. Ober, J. W. Perry and S. R. Marder, Science 2002, 296, 1106-1109. 228 Appendix A. Optical Setup for in-situ Determination of Sample Path Length by Group Velocity Delay Method The optical setup is presented in Figure A.1 (a). The white light continuum was picked off by a flip mirror set in-between the parabolic mirrors (the original traveling path for the 2PA experiment is represented by the dotted curve), and redirected to a second sample. The choice of the second sample is not important – as long as it produces some transient species which absorbs within the probe range. In the case demonstrated in Figure A.1, fluorene was dissolved in ethanol and the kinetics of λ probe = 650 nm was selected by an interference filter (IF) and monitored on a photodiode array (PDA). From the delay of the signal appearance, one can calculate the path length of the liquid experienced by the UV beam. The procedure to measure the path length is as follows. The liquid sample of interest was placed in a l = 1 mm quartz cell. The delay caused by the liquid and the cell was compared to that by the blank cell (+ air). Taking advantage of the unknown pathlength of a commercially available quartz cell, the delay time caused by a unit length of the liquid was calculated for calibration purpose. Next, the delay caused by the same liquid in the gravity jet was compared to that by the air (which is effectively zero), and the thickness of the liquid film was subsequently calculated. The example shown in Figure A.1 (b) and (d) suggested that the thickness of the pure water film formed in the gravity jet is: 10 4 μm × (0.074 ps / 15.8 ps) = 47 μm. 229 The key step of this measurement is finding an area on the liquid film where the beam is distorted to the smallest possible extent after traveling through the sample, since the flow of the liquid determines the concave shape of the surface of the film, which acts as a negative cylindrical lens and elongates the UV beam in the horizontal direction. Therefore, the quality of the flowing film is important, and beams with smaller spot size (and thus experience a smaller curvature of the liquid surface) traveling through the center of the film can help producing a better beam profile after the sample. A poor quality liquid film produces poor spatial profile which can rapidly decrease the TA signal and lead to low signal-to-noise ratio. Figure A.1. (a) Optical setup for in situ measurement of the thickness of the liquid film formed in the gravity jet. (b) Calibration with a standard 10.000 ± 0.002 mm quartz cell. (c) 267 nm delayed by the liquid sample in the gravity jet. 230 Appendix B. Math Reference for Calculating 2PA Cross Section Below is a step-by-step derivation of the two-photon absorbance (ΔA 2PA , observable in our experiment), which can be related to 2PA coefficient (β) and 2PA absolute cross section (σ 2PA ). The probe (I pr ) attenuation at the presence of an intense pump (I p ), for a given position (x,y) of the probe at a given time (t), can be written as the follows (assuming Gaussian pulse in the spatial and time domain for both the pump and probe pules): ) , , ( ) ; , , ( ) , , ( t y x I t y x I z t y x I pr d p pr τ β − = ∂ ∂ where β is the 2PA coefficient (cm/W), τ d is the delay between the pump and probe pulses, and z is the length of travel from the front of the sample. The intensity of the pump and probe pulses in the spatial and time domain can be expressed as the follows. ( ) 2 2 2 2 2 2 ' , 0 2 2 2 exp ) , , ( pr pry prx pr pr t s y s x I t y x I τ − − ⋅ = ( ) 2 2 2 2 2 2 , 2 ) ( 2 2 exp ) ; , , ( p d py px p o d p t s y s x I t y x I τ τ τ − − − ⋅ = where s prx is the half spatial width of the Gaussian probe at e -1/2 maximum in the x- direction, s pry is in the y-direction, and likewise for the pump (s px and s py ), τ p and τ pr are the half widths of the Gaussian pulses at e -1/2 maximum in the time domain. ' , 0 pr I is the initial intensity of the probe, and p o I , is the intensity of the pump which is assumed to be constant as it is much more intense than the probe pulse. Therefore, 231 ( ) ( ) ( ) ( ) 2 2 2 2 2 2 ' , 0 2 2 2 2 2 2 , 2 2 2 2 2 2 ' , 0 2 2 2 exp 2 ) ( 2 2 exp 2 2 2 exp pr pry prx pr p d py px p o pr pry prx pr t s y s x I t s y s x I t s y s x I z τ τ τ β τ − − ⋅ ⋅ − − − ⋅ ⋅ − = − − ⋅ ∂ ∂ Since x and y are completely uncoupled from the traveling distance, time, and time delay, we can integrate both sides with respect to x and y first (detector detects the integration of the Gaussian pulse in the spatial domain, i.e., total energy). We can write: ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )( ) ( ) ( ) ( ) ( ) 2 2 ' , 0 2 2 , 0 2 2 2 2 2 2 ' , 0 2 2 ' , 0 2 2 , 0 2 2 2 2 2 2 ' , 0 2 2 ' , 0 2 2 , 0 2 2 2 2 2 2 2 2 2 2 ' , 0 2 2 2 2 2 exp 2 exp - 2 exp 2 exp 2 exp - 2 2 2 exp 2 2 2 exp 2 exp - 2 2 exp 2 2 exp 2 exp 2 2 exp pr pr p d p pry py prx px py px pr pr pr pr p d p pry py pry py prx px prx px pr pr pry prx pr pr p d p pry prx py px pr pr pry prx t I t I s s s s s s t I z t I t I s s s s s s s s t I z s s t I t I s y s x s y s x dy dx t I z s y s x dy dx τ τ τ β τ τ τ τ β π π τ π π τ τ τ β τ − ⋅ − − ⋅ ⋅ + + = − ⋅ ∂ ∂ ⇒ − ⋅ − − ⋅ ⋅ + ⋅ ⋅ + ⋅ = − ⋅ ∂ ∂ ⋅ ⋅ ⋅ ⋅ ⇒ − ⋅ − − ⋅ ⋅ − − = − ⋅ ∂ ∂         − ∫ ∫ ∫ ∫ ∞ + ∞ − ∞ + ∞ − +∞ ∞ − +∞ ∞ − Now we can integrate over t (convolute): ( ) ( ) ( )( ) ( ) ( ) ( ) ( ) ( ) ∫ ∫ ∞ + ∞ − +∞ ∞ − − ⋅ − − ⋅ ⋅ ⋅ + + = − ⋅ ∂ ∂ 2 2 2 2 , 0 ' , 0 , 0 2 2 2 2 2 2 ' , 0 2 exp 2 exp - - 2 exp pr p d p pr p pry py prx px py px pr pr t t I I I s s s s s s dt t I z τ τ τ β β τ 232 where the integral on the right hand side represents the convolution of the pump and probe. Since ( ) 2 2 2 exp pr t τ − is independent of z, whereas ' , 0 pr I is independent of t, the above equation can be simplified as follows. ( ) ( )( ) ( ) ( ) ( ) ( ) ∫ ∫ ∞ + ∞ − +∞ ∞ − − ⋅ − − ⋅ ⋅ + + = ∂ ∂ ⋅         − dt t t I I s s s s s s I z dt t pr p d pr p pry py prx px py px p pr 2 2 2 2 ' , 0 , 0 2 2 2 2 ' , 0 2 2 2 exp 2 exp - 2 exp τ τ τ β τ After rearranging the above equation, we get: ( )( ) ( ) ( ) ( ) ( ) ( ) ∫ ∫ ∞ + ∞ − ∞ + ∞ − − − ⋅ − − ⋅ ⋅ + + = ∂ ∂ dt t dt t t I I s s s s s s I z pr pr p d pr p pry py prx px py px p 2 2 2 2 2 2 ' , 0 , 0 2 2 2 2 ' , 0 2 exp 2 exp 2 exp - τ τ τ τ β where the denominator represents the normalization of the probe. Carrying out the integration in the fraction results in: ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 exp 2 exp 2 2 2 exp 2 exp 2 exp pr p d pr p p pr p d pr pr p pr p pr pr p d dt t dt t t τ τ τ τ τ τ τ τ τ τ π τ τ τ τ π τ τ τ τ + − ⋅ + = + − ⋅ + = − − ⋅ − − ∫ ∫ ∞ + ∞ − ∞ + ∞ − Collect the terms for the pump (spot size, pulse duration and intensity), we have: ( )( ) ( ) ( ) ( ) p p py px pr p pr p d pry py prx px pr p I s s s s s s I I z , 0 2 2 2 2 2 2 2 2 2 ' , 0 ' , 0 2 exp - τ τ τ τ τ τ β ⋅ + + − ⋅ + + = ∂ ∂ 233 The last term on the right hand side of the equation is related to the total energy (E p , in Joule) of the pump: ( ) ( ) p p py px p p py px p py px p p I s s I s s t s y s x I dt dy dx t y x I dt dy dx , 0 2 / 3 , 0 2 2 2 2 2 2 , 0 2 2 2 2 2 2 2 exp ) , , ( τ π τ π π π τ = ⋅ ⋅ ⋅ = − − − = ∫ ∫ ∫ ∫ ∫ ∫ +∞ ∞ − +∞ ∞ − +∞ ∞ − +∞ ∞ − +∞ ∞ − +∞ ∞ − Therefore, the differential equation can be further simplified to the following: ( )( ) ( ) ( ) P pr p pr p d pry py prx px pr p E s s s s I I z ⋅ + ⋅ + − ⋅ + + = ∂ ∂ 2 2 2 2 2 2 2 2 2 ' , 0 ' , 0 2 2 exp 2 - τ τ π τ τ τ π β We can rearrange, integrate over the whole path length (L) and arrive in the final form in terms of the experimental observable ΔA 2PA . ( )( ) ( ) ( ) ( ) ( )( ) ( ) ( ) L E s s s s I I dz E s s s s I dI P pr p pr p d pry py prx px f i L P pr p pr p d pry py prx px I I p p f i β τ τ π τ τ τ π β τ τ π τ τ τ π ⋅ + ⋅ + − ⋅ + + =         ⇒ ⋅ − ⋅ + ⋅ + − ⋅ + + = ∫ ∫ 2 2 2 2 2 2 2 2 2 , 0 , 0 0 2 2 2 2 2 2 2 2 2 ' , 0 ' , 0 2 2 exp 2 1 log ) 10 ln( 2 2 exp 2 1 , 0 , 0 Since log(I 0,i /I 0,f ) is ΔA 2PA , 2 2 pr p τ τ + is the cross correlation width (τ cc ), the above equation can be written as: ( ) 2 2 2 2 exp 2 1 ) 10 ln( 1 cc d cc P PA L E f A τ τ τ π β − ⋅ ⋅ ⋅ ⋅ ⋅ = Δ (1) where 2PA coefficient β is in cm/W, pump energy E p is in Joules, path length L is in cm, cross correlation width τ cc is in s -1 , τ d is the delay between the pump and probe pulse, and 234 f is a constant factor accounting for the beam size difference between the pump and probe, which is defined as the equation below and has a unit of cm -2 . ( ) ( ) 2 , 2 , 2 , 2 , 2 1 y probe y pump x probe x pump s s s s f + ⋅ + = π (2) where s i,j again is the half width at e -1/2 of the maximum value, in the horizontal (x) or vertical (y) direction, s i,j can be converted to the more commonly used full width at half maximum (FWHM) via s s FWHM ⋅ = ⋅ = 355 . 2 ) 2 ln( 2 2 . 235 Appendix C. MATLAB Script for 2PA Spectrum Processing The MATLAB script “CalTPACrossSection.m” converts the raw data in ΔA 2PA to σ 2PA by integration or time-zero methods. If the user provides parameters to calculate absolute cross sections (pump pulse energy, pump spot size, probe size, sample thickness and concentration), the final result will be in GM, otherwise it will be Arbitrary Units. This script allows the user to define an integration window along the continuum chirp in order to minimize the signal contribution from photoproducts. It also provides a means to test whether the integration window is defined properly by narrowing and widening the defined window and plot the associated error (if the integration method is chosen). It is noteworthy that this script is interactive – the user is urged to read and follow the prompt carefully, and it requires MATLAB script “ContinousTPAGaussFit.m” and “ContinousTPAGaussFit1.m” to function properly (if the time zero cut method is chosen). The complete contents for “CalTPACrossSection.m”, “ContinousTPAGaussFit.m” and “ContinousTPAGaussFit1.m” are presented as follows. Figure C.1 presents the flowchart indicating how “CalTPACrossSection.m” handles the raw data. Figure C.2 illustrates the user-define integration window and how “CalTPACrossSection.m” tests the integration limit. Figure C.3 presents an example of the convergence test. Lastly, Figure C.4 presents a sample output of a 2PA spectrum. All these scripts were written for MATLAB 6.1. 236 CalTPACrossSection.m: function TPAstruct=CalTPACrossSection(DataStruct, WavelengthRange, TimeRange, PumpWavelength, CrossSectionOption, ViewOption, PPStructOption); % % TPAstruct=CalTPACrossSection(DataStruct, WavelengthRange, TimeRange, PumpWavelength, CrossSectionOption, ViewOption, PPStructOption) % % WavelengthRange as [xx:yy] vector (for maximum range type [;], but not recommanded due to bad data points at the first few pixels) % TimeRange as [xx:yy] vector (for maximn range type [;]) % PumpWavelength is the fixed photon frequency in nm % CrossSectionOption is the choice of calculating relative cross section or absolute cross section. % For *relative* cross section, type [;] % For *absolute* cross section, provide the following *seven(7)* quantities (in order) in a square bracket, seperated by space: % PumpPulseEnergy (microjoules) % PumpSpotSize in horizontal direction (FWHM,microns) % PumpSpotSize in vertical direction (FWHM,microns) % ProbeSpotSize in horizontal direction (FWHM,microns) % probeSpotSize in vertical direction (FWHM,microns) % JetThickness (microns) % Concentration (mol/L) % example: [0.20 400 200 100 80 70 55.56] % ViewOption will enable/disable viewing integration graphically (0: disable, 1: enable) % PPStructOption is the file from which you wish to import the integration limit (optional) % % This script is suitable for calculating 2-photon absorption cross section with a TWO-BEAM experiment. % It assumes you have subtracted an pre-time zero signals!!! % It allows you to choose a variable integration window (if it is a broadband experiment). % Dispersion is taken into account upon choosing the integratin window. % It assumes that there is negligible pump absorption. % 1PA at the probe wavelength is already automatically subtracted by measuring the probe signal at pump on and pump off. % If the sample is a pure liquid, then the program directly produce the absolute cross section of the molecule (in GM). % If the sample is a solution, the beta (NOT cross section) of the solvent can be directly subtrated from that of the solution. % % For Single-Color experiment, the 2PA cross section information is stored in: % TPAstruct.AbsCrossSection TPAstruct.AbsCrossSection % Column 1 delta from time zero cut delta from time zero cut % Column 2 delta from integration delta from integration % % For broadband experiment with *Time Zero Cut* method, the 2PA cross section information is stored in: % TPAstruct.AbsCrossSectionByGaussFit TPAstruct.RelCrossSectionByGaussFit % Column 1 probe wavelength probe wavelength % Column 2 2PA peak amplitude (from fitting) 2PA peak amplitude (from fitting) % Column 3 2PA peak "sigma" (from fitting) 2PA peak "sigma" (from fitting) % Column 4 total wavenumber (in 1/cm) total wavenumber (in 1/cm) % Column 5 equivelent one-photon wavelength (in nm) equivelent one-photon wavelength (in nm) % Column 6 total excitation energy (in eV) total excitation energy (in eV) % Column 7 2PA coefficient (beta, in cm/GW) Relative delta: sqrt(2*pi)*amplitude*sigma % Column 8 absolute 2PA cross section (delta, in GM) ---- % % For broadband experiment with *Integration* method, the 2PA cross section information is stored in: % TPAstruct.AbsCrossSectionByIntegration TPAstruct.RelCrossSectionByIntegration % Column 1 probe wavelength probe wavelength % Column 2 mOD value after integration mOD value after integration 237 % Column 3 total wavenumber (in 1/cm) total wavenumber (in 1/cm) % Column 4 equivelent one-photon wavelength (in nm) equivelent one-photon wavelength (in nm) % Column 5 total excitation energy (in eV) total excitation energy (in eV) % Column 6 2PA coefficient (beta, in cm/GW) error bar of the mOD value (optional) % Column 7 absolute 2PA cross section (delta, in GM) ---- % Column 8 error bar of delta (optional) ---- % % Please read the prompt carefully as the program runs! % % TZ 07/10/2008 TPAstruct=DataStruct; warning off; A=find(DataStruct.FileName=='_'); DataStruct.FileName(A)=' '; if isempty(WavelengthRange)==1 WavelengthRange=1:length(DataStruct.Wavelength); end if isempty(TimeRange)==1 TimeRange=1:length(DataStruct.ProbeTime); end DW=DataStruct.Wavelength(WavelengthRange)'; DT=DataStruct.ProbeTime(TimeRange)'; DS=DataStruct.PPSpectra(WavelengthRange,TimeRange); % Determine whether it is a single-color measurement (case #1), or a dispersed pump-probe measurement (case #2) if isfield(DataStruct,'IntegratedDOD') CaseNum=1; else CaseNum=2; end switch CaseNum % For single-Color measurements case 1 disp('Single-Color 2PA Measurement'); PumpPulseWidth=input('Enter the FWHM (in fs) of the pump for initial guess value: '); DT=DataStruct.IntegratedDOD(:,1); DS=DataStruct.IntegratedDOD(:,5); Starting(1)=PumpPulseWidth/2.355/1000; Starting(2)=DT(dsearchn(DS,max(DS))); Starting(3)=max(DS); options=optimset('Display','Off','TolX',1E-500); Estimate=fminsearch(@ContinousTPAGaussFit1,Starting,options,DT(TimeRange),DS(TimeRange)'); Fitting(1)=Estimate(1); Fitting(2)=Estimate(2); 238 Fitting(3)=Estimate(3); FWHM=Fitting(1)*2.355*1000; disp(['2PA FWHM= ',num2str(FWHM),' fs']); % Calculating absolute cross section for single color measurement if length(CrossSectionOption)==7 PumpPulseEnergy=CrossSectionOption(1); PumpSpotSizeX=CrossSectionOption(2)/10000/2.355; PumpSpotSizeY=CrossSectionOption(3)/10000/2.355; ProbeSpotSizeX=CrossSectionOption(4)/10000/2.355; ProbeSpotSizeY=CrossSectionOption(5)/10000/2.355; JetThickness=CrossSectionOption(6); Concentration=CrossSectionOption(7); % Using time zero cut to compute 2PA cross section % Two photon absorption coefficient (Beta) = SpatialFactor * sqrt(2pi) * DeltaOD(delay=0) / (Pump Pulse Energy * Jet Thickness) SpatialFactor=2*pi*((PumpSpotSizeX^2+ProbeSpotSizeX^2)*(PumpSpotSizeY^2+ProbeSpotSizeY^2))^(1/ 2); Beta1=SpatialFactor*log(10.0)*sqrt(2*pi)*Fitting(1)*1000*Fitting(3)/1000/PumpPulseEnergy/(JetThickness*(1E-4)); % Two photon cross section is Beta * pump photon energy / solute concentration. Delta1=6.626*(1E-34)*2.998*(1E+10)/(PumpWavelength*(1E-7))*Beta1*(1E- 9)/(6.02*(1E+20)*Concentration)*(1E+50); disp(['Absolute Cross Section From Gaussian fit: ',num2str(Delta1),' GM']) % Using integration to compute 2PA cross section % Two photon absorption coefficient (Beta) = SpatialFactor * ln(10) * (Integrated DeltaOD) / (Pump Pulse Energy * Jet Thickness) SpatialFactor=2*pi*((PumpSpotSizeX^2+ProbeSpotSizeX^2)*(PumpSpotSizeY^2+ProbeSpotSizeY^2))^(1/ 2); IntDOD=integrate(DT',DS',min(DT(TimeRange)),max(DT(TimeRange))); Beta2=SpatialFactor*log(10.0)*IntDOD/PumpPulseEnergy/(JetThickness*(1E-4)); % Two photon cross section is Beta * pump photon energy / solute concentration. Delta2=6.626*(1E-34)*2.998*(1E+10)/(PumpWavelength*(1E-7))*Beta2*(1E- 9)/(6.02*(1E+20)*Concentration)*(1E+50); disp(['Absolute Cross Section From integration: ',num2str(Delta2),' GM']) TPAstruct.AbsCrossSection=[Delta1 Delta2]; end % Calculating relative cross section for single color measurement if isempty(CrossSectionOption) | length(CrossSectionOption)<7 Delta1=sqrt(2*pi)*Fitting(1)*1000*Fitting(3)/1000; disp(['Relative Cross Section From Gaussian fit: ',num2str(Delta1),' A.U.']); Delta2=integrate(DT',DS',min(DT(TimeRange)),max(DT(TimeRange))); disp(['Relative Cross Section From integration: ',num2str(Delta2),' A.U.']) TPAstruct.RelCrossSection=[Delta1 Delta2]; end clf; plot(DataStruct.IntegratedDOD(:,1),DataStruct.IntegratedDOD(:,5),'ro'); 239 title(['One-Color Measurement: ',DataStruct.FileName,' (baseline corrected)']); xlabel('Delay Time (ps)'); ylabel('\DeltaOD (mOD)'); hold on; for c=1:length(DT) y(c)=Fitting(3)*exp(-(DT(c)-Fitting(2))^2/(2*Fitting(1)^2)); end plot(DT,y,'b-'); legend('Single Color 2PA Measurement','Gaussian Fit') % For dispersed pump-probe measurement case 2 method=input(['Please choose how you want to compute the 2PA cross sections. [1]Gaussian fitting. [2]Integration. [3]Both. Your choice: ']); % For computing cross sections using Gaussian fitting parameters if method==1 | method==3 disp('Computing 2PA cross section from Gaussian fitting parameters.'); % Fitting 2PA spike with Gaussian functions PumpPulseWidth=input('Enter the FWHM (in fs) of the pump for initial guess value: '); disp('Fitting 2PA signal with Gaussian function, please wait, this will take a few minutes...') Starting(1)=PumpPulseWidth/2.355/1000; for a=1:length(DW) Starting(2)=DT(dsearchn(DS(a,:)',max(DS(a,:)))); Starting(3)=max(DS(a,:)); options=optimset('Display','Off','TolX',1E-500); Estimate=fminsearch(@ContinousTPAGaussFit1,Starting,options,DT,DS(a,:)); newmatrix(a,1)=Estimate(1); newmatrix(a,2)=Estimate(2); newmatrix(a,3)=Estimate(3); end SelectedPixels=input('Please choose the pixels (even number) you wish the program to plot: '); disp('Please check the fitting quality. Press any key to continue...') clf; for b=1:length(SelectedPixels) subplot(length(SelectedPixels)/2,2,b); plot(DT,DS(SelectedPixels(b),:),'ro','markersize',4); xlabel('Delay Time (ps)'); ylabel('\Delta OD (mOD)'); title(['pixel',num2str(SelectedPixels(b))]); hold on; for c=1:length(DT) y(c)=newmatrix(SelectedPixels(b),3)*exp(-(DT(c)- newmatrix(SelectedPixels(b),2))^2/(2*newmatrix(SelectedPixels(b),1)^2)); end plot(DT,y,'b-'); hold on; end TPAstruct.TimeZeroCutInfo=newmatrix; pause out(:,1)=DW; out(:,2)=TPAstruct.TimeZeroCutInfo(:,3)/1000; out(:,3)=TPAstruct.TimeZeroCutInfo(:,1)*1000; out(:,4)=10^7./out(:,1)+10^7/PumpWavelength; 240 out(:,5)=10^7./out(:,4); out(:,6)=out(:,4)*1.2398*(1E-4); % Calculating absolute cross sections if length(CrossSectionOption)==7 disp('Computing absolute 2PA cross section...'); PumpPulseEnergy=CrossSectionOption(1); PumpSpotSizeX=CrossSectionOption(2)/10000/2.355; PumpSpotSizeY=CrossSectionOption(3)/10000/2.355; ProbeSpotSizeX=CrossSectionOption(4)/10000/2.355; ProbeSpotSizeY=CrossSectionOption(5)/10000/2.355; JetThickness=CrossSectionOption(6); Concentration=CrossSectionOption(7); % Two photon absorption coefficient (Beta) = SpatialFactor * sqrt(2pi) * DeltaOD(delay=0) / (Pump Pulse Energy * Jet Thickness) SpatialFactor=2*pi*((PumpSpotSizeX^2+ProbeSpotSizeX^2)*(PumpSpotSizeY^2+ProbeSpotSizeY^2))^(1/ 2); disp(['spatial factor:',num2str(SpatialFactor)]); for a=1:length(out(:,1)) out(a,7)=SpatialFactor*log(10.0)*sqrt(2*pi)*out(a,3)*out(a,2)/PumpPulseEnergy/(JetThickness*(1E-4)); end % Two photon cross section is Beta * pump photon energy / solute concentration. Beta is just out(:,6) out(:,8)=6.626*(1E-34)*2.998*(1E+10)/(PumpWavelength*(1E-7))*out(:,7)*(1E- 9)/(6.02*(1E+20)*Concentration)*(1E+50); TPAstruct.AbsCrossSectionByGaussFit=out; clf; subplot(3,1,1); contour(DW,DT,DS',30); xlabel('Wavelength (nm)'); ylabel('Probe Time (ps)'); grid on; colorbar hold on; axis tight; Axe=Axis; axis([Axe(1) Axe(2) min(DataStruct.ProbeTime) max(DataStruct.ProbeTime)]); subplot(3,2,3) plot(out(:,1),out(:,8),'ro-'); grid on; xlabel('Probe Wavelength (nm)'); ylabel('\delta (GM)'); grid on; AxeFigure; subplot(3,2,4) plot(out(:,4),out(:,8),'ro-'); grid on; xlabel('Wavenumber (1/cm)'); ylabel('\delta (GM)'); grid on; AxeFigure; subplot(3,2,5) plot(out(:,5),out(:,8),'ro-'); grid on; xlabel('One-photon Equivelent Wavelength (\lambda_1\lambda_2)/(\lambda_1+\lambda_2) (nm) '); ylabel('\delta (GM)'); grid on; AxeFigure; subplot(3,2,6) plot(out(:,6),out(:,8),'r-'); grid on; hold on; 241 xlabel('Total Excitation Energy E_1+E_2 (eV) '); ylabel('\delta (GM)'); grid on; AxeFigure; suptitle(['Absolute 2PA (time zero cut): ', DataStruct.FileName]); orient landscape; AbsFileName=input('Saving the absolute 2PA information... please name the file (including .txt): ', 's'); SaveAbsFile=['save ' AbsFileName ' out -ascii -tabs']; eval(SaveAbsFile); disp(['Done! File saved in ',pwd,'\',AbsFileName]); end % Calculating relative cross sections if isempty(CrossSectionOption) | length(CrossSectionOption)<7 disp('Computing relative 2PA cross section...') for a=1:length(out(:,1)) out(a,7)=sqrt(2*pi)*out(a,3)*out(a,2); end TPAstruct.RelCrossSectionByGaussFit=out; clf; subplot(3,1,1); contour(DW,DT,DS',30); xlabel('Wavelength (nm)'); ylabel('Probe Time (ps)'); grid on; colorbar hold on; axis tight; Axe=Axis; axis([Axe(1) Axe(2) min(DataStruct.ProbeTime) max(DataStruct.ProbeTime)]); subplot(3,2,3) plot(out(:,1),out(:,7),'ro-'); grid on; xlabel('Probe Wavelength (nm)'); ylabel('A.U.'); grid on; AxeFigure; subplot(3,2,4) plot(out(:,4),out(:,7),'ro-'); grid on; xlabel('Wavenumber (1/cm)'); ylabel('A.U.'); grid on; AxeFigure; subplot(3,2,5) plot(out(:,5),out(:,7),'ro-'); grid on; xlabel('One-photon Equivelent Wavelength (\lambda_1\lambda_2)/(\lambda_1+\lambda_2) (nm) '); ylabel('A.U.'); grid on; AxeFigure; subplot(3,2,6) plot(out(:,6),out(:,7),'r-'); grid on; hold on; xlabel('Total Excitation Energy E_1+E_2 (eV) '); ylabel('A.U.'); grid on; AxeFigure; suptitle(['Relative 2PA (time zero cut): ', DataStruct.FileName]); orient landscape; RelFileName=input('Saving the relative 2PA information... please name the file (including .txt): ', 's'); SaveRelFile=['save ' RelFileName ' out -ascii -tabs']; eval(SaveRelFile); disp(['Done! File saved in ',pwd,'\',RelFileName]); 242 end end if method==2 | method==3; % For computing cross sections using integration % Check existence of the integration window disp('Computing 2PA cross section from integration over delay time.'); if isfield(TPAstruct,'IntegrationUpperLimit') & isfield(TPAstruct,'IntegrationLowerLimit'); disp('Found integration width function. Do you wish to keep it? Y/n'); KeepWindow=input('Warning: Answer [n] will erase all previous 2PA information!!! [Y] ','s'); if isempty(KeepWindow) | KeepWindow=='Y' | KeepWindow=='y' DispersionFrom=TPAstruct.IntegrationLowerLimit; DispersionTo=TPAstruct.IntegrationUpperLimit; elseif KeepWindow=='n' TPAstruct.IntegrationLowerLimit=[]; TPAstruct.IntegrationUpperLimit=[]; TPAstruct.PercentError=[]; TPAstruct.RelCrossSectionByIntegration=[]; TPAstruct.AbsCrossSectionByIntegration=[]; % Disable the following in order to save memory, enable them only when necessary. % TPAstruct.WidthConvergenceTest=[]; % TPAstruct.IntegratedAreasErrorAnalysis=[]; end end % Choose import or construct integration window if it does not already exist % Import integration window if isempty(DispersionFrom) | isempty(DispersionTo); if exist('PPStructOption','var')==1 disp('Importing integration limit from designated file...') DispersionFrom=PPStructOption.IntegrationLowerLimit; DispersionTo=PPStructOption.IntegrationUpperLimit; disp('Done!'); else % Construct integration window disp('Constructing variable integration width...'); for counter=1:2 GoodEnough='n'; clear Point; clf; contourf(DW,DT,DS',20,'k'); xlabel('Probe Wavelength (nm)','fontsize',14), ylabel('ProbeTime (ps)','fontsize',14); set(gca,'fontsize',14); axis tight; Axe=axis; hold on; if counter==2 plot(DispersionFrom(WavelengthRange,1),DispersionFrom(WavelengthRange,2),'g--'); end hold off; j=1; Point(1,1)=mean(Axe(1:2)); Point(1,2)=mean(Axe(3:4)); disp('Use the mouse to point to the point you want to add THEN enter key (for every point)'); 243 while GoodEnough=='n' title('Press Enter inside the box for a point, outside for the end. \newlineconstruct the *lower* limit THEN the *upper* limit!! ','fontsize',12); Point(j,:)=ginput; disp([num2str(Point(j,1)) ' ' num2str(Point(j,2))]); if Point(j,1) < Axe(1) | Point(j,1) > Axe(2) | Point(j,2) < Axe(3) | Point(j,2) > Axe(4) Point(j,:)=[]; break; end Point=sortrows(Point); dispersiondata(:,1) =DataStruct.Wavelength'; [QuarticPoly,S] =polyfit(Point(:,1),Point(:,2),4); dispersiondata(:,2) =polyval(QuarticPoly,DataStruct.Wavelength'); clf; contourf(DW,DT,DS',20,'k'); xlabel('Probe Wavelength (nm)','fontsize',14), ylabel('ProbeTime (ps)','fontsize',14); set(gca,'fontsize',14); title('Press Enter inside the box for a point, outside for the end'); hold on; axis tight; plot(Point(:,1),Point(:,2),'r*'); plot(dispersiondata(WavelengthRange,1),dispersiondata(WavelengthRange,2),'r--'); if counter==2 plot(DispersionFrom(WavelengthRange,1),DispersionFrom(WavelengthRange,2),'g--'); end j=j+1; end disp('Dispersion is based on the Quartic fit'); if counter==1 DispersionFrom=dispersiondata; else DispersionTo=dispersiondata; end end end end TPAstruct.IntegrationUpperLimit=DispersionTo; TPAstruct.IntegrationLowerLimit=DispersionFrom; %Check for convergence by varying integration window width ConvTest=input('Do you wish to test the convergence of the integration windows Y/n? [Y] ','s'); if isempty(ConvTest) | ConvTest=='Y' | ConvTest=='y' GoodEnough='n'; while GoodEnough=='n'; 244 Shift=input('Please enter a *positive* number (in ps), indicating the distance above and below the widths: '); disp('Please wait, computing integration with various integration windows...') for a=2:10 DispersionFromNew(:,2)=DispersionFrom(:,2)+(Shift/4)*(a-6); DispersionToNew(:,2)=DispersionTo(:,2)-(Shift/4)*(a-6); for i=WavelengthRange test(i,1)=DataStruct.Wavelength(i); test(i,a)=integrate(DataStruct.ProbeTime, DataStruct.PPSpectra(i,:), DispersionFromNew(i,2), DispersionToNew(i,2)); end end for i=WavelengthRange RefFrom(i,1)=DataStruct.Wavelength(i); RefFrom(i,2)=DataStruct.ProbeTime(min(TimeRange)+3); RefTo(i,1)=DataStruct.Wavelength(i); RefTo(i,2)=DataStruct.ProbeTime(max(TimeRange)-3); test(i,11)=integrate(DataStruct.ProbeTime,DataStruct.PPSpectra(i,:), RefFrom(i,2), RefTo(i,2)); end Junker=1:length(DataStruct.Wavelength); Junker(WavelengthRange)=[]; test(find(Junker<max(WavelengthRange)),:)=[]; disp('Done!') RefChoice=input('Please choose a reference: [1]maximum possible constant width, or [2]original graphically determined variable width: '); disp('Plotting integrated Area against width of the integration window at six different wavelengths...'); ChooseLambda=input(['Please enter SIX wavelengths (' num2str(round(min(DataStruct.Wavelength(WavelengthRange)))) 'nm-' num2str(round(max(DataStruct.Wavelength(WavelengthRange)))) 'nm), \nOR press Enter for equally divided space: ']); if isempty(ChooseLambda) interval=(max(DataStruct.Wavelength(WavelengthRange))- min(DataStruct.Wavelength(WavelengthRange)))/7; start=min(DataStruct.Wavelength(WavelengthRange)); ChooseLambda=[start+interval start+interval*2 start+interval*3 start+interval*4 start+interval*5 start+interval*6]; end ChooseLambda=ChooseLambda'; LookupList=DataStruct.Wavelength(WavelengthRange)'; LambdaChoice=dsearchn(LookupList,ChooseLambda); % Choose maximum possible fixed widith as reference if RefChoice==1 %Plotting Integrated Area against width of the integration at selected wavelengths clf; subplot(4,1,1); contour(DW,DT,DS',30); ylabel('\DeltaOD'); grid on; colorbar hold on; plot(RefTo(WavelengthRange,1),RefTo(WavelengthRange,2),'r--'); 245 plot(RefFrom(WavelengthRange,1),RefFrom(WavelengthRange,2),'r--'); axis tight; Axe=Axis;axis([Axe(1) Axe(2) min(DataStruct.ProbeTime) max(DataStruct.ProbeTime)]); title('Reference - maximum possible width'); for b=1:6 c=LambdaChoice(b); for a=2:10 IntWinConv(a-1,1)=-(Shift/4)*(a-6); IntWinConv(a-1,b+1)=test(c,a); subplot(4,2,b+2); plot(IntWinConv(a-1,1),IntWinConv(a-1,b+1),'ro-'); text(IntWinConv(a-1,1),IntWinConv(a-1,b+1),[' #',num2str(a)],'fontsize',8); xlabel ('\Delta width'); ylabel('Integrated Area');title([num2str(test(c,1)) 'nm (ref=' num2str(test(c,11)) ')']); grid on; AxeFigure; hold on; end end suptitle(['Convergence Test: ',DataStruct.FileName, '\newline(case numbers are shown next to the points)']) orient tall; disp('Done!'); %Plotting relative error as a function of wavelength RelErr=input('Do you wish to plot relative error as a function of wavelength Y/n? [Y] ','s'); if isempty(RelErr) | RelErr=='Y' disp('Please wait, plotting relative error against probe wavelength...') disp('Relative Error = (A(new range)-A(reference))/A(reference) * 100%') disp('case #2-5 are wider windows, case #6 is the original selection,case #7-10 are narrower windows') subplot(5,2,1); contour(DW,DT,DS',30); xlabel('Wavelength'); ylabel('\DeltaOD'); grid on; colorbar hold on; plot(RefTo(WavelengthRange,1),RefTo(WavelengthRange,2),'r--'); plot(RefFrom(WavelengthRange,1),RefFrom(WavelengthRange,2),'r--'); axis tight; Axe=Axis; title('Reference - maximum possible width'); for f=2:5 diff=(test(:,f)-test(:,11)); den=abs(test(:,11)); for i=1:length(test) y(i)=diff(i)/den(i)*100; end subplot(5,2,f); plot(test(:,1),y,'r-'); ylabel('% error'); title(['Case #',num2str(f),': ',num2str(Shift/4*(6-f)),'ps wider']); grid on; AxeFigure; end 246 diff=(test(:,6)-test(:,11)); den=abs(test(:,11)); for i=1:length(test) y(i)=diff(i)/den(i)*100; end subplot(5,2,6); plot(test(:,1),y,'r-'); ylabel('% error'); title('Case #6: Original Selection'); grid on; AxeFigure; for g=7:10 diff=(test(:,g)-test(:,11)); den=abs(test(:,11)); for i=1:length(test) y(i)=diff(i)/den(i)*100; end subplot(5,2,g); plot(test(:,1),y,'r-'); ylabel('% error'); title(['Case #', num2str(g),': ',num2str(Shift/4*(g-6)),'ps narrower']); grid on; AxeFigure; end suptitle(['Error Analysis: ',DataStruct.FileName, '\newline(negative number means the area is smaller than that of the the reference)']); orient tall; disp('Done!') end end % Choose originally selected variable width as reference if RefChoice==2 %Plotting Integrated Area against width of the integration at selected wavelengths clf; subplot(4,1,1); contour(DW,DT,DS',30); ylabel('\DeltaOD'); grid on; colorbar hold on; plot(DispersionTo(WavelengthRange,1),DispersionTo(WavelengthRange,2),'r--'); plot(DispersionFrom(WavelengthRange,1),DispersionFrom(WavelengthRange,2),'r--'); axis tight; Axe=Axis; axis([Axe(1) Axe(2) min(DataStruct.ProbeTime) max(DataStruct.ProbeTime)]); title('Reference - Original Selection'); for b=1:6 c=LambdaChoice(b); for a=2:10 IntWinConv(a-1,1)=-(Shift/4)*(a-6); IntWinConv(a-1,b+1)=test(c,a); subplot(4,2,b+2); plot(IntWinConv(a-1,1),IntWinConv(a-1,b+1),'ro-'); text(IntWinConv(a-1,1),IntWinConv(a-1,b+1),[' #',num2str(a)],'fontsize',8); xlabel ('\Delta width'); ylabel('Integrated Area');title([num2str(test(c,1)) 'nm']); grid on; AxeFigure; hold on; 247 end end suptitle(['Convergence Test: ',DataStruct.FileName, '\newline(case numbers are shown next to the points)']) orient tall; disp('Done!'); %Plotting relative error as a function of wavelength RelErr=input('Do you wish to plot relative error as a function of wavelength Y/n? [Y] ','s'); if isempty(RelErr) | RelErr=='Y' disp('Please wait, plotting relative error against probe wavelength...') disp('Relative Error = (A(new range)-A(original range))/A(original range) * 100%') disp('case #2-5 are wider windows, case #6 is the original selection,case #7-10 are narrower windows') clf; subplot(5,1,1); contour(DW,DT,DS',30); ylabel('\DeltaOD'); grid on; colorbar hold on; plot(DispersionTo(WavelengthRange,1),DispersionTo(WavelengthRange,2),'r--'); plot(DispersionFrom(WavelengthRange,1),DispersionFrom(WavelengthRange,2),'r--'); axis tight; Axe=Axis; axis([Axe(1) Axe(2) min(DataStruct.ProbeTime) max(DataStruct.ProbeTime)]); for f=2:5 diff=(test(:,f)-test(:,6)); den=abs(test(:,6)); for i=1:length(test) y(i)=diff(i)/den(i)*100; end subplot(5,2,f+1); plot(test(:,1),y,'r-'); ylabel('% error'); title(['Case #',num2str(f),': ',num2str(Shift/4*(6-f)),'ps wider']); grid on; AxeFigure; end for g=7:10 diff=(test(:,g)-test(:,6)); den=abs(test(:,6)); for i=1:length(test) y(i)=diff(i)/den(i)*100; end subplot(5,2,g); plot(test(:,1),y,'r-'); ylabel('% error'); title(['Case #',num2str(g),': ',num2str(Shift/4*(g-6)),'ps narrower']); grid on; AxeFigure; end suptitle(['Error Analysis: ',DataStruct.FileName, '\newline(negative number means the area is smaller than that of the the reference)']); orient tall; disp('Done!') end 248 end Choice=input('Please enter the case # of the best integration window: '); DispersionFrom(WavelengthRange,2)=DispersionFrom(WavelengthRange,2)+(Shift/4)*(Choice-6); DispersionTo(WavelengthRange,2)=DispersionTo(WavelengthRange,2)-(Shift/4)*(Choice-6); out(:,1)=test(:,1); out(:,2)=test(:,Choice); if RefChoice==1 diff=test(:,Choice)-test(:,11); den=test(:,11); for i=1:length(test) error(i)=abs(diff(i)/den(i)); end end if RefChoice==2 diff=test(:,Choice)-test(:,6); den=test(:,6); for i=1:length(test) error(i)=abs(diff(i)/den(i)); end end GoodEnough=input('Is this integration window good enough Y/n? [Y] ','s'); if isempty(GoodEnough) | GoodEnough=='Y' GoodEnough='y'; TPAstruct.IntegrationUpperLimit=DispersionTo; TPAstruct.IntegrationLowerLimit=DispersionFrom; TPAstruct.PercentError=error'; % Disable the following in order to save memory, enable them only when necessary. % TPAstruct.WidthConvergenceTest=IntWinConv; % TPAstruct.IntegratedAreasErrorAnalysis=test; else GoodEnough='n'; end end end %Integration from and to previous chosen lines disp('Scanning over the wavelength range and integrating...'); clf; subplot(3,1,1); contour(DW,DT,DS',30); ylabel('\DeltaOD'); grid on; colorbar hold on; plot(DispersionTo(WavelengthRange,1),DispersionTo(WavelengthRange,2),'r- -'); plot(DispersionFrom(WavelengthRange,1),DispersionFrom(WavelengthRange,2),'r--'); axis tight; Axe=Axis; axis([Axe(1) Axe(2) min(DataStruct.ProbeTime) max(DataStruct.ProbeTime)]); for i=WavelengthRange 249 Range=find(DataStruct.ProbeTime > DispersionFrom(i,2) & DataStruct.ProbeTime < DispersionTo(i,2)); subplot(3,1,2); plot(DataStruct.ProbeTime(Range),DataStruct.PPSpectra(i,Range),'.-'); xlabel('ProbeTime'); ylabel('\DeltaOD'); title([ num2str(round(DataStruct.Wavelength(i))) ' nm']) grid on; AxeFigure; out(i,1)=DataStruct.Wavelength(i); out(i,2)=integrate(DataStruct.ProbeTime, DataStruct.PPSpectra(i,:), DispersionFrom(i,2), DispersionTo(i,2)); subplot(3,1,3) plot(out(i,1),out(i,2),'ro'); grid on; xlabel('Wavelength (nm)'); ylabel('\DeltaOD'); title([ num2str(round(DataStruct.Wavelength(i))) ' nm']) hold on; grid on; AxeFigure; if ViewOption==1 pause(0.01); end end disp('Done!'); out(:,1)=out(:,1); out(:,2)=out(:,2); out(:,3)=10^7./out(:,1)+10^7/PumpWavelength; out(:,4)=10^7./out(:,3); out(:,5)=out(:,3)*1.2398*(1E-4); Junker=1:length(DataStruct.Wavelength); Junker(WavelengthRange)=[]; out(find(Junker<max(WavelengthRange)),:)=[]; % Computing absolute 2PA cross section if length(CrossSectionOption)==7 PumpPulseEnergy=CrossSectionOption(1); PumpSpotSizeX=CrossSectionOption(2)/10000/2.355; PumpSpotSizeY=CrossSectionOption(3)/10000/2.355; ProbeSpotSizeX=CrossSectionOption(4)/10000/2.355; ProbeSpotSizeY=CrossSectionOption(5)/10000/2.355; JetThickness=CrossSectionOption(6); Concentration=CrossSectionOption(7); disp('Calculating Absolute 2PA cross section...'); % Two photon absorption coefficient (Beta) = SpatialFactor * ln(10) * (Integrated DeltaOD) / (Pump Pulse Energy * Jet Thickness) SpatialFactor=2*pi*((PumpSpotSizeX^2+ProbeSpotSizeX^2)*(PumpSpotSizeY^2+ProbeSpotSizeY^2))^(1/ 2); disp(['spatial factor:',num2str(SpatialFactor)]); out(:,6)=SpatialFactor*log(10.0)*out(:,2)/PumpPulseEnergy/(JetThickness*(1E-4)); % Two photon cross section is Beta * pump photon energy / solute concentration. Beta is just out(:,6) out(:,7)=6.626*(1E-34)*2.998*(1E+10)/(PumpWavelength*(1E- 7))*out(:,6)*(1E-9)/(6.02*(1E+20)*Concentration)*(1E+50); if isfield(TPAstruct,'PercentError') for i=1:length(WavelengthRange) ErrorBar(i)=error(i)*out(i,7); 250 end out(:,8)=ErrorBar'; end TPAstruct.AbsCrossSectionByIntegration=out; disp('Done!'); if method==2 clf; subplot(3,1,1); contour(DW,DT,DS',30); xlabel('Wavelength (nm)'); ylabel('Probe Time (ps)'); grid on; colorbar hold on; plot(DispersionTo(WavelengthRange,1),DispersionTo(WavelengthRange,2),'r--'); plot(DispersionFrom(WavelengthRange,1),DispersionFrom(WavelengthRange,2),'r--'); axis tight; Axe=Axis; axis([Axe(1) Axe(2) min(DataStruct.ProbeTime) max(DataStruct.ProbeTime)]); subplot(3,2,3) plot(out(:,1),out(:,7),'ro-'); grid on; xlabel('Probe Wavelength (nm)'); ylabel('\delta (GM)'); grid on; AxeFigure; subplot(3,2,4) plot(out(:,3),out(:,7),'ro-'); grid on; xlabel('Total Wavenumber (1/cm)'); ylabel('\delta (GM)'); grid on; AxeFigure; subplot(3,2,5) plot(out(:,4),out(:,7),'ro-'); grid on; xlabel('One-photon Equivelent Wavelength (\lambda_1\lambda_2)/(\lambda_1+\lambda_2) (nm) '); ylabel('\delta (GM)'); grid on; AxeFigure; subplot(3,2,6) plot(out(:,5),out(:,7),'r-'); grid on; hold on; if isfield(TPAstruct,'PercentError') for m=1:9 ErrInterval(m)=min(out(:,5))+(max(out(:,5))-min(out(:,5)))/10*m; end PixErrInterval=dsearchn(out(:,5),ErrInterval'); Errorbar(out(PixErrInterval,5),out(PixErrInterval,7),ErrorBar(PixErrInterval),'b.'); end xlabel('Total Excitation Energy E_1+E_2 (eV) '); ylabel('\delta (GM)'); grid on; AxeFigure; suptitle(['Absolute 2PA (integration): ', DataStruct.FileName]); orient landscape; end if method==3 clf; subplot(3,2,1); contour(DW,DT,DS',30); 251 xlabel('Wavelength (nm)'); ylabel('Probe Time (ps)'); grid on; colorbar hold on; plot(DispersionTo(WavelengthRange,1),DispersionTo(WavelengthRange,2),'r--'); plot(DispersionFrom(WavelengthRange,1),DispersionFrom(WavelengthRange,2),'r--'); axis tight; Axe=Axis; axis([Axe(1) Axe(2) min(DataStruct.ProbeTime) max(DataStruct.ProbeTime)]); subplot(3,2,2) plot(out(:,1),out(:,7),'ro-'); grid on; xlabel('Probe Wavelength (nm)'); ylabel('\delta (GM)'); grid on; AxeFigure; subplot(3,2,3) plot(out(:,3),out(:,7),'ro-'); grid on; xlabel('Total Wavenumber (1/cm)'); ylabel('\delta (GM)'); grid on; AxeFigure; subplot(3,2,4) plot(out(:,4),out(:,7),'ro-'); grid on; xlabel('One-photon Equivelent Wavelength (\lambda_1\lambda_2)/(\lambda_1+\lambda_2) (nm) '); ylabel('\delta (GM)'); grid on; AxeFigure; subplot(3,2,5) plot(out(:,5),out(:,7),'r-'); grid on; hold on; if isfield(TPAstruct,'PercentError') for m=1:9 ErrInterval(m)=min(out(:,5))+(max(out(:,5))-min(out(:,5)))/10*m; end PixErrInterval=dsearchn(out(:,5),ErrInterval'); Errorbar(out(PixErrInterval,5),out(PixErrInterval,7),ErrorBar(PixErrInterval),'b.'); end xlabel('Total Excitation Energy E_1+E_2 (eV) '); ylabel('\delta (GM)'); grid on; AxeFigure; subplot(3,2,6) plot(out(:,5),out(:,7),'r-'); grid on; hold on; plot(TPAstruct.AbsCrossSectionByGaussFit(:,6),TPAstruct.AbsCrossSectionByGaussFit(:,8),'b-'); xlabel('Total Excitation Energy E_1+E_2 (eV) '); ylabel('\delta (GM)'); legend('Integration','Time Zero Cut') grid on; AxeFigure; suptitle(['Absolute 2PA (integration): ', DataStruct.FileName]); legend('Integration','Time Zero Cut',2); orient landscape; end AbsFileName=input('Saving the absolute 2PA information... please name the file (including .txt): ', 's'); SaveAbsFile=['save ' AbsFileName ' out -ascii -tabs']; eval(SaveAbsFile); disp(['Done! File saved in ',pwd,'\',AbsFileName]); end 252 % Relative 2PA cross section is just deltaOD in arbitrary unit if isempty(CrossSectionOption) | length(CrossSectionOption)<7 disp('Calculating Relative 2PA cross section...'); if isfield(TPAstruct,'PercentError') for i=1:length(error) ErrorBar(i)=error(i)*out(i,2); end out(:,6)=ErrorBar'; end TPAstruct.RelCrossSectionByIntegration=out; disp('Done!'); if method==2 clf; subplot(3,1,1); contour(DW,DT,DS',30); xlabel('Wavelength (nm)'); ylabel('Probe Time (ps)'); grid on; colorbar hold on; plot(DispersionTo(WavelengthRange,1),DispersionTo(WavelengthRange,2),'r--'); plot(DispersionFrom(WavelengthRange,1),DispersionFrom(WavelengthRange,2),'r--'); axis tight; Axe=Axis; axis([Axe(1) Axe(2) min(DataStruct.ProbeTime) max(DataStruct.ProbeTime)]); subplot(3,2,3) plot(out(:,1),out(:,2),'ro-'); grid on; xlabel('Probe Wavelength (nm)'); ylabel('A.U.'); grid on; AxeFigure; subplot(3,2,4) plot(out(:,3),out(:,2),'ro-'); grid on; xlabel('Total Wavenumber (1/cm)'); ylabel('A.U.'); grid on; AxeFigure; subplot(3,2,5) plot(out(:,4),out(:,2),'ro-'); grid on; xlabel('One-photon Equivelent Wavelength (\lambda_1*\lambda_2)/(\lambda_1+\lambda_2) (nm) '); ylabel('A.U'); grid on; AxeFigure; subplot(3,2,6) plot(out(:,5),out(:,2),'r-'); grid on; hold on; if isfield(TPAstruct,'PercentError') for m=1:9 ErrInterval(m)=min(out(:,5))+(max(out(:,5))-min(out(:,5)))/10*m; end PixErrInterval=dsearchn(out(:,5),ErrInterval'); errorbar(out(PixErrInterval,5),out(PixErrInterval,2),ErrorBar(PixErrInterval),'b.'); end xlabel('Total Excitation Energy E_1+E_2 (eV) '); ylabel('A.U'); grid on; AxeFigure; suptitle(['Relative 2PA (integration): ', DataStruct.FileName]); orient landscape; 253 end if method==3 clf; subplot(3,2,1); contour(DW,DT,DS',30); xlabel('Wavelength (nm)'); ylabel('Probe Time (ps)'); grid on; colorbar hold on; plot(DispersionTo(WavelengthRange,1),DispersionTo(WavelengthRange,2),'r--'); plot(DispersionFrom(WavelengthRange,1),DispersionFrom(WavelengthRange,2),'r--'); axis tight; Axe=Axis; axis([Axe(1) Axe(2) min(DataStruct.ProbeTime) max(DataStruct.ProbeTime)]); subplot(3,2,2) plot(out(:,1),out(:,2),'ro-'); grid on; xlabel('Probe Wavelength (nm)'); ylabel('A.U.'); grid on; AxeFigure; subplot(3,2,3) plot(out(:,3),out(:,2),'ro-'); grid on; xlabel('Total Wavenumber (1/cm)'); ylabel('A.U.'); grid on; AxeFigure; subplot(3,2,4) plot(out(:,4),out(:,2),'ro-'); grid on; xlabel('One-photon Equivelent Wavelength (\lambda_1*\lambda_2)/(\lambda_1+\lambda_2) (nm) '); ylabel('A.U'); grid on; AxeFigure; subplot(3,2,5) plot(out(:,5),out(:,2),'r-'); grid on; hold on; if isfield(TPAstruct,'PercentError') for m=1:9 ErrInterval(m)=min(out(:,5))+(max(out(:,5))-min(out(:,5)))/10*m; end PixErrInterval=dsearchn(out(:,5),ErrInterval'); errorbar(out(PixErrInterval,5),out(PixErrInterval,2),ErrorBar(PixErrInterval),'b.'); end xlabel('Total Excitation Energy E_1+E_2 (eV) '); ylabel('A.U'); grid on; AxeFigure; subplot(3,2,6) plot(out(:,5),out(:,2),'r- ',TPAstruct.RelCrossSectionByGaussFit(:,6),TPAstruct.RelCrossSectionByGaussFit(:,7),'b-'); grid on; hold on; xlabel('Total Excitation Energy E_1+E_2 (eV) '); ylabel('A.U.'); grid on; AxeFigure; suptitle(['Relative 2PA (Integration): ', DataStruct.FileName]); legend('Integration','Time Zero Cut',2); orient landscape; end RelFileName=input('Saving the relative 2PA information... please name the file (including .txt): ', 's'); SaveRelFile=['save ' RelFileName ' out -ascii -tabs']; 254 eval(SaveRelFile); disp(['Done! File saved in ',pwd,'\',RelFileName]); end end end ContinousTPAGaussFit.m: function TPAstruct=ContinousTPAGaussFit(DataStruct,PumpPulseWidth,SelectedPixels) % TPAstruct=ContinousTPAGaussFit(DataStruct,PumpPulseWidth,SelectedPixels) % % DataStruct is the data structure created from the raw file. % The input of PumpPulseWidth (FWHM, in fs) gives a rough estimate of the crosscorrelation width % Selected Pixels are the pixels you wish the program to plot, in order to check fitting quality (even numbers). % % This script will plot deltaOD as a function of delay time for every pixel, % fit the 2PA spike in time domain with a Gaussian function, % and append the peak value and the width (sigma) to the data structure: TPAStruct.TimeZeroCutInfo. % Column 1 is sigma; column 2 is t0; column 3 is the peak height. % % Note: This script needs to use ContinousTPAGaussFit1.m % % TZ 07/10/2008 DW=DataStruct.Wavelength'; DT=DataStruct.ProbeTime'; DS=DataStruct.PPSpectra; Starting(1)=PumpPulseWidth/2.355/1000; for a=1:length(DW) Starting(2)=DT(dsearchn(DS(a,:)',max(DS(a,:)))); Starting(3)=max(DS(a,:)); options=optimset('Display','Off','TolX',1E-500); Estimate=fminsearch(@ContinousTPAGaussFit1,Starting,options,DT,DS(a,:)); newmatrix(a,1)=Estimate(1); newmatrix(a,2)=Estimate(2); newmatrix(a,3)=Estimate(3); end clf; for b=1:length(SelectedPixels) subplot(length(SelectedPixels)/2,2,b); plot(DT,DS(SelectedPixels(b),:),'ro','markersize',4); xlabel('Delay Time (ps)'); ylabel('\Delta OD (mOD)'); title(['pixel',num2str(SelectedPixels(b))]); hold on; for c=1:length(DT) y(c)=newmatrix(SelectedPixels(b),3)*exp(-(DT(c)- newmatrix(SelectedPixels(b),2))^2/(2*newmatrix(SelectedPixels(b),1)^2)); end plot(DT,y,'b-'); hold on; end 255 TPAstruct=DataStruct; TPAstruct.TimeZeroCutInfo=newmatrix; hold off ContinousTPAGaussFit1.m: function sse=ContinousTPAGaussFit1(parameter,Input,ActualOutput) sigma=parameter(1); x0=parameter(2); Amp=parameter(3); for b=1:length(Input) FittedCurve(b)=Amp*exp(-(Input(b)-x0)^2/(2*sigma^2)); end ErrorVector=FittedCurve-ActualOutput; sse=sum(ErrorVector.^2); 256 Figure C.1 Flowchart for “CalTPACrossSection.m”. 257 width Figure C.2 Variable integration window defined by the user (red lines), and the integration windows with different upper and lower limits compared to the user-defined limits (green lines). The width is defined by the user if testing integration limits is desirable, then the MATLAB script will automatically create a series of integration limits which are wider or narrower, as defined by the width (increment). 258 Figure C.3 Convergence tests at various wavelengths across the probe spectrum (chosen by user). The integrated area of the 2PA “spikes” were plotted as a function of the integration limit defined in Figure E.2 (e.g., -8 means the integration window is 8 fs narrower than the initially defined limit, whereas +8 means 8 fs wider than the initial limit). 259 Figure C.4 Final output screen of “CalTPACrossSection.m”. The 2PA spectrum of water (227 nm pump + UV probe) was plotted against probe wavelength (nm), 1-photon equivalent wavelength (nm), and total excitation energy (eV). The blue marks on the last screen represent the percent error associated with the integration limit. 260 Appendix D. Knife-edge Technique for Measuring Spot Sizes Traditional spot size measurements with a pin hole can generate serious errors when the spot size of laser beam is approximately the same as the diameter of the pin hole. To circumvent this problem, knife-edge technique is used in all spot size measurements in our laboratory. In this technique, a razor blade is placed at the position of interest, in the case of 2PA experiments, where the pump and probe beams overlap (usually the probe beam is at the focus). The razor blade (mounted on a translation stage) is then translated across the beams and the transmitted light intensity is recorded on a photodiode or a power meter. The transmitted light intensity can be plotted against the razor blade position (mounted on a translation stage with a resolution better than 10 μm), and this raw data can be fitted with a complementary error function (erfc), from which the full- width-at-half-maximum (FWHM) can be extracted (see Figure D.1). Figure D.1. Illustration of the knife-edge technique. The sharp edge of the razor blade is scanned across the cross section of the laser beam and the remaining light intensity is recorded at each increment of the edge position. The resulted intensity curve as a function of edge position is fitted by a complementary error function and the FWHM is extracted. 261 Appendix E. Origin of the Polarization Dependence The two-photon absorption cross section (σ 2PA ) can be expressed in the following form: 2 2 4 2 ) , ( ) ( ) ( ) 2 ( μ λ ν ν ν ν π σ μ λ μ λ of PA S g ch e ⋅ + ⋅ = where ν λ and ν μ are the frequencies of the two incoming photons, g(ν λ + ν μ ) is a normalized line shape function and a quantity S of , which is a certain sum over all molecular states: ∑         − ⋅ ⋅ + − ⋅ ⋅ = i i i of k k k k S μ λ μ λ ) )( ( ) )( ( ) , ( λ P P μ μ P P λ if oi if oi λ and μ are polarization vectors of the two incoming photons in lab coordinate; P oi and P if are transition vectors in molecular coordinate; the value k i denotes the energy of the intermediate state. Upon transformation from molecular coordinate to lab coordinate, the following expression is obtained. ) ]( )[ ( ) , ( * * 2 of of S R B A S R B A of S S l l l l S ρσ αβ σ ρ β α μ λ μ λ μ λ δ = ∝ The first factor contains the polarization of the two photons. The quantity S in the last factor is a tensor of second rank, whose elements are functions of two transition moments induced by the two incoming photons; the middle factor has orientation information and therefore need to be averaged for randomly oriented molecules (angle averaging). The final result can be simplified to: H G F H G F PA σ σ σ σ + + = 2 F, G and H are functions of polarization vectors (see table E.1); σ F , σ G and σ H are certain sums of the elements in the tensor S (see Figure E.1), which are functions of two 262 transition moments induced by the two incoming photons, as well as the energies of the photons and the intermediate state. Table E.1. F, G and H values for some selected polarization combinations. Case Parallel Perpendicular Circular, same direction Circular, opposite direction F 2 -1 -2 3 G 2 4 3 3 H 2 -1 3 -2           = zz zy zx yz yy yx xz xy xx S S S S S S S S S S * * 2 βα αβ αβ αβ αα σ σ σ S S S S S H G F = = = Figure E.1. Two-photon tensor (left) and σs (right). σ F : absolute square of the trace of the tensor; σ G : sum of the absolute squares of all nine tensor elements; σ H : sum of the nine products formed by taking each element times the complex conjugate of its transpose. In general, transitions with different symmetries usually exhibit different polarization ratios (e.g., the value of σ par /σ perp is different, where σ par is the 2PA cross section measured with two laser beams having the same polarization, whereas σ perp is the 2PA cross section measured with the polarizations of the two laser beams being perpendicular to each other). It is generally the case that for a given transition, the absolute 2PA cross sections and thus the polarization ratios will change as a function of the energies of the two incoming photons. Therefore, the prediction of the absolute 2PA cross section and the polarization ratio in a most general case requires performing extensive quantum mechanical calculations. Nevertheless, under some special circumstances, the properties of σ F , σ G and σ H , or even some polarization ratios, can be predicted from symmetry consideration alone, 263 without taking into account the photon energies. McClain and co-workers have identified two important symmetry rules pertaining to 2PA σ values. (i) for transitions that lower the symmetry of the electronic wavefunction, σ F = 0; for those preserving electronic symmetry, σ F > 0 and (ii) for transitions preserving electronic wavefunction symmetry or in cases where the two photons have the same frequency (degenerate) regardless of symmetries involved in the transition, σ G = σ H . Using these rules, one can conclude that for any given point group, in degenerate experiments, the value of σ par /σ perp for non- totally symmetric transitions (assuming the ground state is totally symmetric) is always 4/3. Moreover, in the case where one of the photon has energy very close to that of a molecular eigenstate (near resonance), σ F = σ G . This leads to the conclusion that for a non-totally symmetric transition, σ par /σ perp is always 1/2; whereas that for a totally symmetric transition is always 3. This means that the symmetry of the excited state can be definitively determined by using linearly polarized light only, no circular light is required. The σ values and polarization ratios for some special cases mentioned above are summarized in Table E.2. Table E.2. σ values and polarization ratios for some special cases. Symmetry General Degenerate Near-resonance Totally symmetric σ σ F > 0, σ G = σ H σ F > 0, σ G = σ H σ F = σ G = σ H σ par / σ perp ⁄ >4/3 3 σ par / σ same >2/3 >2/3 >2/3 σ perp / σ same 1/2 1/2 1/2 σ same / σ opp ⁄ ⁄ 1 Non-totally symmetric σ σ F = 0 σ F = 0, σ G = σ H σ F = σ H = 0 σ par / σ perp ⁄ 4/3 1/2 σ par / σ same 2/3 2/3 2/3 σ perp / σ same ⁄ ⁄ ⁄ σ same / σ opp ⁄ ⁄ 1 σ par : parallel; σ perp : perpendicular; σ same : circular with same direction; σ opp : circular with opposite direction; “⁄”: variable ratios. 264 The σ par /σ perp values summarized in Table E.2 shows that from symmetry considerations alone, the totally symmetric transition can be distinguished from the non-totally symmetric cases if one of the two photons is near-resonant with a molecular eigenstate; any other polarization combinations does not result in a constant polarization ratio or results in a constant ratio for only one polarization combination. 265 Appendix F. Silicon Photodiode Array Interface and Parts List † Figure F.1 Home-built timer for interfacing the 256-channel Silicon photodiode array with A/D converter. The red lines are printed on the top layer, whereas the green lines are printed on the bottom layer. All electronic components are on the top layer, with exception of one 10 μF capacitor. Note that the figure shows both the top and bottom side of the printed circuit board (PCB). The latter is printed by ExpressPCB (www.expresspcb.com) – please refer to the original .pcb file in case the timer needs reconstruction. “LASER” is the input of the 1 kHz signal from the laser SDG box. “AD” is the “CL Clock” signal provided to the A/D board. The following table summarized the detail parts list for constructing such a home-built timer. Also included in the table are the photodiode array, power supply, A/D boards, and miscellaneous electronic parts essential for the detection of the broadband continuum, as well as the key optical components for continuum generation and dispersion. † Designed by Frank Niertit, assembled by Chris Elles. 266 Table F.1 Parts list for broadband continuum detection system with Silicon PDA. Optical Setup (key components only) Part # Manufacturer qty Description Notes 020111-1 Koch Crystal Finishing 1 CaF2 excimer windows, 1 inch diameter, 2 mm thickness, crystal axes ‹111›±3º orientation, finish 20/10 S/D, surface λ/4 TWF For continuum generation V8037-101 Janos Technology 1 off-axis parabolic mirror (aluminum protected), 25.4 mm diameter, 25.4mm parent focal length, polished to better than 100nm flatness For collimating the continuum V8037-187 Janos Technology 1 off-axis parabolic mirror (aluminum protected), 25.4 mm diameter, 101.6mm parent focal length, polished to better than 100nm flatness For focusing the continuum 77480 Newport: Oriel Instruments 1 Imaging Spectrograph, 1/8 m, Single Grating, Oriel MS127i 77495 Newport: Oriel Instruments 1 MS127i Ruled Grating, 300 lines/mm, 300nm Blaze, 190-450nm Primary UV grating 77939-3723 Newport: Oriel Instruments 1 Flat Grating, Ruled, 300 lines/mm, 500nm Blaze, 250-1150nm Primary, 30x30x3mm Visible grating 77478 Newport: Oriel Instruments 1 MS127i Ruled Grating, 300 lines/mm, 1000nm Blaze, 575-2500nm Primary NIR grating Silicon PDA Part # Manufacturer qty Description Notes S3901-256Q Hamamatsu 1 NMOS linear Image Sensor 256 Pixels 2 for dual array setup C7884-20 Hamamatsu 1 Multichannel Detector Head for above chip 2 for dual array setup A8226 Hamamatsu 1 Cables for used with above detector head 2 for dual array setup A/D Converter Part # Manufacturer qty Description Notes PD2-MFS-4- 500/16 United Electronic Industries (UEI) 1 500 kS/s, 16-bit resolution, simultaneous sampling PCI multifunction board “fast” board, see footnote ‡ PD2-MFS-4- DG4 United Electronic Industries (UEI) 1 Upgrade 4 single-ended inputs to 4 differential inputs with gains (1, 2, 5, 10) for above board PD-64K-FIFO United Electronic Industries (UEI) 1 Upgrade 1K FIFO to 64K FIFO for above board PDL-MF United Electronic Industries (UEI) 1 50 kS/s, 16-bit resolution, PCI multifunction board “slow” board ‡ The current board includes a chipset upgrade to higher sampling rate 520 kS/s, in order to handle dual 256-channel diode arrays (512 kS/s required). This is a relatively inexpensive upgrade, but UEI does not seem to have this upgrade anymore. 267 Table F.1 continued Breakout Boards and Cables for A/D Converter Part # Manufacturer qty Description Notes PD-STP-9616 United Electronic Industries (UEI) 1 16-channel screw terminal panel with 37-pin, 80-pin and 96-pin connectors Breakout terminal for “fast” board PD-CBL-96 United Electronic Industries (UEI) 1 1m. 96-way, round, shielded cable with molded cover. Cable 1 for above terminal PD-CBL-37TP United Electronic Industries (UEI) 1 37 way D-sub Twisted Pair cable, DIO set w/ bracket Cable 2 for above terminal PDL-STP United Electronic Industries (UEI) 1 Screw-terminal panel for PDL-MF multifunction boards Breakout terminal for “slow” board PDL-CBL-100 United Electronic Industries (UEI) 1 100-way, 1m Y-split flat ribbon cable for PDL- MF boards Cable for above terminal Components for Silicon PDA timer Part# (Mouser) Manufacturer qty Description Notes ATV2500BQ Atmel 1 High-speed high-density UV erasable programmable logic device 595- CD74AC14E Texas Instruments (CD74AC14E) 1 Logic - Buffers, Drivers, and Transceivers Hex Schmitt Triggered Inverters 771- 74HC541N NXP (74HC541N) 1 Buffers & Line Drivers OCTAL BUFF/LDRVR 3ST Obsolete, possible replacement: 771- 74HC541N652 815-ACO-32- EK Abracon (ACO- 32.000MHZ- EK) 1 Standard Clock Oscillators 32MHz 5V 660- CF1/4L680J KOA Speer (CF1/4L680J) 8 1/4Watt Axial Leaded Carbon Film Resistors 68 ohms, 5% tolerance 660- CF1/4L563J KOA Speer (CF1/4L563J) 1 1/4Watt Axial Leaded Carbon Film Resistors 56K ohms, 5% tolerance 660- CF1/4L201J KOA Speer (CF1/4L201J) 8 1/4Watt Axial Leaded Carbon Film Resistors 200 ohms, 5% tolerance 660- CF1/4L102J KOA Speer (CF1/4L102J) 8 1/4Watt Axial Leaded Carbon Film Resistors 1K ohms, 5% tolerance 660- CF1/4L101J KOA Speer (CF1/4L101J) 8 1/4Watt Axial Leaded Carbon Film Resistors 100 ohms, 5% tolerance 660- CF1/4L301J KOA Speer (CF1/4L301J) 1 1/4Watt Axial Leaded Carbon Film Resistors 300 ohms, 5% tolerance 512-1N914B Fairchild Semiconductor (1N914B) 20 Rectifiers, 100V Io/200mA BULK 268 Table F.1 continued Components for Silicon PDA timer (Cont’d) Part# (Mouser) Manufacturer qty Description Notes 581- TAP476K006 SCS AVX (TAP476K006S CS) 1 Radial Tantalum Capacitors, Bulk Packaging, 6.3V, 47μF, 10% tolerance 581- TAP106K035 SCS AVX (TAP106K035S CS) 5 Radial Tantalum Capacitors, Bulk Packaging, 35V, 10μF, 10% tolerance 581- TAP106K016 SCS AVX (TAP106K016S CS) 1 Radial Tantalum Capacitors, Bulk Packaging, 16V, 10μF, 10% tolerance 75- 515D108M035 DK6AE3 Vishay/Sprague (515D108M035 DK6AE3) 1 Radial Aluminum Electrolytic Capacitors - 85 Degree, 1000μF, 35volts, 20% tolerance 50R7637 (Newark) AVX (CK05BX470K) 6 Ceramic Capacitor, 200V, 47 pF (old part only labeled 47 pF, no further information was given) For preventing very sharp rise and fall of signal. 675-HTAA- 16WA Condor/SL Power (HTAA- 16WA) 1 Linear and Switching Power Supplies 5V@2A +/-12/12V@.4A Obsolete, possible replacement: 675- HTAA-16WA-G 575-113640 Mill-Max (110- 13-640-41- 001000) 1 DIP Low Profile Sockets 40P GLD PIN GLD CONT Socket for programmable chip 575-113320 Mill-Max (110- 13-320-41- 001000) 1 DIP Low Profile Sockets 20P GLD PIN GLD CONT 575-113314 Mill-Max(110- 13-314-41- 001000) 1 DIP Low Profile Sockets 14P GLD PIN GLD CONT 535-1107741 Aries (1107741) 1 DIP/SIP Sockets OSCILLATOR FULL 4PIN Socket for 32 MHz oscillator 571-17969490 TE Connectivity (1-796949-0) 1 Eurostyle Terminal Blocks 5.08MM VERTICAL 10P 571-7969498 TE Connectivity (796949-8) 2 Eurostyle Terminal Blocks 5.08MM VERTICAL 8P 571-7969494 TE Connectivity (796949-4) 1 Eurostyle Terminal Blocks 5.08MM VERTICAL 4P 269 Appendix G. Schematics and Control Program (programmed into ATV2500BQ chip) for Silicon PDA timer § § Designed and written by Frank Niertit. 270 Name CCD-timer ; PartNo ATV2500B ; Date 11/02/2004 ; Revision 050218 ; Designer Frank J Niertit ; Company USC Chemistry Dept. ; Assembly ; Location ; Device V2500B ; /*########################################################################## pin out and node numbers ############################################################################ +-------+ clk (input) >--|1 40|--< (input) !rst !laser (input) >--|2 39|--< (input) GND !run (input) >--|3 38|--< (input) GND NC (io) <--|4 37|--< (input) GND NC (io) <--|5 36|--> (io) NC NC (io) <--|6 35|--> (io) NC NC (io) <--|7 34|--> (io) NC NC (io) <--|8 33|--> (io) NC NC (io) <--|9 32|--> (output) clk1 VCC ---|10 31|--> (output) clk0 NC (io) <--|11 30|--- GND NC (io) <--|12 29|--> (output) start0 NC (io) <--|13 28|--> (output) start1 NC (io) <--|14 27|--> (output) atod NC (io) <--|15 26|--> (output) read NC (io) <--|16 25|--> (output) error !tr0 (input) >--|17 24|--> (output) sync eos0 (input) >--|18 23|--< (input) GND !tr1 (input) >--|19 22|--< (input) GND eos1 (input) >--|20 21|--< (input) GND +-------+ ---------------------------------------------------------------------------- pin pin node node pin pin node node input I/O Q1 Q2 input I/O Q1 Q2 01 04 65 41 40 36 88 64 02 05 66 42 39 35 87 63 03 06 67 43 38 34 86 62 07 68 44 37 33 85 61 08 69 45 32 84 60 09 70 46 31 83 59 11 71 47 29 82 58 12 72 48 28 81 57 17 13 73 49 27 80 56 18 14 74 50 23 26 89 55 19 15 75 51 22 25 78 54 20 16 76 52 21 24 77 53 ##########################################################################*/ 271 /*########################################################################## input input pin pin 01 <- _iclk 40 <- !_irst 02 <- !_laser 39 <- GND 03 <- !_run 38 <- GND 37 <- GND 17 <- !_itr0 18 <- _ieos0 23 <- GND 19 <- !_itr1 22 <- GND 20 <- _ieos1 21 <- GND ---------------------------------------------------------------------------- pin node node I/O Q1 Q2 04 -> _o0 65 -> x0 41 -> q0 05 -> _o1 66 -> x1 42 -> q1 06 -> _o2 67 -> x2 43 -> q2 07 -> _o3 68 -> x3 44 -> q3 08 -> _o4 69 -> x4 45 -> q4 09 -> _o5 70 -> NC 46 -> q5 11 -> _o6 71 -> NC 47 -> q6 12 -> _o7 72 -> NC 48 -> tr 13 -> _o8 73 -> NC 49 -> start 14 -> _o9 74 -> NC 50 -> read 15 -> _o10 75 -> NC 51 -> eos 16 -> _o11 76 -> NC 52 -> eosx 36 -> _o12 88 -> NC 64 -> eos1 35 -> _o13 87 -> NC 63 -> eos2 34 -> _o14 86 -> NC 62 -> tr1 33 -> _o15 85 -> NC 61 -> tr0 32 -> _oclk1 84 -> NC 60 -> run 31 -> _oclk0 83 -> NC 59 -> go 29 -> _ostart0 82 -> NC 58 -> g0 28 -> _ostart1 81 -> NC 57 -> g1 27 -> _oatod 80 -> NC 56 -> r3 26 -> _oread 89 -> NC 55 -> r2 25 -> _oerror 78 -> NC 54 -> r1 24 -> _osync 77 -> NC 53 -> r0 ##########################################################################*/ 272 /******************************* INPUT PINS ******************************* input pins 21..23,37..39 are not used and are grounded --------------------------------------------------------------------------*/ PIN 1 = _iclk; /* 32MHz master clock */ PIN 2 = !_irun; /* run signal from computer */ PIN 3 = !_ilaser; /* 1KHz laser clock */ PIN 17 = !_itr0; /* data trigger pulse from channel 0 */ PIN 18 = _ieos0; /* end of scan signal from channel 0 */ PIN 19 = !_itr1; /* data trigger pulse from channel 1 */ PIN 20 = _ieos1; /* end of scan signal from channel 1 */ PIN 40 = !_irst; /* power on hold off */ /****************************** OUTPUT PINS ******************************** output pins 4..9,11..16,33..36 are not used (leave unconnected) --------------------------------------------------------------------------*/ PIN 32 = _oclk1; /* 2MHz data clock to channel0 */ PIN 31 = _oclk0; /* 2MHz data clock to channel1 */ PIN 29 = _ostart0; /* output start pulse to channel0 */ PIN 28 = _ostart1; /* output start pulse to channel1 */ PIN 27 = _oatod; /* conversion clock to A to D */ PIN 26 = _oread; /* system is reading */ PIN 25 = _oerror; /* error detected */ PIN 24 = _osync; /* output of laser pulse */ PIN [9..4] = [_o5.._o0]; /* unused outputs (not connected) */ PIN [16..11] = [_o11.._o6]; /* unused outputs (not connected) */ PIN [33..36] = [_o15.._o12]; /* unused outputs (not connected) */ /******************************* PINNODES *********************************/ PINNODE [47..41] = [q6..q0]; /* internal grey code sequencer */ PINNODE [52..48] = [eosx,eos,read,start,tr]; /* internal logic terms */ PINNODE [56..53] = [r3..r0]; /* initial reset */ PINNODE [64..57] = [eos0,eos1,tr1,tr0,run,go,g0,g1]; /* input logic */ PINNODE [69..65] = [e4..e0]; /* error reset of read */ /*------------------------------------------------------------------------*/ $DEFINE ON 'h'ffffffff $DEFINE OFF 'h'00000000 /**************************************************************************/ /**************************************************************************/ FIELD outs = [_o15.._o0]; /* unconnected output pins */ FIELD cnt = [q6..q0]; /* internal grey code sequencer */ FIELD rst = [r3..r0]; /* initial reset */ FIELD err = [e4..e0]; /* error reset of read */ /*------------------------------------------------------------------------*/ outs = OFF; outs.oe = ON; /*------------------------------------------------------------------------*/ rst.ce = ON; rst.ar = _irst; rst.sp = OFF; 273 r0.d = rst:'b'000x # rst:'b'011x; r1.d = rst:'b'00x1 # rst:'b'0x10; r2.d = rst:'b'0x10 # rst:'b'01xx; r3.d = rst:'b'x100 # rst:'b'1xxx; /*------------------------------------------------------------------------*/ err.ck = q3; err.ar = tr; err.sp = OFF; e0.d = read & (err:'b'0000x # err:'b'0011x # err:'b'0110x # err:'b'0101x); e1.d = read & (err:'b'000x1 # err:'b'0xx10 # err:'b'011x1 ); e2.d = read & (err:'b'00x10 # err:'b'0x1x1 # err:'b'0x10x ); e3.d = read & (err:'b'0x100 # err:'b'01xxx); e4.d = read & (err:'b'x1000 # err:'b'1xxxx); /*------------------------------------------------------------------------*/ cnt.ce = ON; cnt.ar = !r3; cnt.sp = OFF; q0.d = cnt:'b'xxx000x # cnt:'b'xxx011x # cnt:'b'xxx101x # cnt:'b'xxx110x; q1.d = cnt:'b'0101000 /* sequence change start of conversion */ # cnt:'b'xxx00x1 # cnt:'b'xxxxx10 # cnt:'b'xxx11x1; q2.t = cnt:'b'0101000 /* sequence change start of conversion */ # cnt:'b'xxx0010 # cnt:'b'xxx1110; q3.t = cnt:'b'xxx0100 # cnt:'b'xxx1000; q4.t = cnt:'b'0001000 & tr /* sequence change start of transfer */ # cnt:'b'0111000 # cnt:'b'1x1xxxx; q5.t = cnt:'b'0011000 # cnt:'b'1100000; q6.t = cnt:'b'0101000 # cnt:'b'1000110; /*------------------------------------------------------------------------*/ 274 /*------------------------------------------------------------------------*/ /* Internal Logic */ /*------------------------------------------------------------------------*/ tr.d = tr0 & tr1 & cnt:'b'000x0xx # tr & !q4; go.d = run & !start & ( g1 # go ); start.d = go & !read & cnt:'h'02 # start & ( !q2 # !read ); read.d = start & !q2 # read & ( !e4 & !eos # start ); eos.d = read & ( eos0 & eos1 & !eosx # eos ); eosx.d = eos # eosx & ( eos0 # eos1 ); [tr,go,start,read,eos,eosx].ce = ON; [tr,go,start,read,eos,eosx].ar = !r3; [tr,go,start,read,eos,eosx].sp = OFF; /*------------------------------------------------------------------------*/ /* Input logic */ /*------------------------------------------------------------------------*/ g0.d = ON; g0.ck = _ilaser; g0.ar = g1; g1.d = g0; g1.ar = OFF; run.d = _irun & cnt:'b'0001000 # run & (_irun # !q3); tr0.d = _itr0 & cnt:'b'0001xxx # tr0 & (!q6 # q5); tr1.d = _itr1 & cnt:'b'0001xxx # tr1 & (!q6 # q5); eos0.d = _ieos0 & cnt:'h'0A # eos0 & ( q2 # _ieos0 ); eos1.d = _ieos1 & cnt:'h'0A # eos1 & ( q2 # _ieos1 ); [g1,run,tr0,tr1,eos0,eos1].ce = ON; [run,tr0,tr1,eos0,eos1].ar = !r3; [g0,g1,run,tr0,tr1,eos0,eos1].sp = OFF; /*------------------------------------------------------------------------*/ /* Output logic */ /*------------------------------------------------------------------------*/ _oclk0 = cnt:'b'x00x1xx & ( !tr0 # _oclk0 ); _oclk1 = cnt:'b'x00x1xx & ( !tr1 # _oclk1 ); _ostart0 = start & _oclk0 # _ostart0 & !_oclk0; _ostart1 = start & _oclk1 # _ostart1 & !_oclk1; _oatod = cnt:'b'11xxxxx # cnt:'b'1xxxx0x & _oatod; 275 _oread = read; _oerror = (e4 # _oerror) & run; _osync = g1; [_oclk0,_oclk1,_ostart0,_ostart1,_oatod,_oread,_oerror,_osync].oe = ON; /*------------------------------------------------------------------------*/ 276 Appendix H. InGaAs Photodiode Array Interface and Parts List ** Figure H.1 Top layer of the home-built timer for InGaAs array. The red areas represent copper. This layer mainly holds the programmable chips and buffers, with few resistors such as all 100 Ω and some 68 Ω. Different from the Silicon timer, instead of through-hole mounts, this board uses surface mounts for resistors, capacitors and diodes. See Figure H.3 and Table H.1 for detail. ** Designed by Frank Niertit, assembled by Yuyuan Zhang 277 Figure H.2 Bottom layer of the home-built timer for InGaAs array. The green areas represent copper. This layer holds all capacitors, diodes, and most resistors. The connection to the power supply board is also on this layer. 278 Figure H.3 Component layout for the timer board (both top and bottom layers are shown). The PCB is printed by ExpressPCB (www.expresspcb.com) – please refer to the original .pcb file in case the timer needs reconstruction. Refer to the following table for the details of each component. 279 Figure H.4 Top layer of the home-built power supply for InGaAs array detection system. This power board holds a series of DC-DC converters, which will power both the timer board and the InGaAs photodiode array. The red area represents copper and the pats are for surface type capacitors. 280 Figure H.5 Bottom layer of the home-built power supply for InGaAs array detection system. The green area represents copper. 281 Figure H.6 Component layout for the timer board (both top and bottom layers are shown). The PCB is printed by ExpressPCB (www.expresspcb.com) – please refer to the original .pcb file in case the timer needs reconstruction. Refer to the following table for the details of each component. 282 Table H.1 Parts list for broadband continuum detection system with InGaAs PDA. †† InGaAs Photodiode Array Part # Manufacturer qty Description Notes G9213-256S Hamamatsu 1 InGaAs linear image sensor, 256 pixels C8061-01 Hamamatsu 1 Multichannel detector head with TE cooler for above InGaAs sensor Components for InGaAs PDA timer Part# (Digi-Key) Manufacturer qty Description Notes ATF2500C Atmel 1 High-speed high-density UV erasable programmable logic device 2 programmable chips required for 1 board 296-1619-5- ND Texas Instr. (SN74HCT541 N) 4 Logic - Buffers, Drivers, and Transceivers Hex Schmitt Triggered Inverters, 8 bit, 20 dip LTC1099CN# PBF-ND Linear Tech. (LTC1099CN#P BF) 1 A/D converter, 8 bit, high-speed, 20 dip For temperature readout 576-1046-1- ND Micrel (LM4040CYM3 -2.5 TR) 1 IC VREF SHUNT PREC 2.5V SOT-23-3 For temperature readout SER1111-ND Epson Toyocom (SG-531PH 32.0000MC) 1 Oscillator, 32.0000 MHz, pdip “master clock” 641-1005-1- ND Comchip Tech. (CDSF355) 32 Diode switching, 80V 100mA ED3308-ND Mill-Max (110- 93-308-41- 001000) 1 Socket, 8 ps, 0.300" dip, gold Socket for “master clock” ED3640-ND Mill-Max (110-93-640-41- 001000) 2 40 pin socket, gold, 0.600 Sockets for programmable chips ED3320-ND Mill-Max (110-93-320-41- 001000) 5 20 pin socket, gold, 0.300" dip Sockets for buffers & temp control A/D A32073-ND Tyco (5747236- 4) 1 15 pin, D-sub, male pins, gold Connectors to InGaAs array head A32072-ND Tyco (5747250-4) 1 9 pin, D-sub, males pins, gold Connectors to TE cooler in array head †† The optical setup for NIR detection is similar to the UV-Visible detection, with the exception that a sapphire is required to generate NIR up to ~1400 nm. The breakout boards and A/D converters are also identical to those used for UV-Visible detection. Thus, the parts list presented here only include the InGaAs array and the timer. 283 Table H.1 Continued Components for InGaAs PDA timer (cont’d) Part# (Digi-Key) Manufacturer qty Description Notes MHB60K-ND 3M (2560-6002UB) 1 shrouded header 60 pos straight for ribbon cable connecting timer to breakout box WM4210-ND Molex/Waldom (22-23-2121) 1 connector header 12pos, 0.100 vertical, tin for connecting the timer and power supply WM4201-ND Molex/Waldom (22-23-2031) 1 connector header 3pos, 0.100, vertical, tin RHM1.00KCC T-ND Rohm (MCR10EZHF1 001) 17 Resistor, 1.00k ohm, 1/8W, 1% tolerance, 0805 SMD RHM200CCT- ND Rohm (MCR10EZHF2 000) 17 Resistor, 200 ohm, 1/8W, 1% tolerance, 0805 SMD RHM100CCT- ND Rohm (MCR10EZHF1 000) 16 Resistor, 100 ohm, 1/8W, 1% tolerance, 0805 SMD 15 on top layer, 1 on bottom. RHM68.0CCT -ND Rohm (MCR10EZHF6 8R0) 17 Resistor, 68.0 ohm, 1/8W,1% tolerance, 0805 SMD 4 on top layer, 13 on bottom. 718-1391-1- ND Vishay/Sprague (597D108X96R 3R2T) 6 Tantalum Capacitor, 1000 μF, 6.3V, 10% tolerance 587-1963-1- ND Taiyo Yuden (JMK316BJ107 ML-T) 13 Ceramic capacitor, 100 μF, 6.3V, 20% tolerance 709-1175-1- ND Johanson (500R15N470J V4T) 17 Ceramic capacitor, 47 pF, 50V, 5% tolerance 288-1348-ND American Electrical(DSU B-9-MF-CBL) 1 Cable interface module, D-sub, 9 pin Cable connects timer to TE cooler 288-1344-ND American Electrical(DSU B-15-MF-CBL) 1 Cable interface module, D-sub, 15 pin Cable connects timer to detector head Power Supplies Part# (Newark) Manufacturer qty Description Notes EPS355-ND (Digi-Key) Iccnexergy (FW3012-760F) 1 Desktop power supply, 12V, 30W Converts AC to 12V DC 718-1505-1- ND (Digi-Key) Vishay/Sprague (TR3E107K020 C0100) 14 Tantalum capacitor, 100 μF, 20V, 10% tolerance Different from 100 μF for timer 284 Table H.1 Continued ‡‡ Power Supplies (cont’d) Part# (Newark) Manufacturer qty Description Notes 61K2980 Murata (NDY1212C) 1 DC/DC Converter. Input Voltage: 12 V, Output Voltage: 12V DC, Output Current: 250mA, Power rating: 3W for cooler 05M6950 TDK Lambda 1 DC/DC Converter. Input Voltage: 12 V, Output Voltage: 5V DC, Output Current: 2A, Power rating: 10W for cooler 61K2978 Murata (NDY1205C) 1 DC/DC Converter. Input Voltage: 12 V, Output Voltage: 5V DC, Output Current: 600mA, Power rating: 3W for timer 61K2981 Murata (NDY1215C) 1 DC/DC Converter. Input Voltage: 12 V, Output Voltage: 15V DC, Output Current: 200mA, Power rating: 3W for array 61K3055 Murata (NKE1215DC) 1 DC/DC Converter. Input Voltage: 12 V, Output Voltage: 15V DC, Output Current: 66mA, Power rating: 1W # unit failed often, use below (for array) # 51R5308 Traco Power (TMA1215S) 1 DC/DC Converter. Input Voltage: 12 V, Output Voltage: 15V DC, Output Current: 65mA, Power rating: 1W Dimension doesn’t fit, but more stable 61K3049 Murata (NKE1205DC) 2 DC/DC Converter. Input Voltage: 12 V, Output Voltage: 5V DC, Output Current: 200mA, Power rating: 1W 1 for cooler, 1 for array ‡‡ Miscellaneous items such as components needed to construct customized cables are not summarized here. 285 Appendix I. Schematics and Control Program for InGaAs PDA timer §§ §§ Designed and written by Frank Niertit. The squares on the right hand side denotes the 60-pin connection to the A/D cards, the top left pin is #1 and top right pin is #2, whereas the bottom left pin is #59 and bottom right pin is #60. 