University of Southern California Dissertations and Theses

Ultrafast spectroscopic interrogation and simulation of excited states and reactive surfaces

(USC Thesis Other)

Ultrafast spectroscopic interrogation and simulation of excited states and reactive surfaces

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ULTRAFAST SPECTROSCOPIC INTERROGATION AND SIMULATION OF EXCITED STATES AND REACTIVE SURFACES by Christopher Andrew Rivera A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (CHEMISTRY) May 2011 Copyright 2011 Christopher Andrew Rivera ii Dedication Dedicated to my amazing family, my parents Marcus and Cynthia Rivera, my sister Tricia Freeman and my joon Mahsa Rivera. iii Acknowledgements First I would like to express my sincere gratitude to Steve. I could not have asked for a better advisor. I have been so fortunate to have such a challenging and rewarding research project which has led to many collaborations and side adventures. I will never forget dissolving 4 grams of NaCN in 50 ml of water at three in morning! It is a testament to the enthusiasm that you give to your group members and especially me, which allowed me to challenge myself and make new discoveries when most students would have given up. I appreciated all of the advice and encouragement over the years and my training went well beyond the lab. You have given me every opportunity to succeed and I am truly grateful for all the guidance and support. To the Bradforth group, thank you all for sitting through the endless and sometimes painful practice talks. I could always count on getting asked the tough questions (except from Tom!) which made me a better scientist. I also appreciate all of the discussions and help over the years. I want to say a special thanks to Diana Warren. Although we have not always gotten along, we will always be friends. I could always count on you to stay with me all night smelling for almonds if I needed you to. It was nice to know I was not alone through these last 6.5 years. Good luck at New Focus, you will be awesome. I also want to say thank you to Prof. Chris Elles. I probably learned the most about carrying myself as a scientist during the time we worked together and shared an office. Seeing you go through the process of iv becoming a professor and writing papers has really inspired me to publish my first work and obtain a postdoctoral position. I want to wish you all the best in your new position at KU and hopefully we can work together again someday. Finally to my family and friends, thank you all for all of the love and support. You all probably saw the darkest side of graduate school. You had to hear all of the complaints, but I want you all to know I love you back, and it was not nearly as bad as I made it sound! To Matt Black, my roommate for three years, you are truly one of the smartest people I know and you are going to be as good of a husband and researcher as you are a friend. Thanks for the Dr. Mario sessions when I needed to unwind. I really could not have made it without you being around. To Chris Nemirow, thanks for being my surfing buddy. I could always count on you paddling into superfog at Huntington, 8 ft closeouts at Marina or even sketchy shark infested point breaks somewhere in Mexico when no one else would have been there for me. Thanks dude. To Mom and Dad, thanks for answering the phone, even if the answer was “now what?”. You have been so supportive and I couldn’t ask for more. I never thought I would move all over the country but I know I will always have a home to come back to. To my sister, Tricia, I am so immensely proud of the nurse, wife and mother you have become. You might feel like you have been in my shadow but I always knew how successful you would become. You were always better with people and your natural caring instincts make you such an amazing person. Finally, to my wife Mahsa, you are my world. Although I did not start this with you, I could not have finished it without you. I know you have seen the worst of me at times but hopefully this is the start to our amazing future together. I want you to know that I will always be there for you, especially when you are writing one of these for yourself. I love you joon! v Table of Contents Dedication ii Acknowledgements iii List of Tables viii List of Figures ix Abstract xvii Chapter 1. Introduction 1 1.1 Watching reactions in the condensed phase 1 1.2 Future Directions 9 1.3 References for Chapter 1 11 Chapter 2. Gires-Tournois interferometer type negative dispersion mirrors for deep ultraviolet pulse compression 15 2.1 Introduction 16 2.2 Experimental setup 18 2.3 Results 25 2.4 Acknowledgements 33 2.5 References for Chapter 2 34 Chapter 3. Revealing I( 2 P 1/2 ) radicals in ethanol: Evidence of a new detachment pathway from 200 nm photoexcitation of I - and liquid photoelectron spectroscopy 36 3.1 Introduction 37 3.2 Background: Spectroscopy of I - and transient species 42 3.3 Experimental 44 3.3.1 Pump-probe setup 44 3.3.2 Photoelectron setup 47 3.4 Results 48 3.4.1 Water 48 3.4.2 Ethanol 52 3.4.3 Photoelectron spectrum 55 3.5 Discussion 59 vi 3.5.1 Analysis of the photoelectron spectrum 59 3.5.2 Assignment of I* in ethanol 61 3.5.3 Detachment pathways in H 2 O and ethanol 63 3.5.4 Modeling of transients in ethanol 66 3.6 Conclusion 68 3.7 Acknowledgments 69 3.8 References for Chapter 3 70 Chapter 4. The dynamical role of solvent on the ICN photodissociation reaction: Connecting experimental observables directly with molecular dynamics simulations 74 4.1 Introduction 75 4.2 Experimental 79 4.2.1 266 nm pump pulses 80 4.2.2 Broadband continuum probe 80 4.2.3 Sample and spectrometer 81 4.3 Molecular dynamics 82 4.3.1 Potential energy functions 83 4.3.2 MD methods 85 4.4 Spectroscopy 87 4.4.1 ICN Ã spectrum 87 4.4.2 Iodine radical 88 4.4.3 Cyano radical 91 4.5 Experimental results 93 4.5.1 Isotropic results for ICN in water 93 4.5.2 Isotropic results for ICN in ethanol 94 4.6 MD results 96 4.6.1 Spectroscopic sorting 96 4.6.2 Results from 3 Π 0+ 97 4.6.3 Results from 3 Π 1 99 4.7 Discussion 100 4.7.1 Assignments for ICN in ethanol 101 4.7.2 Assignments for ICN in water 102 4.7.3 Kinetic modeling 104 4.8 Comparison to MD 112 4.8.1 Water 112 4.8.2 Ethanol 114 4.9 Conclusions 117 4.10 References for Chapter 4 119 vii Chapter 5. The effect of increasing excitation energy on the condensed phase ICN photodissociation reaction 123 5.1 Introduction 124 5.2 Experimental 127 5.2.1 271 nm pump pulses 128 5.2.2 Probe 128 5.2.3 Sample and spectrometer 129 5.3 Molecular dynamics methods 130 5.4 Results 133 5.4.1 Isotropic results and assignments 133 5.4.2 Anisotropic results 135 5.5 MD results 137 5.5.1 Sorting 138 5.5.2 Results from 3 Π 0+ 139 5.5.3 Results from 3 Π 1 143 5.6 Discussion 145 5.6.1 Isotropic pump energy effects 146 5.6.2 Anisotropic pump energy effects 148 5.6.3 Comparison between MD and experiment 150 5.7 Conclusions 151 5.8 References for Chapter 5 154 Chapter 6. Future directions: Isolating CN radicals in solution 156 6.1 Introduction 156 6.2 2PA of cyanide in solution 161 6.3 Proposed 3 pulse experiment 167 6.4 Concluding remarks 169 6.5 References for Chapter 6 170 Bibliography 172 viii List of Tables Table 3.1 Gaussian fit parameters for the first two iodide contributions to the 1 M NaI in ethanol PE spectrum. 59 Table 4.1 Isotropic data fit parameters for ICN photodissociation in water and ethanol. Transient absorption was fit with multi-exponential functions at 330, 385, and 500 nm. 105 Table 5.1 Relevant energetics and branching ratios from gas phase photodissociation of ICN reproduced from reference 4. 126 Table 5.2 Isotropic data fit parameters for ICN photodissociation. Transient absorption at 385 nm probe. 135 Table 6.1 Gaussian fit parameters for the simultaneous fit of the 2PA spectra of 2 M NaCN in water and the resulting polarization ratio. 164 ix List of Figures Figure 1.1 Photodissociation of a triatomic molecule in the condensed phase. 1) Excitation occurs within typical pulse lengths. 2) Dissociation occurs as molecules exit the Frank-Condon region where energy is distributed as electronic energy, translational, rotational and vibrational kinetic energy. 3) Collision with the solvent cage leads to either 4) recoil and recombination, or cage escape and diffusion. On longer time scales, diffusional recombination or abstraction reactions also lead to depletion of photoproducts. 2 Figure 1.2 Measured transient pump-probe anisotropy after 233- nm photodissociation of ICN in ethanol (circles) and simulated anisotropy in charged liquid Ar reproduced from ref. 10. The inset shows the highly non- Boltzmann initial populations of CN fragments in the rotationally hot (red) and cold (blue) channels resulting from this photodissociation in the gas phase; for comparison, a 300 K thermal rotational distribution is also shown (magenta curve). The gas-phase behavior of the two channels is assumed to carry over to the initial rotor distributions in the condensed phase. 4 Figure 2.1 Experimental setup. Third harmonic light is generated in a hollow core fiber and auto-correlated in a thin film water jet as described in ref. 17. DMA and DMB correspond to the dispersive mirror pair and M4 to the 0 o low dispersion dielectric mirror that makes up the compressor. DMA and M4 can be translated to control the number of bounces per dispersive mirror and the angle of the DM setup can also be varied. W is a pair of Suprasil optical wedges. CHR is a curved low dispersion dielectric high reflector (f=35 cm). M1, M2 and M3 are low dispersion 0 o and 45 o dielectric high reflectors. 20 x Figure 2.2 Dispersive mirror GDD curves designed (thick black) and manufactured (red) for optimal reflectivity at 7 o AOI. Typical 266.5 and 271.5 nm spectra (thin black) with 4.9 and 5.2 nm FWHM respectively produced from FWM in an argon-filled hollow core fiber. 22 Figure 2.3 Dispersive mirror GDD curves for optimal 7 o AOI based on the measured transmission spectrum (black) and simulated curve at 26 o AOI (red) for s-polarized light. 24 Figure 2.4 (a) Autocorrelation of 266.5 nm pulse after DM compression (black) and prism compression (red). The transform limit is calculated to be 21 fs from the corresponding spectral bandwidth assuming a Gaussian shape. (b) Detailed comparison on a log scale showing the deviation of the DM and prism compressed pulses from Gaussian (black dashed). 26 Figure 2.5 (a) Comparison of measured 266.5 nm pulse autocorrelation (black curve) after 24 bounces and 1.0 mm of Suprasil (corresponding to ~ 22 bounces) and 271.5 nm pulse autocorrelation (blue curve) after 24 bounces and 0.9 mm of wedge. Also shown is the simulated 266.5 nm pulse (red curve) compressed by 26 bounces off of the 26 o AOI DM dispersion curve of Figure 2.3. (b) Comparison of the measured deconvoluted pulse width as a function of the number of DM reflections for 266.5 nm pulses (experimental, black circles), 271.5 nm pulses (experimental, blue squares) and 266.5 nm pulses (simulated, red curve). 30 Figure 3.1 (a) Iodide absorption spectrum in water (red) and possible transients from photodetachment including solvated e - (black), I( 2 P 3/2 ) (blue), and the predicted spectrum of I( 2 P 1/2 ) (green). (b) Identical spectra for photodetachment in ethanol. The red shift of the I( 2 P 3/2 ) peaks is predicted by the gas phase ionization potential (ref. 20) and the I( 2 P 1/2 ) transition is shifted by 41 xi the gas phase splitting of 0.94 eV. Colored stripes indicate the probe wavelengths used as described in the text. Figure 3.2 (a) Contour plot of the transient photoproducts from photodetachment in H 2 O, absorption in mOD increases from blue to red. (b) Contour plot after solvated electron is subtracted as described in the text. (c) Contour plot of the transient photoproducts in ethanol. (d) Remaining transient photoproducts after solvated electron subtraction in ethanol. 50 Figure 3.3 Several time cuts showing the transient absorption spectrum and the corresponding electron signal that was subtracted at each time point for (a) water and (b) ethanol. The resulting transient spectra after electron subtraction are shown for (c) water and (d) ethanol. 51 Figure 3.4 Solvated electron data for 750 nm (black, solid) and 920 nm (red, solid) and electron fit (dashed) from the model described in the text. 52 Figure 3.5 The experimental PE spectrum (100 eV photon energy) arising from a 1M NaI/ethanol solution (black line) which contains contributions from both liquid and gas phase ethanol and iodide. Pure liquid ethanol (red line) was recorded with 200 eV photon energy in a separate experiment. Details of the transformation from electron kinetic energy to ionization potential are given in the text. The inset shows the difference spectrum which contains only iodide and gas phase ethanol contributions. 55 Figure 3.6 A plot of the iodide absorption spectrum in ethanol (below, black line), and the photoelectron spectrum from an ethanol liquid jet (above, black squares). The absorption spectrum has been deconvoluted (blue line) as well as the PE spectrum (red line). The overlap of the band associated with I( 2 P 1/2 ) in the absorption 61 xii spectrum has overlap with both the I ( 2 P 3/2 ) and I( 2 P 1/2 ) direct photodetachment channels which leads to formation of both core photoproducts. For reference, the iodide spectrum in water (grey line) and PE spectrum in water (grey circles) are reproduced from ref. 12. Figure 3.7 (a) The transient spectrum after removal of electron signal is compared to a (b) simple simulation of the I and I* dynamics where I* is allowed to relax to I on a 20 ps timescale. An additional decay of I* on a 15 ps timescale is also included which is attributed to I*+e¯ recombination although several other mechanisms for this decay are discussed in the text. A qualitatively identical result arises from a simulation where the extinction coefficient of I* is increased compared to I. In this case I* is allowed to deactivate to I on a 10 ps timescale and no geminate recombination is included other than the timescale of 1.2 ns described by the electron model. 63 Figure 4.1 Potential energy curves corresponding to the ground state (black line) and three lowest excited states (colored lines) reproduced from ref. 21 and plotted as a function of I to CN center-of-mass and linear ICN. The molecular dynamics sorting cut-offs are also shown (gray dashed lines). The inset shows the calculated ICNÃ band from molecular dynamics simulations. 84 Figure 4.2 Transient absorption spectra for products generated from ICN photodissociation in water and ethanol. The ICN (black line), I( 2 P 3/2 ) (solid green line) from ref. 47, I( 2 P 1/2 ) (dashed green line, see text), and gas phase CN B←X (red line) adapted from ref. 52 and 18 are plotted as a function of wavelength. The calculated aqueous CN B←X spectrum (blue line) from ref. 23 is shown in panel (a). The colored bar represents the probe wavelengths available in this experiment. 90 xiii Figure 4.3 Time-dependant transient absorption at magic angle for a) 170 mM ICN in water and b) 260 mM ICN in ethanol excited at 266 nm. Contour plots show the 20 ps time-dependant transient signal, ΔA, with color scale shown to right of each panel increasing in intensity from blue to red. c) Contour plot show only the first picosecond of the experimental data in water and d) ethanol. e) Several spectral slices are shown at time intervals ranging from 70 fs to 20 ps from water and f) ethanol. 92 Figure 4.4 Three individual probe wavelength cuts at 330 nm, 385 nm and 500 nm for ICN in a) water and b) ethanol taken at magic angle showing isotropic transient absorption. 94 Figure 4.5 Transient a) I, I* and b) CN population from MD simulations, insets show magnification of population decay over the first picosecond. 99 Figure 4.6 MD Population decay of I and CN from excitation to 3 Π 1 surface. 100 Figure 4.7 Kinetic scheme for I and I* populations in ethanol. Red arrows indicate processes which are assumed to happen instantaneously. 107 Figure 4.8 a) Fits from kinetic model to 330 and 550 nm experimental data and b) residual CN transient absorption remaining after subtraction of I and I* contributions. 109 Figure 4.9 a) Fit of the 50 fs spectral cut using I, I* and CN (CN not included in kinetic model), and b) residual CN absorption in water after I and I* contributions are removed. The first 500 fs is highlighted (inset) to show the fast shifting of the CN absorption. 111 Figure 4.10 Comparison of a) water experimental spectra over 5 ps and b) reconstruction (see text) of transient spectra 114 xiv from MD simulation of transient populations. Magnification of c) the initial picosecond of ICN in water is compared to d) the first picosecond of the MD reconstruction. Figure 4.11 Comparison of a) ethanol experimental spectra over 20 ps and b) reconstruction (see text) of transient spectra from MD simulation of transient populations stretched in time by a factor of 6. Magnification of c) the initial 5 ps of ICN in ethanol is compared to d) the (stretched) first 5 ps of the MD reconstruction. 116 Figure 5.1 Potential energy curves corresponding to the ground state (black line) and three lowest excited states (colored lines) reproduced from ref. 8 and plotted as a function of I to CN center-of-mass and I-CN angle of 0 degrees. The dashed line corresponds to the approximate solvation shift of the ground state. 125 Figure 5.2 Transient absorption spectra generated from ICN photodissociation in ethanol. The ICN (black line), I( 2 P 3/2 ) (solid green) from ref. 32, I( 2 P 1/2 ) (dashed green), and gas phase CN B←X (red) reconstructed from ref. 11 and 31 are plotted as a function of wavelength. The colored bar represents the probe wavelengths available in this experiment. 132 Figure 5.3 a) Time dependant transient spectral cut at 100 fs for 260 mM ICN in ethanol excited at 271 nm. Contour plots show the time-dependant transient signal increasing in intensity from blue to red. b) The time dependant transient signal measured at 385 nm corresponding to CN signal (black squares) and the fit of a double exponential function (red line). 134 Figure 5.4 Pump wavelength comparison of the long time anisotropy decay observed for ICN photodissociation in ethanol. 255 and 225 from ref. 4 and 233 from ref. 1. 137 xv Figure 5.5 Transient anisotropy decay for 266/400 nm (black triangles) and 233/390 nm (red diamonds) over the first 500 fs. 138 Figure 5.6 Transient a) I, I* and b) CN population from MD simulations of ICN photolysis at 266 nm (black), 255 nm (red) and 233 nm (blue). The insets show an expansion of the first picosecond. 141 Figure 5.7 Average transient rotational anisotropy from MD simulations of ICN photolysis at 266 nm (black), 255 nm (red) and 233 nm (blue). The inset shows the long time anisotropy over the entire 5 ps simulation, the dashed line is a guide. 143 Figure 5.8 Transient population decay of a) I/CN and b) average anisotropy from trajectories started on the 3 Π 1 excited state surface of ICN. The inset shows the average long time anisotropy signal for the entire 5 ps trajectory. 145 Figure 5.9 Comparison of 233 nm (red diamonds) and 266 nm (black triangles) pumped anisotropy measurements and a 50:50 mixture of the 3 Π 0+ to 3 Π 1 (red line) and a 30:70 mixture (black line) from 266 nm simulations. The two curves are offset for clarity. 149 Figure 5.10 The transient anisotropy signal for 266 nm pump a) 355 nm and b) 450 nm probe. The small signal which goes to zero at 1 ps for 450 nm most like comes from small overlap with the tail of the CN absorption spectrum. 152 Figure 6.1 The one-photon absorption (1PA) spectra of aqueous cyanide (black, ref. 19) and cyano radical in water (blue, ref. 4) and gas phase (ref. 12). 158 xvi Figure 6.2 The parallel (red) and perpendicular (blue) 2PA spectrum of 2 M NaCN in water and the parallel 2PA spectrum of water (purple line) from ref. 23. The polarization ratio for 2 M NaCN (green circles) and the polarization ratio for pure water (purple circles) from ref. 23 are also shown. 160 Figure 6.3 Simulated polarization ratio (solid green) and the simulated parallel 2PA spectrum (solid black) overlaid with the experimental data from Figure 6.1. The three excited states (blue, orange, gold) and water (magenta) that were used in the simulation are also shown with their parallel intensities. 165 Figure 6.4 Experimental schematic for proposed 3-pulse experiment where fourth harmonic 200 nm light is generated by doubling 800 nm in a type I BBO and then mixing the 400 nm with 800 nm to generate 266 nm in a type II BBO. The third harmonic 266 nm is then mixed with residual 800 nm in another type I BBO to make 200 nm which is overlapped spatially and temporally with 1200 nm OPA signal in a sample jet. The probe is super continuum generated in a translating CaF 2 disk and collected by a broad band spectrometer. The time delay of the probe is controlled by a motorized delay stage and the 200 nm is chopped to half the repetition rate. The transmission difference of the probe with 200 nm on and off is used to construct the transient absorption in mOD. 168 xvii Abstract Making a movie of a chemical reaction requires intricate knowledge of both the chemical actors and the environment. Furthermore, the subtleties of the reactive surfaces and the effect of external forces such as solvent electrostatics and molecular collisions on these surfaces often play a major role in the reaction outcome. Many of these external effects begin influencing the reaction within tens of femtoseconds, requiring sophisticated experimental techniques in order to observe dynamics on these timescales. Ultrafast pump-probe spectroscopy is employed to study the benchmark ICN Ã photodissociation reaction. This prototypical triatomic system is known to produce a highly non-equilibrium distribution of rotationally excited cyano radicals. The generation of these rotors comes from a non-adiabatic transition which applies extra torque to an already rapidly bending I-CN bond and leads to two product states corresponding to the I*( 2 P 1/2 ) + cold CN and I( 2 P 3/2 ) + hot CN channel. Simulations show that this rotationally excited CN pushes solvent molecules out of the way to create a cavity where free rotation takes place for several picoseconds. Our hypothesis is that this spinning CN resembles a gas phase molecule, and thus should have a gas phase spectral signature. Although the relaxation of the hot CN takes place on relatively long time scales for ultrafast spectroscopists, the initial curve crossing event takes place much earlier (<30 fs) and not much is known about the influence of the solvent on this process. Using sub 40 fs DUV pulses compressed by a novel Gires-Tournois pulse xviii compression system and multiplexed broadband probing, we can interrogate this reaction at much higher levels of detail than previously possible. A clear signature from the gas-phase-like CN B←X transition is observed in both water and ethanol. The spectral contributions of I( 2 P 3/2 ) and I*( 2 P 1/2 ) can be removed from the transient spectrum, because of new and independent experiments also reported here that capture these spectra by photodetachment of iodide in ethanol. This analysis allows us to watch the evolution of the CN band and estimate first pass curve-crossing probabilities for the first time in solution. Semiclassical molecular dynamics simulations of ICN bond breaking in water including the non-adiabatic transition, allow us to microscopically break down the energy flow which takes place immediately after dissociation. Using a spectroscopic sorting criteria which follows only freely dissociated trajectories, we can compare the simulation directly to our experimental observables for the first time. Future work is needed to measure the equilibrium spectrum of the CN radical in the absence of iodine transients as a last reference, but, with our current time resolution and the ability to simultaneously image the absorption spectrum of multiple transient species as a function of time, we are much closer to extracting all pertinent information from a movie of this benchmark solution phase reaction occurring in real time. 1 Chapter 1. Introduction 1.1 Watching reactions in the condensed phase For physical chemists who study reaction dynamics, making a movie of a chemical reaction is considered their “holy grail”. To watch the flow of energy between atoms and molecules and identify the excited states of all reacting species as they move along a reaction coordinate would be invaluable to science, especially in condensed phases and at surfaces where most reactions occur. Today, many sophisticated spectroscopic techniques, coupled with high level theoretical treatments are literally shedding light on some of the earliest dynamical events after excitation, and on a microscopic level. In solution the movie becomes significantly more complicated due the increased number of molecules in the vicinity of the reacting species. Of course, strongly associated solvent environments and their effect on polar and ionic solutes cannot be thought of merely as a confining media. In many cases, the subtle interactions such as stabilization or destabilization of ground and excited states, solvent fluctuations, and side reactions of the solvent with excited products, can greatly influence the observed reaction outcomes and timescales. 1-4 In order to separate out these dynamical processes and identify those most important to the reaction outcomes, measurements must be sensitive to the timescales involved. Due to the compact reaction environments, many of these processes occur on femtosecond timescales, and spectroscopic interrogation requires 2 the use of ultrashort laser pulses. Understanding when, why and how strong solvent effects will be is critical to making this “reaction movie” in the condensed phase. FIG. 1.1. Photodissociation of a triatomic molecule in the condensed phase. 1) Excitation occurs within typical pulse lengths. 2) Dissociation occurs as molecules exit the Frank-Condon region where energy is distributed as electronic energy, translational, rotational and vibrational kinetic energy. 3) Collision with the solvent cage leads to either 4) recoil and recombination, or cage escape and diffusion. On longer time scales, diffusional recombination or abstraction reactions also lead to depletion of photoproducts. Photodissociation has been a method of choice for spectroscopists interested in studying the bond-breaking half reactions, 5 especially in the gas phase. Figure 1.1 shows a schematic for the most typical triatomic photodissociation reaction in the condensed phase, initiated by a laser pulse. A triatomic molecule is the simplest system leading to photoproducts with all available degrees of freedom. As mentioned in the previous paragraph, in solution, the close proximity of the solvent 3 and finite life time of photoproducts necessitates the use of time-resolved spectroscopies in order to deduce the fastest processes. Historically, the processes of interest in the condensed phase include solvent caging, recombination of geminate pairs and rotational, vibrational and kinetic energy release of nascent products. 4, 6-12 Early work by Fornier de Violet et al. 13 and Ottolenghi et al. 14 on polyhalides and iodoaromatic compounds provided insight on the transient lifetimes and radical spectra. Work by Hochstrasser et al., Vohringer et al., and Ruhman et al. on the first triatomics, I 3 – and HgI 2, specifically monitoring the vibrational and rotational dynamics of the diatomic photoproducts. 7, 15-18 Among the triatomics studied, the cyanogens halides, and particularly ICN, are some of the best characterized. As a benchmark system, the ICN photodissociation reaction has been extensively studied in both the gas phase and condensed phase experimentally and theoretically. 9-12, 19-36 The richness of this system comes from a well-known non-adiabatic transition accessible by excitation to the ICN Ã dissociation continuum. 19 The CN fragment from molecules undergoing this transition are imparted with a large non-equilibrium distribution of rotational energy making this an interesting case and has spurred the continuing study of this system. Most recently, work by Crim et al. 37, 38 and Orr Ewing et al. 39 have explored the reactivity of the CN photoproduct soon after ICN bond breaking. Molecular dynamics simulations by Benjamin and co-workers which include Tully’s method describing curve crossing have revealed detailed information regarding the product populations and kinetic energy distributions, but until now, there has been no 4 detailed attempt to compare experimental results to these simulations. 35, 36, 40 Details of the CN rotational relaxation in the solvent are particularly intriguing. Simulation results from Stratt and Tao of the spinning CN fragment in liquid argon were used to determine that anisotropic signals observed experimentally in polar solvents with 30 fs time resolution by Moskun et al. in our laboratory were indeed the result of freely rotating CN (See Figure 1.2). 10 These simulations showed that a cavity is created by the spinning molecule and the solvent is incapable of slowing this rotation down for tens of rotational periods. FIG. 1.2. Measured transient pump-probe anisotropy after 233-nm photodissociation of ICN in ethanol (circles) and simulated anisotropy in charged liquid Ar reproduced from ref. 10. The inset shows the highly non-Boltzmann initial populations of CN fragments in the rotationally hot (red) and cold (blue) channels resulting from this photodissociation in the gas phase; for comparison, a 300 K thermal rotational distribution is also shown (magenta curve). The gas-phase behavior of the two channels is assumed to carry over to the initial rotor distributions in the condensed phase. 5 While much is known about condensed phase ICN photodissociation, there remain important questions that up until now had not been addressed due to experimental limitations. This thesis culminates with the identification of transient absorption spectra, elucidating the effect of solvent on the non-adiabatic transition, and examining the kinetic energy release of photo fragments from ICN photodissociation with sub 40 fs time resolution. Although this work is focused on events relating to one specific dissociation half reaction, the knowledge gained has important and unexpected consequences towards our view of the energy distributions and relaxation resulting from reactions in the condensed phase. 9, 10, 36 We will find that under certain conditions where kinetic energy release is significant enough, the solvent effects can be comparatively weak such that the products are seemingly unaffected for long periods on typical reaction timescales. 9, 10 Remarkably, we can observe a clear signature of the rotationally hot product in the transient absorption spectrum. The primary objective of this work is two-fold; the first is to utilize a sophisticated pump probe apparatus that combines a 30 fs DUV source with broadband simultaneously collected probe pulses to resolve the visible transient absorption spectrum from ICN bond breaking, and second, to connect molecular dynamics simulations with our experimental signals by applying a spectroscopically motivated sort of the resulting trajectories. Typically condensed phase electronic absorption spectra in the visible and UV are quite broad and often significantly 6 overlap, making assignments and deconvolution of individual transitions very difficult. Thus, a portion of this dissertation is focused on determination of the radical absorption spectra corresponding to the expected transient species involved in ICN photodissociation. Utilizing advances in DUV pulse generation 41, 42 and compression implemented in our laboratory, 43 we have shown that pulses with sub-30 fs widths are possible and can be effectively used in time-resolved pump-probe. Transmission of DUV through typical optical materials such as lenses, filters and even air, introduces significant dispersion of the pulse in the time domain. 44 Prism compression is often used to compress pulses in the visible and UV but the thick prism substrates used introduces higher order dispersion which cannot be completely compensated. 41 The result of this dispersion is apparent in the wings of the pulse which deviate from Gaussian pulse shapes at 30% of the peak intensity in some cases. 43 These wings overlap with transient absorption and are difficult to separate from the fastest signals. This is especially problematic in the case photodissociation of ICN which has relatively small extinction coefficients for both the parent molecule (ε < 300 M -1 cm -1 ) and the photofragments generated, which all are expected to have ε max below 3000 M -1 cm -1 depending on the solvent. 25 Good chromophores such as dyes, aromatics and solvated electrons have extinctions that are orders of magnitude stronger and thus, in the case of solvated electrons produce 10 times as much transient signal per photon as ICN. 45 The next chapter of this thesis describes a novel way to compress DUV with superior control of higher order 7 dispersion with good transmission efficiency which improves the overall time resolution of our experimental apparatus. Photodetachment of iodide is the focus of the third chapter. The absorption spectrum of iodide in solution displays two distinct charge-transfer-to-solvent (CTTS) absorption bands. 46 Removing an electron by exciting the lowest CTTS state which corresponds to an I( 2 P 3/2 ) core known to produce iodine radicals and solvated electrons in solution. 47 It was the hope of previous experiments in water, that excitation of the second CTTS state corresponding to a I*( 2 P 1/2 ) core would lead to the observation of spin excited iodine radicals. 48 Unfortunately this was not the case and an autodetachment pathway was shown to outcompete the I* formation. 48 We reconfirm this observation with substantially better signal to noise and time resolution and perform the same experiment on iodide in ethanol. Surprisingly, we see a clear signature of I* in ethanol and propose with evidence from photoelectron spectroscopy, that a direct detachment pathway is responsible for its appearance. This work is motivated by the fact that we expect to produce both I and I* radicals from the photodissociation of ICN and currently there are no reports of resolved I* in any protic solvents. A broadband transient absorption study of the ICN photodissociation at 266 nm is presented in Chapter 4. The chapter focuses on the spectroscopy and populations dynamics. We clearly observe a gas-phase like CN absorption band stemming from the rotationally excited CN population which is nearly identical in both water and ethanol. In water we clearly observe this band shift and broaden 8 similarly to predictions from EOM-CCSD based ab initio calculations. We also observe other absorption features in the regions where we expect to see I and I* and we use kinetic models to follow the populations of these species. This in turn allows us to estimate the branching ratio through a dissociative surface crossing transition in solution for the first time. In this chapter we also present collaborative results with Benjamin and co-workers which reanalyze previous MD results with spectroscopic sorting criteria. We carefully remove trajectories in the ensemble average once the products recombined, thus reproducing the isotropic populations of I, I* and CN which we observe experimentally. We use the newly sorted simulations to reproduce the experimental spectra and make a detailed comparison between theory and experiment for the first time. Chapter 5 compares the effects observed when changing the photolysis energy. It is observed in the gas phase that increasing the pump energy provides a greater amount of available energy for torquing and translational release of the CN radical initially, 9 but in solution it is expected that this will lead to less curve crossing at longer times due to the increased separation of the I* and CN photofragments. We see evidence of this behavior when comparing both the isotropic and anisotropic signals from our experiments and we compare these observations to MD simulations carried out at 233, 255 and 266 nm in water. 9 1.2 Future Directions The spectrum of CN radicals in solution plays an important role in assignments of our transient spectra. While Pieniazek et al. has presented high level condensed phase ab initio calculations in order to predict the spectrum of CN, 49 there remains no clear experimental signature of this species until this dissertation work. To solidify our assignment it would be beneficial to isolate the CN radicals devoid of other overlapping transitions from I and I* radicals. In the final chapter we propose an experiment which aims to detach an electron from cyanide (CN – ) in water to produce CN radicals analogously to the iodide photodetachment experiments in Chapter 3. To detach an electron from CN – via a CTTS transition, one must know the energy of this transition which surprisingly, has not been completely characterized. In fact, not much is known about the electronic structure of cyanide at all as all excited states are unstable in the gas phase due to autodetachment, 50 and all transitions in water lie above 7 eV (177 nm) in energy. 51 Methods for introducing liquids to vacuum environments for spectroscopic study is possible 52 but continues to be a challenging problem and propagation of vacuum-ultraviolet light in air is impossible. Therefore, to reach the CTTS state of CN – in water, which is reported to lie at 7.2 eV (172 nm) from fitting the tail which extends to ~6.6 eV (186 nm), 51 we propose a two-photon pump-probe experiment. In this experiment, two photons overlapped spatially and temporally which add up to 7.2 eV are used to excite the molecule, while a third broadband pulse would be used to probe the radicals 10 produced. In this chapter, we present the two-photon absorption (2PA) spectrum of cyanide and not only show that there is a state at 7.2 eV but that there are at least two other states that absorb at higher energies. Assignments of these higher laying states and implementation of the proposed experiments will require future experimental and theoretical work. The use of 2PA spectroscopy to identify excited states and excite molecules above VUV thresholds, using commercially available amplifiers, has opened a new window to condensed phase pump-probe spectroscopists. 11 1.3 References for Chapter 1 1. D. Raftery, M. Iannone, C. M. Phillips and R. M. Hochstrasser, Chem. Phys. Lett., 1993, 201, 513-520. 2. R. H. Bathgate and E. A. Moelwyn-Hughes, J. Chem. Soc., 1959, 2642-2648. 3. K. Tanaka, G. I. Mackay, J. D. Payzant and D. K. Bohme, Can. J. Chem., 1976, 54, 1643-1659. 4. C. G. Elles, M. J. Cox, G. L. Barnes and F. F. Crim, J. Phys. Chem. A, 2004, 108, 10973-10979. 5. R. Schinke, Photodissociation dynamics: spectroscopy and fragmentation of small polyatomic molecules, Cambridge University Press, Cambridge, 1993. 6. R. W. Anderson and R. M. Hochstrasser, J. Phys. Chem., 1976, 80, 2155- 2159. 7. A. Baratz and S. Ruhman, Chem. Phys. Lett., 2008, 461, 211-217. 8. M. Berg, A. L. Harris and C. B. Harris, Phys. Rev. Lett., 1985, 54, 951-954. 9. A. C. Moskun and S. E. Bradforth, J. Chem. Phys., 2003, 119, 4500-4515. 10. A. C. Moskun, A. E. Jailaubekov, S. E. Bradforth, G. H. Tao and R. M. Stratt, Science, 2006, 311, 1907-1911. 11. C. Z. Wan, M. Gupta and A. H. Zewail, Chem. Phys. Lett., 1996, 256, 279- 287. 12. C. J. Williams, J. W. Qian and D. J. Tannor, J. Chem. Phys., 1991, 95, 1721- 1737. 13. P. Fornier de Violet, Rev. Chem. Intermed., 1981, 4, 121-169. 14. A. Levy, D. Meyerstein and M. Ottolenghi, J. Phys. Chem., 1973, 77, 3044- 3047. 15. H. Bursing, J. Lindner, S. Hess and P. Vohringer, Appl. Phys. B-Lasers O., 2000, 71, 411-417. 16. S. Hess, H. Bursing and P. Vohringer, J. Chem. Phys., 1999, 111, 5461-5473. 12 17. N. Pugliano, S. Gnanakaran and R. M. Hochstrasser, J. Photoch. Photobio. A, 1996, 102, 21-28. 18. U. Banin, A. Waldman and S. Ruhman, J. Chem. Phys., 1992, 96, 2416-2419. 19. Y. Amatatsu, S. Yabushita and K. Morokuma, J. Chem. Phys., 1994, 100, 4894-4909. 20. A. P. Baronavski, Chem. Phys., 1982, 66, 217-225. 21. I. Benjamin and K. R. Wilson, J. Chem. Phys., 1989, 90, 4176-4197. 22. J. A. Beswick, M. Glassmaujean and O. Roncero, J. Chem. Phys., 1992, 96, 7514-7527. 23. E. M. Goldfield, P. L. Houston and G. S. Ezra, J. Chem. Phys., 1986, 84, 3120-3129. 24. W. P. Hess and S. R. Leone, J. Chem. Phys., 1987, 86, 3773-3780. 25. J. Larsen, D. Madsen, J. A. Poulsen, T. D. Poulsen, S. R. Keiding and J. Thøgersen, J. Chem. Phys., 2002, 116, 7997-8005. 26. I. Nadler, D. Mahgerefteh, H. Reisler and C. Wittig, J. Chem. Phys., 1985, 82, 3885-3893. 27. I. Nadler, H. Reisler and C. Wittig, Chem. Phys. Lett., 1984, 103, 451-457. 28. S. W. North, J. Mueller and G. E. Hall, Chem. Phys. Lett., 1997, 276, 103- 109. 29. M. D. Pattengill, Chem. Phys., 1984, 87, 419-429. 30. W. M. Pitts and A. P. Baronavski, Chem. Phys. Lett., 1980, 71, 395-399. 