University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Ultrafast Spectroscopy of Aromatic Amino Acids and their Chromophores in the Condensed Phase by Gaurav Kumar 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) December 2019 ii Dedicated to my parents Asha and Ram Lakhan Sharma iii ACKNOWLEDGEMENTS After spending 6 amazing years at USC, I have come to realize that joining USC to pursue my PhD was one of the best decisions I have made in my life. I am thankful to my advisor Prof. Steve Bradforth and the entire Bradforth group for giving me wonderful memories which I will cherish throughout my life. I am fortunate that I was always surrounded by people who believed in me and kept encouraging me throughout this journey. Steve is an excellent mentor and I am lucky that I got the opportunity to work with him for my PhD. I thank him for sharing his knowledge and experiences, and for providing constant support. I was always impressed by his vast knowledge whether it is science or a technical problem in the lab. Although Steve became chair of the department and then divisional Dean later during my PhD, he always found some time to interact with me. I have taken several short walks with him either from one office to another or office to the USC shuttle stop, seeking my scientific quest or looking for a clue to troubleshoot an issue in the lab. There were several moments in my PhD when I was frustrated and was feeling very low, but Steve always managed to sit with me and have a friendly chat. His word of wisdom during those mental therapy sessions were “We are on a cusp” and “Hang a bit longer” which I will always remember. I’m very grateful to my professors who played an important role in shaping my mind. Many thanks to my undergraduate advisor Prof. K.S. Viswanathan (IISER Mohali) who introduced the world of “Spectroscopy” to me and encouraged me to pursue PhD. At USC, I was fortunate to take classes from Prof. Jahan Dawlaty and Prof. Alex Benderskii. Jahan and Alex are wonderful teachers. They not only taught me spectroscopy, but they also served as my committee members. I owe a lot to them for all the valuable suggestions and comments during my PhD. iv During my first two years in the lab, I worked with Anirban Roy. I will always be grateful to him for training me in the lab and teaching several lab skills. He was phenomenal in explaining the scientific concepts and was always patient with me when I was not able to understand any concept. I also learnt a lot from Saptparna Das. She taught me how to plan and perform experiments in the lab. She taught me how to do something as basic as cleaning the crystal of Micra to a more involved exercise such as aligning a laser beam properly. Both Anirban and Saptparna were inspirational seniors. I am fortunate that I am still in regular touch with them. I will always be grateful for their guidance and sharing the knowledge. Robert Seidel was a postdoc in Bradforth lab when I Joined USC. He made sure that I understand the vacuum technology and TRPES machine before he left the group. Robert not only helped me scientifically, but he remains a regular supplier of chocolates to sweeten our life at USC. Although I did not have much overlap with Robert, he was always there to assist me through email whenever I needed his help. Konstantin was another senior member in the lab who always kept the office alive by his jokes and banter with Robert. After my second year, I was the lead member in the group with not enough knowledge to maintain the lab but was fortunate to be accompanied by Jimmy Joy and Laura Estergreen. We troubleshooted n number of problems in the lab together. They were excellent colleagues and were always there to dirty their hands in the lab. Although Laura was not an active member of my research project, she was always there to help me out in carrying out the experiments and explaining the concepts which I was unaware of. I also want to thank my half lab-mate Dhritiman Bhattacharya. Dhritiman and I shared the apartment in my first year and also took the classes at USC together. With Dhritiman around, life cannot be boring. v I will always be short of words when I have to talk about Mike Kellogg and Ryan McMullen. We worked together during my last phase of the PhD. Ryan was always full of energy in the lab and ready to knock down any equipment in the lab if it required any repairing. He fearlessly took the job to make nozzles by himself which I would have never gathered confidence to do by myself. I am impressed with his engineering skills and have no doubt that he will also improve the photoelectron spectrometer with the help of “powerful” Mike and new postdoc Matt Bain. Mike is one of the humblest people which I have met in my life and I am honored that we worked together. I have spent several nights in the lab doing the experiments with Ryan and Mike. Without their support, I do not think my PhD would have been finished. Apart from my lab mates, I had the opportunity to work with four talented undergrads, Sanjali Padalkar, Tyler Zhang, Shanmukh Kutagulla and Jose Alaras. They were very enthusiastic while working in the lab and also taught me several English phrases. I enjoyed a lot working with them and wish them for their bright career. I would also like to thank our collaborators Dr. Tom Oliver (Uni. Of Bristol, UK) for helping me with data analysis and Prof. Eberhard Riedle (LMU, Munich) and his student Dr. Bastian Baudisch for assisting me in setting up the NOPA in the lab. I also want to thank new members of the Bradforth group, Dr. Matt Bain, Dr. Tillmann Buttersack and Shivalee Dey, who joined recently. I had short but sweet overlap with them. I had several discussions with Matt not only about science but about life in general. I am happy to have met him before leaving USC. Shivalee is very enthusiastic and I am sure she will prove herself a great asset to Bradforth’s group. vi Apart from my lab members, I had several friends at USC who were part of my PhD journey. I want to thank Purnim Dhar, Atanu Acharya, Subhashish Sutradhar, Chayan Dutta, Subodh Tiwari, Deepak Verma, Piyush Deokar, Ankit Khullar, Amit Samantha, Parichita Mazumdar, Angelo Montenegro, Anuj Pennathur and Muhammet Mammetkuliyev. We had several fun activities together during my stay at USC. I shared my apartment with Deepak, Chayan and Subodh and I thoroughly enjoyed their company. Special thanks to Angelo for inviting me to his wedding ceremony and giving the honor to be his “best man”. During my last two years, I had several short trips to Santa Barbara over the weekends and had a get-together with friends from my undergraduate college. I want to thank Shruti Arya, Chinmoy Sarkar and Asif Equbal for keeping me sane and for having fun together. Many thanks to staffs of Machine shop, Seth, Don and Ramon. I do not have enough words to thank Seth. Seth helped us design the magnet assembly and even worked on the “eve before Christmas” to machine the manipulator of the magnet and ensured that our experiments keep running without any delay. I also owe my sincere gratitude to Michele Dea for taking care of all the administrative issues and Magnolia Benitez for being our graduate adviser. Thanks to National Science Foundation (NSF) for providing the financial assistance to carry out the research. Finally, thanks to my parents and siblings. Being youngest in the family, I had the privilege of getting support from everyone. They were always standing with me and encouraged me to take my decisions on my own and for this I am immensely grateful to them. vii TABLE OF CONTENTS List of Figures ix List of Tables xviii Abstract xix Chapter 1 Introduction 1 1.1 Photophysics and Photochemistry in Daily Life 1 1.2 Photophysics and Photochemistry: A Basic Understanding 2 1.3 Photophysical Phenomena in the Excited State 5 1.4 Experimental Techniques: Time Resolved Photoelectron Spectroscopy and Transient Absorption 10 1.5 Thesis Plan 13 1.6 References 15 Chapter 2 Experimental Methods 17 2.1 Overview 17 2.2 Liquid Microjet Photoelectron Spectrometer 19 2.2.1 Brief Description 19 2.2.2 Improvements from the Previous Version 21 2.2.3 New Design of the Magnet 22 2.2.4 Precise Alignment of the Magnet 26 2.3 Calibration of the Spectrometer 29 2.4 New Ultrafast Tunable Laser Source 32 2.4.1 Noncollinear Optical Parametric Amplifier (NOPA) 33 2.4.2 Generation of Tunable UV Source 40 2.5 Long Time Pump-Probe Set-up 42 2.6 References 45 Chapter 3 UV Photoelectron Spectroscopy in Liquid Microjet: Promises and Technical Challenges 47 3.1 Introduction 47 3.2 Experimental 51 3.3 Current Issues in Experimental Measurements 53 3.3.1 Transmission/Cut-off function 53 3.3.2 Inelastic Electron Scattering 57 3.4 Results and Discussions: 59 3.4.1 R2PI Spectra of Aqueous Phenol 59 viii 3.4.2 R2PI Spectra Measured at Various Pump Wavelengths 64 3.4.3 Spectral Comparison between Bradforth’s and Fielding’s Lab 67 3.4.4 Time-Resolved Photoelectron Spectra of Aqueous Phenol 69 3.5 Conclusion 74 3.6 References 75 Chapter 4 The Influence of Aqueous Solvent on the Electronic Structure and Non- Adiabatic Dynamics of Indole Explored by Liquid-jet Photoelectron Spectroscopy 78 4.1 Introduction 78 4.2 Experimental 82 4.3 Results 84 4.3.1 Resonant Two Photon Ionization Photoelectron Spectra 85 4.3.2 Time-Resolved Photoelectron Spectra 88 4.4 Spectral Analysis 94 4.5 Discussions 98 4.6 Conclusion and Outlook 101 4.7 Postscript to Chapter 4 103 4.8 References 108 Chapter 5 Unraveling the Excited State Dynamics of Aqueous Indole from Broadband Femtosecond Transient Absorption and Quenching Studies 111 5.1 Introduction 111 5.2 Experimental 116 5.3 Results and Discussions 118 5.3.1 Steady State Fluorescence Spectroscopy 119 5.3.2 Time-Resolved Spectroscopy of Indole in Ethanol 120 5.3.3 Time-Resolved Spectroscopy of Indole in Water 122 5.3.3.1 Excitation at 200 nm 122 5.3.3.2 Excitation at 266 nm 129 5.3.3.3 Excitation at 292 nm 131 5.4 Conclusion 135 5.5 References 136 ix List of Figures Figure 1.1: Potential energy surfaces featuring adiabatic (left) and nonadiabatic (right) picture 5 Figure 1.2: Schematic representation of the influence of polar solvent on the excited state dynamics of indole 7 Figure 1.3: (a) Transient absorption (TA)- excited state dynamics is monitored by measuring excited state absorption (ESA) (b) TRPES technique- electron kinetic energy (eKE) of photoejected electron from intermediate state is measured to monitor the dynamics of the excited state. 12 Figure 2.1: Schematic representation of TRPES. The potential energy surfaces for the ground state (M), the excited state (M*) and the cationic state (M + ) are illustrated. Right panel depicts the temporal evolution of the excited state in terms of electron kinetic energy of the outgoing electrons. Due to dramatic change in the displacement along nuclear coordinate, the blue colored PES band will be much broader with structured vibrational progression compared to the red colored PES band. 18 Figure 2.2: Newly designed magnet assembly: (a) z-axis actuator, (b) edge-welded bellows, (c) PEEK rod (polyetheretherketone), (d) soft iron rod, (e) permanent magnet inside aluminum jacket, (f) soft iron conical tip, (g) x-axis actuator, (h) y-axis actuator 23 Figure 2.3: Design diagram of the new manipulator along with magnet assembly 25 Figure 2.4: Three-photon ionization of Xenon. (a) eKE spectra measured at multiple values of Z (Z is the distance between the magnetic-tip and the liquid-jet), (b) TOF spectra measured at multiple values of Z. The liquid jet is usually parked at ⁓2mm from the skimmer orifice. The value of Z (in the inset of figure) signifies the distance between the jet and the tip of the soft iron cone of the magnet. 28 x Figure 2.5: Calibration of the photoelectron spectrometer with water vapor in presence of the water jet. a) Schematic of energy levels involved in the three-photon photoionization of water using 266 nm ultrashort pulses (~200 fs). The observed structure in the time-of-flight spectra (b) represents transitions to different vibrational levels on the cationic surface (D0). The time of flight values for each peak is calibrated against the literature kinetic energy values 34, 35 and fitted with eqn. 1 (c). The calibrated kinetic energy spectrum is shown in (d) with the vibrational assignments (v1 v2 v3) of the X 2 B1 state of H2O + . 34 The ionization energies for (000), (010), (100), (110) and (200) bands are 12. 615 eV, 12.792 eV, 13.017 eV, 13.191 eV and 13.408 eV respectively. 35 30 Figure 2.6: Illustration of the position of the skimmer, the ionization point and the liquid microjet while measuring the REMPI of Xe to extract the streaming potential. The equation utilized to extract the streaming potential is also shown in the right panel. 32 Figure 2.7: Amplification of a weak signal with the help of an intense pump beam. (a) the pump (purple) and the weak signal (green) beam entering the crystal in a collinear fashion resulting into temporal walk-off between the signal and the idler (red) (b) the pump (purple) and the signal (green) entering the crystal in a noncollinear geometry and hence while propagating through crystal the signal (green) and the idler (red) are always temporally overlapped resulting into broadband amplification of the signal. 38 The conception of the picture is adapted from the manual of the NOPA. 37 34 Figure 2.8: Reflective telescope (Galilean type), 1: Concave mirror of focal length of 50 cm, 2: A plane mirror (800 nm HR), it was used to provide a compact design by folding the laser beam 3: Plano-concave lens of focal length -25 cm mounted on compact dove-tail linear stage (9.5 mm travel, Newport) for fine adjustment 36 Figure 2.9: (a) Ray schematic of NOPA 37 (b) NOPA on the laser table (c) Superfluorescence ring generated by blue pump after the BBO crystal (d) Amplified and broadband signal generated from the NOPA (bottom in the ring) 37 xi Figure 2.10: (a) Narrowband signal resulting from short pulse-width of the pump beam (b) Broadband signal resulting from chirped pump beam. 39 39 Figure 2.11: The folded geometry of prism pair compressor set-up to compress the visible output of the NOPA. Mirror1: (R= -100 cm SILFLEX broadband mirror for beam collimation) | (2,3,4,6,7,9): EAV mirror | 5,8: Brewster cut fused silica prisms 40 Figure 2.12: (a) UV spectrum measured with fiber optic spectrometer (Ocean Optics 2000) (b) Cross correlation trace between a 268 nm UV pulse and a probe white light continuum. 370 nm, 400 nm and 450 nm were chosen from white light continuum to calculate cross-correlation width. 41 Figure 2.13: Illustration of the wiring inter-linking the devices in the lab (SSC716) 43 Figure 2.14: The scheme to introduce long-time delay 44 Figure 3.1: Illustration of the effect of inelastic scattering on photoelectrons with different eKE. The concept of this figure is adapted from Ref. 18 48 Figure 3.2: (a) UV-Vis absorption spectrum of phenol and tyrosine in water (b) Gas phase PES of phenol along O-H coordinate for S0, S1 and S2 26, 27 . The figure for the gas phase PES of phenol has been adapted from Ref. 26 , authored by Ashfold and his coworkers. 50 Figure 3.3: (a) R2PI spectrum of aqueous indole. The black curve is the experimentally measured spectrum while the red curve is the constructed R2PI spectrum by taking the mirror image of higher eKE region of the spectrum on the lower eKE region. (b) the estimated form of the transmission function generated by taking the ratio of red and blue curve from Figure (a) 55 xii Figure 3.4: (a) Energy diagram for the photoionization of hydrated electron. eBE(genuine), Vo and ΔE represents genuine binding energy for the hydrated electron, the energy gap between the bottom of conduction band of water with respect to the vacuum and eKE loss due to inelastic scattering, respectively. (b) Binding energy distribution for the hydrated electron reported by Luckhaus et al. 35 These figures are taken from Ref 35 56 Figure 3.5: (a) XPS spectrum of aqueous phenol. (b) R2PI spectrum of aqueous phenol where the wavelength is centered at 267 nm. Two Gaussians were required to properly fit both the spectra. Vertical orange line depicts the center of the Gaussians corresponding to VIEs. Note: The x-axis in XPS spectra represents binding energy (BE) while x-axis in R2PI represents electron kinetic energy. (c) R2PI spectrum of aqueous phenol corrected for inelastic scattering after using the transformation scheme suggested by Suzuki and his coworkers. 38 60 Figure 3.6: The presence of two ionization peaks in R2PI-PES experiment emphasizes the contribution of configuration interaction also found in gas phase phenol. 40 63 Figure 3.7: (a) R2PI spectra of aqueous phenol at different photon energy generated by doubling the OPA output, the pulse energy used for 4.46 eV, 4.66 eV, 4.73 eV, 4.88 eV, 4.96 eV, 5 eV and 5.04 eV are 40 nJ, 15.5 nJ, 17.5 nJ, 29 nJ, 36 nJ, 43 nJ and 45 nJ. (b) Comparison of high energy eKE edge for spectra (simply shifted by extra photon energy with respect to 4.46 eV) and experimentally measured eKE edge in higher eKE region for various photon energies. (C) Illustration of effect of inelastic scattering on the measured eKE as a function of photon energy in the gas phase and in solutions. 66 Figure 3.8: R2PI spectra of aqueous phenol recoded in Bradforth’s lab at USC and Fielding’s lab at UCL. R2PI spectra measured in Fielding’s lab is taken from ref 41 . xiii Figure 3.9: R2PI spectrum of phenol (20 mM solution; 266 nm ionization pulse) is compared with the TRPES spectrum of phenol when both 266 nm and 292 nm pulses are temporally overlapped (τ=0) in panel (a). Individual Gaussian fits are shown in magenta and green and the total fit is shown in blue for the R2PI spectrum, while the red line displays a single Gaussian fit for the TRPES spectrum. Panel (b) displays a small shift in the peak maximum (0.08 eV) between the TRPES spectrum of aqueous phenol at t=0 ps and t=500 ps. Panel (c) shows TRPES spectra of aqueous phenol at different pump-probe delays and illustrates no change in spectral shape. Panel (d) shows TRPES spectra at longer pump-probe delays from a separate measurement implementing ns-pump laser centered at 266 nm and 292 nm probe. The 292 nm laser pulse used for the spectra recorded in panel (a), (b) and (c) was generated by doubling the NOPA output while 292 nm used for ns-delays measurement was generated by doubling the OPA output. The polarization of both beams is parallel with respect to each other. 72 Figure 3.10: Illustration of the effect of inelastic scattering in a TRPES experiment measured for aqueous phenol at same pump wavelength (266 nm) but with different probe wavelength (either 292 nm and 200 nm). The time delay is 1 ps. 73 Figure 4.1: (a) UV absorption spectra of indole in water compared with gas phase and aqueous tryptophan. Two peaks (magenta and blue) represents the wavelength used in our R2PI and TRPES experiments. (b) Deconvolution of the relative contribution of L a (red) and L b (blue) from total absorption spectrum of indole in propylene glycol (a hydrogen-bonding solvent) extracted from Ref. 31 80 Figure 4.2: Resonant two-photon ionization spectra of aqueous indole recorded with a (a) 266 nm ionization and (b) 292 nm ionization wavelength. In panel (a) individual Gaussians used to fit the spectrum are shown in red and green; the total Gaussian fit is represented in blue. For panel (b) the spectrum can be fit with a single Gaussian (in 86 xiv red). Rectangles depict the center of the Gaussian. Note that the R2PI spectra are affected by instrument transmission function at eKE < 0.5 eV. Figure 4.3: Time-resolved photoelectron spectra of aqueous indole plotted with pulse time-ordering (a) 292 nm pump with subsequent 266 nm probe and (b) 266 nm pump and subsequent 292 nm probe. Both beams are vertically polarized with respect the laboratory frame. 89 Figure 4.4: TRPES spectra extracted from full dataset in Figure 4.3 (a) where aqueous indole is photoexcited with 292 nm and photoionized with 266 nm subsequently. Panel (a) displays a developing dip around 1.25 eV in the TRPES spectra within half a picosecond while panel (b) mainly demonstrates the shift of higher eKE photoemission bands towards lower eKE with a small decay in overall intensity. Panel (c) displays the comparison between an experimental time slice at 20 ps (orange), for which an attempt to reproduce the spectrum from two Gaussians is also displayed (see text). This does not replicate the experimental slice fully at low electron kinetic energies. But combining these two Gaussians with an estimated cutoff function for our spectrometer now reproduces the rapidly decaying signal intensity at low electron kinetic energies. In panel (d), the PE signal (on logarithmic scale) is plotted at four selected eKEs as a function of time delay. The experimental conditions were different for short time pump- probe delays (upper panel (d)) and for longer delay times (lower panel (d)) measurements (datasets are recorded on different days). 91 Figure 4.5: TRPES spectra extracted from full dataset in Figure 4.3 (b) where aqueous indole is photoexcited with 266 nm and photoionized with 292 nm subsequently. Panel (a) displays the decay of the highest eKE PE band and the rise of other PE bands— depicted with arrows—resulting into an isosbestic point within half a picosecond while panel (b) shows a clear decay of the same PE bands over a longer 20 ps timescale. Panel (c) displays the comparison between an experimental time slice (2 ps delay) with a constructed time slice composed of two individual Gaussians belonging to 1 La, and a third Gaussian representing the solvated electron along with an estimated cutoff 92 xv function of the spectrometer (see text for details). In panel (d), the PE signal (on a logarithmic scale) is plotted at four selected eKEs as a function of time delay. Inset of the top panel shows the rising signal in the first 2 ps with a linear signal scale. Experimental conditions were different for short time pump-probe delay (upper panel (d)) and for long time pump-probe delay (lower panel (d)) measurements (datasets recorded on different days). The mutual polarization of the two beams is parallel. Figure 4.6: TRPES spectra at selected pump-probe delay for (a) 4 mM indole in water (b) 4 mM indole with 0.5 M KNO3 (quencher) in water, both measured with 266 nm excitation and 292 nm ionization. The nitrate anion quenches both solvated electrons as well as the excited state, resulting in an increased decay rate of all the PE bands as displayed in panel (c). Note that the signal axis in (c) is plotted on logarithmic scale. The polarization of the pump and probe are perpendicular with respect to each other. 93 Figure 4.S1: Cross-correlation between pump and probe pulses in the liquid-jet. 103 Figure 4.S2: Comparison between R2PI spectra with different one-color excitation along with time-resolved spectrum obtained near zero pump-probe delay (resonant 1+1' ionization). The 266 nm beam was horizontally polarized while 292 nm beam was vertically polarized with respect to the laboratory frame. 104 Figure 4.S3: Resonant one-color two-photon ionization spectra of aqueous tryptophan (10 mM) recorded with (a) 266 nm and 292 nm ionization (b) pulses. Individual Gaussian fits are shown in red and green. Note that the R2PI spectra are affected by the instrument cut-off function at eKE < 0.5 eV. 105 xvi Figure 4.S4: TRPES spectra of aqueous tryptophan excited with 292 nm and photoionized with 266 nm are shown in panel (a) while panel (b) displays the TRPES spectra when aqueous tryptophan is photoexcited with 266 nm and photoionized with 292 nm. The arrows in the figure displays the decay of the PE bands. Experimental conditions were different for short pump-probe delay (up to 20 ps) and long pump- probe delay (50 -400 ps) since the measurements were performed on different days. 106 Figure 5.1: Steady state fluorescence spectra of indole in (a) water and in (b) ethanol, Excitation and Emission spectra for indole in (c) water and in (d) ethanol. The different Stokes shift in the two solvents is illustrated by the red dashed lines. The pump wavelengths used for time resolved experiments are marked by the black dotted line in 2D plot. 119 Figure 5.2. Transient absorption spectra of indole in ethanol measured with excitation wavelengths of (a) 292 nm, (b) 266 nm and (c) 200 nm, (d) Comparison of experimental spectrum at 292 nm pump (delay = 1 ps) to the computed S1 ESA spectrum Note: In panel (d), the y-axis for experimental spectrum (wine color) is in absorbance (mOD) while for computed spectrum the y-axis is in TDM 2 121 Figure 5.3: (a) Contour plot of the full 2D transient absorption data set of 16 mM indole in water at 200 nm excitation (b) Spectral slices at a series of early time delays and (c) later time delays (d) experimentally measured spectrum of indole cation, radical and the solvated electron; solvated electron spectrum is scaled by ¼. The black line in fig (b) represents the solvent only signal. The black arrows display the rise of the signal in panel (b) and decay of the signal in panel (c). 123 Figure 5.4: Spectral slices for the aqueous indole (10 mM) when 0.2 M HCl is added in (a) early time and (b) at later time, (c) Temporal profile recorded at 633 nm for aqueous indole with and without HCl added 125 xvii Figure 5.5 Computed spectrum of indole cation and radical in the gas phase at CASPT2 level of theory. 126 Figure 5.6: Comparison between the experimentally measured spectrum of 16 mM indole in water at 200 nm excitation with a constructed spectrum by adding the reported spectra of solvated electron and the indole cation. The ratio of solvated electron and cation is kept same while constructing the spectrum, and a scaling factor is used to scale the spectra of solvated electron and cation before summing them to produce the constructed spectrum (Cation + e - aq). 128 Figure 5.7: (a) Contour plot of the full 2D transient absorption data set of 17 mM indole in water at 266 nm (b) Spectral slices at a series of time delays (c) Contour plot of the TA data set of 10 mM Indole in water when 0.2 M HCl is added to the solution at 266 nm (d) Spectral slices for the aqueous indole when 0.2 M HCl is added to the solution 130 Figure 5.8: (a) Contour plot of the full 2D transient absorption data set of 17 mM indole in water at 292 nm excitation (b) Spectral slices at a series of time delays (c) Contour plot of the TA data set of 10 mM Indole in water when 0.2 M HCl is added to the solution at 292 nm (d) Spectral slices for the aqueous indole when 0.2 M HCl is added 132 xviii List of Tables Table 2.1: Details of the commercial parts for high precision (+/- 0.1 mm) magnet manipulator using ball-bearings linear guides 24 Table 3.1: Influence of inelastic scattering on photoelectron kinetic energy 61 Table 4.1: R2PI peak positions and derived vertical ionization energies for aqueous indole compared with the isolated molecule in the gas phase. All energies in eV . 87 Table 4.2: Binding energies (BE) extracted from corresponding eKE peaks in TRPES spectrum. All energies in eV . 98 Table 4. S1: Photoionization of aqueous tryptophan. All energies are in eV. 105 xix Abstract Photochemistry in the condensed phase is often complicated due to additional solute- solvent interaction compared to isolated environment. Understanding how the electronic structure of a solute in liquid is intricately bound up with the arrangement of a host liquid provides insight into how non-adiabatic photochemistry takes place in the condensed phase. For example, the presence of water provides additional solute-solvent interactions compared to non-polar solvents, changing the stability of ionized products; modifying the energies of low-lying excited valence states as well as moving the point of intersection between potential surfaces. Thus, the locations and topology of conical intersections between these surfaces also change. The overall impact of the polar environment can be to modify the intricate photochemical and non-radiative pathways taking place after photoexcitation. Time-resolved photoelectron spectroscopy (TRPES) in a liquid microjet and Transient absorption (TA) are implemented here to investigate the influence of polar solvents on the electronic structure and dynamics in the excited states for aromatic amino acids and their chromophores. The ultrafast relaxation pathways are also established by varying the excitation wavelengths and polarity of the solvent. 1 Chapter 1. Introduction 1.1 Photophysics and Photochemistry in Daily Life A light driven chemical reaction is very common in the field of photochemistry or photophysics. Photochemical and photophysical processes that we are exposed to in everyday life are typically due to the absorption of photons whose energies lie in the visible to ultraviolet (UV) region of the electromagnetic spectrum. Some of the examples are – perception of our surroundings by photochemical processes occurring in our eyes, the photosynthesis happening in plants, photodamage of DNA by UV light, and the conversion of solar energy into electrical energy by organic solar cells. The above processes are used in chemistry for a multitude of applications, though they are not well understood. By using time-resolved electronic spectroscopy, the multitude of complex pathways that govern these processes can be understood and utilized towards a range of applications. In organic chemistry, photochemists study many types of reactions driven by light such as photoisomerization, photo substitution, photorearrangement, photooxidation-reduction reactions, photodissociation and many more. 1, 2 These organic chemistry reactions are observed in biology as well. For example, vision is result of a natural photochemical reaction due to cis-trans photoisomerization about a double bond in 11-cis retinal present in the visual pigment rhodopsin, a well-studied problem in the scientific literature. 3-5 The sun plays a significant role in chemical processes observed on a daily basis. The ultraviolet part of the solar radiation is mainly responsible for the photochemical reactions happening in most biomolecules. The UV region of the solar spectrum can be divided in three sub regions; UV-A (320 - 400 nm), UV-B (280-320 nm) and UV-C (200-20 nm). The ozone layer in the stratosphere of the earth atmosphere filters out a significant portion of higher energy region 2 UV-B and UV-C which are responsible for the skin cancer and damage of the human immune system. 6 The majority of the UV-A part of the UV spectrum is able to make it to the earth’s surface. Photons in the UV-A can penetrate deep under the skin and cause significant chemical damage. 7 Our DNA bases absorbs the light in UV-B and UV-C region strongly and hence even the residual radiation can cause serious damage to our DNA. 8-12 Similar to DNA, this higher energy UV light also causes the photochemical damage to the proteins which contain strong chromophores such as aromatic amino acids tryptophan and tyrosine. 13 During the past three decades there has been phenomenal growth in the use of ultrafast techniques to understand the mechanistic insight of these reactions happening on molecular level following photoabsorption. Extensive research has been done to reveal the photophysical behavior of nucleobases, nucleotides, aromatic amino acids and their derivatives after irradiating them with ultraviolet light. 14-21 1.2 Photophysics and Photochemistry: A Basic Understanding Following the absorption of light of a suitable energy, the molecule gets promoted from the ground state (minimum energy state) to an electronically excited state. Each excited state has a different electronic structure. The electronic structure determines the physical and chemical properties of the molecule which means that each excited state possess unique chemical and physical properties. The standard UV-Vis spectrum reveals the energy requirement of directly populating an excited state, where the longest wavelength of maximal absorption is typically associated with a HOMO to LUMO transition. The picture becomes complicated when we try to study the properties of the excited state because the excited state is always accompanied by deactivation processes. These processes can be either a photochemical reaction such as 3 dissociation of the molecules, when a bond between atoms breaks, or relaxation/deactivation from the excited state to the ground state with or without the emission of a new photon of different wavelength. To understand a photochemical reaction at the molecular level, the starting point is to separate the nuclear motion from electronic motion. The idea of separation of these two motions is logical in the sense that the masses of electrons are very different than the nuclei. This famous approximation is called Born-Oppenheimer (BO) approximation and scientists almost always use it as a stepping stone to understand chemical processes. 22 Although, this approximation is adequate to describe many chemical processes, it breaks down in many scenarios when there is a coupling between the nuclear motion and an electronic motion. In these circumstances, we cannot visualize these two motions separately. Within the BO approximation, we get an energy, E, when we solve the electronic Schrodinger equation with the coordinates of the nuclei fixed. Now, let’s assume that all the nuclear positions are represented by a collective coordinate, Q. By changing Q, the distances between the nuclei, we obtain different values of E since the electrostatic interaction will also change. Therefore, we get electronic energies as a function of nuclear geometries and its graphical form is called a potential energy surface (PES). To understand the coupling between these two motions in a better way, it is important to understand the concept of the potential energy surfaces (PESs). The PES uniquely represents an electronic state of a molecule, created by the motion of electrons, and it provides the information of change in the energy of the system as a function of change along a nuclear coordinate. The two PESs corresponding to two electronic states may be very far from each other or very close and sometimes they can even cross at a particular nuclear coordinate. Within the BO approximation, the two surfaces remain far apart and there is no 4 coupling between them. These surfaces are often referred to as adiabatic potential energy surfaces (Figure 1, left). However, when two PESs are close to each other, coupling between the two surfaces requires consideration as the true wavefunction can involve mixing the two surfaces (Figure 1, right). In other words, we can say that when nuclei move, the electronic configuration of an excited state may change since electronic motions are coupled with nuclear motions. The latter is the situation where the BO approximation breaks down and the coupling between nuclear and electronic motions need to be considered. These events are known as nonadiabatic transitions. When there is a crossing between two adiabatic PESs, a conical intersection (CI) forms. In case of diatomic molecule, the symmetry of the ground and the excited state must be different to form a CI otherwise avoided crossing occurs. Unlike diatomic system, two surfaces can cross in polyatomic system even if they possess same symmetry and spin multiplicity. The conical intersection assists in making the transition of an excited molecule from one electronic surface to the other very efficient for specific geometric configurations, without absorbing or emitting a photon. Hence, the nonadiabatic transition is also referred as a radiationless transition. The extent of coupling between two surfaces depends on the energy gap between them. The rate of nonadiabatic transition depends on the energy gap. A term known as “derivative coupling” explains the degree of coupling of different electronic states, which is inversely proportional to the energy difference of the two electronic states. 23 Generally, the molecule moves through a CI in tens of femtoseconds. The CI basically acts as a “funnel” for transferring a molecule from one state to the other. In Figure 1.1, S0 and S1 illustrate the ground electronic and first excited state of the molecule and Q is the nuclear coordinate. The purple arrow depicts the electronic transition from ground state to the first excited state while the horizontal lines within each PES depict the 5 vibrational states of the molecule within an electronic state. If the molecule is diatomic then the PES would be a curve and Q signifies the interatomic distance. If the molecule is polyatomic then the curve will transform into a multi-dimensional surface and Q will become a specific coordinate along which a specific transition is allowed. Figure 1.1: Potential energy surfaces featuring adiabatic (left) and nonadiabatic (right) picture 1.3 Photophysical Phenomena in the Excited State In recent years, a lot of scientific and technological efforts have been invested towards the study of molecular behavior in the excited state in both the gaseous and condensed phase. In the gas phase, one can study the intrinsic properties of the molecule free from interactions with any environment, while in solution the photochemistry becomes complicated due to additional solute/solvent interactions. The electronic structure of the solute molecule in solution can be very 6 different than in its isolated form depending on how strongly the solvent interacts with the solute. The solute/solvent interactions can easily alter the electronic structure of the solute molecule. The excited state molecule undergoes various relaxation pathways after electronic excitation due to the absorption of a photon of an appropriate energy. In gas phase, the excited molecule distributes its deposited energy in various vibrational modes by a process known as intramolecular vibrational relaxation (IVR) while in condensed phase the solute molecule can additionally distribute its vibrational energy to the solvent by vibrational energy relaxation (VER). The influence of solvent on the electronic structure of the solute molecule can turn on various competing deactivation pathways which do not exist in the gas phase. 