University of Southern California Dissertations and Theses

Unraveling photodissociation pathways in pyruvic acid and the role of methylhydroxycarbene

(USC Thesis Other)

Unraveling photodissociation pathways in pyruvic acid and the role of methylhydroxycarbene

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Title Page UNRAVELING PHOTODISSOCIATION PATHWAYS IN PYRUVIC ACID AND THE ROLE OF METHYLHYDROXYCARBENE by BIBEK RANJAN SAMANTA 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) August 2020 Copyright 2020 Bibek Ranjan Samanta ii Dedication To my mom, with love. While the other kids played cricket, you made sure I stay and finish my homework on time. Today I feel lucky. Thank you! iii Acknowledgements I still cannot believe that a decision made after I was invited to a group dinner party in the first year of my PhD program, would result in the most rewarding experience of in my life so far. As a confused grad student from an entirely different academic culture, I was drawn to a certain psyche in the Reisler group which I could not quantify as well back then. I can now confidently say that it was a mixture of — welcoming nature of a group of physical chemists who could eat, laugh and solve problems together (30%), my fascination for learning new techniques and instruments (10%), and professor Hanna Reisler’s ability to explain complex scientific problems in creative ways (60%). I knew I had to join this group because I would learn a lot more than just gas-phase spectroscopy and dynamics. As someone who has always second guessed his decisions, I am glad I stuck with this one. Dr. Reisler, you have taught me to never stop questioning until I understand the science behind any problem. You have also helped me identify my weaknesses which I promise to work on. More importantly, you have been the most amazing mentor and if you ever read this, I just want you know how grateful I am to you for the confidence I walk away with from grad school. I am unsure whether my career will lead me to academia or industry, but I promise to pay it forward whenever I get an opportunity to mentor or teach. iv I have received immense help from my labmates, for which I remain forever thankful. Dr. Amit Samanta explained to me the basic instrumentation when I joined the lab, which made for a smoother transition into my research. I am also indebted to him for going over my qualifying exam material with great scrutiny which helped me refine it to acceptable standards. I was lucky to have worked with Dr. Ravin Fernando, a friend, a guide and the first pair of ears to all my crazy experimental ideas. Ravin, you have helped me so much all through these years with setting up experiments, analyzing data and troubleshooting lab equipment. You have also advised me numerous times on how to navigate some of my tougher moments and for that, I am grateful to you. I found an inquisitive, witty and supportive lab partner in Dr. Subhasish Sutradhar. He has held my hand through my screening and qualifying exams, and always had a light moment to spare. I am also glad I shared offices with Daniel Kwasniewski. Dan has been a source for jokes, funny memes and endless encouragement, which made going to lab fun and extensive periods of data analysis in the office seem less tedious. My heartfelt thanks go to the professors who have taught me at USC, in particular, professors Curt Wittig and Jahan Dawlaty. Dr. Wittig always encouraged and challenged my ability to think outside my comfort zone through his courses. Jahan taught me so many important aspects of teaching a graduate level course and was always available to explain tough concepts regardless of their topic of origin. The work presented here would be incomplete without the help of our collaborator Dr. David Osborn. Dr. Osborn’s guidance and thoughtful insights have led me to conduct experiments which at one point seemed futile. His support with the MPIMS machinery v through the late hours of beamtime shifts, and his patience and enthusiasm for discussing scientific ideas form the backbone for a lot of data and analysis presented here. I am extremely grateful to the USC chemistry staff for their continued support. Michele Dea had answers to every administrative problem and I would not have been able to contribute to the Chemistry Graduate Student Organization without her help. Magnolia Benitez would happily entertain my last-minute requests for forms and signatures. I was also fortunate enough to have forged good friendships over these years which made this journey enjoyable. The bond with my roommates, Debanjan Mitra and Dibyendu Mondal, both chemistry grad students, was developed over shared experiences. Myungjin Lee was always available for impromptu boba runs, Gozde Sahin helped me explore the diverse Los Angeles food scene, and a phone call with Avichal Vaish would always keep my spirits high. I am also thankful to my parents and my brother, who endeared me with their love and support. I am especially grateful to my father, who emphasized the importance of perseverance while learning. Finally, I would like to thank Priyom Adhyapok for being my support system through all my smiles and tears over the last 7 years. Thank you for loving me and making me feel ever so brave with every major step I take. vi Table of Contents Dedication ii Acknowledgements iii List of Tables x List of Figures xii Abstract xix Introduction 1 Pyruvic Acid .............................................................................................................................................. 4 Atmospheric implications .................................................................................................... 4 What’s (un)known? ................................................................................................................ 6 Methylhydroxycarbene..................................................................................................................... 12 Implications: an elusive intermediate ......................................................................... 12 Overview of experimental observations and challenges .................................... 14 References .............................................................................................................................................. 15 Experimental techniques 22 Introduction........................................................................................................................................... 22 Molecular beam and VMI instrument ......................................................................................... 25 Molecular beam through a pulsed nozzle .................................................................. 26 Photolysis and ionization .................................................................................................. 27 Time-of-flight (TOF) mass spectrometer ................................................................... 28 Sliced velocity map imaging (SVMI) ............................................................................. 30 vii Third-harmonic vacuum ultraviolet (VUV) generation using a gas cell ....... 35 Low pressure flow reactor and MPIMS ..................................................................................... 36 Flow tube reactor ................................................................................................................. 38 Photoionization mass spectrometer ............................................................................ 39 Multiplexed data acquisition ........................................................................................... 40 References .............................................................................................................................................. 44 Two photon absorption and fragmentation processes 48 Introduction........................................................................................................................................... 48 Experimental section ......................................................................................................................... 50 Results ...................................................................................................................................................... 52 Photofragment yield spectra ........................................................................................... 52 Fragments’ kinetic energy release ................................................................................ 55 Discussion ............................................................................................................................................... 60 Excited states of PA .............................................................................................................. 60 Energetically allowed reactions ..................................................................................... 61 Two-body fragmentation pathways ............................................................................. 63 Three-body fragmentation pathways .......................................................................... 64 Mechanistic interpretations ............................................................................................. 76 Summary and conclusions .............................................................................................................. 79 References .............................................................................................................................................. 80 Pyruvic acid photodissociation at 193 nm 85 Introduction........................................................................................................................................... 85 Experimental section ......................................................................................................................... 89 VMI experiments ................................................................................................................... 89 MPIMS experiments ............................................................................................................ 91 Results and Analysis .......................................................................................................................... 92 Molecular beam studies using VMI ............................................................................... 92 Low pressure flow reactor studies using MPIMS ................................................... 95 Determination of product yields .................................................................................. 104 viii Multivariate linear regression for pathway contributions ............................... 114 Discussion ............................................................................................................................................. 118 Excited States of PA ........................................................................................................... 118 Dissociation pathways...................................................................................................... 118 Formation Mechanisms of Main Products ............................................................... 122 Summary and Conclusions ............................................................................................................ 126 References ............................................................................................................................................ 128 Pyruvic acid photochemistry initiated on S1 state 133 Introduction......................................................................................................................................... 133 Experimental section ....................................................................................................................... 136 Results and Discussion ................................................................................................................... 137 S1←S0 spectroscopy in a molecular beam ................................................................ 137 Decarboxylation reaction ................................................................................................ 139 MHC formation and decay .............................................................................................. 144 Possible secondary reaction of MHC .......................................................................... 152 Conclusion ............................................................................................................................................ 156 References ............................................................................................................................................ 157 Electronic states of hydroxycarbenes 160 Introduction......................................................................................................................................... 160 Computational details ..................................................................................................................... 162 Results and discussion .................................................................................................................... 164 Molecular orbital framework, electronic configurations, and relevant geometries ............................................................................................................................................... 164 Excited electronic states .................................................................................................. 166 Cations and Rydberg states ............................................................................................ 169 Detection of hydroxycarbenes ...................................................................................... 176 Conclusions .......................................................................................................................................... 178 References ............................................................................................................................................ 179 ix Future work 183 Introduction......................................................................................................................................... 183 Investigating PA + H2O interactions .......................................................................................... 185 Photoinduced reactions ................................................................................................... 185 Photodissociation dynamics of dimers and higher order clusters ............... 187 Nature of the S1 state ....................................................................................................................... 189 Light emission studies ...................................................................................................... 189 Experiments with deuterated PA ................................................................................ 191 References ............................................................................................................................................ 191 Appendices 196 A Three-body fragmentation KEmax simulations ..................................................................... 196 A.1 Synchronous three-body fragmentation MATLAB code .................................... 196 A.2 Synchronous three-body fragmentation MATLAB code .................................... 197 B HC and MHC geometries ................................................................................................................ 199 B.1 Hydroxycarbene (HC) ....................................................................................................... 199 B.2 Methylhydroxycarbene (MHC) ..................................................................................... 201 C MPIMS gas connections .................................................................................................................. 205 C.1 Bubbler dimensions for MPIMS use ........................................................................... 205 C.2 MPIMS manifold and flow setup .................................................................................. 207 D KER distributions from co-fragment internal energy states .......................................... 209 D.1 Benzene-water MATLAB code (detecting water) ................................................. 209 E Absolute photoionization cross-section calculations ........................................................ 211 E.1 Absolute photoionization cross-section tables ..................................................... 212 F Eckart simulations ............................................................................................................................ 216 F.1 Function file for defining Eckart barrier .................................................................. 216 F.2 Main code for plots and integration ........................................................................... 217 x List of Tables Table 3.1 Calculated transition dipole moments between the indicated states, their oscillator strengths and vertical excitation energies are obtained using EOM-EE-CCSD/6- 311(2+)G**. The X, Y, and Z axes are defined with respect to the molecular plane (XY) containing the CC(O)C(O)O skeleton. ........................................................................................................... 61 Table 3.2 Estimated angle α and corresponding KEmax values of fragments produced via synchronous three body-fragmentation reactions 1a–1c. The error bars on KEmax represent the deviation from the average value for a change of ± 5° in the corresponding α value. Excitation energy, ℎ𝜈𝜈 , corresponding to the employed probe wavelength ais used for this table. For 1a, 1b and 1d, ℎ𝜈𝜈 = 54000 cm -1 (~370 nm). For 1c, during CH3 detection ℎ𝜈𝜈 = 59500 cm - 1 (~ 335 nm). ...................................................................................................................................... 69 Table 3.3 Computed KEmax values of fragments produced in the secondary reactions 5 – 8, correlated with the observed KEmax of CH3CO or HOCO generated in reaction 1. .................... 75 Table 4.1 Observed photoproducts in PA photodissociation, their IEs, mole fraction yields 𝑓𝑓𝑓𝑓 , and the photon energies employed for mole fraction yield determination. ....................... 111 Table 4.2 Reaction mechanisms, primary two-body steps, and final fragmentation channels, their ΔHrxn values,[2,19,39] and percent contributions. A 193 nm photon supplies 51,800 cm -1 of energy. The percent contribution of each mechanistic class is generated by summing contributions from Table 4.3. .................................................................................................... 113 Table 4.3 Possible reaction pathways and their percent contribution. ..................................... 117 Table 5.1 Observed S1←S0 rovibtonic bands and the notations of the normal modes are based on the frequencies computed in reference [8]. The unassigned transitions were attributed to splittings in internal rotor transitions of the CH3 moiety. ..................................... 138 Table 5.2 Summary of thermodynamic values for MHC isomerization reaction. [12] ........ 145 Table 6.1 Comparison of S-T gap energy (in kcal/mol) calculated in this work at the EOM- EE-CCSD/aug-cc-pVTZ basis set with previously reported values.[11-12,18] ......................... 165 Table 6.2 Vertical singlet excitation energy results from EOM-EE-CCSD calculation. Five A’ states (white) and five A” states (gray) are calculated. ...................................................................... 167 xi Table 6.3 Vertical and adiabatic excitation energies of singlet Rydberg states, oscillator strengths and quantum defects for HC and MHC. Quantum defects are estimated from the Rydberg formula (see text). ............................................................................................................................ 171 Table E.1 PA PI cross-section values, 𝜎𝜎 𝑃𝑃 𝑃𝑃 𝑎𝑎 𝑎𝑎 𝑎𝑎 , in Mb (1 Mb = 10 -18 cm 2 molecule -1 ). The uncertainty in the reported values is ~25%. .......................................................................................... 212 Table E.2 CO PI cross-section values, 𝜎𝜎 𝐶𝐶𝐶𝐶 𝑎𝑎 𝑎𝑎 𝑎𝑎 , in Mb. The uncertainty in the reported values is ~20%. ....................................................................................................................................................................... 213 Table E.3 CO2 PI cross-section values, 𝜎𝜎 𝐶𝐶𝐶𝐶 2 𝑎𝑎 𝑎𝑎 𝑎𝑎 , in Mb. The uncertainty in the reported values is ~20%. ....................................................................................................................................................................... 214 xii List of Figures Figure 1.1 While most photodissociation studies of PA reported that the decarboxylation process yields acetaldehyde as the major product, this study unearths several other photodissociation products unique to higher excited states. Although the exact mechanism of their formation needs more theoretical analysis, this study presents the first comprehensive map of reaction pathways for PA. .................................................................................... 3 Figure 1.2 The cascading series of oxidation reactions which stem from isoprene and contribute to biospheric SOA formation (adapted from reference [24]). ....................................... 5 Figure 1.3 UV-Vis absorption spectrum PA at room temperature (298 K). (adapted from reference [1]). This transition has a λmax near 350 nm and can be readily accessed by the near-UV solar radiation near the Earth’s surface.[46] ............................................................................. 7 Figure 1.4 The ground state structure of PA optimized at MP2/6-31+G* level (top) and a canonical representation of its HOMO and LUMO orbitals. ................................................................... 8 Figure 1.5 Acetaldehyde PES with stationary points involving its isomerization (black line) and dissociation pathways (blue line). Energies are in kcal/mol, relative to the global minimum acetaldehyde. (Adapted from ref [73]) ................................................................................... 13 Figure 2.1 Photodissociation dynamics in a molecular beam using VMI and chemical kinetic studies using the MPIMS have been used together to explore the complicated chemical systems described in this dissertation. .................................................................................... 22 Figure 2.2 Schematic of the VMI apparatus is shown along with its main components (bold face) and their intended purpose (regular font). The source and detection chambers are maintained under high vacuum conditions. The PA/He partial pressures used in this apparatus are also included. ............................................................................................................................. 25 Figure 2.3 A schematic representation of a VMI steps for an example molecule (here, CO2). (a) Adiabatic expansion of the reactant molecules in a buffer gas (He or Ar) through a pulsed nozzle creates a moleuclar beam. (b) The molecular beam is irradiated using a photolysis laser which enables (c) photoexcitation and (d) subsequent dissociation of the molecule. Additional laser(s) can be used to resonantly ionize a fragment of interest (for example, O( 3 P) using 225 nm) for state-selective detection. (e) The resulting spatial xiii distribution of ions contains information about the dissociation process (KE, anisotropy, etc.). ............................................................................................................................................................................. 28 Figure 2.4 (a) The basic principle of VMI is depicted using an exampe molecule (CO2). The corresponding 3D ion spheroid distribution and the desired final energy distribution is also shown. (b) The projection of the ion spheroid on the detector in a conventional VMI setup with a DC voltage supplied to the MCP plates. (c) Only the central 2D slice of the ion spheroid is projected by pulse gating the detector voltage, minimizing the need for image reconstruction. ....................................................................................................................................................... 32 Figure 2.5 Schematic of the MPIMS apparatus used at the ALS. The main components of the apparatus are labeled. He bubbled through a glass bulb containing PA is flown into the flow-tube reactor. After photolysis, the molecules are photoionzied using synchrotron radiation. The attached OA-TOF arrangement provides enhanced mass resolution. ............. 38 Figure 2.6 Three-dimensional dataset measured from the photoinitiated reactions in the flow tube (adapted from ref. [6]) The position on the detector linearly correlates with an ion’s TOF and is related to the m/z ratio of the ion. Integrated cross-sectional slices through the data can be taken to generate 2D plots which show the TOF spectrum as a function of reaction time (left, middle) or as a function of the photoionization energy (right, middle). From these 2D plots, integrated photoionization efficiency curves can be generated as functions of reaction time (left, bottom) or photoionization energy (right, bottom) for a given m/z ratio. ...................................................................................................................................................... 42 Figure 3.1 The two photon-absorption via the long lived S1 state to the S2 state and subsequent three-body fragmentation into smaller fragments discussed in this chapter are depecited. Norrish type I bond fission behavior in keto-acids like PA have not been reported previouly. ............................................................................................................................................... 49 Figure 3.2 The jet-cooled spectrum of the first excited state of PA is inferred from the H- PFY spectrum obtained in He (red). The corresponding broad features in the room temperature absorption spectrum[10] (black) is shown for comparison. .................................. 53 Figure 3.3 The Tc and Tt conformers of PA are present in the ground state with Boltzmann population values of 97% and 3%, respectively, at 298 K. Intramolecular H-bonding makes the Tc conformer ∼2.0 kcal/mol more stable than the Tt conformer.[11-12] ........................... 54 Figure 3.4 PFY spectra of the (a) CH3CO and (b) HOCO fragments produced from PA photodissociation. ................................................................................................................................................. 55 Figure 3.5 KER distributions (top) and corresponding recoil anisotropy parameters, β, (bottom) obtained by monitoring CH3CO fragments following S1 excitation at two different peaks of the PFY spectrum in Figure 3.2. The simulated KEmax values based on three-body fragmentation channels are marked by arrows. ...................................................................................... 56 xiv Figure 3.6 KER distributions (top) and corresponding recoil anisotropy parameters, β, (bottom) of H fragments following S1 excitation at two different peaks of the H-PFY spectrum. The simulated KEmax values based on the indicated three-body fragmentation channels are marked by arrows. ..................................................................................................................... 57 Figure 3.7 KER distribution of CO(X 1 Σ + , v’’=0, rotational bandhead) products detected by (2+1) REMPI at 230.1 nm following PA excitation at 374.4 nm. The predicted KEmax values for synchronous and sequential three-body fragmentation processes are marked by arrows. ....................................................................................................................................................................... 58 Figure 3.8 KER distribution of HOCO products detected by non-resonant ionization following PA excitation at 369.7 nm. The predicted KEmax values for synchronous and sequential three-body fragmentation processes are marked by arrows. ..................................... 59 Figure 3.9 A synchronous three body fragmentation of molecule ABC is depicted. The X- axis is aligned in the direction of 𝑝𝑝 𝐵𝐵 � � � � ⃗ and as a result it divides the critical A–B–C angle, α, equally. ....................................................................................................................................................................... 65 Figure 3.10 KEmax of the final products of synchronous three-body fragmentation reactions are simulated for (a) reaction 1a, (b) reaction 1b, (c) reaction 1c, and (d) reaction 1d as a function of α. These plots are calculated for an excitation energy of 54000 cm -1 . .................... 68 Figure 3.11 A schematic of a sequential three-body fragmentation of ABC where the B-C bond breaks first at time τ1 producing AB and C fragments. The AB fragment can further dissociate after time τ2, which is longer than the mean rotational period of the AB intermediate. The observed laboratory-frame velocities of the A and B fragments depend on the velocity of AB. ........................................................................................................................................... 71 Figure 3.12 KEmax of the final products of sequential three-body fragmentation reactions are simulated for reaction 1 followed by (a) reaction 5, (b) reaction 6, (c) reaction 7, and (d) reaction 8 as a function of the KER in reaction 1. These plots are simulated for an excitation energy of 54000 cm -1 . .................................................................................................................... 74 Figure 4.1 The results in this chapter indicate that following 193 nm excitation, PA can undergo several types of reaction. The branching ratios are dictated by the nature of the electronic excited states ..................................................................................................................................... 88 Figure 4.2 H-fragment KER distribution (one-laser background subtracted) obtained at 193 nm (black; solid) compared to the previously reported distribution obtained for H- fragments following 2-photon photodissociation of PA with excitation wavelength 374 nm (red; dashed).[2] The predicted maximum KER values for several possible synchronous three-body fragmentation (blue) and two-body fragmentation (green) reactions are marked by arrows. ................................................................................................................................................ 93 xv Figure 4.3 KER distributions (top) and corresponding recoil anisotropy parameters β (bottom) of CO fragments in v=0 (black) and v=1 (red). ..................................................................... 95 Figure 4.4 Kinetic time trace of m/z = 88.02 showing the depletion in the PA ion signal upon 193 nm photolysis. The pre- and post-photolysis signal averages are represented by blue and green solid lines, respectively. ...................................................................................................... 96 Figure 4.5 Integrated TOF mass spectra in the range 8.9 – 9.2 eV (a), and 10 – 10.7 eV (b). Products that were identified using their kinetic time traces and PI spectra are labelled. .. 98 Figure 4.6 PI spectra of (a) m/z = 17; (b) m/z = 42; (c) m/z = 15; and (d) m/z = 44 observed in PA photodissociation (red), compared to published PI spectra (black) of the hydroxy radical,[23] ketene,[24] methyl radical,[25] vinyl alcohol, and acetaldehyde.[26] ..................................................................................................................................................................................... 100 Figure 4.7 Early time (0–3 ms) TOF mass spectrum highlighting short-lived photoproducts of deuterated PA (red) and PA (blue) for (a) m/z=28–47 and (b) m/z=15–19. The negative scale for the PA mass spectrum is used for ease of comparison. m/z=18 from d1-PA identified as OD from its (c) kinetic time trace which resembles that of an unstable product and (d) its ionization spectrum which matches the previously reported absolute PI spectrum for OH radical.[23].......................................................................................................................... 102 Figure 4.8 (a) Kinetic time traces of m/z = 42, 44 and 58 compared to the instrument response time. The bimolecular reaction product (m/z = 58) has a slower rise time than that of primary photoproducts. (b) The kinetic time trace (red) of ketene photoproduct from acetone photodissociation is shown. The 10% to 90% signal rise is shown by the shaded region and the corresponding response time period is marked by arrows. ............. 104 Figure 4.9 (a) Plot of the ion detection efficiency vs. mass for the MPIMS setup. The mass discrimination factor is extracted from a power function fit. (b) The absolute PI cross section of PA from threshold to 14 eV (black; circles), with uncertainty (gray; shaded) shown. Absolute photoionization cross-sections of (c) CO and (d) CO2 from threshold to around 14 eV obtained at 0.005 eV steps. ................................................................................................ 106 Figure 4.10 Kinetic time traces and their decay fits (given by equation 4.3) of (a) methyl radical at 10.25 eV; (b) hydroxyl radical at 13.25 eV; (c) formyl radical at 10.25 eV; and (d) acetyl radical at 9.7 eV. ...................................................................................................................................... 108 Figure 5.1 A cartoon highlighting the slow H tunneling of MHC to its isomers: acetaldehyde and vinyl alcohol, because of the existence of significant barriers to H exchange. A relatively fast MHC removal in our experiment raises the possibility of secondary reactions.............. 135 Figure 5.2 H-PFY spectrum of PA in Ar. Peak assignments are labeled based on tentative assignments carried out in reference [11]. .............................................................................................. 137 xvi Figure 5.3 Photoionization (PI) spectra measured at a) 9.2–10.4 eV showing the presence of both vinyl alcohol (VA) and acetaldehyde (Ac) photoproducts, and b)13.7–14.1 eV showing the CO2 photoproduct. The photoproducts are identified by comparing the PI spectra obtained in this work to known absolute PI curves of these molecules (Daniel Rösch, private communication). ................................................................................................................... 142 Figure 5.4 Fluence dependent signals of PA photoproducts a) acetaldehyde and vinyl alcohol at 10.25 eV b) vinyl alcohol at 9.65 eV c) CO2 at 14.15 eV and d) species M at 9.65 eV. The plotted signal intensities, fit to a quadratic equation with (0,0) as intercept, show that the fits are nearly linear. ......................................................................................................................... 143 Figure 5.5 The computed Eckart potential (top panel) is plotted for both acetaldehyde and vinyl alcohol. Corresponding plots of transmission factors at T = 20, 215 and 400 K are also shown (bottom panel). The acetaldehyde to vinyl alcohol ratio of integrated areas under the plot of transmission factors at 215 K is 2:1. ..................................................................................... 146 Figure 5.6 (a) Integrated (over 8.4–9.1 eV ionization) kinetic time traces of m/z=44.03 (MHC) and m/z=45.03 (d1-MHC) from PA and d1-PA photodissociation, respectively. The blue dashed line represents the noise baseline averaged at time 2 – 20 ms. (b) The PI spectrum of m/z=45.03 obtained in the photodissociation of d1-PA. .......................................... 148 Figure 5.7 Kinetic time traces of acetyl (left) and DOCO (right) from d1-PA photodissociation at 9.65 eV PI energy of. The 13 C background of the dominant d1-vinyl alcohol product is shown by the blue dashed line on the right panel. ......................................... 150 Figure 5.8 Possible state-specific dissociation pathways of PA excited at 351 nm. Energies are in kcal/mol. A 351 nm photon supplies 81.5 kcal/mol of energy. ......................................... 151 Figure 5.9 a) Plot of kinetic time vs. m/z in the vicinity of m/z=88. The pre- photodissociation background is not subtracted. Integrating over kinetic times b) when all products have been pumped out (65 – 84 ms) c) when photodissociation products are present in the tube (0 – 30 ms) and shows two different mass peaks. Integrating over the lighter mass gives time trace d) showing 0.3% depletion consistent with PA and e) the heavier mass (species M) shows time trace resembling a stable photoproduct. .................... 153 Figure 5.10 a) TOF spectra of PA (red trace) and d1-PA (blue trace, inverted for clarity) showing the observed change in mass peaks due to deuteration. b) Species M signal changes linearly upon change in PA concentration and is independent of the mode of PA sample preparation demonstrated by two different methods – He bubbled through a glass bulb containing PA (red circles) and another delivered via a cylinder (blue triangles) c) At low PA partial pressure (~0.2 mTorr) species M shows a slower rise time while, VA, a primary photoproduct shows a rise time limited by the instrument response time. The plots shown in this figure are all obtained at a photon energy of 9.65 eV. ................................ 154 xvii Figure 6.1 Photolytic generation and stabilization of hydroxycarbenes is extremely difficult. They quickly isomerize to more stable conformers and in most cases only the aldehydic isomer is spectroscpically observed ...................................................................................... 161 Figure 6.2 A molecular orbital diagram for trans-HC. The sp 2 hybridized C atom orbitals are shown on the left, and the sp 3 hybridized O atom orbitals are shown on the right. LP: Lone Pair; LPC is the sp 2 hydridized carbon orbital in the y-direction which does not mix with any other orbital. LPO is formed from two of the sp 3 hydridized orbitals on oxygen. 164 Figure 6.3 Representation of the orbitals in going from a planar (i) to a twisted geometry (ii) in the first excited state of HC. The first row, represents the stabilization of the electron deficient C orbitals by the O lone pairs. This stabilization can also be explained from the molecular orbital picture (second row). ................................................................................................... 166 Figure 6.4 Vertical excitation energies and oscillator strengths calculated using EOM- CCSD/aug-cc-pVTZ for the cis- and trans-isomers of HC (left) and MHC (right). Vertical singlet electronic energies (regular font) and oscillator strengths (bold font, square bracket) are calculated at the ground-state optimized geometries. Vertical IE (black dashed line) and ΔEcis-trans values are also listed. .................................................................................................. 170 Figure 6.5 Particle NTOs, and the hole NTO (labeled HOMO) for the lowest excited states of trans-MHC (left) and trans-HC (right); their directional spatial extents in Å are labeled as 〈 𝑟𝑟 〉( 〈 𝑟𝑟 𝑥𝑥 〉, 〈 𝑟𝑟 𝑦𝑦 〉, 〈 𝑟𝑟 𝑧𝑧 〉); The orientation of the molecules has been kept the same in all pictures. ..................................................................................................................................................................................... 174 Figure 6.6 Particle NTOs, and the hole NTO (labeled HOMO) for the lowest excited states of cis-MHC (left) and cis-HC (right); their directional spatial extents in Å are labeled as 〈 𝑟𝑟 〉( 〈 𝑟𝑟 𝑥𝑥 〉, 〈 𝑟𝑟 𝑦𝑦 〉, 〈 𝑟𝑟 𝑧𝑧 〉); The orientation of the molecules has been kept the same in all pictures. ..................................................................................................................................................................................... 175 Figure 7.1 Molecular structures of (a) keto-pyruvic acid; (b) enol-pyruvic acid; (c) 2,2- dihydroxypropanoic acid (gem-diol); and (d) parapyruvic acid. ................................................... 185 Figure 7.2 Vibrational spectrum of aqueous PA in the mid-IR range adapted from reference [23]. Several characteristic OH stretching frequencies in the 2800–3800 cm -1 range are labelled. .................................................................................................................................................................... 188 Figure C.1 Schematic of two glass bubbler setups for standard 10–15 ml sample volume (left) and for 2–3 ml small sample volume (right). The conical bottom of the small sample bubbler is more efficient for gas dispersion when samples are sparse, for example, in the case of deuterated PA. ....................................................................................................................................... 205 Figure C.2 Complete flow setup and connections to the reactor tube for a single bubbler, with the current 5-inlet Sandia yoke setup. All the inlets have independently controlled mass flow controllers. ....................................................................................................................................... 207 xviii Figure C.3 Complete flow setup and connections to the reactor tube for a double bubbler setup, with independently controlled flow-restricting needle valves. ......................................... 208 Figure E.1 Absolute photoionization cross-sections of (a) CO and (b) CO2 from threshold to around 14 eV obtained at 0.005 eV steps. ................................................................................................ 212 xix Abstract The photodissociation of pyruvic acid (PA) was studied in the gas-phase using two complementary techniques. The time-sliced velocity map imaging (VMI) arrangement was used to determine kinetic energy release distributions of fragments and estimate dissociation timescales. The multiplexed photoionization mass spectrometer (MPIMS) setup was used to identify and quantify photoproducts, including isomers and free radicals, by their mass-to-charge ratios, photoionization spectra, and kinetic time profiles. PA’s dissociation behavior is probed following excitation to the first three singlet excited states — S1, S2 and S3. PA is introduced in a molecular beam, and following two-photon excitation via the first absorption band (S1←S0) at 330−380 nm, CH3CO, HOCO, CO, CH3, and H fragments are detected as photodissociation products. The sharp vibrational features in the H photofragment yield spectrum match well the broad features observed in the room temperature absorption spectrum, indicating that the S1 state is long-lived (> 1 picosecond). VMI is used to determine kinetic energy release (KER) and angular distributions of CH3CO, HOCO, CO, CH3, and H fragments, which show that an additional photon absorption from S2←S1 is facile and is followed by rapid dissociation to the observed fragments. On the basis of the energetics of the different dissociation pathways and analyses of the observed KER xx distributions, three-body fragmentation processes are proposed as major contributors to the formation of the observed products. Upon excitation of the S3 state using a single 193 nm photon, CO2, CO, H, OH, HCO, CH2CO, CH3CO and CH3 are determined to be major photodissociation products of PA at 193 nm. Experiments using both VMI and MPIMS reveal that three-body fragmentation processes are dominant. Some dissociation pathways are believed to proceed via internal conversion to lower excited states on which decarboxylation is favored. Acetaldehyde and vinyl alcohol are only minor co-products from the decarboxylation process at this wavelength, but products that are known to arise from their unimolecular dissociation, such as HCO, H2CO and CH4, are identified and quantified. In fact, most of the decarboxylation pathways end up in three-body fragmentation. A multivariate analysis, which takes into account the yields of the observed products and assumes a set of feasible primary dissociation reactions, offers the first comprehensive description of the dissociation pathways of PA initiated on the S3 excited state. Most of the observed products and yields are rationalized on the basis of three reaction mechanisms: (i) decarboxylation terminating in CO2 + other primary products (~50%); (ii) Norrish type I dissociation typical of carbonyls (~30%); and (iii) O-H and C-H bond fission reactions generating H atom (~10%). The analysis shows that most of the dissociation reactions create more than two products. This observation is not surprising considering the high excitation energy (~51,800 cm -1 ) and fairly low energy required for dissociation of PA. Experiments with d1-PA (CH3COCOOD) support the interpretations. The dissociation on S3 is fast, as indicated by the products’ recoil angular anisotropy, but the roles of internal conversion and intersystem crossing to lower states are yet to be determined. xxi Excitation to the S1 state via a single photon of 351 nm confirms that decarboxylation is favored at low excitation energies. Previous experiments have shown that following excitation to the S1 state, acetaldehyde and CO2 are produced. Using the MPIMS, we show that the hypothesized methylhydroxycarbene (CH3COH; MHC) intermediate is a nascent product of the photodissociation process at 351 nm. We obtain its kinetic time trace and attribute its rapid disappearance to isomerization and secondary reactions. MHC is produced with significant excitation energy (>14,000 cm -1 ) and is observed to form both the kinetically favored vinyl alcohol isomer and the thermodynamically favored acetaldehyde isomer. MHC is believed to live sufficiently long to undergo secondary reactions with other molecules. To aid in designing strategies for state-selective detection of hydroxycarbenes via ionization, vertical and adiabatic excitation energies and oscillator strengths for valence and Rydberg states of HC and MHC are reported. The electronic states’ characters are analyzed by plotting natural transition orbitals (NTOs). The calculations demonstrate that the shape, size, and energy of each Rydberg orbital are affected to varying degrees by their interaction with the ion core. 1 Introduction The UV photochemistry of organic molecules and the dissociation dynamics that lead to final products are fundamental processes that govern reactions in the atmosphere, reactivity on the surface of organic aerosols, and biological damage caused by UV irradiation. In many excited molecules, the ensuing dynamics include processes that are in competition with direct dissociation on the excited state, e.g., couplings to lower potential energy surfaces (PESs), isomerization, phototautomerization, and secondary collisions. These excited state pathways, and dynamics of product formation are often of more pedagogical interest to physical chemists than just the identity of products. This is because the overall photodissociation is usually fast (picoseconds to microseconds), far from equilibrium, and kinetic competition often determines final product outcomes. Therefore, understanding this complex behavior often requires the use of multiple experimental approaches and the support of theory. In our work, as well as that of others, advances have been achieved by first using simpler model molecules that allow detailed exploration, and then projecting the results to larger members of the series. 2 This dissertation focuses on Pyruvic acid (PA, CH3COCOOH), which is a proxy molecule for studying the behavior of α-keto carboxylic acids in the atmosphere. PA is a deceptively simple molecule whose UV photodissociation dynamics is complex, and remains poorly understood. It has only three carbon atoms, comprising a carbonyl, carboxylic, and methyl groups (Figure 1.4). Following π*←n excitation to the first excited state,[1] PA can behave like a carbonyl, with dominant internal conversion and intersystem crossing pathways. It can also behave as a carboxylic acid undergoing decarboxylation to yield the most commonly observed photodissociation product — acetaldehyde. In fact, the first photolysis studies of PA, concerned with the first excited state, determined acetaldehyde and CO2 as major products.[2-10] Because of its rich photochemistry, PA has attracted much research effort. However, our understanding of its photoinitiated chemistry is incomplete, and some of the results are the subject of controversy. This dissertation addresses several gaps which existed in the understanding of the nature of its excited states, and in the identification of its primary photoinduced reaction products in the gas phase. This work establishes that PA photochemistry is in fact more complicated than initially expected, and upon excitation to higher excited states, several photodissociation pathways leading to radical and neutral fragments become accessible. These products are depicted pictorially in Figure 1.1. We became interested in PA as a possible source of the methylhydroxycarbene (MHC), the elusive isomer of acetaldehyde. α-keto carboxylic acids like PA are considered to be a good source of hydroxycarbenes like MHC.[2,11-14] In recent times, decarboxylation of these acids via pyrolysis has been a preferred method for hydroxycarbene generation below 3 their isomerization barriers.[11-12] It has only been implicated in photolysis studies and had not been directly observed in a photoinduced reaction. This work describes possible detection schemes for hydroxycarbenes.[15] This work also reports the first detection of MHC in a photolysis experiment and establishes its participation in unimolecular dissociation, isomerization and possible secondary reactions of PA. Figure 1.1 While most photodissociation studies of PA reported that the decarboxylation process yields acetaldehyde as the major product, this study unearths several other photodissociation products unique to higher excited states. Although the exact mechanism of their formation needs more theoretical analysis, this study presents the first comprehensive map of reaction pathways for PA. 4 Pyruvic Acid Atmospheric implications Organic molecules and their photochemistry have a decisive impact on Earth’s atmospheric composition and chemistry. These processes play an important role in determining the concentration of pollutants, aerosols, and trace gases.[16-19] As such, volatile organic compounds (VOCs) account for a significant majority of carbon concentration in the troposphere, with an estimated worldwide flux of the order of 1000 TgC/yr a .[20-22] Biogenically emitted isoprene single-handedly contributes to the majority of the total VOC budget (∼500 TgC/yr),[20,22] and when oxidized, leads to several oxygenated and hydroxylated species that are associated with the formation of secondary organic aerosols (see Figure 1.2).[23-30] PA, a keto-acid intermediate among the cascading network of isoprene oxidation reactions,[25,27,30] is abundant in the atmosphere, both in the gas and aqueous (aerosols) phases. Field and laboratory studies confirm that isoprene released from forested areas like the Amazon basin and the eastern United States is a major biogenic source of PA.[31-32] These studies also concluded that biogenic isoprene oxidation is a global a phenomenon and not confined to those regions. On average 600 Tg/yr are emitted from trees and shrubs which result in a substantial amount of pyruvic and glyoxylic acid formation.[22] a Unit of mass: 1 TgC (Teragrams Carbon) = 1 MtC (Milliontonnes Carbon) 5 Figure 1.2 The cascading series of oxidation reactions which stem from isoprene and contribute to biospheric SOA formation (adapted from reference [24]). 6 The atmospheric degradation pathway of PA is not akin to other VOCs which react mainly with hydroxyl radicals (OH) in the atmosphere. PA absorbs near-UV radiation from the solar actinic flux reaching the Earth’s surface, and is oxidized relatively slowly by H2O2 or the OH radical. In fact, under atmospheric conditions, direct photolysis of PA dominates over OH oxidation by several orders of magnitude.[8,33-34] In the aqueous phase, PA photochemistry is found to be responsible for oligomer formation and the production of secondary organic aerosols (SOA).[6-7,33,35-37] As previously mentioned, PA is a deceivingly simple molecule with just three carbon atoms and yet much of its photochemistry is either unknown or controversial. PA chemistry spans multiple fields and is relevant to understanding several key photochemical reactions of ketoacids in the environment. Some examples include — its gas and aqueous photochemistry,[8,25,33,38] thermal decomposition,[39-41] and infrared multiphoton pyrolysis.[42] Room temperature PA (in liquid phase) can spontaneously oligomerize, even in the dark to form several dimer species or adducts like zymonic acid and parapyruvic acid.[43] What’s (un)known? Much research has been devoted to the detection and degradation of PA in the gas and condensed phases under solar irradiation.[3-4,6,8,10,14,27,33,44-45] PA readily undergoes photolysis in the troposphere because the absorption spectrum (see Figure 1.3) to its first excited state (S1), an n to π* electronic transition (see Figure 1.4) involving the 7 carbonyl functionality,[1] is in near-complete overlap with the solar radiation available near the Earth’s surface.[46] Figure 1.3 UV-Vis absorption spectrum PA at room temperature (298 K). (adapted from reference [1]). This transition has a λmax near 350 nm and can be readily accessed by the near-UV solar radiation near the Earth’s surface.[46] Early literature on the photochemistry of PA vapor at low pressures (less than 150 Torr) reveals products from its photodissociation using narrowband radiation (λ ≤ 366 nm).[2,4,44] These studies reported a quantum yield of unity for PA degradation upon photolysis, and CO2 (∼100% yield) and acetaldehyde (45−80% yield) were identified as major products.[2,4] Whereas C-C bond cleavage is generally associated with molecules containing the carbonyl functionality, such pathways have not been observed for PA in the troposphere.[47-51] The lowest atmospherically feasible bond fission pathway in PA is breaking of the C-C bond between the keto and acid functional groups (PA → CH3CO + 8 COOH).[52-54] Although this pathway can be accessed with λ ≤ 344 nm, barriers to this bond fission reaction are currently unknown, and decarboxylation involving H-transfer coupled with C-C cleavage is believed to be significantly more favorable.[2,44] As a result, the deviation of PA from the conventional photochemistry of RCOR’ compounds can possibly be attributed to the internal hydrogen bonding between the keto and the acetyl moieties in the cis-keto (Tc) lowest energy conformer. The barrier to C-C cleavage is significantly reduced due to a concerted or step-wise H transfer, believed to progress through a five-membered, cyclic transition state.[4,41,52] Figure 1.4 The ground state structure of PA optimized at MP2/6-31+G* level (top) and a canonical representation of its HOMO and LUMO orbitals. b b Only canonical representations of the orbitals are shown here. Natural transition orbitals (NTOs) describe transitions more accurately. In PA two seemingly degenerate lone-pair type HOMO levels are split into n+ and n− levels. This behavior is captured in the calculations by Dhanya et al.[54] 9 In addition to the low-pressure work, previous studies have also investigated the photolysis of gas-phase PA under atmospheric pressure (760 Torr). In 1992, Berges and Warneck[3] photolyzed PA (50–100 ppm) at 350 nm and observed a quantum yield of Φ = 0.85, smaller than the low-pressure studies. Surprisingly, the product yields of acetaldehyde (Φ = 0.48) and CO2 (Φ = 1.27) remained the same and additional products like acetic acid, CO and CH4 were also observed.[3] A later analysis using a smog chamber found much lower PA photolysis quantum yield (Φ = 0.43) and new products like formic acid and formaldehyde, in addition to the previously detected acetaldehyde, acetic acid, and CO products.[8] More recently, Vaida and coworkers have investigated the photodissociation under near-atmospheric conditions with a photolysis source whose distributed flux is similar to the solar radiation near the Earth’s surface. These studies, performed under different PA concentrations (0.04 – 0.3 Torr) and different buffer gases (N2, O2, synthetic air), showed that the quantum yield of the observed CO2, CO, CH3CO, HCHO, CH4, CH3OH and HCOOH products depended sensitively on the partial pressure of PA and the pressure and nature of the buffer gas.[9-10] Most of these experiments, however, were carried out in static gas cells where nascent products were subject to prolonged irradiation, secondary reactions, and multiple collisions. Under these conditions, the final products and their yields depended also on the specific experimental arrangement used in each study. Understandably, these results did not always agree with one another on the relative abundances and chemical nature of the final products.[3,8-10] 10 Studies beyond the ground and first excited states of PA at shorter wavelengths, which reach higher electronic states, are extremely limited. As a result, the excited state picture of PA remained largely unmapped, not to mention the lack of studies which explore the interactions between these excited states. Even on S0 and S1, the mechanism of product formation and the role of important intermediates like MHC remain unclear.