31. J. W. Qian, C. J. Williams and D. J. Tannor, J. Chem. Phys., 1992, 97, 6300- 6308. 32. D. Raftery, E. Gooding, A. Romanovsky and R. M. Hochstrasser, J. Chem. Phys., 1994, 101, 8572-8579. 33. J. Vieceli, I. Chorny and I. Benjamin, J. Chem. Phys., 2001, 115, 4819-4828. 34. J. Vieceli, I. Chorny and I. Benjamin, Chem. Phys. Lett., 2002, 364, 446-453. 13 35. N. Winter and I. Benjamin, J. Chem. Phys., 2004, 121, 2253-2263. 36. N. Winter, I. Chorny, J. Vieceli and I. Benjamin, J. Chem. Phys., 2003, 119, 2127-2143. 37. A. C. Crowther, S. L. Carrier, T. J. Preston and F. F. Crim, J. Phys. Chem. A, 2008, 112, 12081-12089. 38. A. C. Crowther, S. L. Carrier, T. J. Preston and F. F. Crim, J. Phys. Chem. A, 2009, 113, 3758-3764. 39. S. J. Greaves, R. A. Rose, T. A. A. Oliver, M. N. R. Ashfold, I. P. Clark, G. M. Greetham, A. W. Walker, M. Towrie and A. J. Orr-Ewing, private communication. 40. I. Benjamin, J. Chem. Phys., 1995, 103, 2459-2471. 41. A. E. Jailaubekov and S. E. Bradforth, Appl. Phys. Lett., 2005, 87, 021107. 42. C. G. Durfee, S. Backus, M. M. Murnane and H. C. Kapteyn, Opt. Lett., 1997, 22, 1565-1567. 43. C. A. Rivera, S. E. Bradforth and G. Tempea, Opt. Express, 2010, 18, 18615- 18624. 44. I. Walmsley, L. Waxer and C. Dorrer, Rev. Sci. Instrum., 2001, 72, 1-29. 45. P. M. Hare, E. A. Price and D. M. Bartels, J. Phys. Chem. A, 2008, 112, 6800-6802. 46. J. Jortner, B. Raz and G. Stein, T. Faraday Soc., 1960, 56, 1273-1275. 47. X. Y. Chen and S. E. Bradforth, Annu. Rev. Phys. Chem., 2008, 59, 203-231. 48. A. C. Moskun, S. E. Bradforth, J. Thogersen and S. Keiding, J. Phys. Chem. A, 2006, 110, 10947-10955. 49. P. A. Pieniazek, S. E. Bradforth and A. I. Krylov, J. Phys. Chem. A, 2006, 110, 4854-4865. 14 50. C. S. Ewig and J. Tellinghuisen, Chem. Phys. Lett., 1988, 153, 160-165. 51. M. F. Fox and E. Hayon, J. Chem. Soc. Faraday Trans., 1990, 86, 257-263. 52. M. Faubel, B. Steiner and J. P. Toennies, J. Chem. Phys, 1997, 106, 9013- 9031. 15 Chapter 2. Gires-Tournois interferometer type negative dispersion mirrors for deep ultraviolet pulse compression Reproduced from: Christopher A. Rivera, Stephen E. Bradforth, and Gabriel Tempea, "Gires- Tournois interferometer type negative dispersion mirrors for deep ultraviolet pulse compression," Opt. Express 18, 18615-18624 (2010) Abstract for Chapter 2 Typical femtosecond pulse compression of deep ultraviolet radiation consists of prism or diffraction grating pair chirp compensation but, both techniques introduce higher-order dispersion, spatial-spectral beam distortion and poor transmission. While negatively chirped dielectric mirrors have been used to compress near infrared and visible pulses to <10 fs, there has been no extension of this technique below 300 nm. We demonstrate the use of Gires-Tournois interferometer (GTI) negative dispersion multilayer dielectric mirrors designed for pulse compression in the deep ultraviolet region. GTI mirror designs are more robust than chirped mirrors and, can provide sufficient bandwidth for the compression of sub-30-fs pulses in the UV wavelength range. Compression of a 5 nm (FWHM) pulse centered between 266 and 271 nm to 30 fs has been achieved with less pulse broadening due to high-order dispersion and no noticeable spatial 16 deformation, thereby improving the resolution of ultrafast techniques used to study problems such as fast photochemical reaction dynamics. 2.1 Introduction Dispersive mirrors (DMs) are becoming increasingly popular for any application where dispersion control, especially group delay dispersion (GDD) compensation for compression of ultrashort pulses is required. The use of DMs as intracavity mirrors in ultrafast laser systems, 1 and for pulse compression of high energy Ti:S amplified systems 2-4 remain the most common applications due to the high damage thresholds, high reflectivity, and good spatial mode preservation of DM-coatings. 5 Furthermore, DM-based dispersion compensators are compact, robust and user-friendly. Manipulation of the linear chirp rate can also be used for quantum coherent control resulting in the observation of, for example, chirp- dependent fluorescence. 6 Recently, advances in engineering and the introduction of various high and low index materials, such as HfO 2 and SiO 2 , have extended DM applications and commercial products to the UV spectral range down to 350 nm. 7, 8 Additionally, designs for Mo/Si chirped mirrors (CMs) have been published for attosecond pulses in the XUV. 9 For chemists and physicist working in the deep ultraviolet (200-300 nm), there has been significant progress made in the efficient generation of broadband pulses by four wave mixing in fibers or in gas cells which have allowed for the 17 observation of dynamics on a timescale hitherto unobtainable. 10-16 Unfortunately, delivery of laser light to an experimental apparatus generally requires transmissive optics such as lenses, windows, waveplates, etc. which introduce significant temporal broadening. 17 There are several ways to control phase dispersion of optical pulses each with advantages and drawbacks. Prism compression generally results in good efficiency (~75%) but substantial third-order dispersion (TOD) is observed 18 and precise prism matching is needed in order to avoid significant spatial-spectral and mode distortion. Grating compression generally results in less TOD but increased fourth order dispersion (FOD) and poor transmission in the UV (< 50% even in a single pass configuration) and similar problems from spatial chirp are also unavoidable. 11 Theoretically, combinations of prisms and gratings could be used to compensate both second and third order dispersion 19 but will improve neither spatial dispersion, transmission losses nor FOD-compensation. Pulse shapers use long (50 - 75 mm) KDP crystals to precisely control the phase of pulses throughout the UV range (250 - 400 nm). These can be used to achieve near perfect Fourier-limited pulses (< 20 fs) or can be used to generate multi pulse schemes in the time domain for 2D spectroscopies. 12, 20, 21 Unfortunately, very low transmission efficiency (~20%), significant spatial chirp, parallel displacement of diffracted sub-pulses and high cost all complicate the implementation and usefulness of pulse shapers for typical pump-probe experiments where the main objective is maximum time resolution. 18 In this paper, we present to our knowledge the first set of negative dispersion mirrors designed to compress pulses in the DUV. We show that we can produce 30 fs pulses while taking advantage of the high reflectivity and minimal spatial dispersion which typically make dispersive mirrors attractive for applications in the visible and IR, 22, 23 and the added control over higher order dispersion improves the compression efficiency over standard prism and grating compensating methods. We show that this DM compressor is effective used in combination with a hollow core fiber DUV frequency source pumped by 35 - 100 fs Ti:Sa amplifier systems and that the compressor can be tuned over at least a 6 nm range from 266-272 nm for pulses that have ~5 nm of bandwidth. 2.2 Experimental setup Two amplified laser systems were employed to produce deep UV pulses and fully test the capabilities of the DM compressor. (i) A portion of the output of a 800 μJ, 110 fs, 1 kHz Ti:sapphire regenerative amplifier (Spectra Physics Hurricane) shown in Fig. 2.1 was doubled in a long (500 μm thick) BBO which was combined with the residual 800 nm to drive a hollow core fiber four wave mixing (FWM) apparatus to generate pulses having broad bandwidth in the deep UV. This difference-frequency mixing of second-harmonic light (65 μJ) with residual fundamental (65 μJ) was used to generate 4 μJ of 266 nm third-harmonic (3ω = 2ω + 2ω – ω) as demonstrated previously. 18 A typical pulse centered at 266.5 nm with 4.9 nm FWHM, assuming Gaussian pulse shape, was measured with an EPP2000 UV2 19 200-400 nm spectrometer (StellarNet Inc.) and is shown in Fig. 2.2. (ii) Similarly, a portion of output of a 3.5 mJ, 35 fs, 1 kHz Ti:sapphire regenerative amplifier (Coherent Legend USP-HE) was used to pump an identical hollow core fiber system. In this case the same 500 μm BBO was used to produce 70 μJ of 400 nm and this was combined with 115 μJ of residual fundamental to produce 5 μJ of 271.5 nm. The spectrum of a pulse with 5.2 nm of bandwidth from this latter system is also shown in Fig. 2.2. The difference in center frequency of the generated third-harmonic light is a result of the larger fundamental bandwidth of system (ii), resulting in a small amount of tunability when overlapping with the second-harmonic in the hollow core fiber. The output of the hollow waveguide DUV source was collimated by a custom low-dispersion curved dielectric mirror with a radius of curvature of -70 cm and steered into a DM pair before being sent into an autocorrelator. The DUV beam was split, then characterized in an interferometer by measuring the simultaneous two- photon absorption generated by overlapping the two beams spatially and temporally in a 100 μm jet of flowing liquid water. 18, 24 The autocorrelation traces were collected by scanning a delay stage in one arm of the interferometer and measuring the change in absorption as a function of delay time. Identical material and number of coated surfaces are present in each arm. 20 FIG. 2.1. Experimental setup. Third harmonic light is generated in a hollow core fiber and auto-correlated in a thin film water jet as described in ref. 17. DMA and DMB correspond to the dispersive mirror pair and M4 to the 0 o low dispersion dielectric mirror that makes up the compressor. DMA and M4 can be translated to control the number of bounces per dispersive mirror and the angle of the DM setup can also be varied. W is a pair of Suprasil optical wedges. CHR is a curved low dispersion dielectric high reflector (f=35 cm). M1, M2 and M3 are low dispersion 0 o and 45 o dielectric high reflectors. The DM mirrors are 20x40x10 mm and were designed to compensate for ~50 fs 2 of GDD and ~12.5 fs 3 of TOD per bounce, as well as to have 99% reflectivity at 268 nm at 7 o angle of incidence (AOI) from perpendicular to the mirror substrate. The GTI-like design consisted of a high-reflecting 42-layer quarter-wave stack, a half-wave high-index spacing layer and a partially-reflecting two-layer quarter-wave section. HfO 2 and SiO 2 were employed as coating materials. In chirped mirror designs the frequency dependence of the group delay imparted upon reflection is controlled by means of the penetration of the different wavepackets into the multilayer; consequently, all layer thicknesses will sensitively affect the GDD of the 21 mirror. Layer thickness accuracies in the range of 0.5 nm are required for the manufacturing of CMs for the visible and infrared spectral range. Since the average layer thickness of CMs (and consequently the acceptable layer thickness tolerance) scales roughly linearly with the central wavelength, an absolute layer thickness accuracy in the range of 1 Angstrom would be required in order to manufacture CMs for the sub-300-nm wavelength range. This is hardly achievable with any state of the art deposition technique for dielectric optical layers. In contrast to CMs, GTI-like dispersive mirrors are much more robust; deviations in the layers thicknesses of the two quarter-wave stacks hardly affect the GDD of the mirror at all, while deviations of the spacer layer thickness from the theoretical design value merely result in a spectral shift of the mirror characteristics. This spectral shift can be simply determined from a transmittance measurement where the position of side bands of the highly-reflective region of the mirror can be used to estimate the error in the thickness of the resonator layer. The designed GDD curve and actual GDD curve derived from the theoretical and measured transmittance of the GTI-DMs are shown in Fig. 2.2. The GTI mirrors were designed to compress a DUV pulse with 5 nm FWHM to ~1.22x the transform limit. By tuning the angle of incidence it is possible to shift the optimal wavelength of GDD compensation although this will result in a change of the compression capabilities of the DMs. An increase in the AOI results in a blue shift of this curve, while decreasing the AOI causes a red shift (Fig. 2.3). The amount of GDD compensation at the optimal wavelength also varies depending on the polarization of 22 the incoming pulse. For p-polarized light the amount of GDD decreases with increasing AOI while for s-polarized light the GDD compensation increases. In our setup, the third harmonic out of the waveguide is s-polarized with respect to the DM. In the case of both s- and p-polarization, an increase in AOI results in a decrease in the reflectivity. The coating process was expected to have a ±1% error with respect to the optimal compensation wavelength which is well substantiated by Fig. 2.2 indicating a 2.5 nm shift between the theoretical and the reverse-engineered dispersion curves. Since a spectral interferogram from a broadband Michelson interferometer would be required to measure the actual GDD curves, a capability we do not currently have, the actual GDD properties of the DMs still have to be thoroughly tested by performing pulse compression on a DUV ultrafast pulse. FIG. 2.2. Dispersive mirror GDD curves designed (thick black) and manufactured (red) for optimal reflectivity at 7 o AOI. Typical 266.5 and 271.5 nm spectra (thin black) with 4.9 and 5.2 nm FWHM respectively produced from FWM in an argon- filled hollow core fiber. 23 To minimize the total amount of group delay dispersion requiring compensation from the addition of the necessary optical components in the beam path, all reflective surfaces are custom low-dispersion 268 nm dielectric mirrors and all focusing was done with custom low-dispersion curved 268 nm dielectric mirrors supplied by Femtolasers. Only two transmissive optics were used in the DUV optical path: the beam splitter in the interferometer consisted of a 1.6 mm thick CaF 2 window (CVI) with a UV anti-reflection coating on the second surface, and the back window of the fiber cell was a 1 mm piece of uncoated CaF 2 . Using the estimated dispersion characteristics of air, and all optical components in the DUV optical line (except the DM compensating components) the total estimated GDD and TOD at 266.5 nm are 865±50 fs 2 and 260±12 fs 3 , and at 271.5 nm, 914±50 fs 2 and 280±10 fs 3 respectively. The DUV path length was approximately 0.84 m longer in the latter setup accounting for the higher dispersion. The dispersion characteristics of the hollow core fiber were not included in this estimated value and these should depend on the fiber diameter and phase matching pressure 25 (GDD from the argon filled hollow core is ~30 fs 2 ). Previous work by Durfee et al. estimated that their DUV pulses emerge positively chirped from the end of the fiber; pulses with a transform limit of ~8 fs are stretched to 41 fs in a 70 cm long fiber with a 140 μm inner diameter. 11 Although there are differences in our fiber setup, we can still expect that significant accumulated phase arises from propagation inside the fiber. Starting with the GDD and TOD of the optical components while ignoring that incurred in the fiber therefore sets a lower limit on the expected number of sets of DM bounces 24 needed to compensate the GDD: that value is approximately 5 bounces per mirror (20 total reflections) based on operation at the designed geometry (7 o AOI). Since the layout of the DM compressor allows for variation of the number of bounces by multiples of 4 or ~200 fs 2 at 7 o AOI, a set of Suprasil wedges are used to fine tune the amount of dispersion by varying the thickness of the inserted substrate. The thickness of each wedge varies from 0.2-1 mm allowing for 74-388 fs 2 of variable dispersion at 268 nm. Overcompensation by one or more sets of DM bounces and adding dispersion from the wedges, results in optimum compression of the pulse. 255 260 265 270 275 280 -80 -60 -40 -20 0 20 40 60 80 7 o AOI 26 o AOI GDD, fs 2 Wavelength, nm FIG. 2.3. Dispersive mirror GDD curves for optimal 7 o AOI based on the measured transmission spectrum (black) and simulated curve at 26 o AOI (red) for s-polarized light. 25 2.3 Results Figure 2.4(a) shows the optimally compressed pulse centered at 266.5 nm after a total of 24 DM reflections and 1.03 mm of inserted wedge. The spatial/spectral homogeneity was excellent in the far field, verified by scanning a 50 μm pinhole through the 3 mm 1/e 2 vertical and horizontal beam axis several meters downstream of the compressor. The FWHM of the autocorrelation is 42 fs corresponding to a deconvoluted pulse FWHM of 30 fs (1.4 times transform limited) assuming Gaussian temporal pulses. A transform limited pulse width centered at 266.5 nm with 4.9 nm of bandwidth would be 21 fs (again assuming Gaussian spectral/temporal shape). Figure 2.4(b) shows that prism compression of a 267 nm pulse with similar spectral shape and bandwidth, leads to an increase in higher order dispersion (mostly TOD) due to the long transmission lengths through the CaF 2 prism substrate and the residual dispersion from the double pass configuration resulting in greater departure from the transform limit. 14, 18 It should be noted that prisms made from MgF 2 would result in less (~40% less compared to CaF 2 ) but still significant amount of accumulated TOD. The prism-compressed pulse significantly deviates from a perfect Gaussian at the 50% level with respect to the peak, and this wing structure could potentially overlap experimentally with fast transient signals from photoproducts in photochemical pump-probe experiments. 26 In contrast, the DM autocorrelation does not deviate significantly from Gaussian until below 10%. Optimal performance at 266.5 nm required rotation of the DM AOI to 37 o where the reflectivity per bounce was measured to be 96% (compare to reflectivity > 99% 26 verified at 7 o AOI). The large departure from the design AOI is responsible for the loss in reflectivity and also explains why the compression achieved is not closer to the transform limit. This performance will now be analyzed and discussed. FIG. 2.4. (a) Autocorrelation of 266.5 nm pulse after DM compression (black) and prism compression (red). The transform limit is calculated to be 21 fs from the corresponding spectral bandwidth assuming a Gaussian shape. (b) Detailed comparison on a log scale showing the deviation of the DM and prism compressed pulses from Gaussian (black dashed). To estimate the actual GDD in our set up and determine the effectiveness of each DM reflection, a pulse autocorrelation measurement was made without any bounces off of the DM pair to measure the true accumulated phase in an uncompressed pulse. For pulses centered at 266.5 nm, the uncompressed pulse FWHM was measured to be 199 fs after Gaussian deconvolution. The GDD is estimated by the following relationship, 1 2 ln 4 2 2 −         = TL o TL GDD τ τ τ 2.1 27 where TL τ corresponds to the transform limited pulse FWHM and o τ is the measured pulse FWHM. Although this formula assumes zero TOD and higher order dispersion, the amount of TOD needed to introduce significant error at this stage would need to exceed 1000 fs 3 . Based on the measured width, the estimated GDD of the system for the 266.5 nm pulse is ~1490 fs 2 which is nearly 650 fs 2 larger than that estimated for the optical layout of the system. The same measurement was made for the 271.5 nm pulses generated from the 35 fs amplifier system, resulting similarly in ~1390 fs 2 of estimated GDD. These values indicate that there is indeed significant GDD in the light emerging at the output of the hollow core fiber; 11 this additional dispersion would require an additional 3-4 sets of bounces according to the original DM design. The conditions that lead to the optimal pulse compression shown in Fig. 2.2 therefore correspond to ~68 fs 2 of GDD compensation per DM reflection. As highlighted in Fig. 2.3, the increase in the compensation per bounce with respect to the designed curve and decrease in reflectivity are as expected for s-polarized light as the AOI is increased. This is a result of the fact that the reflectance of the partial reflector placed on the top of the GTI-like dispersive mirror increases with the angle of incidence for s-polarized light, leading to an increased storage time of the incident radiation at the resonant wavelength which results in a more negative GDD minimum. At the same time scattering losses grow significantly as a consequence of the extended storage time in the resonant spacer layer. 28 With the reverse-engineered GDD curves in hand, a simple model based on Gaussian pulse propagation can be used to describe the dispersion properties of our optical apparatus as well as the compression characteristics of the DM pair as used. The 266.5 nm laser pulse generated by the hollow core fiber is represented by a transform limited Gaussian function t i t t o e e t E ω −       Δ − = 2 2 ln 2 ) ( 2.2 where o ω , is the center frequency of the pulse. Complex fast-Fourier transformation to the frequency domain is followed by multiplication with the measured GDD of 1490 fs 2 and estimated material TOD of 280 fs 3 representing all accumulated phase between generation and the autocorrelating medium: ( ) ( ) ( ) ( ) 3 3 2 2 3 4 2 ) ( o TOD o GDD i i in e e t E cfft E ω ω π φ ω ω π φ ω − − = 2.3 This results in our best estimate of the uncompressed DUV pulse, where GDD φ corresponds to the material GDD and TOD φ corresponds to the TOD. The DM GDD curve shown in Fig. 2.3 (26 o AOI) which displays a maximum negative GDD at 266.5 nm was used to calculate the phase compensation of the DMs at the optimal AOI for compression found experimentally (37 o ). This AOI discrepancy between the curves reverse-engineered from the transmission measurement and our experimental result highlights the challenges involved in the fabrication process of the mirrors, but a negative GDD per bounce near our measured 29 value of -68 fs 2 indicates that this a reasonable starting point for our DM simulation. The DM dispersion curve can be simply fit to two opposite sign Lorentzians which result in an R 2 value of 0.9999; the resulting function was integrated twice to reproduce the appropriate expression for the phase compensation of each reflection. Finally, the DM spectral phase is multiplied by the total number of reflections and this expression is combined with the starting input pulse, resulting in the compressed pulse, ( ) ( ) N i in out o DM e E E ⋅ − = ) ( ω ω φ ω ω 2.4 where DM φ is the phase compensated by an individual DM reflection and N is the total number of reflections. The pulse autocorrelation is computed in the frequency domain and, after inverse Fourier-transformation, compared with experiment. Figure 2.5(a) shows this comparison of the simulated autocorrelation after 26 bounces along with the experimental measurement. Although both pulses show very little intensity in the wings of the pulse, in the experimentally measured pulse they are somewhat more significant and the FWHM is not as short as for the simulated pulse. The higher order dispersion generated by the DMs, which is responsible for both of these features, is due to the fact that the GDD curve (Fig. 2.3) is not flat over the entire spectrum of the DUV pulse. This curvature, specifically the quadratic shape of the curve gives rise to significant fourth-order dispersion (FOD). Since this is less evident in the simulated pulse, it can be expected that the real DM GDD curve is somewhat steeper in this spectral region than the simulated curves predict. 30 Furthermore, the minimum of the compression curve (Fig. 2.5(b)) occurs at 26 bounces which is slightly more than expected accounting only for the measured GDD. This is due to the fact that some of the accumulated FOD can be approximately nulled with additional negative GDD, thus requiring a few extra reflections to achieve the best measured pulse compression. Experimentally, we must also be suppressing the broadening from FOD with extra reflections, meaning that, the minimum value in the actual GDD curve must be slightly more negative than our estimate of -68 fs 2 . FIG. 2.5. (a) Comparison of measured 266.5 nm pulse autocorrelation (black curve) after 24 bounces and 1.0 mm of Suprasil (corresponding to ~ 22 bounces) and 271.5 nm pulse autocorrelation (blue curve) after 24 bounces and 0.9 mm of wedge. Also shown is the simulated 266.5 nm pulse (red curve) compressed by 26 bounces off of the 26 o AOI DM dispersion curve of Figure 2.3. (b) Comparison of the measured deconvoluted pulse width as a function of the number of DM reflections for 266.5 nm pulses (experimental, black circles), 271.5 nm pulses (experimental, blue squares) and 266.5 nm pulses (simulated, red curve). Shifting the center wavelength to 271.5 nm should alleviate the need to reflect off of the DMs with such a high AOI and bring us closer to the minimum in the manufactured GDD curves. Figure 2.5(a) shows a comparison of the compressed 31 266.5 and 271.5 nm pulses. The optimized deconvoluted pulse width at 271.5 nm (24 bounces and 0.9 mm of inserted wedge) was found to be 34 fs (1.6 times transform limited) with an optimal AOI of 24.5 o corresponding to 97% reflectivity per bounce. While the AOI is substantially decreased and reflectivity increased as expected, it is still significantly larger than the predicted AOI from the transmission measurement of the manufactured DMs. It can also be seen from Fig. 2.5(b) that maximum pulse compression also occurs between 20 and 24 total reflections indicating that the DMs are compensating for approximately 63 fs 2 of GDD per bounce which is also larger than the designed compensation of 50 fs 2 per reflection as expected for the high AOI. The 271.5 nm pulse appears to have more high order dispersion than the 266.5 nm pulse, indicating that the GDD minimum is slightly shifted away from the center frequency of the light leading to greater fourth order dispersion. In this case, a more thorough scan of the AOI space would probably alleviate some residual dispersion. Although the precise phase structure in the pulse is sensitive to AOI and central wavelength, experimentally, once the optimum AOI is reached, as long as the center frequency and bandwidth of the pulse remain stable, the DM compressor system requires very little day to day adjustment. It should be noted that although we have concentrated here on performance of the DMs under carefully controlled conditions to test the compressor properties, sub-30 fs pulses have been obtained for pulses at 271.5 nm when 7 nm of DUV bandwidth is employed, but, as expected, they have greater high-order dispersion. 32 In conclusion, we have demonstrated that despite the challenges of working with dispersive multilayer coatings at deep UV wavelengths, a GTI mirror compressor designed to support ~5 nm of bandwidth can successfully compress DUV pulses to 30 fs. At ~1.5x the transform limit, these pulses have less higher order dispersion when the DMs are properly optimized compared to similarly generated pulses compressed by a prism pair. The performance of these DMs, while encouraging, highlights the difficulties in achieving compression and pulse characteristics in the DUV comparable to those routinely accomplished in the visible and NIR. This is a consequence of the fact that the sensitivity of dispersive mirror designs to layer thickness manufacturing errors increases by a factor of approximately 3 in the DUV range as compared to the NIR. Although difficulties in the manufacturing tolerances lead to a significant deviation away from the optimal working AOI, even considering the increased losses with this prototype set of DUV DMs the ~45% overall transmission efficiency puts this system on par with grating compressors. It is reasonable to expect that revision in design based on the current characterization should allow for 80% overall transmission with a broader GDD minimum. Furthermore, improved manufacturing accuracy along with more complex designs based for instance on two-cavity GTI structures might enable increasing the effective GDD compensation bandwidth by a factor of two. This bandwidth would correspond to bandwidth-limited pulse durations closely approaching 10 fs. Along with this, the absence of spatial/spectral dispersion problems associated with typical compression methods and excellent transmitted wavefront, makes this an attractive 33 new method for DUV pulse compression. Although not thoroughly tested in this case, we see no damage of the DM coating with the pulse powers used. The potential for high damage thresholds offered by dielectric mirrors should allow for compression of the several μJs needed for pump-probe and other applications without sustaining optical damage over time, as is typical for prism methods in the DUV. With less residual dispersion, the improved temporal characteristics of broad bandwidth DUV pulses will effectively widen the window of observation available for time resolved studies, as well as provide suitable pulses for two-dimensional and other non-linear spectroscopies. 27, 28 2.4 Acknowledgements The work at USC is supported by the National Science Foundation under grant CHE-0617060. Christopher Rivera was supported by a Ford Foundation Graduate Fellowship. We thank Andreas Isemann for his help in starting the development of the DM system and Christopher Elles for assistance in implementation as well as comments on this manuscript. 34 2.5 References for Chapter 2 1. M. Yamashita, M. Ishikawa, K. Torizuka and T. Sato, Opt. Lett., 1986, 11, 504-506. 2. P. Dombi, P. Racz, M. Lenner, V. Pervak and F. Krausz, Opt. Express, 2009, 17, 20598-20604. 3. X. Chen, L. Canova, A. Malvache, A. Jullien, R. Lopez-Martens, C. Durfee, D. Papadopoulos and F. Druon, Appl. Phys. B, 2010, 99, 149-157. 4. C. F. Dutin, A. Dubrouil, S. Petit, E. Mevel, E. Constant and D. Descamps, Opt. Lett., 2010, 35, 253-255. 5. R. Szipocs, K. Ferencz, C. Spielmann and F. Krausz, Opt. Lett., 1994, 19, 201-203. 6. I. Matsuda, K. Misawa and R. Lang, Opt. Comm., 2004, 239, 181-186. 7. V. Pervak, F. Krausz and A. Apolonski, Opt. Lett., 2007, 32, 1183-1185. 8. Y. Kida, S. Zaitsu and T. Imasaka, Opt. Express, 2008, 16, 13492-13498. 9. A. S. Morlens, P. Balcou, P. Zeitoun, C. Valentin, V. Laude and S. Kazamias, Opt. Lett., 2005, 30, 1554-1556. 10. C. G. Durfee, S. Backus, M. M. Murnane and H. C. Kapteyn, Opt. Lett., 1997, 22, 1565-1567. 11. C. G. Durfee, S. Backus, H. C. Kapteyn and M. M. Murnane, Opt. Lett., 1999, 24, 697-699. 12. N. Krebs, R. A. Probst and E. Riedle, Opt. Express, 2010, 18, 6164-6171. 13. K. Kosma, S. A. Trushin, W. E. Schmid and W. Fuss, Opt. Lett., 2008, 33, 723-725. 14. S. A. Trushin, W. Fuss, K. Kosma and W. E. Schmid, Appl. Phys. B, 2006, 85, 1-5. 15. S. A. Trushin, K. Kosma, W. Fuss and W. E. Schmid, Opt. Lett., 2007, 32, 2432-2434. 35 16. M. Beutler, M. Ghotbi, F. Noack, D. Brida, C. Manzoni and G. Cerullo, Opt. Lett., 2009, 34, 710-712. 17. I. Walmsley, L. Waxer and C. Dorrer, Rev. Sci. Instrum., 2001, 72, 1-29. 18. A. E. Jailaubekov and S. E. Bradforth, Appl. Phys. Lett., 2005, 87, 021107. 19. C. H. B. Cruz, P. C. Becker, R. L. Fork and C. V. Shank, Opt. Lett., 1988, 13, 123-125. 20. B. J. Pearson and T. C. Weinacht, Opt. Express, 2007, 15, 4385-4388. 21. C. H. Tseng, S. Matsika and T. C. Weinacht, Opt. Express, 2009, 17, 18788- 18793. 22. G. Steinmeyer, Appl. Phys. A, 2004, 79, 1663-1671. 23. G. Steinmeyer, Appl. Optics, 2006, 45, 1484-1490. 24. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum., 2003, 74, 4958-4960. 25. A. W. Snyder and J. D. Love, Optical Waveguide Theory, Chapman and Hall, New York, 1983. 26. A. C. Moskun, A. E. Jailaubekov, S. E. Bradforth, G. H. Tao and R. M. Stratt, Science, 2006, 311, 1907-1911. 27. Z. Y. Li, D. Abramavicius, W. Zhuang and S. Mukamel, Chem. Phys., 2007, 341, 29-36. 28. D. Abramavicius, J. Jiang, B. M. Bulheller, J. D. Hirst and S. Mukamel, J. Am. Chem. Soc., 2010, 132, 7769-7775. 36 Chapter 3. Revealing I( 2 P 1/2 ) radicals in ethanol: Evidence of a new detachment pathway from 200 nm photoexcitation of I - and liquid photoelectron spectroscopy In collaboration with: Bernd Winter Max-Born-Institut für Nichtlineare Optik und Kurzzeitspektroskopie, Max-Born-Str. 2A D-12489 Berlin, Germany Niklas Ottoson Department of Physics and Astronomy, Uppsala University, Box 516, SE-751 20 Uppsala, Sweden Abstract for Chapter 3 Pump-probe spectroscopy was used to photodetach electrons from the second charge-transfer-to-solvent excited state of iodide in water and ethanol. Liquid-jet photoelectron spectroscopy was employed to rationalize the observed transient species. Photodetachment in ethanol revealed a signature which has been assigned as coming from the charge-transfer absorption of neutral iodine radicals in their spin- orbit excited state, I*( 2 P 1/2 ), which had previously never been observed from photodetachment in any polar-protic solvent. We confirm that a similar signature is not observed in water. The lower vertical detachment energies corresponding to I and I* cores obtained in the photoemission experiments reveals a possible coupling 37 between the second CTTS band of iodide and both detachment continua, resulting in competing autodetachment and direct detachment pathways. We use simple kinetic models to extract information on the fate of all transient features and derive lifetimes and possible decay pathways for the newly observed I* charge transfer species in ethanol. 3.1 Introduction The photodetachment of iodide in solution has long been used as a means to generate and study solvated electrons in solution. 1-11 The production of a neutral radical and a free solvated electron by excitation into low lying charge-transfer-to- solvent (CTTS) states is advantageous since typically these bands are accessible by UV laser light and, in the case of water and alcohols, produce prompt electrons with high quantum yields. 9 While the solvated electrons have been the subject of intense study, recent work has examined the detachment mechanism itself, both in small cluster and in the bulk liquid. 4, 7, 12-15 The autodetachment pathway in solution is analogous to that observed from dipole-bound states below the corresponding ionization thresholds in iodide-CH 3 CN clusters and FeO¯ photodetachment. 15, 16 Even more recent, time-resolved photoelectron spectroscopy (TREPS) studies of clusters and the bulk and even surface SHG show that electron ejection in liquids and interfaces both display similar detachment signatures. 11, 13, 14, 17 In our group we have more recently been interested in the creation and reaction of the neutral radical species formed and what this can reveal about the dynamics of detachment at higher 38 energies. 12, 18, 19 While the spectrum and diffusional reaction dynamics of the neutral iodine radical is well known in water and most alcohols from previous flash photolysis studies, 20-23 the formation and subsequent relaxation of this species is poorly understood. Furthermore, the spin-orbit excited radical (I*) which is produced in gas phase photolysis experiments, has never been observed in water or alcohols from photodetachment experiments. Our group’s previous attempt to produce I* in water by excitation of the second CTTS band of iodide, presumed to arise from an I*-core CTTS band, yielded no spectroscopic signature of the spin- orbit excited form. 18 Three possible deactivation pathways of the I* species was therefore postulated: i) I* and e¯ undergo very fast recombination; ii) I* is deactivated by the solvent, possibly enhanced by the nearby electron; iii) excitation into the second CTTS band does not actually yield iodine in its spin-orbit excited state as a product because of non-adiabatic decay pathways. It was concluded that the last was the most likely scenario, supported by the liquid photoelectron spectrum of iodide in water. 12 Autodetachment, or the ejection of an electron via a non- adiabatic change of electronic surface, is taking place due to coupling of the I* CTTS electronic state to the electron detachment continuum corresponding to ground state I radicals. To our knowledge, the only known claim of the I* radical spectrum in water has been produced from ICN photodissociation experiments, 24 although there is still some confusion as to the assignments of transients in this spectral range. 18 Our own ICN photodissociation experiments provide the major motivation for the present study. Recent publications concentrating on the rotational dynamics 39 of the CN radical stemming from photodissociation in water and alcohols, 25, 26 along with the recent development of a broadband time-resolved spectrometer, 19, 27 has allowed us to produce time-resolved transient spectra from ICN photodissociation. Our goal is to definitively assign transient species and observe their individual time- dependent evolution. As pointed out previously by Larsen et al., the spectral signatures of the transient species are relatively broad and span nearly the entire visible and UV spectrum and are expected to significantly overlap with each other, making it difficult to separate and assign the transient features observed. 24 As we plan to focus on ICN in water and ethanol in upcoming experiments, it makes sense to examine the iodide photodetachment experiments in these solvents in the hopes of obtaining iodine radical transient spectra, especially since there have been no recent studies using ethanol solutions with this motivation. Also, we can simultaneously explore whether or not the ultrafast autodetachment pathway is common in different solvents and how this depends on the specific energetic of the detachment process. In ethanol, excitation into the first CTTS band of iodide results in final trapping of a solvated electron after detachment occurring on a timescale which is approximately an order of magnitude slower than in water. 9 To our knowledge, no study has shown that the same holds for excitation of the second CTTS transition of iodide in ethanol. If this is indeed the case, it is possible that the persistence of the electron-radical contact pair will allow for a longer window of observation of I* radicals before undergoing deactivation through autodetachment or relaxation. Along with this, one further practical motivation comes from the fact that the known radical CT band of 40 iodine in ethanol is red shifted to ~360nm compared to 260 nm in water, putting it well within our visible probing capabilities. 23, 28 Figure 3.1a and 3.1b show the ground state absorption spectra for iodide in water and ethanol respectively. The first CTTS band peaks at 226 nm in water and 220 nm in ethanol. The second CTTS band which has been resolved in water peaks at 195 nm, 29 while in ethanol we estimate this band peaks at ~190 nm based on fitting the red side of the absorption spectrum to a Gaussian. The predicted absorption maximum is in good agreement with the reported deep-UV spectra for a series of alcohols, from which ethanol is a surprisingly absent member. 30 The splitting of ~0.86 eV between the CTTS bands in water and all of the alcohols is very close to the spin-orbit splitting observed for iodine radicals in vacuum 31 . Thus, these first two CTTS transitions are expected to be associated with upper states that have a diffuse 6s type electron 32 associated with the neutral radical cores, I( 2 P 3/2 ) and I*( 2 P 1/2 ) respectively. 29 In this paper we follow the transient species generated by pumping the second CTTS transition of iodide in water and ethanol and probing these transients simultaneously at all wavelengths across the visible and UV spectrum (320 to 625 nm). We will demonstrate that there is a signature arising in ethanol from I*( 2 P 1/2 ) absorption, which is observed here for the first time in any polar protic solvent. By observing the dynamics of the I and I* populations along with the measurements of vertical detachment energetics to vacuum using liquid photoelectron spectroscopy we are able to examine the detachment pathways, which 41 have been suggested previously to account for the absence of I* in water, in greater detail. FIG. 3.1. (a) Iodide absorption spectrum in water (red) and possible transients from photodetachment including solvated e - (black), I( 2 P 3/2 ) (blue), and the predicted spectrum of I( 2 P 1/2 ) (green). (b) Identical spectra for photodetachment in ethanol. The red shift of the I( 2 P 3/2 ) peaks is predicted by the gas phase ionization potential (ref. 20) and the I( 2 P 1/2 ) transition is shifted by the gas phase splitting of 0.94 eV. Colored stripes indicate the probe wavelengths used as described in the text. 6 5 4 3 2 0 5000 10000 15000 pum p (200 nm ) Probe 2 Probe 1 e - in ethanol I - in ethanol I*( 2 P 1/2 )x5 I( 2 P 3/2 )x5 (b) ethanol ε ε ε ε, M -1 cm -1 Energy, eV 0 5000 10000 15000 20000 200 300 400 500 600 λ λ λ λ, nm ε ε ε ε, M -1 cm -1 pum p (200 nm ) I - (aq) I( 2 P 3/2 )x10 I*( 2 P 1/2 )x10 e - (aq) Probe 2 Probe 1 (a) H 2 O 42 3.2 Background: Spectroscopy of I - and transient species Excitation of the second CTTS band of iodide is achieved by 200 nm excitation. Figure 3.1 shows that, in both solvents, this wavelength primarily overlaps with the second CTTS band but will also excite the tail of the lowest CTTS band; this effect is slightly larger in ethanol. While excitation into the second CTTS state leads to an I* core ground state, I was observed within the time resolution of a previous 200 nm experiment in water. We therefore anticipate the possibility to observe three transient species within our visible probing window. The spectra of I( 2 P 3/2 ), I*( 2 P 1/2 ), and e¯ in solutions of water and ethanol are also shown in Figure 3.1a and 3.1b respectively. The I¯ absorption spectrum lies much lower in energy than our probing window which cuts off below 320 nm, so no bleach signal or bleach recovery is expected to be observed in our experiment. The most dominant feature in our probing window is that of the solvated electron, which is expected to have an extinction coefficient one order of magnitude greater than that of the iodine radicals in water and five times greater in ethanol. 