24-26 The extent of solute/solvent interactions depends on the nature of the solvent. Polar solvents affect the electronic structure more compared to non-polar solvents due to the interaction between the solvent dipole and the dipole of the solute electronic states. Each electronic state of the solute has a different geometry and dipole strength with respect to one another and so responds differently to the solvent dipole. For example, a molecule can be easily ionized in water compared to the gas phase because the resultant ionized species would be more stable in water and hence the ionization threshold becomes lower. The influence of solvent (mainly polar) can also be seen by the change in energy of the excited valence states and the movement of the CI with respect to an internal molecular coordinate. Thus, the locations and topology of conical intersections between these surfaces also change, which modify the intricate photochemical and non-radiative pathways taking place after photoexcitation. In Figure 1.2, we can see the influence of solvent on the electronic structure and the mode of relaxations of an excited solute chromophore. The dashed curve represents a repulsive electronic surface with a higher dipole moment and hence this state would be strongly influenced by the presence of the solvent compared to the other states. It can be 7 seen in the picture that due to interactions with its polar environment, the molecule experiences a lowering of the repulsive surface where the population of the excited state can return to the ground electronic state without emitting any photon (non-radiative decay). In the gas phase this relaxation channel is absent. Figure 1.2: Schematic representation of the influence of polar solvent on the excited state dynamics of indole 27 In the following chapters, we seek an understanding of the excited state photophysics and photochemistry happening in the aromatic amino acids (tryptophan and tyrosine) and their intrinsic chromophores, indole and phenol, in the condensed phase. 8 Tryptophan, one of the essential amino acids, is often found in active sites of many enzymes. It absorbs strongly in the ultraviolet region and is fluorescent. The absorption and emission bands for tryptophan are well separated and its fluorescence is highly sensitive to its local environment. 28 Depending on the nature of the surroundings, the fluorescence maximum can change by several nanometers. Along with this change, more importantly the fluorescence quantum yield and the lifetime in the excited state are also subject to change depending on the polarity of the surroundings. These The emission maximum of tryptophan in azurin appears at 308 nm while in glucagon, it shifts to 355 nm due to the difference in solvent exposure of the chromophore. 28 Therefore, it is used as a probe to study the protein structure and dynamics in different environments. 29 Recently, the strong fluorescence property of tryptophan is also being exploited in astrobiological studies to determine the presence of amino acids. 30-32 The complex photophysical behavior of tryptophan is governed by its chromophore indole. Indole is a heteroaromatic molecule composed of a benzene ring fused to a five-membered nitrogen containing pyrrole ring. Indole is also a major constituent of the complex polymer eumelanin which is responsible for the color of our skin, hair and retina and defends against the harmful effects of UV exposure. 33, 34 The other system which draws our attention is phenol. Phenol is the chromophore of the amino acid tyrosine and plays an important role in biological and radiation chemistry. The photophysical behavior of phenol is heavily involved in the reaction center of Photosystem II and in enzymatic catalysis. 35-37 The oxidation of phenolic compounds to its radical forms is a crucial step in many biological and industrial processes. For instance, in the early stages of oxidation of benzene, which is a common aromatic additive of unleaded fuel, phenol and its oxidative product phenoxyl are vital products at combustion temperature. 38, 39 The electronically excited states are 9 very close in energy for these UV-absorbing amino acids and their molecular precursors (indole for tryptophan; phenol for tyrosine). Due to these close lying excited states, the complexity of the excited state dynamics is vastly increased. On top of that, the presence of solvent offers additional complexity by giving rise to solute-solvent interactions and other types of bonding, such as hydrogen-bonding. In order to understand the fast electronic and nuclear dynamics, uncovering the time-evolving electronic character of the excited state turns out to be very important. To understand the complex behavior of these molecules in the condensed phase, it is necessary to have prior knowledge of the excited electronic character. In the past, a lot of efforts (theoretically and experimentally) have been put into the deciphering the intricate photophysical behavior and electronic structure of the excited states for these molecules. For example, Townsend’s group studied indole in a molecular beam by implementing time-resolved photoelectron spectroscopy to establish the electronic structure of an important dissociative state (πσ*). 40 Their work suggests the involvement of this dissociative state at photolysis wavelengths from 249-273 nm. This study is not consistent with the results of other groups who come up with a picture that the direct dissociation of the excited state of indole is not operative via πσ* state only below 4.7 eV excitation energy (263 nm). 41 Kohler’s group utilized pump-probe spectroscopy to study the photoionization dynamics of indole in water at 260 nm excitation and confirmed the generation of solvated electron within their experimental time resolution of 200 fs. 42 The solvated electron is the simplest quantum solute, formed by the trapping of electron in the solvent cage after the ionization event in the excited state. There is a growing body of research pertaining to the photochemistry of phenols, thiophenols and other heteroaromatic molecules such as indole, tryptophan, DNA bases in the gas as well as condensed phase. 43-51 These papers describe the radiative and non-radiative pathways responsible for excites state relaxation in details in a series of publications. 10 The main difficulty in studying these molecules is the proper assignment of the spectral features of the associated species. The difficulty lies in the fact that the spectrum of one species either partially or fully overlap on the other one. For example, in the time-resolved spectra of indole in water measured by pump-probe spectroscopy, the absorption bands attributed to indole radical cations, solvated electrons, radicals, triplets, and the parent excited state absorption all overlap. To reveal the signature of some species, it is possible to use quenching/scavenging experiments. As an example, the spectral signature of the solvated electron can be affirmed by adding a strong acid HCl. The H + coming from the acid will react with solvated electron and as a result the band associated with this species will decrease in intensity. Now, if the spectrum of the geminate partner of solvated electron, cation, is overlapped with the solvated electron, then it is possible to observe the clear feature of cation in the acidic solution since the solvated electron is quenched by the H + (the word “geminate” tells that the source of origin is the same for both species). Now the problem is, we cannot use the trick of quenching for every species. In some cases, theoretical calculation helps to explain the experimental results if the excited electronic states are not very close in energy. In the case of our molecules of interest, theoretical modeling has been proven to be quite difficult. 1.4 Experimental Techniques: Time Resolved Photoelectron spectroscopy and Transient Absorption To study the photophysics of these molecules, we implemented two experimental techniques - transient absorption (TA) using a gravity jet and time-resolved photoelectron spectroscopy (TRPES) using a liquid micro-jet. These two techniques are complementary in nature. In transient absorption, two laser beams are used – one beam (pump) excites some of the molecules from ground state to the excited state and the second beam (probe) probes the excited 11 state. By changing the time delay between pump and probe, the dynamics of the excited state is monitored by measuring the excited state absorption. The selection rule for this electronic transition determines whether the transition is allowed or forbidden. The basis of the selection rules is the “symmetry” of each state. An electronic transition is allowed between two states of different parity if a molecule has a center of inversion. This is known as the Laporte rule. In a lower point group, a specific selection rule exists for each point group. A second rule states that there must be no change in the spin multiplicity (ΔS=0) during an electronic transition. Sometimes, events like vibronic coupling, spin-orbit coupling, and state mixing can make a transition allowed which was previously forbidden by the Laporte and spin selection rules. Apart from having a strict selection rule, transient absorption provides the information of relative energy gap between the electronic states. In the case of TRPES, the first step is same as in TA but in second step the dynamics is monitored by photoionizing the excited state instead of measuring the excited state absorption (Figure 1.3). Although, photoionization is an electronic transition, selection rules are relaxed considerably. Here a ‘one-electron” picture is used to describe the transition. Any transition between neutral and ion is allowed which is associated with the removal of an electron from a single orbital without disturbing the electron occupation of other orbitals during the absorption event. The principal observable of photoelectron spectroscopy is the energy of the valence orbitals from which the electron is ejected through Koopman’s theorem. The Koopman’s theorem states that the binding energy of the outgoing electron is equal to the parent orbital energy. The valence orbitals are directly connected with chemical reactivity and we attempt to read out how these orbitals change along the excited state relaxation pathway within the relevant medium. Hence, the 12 whole idea of using these two techniques was to take advantage of the complementarity of their selection rule in an attempt to reveal the rich photophysics of the biomolecules of interest. Figure 1.3: (a) Transient absorption (TA)- excited state dynamics is monitored by measuring excited state absorption (ESA) (b) TRPES technique- electron kinetic energy (eKE) of photoejected electron from intermediate state is measured to monitor the dynamics of the excited state In this thesis, we tried to assess the role of solvation in controlling the excited state dynamics of phenol, indole and tryptophan as manifested in (i) solvent-induced changes to the solute valence electronic structure, (ii) the role the solvent plays in stabilizing charge-separated states relative to the gas-phase, and (iii) how the presence of solvent molecules modifies the approach to the conical intersections which in turn controls photochemical branching. The experiments were performed at multiple wavelengths spanning the UV-B and UV-C in two solvents ethanol and water. One of the key findings is an observation of ionized products after 13 exciting aqueous indole even if the molecules are pumped with the excitation energy below the reported ionization threshold in the literature. In contrast, no observation of ionized products is observed for indole in ethanol even when the used excitation energy is higher than the ionization threshold. The details of the excited state dynamics of the molecules of interest is further described in the relevant chapters. 1.5 Thesis Plan In chapter 2 of this thesis, recent improvements in the design of the liquid-jet photoelectron spectrometer is described. It also explains the newly built table-top tunable UV source, coupled with TRPES or TA, to achieve higher temporal resolution in experiments. Chapter 3 describes the resonant two-photon ionization and transient photoemission spectroscopy of aqueous phenol. The influence of polarizable solvent on the electronic structure and the dynamics of excited state for phenol is revealed. This chapter also explains the current issues with UV photoelectron spectroscopy such as inelastic electron scattering and transmission function, which need to be addressed to overcome the ambiguities in the experimental results measured by several research groups. In chapter 4 and 5, the rich photophysics of indole is studied by implementing liquid-jet photoelectron spectroscopy and transient absorption. By varying the excitation wavelength, the ultrafast relaxation pathways for aqueous indole have been established. The spectral signature of the fluorescent state and the evidence of solvated electron formation is observed on an ultrafast timescale. Our results also confirm the contribution of the dissociative  * state, which has a vital role in the generation of solvated electron and nonadiabatic dynamics. At this point of time, the whole community using UV photoelectron spectroscopy in liquid microjet is facing the challenge to overcome the inelastic electron scattering. One way to address this issue 14 is by implementing a EUV (extreme UV) probe photon energy, but the signal is lower with EUV probe due to lower probe depth and different ionization cross section. More importantly, UV photon helps to exploits the advantage of chemical selectivity and sensitivity if we compare R2PI with XPS. Therefore, the community needs to come up with a transformation method to take care of inelastic electron scattering in order to extract the maximum information out of the UV photoelectron spectrum. 15 1.6 References 1. V. Balzani; P. Ceroni; A. Juris, John Wiley & Sons: 2014. 2. P. Klán; J. Wirz, John Wiley & Sons: 2009. 3. F. Gai; K. Hasson; J. C. McDonald; P. A. Anfinrud, Science 1998, 279, 1886-1891. 4. K. Hasson; F. Gai; P. A. Anfinrud, Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 15124-15129. 5. S. Schenkl; F. Van Mourik; G. Van der Zwan; S. Haacke; M. Chergui, Science 2005, 309, 917-920. 6. D. González Maglio; M. Paz; J. Leoni, BioMed research international 2016, 2016. 7. D. L. Narayanan; R. N. Saladi; J. L. Fox, Int. J. Dermatol. 2010, 49, 978-986. 8. R. P. Rastogi; A. Kumar; M. B. Tyagi; R. P. Sinha, J Nucleic Acids 2010, 2010. 9. J. Cadet; E. Sage; T. Douki, Mutat. Res-Fund. Mol. M. 2005, 571, 3-17. 10. D.-P. Häder; R. P. Sinha, Mutat. Res-Fund. Mol. M. 2005, 571, 221-233. 11. J. Cadet; M. Berger; T. Douki; B. Morin; S. Raoul; J.-l. Ravanat; S. Spinelli, Biol. Chem. 1997, 378, 1275-1286. 12. E. C. Friedberg, Nature 2003, 421, 436. 13. L. I. Grossweiner, Curr. Eye Res. 1984, 3, 137-144. 14. M. Barbatti; A. J. A. Aquino; J. J. Szymczak; D. Nachtigallová; P. Hobza; H. Lischka, Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 21453-21458. 15. I. Hünig; C. Plützer; K. A. Seefeld; D. Löwenich; M. Nispel; K. Kleinermanns, ChemPhysChem 2004, 5, 1427-1431. 16. E. Nir; K. Kleinermanns; L. Grace; M. S. de Vries, J. Phys. Chem. A 2001, 105, 5106- 5110. 17. E. Pluhařová; P. Jungwirth; S. E. Bradforth; P. Slavíček, J. Phys. Chem. B 2011, 115, 1294–1305. 18. E. Pluhařová; C. Schroeder; R. Seidel; S. E. Bradforth; B. Winter; M. Faubel; P. Slavíček; P. Jungwirth, J. Phys. Chem. Lett. 2013, 4, 3766. 19. M. A. Yandell; S. B. King; D. M. Neumark, J. Am. Chem. Soc. 2013, 135, 2128-2131. 20. D. Creed, Photochem. Photobiol. 1984, 39, 537-562. 21. D. Creed, Photochem. Photobiol. 1984, 39, 563-575. 22. M. Born; R. Oppenheimer, Annalen der physik 1927, 389, 457-484. 23. S. Matsika; P. Krause, Annu. Rev. Phys. Chem. 2011, 62, 621-643. 24. G. Kumar; A. Roy; R. S. McMullen; S. Kutagulla; S. E. Bradforth, Faraday Discuss. 2018, 212, 359-381. 25. T. A. Oliver; Y. Zhang; A. Roy; M. N. Ashfold; S. E. Bradforth, J. Phys. Chem. Lett. 2015, 6, 4159-4164. 26. Y. Zhang; A. Roy; S. Das; T. A. A. Oliver; M. N. R. Ashfold; S. E. Bradforth, 2010, in preparation. 27. A. Sobolewski; W. Domcke; C. Dedonder-Lardeux; C. Jouvet, Phys. Chem. Chem. Phys. 2002, 4, 1093-1100. 28. J. T. Vivian; P. R. Callis, Biophys. J. 2001, 80, 2093-2109. 16 29. P. R. Callis; T. Liu, J. Phys. Chem. B 2004, 108, 4248-4259 30. F. Scappini; M. Capobianco; F. Casadei; R. Zamboni; P. Giorgianni, Int. J. Astrobiol. 2007, 6, 281-289. 31. E. Eshelman; M. Daly; G. Slater; E. Cloutis, Planet. Space Sci. 2015, 119, 200-207. 32. L. R. Dartnell; T. A. Roberts; G. Moore; J. M. Ward; J.-P. Muller, PLoS One 2013, 8, 75270. 33. F. Solano, New J. Sci. 2014, 2014. 34. K. Wakamatsu; Y. Nakanishi; N. Miyazaki; L. Kolbe; S. Ito, Pigm Cell Melanoma R 2012, 25, 434-445. 35. B. Kok; B. Forbush; M. McGloin, Photochem. Photobiol. 1970, 11, 457-475. 36. B. A. Diner; D. A. Force; D. W. Randall; R. D. Britt, Biochemistry 1998, 37, 17931-17943. 37. T. J. Meyer; M. H. V. Huynh; H. H. Thorp, Angew. Chem. Int. Ed. 2007, 46, 5284-5304. 38. C.-Y. Lin; M. Lin, J. Phys. Chem. 1986, 90, 425-431. 39. C. Y. Lin; M. Lin, Int. J. Chem. Kinet. 1985, 17, 1025-1028. 40. R. Livingstone; O. Schalk; A. E. Boguslavskiy; G. Wu; L. Therese Bergendahl; A. Stolow; M. J. Paterson; D. Townsend, J. Chem. Phys. 2011, 135, 194307. 41. B. Cronin; M. G. Nix; R. H. Qadiri; M. N. Ashfold, Phys. Chem. Chem. Phys. 2004, 6, 5031-5041. 42. J. Peon; G. C. Hess; J.-M. L. Pecourt; T. Yuzawa; B. Kohler, J. Phys. Chem. A 1999, 103, 2460-2466. 43. A. L. Devine; M. G. D. Nix; B. Cronin; M. N. R. Ashfold, Phys. Chem. Chem. Phys. 2007, 9, 3749-3762. 44. R. N. Dixon; T. A. A. Oliver; M. N. R. Ashfold, J. Chem. Phys. 2011, 134, 194303. 45. S. J. Harris; D. Murdock; Y. Zhang; T. A. A. Oliver; M. P. Grubb; A. J. Orr-Ewing; G. M. Greetham; I. P. Clark; M. Towrie; S. E. Bradforth; M. N. R. Ashfold, Phys. Chem. Chem. Phys. 2013, 15, 6567-6582. 46. G. A. King; A. L. Devine; M. G. D. Nix; D. E. Kelly; M. N. R. Ashfold, Phys. Chem. Chem. Phys. 2008, 10, 6417-6429. 47. G. A. King; T. A. A. Oliver; M. G. D. Nix; M. N. R. Ashfold, J. Phys. Chem. A 2009, 113, 7984-7993. 48. T. A. A. Oliver; Y. Zhang; M. N. R. Ashfold; S. E. Bradforth, Faraday Discuss. 2011, 150, 439 - 458. 49. Y. Zhang; T. A. A. Oliver; M. N. R. Ashfold; S. E. Bradforth, J. Phys. Chem. Lett. 2015, 6, 4159-4164. 50. Y. Y. Zhang; T. A. A. Oliver; M. N. R. Ashfold; S. E. Bradforth, Faraday Discuss. 2012, 157, 141. 51. M. G. D. Nix; A. L. Devine; B. Cronin; M. N. R. Ashfold, J. Chem. Phys. 2007, 126, 124312. 17 2. Experimental Methods 2.1 Overview Photoelectron spectroscopy in a liquid jet is a strong technique to investigate the excited state dynamics of polyatomic molecules in the condensed phase. 1-6 Although photoelectron spectroscopy in the gas phase is a well-established technique, it’s realization in the condensed phase has gained momentum very recently. 7, 8 Photoelectron signals are sensitive to changes in the electronic configuration of the molecules in excited state, which makes it ideal to study the excited state processes. The principle of photoelectron spectroscopy is rather simple. If a photon carries higher energy than the binding energy of an electron in a molecule or atom, then the electron will be ejected, and the kinetic energy of the ejected electron will be equal to the difference between the photon energy and the binding energy based on the conservation of energy principle, via the following equation: 𝑒𝐾𝐸 = ℎ𝑣 − 𝑒𝐵𝐸 ………………………………. (1) The term eKE stands for the electron’s kinetic energy and eBE stands for electron binding energy in Equation 1. Hence, the photoelectron spectrum will represent the intensity of ejected electrons as a function of their kinetic energy. A simple picture to understand the photoelectron spectrum is based on “Koopmans’ theorem” which states that the emission of an independent valence electron occurs without any simultaneous reorganization of the electrons in the resulting cation core. The relative strengths of the different electronic bands depend on their ionization cross sections. The ionization cross section, determines the ionization probability to different ionic states, depends on the nature of the molecular orbital. 9 The intensity within a given electronic band will depend on the Franck- Condon factor between the neutral wave function and the cationic wave function. 18 Photoelectron spectroscopy can also be used in a time resolved fashion, known as time- resolved photoelectron spectroscopy or TRPES, such that a system can be initially excited via a pump pulse and then subsequently probed by another laser pulse leading to electron ejection after a given amount of pump-probe delay. There are multiple reviews in the literature which explain this technique in detail. 10-12 Figure 2.1 provides a pictorial representation for the excited state dynamics probed via TRPES. The change in the electronic structure of a given electronic state can be reflected in kinetic energy of the ejected electron from that state. Hence, by monitoring the binding energy as a function of time during a photophysical phenomenon, the excited state evolution can be revealed. Figure 2.1: Schematic representation of TRPES. The potential energy surfaces for the ground state (M), the excited state (M*) and the cationic state (M + ) are illustrated. Right panel depicts the temporal evolution of the excited state in terms of electron kinetic energy of the outgoing electrons. Due to dramatic change in the displacement along nuclear coordinate, the blue colored PES band will be much broader with structured vibrational progression compared to the red colored PES band. 19 We implemented photoelectron spectroscopy (PES) in liquid phase. Photoelectron spectroscopy has been known for a long time as a popular technique to study molecules in both isolated environments and inside solid substrates. 13-18 Implementation of PES with liquids has been revolutionized in last 10-15 years. In 1973, Siegbahn and coworkers were the first to perform photoelectron spectroscopy in liquids, but their technique was only limited to liquids with low vapor pressure. 7 Later, Faubel and coworkers demonstrated the use of the liquid microjet technique, inserted into vacuum for use with highly volatile liquids. 19 Currently, several research groups have successfully set up liquid microjet photoelectron spectroscopy apparatuses and widely use this technique to measure the binding energies and dynamics of choice molecules in the condensed phase. 11, 12, 20, 21 A brief description of the spectrometer is presented in this chapter. The complete description of the design of the spectrometer and the implementation of liquid jet is given in Anirban Roy’s thesis, a previous group member . 22 2.2 Liquid Microjet Photoelectron Spectrometer 2.2.1 Brief Description of the Spectrometer The current experimental set up has been improved in several ways from the version described earlier. A dual-piston reciprocating solvent pump designed to eliminate mobile-phase pulsed flow in high pressure liquid chromatography (Shimadzu LC-20AD) is used to inject solution through a 25 μm diameter fused silica jet nozzle into a vacuum chamber, forming a jet with a flow velocity of ~20 m/s. It is important to use a dual-piston assembly rather than a single- piston pump since dual piston models reduce pulsation in the flow as the piston retracts. The solvent pump is run in constant flow rate mode (0.5 ml min -1 ) to prevent fluctuations in flow which 20 may bring about signal breathing. Typical pressures supplied by the solvent pump varies between 0.1 - 0.3 MPa to maintain target flowrate. The vacuum chamber is split into two sections, a main chamber and a time of flight tube; both are evacuated by separate turbomolecular pumps. The main vacuum chamber is pumped by a large turbomolecular pump (Hi Pace Turbo 1500; 1500 l/s; Pfeiffer Vacuum) and two liquid nitrogen cooled cryotraps. During liquid jet operation in the main chamber, the traps combined with the turbomolecular pump achieve an operating pressure of ~3x10 -4 mbar in an aqueous liquid- jet experiment, though this pressure has been observed to increase with more volatile solvents such as ethanol and methanol. The time-of-flight (TOF) tube of ~ 50 cm length (diameter 10 cm) is evacuated with two turbomolecular pumps (Hi Pace 300; 300 l/s, Pfeiffer) situated behind the detector at the end of the drift tube. The TOF tube is separated from the main chamber by a skimmer and a manually-operated mechanical shut-off valve, which allows the flight tube to be kept under vacuum at all times. Photoelectrons are detected and analyzed with a home-built magnetic bottle TOF photoelectron spectrometer. Utilization of a magnetic bottle helps to ensure high electron collection. The permanent magnet in the magnetic bottle configuration is composed of a soft-iron tip (7 mm height, 45 o angle at the tip, Sargent-Welch) mounted in the center of a cylindrical 0.7 T SmCo magnet in the main chamber (two SmCo magnets of 25 mm diameter and 15 mm length). The soft-iron tip is secured in the center of the SmCo magnet by an aluminum jacket such that the position does not change between experiments. The SmCo/soft-iron tip assembly is mounted on a PEEK rod and is attached to a load-bearing XYZ manipulator via an edge-welded bellows for alignment purposes as sown in Figure 2. The second part of the magnetic bottle concerns the magnetic field inside the flight tube and consists of a copper solenoid wound directly around 21 exterior the flight tube, from which a 1 mT magnetic field (Low Magnetic Field) emanates when 3 amps of current is passed through the solenoid. The magnitude of the fields is sufficient to screen the electron from the Earth’s magnetic field. 2.2.2 Improvements from the Previous Version As mentioned above, the permanent magnet is mounted on an XYZ manipulator which allows for precise alignment of the magnet with the skimmer orifice and jet. A proper alignment increases the collection efficiency of the photoelectrons. Previously, the magnet was mounted on a low-quality manipulator which did not allow the translation of the magnet in all three-direction and did not reliably position the magnet repeatedly after each pump down of the chamber. These issues led to reduced collection efficiency and poor resolution of vibronic structure in vapor phase photoelectron spectra. In order to understand the influence of misalignment of the magnet, first we have to understand the working principle of the magnetic bottle time of flight. The main advantage of the magnetic bottle is enhanced collection efficiency without altering the kinetic energy of the photoelectrons. In a magnetic bottle geometry, a strong inhomogeneous diverging magnetic field is generated in the source chamber by a permanent magnet, or an electromagnet, and a weak homogeneous field is generated in the time of flight tube by a solenoid coil. Upon generation, a given ejected electron will spiral around the magnetic field lines due to the applied Lorenz force. It is guided by the field of the strong magnet through the skimmer into the region of uniform parallel magnetic field lines inside the TOF tube. One component of the Lorentz force acts along the TOF axis which increases the longitudinal component of the velocity and subsequently the transverse component decreases in order to conserve the energy and the momentum of the electrons. This leads to the parallelization of the electrons in the low field region within the flight 22 tube. The important parameter to maintain here is the magnetic field gradient in the two regions such that even the electrons which are emitted perpendicularly to the field lines in the high field region become almost parallel to the field lines in the low field region. The advantage of enhanced collection efficiency of the magnetic bottle geometry comes at the cost of losing the angular information of the ejected electrons. The energy resolution is also sacrificed to some extent due to the extra distance traversed by some of the electrons before heading towards skimmer orifice by the applied Lorentz force. Since the interaction region is a small distance away from the highest B-field, most of the electrons which are emitted away from the detector can be turned around in the direction of detector by the influence of the highest B-field. In this way the collection efficiency of electrons increases significantly compared to field-free TOF spectrometer. In our spectrometer, the efficiency improved from 0.04% (estimated based on the geometry of TOF tube and the diameter of MCP) to 50% after implementing the magnetic bottle geometry. The electrons which are ejected away from the detector and parallel to the B-field lines cannot be turned away. 21 2.2.3 New Design of the Magnet Paraphrasing the words of Kruit and Read, 23 three conditions must be fulfilled for good energy resolution in a time of flight experiment: (i) to achieve parallelization of electron trajectories, the magnetic field in the source chamber should be much stronger than the magnetic field in the flight tube (ii) the distance of the magnetic field gradient (from high to low) must be shorter than the flight tube (iii) the magnitude of change in magnetic field during one orbit (one circular rotation) of electron must be negligible compared to the total field. Additionally, the magnetic field in the drift/flight tube should be at least 0.6 mT to overrule the effect of any stray electrical or magnetic field lines. 24 The design of the magnet and new manipulator implemented in our experiment is sketched in Figure 2.2 and Figure 2.3. To increase the magnetic flux, a soft 23 iron cone has been placed on top of the permanent magnet. To guide the magnetic field lines properly, the base of the soft iron cone has been matched to be concentric with the center of the magnet and the other end of the magnet is matched with the base of the soft iron rod with the help of the aluminum jacket (4 cm length). In this way, the magnetic field lines are not distorted. The strength of the magnetic field (0.6 T, measured using Hall probe) is maximum at the tip of iron cone and decreases gradually along the distance towards the flight tube. Figure 2.2: Newly designed magnet assembly: (a) z-axis actuator, (b) edge-welded bellows, (c) PEEK rod (polyetheretherketone), (d) soft iron rod, (e) permanent magnet inside aluminum jacket, (f) soft iron conical tip, (g) x-axis actuator, (h) y-axis actuator 24 Table 2.1: Details of the commercial parts for high precision (+/- 0.1 mm) magnet manipulator using ball-bearings linear guides Item Qty: Catalog # Description Catalog # Vendor 1 1 471022 Welded Bellows, 1.500"ID, 2"Stroke 471022 Mdc Vacuum Products LLC 2 1 BM25.40 Standard Resolution Micrometer, 40 mm Travel, 100 lb. Load Capacity BM25.40 Newport Corporation 3 2 00901223 Mitutoyo - 1 Inch, 49mm Zero- Adjustable Thimble, 8mm Diameter x 38.7mm Long Spindle, Mechanical Micrometer Head 00901223 MSC Industrial Supply 4 12 8438K3 Corrosion-Resistant Ball Bearing Carriage, Load Fastening from Top, for 9mm Wide Rail 8438K3 McMaster- Carr 5 1 9246K684 6061 Aluminum, 2" Thick, 6" x 48" 9246K684 McMaster- Carr 6 4 7806K63 High-Performance Bronze Thrust Ball Bearing for 8 mm Shaft Diameter, 16 mm Outside Diameter 7806K63 McMaster- Carr 7 3 6725K43 9mm Wide Guide Rail for Ball Bearing Carriage, Length 220 mm 6725K43 McMaster- Carr 8 1 7269K31 PEEK Rod, Beige, 9" Long, 1" Diameter 7269K31 McMaster- Carr 9 2 9088K15 Flange for Stainless Steel Tubing, Socket-Connect Adapter, 2-3/4" Flange OD 9088K15 McMaster- Carr 10 1 8909K17 Easy-to-Machine Cast Iron Rod, Oversized, 1" Diameter 8909K17 McMaster- Carr 25 Figure 2.3: Design diagram of the new manipulator along with magnet assembly The homogeneous magnetic field (1 mT) is produced by the solenoid coil in the flight tube by sending a current of nearly 3 Amperes. Although the homogeneous magnetic field can be bit lower than 1 mT, but it should not be too low (⁓ 0.25 mT) otherwise slow-moving electrons will be strongly affected by the stray fields present in the surroundings. In order to avoid the interference of the weak magnetic field inside the flight tube from the earth magnetic field, the solenoid has been wrapped by Mu-metal foil without re-annealing once bent around the flight tube circumference. 26 2.2.4 Precise Alignment of the Magnet The magnet should be properly aligned along the axis of the flight tube mounted on the other side of the interaction region with the help of high precision manipulator (+/- 0.1 mm precision) by looking at the REMPI photoelectron signal at the computer. The collection efficiency of the photoejected electrons varies with the distance of the magnetic tip from the interaction region. We observed that the spectral shape of the photoelectron spectrum changes depending on where the magnetic tip has been positioned with respect to the interaction region. To understand the change in spectral shape due to change in position of the magnet, Xe gas was pumped into the chamber through a jet nozzle of 35 μm diameter nearby the skimmer. It was possible to ionize Xe to its cationic states by multi-photon ionization using the UV photon energies available by harmonic generation of the Ti:Sa amplifier. The three-photon ionization of Xe was recorded using 266 nm (Ehѵ = 4.66 eV, power = 2 μJ). With 266 nm laser pulses, Xe can be ionized to its cationic states 2 P3/2 (ionization potential Ip =12.1 eV) and 2 P1/2 (Ip =13.4 eV). 25 Therefore, the electron kinetic energy (eKE) of the peaks corresponding to these two states should occur around 1.9 eV and 0.6 eV respectively. The TOF spectra measured as a function of translating the magnet away from the liquid-jet along the time of flight axis, henceforth defined as Z. To increase the translation range of the z-axis actuator, we also placed a cylindrical disc of 0.5-inch thickness (aluminum disc from Newport which we use to increase the length of optical pedestal) before the tip of the actuator. The TOF spectra recorded at multiple values of Z are converted into the eKE domain (shown in Figure 3) with crude calibration parameters (length of flight tube = 50 cm, timing offset = 2 ns and energy offset = 0 eV) and employing the appropriate Jacobian. In our spectra, we mainly observed a peak corresponding to 2 P3/2 state around 1.7 eV in close agreement with the reported value in the literature. 25 The peak corresponding to 2 P1/2 state is apparently masked by the transmission function 27 of the instrument or has very low cross section. The transmission/cut-off function determines the detection efficiency as a function of kinetic energy of the ejected electrons. Typically, the detection efficiency of a magnetic bottle spectrometer drops at low kinetic energy from slow electrons not reaching the MCP detector, which is described by the transmission function/cutoff-threshold. The concept of transmission function is explained in detail in Chapter 3. Although, the actual form of the transmission function (or more crudely the cut-off threshold) is so far undetermined in our experimental set-up, it is expected, based on observed peak shapes from multiple differing experimental target systems, that the transmission drops rapidly below 0.5 eV electron kinetic energy. From the Figure 2.4 (a), it is apparent that the peak at 1.7 eV starts splitting into two peaks with increase in the distance between the tip of the magnet and the interaction region. Perhaps the simulation of electron trajectories, as dictated by the magnetic field lines, could have helped us to figure out the reason of this peak splitting, but this is a separate project outside the scope this thesis. At this point of time, we speculate that the difference in the time of flight (TOF) between the electrons emitted towards the detector and away from the detector may result in the splitting of the peak. When the magnet is retracted back (away from the interaction region), the electrons ejected in the 2π solid angle opposite to the detector are required to travel a greater distance before they are turned back by the region of highest magnetic field and this results into the spread in the TOF. 24 Due to this scenario, a dip starts appearing in the overall photoelectron signal separating the photoelectrons ejected towards the detector (lower TOF time) and away from the detector (higher TOF time). The appearance of the dip is also accompanied by a drop in the collection efficiency of the photoelectrons. The photoelectrons which have larger velocity component, perpendicular to the TOF axis, may fail to enter the TOF tube resulting into a decrease in the photoelectron signal. 28 Riley et al. also observed an increase in signal due to scatter photoelectrons when the magnet is translated away from the detector, along with peak splitting in the Xe photoionization signal. 26 Our expectation was that the peak corresponding to the photoelectrons emitted in the forward direction will be unaffected when the magnet is translated away from the detector while the TOF for the photoelectrons ejected in the backward direction will increase. But, from our experimental spectra, it seems like forward photoelectrons get an extra push resulting into higher eKE while the eKE for the backward electrons decreased as expected. Separately, in a liquid-jet environment, where photoelectron bands are already much broader (0.7 -1 eV), the difference in electron trajectories for electrons emitted directly toward the detector and those turned around by the permanent magnet will be manifested as additional spectral broadening. Figure 2.4: Three-photon ionization of Xenon. (a) eKE spectra measured at multiple values of Z (Z is the distance between the magnetic-tip and the liquid-jet), (b) TOF spectra measured at multiple values of Z. The liquid jet is usually parked at ⁓2mm from the skimmer orifice. The value of Z (in the inset of figure) signifies the distance between the jet and the tip of the soft iron cone of the magnet. 29 Another effect which we have noticed while running liquid-jet experiments is the change in electron counts with change in distance between the magnet and the interaction region. We observed the maximum signal when the magnet is 4-5 mm away from the jet. The total counts decrease both when i.) the magnet is close (<1 mm) to the jet and ii.) far away from the jet (>5 mm). Since, the magnetic field is properly confined nearby the tip so when we move the magnet away from the jet, the electrons travel larger distances initially away from the skimmer before being turned around. The electrons, which are ejected with their velocity vectors close to the TOF axis, make it to through the nozzle whereas those ejected perpendicular to the TOF axis are also parallelized along the TOF axis but at a larger off-axis distance not falling in the width allowed by the skimmer. Thus, the center of the peak (corresponding to intermediate pathlengths) are eliminated but the purely forward and purely back ejected electrons (representing the shortest and longest pathlengths) are still captured, leading to the observed splitting in gas phase peaks and broadening in the liquid phase peaks. Due to this effect, overall electron counts will also decrease. When the magnet is moved very close to the interaction region, the size of the interaction volume which is being probed (actual interaction volume is determined by the overlap of the jet and the laser beam) gets smaller and thus the number of counts. 24 For all our experimental measurements, we positioned the magnet (⁓4 mm from the liquid jet) where we observed the maximum photoelectron counts. 2.3 Calibration of the Spectrometer The calibration of the photoelectron spectrometer is performed on a day to day basis, by using three-photon photoemission of water vapor. To extract the calibration parameters, the vibronic bands are fitted with the equation given below. 27 30 𝐸 = 1 2 𝑚 𝑙 2 (𝑡 − 𝑡 0 ) 2 − 𝐸 0 The typical values of the parameters we get in our experiments are: flight length (𝑙 ) = 47±2 cm, timing offset 𝑡 0 = 50±3 ns and energy offset 𝐸 0 = 40±10 meV (Figure 2.5). Figure 2.5: Calibration of the photoelectron spectrometer with water vapor in presence of the water jet. a) Schematic of energy levels involved in the three-photon photoionization of water using 266 nm ultrashort pulses (~200 fs). The observed structure in the time-of-flight spectra (b) represents transitions to different vibrational levels on the cationic surface (D0). The time of flight values for each peak is calibrated against the literature kinetic energy values 28, 29 and fitted with eqn. 1 (c). The calibrated kinetic energy spectrum is shown in (d) with the vibrational assignments (v1 v2 v3) of the X 2 B1 state of H2O + . 28 The ionization energies for (000), (010), (100), (110) and (200) bands are 12. 615 eV, 12.792 eV, 13.017 eV, 13.191 eV and 13.408 eV respectively. 29 31 The spectrum for the calibration is recorded with the water jet flowing inside the source chamber. However, the jet is moved away from the laser focus (towards the permanent magnet) when calibrating the spectrometer to only ionize the water vapor jacket that surrounds the liquid jet. By implementing this calibration procedure, we do not address the issue of electrokinetic charging of the liquid separately. There is a buildup of surface charge, named as streaming potential, when an electrolyte flows through a nozzle. This surface charge can accelerate or deaccelerate the electron ejected from the liquid impacting its true kinetic energy. The sign and magnitude of the streaming potential depends upon the diameter of the nozzle, velocity of the liquid jet, nature of the solvent and ion concentration. To reduce the effect of the streaming potential, a small amount of the electrolytes (e.g., 30 mM NaCl) is always added. 30 The magnitude of the streaming potential can vary in the range of 0.1-0.3 eV for aqueous and alcohol solutions, as reported by several studies. 31-34 Ideally, calibration should be carried out in the complete absence of a liquid streaming potential by introducing the gas as in the xenon experiments above and then separately accounting for the streaming potential. 35, 36 Tang et al. and Riley et al. multi-photon ionization of NO to calibrate their TOF. 35, 36 They extract the calibration parameters in a similar way like ours but instead of ionizing water vapor and fitting vibronic states of H2O + , they fit vibronic states of NO + . The measurement of streaming potential was also performed separately in these two studies. Riley et al. adopted the same method as suggested by Tang et al. to quantify the streaming potential. 35 The 2+1 REMPI spectra of Xe was recorded at the original ionization point L, where L is the distance between the skimmer and the ionization point where the liquid jet was originally placed. The liquid jet was moved away from the original ionization point by x and the REMPI spectra of Xe was recorded at various distances between the ionization point and the 32 liquid-jet. The eKE of photoelectron was plotted as a function of x and the value of the streaming potential was extracted by using the equation given below (Figure 2.6). Figure 2.6: Illustration of the position of the skimmer, the ionization point and the liquid microjet while measuring the REMPI of Xe to extract the streaming potential. The equation utilized to extract the streaming potential is also shown in the right panel. In the above equation: eKEtrue = hѵ – VBE, L= distance between the skimmer and the ionization point, x = position of liquid-jet from the ionization point, V = unwanted field which can lead to PKE shift. It should be noted that while measuring the streaming potential, the magnet was moved away from the skimmer and hence a new calibration was performed with the new magnet position. 2.4 New Ultrafast Tunable Laser Source The advancement of ultrafast techniques has revolutionized the field of molecular dynamics. Ultraviolet (UV) laser pulses are used in all the experiments performed in this thesis. In time-resolved photoelectron spectroscopy, both pump and probe pulses are UV. One UV source is tunable (235 nm – 350 nm), generated by the second harmonic of a home-built non-collinear optical parametric amplifier (NOPA) output (470 nm – 750 nm, NOPA manual) 37 while the other UV source has a fixed wavelength of 266 nm or 200 nm, generated by the third or fourth harmonic 33 of the output of the output of a 35-fs Ti:Sapphire amplifier system (Coherent Legend Elite USP, 1 kHz repetition rate). 2.4.1 Non-Collinear Optical parametric amplifier (NOPA) Working Principle: Wavelength tunability of a laser source is very important in the field of ultrafast spectroscopy. Tunability can be achieved by utilizing a nonlinear process called three- wave mixing. An optical parametric amplifier (OPA) uses this phenomenon to generate wavelengths of different colors. During the parametric amplification in an OPA, an intense pump beam (𝜔 𝑝 ) amplifies a low intensity beam, signal (𝜔 𝑠 ) in addition to a third beam, idler (𝜔 𝑖 ), at the difference frequency of pump and signal inside a nonlinear optical crystal. There are two conditions which needs to be fulfilled for efficient three-wave mixing: (i) Energy conservation: ℏ𝜔 𝑝 = ℏ𝜔 𝑠 + ℏ𝜔 𝑖 (ii) Momentum conservation (phase matching condition): ℏ𝑘 𝑝 = ℏ𝑘 𝑠 + ℏ𝑘 𝑖 In the above equations, 𝑘 𝑝 , 𝑘 𝑠 and 𝑘 𝑖 represent the wavevectors of pump, signal, and idler, respectively. By rotating the nonlinear crystal, the phase velocities of the pump, signal and idler can be matched. However, the efficient phase matching of these three beams cannot ensure the matching of their group velocities. These three beams propagate at different speed inside the crystal. It has been observed that if the pump and signal travel through the crystal in a collinear fashion, they become quite broad temporally by the time the beams exit the crystal. Figure 2.7(a) represents the temporal walk off between the signal (green) and the idler (red). The idler has higher group velocity than the signal and hence it travels faster than signal. The temporal walk off between these beams result into an amplified signal with narrow bandwidth, which will correspond to a long, transform limited pulse width of the signal. To address the issue of pulse lengthening, a 34 noncollinear geometry between the pump and the seed (a signal from the white light often referred as the seed due to weak intensity) can be adopted. Due to the noncollinear geometry between the pump and the signal when entering the crystal, the temporal walk off between these beams can be alleviated resulting into broadband amplification of the signal beam (Figure 2.7(b)). 38, 39 Figure 2.7: Amplification of a weak signal with the help of an intense pump beam. (a) the pump (purple) and the weak signal (green) beam entering the crystal in a collinear fashion resulting into temporal walk-off between the signal and the idler (red) (b) the pump (purple) and the signal (green) entering the crystal in a noncollinear geometry and hence while propagating through crystal the signal (green) and the idler (red) are always temporally overlapped resulting into broadband amplification of the signal. 38 The conception of the picture is adapted from the manual of the NOPA. 37 The angle between the pump and the seed (ψ) in a noncollinear geometry creates a different angle (Ω) between the signal and the idler, as shown in equation below. 𝛺 ≈ Ѱ. (1 + 𝜆 𝑖𝑑𝑙𝑒𝑟 /𝜆 𝑠𝑖𝑔𝑛𝑎𝑙 ) At angle Ω, the projection of the idler group velocity onto the signal k vector, which is the propagation direction, matches the signal group velocity. 𝑣 𝑔𝑟𝑜𝑢𝑝 ,𝑖𝑑𝑙𝑒𝑟 . cos(𝛺 ) = 𝑣 𝑔𝑟𝑜𝑢𝑝 ,𝑠𝑖𝑔𝑛𝑎𝑙 35 In this scenario, the idler photons are generated at the same location where signal photons are being amplified. This results in no lengthening of the pulses coming out from the crystal. Although, a broadband amplification of the signal can be achieved by the noncollinear geometry, but it also results into distortion in the aspect ratio of the signal. The signal can look a bit elliptical (vertically) in shape since the idler remains displaced transversally during the propagation inside the crystal. The resulting output pulse in this noncollinear geometry can be compressed to as short as 10 fs, successfully shown by Riedle et al. 38 Installation and Alignment of NOPA: The beam size of the 800 nm coming out of the Legend amplifier is around 4.6 mm in diameter. This beam size was reduced to about one half before entering the NOPA by using a compact folded reflective Galilean-type telescope. A reflective geometry (as shown in the Figure 2.8) was preferred over refractive to avoid the beam passing through more glass material which temporally broadens the pulse. The reason to choose Galilean configuration rather than a Keplerian one, is to make a telescope that has no focal point between the eyepiece (concave mirror, f=50 cm) and the objective lens (Plano-concave lens, f = -25 cm). An intermediate focus between the eyepiece and objective optics is undesirable as it may lead to mode collapse when the focused intensity of the 800 nm is above critical intensity for self-focusing in air. A down-collimated beam free of spatial distortion is possible with a Galilean telescope. 36 Figure 2.8: Reflective telescope (Galilean type), 1: Concave mirror of focal length of 50 cm, 2: A plane mirror (800 nm HR), it was used to provide a compact design by folding the laser beam 3: Plano-concave lens of focal length -25 cm mounted on compact dove-tail linear stage (9.5 mm travel, Newport) for fine adjustment The optical layout of NOPA is shown in Figure 2.9 with all the components. It is a two- stage NOPA. In the first stage, the seed is amplified in a 1mm thick BBO crystal. Here, the broadband phase matching is achieved by the noncollinear configuration between the 400 nm (pump) and the seed (white light). While aligning the beams, the goal should be to maximize the bandwidth measured on a spectrometer rather than power by adjusting the temporal delay between the pump and seed with the help of translation stage (mounted at 8 in Figure 2.9(a)) and optimizing the angle between the pump and white light. In the second stage, a 2 mm thick BBO crystal is used to amplify the signal generated in first stage. The pump pulse should be focused before the BBO crystal while setting up the NOPA. To find the optimal position of the BBO crystal, the BBO should be moved from the maximum possible distance towards the pump focus. At the optimal position, a visible superfluorescence ring (Figure 2.9(c)) appears, which should be very faint and barely visible with eyes in the dark room. 37 Figure 2.9: (a) Ray schematic of NOPA 37 (b) NOPA on the laser table (c) Superfluorescence ring generated by blue pump after the BBO crystal (d) Amplified and broadband signal generated from the NOPA (bottom in the ring) To generate stable NOPA output, it is necessary to reduce the intensity of the ring by moving the crystal away from the pump focus if the ring is very intense. Here, the only input beam is the blue pump. The conversion of a single pump photon to two lower-frequency photons is spontaneous. There are multiple solutions for the frequency- and phase-matching conditions in this scenario and each solution generates a pair of photons with specific frequencies and directions giving rise to the cone shape of all the multispectral light. This process is called spontaneous parametric downconversion, which is responsible for the generation of superfluorescence ring. The key for the optimal operation of NOPA is generating a good seed continuum (white light). Here, the white light is being generated by focusing a small portion of the 800 nm on sapphire crystal window. Sapphire is preferred for continuum generation because of its better 38 stability and high damage threshold. Due to the dispersion in the sapphire crystal and the collimating optics, the seed continuum is found to be strongly chirped. Apart from the chirp, the seed continuum also travels faster than the 400 nm pump pulse in the amplifier crystal (BBO) and hence a proper choice of delay between the pump and the seed is required to select the central wavelength of interest and the bandwidth of the pulses. In a parametric amplification, the energy is borrowed from the intense pump pulse into the less intense seed. So, the careful adjustment of pump pulse parameters such as wavelength, energy and duration are very important. The wavelength of the pump determines the tuning range while the pulse energy and duration influence the interaction length, the bandwidth gain and the maximum energy of the amplified signal. Apart from pump parameters, the parameters of the nonlinear crystal such as type, length, amplification bandwidth and transparency range (crystal should be transparent in the wavelength range where pump, signal and idler falls) are also important. Commercially available NOPAs are designed to work with longer pulse width amplifiers (around 100 fs). In our lab, the pulse width of 800 nm, output of the amplifier laser, is 35 fs. Ironically, the short pulse width in the 400 nm pump results in a narrower bandwidth of the signal and so longer amplified signal pulses. Under these conditions, a typical bandwidth observed at 550 nm was 15 nm (≈ 30 fs transform limited pulse). For the optimal energy exchange between the pump and the seed, both need to be temporally overlapped throughout the gain medium. As the seed is strongly chirped, the different colors in the white light are arriving at the BBO crystal at different times and the pump pulse does not temporally overlap all the colors - resulting in a narrower than optimal bandwidth in the amplified signal. To overcome this issue, we added two fused silica windows (each 1 cm thick) to temporally stretch the pump beam making sure that the seed duration is shorter or comparable to the stretched pump pulse (Figure 2.10). Now, the 39 bandwidth increased from 15 nm to 46-50 nm at a 550 nm center wavelength, corresponding to 10 fs transform limited (TL) pulse width. One should therefore not think that the duration of the NOPA output pulses will depend upon the pump pulse duration; the output pulse duration is determined instead by the broadband phase-matching. Figure 2.10: (a) Narrowband signal resulting from short pulse-width of the pump beam (b) Broadband signal resulting from chirped pump beam. 39 Compression of NOPA Output Pulses: The output pulses from the NOPA are generally lengthened in the time domain, typically around 200 fs. The output has pretty large pulse width due to positive chirp acquired by the material dispersion introduced during WL continuum generation, several optical elements (beam splitter, lenses, filters) in the path of the signal and the BBO crystals. Therefore, the pulse must be compressed to achieve a width nearing the transform- limit. Some of the well-known optical systems known to provide negative dispersion are prism pairs, grating pair and chirped mirrors. Generally, a pair of gratings is less preferable to compress the NOPA/OPA output pulses due to their higher power loss and large higher-order dispersion. In our lab, we built a Brewster-cut prism pair compressor in a folded geometry. The folded geometry was adopted to fit in a compact space on the laser table (Figure 2.11). The material chosen for prisms is fused silica due to its low material dispersion. One should keep in mind that the separation between the prisms is dictated by both the second and third order of dispersion (TOD) and hence 40 the prism compressor can be applicable for ⁓20 fs TL pulse but not very short TL pulse. 39 The negative TOD should be handled separately if a broader bandwidth is being compressed. The proper choice of the prism material depends on the ratio of GDD and TOD and the wavelength. Whenever the NOPA is tuned to different wavelength, the inter-prism distance should be adjusted accordingly. By changing the distance between the two prisms, a crude adjustment to balance the GDD of NOPA is possible. To fine tune the degree of dispersion, the prism insertion (in and out of the beam) is also adjusted. Since the SHG conversion efficiency depends on peak incident power, the qualitative alignment of the prism compressor can be done just by optimizing the SHG of the NOPA beam after the compressor. The output of NOPA is horizontally polarized with respect to the lab reference. Figure 2.11: The folded geometry of prism pair compressor set-up to compress the visible output of the NOPA. Mirror1: (R= -100 cm SILFLEX broadband mirror for beam collimation) | (2,3,4,6,7,9): EAV mirror | 5,8: Brewster cut fused silica prisms 2.4.2 Generation of Tunable Ultraviolet Pulses To access the tunability in the UV region, the NOPA output (470 – 750 nm) is frequency doubled in a thin BBO crystal. The compressed pulse of the NOPA output (visible range) is passed through a 90 ͦ periscope which serves both to change the beam height and the beam polarization. 41 The vertically polarized light was focused onto 150 μm thick BBO crystal by using a 10 cm plano- convex lens (fused silica). A collimating lens (MgF2, f =15 cm) is placed on a compact dove-tail linear stage after the BBO for fine adjustment to properly collimate the beam. The focal length of collimating lens is chosen larger than the focusing lens, so that this pair of lenses can act as a beam- expanding telescope. The beam size of second harmonic of NOPA output (UV beam) is increased on purpose to achieve a tighter focus at the sample position inside the TRPES vacuum chamber. The horizontally polarized SHG signal is passed through a second Brewster-cut prism pair compressor, this time with prisms made of CaF2. This pair serves to pre-compensate for dispersion that the UV pulses pick up from travelling from the SHG crystal to the sample position. Generally, we can achieve 3.5 - 4 nm of bandwidth in UV region. For example, the bandwidth at 268 nm center wavelength was around 3.8 nm (or 28 fs TL pulse). The cross correlation between 268 nm and white light continuum was performed in neat ethanol, using a gravity jet to generate a thin film of ethanol. The FWHM of the cross-correlation is found to be 50 fs (Figure 2.12). Figure 2.12: (a) UV spectrum measured with fiber optic spectrometer (Ocean Optics 2000) (b) Cross correlation trace between a 268 nm UV pulse and a probe white light continuum. 370 nm, 400 nm and 450 nm were chosen from white light continuum to calculate cross-correlation width. 42 2.5 Long Time Pump-Probe Set-up To explore the dynamics of a photophysical process, we need a time delay between the pump and the probe pulses. A mechanical translation stage generally provides a delay up to 2 ns since the can translate only up to 60 cm. Even over this range, beam pointing walks off and it is less and less desirable to use additional physical path length to introduce variable time delay. In order to study longer time dynamics (ns – ms), temporal delays must be approached a different way. For example, to achieve a ms delay between pump and probe, one pulse would have to travel an additional 300 kilometers path length! Fortunately, a delay can be introduced between the pump and probe pulses if they are derived from different electronically synchronized sources where the delay is introduced by a controlled delay generator (Stanford Research Systems, DG645). To study such long-time dynamics, a pump pulse is generated by a separate picosecond Nd: YAG laser (Alphalas, PULSELAS-A-266-300-SP, pulse duration < 800 ps) and synchronized to our femtosecond laser system, the source of the probe pulses, by the delay generator. To synchronize the ns-laser and fs-laser, the delay generator and fs-laser were triggered by the oscillator (Micra) and were synchronized to the 1 kHz repetition rate of the amplifier with the help of a synchronization and delay generator (SDG) device. The SDG monitors the timing between a mode-locked seed source (Micra) and an amplifier (Legend). The user is responsible to provide an external well-conditioned TTL (transistor-transistor logic) pulse to trigger the SDG device. Sometimes, an internal trigger is available in the SDG, but we do not have in our case. The SDG provides separate trigger signals, which can be adjusted in the steps of 250 ps. Two of the trigger signals are being used to trigger the Pockel cells inside the amplifier and the third one can be used to trigger oscilloscope or any other device at a repetition rate of 1kHz. 40 The firing of the Pockel cells at correct timing is very important to get a single output pulses from the amplifier. A small 43 error (2 or 3 ns) can result in multiple output pulses. We used the third signal to trigger the delay generator (DG645) but we encountered some issues like a jitter (⁓10 ns) and double-pulses in ns- laser output. The issue was coming from improper synchronization between the seed laser and the pump laser of the amplifier. To encounter these experimental issues, a master trigger source was implemented, named as Countdown box. Figure 2.13: Illustration of the wiring inter-linking the devices in the lab (SSC716) The Countdown box triggers the SDG and the amplifier’s Q-switched pump laser (Evolution QSW) via a pulse generator device (Quantum composer, QC 9512). Further, the SDG triggers the ns-laser via the delay generator. To accomplish the synchronization between the seed (Micra) and the amplifier (Q-switch driver, HSD), the RF signal coming from the mode-locked 44 output of the Micra is coupled to the SDG via the Countdown box. A proper trigger pulse is critical for the normal operation of the devices/equipment in the lab. The splitting of trigger pulses (by using a T-joint) must be avoided for the proper functioning of the instruments. Figure 2.13 shows the wiring diagram currently in use in SSC 716. The 1 kHz repetition rate caps the maximum delay to one millisecond between ns-laser and the fs-laser. To perform the experiments, a variable delay is applied to the pump pulse relative to the previous pulse in the pulse train, approaching the 1 ms inter-pulse separation to achieve shortest pump-probe delay times (shown in Figure 2.14). The fundamental of Alphalas laser is 1064 nm. Hence, the pump pulses can only be generated at 532 nm, 355 nm and 266 nm by second harmonic, third harmonic and fourth harmonic generation respectively. The ns-laser has been successfully coupled to our transient absorption (TA) and time-resolved photoelectron spectroscopy (TRPES) by a graphical interface (LabView program). The name of the LabView program for TA measurement is “Longtime Main 3.0.1.VI” and “LongTime Pump Probe” for TRPES measurement on the assigned computer in the lab. Figure 2.14: The scheme to introduce long-time delay The name of the LabView program for TA measurement is “Longtime Main 3.0.1.VI” and “LongTime Pump Probe” for TRPES measurement on the assigned computer in the lab. 45 2.6 References 1. F. Buchner; H.-H. Ritze; J. Lahl; A. Lübcke, Phys. Chem. Chem. Phys. 2013, 15, 11402- 11408. 2. F. Buchner; B. Heggen; H.-H. Ritze; W. Thiel; A. Lübcke, Phys. Chem. Chem. Phys. 2015, 17, 31978-31987. 3. F. Buchner; A. Nakayama; S. Yamazaki; H.-H. Ritze; A. Lü bcke, J. Am. Chem. Soc. 2015, 137, 2931-2938. 4. H. L. Williams; B. A. Erickson; D. M. Neumark, J. Chem. Phys. 2018, 148, 194303. 5. T. Suzuki, Bull. Chem. Soc. Jpn. 2014, 87, 341-354. 6. G. Kumar; A. Roy; R. S. McMullen; S. Kutagulla; S. E. Bradforth, Faraday Discuss. 2018, 212, 359-381. 7. H. Siegbahn; K. Siegbahn, J. Electron. Spectrosc. Relat. Phenom. 1973, 2, 319-325. 8. M. Faubel; B. Steiner; J. P. Toennies, J. Chem. Phys. 1997, 106, 9013-9031. 9. J. H. D. Eland, Elsevier: 2013. 10. I. Jordan; A. Jain; T. Gaumnitz; J. Ma; H. J. Wörner, Rev. Sci. Instrum. 2018, 89, 053103. 11. F. Buchner; A. Lübcke; N. Heine; T. Schultz, Rev. Sci. Instrum. 2010, 81, 113107. 12. A. Kothe; J. Metje; M. Wilke; A. Moguilevski; N. Engel; R. Al-Obaidi; C. Richter; R. Golnak; I. Y. Kiyan; E. F. Aziz, Rev. Sci. Instrum. 2013, 84, 023106. 13. N.-H. Ge; C. M. Wong; R. L. Lingle; J. D. McNeill; K. J. Gaffney; C. B. Harris, Science 1998, 279, 202-205. 14. T. Hertel; E. Knoesel; M. Wolf; G. Ertl, Phys. Rev. Lett. 1996, 76, 535-538. 15. R. Haight; J. Bokor; J. Stark; R. H. Storz; R. R. Freeman; P. H. Bucksbaum, Phys. Rev. Lett. 1985, 54, 1302-1305. 16. A. W. Potts; W. C. Price, Phys. Scr. 1977, 16, 191. 17. J. E. Reutt; L. S. Wang; Y. T. Lee; D. A. Shirley, J. Chem. Phys. 1986, 85, 6928. 18. L.-S. Wang; J. E. Reutt; Y. T. Lee; D. A. Shirley, J. Electron. Spectrosc. Relat. Phenom. 1988, 47, 167-186. 19. M. Faubel; B. Steiner; J. P. Toennies, J. Chem. Phys. 1997, 106, 9013. 20. I. Jordan; M. Huppert; M. Brown; J. A. van Bokhoven; H. J. Wörner, Rev. Sci. Instrum. 2015, 86, 123905. 21. M. Mucke; M. Förstel; T. Lischke; T. Arion; A. M. Bradshaw; U. Hergenhahn, Rev. Sci. Instrum. 2012, 83, 063106. 22. A. Roy. Following Redox Chemistry and Excited State Dynamics in Solution Using Liquid Jet Photoelectron Spectroscopy. University of Southern California, 2015. 23. P. Kruit; F. H. Read, J. Phys. E: Sci. Instrum. 1983, 16, 313. 24. M. Mucke; M. Förstel; T. Lischke; T. Arion; A. M. Bradshaw; U. Hergenhahn, Rev. Sci. Instrum. 2012, 83, 063106. 25. D. Trabert; A. Hartung; S. Eckart; F. Trinter; A. Kalinin; M. Schöffler; L. P. H. Schmidt; T. Jahnke; M. Kunitski; R. Dörner, Phys. Rev. Lett. 2018, 120, 043202. 26. J. W. Riley; B. Wang; M. A. Parkes; H. H. Fielding, Rev. Sci. Instrum. 2019, 90, 083104. 46 27. A. Roy; R. Seidel; G. Kumar; S. E. Bradforth, J. Phys. Chem. B 2018, 122, 3723-3733. 28. S. Truong; A. Yencha; A. Juarez; S. Cavanagh; P. Bolognesi; G. King, Chem. Phys. 2009, 355, 183-193. 29. L. Karlsson; L. Mattsson; R. Jadrny; R. Albridge; S. Pinchas; T. Bergmark; K. Siegbahn, J. Chem. Phys. 1975, 62, 4745-4752. 30. N. Preissler; F. Buchner; T. Schultz; A. Lübcke, J. Phys. Chem. B 2013, 117, 2422-2428. 31. Y.-i. Yamamoto; S. Karashima; S. Adachi; T. Suzuki, J. Phys. Chem. A 2016, 120, 1153- 1159. 32. H. Shen; N. Kurahashi; T. Horio; K. Sekiguchi; T. Suzuki, Chem. Lett. 2010, 39, 668-670. 33. A. T. Shreve; M. H. Elkins; D. M. Neumark, Chemical Science 2013, 4, 1633-1639. 34. N. Kurahashi; S. Karashima; Y. Tang; T. Horio; B. Abulimiti; Y.-I. Suzuki; Y. Ogi; M. Oura; T. Suzuki, J. Chem. Phys. 2014, 140, 174506. 35. Y. Tang; Y. I. Suzuki; H. Shen; K. Sekiguchi; N. Kurahashi; K. Nishizawa; P. Zuo; T. Suzuki, Chem. Phys. Lett. 2010, 494, 111-116. 36. J. W. Riley; B. Wang; J. L. Woodhouse; M. Assmann; G. A. Worth; H. H. Fielding, J. Phys. Chem. Lett. 2018, 9, 678-682. 37. E. Riedle, NOPA: Fundamental and Instructions Manual. 38. E. Riedle; M. Beutter; S. Lochbrunner; J. Piel; S. Schenkl; S. Spörlein; W. Zinth, Appl. Phys. B 2000, 71, 457-465. 39. C. Manzoni; G. Cerullo, J. Opt. 2016, 18, 103501. 40. Operator's Manual, SDG Synchronization and Delay Generator Coherent. 47 Chapter 3. UV Photoelectron Spectroscopy in Liquid Microjet: Promises and Technical Challenges 3.1 Introduction Photoelectron spectroscopy (PES) is a powerful tool to explore molecular electronic structure and its evolution in the field of molecular structure and dynamics. For decades, resonance enhanced multiphoton ionization (REMPI) has been used to reveal the electronic and vibrational structure of polyatomic molecules in the gas phase in conjunction with computational calculations. 1-6 REMPI offers far greater sensitivity over nonresonant photoionization. 7 Most often ions are detected in gas phase REMPI, but photoelectrons may also be collected efficiently. With recent advent of the liquid microjet technique by Faubel and coworker, photoelectron measurements are also possible in the solution phase. 8 The incompatibility of vacuum technology with liquids was the main concern for the feasibility of liquid phase photoelectron measurements. In the last ten years, a series of papers have been published on different polyatomic molecules in aqueous and nonaqueous solution by several groups using REMPI and its time resolved analogue, time resolved photoelectron spectroscopy (TRPES). 9-17 Although the introduction of liquid micro-jet techniques in conjunction with photoelectron spectroscopy is a great success for the molecular structure and dynamics community, but there are still some critical issues which impose difficulty in analyzing photoelectron spectra obtained from liquid microjets. The most critical problem encountered, in probing liquids by PES, is the inelastic scattering of ejected photoelectrons by the solvent molecules. Inelastic scattering reduces the observed electron flux from the liquid microjet and at the same time the eKE of the photoelectrons 48 is also reduced by the collisions with the solvent molecules before a successful ejection into the vacuum as shown in Figure 3.1. Figure 3.1: Illustration of the effect of inelastic scattering on photoelectrons with different eKE. The concept of this figure is adapted from Ref. 18 Since the photoelectron bands in liquids are quite broad (⁓ 1eV), the low and high eKE regions of a given band are affected to different extents. As a result, the peak positions of photoelectron bands change along with spectral distortions. Currently, it is known that the issue of inelastic scattering can be resolved by implementing a very high ionization photon energies such as EUV source (photons in the range of >150 eV) generated by high harmonic generation. 19-21 But the application of EUV photoelectron spectroscopy is limited in solutions due to a complicated instrumental set- up and extremely low signal level. The UV photon source is easier to implement, and the application of UV photoelectron spectroscopy is very promising in exploring the electronic 49 structure and the dynamics of molecular excited states in solution. The main concern is degradation of the quality of spectra measured by UV photons due to inelastic scattering. Another critical issue in UV photoelectron spectroscopy is the loss of photoelectrons with low eKEs termed the transmission function, or cut-off threshold. The transmission function is dependent on the kinetic energy of the photoelectrons; this is determined by two parts: loss of photoelectrons due to the presence of stray fields and the loss of photoelectrons within the bulk of the liquid inside the liquid microjet. The loss of lower eKE electrons within the liquid is governed by the energy gap between the bottom edge of the conduction band of the liquid and the vacuum. The photoelectron spectrum gets distorted in the low eKE region due to a failure to capture the photoelectrons with low eKEs. The transmission function does not affect the time constants of the excited state dynamics, but it has to be taken into account to explain the shape of the spectrum and the “genuine” binding energy of the excited state intermediates. To understand how these theoretical concepts manifest themselves experimentally, phenol is chosen as a prototype molecule since its photophysics is well-studied. Ashfold and coworkers and Bradforth and coworkers published a series of papers on the electronic structure and possible relaxation pathways of isolated phenol. 22-25 Ashfold and coworkers reported that O-H bond fission is the primary channel of relaxation when isolated phenol is excited to the first electronically excited state (S1 ←S0). Phenol has a strong UV absorption band due to population of the 1 ππ * state shown in Figure 3.2(a). The 1 ππ * (S1) state is optically bright in character and bound in all coordinates. This molecule also has a low-lying dissociative state, 1 πσ * , which forms a conical intersection with the 1 ππ * state along the O-H coordinate. The crossing point is 5 eV with the O-H bond length increased to 1.2 A o and other 50 coordinates as at Franck Condon region (Figure 3.2(b)). The rate of dissociation of the O-H bond varies as a function of excitation wavelength. The O-H fission above the CI occurs on a femtosecond timescale through direct dissociation while it takes nanosecond below the CI due to tunneling under the barrier to access the 1 πσ * state. 22, 26 Figure 3.2: (a) UV-Vis absorption spectrum of phenol and tyrosine in water (b) Gas phase PES of phenol along O-H coordinate for S0, S1 and S2 26, 27 . The figure for the gas phase PES of phenol has been adapted from Ref. 26 , authored by Ashfold and his coworkers. The landscape of the PES changes in polar liquids and the magnitude of this change depends on the strength of dipole-dipole interaction between the excited electronic states and the surrounding medium. Therefore, the energy minimum or onset of the CI in polar media is altered, resulting in different relaxation pathways compared to the gas phase. More importantly, new surfaces become lowered in energy and also contribute to the excited state dynamics. Recently, Bradforth and coworkers have reported the formation of solvated electron in aqueous phenol solution when excited with 266 nm and 200 nm. 28 At 200 nm excitation, autoionization is the only mechanism to generate the solvated electron while at 266 nm excitation autoionization and proton coupled electron transfer are the responsible mechanisms. 28 51 The redox properties of phenol in water have already been explored by the Bradforth group using X-ray photoelectron spectroscopy and ab-initio calculations by Krylov and coworkers. 29 In this chapter, we utilize the tunability of ultrafast lasers to measure the REMPI of aqueous phenol at a series of excitation energies to quantify the effect of inelastic scattering at different ionization photon energies. Although the utilization of photoelectron spectroscopy in the liquid phase is quite promising in revealing the effects of solvation on the dynamics of the excited state, the current experimental challenges in UV photoelectron spectroscopy degrades the information carried by the photoelectrons. We have also performed time-resolved experiments of aqueous phenol to understand how these issues are present in a dynamic scenario 3.2 Experimental All experiments were performed using our home-built liquid microjet photoelectron spectrometer coupled with tunable femtosecond ultraviolet laser pulses. The details of the spectrometer are reported elsewhere. 7 To perform resonant enhanced two photon ionization of aqueous phenol, the UV pulses at a series of wavelengths were generated by the second harmonic of visible output from an optical parametric amplifier (OPA 800C, Spectra Physics) in a 150 μm thick type I BBO crystal. The polarization of the UV laser beam was vertical with respect to the laboratory frame and perpendicular to the time of flight axis. The experiment was also performed at 200 nm, which was generated by the fourth harmonic of 800 nm. The 200 nm beam was generated by sum frequency mixing of the fundamental beam (800 nm, 30 fs) of Ti:sapphire amplifier system (Coherent Legend Elite, 1 kHz repetition rate) and its third harmonic in a 75 μm thick BBO type I crystal. The third harmonic was generated by sum frequency mixing of the 800 nm fundamental and its second harmonic in a 150 μm thick BBO type II crystal. The spot size of all laser beams utilized at the interaction point with the sample was less than 80 μm. The spot size 52 was determined using a translating knife edge. The temporal evolution of the excited state was also monitored by pumping the aqueous solution of phenol at 266 nm and probing at 292 nm. The probe beam centered at 292 nm was generated from the second harmonic of a NOPA output for short-time pump-probe delay measurement (<500 ps) while for ns-delay the second harmonic of OPA was utilized. We also performed an experiment where the pump excitation was 266 nm, but the probe was changed from 292 nm to 200nm to explore the influence of inelastic scattering of photoelectrons. The laser beam was directed onto the laminar region of the liquid jet. An HPLC pump (Shimadzu LC-20AD) in constant flow mode was used to inject the aqueous phenol solution inside the vacuum chamber via a fused silica nozzle (20 μm diameter) with a backing pressure of 0.1-0.3 MPa. During operation, the typical pressure in the main chamber was maintained around 2x10 -4 mbar and 3x10 -6 mbar in the drift tube with the aid of turbomolecular pumps and cryopumping. The photoelectrons were measured at the end of 50 cm drift/time of flight tube by using a pair of microchannel plate (40 cm diameter) in a chevron configuration. The detector was operated in counting mode and the count rate was maintained at <10 electrons/pulse to avoid any statistical issue, e.g. counting two electrons as one. The count rate should also be sufficiently low to avoid space charge effect. If there is a high density of photoelectrons generated, a coulombic repulsion between electrons occur which causes a change in their kinetic energies and broadening of the eKE distribution. Therefore, the number of electrons generated per laser pulse should be as low as possible. In a pump-probe geometry, the effect of space charge can be more severe: in R2PI the Coulombic repulsion involves electrons generated from the same pulse envelope while in pump- probe mode, electrons will also be generated by the combination of individual photons coming from pulse and probe envelopes in addition to electrons generated via R2PI 53 We calibrated the spectrometer using vibrationally resolved three photon emission spectra of water vapor as described in Chapter 2 chapter. The aqueous solution of phenol (≥99.5%; Sigma-Aldrich) was prepared in ultrapure water (18.2 M Ω.cm, Millipore) without any further phenol purification. The concentration was maintained at 20 mM. During all the experiments, the temperature of the solution outside the vacuum chamber was kept at 20 o C however the solution temperature will rapidly decrease at the interaction point inside the vacuum chamber due to extreme evaporative cooling. 8, 30 3.3 Current Issues in Experimental Measurements Before we start discussing the experimental results of phenol, it is prudent to revisit the concept of inelastic scattering of photoelectrons in the liquid-jet as well as the spectral limitation caused by the transmission function of the instrument. 3.3.1 Transmission/Cut-off function The energy dependent transmission function affects the photoelectron spectra significantly. The measured eKE distribution is the result of the product of the true eKE distribution and the transmission function. The transmission function can be defined as the ratio of the number of electrons detected by the MCP at a given kinetic energy to the number of electrons generated in the conduction band of the liquid microjet solution. Based on a number of experimental measurements, we have understood that there are two factors that contribute to this transmission function. The first one is the presence of stray fields, which reduces the number of slow-moving electrons (low eKE electrons) as they fail to reach the detector. This issue is common to all photoelectron spectrometers but varies with different design. Another factor which contributes to the transmission function is the inherent presence of the liquid-gas interface. Some 54 of the low eKE electrons are lost inside the liquid-jet without successful ejection into the vacuum, leading to a cut-off at low eKEs. The energy gap between the bottom of the conduction band and the vacuum level determines the probability for the electrons, especially those with low kinetic energy, to make successful ejection into the vacuum. For the correct analysis of the spectral shape, it is necessary to account for the transmission function of the instrument. In our experiment, we have not determined the correct form of transmission function however it appears that the spectrum below 0.5 eV is severely affected in our water-based spectra. There is a sharp decline in the number of electrons detected below 0.5 eV. This is not to say that these electrons do not exist, but rather that they have not been successfully detected. The crude estimated graphical form of the transmission function (for water-based measurements) by our photoelectron spectrometer is shown in Figure 3.3 (b). To get the graphical form, the R2PI spectrum of aqueous indole (4 mM solution) was measured with a 292 nm ionization pulse. Recognizing the existence of the transmission function the spectrum was fit with only one Gaussian centered at 1 eV and truncated to ~ 0.7 eV, avoiding the low eKE region. One expects to see another band around ⁓ 0.4 eV, which seems to be missing here. Currently we are assuming that this band will fall under the cut-off threshold of the spectrometer. We further assumed that there is no loss of higher eKEs electrons, with these electrons corresponding to the signal from the right side of the spectrum relative to the peak maximum. Even though the low-eKE band was not observed, this should not inhibit our ability to fit the second, observed electronic band (high eKE) with a Gaussian profile, spanning both the low and high eKE regions. To this aim, the higher eKE side of the spectrum was mirror-imaged towards low eKE side as shown in Figure 3.3(a). Now, the newly constructed R2PI spectrum, created by taking the mirror image of higher eKE region towards lower eKE, was divided by the actual measured R2PI spectrum (truncated to the region 55 where data was recorded) to generate the graphical form given in Figure 3.3(b). It clearly shows that the spectral region with eKE <0.5 eV is significantly affected due to the loss of low eKEs electrons. Figure 3.3: (a) R2PI spectrum of aqueous indole. The black curve is the experimentally measured spectrum while the red curve is the constructed R2PI spectrum by taking the mirror image of higher eKE region of the spectrum on the lower eKE region. (b) the estimated form of the transmission function generated by taking the ratio of red and blue curve from Figure (a) An electron can be successfully ejected into the vacuum if the eKE is greater than the energy difference of the conduction band and the vacuum. In Figure 3.4, the energy difference between the bottom of the conduction band and the vacuum is denoted by V0 and eKE signifies the measured kinetic energy of ejected electrons. The correct estimate of V0 is very critical to extract the genuine binding energy of any species in a liquid microjet study. Currently, there are multiple values reported for V0 for water in literatures. Bernas et al. reported 0.75 eV 31 , Gaiduk et al. reported 0.1-0.3 eV 32 , Coe et al. reported -0.12<V0<0 33 , and Ziaei et al. reported relatively larger value V0⁓1.1 eV. 34 56 Luckhaus et al. reported the “genuine” binding energy for the hydrated electron. 35 The distribution for the binding energy for the hydrated electron is found to be bimodal in nature, as shown in Figure 3.4 (b), with the average value of 3.7 eV. Suzuki and coworkers emphasized in their studies that the average value for the binding energy of the hydrated electron strongly depends on the energy of the ionizing photon especially for photons falling in the range of 3.6 – 5.8 eV. Depending on the different photon energies, the inelastic scattering of the electron within the liquid microjet is different, which causes the inaccuracy in the true binding energy. To overcome the issue of inelastic scattering, Luckhaus et al. performed Monte Carlo simulations on the data recorded in liquid microjet at several ionization photon energies. 35 To correct for loss in eKE due to scattering, they considered the low energy scattering cross-sections measured in microdroplets 36 and amorphous ice. 37 Based on simulations (in conjunction with experimental data), they report the bimodal distribution for the binding energy of the hydrated electron. Figure 3.4: (a) Energy diagram for the photoionization of hydrated electron. eBE(genuine), Vo and ΔE represents genuine binding energy for the hydrated electron, the energy gap between the bottom of conduction band of water with respect to the vacuum and eKE loss due to inelastic scattering, respectively. (b) Binding energy distribution for the hydrated electron reported by Luckhaus et al. 35 These figures are taken from Ref 35 57 The energy diagram in Figure 3.4(a) depicts the concept behind analyzing the data. The reason behind the bimodal distribution for the hydrated electron is not clearly described by Luckhaus et al. but it may be related to incorrect estimate of V0. 3.3.2 Inelastic Electron Scattering Typically, in photoelectron spectroscopy the electron binding energy is computed from the difference between the photon energy employed and the electron kinetic energies recorded. One important factor, which can lead to potential misinterpretation of the experimental data is the influence of inelastic scattering. Due to scattering before the electron leaves the liquid, the kinetic energy imparted immediately to the electron upon ejection from the parent molecule may be reduced before entering the vacuum. Consequently, the recorded value at the detector will be correspondingly lowered, leading to an overestimation in binding energy. The probability of scattering and the average energy loss depends on the initial kinetic energy of the electron. It should be remembered that as the ionizing light is tunable, even the same PE band recorded with different photon energies may change. 35 The electrons ejected through the liquid in the energy range of 0.5 < eKE < 3 eV in liquids are significantly affected by scattering and energy loss, which results in the distortion of the photoelectron spectra and ambiguity in the accurate determination of electron binding energies. Luckhaus et al. carefully examined this issue and reported the contribution of inelastic scattering by comparing the experimental photoelectron spectra (3.6 eV ≤ hν ≤13.6 eV ) of the hydrated electron—which exhibits a spectrum with only one feature which can be fit with a single Gaussian—to Monte Carlo simulations that include inelastic scattering. 35 This study finds that outgoing electrons with eKE ≈ 0.75 eV are minimally affected by scattering while electrons departing with 1.5 eV eKE are shifted by as much as 0.5 eV. Thus, to fully recover a genuine 58 binding energy spectrum from a measured liquid jet PE spectrum, the authors state it is necessary to simulate the expected photoelectron spectrum to account for contributions from inelastic electron scattering. Another approach to correct for inelastic scattering would be to use a much higher probe energy at the cost of low signal level resulting from low photon flux. It has been shown experimentally that there is minimal contribution from inelastic electron scattering if the probe photon with more than 30 eV is used. 19, 21 Suzuki and coworkers have demonstrated a spectral retrieval method by measuring the binding energy distribution of the hydrated electron using both UV and extreme UV (EUV) probe energies. 38 An EUV source based on high harmonic generation is required to measure the eBE distributions with minimal or zero contribution from inelastic scattering. This measurement helps to determine the initial eKE distribution, Gℏω (E), unperturbed by the inelastic scattering for various UV photon energies by taking into account the eBE measured by EUV. The FWHM of a given photoelectron band is also determined from the EUV measurements. Using a UV source with different photon energies, the eKE distribution (gℏω(E)) is recorded which is affected by inelastic scattering. Now, for every ℏω there is a gℏω(E) and a corresponding Gℏω(E). Several pairs of gℏω(E)← Gℏω(E) provide a one-to-one mapping (a linear transformation). The inverse mapping of this linear transformation allows retrieval of the electron kinetic energy distributions unaffected by the inelastic scattering. The set of gℏω(E) and Gℏω(E) pairs should cover the entire range of kinetic energy of interest (spectral range). The measured kinetic energy distribution from UV source can be expanded in terms of their basis functions (gℏω(E)) along with their expansion coefficients. To retrieve the true kinetic energy distributions, the (gℏω(E)) should be replaced by 59 Gℏω(E). The photoelectron measurements performed in liquid microjet by implementing UV photons should be corrected to avoid the wrong interpretation of the experimental results. 3.4 Results and Discussions: To explore the electronic structure and the time evolution of the photoexcited state of phenol and the effect of the transmission function and inelastic scattering of electrons on the photoelectron spectra, we performed one color two-photon photoelectron spectroscopy (R2PI) on aqueous phenol and we also carried out TRPES measurements. 3.4.1 R2PI Spectra of Aqueous Phenol The vertical ionization energy (VIE) of aqueous phenol is measured by implementing resonant two photon ionization (R2PI) at USC and from single photon ionization (XPS) measured at the U41-PGM beam line at the BESSY II synchrotron facility in Berlin (end station needed). The photon energy used at USC was 4.64 eV while it was175 eV at BESSY. Since, it is known that the spectrum recorded with XPS will be minimally affected by the inelastic scattering and the transmission function of the instrument, so the binding energies will be more accurate compared to the values extracted from the R2PI spectrum. Figure 3.5 (a) shows the XPS spectrum of aqueous phenol which can be fitted with two Gaussians centered at 7.8 0.1 eV and 8.6  0.1 eV. 29 These values represent the two lower VIEs corresponding to D0←S0 (VIE I) and D1←S0 (VIE II) transitions, where D0 and D1 signifies the first and second cationic states. Now coming back to Figure 3.5 (b), which is R2PI spectra of aqueous phenol, two Gaussians are still required to properly fit the experimental spectrum. The center of these Gaussians fits is at 1.3  0.1 eV and 0.7 0.1 eV respectively, which correspond to 8.0  0.1 eV and 8.6  0.1 eV in binding energies 60 (BE = 2hѵ – eKE). After comparing the binding energies extracted from XPS and R2PI, we can say that VIE I is larger by 0.2 eV in R2PI while the VIE II is equal or within experimental error. Figure 3.5: (a) XPS spectrum of aqueous phenol. (b) R2PI spectrum of aqueous phenol where the wavelength is centered at 267 nm. Two Gaussians were required to properly fit both the spectra. Vertical orange line depicts the center of the Gaussians corresponding to VIEs. Note: The x-axis in XPS spectra represents binding energy (BE) while x-axis in R2PI represents electron kinetic energy. (c) R2PI spectrum of aqueous phenol corrected for inelastic scattering after using the transformation scheme suggested by Suzuki and his coworkers. 38 As mentioned earlier, Luckhaus et al. performed simulations to quantify the magnitude of inelastic scattering as a function of different photon energies for the photoelectron band of the hydrated electron 35 . They reported the contribution of inelastic scattering in the experimental data by comparing the experimental photoelectron spectra (3.6 eV ≤ hν ≤13.6 eV) of the same aqueous species with the Monte Carlo simulations that include inelastic scattering. The different photon energies provide different kinetic energies of the outgoing photoelectrons since the binding energy is the same in each case. We referred to the supplementary information of this paper and tabulated the kinetic energy shift between the experimental spectra and the simulated spectra for given ionization photon energies. A ‘look-up’ table, given below, is generated so that we can crudely correct our photoelectron spectra. 61 Table 3.1: Influence of inelastic scattering on photoelectron kinetic energy eKE peak (in experimental spectra) Shift towards high eKE (after correcting for inelastic scattering) 0.75 0 0.83 0.08 1 0.15 1.15 0.35 1.4 0.4 1.5 0.45 Based on the values provided in Table 3.1, we can say that there will be almost zero shift in eKE due to inelastic scattering for the Gaussian centered at 0.7 eV in R2PI spectrum while the Gaussian centered at 1.3 eV can shift by ⁓0.3 eV. After taking the effect of inelastic scattering into account, the binding energy corresponding to VIE I will change from 8 eV to 7.7 eV, in closer agreement to the value obtained from XPS. Since the change in eKE due to inelastic scattering is minimal for the peak centered at 0.7 eV, the value of binding energy found by both techniques is almost identical. We also see a sharp decline in the number of low eKE (<0.5 eV) electrons in R2PI spectrum, which can be explained by the transmission function of our instrument. Figure 3.5 (c) shows the R2PI spectrum of aqueous phenol after using the transformation scheme suggested by Suzuki and his coworkers to correct for inelastic scattering. 38 Let’s ignore the part of the spectrum with negative amplitude, which is the result of bad fitting of the basis spectra used in the transformation. In the transformed spectra, we can clearly see two Gaussians with their maxima peaking at 1.8 eV and 0.8 eV. The energy gap between these two Gaussians in the transformed spectra is ⁓1 eV, slightly larger than the energy gap of 0.8 eV in XPS (Figure 3.5(a)). Certainly, the value of peak centers of eKE distributions in the transformed spectra produces VIEs at 7.5 eV and 8.5 eV in closer agreement with the binding energies extracted from 62 XPS. Probably, the spectral shape of the transformed R2PI spectra can be improved by better handling of the basis spectra, for example choosing another function for the fitting instead of using higher order polynomials which have been used here. In R2PI, the first photon absorbed excites the molecule to an intermediate excited state, and a second photon absorbed under the same pulse envelope subsequently ionizes the molecule from the intermediate excited state. XPS, on the other hand, does not involve any intermediate state, so any one electron ejection is allowed. Considering the simplest molecular orbital picture to understand the R2PI spectrum, one would expect the first photon to promote the ring π electron from the HOMO to the LUMO and a subsequent photon to ionize the intermediate excited state to the manifold of cationic states that are connected by one-electron ionization events, i.e., with Koopman's theorem correlations. 39 For a HOMO →LUMO excitation by the pump, only the D0 cation state would be expected to have a non-negligible intensity (i.e, a single band observed) since the electron configuration of D0 is best described as a hole in the neutral HOMO. With 267 nm excitation, phenol is excited to the S1 state, which is the promotion from a π3 orbital. So the initial expectation was to see only one peak corresponding to π3 -1 but we observed two peaks corresponding to final states with π3 -1 and π2 -1 electronic configurations, also observed for tryptophan and tyrosine. 10 This observation of two bands can be explained based on the configuration interaction in the excited state (Figure 3.6). 40 63 Figure 3.6: The presence of two ionization peaks in R2PI-PES experiment emphasizes the contribution of configuration interaction also found in gas phase phenol. 40 Another noteworthy point is significant lowering of VIE in aqueous phenol compared to the gas phase. The VIE corresponding to the ground state to first cationic state (D0-←S0) decreases by 0.7 eV in water compared to the isolated molecule in vacuum. The VIE is affected by the relative stabilization of the neutral ground state and the final cationic state in water due to the electronic polarization provided by the solvent. Since the VIEs extracted from XPS and R2PI s are similar, we can say that the ionization from the intermediate state in R2PI is instantaneous with respect to the relaxation of the solvent or intermediate structure. The electronic response to the change in change for the final cationic state is principally responsible for the shift to lower VIE. Fielding and coworkers performed R2PI measurement on aqueous phenol as a function of pump wavelength and reported that the VIEs corresponding to D0-S0 and D1-S0 ionizations are 7.9  0.1 and 8.4  0.1 eV, respectively. They also observed that D0-S0 is lowered by 0.8 eV, 41 which is in good agreement with the current measurement and also reported by Ghosh et al. 29 The value of 64 0.7 eV for the gas-liquid shift is in the same range as for similar aromatic systems previously measured by water jet photoelectron spectroscopy. For example, 0.5 – 0.6 eV decreases in vertical ionization energies are reported for aniline, veratrole alcohol, imidazole and thymidine 42 and even lower values are reported for cytidine. 43 For an R2PI process, the peak shape depends on ultrafast dynamics (both intramolecular and solvation), on the intermediate state in relation to the pulse width and to the ionization rate on absorbing the second photon. 10 The ionization rate by the second photon is intensity-dependent and kinetically competes with electronic and nuclear relaxation from the initially excited state. Therefore, the observed vertical ionization energy (VIE) as well as the ionization onset/threshold (AIE) may manifest pulse-width-dependent (and intensity-dependent) character if the relaxation is fast compared to the pulse envelope. 44, 45 From the similarity in the onsets of the R2PI-PES and XPS photoelectron spectra of phenol, we can say that the extent of relaxation from solvation and/or intramolecular vibrational redistribution/vibrational energy relaxation is relatively small for phenol. As far as we are aware, no one has considered non-Condon effects on aqueous phase photoionization to date and such a study would be of considerable merit. 3.4.2 R2PI Spectra Measured at Various Pump Wavelengths R2PI measurements were performed on aqueous phenol by varying the photon energy (Figure 3.7(a)). These wavelengths were generated by doubling the OPA output as mentioned in the experimental section. Here, we are assuming that there is no change in the dynamics of the intermediate surface if photon energies are varied from 4.46 eV to 5.04 eV, since VIEs extracted from R2PI (with 4.66 eV) and XPS are close in agreement. By comparing all the spectra, the high eKE edge is not simply shifting by the extra energy deposited due to the higher probe energy used. 65 In fact, the trend does not look linear with respect to the photon energy. Figure 3.7(b) illustrates the shift in a graphical form. The abscissa represents the photon energy used and ordinate represents the actual eKE measured at the higher eKE edge. The red line shows the trend when inelastic scattering is completely neglected and the eKE for the high energy edge is shifted according to extra photon energy used to ionize the molecule. The black line illustrates the eKE measured as a function of probe photon energy implemented in our R2PI experiments. It is apparent from the Figure 3.7(b) that the magnitude of eKE shift due to inelastic scattering cannot be described by a simple function. The probability of an inelastic scattering event is determined by its cross-section which varies as a function of eKE and is inversely proportional to the inelastic mean free path (IMFP). The IMFP is the distance traversed by the photoelectron without encountering any inelastic collisions. The IMFP as a function of eKE is represented in the form of an inverse-bell-shaped curve known as the “universal curve”. The minimum of the curve lies in the range of 50-100 eV and it starts increasing monotonically after 100 eV. For kinetic energy range <10 eV, there is still a necessity to revisit this problem either experimentally or theoretically as the picture is not quite clear. Now, before we try to explain the nature of the graph in Figure 3.7(b) as to why it is not nonlinear for lower photon energies (which will result into lower eKE) but at higher photon energies the red curve and black curve are converging, let us think what we should expect. In the gas phase, there is no scattering events involved, so the curve should be a straight line as shown in Figure 3.7(c) with the visual help of the red line. In the liquid phase, the curve for eKE as a function of photon energy can be represented by the blue curve, keeping in mind the general shape of the inelastic cross-section versus eKE. But, the nature of the graph in Figure 3.7(b) is different to the 66 one expected. At this point, we are not able to comment further but if we assume that at lower photon energy the IMFP curve still near its minimum, then at least we can say that the number of inelastic collisions of photoelectrons with solvent molecule before being ejected into vacuum will be more (which means more loss in eKE) at lower photon energies. Figure 3.7: (a) R2PI spectra of aqueous phenol at different photon energy generated by doubling the OPA output, the pulse energy used for 4.46 eV, 4.66 eV, 4.73 eV, 4.88 eV, 4.96 eV, 5 eV and 5.04 eV are 40 nJ, 15.5 nJ, 17.5 nJ, 29 nJ, 36 nJ, 43 nJ and 45 nJ. (b) Comparison of high energy eKE edge for spectra (simply shifted by extra photon energy with respect to 4.46 eV) and experimentally measured eKE edge in higher eKE region for various photon energies. (C) Illustration of effect of inelastic scattering on the measured eKE as a function of photon energy in the gas phase and in solutions. 67 3.4.3 Spectral Comparison between Bradforth’s and Fielding’s Lab Figure 3.8 illustrates the R2PI spectra of aqueous phenol measured at 266 nm, 254 nm and 250 nm. The main reason to compare these spectra is to see the reproducibility and change in the spectral shape of R2PI when measured in two different labs with two different spectrometer designs. 41 There are three differences between the spectra measured in two labs, which one can notice: (i) There is a sharp decline in low eKE electrons in Bradforth’s lab but not in Fielding’s lab (ii) The spectral shape is different. The spectra recorded in Fielding’s lab are a single Gaussian while the spectra are asymmetric for Bradforth’s data. (iii) the spectral tail at higher eKE is more pronounced in the spectra recorded in Bradforth’s lab. The loss of low eKE electrons in our lab can be explained based on the concept of the transmission function (<0.5 eV). It looks like the spectra is not at all affected by the cutoff threshold measured in Helen’s lab. It’s bit puzzling since one part of the transmission function depends on the energy gap between the conduction and the vacuum which is inhere to both experiments. To properly fit the R2PI spectra measured in Bradforth’s lab, two Gaussians are required (as shown in Figure 3.5(b)) while the spectra measured in Fielding’s lab requires only one Gaussian. Recently, Fielding and coworkers have implemented a recirculation system for the sample delivery in their liquid microjet set-up and claims that the streaming potential is almost zero, which forces them to use two Gaussians to properly fit the spectrum. They claim that by taking care of streaming potential, the spectral quality in the low eKE region is improved, and they subsequently needed two Gaussians to obtain a better fit corresponding to 1 1 ππ* - D0 and 1 1 ππ* - D1 ionization. 46 Currently, it is difficult to explain why the spectral shape is different, but it appears that these experiments are very sensitive to experimental conditions. 68 Figure 3.8: R2PI spectra of aqueous phenol recoded in Bradforth’s lab at USC and Fielding’s lab at UCL. R2PI spectra measured in Fielding’s lab is taken from Ref 41 . 69 3.4.4 Time-Resolved Photoelectron Spectra of Aqueous Phenol In the time-resolved photoelectron measurements, the excitation wavelengths are chosen such that the total photon energies (8.9 eV) is considerably greater than the VIE to the lowest cation state, D0. Depending on how severe the spectrometer cut-off function is, the cation D1 is also within reach. Phenol is a well-studied system by our group as well as several others. 22, 28, 47- 53 With the extensive research done on phenol, the outcome of our experiments can be interpreted by reference to previous results, making it a strong model system for this experiment. Aqueous phenol's S1 excited state is well separated from higher electronic excitations and the excited state dynamics is relatively simple over the first few hundred picoseconds, with very little spectral relaxation observed in water by transient absorption. 52, 54 Therefore, we will attempt to draw a connection between time-resolved spectra nearby zero pump-probe delay and the R2PI spectrum. Another simplification for aqueous phenol is that the S1 S0 transition peaks near 266 nm and is non-resonant for the probe pulse wavelength used in our experiments, which is 292 nm. So, for R2PI, two-photons of 266 nm are used to ionize, while for the TRPES, 266 and 292 nm are used for excitation and ionization, respectively. In Fig 3.9 (a), the 266 nm/292nm PE spectrum is obtained at zero pump-probe delay, where the probe photon is less energetic, the PE intensity can be fit by a single Gaussian; (eKE = 0.73 eV center, FWHM ~ 1 eV). We cannot at this point comment on whether a second higher energy PE band is missing or is weak, because if present, it may well fall under the low eKE cut- off function of our spectrometer. One thing can be said that the peaks will be certainly sensitive to the inelastic scattering which will affect the peaks differently because of their different eKEs. If we compare more carefully the high eKE sides of the R2PI-PES and the 266 nm/292nm PE spectrum obtained at zero pump-probe delay, we might expect that the peaks would be identical 70 in shape but simply be positioned at different eKEs in accordance with the extra photon energy. Looking more carefully at Figure 3.9(a), the shape of R2PI and the 266nm/292nm zero-delay spectra indeed are very similar with a shift of 0.4 eV between the high energy edges. This shift is equal to the difference of the photon energy (4.66 - 4.24 = 0.42 eV) absorbed in the ionization step. This further implies the threshold ionization energies are essentially the same from both spectra. An important observation is to compare the shape, particularly on the high eKE side of the spectrum for the zero and long inter-pulse delay limits. Molecules excited to the intermediate state have only a finite time to evolve in the excited state before ionization in the zero-delay limit, essentially governed by the pulse length(s) and the intensity-dependent ionization rate. In contrast, in the long delay limit, the excited state molecules can fully relax. Now, if we compare the time- resolved spectrum at the pump-probe delay of 0 and 500 ps (Figure 3.9(b)), we observe that the peak maximum is shifted by a minimal amount of 0.08 eV. At 500 ps, we can assign the PE band as originating from the ionization of relaxed S1 and hence the vertical gap corresponding to D0 1 1 ππ* is the difference between the probe photon energy and the peak eKE, 4.24 - 0.73 eV = 3.51 eV. From table 3.1, it can be inferred that the peak at 0.73 eV will not be affected due to the inelastic scattering. It can be seen even more clearly from Figure 3.9 (c) that the S1 state undergoes very little relaxation over 0.5 ns that impacts the VIE; the VIE changes from 3.43 eV to 3.51 eV, almost within instrumental eKE error. To explore the excited state dynamics occurring in long time (ns- μs), we implemented a fourth harmonic (266 nm) of a picosecond Nd:YAG laser (pulse width around 700 ps) as an excitation pulse synchronized to femtosecond probe (292 nm) as an ionization pulse. Oliver and Zhang et al. suggested the formation of the solvated electron and the phenoxyl radical at 266 nm pump excitation for aqueous phenol with nanosecond formation time scale . 28 71 The mechanism proposed for the formation of these charged species is PCET and autoionization. They confirmed the presence of solvated electron experimentally by using electron scavenger (HCl) even at 13 ns. With the ns-TRPES, we were expecting to find the spectral profile of the solvated electron. The binding energy of solvated electron is 3.7 eV, so we expected to have a peak at 0.5 eV (4.24-3.7 =0.54 eV). Although we did not observe any peak corresponding to solvated electron in our TRPES spectra, but we did observe a new photoelectron band at 2 eV eKE which decays with time as shown in Figure 3.9(d). Currently, we are unable to assign this feature, but we are speculating that this feature might be coming from two-photon probe ionization of the phenoxyl radical since the asymptotic value to generate phenoxyl radical in water is ⁓6 eV which can give rise to a spectral band at ⁓2 eV (2 x 4.2 eV- 6 eV). 55 Our analysis of the simple R2PI and time-resolved PE spectrum of phenol hinges on the fact that there was very little solvation or electronic dynamics occurring in the excited state (these changes would be seen in the photoelectron signature) in line with literature expectations. In the case of phenol, most of the excitation energy in excess of the electronic origin goes into perpendicular vibrational modes 56 as ionization is diagonal with respect to these coordinates. It is understandable that minimal evolution is observed in the aqueous phenol VIE even if vibrational relaxation takes place. However, if nuclear wavepacket motion does occur along a coordinate where there is a significant displacement between the excited state and cation surfaces, solvation dynamics, or an electronic state change occurs prior to the ionization pulse, then the peak eKE in an R2PI band cannot be simply assigned as the vertical ionization energy of the intermediate excited state. 72 Figure 3.9: R2PI spectrum of phenol (20 mM solution; 266 nm ionization pulse) is compared with the TRPES spectrum of phenol when both 266 nm and 292 nm pulses are temporally overlapped (τ=0) in panel (a). Individual Gaussian fits are shown in magenta and green and the total fit is shown in blue for the R2PI spectrum, while the red line displays a single Gaussian fit for the TRPES spectrum. Panel (b) displays a small shift in the peak maximum (0.08 eV) between the TRPES spectrum of aqueous phenol at t=0 ps and t=500 ps. Panel (c) shows TRPES spectra of aqueous phenol at different pump-probe delays and illustrates no change in spectral shape. Panel (d) shows TRPES spectra at longer pump-probe delays from a separate measurement implementing ns-pump laser centered at 266 nm and 292 nm probe. The 292 nm laser pulse used for the spectra recorded in panel (a), (b) and (c) was generated by doubling the NOPA output while 292 nm used for ns-delays measurement was generated by doubling the OPA output. The polarization of both beams is parallel with respect to each other. In fact, the peak eKE obtained would be lower than the expected value resulting in overestimation of VIE. Buchner et al. has observed this effect for aqueous adenine. 13 Furthermore, 73 there will be substantial changes in the time-resolved PE spectrum matching the relevant timescales of these dynamics. Now, returning to effect of inelastic scattering on TRPES spectra, Figure 3.10 represents the effect of inelastic scattering, but here it has been shown for a time resolved experiment. The pump wavelength was 266 nm and the probe wavelength were 292 nm and 200 nm. Figure 3.10: Illustration of the effect of inelastic scattering in a TRPES experiment measured for aqueous phenol at same pump wavelength (266 nm) but with different probe wavelength (either 292 nm and 200 nm). The time delay is 1 ps. Our expectation was that by introducing higher probe energy photon, the overall spectrum will simply shift towards higher eKE by the difference in the photon energies of probe (hѵ200 nm – hѵ292 nm) since the pump (excitation) wavelength is the same. The black curve represents the experimentally measured spectrum when the probe wavelength is 292 nm. The red curve is just a 74 shifted form of the black curve with extra probe energy which is (hѵ200 - hѵ292 = 6.2 eV- 4.2 eV = 2 eV), which was expected. In reality, we observed the blue spectrum experimentally when the pump was the same and the probe was changed from 292 nm to 200 nm. By comparing the red and blue curve, it is clear that the inelastic scattering is not only changing the true eKE distributions but also distorting the spectral shape. 3.5 Conclusion UV photoelectron spectroscopy in liquid microjet provides a great promise to unravel dynamical features and the electronic structure of molecules in the solution phase. However, there are technical challenges, such as inelastic scattering of photoelectrons within the liquid microjet and loss of low eKE electrons expressed by the transmission function, which needs to be addressed. We demonstrated that inelastic scattering not only affects the absolute value of binding energy of the transients but also causes spectral distortion in the photoelectron spectra. The loss of low eKE electrons due to the transmission function of the spectrometer obscures photoelectron bands appearing towards the lower eKE region, significantly degrading the information carried by the photoelectrons. From the transformation method proposed by Suzuki and his coworkers, it is possible to retrieve the true spectral features particularly for spectra possessing a single photoelectron band. In the future, we want to improve on this spectral retrieval method for the photoelectron spectra containing more than one bands. 75 3.6 References 1. B. N. Papas; M. S. Schuurman; D. R. Yarkony, J. Chem. Phys. 2009, 130. 2. C. P. Schick; S. D. Carpenter; P. M. Weber, J. Phys. Chem. A 1999, 103, 10470-10476. 3. K. Seki; H. Inokuchi, Chem. Phys. Lett. 1979, 65, 158-160. 4. G. A. Worth; R. E. Carley; H. H. Fielding, Chem. Phys. 2007, 338, 220-227. 5. M. T. Zanni; V. S. Batista; B. J. Greenblatt; W. H. Miller; D. M. Neumark, J. Chem. Phys. 1999, 110, 3748-3755. 6. M. Gochel-Dupuis; J. Delwiche; M.-J. Hubin-Franskin; J. E. Collin, Chem. Phys. Lett. 1992, 193, 41-48. 7. A. Roy. Following Redox Chemistry and Excited State Dynamics in Solution Using Liquid Jet Photoelectron Spectroscopy. University of Southern California, 2015. 8. M. Faubel; S. Schlemmer; J. P. Toennies, Z. Phys. D: At., Mol. Clusters 1988, 10, 269. 9. G. Kumar; A. Roy; R. S. McMullen; S. Kutagulla; S. E. Bradforth, Faraday Discuss. 2018, 212, 359-381. 10. A. Roy; R. Seidel; G. Kumar; S. E. Bradforth, J. Phys. Chem. B 2018, 122, 3723-3733. 11. F. Buchner; B. Heggen; H.-H. Ritze; W. Thiel; A. Lübcke, Phys. Chem. Chem. Phys. 2015, 17, 31978-31987. 12. F. Buchner; A. Nakayama; S. Yamazaki; H.-H. Ritze; A. Lübcke, J. Am. Chem. Soc. 2015, 137, 2931-2938. 13. F. Buchner; H.-H. Ritze; J. Lahl; A. Lübcke, Phys. Chem. Chem. Phys. 2013, 15, 11402- 11408. 14. A. Lübcke; F. Buchner; N. Heine; I. V. Hertel; T. Schultz, Phys. Chem. Chem. Phys. 2010, 12, 14629. 15. T. Suzuki, Bull. Chem. Soc. Jpn. 2014, 87, 341-354. 16. M. H. Elkins; H. L. Williams; D. M. Neumark, J. Chem. Phys. 2015, 142, 234501. 17. M. H. Elkins; H. L. Williams; A. T. Shreve; D. M. Neumark, Science. 2013, 342, 1496- 1499. 18. Y.-I. Suzuki; K. Nishizawa; N. Kurahashi; T. Suzuki, Phys. Rev. E. 2014, 90, 010302. 19. B. Winter; R. Weber; W. Widdra; M. Dittmar; M. Faubel; I. Hertel, J. Phys. Chem. A 2004, 108, 2625-2632. 20. B. Winter; M. Faubel, Chem. Rev. 2006, 106, 1176-1211. 21. K. Nishizawa; N. Kurahashi; K. Sekiguchi; T. Mizuno; Y. Ogi; T. Horio; M. Oura; N. Kosugi; T. Suzuki, Phys. Chem. Chem. Phys. 2011, 13, 413-417. 22. M. N. R. Ashfold; A. L. Devine; R. N. Dixon; G. A. King; M. G. D. Nix; T. A. A. Oliver, Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 12701-12706. 23. G. A. King; T. A. A. Oliver; M. G. D. Nix; M. N. R. Ashfold, J. Phys. Chem. A 2009, 113, 7984-7993. 24. T. A. A. Oliver; Y. Zhang; M. N. R. Ashfold; S. E. Bradforth, Faraday Discuss. 2011, 150, 439 - 458. 76 25. G. A. K. Thomas A.A. Oliver, David P. Tew, Richard N. Dixon and Michael N.R. Ashfold, In preparation 2012. 26. M. G. D. Nix; A. L. Devine; B. Cronin; R. N. Dixon; M. N. R. Ashfold, J. Chem. Phys. 2006, 125, 133318. 27. Z. Lan; W. Domcke; V. Vallet; A. L. Sobolewski; S. Mahapatra, J. Chem. Phys. 2005, 122, 224315. 28. T. A. Oliver; Y. Zhang; A. Roy; M. N. Ashfold; S. E. Bradforth, J. Phys. Chem. Lett. 2015, 6, 4159-4164. 29. D. Ghosh; A. Roy; R. Seidel; B. Winter; S. E. Bradforth; A. I. Krylov, J. Phys. Chem. B 2012, 116, 7269-7280. 30. K. R. Wilson; B. S. Rude; J. Smith; C. Cappa; D. T. Co; R. D. Schaller; M. Larsson; T. Catalano; R. J. Saykally, Rev. Sci. Instrum. 2004, 75, 725. 31. A. Bernas; C. Ferradini; J.-P. Jay-Gerin, Chem. Phys. 1997, 222, 151-160. 32. A. P. Gaiduk; T. A. Pham; M. Govoni; F. Paesani; G. Galli, Nat. Commun. 2018, 9, 247. 33. J. V. Coe; A. D. Earhart; M. H. Cohen; G. J. Hoffman; H. W. Sarkas; K. H. Bowen, J. Chem. Phys. 1997, 107, 6023-6031. 34. V. Ziaei; T. Bredow, J. Phys.: Condens. Matter 2018, 30, 215502. 35. D. Luckhaus; Y.-i. Yamamoto; T. Suzuki; R. Signorell, Sci. Adv. 2017, 3, e1603224. 36. R. Signorell; M. Goldmann; B. L. Yoder; A. Bodi; E. Chasovskikh; L. Lang; D. Luckhaus, Chem. Phys. Lett. 2016, 658, 1-6. 37. M. Michaud; A. Wen; L. Sanche, Radiat. Res. 2003, 159, 3-22. 38. S. Karashima; Y.-i. Yamamoto; T. Suzuki, J. Phys. Chem. Lett. 2019, 10, 4499-4504. 39. M. J. Vrakking; D. Villeneuve; A. Stolow, Phys. Rev. A. 1996, 54, R37. 40. C. P. Schick; S. D. Carpenter; P. M. Weber, J. Phys. Chem. A 1999, 103, 10470-10476. 41. J. W. Riley; B. Wang; J. L. Woodhouse; M. Assmann; G. A. Worth; H. H. Fielding, J. Phys. Chem. Lett. 2018, 9, 678-682. 42. P. R. Tentscher; R. Seidel; B. Winter; J. J. Guerard; J. S. Arey, J. Phys. Chem. B 2014, 119, 238-256. 43. P. Slavíček; B. Winter; M. Faubel; S. E. Bradforth; P. Jungwirth, J. Am. Chem. Soc. 2009, 131, 6460-6467. 44. G. Mainfray; C. Manus, Appl. Opt. 1980, 19, 3934-3940. 45. L. Lompre; G. Mainfray; C. Manus, J. Phys. B: At., Mol. Opt. Phys. 1980, 13, 85. 46. A. Henley; J. Riley; B. Wang; H. Fielding, Faraday Discuss. 2019. 47. T. N. Das, J. Phys. Chem. A 2005, 109, 3344-3351. 48. A. L. Devine; M. G. D. Nix; B. Cronin; M. N. R. Ashfold, Phys. Chem. Chem. Phys. 2007, 9, 3749-3762. 49. R. N. Dixon; T. A. A. Oliver; M. N. R. Ashfold, J. Chem. Phys. 2011, 134, 194303. 50. G. Grabner; G. Koehler; G. Marconi; S. Monti; E. Venuti, J. Phys. Chem. 1990, 94, 3609- 3613. 51. C. P. Schick; P. M. Weber, J. Phys. Chem. A 2001, 105, 3725-3734. 77 52. Y. Zhang; T. A. A. Oliver; M. N. R. Ashfold; S. E. Bradforth, J. Phys. Chem. Lett. 2015, 6, 4159-4164. 53. Y. Zhang; A. Roy; S. Das; T. A. A. Oliver; M. N. R. Ashfold; S. E. Bradforth, 2010, in preparation. 54. Y. Zhang; T. A. A. Oliver; M. N. R. Ashfold; S. E. Bradforth, Faraday Discuss. 2012, 157, 141-163. 55. T. N. Das, J. Phys. Chem. A 2005, 109, 3344-3351. 56. M. G. Nix; A. L. Devine; B. Cronin; R. N. Dixon; M. N. Ashfold, J. Chem. Phys. 2006, 125, 133318. 78 Chapter 4. The Influence of Aqueous Solvent on the Electronic Structure and Non- Adiabatic Dynamics of Indole Explored by Liquid-jet Photoelectron Spectroscopy 4.1. Introduction The complex yet fundamental photophysics and photochemistry of tryptophan, one of the essential amino acids, is of considerable importance due to its pivotal role in numerous biological processes/pathways. 1-4 Tryptophan absorbs strongly in the ultraviolet region of the solar spectrum. The absorption and emission bands for tryptophan are well separated and its fluorescence emission is highly sensitive to its local environment which makes it a good probe to study the dynamics of proteins in different environments. 5, 6 In addition to being an important intrinsic probe for studying protein dynamics, it also provides a spectroscopic handle for the determination of the presence of amino acids in astrobiological studies. 7-9 The photophysics of tryptophan is governed by its chromophore indole. Because of the importance of the tryptophan residue, there is an extensive spectroscopic literature on indole ranging from high-resolution molecular-beam electronic spectroscopy to sub-picosecond ultrafast spectroscopy in a variety of solvents, including water. 10-16 The fluorescence decay of indole in water shows a mono-exponential decrease fluorescence decay with a 4.3 ns lifetime, while in contrast aqueous tryptophan is described by a double exponential decay with lifetimes of 3.14 ns and 0.51 ns. The additional complexity for tryptophan results from the multiple rotamers existing at room temperature with different deployment of the charged amino acid elements attached to the indole chromophore. 17, 18 The single fluorescence lifetime for indole belies the fact that considerable picosecond and sub-picosecond dynamics have been implicated for the excited state chromophore in water, including a significant fraction of the excited state population branching into a photoionization channel to generate solvated electrons. 79 It is the presence of close-lying electronically excited states in the UV-absorbing amino acids and their molecular precursors (indole for tryptophan; phenol for tyrosine) that vastly increase the complexity of their excited state dynamics. 19-22 In order to understand the fast electronic and nuclear dynamics, uncovering the time-evolving electronic character of the excited state has turned out to be very important. The Born-Oppenheimer approximation breaks down as two potential energy surfaces, belonging to different electronic states, approach each other. 23 In other words, the coupling between the electronic and nuclear degrees of freedoms results in nonadiabatic pathways for electronic relaxation. Non-adiabatic dynamics is often manifested at conical intersections where two potential energy surfaces would otherwise cross. In indole, the absorption spectrum is dominated by two bound 1 ππ* valence states, known as the 1 La and 1 Lb (in accordance with Platt-Murrell nomenclature 24, 25 ), which cross a dissociative surface with 𝜋𝜎 * character along the N-H coordinate. The 𝜋𝜎 * state also exhibits a conical intersection (CI) with the ground electronic state at elongated N-H distance. 12, 26-28 Due to presence of multiple conical intersections, the excited state relaxation pathways, even in the isolated molecule, can become quite rich. According to UV-absorption spectra (Figure 4.1), excitation in the longest wavelength UV region (which is the focus of most experimental studies including this one) involves the optically accessible low-lying excited states which are of ππ* and πσ* character. The broad peak in the range of 220 - 280 nm is primarily attributed to the absorption from the ground state to one of the ππ* states called 1 La, in accordance with Platt-Murrell nomenclature. 24, 25 Another ππ* state, 1 Lb, mainly contributes in the region longer than 280 nm. In water, the maximum of the absorption spectrum shifts towards longer wavelength by ~20 nm. It has been determined experimentally that the band origin of 1 Lb and 1 La lies around 283 nm and 273 nm in the gas phase respectively. 29, 30 80 Figure 4.1: (a) UV absorption spectra of indole in water compared with gas phase and aqueous tryptophan. Two peaks (magenta and blue) represents the wavelength used in our R2PI and TRPES experiments. (b) Deconvolution of the relative contribution of L a (red) and L b (blue) from total absorption spectrum of indole in propylene glycol (a hydrogen-bonding solvent) extracted from Ref. 31 Although both 1 La and 1 Lb are of ππ* character, their static dipoles are markedly different. 32 In the gas phase, the maximum of 1 La absorption always lies above the maximum of 1 Lb which makes 1 Lb the fluorescent state according to Kasha’s rule. But dissolved in solvent, the relative energy separation between 1 La and 1 Lb in indole strongly depends on the dielectric constant of the medium. Due to the large static dipole (6.12 D) of the 1 La state, it is stabilized in polar environments, like water, due to dipole-dipole interactions. This results in the lowering of 1 La relative to 1 Lb to an extent where 1 La now becomes the state where emission originates and the fluorescence spectrum exhibits a large Stokes shift. 32-34 Figure 4.1(b) shows the resolution of the excitation spectrum of indole along with the relative contribution of the 1 La and 1 Lb states. 31, 35 There is still a debate in the literature whether the reversal of energy order for 1 La and 1 Lb happens before or after excitation. Roos and coworkers have shown computationally that the vertical excitation energy to 1 La and 1 Lb changes by 0.06 eV and 0.03 eV, respectively, in water but the 81 vertical Franck Condon region in the 1 La state remains higher in energy than 1 Lb. In other words, the solvent affects mainly the emission properties of indole. 33 Despite the wealth of literature on gas-phase spectroscopy and electronic structure theory, aqueous phase studies are still not conclusive. Excited state studies of indole and tryptophan have been carried out using pump-probe spectroscopy and fluorescence upconversion, however a clear unified picture has not emerged. 11, 13 Kohler and coworkers focused on the photoionization dynamics of indole in water ensuing excitation at 260 nm and confirmed the generation of the solvated electron within their experimental time resolution of 200 fs. They also suggested that the geminate recombination between the indole radical cation and the solvated electron appears to be surprisingly slow, taking place only on a near-nanosecond timescale. 10 Chergui and coworkers reported the relaxation dynamics of tryptophan in water using fluorescence up-conversion and concluded that the internal conversion between 1 La to 1 Lb occurs in only 45 fs while the solvation dynamics has two characteristic time scales, 160 fs and 1 ps. 13 Haacke and coworkers have also reported a similar time scale for the solvent relaxation. 11 There has been no report implicating the involvement of the 𝜋𝜎 * state of indole (or tryptophan) in the liquid solution. On the other hand, time-resolved photoelectron spectroscopy (TRPES) of gas-phase indole is strongly suggestive of the involvement of the 𝜋𝜎 * state at four different excitation wavelengths from 249 nm and 273 nm. 26, 36 Recently, the labs of Lübcke and Neumark have published a series of papers on aqueous nucleic acid monomers (bases, nucleosides and nucleotides) by using liquid jet implementation of TRPES, with the goal of unraveling the excited state population dynamics by mapping the electronic character of the populated excited state onto the ionization continuum. 37-40 These 82 pioneering experiments suggest the promise, and some limitations, of this technique in studying excited state dynamics of solvated molecules. In this chapter, our goal is to apply TRPES in a liquid micro-jet to elucidate the pathways of electronic relaxation for both aqueous indole and tryptophan. In many ways, the excited state dynamics of the chromophore in these systems are less complex than the nucleobases, which are very short-lived in their electronic excited states. In this study, we explore the influence of polar solvent on the excited state dynamics of indole as reflected in change in the solute valence electronic structure, stabilization of charge separated states relative to the gas-phase and change in the onset of conical intersection (CI) which determine the photochemical branching. 4.2. Experimental The details about the liquid-jet photoelectron spectrometer is already given in the chapter 2 of this thesis and therefore, it will not be repeated here. In time-resolved configuration, the two laser beams are spatially overlapped on the laminar flow region of the liquid jet. The excitation wavelengths and the pulse energies used in the experiment are chosen carefully to ensure that no three-photon ionization from water occurs. A tunable pump beam (240-350 nm) is generated by doubling the visible output from a homemade Non-collinear Optical Parametric Amplifier (NOPA) following the design of Riedle et. al. 41, 42 The NOPA is pumped with the output of a 35-fs Ti:Sapphire amplifier system (Coherent Legend Elite USP, 1 kHz repetition rate). To accomplish optimal temporal resolution in our experimental setup, the visible output and the second harmonic of the NOPA are compressed using prism-pair compressors, with prisms of fused silica and calcium fluoride respectively. The second, probe, beam fixed at 266 nm is generated by sum frequency of the regenerative amplifier fundamental (centered at 800 nm) and its second harmonic in a BBO Type II crystal of 150 m thickness. To 83 compensate for dispersion in the optics making up the probe beam path, the 266 nm is also compressed using a calcium fluoride prism-pair compressor before directing to the liquid jet interaction region inside the vacuum chamber. For optimal signal, the spot size of both beams was maintained below 80 m at the interaction point, as determined by a translating knife edge. Typical pulse energies are 15-25 and 3-5 nJ in the pump and probe, respectively. Unless otherwise stated, the polarization of both beams was vertical with respect to the laboratory frame and perpendicular to the time of flight axis. The experiments using perpendicular polarization were also carried out, with no obvious differences in the signal shapes in the temporal or kinetic energy domains. In a single-color resonant two photon ionization (R2PI) experiment, a single beam of ultrafast pulses is used to photoionize a chromophore bearing molecule. The first photon absorbed excites the molecule to the intermediate excited state, which is on resonance with the pulse energy, and a second photon absorbed under the same pulse envelope subsequently ionizes the molecule. The R2PI experiments were performed using either at 266 nm or at 292 nm. The typical energies used for 266 nm and 292 nm pulses were 6 nJ and 35 nJ respectively; it was ensured that the electron count rate was < 10 electrons per shot. Under our experimental conditions, both the pump and the probe beam individually yield photoelectrons due to resonant one-color two-photon ionization. 43 To eliminate the resonant two photon ionization signals from the transient pump-probe signal, a combination of two identical mechanical choppers (Thorlabs), each operating at 250 Hz with /2 phase shift, has been exploited. Using this configuration, one can acquire a complete transient photoemission spectrum from 4 consecutive laser shots without worrying about longer time laser intensity drift and 1/f noise. The pump-probe cross-correlation width was around 190 fs (Figure 4.S1 in the postscript of this chapter). The non-resonant two color (1+1') multiphoton ionization of sodium iodide solution in 84 water (100 mM) was performed in the microjet. At each delay point, the signal from the pump alone and the probe alone, which were invariant in time, were subtracted from the overall pump- probe signal. The FWHM was extracted after plotting the total PE integrated signal as a function of pump-probe delay time. Sample preparation: Phenol (≥99.5%; Sigma-Aldrich) and L-tryptophan (>99%; Beantown Chemical) were purchased and used without further purification. Indole (99%; Sigma) was purified to remove brown crystals of presumably oligomers of indole. An amount of 40/60 petroleum ether was added to the impure indole and heated to boiling. Pure indole was dissolved in the top, clear layer of petroleum ether, whilst a layer of molten brown impurity was present in the lower layer. Prior to cooling, the top layer was decanted off from the brown mother liquor, recrystallized and dried under reduced pressure. Solutions of purified indole remain colorless if prepared at a pH 7 buffer and turn yellow if not purified and buffered. Solutions of indole were prepared in a buffer consisting of monobasic potassium phosphate and dibasic sodium phosphate. Two types of stock solutions (4 mM) of the purified indole were then prepared, one with 30 mM NaCl to reduce the jet streaming potential caused by electrokinetic charging and the other with 0.5 M KNO3, where the nitrate ion serves both as a diffusive scavenger for solvated electrons and also as a static quencher for the primary excited state. 44, 45 4.3 Results Our current study aims to explore the time evolution of the photoexcited state of indole and its parent molecule, tryptophan, as the molecule undergoes non-adiabatic electronic and structural relaxation. PE spectroscopy approaches the problem by mapping out the change in the excited state electronic wavefunction (more correctly for an ensemble measurement, the system density) by projecting it on the basis of cationic states. 85 4.3.1 Resonant Two-Photon Ionization Photoelectron Spectra R2PI-PES of aqueous indole have been recorded under conditions of two photon ionization at 266 nm or 292 nm. The R2PI-PES of indole are shown in Figure 4.2; the equivalent spectra of tryptophan recorded on the same day are shown in the postscript of this chapter (Fig. 4.S2). Our main purpose for recording the R2PI spectra is to extract the vertical ionization energies (VIE) of aqueous indole and compare it with indole in the gas phase and aqueous tryptophan. These spectra provide a t = 0 slice of the overall dynamics. But they also serve to illustrate the complications and limitations with energy resolving the outgoing electrons emerging from a liquid jet, as discussed in the previous section: namely, that inelastic scattering distorts the PE spectrum, and this does not lead to a simple additive shift because of the electron kinetic energy dependence of the energy loss. A single very broad photoemission band that cannot be represented by a single Gaussian was observed in the R2PI-PES recorded at 266 nm ionization. It can be well fit by two Gaussians centered at 0.8 and 1.5 eV with an onset at high eKE of ~3.1 eV (the latter is only obvious on a logarithmic eKE scale; the experiment has at least three decades of signal sensitivity). The peak spacing is similar to that fitted for the first and second ionization energies in the XPS spectrum of tryptophan (see below). 43 On the other hand, a narrower PE band, which can be fit by a single Gaussian centered at 1 eV and an onset of ~2.5 eV, is observed when 292 nm is employed as the ionizing light. The centers of the higher eKE Gaussians are assigned as the (first) vertical ionization energies and are listed in Table 4.1. In the case of 292 nm, the second ionization band appears to be missing; it either falls well under the transmission cut-off of our instrument or is not present in the data. 86 Figure 4.2: Resonant two-photon ionization spectra of aqueous indole recorded with a (a) 266 nm ionization and (b) 292 nm ionization wavelength. In panel (a) individual Gaussians used to fit the spectrum are shown in red and green; the total Gaussian fit is represented in blue. For panel (b) the spectrum can be fit with a single Gaussian (in red). Rectangles depict the center of the Gaussian. Note that the R2PI spectra are affected by instrument transmission function at eKE < 0.5 eV . The comparison of two R2PI spectra indicates the importance of applying a correction for inelastic scattering to the peaks. According to the simulations of Luckhaus et al., photoelectrons departing with an eKE of 0.8 eV (the center of the 266nm lower band) should be minimally affected by electron inelastic scattering while those at 1.0 eV (292 nm PE center) or 1.5 eV (the upper band center at 266 nm) would be expected to be affected more significantly. 46 After applying an inelastic scattering correction inferred from ref. 46 , the bands at 0.8, 1.0 and 1.5 eV are shifted to the right (towards higher eKE) by 0.05, 0.1 eV and 0.4 eV, respectively. (See postscript of this chapter for approximate inelastic energy loss shifts). The computed difference in the corrected eKE between the higher eKE peaks obtained at 266 nm and 292 nm respectively now comes to 0.8 eV; essentially equal to the difference of the doubled photon energy (9.3 eV – 8.5 eV). 87 Table 4.1: R2PI peak positions and derived vertical ionization energies for aqueous indole compared with the isolated molecule in the gas phase. All energies in eV . Ionized orbital eKE a (266 nm R2PI/ 292 nm R2PI) Corrected eKE b (266 nm R2PI/ 292 nm R2PI) VIE c (266 nm R2PI/ 292 nm R2PI) VIE d (gas phase) HOMO (π3) 1.5 / 1 1.9 / 1.1 7.4 / 7.4 7.9 HOMO-1(π2) 0.8 / - 0.9 / - 8.4 / - 8.4 a electron kinetic energies b eKE reported after correcting for electron inelastic scattering c vertical ionization energies d VIE reported in gas phase 47-50 Recently, our group reported the vertical ionization energies (VIE) of aqueous tryptophan corresponding to D0-S0 and D1-S0 with one-photon non-resonant ionization using soft X-rays. We reported 7.3 ± 0.1 eV and 8.0 ± 0.1 eV using XUV synchrotron radiation at BESSY II. 43 The ionization energies corresponding to D0-S0 and D1-S0 for indole in water are 7.4 ± 0.1 eV and 8.4 ± 0.1 eV after correction for inelastic scattering for the low eKEs in the R2PI experiments, quite similar to the values reported for tryptophan. As expected, on introducing a condensed polarizable medium, there is a significant decrease in VIE compared to the isolated molecule in vacuum (0.5 eV). The VIE is influenced by the relative stabilization in water of the neutral ground state and the final cationic state (but at the neutral solvent shell structure); the fast electronic polarization of the environment also plays a role. 43, 51 Upon further examination of Table 4.1, it is evident that hydration has a large effect on the ionization corresponding to D0-S0. However, a corresponding increase in the D1-D0 gap, at the neutral indole configuration, keeps the D1-S0 second ionization energy almost unaltered. A similar effect is observed for tryptophan (see postscript of this chapter). From the similarity in the onsets of the R2PI-PES and XPS photoelectron spectra of tryptophan, we previously concluded that the extent of relaxation from solvation and/or 88 intramolecular vibrational redistribution/vibrational energy relaxation is relatively small for the indole moiety. A time-resolved study can test this notion. 4.3.2 Time-Resolved Photoelectron Spectra In time-resolved photoelectron measurement, the excitation wavelengths were chosen such that the total photon energy (8.9 eV) is considerably greater than the VIE to the lowest cation state, D0. Depending on how severe the spectrometer cut-off function is, the cation D1 is also within reach. With an excitation pulse of 292 nm, the two singlet excited states ( 1 La and 1 Lb) are approximately equally populated as judged by the ratio of absorption strength ( 1 La / 1 Lb) at this wavelength. 13, 31 A contour plot showing the photoelectron spectrum as a function of pump-probe delay is displayed in Figure 4.3(a) for aqueous indole following 292 nm excitation; 266 nm here is the probe wavelength. From the spectrum, it is quite clear that a long-lived species is present in the region of 0.5 – 1 eV with a peak maximum at 0.75 eV. In addition to the photoemission band at 0.75 eV, a second distinct but broader band is also apparent in the higher kinetic energy range between 1.2 to 2.3 eV. The photoemission intensity in the higher kinetic range also remains to the longest delays but with less intensity comparatively. Figure 4.3(b) represents a contour plot when reversing the pump and probe in terms of the order they intersect the liquid jet. With this ordering, the dynamics recorded reflects an excitation wavelength centered at 266 nm, with the later 292 nm pulse now acting as the ionizing probe pulse. From Figure 4.1(b), it can be inferred that at 266 nm, the initially excited population now favors the 1 La state by about 3:1. The PE spectrum in Figure 4.3(b) shows a long-lived structure that is weighted toward the lowest kinetic energy range, but the total PE bandwidth is smaller; the spectrum does not extend to as high eKE in accordance with the smaller energy in the probe photon. 89 Figure 4.3: Time-resolved photoelectron spectra of aqueous indole plotted with pulse time- ordering (a) 292 nm pump with subsequent 266 nm probe and (b) 266 nm pump and subsequent 292 nm probe. Both beams are vertically polarized with respect the laboratory frame. Figure 4.4 shows the same information as a set of PE spectra at fixed time delays between pulses but is more revealing as to the detail of the spectral dynamics. With 292 nm used for excitation, the PE spectrum shows three time epochs: (i) faster than 500 fs, there is also a developing dip in the PE spectrum around 1.25 eV, (ii) between 1 and 20 ps (Fig. 4.4(a,b)), it is difficult to see any signal change in the range of 0.5 to 1 eV eKE but the photoemission bands in the higher kinetic range are affected significantly. The colored rectangles illustrate the centers of the three Gaussians which can be used to fit an early time slice. Figure 4.4(b) clearly reveals that the higher eKE photoemission (PE) band which starts centered around 1.8 eV shifts towards lower kinetic energy, converging at around 1.55 eV eKE peak center, with a small decay in overall intensity over 20 ps. There is no accompanying rise anywhere in the spectrum. At longer pump- probe delay (Figure 4.4(d)), there is a clear decay of all PE bands but there is no further discernible spectral evolution. 90 Time slices of the PE spectrum are shown in Figure 4.5 when indole is excited instead at 266 nm. It is evident that the spectral shape is completely different compared to 292 nm excitation. A resolved peak at higher eKE is not so evident. In fact, one can observe an isosbestic point (Figure 4.5(a)) at earliest times which reveals that highest most eKE intensity is decaying while the other PE bands are growing. The spectral feature at lowest eKE appears to be similar to that observed with 292 nm excitation. Because the 4.66 eV excitation photon energy lies above the ionization threshold of indole (4.35 eV) by 0.3 eV, 52 we can expect the generation of solvated electrons to be part of the dynamics. The prompt photoionization yield reported in the literature is 45±15%. 53 Kohler and coworkers have reported the appearance of solvated electrons within 200 fs in aqueous indole with an excitation wavelength of 260 nm. So, a clear possibility for the rise at < 300 fs delay is a contribution of solvated electrons which are expected to overlap in outgoing eKE. As the electrons first trap and then fully solvate, the band will grow first at lower binding energies (high eKE) and then shift from high eKE to lower eKEs, with the lower eKE region expected to grow over the solvation timescale (as observed). Once fully relaxed, we anticipate a binding energy for the solvated electron, Eb ≈ 3.7 eV, 46 and so an eKE peak center at hνprobe - 3.7 ~ 0.5 eV (although see below about peak shape). To verify the signature of the solvated electron, KNO3 was used as an electron scavenger. Although a strong acid such as HCl is more typically used as a diagnostic solvated electron scavenger, KNO3 is preferred here over HCl because acidic solutions can initiate the acid catalyzed polymerization of indole. 54 91 Figure 4.4: TRPES spectra extracted from full dataset in Figure 4.3 (a) where aqueous indole is photoexcited with 292 nm and photoionized with 266 nm subsequently. Panel (a) displays a developing dip around 1.25 eV in the TRPES spectra within half a picosecond while panel (b) mainly demonstrates the shift of higher eKE photoemission bands towards lower eKE with a small decay in overall intensity. Panel (c) displays the comparison between an experimental time slice at 20 ps (orange), for which an attempt to reproduce the spectrum from two Gaussians is also displayed (see text). This does not replicate the experimental slice fully at low electron kinetic energies. But combining these two Gaussians with an estimated cutoff function for our spectrometer now reproduces the rapidly decaying signal intensity at low electron kinetic energies. In panel (d), the PE signal (on logarithmic scale) is plotted at four selected eKEs as a function of time delay. The experimental conditions were different for short time pump-probe delays (upper panel (d)) and for longer delay times (lower panel (d)) measurements (datasets are recorded on different days). 92 Figure 4.5: TRPES spectra extracted from full dataset in Figure 4.3 (b) where aqueous indole is photoexcited with 266 nm and photoionized with 292 nm subsequently. Panel (a) displays the decay of the highest eKE PE band and the rise of other PE bands—depicted with arrows—resulting into an isosbestic point within half a picosecond while panel (b) shows a clear decay of the same PE bands over a longer 20 ps timescale. Panel (c) displays the comparison between an experimental time slice (2 ps delay) with a constructed time slice composed of two individual Gaussians belonging to 1 La, and a third Gaussian representing the solvated electron along with an estimated cutoff function of the spectrometer (see text for details). In panel (d), the PE signal (on a logarithmic scale) is plotted at four selected eKEs as a function of time delay. Inset of the top panel shows the rising signal in the first 2 ps with a linear signal scale. Experimental conditions were different for short time pump-probe delay (upper panel (d)) and for long time pump-probe delay (lower panel (d)) measurements (datasets recorded on different days). The mutual polarization of the two beams is parallel. 93 (NO3 - is also known to act as a scavenger that can quench directly from the optically excited state if the excited state has a spatially extended electronic wavefunction. NO3 - therefore scavenges the precursors to solvated electrons as well as diffusively scavenging solvated electrons, while H + only diffusively scavenges solvated electrons.) 44 From the PE spectra shown in Figure 4.6, we find that the addition of KNO3 leads to a noticeable increase in the signal decay over the hundreds-of-picosecond timescale for all the PE bands and with a rate commensurate with the known e - aq/NO3 - scavenging rate constant and [NO3 - ] used in the experiment. The impact of scavenging on the PE spectrum is most obvious at the last time slice (300 ps) in Figure 4.6(b) where a more structured high eKE peak becomes visible once all the solvated electrons have been scavenged, reminiscent of the 292 nm pump data. Analysis of the quenching experiment is complicated by the fact that nitrate also quenches the excited state directly, 44 a result confirmed by a large quenching of the fluorescence quantum yield in steady state fluorescence measurements on aqueous indole (data not shown). Polarized pump-probe measurements have also been performed; no anisotropy evolution was observed. Figure 4.6: TRPES spectra at selected pump-probe delay for (a) 4 mM indole in water (b) 4 mM indole with 0.5 M KNO3 (quencher) in water, both measured with 266 nm excitation and 292 nm ionization. The nitrate anion quenches both solvated electrons as well as the excited state, resulting in an increased decay rate of all the PE bands as displayed in panel (c). Note that the signal axis in (c) is plotted on logarithmic scale. The polarization of the pump and probe are perpendicular with respect to each other. 94 4.4 Spectral Analysis After careful consideration of the results for aqueous indole shown in Figures 4.2 – 4.5, we note the following. At near zero delay: (1) The 266 nm R2PI spectra (Fig. 4.2) has been assigned with two Gaussian component bands underlying the broadened peak, based on assignment of non-resonant XPS valence spectra of tryptophan. We therefore assign the two peaks as ionization originating from the HOMO and HOMO-1 orbitals of aqueous indole (Table 4.1). Unlike the case of phenol, the two bands aren't necessarily the result of configuration interaction in the intermediate state. That is because two overlapping excitation transitions in indole are resonant at 266 nm; Roos et al. have assigned as primarily HOMO →LUMO and antisymmetric admixture of HOMO-1→LUMO and HOMO →LUMO+1 for 1 La S0 and 1 Lb S0, respectively. 33 (2) Because of the 0.81 eV lower total photon energy, the 292 nm R2PI spectra reports on just one component. and as commented in Chapter 3, The impact of the cut-off/transmission function and the larger differences in inelastic scattering shifts leads to greater distortion between the two spectra and poorer matching of the high eKE sides of the PE band shape compared to the model case of phenol, described in Chapter3. It should also be recognized that different (and non-trivial) relaxation dynamics are taking place within the pulse duration at the two wavelengths, which can be expected to lead to a more physical change in peak shape. (3) Unlike aqueous phenol, where the two-color time-resolved spectrum obtained at t=0 displayed an identical shape and high-energy threshold to the R2PI-PES simply with a shift in eKE accounted for by the probe photon energy difference, there is a significant difference between R2PI-PES and TRPES evident for aqueous indole (see Fig. 4. S3 in the 95 postscript of this chapter). Even after accounting for differences in inelastic scattering, there is an introduction of a clear shoulder at higher eKE in the TRPES spectrum. If we now consider the 292 nm pump TRPES spectra at the longest pump-probe delays where we expect indole to be relaxed electronically and for solvation to be complete, one might naively expect a single spectral feature present corresponding to ionizing the relaxed 1 La state, the state responsible for the fluorescence emission of indole. A quick calculation based on the anticipated energy for relaxed 1 La (~3.8 eV above the ground state), 34 the corrected VIE of the ground state (Table 4.1) and application of Koopman's theorem, would estimate an approximate ionization energy (to D0 cation) for the relaxed 1 La of 3.6 eV. So, with a 4.66 eV probe photon this would give rise to a peak at ~ 0.9 eV (after making a small inelastic correction of 0.1 eV). Therefore, the Koopmans' expected feature is the dominant lower eKE peak at 0.75 eV. However, there is a noticeable shoulder at much higher eKE present to the longest pump-probe delays (even well beyond 20 ps trace shown in Fig. 4.4). From the lower panel of Fig. 4.4(d) we observe that the slope of all the decay traces over longer time are similar. These observations suggest that the whole PE band belongs to the same electronic state 1 La. Now considering the assignment of the complete evolution of the 292 nm TRPES dataset, it has been reported that the 1 La → 1 Lb internal conversion (IC) happens in 45  15 fs. 13 Hence, even though at this pump wavelength the excitation approximately equally populates the 1 Lb state, it would be surprising if a signature for the 1 Lb was discernable in our measurement, since this state-switching is faster than the current instrument temporal resolution. If it were, and the ionization cross sections were approximately equal for the two states, one would expect to see a region of the spectrum rising as well as another region decaying. It is possible that the very fast 96 band sharpening and the development of a dip around 1.2 eV is associated with the state switching but no equivalent gain of PE intensity is seen anywhere else in the spectral window probed. The time constants of solvation dynamics on the 1 La surface on the other hand for tryptophan in water are 160 fs and 1 ps, which means that the signature of solvent relaxation should be observable. 11, 13 We clearly see shifting to lower eKE in the band around 1.8 eV but with a longer associated time constant of ~ 2 ps and surprisingly no apparent shifting of the lower peak centered at 0.75 eV. We will return to this point later. At lower eKE we have already commented that the spectral shape is being distorted by the decreasing transmission efficiency at these energies. In Fig. 4.4(c), we have constructed a simulated time slice using two Gaussians (centered at 0.66 eV and 1.6 eV) designed to match the experimental time slice measured at t=20 ps using an estimated cutoff function. The constructed spectrum is quite similar to the experimental time slice and demonstrates how the spectrometer transmission cutoff function affects the spectral feature appearing lower than 0.6 eV. This will become important in making a cross-comparison of the long time PE spectra recorded with two different probe photon energies (Figure 4.4 and 4.5). Switching now to considering the 266 nm pumped experiments (Figure 4.5), the major impact on the spectroscopy is the ~ 0.4 eV lower probe photon energy. Although 266 nm excitation excites substantially less population on 1 Lb (Fig. 4.1(b)), as the state switching is not resolved, this will only change the disposition of excess energy. As already discussed, the other major change is that the pump photon energy exceeds the threshold for the photoionization channel; the appearance of solvated electrons in the time-resolved dataset is confirmed by the changes to the TRPES on adding an electron scavenger (NO3 - ) to the indole solution. Other evidence which strengthens the claim of solvated electron formation comes from 266 nm experiments (shown in 97 the postscript of this chapter) conducted on aqueous tryptophan where there is a much less apparent rise in intensity in the lower eKE region of the spectrum that we have assigned to appearance of solvated electrons. Since, the quantum yield of photoionization is three to four times lower in tryptophan compared to aqueous indole, 55-57 it is consistent that a rising feature would not be as noticeable as observed in aqueous indole. The final puzzle is in considering the long-time PE spectrum. Taking into account that only signatures for the solvated electron and the 1 La should remain, and that the 1 La features should be shifted to ~ 0.4 eV lower eKE, it is hard to understand why the dominant feature again appears at 0.75 eV, the same position recorded with a higher probe photon in Figure 4.4. Figure 4.5(c) goes some way to explain this observation. The equivalent two peak representation of the 1 La signature is used; the center positions of the Gaussians were shifted based on the photon energy used in the ionization step and the contribution coming from inelastic scattering. We also incorporated a third Gaussian here which represents the PE band of solvated electrons. The photoelectron kinetic energy position and FWHM of solvated electron PE band is taken from the ref. 46 and with area consistent with the literature solvated electron yield (assuming equivalent photoionization cross section for e - aq and 1 La). The shape of the constructed time slice (Fig. 4.5(c)) matches qualitatively with the experimental time slice and clearly shows that this experimental spectrum is severely affected by the transmission cut-off function of the spectrometer resulting into spectral distortion. The peak positions of all features in the Figure 4.4 and 4.5 are given in Table 4.2. 98 Table 4.2: Binding energies (BE) extracted from corresponding eKE peaks in TRPES spectrum. All energies in eV . 292 nm pump (excitation) and 266 nm probe (ionization) PE Bands eKE Corrected eKE BE (hνprobe - eKE) PE I a 0.75 0.75 3.9 PE II 1.4 1.8 2.9 PE III 1.8 2.3 2.3 266 nm pump (excitation) and 292 nm probe (ionization) PE Bands eKE Corrected eKE BE (hνprobe - eKE) PE I a 0.75 0.75 3.5 PE II 1.2 1.5 2.7 PE III 1.6 2 2.2 a Peak position strongly affected by spectrometer cut-off function on low eKE side. 4.5 Discussion To understand the non-adiabatic dynamics of indole, several time-resolved studies have been performed. Most of the experimental studies focus on the gas phase behavior of this molecule and only a few of them consider the influence of water on the non-adiabatic behavior. 10, 11, 13 Townsend and his coworkers investigated the spectroscopic behavior of gas phase indole at four excitation wavelengths between 249 nm and 273 nm with the same ionization wavelength of 300 nm using time resolved photoelectron spectroscopy. They concluded that at all wavelengths, the 1 La state decays by branching within 100 fs to 1 Lb and to 𝜋𝜎 * ; the latter population decays on the order of picosecond residing on the 𝜋𝜎 * for a surprisingly long time at long R(N-H). 26 The shape and the position of the spectral features were found to be identical for all pump wavelengths. The only difference was the relative amplitude of the features associated with the decay constants of 99 the electronic states, excited directly or indirectly. It was interesting that these authors found two parallel relaxation pathways at all excitation energies. The decay constant associated with 𝜋𝜎 * decreases slowly and monotonically from 1.2 ps at 273 nm excitation to 0.7 ps at 249 nm excitation. This difference of 𝜋𝜎 * lifetime may be due to the excess of excitation energy at 249 nm which launches the wavepacket above the 0.45 eV barrier calculated to lie along the N-H coordinate outside the FC region. An interesting outcome of the experiments from Townsend's group was the apparent involvement of 𝜋𝜎 * even at the longest excitation wavelength 273 nm. 26 In the aqueous phase, the landscape of the excited state potential energy surfaces is influenced by the presence of the solvent, which can change the position of conical intersections that drive efficient radiation-less transitions. Several computational studies performed on indole in water reveal that the polar nature of water mainly affects the emission properties of indole leading to inversion of the energy order of 1 La and 1 Lb; 1 La becomes the emitting state in water. 32, 33, 58 The inversion of the solvent-relaxed electronic states can be explained based on the argument of the higher permanent dipole moment of 1 La state; the 1 La state becomes lower in energy due to enhanced dipole-dipole interaction in water. 33 Although the relaxed state ordering is inverted compared to the gas phase, the state ordering is unchanged at the level of the vertical excitation energy from the ground state of indole. 33 34 What has not been as well established is the energy of 𝜋𝜎 * . The 𝜋𝜎 * state has the highest permanent dipole moment (12.4 D) which is almost double than the static dipole of the 1 La state (6.2 D), but it also has Rydberg character which can push its energy up in the order due to Pauli repulsion. 26 In this work, we are interested to see if we can capture the involvement of 𝜋𝜎 * in the relaxation dynamics of indole in water using TRPES. If we were to follow the proposal by Domcke and coworkers, 58 the fact that solvated electrons are produced with high quantum yields 59 100 immediately implicates this pathway and we should be looking to connect the spectral signatures of the 𝜋𝜎 * as it evolves into a solvated electron. Figure 4.1(a) reveals that the indole absorption maximum is red-shifted by almost 20 nm in water. Therefore, the experiments here at 292 nm are the closest analogue to the long wavelength excitation experiments in the gas phase, 273 nm, where the 𝜋𝜎 * is assigned to have its longest lifetime, ~ 1.2 ps. Based on our analysis so far, it is tempting to assign the very highest eKE feature to the  * channel. This may be reasonable, as this timescale does not match well with the 1 La solvation timescales in water 13 and as commented earlier we see no band shifting for the low eKE feature also assigned to PE signature from the 1 La. At 292 nm, we do see a loss of integrated signal with a timescale of 10 ps, and a narrowing of the higher energy side of the 1.8 eV peak without evidence of a rise anywhere else in the PE spectrum. According to Bernas et al., there should be no solvated electron production as the ionization threshold of indole has been estimated at 4.35 eV using the photoconductivity measurements, which is higher than the excitation energy 4.24 eV (292 nm). 52 Contradicting this observation, Ryuzi Katoh 55 performed nanosecond time resolved transient absorption on aqueous indole to examine the photoionization yield as a function of excitation wavelength (220-290 nm) and concluded that the solvated electrons are produced through a monophotonic process over the entire wavelength range. The long-time quantum yield remained approximately constant at 0.2 for all wavelengths longer than 250 nm. We are unable to observe any concomitant rise in any other PE bands at 292 nm excitation that might be assignable to the solvated electron, unlike at 266 nm excitation. One possibility is the electron being produced on a much slower timescale (but with similar yield to satisfy the hundreds of nanoseconds data of Katoh) and concomitantly with the slow decay of the 1 La state, similarly to the case of aqueous phenol at 266 nm excitation. 20 101 4.6 Conclusion and Outlook The vertical ionization energies of indole (7.4  0.1 eV, 8.4  0.1 eV) have been measured in water. The presence of aqueous environment lowers the first VIE by 0.5 eV. In our TRPES measurements, we observed clear spectral signatures that can be assigned to 1 La state and, with a 266 nm pump, solvated electrons. With 292 nm excitation, evidence for relaxation within the 1 La state is observable in the PE spectrum shifting to lower eKE, however, such dynamics is captured only in one of the two peaks that make up the fingerprint of the 1 La state. We speculate that rather than reflecting solvation dynamics, the highest eKE features and their disappearance may be connected with passage via the 𝜋𝜎 * state of a small portion of the overall excited state population. The picosecond disappearance of this high eKE signal (and the overall PE intensity) corresponds to dissociation (to H atoms) over a 𝜋𝜎 * barrier without the obvious production of solvated electrons, at least within the hundreds of picosecond delay window of the current experiment. Alternatively, this pathway provides efficient return to the ground state. Results for a 266 nm pump are consistent also with branching of the population but now on a femtosecond timescale from the initially excited (or rapidly populated) 1 La state. Part of the 1 La state population relaxes and decays on the nanosecond time scale via internal conversion while a smaller portion undergoes ultrafast photoionization within 200 fs. 10 Making a connection to the 𝜋𝜎 * pathway (where there is more evidence in the 292 nm dataset), it is possible that with the higher energy excitation, there is simply no energy barrier on the 𝜋𝜎 * intermediate state and the larger resulting H-atom translational energy results in water ionization to generate solvated electrons. 58, 60 102 In future work, we will follow up by probing for the signature of solvated electrons with electronic transient absorption and their appearance time at 292 nm out to nanosecond time delays. To reinforce the argument for an excitation energy dependent solvated electron generation, solvents that yield higher threshold energies for indole photoionization, (e.g., ethanol) will be tested; we expect features such as the early time rise in the excited state dynamics to be absent. In TRPES, a higher probe energy will be employed in future experiments to avoid the spectrometer cut-off function and fully resolve the 1 La signature and solvated electron signatures. A careful choice of probe/ionization photon energy is also necessary to access the complete manifold of FC window, as the molecular system evolves in the excited state. 40 If the probe photon energy would be insufficient to access the FC window for photoionization then the measured lifetime would be less than the actual value as pointed out by Martinez and co-workers. 61 103 4.7 Postscript to Chapter 4 4.7.1 Cross-Correlation Measurement for Determining Instrument Response Function The cross-correlation between pump and probe was measured by recording non-resonant two photon ionization of 100 mM sodium iodide in water. For both 266 nm and 292 nm, the wavelengths are off-resonant with the iodide charge-transfer-to-solvent (CTTS) band, although 266 nm is very nearly resonant. 62 The vertical detachment energy of iodide is 8.1 eV 63 so ejected electron signal should require absorption of two photons, either both coming from the 292 nm beam, both from the 266 nm beam, or one from each. The pre-resonance at 266 nm will strongly favor the latter two paths; only the last path will depend on the pump-probe delay. Therefore, at each delay point, time of flight (TOF) spectrum was recorded and the time-invariant signal of pump alone and probe alone were subtracted from the total photoelectron signal. In order to extract the cross-correlation width, the integrated signal at each delay point with increment of 5 fs was plotted as a function of pump-probe delay time and was fitted with a Gaussian function (Figure 4.S1). Figure 4.S1: Cross-correlation between pump and probe pulses in the liquid-jet. 104 4.7.2 Comparison between R2PI Spectra and TRPES Spectrum Near Zero Pump-Probe Delay Figure 4.S2 shows the resonant two-photon enhanced photoelectron spectrum ionized 1+1 with one-color 266 nm and one-color 292 nm, respectively, in comparison with resonant two-color 266/292 nm 1+1' PE spectrum obtained at zero pump-probe delay in the TRPES for aqueous indole. Aqueous indole absorbs at the one-photon level at both 266 and 292 nm, and hence the 1+1' spectrum has contributions from both time-orderings. The total photon energies involved are 8.5 eV , 9.4 eV and 8.9 eV , respectively. The 266 nm R2PI spectrum is shifted towards higher eKE because of extra photon energy. The time-resolved spectrum clearly shows a well resolved shoulder in higher eKE and nowhere close in shape with either of R2PI spectrum. Figure 4.S2: Comparison between R2PI spectra with different one-color excitation along with time-resolved spectrum obtained near zero pump-probe delay (resonant 1+1' ionization). The 266 nm beam was horizontally polarized while 292 nm beam was vertically polarized with respect to the laboratory frame. 105 4.7.3 R2PI Measurement of Aqueous Tryptophan The vertical ionization energies (VIE) of aqueous tryptophan corresponding to D0-S0 and D1-S0 using two photon ionizations from 266 nm and 292 nm are 7.5 ± 0.1 eV and 8.4 ±0.1 eV (Table 4.S1) after correction for inelastic scattering, which are not significantly different to the values reported for aqueous indole in this article (7.4 and 8.4 ±0.1 eV). Figure 4.S3: Resonant one-color two-photon ionization spectra of aqueous tryptophan (10 mM) recorded with (a) 266 nm and 292 nm ionization (b) pulses. Individual Gaussian fits are shown in red and green. Note that the R2PI spectra are affected by the instrument cut-off function at eKE < 0.5 eV. From Table 4.S1, it is clear that solvation has had a large impact on the ionization corresponding to D0-S0 while the D1-S0 ionization energy remains relatively unchanged in comparison to the gas phase values. This implies a large change in the D0-D1 gap in aqueous solution. Table 4. S1: Photoionization of aqueous tryptophan. All energies are in eV. Ionized orbital eKE a (266 nm/292 nm) Corrected eKE b (266 nm/292 nm) VIE c (266 nm/292 nm) VIE d (gas phase) HOMO (π3) 1.4 / 0.9 1.8 / 1 7.5 / 7.5 7.9 HOMO-1(π2) 0.8 / - 0.9 / - 8.4 / - 8.3 a eKE - electron kinetic energy b eKE reported after correcting for electron inelastic scattering. c VIE - vertical ionization energy d VIE reported in gas phase 64-66 106 4.7.4 Time-Resolved Photoelectron Spectra of Aqueous Tryptophan The time resolved photoelectron spectra of aqueous tryptophan (Fig 4.S4) was also measured at 292 nm and 266 nm excitations in a similar fashion as described for aqueous indole in this chapter. At 292 nm excitation, we did not observe any change in the spectral shape or decay of PE bands within 0.5 ps unlike aqueous indole where a dip was developing around 1.25 eV. After 1 ps, the whole PE spectra is decaying without any rise of any PE band anywhere in the spectra similar to aqueous indole. Another spectral difference in aqueous tryptophan is not much resolved PE bands unlike aqueous indole although a developing shoulder in higher eKE range can be observed. At 266 nm excitation, the higher eKE spectral band is decaying within half a picosecond but there is no change in the PE band at 0.75 eV. In case of indole we observed that PE bands at 0.75 eV and 1.2 eV was rising while the highest eKE band was decaying which resulted into an isosbestic point. After 1 ps, the whole PE spectra was decaying similar to aqueous indole, but the overall spectral width seems to be less than the aqueous indole. We suspect that the since the photoionization quantum yield is three to four times lower for aqueous tryptophan compared to aqueous indole, the rise of lower eKE band is not that much pronounced here. 107 Figure 4.S4: TRPES spectra of aqueous tryptophan excited with 292 nm and photoionized with 266 nm are shown in panel (a) while panel (b) displays the TRPES spectra when aqueous tryptophan is photoexcited with 266 nm and photoionized with 292 nm. The arrows in the figure displays the decay of the PE bands. Experimental conditions were different for short pump-probe delay (up to 20 ps) and long pump-probe delay (50 -400 ps) since the measurements were performed on different days. 108 4.8 References 1. S. K. Pal; J. Peon; A. H. Zewail, Proc. Natl. Acad. Sci. 2002, 99, 1763-1768. 2. S. Schenkl; F. Van Mourik; G. Van der Zwan; S. Haacke; M. Chergui, Science 2005, 309, 917-920. 3. B. L. Vallee; J. F. Riordan, Annu. Rev. Biochem. 1969, 38, 733-794. 4. K. S. Sarkisyan; I. V. Yampolsky; K. M. Solntsev; S. A. Lukyanov; K. A. Lukyanov; A. S. Mishin, Sci. Rep. 2012, 2, 608. 5. P. R. Callis, [7] 1La and 1Lb transitions of tryptophan: Applications of theory and experimental observations to fluorescence of proteins. In Methods Enzymol., Elsevier: 1997; Vol. 278, pp 113-150. 6. J. Lakowicz, Principles of fluorescence spectroscopy. Kluwer Academic Press, New York 1999, -. 7. E. Eshelman; M. Daly; G. Slater; E. Cloutis, Planet. Space Sci. 2015, 119, 200-207. 8. L. R. Dartnell; T. A. Roberts; G. Moore; J. M. Ward; J.-P. Muller, PLoS One 2013, 8, 75270. 9. F. Scappini; M. Capobianco; F. Casadei; R. Zamboni; P. Giorgianni, Int. J. Astrobiol. 2007, 6, 281-289. 10. J. Peon; G. C. Hess; J.-M. L. Pecourt; T. Yuzawa; B. Kohler, J. Phys. Chem. A 1999, 103, 2460-2466. 11. D. Sharma; J. Léonard; S. Haacke, Chem. Phys. Lett. 2010, 489, 99-102. 12. R. Montero; Á. P. Conde; V. Ovejas; F. Castañ o; A. Longarte, J. Phys. Chem. A 2011, 116, 2698-2703. 13. O. Bram; A. A. Oskouei; A. Tortschanoff; F. van Mourik; M. Madrid; J. Echave; A. Cannizzo; M. Chergui, J. Phys. Chem. A 2010, 114, 9034-9042. 14. E. Gudgin; R. Lopez-Delgado; W. R. Ware, J. Phys. Chem. 1983, 87, 1559-1565. 15. M. R. Eftink; J. Jia; D. Hu; C. A. Ghiron, J. Phys. Chem. 1995, 99, 5713-5723. 16. J. W. Petrich; M. Chang; D. McDonald; G. Fleming, J. Am. Chem. Soc. 1983, 105, 3824- 3832. 17. N. Periasamy; G. Joshi, Nuovo Cimento Soc. Ital. Fis., D 1990, 12, 1691-1695. 18. D. Rayner; A. Szabo, Can. J. Chem. 1978, 56, 743-745. 19. K. Guzow; R. Ganzynkowicz; A. Rzeska; J. Mrozek; M. Szabelski; J. Karolczak; A. Liwo; W. Wiczk, J. Phys. Chem. B 2004, 108, 3879-3889. 20. T. A. Oliver; Y. Zhang; A. Roy; M. N. Ashfold; S. E. Bradforth, J. Phys. Chem. Lett. 2015, 6, 4159-4164. 21. Y. Zhang; T. A. Oliver; M. N. Ashfold; S. E. Bradforth, Faraday Discuss. 2012, 157, 141- 163. 22. O. V. Boyarkin; S. R. Mercier; A. Kamariotis; T. R. Rizzo, J. Am. Chem. Soc. 2006, 128, 2816-2817. 23. M. Born; R. Oppenheimer, Annalen der physik 1927, 389, 457-484. 109 24. J. R. Platt, J. Chem. Phys. 1949, 17, 484-495. 25. J. N. Murrell, 1963. 26. R. Livingstone; O. Schalk; A. E. Boguslavskiy; G. Wu; L. Therese Bergendahl; A. Stolow; M. J. Paterson; D. Townsend, J. Chem. Phys. 2011, 135, 194307. 27. G. M. Roberts; V. G. Stavros, Chem. Sci. 2014, 5, 1698-1722. 28. M. G. Nix; A. L. Devine; B. Cronin; M. N. Ashfold, J. Chem. Phys. 2007, 126, 124312. 29. V. Povedailo; D. Yakovlev, J. Appl. Spectrosc. 2008, 75, 336-340. 30. B. J. Fender; D. M. Sammeth; P. R. Callis, Chem. Phys. Lett. 1995, 239, 31-37. 31. B. Valeur; G. Weber, Photochem. Photobiol. 1977, 25, 441-444. 32. A. L. Sobolewski; W. Domcke, Chem. Phys. Lett. 1999, 315, 293-298. 33. L. Serrano-Andrés; B. O. Roos, J. Am. Chem. Soc. 1996, 118, 185-195. 34. D. Brisker‐Klaiman; A. Dreuw, ChemPhysChem. 2015, 16, 1695-1702. 35. M. R. Eftink; L. A. Selvidge; P. R. Callis; A. A. Rehms, J. Phys. Chem. 1990, 94, 3469- 3479. 36. M. M. Zawadzki; J. O. Thompson; E. A. Burgess; M. J. Paterson; D. Townsend, Phys. Chem. Chem. Phys. 2015, 17, 26659-26669. 37. F. Buchner; H.-H. Ritze; J. Lahl; A. Lübcke, Phys. Chem. Chem. Phys. 2013, 15, 11402- 11408. 38. F. Buchner; B. Heggen; H.-H. Ritze; W. Thiel; A. Lübcke, Phys. Chem. Chem. Phys. 2015, 17, 31978-31987. 39. F. Buchner; A. Nakayama; S. Yamazaki; H.-H. Ritze; A. Lü bcke, J. Am. Chem. Soc. 2015, 137, 2931-2938. 40. H. L. Williams; B. A. Erickson; D. M. Neumark, J. Chem. Phys. 2018, 148, 194303. 41. S. Lochbrunner; T. Wilhelm; J. Piel; S. Spörlein; E. Riedle In Sub-20-fs tunable pulses in the visible and NIR by noncollinear optical parametric amplification (NOPA), Advanced Solid State Lasers, Boston, Massachusetts United States, Optical Society of America: Boston, Massachusetts United States, 1999; p TuA4. 42. E. Riedle; M. Beutter; S. Lochbrunner; J. Piel; S. Schenkl; S. Spörlein; W. Zinth, Appl. Phys. B 2000, 71, 457-465. 43. A. Roy; R. Seidel; G. Kumar; S. E. Bradforth, J. Phys. Chem. B 2018, 209. 44. X. Chen; D. S. Larsen; S. E. Bradforth; I. H. van Stokkum, J. Phys. Chem. A 2011, 115, 3807-3819. 45. T. W. Kee; D. H. Son; P. Kambhampati; P. F. Barbara, J. Phys. Chem. A 2001, 105, 8434- 8439. 46. D. Luckhaus; Y.-i. Yamamoto; T. Suzuki; R. Signorell, Sci. Adv. 2017, 3, e1603224. 47. J. Eland, Int. J. Mass Spectrom. Ion Processes 1969, 2, 471-484. 48. L. Dolby; G. Hanson; T. Koenig, J. Org. Chem. 1976, 41, 3537-3539. 49. L. Domelsmith; L. L. Munchausen; K. Houk, J. Am. Chem. Soc. 1977, 99, 4311-4321. 50. M. Kubota; T. Kobayashi, J. Electron. Spectrosc. Relat. Phenom. 2003, 128, 165-178. 110 51. B. Jagoda-Cwiklik; P. Slavíček; L. Cwiklik; D. Nolting; B. Winter; P. Jungwirth, J. Phys. Chem. A 2008, 112, 3499-3505. 52. A. Bernas; D. Grand; E. Amouyal, J. Phys. Chem. 1980, 84, 1259-1262. 53. T. Bizjak; P. Bisht; S. Lochbrunner; E. Riedle, Femtosecond transient spectroscopy of the photoionization of indole in water. In Femtochemistry and Femtobiology, Elsevier: 2004; pp 229- 232. 54. G. Smith, The Acid-Catalyzed Polymerization of Pyrroles and Indoles. In Adv. Heterocycl. Chem., Elsevier: 1963; Vol. 2, pp 287-309. 55. R. Katoh, J. Photochem. Photobio., A 2007, 189, 211-217. 56. F. Saito; S. Tobita; H. Shizuka, J. Photochem. Photobiol., A 1997, 106, 119-126. 57. K. L. Stevenson; G. A. Papadantonakis; P. R. LeBreton, J. Photochem. Photobiol., A 2000, 133, 159-167. 58. A. L. Sobolewski; W. Domcke, Chem. Phys. Lett. 2000, 329, 130-137. 59. W. G. McGimpsey; H. Görner, Photochem. Photobiol. 1996, 64, 501-509. 60. C. G. Elles; I. A. Shkrob; R. A. Crowell; S. E. Bradforth, J. Chem. Phys. 2007, 126, 164503. 61. H. R. Hudock; B. G. Levine; A. L. Thompson; H. Satzger; D. Townsend; N. Gador; S. Ullrich; A. Stolow; T. J. Martinez, J. Phys. Chem. A 2007, 111, 8500-8508. 62. M. J. Blandamer; M. F. Fox, Chem. Rev. 1970, 70, 59-93. 63. R. Seidel; B. Winter; S. E. Bradforth, Annu. Rev. Phys. Chem. 2016, 67, 283-305. 64. S. Campbell; J. Beauchamp; M. Rempe; D. Lichtenberger, Int. J. Mass Spectrom. Ion Processes 1992, 117, 83-99. 65. P. Cannington; N. S. Ham, J. Electron. Spectrosc. Relat. Phenom. 1983, 32, 139-151. 66. K. Seki; H. Inokuchi, Chem. Phys. Lett. 1979, 65, 158-160. 111 Chapter 5. Unraveling the Excited State Dynamics of Aqueous Indole with Broadband Femtosecond Transient Absorption and Quenching Studies 5.1 Introduction Indole is the absorbing chromophore of the essential amino acid, tryptophan. Tryptophan is considered an important probe to study protein dynamics and associated structural changes since the fluorescence properties (i.e. emission maximum and quantum yield) of its chromophore, indole, is highly sensitive to the local environment. 1-4 Indole is also a building block of the eumelanin, a skin pigment, which is known for its photoprotective role from the ultraviolet part of the solar radiation. 5, 6 Since the photophysical and photochemical behavior of indole has been central to the understanding of important biological functions/pathways, it has drawn considerable attention from researchers, which has triggered the molecular level studies pertaining to its electronic structure and intricate excited state dynamics. A large number of theoretical and gas phase experimental studies are available in the literature with a few in the condensed phase; in most studies, water. The photophysics of indole is primarily governed by the low-lying excited states 1 La, 1 Lb and πσ* states, which fall within 0.5 – 0.7 eV of each other. The 1 La and 1 Lb states, named in accordance with Platt’s nomenclature 7 , are optically bright with a significant oscillator strength having ππ* character, while the 1 πσ* state possesses a low oscillator strength. 8, 9 The adiabatic 1 πσ* state is formed via an avoided crossing between the N (3s) Rydberg state and the valence 1 πσ* state. In the vertical Franck-Condon region this state has extensive 3s Rydberg character but becomes purely of 1 πσ* character upon elongation of the N-H bond; the state can be populated only indirectly via transfer from an optically bright state, unlike 1 La and 1 Lb. The relaxation pathways of even isolated indole are quite complex due to presence of multiple conical intersections (CIs) such as 1 La/ 1 πσ*, 1 Lb/ 1 πσ* and 1 πσ*/S0 along NH 112 coordinate and 1 La/ 1 Lb along another coordinate (close to 1 La origin), which involves ring puckering or ring distortion modes. The repulsive nature of the 1 πσ* at extended N-H coordinate imparts two possible deactivation pathways, which include repopulating the ground state through existing CI at long N-H distance and generation of H-atom through dissociation in the excited state. There have been a wide range of gas phase experiments designed to explore and understand the H-atom photodissociation pathway on photoexcitation of indole. These have been thoroughly reviewed in the literature. Lin et al. using photofragment translation spectroscopy, found evidence of two dissociative mechanisms; one which occurs on a fast timescale and one on a slower timescale. 10 The faster process is attributed to direct dissociation upon photoexcitation to an electronically excited state while the slower process occurs from ground electronic state surface. They reported that more than 80% of indole dissociation occurred from the excited state upon excitation at 248 nm. Using H-atom Rydberg time-of-flight spectroscopy at multiple longer pump wavelengths, Ashfold and coworkers found a threshold for fast H-atom fragmentation 4.7 eV (at 263 nm). 11 They considered the dynamics revealed in their experiment to be consistent with the molecules on the excited ππ* state, having sufficient energy to access a conical intersection with the dissociative 1 πσ* surface. Ullrich and her coworkers used femtosecond resolution of the H- atom kinetic energy release and time-resolved photoelectron spectroscopy in combination with the time-resolved ion yield measurements of Longarte and coworkers. 12 They came up with a picture more consistent with the experiment done by the Ashfold group, where direct dissociation via the 1 πσ* channel is not operative below 4.7 eV excitation energy. 12, 13 Townsend's group have applied time-resolved photoelectron spectroscopy of indole in a molecular beam to establish the electronic signature of the 1 πσ* in the excited state photoelectron spectrum and through the photoelectron angular distribution. 8 Their work suggests the involvement of the 1 πσ* state at photolysis 113 wavelengths longer that 263 nm and only a slight slowing of the passage time of the system over a small barrier along the H-atom dissociation coordinate on the 1 πσ* surface as the wavelength is increased to 273 nm. The complexity of dynamics in the excited state increases when indole is dissolved in solvents due to the change in topography of potential energy surfaces. This leads to changes in the position of CIs, because the relative energy separation of these close lying excited states gets affected by the dielectric constant of the solvent. Several computational and experimental studies confirm the inversion of energy ordering of 1 La and 1 Lb in water in contrast to the gas phase. 14-16 This argument is based upon the large difference in the static dipole moments of 1 La and 1 Lb where 1 La gets stabilized in the polar medium due to enhanced dipole-dipole interaction making 1 La the fluorescent state. The fluorescence of aqueous indole shows a mono-exponential decay of 4.1 ns. 17 The parent molecule of indole, tryptophan, shows a biexponential decay with lifetimes of 3.14 ns and 0.51 ns resulting from the existence of multiple conformers at room temperature. 18 Kohler and coworkers studied the near-threshold photoionization dynamics of aqueous indole by using transient absorption spectroscopy. 19 After exciting the indole solution with 260 nm pump pulses, they observed the formation of solvated electrons within their instrument response (200 fs). The possible mechanism for the generation of solvated electron is electron transfer where the electron is directly transferred to a trapping site in water without formation of an intermediate delocalized electron in the conduction band. Since the separation between the geminate pair is expected to be small near the photoionization threshold, 20 it is expected that there would be a significant geminate recombination yield However, they claim that there is no geminate recombination between the solvated electron and the indole radical cation in their experimental measurement, which scans up to 600 ps time delay. Based on this study, no conclusion was reached whether the electron and 114 indole radical cation remain as a tight ion pair for a significant amount of time or if they quickly diffuse away from one another. 19 Sobolewski et. al. performed ab initio calculations on indole-water cluster to explore the influence of micro solvation on the gas phase 1 πσ* indole dynamics. 21, 22 Their calculations reveal that the adiabatic excitation energy of the πσ* state converges towards the experimental threshold of photoionization (4.35 eV) in water and they claim this state to be the precursor for the hydrated electron. In indole-water clusters, the state evolves into charge transfer to solvent (CTTS) state in which solvated electron is hydrogen bonded to the indole radical cation. To quantify the yield of solvated electron as a function of excitation energy, Bernas et.al. measured the efficiency of the photoionization process with increasing photon energy through photoconductivity measurement and concluded that the quantum yield of photoionization increases with photon energy with a photoionization threshold of 4.35 eV (285 nm). 23 Katoh et. al. performed nanosecond transient absorption measurements on aqueous indole to explore the wavelength dependence of photoionization quantum yield. 24 They reported that the quantum yield for photoionization is independent of excitation wavelength in the range of 250 nm – 300 nm with a constant value of 0.2. The quantum yield increases below 250 nm excitation and reaches 0.26 at 220 nm. The conclusions of this study are inconsistent with the results obtained by Bernas et al. 23 Another possible species encountered during the study of indole photophysics in the condensed phase is “triplet” state of indole. There are several studies available in the literature which report the lifetime of the triplet state of aqueous indole and the possible pathways of its generation by implementing different techniques such as flash photolysis and sensitive photon counting phosphorescence measurements. The triplet state lifetime varies from μs to ms and the two of the most contrasting hypotheses with respect to the lifetime were presented by Fischer et 115 al. and Strambini et al.. 25-27 Fischer et al. proposed that the short lifetime (12- 40 μs) of triplet state is due to the normal intersystem crossing with a smaller quantum yield while the long lived “ms” lifetime observed by Strambini et al. was presumably arising from the geminate recombination of the long-lived species, indole radical cation and the solvated electron after spin dephasing, with a two- to three-times larger quantum yield. 26, 27 In a later study, Strambini et al. postulated that the intrinsic lifetime of triplet is ms long and the shorter lifetime is coming from the diffusional quenching of the photoproducts or the possible quenching introduced by the impurities present in the water or the contaminated sample. 28 In this work, we are investigating the influence of solvation on the photophysics of indole via ultrafast transient absorption spectroscopy in a more detailed way. In general, the difficulty has been in assigning the transient absorption spectra as absorption bands attributed with indole radical cations, solvated electrons, indolyl radical, triplets, and the parent excited state absorption all overlap. To gain better insight, we have implemented time resolved femtosecond pump- broadband probe spectroscopy with sub-100 fs resolution. Excitation was performed at several pump wavelengths and probed with a broad-band white light continuum ranging from near-UV to visible wavelengths. Various excited state scavengers were used to help with spectral assignment. Herein transient absorption is used to gain insight into the effects of the solvent environment and change in excitation energy on the excited state landscape with respect to transfer and decay of excited state population to different pathways, change in the position of conical intersection (CI) and generation of possible charged species. 116 5.2 Experimental Indole (99%; Sigma) was purchased and used without further purification. The steady state UV absorption spectrum of indole (50 μM) in water and ethanol was recorded using the Cary 50 UV-Visible spectrophotometer while the steady state fluorescence measurements were carried out by fluorimeter (JOBIN YVON FLUOROMAX 3) in a 1 mm path length cuvette. The transient absorption experiments were performed by exciting the indole solution with UV pulses (λpump = 200, 267 and 292 nm) and probed using a white light super-continuum probe. To generate 200 nm pump, the fundamental wavelength (800 nm) was gently focused by a f = 2 m lens into a 500 μm type-I β-barium borate (BBO) crystal (Red Optronics) to generate 400 nm. The resulting 400 nm laser pulse was used to generate 267 nm by sum frequency mixing of 400 nm with a part of the residual 800 nm in a 150 μm type-II BBO crystal (Red Optronics). The 267 nm had a bandwidth of 3-4 nm with a maximum pulse energy of 8-9 μJ and was used again to generate 200 nm (1 μJ) via sum frequency generation with the residual 800 nm (60 μJ) in a 75 μm thick type-I BBO crystal (Red Optronics), which was placed at the focus of the 267 nm and the 800 nm. The deconvoluted temporal width of the 200 nm was around 240 fs, as determined by the cross- correlation with the continuum in ethanol solution. The other pump wavelengths 267 nm and 292 nm with a typical bandwidth of 4.5-5 nm were generated by doubling (using a 150 μm type-I BBO crystal (Red Optronics)) the visible output from a homemade Non-Collinear Optical Parametric Amplifier (NOPA) following the design of Riedle et. al. 29 The NOPA is pumped with the output of a 35-fs Ti:Sapphire amplifier system (Coherent Legend Elite USP, 1 kHz repetition rate). To achieve the optimal temporal resolution in our experimental set-up, the visible output and the second harmonic of the NOPA 117 were compressed using prism-pair compressors, with prisms of fused silica and calcium fluoride respectively. The deconvoluted temporal width of the 267 nm was determined by the cross- correlation with the continuum in ethanol solution and is ~70 fs. The probe continuum broadband, ranging from 320 nm to 700 nm, was produced by focusing a small fraction of 800 nm onto a rotating calcium fluoride window (2 mm thick, Koch Crystal Finishing). A pair of aluminum-coated off-axis parabolic mirrors (Janos Technology) was used to first collimate the beam and then focus on the sample. The relative polarization between the pump and probe pulses was controlled by rotating the 800 nm polarization prior to continuum generation with an air-spaced zero-order half waveplate (Karl Lambrecht Corporation). All the experiments reported here were performed at a magic angle between the pump and probe pulses. In all the experiments, the aqueous indole (10-20 mM) was flowed through a recirculating wire-guided gravity jet which produced a thin film of the liquid with thickness of 115 μm. 30 The exact path length of the sample varies, depending upon the liquid property such as density, viscosity, and the distance of the interaction region from the nozzle opening. A variety of advantages are realized when using liquid film to carry out experiments: (a) to minimize the group velocity walk-off between the deep-UV pump and the broadband continuum, (b) avoid the contamination of the TA signal from non-resonant coherent signals (such as two-photon absorption) from the quartz/glass (of a cuvette) and (c) to ensure a new sample volume is presented at the laser interaction region for every laser shot, which ensures no signal is recorded that corresponds to photo-degraded sample. The dispersion in the TA signal (chirp correction) was corrected by measuring the signal from the pure solvent under the same experimental conditions and adjusted during data workup by interpolation to a third order polynomial. 118 5.3 Results and Discussions Our main goal here is to investigate the excited state dynamics of indole by changing the excitation energy. These energies span the region of low-lying excited states such as 1 La, 1 Lb and 1 πσ* states as well as higher lying Ba and Bb states. The relaxation dynamics following excitations at 200, 267 and 292 nm are explored in water and ethanol using transient absorption, ensuring the access above and below the 1 πσ* state. Due to differing strengths of the permanent dipole moments in each electronic state, the solvent will have a differential effect on the relative stabilization of these states compared to isolated gas phase indole. The experimental dipole moments for the ground state, 1 Lb, and 1 La are 2.09, 2.3 and 5.4 Debye in the gas phase 31-33 respectively while the computed gas phase dipole moments for 1 πσ*, 1 Ba, 1 Bb are 12.4, 1.8 and 3.8 Debye. 14, 21 Based on the strength of dipole moments, we expect that the 1 La, 1 Bb and 1 πσ* state will be most influenced by the presence of water due to enhanced dipole-dipole interaction. 14 Before we start discussing the photoionization and temporal evolution of the excited states, it would be a good idea to look at the effect of the excitation energy on the fluorescence emission in both water and ethanol, the solvents used in these experiments. 5.3.1 Steady State Fluorescence Spectroscopy The steady state excitation and emission spectra of indole were measured in water and ethanol (Figure 5.1). The unstructured fluorescence spectra at the excitation wavelengths of 266 nm and 292 nm are both attributed to emission from the 1 La state, the fluorescent state in polar solvents as oppose to the gas phase or nonpolar solvents where fluorescence is observed from the 1 Lb state. The 1 La state fluorescence shows large solvatochromism on changing from ethanol to water due to stabilization effects from the solvent, as the permanent dipole of 1 La is relatively large. 119 Since water is more polar than ethanol, the Stokes shift is greater in water by 25-30 nm compared to ethanol (Fig. 5.1(c), 5.1(d))). This observation is consistent with the reported literature values explaining the state reversal of 1 La and 1 Lb which occurs in polar solvents. 14, 21 The other noteworthy point from the steady state measurement is identical fluorescence emission (peaks around 350 nm), e.g. Kasha-type fluorescence for 250-300 nm excitation signifying the same fluorescent state irrespective of the excitation energy (Fig. 5.1(a), 5.1(b)). Figure 5.1: Steady state fluorescence spectra of indole in (a) water and in (b) ethanol, Excitation and Emission spectra for indole in (c) water and in (d) ethanol. The different Stokes shift in the two solvents is illustrated by the red dashed lines. The pump wavelengths used for time resolved experiments are marked by the black dotted line in 2D plot. 120 5.3.2 Time-Resolved Spectroscopy of Indole in Ethanol The ultrafast relaxation dynamics of indole were measured in ethanol and water at different excitation energies using transient absorption. In Figure 5.2, we show a two-dimensional representation of the transient absorption spectra of indole in ethanol as probed with broadband continuum and measured at three different pump wavelengths. We notice immediately that there is a strong signal at the blue edge at all three pump wavelengths. The ionization threshold of indole in ethanol is 4.85 eV (256 nm), 23 which means that the indole cannot be ionized at 267 and 292 nm excitations, but that there might be ionization at 200 nm. In the 200 nm pump wavelength spectrum, there is an intense feature in the range of 350-450 nm with a maximum around 360 nm which shifts towards the blue with increasing pump-probe delay and a weak feature is present throughout the probe window without any time evolution, a region dominated by the solvated electron signal. 34 Therefore, we can say that the signal is mainly coming from the excited state absorption. Now, the other point which is apparent in all these spectra is the wavelength dependent shifting and narrowing of the excited state absorption feature (ESA). This shift is largest at the 200 nm pump (maximum shifting by 30 nm) followed by the 266 nm (10 nm shift) and 292 nm (8 nm shift). The reason for the spectral shift cannot be the population transfer between the 1 Lb to 1 La because the time scale of the internal conversion should be faster than the cross correlation or instrument response of this experiment (<100 fs). 12, 35 We propose that this shifting/narrowing is mainly occurring due to vibrational cooling on the excited state surface. In a purely Kasha-type picture, one can expect the largest shift at the 200 nm pump because of large internal energy followed by the 266 nm and 292 nm pump, the latter near the S1 band origin. 121 Figure 5.2: Transient absorption spectra of indole in ethanol measured with excitation wavelengths of (a) 292 nm, (b) 266 nm and (c) 200 nm, (d) Comparison of experimental spectrum at 292 nm pump (delay = 1 ps) to the computed S1 ESA spectrum Note: In panel (d), the y-axis for experimental spectrum (wine color) is in absorbance (mOD) while for computed spectrum the y-axis is in TDM 2 In order to confirm the spectral assignment, the ESA spectra for S1-Sn is computed (sticks in Figure 5.2(d)) and compared with the experimental transient absorption spectrum (at 1 ps delay) recorded with 292 nm excitation wavelength (Figure 5.2(d)). The ESA spectra for S2-Sn is also computed and look identical to S1-Sn (not shown here). The measured spectrum matches well with the computed S1 ESA and hence we can conclude that there is negligible formation of solvated electron in ethanol. 122 5.3.3 Time-Resolved Spectroscopy of Indole in Water In water, the ionization threshold of indole is 4.35 eV (287 nm), 23 which suggests that photoionization should be an additional channel of relaxation at the excitation wavelengths of 266 and 200 nm but not at 292 nm. Since, 200 nm excitation is well above the ionization threshold, in analogy to the photochemistry of aqueous phenol, it is expected that most of the excited state population will prefer the photoionization channel as a relaxation pathway. 36 5.3.3.1 Excitation at 200 nm Figure 5.3 displays the pump induced transient absorption spectra of indole in water at 200 nm excitation. The contour plot in Figure 5.3(a) depicts the full transient absorption dataset up to 10 ps. The time and spectral slices represent the feature of transient species and their kinetics in a better way and hence the data has been plotted at a series of time delays, from 0.25 ps to 500 ps. After inspecting the transient absorption spectra, two key features stand out. (i)The most intense band is in the region of 500-650 nm followed by the wavelength region of <375 nm (Fig 5.3 (a)) (ii) After 1 ps, a shoulder in the range of 400-475 nm (peak maximum around 440 nm) with intermediate intensity is appearing within the instrument response time and the blue edge of the spectrum shifts even further by ⁓10 nm (the minimum shifts from 394nm to 383 nm) which appears after 1 ps (Fig 5.3 (b)). After 2 ps, there is minimal narrowing or shifting of the blue edge (Fig. 5.3(c)). 123 Figure 5.3: (a) Contour plot of the full 2D transient absorption data set of 16 mM indole in water at 200 nm excitation (b) Spectral slices at a series of early time delays and (c) later time delays (d) experimentally measured spectrum of indole cation, radical and the solvated electron; solvated electron spectrum is scaled by ¼. The black line in fig (b) represents the solvent only signal. The black arrows display the rise of the signal in panel (b) and decay of the signal in panel (c). Since the excitation energy at 200 nm is well above the ionization threshold (4.35 eV), the ionization channel will be a dominant pathway of relaxation for the excited state of aqueous indole. This results in, a significant amount of solvated electron signal. Each ejected electron will be accompanied by its geminate partner; indole cation. McGimpsey et al. measured the ionization quantum yield of aqueous indole at 193 nm excitation with a pulse duration of 20 ns in the range of 0.3-0.4 from Ar-saturated aqueous indole by laser photolysis using optical and conductometric 124 detection methods. 37 We suspect that since the pulse width in the study performed by McGimpsey et. al. was 20 ns, they would not be able to capture any recombination reaction happening just after photoionization. At 200 nm, it’s possible that the instantaneous photoionization quantum yield is close to unity and the TA spectrum will arise from a time dependent mixture of cation and solvated electron in addition to other photoproducts which may appear due to the recombination between the geminate pair. The recombination between the geminate pair can produce ground state indole, indole anion or triplet state indole as photoproducts. The most intense broad band in the red part of the spectrum indicates the presence of solvated electron, which peaks around 720 nm along with the radical cation which peaks around 580 nm. The extinction coefficient of the indole cation is 3000M -1 cm -1 at 580 nm, while the maximum extinction coefficient of solvated electron is 19000 M -1 cm -1 at 720 nm as shown in Fig 5.3 (d). 38, 39 Based on the molar extinction coefficient, we can say that the cation band will be masked by the solvated electron signal in the recorded spectrum. Even at 580 nm, the ratio of extinction coefficient of cation versus solvated electron is 1:4. To confirm the presence of the solvated electron, an electron quenching experiment was performed. Hydrochloric acid was added to the aqueous indole solution. The generated solvated electron is diffusively quenched by the H + . 36 The concentration (0.2 M) of HCl has been precisely chosen such that acid catalyzed polymerization of ground state indole can be avoided. 40 In the acidic solution, the solvated electrons are quenched with appropriate kinetics leaving the indole cations which possess a μs long lifetime. The rising feature of photoionized products can be seen within 2 ps in the Figure 5.4 (a) while a strong decay of the intensity can be seen in the red part of the spectrum in the Figure 5.4 (b), a region dominated by the solvated electron signal with a partial contribution from the cationic signal. If we compare the 633 nm spectral slice from indole solution and acidic indole solution as 125 shown in the Figure 5.4(c), we can clearly see a significant decay at 633 nm in acidic solution compared to aqueous indole. This observation confirms the quenching of solvated electron by H + . The presence of solvated electrons in our data indicates that the cation is also contributing to the TA signal, which is the immediate partner of the photoionized electron. Figure 5.4: Spectral slices for the aqueous indole (10 mM) when 0.2 M HCl is added in (a) early time and (b) at later time, (c) Temporal profile recorded at 633 nm for aqueous indole with and without HCl added To confirm the presence of the cation band, our experimental spectrum at long time delay (500 ps) was compared with the cation feature measured by Steenken group using the pulse radiolysis technique in aqueous solutions at 20 o C. 38 The cationic feature reported by the Steenken group shows two intense regions which appear around 330 nm and 580 nm (Figure 5.3(d)) in close agreement with our spectrum, shown in Figure 5.4(b). 38 To reinforce the cationic signature in our measurements, the spectra corresponding to the indole cation and the indolyl radical were also computed by CASPT2 level of theory and displayed in Figure 5.5. The computed spectra for both species are in good agreement with the experimental spectra measured by Steenken and coworkers, reinforcing our assignments. 38 126 Figure 5.5: Computed spectrum of indole cation and radical in the gas phase at CASPT2 level of theory. As described above, the cation band at 580 nm can be masked by the solvated electron signal but the signal at 380 nm will have a much smaller contribution from the solvated electron. The blue region could be dominated by the signal coming from ESA (if photoionization yield is not unity) and the cationic signal. If there is signal from ESA, then the band in analogy to our observation in ethanol (Fig 5.2) would be expected to further shift/narrow due to vibrational cooling in the excited electronic state within 1-2 ps. The absence of shifting in the blue region indicates that the signal is mainly coming from the absorbance of the cationic band (Figure 5.3) and not from ESA. After 1 ps, a new band is appearing in the spectral region of 400 – 450 nm and the bands in the blue and red region started going down with the similar time constant. Based on this observation, we feel comfortable to say that there is a small to negligible contribution from the ESA in our data measured with 200 nm pump and hence the photoionization is the dominant channel of relaxation here. 127 Assuming that at 200 nm the quantum yield of ionization is close to unity at 200 nm excitation, we attempted to construct the experimental spectrum simply by taking the sum of the experimentally measured spectrum of indole cation reported by Steenken’s group and reported spectrum of the solvated electron. 38 We observe in Fig 5.3 that the band in the red side of the spectrum is rising up to ⁓ 1.5 ps, which is in agreement with the rise of the solvated electron. Within the same time scale, we also observed a sudden shift of 10 nm towards further blue but not a gradual shift towards the blue which generally indicates the vibrational cooling of the ESA as observed in ethanol spectrum. The absence of vibrational cooling demonstrates the absence of ESA. Therefore, we treated the solvated electron spectrum and the indole cation spectrum as basis spectra to construct the resultant spectrum analogous to the experimental one. By comparing the constructed and experimentally measured spectrum (Fig 5.6), it can be noted that a significant intensity is missing in the range of 400-450 nm in the constructed spectrum. To address the missing intensity, the spectrum corresponding to indolyl radical as reported by the Steenken group was also taken into account but still the experimental spectrum could not be qualitatively reproduced. 38 Therefore, there might be another transient species coming from a process besides photoionization. McGimpsy et al. has reported the photoionization channel as a dominant relaxation pathway in aqueous indole and also indicated the presence of the triplet band in their measurement after 5 μs. 37 In another study, Bent and Hayon reported the triplet-triplet absorption spectra at ⁓ 440 nm with 265 nm excitation using the technique of kinetic absorption spectrophotometry. 41 In our data, we speculate that there might be a triplet band even at 200 nm excitation in the wavelength region of 400-450 nm where we could not account for the missing intensity. Most of the ultrafast studies have been performed by implementing the 266 nm excitation which mainly populates the 1 La state. 128 Figure 5.6: Comparison between the experimentally measured spectrum of 16 mM indole in water at 200 nm excitation with a constructed spectrum by adding the reported spectra of solvated electron and the indole cation. The ratio of solvated electron and cation is kept same while constructing the spectrum, and a scaling factor is used to scale the spectra of solvated electron and cation before summing them to produce the constructed spectrum (Cation + e - aq). With higher excitation energy at 200 nm, the upper excited electronic states ( 1 Bb, 1 Ba) will be accessed. At such a large photon energy, the decay processes associated with 1 S1 state population, such as fluorescence and intersystem crossing, become less efficient. The ultrafast formation of the triplet band in our data can occur by the geminate recombination of the solvated electron and the cation. It’s possible that some population of the solvated electron and the cation remain associated in a close proximity and react to form the triplet after 1 ps, while the rest of the population gets separated by the diffusion. Fischer et al. proposed that the recombination of the solvated electron and cation radical is the dominant pathway to produce triplet and estimated to be two to three times larger than the direct intersystem crossing. 26 An ultrafast transient absorption 129 study by the Haacke’s group was performed on aqueous tryptophan at the pump wavelength of 266 nm. They reported a photoproduct at 415 nm which possess the triplet signature. 42 5.3.3.2 Excitation at 266 nm At 266 nm excitation, the ratio of excited state population of the 1 La and 1 Lb is approximately 3:1 as judged by the ratio of absorption strength ( 1 La / 1 Lb) determined by polarized fluorescence excitation spectrum of indole in propylene glycol (a hydrogen-bonding solvent). 43 A contour plot showing the transient absorption spectrum at 266 nm excitation as a function of pump- probe delay is displayed in Figure 5.7(a). It is apparent from the spectrum that there are three distinct regions based on the signal intensity. The blue and red part of the spectrum exhibit highest intensity while the wavelength region in the middle (410-510 nm) shows a plateau but at intermediate absorbance. A more revealing picture of the spectral dynamics can be seen in the Figure 5.7(b) as a set of time slices recorded at different time delays. In Figure 5.7(b), the maximum of the electronic band in the blue edge shifts towards shorter wavelength by 20 nm over the pump-probe delay of 0.2 ps -1.5 ps unlike the spectrum recorded at 200 nm pump. This time scale matches with the characteristic time scale of the vibrational cooling of the ESA seen in the ethanol data where the extra vibrational energy is transferred into the modes of the surrounding solvent. However, the extent of the shift is greater in H2O than EtOH. A slight decay in the intensity is observed throughout the spectral region at longer pump-probe delay. With 266 nm excitation, there is a photo branching of the excited state population into different relaxation pathways on a femtosecond timescale. One part of the population decays via internal conversion while the other undergoes photoionization within 200 fs. 44 130 To confirm the presence of solvated electron at 266 nm excitation, we again performed scavenging experiments. A strong acid HCl and nitrate salt KNO3 were chosen as electron scavengers. The H + quenches the solvated electrons diffusively while NO3 - quenches statically as well as diffusively. The nitrate acts as static quencher by quenching the excited state directly if it has a spatially extended electronic wavefunction. Figure 5.7: (a) Contour plot of the full 2D transient absorption data set of 17 mM indole in water at 266 nm (b) Spectral slices at a series of time delays (c) Contour plot of the TA data set of 10 mM Indole in water when 0.2 M HCl is added to the solution at 266 nm (d) Spectral slices for the aqueous indole when 0.2 M HCl is added to the solution 131 Figure 5.7 (c) represents the transient absorption data of acidic indole in a contour form. It is visible from the contour plot that the red part of the spectrum is decaying but there are two new features appearing at 367 nm and 400 nm after 80 ps. In Figure 5.7(d), a set of time slices extracted from the TA dataset measured in the acidic indole is presented. It is observed that the intensity in the red part of the spectrum is dramatically reduced confirming the presence of the solvated electrons. At longer pump probe delays (e.g 780 ps), the visible region in the spectrum resembles the spectrum feature of the cation. A band peaking at 580 nm is clearly visible at 780 ps pump probe delay in Figure 5.7(d). The 580 nm band resembles the spectral feature recorded with 200 nm pump wavelength in acidic indole and the computed spectrum. (needs to merge). In the acidic indole data, two new electronic bands centered at 360 nm and 400 nm are quite prominent and obscure the cationic band which comes around 330-340 nm. These two bands are only visible when HCl is added to the aqueous indole solution. The equilibrium of the reaction between H + and e - lies towards the side of neutral hydrogen atom. Therefore, it’s possible that the hydrogen atom can react with indole cation and form an adduct. To check our supposition, calculations were performed at the DFT level of theory which supports the argument that the hydrogen atom can indeed react with cation and the resultant adduct would be more stable than the infinitely separated indole cations and H atoms. Due to the adduct formation, we are also observing a significant decrease in the cationic signal intensity. Since the electron ejection radius is much larger at the 200 nm compared to the 266 nm pump wavelength, the electron will be ejected at relatively larger distance from its geminate partner cation with 200 nm excitation. Therefore, in the 200 nm pumped experiment the hydrogen atoms did not encounter the indole cation diffusively within the timescale of the experiment. 132 5.3.3.3 Excitation at 292 nm We also performed experiments with 292 nm pump wavelength. The two singlet excited states ( 1 La and 1 Lb) are approximately equally populated at this excitation wavelength. 43 Figure 5.8: (a) Contour plot of the full 2D transient absorption data set of 17 mM indole in water at 292 nm excitation (b) Spectral slices at a series of time delays (c) Contour plot of the TA data set of 10 mM Indole in water when 0.2 M HCl is added to the solution at 292 nm (d) Spectral slices for the aqueous indole when 0.2 M HCl is added Figure 5.8 shows the same information as Figure 5.7 but here the measurements have been performed at 292 nm pump wavelength. Bernas et al. estimated the ionization threshold of aqueous indole to be 285 nm (4.35 eV) which is higher in energy than the photon energy of 292 nm (4.24 133 eV). 23 Therefore, the photoionization channel should be expected to be inactive at 292 nm excitation. However, the spectra recorded with 292 nm pump looks qualitatively identical to the spectra recorded with 266 nm pump. As we already confirmed by quenching that the solvated electron is generated at 266 nm excitation, 19, 44, 45 based on the resemblance in the spectra and the quenching studies performed at these two pump wavelengths, it can be inferred that the solvated electron is also being formed at 292 nm pump wavelength contradicting the ionization threshold value reported by the Bernas et al. 23 To confirm the presence of the solvated electron at 292 nm excitation, the scavenging experiments were performed. It can be seen in the Figure 5.7(d) that the addition of HCl in the indole solution resulted into the decrement in the intensity in the red part of the spectrum. This observation confirms the presence of the solvated electron. The outcome of our measurements supports the result obtained by the Katoh et al. 24 He reported that the photoionization quantum yield is same in the excitation wavelength range of 250-290 nm. The main difference between the data recorded at 266 nm and 292 nm pump wavelengths is observed in the quenching experiments (Figure 5.7(d) and Figure 5.8(d). If we compare the 780 ps time slice in Figure 5.7(d) and Figure 5.8(d), it looks like the band associated with the cation at 580 nm has completely disappeared at 292 nm pump unlike 200 nm and 266 nm. This situation is possible since the electron ejection distance will be shorter at 292 nm pump compared to 266 nm pump because of lower photon energy. Hence, we can say that at 292 nm pump, all the solvated electrons will react with H + forming H atoms, which will subsequently react with the indole cation to form the adduct. At 266 nm, the electron ejection distance would more likely be greater and therefore a longer diffusion time is required to remove all the indole cations. Hence, a small population of cation would still be present at 266 nm excitation but not at 292 nm at the same pump-probe delay. 134 Now coming back to the triplet band at 440 nm, it is clear from the spectrum that this band is more prominent at 266 nm and 292 nm excitations compared to the 200 nm experiments. At 266 nm and 292 nm excitation, the triplet band is visible within 250 fs. This ultrafast formation of the triplet is possible when the cation and the solvated electron are in close contact known as the contact pair and the recombination of these two species produces the triplet. 46, 47 Although one can argue that the recombination within the contact pair can also form singlet, but it is not as energetically favorable since the energy of the excited singlet state is higher than the contact pair. We propose that the triplet and the contact pair are very close in energy and therefore they establish a dynamic equilibrium. The H + quenches the presolvated electron within the contact pair resulting into a decrease in the population of contact pairs. In this situation, the equilibrium between the triplet and the contact pair would shift towards the side of the contact pair and hence, the triplet population will decrease. Th spectrum of the contact pair should look like the solvated electron spectrum. Based on the experimental observation, we can say that very few or almost negligible amount of trapped and thermally equilibrated solvated electrons are being generated. This will result into no geminate recombination between the well separated indole cation and the equilibrated solvated electron supporting the argument given by Kohler and his coworkers. 19, 44 At 200 nm excitation, some of the contact pairs recombine to form the triplet while most of the contact pair population form into well separated indole cations and the trapped solvated electrons. The H + here is quenching both the solvated and the presolvated electrons so therefore, we observed no formation of triplet band at early time and also, we observe decay of the red part of the spectrum which is a characteristic of the solvated electron. In chapter 4, we suggested that there is no photoionization at 292 nm excitation based on our TRPES spectra, which does not support the result obtained from TA spectra here. Bent and 135 Hayon demonstrated that the yield of the photoionization quantum yield for aqueous tryptophan and aqueous indole is temperature dependent. 41 The temperature of the liquid micro-jet in TRPES experiments decreases as the solution flows down from the jet-nozzle due to evaporative cooling. Evaporating molecules take the heat of vaporization away from the liquid jet molecules present in the core. This transfer of heat results into lowering of temperature of the liquid microjet. 48 During TRPES experiments, we probe the molecules in the liquid-jet 1-2 mm down from the orifice of the nozzle where a significant lowering of the temperature is expected. This might explain the low or negligible yield of solvated electrons in the TRPES experiments of aqueous indole at 292 nm excitation. 5.4 Conclusion Using broadband transient absorption technique at different excitation energy and different solvents along with computations, the spectral shape of each transient species has been successfully reported. Although photoionization channel is active for aqueous indole at all three pump wavelengths, it is the major relaxational channel at 200 nm. There is a photobranching ratio of excited state population between ionization and internal conversion. We also confirmed the ultrafast formation of triplet band due to recombination between the cation and the presolvated electron within the contact pair. Our study confirms that there is no geminate recombination at 292 nm and 266 nm excitations. At 200 nm, one part of the population of the contact pairs recombine to form a triplet band while the other results into well separated charged products. We also established the photoionization relaxation pathways at 292 nm in accordance with the results obtained by Katoh. 136 5.5 References 1. P. R. Callis; B. K. Burgess, J. Phys. Chem. B 1997, 101, 9429-9432. 2. P. R. Callis; T. Liu, J. Phys. Chem. B: 2004; Vol. 108, pp 4248-4259. 3. J. T. Vivian; P. R. Callis, Biophys. J. 2001, 80, 2093-2109. 4. D. Creed, Photochem. Photobiol. 1984, 39, 537-562. 5. P. Meredith; T. Sarna, Pigment cell res. 2006, 19, 572-594. 6. A. Huijser; A. Pezzella; V. Sundström, Phys. Chem. Chem. Phys. 2011, 13, 9119-9127. 7. J. R. Platt, J. Chem. Phys. 1949, 17, 484-495. 8. R. Livingstone; O. Schalk; A. E. Boguslavskiy; G. Wu; L. Therese Bergendahl; A. Stolow; M. J. Paterson; D. Townsend, J. Chem. Phys. 2011, 135, 194307. 9. A. Iqbal; V. G. Stavros, J. Phys. Chem. A 2009, 114, 68-72. 10. M.-F. Lin; C.-M. Tseng; Y. T. Lee; C.-K. Ni, J. Chem. Phys. 2005, 123, 124303. 11. M. Nix; A. Devine; B. Cronin; M. Ashfold, Phys. Chem. Chem. Phys. 2006, 8, 2610-2618. 12. R. Montero; Á. P. Conde; V. Ovejas; F. Castañ o; A. Longarte, J. Phys. Chem. A 2011, 116, 2698-2703. 13. T. Godfrey; H. Yu; M. S. Biddle; S. Ullrich, Phys. Chem. Chem. Phys. 2015, 17, 25197- 25209. 14. L. Serrano-Andrés; B. O. Roos, J. Am. Chem. Soc. 1996, 118, 185-195. 15. A. L. Sobolewski; W. Domcke, Chem. Phys. Lett. 1999, 315, 293. 16. D. Brisker‐Klaiman; A. Dreuw, ChemPhysChem. 2015, 16, 1695-1702. 17. M. S. Walker; T. W. Bednar; R. Lumry III, Benjamin: New York: 1969. 18. D. Rayner; A. Szabo, Can. J. Chem. 1978, 56, 743-745. 19. J. Peon; J. D. Hoerner; B. Kohler, Liquid Dynamics: Experiment, Simulation, and Theory 2002, 820, 122-135. 20. C. G. Elles; A. E. Jailaubekov; R. A. Crowell; S. E. Bradforth, J. Chem. Phys. 2006, 125, 044515 (1-12). 21. A. L. Sobolewski; W. Domcke, Chem. Phys. Lett. 2000, 329, 130-137. 22. A. L. Sobolewski; W. Domcke, Chem. Phys. Lett. 2000, 321, 479-484. 23. A. Bernas; D. Grand; E. Amouyal, J. Phys. Chem. 1980, 84, 1259-1262. 24. R. Katoh, J. Photochem. Photobio. A: Chem. 2007, 189, 211-217. 25. D. Bent; E. Hayon, J. Am. Chem. Soc. 1975, 97, 2612-2619. 26. C. J. Fischer; A. Gafni; D. G. Steel; J. A. Schauerte, J. Am. Chem. Soc. 2002, 124, 10359- 10366. 27. G. B. Strambini; M. Gonnelli, J. Am. Chem. Soc. 1995, 117, 7646-7651. 28. G. B. Strambini; B. A. Kerwin; B. D. Mason; M. Gonnelli, Photochem. Photobiol. 2004, 80, 462-470. 29. E. Riedle; M. Beutter; S. Lochbrunner; J. Piel; S. Schenkl; S. Spörlein; W. Zinth, Appl. Phys. B 2000, 71, 457-465. 30. M. J. Tauber; R. A. Mathies; X. Chen; S. E. Bradforth, Rev. Sci. Instrum. 2003, 74, 4958- 4960. 137 31. C. T. Chang; C. Y. Wu; A. R. Muirhead; J. R. Lombardi, Photochem. Photobiol. 1974, 19, 347-351. 32. H. Lami; N. Glasser, J. Chem. Phys. 1986, 84, 597-604. 33. W. Caminati; S. Di Bernardo, J. Mol. Struct. 1990, 240, 253-262. 34. A. Vannikov; V. Marevtsev, Int. J. Rad. Phys.Chem. 1973, 5, 453-456. 35. O. Bräm; A. A. Oskouei; A. Tortschanoff; F. van Mourik; M. Madrid; J. Echave; A. Cannizzo; M. Chergui, J. Phys. Chem. A 2010, 114, 9034-9042. 36. T. A. Oliver; Y. Zhang; A. Roy; M. N. Ashfold; S. E. Bradforth, J. Phys. Chem. Lett. 2015, 6, 4159-4164. 37. W. G. McGimpsey; H. Görner, Photochem. Photobiol. 1996, 64, 501-509. 38. S. V. Jovanovic; S. Steenken, J. Phys. Chem. 1992, 96, 6674-6679. 39. Y. Kimura; J. C. Alfano; P. Walhout; P. F. Barbara, J. Phys. Chem. 1994, 98, 3450-3458. 40. G. Smith, The Acid-Catalyzed Polymerization of Pyrroles and Indoles. In Adv. Heterocycl. Chem., Elsevier: 1963; Vol. 2, pp 287-309. 41. D. V. Bent; E. Hayon, J. Am. Chem. Soc. 1975, 97, 2612-2619. 42. D. Sharma; J. Léonard; S. Haacke, Chem. Phys. Lett. 2010, 489, 99-102. 43. B. Valeur; G. Weber, Photochem. Photobiol. 1977, 25, 441-444. 44. J. Peon; G. C. Hess; J.-M. L. Pecourt; T. Yuzawa; B. Kohler, J. Phys. Chem. A 1999, 103, 2460-2466. 45. G. Kumar; A. Roy; R. S. McMullen; S. Kutagulla; S. E. Bradforth, Faraday Discuss. 2018, 212, 359-381. 46. J. A. Kloepfer; V. H. Vilchiz; V. A. Lenchenkov; X. Y. Chen; S. E. Bradforth, J. Chem. Phys. 2002, 117, 766-778. 47. J. A. Kloepfer; V. H. Vilchiz; V. A. Lenchenkov; A. C. Germaine; S. E. Bradforth, J. Chem. Phys. 2000, submitted (May, 2000). 48. M. Faubel; S. Schlemmer; J. Toennies, Zeitschrift für Physik D Atoms, Molecules and Clusters 1988, 10, 269-277.

Abstract (if available)

Abstract Photochemistry in the condensed phase is often complicated due to additional solute-solvent interaction compared to isolated environment. Understanding how the electronic structure of a solute in liquid is intricately bound up with the arrangement of a host liquid provides insight into how non-adiabatic photochemistry takes place in the condensed phase. For example, the presence of water provides additional solute-solvent interactions compared to non-polar solvents, changing the stability of ionized products

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

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

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

PDF

Computational spectroscopy in gas and condensed phases

PDF

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

PDF

Spectroscopic signatures and dynamic consequences of multiple interacting states in molecular systems

PDF

Unraveling photodissociation pathways in pyruvic acid and the role of methylhydroxycarbene

PDF

Adiabatic and non-adiabatic molecular dynamics in nanoscale systems: theory and applications

PDF

Nonlinear spectroscopy of chemical phenomena at interfaces: carrier trapping and local electric fields

PDF

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

PDF

Ultrafast dynamics of excited state intra- and intermolecular proton transfer in nitrogen containing photobases

PDF

Probing charge transfer and electric fields at the bulk and interface using vibrational spectroscopy

PDF

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

PDF

Water partitioning between the bulk and an electrode surface

PDF

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

PDF

Interfacial polarization and photobasicity: spectroscopic studies of the electrode environment and photochemical effects

PDF

Excited state proton transfer in quinoline photobases

PDF

Modeling x-ray spectroscopy in condensed phase

PDF

Capturing the sun: leveraging excited state dynamics

PDF

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

PDF

Dissociation energy and dynamics of HCl and aromatic-water clusters

Asset Metadata

Creator Kumar, Gaurav (author)

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

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Publication Date 10/10/2019

Defense Date 10/08/2019

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

Tag conical intersection,indole,liquid microjet,nonadibatic,OAI-PMH Harvest,photoelectron spectroscopy,solvation dynamics,transient absorption

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Bradforth, Stephen (committee chair), Dawlaty, Jahan (committee member), El-Naggar, Moh (committee member)

Creator Email gauravk@usc.edu,iiser.gaurav@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-223481

Unique identifier UC11675457

Identifier etd-KumarGaura-7855.pdf (filename),usctheses-c89-223481 (legacy record id)

Legacy Identifier etd-KumarGaura-7855.pdf

Dmrecord 223481

Document Type Dissertation

Rights Kumar, Gaurav

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