[2,9- 10,41,44,52,55-56] The experiments reported in this dissertation are aimed at identifying nascent products and unraveling primary dissociation mechanisms on the lowest singlet excited states of PA, S1, S2 and S3, and understanding the effects of collisions and secondary reactions on dissociation pathways and branching ratios. As mentioned earlier, PA has a carbonyl group, a carboxylic group, and a labile H (due to internal H-bonding) in the Tc isomer. Therefore, PA presents a unique test case to understand how the presence of multiple functional moieties in an organic molecule manifest in the nature of its excited states, and consequently in its photochemical pathways. The experiments reported here are conducted in molecular beams and low-pressure flow reactors with nanosecond pulsed laser radiation. Under these experimental conditions, we can identify and characterize primary photodissociation products, especially reactive intermediates. The results described in this dissertation give new fundamental insights into the complex PA photochemistry. We find that the “acid-like” decarboxylation pathway, which is favored on the S0 and S1 states, is no longer dominant at higher excitation energies. In fact, 11 following excitation to the S2 and S3 states, PA also starts to exhibit “keto-like” bond cleavage around the carbonyl bonds, resulting in several open-shell radical fragments. Chapter 3 describes bond fission products of PA arising from fast dissociation on the S2 surface accessed using two-photon excitation via the S1 state. We observed that PA undergoes three-body fragmentation at such high excitation energies and invoked synchronous and sequential fragmentation models to describe the results. We are the first to carry out studies with expansion-cooled PA and obtain a spectrum of the S1 state under these conditions.[53] Chapter 4 focuses on the PA photoproducts following one-photon 193 nm excitation to the S3 state. Studies using both the molecular beam and low-pressure flow reactor setups show that three-body fragmentation processes dominate the dissociation pathways, even for those arising from a decarboxylation first step. We have quantified 15 primary photoproducts and predicted 5 more based on our analysis. Using a simple multivariate regression fit we identify and quantify the first comprehensive set of PA photodissociation pathways. Chapter 5 explores photodissociation dynamics on the S1 state with the goal of characterizing and investigating the participation of the MHC intermediate. Studies using singly deuterated PA reveal additional insights about MHC’s isomerization to the stable acetaldehyde and vinyl alcohol products. We also obtain preliminary evidence of MHC participation in secondary reactions. 12 Methylhydroxycarbene Implications: an elusive intermediate Hydroxycarbenes, the tautomers of aldehydes, are known to be particularly reactive and therefore are difficult to isolate and study. These carbenes have been implicated early on in the synthesis of carbohydrates in space through photocatalysis.[57] Hydroxycarbenes have also generated much interest as high-lying intermediates in the excited-state dynamics of aldehydes.[58-59] The simplest member of the series, hydroxycarbene (HCOH), became the focal point of theoretical attention over 30 years ago as a possible participant in the unimolecular decomposition of formaldehyde.[60-63] More recent research involving the H2CO/HCOH system was driven by experimental results that identified a roaming mechanism in the photodissociation of formaldehyde to CO + H2.[64-66] Likewise, the role of MHC (CH3COH) was explored from the point of view of acetaldehyde photochemistry. Considerable experimental and theoretical interest has been generated towards quantifying the relative magnitudes of the barriers to intramolecular rearrangements connecting acetaldehyde, vinyl alcohol, and MHC.[11,67-72] 13 Figure 1.5 Acetaldehyde PES with stationary points involving its isomerization (black line) and dissociation pathways (blue line). Energies are in kcal/mol, relative to the global minimum acetaldehyde. (Adapted from ref [73]) Many products are possible from unimolecular dissociation of MHC and isomers (see Figure 1.5). In particular, the CH4 + CO and CH3 + HCO channels have been extensively investigated.[74-80] Vasiliou et al. conducted thermal decomposition of CH3CHO at 1700 K and proposed several more product channels from their observation of decomposition products like CH3, CO, H, H2, CH2CO, CH2CHOH, H2O, and C2H2.[81] Although quasiclassical trajectory calculations on the acetaldehyde PES have been instructive in explaining several observed results, including the “roaming mechanism”, the fate of trajectories starting from the high energy MHC isomer has not been investigated.[73,82] As a result, MHC photochemistry, its unimolecular dissociation and the nature of its excited states remained largely unexplored. 14 Overview of experimental observations and challenges More recently, these carbenes have been isolated and spectroscopically characterized in the ultracold environment of an Ar matrix. [11-12] Although, pyrolysis of α-keto carboxylic acids like PA has proven to be a good source of hydroxycarbenes like MHC, it is still impossible to identify them unambiguously and stabilize them as photolysis products of stable molecules.[2,11-14] Previous experiments which have been successful in isolating and characterizing hydroxycarbenes have generated them via pyrolysis of glyoxylic acid (for HC) and PA (for MHC).[11-12] The carbenes resulting from thermal decomposition are expected to have low internal energies and as a result, the findings from these studies cannot be extrapolated to the behavior of hydroxycarbenes in the atmosphere. For example, MHC produced from decarboxylation of PA following excitation to the S1 state using 340–370 nm radiation is expected to have internal energy in excess of 14,000 cm -1 .[56] This scenario, however, presents additional complications in the form of possibilities of further dissociation or isomerization to its stable conformers. This is because barriers to isomerization or further dissociation are exceeded at higher photon energies (see Figure 1.5). Therefore, generating MHC via photolysis would require pristine collision-less environments (for example, in a molecular beam) and sensitive detection schemes which allow for detection within its short lifetime. Resonance enhanced multiphoton ionization (REMPI) detection via Rydberg states is known to be a sensitive and state-selective detection method in molecular beams. Chapter 6 presents electronic structure calculations which help derive these REMPI schemes for HC 15 and MHC. We gained insights into the nature of the excited states, oscillator strengths, and the direction of the transition dipole moment, from the shapes of the orbitals involved in the transitions and their symmetries. Chapter 5 describes the first detection of MHC produced in PA photolysis. The low-pressure flow reactor setup used for experiments described in Chapter 5 is ideal for detecting nascent photoproducts including unstable intermediates like MHC. We characterized the MHC products by its ionization energy, mass, and kinetic time trace. References [1] Horowitz, A.; Meller, R.; Moortgat, G. K., The UV–VIS absorption cross sections of the α- dicarbonyl compounds: pyruvic acid, biacetyl and glyoxal. J. Photochem. Photobiol. A: Chem. 2001, 146 (1-2), 19-27. [2] Yamamoto, S.; Back, R., The photolysis and thermal decomposition of pyruvic acid in the gas phase. Can. J. Chem. 1985, 63 (2), 549-554. [3] Berges, M. G.; Warneck, P., Product quantum yields for the 350 nm photodecomposition of pyruvic acid in air. Ber. Bunsenges. Phys. Chem. 1992, 96 (3), 413-416. [4] Vesley, G. F.; Leermakers, P. A., The photochemistry of α-keto acids and α-keto esters. III. Photolysis of pyruvic acid in the vapor phase. J. Phys. Chem. 1964, 68 (8), 2364- 2366. [5] Arnett, J.; Larson, D.; McGlynn, S., Absorption and emission spectroscopy of pyruvic acids and pyruvate esters. J. Am. Chem. Soc. 1973, 95 (23), 7599-7603. [6] Griffith, E. C.; Carpenter, B. K.; Shoemaker, R. K.; Vaida, V., Photochemistry of aqueous pyruvic acid. Proc. Nat. Acad. Sci. 2013, 110 (29), 11714-11719. [7] Leermakers, P. A.; Vesley, G. F., The photochemistry of α-keto acids and α-keto esters. I. Photolysis of pyruvic acid and benzoylformic acid. J. Am. Chem. Soc. 1963, 85 (23), 3776-3779. [8] Mellouki, A.; Mu, Y., On the atmospheric degradation of pyruvic acid in the gas phase. J. Photochem. Photobiol. A: Chem. 2003, 157 (2-3), 295-300. 16 [9] Reed Harris, A. E.; Cazaunau, M.; Gratien, A.; Pangui, E.; Doussin, J.-F.; Vaida, V., Atmospheric Simulation Chamber Studies of the Gas-Phase Photolysis of Pyruvic Acid. J. Phys. Chem. A 2017, 121 (44), 8348-8358. [10] Reed Harris, A. E.; Doussin, J.-F.; Carpenter, B. K.; Vaida, V., Gas-phase photolysis of pyruvic acid: The effect of pressure on reaction rates and products. J. Phys. Chem. A 2016, 120 (51), 10123-10133. [11] Schreiner, P. R.; Reisenauer, H. P.; Ley, D.; Gerbig, D.; Wu, C.-H.; Allen, W. D., Methylhydroxycarbene: Tunneling control of a chemical reaction. Science 2011, 332 (6035), 1300-1303. [12] Schreiner, P. R.; Reisenauer, H. P.; Pickard IV, F. C.; Simmonett, A. C.; Allen, W. D.; Mátyus, E.; Császár, A. G., Capture of hydroxymethylene and its fast disappearance through tunnelling. Nature 2008, 453 (7197), 906. [13] Leavitt, C. M.; Moradi, C. P.; Stanton, J. F.; Douberly, G. E., Communication: Helium nanodroplet isolation and rovibrational spectroscopy of hydroxymethylene. J. Chem. Phys. 2014, 140 (17), 171102. [14] Back, R.; Yamamoto, S., The gas-phase photochemistry and thermal decomposition of glyoxylic acid. Can. J. Chem. 1985, 63 (2), 542-548. [15] Samanta, B. R.; Sutradhar, S.; Fernando, R.; Krylov, A. I.; Reisler, H., Electronic Structure and Rydberg–Core Interactions in Hydroxycarbene and Methylhydroxycarbene. J. Phys. Chem. A 2018, 122 (30), 6176-6182. [16] Ito, A.; Sillman, S.; Penner, J. E., Effects of additional nonmethane volatile organic compounds, organic nitrates, and direct emissions of oxygenated organic species on global tropospheric chemistry. J. Geophys. Res. Atmos. 2007, 112 (D6). [17] Poisson, N.; Kanakidou, M.; Crutzen, P. J., Impact of non-methane hydrocarbons on tropospheric chemistry and the oxidizing power of the global troposphere: 3- dimensional modelling results. J. Atmos. Chem. 2000, 36 (2), 157-230. [18] Atkinson, R., Atmospheric chemistry of VOCs and NOx. Atmos. Environ. 2000, 34 (12- 14), 2063-2101. [19] Mellouki, A.; Wallington, T.; Chen, J., Atmospheric chemistry of oxygenated volatile organic compounds: impacts on air quality and climate. Chem. Rev. 2015, 115 (10), 3984-4014. [20] Sindelarova, K.; Granier, C.; Bouarar, I.; Guenther, A.; Tilmes, S.; Stavrakou, T.; Muller, J.-F.; Kuhn, U.; Stefani, P.; Knorr, W., Global data set of biogenic VOC emissions calculated by the MEGAN model over the last 30 years. 2014. 17 [21] Goldstein, A. H.; Galbally, I. E., Known and unexplored organic constituents in the earth's atmosphere. Environ. Sci. Technol. 2007, 41 (5), 1514-1521. [22] Guenther, A.; Jiang, X.; Heald, C.; Sakulyanontvittaya, T.; Duhl, T.; Emmons, L.; Wang, X., The Model of Emissions of Gases and Aerosols from Nature version 2.1 (MEGAN2.1): an extended and updated framework for modeling biogenic emissions. 2012. [23] Altieri, K. E.; Carlton, A. G.; Lim, H.-J.; Turpin, B. J.; Seitzinger, S. P., Evidence for oligomer formation in clouds: Reactions of isoprene oxidation products. Environ. Sci. Technol. 2006, 40 (16), 4956-4960. [24] Carlton, A.; Wiedinmyer, C.; Kroll, J., A review of Secondary Organic Aerosol (SOA) formation from isoprene. Atmos. Chem. Phys. 2009, 9, 4987–5005. [25] Carlton, A. G.; Turpin, B. J.; Lim, H. J.; Altieri, K. E.; Seitzinger, S., Link between isoprene and secondary organic aerosol (SOA): Pyruvic acid oxidation yields low volatility organic acids in clouds. Geophys. Res. Lett. 2006, 33 (6). [26] Ervens, B.; Carlton, A. G.; Turpin, B. J.; Altieri, K. E.; Kreidenweis, S. M.; Feingold, G., Secondary organic aerosol yields from cloud-processing of isoprene oxidation products. Geophys. Res. Lett. 2008, 35 (2). [27] Grosjean, D.; Williams, E. L.; Grosjean, E., Atmospheric chemistry of isoprene and of its carbonyl products. Environ. Sci. Technol. 1993, 27 (5), 830-840. [28] Kroll, J. H.; Ng, N. L.; Murphy, S. M.; Flagan, R. C.; Seinfeld, J. H., Secondary organic aerosol formation from isoprene photooxidation. Environ. Sci. Technol. 2006, 40 (6), 1869- 1877. [29] Lim, H.-J.; Carlton, A. G.; Turpin, B. J., Isoprene forms secondary organic aerosol through cloud processing: Model simulations. Environ. Sci. Technol. 2005, 39 (12), 4441-4446. [30] Nguyen, T. B.; Bateman, A. P.; Bones, D. L.; Nizkorodov, S. A.; Laskin, J.; Laskin, A., High- resolution mass spectrometry analysis of secondary organic aerosol generated by ozonolysis of isoprene. Atmos. Environ. 2010, 44 (8), 1032-1042. [31] Andreae, M. O.; Talbot, R. W.; Li, S. M., Atmospheric measurements of pyruvic and formic acid. J. Geophys. Res. Atmos. 1987, 92 (D6), 6635-6641. [32] Talbot, R.; Andreae, M.; Berresheim, H.; Jacob, D. J.; Beecher, K., Sources and sinks of formic, acetic, and pyruvic acids over Central Amazonia: 2. Wet season. J. Geophys. Res. Atmos. 1990, 95 (D10), 16799-16811. 18 [33] Reed Harris, A. E.; Ervens, B.; Shoemaker, R. K.; Kroll, J. A.; Rapf, R. J.; Griffith, E. C.; Monod, A.; Vaida, V., Photochemical kinetics of pyruvic acid in aqueous solution. J. Phys. Chem. A 2014, 118 (37), 8505-8516. [34] Epstein, S.; Nizkorodov, S., A comparison of the chemical sinks of atmospheric organics in the gas and aqueous phase. Atmos. Chem. Phys. 2012, 12 (17), 8205-8222. [35] Boris, A. J.; Desyaterik, Y.; Collett Jr, J. L., How do components of real cloud water affect aqueous pyruvate oxidation? Atmos. Res. 2014, 143, 95-106. [36] Guzman, M.; Colussi, A.; Hoffmann, M., Photoinduced oligomerization of aqueous pyruvic acid. J. Phys. Chem. A 2006, 110 (10), 3619-3626. [37] Leermakers, P.; Vesley, G., Photolysis of pyruvic acid in solution. J. Org. Chem. 1963, 28 (4), 1160-&. [38] Lopalco, A.; Dalwadi, G.; Niu, S.; Schowen, R. L.; Douglas, J.; Stella, V. J., Mechanism of decarboxylation of pyruvic acid in the presence of hydrogen peroxide. J. Pharm. Sci. 2016, 105 (2), 705-713. [39] Saito, K.; Sasaki, G.; Okada, K.; Tanaka, S., Unimolecular decomposition of pyruvic acid: an experimental and theoretical study. J. Phys. Chem. 1994, 98 (14), 3756-3761. [40] Taylor, R., The mechanism of thermal eliminations Part XXIII:[1] The thermal decomposition of pyruvic acid. Int. J. Chem. Kinet. 1987, 19 (8), 709-713. [41] da Silva, G., Decomposition of pyruvic acid on the ground-state potential energy surface. J. Phys. Chem. A 2016, 120 (2), 276-283. [42] Colberg, M. R.; Watkins, R. J.; Krogh, O. D., Vibrationally excited carbon dioxide produced by infrared multiphoton pyrolysis. J. Phys. Chem. 1984, 88 (13), 2817-2821. [43] Perkins, R. J.; Shoemaker, R. K.; Carpenter, B. K.; Vaida, V., Chemical equilibria and kinetics in aqueous solutions of zymonic acid. J. Phys. Chem. A 2016, 120 (51), 10096- 10107. [44] Rosenfeld, R. N.; Weiner, B., Energy disposal in the photofragmentation of pyruvic acid in the gas phase. J. Am. Chem. Soc. 1983, 105 (11), 3485-3488. [45] Reed Harris, A. E.; Pajunoja, A.; Cazaunau, M.; Gratien, A.; Pangui, E.; Monod, A.; Griffith, E. C.; Virtanen, A.; Doussin, J.-F.; Vaida, V., Multiphase photochemistry of pyruvic acid under atmospheric conditions. J. Phys. Chem. A 2017, 121 (18), 3327-3339. [46] Rottman, G. J.; Woods, T. N. In Upper atmosphere research satellite (UARS) solar stellar irradiance comparison experiment (SOLSTICE), Optical Spectroscopic Techniques and 19 Instrumentation for Atmospheric and Space Research, International Society for Optics and Photonics: 1994; pp 317-327. [47] Hunnicutt, S. S.; Waits, L. D.; Guest, J. A., 1 (n,π*)-Photochemistry of acetic acid at 200 nm: further evidence for an exit channel barrier and reaction selectivity. J. Phys. Chem. 1991, 95 (2), 562-570. [48] North, S. W.; Blank, D. A.; Gezelter, J. D.; Longfellow, C. A.; Lee, Y. T., Evidence for stepwise dissociation dynamics in acetone at 248 and 193 nm. J. Chem. Phys. 1995, 102 (11), 4447-4460. [49] Bagchi, A.; Dyakov, Y. A.; Ni, C.-K., Photodissociation and photoionization of 2,5- dihydroxybenzoic acid at 193 and 355 nm. J. Chem. Phys. 2010, 133 (24), 244309. [50] North, S. W.; Marr, A. J.; Furlan, A.; Hall, G. E., Nonintuitive asymmetry in the three-body photodissociation of CH3COCN. J. Phys. Chem. A 1997, 101 (49), 9224-9232. [51] Deshmukh, S.; Hess, W. P., Photodissociation of acetyl chloride: Cl and CH3 quantum yields and energy distributions. J. Chem. Phys. 1994, 100 (9), 6429-6433. [52] Chang, X.-P.; Fang, Q.; Cui, G., Mechanistic photodecarboxylation of pyruvic acid: Excited-state proton transfer and three-state intersection. J. Chem. Phys. 2014, 141 (15), 154311. [53] Sutradhar, S.; Samanta, B. R.; Fernando, R.; Reisler, H., Spectroscopy and Two-Photon Dissociation of Jet-Cooled Pyruvic Acid. J. Phys. Chem. A 2019, 123 (28), 5906-5917. [54] Dhanya, S.; Maity, D. K.; Upadhyaya, H. P.; Kumar, A.; Naik, P. D.; Saini, R. D., Dynamics of OH formation in photodissociation of pyruvic acid at 193 nm. J. Chem. Phys. 2003, 118 (22), 10093-10100. [55] Plath, K. L.; Takahashi, K.; Skodje, R. T.; Vaida, V., Fundamental and overtone vibrational spectra of gas-phase pyruvic acid. J. Phys. Chem. A 2009, 113 (26), 7294-7303. [56] Takahashi, K.; Plath, K. L.; Skodje, R. T.; Vaida, V., Dynamics of vibrational overtone excited pyruvic acid in the gas phase: Line broadening through hydrogen-atom chattering. J. Phys. Chem. A 2008, 112 (32), 7321-7331. [57] Baly, E. C. C.; Heilbron, I. M.; Barker, W. F., CX.—Photocatalysis. Part I. The synthesis of formaldehyde and carbohydrates from carbon dioxide and water. J. Chem. Soc., Trans. 1921, 119, 1025-1035. [58] Kemper, M.; Van Dijk, J.; Buck, H., Ab initio calculation on the photochemistry of formaldehyde. The search for a hydroxycarbene intermediate. J. Am. Chem. Soc. 1978, 100 (25), 7841-7846. 20 [59] Shepler, B. C.; Epifanovsky, E.; Zhang, P.; Bowman, J. M.; Krylov, A. I.; Morokuma, K., Photodissociation dynamics of formaldehyde initiated at the T1/S0 minimum energy crossing configurations. J. Phys. Chem. A 2008, 112 (51), 13267-13270. [60] Osamura, Y.; Goddard, J. D.; Schaefer III, H. F.; Kim, K. S., Near degenerate rearrangement between the radical cations of formaldehyde and hydroxymethylene. J. Chem. Phys. 1981, 74 (1), 617-621. [61] Goddard, J. D.; Yamaguchi, Y.; Schaefer III, H. F., Features of the H2CO potential energy hypersurface pertinent to formaldehyde photodissociation. J. Chem. Phys. 1981, 75 (7), 3459-3465. [62] Moore, C. B.; Weisshaar, J. C., Formaldehyde photochemistry. Annu. Rev. Phys. Chem. 1983, 34 (1), 525-555. [63] Lucchese, R. R.; Schaefer III, H. F., Metal-carbene complexes and the possible role of hydroxycarbene in formaldehyde laser photochemistry. J. Am. Chem. Soc. 1978, 100 (1), 298-299. [64] Lahankar, S. A.; Chambreau, S. D.; Townsend, D.; Suits, F.; Farnum, J.; Zhang, X.; Bowman, J. M.; Suits, A. G., The roaming atom pathway in formaldehyde decomposition. J. Chem. Phys. 2006, 125 (4), 044303. [65] Suits, A. G., Roaming atoms and radicals: a new mechanism in molecular dissociation. Acc. Chem. Res. 2008, 41 (7), 873-881. [66] van Zee, R. D.; Foltz, M. F.; Moore, C. B., Evidence for a second molecular channel in the fragmentation of formaldehyde. J. Chem. Phys. 1993, 99 (3), 1664-1673. [67] Bouma, W. J.; Poppinger, D.; Radom, L., Vinyl alcohol. A stable molecule. J. Am. Chem. Soc. 1977, 99 (19), 6443-6444. [68] Bouma, W. J.; Vincent, M. A.; Radom, L., Ab initio molecular orbital studies of sigmatropic rearrangements. Int. J. Quantum Chem. 1978, 14 (6), 767-777. [69] Rodwell, W. R.; Bouma, W. J.; Radom, L., Ab initio studies of 1,3-sigmatropic rearrangements: Effect of basis set and electron correlation. Int. J. Quantum Chem. 1980, 18 (1), 107-116. [70] Smith, B. J.; Nguyen, M. T.; Bouma, W. J.; Radom, L., Unimolecular rearrangements connecting hydroxyethylidene (CH3-C-OH), acetaldehyde (CH3-CH=O), and vinyl alcohol (CH2=CH-OH). J. Am. Chem. Soc. 1991, 113 (17), 6452-6458. 21 [71] Räsänen, M.; Raaska, T.; Kunttu, H.; Murto, J., Ab initio studies on carbenes; singlet and triplet conformers and vibrational spectra of hydroxy-, dihydroxy-and methylhydroxy- carbene. J. Mol. Struc.: THEOCHEM. 1990, 208 (1-2), 79-90. [72] Yadav, J. S.; Goddard, J. D., Methoxycarbene and methylhydroxycarbene: energies, structures, vibrational frequencies, and unimolecular reactivities. J. Chem. Phys. 1986, 85 (7), 3975-3984. [73] Han, Y.-C.; Shepler, B. C.; Bowman, J. M., Quasiclassical trajectory calculations of the dissociation dynamics of CH3CHO at high energy yield many products. J. Phys. Chem. Lett. 2011, 2 (14), 1715-1719. [74] Terentis, A. C.; Stone, M.; Kable, S. H., Dynamics of acetaldehyde dissociation at 308 nm: rotational (N, Ka) and translational distributions of the HCO photoproduct. J. Phys. Chem. 1994, 98 (42), 10802-10808. [75] Lee, S. H.; Chen, I. C., Photofragments CH3(X� 2 A2’’) + HCO(X� 2 A’) from acetaldehyde: Distributions of rotational states and preferential population of K doublets of HCO. J. Chem. Phys. 1996, 105 (11), 4597-4604. [76] Gherman, B. F.; Friesner, R. A.; Wong, T.-H.; Min, Z.; Bersohn, R., Photodissociation of acetaldehyde: the CH4 + CO channel. J. Chem. Phys. 2001, 114 (14), 6128-6133. [77] Houston, P.; Kable, S., Photodissociation of acetaldehyde as a second example of the roaming mechanism. Proc. Nat. Acad. Sci. 2006, 103 (44), 16079-16082. [78] Rubio-Lago, L.; Amaral, G.; Arregui, A.; Izquierdo, J.; Wang, F.; Zaouris, D.; Kitsopoulos, T.; Banares, L., Slice imaging of the photodissociation of acetaldehyde at 248 nm. Evidence of a roaming mechanism. Phys. Chem. Chem. Phys. 2007, 9 (46), 6123-6127. [79] Heazlewood, B. R.; Rowling, S. J.; Maccarone, A. T.; Jordan, M. J.; Kable, S. H., Photochemical formation of HCO and CH3 on the ground S0( 1 A’) state of CH3CHO. J. Chem. Phys. 2009, 130 (5), 054310. [80] Heazlewood, B. R.; Maccarone, A. T.; Andrews, D. U.; Osborn, D. L.; Harding, L. B.; Klippenstein, S. J.; Jordan, M. J.; Kable, S. H., Near-threshold H/D exchange in CD3CHO photodissociation. Nat. Chem. 2011, 3 (6), 443. [81] Vasiliou, A.; Piech, K. M.; Zhang, X.; Nimlos, M. R.; Ahmed, M.; Golan, A.; Kostko, O.; Osborn, D. L.; Daily, J. W.; Stanton, J. F., The products of the thermal decomposition of CH3CHO. J. Chem. Phys. 2011, 135 (1), 014306. [82] Shepler, B. C.; Braams, B. J.; Bowman, J. M., “Roaming” dynamics in CH3CHO photodissociation revealed on a global potential energy surface. J. Phys. Chem. A 2008, 112 (39), 9344-9351. 22 Experimental techniques Introduction Two different experimental setups were used for the present study, each with its own advantages and limitations. By using both these experimental setups in tandem, we can circumvent several experimental challenges highlighted in the previous section. In doing so, we get several complementary measurements which enable key insights into complicated dissociation mechanisms of molecular systems. Figure 2.1 Photodissociation dynamics in a molecular beam using VMI and chemical kinetic studies using the MPIMS have been used together to explore the complicated chemical systems described in this dissertation. 23 The first setup employs the velocity map imaging (VMI) technique. VMI is an excellent technique for quantitative assessment of dissociation energetics. It is used in experiments involving investigation of classical momentum distributions of products (neutrals or ions) from disintegration (dissociation, ionization, detachment) processes that are common in scattering of atomic and molecular systems. Velocity distributions measured using VMI can provide mechanistic insights about the dissociation process, including speed and angular distributions of the products, product branching ratios, dissociation geometries and timescales.[1-5] Together with cold molecular beams and narrow-bandwidth lasers, this method also allows determination of dissociation energies with significant accuracy. VMI is routinely employed to study the nature of excited states, including quantum effects like resonances, conical intersections and tunneling.[3,5] Although the imaging aspect of VMI, and as a result the entire method, is inherently applicable to electrically charged particles only, by using suitable ionization methods (under the assumption that the ion retains the initial momentum of the neutral particle) it becomes suitable for neutral products as well. Highly selective ionization methods, such as resonance-enhanced multiphoton ionization (REMPI), allow for quantum-state selective detection of the products. The experimental setup and the principle of data analysis are described in the following subsection. The second setup used for data acquisition was the multiplexed photoionization mass spectrometer (MPIMS) built by Dr. David Osborn and coworkers of the Combustion Research Facility at Sandia National Labs, Livermore, CA. This instrument has multiple sections and can function with any ionization source. For the purposes of this dissertation, this instrument was employed as a roll-up end-station at beamline 9.0.2.3 (Chemical Physics) of the 24 Advanced Light Source (ALS) at the Lawrence Berkeley National Lab. The ALS is a third- generation synchrotron that can produce widely tunable VUV radiation in the 7.2–25 eV range with high brightness (10 21 photons cm −2 sr −1 ) and intermediate resolution (E/ΔE ~ 1000). The narrow bandwidth and high brightness of this synchrotron source combined with the high sensitivity and the broad tunability makes the MPIMS highly selective over a wide range of photon energies. This work utilized tunable VUV radiation in the 8–19 eV range from the synchrotron for ionization of species introduced or produced in the MPIMS instrument. The MPIMS instrument was primarily designed to study chemical kinetics of neutral reactants, intermediates and products.[6] Contrary to the quantum-state specificity of REMPI techniques employed in VMI, the MPIMS is universally applicable to all atoms and molecules in the reactor tube irrespective of the quantum state they exist in. Historically, single-photon photoionization is difficult to achieve due to the complexity of generating vacuum ultraviolet photons in the 6.5–16 eV range (where most atoms and molecules ionize) that are simultaneously intense, widely tunable, and spectrally narrow. The MPIMS allows for threshold ionization of different structural isomers by virtue of their different photoionization curves while minimizing ion-fragmentation. With this in mind, we use the MPIMS as an identification tool for photodissociated products and to monitor their kinetics over several milliseconds. The VMI technique is used simultaneously to spectroscopically identify key products and monitor their internal energy distributions. This allows us to track the unimolecular photodissociation dynamics on different excited states of the molecule. Compared to the slow instrument response time of 25 the MPIMS (480 μs), the VMI technique gives us a “faster” snapshot of the reaction timescale (several picoseconds) and insights into molecular motions like vibration and rotations. Molecular beam and VMI instrument The initial assembly and subsequent modifications of the VMI setup have been described in previous dissertations[7-10] and only the working principle is highlighted in this dissertation in the context of the present study. The main parts of the setup are the pulsed nozzle, the interaction region, the time-of-flight (TOF) mass spectrometer and the detector. Each of these parts are further explained in subsequent subsections and a simple schematic is shown in Figure 2.2. Figure 2.2 Schematic of the VMI apparatus is shown along with its main components (bold face) and their intended purpose (regular font). The source and detection chambers are maintained under high vacuum conditions. The PA/He partial pressures used in this apparatus are also included. 26 Molecular beam through a pulsed nozzle The sample is introduced into a vacuum chamber using a piezo-electrically pulsed nozzle. The vacuum chamber is divided by a molecular beam skimmer into two regions: source and detection. The base pressure of the source chamber is 2×10 -7 Torr while that of the detection region is 2×10 -8 Torr. Both regions are pumped using turbomolecular pumps which are connected to oil-based roughing pumps. The efficiency of the vacuum systems can be compromised when using volatile organic compounds (for example PA). PA is found to be notoriously sticky and adsorbs on the chamber’s inner surface, causing a buildup of background signal over time. The chamber was periodically “baked” with differential heating (80–100 °C, increasing from source to detection) to minimize this issue. In order to create a seeded molecular beam, the sample is mixed with a carrier gas (usually He or Ar) and introduced into the source chamber. Gas is flown into a piezo housing (consisting of a piezoelectric actuator disk attached to a Teflon plunger) via a ¼’’ Teflon tube. The plunger has a Kalrez O-ring (which resists corrosive gases) at its tip which acts as a valve, sealing the nozzle orifice from vacuum. When a 100–300 μs wide negative 200–250 V DC voltage pulse is applied, the piezoelectric actuator delivers a controlled amount of reactant gases into vacuum via a quartz tube (1.5 cm long, 2 mm o.d., 1 mm i.d.) attached to the nozzle plate. The pressure difference between the sample pressure (1400 Torr) and the source chamber (10 -7 Torr) causes the gases to adiabatically cool down to ~10 K. Rovibrational cooling to 10–15 K, achieved in this setup, has been determined from the rotational distribution of CO (v=0) when using He as the carrier gas. The nozzle directs the effusing 27 gases towards the center of a circular detector ( 2’’ dia.). The gas mixture, due to collisions in the nozzle, results in a molecular beam of low divergence and narrow velocity distribution. Using He, supersonic molecular speeds of ~1600 m/s are obtained. This is represented by step (a) in Figure 2.3. The molecular beam is skimmed (Beam Dynamics, 1.51 mm dia.) 2–3 cm downstream from the nozzle to further narrow its divergence. Photolysis and ionization The section after the skimmer is referred to as the detection region. This section houses the TOF mass spectrometer assembly, Einzel lens assembly and the detector. The instrument has two diametrically opposite ports sealed with quartz windows to allow laser irradiation into the chamber. Pump and probe lasers can be introduced from either side of the chamber (crossed pump and probe beam setup is commonly used). The laser radiation is focused on the molecular beam to photoexcite and photoionize the molecules as seen in steps (b)–(d) of Figure 2.3. During unimolecular dissociation processes, the first photon often accesses electronically excited states of the molecule (here, CO2) from where it sometimes dissociates into fragments. a Additional photon(s), resonant with an excited state of the molecule, are used to state-selectively ionize a fragment of choice (henceforth referred to as a REMPI scheme). The REMPI schemes employed for each species detected in this dissertation are mentioned in later chapters under the subheading “experimental details". In the example shown in step (d) of Figure 2.3, O( 3 P) is detected using ~225 nm via (2+1) a Unlike condensed phase studies, collisional relaxation is not feasible in the collisionless gas phase environment of this study. As a result, the excitation energy is utilized in bond dissociation. 28 REMPI.[11] CO2 molecules can be oriented in different directions at the moment of dissociation, leading to a distribution of positively charged ions which fly radially outward because of their acquired kinetic energy (KE) from the dissociation process, as depicted in step (e) of Figure 2.3. The competition between the rotational time period and the molecule’s dissociation timescale determines the specific shape of the spatial distribution. In the examples shown in Figures 2.3 and 2.4, we assume a spherical distribution for the sake of simplicity. Figure 2.3 A schematic representation of a VMI steps for an example molecule (here, CO2). (a) Adiabatic expansion of the reactant molecules in a buffer gas (He or Ar) through a pulsed nozzle creates a moleuclar beam. (b) The molecular beam is irradiated using a photolysis laser which enables (c) photoexcitation and (d) subsequent dissociation of the molecule. Additional laser(s) can be used to resonantly ionize a fragment of interest (for example, O( 3 P) using 225 nm) for state-selective detection. (e) The resulting spatial distribution of ions contains information about the dissociation process (KE, anisotropy, etc.). Time-of-flight (TOF) mass spectrometer The TOF arrangement built at USC utilizes the Wiley-McLaren TOF setup, which requires an ionization, ion-acceleration, and a field-free drift region.[12] These regions are defined with respect to the positions of three electrostatic plates (repeller, extractor and 29 ground) and a multichannel plate (MCP) detector. The ionization point (focal point of the ionization laser) which completely or partially overlaps the photolysis point, is positioned in the middle of repeller and extractor plates. The ion optics of our instrument is designed as the best compromise between space focusing, required for high resolution TOF spectroscopy, and velocity focusing required for time-sliced velocity map imaging.[7] The ion plates define an electric field distribution that allows ions with the same initial velocity direction to have the same flight trajectory to the detector. The current setup employs 14 ion plates, evenly spaced 1 cm apart.[7] Plate 1 is reserved for the repeller. The extractor and ground electrostatic plates can be varied to change the distance between the plates and as a result, the overall resolution of the instrument. This feature allows a system specific TOF setting to tune the maximum KE range and TOF mass range. An additional Einzel lens system, mounted in the field-free drift region, provides more tunability on the maximum KE range that can fit into the detector. The general design philosophy and the range of electrostatic parameters chosen for optimal operation of the TOF setup in conjunction with ion imaging has been summarized by Ryazanov.[7] The maximum detectable KE, KErange, for a particular voltage setting is optimized using SIMION. For the VMI experiments reported in this dissertation, specific values are set for each fragment ion for the repeller to extractor distance (L0), extractor to ground distance (L0), repeller voltage (V0), extractor voltage (V1) and Einzel lens voltage (VL). For H atom detection, L0 = 8.5 cm, L1 = 3 cm, V0 = 1000 V, V1 = 475 V, VL = −2600 V, KErange = 3.26 eV. For all the other fragments, namely, CH3CO, CO, CH3 and HOCO, we used L0 = 5 cm, L1 = 5 cm, V0 = 3000 V, V1 = 2115 V, VL = 0 V, KErange = 1.68 eV. 30 Sliced velocity map imaging (SVMI) As mentioned earlier the VMI technique is used in experiments involving dissociation of molecules. Ion imaging was first introduced by Chandler and Houston in 1987.[1] Later, Eppink and Parker[13] demonstrated an improved version of the technique by introducing velocity mapping. The VMI technique is widely used to investigate kinetic energy distribution and infer internal energy distributions of products formed in photodissociation processes. The basic principle of VMI is illustrated in Figure 2.4(a) for photodissociation of CO2. For the sake of simplicity, it is assumed that upon excitation with a photon that imparts energy 𝐸𝐸 , CO2 dissociates into CO and O( 3 P). In the illustration, the CO fragment is formed in two vibrational states — CO(v=0) and CO(v=1). The total KE of the fragments, 𝐸𝐸 k in , can be derived using conservation of energy as 𝐸𝐸 = Δ 𝐻𝐻 rx n + 𝐸𝐸 avl , (2.1) where Δ 𝐻𝐻 rx n is the bond dissociation energy and 𝐸𝐸 avl is the maximum available energy. As a result of the dissociation, this available energy is partitioned among all degrees of freedom of the fragments. Based on the schematic shown in Figure 2.4(a), 𝐸𝐸 av l = 𝐸𝐸 k in C O + 𝐸𝐸 k in O + 𝐸𝐸 in t C O + 𝐸𝐸 in t O . (2.2) Here 𝐸𝐸 k in and 𝐸𝐸 in t represent the kinetic and internal energies of the fragments, respectively. Equation 2.2 can be further simplified for the example case where the rovibrational modes of CO (with energy 𝐸𝐸 rov ib C O ) are excited, and the atomic O fragment does not possess internal modes. As a result, 31 𝐸𝐸 av l = 𝐸𝐸 k in C O + 𝐸𝐸 k in O + 𝐸𝐸 rov ib C O . (2.3) Furthermore, due to conservation of linear momentum, the KEs of the two fragments in a two-body fragmentation process are related by the following equation. 𝐸𝐸 k in C O ∙ 𝑚𝑚 C O = 𝐸𝐸 k in O ∙ 𝑚𝑚 O (2.4) The primary variable of interest in a photodissociation experiment is often the total kinetic energy release (KER) in the combined system of fragments. In the center of mass (c.m.) reference frame, the total KER is simply equal to the sum of fragment KEs. Equation 2.1 as a result, can be simplified to obtain the c.m. KER distribution from the KE of O products. 𝐸𝐸 = Δ 𝐻𝐻 rx n + 𝐸𝐸 k in O ∙ 𝑚𝑚 C O 2 𝑚𝑚 C O + 𝐸𝐸 rov ib C O . (2.5) As a consequence of equation 2.5, the KE of the detected O fragments will reflect the rovibrational excitation of its cofragment — CO. This inter-dependency of the internal energies of the two fragments is called pair-correlation. A maximum c.m. KE is obtained when CO is detected in its ground rovibrational state. More rovibrational excitation in CO corresponds to a lower value of c.m. KE. The radial size of the ion spheroid at any given time is proportional to the velocity of its constituent ions which is dependent on its KE determined by equation 2.5 above. This is demonstrated in the right panel of Figure 2.4(a) by the two spheres in red and blue representing the O fragment KE’s associated with CO (v=0) and CO (v=1), respectively. The 32 ion spheroid is projected on an MCP detector coupled to a phosphor screen. The ion hits are recorded using a camera to reveal a 2D projection of the 3D ion spheroid, as seen in Figure 2.4(b). In a conventional VMI, reconstructing the 3D velocity distribution from a 2D image can be mathematically challenging.[14-17] Figure 2.4 (a) The basic principle of VMI is depicted using an exampe molecule (CO2). The corresponding 3D ion spheroid distribution and the desired final energy distribution is also shown. (b) The projection of the ion spheroid on the detector in a conventional VMI setup with a DC voltage supplied to the MCP plates. (c) Only the central 2D slice of the ion spheroid is projected by pulse gating the detector voltage, minimizing the need for image reconstruction. To circumvent the need for image reconstruction and to directly measure just the central cross-section of the ion spatial distribution, time-sliced velocity map imaging (SVMI) methods were developed in the early 2000s.[2,4] The obtained sliced image contains information on the ion intensity distribution over a 2D pixel space. In this technique, a pulsed voltage gate is applied to the detector in contrast to a conventional VMI which supplies a constant (or a broad pulse-width) DC voltage to the MCP. The voltage gate pulse is 33 synchronized to the ion’s TOF to obtain a thin central time slice of the ion spheroid, retaining the radial velocity distribution of the ions, as depicted in Figure 2.4(c). This is converted to a radial distribution of intensities by integrating over all angles for a fixed radius giving 𝐼𝐼 ( 𝑅𝑅 ) = � 𝐼𝐼 ( 𝑅𝑅 , 𝜃𝜃 ) 𝑅𝑅 2 |sin 𝜃𝜃 | d𝜃𝜃 2 𝜋𝜋 0 , (2.6) which can be converted to a speed distribution because speed is proportional to radius. The proportionality constant depends on instrumental parameters like the length of the drift tube and total accelerating voltages. More simply, 𝑅𝑅 = 𝛾𝛾 ∙ 𝑣𝑣 𝑟𝑟 or 𝐼𝐼 (𝑅𝑅 ) ~ 𝐼𝐼 (𝑣𝑣 ) (2.7) where 𝛾𝛾 , proportional to √ 𝑚𝑚 , is the constant TOF of an ion with mass 𝑚𝑚 . 𝛾𝛾 can be calibrated experimentally by using a molecule with known dissociation energy and state distribution of the observed fragment. The value of 𝛾𝛾 has also been simulated for the current SVMI apparatus by Ryazanov.[7] The distribution of KEs, 𝐼𝐼 (𝐸𝐸 ), of the detected fragment can be obtained from its speed distribution using 𝐸𝐸 = 1 2 𝑚𝑚 𝑣𝑣 2 or d𝐸𝐸 = 𝑚𝑚 𝑣𝑣 𝑚𝑚𝑣𝑣 . (2.8) Because the total intensity is conserved, 𝐼𝐼 (𝑣𝑣 )𝑚𝑚𝑣𝑣 = 𝐼𝐼 (𝐸𝐸 )𝑚𝑚𝐸𝐸 . Therefore, 𝐼𝐼 ( 𝐸𝐸 ) = 1 𝑣𝑣 𝐼𝐼 ( 𝑣𝑣 ) ≈ 1 𝑅𝑅 � 𝐼𝐼 ( 𝑅𝑅 , 𝜃𝜃 ) 𝑅𝑅 2 |sin 𝜃𝜃 | d𝜃𝜃 2 𝜋𝜋 0 (2.9) 34 The kinetic energy resolution of the distribution depends on the pulse-width of the voltage gate. A short pulse creates a thin slice, thereby eliminating overlap between concentric spheroids. Although a fast gate with sub-nanosecond rise and fall times might seem like an ideal choice, it would also sacrifice most of the signal, thereby requiring prolonged acquisition times to achieve good signal-to-noise ratio. Typical MCP operation requires up to 2 kV of DC voltage for optimal performance. Driving such high voltages over a fast circuit also places limitations on its components. As a result, most commercially available pulse generators, which can be coupled to high voltage supplies, work in the 30– 300 ns range. In order to simultaneously maintain good signal-to-noise ratio (SNR) and a good energy resolution of the image, a slice thickness of roughly a tenth of the ion spheroid’s axial size is recommended. While a conventional SVMI setup with a commercial pulse generator is convenient for most ions, it is not suitable for the light H atoms whose TOF stretch can be as low as 20 ns. To eliminate this disadvantage, a special technique was devised by Ryazanov for detecting H atom fragments, which utilizes a custom-built 5 ns high voltage pulse generator for fast gating of the MCP along with a combination of optimized lens voltages which increase the TOF stretch of H fragment ions to ~100 ns.[7,18-19] By changing these lens voltages, heavier atom fragments could also be imaged with the same 5 ns pulse generator with high overall resolution without sacrificing SNR. 35 Third-harmonic vacuum ultraviolet (VUV) generation using a gas cell VUV radiation of wavelengths around 120 nm, used in this dissertation, is generated using inert gas mixtures in a gas cell where the incident UV or near-visible pump radiation is frequency tripled in a process called third harmonic generation (THG). The principle of THG of VUV radiation using inert gas mixture relies on phase matching between the fundamental and its generated third harmonic.[20-21] Inert gases show relatively strong nonlinear response under strong electric fields due to the existence of metastable electronic states. Because of their non-linear refractive indices, there is a relative phase mismatch between the incident fundamental and the generated third harmonic frequencies. As a result, the phase matching condition requires a mixture of both positive and negative dispersion mediums. The relative phase shift is proportional to 𝑁𝑁 2 𝑃𝑃 3 where N is the number density of Ar and Kr gas molecules (or partial pressures) and P is the intensity of the incident power. Mahon et al. observed several small wavelength regions where VUV could be generated via Ar/Kr gas mixtures, resulting in narrow regions of tunability.[20] Although the intensity of the third harmonic increases with an increase in total pressure in the gas cell or at higher incident powers, total pressure above 900 Torr and incident UV energies above 10 mJ/pulse can cause optical breakdown of the gas mixture in the current setup. A schematic of the gas cell and its wavelength tunability under different pressure ratios has been documented by Sutradhar[22] and only the experimental conditions used in this study are discussed in the next paragraph. VUV wavelengths as low as 114 nm can be generated using the current setup. 36 VUV radiation at 121.6 nm for H detection via its Lyman-α transition is generated by focusing ~365 nm laser radiation inside the gas cell. The produced 121.6 nm is focused inside the source chamber along with the residual incident 365 nm using MgF2 lens (75 mm f.l.). A mixture of 200 Torr Kr and 590 Torr Ar is used to generate 121.6 nm for H detection. VUV radiation at ~118 nm for OH detection using the OH A 2 Π state[23] was generated by focusing 355 nm output of a Nd:YAG laser in the gas cell filler with 30 Torr Xe and 270 Torr Ar. A total pressure of 300 – 350 Torr was found to be most effective for VUV production in this range. The 118 nm intensity decreases drastically at total pressures below 250 Torr and almost disappears at pressures above 600 Torr. Low pressure flow reactor and MPIMS The MPIMS was built primarily for the study of chemical kinetics in combustion and flame chemistry. Based on the original photoionization mass-spectrometry technique developed by Gutman et al.,[24] the MPIMS was modified to include multiple-mass detection and tunable-synchrotron photoionization. Recent approaches to study reaction kinetics and product branching with multiplexed b mass spectrometry and synchrotron photoionization have considerable advantage over single-mass detection.[25-26] The instrument includes a flow tube reactor where photolysis or combustion products are generated, and is designed with the goal of resolving isomer specific product branching ratios in gas phase reactions. The reaction is initiated by a pulsed laser directed along the length of the tube. The flow tube b Data is represented as a function of three dimensions: mass, photoionization energy, and the physical dimension of the experiment (e.g. distance from a flame or time after photolysis) 37 reactor enables continuous side-sampling of molecular species using tunable synchrotron radiation generated at the ALS for photoionization. The ions then pass through an orthogonal axis time-of-flight (OA-TOF) tube for multi-mass detection using a time- and position- sensitive detector capable of single ion counting. The flow tube reactor allows studying chemical systems under varying conditions of pressure, concentration and temperature, thereby providing kinetic parameters like rate coefficients. Additionally, different structural isomers and short-lived reaction intermediates can be identified based on their photoionization spectra, providing more insights about reaction mechanisms.[27-28] The 3D data set from the MPIMS provides new and unique insights because of the instrument’s high sensitivity, mass resolution, and selectivity. A simplified schematic of the instrument is shown in Figure 2.5 and a more detailed description of its parts and features is presented in subsequent sections. 38 Figure 2.5 Schematic of the MPIMS apparatus used at the ALS. The main components of the apparatus are labeled. He bubbled through a glass bulb containing PA is flown into the flow- tube reactor. After photolysis, the molecules are photoionzied using synchrotron radiation. The attached OA-TOF arrangement provides enhanced mass resolution. Flow tube reactor Sample gas mixtures, diluent gas (He) and calibration gases are flown into the reactor, metered by individually calibrated mass flow controllers. The pressure in the reactor tube (62 cm long, 1.27 cm o.d., 1.05 cm i.d.) is controlled using a throttle valve connected to a Roots pump. Gases are sampled through a 650 μm pinhole into the source chamber, forming a near-effusive molecular beam. The molecular beam is then skimmed using a 1.5 mm diameter skimmer approximately 0.3 cm downstream from the pinhole before entering the ionization chamber. The slow-flow conditions in the flow tube ensure that the initial distribution of the photoproducts becomes homogenous faster than the kinetic timescales of 39 the reaction, resulting in concentration changes that are independent of the radial and axial positions from which a sample is extracted. The inner surface of the tube is coated with polychlorotrifluroethylene 2300 wax (Halocarbon Products Corporation) to suppress first- order radical loss on the reactor walls. Photoproducts consisting of neutrals and free radicals are generated by photodissociation of precursor molecules with an unfocused 193 or 351 nm excimer laser pulse (10 Hz, 20 ns pulsewidth, 10–60 mJ cm −2 fluence) propagating collinearly down the reactor tube. At a laser repetition rate of 10 Hz, gas flow speed of ~1000 cm/s provides a fresh sample of gas for each laser firing. Optimal operating conditions to ensure low-density of molecules to maintain homogeneity are observed over the pressure range of 1–10 Torr. The ionization and TOF sections are each pumped using turbomolecular pumps, which are backed by oil-free scroll pumps. Photoionization mass spectrometer Continuous probing of the photoinitiated reaction products requires the use of continuous ionization sources. The ALS synchrotron radiation at the Chemical Dynamics Beamline (9.0.2.3) provides a continuous and rapidly tunable ionization source. VUV radiation (7 – 25 eV) from the undulator passes through a gas cell filled with an inert gas to filter out harmonics of the undulator radiation with energies above their IEs. This radiation is dispersed using a monochromator to achieve a narrower spectral bandwidth (10 – 50 meV). The radiation then goes through an exit slit, controlled using a screw caliper, and intersects the molecular beam perpendicularly, roughly 2.2 cm downstream from the skimmer inside the ionization region of an OA-TOF mass spectrometer. The MPIMS used in 40 conjunction with the ALS radiation enables threshold ionization of fragments along with isomer-specific species identification. Due to its high sensitivity, selectivity, and mass resolution, the MPIMS provides a powerful tool for universal detection and monitoring of fragment kinetics. The VUV photon energy is calibrated using the known 8s←5p atomic absorption in Xe. The OA-TOF continuously probes reactants and products simultaneously over a wide range of masses. In this TOF setup, the ions formed in the ionization region enter the orthogonal accelerator (OA) at nearly right angles to the TOF direction. At regular time intervals (typically 20 μs or 50 kHz operation), ions are orthogonally accelerated using a gate pulse which allows ions into an acceleration region and a drift region. Ion arrival events at the detector are timed relative to the gating event by a transient digitizer. This method of gating ions into a TOFMS partially eliminates peak broadening effects from temporal spread in the ion acceleration time.[29] The position and arrival time with respect to the photodissociation laser are recorded for each ion and laser pulse (a data cycle). The raw data consist of an (x, y, t) information for each detected ion. These raw data are binned in a variety of ways to extract desired experimental observables as described in the next section. Multiplexed data acquisition Following photolysis laser firing, the chemical composition of the gas mixture in the reaction tube evolves in time. The raw data for each individual ion is described by its position on the detector (proportional to its m/z ratio), arrival time, and the VUV photon energy at which it was created. This cycle is repeated for each laser firing and the photon energy is 41 incremented typically in 25 meV steps after roughly 500 laser firings at each photon energy. The result is a 3D data cube illustrated in the top panel of Figure 2.6. For ease of visualization, the 3D data is sliced into various 2D images which aid data analysis. The 2D data can be further sliced to obtain signal intensities as a function of reaction time or photoionization energy for a particular mass. The 1D plots are more suitable for quantitative analysis and species identification. 2D time-resolved mass spectra at a single photon energy can also be directly obtained from the MPIMS by signal-averaging at a single photon energy rather than scanning over a range. This mode is utilized in this work for quantifying free-radical intermediates. 42 Figure 2.6 Three-dimensional dataset measured from the photoinitiated reactions in the flow tube (adapted from ref. [6]) The position on the detector linearly correlates with an ion’s TOF and is related to the m/z ratio of the ion. Integrated cross-sectional slices through the data can be taken to generate 2D plots which show the TOF spectrum as a function of reaction time (left, middle) or as a function of the photoionization energy (right, middle). From these 2D plots, integrated photoionization efficiency curves can be generated as functions of reaction time (left, bottom) or photoionization energy (right, bottom) for a given m/z ratio. m/z m/z Kinetic time (ms) Kinetic time (ms) Photon energy (eV) Photon energy (eV) 43 General workflow and “good practices” of data analysis for the MPIMS data set are described below. The raw data is analyzed using the IGOR Pro (WaveMetrics Inc.) numerical analysis and computation software. An IGOR procedure file with prefix “ALS_kin_tools_TOF_”, written by scientists at the Combustion Research Facility of Sandia National Laboratories, and containing a compendium of data analysis codes, is used to bin data into the above mentioned 2D and 1D plots. The m/z axis is obtained from the primary TOF axis, T, using the equation 𝑇𝑇 = 𝛼𝛼 � 𝑚𝑚 𝑧𝑧 � 1 2 + 𝛽𝛽 , (2.10) where 𝛼𝛼 and 𝛽𝛽 are scalars calibrated using a gas mixture of ethene, propene, 1-butene and Xe, whose masses are precisely known. The same mixture is used to calculate the photon energy calibration offset (deviation from the correct photon energy value) from a xenon scan over the (8s) Rydberg resonance, which has an absolute center value of 12.5752 eV. The offset value can change on a daily basis when the beamline mirrors are adjusted prior to each single-day beamtime shift operation, and therefore, it is recommended that this value be calculated at the beginning of each shift. In a single energy time-resolved mass spectrum data set, the data set is loaded as a XT plot (ion intensity vs. TOF (X) and kinetic time (T)). This function also creates a mass spectrum integrated over all kinetic time. For single energy (E) data, it is meaningful to subtract the average background signal measured before the laser fires. After background subtraction, negative intensity peaks in the mass spectrum represent species destroyed by 44 the laser radiation; positive intensity peaks are species created due to the laser’s firing. The TOF axis in both 1D (a mass spectrum) and 2D data (either XT or EX plots) can be scaled to a m/z axis, using equation 2.11. From a XT plot, a vertical slice (VS) can be used to extract the time trace for a specific mass, or a horizontal slice (HS) to extract the mass spectrum at a specific kinetic time. The PI scans involve a rather large 3D dataset that consists of ion intensity as a function of X, T and E. Because it is hard to visualize this 3D dataset directly, one or more 2D slices through the data are obtained to extract EX for a particular kinetic time, ET for a particular mass, or XT integrated over a small (or entire) photon energy range. Pre- photolysis background can be subtracted from the 3D dataset prior to 2D slicing. However, background subtraction on the individual 2D slices accomplishes the same task and uses less computational resources. EX and ET scans are normalized to the VUV photon flux, measured using a standardized photodiode. Appropriate vertical or horizontal slices are then extracted to create the desired 1D plots. It is not good practice to integrate data over more than one m/z peak (for example, integrate over X from m/z = 20–80 amu). The only exception might be integrating over all the isotopes of xenon for calibration purposes, provided there are no other signals from other species in this region. References [1] Chandler, D. W.; Houston, P. L., Two-dimensional imaging of state-selected photodissociation products detected by multiphoton ionization. J. Chem. Phys. 1987, 87 (2), 1445-1447. 45 [2] Gebhardt, C. R.; Rakitzis, T. P.; Samartzis, P. C.; Ladopoulos, V.; Kitsopoulos, T. N., Slice imaging: A new approach to ion imaging and velocity mapping. Rev. Sci. Instrum. 2001, 72 (10), 3848-3853. [3] Suits, A. G.; Continetti, R. E., Imaging in chemical dynamics: The state of the art. ACS Publications: 2001. [4] Townsend, D.; Minitti, M. P.; Suits, A. G., Direct current slice imaging. Rev. Sci. Instrum. 2003, 74 (4), 2530-2539. [5] Whitaker, B. J., Imaging in molecular dynamics: technology and applications. Cambridge university press: 2003. [6] Taatjes, C. A.; Hansen, N.; Osborn, D. L.; Kohse-Höinghaus, K.; Cool, T. A.; Westmoreland, P. R., “Imaging” combustion chemistry via multiplexed synchrotron-photoionization mass spectrometry. Phys. Chem. Chem. Phys. 2008, 10 (1), 20-34. [7] Ryazanov, M. Development and implementation of methods for sliced velocitymap imaging. Studies of overtone-induced dissociation and isomerization dynamicsof hydroxymethyl radical (CH2OH and CD2OH). Ph.D. Thesis, University of Southern California, 2012. [8] Aristov, V. Spectroscopy, Unimolecular and Bimolecular Reactions of Methyl and Hydroxymethyl Radicals. Ph.D. Thesis, University of Southern California, 2000. [9] Conroy, D. Rydberg State of an Open Shell Species: Characterization and Photophysics of the 3pz State of CH2OH. Ph.D. Thesis, University of Southern California, 2000. [10] Feng, L. Spectroscopy and photodissociation dynamics of the hydroxymethyl radical (CH2OH). Ph.D. Thesis, University of Southern California, 2004. [11] Bamford, D. J.; Dyer, M. J.; Bischel, W. K., Single-frequency laser measurements of two- photon cross sections and Doppler-free spectra for atomic oxygen. Phys. Rev. A 1987, 36 (7), 3497. [12] Wiley, W.; McLaren, I. H., Time-of-flight mass spectrometer with improved resolution. Rev. Sci. Instrum. 1955, 26 (12), 1150-1157. [13] Eppink, A. T.; Parker, D. H., Velocity map imaging of ions and electrons using electrostatic lenses: Application in photoelectron and photofragment ion imaging of molecular oxygen. Rev. Sci. Instrum. 1997, 68 (9), 3477-3484. [14] Bordas, C.; Paulig, F.; Helm, H.; Huestis, D., Photoelectron imaging spectrometry: Principle and inversion method. Rev. Sci. Instrum. 1996, 67 (6), 2257-2268. 46 [15] Vrakking, M. J., An iterative procedure for the inversion of two-dimensional ion/photoelectron imaging experiments. Rev. Sci. Instrum. 