286 Name InGaAs-v2 Interface; PartNo ATF2500C; Date 5/7/2009; Revision 0; Designer Frank J Niertit ; Company USC Chemistry Dept. ; Assembly U1; Location Left side; Device v2500C ; /*########################################################################## pinout and node numbers ############################################################################ ATF2500C +-------+ osc >--|1 40|--< reg3 {U2 pin4 output} {InGaAs output !eos} eos >--|2 39|--< reg2 {U2 pin5 output} {InGaAs output trigger} trig >--|3 38|--< reg1 {U2 pin6 output} {InGaAs input clock} clock <--|4 37|--< reg0 {U2 pin7 output} {InGaAs input start} start <--|5 36|--> scan {U2 pin2 input} {zero} io19 <--|6 35|--> samp {U2 pin3 input} {scla} io18 <--|7 34|--> io9 {sclb} io17 <--|8 33|--> io8 {hold} io16 <--|9 32|--> io7 VCC ---|10 31|--> io6 {s4} io15 <--|11 30|--- GND {s3} io14 <--|12 29|--> io5 {s2} io13 <--|13 28|--> io4 {s1} io12 <--|14 27|--> io3 {s0} io11 <--|15 26|--> io2 {U2 pin 14} io10 <--|16 25|--> io1 {U2 pin 15} {U2 pin9 output} nin0 >--|17 24|--> io0 {U2 pin 16} {U2 pin11 output} clck >--|18 23|--< enabl {computer output} {U2 pin12 output} cary >--|19 22|--< clear {computer output} {U2 pin13 output} strt >--|20 21|--< laser {computer output} +-------+ ---------------------------------------------------------------------------- pin pin node node pin pin node node input I/O Q1 Q2 input I/O Q1 Q2 01 04 65 41 40 36 88 64 02 05 66 42 39 35 87 63 03 06 67 43 38 34 86 62 07 68 44 37 33 85 61 08 69 45 32 84 60 09 70 46 31 83 59 11 71 47 29 82 58 12 72 48 28 81 57 17 13 73 49 17 27 80 56 18 14 74 50 18 26 79 55 19 15 75 51 19 25 78 54 20 16 76 52 20 24 77 53 ############################################################################ 287 ############################################################################ input pins input_pins 01 <- osc 40 <- reg3 {U2_4) 02 <- eos {InGaAs} 39 <- reg2 (U2_5} 03 <- trig {InGaAs} 38 <- reg1 {U2_6} 37 <- reg0 {U2_7} 17 <- nin0 {U2_9) 18 <- clck (U2_11} 23 <- enabl {computer} 19 <- cary {U2_12} 22 <- clear {computer} 20 <- strt {U2_13} 21 <- laser {computer} ---------------------------------------------------------------------------- io_pins node node 04 -> clock {InGaAs} 65 -> Q1 clock.t 41 -> Q2 no0.d 05 -> start {InGaAs} 66 -> Q1 start.t 42 -> Q2 no1.d 06 -> io19 {zero} 67 -> Q1 strte.t 43 -> Q2 no2.d 07 -> io18 {scla} 68 -> Q1 hold.t 44 -> Q2 no3.d 08 -> io17 {sclb} 69 -> Q1 sclb.d 45 -> Q2 no4.d 09 -> io16 {hold} 70 -> Q1 scla.d 46 -> Q2 no5.d 11 -> io15 {s4} 71 -> Q1 zero.d 47 -> Q2 no6.d 12 -> io14 {s3} 72 -> Q1 cl0.t 48 -> Q2 no7.d 13 -> io13 {s2} 73 -> Q1 cl1.t 49 -> Q2 no8.d 14 -> io12 {s1} 74 -> Q1 cl2.t 50 -> Q2 no9.d 15 -> io11 {s0} 75 -> Q1 cl3.t 51 -> Q2 no10.d 16 -> io10 76 -> Q1 cl4.t 52 -> Q2 no11.d 36 -> scan {U2_2} 88 -> Q1 scan.t 64 -> Q2 no23.d 35 -> samp {u2_3} 87 -> Q1 samp.t 63 -> Q2 no22.d 34 -> io9 86 -> Q1 s0.t 62 -> Q2 no21.d 33 -> io8 85 -> Q1 s1.t 61 -> Q2 no20.d 32 -> io7 84 -> Q1 s2.t 60 -> Q2 no19.d 31 -> io6 83 -> Q1 s3.t 59 -> Q2 no18.d 29 -> io5 82 -> Q1 s4.t 58 -> Q2 no17.d 28 -> io4 81 -> Q1 no24.d 57 -> Q2 no16.d 27 -> io3 80 -> Q1 no25.d 56 -> Q2 no15.d 26 -> io2 {U2_14} 79 -> Q1 no26.d 55 -> Q2 no14.d 25 -> io1 {U2_15} 78 -> Q1 no27.d 54 -> Q2 no13.d 24 -> io0 {U2_16} 77 -> Q1 no28.d 53 -> Q2 no12.d ########################################################################### 288 ***************************** INPUT PINS *********************************/ /* input from oscillator */ pin 1 = osc; /* 32MHz system clock */ /* inputs from InGaAs array */ pin 2 = eos; /* array scan signal */ pin 3 = trig; /* data sample trig */ /* inputs from U2 */ pin [18..20,40..37] = [clck,cary,strt,reg3..0]; /* 5bit counter values */ pin 17 = U2_9; /* unused input pin */ /* inputs from computer interface */ pin 23 = enabl; /* enable start pulse */ pin 22 = clear; /* reset error signal */ pin 21 = laser; /* laser timing pulse */ /**************************** OUTPUT PINS ********************************/ /* outputs to InGaAs array */ pin 4 = clock; /* variable 4MHz to 250KHz clock */ pin 5 = start; /* 250ns start signal */ /* outputs to period counter */ pin 36 = scan; /* signal analogous to the InGaAs reset signal */ pin 35 = samp; /* 250ns sample pulse */ pin [7..9,11..16,34..31,29..24] = [io18..0]; /* unused output io pins */ /*************************** PINNODES ************************************/ pinnode 67 = strte; /* synchronizes the start pulse */ pinnode [70,69] = [scla,sclb]; /* synchronizes base clock to U2 */ pinnode [86..82,68] = [s0..4,hold]; /* state machine */ pinnode [76..71] = [cl4..0,zero]; /* variable rate clock */ pinnode [41..64,81..77] = [no0..no28]; /* unused nodes */ /************************************************************************** 289 **************************************************************************/ $DEFINE ON 'h'ffffffff $DEFINE OFF 'h'00000000 [clock,start,scan,samp,io19..0].oe = ON; [io19..11] = [zero,scla,sclb,hold,s4..0]; [io10..0] = OFF; [no28..0].ar = ON; [no28..0].ce = OFF; [no28..0].d = OFF; [no28..0].sp = OFF; [clock,start,strte,hold,scan,samp].sp = OFF; [zero,sclb,scla,cl4..0,s4..0].sp = OFF; [scan,clock,scla,sclb,zero,cl4..0].ar = OFF; [samp,hold,s0..4].ar = !scan; [start,strte].ar = scan; [clock,strte,scla,sclb,zero,cl4..0].ce = ON; [start,samp,hold,scan,s4..0].ck = clock; /************************************************************************/ FIELD vcls = [hold,zero,sclb,scla]; FIELD vclr = [cl4..cl0]; /* variable frequency clock generator */ scla.d = !clck & !sclb; /* sychronize the base clocks with U2 */ sclb.d = !clck & scla; load = vcls:'b'1100; decr = vcls:'b'1000; cl0.t = load & (cl0 $ cary) # decr & vclr:'b'xxxx1; cl1.t = load & (cl1 $ reg0) # decr & vclr:'b'xxxx0; cl2.t = load & (cl2 $ reg1) # decr & vclr:'b'xxx00; cl3.t = load & (cl3 $ reg2) # decr & vclr:'b'xx000; cl4.t = load & (cl4 $ reg3) # decr & vclr:'b'x0000; zero.d = !hold # vclr:'h'0 # !cary & [reg3..0]:'h'0; clock.t = vcls:'b'x111; /* state machine */ scan.d = eos # s0 # s1; s0.d = eos & (scan & !s1 # s0 & !s2) # !eos & !s4 & !s2 & s1; s1.d = eos & s0 # s1 & !s0; s2.d = eos & s0 & s1 # !s3 & ( s2 # s4 ) # samp ; s3.d = !samp & !s0 & s2; s4.d = !samp & ( !s4 & s3 & s2 # s4 & (!s3 # !s2) ); hold.d = !samp & eos & s4; samp.d = trig; strte.d = !start & ( strt # strte ); start.d = strte; /************************************************************************/ 290 Name Laser Interface; PartNo ATF2500C ; Date 5/7/2009; Revision 0; Designer Frank J Niertit ; Company USC Chemistry Dept. ; Assembly U2; Location Right side; Device v2500C ; /*########################################################################## pinout and node numbers ############################################################################ ATF2500C +-------+ osc >--|1 40|--< inp7 {computer output} {U1 pin36 output} scan >--|2 39|--< inp6 {computer output} {U1 pin35 output} samp >--|3 38|--< inp5 {computer output} {U1 pin40 input} hq11 <--|4 37|--< inp4 {computer output} {U1 pin39 input} hq10 <--|5 36|--> diag5 { !psb & !psa } {U1 pin38 input} hq9 <--|6 35|--> diag4 { !psb & psa } {U1 pin37 input} hq8 <--|7 34|--> diag3 { psb & psa } io0 <--|8 33|--> diag2 { psb & !psa } {U1 pin17 input} io1 <--|9 32|--> diag1 { las2 } VCC ---|10 31|--> diag0 { las1 } {U1 pin18 output} clck <--|11 30|--- GND {U1 pin19 output} cary <--|12 29|--> sampl {computer input} {U1 pin20 output} strt <--|13 28|--> track {computer input} {U1 pin 26} io2 <--|14 27|--> lockd {computer input} {U1 pin 25} io3 <--|15 26|--> error {computer input} {U1 pin 24} io4 <--|16 25|--> oflow {computer input} {computer output} inp3 >--|17 24|--> io5 {computer output} inp2 >--|18 23|--< enabl {computer output} {computer output} inp1 >--|19 22|--< clear {computer output} {computer output} inp0 >--|20 21|--< laser {computer output} +-------+ ---------------------------------------------------------------------------- pin pin node node pin pin node node input I/O Q1 Q2 input I/O Q1 Q2 01 04 65 41 40 36 88 64 02 05 66 42 39 35 87 63 03 06 67 43 38 34 86 62 07 68 44 37 33 85 61 08 69 45 32 84 60 09 70 46 31 83 59 11 71 47 29 82 58 12 72 48 28 81 57 17 13 73 49 17 27 80 56 18 14 74 50 18 26 79 55 19 15 75 51 19 25 78 54 20 16 76 52 20 24 77 53 ############################################################################ 291 ############################################################################ input pins input_pins 01 <- osc 40 <- inp7 {computer} 02 <- scan {U1_36} 39 <- inp6 {computer} 03 <- samp (U1_35} 38 <- inp5 {computer} 37 <- inp4 {computer} 17 <- inp3 {computer} 18 <- inp2 {computer} 23 <- enabl {computer} 19 <- inp1 {computer} 22 <- clear {computer} 20 <- inp0 {computer} 21 <- laser {computer} ---------------------------------------------------------------------------- io_pins node node 04 -> hq11 {U2_40} 65 -> Q1 hq11 41 -> Q2 hq0 05 -> hq10 {U2_39} 66 -> Q1 hq10 42 -> Q2 hq1 06 -> hq9 {U2_38} 67 -> Q1 hq9 43 -> Q2 hq2 07 -> hq8 {U2_37} 68 -> Q1 hq8 44 -> Q2 hq3 08 -> io0 69 -> Q1 hq7 45 -> Q2 hq4 09 -> io1 {U2_17} 70 -> Q1 decd 46 -> Q2 hq5 11 -> clck {U2_18} 71 -> Q1 clck 47 -> Q2 hq6 12 -> cary {U2_19} 72 -> Q1 cary 48 -> Q2 pq0 13 -> strt {U2_20} 73 -> Q1 strt 49 -> Q2 pq1 14 -> io2 {U2_26} 74 -> Q1 las2 50 -> Q2 pq2 15 -> io3 {U2_25} 75 -> Q1 las1 51 -> Q2 pq3 16 -> io4 {U2_24} 76 -> Q1 las0 52 -> Q2 pq4 36 -> diag5 { 00 } 88 -> Q1 no88 64 -> Q2 dq11 35 -> diag4 { 01 } 87 -> Q1 en1 63 -> Q2 dq10 34 -> diag3 { 11 } 86 -> Q1 en0 62 -> Q2 dq9 33 -> diag2 { 10 } 85 -> Q1 decr 61 -> Q2 dq8 32 -> diag1 {las2 } 84 -> Q1 psb 60 -> Q2 dq7 31 -> diag0 {las1 } 83 -> Q1 psa 59 -> Q2 dq6 29 -> sampl 82 -> Q1 sampl 58 -> Q2 dq5 28 -> track 81 -> Q1 track 57 -> Q2 dq4 27 -> lockd 80 -> Q1 lockd 56 -> Q2 dq3 26 -> error 79 -> Q1 error 55 -> Q2 dq2 25 -> oflow 78 -> Q1 oflow 54 -> Q2 dq1 24 -> io5 77 -> Q1 sampd 53 -> Q2 dq0 ########################################################################### **************************************************************************/ $DEFINE ON 'h'ffffffff $DEFINE OFF 'h'00000000 /* input from oscillator */ pin 1 = osc; /* 32MHz system clock */ /* inputs from U1 */ pin [2,3] = [scan,samp]; 292 /* inputs from computer interface */ pin [23..21,40..37,17..20] = [enabl,clear,laser,inp7..0]; /*************************************************************************/ /* force all unused outputs low and enable their outputs */ pin [8,9,14,15,16,24] = [io0..5]; [io0..5].oe = ON; [io0..5] = OFF; /* diagnostic outputs */ pin [31..36]= [diag0..5]; [diag0..5].oe = ON; /* outputs to computer interface */ pin [29..25] = [sampl,track,lockd,error,oflow]; [sampl,track,lockd,error,oflow].oe = ON; /* outputs to U1 */ pin [11..13,4..7] = [clck,cary,strt,hq11..8]; [clck,cary,strt,hq11..8].oe = ON; /*************************************************************************/ /* unused nodes disabled */ pinnode 88 = no88; no88.sp = OFF; no88.ar = ON; no88.ce = OFF; no88.d = OFF; /* used nodes */ pinnode [87,86] = [en1,en0]; pinnode [70,85..83,52..48,64..53] = [decd,decr,psb,psa,pq4..0,dq11..0]; pinnode [69,47..41,74..77] = [hq7..0,las2..0,sampd]; /*************************************************************************/ /* [.sp] all synchronous presets off */ [en0..1,dq0..11,hq0..11,pq0..4,las0..2].sp = OFF; [clck,cary,strt,decd,decr,psb,psa].sp = OFF; [sampl,track,lockd,error,oflow,sampd].sp = OFF; /* [.ar] all asynchronous resets off */ [sampl,track,lockd,error,oflow,sampd].ar = OFF; [psb,psa,strt,cary,clck,decd,decr].ar = OFF; [en1..0,hq11..0,dq11..0,las2..0,pq4..0].ar = OFF; /* [.ce] all clocks set synchronous */ [sampl,track,lockd,error,oflow,sampd].ce = ON; [decd,decr,psb,psa,strt,cary,clck].ce = ON; [en1..0,hq11..0,dq11..0,pq4..0,las2..0].ce = ON; 293 /************************************************************************** **************************************************************************/ FIELD pcnt = [pq4..0]; FIELD hreg = [hq11..0]; FIELD dreg = [dq11..0]; FIELD las = [las2..0]; /* diagnostic outputs */ diag5 = (!psb & !psa); diag4 = (!psb & psa); diag3 = ( psb & psa); diag2 = ( psb & !psa); diag1 = las:'b'1xx; diag0 = las:'b'x1x; /* Laser trigger sequence [laser,las2,las1,las0] 1000:1001:x011:x010:x110:x100:1100..0100:0000 */ las0.d = laser & las:'b'00x; las1.d = clck & las:'b'001 # las:'b'01x; las2.d = laser & las:'b'1xx # las:'b'x10; /* 5 bit prescaler */ pq0.d = !pq1; pq1.d = pq0; pq2.t = pcnt:'b'xxx11 & !las0 # pq2 & las0; pq3.t = pcnt:'b'xx100 & !las0 # pq3 & las0; pq4.t = pcnt:'b'x1000 & !las0 # pq4 & las0; dcl = pcnt:'b'10000; /* 1MHz clock pulse */ /* state machine */ psa.t = las:'b'110 & psa # dcl & dreg:'h'fff & !(psb $ psa); psb.t = las:'b'110 & psb # dcl & dreg:'h'fff & !psb & psa; /* outputs to U1 */ clck.d = pcnt:'b'xxx11; /* 8MHz clock pulse */ cary.d = [hq7..0]:#; /* cary from 8bit down counter */ strt.d = las:'b'01x; /* start pulse */ /* outputs to computer */ en0.d = enabl & !scan & !en1; /* synchronize enable and scan */ en1.d = enabl & !scan & en0 # en1 & (enabl # scan); sampd.d = en1 & (sampl & clck # samp & sampd); sampl.d = en1 & !sampd & samp; /* 125ns pulse to trigger AtoD */ track.d = !las1 & track # las1 & psa; lockd.d = !las1 & lockd # las1 & psb & psa; error.d = !clear & error # las1 & scan; /* oflow.d = !las1 & oflow # las1 & psb & !psa; */ oflow.d = en1 & scan; 294 /* control arguments */ hq_clr = las:'b'011; hq_load = las:'b'010 & psb; hq_decr = las:'b'x00 & samp & cary; dq_load = las:'b'010; dq_expo = las:'b'x00 & !psb & psa & dreg:'h'0; dq_incr = dcl & las:'b'x00 & (!psb # psa); /************************************************************************** **************************************************************************/ /* 12 bit delay counter */ /* load */ $repeat i = [0..11] APPEND hq{i}.t = hq_clr & hq{i}; APPEND hq{i}.t = hq_load & ( psa & dq{i} # !psa ); $repend /* decrement */ decd.d = hq_decr # decd & samp; decr.d = hq_decr & !decd; APPEND hq0.t = decr; APPEND hq1.t = decr & hreg:'b'xxxxxxxxxxx0; APPEND hq2.t = decr & hreg:'b'xxxxxxxxxx00; APPEND hq3.t = decr & hreg:'b'xxxxxxxxx000; APPEND hq4.t = decr & hreg:'b'xxxxxxxx0000; APPEND hq5.t = decr & hreg:'b'xxxxxxx00000; APPEND hq6.t = decr & hreg:'b'xxxxxx000000; APPEND hq7.t = decr & hreg:'b'xxxxx0000000; /*************************************************************************/ /* 12 bit period counter */ /* load minimum (-516us) scan time */ APPEND dq0.t = dq_load & !dq0; APPEND dq1.t = dq_load & !dq1; APPEND dq2.t = dq_load & !dq2; APPEND dq3.t = dq_load & dq3; APPEND dq4.t = dq_load & !dq4; APPEND dq5.t = dq_load & !dq5; APPEND dq6.t = dq_load & !dq6; APPEND dq7.t = dq_load & !dq7; APPEND dq8.t = dq_load & !dq8; APPEND dq9.t = dq_load & dq9; APPEND dq10.t = dq_load & !dq10; APPEND dq11.t = dq_load & !dq11; /* load exposure time */ 295 $repeat i = [0..7] APPEND dq{i}.t = dq_expo & !inp{i}; $repend APPEND dq8.t = dq_expo; APPEND dq9.t = dq_expo; APPEND dq10.t = dq_expo; APPEND dq11.t = dq_expo; /* increment */ APPEND dq0.t = dq_incr; APPEND dq1.t = dq_incr & dreg:'b'xxxxxxxxxxx1; APPEND dq2.t = dq_incr & dreg:'b'xxxxxxxxxx11; APPEND dq3.t = dq_incr & dreg:'b'xxxxxxxxx111; APPEND dq4.t = dq_incr & dreg:'b'xxxxxxxx1111; APPEND dq5.t = dq_incr & dreg:'b'xxxxxxx11111; APPEND dq6.t = dq_incr & dreg:'b'xxxxxx111111; APPEND dq7.t = dq_incr & dreg:'b'xxxxx1111111; APPEND dq8.t = dq_incr & dreg:'b'xxxx11111111; APPEND dq9.t = dq_incr & dreg:'b'xxx111111111; APPEND dq10.t = dq_incr & dreg:'b'xx1111111111; APPEND dq11.t = dq_incr & dreg:'b'x11111111111; /*************************************************************************/ 296 Appendix J. Interface Operators Manual for InGaAs PDA *** In this manual I will refer to the C806101 InGaAs multichannel detector head as the “array”. The interface unit, the electronic circuit we devised, as the “device”; the breakout connections as the “computer”; and the laser's triggering pulses as the “laser”. All signal levels except one, the analog signal from the array, are digital and I will refer to them as low (0.7 to 0.0V), or high (2.4 to 5.0V). Any signal line left open will be held low by the device. The devices function is to allow control of the InGaAs arrays exposure time and coordinate the lasers pulses with the transfer of data to the computer. The device has four basic connectors. A 5mm power jack connected to a 12Volt 20Watt power supply. A 9pin D connector to the InGaAs cooling system. A 15pin D connector to the InGaAs diode array. And a 60pin IDC connector to the computer interface. For the purposes of the operator only the signals associated with the 60pin IDC connector are of significance since the computer has no direct control of the other connectors and it is the 60pin IDC connector that has all the signals of interest. Below is a list of the signals carried by this connector and their functions. *** Written by Frank Niertit, expanded by Yuyuan Zhang 297 Pins [2,4,6,8,10..54] are all connected to digital ground. Pins [55,56,59,60] are connected to the analog ground. Pins [57,58] ( Analog Data ) from the array. Pin 1 ( Cooling ) Is the signal controlling the cooling unit of the array. A high level turns the cooling unit on. Pin 3 ( Gain ) Is the gain select for the array. A low level signal sets the gain to maximum and a high level signal sets it to its minimum. Pins [5,7,9,11,13,15,17,19] ( Exposure Time ) Are input signals from the computer to the device. They represent a value of 0 to 255 microseconds and set the exposure time of the array. The most significant bit with a value of 128 microseconds is pin 5. The least significant bit with a value of 1 microsecond is pin 19. A high level signal activates the pin. The signals on these pins need to be held stable during the operation of the device. The actual exposure time is the value impressed on these pins plus 2.5 ± 1.0 microseconds. Pins [21,23,25,27,29] Are output signals from the device to the computer. Pin 21 ( SAMPLE ) The sample signal is a 125nanosecond pulse that signals the computer to sample the data. 256 such pulses will occur per scan. Pin 23 ( Tracking ) The tracking signal indicates that the device is successfully tracking the laser pulse stream. This does not mean that the exposure time is correct, only that the laser frequency is within range of the tracking capability of the device, 1.9 kHz to 212 Hz. 298 Pin 25 ( Locked ) The locked signal indicates that the device is controlling the exposure time of the array. Pin 27 ( Error ) The error signal is set whenever the device receives a laser pulse during the output scan cycle of the array. It means that the device was unable to generate a corresponding start pulse for the array and that there was no data taken for that laser pulse. This signal will stay high until the computer asserts a clear signal on pin 33. Pin 29 ( Overflow ) The overflow signal indicates that the laser pulse repetition rate has fallen below 212Hz. Note: this pin has been reconfigured later to give the trigger to the slow board (equivalent to the “sample” signal to the fast board) in order to read the chopper signal. Pins [31,33,35] Are input signals from the computer to the device. Pin 31 ( Enable ) The enable signal allows the computer to control the sample signal. A high signal allows sample signals to be sent to the computer. This signal is synchronized to the array scan in such a manner that it will always enable sample signals starting with the first bit of scan data and will always allow the completion of a scan. This relieves the computer from having to coordinate with the array. Pin 33 ( Clear ) The high clear signal will reset the error signal if it is set or if the high level is maintained it will prevent the error signal from being set. 299 Pin 35 ( Laser ) The laser pulse will initiate a start pulse for the array if a scan cycle is not active. Otherwise it will set the error signal which will remain high until reset with the assertion of a clear signal from the computer. Pins [37,39,41,43,45,47,49,51] ( Temperature ) Are output signals to the computer. They represent an 8bit binary number from 0 to 255 which corresponds to 0 to 2.55 volts from the arrays temperature sensing diode. Pin 37 is the most significant bit corresponding to 1.28 volts. Pin 51 is the least significant bit corresponding to 0.01 volts. These lines are updated on every pulse asserted on pin 53. The levels are held until the next pulse on pin 53. Pin 53 ( A/D Sample for temperature readout ) An output pulse from the computer that acts as the sample signal to the on board A/D converter (on the timer). This signal should be at least 2.5 microseconds in duration. The A/D converter will hold the output value read during this pulse until the next pulse arrives. Basic operating procedure: The computer should set the clear line high and the enable line low. It should then interrogate the tracking signal to see if the laser is in range and that the device is tracking its signal. Once the system is tracking an exposure time should be set and the locked signal interrogated to see if we are holding lock. If lock has been established the clear 300 line should be de-asserted to see if we received any errors. These errors are usually caused by the jitter of the laser. If we are experiencing errors the exposure time should be increased. Repeat this procedure until you have a stable lock with no errors. Once the system is stable assert the enable line to start collecting data. In the most simplistic fashion, the InGaAs PDA detection system requires five signals to function properly: 1) the “laser” signal from pin 35; 2) the “enable” signal from pin 31; 3) the “sample” signal from pin 21; 4) the chopper signal input directly to the slow A/D board; and obviously 5) the analog signal from pin 57. The users are urged to ensure these lines connected to their corresponding post in the breakout board properly before any experiments. In addition, a +5V voltage should be applied to pin 1 in order to turn on the TE cooler (and fan) for the InGaAs PDA (the green indicator on the detector head will illuminate after cooler is on). Moreover, the users are urged to set the exposure time properly by applying +5V to pins 5, 7, 9, 11, 13, 15, 17 or 19, or any combination. A smaller exposure time can reduce the amount of stray NIR light, but increase the risk of encountering an error caused by the jitter of the laser trigger signal. Role of the programmable chips: The principal operating mechanisms of the silicon and InGaAs timers are similar. However, an important difference between the two is the capability of manipulating the exposure time in the InGaAs timer. This is of fundamental importance for NIR detection as this wavelength region is susceptible to the ambient environment (light and heat), and 301 thus reducing the exposure time will reduce the amount of background signal and ultimately improve the S/N in experiments with NIR probing. The manipulation of the exposure time, unfortunately, is not straightforward with the current InGaAs detector head and therefore requires the manipulation of data transfer time to achieve this goal – see Figure J.1 for illustration. Figure J.1 Illustration of data transfer process for a 256-channel detection in a 1 kHz system. (a) laser pulse train; (b) laser trigger pulse train (from SDG box); (c) transfer time and exposure time in between every laser pulse; (d) “sample and hold” for each channel (total 256 channels). As Figure J.1 illustrated, the data transfer occurs in-between every laser trigger pulse, i.e., 1 ms for 1 kHz system. As each channel requires a minimum of 2 μs to transfer the analog data, a minimum of 512 μs is needed to transfer 256 such channels (the actual 302 transfer time for one InGaAs array is ~530 μs). The remaining ~470 μs is the exposure time where the detector is collecting photons. This transfer time can be lengthened and as a consequence, the exposure time is shortened. This is achieved by stretching the “hold” time (see Figure J.1 (d)). For example, to set an exposure time of 255 μs, the transfer time needs to be changed from 530 μs to 745 μs – the extra 215 μs can be realized by stretching the transfer time of the first 215 channels (total 256 channels) from 2 μs to 3 μs (see Figure J.2 for illustration). Figure J.2 Illustration of manipulating the exposure time indirectly by manipulating the data transfer time. For demonstration only, the mechanism shown here is not necessarily identical to how the program divides the extra 215 μs among the 256 channels. The programmable chip labeled “U2” takes the user input of the exposure time and send to a second programmable chip labeled “U1”, which determine how much extra time needs to be added to the transfer time and exactly which channel(s) to stretch. The latter also provides the necessary signals to drive the InGaAs PDA, receives the feedback from the array and prepares the 256 trigger signals (“sample” / “CL Clock”) for the A/D. “U2” on the other hand, gates the “sample” signal and provides the necessary diagnostic signals for the overall performance of the timer. 303 Laser Repetition Rate and Dual-Array detection As a final remark, it is worth pointing out that dual-array detection cannot be implemented for the 256-channel InGaAs array system if the laser repetition rate is 1 kHz. Distinct from the dual-Silicon array where the total data transfer time is ~1 ms (500 μs/array at 520 kS/s), the transfer time for dual-InGaAs array is 1.06 ms, as limited by the speed of the A/D board (~530 μs/array at 520 kS/s). It is noteworthy that this data transfer time in the dual-InGaAs array system does not depend on the user-defined exposure time, which is different from the single-InGaAs array system where the data transfer time is deliberately lengthened in order to decrease the exposure time. In the case of dual-array, the transfer time is a constant (1.06 ms), and together with the variable exposure time, it defines the total time required from data collection – this time would increase if the user chooses a longer exposure time (it is usually set at 4 μs). It is immediately apparent that the laser cannot be set at 1 kHz as 1 ms is not enough to completely transfer data; the minimum separation time for two adjacent pulses must be ~1.06 ms, which means that the repetition rate of the system must be set to < 940 Hz in case of dual-InGaAs array detection so that the data from both arrays can be transferred before the next incoming pulse. A similar idea applies to a dual-array setup in which one InGaAs array and one Silicon array are used (for simultaneous detection of UV-Visible and NIR regions). Since a minimum of ~1.03 ms is needed (~530 μs for InGaAs, ~500 μs for Silicon), a maximum laser repetition rate of ~970 Hz is required. 304 If dual-array detection is not needed, the second channel must be turned off (as can be done by our LabView software ††† ) when the InGaAs array is used for NIR probing. In this case, the maximum repetition rate of the laser can be set to ~1.9 kHz (1/530μs), and faster repetition is not possible with the current A/D converter speed – see description on Pin 23 for the operating range of the laser repetition rate. ††† Under “Setting” →“DAQ Hardware Setup”, or go directly to “Hardware Settings 3.0.0.vi”, click the green indicators at “channel list” to turn on or off a particular channel. 305 Appendix K. 2PA Spectra of Chlorinated Alkanes Figure K.1 presents the polarized 2PA spectra for two commonly used chlorinated solvents – chloroform and carbon tetrachloride, obtained with similar experimental parameters as those for the alkanes. 2PA spectra for both molecules show similar behavior as those of water, alcohols and alkanes, in that the 2PA cross sections for the chlorinated alkanes increase monotonically from 6.7 to 8.7 eV, and the cross section for absorbing two parallel polarized photons is larger than absorbing two perpendicularly polarized photons. Both molecules show relatively small cross sections in the low energy region but a significantly larger ones at E 2PA > 7.5 eV, which is indicative of two different transitions in the two energy regions. This is further confirmed by the varying polarization ratios across the 2PA energy range. Chloroform, belonging to the C 3v point group, has a very high ratio (> 3) in the low energy region, but it decreases quickly to ~2.7 and stays constant from ~7.2 to 8.5 eV. The polarization ratio for carbon tetrachloride (T d point group) is observed to vary from ~1.7 in the low energy region to ~3 in the high energy region. Compared to the condensed phase 1PA spectrum of carbon tetrachloride, 58 the feature at ~7.0 eV due to the T 2 ← A 1 transition is much less pronounced in the 2PA spectrum. 306 Figure K.1 Relative 2PA spectra for (a) chloroform and (a) carbon tetrachloride (CCl 4 ). Both parallel and perpendicular spectra, as well as the polarization ratios, were shown. 307 Appendix L. Preliminary Results for 2PA Spectra of Benzene The aromatic solvent benzene has a much lower absorption threshold as compared to all molecules discussed above, thereby requires a lower energy pump to access E 2PA < 6.5 eV. Figure L.1 presents the 2PA spectra of neat benzene obtained with 1.84 eV (675 nm) and 3.1 eV (400 nm) pump wavelengths. The 2PA cross section is extremely weak in the region of 4.5 – 6.25 eV, but dramatically increased by ~100x at 7.0 eV. However, perusing the low energy region reveals a subtle structure with vibronic progression reminiscent to that observed in the 1PA spectrum. 59 Figure L.1. 2PA spectra of benzene obtained with 3.1 eV (red curve) and 1.84 eV (blue curve) pump. The Raman response and the pump scattering were removed from the spectra and result in a gap in the 5.5 – 6.25 eV region of the red curve. The spectra were measured at the polarization of the pump and probe set to parallel. The curve for the 1.84 eV pump was measured in an absolute scale in GM, but the curve for 3.1 eV pump was scaled arbitrarily for ease of comparison (as shown in dash line in the inset). The slight difference in the peak positions of the vibronic progression is due to the calibration error (in measuring the center wavelengths for both the pump and the probe) and it is not scientifically meaningful). 308 Appendix M. Evidence for Excited State Radical The observation of the fast decay A % state is strongly hindered by the large 2PA signals at early time. This contribution is partially reduced by using a shorter pump pulse afforded by the hollow core fiber apparatus. An instrument response time of ~56 fs, defined as the FWHM of the 2PA “spike” (convolution of the pump and probe pulses), seems to be reasonable to extract dynamical information at 100 fs. However, if the TA signal from the A % state radical is outsized by the 2PA “spike” (which is usually the case), the residual 2PA at 100 fs could still interfere with the TA signal. Examining the spectrum at t = 0 fs reveals a band centered right at 363 nm, the expected absorption wavelength for the radical D ~ ← A % transition. Let’s first assume this t = 0 fs spectrum is the 2PA spectrum of p-MePhSH and no photoproduct contribution at such early time. If this was true, tuning the pump wavelength would subsequently change the absorbing wavelength of the probe, since the total 2PA energy (pump + probe) must not change for a given transition. Specifically, if the λ probe = 363 nm feature observed in λ pump = 267 nm is indeed due to a transition to an unidentified higher excited state via 2PA, the transition energy would be 8.06 eV. Thus, tuning λ pump to 295 nm would dramatically shift the probe absorption to a much shorter wavelength of 321 nm – see Figure M.1 for illustration. In light of this, a ~40 nm blue shift of the probe absorption should be expected when the 295 nm pump is used. 309 Figure M.1 Left: Illustration of pump-probe combinations for a 2-photon transition to an unidentified excited state (S n ), the transition energy is assumed to be 8.04 eV (calculated from the combination of 267 nm + 363 nm). Right: spectra at t = 0 fs for p-MePhSH in ethanol, excited with three different wavelengths. The ΔOD signals of the three traces were normalized at 363 nm. However, from examining the t = 0 fs spectra with three different pump wavelengths, it is apparent that there is no such shift when tuning the pump wavelength from 267 to 295 nm. Instead, the feature is always centered at 363 nm with λ pump = 267, 285 and 295 nm (see Figure M.1). Therefore, the contribution of 2PA signal can be ruled out and this feature must originate from a transient species being produced within our instrument response time. p-MePhS radical is the leading contender for such transient species as the ballistic bond fission usually occurs within ~10 fs (the time scale for an O–H stretch vibration). Moreover, according to ab initio calculations, the X % state radical does not absorb at ~363 nm, but the A % radical does show a transition around this energy. This is a strong evidence to support assigning the 363 nm feature to A % state radical absorption. 310 Further, perusing the spectra around t = 0 fs reveals a band in the region of 450 – 500 nm, where the X % state radical shows dominant absorption. This reinforces the idea of observing the A % state radical within our instrument response time, since, according to the gas phase HRA-PTS experiments, both X % and A % state radicals are generated upon UV irradiation. Interestingly, the TA spectra at 0 ≤ t ≤ 200 fs even show hints of the A % → X % electronic quenching (occurred via re-crossing the S 2 /S 0 CI). From Figure M.2(a) we can see that the area of the X % state radical absorption (450 – 550 nm) increases from 0 to 200 fs delay time (after subtracting the underlying background due to S 1 ESA), and this is commensurate, at least qualitatively, with the decay of the A % state radical absorption (~363 nm). Figure M.2 TA spectra of p-MePhSH in (a) ethanol and (b) cyclohexane at 0 ≤ t ≤ 200 fs, excited with 267 nm and probed by broadband continuum. The relative polarization between the pump and probe is set to magic angle. Note that these two experiments were not performed back-to- back. 311 As we provided strong evidence for the generation of A % state radical, we are now poised to explore the reaction dynamics in two different solvents. In ethanol solution, both radicals were produced initially, but the A % state radical undergoes fast electronic quenching which contributes to the population growth of the X % state radical. However, ab initio calculations predict that the TDM of the D ~ ← A % transition (~363 nm) is ~10x smaller than the transition B ~ ← X ~ (~500 nm), which means that the oscillator strengths of these two transitions differ by two orders of magnitude (see Table 4.1). If the TDMs calculated for an isolated molecule is transferable to the solution phase, our TA spectra would then suggest almost all radicals were generated in the A % state. This situation is only favorable when the S–H bond lies out-of-plane as the dissociating p-MePhSH approaching the S 2 /S 0 CI. Although the torsional barrier of the parent molecule is low, 47 it is highly improbable that the S-H bond angular distribution is predominantly out-of-plane even for a free rotor. Thus, our observation of the A % state radical casts sufficient doubt on the calculated TDM, and unfortunately, without the knowledge of the extinction coefficient of the A % state, we cannot comment on the Ã / X % product branching ratio. The electronic branching of the Ã and X % radicals in cyclohexane seems to be similar to that in ethanol. Ã state product is observed, but the X % state absorption is absent in cyclohexane solution. This is even more extreme than the ethanol case and contradicts with the gas phase results, where both the Ã and X % radicals are observed upon 267 nm photolysis. However, the difference between the ethanol and cyclohexane solutions is 312 small, especially when comparing the relative yield between the Ã and X % radicals at t ≥ 50 fs. The quantitative analysis of the yield is further complicated by the interference of the evolving underlying feature. Thus, to the first order approximation, it can be concluded that the reaction dynamics of the dissociating p-MePhSH is similar in both cyclohexane and ethanol; the strongly interacting ethanol solvent provides no change (or no additional change) to how the molecule approaches the S 2 /S 0 CI, as compared to cyclohexane.

Abstract (if available)

Abstract Photochemistry occurring in a solution phase environment is often complicated by the solute-solvent interaction. Depending on how strongly the solvent interacts with the solute, the electronic structure of an excited state solute molecule can be very different from that in an isolated environment. This change in the electronic character of the excited state can turn on various competing deactivation processes which are otherwise absent in the gas phase. To decipher this complex condensed phase behavior, prior knowledge on the excited state electronic character is crucial. Two-photon absorption (2PA) spectra contain rich information on the excited state symmetry of a given molecule. The measurement of a polarization dependent broadband 2PA spectrum is realized by the pump-dispersed-probe technique, from which the zero-delay spectrum is extracted by removing the white light chirp. The experimental technique is described, and the 2PA spectra of several alcohols and liquid alkanes commonly used as solvents are presented for the first time. This provides information on the window of transparency for these liquids, and further, the symmetry of low lying electronically excited states. 2PA spectra are extremely valuable in deciphering broad and overlapping transitions usually observed in the condensed phase spectra. ❧ The ubiquitous nature of πσ* excited states has attracted much attention in the recent years. These dissociative states play an important role in the photochemistry of phenols, thiophenols and other heteroaromatic molecules, as they lead to heteroatom–H homolytic bond fission upon UV irradiation. There has been a surge of frequency-resolved studies for this class of molecules, but as much as the nuances of the photodissociation reaction can be extracted from these gas phase experiments, important photochemical reactions of such molecules of biological importance occur in condensed environments, e.g., in water where strong solute-solvent interactions are expected. In a collaborative project, the ultrafast pump-probe technique is used to study the non-radiative deactivation pathways of phenol and thiophenol in solution upon deep-UV excitation. It is found that thiophenol molecules undergo prompt S-H bond fission exclusively in both cyclohexane and ethanol. In the case of phenol, gas phase-like bond fission is also observed in cyclohexane, but competing pathways such as autoionization and proton-coupled electron transfer come into play in aqueous solution.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Ultrafast spectroscopy of aromatic amino acids and their chromophores in the condensed phase

PDF

Capturing the sun: leveraging excited state dynamics

PDF

Following redox chemistry and excited state dynamics in solution using liquid jet photoelectron spectroscopy

PDF

A multifaceted investigation of tris(2-phenylpyridine)iridium

PDF

Electronic structure and spectroscopy of radicals in the gas and condensed phases

PDF

Excited state dynamics of organic photovoltaics probed using ultrafast spectroscopy

PDF

Ultrafast electronic deactivation of DNA bases in aqueous solution

PDF

Electronically excited and ionized states in condensed phase: theory and applications

PDF

Interfacial vibrational and electronic excited states of photovoltaic materials probed using ultrafast nonlinear spectroscopy

PDF

Vibrational sum frequency spectroscopy of molecules on metal, semiconductor, and aqueous surfaces

PDF

Modeling x-ray spectroscopy in condensed phase

PDF

First steps of solar energy conversion; primary charge generation processes in condensed phase coupled chromophores

PDF

Quantifying aspects of DNA damage

PDF

Electronic structure and spectroscopy in the gas and condensed phase: methodology and applications

PDF

Dissociation energy and dynamics of water clusters

PDF

Photodissociation dynamics and energetics of HCl-(H₂O)₃

PDF

Development of robust ab initio methods for description of excited states and autoionizing resonances

PDF

Liquid phase photoelectron spectroscopy as a tool to probe excess electrons in liquid ammonia along with Birch chemistry carbanions: from dilute electrolytic solutions to the metallic regime

PDF

An investigation of morphology and transport in amorphous solid water via guest-host interactions

PDF

Complementary hydrogen bonding enables efficient and mechanically robust polymer solar cells

Asset Metadata

Creator Zhang, Yuyuan (author)

Core Title Two-photon absorption spectroscopy and excited state photochemistry of small molecules

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Publication Date 05/01/2013

Defense Date 03/09/2012

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag methylthiophenol,OAI-PMH Harvest,phenol,photochemistry,two-photon absorption

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Bradforth, Stephen E. (committee chair), Wittig, Curt F. (committee member), Wu, Wei (committee member)

Creator Email tommyzh@gmail.com,yuyuanzh@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-21099

Unique identifier UC11289849

Identifier usctheses-c3-21099 (legacy record id)

Legacy Identifier etd-ZhangYuyua-703.pdf

Dmrecord 21099

Document Type Dissertation

Rights Zhang, Yuyuan

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA