University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Interfacial Polarization and Photobasicity: Spectroscopic Studies of the Electrode Environment and Photochemical Effects by Matthew J. V oegtle A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (CHEMISTRY) December 2022 Copyright 2022 Matthew J. V oegtle Contents Abstract ix Chapter 1: Introduction 1 1.1 Interfacial Structure at Electrode/Liquid Interfaces 1 1.2 Photobasicity - Controlling Acid/Base Chemistry with Light 4 Chapter 2: Electrodes as Polarizing Functional Groups: Correlation between Ham- mett Parameters and Electrochemical Polarization 7 2.1 Introduction 7 2.2 The Hammett Parameters, Interfacial Fields, and their Vibrational Signature 8 2.3 Experimental and Computational Methods 11 2.4 Results 15 2.5 Discussion 19 Chapter 3: Electric Fields and Structure at the Ionic Liquid/Metal Interface 26 3.1 Approach 28 3.2 Experimental Methods 30 3.3 Computational Methods 33 3.4 Results 36 3.5 Discussion 43 3.6 Concluding remarks 47 Chapter 4: Can Bronsted Photobases act as Lewis Photobases? Quinolines Enhance Binding Affinity to BF 3 in the Excited State 48 4.1 Introduction 48 4.2 Concepts 50 4.3 Experimental and Computational Methods 51 4.4 Results and Discussion 54 4.5 Summary and Outlook 61 Chapter 5: Conclusions and Further Work 63 References 66 Appendices 86 A Supporting Information for: Electrochemical Control of Molecular Polarization 87 A.1 Substituents, Hammett parameters and Nitrile frequencies 88 A.2 Linewidths with changingσ ρ andω CN 89 A.3 V oltage Dependent SFG data 90 ii B Supporting Information for: Electric Fields and Structure at the Ionic Liquid/Metal Interface 99 B.1 Raw SFG Spectra of the 4-MBN Monolayer in the Presence of Different ILs 99 B.2 Raw SERS Spectra of the 4-MBN Monolayer in the Presence of Different ILs 100 B.3 Linewidths of Extracted Lorenztians from SFG Spectra 101 B.4 Temperature Dependent SFG Data 102 B.5 FTIR Spectra of [EMIM] + [BF 4 ] − Before and After Microwave Purification103 B.6 Summary of Key Data 104 B.7 Selection of Radial Cut-off (R cut ) for Electric Field Calculations 109 C Supporting Information for: Can Bronsted Photobases Act as Lewis Photobases? 118 C.1 Brønsted Photobasicity in Quinolines 118 C.2 Method Dependence of Computational F¨ orster Cycle 120 C.3 Electronic Spectra of 5-R-Quinolines and Their BF 3 Adducts 122 C.4 Additional View of 5-MeOQBF 3 EDD Map 127 C.5 Summary of Computational Data 128 D Further Details on the Computational Forster Cycle Applied to Lewis Adducts 133 D.1 Background 133 D.2 Comments on EDA 133 D.3 Some tips for TD-DFT Geometry Calculations 136 E Rotations, Quaternions, and You 139 List of Tables 5.1 Substituents used, respective Hammett parameters and Nitrile frequencies 88 5.2 Fit Params. with equation: I SFG (ω)= A NR e iφ + B CN ω− ω CN +iΓ CN 2 exp − (ω− ω g) 2 σ 2 g 91 5.3 Summary of anion series from SFG data reported in the main text. We have in- cluded the estimated charge density, average values ∆ν CN , average values Γ CN , along with error bars for measurements and average R 2 values for the fits. 104 5.4 Summary of anion series from SERS data reported in the main text. We have included the estimated charge density, measured values of ∆ν CN and Γ CN , along with error bars for measurements and R 2 values for the fits. 105 5.5 System setups for functionalized gold surface - ionic liquid assembly a 115 5.6 Densities and diffusion coefficients of pure ionic liquids a,b 115 5.7 Experimental and previously calculated densities and diffusion coefficients of ionic liquids 117 5.8 EDA results and calculated absorption energy for the 5-aminoquinolineBF 3 adduct with BF 3 bound to the amino group. Geometries and energy calculations were per- formed usingωB97X-D/def2-svpd. We observe a blue shift in the spectra relative to free 5-aminoquinoline, indicating that the amino-bound BF 3 acts as an electron withdrawing group 123 5.9 Experimental estimates for 0-0 energies between the ground and photobasic elec- tronic states. 126 iii 5.10 Direct output of EDA2 calculations of 5-R-Quinoline-BF 3 Adducts.Energies of the same geometries calculated using ωB97X-D/def2-tzvppd, along with a PCM, are reported below the break in the table. 128 5.11 Energies (ωB97X-D/def2-svpd) corresponding to the optimized S0 and photobasic (PB) excited states for free and adducted quinolines. The differences between the S0 and PB energies for the free and adducted states are the 0-0 energies used in the F¨ orster cycle analysis (i.e., the enhancement in BF 3 binding energy in the excited state), which are reported in Table Table 5.12.Energies of the same geometries calculated using ωB97X-D/def2-tzvppd, along with a PCM, are reported below the break in the table. 129 5.12 Summary of computed absorption and emission lines for the S0-PB state transi- tion. Geometries and energies were calculated using ωB97X-D/def2-svpd, and TD-DFT calculations were performed using TDA and triplet states were excluded. Energies of the same geometries calculated using ωB97X-D/def2-tzvppd, along with a PCM, are reported below the break in the table. 130 5.13 This table displays the energies corresponding to the rehybridization/strain of BF 3 and quinolines in the adducted state. All energies reported in this table are from calculations performed at the ωB97X-D/def2-svpd level of theory. As discussed in the main text, single point calculations were performed for isolated adduct frag- ments (without relaxation) to get a more accurate accounting of the B-N bond en- ergy. The geometry change in BF 3 from planar to pyramidal requires a great deal of energy, and default EDA procedures (which perform no geometry optimization) do not account for this. Mao and Head-Gordon have introduced a method called adiabatic EDA to better treat these cases 1 . Energies of the same geometries cal- culated using ωB97X-D/def2-tzvppd, along with a PCM, are reported below the break in the table. 131 5.14 Results and associated energy values to determine the TEA-BF 3 binding energy. In an identical procedure for the quinoline data, the fragments (TEA and BF 3 ) of the adduct were optimized independently and the relaxation energy was subtracted from the raw EDA value. 132 List of Figures 2.1 Picture depicting the effect of polarizing electric field at electrochemical interface and an electron withdrawing functional group on the nitrile stretch. 11 2.2 (a) Diagram showing a cross section of the cell used in experiments. (b) Cartoon diagram of SAM-modified Au sample in [EMIM] + [BF4] − . (c) Representative spectrum (blue) and its fit (red) at V = 0.0 V . The dip around 2230 cm − 1 is due to the nitrile stretch. 13 2.3 (a) Representative FTIR absorption spectra of the compounds in the nitrile stretch region. (b) Experimentally determined central frequencies of the nitrile stretch in a series of para-substituted benzonitriles versus the Hammett parameters of the sub- stituents. (c) Computed central frequencies of the nitrile stretch for the Hammett series. 16 iv 2.4 (a) The nitrile stretch spectra isolated from vSFG spectra. The data shows variation of the spectra as a function of applied potential. (b) Experimentally obtained nitrile stretch frequency as a function of applied potential. (c) Computed nitrile stretch frequency with varying applied field. 17 2.5 Experimentally obtained frequencies as a function of (a) applied field (b) Hammett parametersσ p ranging from -0.37 to +0.45. (c) Relation between the applied field and the Hammett parameterσ p . 20 2.6 Computed nitrile frequencies as a function of (a) applied field (b) Hammett param- eters σ p . Red circles are the data points and black lines are the fits. (c) Relation between electric field and σ p . 21 2.7 Computed ρ N as a function of (a) applied field (b) Hammett parameters σ p . Red circles are the data points and black lines are the fits. (c) Relation between electric field and σ p . 23 2.8 Relation between σ p and electric field based on experimental frequency changes (blue), computational frequency changes (green), and charge density changes on nitrogenρ N (red) 24 3.1 Chemical structure and name abbreviations of the anions studied in this work and their corresponding molecular charge density. The volume term in the charge den- sities was estimated using DFT-optimized structures for each anion. 27 3.2 Overview of experimental work. Panel Figure 3.2a: A diagram of the cell used to acquire SFG spectra. Panel Figure 3.2b: A cartoon depicting SFG generation from the 4-MBN SAM at the gold-IL interface. Panel Figure 3.2c: Representative SFG spectra of a 4-MBN SAM showing a broad non-resonant background the narrow CN stretch. 30 3.3 Room temperature frequency shifts of 4-MBN monolayer on the electrode surface in the presence of different ILs. Figure Figure 3.3a shows extracted Lorenzians from raw SFG spectra, and Figure Figure 3.3a shows the center nitrile frequency plotted against the charge density of the anion. The monolayer is strongly solvated in the presence of smaller anions, with a large field of ∼ 3.6 V/nm observed using [EMIM] + [Br] − . 38 3.4 Surface enhanced Raman (SERS) data of the nitrile center frequency in the pres- ence of different ILs. Chemically etched silver was used as a substrate. We ob- serve an increase in polarization at the interface correlated with smaller anion size, in agreement with the SFG results. Point to point differences between SERS and SFG measurements may be related to heterogeneity at the SERS surface. 39 3.5 Calculated frequency shifts at nitrile carbon of 4-MBN monolayer on the gold slab in presence of different ILs. (a) Realistic anions with [EMIM] + cations, and (b) Modified Cl – anions with[EMIM] + cations. Lennard-Jonesσ for Cl − = 4.04 ˚ A. 40 3.6 Snapshots of representative systems after 200 ns of production MD. Only top half of the functionalized gold slab has been shown. For realistic anions, full IL residues have been shown for atoms within 4 ˚ A of nitrile nitrogens. For[EMIM] + [Cl 2σ ] − and[EMIM] + [Cl 3σ ] − , the selection thresholds are 6 ˚ A and 8 ˚ A respectively. 41 3.7 Symmetrized partial number density of representative atoms plotted against aver- age relative position from the center of gold layer. As a reference, the positions of nitrile nitrogen (N CN ) atoms of 4-MBN monolayer are shown in green. [EMIM] + hydrogens from imidazole ring are labelled as H3 (H atom at 3 rd position of imi- dazole ring), H4 and H5. All four F atoms of[BF 4 ] − anions are labelled F. The F atoms directly attached to phosphorus center in[FAP] − are labelled F P . The atom names for larger ions are illustrated in Scheme Scheme 3.1. 46 4.1 A diagram of the F¨ orster cycle applied to the quinoline-BF 3 adduct. The excited state drive∆G ES is inferred from spectroscopic estimates of the S 0 -S 1 gaps for the adducted and unadducted forms, and the ground state energy for adduct formation. 51 v 4.2 (a) Absorption and emission spectra for free and BF 3 -adducted 5-methoxyquinoline. The large redshift for the absorption and emission between the adducted and unad- ducted forms indicates a substantial enhancement for 5-MeOQ to bind BF 3 in the excited state. (b) Calculated absorption and emission spectra (S 0 - S 1 states) for free and BF 3 -adducted 5-methoxyquinoline show similar red shifts. 55 4.3 (a) A schematic of the 5-MeOQBF 3 adduct and the flow of electrons to the B-N bond in the S 1 state. (b) An electron density difference plot corresponding to the S 0 -S 1 excitation for the 5-MeOQBF 3 adduct. The plot shows that the excitation builds up electronic charge density on nitrogen, which strengthens its bond with BF 3 . 57 4.4 The excited state affinity change for adduct formation expressed both in terms of energy and equilibrium constant change, plotted against the electron withdrawing strength (Hammett Parameter) of the substituent at the 5 position. The data show that the measured affinity change strongly depends on the substituent. 58 4.5 The binding affinity in the ground and excited state using a computational F ¨ orster cycle showing similar trend with respect to the Hammett parameter as our experi- mental data. 60 5.1 (a) Plot of the widths of Lorentzians (obtained from fits) as a function of applied voltage. (b) Full width half maxima vs Hammett parameter 89 5.2 (a) Dipole moment of the molecule as a function of Applied Field.(b) Dipole mo- ment of the molecule as a function of Hammett parameter. 89 5.3 Representative SFG spectra at different voltages. 90 5.4 (a) V oltage (with respect to Ag/AgCl) vs Time. (b) Current vs Time. 91 5.5 SFG spectra of 4-MBN SAMs representative of the data reported in the main text (Figure Figure 3.3). Data from SFG experiments are in general complex and must be fitted to correctly extract spectral information. The black curve in the plot is of the 4-MBN monolayer on gold in the presence of air, and has been scaled for better visual reference. 99 5.6 Unprocessed SERS spectra of 4-MBN SAM for the data reported in the main text (Figure Figure 3.4). The dotted line emphasizes the nitrile stretch of the SAM with no ionic liquid added over the SERS substrate (shown in the black curve). Data reported in the main text are frequency shifts referenced for each SERS substrate to correct for variation over time and between substrates. 100 5.7 Averaged nitrile linewidth from SFG experiments plotted vs varying anion identity. We do not observe any clear dependence of the nitrile linewidth on the identity of the IL anion. 101 5.8 Plot of the change in 4-MBN CN center frequency in the presence of [EMIM] + [Br 4 ] − units of wavenumbers (cm − 1 ), measured as a function of temperature. This data is relevant because the MD simulations reported in this work were done at 400K to prevent freezing. The data show that the CN center frequency is invariant with respect to temperature in this ionic liquid - indicating that the higher simula- tion temperatures are no cause for concern. 102 5.9 These spectra (overlaid to emphasize their similarity across treatments) show that microwave purification does not result in damage to the [EMIM] + [BF 4 ] − IL. IR spectra were taken in an unmodified FTIR liquid cell (using a 25 µm spacer), similar to the one used for SFG. These spectra also show the reasonably low level of water in our ionic liquids after storage in dry air. The gold spectra is of[EMIM] + [BF 4 ] − after exposure to atmosphere for 10 minutes. The bands growing in after this time are likely from water, though they are shifted from the pure water values. 103 5.10 Calculated frequency shifts using electric field at nitrile carbon of 4-MBN mono- layer on the gold slab in presence of different ILs. 106 5.11 (a) Symmetrized partial number density of representative atoms vs Z (b) Sym- metrized charge density of representative atoms vs Z, for all realistic anions 107 vi 5.12 (a) Symmetrized partial number density of representative atoms vs Z (b) Sym- metrized charge density of representative atoms vs Z, for all model Cl – anions 108 5.13 Convergence of electric field on a snapshot at t = 200 ns for representative sys- tems. For each system, top panel is the averaged field components over 200 nitrile nitrogens. Bottom panel shows the standard deviation of electric field over the 200 nitrogens, as a function of R cut used for field calculations. Elelctric field calcula- tions have been performed every 0.05 nm from R cut value of 0.5 nm to 4 nm. 110 5.14 Gradient of spatial average a and gradient of standard deviation b for calculated elec- tric field shown in Figure Figure 5.13. Gradient is calculated c with respect to R cut and are shown using dots. Furthermore, smoothed interpolations d of the gradients have been shown using overlaid lines. At R cut = 3.5 nm, average of gradients over components and three representative systems e is 3.6× 10 − 2 V nm − 2 . 111 5.15 Components of electric field and their gradients calculated at a specific nitrile ni- trogen. At R cut = 3.5 nm, the average of gradients over components and three representative systems is 2.2× 10 − 2 V nm − 2 . 112 5.16 Histogram of electric field components calculated on 200 nitrile nitrogens every 50 ps from 100-200 ns, for ILs with realistic anions 113 5.17 Histogram of electric field components calculated on 200 nitrile nitrogens every 50 ps from 100-200 ns, for ILs with modified chloride anions 114 5.18 Field autocorrelation function for different IL systems 116 5.19 We have reproduced a figure from a prior paper published by our group, which fo- cused on the electronic origin of the Brønsted photobasic effect in 5-R-Quinolines. We observe strong parallels between this figure and the compuational data pub- lished in the main text (Figure Figure 4.5). 2 118 5.20 Panel A shows a schematic of electron displacement in 5-MeoQ between the ground and excited state. Panel B displays an electron density difference map between the ground and first excited state of 5-MeoQ. Blue isosurfaces represent volumes which lose charge density in the excited state, and red isosurfaces correspond to an increase in charge density. As discussed in the main text, we observe an marked in- crease in charge density at the quinoline’s nitrogen. This change in charge density stabilizes the binding of both protons and Lewis acids like BF 3 119 5.21 This figure supplements the computational F ¨ orster cycle shown in Figure Fig- ure 4.5. In order to determine the robustness and method dependence of our results, single point and TDDFT energies were recalculated usingωB97X-D/def2-tzvppd for the geometries used in the main text (calculated using ωB97X-D/def2-svpd). In addition, these energy calculations were performed with a PCM model set to the parameters for acetonitrile, the solvent used for the experimental work. The dotted lines are provided as a guide. 120 5.22 Displayed here is the UV-vis absorption spectra of 5-aminoquinoline and its BF 3 adduct. As opposed to to the other quinolines studied in this work, the aminoquino- line’s absorption blue-shifted from its unadducted form. This result is consistent with BF 3 binding to the free amino group and not the ring nitrogen. We rational- ize this by noting that attaching a BF 3 to the amino group will result in a strong electron withdrawing effect, thereby increasing the L A state. 123 5.23 Displayed here is the UV-vis and fluorescence spectra of 5-methylquinoline and its BF 3 adduct. The sharp rise in intensity in the dotted blue curve is due to a solvent Raman peak. 124 5.24 Displayed here is the UV-vis and fluorescence spectra of quinoline and its BF 3 adduct in dry acetonitrile. 124 5.25 Displayed here is the UV-vis and fluorescence spectra of 5-bromoquinoline and its BF 3 adduct in dry acetonitrile. 125 vii 5.26 Displayed here is the UV-vis and fluorescence spectra of 5-cyanoquinoline and its BF 3 adduct in dry acetonitrile. 125 5.27 This is the same electron density difference map shown in Figure Figure 4.3, but viewed from a different angle to better accentuate the increase of electron density in the vicinity of the B-N bond. As a guide, note that the BF 3 group is towards the right of the image (the boron is colored pink and the fluorides are a light blue-green. 127 5.28 A schematic showing the change in energy as BF 3 goes from a planar geometry to a pyramidal geometry. The pyramidal BF 3 represents the BF 3 fragment’s geometry when it is dative bonded to a Lewis base. The energy of this geometry change lowers the overall dative bond binding enthalpy. A similar comment can be made about the Lewis base, but the engergetic magnitude of this strain is considerably smaller 135 viii Abstract My work as a graduate student at USC focused - in a general sense - on the fundamental physical chemistry of systems important for energy applications. Under this wide umbrella, I spent my time researching two separate areas: chemical structure and electric fields at the electrode/liquid interface, and the photobasic properties of a family of quinolines, which undergo a large increase in proton affinity in the excited state. For the first part of my PhD, I learned a great deal about surface spectroscopy, particularly through the use of surface vibrational Stark shift spectroscopy to investigate the local electric fields that exist at metal interfaces, with and without the application of potential. Ionic liquid structure at the metal surface was a specific area I worked in, and there are many more questions that remain to be answered about the properties of highly concentrated electrolytes. In the second part of my PhD, I studied the photobasic properties of quinolines. My work here largely served to indulge some fundamental curiosity I (and my group) had about these compounds. Prior work in the Dawlaty lab had established that quinolines greatly enhance their proton binding strength in the excited state. I wondered if quinolines could also increase their affinity towards a greater variety of electron acceptors, which would make them Lewis (or dative) photobases as well as Bronsted photobases. My work established a fundamental basis for studying photobases in this way, and the Dawlaty group is currently working to extend these ideas in a more applied context. ix Chapter 1 Introduction A tremendous amount of research in the physical sciences is currently devoted to addressing the world’s energy problems 3 . Our energy problems do not stem from a lack of available energy, but issues with the capture, storage, and use of that energy to perform work. The problem is one of energy conversion, essentially. The earth is constantly bathed in light energy from the sun, and we can readily use electrical power to perform many useful tasks. However, taking that energy and converting it to a useful form (changing electrical energy to a chemical potential energy, for example) at a global scale is still a technological challenge. In order to study these problems as chemists, we need to delve into local environment and molecular physics that dictate critical energetic processes, such as electron transfer between an electrode and a redox active species. Or, if we are interested in a light-activated chemical process, we need to understand the excited state reactivity of the chromophore, and the way that state evolves over time. 1.1 Interfacial Structure at Electrode/Liquid Interfaces Electrocatalysis and electrochemistry hold the potential to drastically change our energy economy - by activation of small molecules like CO 2 or CO to higher value compounds, or by efficiently extracting energy from electrochemical reactions to generate current. Despite the increasing im- portance of electrochemistry and heterogeneous catalysis, understanding of reactions at electrode surfaces is still limited by theoretical and experimental challenges 4 . The essential problem with 1 understanding electrochemical interfaces is a dimensional mismatch: Most electrochemistry is per- formed using a bulk electrode (platinum, or copper, perhaps), often in an aqueous environment, and the potentiostat reads out the total current passed between your counter and working electrodes. However, if you wish to design an electrocatalyst to perform alcohol oxidation, you need to under- stand the specific, molecular interaction between the alcohol, the electrode, and the surrounding environment. On the electrode surface, different metal facets can have distinctly different prop- erties, and the metal surface can undergo reconstruction during an electrochemical experiment 5 . The application of potential to the electrode can directly influence surface pH, thereby changing the reactivity of proton-coupled processes 6 . All of these issues exist before we even add reactant to the solution! But still, we want to do some decent chemistry in this environment. Once you come to terms with the fact that electrocatalysis has to be understood in a molecular fashion, the scope of the problem becomes a little more clear. The most widely known and used theories for interfacial structure tend to be continuum theories 7 , and they are fundamentally unable to provide a molecular picture of the local environment experienced by a chemical species at the surface. The Dawlaty group’s main method of addressing these experimental challenges is to perform surface spectroscopy of a vibrational probe, 4-mercaptobenzonitrile, tethered to the electrode sur- face by a metal-thiol bond. The nitrile stretch of this compound is very well studied by our group and others, and its CN frequency shifts to different degrees depending on the molecular environ- ment 8;9;10 . This phenomenon is known as the vibrational Stark effect, which provides a functional relationship between a perturbing electric field and a measured vibrational frequency 11 . If you understand the Stark response of a vibrational chromophore, you can then use that chromophore as a reporter on its local surroundings, which is particularly helpful for understanding difficult chemical environments. This has allowed researchers like Boxer to use the vibrational Stark effect to understand the catalytic properties of enzymes 12 , and of course, the Dawlaty group has also found the Stark effect to be a productive way of understanding electrode interfacial structure and polarization. For a useful summary of all our work on this topic to-date, I recommend looking at the book chapter recently authored by our group 11 . 2 My research into the structure of electrochemical interfaces is primarily contained in two pub- lished articles, “Electrodes as polarizing functional groups: Correlation between Hammett param- eters and electrochemical polarization”, and “Interfacial Polarization and Ionic Structure at the Ionic Liquid–Metal Interface Studied by Vibrational Spectroscopy and Molecular Dynamics Sim- ulations” 13;14 . The first was authored by my coworker Sohini Sarkar, with supporting authorship including myself, Anuj Pennathur and Joel Patrow. I hold first authorship for the second, with So- hini, Anuj and Joel also on the paper, along with Sevan Menachekanian from the Dawlaty group, and theoretical support from Tanmoy Pal and Qiang Cui from Boston University. The Hammett parameter paper was focused on quantifying (using the nitrile stretch) the fun- damental polarizing influence of an electrode on surface adsorbed species and using that quantifi- cation to hypothesize an empirical relationship to the Hammett parameter, a metric for estimating the effect of a chemical substituent on a chemical property of interest. The work for this paper involved a lot of time in the laser lab setting up hours-long SFG experiments, and developing the scientific concepts for this paper was my first taste of thinking and communicating like a graduate student. The experimental work for the ionic liquid paper was relatively more laborious. We learned that ionic liquids tend to produce horrendous background signals during SFG experiments - mak- ing it nearly impossible to fit CN peaks. We (multiple, sequential researchers in the group) also went through a long period of poor experimental reproducability, which is basically a death sen- tence for a project. After a considerable amount of struggle, we were able to figure out that ILs absorb atmospheric water in high quantities, which changes the surface structure at the 4-MBN monolayer. As an aside, Anuj, Sevan and myself then published a paper studying IL-water mix- tures to understand the impact of bulk water on interfacial IL structure 15 . After clearing up these experimental issues, we were able to produce the experimental trend that we originally hoped for, which was a relationship between the anion size of an ionic liquid and the measured CN stretch at the gold surface. We then struggled to interpret this data for a further amount of time, until our group connected with Qiang Cui, a professor at Boston University. His student Tanmoy ran molec- ular dynamics simulations of our experimental environment, and he found that our experimental 3 observations could be explained by partial intercalation of the anion within the 4-MBN monolayer. This finding randirectly counter to our inital hypothesis of interfacial structure, so we were glad to have their support for this paper. Despite the importance of electrochemical interfaces and the number of researchers working in the field, there are absolutely basic questions that are yet to be fully understood. One, for a tethered species (say thiol tethered, for example) at a charged electrode surface, that species will experience two, somewhat distinguishable effects: electronic polarization due to coupling to the electrode, and electronic polarization due to ionic structure at the interface. Which of these effects tends to dominate surface reactions in general, and what implications does this have for electrocatalysis? Classical electrochemical experiments are necessary tools for tackling these questions, but surface spectroscopy offers a complementary and somewhat finer picture of this environment. 1.2 Photobasicity - Controlling Acid/Base Chemistry with Light The other research half of the Dawlaty group focuses on excited state phenomena, and within that field, our group has primarily researched a class of compounds called photobases 2 . Photobases are the counterpart to photoacids - the elevator pitch for these molecules are that they are weakly basic (acidic) in the ground state, and then become strong bases (acids) upon interaction with light. The photoacids and photobases that are most studied tend to undergo an excited state pKa jump of over 10 units 2 . Proton-coupled reactions are fundamentally important to basic chemistry and biology, and the development of a reliable photochemical platform to control the motion of protons could easily be of use in many applications 16 . Our group has shown that bulk conductivity can be modulated by a photoacidic polymer, and photobases have been installed on a functional iridium catalysts as a proof-of-concept 17;18 The photobasic work first began in the Dawlaty group through the work of Eric Driscoll, and then carried forward by my co-worker Ryan Hunt. Though photobases function through a remark- ably similar mechanism to photoacids, photobases are markedly less studied, and a fundamental 4 understanding of their properties was not as developed. Eric and Ryan initially started by inves- tigating the thermodynamics of quinoline photobases using the F¨ orster cycle, and found that the excited state basicity of quinolines depended linearly on the Hammett parameter of a substituent at the 5-position, similar to the ground state basicity 2 . However, the excited state pKa depends more strongly on the substituent than the ground state pKa (in other words, you get more bang for your buck in the excited state). After this work, they used transient-absorption spectroscopy to track the kinetics of proton transfer to photobases at ultrafast timescales 19 , over the same range of substituents. Intriguingly, they found that even if a photobase has the thermodyanamic drive to capture a proton in the excited state, that does not always translate to this process being observed kinetically. Ryan’s work after this did an excellent job of framing the function of photobases in a more functional, practical way. For me in particular, seeing him work through and present his results helped me transition from thinking about photobasicity as an abstract concept to a more physical one. His first paper showed that quinoline photobases are able to deprotonate alcohols, generating an excited state titration by tuning the acidity of the proton donor 20 . In addition, his transient absorption data showed that a free energy relationship exists for quinoline proton transfer kinetics. Following that, he and Cindy Tseng did some work to explore the overall design space of aromatic photobases, along with further kinetic studies 21 . He followed this up with an additional kinetics paper, which showed that even if a photobase has the thermodynamic drive to deprotonate a single proton donor, two proton donors are kinetically required to allow the process to take place 22 . I started my work on the photobase concept as a relative outsider, and my paper reflects that to a large degree. After seeing all of this great, fundamental work by Ryan, I wondered if the fundamental properties of photobases could apply to a wider class of chemistry than protons alone. This question is summed up in the title of the paper we published, ”Can Bronsted Photobases act as Lewis Photobases?” 23 . This work was inspired by a combination of Joel’s work on interfacial Lewis acid-base chemistry 24 and all of Ryan’s work on photobases. The clear communication within our group and in our old papers allowed me to take the right experimental approach to generate data in support of the initial hypothesis. In addition, Andrew Petit’s group published 5 theoretical work on quinoline photobases 25 which made it fairly easy for me to adapt in order to calculate a theoretical F¨ orster cycle for quinolines and BF 3 . 6 Chapter 2 Electrodes as Polarizing Functional Groups: Correlation between Hammett Parameters and Electrochemical Polarization 2.1 Introduction We create a fundamental link between seemingly two distinct concepts in two different fields of chemistry− the concept of electron withdrawing by a functional group as measured by the Ham- mett parameter and the concept of polarizing electric fields at electrochemical interfaces. The for- mer is used in synthetic organic and organometallic chemistry and homogeneous catalysis, while the latter is used in electrochemistry and heterogeneous catalysis. The induced polarization is intimately related to reactivity via influencing the stability of reactive intermediates, and the ener- getics of transition states. Molecular polarization is controlled by either functional groups or by an external electrochemical potential. While it seems natural and intuitive that a fundamental link must exist between the polarization induced by a functional group and that created by an external field, surprisingly, no work has been done to establish this connection. We present experimental and computational evidence to create such a link that we anticipate will benefit several areas of chemistry. The link will serve as part of a common language between heterogeneous and homogeneous chemistry to facilitate transfer of ideas between the two fields, and connect seemingly disparate phenomena in these 7 realms. For example, if one understands how an organometallic catalyst responds to the Hammett parameter of a functional group within its ligands, one can extrapolate its behavior when it is attached to an electrochemical interface. The link can also shed light on the persistent and difficult electrochemical problem of estimating pK a of molecules near a biased interface. Similarly, the influence of a substituent on a homogeneous reactive center can be interpreted as an effective electrochemical potential, and therefore its redox properties can be estimated. This paper is organized as follows. First, we introduce the concepts of Hammett parameter, interfacial fields, and their interconnection with particular reference to their measurement using vibrational spectroscopy. Second, we present our experimental results, supported by computa- tions, showing vibrational frequency shifts of benzonitrile as a function of Hammett parameter and applied electrochemical potential. Third, we will discuss how Hammett parameter and electric field are fundamentally connected to each other. Finally we will discuss the consequences and limitations of this correlation for understanding chemical phenomena both at interfaces and in the bulk. 2.2 The Hammett Parameters, Interfacial Fields, and their Vibrational Signature In this section we set the scene by introducing the concepts and tools that are used in this work. The Hammett parameter (σ) is a metric commonly used by organic and organometallic chemists to express the polarizing influence that a functional group exerts on a chemical system. It is tradition- ally defined by the influence that a functional group R exerts on an acid dissociation reaction 26;27 . The acid dissociation constant K 0 of unsubstituted benzoic acid is often taken as a reference (pK a = 4.2). Then the acid dissociation constants K of a series of R-substituted benzoic acids are mea- sured. Electron-withdrawing (EW) groups help stabilize the anionic benzoate and therefore fa- vor the reaction towards more dissociated or larger equilibrium constants K. Electron-donating (ED) functional groups exert the opposite influence. The Hammett parameter sigma is defined as σ =(1/ρ)logK/K 0 , where ρ is a constant (that is set to 1 for benzoic acid). Positive σ values 8 indicate EW substituents, while negativeσ value indicate ED substituents. Several variants of the Hammett parameter exist, with some that attempt to distinguish between the resonance and induc- tive effects of a substituent 28 . In this work we consider the commonly used Hammett parameterσ p based on para-substituted benzoic acids. Both experimental and computational works show corre- lation between a substituent’sσ p value and an extraordinarily diverse range of chemical properties, including catalytic activity 29 , photoacidity 30 , photobasicity 2 , π-conjugation strength in aromatic systems 31 , along with many others 32;33;34;35;36;37;38;39;40;41;42 . This general applicability justifies fundamental work to further understand its relation to other fields, in particular to electrochemistry. Similar to the polarizing effect of a functional group, a biased electrochemical interface has a polarizing influence on the nearby molecules. The applied potential on an electrode in contact with a high ionic strength electrolyte decays rapidly due to screening by ions, often over length scales on the order of∼ 10 nm. Therefore, modest electrochemical potentials (∼ 1 V) correspond to very large potential gradients or electric fields near the electrode ( ∼ 1 MV/cm) 8;43 . Such large fields polarize the molecules near the surface, and when large enough, they drive redox reactions. Inter- estingly, similar to the electrochemical fields, it is proposed that enzymes maintain large electric fields within their reactive sites that help catalyze reactions 12 . While the importance of electro- static fields in chemical reactions is known for several decades 44;45 , the topic has gained special attention recently 46;47;48;49 , with particular emphasis towards creating oriented electric fields for driving catalytic reactions 50 . Thus, viewing chemical reactivity from the electric field perspective has wider utility and is not limited to interfacial electrochemistry alone. Measuring electric fields and correlating them to existing chemical concepts, as is done in this work, is a necessary endeavor. To measure local electric fields in a chemical environment often Vibrational Stark shift spec- troscopy is used. Probe molecules, bearing vibrational tags, are placed at desired locations 51;52;53;54;55 . Their measured frequency shifts with respect to reference environments is related to the local elec- tric fields within the linear Stark shift approximation as hν=− ∆⃗ µ· ⃗ F, where ⃗ F is the electric field and∆⃗ µ, also known as the Stark tuning rate, is the change in dipole moment between the ν 0 and 9 ν 1 vibrational levels. Often the nitrile group, which has a relatively isolated and narrow vibra- tional signature, is used as a probe 56 . Benzonitriles, with a Stark tuning rate of∆⃗ µ = 0.022D∼ 0.36cm − 1 /(MV/cm) have been used by us and others for measuring local fields 55;10;57;58;59 . To make the connection between the two central concepts in this work, vibrational spectroscopy is a natural choice as will be explained below. We choose benzonitriles as our probe molecule as shown in figure Figure 2.1. It is well-known that the frequency of the nitrile stretch in benzonitrile is sensitive to changes in Hammett parameter. 52;53;60;55;12 This is likely because the substituent po- larizes the benzeneπ electrons which have some overlap with the antibonding orbital of the nitrile moiety. Therefore any agent, be it an electron donating substituent or an external field, that pushes charge density from the ring towards the nitrile group will red-shift the stretch frequency 61;62 . Similarly any effect that polarizes the charge density in the ring away from the nitrile group will blue-shift its stretch frequency. Thus, the nitrile stretch is a natural, and convenient choice for as- sessing polarizing forces both arising from functional groups and from external fields. Both of the above properties of benzonitrile are established from extensive earlier work. The effect of substi- tution on the nitrile stretch of 4-R-benzonitrile demonstrated a clear frequency shift with respect to σ p 60 . Additionally, 4-mercaptobenzonitrile has been used previously to probe fields at electrified metal interfaces, and Stark behavior of the CN moiety at the surface is well documented by us and others 63;8;55;43 . Our purpose here is to unite these two views, which has not been performed so far. To understand the influence of the Hammett parameter of a substituent on the nitrile fre- quency, we measure the vibrational frequencies of a series of benzonitrile molecules that are substituted at the para-position with functional groups spanning a wide range of Hammett pa- rameters. To understand the influence of electrochemical potential on the nitrile frequency, we tether 4-mercaptobenzonitriles on a gold surface and polarize it in an electrochemical cell. Us- ing vibrational Sum Frequency Generation (vSFG) spectroscopy, we measure the nitrile frequency shift as a function of applied potential. Nitrile vibrational frequencies are sensitive to solvation environments both in the bulk and near a surface as noted by us and others 8;64;65 . Therefore, it is critically important to keep the solvent the same between the electrochemical and Hammett mea- surements. This restriction precludes aqueous electrolytes since many of the substituted molecules 10 Hammett Electron Withdrawing Parameter Electric Field Field Figure 2.1: Picture depicting the effect of polarizing electric field at electrochemical interface and an electron withdrawing functional group on the nitrile stretch. in the Hammett series are not soluble in water. Furthermore, the relatively narrow electrochemical window of water limits the range of potential that can be applied. We chose an ionic liquid as a solvent and electrolyte, since ionic liquids exhibit wider electrochemical windows and can easily dissolve a wide range of organic molecules, including all of our Hammett series molecules. 2.3 Experimental and Computational Methods FTIR spectra of benzonitrile and para-substituted benzonitriles were recorded using a Bruker Ver- tex 80 FTIR spectrometer. A series of compounds spanning the Hammett parameter σ p from -0.83 to +1.11 were used. The names of the molecules and their Hammett parameters are listed in the SI and also appear in figure 3. The Hammett parameters values are based on previous literature 27;28 . For each of the compounds, the concentration was maintained to be∼ 2 mmol. All compounds were dissolved in 1-Ethyl-3-methylimidazolium tetrafluoroborate abbreviated as [EMIM] + [BF4] − . The ionic liquid (IL), [EMIM] + [BF4] − , was purchased from Sigma Aldrich and nitrogen was bubbled through the liquid at 60 0 C to remove oxygen and water 66 . The solutions in IL were injected between two calcium fluoride plates separated by a 100 µm spacer and held 11 together using a demountable liquid FTIR cell (International Crystal Laboratories). A table with the names of substituents, their Hammett parameters, and the -CN stretch frequencies are given in the SI. Silicon wafers with a 10 nm Ti adhesion layer and 100 nm of Au purchased from LGA Thin Films, Inc were used to prepare self-assembled monolayers (SAMs) of 4-mercaptobenzonitrile (4- MBN). The wafers were sonicated in acetone and then in ethanol twice for 8 min each time to ensure cleaning. The cleaned wafers were immersed in a 0.03 M solution of 4-MBN in ethanol for at least 24 h, which ensures full surface coverage for good signal quality. After soaking in the 4-MBN solution, the wafers were removed and again sonicated twice in fresh ethanol for 8 min each. For recording the electrochemical sum frequency generation spectra, the SAM-modified Au samples were mounted in the liquid FTIR cells (International Crystal Laboratories) adapted for this purpose. The front window comprised a 4 mm thick CaF 2 window with two holes using for filling. The back window of the cell was replaced with the SAM-containing wafer. The two windows were separated by a 25 µm Teflon spacer. The entire assembly was held firmly together using stainless steel plates and screws (Figure Figure 2.2 b.). The cell was then placed under a N 2 atmosphere where the IL was then injected via a large syringe into the cell. After injection, the syringe was disconnected from the cell and its plunger was removed. Then, syringe (now without a plunger) was reconnected to one of the cell’s filling ports and used as a reservoir to hold the counter and reference electrodes. The working electrode consisted of the SAM-modified Au samples with attached wires to connect to the potentiostat; The reference electrode was Ag/AgCl (purchased from Gamry); A Pt wire was used as the counter electrode. A Gamry Reference 3000 potentiostat was used for applying potentials. The optical light source was a Ti-sapphire amplified laser (Coherent) that generates ultrafast near-IR pulses at 1 KHz repetition rate. 1 W of the average power was directed to an optical de- lay stage followed by a 4f filter that significantly narrowed the spectrum. The 4f filter built using two transmissive volume phase gratings (BaySpec, Inc.), two cylindrical lenses, and a variable 12 width slit could filter near-IR pulse to a spectral width of 8.0 cm − 1 , centered at 783.35 nm. An- other portion (2 W) of the Ti-sapphire average power was fed into an optical parametric amplifier (OPA;Coherent OPerA Solo) that generated mid IR pulses via difference frequency generation in a AgGaS 2 crystal. Pulse energies at sample position for near IR and mid IR pulses were∼ 8 µJ and∼ 6.5 µJ respectively. vSFG spectra were acquired by spatially and temporally overlapping these two pulses after they were focused using a common parabolic mirror. The resulting vSFG signal was collected with a second parabolic mirror and passed through a 750 nm short-pass filter to reject the majority of the scattered near-IR photons prior to being sent to a spectrometer (Horiba iHR320)with a CCD camera (Syncerity) for spectral analysis. All spectra were acquired under a CO 2 and H 2 O free atmosphere. Each spectrum was obtained by integrating for 120 seconds. The spectral resolution of this setup is discussed in our previous work 8 . (b) V A (a) (c) + - Contact to working electrode Counter electrode Reference electrode Filling Syringe Metal scaffold 25μm Teflon spacer 100nm gold layer on Si Wafer 5 mm thick CaF₂ window + - - - Figure 2.2: (a) Diagram showing a cross section of the cell used in experiments. (b) Cartoon diagram of SAM-modified Au sample in [EMIM] + [BF4] − . (c) Representative spectrum (blue) and its fit (red) at V = 0.0 V . The dip around 2230 cm − 1 is due to the nitrile stretch. The electrochemical vSFG studies were carried out as a function of the applied potential. The electrolyte used was [EMIM] + [BF4] − which is a room temperature ionic liquid. Ideal room tem- perature ionic liquids are organic salts with mobile anions and cations and no solvent. The ionic liquids have decent ionic and electrical conductivity, electrochemical stability, low vapour pressure and ability to dissolve a wide range of substances. Correspondingly they have found applications in batteries, fuel cells, solar cells, and electrochemistry 67;68;69 . We chose [EMIM] + [BF4] − as an electrolyte for performing electrochemical vSFG owing to its wider electrochemical window (∼ 13 4V) compared to that of water 70;71 . The SAM has a narrow potential window from -1.4 V to +0.6 V where the adsorbed thiol on gold is electrochemically stable. A cyclic potential scan was per- formed where the applied potential was scanned from 0.0 to +0.6 V then to -1.4 V and then back to 0.0 V with a step size of 0.2 V . After application of each potential step, a transient capacitive current was observed. Six minutes was allowed to let the transient current decay before acquisition of the vSFG spectra. This ensures that the steady state of the electrode is measured for each poten- tial value. At each potential, three vSFG spectra were obtained each with 120 s integration times. As known from prior literature, the nitrile stretch appears as a negative narrow Lorentzian line in- terfering with a broader Gaussian background arising from the non-resonant response of the gold electrode 8;55;72 . Following previous work, we used the following equation to fit our experimental data and obtain the centre frequencies of the nitrile stretch: I SFG (ω)= A NR e iφ + B ω− ω CN + iΓ CN 2 exp − (ω− ω g ) 2 σ 2 g The above model utilized by Benderskii et al. 73 is for the total vSFG signal arising from the interference of a nonresonant background signal from gold, and a resonant signal from the ben- zonitrile adhering to the gold surface. An amplitude A NR with relative phaseφ is used to model the constant non-resonant background. The resonant signal adopts a Lorentzian line shape with ampli- tude B, center frequency ω CN , and widthΓ CN . Both the signals are multiplied by an Gaussian IR pulse, with center frequencyω g andσ g . A background spectrum was obtained by walking off the temporal overlap of the IR-near-IR pulses under otherwise identical conditions. This background was subtracted from the raw vSFG spectra. All of the fitting parameters for the vSFG spectra are listed in Supporting Information (SI). Each vSFG spectrum at each potential was independently fit to the above equation using Matlab’s nonlinear least-squares fitting algorithm and same initial guess. A representative spectrum with its fit is shown in Figure 2.c. As the applied potential is scanned from 0.0 V to +0.6 V , the electrode is positively polarized which in turn polarizes the probe molecule that stiffens the CN vibration resulting in blue shift. For potentials much larger than +0.6 V the nitrile stretch begins to disappear due to the instability of the SAM to oxidative chemistry. Thus +0.6 V is the upper limit of our electrochemical window. 14 In the positive potential regime negligible current traverses the interface. For negative potentials, some electrochemical current along with some hysteresis both in the current and in the nitrile frequencies is observed. A plot showing current vs applied potential is given in SI. Figure 4.b. shows the frequency of the nitrile stretch as a function of applied potential. Variations due to hysteresis, along with variations within several replication of the experiment (see SI for further detail), are all accounted for in the error bars. Computational work was carried out using the quantum chemistry package Q-Chem 74 . Den- sity functional methods were used for geometry and frequency calculation using B3LYP functional and the 6-311G(d,p) basis set for the substituent and field studies. The field calculations were performed by generating arrays of point charges to produce a uniform electric field across a ben- zonitrile molecule. The method is described in more detail in our previous work 72 . Briefly, the arrays were generated to mimic the two oppositely charged plates of a capacitor. Fields of different strengths were simulated by uniformly altering the magnitude associated with the charges within the array. Geometry and frequency calculations were completed as described above in fields rang- ing in strength from -100 MV cm to 100 MV cm . Charge densities on the nitrile nitrogen were obtained through CHELPG charge analysis as printed in the Q-Chem output file. 2.4 Results FTIR spectra of p-substituted benzonitriles are shown in Figure Figure 2.3. The substituents have been varied spanning a wide range of Hammett parameters (σ p ) from -0.83 to +1.11, with in- creasing Hammett parameter corresponding to more electron withdrawing by the substituent. As mentioned earlier,σ p = 0 corresponds to unsubstituted (i.e. hydrogen-substituted) benzonitrile and the positive (negative) values correspond to more (less) electron-withdrawing groups with respect to hydrogen. Consistent with previous work, 52;53;60 we observe that as the Hammett parameter increases, the nitrile frequency blue shifts (Figure 2.3.b.). This is likely because the charge den- sity on the benzene ring has some overlap with the CN antibonding orbital and that any external influence that polarizes this charge density away (towards) the nitrile group will result into blue 15 (red) shift of its vibrational frequency. 75 This trend is clearly seen in figure (Figure 2.3.b.). where nearly 25 cm − 1 of shift is observed within the studied range ofσ p . A certain level of scatter is ob- served in the data. It is natural to suspect whether interaction with the solvent results into deviation from an otherwise stronger correlation. To understand whether the scatter arises from variation of solvatochromic shift within the series, we computed the nitrile frequencies for the molecules within this series in the gas phase. The computed frequencies (figure 3.c.) confirm the overall trend of increasing frequency with increasing Hammett parameter. In addition, they also show a qualitatively similar range of scatter as in the experimental data. Therefore, the level of correla- tion between the frequency and the Hammett parameter is internal to the molecule and does not arise from variations in the solvatochromic shifts. Given that the data is a correlation between very different observables- frequency in benzonitrile obtained from spectroscopy and Hammett parameter in benzoic acids obtained from titration experiments- the observed level of correlation is quite reasonable. Since the purpose of the computation here was to understand the trend and not the absolute frequencies, exact matching of each data point between experiment and computa- tion is unnecessary and unexpected at this level of theory. As expected and known for DFT-level computations, frequencies are always overestimated 76 . This is reflected in the data shown in the figure. (c) (b) (a) Figure 2.3: (a) Representative FTIR absorption spectra of the compounds in the nitrile stretch region. (b) Experimentally determined central frequencies of the nitrile stretch in a series of para- substituted benzonitriles versus the Hammett parameters of the substituents. (c) Computed central frequencies of the nitrile stretch for the Hammett series. 16 For the nitriles tethered on the electrochemically biased interface, vSFG spectra were obtained as discussed in the experimental section. The nitrile stretch frequenciesν CN were retrieved from the fitting procedure as discussed earlier. Spectra of the nitrile stretch segregated from the nonresonant background are plotted at different applied potentials (Figure 4.a). The nitrile frequency at 0.0 V with respect to the Ag/AgCl reference electrode is 2229.4 cm − 1 . Positive potentials polarize the electron density in the benzonitriles towards the electrode and away from the nitrile group. Therefore, similar to the influence of the EW groups, electron density on the antibonding orbital of the CN decreases and blue shifts the CN stretch. At negative potentials the molecule is polarized such that electron density is pushed on to the CN antibonding orbital, resulting into softening of the CN bond. This behavior is manifested in the data in figure 4.a-b. The error bars in the figure include variations over several iterations of the experiment and account for the slight hysteresis in electrochemical cycles. Further details on the error bars is given in the SI. Within the potential window of 2 V we observe a 9.1 cm − 1 shift or nearly 4.5 cm − 1 /V . Such frequency change with respect to potential is comparable to that observed in our previous study 55 when the electrolyte was∼ 100 mM KCl solution. (a) (b) (c) Figure 2.4: (a) The nitrile stretch spectra isolated from vSFG spectra. The data shows variation of the spectra as a function of applied potential. (b) Experimentally obtained nitrile stretch frequency as a function of applied potential. (c) Computed nitrile stretch frequency with varying applied field. It is important to note that the interfacial electric field, and not the electrochemical potential, is the fundamentally important quantity in considering the polarization experienced by the tethered 17 molecules. The applied electrochemical potential decays away from the electrode into the elec- trolyte in a way that is dependent on the electrolyte type and concentration. The electric field, or the gradient of this potential, is not only a function of the externally applied potential, but also a function of the electrolyte type and concentration. Conventionally, while the potential can be easily controlled and measured electrochemically, interfacial fields can only be inferred from a model of how the applied potential decays. Fortunately, placing a Stark shift reporter with known properties at the interface makes it possible to measure the field spectroscopically, without heavily relying on a model. A body of literature on the Stark properties of benzonitrile, including our work for the case of electrode-electrolyte interfaces and interfacial solvation 61;55 , exists. Therefore, it is possible to convert our spectroscopic frequency shifts to electric field values as will be done in the next section. To further verify the validity of using linear Stark shift within this range, we computed frequencies for benzonitrile in the gas phase in electric fields as explained in the com- putational methods. Just as with the computations for the Hammett series, it is neither expected nor necessary to match the experimental and computed frequencies exactly. The result verifies that linear Stark shift is a reasonable model within this range and that computationally retrieved Stark tuning rate∆µ= 0.35cm − 1 is very close to the previously reported experimental values 72 . As will be shown in the next section, to match the entire range of the Hammett series to electric fields it will become necessary to go beyond the linear Stark approximation. Finally, as shown in Figure 4.a., the line width of the nitrile stretch is a function of the applied potential, with the line width narrowing for more positive potentials. Interestingly, a similar effect can be seen in the line widths of the Hammett series (figure 3.a.), where larger Hammett parameters also correspond to narrower lines. This intriguing observation further confirms the common origin for not only the central frequency of the nitrile between the two sets of experiments, but also its line width. Spectral broadening associated with substituents has been observed in previous work investigating the carbonyl stretches of substituted acetophenones 77 . The observed broadening was associated with the thermal population of rotational states and interconversion between cis and trans isomers. However, due to the symmetry and structure of the benzonitrile molecule in the bulk ionic liquid and within the Au monolayer, we can rule out the above as sources of our observed 18 broadening. Detailed study of the line widths in these experiments is the subject of future studies since they relate to solvation and spectral dynamics which are outside the realm of this paper. 2.5 Discussion Based on the experimental and computational data presented so far, we will arrive at a relation between the Hammett parameterσ p and the electric field F using the nitrile stretch frequency as a tool. We will perform this task at three levels. First we will use the range of frequencies for which we have overlapping experimental data for both electrochemistry and the Hammett series. Second, due to the limited range of the electrochemical data, we will use the entire range of frequencies spanned by the Hammett series against computed Stark shifts, including nonlinear behavior of frequency at large field values. Finally, we discuss a direct relation between the two concepts that does not rely on the nitrile frequency and is based on computed charge densities. First, we use the known Stark tuning parameter∆µ = 0.36cm − 1 /(MV/cm) 10;55 to convert the experimentally measured frequencies as a function of potential to electric field values F. The use of linear Stark shift in the relatively narrow range of experimentally observed frequency changes is justified as was explained in the previous section. If we measure the electric field relative to the field when the applied potential is zero, we can write the following relation: ν CN (F)=(0.36 cm − 1 MV/cm )F+ 2229.4cm − 1 (2.1) where ν 0 = 2229.4cm − 1 is the frequency at zero applied potential relative to the reference electrode. The above relation is applied to the experimental data and plotted in figure 5.a. The data spans a range of frequencies from 2224 cm − 1 to 2232 cm − 1 . Although it is desired to experimen- tally apply a wider range of electric fields, but unfortunately this range is limited by the tolerance of the SAM to redox chemistry and the electrochemical window of the electrolyte. Second, within the same range of frequencies that were spanned by the electrochemical mea- surements, we collect the frequencies of all of the molecules in the Hammett series and plot them versus their Hammett parameter (Figure 5.b), yielding essentially a zoomed in version of figure 19 (a) (b) (c) Figure 2.5: Experimentally obtained frequencies as a function of (a) applied field (b) Hammett parametersσ p ranging from -0.37 to +0.45. (c) Relation between the applied field and the Hammett parameterσ p . 3.b. As commented in the earlier section, the scatter in this data is not a consequence of experi- mental noise or solvatochromic differences between different molecules, but rather inherent to the molecules and is reproduced in computations. Fitting this data to a linear relation yields: ν CN =(10.15± 4cm − 1 )σ p +(2228± 0.86cm − 1 ) (2.2) As argued before, the CN frequency responds to the polarization of the molecule, whether that polarization is induced by an external field or by a functional group. Therefore, it justifies comparing equation 2.1 and equation 2.2 to yield a relation between electric fields F andσ p as: F = 27.5(MV/cm)σ p − 2.71(MV/cm) (2.3) This relation is our first main result that can be used to inter-convert between the Hammett parameterσ p and the electric field F. Two main limitations should be considered when using this relation. First, the range of experimentally applied fields is relatively narrow, spanning a Hammett range of − 0.37≤ σ p ≤ 0.45 as seen in figure Figure 2.5.b. Second, the assumption of linear fit of frequencies to Hammett parameter clearly does not hold for larger values as seen in figure Figure 2.3.b. 20 To understand this problem, we extend the range of electric fields in our computation to span a wider range of Stark shifts (figure Figure 2.6.a). Interestingly, the computed results in figure Figure 2.6.a show that while a linear dependence of frequency versus field is valid over a narrow range, the computed frequencies deviate from linear dependence in a way that is very similar to the dependence of frequency on the Hammett parameter. At large positive fields and Hammett parameters the nitrile frequency change tapers off and does not change as rapidly as for large negative fields and Hammett parameters. This observation is yet another piece of evidence in favor of the fundamental relation between these two concepts over a wide range. To create a ν CN (F) relation that would be valid over a larger range, we fit the computational results presented in figure Figure 2.6.a to a second order polynomial: ν CN (F)= AF 2 + BF+C (2.4) with the fit yielding A=− 0.003 cm − 1 (MV/cm) 2 B= 0.37 cm − 1 (MV/cm) and C =− 0.19cm − 1 . This fit, along with the computed frequencies, is shown in figure Figure 2.6.a. (a) (b) (c) Figure 2.6: Computed nitrile frequencies as a function of (a) applied field (b) Hammett parameters σ p . Red circles are the data points and black lines are the fits. (c) Relation between electric field andσ p . Next, to create the ν CN (σ p ) relation, we return to the computed frequencies for the Hammett series presented in figure Figure 2.6.b. Use of computed frequencies is reasonable since they will be correlated with computed Stark shifted frequencies. Fitting the frequencies versus the Hammett parameters to a second order polynomial yields: 21 ν CN (σ p )= Aσ 2 p + Bσ p +C (2.5) where A=− 3.94± 2.75cm − 1 , B= 9.90± 1.82cm − 1 and C= 2349.5± 1cm − 1 . Figure Fig- ure 2.6.b shows this fit superposed on the computed data. Equipped with equations 2.4 and 2.5, we arrive at a relation between electric field and Hammett parameter that is based on a broader range of parameters and accounts for the nonlinear behavior of frequency versus either of the two parameters. This relation is presented in figure Figure 2.6.c. When fitted to a second order polynomial, it yields: F = Aσ 2 p + Bσ p +C (2.6) where A=− 4.84± 0.12, B= 24.56± 0.08 and C=− 3.25± 0.05. This is our second and refined result for connecting the Hammett parameter to electric field. A plot of this relation is shown in figure 6.c. Finally, unlike in the experiments, computation allows for directly estimating the charge den- sity ρ on any atom in the molecule either as a function of substituent or external field. Just like the frequency, charge density is also sensitive to the external field and Hammett parameter 41;49 . Therefore, as an alternate means of connecting the Hammett parameter to electric field, we choose to use the charge density on the nitrogen atom of the CN moiety as a reference. Towards this goal, we plot the charge density on nitrogenρ N as a function of external field and as a function of Ham- mett parameter. The plots show that ρ N has a reasonably strong linear correlation with respect to both field and the Hammett parameter. This is in accordance with the intended wide usage of the Hammett parameter in chemistry and is consistent with previous studies 29;32;37 . However, here we take an additional step by studying ρ N with respect to an external field to allow linking external fields with the Hammett parameter. Following a similar approach as for the frequency data, ρ N can be fitted as a function of F andσ p , and the two resulting equations correlated with each other and fitted to the following linear relation (see SI for details): 22 F = 15.3(MV/cm)σ p (2.7) (a) (b) (c) Figure 2.7: Computedρ N as a function of (a) applied field (b) Hammett parameters σ p . Red circles are the data points and black lines are the fits. (c) Relation between electric field and σ p . Therefore, based on this method, one unit of Hammett parameter corresponds to 15.3 MV/cm of polarizing external field. There is no offset in this relation because the reference molecule for both the Hammett-ρ N and field- ρ N relations is unsubstituted benzonitrile in zero field. Figure Figure 2.8 summarizes the results of the three methods used in our study to correlate Hammett parameter σ p and electric field. To reiterate, these methods are based on experimental frequency changes, computational frequency changes, and charge density changes on nitrogen in response to bothσ p and electric field. It is important to note that the relation retrieved based on experimental results matches very closely with the relation retrieved from computational results for frequencies. The experimental relation (blue line) is retrieved from two very different experiments - electrochemical vSFG Stark shift spectroscopy of molecules tethered on a surface and bulk FTIR spectroscopy of the Hammett series molecules. The computational relation (green line) is based on frequency calculations at DFT level within an electric field and for the Hammett series. The fact that they corroborate each other is a testament to the reasonability of our approach. 23 From calc. N From N From expt. CN C Figure 2.8: Relation betweenσ p and electric field based on experimental frequency changes (blue), computational frequency changes (green), and charge density changes on nitrogenρ N (red) We also note that the relation based on charge densityρ N (red line) deviates from that retrieved based on frequency measurements and calculations (blue and green lines), especially in the nega- tive Hammett parameter region. This deviation suggests that the results should be used with some restriction. To understand this better, we clarify with an example. An electric field of about - 8 MV/cm would produce the same charge density on nitrogen as that produced by a functional group at the para position with Hammett parameter ofσ p =− 0.5 (red line in the figure). However, a much larger field (about -18 MV/cm) would be necessary to produce a frequency shift that is equivalent to a functional group ofσ p =− 0.5 (green line in the figure). Therefore, in assessing the equivalence of the electric fields to the Hammett parameters, one should consider the molecular property that is expected to be influenced. Finally, we highlight the utility of the relations between field and Hammett parameter that we have discussed so far. These relations can be used to connect seemingly different homogeneous and heterogeneous chemistry phenomena. For example, charge density on an atom, such as a hete- rocyclic nitrogen in a conjugated system, is clearly related to the Hammett parameter of functional groups in that system and this charge density influences its pK a . If such a system is placed at an electrochemical interface, the interfacial field can also change the pK a of that nitrogen. Using the 24 relations in this work, it will be possible to estimate the pK a as a function of interfacial field, and consequently as a function of electrochemical potential. Similarly, the substituents in the periph- ery of a homogeneous organometallic system affect the energetics of the frontier orbitals of the metal center. If that system is placed near an electrode, similar changes can be achieved using a polarizing field. The relations discussed in this work are useful in estimating the fields necessary to achieve these changes. Caution should be exercised when using these relations in cases where inhomogeneous fields are applied, or when the molecules have a more complicated substitution pattern. Further work is needed to understand such intricacies. 25 Chapter 3 Electric Fields and Structure at the Ionic Liquid/Metal Interface Ideal room temperature ionic liquids (ILs) are liquid organic salts with relatively mobile anions and cations. They have opened and continue to open new frontiers in applications 78;79 such as in batteries, fuel cells, solar cells, electrochemistry 80 , and ionic thrusters for propulsion in space missions 81 due to their unique properties, including electrochemical and thermal stability, low vapor pressure, and the ability to dissolve a wide range of substances. Most of these applications rely upon understanding and engineering the behavior of ILs at interfaces with other materials 82;83 . However, their behavior at an interface is often a far cry from the dilute electrolyte interfaces. A few of the central differences are highlighted below. They have strong Coulomb correlations. Even in a concentrated conventional electrolyte (e.g. 1 M aqueous HCl) the ratio of ions to solvent molecules are in the order of∼ 1:50. In a pure IL the entire liquid is made of ions and strong interaction between ions can not be ignored. For example, ion pairing can change the essential properties that are of value to applications such as conductivity, viscosity, and interfacial kinetics. The degree of ion-pairing in ILs, especially near an interface is a subject of current research 82;84 . Ionic liquids may be thought of as “liquid plasma” and have even been suggested as a test bed for understanding the complexities of plasma physics theories 85;86;87 . Furthermore, the ions have non-negligible sizes, and size difference between the cations and anions can vary largely. This leads to complex structure formation near the interface. 26 Coulomb interactions compete with other specific intermolecular forces such as hydrogen bond- ing, van der Waals, and hydrophobic/hydrophilic interactions. Such competition affects physical properties such as melting point and viscosity. Albeit complicated, changing the composition of the ions can promote or demote the importance of a given interaction and serve as a handle in tuning their properties. Given the above, it is not surprising that conventional theories of electrolytes near interfaces (analogues of Gouy-Chapman theory and its variants) do not generally hold for ILs. Of the many facets of structure and dynamics of ILs, we are interested in only one: local electric field near a surface, which we argue is complicated, and crucial for many applications of ILs. Experimental and computational work 88;89;90 has shown the potential profile away from an electrode into the IL is non-monotonic and oscillatory, reflecting the underlying layered structure of ions near the electrode. Increasing Charge Density (e-/nm 3 ) 4 6 8 10 12 14 16 Figure 3.1: Chemical structure and name abbreviations of the anions studied in this work and their corresponding molecular charge density. The volume term in the charge densities was estimated using DFT-optimized structures for each anion. Ionic liquids have a vast chemical space 82;91 . Their properties can be tuned by somewhat independent choice of anions and cations, with a variety of sizes and substituent groups, leading to millions of possible ILs even by conservative estimates, and thousands already reported in the literature. Rather than randomly searching the parameter space, it is necessary to identify themes and build a general understanding of their behavior at interfaces. In this study, we investigate the surface structure as a function of anion size. 27 Electric fields are increasingly understood to be fundamentally important in understanding and modifying chemical reactions 92;93;94 . In electrochemical contexts, the rapid drop-off of potential away from the surface can result in electric fields on the order of 1 V/nm, providing the driv- ing force for electron transfer to the electrode. Despite the importance of these interfacial fields, their direct experimental measurement is nearly impossible with many conventional techniques. The usual experimental approach is the measurement of differential capacitance as a function of potential. Often such measurements fall under the general umbrella of impedance spectroscopy, in which a static background potential with an added small amplitude oscillatory potential is ap- plied to the interface and the complex response of the interface (consisting of both resistive and capacitive components) is measured as a function of frequency. Then the data is modeled using an assumed equivalent circuit for the interface and the capacitance of the interface is inferred. Even if we ignore the inherent reliance of this method on an assumed equivalent circuit model, these mea- surements probe the entire double layer and necessarily average out the intricate field variations near the surface. The interpretation of such measurements for ILs is heavily debated in the litera- ture 95;96;97;98;99 and a clear microscopic picture is yet to emerge. We propose that more localized measurements of the solid/IL interface may better resolve the nuances of double layer structure in concentrated electrolytes. 3.1 Approach Our work provides a new and independent outlook to this problem by combining a spectroscopic method, Vibrational Sum Frequency Generation (vSFG), with molecular dynamics (MD) simula- tions. Together, our results provide a detailed view of interfacial structure grounded in experimen- tal observables. To understand the electrostatic polarization of molecules at the metal-IL interface, we measure the nitrile frequency of 4-mercaptobenzonitrile (4-MBN) SAMs adsorbed at the gold surface in the presence of a range of ILs varied by their molecular anion volume (Figure Fig- ure 3.1). 4-MBN is a well-understood vibrational Stark probe and has been used by us and others to probe electric fields at the surface of electrodes 13;9;100 . Nitrile groups (and other Stark probes) 28 are useful in providing a picture of the local electrostatic environment, but specific interactions such as hydrogen bonding to the nitrile shift the CN frequency in ways that are not explained by a simple mean-field picture 8;24;101 . More broadly, heterogeneous polarization across the body of the probe complicates the use of Stark spectroscopy for electric field measurements. Electric fields can and do vary over molecular length scales 102 , but the probe only reports a single frequency reflective of an averaged field. Therefore, while the vibrational frequency of the nitrile probe is an impor- tant reporter of the local electrostatic environment, it does not imply homogeneity of field at the molecule scale. A molecular scale picture when interpreting vibrational frequencies as arising due to Stark shift is needed. For that reason, we used MD simulations to understand how the measured frequency shifts relate to a detailed picture of the ionic structure at the interface. Using MD sim- ulations, ILs of the imidazolium family have been studied at the interface of vacuum 103;104;105;106 and silica 107;108 . Kislenko et al. 109 studied the electrical double layer in[BMIM] + [PF 6 ] − IL at un- charged, positively charged, and negatively charged graphite surfaces. Recently,[BMIM] + [BF 4 ] − in a confined environment between two gold electrode surfaces has been studied 110 . The frequency shift of nitrile has been used to report the solvation field strength in the bulk of ILs, and the field strength was found to depend on the size of the anion, but little to no dependence on the size of the cations 56 . In this paper, we will refer to the fields reported by the nitrile probe as interfacial solvation fields. In the spirit of our earlier work, the nitrile probe is effectively solvated by the surrounding ionic environment and the metal which responds both to the probe molecule and the ions. The frequency of the probe responds to the total solvation environment. However, we must note that modifying the metal surface with a SAM inherently perturbs the environment that the probe reports on. The presence of a SAM on a metal can screen the potential drop away from the surface, polarize the metal interface, and modulate the hydrophobicity of the surface, among other effects 111;102;112;113 . This modification is an inherent trade-off when using SAMs to investigate structure at metal surfaces, but the fundamental and applied importance of metal interfaces - with or without modification - makes this trade-off more than worthwhile. IL structure have also been studied extensively by several groups, including those of Baldelli 114;115;116;117 , Dlott 118;119;120 , and Fayer 121;122;123 . Their work has revealed significant insight into the structure 29 and dynamics of ILs. This work is distinct from the mentioned efforts in that we measure the change in frequency of a Stark reporter, and not the IL itself. In doing so, we gain spatial speci- ficity, since we interact with the Stark probe only in one location, namely near the interface, as opposed to interacting with several layers of liquid, or the bulk of the liquid, all at once. The experimental concept of this study is outlined in Figure Figure 3.2. The spatial specificity of the Stark probe allows it to provide complementary information on surface structure relative to other commonly used surface techniques, including impedance spectroscopy. We have discussed the useful conjunction of vibrational Stark spectroscopy and EIS in a previous paper 124 . (b) (a) (c) Filling Syringe Metal scaffold 25μm Teflon spacer 100nm gold layer on Si Wafer 5 mm thick CaF₂ window Ionic Liquid Figure 3.2: Overview of experimental work. Panel Figure 3.2a: A diagram of the cell used to acquire SFG spectra. Panel Figure 3.2b: A cartoon depicting SFG generation from the 4-MBN SAM at the gold-IL interface. Panel Figure 3.2c: Representative SFG spectra of a 4-MBN SAM showing a broad non-resonant background the narrow CN stretch. 3.2 Experimental Methods Self-assembled monolayers (SAMs) of 4-Mercaptobenzonitrile (4-MBN) were prepared on silicon wafers with a 10 nm Ti adhesion layer and 100 nm of Au purchased from LGA Thin Films, Inc. Wafers were cleaned by sonication in ethanol twice, then in methanol twice for 8 minutes each time, then immersed in a 0.03 M solution of 4-MBN in ethanol overnight. This results in a dense monolayer with full surface coverage for a reproducable spectrum with a high signal to noise ratio. 125 . After soaking in the 4-MBN solution, the wafers were removed and again sonicated in ethanol and then methanol for 8 minutes each. 30 A 1 kHz regeneratively amplified Ti:Sapph laser (Coherent) was used to generate ultrafast near IR pulses. A portion (1 W) of this was directed to an optical delay stage followed by a 4f filter to significantly narrow the spectrum, while another portion (2 W) was directed to an OPA (Coherent OPerA Solo) equipped with a AgGaS 2 crystal for difference frequency generation of mid IR pulses. The 4f filter incorporates two volume phase gratings (BaySpec, Inc), two cylindrical lenses and a variable width slit to filter the near IR pulse to a spectral width of 8.0 cm − 1 , centered at 784.62 nm. Typical spectra of both the near IR upconversion pulses as well as the broadband mid IR pulses can be found in previous work 8 . Pulse energies were measured at the sample position to be≈ 8µJ for the near IR and≈ 7.56 µJ for the mid IR. VSFG spectra were acquired by focusing these two pulses together on the sample using a parabolic mirror and overlapping them in time. The resulting VSFG signal was collected with a second parabolic mirror and passed through a short pass filter to reject the majority of the scattered near-IR photons. The SFG was then sent to a spectrometer (Horiba iHR320) with a CCD camera (Sycerity, JY) for spectral analysis. With the input slit of the spectrometer set to 0.05 mm, and using an 1800 gr/mm grating, the theoretically achievable spectral resolution was 0.05 nm (about 1 cm − 1 in the spectral range of interest), which is well below the width of the near IR upconversion pulse. Spec- tral resolution of the SFG spectra are, thus, limited by the 8 cm − 1 width of the near IR upconversion pulses. We point out that peak shifts smaller than this value can be measured as has been discussed in a previous publication. 8 Vibrational Sum Frequency Generation (VSFG) spectra were taken before and after application of each IL to the gold wafer. Spectra were obtained from three acquisitions, each integrating for 180 seconds. Final raw spectra are a simple average of the three acquisitions. Experiments were conducted in a demountable liquid FTIR cell (International Crystal Laboratories) modified for this purpose (see Figure Figure 3.2). The back window of the cell was removed and replaced with the SAM containing wafer and a 25 µm Teflon spacer was placed directly on the sample surface. Ionic liquid was injected into the cell between the gold and a CaF 2 window. The entire assembly is then held firmly together using stainless steel plates and screws. All SFG measurements were taken in a purged environment, free of CO 2 and water. Raw SFG spectra is processed using a 31 fitting equation described by Benderskii et al. 126 , which is comprised of a resonant Lorentzian, a non-resonant Gaussian, and variable phase between the two signals. Temperature dependent SFG measurements were completed using a Lakeshore Model 325 Temperature controller and a Lakeshore DT-670 temperature sensor for temperature measurements. The normal SFG cell was slightly modified with a custom-machined aluminum base which allows for good temperature contact between the heater lead, cell environment and temperature probe. Final data reported from SFG experiments such as center frequencies and line widths are reported from an average of three spectra. Representative SFG spectra for all IL systems studied are reported in the SI (Figure Figure 5.5). SERS studies were carried out using a Horiba XploRA Raman Microscope System using a 532 nm fundamental beam. Spectra were taken using a 1800 groove/mm grating in 10 second incre- ments and averaged over six scans. SERS substrates were prepared with the following method, adapted from the literature 127 : A silver strip was sonicated in distilled water for 8 minutes, then submerged in 60% ammonium hydroxide solution for 1 minute, followed by inserting the silver strip in concentrated nitric acid for 10 seconds. The SERS substrates were then sonicated in water again for 8 minutes for a final cleaning before monolayer adsorption. Nitrile peaks from SERS studies were fit to Lorenzians and the error bars shown in the figure are from the 95% confidence interval of the fit. Each IL measurement was taken on a fresh piece of roughened silver and reported frequency shifts represent a subtraction between the ’neat’ 4-MBN monolayer and the CN stretch of 4-MBN after application of IL. Raw SERS spectra are reported in the SI (Figure Figure 5.6). To investigate the effect of anion size on the surface solvation environment, we used a se- ries of six ILs where the cation identity was fixed and the anion was changed. The cation used for this series was [EMIM] + . The anions used for this series are —in decreasing size order— tris(pentafluoroethyl)trifluorophosphate ( FAP − ), dimethylphosphate (DMP − ), ethyl sulfate (EtSO − 4 ), acetate (AcO − ), boron tetrafluoride ( BF − 4 ), and bromide (Br − ). Ionic liquids were purchased from Sigma Aldrich with purities higher than 98%. Structural information of the anion anion series used in this experiment are provided in Figure Figure 3.1. Ionic liquids were stored under moisture-free air and dried before measurement using a microwave purification method adapted from Ha et al 128 . 32 This method was shown to remove water to levels below 0.5 wt% rapidly and without damage to the ions. In short, aliquots of IL were heated in a lab microwave until IL temperatures reached 120°. We used a Nicolet iS50 FTIR Spectrometer to confirm that this treatment removes water and does not alter IL structure (associated figure is in the SI). The data show that a sample IL is not compromised or damaged by this treatment and that our storage and purification methods result in low water levels. 3.3 Computational Methods The 4-MBN functionalized slab of nanomaterial was generated in the following steps. First, we used CHARMM-GUI 129 nanomaterial modeler to build a 4 nm× 4 nm× 1 nm gold (100) surface with 100 % SCH 2 CH 3 ligand coverage (ligand density∼ 6.25nm 2 ). An initial energy minimiza- tion was performed using the Steepest Descent algorithm in the CHARMM 130 package. Then the system was translated, rotated and minimized again to obtain a 4 nm× 4 nm× 2 nm gold surface with ligands on both positive and negative Z directions. Next, the ethyl part of the ligands was replaced (patched 130 ) with 4-benzonitrile group to obtain the 4-MBN functionalized gold slab. Enegy minimization was performed again before packing the system with ILs. Detailed system dimensions and number of ions are included in SI Table Table 5.5. It is well known that consideration of electronic polarization is important for studying the struc- ture and dynamics of ILs. Electronic polarization is known to reduce enthalpy of vaporization, and accelerate the ion diffusion 131 . Studies have found that diffusion coefficients simulated using non- polarizable force fields are smaller than the experimental values 132;133;134 . Yan et al. 135 showed that for[EMIM] + [NO 3 ] − , introducing electronic polarization increases the diffusion coefficient to three times of non-polarizable model. They also observed that due to higher ion mobility, the shear viscosity calculated from the polarizable model was in better agreement with the experimental val- ues 135 . However, the effect of electronic polarization on IL structure is more subtle 131 , mainly in terms of anion-anion pair correlations 131;136;135 . Polarization is shown to relax long-range ion structuring in[BMIM] + [BF 4 ] − , and the influence propagates to short-range ion-ion correlation 136 . 33 The effect of polarization is known to be more pronounced for asymmetric ions 136 . Polarizable force fields are known to develop induced dipoles at the metal-IL interface, which soften the cor- relation between interfacial ions 137 . Nevertheless, a recent study by Ntim and Sulpizi 110 demon- strated that the density profiles and cation/anion orientation of IL ( [BMIM] + [BF 4 ] − ) was negligibly affected by gold polarization. In our simulations, the presence of 4-MBN monolayer on uncharged gold surface is expected to further minimize the effect of interfacial polarization. Moreover, non- polarizable force fields are shown to reproduce IL structure quite well 136;135;138;139;140 , due to the dominant effect of electrostatics. Since we focus on the trends in structural features across a series of ILs in this study, we use non-polarizable force fields due to their higher computational efficiency and broader range of availability for different ILs. Unless stated otherwise, non-polarizable CHARMM36 141;142 and CHARMM General Force Field 143;144;145;146 were used to describe the ligands and the ILs. Lennard-Jones parameters for Br – anions were taken from Canongia Lopes et al. 147 The structure of BF – 4 anion was optimized with B3LYP 148;149;150;151;152;153 and the aug-cc-pVDZ 154 basis set in the Gaussian 16 program 155 . The force field parameters for BF – 4 were taken from de Andrade et al. 156 . The authors develop the intra-molecular potential parameters using AMBER methodology 157 , and the Van der Waals parameters for fluorine and boron were sourced from AMBER 158 and DREIDING 159 force fields respectively. The [DMP] − anion has been modeled using the ligand modeler 160 in CHARMM- GUI 129 . Finally, the geometry of [FAP] − anion was optimized using the same DFT method and basis set as in BF – 4 . The force field parameters used for [FAP] − are developed by Shimizu et al. 161 based on the OPLS-AA molecular force field. In our implementation, the harmonic force constant for F-P-F, C-P-F, and C-P-C angles have been increased to 1000 KJ mol − 1 rad − 2 for additional rigidity around the phosphorus center. The diffusion coefficients calculated in this work (SI Table Table 5.6) are of the order of 10 − 11 m 2 s − 1 , consistent with previous experimental 162;163 and simulation studies 164;165;166;167 The computed densities are also in decent agreement with those reported in previous work (see SI Table Table 5.7). 34 For the gold surface, we use the INTERFACE force field 168;169 , which has been successfully applied to gold surface and gold nanoparticles. As mentioned above, this choice is further sup- ported by the recent study of Ntim and Sulpizi 110 , who demonstrated that the density profiles and cation/anion orientation of IL ([BMIM] + [BF 4 ] − ) was negligibly affected by gold polarization. The functionalized surface was packed with 550-1130 pairs of different IL cations and an- ions, and molecular dynamics simulations were performed using the GPU version of the GRO- MACS 170;171;172;173;174;175;176 2018 package. Position restraints of 200,000 KJ mol − 1 nm − 2 were applied on the gold atoms in all three dimensions. Periodic boundary conditions were employed in three dimensions as well. After energy minimization, the systems were equilibrated for 100 ps in the NVT ensemble (with 0.5 fs timestep), and 200 ns in the NPT ensemble (with 2 fs timestep). 200 ns of production run was performed thereafter, with a timestep of 2 fs. The particle-mesh-Ewald 177 method with a Fourier spacing of 0.12 nm was used to calculate electrostatic interactions. Real space non-bonded interactions were treated with a cut-off distance of 1.2 nm and a force-switch modifier. LINCS 178;179 algorithm was used to constrain all bonds involving hydrogen atoms. We used the Berendsen 180 thermostat with a time constant of 1 ps and a target temperature of 400 K for equilibration NPT runs, and the Nos´ e-Hoover 181;182 thermostat with same parameters for production runs. Semi-isotropic pressure coupling was applied for all NPT simulations with a target pressure of 1.0 atm. The Berendsen 180 pressure-coupling scheme with a time constant of 5.0 ps and compressibility of 4.5× 10 − 5 bar − 1 was used for equilibration, while we used the Parrinello-Rahman 183;184 method with a time constant of 10.0 ps and compressibility of 4.5× 10 − 5 bar − 1 for production simulations. To probe local electrostatics at the interface, we calculate the electric field at the nitrile nitrogen of the 4-MBN probe. The electrostatic field at position r due to point charges q i at positions r i is given by: E(r)= n ∑ i=1 q i (r− r i ) |r− r i | 3 (3.1) 35 In this study, an atom-based cut off distance of 3.5 nm is used for field calculation. The choice of cut-off distance is based on the convergence of averaged electric field gradients to 3.6× 10 − 2 V nm − 2 (SI Figure Figure 5.13, Figure 5.14, Figure 5.15). We calculate the electric field exclusively from ILs, and it’s projection along C-N axis for all 200 nitrile nitrogens. The electric field is sampled every 50 ps for the last 100 ns of production run. (Histograms of electric field components and projection are shown in SI Figures Figure 5.16, Figure 5.17). At that timescale, the components and the projection of electric field are decorrelated (SI Figure Figure 5.18). The component of the electric field on nitrile nitrogens along C-N is averaged over space and time, and is used next to calculate the estimated Stark frequency shift. Assuming a linear Stark tuning rate of ∆⃗ µ = 0.36cm − 1 (MV/cm) − 1 for benzonitriles 9;10;185;186;187;188;189 , we report the estimated Stark frequency shift∆ν CN (SI Figure 3.5a, 3.5b). The volume of anions was estimated using the quantum chemistry package QChem 5.1 74 . Anion structures were optimized using the B3LYP functional and the 6-31G* basis set. Following optimization, the volume corresponding to 99% of the anion’s electron density was extracted from the cube file associated with the final structure. The volumes estimated by this method match those published in work by others 56 . 3.4 Results We first present experimental results showing the dependence of the probe vibrational frequency on the anion charge density, followed by computational results confirming the experimental trend. Figure Figure 3.3a shows the extracted Lorenzian fits to the nitrile SFG spectra, with the dotted line representing the nitrile frequency of the monolayer in contact with air (see SI for details). All center frequency shifts with respect to air are plotted in Figure Figure 3.3b. Our main result is the observation of a systematic shift in the central frequency with decreasing anion size (i.e. increasing charge density). Over the range of ILs studied, we observe the smallest nitrile shift (∼ 2 cm − 1 ) from the IL with the largest anion ([EMIM] + [FAP] − ), and the largest shift (∼ 12.5 cm − 1 ) from the IL with the smallest anion ([EMIM] + [Br] − ). Under the assumption of a linear Stark tuning rate 36 (discussed in our previous work) 13 , this corresponds to a considerable interfacial solvation field of∼ 3.6 V/nm. The reason for this behavior could only be explained after our MD simulations, revealing a structure arising from a balance between size, ion packing and electrostatics near the surface, as will be explained in the discussion section below. Because frequencies extracted from SFG are inherently convoluted with a non-resonant back- ground and are highly dependent on reliable fitting 190 , we supplemented the SFG measurements with surface-enhanced Raman spectroscopy (SERS). Though the roughened surfaces required for SERS introduce additional complexity 191;192 , it can still serve as a useful comparison and allevi- ates concerns regarding phase-amplitude mixing in recovering the SFG central frequencies. Our measured SERS results (shown in Figure Figure 3.4) are generally in agreement with the SFG re- sults and the overall trend with respect to anion size is consistent between the two. We speculate that the differences between the SERS and SFG central frequencies beyond the estimated∼ 1cm − 1 experimental error may arise from the nanostructured hot spots in silver which gives rise to signal enhancement. Such nanostructuring may have a small influence on ionic structure. The qualita- tive agreement between these two experimental techniques indicates that ∆ν CN values extracted from SFG are not significantly impacted by fitting and background errors, thereby confirming the observed trend in frequencies with respect to anionic charge density. As an additional experimental check, the temperature dependence of the benzonitrile mono- layer’s CN stretch in ([EMIM] + [Br] − ) was measured using VSFG. Varying the temperature from 295K to 380K did not change the nitrile center frequency greater than an approximate bounds of ± 0.5 cm − 1 . The results of this experiment are shown in Figure Figure 5.8. This confirms that the ionic structures studied, at least in the range of temperatures studied, are not in a metastable configuration and represent an equilibrium arrangement. The computational results comprise of two parts. First, the calculated frequency shift of the probe molecule in the presence of an equilibrated configurational ensemble of ILs, and second, analysis of the arrangement of ions as a function of distance from the surface. Following procedures explained in the computational methods section, the calculated fre- quency shifts based on the nitrile Stark response and the field at the nitrile nitrogen are plotted 37 (a) Smaller Anions (b) Figure 3.3: Room temperature frequency shifts of 4-MBN monolayer on the electrode surface in the presence of different ILs. Figure Figure 3.3a shows extracted Lorenzians from raw SFG spectra, and Figure Figure 3.3a shows the center nitrile frequency plotted against the charge density of the anion. The monolayer is strongly solvated in the presence of smaller anions, with a large field of ∼ 3.6 V/nm observed using[EMIM] + [Br] − . 38 Figure 3.4: Surface enhanced Raman (SERS) data of the nitrile center frequency in the presence of different ILs. Chemically etched silver was used as a substrate. We observe an increase in polarization at the interface correlated with smaller anion size, in agreement with the SFG results. Point to point differences between SERS and SFG measurements may be related to heterogeneity at the SERS surface. 39 (a) (b) Figure 3.5: Calculated frequency shifts at nitrile carbon of 4-MBN monolayer on the gold slab in presence of different ILs. (a) Realistic anions with[EMIM] + cations, and (b) Modified Cl – anions with[EMIM] + cations. Lennard-Jonesσ for Cl − = 4.04 ˚ A. against the charge density of anions in Figure 3.5a. Consistent with experimental results, a red shift with respect to increasing charge density is observed. Frequency shifts based on electric field values at the nitrile carbon atom are included in SI Figure 5.10a and show a similar trend. While the trend in experimental data is reproduced computationally, the magnitude of the computed fre- quency shifts is larger than experimentally observed ones. The possible origins of this will be discussed in the discussion section. One may argue that the observed trend is not necessarily a consequence of ionic size, but rather majorly their structural and chemical details. To gain insight into this issue, we constructed a sim- plified model, in which only the size of the anion was varied, without affecting their shape. Model systems were constructed by starting from a Cl – anion and modifying its Lennard-Jonesσ param- eter, while keeping all other force field parameters the same. This gives rise to various Cl – anions with artificially enlarged volumes. We use 1.5, 2, and 3 times the original value of σ (4.04 ˚ A) to construct the IL systems, named [EMIM] + [Cl 1.5σ ] − , [EMIM] + [Cl 2σ ] − , and [EMIM] + [Cl 3σ ] − respectively. The effective radii of the anions are, therefore scaled by a factor of 1.5, 2, and 3 while maintaining the spherical shape. MD simulations are performed for this anion series, while the cation is still kept to be [EMIM] + . The frequency shifts of the probe molecule, based on field at 40 the nitrogen of nitrile, are plotted against the relative charge density of this chloride series, calcu- lated using the ratio of ionic radii (Figure 3.5b). The figure shows that the trend with respect to size indeed holds when only the size of the anion is changed. The origin of this change with respect to ion size will be discussed in discussion section. Frequency shifts for the chloride series calculated based on the field on the carbon atoms of nitrile are included in the SI (Figure 5.10b) and show a similar behavior. Figure 3.6: Snapshots of representative systems after 200 ns of production MD. Only top half of the functionalized gold slab has been shown. For realistic anions, full IL residues have been shown for atoms within 4 ˚ A of nitrile nitrogens. For[EMIM] + [Cl 2σ ] − and[EMIM] + [Cl 3σ ] − , the selection thresholds are 6 ˚ A and 8 ˚ A respectively. 41 Figure Figure 3.6 shows snapshots of representative systems at the end of production runs. Inspection of these snapshots for the experimental anion series (left panel) and the model chloride series (right panel) reveal the relative positions of cations and anions at the interface. We note that for anion series used in the experiments, the closest layer to the 4-MBN ligands is predominantly made of anions. Moreover, the anions show different degrees of intercalation into the 4-MBN layer. Smaller anions such as Br – and Cl – fully insert themselves into the 4-MBN layer and lie in the same plane as the nitrile nitrogens. Larger[BF 4 ] − anions show partial insertions, and even larger [FAP] − anions cannot intercalate and are excluded from the monolayer. The density of anions at the interface is also less for larger anions, as expected. A similar behavior is observed for the model chloride series (Figure Figure 3.6, right panel). The normal size chloride intercalates into the monolayer, while the largest model chloride is excluded from the surface. The ionic density near the surface is also the smallest for the largest of the chloride series anion. To rationalize the frequency shift-size trends, we calculated the symmetrized number density of a few representative atoms (see Scheme Scheme 3.1 for special atom names). Figure Figure 3.7 shows the distribution of these atoms along the box’s positive z-axis, centered around the gold slab. Symmetrized number densities and charge densities of all realistic and model systems are included in the SI (Figure 5.11a,5.11b,5.12a,5.12b). We should note that the density profiles are of unique atoms in IL residues, they are not cumulative of any atom type (such as F P in[FAP] − ). The position distributions of nitrile nitrogen atoms have been shaded green for reference. The overlap of the black lines (representing key anionic atoms such as Br, F, F P , and Cl) with green region depicts the extent of anionic intercalation into the 4-MBN layer. Number density of three key imidazole hydrogens (see Scheme Scheme 3.1) from[EMIM] + cations have also been shown in the figure. The hydrogen with highest partial charge (H3) is shown in red; whereas the other two (H4, H5) have been shown in orange. Higher density of one type of hydrogen over others, as seen in [EMIM] + [Br] − , [EMIM] + [Cl] + , and [EMIM] + [BF 4 ] − indicate preferential orientation of cations near the interface. 42 3.5 Discussion The main objective of this study is to understand the organization of ILs near the interface by measuring the frequency shift of the probe molecule adsorbed on the metal. Our results show a systematic change in the vibrational frequency of the probe with increasing anion size. However, this information alone cannot provide a complete molecular picture of the local ionic structure, since ordering of ions of varying sizes is complex and not uniquely associated with a single value of frequency shift. Therefore, we complemented the experimental work with MD simulations with two purposes - first to find out whether the experimental trends were reproduced by the simulations, and second to identify the underlying structural origins of the observed trends. The primary takeaway from these combined efforts is that IL molecular structure at the inter- face is dictated by the ability of ions to pack and organize near the surface. The most important result from MD simulations is the tendency for the anions to partially intercalate within the SAM. The smallest anion (bromide) readily fits between the 4-MBN molecules and intercalates into the monolayer, residing at the same depth as the nitrogen atoms of the nitrile (Figure Figure 3.7). This insertion, in turn, supports a high packing density of EMIM + in the next layer. The large interfacial charge density strongly polarizes the SAM, resulting in a considerable shift in the nitrile frequency. However, as the size of the anions in the IL is increased, their ability to insert within the SAM is diminished. Furthermore, the large anion sizes exclude some volume near the surface and push the cation-dominated layer away from the SAM. This results in low density of the cations and larger distance between the cations and the monolayer. Therefore, a smaller field is experienced by the probe, leading to smaller frequency shifts. The above scenario even holds to explain the slightly out-of-order behavior of acetate in[EMIM] + [AcO] − , which is observed experimentally (Figures Figure 3.3,Figure 3.4) and confirmed computationally (Figure 5.11a,5.11b). In both cases, the acetate anion causes a frequency shift in the probe that is larger than anticipated from its effective size, appearing in both the experimental and compu- tational results as a slight non-monotonic deviation from the trend. In this case, the molecular shape of the acetate together with the intercalation description given above explains this effect. 43 The charged portion of the acetate ion (delocalized mainly over the oxygens) is relatively com- pact and available for insertion in the monolayer. Similar to the behavior of [EMIM] + [Br] − or [EMIM] + [Cl] − , this insertion supports the formation of a high-density layer of[EMIM] + cations near the SAM and the resulting structure polarizes the nitrile more than anticipated based on the total net size of the acetate. The use of the chloride model system is a way to investigate the fundamental dependence of interfacial solvation on anion size without the confounding effects of molecular structure. The results from these simulations show a similar overall trend but the calculated fields depend mono- tonically on anion size, thereby confirming the hypothesis that insertion and packing is at the core of this behavior. We observe the same ion intercalation structure in the model chloride system as the atomistic anions (Figure Figure 3.6 and Figure 3.7). The center of anionic charge is pushed away from the interface as the anion size increases. However, at the limit of chlorides with large values ofσ, the first ionic layer becomes predominantly cationic, which we do not observe for ILs like[EMIM] + [FAP] − . In both cases, the packing density and the induced field are small, therefore confirming that packing density may be a stronger factor in dictating the interfacial field compared to solely relying on the ordering of the layers. Scheme 3.1: Atom names used in simulations for[EMIM] + (left) and[FAP] − (right). The hydro- gen at 3 rd position of imidazole ring in[EMIM] + is named H3. The other two ring hydrogens are named H4 and H5. In [FAP] − molecule, fluorine atoms bonded to phosphorus center are named F P . We note that some parts of the larger anions (e.g. the fluorines in [FAP] − ) also penetrate the 4-MBN layer to a similar extent (Figure Figure 3.7) as Br − . However, a second fluorine peak is observed for both [EMIM] + [BF 4 ] − and [EMIM] + [FAP] − oustide the monolayer, indicating that 44 large anions are only partially interecalated in the monolayer and mainly reside outside. This behavior is also observed for[EMIM] + [DMP] − as seen in the SI. Due to this effect, and the larger sizes of these anions, the overall charge density near the surface is much smaller compared to [EMIM] + [Br] − , thereby producing a smaller field experienced by the probe molecule. Though the overall trend between interfacial fields and anion size holds between theory and experiment, the computed fields exceed the measured fields for every IL we studied. This dis- crepancy is the greatest for larger anions (factor of∼ 7 for [FAP] − ). We hypothesize that this is a limitation of the fixed-charge non-polarizable nature of our force field. Inclusion of polarizable force fields will modify metal-ion and ion-ion interactions and may result in an overall decrease in the calculated interfacial fields. Furthermore, the location at which field values are calculated is chosen at a single point within the probe molecule. Our previous work shows that fields near inter- faces are not uniform, and vary across the length of the probe molecule 124 . Therefore, choosing a single point within the molecule only approximately emulates the homogeneous field Stark effect. For the purposes of this study, which is identifying the size effects trends, this does not pose any problems. A comment related to the water content of ILs is necessary for our work. As many previous work have shown, ILs are hygroscopic and water adsorbed from the atmosphere alters their prop- erties in important ways. The partition of water between interface and bulk in ILs probed by a 4-MBN SAM was reported by us recently 15 . Based on that work, even though ILs adsorb water readily, if the quantity of the water is small it is favorably solvated within the bulk and it hardly appears at the interface. A significantly large critical threshold (above ∼ 0.8 mole fraction) must be reached before water has a significant partition at the surface. As discussed in the experimental section, we have taken steps to dry the ILs studied in this work and therefore the water content should be minimal. Based on the results from our previous study, the measurements reported in this work are representative of water-free ILs. Further characterization of IL-metal interfaces using this joint experimental-computational ap- proach could include the measurement and simulation of charged interfaces along a similar series of ion structure. We have used VSFG to probe the electrified metal-IL interface, 13 but that previous 45 Figure 3.7: Symmetrized partial number density of representative atoms plotted against average relative position from the center of gold layer. As a reference, the positions of nitrile nitrogen (N CN ) atoms of 4-MBN monolayer are shown in green. [EMIM] + hydrogens from imidazole ring are labelled as H3 (H atom at 3 rd position of imidazole ring), H4 and H5. All four F atoms of [BF 4 ] − anions are labelled F. The F atoms directly attached to phosphorus center in [FAP] − are labelled F P . The atom names for larger ions are illustrated in Scheme Scheme 3.1. 46 work did not cover systematic variation of ion composition. Modifying particular chemical proper- ties of ILs (such as introducing hydrophobic moieties) will also provide fundamental information about ionic structure at the interface. 3.6 Concluding remarks In this study, we take a detailed look at the structure and electrostatics of ILs near a metal interface functionalized by a layer of probe molecules. The main takeaway from our work is that the local electric field sensed by the probes varies significantly with the size of the anions in the imidazolium family of ILs, with larger anions producing smaller interfacial fields. The origin of this effect was revealed by our molecular dynamics simulations. We observe that small anions intercalate into the nitrile probe layer which helps with tighter packing of the nearby cation layer and results in large electric fields. As the size of the anions increase, the extent of surface penetration diminishes, leading to disappearance of the ordered structure and looser packing of ions near the surface. We also emphasize that not only size, but also the shape of the anions is important in dictating the local electric field. Finally, the trends in calculated Stark frequency shifts qualitatively agree with experiments. This work is a stepping stone towards understanding and modifying interfacial fields in the presence of complex ionic environments. 47 Chapter 4 Can Bronsted Photobases act as Lewis Photobases? Quinolines Enhance Binding Affinity to BF 3 in the Excited State 4.1 Introduction Light induced proton transfer is a well studied excited state phenomenon. This is often made possible by photoacids and photobases, which are molecules that undergo a pKa change upon absorption of a photon. For the strongest photoacids and photobases, the excited state pKa (often denoted as pKa*) can vary from the ground state by more than 10 units 193;2 . This photochemical handle not only allows for fundamental study of proton transfer at ultrafast timescales, but also holds promise for controlling proton-requiring processes such as acid-catalyzed reactions with light 194;195 , for creating a net photovoltage in membranes embedded with photoacids 196;197 , and modulating ionic conductivity in polymers 17 . Studies of photoacids and photobases to date have mostly focused on control of Brønsted (pro- tonic) acidity and basicity with light. Photo-enhanced acidity and basicity can either be reversible, where the acid or base returns to its original form following electronic relaxation 2 , or irreversible, where the photoexcited species generates an acid or base in the excited state and does not return to its original form. The subject of this paper is reversible photobasicity, in which the photoactive molecule can repeat the photocycle after relaxation to the ground state. It is distinct from pho- toacid/base generators that are commonly used and involve a light-triggered irreversible release of an acid/base 198;199 . The underlying mechanism of reversible photobases is a build up of charge 48 density on the basic site, which lowers the energy of the conjugate acid 2 . To further illustrate this point, we have provided a diagram of the excited state charge relocalization for a representative quinoline in the SI (Figure Figure 5.20). In this work, we will lay the foundations of extending the concept of photobasicity to Lewis interactions, thereby increasing the scope of applications where light can be used to drive Lewis catalyzed reactions. The concepts of Brønsted acidity/basicity were generalized by Lewis, who defined electron pair acceptors and donors as Lewis acids and bases respectively. A classic example of Lewis acid/base interaction in the main group elements is the dative bond formed in the NH 3 - BH 3 adduct. The formation of this bond is highly favorable due to the interaction between boron’s empty P orbitals and ammonia’s lone pair, and stronger electron donors produce a more stable adduct. As described in the previous paragraph, quinolines build up charge density on nitrogen in the excited state and become stronger electron donors. Knowing this, we hypothesized that this build-up of charge would correspond to photobasicity for a general class of acids beyond protons. Our main goal in this work is to present experimental and computational evidence in support of this idea. Lewis’ framework describes a large number of chemical interactions. In organometallic chem- istry, often the metal ions are Lewis acids, which interact with organic ligands which act as Lewis bases 200;201 . Strong electron acceptors like BF 3 and AlCl 3 find wide use as catalysts in organic synthesis due to their stabilizing effect on intermediates 202 . The concept of Frustrated Lewis Pairs (FLPs), or an unfulfilled Lewis bond due to steric hinderance, has gained attention as a potent catalyst for activation of small molecules 203 . Furthermore, a large number of surface interac- tions, and electrochemical phenomena depend on adsorption/desorption that are based on Lewis bonds 204;205;206 . A previous study on the electronic spectra of the quinoline-BCl 3 complex noted that the quinolinium ion and the dative-bonded quinoline have remarkably similar spectral fea- tures 207 . Given the wide relevance of Lewis bonds to many areas of chemistry, understanding their behavior in the excited states is of both fundamental and practical value. If a Lewis interaction could be turned on or off by light, or if a Lewis acid could be reversibly released with light for the 49 purpose of catalyzing a reaction, it could be of use in numerous applications. A growing number of authors are recognizing the potential usefulness of light-controlled dative bonding 208;209;210;211 , but no explicit formalism has developed around these ideas. Prior work on light-responsive Lewis pairs has largely focused on weakening dative bonds with light, not enhancing the bond strength. Charge-transfer excitations in B-N bonds have been extensively studied in the BODIPY family of dyes 212 . The photophysics of B-N bond containing systems more generally is also a field of interest 213;214 .This paper aims to broaden these pre-existing ideas to a wider range of chemical applications by connecting to concepts in the photoacid/photobase literature. In this work, we will show experimental and computational results to understand the founda- tions of Lewis basicity in the excited state, and describe the similarities and differences between the excited Lewis and Brønsted counterparts. Developing control of Lewis bonding through the use of photoacids and photobases has wide implications for catalysis and may open up the study of transient species in the transfer of a dative bond from one adduct to another. 4.2 Concepts This section will provide an overview of the F¨ orster cycle and its use in studying photoacids and photobases 215 , followed by a discussion on applying the F¨ orster cycle in the specific context of excited state Lewis acidity and basicity. The F¨ orster cycle is a thermodynamic cycle most com- monly used to estimate the excited state pKa* from the absorption, emission, and ground state pKa values. Just as expected from any thermodynamic cycle, the predictions of F¨ orster cycle does not guarantee kinetic feasibility of a process. Separate experiments are needed to establish that a ther- modynamically possible proton transfer process is also kinetically feasible. Previous work by us and others have demonstrated that quinolines are indeed kinetically competent for proton capture from a variety of proton donors 2;216;19;20;22 . The idea of F¨ orster cycle analysis is, of course, general and has been employed to explain other excited state processes before, including for metal-ligand interactions 217;218 . 50 Interaction of the lone-pair of the heterocyclic nitrogen of quinoline with BF3 bears resem- blance to its protonation. Therefore, one can apply the F¨ orster cycle to the quinoline-BF 3 as depicted in Figure Figure 4.1. In direct analogy to excited state proton capture, the drive for the photoexcited quinoline to capture a BF3 molecule from a donor can be inferred from this cycle. Experimental and computational determination of this excited state drive is the main focus of our work. Figure 4.1: A diagram of the F¨ orster cycle applied to the quinoline-BF 3 adduct. The excited state drive∆G ES is inferred from spectroscopic estimates of the S 0 -S 1 gaps for the adducted and unadducted forms, and the ground state energy for adduct formation. 4.3 Experimental and Computational Methods To determine the excited state drive for adduct formation, it is necessary to measure the absorption and emission spectra of the adducted and unadducted forms. It is necessary that the solvent is not strongly Lewis basic to cause significant competition with photoexcited quinoline. The solvent 51 also needs to be dry since water is a proton donor and a strong Lewis base - meaning it can in- teract with both the Lewis acid BF 3 and the photobasic quinoline. All solvents were dried for at least 24 hours using 3 ˚ A molecular sieves (>10% m/v) prior to their use. This procedure was re- ported to lower the water content in acetonitrile to sub-ppm levels 219 , which is far below the water sensitivity threshold for a pyridine-BF 3 complex, as reported previously in a similar system 220 . Additionally, all solution preparation was performed in a glovebox purged with dry air. Measures taken to minimize the exposure of our quinoline-BF 3 complexes to water give us confidence that any remaining trace moisture will not impact our results and conclusions. UV-Vis spectra were taken using a Cary 50 UV-vis spectrophotometer. Emission data was acquired using a Jobin-Yvon Fluoromax 3 fluorometer. Acetonitrile solutions of 5-R-quinolines were prepared at 4 x 10 − 5 M for all absorption and emission experiments, with the exception of the unadducted fluorescence of quinoline, 5-bromoquinoline and 5-cyanoquinoline. The reason for this choice is that quinolines with electron donating substituents are known to have easily measur- able emission 21 , but unadducted quinoline and quinolines with electron withdrawing substituents emit quite poorly in aprotic (particularly non-polar) media 221;222 . Emission spectra for these com- pounds were taken using quinoline concentrations of 4 x 10 − 3 M. We used electronic structure calculations to further characterize the Lewis photobasicity of 5- R-quinolines. Our approach is similar to the one used by the Petit group studying proton transfer energetics of a series of photobases 223;25 with some small modifications made to suit the study of dative bond complexes. Their computational results were shown to correlate quite well with experimental values for the excited state drive for proton transfer in quinoline photobases. 223 We will outline our procedure here and refer the reader to Petit’s prior work for a more detailed dis- cussion. Paralleling the experimental work, a computational F¨ orster cycle requires the calculation of three energies: the 0-0 gap between the ground state and the photobasic electronic state of the unadducted quinoline (∆E Free 00 ), the 0-0 gap between the ground state and the photobasic electronic state of the adducted quinoline (∆E Adduct 00 ), and the ground state energy for adduct dissociation (∆E GS ). These three energies determine∆E ES which is the excited state energy for adduct forma- tion through the equation∆E Free 00 =∆E Adduct 00 +∆E GS +∆E ES . All calculations were performed at 52 theωB97X-D/def2-svpd level of theory. This method was used by Petit et al. 223 because it is well- suited for the calculation of energies in charge transfer complexes 224 . TD-DFT calculations were performed using the Tamm-Dancoff approximation, and as the photobasic states of quiniolines are typically singlets, triplet states were excluded from the calculation. The calculations displayed in the main text were performed in the gas phase, but we performed additional calculations to under- stand the method and solvent dependence of our results. We repeated the single point calculations at the optimized geometries using a larger basis set along with inclusion of polarizable continuum (see SI, Figure Figure 5.21). The trend with respect to the Hammett parameter did not change at this level of theory. The two energy differences, ∆E Free 00 and ∆E Adduct 00 , were obtained by performing ground and excited-state geometry optimizations of quinoline for the free (unadducted) structure and for the BF 3 adducted quinoline. To compute the ground state energy for adduct formation, ∆E GS , we used the energy decomposition routine (EDA2) implemented in the QChem 5 package 225 , which accounts for the basis set superposition error and decomposes the total energy for adduct into separate, physically insightful quantities. The use of energy decomposition methods is common in the study of Lewis acid-base complexes 226;227;228 . Details of the energy decomposition results are provided in the Supporting Information. The formation of a BF 3 adduct also involves a significant change in the BF 3 structure. This rehybridization/strain energy is not accounted for in QChem’s default EDA method, which computes bond formation energies in a “vertical” fashion 1 . Therefore, the isolated BF 3 and quinoline fragments were re-optimized from the adducted structures and the difference between these strained and relaxed geometries were subtracted from the EDA output to find the total complexation energy. To understand the charge migration in the excited state of the adduct, we calculated the electron density difference (EDD) maps. The EDD plot of the 5-MeOQBF 3 adduct was generated using a cube file resulting from TDDFT calculations performed in QChem (using ωB97X-D/def2-svpd). The cube file was loaded into VMD 229 to generate positive and negative isosurfaces, and those images were then rendered using POVRay 230 . 53 Changes in proton affinity for photoacids and photobases upon light absorption are typically reported as∆pKa, which related to energy differences via pK a =− ∆G/2.3k B T . We express the excited state drive for Lewis basicity using both of these conventions. For the experimental data, we chose to set the ground-state∆G in the F¨ orster cycle to zero. Therefore, the energies reported correspond to a quinoline’s ∆∆G for binding BF 3 in the excited state. We had more freedom in determining BF 3 binding energies for the computational work, and we decided to reference these energies to a strong Lewis base, triethylamine (TEA). Functionally, this means that we set the energy of the TEA-BF 3 complex as zero and reference all other energies to that value. In analogy to the Brønsted pKa we use pK BF 3 = -∆G BF 3 /2.3k B T as a scale in Figure Figure 4.5 in addition to expressing it in energy scale. 4.4 Results and Discussion We begin by showing representative absorption and emission spectra of one of the quinolines (5- methoxyquinoline) in free and adducted forms (figure Figure 4.2)a in acetonitrile solvent. We have measured the experimental Stokes shifts in 5-MeoQ for both the adducted and unadducted species to be∼ 1 eV , while the computed Stokes shifts for both species are somewhat smaller in magnitude, ∼ 0.6 eV . Corresponding spectra for the other quinolines used in this study are provided in the SI. The lowest-lying electronic states of quinolines are labelled following Platt’s notation 231;2 , with the L A and L B corresponding to atom-centered and bond-centered excitations, respectively. The L A excitation (or the photobasic excitation) has a broad lineshape and its energy is very sensitive to hydrogen bonding and protonation as reported previously 2 . As seen in Figure Figure 4.2a, a similar effect applies when BF 3 bonds to 5-MeOQ to form a Lewis adduct. Relative to free 5- MeOQ both the absorption and emission maxima for 5-MeOQBF 3 redshift by ∼ 100 nm. As discussed previously, interpreting these measurements in the context of the F¨ orster cycle leads us to conclude that 5-MeOQ enhances its binding affinity towards BF 3 in the excited state. In order to easily compare the experimental and computational results for the electronic spectra of 5-MeOQ and its BF 3 adduct, we generated Figure Figure 4.2b from the TDDFT stick spectra. 54 (a) (b) Figure 4.2: (a) Absorption and emission spectra for free and BF 3 -adducted 5-methoxyquinoline. The large redshift for the absorption and emission between the adducted and unadducted forms indicates a substantial enhancement for 5-MeOQ to bind BF 3 in the excited state. (b) Calculated absorption and emission spectra (S 0 - S 1 states) for free and BF 3 -adducted 5-methoxyquinoline show similar red shifts. 55 The Gaussian lineshapes are drawn to visually correspond to the experimental linewidths. The first takeaway from this data is that TDDFT is able to corroborate the Lewis photobasic effect seen in the experimental spectra, as evidenced by the redshifted absorption and emission lines in the adduct structure compared to free 5-MeOQ. Though the relative order of the bands is common between experiment and theory, the computational lines are all blue shifted with respect to experimental results. In order to rationalize and understand this data in an intuitive way, we generated an elec- tron density difference (EDD) map for the ground and first excited states for the 5-MeOQBF 3 adduct. EDD maps are useful in analyzing charge transfer-like excitations because they represent the motion of electron density between the ground and excited states. Figure Figure 4.3 displays a schematic of the 5-MeOQBF 3 adduct and the net build-up of electrons in the B-N bond upon excitation. The EDD plot is shown in Figure Figure 4.3b, where the red (blue) isosurfaces repre- sent regions of increased (decreased) charge density in the S 1 state. The EDD plot shows that the photobasic state (or LA state) is primarily ofπ-π ∗ character, and that charge density tends to build up on the atomic centers, including nitrogen. This is in line with work previously published by our group 2 . Near the B-N bond, the electron density becomes markedlyσ-like and is centered around the dative bond. We have denoted the B-N bond in Figure Figure 4.3 as a guide, and a side view of this EDD plot is included in the SI for additional visualization. This data shows that the increased charge density on nitrogen in the photobasic state of 5-methoxyquinoline readily participates in a dative bond with BF 3 . Next we discuss the sensitivity of the Lewis photobasicity to the Hammett parameter (electron- withdrawing/donating strength) of a functional group at the 5 position of quinoline. The purpose is to undestand whether the data and the proposed mechanism for photobasicity reasonably follows the Hammett parameter, as it has been shown for Brosted photobasicity 2 . For all the quinolines studied in this work, we observe a large change in the L A energy upon addition of BF 3 , where the narrow bands corresponding to the L B states change minimally. For all quinolines except 5-aminoquinoline, we observe a redshift in the absorption and emission spectra for the adducted species. These spectral changes draw a strong analogy to the response of quinolines to protonation, 56 B-N Bond (a) (b) Figure 4.3: (a) A schematic of the 5-MeOQBF 3 adduct and the flow of electrons to the B-N bond in the S 1 state. (b) An electron density difference plot corresponding to the S 0 -S 1 excitation for the 5-MeOQBF 3 adduct. The plot shows that the excitation builds up electronic charge density on nitrogen, which strengthens its bond with BF 3 . meaning that the spectra indicate that the photobasic effect in quinolines applies to both protons and BF 3 . As shown in the SI (Figure Figure 5.22), 5-aminoquinoline’s absorption spectra blue shifted upon addition of BF 3 . This observation is consistent with BF 3 binding to the amino group, as supported by the results of DFT calculations provided in the SI. Based on these observations, tuning the photobasicity of quinolines with groups that involve competing strong Lewis bases is not recommended. Because we are only interested in studying BF 3 binding to the heterocyclic nitrogen, we have excluded the experimental data of 5-aminoquinoline from F¨ orster cycle analysis. However, we retain 5-aminoquinoline in the computational section, where we can control over where the BF 3 binds. Because no other quinoline showed a blue shift upon addition of BF 3 , we are confident that the rest of our data reflects binding between the heterocyclic nitrogen and BF 3 , as intended. Figure Figure 4.4 shows the difference between the 0-0 S 0 -S 1 energies between free and ad- ducted quinolines based on experimental data and plotted against the Hammett parameter σ P for the substituent at the 5 position. This∆E is proportional to the energetic relaxation for a quinoline 57 OCH 3 CH 3 H Br CN Figure 4.4: The excited state affinity change for adduct formation expressed both in terms of en- ergy and equilibrium constant change, plotted against the electron withdrawing strength (Hammett Parameter) of the substituent at the 5 position. The data show that the measured affinity change strongly depends on the substituent. 58 upon binding with BF 3 in the excited state. Similar to our group’s prior work on the Brønsted pho- tobasicity of quinoline 2 , all quinolines studied in this work develop an increased thermodynamic drive to bind BF 3 in the excited state as evidenced by the experimentally determined F¨ orster cycle. We find that the excited state binding affinities for quinolines increase with the electron donating strength of the substituent, and that electron withdrawing groups reduce the strength of this ef- fect. An exception is unsubstituted quinoline (σ P = 0), whose data point is notably off the linear trend. Again, our prior work on Brønsted photobasicity also found unsubstituted quinoline to be somewhat of an outlier 2 . In that work, we hypothesized that the smaller polarizability volume of unsubstituted quinoline may limit the charge transfer character of the photobasic state, which is not accounted for by the Hammett parameter alone. The computational F¨ orster cycle mirrors the experimental data, as presented in Figure Fig- ure 4.5. In this figure, the energy of the TEA-BF 3 adduct as determined by fragment analysis is taken as zero. The photobasic effect is expressed both in terms of energy and changes in equilib- rium constants. Over the range of substituents used in the calculations, we see a clear dependence in both the ground state and excited state BF 3 affinity. The linear relation between the ground state BF 3 and the Hammett parameter of the substituent is to be expected from a simple electron do- nating/withdrawing argument, and from published data on BF 3 dative bonds 226 . The excited state quinolines all increase their BF 3 affinity relative to the ground state, and quinolines with strong electron donating groups are calculated to increase their BF 3 binding constants by∼ 10 orders of magnitude. Initial calculations on basis set and solvent dependence (shown in Figure Figure 5.21) suggest that the photobasic effect in quinolines towards BF 3 is not method dependent. The Lewis adducts have a slight charge transfer and as a result we should expect them to stabilize in higher dielectric solvents. As we discuss in the SI, a more systematic theoretical and experimental study will be necessary to draw broader conclusions. A second important consideration is that the Lewis basicity of the solvent must be compatible with the photobases. This is not a heavily limiting fac- tor, since in Bronsted photacids the acidity or basicity of the solvents always play a key role when designing excited state proton transfer pathways. 59 These results are analogous to our group’s prior work on excited state proton affinity in a similar quinoline series 2 . We have reproduced the corresponding figure for Brønsted photobasicity in the SI (Figure Figure 5.19). The similarity of the photobasic effect towards protons and BF 3 indicates that they share a similar electronic origin. Finally, the figure shows the excited state bond has a stronger dependence on the Hammett parameter compared to the ground state. This, again, is consistent with the Brønsted photobasicity reported before, and is ascribed to the larger polarizability of the excited state. NH 2 OCH 3 H CF 3 CN Figure 4.5: The binding affinity in the ground and excited state using a computational F ¨ orster cycle showing similar trend with respect to the Hammett parameter as our experimental data. 60 4.5 Summary and Outlook In summary, the central value of this study is that it places the excited-state BF 3 binding affinity of quinolines on a sound thermodynamic footing while offering an intuitive physical picture for its en- ergetic control. As mentioned in the introduction, some researchers have investigated photoswitch- ing behavior of some Lewis pairs 208;209;210;211 , but these papers have not extensively explored the energetics of these systems. The application of a F¨ orster cycle analysis to light-controlled Lewis acids is an excellent foundation for future work in this field. Prior work on photoacids and photobases has established that kinetic factors can prevent proton transfer even if there is sufficient thermodynamic drive 21;19 . Therefore, establishing a similar kinetic understanding of Lewis acid transfer in the excited state is a clear next step. Application of this concept will rely upon the kinetic details of the excited state in addition to the thermodynamic arguments presented here. This will be the subject of future work. However, we can extrapolate the kinetics from our previous work on the excited state dynamics of protonated and unprotonated quinolines 19;20;22 . Typically, the excited state lifetime of quinoline after proton capture and within the singlet excited state is 5-15 ns. In some cases, the triplet states may be involved, and therefore a longer lifetime is expected. The adducted form of quinoline will also likely exhibit similar excited state lifetimes. From the application point of view, to drive a diffusion-limited reaction, in general, longer lifetimes is desired. However, in situations where a pre-associated complex is involved, fast relaxation does not pose a significant challenge. Pre-associated complexes of photobases with a target molecule can be created via hydrogen bonding or covalent tethering of the photobase in the coordination sphere of an organometallic catalyst 21;18 . We hypothesize that the synthesis of a chemical platform designed for intramolecular Lewis acid transfer may facilitate such research. Beyond this, it is valuable to study the excited state interaction of quinolines with other Lewis acids, including different types of borohalides, and Lewis acids with different electron-accepting centers, such as SO 3 or metal ions. Other relevant areas for study are the effects of environment on these excited state complexes, such as solvation or electrochemical polarization. 61 In this work we have laid the foundations of excited state thermodynamics of Lewis acid-base exchange. We anticipate that in analogy to proton transfer, this concept can be used to control Lewis interactions. For example, if a reaction requires BF3 as a catalyst, a Lewis photoacidic adduct, such as a naphtholate, could deliver the catalyst upon optical excitation. Given that BF 3 , AlCl 3 and other Lewis acids are widely used in organic synthesis as catalysts, this will open a new framework for controlling reactivity. Our current work, which is focused on photobasicity, can be easily extended to cover Lewis photoacidity. Additionally, a number of important small molecules relevant to the environment and energy such as CO 2 and CO interact as Lewis acids or bases. Therefore, in principle, it may be possible to capture such molecules when they are dissolved in a solution using photobases. One possible applied avenue following this work is the incorporation of light-activated Lewis acids and bases into frustrated Lewis pairs (FLP) platforms. Some recent publications in this field have described ”inverse” FLPs, where a strong Lewis base cooperates with a relatively weak Lewis acid to catalyze reactions 232;233 . Incorporating photobases such as quinolines into FLP frameworks could lead to further enhancement of catalytic activity. Additionally, an FLP whose catalytic strength can be easily turned on and off with light may allow for improved cycling and control of reactivity. To our knowledge, no formal study of FLPs in the excited state have been conducted. 62 Chapter 5 Conclusions and Further Work During my time as a graduate student, I have been extremely grateful to be able to learn from so many different researchers. One of the more valuable things I’ve learned is that you simply can’t know it all! It’s an easy thing to say, but remarkably true. This fact helps frame learning as more of a choice - you have to pick the right ideas to hold onto and avoid getting bogged down by the sheer volume of stuff there is to know. I have to thank Jahan Dawlaty in particular for this - in teaching his graduate students he focuses on critically important ideas - basic wave mechanics, Fourier transforms, the Boltzmann constant and thermodynamics. In understanding these ideas, you can have fairly deep conversations with a large number of people. As you might guess, I didn’t manage to solve the world’s most pressing scientific problems during my PhD. The silver lining here is that there’s plenty of work left over for future graduate students in the Dawlaty lab. One project I became interested in during the last few months is the idea of understanding and controlling interfacial halogen bonding at the surface of an electrode. This type of work is similar to the Lewis photobase work in that it would seek to push the boundaries for the types of chemistry that we are able to control with a useful experimental handle such as light or applied potential. Shayne and Joel’s work on interfacial dielecric solvation already showed some evidence for halogen bonding to a nitrile monolayer, in that halogenatedsolvents did not perfectly follow the model for pure dielectric solvation (similar to protic solvents). However, nitriles are not terribly strong halogen bond donors and it might be challenging to study a nitrile-halogen bond over a wide range of conditions. As a solution to this, Sevan and I made some progress towards surface 63 enhanced raman studies of 4-mercaptopyridine monolayers for a follow-up paper on interfacial Lewis pairs. People who study these monolayers tend to look at the ring C=C stretching modes near 1600 cm − 1 as a reporter mode, and it is very sensitive to any interaction with the pyridine’s lone pair, including halogen bonding 234 . I was able to reproduce this effect on my own in the bulk (using perfluoroiodobutane as a halogen bond donor), and an initial surface study gave me hope that it is a feasible experiment. Halogen bonding can be used for nano-scale patterning 235 , so any degree of surface control for this bond could be highly useful. Continuing this thought, Sevan and I conducted many experiments to learn more about elec- trochemical control of dative bonding - again, similar to the Lewis photobase paper, but at the electrode surface, with a pyridine monolayer as Lewis base and BF 3 as a Lewis acid. Beguilingly, we see the exact opposite trend than we were expecting based on a reasonable Hammett parameter argument. We hypothesized that as the electrode polarizes the pyridine and pushes electron density into the lone pair, those electrons will be better able to participate in dative bonding and the N-B bond will become stronger. This is borne out by simple chemical arguments and quantum chemi- cal calculations. However, when you tether pyridine to the surface, make a pyridine-BF 3 bond and apply potential, the potentials that ought to stabilize the adduct ends up destroying it. Our current hypothesis is that some auxillary chemical effect, perhaps with the electrolyte, is responsible for this unexpected behavior, and we are currently working with the Hammes-Schiffer group to better understand what’s going on. With regard to extending the work of the Lewis photobase paper, I think there is a great deal of work that could be done to develop those ideas further. A key question from an application point of view is if a quinoline is kinetically capable of capturing BF 3 in the excited state - as a true analog to proton transfer. I did a considerable number of experiments to observe this behavior, but ran into a limitation: as I increased the concentration of a BF 3 donor to allow 5-MeOQ to capture the Lewis acid in the excited state, 5-MeOQ became completely adducted in the ground state - far be- low concentrations needed to overcome diffusional timescales in the excited state. Basically, I was unable to find a BF 3 donor strong enough to prevent transfer to 5-MeOQ in the ground state but allow it in the excited state. Unfortunately, I think considerable component of solving this problem 64 is synthetic: creating a chemical platform that gets around these kinetic issues, by allowing for in- tramolecular Lewis acid transfer, for example, could demonstrate the effect satisfactorily. Further, I am very much interested in seeing how Lewis photobases might be successfully implemented to- wards Frustrated Lewis Pair catalysis. The excited state change in Lewis basicity that we estimated for quinolines could be used to impart a photoswitching character to these catalysts. Along these lines, the Dawlaty group is also currently funded to investigate the fundamentals of CO 2 capture, focus right now on amine based CO 2 capture. A proton transfer step is critical to generate the activated amino anion 236 , which is an easy place to incorporate a photobase. However, there is a chance that CO 2 could be coaxed to bind with the lone pair of a quinoline if the energetics are tuned correctly. I would be very much interested to either of those chemistries developed further. 65 Bibliography [1] Yuezhi Mao, Paul R Horn, and Martin Head-Gordon. Energy decomposition analysis in an adiabatic picture. Physical Chemistry Chemical Physics, 19(8):5944–5958, 2017. [2] Eric William Driscoll, Jonathan Ryan Hunt, and Jahan M Dawlaty. Photobasicity in quino- lines: origin and tunability via the substituents’ hammett parameters. The journal of physical chemistry letters, 7(11):2093–2099, 2016. [3] Francesco Rizzi, Nees Jan van Eck, and Marco Frey. The production of scientific knowledge on renewable energies: Worldwide trends, dynamics and challenges and implications for management. Renewable Energy, 62:657–671, 2014. [4] Sharon Hammes-Schiffer and Giulia Galli. Integration of theory and experiment in the modelling of heterogeneous electrocatalysis. Nature Energy, 6(7):700–705, 2021. [5] DM Kolb and J Schneider. Surface reconstruction in electrochemistry: Au (100-(5× 20), au (111)-(1× 23) and au (110)-(1× 2). Electrochimica Acta, 31(8):929–936, 1986. [6] Jaeyune Ryu and Yogesh Surendranath. Polarization-induced local ph swing promotes pd- catalyzed co2 hydrogenation. Journal of the American Chemical Society, 142(31):13384– 13390, 2020. [7] David Leonard Chapman. Li. a contribution to the theory of electrocapillarity. The London, Edinburgh, and Dublin philosophical magazine and journal of science, 25(148):475–481, 1913. [8] Shayne A Sorenson, Joel G Patrow, and Jahan M Dawlaty. Solvation reaction field at the interface measured by vibrational sum frequency generation spectroscopy. J. Am. Chem. Soc., 139(6):2369–2378, 2017. [9] Joel G Patrow, Shayne A Sorenson, and Jahan M Dawlaty. Direct spectroscopic measure- ment of interfacial electric fields near an electrode under polarizing or current-carrying con- ditions. J. Phys. Chem. C, 121(21):11585–11592, 2017. [10] Stephen D Fried and Steven G Boxer. Measuring electric fields and noncovalent interactions using the vibrational stark effect. Acc. Chem. Res., 48(4):998–1006, 2015. [11] Sohini Sarkar, Cindy Tseng, Anwesha Maitra, Matthew J V oegtle, and Jahan M Dawlaty. Advances in vibrational stark shift spectroscopy for measuring interfacial electric fields. In Emerging Trends in Chemical Applications of Lasers, pages 199–224. ACS Publications, 2021. 66 [12] Stephen D Fried and Steven G Boxer. Electric fields and enzyme catalysis. Annual review of biochemistry, 86:387–415, 2017. [13] Sohini Sarkar, Joel G Patrow, Matthew J V oegtle, Anuj K Pennathur, and Jahan M Dawlaty. Electrodes as polarizing functional groups: Correlation between hammett parameters and electrochemical polarization. J. Phys. Chem. C, 123(8):4926–4937, 2019. [14] Matthew J V oegtle, Tanmoy Pal, Anuj K Pennathur, Sevan Menachekanian, Joel G Patrow, Sohini Sarkar, Qiang Cui, and Jahan M Dawlaty. Interfacial polarization and ionic struc- ture at the ionic liquid–metal interface studied by vibrational spectroscopy and molecular dynamics simulations. The Journal of Physical Chemistry B, 125(10):2741–2753, 2021. [15] Anuj K Pennathur, Matthew J V oegtle, Sevan Menachekanian, and Jahan M Dawlaty. Strong propensity of ionic liquids in their aqueous solutions for an organic-modified metal surface. J. Phys. Chem. B, 124(34):7500–7507, 2020. [16] James M Mayer. Proton-coupled electron transfer: a reaction chemist’s view. Annu. Rev. Phys. Chem., 55:363–390, 2004. [17] Shima Haghighat, Sarah Ostresh, and Jahan M Dawlaty. Controlling proton conductivity with light: a scheme based on photoacid doping of materials. The Journal of Physical Chemistry B, 120(5):1002–1007, 2016. [18] Ivan Demianets, Jonathan R Hunt, Jahan M Dawlaty, and Travis J Williams. Optical p k a control in a bifunctional iridium complex. Organometallics, 38(2):200–204, 2019. [19] Eric William Driscoll, Jonathan Ryan Hunt, and Jahan M Dawlaty. Proton capture dynamics in quinoline photobases: substituent effect and involvement of triplet states. The Journal of Physical Chemistry A, 121(38):7099–7107, 2017. [20] Jonathan Ryan Hunt and Jahan M Dawlaty. Photodriven deprotonation of alcohols by a quinoline photobase. The Journal of Physical Chemistry A, 122(40):7931–7940, 2018. [21] Jonathan Ryan Hunt, Cindy Tseng, and Jahan M Dawlaty. Donor–acceptor preassociation, excited state solvation threshold, and optical energy cost as challenges in chemical applica- tions of photobases. Faraday discussions, 216:252–268, 2019. [22] Jonathan Ryan Hunt and Jahan M Dawlaty. Kinetic evidence for the necessity of two proton donor molecules for successful excited state proton transfer by a photobase. The Journal of Physical Chemistry A, 123(48):10372–10380, 2019. [23] Matthew J V oegtle and Jahan M Dawlaty. Can brønsted photobases act as lewis photobases? Journal of the American Chemical Society, 2022. [24] Joel G Patrow, Yi Wang, and Jahan M Dawlaty. Interfacial lewis acid–base adduct formation probed by vibrational spectroscopy. J. Phys. Chem. Lett., 9(13):3631–3638, 2018. 67 [25] Sophya F Alamudun, Kyle Tanovitz, Lanette Espinosa, April Fajardo, John Galvan, and Andrew S Petit. Structure–photochemical function relationships in the photobasicity of aromatic heterocycles containing multiple ring nitrogen atoms. The Journal of Physical Chemistry A, 125(1):13–24, 2020. [26] Louis P Hammett. The effect of structure upon the reactions of organic compounds. benzene derivatives. Journal of the American Chemical Society, 59(1):96–103, 1937. [27] Darl H McDaniel and Herbert C Brown. An extended table of hammett substitutent con- stants based on the ionization of substituted benzoic acids. The Journal of Organic Chem- istry, 23(3):420–427, 1958. [28] Corwin Hansch, A Leo, and RW Taft. A survey of hammett substituent constants and resonance and field parameters. Chemical Reviews, 91(2):165–195, 1991. [29] Hang Chen, Lu Jia, Jia Yao, Jinghui Hu, Kexian Chen, Zhirong Chen, and Haoran Li. The relationships of catalytic activity of metal schiff base catalysts and the hammett constants of their anion-cationic substituents on ligand. Journal of Physical Organic Chemistry, 28(8): 570–574, Aug 2015. [30] Malcolm S Groves, Kacie J Nelson, Ryan C Nelson, and Kana Takematsu. ph switch for oh- photoacidity in 5-amino-2-naphthol and 8-amino-2-naphthol. Physical Chemistry Chemical Physics, 2018. [31] Israel Fern´ andez and Gernot Frenking. Correlation between hammett substituent constants and directly calculatedπ-conjugation strength. J. Org. Chem., 71(6):2251–2256, Mar 2006. [32] Alexander Star, Tzong-Ru Han, Jean-Christophe P. Gabriel, Keith Bradley, and George Gr¨ uner. Interaction of aromatic compounds with carbon nanotubes: Correlation to the ham- mett parameter of the substituent and measured carbon nanotube fet response. Nano Lett., 3(10):1421–1423, Oct 2003. [33] P. J. Bray and R. G. Barnes. Estimates of hammett’s sigma values from quadrupole reso- nance studies. J. Chem. Phys., 27(2):551–560, Aug 1957. [34] Velaug Myrseth, Leif J. Saethre, Knut J. Børve, and T. D. Thomas. The substituent effect of the methyl group. carbon 1s ionization energies, proton affinities, and reactivities of the methylbenzenes. The Journal of organic chemistry, 72(15):5715–5723, Jul 20, 2007. URL https://www.ncbi.nlm.nih.gov/pubmed/17595144. [35] Paresh C. Ray and Jerzy Leszczynski. Two-photon absorption and first nonlinear optical properties of ionic octupolar molecules: structure-function relationships and solvent effects. The journal of physical chemistry. A, 109(30):6689–6696, Aug 4, 2005. URL https:// www.ncbi.nlm.nih.gov/pubmed/16834021. [36] Julie N. Graff, Anne E. McElhaney, Partha Basu, Nadine E. Gruhn, Chaung-Sheng J. Chang, and John H. Enemark. Electrochemistry and photoelectron spectroscopy of oxomolybde- num(v) complexes with phenoxide ligands: effect of para substituents on redox potentials, 68 heterogeneous electron transfer rates, and ionization energies. Inorganic chemistry, 41(10): 2642, May 20, 2002. URLhttps://www.ncbi.nlm.nih.gov/pubmed/12005487. [37] Richard Lincoln, Lana E. Greene, Katerina Krumova, Zhutian Ding, and Gonzalo Cosa. Electronic excited state redox properties for bodipy dyes predicted from hammett constants: estimating the driving force of photoinduced electron transfer. The journal of physical chem- istry. A, 118(45):10622–10630, Nov 13, 2014. URL https://www.ncbi.nlm.nih.gov/ pubmed/25066755. [38] Julien Frey, Basile F. E. Curchod, Rosario Scopelliti, Ivano Tavernelli, Ursula Rothlis- berger, Mohammad K. Nazeeruddin, and Etienne Baranoff. Structure-property relation- ships based on hammett constants in cyclometalated iridium(iii) complexes: their applica- tion to the design of a fluorine-free firpic-like emitter. Dalton transactions (Cambridge, England : 2003), 43(15):5667–5679, Apr 21, 2014. URL https://www.ncbi.nlm.nih. gov/pubmed/24345847. [39] K. C. Gross, P. G. Seybold, Z. Peralta-Inga, J. S. Murray, and P. Politzer. Comparison of quantum chemical parameters and hammett constants in correlating pk(a) values of sub- stituted anilines. The Journal of organic chemistry, 66(21):6919, Oct 19, 2001. URL https://www.ncbi.nlm.nih.gov/pubmed/11597210. [40] Florian M. Wisser, Pierrick Berruyer, Luis Cardenas, Yorck Mohr, Elsje A. Quadrelli, Anne Lesage, David Farrusseng, and J´ erˆ ome Canivet. Hammett parameter in microporous solids as macroligands for heterogenized photocatalysts. ACS Catalysis, 8(3):1653–1661, Mar 2, 2018. [41] T. Krygowski and N. Sadlej-Sosnowska. Towards physical interpretation of hammett con- stants: charge transferred between active regions of substituents and a functional group. Structural Chemistry, 22(1):17–22, Feb 2011. [42] Meredith Hyatt Barbee, Tatiana B Kouznetsova, Scott L Barrett, Gregory R Gossweiler, Yangju Lin, Shiva Kumar Rastogi, William J Brittain, and Stephen L Craig. Substituent effects and mechanism in a mechanochemical reaction. Journal of the American Chemical Society, 140(40):12746, 2018. [43] Jana K Staffa, Lisa Lorenz, Michael Stolarski, Daniel H Murgida, Ingo Zebger, Tillmann Utesch, Jacek Kozuch, and Peter Hildebrandt. Determination of the local electric field at au/sam interfaces using the vibrational stark effect. The Journal of Physical Chemistry C, 121(40):22274–22285, 2017. [44] Arieh Warshel, Pankaz K Sharma, Mitsunori Kato, and William W Parson. Modeling elec- trostatic effects in proteins. Biochimica et Biophysica Acta (BBA)-Proteins and Proteomics, 1764(11):1647–1676, 2006. [45] HJ Kreuzer. Surface physics and chemistry in high electric fields. In Chemistry and Physics of Solid Surfaces VIII, pages 133–158. Springer, 1990. 69 [46] Fanglin Che, Jake T Gray, Su Ha, and Jean-Sabin McEwen. Catalytic water dehydrogena- tion and formation on nickel: Dual path mechanism in high electric fields. Journal of Catalysis, 332:187–200, 2015. [47] Eric M Stuve. Ionization of water in interfacial electric fields: An electrochemical view. Chemical Physics Letters, 519:1–17, 2012. [48] Stephen D Fried, Sayan Bagchi, and Steven G Boxer. Extreme electric fields power catalysis in the active site of ketosteroid isomerase. Science, 346(6216):1510–1514, 2014. [49] Sason Shaik, Debasish Mandal, and Rajeev Ramanan. Oriented electric fields as future smart reagents in chemistry. Nature chemistry, 8(12):1091, 2016. [50] Fanglin Che, Jake T Gray, Su Ha, Norbert Kruse, Susannah L Scott, and Jean-Sabin McEwen. Elucidating the roles of electric fields in catalysis: A perspective. ACS Catal- ysis, 8:5153–5174, 2018. [51] Arun Chattopadhyay and Steven G Boxer. Vibrational stark effect spectroscopy. Journal of the American Chemical Society, 117(4):1449–1450, 1995. [52] Steven S. Andrews and Steven G. Boxer. Vibrational Stark Effects of Nitriles. I. Methods and Experimental Results. J. Phys. Chem. A, 104(51):11853–11863, 2000. ISSN 10895639. doi: 10.1021/jp002242r. [53] Steven S. Andrews and Steven G. Boxer. Vibrational Stark Effects of Nitriles II. Physical Origins of Stark Effects from Experiment and Perturbation Models. J. Phys. Chem. A, 106: 466–477, 2002. [54] Eun Sun Park, Steven S Andrews, Robert B Hu, and Steven G Boxer. Vibrational stark spectroscopy in proteins: A probe and calibration for electrostatic fields. The Journal of Physical Chemistry B, 103(45):9813–9817, 1999. [55] Joel G. Patrow, Shayne A. Sorenson, and Jahan M. Dawlaty. Direct spectroscopic mea- surement of interfacial electric fields near an electrode under polarizing or current-carrying conditions. The Journal of Physical Chemistry C, 121(21):11585–11592, Jun 2017. [56] Shiguo Zhang, Yan Zhang, Xiangyuan Ma, Liujin Lu, Yude He, and Youquan Deng. Ben- zonitrile as a probe of local environment in ionic liquids. J. Phys. Chem. B, 117(9):2764– 2772, 2013. [57] Daniel J. Aschaffenburg and Richard S. Moog. Probing Hydrogen Bonding Environments: Solvatochromic Effects on the CN Vibration of Benzonitrile. J. Phys. Chem. B, 113(38): 12736–12743, 2009. ISSN 15206106. doi: 10.1021/jp905802a. [58] Sayan Bagchi, Stephen D Fried, and Steven G Boxer. A solvatochromic model calibrates nitriles’ vibrational frequencies to electrostatic fields. Journal of the American Chemical Society, 134(25):10373–10376, 2012. 70 [59] Nirmalya Ballav, Bj¨ orn Sch¨ upbach, Ole Dethloff, Peter Feulner, Andreas Terfort, and Michael Zharnikov. Direct Probing Molecular Twist and Tilt in Aromatic Self-Assembled Monolayers. J. Am. Chem. Soc., 129(50):15416–15417, 2007. ISSN 00027863. doi: 10.1021/ja0751882. [60] P. J. Krueger and H. W. Thompson. Vibrational band intensities and structural factors. Proc. R. Soc. London, Ser. A, 250(1260):22–38, Feb 24, 1959. URL http://rspa. royalsocietypublishing.org/content/250/1260/22.abstract. [61] Z Getahun, C Y Huang, T Wang, B De Leon, W F DeGrado, and F Gai. Using Nitrile- Derivated Amino Acids as Infrared Probes of Local Environment. J. Am. Chem. Soc., 125 (405-406):405–411, 2003. [62] Keith F Purcell and Russell S Drago. Studies of the bonding in acetonitrile adducts1. Journal of the American Chemical Society, 88(5):919–924, 1966. [63] Vanessa Oklejas and Joel M. Harris. In-Situ Investigation of Binary-Component Self- Assembled Monolayers: A SERS-Based Spectroelectrochemical Study of The Effects of Monolayer Composition on Interfacial Structure. Langmuir, 19(14):5794–5801, 2003. ISSN 07437463. doi: 10.1021/la020916e. [64] Sayan Bagchi, Stephen D Fried, and Steven G Boxer. A Solvatochromic Model Calibrates Nitriles’ Vibrational Frequencies to Electrostatic Fields. J. Am. Chem. Soc., 134:10373– 10376, 2012. [65] Nicholas M Levinson, Stephen D Fried, and Steven G Boxer. Solvent-Induced Infrared Frequency Shifts in Aromatic Nitriles Are Quantitatively Described by the Vibrational Stark Effect. J. Phys. Chem. B, 116:10470–10476, 2012. [66] Chiara H Giammanco, Patrick L Kramer, Daryl B Wong, and Michael D Fayer. Water dynamics in 1-alkyl-3-methylimidazolium tetrafluoroborate ionic liquids. The Journal of Physical Chemistry B, 120(44):11523–11538, 2016. [67] Divia Dinesh Patel and Jong-Min Lee. Applications of ionic liquids. The Chemical Record, 12(3):329–355, 2012. [68] Natalia V Plechkova and Kenneth R Seddon. Applications of ionic liquids in the chemical industry. Chemical Society Reviews, 37(1):123–150, 2008. [69] Hiroyuki Ohno. Electrochemical aspects of ionic liquids. John Wiley & Sons, 2005. [70] T Romann, O Oll, P Pikma, H Tamme, and E Lust. Surface chemistry of carbon electrodes in 1-ethyl-3-methylimidazolium tetrafluoroborate ionic liquid–an in situ infrared study. Elec- trochimica Acta, 125:183–190, 2014. [71] Maan Hayyan, Farouq S Mjalli, Mohd Ali Hashim, Inas M AlNashef, and Tan Xue Mei. Investigating the electrochemical windows of ionic liquids. Journal of Industrial and Engi- neering Chemistry, 19(1):106–112, 2013. 71 [72] Joel G. Patrow, Yi Wang, and Jahan M. Dawlaty. Interfacial lewis acid-base adduct forma- tion probed by vibrational spectroscopy. The journal of physical chemistry letters, 9(13): 3631–3638, Jul 5, 2018. URLhttps://www.ncbi.nlm.nih.gov/pubmed/29897764. [73] Fadel Y . Shalhout, Sergey Malyk, and Alexander V . Benderskii. Relative Phase Change of Nearby Resonances in Temporally Delayed Sum Frequency Spectra. J. Phys. Chem. Lett., 3(23):3493–3497, 2012. ISSN 19487185. doi: 10.1021/jz3014437. [74] Yihan Shao, Zhengting Gan, Evgeny Epifanovsky, Andrew TB Gilbert, Michael Wormit, Joerg Kussmann, Adrian W Lange, Andrew Behn, Jia Deng, Xintian Feng, et al. Advances in molecular quantum chemistry contained in the q-chem 4 program package. Mol. Phys., 113(2):184–215, 2015. [75] Han-Gook Cho and Byeong-Sea Cheong. A theoretical investigation of the structure and vibrational frequencies of ch3cn–bf3 and ch3cn–bcl3. Journal of Molecular Structure: THEOCHEM, 496(1-3):185–198, 2000. [76] Mathew D Halls, Julia Velkovski, and H Bernhard Schlegel. Harmonic frequency scaling factors for hartree-fock, s-vwn, b-lyp, b3-lyp, b3-pw91 and mp2 with the sadlej pvtz electric property basis set. Theoretical Chemistry Accounts, 105(6):413–421, 2001. [77] R. N. Jones, W. F. Forbes, and W. A. Mueller. The infrared carbonyl stretching bands of ring substituted acetophenones. Canadian Journal of Chemistry, 35(5):504–514, May 1, 1957. URLhttp://www.nrcresearchpress.com/doi/abs/10.1139/v57-071. [78] Divia Dinesh Patel and Jong-Min Lee. Applications of ionic liquids. Chem. Rec., 12(3): 329–355, 2012. [79] Natalia V Plechkova and Kenneth R Seddon. Applications of ionic liquids in the chemical industry. Chem. Soc. Rev., 37(1):123–150, 2008. [80] Hiroyuki Ohno. Electrochemical aspects of ionic liquids. John Wiley & Sons, 2011. [81] Qinghua Zhang and Jeanne M Shreeve. Ionic liquid propellants: Future fuels for space propulsion. Chem. Eur. J., 19(46):15446–15451, 2013. [82] Maxim V Fedorov and Alexei A Kornyshev. Ionic liquids at electrified interfaces. Chem. Rev., 114(5):2978–3036, 2014. [83] Peter Reichert, Kasper Skov Kjær, Tim Brandt van Driel, Julian Mars, Jannis Walther Ochsmann, Diego Pontoni, Moshe Deutsch, Martin Meedom Nielsen, and Markus Mezger. Molecular scale structure and dynamics at an ionic liquid/electrode interface. Faraday Dis- cuss., 206:141–157, 2017. [84] Hiroyuki Tokuda, Kikuko Hayamizu, Kunikazu Ishii, Md Abu Bin Hasan Susan, and Masayoshi Watanabe. Physicochemical properties and structures of room temperature ionic liquids. 1. variation of anionic species. J. Phys. Chem. B, 108(42):16593–16600, 2004. [85] Yan Levin. Electrostatic correlations: from plasma to biology. Rep. Prog. Phys, 65(11): 1577, 2002. 72 [86] Yaakov Rosenfeld. Free energy model for inhomogeneous fluid mixtures: Yukawa-charged hard spheres, general interactions, and plasmas. J. Chem. Phys., 98(10):8126–8148, 1993. [87] Werner Freyland. Coulombic fluids: bulk and interfaces , volume 168. Springer Science & Business Media, 2011. [88] Ruth M Lynden-Bell, AI Frolov, and Maxim V Fedorov. Electrode screening by ionic liquids. Phys. Chem. Chem. Phys., 14(8):2693–2701, 2012. [89] Naoya Nishi, Junya Uchiyashiki, Yoichi Ikeda, Seiji Katakura, Tatsuro Oda, Masahiro Hino, and Norifumi L Yamada. Potential-dependent structure of the ionic layer at the electrode interface of an ionic liquid probed using neutron reflectometry. J. Phys. Chem. C, 123(14): 9223–9230, 2019. [90] Seiya Watanabe, Georgia A Pilkington, Anna Oleshkevych, Patricia Pedraz, Milad Radiom, Rebecca Welbourn, Sergei Glavatskih, and Mark W Rutland. Interfacial structuring of non- halogenated imidazolium ionic liquids at charged surfaces: effect of alkyl chain length. Phys. Chem. Chem. Phys., 22(16):8450–8460, 2020. [91] Rajni Ratti. Ionic liquids: Synthesis and applications in catalysis. Adv. Chem., 2014, 2014. [92] Sason Shaik, Debasish Mandal, and Rajeev Ramanan. Oriented electric fields as future smart reagents in chemistry. Nat. Chem., 8(12):1091, 2016. [93] Arieh Warshel and Michael Levitt. Theoretical studies of enzymic reactions: dielectric, electrostatic and steric stabilization of the carbonium ion in the reaction of lysozyme. J. Mol. Biol., 103(2):227–249, 1976. [94] Ian T Suydam, Christopher D Snow, Vijay S Pande, and Steven G Boxer. Electric fields at the active site of an enzyme: Direct comparison of experiment with theory. Science, 313 (5784):200–204, 2006. [95] Alasdair W Taylor, Peter Licence, and Andrew P Abbott. Non-classical diffusion in ionic liquids. Phys. Chem. Chem. Phys., 13(21):10147–10154, 2011. [96] Hainan Wang and Laurent Pilon. Intrinsic limitations of impedance measurements in deter- mining electric double layer capacitances. Electrochim. Acta, 63:55–63, 2012. [97] M Gnahm, T Pajkossy, and DM Kolb. The interface between au (111) and an ionic liquid. Electrochim. Acta, 55(21):6212–6217, 2010. [98] Marcel Dr¨ uschler and Bernhard Roling. Commentary on ‘The interface between Au (111) and an ionic liquid’. Electrochim. Acta, 56(20):7243–7245, 2011. [99] Bernhard Roling and Marcel Dr¨ uschler. Comments on “Intrinsic limitations of impedance measurements in determining electric double layer capacitances” by h. wang and l. pilon [electrochim. acta 63 (2012) 55]. Electrochim. Acta, 76:526–528, 2012. 73 [100] Jana K Staffa, Lisa Lorenz, Michael Stolarski, Daniel H Murgida, Ingo Zebger, Tillmann Utesch, Jacek Kozuch, and Peter Hildebrandt. Determination of the local electric field at au/sam interfaces using the vibrational stark effect. J. Phys. Chem. C, 121(40):22274– 22285, 2017. [101] Nicholas HC Lewis, Aysenur Iscen, Alanna Felts, Bogdan Dereka, George C Schatz, and Andrei Tokmakoff. Vibrational probe of aqueous electrolytes: The field is not enough. J. Phys. Chem. B, 124(32):7013–7026, 2020. [102] Zachary K Goldsmith, Maxim Secor, and Sharon Hammes-Schiffer. Inhomogeneity of in- terfacial electric fields at vibrational probes on electrode surfaces. ACS Cent. Sci., 6(2): 304–311, 2020. [103] RM Lynden-Bell. Gas—liquid interfaces of room temperature ionic liquids. Mol. Phys., 101(16):2625–2633, 2003. [104] RM Lynden-Bell and M Del Popolo. Simulation of the surface structure of butylmethylim- idazolium ionic liquids. Phys. Chem. Chem. Phys., 8(8):949–954, 2006. [105] E Sloutskin, RM Lynden-Bell, S Balasubramanian, and M Deutsch. The surface structure of ionic liquids: Comparing simulations with x-ray measurements. J. Chem. Phys., 125(17): 174715, 2006. [106] Tianying Yan, Shu Li, Wei Jiang, Xueping Gao, Bing Xiang, and Gregory A V oth. Structure of the liquid- vacuum interface of room-temperature ionic liquids: A molecular dynamics study. J. Phys. Chem. B, 110(4):1800–1806, 2006. [107] Carlos Pinilla, Mario G Del Popolo, Ruth M Lynden-Bell, and Jorge Kohanoff. Structure and dynamics of a confined ionic liquid. topics of relevance to dye-sensitized solar cells. J. Phys. Chem. B, 109(38):17922–17927, 2005. [108] Carlos Pinilla, MG Del P´ opolo, Jorge Kohanoff, and RM Lynden-Bell. Polarization re- laxation in an ionic liquid confined between electrified walls. J. Phys. Chem. B, 111(18): 4877–4884, 2007. [109] Sergey A Kislenko, Igor S Samoylov, and Ravil H Amirov. Molecular dynamics simulation of the electrochemical interface between a graphite surface and the ionic liquid [bmim][pf 6]. Phys. Chem. Chem. Phys., 11(27):5584–5590, 2009. [110] Samuel Ntim and Marialore Sulpizi. Role of image charges in ionic liquid confined between metallic interfaces. Phys. Chem. Chem. Phys., 22(19):10786–10791, 2020. [111] John F Smalley, Stephen W Feldberg, Christopher ED Chidsey, Matthew R Linford, Mar- shall D Newton, and Yi-Ping Liu. The kinetics of electron transfer through ferrocene- terminated alkanethiol monolayers on gold. J. Phys. Chem., 99(35):13141–13149, 1995. [112] Christopher P Smith and Henry S White. Theory of the interfacial potential distribution and reversible voltammetric response of electrodes coated with electroactive molecular films. Anal. Chem., 64(20):2398–2405, 1992. 74 [113] John G Van Alsten. Self-assembled monolayers on engineering metals: structure, derivati- zation, and utility. Langmuir, 15(22):7605–7614, 1999. [114] Chariz Y Pe˜ nalber, Gary A Baker, and Steven Baldelli. Sum frequency generation spec- troscopy of imidazolium-based ionic liquids with cyano-functionalized anions at the solid salt–liquid interface. J. Phys. Chem. B, 117(19):5939–5949, 2013. [115] Selimar Rivera-Rubero and Steven Baldelli. Surface characterization of 1-butyl-3- methylimidazolium br-, i-, pf6-, bf4-,(cf3so2) 2n-, scn-, ch3so3-, ch3so4-, and (cn) 2n-ionic liquids by sum frequency generation. J. Phys. Chem. B, 110(10):4756–4765, 2006. [116] Steven Baldelli. Surface structure at the ionic liquid- electrified metal interface. Acc. Chem. Res., 41(3):421–431, 2008. [117] Selimar Rivera-Rubero and Steven Baldelli. Influence of water on the surface of hydrophilic and hydrophobic room-temperature ionic liquids. J. Am. Chem. Soc., 126(38):11788–11789, 2004. [118] Natalia Garc´ ıa Rey and Dana D Dlott. Effects of water on low-overpotential co 2 reduc- tion in ionic liquid studied by sum-frequency generation spectroscopy. Phys. Chem. Chem. Phys., 19(16):10491–10501, 2017. [119] Natalia Garc´ ıa Rey and Dana D Dlott. Structural transition in an ionic liquid controls co2 electrochemical reduction. J. Phys. Chem. C, 119(36):20892–20899, 2015. [120] Brian A Rosen, John L Haan, Prabuddha Mukherjee, Bjorn Braunschweig, Wei Zhu, Amin Salehi-Khojin, Dana D Dlott, and Richard I Masel. In situ spectroscopic examination of a low overpotential pathway for carbon dioxide conversion to carbon monoxide. J. Phys. Chem. C, 116(29):15307–15312, 2012. [121] Amr Tamimi, Heather E Bailey, and Michael D Fayer. Alkyl chain length dependence of the dynamics and structure in the ionic regions of room-temperature ionic liquids. J. Phys. Chem. B, 120(30):7488–7501, 2016. [122] Chiara H Giammanco, Patrick L Kramer, Steven A Yamada, Jun Nishida, Amr Tamimi, and Michael D Fayer. Carbon dioxide in an ionic liquid: Structural and rotational dynamics. J. Chem. Phys., 144(10):104506, 2016. [123] Michael D Fayer. Dynamics and structure of room temperature ionic liquids. Chem. Phys. Lett., 616:259–274, 2014. [124] Sohini Sarkar, Anwesha Maitra, Soumyodip Banerjee, V Sara Thoi, and Jahan M Dawlaty. Electric fields at metal–surfactant interfaces: A combined vibrational spectroscopy and ca- pacitance study. J. Phys. Chem. B, 124(7):1311–1321, 2020. [125] C. Humbert, B. Busson, C. Six, A. Gayral, M. Gruselle, F. Villain, and A. Tadjeddine. Sum-Frequency Generation as a Vibrational and Electronic Probe of The Electrochemical Interface and Thin Films. J. Electroanal. Chem., 621(2):314–321, 2008. ISSN 00220728. doi: 10.1016/j.jelechem.2008.02.008. 75 [126] Fadel Y Shalhout, Sergey Malyk, and Alexander V Benderskii. Relative phase change of nearby resonances in temporally delayed sum frequency spectra. J. Phys. Chem. Lett., 3 (23):3493–3497, 2012. [127] Shavini Wijesuriya, Krishna Burugapalli, Ruth Mackay, Godwin Ajaezi, and Wamadeva Balachandran. Chemically roughened solid silver: A simple, robust and broadband sers substrate. Sensors, 16(10):1742, 2016. [128] Sung Ho Ha, Ngoc Lan Mai, and Yoon-Mo Koo. Microwave-assisted separation of ionic liquids from aqueous solution of ionic liquids. J. Chromatogr. A, 1217(49):7638–7641, 2010. [129] Sunhwan Jo, Taehoon Kim, Vidyashankara G Iyer, and Wonpil Im. Charmm-gui: a web- based graphical user interface for charmm. J. Comput. Chem., 29(11):1859–1865, 2008. [130] Bernard R Brooks, Charles L Brooks III, Alexander D Mackerell Jr, Lennart Nilsson, Robert J Petrella, Benoˆ ıt Roux, Youngdo Won, Georgios Archontis, Christian Bartels, Ste- fan Boresch, et al. Charmm: the biomolecular simulation program. J. Comput. Chem., 30 (10):1545–1614, 2009. [131] Jean-Pierre Hansen and Ian R McDonald. Theory of simple liquids. Elsevier, 1990. [132] CJ Margulis, HA Stern, and BJ Berne. Computer simulation of a “green chemistry” room- temperature ionic solvent. J. Phys. Chem. B, 106(46):12017–12021, 2002. [133] Timothy I Morrow and Edward J Maginn. Molecular dynamics study of the ionic liquid 1-n- butyl-3-methylimidazolium hexafluorophosphate. J. Phys. Chem. B, 106(49):12807–12813, 2002. [134] Mario G Del P´ opolo and Gregory A V oth. On the structure and dynamics of ionic liquids. J. Phys. Chem. B, 108(5):1744–1752, 2004. [135] Tianying Yan, Christian J Burnham, Mario G Del P´ opolo, and Gregory A V oth. Molecular dynamics simulation of ionic liquids: The effect of electronic polarizability. J. Phys. Chem. B, 108(32):11877–11881, 2004. [136] Jesse G McDaniel and Arun Yethiraj. Influence of electronic polarization on the structure of ionic liquids. J. Phys. Chem. Lett., 9(16):4765–4770, 2018. [137] Justin B Haskins and John W Lawson. Evaluation of molecular dynamics simulation meth- ods for ionic liquid electric double layers. J. Chem. Phys., 144(18):184707, 2016. [138] Dmitry Bedrov, Oleg Borodin, Zhe Li, and Grant D Smith. Influence of polarization on structural, thermodynamic, and dynamic properties of ionic liquids obtained from molecular dynamics simulations. J. Phys. Chem. B, 114(15):4984–4997, 2010. [139] Kenta Fujii, Yasufumi Soejima, Yasuhiro Kyoshoin, Shuhei Fukuda, Ryo Kanzaki, Ya- suhiro Umebayashi, Toshio Yamaguchi, Shin-ichi Ishiguro, and Toshiyuki Takamuku. Liquid structure of room-temperature ionic liquid, 1-ethyl-3-methylimidazolium bis- (trifluoromethanesulfonyl) imide. J. Phys. Chem. B, 112(14):4329–4336, 2008. 76 [140] Jos´ e N Canongia Lopes, Johnny Deschamps, and Ag´ ılio AH P´ adua. Modeling ionic liquids using a systematic all-atom force field. J. Phys. Chem. B, 108(6):2038–2047, 2004. [141] Alex D MacKerell Jr, Donald Bashford, MLDR Bellott, Roland Leslie Dunbrack Jr, Jef- frey D Evanseck, Martin J Field, Stefan Fischer, Jiali Gao, H Guo, Sookhee Ha, et al. All-atom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B, 102(18):3586–3616, 1998. [142] Jing Huang and Alexander D MacKerell Jr. Charmm36 all-atom additive protein force field: Validation based on comparison to nmr data. J. Comput. Chem., 34(25):2135–2145, 2013. [143] Kenno Vanommeslaeghe, Elizabeth Hatcher, Chayan Acharya, Sibsankar Kundu, Shijun Zhong, Jihyun Shim, Eva Darian, Olgun Guvench, P Lopes, Igor V orobyov, et al. Charmm general force field: A force field for drug-like molecules compatible with the charmm all- atom additive biological force fields. J. Comput. Chem., 31(4):671–690, 2010. [144] Kenno Vanommeslaeghe and Alexander D MacKerell Jr. Automation of the charmm general force field (cgenff) i: bond perception and atom typing. J. Chem. Inf. Model., 52(12):3144– 3154, 2012. [145] Kenno Vanommeslaeghe, E Prabhu Raman, and Alexander D MacKerell Jr. Automation of the charmm general force field (cgenff) ii: assignment of bonded parameters and partial atomic charges. J. Chem. Inf. Model., 52(12):3155–3168, 2012. [146] Wenbo Yu, Xibing He, Kenno Vanommeslaeghe, and Alexander D MacKerell Jr. Exten- sion of the charmm general force field to sulfonyl-containing compounds and its utility in biomolecular simulations. J. Comput. Chem., 33(31):2451–2468, 2012. [147] Jos´ e N Canongia Lopes and Ag´ ılio AH P´ adua. Molecular force field for ionic liquids iii: Imidazolium, pyridinium, and phosphonium cations; chloride, bromide, and dicyanamide anions. J. Phys. Chem. B, 110(39):19586–19592, 2006. [148] John P Perdew, John A Chevary, Sy H V osko, Koblar A Jackson, Mark R Pederson, Dig J Singh, and Carlos Fiolhais. Atoms, molecules, solids, and surfaces: Applications of the generalized gradient approximation for exchange and correlation. Phys. Rev. B, 46(11): 6671, 1992. [149] John P Perdew, JA Chevary, SH V osko, Koblar A Jackson, Mark R Pederson, DJ Singh, and Carlos Fiolhais. Erratum: Atoms, molecules, solids, and surfaces: Applications of the generalized gradient approximation for exchange and correlation. Phys. Rev. B, 48(7):4978, 1993. [150] John P. Perdew, Kieron Burke, and Yue Wang. Generalized gradient approximation for the exchange-correlation hole of a many-electron system. Phys. Rev. B, 54:16533–16539, Dec 1996. doi: 10.1103/PhysRevB.54.16533. URL https://link.aps.org/doi/10.1103/ PhysRevB.54.16533. 77 [151] Burkhard Miehlich, Andreas Savin, Hermann Stoll, and Heinzwerner Preuss. Results ob- tained with the correlation energy density functionals of becke and lee, yang and parr. Chem. Phys. Lett., 157(3):200 – 206, 1989. ISSN 0009-2614. doi: https://doi.org/10.1016/ 0009-2614(89)87234-3. URL http://www.sciencedirect.com/science/article/ pii/0009261489872343. [152] Chengteh Lee, Weitao Yang, and Robert G. Parr. Development of the colle-salvetti correlation-energy formula into a functional of the electron density. Phys. Rev. B, 37:785– 789, Jan 1988. doi: 10.1103/PhysRevB.37.785. URL https://link.aps.org/doi/10. 1103/PhysRevB.37.785. [153] A. D. Becke. Density-functional exchange-energy approximation with correct asymptotic behavior. Phys. Rev. A, 38:3098–3100, Sep 1988. doi: 10.1103/PhysRevA.38.3098. URL https://link.aps.org/doi/10.1103/PhysRevA.38.3098. [154] Angela K Wilson, Tanja van Mourik, and Thom H Dunning Jr. Gaussian basis sets for use in correlated molecular calculations. vi. sextuple zeta correlation consistent basis sets for boron through neon. Comput. Theor. Chem., 388:339–349, 1996. [155] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V . Barone, G. A. Petersson, H. Nakatsuji, X. Li, M. Caricato, A. V . Marenich, J. Bloino, B. G. Janesko, R. Gomperts, B. Mennucci, H. P. Hratchian, J. V . Ortiz, A. F. Izmaylov, J. L. Sonnenberg, D. Williams-Young, F. Ding, F. Lipparini, F. Egidi, J. Goings, B. Peng, A. Petrone, T. Henderson, D. Ranasinghe, V . G. Zakrzewski, J. Gao, N. Rega, G. Zheng, W. Liang, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y . Honda, O. Kitao, H. Nakai, T. Vreven, K. Throssell, J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. J. Bearpark, J. J. Heyd, E. N. Brothers, K. N. Kudin, V . N. Staroverov, T. A. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, J. M. Millam, M. Klene, C. Adamo, R. Cammi, J. W. Ochterski, R. L. Martin, K. Morokuma, O. Farkas, J. B. Foresman, and D. J. Fox. Gaussian˜16 Revision C.01, 2016. Gaussian Inc. Wallingford CT. [156] Jones de Andrade, Elvis S B¨ oes, and Hubert Stassen. Computational study of room temper- ature molten salts composed by 1-alkyl-3-methylimidazolium cations force-field proposal and validation. J. Phys. Chem. B, 106(51):13344–13351, 2002. [157] Thomas Fox and Peter A Kollman. Application of the resp methodology in the parametriza- tion of organic solvents. J. Phys. Chem. B, 102(41):8070–8079, 1998. [158] Craig A Gough, Stephen E Debolt, and Peter A Kollman. Derivation of fluorine and hydro- gen atom parameters using liquid simulations. J. Comput. Chem., 13(8):963–970, 1992. [159] Stephen L Mayo, Barry D Olafson, and William A Goddard. Dreiding: a generic force field for molecular simulations. J. Phys. Chem., 94(26):8897–8909, 1990. [160] Seonghoon Kim, Jumin Lee, Sunhwan Jo, Charles L. Brooks III, Hui Sun Lee, and Wonpil Im. Charmm-gui ligand reader and modeler for charmm force field generation of small 78 molecules. J. Comput. Chem., 38(21):1879–1886, 2017. doi: 10.1002/jcc.24829. URL https://onlinelibrary.wiley.com/doi/abs/10.1002/jcc.24829. [161] Karina Shimizu, Dimitrios Almantariotis, Margarida F Costa Gomes, Agilio AH Padua, and Jos´ e N Canongia Lopes. Molecular force field for ionic liquids v: Hydroxyethylimida- zolium, dimethoxy-2-methylimidazolium, and fluoroalkylimidazolium cations and bis (flu- orosulfonyl) amide, perfluoroalkanesulfonylamide, and fluoroalkylfluorophosphate anions. J. Phys. Chem. B, 114(10):3592–3600, 2010. [162] Akihiro Noda, Kikuko Hayamizu, and Masayoshi Watanabe. Pulsed-gradient spin- echo 1h and 19f nmr ionic diffusion coefficient, viscosity, and ionic conductivity of non- chloroaluminate room-temperature ionic liquids. J. Phys. Chem. B, 105(20):4603–4610, 2001. [163] Hayley A Every, Andrea G Bishop, Douglas R MacFarlane, Greger Or¨ add, and Maria Forsyth. Transport properties in a family of dialkylimidazolium ionic liquids. Phys. Chem. Chem. Phys., 6(8):1758–1765, 2004. [164] S´ ergio M Urahata and Mauro CC Ribeiro. Single particle dynamics in ionic liquids of 1-alkyl-3-methylimidazolium cations. J. Chem. Phys., 122(2):024511, 2005. [165] Carlos Rey-Castro, AL Tormo, and Lourdes F Vega. Effect of the flexibility and the anion in the structural and transport properties of ethyl-methyl-imidazolium ionic liquids. Fluid Ph. Equilibria, 256(1-2):62–69, 2007. [166] Carlos Rey-Castro and Lourdes F Vega. Transport properties of the ionic liquid 1-ethyl-3- methylimidazolium chloride from equilibrium molecular dynamics simulation. the effect of temperature. J. Phys. Chem. B, 110(29):14426–14435, 2006. [167] MH Kowsari, Saman Alavi, Mahmud Ashrafizaadeh, and Bijan Najafi. Molecular dynamics simulation of imidazolium-based ionic liquids. i. dynamics and diffusion coefficient. J. Chem. Phys., 129(22):224508, 2008. [168] Hendrik Heinz, RA Vaia, BL Farmer, and RR Naik. Accurate simulation of surfaces and interfaces of face-centered cubic metals using 12- 6 and 9- 6 lennard-jones potentials. J. Phys. Chem. C, 112(44):17281–17290, 2008. [169] Hendrik Heinz, Tzu-Jen Lin, Ratan Kishore Mishra, and Fateme S Emami. Thermodynam- ically consistent force fields for the assembly of inorganic, organic, and biological nanos- tructures: the interface force field. Langmuir, 29(6):1754–1765, 2013. [170] Mark James Abraham, Teemu Murtola, Roland Schulz, Szil´ ard P´ all, Jeremy C Smith, Berk Hess, and Erik Lindahl. Gromacs: High performance molecular simulations through multi- level parallelism from laptops to supercomputers. Softwarex, 1:19–25, 2015. [171] Szil´ ard P´ all, Mark James Abraham, Carsten Kutzner, Berk Hess, and Erik Lindahl. Tackling exascale software challenges in molecular dynamics simulations with gromacs. In Interna- tional conference on exascale applications and software, pages 3–27. Springer, 2014. 79 [172] Sander Pronk, Szil´ ard P´ all, Roland Schulz, Per Larsson, P¨ ar Bjelkmar, Rossen Apostolov, Michael R Shirts, Jeremy C Smith, Peter M Kasson, David van der Spoel, et al. Gromacs 4.5: a high-throughput and highly parallel open source molecular simulation toolkit. Bioin- formatics, 29(7):845–854, 2013. [173] Berk Hess, Carsten Kutzner, David Van Der Spoel, and Erik Lindahl. Gromacs 4: algorithms for highly efficient, load-balanced, and scalable molecular simulation. J. Chem. Theory Comput., 4(3):435–447, 2008. [174] David Van Der Spoel, Erik Lindahl, Berk Hess, Gerrit Groenhof, Alan E Mark, and Her- man JC Berendsen. Gromacs: fast, flexible, and free. J. Comput. Chem., 26(16):1701–1718, 2005. [175] Erik Lindahl, Berk Hess, and David Van Der Spoel. Gromacs 3.0: a package for molecular simulation and trajectory analysis. J. Mol. Model., 7(8):306–317, 2001. [176] Herman JC Berendsen, David van der Spoel, and Rudi van Drunen. Gromacs: a message- passing parallel molecular dynamics implementation. Comput. Phys. Commun., 91(1-3): 43–56, 1995. [177] T Darden, D York, and L Pedersen. An n.log (n) method for ewald sums in large systems. J. Chem. Phys., 98:10089–10092, 1993. [178] Berk Hess, Henk Bekker, Herman JC Berendsen, and Johannes GEM Fraaije. Lincs: a linear constraint solver for molecular simulations. J. Comput. Chem., 18(12):1463–1472, 1997. [179] Berk Hess. P-lincs: A parallel linear constraint solver for molecular simulation. J. Chem. Theory Comput., 4(1):116–122, 2008. [180] Herman JC Berendsen, JPM van Postma, Wilfred F van Gunsteren, ARHJ DiNola, and Jan R Haak. Molecular dynamics with coupling to an external bath. J. Chem. Phys., 81(8): 3684–3690, 1984. [181] Sh¯ uichi Nos´ e. A molecular dynamics method for simulations in the canonical ensemble. Mol. Phys., 52(2):255–268, 1984. [182] William G Hoover. Canonical dynamics: Equilibrium phase-space distributions. Phys. Rev. A, 31(3):1695, 1985. [183] Michele Parrinello and Aneesur Rahman. Polymorphic transitions in single crystals: A new molecular dynamics method. J. Appl. Phys., 52(12):7182–7190, 1981. [184] Shuichi Nos´ e and ML Klein. Constant pressure molecular dynamics for molecular systems. Mol. Phys., 50(5):1055–1076, 1983. [185] Sayan Bagchi, Stephen D. Fried, and Steven G. Boxer. A solvatochromic model calibrates nitriles’ vibrational frequencies to electrostatic fields. J. Am. Chem. Soc., 134(25):10373– 10376, 2012. doi: 10.1021/ja303895k. URL https://doi.org/10.1021/ja303895k. PMID: 22694663. 80 [186] Seonghoon Kim, Jumin Lee, Sunhwan Jo, Charles L. Brooks III, Hui Sun Lee, and Wonpil Im. Charmm-gui ligand reader and modeler for charmm force field generation of small molecules. J. Comput. Chem., 38(21):1879–1886, 2017. doi: 10.1002/jcc.24829. URL https://onlinelibrary.wiley.com/doi/abs/10.1002/jcc.24829. [187] Tomoyasu Mani and David C. Grills. Nitrile vibration reports induced electric field and delocalization of electron in the charge-transfer state of aryl nitriles. J. Phys. Chem. A, 122(37):7293–7300, 2018. doi: 10.1021/acs.jpca.8b08025. URL https://doi.org/10. 1021/acs.jpca.8b08025. PMID: 30141944. [188] Nirmalya Ballav, Bj¨ orn Sch¨ upbach, Ole Dethloff, Peter Feulner, Andreas Terfort, and Michael Zharnikov. Direct probing molecular twist and tilt in aromatic self-assembled monolayers. J. Am. Chem. Soc., 129(50):15416–15417, 2007. doi: 10.1021/ja0751882. URLhttps://doi.org/10.1021/ja0751882. PMID: 18041835. [189] Daniel J. Aschaffenburg and Richard S. Moog. Probing hydrogen bonding environ- ments: Solvatochromic effects on the cn vibration of benzonitrile. J. Phys. Chem. B, 113 (38):12736–12743, 2009. doi: 10.1021/jp905802a. URL https://doi.org/10.1021/ jp905802a. PMID: 19711975. [190] Paul E Ohno, HanByul Chang, Austin P Spencer, Yangdongling Liu, Mavis D Boamah, Hong-fei Wang, and Franz M Geiger. Beyond the gouy-chapman model with heterodyne- detected second harmonic generation. J. Phys. Chem. Lett., 2019. [191] Katherine A Willets. Super-resolution imaging of sers hot spots. Chem. Soc. Rev., 43(11): 3854–3864, 2014. [192] Clare E Harvey and Bert M Weckhuysen. Surface-and tip-enhanced raman spectroscopy as operando probes for monitoring and understanding heterogeneous catalysis. Catal. Lett., 145(1):40–57, 2015. [193] Naama Karton-Lifshin, Itay Presiado, Yuval Erez, Rinat Gepshtein, Doron Shabat, and Dan Huppert. Ultrafast excited-state intermolecular proton transfer of cyanine fluorochrome dyes. The Journal of Physical Chemistry A, 116(1):85–92, 2012. [194] Yi Liao. Design and applications of metastable-state photoacids. Accounts of chemical research, 50(8):1956–1964, 2017. [195] Gaoyuan Zhao and Ting Wang. Stereoselective synthesis of 2-deoxyglycosides from glycals by visible-light-induced photoacid catalysis. Angewandte Chemie International Edition, 57 (21):6120–6124, 2018. [196] William White, Christopher D Sanborn, Ronald S Reiter, David M Fabian, and Shane Ardo. Observation of photovoltaic action from photoacid-modified nafion due to light-driven ion transport. Journal of the American Chemical Society, 139(34):11726–11733, 2017. [197] Simon Luo, William White, Joseph M Cardon, and Shane Ardo. Clarification of mech- anisms of protonic photovoltaic action initiated by photoexcitation of strong photoacids 81 covalently bound to hydrated nafion cation-exchange membranes wetted by aqueous elec- trolytes. Energy & Environmental Science, 14(9):4961–4978, 2021. [198] Benjamin K Keitz and Robert H Grubbs. A tandem approach to photoactivated olefin metathesis: combining a photoacid generator with an acid activated catalyst. Journal of the American Chemical Society, 131(6):2038–2039, 2009. [199] Pawel J Serafinowski and Peter B Garland. Novel photoacid generators for photodirected oligonucleotide synthesis. Journal of the American Chemical Society, 125(4):962–965, 2003. [200] Robert H Crabtree. The organometallic chemistry of the transition metals. John Wiley & Sons, 2009. [201] Ashim Nandi and Sebastian Kozuch. History and future of dative bonds. Chemistry–A European Journal, 26(4):759–772, 2020. [202] Daniel J Parks and Warren E Piers. Tris (pentafluorophenyl) boron-catalyzed hydrosilation of aromatic aldehydes, ketones, and esters. Journal of the American Chemical Society, 118 (39):9440–9441, 1996. [203] Douglas W Stephan. Frustrated lewis pairs. Journal of the American Chemical Society, 137 (32):10018–10032, 2015. [204] PC Stair. The concept of lewis acids and bases applied to surfaces. Journal of the American Chemical Society, 104(15):4044–4052, 1982. [205] Horia Metiu, Steve Cr` etian, Zhenpeng Hu, Bo Li, and XiaoYing Sun. Chemistry of lewis acid–base pairs on oxide surfaces. The Journal of Physical Chemistry C, 116(19):10439– 10450, 2012. [206] Gianmario Martra. Lewis acid and base sites at the surface of microcrystalline tio2 anatase: relationships between surface morphology and chemical behaviour. Applied Catalysis A: General, 200(1-2):275–285, 2000. [207] R Snyder and AC Testa. Influence of electron-donor-acceptor complexation on electronic relaxation of quinoline. The Journal of Physical Chemistry, 88(24):5948–5950, 1984. [208] Naoki Ando, Takuya Yamada, Hiroki Narita, Niels N Oehlmann, Matthias Wagner, and Shigehiro Yamaguchi. Boron-doped polycyclic π-electron systems with an antiaromatic borole substructure that forms photoresponsive b–p lewis adducts. Journal of the American Chemical Society, 143(26):9944–9951, 2021. [209] Andrey Y Khalimon, Warren E Piers, James M Blackwell, David J Michalak, and Masood Parvez. A photo lewis acid generator (phlag): Controlled photorelease of b (c6f5) 3. Journal of the American Chemical Society, 134(23):9601–9604, 2012. [210] S Joel Cassidy, Ian Brettell-Adams, Louis E McNamara, Mallory F Smith, Michael Bautista, Hongda Cao, Monica Vasiliu, Deidra L Gerlach, Fengrui Qu, Nathan I Hammer, et al. Bo- ranes with ultra-high stokes shift fluorescence. Organometallics, 37(21):3732–3741, 2018. 82 [211] Kyohei Matsuo, Shohei Saito, and Shigehiro Yamaguchi. Photodissociation of b–n lewis adducts: a partially fused trinaphthylborane with dual fluorescence. Journal of the American Chemical Society, 136(36):12580–12583, 2014. [212] No¨ el Boens, V olker Leen, and Wim Dehaen. Fluorescent indicators based on bodipy. Chem- ical Society Reviews, 41(3):1130–1172, 2012. [213] Joshua A Snyder, Peter Gr¨ uninger, Holger F Bettinger, and Arthur E Bragg. Excited-state deactivation pathways and the photocyclization of bn-doped polyaromatics. The Journal of Physical Chemistry A, 121(27):5136–5146, 2017. [214] Heidi L van de Wouw and Rebekka S Klausen. Bn polystyrenes: Emerging optical materials and versatile intermediates. The Journal of organic chemistry, 84(3):1117–1125, 2019. [215] Th F¨ orster. Elektrolytische dissoziation angeregter molek¨ ule. Zeitschrift f¨ ur Elektrochemie und angewandte physikalische Chemie, 54(1):42–46, 1950. [216] Maria Sittig, Jessica C Tom, Johanna K Elter, Felix H Schacher, and Benjamin Dietzek. Quinoline photobasicity: Investigation within water-soluble light-responsive copolymers. Chemistry–A European Journal, 27(3):1072–1079, 2021. [217] Zbigniew R Grabowski and Wiesława Rubaszewska. Generalised f¨ orster cycle. thermody- namic and extrathermodynamic relationships between proton transfer, electron transfer and electronic excitation. Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases, 73:11–28, 1977. [218] Zbigniew R Grabowski. Generalized f¨ orster cycle applied to coordination compounds. Jour- nal of Luminescence, 24:559–562, 1981. [219] D Bradley G Williams and Michelle Lawton. Drying of organic solvents: quantitative eval- uation of the efficiency of several desiccants. The Journal of organic chemistry, 75(24): 8351–8354, 2010. [220] Shuhei Tsumura, Kazuki Ohira, Keiichi Imato, and Yousuke Ooyama. Development of optical sensor for water in acetonitrile based on propeller-structured bodipy-type pyridine– boron trifluoride complex. RSC Advances, 10(56):33836–33843, 2020. [221] CM O’Donnell, GA Knesel, TS Spencer, and FR Stermitz. Substitution effects on the emissive properties of n-heteroaromatics. i. substituted quinolines. The Journal of Physical Chemistry, 74(20):3555–3559, 1970. [222] Noboru Mataga and Shizuyo Tsuno. Hydrogen bonding effect on the fluorescence of some nitrogen heterocycles. i. Bulletin of the Chemical Society of Japan, 30(4):368–374, 1957. [223] Sophya F Alamudun, Kyle Tanovitz, April Fajardo, Kaitlind Johnson, Andy Pham, Tina Jamshidi Araghi, and Andrew S Petit. Structure–photochemical function relationships in nitrogen-containing heterocyclic aromatic photobases derived from quinoline. J. Phys. Chem. A, 124(13):2537–2546, 2020. 83 [224] Jeng-Da Chai and Martin Head-Gordon. Long-range corrected hybrid density functionals with damped atom–atom dispersion corrections. Physical Chemistry Chemical Physics, 10 (44):6615–6620, 2008. [225] Paul R Horn, Yuezhi Mao, and Martin Head-Gordon. Probing non-covalent interactions with a second generation energy decomposition analysis using absolutely localized molecular orbitals. Physical Chemistry Chemical Physics, 18(33):23067–23079, 2016. [226] Christian Laurence and Jean-Franc ¸ois Gal. Lewis basicity and affinity scales: data and measurement. John Wiley & Sons, 2009. [227] Yirong Mo and Jiali Gao. Polarization and charge-transfer effects in lewis acid- base com- plexes. The Journal of Physical Chemistry A, 105(26):6530–6536, 2001. [228] John P Wrass, Daniel Sadowsky, Kaitlin M Bloomgren, Christopher J Cramer, and James A Phillips. Quantum chemical and matrix-ir characterization of ch 3 cn–bcl 3: a complex with two distinct minima along the b–n bond potential. Physical Chemistry Chemical Physics, 16(31):16480–16491, 2014. [229] William Humphrey, Andrew Dalke, and Klaus Schulten. VMD – Visual Molecular Dynam- ics. Journal of Molecular Graphics, 14:33–38, 1996. [230] Tomas Plachetka. Pov ray: persistence of vision parallel raytracer. In Proc. of Spring Conf. on Computer Graphics, Budmerice, Slovakia, volume 123, page 129, 1998. [231] John R Platt. Classification of spectra of cata-condensed hydrocarbons. The Journal of chemical physics, 17(5):484–495, 1949. [232] Suresh Mummadi, Daniel K Unruh, Jiyang Zhao, Shuhua Li, and Clemens Krempner. “in- verse” frustrated lewis pairs–activation of dihydrogen with organosuperbases and moderate to weak lewis acids. Journal of the American Chemical Society, 138(10):3286–3289, 2016. [233] Suresh Mummadi, Amandeep Brar, Guoqiang Wang, Dustin Kenefake, Rony Diaz, Daniel K Unruh, Shuhua Li, and Clemens Krempner. “inverse” frustrated lewis pairs: An inverse flp approach to the catalytic metal free hydrogenation of ketones. Chemistry–A European Journal, 24(62):16526–16531, 2018. [234] Haiyan Fan, Jeffrey K Eliason, C Diane Moliva A, Jason L Olson, Scott M Flancher, MW Gealy, and Darin J Ulness. Halogen bonding in iodo-perfluoroalkane/pyridine mix- tures. The Journal of Physical Chemistry A, 113(51):14052–14059, 2009. [235] Marco Saccone, Valentina Dichiarante, Alessandra Forni, Alexis Goulet-Hanssens, Gabriella Cavallo, Jaana Vapaavuori, Giancarlo Terraneo, Christopher J Barrett, Giuseppe Resnati, Pierangelo Metrangolo, et al. Supramolecular hierarchy among halogen and hy- drogen bond donors in light-induced surface patterning. Journal of Materials Chemistry C, 3(4):759–768, 2015. [236] Sheila P Ewing, Daniel Lockshon, and William P Jencks. Mechanism of cleavage of carba- mate anions. Journal of the American Chemical Society, 102(9):3072–3084, 1980. 84 [237] S´ ergio M Urahata and Mauro CC Ribeiro. Structure of ionic liquids of 1-alkyl-3- methylimidazolium cations: A systematic computer simulation study. J. Chem. Phys., 120 (4):1855–1863, 2004. [238] William L Jorgensen, Jeffry D Madura, and Carol J Swenson. Optimized intermolecular potential functions for liquid hydrocarbons. Journal of the American Chemical Society, 106 (22):6638–6646, 1984. [239] Y Shim, MY Choi, and Hyung J Kim. A molecular dynamics computer simulation study of room-temperature ionic liquids. i. equilibrium solvation structure and free energetics. J. Chem. Phys., 122(4):044510, 2005. [240] B Th Thole. Molecular polarizabilities calculated with a modified dipole interaction. Chem- ical Physics, 59(3):341–350, 1981. [241] Alessandro Bagno, Fabio D’Amico, and Giacomo Saielli. Computer simulation of diffu- sion coefficients of the room-temperature ionic liquid [bmim][bf4]: Problems with classical simulation techniques. Journal of molecular liquids, 131:17–23, 2007. [242] Michel Picquet, Igor Tkatchenko, Immacolata Tommasi, Peter Wasserscheid, and J¨ org Zim- mermann. Ionic liquids, 3. synthesis and utilisation of protic imidazolium salts in homoge- neous catalysis. Advanced Synthesis & Catalysis, 345(8):959–962, 2003. [243] Oleg Borodin. Polarizable force field development and molecular dynamics simulations of ionic liquids. The Journal of Physical Chemistry B, 113(33):11463–11478, 2009. ISSN 1520-6106. doi: 10.1021/jp905220k. [244] Mark B Shiflett and A Yokozeki. Liquid- liquid equilibria in binary mixtures of 1, 3- propanediol+ ionic liquids [bmim][pf6],[bmim][bf4], and [emim][bf4]. Journal of Chemical & Engineering Data, 52(4):1302–1306, 2007. [245] Andreas Nazet, Sophia Sokolov, Thomas Sonnleitner, Takashi Makino, Mitsuhiro Kanakubo, and Richard Buchner. Densities, viscosities, and conductivities of the imi- dazolium ionic liquids [emim][ac],[emim][fap],[bmim][beti],[bmim][fsi],[hmim][tfsi], and [omim][tfsi]. Journal of Chemical & Engineering Data, 60(8):2400–2411, 2015. [246] Shiro Seki, Nobuyuki Serizawa, Kikuko Hayamizu, Seiji Tsuzuki, Yasuhiro Umebayashi, Katsuhito Takei, and Hajime Miyashiro. Physicochemical and electrochemical proper- ties of 1-ethyl-3-methylimidazolium tris (pentafluoroethyl) trifluorophosphate and 1-ethyl- 3-methylimidazolium tetracyanoborate. Journal of The Electrochemical Society, 159(7): A967, 2012. [247] Bernard Valeur and M´ ario Nuno Berberan-Santos. Molecular fluorescence: principles and applications. John Wiley & Sons, 2012. [248] Keiji Morokuma and Kazuo Kitaura. Energy decomposition analysis of molecular inter- actions. In Chemical applications of atomic and molecular electrostatic potentials, pages 215–242. Springer, 1981. 85 [249] Lili Zhao, Moritz von Hopffgarten, Diego M Andrada, and Gernot Frenking. Energy de- composition analysis. Wiley Interdisciplinary Reviews: Computational Molecular Science, 8(3):e1345, 2018. [250] Yuezhi Mao, Matthias Loipersberger, Paul R Horn, Akshaya Das, Omar Demerdash, Daniel S Levine, Srimukh Prasad Veccham, Teresa Head-Gordon, and Martin Head-Gordon. From intermolecular interaction energies and observable shifts to component contributions and back again: A tale of variational energy decomposition analysis. Annual review of physical chemistry, 72:641–666, 2021. [251] Leonardo J Duarte, Wagner E Richter, Roy E Bruns, and Paul LA Popelier. Electrostatics explains the reverse lewis acidity of bh3 and boron trihalides: Infrared intensities and a relative energy gradient (reg) analysis of iqa energies. The Journal of Physical Chemistry A, 125(39):8615–8625, 2021. [252] William R Hamilton. Theory of quaternions. Proceedings of the Royal Irish Academy (1836-1869), 3:1–16, 1844. [253] Rosanna Heise and Bruce A MacDonald. Quaternions and motion interpolation: A tutorial. New advances in computer graphics, pages 229–243, 1989. 86 Appendices A Supporting Information for: Electrochemical Control of Molecular Polarization 87 A.1 Substituents, Hammett parameters and Nitrile frequencies Table 5.1: Substituents used, respective Hammett parameters and Nitrile frequencies Substituent Hammett parameter (σ p ) Nitrile frequency (ω CN )(in cm − 1 ) -(NMe 2 ) -0.83 2212 -NH 2 -0.66 2212 -OH -0.37 2224 -OMe -0.27 2225 -CH 3 -0.17 2228 -OPh -0.03 2228 -H 0 2229 -F 0.06 2232 -B(OH) 2 0.12 2223.7 -SH 0.15 2228 -I 0.18 2230 -Cl 0.23 2232 -Br 0.23 2231 -OCOCH 3 0.31 2230.5 -COOH 0.45 2229 -COCH 3 0.50 2231.8 -CF 3 0.54 2234 -CN 0.66 2233 -NO 2 0.78 2235 -SO 2 Cl 1.11 2234 88 A.2 Linewidths with changingσ ρ andω CN Figure 5.1: (a) Plot of the widths of Lorentzians (obtained from fits) as a function of applied voltage. (b) Full width half maxima vs Hammett parameter Computational Data All the calculations were obtained through Q-Chem,using 6-31g(d,p) basis and B3LYP. Figure 5.2: (a) Dipole moment of the molecule as a function of Applied Field.(b) Dipole moment of the molecule as a function of Hammett parameter. The two plots were fit to a linear equation of the form y= p 1 x+ p 2 . The fit results were as follows (± was obtained from the 95% confidence interval): 89 µ vs Applied Field: p 1 =− 0.05624± 0.00023 p 2 = 4.555± 0.011 µ =− 0.05624F+ 4.555 µ vs Hammett Parameter: p 1 =− 4.34± 1.04 p 2 = 4.442± 0.51 µ =− 4.34F+ 4.44 A.3 Voltage Dependent SFG data Figure 5.3: Representative SFG spectra at different voltages. 90 (a) (b) Figure 5.4: (a) V oltage (with respect to Ag/AgCl) vs Time. (b) Current vs Time. The error bars in Figure 4b. in main text denote the standard deviations. The error for nitrile stretch from -1.2 V to +0.6 V includes fluctuations over different days of data collection and slight hysteresis observed due to electrochemical current in the negative potential regime. The error for the nitrile stretch at -1.4 V includes the 95 % confidence interval since we have only two data points corresponding to forward and reverse scan. Table 5.2: Fit Params. with equation: I SFG (ω)= A NR e iφ + B CN ω− ω CN +iΓ CN 2 exp − (ω− ω g) 2 σ 2 g Mean 95 % Conf. Int. (± ) Voltage Parameter Value from fits (avg.) 0V A NR 5.48 0.05 B CN 16.13 1.27 Γ CN 5.18 0.48 φ 0.482 0.081 ω CN 2228.8 0.48 σ g 102.2 1.7 ω g 2223.9 1.2 0.2V A NR 5.17 0.05 91 Fit Parameters Cont. Voltage Parameter Average 95 % Con. Int. (± ) B CN 15.74 1.27 Γ CN 4.88 0.43 φ 0.639 0.074 ω CN 2229.4 0.44 σ g 104.6 1.7 ω g 2223.8 1.2 0.4V A NR 5.08 0.08 B CN 12.90 2.21 Γ CN 4.04 0.86 φ 0.654 0.173 ω CN 2230.4 0.85 σ g 110.8 3.7 ω g 2225.0 2.6 0.6V A NR 5.58 0.86 B CN 6.33 19.27 Γ CN 4.05 18.11 φ 0.654 3.29 ω CN 2230.9 16.25 σ g 112.8 34.4 ω g 2223.8 24.2 0.6V A NR 5.77 0.03 B CN 3.80 0.69 Γ CN 2.84 1.09 92 Fit Parameters Cont. Voltage Parameter Average 95 % Con. Int. (± ) φ 0.543 0.229 ω CN 2231.4 0.75 σ g 109.6 1.5 ω g 2240.5 1.1 0.4V A NR 4.99 0.04 B CN 7.36 1.04 Γ CN 4.19 0.72 φ 0.490 0.14 ω CN 2230.2 0.74 σ g 114.7 2.0 ω g 2242.1 1.4 0.2V A NR 4.59 0.03 B CN 12.10 0.92 Γ CN 5.18 0.43 φ 0.589 0.076 ω CN 2229.6 0.47 σ g 106.4 1.7 ω g 2237.1 1.1 0V A NR 4.56 0.03 B CN 14.92 0.93 Γ CN 5.60 0.38 φ 0.499 0.064 ω CN 2229.1 0.41 93 Fit Parameters Cont. Voltage Parameter Average 95 % Con. Int. (± ) σ g 102.1 1.6 ω g 2235.7 1.0 -0.2V A NR 4.25 0.03 B CN 15.18 0.79 Γ CN 5.87 0.34 φ 0.277 0.059 ω CN 2228.5 0.36 σ g 98.7 1.5 ω g 2235.7 1.0 -0.4V A NR 3.71 0.04 B CN 15.84 0.90 Γ CN 6.28 0.39 φ 0.048 0.068 ω CN 2228.0 0.40 σ g 104.7 2.1 ω g 2238.4 1.3 -0.6V A NR 3.22 0.04 B CN 19.94 0.92 Γ CN 6.79 0.32 φ -0.264 0.057 ω CN 2227.5 0.32 σ g 109.8 2.4 94 Fit Parameters Cont. Voltage Parameter Average 95 % Con. Int. (± ) ω g 2236.4 1.6 -0.8V A NR 2.78 0.04 B CN 25.34 1.15 Γ CN 7.68 0.36 φ -0.479 0.063 ω CN 2226.3 0.35 σ g 134.7 4.0 ω g 2230.6 2.8 -1V A NR 2.87 0.07 B CN 26.09 1.64 Γ CN 7.67 0.47 φ -0.492 0.084 ω CN 2224.0 0.47 σ g 119.8 5.2 ω g 2230.4 3.5 -1.2V A NR 2.91 0.02 B CN 27.19 0.54 Γ CN 8.22 0.08 φ -0.591 0.183 ω CN 2222.0 2.44 σ g 122.3 2.7 ω g 2237.4 3.5 -1.2V 95 Fit Parameters Cont. Voltage Parameter Average 95 % Con. Int. (± ) A NR 3.05 0.14 B CN 25.30 3.98 Γ CN 8.60 1.32 φ -0.629 0.192 ω CN 2221.9 1.25 σ g 117.8 9.1 ω g 2238.3 6.7 -1V A NR 3.07 0.05 B CN 21.51 1.29 Γ CN 7.97 0.48 φ -0.497 0.074 ω CN 2223.2 0.48 σ g 110.3 2.9 ω g 2240.0 2.2 -0.8V A NR 2.92 0.17 B CN 18.02 4.68 Γ CN 7.68 1.99 φ -0.373 0.306 ω CN 2224.0 1.96 σ g 108.3 11.6 ω g 2242.2 8.0 -0.6V A NR 3.04 0.05 B CN 16.80 1.51 96 Fit Parameters Cont. Voltage Parameter Average 95 % Con. Int. (± ) Γ CN 8.06 0.75 φ -0.321 0.104 ω CN 2225.7 0.74 σ g 111.9 3.2 ω g 2239.5 2.4 -0.4V A NR 3.34 0.04 B CN 12.24 1.16 Γ CN 7.29 0.76 φ -0.152 0.108 ω CN 2226.8 0.76 σ g 106.1 2.3 ω g 2238.4 1.7 -0.2V A NR 3.33 0.04 B CN 9.98 1.21 Γ CN 7.51 0.98 φ -0.026 0.131 ω CN 2228.0 1.02 σ g 108.2 2.3 ω g 2239.9 1.7 0V A NR 3.55 0.04 B CN 6.69 1.59 Γ CN 7.41 1.76 φ 0.093 0.182 97 Fit Parameters Cont. Voltage Parameter Average 95 % Con. Int. (± ) ω CN 2228.7 1.75 σ g 108.9 2.6 ω g 2243.6 1.8 98 B Supporting Information for: Electric Fields and Structure at the Ionic Liquid/Metal Interface B.1 Raw SFG Spectra of the 4-MBN Monolayer in the Presence of Different ILs Figure 5.5: SFG spectra of 4-MBN SAMs representative of the data reported in the main text (Figure Figure 3.3). Data from SFG experiments are in general complex and must be fitted to correctly extract spectral information. The black curve in the plot is of the 4-MBN monolayer on gold in the presence of air, and has been scaled for better visual reference. 99 B.2 Raw SERS Spectra of the 4-MBN Monolayer in the Presence of Different ILs Figure 5.6: Unprocessed SERS spectra of 4-MBN SAM for the data reported in the main text (Figure Figure 3.4). The dotted line emphasizes the nitrile stretch of the SAM with no ionic liquid added over the SERS substrate (shown in the black curve). Data reported in the main text are frequency shifts referenced for each SERS substrate to correct for variation over time and between substrates. 100 B.3 Linewidths of Extracted Lorenztians from SFG Spectra Figure 5.7: Averaged nitrile linewidth from SFG experiments plotted vs varying anion identity. We do not observe any clear dependence of the nitrile linewidth on the identity of the IL anion. 101 B.4 Temperature Dependent SFG Data ∆ν CN vs. Temperature: [EMIM] + [Br] − Figure 5.8: Plot of the change in 4-MBN CN center frequency in the presence of [EMIM] + [Br 4 ] − units of wavenumbers (cm − 1 ), measured as a function of temperature. This data is relevant because the MD simulations reported in this work were done at 400K to prevent freezing. The data show that the CN center frequency is invariant with respect to temperature in this ionic liquid - indicating that the higher simulation temperatures are no cause for concern. 102 B.5 FTIR Spectra of [EMIM] + [BF 4 ] − Before and After Microwave Purification Figure 5.9: These spectra (overlaid to emphasize their similarity across treatments) show that microwave purification does not result in damage to the [EMIM] + [BF 4 ] − IL. IR spectra were taken in an unmodified FTIR liquid cell (using a 25 µm spacer), similar to the one used for SFG. These spectra also show the reasonably low level of water in our ionic liquids after storage in dry air. The gold spectra is of [EMIM] + [BF 4 ] − after exposure to atmosphere for 10 minutes. The bands growing in after this time are likely from water, though they are shifted from the pure water values. 103 B.6 Summary of Key Data The following tables organize data presented as plots in the main text and SI. Our fitting equation used to interpret SFG spectra includes more values than just the nitrile center frequency and nitrile linewidth. The other parameters include the intensity of the nitrile vibration, non-resonant back- ground (fit as a Gaussian) σ and intensity, and phase offset between the two signals. These values are highly dependent on a wide range of experimental parameters, and for the purposes of this work we have not interpreted their values in the context of the ionic liquid solvation environment. Therefore, we only include the values we have presented in the text and SI,∆ν CN andΓ CN . Table 5.3: Summary of anion series from SFG data reported in the main text. We have included the estimated charge density, average values∆ν CN , average valuesΓ CN , along with error bars for measurements and average R 2 values for the fits. IL Name 1/V (nm − 3 ) (∆ν CN )(cm − 1 ) (Γ CN )(cm − 1 ) R 2 [EMIM] + [FAP] − 3.226 -2.31± 0.56 7.00± 0.55 0.9808 [EMIM] + [DMP] − 6.578 -4.43± 0.33 7.54± 0.07 0.9888 [EMIM] + [EtSO 4 ] − 7.299 -2.79± 0.37 5.64± 0.67 0.9838 [EMIM] + [AcO] − 11.765 -6.66± 1.75 5.35± 0.16 0.9723 [EMIM] + [BF 4 ] − 13.889 -4.35± 0.74 7.31± 0.37 0.9861 [EMIM] + [Br] − 16.949 -12.53± 0.35 7.83± 3.6 0.9895 104 Table 5.4: Summary of anion series from SERS data reported in the main text. We have included the estimated charge density, measured values of ∆ν CN and Γ CN , along with error bars for mea- surements and R 2 values for the fits. IL Name 1/V (nm − 3 ) (∆ν CN SERS)(cm − 1 ) (Γ CN )(cm − 1 ) R 2 [EMIM] + [FAP] − 3.226 -2.16± 0.28 5.95± 0.49 0.9651 [EMIM] + [DMP] − 6.578 -2.11± 0.21 5.367± 0.311 0.9763 [EMIM] + [EtSO 4 ] − 7.299 − 2.17± 0.17 5.51± .025 0.9786 [EMIM] + [AcO] − 11.765 − 6.82± 0.23 4.975± 0.32 0.9620 [EMIM] + [BF 4 ] − 13.889 -5.04± 0.15 5.42± 0.30 0.9763 [EMIM] + [Br] − 16.949 -8.04± 0.44 6.36± 0.69 0.9209 105 (a) Realistic anions with[EMIM] + cations (b) Modified Cl – anions with[EMIM] + cations Figure 5.10: Calculated frequency shifts using electric field at nitrile carbon of 4-MBN monolayer on the gold slab in presence of different ILs. 106 (a) (b) Figure 5.11: (a) Symmetrized partial number density of representative atoms vs Z (b) Symmetrized charge density of representative atoms vs Z, for all realistic anions 107 (a) (b) Figure 5.12: (a) Symmetrized partial number density of representative atoms vs Z (b) Symmetrized charge density of representative atoms vs Z, for all model Cl – anions 108 B.7 Selection of Radial Cut-off (R cut ) for Electric Field Calculations The cut-off distance of 3.5 nm has been chosen using the scheme detailed below (supported by Figure Figure 5.13, Figure 5.14 and Figure 5.15). For three representative systems, namely [EMIM] + [AcO] − ,[EMIM] + [Cl] − , and[EMIM] + [FAP] − we ran the following calculations. 1. We calculated the electric field components centered at each nitrile nitrogen on a cut-off distance grid ranging from 0.5 nm to 4 nm with a grid-spacing of 0.05 nm. The electric field components calculated on one such nitrile nitrogen are shown in Figure Figure 5.15 (top panels). 2. Next, the electric field components were averaged over nitrile nitrogen positions. The av- erage and standard deviation of field components, plotted against R cut (Figure Figure 5.13) illustrate the degree of charge heterogeneity in the immediate IL environments experienced by the nitrile nitrogens. Qualitatively we observe that the fluctuations in electric field com- ponents diminish beyond R cut = 3.0 nm. 3. To achieve a safe cut-off distance, we further calculated the gradient of the electric field components with respect to the cut-off distances. (a) Figure Figure 5.15 (bottom three panels) shows the gradients of electric field compo- nents calculated at one nitrile nitrogen. At R cut = 3.5 nm, the average of gradient taken over three components for all three representative systems is 2.2× 10 − 2 V nm − 2 . A longer R cut in this case shows negligible improvement. (b) The gradients of mean and standard deviation of electric field components with respect to R cut are shown in Figure Figure 5.14. At R cut = 3.5 nm, average of gradients over components and three representative systems is 3.6× 10 − 2 V nm − 2 . Significant im- provement in this metric is also not seen with increased R cut . Further details regarding cut-off selection are provided in the description of Figure Figure 5.14. 109 4. Finally, we note that cut-off radii beyond 4.0 nm could lead to interference of IL charges from the opposite side of the slab. Therefore choosing an R cut = 3.5 nm was convenient to keep the implementation simple. (a) (b) (c) Figure 5.13: Convergence of electric field on a snapshot at t = 200 ns for representative systems. For each system, top panel is the averaged field components over 200 nitrile nitrogens. Bottom panel shows the standard deviation of electric field over the 200 nitrogens, as a function of R cut used for field calculations. Elelctric field calculations have been performed every 0.05 nm from R cut value of 0.5 nm to 4 nm. 110 (a) (b) (c) a.⟨·⟩ denotes average over 200 nitrile nitrogen centers.⟨F i ⟩ ′ = d⟨F i ⟩ dR cut , i={X,Y,Z} b. σ ′ i = dσ i dR cut , i={X,Y,Z} c. Calculated usingnumpy.gradient() d. Calculated usingscipy.Bspline(,k=3) e. 1 3 ∑ s 1 3 ∑ i ⟨F i ⟩ ′ s ,i={X,Y,Z},s={[EMIM] + [AcO] − ,[EMIM] + [Cl] − ,[EMIM] + [FAP] − } Figure 5.14: Gradient of spatial average a and gradient of standard deviation b for calculated electric field shown in Figure Figure 5.13. Gradient is calculated c with respect to R cut and are shown using dots. Furthermore, smoothed interpolations d of the gradients have been shown using overlaid lines. At R cut = 3.5 nm, average of gradients over components and three representative systems e is 3.6× 10 − 2 V nm − 2 . 111 (a) (b) (c) Figure 5.15: Components of electric field and their gradients calculated at a specific nitrile nitro- gen. At R cut = 3.5 nm, the average of gradients over components and three representative systems is 2.2× 10 − 2 V nm − 2 . 112 Components of Electric Field(V nm − 1 ) Figure 5.16: Histogram of electric field components calculated on 200 nitrile nitrogens every 50 ps from 100-200 ns, for ILs with realistic anions 113 Components of Electric Field(V nm − 1 ) Figure 5.17: Histogram of electric field components calculated on 200 nitrile nitrogens every 50 ps from 100-200 ns, for ILs with modified chloride anions 114 Table 5.5: System setups for functionalized gold surface - ionic liquid assembly a Ionic Liquid Average Box dimensions Number of Cations/Anions (nm× nm× nm) [EMIM] + [Br] − (4.23× 4.23× 19.71) 1170 [EMIM] + [BF 4 ] − (4.24× 4.24× 20.01) 1050 [EMIM] + [AcO] − (4.21× 4.21× 13.22) 600 [EMIM] + [DMP] − (4.23× 4.23× 20.09) 820 [EMIM] + [FAP] − (4.23× 4.23× 22.62) 550 [EMIM] + [Cl 1σ ] − (4.23× 4.23× 20.34) 1210 [EMIM] + [Cl 1.5σ ] − (4.25× 4.25× 24.94) 900 [EMIM] + [Cl 2σ ] − (4.25× 4.25× 24.93) 700 [EMIM] + [Cl 3σ ] − (4.76× 4.76× 28.25) 550 a. Production run 200 ns for all systems. Table 5.6: Densities and diffusion coefficients of pure ionic liquids a,b Ionic Liquid ρ m D + D − [EMIM] + [Br] − 1.34± 8.0× 10 − 4 0.51± 6.06× 10 − 2 0.34± 1.72× 10 − 3 [EMIM] + [BF 4 ] − 1.18± 6.0× 10 − 5 8.08± 1.50× 10 − 1 5.49± 1.35× 10 − 1 [EMIM] + [AcO] − 1.03± 6.0× 10 − 5 5.31± 5.50× 10 − 1 2.87± 1.60× 10 − 1 [EMIM] + [DMP] − 1.15± 6.0× 10 − 5 7.64± 1.50× 10 − 1 3.98± 2.60× 10 − 1 [EMIM] + [FAP] − 1.51± 1.6× 10 − 4 7.77± 3.90× 10 − 1 6.74± 7.00× 10 − 2 [EMIM] + [Cl 1σ ] − 1.02± 7.0× 10 − 5 0.62± 3.36× 10 − 2 0.35± 3.31× 10 − 2 [EMIM] + [Cl 1σ ] − (500K) 0.97± 5.0× 10 − 5 21.91± 1.02 17.51± 0.44 [EMIM] + [Cl 1.5σ ] − 0.61± 2.0× 10 − 5 3.23× 10 1 ± 0.04 2.07× 10 1 ± 0.79 [EMIM] + [Cl 2σ ] − 0.34± 2.0× 10 − 5 1.27× 10 2 ± 6.22 8.25× 10 1 ± 4.74 [EMIM] + [Cl 3σ ] − 0.10± 2.0× 10 − 5 6.71× 10 2 ± 8.96 4.81× 10 2 ± 1.30 a. The unit of diffusion coefficient is 10 − 11 m 2 s − 1 . Mass densityρ m is shown in gcm − 3 or 10 3 kgm − 3 b. Calculated using a 200-400 ns trajectory of only ionic liquid in a cubic box. Unless otherwise stated, temperature is 400 K. 115 (a) Field autocorrelation function for field at all 200 nitrile nitrogens for [EMIM] + [Br] − IL system (b) Average Field autocorrelation function over 200 nitrile nitrogens for [EMIM] + [Br] − IL system, error bars showed in yellow are standard deviations (c) Field autocorrelation function for field at all 200 nitrile nitrogens for for [EMIM] + [FAP] − IL system (d) Average Field autocorrelation function over 200 nitrile nitrogens for[EMIM] + [FAP] − IL system, error bars showed in yellow are standard deviations Figure 5.18: Field autocorrelation function for different IL systems 116 Table 5.7: Experimental and previously calculated densities and diffusion coefficients of ionic liquids Ionic Liquid Reference Force Field/Method ρ m (gcm − 3 ) D + D − [EMIM] + [Cl] − (400 K) 167 OPLS-AA/AMBER based 140 3.78± 0.74 2.59± 0.62 [EMIM] + [Cl] − (400 K) 164 OPLS 237;238 16.96± 1.15 9.36± 1.66 [EMIM] + [Cl] − (400 K) 165 OPLS 237;238 34± 0.5 19.9± 0.6 [EMIM] + [Cl] − (404 K) 166 Shim et al. 239 3.61± 0.64 1.43± 0.68 [EMIM] + [Cl] − (400 K) 134 AMBER based 134 9.6 5.8 [EMIM] + [Cl] − (400 K) 135 AMBER based 134 non-polarizable 1.174 5.1 4.8 [EMIM] + [Cl] − (400 K) 135 AMBER based 134 polarizable 240 1.177 14.9 15.5 [EMIM] + [Cl] − (323 K) 237 OPLS 237;238 1.14 a [MMIM] + [Cl] − (323 K) 237 OPLS 237;238 1.21 b [BMIM] + [Cl] − (323 K) 237 OPLS 237;238 1.08 [EMIM] + [Br] − (323 K) 237 OPLS 237;238 1.48 a [MMIM] + [Br] − (323 K) 237 OPLS 237;238 1.26 c [BMIM] + [BF 4 ] − (315 K) 241 NMR 242 D= 3.5± 0.1 [BMIM] + [BF 4 ] − (358 K) 241 Tokuda et al. 84 1.150 1.30 1.30 [EMIM] + [BF 4 ] − (393 K) 243 Polarizable 243 1.198± 0.6 38.9 32.1 c [EMIM] + [BF 4 ] − (393 K) 243 Experiment 244 1.206 c [EMIM] + [BF 4 ] − (333 K) 243 Experiment 244 13.5 11.6 c [EMIM] + [OAc] − (358 K) 245 Experiment 245 1.063 c [EMIM] + [FAP] − (358 K) 245 Experiment 245 1.638 c [EMIM] + [FAP] − (303 K) 246 PGSE-NMR 246 1.7020 3.09 1.64 a. 1,3-dimethylimidazolium cation abbreviated as[MMIM] + b. 1-n-butyl-3-methylimidazolium cation abbreviated as[BMIM] + c. Experimental, 117 C Supporting Information for: Can Bronsted Photobases Act as Lewis Photobases? C.1 Brønsted Photobasicity in Quinolines Figure 5.19: We have reproduced a figure from a prior paper published by our group, which fo- cused on the electronic origin of the Brønsted photobasic effect in 5-R-Quinolines. We observe strong parallels between this figure and the compuational data published in the main text (Figure Figure 4.5). 2 As discussed in our prior work 2 , the mechanism of reversible photobases(acids) is an increase (decrease) of electron density at the proton-accepting site, which stabilizes (destabilizes) the pro- tonated species. This study describes photobasicity in quinolines as applying to a more general definition of acids, like Lewis acids. Comparing Figure Figure 5.19 and Figure Figure 4.5, we observe that the energetics of the photobasic effect towards Brønsted acids and Lewis acids scales quite similarly with respect to electron donating/withdrawing substituents. To outline this charge displacement in a free quinoline, we have included Figure Figure 5.20, which shows a charge density difference map for free 5-methoxyquinoline. This figure clearly shows that the 5-MeoQ’s 118 nitrogen gains charge in the excited state. Further, the EDD plot shown in the main text (Figure Figure 4.3) shows that this increased charge density participates in binding with BF 3 . (a) (b) Figure 5.20: Panel A shows a schematic of electron displacement in 5-MeoQ between the ground and excited state. Panel B displays an electron density difference map between the ground and first excited state of 5-MeoQ. Blue isosurfaces represent volumes which lose charge density in the excited state, and red isosurfaces correspond to an increase in charge density. As discussed in the main text, we observe an marked increase in charge density at the quinoline’s nitrogen. This change in charge density stabilizes the binding of both protons and Lewis acids like BF 3 119 C.2 Method Dependence of Computational F¨ orster Cycle NH 2 OCH 3 H CF 3 CN Figure 5.21: This figure supplements the computational F ¨ orster cycle shown in Figure Figure 4.5. In order to determine the robustness and method dependence of our results, single point and TDDFT energies were recalculated using ωB97X-D/def2-tzvppd for the geometries used in the main text (calculated using ωB97X-D/def2-svpd). In addition, these energy calculations were performed with a PCM model set to the parameters for acetonitrile, the solvent used for the exper- imental work. The dotted lines are provided as a guide. Understanding the methodological dependence of our computational data will provide strength to the results discussed in the main text, and help to clarify further work that needs to be done in understanding Lewis interactions of photoacids and photobases. Figure Figure 5.21 displays the results of single point and TDDFT energy calculations (ωB97X-D/def2-tzvppd, acetonitrile PCM) performed on minimized structures calculated with ωB97X-D/def2-svpd. A similar method was used by Petit’s group in their theoretical study of Brønsted photobasicity 223 . In comparing the two data sets, we observe some differences in the relative B-N energy ordering between different 5-R-Quinolines, both in the ground and excited state. The relative trend and the overall photobasic effect are clearly consistent between the two datasets. This gives us confidence that our conclusions 120 are not specific to one method. We are interested to note that the inclusion of solvent effects did not markedly stabilize the excited quinoline-BF 3 complexes. Because the photobasic excitation in quinolines is similar to a charge-transfer excitation, we expected the solvent to greatly stabilize the adducts, particularly in the excited state, which would lead to an enhancement of the photobasic effect. In order to draw stonger conclusions on these results, a more systematic study of solvation effects on these complexes is required. 121 C.3 Electronic Spectra of 5-R-Quinolines and Their BF 3 Adducts 122 UV-vis of 5-Aminoquinoline-BF 3 and its BF 3 Adduct Figure 5.22: Displayed here is the UV-vis absorption spectra of 5-aminoquinoline and its BF 3 adduct. As opposed to to the other quinolines studied in this work, the aminoquinoline’s absorption blue-shifted from its unadducted form. This result is consistent with BF 3 binding to the free amino group and not the ring nitrogen. We rationalize this by noting that attaching a BF 3 to the amino group will result in a strong electron withdrawing effect, thereby increasing the L A state. Table 5.8: EDA results and calculated absorption energy for the 5-aminoquinolineBF 3 adduct with BF 3 bound to the amino group. Geometries and energy calculations were performed usingωB97X- D/def2-svpd. We observe a blue shift in the spectra relative to free 5-aminoquinoline, indicating that the amino-bound BF 3 acts as an electron withdrawing group R σ P EDA (kJ/mol) FROZ (kJ/mol) POL (kJ/mol) CT (kJ/mol) NH 2 Reverse -0.66 -200.7615 137.6036 -205.0884 -133.2767 NH 2 Reverse -0.66 Absorption (eV) 4.7099 123 UV-Vis and Fluorescence of 5-Methylquinoline and its BF 3 adduct Figure 5.23: Displayed here is the UV-vis and fluorescence spectra of 5-methylquinoline and its BF 3 adduct. The sharp rise in intensity in the dotted blue curve is due to a solvent Raman peak. UV-Vis and Fluorescence of Quinoline and its BF 3 adduct Figure 5.24: Displayed here is the UV-vis and fluorescence spectra of quinoline and its BF 3 adduct in dry acetonitrile. 124 UV-Vis and Fluorescence of 5-Bromoquinoline and its BF 3 adduct Figure 5.25: Displayed here is the UV-vis and fluorescence spectra of 5-bromoquinoline and its BF 3 adduct in dry acetonitrile. UV-Vis and Fluorescence of 5-Cyanoquinoline and its BF 3 adduct Figure 5.26: Displayed here is the UV-vis and fluorescence spectra of 5-cyanoquinoline and its BF 3 adduct in dry acetonitrile. 125 Tabulated 0-0 Energies for the 5-R-Quinolines(-BF 3 ) Studied in this Work As discussed in the main text, 0-0 energies corresponding to the 0-0 S0-S1 transition for the quinoline photobases (and their adducted counterparts) were estimated by fitting the spectra (where necessary) and finding the point of intersection between the normalized absorbance and emission traces. This method is typically considered to be a better estimate for 0-0 values between electronic transitions compared to simply the using absorption and emission maxima, according to Valeur 247 . Further helpful discussion can be found in the cited textbook. We have estimated the error in this procedure by assuming an uncertainty of± 4nm for all reported 0-0 gaps. Table 5.9: Experimental estimates for 0-0 energies between the ground and photobasic electronic states. 0-0 Energy Unadducted (nm) Adducted (nm) Unadducted (eV) Adducted (eV) 5MeOQ 339± 4 422± 4 3.657± 0.043 2.938± 0.028 5MeQ 320± 4 381± 4 3.874± 0.048 3.254± 0.034 Q 322± 4 354± 4 3.85± 0.048 3.502± 0.039 BrQ 323± 4 381± 4 3.838± 0.047 3.254± 0.034 CNQ 319± 4 339± 4 3.888± 0.049 3.657± 0.043 126 C.4 Additional View of 5-MeOQBF 3 EDD Map Figure 5.27: This is the same electron density difference map shown in Figure Figure 4.3, but viewed from a different angle to better accentuate the increase of electron density in the vicinity of the B-N bond. As a guide, note that the BF 3 group is towards the right of the image (the boron is colored pink and the fluorides are a light blue-green. 127 C.5 Summary of Computational Data The following tables summarize the results of calculations discussed in the main text. Raw EDA results, structural energies and calculated absorption/emission lines are provided below. Table 5.10: Direct output of EDA2 calculations of 5-R-Quinoline-BF 3 Adducts.Energies of the same geometries calculated using ωB97X-D/def2-tzvppd, along with a PCM, are reported below the break in the table. R σ P EDA (kJ/mol) FROZ (kJ/mol) POL (kJ/mol) CT (kJ/mol) SOLV (kJ/mol) NH 2 -0.66 -253.4989 123.0048 -232.7562 -143.7475 MeO -0.268 -261.2115 126.3485 -241.3347 -146.2253 H 0 -250.1975 126.2774 -232.313 -144.162 TFM 0.54 -239.4935 128.926 -226.3555 -142.064 CN 0.66 -235.5906 131.4034 -225.0359 -141.9581 NH 2 -0.66 -246.6331 96.1726 -224.2288 -138.7682 20.1914 MeO -0.268 -244.3892 99.6931 -225.0332 -138.4901 19.4409 H 0 -241.7365 102.6466 -224.029 -138.7136 18.3595 TFM 0.54 -229.8064 108.587 -217.5598 -136.6106 15.7771 CN 0.66 -225.9623 113.219 -216.9769 -136.4097 14.2052 128 Table 5.11: Energies (ωB97X-D/def2-svpd) corresponding to the optimized S0 and photobasic (PB) excited states for free and adducted quinolines. The differences between the S0 and PB energies for the free and adducted states are the 0-0 energies used in the F¨ orster cycle analysis (i.e., the enhancement in BF 3 binding energy in the excited state), which are reported in Table Table 5.12.Energies of the same geometries calculated usingωB97X-D/def2-tzvppd, along with a PCM, are reported below the break in the table. R Free Q E (Ha) Free Q PB E (Ha) QBF 3 E (Ha) QBF 3 PB E (Ha) NH2 -456.8177553 -456.6793401 -781.0975502 -780.9801694 MeO -515.9232994 -515.766436 -840.203298 -840.0665917 H -401.5108095 -401.3407481 -725.7899534 -725.6331418 TFM -738.2237718 -738.0537847 -1062.500012 -1062.336298 CN -493.6495839 -493.4827778 -817.9251259 -817.7673958 NH2 -457.3011039 -457.1682191 -781.9619223 -781.8481152 MeO -516.4682614 -516.3166849 -841.1280293 -840.9945778 H -401.9318121 -401.756541 -726.5912002 -726.4342909 TFM -739.0281866 -738.8557226 -1063.684031 -1063.520568 CN -494.1751647 -494.0085249 -818.8303463 -818.6718887 129 Table 5.12: Summary of computed absorption and emission lines for the S0-PB state transition. Geometries and energies were calculated using ωB97X-D/def2-svpd, and TD-DFT calculations were performed using TDA and triplet states were excluded. Energies of the same geometries calculated using ωB97X-D/def2-tzvppd, along with a PCM, are reported below the break in the table. R Q Abs (eV) Q Em(eV) Add Abs (eV) Add Em (eV) Free 0-0 (Ha) Add 0-0 (Ha) NH 2 4.0701 3.4538 3.535 2.8328 0.138415 0.117380 MeO 4.5862 3.9458 4.0782 3.3473 0.1568633 0.1367062 H 4.563 3.3146 4.6011 3.919 0.1700614 0.1568116 TFM 4.5114 3.2623 4.7245 4.1237 0.1699870 0.1637144 CN 4.4615 3.2002 4.5884 3.9822 0.1668060 0.1577300 NH2 3.9178 3.328 3.4441 2.8186 0.1328847678 0.113807137 MeO 4.4452 3.8058 3.9952 3.3147 0.1515764495 0.1334514197 H 4.7446 4.4889 4.5913 3.9514 0.1752710926 0.15690934 TFM 4.6886 4.5205 4.6832 4.1322 0.1724639168 0.1634631318 CN 4.6223 4.371 4.5732 4.0274 0.1666397754 0.1584576279 130 Table 5.13: This table displays the energies corresponding to the rehybridization/strain of BF 3 and quinolines in the adducted state. All energies reported in this table are from calculations performed at theωB97X-D/def2-svpd level of theory. As discussed in the main text, single point calculations were performed for isolated adduct fragments (without relaxation) to get a more accurate account- ing of the B-N bond energy. The geometry change in BF 3 from planar to pyramidal requires a great deal of energy, and default EDA procedures (which perform no geometry optimization) do not account for this. Mao and Head-Gordon have introduced a method called adiabatic EDA to better treat these cases 1 . Energies of the same geometries calculated usingωB97X-D/def2-tzvppd, along with a PCM, are reported below the break in the table. R-Group σ P BF 3 Strain BF 3 Relax Q Strain Q Relax B-N Bond E (ev) NH2 -0.66 -324.184839 -324.2321077 -456.81608 -456.81776 -1.295294327 MeO -0.268 -324.1820265 -324.2321077 -515.92146 -515.92331 -1.294221795 H 0 -324.1854636 -324.2321077 -401.50917 -401.51080 -1.279554624 TFM 0.54 -324.1867315 -324.2321077 -738.22207 -738.22376 -1.201300206 CN 0.66 -324.1873067 -324.2321077 -493.64799 -493.64960 -1.178884525 nh2 -0.66 -324.567763 -324.6138181 -457.3001656 -457.3011039 -1.277440617 meo -0.268 -324.5648624 -324.6138181 -516.4671812 -516.4682614 -1.143905694 h 0 -324.5683722 -324.6138181 -401.9307099 -401.9318121 -1.115163607 tfm 0.54 -324.5695833 -324.6138181 -739.0268936 -739.0281866 -1.142925856 cn 0.66 -324.5701412 -324.6138181 -494.1741269 -494.1751647 -1.125212007 131 Table 5.14: Results and associated energy values to determine the TEA-BF 3 binding energy. In an identical procedure for the quinoline data, the fragments (TEA and BF 3 ) of the adduct were optimized independently and the relaxation energy was subtracted from the raw EDA value. TEA EDA (eV) -2.942250876 BF 3 Strained (Ha) -324.1827398 BF 3 Relaxed (Ha) -324.2321077 ∆E BF 3 Relaxation (eV) 1.343300055 TEA Strained (Ha) -292.1261256 TEA Relaxed (Ha) -292.1342445 ∆E TEA Relaxation (eV) 0.2209151027 TEA EDA ad justed (ev) -1.378035718 With def2-tzvppd, PCM TEA EDA (eV) -2.847960367 BF 3 Strained (Ha) -324.5657248 BF 3 Relaxed (Ha) -324.6138181 ∆E BF 3 Relaxation (eV) 1.308619221 TEA Strained (Ha) -292.4332073 TEA Relaxed (Ha) -292.4386179 ∆E TEA Relaxation (eV) 0.1472209104 TEA EDA ad justed (ev) -1.392120235 132 D Further Details on the Computational Forster Cycle Applied to Lewis Adducts D.1 Background The theoretical work presented in Chapter 4 to create a computational Forster Cycle for quinolines and Lewis acids was created by adapting/synthesizing ideas from a few different sources in the literature. My overall goal was to calculate the excited state change in N-B bond energy of a quinoline-BF 3 adduct, and I used a combination of Time-Dependent Density Functional Theory (TDDFT) and energy decomposition analysis (EDA) to find that information. The procedure I developed is nothing terribly complicated, but explaining it in more detail here might help clarify the concepts for anyone looking to reproduce or extend my results. D.2 Comments on EDA The first computational method I’d like to introduce is called energy decomposition analysis (EDA). This procedure was first introduced by Morokuma, but many have further developed the idea, in- cluding the Head-Gordon group 248;225;1 . EDA is a fragment analysis method which computes the total interaction energy between two chemical fragments (i.e., the bonding energy), and then it partitions -decomposes- that energy into separate quantities. These quantities correspond to dif- ferent types of bonding energy - which I’ll explain in a moment - and essentially, they serve to provide a quantitative basis for the qualitative identification of bonds as dative, covalent, etc., or any mixture of the various types. So, when we intuitively identify bonding between an amine and carbon dioxide (an N-C bond) as reasonably dative but with some covalent character, EDA seeks to put some hard numbers on the relative contribution of these different flavors of bonding. Because EDA serves to classify types of bonds in a more human fashion, I find that it is very commonly applied to analyze dative-bonded fragments, as that classification actually applies to a variety of bond types which have distinct chemical origins 249 . However, it can be used on (seemingly) any 133 chemical interaction that can be separated into distinct fragments, including water dimers, DNA base pairing, radical complexes, for just a few examples 225 . The EDA routine I have the most familiarity with is the one developed by the Head-Gordon group(Absolutely localized molecular orbital EDA, or ALMO-EDA), which is the default EDA procedure used in the QChem quantum chemistry software package 225 . The very basic function of EDA is to partition the total interaction energy ∆E INT into separate, physically meaningful quantities: ∆E INT =∆E GD +∆E FRZ +∆E POL +∆E CT ∆E GD is the energy associated with geometric distortion of the fragments upon bonding (more on this later - most EDA methods freeze the atomic coordinates and therefore do not account for this energy). ∆E FRZ corresponds to the energy of the separated fragments if the electron densities are not allow to relax (they are “frozen”). ∆E POL is the electron polarization energy, the energy associated with intra-fragment electron reorganization (meaning, electron density can move within each fragment, but cannot move between fragments). The remaining energy,∆E CT is the charge transfer energy; if the total energy changes when electron density moves between fragments, it is assigned to this term. EDA methods created by various groups are differentiated by the way that they calculate these terms. I don’t have sufficient understanding to describe the methods at this level of detail, so I won’t make an attempt. A nice review of EDA and ALMO-EDA was pub- lished last year by the Head-Gordon group 250 , which I recommend looking at for a more detailed explanation. As I alluded to in the last paragraph, most EDA methods (including the default EDA in QChem) can be considered “vertical” EDA methods (vertical means that the nuclear coordinates are fixed for the calculation, identical to the way we think about electronic absorption and emission processes). This means that if the geometry of fragments distort significantly upon bonding, the output of a typical EDA routine will not reflect this energy. Bonding-induced geometry changes (and their associated energies) should not be neglected as a rule, but BF 3 changes from a planar to pyramidal geometry when it binds to an electron donor. The energy for this change is positive (disfavored) 134 Increase in Fragment Energy Figure 5.28: A schematic showing the change in energy as BF 3 goes from a planar geometry to a pyramidal geometry. The pyramidal BF 3 represents the BF 3 fragment’s geometry when it is dative bonded to a Lewis base. The energy of this geometry change lowers the overall dative bond binding enthalpy. A similar comment can be made about the Lewis base, but the engergetic magnitude of this strain is considerably smaller and reasonably large, so using EDA alone will result in an substantial overestimate of the binding energy for BF 3 adducts 251 . Mao and coworkers have published what they call an “adiabatic” EDA method which includes geometric relaxation, and this method is available in QChem 1 . Its implementation isn’t as automated as some other QChem features, so I decided to account for this energy by manually setting up fragment geometry optimization jobs to determine the fragment relaxation energy. This allowed me to find the difference in energy between the adducted BF 3 fragment and a fully relaxed, free BF 3 and subtract that from the EDA results I obtained for each quinoline-BF 3 adduct. To give a clear example of how I run EDA jobs, I’m including an example EDA input file that I used for the Quinoline-BF 3 paper, which is at the end of this appendix section. The important thing to include (which you do not use in a typical Qchem job) are the double hyphens which bracket the text containing the atomic coordinates. These double hyphens indicate separate molecular fragments, and note that I have to specify the charge and multiplicity for the whole structure and each fragment individually. The job will not run otherwise. 135 If you take a look at the input file, you’ll also note the three “@@@” before the second section. This allows the jobs to run sequentially, and pass information from the first job to the second through the .fchk file. You can see this where I specify “read” in the molecule section instead of writing out more coordinates. By writing “read”, I can use the optimized geometry that Qchem found at the end of the first job. Running your jobs in this way isn’t better in every case, but it can save you a lot of legwork in many situations. D.3 Some tips for TD-DFT Geometry Calculations I also had to run a decent amount of TD-DFT and TD-DFT geometry optimizations for the Quinoline- BF 3 paper. I am not an expert in the mathematical details of TD-DFT, but I was able to find a few key input file parameters which allow these jobs to converge in a reasonable amount of time. One issue I found with TD-DFT calculations is that when you are using larger basis sets and functionals like ωB97X-D (and others in that family) I recieved warnings in my Qchem output file that the “THRESH” setting was too low. My recollection from looking this up is that the SCF algorithm can run into numerical issues with basis set overlap unless you tell it to run at a higher level of precision. I typically just tightened this parameter by writing “THRESH = 12”, for example. In situations where this helped me get convergence, I would often see the SCF energy start oscillat- ing, instead of exhibiting convergent behavior. This problem seems to happen more often with TD-DFT jobs. In a TD-DFT geometry optimization, QChem tries to minimize the vertical excitation energy between your state of interest and the ground state. You may, for example, be interested in the sec- ond excited state of a chromophore, and tell QChem to optimize the energy of that state. However, if there are other close-lying excited states along your optimization path, you can run into a good deal of trouble, especially if the identity of the TD-DFT states changes over the optimization path. Even without discussing conical intersections, the TD-DFT algorithm can get stuck flip-flopping between two different states that are intersecting and your optimization will never converge. I ran into this when trying to perform excited state optimizations on quinolines. Thankfully, this 136 problem was solved by setting “CIS TRIPLETS = FALSE”, which tells QChem to ignore the cal- culation of triplet states for the TD-DFT job. Since I was only interested in the singlet photobasic states of quinoline, this was an ideal choice. If you wish to do TD-DFT geometry optimizations and are still running into some trouble, I would also recommend looking into the ‘state following’ functions included in QChem. The two main adjustments here are to narrow the energy threshold for searching between different states (“FOLLOW ENERGY”, the QChem manual says this can help you traverse either an adiabatic or diabatic path), and to check the electronic character of the state you’re optimizing (“FOLLOW OVERLAP”, which uses overlap integrals to make sure the electronic structure doesn’t change wildly along optimization steps). If you’re still stuck after all this, either you’re missing something simple or you’re tackling a somewhat difficult problem! Either way, good luck. Listing 5.1: QChem Geometry job preceding EDA routine 1 $molecule 2 0 1 3 −− 4 0 1 5 N −0.3402289 0.4506070 −0.3554335 6 C −1.3930588 0.5939871 0.7148063 7 C −0.8400754 −0.2587466 −1.5838306 8 C 0.2448034 1.7816048 −0.7616081 9 C −1.3175061 −1.7117676 −1.3869914 10 C −2.8323673 0.9398199 0.2654294 11 C −0.6970459 2.8268374 −1.3738333 12 H −1.0654376 1.3448073 1.4691726 13 H −1.5254809 −0.3643697 1.2653255 14 H −0.0503515 −0.2826216 −2.3691384 15 H −1.6895523 0.2975482 −2.0285616 16 H 0.7222251 2.2943235 0.1050161 17 H 1.0465339 1.6186508 −1.5202860 18 H −0.4837854 −2.3849466 −1.1093309 137 19 H −1.7185773 −2.0908442 −2.3506692 20 H −2.1200994 −1.7897237 −0.6281038 21 H −2.9387087 1.8946250 −0.2654561 22 H −3.4761562 1.0046566 1.1682217 23 H −3.2684803 0.1499661 −0.3756735 24 H −1.3931561 3.2327270 −0.6195974 25 H −1.2325455 2.4483842 −2.2663805 26 H −0.0890578 3.6968850 −1.7016976 27 −− 28 0 1 29 B 0.8828801 −0.3489296 0.2794020 30 F 0.4575624 −1.6009897 0.8939198 31 F 1.5282676 0.4320178 1.3301025 32 F 1.8816019 −0.6579785 −0.7380895 33 $end 34 35 $rem 36 BASIS = 6−311+G ** 37 GUI = 2 38 JOB TYPE = Optimization 39 METHOD = B3LYP 40 SCF ALGORITHM = DIIS 41 SCF CONVERGENCE = 8 42 $end Listing 5.2: EDA Job following Geometry optimization -these groups of code should be submitted in a single input file. 1 @@@ 2 3 $molecule 4 read 5 $end 138 6 7 $rem 8 BASIS = 6−311+G ** 9 GUI = 2 10 JOB TYPE = EDA 11 METHOD = B3LYP 12 SCF ALGORITHM = DIIS 13 $end E Rotations, Quaternions, and You I decided to teach myself how to use quaternions because I was interested in rotating the atomic coordinates for QChem jobs to have specific orientations with respect to an applied electric field. For the Dawlaty lab’s purposes, rotation matrices are fully capable of doing the job, but I wanted to use quirkier mathematics. But hold on - what are quaternions? Let me attempt an informal introduction. Through the course of living life and contemplating the natural world, you may have con- sidered the rotation of vectors in three dimensions and thought to yourself: “Hm! This problem doesn’t seem ‘complex’ enough!” If that’s the case, I’m excited to introduce quaternions to you. Quaternions were first conceived by William Hamilton in 1843 as a way to arbitrarily transform vectors in three-dimensional space 252 . I encourage you to do some searching on your own to learn more about the history of quaternions. Though quaternions are studied own their own, most computational and instructional approaches tend to use matrices to transform vectors instead of quaternions, certain fields such as robotics and video rendering prefer to use quaternions because they allow for easy interpolation within a rotation operation and tend to be more numerically sta- ble 253 . In addition, they fundamentally avoid the problem known as ‘gimbal lock’. A quaternion is defined as: q= a+ bi+ cj+ dk 139 As is apparent from the definition and can be surmised from the name, quaternions are com- prised of four elements, and they can be thought of an extension of a typical complex number to multiple complex planes. If you like, the quaternion is a real number along with three mutually orthogonal complex values. Addition and subtraction of quaternions works equivalently to vector (or complex) algebra. Before getting too worried about how to think of all these complex planes, just think of what Hamilton was originally looking for: he wanted a way to transform vectors. To transform (multiply) a scalar, you need another scalar. 3-D vectors in general have to be described by 4 quantities - a magnitude, and their direction in each of the three coordinates. Therefore, to transform them in general, you need to specify the change in magnitude of the vector, along with two reference axes for rotation and the magnitude of the rotation - a minimum, essentially, of four pieces of information. In short, simply think of the quaternion as a concise way of storing the in- formation needed to transform vectors, as opposed to an even worse version of complex numbers. On a practical note, don’t let yourself get too wrapped up in trying to get an extremely intuitive picture of four-dimensional objects like quaternions. Our brains aren’t good at it, so just do your best and then give yourself a break! An important distinction that applies to quaternions is the way that its basis vectors multiply. With a basic understanding of complex numbers, you know by definition that i 2 =− 1. This is still true, but multiplication by each basis vector (i, j, k) corresponds to a unique rotation in this space, and these multiplications do not commute. Along with the above definition of quaternions, this further definition is necessary: i 2 = j 2 = k 2 = ijk=− 1 I believe the final equality in the prior equation is enough to specify all the multiplicative identities of the basis vectors, but I’ll include some here, as they won’t be immediately obvious. Others are ij= k,ji=− k,ij=− ji. I find it’s much easier to get a handle on these operations if you think of them as rotations in space, which also makes it relatively easy to understand why these multiplications don’t commute. For a more modern way of describing quaternion multiplication, we can frame these opera- tions as a combination of scalar multiplication, a dot product, and a cross product. This is pretty 140 reasonable! Let us define two quaternions, q 1 = s+⃗ u and q 2 = t+⃗ v, where we are separating the quaternions into their scalar and vector components. Then, their multiplication is given by: q 1 q 2 =(s+⃗ u)(t+⃗ v)=(st+⃗ u·⃗ v)+(s⃗ v+t⃗ u+⃗ u× ⃗ v) You’ll note that the parenthesis gather the new scalar and vector terms that arise from the multiplication, and we regain the original quaternion structure. Now I’ll try to explain how to rotate a real, 3D vector using quaternions. Somewhat confusingly (but also, not), if you have a real vector⃗ v: ⃗ v= v 1 i+ v 2 j+ v 3 k the corresponding position quaternion p looks like: p= 0+ v 1 i+ v 2 j+ v 3 k I’ve kept the zero included to emphasize that the corresponding quaternion has no real com- ponent, and all of the vector information goes to the imaginary terms. A quaternion with only imaginary magnitude is called a ‘pure’ quaternion. I won’t prove these quaternion operations to you, but in order to rotate p to a new position, q ′ you have to multiply p by q, a quaternion that holds the information on the rotation you’d like to carry out. The operation looks like this: p ′ = q − 1 pq And remember, because quaternion multiplication includes a cross product, this operation doesn’t commute. There are a variety of ways to find the rotation quaternion, and you can some- what readily go between quaternions and the language of rotation matrices, Euler angles, etc. However, if you know the axis ⃗ w = w 1 i+ w 2 j+ w 3 k about which you’d like to rotate and the corresponding degree of rotation (θ), the formula for q is pretty simple: q= e θ(w 1 i+w 2 j+w 3 k) = cos θ 2 +(w 1 i+ w 2 j+ w 3 k)sin θ 2 141 This equation also shows the extension of Euler’s formula with multiple imaginary terms. This section finishes out with some Matlab code that applies the operation of quaternions to the rotation of atomic coordinates to align a molecule with respect to an electric field in a QChem calculation. It looks like I did an okay job commenting my code, but I leave out some details when it comes to parsing the atomic coordinate file. If you’re struggling to understand that code, I’d take a look at the “fopen”, “fgets”, “fclose” family (“fopen” = file open, if I’m not mistaken), and learn how they work. The third segment of code contains the actual script that you run, the other two are functions called by the script. There are many resources available describing the fundamentals and use of quaternions (including Hamilton’s own text!) 253;252 . I encourage readers to seek those out for a more detailed discussion. The Matlab function below generates the rotation quaternion. Listing 5.3: Matlab code to generate a rotation quaternion 1 function quat out = quat rot(axis in, axis out) 2 % This function outputs a rotation quaternion corresponding to a 3 % rotation from a specified input vector to another specified 4 % output vector in 3D space. 5 6 % Computes the proper rotation axis with cross product 7 n = cross(axis in,axis out)./norm(cross(axis in,axis out)); 8 %Computes degree of rotation 9 theta = acosd(dot(axis in,axis out)/norm(axis in)/norm(axis out)); 10 quat init = zeros(1,4); 11 quat init(1) = cosd(theta/2); 12 quat init(2:4) = sind(theta/2) * n; 13 14 conj = [1,−1,−1,−1]; 15 quat init = quat init/sqrt(norm(mult quat(quat init, quat init. * conj ))); 16 17 18 quat out = quat init; 142 The following function multiplies two quaternions together. Listing 5.4: Matlab code that carries out quaternion multiplication, as described in the chapter 1 function quaternion out = mult quat(quat1,quat2) 2 quat init = zeros(1,4); 3 %Extracts out the real and imaginary parts of the quaternion, 4 %primarily for code legibility. 5 6 p0 = quat1(1); 7 q0 = quat2(1); 8 9 p = quat1(2:4); 10 q = quat2(2:4); 11 %computes the real part of the product quaternion 12 quat real = p0 * q0 − dot(p,q); 13 14 %computes the imaginary part of the product quaternion 15 quat imag = p0 * q + q0 * p + cross(p,q); 16 17 quat init(1) = quat real; 18 quat init(2:4) = quat imag; 19 20 quaternion out = quat init; This last bit of Matlab code reads and writes a .XYZ atomic coordinate input file, generates a rotation quaternion, rotates the atomic coordinates using the quaternion and reprints the rotated coordinates in a new .XYZ file. Listing 5.5: Matlab code to read, rotate and write atomic coordinates along an electric field oriented along the z axis 143 1 fid read = fopen(MoleculeXYZ in,'r'); 2 fid out = fopen(MoleculeXYZ out,'wt'); 3 4 5 % This loop reads through the XYZ file and pulls out the coordinates of 6 % atom 1 and atom 2, as designated above. 7 counter = 1; 8 readline = fgets(fid read); 9 readline = fgets(fid read); 10 11 while ¬feof(fid read) 12 readline = fgets(fid read); 13 if counter == rotationAtom1 14 xyz1 = str2num(readline(2:end)); %#ok<ST2NM> 15 end 16 17 if counter == rotationAtom2 18 xyz2 = str2num(readline(2:end)); %#ok<ST2NM> 19 end 20 counter = counter + 1; 21 22 end 23 24 %This step generates a rotation quaternion which performs the correct 25 %rotation. 26 referenceVector = xyz2 − xyz1; 27 rotation quat = quat rot(referenceVector,[0,0,norm(referenceVector)]); 28 29 % Resets the readfile to the beginning. 30 fclose(fid read); 31 fid read = fopen(MoleculeXYZ in,'r'); 32 33 readline = fgets(fid read); 34 readline = fgets(fid read); 144 35 %Optional loop which allows the user to recenter the geometry 36 %to a different atom than the one used for rotation. 37 counter = 1; 38 39 if shiftagain == true 40 while ¬feof(fid read) 41 readline = fgets(fid read); 42 if counter == shiftatom 43 atomcoords = str2num(readline(2:end)) − xyz1; %#ok<ST2NM> 44 atomrot1 = mult quat(rotation quat,[0,atomcoords]); 45 atomrot2 = mult quat(atomrot1, rotation quat. * [1,−1,−1,−1]); 46 xyzshift = atomrot2(2:4); 47 48 end 49 counter = counter + 1; 50 51 end 52 end 53 54 % Resets the readfile to the beginning. 55 fclose(fid read); 56 fid read = fopen(MoleculeXYZ in,'r'); 57 58 %Throws out the first two lines of the .xyz file for rotation, 59 %but writes them to the new .xyz file 60 readline = fgets(fid read); 61 fprintf(fid out, readline); 62 63 readline = fgets(fid read); 64 fprintf(fid out, readline); 65 66 while ¬feof(fid read) 67 68 readline = fgets(fid read); 145 69 atomname = readline(1); 70 atomcoords = str2num(readline(2:end)) − xyz1; %#ok<ST2NM> 71 atomrot1 = mult quat(rotation quat,[0,atomcoords]); 72 73 atomrot2 = mult quat(atomrot1, rotation quat. * [1,−1,−1,−1]); 74 75 %Optional shift in geometry 76 if shiftagain == true 77 atomcoords = atomrot2(2:4) − xyzshift; 78 elseif ¬shiftagain == true 79 atomcoords = atomrot2(2:4); 80 end 81 fprintf(fid out, [atomname '\t' num2str(atomcoords(1), '%+8.5f') ... '\t' num2str(atomcoords(2), '%+8.5f') '\t' ... num2str(atomcoords(3), '%+8.5f') '\t' '\n']); 82 end 83 84 fclose('all'); 146

Abstract (if available)

Abstract My work as a graduate student at USC focused - in a general sense - on the fundamental physical chemistry of systems important for energy applications. Under this wide umbrella, I spent my time researching two separate areas: chemical structure and electric fields at the electrode/liquid interface, and the photobasic properties of a family of quinolines, which undergo a large increase in proton affinity in the excited state. For the first part of my PhD, I learned a great deal about surface spectroscopy, particularly through the use of surface vibrational Stark shift spectroscopy to investigate the local electric fields that exist at metal interfaces, with and without the application of potential. Ionic liquid structure at the metal surface was a specific area I worked in, and there are many more questions that remain to be answered about the properties of highly concentrated electrolytes. In the second part of my PhD, I studied the photobasic properties of quinolines. My work here largely served to indulge some fundamental curiosity I (and my group) had about these compounds. Prior work in the Dawlaty lab had established that quinolines greatly enhance their proton binding strength in the excited state. I wondered if quinolines could also increase their affinity towards a greater variety of electron acceptors, which would make them Lewis (or dative) photobases as well as Bronsted photobases. My work established a fundamental basis for studying photobases in this way, and the Dawlaty group is currently working to extend these ideas in a more applied context.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Excited state proton transfer in quinoline photobases

PDF

Water partitioning between the bulk and an electrode surface

PDF

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

PDF

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

PDF

Proton kinetics in electrochemistry: new directions and mechanistic analysis

PDF

Vibrational sum frequency generation studies of electrode interfaces

PDF

Molecular orientation and structure of hydrogen bonds at water interfaces

PDF

Hot electron driven photocatalysis and plasmonic sensing using metallic gratings

PDF

Asymmetric response and the reorientation of interfacial water with respect to applied electric fields

PDF

Vibrational spectroscopy and molecular orientation at interfaces: correlating the structure of a material's surface to its properties

PDF

Elucidating local electric fields and ion adsorption using benzonitrile: vibrational spectroscopic studies at interfaces and in bulk

PDF

Molecular orientation from sum frequency generation spectroscopy: case study of water and methyl vibrations

PDF

The adsorption and selective deposition of molecular and nanocluster ions on carbon based devices

PDF

Potential-induced formation and dissociation of adducts at interfaces and adsorption of phenols onto metal electrodes

PDF

Understanding organic and inorganic semiconductor surfaces by coherent nonlinear spectroscopy

PDF

Dissociation energy and dynamics of HCl and aromatic-water clusters

PDF

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

PDF

Monitoring local electric fields, charge densities at electrode surfaces using in-situ graphene/surface enhanced Raman spectroscopy (GERS/SERS) based Stark-shifts

PDF

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

PDF

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

Asset Metadata

Creator Voegtle, Matthew Joseph (author)

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

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Degree Conferral Date 2022-12

Publication Date 08/26/2022

Defense Date 08/25/2022

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

Tag electronic spectroscopy,ionic liquids,OAI-PMH Harvest,photobase,photochemistry,proton transfer,solvation,surface spectroscopy,vibrational spectroscopy

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Dawlaty, Jahan Mansoor (committee chair), Benderskii, Alexander (committee member), Ravichandran, Jayakanth (committee member)

Creator Email voegtle@usc.edu,voegtlematthew@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111384553

Unique identifier UC111384553

Legacy Identifier etd-VoegtleMat-11160

Document Type Thesis

Format application/pdf (imt)

Rights Voegtle, Matthew Joseph

Type texts

Source 20220901-usctheses-batch-976 (batch), 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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

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

Repository Email cisadmin@lib.usc.edu