2001, 72 (11), 4084-4089. [16] Rallis, C.; Burwitz, T.; Andrews, P.; Zohrabi, M.; Averin, R.; De, S.; Bergues, B.; Jochim, B.; Voznyuk, A.; Gregerson, N., Incorporating real time velocity map image reconstruction into closed-loop coherent control. Rev. Sci. Instrum. 2014, 85 (11), 113105. [17] Dribinski, V.; Ossadtchi, A.; Mandelshtam, V. A.; Reisler, H., Reconstruction of Abel- transformable images: The Gaussian basis-set expansion Abel transform method. Rev. Sci. Instrum. 2002, 73 (7), 2634-2642. [18] Ryazanov, M.; Reisler, H., Improved sliced velocity map imaging apparatus optimized for H photofragments. J. Chem. Phys. 2013, 138 (14), 144201. [19] Ryazanov, M.; Rodrigo, C.; Reisler, H., Overtone-induced dissociation and isomerization dynamics of the hydroxymethyl radical (CH2OH and CD2OH). II. Velocity map imaging studies. J. Chem. Phys. 2012, 136 (8), 084305. [20] Mahon, R.; McIlrath, T.; Myerscough, V.; Koopman, D., Third-harmonic generation in argon, krypton, and xenon: bandwidth limitations in the vicinity of Lyman-α. IEEE J. Quantum Electron. 1979, 15 (6), 444-451. [21] Delone, N. B.; Kraĭnov, V. P., Fundamentals of nonlinear optics of atomic gases. John Wiley & Sons Inc: 1988. [22] Sutradhar, S. Photodissociation Dynamics of Atmospherically Relevant Small Molecules in Molecular Beams. Ph.D. Thesis, University of Southern California, 2019. [23] Beames, J. M.; Liu, F.; Lester, M. I., 1+1’ resonant multiphoton ionisation of OH radicals via the A 2 Σ + state: insights from direct comparison with A-X laser-induced fluorescence detection. Mol. Phys. 2014, 112 (7), 897-903. [24] Slagle, I. R.; Yamada, F.; Gutman, D., Kinetics of free radicals produced by infrared multiphoton-induced decompositions. 1. Reactions of allyl radicals with nitrogen dioxide and bromine. J. Am. Chem. Soc. 1981, 103 (1), 149-153. [25] Goulay, F.; Osborn, D. L.; Taatjes, C. A.; Zou, P.; Meloni, G., Synergies between Experimental and Theoretical Studies of Gas Phase Reactions. Phys. Chem. Chem. Phys. 2007, 9, 4291-4300. [26] Taatjes, C. A., Uncovering the fundamental chemistry of alkyl + O2 reactions via measurements of product formation. J. Phys. Chem. A 2006, 110 (13), 4299-4312. 47 [27] Meloni, G.; Zou, P.; Klippenstein, S. J.; Ahmed, M.; Leone, S. R.; Taatjes, C. A.; Osborn, D. L., Energy-resolved photoionization of alkylperoxy radicals and the stability of their cations. J. Am. Chem. Soc. 2006, 128 (41), 13559-13567. [28] Osborn, D. L.; Taatjes, C. A., The physical chemistry of Criegee intermediates in the gas phase. Int. Rev. Phys. Chem. 2015, 34 (3), 309-360. [29] Dawson, J.; Guilhaus, M., Orthogonal-acceleration time-of-flight mass spectrometer. Rapid Commun. Mass Spectrom. 1989, 3 (5), 155-159. 48 Two photon absorption and fragmentation processes Introduction The UV photodissociation of PA from its second excited state is studied in molecular beams using time of flight (TOF) mass spectroscopy and time-sliced VMI following excitation to the first absorption band (S1←S0) at 330-380 nm. However, due to its relatively long lifetime, and a relatively high S2←S1 transition dipole moment, a second photon is absorbed, leading to several photoproducts that can be detected by resonant or non-resonant ionization. The vibrational features of the S1 state are characterized by the photofragment yield (PFY) spectrum of H, CH3CO and HOCO products obtained using He and Ar carrier gases in a molecular beam. The PFY spectrum, recorded in the 350–380 nm range, shows sharp vibrational features reflecting the significant rotational cooling achieved in the molecular beam. The spectrum matches quite well with the broad features observed in the room temperature absorption spectrum, and the narrow peak linewidths indicate that the S1 state lives longer than a picosecond. We identify the origin band of the S1←S0 transition at 26,710 cm -1 , and tentatively assign progressions in the CH3 and C-C torsional modes. The two- photon nature of formation of the fragments was confirmed by their KER distributions 49 measured by monitoring CH3CO, HOCO, CO, CH3, and H products. With the help of classical three-body bond breaking simulations, we conclude that three-body fragmentation is a major dissociation mechanism leading to the observed products. Due to the high excitation energy in the two-photon excitation (~54,000 cm -1 ) and the relatively low bond dissociation energies, it is not surprising that multiple bond fragmentations occur and small fragments like CH3CO, HOCO, CO, CH3, and H are detected as photodissociation products. KER distributions and angular anisotropy distributions of CH3CO, HOCO, CO, CH3, and H fragments indicate that additional photon absorption from S1 to the S2/S3 states is facile and is followed by rapid dissociation to the observed fragments. Based on the energetics of the different dissociation pathways and analyses of the observed KER distributions, three-body fragmentation processes are proposed as major contributors to the formation of the observed products. Figure 3.1 The two photon-absorption via the long lived S1 state to the S2 state and subsequent three-body fragmentation into smaller fragments discussed in this chapter are depecited. Norrish type I bond fission behavior in keto-acids like PA have not been reported previouly. 50 Experimental section PA, procured as a yellow viscous liquid (98%, Sigma-Aldrich), is purified by vacuum distillation at 50-60 °C twice and the distilled colorless sample is then degassed using several freeze-pump-thaw cycles.[1] A carrier gas (He or Ar) is bubbled through a glass bulb with 5– 10 ml PA to generate a gas phase mixture of 1 Torr PA vapors in 1300 Torr He. The gas mixture is introduced into the source region of the vacuum chamber and supersonically expanded through a piezoelectrically driven pulsed nozzle operating at 10 Hz. After passing through a skimmer (Beam Dynamics, 1.0 mm diameter), PA is excited by UV laser radiation (330-380 nm, 2–3 mJ/pulse) produced by frequency doubling the output of a dye laser (Continuum ND6000, DCM dye) pumped by the second harmonic output of a Nd:YAG laser (Continuum, PL8000, 10 Hz). TOF mass spectroscopy and VMI is used to characterize H, CH3CO, HOCO, CO and CH3 photofragments ionized using appropriate probe laser radiation. H photofragments are ionized using 1+1’ two-color REMPI scheme via the Lyman-α transition. The required vacuum ultraviolet (VUV) laser radiation at 121.567 nm is generated by frequency tripling ∼365-nm (2–3 mJ) radiation focused into a gas cell (see Chapter 2) filled with Kr and Ar gases in 200:590 Torr ratio. The 365 nm radiation is generated by frequency doubling the output of a dye laser (Continuum, ND6000, LDS 722 dye) pumped with the second harmonic of a Nd:YAG laser(Continuum, NY-81C). The tripled VUV and residual 365 nm radiation are focused into the interaction region using a MgF2 lens (f.l. = 7.5 cm). 51 CH3CO and HOCO photofragments are detected by one-color multiphoton non- resonant ionization (pump radiation also ionizes the photofragments). There are no known REMPI schemes for CH3CO and HOCO and the non-state-selective detection can encompass all the internal states of these species that are below their barriers to dissociation. CO (X 1 Σ + , v’’=0) products are detected by 2+1 REMPI at 230.1–230.5 nm via the B 1 Σ + ←X 1 Σ + transition.[2] UV radiation in this range is generated by the frequency doubling the output of a dye laser (Continuum ND6000, Coumarin dyes, 0.5–1 mJ/pulse) pumped by the third harmonic output of a pulsed Nd:YAG laser (Continuum, PL8000, 10 Hz). CH3 ( 𝑋𝑋 � 2 A2’’) products are detected by 2+1 REMPI at 325–336 nm (~1 mJ/pulse) via the 3p 2 A2’’← 𝑋𝑋 � 2 A2’’ transition.[3-4] The KER distributions of the product ions were monitored using SVMI, which has been discussed in Chapter 2, and detailed description of the appartus is given elsewhere.[5- 7] Briefly, a series of electrostatic lenses accelerate the ions produced in the detection region through a TOF tube towards a position sensitive detector (40 mm diameter MCP coupled to a phosphor screen). Ion hit events on the detector are recorded by a PixeLINK PL-B741F video-camera located behind the phosphor screen, and intensity vs. 2D pixel data is converted to intensity vs. energy using appropriate VMI principles. 52 Results Photofragment yield spectra H photofragments, produced by PA fragmentation on the S2 electronic state, showed sharp vibrational features. These features were in good agreement with the broad room- temperature absorption spectrum of the S1 state as shown in Figure 3.2 below. The H-PFY spectra were recorded using He and Ar carrier gases by scanning the pump laser radiation in the 350–380 nm range while the probe laser was fixed at the Lyman-α transition for H detection via REMPI. A comparison with the room temperature absorption spectrum reveals significant rovibrational cooling of PA in the molecular beam under He. The corresponding spectrum recorded in Ar showed even greater rotational cooling of the broad S1←S0 absorption features originally recorded at 298 K. Both spectra become progressively congested towards shorter wavelengths. The last vibrational band is observed at 374.4 nm (26710 cm -1 ) and is assigned as 0–0 band of the S1←S0 transition. A plot highlighting the extent of cooling upon changing carrier gases has been reported elsewhere.[8-9] Despite the excellent agreement between the room temperature absorption spectrum and the H-PFY spectrum, it is important to note the difference between the two. Peak intensities in the room temperature PA absorption spectrum reflects the rovibrational population of S0 at 298 K and the Franck–Condon factors for the S1←S0 transition. Although the H-PFY spectrum originates on S0, PA is assumed to be mainly in ground vibrational state due to collisional cooling in the molecular beam. The observed peak intensities in the PFY 53 spectrum depend on the H-dissociation efficiency from each vibrational level of the S1 state, in addition to the Franck–Condon overlap. Figure 3.2 The jet-cooled spectrum of the first excited state of PA is inferred from the H- PFY spectrum obtained in He (red). The corresponding broad features in the room temperature absorption spectrum[10] (black) is shown for comparison. PA has two stable conformers in the ground state. The most stable conformer, Tc, comprises 97% of the population at room temperature, and lies 730 cm -1 below the Tt conformer (the barrier height for isomerization is unknown).[11] The two relevant conformers of ground state PA are shown in Figure 3.3. The participation of both these conformers are assessed in order to assign the H-PFY spectrum. 54 Figure 3.3 The Tc and Tt conformers of PA are present in the ground state with Boltzmann population values of 97% and 3%, respectively, at 298 K. Intramolecular H-bonding makes the Tc conformer ∼2.0 kcal/mol more stable than the Tt conformer.[11-12] In order to characterize the participation of the Tt isomer, the H-PFY spectrum was obtained under different heating conditions of the sample glass bulb. Measurements at 318 K and 338 K did not exhibit any changes to the original spectrum obtained with the glass bulb at 298 K. This, along with the large population difference between the two confomers, showed that the recorded H-PFY spectrum represents primarily absorption by the more stable Tc conformer. If the measured PFY spectrum for H is a fingerprint of the S1 state, then it should not depend on the identity of the fragment detected. In order to confirm this, we obtained PFY spectra of additional photoproducts, CH3CO (m/z=43) and HOCO (m/z=45) via one-color multiphoton ionization (shown in Figure 3.4), which match well the H-PFY spectrum. 55 Figure 3.4 PFY spectra of the (a) CH3CO and (b) HOCO fragments produced from PA photodissociation. A more detailed analysis of the rovibronic features in these PFY spectra and their tentative assignments are discussed in a later chapter focused on the S1 state. Regardless, it suffices to say that the S1 state plays an important role in the photodissociation of PA. The mechanistic origin of the obtained photofragments is discussed in subsequent sections in this chapter. Fragments’ kinetic energy release We recorded time sliced images of the fragments at excitation energies corresponding to several peaks in the H-PFY spectrum. We observed changes in the KER distributions of some fragments as we varied the excitation energy of PA. This suggests that 56 several dissociation mechanisms are involved. The VMI technique helps obtain a comprehensive view of the photodissociation dynamics of PA. The KER distributions of CH3CO and H fragments and their corresponding recoil anisotropy parameters are shown in Figures 3.5 and 3.6. The KER distributions shown here are the individual fragment KEs and not the customary center-of-mass (c.m.) KER. This is because each product can be formed via more than one pathway. Figure 3.5 KER distributions (top) and corresponding recoil anisotropy parameters, β, (bottom) obtained by monitoring CH3CO fragments following S1 excitation at two different peaks of the PFY spectrum in Figure 3.2. The simulated KEmax values based on three-body fragmentation channels are marked by arrows. The KER distributions of CH3CO (Figure 3.5) exhibit two components: a slow component peaking at around 200 cm -1 , and a fast component with a maximum around 2500 57 cm -1 and a long tail extending up to ~6500 cm -1 . The relative contribution of the high KER component increases with excitation energy, and the distribution becomes clearly bimodal. The recoil anisotropy parameter (β) calculated for each of the distributions is shown as a function of KER in the lower panel. The low KE component has an average β value of 0.2–0.5. The high KE component, observed at higher excitation energies, is more anisotropic with β value in the 1.4–1.5 range. Figure 3.6 KER distributions (top) and corresponding recoil anisotropy parameters, β, (bottom) of H fragments following S1 excitation at two different peaks of the H-PFY spectrum. The simulated KEmax values based on the indicated three-body fragmentation channels are marked by arrows. 58 Like CH3CO, the H-photofragment KER distribution is also broad and multimodal, as seen in Figure 3.6. It extends up to ~25,000 cm -1 with three apparent features centered at about 1500, 5000 and 13,000 cm -1 . The angular distribution of the lowest KER component is nearly isotropic with β values in the range −0.3 – 0, but the high KER component is mostly anisotropic with β values in the −0.6 – −0.4 range. The other detected products — CO, HOCO, and CH3, showed mainly a low KE component. The KER distributions of ground state CO(X 1 Σ + , v’’=0) was recorded following excitation of the PA 0-0 bandhead and is shown in Figure 3.7. Figure 3.7 KER distribution of CO(X 1 Σ + , v’’=0, rotational bandhead) products detected by (2+1) REMPI at 230.1 nm following PA excitation at 374.4 nm. The predicted KEmax values for synchronous and sequential three-body fragmentation processes are marked by arrows. A small signal at m/z = 45 was observed in a one-color experiment and assigned to HOCO. The low relative yield of the HOCO radical could be because of its low dissociation 59 threshold or poor ionization cross-section.[13] Its KER distribution is shown in Figure 3.8 below. Figure 3.8 KER distribution of HOCO products detected by non-resonant ionization following PA excitation at 369.7 nm. The predicted KEmax values for synchronous and sequential three-body fragmentation processes are marked by arrows. In addition, we also detected CH3 radicals by using one-color (2+1) REMPI at 333.77 and 329.95 nm to probe their formation in the ground (v1’’v2’’v3’’v4’’ = 0000) and excited (v1’’v2’’v3’’v4’’ = 0100) vibrational levels, respectively. The corresponding KER distributions of these CH3 vibrational states were also recorded. However, the obtained CH3 signal originated from PA excitation at the probe wavelengths (one-color experiment) and attempts to detect CH3 in a two-color pump-probe experiment were unsuccessful. This is because the absorption cross section of PA at the probe wavelengths (325–336 nm) is much greater than that at the 369–374 nm pump wavelength.[10] All the detected fragments exhibit high KEs, much higher than allowed in one-photon absorption, while their PFY spectra show features characteristic of absorption to the S1 state. 60 A two-photon absorption followed by dissociation from a higher excited state is the only possible explanation for the observed results. Electronic structure calculations and three- body fragmentation simulations bolster our reasoning. The mechanistic interpretation and simulation details of the channel KEmax values marked in the figures above, are discussed in the next few sections. Discussion Excited states of PA To ascertain the feasibility of two-photon excitation in PA the excitation energies and the nature of the electronic excited states are calculated at the EOM-EE-CCSD/6-311(2+)G** level of theory. The vertical excitation energies from S0 to S1, S2 and S3, the oscillator strengths of the transitions, and the directions of the transition dipole moments are in good agreement with previous calculations,[14] and are listed in Table 3.1. Based on the calculated vertical S2←S0 excitation energy (45,160 cm -1 ), a two-photon excitation via the origin (0-0) band of S1 at 374.4 nm, which imparts 53,420 cm -1 of total energy, can easily access the S2 state. The oscillator strength for the S2←S1 transition is approximately a factor to 10 3 larger than that for S3←S1, making this the most likely process. Dhanya et al. have previously observed the formation of OH products in the photodissociation of PA at 193 nm (~52,000 cm -1 ) and concluded that both the S2 and S3 states can be reached using a single 193 nm photon.[14] More proof in favor of the participation of the S2 state is drawn from the direction of the transition dipole moment. The 61 S2←S1 transition is parallel and is expected to give rise to a recoil anisotropy parameter βmax = 2 in the case of fast dissociation. On the other hand, the S3←S1 transition is perpendicular and βmax = −1 is expected in this case. Table 3.1 Calculated transition dipole moments between the indicated states, their oscillator strengths and vertical excitation energies are obtained using EOM-EE-CCSD/6- 311(2+)G**. The X, Y, and Z axes are defined with respect to the molecular plane (XY) containing the CC(O)C(O)O skeleton. Transition Transition dipole direction Oscillator Strength Vertical excitation energy (cm -1 ) Character S1←S0 Perpendicular (Z) 6.0x10 -5 29,850 π*←n+ S2←S0 Perpendicular (Z) 1.6x10 -4 45,160 π*←n– S3←S0 Parallel (X,Y) 7.0x10 -2 56,500 π*←π S2←S1 Parallel (X,Y) 4.3x10 -2 11,600 S3←S1 Perpendicular (Z) 2.8x10 -5 26,050 Energetically allowed reactions Several bond dissociation pathways become energetically accessible at the two- photon energies used in this experiment. The estimated thermochemical heats of reaction (∆Hrxn) that may give rise to the observed products is tabulated below. In this chapter we focus on reactions arising from the Norrish type I and H-fragmentation reactions. ΔHrxn values for the relevant reactions from these two processes are listed in cm −1 below and are accurate to within 400 cm −1 . As discussed below, both two- and three-body fragmentation processes are energetically allowed. 62 CH3COCOOH → CH3CO + HOCO ∆Hrxn = 28,600 1 CH3COCOOH → CH3CO + H + CO2 ∆Hrxn = 28,900 1a CH3COCOOH → CH3CO + OH + CO ∆Hrxn = 37,600 1b CH3COCOOH → CH3 + CO + HOCO ∆Hrxn = 31,600 1c CH3COCOOH → CH2CO + H + HOCO ∆Hrxn = 42,600 1c CH3COCOOH → CH3COCOO + H ∆Hrxn = 39,000 2 CH3COCOOH → CH2COCOOH + H ∆Hrxn = 33,000 3 CH3COCOOH → CH3COCO + OH ∆Hrxn = 36,700 4 ∆Hrxn for reaction 1 has been calculated by Gabriel da Silva (private communication) and for reactions 2, 3 by David Osborn (private communication). ΔHrxn for reactions 1a−1c are estimated, respectively, by adding the reported ΔHrxn values of HOCO → H + CO2, HOCO → OH + CO, and CH3CO → CH3 + CO reactions to ΔHrxn of reaction 1.[13,15] Alternatively, ∆Hrxn for these reactions can be estimated with similar results (within 400 cm -1 ) from the heats of formation (∆Hf 0 ) of reactants and products.[16] The ∆Hrxn for reaction 4 is estimated based on the reported calculations of Dhanya et. al.[14] Clearly, both two- and three-body fragmentation processes are energetically feasible following two-photon absorption. The contributions of these processes can be distinguished based on the KER plots shown in Figures 3.5–3.8, focusing primarily on the observed maximum KER values, KEmax, for each fragment. 63 Two-body fragmentation pathways The maximum value, KEmax, of each fragment in the two-body fragmentation reactions 1, 2 – 4 will be observed when the fragments have no internal energy. For reaction 1, the maximum allowed total KE in both fragments is 53,400 – 28,600 cm -1 = 24,800 cm -1 for excitation via the 0–0 band of S1←S0. In accordance with momentum conservation for a two- body fragmentation, the CH3CO fragment can have KEmax = 12,700 cm -1 . The KE of the CH3CO product should then reach a minimum of 9600 cm -1 because CH3CO generated with internal energy above its dissociation barrier of ~6000 cm -1 will not be observed via ionization. Therefore, the observed KEmax of only 6500 ± 500 cm -1 (Figure 3.5) in the CH3CO fragment would indicate that the HOCO cofragment possesses significant internal energy. In the complementary KER measurement of HOCO shown in Figure 3.8, we find that its observed KEmax is 5000 ± 500 cm -1 , less than the 12,100 cm -1 value one would expect for cold CH3CO cofragments. This confirms that the CH3CO cofragments are produced internally hot. HOCO and CH3CO fragments with internal energies exceeding their dissociation barriers will further dissociate leading to three-body fragmentation and these processes are discussed in the next section. Similarly, primary H photofragments generated by reactions 2 and 3 have expected KEmax values of 14,000 ± 400 and 20,000 ± 400 cm -1 , respectively. While reactions 2 and 3, which capture C-H and O-H bond fission reactions, respectively, might contribute to the observed KER distribution of H fragments (Figure 3.6), they are not likely to explain the observed KEmax of 23,000 cm -1 . Therefore, two-body fragmentation processes alone cannot 64 explain the observed KEmax of CH3CO, HOCO and H products. In the next section we obtain KEmax values for three-body fragmentation processes which consider dissociation of more than one bond. Three-body fragmentation pathways Three-body fragmentation processes have been discussed in the literature extensively, and here we follow the treatment of Maul and Gericke,[17-18] who distinguish between what they term synchronous, sequential, and asynchronous processes. Synchronous three-body fragmentation refers to simultaneous breaking of two bonds to produce three fragments. Sequential dissociation involves the initial breaking of one bond that results in two primary products, one of which has enough internal excitation to further dissociate. The asynchronous dissociation process is less well defined; it involves breaking of the two bonds on time scales that differ by less than a rotational period of the molecule. The treatment of Maul and Gericke is described here for three-body dissociation of molecule A–B–C to fragments A, B and C. The corresponding generalized equations for KEmax of the fragments are then derived for the synchronous and sequential processes. The parametric dependence of the KEmax value is different for the two dissociation processes; it depends on the critical angle of dissociation, α, for synchronous processes, whereas for sequential fragmentation it depends on the KE in the fragments produced during the first dissociation step. The KEmax value for each observed product is calculated for each of its formation pathways and compared with the observed KER distribution of each fragment as shown in section 3.4.2 65 3.4.4.I Synchronous three-body fragmentation The fragmentation of an ABC molecule into fragments A, B, and C is discussed here. For synchronous dissociation, the fragments’ KER distributions depend on the central A-B-C angle at the critical molecular geometry at the moment of fragmentation. Figure 3.9 A synchronous three body fragmentation of molecule ABC is depicted. The X- axis is aligned in the direction of 𝑝𝑝 𝐵𝐵 � � � � ⃗ and as a result it divides the critical A–B–C angle, α, equally. In the case of PA, this geometry is unknown, but the anisotropy of the angular distributions (Figures 3.5 and 3.6) suggests that the final dissociation on S2 is fast and direct. Furthermore, Dhanya et al. calculated the geometries of the S0, S1, S2 and S3 states to be largely similar.[14] It can therefore be assumed that α during dissociation is similar to the corresponding bond angle in the S1 state. Based on Figure 3.9, energy and linear momentum conservation dictates that 𝑝𝑝 A � � � � ⃗ + 𝑝𝑝 B � � � � ⃗ + 𝑝𝑝 C � � � � ⃗ = 0 (3.1) 66 𝐸𝐸 A k in + 𝐸𝐸 B k in + 𝐸𝐸 C k in = ℎ𝜈𝜈 − Δ 𝐻𝐻 rx n − � 𝐸𝐸 in t A, B, C = 𝜖𝜖 . (3.2) Here ℎ𝜈𝜈 refers to the two photon energy; Δ 𝐻𝐻 rx n is the heat of reaction of the ABC molecule fragmenting into A, B and C and leading to KER of 𝐸𝐸 A k in , 𝐸𝐸 B k in , and 𝐸𝐸 C k in , respectively; 𝐸𝐸 in t is the internal energy of the fragments, and 𝜖𝜖 is the total available KE or the c.m. KE. The fragments A, B and C have masses 𝑚𝑚 𝑃𝑃 , 𝑚𝑚 𝐵𝐵 , and 𝑚𝑚 𝐶𝐶 , respectively. Obviously, KEmax will be observed when the three fragments are formed with no internal energy. Because of the choice of coordinate system for this system, 𝑝𝑝 A Y = − 𝑝𝑝 C Y or | 𝑝𝑝 A | sin(𝛼𝛼 /2) = | 𝑝𝑝 C | sin(𝛼𝛼 /2), which simplifies to | 𝑝𝑝 A | = | 𝑝𝑝 C | = 𝑝𝑝 . As a result, equations 3.1 and 3.2 can be written as 𝑝𝑝 B = −2 𝑝𝑝 cos( 𝛼𝛼 /2) (3.3) 𝐸𝐸 B k in = 2 𝑝𝑝 2 2 𝑚𝑚 B cos 2 (𝛼𝛼 /2). (3.4) Solving the above system of equation yields the following for the KEmax values of fragments A, B and C: 𝐸𝐸 B k in = 𝜖𝜖 [(𝑚𝑚 B (𝑚𝑚 A + 𝑚𝑚 C )/4 𝑚𝑚 A 𝑚𝑚 C )sec 2 (𝛼𝛼 /2)] + 1 , (3.5) 𝐸𝐸 A k in = 𝜖𝜖 [(𝑚𝑚 A + 𝑚𝑚 C )/𝑚𝑚 C ] + 4(𝑚𝑚 A /𝑚𝑚 B )cos 2 (α/2) , (3.6) 𝐸𝐸 C k in = 𝜖𝜖 [(𝑚𝑚 A + 𝑚𝑚 C )/𝑚𝑚 A ] + 4(𝑚𝑚 C /𝑚𝑚 B )cos 2 (α/2) . (3.7) 67 Clearly, the KEmax value for each fragment will depend on the choice of 𝛼𝛼 around the central fragment. For reactions 1b and 1c, α is defined as the angles C4C5C7 and C5C7O9, respectively (See Figure 3.10 for the numbering of the atoms in PA). For reaction 1a where CO2 elimination occurs, α can be estimated by the angle subtended by the participating C5C7 and O9H10 bonds (Figure 3.10). A previous theoretical study reported that the C4C5C7 and C5C7O9 angles in S1 are 119° and 115°, respectively, and these define the values of α for reactions 1b and 1c.[19] The C7O9H10 angle was calculated to be 107°, leading to α = 42° for reaction 1a (Figure 3.10).[19] Similarly, α for reaction 1d can be estimated by assuming that bond cleavage happens along the in-plane C4H1 and C5C7 bonds. The angle subtended by the two estimated to be 49° based on the H1C4C5 (109°) and the C4C5C7(120°) angles. We note, that reaction 1d can also be formed from breaking of any of the out-of-plane C4H2 and C4H3 bonds as well and can result in a different α value. However, the KEmax for H1, H2 or H3 fragments from reaction 1d will be similar, because these values only slightly change with α (Figure 3.10). The KEmax value for each fragment in Figure 3.10 is obtained in this model by using ℎ𝜈𝜈 = 54000 and ∆Hrxn for the appropriate reaction producing the fragment. The calculated KEmax values are marked by arrows in Figures 3.5–3.8, listed in Table 3.2, and also shown in green vertical dashed lines in Figure 3.10. 68 Figure 3.10 KEmax of the final products of synchronous three-body fragmentation reactions are simulated for (a) reaction 1a, (b) reaction 1b, (c) reaction 1c, and (d) reaction 1d as a function of α. These plots are calculated for an excitation energy of 54000 cm -1 . The estimated KEmax value of the H fragment generated in reaction 1a is close to the observed one, as shown in Figure 3.6, and so is the corresponding value for HOCO (Figure 3.8) generated in reaction 1c. The KEmax values for H and HOCO corresponding to reaction 1d are indicated in Figures 3.6 and 3.8, respectively. The observed values for CO (Figure 3.7) and CH3CO (Figure 3.4), however, are lower than the predicted KEmax, and might signify that 69 either the fragments possess significant internal energy, and/or they are generated by sequential processes. Table 3.2 Estimated angle α and corresponding KEmax values of fragments produced via synchronous three body-fragmentation reactions 1a–1c. The error bars on KEmax represent the deviation from the average value for a change of ± 5° in the corresponding α value. Excitation energy, ℎ𝜈𝜈 , corresponding to the employed probe wavelength ais used for this table. For 1a, 1b and 1d, ℎ𝜈𝜈 = 54000 cm -1 (~370 nm). For 1c, during CH3 detection ℎ 𝜈𝜈 = 59500 cm -1 (~ 335 nm). Reaction α Fragments KEmax (cm -1 ) 1a 42° CH3CO 529 ± 1 CO2 1800 ± 50 H 22,750 ± 50 1b 115° CH3CO 3140 ± 140 CO 5560 ± 500 OH 7940 ± 360 1c 119° HOCO 4960 ± 50 CO 8200 ± 860 CH3 14,900 ± 50 1d 49° HOCO 229 ± 1 CH2CO 810 ± 29 H 10300 ± 30 70 3.4.4.II Sequential three-body fragmentation As discussed above, the KER plots displayed in Figures 3.5 and 3.8 indicate that CH3CO and HOCO fragments produced via reaction 1 have significant internal energies. The observed KEmax values are 6500 and 5000 cm -1 , respectively, leaving up to ~ 13,300 cm -1 to distribute in internal energy of these fragments. In addition, noting that most CH3CO and HOCO fragments have fairly low KEs, some of their cofragments must have internal energies exceeding their dissociation barriers. ∆Hrxn for CH3CO and HOCO dissociation reactions are listed below in cm -1 , with barrier heights listed in parentheses.[13,15] CH3CO → CH3 + CO ∆Hrxn = 3000 (6000) 5 CH3CO → CH2CO + H ∆Hrxn = 14,500 6 HOCO → OH + CO ∆Hrxn = 9000 (barrierless) 7 HOCO → H + CO2 ∆Hrxn = 300 (12,000) 8 We consider again a triatomic ABC molecular analogue and assume that to achieve KEmax for each fragment, the final A, B, and C fragments must have no internal energy. A sequential fragmentation of ABC proceeds in two steps: ABC → AB + C → A + B + C. Unlike the synchronous three-body fragmentation case, where the calculated KEmax values of the products are determined solely by the angle α, the sequential case depends on the fragment KEs after the first step.[17-18] The kinematics of a sequential three body fragmentation are shown for molecule ABC in Figure 3.11. 71 Figure 3.11 A schematic of a sequential three-body fragmentation of ABC where the B-C bond breaks first at time τ1 producing AB and C fragments. The AB fragment can further dissociate after time τ2, which is longer than the mean rotational period of the AB intermediate. The observed laboratory-frame velocities of the A and B fragments depend on the velocity of AB. Referring to Figure 3.11, the maximum KER in the final products of a sequential process are derived by using, ℎ𝜈𝜈 − Δ 𝐻𝐻 rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 1 = 𝐸𝐸 C k in + 𝐸𝐸 C in t + 𝐸𝐸 AB k in + 𝐸𝐸 AB in t (3.8) 𝐸𝐸 AB in t − Δ 𝐻𝐻 rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 2 = 𝐸𝐸 A in t + 𝐸𝐸 A k in,C M + 𝐸𝐸 B in t + 𝐸𝐸 B k in,C M . (3.9) Here Δ 𝐻𝐻 rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 1 and Δ 𝐻𝐻 rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 2 are the heats of reaction for the first and second fragmentation steps, respectively. 𝐸𝐸 A k in,C M and 𝐸𝐸 B k in,C M represent the KE of A and B fragments, respectively, in the c.m. frame of the AB fragment. Obviously, the maximum KER release 72 corresponds to zero internal energy in the fragments, and the equations above can be simplified to yield ℎ𝜈𝜈 − Δ 𝐻𝐻 rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 1 = 𝐸𝐸 C k in + 𝐸𝐸 AB k in + 𝐸𝐸 AB in t (3.10) 𝐸𝐸 AB in t − 𝛥𝛥 H rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 2 = 𝐸𝐸 A k in,C M + 𝐸𝐸 B k in,C M . (3.11) During each fragmentation step, linear momentum is also conserved in the frame of reference of each step of bond breaking into two fragments. Therefore, 𝑚𝑚 C 𝑣𝑣 C � � � � ⃗ + 𝑚𝑚 AB 𝑣𝑣 AB � � � � � � � ⃗ = 0 (3.12) 𝑚𝑚 A 𝑣𝑣 A C M � � � � � � � ⃗ + 𝑚𝑚 B 𝑣𝑣 B C M � � � � � � � ⃗ = 0. (3.13) Whereas equation 3.12 gives the velocities of A and B in the c.m. frame of AB, their laboratory frame velocities are obtained from the vector additions, 𝑣𝑣 A � � � � ⃗ = 𝑣𝑣 A C M � � � � � � � ⃗ + 𝑣𝑣 AB � � � � � � � ⃗ (3.14) 𝑣𝑣 𝐵𝐵 � � � � ⃗ = 𝑣𝑣 𝐵𝐵 C M � � � � � � � ⃗ + 𝑣𝑣 AB � � � � � � � ⃗. (3.15) In the limiting case where the fragments’ velocity vectors in both dissociation steps are collinear (θ=0 in Figure 3.11), the maximum velocities of A and B will be achieved. For maximum KE in fragment A, | 𝑣𝑣 A | = � 𝑣𝑣 A C M + 𝑣𝑣 AB � , and for a maximum KE in fragment B, | 𝑣𝑣 B | = � 𝑣𝑣 B C M + 𝑣𝑣 AB � . It must be noted that both these conditions cannot be satisfied simultaneously, i.e., a maximum KE of fragment A would require a minimum c.m energy in fragment B. 73 Thus, the maximum fragment kinetic energies limiting cases for a given first step AB with internal energy 𝐸𝐸 AB in t can be written as 𝐾𝐾 𝐸𝐸 𝑚𝑚 𝑎𝑎𝑥𝑥 𝐶𝐶 = (𝑚𝑚 A + 𝑚𝑚 𝐵𝐵 )(ℎ𝜈𝜈 − Δ 𝐻𝐻 rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 1 − 𝐸𝐸 AB in t ) ( 𝑚𝑚 A + 𝑚𝑚 𝐵𝐵 + 𝑚𝑚 C ) (3.16) 𝐾𝐾 𝐸𝐸 𝑚𝑚 𝑎𝑎𝑥𝑥 𝑃𝑃 = 1 2 𝑚𝑚 A � � 2 𝑚𝑚 C (ℎ𝜈𝜈 − 𝛥𝛥 H rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 2 − 𝐸𝐸 AB in t ) (𝑚𝑚 A + 𝑚𝑚 𝐵𝐵 + 𝑚𝑚 C )(𝑚𝑚 A + 𝑚𝑚 𝐵𝐵 ) + � 𝑚𝑚 B (𝐸𝐸 AB in t − 𝛥𝛥 H rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 1 ) 𝑚𝑚 A (𝑚𝑚 A + 𝑚𝑚 𝐵𝐵 ) � 2 (3.17) 𝐾𝐾 𝐸𝐸 m ax B = 1 2 𝑚𝑚 B � � 2 𝑚𝑚 C (ℎ𝜈𝜈 − 𝛥𝛥 H rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 2 − 𝐸𝐸 AB in t ) (𝑚𝑚 A + 𝑚𝑚 𝐵𝐵 + 𝑚𝑚 C )(𝑚𝑚 A + 𝑚𝑚 𝐵𝐵 ) + � 𝑚𝑚 A (𝐸𝐸 AB in t − 𝛥𝛥 H rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 1 ) 𝑚𝑚 B (𝑚𝑚 A + 𝑚𝑚 𝐵𝐵 ) � 2 . (3.18) These equations are used to generate the values listed in Table 3.3, and the plots in Figure 3.12 based on the known values of 𝛥𝛥 H rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 1 and 𝛥𝛥 H rx n 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑠𝑠 2 . The internal energy of AB ( 𝐸𝐸 AB in t ) can be estimated from the c.m. KER in the first step. Therefore, the predicted KEmax value of each final product is plotted as a function of the c.m. KER in the first step in Figure 3.12. The maximum possible KER in the first step refers to a scenario where the non- dissociating fragment has zero internal energy and the dissociating fragment has internal energy just above its dissociation barrier. As the dissociating fragment internal energy increases above this value, the KE in the first step decreases concomitantly. 74 Figure 3.12 KEmax of the final products of sequential three-body fragmentation reactions are simulated for reaction 1 followed by (a) reaction 5, (b) reaction 6, (c) reaction 7, and (d) reaction 8 as a function of the KER in reaction 1. These plots are simulated for an excitation energy of 54000 cm -1 . A particular three-body fragmentation reaction is deemed acceptable by comparing the observed and computed values of KEmax for the fragments in the first and second steps. For example, the observed KEmax values of the CH3CO fragment is ~6500 (Figure 3.5). For this value, the correlated KEmax values expected for the products of the secondary dissociation of the HOCO cofragment to CO + OH and CO2 + H (reactions 7 and 8) are listed 75 in Table 3.3. The KEmax values of the products of the secondary dissociation of CH3CO (reactions 5 and 6), which are correlated with the observed ~5000 cm -1 KEmax value of the HOCO fragment (Figure 3.8), are also listed in Table 3.3. The predicted KEmax values are indicated by arrows in the KER plots of the CH3CO, HOCO, CO and H photofragments (Figures 3.5–3.8). Clearly, the sequential mechanism is consistent with the observed KEmax values of CH3CO, HOCO and possibly also CO. With the KEmax values of the synchronous and sequential fragmentation known, the likely dissociation pathways following two-photon absorption of PA is discussed next. Table 3.3 Computed KEmax values of fragments produced in the secondary reactions 5 – 8, correlated with the observed KEmax of CH3CO or HOCO generated in reaction 1. Reaction Observed KEmax of non- dissociating fragment (cm -1 ) Correlated KEmax of fragments from secondary dissociation step (cm -1 ) 5 5000 (HOCO) 15,000 (CO) 17,000 (CH3) 6 5000 (HOCO) 5860 (CH2CO) 1940 (H) 7 6500 (CH3CO) 10,000 (CO) 9500 (OH) 8 6500 (CH3CO) 9000 (CO2) 15,000 (H) 76 Mechanistic interpretations The multimodal appearance of the KER distributions of the H and CH3CO fragments suggests that more than one dissociation pathway contributes to the observed products. As discussed above, the FWHM of the H-PFY rovibronic bands (~8 cm -1 ) indicates (taking into account that each band includes an unresolved rotational envelope) that the lifetime of the S1 state is roughly 1 picosecond or more. This lifetime is also consistent with the observation of light emission when PA is excited with λ > 366 nm.[20] Consequently, PA in the S1 state may undergo rotational and vibrational motions before absorbing another photon, and these motions will affect the product recoil anisotropy. Even though the interpretation of product angular distributions in multiple-photon dissociation is not straightforward, it is likely that the observed anisotropy is indicative of fast dissociation on S2. As shown in Table 3.1, the direction of the transition dipole moment for the S2←S1 transition is parallel with respect to the molecular plane whereas it is perpendicular for S3←S1. The high KER component of the CH3CO distribution has a positive angular anisotropy parameter, β = 1.4–1.5, which is consistent with fast dissociation from the S2 state. The slow KER component with smaller anisotropy, β = 0.2–0.5, may derive from dissociation of vibrationally excited PA in the S1 or T1 states reached via nonradiative transitions from S2. The vibrational motions of PA involved in the couplings to lower electronic state(s) prior to dissociation can lead to reduction of the angular anisotropy of products. Internal conversion and intersystem crossings prior to dissociation are common in photodissociation of radicals 77 and large polyatomic molecules, which often exhibit fast and slow KE components that are anisotropic and isotropic, respectively.[21-23] This scenario is also consistent with the increase in the fraction of the fast and anisotropic CH3CO component with increasing excitation energy; at higher excitation energies, direct dissociation on S2 may compete effectively with couplings to the lower electronic states. Analysis of the angular distributions of the H fragment, which show an overall negative anisotropy (β = −0.2 – −0.6), is more complicated because several two- and three- body fragmentation processes can contribute to H formation. For example, the angular distributions of H fragments produced via two-body O-H bond fission (reaction 2) and synchronous three-body fragmentation (reaction 1a) will depend on the angle between the transition dipole moment vector and the dissociating bond. Similarly, the angular distribution of H fragments produced via reaction 3 will be determined by the orientation of the dissociating methyl C-H bond relative to the transition dipole moment vector. On the other hand, the dissociation of rovibrationally hot HOCO to H + CO2 (reaction 8) is likely to lead to H products with an isotropic angular distribution and a broad range of KEs. Taking into account the available energies for the different reactions, the KEmax values of the fragments, and the observed KER and angular distributions, it is clear that three-body fragmentations processes, synchronous and sequential, are major contributors to the CH3CO, HOCO, H, and CO products generated by two-photon dissociation. At the two-photon energy of ~54,000 cm -1 , all the listed three-body fragmentation pathways are energetically allowed. While the above discussion of the predicted KEmax values is restricted to synchronous and 78 sequential three-body fragmentation processes, asynchronous three body fragmentation might also contribute to the observed KER distributions, but this mechanism is more difficult to evaluate without theoretical calculations of dissociation timescales. Nevertheless, analyses of the observed KEmax values show that three-body fragmentation pathways, both synchronous and sequential, can explain the observed KER plots of all the products most consistently. Photoinduced three-body fragmentation has been observed previously in the photodissociation of molecules such as alkanes, alkenes, carbonyls, alcohols, and carboxylic acids.[24-37] In fact, it is quite common in molecules of the general formula CH3COR, where R includes H, alkyl, halogens, cyanide, hydroxyl, and alkoxy.[4,17-18,24-26,28-33,38] For example, photodissociation studies of acetone reveal a sequential three-body fragmentation pathway, where nascent CH3CO fragments further dissociate to CH3 and CO.[4,24,34-35] Similar pathways have been observed in organic aldehydes such as formaldehyde, acetaldehyde and higher analogues,[27-28,36] as well as in acetic acid and dihydroxybenzoic acid.[31,38] Synchronous three-body fragmentation has also been invoked in photodissociation of glyoxal, which is structurally similar to PA.[29-30,37] In the case of PA, R = COOH and both CO and CH3 fragments can be observed in a sequential mechanism when the internal energy of the CH3CO cofragment is above the barrier. In addition, as seen in Figure 3.5, synchronous processes can contribute to the lower KE component of the CH3CO KER distribution. 79 It should also be noted that the predicted KEmax values refer to cases when all the final dissociation fragments have zero internal energy. It is more realistic that many of the CH3CO and HOCO products of reaction 1 would have non-zero internal energies. When this internal energy is lower than their dissociation barriers, stable HOCO and CH3CO fragments will be observed. In general, the observed KER decreases when the reaction products have internal energy; therefore, internally hot fragments are correlated with KER values lower than the predicted KEmax. The results in this chapter point mainly to the S2 state as playing a major role in the formation of the observed products, and although, in previous studies, CO2 was identified[20,39-42] as a final dissociation product following one-photon excitation to S1, attempts to detect CO2 using 3+1 REMPI schemes were unsuccessful in this study due to limitations of the detection method. In a later chapter we describe the use of the MPIMS flow reactor setup in following the dissociation dynamics on the S1 state where CO2 is observed as a major product. Summary and conclusions The first study of the photodissociation of PA in molecular beams is described providing information on the S1←S0 electronic spectroscopy of the Tc conformer, and the photodissociation products generated by two-photon dissociation via the S1 state. Electronic structure calculations show that the oscillator strength of the S2←S1 transition is more than two orders of magnitude larger than that of the S1←S0 transition, while the S3←S1 oscillator strength is much smaller (Table 3.1). As a result, two-photon transitions via S1 (mainly to S2) 80 are facile, and in fact are difficult to suppress in focused laser experiments. This explains the generation of H, HOCO, CH3CO, CO and CH3 photodissociation products in this study. KER distributions of these fragments, obtained using VMI, confirmed the two-photon nature of the dissociation. The anisotropy in the angular distributions of the H and CH3CO indicates that dissociation on the S2 surface is fast. Analyses of the energetically allowed dissociation pathways show that both two- and three-body fragmentation processes are feasible. While this study was unable to quantify branching ratios of the observed pathways, it revealed new insights into the dissociation mechanisms of PA at high excitation energies. The maximum allowed KER values, KEmax, of the fragments for synchronous and sequential three-body fragmentation pathways show that several three-body fragmentation processes, both synchronous and sequential, contribute significantly to the observed products. Also, most of the CH3CO and HOCO fragments are generated with significant internal energy and many of them further dissociate. References [1] Reed Harris, A. E.; Doussin, J.-F.; Carpenter, B. K.; Vaida, V., Gas-phase photolysis of pyruvic acid: The effect of pressure on reaction rates and products. J. Phys. Chem. A 2016, 120 (51), 10123-10133. [2] Tjossem, P. J.; Smyth, K. C., Multiphoton excitation spectroscopy of the B 1 Σ + and C 1 Σ + Rydberg states of CO. J. Chem. Phys. 1989, 91 (4), 2041-2048. [3] Hudgens, J. W.; DiGiuseppe, T.; Lin, M.-C., Two photon resonance enhanced multiphoton ionization spectroscopy and state assignments of the methyl radical. J. Chem. Phys. 1983, 79 (2), 571-582. 81 [4] Trentelman, K. A.; Kable, S. H.; Moss, D. B.; Houston, P. L., Photodissociation dynamics of acetone at 193 nm: Photofragment internal and translational energy distributions. J. Chem. Phys. 1989, 91 (12), 7498-7513. [5] Ryazanov, M. Development and implementation of methods for sliced velocitymap imaging. Studies of overtone-induced dissociation and isomerization dynamicsof hydroxymethyl radical (CH2OH and CD2OH). Ph.D. Thesis, University of Southern California, 2012. [6] Ryazanov, M.; Reisler, H., Improved sliced velocity map imaging apparatus optimized for H photofragments. J. Chem. Phys. 2013, 138 (14), 144201. [7] Ryazanov, M.; Rodrigo, C.; Reisler, H., Overtone-induced dissociation and isomerization dynamics of the hydroxymethyl radical (CH2OH and CD2OH). II. Velocity map imaging studies. J. Chem. Phys. 2012, 136 (8), 084305. [8] Sutradhar, S. Photodissociation Dynamics of Atmospherically Relevant Small Molecules in Molecular Beams. Ph.D. Thesis, University of Southern California, 2019. [9] Sutradhar, S.; Samanta, B. R.; Fernando, R.; Reisler, H., Spectroscopy and Two-Photon Dissociation of Jet-Cooled Pyruvic Acid. J. Phys. Chem. A 2019, 123 (28), 5906-5917. [10] Horowitz, A.; Meller, R.; Moortgat, G. K., The UV–VIS absorption cross sections of the α- dicarbonyl compounds: pyruvic acid, biacetyl and glyoxal. J. Photochem. Photobiol. A: Chem. 2001, 146 (1-2), 19-27. [11] Reva, I. D.; Stepanian, S. G.; Adamowicz, L.; Fausto, R., Combined FTIR matrix isolation and ab initio studies of pyruvic acid: proof for existence of the second conformer. J. Phys. Chem. A 2001, 105 (19), 4773-4780. [12] Plath, K. L.; Takahashi, K.; Skodje, R. T.; Vaida, V., Fundamental and overtone vibrational spectra of gas-phase pyruvic acid. J. Phys. Chem. A 2009, 113 (26), 7294-7303. [13] Johnson, C. J.; Otto, R.; Continetti, R. E., Spectroscopy and dynamics of the HOCO radical: Insights into the OH + CO → H + CO2 reaction. Phys. Chem. Chem. Phys. 2014, 16 (36), 19091-19105. [14] Dhanya, S.; Maity, D. K.; Upadhyaya, H. P.; Kumar, A.; Naik, P. D.; Saini, R. D., Dynamics of OH formation in photodissociation of pyruvic acid at 193 nm. J. Chem. Phys. 2003, 118 (22), 10093-10100. [15] Mordaunt, D. H.; Osborn, D. L.; Neumark, D. M., Nonstatistical unimolecular dissociation over a barrier. J. Chem. Phys. 1998, 108 (6), 2448-2457. 82 [16] Ruscic, B.; Bross, D., Active Thermochemical Tables (ATcT) values based on ver. 1.122d of the Thermochemical Network (2018); available at ATcT.anl.gov. [17] Maul, C.; Gericke, K.-H., Aspects of photoinduced molecular three-body decay. J. Phys. Chem. A 2000, 104 (12), 2531-2541. [18] Maul, C.; Gericke, K.-H., Photo induced three body decay. Int. Rev. Phys. Chem. 1997, 16 (1), 1-79. [19] Chang, X.-P.; Fang, Q.; Cui, G., Mechanistic photodecarboxylation of pyruvic acid: Excited-state proton transfer and three-state intersection. J. Chem. Phys. 2014, 141 (15), 154311. [20] Yamamoto, S.; Back, R., The photolysis and thermal decomposition of pyruvic acid in the gas phase. Can. J. Chem. 1985, 63 (2), 549-554. [21] Amaral, G.; Xu, K.; Zhang, J., UV photodissociation dynamics of ethyl radical via the Ã 2 A’(3s) state. J. Chem. Phys. 2001, 114 (12), 5164-5169. [22] Song, Y.; Zheng, X.; Zhou, W.; Lucas, M.; Zhang, J., Ultraviolet photodissociation dynamics of the n-propyl and i-propyl radicals. J. Chem. Phys. 2015, 142 (22), 224306. [23] Sun, G.; Song, Y.; Zhang, J., Ultraviolet photodissociation dynamics of 1-pentyl radical. Chin. J. Chem. Phys 2018, 31 (4), 439. [24] North, S. W.; Blank, D. A.; Gezelter, J. D.; Longfellow, C. A.; Lee, Y. T., Evidence for stepwise dissociation dynamics in acetone at 248 and 193 nm. J. Chem. Phys. 1995, 102 (11), 4447-4460. [25] North, S. W.; Marr, A. J.; Furlan, A.; Hall, G. E., Nonintuitive asymmetry in the three-body photodissociation of CH3COCN. J. Phys. Chem. A 1997, 101 (49), 9224-9232. [26] Tsai, P.-Y.; Chao, M.-H.; Kasai, T.; Lin, K.-C.; Lombardi, A.; Palazzetti, F.; Aquilanti, V., Roads leading to roam. Role of triple fragmentation and of conical intersections in photochemical reactions: experiments and theory on methyl formate. Phys. Chem. Chem. Phys. 2014, 16 (7), 2854-2865. [27] Hobday, N.; Quinn, M. S.; Nauta, K.; Andrews, D. U.; Jordan, M. J.; Kable, S. H., Experimental and theoretical investigation of triple fragmentation in the photodissociation dynamics of H2CO. J. Phys. Chem. A 2013, 117 (46), 12091-12103. [28] Morajkar, P.; Bossolasco, A.; Schoemaecker, C.; Fittschen, C., Photolysis of CH3CHO at 248 nm: Evidence of triple fragmentation from primary quantum yield of CH3 and HCO radicals and H atoms. J. Chem. Phys. 2014, 140 (21), 214308. 83 [29] Osamura, Y.; Schaefer III, H. F.; Dupuis, M.; Lester Jr, W. A., A unimolecular reaction ABC → A + B + C involving three product molecules and a single transition state. Photodissociation of glyoxal: HCOHCO → H2 + CO + CO. J. Chem. Phys. 1981, 75 (12), 5828-5836. [30] Scuseria, G. E.; Schaefer III, H. F., The unimolecular triple dissociation of glyoxal: transition-state structures optimized by configuration interaction and coupled cluster methods. J. Am. Chem. Soc. 1989, 111 (20), 7761-7765. [31] Bagchi, A.; Dyakov, Y. A.; Ni, C.-K., Photodissociation and photoionization of 2,5- dihydroxybenzoic acid at 193 and 355 nm. J. Chem. Phys. 2010, 133 (24), 244309. [32] Deshmukh, S.; Hess, W. P., Photodissociation of acetyl chloride: Cl and CH3 quantum yields and energy distributions. J. Chem. Phys. 1994, 100 (9), 6429-6433. [33] de Wit, G.; Heazlewood, B.; Quinn, M.; Maccarone, A.; Nauta, K.; Reid, S.; Jordan, M.; Kable, S., Product state and speed distributions in photochemical triple fragmentations. Faraday Discuss. 2012, 157, 227-241. [34] Kim, S. K.; Pedersen, S.; Zewail, A. H., Direct femtosecond observation of the transient intermediate in the α‐cleavage reaction of (CH3)2CO to 2CH3 + CO: Resolving the issue of concertedness. J. Chem. Phys. 1995, 103 (1), 477-480. [35] Zhong, Q.; Poth, L.; Castleman Jr, A., Ultrafast dissociation dynamics of acetone: A revisit to the S 1 state and 3s Rydberg state. J. Chem. Phys. 1999, 110 (1), 192-196. [36] Chin, C.-H.; Lee, S.-H., Comparison of two-body and three-body decomposition of ethanedial, propanal, propenal, n-butane, 1-butene, and 1, 3-butadiene. J. Chem. Phys. 2012, 136 (2), 024308. [37] Hepburn, J.; Buss, R.; Butler, L.; Lee, Y.-T., Molecular beam study of the photochemistry of S1 glyoxal. J. Phys. Chem. 1983, 87 (19), 3638-3641. [38] Hunnicutt, S. S.; Waits, L. D.; Guest, J. A., 1 (n,π*)-Photochemistry of acetic acid at 200 nm: further evidence for an exit channel barrier and reaction selectivity. J. Phys. Chem. 1991, 95 (2), 562-570. [39] Reed Harris, A. E.; Cazaunau, M.; Gratien, A.; Pangui, E.; Doussin, J.-F.; Vaida, V., Atmospheric Simulation Chamber Studies of the Gas-Phase Photolysis of Pyruvic Acid. J. Phys. Chem. A 2017, 121 (44), 8348-8358. [40] Reed Harris, A. E.; Pajunoja, A.; Cazaunau, M.; Gratien, A.; Pangui, E.; Monod, A.; Griffith, E. C.; Virtanen, A.; Doussin, J.-F.; Vaida, V., Multiphase photochemistry of pyruvic acid under atmospheric conditions. J. Phys. Chem. A 2017, 121 (18), 3327-3339. 84 [41] Vesley, G. F.; Leermakers, P. A., The photochemistry of α-keto acids and α-keto esters. III. Photolysis of pyruvic acid in the vapor phase. J. Phys. Chem. 1964, 68 (8), 2364- 2366. [42] Rosenfeld, R. N.; Weiner, B., Energy disposal in the photofragmentation of pyruvic acid in the gas phase. J. Am. Chem. Soc. 1983, 105 (11), 3485-3488. 85 Pyruvic acid photodissociation at 193 nm Introduction Based on the electronic structure calculations reported in Table 3.1, the lowest energy observed transition of PA, S1←S0, is of π*←n+ character. The observed origin band of this perpendicular transition lies at 26,710 cm -1 and the band system extends to about 32,000 cm -1 .[1-4] The next absorption system, S2←S0, is of π*←n− character and can be reached via a perpendicular transition at a calculated vertical energy of 45,160 cm -1 .[2,5] Excitation to S3 is of π*←π character and is reached via a parallel transition at a calculated vertical energy of 56,500 cm -1 . The S3←S0 transition has been implicated in 193 nm dissociation, because its oscillator strength, 7.0×10 −2 , is much larger than that for excitation to S2 (1.6×10 −4 ) or S1 (6.0×10 −5 ).[2] Very little information exists on the 193 nm photodissociation of PA. In spite of the high excitation energy reached (~51,800 cm -1 ), the only products identified to date with certainty are CH4, CO2 and OH, but the mechanisms involved in their formation are unknown.[3,6-9] Flynn and coworkers used time domain infrared (IR) absorption 86 spectroscopy with a tunable diode laser to detect vibrationally excited CO2 at about 3 Torr partial pressure.[7] They initially identified CO2 with up to 4 quanta of bending excitation and one quantum of asymmetric stretch but concluded that 97% of the CO2 products were formed in the ground vibrational state. (In later work, Flynn and coworkers speculated that this large fraction of CO2(0,0,0) was perhaps a spurious result.)[8] They proposed that direct decarboxylation of PA was not the only pathway leading to CO2 formation, and invoked pathways in which CO2 is a secondary dissociation product. In the latter study they used time-resolved FTIR spectroscopy to observe IR emission from CO2 in flowing samples of 30 mTorr CO2 in Ar and followed the emission for 25 µs.[8] Rosenfeld and Wiener also observed IR fluorescence from CO2 when irradiating PA at 193 nm in a cell.[9] Preliminary results regarding CO and CH4 products have been reported in the IR studies.[10] In separate experiments, Dhanya et al. excited 40 mTorr of PA at 193 nm and, using laser induced fluorescence (LIF), observed OH products in rotational levels N up to 6.[3] The OH rotational temperature was 635 K, and from the Doppler profiles, the average translational energy was estimated at 18.7 kcal mol -1 (~6500 cm -1 ). Surprisingly, the appearance time of the OH radical was long (660 ns). The authors proposed that C-OH bond fission in PA was the main source of OH, rather than secondary dissociation of HOCO, though they could not rule out dissociation of MHC as a source of OH and did not consider other possibilities. They suggested that excitation to S3 is rapidly followed by couplings to the lower-lying S1, T2, and T1 states, and that the long appearance time of OH is due to slow 87 dissociation on T1. Their calculations reveal a late barrier of 14.6 kcal mol -1 for C-OH dissociation on T1.[3] Also relevant to the study reported here are those in the previous chapter which reports the first molecular beam photodissociation study of PA, exploiting VMI and TOF mass spectrometry to detect nascent products.[2] We discovered an unexpectedly facile 2-photon transition in PA via S1 to S2 from which dissociation occurs. Our electronic structure calculations show that the oscillator strength of the S2←S1 transition (7.0×10 −2 ) is more than two orders of magnitude larger than that for S1←S0, while the S3←S1 oscillator strength is much smaller (2.8×10 −5 ). We observed H, HOCO, CH3CO, CO and CH3 as nascent dissociation products following 2-photon excitation, and KER distributions of these fragments were calculated from speed distributions recorded by VMI. The observed anisotropic angular distributions of the products indicate that dissociation is faster than the rotational period of PA. Several of the product KER distributions appeared multimodal, suggesting that more than one pathway led to their formation.[2] Therefore, the feasibility of energetically allowed two- and three-body fragmentation pathways were analyzed using the treatment developed by Maul and Gericke.[11] These interpretations relied on calculations of the maximum allowed KER, KEmax, for each product for synchronous and sequential three-body fragmentation pathways, and comparisons of these maximum values to the experimental ones and the KEmax values for two-body fragmentation. The KER distributions of CH3CO and HOCO fragments, for example, implied internal energies that far exceed the values required for their further dissociation. Considering the two-photon energy of >54,000 cm -1 in these experiments, it is hardly surprising that three-body fragmentation pathways are important. 88 In the experiments described herein, both the time-sliced VMI instrument at USC and the Sandia MPIMS setup were used. The latter method can distinguish unimolecular from bimolecular photoproducts, and can provide quantitative determination of product yields and branching ratios. Experiments with partially deuterated PA aid in interpretations. The results provide the first comprehensive description of the dissociation pathways of PA following 193 nm excitation. Although the MPIMS experiments are carried out at higher temperature and pressure than the molecular beam experiments, comparisons of the results obtained by the two methods allow us to propose dissociation mechanisms that describe the Figure 4.1 The results in this chapter indicate that following 193 nm excitation, PA can undergo several types of reaction. The branching ratios are dictated by the nature of the electronic excited states 89 observed products and their yields. Considering the high excitation energy of 51,800 cm -1 and the low dissociation energy of PA, it is not surprising that many reaction channels are energetically open, consistent with the large number of observed products. Even though we had to use some simplification in the data analysis, we were able to determine the yields of major products in a self-consistent way and assess quantitatively the contributions of different dissociation pathways. This approach enabled us to propose a global dissociation scheme for the 193 nm photodissociation of PA. Analyses of absolute product yields and branching ratios show, for example, that the decarboxylation channel leading to stable CO2 + C2H4O products is minor, whereas many observed products result from three-body fragmentation processes. The measurements also indicate that dissociation initiated from the S3 state is fast, and experiments with d1-PA exhibit little isotope exchange. Experimental section VMI experiments The experimental arrangement used in the imaging studies has been described before,[2] and is briefly summarized here with emphasis on details specific to the present experiments. He gas (1300 Torr) is flown through a glass bubbler containing doubly distilled PA liquid (98%, Sigma-Aldrich), and the seeded gas mixture is supersonically expanded through a piezoelectrically driven pulsed nozzle in the source chamber of the instrument. PA in the skimmed molecular beam is excited by the 193 nm output of an ArF laser (GAM EX100) operated at 10 Hz. The unpolarized output is reflected off a quartz window at Brewster angle 90 resulting in a s-polarized radiation. After passing through multiple irises, the output (~0.2 mJ) is focused by a 30 cm focal length (f.l.) lens onto the molecular beam at the center of the differentially pumped detection chamber. The laser fluence in these experiments was about 2×10 3 mJ cm -2 . H and CO photofragments are detected by TOF mass spectrometry and VMI using REMPI. H photofragments are detected by 1+1′ REMPI via the Lyman-α transition. The first photon at 121.567 nm results from frequency tripling ∼365 nm radiation (~3 mJ) generated by the frequency doubled output of a dye laser (532 pumped Continuum ND6000; LDS 722 dye), and focusing the beam by a 30 cm f.l. lens into a gas cell containing a mixture of Kr and Ar in the ratio 200:590 Torr. The generated VUV radiation and residual 365 nm beams are focused at the molecular beam (confocal with 193 nm) by using a MgF2 lens (f.l.=7.5 cm). CO (X 1 Σ + , v=0,1) products are detected by 2+1 REMPI at 230.1−230.5 nm via the B 1 Σ + ←X 1 Σ + transition.[12] The UV focused radiation (~0.3 mJ, 30 cm f.l. lens) is generated by mixing 355 nm and 653−655 nm light (532 pumped Continuum ND6000; DCM dye) in a BBO crystal. Ions produced in the detection region are accelerated through a series of electrostatic lenses toward a position sensitive detector (40 mm diameter double-stack microchannel plates (MCP)) coupled to a phosphor screen (Galileo ElectroOptics 3040FM series). A one- megapixel camera behind the phosphor screen records ion hit events on the detector. Velocity map images of the central slice of the ion spheres are obtained by fast gating of the detector using a home-built 2kV pulse generator (5 ns fwhm).[13-14] 91 MPIMS experiments The MPIMS apparatus, developed by Osborn and coworkers, was employed in these experiments at the ALS of the Lawrence Berkeley National Laboratory. The MPIMS apparatus has been described in detail in previous publications and also in Chapter 2 and is briefly reviewed here.[15-16] He gas flows through a doubly distilled sample of PA in a glass bulb maintained at 21°C and a total pressure of 35 Torr. Deuterated PA samples (d1-PA) are synthesized by first preparing a slurry of 3 g of sodium pyruvate (CH3COCOONa) in 2 ml D2O and then adding 0.6 ml of D2SO4 dropwise just before using the sample. This yields 78% d1-PA confirmed by the MPIMS mass spectrum. The PA(d1-PA)/He mixture is further diluted to roughly 0.1% PA in He in the reactor tube (1.05 cm i.d., 62.5 cm long). Typical experimental flow conditions inside the reactor tube (4 Torr at 250 sccm flow) yield a total number density of ⁓1.3×10 17 molecules cm -3 . Under these conditions the maximum partial pressure of PA(d1-PA) in the reactor tube is 4 mTorr. Photodissociation is achieved by 193 nm radiation generated by an ArF excimer laser operated at 10 Hz. Unfocused laser radiation (~23 mJ cm -2 ) is propagated along the length of the tube parallel to the gas flow direction. The laser fluence was minimized by using wire meshes with different mesh counts at the excimer exit window, or by reducing the operating voltage, or a combination of the two. The quartz tube is coated with halocarbon wax to suppress wall loss of reactive radicals. Reactants, intermediates, and products are sampled through a ⁓650 μm diameter orifice in the tube and the resulting molecular beam is 92 intersected by tunable VUV synchrotron radiation at the Chemical Dynamics Beamline of the ALS synchrotron. The cations produced by VUV ionization are analyzed with an orthogonal TOF mass spectrometer coupled to a time-sensitive MCP detector. Xe + ions formed by autoionization following 8s←5p atomic absorption in Xe are used to calibrate the VUV photon energy. The calibration gas mixture contains trace amounts of ethylene, propene, and 1-butene, which are used for mass calibration. TOF data from characterized molecular products are used to characterize mass accuracy and sensitivity. Using the MPIMS apparatus we obtain time- and mass-resolved PI spectra that provide ion intensity as a function of m/z ratio, kinetic time, and VUV photon energy. We signal average at selected photon energies to obtain ion intensity as a function of m/z and kinetic time. The experimental PI spectra are normalized for changes in photon-flux as a function of VUV energy using the output of SXUV-100 Photodiode (International Radiation Detectors, Inc.) that intercepts the VUV beam after the ionization point. Normalization is needed for relative and absolute determinations of ionization cross-sections obtained at different photon energies. Results and Analysis Molecular beam studies using VMI We determined the KER distribution of H fragments generated by 193 nm (51,800 cm -1 ) photodissociation of PA from its speed distribution derived from the VMI image. Because there is more than one pathway that can generate each fragment, we report the 93 results as KER distribution of that fragment, rather than the customary c.m. KER distribution.[2] The H-fragment KER distribution shown in Figure 4.2 appears rather similar at low kinetic energies to that determined in our previous studies using 2-photon excitation at 374 nm (~53,500 cm -1 excitation energy). In contrast, however, it does not show a clear bimodal structure. Figure 4.2 H-fragment KER distribution (one-laser background subtracted) obtained at 193 nm (black; solid) compared to the previously reported distribution obtained for H- fragments following 2-photon photodissociation of PA with excitation wavelength 374 nm (red; dashed).[2] The predicted maximum KER values for several possible synchronous three-body fragmentation (blue) and two-body fragmentation (green) reactions are marked by arrows. At 193 nm we observe a broad hump consisting of a peak around 2000 cm -1 with a small shoulder at ~5000 cm -1 . The observed maximum KER of the H fragments is <12,000 cm -1 . The noisy features extending from 12,500 – 25,000 cm -1 are a result of background 94 subtraction of the fairly large 193-nm-only signal. The higher KE feature appearing following 2-photon excitation, which ended at ~25,000 cm -1 , is largely missing with 193 nm excitation. The recoil anisotropy parameter, β, is zero, indicating that the H distribution is isotropic. In contrast, 2-photon excitation at 374 nm results in an anisotropic angular distribution for all kinetic energies, reaching β = −0.6 at KE > 10,000 cm -1 .[2] In Figure 4.3 we display the CO-fragment KER distributions obtained by monitoring CO(v=0) and CO(v=1) at their rotational bandheads. The KER distribution of CO(v=0) is similar to the one obtained in the previous 2-photon studies;[2] it peaks at ~700 cm -1 and has a maximum KER at ~8000 cm -1 . Its angular distribution is isotropic (β=0). The CO(v=1) distribution is remarkably different. It peaks at ~3800 cm -1 , and is isotropic at low KER, but becomes strongly anisotropic at higher kinetic energies, reaching β≈1, as seen in the bottom panel of Figure 4.3. 95 Figure 4.3 KER distributions (top) and corresponding recoil anisotropy parameters β (bottom) of CO fragments in v=0 (black) and v=1 (red). Low pressure flow reactor studies using MPIMS 4.3.2.I PA photodissociation at 193 nm PA ionizes at photon energies above ~10 eV. Therefore, the kinetic time trace of m/z = 88.02 is proportional to the concentration of PA in the flow reactor before and after 193 nm photodissociation. Figure 4.4 shows clearly the decrease in the PA ion signal intensity, using 10.25 eV ionizing radiation, upon photodissociation at 193 nm at a fluence of ~23 mJ cm -2 . The fraction of photodissociated or depleted PA is calculated by using 96 𝑓𝑓 𝑑𝑑 𝑠𝑠𝑠𝑠 𝑑𝑑 𝑠𝑠𝑠𝑠 𝑠𝑠 𝑑𝑑 = 𝑁𝑁 𝑑𝑑 𝑠𝑠𝑠𝑠 𝑑𝑑 𝑠𝑠𝑠𝑠 𝑠𝑠𝑑𝑑 𝑃𝑃 𝑃𝑃 𝑁𝑁 𝑃𝑃 𝑃𝑃 = �1 − 𝑚𝑚 𝑚𝑚𝑚𝑚𝑚𝑚 ( 2 𝑡𝑡𝑡𝑡 12 𝑚𝑚𝑚𝑚 ) 𝑚𝑚 𝑚𝑚𝑚𝑚𝑚𝑚 ( −10 𝑡𝑡𝑡𝑡 − 0.5 𝑚𝑚𝑚𝑚 ) �. (4.1) An average 𝑓𝑓 𝑑𝑑 𝑠𝑠𝑠𝑠 𝑑𝑑 𝑠𝑠𝑠𝑠 𝑠𝑠 𝑑𝑑 value of 0.12 ± 0.01 is obtained under these conditions. The 𝑓𝑓 𝑑𝑑 𝑠𝑠𝑠𝑠 𝑑𝑑 𝑠𝑠𝑠𝑠 𝑠𝑠𝑑𝑑 value depends linearly on the 193 nm fluence, which confirms that PA undergoes one-photon absorption at this wavelength. Figure 4.4 Kinetic time trace of m/z = 88.02 showing the depletion in the PA ion signal upon 193 nm photolysis. The pre- and post-photolysis signal averages are represented by blue and green solid lines, respectively. The depletion of 12% is consistent, within experimental uncertainties, with previous estimates of the absorption cross-section of PA at 193 nm. Assuming a dissociation quantum yield of unity, we arrive at an absorption cross-section (base e) of (5.4 ± 1.8)×10 -18 cm 2 (approximate absorption coefficient 0.18 ± 0.06 cm -1 Torr -1 ). Wood et al. had estimated a 193 nm PA absorption coefficient of 0.05 ± 0.015 cm -1 Torr -1 in a static gas cell.[6] 97 4.3.2.II Product identification The 3D capabilities (kinetic time, photon energy, m/z) of the MPIMS instrument allow us to detect products with high sensitivity and identify them conclusively by their PI spectra and kinetic time traces. The full range TOF mass spectrum following photodissociation shows no peaks above m/z = 88 or below m/z = 14. The pre-photodissociation signal has been subtracted from the two representative TOF mass spectra shown in Figure 4.5. Positive peaks are observed at m/z = 14, 15, 16, 17, 18, 26, 28, 29, 30, 42, 43, 44, 45, 58, 60, 72, 74, 86 and 88. Because PA is depleted by the 193 nm laser, parent (m/z = 88.02) and daughter ions (e.g., m/z = 43.02) generate negative peaks after background subtraction. Positive peaks are primary photoproducts or secondary reaction products. Based on their m/z values, we identified the molecular formula of products formed in the 193 nm photodissociation of PA. PI spectra enable us to identify specific molecular species and their isomers, some of which are labelled in Figure 4.5. Figure 4.5(a) shows mass spectra obtained by integrating the signal over 8.9–9.2 eV photon energies, highlighting radical products with ionization energies (IE) < 9.2 eV, e.g., formyl (HCO), acetyl (CH3CO), and hydroxyformyl (HOCO). The acetyl radical, observed at photon energies below 10 eV, has a time trace characteristic of a reactive species (see Figure 4.10 below) and is distinguished from the m/z = 43.02 ion produced by dissociative ionization of PA at photon energies above 10.25 eV, whose time trace is identical to that of PA (Figure 4.4). At higher photon energies (10–10.7 eV) we obtained the spectrum shown in Figure 4.5(b) and identified stable products, such as ketene, vinyl alcohol, and acetaldehyde. 98 As described earlier, in previous 193 nm photodissociation studies, CO2 was a predominant product, and low yields of CH4 and OH were also observed.[3,10] Wood et al.[6] measured a quantum yield for CO2 production of 1.03 ± 0.2. We observe a surprising number of additional products that we attribute to a variety of photodissociation pathways. Figure 4.5 Integrated TOF mass spectra in the range 8.9 – 9.2 eV (a), and 10 – 10.7 eV (b). Products that were identified using their kinetic time traces and PI spectra are labelled. Molecular species and their isomers were identified by comparing their PI spectra to previously published results. Several examples are shown in Figure 4.6. As an example of how this information may provide new insights, consider the following case. The MPIMS 99 instrument’s mass resolution of m/Δm ≈ 1500 allows for the distinction between CO2 (43.99) and C2H4O (44.03), and the PI spectra of these two peaks confirm this assignment. As seen in Figure 4.6(d), the PI energy scan of m/z = 44 at 9–10.3 eV identifies the two stable isomers of MHC — acetaldehyde and vinyl alcohol (m/z = 44.03). This observation contrasts with earlier studies, which reported only acetaldehyde (the most stable C2H4O isomer) as the observed co-fragment in the decarboxylation pathway.[5,9,17-21] Therefore vinyl alcohol, which we observe for the first time in PA photodissociation, must be kinetically favored. Indeed, on the C2H4O potential energy surface, the isomerization barrier for MHC → vinyl alcohol is ~6 kcal mol -1 lower than the barrier for MHC → acetaldehyde.[22] This observation serves as indirect evidence for the formation of MHC as a primary product that isomerizes quickly to the more stable structural isomers. Figure 4.6 also conclusively identifies ketene (H2C=C=O), methyl (CH3) and OH radicals as photoproducts of PA photodissociation. In a similar way we confirm isomeric assignments of other products by their PI spectra. 100 Figure 4.6 PI spectra of (a) m/z = 17; (b) m/z = 42; (c) m/z = 15; and (d) m/z = 44 observed in PA photodissociation (red), compared to published PI spectra (black) of the hydroxy radical,[23] ketene,[24] methyl radical,[25] vinyl alcohol, and acetaldehyde.[26] In order to further characterize the dissociation mechanisms leading to the observed primary products we repeated the measurements with partially deuterated PA (the mass spectrum of this sample confirmed its composition as 78% d1-PA, 13% d2-PA, ~3% d3-PA and d4-PA, and <3% d0-PA). Figure 4.7 shows a comparison of the early time TOF spectra of the non-deuterated sample (d0-PA) and the deuterated sample (mainly d1-PA). We also calculated the percent deuteration in each product, defined as the percentage ratio of a single 101 isotopologue to the sum of all isotopologues of the product. We observed 78% d0-acetyl, 13% d1-acetyl, more than 90 % DOCO and almost 100% OD products, confirming both that our synthesis produced mainly CH3COCOOD (as expected based on the starting salt), and that there is little isotopic scrambling during photodissociation. The OD product could be distinguished from the H2O background based on its time-trace and PI spectrum (see Figure 4.7). We observe 76% d1-methane and 80% d1-vinyl alcohol products, which preserve all H/D atoms of d1-PA(78%). We also observe 19% d1-ketene, 30% DCO and 31% d1-methyl, which show a propensity to form their d0 isotopologues. 102 Figure 4.7 Early time (0–3 ms) TOF mass spectrum highlighting short-lived photoproducts of deuterated PA (red) and PA (blue) for (a) m/z=28–47 and (b) m/z=15–19. The negative scale for the PA mass spectrum is used for ease of comparison. m/z=18 from d1-PA identified as OD from its (c) kinetic time trace which resembles that of an unstable product and (d) its ionization spectrum which matches the previously reported absolute PI spectrum for OH radical.[23] Some products we observe are not primary photoproducts of PA photodissociation. We can distinguish primary photoproducts from secondary (bimolecular) reaction products based on kinetic time traces. Secondary products, e.g., acetone (58), methyl glyoxal (72), and glyoxylic acid (74), can arise from bimolecular reactions between two radical species or from 103 a radical species reacting with PA. Assuming that photodissociation is faster than the 15 nanosecond laser pulse, primary photodissociation products should appear with the 480 µs instrument response time (10 – 90% rise time) of the instrument. The instrument response has been characterized by the rise-time of ketene (m/z = 42), a prompt product in the 193 nm photodissociation of acetone,[27] as described below. Products of bimolecular reactions have slower rise times than the instrument response time (see Figure 4.8(a)). The appearance times of some products, e.g. H2O and H2CO, have an initial fast rise of <1 ms followed by a slower component, indicating that they originate in both unimolecular and bimolecular reactions. In this chapter, we focus on primary photoproducts of PA dissociation at 193 nm and their formation pathways. The instrument response function is estimated based on the rise time of ketene photoproducts from 193 nm dissociation of acetone. Toulson et al. observed that ketene production from acetone is complete within the 37 ps instrument response time.[27] Because of its short appearance time after photolysis relative to the nanosecond laser pulse, the formation of ketene can be considered as instantaneous compared to the rise time of our instrument (several hundred microseconds). Figure 4.8(b) shows the observed rise time of the ketene photoproduct after photolysis at time t=0. Assuming a heavyside step function as the input current, the response time of the instrument is characterized by the commonly- used 10% to 90% rise time of the observed signal. We obtain a 480 µs response time for our instrument. 104 Figure 4.8 (a) Kinetic time traces of m/z = 42, 44 and 58 compared to the instrument response time. The bimolecular reaction product (m/z = 58) has a slower rise time than that of primary photoproducts. (b) The kinetic time trace (red) of ketene photoproduct from acetone photodissociation is shown. The 10% to 90% signal rise is shown by the shaded region and the corresponding response time period is marked by arrows. Determination of product yields A main goal of this chapter is to quantify the yields of primary photodissociation products. We will use this information to determine which dissociation channels are consistent with the individual products we detect. This section describes the procedure for calculating product yields. Because the signal strength of a molecular ion is proportional to its concentration and ionization cross-section, we can use equation 4.2 to calculate the concentration of photoproducts compared to a reference molecule. This comparison needs to be done with careful attention to variations in sensitivity of the instrument with mass.[25] 105 � 𝜎𝜎 𝑢𝑢𝑢𝑢𝑢𝑢 𝑖𝑖 𝑖𝑖 𝑢𝑢 (𝜖𝜖 ) 𝜎𝜎 𝑟𝑟 𝑠𝑠 𝑟𝑟 𝑖𝑖 𝑖𝑖 𝑢𝑢 (𝜖𝜖 ) � = � 𝑆𝑆 𝑢𝑢𝑢𝑢𝑢𝑢 (𝜖𝜖 ) 𝑆𝑆 𝑟𝑟 𝑠𝑠 𝑟𝑟 (𝜖𝜖 ) � � 𝑁𝑁 𝑟𝑟 𝑠𝑠 𝑟𝑟 𝑁𝑁 𝑢𝑢𝑢𝑢𝑢𝑢 � � ∝ 𝑟𝑟 𝑠𝑠 𝑟𝑟 ∝ 𝑢𝑢𝑢𝑢𝑢𝑢 �. (4.2) Here 𝜎𝜎 𝑢𝑢𝑢𝑢𝑢𝑢 𝑖𝑖 𝑖𝑖 𝑢𝑢 (𝜖𝜖 ) and 𝜎𝜎 𝑟𝑟 𝑠𝑠 𝑟𝑟 𝑖𝑖 𝑖𝑖 𝑢𝑢 (𝜖𝜖 ) are the absolute ionization cross-sections of the unknown and reference molecule at the reference photon energy 𝜖𝜖 , Ni are the molar concentrations of the two species, and 𝑆𝑆 𝑢𝑢𝑢𝑢𝑢𝑢 (𝜖𝜖 ) and 𝑆𝑆 𝑟𝑟 𝑠𝑠 𝑟𝑟 (𝜖𝜖 ) are the spectra of measured signals of the unknown and reference molecules. ∝ 𝑟𝑟𝑟𝑟𝑟𝑟 ∝ 𝑢𝑢 𝑢𝑢 𝑢𝑢 represents the ratio of their mass discrimination factors[25]. We obtain � ∝ 𝑟𝑟𝑟𝑟𝑟𝑟 ∝ 𝑢𝑢 𝑢𝑢 𝑢𝑢 � = � 𝑚𝑚 𝑟𝑟𝑟𝑟𝑟𝑟 𝑚𝑚 𝑢𝑢𝑢𝑢𝑢𝑢 � 0. 5 3 . To determine the mass discrimination factor, which is a measure of the relative ion detection efficiency of the instrument, we used a calibration gas consisting of a mixture of 1000 ppm Kr, 1000 ppm Xe, 0.50% Ar, 1.00% CH4, 10.00% H2 with balance of He. We obtained a TOF spectrum of the calibration mixture using 19 eV ionizing radiation. From this spectrum, the integrated peak intensities, 𝑆𝑆 𝑖𝑖 , for H2, 40 Ar, 82 Kr, 83 Kr, 84 Kr, 86 Kr, 129 Xe, 131 Xe, 132 Xe, 134 Xe and 136 Xe were obtained. The ionization cross-sections, σi, for each of these species at 19 eV can be found in the results summarized by Samson and Stolte.[28] The concentration of each species in the calibration mixture is scaled by their isotope ratios to obtain their individual concentrations, 𝑁𝑁 𝑖𝑖 . Using the above quantities, the signal for each species is divided by its concentration and ionization cross-section to calculate an adjusted signal, 𝑆𝑆 𝑖𝑖 𝑎𝑎 𝑑𝑑 𝑎𝑎 , which is plotted against its mass, 𝑚𝑚 𝑖𝑖 . If there were no mass discrimination, this plot would be a straight line with slope of zero. Savee et al. argued that the mass discrimination factor[25] can be described by the 106 functional form 𝛼𝛼 𝑖𝑖 = 𝑚𝑚 𝑖𝑖 𝑢𝑢 . The data points are fit to a curve described by 𝑆𝑆 𝑖𝑖 𝑎𝑎 𝑑𝑑 𝑎𝑎 = 𝐴𝐴 ∙ 𝑚𝑚 𝑖𝑖 𝑢𝑢 as shown in Figure 4.9(a). For the present experiment a value on 𝑚𝑚 = (0.53 ± 0.03) is obtained. Figure 4.9 (a) Plot of the ion detection efficiency vs. mass for the MPIMS setup. The mass discrimination factor is extracted from a power function fit. (b) The absolute PI cross section of PA from threshold to 14 eV (black; circles), with uncertainty (gray; shaded) shown. Absolute photoionization cross-sections of (c) CO and (d) CO2 from threshold to around 14 eV obtained at 0.005 eV steps. We calculated the mole fraction yields of photoproducts by measuring their concentration ratios relative to PA depletion. To use PA as a reference, we measured its absolute PI cross-section from 9.5 eV to 14 eV; the absolute PI spectrum is displayed in Figure 4.9(b) and tabulated in the appendices. For calibration, we flow measured quantities of both the PA/He mixture and a reference gas (with known mole fractions of ethene, 107 propene and 1-butene) into the instrument and scale the PI spectrum of PA with respect to propene, which has a known absolute PI cross-section spectrum.[29] The concentration of PA used for this calibration is estimated using its vapor pressure at 21.5 °C, calculated at 1 Torr from the Antoine equation parameters.[30] The ionization cross section appears nearly constant at 11–14 eV with a value of about 3.5×10 -18 cm 2 . The uncertainty in this value is ~25 %. We also determined the absolute PI cross-sections around 14 eV for CO and CO2 (see Figure 4.9(c) and (d)). We calibrated the CO and CO2 PI spectra using Xe as the reference gas. The known ionization cross-sections and measured signals of the photoproducts are used in equation 4.2 to calculate concentrations relative to PA. In order to determine a photoproduct’s nascent signal, 𝑆𝑆 𝑠𝑠 𝑟𝑟𝑖𝑖 𝑑𝑑 𝑢𝑢 𝑝𝑝 𝑠𝑠 (𝜖𝜖 ), its kinetic time trace is extrapolated to t = 0. For stable products the signal is estimated using the instrument response function (see Figure 4.8). The instrument response function is fit to the first 5–10 ms to extract only the primary photodissociation component and discard any contribution from secondary bimolecular reactions. The PA signal is calculated as the mean signal intensity from −10 – −1 ms (Figure 4.4). Because the signals from reactive radical species decrease as they are consumed by wall collisions and/or secondary reactions, their 𝑆𝑆 𝑠𝑠 𝑟𝑟𝑖𝑖 𝑑𝑑 𝑢𝑢 𝑝𝑝 𝑠𝑠 (𝜖𝜖 ) values at t = 0 are obtained from a fit to a first- and second-order decay kinetics profile. Equation 4.3, which describes the loss processes due to a combination of first and second order kinetics, is used to fit the time- traces 𝑆𝑆 𝑖𝑖 (𝑡𝑡 ) for species i.[31] 108 𝑆𝑆 𝑖𝑖 (𝑡𝑡 ) = 𝑘𝑘 1 𝑆𝑆 𝑖𝑖 (𝑡𝑡 = 0) (2 𝑘𝑘 ′ + 𝑘𝑘 1 )𝑚𝑚 𝑢𝑢 1 ∙ 𝑠𝑠 − 2 𝑘𝑘 ′ . (4.3) Here 𝑆𝑆 𝑖𝑖 (𝑡𝑡 = 0) is the desired extrapolated signal to time zero, 𝑘𝑘 1 is the rate constant representing the sum of all first order losses (including wall-loss), and 𝑘𝑘 ′ = ∑ 𝑘𝑘 2 ∙ [𝑗𝑗 ] 𝑠𝑠 = 0 represents the sum of all second order reactions where 𝑗𝑗 is any species with which the product reacts, including self-reaction. Figure 4.10 shows examples of the fit described by equation 4.3 to the measured time traces of several radical products. Figure 4.10 Kinetic time traces and their decay fits (given by equation 4.3) of (a) methyl radical at 10.25 eV; (b) hydroxyl radical at 13.25 eV; (c) formyl radical at 10.25 eV; and (d) acetyl radical at 9.7 eV. 109 We calculate the concentration ratio of each photoproduct by comparing its nascent signal and photoionization cross-section using the relationship given in equation 4.2 to that of PA pre-photolysis signal. At 14.15 eV, we obtain the concentration ratios 𝑁𝑁 𝐶𝐶𝐶𝐶 2 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ and 𝑁𝑁 𝐶𝐶𝐶𝐶 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ . The ionization energy of PA at 14.15 eV is approximated by the measured value at 14.025 eV. The integrated signal around m/z = 28 at 14.15 eV, however, is found to be a mixture of CO (m/z =27.99) and ethene (m/z =28.03). We obtain the integrated signal intensity of just the CO photoproduct, 𝑆𝑆 𝐶𝐶𝐶𝐶 , by multiplying the total m/z=28 signal by an appropriate scaling factor obtained from Gaussian fits used to resolve the two peaks in the TOF mass spectrum. The concertation ratios 𝑁𝑁 𝑀𝑀𝑠𝑠𝑠𝑠 ℎ 𝑎𝑎 𝑢𝑢𝑠𝑠 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ , 𝑁𝑁 𝐶𝐶 𝑂𝑂 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ , 𝑁𝑁 𝑃𝑃𝑝𝑝 𝑠𝑠𝑠𝑠 𝑖𝑖 𝑝𝑝 𝑃𝑃𝑝𝑝𝑖𝑖 𝑑𝑑 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ , 𝑁𝑁 𝑊𝑊 𝑎𝑎 𝑠𝑠 𝑠𝑠𝑟𝑟 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ , 𝑁𝑁 𝐸𝐸 𝑠𝑠 ℎ 𝑠𝑠𝑢𝑢𝑠𝑠 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ , 𝑁𝑁 𝑃𝑃𝑝𝑝 𝑠𝑠𝑠𝑠 𝑦𝑦 𝑑𝑑 𝑠𝑠𝑢𝑢 𝑠𝑠 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ and 𝑁𝑁 𝐹𝐹 𝑖𝑖 𝑟𝑟𝑚𝑚𝑎𝑎𝑑𝑑 𝑑𝑑𝑠𝑠 ℎ 𝑦𝑦 𝑑𝑑 𝑠𝑠 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ are obtained from nascent photoproduct signals at 13.25 eV.[23,32-36] The integrated signal around m/z = 30 at 13.25 eV is found to be a mixture of formaldehyde (m/z=30.01) and ethane (m/z=30.05). We obtain integrated signal intensity of just the formaldehyde photoproduct, 𝑆𝑆 𝐹𝐹 𝑖𝑖 𝑟𝑟𝑚𝑚𝑎𝑎𝑑𝑑𝑑𝑑𝑠𝑠 ℎ𝑦𝑦 𝑑𝑑𝑠𝑠 , by multiplying the total m/z=30 signal intensity by an appropriate scaling factor obtained from Gaussian fits used to resolve the two peaks in the TOF mass spectrum. The formaldehyde time trace is found to resemble that of a primary photoproduct (instrument-limited rise time), whereas the ethane’s rise time is slower and resembles that of a secondary reaction product. The ethane time trace shows both photolysis and secondary reaction components. The photolysis part is extracted by fitting the instrument response function to the first 5 milliseconds. 110 The cation of acetic acid produced by photodissociation and as the daughter ion of PA are both observed at m/z = 60. Because the observed time trace is a combination of both processes, the photodissociation signal, 𝑆𝑆 𝑃𝑃𝑝𝑝 𝑠𝑠𝑠𝑠 𝑖𝑖 𝑝𝑝 𝑃𝑃𝑝𝑝𝑖𝑖 𝑑𝑑 , is obtained by adding the estimated depleted percentage of PA to the observed signal. The nascent OH radical signal, 𝑆𝑆 𝐶𝐶 𝑂𝑂 , is obtained from the fits shown in Figure 4.10. Similarly, at 10.25 eV we obtain 𝑁𝑁 𝐾𝐾 𝑠𝑠𝑠𝑠 𝑠𝑠𝑢𝑢 𝑠𝑠 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ , 𝑁𝑁 𝑃𝑃𝑝𝑝 𝑠𝑠𝑠𝑠 𝑎𝑎 𝑑𝑑 𝑑𝑑 𝑠𝑠ℎ 𝑦𝑦 𝑑𝑑𝑠𝑠 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ , 𝑁𝑁 𝑀𝑀𝑠𝑠𝑠𝑠 ℎ 𝑦𝑦 𝑑𝑑 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ and 𝑁𝑁 𝑂𝑂 𝐶𝐶𝐶𝐶 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ from nascent photoproduct signals.[24,26,37-38] Calculating the ratio of acetaldehyde to PA at 10.25 eV is a two-step process, because the signal intensity at m/z = 44 is the sum of both the acetaldehyde and vinyl alcohol fragments. First, using the ratio 𝑁𝑁 𝑉𝑉 𝑖𝑖 𝑢𝑢𝑦𝑦 𝑑𝑑 𝑁𝑁 𝐾𝐾 𝑠𝑠𝑠𝑠 𝑠𝑠𝑢𝑢 𝑠𝑠 ⁄ obtained at 9.7 eV, an estimated signal intensity 𝑆𝑆 𝑉𝑉 𝑖𝑖 𝑢𝑢𝑦𝑦 𝑑𝑑 𝑃𝑃𝑑𝑑 𝑝𝑝 𝑖𝑖 ℎ𝑖𝑖 𝑑𝑑 (10.25) is calculated from known ionization cross-sections of vinyl alcohol and ketene at 9.7 eV. This value is subtracted from total m/z=44 signal to get 𝑆𝑆 𝑃𝑃𝑝𝑝 𝑠𝑠𝑠𝑠 𝑎𝑎 𝑑𝑑 𝑑𝑑 𝑠𝑠ℎ 𝑦𝑦 𝑑𝑑 𝑠𝑠 (10.25). At 9.7 eV we also obtain the concentration ratio 𝑁𝑁 𝑃𝑃𝑝𝑝 𝑠𝑠𝑠𝑠 𝑦𝑦 𝑑𝑑 𝑁𝑁 𝑃𝑃 𝑃𝑃 ⁄ (Sheps L., private communication for acetyl cross-section). The nascent signals of the radicals CH3CO, CH3 and HCO are obtained from the exponential fits depicted in Figure 4.10. The calculated concentration ratios are converted into product mole fraction yields, 𝑓𝑓 𝑖𝑖 , by scaling them to the concentration of photodissociated PA, such that for any product 𝑓𝑓 , 𝑓𝑓 𝑖𝑖 = 𝑁𝑁 𝑖𝑖 𝑁𝑁 𝑃𝑃 𝑃𝑃 𝑑𝑑 𝑠𝑠𝑠𝑠 𝑑𝑑 𝑠𝑠𝑠𝑠 𝑠𝑠𝑑𝑑 ⁄ . These results are summarized in Table 4.1. The uncertainty in the 𝑓𝑓 𝑖𝑖 measurements is 50–80%, reflecting mainly uncertainties in the published ionization cross- sections and the signal-to-noise ratios in our measurements. 111 Table 4.1 Observed photoproducts in PA photodissociation, their IEs, mole fraction yields 𝑓𝑓 𝑖𝑖 , and the photon energies employed for mole fraction yield determination. Photoproducts IE (eV) 𝑓𝑓 𝑖𝑖 Photon energy (eV) 1 CO 13.78 0.70 14.15 2 CO2 14.00 0.66 14.15 3 CH4 12.61 0.07 13.25 4 CH3 9.84 0.24 10.25 5 OH 13.02 0.17 13.25 6 CH3COOH 10.65 0.03 13.25 7 HCO 8.12 0.07 10.25 8 CH2CO 9.62 0.11 10.25 9 CH3CHO (Ac) 10.23 0.02 10.25 10 CH3CO 7.00 0.12 9.70 11 CH2CHOH (VA) 9.17 0.01 9.70 12 C2H2 11.4 0.01 13.25 13 H2O 12.62 0.14 13.25 14 H2CO 10.88 0.07 13.25 15 C2H4 10.51 0.11 13.25 16 HOCO 8.48 n/a n/a 17 CH2 10.40 n/a n/a 112 As evident from column 3 of Table 4.1, CO and CO2 are the major stable products but there are significant mole fraction yields of radicals like CH3, OH, and CH3CO. H photoproducts, which were observed in our molecular beam studies, are challenging to observe in the MPIMS experiments and we did not attempt to quantify this product. We also anticipate products like H2 and O( 3 P) based on the photodissociation reactions discussed in section IV. To estimate the contributions of specific dissociation reactions we first analyzed the atomic balance of products, i.e., the percentage of atoms in the dissociated PA that we recovered. For example, the total carbon balance, 𝐶𝐶 𝑎𝑎𝑎𝑎𝑑𝑑𝑎𝑎𝑢𝑢 𝑝𝑝𝑠𝑠 , of the products is defined as 𝐶𝐶 𝑎𝑎𝑎𝑎𝑑𝑑𝑎𝑎𝑢𝑢 𝑝𝑝𝑠𝑠 = � 𝐶𝐶 𝑖𝑖 𝑓𝑓 𝑖𝑖 𝐶𝐶 𝑃𝑃 𝑃𝑃 𝑖𝑖 , (4.4) where 𝐶𝐶 𝑖𝑖 is the number of carbon atoms in product 𝑓𝑓 , and 𝑓𝑓 𝑖𝑖 is its mole fraction yield. The number of carbon atoms in PA is 𝐶𝐶 𝑃𝑃 𝑃𝑃 = 3. We obtain 𝐶𝐶 𝑎𝑎𝑎𝑎𝑑𝑑𝑎𝑎𝑢𝑢 𝑝𝑝𝑠𝑠 = 0.88 for species 1–15 listed in Table 4.1. Similarly, we calculated 𝑂𝑂 𝑎𝑎𝑎𝑎𝑑𝑑𝑎𝑎𝑢𝑢 𝑝𝑝𝑠𝑠 = 0.93 and 𝐻𝐻 𝑎𝑎𝑎𝑎 𝑑𝑑 𝑎𝑎 𝑢𝑢 𝑝𝑝 𝑠𝑠 = 0.74. 113 Table 4.2 Reaction mechanisms, primary two-body steps, and final fragmentation channels, their ΔHrxn values,[2,19,39] and percent contributions. A 193 nm photon supplies 51,800 cm -1 of energy. The percent contribution of each mechanistic class is generated by summing contributions from Table 4.3. Primary step Fragmentation reaction ΔHrxn (cm -1 ) % contribution Decarboxylation CH3COH + CO2 CH3CHO + CO2 − 3200 51 CH2CHOH + CO2 890 CH4 + CO + CO2 − 5020 CH2CO + H2 + CO2 5900 CH3 + HCO +CO2 25800 C2H2 + H2O + CO2 9800 C2H4 + O( 3 P) + CO2 35700 CH2 + H2CO + CO2 34600 CH2 + CO + H2 + CO2 33900 Norrish Type I Bond fission CH3CO + HOCO 28600 27 CH3 + COCOOH CH3 + CO + HOCO 31400 CH3 + CO + H + CO2 32000 CH3COCO + OH CH3 + CO + CO + OH 39000 CH3CO + CO + OH 37600 H fragmentation H + CH2COCOOH H + CH2CO + H + CO2 43800 9 H + CH2CO + HOCO 43200 CH3COCOO + H CH3CO + CO2 + H 29200 Others CH3COOH + CO − 1000 13 CH2CO + CO + H2O 10300 CH2 + CO + CO + H2O 37300 114 Having recovered most of the atomic balance, we consider possible photodissociation channels of PA that can contribute to the observed product yields. In order to simplify our analysis of the contributions of specific product channels, we assigned tentative yields to observed products whose yields could not be calculated, and to expected products that were not observed. For example, with the assumption that HOCO and O are the only missing species contributing to 𝑂𝑂 𝑎𝑎𝑎𝑎𝑑𝑑𝑎𝑎𝑢𝑢 𝑝𝑝𝑠𝑠 , and based on the stoichiometry of the proposed dissociation reactions (see column 3 in Table 4.2), we expect 𝑓𝑓 𝐶𝐶 = 𝑓𝑓 𝐸𝐸 𝑠𝑠 ℎ 𝑠𝑠𝑢𝑢𝑠𝑠 = 0.11, and as a result, the remainder 𝑓𝑓 𝑂𝑂 𝐶𝐶𝐶𝐶𝐶𝐶 = 0.05. We then proceed to assign the remaining 𝑓𝑓 𝐶𝐶 𝑂𝑂 2 , 𝑓𝑓 𝑂𝑂 and 𝑓𝑓 𝑂𝑂 2 values, such that the final 𝐶𝐶 𝑎𝑎𝑎𝑎𝑑𝑑𝑎𝑎𝑢𝑢 𝑝𝑝𝑠𝑠 = 𝑂𝑂 𝑎𝑎𝑎𝑎 𝑑𝑑 𝑎𝑎 𝑢𝑢 𝑝𝑝 𝑠𝑠 = 𝐻𝐻 𝑎𝑎𝑎𝑎 𝑑𝑑 𝑎𝑎 𝑢𝑢 𝑝𝑝 𝑠𝑠 ≈ 1. These tentative assignments helped us in performing the multivariate fit described below that estimated the contribution of each dissociation channel. Importantly, we did not find significant changes to the reaction channel contributions upon the addition of these tentative values beyond the reported uncertainties. With the product mole fraction yields known and total atomic balance achieved, we can now proceed to the next step in the analysis, namely, determination of the energetically feasible two-, three-, and four-body reaction channels that may result in the observed product yields. Multivariate linear regression for pathway contributions To determine the contribution of the fragmentation channels listed in Table 4.3 below to the yields of the detected products (Table 4.1) and those expected from the listed reactions, a multivariate linear regression analysis is set up in the following way. We define 115 𝑌𝑌 = � 𝑌𝑌 1 𝑌𝑌 2 ⋮ 𝑌𝑌 2 0 � , as the column of product yields, where 𝑌𝑌 𝑖𝑖 represents the yield of the i th product. 𝑋𝑋 = � 𝑋𝑋 1 1 … 𝑋𝑋 1 2 0 ⋮ ⋱ ⋮ 𝑋𝑋 2 0 1 … 𝑋𝑋 2 0 2 0 � , is a matrix representation of the dissociation reactions, where each column represents one reaction channel in Table 4.2, and each row represents a distinct product atom or molecule. For the column representing reaction channel j, 𝑋𝑋 𝑖𝑖 𝑎𝑎 = n if 1 mole of PA leads to n moles of product i. For example, for reaction 16 (PA → CH3CO + CO + OH) the 16 th column of X has ones for i = CH3CO, CO, OH and zero for the other elements in the column. The contribution of each reaction (i.e. the reaction coefficient in Table 4.3) is represented by 𝜃𝜃 𝑎𝑎 so that the condition, 𝑌𝑌 𝑖𝑖 = � 𝑋𝑋 𝑖𝑖 𝑎𝑎 𝑎𝑎 𝜃𝜃 𝑎𝑎 + 𝛿𝛿 , (4.5) is satisfied for every species i where we seek to minimize the error parameter, 𝛿𝛿 . The vector of coefficient 𝜃𝜃 represents the contribution of each reaction to product 𝑌𝑌 𝑖𝑖 . For example, 𝜃𝜃 1 5 and 𝜃𝜃 1 6 should sum to 17.1% ( 𝑓𝑓 𝑖𝑖 = 0.17) because they are the only listed reactions that produce OH. This system of equations is solved for every product by minimizing the sum of squared residuals, 𝜖𝜖 , using the normal equation. 116 𝜃𝜃 = (𝑋𝑋′ 𝑋𝑋 ) − 1 𝑋𝑋′ 𝑌𝑌 (4.6) The coefficients 𝜃𝜃 for the possible pathways are included in Table 4.3. The error bars listed in the 3 rd column of Table 4.3 reflect the 50–80% uncertainty in the observed yields listed in Table 4.1. Because we began the fit with the assumption that we recover the total atomic balance from all possible photoproducts – both detected and implicated based on stoichiometry – the total percent contributions sum up to one hundred percent, as required. Although the individual contribution of each dissociation channel may be inaccurate, we find that the reactions can be divided into several distinct groups, which reflect known primary photodissociation steps of carboxylic acids and carbonyls. 117 Table 4.3 Possible reaction pathways and their percent contribution. Fragmentation pathway % contribution 1 Ac + CO2 3.0 ± 1.6 2 VA + CO2 1.7 ± 1.6 3 CH4 + CO + CO2 7.2 ± 2.7 4 CH2CO + H2 + CO2 2.1 ± 1.8 5 C2H2 + H2O + CO2 2.2 ± 2.0 6 C2H4 + O( 3 P) + CO2 10.8 ± 1.6 7 CH2 + H2CO + CO2 8.2 ± 2.1 8 CH2 + CO + H2 + CO2 8.5 ± 1.8 9 CH2CO + CO + H2O 2.5 ± 1.9 10 CH2+ CO + CO + H2O 8.8 ± 2.6 11 CH3CO + HOCO 0.9 ± 1.0 12 CH3CO + CO2 + H 4.0 ± 1.6 13 CH3 + CO + HOCO 2.1 ± 1.6 14 CH3 + CO + H + CO2 5.2 ± 2.3 15 CH3 + CO + CO + OH 8.7 ± 1.9 16 CH3CO + CO + OH 7.5 ± 1.8 17 H + CH2CO + HOCO 1.7 ± 1.6 18 H + CH2CO + CO2 + H 4.7 ± 1.5 19 CH3COOH + CO 2.5 ± 1.7 20 CH3 + HCO + CO2 7.8 ± 1.9 118 Discussion Excited States of PA The electronic structure calculations reported in Chapter 3 indicate that direct excitation of PA at 193 nm accesses the S3 state via a parallel transition centered on the carbonyl CO group. This assignment is based on calculations of the oscillator strengths and directions of transition dipole moments for transitions to S1-S3,[2] and is in agreement with previous work.[3,8] The S3 state is reached via a π*←π transition, where the HOMO is localized on the ketonic carbonyl and the LUMO is delocalized on both the ketonic and acidic carbonyls. We have shown that the S2 state, which cannot be reached efficiently by 1-photon excitation, is accessed efficiently by a 2-photon transition via S1, giving rise to many of the same products observed in this work.[2] It is also known that in photodissociation of carbonyls, internal conversion and intersystem crossing are common.[40] The first three singlet excited states of PA are similar in geometry and maintain the planarity of the ground electronic state.[3] As a result, rapid non-radiative transitions to vibrationally excited S0, T1, or S1 states after excitation to S3 may be facile, though as of yet no experimental evidence for their participation exists. Dissociation pathways Earlier studies have focused on decarboxylation of PA on S0 and S1 where it is believed to be the major pathway.[4,20-21] Clearly, more dissociation reactions contribute at 193 nm. As a starting point we propose two-body fragmentation pathways to help rationalize the 119 products observed in our experiments, and which correspond to the first three mechanistic classes in Table 4.2: CH3COCOOH* → CH3COH + CO2 ΔHrxn = 14200 cm -1 a) CH3COCOOH* → CH3CO + HOCO ΔHrxn = 28600 cm -1 b1) CH3COCOOH* → CH3 + COCOOH ΔHrxn = 32400 cm -1 b2) CH3COCOOH* → CH3COCO + OH ΔHrxn = 36700 cm -1 b3) CH3COCOOH* → CH3COCOO + H ΔHrxn = 39000 cm -1 c1) CH3COCOOH* → H + CH2COCOOH ΔHrxn = 33000 cm -1 c2) The decarboxylation pathway a has been discussed in the literature and is common in carboxylic acids.[17,20,41-42] Pathways b1- b3 break a bond α to a carbonyl bond, similar to the Norrish type I reactions typical of carbonyls. Although this is commonly observed in compounds such as acetone, acetyl cyanide, etc.,[43-47] it was only recently implicated in the photodissociation of PA, where CH3, CH3CO and HOCO products were observed following 2-photon excitation of PA to S2.[2] Pathways c1 and c2 derive from C-H and O-H bond fission reactions commonly observed in aliphatic hydrocarbons at high excitation energies, and indeed H photofragments have been reported following excitation to S2,[2] and we observe them using VMI in this work following excitation to S3. However, as seen in Table 4.2, most of the proposed final dissociation channels involve three-body (and even four-body) fragmentation of PA. Considering the high excitation energy used in this study (~51,800 cm -1 ) this is not surprising. PA is a closed-shell singlet, and therefore the only spin-allowed outcomes of two-body dissociation are doublet 120 + doublet, and singlet CO2 + singlet carbene (MHC). The doublets are all free radicals, and we expect them to have at least one rather weak bond, making their subsequent fragmentation likely. MHC is the highest energy C2H4O isomer, and we expect it to isomerize efficiently and/or dissociate. Therefore, two-body fragmentation processes leading to stable products should be rather unlikely. Indeed, we only observe a small fraction (~3%) of the stable isomers of MHC even though, based on our estimates, the decarboxylation pathway a contributes about 50% of the total flux. In earlier 193 nm photodissociation studies of both PA and d1-PA, Hall et al. attributed the lack of IR emission bands corresponding to MHC, d1- MHC or their isomers to the fairly weak oscillator strengths of C-H and O-H vibrations compared to CO2.[8] We observe that MHC and/or its isomers can undergo further dissociation (Table 4.3) to yield CH3 + HCO (~8%), C2H4 + O (~10%), CH4 + CO (~8%) and possibly also CH2 + H2CO (~8%). Our observations agree with the previously reported 9% yield of CH4.[10] Our experiments using partially deuterated PA, mentioned above, provides some evidence for MHC production, and additional support for our proposed fragmentation pathways of this intermediate. From these experiments, the DCO isotopologue represents 30% of the formyl radical yield, which is similar to the 25% DCO yield one would expect with complete H/D scrambling if MHC is a primary photoproduct that isomerizes and subsequently fragments to methyl + formyl. Heazlewood et al.[22] documented H/D scrambling in photoexcited partially deuterated acetaldehyde that required passage through an MHC intermediate to rationalize their experiments with master equation modelling. The extent of H/D scrambling increased with the excess energy imparted to acetaldehyde in that study, and hence at the higher excess energies available to MHC in PA photodissociation, it 121 is reasonable to expect a nearly statistical H/D ratio in the observed methyl and formyl radicals. The decarboxylation pathway may happen primarily on the S0, S1 or T1 states where dissociation is slower than on a purely repulsive surface, and d1-PA possesses significant internal energy that would facilitate H/D exchange. A second group of reactions likely originates in a Norrish Type I photodissociation mechanism typical of carbonyls (pathway b) with C-C or C-OH fission involving the carbonyl groups. We observe almost 30% contribution of this mechanism through reaction channels resulting in several radical products, e.g., methyl, hydroxyl, and hydroxyformyl. This mechanism was also suggested by Sutradhar et al. in their study of the photodissociation of PA from S2 at excitation energies > 53,000 cm -1 . They concluded that indeed, three- and even four-body fragmentation pathways contributed significantly to the observed products.[2] For example, the observation of CH3CO and HOCO both with very low KE could be explained only by three-body fragmentation reactions. We note that with CH3COCOOD, we observe primarily DOCO and CH3CO (see Figure 4.7(a)), suggesting that this primary bond fission reaction happens on a timescale faster than H/D exchange. We also observe OD products (Figure 4.7(b)) from d1-PA and no OH. In addition, we see evidence of dissociation reactions involving H fragmentation (pathway c). The low D content in the methyl and ketene products is consistent with our hypothesis that the first bond-fission step is sufficiently fast that H/D exchange does not occur to a significant degree. By contrast, the products from two-body fragmentation will always have less internal energy than the initially excited PA molecule, and therefore secondary 122 fragmentation reactions may occur on a longer timescale, making H/D exchange more competitive with bond breaking. Formation Mechanisms of Main Products As discussed above, the excitation energy available in 193 nm photodissociation is sufficiently high to allow the sequential and synchronous (or asynchronous) three-body fragmentation reactions listed in Table 4.2. Below we discuss how the different dissociation mechanisms and reactions contribute to the observed yields of specific products. 4.4.3.I H products H photofragments were detected by REMPI, and their observed KER distribution is shown in Figure 4.2. Although we did not attempt to observe H atoms using MPIMS, we succeeded in detecting some of the co-fragments of their proposed formation reactions. In Figure 4.2 we mark the maximum KER, KEmax, for several of these reactions, taking into account both two-body and sequential and synchronous three-body fragmentation reactions. We have used the same model previously for the 2-photon dissociation on S2.[2] The propensity towards low H-atom kinetic energies and the observation of a significant ketene quantum yield (Table 4.1) support the contribution of the H + CH2CO + HOCO reaction, for which KEmax is ~ 9000 cm -1 . This reaction can proceed via a synchronous (or asynchronous) three-body fragmentation of PA or by sequential fragmentation of CH2COCOOH. We note that ketene has been observed in high energy photodissociation of other carbonyls.[48] The four-body fragmentation reaction generating ketene (2H + CH2CO + CO2) is hard to model but based on energetics its KEmax value would be lower than the 123 corresponding three-body fragmentation reaction. In our VMI measurements we observed H products with high sensitivity, and from the MPIMS experiments we estimate that the contributions of all dissociation reactions producing H photofragments add up to about 15% (see Table 4.3). 4.4.3.II CO products CO has been observed by both REMPI and tunable VUV ionization, and it is one of the two most abundant dissociation products. It is formed both by decarboxylation and Norrish type I reactions. Its origin is likely in sequential three-body fragmentation reactions, such as CH4 + CO + CO2 consistent with previous observation of CH4 using IR absorption by Flynn and coworkers.[10] The internal state distributions of the CO products arising from different pathways are likely to be different and, indeed, the different KER distributions of CO in v=0 and v=1 shown in Figure 4.3 indicate that there is more than one contributing pathway to its formation. CO(v=0) exhibits an isotropic angular distribution with kinetic energies strongly skewed towards the lowest values. CO generated via secondary dissociation of acetaldehyde or MHC products may result in such isotropic angular distribution following initial couplings to lower lying excited states, on which decarboxylation is known to be efficient.[4,49] CO(v=1) exhibits larger KER and an anisotropic angular distribution typical of a parallel transition and fast dissociation. It may derive from synchronous (asynchronous) three-body fragmentation process directly on S3. We note that excitation to S3 involves populating the π* orbital of CO, with excitation delocalized on both the ketonic and carboxylic CO groups. 124 This electronic structure may promote fast dissociation via the Norrish type I reaction mechanism. 4.4.3.III CO2 products CO2 is the product with the second highest mole fraction yield in 193 nm photodissociation of PA, and our analysis concludes that decarboxylation is the dominant process, although only a minor portion of this decarboxylation pathway stops at two-body fragmentation. Decarboxylation leading to acetaldehyde is considered to be a dominant channel in dissociation on S0 and S1, but that is not the case at 193 nm. We observe low yields of acetaldehyde and vinyl alcohol photoproducts and have to invoke reactions that include secondary dissociation of MHC and its isomers to explain the observed high yield of CO2.[50] In our unpublished work on the 351 nm photodissociation of PA following S1 excitation, we observed short lived MHC at IE <9.2 eV, and MHC is expected to be even less stable following 193 nm photodissociation. This conclusion is supported by the observation of products that correspond to further dissociation of MHC and its isomers, such as CH3, HCO, CH2CO, and CH4. As described earlier, Hall and coworkers used IR emission to observe several overtones of the CO2 bending mode from 193 nm PA photodissociation.[8] Contradicting the earlier conclusion by O’Neill et al.,[7] who concluded that 97% of the CO2 is produced in its zero point level, Hall et al.[8] concluded that only ~1.4% of the CO2 products are produced in their zero point level. Hall’s conclusion seems much more plausible based on Franck- Condon arguments, because the CO2 moiety in PA is strongly bent, whereas free CO2 is linear. 125 Decarboxylation may take place on S3 directly or following internal conversion and/or intersystem crossing to lower-lying states. As seen in Table 4.2, photodissociation reactions that include secondary dissociation of HOCO appear to have smaller, yet significant, contributions to the yield of CO2 (see also Table 4.3). 4.4.3.IV OH products As evident in the kinetic time plot shown in Figure 4.10, OH observed in the MPIMS experiments is removed rapidly by secondary reactions and/or collisions with the wall. Based on the analysis presented above, OH is generated by bond fission reactions. Although we cannot distinguish between C-OH fission and initial C-C bond fission, experiments with d1-PA, where we observe 100% OD, confirm that the dissociation reaction happens without significant H/D exchange, and is likely to involve either direct C-OD primary dissociation or fast secondary dissociation of an unstable product such as COCOOD. OH (N = 0–5) has been detected previously following 193 nm photodissociation using LIF, and had a surprisingly slow (~ 600 ns) appearance time.[3] This observation led the authors to propose that OH is generated by breaking the C-OH bond of PA on the T1 PES following a series of internal conversions and intersystem crossings after S3 excitation. They hypothesized that the calculated barrier of 14.7 kcal mol -1 on T1 might be the reason for the slow dissociation. A conclusive explanation for the long OH appearance time awaits further experimental verification and theoretical investigation. Our results, however, show that the OH yield is significant (~17%). 126 4.4.3.V CH3CO products Acetyl, like OH, is produced mainly via pathway b, the Norrish Type I mechanism. The contribution of two-body fragmentation to give stable CH3CO + HOCO radicals is very small, indicating that these products have sufficient internal energy to undergo further dissociation. A similar conclusion was reached for acetyl products formed in the 2-photon dissociation experiments described above.[2] The most likely sources of the observed acetyl radicals in this work are the three-body fragmentation reactions listed in Table 4.2. Summary and Conclusions The 193 nm photodissociation of PA has been studied using the complementary VMI and MPIMS techniques. MPIMS, which employs tunable ionization, has allowed us to detect products that are stable for longer than 1 ms (at 4 Torr and 300 K) and determine their mole fraction yields with reasonable accuracy. The VMI technique identifies nascent products created in the collision-free environment of a molecular beam but is limited to products that have efficient REMPI detection schemes. With these two techniques, we observed all the major dissociation products of PA (see Table 4.1). Acetaldehyde and vinyl alcohol are minor products, but products that are known to arise from their dissociation, such as HCO, H2CO and CH4, were identified and quantified. The multivariate analysis described above provided a reasonable description of the complicated photoinitiated chemistry of PA at 193 nm, albeit with some simplifications. We recognize that a full kinetic model, which addresses all the intermediates and their bimolecular reactions with PA and other photoproducts, is needed to capture the full reaction kinetics taking place in the MPIMS flow tube. However, we believe 127 that the reaction mechanisms and specific dissociation reactions presented in Table 4.2 provide the most complete reaction scheme proposed to date to explain the photodissociation of PA initiated on the S3 state. The observed products and yields are rationalized based on three dissociation mechanisms: (i) decarboxylation creating CO2 and a co-product that usually undergoes secondary fragmentation; (ii) Norrish type I dissociation typical of carbonyls; and (iii) O-H and C-H bond fission reactions generating H atoms. Some radicals and molecules generated by two-body fragmentation reactions via these pathways are unobserved due to their short lifetimes. Our analysis shows that most of the dissociation reactions listed in Table 4.2 have more than two products, which is not surprising considering the large excitation energy (~51,800 cm -1 ) and fairly low energy required for decarboxylation (14,000 cm -1 ) and breaking the CH3C(O)-C(O)OH bond (28,000 cm -1 ). In addition, we note the low dissociation energies of CH3CO, HOCO, and MHC. The underlying mechanisms (i) – (iii) all contribute, and sequential dissociation of one or both primary products is probable. Synchronous (or asynchronous) three-body dissociation of PA is likely also important. Experiments with partially deuterated PA (CH3COCOOD) support our interpretation. We recognize that some reactions, such the one yielding CH3CO + CO2 + H, may be assigned to more than one mechanism, but we find that the reactions listed in Table 4.2 give a reasonable description of the overall photodissociation mechanism of PA at 193 nm. A similar conclusion, namely, that three-body fragmentation reactions are important, was also reached in a previous 128 publication on the photodissociation of PA on S2,[2] though the number of reaction pathways that could be observed in that study was limited. Initial bond-breaking events on the S3 surface are likely fast, as indicated by the anisotropic angular distribution of CO(v=1). However, H atoms and CO(v=0) have isotropic angular distributions, hinting that the roles of internal conversion and intersystem crossing to lower states are non-negligible, but yet to be assessed. An intriguing issue concerns the probability of H-transfer from the carboxylic group to the ketonic CO group followed by decarboxylation, which requires a 5-member cyclic transition state.[9,19-21] If the dissociation on S3 is fast and the available energy is large, then such H-transfer via a tight transition state would be less likely than synchronous (or asynchronous) three-body fragmentation via the reactions listed in Table 4.2. Dynamical calculations initiated on the S3 potential energy surface are needed to determine the relative contributions of the proposed dissociation reactions and clarify the roles of lower electronic states. References [1] Horowitz, A.; Meller, R.; Moortgat, G. K., The UV–VIS absorption cross sections of the α- dicarbonyl compounds: pyruvic acid, biacetyl and glyoxal. J. Photochem. Photobiol. A: Chem. 2001, 146 (1-2), 19-27. [2] Sutradhar, S.; Samanta, B. R.; Fernando, R.; Reisler, H., Spectroscopy and Two-Photon Dissociation of Jet-Cooled Pyruvic Acid. J. Phys. Chem. A 2019, 123 (28), 5906-5917. [3] Dhanya, S.; Maity, D. K.; Upadhyaya, H. P.; Kumar, A.; Naik, P. D.; Saini, R. D., Dynamics of OH formation in photodissociation of pyruvic acid at 193 nm. J. Chem. Phys. 2003, 118 (22), 10093-10100. 129 [4] Chang, X.-P.; Fang, Q.; Cui, G., Mechanistic photodecarboxylation of pyruvic acid: Excited-state proton transfer and three-state intersection. J. Chem. Phys. 2014, 141 (15), 154311. [5] Arnett, J.; Larson, D.; McGlynn, S., Absorption and emission spectroscopy of pyruvic acids and pyruvate esters. J. Am. Chem. Soc. 1973, 95 (23), 7599-7603. [6] Wood, C. F.; O'Neill, J. A.; Flynn, G. W., Infrared diode laser probes of photofragmentation products: Bending excitation in CO2 produced by excimer laser photolysis of pyruvic acid. Chem. Phys. Lett. 1984, 109 (4), 317-323. [7] O’Neill, J. A.; Kreutz, T. G.; Flynn, G. W., IR diode laser study of vibrational energy distribution in CO2 produced by UV excimer laser photofragmentation of pyruvic acid. J. Chem. Phys. 1987, 87 (8), 4598-4605. [8] Hall, G. E.; Muckerman, J. T.; Preses, J. M.; Weston Jr, R. E.; Flynn, G. W., Time-resolved FTIR studies of the photodissociation of pyruvic acid at 193 nm. Chem. Phys. Lett. 1992, 193 (1-3), 77-83. [9] Rosenfeld, R. N.; Weiner, B., Energy disposal in the photofragmentation of pyruvic acid in the gas phase. J. Am. Chem. Soc. 1983, 105 (11), 3485-3488. [10] Sarkar, S. K.; Mittal, J. P.; Flynn, G. W., Infrared diode laser probing of methane in excimer laser photolysis of pyruvic acid. Proc. Indian Acad. Sci. (Chem. Sci.) 1994, 106, 1307-1314. [11] Maul, C.; Gericke, K.-H., Aspects of photoinduced molecular three-body decay. J. Phys. Chem. A 2000, 104 (12), 2531-2541. [12] Tjossem, P. J.; Smyth, K. C., Multiphoton excitation spectroscopy of the B 1 Σ + and C 1 Σ + Rydberg states of CO. J. Chem. Phys. 1989, 91 (4), 2041-2048. [13] Ryazanov, M.; Reisler, H., Improved sliced velocity map imaging apparatus optimized for H photofragments. J. Chem. Phys. 2013, 138 (14), 144201. [14] Rodrigo, C. P.; Sutradhar, S.; Reisler, H., Imaging Studies of Excited and Dissociative States of Hydroxymethylene Produced in the Photodissociation of the Hydroxymethyl Radical. J. Phys. Chem. A 2014, 118 (51), 11916-11925. [15] Osborn, D. L.; Zou, P.; Johnsen, H.; Hayden, C. C.; Taatjes, C. A.; Knyazev, V. D.; North, S. W.; Peterka, D. S.; Ahmed, M.; Leone, S. R., The multiplexed chemical kinetic photoionization mass spectrometer: A new approach to isomer-resolved chemical kinetics. Rev. Sci. Instrum. 2008, 79 (10), 104103. 130 [16] Taatjes, C. A.; Hansen, N.; Osborn, D. L.; Kohse-Höinghaus, K.; Cool, T. A.; Westmoreland, P. R., “Imaging” combustion chemistry via multiplexed synchrotron-photoionization mass spectrometry. Phys. Chem. Chem. Phys. 2008, 10 (1), 20-34. [17] Leermakers, P. A.; Vesley, G. F., The photochemistry of α-keto acids and α-keto esters. I. Photolysis of pyruvic acid and benzoylformic acid. J. Am. Chem. Soc. 1963, 85 (23), 3776-3779. [18] Griffith, E. C.; Carpenter, B. K.; Shoemaker, R. K.; Vaida, V., Photochemistry of aqueous pyruvic acid. Proc. Nat. Acad. Sci. 2013, 110 (29), 11714-11719. [19] Takahashi, K.; Plath, K. L.; Skodje, R. T.; Vaida, V., Dynamics of vibrational overtone excited pyruvic acid in the gas phase: Line broadening through hydrogen-atom chattering. J. Phys. Chem. A 2008, 112 (32), 7321-7331. [20] Vesley, G. F.; Leermakers, P. A., The photochemistry of α-keto acids and α-keto esters. III. Photolysis of pyruvic acid in the vapor phase. J. Phys. Chem. 1964, 68 (8), 2364- 2366. [21] Yamamoto, S.; Back, R., The photolysis and thermal decomposition of pyruvic acid in the gas phase. Can. J. Chem. 1985, 63 (2), 549-554. [22] Heazlewood, B. R.; Maccarone, A. T.; Andrews, D. U.; Osborn, D. L.; Harding, L. B.; Klippenstein, S. J.; Jordan, M. J. T.; Kable, S. H., Near-threshold H/D exchange in CD3CHO photodissociation. Nat. Chem. 2011, 3 (6), 443-448. [23] Dodson, L. G.; Savee, J. D.; Gozem, S.; Shen, L.; Krylov, A. I.; Taatjes, C. A.; Osborn, D. L.; Okumura, M., Vacuum ultraviolet photoionization cross section of the hydroxyl radical. J. Chem. Phys. 2018, 148 (18), 184302. [24] Yang, B.; Wang, J.; Cool, T. A.; Hansen, N.; Skeen, S.; Osborn, D. L., Absolute photoionization cross-sections of some combustion intermediates. Int. J. Mass Spectrom. 2012, 309, 118-128. [25] Savee, J. D.; Soorkia, S.; Welz, O.; Selby, T. M.; Taatjes, C. A.; Osborn, D. L., Absolute photoionization cross-section of the propargyl radical. J. Chem. Phys. 2012, 136 (13), 134307. [26] Cool, T. A.; Nakajima, K.; Mostefaoui, T. A.; Qi, F.; McIlroy, A.; Westmoreland, P. R.; Law, M. E.; Poisson, L.; Peterka, D. S.; Ahmed, M., Selective detection of isomers with photoionization mass spectrometry for studies of hydrocarbon flame chemistry. J. Chem. Phys. 2003, 119 (16), 8356-8365. 131 [27] Toulson, B. W.; Fishman, D. A.; Murray, C., Photodissociation dynamics of acetone studied by time-resolved ion imaging and photofragment excitation spectroscopy. Phys. Chem. Chem. Phys. 2018, 20 (4), 2457-2469. [28] Samson, J.; Stolte, W. C., Precision measurements of the total photoionization cross- sections of He, Ne, Ar, Kr, and Xe. J. Electron. Spectrosc. Relat. Phenom. 2002, 123 (2-3), 265-276. [29] Person, J. C.; Nicole, P. P., Isotope effects in the photoionization yields and the absorption cross sections for acetylene, propyne, and propene. J. Chem. Phys. 1970, 53 (5), 1767-1774. [30] Stull, D. R., Vapor pressure of pure substances. Organic and inorganic compounds. Ind. Eng. Chem. 1947, 39 (4), 517-540. [31] Shafir, E. V.; Slagle, I. R.; Knyazev, V. D., Kinetics and products of the self-reaction of propargyl radicals. J. Phys. Chem. A 2003, 107 (42), 8893-8903. [32] Wang, J.; Yang, B.; Cool, T. A.; Hansen, N.; Kasper, T., Near-threshold absolute photoionization cross-sections of some reaction intermediates in combustion. Int. J. Mass spectrom. 2008, 269 (3), 210-220. [33] Leach, S.; Schwell, M.; Jochims, H.-W.; Baumgńrtel, H., VUV photophysics of acetic acid: Fragmentation, fluorescence and ionization in the 6–23 eV region. Chem. Phys. 2006, 321 (1-2), 171-182. [34] Katayama, D.; Huffman, R.; O'Bryan, C., Absorption and photoionization cross sections for H2O and D2O in the vacuum ultraviolet. J. Chem. Phys. 1973, 59 (8), 4309-4319. [35] Cool, T. A.; McIlroy, A.; Qi, F.; Westmoreland, P. R.; Poisson, L.; Peterka, D. S.; Ahmed, M., Photoionization mass spectrometer for studies of flame chemistry with a synchrotron light source. Rev. Sci. Instrum. 2005, 76 (9), 094102. [36] Cool, T. A.; Wang, J.; Nakajima, K.; Taatjes, C. A.; Mcllroy, A., Photoionization cross sections for reaction intermediates in hydrocarbon combustion. Int. J. Mass spectrom. 2005, 247 (1-3), 18-27. [37] Taatjes, C. A.; Osborn, D. L.; Selby, T. M.; Meloni, G.; Fan, H.; Pratt, S. T., Absolute photoionization cross-section of the methyl radical. J. Phys. Chem. A 2008, 112 (39), 9336-9343. [38] Shubert, V. A.; Pratt, S. T., Photodissociation of acetaldehyde and the absolute photoionization cross section of HCO. J. Phys. Chem. A 2010, 114 (42), 11238-11243. 132 [39] Ruscic, B.; Bross, D., Active Thermochemical Tables (ATcT) values based on ver. 1.122d of the Thermochemical Network (2018); available at ATcT.anl.gov. [40] Finlayson-Pitts, B. J.; Pitts Jr, J. N., Chemistry of the upper and lower atmosphere: theory, experiments, and applications. Elsevier: 1999. [41] Huang, C.; Zhang, C.; Yang, X., State-selected imaging studies of formic acid photodissociation dynamics. J. Chem. Phys. 2010, 132 (15), 154306. [42] Hunnicutt, S. S.; Waits, L. D.; Guest, J. A., 1 (n,π*)-Photochemistry of acetic acid at 200 nm: further evidence for an exit channel barrier and reaction selectivity. J. Phys. Chem. 1991, 95 (2), 562-570. [43] Kim, S. K.; Pedersen, S.; Zewail, A. H., Direct femtosecond observation of the transient intermediate in the α‐cleavage reaction of (CH3)2CO to 2CH3 + CO: Resolving the issue of concertedness. J. Chem. Phys. 1995, 103 (1), 477-480. [44] North, S. W.; Marr, A. J.; Furlan, A.; Hall, G. E., Nonintuitive asymmetry in the three-body photodissociation of CH3COCN. J. Phys. Chem. A 1997, 101 (49), 9224-9232. [45] Osamura, Y.; Schaefer III, H. F.; Dupuis, M.; Lester Jr, W. A., A unimolecular reaction ABC → A + B + C involving three product molecules and a single transition state. Photodissociation of glyoxal: HCOHCO → H2 + CO + CO. J. Chem. Phys. 1981, 75 (12), 5828-5836. [46] Zhong, Q.; Poth, L.; Castleman Jr, A., Ultrafast dissociation dynamics of acetone: A revisit to the S 1 state and 3s Rydberg state. J. Chem. Phys. 1999, 110 (1), 192-196. [47] Deshmukh, S.; Hess, W. P., Photodissociation of acetyl chloride: Cl and CH3 quantum yields and energy distributions. J. Chem. Phys. 1994, 100 (9), 6429-6433. [48] Harrison, A. W.; Kharazmi, A.; Shaw, M. F.; Quinn, M. S.; Lee, K. K.; Nauta, K.; Rowell, K. N.; Jordan, M. J.; Kable, S. H., Dynamics and quantum yields of H2 + CH2CO as a primary photolysis channel in CH3CHO. Phys. Chem. Chem. Phys. 2019, 21 (26), 14284-14295. [49] da Silva, G., Carboxylic Acid Catalyzed Keto-Enol Tautomerizations in the Gas Phase. Angew. Chem. Int. Ed. 2010, 49 (41), 7523-7525. [50] Yadav, J. S.; Goddard, J. D., Methoxycarbene and methylhydroxycarbene: energies, structures, vibrational frequencies, and unimolecular reactivities. J. Chem. Phys. 1986, 85 (7), 3975-3984. 133 Pyruvic acid photochemistry initiated on S 1 state Introduction As discussed earlier, dissociation via the π*←n+ band at 300–380 nm, which accesses the S1 excited state, has attracted considerable research interest but also some controversy. The primary photoproducts are believed to be CO2 and methylhydroxycarbene (MHC), which rapidly isomerizes to the most stable isomer — acetaldehyde.[1-4] When PA’s photodissociation is investigated in the presence of buffer gases such as O2, N2, air, and water in an atmospheric simulator, several minor products, e.g., acetic acid and CO, are also observed.[5] The photodissociation quantum yield and relative product yields in these studies depend sensitively on the nature of the experiment and sometimes disagree amongst one another on the mechanism, product yields, and the roles of the electronic ground and excited S1 states. [1,3,5-7] In spite of significant effort, the question still remains — how does PA photodissociate from the S1 excited state and what are the nascent photoproducts? On the ground electronic state, the lowest energy conformer, Tc, is stabilized by a hydrogen bond (H-bond) between hydrogen on the carboxylic group and the oxygen of the 134 ketonic group(see Figure 3.3).[8-9] Theoretical studies on the S1 state of PA attest to the complexity of the dissociation dynamics.[10] Chang et al. calculated the height of the barrier to H-transfer from the carboxylic to the ketonic group on the S1 potential energy surface (PES) of the Tc (cis-keto) conformer of PA.[10] They report that H-transfer is the first step in the decarboxylation reaction, which takes place on S0 following passage through a conical intersection. They also find efficient spin-orbit coupling to the close-lying T1 triplet state, which may lead to the Norrish type I reaction products typical in carbonyls — acetyl (CH3CO) and hydroxyformyl (HOCO) radicals. Chapter 3 shows the S1←S0 absorption spectrum of PA in molecular beams which indicates that the S1 state lives for a time longer than a vibrational (or even rotational) period, and that several low frequency skeletal modes are efficiently excited in the S1 excited state.[11] The excess energy available to products in the decarboxylation pathway at 351 nm is about ~14,300 cm -1 ,[9] which is above the isomerization barriers to both the acetaldehyde and vinyl alcohol tautomers. Yet, the only MHC isomer observed to date following S1 photodissociation is acetaldehyde. In pyrolysis of PA on S0, a small amount of MHC was stabilized when captured fast in a cold Ar matrix,[12] but MHC formed in this way has little rovibrational energy, and thus its fate may not be the same as in PA’s photodissociation. We report here the first direct observation of MHC following S1 excitation and, moreover, we show that it decays within less than a millisecond to its stable isomers, acetaldehyde and vinyl alcohol. We obtain similar results with CH3COCOOD (d1-PA) and observe that the lifetime of the deuterated MHC is slightly longer than the fully hydrogenated 135 isotopologue. In addition, we present indirect evidence of d1-PA dissociation from the triplet T1 PES, which terminates with the Norrish Type I bond fission products CH3CO and DOCO. These findings, combined with insights from theoretical work, lead us to suggest a mechanism for the photodissociation initiated on S1. Figure 5.1 A cartoon highlighting the slow H tunneling of MHC to its isomers: acetaldehyde and vinyl alcohol, because of the existence of significant barriers to H exchange. A relatively fast MHC removal in our experiment raises the possibility of secondary reactions. Because of its hypothesized participation in sugar formation in prebiotic environments, bimolecular reactions of hydroxycarbenes have been of particular interest.[13] Recently, Schreiner and coworkers showed that the reaction between hydroxycarbene (trapped in an Ar matrix) and formaldehyde can form glycolaldehyde via a barrierless mechanism.[14] Here we show prima facie indication of decarboxylative benzoin condensation type reaction between MHC and/or its isomers and PA. 136 Experimental section PA (Sigma-Aldrich; 98%) is doubly distilled at 60 °C at <1 Torr vacuum to eliminate volatile impurities. Only the clear distillate is used, leaving behind a yellow crude liquid consisting mostly of photocatalyzed oligomers.[15-16] A freshly distilled sample is used in each experiment. Experimental details pertaining to the H-PFY spectrum have been explained in Chapter 3, and gas flow conditions in the MPIMS have been explained in Chapter 4. In the MPIMS apparatus, PA photodissociation is achieved using 351 nm radiation generated by a XeF excimer laser operated at 10 Hz. Unfocused laser radiation (~ 35 mJ cm - 2 ) is propagated along the length of the tube parallel to the gas flow direction. The quartz reactor tube is coated with polychlorotrifluroethylene 2300 wax (Halocarbon Products Corporation) to suppress loss of reactive radicals due to wall loss. Singly deuterated PA samples (d1-PA, CH3COCOOD) are prepared in-situ by combining a 20 sccm He flow through a glass bulb containing PA (21 °C; 36 Torr total with He) and a 10 sccm He flow through another glass bulb containing D2O (21 °C; 54 Torr total with He). The total flow is balanced to 250 sccm flow with He. This method yields ~97% pure d1-PA, confirmed using a time-of-flight (TOF) mass spectrum obtained at the ALS 9.0.2.3 beamline. 137 Results and Discussion S 1 ←S 0 spectroscopy in a molecular beam Figure 5.2 H-PFY spectrum of PA in Ar. Peak assignments are labeled based on tentative assignments carried out in reference [11]. As mentioned in Chapter 3, the H-PFY spectra were recorded at different sample temperatures, carrier gases, and backing pressures. The lowest energy peak (peak 1 in Figure 5.2) appears consistently at 26,710 cm -1 under all conditions. Subhashish et al. assigned this peak as the origin (0-0) band of the S1←S0 transition of the Tc conformer.[11] The other resolved transitions and their positions relative to the 0-0 band are listed in Table 5.1. 138 Table 5.1 Observed S1←S0 rovibtonic bands and the notations of the normal modes are based on the frequencies computed in reference [8]. The unassigned transitions were attributed to splittings in internal rotor transitions of the CH3 moiety. Peak number Positions (cm -1 ) Relative position from peak 1 (cm -1 ) Tentative assignment 1 26,710 0 0-0 2 26,832 122 υ23 (CH3 torsion) 3 26,847 137 υ24 (C-C torsion) 4 26,944 234 unassigned 5 26,952 242 2υ23 6 26,960 250 unassigned 7 26,979 269 2υ24 8 27,006 296 υ16 (CCC bending) 9 27,050 340 unassigned 10 27,063 353 unassigned 11 27,072 362 unassigned 12 27,078 368 3υ23 13 27,084 374 unassigned 14 27,094 384 CH3 torsion/CCO bending Based on vibrational frequencies in the S0 state, we made tentative assignments to the peaks in the H-PFY spectrum.[8] The frequencies of the OCCO (υ24) and CH3 (υ23) torsional modes in S0 are reported to be 90 and 134 cm -1 , respectively.[17] The electronic 139 structure calculations by Chang et al. show that upon excitation to the S1 state, the central C- C bond is shortened compared to S0, and therefore, it is reasonable to expect that OCCO torsion and CCC or CCO bending modes will be populated in the Frank-Condon region. It is evident from Figure 5.2 that the two low frequency torsional modes cannot account for all the observed peaks in the PFY spectrum up to 400 cm -1 above the 0-0 peak. Chang et al. predict an eclipse to staggered conformation change upon electronic excitation to the S1 state. As a result, the methyl group internal rotation, which causes additional splitting of the bands, is expected to account for several unassigned peaks. Non-degenerate internal rotational modes of the methyl group, which give rise to split peaks, have been reported both experimentally and theoretically.[18-20] High-level electronic structure calculations that include anharmonic vibrational frequencies and internal rotation of the methyl group are required in order to offer definite assignments. Decarboxylation reaction We sample molecules continuously in the MPIMS to record the time evolution of products formed in the photodissociation reaction. We identify molecular species using their TOF spectra obtained at selected ionization energies. Unless stated otherwise, the pre- photodissociation background of PA and its daughter ions is subtracted in the plots shown below to highlight signals arising only from photoproducts. A TOF spectrum obtained at 10.25 eV ionization energy shows that m/z = 44.03 ± 0.05 is the most intense peak. At 14.00 eV ionization energy, the most intense peak is at m/z = 44.00 ± 0.05. We have confirmed that the former peak corresponds to the molecular formula C2H4O and the latter as CO2 by 140 comparing their PI spectra to the known species, as shown in Figure 5.3. We find that the C2H4O signal (m/z =44.03) is observed even below the reported ionization energies of acetaldehyde, and the cumulative signal is, in fact, a mixture of the two tautomers, acetaldehyde and vinyl alcohol. We circumvent the similarity in mass of the CO2 and C2H4O photoproducts by repeating the measurements with d1-PA. At 10.25 eV we obtain the most intense peak at m/z = 45.03 ± 0.05, whereas m/z = 44.00 ± 0.05 remains the most intense peak at 14 eV. The dominant fragment ion formed from dissociative ionization of both PA and d1-PA above 10.2 eV is m/z = 43.02 (acetyl). The absence of deuteration in the alkyl moiety of the acetyl ion indicates that the deuteration in d1-PA is indeed achieved on the acidic H. The decarboxylation of PA yields the C2H4O isomers as co-fragments of CO2. We rule out contributions of secondary processes to the observed signals because the rise times are limited by the instrument response time (480 μs). The instrument response time was determined from the rise time of NO photoproducts, which are formed promptly following 351 nm photodissociation of NO2. Products of secondary bimolecular reactions are expected to rise more slowly, e.g. over several milliseconds. This suggests that the nascent products of PA photodissociation at 351 nm are MHC and CO2, even though the observed products are acetaldehyde and vinyl alcohol, the more stable isomers of MHC. In calculations carried out on the S0 PES, Vaida and coworkers found that a complete H-transfer process near the dissociation threshold of the Tc-conformer takes place within 200 fs.[9] This time scale was based on a combination of measurements of the linewidth of 141 OH overtones in the PA vibrational absorption spectrum and dynamical calculations on the S0 PES. The calculations suggested that rapid H exchange followed by its transfer to the carbonyl group was the first step in the decarboxylation reaction.[9] Using the known absolute PI cross-sections of acetaldehyde and vinyl alcohol, we determine that they are formed in the ratio of 2.1 ± 0.4 (favoring acetaldehyde) in the present experiments.[21] This ratio is the same for both the d0 and d1 isotopologues of C2H4O. Previous studies have concluded that MHC products with low internal energy isomerize entirely to acetaldehyde, even though the isomerization barrier for this process is higher by 6 kcal/mol than that to vinyl alcohol, because the barrier to acetaldehyde is narrower than the corresponding one to vinyl alcohol.[12] This interpretation was confirmed in studies in which MHC was trapped in a cold Ar matrix where its internal energy was only the zero- point energy.[12] MHC formed with more than 22 kcal/mol internal energy, as is possible in our studies, will favor isomerization to vinyl alcohol because of more efficient transmission. Given the low dissociation threshold for decarboxylation of PA (14,200 cm -1 )[9] and the 351 nm energy of 28,500 cm -1 , we expect MHC to be formed with significant internal energy (up to 14,300 cm -1 ). A previous experiment by Rosenfeld et al. at 351 nm, which reported detection of the CO2 co-fragment, showed no IR fluorescence, and thus CO2 was believed to be formed with little, if any, vibrational excitation.[22] Although collisional relaxation of the nascent CO2 excitation in that experiment might have affected the observations, these results suggest that the internal energy in the MHC fragment could be significant. 142 Figure 5.3 Photoionization (PI) spectra measured at a) 9.2–10.4 eV showing the presence of both vinyl alcohol (VA) and acetaldehyde (Ac) photoproducts, and b)13.7–14.1 eV showing the CO2 photoproduct. The photoproducts are identified by comparing the PI spectra obtained in this work to known absolute PI curves of these molecules (Daniel Rösch, private communication). In a previous study we discovered a facile two-photon excitation process via S1 to S2 that was competitive with dissociation on S1 at high laser fluence.[11] Therefore, we measured the dependence of the yields of the C2H4O and CO2 photoproducts on laser fluence to confirm that in the present experiments, carried out at much lower laser fluence, they are 143 formed by one-photon dissociation. We found that these product signals varied nearly linearly with the 351 nm laser fluence (See Figure 5.4). Figure 5.4 Fluence dependent signals of PA photoproducts a) acetaldehyde and vinyl alcohol at 10.25 eV b) vinyl alcohol at 9.65 eV c) CO2 at 14.15 eV and d) species M at 9.65 eV. The plotted signal intensities, fit to a quadratic equation with (0,0) as intercept, show that the fits are nearly linear. In addition, the time trace of m/z = 88.02 showed depletion in the PA ion signal of (0.33 ± 0.15)% upon photolysis at 30 mJ/cm 2 . Assuming a photodissociation quantum yield of unity, this result is consistent with the estimates of 0.34% absorption based on the reported absorption cross-section at 351 nm. We conclude that MHC, its isomers, and CO2 are formed in these experiments by one-photon excitation to the S1 state of PA. 144 MHC formation and decay The nascent MHC photoproduct is expected to undergo up to ~20,000 collisions during the MPIMS instrument response time (~0.5 ms). However, 99% of these collisions will be with He atoms, and these are unlikely to cause complete vibrational cooling. As a result, we expect the MHC products to have a broad range of internal energies. As discussed above, those with high internal energies will isomerize rapidly favoring vinyl alcohol, while products formed with lower internal energies will do so via tunneling favoring acetaldehyde. We estimate an integrated transmission factor for the isomerization coordinate by using a simple Eckart potential, 𝑉𝑉 ( 𝑥𝑥 ), of the form,[23-24] 𝑉𝑉 ( 𝑥𝑥 ) = 𝐴𝐴𝐴𝐴 (1 + 𝐴𝐴 ) 2 + 𝐵𝐵 𝐴𝐴 (1 + 𝐴𝐴 ) , 𝐴𝐴 = 𝑚𝑚 𝑥𝑥 ∕𝑎𝑎 . (5.1) This potential has limiting values of V→0 as y→0 ( 𝑥𝑥 → – ∞) and V→0 as y→∞ ( 𝑥𝑥 → + ∞). The values of A, B, and b depend on barrier height (maximum value of 𝑉𝑉 ( 𝑥𝑥 ), 𝑉𝑉 ‡ ), exothermicity (ΔHrxn) and imaginary frequency (measure of barrier curvature, 𝜈𝜈 ‡ ). 𝐵𝐵 = Δ 𝐻𝐻 rx n (5.2) 𝐴𝐴 = 2 𝑉𝑉 ‡ − 𝐵𝐵 + 2 � 𝑉𝑉 ‡ ( 𝑉𝑉 ‡ − 𝐵𝐵 ) (5.3) 𝑏𝑏 = 1 2 𝜋𝜋 𝜈𝜈 ‡ √ 𝑚𝑚 ( 𝐴𝐴 2 − 𝐵𝐵 2 ) � √2 𝐴𝐴 � 3 (5.4) The values of above parameters have been theoretically computed by Schreiner et al.[12] and summarized in Table 5.2 below. 145 Table 5.2 Summary of thermodynamic values for MHC isomerization reaction. [12] Acetaldehyde Vinyl alcohol 𝑉𝑉 ‡ (kcal/mol) 29.99 23.94 ΔHrxn(kcal/mol) − 50.7 − 39.8 𝜈𝜈 ‡ (cm -1 ) 2031i 1443i Equation 5.1 can be inserted in the Schrödinger equation and exact solutions can be obtained to determine the tunneling probability across the barrier, 𝑃𝑃 (𝐸𝐸 ), as a function of the incident energy, E. [23-24] Introducing the abbreviations 𝑘𝑘 = √2 𝑚𝑚 𝐸𝐸 ℏ , 𝑚𝑚 = � ( 𝐸𝐸 − 𝐵𝐵 ) 𝐸𝐸 , (5.5) 𝑃𝑃 ( 𝐸𝐸 ) = sinh 2 [ 𝜋𝜋 𝑘𝑘𝑏𝑏 (1 + 𝑚𝑚 )] − sinh 2 [ 𝜋𝜋 𝑘𝑘𝑏𝑏 (1 − 𝑚𝑚 )] sinh 2 [ 𝜋𝜋 𝑘𝑘𝑏𝑏 (1 + 𝑚𝑚 )] + cosh 2 � 𝜋𝜋 2 � 8 𝑚𝑚 𝑏𝑏 2 𝐴𝐴 ℏ 2 − 1 � 1 2 � (5.6) The total integrated transmission factor, 𝛤𝛤 , can then be obtained by calculating the tunneling probability for the entire available energy range, weighted by a Boltzmann distribution of the total population of internal states. 𝛤𝛤 = � 𝑃𝑃 (𝐸𝐸 ) 𝑚𝑚 − 𝐸𝐸 𝑢𝑢 𝑘𝑘 ⁄ 𝑘𝑘 𝑇𝑇 𝐸𝐸 𝑚𝑚 𝑚𝑚𝑚𝑚 0 𝑚𝑚𝐸𝐸 (5.7) 146 Here, 𝐸𝐸 𝑚𝑚 𝑎𝑎𝑥𝑥 is the maximum available energy (14,300 cm -1 ) and 𝑃𝑃 (𝐸𝐸 ) is the tunneling probability at energy 𝐸𝐸 through the Eckart barrier for each isomer, which depends on the barrier width and height at each energy. Figure 5.5 The computed Eckart potential (top panel) is plotted for both acetaldehyde and vinyl alcohol. Corresponding plots of transmission factors at T = 20, 215 and 400 K are also shown (bottom panel). The acetaldehyde to vinyl alcohol ratio of integrated areas under the plot of transmission factors at 215 K is 2:1. With equation 5.7, we reproduce the observed acetaldehyde to vinyl alcohol ratio of 2.1:1 by assuming a 215 K Boltzmann distribution of MHC internal states. The results of the 147 simplified integrated transmission calculations (equation 5.7) indicate that the observed ratio of acetaldehyde to vinyl alcohol can be rationalized by assuming that only a small fraction of the MHC photoproducts has high internal energies (see Figure 5.5). We believe that MHC products with little internal energy will be stable enough to be detected within the time-resolution of our instrument (480 μs), because cold MHC has measured tunneling lifetime of up to hours when trapped in a 4 K Argon matrix in the absence of secondary reactions. Because of the excess energy available in our experiments, the isomer ratio is not expected to change significantly in going from MHC to d1-MHC but the tunneling rate is expected to be slower for d1-MHC. We searched for MHC products at photon energies below 9.1 eV to minimize the contribution of vinyl alcohol products. Significant signal averaging revealed the kinetic time traces of an extremely reactive species at m/z=44.03 for PA and m/z=45.03 for d1-PA, as shown in Figure 5.6(a). We could record the m/z=45.03 signal with acceptable signal-to- noise ratio (S/N) above 8.44 eV, and the signal improved around 8.94 eV, as seen in the PI spectrum displayed in Figure 5.6(b). In a previous theoretical study, we had estimated the adiabatic and vertical ionization energies of trans-MHC at 8.21 eV and 8.94 eV, respectively, with an uncertainty of 0.3 eV, which bolsters our confidence that the observed species is indeed MHC. Furthermore, the S/N is better and the removal time is slower for d1-MHC compared to MHC, as evident in the decay curves shown in Figure 5.6(a). This is attributed to slower isomerization reactions involving D exchange compared to H exchange. 148 Figure 5.6 (a) Integrated (over 8.4–9.1 eV ionization) kinetic time traces of m/z=44.03 (MHC) and m/z=45.03 (d1-MHC) from PA and d1-PA photodissociation, respectively. The blue dashed line represents the noise baseline averaged at time 2 – 20 ms. (b) The PI spectrum of m/z=45.03 obtained in the photodissociation of d1-PA. Chang et al. carried out the most comprehensive theoretical study to date of the dissociation of PA initiated on S1.[10] They find that H transfer from the carboxylic group to the ketonic group takes place on S1 over a fairly small barrier of 7 kcal/mol, reaching an isomer they call S1-HT. The minimum energy and geometry of this isomer are very close to the triple conical intersection of S1-HT with S0 and T1. They envision an efficient transfer to S0 via this conical intersection from which decarboxylation takes place by two routes. Some 149 of the flux dissociates directly after passing through the conical intersection, while the other part propagates first towards the minimum geometry of S0 and then decarboxylates.[10] These two pathways may result in different quantum state distributions in the products, with the direct component being nonstatistical and the indirect route leading to a more statistical-like distribution. While the S1 and S0 states are key to understanding the decarboxylation of PA following 351 nm excitation, the participation of the T1 state is also invoked. Chang et al. find that the S1 and T1 states are close in energy, with respective minima at 84 and 78 kcal/mol, and these states are coupled by a strong spin-orbit interaction that facilitates intersystem crossing.[10] Da Silva has later placed the corresponding values at 75 and 69 kcal/mol (private communication). In our previous experimental studies, we showed that the S1←S0 band origin is at 374 nm (76 kcal/mol), in good agreement with the more recent calculation.[6,11] The linewidths of the vibronic bands measured in the molecular beam indicate that the S1 state is fairly long-lived (>1 ps), making the absorption of a second photon to the S2 state at high laser fluences competitive with dissociation on S1. We hypothesize that the long lifetime of S1 could also make the S1/T1 intersystem crossing near the S1 minimum configuration competitive with the transition to the S1-HT state and subsequent internal conversion to S0, at least at energies close to the S1→S1-HT barrier. 150 Figure 5.7 Kinetic time traces of acetyl (left) and DOCO (right) from d1-PA photodissociation at 9.65 eV PI energy of. The 13 C background of the dominant d1-vinyl alcohol product is shown by the blue dashed line on the right panel. Indeed, there exists some experimental evidence for the participation of T1 in the photodissociation. Yamamoto and Back find that the light emission observed following 340- 380 nm excitation to S1 can be partially quenched by adding oxygen, which is known to quench triplet states. They propose that T1 gives rise to phosphorescence.[1] In the present MPIMS study, we are able to observe very small but reproducible signals of CH3CO and DOCO in CH3COCOOD (d1-PA) photodissociation (see Figure 5.7). C-C bond fission near the carbonyl group is common in dissociation of many carbonyls (called Norris Type I reaction), and has been seen in PA following excitation to S2 and S3. Norrish type I reactions typically have relatively low barriers to dissociation on T1. Once on T1, PA can dissociate via spin-allowed channels to singlet CO2 + triplet MHC and doublet CH3CO + doublet DOCO. These two channels have comparable ∆H values: 83.4 kcal/mol for the former and 81.8 kcal/mol for the latter, which are also similar, within their error bars, to the excitation energy of the present experiment (81.4 kcal/mol). The barrier height on T1 to give CH3CO + HOCO has been 151 estimated at 14.7 kcal/mol, [25-26] slightly above the energy available in our experiments. However, PA molecules at the Boltzmann tail of the energy distribution at 294 K may be able to react on T1 to give rise to the small observed signals of the radical products. We find it intriguing that we observe the Norrish Type I reaction channel only with CH3COCOOD. The reason for this might be that the carboxylic D atom tunnels less efficiently through the barrier to S1-HT than the H counterpart, thereby lengthening the lifetime of S1 and enhancing the probability of intersystem crossing to T1 near the S1 minimum. This supports the hypothesis that the H (or D) transfer from S1 to S1-HT through the barrier plays a pivotal role in the ensuing dynamics. Experiments that explore the dependence of the triplet pathway yield on excitation energy would be enlightening. Figure 5.8 Possible state-specific dissociation pathways of PA excited at 351 nm. Energies are in kcal/mol. A 351 nm photon supplies 81.5 kcal/mol of energy. While Norrish Type I reaction on T1 is a minor channel at 351 nm, decarboxylation is the major dissociation pathway for PA and d1-PA. Based on the relative product yields obtained in our measurements, we estimate a C2H4O to CO2 ratio of 0.5 ± 0.07. A CH3CHO to 152 CO2 product ratio less than unity (0.2–0.6, depending on pressure) was estimated by Yamamoto and Back who attributed it to undetected vinyl alcohol or loss of MHC by reaction with O2.[1] Although in our experiments we can now rule out the former, secondary (bimolecular) reactions of MHC might account for some of its loss. MHC lifetimes in the order of hours have been previously observed in the cold Ar matrix, but the low observed yield of MHC and its rapid disappearance in the MPIMS experiments cannot be attributed solely to tunneling and isomerization. In the next section we present the first evidence of participation of MHC in a bimolecular secondary reaction. Possible secondary reaction of MHC In addition to the mass peak at m/z=88.02, which is assigned to PA, we observe a peak at m/z=88.05 in the TOF spectra obtained at PI energies less than 9.95 eV (where PA does not ionize). This new mass peak corresponds to the molecular formula C4H8O2, i.e. double the mass of MHC, and is henceforth denoted M because its identity is unknown. It is slightly heavier than the parent mass of PA (C3H4O3) and, furthermore, it is formed only following 351 nm excitation, as shown in Figure 5.9. We are able to observe species M at ionization energies as low as 7.7 eV. Several molecular species can be assigned the molecular formula C4H8O2. We tested acetoin as a candidate because it has been previously observed in photolysis of PA in aqueous solutions.[27-28] We were not able to conclusively determine the identity of species M based on the obtained PI spectrum of acetoin, and believe that acetoin or its enol form (2-butene- 2,3-diol) maybe be involved. We found that in photodissociation of d1-PA, the M peak shifted 153 to m/z=90, as shown in Figure 5.10(a). The double deuteration suggests that M is a result of a bimolecular reaction between two singly deuterated species. Figure 5.9 a) Plot of kinetic time vs. m/z in the vicinity of m/z=88. The pre- photodissociation background is not subtracted. Integrating over kinetic times b) when all products have been pumped out (65 – 84 ms) c) when photodissociation products are present in the tube (0 – 30 ms) and shows two different mass peaks. Integrating over the lighter mass gives time trace d) showing 0.3% depletion consistent with PA and e) the heavier mass (species M) shows time trace resembling a stable photoproduct. Schreiner and coworkers have recently demonstrated the formation of glycoaldehyde in an Ar matrix, and proposed a reaction between hydroxycarbene and formaldehyde as a 154 possible source.[14] A reaction between vibrationally excited MHC produced in the photodissociation and one of the C2H4O photoproducts is unlikely under our experimental conditions. The number of collisions between MHC and the C2H4O isomers (including another MHC) is less than one per millisecond (~8×10 −2 ), assuming that photoexcited PA results in MHC with a unity quantum yield. Thus, the possibility of product formation by recombination of two colliding C2H4O molecules is ruled out. Figure 5.10 a) TOF spectra of PA (red trace) and d1-PA (blue trace, inverted for clarity) showing the observed change in mass peaks due to deuteration. b) Species M signal changes linearly upon change in PA concentration and is independent of the mode of PA sample preparation demonstrated by two different methods – He bubbled through a glass bulb containing PA (red circles) and another delivered via a cylinder (blue triangles) c) At low PA partial pressure (~0.2 mTorr) species M shows a slower rise time while, VA, a primary photoproduct shows a rise time limited by the instrument response time. The plots shown in this figure are all obtained at a photon energy of 9.65 eV. 155 The number of collisions between MHC and PA, however, can be as high as 25 per millisecond under our experimental conditions. Such collisions may result in decarboxylative condensation to create the covalently bonded species M. In order to determine the order of the reaction producing M, we studied it as a function PA concentration. At higher partial pressures we used a bubbler, as described in section 5.2 , and to reach lower partial pressures we diluted PA further in a gas cylinder. This allowed us to vary the concentration by a factor of 10. We find that, as expected, the signals of primary photoproducts such as vinyl alcohol, acetaldehyde and CO2 depend linearly on PA partial pressure (see Figure 5.10(b)). We also observed a linear concentration dependence of the signal of species M on PA concentration. Because of the low absorption cross section at 351 nm, the initial concentration of the MHC photoproduct is very low, and thus the concentration of PA does not change significantly even when MHC is completely consumed by the reaction. Thus, we can treat this reaction as pseudo first order, for which the observed linear dependence of the signal of the M product on PA concentration is expected. However, when the rate of the MHC reaction slows down at low PA concentrations, the rise time of M can become longer than the instrument response time, as seen in Figure 5.10(c). This proves that M is indeed formed by a bimolecular reaction. Assuming a reaction coefficient, k = 1x10 -11 cm 3 molecules -1 s -1 , ten times lower than the maximum gas phase kinetic rate between two neutrals, we obtain a first order reaction rise time of t1/2 ≈ 500 μs for PA partial pressure of 5 mTorr, which is similar to the response time of the instrument. This explains also the fast removal time of MHC shown in Figure 5.6. However, at the lowest 156 partial pressure used in our experiments, the rise time of the M product finally becomes longer than the response time, as seen in Figure 5.10(c). In considering possible bimolecular reaction mechanisms, we recall that carboxylic acids are known to form H-bonded dimers. A recent study by Vaida and coworkers showed that the Tt-Tt dimer is relatively stable.[6] The low concentration of the Tt isomer (3%) however, indicates that the gas phase results in the present work are dominated by the Tc isomer. MHC and PA may also be H-bonded, which may catalyze their bimolecular reaction. A condensation reaction ending in decarboxylation could be energetically feasible due to the nucleophilicity of MHC and the electrophilicity of carbon centers on PA. We also cannot rule out the possibility that acetaldehyde and vinyl alcohol products may be formed with significant internal energies, because of their much lower enthalpies relative to MHC, and they may react with PA. Theoretical calculations of the mechanism of the reaction between PA and the C2H4O isomers are needed to better understand nature of the bimolecular reaction producing species M. Conclusion We show that the photodissociation dynamics of a deceptively simple molecule like PA can be extremely complicated because of a) the interactions among excited states and b) participation of reactive intermediates whose final outcomes can be governed by the nature of these excited states. At 351 nm, PA photodissociation results in MHC and CO2 but the fate of MHC can be dictated by the environment. Following isomerization, both acetaldehyde and vinyl alcohol, the thermodynamically and kinetically favored products, respectively, are 157 formed. While most MHC products isomerize to these stable isomers, some can also live sufficiently long to undergo secondary reactions with PA. Preliminary evidence suggests the existence of a pseudo-first order reaction between PA and MHC (or its isomers) to form a new species with molecular formula C4H8O2. Following excitation close to the S1 state minimum, we see evidence that PA lives longer than a picosecond. At 351 nm, some of the photoexcited PA is expected to relax to the ground state via a conical intersection where it decarboxylates.[10] Existence of a possible barrier to H-exchange in the excited state is inferred from d1-PA photodissociation. The slower tunneling rate of the D atom in d1-PA to this state may enhance the probability of intersystem crossing from the S1 state to the close- lying T1 surface leading to CH3CO and DOCO products. Theoretical insights into the nature of PA’s first singlet excited state and quasi-classical trajectory simulations will aid in quantifying PA’s photodissociation pathways. References [1] Yamamoto, S.; Back, R., The photolysis and thermal decomposition of pyruvic acid in the gas phase. Can. J. Chem. 1985, 63 (2), 549-554. [2] Horowitz, A.; Meller, R.; Moortgat, G. K., The UV–VIS absorption cross sections of the α- dicarbonyl compounds: pyruvic acid, biacetyl and glyoxal. J. Photochem. Photobiol. A: Chem. 2001, 146 (1-2), 19-27. [3] Berges, M. G.; Warneck, P., Product quantum yields for the 350 nm photodecomposition of pyruvic acid in air. Ber. Bunsenges. Phys. Chem. 1992, 96 (3), 413-416. [4] Vesley, G. F.; Leermakers, P. A., The photochemistry of α-keto acids and α-keto esters. III. Photolysis of pyruvic acid in the vapor phase. J. Phys. Chem. 1964, 68 (8), 2364- 2366. 158 [5] Reed Harris, A. E.; Doussin, J.-F.; Carpenter, B. K.; Vaida, V., Gas-phase photolysis of pyruvic acid: The effect of pressure on reaction rates and products. J. Phys. Chem. A 2016, 120 (51), 10123-10133. [6] Blair, S. L.; Reed Harris, A. E.; Frandsen, B. N.; Kjaergaard, H. G.; Pangui, E.; Cazaunau, M.; Doussin, J.-F.; Vaida, V., Conformer-Specific Photolysis of Pyruvic Acid and the Effect of Water. J. Phys. Chem. A 2020, 124 (7), 1240-1252. [7] Reed Harris, A. E.; Cazaunau, M.; Gratien, A.; Pangui, E.; Doussin, J.-F.; Vaida, V., Atmospheric Simulation Chamber Studies of the Gas-Phase Photolysis of Pyruvic Acid. J. Phys. Chem. A 2017, 121 (44), 8348-8358. [8] Plath, K. L.; Takahashi, K.; Skodje, R. T.; Vaida, V., Fundamental and overtone vibrational spectra of gas-phase pyruvic acid. J. Phys. Chem. A 2009, 113 (26), 7294-7303. [9] Takahashi, K.; Plath, K. L.; Skodje, R. T.; Vaida, V., Dynamics of vibrational overtone excited pyruvic acid in the gas phase: Line broadening through hydrogen-atom chattering. J. Phys. Chem. A 2008, 112 (32), 7321-7331. [10] Chang, X.-P.; Fang, Q.; Cui, G., Mechanistic photodecarboxylation of pyruvic acid: Excited-state proton transfer and three-state intersection. J. Chem. Phys. 2014, 141 (15), 154311. [11] Sutradhar, S.; Samanta, B. R.; Fernando, R.; Reisler, H., Spectroscopy and Two-Photon Dissociation of Jet-Cooled Pyruvic Acid. J. Phys. Chem. A 2019, 123 (28), 5906-5917. [12] Schreiner, P. R.; Reisenauer, H. P.; Ley, D.; Gerbig, D.; Wu, C.-H.; Allen, W. D., Methylhydroxycarbene: Tunneling control of a chemical reaction. Science 2011, 332 (6035), 1300-1303. [13] Baly, E. C. C.; Heilbron, I. M.; Barker, W. F., CX.—Photocatalysis. Part I. The synthesis of formaldehyde and carbohydrates from carbon dioxide and water. J. Chem. Soc., Trans. 1921, 119, 1025-1035. [14] Eckhardt, A. K.; Linden, M. M.; Wende, R. C.; Bernhardt, B.; Schreiner, P. R., Gas-phase sugar formation using hydroxymethylene as the reactive formaldehyde isomer. Nat. Chem. 2018, 10 (11), 1141-1147. [15] Perkins, R. J.; Shoemaker, R. K.; Carpenter, B. K.; Vaida, V., Chemical equilibria and kinetics in aqueous solutions of zymonic acid. J. Phys. Chem. A 2016, 120 (51), 10096- 10107. [16] Rapf, R. J.; Perkins, R. J.; Carpenter, B. K.; Vaida, V., Mechanistic description of photochemical oligomer formation from aqueous pyruvic acid. J. Phys. Chem. A 2017, 121 (22), 4272-4282. 159 [17] Hollenstein, H.; Akermann, F.; Günthard, H. H., Vibrational analysis of pyruvic acid and D-, 13 C- and 18 O-labelled species: Matrix spectra, assignments, valence force field and normal coordinate analysis. Spectrochimica Acta Part A: Molecular Spectroscopy 1978, 34 (11), 1041-1063. [18] Kamei, S.; Okuyama, K.; Abe, H.; Mikami, N.; Ito, M., Mode selectivity in intersystem crossing: glyoxal, methylglyoxal, and biacetyl. J. Phys. Chem. 1986, 90 (1), 93-100. [19] Wang, X.; Perry, D. S., An internal coordinate model of coupling between the torsion and C–H vibrations in methanol. J. Chem. Phys. 1998, 109 (24), 10795-10805. [20] Gurnick, M.; Chaiken, J.; Benson, T.; McDonald, J., Vibrational and rotational spectroscopy of the first electronically allowed transition of α‐dicarbonyls. J. Chem. Phys. 1981, 74 (1), 99-105. [21] Cool, T. A.; Nakajima, K.; Mostefaoui, T. A.; Qi, F.; McIlroy, A.; Westmoreland, P. R.; Law, M. E.; Poisson, L.; Peterka, D. S.; Ahmed, M., Selective detection of isomers with photoionization mass spectrometry for studies of hydrocarbon flame chemistry. J. Chem. Phys. 2003, 119 (16), 8356-8365. [22] Rosenfeld, R. N.; Weiner, B., Energy disposal in the photofragmentation of pyruvic acid in the gas phase. J. Am. Chem. Soc. 1983, 105 (11), 3485-3488. [23] Johnston, H. S.; Heicklen, J., Tunnelling corrections for unsymmetrical Eckart potential energy barriers. J. Phys. Chem. 1962, 66 (3), 532-533. [24] Bell, R. P., The tunnel effect in chemistry. Springer: 2013. [25] Dhanya, S.; Maity, D. K.; Upadhyaya, H. P.; Kumar, A.; Naik, P. D.; Saini, R. D., Dynamics of OH formation in photodissociation of pyruvic acid at 193 nm. J. Chem. Phys. 2003, 118 (22), 10093-10100. [26] da Silva, G., Decomposition of pyruvic acid on the ground-state potential energy surface. J. Phys. Chem. A 2016, 120 (2), 276-283. [27] Leermakers, P. A.; Vesley, G. F., The photochemistry of α-keto acids and α-keto esters. I. Photolysis of pyruvic acid and benzoylformic acid. J. Am. Chem. Soc. 1963, 85 (23), 3776-3779. [28] Griffith, E. C.; Carpenter, B. K.; Shoemaker, R. K.; Vaida, V., Photochemistry of aqueous pyruvic acid. Proc. Nat. Acad. Sci. 2013, 110 (29), 11714-11719. 160 Electronic states of hydroxycarbenes Introduction Because of their high reactivity, hydroxycarbenes have been difficult to isolate for detailed studies of their photophysics and photochemistry. Reisler and coworkers generated HCOH in the gas-phase photodissociation of CH2OH via the H + HCOH channel and characterized its internal states from the KE distributions of the H atom cofragments.[1-4] Although indirect, so far this has been the only measurement of hydroxycarbenes produced via photolysis and other attempts to photolytically produce and detect the carbenes have resulted in the detection of only their most stable isomers — formaldehyde(for hydroxycarbene) and acetaldehyde (for methylhydroxycarbene).[5-10] To that end, decarboxylation of α-keto acids such as glyoxylic acid and PA has proven to be a reliable source of HC and MHC, respectively, especially for spectroscopic studies.[5- 6,11-15] Schreiner and coworkers succeeded in trapping HC and MHC in an argon matrix. Both carbenes were probed by UV-Vis absorption spectroscopy and IR spectroscopy.[11-12] Their experiments validated the existence of a high potential barrier to isomerization to the 161 respective aldehydes. Furthermore, in a joint experimental and theoretical study, they found that the lifetimes were limited by H atom tunneling through the barrier to form the aldehydes. Lifetimes of 2 and 1 hour were measured for HC and MHC, respectively.[11-12] Figure 6.1 Photolytic generation and stabilization of hydroxycarbenes is extremely difficult. They quickly isomerize to more stable conformers and in most cases only the aldehydic isomer is spectroscpically observed The infrared (IR) vibrational spectra of HC and MHC in the Ar matrix were in good agreement with a later study carried out in He nanodroplets.[13] Koziol et al. computed the IR and photoelectron spectra of cis- and trans-HC by accurate vibrational configuration interaction calculations.[16-17] Ionization of HC and MHC was also studied theoretically, and vertical and adiabatic energies were reported.[18] However, challenges remain in generating, isolating and detecting these intermediates in the collisionless environment of molecular beams, and studying their dissociation dynamics. Resonance enhanced multiphoton ionization (REMPI) detection via Rydberg states is known to be a sensitive and state-selective detection method in molecular beams. Hydroxycarbenes should have several Rydberg states centered on carbon in addition 162 to the S1 states described above; these can be used to facilitate state-selective REMPI detection. This chapter describes the electronic structure calculations of electronically excited states of the first two members of the hydroxycarbene family, namely, the cis- and trans- isomers of HC and MHC. It is found that above the S1 state and below the ionization threshold, HC and MHC have several Rydberg states in the n=3 manifold. The presence of an additional methyl group in MHC alters the charge distributions, and excitation and ionization energies in predictable ways. To aid in the qualitative understanding of the nature of the Rydberg states and their electron-core interactions, we also compute the Natural Transition Orbitals (NTOs). We gain further information on the oscillator strengths and the direction of the transition dipole moments, which are useful in designing detection schemes, from the shapes of the orbitals involved in the transitions and their symmetries. We then discuss the orbital shapes and how they relate to the computed excitation energies. Possible experimental detection schemes for HC and MHC are also mentioned towards the end. Computational details The ground-state geometries of the neutral and cationic forms of the hydroxycarbenes are optimized at the MP2/aug-cc-pVTZ level. We used the equation-of- motion coupled-cluster method with single and double substitutions (EOM-CCSD). We computed the excitation energies (vertical and adiabatic) and oscillator strengths by EOM- EE-CCSD, and ionization energies (IEs) by EOM-IP-CCSD.[19-21] Core electrons were frozen, and the aug-cc-pVTZ basis set was employed in all EOM calculations. The results agree well 163 with previously reported results for HC and MHC of the excitation energies to S1 and T1, neutral and cation geometries, cis-trans energy gaps, and ionization energies (presented in the next section). All the electronic structure calculations were carried out using Q-CHEM.[27- 28] To assign excited state characters, we carried out NTO analyses of the EOM-EE-CCSD wavefunctions.[22-25] NTOs provide the most compact representation of the electronic transition. The NTO analysis allows one to easily determine the character of the transition, e.g., to distinguish between valence and Rydberg states, identify charge-transfer states, and assign specific character (π–π *, n–π*, etc.) to the transitions. The Gabedit interface is used to visualize the NTOs. Valence and Rydberg characters to the excited states were assigned by considering the symmetry of the transitions, the character of the NTOs, and the spatial extent of the particle orbitals [defined as the expectation value of the position operators, 〈 𝑟𝑟 𝑥𝑥 〉, 〈 𝑟𝑟 𝑦𝑦 〉, 〈 𝑟𝑟 𝑧𝑧 〉; and an average radius of an orbital computed as 〈 𝑟𝑟 〉= � 〈 𝑟𝑟 𝑥𝑥 〉 2 + 〈 𝑟𝑟 𝑦𝑦 〉 2 + 〈 𝑟𝑟 𝑧𝑧 〉 2 � 0.5 ]. The Mulliken charge distributions in the ground and excited states were also analyzed. To compute adiabatic excitation energies for Rydberg states, the ground-state geometry of the cation was used, which is a common approximation.[26] For the valence states, geometry optimizations were carried out using EOM-EE-CCSD/aug-cc-pVTZ. Throughout the chapter, the axis definition is the following: X and Y lie along the long and short axes of the molecule, respectively, and Z is perpendicular to the plane of the molecule. 164 Results and discussion Molecular orbital framework, electronic configurations, and relevant geometries The bonding in hydroxycarbenes can be rationalized within the molecular orbital (MO) framework shown in Figure 6.2 for trans-HC (similar description applies to cis-HC). The electronic configuration of HC in the ground electronic state can be written as (1sO) 2 (1sC) 2 (σOH) 2 (σCH) 2 (LPO) 2 (σCO) 2 (πCO) 2 (LPC) 2 And the electronic configuration of the lowest singlet and triplet excited states is (1sO) 2 (1sC) 2 (σOH) 2 (σCH) 2 (LPO) 2 (σCO) 2 (πCO) 2 (LPC) 1 (π*CO) 1 . Figure 6.2 A molecular orbital diagram for trans-HC. The sp 2 hybridized C atom orbitals are shown on the left, and the sp 3 hybridized O atom orbitals are shown on the right. LP: Lone Pair; LPC is the sp 2 hydridized carbon orbital in the y-direction which does not mix with any other orbital. LPO is formed from two of the sp 3 hydridized orbitals on oxygen. 165 While methylene has a triplet ground state with a vertical singlet-triplet gap of 0.2 eV,[29] hydroxycarbenes have singlet ground states resulting from the stabilizing interaction between the 2pz orbital on carbon and the lone pair of oxygen.[30] The calculations in this study give vertical S-T gaps of 1.54 and 1.85 eV for HC and MHC, respectively; the corresponding adiabatic values are 1.07 and 1.27 eV. These values agree well with previously reported S-T gaps calculated at the CCSD(T)/aug-cc-pVnZ level for these carbenes (see Table 6.1). Table 6.1 Comparison of S-T gap energy (in kcal/mol) calculated in this work at the EOM- EE-CCSD/aug-cc-pVTZ basis set with previously reported values.[11-12,18] Δ 𝐸𝐸 S − T ad iab at ic (kcal/mol) This work Schreiner et al. a Matus et al. b HC 24.61 27.97 25.3 MHC 29.20 30.64 30.5 Section B in the appendices shows the equilibrium structures of HC and MHC in their neutral, first excited singlet, and cationic states; these parameters agree well with previous studies.[11-12,16-18] In both carbenes, the trans-isomer is the most stable, and in Figure 6.4 the energy differences between the isomers of HC and MHC are listed, which are in good agreement with previously published results.[31-32] The cis-isomers are higher in energy in both HC and MHC due to repulsive interactions between the oxygen lone pairs and the non-bonding carbon electrons. a Calculated using (aug)-cc-p(C)VXZ basis sets with single-reference coupled-cluster theory, incorporating triple excitation and higher-order correlation effects at the CCSDT(Q) level. b Thermochemical calculation at CCSD(T)/aug-cc-pVnZ level of theory where n = D, T, Q, and 5, extrapolated to the complete basis set limit. 166 Excited electronic states The vertical and adiabatic electronic excitation energies of the cis- and trans-isomers of HC and MHC, calculated by EOM-EE-CCSD, are summarized in Table 6.3 and Figure 6.4. Here, attention is drawn to the lowest-lying (n=3) Rydberg states and their possible use for detection of hydroxycarbenes. For completeness, we also report the vertical excitation energies and oscillator strengths of five A’ and five A’’ excited singlet states up to the ionization energy in Table 6.2. The S1 energy of MHC is higher than that of HC due to the destabilization of the particle NTO (LUMO) by antibonding contributions from the C-H bonds of methyl (Figure 6.5). The optimized relaxed geometry of S1 involves a change in the H-O-C- C/H dihedral angle. Figure 6.3 Representation of the orbitals in going from a planar (i) to a twisted geometry (ii) in the first excited state of HC. The first row, represents the stabilization of the electron deficient C orbitals by the O lone pairs. This stabilization can also be explained from the molecular orbital picture (second row). 167 This can be explained by considering changes in the nodal plains and overlap of the relevant molecular orbitals. In the planar ground state geometry, the two filled lone pairs orbitals on oxygen are opposite in phase to the sp 2 hybridized lone pair orbitals on carbon. When the p-type LUMO on carbon is singly occupied in the S1 state, rotation about the CO bond axis enhances the in-phase spatial overlap with the oxygen lone pairs, which has a stabilizing effect on S1.[33] This is illustrated schematically in Figure 6.3 which shows the increased in-phase orbital overlap in going from the Cs symmetry of S0 to the C1 symmetry of S1. Table 6.2 Vertical singlet excitation energy results from EOM-EE-CCSD calculation. Five A’ states (white) and five A” states (gray) are calculated. a) HC cis-HC trans-HC States Energy (eV) Oscillator Strength Type of Excitation Energy (eV) Oscillator Strength Type of Excitation 2 1 A’ 5.34 0.025 3s←n (Ryd) 5.67 0.010 3s←n (Ryd) 3 1 A’ 6.77 0.065 3p←n (Ryd) 6.65 0.036 3p←n (Ryd) 4 1 A’ 7.65 0.057 7.51 0.089 5 1 A’ 8.14 0.026 3d←n (Ryd) 8.23 0.033 3d←n (Ryd) 6 1 A’ 8.63 0.023 8.86 0.008 1 1 A” 3.08 0.010 π*←n (Val) 3.05 0.009 π*←n (Val) 2 1 A” 7.54 0.013 3p←n (Ryd) 7.54 0.026 3p←n (Ryd) 3 1 A” 8.09 0.007 π*←lp(O) (Val) 9.57 0.007 π*←lp(O) (Val) 4 1 A” 9.81 0.002 3d←n (Ryd) 9.80 0.001 3d←n (Ryd) 5 1 A” 10.08 0.005 3s← π (Ryd) 10.25 0.015 168 b) MHC cis-MHC trans-MHC States Energy Oscillator Strength Type of Excitation Energy Oscillator Strength Type of Excitation 2 1 A’ 5.59 0.062 3s←n (Ryd) 5.47 0.007 3s←n (Ryd) 3 1 A’ 6.42 0.065 3p←n (Ryd) 6.37 0.023 3p←n (Ryd) 4 1 A’ 6.74 0.124 6.79 0.137 5 1 A’ 7.61 0.001 3d←n (Ryd) 7.57 0.021 3d←n (Ryd) 6 1 A’ 7.68 0.024 7.67 0.008 1 1 A” 3.49 0.011 π*←n (Val) 3.40 0.010 π*←n (Val) 2 1 A” 6.75 0.042 3p←n (Ryd) 6.72 0.047 3p←n (Ryd) 3 1 A” 7.93 0.006 3d←n (Ryd) 7.90 0.008 3d←n (Ryd) 4 1 A” 8.00 0.009 π*←lp(O) (Val) 8.34 0.004 5 1 A” 8.12 0.004 3d←n (Ryd) 9.06 0.011 π*←lp(O) (Val) The UV absorption spectra of HC and MHC measured in Ar matrix for the trans- isomers involve excitation to S1.[11-12] The absorption spectrum of HC shows an onset at around 2.48 eV (500 nm) and a maximum at 2.90 eV (427 nm). The authors also calculated adiabatic and vertical excitation energies to S1 at 2.40 eV (516 nm) and 2.99 eV (415 nm), respectively.[12] These values are in excellent agreement with the adiabatic and vertical energies [excluding zero-point energy (ZPE) corrections] of 2.46 eV (504 nm) and 3.05 eV (406 nm), obtained in this work. The absorption spectrum of MHC shows an onset at around 460 nm and a maximum absorption at about 393 nm, and the calculated excitation energies to S1 are 2.70 eV (459 nm) and 3.38 eV (367 nm) for the adiabatic and vertical transitions, 169 respectively.[11] They agree with our calculations (excluding ZPE corrections) of 2.69 eV (461 nm) and 3.4 eV (365 nm), respectively. The broad absorption spectrum of HC shows several vibronic bands consistent with a change in the dihedral angle.[12] Moreover, the calculated oscillator strengths for the S1←S0 transitions of these carbenes are small (Table 6.3 and Figure 6.4), in agreement with experiments.[11-12] This is expected due to a poor overlap between the ground and excited state electronic wavefunctions. Cations and Rydberg states Promotion of an electron from the non-bonding doubly occupied HOMO leads to a series of Rydberg states converging to the ground state of the cation. The vertical and adiabatic excitation energies for the n=3 Rydberg manifold are summarized in Table 6.3 and Figure 6.4. The energy order of the states — a low lying valence state, followed by the 3s, 3px, and nearly degenerate 3py and 3pz states (in increasing order of energy), does not change in going from HC to MHC, and is the same for the cis- and trans-isomers. The electronic absorption spectrum of these carbenes is expected to be dominated by transitions to Rydberg states. Except for the S1 state discussed above, no other singlet valence state was found near the n=3 manifold. In fact, only one additional valence state, close to the IE, was found which does not mix with these states (Table 6.2). As expected, the IE of MHC is lower than that of HC due to hyperconjugation with the C-H bonds in the α-position. 170 Figure 6.4 Vertical excitation energies and oscillator strengths calculated using EOM- CCSD/aug-cc-pVTZ for the cis- and trans-isomers of HC (left) and MHC (right). Vertical singlet electronic energies (regular font) and oscillator strengths (bold font, square bracket) are calculated at the ground-state optimized geometries. Vertical IE (black dashed line) and ΔEcis-trans values are also listed. 171 Table 6.3 Vertical and adiabatic excitation energies of singlet Rydberg states, oscillator strengths and quantum defects for HC and MHC. Quantum defects are estimated from the Rydberg formula (see text). State Evert (eV) Eadiab (eV) Oscillator Strength Quantum defect (δ) HC trans-HC 2 1 A’(3s) 5.67 4.87 0.010 1.141 3 1 A’(3px) 6.65 5.96 0.036 0.856 4 1 A’(3py) 7.51 6.88 0.089 0.454 2 1 A’’(3pz) 7.54 6.86 0.026 0.436 1 2 A’(IE) 9.61 8.94 cis-HC 2 1 A’(3s) 5.34 4.76 0.025 1.219 3 1 A’(3px) 6.77 6.01 0.065 0.818 4 1 A’(3py) 7.65 6.93 0.057 0.378 2 1 A’’(3pz) 7.54 6.81 0.013 0.448 1 2 A’(IE) 9.63 8.91 MHC trans-MHC 2 1 A’(3s) 5.47 4.59 0.007 1.017 3 1 A’(3px) 6.37 5.65 0.023 0.694 2 1 A’’(3pz) 6.72 5.99 0.047 0.518 4 1 A’(3py) 6.79 6.11 0.137 0.478 1 2 A’(IE) 8.93 8.21 cis-MHC 2 1 A’(3s) 5.59 4.80 0.085 1.002 3 1 A’(3px) 6.42 5.62 0.073 0.703 2 1 A’’(3pz) 6.74 6.03 0.051 0.546 4 1 A’(3py) 6.75 6.08 0.130 0.540 1 2 A’(IE) 9.00 8.24 172 The electronic excitation energies of the trans- and cis-isomers of the two carbenes are similar. The orientation of the O-H bond appears to have a minimal, if any, effect on the charge distribution in the ion core. In contrast to the trend in S1, the Rydberg excitation energies of MHC are lower than those of HC. As discussed above, the lowering of the IE of MHC is caused by hyperconjugation that leads to stabilization of the cation. As a result, its Rydberg series is lower in energy than in HC. Just the IE, however, does not explain the relative ordering of the 3px, 3py, and 3pz states. It is found that the trend in relative ordering of the 3lm Rydberg energies is similar for HC and MHC and for the cis- and trans-isomers, as discussed below. The excitation energies of Rydberg states of small polyatomic molecules can be approximated by the well-known Rydberg formula, 𝐸𝐸 𝑠𝑠𝑥𝑥 = 𝐼𝐼 𝐸𝐸 − 𝑅𝑅 𝐴𝐴𝑚𝑚 ( 𝑚𝑚 − 𝛿𝛿 ) 2 (6.1) where 𝐸𝐸 𝑠𝑠𝑥𝑥 is the vertical excitation energy, 𝐼𝐼 𝐸𝐸 is the vertical ionization energy of the molecule, 𝑅𝑅 𝐴𝐴𝑚𝑚 = 13.61 eV is the Rydberg constant, 𝑚𝑚 is the principal quantum number of the excited atomic orbital, and 𝛿𝛿 is the so-called quantum defect, which accounts for the interaction of the Rydberg electron with the cation core. The vertical 𝐸𝐸 𝑠𝑠𝑥𝑥 energies obtained from the EOM-CCSD calculations were used in the Rydberg formula to obtain the 𝛿𝛿 values listed in Table 6.3. They represent the strength of the interaction with the ion core, and depend on the shape and size of the ion core and the angular momentum of the participating orbitals. Typical values are 𝛿𝛿 = 0.9–1.2 for s states and 0.3–0.6 for p states. The quantum 173 defect 𝛿𝛿 depends primarily on the size and shape of the ion core, whose geometry is similar to that of the ion. The IE of the molecule and the geometry of its cation are both determined by the redistribution of electron density in the ion core upon ionization. However, one should bear in mind that in neutral molecules Rydberg-valence interactions are common,[26] and can affect the geometry and electronic structure of the excited Rydberg state. It is instructive to compare the 𝛿𝛿 values for the 3p Rydberg states of HC and MHC. Compared to atoms, the Rydberg states of molecules have larger and more delocalized ion cores. While the nodes in the 3p orbitals are still centered on the C atom of the hydroxycarbenes, the Rydberg electron interacts with the entire ion core, and the strength of the electron-core interaction depends on the specific ion core charge distribution. The 3px state is the most affected because it is exposed to the entirety of the positive charge along the C-O-H skeleton, defined here in the X direction. Indeed, the 3px state has the highest quantum defect in both HC and MHC, with the largest 𝛿𝛿 value for HC. The positive charge distribution of the 3px ion core is directed towards H (in HC) and methyl (in MHC). However, the methyl group in MHC carries more of the positive charge than the sole hydrogen in HC. Naively, one would expect the additional methyl group to have an added contribution to the charge delocalization and thus to the quantum defect. However, two opposing factors come into play here. While the distribution of positive charge along several nuclei causes the Rydberg electron to interact with charges distributed over several atoms, as opposed to a single atom, the distribution of electron density throughout the core also induces greater screening by bonding and valence electrons, such as σCC. The final outcome is that the quantum defect of 3px is larger for HC than for MHC. Koziol et al.[34] explained the trends in 174 quantum defects observed in a series of substituted vinyl radicals by similar arguments involving increased screening. Figure 6.5 Particle NTOs, and the hole NTO (labeled HOMO) for the lowest excited states of trans-MHC (left) and trans-HC (right); their directional spatial extents in Å are labeled as 〈 𝑟𝑟 〉( 〈 𝑟𝑟 𝑥𝑥 〉, 〈 𝑟𝑟 𝑦𝑦 〉, 〈 𝑟𝑟 𝑧𝑧 〉); The orientation of the molecules has been kept the same in all pictures. A clear distinction between Rydberg and valence orbitals is provided by inspection of the sizes of the NTOs shown in Figure 6.5. The energies and geometries of the S1 states in HC and MHC were discussed above and illustrated by the LUMO particle NTO. The HOMO and first excited valence orbital are of similar size (1.2-1.3 Å radius). The radius of the Rydberg 175 orbitals ranges from 2.5 to 4.1 Å. It is useful to examine how the quantum defect manifests itself in the shapes of the Rydberg orbitals. The shapes are similar for both isomers (except for the 3py orbital discussed below), and therefore the orbitals of the trans-isomer are used as illustrative examples in Figure 6.5. The Rydberg orbital of the 3s state, the lowest in the series, is not as spherically symmetric as expected for an orbital with zero orbital angular momentum. Figure 6.6 Particle NTOs, and the hole NTO (labeled HOMO) for the lowest excited states of cis-MHC (left) and cis-HC (right); their directional spatial extents in Å are labeled as 〈 𝑟𝑟 〉( 〈 𝑟𝑟 𝑥𝑥 〉, 〈 𝑟𝑟 𝑦𝑦 〉, 〈 𝑟𝑟 𝑧𝑧 〉); The orientation of the molecules has been kept the same in all pictures. 176 The stronger interaction with the core reduces the size of this orbital compared to the 3p orbitals and deforms its shape. The 3px orbital is deformed due to interactions with the ion core along the X axis. The 3py orbitals of the trans-isomer of HC and MHC are the most extended, because the single electron in the excited HOMO provides additional shielding from the positive core in the Y direction. As shown in Figure 6.6, the shapes of the py orbitals of the cis- and trans-isomers of HC are slightly different. This is because the extent of the ion core in the Y direction changes slightly with the orientation of the OH bond. Detection of hydroxycarbenes One of the goals of this study was to devise spectroscopic detection schemes for hydroxycarbenes. For molecular beam studies, the most sensitive technique is REMPI detection. Because their geometries are similar to the corresponding cations, Rydberg states often serve as intermediates in the initial excitation, followed by an efficient ionization step. These results show that there are no valence states close in energy to the 3p Rydberg states. Only weak Rydberg-valence couplings are expected, and thus, REMPI detection via these states should be feasible. In order to assess the detection sensitivity, the oscillator strengths of the transitions to the Rydberg states are also calculated (Table 6.3). One can get an idea about the oscillator strength by inspection of the participating molecular orbitals. The NTOs in Figure 6.5 show that the ground state orbital density of the lone pair on carbon is mostly distributed along the Y axis. Therefore, it is expected that the 3py state would have the highest oscillator strength in both HC and MHC. This is indeed the case for trans-HC and both isomers of MHC, 177 where excitation to the 3py state has a much larger oscillator strength than the other states. In cis-HC, however, the oscillator strengths for transitions to 3py and 3px are comparable (see Table 6.3). In deciding on the feasibility of detection schemes, the stability of the corresponding ion must be considered. Previous theoretical studies of the HC cation found a barrier of ~25 kcal/mol for H elimination reaction.[35-38] The rearrangement barrier to the formaldehyde cation was found to be ~50 kcal/mol. Similar studies of the MHC cation show a barrier height of ~40 kcal/mol for isomerization to the acetaldehyde cation.[39-41] The barriers to hydrogen dissociation and isomerization to vinyl alcohol cation are about 30 kcal/mol. In conclusion, the cations are fairly stable with significant barriers to dissociation and isomerization. The goal, therefore, is to first reach the Rydberg states and from there, assuming that the Rydberg ion core geometry is similar to that of the cation, diagonal Franck- Condon factors for ionization are expected. An estimate of the Franck-Condon factors for excitation from the ground to the Rydberg state can be obtained from the calculated photoelectron spectrum of HCOH reported by Koziol et al.[16] They obtained vibronic progressions in the H-O-C and O-C-H bends, which is expected because in the cation the H-O-C and O-C-H angles are larger than in the ground state due to the increased s-type hybridization on C. In addition, the CO bond is shortened in the cation, which leads to a progression in the CO stretch. An absorption spectrum similar to the photoelectron spectrum is expected for the 3py and 3pz states of HCOH, which have the smallest interactions with the core. Although similar calculations have 178 not been reported for MHC, a behavior similar to HC is expected, because the geometry changes upon ionization are similar to those in HC. Upon excitation to 3py, it is predicted that vibrational progressions will be built on H-O-C and O-C-H bends and the CO stretch. The spectra, of course, will depend also on the internal energy of the generated carbenes, but the stability of the ion and the reasonable cross sections for excitation to the Rydberg states are promising for sensitive REMPI detection. Based on the data summarized in Table 6.3, several detection schemes should be feasible for these carbenes. In using one-color schemes, we prefer that the excitation would reach not far above the ground state of the ion to avoid its dissociation. For HC a (2+1) REMPI scheme via the 3py state using 330 – 360 nm radiation should be feasible. Two color schemes are, of course, also possible with several different combinations. For MHC, a (1+1+1) REMPI scheme via the S1 and 3py states can be achieved at 364 – 406 nm, and looks promising. Conclusions Calculations of excited and ionized states of the cis and trans-isomers of HC and MHC are reported using EOM-CCSD and focusing on low-lying Rydberg states. Vertical and adiabatic excitation energies are reported along with the corresponding oscillator strengths for transitions from the ground electronic state. The shapes and energy ordering of the low- lying Rydberg orbitals using NTO analysis and a simplified molecular orbital scheme are described. The interaction of the Rydberg electrons with the ion core are also discussed. Both HC and MHC have Rydberg 3p states in the 5–8 eV range, which can be accessed via multiphoton excitation schemes, and used for REMPI detection of this carbenes. 179 References [1] Feng, L.; Demyanenko, A. V.; Reisler, H., Competitive C–H and O–D bond fission channels in the UV photodissociation of the deuterated hydroxymethyl radical CH2OD. J. Chem. Phys. 2004, 120 (14), 6524-6530. [2] Feng, L.; Reisler, H., Photodissociation of the Hydroxymethyl Radical from the 2 2 A’’(3 pz) State: H2CO and HCOH Products. J. Phys. Chem. A 2004, 108 (45), 9847-9852. [3] Rodrigo, C. P.; Sutradhar, S.; Reisler, H., Imaging Studies of Excited and Dissociative States of Hydroxymethylene Produced in the Photodissociation of the Hydroxymethyl Radical. J. Phys. Chem. A 2014, 118 (51), 11916-11925. [4] Rodrigo, C. P.; Zhou, C.; Reisler, H., Accessing Multiple Conical Intersections in the 3s and 3px Photodissociation of the Hydroxymethyl Radical. J. Phys. Chem. A 2013, 117 (46), 12049-12059. [5] Back, R.; Yamamoto, S., The gas-phase photochemistry and thermal decomposition of glyoxylic acid. Can. J. Chem. 1985, 63 (2), 542-548. [6] Yamamoto, S.; Back, R., The photolysis and thermal decomposition of pyruvic acid in the gas phase. Can. J. Chem. 1985, 63 (2), 549-554. [7] Mellouki, A.; Mu, Y., On the atmospheric degradation of pyruvic acid in the gas phase. J. Photochem. Photobiol. A: Chem. 2003, 157 (2-3), 295-300. [8] Rosenfeld, R. N.; Weiner, B., Energy disposal in the photofragmentation of pyruvic acid in the gas phase. J. Am. Chem. Soc. 1983, 105 (11), 3485-3488. [9] Berges, M. G.; Warneck, P., Product quantum yields for the 350 nm photodecomposition of pyruvic acid in air. Ber. Bunsenges. Phys. Chem. 1992, 96 (3), 413-416. [10] Reed Harris, A. E.; Doussin, J.-F.; Carpenter, B. K.; Vaida, V., Gas-phase photolysis of pyruvic acid: The effect of pressure on reaction rates and products. J. Phys. Chem. A 2016, 120 (51), 10123-10133. [11] Schreiner, P. R.; Reisenauer, H. P.; Ley, D.; Gerbig, D.; Wu, C.-H.; Allen, W. D., Methylhydroxycarbene: Tunneling control of a chemical reaction. Science 2011, 332 (6035), 1300-1303. [12] Schreiner, P. R.; Reisenauer, H. P.; Pickard IV, F. C.; Simmonett, A. C.; Allen, W. D.; Mátyus, E.; Császár, A. G., Capture of hydroxymethylene and its fast disappearance through tunnelling. Nature 2008, 453 (7197), 906. 180 [13] Leavitt, C. M.; Moradi, C. P.; Stanton, J. F.; Douberly, G. E., Communication: Helium nanodroplet isolation and rovibrational spectroscopy of hydroxymethylene. J. Chem. Phys. 2014, 140 (17), 171102. [14] Takahashi, K.; Plath, K. L.; Axson, J. L.; Nelson, G. C.; Skodje, R. T.; Vaida, V., Dynamics and spectroscopy of vibrational overtone excited glyoxylic acid and 2, 2- dihydroxyacetic acid in the gas-phase. J. Chem. Phys. 2010, 132 (9), 094305. [15] Takahashi, K.; Plath, K. L.; Skodje, R. T.; Vaida, V., Dynamics of vibrational overtone excited pyruvic acid in the gas phase: Line broadening through hydrogen-atom chattering. J. Phys. Chem. A 2008, 112 (32), 7321-7331. [16] Koziol, L.; Mozhayskiy, V. A.; Braams, B. J.; Bowman, J. M.; Krylov, A. I., Ab Initio Calculation of the Photoelectron Spectra of the Hydroxycarbene Diradicals. J. Phys. Chem. A 2009, 113 (27), 7802-7809. [17] Koziol, L.; Wang, Y.; Braams, B. J.; Bowman, J. M.; Krylov, A. I., The theoretical prediction of infrared spectra of trans-and cis-hydroxycarbene calculated using full dimensional ab initio potential energy and dipole moment surfaces. J. Chem. Phys. 2008, 128 (20), 204310. [18] Matus, M. H.; Nguyen, M. T.; Dixon, D. A., Heats of Formation and Singlet− Triplet Separations of Hydroxymethylene and 1-Hydroxyethylidene. J. Phys. Chem. A 2006, 110 (28), 8864-8871. [19] Bartlett, R. J., Coupled-cluster theory and its equation-of-motion extensions. Wiley Interdiscip. Rev. Comput. Mol. Sci. 2012, 2 (1), 126-138. [20] Krylov, A. I., Equation-of-motion coupled-cluster methods for open-shell and electronically excited species: The hitchhiker's guide to Fock space. Annu. Rev. Phys. Chem. 2008, 59. [21] Sneskov, K.; Christiansen, O., Excited state coupled cluster methods. Wiley Interdiscip. Rev. Comput. Mol. Sci. 2012, 2 (4), 566-584. [22] Luzanov, A.; Sukhorukov, A.; Umanskii, V., Application of transition density matrix for analysis of excited states. Theor. Exp. Chem. 1976, 10 (4), 354-361. [23] Mewes, S. A.; Plasser, F.; Krylov, A.; Dreuw, A., Benchmarking excited-state calculations using exciton properties. J. Chem. Theory Comput. 2018, 14 (2), 710–725. [24] Plasser, F.; Bäppler, S. A.; Wormit, M.; Dreuw, A., New tools for the systematic analysis and visualization of electronic excitations. II. Applications. J. Chem. Phys. 2014, 141 (2), 024107. 181 [25] Plasser, F.; Wormit, M.; Dreuw, A., New tools for the systematic analysis and visualization of electronic excitations. I. Formalism. J. Chem. Phys. 2014, 141 (2), 024106. [26] Reisler, H.; Krylov, A. I., Interacting Rydberg and valence states in radicals and molecules: experimental and theoretical studies. Int. Rev. Phys. Chem. 2009, 28 (2), 267-308. [27] Krylov, A. I.; Gill, P. M., Q-Chem: an engine for innovation. Wiley Interdiscip. Rev. Comput. Mol. Sci. 2013, 3 (3), 317-326. [28] Shao, Y.; Gan, Z.; Epifanovsky, E.; Gilbert, A. T.; Wormit, M.; Kussmann, J.; Lange, A. W.; Behn, A.; Deng, J.; Feng, X., Advances in molecular quantum chemistry contained in the Q-Chem 4 program package. Mol. Phys. 2015, 113 (2), 184-215. [29] Vargas, R.; Galván, M.; Vela, A., Singlet− Triplet Gaps and Spin Potentials. J. Phys. Chem. A 1998, 102 (18), 3134-3140. [30] Mueller, P. H.; Rondan, N. G.; Houk, K.; Harrison, J. F.; Hooper, D.; Willen, B. H.; Liebman, J. F., Carbene singlet-triplet gaps. Linear correlations with substituent. pi.-donation. J. Am. Chem. Soc. 1981, 103 (17), 5049-5052. [31] Evanseck, J.; Houk, K., Theoretical predictions of activation energies for 1, 2-hydrogen shifts in singlet carbenes. J. Phys. Chem. 1990, 94 (14), 5518-5523. [32] Räsänen, M.; Raaska, T.; Kunttu, H.; Murto, J., Ab initio studies on carbenes; singlet and triplet conformers and vibrational spectra of hydroxy-, dihydroxy-and methylhydroxy- carbene. J. Mol. Struc.: THEOCHEM. 1990, 208 (1-2), 79-90. [33] Hwang, D.-Y.; Mebel, A. M.; Wang, B.-C., Ab initio study of the addition of atomic carbon with water. Chem. Phys. 1999, 244 (2-3), 143-149. [34] Koziol, L.; Levchenko, S. V.; Krylov, A. I., Beyond vinyl: electronic structure of unsaturated propen-1-yl, propen-2-yl, 1-buten-2-yl, and trans-2-buten-2-yl hydrocarbon radicals. J. Phys. Chem. A 2006, 110 (8), 2746-2758. [35] Osamura, Y.; Goddard, J. D.; Schaefer III, H. F.; Kim, K. S., Near degenerate rearrangement between the radical cations of formaldehyde and hydroxymethylene. J. Chem. Phys. 1981, 74 (1), 617-621. [36] Bouma, W. J.; Macleod, J. K.; Radom, L., An ab initio molecular orbital study of the CH2O +• isomers: The stability of the hydroxymethylene radical cation. International Journal of Mass Spectrometry and Ion Physics 1980, 33 (1), 87-93. 182 [37] Burgers, P. C.; Mommers, A. A.; Holmes, J., Ionized oxycarbenes: hydroxymethylidyne cation ([COH] + ), hydroxymethylene cation ([HCOH] +• ), dihydroxymethylene cation ([C(OH)2] +• ), formoxylium cation ([HCO2] + ) and carboxyl cation ([COOH] + ), their generation, identification, heat of formation, and dissociation characteristics. J. Am. Chem. Soc. 1983, 105 (19), 5976-5979. [38] Wagner, J. P.; Bartlett, M. A.; Allen, W. D.; Duncan, M. A., Tunneling Isomerizations on the Potential Energy Surfaces of Formaldehyde and Methanol Radical Cations. ACS Earth and Space Chemistry 2017, 1 (6), 361-367. [39] Apeloig, Y.; Karni, M.; Ciommer, B.; Depke, G.; Frenking, G.; Meyn, S.; Schmidt, J.; Schwarz, H., [CH3COH] •+ , the central intermediate in the isomerization–dissociation reactions of ionized vinyl alcohol. J. Chem. Soc., Chem. Commun. 1983, (24), 1497-1499. [40] Smith, B. J.; Nguyen, M. T.; Bouma, W. J.; Radom, L., Unimolecular rearrangements connecting hydroxyethylidene (CH3-C-OH), acetaldehyde (CH3-CH=O), and vinyl alcohol (CH2=CH-OH). J. Am. Chem. Soc. 1991, 113 (17), 6452-6458. [41] Terlouw, J. K.; Wezenberg, J.; Burgers, P. C.; Holmes, J. L., New, stable isomers of [C2H4O] +• , and [C2H4O2] +• , the radical cations [CH3COH] +• and [CH3OCOH] +• . J. Chem. Soc., Chem. Commun. 1983, (20), 1121-1123. 183 Future work Introduction This chapter builds on the experimental foundation laid out in the previous chapters which explored the complicated photochemistry of PA. Our initial goal was to study only the unimolecular dissociation dynamics of PA, focusing mainly on the detection and isolation of MHC. Based on the promising preliminary results obtained from bimolecular reactions in the MPIMS measurements, PA photochemistry studies are being extended to include its bimolecular reaction dynamics in the presence of water. We learnt in Chapter 5 that MHC and PA can react to form a higher homologue of MHC, although a concrete evidence for the reaction mechanism is yet to be discovered. As mentioned earlier, PA in the atmosphere is found in both the gas and aqueous phases, and its photochemistry can shed light into the behavior of other carboxylic acids and VOCs.[1-4] Although photolysis under solar radiation is a primary degradation pathway for this molecule, it has been shown to form higher-molecular-weight photoproducts in aqueous medium.[5] Such secondary photoproducts may contribute to the formation of secondary 184 organic aerosols (SOAs).[6-7] Current understanding SOAs, and as a result, their inclusion in climate and atmospheric modeling, is limited due to the lack of lab-based studies on such molecular clusters.[6-11] Recent studies have shown that high-molecular-weight oligomers, responsible for SOAs, can be generated from organic reactions in an aqueous medium.[6,12- 14] However several gaps still remain in the understanding of photoinduced reactions of these molecules.[15-16] At the same time, the interplay between the singlet excited states of PA following photoexcitation is still unclear. In order to identify and characterize the product contribution from each singlet excited state, additional measurements to track the excitation energy is required. The limitation of accessing the S1 state via focused laser radiation in the 350–375 nm range has been discussed in previous chapters. Emission studies from the S1 or higher excited states is another promising alternative to quantify the nature of excited. Because the vibronic bands of the S1←S0 state (from H-PFY) were only tentatively assigned, we also propose doing additional measurements with deuterated PA in order to identify the observed vibrational fingerprints. Therefore, these future experiments present a unique possibility of identifying nascent photoproducts and secondary reactions of PA, arising from its unimolecular dissociation or bimolecular reactions with water or other primary photoproducts. 185 Investigating PA + H 2 O interactions Photoinduced reactions The interaction of PA with water is one of the most fascinating and multi-faceted aspects of its photochemistry. As such, H-bonding facilitates the formation of large clusters of PA with itself and/or with water,[17] though the interactions with water in the gas phase have not been elucidated. Theoretical papers characterized the geometry of H-bonded com- plexes of PA with one and two water molecules and showed that water serves both as donor and acceptor.[18-22] Figure 7.1 Molecular structures of (a) keto-pyruvic acid; (b) enol-pyruvic acid; (c) 2,2- dihydroxypropanoic acid (gem-diol); and (d) parapyruvic acid. H-bonding with water can also facilitate the formation of the covalently-bonded germinal (gem)-diol of PA, 2,2-dihydroxypropanoic acid (Figure 7.1(c)).[22-23] In fact equilibrium exists between the gem-diol and the keto-PA forms, such that at 298 K only 45% of its aqueous solution is in the keto form.[5,24] In addition, H-bonded PA dimers can convert 186 to their covalently bonded form, parapyruvic acid (Figure 7.1(d)) upon UV irradiation in aqueous solution.[25] Very little, however, is known on the primary steps of these reactions in the gas phase under well-controlled conditions. The MPIMS arrangement boasts high sensitivity, species specificity, and the ability to follow in time bimolecular reactions up to about 100 ms. We plan to use this method to study the 351 and 193 nm photochemistry of gas phase PA in the presence of water vapor and other collision partners. Our goal is to identify new species using mass spectroscopy and explore their kinetic behavior. Measurements done at different relative water concentrations will be compared to those obtained in the absence of water in order to identify changes in mass spectra that point to the formation of hydrated products. We will watch especially for changes in the observed mass 61, which has a molecular formula identical to protonated acetic acid, but its formation has not yet been fully explained. Protonated acetic acid has been shown to be a prominent product in the multiphoton ionization of PA clusters at 193 nm, and this was rationalized as indicating proton transfer in the ionization process of dimers and larger clusters.[17] Once the background photoionization signals are characterized, we will examine the influence of UV excitation on the photochemistry of PA at 351 and 193 nm in the presence of gaseous water. Specifically, we will look for kinetic evidence of bimolecular reactions that lead to the formation of the geminal-diol and identify their sources and mechanisms. We also plan on evaluating how the interaction with water vapor affects the acetaldehyde to vinyl alcohol ratio following 351 and 193 nm photolysis. 187 Photodissociation dynamics of dimers and higher order clusters Theoretical studies have suggested that keto-enol tautomerization can be catalyzed by water, again through the formation of H-bonded cyclic complexes that facilitate hydrogen exchange. In fact, Valadbeigi and Farrokhpour have shown the dramatic effect of one water molecule on lowering the barrier in keto-enol tautomerization in PA from about 68 kcal/mole (24,000 cm -1 ) to about 31.6 kcal/mole (11,000 cm -1 ).[18] One can imagine that even greater reduction in the keto-enol barrier may be reached with a cyclic cluster with two water molecules, which would reduce ring strain. More recent calculations by Blair et al. revealed that addition of water changes the conformer population of gas-phase PA. While the Tc conformer (Figure 7.1(a)) is more abundant in the gas-phase, the Tt conformer (Figure 7.1(b)) is favored in the aqueous phase.[26] Because photolysis rates are dominated by the Tc conformer, the gas-phase photolysis rate constants of PA are decreased in the presence of water. It will be useful to study the nature and extent of H-bonding in PA-H2O dimer and possibly PA-(H2O)2 trimer. In the past, the Reisler group has studied vibrational predissociation of several H-bonded clusters.[27-28] Recently, Kwasniewski et al. addressed vibrational predissociation of H-bonded clusters of water with organic molecules. From their experiments on phenol-H2O clusters, they observed that the water fragments formed from excitation of the H-bonded OH bond of the phenol moiety had a different dissociation mechanism compared to those formed upon excitation of the free OH-bond of water.[29] 188 Figure 7.2 Vibrational spectrum of aqueous PA in the mid-IR range adapted from reference [23]. Several characteristic OH stretching frequencies in the 2800–3800 cm -1 range are labelled. Based on their calculations, Maroń et al. identified several OH stretching frequencies in aqueous solutions of PA as shown in Figure 7.2.[23] The UV laser can be parked at a wavelength corresponding to the 2+1 REMPI via the C � 1 B1(000) ← X �1 A1 (000) transition for a low rotational level of water. The IR laser will can then be scanned in the 2800–3800 cm -1 range with the goal of looking for enhancements around the identified ν OH positions in Figure 7.2. Different product state distributions are expected for H2O products arising from the two different PA conformers. The nature of PA-H2O interaction can be insightful in understanding its behavior in aerosols. 189 Nature of the S 1 state Light emission studies To date there has been only one published report of light emission in the gas phase photodissociation of PA. Yamamoto and Back recorded the S1← S0 absorption spectrum of 5 Torr of PA by using wavelength selected Hg-Xe high pressure arc lamp, and monitored light emission as a function of excitation wavelength with a photomultiplier tube (PMT).[30] They found that emission was most intense at the long wavelength edge of the absorption curve (~380 nm) and the light intensity decreased gradually at shorter wavelengths, e.g. to about a third at 350 nm. There are no reports of the wavelengths or lifetimes of the emitted light. There is one report of emission in solution and thin films of PA, where the absorption is blue shifted with a peak at 320 nm, and it was attributed to phosphorescence from T1.[31] To propose a mechanism for the one- and two-photon dissociation dynamics via S1, we need to characterize the light emission under the same conditions employed in our present experiments. Previous studies report that products are observed close to the band origin, where emission was also observed.[30,32] However, these studies were all done at room temperature or above, where excitation can involve higher rovibrational levels of PA than in the molecular beam experiments that we carried out. We wish to find out whether the persistence of light emission even at 350 nm indicates, as suggested by theory,[33] that there 190 are two decaying populations on S1, the keto-form that decays to T1/S0 and the H-transferred form that decarboxylates efficiently. To study light emission, the existing LIF arrangement at USC can be reinstalled to detect lifetimes as long as 100 μs. In the past, Qian et al. [34] have succeeded in doing so for jet-cooled NCNO following S1←S0 (π*←n) excitation by locating the surface of the photomultiplier tube (PMT) 30 cm downstream from and in line with the supersonic jet exit. In the proposed experiments, PA samples can be premixed with He or Ar to total pressures of 300–1000 Torr, and expanded through the pulsed nozzle into the LIF chamber pumped to ~10 -6 Torr. Following laser excitation in the collision free region downstream from the nozzle orifice, the time-resolved total emission and lifetimes can be measured. Observation of long lifetimes correlated with longer emission wavelengths will suggest phosphorescence from T1. The PMT will have a range of 300-800 nm, and a series of existing baffles and slits will reduce scattered light from the excitation laser. These experiments can be used to estimate how dissociation competes with light emission. A second LIF arrangement[34-35] can be employed to study the effect of S1 relaxation on light emission in the presence of collisions under conditions similar to the ones employed in the ALS experiment. Slow flowing gaseous mixtures of 10-50 mTorr PA in varying pressures of He, Ar, N2, O2 and H2O can be used to measure light emission as a function of experimental conditions. These experiments will aid in understanding the relevant relaxation processes. 191 Experiments with deuterated PA The rovibrational bands in the jet-cooled spectrum of the S1←S0 state were discussed in Chapters 3 and 5 but these bands were only tentatively assigned. In order to fully identify them, additional spectroscopic measurements are required. This can be accomplished by isotope labeling of the hydrogens of PA. Because the observed progressions near the 0-0 band were assigned to those arising from CH3 torsion and OCCO torsion, their frequencies are expected to change upon partial or full deuteration of PA. Historically this has been used to identify the spectral signatures of several gas phase molecules.[36-39] Furthermore, we can also selectively record the PFY spectra and KER distributions of H and D fragments from partially deuterated PA with laser radiation in the 350–375 nm range. The PFY spectrum will reflect changes in the structural parameters which will assist us in assignment of the S1 state spectrum. The KER distributions for H vs D fragments will help distinguish between the fragmentation pathways arising from C-H or O-H dissociation. The ΔHrxn for O-H fragmentation is 6000 cm -1 higher than that of C-H fragmentation (see chapter 4). The barriers to their dissociation, however, are different and the breaking of the stronger O-H bond may be facilitated by H-bonding in Tc isomer of PA. As a result, not just the KEmax values, but the KER distributions and the corresponding anisotropy parameters are expected be different for both H and D fragments from partially deuterated PA. References [1] Kawamura, K.; Tachibana, E.; Okuzawa, K.; Aggarwal, S.; Kanaya, Y.; Wang, Z., High abundances of water-soluble dicarboxylic acids, ketocarboxylic acids and alpha- 192 dicarbonyls in the mountaintop aerosols over the North China Plain during wheat burning season. Atmos. Chem. Phys. 2013, 13 (16), 8285-8302. [2] Carlton, A. G.; Turpin, B. J.; Lim, H. J.; Altieri, K. E.; Seitzinger, S., Link between isoprene and secondary organic aerosol (SOA): Pyruvic acid oxidation yields low volatility organic acids in clouds. Geophys. Res. Lett. 2006, 33 (6), L06822. [3] Eger, P. G.; Schuladen, J.; Sobanski, N.; Fischer, H.; Karu, E.; Williams, J.; Riva, M.; Zha, Q.; Ehn, M.; Quéléver, L. L., Pyruvic acid in the boreal forest: gas-phase mixing ratios and impact on radical chemistry. Atmos. Chem. Phys. 2020, 20 (6), 3697–3711. [4] Warneck, P., Multi-phase chemistry of C 2 and C 3 organic compounds in the marine atmosphere. J. Atmos. Chem. 2005, 51 (2), 119-159. [5] Griffith, E. C.; Carpenter, B. K.; Shoemaker, R. K.; Vaida, V., Photochemistry of aqueous pyruvic acid. Proc. Nat. Acad. Sci. 2013, 110 (29), 11714-11719. [6] Ervens, B.; Turpin, B.; Weber, R., Secondary organic aerosol formation in cloud droplets and aqueous particles (aqSOA): a review of laboratory, field and model studies. Atmos. Chem. Phys. Discuss. 2011, 11 (8). [7] Kanakidou, M.; Seinfeld, J.; Pandis, S.; Barnes, I.; Dentener, F.; Facchini, M.; Van Dingenen, R.; Ervens, B.; Nenes, A.; Nielsen, C., Organic aerosol and global climate modelling: a review. 2005. [8] Perraud, V.; Bruns, E. A.; Ezell, M. J.; Johnson, S. N.; Yu, Y.; Alexander, M. L.; Zelenyuk, A.; Imre, D.; Chang, W. L.; Dabdub, D., Nonequilibrium atmospheric secondary organic aerosol formation and growth. Proc. Nat. Acad. Sci. 2012, 109 (8), 2836-2841. [9] Ervens, B.; Carlton, A. G.; Turpin, B. J.; Altieri, K. E.; Kreidenweis, S. M.; Feingold, G., Secondary organic aerosol yields from cloud-processing of isoprene oxidation products. Geophys. Res. Lett. 2008, 35 (2), L02816. [10] Lim, Y.; Tan, Y.; Perri, M.; Seitzinger, S.; Turpin, B., Aqueous chemistry and its role in secondary organic aerosol (SOA) formation. Atmos. Chem. Phys. Discuss. 2010, 10 (6), 10521–10539. [11] Ervens, B.; George, C.; Williams, J.; Buxton, G.; Salmon, G.; Bydder, M.; Wilkinson, F.; Dentener, F.; Mirabel, P.; Wolke, R., CAPRAM 2.4 (MODAC mechanism): An extended and condensed tropospheric aqueous phase mechanism and its application. J. Geophys. Res. Atmos. 2003, 108 (D14). [12] Tan, Y.; Lim, Y.; Altieri, K.; Seitzinger, S.; Turpin, B., Mechanisms leading to oligomers and SOA through aqueous photooxidation: insights from OH radical oxidation of acetic acid. Atmos. Chem. Phys. Discuss. 2012, 12 (2), 801–813. 193 [13] De Haan, D. O.; Corrigan, A. L.; Tolbert, M. A.; Jimenez, J. L.; Wood, S. E.; Turley, J. J., Secondary organic aerosol formation by self-reactions of methylglyoxal and glyoxal in evaporating droplets. Environ. Sci. Technol. 2009, 43 (21), 8184-8190. [14] Ervens, B.; Gligorovski, S.; Herrmann, H., Temperature-dependent rate constants for hydroxyl radical reactions with organic compounds in aqueous solutions. Phys. Chem. Chem. Phys. 2003, 5 (9), 1811-1824. [15] Guzman, M.; Colussi, A.; Hoffmann, M., Photoinduced oligomerization of aqueous pyruvic acid. J. Phys. Chem. A 2006, 110 (10), 3619-3626. [16] Donaldson, D.; George, C.; Vaida, V., Red sky at night: Long-wavelength photochemistry in the atmosphere. Environ. Sci. Technol. 2010, 44 (14), 5321–5326. [17] Grygoryeva, K.; Ončák, M.; Pysanenko, A.; Fárník, M., Pyruvic acid proton and hydrogen transfer reactions in clusters. Phys. Chem. Chem. Phys. 2019, 21 (16), 8221-8227. [18] Valadbeigi, Y.; Farrokhpour, H., Theoretical study on keto–enol tautomerism and isomerization in pyruvic acid. Int. J. Quantum Chem 2013, 113 (21), 2372-2378. [19] Buemi, G., DFT study of the hydrogen bond strength and IR spectra of formic, oxalic, glyoxylic and pyruvic acids in vacuum, acetone and water solution. J. Phys. Org. Chem. 2009, 22 (10), 933-947. [20] Schnitzler, E. G.; Seifert, N. A.; Ghosh, S.; Thomas, J.; Xu, Y.; Jäger, W., Hydration of the simplest α-keto acid: a rotational spectroscopic and ab initio study of the pyruvic acid– water complex. Phys. Chem. Chem. Phys. 2017, 19 (6), 4440-4446. [21] Yang, X.; Orlova, G.; Zhou, X.; Leung, K., A DFT study on the radical, monomer and dimer of α-keto pyruvic acid: equilibrium structures and vibrational analysis of stable conformers. CPL 2003, 380 (1-2), 34-41. [22] Yang, D.; Zhang, L., Excited-state hydrogen bonding dynamics of pyruvic acid and geminal-diol, 2, 2-dihydroxypropanoic acid in aqueous solution: a DFT/TDDFT study. J. Phys. Org. Chem. 2012, 25 (12), 1391-1394. [23] Maroń, M. K.; Takahashi, K.; Shoemaker, R. K.; Vaida, V., Hydration of pyruvic acid to its geminal-diol, 2, 2-dihydroxypropanoic acid, in a water-restricted environment. Chem. Phys. Lett. 2011, 513 (4-6), 184-190. [24] Eugene, A. J.; Pillar, E. A.; Colussi, A. J.; Guzman, M. I., Enhanced acidity of acetic and pyruvic acids on the surface of water. Langmuir 2018, 34 (31), 9307-9313. 194 [25] Rapf, R. J.; Perkins, R. J.; Carpenter, B. K.; Vaida, V., Mechanistic description of photochemical oligomer formation from aqueous pyruvic acid. J. Phys. Chem. A 2017, 121 (22), 4272-4282. [26] Blair, S. L.; Reed Harris, A. E.; Frandsen, B. N.; Kjaergaard, H. G.; Pangui, E.; Cazaunau, M.; Doussin, J.-F.; Vaida, V., Conformer-Specific Photolysis of Pyruvic Acid and the Effect of Water. J. Phys. Chem. A 2020, 124 (7), 1240-1252. [27] Samanta, A. K.; Czakó, G. b.; Wang, Y.; Mancini, J. S.; Bowman, J. M.; Reisler, H., Experimental and theoretical investigations of energy transfer and hydrogen-bond breaking in small water and HCl clusters. Acc. Chem. Res. 2014, 47 (8), 2700-2709. [28] Zuraski, K.; Kwasniewski, D.; Samanta, A. K.; Reisler, H., Vibrational Predissociation of the HCl–(H2O)3 Tetramer. J. Phys. Chem. letters 2016, 7 (21), 4243-4247. [29] Kwasniewski, D.; Butler, M.; Reisler, H., Vibrational predissociation of the phenol–water dimer: a view from the water. Phys. Chem. Chem. Phys. 2019, 21 (26), 13968-13976. [30] Yamamoto, S.; Back, R., The photolysis and thermal decomposition of pyruvic acid in the gas phase. Can. J. Chem. 1985, 63 (2), 549-554. [31] Arnett, J.; Larson, D.; McGlynn, S., Absorption and emission spectroscopy of pyruvic acids and pyruvate esters. J. Am. Chem. Soc. 1973, 95 (23), 7599-7603. [32] Mellouki, A.; Mu, Y., On the atmospheric degradation of pyruvic acid in the gas phase. J. Photochem. Photobiol. A: Chem. 2003, 157 (2-3), 295-300. [33] Chang, X.-P.; Fang, Q.; Cui, G., Mechanistic photodecarboxylation of pyruvic acid: Excited-state proton transfer and three-state intersection. J. Chem. Phys. 2014, 141 (15), 154311. [34] Qian, C.; Ogai, A.; Reisler, H.; Wittig, C., Correlated product state distributions in the unimolecular reaction of NCNO. J. Chem. Phys. 1989, 90 (1), 209-218. [35] Reisler, H.; Mangir, M.; Wittig, C., LASER KINETIC SPECTROSCOPY OF ELEMENTARY PROCESSES. In Chemical and Biochemical Applications of Lasers, Academic Press: 1980; pp 139-174. [36] Cole, A.; Osborne, G., Vibrational spectra of glyoxal, monodeuteroglyoxal and dideuteroglyoxal. Spectrochim. Acta A 1971, 27 (12), 2461-2490. [37] Gray, A. P.; Lord, R., Rotation-Vibration Spectra of Methyl Amine and Its Deuterium Derivatives. J. Chem. Phys. 1957, 26 (3), 690-705. 195 [38] Uno, T.; Machida, K.; Saito, Y., Out-of-plane vibrations of acetamide and partially N- deuterated acetamide. Spectrochim. Acta A 1971, 27 (6), 833-844. [39] Marushkevich, K.; Khriachtchev, L.; Lundell, J.; Domanskaya, A.; Räsänen, M., Vibrational spectroscopy of trans and cis deuterated formic acid (HCOOD): Anharmonic calculations and experiments in argon and neon matrices. J. Mol. Spectrosc. 2010, 259 (2), 105-110. 196 Appendices A Three-body fragmentation KE max simulations A.1 Synchronous three-body fragmentation MATLAB code 1 close all 2 clear all 3 hv=53400; % total excitation energy 4 dH=31600; % total BDE of both bonds 5 6 % Define masses A, B and C. Important! B should be the central fragment! 7 ma=45; 8 mb=28; 9 mc=15; 10 mac=ma*mc/(ma+mc); % reduced mass 11 E=hv-dH; 12 13 n=1; % define angle step size 14 E_a=zeros(1,180/n+1); 15 E_b=zeros(1,180/n+1); 16 E_c=zeros(1,180/n+1); 17 18 % loop over angles 0 to 180. Can make the values smaller if exact angles 19 % known 20 for th=0:n:180 21 E_b(th/n+1)=E/(1+(mb/mac/4)*(1+(tan(th/2*pi/180))^2)); 197 22 E_c(th/n+1)=E/(4*mc/mb*(cos(th/2*pi/180))^2+(ma+mc)/ma); 23 E_a(th/n+1)=E/(4*ma/mb*(cos(th/2*pi/180))^2+(ma+mc)/mc); 24 end 25 hold on 26 plot([0:n:180],E_a,'r'); 27 plot([0:n:180],E_b,'g'); 28 plot([0:n:180],E_c,'b'); 29 xlabel('TS angle A-B-C') 30 ylabel('max fragment KER') 31 hold off 32 % saves values in an excel file. Replace .xlsx by .txt if you want UTF- 8 33 % encoded text file 34 filename = 'sync_HOCO_CO_CH3.xlsx'; 35 A=[]; 36 A = [[0:n:180]' E_a' E_b' E_c']; % columns are arranged [A, B, C] 37 xlswrite(filename,A); A.2 Synchronous three-body fragmentation MATLAB code 1 close all 2 clear all 3 hv=54000; % total excitation energy 4 dH=42600; % total BDE of both bonds 5 % ABC ==> A + BC ==> A + B + C 6 Int_a=0; 7 ma=45; %order of masses important! See the reaction process above 8 mb=42; 9 mc=1; 10 E1=hv-dH-Int_a; % For step 2 happen min E1 = D_BC 11 12 %initialize energy arrays 13 E_a=zeros(1,E1/10+1); 14 E_b=zeros(1,E1/10+1); 15 E_c=zeros(1,E1/10+1); 16 198 17 %loop over the energy in first step 18 for i=0:10:E1 19 E_a(i/10+1)=i*(mb+mc)/(ma+mb+mc); 20 Ebc=i*(ma)/(ma+mb+mc); 21 Int_bc=E1-i; 22 Eb=Int_bc*mc/(mb+mc); 23 Ec=Int_bc*mb/(mb+mc); 24 E_b(i/10+1)=(((2*Ebc/(mb+mc))^.5+(2*Eb/mb)^.5)^2)*mb/2; 25 E_c(i/10+1)=(((2*Ebc/(mb+mc))^.5+(2*Ec/mc)^.5)^2)*mc/2; 26 end 27 hold on 28 plot([0:10:E1],E_a,'r'); 29 plot([0:10:E1],E_b,'g'); 30 plot([0:10:E1],E_c,'b'); 31 xlabel('KER in first step') 32 ylabel('max fragment KER') 33 hold off 34 35 % saves values in an excel file. Replace .xlsx by .txt if you want UTF- 8 36 % encoded text file 37 filename = 'HOCO_ketene_H.xlsx'; 38 A=[]; 39 A = [[0:10:E1]' E_a' E_b' E_c']; % columns are arranged [A, B, C] 40 xlswrite(filename,A); 199 B HC and MHC geometries B.1 Hydroxycarbene (HC) S0 geometry of cis-hydroxycarbene optimized at the MP2/aug-cc-pVTZ level H 0.9316120240 -0.8075743777 0.0000000000 O 0.5741481585 0.0970737059 -0.0000000000 C -0.7368118639 0.1546627254 -0.0000000000 H -1.1039261084 -0.8969916220 0.0000000000 Nuclear Repulsion Energy = 30.638201 MP2 energy = -114.220014 S0 geometry of trans-hydroxycarbene optimized at the MP2/aug-cc-pVTZ level H 1.0946972316 -0.6448108655 -0.0000000000 O 0.5495211998 0.1552848846 0.0000000000 C -0.7077398040 -0.2236842521 -0.0000000000 H -1.2444280064 0.7446373011 0.0000000000 Nuclear Repulsion Energy = 30.735659 MP2 energy = -114.228119 D0 geometry of cis-hydroxycarbene cation optimized at the MP2/aug-cc-pVTZ level 200 H -1.0911149057 -0.7511035811 0.0000000000 O -0.5545062056 0.0844539309 0.0000000000 C 0.6605549330 0.0150990918 0.0000000000 H 1.3430928313 -0.8436456674 0.0000000000 Nuclear Repulsion Energy = 31.910311 MP2 energy = -113.900137 D0 geometry of trans-hydroxycarbene cation optimized at the MP2/aug-cc-pVTZ level H 1.1457613514 -0.7051695695 0.0000000000 O 0.5690470822 0.1011885423 0.0000000000 C -0.6396763487 -0.0803383286 0.0000000000 H -1.3877214100 0.7202321970 0.0000000000 Nuclear Repulsion Energy = 31.872637 MP2 energy = -113.906021 S1 geometry of hydroxycarbene optimized at the EOM-EE-CCSD/aug-cc-pVTZ level H 1.0607771681 0.6860769042 -0.2594430355 O 0.5810032732 -0.0718121667 0.1078402207 C -0.7115247527 -0.0008460332 -0.1607811870 201 H -1.5056601577 0.2874068822 0.5354763693 Nuclear Repulsion Energy = 30.303616 S1 CCSD total energy = -114.154259 T1 geometry of hydroxycarbene optimized at the EOM-EE-CCSD/aug-cc-pVTZ level H 1.0378363542 0.6974439577 -0.2444137309 O 0.5882009344 -0.0798388592 0.1044610699 C -0.7226165780 -0.0258055921 -0.1724375227 H -1.4788251797 0.3090260801 0.5354825511 Nuclear Repulsion Energy = 30.092369 T1 CCSD total energy = -114.205276 B.2 Methylhydroxycarbene (MHC) S0 geometry of cis-methylhydroxycarbene optimized at the MP2/aug-cc-pVTZ level O 1.1868344540 0.1237451207 -0.0000000000 H 0.9564841243 1.0802177211 -0.0000000000 C 0.1311852606 -0.6462793952 0.0000000000 C -1.1343508851 0.1462742942 -0.0000000000 H -1.7177732417 -0.1544954477 0.8716029000 H -1.7177732417 -0.1544954477 -0.8716029000 H -0.9966195255 1.2388428153 -0.0000000000 Nuclear Repulsion Energy = 69.672121 MP2 energy = -153.462104 202 S0 geometry of trans-methylhydroxycarbene optimized at the MP2/aug-cc-pVTZ level C 1.1574204338 0.1371534061 -0.0000000000 O -1.0964950192 0.2778337929 -0.0000000000 H 1.7253406425 -0.1978982463 0.8645856348 H 1.0584925919 1.2202974024 -0.0000000000 H 1.7253406425 -0.1978982463 -0.8645856348 C -0.1260902644 -0.6090798883 0.0000000000 H -1.9251947395 -0.2156123598 0.0000000000 Nuclear Repulsion Energy = 69.770894 MP2 energy = -153.468105 D0 geometry of cis-methylhydroxycarbene cation optimized at the MP2/aug-cc-pVTZ level O 1.2442734591 0.0538163374 0.0000000000 H 1.3512202539 1.0373724474 0.0000000000 C 0.0971118289 -0.3955794103 0.0000000000 C -1.2385297899 0.1610514884 0.0000000000 H -1.7577765415 -0.2376069072 0.8769859738 H -1.7577765415 -0.2376069072 -0.8769859738 H -1.2305357242 1.2523626121 0.0000000000 Nuclear Repulsion Energy = 70.040142 MP2 energy = -153.165275 D0 geometry of trans-methylhydroxycarbene cation optimized at the MP2/aug-cc-pVTZ level 203 C 1.2272647677 0.1147545737 -0.0143911962 O -1.1956380277 0.0975723190 -0.0625750253 H 1.7430401758 -0.2435518156 0.8816480176 H 1.1765686242 1.2033889786 -0.0499703210 H 1.7699120156 -0.2982983222 -0.8704671172 C -0.0937864517 -0.4675695343 -0.0184102140 H -1.9761801169 -0.5039153508 -0.0561023428 Nuclear Repulsion Energy = 70.100590 MP2 energy = -153.171879 S1 geometry of methylhydroxycarbene optimized at the EOM-EE-CCSD/aug-cc-pVTZ level C -0.1199435478 -0.3855470494 -0.0716671909 O -1.2575253575 0.3042131902 -0.0410852365 H -1.7084351335 0.1974482805 0.8102310123 C 1.2716697780 0.1429915721 -0.0414360962 H 1.9186349669 -0.4544178957 -0.6827316237 H 1.7055698473 0.1344466800 0.9689033581 H 1.2618230596 1.1750262329 -0.4095116992 Nuclear Repulsion Energy = 67.657603 S1 CCSD total energy = -153.394659 T1 geometry of methylhydroxycarbene optimized at the EOM-EE-CCSD/aug-cc-pVTZ level 204 C -0.0931884049 -0.5103692318 0.0262351846 O -1.2564854601 0.1583647597 -0.1170889652 H -1.6597206035 0.3153306331 0.7438577867 C 1.2519732928 0.1267875746 0.0009698765 H 2.0218102457 -0.6393114675 0.0781668010 H 1.3809680152 0.8281948100 0.8332018243 H 1.4029682448 0.6756365454 -0.9342127736 Nuclear Repulsion Energy = 67.590893 T1 CCSD total energy = -153.446982 205 C MPIMS gas connections C.1 Bubbler dimensions for MPIMS use Figure C.1 Schematic of two glass bubbler setups for standard 10–15 ml sample volume (left) and for 2–3 ml small sample volume (right). The conical bottom of the small sample bubbler is more efficient for gas dispersion when samples are sparse, for example, in the case of deuterated PA. The bubblers shown in Figure C.1 are a general schematic which was designed to adhere to the restrictions for chemical usage at the ALS beamline. First and foremost, the safe operation zone for these bubblers is below 1 atm. The ALS permits no open use of toxic or corrosive chemicals at the beamline. Therefore, PA – a corrosive, combustible and flammable liquid should not be allowed to leak out at the beamline under any circumstances. 206 An operating pressure below 760 Torr in the bulb ensures that leaks, if any, would draw gasses into the bubbler and minimize sample exposure to the beamline atmosphere. Sample preparation was done at the ALS chemistry laboratory under a fume hood ensuring proper waste disposal guidelines. Sample line connections on the upper half of these bubblers have multiple high-vacuum Teflon stopcocks to ensure that both the Helium inlet line and the sample bulb can be independently evacuated. Liquid samples should be filled to only 30–40% volume capacity of the lower half to avoid frothing or spillage at fast gas flow rates. The ¼’’ inlet and outlet connections should be at least 1 inch in length to accommodate Ultra-Torr fittings. At the same time, the total lateral dimension of the bubbler setup has to be minimized for easy connectivity with the Sandia yoke setup. Additional modifications can be made to the bottom part for sample loading/unloading or cleaning purposes but changes to the top part should be made only after verifying compatibility with the MPIMS setup during the beamtime run. 207 C.2 MPIMS manifold and flow setup Figure C.2 Complete flow setup and connections to the reactor tube for a single bubbler, with the current 5-inlet Sandia yoke setup. All the inlets have independently controlled mass flow controllers. The flow schematics shown in Figures C.2 and C.3 have been designed with appropriate number of valves before and after every connection to aid in section-wise leak checks. This is necessitated by the earlier mentioned condition of creating leak-free gas flow connections. The 5-inlet glass yoke supplied by Sandia National Labs (SNL) is used to complete the sample connection to the reactor tube. Inlet 1 is reserved for ultra-high-pure He. Inlet 2 is connected to SNL’s reactive gases cabinet, and is usually never disconnected once they are made at the beginning of each six-month period. Lines 3,4 and 5 were blank 208 and therefore available for our use. Among these, inlet 3 is used for a calibration gas mixture (usually mass calibration). Inlet 4 was used for gas mixtures made in cylinders. Inlet 5 has Teflon connections and was used for the PA sample flows. Figure C.3 Complete flow setup and connections to the reactor tube for a double bubbler setup, with independently controlled flow-restricting needle valves. In both setups the water bath temperatures were maintained 1–2 ˚C below room temperature to avoid sample condensation inside the lines. A double bubbler setup was employed for deuteration experiments described in chapter 5, where D2O and PA vapors from separate bubblers were combined to make in-situ d1-PA. Helium flow rate and the total pressure in each bubbler was controlled using individual MFCs and needle valves. 209 D KER distributions from co-fragment internal energy states D.1 Benzene-water MATLAB code (detecting water) 1 clear all; 2 close all; 3 IR=3623; 4 Efrag=1600; 5 D0=900; 6 avl=1200; %ideally IR-D0-int_energy but can use an experimental number 7 B=0.1897651285; 8 A=B; 9 C=B/2; 10 % benzene vibrations 11 vib=[0 448 539 595 673 779 1003 1037 1106 1192 1212 1242 1388 1482 1522 1622 1667 1673 1716 1755 1811 1888 1917 1958 1989 2009 2005 2144 2214 2328 2326 2386]; 12 %%------calculate rotations------%% 13 J=30; 14 vib=vib(vib<avl); % only count the accessible vibrations 15 dist=zeros(((J+1)*(J+2)/2),4); 16 c=1; 17 T=200; 18 kT=207/298*T; 19 for j=0:J; 20 for k=j:-1:0; 21 Ej=B*j*(j+1)+(C-B)*k*k; 22 %dist(c,:)=[j k Ej (2*j+1)*((k>0)+1)]; %%can add degeneracy 23 dist(c,:)=[j k Ej (2*j+1)*((k>0)+1)*exp(-Ej/kT)]; %%rotational envelop function 24 c=c+1; 25 end 26 end 27 dist=sortrows(dist,3); 28 Erot=(dist(:,3)); 210 29 Int1=dist(:,4); 30 %plot(Erot,Int,'.') 31 vec=zeros((length(dist)*length(vib)),2); 32 r=length(dist); 33 %%-------add up vibrations----%% 34 c=1; 35 for v=1:length(vib) 36 vec(c:(c+r-1),1)=avl-vib(v)-Erot; 37 vec(c:(c+r-1),2)=Int1; 38 c=c+r; 39 end 40 sc=7.02728*42.4/8; % experimentally scaling factor for pixel to velocity 41 Evec=vec(:,1); 42 Int=vec(:,2); 43 Int(Evec<0)=0; 44 Evec(Evec<0)=0; 45 Vvec=sc*(Evec*78/96).^.5; %scaled to velocity from COM energy 46 hold on 47 %plot(Vvec,Int,'.r') 48 width=40/2.355; 49 xdist=0:0.1:max(Vvec)+2.355*width; 50 ydist=zeros(1,length(xdist)); 51 for l=1:length(vec) 52 ydist=ydist+Int(l)*normpdf(xdist,Vvec(l),width); 53 end 54 plot(xdist,ydist); % velocity distribution 55 hold off 56 figure() 57 hold on 58 stem(avl-vib,100*ones(1,length(vib)),'r','MarkerSize',0) 59 plot((xdist/sc).^2*96/78,ydist); % energy dist with vibration markers 60 xlim([0 1400]) 61 hold off 211 E Absolute photoionization cross-section calculations We obtained photoionization spectra of PA, CO and CO2 as a function of energy and normalized them to the VUV photon flux. We then scaled them to their respective absolute cross-sections using equation 4.2 in Chapter 4. For PA, we flow 10 sccm of the PA/He mixture (1 Torr PA vapor in 56 Torr He) and 4 sccm of reference gas (with 1% each of ethene, propene and 1-butene) balanced with He to a total flow of 250 sccm into the reactor tube maintained at 4 Torr. Photoionization spectra of the PA and the reference mixture are obtained in the 9.6 to 10.6 eV range. The PA photoionization spectrum is scaled using the known ionization cross-section of propene at 10.2 eV. The absolute PI cross-section for PA is shown in Table E.1. For CO, we prepare a mixture of 5.95 Torr CO and 1.97 Torr Xe balanced to 1000 Torr with He. We flow 5 sccm of the CO/Xe mixture and 495 sccm of pure He into the reactor tube maintained at 2 Torr and obtain a photoionization scan between 13.9–14.4 eV. The photoionization spectrum of CO is scaled using the known ionization cross-section of Xe at 14.15 eV, and the absolute cross-section values are shown in Table E.2 and plotted in Figure E.1(a). For CO2, we prepare a cylinder with 5.98 Torr CO2 and 1.94 Torr Xe balanced to 1000 Torr with He. We flow 1 sccm of the CO2/Xe mixture and 249 sccm of pure He into the reactor tube maintained at 4 Torr and obtain a photoionization scan at 13.7–14.25 eV. The photoionization spectrum of CO2 is also scaled using the known ionization cross-section of 212 Xe at 14.15 eV, and its absolute cross-section values are shown in Table E.3 and plotted in Figure E.1(b). Figure E.1 Absolute photoionization cross-sections of (a) CO and (b) CO2 from threshold to around 14 eV obtained at 0.005 eV steps. E.1 Absolute photoionization cross-section tables Table E.1 PA PI cross-section values, 𝜎𝜎 𝑃𝑃 𝑃𝑃 𝑎𝑎 𝑎𝑎 𝑎𝑎 , in Mb (1 Mb = 10 -18 cm 2 molecule -1 ). The uncertainty in the reported values is ~25%. Photon energy (eV) 𝜎𝜎 𝑃𝑃 𝑃𝑃 𝑎𝑎 𝑎𝑎 𝑎𝑎 Photon energy (eV) 𝜎𝜎 𝑃𝑃 𝑃𝑃 𝑎𝑎 𝑎𝑎 𝑎𝑎 (cont.) Photon energy (eV) 𝜎𝜎 𝑃𝑃 𝑃𝑃 𝑎𝑎 𝑎𝑎 𝑎𝑎 (cont.) Photon energy (eV) 𝜎𝜎 𝑃𝑃 𝑃𝑃 𝑎𝑎 𝑎𝑎 𝑎𝑎 (cont.) 9.525 0.000 9.968 0.023 10.293 2.090 10.643 3.160 9.643 0.000 9.993 0.027 10.318 2.410 10.775 3.128 9.668 0.000 10.018 0.052 10.343 2.760 11.025 3.399 9.693 0.000 10.025 0.075 10.368 2.900 11.275 3.724 9.718 0.000 10.043 0.070 10.393 3.030 11.525 3.735 9.743 0.000 10.068 0.126 10.418 3.040 11.775 3.500 9.768 0.000 10.093 0.205 10.443 3.100 12.025 3.178 9.775 0.000 10.118 0.277 10.468 3.150 12.275 3.159 213 9.793 0.000 10.143 0.427 10.493 3.230 12.525 3.301 9.818 0.001 10.168 0.584 10.518 3.140 12.775 3.401 9.843 0.001 10.193 0.843 10.525 3.167 13.025 3.374 9.868 0.002 10.218 1.090 10.543 3.200 13.275 3.422 9.893 0.003 10.243 1.410 10.568 3.180 13.525 3.536 9.918 0.006 10.268 1.750 10.593 3.180 13.775 3.822 9.943 0.013 10.275 2.058 10.618 3.180 14.025 4.210 Table E.2 CO PI cross-section values, 𝜎𝜎 𝐶𝐶𝐶𝐶 𝑎𝑎 𝑎𝑎 𝑎𝑎 , in Mb. The uncertainty in the reported values is ~20%. Photon energy (eV) 𝜎𝜎 𝐶𝐶𝐶𝐶 𝑎𝑎 𝑎𝑎 𝑎𝑎 Photon energy (eV) 𝜎𝜎 𝐶𝐶𝐶𝐶 𝑎𝑎 𝑎𝑎 𝑎𝑎 (cont.) Photon energy (eV) 𝜎𝜎 𝐶𝐶𝐶𝐶 𝑎𝑎 𝑎𝑎 𝑎𝑎 (cont.) Photon energy (eV) 𝜎𝜎 𝐶𝐶𝐶𝐶 𝑎𝑎 𝑎𝑎 𝑎𝑎 (cont.) 13.935 0.852 14.065 49.562 14.195 29.042 14.325 32.006 13.940 0.742 14.070 50.912 14.200 33.304 14.330 31.256 13.945 0.888 14.075 47.632 14.205 38.434 14.335 30.174 13.950 0.898 14.080 43.809 14.210 45.142 14.340 32.636 13.955 0.918 14.085 39.149 14.215 51.992 14.345 30.920 13.960 1.164 14.090 34.970 14.220 52.747 14.350 30.368 13.965 1.315 14.095 31.049 14.225 49.011 14.355 27.641 13.970 1.394 14.100 29.084 14.230 43.158 14.360 26.571 13.975 1.942 14.105 28.146 14.235 43.800 14.365 26.855 13.980 3.031 14.110 30.899 14.240 42.489 14.370 29.001 13.985 5.164 14.115 31.037 14.245 43.060 14.375 32.979 13.990 8.100 14.120 27.975 14.250 36.620 14.380 33.422 214 13.995 13.126 14.125 24.913 14.255 29.989 14.385 36.397 14.000 18.402 14.130 21.228 14.260 26.499 14.390 34.342 14.005 25.218 14.135 21.142 14.265 26.955 14.395 30.663 14.010 28.190 14.140 21.337 14.270 30.055 14.400 27.728 14.015 29.212 14.145 22.949 14.275 32.336 14.405 26.386 14.020 26.090 14.150 23.147 14.280 35.339 14.410 26.234 14.025 25.127 14.155 21.995 14.285 37.636 14.415 27.391 14.030 25.391 14.160 22.561 14.290 37.083 14.420 26.659 14.035 30.439 14.165 21.828 14.295 37.469 14.425 24.062 14.040 34.829 14.170 22.031 14.300 35.915 14.430 24.437 14.045 39.514 14.175 21.889 14.305 36.297 14.435 24.673 14.050 44.012 14.180 23.109 14.310 34.593 14.055 47.514 14.185 24.260 14.315 31.509 14.060 49.445 14.190 26.618 14.320 31.748 Table E.3 CO2 PI cross-section values, 𝜎𝜎 𝐶𝐶𝐶𝐶 2 𝑎𝑎 𝑎𝑎 𝑎𝑎 , in Mb. The uncertainty in the reported values is ~20%. Photon energy (eV) 𝜎𝜎 𝐶𝐶𝐶𝐶 2 𝑎𝑎 𝑎𝑎 𝑎𝑎 Photon energy (eV) 𝜎𝜎 𝐶𝐶𝐶𝐶 2 𝑎𝑎 𝑎𝑎 𝑎𝑎 (cont.) Photon energy (eV) 𝜎𝜎 𝐶𝐶𝐶𝐶 2 𝑎𝑎 𝑎𝑎 𝑎𝑎 (cont.) Photon energy (eV) 𝜎𝜎 𝐶𝐶𝐶𝐶 2 𝑎𝑎 𝑎𝑎 𝑎𝑎 (cont.) 13.667 0.311 13.817 26.342 13.967 28.778 14.117 22.467 13.672 0.310 13.822 26.586 13.972 28.100 14.122 23.456 13.677 0.368 13.827 25.702 13.977 27.552 14.127 21.823 13.682 0.486 13.832 25.954 13.982 27.561 14.132 21.794 13.687 0.570 13.837 26.466 13.987 27.827 14.137 21.982 215 13.692 0.561 13.842 28.992 13.992 26.441 14.142 21.168 13.697 0.774 13.847 30.585 13.997 27.157 14.147 21.555 13.702 0.860 13.852 28.888 14.002 27.531 14.152 21.312 13.707 0.940 13.857 29.830 14.007 26.960 14.157 20.898 13.712 1.032 13.862 27.924 14.012 27.080 14.162 21.135 13.717 1.065 13.867 30.156 14.017 27.888 14.167 20.787 13.722 1.189 13.872 30.385 14.022 26.864 14.172 20.325 13.727 1.225 13.877 28.672 14.027 26.793 14.177 19.916 13.732 1.440 13.882 27.112 14.032 27.188 14.182 19.627 13.737 1.334 13.887 26.380 14.037 26.500 14.187 19.563 13.742 1.607 13.892 28.587 14.042 26.267 14.192 19.290 13.747 1.972 13.897 26.995 14.047 25.864 14.197 19.014 13.752 2.619 13.902 26.779 14.052 24.981 14.202 18.932 13.757 3.198 13.907 27.531 14.057 25.132 14.207 18.922 13.762 4.601 13.912 27.696 14.062 25.143 14.212 18.388 13.767 7.583 13.917 27.735 14.067 23.663 14.217 17.733 13.772 12.409 13.922 27.591 14.072 24.009 14.222 18.232 13.777 15.093 13.927 27.505 14.077 23.596 14.227 18.672 13.782 18.099 13.932 27.826 14.082 23.524 14.232 17.836 13.787 21.221 13.937 28.570 14.087 24.052 14.237 17.935 13.792 24.332 13.942 28.998 14.092 23.677 14.242 17.738 13.797 26.094 13.947 29.285 14.097 23.771 14.247 17.319 13.802 24.924 13.952 29.324 14.102 22.332 14.252 17.429 216 13.807 25.824 13.957 28.305 14.107 23.290 14.257 16.707 13.812 25.172 13.962 29.289 14.112 22.562 F Eckart simulations F.1 Function file for defining Eckart barrier 1 function [V,G]=eckart_unsym(Vb,delH,mass,nubarrier,x,Emax) 2 %----define all unit conversions----% 3 bohr = 5.3e-11; %distance 4 amu = 1.66e-27; %mass 5 hbar = 1.0546e-34; 6 q = 1.6e-19; 7 mol = 6.023e23; 8 Vcon = 6.9477e-21; %energy in kcal/mol to J 9 m=mass*amu; 10 %-----------------------------------% 11 x = x*bohr; 12 V1 = Vb*Vcon; 13 V2 = delH*Vcon; 14 B = V2; 15 A = 2*V1 - B + 2*(-V1*(B - V1))^(1/2); 16 nu = nubarrier*3e10; %cm-1 to s-1 17 b = (A-B)*(A+B)/(2*pi*nu)/(2*A)^1.5/sqrt(m); 18 y = exp(x/b); 19 V = A*y./(1+y).^2 + B*y./(1+y); 20 dE = 0.1; %kcal/mol 21 22 G=zeros(1,Emax/dE+1); 23 for i=1:length(G) 24 e = (i - 1)*dE*Vcon; 25 delta = ((e-B)/e)^.5; 217 26 k = (2*m*e)^.5/hbar; 27 G(i) = ((sinh(pi*k*b*(1+delta)))^2 - (sinh(pi*k*b*(1- delta)))^2)/((sinh(pi*k*b*(1+delta)))^2+(cosh(pi/2*(8*m*b^2*A/hbar^2- 1)^0.5))^2); %transmission 28 end 29 end F.2 Main code for plots and integration 1 clear all 2 close all 3 x = [-5:0.01:5]; %in units of bohr 4 mass = 1; %amu 5 Emax = (81 - 40.5); %kcal/mol 6 dE = 0.1; %default in the function so change in both places 7 v = [3445,2917,2836,1452,1382,1294,1182,1011,814,460,2885,1451,951,697,73]; 8 dos = bs(v,Emax*350,dE*350); %Can use if degeneracy required 9 [Vac,Gac]=eckart_unsym(29.99,-50.7,mass,2031,x,Emax); 10 [Vvy,Gvy]=eckart_unsym(23.94,-39.8,mass,1443,x,Emax); %all energies in kcal/mol, freq in cm-1(absolute) 11 Vcon = 6.9477e-21; 12 T = 173; %Enter T in K 13 kT = T/298*4.11e-21; %at 298K in J 14 enaxis = [0:dE:Emax]; 15 l = length(enaxis); 16 Er = enaxis*Vcon; 17 transvy = exp(-Er/kT).*Gvy; 18 transac = exp(-Er/kT).*Gac; 19 sprintf('ac:vy = %d',sum(transac(2:l))/sum(transvy(2:l))) 20 % subplot(1,3,3); 21 figure() 22 hold on 23 plot(enaxis,transac,'r') 24 plot(enaxis,transvy) 25 hold off 218 26 27 % subplot(1,3,1); 28 figure() 29 hold on 30 plot(x,Vac/Vcon,'r') 31 plot(x,Vvy/Vcon) 32 ylim([-50 40]) 33 xlim([-2.5 2.5]) 34 hold off

Abstract (if available)

Abstract The UV photochemistry of organic molecules is a fundamental process that governs reactions in the atmosphere, synthetic chemistry, processing of organic aerosols, and biological damage in living tissues. In many organic molecules, the ensuing dynamics involve pathways that are in competition with each other. Photodissociation is usually fast (picoseconds to microseconds), and these non-equilibrium processes are controlled by kinetic competition and dynamical forces. Pyruvic acid (PA), a proxy for volatile organic compounds in the atmosphere, is one of the few carboxylic acids that absorbs solar radiation and reacts slowly with OH radicals in the atmosphere. It is believed to form an extremely short-lived methylhydroxycarbene (MHC) radical under UV irradiation. ❧ Excitation to the S₁ state via a single photon of 351 nm confirms that decarboxylation is favored at low excitation energies. Previous experiments have shown that following excitation to the S₁ state, acetaldehyde and CO₂ are produced. Using the MPIMS, we show that the hypothesized MHC intermediate is a nascent product of the photodissociation process at 351 nm. We obtain its kinetic time trace and attribute its rapid disappearance to isomerization and secondary reactions. MHC is produced with significant excitation energy (>14,000 cm⁻¹) and is observed to form both the kinetically favored vinyl alcohol isomer and the thermodynamically favored acetaldehyde isomer. MHC is believed to live sufficiently long to undergo secondary reactions with other molecules. ❧ Upon excitation using a single 193 nm photon, CO₂, CO, H, OH, HCO, CH₂CO, CH₃CO and CH₃ are determined to be major photodissociation products of PA at 193 nm. Experiments using both VMI and MPIMS reveal that three-body fragmentation processes are dominant. Some dissociation pathways are believed to proceed via internal conversion to lower excited states on which decarboxylation is favored. Acetaldehyde and vinyl alcohol are only minor co-products from the decarboxylation process at this wavelength, but products that are known to arise from their unimolecular dissociation, such as HCO, H₂CO and CH₄, are identified and quantified. A multivariate analysis offers the first comprehensive description of the dissociation pathways of PA initiated on the S₃ excited state. Most of the observed products and yields are rationalized on the basis of three reaction mechanisms: (i) decarboxylation terminating in CO₂ + other primary products (∼50%)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Photodissociation dynamics of atmospherically relevant small molecules in molecular beams

PDF

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

PDF

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

PDF

Photoelectron and ion imaging investigations of spectroscopy, photoionization, and photodissociation dynamics of diazomethane and diazirine

PDF

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

PDF

Development and implementation of methods for sliced velocity map imaging. Studies of overtone-induced dissociation and isomerization dynamics of hydroxymethyl radical (CH₂OH and CD₂OH)

PDF

Photoexcitation and nonradiative relaxation in molecular systems: methodology, optical properties, and dynamics

PDF

Quantum molecular dynamics and machine learning for non-equilibrium processes in quantum materials

PDF

The role of the environment around the chromophore in controlling the photophysics of fluorescent proteins

PDF

Surface plasmon resonant enhancement of photocatalysis

$Study of rotation and phase separation in ³He, ⁴He, and mixed ³He/⁴He droplets by X-ray diffraction$

PDF

Study of rotation and phase separation in ³He, ⁴He, and mixed ³He/⁴He droplets by X-ray diffraction

PDF

Harnessing fluorinated C1 nucleophilic reagents for the direct fluoroalkylation of ubiquitous C(sp2)-X and C(sp)-H centers

PDF

Electrochemical pathways for sustainable energy storage and energy conversion

PDF

Reforming of green-house gases: a step towards the sustainable methanol economy; and, One-step deoxygenative fluorination and trifluoromethylthiolation of carboxylic acids

PDF

From single molecules to bacterial nanowires: functional and dynamic imaging of the extracellular electron transfer network in Shewanella oneidensis MR-1

Asset Metadata

Creator Samanta, Bibek Ranjan (author)

Core Title Unraveling photodissociation pathways in pyruvic acid and the role of methylhydroxycarbene

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Publication Date 01/16/2021

Defense Date 06/22/2020

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

Tag gas-phase dynamics,hydroxycarbene,molecular dynamics,OAI-PMH Harvest,photochemistry,photodissociation,pyruvic acid

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Reisler, Hanna (committee chair), Dawlaty, Jahan (committee member), El-Naggar, Moh (committee member)

Creator Email bibeksam@gmail.com,bsamanta@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-333340

Unique identifier UC11664358

Identifier etd-SamantaBib-8693.pdf (filename),usctheses-c89-333340 (legacy record id)

Legacy Identifier etd-SamantaBib-8693.pdf

Dmrecord 333340

Document Type Dissertation

Rights Samanta, Bibek Ranjan

Type texts

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

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

Repository Name University of Southern California Digital Library

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