28, 33, 34 The thermalized solvated electron absorption spectrum in water and ethanol has been well characterized by Jou and Freeman and a significant revision to the aqueous electron extinction coefficient was recently reported by Hare et. al. 33, 34 The band shape is described by a Lorentzian on the high energy side of the spectrum and a Gaussian on low energy side. The spectrum at 298 K stretches across the probing window peaking at 720 nm in water and 690 nm in ethanol. Many past studies have shown that electron ejection results in a red-shifted absorption for the newly trapped 43 electron which then undergoes relaxation to the equilibrated spectrum, 9, 35, 36 although alternative assignment of this shifting process have been suggested. 37, 38 Whatever the exact mechanism, this process results in a 0.36 eV shift with an 850 fs characteristic time after excitation of the first CTTS band in water. Excitation of the second CTTS state in water results in qualitatively similar results although the shift and characteristic time is smaller by ~40% and 20% respectively. 18 In ethanol, excitation in the first CTTS band produces electrons with a 0.18 eV shift and 17 ps characteristic relaxation time. While this relaxation time agrees with previous time scales for electron solvation, 39 the model employed to describe the electron signal was not sufficient to examine dynamics within the first 5 ps. 9 In solution, I( 2 P 3/2 ) has a broad absorption band corresponding to a charge- transfer interaction with the solvent where an adjacent solvent molecule is considered to be the electron donor and the iodine radical the charge acceptor, 40 which notably, is opposite to the direction of CTTS. Because a solvent molecule is the electron donor, the center frequency of the CT band depends on the ionization potential of the solvent. Such halogen solvent CT bands are documented for a number of solvents. 28, 30 As water has a relatively high ionization potential, the CT band is relatively deep into the UV (260 nm) while in ethanol the CT band is at ~360 nm, although this absorption band has never been reported devoid of convolution with I 2 ¯. The maximum extinction coefficients of I( 2 P 3/2 ) in water and ethanol are 1040 and 1900 M¯ 1 cm¯ 1 respectively. The I( 2 P 3/2 ) bandwidth in water is ~1.2 eV while in ethanol, the bandwidth has not been reported to our knowledge. The 44 bandwidth of the I( 2 P 3/2 ) in ethanol was estimated to be ~0.8 eV from the spectral series reported in water, methanol and isopropyl alcohol. 21 Figure 3.1 (blue curves) shows the expected I( 2 P 3/2 ) CT bands in water and ethanol respectively. The I*( 2 P 1/2 ) species is assumed also to give rise to CT absorptions, but clearly assigned bands for these species have not been reported and their transition energies have therefore only been predicted on the basis of Mulliken’s theory of charge-transfer absorption. 18, 24, 40 Assuming that the spin-orbit states of iodine are identical except for the increased electron affinity of I* due to the spin-orbit separation. The gas-phase splitting of the neutral spin-orbit states is 0.94 eV, 31 and this is used to estimate the splitting of I and I* in water and ethanol. Of course, deviation from the gas phase separation may be observed, as seen in the CTTS splitting (~0.91) which has been attributed to slight differences in the heat of solution between 2 P 1/2 and 2 P 3/2 states by Jortner. 29 Figure 3.1 (green curves) shows the I*( 2 P 1/2 ) CT bands which are predicted to peak at 315 and 500 nm in water and ethanol respectively. In the absence of additional information, the same extinction coefficients and bandwidths of the I( 2 P 3/2 ) bands are assumed. 3.3 Experimetnal 3.3.1 Pump-probe setup The transient absorption photodetachment results presented in this paper were acquired using two similar experimental setups at USC but at different times. Two laser systems were used: (a) a 1 W, 110 fs, 1 kHz Ti:S regenerative amplifier (Spectra Physics Hurricane) and (b) a 3.5 W, 35 fs, 1 kHz Ti:S regenerative amplifier 45 (Coherent Legend Elite USP-HE). In both cases, the fourth-harmonic light was created by gently focusing a 500 μJ portion of 800 nm into a 100 μm thick type-I BBO to produce 40 μJ of 400 nm. The 400 nm was then mixed with the residual 800 nm in a 100 μm type-II BBO to generate 8 μJ of 266 nm. Finally, 266 nm light was mixed with another 50 μJ portion of residual 800 nm in a 150 μm (setup (a)) and 80μm (setup (b)) type-I BBO to generate 2 μJ of 200 nm (6.2 eV). A pair of 200 nm dielectric mirrors (CVI) was used to steer the pump beam to the sample where the light was focused by a 25 cm MgF 2 lens to an ~300 μm waist. The pump light was chopped at 500 Hz. Probe light was generated by focusing a small portion of the fundamental 800 nm into a 2 mm thick piece of CaF 2 to generate super-continuum spanning 320-425 nm (setup (a), Probe 1) and 425-625 nm (setup (b), Probe 2). Both setups generate super-continuum spanning 310-960 nm but due to slight differences in the experimental conditions (800 nm stability, focusing conditions, super-continuum intensity), the intensity of the UV portion generated in setup (a) was more stable. For wavelengths above 625 nm only the relative transient signals were used due to slight deviations in pump-probe overlap which would require additional adjustment in order to provide an absolute scale. Two aluminum off-axis parabolic mirrors were used to focus the super continuum probe to a ~75 μm diameter waist at the sample. After the sample the super-continuum was dispersed onto a 256 channel diode array with a resolution of ~1.25 nm per pixel. A UV diffraction grating (300 gr/mm, 300 nm blaze, Oriel) was used to disperse probe wavelengths from 320-550 46 nm and a visible diffraction grating (300 gr/mm, 500 nm blaze) was used to disperse wavelengths >550 nm. Transient dynamics were measured to 20 ps by delaying probe pulses via a computer-controlled motorized translation stage. The ΔA signal collected at each pixel simultaneously for each time delay is defined as         − = Δ off pump on pump T T A _ _ log 3.1 where T corresponds to the relative transmitted light with and without the pump pulse. ΔA is reported in the familiar units of mOD. The time resolution of the spectrometer was estimated from cross-correlation measurements of the 2-photon absorption (pump+probe) signal from water and ethanol to be ~200 fs (setup (a)) and ~150 fs (setup (b)). 27 The difference between the two setups is a result of the thinner sum-frequency crystal used to generate 200 nm, resulting in a slightly shorter pulse duration. A time-dependent chirp of the super-continuum is expected 27 and all data were calibrated in the time-domain such that the two photon absorption signal from neat solvent is all aligned at zero delay. Neat solvent background scans were taken to ensure that the pump intensity was small enough such that no electrons were generated by 2-photon absorption (pump+pump) and ionization of the solvent. The samples consisted of 20 mM potassium iodide (99.3%, Mallinckrodt) in water and ethanol, respectively. The water used was either distilled (Arrowhead) or ultra- purified (14 MΩ·cm, Millipore), and the ethanol was 200 proof, anhydrous (PHARMCO-AAPER). The sample was delivered to be irradiated by the laser in a thin film gravity drop jet. 41 The liquid film was approximately 100 μm thick. 47 3.3.2 Photoelectron setup The liquid jet photoelectron (PE) spectroscopy setup at the BESSY-II synchrotron facility in Berlin has been described previously. 42 For collection of the iodide photoelectron spectrum, synchrotron radiation from the BESSY II U-125 undulator beamline was used as the ionizing radiation and, for the valence band photoemission spectra reported here, the photon energy was set to 100 eV. This experimental setup allows for collection of photoelectrons in the range of 0-100 eV. The total experimental resolution was ~0.3 eV, estimated from the width of the individual vibrational components of the gas phase water 1b 1 photoline, recorded the same day. A 1 M sodium iodide in ethanol solution was flowed at approximately 20 bar backpressure through a 20 μm diameter glass nozzle to form a liquid jet in vacuum at a velocity of approximately 100 m/s. The liquid is cooled to 4 o C prior to flowing into the vacuum chamber. The X-ray spot size at this beam line is 100x150 μm 2 (much larger than the diameter of the jet), and thus the spectrum necessarily contains significant contributions from both evaporated gas and liquid phase alcohol. As will be described below, in order to extract the iodide electron binding energies in ethanol, the PE spectrum must first be calibrated using literature values for either the gas or liquid ethanol peaks and then the solvent background must be removed. Because a spectrum of the pure liquid was not taken at the time of the iodide experimental run, a photoelectron spectrum of the pure liquid was recorded in a later experiment using the BESSY II U41-PGM beamline with a photon energy of 200 eV. In this case the exact same apparatus was used to deliver the ethanol sample 48 except the beamline has a considerably smaller X-ray spot size (23x12 μm 2 ) allowing just the photoelectron features of liquid ethanol to be observed. 3.4 Results The results from photodetachment of iodide in water and ethanol will be presented followed by the liquid-jet photoemission data for iodide solutions. For the ultrafast measurements, we will describe time dependent spectral features but leave the rigorous assignment of these features to the discussion. Figure 3.2a and c show the entire 20 ps contoured transient absorption data sets for 20 mM KI in water and ethanol. On first view, the large absorption signal of the solvated electron dominates the transient absorption spectrum in both the ethanol and water experiments. While little to no transient absorption is observable on the tail of the solvated electron absorption in water, in ethanol, the tail extends much farther into the UV side of the visible spectrum. Despite the small shift to the blue for the relaxed electron in ethanol, this absorption cannot be attributed purely to the solvated electron and we expect that this comes from another transient species such as I or I*, which indeed should absorb in the visible region in ethanol. 3.4.1 Water The strong transient signal due to solvated electrons is observed to rise and spectrally shift over the first 2 ps and then decay over tens of picoseconds. There are no obvious transient species observed over these times scales which can be attributed to absorption from either I or I*, which would be observed as tails on the blue side of 49 the solvated electron spectrum. Removal of the strong solvated electron signal is necessary in order to look for the spectrum of any weakly absorbing transients which might overlap with the absorption features of the solvated electron. In order to do so, a spectrally shifting absorption band 9, 36, 43 was subtracted from the data which accounts both for the thermalization and population dynamics expected for the solvated electron product. As in our previous analyses, we assume that the shape of the electron spectrum remains constant (the so called “shape stability” 34, 44 ) while the spectrum blue shifts 0.21 eV with a relaxation time of 850 fs. The time-dependant population dynamics of the solvated electrons were described by a simple competitive kinetics model that assumes a contact pair. 45 We have described more detailed physical models for the recombination dynamics in earlier work, 9 and there has been discussion in the literature about whether there are multiple types of pair intermediates for these reactions. 37, 43 However, for the purposes of this paper a simple kinetic model is sufficient to provide an adequate mathematical description. Such a model requires three parameters: a contact pair formation rate, k p ; a solvated electron escape rate, k d ; and a geminate recombination rate back to reform I - ; k n . Nearly the same parameters could be used here as for earlier work on 200 nm photodetachment: k p ¯ 1 = 0.38 ps, k n ¯ 1 = 42 ps, and k d ¯ 1 = 32 ps. This model provides a good fit to the solvated electron spectrum at all probe wavelengths and delay times of the experiment. Figure 3.2b shows the transient data remaining after the electron is subtracted. Transient spectra for several delay times are shown in Figure 3.3a along with the solvated electron signal that was subtracted. 50 FIG. 3.2. (a) Contour plot of the transient photoproducts from photodetachment in H 2 O, absorption in mOD increases from blue to red. (b) Contour plot after solvated electron is subtracted as described in the text. (c) Contour plot of the transient photoproducts in ethanol. (d) Remaining transient photoproducts after solvated electron subtraction in ethanol. Figure 3.3c shows the resulting transient spectra after removal of the electron signal at several delay times. No significant transient signals are observed between 350 and 500 nm consistent with previous experimental measurements in water. Only a very small (<0.2mOD) transient signal can be seen to absorb at probe wavelengths greater than 500 nm, which we take to be imperfect fitting of the electron solvation 350 400 450 500 550 600 0 2 4 6 8 10 12 14 16 18 20 (a) Probe Wavelength, nm Probe Delay, ps 0 2 4 6 8 10 12 350 400 450 500 550 600 0 2 4 6 8 10 12 14 16 18 20 (b) Probe Wavelength, nm Probe Delay, ps -0.2 1 2 350 400 450 500 550 600 0 2 4 6 8 10 12 14 16 18 20 (c) Probe Wavelength, nm Probe Delay, ps 0 1 2 3 4 5 6 7 8 350 400 450 500 550 600 0 2 4 6 8 10 12 14 16 18 20 (d) Probe Wavelength, nm Probe Delay, ps 0 1 2 51 dynamics. Below 350 nm there appears to be a tail of an absorption band peaking further into the UV than 320 nm (the shortest wavelength measured here) and this spectral feature decays on roughly the same timescale as the solvated electron. A very weak tail which extends to ~450 nm (Fig. 3.3c 600 fs) and decays very quickly over the first 500 fs is similar to small residual absorption measured in previous experiments. 18 FIG. 3.3 Several time cuts showing the transient absorption spectrum and the corresponding electron signal that was subtracted at each time point for (a) water and (b) ethanol. The resulting transient spectra after electron subtraction are shown for (c) water and (d) ethanol. 350 400 450 500 550 600 0 2 4 6 8 10 12 (a) 20 ps 10 ps 5 ps 3 ps 1 ps 600 fs Δ Δ Δ ΔmOD Probe Wavelength, nm 350 400 450 500 550 600 0 1 2 3 4 5 6 7 8 9 Probe Wavelength, nm Δ Δ Δ ΔmOD (b) 20 ps 10 ps 5 ps 3 ps 1 ps 600 fs 350 400 450 500 550 600 0 1 2 Probe Wavelength, nm Δ Δ Δ ΔmOD (d) 20 ps 10 ps 5 ps 3 ps 1 ps 600 fs 350 400 450 500 550 600 0 1 2 Probe Wavelength, nm Δ Δ Δ ΔmOD (c) 20 ps 10 ps 5 ps 3 ps 1 ps 600 fs 52 3.4.2 Ethanol Similarly to water, the strong solvated electron absorption overshadows any other transient signals when iodide is photodetached in ethanol (Fig 3.2c) although the transient signal on the blue side of the spectrum clearly indicates that electrons do not account for all of the signal. 0 5 10 15 20 2 4 6 8 10 920 nm 750 nm Δ Δ Δ ΔA, mOD Probe Delay, ps FIG. 3.4. Solvated electron data for 750 nm (black, solid) and 920 nm (red, solid) and electron fit (dashed) from the model described in the text. The solvated electron contribution in ethanol was fitted with a similar method to water. Previous work using a single exponential to describe the evolution of electrons generated by 225 nm excitation of iodide in ethanol did not adequately describe the behavior below 5 ps. Because, we would expect that electrons in our 53 probing window would be overlapped with at least the ground state iodine radical and possibly the I* transient species we need to use a probe wavelength which is dominated by the solvated electron. We therefore recorded additional data between 715 and 920 nm from the white light super-continuum and used this spectral region alone to simultaneously fit the solvated electron dynamics. The fit to these two wavelengths is shown in Fig. 3.4. At these wavelengths we expect that the solvated electron signal will be at least one order of magnitude greater than the signal from I or I* based on the predicted spectrum. Over the 20 ps time scale of the experiment, no obvious recombination is observed from the 750 nm signal which is consistent with previous studies, reporting no cage recombination (k n ¯ 1 ) occurring over a 1 ns time window. 9 The decay observed at 920 nm can be attributed almost entirely to the shifting of the electron spectrum. Since spectral shifting described by a single exponential does not provide an adequate simultaneous description of both the early and long time electron dynamics, we have allowed the time dependant thermalization to be described with a double exponential function. The shape of the electron spectrum, which is well characterized in ethanol, 34 was held constant. The relaxation times (and relative weights), which provide a good fit to both wavelengths, are 1.9 ps (67%) and 17 ps (33%), with a total shift of the solvated electron of 0.83 eV. The average electron solvation time is 6.9 ps which is lower than other values reported in the literature which vary from 11-26 ps, 9, 46-48 but this agrees well with the predicted electron solvation times calculated from a stepwise model, only depending on solvent timescales for reorientation and longitudinal dielectric relaxation. 39 The 54 significant increase in the spectral shift compared to the previous study at 225 nm is surprising and suggests that the electron ejection mechanism may be different at 200 nm excitation. 9 The contact pair formation rate (k p ¯ 1 ) was found to be ~200 fs which is approximately the time resolution of the instrument. The solvated electron signal was scaled to the transient spectra such that the maximum signal was removed. Figure 3.3b shows the transient spectra at several time cuts between 600 fs and 20 ps and the amount of solvated electron signal that was removed. After removal of the solvated electron, the remaining transient signal observed (Fig. 3.2d) peaks near 370 nm and a broad shoulder in the visible can be observed decaying over the first 10 ps. Figure 3.3d shows the transient spectra at several time cuts with the solvated electron contribution removed. Assuming that all of the transient absorption signal remaining at 20 ps can be attributed to I( 2 P 3/2 ) and using the known extinction coefficient for e¯ aq in ethanol, 34 we can determine the molar extinction of the radical CT band to be ~3000±600 M¯ 1 cm¯ 1 with a FWHM of 0.8 eV assuming a Gaussian shape. We estimate a ~20% error in ΔOD for transient signals above 3 eV (<400 nm) and below 2 eV (625 nm) in water and ethanol due to variations in pump-probe overlap in the sample at the extremities of the continuum. This value is derived from comparing the shape of the water 2-photon absorption (2PA) from background scans to the known 2PA spectrum of water. 27 This peak extinction is approximately a factor of two greater than that reported previously, but as stated earlier, the I atom band was overlapped with absorption from I 2 ¯ in these earlier measurements which might account for some of the discrepancy. 28 The 55 bandwidth however, is in very good agreement with I( 2 P 3/2 ) spectra reported in the series of alcohols. 22 Assignment of the shoulder at longer wavelengths, and its subsequent decay, will be thoroughly presented in the discussion section. 11 10 9 8 7 6 5 0 5 10 15 20 25 I( 2 P 3/2 ) I( 2 P 1/2 ) 3a'' (liq) Iodide (5p) 3a'' (gas) 11 10 9 8 7 6 0.0 0.5 1.0 PE Signal Electron Binding Energy (eV) PE Signal Electron Binding Energy, eV FIG. 3.5. The experimental PE spectrum (100 eV photon energy) arising from a 1M NaI/ethanol solution (black line) which contains contributions from both liquid and gas phase ethanol and iodide. Pure liquid ethanol (red line) was recorded with 200 eV photon energy in a separate experiment. Details of the transformation from electron kinetic energy to ionization potential are given in the text. The inset shows the difference spectrum which contains only iodide and gas phase ethanol contributions. 3.4.3 Photoelectron spectrum Figure 3.5 (black curve) shows the experimental photoelectron spectrum of a 1 M sodium iodide solution in ethanol. There are three distinct peaks at low electron 56 binding energy. The first, which is much reduced in spectra of liquid ethanol (shown in red) can be readily identified as ionization from iodide 5p by analogy with the 5p energy in water. 49 The sharpest feature near 11 eV corresponds to the gas phase 3a" IP of ethanol, the broader feature near 9 eV corresponds to the 3a" IP of liquid ethanol (as seen in the red curve). Recall, as explained in the experimental section, the gas phase peak is absent in the red curve as the liquid spectrum is recorded with a smaller X-ray spot. As will be shown later in this section, subtraction of a the PE spectrum of liquid ethanol from that of the iodide in ethanol reveals a shoulder at lower binding energy, between the 7.5 and 9 eV peaks, which arises due to the ionization of iodide. Calibrating the absolute binding energy scale of the PE spectrum in Figure 3.5 is not straightforward. While the IP of the 3a" HOMO of ethanol has been reported previously for both liquid 50 and vapor, 50, 51 significant charging effects on the pure liquid due to the ionizing radiation, as well as streaming potentials, are known to cause the gas and liquid phase photoelectron peaks to shift relative to each other. 42 This splitting (ΔE g-l ) was reported in Faubel et al.’s pioneering work as 0.97 eV using a helium lamp, putting the lowest liquid feature at 9.66 eV but a more recent study from the Göttingen group using high harmonic pulses in the XUV reports ΔE g-l = 1.33 eV and the lowest liquid peak is thus 9.3 eV. 52 In our data (Fig. 3.5) ΔE g-l = 1.23 eV in relatively good agreement with the newer XUV measurements. In the current work, the high ion concentration helps to reduce surface charging in our experiment. 42 We have chosen to align the sharpest feature 57 of the experimental spectrum at 10.64 eV BE, i.e. at the IP for gas-phase ethanol, which has been very accurately determined in the literature. 51 Given that the residual charging of the liquid is small in our experiments, this procedure provides us with a calibrated BE scale for the liquid photoemission features against vacuum. Furthermore, a separate experiment done on a 1 M NaI in a 50/50 water to ethanol mixture under the same experimental conditions independently confirmed that: a) the water and ethanol gas phase 1b 1 and 3a" peaks were separated by the literature value of 1.97 eV, 51 b) the water gas and liquid 1b 1 splitting matches the literature value of 1.45 eV, 42 and c) the ethanol gas to liquid 3a" splitting was consistent with the splitting in the NaI/ethanol data. These findings all suggest that any residual effect of surface charging on these experiments does not affect the gas-liquid splitting significantly. The alignment of the 3a" HOMO of gas-phase ethanol to the literature value thus provides us with a calibrated electron binding energy scale of the liquid PE lines in Figure 3.5, relative to vacuum. If we were to calibrate to the first liquid IP originally reported by Faubel et al., the absolute energy of all features in the spectrum would be shifted 0.26 eV to higher ionization energy. This provides an estimate of the maximum error in our reported peak positions for iodide. In order to remove the background contributions from liquid ethanol which overlaps with that of iodide, the pure liquid photoelectron spectrum was scaled to match the liquid 3a" peak in the 1M NaI/ethanol spectrum (Fig. 3.5, red curve). This is acceptable despite the pure liquid spectrum being recorded at higher photon energy (which leads to variations in the respective photoionization cross sections) 53 58 since we only want to remove a single band’s contribution; that of the liquid 3a" state. The resulting spectrum containing only gas phase and iodide contributions is shown in the inset of Fig. 3.5. Although not shown here, a reconstruction of the liquid spectrum from the previous work by Faubel et al. shifted by 0.26 eV and narrowed by 0.8 eV (reflecting the difference in ΔE g-l and instrument resolution, respectively, in this experiment) produced an qualitatively identical result upon subtraction. Upon removal of the liquid background, it is clear that the lowest energy feature in the original spectrum is made up of two distinct transitions as also observed for aqueous iodide. 12, 49 This lowest band observed in the PE spectrum has been fitted with two Gaussians of equal widths. The peak near 7.3 eV and shoulder at 8.2 eV correspond to the two direct channels expected for 5p I¯ photodetachment. The resolved doublet (spaced by 0.95 eV) corresponds to the spin-orbit splitting of the final neutral iodine atom states, and is close to the splitting observed in the gas phase (0.94 eV) 54 and in water (1.1 eV). 49 This agreement, along with the fact that the bandwidths and relative weights are very similar to previous studies in water, gives us confidence in the overall subtraction procedure used here. The fit parameters for the iodide contribution to the PE spectrum are summarized in Table 1. 59 TABLE 3.1. Gaussian fit parameters for the first two iodide contributions to the 1 M NaI in ethanol PE spectrum. Center (eV) FWHM (eV) Area 7.26 0.88 0.57 8.21 0.88 0.43 3.5 Discussion In water, besides the solvated electron signal and a small transient tail between 320 and 400 nm, there is no clear spectral signature in the region where I*( 2 P 1/2 ) is predicted to absorb. This agrees well with our previous experiments where no signature was observed with 300 fs time resolution suggesting that nothing further is going on down to 150 fs. As before, we can conclude that if formed at all I* it is deactivated on a 100 fs timescale. However, in ethanol, the strong transient absorption at 360 nm is in good agreement with the expected absorption for I( 2 P 3/2 ) while a broad transient shoulder is also observed evolving on a different time scale than either the I( 2 P 3/2 ) or the thermalization of solvated electrons. The nature of this absorption feature and the mechanism for the clear differences between photodetachment in water and ethanol will be addressed in this discussion. 3.5.1 Analysis of the photoelectron spectrum Before assignments to the transient spectra are made, it is useful to discuss the PE spectra relative to the CTTS transition energies and compare the results obtained for ethanol to those previously observed in water. Figure 3.6 shows a comparison of the first two CTTS bands in the absorption spectrum and the two 60 lowest ionization potentials of iodide in ethanol (shown in black, top tier) and water (shown in grey). The binding energy for the 5p electron of iodide in ethanol is significantly lower than that in water; both the I( 2 P 3/2 ) and I*( 2 P 1/2 ) photoelectron peaks are 0.4-0.6 eV to lower eBE. The lower binding energy is presumably attributable to less effective charge stabilization by ethanol, which increases the initial-state energy. 32 Not really sure about this one In contrast both CTTS transitions shift to higher energy and so crucially the CTTS to vacuum gap is much smaller in ethanol than in water. As a result, in ethanol the I* CTTS band overlaps strongly with the I( 2 P 3/2 ) detachment continuum, and the high energy tail of the I* CTTS band also has a very small but non-trivial overlap with the I*( 2 P 1/2 ) opening. These observations may have important consequences in the ultrafast mechanisms for the two liquids. Improved energetic coupling to the lowest detachment continuum would enhance the fast autodetachment pathway but the improved overlap with the second I*( 2 P 1/2 ) continuum now opens up the possibility of direct detachment without change of electronic state of the core neutral atom. Thus, the I* CTTS state can couple to both detachment channels in ethanol. Although the degree of overlap between the I* CTTS band and I( 2 P 3/2 ) continuum is much greater than the I*( 2 P 1/2 ) continuum, a non-adiabatic transition is needed to undergo autodetachment. Thus, even with a relatively small amount of coupling both of these detachment pathways could become competitive. The relative populations of I and I* and a full discussion of the possible detachment pathways which result from the PES analysis will constitute the remaining portions of this work. 61 5 6 7 8 9 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Ethanol Water Water Electron Binding Energy, eV Intensity, arb. units Excitation Energy, eV FIG. 3.6. A plot of the iodide absorption spectrum in ethanol (below, black line), and the photoelectron spectrum from an ethanol liquid jet (above, black squares). The absorption spectrum has been deconvoluted (blue line) as well as the PE spectrum (red line). The overlap of the band associated with I( 2 P 1/2 ) in the absorption spectrum has overlap with both the I ( 2 P 3/2 ) and I( 2 P 1/2 ) direct photodetachment channels which leads to formation of both core photoproducts. For reference, the iodide spectrum in water (grey line) and PE spectrum in water (grey circles) are reproduced from ref. 12. 3.5.2 Assignment of I* in ethanol The shoulder feature adjacent to the I( 2 P 3/2 ) absorption seen in Fig. 3.3d arises from an absorption band peaking at 520 nm with similar bandwidth as the I( 2 P 3/2 ) absorption band. This absorption band peaks at approximately the wavelength predicted in section 3.2 for the I*( 2 P 1/2 ) CT band of 500 nm and clearly along with the decay of this band there is a rise in the ground state I radical 62 population. This is consistent with an inter-conversion of the two forms of iodine. We can easily rule out that the observed spectral evolution is a result of narrowing or shifting of the I( 2 P 3/2 ) charge-transfer band. If this was indeed happening we would observe a time-dependent shift of the entire band but instead there is a clear isosbestic point at 380 nm which is best explained by two state iodine population dynamics. With no other transients expected to absorb in this spectral region or on this timescale, these observations leads us to the natural conclusion that the 520 nm absorption band is due to I*( 2 P 1/2 ). This is the first time the I* species has been captured via its charge transfer absorption in liquid solution. We will now use a simple kinetic scheme to model the observed I and I* kinetics. This model does not represent the only explanation of the experimental results but does provide a maximum initial ratio for the I to I* population and provides a good starting point for the discussion. Alternative descriptions and caveats of this model will be presented later. Figure 3.7 shows the results of a kinetic scheme where an instantaneous starting populations of 70% I and 30% I* are allowed to evolve based on the competitive kinetics model 45 described in section 3.4.2 13 , where ground state iodine undergoes geminate recombination on a 1.2 ns time scale. The I and I* bandwidths and extinction are identical (as in Figure 3.1) but the I* center frequency is slightly shifted from 510 nm to 520 nm. The I* signal is allowed to decay exponentially on a 15 ps time scale. The small increase in I absorption corresponds to I*→I conversion competes with I* decay on a similar 20 63 ps timescale, thus reflecting the total increase in I( 2 P 3/2 ) by ~11%. As seen in figure 3.7b, the model seems to reproduce the experimental results well. 55 FIG. 3.7. (a) The transient spectrum after removal of electron signal is compared to a (b) simple simulation of the I and I* dynamics where I* is allowed to relax to I on a 20 ps timescale. An additional decay of I* on a 15 ps timescale is also included which is attributed to I*+e¯ recombination although several other mechanisms for this decay are discussed in the text. A qualitatively identical result arises from a simulation where the extinction coefficient of I* is increased compared to I. In this case I* is allowed to deactivate to I on a 10 ps timescale and no geminate recombination is included other than the timescale of 1.2 ns described by the electron model. 3.5.3 Detachment pathways in H 2 O and ethanol Several explanations for prompt appearance of both I and I* CT absorption are possible which will be considered, although a likely mechanism has already been suggested in the analysis of the PE spectrum. With reference to the introduction and the PE spectrum, i) competitiveness between the autodetachment pathway originally proposed in water 18 with coupling to the I* direct detachment channel in ethanol, thereby leading to the production of both I and I* in the latter solvent. Else ii) the 350 400 450 500 550 600 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 (a) Probe Wavelength, nm 600 fs 1 ps 3 ps 5 ps 10 ps 20 ps Δ Δ Δ ΔmOD 350 400 450 500 550 600 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 (b) Probe Wavelength, nm 600 fs 1 ps 3 ps 5 ps 10 ps 20 ps Δ Δ Δ ΔmOD 64 slower (by ~3 orders of magnitude) geminate recombination observed in ethanol is also slowing the recombination of I* with electrons long enough to capture the I* species. Finally, along this same line, iii) the autodetachment mechanism is substantially slowed such that absorption of I* is observed prior to undergoing any energetic relaxation to I or recombination reaction. These three scenarios will be examined below. The autodetachment mechanism proposed as an explanation for lack of I* in water is most likely responsible for the appearance of I CT absorption in ethanol. We observe significant overlap between the I* CTTS band and I detachment continuum from the PE spectrum. In fact, at 200 nm (6.2 eV) direct optical coupling to the I( 2 P 3/2 ) continuum is also possible and may account for some of the observed ground state iodine as the first PE peak begins to rise at 6.3 eV. It is also worth noting that a small fraction (~5%) of molecules are excited into the first CTTS band, estimated from the deconvoluted static absorption spectrum, which are also likely to undergo direct detachment by coupling between the lowest CTTS state and the tail of the I( 2 P 3/2 ) direct detachment continuum. This may indeed explain the prompt electron formation observed in previous detachment experiments done at 225 nm. 9 Analogously, we propose that the small coupling between the I* CTTS state and I* direct detachment continuum leads to a prompt formation of I* and I* CT absorption (case i). Changing coupling to the two detachment continua due to the decreased vertical detachment energies (VDE) and CTTS gaps observed in ethanol provides a nice picture for explaining the appearance of I* in ethanol and the overall differences 65 in the transient absorption data in water and ethanol. We could expect that as the solvent environment became less capable of stabilizing charges, i.e. as the alcohol chain length increases, this gap would continue to decrease and more I* would be produced. Although this may be the case, it should be noted that the I and I* CT bands would also shift to lower energies 28 which would further convolute the CT bands with solvated electron absorption. In this interpretation, the I* CTTS state in water can couple to the I( 2 P 3/2 ) only, and thus the lack of I* product must mean that autodetachment outcompetes adiabatic separation of the I*:e¯ contact pair. It is alternatively possible that autodetachment to the I( 2 P 3/2 ) continuum is slowed down enough in ethanol such that adiabatic separation outcompetes and the result is I* formation by a CTTS pathway (case ii). We doubt this interpretation not only because the better energetic overlap of the states but also because this would enhance geminate recombination since CTTS electron detachment pathways typically result in a contact pair, which is poised for recombination. In contrast, pathways that utilize the conduction band lead to longer trapping lengths. 56 Another mechanism for I* appearance and subsequent decay is related to slow electron mediated relaxation (case iii). In water, it has been suggested that fast electronic relaxation of I* by collisions with the surrounding solvent molecules and subsequent energy transfer to solvent vibrational modes could be responsible for the absence of an I* signature. 18 In ethanol, this process would have to be significantly slowed down in order to account for the observation of I* and subsequent decay on a 66 ps timescale. A simple calculation of room-temperature water from the use of a binary collision model and the I* gas phase deactivation rate leads to a I* lifetime of ~16 ps. The observed decay of the I* species in ethanol is remarkably similar to the estimated collisional lifetime in water suggesting that I* deactivation could be facilitated in this way. The electronic to vibrational energy transfer rate depends strongly on the overlap of the acceptor vibrational spectrum and the I* to I energy gap. Resonant Raman studies show that in the presence of electrons a strong enhancement of the Raman stretch and bend modes of the nearby solvent molecules occur, 57 which may enhance the electronic-to-vibrational transfer rate resulting in a much shorter I* lifetime. Although the exact quantification of this enhancement requires knowledge of the accepting solvent vibrational dipole strength, the resonant enhancement of the vibrational modes observed in water and ethanol are similar (~ 4 orders of magnitude). 57, 58 Thus, it is not apparent that the collisional relaxation time in ethanol should be three orders of magnitude slower than in water. 3.5.4 Modeling of transients in ethanol The suggested mechanism of two or more competing detachment pathways could provide a rationale for the need of a biexponential function to describe electron solvation in ethanol. The faster, 1.9 ps time constant would correspond to the less energetic electrons directly detached into the I* continuum while the 17 ps relaxation fraction would correspond to the higher energy electrons produced via autodetachment. The relative weights of 67% and 33% also correspond well with the observed I and I* band intensities at early times. Although this description is 67 convenient, it is certainly possible that electrons produced via autodetachment or direct detachment undergo similar solvation dynamics as both involve detachment into the water conduction band. Furthermore, it is common for polar solvation dynamics to be described by multi-exponential functions 59 as solvent molecules must reorient to accommodate changes in charge distribution. Therefore, the use of a biexponential function does not necessarily indicate multiple detachment mechanisms. The kinetic model used to describe the I and I* population presented in section 3.5.2 and Fig. 3.7 indicates that I* is allowed to decay on an additional 15 ps timescale which corresponds to ~63% of the I* population and 19% of the electron population. This decay was not included in the modeling of the electron absorption as no decay of electrons was observed, and it is not immediately clear why decay of I* from recombination of an electron (I* + e¯ → I¯) is likely to occur much fast than the I/e¯ pair. In the absence of bleach recovery information to confirm the lack of recombination, we would still expect that I*( 2 P 1/2 ) would react slower than I( 2 P 3/2 ). The recombination reaction is expected to be in the Marcus inverted regime, thus making electron transfer more downhill for I*. An abstraction reaction with the solvent is also possible. The most likely abstraction reactions involve removal of a hydrogen from ethanol to produce HI and either CH 3 CH 2 O, CH 3 CHOH or CH 2 CH 2 OH. The B←X electronic transitions of these radicals in the gas phase 60, 61 all lie above 342 nm and neither the positions, nor extinction coefficients of these species in solution are known. It is possible that one 68 of these species is responsible for the small rise observed in the region of the I( 2 P 3/2 ) absorption thus attributing all decay of the I* signal to hydrogen abstraction. One final scenario involves relaxation of the constraint that the extinction coefficients of I and I* are equivalent. A qualitatively identical result as that presented in Figure 3.7 can be obtained using a model where a 3:2 extinction coefficient is used for I* to I. In this case the initial I and I* are generated in an 4:1 ratio and I* is allowed to deactivate to I with a 10 ps time constant. An electron recombination rate of 1.2 ns¯ 1 is given to both species, consistent with the electron fitting model and all other parameters were unchanged. 3.6 Conclusion The identification of the I* spectrum and the proposed detachment mechanism is significant. To our knowledge this is the first report of a resolved I* spectrum from photodetachment in the condensed phase. Although I* is observed, autodetachment leading to ground state iodine is still the dominant detachment pathway. The photoelectron spectrum for iodine in ethanol, is also reported for the first time and show that the binding energies corresponding to I and I* cores are significantly red shifted from those observed in water. The result is that we observe overlap of the I* CTTS state with both I and I* direct ionization states opening a pathway for detachment through either channel, supporting our assignment of competing direct and autodetachment pathways. In water, no overlap is observed between the I* CTTS and direct ionization state leading to only observation of the 69 ground state radical. This connection between the photoelectron spectrum and the CTTS states of iodine in water and ethanol gives us new insight into the electron formation and recombination dynamics observed in longer chain alcohols. 9 Further experiments are required to confirm that the ionization potentials of iodide in longer chain alcohols such as proponol and butanol are further red shifted but near zero recombination of electrons from excitation of the lowest CTTS state in these solvents is observed. This indicates that a connection between the lowest I( 2 P 3/2 ) CTTS state and the corresponding I( 2 P 3/2 ) direct detachment state could put the electron directly into the solvent conduction band and be responsible for the lack of recombination. The identification of the I* CT spectrum has important consequences for other experiments, especially ICN photodissociation. Photodissociation of gas phase ICN leads to the formation of I* and it is expected that the same reaction in the condensed phase will also lead to I* formation. New dispersed probe experiments in combination with the I and I* spectra observed here will help clarify the transient species observed. 3.7 Acknowledgments The work at USC was supported by the U.S. National Science Foundation (CHE-0617060). Additional support was provided to Christopher Rivera by the Ford Foundation Diversity Fellowship. 70 3.8 References for Chapter 3 1. N. Chandrasekhar and P. Krebs, J. Chem. Phys., 2000, 112, 5910-5914. 2. G. Dobson and L. I. Grossweiner, Radiat. Res., 1964, 23, 290-299. 3. J. Jortner, M. Ottolenghi and G. Stein, J. Phys. Chem., 1963, 67, 1271-1274. 4. R. Lian, D. A. Oulianov, R. A. Crowell, I. A. Shkrob, X. Y. Chen and S. E. Bradforth, J. Phys. Chem. A, 2006, 110, 9071-9078. 5. F. H. Long, H. Lu, X. L. Shi and K. B. Eisenthal, Chem. Phys. Lett., 1990, 169, 165-171. 6. H. Seki and M. Imamura, Bull. Chem. Soc. Jpn., 1971, 44, 1538-1543. 7. A. T. Shreve, T. A. Yen and D. M. Neumark, Chem. Phys. Lett., 2010, 493, 216-219. 8. K. Takahashi, K. Suda, T. Seto, Y. Katsumura, R. Katoh, R. A. Crowell and J. F. Wishart, Radiat. Phys. Chem., 2009, 78, 1129-1132. 9. V. H. Vilchiz, X. Y. Chen, J. A. Kloepfer and S. E. Bradforth, Radiat. Phys. Chem., 2005, 72, 159-167. 10. C. G. Xia, J. Peon and B. Kohler, J. Chem. Phys., 2002, 117, 8855-8866. 11. H. A. Shen, N. Kurahashi, T. Horio, K. Sekiguchi and T. Suzuki, Chem. Lett., 2010, 39, 668-670. 12. X. Y. Chen and S. E. Bradforth, Annu. Rev. Phys. Chem., 2008, 59, 203-231. 13. O. T. Ehrler, G. B. Griffin, R. M. Young and D. M. Neumark, J. Phys. Chem. B, 2009, 113, 4031-4037. 14. O. T. Ehrler and D. M. Neumark, Acc. Chem. Res., 2009, 42, 769-777. 15. F. Mbaiwa, J. Wei, M. Van Duzor and R. Mabbs, J. Chem. Phys., 2010, 132, 134304. 16. T. Andersen, K. R. Lykke, D. M. Neumark and W. C. Lineberger, J. Chem. Phys., 1987, 86, 1858-1867. 71 17. D. M. Sagar, C. D. Bain and J. R. R. Verlet, J. Am. Chem. Soc., 2010, 132, 6917-6919. 18. A. C. Moskun, S. E. Bradforth, J. Thogersen and S. Keiding, J. Phys. Chem. A, 2006, 110, 10947-10955. 19. X. Y. Chen, D. S. Larsen, S. E. Bradforth and I. H. M. Stokkum, J. Phys. Chem. A, 2010, submitted. 20. M. Besnard, N. Delcampo, P. Fornier de Violet and C. Rulliere, Laser Chem., 1991, 11, 109-118. 21. P. Fornier de Violet, R. Bonneau and J. Joussot-Dubien, Chem. Phys. Lett., 1973, 19, 251-253. 22. P. Fornier de Violet, R. Bonneau and J. Joussot-Dubien, Mol. Photochem., 1973, 5, 61-67. 23. T. A. Gover and G. Porter, Proc. R. Soc. London Ser. A, 1961, 262, 476-488. 24. J. Larsen, D. Madsen, J. A. Poulsen, T. D. Poulsen, S. R. Keiding and J. Thøgersen, J. Chem. Phys., 2002, 116, 7997-8005. 25. A. C. Moskun, A. E. Jailaubekov, S. E. Bradforth, G. H. Tao and R. M. Stratt, Science, 2006, 311, 1907-1911. 26. A. C. Moskun and S. E. Bradforth, J. Chem. Phys., 2003, 119, 4500-4515. 27. C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys., 2009, 130, 084501. 28. P. Fornier de Violet, Rev. Chem. Intermed., 1981, 4, 121-169. 29. J. Jortner, B. Raz and G. Stein, Trans. Faraday Soc., 1960, 56, 1273-1275. 30. M. F. Fox and E. Hayon, J. Chem. Soc., Faraday Trans. 1, 1977, 73, 1003- 1016. 31. R. S. Berry, G. N. Spokes and C. W. Reimann, J. Chem. Phys., 1962, 37, 2278-2290. 32. S. E. Bradforth and P. Jungwirth, J. Phys. Chem. A, 2002, 106, 1286-1298. 33. P. M. Hare, E. A. Price and D. M. Bartels, J. Phys. Chem. A, 2008, 112, 6800-6802. 72 34. F. Y. Jou and G. R. Freeman, Can. J. Chem., 1979, 57, 591-597. 35. A. R. Menzeleev and T. F. Miller, J. Chem. Phys., 2010, 132, 034106. 36. V. H. Vilchiz, J. A. Kloepfer, A. C. Germaine, V. A. Lenchenkov and S. E. Bradforth, J. Phys. Chem. A, 2001, 105, 1711-1723. 37. H. Iglev, R. Laenen and A. Laubereau, Chem. Phys. Lett., 2004, 389, 427- 432. 38. H. Iglev, A. Trifonov, A. Thaller, I. Buchvarov, T. Fiebig and A. Laubereau, Chem. Phys. Lett., 2005, 403, 198-204. 39. J. P. Jay-Gerin, Can. J. Chem., 1997, 75, 1310-1314. 40. R. S. Mulliken, J. Phys. Chem., 1952, 56, 801-822. 41. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum., 2003, 74, 4958-4960. 42. B. Winter and M. Faubel, Chem. Rev., 2006, 106, 1176-1211. 43. T. Goulet, C. Pepin, D. Houde and J. P. Jay-Gerin, Radiat. Phys. Chem., 1999, 54, 441-448. 44. D. M. Bartels, K. Takahashi, J. A. Cline, T. W. Marin and C. D. Jonah, J. Phys. Chem. A, 2005, 109, 1299-1307. 45. A. Staib and D. Borgis, J. Chem. Phys., 1996, 104, 9027-9039. 46. Y. Hirata and N. Mataga, Prog. React. Kinet., 1993, 18, 273-308. 47. Y. Wang, M. K. Crawford, M. J. McAuliffe and K. B. Eisenthal, Chem. Phys. Lett., 1980, 74, 160-165. 48. X. L. Shi, F. H. Long and K. B. Eisenthal, J. Phys. Chem., 1995, 99, 6917- 6922. 49. R. Weber, B. Winter, P. M. Schmidt, W. Widdra, I. V. Hertel, M. Dittmar and M. Faubel, J. Phys. Chem. B, 2004, 108, 4729-4736. 50. M. Faubel, B. Steiner and J. P. Toennies, J. Chem. Phys, 1997, 106, 9013- 9031. 73 51. K. Kimura, Handbook of HeI photoelectron spectra of fundamental organic molecules : ionization energies, ab initio assignments, and valence electronic structure for 200 molecules, Japan Scientific Societies Press, Tokyo, 1981. 52. O. Link, Georg-August-Universitat Göttingen, 2007. 53. N. Ottosson, M. Faubel, S. E. Bradforth, P. Jungwirth and B. Winter, J. Electron Spectrosc., 2010, 177, 60-70. 54. C. E. Moore, Atomic energy levels as derived from the analyses of optical spectra, U.S. National Bureau of Standards, Washington, 1971. 55. The experimental data seems to slope very slightly more on the red side of the spectrum the model predicts. This is most likely the result of imperfections in pump/probe overlap seen below 2 eV discussed earlier. A similar feature can be seen in the small residual electron signal remaining in the water transient spectra. 56. C. G. Elles, A. E. Jailaubekov, R. A. Crowell and S. E. Bradforth, J. Chem. Phys., 2006, 125, 044515. 57. M. J. Tauber and R. A. Mathies, J. Am. Chem. Soc., 2003, 125, 1394-1402. 58. C. M. Stuart, M. J. Tauber and R. A. Mathies, J. Phys. Chem. A, 2007, 111, 8390-8400. 59. M. L. Horng, J. A. Gardecki, A. Papazyan and M. Maroncelli, J. Phys. Chem., 1995, 99, 17311-17337. 60. S. Gopalakrishnan, L. Zu and T. A. Miller, Chem. Phys. Lett., 2003, 380, 749-757. 61. C. Anastasi, V. Simpson, J. Munk and P. Pagsberg, Chem. Phys. Lett., 1989, 164, 18-22. 74 Chapter 4. The dynamical role of solvent on the ICN photodissociation reaction: Connecting experimental observables directly with molecular dynamics simulations In collaboration with: Nicolas Winter, Rachael V. Harper and Ilan Benjamin Department of Chemistry, University of California, Santa Cruz, California, 95064 Abstract for Chapter 4 ICN Ã photodissociation in water and ethanol is explored using sub 40 fs laser pulses and broadband (325-625 nm) simultaneously collected probing. It is well known that a nonadiabatic transition leads to rotationally hot cyano radicals from a multitude of gas phase studies and recent experiments suggest that this rotational energy is ineffectively relaxed by the solvent. The experimental study shows a distinct spectroscopic signature from rotationally excited CN radicals dominates the spectrum in both water and ethanol. The transient spectra also contain contributions from iodine radicals in both the ground state, I( 2 P 3/2 ) and spin excited state, I*( 2 P 1/2 ). Kinetic modeling helps remove the iodine contributions revealing a clear spectral shift of the CN spectrum in water as the rotational energy is relaxed by the solvent. Semi-classical molecular dynamics simulations which include surface hopping have been used to characterized the microscopic properties of this benchmark system but new spectroscopically motivated sorting criteria allows us to 75 reanalyze the simulation results and directly compare them to our experimental observables for the first time. With the improved time resolution of the experiment we are also able to examine the effect of the solvent on the curve crossing probability and estimate the first-pass crossing probability to be 34-47% in ethanol. 4.1 Introduction The effect of photodissociating a molecule in solvent and studying the transients produced has been a topic of great interest both experimentally 1-10 and theoretically 11-15 for many years. While early experiments and theory focused on simpler systems like I 2 and iodoaromatics, more recent work has focused on complex systems such as HgI 2 , I 3 − and ICN which exhibit significant and measurable energy release into vibrational and rotational channels of the photoproducts. 6-8, 10, 15-19 Photodissociation of ICN leads to photoproducts which have been shown to exhibit a highly non-equilibrium distribution of rotational energy imparted into the CN photofragment regardless of ICN being solvated or not. 17, 18, 20 Recent condensed phase experimental and theoretical work has focused on the inability of early solvent collisions to effectively cool the spinning CN, resulting in isotropic signatures when probing at the gas-phase CN B←X transition energy, and gas-phase like anisotropic signals. 17, 18 High level gas-phase ab initio potential surfaces reveal that the burst of energy responsible for this behavior comes from a non-adiabatic transition that amplifies the rapidly bending molecule 76 evolving on the excited state. 21 These surfaces facilitate molecular dynamics (MD) simulations that show rotational energy distributions of N > 40 for molecules that change electronic states through the conical intersection. 21 Placing this reaction in a cage of solvent molecules produces many interesting and challenging phenomenological questions which are still under investigation. One of the most challenging questions, and also one of the most conceptually basic, involves identification of the absorption spectra of the transient species. The transient radical absorption bands are expected to consist of both electronic and charge-transfer transitions. 16, 17, 22, 23 Like most electronic spectra, these are expected to have broad bandwidths which strongly overlap with each other. Most of the species produced in this reaction have not been observed with time-resolution that allows for clear resolution of the oscillator strengths and band positions. In particular, the excited spin state of iodine I*( 2 P 1/2 ) and CN radicals have never been observed in ethanol. The CN radical is of particular interest as we expect to see both solvated and “free” spinning CN, the latter of which was not observed in previous experiments which followed the transients produced with 266 nm photolysis energy and 500 fs time resolution. 16 We provide a detailed spectroscopic section in this article which details the information available both experimentally and theoretically to predict and assign the spectral features we observe from our broadband transient measurements. The evolution of the CN band as rotors cool via solvation, will be especially important for understanding and interpreting our experimental observations. 77 Dynamically, there are many processes influenced by the solvent, and expected on many different timescales. These include recombination from recoil off of the solvent cage, escape out of the solvent cage, diffusive recombination, abstraction reactions with both solutes and solvent, rotational and vibrational relaxation. One of the most interesting aspects, and experimentally difficult to measure, involves following molecules through the conical intersection. Previous MD studies indicate that after excitation the molecule passes the curve-crossing region within 30 fs. 12, 15, 21 It is unclear whether the solvent influences this timescale and the effect on the coupling of the two surfaces. Classically, electrostatic drag from dipolar hydrogen-bonded solvents such as water and ethanol could slow down the molecules exiting the Frank-Condon region and amplify the curve-crossing probability while this same effect may also provide an energetic barrier preventing first pass curve crossing. Furthermore, this process happens faster than typical solvent reorientation time scales making it not entirely clear that the solvent will even have an effect on curve crossing probability. It is expected that cage recombination begins within 100 fs of excitation 12, 15 so the challenge experimentally is to observe the product population prior to significant cage recombination and elucidate the possibility of undergoing a first pass curve crossing. Uncovering this microscopic aspect of the experiment which is responsible for many of the longer timescale observations has never been attempted in solution. While sub-30 fs DUV with good dispersion characteristics is still not easy to produce, 24, 25 we have developed a robust method to produce 78 pulses resulting in ~40 fs time resolution which can begin to uncover some of these early features for the first time. As mentioned previously, condensed phase MD has played a valuable role in breaking down the energy release upon photodissociation. 12, 15, 18 The work by Benjamin and Wilson 14 provided the major motivation for the early experimental measurements and utilizing the surfaces calculated by Amatatsu, Yabushita and Morokuma, 21 along with a fully flexible water potential, provided the ground work for the MD simulations presented in this work. The inclusion of Tully’s method 26 for modeling curve crossing dynamics, makes these quantum-classical simulations the best comparison to our experimental results and allow us to follow energy release and nonadiabatic behavior in great detail. Previous analysis of the MD trajectories dissociated at 266 nm provided details on the first pass curve crossing probability, translational, rotational and vibrational energy release, and average recombination time, among many other properties. 15 In this work we have increased statistics from the initial 800 trajectories to 1600 trajectories per excited state and most importantly included a spectroscopically motivated sorting technique. This sorting method evaluates the position of the I and CN on the excited state and determines whether the photoproducts are separated or recombined at every time point for each trajectory. The details and justification of this sorting is explained in detail below, but essentially we believe that we measure primarily “dissociated” photoproducts experimentally. This sorting method removes recombined trajectories from the ensemble average 79 which were previously left in. The result is that we can now determine free I, I* and CN populations as a function of time which are directly related to the experimental isotropic signals. In this chapter, the broadband transient spectra using 266 nm pump excitation of the ICN Ã continuum in water and ethanol is reported. This is the first time that the broadband visible spectrum has been probed simultaneously for ICN photodissociation. Several transient signals are revealed and we discuss both the time and spectral evolution of these species over the 20 ps timescale of the experiment with unparalleled sub-50 fs time resolution. We can pair our transient signals with those pertaining to I, I* and CN from sophisticated molecular dynamics simulations. These simulations which include fully flexible water molecules and all of the atoms included in the experimental system provide the most realistic comparison to the experimental measurements and help confirm our spectral assignments. 4.2 Experimental The methods for generating 266 nm pump and supercontinuum probe employed for this work are identical to those presented by Jailaubekov et al., and Elles et al. 24, 27 The broadband spectrometer setup has also been presented in previous publications. 27, 28 For brevity, only a discussion the most crucial details of the experimental methods are presented here and we refer to previous works for a more thorough discussion. 80 4.2.1 266 nm pump pulses A 250 μJ portion of a 1 kHz regeneratively amplified Ti:sapphire laser system (Spectra-Physics Hurricane, 750 μJ, 800 nm, ~110 fs pulse width) was frequency doubled to produce 70 μJ of 400 nm. The residual fundamental and 400 nm were combined in a hollow-core four wave mixing (FWM) apparatus described previously 24 to generate 3.5 μJ of third-harmonic 266 nm light. Prism compression and autocorrelation resulted in a 36 fs FWHM 266 nm pulses. These pulses were used to measure the dispersed pump-probe isotropic data presented for both solvents. The pump beam was then chopped to 500 Hz and focused to ~100 μm at the sample. The polarization of the pump was controlled by an air-spaced 266 nm zero-order quarter-wave waveplate (CVI). 4.2.2 Broadband continuum probe Generation of the probe pulses consisted of focusing a small portion of the fundamental into a circularly translating piece of 2 mm thick calcium fluoride to generate super-continuum pulses that range from 300–700 nm. The most stable portion of this spectrum ranged from 325–625 nm and produced the probing window available in these experiments. After generation in the CaF 2 disk, the super-continuum was collimated and then focused into the sample using a pair of protected aluminum off-axis parabolic mirrors (Janos Technology) to ~60 μm diameter. The polarization of the 800 nm was also controlled by a zero-order half-wave waveplate (CVI) and a quartz polarizer cube was used to check that that polarization was better than 200:1 across the entire probe spectrum. 81 4.2.3 Sample and spectrometer ICN was synthesized and purified in-house my the same method described by Moskun et al. 17 A 170 mM solution of ICN in water (distilled, Arrowhead) and 260 mM solution of ICN in ethanol (200 proof, KOPTEC) were used for all experiments at 266 and 271 nm. These solutions were delivered to the laser beams by a flowing thin-film wire guided gravity drop jet. 29 The liquid film thickness was ~60 μm. The pump and probe beams are overlapped spatially in the flowing sample. For dispersed pump-probe measurements, the probe is collimated after the sample and subsequently focused into a spectrometer. A UV diffraction grating (300 gr/mm, 300 nm blaze, Oriel) is used to disperse the probe onto a 256 pixel diode array. Transient dynamics were measured out to 20 ps for all 266 nm experiments and 100 ps for 271 nm pumped experiments. The temporal delay between the pump and probe pulses is achieved by a computer controlled motorized linear delay stage (Newport) mounted with a retro-reflector in the path of the probe before super-continuum generation. The ΔA signal collected at each pixel simultaneously for each time delay is defined as ∆ log _ _ 4.1 where T corresponds to the relative transmitted light with and without the pump pulse. ΔA is reported in units of mOD. In order to collect the isotropic transient dynamics the probe polarization was rotated independently until the polarization 82 angle between the pump and probe was 54.7 o (magic angle), which excludes reorientation dependant signals. The time resolution for these experiments was estimated by measuring the 1/e 2 width of the two-photon absorption signal in pure water. 24, 27 The time resolution is defined as the point where this signal reaches 1/e 2 of the peak intensity centered at zero. This criterion was chosen since Gaussian fits did not always provide sufficient descriptions of the cross-correlation function resulting in shorter time-resolution estimates than what is observed due to higher order dispersion and cross-phase modulation in the pulse wings. The time resolution across the spectrum was 37 fs in water (although a small amount of negative cross-phase modulation signal was observed until 70 fs) and 50 fs in ethanol. The solvent backgrounds were also taken to ensure that the pump intensity was small enough such that no electrons or other photoproducts were generated by 2- photon absorption (pump+pump) of the solvent. 4.3 Molecular dynamics The photodissociation of ICN has been studied by many theoretical methods including both classical and semiclassical trajectory calculations 21, 30-33 and time-dependent quantum mechanical simulations. 34-39 The development of high level ab initio potential energy surfaces 21, 31 in combination with Leonard- Jones solute-solvent and solvent-solvent potentials provided the basis for extension of semiclassical simulations to the condensed phase. 12, 15, 40-43 We have 83 built upon existing methodology for ICN photodissociation in water and thus we refer to this previous work for a detailed description of our semiclassical approach. 15 We will only briefly describe the methods used and emphasize the modifications we have implemented for this work. 4.3.1 Potential energy functions The ICN potential energy functions, which have been studied extensively by Morokuma and co-workers 21 have been employed in several condensed phase studies previously. 12, 15, 40, 41, 43 There are three excited states ( 3 Π 1 , 3 Π 0+ and 1 Π 1 ) which can be accessed by excitation from the ground state ( 1 Σ + ) of ICN which make up the continuous Ã band. Excitation to all three surfaces leads to photodissociation in the gas phase. The potential energy functions for these four states are shown in Fig. 4.1 for the ICN linear geometry. The gas phase ICNÃ absorption continuum is accessible by photon energies ranging from 350–200 nm and lead to the production of a CN radical in its ground state ( 2 Σ + ) and an atomic iodine radical in its ground state I( 2 P 3/2 ) or its spin-orbit excited state I*( 2 P 1/2 ). These two channels, which are represented by the two adiabatic limits at rI–CN of 6Å in Fig. 4.1, are split by ~0.9 eV which corresponds to the spin-orbit splitting of iodine. The non-adiabatic curve crossing between the 3 Π 0+ and 1 Π 1 occurs at rI–CN of ~3.3Å in Fig. 4.1 and the details of the quantum mechanical treatment of transitions between these states will be discussed below. 84 2 3 4 5 6 -4 -2 0 2 4 6 160 180 200 220 240 260 280 0 3 Π Π Π Π 0+ 1 Π Π Π Π 1 3 Π Π Π Π 1 Absorption (A.U.) Wavelength (nm) 1 Σ Σ Σ Σ + 0+ 3 Π Π Π Π 1 1 Π Π Π Π 1 3 Π Π Π Π 0+ rI-CN (Å) E (eV) FIG. 4.1. Potential energy curves corresponding to the ground state (black line) and three lowest excited states (colored lines) reproduced from ref. 21 and plotted as a function of I to CN center-of-mass and linear ICN. The molecular dynamics sorting cut-offs are also shown (gray dashed lines). The inset shows the calculated ICNÃ band from molecular dynamics simulations. The water potential energy functions consist of a fully flexible intramolecular portion adapted from the potential described by Kichitsu and Morino 44 and a pair-wise sum of Lennard-Jones and Coulombic potentials (for water-water interactions). This potential has been shown to reasonably reproduce the bulk properties of water. 45 Similarly, the interaction of ICN and water molecules for both the ground and excited states is also modeled with Lennard- 85 Jones plus Coulomb terms. The details of the parameters was described previously 15 but is worth noting that because of the blue shift of the absorption spectrum of the ICNÃ band in water and alcohols, a significant change in the excited-state charge distribution is expected. Therefore, great care was taken to include a charge switching function which accounts for the charge on each atom (I, C and N) both in the Franck-Condon and in the product channels. The parameters included were based on atomic charge calculations, as well as experimental and ab initio calculations if the ICN, INC and CN dipole moments. 4.3.2 MD methods The total system simulated consisted of one ICN molecule embedded in a 1000 water molecules confined in a cubic box of fixed pressure at the experimental density at 293 K with three-dimensional periodic boundary conditions. A total of 1600 trajectories with 0.5 fs steps (values were reported every 5 fs) were run on each excited electronic state of ICN. At 266 nm, the relative weights of 3 Π 0+ and 3 Π 1 are ~2:3 with 1 Π 1 contributing less than 1% based on our calculated absorption spectrum. The absorption spectrum was calculated using the classical Frank-Condon (FC) approximation 1 |"r| # $% Ωr&' 4.2 where ω is the incident photon frequency and μ(r) is the transition dipole moment which is taken to be a constant. The expression ћΩ(r) is the energy gap between 86 the ground and excited states. Two 320 ps trajectories were used to calculate the average absorption spectrum at constant temperature. At every time point the energy gap for the ground to excited states was calculated for the specific geometry and then binned by energy producing normalized line shapes for each excited state. Then each line shape was multiplied by the calculated oscillator strength. 21 The resulting spectrum is shown in the inset in Fig. 4.1. Every 20 ps a different configuration along the equilibrium trajectory is saved for use as an initial configuration when searching for proper Franck-Condon conditions for the laser-induced dynamics. An iterative procedure which scans each of the 32 saved initial conditions until a proper FC configuration was found where ћω(r)=V ex (r)-V gr (r) depending on the photodissociation wavelength of interest. Five of these configurations were extracted from each saved configuration producing 160 FC configurations. Because we find that the transition strength to 1 Π 1 is very small no intitial FC configurations are evaluated for this surface and we concentrate excited state dynamics to trajectories initiated only to 3 Π 0+ and 3 Π 1 . The inclusion of non-adiabatic surface hopping between the 3 Π 0+ and the 1 Π 1 excited states of this system is used to describe the mixed quantum and classical nature of the experimentally observed photodissociation dynamics. The surface hopping method developed by Tully, 26 using coupling from the excited states calculated by Amatsu et al. 21 and the instantaneous ICN atomic positions, is employed at every step of the MD trajectories. The details of this treatment are 87 described previously 15, 43 and this method is used in exactly the same way. Coupling between the ground and excited states at the asymptotic limit is also included such that when the ground-excited state energy gap is less than kT, the transition probability to the ground state becomes 1. Trajectory calculations compute various properties of the system including rotational, translational, and vibrational energy distributions, potential energy of the system, and atomic positions for each trajectory, at each time step. These properties can be used to compute various ensemble averages of the system and the most pertinent are reported and discussed in the following sections. 4.4 Spectroscopy Figure 4.2 shows the spectrum for ICN all of the relevant transient spectra we would expect to observe from photodissociation in water and ethanol respectively. These spectra are comprised of various theoretical and experimental studies which will be discussed in this section. 4.4.1 ICN Ã spectrum The ICN Ã band is plotted in Fig. 4.2 as a function of molar extinction coefficient dissolved in both water and ethanol. The spectrum peaks at approximately 225 nm in both solvents. The is a ~5000 cm –1 shift with respect to the gas phase absorption band. 20 We have postulated previously 17 that this shift is due to the stabilization of the ground state from polar solvation of the ground state of ICN and destabilization of the excited states. Although the tail of the 88 ICN absorption band stretches above 300 nm, due to the relatively weak absorption cross-section compared to the other transients species that will be discussed in this section, we do not expect to observe any bleach or subsequent recombination signal in our probe wavelength range. The absorption feature rising on the blue side of the spectrum is due to the α-band absorption of ICN which begins at 205 nm in water and 212 nm in ethanol, well outside our pump energies. The similarity of the shape of the absorption spectra between the gas and liquid phase (and water to alcohol), suggests that the ICN environment has little effect on the shape of the ground and excited state surfaces in the FC region. The isomer of ICN, INC has been observed in a solid Ar matrix and absorbs near 245 nm with an ε max of ~3700 M –1 cm –1 . The previous work dissociating ICN at 266 nm by Larsen et al. showed a strong absorption feature produced <1 ps peaking below 227 nm and stretching to ~250 nm which was ascribed to INC but we expect that this absorption feature is also outside of our probe window. 4.4.2 Iodine radical The two photodissociation channels are expected to lead to two atomic species of iodine. These two species, I( 2 P 3/2 ) and I*( 2 P 1/2 ) are known to be produced from the gas phase reaction 20, 46 and I( 2 P 3/2 ) has a well known condensed phase charge transfer band. The center frequency of this band is highly solvent dependent as the I species acts as the electron acceptor and thus largely depends on the ionization potential of the solvent. 9, 47, 48 For water the I( 2 P 3/2 ) CT band has been observed by both flash photolysis of I 2 and ultrafast 89 pump probe measurements 22, 47 and is shown in Fig. 4.2a centered at ~260 nm. The I* CT band has never been definitively resolved in a polar liquid. Estimation of the nature of this band has been done using Mulliken’s theory of charge transfer states which places the CT transition of I* approximately 0.94 eV lower in energy than the I CT band. 16, 17, 22, 49 Larsen et al. assigned all transient absorption in the region of 250 to 377 nm to I and I* absorption although these features could not be clearly resolved from each other. A collaborative experiment done with the Danish group to identify the I*(aq) CT band by photodetachment of an electron from I¯(aq) with 200 nm pump pulses did not yield any signature from I* although these experiments did confirm that I( 2 P 3/2 ) CT absorption is indeed centered at 260 nm with approximately the same bandwidth of 1.2 eV FWHM assigned from flash photolysis measurements. 22 The photoelectron spectrum of I¯(aq) and indicates that I* can quickly deactivate to ground state I through an autodetachment mechanism. Without evidence to the contrary, it is assumed that the bandwidth and extinction of the I* CT band in Fig. 4.2a is identical to that of the I CT band. In ethanol the exact position of the I( 2 P 3/2 ) is less well-defined. Flash photolysis measurements estimated a band position of ~360 nm but convolution with absorption from I 2 ¯ make determination of the exact center frequency and bandwidth difficult. 47 Our recent experiments photodissociating I¯ at 200 nm in ethanol definitively confirm that the I( 2 P 3/2 ) CT band is indeed centered at 360 nm and a broad shoulder with the same bandwidth is also observed centered at 90 ~500 nm. 50 The splitting of 0.97 eV is nearly identical to the gas-phase spin-orbit splitting observed for I and I*, and the bandwidths of 1.1 eV are in good agreement with the CT spectra of I in other alcohols. Gaussian representations of these two bands are shown in Fig. 4.2b as a function of molar extinction. The relative extinction coefficients in ethanol are ~2–3 times greater than in water and there is significant overlap of both absorption bands with our probe spectrum. 200 300 400 500 600 0 500 1000 1500 0 500 1000 1500 2000 2500 ICN(aq) ICN(EtOH) ε ε ε ε (M -1 cm -1 ) Wavelength (nm) 0.1xCN(g) v=1 (b) ethanol (a) H 2 O Probing Region I( 2 P 1/2 )(EtOH) I( 2 P 3/2 )(EtOH) 0.1xCN(g) CN(aq) 0.1xCN(g) v=0 I( 2 P 1/2 )(aq) I( 2 P 3/2 )(aq) FIG. 4.2. Transient absorption spectra for products generated from ICN photodissociation in water and ethanol. The ICN (black line), I( 2 P 3/2 ) (solid green line) from ref. 47, I( 2 P 1/2 ) (dashed green line, see text), and gas phase CN B←X (red line) adapted from ref. 52 and 18 are plotted as a function of wavelength. The calculated aqueous CN B←X spectrum (blue line) from ref. 23 is shown in panel (a). The colored bar represents the probe wavelengths available in this experiment. 4.4.3 Cyano radical 91 The cyano radical in the condensed phase is the least characterized of these transient species. The most detailed investigation of the CN aqueous phase electronic spectrum are theoretical. 23 It is known from gas phase calculations and experiments that CN(g) displays a transition at 390 nm (Fig. 4.2) corresponding to the CN B←X excitation. 51, 52 It is expected that the CN B state would display a blue shift upon solvation due to its smaller permanent electric dipole and opposite sign with respect to the ground state dipole. 53 The work by Pieniazek et al. which combined EOM-CCSD and molecular dynamics to calculate CN excitation energies, estimated the magnitude of the blue shift and effect on the band shape of CN B←X transition. The solvated CN is predicted to absorb at 340 nm and the broadened spectrum is shown in Fig. 4.2a. 23 Experimentally, Larsen et al. assigned a broad shoulder on the solvated electron signal to absorption of CN in water centered at 450 nm. 16 Our recent work probing between 385–400 nm displayed long lived anisotropic signal as well as transient signal that we attribute to free CN B←X absorption. 16-18 In ethanol, the blue shift is not as well defined as the dipole moment of water (1.85 D) is ~10% larger than that of ethanol (1.69 D) 54 and no experimental or theoretical attempts to resolve the CN spectrum have been carried out to our knowledge. It is expected that the blue shift in ethanol would not be as large due to the smaller dipole moment of the solvent compared to water as well as the decrease in hydrogen bonding strength and number of ethanol molecules in the solvent shell that have to accommodate the change in dipole moment upon excitation. One of the major motivations for reexamining 92 this reaction with good time resolution and dispersed pump-probe is not only to resolve the spectrum of CN radicals, but also to examine how this band evolves in time through molecular collisions and subsequent solvation. FIG. 4.3. Time-dependant transient absorption at magic angle for a) 170 mM ICN in water and b) 260 mM ICN in ethanol excited at 266 nm. Contour plots show the 20 ps time-dependant transient signal, ΔA, with color scale shown to right of each panel increasing in intensity from blue to red. c) Contour plot show only the first picosecond of the experimental data in water and d) ethanol. e) Several spectral slices are shown at time intervals ranging from 70 fs to 20 ps from water and f) ethanol. 4.5 Experimental results 350 400 450 500 550 600 0.0 0.5 1.0 350 400 450 500 550 600 0.0 0.5 1.0 Δτ Δτ Δτ Δτ (ps) 350 400 450 500 550 600 0 5 10 15 20 (e) (c) (a) Δ Δ Δ ΔA (mOD) Δ Δ Δ ΔA (mOD) Δτ Δτ Δτ Δτ (ps) -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 170 mM ICN in water Wavelength (nm) 70 fs 100 fs 500 fs 1 ps 5 ps 10 ps 20 ps 350 400 450 500 550 600 0.0 0.5 1.0 1.5 2.0 350 400 450 500 550 600 0.0 0.5 1.0 Δτ Δτ Δτ Δτ (ps) 350 400 450 500 550 600 0 5 10 15 20 (d) (b) Δ Δ Δ ΔA (mOD) Δ Δ Δ ΔA (mOD) Δτ Δτ Δτ Δτ (ps) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 (f) 260 mM ICN in ethanol Wavelength (nm) 70 fs 100 fs 500 fs 1 ps 5 ps 10 ps 20 ps 93 The experimental results consist of the isotropic data for ICN photodissociation at 266 nm in water and ethanol. The complete 2D isotropic transient absorption spectra are shown in Fig. 4.3a,b to 20 ps and Fig. 4.3c,d shows an enlargement of the first picosecond for each solvent system. Several slices of the transient spectrum which range between 70 fs and 20 ps are shown in Fig. 4.3e,f. The contour plots in Fig. 4.3a–d were smoothed by adjacent point averaging for clarity, while the spectral slices in Fig. 4.3e,f show the raw data. 4.5.1 Isotropic results for ICN in water The transient absorption spectrum in water shows two distinct transient features evolving in time. At early times (<200 fs) the spectrum is dominated by a transient feature peaking at ~385 nm with a long tail stretching across the visible (Fig. 4.3c and e, red/orange spectra). This absorption peak decays quickly as another absorption band peaking below 325 nm can be observed rising on the blue edge of the spectrum. At 20 ps, only the tail of an absorption feature can be observed which peaks below 325 nm. Figure 4.4(a) shows the time-dependent transient signal for three selected wavelengths of 330, 385 and 500 nm. All three wavelengths can be adequately fit with double exponential functions. A fast decay of ~100 fs (58%) is observed at 500 nm while a similar ~80 fs (43%) rise is observed at 500 nm (Fig. 4.4 a, inset). The signal at 385 nm corresponding to the absorption peak at early times shows a much slower initial decay of 520 fs (58%). A slower decay of 6 (50%) and 6.5 ps (36%) at 330 and 385 nm respectively, are observed while at 500 nm the signal decays significantly faster with a 1.6 ps 94 (35%). At all wavelengths a residual fraction of 5–7% is estimated to remain at long times. Assignment and consequences of these observations will be discussed in detail in the discussion. For now we will compare these results to those in ethanol. 0 5 10 15 20 0.0 0.5 1.0 1.5 2.0 0 5 10 15 20 0.0 0.5 1.0 0 1 2 0.0 0.5 1.0 500 nm Δ Δ Δ ΔA (mOD) Δτ Δτ Δτ Δτ (ps) (b) ethanol 500 nm 330 nm Δ Δ Δ ΔA (mOD) Δτ Δτ Δτ Δτ (ps) 385 nm 0 1 2 0.0 0.5 1.0 Δτ Δτ Δτ Δτ (ps) Δ Δ Δ ΔA (mOD) (a) H 2 O 330 nm 385 nm 500 nm Δτ Δτ Δτ Δτ (ps) Δ Δ Δ ΔA (mOD) FIG. 4.4. Three individual probe wavelength cuts at 330 nm, 385 nm and 500 nm for ICN in a) water and b) ethanol taken at magic angle showing isotropic transient absorption. 4.5.2 Isotropic results for ICN in ethanol In ethanol, the dissociation dynamics occur on a much different time scales although much of the observed dynamics appear to be rather similar. At early times (<1 ps) a sharp peak is observed at ~385 nm with a long tail stretching out across the visible wavelengths of our probe. This band structure is almost identical to that observed in water (compare red and orange spectra in Fig. 4.3e and f) although only a small decay of this absorption band is observed over the first ps (Fig. 4.3d). This band, although not as sharp at later times, is clearly 95 still present at 20 ps, whereas in water this absorption band is not observed in the probe window beyond 10 ps. A separate absorption feature appears to rise on the farthest UV side of the spectrum. The broad absorption feature which comprises the visible tail, extends beyond 600 nm and decays over the first 5 ps in ethanol. Figure 4.4b shows the time-dependent transient signal at 330, 385 and 500 nm. In ethanol, three exponentials are needed to describe the observed transient behavior at 330 and 385 nm. At all three wavelengths, an initial fast decay of ~60 fs constitutes 45–74% of the transient signal. The actual contribution to the overall transient signal was difficult to determine as the magnitude of this signal was influenced by convolution with the tail of the instrument response function. The fact that the time constant is identical across the spectrum indicates the same dynamical process is responsible for decay of all of the observed signals at early times. A 1 ps (17%) rise and 12 ps (26%) decay is observed at 330 nm (Fig. 4.4b, black curve). A small 1.4 ps (7%) decay of the 385 nm signal is observed corresponding to the relatively small change observed in the absorption peak over the first ps. The slower decay component of 14 ps (29%) is similar to the decay at 330 nm although it does appear in Fig. 4.4b (red and black) to be slightly slower. At 500 nm, only a double exponential is needed to fit the decay. A 2.9 ps (19%) decay is observed which does not correspond to either the 1 ps rise at 330 nm or the 1.4 ps decay at 385. It is more likely that this corresponds to the average of a shorter and longer decay component but the details and evidence for this will be presented in the next section. 96 4.6 MD results The MD results are separated based on the initial excited state of the ICN trajectory. An important spectroscopically motivating sorting criteria is employed to examine the data and produce ensemble averages for which pertain to the transient photoproducts. 4.6.1 Spectroscopic sorting Spectroscopically, we are only interested in trajectories where CN remains separated from iodine and following the energetics until recombination since we believe ICN and INC will be outside our probing window. Therefore, we apply two sorting criteria to each trajectory at each time point to decide whether to include a trajectory in the ensemble average. First a cut-off is set at an I to CN center of mass distance of 3.5 Å where we consider the I and CN molecules separated by more than 3.5 Å to be dissociated and less to be recombined. Molecules which recombine on the excited state and later redissociate are allowed to move in and out of the ensemble average. Second, a cut-off is set to –13 kcal/mol (–0.56 eV) corresponding to recombination on the ground state. With this new spectroscopically motivated filtering of the ensemble we can follow the lifetime of the transient I and CN products and remove the large population of recombined trajectories, making our ensemble averages comparable to the transient signals obtained. 97 The previous analysis of the first 800 trajectories produced many spectroscopically interesting results, especially at early times where nearly all trajectories are still free. Some of these results, including branching ratios, recombination time, and final I and I* populations, which provided interesting results to compare to the experimental observations. This analysis showed that the majority of trajectories (85%) ultimately recombined on the ground state, but were not removed from the ensemble averages. 15 Thus, the ensemble averages were comprised of every trajectory at every time point. Trajectories which undergo a non-adiabatic transition from 3 Π 0+ to 1 Π 1 , which is responsible for imparting the rotational energy on the CN fragment, were uniformly grouped together with all trajectories that finish on the ground state, including those that start on the 3 Π 1 , which do not receive the same rotational torquing. It was the goal of this collaborative effort to sort out these trajectories once recombined, and uncover the energetic properties of the “free” ensemble of trajectories at each time point. 4.6.2 Results from 3 Π 0+ The results from the MD are separated initially by the initial excited state, 3 Π 0+ or 3 Π 1 . For trajectories starting on the 3 Π 0+ where nonadiabatic transitions to the 1 Π 1 surface are expected, the time dependant populations of I, I* and CN are shown in Figure 4.5a and b. The rise in the I and I* signal occurs over the first 100 fs as seen in Figure 4.5a. The rise corresponding to 27% for the I population corresponds to the initial percentage of trajectories that curve cross on the first 98 pass. After the sub 100 fs rise, a fast initial decay over the first 100–200 fs of both I and I* corresponds to collision with the solvent cage. Solvent recoil on the I* surface leads to an increase in the I population but cage recombination on the ground state is significant enough that it is still observed (Fig. 4.5a, inset). This solvent caging behavior is corroborated by a spike in the water–CN potential energy which is observed at 50 fs. 15 A second rise of the I population is observed between 200–400 fs corresponding to later nonadiabatic transitions on second or more passes through the curve crossing region. The second rise in I population peaks at 23% (450 fs). Finally a slower decay is observed in both I and I* populations as the trajectories undergo later curve crossings and diffusional recombination. It is important to note that I* cannot recombine on the ground state without undergoing a nonadiabatic transition induced by a nearby CN radical. At the end of the 5 ps trajectories, the fractional I and I* populations correspond to 8% and 12% respectively. The CN population, which is just the sum of the I and I* populations, shows a similar fast decay corresponding to cage recoil and recombination (although this is a relatively small fraction of the total population) and longer ps decays which correspond to secondary recombination through diffusion once most of the I* has curve-crossed. Initially the CN population peaks at ~95–97% before the strong caging decay takes over. Interestingly a small rise at 350 fs can be seen. The only way for this to occur is by trajectories that temporarily recombine on the excited state ( 3 Π 0+ does have a shallow minimum at linear rI– 99 CN = 2.5 Å, and less than 1 kT) which causes a drop in the CN population, but upon eventual movement either back onto the dissociative part of the 3 Π 0+ , or nonadiabatically on the 1 Π 1 surface, the “free” CN population will include these trajectories again. How such a trapped population appears spectroscopically will be revisited below. 0 1 2 3 4 5 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0 1 2 3 4 5 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 Population Δτ Δτ Δτ Δτ (ps) I I* Population Δτ Δτ Δτ Δτ (ps) I I* CN Population CN Population Δτ Δτ Δτ Δτ (ps) Δτ Δτ Δτ Δτ (ps) FIG. 4.5. Transient a) I, I* and b) CN population from MD simulations, insets show magnification of population decay over the first picosecond. 4.6.3 Results from 3 Π 1 For trajectories starting on the 3 Π 1 excited state, only dissociation as CN and ground state I is possible, thus using the same sorting criteria, every free CN corresponds to a free iodine atom in the ground state. The I/CN population as a function of time is shown in figure 4.6a with 17% of the trajectories surviving after 5 ps. Interestingly, there is only a very small amount of fast cage recombination observed (~5%) over the first 500 fs indicating that most of the 100 CN molecules must survive on the flat section of the potential surface after collision with the solvent before quickly recombining on the ground state over the next picosecond. 0 1 2 3 4 5 0.0 0.2 0.4 0.6 0.8 1.0 I/CN Population Δτ Δτ Δτ Δτ (ps) FIG. 4.6. MD Population decay of I and CN from excitation to 3 Π 1 surface. 4.7 Discussion This discussion will address many aspects of the ICN photodissociation results. We will begin by discussing the assignment of the transient absorption spectra. We will interpret the time-dependant features of the spectrum using a simple kinetic model to compare isotropic data. We will use the results from the molecular dynamics simulations to aid us in our interpretation of the transient behavior and make a direct comparison to the experimental observables. Finally 101 we will summarize our results by discussing the striking differences between photodissociation in water and ethanol and the role of both molecular collisions and electrostatic effects which we can infer from our data. 4.7.1 Assignments for ICN in ethanol The transient spectra obtained in ethanol reveal the most information as transient species evolve on an apparently much slower time scale than in water, as well as within our probe window. We will begin by making assignments of these features first and then discuss the time dependent nature of these species. The absorption signal which appears at between 385 and 400 nm in polar solvents has been assigned to “gas-like” CN radicals in previous work. 17 The revelation in this work of a distinct and sharp transient peak in this spectral region clearly agrees with that assignment which is also in the exact position corresponding to the gas-phase (although substantially broader than the actual gas phase spectrum) CN B←X transition shown in Figure 4.2. The broad tail which stretches across the visible probe window (450–600 nm) is in excellent agreement with the predicted and recently measured I* CT absorption peaking at ~500 nm. The decay of this absorption band also coincides with a rise in absorption observed at the bluest probe wavelengths which is precisely where ground state I CT is expected to absorb (360 nm). Therefore, we assign these bands to I and I* charge-transfer absorption. 102 4.7.2 Assignments for ICN in water In water, the assignments are more difficult to definitively assign. Very similar transient signatures are observed as in ethanol. A very distinct absorption peak at early times corresponds to the same absorption species assigned to the CN B←X transition observed in ethanol. As the center frequency and bandwidth does not seem affected by the solvent at early times, this reaffirms the assignment that this corresponds to an unrelaxed CN molecule. The absorption from I and I* however are not as clearly resolved. As in ethanol, a long tail stretches across the visible spectrum from 450–600 nm. This feature decays very quickly over the first 300 fs (Fig. 4.4) and a corresponding rise is observed at the bluest probe wavelengths. However, the I* CT absorption is expected at ~350 nm in water and thus, it is not apparent that this absorption can be assigned to I* as readily as in ethanol. In addition, some of this rise at 330 nm may also be due to blue shifting of the CN absorption band upon solvation as indicated in Figure 4.2 (blue spectrum). A similar absorption feature between 450‒ 600 nm was observed in previous work by Larsen et. al. which appears as a shoulder on the large solvated electron signal generated by two-photon ionization of the solvent. They attribute this to CN absorption although overlap with the electron spectrum made it difficult to evaluate the transient lifetime of this species and no corresponding rises were observed in at 330 nm as in our experiment due to their time resolution. Using Mulliken’s theory of charge transfer, a CN charge transfer absorption is estimated to peak at below 550 nm which we believe could account 103 for this absorption. 17, 49 Although, the fast disappearance of this band is not what we would expect to observe for relaxing CN undergoing solvation. We would instead expect to see a corresponding rise as the signal at 385 nm decreases. It is possible the CN charge transfer band could be disappearing via an abstraction reaction with the solvent where CN removes a hydrogen from water to form HCN. This scenario also seems improbable as the H–CN and H–OH gas phase bond energies are nearly identical at 518 kJ/mol and 498 kJ/mol although we cannot rule out this scenario without probing for these transient photoproducts. 55 In other experiments which probe HCN formation from CN abstraction from CH 2 Cl 2 , which has a weaker H–CHCl 2 bond energy of 414 kJ/mol, the time scale for HCN growth is ~313 ps. 56, 57 In chloroform, the abstraction rate was measured to 194 ps. 58 Crowther et. al. actually observe two timescales for CN signal, 4 ps and 1500 ps, they attribute the 4 ps decay to a combination of caging and diffusional recombination, and they attribute the 1500 ps time constant to CN–solvent complex formation. The 313 ps rise observed in the HCN signal is attributed to certain complexes that quickly go on to form HCN which provides an explanation for the two time scales. Formation of HCN as fast as 300 fs needed to account for our observed decay, has never been observed. Although the experiments that follow HCN appearance do not support fast abstraction as being responsible for the fast decay observed in our experiment, the CN transient spectrum for 5 ps delay presented by Crowther et. al. does also display a long visible tail which stretches across the probe region beyond 650 104 nm. 57 This absorption is nearly identical to that which we observed in water but they do not make an assignment of this feature, nor measure the transient dynamics beyond 400 nm. It is possible that the absorption feature in the visible could come from excited state absorption (ESA) from the Ã continuum to a higher lying excited state of ICN. If this were the case, the most likely scenario is that molecules are trapped in the weak minimum of the 3 Π 0+ surface and essentially wait there for later curve crossing. The most likely candidate for this absorption would be to the B state which lies ~2.17 eV above the Franck Condon region of the Ã continuum. 59 Without a resolved ESA spectrum and due to the decay time scale, the “best” explanation for this feature appears to be from I* absorption which is significantly red shifted from its expected position of 330 nm seen in Figure 4.2a. We can justify this tentative assignment using a simple kinetic model discussed in the next section but currently have no explanation for the observation of such a large red shift of this band. 4.7.3 Kinetic modeling Modeling the photodissociation dynamics for this reaction are remarkably complex, with multiple broad absorptions spanning the probe wavelengths, shifting absorption bands, various reaction pathways and nonequilibrium energy release. We begin this analysis by building a kinetic model based on previous spectroscopic evidence, our spectral assignments and the exponential decays observed in Table 4.1. 105 TABLE 4.1 Isotropic data fit parameters for ICN photodissociation in water and ethanol. Transient absorption was fit with multi-exponential functions at 330, 385, and 500 nm. Probe (nm) A fast (%) T fast (ps) A 1 (%) T 1 (ps) A 2 (%) T 2 (ps) A ∞ (%) H 2 O 330 43(rise) ± 6 0.08 ± 0.01 50 ± 1 6.0 ± 0.2 6 ± 1 385 58 ± 1 0.52 ± 0.02 36 ± 1 6.5 ± 0.5 5 ± 1 500 58 ± 7 0.10 ± 0.01 35 ± 1 1.6 ± 0.1 7 ± 3 EtOH 330 46 ± 10 0.06 ± 0.01 17 (rise) ± 1 1.0 ± 0.1 26 ± 1 12 ± 2 10 ± 1 385 45 ± 2 0.06 ± 0.01 7 ± 1 1.4 ± 0.4 29 ± 1 14 ± 2 18 ± 2 500 74 ± 5 0.06 ± 0.01 19± 1 2.9 ± 0.1 7 ± 1 The kinetic scheme for ICN in ethanol is shown in Figure 4.7. In this model, instantaneous populations are assigned to one of three initial dissociation channels: i) cage recombination, ii) cage escape and iii) permanent escape. Based on the predicted transient spectra in Figure 4.2b, probe wavelengths of 330 and 550 nm were chosen to represent the relative I and I* decays as these wavelengths do not significantly overlap with the expected CN spectrum and both lie near the center of the predicted I and I* absorption bands. Exponential decays were set based on exponential decays presented in Table 4.1. The caging for both I+CN and I*+CN was set to 60 fs and the diffusional recombination time was set to 16 ps. There is an assumption made that CT bands of I and I* appear instantaneously. From our previous photodetachment experiments we see the 106 immediate appearance of the CT absorption in both water and ethanol. Other experiments by Elles et al have shown that Cl radicals exhibit a small red shift and broadening upon solvation but we do not see clear evidence that this is occurring in our ICN experiments. The exponential fits of the rise observed at 330 nm resulted in a 1 ps time constant and the decay at 500 nm to a 2.9 ps time constant. We found that in the kinetic model the rise at 330 nm and decay at 550 nm fit well to a 1.6 ps time constant for I* to I conversion. Relaxation of I* from both cage recombination and diffusion is assumed to feed the channel which corresponds to diffusive recombination on the ground state (i.e. no cage recoil is allowed for molecules that do not undergo a nonadiabatic transition on the first pass through the curve–crossing region). This would lead to an underestimate of the amount of fast recombination we observe in the I signal but modeling this would involve adding another parameter which would account for splitting the I* decay channels to different population of I caging and diffusion. In practice we found that this simpler model was effective for reproducing the experimental signals and thus, we will interpret our results under the assumption that the majority of later nonadiabatic transitions leads do diffusive recombination. The initial relative populations of each channel were adjusted such that a simultaneous fit of the two experimental probe wavelengths was achieved. The relative instantaneous populations for each channel are shown in Fig. 4.7 and are as follows: i) cage recombination, I = 39%, I* = 19%, ii) cage escape, I = 10%, I* = 22% and iii) permanent escape, I = 5%, and I* = 5%. 107 FIG. 4.7. Kinetic scheme for I and I* populations in ethanol. Red arrows indicate processes which are assumed to happen instantaneously. Using this model and the spectra shown for I and I* in Figure 4.2, we can reconstruct the I and I* signals and subtract them from the entire 2D data set presented in Figure 4.3b, revealing the residual absorption presumably from CN only. Figure 4.8 shows the relative fits to the transient signal at 330 and 550 nm and the residual signal left after subtraction of the I and I*. The sharp CN absorption band remaining, which peaks at 390 nm can be seen decaying and slightly broadening over the 20 ps experiment. A very slight red shift is also observed to 400 nm at 20 ps. At early times, the model slightly underestimates the I and I* populations assigned to cage recombination and the model slightly overestimates the I population at 5 ps but overall this simple model reproduces the experimental transient I and I* features remarkably well. Using this residual 108 spectrum we can estimate the approximate band width and maximum extinction coefficient for CN. At 100 fs we estimate an extinction coefficient to be 650 M – 1 cm –1 with a FWHM of 0.47 eV. If we scale the calculated CN(aq) spectrum in Figure 4.2a by increasing the bandwidth to match our experimental value, the maximum extinction becomes ~750 M –1 cm –1 which agrees very well with our measurement. Furthermore, our recent photodetachment experiments in ethanol indicate that the estimated extinction coefficient for I and I* in ethanol may be as much as 60% larger than reported from flash photolysis studies which would also lead to an increase in our estimated CN extinction coefficient. This model also allows for the estimation of the I and I* branching ratios from the initial populations used. The relative populations from Figure 4.7 are 46% I* to 54% I. The initial populations, however, are estimated based on the transient signal from our experiment which has 40 fs time resolution. Excited ICN molecules are expected to reach the curve crossing region within 30 fs and reach the first solvent shell in ~50 fs. 15, 17, 21 This means that our initial populations used in the kinetic model do not account for the actual starting populations but rather, the population after passing the curve crossing region. For the equivalent gas phase photodissociation energy of 308 nm, estimates from β anisotropy parameter, estimate the population excited to the 3 Π 0+ surface to be 70–87%. 60 Making the assumption that the excitation with 266 nm produces the same initial population, we can estimate that 34–47% of ICN molecules which are initially excited to 3 Π 0+ curve cross on the first pass through the curve 109 crossing region. This is in fairly good agreement with the condensed phase molecular dynamics simulations which predict that 29% curve cross on the first pass in water. 15 An estimation of branching ratio from the models presented by Amatatsu et al weighted to 70% 3 Π 0+ excitation, yields a branching ratio of 58% I* although this model is not useful for examining the first pass branching ratio, only the final outcome branching ratio. 350 400 450 500 550 600 0.0 0.2 0.4 0.6 0.8 1.0 0 5 10 15 20 0.0 0.2 0.4 0.6 0.8 1.0 Δ Δ Δ ΔA (mOD) Wavelength (nm) 70 fs 100 fs 500 fs 1 ps 5 ps 10 ps 20 ps 550 nm 330 nm Δ Δ Δ ΔA (mOD) Δτ Δτ Δτ Δτ (ps) FIG. 4.8. a) Fits from kinetic model to 330 and 550 nm experimental data and b) residual CN transient absorption remaining after subtraction of I and I* contributions. The same kinetic model was used in water although the fit could not be done as rigorously since only the tail of I CT band can be observed and we expect the CN B←X transition to shift to the blue making it difficult to separate the transient signal coming from just one species. We can however make some assumptions based on our observations in ethanol and the transients decay fits from Table 4.1. To approximate the initial concentrations of I and I* the 60 fs transient absorption spectrum was used to estimate the amount I* from the 520 110 nm absorption feature and CN using the 650 M –1 cm –1 estimation for the extinction coefficient at 385 nm. The amount of I to I* is 61% to 39% respectively (compared to 54% and 47% in ethanol) and the combined I and I* spectra are shown with the 50 fs transient signal in Figure 4.9. Using the time constant and percentages for the 500 nm decay in Table 4.1, the caging for I and I* was set to 100 fs and diffusive relaxation to I was assigned a 1.6 ps time constant. The diffusive recombination time for I was assigned to 6 ps since this decay was observed at both 330 nm and 385 nm. The relative I cage and diffusive populations were impossible to estimate precisely without probing at higher energies although the CN population must agree stoichiometrically with the amount of I and I* population at all times. Unfortunately, this is also difficult to ascertain as the entire CN spectrum must also be resolved. We have chosen a cage recombination fraction of 34% and diffusive fraction of 65% estimated the amount of transient CN signal at 50 fs. Since we do not observe resolved I and CN populations from the experiment, we cannot provide a unique fit, but based on the available data we present a solution where the instantaneous populations in the kinetic model were set as follows: i) cage recombination, I = 11%, I* = 24%, ii) cage escape, I = 42%, I* = 13% and iii) permanent escape, I = 8%, and I* = 2%. 111 350 400 450 500 550 600 0.0 0.2 0.4 0.6 0.8 1.0 200 300 400 500 600 0.0 0.2 0.4 0.6 0.8 1.0 (b) 350 400 0.0 0.2 0.4 0.6 Δ Δ Δ ΔA (mOD) Wavelength (nm) 100 fs 200 fs 300 fs 400 fs 500 fs Δ Δ Δ ΔA (mOD) Wavelength (nm) 70 fs 100 fs 500 fs 1 ps 5 ps 10 ps 20 ps (a) CN I I* Δ Δ Δ ΔA (mOD) Wavelength (nm) FIG. 4.9. a) Fit of the 50 fs spectral cut using I, I* and CN (CN not included in kinetic model), and b) residual CN absorption in water after I and I* contributions are removed. The first 500 fs is highlighted (inset) to show the fast shifting of the CN absorption. The remaining CN residual signal is shown in Figure 4.9b. We can see that the CN signal observed at 385 nm quickly shifts to higher transitions energies over the first 500 fs (inset Fig. 4.9b). This is expected for solvation of 112 CN in water but the residual tail indicates that not all of the CN is solvated as fast. It is interesting to note that the 100 fs rise at 330 nm seems to be due almost entirely to the shifting CN signal and this is the first time that this behavior has been observed from condensed phase photodissociation of ICN. 4.8 Comparison to MD The population decays extracted from the MD simulations and shown in Figures 4.5 and 4.6, are used to reconstruct the transient 2D data sets by applying a Gaussian function to each species and summing across the spectrum. The results for water and ethanol are discussed in this section. 4.8.1 Water Gaussian functions representing the different populations were generated from the assignments made in previous sections. The peaks corresponding to I and I* were centered at 4.7 eV (263 nm) and 2.4 eV (516 nm) respectively with FWHM of 1.2 eV. The CN was described by both “hot” and “cold” distributions where “hot” CN corresponded to trajectories that start on the 3 Π 0+ potential energy surface and then curve cross to 1 Π 1 , while the “cold” CN corresponds to trajectories started stay on the 3 Π 0+ surface or start on the 3 Π 1 surface. The cold CN absorption was centered at 3.8 eV (326 nm) and has a 1.2 eV FWHM, while the “hot” CN is centered at 3.25 eV (381 nm) and has a much narrower 0.47 eV FWHM. A cooling function consisting of an exponential decay with a 350 fs time constant was used to transfer “hot” CN to “cold” CN as a function of time. 113 This was the only modification made to the simulation results. The intensities of each band were defined such that the overall integrated intensity was held constant and no adjustments were made for individual transitions. An equal amount of 3 Π 0+ and 3 Π 1 trajectories were used as in previous analysis of the MD simulations. A comparison between the experimental contour over 1 ps and 5 ps and the simulated spectra are shown in Figure 4.10. The simulation does a fairly good job reproducing the transient signals observed over the first picosecond. We did not include a spectral shift of the 390 nm CN band into the simulation results which is clear in the experimental data over the first 500 fs but simply converting “hot” CN to “cold” CN does reproduce the relative intensities and position of the transient absorption features. The most obvious difference between the simulations and experiments arise from the overall faster decay observed of all the transient signals after 1 ps. The blue side of the experimental data shows a much longer decay over the 5 ps time window. This would indicate that the diffusive recombination in water occurs slower than predicted by the simulation. As trajectories approach the ground state asymptote on either the 3 Π 1 or the 1 Π 1 , a simplification in the code transfers the trajectory to the 1 Σ + ground state when the energy gap is less than kT. This results in slightly faster recombination than expected and may be responsible for some of the discrepancy. It is very likely that this faster decay is simply comes from a slight underestimate of the translational kinetic energy release imparted on the CN fragment. A more thorough analysis which includes simulations done at multiple 114 pump wavelengths is forthcoming and may shed light on the relative product energy distributions and how they compare to experimentally measured CN decay. 61 350 400 450 500 550 600 0.0 0.5 1.0 Δτ Δτ Δτ Δτ (ps) 350 400 450 500 550 600 0 1 2 3 4 5 (d) (c) (b) (a) Wavelength (nm) Wavelength (nm) Δ Δ Δ ΔA (mOD) Δτ Δτ Δτ Δτ (ps) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 350 400 450 500 550 600 0 1 2 3 4 5 Δτ Δτ Δτ Δτ (ps) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 350 400 450 500 550 600 0.0 0.5 1.0 Δτ Δτ Δτ Δτ (ps) Δ Δ Δ ΔA (mOD) FIG. 4.10. Comparison of a) water experimental spectra over 5 ps and b) reconstruction (see text) of transient spectra from MD simulation of transient populations. Magnification of c) the initial picosecond of ICN in water is compared to d) the first picosecond of the MD reconstruction. 4.8.2 Ethanol The transient decays observed in ethanol are much slower, almost a factor 4 in T 1 , so directly comparing the MD simulation results to water is not going to reproduce the results we observe in the experimental measurement. Therefore we 115 have stretched MD simulation uniformly in time in order to match the timescales observed in ethanol. We do not expect that any of the timescales, especially cage recombination to scale linearly with respect to one another. However, it is worth extending at least the diffusional part of the simulation and making a comparison to our experimental results. In the case of ethanol, the center wavelengths of the I and I* transitions were exactly the same as Figure 4.2 and had FWHMs of 1 eV. The “hot” and “cold” CN were set to 3.2 and 3.0 eV respectively. The FWHM of the “hot” CN band was slightly narrower than water and set to 0.35 eV, while the “cold” CN was also narrower at 0.5 eV. The hot CN extinction was reduced by 50% to that of I and I* while the cold CN was reduced by 30% (reflecting the change in bandwidth). The simulated spectra are compared to the experimental spectra in Figure 4.11. In order to produce this spectrum from the simulation is to extend the time scale by a factor of six, including the cooling function. This is greater than the increase in either T 1 or T 2 between water and ethanol, but due to the faster decays across the spectrum observed in the water simulation, it is not surprising that the best fit comes from overstretching. Again, the simulation does a reasonably good job reproducing the experimental results. From the simulation, we see a slightly faster decay of transient signal between 325 and 450 nm, but longer decay between 500 and 600 nm. This indicates that the I* to I conversion rate is most likely underestimated when stretching out the simulations. This is also the reason why the pronounced 116 rise noted at 340 nm in the experiment, is absent from the simulated plot. We also observe that the caging decay in the simulation is too long which is reasonable since we do not expect that the caging dynamics should be markedly longer in ethanol, especially by a factor of 6, although the ethanol cage should be slightly more relaxed. 350 400 450 500 550 600 0 1 2 3 4 5 Δτ Δτ Δτ Δτ (ps) 350 400 450 500 550 600 0 5 10 15 20 (b) (a) Δ Δ Δ ΔA (mOD) Δτ Δτ Δτ Δτ (ps) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 350 400 450 500 550 600 0 5 10 15 20 (c) Δτ Δτ Δτ Δτ (ps) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 350 400 450 500 550 600 0 1 2 3 4 5 Wavelength (nm) Wavelength (nm) (d) Δτ Δτ Δτ Δτ (ps) Δ Δ Δ ΔA (mOD) FIG. 4.11. Comparison of a) ethanol experimental spectra over 20 ps and b) reconstruction (see text) of transient spectra from MD simulation of transient populations stretched in time by a factor of 6. Magnification of c) the initial 5 ps of ICN in ethanol is compared to d) the (stretched) first 5 ps of the MD reconstruction. 117 4.9 Conclusions In summary, several important and previously unobserved dynamics of ICN photodissociation were made possible by the coupling of sub 40 fs DUV pulses and simultaneously collected broadband probe pulses. The transient spectra display signatures from all three radical transient species, I, I* and CN in ethanol. The CN band is resolved for the first time and a sharp feature at early times indicates that CN is highly rotationally excited. Over the first few picoseconds this absorption broadens and decays as solvation and secondary recombination take over. Through kinetic modeling the first pass curve crossing is estimated to be 34-47% which is good agreement with MD simulations. In water, we also observe a signature from rotationally excited CN and a clear blue shift is observed over the fist 500 fs. The blue shift of the solvated CN species is expected from EOM-CCSD calculation and this is the first time this signature has been observed experimentally. With the significantly improved time resolution, cage recombination of the transient photoproducts is observed with a sub 100 fs time constant and appears as a decay of all transient species in our probing window. Reanalysis of the semiclassical molecular dynamics simulations using a sorting criteria which removes recombined trajectories from ensemble averages, is used to reproduce the free I, I* and CN populations. The transient populations are analogous to our isotropic experimental spectra making the MD results directly comparable to our experimentally measured spectra for the first time. 118 At early times the MD clearly displays a fast decay from cage recombination and a 27% first pass probability. Using a cooling function to describe the transition from hot to cold CN and our transient assignments to reproduce the 2D data, we see excellent agreement with the water experiment. Although, the MD slightly overestimates the diffusional recombination timescales, this may be an artifact of the transition between the excited states and ground state at the asymptotic limit. The effect of the increasing the pump energy with be the topic of the next chapter but these experiments and transient assignments help solidify assignments made in previous work, including the assignment of CN absorption at 385 nm and persistent free rotation of this species. 119 4.10 References for Chapter 4 1. R. W. Anderson and R. M. Hochstrasser, J. Phys. Chem., 1976, 80, 2155- 2159. 2. A. Levy, D. Meyerstein and M. Ottolenghi, J. Phys. Chem, 1973, 77, 3044-3047. 3. M. Berg, A. L. Harris and C. B. Harris, Phys. Rev. Lett., 1985, 54, 951- 954. 4. R. Lingle, X. B. Xu, S. C. Yu, Y. J. Chang and J. B. Hopkins, J. Chem. Phys., 1990, 92, 4628-4630. 5. C. G. Elles, M. J. Cox, G. L. Barnes and F. F. Crim, J. Phys. Chem. A, 2004, 108, 10973-10979. 6. H. Bursing, J. Lindner, S. Hess and P. Vohringer, Appl. Phys. B-Lasers O., 2000, 71, 411-417. 7. S. Hess, H. Bursing and P. Vohringer, J. Chem. Phys., 1999, 111, 5461- 5473. 8. A. Baratz and S. Ruhman, Chem. Phys. Lett., 2008, 461, 211-217. 9. P. Fornier de Violet, R. Bonneau and J. Joussot-Dubien, Mol. Photochem., 1973, 5, 61-67. 10. C. Z. Wan, M. Gupta and A. H. Zewail, Chem. Phys. Lett., 1996, 256, 279-287. 11. P. S. Dardi and J. S. Dahler, J. Chem. Phys., 1993, 98, 363-372. 12. I. Benjamin, J. Chem. Phys., 1995, 103, 2459-2471. 13. I. Benjamin, U. Banin and S. Ruhman, J. Chem. Phys., 1993, 98, 8337- 8340. 14. I. Benjamin and K. R. Wilson, J. Chem. Phys., 1989, 90, 4176-4197. 15. N. Winter, I. Chorny, J. Vieceli and I. Benjamin, J. Chem. Phys., 2003, 119, 2127-2143. 120 16. J. Larsen, D. Madsen, J. A. Poulsen, T. D. Poulsen, S. R. Keiding and J. Thøgersen, J. Chem. Phys., 2002, 116, 7997-8005. 17. A. C. Moskun and S. E. Bradforth, J. Chem. Phys., 2003, 119, 4500-4515. 18. A. C. Moskun, A. E. Jailaubekov, S. E. Bradforth, G. H. Tao and R. M. Stratt, Science, 2006, 311, 1907-1911. 19. N. Pugliano, S. Gnanakaran and R. M. Hochstrasser, J. Photoch. Photobio. A, 1996, 102, 21-28. 20. W. P. Hess and S. R. Leone, J. Chem. Phys., 1987, 86, 3773-3780. 21. Y. Amatatsu, S. Yabushita and K. Morokuma, J. Chem. Phys., 1994, 100, 4894-4909. 22. A. C. Moskun, S. E. Bradforth, J. Thogersen and S. Keiding, J. Phys. Chem. A, 2006, 110, 10947-10955. 23. P. A. Pieniazek, S. E. Bradforth and A. I. Krylov, J. Phys. Chem. A, 2006, 110, 4854-4865. 24. A. E. Jailaubekov and S. E. Bradforth, Appl. Phys. Lett., 2005, 87, 021107. 25. C. A. Rivera, S. E. Bradforth and G. Tempea, Opt. Express, 2010, 18, 18615-18624. 26. J. C. Tully, J. Chem. Phys., 1990, 93, 1061-1071. 27. C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys., 2009, 130, 084501. 28. X. Y. Chen, D. S. Larsen, S. E. Bradforth and I. H. M. Stokkum, J. Phys. Chem. A, 2010, submitted. 29. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum., 2003, 74, 4958-4960. 30. A. P. Baronavski, Chem. Phys., 1982, 66, 217-225. 31. E. M. Goldfield, P. L. Houston and G. S. Ezra, J. Chem. Phys., 1986, 84, 3120-3129. 32. M. D. Pattengill, Chem. Phys., 1984, 87, 419-429. 121 33. B. A. Waite, H. Helvajian, B. I. Dunlap and A. P. Baronavski, Chem. Phys. Lett., 1984, 111, 544-548. 34. R. Baer and R. Kosloff, Chem. Phys. Lett., 1992, 200, 183-191. 35. J. A. Beswick, M. Glassmaujean and O. Roncero, J. Chem. Phys., 1992, 96, 7514-7527. 36. R. D. Coalson and M. Karplus, J. Chem. Phys., 1990, 93, 3919-3930. 37. S. Y. Lee, J. Chem. Phys., 1992, 97, 227-235. 38. J. W. Qian, C. J. Williams and D. J. Tannor, J. Chem. Phys., 1992, 97, 6300-6308. 39. C. J. Williams, J. W. Qian and D. J. Tannor, J. Chem. Phys., 1991, 95, 1721-1737. 40. M. L. Johnson and I. Benjamin, J. Phys. Chem. A, 2009, 113, 7403-7411. 41. J. Vieceli, I. Chorny and I. Benjamin, J. Chem. Phys., 2001, 115, 4819- 4828. 42. J. Vieceli, I. Chorny and I. Benjamin, Chem. Phys. Lett., 2002, 364, 446- 453. 43. N. Winter and I. Benjamin, J. Chem. Phys., 2004, 121, 2253-2263. 44. K. Kuchitsu and Y. Morino, B. Chem. Soc. Jpn., 1965, 38, 814-824. 45. D. L. Thompson, Modern methods for multidimensional dynamics computations in chemistry, World Scientific, Singapore; River Edge, NJ, 1998. 46. W. M. Pitts and A. P. Baronavski, Chem. Phys. Lett., 1980, 71, 395-399. 47. P. Fornier de Violet, Rev. Chem. Intermed., 1981, 4, 121-169. 48. T. A. Gover and G. Porter, Proc. R. Soc. Lon. Ser.-A, 1961, 262, 476-488. 49. R. S. Mulliken, J. Phys. Chem., 1952, 56, 801-822. 50. C. A. Rivera, Y. Zhang, N. Ottosson, B. Winter and B. S. E., in prep., 2010. 122 51. H. F. Schaefer and T. G. Heil, J. Chem. Phys., 1971, 54, 2573-2580. 52. D. D. Davis and H. Okabe, J. Chem. Phys., 1968, 49, 5526-5531. 53. R. Thomson and F. W. Dalby, Can. J. Phys., 1968, 46, 2815. 54. CRC Handbook of Chemistry and Physics, CRC Press, Cleveland, Ohio, 2010. 55. D. F. McMillen and D. M. Golden, Annu. Rev. Phys. Chem., 1982, 33, 493-532. 56. A. C. Crowther, S. L. Carrier, T. J. Preston and F. F. Crim, J. Phys. Chem. A, 2008, 112, 12081-12089. 57. A. C. Crowther, S. L. Carrier, T. J. Preston and F. F. Crim, J. Phys. Chem. A, 2009, 113, 3758-3764. 58. D. Raftery, E. Gooding, A. Romanovsky and R. M. Hochstrasser, J. Chem. Phys., 1994, 101, 8572-8579. 59. W. S. Felps, K. Rupnik and S. P. McGlynn, J. Phys. Chem., 1991, 95, 639-656. 60. S. W. North, J. Mueller and G. E. Hall, Chem. Phys. Lett., 1997, 276, 103- 109. 61. C. A. Rivera, S. E. Bradforth, N. Winter, R. Harper and I. Benjamin, in prep, 2011. 123 Chapter 5. The effect of increasing excitation energy on the condensed phase ICN photodissociation reaction In collaboration with: Nicolas Winter, Rachael V. Harper and Ilan Benjamin Department of Chemistry, University of California, Santa Cruz, California, 95064 Abstract for Chapter 5 The benchmark photodissociation reaction of ICN is investigated in ethanol using five photolysis wavelengths which span the red side of the ICN Ã absorption spectrum. The decay of the transient signal at 385 nm is due primarily to absorption from CN radicals and thus the decay lifetimes describe the dynamical properties of the system which include cage recombination, diffusion, abstraction. Anisotropic studies done at four of these wavelengths also indicate long lived rotational coherence of the CN photofragment. The large amount of rotational energy is due to transitions through a well known conical intersection. As we have seen in previous studies, relaxation of the rotationally hot photoproduct leads to broadening of the CN absorption band as a function of time which should also correspond to a decrease in the peak intensity of the transient signal. Semiclassical MD performed for ICN in water has already been shown to reproduce the isotropic dynamics of the transient species. A more comprehensive analysis of the isotropic CN transient decay and rotational dynamics using multiple photolysis wavelengths is presented here along 124 with the results of the MD simulations using photolysis energies of 266, 255 and 233 nm. 5.1 Introduction In the previous chapter, we focussed on the isotropic dynamics observed from photolysis of ICN with 266 nm photons. That work examined the effects of changing solvents from water to ethanol. Kinetic models and molecular dynamics simulations (MD) were used to make assignments to the transient spectra observed. While in water the assignment of some of the transient species is not clear cut, the sharp feature observed in both solvents clearly corresponds to CN radicals. Furthermore, the sharp, almost Lorentzian shaped band and subsequent broadening (and significant blue shift in water) is indicative of the fact that the CN is highly rotationally excited through a nonadiabatic transition (see Fig 5.1). The details of this transition and the effects of the solvent on curve crossing dynamics have been addressed however the effect of photolysis wavelength on curve crossing is still an open question. It is clear from gas phase measurements that increasing the pump energy decreases the curve crossing probability, but increases the average rotational energy for molecules that do cross (Table 5.1). 1 The larger the pump energy the higher up in the Frank Condon region the molecules begin and the faster they move through the curve crossing region. According to Landau-Zener, this should lead to decreased curve crossing probability. 2, 3 For molecules that do curve cross, the increased available energy results in a larger fraction partitioned into the rotational channel. 4 In the condensed phase, the surrounding solvent complicates the matter. 125 Solvent friction from electrostatic forces could effectively slow down molecules as they reach the curve crossing region, thereby increasing the probability of making a transition and negate the effect of increasing pump energy. Collision with the solvent cage and secondary diffusive recombination lead to multiple passes through the curve crossing region and thus may contribute to curve crossing at much longer delay times. 5, 6 The combined effects of the solvent environment and the increased pump energy will be the primary focus of this chapter although the findings may have significant impact on previous analyses. 2 3 4 5 6 -4 -2 0 2 4 1 Σ Σ Σ Σ + 0+ 3 Π Π Π Π 1 1 Π Π Π Π 1 3 Π Π Π Π 0+ rI-CN (Å) E (eV) FIG. 5.1. Potential energy curves corresponding to the ground state (black line) and three lowest excited states (colored lines) reproduced from ref. 8 and plotted as a function of I to CN center- of-mass and I-CN angle of 0 degrees. The dashed line corresponds to the approximate solvation shift of the ground state. 126 TABLE 5.1 Relevant energetics and branching ratios from gas phase photodissociation of ICN reproduced from reference 4. I* yield 44 66 58 < 30 < 20 0 I* channel <N> 12 10 5 7 Not accessible <E rot > (cm -1 ) 276 172 51 96 I channel <N> 40 42 36 31 26 <E rot > (cm -1 ) 3039 3274 2466 1788 1252 % parallel 86 85 77 83 > 85 70 - 87 λ exc (nm) H 2 O/ alcohols 221 235 245 253 264 267 λ exc (nm) gas 248 266 280 290 304 308 E above D 0 (cm -1 ) 15240 12512 10632 9400 7812 7386 127 As with the investigation of the isotropic dynamics, the semiclassical molecular dynamics simulations developed by Benajmin et al. 5-7 provide a parallel result which helps us identify and separate some of the features we observe in our experimental data. These simulations which include the ICN potentials calculated by Morokuma et al., 8 Tully’s surface hopping correction 9 and fully flexible water molecules 10 have been shown to reproduce the experimental isotropic dynamics very well but no detailed investigation of the effect of changing pump energy or comparison of the free CN rotational dynamics have not been attempted until this work. It is the goal of this chapter to connect almost a decade worth of work done in our lab 4, 11 and compare the broadest range of isotropic and anisotropic data available for this historically important system. 5.2 Experimental Several experimental layouts and laser systems were employed to produce the experimental data presented in this work. The four photodissociation pump wavelengths referred to in this work consist of 271, 266, 255, 233 and 224 nm. For brevity, we will refer to the experimental sections of previous work for a detailed description of the experimental conditions used for pump wavelengths of 255 and 224 nm. 4 Previous work has also been published for the photodissociation reaction at 233 nm by Moskun et al. 11 although that work primarily dealt with transient rotational dynamics and the isotropic dynamics have not been previously reported until now. The light generated at 266 nm was described in the previous chapter and 128 was also used to measure the experimental anisotropic decay which is reported in this work. 5.2.1 271 nm A 300 μJ portion of a 1kHz amplifier system (Coherent Legend Elite USP- HE, 3.5 mJ, 800 nm, 36 fs pulse width) was frequency doubled to produce 70 μJ of 400 nm. The residual fundamental and 400 nm was combined in a FWM hollow- core fiber to produce 4.5 μJ of 271 nm. 12 The difference in center frequency from pulses generated at 266 nm described in the previous chapter is a result of the increased bandwidth of the 36 fs fundamental pulses. The 271 nm was compressed by a novel Gires-Tournois type compressor to produce 32 fs FWHM pulses. 13 This pump beam was chopped to 500 Hz focused to ~100 μm at the sample. 5.2.2 Probe Generation of the probe pulses consisted of focusing of a small portion of the fundamental into a circularly translating piece of 2 mm thick calcium fluoride to generate super-continuum pulses that range from 325-625 nm. 14, 15 After generation in the CaF 2 disk, the super-continuum was collimated and then focused into the sample using a pair of protected aluminum off-axis parabolic mirrors (Janos Technology) to ~60 μm diameter. The polarization of the 800 nm was also controlled prior to continuum generation by a zero-order half-wave waveplate (CVI) and a quartz polarizer cube was used to test that the continuum polarization was better than 200:1 across the entire probe spectrum. Although dispersed transient spectra are obtained here, previous work done at 255, 233, and 224 were only taken 129 using individual probe wavelengths and thus, we will concentrate on the identical portion of the transient spectrum as the earlier work for this chapter. 5.2.3 Sample and spectrometer The method for ICN synthesis was presented earlier 4 and reproduced in Chapter 4. For the work presented here from 271 nm photolysis, 260 mM solution of ICN in ethanol (200 proof, KOPTEC) were used for all experiments. This solution was delivered to the laser beams by a flowing thin-film wire guided gravity drop jet. 16 The liquid film thickness was ~60 μJ. The pump and probe beams are overlapped spatially and temporally in the flowing sample to generate the pump- probe transient signals. For dispersed pump-probe measurements, the probe is collimated after the sample and subsequently focused into a spectrometer. The details of the spectrometer have also been described in detail previously. 14, 15 A UV diffraction grating (300 gr/mm, 300 nm blaze, Oriel) is used to disperse the probe onto a 256 pixel diode array. Transient dynamics were measured out to 100 ps for 271 nm pumped experiments and the anisotropy was measured to 5 ps. The temporal delay between the pump and probe pulses is achieved by a computer controlled motorized linear delay stage (Newport) mounted with a retro-reflector in the path of the probe before super-continuum generation. In order to collect the isotropic transient dynamics either the probe polarization was rotated independently until the polarization angle between pulses was 54.7 o (magic angle). 4 For evaluation of the rotational anisotropy function certain spectral regions of the transient signal are 130 selected using three separate bandpass filters; 355 nm (10 nm FWHM), 400 nm (25 nm FWHM) and 450 nm (25 nm FWHM). For each different wavelength region the filtered probe light was rotated to 45 degrees with respect to the pump pulse prior to the sample. After the sample the probe light was transmitted through a Wollaston polarizing prism (Karl Lambrecht) in order to separate the parallel and perpendicular contributions to the signal. These two signals were tightly focused onto a diode array such that >90% of the probe light was imaged on a single pixel and both the parallel and perpendicular transient signals were collected simultaneously. The time resolution for all of these experiments was estimated by the cross- correlation of pump and probe pulses using the 1/e 2 width of the two-photon absorption signal from the pure solvent. The time resolution is defined as the point where this signal reaches 1/e 2 of the peak intensity centered at zero. For the 266 nm anisotropy measurements the time resolution was 30 fs. For isotropic measurements made using a 271 nm pump the time resolution was 27 fs. Solvent backgrounds were also taken to ensure that the pump intensity was small enough such that no electrons or other photoproducts were generated by 2-photon absorption (pump+pump) of the solvent. 5.3 Molecular dynamics methods There have been several theoretical studies treating ICN both classically and semiclassically 8, 17-20 and using time-dependent quantum mechanical simulations. 21-26 The high level ab initio potential surfaces calculated by Morokuma et al. 8, 18 in combination with Coulombic and Leonard-Jones solute-solvent and solvent-solvent 131 potentials provided for extension of simulations which include the non-adiabatic transition to a condensed phase environment. 5, 6, 27-30 In the previous chapter we have built upon existing methodology for ICN photodissociation simulations in water at 266 nm thus we refer to this previous work for a detailed description of our semiclassical approach. 6 We have extended this work to encompass three dissociation wavelengths of 233, 255, and 266 nm. The details of the potentials used to govern both intermolecular and intramolecular forces do not depend on the excitation wavelengths used and thus, we will refer to previous work for a detailed description. The simulated system consisted of one ICN molecule and 1000 water molecules confined in a cubic box of fixed pressure at the experimental density at 293 K with three-dimensional periodic boundary conditions. 800-1600 trajectories with 5 fs steps were calculated per excited state depending on the dissociation wavelength. To generate the initial Frank-Condon (FC) conditions for excitation to a given excited state, a 320 ps equilibrium trajectory was initiated. Every 20 ps a different configuration along the equilibrium trajectory is saved for use as an initial configuration. An iterative procedure which scans each of the 16 (for 266 nm two sets of 16) saved initial conditions until a proper FC configuration was found where ћω(r)=V ex (r)-V gr (r) depending on the photodissociation wavelength of interest. From each saved configuration 10 FC configurations were extracted producing 80 FC configurations (160 for 266 nm) per dissociation wavelength. Only the 3 Π 0+ and 132 3 Π 1 excited states were used to produce FC configurations since at all three excitation wavelengths overlap with the 1 Π 1 excited state is very small. 200 300 400 500 600 0 500 1000 1500 2000 probe window ethanol 0.1xCN(g) I( 2 P 1/2 )(aq) I( 2 P 3/2 )(EtOH) ICN(EtOH) ε ε ε ε (M -1 cm -1 ) Wavelength (nm) FIG. 5.2. Transient absorption spectra generated from ICN photodissociation in ethanol. The ICN (black line), I( 2 P 3/2 ) (solid green) from ref. 32, I( 2 P 1/2 ) (dashed green), and gas phase CN B←X (red) reconstructed from ref. 11 and 31 are plotted as a function of wavelength. The colored bar represents the probe wavelengths available in this experiment. The semiclassical nature of these simulations comes from the inclusion of non-adiabatic surface hopping between the 3 Π 0+ and the 1 Π 1 excited states of this system. The surface hopping method developed by Tully, 9 which uses a probability function to determine the chances of a non-adiabatic transition based on the coupling between the excited states calculated by Morokuma et al. 8 and the instantaneous ICN atomic positions, is employed at every step of the MD trajectories. The implementation of this method was described in detail previously 6, 30 and done in exactly the same way. Coupling between the ground and excited states at the 133 asymptotic limit is also included such that when the ground-excited state energy gap is less than kT, the transition probability the ground state becomes 1. 5.4 Results The isotropic and anisotropic decays are presented for ICN photoproducts. The isotropic decays presented here correspond to a photolysis wavelength of 271 nm while the anisotropic decay is generated using a photolysis wavelength of 266 nm. A rigorous comparison between all photolysis wavelengths will be left for the discussions, however it is helpful to compare the isotropic results from 271 nm to the results in Chapter 4 collected at 266 nm as they are nearly identical. 5.4.1 Isotropic results and assignments The dispersed transient spectra displays nearly identical features to those observed using 266 nm and due to decreased signal to noise in the 271 nm spectra, only the 100 fs spectral slice and transient decay at 385 nm are plotted here. At early times (<1 ps) a sharp transient peak is observed at ~385 nm with a long tail stretching out across the visible wavelengths of our probe. The sharp peak displays the same spectral features assigned to rotationally excited CN generated by 266 nm excitation and can be seen clearly in the 100 fs spectrum. The tail on the visible side of the spectrum aligns well with the I* spectrum presented in Figure 5.2 and assignments in Chapter 4. 31, 32 We observe the same 5 ps decay of this band corresponding to another band on the UV side of the spectrum which rises over the first 5 ps. This feature is assigned to iodine radicals and the 5 ps decay in the visible 134 and rise in the UV is assigned to I* to I secondary conversion through the non- adiabatic transition. Although less pronounced than that observed from 266 nm, this is likely due to poorer overlap of the pump and probe beams on the UV side of the spectrum. The 385 nm absorption appears to broaden as a function of time and decays over the 100 ps timescale of the experiment, although this signal is clearly still present at 20 ps as observed previously. The improved time resolution also uncovers an initial fast decay occurring within the first 300 fs across the entire transient spectrum which is assigned to cage recombination although this decay seems slower than the 60 fs decay observed at 266 nm. The origin of this discrepancy will be a topic of discussion in the next section. 350 400 450 500 550 600 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 5 10 15 20 0.0 0.2 0.4 0.6 0.8 1.0 1.2 (a) Δ Δ Δ ΔA, mOD Wavelength, nm (b) Δ Δ Δ ΔA, mOD Time Delay, ps FIG. 5.3. a) Time dependant transient spectral cut at 100 fs for 260 mM ICN in ethanol excited at 271 nm. Contour plots show the time-dependant transient signal increasing in intensity from blue to red. b) The time dependant transient signal measured at 385 nm corresponding to CN signal (black squares) and the fit of a double exponential function (red line). 135 Since only single probe wavelengths are available for most of the photodissociation studies, we only exponentially fit the transient decay at 385 nm and display the transient decay in Fig. 5.3. The results of exponential fitting are presented in Table 5.2. A full comparison is presented in the discussion section, but we can see that only two exponentials are needed to adequately fit the data requiring a 310 fs time constant (44%) and an 11 ps time constant (38%) with 17% surviving at infinite time delay. It should be noted that results using a third-order decay produced time constants almost identical to those obtained at 266 nm although the fastest decay was still slower at ~200 fs. TABLE 5.2 Isotropic data fit parameters for ICN photodissociation. Transient absorption at 385 nm probe. Pump (nm) A fast (%) T fast (ps) A 1 (%) T 1 (ps) A 2 (%) T 2 (ps) A ∞ (%) 271 44 ± 1 0.31 ± 0.04 38± 1 11 ± 1 17 ± 1 266 45 ± 2 0.06 ± 0.01 7 ± 1 1.4 ± 0.4 29 ± 1 14 ± 2 18 ± 2 255 a 42 2.0 42 22 16 233 b 57 ± 25 0.08 ± 0.01 23 ± 1 4.3 ± 0.4 20 ± 1 224 a 46 1.4 32 22 22 a ref. 4, b ref. 1 5.4.2 Anisotropic results The transient pump-probe anisotropy signal depends on the angle between the pumped and probed transition dipole moments of the chromophore. The time- dependent transient anisotropy, R(t), is an ensemble average defined as 136 )] ( [cos( )] ( ) 0 ( [ ) ( 2 5 2 2 5 2 t P t P t R probe pump θ μ μ = ⋅ = 5.1 where μ pump and μ probe correspond to the pumped and probed transition dipole moments, θ is the angle between them and P 2 is the second Legendre polynomial. 4 The collected parallel and perpendicular transient components are used to reconstruct R(t) by the equivalent relationship ) ( 2 ) ( ) ( ) ( ) ( || || t I t I t I t I t R ⊥ ⊥ + − = 5.2 where ) ( || t I the parallel component and ) (t I ⊥ corresponds to the perpendicular component. Since P 2 (cosθ) ranges from 1 to -1/2, when cosθ = 1, R(t) = 0.4 (transition dipoles are parallel) and when cosθ = -1, R(t) = -0.2 (transition dipoles are perpendicular). A more rigorous background with regard to the free-rotor gas phase limit and extended diffusion models has been presented previously 4 and our results will be analyzed in the discussion section with respect to the previous interpretations of the anisotropic measurements. The observed anisotropic decay for 271 nm pump/400 nm probe of ICN in ethanol is shown in Figure 5.4 and 5.5 (black triangles). This transient signal displays the characteristic features previously assigned to a long lived ensemble of rotationally excited CN molecules. The features of this signal can be separated into two main components; a long ~7 ps decay (Figure 5.4) which is assigned to relaxation of the excited CN rotor 4 and the initial fast decay which has been assigned to “free” spinning of the CN radicals 11 which include a dip in the anisotropy decay and subsequent recovery to a value near 137 0.1 (Figure 5.5). The long lived anisotropic signal observed at 225, 233, 255 and 271 nm in Figure 5.4 shows that the long lived signal does not scale with pump wavelength. The fast decay from 271 nm observed in Figure 5.5 is seen to follow the previous result at 233 nm pump wavelengths, while the clearly observed dip at 90 fs in the 233 nm data is shifted to ~110 fs and the recovery is not as pronounced. A detailed analysis of these observations is presented in the discussion. 0 1 2 3 4 5 0.01 0.1 266 nm 225 nm 255 nm 233 nm ICN in EtO H Δτ Δτ Δτ Δτ (ps) R(Δτ Δτ Δτ Δτ) FIG. 5.4. Pump wavelength comparison of the long time anisotropy decay observed for ICN photodissociation in ethanol. 255 and 225 from ref. 4 and 233 from ref. 1. 5.5 MD results The molecular dynamics simulations provide a wealth of data and allow for multiple different breakdowns of the experimental observables. These simulations are nearly identical to those presented in earlier work for 266 nm photon energy 6 except that we have extended the photodissociation energies to include 233 and 255 nm, and we have included a spectroscopically motivated sorting criteria. 138 0.0 0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 ICN in EtOH Δτ Δτ Δτ Δτ (ps) R(Δτ Δτ Δτ Δτ) FIG. 5.5. Transient anisotropy decay for 266/400 nm (black triangles) and 233/390 nm (red diamonds) over the first 500 fs. 5.5.1 Sorting Previous analysis for ICN in water and chloroform sorted trajectories by final outcome and ensemble averages were calculated using all trajectories including those of trajectories which were recombined. Separating the recombined trajectories from the free trajectories is critically important as the recombined trajectories can generate observables that cover experimentally interesting features. This is most evident when examining things like the average rotational energy, where the inclusion of recombined molecules produces an increase of the average rotational energy due to the bending mode of ICN. Therefore it is important to reemphasize the importance of sorting out recombined trajectories at each time step. Once recombination occurs 139 (either temporarily on an excited surface, or on the ground state as ICN or INC), we expect this to coincide with a decay of the CN and iodine signals we hope to observe experimentally. The previous results showed that the majority of trajectories (85%) ultimately recombined on the ground state. 6 Spectroscopically, we are only interested in trajectories where CN remains separated from iodine and following the energetics until recombination where the trajectory is then removed from the ensemble. Therefore we apply two sorting criteria to each trajectory at each time point to decide whether to include a trajectory in the ensemble average. First a cut- off is set at an I to CN center of mass distance of 3.5 Å where we consider the I and CN molecules separated by more than 3.5 Å to be dissociated and less to be recombined. Molecules which recombine on the excited state and later redissociate are allowed to move in and out of the ensemble average. Second, a cut-off is set to - 13 kcal/mol (-0.56 eV) corresponding to recombination on the ground state. With this new cut-off justification we can follow the lifetime of the transient I and CN products and removal of the large population of recombined trajectories, make our ensemble averages much more sensitive to the smaller population of reactions which undergo a non-adiabatic transition and remain unrecombined. 5.5.2 Results from 3 Π 0+ The results from the MD are separated initially by the initial excited state, 3 Π 0+ or 3 Π 1 . For trajectories starting on the 3 Π 0+ where nonadiabatic transitions to the 1 Π 1 surface are expected, the time dependant populations of I, I* and CN are shown 140 in Figure 5.6a and b for pump wavelengths of 233, 255 and 266 nm. The rise in the I and I* signal occurs over the first 50 fs as seen in Figure 5.6a. The 17%, 20% and 27% rise at 233, 255 and 266 nm of the I population corresponds to the initial percentage of trajectories that curve cross on the first pass. A fast initial decay after the rise over the first 100-200 fs of both I and I* corresponds to collision with the solvent cage. Solvent recoil on the I* surface leads to an increase in the I population but cage recombination on the ground state competes making the ground I population appear almost flat for 233 and 255 nm over the first 300 fs with only a slight dip near 200 fs (Fig. 5.6a, inset). This solvent caging behavior is corroborated by a spike in the water-CN potential energy which is observed at 50 fs. 6 A second rise of the I population is observed between 300 fs and 1 ps corresponding to later nonadiabatic transitions on second and third passes through the curve crossing region. This secondary rise appears to rise earlier with decreasing pump energy. The second rise in I population peaks at 20% (1 ps) at 233 nm, 25% (750 fs) at 255 nm and 27% (550 fs) at 266 nm. Finally a slower decay is in both I and I* populations as the trajectories undergo late curve crossings and diffusional recombination. At the end of the 5 ps trajectories, the fractional I and I* populations correspond to 10% and 22% at 233 nm, 6% and 12% at 255 nm and 8% and 11% at 266 nm. The CN population, which is just the sum of the I and I* populations, shows a similar fast decay corresponding to cage recoil and recombination and longer ps decays which correspond to secondary recombination through diffusion. Initially, 141 the CN populations are nearly identical peaking at ~95-97% but clearly the caging behavior is much stronger in the 255 nm and 266 nm simulation. Also a small rise at ~300 fs can be seen in both the 255 nm and 266 nm data. The only way for this to occur is by trajectories that temporarily recombine on the excited state ( 3 Π 0+ does have a slight minimum at linear rI-CN = 2.5 Å, and less than 1 kT) which causes a drop in the CN population, but upon eventual movement either back onto the dissociative part of the 3 Π 0+ , or nonadiabatically on the 1 Π 1 surface, the “free” CN population will include these trajectories again. Clearly, the decreasing pump and subsequent increase in caging makes this behavior more apparent. It is also interesting to note that the long time population of CN from 255 nm and 266 nm simulations are almost identical, although this may also be due to the decreased statistics as only ~20% of the trajectories remain after 5 ps. 0 1 2 3 4 5 0.0 0.2 0.4 0.6 0.8 1.0 0 1 2 3 4 5 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 0.0 0.2 0.4 0.6 0.8 CN Population Δτ Δτ Δτ Δτ (ps) CN Population Δτ Δτ Δτ Δτ (ps) b) a) 0.0 0.5 0.0 0.2 0.4 0.6 0.8 Δτ Δτ Δτ Δτ (ps) I I* Population I I* Population Δτ Δτ Δτ Δτ (ps) FIG. 5.6. Transient a) I, I* and b) CN population from MD simulations of ICN photolysis at 266 nm (black), 255 nm (red) and 233 nm (blue). The insets show an expansion of the first picosecond. 142 The CN rotational dynamics, and in particular the rotational anisotropy functions has also been evaluated at the three pump wavelengths. For each trajectory θ(t) is monitored at each time delay and used to calculate the anisotropy correlation function using Eq. 5.1 and the same sorting criteria as above. Figure 5.7a and b shows the anisotropy decays for trajectories that start on the 3 Π 0+ excited state. The decays contain a fast and slow decay component similar to the experimentally observed signal, although we will make a detailed comparison in the discussion. Comparatively, the early time fast decay 266 nm and 255 nm curves are nearly identical with the 233 nm decaying slightly faster (Fig. 5.7a). The dip appears at ~70 fs (in the 266 nm simulation the dip appears slightly earlier at 65 fs) independent of pump energy, although the depth of this feature is deeper at 233 nm. The recovery peaks at ~100 fs for all pump wavelengths although the 266 nm recovery is sharper than that observed in either the 255 nm or 233 nm data. The relative long time decays appear to be nearly identical within the statistical noise beyond ~500 fs, although the anisotropy value never decays fully to zero over the 5 ps simulation indicating significant long lived rotational dynamics at all pump energies. 143 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0 1 2 3 4 1E-3 0.01 0.1 R(Δτ Δτ Δτ Δτ) Δτ Δτ Δτ Δτ (ps) R(Δτ Δτ Δτ Δτ) Δτ Δτ Δτ Δτ (ps) FIG. 5.7. Average transient rotational anisotropy from MD simulations of ICN photolysis at 266 nm (black), 255 nm (red) and 233 nm (blue). The inset shows the long time anisotropy over the entire 5 ps simulation, the dashed line is a guide. 5.5.3 Results from 3 Π 1 For trajectories starting on the 3 Π 1 excited state, only dissociation as CN and ground state I is possible, thus using the same sorting criteria, every free CN corresponds to a free iodine atom in the ground state. The I/CN population as a function of time is shown in Figure 5.8a for all three pump wavelengths. No significant difference in the transient decay is observed for trajectories starting on this surface. A slight difference in the final populations of 13%, 15% and 17% for 266, 255, and 233 nm is observed respectively, indicating that there may be a slight difference in the diffusion controlled recombination time stemming from a small increase in translational energy of the CN molecule with increasing pump 144 wavelength. Interestingly, there is only a very small amount of fast caging observed (~5%) over the first 500 fs indicating that most of the CN molecules must survive on the flat section of the potential surface after collision with the solvent but then quickly recombine on the ground state over the first 2 ps. The rotational anisotropy function also displays markedly different features than for 3 Π 0+ trajectories. A sharp decay to almost zero is observed within the first 100 fs but the recovery is very slow over the first picosecond. Only a small recovery of the 233 nm can be seen at 120 fs. This indicates that initial free rotating CN molecules may be present for trajectories on this surface but then the rotational energy is quickly being quenched by the solvent. The broad recovery of the anisotropy signal is observed peaking at 1 ps, which is very similar to the peak observed in the unsorted average rotational energy. This also coincides well with the steepest part of the I/CN population decay and could be an artifact of CN molecules that preferentially recombine in the same plane as the initial excitation. These trajectories will have a cos(θ) value of 1 contributing a positive value to the average anisotropy signal for one or two time steps before it is removed. This may also be playing a role in the long time anisotropic decay observed from 3 Π 0+ trajectories, although no corresponding rise in the anisotropy is observed after 200 fs. In contrast to the 3 Π 0+ anisotropy function, no anisotropy is observed beyond 2 ps. 145 0 1 2 3 4 5 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 b) a) I/CN Population Δτ Δτ Δτ Δτ (ps) 0 1 2 3 4 1E-3 0.01 0.1 R(Δτ Δτ Δτ Δτ) Δτ Δτ Δτ Δτ (ps) R(Δτ Δτ Δτ Δτ) Δτ Δτ Δτ Δτ (ps) FIG. 5.8. Transient population decay of a) I/CN and b) average anisotropy from trajectories started on the 3 Π 1 excited state surface of ICN. The inset shows the average long time anisotropy signal for the entire 5 ps trajectory. 5.6 Discussion This discussion section will address the observed pump energy effects for all available experimental measurements. We will begin by examining the experimental CN transient decay and then discuss the experimental anisotropic observations. Although the simulations take place in a water environment where evidence shows that recombination occurs on significantly different timescales than for ethanol, we have shown in the previous chapter that the general features of the simulated transient decay are consistent for both solvents and we can make qualitative comparisons for the isotropic signals. The simulated anisotropic signals do not vary with solvent over the first picosecond and thus we will draw a more direct comparison between experiment and theory for the features observed on this time scale. 146 5.6.1 Isotropic pump energy effects The isotropic decays for ICN photodissociation in ethanol, measured at pump wavelengths between 271 and 224 nm and probed at 385 nm (390 nm for 233 nm photolysis) are shown in Table 5.2. We fully expect that the CN signal is overlapped by the signal from ground state iodine which will contribute some to the measured transient signal but indications from molecular dynamics and the dispersed isotropic signal, at early times the I signal does not change much. This is a consequence of competing decay via recombination and growth from non-adiabatic transitions (see Fig. 5.6a. Another reason comparison in ethanol is preferable is based on our observations from kinetic modeling, where no significant shift of the CN band is observed in ethanol and only slight broadening occurs. Improved time resolution at 271, 266 and 233 nm has uncovered the fast decay associated with fast cage recombination. Although not uncovered at 255 and 224 nm we would expect to observe cage recombination at these wavelengths as well. It is interesting to note that for all three wavelengths where this fast timescale is observed, it corresponds to ~50% of the total decay (although at 233 nm the signal to noise leads to a rather large uncertainty). Remarkably, T fast appears to be ~5 times slower with 271 nm excitation than 266 nm excitation. While, we expect that decreasing pump energy would lead to slower cage recombination as the average translational energy should decrease, thus causing the CN to reach the cage later and recombine later, a factor of 5 is unexpectedly large. Fitting the 271 nm decay to three exponentials, resulted in T fast of 200 fs which seems much more consistent with the other T fast components. 147 T 1 in ethanol was previously assigned to a combination of cage recombination and diffusive recombination since a slight decrease was observed between 255 nm and 224 nm excitation and this timescale is fairly long for cage recombination. 4 At 233 nm excitation, where T fast and T 1 are separated, we see a significant increase in T 1 to 4.3 ps compared to 1.4 ps for 266 nm excitation which is consistent with the expected recombination rate as more energy is supplied, leading to greater translational kinetic energy release. In our kinetic model, this time constant was assigned to diffusional relaxation of I* to I. But it is probable that the overestimation of I absorption predicted between 2-7 ps comes from a faster decay component of the I which is masked by the rise as I* decays to I and not included in the kinetic model. A faster decay component of I would also lead to faster decay of CN and could be responsible for the observed timescale. The assignment of T 2 is more difficult. It has been suggested that this decay in chloroform and alcohols could be due to abstraction as mentioned previously. The fastest rate measured for abstraction in a pure solvent is 194 ps in chloroform by HCN formation. 33 A simple scaling argument based on the number of hydrogens available was used to explain the ~20 ps decays observed in alcohols. Evidence of two diffusive timescales from the MD will be discussed in more detail below but it seems more likely that T 2 comes from a second slower diffusive recombination process. There are two scenarios that could lead to two decay timescales. The most obvious comes from excitation to the two ICN excited states. Molecules that are excited to the 3 Π 1 surface do not receive the same rotational and translational kick as 148 those excited to the 3 Π 0+ surface, thus explaining the two time constants. If this were the case however, the percentage of T 1 would decrease with pump energy as more excitation occurs to the upper surface at shorter pump wavelength. The seemingly more likely, scenario is that T 1 comes from relaxation of the rotationally excited CN molecules causing the spectrum to broaden which appears as a slight decay in the CN signal. This corresponds well with the relatively small amount of decay at longer pump wavelengths and the increase with increasing pump energy, as excitation to the 3 Π 0+ increases. This leaves T 2 as the diffusive recombination timescale which corroborates the assignment in the kinetic model in Chapter 4. 5.6.2 Anisotropic pump energy effects As with the isotropic decays, the anisotropic decays are complex and difficult to interpret definitively. But clearly, long lived anisotropy seen in Fig. 5.4 and the dip and recovery observed within the first 200 fs in Fig. 5.5 indicates significant rotational coherence. Suprisingly, the long time decay of the anisotropic signal fastest at 233 and 224 nm which are near the peak of the absorption spectrum. This is counter intuitive because increasing the pump energy leads to greater rotational energy distributed to the CN fragment upon curve crossing. However, this also leads to less curve crossing and greater fraction escaping in the rotationally cold I* channel. Although the potential for later curve crossings increases with a greater fraction dissociating on the I* surface, these later curve crossings will most likely have lost the initial plane of rotation corresponding to their orientation upon excitation. Molecules excited with lower energy are more likely to curve cross on 149 the first pass or after direct recoil from the solvent cage thus maintaining the initial plane of rotation and leading to longer lived anisotropic signals. Figure 5.5 compares the anisotropic signals generated from 233 and 266 nm pump wavelengths. Although the long lived anisotropic signal from 266 nm photolysis is greater than 233, the early time signal is indicative of a greater average kinetic energy release at 233 nm. The earlier dip and faster recovery are indicative of the shorter rotational period coming from the larger kick at longer wavelengths. The later and shallower recovery observed from 266 nm most likely indicates a smaller rotational distribution, and possibly a contribution from the larger fraction of excitation to 3 Π 1 . The next section compares the MD anisotropy curves which will shed considerable light on the shape of the observed signals. 0.0 0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 ICN in EtOH Δτ Δτ Δτ Δτ (ps) R(Δτ Δτ Δτ Δτ) FIG. 5.9. Comparison of 233 nm (red diamonds) and 266 nm (black triangles) pumped anisotropy measurements and a 50:50 mixture of the 3 Π 0+ to 3 Π 1 (red line) and a 30:70 mixture (black line) from 266 nm simulations. The two curves are offset for clarity. 150 5.6.3 Comparison between MD and experiment As mentioned earlier, directly comparing the isotropic decays from the MD simulations to the experimental decays is not practical since the recombination timescales in water are much faster than ethanol. In general however, we observe very similar trends in the CN populations observed in Figure 5.6 and the experimental CN decays reported in Table 5.2. Increased pump energy appears to lead to a slightly longer decay timescale over the first 2-3 ps. This is very similar to what we observe in T 1 as pump energy is increased although the MD decay does not account for spectral evolution of the CN population. This population decay may still be related to curve crossing and the fate of single pass versus later curve crossing but a more detailed breakdown of the MD is required to completely detangle the corresponding dynamics to the population decays. A simple fit of the CN population decay to a single exponential, results in a time constant of ~1.5 ps. Previous stretching of the MD simulations to match the ethanol signal in Chapter 4 revealed that scaling up by >5 ps was needed to match the CN isotropic signal which indicates that this decay is much closer to matching T 2 timescales. The anisotropic data collected in ethanol we will compare directly to the simulated result. As we have already shown, the shape of the anisotropy function over the first 500 fs is independent of solvent and thus corresponds to freely rotating CN. 11 As in the spectral simulations in chapter 4, a 50:50 contribution from 3 Π 0+ and 3 Π 1 trajectories was used to reconstruct the anisotropic correlation function. The result is shown in Figure 5.9 (red line) seems to reproduce the 233 nm signal 151 observed previously 11 and clearly indicates that there is freely rotating CN in both the simulation and experiment. The 266 nm data which does not have as pronounced a dip and recovery can still be well reproduced by using a 30:70 contribution from 3 Π 0+ and 3 Π 1 trajectories. The deeper dip in the experimental anisotropy indicates that the amount of rotational energy imparted to the CN is underestimated in the simulations. It is possible that the intermolecular potential between water and ICN is over estimated. This may also contribute to the shorter recombination dynamics observed. It is not surprising that a greater contribution of 3 Π 1 is needed to reproduce the experimental result as we expect that towards the tail of the absorption band, this surface provides a greater contribution to the overall CN population. 6, 8 However, there are other contributions to the signal from I and I* which would have zero anisotropy effectively pulling the experimental signal down. Accounting for this signal, one would expect that amount of 3 Π 1 contribution would also effectively decrease. It should be noted that using a 50:50 mixture using the 233 nm anisotropy curves gives a qualitatively similar result and thus only the 266 nm simulations are used in this comparison. 5.7 Conclusions The purpose of this chapter was to make connections between the transient signals measured at several different pump wavelengths for the ICN photodissociation reaction in ethanol. Three distinct exponential components comprise the CN isotropic decay. These timescales correspond to cage 152 recombination (100-200 fs), rotational relaxation (1-5 ps) and diffusive recombination (10-20 ps). We see a distinct increase of the both the rotational and diffusive timescales with increasing pump energy corresponding to the increase of kinetic energy available to these channels. Although molecular dynamics simulations are run in a water environment, we see very similar population decays and a trend of increasing diffusive timescales with increasing pump energy. 0 1 2 3 4 5 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0 1 2 3 4 5 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 266/355 nm Δτ Δτ Δτ Δτ (ps) R(Δτ Δτ Δτ Δτ) 266/450 nm R(Δτ Δτ Δτ Δτ) Δτ Δτ Δτ Δτ (ps) FIG. 5.10 The transient anisotropy signal for 266 nm pump a) 355 nm and b) 450 nm probe. The small signal which goes to zero at 1 ps for 450 nm most like comes from small overlap with the tail of the CN absorption spectrum. The spectroscopically motivated sorting reveals that the MD also shows a clear indication of rotationally excited CN at early times. Even though the solvent seems to relax this energy faster than what we observe experimentally, it is still remarkable that this behavior is observed using a highly associated solvent. Experimentally, we observe the long lived anisotropic signal which indicates that the rotationally excited CN is not relaxed for several picoseconds although the decay scales inversely with pump energy which is quite surprising. This explained by the fact that longer pump wavelengths lead to a greater initial population of curve 153 crossed (rotationally excited CN) which maintains the initial plane of rotation. At longer wavelengths, less curve cross on the first pass and later curve crossing occurs after CN has had more opportunity to collide and interact with the solvent, erasing the initial plane of rotation. Comparing the anisotropy generated by the experiment to the MD simulations shows good agreement although the simulations underestimate the amount of initial rotational energy based on the shallower dip and recovery observed over the first 500 fs and the underestimate of the long time anisotropic decay. The simulations also require significant contribution from the 3 Π 1 trajectories to reproduce the experimental anisotropy. Although this makes sense for the 271 nm data, there should be very little contribution from this state at 233 nm. It is more likely that the experimental signal is pulled down by other overlapping transients which have anisotropy value of 0, specifically I and I*. Figure 5.10 shows the anisotropy measured at probe wavelengths of 355 nm and 450 nm where we expect to be measuring the anisotropy of I and I*. We see no anisotropic signal beyond the first picosecond. Introduction of the solvent to this reaction adds complexity but the flexibility of our experimental technique and the reanalysis of the quantum-classical MD makes a nice story and confirms our previous assignments and allows to explore subtleties of this reaction which were previously impossible. While several problems remain open, including a concrete assignment of T 1 and the visible absorption band observed in water, we feel that we have a good feel for the majority of trends and assignments for ICN bond breaking dynamics in water and ethanol. 154 5.8 References for Chapter 5 1. A. C. Moskun, University of Southern California, 2005. 2. L. Landau, Phys. Sov. Union, 1932, 2, 46-51. 3. C. Zener, Proc. R. Soc. London Ser. A, 1932, 137, 696-702. 4. A. C. Moskun and S. E. Bradforth, J. Chem. Phys., 2003, 119, 4500-4515. 5. I. Benjamin, J. Chem. Phys., 1995, 103, 2459-2471. 6. N. Winter, I. Chorny, J. Vieceli and I. Benjamin, J. Chem. Phys., 2003, 119, 2127-2143. 7. I. Benjamin and K. R. Wilson, J. Chem. Phys., 1989, 90, 4176-4197. 8. Y. Amatatsu, S. Yabushita and K. Morokuma, J. Chem. Phys., 1994, 100, 4894-4909. 9. J. C. Tully, J. Chem. Phys., 1990, 93, 1061-1071. 10. K. Kuchitsu and Y. Morino, B. Chem. Soc. Jpn., 1965, 38, 814-824. 11. A. C. Moskun, A. E. Jailaubekov, S. E. Bradforth, G. H. Tao and R. M. Stratt, Science, 2006, 311, 1907-1911. 12. A. E. Jailaubekov and S. E. Bradforth, Appl. Phys. Lett., 2005, 87, 021107. 13. C. A. Rivera, S. E. Bradforth and G. Tempea, Opt. Express, 2010, 18, 18615- 18624. 14. X. Y. Chen, D. S. Larsen, S. E. Bradforth and I. H. M. Stokkum, J. Phys. Chem. A, 2010, submitted. 15. C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys., 2009, 130, 084501. 16. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum., 2003, 74, 4958-4960. 17. A. P. Baronavski, Chem. Phys., 1982, 66, 217-225. 155 18. E. M. Goldfield, P. L. Houston and G. S. Ezra, J. Chem. Phys., 1986, 84, 3120-3129. 19. M. D. Pattengill, Chem. Phys., 1984, 87, 419-429. 20. B. A. Waite, H. Helvajian, B. I. Dunlap and A. P. Baronavski, Chem. Phys. Lett., 1984, 111, 544-548. 21. R. Baer and R. Kosloff, Chem. Phys. Lett., 1992, 200, 183-191. 22. J. A. Beswick, M. Glassmaujean and O. Roncero, J. Chem. Phys., 1992, 96, 7514-7527. 23. R. D. Coalson and M. Karplus, J. Chem. Phys., 1990, 93, 3919-3930. 24. S. Y. Lee, J. Chem. Phys., 1992, 97, 227-235. 25. J. W. Qian, C. J. Williams and D. J. Tannor, J. Chem. Phys., 1992, 97, 6300- 6308. 26. C. J. Williams, J. W. Qian and D. J. Tannor, J. Chem. Phys., 1991, 95, 1721- 1737. 27. M. L. Johnson and I. Benjamin, J. Phys. Chem. A, 2009, 113, 7403-7411. 28. J. Vieceli, I. Chorny and I. Benjamin, J. Chem. Phys., 2001, 115, 4819-4828. 29. J. Vieceli, I. Chorny and I. Benjamin, Chem. Phys. Lett., 2002, 364, 446-453. 30. N. Winter and I. Benjamin, J. Chem. Phys., 2004, 121, 2253-2263. 31. D. D. Davis and H. Okabe, J. Chem. Phys., 1968, 49, 5526-5531. 32. P. Fornier de Violet, Rev. Chem. Intermed., 1981, 4, 121-169. 33. D. Raftery, E. Gooding, A. Romanovsky and R. M. Hochstrasser, J. Chem. Phys., 1994, 101, 8572-8579. 156 Chapter 6. Future directions: Isolating CN radicals in solution Abstract for Chapter 6 The intramolecular excitation of CN(aq) have been studied recently both theoretically and by time resolved photodissociation of ICN. Broad overlapping absorptions of parent iodine radicals convolute the CN(aq) absorption bands making it difficult to assign absorption to specific radical species. Removing other radical absorption species and isolating CN radicals in solution would help to alleviate the ambiguity of some of the assignments made experimentally. Removing an electron from CN – in solution would require knowledge of the CTTS transition or direct detachment energies. The vertical binding energy lies above 9 eV and although a CTTS transition is reported at 170 nm the VUV absorption spectrum of cyanide has only ever been reported to 180 nm in due to the solvent window cutoff and the difficulties involved with exposing liquids to vacuum environments. In this work, the VUV absorption spectrum of CN – is reported to 8.6 eV (144 nm) for the first time and a signature of the cyanide CTTS state is reported. A three pulse experiment is devised to access the CTTS transition and probe the detached solvated CN radical and electron. 6.1 Introduction The electronic spectrum of aqueous cyano radical, CN, until recently 1 had never been reported experimentally, and exact assignments are still under dispute. 1-4 This is quite surprising since this is one of the most extensively studied gas phase 157 radicals 5-12 and is particularly important for identifying transient spectra generated from ICN photodissociation in solution performed by several groups, 1-3, 13, 14 including this dissertation work. Our previous reports have indicated that a mixture of highly rotationally excited “hot” CN radicals and comparatively rotationally “cold” CN is generated from this reaction 2, 3 and a clear absorption band at 385 nm is observed which is centered near the gas phase CN B←X transition. It is believed that the spinning molecule creates a cavity in the solvent which is responsible for the gas-phase like signature. 3 Over the next few picoseconds, depending on the solvent, friction finally relaxes the “hot” CN and solvation occurrs. 2, 3 A high level calculation of the solvated CN B←X in an aqueous environment has been carried out by Pieniazek et al, and provides the best estimation for the effect of the water environment on this electronic transition. 4 These calculations utilized a combination of equation-of-motion coupled-cluster with single and double substitutions (EOM- CCSD), and time-dependant density functional theory (TD-DFT) to evaluate the two lowest electronic transitions of the cyano radical. The results indicate that the solvent induces a blue-shift and broadening of both the CN A←X and CN B←X transitions. The CN A←X absorption is expected to lie in the near-IR region, centered at 1.45 eV (855 nm) with 10 times weaker gas-phase oscillator strength 15 than the CN B←X transition centered at 3.6 eV (344 nm). 158 200 300 400 500 0 1000 2000 3000 6 5 4 3 CN _ (aq) CN(aq) CN(g) x 0.1 1PA Energy (eV) Wavelength (nm) ε ε ε ε (M -1 cm -1 ) FIG. 6.1. The one-photon absorption (1PA) spectra of aqueous cyanide (black, ref. 19) and cyano radical in water (blue, ref. 4) and gas phase (ref. 12). While our ICN experiments exhibit transient absorption features which resemble shifting and broadening of the CN B←X absorption (the CN A←X absorption lies outside our probing window), other transients such as I and I* are expected to overlap in the same region as CN, 1, 2, 16, 17 making definitive assignments difficult. Ideally, isolating a CN radical in solution devoid of other absorbing molecules is the only way to definitively identify the CN spectrum. Photodetachment of cyanide to form CN radicals and solvated electrons by a similar method to our iodide photodetachment experiments 16-18 could provide a pathway to 159 identifying the CN spectrum. The shifting electron spectrum would overlap with the CN absorption bands but this signal can be removed by methods presented in detail in these previous studies. As with the halogens, CN is expected to display a charge- transfer-to-solvent, or CTTS absorption band, and the reported band center is ~170 nm (7.3 eV). 19 Due to the high photon energies needed to resolve this absorption band, currently there is no one-photon absorption (1PA) spectrum of aqueous cyanide reported at energies above 180 nm and the reported CTTS transition is known only from numerical fitting of the observed tail. The furthest photodetachment wavelengths available from our experimental apparatus only extend to 200 nm (6.2 eV). The CTTS spectrum of cyanide is shown in Figure 6.1 and the tail of the spectrum has no absorption at 200 nm making one-photon detachment impossible without generating light in the vacuum-ultraviolet (VUV). An alternative method for reaching VUV wavelengths via 2-photon (2PA) spectroscopy has already proven useful for probing the electronic spectrum of various solvents and solutes including water and iodide. 20-23 We have already presented a 2PA apparatus which can access 1PA energies ranging from ~7 – 10 eV by overlapping two lower energy pulses spatially and temporally. 23 Furthermore, the 2PA spectrum measured at different relative polarizations between the two pulses can be used to divulge information on the symmetry of the excited state wave function and provide information for assignments of the excited states. 24 Although the 2PA selection rules are not equivalent to the 1PA case, we have already observed that solvation may cause a relaxation of rigorous selection rules by helping to 160 delocalize excited state wave functions allowing observation of both 2PA forbidden and 1PA allowed transitions to appear in the 2PA spectrum. 23 Applying 2PA spectroscopy to the aqueous CN − system, provides a picture of the previously unobserved VUV transitions and provides a rationale for possible 3-photon experiments where two pulses are used to reach the CN CTTS state and detach an electron and a third delayed pulse probes the CN radical spectrum. In this chapter, the preliminary aqueous CN − 2PA spectrum is presented and discussed in terms of the possible excited states which are elucidated and a 3-photon experiment to determine the CN radical B←X spectrum is proposed as the future direction of this dissertation work. 6.5 7.0 7.5 8.0 8.5 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 -2 0 2 4 6 8 10 190 180 170 160 150 Polarization Ratio Relative Cross Section (arbitrary units) Total 2PA Energy (eV) Wavelength (nm) FIG. 6.2. The parallel (red) and perpendicular (blue) 2PA spectrum of 2 M NaCN in water and the parallel 2PA spectrum of water (purple line) from ref. 23. The polarization ratio for 2 M NaCN (green circles) and the polarization ratio for pure water (purple circles) from ref. 23 are also shown. 161 6.2 2PA of cyanide in solution The details of the 2PA experimental setup and generation of 271 nm was presented previously and the exact same setup was used to produce the 2PA spectrum of CN − in water. 23, 25, 26 The pump pulses consisted of 4 μJ, ~40 fs, 271 nm (4.6) third-harmonic of a 3.5 mJ, 35 fs, 800 nm, 1 kHz Ti:S regenerative amplifier (Coherent Legend Elite USP-HE). The super-continuum probe pulses were generated by focusing a small portion of the fundamental in a 2 mm thick circularly translating CaF 2 plate and spanned 688 – 309 nm (1.8 – 4.0 eV). Thus, the total 2PA energies achieved are 194 – 144 nm (6.4 – 8.6 eV). The 2PA signal was generated by temporally scanning the probe pulse which is delayed by a computer controlled motorized translation stage. When the pump and probe are overlapped temporally, the strong 2PA signal appears as a cross-correlation of the pump and probe as a function of time. The probe is transmitted through the sample and focused into a broadband spectrometer which disperses the super-continuum onto a 256 pixel diode array. Two diffraction gratings were used in the spectrometer to cover both the UV and visible portions of the probe and no difference in the spectrum was observed in the data where the probe wavelengths overlap. By chopping the pump pulses to 500 Hz and measuring the change in probe transmission with the pump on vs. pump off, the result is collected and integrated at each diode pixel resulting in the relative 2PA signal. The pump spot size is maximized such that the transient signal produced by 2PA (pump + pump) of either the solvent or solute is small. Finally, the 2PA is 162 collected at two different polarization schemes: when the pump and probe are polarized parallel with respect to each other, this constitutes parallel 2PA, and when the probe is polarized by 90 degrees with respect to the pump, this constitutes perpendicular 2PA. The sample consists of 2 M NaCN (> 95%, EMD) in water (15 MΩ·cm, Millipore) and was delivered by a ~100 μm thick film wire-guided gravity drop jet. 27 The parallel and perpendicular cyanide 2PA spectra are shown in Figure 6.2 (solid red and blue respectively) along with the parallel 2PA spectrum of water 23 (solid purple). Both the parallel and perpendicular 2PA spectra monotonically rise to 7.6 eV. The parallel spectrum has a distinct absorption maximum at 7.6 eV, while no peak is observed in the perpendicular spectrum. Another small rise is observed in the parallel spectrum between 7.9 eV and 8.1 eV and then both spectra are essentially flat at 2PA energies above 8.1 eV. The parallel 2PA water is shown as we do expect some water absorption from the solvent, but clearly, even if we maximize the water contribution to match the highest energy side of the 2PA spectrum, the water cannot account for the absorption signals observed in the CN − experiments. As with the previous studies in water, it is difficult to determine the number of excited states which contribute to the total absorption signature as these bands are often quite broad in solution. 23 However, from the work by Monson and McClain we know that the ratio between the parallel and perpendicular signals (or polarization ratio) is very sensitive to the symmetry of the excited states. 24 Even if we cannot definitively assign the symmetry of the excited state, subtle changes in the 163 polarization ratio can tell us the minimum number of states needed to simultaneously reconstruct the 2PA spectra and the shape of the polarization ratio. The polarization ratio for 2 M NaCN in water is shown in Figure 6.2 (green circles) which has a value of 2 at 6.5 eV and then increases to a maximum of 8 at 7.5 eV, only to drop down to a value of 3 at higher 2PA energies. For comparison, the water ratio is also shown in Figure 6.2 (purple circles) which is relatively flat but does gently increase with increasing 2PA energy. For water, this gentle increase was assigned to the weakly absorbing Ã 1 1 B 1 transition of water in the tail of the spectrum which has a ratio of ~0.9 overlapping with higher energy transitions with higher polarization ratios. In order to reproduce the shape just the polarization ratio, we know that a minimum of three states must be used. The reason for this is fairly straightforward: starting at the lowest 2PA energy a) the ratio must correspond to a state with a relatively low ratio responsible for the starting value of 2, then b) second state with a much higher polarization ratio begins to absorb and we observe a sharp increase in the ratio, finally c) above 7.5 eV a third state with relatively lower polarization ratio causes the dip to ~3. A simulation which consisted of three excited state transitions and water background was used to simultaneously simulate the polarization ratio and the individual 2PA spectra. Gaussians were used to represent the three CN − transitions and the water was reconstructed from Elles et al. 23 The known CTTS transition from Fox and Hayon was centered at 7.28 eV and given a bandwidth of 0.56 eV FWHM, 19 while the center frequency and bandwidth of the other two states were allowed to 164 vary in energy. Using these three states plus the water spectrum, the parallel spectrum was fit and the relative polarization ratio of each state was then adjusted to reproduce the experimental ratio. The result is shown in Figure 6.3 and the fitting parameters for the CN − transitions are TABLE 6.1. Gaussian fit parameters for the simultaneous fit of the 2PA spectra of 2 M NaCN in water and the resulting polarization ratio. State Energy (eV) FWHM (eV) Parallel int. (arb.) Ratio CTTS 7.28 0.56 0.18 4 1 7.67 0.56 0.56 9 2 8.15 0.40 0.26 2.5 presented in Table 6.1. Overall, the simulated spectrum and ratio reproduces the shape of both the 2PA spectra and the ratio, successfully reproducing the sharp peak in the polarization ratio and both peaks in the 2PA spectrum. At low 2PA energy the simulated ratio flattens to a value of ~4 while the experimental ratio drops to a value of 2. Careful examination of the tail of the 2PA spectrum reveals that there is a small contribution of signal presumably from transient solvated electrons created by 2PA (pump + pump) ionization of the solvent which gets counted from the integration of the 2PA peak. Assuming that the strength of the transient electron absorption is independent of polarization, the resulting signal should have a ratio of 1 which would pull the total ratio down even further, although we did not attempt to add electron signal to the simulation. Also, the peak in the experimental and simulated spectra do not match up perfectly, which could also be a result of not including the solvated electron which would cause an increase the ratio of the CTTS 165 state, thus shifting the peak of the total ratio lower in energy. Alternatively, there could also be a fourth state absorbing in the 2PA spectrum between the CTTS state and state 2 (Fig. 6.3, orange), which is not accounted for in the simulation. 6.5 7.0 7.5 8.0 8.5 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 -2 0 2 4 6 8 10 Polarization Ratio Relative Cross Section (arbitrary units) Total 2PA Energy (eV) H 2 O 3 2 CTTS FIG. 6.3. Simulated polarization ratio (solid green) and the simulated parallel 2PA spectrum (solid black) overlaid with the experimental data from Figure 6.1. The three excited states (blue, orange, gold) and water (magenta) that were used in the simulation are also shown with their parallel intensities. Other than the CTTS state, there is very little information on the excited states of cyanide. The gas phase excited states are unstable due to autodetachment 28 and have only been investigated theoretically. 28-30 The vertical detachment energy (VDE) of small water-cyanide clusters have been examined 31 and the binding energy measured by photoelectron spectroscopy. 32 The VDE of CN − in water is has been 166 measured to be 9.33 eV by liquid jet photoelectron spectroscopy, so it can be expected that all excited states measured in this study should peak below the detachment continuum. The valence electron configuration for cyanide is as follows: (π2p) 4 (σ2p) 2 (π*2p) 0 (σ*2p) 0 The ground state is 1 Σ + and the three lowest singlet excited states in the gas phase calculated by Musiał 29 are 1 Σ, 1 Π, and 1 Δ which would correspond to excitations from the σ bonding orbital to the π* and σ* and excitation from the π to the π* orbital respectively. The 1 Δ state is approximately 1 eV higher in energy than the 1 Σ and 1 Π states. This difference in energy is essentially the energy gap between the (σ2p) and (π2p) orbitals and is in good agreement with the calculation of the CN radical A←X transition which corresponds to excitation between these orbitals. 4 Upon solvation, Pieniazek et al calculate an additional 0.15 eV blue shift for the A←X transition. Although, the effect of solvation on the valence orbitals of cyanide is not known, if state 2 in Figure 6.3 corresponds to either the 1 Σ or 1 Π states, the 1 Δ state would most likely lie at least 1 eV higher in energy (> 8.67 eV) and thus is not responsible for state 3. It is more likely that upon solvation, splitting between the (σ2p) and (π2p) increases and is responsible for the 0.4 eV separation of states 2 and 3. These assignments rely on the fact that the state at 7.2 eV is in fact a CTTS state and that no other intramolecular transitions occur below 7.6 eV which may not be the case although the assignment of the CTTS transition should be relatively easy to verify. Mulliken’s theory of charge-transfer absorption bands predicts that the 167 position of the CTTS band is highly dependent on the IP of the solvent 33 and thus a simple change of solvent should cause the CTTS to shift significantly. 6.3 Proposed 3 pulse experiment In order to isolate CN radicals in solution, a three pulse experiment can be devised to remove an electron from CN – and probing the transient solvated products. Two photons with a combined energy 7.2 eV should be sufficient to excite the CTTS transition of aqueous cyanide and then a third super-continuum probe pulse is used to observe the CN B←X transition which is expected to occur near 320–340 nm based on both photodissociation and ab initio results. Figure 6.4 shows a schematic where fourth harmonic (200 nm) light is generated similarly to previous 2PA experiments, and the 1200 nm signal output from an OPA are overlapped spatially and temporally in an aqueous thin film jet of 2 M NaCN. The total energy of these two pulses acting as the pump pulses total 7.2 eV, while a third super-continuum probe pulse measures the transient spectrum as a function of time. This relatively simple experiment should produce CN radicals and solvated electrons within the probe window of 320– 625 nm. The estimated CN B←X extinction coefficient is similar to those reported for iodine radicals 4 and thus, should not be difficult to separate from the well known solvated electron spectrum. 34 It is also feasible to propose detaching an electron directly to the water conduction band by using to photons of 200 nm (6.2 eV) and 400 nm (3.1 eV) which would correspond to the liquid VBE. 32 Although this route would most likely lead to larger yields of CN + e – similar to what in observed from iodide photodetachment, the 2PA cross-section of water is known to rise monotonically and peak near 9.5 eV and the below 10 eV which could compete with CN FIG. 6.4. Experimental schematic for proposed 3 harmonic 200 nm light is generated by doubling 800 nm in a type I BBO and then mixing the 400 nm with 800 nm to generate 266 nm in a type II BBO. The third harmonic 266 nm is then mixed make 200 nm which is overlapped signal in a sample jet. The probe CaF 2 disk and collected by a broad band spectrometer. is controlled by a motorized delay stage and the 200 nm is chopped to half the repetition rate. The transmission difference used to construct the lly and peak near 9.5 eV and the vertical detachment threshold extends below 10 eV which could compete with CN – absorption. 23, 35 Experimental schematic for proposed 3-pulse experiment where fourth harmonic 200 nm light is generated by doubling 800 nm in a type I BBO and then mixing the 400 nm with 800 nm to generate 266 nm in a type II BBO. The third harmonic 266 nm is then mixed with residual 800 nm in another type I BBO to e 200 nm which is overlapped spatially and temporally with 1200 nm OPA ignal in a sample jet. The probe is super continuum generated in a translating disk and collected by a broad band spectrometer. The time delay of the probe is controlled by a motorized delay stage and the 200 nm is chopped to half the repetition rate. The transmission difference of the probe with 200 nm on and off construct the transient absorption in mOD. 168 vertical detachment threshold extends pulse experiment where fourth harmonic 200 nm light is generated by doubling 800 nm in a type I BBO and then mixing the 400 nm with 800 nm to generate 266 nm in a type II BBO. The third with residual 800 nm in another type I BBO to and temporally with 1200 nm OPA is super continuum generated in a translating The time delay of the probe is controlled by a motorized delay stage and the 200 nm is chopped to half the of the probe with 200 nm on and off is 169 6.4 Concluding remarks Although we have measured the 2PA spectrum of CN – in water only, the range of polar solvents and ionic solutes which can be interrogated by 2PA is almost endless and will be a major goal of our lab in the future. Understanding and assigning 2PA spectra will open the possibility of accessing excited states of molecules well beyond UV photon energies without the need for vacuum chambers and high energy radiation sources. Accessing high lying CTTS transitions in the case of CN – and low lying vertical detachment channels in other molecules pave a new pathway for the creation and study of radical species with high time resolution. The relative ease moving between typical pump-probe experiments and 2PA experiments makes this technique even more attractive to labs which already have broadband probing capabilities. 170 6.5 References for Chapter 6 1. J. Larsen, D. Madsen, J. A. Poulsen, T. D. Poulsen, S. R. Keiding and J. Thogersen, J. Chem. Phys., 2002, 116, 7997-8005. 2. A. C. Moskun and S. E. Bradforth, J. Chem. Phys., 2003, 119, 4500-4515. 3. A. C. Moskun, A. E. Jailaubekov, S. E. Bradforth, G. H. Tao and R. M. Stratt, Science, 2006, 311, 1907-1911. 4. P. A. Pieniazek, S. E. Bradforth and A. I. Krylov, J. Phys. Chem. A, 2006, 110, 4854-4865. 5. B. Brocklehurst, G. R. Hebert, S. H. Innanen, R. M. Seel and R. W. Nicholls, The identification atlas of molecular spectra: The CN CN B2Σ+←X2Σ+ Red System, York University, Centre for Research in Experimental Space Science, Toronto, 1972. 6. I. Nadler, D. Mahgerefteh, H. Reisler and C. Wittig, J. Chem. Phys., 1985, 82, 3885-3893. 7. I. Nadler, H. Reisler and C. Wittig, Chem. Phys. Lett., 1984, 103, 451-457. 8. P. Casavecchia, N. Balucani, L. Cartechini, G. Capozza, A. Bergeat and G. G. Volpi, Faraday Discuss., 2001, 119, 27-49. 9. J. H. Ling and K. R. Wilson, J. Chem. Phys., 1975, 63, 101-109. 10. E. M. Goldfield, P. L. Houston and G. S. Ezra, J. Chem. Phys., 1986, 84, 3120-3129. 11. M. Dantus, M. J. Rosker and A. H. Zewail, J. Chem. Phys., 1987, 87, 2395- 2397. 12. D. D. Davis and H. Okabe, J. Chem. Phys., 1968, 49, 5526-5531. 13. A. C. Crowther, S. L. Carrier, T. J. Preston and F. F. Crim, J. Phys. Chem. A, 2008, 112, 12081-12089. 14. A. C. Crowther, S. L. Carrier, T. J. Preston and F. F. Crim, J. Phys. Chem. A, 2009, 113, 3758-3764. 15. K. P. Huber and G. Herzberg, Constants of diatomic molecules, Van Nostrand Reinhold, New York, 1979. 171 16. A. C. Moskun, S. E. Bradforth, J. Thogersen and S. Keiding, J. Phys. Chem. A, 2006, 110, 10947-10955. 17. C. A. Rivera, Y.; Ottosson, N.; Winter, B.; Bradforth S. E., in prep., 2010. 18. V. H. Vilchiz, X. Y. Chen, J. A. Kloepfer and S. E. Bradforth, Rad. Phys. Chem., 2005, 72, 159-167. 19. M. F. Fox and E. Hayon, J. Chem. Soc. Faraday Trans., 1990, 86, 257-263. 20. Y. B. Zhang, S. E., in prep., 2011. 21. S. Yamaguchi and T. Tahara, Chem. Phys. Lett., 2003, 376, 237-243. 22. S. Yamaguchi and T. Tahara, Chem. Phys. Lett., 2004, 390, 136-139. 23. C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys., 2009, 130, 084501. 24. P. R. Monson and W. M. Mcclain, J. Chem. Phys., 1970, 53, 29-&. 25. C. A. Rivera, S. E. Bradforth and G. Tempea, Opt. Express, 2010, 18, 18615- 18624. 26. A. E. Jailaubekov and S. E. Bradforth, Appl Phys Lett, 2005, 87, -. 27. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum., 2003, 74, 4958-4960. 28. C. S. Ewig and J. Tellinghuisen, Chem. Phys. Lett., 1988, 153, 160-165. 29. M. Musial, Mol. Phys., 2005, 103, 2055-2060. 30. R. Polak and J. Fiser, J. Mol. Struct. THEOCHEM, 2002, 584, 69-77. 31. X. B. Wang, K. Kowalski, L. S. Wang and S. S. Xantheas, J. Chem. Phys., 2010, 132, 124306. 32. S. E. Bradforth, in prep., 2011. 33. R. S. Mulliken, J. Phys. Chem., 1952, 56, 801-822. 34. F. Y. Jou and G. R. Freeman, Can. J. Chem., 1979, 57, 591-597. 35. B. Winter and M. Faubel, Chem. Rev., 2006, 106, 1176-1211. 172 Biliography D. Abramavicius, J. Jiang, B. M. Bulheller, J. D. Hirst and S. Mukamel, J. Am. Chem. Soc., 2010, 132, 7769-7775. Y. Amatatsu, S. Yabushita and K. Morokuma, J. Chem. Phys., 1994, 100, 4894- 4909. C. Anastasi, V. Simpson, J. Munk and P. Pagsberg, Chem. Phys. Lett., 1989, 164, 18- 22. T. Andersen, K. R. Lykke, D. M. Neumark and W. C. Lineberger, J. Chem. Phys., 1987, 86, 1858-1867. R. W. Anderson and R. M. Hochstrasser, J. Phys. Chem., 1976, 80, 2155-2159. R. Baer and R. Kosloff, Chem. Phys. Lett., 1992, 200, 183-191. U. Banin, A. Waldman and S. Ruhman, J. Chem. Phys., 1992, 96, 2416-2419. A. Baratz and S. Ruhman, Chem. Phys. Lett., 2008, 461, 211-217. A. P. Baronavski, Chem. Phys., 1982, 66, 217-225. D. M. Bartels, K. Takahashi, J. A. Cline, T. W. Marin and C. D. Jonah, J. Phys. Chem. A, 2005, 109, 1299-1307. R. H. Bathgate and E. A. Moelwyn-Hughes, J. Chem. Soc., 1959, 2642-2648. I. Benjamin, J. Chem. Phys., 1995, 103, 2459-2471. I. Benjamin, U. Banin and S. Ruhman, J. Chem. Phys., 1993, 98, 8337-8340. I. Benjamin and K. R. Wilson, J. Chem. Phys., 1989, 90, 4176-4197. M. Berg, A. L. Harris and C. B. Harris, Phys. Rev. Lett., 1985, 54, 951-954. R. S. Berry, G. N. Spokes and C. W. Reimann, J. Chem. Phys., 1962, 37, 2278-2290. M. Besnard, N. Delcampo, P. Fornier de Violet and C. Rulliere, Laser Chem., 1991, 11, 109-118. 173 J. A. Beswick, M. Glassmaujean and O. Roncero, J. Chem. Phys., 1992, 96, 7514- 7527. M. Beutler, M. Ghotbi, F. Noack, D. Brida, C. Manzoni and G. Cerullo, Opt. Lett., 2009, 34, 710-712. S. E. Bradforth, in prep., 2011. S. E. Bradforth and P. Jungwirth, J. Phys. Chem. A, 2002, 106, 1286-1298. B. Brocklehurst, G. R. Hebert, S. H. Innanen, R. M. Seel and R. W. Nicholls, The identification atlas of molecular spectra: The CN CN B2Σ+←X2Σ+ Red System, York University, Centre for Research in Experimental Space Science, Toronto, 1972. H. Bursing, J. Lindner, S. Hess and P. Vohringer, Appl. Phys. B-Lasers O., 2000, 71, 411-417. P. Casavecchia, N. Balucani, L. Cartechini, G. Capozza, A. Bergeat and G. G. Volpi, Faraday Discuss., 2001, 119, 27-49. N. Chandrasekhar and P. Krebs, J. Chem. Phys., 2000, 112, 5910-5914. X. Chen, L. Canova, A. Malvache, A. Jullien, R. Lopez-Martens, C. Durfee, D. Papadopoulos and F. Druon, Appl. Phys. B, 2010, 99, 149-157. X. Y. Chen and S. E. Bradforth, Annu. Rev. Phys. Chem., 2008, 59, 203-231. X. Y. Chen, D. S. Larsen, S. E. Bradforth and I. H. M. Stokkum, J. Phys. Chem. A, 2010, submitted. R. D. Coalson and M. Karplus, J. Chem. Phys., 1990, 93, 3919-3930. CRC Handbook of Chemistry and Physics, CRC Press, Cleveland, Ohio, 2010. A. C. Crowther, S. L. Carrier, T. J. Preston and F. F. Crim, J. Phys. Chem. A, 2008, 112, 12081-12089. A. C. Crowther, S. L. Carrier, T. J. Preston and F. F. Crim, J. Phys. Chem. A, 2009, 113, 3758-3764. C. H. B. Cruz, P. C. Becker, R. L. Fork and C. V. Shank, Opt. Lett., 1988, 13, 123- 125. M. Dantus, M. J. Rosker and A. H. Zewail, J. Chem. Phys., 1987, 87, 2395-2397. 174 P. S. Dardi and J. S. Dahler, J. Chem. Phys., 1993, 98, 363-372. D. D. Davis and H. Okabe, J. Chem. Phys., 1968, 49, 5526-5531. G. Dobson and L. I. Grossweiner, Radiat. Res., 1964, 23, 290-299. P. Dombi, P. Racz, M. Lenner, V. Pervak and F. Krausz, Opt. Express, 2009, 17, 20598-20604. C. G. Durfee, S. Backus, H. C. Kapteyn and M. M. Murnane, Opt. Lett., 1999, 24, 697-699. C. G. Durfee, S. Backus, M. M. Murnane and H. C. Kapteyn, Opt. Lett., 1997, 22, 1565-1567. C. F. Dutin, A. Dubrouil, S. Petit, E. Mevel, E. Constant and D. Descamps, Opt. Lett., 2010, 35, 253-255. O. T. Ehrler, G. B. Griffin, R. M. Young and D. M. Neumark, J. Phys. Chem. B, 2009, 113, 4031-4037. O. T. Ehrler and D. M. Neumark, Acc. Chem. Res., 2009, 42, 769-777. C. G. Elles, M. J. Cox, G. L. Barnes and F. F. Crim, J. Phys. Chem. A, 2004, 108, 10973-10979. C. G. Elles, A. E. Jailaubekov, R. A. Crowell and S. E. Bradforth, J. Chem. Phys., 2006, 125, 044515. C. G. Elles, C. A. Rivera, Y. Zhang, P. A. Pieniazek and S. E. Bradforth, J. Chem. Phys., 2009, 130, 084501. C. S. Ewig and J. Tellinghuisen, Chem. Phys. Lett., 1988, 153, 160-165. M. Faubel, B. Steiner and J. P. Toennies, J. Chem. Phys., 1997, 106, 9013-9031. W. S. Felps, K. Rupnik and S. P. McGlynn, J. Phys. Chem., 1991, 95, 639-656. P. Fornier de Violet, Rev. Chem. Intermed., 1981, 4, 121-169. P. Fornier de Violet, R. Bonneau and J. Joussot-Dubien, Chem. Phys. Lett., 1973, 19, 251-253. P. Fornier de Violet, R. Bonneau and J. Joussot-Dubien, Mol. Photochem., 1973, 5, 61-67. 175 M. F. Fox and E. Hayon, J. Chem. Soc., Faraday Trans. 1, 1977, 73, 1003-1016. M. F. Fox and E. Hayon, J. Chem. Soc. Faraday Trans., 1990, 86, 257-263. E. M. Goldfield, P. L. Houston and G. S. Ezra, J. Chem. Phys., 1986, 84, 3120-3129. S. Gopalakrishnan, L. Zu and T. A. Miller, Chem. Phys. Lett., 2003, 380, 749-757. T. Goulet, C. Pepin, D. Houde and J. P. Jay-Gerin, Radiat. Phys. Chem., 1999, 54, 441-448. T. A. Gover and G. Porter, Proc. R. Soc. London Ser. A, 1961, 262, 476-488. S. J. Greaves, R. A. Rose, T. A. A. Oliver, M. N. R. Ashfold, I. P. Clark, G. M. Greetham, A. W. Walker, M. Towrie and A. J. Orr-Ewing, private communication. P. M. Hare, E. A. Price and D. M. Bartels, J. Phys. Chem. A, 2008, 112, 6800-6802. S. Hess, H. Bursing and P. Vohringer, J. Chem. Phys., 1999, 111, 5461-5473. W. P. Hess and S. R. Leone, J. Chem. Phys., 1987, 86, 3773-3780. Y. Hirata and N. Mataga, Prog. React. Kinet., 1993, 18, 273-308. M. L. Horng, J. A. Gardecki, A. Papazyan and M. Maroncelli, J. Phys. Chem., 1995, 99, 17311-17337. K. P. Huber and G. Herzberg, Constants of diatomic molecules, Van Nostrand Reinhold, New York, 1979. H. Iglev, R. Laenen and A. Laubereau, Chem. Phys. Lett., 2004, 389, 427-432. H. Iglev, A. Trifonov, A. Thaller, I. Buchvarov, T. Fiebig and A. Laubereau, Chem. Phys. Lett., 2005, 403, 198-204. A. E. Jailaubekov and S. E. Bradforth, Appl. Phys. Lett., 2005, 87, 021107. J. P. Jay-Gerin, Can. J. Chem., 1997, 75, 1310-1314. M. L. Johnson and I. Benjamin, J. Phys. Chem. A, 2009, 113, 7403-7411. J. Jortner, M. Ottolenghi and G. Stein, J. Phys. Chem., 1963, 67, 1271-1274. J. Jortner, B. Raz and G. Stein, T. Faraday Soc., 1960, 56, 1273-1275. 176 F. Y. Jou and G. R. Freeman, Can. J. Chem., 1979, 57, 591-597. Y. Kida, S. Zaitsu and T. Imasaka, Opt. Express, 2008, 16, 13492-13498. K. Kimura, Handbook of HeI photoelectron spectra of fundamental organic molecules : ionization energies, ab initio assignments, and valence electronic structure for 200 molecules, Japan Scientific Societies Press, Tokyo, 1981. K. Kosma, S. A. Trushin, W. E. Schmid and W. Fuss, Opt. Lett., 2008, 33, 723-725. N. Krebs, R. A. Probst and E. Riedle, Opt. Express, 2010, 18, 6164-6171. K. Kuchitsu and Y. Morino, B. Chem. Soc. Jpn., 1965, 38, 814-824. L. Landau, Phys. Sov. Union, 1932, 2, 46-51. J. Larsen, D. Madsen, J. A. Poulsen, T. D. Poulsen, S. R. Keiding and J. Thogersen, J. Chem. Phys., 2002, 116, 7997-8005. J. Larsen, D. Madsen, J. A. Poulsen, T. D. Poulsen, S. R. Keiding and J. Thøgersen, J. Chem. Phys., 2002, 116, 7997-8005. S. Y. Lee, J. Chem. Phys., 1992, 97, 227-235. A. Levy, Meyerste.D and Ottoleng.M, J. Phys. Chem., 1973, 77, 3044-3047. A. Levy, D. Meyerstein and M. Ottolenghi, J. Phys. Chem, 1973, 77, 3044-3047. Z. Y. Li, D. Abramavicius, W. Zhuang and S. Mukamel, Chem. Phys., 2007, 341, 29-36. R. Lian, D. A. Oulianov, R. A. Crowell, I. A. Shkrob, X. Y. Chen and S. E. Bradforth, J. Phys. Chem. A, 2006, 110, 9071-9078. J. H. Ling and K. R. Wilson, J. Chem. Phys., 1975, 63, 101-109. R. Lingle, X. B. Xu, S. C. Yu, Y. J. Chang and J. B. Hopkins, J. Chem. Phys., 1990, 92, 4628-4630. O. Link, Georg-August-Universitat Göttingen, 2007. F. H. Long, H. Lu, X. L. Shi and K. B. Eisenthal, Chem. Phys. Lett., 1990, 169, 165- 171. I. Matsuda, K. Misawa and R. Lang, Opt. Comm., 2004, 239, 181-186. 177 F. Mbaiwa, J. Wei, M. Van Duzor and R. Mabbs, J. Chem. Phys., 2010, 132, 134304. D. F. McMillen and D. M. Golden, Annu. Rev. Phys. Chem., 1982, 33, 493-532. A. R. Menzeleev and T. F. Miller, J. Chem. Phys., 2010, 132, 034106. P. R. Monson and W. M. Mcclain, J. Chem. Phys., 1970, 53, 29-&. C. E. Moore, Atomic energy levels as derived from the analyses of optical spectra, U.S. National Bureau of Standards, Washington, 1971. A. S. Morlens, P. Balcou, P. Zeitoun, C. Valentin, V. Laude and S. Kazamias, Opt. Lett., 2005, 30, 1554-1556. A. C. Moskun, University of Southern California, 2005. A. C. Moskun and S. E. Bradforth, J. Chem. Phys., 2003, 119, 4500-4515. A. C. Moskun, S. E. Bradforth, J. Thogersen and S. Keiding, J. Phys. Chem. A, 2006, 110, 10947-10955. A. C. Moskun, A. E. Jailaubekov, S. E. Bradforth, G. H. Tao and R. M. Stratt, Science, 2006, 311, 1907-1911. R. S. Mulliken, J. Phys. Chem., 1952, 56, 801-822. 103. M. Musial, Mol. Phys., 2005, 103, 2055-2060. I. Nadler, D. Mahgerefteh, H. Reisler and C. Wittig, J. Chem. Phys., 1985, 82, 3885- 3893. I. Nadler, H. Reisler and C. Wittig, Chem. Phys. Lett., 1984, 103, 451-457. S. W. North, J. Mueller and G. E. Hall, Chem. Phys. Lett., 1997, 276, 103-109. N. Ottosson, M. Faubel, S. E. Bradforth, P. Jungwirth and B. Winter, J. Electron Spectrosc., 2010, 177, 60-70. M. D. Pattengill, Chem. Phys., 1984, 87, 419-429. B. J. Pearson and T. C. Weinacht, Opt. Express, 2007, 15, 4385-4388. V. Pervak, F. Krausz and A. Apolonski, Opt. Lett., 2007, 32, 1183-1185. 178 P. A. Pieniazek, S. E. Bradforth and A. I. Krylov, J. Phys. Chem. A, 2006, 110, 4854-4865. W. M. Pitts and A. P. Baronavski, Chem. Phys. Lett., 1980, 71, 395-399. R. Polak and J. Fiser, J. Mol. Struct. THEOCHEM, 2002, 584, 69-77. N. Pugliano, S. Gnanakaran and R. M. Hochstrasser, J. Photoch. Photobio. A, 1996, 102, 21-28. J. W. Qian, C. J. Williams and D. J. Tannor, J. Chem. Phys., 1992, 97, 6300-6308. D. Raftery, E. Gooding, A. Romanovsky and R. M. Hochstrasser, J. Chem. Phys., 1994, 101, 8572-8579. D. Raftery, M. Iannone, C. M. Phillips and R. M. Hochstrasser, Chem. Phys. Lett., 1993, 201, 513-520. C. A. Rivera, S. E. Bradforth and G. Tempea, Opt. Express, 2010, 18, 18615-18624. C. A. Rivera, S. E. Bradforth, N. Winter, R. Harper and I. Benjamin, in prep, 2011. C. A. Rivera, Y. Zhang, N. Ottosson, B. Winter and B. S. E., in prep., 2010. D. M. Sagar, C. D. Bain and J. R. R. Verlet, J. Am. Chem. Soc., 2010, 132, 6917- 6919. H. F. Schaefer and T. G. Heil, J. Chem. Phys., 1971, 54, 2573-2580. R. Schinke, Photodissociation dynamics: spectroscopy and fragmentation of small polyatomic molecules, Cambridge University Press, Cambridge, 1993. H. Seki and M. Imamura, Bull. Chem. Soc. Jpn., 1971, 44, 1538-1543. H. A. Shen, N. Kurahashi, T. Horio, K. Sekiguchi and T. Suzuki, Chem. Lett., 2010, 39, 668-670. X. L. Shi, F. H. Long and K. B. Eisenthal, J. Phys. Chem., 1995, 99, 6917-6922. A. T. Shreve, T. A. Yen and D. M. Neumark, Chem. Phys. Lett., 2010, 493, 216-219. A. W. Snyder and J. D. Love, Optical Waveguide Theory, Chapman and Hall, New York, 1983. A. Staib and D. Borgis, J. Chem. Phys., 1996, 104, 9027-9039. 179 G. Steinmeyer, Applied Physics A, 2004, 79, 1663-1671. G. Steinmeyer, Appl. Optics, 2006, 45, 1484-1490. C. M. Stuart, M. J. Tauber and R. A. Mathies, J. Phys. Chem. A, 2007, 111, 8390- 8400. R. Szipocs, K. Ferencz, C. Spielmann and F. Krausz, Opt. Lett., 1994, 19, 201-203. K. Takahashi, K. Suda, T. Seto, Y. Katsumura, R. Katoh, R. A. Crowell and J. F. Wishart, Radiat. Phys. Chem., 2009, 78, 1129-1132. K. Tanaka, G. I. Mackay, J. D. Payzant and D. K. Bohme, Can. J. Chem., 1976, 54, 1643-1659. M. J. Tauber and R. A. Mathies, J. Am. Chem. Soc., 2003, 125, 1394-1402. M. J. Tauber, R. A. Mathies, X. Y. Chen and S. E. Bradforth, Rev. Sci. Instrum., 2003, 74, 4958-4960. D. L. Thompson, Modern methods for multidimensional dynamics computations in chemistry, World Scientific, Singapore; River Edge, NJ, 1998. R. Thomson and F. W. Dalby, Can. J. Phys., 1968, 46, 2815. S. A. Trushin, W. Fuss, K. Kosma and W. E. Schmid, Appl. Phys. B, 2006, 85, 1-5. S. A. Trushin, K. Kosma, W. Fuss and W. E. Schmid, Opt. Lett., 2007, 32, 2432- 2434. C. H. Tseng, S. Matsika and T. C. Weinacht, Opt. Express, 2009, 17, 18788-18793. J. C. Tully, J. Chem. Phys., 1990, 93, 1061-1071. J. Vieceli, I. Chorny and I. Benjamin, J. Chem. Phys., 2001, 115, 4819-4828. J. Vieceli, I. Chorny and I. Benjamin, Chem. Phys. Lett., 2002, 364, 446-453. V. H. Vilchiz, X. Y. Chen, J. A. Kloepfer and S. E. Bradforth, Radiat. Phys. Chem., 2005, 72, 159-167. V. H. Vilchiz, J. A. Kloepfer, A. C. Germaine, V. A. Lenchenkov and S. E. Bradforth, J. Phys. Chem. A, 2001, 105, 1711-1723. 180 B. A. Waite, H. Helvajian, B. I. Dunlap and A. P. Baronavski, Chem. Phys. Lett., 1984, 111, 544-548. I. Walmsley, L. Waxer and C. Dorrer, Rev. Sci. Instrum., 2001, 72, 1-29. C. Z. Wan, M. Gupta and A. H. Zewail, Chem. Phys. Lett., 1996, 256, 279-287. X. B. Wang, K. Kowalski, L. S. Wang and S. S. Xantheas, J. Chem. Phys., 2010, 132, 124306. Y. Wang, M. K. Crawford, M. J. McAuliffe and K. B. Eisenthal, Chem. Phys. Lett., 1980, 74, 160-165. R. Weber, B. Winter, P. M. Schmidt, W. Widdra, I. V. Hertel, M. Dittmar and M. Faubel, J. Phys. Chem. B, 2004, 108, 4729-4736. C. J. Williams, J. W. Qian and D. J. Tannor, J. Chem. Phys., 1991, 95, 1721-1737. B. Winter and M. Faubel, Chem. Rev., 2006, 106, 1176-1211. N. Winter and I. Benjamin, J. Chem. Phys., 2004, 121, 2253-2263. N. Winter, I. Chorny, J. Vieceli and I. Benjamin, J. Chem. Phys., 2003, 119, 2127- 2143. C. G. Xia, J. Peon and B. Kohler, J. Chem. Phys., 2002, 117, 8855-8866. S. Yamaguchi and T. Tahara, Chem. Phys. Lett., 2003, 376, 237-243. S. Yamaguchi and T. Tahara, Chem. Phys. Lett., 2004, 390, 136-139. M. Yamashita, M. Ishikawa, K. Torizuka and T. Sato, Opt. Lett., 1986, 11, 504-506. C. Zener, Proc. R. Soc. London Ser. A, 1932, 137, 696-702. Y. Zhang and S. E. Bradforth, in prep., 2011.

Abstract (if available)

Abstract Making a movie of a chemical reaction requires intricate knowledge of both the chemical actors and the environment. Furthermore, the subtleties of the reactive surfaces and the effect of external forces such as solvent electrostatics and molecular collisions on these surfaces often play a major role in the reaction outcome. Many of these external effects begin influencing the reaction within tens of femtoseconds, requiring sophisticated experimental techniques in order to observe dynamics on these timescales.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Water partitioning between the bulk and an electrode surface

PDF

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

PDF

Advancing ab initio QM/MM free energy calculations: refining, validating and quantifying the reference potential approach

PDF

Nanomaterials under extreme environments: a study of structural and dynamic properties using reactive molecular dynamics simulations

PDF

Interactions of planetary surfaces with space environments and their effects on volatile formation and transport: atomic scale simulations

PDF

Photoinduced redox reactions in biologically relevant systems

Asset Metadata

Creator Rivera, Christopher Andrew (author)

Core Title Ultrafast spectroscopic interrogation and simulation of excited states and reactive surfaces

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry (Chemical Physics)

Publication Date 02/02/2011

Defense Date 12/09/2010

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

Tag molecular dynamics,OAI-PMH Harvest,photodissociation,pump-probe,spectroscopy,time-resolved,ultrafast

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Bradforth, Stephen E. (committee chair), Haworth, Ian S. (committee member), Reisler, Hannah (committee member)

Creator Email rivera.chris.a@gmail.com,riverac@umd.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m3636

Unique identifier UC1187271

Identifier etd-Rivera-4289 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-433284 (legacy record id),usctheses-m3636 (legacy record id)

Legacy Identifier etd-Rivera-4289.pdf

Dmrecord 433284

Document Type Dissertation

Rights Rivera, Christopher Andrew

Type texts

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

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu