Capturing the sun: leveraging excited state dynamics

PDF

Download

Open document

Flip pages

Summarize

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

CAPTURING THE SUN: LEVERAGING EXCITED STATE DYNAMICS

By Michael Sean Kellogg

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMISTRY)

August 2023

Copyright 2023 Michael Sean Kellogg

ii

Dedicated to my parents,
James and Patricia Kellogg
iii

Acknowledgements
My biggest acknowledgments and appreciations are to Professor Steve Bradforth, my PhD
advisor, PI, and first ever boss. It goes without saying that nothing would have been possible
without Steve’s insight, support, and advice. More than just helping me understand raw data that
I just took to him to understand, he provided and molded a manner of thinking that goes beyond a
single dataset. I would utilize his advice to form my thought processes into proper critical and
scientific thinking, a skill that is invaluable. He taught me to piece apart a dataset and to not just
throw it away if it did not produce the intended outcome, even if that was the personal desire. And
beyond science, he helped me curate my speech to be effective, descriptive, and informative to
transfer information to other people. Diplomacy and tailored speech with people to achieve a
mutually beneficial discussion and outcome was also a strong skill I learned from Steve. I am
naturally impulsive. Steve has taught me to “hold on” when methodical thinking was required.
For Steve, I’m ever grateful for launching my scientific career in a great and promising direction.
I would like to acknowledge Professors Jahan Dawlaty and Alex Benderksii. Jahan and
Alex, alongside Steve and the rest of the ultrafast group meeting congregation, provided an
exceptional atmosphere of learning and mutual understanding. I looked forward to group meeting
because if there was a topic that the collective group struggled to understand, I could always count
on Jahan, Alex, or Steve to explain it to us. I had the linear spectroscopy class as a joint student
between Jahan and Alex. I believe that Jahan was my best professor I ever had in terms of
understanding fundamentals and spectroscopy. And Alex has a grasp on physics that I have not
seen matched. Jahan and Alex were instrumental to my PhD career.

iv

I would also like to thank the professors who lead the classes in my first and second years.
I would like to acknowledge Susumu Takahashi, Andrey Vilesov, Jahan and Alex, Oleg Prezhdo,
and Anna Krylov for their wonderfully instructive and instrumental classes. These classes formed
the basis of my grad school education and I thank you all for them. I’d also like to thank them for
providing a solid foundation for my physical chemistry education.
First, to my fellow 2016 grad student cohort, Ryan, thank you and we’ve had a both
challenging but fulfilling time in grad school. I came into grad school with far less confidence
than I have now, and I’d like to thank him for invigorating and just by virtue of being there, helped
me. While our projects never overlapped strongly, I think we shared some lovely lab space and
time with fantastic discussions about science. We will see if his prediction of unified scientific
theory of quantum versus relativity comes true! Good luck on your thesis and defense, I’m sure
you’ll knock it right out of the park and I look forward to reading and attending them, respectively.
To doctors Matt Bain and Fabiola Cardoso Delgado, thank you. I’ve had the pleasure to
work with both of them on fascinating projects. We’ve had many good conversations, shared
many laughs and thoughts and thanks to Matt, I had my first introduction to the exciting world of
rugby! Fab is absolutely wonderful and kind and just great to work with. Her efforts with the cMa
project helped it see fruition. To my junior students, Shivalee, José, Brandon, and Thabassum:
they’ve been all an absolute pleasure to work with and share my PhD time with. I had the
opportunity to stretch both my knowledge and explanation skills and often I found out that I didn’t
know a physical concept as well as I thought I did. Additionally, as well as learning physical
concepts from them because I didn’t and still don’t know everything! These lovely folk were
instrumental in helping me complete some of my craziest and most complex experiments. Shivalee
Dey, I thank her for assisting with the preliminary copper experiments that required freeze-pump-
v

thaws and glove bag usage, something very unique for both of us. She’s also been the go-to person
for the Magnitude Instruments work for 266 nm. And just an all-around wonderful and extremely
kind person. José Godínez has been and will continue to be a fantastic scientist and I look forward
to how he continues on his work. His constant upbeat attitude along with his devilish sense of
humor and mischievous pictures have always been a treat and my biggest sorrow is not seeing his
complete PowerPoint deck of pictures and memes. I have full faith and pride in saying José and
Shivalee will carry on the greatness of the Bradforth lab as the new senior students. Thabassum,
his infectious smile, healthy drive for questions, and strong work ethic have been an inspiration to
experience since his time here. His work with Mark’s group will create great science, I’m sure of
it. And Brandon, I want to thank him for helping me with exciting copper-olefin synthesis for
without his synthetic assistance, I’d have no possibility of completing it. For the few times we
did, it was great to talk about music with Brandon. So, I thank again my junior students and I will
absolutely miss them dearly. And to Kyro Grace, my undergrad student who helped me take the
first round of fluorescence quantum yield measurements for the SBCT stuff, thank you. Good luck
in med school, good sir!
I’d like to acknowledge and thank my senior students Gaurav Kumar, Laura Estergreen,
and Jimmy Joy. They were very helpful and instructive throughout the first few years of my
graduate degree. Laura taught me about the basics of transient absorption, TCSPC, and material
science in general as well as being my materials mentor. We’d have great discussions about
graduate school, what to expect, imposter syndrome, etc. For Garuav, I assisted him with all
manners of experimental procedures like REMPI, TRPES, TA and general optics. Most of what I
learned about optics was taken from Gaurav’s mentorship. And many a day I would spend with
vi

Gaurav talking about life and what it can hold for us. And it’s great to finally work to get the
Indole TA manuscript out! I’d like to thank Jimmy for assistance with my TCSPC experiments.
To my USC collaborators, thank you. The organic photovoltaics and novel photosensitizer
work I had the pleasure of working on with Professor Mark Thompson has been scientifically
invigorating and personally fulfilling. Mark is the legend in the OLED field. His insight into
organo-metallics and photophysical knowledge of these same compounds is unbelievable. And
even with such an impressive pedigree, Mark is approachable, patient, and understanding. Mark is
my unofficial second PI, I must say, and it is an honor and humbling to work with him. To my
synthetic student colleagues, Collin, and Austin, thank you. These gentlemen are exceptional in
organic and organo-metallic synthesis as well as photophysics. They’ve both been great to work
with on the cMa compounds and we’ve learned and had great fun, going to Penn State and UCR.
Even if the car we took to UCR was haunted! Any synthetic question, which may seem obvious
to them, they were quick and helpful to answer. I’d also like to thank the rest of the Thompson
group I got to work and talk with, Nina, Kelly, Peter, Jonas, and Allen. Especially thank you to
Nina and Kelly for their work on the cMa project. They’re bright young students with lots of
potential which I’m sure will be utilized in full. I wish them the best of luck in their experiments,
cryostat measurements, and the soon explosion of nsTA data from our Magnitude enVISion.
To my MURI collaborators in theory, thank you. Acknowledgements to Professor Steven
Lopez and his post-doc, Jordan Cox. I was able to have insightful and effective discussions with
them and bring about my first ever true deep dive into theory and fundamental photophysics of a
simple geometric molecule but one that is indiscernibly fascinating, HFB. With them and Matt and
Steve, we were able to uncover the true nature of this bizarre molecule and it’s due to Jordan and
Steven’s theory skills. To my MURI collaborators in synthesis, Professor Noah Burns and his
vii

students Ben Boswell and Carl Mansson, thank you. I appreciate them in helping me work through
both the HFB-olefin work and the copper-olefin synthesis, regardless of how inane my questions
might appear to an organic chemist. And finally, I’d like to thank Professor Todd Martinez and his
group for organizing a fruitful and productive MURI team as well as being a beacon of theoretical
and chemically physical knowledge.
To Professor John Asbury and Dr. Chris Grieco at Penn State and Auburn State,
respectively, thank you. I greatly appreciate the opportunity to partake in my first ever physical
chemistry in John’s lab. And thank you Chris for being my graduate student mentor at Penn State.
I had the opportunity to perform nanosecond laser-based experiments with novel block copolymers
as well as experience graduate school firsthand. I learned the necessity of absorption experiments,
sample preparation and began my trek with transient absorption. Also, thank you to Dr. Eric
Kennehan of Magnitude Instruments for not only developing a very exciting instrument but also
being fantastic to work with closely and being highly available, regardless of his insanely busy
schedule. He’s I also had the chance to meet John’s other wonderful students: Rob, Alec, Grayson,
and Kyle. The group of students I interacted with and the conversations I had solidified my
decision to go into graduate school in the first place.
To the 6
th
and 7
th
floor physical chemistry and spectroscopy groups, thank you. It was
great interacting with everyone throughout the years especially to Eric, Joel, Ryan, Anuj, Cindy,
and Matt from Jahan’s lab and Dhritiman and Muhammet from Alex’s lab. To the faculty and staff
that keep USC running, I’d like to thank them as well. Thanks to Yasaman Dayani for her
continued assistance at the Agilent Center, helping me gain access to useful instruments like a UV-
vis spectrometer and fluorimeter when our lab has it down as well as providing upkeep to these
tools. Thank you to Shawn Wagner for his work in our chemistry department as the NMR legend
viii

and personally teaching me how to do and take an NMR when I had no idea, especially a rarer one
like
19
F NMR. Without you, chapter 6 simply wouldn’t exist. Thank you to Joe Lim for helping
us analyze and fix all sorts of strange electronics we brought to him.
To those outside USC but my closest family and friends, absolutely thank you. To my
closest friends Octavio, Lauren, Sarah, and Alu, I adore you all and thank you for being my
absolute closest friends for years now. I’m ecstatic to have such kind, funny, sweet, understanding,
and talented artists as friends. After submission of this thesis, I cannot wait to see you in a week.
To my brothers, Tim and Kevin, thank you. Thank you for providing the utmost support and belief
in me along with adding healthy competition and the drive to continue onward, even in the hardest
of times in my PhD. They’re both an inspiration to always strive to do more in my career and life.
To my parents, throughout this they’ve been my biggest supporters and confidents when things
got rough. I loved their visits every weekend just to check up on me and to have lunch with on
Sunday and now letting me live at home while I sort the rest of my life out. I don’t think there’s
enough words to express my gratitude not only these past 7 years but for all my life. Mom and Dad
have been there every step of the way. I love you both dearly.

ix

Table of Contents
Dedication ...................................................................................................................................... ii

Acknowledgements ...................................................................................................................... iii
List of Tables .............................................................................................................................. xiv
List of Figures .............................................................................................................................. xv
Abstract ..................................................................................................................................... xxiv
Chapter 1. General Introduction ................................................................................................. 1
1.1. The Need to Capture the Sun ............................................................................................... 1
1.2. The Case for Photovoltaics .................................................................................................. 4
1.3. Generating solar fuels .......................................................................................................... 9
1.4. Fundamental Photophysics and Photochemistry ............................................................... 14
1.5. References .......................................................................................................................... 22
Chapter 2. Description, Design and Additions to a Nanosecond Transient
Absorption Spectrometer ........................................................................................................... 28
2.1. Introduction ........................................................................................................................ 28
2.2. Magnitude Instruments for nsTA ....................................................................................... 29
2.2.1. Method Comparison of psTA and nsTA ..................................................................... 31
2.2.1.1. Experimental method of psTA ............................................................................. 31
2.2.1.2. Experimental Methods for nsTA ......................................................................... 33
2.2.2. Photoluminescence Correction in Magnitude Instruments ......................................... 36
2.2.2.1. Counterpoint: The impact of PL in psTA and approaches used to correct
for PL in psTA. ................................................................................................................. 40
2.2.2.2. Correcting for PL in the nsTA Dataset ................................................................ 41
2.2.3. Second Harmonic Generation using Magnitude Instruments ..................................... 47
2.2.3.1. Prior SHG Design with the enVISTA .................................................................. 48
2.2.4. Current Setup in the enVISion .................................................................................... 57
2.2.4.1. Current SHG Design ............................................................................................ 65
2.2.4.2. Comparison of Data Obtained from enVISTA and enVISion SHG Setups ........ 71
2.3. References .......................................................................................................................... 74
2.4. Appendix ............................................................................................................................ 75
2.4.1. Data Structure of Magnitude Instruments Raw Data File (MATLAB MAT-
file) ........................................................................................................................................ 75
2.4.2. Dual Phase-Locked Choppers Scheme ....................................................................... 79
2.4.3. Subtraction of Two Exponentials ................................................................................ 80
2.4.4. Subtraction of a Raw Dataset from a Recreated Dataset ............................................ 81
2.4.5. Instrument Response Function of a Magnitude enVISTA .......................................... 82
Chapter 3. Symmetry Breaking Charge Transfer as a Means to Study Electron
Transfer with No Driving Force
‡
............................................................................................... 83
3.1. Abstract .............................................................................................................................. 83
3.2. Introduction ........................................................................................................................ 84
3.3. Experimental ...................................................................................................................... 89
x

3.3.1. Synthetic methods ....................................................................................................... 89
3.3.2. Quantum Yield Measurements.................................................................................... 89
3.3.3. Transient Absorption .................................................................................................. 91
3.3.4. Time-Correlated Single Photon Counting .................................................................. 92
3.4. Results and Discussion ...................................................................................................... 92
3.4.1. Electron Transfer at Zero Driving Force..................................................................... 92
3.4.2. Modeling LE and SBCT states.................................................................................... 97
3.4.3. Comparing the SBCT process in meso-BODIPY (1) and Zn(dpy)2 (2) ................... 101
3.5. Conclusion ....................................................................................................................... 104
3.6. References ........................................................................................................................ 105
3.7. Supplement and Appendix ............................................................................................... 110
3.7.1. Steady State Absorption and Emission ..................................................................... 110
3.7.1.1. Pure Solvents ..................................................................................................... 110
3.7.1.2. CHX/THF Mixtures ........................................................................................... 110
3.7.1.3. TOL/THF Mixtures ............................................................................................ 111
3.7.1.4. TOL/CHX Mixtures ........................................................................................... 111
3.7.2. Transient Absorption ................................................................................................ 112
3.7.2.1. Pure Solvents ..................................................................................................... 112
3.7.2.2. CHX/THF Mixtures ........................................................................................... 114
3.7.2.3. TOL/THF Mixtures ............................................................................................ 115
3.7.2.4. TOL/CHX Mixtures ........................................................................................... 116
3.7.3. Time-Correlated Single Photon Counting ................................................................ 117
3.7.3.1. Pure Solvents ..................................................................................................... 117
3.7.3.2. Fluorescein ......................................................................................................... 118
3.7.3.3. Table of Fitting Values ...................................................................................... 118
3.7.4. Summary of Rate Constants ...................................................................................... 119
3.7.4.1. TA Fitting Protocol – Abstraction of Rate Constants ........................................ 119
3.7.4.2. Summary of Rate Constants............................................................................... 120
Chapter 4. Intra- and Inter-Molecular Charge Transfer Dynamics of Carbene-
Metal-Amide Photosensitizers
‡
................................................................................................ 121
4.1. Abstract ............................................................................................................................ 121
4.2. Introduction ...................................................................................................................... 122
4.3. Experimental .................................................................................................................... 125
4.3.1. Bulk Electrolysis ....................................................................................................... 125
4.3.2. Pulse Radiolysis ........................................................................................................ 125
4.3.3. Sample Preparation for Optical Measurements ........................................................ 127
4.3.4. Time-Correlated Single Photon Counting ................................................................ 128
4.3.5. Picosecond Transient Absorption ............................................................................. 129
4.3.6. Nanosecond Transient Absorption ............................................................................ 130
4.4. Results and Discussion .................................................................................................... 132
4.4.1. Ground and Excited State Spectra of cMa Complexes ............................................. 132
4.4.2. Spectroelectrochemistry ............................................................................................ 134
4.4.3. ISC Rates Determined by TCSPC ............................................................................ 138
4.4.4. Spectrally Resolving ISC with psTA ........................................................................ 144
4.4.5. Excited State Spectra Recreation .............................................................................. 149
4.4.6. Observing Intermolecular CT via nsTA.................................................................... 152
xi

4.5. Conclusion ....................................................................................................................... 157
4.6. References ........................................................................................................................ 160
4.7. Appendix .......................................................................................................................... 163
4.7.1. Absorptivity Spectra of cMa ions from Pulse Radiolysis ......................................... 163
4.7.2. TCSPC Curves .......................................................................................................... 165
4.7.3. Picosecond Transient Absorption ............................................................................. 168
4.7.3.1. psTA Data on Changing the Carbazole Unit and Solvent ................................. 168
4.7.3.2. Charge Transfer Within a 1 ns Window with Concentrated Quencher ............. 170
4.7.3.2.1. Electron Transfer to MePI .......................................................................... 170
4.7.3.2.2. Hole Transfer to BIH .................................................................................. 173
4.7.3.3. Contour Plots of psTA spectra ........................................................................... 175
4.7.3.4. Operational Details for psTA and Target Analysis TA Fitting.......................... 181
4.7.3.5. Species Associated Decay Spectra from psTA Datasets ................................... 184
4.7.4. Nanosecond Transient Absorption ............................................................................ 187
4.7.4.1. nsTA Spectra ...................................................................................................... 187
4.7.4.2. Fitting Schemes for Quenching nsTA................................................................ 193
4.7.4.3. Species Associated Decay Spectra (SADS)/Triplet Spectra of cMa ................. 195
4.7.4.4. SADS of nsTA Quenching Experiments ........................................................... 196
4.7.5. Comparison of Triplet Spectra from psTA and nsTA............................................... 198
4.7.6. Simulation of the 𝑆𝑆 1 and 𝑇𝑇 1 ESA utilizing Pulse Radiolysis Spectra ...................... 200
4.7.7. Simulation of the 𝑆𝑆 1 and 𝑇𝑇 1 ESA utilizing Bulk Electrolysis Spectra ..................... 202
4.7.8. Stern-Volmer plots of 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 with BIH in THF .............................................. 204
Chapter 5. Role of the Perfluoro Effect in the Selective Photochemical Isomerization
of Hexafluorobenzene
‡
.............................................................................................................. 205
5.1. Abstract ............................................................................................................................ 205
5.2. Introduction ...................................................................................................................... 206
5.3. Experimental .................................................................................................................... 209
5.3.1. Computational Methods ............................................................................................ 209
5.3.2. Sample Preparation and Purification ........................................................................ 209
5.3.3. Absorption and Fluorescence Measurements ........................................................... 209
5.3.4. Temperature-Dependent Fluorescence ..................................................................... 210
5.3.5. Femtosecond-nanosecond Tranisent Absorption ...................................................... 211
5.3.6. Time-Correlated Single Photon Counting ................................................................ 212
5.4. Results and Discussion .................................................................................................... 213
5.4.1. Absorption Spectrum ................................................................................................ 215
5.4.2. Vibrationally Resolved Electronic Spectra ............................................................... 217
5.4.3. Electronic State Coupling ......................................................................................... 218
5.4.4. Transient Absorption Spectroscopy .......................................................................... 220
5.4.5. Potential Energy Surface ........................................................................................... 226
5.4.6. Simulated Photochemical Dynamics ........................................................................ 229
5.5. Conclusion ....................................................................................................................... 232
5.6. References ........................................................................................................................ 234
5.7. Supplement and Appendix ............................................................................................... 238
5.7.1. Temperature Dependent Fluorescence ...................................................................... 238
5.7.2. psTA Data ................................................................................................................. 239
5.7.2.1. HFB psTA in Cyclohexane ................................................................................ 239
xii

5.7.2.2. HFB psTA in Ethanol, excitation at 266 nm...................................................... 240
5.7.2.3. Contour Plots of HFB ........................................................................................ 241
5.7.2.4. psTA Data Time Delays..................................................................................... 242
5.7.2.5. Integrated Time Traces of psTA ........................................................................ 243
5.7.2.6. Concentration Dependent Transient Absorption ............................................... 246
5.7.3. Time-Correlated Single Photon Counting ................................................................ 247
5.7.4. Wavelength Dependence of HFB Fluorescence Lifetime......................................... 248
Chapter 6. Photochemistry of Hexafluorobenzene and Norbornene ................................... 256
6.1. Abstract ............................................................................................................................ 256
6.2. Introduction ...................................................................................................................... 256
6.3. Experimental .................................................................................................................... 260
6.3.1. Sample Preparation for Stern-Volmer Experiments ................................................. 260
6.3.2. Sample Preparation for Irradiation and NMR Studies .............................................. 260
6.3.3. Prediction of NMR of Irradiated Samples ................................................................ 261
6.3.4. Steady State Fluorescence Measurements ................................................................ 261
6.3.5. HFB Lifetime Quenching Measurements via TCSPC .............................................. 262
6.3.6. Irradiation Setup of HFB with and without NB ........................................................ 263
6.4. Results and Discussion .................................................................................................... 265
6.4.1. Reaction Efficiency Through Steady State Measurements ....................................... 265
6.4.2. Reaction Efficiency Through Lifetime Quenching ................................................... 269
6.4.3. Irradiation of HFB to Afford Dewar-HFB ................................................................ 273
6.4.4. Irradiation of HFB and NB to Afford Photoproducts ............................................... 278
6.5. Conclusion ....................................................................................................................... 284
6.6. References ........................................................................................................................ 286
6.7. Supplement and Appendix ............................................................................................... 288
6.7.1. Absorption Spectra .................................................................................................... 288
6.7.2. Fluorescence Spectra of Norbornene Samples.......................................................... 290
6.7.3. Stern-Volmer Theory and Simulation ....................................................................... 291
6.7.4. Assignment of Error Bars in SV plots ...................................................................... 297
6.7.4.1. Error Bar Assignment from Instrument Uncertainty Only ................................ 297
6.7.4.2. Error Bar Assignment Incorporating Human Error ........................................... 301
6.7.5. HFB and Norbornene Results from Burns Group ..................................................... 313
6.7.6. Photograph of Irradiated Samples ............................................................................. 314
Appendix A: Derivations and Calculations ............................................................................ 315
1. Calculated QY from Rate Constants (Equation 3.2) ................................................... 315
2. Measured QY from Steady State Measurements (Equation 3.1) ................................ 316
3. Excited State Population Calculation from Experimental Parameters ....................... 320
4. Average Distance Between Molecules Using Concentration ..................................... 321
Appendix B: Matlab Codes ...................................................................................................... 322
1. Colormap generation using HSV values for use in contour plots/2DTA ................... 322
2. Brewster Angle Calculations ...................................................................................... 324
3. Photoluminescence Correction of nsTA Data from using psTA data ......................... 330
4. Transient Absorption Simulation for a System with Two Excited States .................. 335
5. Appendix References .................................................................................................. 352
xiii

xiv

List of Tables
Table 2.1 – Spectral Range Details and Components for enVISion ............................................ 58
Table 2.2 – Experimental Parameters of enVISion with Benzophenone in Cyclohexane ........... 73
Table 3.1 – Kinetic parameters for compounds 1 and 2 ............................................................. 102
Table 4.1 – Summary of measured TCSPC fluorescence decays in toluene and derived
ISC rates and singlet-triplet energy gap ...................................................................................... 143
Table 6.1 – HFB Lifetimes from TCSPC of HFB-NB mixtures ................................................ 272

xv

List of Figures
Figure 1.1 – (a) Graph of the global surface temperature anomaly as recorded by
GISS/NASA. b) Recreated visualization of surface temperature for past years and the
current year, clockwise from top left: 1890, 1950, 2022, 1990. Temperature differences
(°F) colder than the average, below 0°C tend blue and hotter temperatures tend orange then
red. Short-term changes are smoothed on a five-year running point average between
consecutive years. Figures in b) obtained from climate.nasa.gov. Data acquisition and
smoothing parameters listed in Lenssen et al. ................................................................................ 3
Figure 1.2 – Comparing finite and renewable planetary energy reserves (Terawatt-years).
Total recoverable reserves are shown for the finite resources (outside yellow circle).
Yearly potential is shown for the renewables (inside yellow circle). Adapted from Perez
and Perez et al.(Perez 2022) ........................................................................................................... 4
Figure 1.3 – Chart of the highest efficiency research solar cells as a function of year and
class of material and design as of September 2022. Figure is courtesy of National
Renewable Energy Laboratory, NREL. Data adapted by NREL from Green 2023. ...................... 7
Figure 1.4 – Standardized spectral irradience of the sun, collected above earth’s
atmosphere (Extraterrestrial, red) on earth’s surface (global, blue) with an air mass of 1.5
(solar zenith angle of 48°). Integration of the extraterrestial spectrum yields 1366.1 W/m
2

and the global spectrum yields 1000 W/m
2
. Dips in the direct spectrum are attributed to
absorption by the overtones of vibrations of the indicated molecules. The pink spectrum
displays the photon flux (number of photons per unit wavelength per second) of AM1.5 by
dividing the solar irradiance by the photon’s energy. ..................................................................... 9
Figure 1.5 – Cartoon representation of a reductive catalytic cycle initiated by a
photosensitizer. Blue box – photosensitizer (PS); red circle – electrocatalyst (EC); green
pentagon – sacrificial reductant (SAC). Charged states of the species are labeled as well,
i.e. PS
+
is the oxidized photosensitizer. The stages of the cycle are denoted by their
respective numbers below the PS-EC pair as circled numbers. Activity between the PS –
EC pair is denoted by labeled black arrows. See text for full description of catalytic cycle. ....... 13
Figure 1.6 – a) Cartoon representation of web of organic molecules (letters), linked by
known reactions (blue arrows), undiscovered reactions (red arrows), and reversible
reactions (double headed arrows). A synthesis path for molecule X from the possible
starting materials is currently unavailable in this diagram. b) Cartoon diagram of Mt Baldy,
CA, adapted from the official map with all overlays and markers removed for clarity.
Letters – various summits; red line – helicopter path; black line – skiing path............................ 15
Figure 1.7 – a) Energy surface plot, representing a pair of potential energy surfaces (ground
state – blue, excited state – red) as a function of two reaction coordinates (Rxn Coord). All
units are arbitrary. Reactant – A, product – X, excited state reactant – A*. Excitation to A*
from A is indicated by h 𝜈𝜈 . Thermal pathway is indicated by the dotted arrow “1” and
excited state reaction is indicated by the dotted arrow “2”. b) The plot in a) at a more
xvi

extreme tilt angle with emphasis on the excited state pathway. Product X lies on the ground
state surface as indicated by the intercepting blue lines over the red lines................................... 17
Figure 1.7 – Cartoon iterative feedback loop of the Synthesis Planning and Reaction
Discovery for Photochemistry and Chemistry in Novel Environments Project. .......................... 19
Figure 1.8 – a) A summary of energy-dense polycyclobutane targets. b) Summary of target
polycyclobutanes in the Syntheis Planning project. c) Photochemistry of an activated 𝛼𝛼𝛼𝛼
ketone to form a cyclobutane aldehyde. d) Photochemistry of a Cu(I)-cyclobutene2
complex to form a ladderane......................................................................................................... 21
Figure 2.1 – Cartoon depiction of experimental timescale capabilities in this work. .................. 28
Figure 2.2 – nsTA of 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in toluene under varying values of PL subtraction
parameter. (a) No PL subtraction, (b) auto-subtraction and (c) reported time-domain nsTA
after auto-PL subtraction at varying wavelengths and time-dependent PL. The 440 nm is
a trace centered on the GSB has been inverted. The PL (also inverted) is plotted as the
circle-blue line. (d) Simulation of two exponentials with identical time constants and the
resulting difference. ...................................................................................................................... 38
Figure 2.3 - Power dependent nsTA spectra of 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in toluene, excited at 355 nm.
The y-axis is kept constant to demonstrate the decrease in signal as the power is decreased.
The absorption and emission spectra are scaled arbitrarily for clarity. ........................................ 41
Figure 2.4 – The transient spectra of 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in toluene with 355 nm excitation at
various laser pulse energies with a) black – 75 uJ or 100% power, b) red – 35 uJ or 47%
power and c) orange - 8 uJ or 11% power. These spectra were normalized at 500 nm. ............... 42
Figure 2.5 – Cartoon schematic of sample chamber geometry used in the pump
repositioning experiment used in understanding the source of the PL undersubtraction. The
black ellipses at either end of the chamber are lens. ..................................................................... 43
Figure 2.6 – Photograph of sample chamber with OD filter in place before probe during
the probe attenuation experiment. ................................................................................................. 44
Figure 2.7 – SWTA of 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in toluene, excitation at 355 nm, detection at 580 nm
with PL subtraction engaged. a) SWTA for all four ND filters and b) inset of left with 2.1
OD removed for visibility. ............................................................................................................ 44
Figure 2.8 – nsTA of 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in toluene with the PL subtraction method described in
the text engaged here. a) 2DTA of 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in toluene and b) Normalized nsTA time
traces at 570 nm under varying s values. The PL at 570 nm is plotted as a dashed black
line................................................................................................................................................. 46
Figure 2.9 - a) Previous setup of Magnitude enVISTA, external laser, and SHG. b)
Photograph of entire enVISTA setup and c) a closer view of the SHG setup. There are
minor differences between the schematic in a) and the realized setup in b) and c): the
removal of lenses L2 and L3, the dovetail stage for L1 and X1 replacement with micrometer
xvii

stages for X1. The white circled optic in c) is only in place for viewing the beam mode of
266 nm and was removed when collecting data. .......................................................................... 50
Figure 2.10 – Photograph of previous external laser housing design for pairing with the
enVISTA. Along the beam propagation the optical components are as follows: In – input
from external laser cavity; BS – beamsplitter for delivering a small fraction to the triggering
photodiode, TPD; WP – mechanical waveplate; P – cube polarizer; BD – beam dump; L1a
– lens for focusing through chopper blades; Ch – chopper; PH – 1 mm pinhole; L2a –
recollimating lens; M1a – mirror for directing output, removed in previous setup; Out –
output to SHG setup and enVISTA. ............................................................................................. 51
Figure 2.11 – Photographs of CLBO crystal immediately after unpacking from Eksma and
housing left) front facing, center) side facing and on the right) in the SHG setup. The small
arrow in the right-hand photo indicates the fused silica window with visible AR coating.
This arrow points in the opposite direction to the laser beam propagation. The CLBO
crystal was placed such that the flattened side with the arrow facing away from the camera.
....................................................................................................................................................... 52
Figure 2.12 – Photograph of 266 nm beam with the CLBO crystal. Taken after the circled
optic in Figure 2.10c. ................................................................................................................... 55
Figure 2.13 – a) Transmission spectra of the probe for differing instruments (enVISion or
enVISTA), G – grating number, and lamp (Xenon or Halogen). Not every combination was
explored. b) Transmission spectra of the probe in the NIR for the enVISion. D – Detector. ....... 61
Figure 2.14 – Laser pump spectra for blue – internal 355 nm and green – external 532 nm
laser, normalized to 100 at peak height. ....................................................................................... 62
Figure 2.15 – Photographs of outputs of 532 nm, left and 355 nm, right, taken May 2
nd
,
2023............................................................................................................................................... 62
Figure 2.16 – Laser powers (left axis, mW) and pulse energies (right axis, 𝜇𝜇 J) as a function
of the laser power attenuator transmission for the 355 nm laser at a) 100% diode current, 2
kHz repetition rate and b) 80% diode current, 5 kHz repetition rate. A linear fit is displayed
as well for each (red line). c) Recorded internal laser power and pulse energies as a function
of laser diode current and OD filter, collected at 5 kHz laser repetition rate. Pulse energies
were obtained through dividing the power by half the laser rep rate. The powers were
recorded after the chopper, at the sample stage. Solid symbols – current enVISion values;
open symbols – previous enVISTA values. d) Recorded laser powers as a function of the
laser power attenuator for the 532 nm laser at 100% diode current, 2 kHz repetition rate.
A linear fit is displayed as well for each (red line). Errors in power are ~0.4 mW but not
displayed as they are smaller than the size of the symbol. ........................................................... 63
Figure 2.17 – Photograph of enVISion (right hand side, sample chamber open) and
external laser (left hand side) as of April 2023 in the Mark Thompson lab space. ...................... 64
Figure 2.18 – Photograph of external laser and SHG setup as of April 2023. The naming
convention for this setup is adopted from the previous external laser housing and SHG
xviii

setup. Modifications: Atten – laser power attenuator; M1-M4 – 532 nm HRs; M5 and M6
– 266 nm HRs; FM2 – Magnetic kinematic mount with 266 nm HR to send SHG into
enVISion. Mounts are in place for 266 nm generation. FM1 would be placed on the pedestal
in front of L1 and FM2 would be removed to pump 532 nm for the enVISion. .......................... 65
Figure 2.19 – Photograph of 266 nm laser mode on a ruled card in the sample chamber in
the left) horizontal and right) vertical directions. The laser power was attenuated to ~1% to
avoid camera saturation. ............................................................................................................... 67
Figure 2.20 – Laser pump spectrum 266 nm laser output, normalized to 100 at peak height.
The ‘ * ’ above the peak at 270 are an artifact from the spectrometer, confirmed by
comparing the mercury spectrum on both the USB 650 and USB 2000 spectrometers where
every peak in the USB 2000 possessed a ~1-10% shoulder, ~4 nm red of the main feature.
The USB 650 did not and the artifacts in the USB 2000 were not known mercury lines. ........... 69
Figure 2.21 – Laser powers (left axis, mW) and pulse energies (right axis, 𝜇𝜇 J) as a function
of the laser power attenuator transmission for the 266 nm laser at a) 100% diode current, 2
kHz repetition rate. A quadratic fit is displayed as well for each (red line). ............................... 70
Figure 2.22 – nsTA data obtained on benzophenone pumped via 266 nm. Comparison of
BP in EtOH data obtained from a) the enVISTA and b) the enVISion. See text for
experimental differences. d) A SWTA of BP in CHX and d) a 2DTA of the same sample. ........ 72
Figure 3.1 - (top) A schematic representation of the decay process for SBCT materials in
polar and non-polar media. Two identical chromophores are represented by the open
rectangles and the asterisk indicates an excited chromophore (singlet or triplet). (bottom)
A kinetic scheme is shown for the SBCT process. The energy of the CT state is strongly
solvent independent, while that of the chromophore localized singlet (labelled LE) is
largely solvent independent. ......................................................................................................... 85
Figure 3.2 – (a) Dipyrrin structures used to study symmetry breaking charge transfer in
this paper. Ar = 1-mesityl. (b) The HOMO and LUMO orbitals of BODIPY are shown.
The surfaces are the same for Zn coordinated to the dipyrrin. ..................................................... 87
Figure 3.3 – Photoluminescence quantum yields in a range of solvents and solvent
mixtures. Both measured (open symbols) and values calculated from the TA measurements
using Equation 1.2 (closed symbols) are shown. The Φ 𝑓𝑓𝑓𝑓 values are plotted versus both
solvent dielectric (left) and solvent polarity, using the ET(30) scale (right). ................................ 92
Figure 3.4 – (left) Rate constants from TA experiments are shown for 2 in a range of
solvents, from nonpolar to polar. (right) Rate constants from TA experiments are shown
for both pure solvent sand solvent mixtures, focusing on the less polar solvents and
mixtures. Filled symbols are 𝑘𝑘𝑘𝑘𝑘𝑘 and open symbols are 𝑘𝑘 𝑏𝑏𝑏𝑏 𝑘𝑘 (see Figure 3.1). ......................... 95
Figure 3.5 – ∆G for the equilibrium LE ⇄ SBCT for 2 as a function of solvent ET(30). A
line has been added at Δ 𝐺𝐺 = 0 as a guide. .................................................................................... 96
Figure 3.6 – Molecular orbitals for 1 and 2 (DFT: B3LYP functional, 6-31+G* basis). ............ 97
xix

Figure 3.7 – SBCT and LE natural transition orbitals (NTOs) for 1 and 2. The hole orbitals
are in blue, and the electron orbitals are in green. ........................................................................ 98
Figure 3.8 – Geometry optimized excited state structures of 2 differ in media with different
dielectric constants. The structural change here is largely a displacement of the Zn
2+
ion
along the axis containing the two meso-carbons and the Zn
2+
ion, given by a and b in the
image. The effect is marked in the SBCT state, where the Zn
2+
ion moves closer to the
dipyrrin carrying the electron. The effect is most pronounced in nonpolar media. .................... 100
Figure 4.1 – Schematic representation of a carbene metal amide (cMa, 𝐴𝐴 𝑀𝑀𝑀𝑀 𝑀𝑀𝑀𝑀 𝑏𝑏 𝑘𝑘 𝑀𝑀𝑀𝑀 𝑏𝑏 𝑏𝑏 𝑀𝑀 𝑏𝑏 )
being used as a sensitizer for a photoelectrocatalytic reduction reaction. The cycle begins
with light excitation of the 𝐴𝐴 𝑀𝑀𝑀𝑀 𝑀𝑀𝑀𝑀 𝑏𝑏 𝑘𝑘 𝑀𝑀𝑀𝑀 𝑏𝑏 𝑏𝑏 𝑀𝑀 𝑏𝑏 to form the 𝐴𝐴 𝑀𝑀𝑀𝑀 𝑀𝑀𝑀𝑀 𝑏𝑏 + 𝑘𝑘 𝑀𝑀 𝑀𝑀𝑏𝑏𝑏𝑏𝑀𝑀 𝑏𝑏 −

excited
state. The excited state then either donates an electron to a catalyst (CAT) (right path) and
a reductant (RED) returns 𝐴𝐴 𝑀𝑀𝑀𝑀 𝑀𝑀𝑀𝑀 𝑏𝑏 + 𝑘𝑘 𝑀𝑀 𝑀𝑀𝑏𝑏𝑏𝑏𝑀𝑀 𝑏𝑏 to the ground state. In alternate path, (left
path) the reductant captures the 𝐴𝐴 𝑀𝑀𝑀𝑀 𝑀𝑀𝑀𝑀 𝑏𝑏 + 𝑘𝑘 𝑀𝑀 𝑀𝑀𝑏𝑏𝑏𝑏𝑀𝑀 𝑏𝑏 − excited state, forming
𝐴𝐴 𝑀𝑀𝑀𝑀 𝑀𝑀𝑀𝑀 𝑏𝑏 𝑘𝑘 𝑀𝑀𝑀𝑀 𝑏𝑏 𝑏𝑏 𝑀𝑀 𝑏𝑏 −, which has sufficient reducing potential to reduce the catalyst. The
reductant can be either an electrode or a chemical reductant. .................................................... 122
Figure 4.2 – Left) General structure of the cMa compounds in this article. The cMa on the
left is displayed in its excited ICT state. The carbene ligand is displayed in blue, the amide
is displayed in red. Center) R-group substitutions on carbazole. Right) Carbene ligands
used in this study. Bottom) N-methylphthalimide (MePI) and dihydrobenzimidazole (BIH)
used in electron and hole transfer studies, respectively. ............................................................. 132
Figure 4.3 – (a) Steady state molar absorptivity spectra in THF for 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴 and
𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 (top) and 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 , 𝐴𝐴 𝐴𝐴 𝐶𝐶 ℎ 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 , 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐶𝐶 𝐴𝐴𝐴𝐴 𝐶𝐶 𝐴𝐴 𝐴𝐴 (bottom). 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐶𝐶 𝐴𝐴𝐴𝐴 𝐶𝐶 𝐴𝐴 𝐴𝐴 is
in 2-MeTHF. The pump wavelengths used in the laser experiments are indicated as dashed
lines: nsTA at 355 nm, TCSPC at 400 nm and psTA at 405 nm. (b) Steady state emission
spectra of 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴 and 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in 2-MeTHF (top) and 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 , 𝐴𝐴 𝐴𝐴 𝐶𝐶 ℎ 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 ,
𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in 2-MeTHF. 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 is in THF. ................................................................... 133
Figure 4.4 – (a) Excited state absorptivity spectra of 10 mM 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 , 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 , and
𝐴𝐴 𝐴𝐴 𝐶𝐶 ℎ 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in a solution of 20 mM triphenylamine in o-xylene 5 ns after electron pulse.
(b) Excited state absorptivity spectra of 10 mM 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 , 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐶𝐶 𝐴𝐴𝐴𝐴 𝐶𝐶 𝐴𝐴 𝐴𝐴 , and 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴
in o-xylene 10 ms after electron pulse. Minor differences between the two experimental
conditions are observed in the spectra for 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 . Data collected is displayed as
symbols and the lines are interpolations between data. .............................................................. 135
Figure 4.5 – The bulk electrolysis spectra of 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 and 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in THF.
Neutral (0 V) – black, Cation (+0.9 V) – Red, Anion (-2.1 V) – Blue. The voltages were
referenced to a silver pseudo-reference electrode. The peak centers of the cations are also
indicated by arrows. A 7-point smooth was applied to the data to remove high noise in the
< 500 nm region. ......................................................................................................................... 136
Figure 4.6 – (a) Cation molar absorptivity spectra of 10 mM 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 (blue),
𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 (red), and 𝐴𝐴 𝐴𝐴 𝐶𝐶 ℎ 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 (black) in a 20 mM solution of triphenylamine in o-
xylene. (b) Anion molar absorptivity spectra of 10 mM 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 (blue), 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴
xx

(green), and 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐶𝐶 𝐴𝐴𝐴𝐴 𝐶𝐶 𝐴𝐴 𝐴𝐴 (orange) in 10 mM TADPF in THF. Data collected is displayed
as symbols and the lines are interpolations between data. .......................................................... 137
Figure 4.7 – Kinetic model of the cMa compounds. Rates: 𝑘𝑘 𝑀𝑀𝑏𝑏𝑓𝑓 – solvent and
intramolecular vibrational relaxation to 𝑆𝑆 1 minimum, 𝑘𝑘𝑘𝑘 𝑆𝑆 𝐴𝐴 𝑏𝑏 𝑘𝑘 𝑏𝑏 𝑀𝑀 – exothermic intersystem
crossing from 𝑆𝑆 1 to 𝑇𝑇 1, 𝑘𝑘 𝑘𝑘 𝑆𝑆𝐴𝐴 𝑏𝑏 𝑀𝑀 𝑀𝑀𝑏𝑏 𝑀𝑀 – endothermic intersystem crossing from 𝑇𝑇 1 to 𝑆𝑆 1,
𝑘𝑘 𝑀𝑀 𝑆𝑆 + 𝑘𝑘 𝑀𝑀𝑀𝑀 𝑆𝑆 / 𝑘𝑘 𝑀𝑀 𝑇𝑇 + 𝑘𝑘 𝑀𝑀𝑀𝑀 𝑇𝑇 – the sum of radiative and non-radiative relaxation rates from
𝑆𝑆 1/ 𝑇𝑇 1 . Additionally, 𝜏𝜏 𝑇𝑇 𝐴𝐴𝐶𝐶 𝜏𝜏 = 1 𝑘𝑘 𝑀𝑀 + 𝑘𝑘 𝑀𝑀𝑀𝑀 . ............................................................................. 140
Figure 4.8 – The psTA spectra of 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 in toluene (a) and THF (b) with 405 nm
pump. The temporal evolution is indicated by rainbow color coded spectral traces. The
inverted steady state absorption (blue dash) and emission spectra (red dash) are displayed.
The regions of 400 and 800 nm are removed due to large pump scatter. The major spectral
evolutions are depicted with arrows: Initial ultrafast ESA redshift (1), subsequent ESA
blueshift (2), and simultaneous SE recovery (3). ........................................................................ 145
Figure 4.9 – The psTA spectra of 𝐴𝐴 𝐴𝐴𝐴𝐴 class of cMa compounds with excitation at 405 nm
in toluene with Cz kept constant as the amide ligand. The gold and copper complexes are
the top and bottom rows. 𝐴𝐴𝐴𝐴𝐴𝐴 𝐴𝐴 and 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴 compounds are the left and right columns,
respectively. ................................................................................................................................ 148
Figure 4.10 – Simulation of the ESA by summation of cation and anion basis spectra
(black) of (a) C 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴 using PR data and (b) 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 using BE data with
comparisons made to the triplet (red) and singlet (blue) SADS from psTA. The PR triplet
data for 𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴 was used in place of the triplet spectra. The inverted absorption
(dashed purple) and emission spectra (dashed orange) are displayed. ....................................... 151
Figure 4.11 – 𝐴𝐴𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 nsTA spectra as neat solutions (top row), with MePI as an
electron acceptor (middle row), and BIH as a hole acceptor (bottom row) in THF (left
column) and toluene (right column, except (f) which is in THF). Plot (f) is generated from
the experiment in (e) after averaging together the long time spectra (>100 ns) for the spectra
with (black) and without BIH (red). All spectra were collected on a Magnitude enVISion
except for (c) which was performed on an enVISTA (see Experimental). All data was
pumped with 355 nm except for (e) and (f) which were pumped with 450 nm. ........................ 154
Figure 5.1 – (a) Photochemical isomerization reactions of benzene and HFB yielding
different product distributions.
5
(b) Photochemical cascade reaction of Burns and co-
workers to form a fluorinated ladderene.
4
(c) Subsequent polymerization and
mechanochemical unzipping of the fluorinated ladderene to form fluorinated
polyacetylene. ............................................................................................................................. 207
Figure 5.2 – Theoretically computed gas-phase absorption of HFB in black and the
measured UV−vis absorption spectrum of HFB solvated in ethanol in dashed gray. The
transition-specific computed spectra are rendered below in solid-colored lines with upper
state symmetry assignments shown in the legend. The inset panel shows the vertical
excitation energies for the four lowest excited states of HFB. All calculations used the aug-
xxi

cc-pVDZ basis set, CAS and (XMS-CASPT2) calculations included a (6, 7) active space
described in the Experimental Section in Lopez et al.
2‡
............................................................. 216
Figure 5.3 – Absorption and emission spectra for HFB (red, main panel) and benzene
(blue, inset panel). Absorption spectra are rendered in solid lines and emission spectra in
filled traces. Both absorption and emission spectra for HFB are measured using a 75 μM
solution in ethanol, while the analogous benzene spectra are adapted from Du et al.
24
and
the extinction values calibrated by Berlman.
25
The black stick spectra represent
vibrationally resolved absorption and emission spectra computed from Franck−Condon
factors of the ground- and
1
B2u excited state gas phase minimum structures.
Franck−Condon factors are computed using the parallel mode approximation at 0 K. ............. 216
Figure 5.4 – (Left) Side and top elevations of the minimum energy geometries for the 𝑆𝑆 0
(
1
A1g) and 𝑆𝑆 1 (
1
B2u) states. (Right) Cuts through the ground,
1
A1g (black), and first four
excited,
1
B2u (orange),
1
E1g (green),
1
E1g (blue), and
1
B1u (red), potential energy surfaces,
along the ν18 e2u motion corresponding to opposing carbons moving symmetrically out of
plane. ........................................................................................................................................... 218
Figure 5.5 – Transient absorption spectra for 200 mM HFB in ethanol. The main panel
shows transient absorption spectra at pump−probe delays indicated by the color-coded
legend (in picoseconds) following photoexcitation at 255 nm (4.86 eV). The inset shows
the time evolution on a linear scale from −1 to 1 ps pump−probe delay and a logarithmic
scale from 1 to 1000 ps delays. For sticks and simulated bands see the text.............................. 220
Figure 5.6 – Transient absorption spectra of benzene (green) and HFB (blue) taken 1 ps
after excitation with 266 nm. The concentrations of benzene and HFB were ∼4 M and 400
mM, respectively. These spectra were measured at 1 ps to avoid the rise time of the benzene
excimer and intermolecular dynamics. ....................................................................................... 224
Figure 5.7 – A 2D cut through the 𝑆𝑆 0
1
A1g (blue) and 𝑆𝑆 1
1
B2u (red) gas-phase potential
energy surfaces. The two coordinates correspond to the angle of the two para carbons,
indicated by the orange and green arrows, rising out of the plane defined by the other four
carbons. Motion diagonally across the PES from the D6h minimum in the left-hand corner
(A) to the right-hand corner corresponds to both carbons rising symmetrically out of the
plane to form the bridgehead bond and yield the Dewar-HFB isomer (B). Here both carbons
are out of plane by 76°. Partial motion along the diagonal, where carbons are out-of-plane
by 8° and corresponds to the minimum energy geometry of the
1
B2u state (C). Motion along
only one axis, where only one carbon rises out of plane, leads to the MECI at 65° (D). From
this MECI, internal conversion to the ground state leads exclusively to the D6h minimum
(A). However, from the MECI motion of the other carbon is not degeneracy breaking until
40°. Molecules which pass through the CI at the upper portion of the seam are on a portion
of the ground state surface which bifurcates. One trajectory leading back to (A) while the
other trajectory leads to the Dewar form (B). ............................................................................. 226
Figure 5.8 – (Left) Minimum Energy CI geometry in HFB with atomwise contributions to
the ĝ vector. (Right) Cut of the ground (blue) and first excited (red) states along the
branching space ĝ vector for HFB (solid) and benzene (dashed). .............................................. 228
xxii

Figure 5.9 – (Top) Time evolution of the population of each electronic state. Green trace
shows S2, orange trace shows S1, and black trace shows S0. (Bottom) Time evolution of
reactive C−C bond distance for full ensemble of 400 trajectories. ............................................. 230
Figure 6.1 – a) Reaction scheme leading to PFA. Intermolecular [2+2] photocycloaddition
between HFB and Dewar benzene followed by an additional intramolecular [2+2]
photocycloaddition to form the fluorinated ladderane. Then a ROMP is performed, and
mechanical force opens the ladderane into a polymer, PFA. b) Similar reaction scheme of
HFB with norbornene as to b). The full saturation of the backbone of the norbornene unit
prevents the photoproduct from undergoing ROMP. ................................................................. 257
Figure 6.2 – a) Steady state fluorescence spectra of HFB at varying norbornene
concentrations. The semi-transparent region in grey was removed from integration due to
emission from NB impurities. The unscaled emission from the NB stock solutions is
displayed as dashed lines. b) corresponding Stern-Volmer plot. An inset of the low [NB]
is displayed as well. The error bars for all [NB] are present but only easily visible in the
inset and for the maximum [NB], 0.5 M NB, where they are ± 0.006 M. Error bars for b)
are discussed in section 6.7.4. A best fit line is applied in red with fitting parameters
displayed in the legend................................................................................................................ 265
Figure 6.3 – TCSPC trace of a) 1 mM HFB and b) 500 mM NB stock solutions in ethanol
with excitation at 266 nm under identical conditions. The time constants for the decays are
displayed on the figure with the amplitude percentages of the decays included after the
semicolon. ................................................................................................................................... 270
Figure 6.4 – a) The set of TCSPC decay traces of HFB with varying concentrations of
norbornene in ethanol. The [NB] of 0.2 to 0.5 are displayed with open symbols to improve
visibility with t > 15 ns. The HFB with [NB] = 0 M trace is the same trace from Figure
6.3a. b) corresponding SV plot using lifetimes from a) where the fitting parameters for the
linear fit are displayed. The error bars for all [NB] are present but only easily visible in
the inset and for the maximum [NB], 0.5 M NB, where they are ± 0.006 M. Error bars for
b) are discussed in section 6.7.4. A best fit line is applied in red with fitting parameters
displayed in the legend................................................................................................................ 271
Figure 6.5 –
19
F NMR of 40 mM HFB in CD3OD displayed as the a) full spectrum and b)
focus on <0.5% intensity of a). c) Focus on the region of Dewar-HFB. Peak areas are given
by 𝐴𝐴 𝜏𝜏 1,2. Integration limits are given by the dotted black lines. Red – before irradiation,
blue – after irradiation. All spectra were normalized at the HFB peak maximum in their
respective spectra. ....................................................................................................................... 275
Figure 6.6 –
19
F NMR of 40 mM HFB and 250 mM NB in CD3OD displayed as the a) full
spectrum and b) focus on <2% intensity of a). Red – before irradiation, blue – after
irradiation. Both spectra were normalized to 1 at the HFB peak maximum of the before
experiment. c)
19
F NMR spectra of HFB-only (blue) and HFB with NB (red). ......................... 279
Figure 6.7 –
19
F NMR of HFB and norbornene in CD3OD after irradiation (blue) with
predicted photoproducts. The predicted
19
F NMR spectral lines for photoproducts are
xxiii

displayed as drop lines with colors (black squares for P1, red circles for P2 and green
triangles for P3), normalized to 0.01 intensity. The corresponding fluorines are displayed
above their predicted positions. .................................................................................................. 280
Figure 6.8 – Possible fluorinated photoproducts in the irradiation study. The predicted
chemical shifts are displayed below each compound ordered by the indicated fluorine
numerical label, e.g., for P4, F1 is -153.41, F2 is -158.75, and F3 is -160.73. ............................. 282

xxiv

Abstract
The photophysics, excited state kinetics, charge transfer dynamics, and reaction
mechanisms of several photochemically active systems were studied with ultrafast spectroscopy
techniques. The process of symmetry breaking charge transfer (SBCT) for a zinc dipyrrin was
explored via a solvent dependent study which modulated the intramolecular charge transfer rates
to establish an energy regime where the driving force was zero (∆G = 0). This was established via
femto- to nanosecond transient absorption (psTA) and fluorescence quantum yield. Subsequently,
intermolecular charge transfer conditions were mapped out for an entire class of carbene-metal-
amide (cMa) compounds for use as solar fuel photosensitizers, assigned by spectral descriptions
of the neutral and charged species via pulse radiolysis and bulk electrolysis. The intersystem
crossing rates (~3 − 20 ∙ 10
9
𝑠𝑠 − 1
) were determined by time-correlated single photon counting
(TCPSC) and corroborated by psTA while the charge transfer kinetics to a diffusive electron donor
or acceptor was mapped with a new nano- to microsecond transient absorption spectrometer, an
instrument whose use in the UV is described herein.
In a project aimed at strained ring formation via photochemical pathways, the photophysics
and ultraviolet photochemistry of hexafluorobenzene (HFB) was explored using absorption,
emission, psTA, multiconfigurational computations, and non-adiabatic dynamics simulations to
develop a complete description of the dynamics of the 4π-electrocyclic intramolecular ring-closing
of HFB and the per-fluoro effect on the reaction. The intermolecular [2+2] cycloaddition of HFB
with an olefin was then explored via fluorescence lifetime quenching and
19
F NMR after UV
irradiation which successfully produced solid photoproduct.

1

Chapter 1. General Introduction
1.1. The Need to Capture the Sun
Currently, as a globally wide community, we stand on the precipice of world-changing
environmental phenomena caused by decades of use of unsustainable, unrenewable, and
atmospherically damaging energy sources. Earth surface temperature recordings by the National
Aeronautics and Space Administration (NASA) and the Goddard Institute for Space Studies
(GISS) indicate the temperature anomaly has risen to 0.9°C, a steady linear increase from 1980 to
now (Figure 1.1).
1,2
The GISS-NASA values are in conjecture with the Climatic Research Unit
(CRU) and the National Oceanic and Atmospheric Administration (NOAA). The scientific
community is in excellent agreement (97-100%)
3–6
that the current rise in global temperature past
1880 is human driven via fossil fuel use. The consensus value reaches ~100% among climate
scientists.
4
Higher Earth surface temperatures introduce a larger amount of internal energy into
Earth’s atmosphere which opens up a myriad of negative effects such as sea level rise,
7,8
ocean
acidification,
9,10
more dramatic weather patterns,
11
and food chain destruction.
12,13
Therefore, if
humankind wants at most to avoid and at least to adapt to the most negative effects of climate
change, we must abate climate change effectively and quickly. Fortunately, the scientific
community has recognized the dangers of unmitigated fossil fuel use and has been pursuing
alternative electricity generation and fuel sources to renovate and overhaul our energy
infrastructure.
Life on Earth continues to thrive through the harnessing of energy from the sun and would
cease without it. Every second, a vast number of organisms utilize the sun’s power to live through
the process of photosynthesis. In photosynthesis for most organisms, carbon dioxide and water are
converted into sugars and oxygen via the capturing of light energy from the sun.
14,15
The sugars
2

provide energy in the form of chemical bonds for the plants, serving as nature’s essential fuels.
Photosynthesis is such a useful process that humans have attempted to recreate it synthetically and
industrially through various means.
16,17
While photosynthesis requires a complex arrangement of
plant-based chromophores, proteins, enzymes, etc., the scientific research community at large can
still borrow design principles from nature for renewable energy sources to recreate nature’s
success, such as hydrogen equivalents from NADPH production.
14

The sun provides the largest source of renewable energy at an output of 23,000 TW, in
comparison to global human consumption, 16 TW (Figure 1.2).
18
The possible amount of solar
energy supply far outcompetes the current power demand and compared with the finite resources
of fossil fuels, solar energy has an essentially infinite supply.
18
Additionally, solar energy has the
possibility to generate energy without major environmental drawbacks such as toxic waste from
nuclear energy or large changes to the environment such as in hydro power generation.
19

Therefore, employing the sun as a renewable source of energy is an attractive solution to the
current energy and climate crisis. However, a large caveat exists wherein utilizing the sun’s energy
reserve is only possible if humankind is able to harness the energy from the sun effectively.
Fortunately, several avenues of solar energy capture exist with current and rapid improvements
made in many sectors. In this thesis, we will explore two of those avenues for attacking the
problems for both stationary (organic photovoltaics) and mobile energy demands (solar fuel
photosensitizers).

3

1880
1900
1920
1940
1960
1980
2000
2020
-0.5
0
0.5
1
Temperature Anamoly (
o
C)
Time (Year)
Annual Mean
Lowess(5)
(a)
-1
0
1
2
Temperature Anamoly (
o
F)

----------------------------------------------------------------------------------------------------------------
b)

Figure 1.1 – (a) Graph of the global surface temperature anomaly as recorded by GISS/NASA.
b) Recreated visualization of surface temperature for past years and the current year, clockwise
from top left: 1890, 1950, 2022, 1990. Temperature differences (°F) colder than the average,
below 0°C tend blue and hotter temperatures tend orange then red. Short-term changes are
smoothed on a five-year running point average between consecutive years. Figures in b)
obtained from climate.nasa.gov. Data acquisition and smoothing parameters listed in Lenssen
et al.
4

1.2. The Case for Photovoltaics
The sun’s light energy can be transferred into electrical energy through a device called a
photovoltaic (PV) solar cell. A full description of a PV cell will only be explained briefly here. A
standard PV cell is composed of several molecular layers. For a standard PV cell, the active layer
is composed of semiconducting materials which absorb incident light, and through the
photoelectric effect, electron-hole pairs are created, known as excitons. These excitons then
migrate to grain boundaries within the PV cell which provides a sufficient and inherent electric
field to split the coulombically attracted particles into the respective electrons and holes. These

Figure 1.2 – Comparing finite and renewable planetary energy reserves (Terawatt-years). Total
recoverable reserves are shown for the finite resources (outside yellow circle). Yearly potential
is shown for the renewables (inside yellow circle). Adapted from Perez and Perez et al.(Perez
2022)
5

carriers then migrate to the electrodes and are harvested as electricity. Various iterations of PV
cells exist, with differing active layer thickness, geometry, and material makeup.
A seminal paper by Shockley and Queisser detailed the ideal limit for a single junction
solar cell’s efficiency under standard thermodynamic and solar conditions.
20
For a cell with a
bandgap of 1.1 eV (silicon), the maximum detailed balance limit efficiency is 30%. The efficiency
of a solar cell has become the gold standard for determining how upcoming solar cells can be
viable in the renewable energy market.
For some historical background information, the photovoltaic effect was discovered in
1839 by Edmond Becquerel
21
. The first solar cell was created by Charles Fritts in 1883 using
selenium to an efficiency of <1%.
22
In 1954, Bell Labs introduced the p-n junction silicon solar
cell and 3 years later an efficiency of 8% was recorded by Pearson, Chapin, and Fuller.
23
As of
September 2022, silicon-based PV cells account for 95% of renewables, of which there is 0.9% of
total energy generation in 2015
24
with efficiencies reaching 24-26.6% (Figure 1.3, purple
symbols).
25
The prevailing use of silicon is due to being abundant raw materials, well understood,
and heavily researched and displacing them is highly unfeasible.
26
However, silicon suffers from
several drawbacks. Under standard production rates, most current silicon solar cells require large
areas of land, often placed in regions of desert, obstructing wildlife and residents.
27
Recently,
advancements in ultrathin technology by MIT for silicon have been made in which the previously
required thick solar cells can be replaced with 50 µm thin PV systems.
28
This limitation was one
of the primary issues plaguing silicon solar cells.
While silicon based solar cells dominate the renewable energy market, there have been
significant strides made in materials research using other material classes such as GaAs, GaInP,
CdTe, perovskite, and quantum dots (Figure 1.3).
25,29
Organic-inorganic hybrid perovskites
6

specifically have seen an unmatched increase of efficiency per year (Figure 1.3, yellow circles),
achieving ~26% in only 10 years. Researchers have also utilized multi-junction solar cells,
combining multiple materials to achieve record efficiencies of 48% (Figure 1.3, purple dotted-
squares), far surpassing the Shockley-Queisser limit.
Another possibility has risen within recent years: organic photovoltaics (OPV).
30
OPVs
are fabricated where the main light harvesting layer(s) consists organic chromophores, usually
made mostly of organic dyes or conductive organic polymers. High efficiency OPVs are fabricated
with dyes that possess strong absorptive characteristics ( 𝜖𝜖 ~10
4
− 10
5
𝐴𝐴 − 1
𝑘𝑘 𝑀𝑀 − 1
), tunable
bandgaps across the visible region, and long excited state charge transfer lifetimes. Long excited
state lifetimes provide sufficient time for excitons to diffuse to grain boundaries so they can be
split into respective electrons and holes. The strong absorptivity and tunability allows for short
enough pathlengths so that OPVs can be fabricated as flexible, thin, partially transmissive sheets,
an advantage over silicon cells which are opaque in the visible region. These sheets can then be
placed upon windows, able to harvest light during the day while still permitting light to enter
inside.
7

Figure 1.3 – Chart of the highest efficiency research solar cells as a function of year and class of material and design as of September
2022. Figure is courtesy of National Renewable Energy Laboratory, NREL. Data adapted by NREL from Green 2023.
8

While promising, OPVs still suffer from several other drawbacks leading to low solar cell
efficiency, best at 15-18% (Figure 1.3, red closed cells) when compared to other leading materials.
Here we will focus on two: 1) charge separation after exciton formation and 2) absorptivity in the
NIR region of the spectrum. The electron and hole in OPVs are generated on single chromophores
and due to weak dielectric screening in the low dielectric, the hole and electron possess strong
Coulombic attraction, limiting efficient charge separation (CS). Poor CS leads to geminate charge
recombination back to the ground state, either radiative or non-radiatively, leading to efficiency
loss. Some progress has been made here by fabricating solar cells as finely mixed regions (1-10
nm) of donor and accepter – known as bulk heterojunction solar cells. Here, diffusion to a
boundary to promote CS is more effective, reducing the need for large diffusion lengths.
Additionally, most OPV based chromophores absorption maximum lies in the UV-vis region of
the solar spectrum (~400 to 600 nm). While the highest intensity irradiance of sun is positioned in
the visible around 500 nm (Figure 1.4, blue spectrum), the number of photons maximizes at 600
nm (Figure 1.4). More importantly, the number of available photons is greatest in the red-NIR
region of the solar spectrum. OPVs with chromophores absorbing in the UV-vis region are
therefore losing most of the impingent solar photons. Several research groups have worked to
extend chromophore absorption and emission (for LEDs).
31,32
Conversely, in the NIR, excitons in
OPVs (and any chromophore) struggle to achieve long lifetimes and high efficiencies due to
competition with deactivation as dictated by the energy-gap law.
31,33

We introduce our work here. The goal of the first project was to couple synthetic work
which extended the absorption of the chromophore further red while not losing redox power, to
physical chemistry to understand the excited state dynamics of these dyes. It has become well-
known that difluoroboron dyprrins (BODIPY) are a promising class to address the aforementioned
9

issues due to strong synthetic tunability, long lifetime and strong absorptive capability.
34

Therefore, chapter 3 explores the use of a molecule that is a promising system to address both OPV
shortcomings.

1.3. Generating solar fuels
Fossil fuels have been in use since the Industrial Revolution and are the major sources of
humankind’s energy infrastructure. One of the reasons for their continued use is because fuels like
gasoline (hydrocarbons made of chains of 5-11 carbon atoms) exist as liquids and can be easily
transported. Methane can be compressed and transported as liquified natural gas (LNG). These
500 1000 1500 2000 2500
0
0.5
1
1.5
2
2.5
Solar Irradiance (W m
-2
nm
-1
)
Wavelength (nm)
AM0
AM1.5
O
3
O
2
H
2
O
H
2
O
H
2
O
CO
2
Silicon
VIS UV NIR
0
1
2
3
4
5
Solar Flux (10
18
Photons m
-2
nm
-1
)
6 4 3 2 1 0.5
Energy (eV)

Figure 1.4 – Standardized spectral irradience of the sun, collected above earth’s atmosphere
(Extraterrestrial, red) on earth’s surface (global, blue) with an air mass of 1.5 (solar zenith angle
of 48°). Integration of the extraterrestial spectrum yields 1366.1 W/m
2
and the global spectrum
yields 1000 W/m
2
. Dips in the direct spectrum are attributed to absorption by the overtones of
vibrations of the indicated molecules. The pink spectrum displays the photon flux (number of
photons per unit wavelength per second) of AM1.5 by dividing the solar irradiance by the
photon’s energy.
10

fuels are usually produced via fractal distillation of crude oil, cracking (splitting large chains into
smaller ones), or isomerization (converting various C-chains into smaller, more efficiently burned
fuels).
35
Renewable electricity from photovoltaics address the complications of replacing fossil
fuels for built infrastructure, e.g., homes, offices and industrial plants, but are not well suited for
the transportation sector such as automobiles, trucks, boats, and airplanes.
36
A goal of renewable
energy work is to replace fossil fuels leveraging the infrastructure already built out for liquid fuels.
To accomplish this, research has been performed to create catalytic systems that capture the sun’s
energy to make liquid solar fuels. Renewable fuels are generated from abundant materials like
CO2 and H2O in carbon dioxide reduction and water splitting processes, respectively.
37
Fuels such
as methanol and hydrogen are generated and can be used as fuels. Additionally, via the Fischer
Tropsch process, liquid hydrocarbons can be produced from syngas, hydrogen and carbon
monoxide, or organic type-wastes.
38,39

The process for generating fuels from the sun can be performed in several ways. A PV cell
can be wired to drive current to electrodes immersed in water. This is the process of electrolysis
and is how hydrogen is often made while also generating oxygen. However, these electrodes are
often made of platinum, a rare and expensive metal; avoiding using Pt is ideal for widespread use
and lower costs. The electrodes can be replaced with molecular catalysts. Here, the PV cell is wired
to an electrode where the catalyst is deposited on the surface and solar generated charges from the
PV supplied to the catalyst for hydrogen production or carbon dioxide reduction.
40

Alternatively, a simpler wire-free approach would be preferable.
41
In this arrangement, a
molecule specifically designed to absorb light and donate charge can be used, known as a
photosensitizer (PS).
42,43
Photosensitizers are possible solutions to not only energy applications
but biomedical as well as in photodynamic therapy.
44
The PS can then be used to reduce/oxidize
11

an electrocatalyst (EC). The PS and EC can either be dissolved in solution
45
as diffuse monomers
or linked together covalently. We will explore the PS-EC construct for generation of fuels with
Figure 1.5. We begin at stage 1 wherein, upon absorption of light, the PS is promoted to its excited
state (PS*) to transition to stage 2. Usually for PS materials, the lowest energy excited state is an
intramolecular charge transfer (ICT) state, commonly a metal to ligand (MLCT) or ligand to ligand
(LLCT) charge transfer state.
42,46
The inherent ICT is useful as it prepares the system for
intermolecular charge transfer. The PS* possesses simultaneously stronger oxidizing and reducing
capabilities than ground state PS.
47
In stage 3, the PS* diffuses to the EC and donates an electron
to produce EC
–
; PS* is oxidized to PS
+
. In turn, the PS* provides sufficient chemical potential for
the EC to do useful catalytic reduction. With appropriate choice of PS-EC pair, the system can be
fed feedstock such as CO2 and H2O to produce useful fuels such as H2, formaldehyde, and
methanol. Finally, in stage 4, the PS
+
encounters a third molecule, a sacrificial reductant (SAC)
to which the PS
+
donates the remaining hole to recover the ground state PS. More ideally, PS
+
can
deliver the holes to oxidize water to liberate oxygen so as not to consume stoichiometric amounts
of the sacrificial agent. Regardless, the cycle is then primed to begin again.
Use of the solar fuels, for example in transportation, provide a second closed loop leading
to a more carbon neutral economy. Carbon dioxide capture, reduction and combustion leads to a
net carbon-neutral fuel economy as the combusted fuels (Methanol, methane, acetone) serve as
fuel for the next carbon capture part of the cycle.
48
Water electrolysis yields hydrogen gas,
combustion of which leads back to water and thus no carbon dioxide production at all, hence its
attractive solution to the fuel discussion.
There are several notes we wish to make. First, this is not the only way the PS-EC pair can
interact. The SAC can act as the electron donor after stage 2, to produce a ground state PS anion,
12

PS
–
, which then proceeds to reduce the catalyst. Additionally, the charges on the PS* can be
reversed where the PS* can act as an oxidizing agent, donating a hole to an EC to produce PS
–
and
EC
+,
ready to oxidize a substrate like water to oxygen. For a full solar fuel generation system, the
PS* should be able to donate both an electron and a hole to two different ECs.
Most photosensitizers (and electrocatalysts) to date have used expensive, rare, second and
third row transition metals like ruthenium, rhodium, and iridium. Their continued use is due to
prototypical complexes like 𝑅𝑅 𝐴𝐴 ( 𝑏𝑏𝑏𝑏𝑏𝑏 )
3
2 +
and 𝑘𝑘𝑀𝑀 ( 𝑝𝑝𝑝𝑝 𝑏𝑏 )
3
possessing long lifetimes, strong absorption
capabilities and powerful redox potentials. These materials effectively utilize an MLCT state as
the redox active species. While the most effective PS complexes use platinum group metals (Ru,
Ir, Ru), the desire is to move to more abundant metals, such as the first-row transition metals. For
2
nd
and 3
rd
row transition metals, an MLCT state is the lowest energy excited state. However, for
first row transition metals, the lowest energy excited state is not an MLCT band but rather a d-d
transition as the MLCT state now lies higher in energy. For metal atoms with d orbital vacancies,
such as Fe(II) and Co(II), these lower lying d-d states lead to rapid (<1 ps) deactivation of the
MLCT state as explored by McCusker et al.
49
One way around this and still use a first-row
transition metal, is to exploit a metal with d
10
configuration, like Cu(I). As all the d orbitals are
filled, no d-d excitations exist in the system. In these complexes, both MLCT and LLCT transitions
are possible and may lie lowest. By this approach, we can potentially avoid rare metals and still
obtain long excited state lifetimes. We will show that by using specific LLCT transitions we can
achieve record excited state lifetimes ~ 1 µs, that are long enough to allow charge transfer to an
EC simply by bimolecular diffusion.
In chapter 4, we will focus on a specific class of carbene-metal-amide (cMa) complexes
which use copper and gold, more common and less expensive central metals. These cMa
13

complexes boast strong absorption tunability,
50
high excited state redox potentials
47
and long
lifetimes.
51
We will build off previous work and currently examine the charge transfer capabilities
of these species with which we demonstrate their usefulness as possible PS materials. Preliminary
work in our groups is now demonstrating their efficacy and competitiveness as photosensitizers
for hydrogen production upon incorporation with a nickel catalyst.

Figure 1.5 – Cartoon representation of a reductive catalytic cycle initiated by a photosensitizer.
Blue box – photosensitizer (PS); red circle – electrocatalyst (EC); green pentagon – sacrificial
reductant (SAC). Charged states of the species are labeled as well, i.e. PS
+
is the oxidized
photosensitizer. The stages of the cycle are denoted by their respective numbers below the PS-
EC pair as circled numbers. Activity between the PS – EC pair is denoted by labeled black
arrows. See text for full description of catalytic cycle.

PS Recovery
EC
–
PS
+
EC
PS
+
EC
PS
C
O
O
O
H H
H
H
HO
CH
3
SAC
+
SAC
Renewable
Solar Fuels

4
C
O
H H
14

1.4. Fundamental Photophysics and Photochemistry
In addition to large organic complexes, thesis work has also focused on more fundamental
organic photochemistry, still utilizing light as a reagent as introduced in the previous section, 1.3.
Organic chemistry can be thought of as a vast web of interconnected reactants, products, reaction
conditions, temperatures, pH measures, etc
52,53
(Figure 1.6a, blue circles). This hypergraph of
largely infinite possible combinations is produced from only a select group of main block elements
(carbon, oxygen, nitrogen, hydrogen, and fluorine). Nodes in this network can be thought of as
stable organic molecules and the lines between these molecules are reaction conditions and
experiments (Figure 1.6a, arrows). Organic chemists have been working for centuries to unravel
this web by iterating upon reaction conditions to synthesize molecules of interest. In this
succeeding example, we will focus on a small subset of these molecules.
Organic synthesis is also analogous to an infinite mountain range where the height of the
peak corresponds to energy levels. We can imagine a scenario in which a mountain skier is scaling
a large mountain range with large collections of pine trees or hills that greatly reduce the pathways
a climber can see certain part of the range (Figure 1.6b). Transversing this range is often possible
by incorporating thermal energy to overcome certain energetic barriers; just as the mountain
climber must input energy into their climb to scale the mountain. After a day of skiing, the climber
then realizes they are trapped within a small valley along the range (A), with no visible escape but
can still setup camp safely. Chemically, we see this scenario, A, as our starting reactant. Without
a possible escape by foot, as the trees prevent reaching base camp, the climber calls in a helicopter
for rescue. This helicopter arrives and can transport the skier to an even higher peak (A
*
), from
which the climber can safely ski down the highest mountain into B and finally X.
15

Figure 1.6 – a) Cartoon representation of web of organic molecules (letters), linked by known
reactions (blue arrows), undiscovered reactions (red arrows), and reversible reactions (double
headed arrows). A synthesis path for molecule X from the possible starting materials is
currently unavailable in this diagram. b) Cartoon diagram of Mt Baldy, CA, adapted from the
official map with all overlays and markers removed for clarity. Letters – various summits; red
line – helicopter path; black line – skiing path.
A
B
C
D
E
F
G
H
I
X
a)
X
A A
A
*
B B
Base camp
b)
16

In the chemical picture, we see B as some excited state reaction intermediate and X as our
desired product. Certain molecules do not possess thermal synthesis pathways because of large
energy barriers between reactant and products and thus not accessible using thermal energy alone.
Energy required to perform chemical reactions need not be supplied solely by thermal energy
alone. Energy, in the form of light can provide a route to molecules previously prohibited by large
thermal barriers. From the mountain analogy, the large altitude and fast speed of the helicopter
are analogous to the instantaneous and verticality light provides to molecules to reach the excited
state. The excited state provides the ability for molecules to see new synthetic pathways that were
previously prohibited.
Now, we turn to a more physical chemistry approach here. We return to molecules A and
X our reactant and products, respectively along a subset of the mutual potential energy surface
(Figure 1.7a). The ground state potential energy hypersurface is in blue; we introduce the excited
state of the A-X potential energy hypersurface in red. The ground state barrier prevents a ground
state pathway from A leading to X (reaction path 1). Excitation of A leads to A* which follows
the excited state potential energy surface into the energy well of X (reaction path 2). A* flows
across the surfaces from excited to ground state, affording strong flux from excited reactant to
ground state photoproduct (Figure 1.7b). A
*
to X now glides over the previously prohibited
ground state barrier which forms the potential energy well. An emphasis is placed on the barriers
that determine the need for excited state pathways, not absolute energy levels of molecules: it is
the barrier and the lack of a direct pathway that prevents ground state thermal chemistry and
requires the excited state. It should be noted this is a classically endergonic process. This need not
be the case and the photoproduct may indeed be lower in total energy compared to the reactant.

17

Light, most often in the visible and UV portions of the electromagnetic spectrum, can be
absorbed by a molecule to do useful photochemical work. We turn to photochemistry for new
pathways between reactant and product, providing synthetic access to molecules which cannot be
prepared by thermal means. Light driven reactions begin with the absorption of a photon by a
molecule to generate an excited state. While the focus of photochemistry is on the rearrangement
of the molecule itself (unimolecular) or reaction of a molecule with another molecule
(bimolecular), the understanding of the evolution of the molecular excited state must first be
understood. For a small group of molecules, the photophysics may be understood already and the
focus can be move to the photochemical reaction. However, when a new molecule is introduced,
first care must be taken understand its photophysical characteristics as a chromophore first and
reagent second so as to build understanding from the ground up. A simple example is
hexafluorobenzene or HFB. The replacement of the hydrogens with fluorines on the aromatic rings

Figure 1.7 – a) Energy surface plot, representing a pair of potential energy surfaces (ground
state – blue, excited state – red) as a function of two reaction coordinates (Rxn Coord). All
units are arbitrary. Reactant – A, product – X, excited state reactant – A*. Excitation to A*
from A is indicated by h 𝜈𝜈 . Thermal pathway is indicated by the dotted arrow “1” and excited
state reaction is indicated by the dotted arrow “2”. b) The plot in a) at a more extreme tilt angle
with emphasis on the excited state pathway. Product X lies on the ground state surface as
indicated by the intercepting blue lines over the red lines.

2
h
A
A*
X
A*
A
X
h
2
1
b) a)
18

leads to a dramatic change in both the photophysics and photochemistry going from benzene to
HFB. The absorption and emission spectra become broadened; the lifetime drops from 60 ns to
2.2 ns and the Stokes shift widens to ~110 nm!
54,55
And instead of several photoproducts produced
upon irradiation as in benzene, a single product from HFB, Dewar-HFB is formed. A complete
description of HFB’s unique light-dependent properties with emphasis on its
photochemoselectivity will be explored in chapter 5.
Organic chemistry has been historically approached with a rather practical philosophy: a
chemical product is desired from a set of reagents. Reaction conditions are then tuned exploiting
prior wisdom, chemical intuition, and essentially, trial and error determine whether a good
pathway exists. Navigating this network is the training for a budding organic chemist. However,
photochemistry presents with an even larger network of possibilities and variables and guiding
rules and intuition is more limited. This motivated a 5-year project involving several groups in
theory (Todd Martinez and Steven Lopez), organic synthesis (Noah Burns and Matt Kanan), and
experimental reaction dynamics (Stephen Bradforth) to apply a new overarching approach,
exploiting machine learning and high throughput spectroscopy, to impart some structure to
photochemical reaction discovery. The goal of this project was to create a wholistic design
approach to chemical synthesis by incorporating informative elements from organic chemistry
while being informed by theory and experiment (Figure 1.7). Essentially, the Martinez group
created a nanoreactor, a theoretical calculational construct, where starting materials are put into
harsh, high-energy conditions.
56
From it, a series of products are produced which are analyzed
theoretically to determine if they possess interesting characteristics (Martinez and Lopez). The
organic chemists (Burns), with the help of AI trained on a large set of reaction conditions can then
suggest possible improvements. The organic synthesis counterpart would also produce possible
19

materials for analysis (Kanan). And then high throughput spectroscopy can probe a series of
organically produced products for interesting properties (Bradforth), the results of which would be
feed back into the nanoreactor and AI to form a wholistic cycle to reaction discovery.

By incorporating photochemistry, the ability to reach previously unobtainable high energy
density molecules presents itself (Figure 1.8a,b) for high energy applications. An example is
cubane: a highly strained and therefore, highly energy dense molecule the emblematic target for
strained molecular synthesis(Figure 1.8a, first species).
57
Another set of photochemistry that our
project focused on was several [2+2] photocycloadditions
58
as these reactions can lead to strained

Figure 1.8 – Cartoon iterative feedback loop of the Synthesis Planning and Reaction Discovery
for Photochemistry and Chemistry in Novel Environments Project.
20

ladderanes (Figure 1.8a, third, fourth species). The Burns group have shown efficient and gram
scale synthesis of a ladderane species, present in the phospholipid layers of certain anammox
bacteria.
59
In general [2+2] reactions, upon absorption of light, a set of two double bonds, on
separate molecules transform into a pair of single bounds, yielding a cyclobutane ring. There are
two possible [2+2] cycloadditions wherein one of the olefins directly absorbs a photon (Figure
1.8c) or a complex formed between a metal atom, often a copper (I) atom serving as a catalyst,
absorbs a photon (Figure 1.8d).
60–62
We exam the first of these conditions. In chapter 6, HFB
serves the role as the “olefin” which absorbs light to do the [2+2] photochemistry. We extend our
exploration of HFB unimolecular photochemistry of chapter 5 by assessing isomerization quantum
yields and its bimolecular chemistry with a representative olefin to study [2 + 2]
photocycloaddition reaction mechanisms.

21

Figure 1.9 – a) A summary of energy-dense polycyclobutane targets. b) Summary of target
polycyclobutanes in the Syntheis Planning project. c) Photochemistry of an activated 𝛼𝛼𝛼𝛼 ketone
to form a cyclobutane aldehyde. d) Photochemistry of a Cu(I)-cyclobutene2 complex to form a
ladderane.
a)
b)
c) d)
22

1.5. References
(1) Schmunk, R. B.; Team, G. NASA Goddard Institute for Space Studies.
https://climate.nasa.gov/vital-signs/global-temperature/ (accessed 2023-04-10).
(2) Lenssen, N. J. L.; Schmidt, G. A.; Hansen, J. E.; Menne, M. J.; Persin, A.; Ruedy, R.; Zyss,
D. Improvements in the GISTEMP Uncertainty Model. J Geophys Res Atmospheres 2019, 124
(12), 6307–6326. https://doi.org/10.1029/2018jd029522.
(3) Cook, J.; Oreskes, N.; Doran, P. T.; Anderegg, W. R. L.; Verheggen, B.; Maibach, E. W.;
Carlton, J. S.; Lewandowsky, S.; Skuce, A. G.; Green, S. A.; Nuccitelli, D.; Jacobs, P.;
Richardson, M.; Winkler, B.; Painting, R.; Rice, K. Consensus on Consensus: A Synthesis of
Consensus Estimates on Human-Caused Global Warming. Environ Res Lett 2016, 11 (4),
048002. https://doi.org/10.1088/1748-9326/11/4/048002.
(4) Powell, J. Scientists Reach 100% Consensus on Anthropogenic Global Warming. Bulletin Sci
Technology Soc 2017, 37 (4), 183–184. https://doi.org/10.1177/0270467619886266.
(5) Lynas, M.; Houlton, B. Z.; Perry, S. Greater than 99% Consensus on Human Caused Climate
Change in the Peer-Reviewed Scientific Literature. Environ Res Lett 2021, 16 (11), 114005.
https://doi.org/10.1088/1748-9326/ac2966.
(6) Myers, K. F.; Doran, P. T.; Cook, J.; Kotcher, J. E.; Myers, T. A. Consensus Revisited:
Quantifying Scientific Agreement on Climate Change and Climate Expertise among Earth
Scientists 10 Years Later. Environ Res Lett 2021, 16 (10), 104030. https://doi.org/10.1088/1748-
9326/ac2774.
(7) MIMURA, N. Sea-Level Rise Caused by Climate Change and Its Implications for Society.
Proc Jpn Acad Ser B 2013, 89 (7), 281–301. https://doi.org/10.2183/pjab.89.281.
(8) Kwadijk, J. C. J.; Haasnoot, M.; Mulder, J. P. M.; Hoogvliet, M. M. C.; Jeuken, A. B. M.;
Krogt, R. A. A. van der; Oostrom, N. G. C. van; Schelfhout, H. A.; Velzen, E. H. van; Waveren,
H. van; Wit, M. J. M. de. Using Adaptation Tipping Points to Prepare for Climate Change and
Sea Level Rise: A Case Study in the Netherlands. Wiley Interdiscip Rev Clim Change 2010, 1
(5), 729–740. https://doi.org/10.1002/wcc.64.
(9) Koch, M.; Bowes, G.; Ross, C.; Zhang, X. Climate Change and Ocean Acidification Effects
on Seagrasses and Marine Macroalgae. Global Change Biol 2013, 19 (1), 103–132.
https://doi.org/10.1111/j.1365-2486.2012.02791.x.
(10) Hoegh-Guldberg, O.; Mumby, P. J.; Hooten, A. J.; Steneck, R. S.; Greenfield, P.; Gomez,
E.; Harvell, C. D.; Sale, P. F.; Edwards, A. J.; Caldeira, K.; Knowlton, N.; Eakin, C. M.; Iglesias-
Prieto, R.; Muthiga, N.; Bradbury, R. H.; Dubi, A.; Hatziolos, M. E. Coral Reefs Under Rapid
Climate Change and Ocean Acidification. Science 2007, 318 (5857), 1737–1742.
https://doi.org/10.1126/science.1152509.
23

(11) Koetse, M. J.; Rietveld, P. The Impact of Climate Change and Weather on Transport: An
Overview of Empirical Findings. Transp Res Part D Transp Environ 2009, 14 (3), 205–221.
https://doi.org/10.1016/j.trd.2008.12.004.
(12) Tirado, M. C.; Clarke, R.; Jaykus, L. A.; McQuatters-Gollop, A.; Frank, J. M. Climate
Change and Food Safety: A Review. Food Res Int 2010, 43 (7), 1745–1765.
https://doi.org/10.1016/j.foodres.2010.07.003.
(13) Binzer, A.; Guill, C.; Brose, U.; Rall, B. C. The Dynamics of Food Chains under Climate
Change and Nutrient Enrichment. Philosophical Transactions Royal Soc B Biological Sci 2012,
367 (1605), 2935–2944. https://doi.org/10.1098/rstb.2012.0230.
(14) Eberhard, S.; Finazzi, G.; Wollman, F.-A. The Dynamics of Photosynthesis. Annu Rev
Genet 2008, 42 (1), 463–515. https://doi.org/10.1146/annurev.genet.42.110807.091452.
(15) Farquhar, G. D.; Caemmerer, S. von; Berry, J. A. Models of Photosynthesis. Plant Physiol
2001, 125 (1), 42–45. https://doi.org/10.1104/pp.125.1.42.
(16) Fleming, G. R.; Schlau-Cohen, G. S.; Amarnath, K.; Zaks, J. Design Principles of
Photosynthetic Light-Harvesting. Faraday Discuss 2011, 155 (0), 27–41.
https://doi.org/10.1039/c1fd00078k.
(17) Benniston, A. C.; Harriman, A. Artificial Photosynthesis. Mater Today 2008, 11 (12), 26–
34. https://doi.org/10.1016/s1369-7021(08)70250-5.
(18) Perez, M.; Perez, R. Update 2022 – A Fundamental Look at Supply Side Energy Reserves
for the Planet. Sol Energy Adv 2022, 2, 100014. https://doi.org/10.1016/j.seja.2022.100014.
(19) Balzannikov, M. I.; Vyshkin, E. G. Hydroelectric Power Plants ’ Re Servoirs And Their
Impact On The Environment. Environ Technology Resour Proc Int Sci Pract Conf 2015, 1, 171–
174. https://doi.org/10.17770/etr2011vol1.885.
(20) Shockley, W.; Queisser, H. J. Detailed Balance Limit of Efficiency of P‐n Junction Solar
Cells. J Appl Phys 1961, 32 (3), 510–519. https://doi.org/10.1063/1.1736034.
(21) Becquerel, E. Recherche Sur Les Effets de La Radiation Chimique de La Lumière Solaire,
Au Moyen Des Courants Électriques. Comptes Rendus 1839, 9, 145–149.
(22) Fritts, C. E. On a New Form of Selenium Cell, and Some Electrical Discoveries Made by Its
Use. Am J Sci 1883, s3-26 (156), 465–472. https://doi.org/10.2475/ajs.s3-26.156.465.
(23) CHAPIN, D. M.; PEARSON, G. L.; FULLER, C. S. Semiconductor Devices: Pioneering
Papers. 1991, 969–970. https://doi.org/10.1142/9789814503464_0138.
24

(24) Woodhouse, M.; Jones-Albertus, R.; Feldman, D.; Fu, R.; Horowitz, K.; Chung, D.; Jordan,
D.; Kurtz, S. The Role of Advancements in Solar Photovoltaic Efficiency, Reliability, and Costs;
Golden, CO: National Renewable Energy Laboratory, 2016.
(25) Green, M. A.; Dunlop, E. D.; Siefer, G.; Yoshita, M.; Kopidakis, N.; Bothe, K.; Hao, X.
Solar Cell Efficiency Tables (Version 61). Prog. Photovolt: Res. Appl. 2023, 31 (1), 3–16.
https://doi.org/10.1002/pip.3646.
(26) Chaar, L. E.; lamont, L. A.; Zein, N. E. Review of Photovoltaic Technologies. Renew
Sustain Energy Rev 2011, 15 (5), 2165–2175. https://doi.org/10.1016/j.rser.2011.01.004.
(27) Wainwright, O. How Solar Farms Took over the California Desert: ‘An Oasis Has Become
a Dead Sea.’ The Guardian. Desert Center, CA May 21, 2023. https://www.theguardian.com/us-
news/2023/may/21/solar-farms-energy-power-california-mojave-desert.
(28) Saravanapavanantham, M.; Mwaura, J.; Bulović, V. Printed Organic Photovoltaic Modules
on Transferable Ultra‐thin Substrates as Additive Power Sources. Small Methods 2023, 7 (1),
2200940. https://doi.org/10.1002/smtd.202200940.
(29) Bahrami, A.; Mohammadnejad, S.; Soleimaninezhad, S. Photovoltaic Cells Technology:
Principles and Recent Developments. Opt Quant Electron 2013, 45 (2), 161–197.
https://doi.org/10.1007/s11082-012-9613-9.
(30) Kippelen, B.; Brédas, J.-L. Organic Photovoltaics. Energ Environ Sci 2009, 2 (3), 251–261.
https://doi.org/10.1039/b812502n.
(31) Zampetti, A.; Minotto, A.; Cacialli, F. Near‐Infrared (NIR) Organic Light‐Emitting Diodes
(OLEDs): Challenges and Opportunities. Adv. Funct. Mater. 2019, 29 (21), 1807623.
https://doi.org/10.1002/adfm.201807623.
(32) Lee, W. W. H.; Zhao, Z.; Cai, Y.; Xu, Z.; Yu, Y.; Xiong, Y.; Kwok, R. T. K.; Chen, Y.;
Leung, N. L. C.; Ma, D.; Lam, J. W. Y.; Qin, A.; Tang, B. Z. Facile Access to Deep Red/near-
Infrared Emissive AIEgens for Efficient Non-Doped OLEDs. Chem Sci 2018, 9 (28), 6118–
6125. https://doi.org/10.1039/c8sc01377b.
(33) Englman, R.; Jortner, J. The Energy Gap Law for Radiationless Transitions in Large
Molecules. Mol Phys 1970, 18 (2), 145–164. https://doi.org/10.1080/00268977000100171.
(34) Loudet, A.; Burgess, K. BODIPY Dyes and Their Derivatives: Syntheses and Spectroscopic
Properties. Chem Rev 2007, 107 (11), 4891–4932. https://doi.org/10.1021/cr078381n.
(35) Britannica. Gasoline; Encyclopaedia, T. E. of, Ed.; Britannica, E., Series Ed.; 2023.
(36) Auttawaitkul, Y.; Pungsiri, B.; Chammongthai, K.; Okuda, M. A Method of Appropriate
Electrical Array Reconfiguration Management for Photovoltaic Powered Car. Ieee Apccas 1998
25

1998 Ieee Asia-pacific Conf Circuits Syst Microelectron Integrating Syst Proc Cat 98ex242
1998, 201–204. https://doi.org/10.1109/apccas.1998.743713.
(37) Walter, M. G.; Warren, E. L.; McKone, J. R.; Boettcher, S. W.; Mi, Q.; Santori, E. A.;
Lewis, N. S. Solar Water Splitting Cells. Chem Rev 2010, 110 (11), 6446–6473.
https://doi.org/10.1021/cr1002326.
(38) Dry, M. E. The Fischer–Tropsch Process: 1950–2000. Catal Today 2002, 71 (3–4), 227–
241. https://doi.org/10.1016/s0920-5861(01)00453-9.
(39) Tucker, C. L.; Bordoloi, A.; Steen, E. van. Novel Single Pass Biogas-to-Diesel Process
Using a Fischer–Tropsch Catalyst Designed for High Conversion. Sustain Energy Fuels 2021, 5
(22), 5717–5732. https://doi.org/10.1039/d1se01299a.
(40) Uhm, S.; Kim, Y. D. Electrochemical Conversion of Carbon Dioxide in a Solid Oxide
Electrolysis Cell. Curr Appl Phys 2014, 14 (5), 672–679.
https://doi.org/10.1016/j.cap.2014.02.013.
(41) Ganji, P.; Borse, R. A.; Xie, J.; Mohamed, A. G. A.; Wang, Y. Toward Commercial Carbon
Dioxide Electrolysis. Adv Sustain Syst 2020, 4 (8), 2000096.
https://doi.org/10.1002/adsu.202000096.
(42) Wu, Y.; Kim, D.; Teets, T. S. Photophysical Properties and Redox Potentials of
Photosensitizers for Organic Photoredox Transformations. Synlett 2021, 33 (12), 1154–1179.
https://doi.org/10.1055/a-1390-9065.
(43) Kawawaki, T.; Negishi, Y.; Kawasaki, H. Photo/Electrocatalysis and Photosensitization
Using Metal Nanoclusters for Green Energy and Medical Applications. Nanoscale Adv 2019, 2
(1), 17–36. https://doi.org/10.1039/c9na00583h.
(44) Hu, F.; Xu, S.; Liu, B. Photosensitizers with Aggregation‐Induced Emission: Materials and
Biomedical Applications. Adv. Mater. 2018, 30 (45), 1801350.
https://doi.org/10.1002/adma.201801350.
(45) Gerdes, R.; Wöhrle, D.; Spiller, W.; Schneider, G.; Schnurpfeil, G.; Schulz-Ekloff, G.
Photo-Oxidation of Phenol and Monochlorophenols in Oxygen-Saturated Aqueous Solutions by
Different Photosensitizers. J Photochem Photobiology Chem 1997, 111 (1–3), 65–74.
https://doi.org/10.1016/s1010-6030(97)00209-8.
(46) Shon, J.-H.; Teets, T. S. Molecular Photosensitizers in Energy Research and Catalysis:
Design Principles and Recent Developments. Acs Energy Lett 2019, 4 (2), 558–566.
https://doi.org/10.1021/acsenergylett.8b02388.
(47) Muniz, C.; Archer, C.; Applebaum, J.; Schaab, J.; Alagaratnam, A.; Djurovich, P.;
Thompson, M. Two-Coordinate Coinage Metal Complexes as Solar Photosensitizers. 2023.
https://doi.org/10.26434/chemrxiv-2023-fkkfz-v2.
26

(48) Olah, G. A.; Goeppert, A.; Prakash, G. K. S. Beyond Oil and Gas: The Methanol Economy,
Third, Updated and Enlarged.; Wiley-VCH Verlag GmbH & Co.: Weinheim, Germany, 2018.
(49) McCusker, J. K. Femtosecond Absorption Spectroscopy of Transition Metal Charge-
Transfer Complexes. Accounts Chem Res 2003, 36 (12), 876–887.
https://doi.org/10.1021/ar030111d.
(50) Shi, S.; Jung, M. C.; Coburn, C.; Tadle, A.; R., D. S. M.; Djurovich, P. I.; Forrest, S. R.;
Thompson, M. E. Highly Efficient Photo- and Electroluminescence from Two-Coordinate Cu(I)
Complexes Featuring Nonconventional N-Heterocyclic Carbenes. J Am Chem Soc 2019, 141 (8),
3576–3588. https://doi.org/10.1021/jacs.8b12397.
(51) Hamze, R.; Shi, S.; Kapper, S. C.; Ravinson, D. S. M.; Estergreen, L.; Jung, M.-C.; Tadle,
A. C.; Haiges, R.; Djurovich, P. I.; Peltier, J. L.; Jazzar, R.; Bertrand, G.; Bradforth, S. E.;
Thompson, M. E. “Quick-Silver” from a Systematic Study of Highly Luminescent, Two-
Coordinate, D10 Coinage Metal Complexes. J Am Chem Soc 2019, 141 (21), 8616–8626.
https://doi.org/10.1021/jacs.9b03657.
(52) Sinanoglu, O. Theory of Chemical Reaction Networks. All Possible Mechanisms or
Synthetic Pathways with given Number of Reaction Steps or Species. J Am Chem Soc 1975, 97
(9), 2309–2320. https://doi.org/10.1021/ja00842a001.
(53) Sellers, P. H. An Introduction to a Mathematical Theory of Chemical Reaction Networks I.
Arch. Rational Mech. Anal. 1971, 44 (1), 23–40. https://doi.org/10.1007/bf00250826.
(54) Nijegorodov, N.; Mabbs, R.; Winkoun, D. P. Influence of Weak and Strong Donor Groups
on the Fluorescence Parameters and the Intersystem Crossing Rate Constant. Spectrochimica
Acta Part Mol Biomol Spectrosc 2003, 59 (3), 595–606. https://doi.org/10.1016/s1386-
1425(02)00207-x.
(55) Cox, J. M.; Bain, M.; Kellogg, M.; Bradforth, S. E.; Lopez, S. A. Role of the Perfluoro
Effect in the Selective Photochemical Isomerization of Hexafluorobenzene. J Am Chem Soc
2021, 143 (18), 7002–7012. https://doi.org/10.1021/jacs.1c01506.
(56) Wang, L.-P.; Titov, A.; McGibbon, R.; Liu, F.; Pande, V. S.; Martínez, T. J. Discovering
Chemistry with an Ab Initio Nanoreactor. Nat Chem 2014, 6 (12), 1044–1048.
https://doi.org/10.1038/nchem.2099.
(57) Eaton, P. E.; Cole, T. W. Cubane. J Am Chem Soc 1964, 86 (15), 3157–3158.
https://doi.org/10.1021/ja01069a041.
(58) Sarkar, D.; Bera, N.; Ghosh, S. [2+2] Photochemical Cycloaddition in Organic Synthesis.
Eur. J. Org. Chem. 2020, 2020 (10), 1310–1326. https://doi.org/10.1002/ejoc.201901143.
(59) Mercer, J. A. M.; Cohen, C. M.; Shuken, S. R.; Wagner, A. M.; Smith, M. W.; Moss, F. R.;
Smith, M. D.; Vahala, R.; Gonzalez-Martinez, A.; Boxer, S. G.; Burns, N. Z. Chemical Synthesis
27

and Self-Assembly of a Ladderane Phospholipid. J Am Chem Soc 2016, 138 (49), 15845–15848.
https://doi.org/10.1021/jacs.6b10706.
(60) Chen, Z.; Mercer, J. A. M.; Zhu, X.; Romaniuk, J. A. H.; Pfattner, R.; Cegelski, L.;
Martinez, T. J.; Burns, N. Z.; Xia, Y. Mechanochemical Unzipping of Insulating Polyladderene
to Semiconducting Polyacetylene. Science 2017, 357 (6350), 475–479.
https://doi.org/10.1126/science.aan2797.
(61) Salomon, R. G.; Kochi, J. K. Copper(I) Catalysis in Photocycloadditions. I. Norbornene. J
Am Chem Soc 1974, 96 (4), 1137–1144. https://doi.org/10.1021/ja00811a030.
(62) Salomon, R. G.; Folting, K.; Streib, W. E.; Kochi, J. K. Copper(I) Catalysis in
Photocycloadditions. II. Cyclopentene, Cyclohexene, and Cycloheptene. J Am Chem Soc 1974,
96 (4), 1145–1152. https://doi.org/10.1021/ja00811a031.

28

Chapter 2. Description, Design and Additions to a Nanosecond Transient Absorption
Spectrometer
2.1. Introduction
The life of an excited state organic molecule, specifically one synthetically designed to
engage in multiple reactions with other nearby molecules of interest, often involves multiple
phases, living from a few nanoseconds to even minutes for certain triplet or charge separated
species. Capturing snapshots of these photochemical dynamics as well as understanding them is
central to physical-organic chemistry and spectroscopy. While the Bradforth lab has an excellent
suite of tools to observe various aspects of the ultrafast dynamics, ranging from femto- to nano-
seconds, the detailed evolution of the molecular clock for longer-lived species with >1 ns lifetimes
have remained largely invisible to us (Figure 2.1). These events include molecular translations,
diffusion, slower chemical reactions, etc., all events we are interested in for projects described in
this thesis. The primary timescale we are interested in is the diffusion of molecular quenchers

Figure 2.1 – Cartoon depiction of experimental timescale capabilities in this work.
29

which serve to either reduce or oxidize a photosensitizer for use in catalysis (chapter 4) or undergo
a bimolecular photochemical reaction (chapter 5). Diffusion happens on the order of a few to tens
of nanoseconds with ~1-50 mM quencher concentration.
This thesis describes several steady-state and time-resolved (transient) spectroscopy
experiments. The documentation of absorption, emission, time-correlated single photon counting
(TCSPC), as well as femto-/pico- to nano-second transient absorption (psTA) experiments are
described in each chapter where each is used. These methods and how we implement them have
been described in detail by our group before. However, the implementation of nano- to micro-
second TA (nsTA) used in this work is new to our group and thus warrants a chapter due to the
technical detail in how this was set up and how the apparatus works. While the nsTA implemented
is based on a commercial instrument, significant additions providing novel capabilities are
described. Therefore, this chapter is dedicated to the description of the recent-to-market
Magnitude Instruments nsTA instrument with comparison to the psTA setup in the Bradforth lab.
2.2. Magnitude Instruments for nsTA
Previously, we captured long time TA events (> 1 ns) with a ~700 ps IRF laser (Alphalas-
Pulselas-A Nd:YAG) coupled to the femtosecond regenerative amplifier by scanning an electronic
delay.
1,2
However, poor reliability, shot to shot stability, mode quality, and pulse energies lead to
our group researching different avenues. Here we describe Magnitude Instruments’ enVISTA and
enVISion spectrometers. The Magnitude Instruments (Innovation Park, PA) enVISTA and its
cousins the enVISion and the inspIRe are a class of novel benchtop, turn-key, nanosecond transient
absorption spectrometers. The enVISion allows us to probe chemical dynamics and temporal
regimes at high acquisition efficiencies and exceptional signal to noise levels without the high
peak powers afforded by ultrafast TA. This instrument utilizes state of the art noise cancellation
30

to achieve a >10
4
x sensitivity compared to other nsTA instruments that have been on the market
for 20+ years. Data from the Magnitude nsTA instruments has appeared in the last three years
published in JACS,
3
J. Phys. Chem. Lett.,
4
and J. Chem. Phys.
5,6

While the exact design by which the Magnitude instrument achieves sub-µOD signal to
noise is kept as a company secret, the patent filed by Rimshaw et al,
7
is revealing. What is detailed
is that there are 4 noise cancellation techniques that may be used to achieve high SNR. At least
three of the following are used at any time. In the preferred setup, there exists 1) a pump laser
operating with at least a 100 Hz repetition rate; 2) a probe with average irradiances at least or
greater than 1 µW m
-2
nm
-1
; 3) a DC-coupled detector for measurement of the probe’s transmission
on a pump off shot which can then sequentially switch an AC-coupling for measurement of the
change in transmission of the signal on a pump on shot; 4) a digital oscilloscope with fast trigger
rearm time and large enough band width (>1 MHz) to capture every trigger event for sequential
pump on, pump off shots. Again, not all 4 techniques are required, and they may be present in any
iteration to achieve high noise cancellation.
For chronological bookkeeping, we will briefly describe here the history of use of
Magnitude Instruments at USC. An enVISTA with an external 532 nm laser housing was
purchased first and installed in June 2022. The first iteration of UV capabilities was performed
during this period (section 2.2.2.1). The enVISTA was utilized until December 2022 where it was
shipped back to Magnitude Instruments for upgrade to an enVISion, which has a light source and
detectors that extend its probing range into the NIR. The internal 355 nm laser and external 532
laser remain identical between the instruments. To improve probe transmission into the UV, the
fused silica optics of the enVISTA are replaced with calcium fluoride in the enVISion. The
enVISion instrument arrived late February 2023 with the external laser, complete with the SHG to
31

266 nm (section 2.2.3.1), arriving later with full installation and availability for use on March 15
th
,
2023. Gratings 1 and 3 were found to be miscalibrated in March 2023 and the monochromator
was replaced in April 2023. The lamp unit, which suffered from periodic overheating, was
replaced in June 2023. According to Magnitude Instruments, the instrument and lamp were run at
28
o
C for five hours and the bulb did not turn off or overheat during this period. Additionally,
while at Magnitude Instruments, the following modifications and improvements were made. The
Xe lamp power supply and mounting were replaced. A second 140 mm fan was added to the
enVISion to increase airflow to the lamp compartment. A surge suppressor to the AC mains was
added. The IO and laser control boards were replaced with the latest version. The filter wheel #2
(after the sample chamber) was changed from a 1400 nm LP filter to a 1300 nm longpass filter,
increasing transmission level of the probe light. The instrument was shipped back to USC by July
6
th
, 2023, during the submission of this thesis.
2.2.1. Method Comparison of psTA and nsTA
To explain the nsTA method of data capture, it is useful to compare the nsTA with the
workhorse psTA in the ultrafast laboratory. In transmission geometry setup of a TA experiment,
the probe is sent through focusing optics, impingent upon the sample and then sent into a
monochromator or spectrometer. The probe light is separated into its constituent colors and the
relative intensity of the probe with and without the pump is determined. The probe colors are
detected with a dispersed diode-array spectrometer (in the psTA case) or with a scanning
monochromator (in the Magnitude Instruments case).
2.2.1.1. Experimental method of psTA
The psTA best observes dynamics from 50 fs – 1.5 ns, i.e, recording information about a
timescale up to 10
6
faster than a nsTA spectrometer. However, the detector itself does not provide
32

this time resolution; therefore, the ultrafast community utilizes time-integrating detectors and
optical delay to achieve the timing information. In psTA setups implemented in labs across the
world, the pump-probe time delay is achieved by varying the optical path between the two light
beams. Specifically in our lab, a gold corner cube retroreflector is translated via a mechanically
controlled delay stage situated within the probe path; this changes the relative pathlengths between
the pump and probe to 100 nm precision. Positive time delay points are when the probe pathlength
is longer, negative times for when the pump path is longer, and we define the quantity t0 as the
position of the translation stage when there is zero pathlength difference. Mechanically delayed
stages are effective for time delays much smaller than 10 ns: a time delay of only 10 ns would
require a 3 m long stage! Maintaining collimation and pointing over greater distances is
unreasonable for most labs, hence the 1-1.5 ns time limit. Some labs have achieved 2-5 ns delays
by having 2+ passes on the retroreflector.
To record a psTA dataset, spectral snapshots at a given time delay are recorded for 250-
500 shots. The probe intensity on the array detector is collected every shot at 1 kHz and processed
in pairs of pump-on ( 𝑇𝑇 ( 𝑘𝑘 , 𝜆𝜆 )
𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝 𝑜𝑜 𝑛𝑛 ), pump-off shots ( 𝑇𝑇 ( 𝑘𝑘 , 𝜆𝜆 )
𝑝𝑝 𝑝𝑝𝑝𝑝 𝑝𝑝 𝑜𝑜 𝑜𝑜 𝑜𝑜 ); the ratio is taken and the
Δ 𝐴𝐴 𝑏𝑏𝑠𝑠 for psTA is calculated for each shot pair, demonstrated in Equation 2.1.
Δ 𝐴𝐴 𝑏𝑏𝑠𝑠 ( 𝑘𝑘 , 𝜆𝜆 ) = − log
1 0
�
𝑇𝑇 ( 𝑡𝑡 , 𝜆𝜆 )
𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝 𝑜𝑜𝑜𝑜
𝑇𝑇 ( 𝑡𝑡 , 𝜆𝜆 )
𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝 𝑜𝑜 𝑜𝑜𝑜𝑜
� Equation 2.1
Importantly, in psTA all probe wavelengths are captured simultaneously every laser shot but the
time evolution information is captured more slowly via scanning the delay stage (over seconds in
real time). The stage position stays fixed at a given time delay for a given number of shots. Then
the stage is stepped sequentially as defined in delay file and each time point is recorded as defined
above, completing a scan when the final, most positive delay point is recorded.
33

2.2.1.2. Experimental Methods for nsTA
In comparison, the nsTA captures the time-wavelength two-dimensional dataset in an
operationally orthogonal process compared to the psTA. The monochromator is set to a single
specified probe wavelength where the full diode response in time is collected on a digital
oscilloscope. The single wavelength TA experiment is carried out for a given number of shots and
then the monochromator moves to the next probe wavelength to execute a 2DTA scan. To
emphasize, for the nsTA, the full temporal information is captured each and every shot and the
spectral information is captured more slowly via scanning the monochromator. Iterations refer to
how many times a single wavelength time-trace is repeatedly recorded and accumulated before
moving to the next wavelength.
In the nsTA apparatus, while the pump comes from a 0.12 - 6 kHz (internal 355 nm laser)
variable repetition rate laser, the probe comes from a continuous lamp light source. This is different
from the psTA configuration, as the broadband probe is not pulsed and synchronized at the same
repetition rate as the pump laser. Therefore, the change of the few nanoseconds duration pump
pulse has on the continuous probe intensity can be recorded. The fast photodiode detector is
coupled to a fast digital oscilloscope (200 MHz). The digital oscilloscope records the voltage in
mV. Note, there is a tradeoff between the sampling bandwidth (determining the time resolution)
and the voltage resolution; an increase in bit resolution (8 bit to 12 bit to 14 bit) in digitizing the
voltage on the oscilloscope will lead to a decrease in temporal resolution (2 ns to 4 ns to 8 ns).
As stated previously, ultrafast pump-probe experiments use time-integrating detectors with
translation of an optical delay stage to recover the lost time information. To achieve sufficient
excited state population which determines signal level, an ultrafast experiment incorporates small
spot sizes (~300 µm) to achieve high population transferred to the excited state for 0.1 to 2 µJ
34

pulse lasers. A nanosecond pump probe experiment uses a larger pulse energy (~100 µJ) which
allows for use of a larger spotsize (~8 mm) to achieve similar excited state concentrations, and
therefore similar Δ 𝐴𝐴 . The larger pump spotsize in the nanosecond experiment accommodates the
larger probe spotsize (~2x4 mm), which is now larger as the probe is a lamp and cannot be focused
as tightly as a laser spot.
Before we discuss the way Magnitude records transient data, it is useful to discuss how the
Δ 𝐴𝐴 value might be recorded to achieve optimal noise reduction. In most TA setups, to capture the
difference measurement between the pump and probe for Δ 𝐴𝐴 , an optical chopper is placed after
pump generation and before the sample. The chopper serves to provide a series of pump on and
off shots by blocking the pump when one of the blades intercepts the laser beam and letting the
pump through as the blade no longer intercepts on a later shot. To minimize noise fluctuations
from
1
𝑓𝑓 �
noise, the chopping frequency is set at half the trigger/probe repetition rate to provide
sequential pairs of pump on-pump off shots. In the vast number of cases, the probe is not optically
chopped and is present at the sample on every trigger shot. However, as the probe is ever present,
the dark count noise is not accounted for and noise drifts in the environment can play a large role.
A possible solution uses two phase-locked choppers, with a chopping frequency now
1
4
�
the trigger repetition rate and at a
𝜋𝜋 2
�
phase difference (Figure S2.3).
8
This scheme provides the
ability to record the TA (probe on/pump off and probe on/pump on); the pump only signal which
may isolate pump scatter and photoluminescence (PL) artifacts (probe off/pump on); and finally,
the dark offset (probe off/pump off). This scheme isolates the noise coming from each component
and specifically, the probe fluctuations. As we will see in the next section, isolating the pump
on/probe off may possibly increase the effectiveness of PL subtraction.
35

As part of the noise reduction in their instruments, Magnitude incorporates a detector that
can switch between coupling modes, AC to DC, with a fast rearm time. On a pump-on shot, the
detector is in AC-coupled mode and the change in transmission of the sample on a 2-8 ns per
trigger shot is recorded � Δ 𝑇𝑇 ( 𝑘𝑘 , 𝜆𝜆 ) � . AC-coupling mode allows the detector to maximize the
voltage resolution. Recording Δ 𝑇𝑇 allows for subtraction of any non-white noise from the detected
voltage. Then, on a pump-off shot, the detector is switched to DC-coupled mode and records the
voltage for the full intensity of the transmitted light � 𝑇𝑇 ( 𝜆𝜆 ) � . Currently, we believe 𝑇𝑇 ( 𝜆𝜆 ) is
collected once or at a much lower collection frequency than Δ 𝑇𝑇 due to the generation of a time-
averaged 𝑇𝑇 ( 𝜆𝜆 ) in the data file (section 2.4.1) which when used alongside an abstracted Δ 𝑇𝑇 ( 𝜆𝜆 ), can
fully recreate a nsTA dataset, down to floating point error (Figure S2.4). It is highly unlikely the
𝑇𝑇 ( 𝜆𝜆 ) is collected at the same frequency as Δ 𝑇𝑇 ( 𝜆𝜆 ) as this spectral recreation should not be possible
if 𝑇𝑇 ( 𝜆𝜆 ) is collected alongside Δ 𝑇𝑇 ( 𝜆𝜆 ). The lamp does not have this low of a noise floor. The Δ 𝐴𝐴 𝑏𝑏𝑠𝑠
for nsTA is calculated with Equation 2.2:
Δ 𝐴𝐴 𝑏𝑏𝑠𝑠 ( 𝑘𝑘 , 𝜆𝜆 ) = − log
1 0
�
𝑇𝑇 ( 𝜆𝜆 ) + Δ 𝑇𝑇 ( 𝑡𝑡 , 𝜆𝜆 )
𝑇𝑇 ( 𝜆𝜆 )
�= − log
1 0
� 1 +
Δ 𝑇𝑇 ( 𝑡𝑡 , 𝜆𝜆 )
𝑇𝑇 ( 𝜆𝜆 )
� Equation 2.2
Note the similarities between Equation 2.1 and 2.2, where Equation 2.2 is a ratio of
transmissions like Equation 2.1 but uses the change in transmission rather than a ratio of two
absolute transmitted light intensities. This difference in data digitization provides a key to the
increased sensitivity of the instrument and it is also important preamble to understand the method
we use for removing an artifact present in nsTA dataset from both Magnitude instruments for
samples with strong photoluminescence, which is discussed in the next section.

36

2.2.2. Photoluminescence Correction in Magnitude Instruments
The Magnitude nsTA uses a different optical geometry from the psTA experimental setup
with the latter almost collinear and the former using an orthogonal beam geometry. Excitation of
emissive compounds leads to luminescent emission in all directions including along the probe path
(even in the absence of probe light). Therefore, undesired detection of photoluminescence (PL)
occurs along with the probe light. When detected, pump-induced PL is therefore additional light
intensity, with its own time dependence. Because it is adding to the detected light, it appears as a
negative going artifact in the absorbance defined by Equation 2.2. This PL artefact can both
qualitatively distort the TA spectrum and quantitatively modify the kinetics. In the most extreme
cases, PL can completely overpower the TA spectrum leading to only a large, negative
“absorbance” signal. So, to extract meaningful nsTA spectra, especially for optically emissive
materials with long lifetimes, the PL must be removed.
Before we proceed, a distinction is warranted: stimulated emission is not PL. Stimulated
emission is induced by the probe beam, stimulating population back to the ground state with
different probability depending on the delay from the pump pulse. It is therefore a real part of the
TA signal, as it is in a psTA experiment, whereas detection of pump-induced PL is an experimental
artifact which would not be observed if the probe light was perfectly collimated, and the detector
separated from the sample chamber by a very long distance. PL is not as troubling issue with a
psTA apparatus; as it is induced by the pump alone, it contributes equally for every optical delay,
including negative delays, and so cannot influence the recorded time dynamics. In the rare cases
that the PL gets to comparable intensity as the (pulsed) white light continuum, the PL can be
removed by recording a pump-only “baseline” spectrum.
37

Magnitude Instruments has provided a means for auto-PL subtraction by a simple button
press in software. The PL is subtracted by performing a paired pump-laser only measurement
taken after the main scan along with the TA. In a PL-subtracted experiment, the standard TA with
the pump and probe unblocked is recorded for a given number of shots at a given wavelength.
Then, a PL measurement is performed by blocking the probe with a shutter, not a chopper, and the
intensity of the diode is recorded for the same number of shots. This requires the probe to not drift
over the course of a few seconds during the PL only collection period as its sample interaction
time between the TA and PL only is lengthened. To produce the PL-clean nsTA, the intensity of
the PL is subtracted in the following equation, adapted from Equation 2:
Δ 𝐴𝐴 𝑏𝑏𝑠𝑠 ( 𝑘𝑘 , 𝜆𝜆 ) = − log
1 0
� 1 +
Δ 𝑇𝑇 ( 𝑡𝑡 , 𝜆𝜆 ) − 𝐼𝐼 𝑃𝑃 𝑃𝑃 ( 𝑡𝑡 , 𝜆𝜆 )
𝑇𝑇 ( 𝜆𝜆 )
� Equation 2.3
where 𝑘𝑘 𝑃𝑃 𝑃𝑃 ( 𝑘𝑘 , 𝜆𝜆 ) is the time-dependent PL signal recorded on the same photodiode at every
wavelength.
For most organic chromophore systems, the PL comes from the singlet state and thus will
be strongest in the first few nanoseconds, producing a sharp, single spike to several digital delay
bins around t0. The large flux of light on the photodiode at early times can also produce a ringing
in the time signal. This can then affect time points out to ~20 ns depending on the size of the
ringing signal. This t0 artifact can be accounted for with deconvolution the instrument response.
However, for compounds with strong emission throughout their lifetimes, PL remains an issue for
all time points where nsTA is also present. For those purposes here, our example is 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , a
compound, because of the thermally activated delayed fluorescence, with a long luminescent (~440
ns) lifetime with significant PL emission. In the nanosecond domain, the system has evolved to a
predominantly triplet population (~99%) and such should show negligible stimulated emission.
38

See Chapter 4 for a more complete description of these compounds which is not required for this
section. This methodology can be applied to any system with similar attributes.
If we do not have any form of PL subtraction, the nsTA is overwhelmed by the PL artifact
(Figure 2.2a). However, we have found that when the manufacturer’s auto-PL subtract option is
used it under subtracts the amount of PL (Figure 2.2b), the dip formed at 570 nm. We know it is
under-subtracted because we have recorded psTA data for each of these compounds which contain
400 500 600 700 800 900
-70
-60
-50
-40
-30
-20
-10
0
∆Abs (mOD)
Wavelength (nm)
10 ns
100 ns
250 ns
500 ns
1000 ns
2000 ns
(a) No PL Subtraction
400 500 600 700 800 900
-2
-1
0
1
2
3
4
5
∆Abs (mOD)
Wavelength (nm)
10 ns 500 ns
100 ns 1000 ns
250 ns 2000 ns
(b)
Auto-Subtraction

0 1000 2000 3000
0
0.2
0.4
0.6
0.8
1
Signal (arb. u.)
Time (ns)
A
1
= 1.0, τ
1
= 500 ns
A
2
= 0.6, τ
2
= 500 ns
Difference
(d)

Figure 2.2 – nsTA of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene under varying values of PL subtraction parameter. (a)
No PL subtraction, (b) auto-subtraction and (c) reported time-domain nsTA after auto-PL
subtraction at varying wavelengths and time-dependent PL. The 440 nm is a trace centered on
the GSB has been inverted. The PL (also inverted) is plotted as the circle-blue line. (d)
Simulation of two exponentials with identical time constants and the resulting difference.

39

no dip in the region of their respective emission bands. See succeeding section, 2.2.2.1, for more
details.
Further, we observed a time trace in the region of maximal PL emission that showed a fast
decay, a rise then followed by another slower decay (Figure 2.2c, red trace). We know 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵
obeys single state first order decay in the ns-µs regime. It follows that time dynamics should be
identical for every probed wavelength in transient absorption and the PL lifetime, i.e., the nsTA
signal and the PL trace should be described by an exponential with the same time constant but
differing prefactors. Subtraction of a set of exponentials one from another should produce another
exponential with the exact same lifetime but the prefactor is a difference of two starting prefactors
(Figure 2.2d, section 2.4.3). On this basis, we consider, the red curve with its demonstrably
different time behavior to be distorted, presumably by imperfect correction for detected PL.
Therefore, we have developed an alternate procedure to remove independently the PL
artifact. This requires comparison to a psTA spectrum recorded at 1.5 ns (the longest time delay
measurable in the psTA setup) for the same compound and solvent that is the best representation
of the PL-free TA spectrum. We argue that it is best representation because of the design of
femtosecond pump probe experiment as stated above: any pump-only PL signal will appear
equally for all optical time delays and cannot imprint a time-dependent artifact. This (relatively
small) PL signal can also be assessed and removed from the psTA dataset. The following section
provides greater detail, but we detail this simply to provide a counterpoint to the PL-subtract
problem in the nsTA apparatus.

40

2.2.2.1. Counterpoint: The impact of PL in psTA and approaches used to correct for
PL in psTA.
The psTA and nsTA setups are markedly different. In this context of the magnitude of PL
contamination, the most noticeable difference is the distance between the sample and the collection
lens for the probe in the two setups. The Bradforth lab collection lens is a ∅1” lens, approximately
25 cm away from the sample. The Magnitude instrument also contains a ∅1” lens but the distance
is only 10 cm. For samples under identical conditions, the PL detection efficiency is ~6x stronger
in the Magnitude instrument. Here, we assume for a sample with isotropic emission from a point
source thereby finding the solid angle of light, dropping by d
2
over distance. Additionally, the
Bradforth lab has two software methods of removing PL as well. PL manifests at all times in the
psTA due to the time integrating detectors, including pre-t0. So, utilizing its presence in negative
times is both possible and advantageous. Within the psTA data acquisition software, there is a
function that allows the user to perform background light subtraction. To enable, the probe is
blocked, and the intensity of pump light and pump-induced PL is recorded and subtracted off all
intensity values recorded from the diode array. When the probe light is unblocked, the PL and
pump scatter is removed. A second method of determining PL contamination is done by sending
the translation stage to negative times to decide if there appears to be any negative going signal in
the region of the PL. A post processing workup method using Matlab that can remove PL is also
implemented. During this routine, the average of the pre-time zero signal is calculated for every
wavelength. The artifact is then subtracted from the signal, ensuring any pre-time zero signal is
removed like PL. Qualitatively, even for high PL compounds ( Φ > 0.9), the PL artifact is
demonstrably smaller than the pump scatter which typically presents a much larger issue.
Therefore, the magnitude of PL signal can be inspected at delays before time zero and then, if
41

present at all, removed in equal measure from the entire TA dataset. For these reasons, we consider
the 1.5 ns delay spectrum from psTA to be free of PL contamination.
2.2.2.2. Correcting for PL in the nsTA Dataset
We will approach several avenues for an attempt at understanding the origin of the PL
artifact. These methods were discussed with and suggested by Magnitude Instruments, forming an
effective partnership. Understanding the source of the PL-subtraction artifact should assist the
USC groups, Magnitude Instruments, and other Magnitude Instrument users encountering this
problem, especially researchers and groups with long lived, emissive samples.

Our first method is a pump power study. The hypothesis is that the detector is overwhelmed
at high PL leading to a failure of PL subtraction and that the PL artifact is better removed at low
powers. While the detector may behave linearly with only impingent probe light, the addition of
the large PL signal may saturate the detector, leading to nonlinear behavior. While attenuation of
the pump power will lead to a reduction of TA signal, reducing the laser power will reduce the PL
as well and at low powers, the PL subtract should operate appropriately according to advice from
Magnitude. We should expect the spectral dip artifact in the time-dependent PL-subtracted TA to
be less evident. The result of the power dependence, located in Figure 2.3, indicates this method

Figure 2.3 - Power dependent nsTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene, excited at 355 nm. The y-
axis is kept constant to demonstrate the decrease in signal as the power is decreased. The
absorption and emission spectra are scaled arbitrarily for clarity.

42

does not remove the PL artifact. The dip,
even after attenuation to 10% of the original
power, is still as prominent as the 100%
pump power. Normalization of Figure 2.3
to Figure 2.4 for each power is further
indicative of this result. Positively, this does
indicate that the nsTA and PL is linearly
dependent on power. This experiment was
performed with the enVISTA utilizing the
in-place OD 0.3 and OD 1 filters.
A second method takes advantage of the difference between the probe and PL light. The
probe is a well-formed, nearly collimated beam of lamp light. The PL can be approximated as a
point source, emanating from the spot where the pump hits the sample. The PL, unlike the probe
light, which is collected efficiently by the second probe lens, is emitted isotropically in all
directions, with only a small fraction reaching the detector, determined by the distance to the
collimating lens and solid angle cut out from the lens. Positioning the pump further from the
collection lens and detector should reduce the overall detected intensity of the PL with minimal
reduction in TA signal due to the sample out of focus of the probe (Figure 2.5). Two mirrors were
introduced to direct the pump into the sample which was placed ~7 cm away from the sample’s
original position. The data obtained on 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in this geometry did not appreciably remove the PL
artifact, instead only worsening the SNR.
A third attempt to unravel the PL artifact addressed the possibility the PL artifact is a result
of the probe. Compounds can have absorptions that overlap with the output of the probe lamp and
450 500 550 600 650 700
-0.5
0
0.5
1
∆Abs (mOD)
Wavelength (nm)
100% Power
47% Power
11% Power

Figure 2.4 – The transient spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in
toluene with 355 nm excitation at various laser
pulse energies with a) black – 75 uJ or 100%
power, b) red – 35 uJ or 47% power and c) orange
- 8 uJ or 11% power. These spectra were
normalized at 500 nm.
43

so it is possible to excite some population of
the molecules of interest via the probe. The
excitation density created from the probe
will be far less than the pump due to the
latter’s high fluence, but we have found the
PL subtraction by Magnitude is ~2%
insufficient. Therefore, we hypothesized
that the weak intensity probe continually
excites molecules which provides a steady
state of fluorescing molecules, contributing
to the PL that is not corrected for by the PL
only experiment. The probe is blocked
during the PL only experiment and therefore its effect is not represented. Here, the hypothesis
being that if the probe intensity is attenuated for a TA scan, the effect of the PL generated by the
probe will be reduced and the afforded time trace will be less distorted.

Figure 2.5 – Cartoon schematic of sample
chamber geometry used in the pump
repositioning experiment used in understanding
the source of the PL undersubtraction. The black
ellipses at either end of the chamber are lens.
44

To isolate the effect of the probe on the PL artifact, several single wavelength nsTA
experiments were run where the probe intensity was attenuated via neutral density filters ranging
two orders of magnitude (0 OD to 2 OD). The setup is depicted in Figure 2.6. The traces were
taken with detection at 580 nm, near the peak of the emission of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene to capture the
region where the PL subtraction fails most strongly. The results are displayed in Figure 2.6.
Rather than improve the PL subtraction process, attenuation of the probe decreases the
effectiveness of the PL subtraction. This is visible as the dip at ~500 ns increases, rather than

Figure 2.6 – Photograph of sample chamber with OD filter in place before probe during the
probe attenuation experiment.
0 2000 4000 6000 8000 10000
-4
-2
0
2
∆Abs (mOD)
Time (ns)
0.0 OD
0.3 OD
1.0 OD
2.1 OD
Varying Probe Intensity for PL Artifact a)
0 2000 4000 6000 8000 10000
-1
-0.5
0
0.5
1
∆Abs (mOD)
Time (ns)
0.0 OD
0.3 OD
1.0 OD
Varying Probe Intensity - PL Artifact, Inset b)

Figure 2.7 – SWTA of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene, excitation at 355 nm, detection at 580 nm with PL
subtraction engaged. a) SWTA for all four ND filters and b) inset of left with 2.1 OD removed
for visibility.
45

decreases in intensity. The SNR ratio worsens as expected as the amount of transmitted light
decreases, leading to a larger deviation for Δ 𝐴𝐴 𝑏𝑏𝑠𝑠 . These findings are logical as the TA experiment
records the transmission of the probe every other shot. If the probe is producing PL active
molecules, the emission from them will be incorporated into the transmission from equation 3.
Therefore, the presence of the probe is not responsible for the PL artifact.
After attempting to find the source of the PL artifact, we will return to how data is currently
PL corrected. The current method assumes that the PL induced by the laser is not enough but can
be corrected for with a linear factor, 𝒔𝒔 , demonstrated in Equation 2.4:
Δ 𝐴𝐴 𝑏𝑏𝑠𝑠 ( 𝑘𝑘 , 𝜆𝜆 ) = − log
1 0
� 1 +
Δ 𝑇𝑇 ( 𝑡𝑡 , 𝜆𝜆 ) − 𝒔𝒔 ∙ 𝐼𝐼 𝑃𝑃 𝑃𝑃 ( 𝑡𝑡 , 𝜆𝜆 )
𝑇𝑇 ( 𝜆𝜆 )
� Equation 2.4
where 𝒔𝒔 is the linear factor and now scales the overall amount of PL recorded in the nsTA so that
we best match first (10 ns) transient spectrum for the nsTA to the 1.5 ns cut obtained from psTA.
The effect of this PL subtraction is demonstrated in Figure 2.8. Here, most efficient PL removal
occurs when 𝒔𝒔 = 1.02 (Figure 2.8a), different enough from the auto-subtracted feature to change
the recovered dynamics (Figure 2.8b, red curve vs blue curve). This procedure of scaling the PL
subtraction in the nsTA to match the 1.5 ns trace of the psTA was carried through for nsTA spectra
for each compound. For better PL subtraction, this value varies from 1.01 to 1.07 for each nsTA
spectra for the cMa compounds. This value may vary more strongly depending on the strength of
the emission for other compounds. While the subtraction is imperfect as evidenced by temporal
deviations of the nsTA traces from the PL at the same wavelength (Figure 2.8b, comparing blue
and dotted black traces) and the spectral shape is recovered. However, the temporal trace is not
recovered as would be expected for a subtraction between two exponentials with only differing
prefactors. It was found that, while currently unexplained, performing a nsTA with PL subtraction
is more effective than running a separate PL scan after the nsTA scan. However, the nsTA matrix
46

without PL subtract is not obtained, only the nsTA with PL subtraction and raw PL intensities.
Therefore, we use Δ 𝐴𝐴 with PL sub, 𝑘𝑘 𝑃𝑃 𝑃𝑃 and 𝑇𝑇 values Equation 4 to solve for Δ 𝑇𝑇 in Equation 2.5:
Δ 𝑇𝑇 ( 𝑘𝑘 , 𝜆𝜆 ) = 𝑇𝑇 ( 𝜆𝜆 ) ∙ � 10
− Δ 𝑀𝑀 𝑃𝑃𝑃𝑃𝑃𝑃 𝑝𝑝 𝑃𝑃 ( 𝑡𝑡 , 𝜆𝜆 )
− 1 �+ 𝑘𝑘 𝑃𝑃 𝑃𝑃 ( 𝑘𝑘 , 𝜆𝜆 ) Equation 2.5
where Δ 𝐴𝐴 𝑃𝑃𝑃𝑃𝑃𝑃 𝑝𝑝 𝑃𝑃 ( 𝑘𝑘 , 𝜆𝜆 ) is the data obtained from automatic PL subtraction. Essentially, the
previously subtracted PL values are readded. Here, Δ 𝑇𝑇 ( 𝑘𝑘 , 𝜆𝜆 ) can be substituted back into Equation
2.4 to perform the PL subtraction manually through variation of 𝒔𝒔 . A metric to test the validity of
using this reconstruction of the dataset from Equations 2.3 and 2.4, a subtraction between a raw
dataset matrix ( 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF) and a dataset matrix reconstructed from the same raw data while
using Equation 2.5 and an value of 𝑠𝑠 = 1 produces a double precision matrix with no values larger
than 10
-17
OD (section 2.4.4). The highest precision MATLAB records between two double
precision numbers is 2∙10
-16
and therefore the raw data and the reconstructed data using 𝑠𝑠 = 1 are
identical to floating point error.
With this information, we are hopeful that the Magnitude team can address the PL artifact.
While post-processing the data to adjust the subtraction is not encumbering, the requirement to use
400 500 600 700 800 900
-2
-1
0
1
2
3
4
5
∆Abs (mOD)
Wavelength (nm)
10 ns 500 ns
100 ns 1000 ns
250 ns 2000 ns
1.5 ns, psTA
s = 1.02
(a)

Figure 2.8 – nsTA of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene with the PL subtraction method described in the text
engaged here. a) 2DTA of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene and b) Normalized nsTA time traces at 570 nm
under varying s values. The PL at 570 nm is plotted as a dashed black line.

0 1000 2000 3000
0
0.2
0.4
0.6
0.8
1
∆Abs (norm.)
Time (ns)
s = 0.00
s = 1.00
s = 1.02
PL
λ = 570 nm
(b)
47

an analogue psTA experiment to recover the data could be difficult for labs which are not equipped
with such instrumentation. An ultrafast experiment to compliment the nsTA lengthens
experimental time and only works for systems operating with the assumption the spectral evolution
between 1 and 10 ns is minor which is often not the case.
The PL subtraction artifact strongly affects long lived (>10 ns) emissive samples with
excitation and probing performed in the visible. While only visibly/NUV absorbing compounds
(cMa) were probed here for the discussion of PL subtraction, this artifact remains not only an issue
for UV absorbing compounds (i.e., indole, hexafluorobenzene, and pyrazine) but is in fact
overwhelming the nsTA signal produced from bluer absorbing compounds. The PL produced is
strong enough to cause photodiode ringing out to ~50 ns, far beyond the instrument response (5
ns, section 2.4.5). This issue is exacerbated for compounds with short (1-10 ns) lifetimes as the
entire nsTA signal is subject to PL washing the already low signal which can have decayed
significantly within even the IRF. Discussion of how UV is generated for the enVISTA and
enVISion setups is detailed in the following section.
2.2.3. Second Harmonic Generation using Magnitude Instruments
In addition to emissive, visibly/NUV pumped samples, the wavelength options for the
pump can be extended as well. Using the 532 nm external laser, the excitation wavelengths can be
extended into the DUV by converting 532 nm to 266 nm. Photochemistry experiments can be
probed by nsTA experiments through use of Magnitude Instruments and is current ongoing work
in our lab. The photochemistry is initiated by absorption in the deep ultraviolet (DUV), here actinic
action is provided by the pump laser. The DUV can be accessed in a variety ways, traditionally
achieved in the nanosecond regime by quadrupling the output of a Nd:YAG (or similar mediums)
laser (1064 nm). Here, the 1064 nm is doubled internally to produce 532 nm output via second
48

harmonic generation (SHG). Ideally, a vendor would provide a module that then doubles the 532
nm to produce 266 nm. However, the lack of vendor supplied modules for the external laser
requires an in-house design to provide access to the DUV. This limitation comes from the original
laser manufacturer, not Magnitude Instruments. This section is devoted to describing the SHG of
266 nm from the output of an external laser at 532 nm for use in the Magnitude instruments. We
first describe a design for the SHG setup of the enVISTA which was then iterated and improved
upon for the enVISion upgrade.
The optical physics describing SHG is not discussed presently, rather the practical aspects
of generating such a setup in two iterations. We discuss the first iteration using the enVISTA and
then second the improvements made by Magnitude Instruments for the current setup with the
enVISion. The instrument itself does not impact the SHG setup but reference to either iteration is
made simpler by referring to which instrument the SHG supplied/supplies.
2.2.3.1. Prior SHG Design with the enVISTA
This section provides a detailed description of the SHG setup using the Magnitude
Instrument of the enVISTA. The enVISTA setup used in this thesis is equipped with two lasers:
one internal, output at λ = 355 nm and one external laser, λ = 532 nm. The external laser is the
same between both instruments, but the external laser setups are different.
For the enVISTA, the experimental design is displayed below in Figure 2.9. There are
three major components to this setup. First there is the housing for the external 532 nm laser which
houses the eponymous laser, a mechanical chopper, a photodiode for laser power reference, and a
telescope. The second component is the open to air SHG setup for 266 nm generation, later
surrounded by black cardstock to prevent stray light leaving the setup. And finally, the third
49

compartment is the enVISTA instrument itself. This contains the internal 355 nm laser, the internal
laser components, sample chamber, etc.
To securely position and lock the laser, instrument and optics, an optical breadboard was
used. Two solid, aluminum, optical, 24 x 36 x 0.5”, 1” spaced grid of ¼” – 20 tapped holes
breadboards from Newport Corp (SA2-23) are used to mount this setup, providing a stable base.
To lock the breadboards with respect to each other, three “straps” were used (Thorlabs BA2L).
These are indicated in the figure below by the dark grey rectangles.

50

Figure 2.9 - a) Previous setup of Magnitude enVISTA, external laser, and SHG. b) Photograph of
entire enVISTA setup and c) a closer view of the SHG setup. There are minor differences between
the schematic in a) and the realized setup in b) and c): the removal of lenses L2 and L3, the dovetail
stage for L1 and X1 replacement with micrometer stages for X1. The white circled optic in c) is
only in place for viewing the beam mode of 266 nm and was removed when collecting data.
1”
enVISTA
10”
External Laser
(532 nm)
M1
M2
M3
M4
FM1
L1 L2
L3
L4
532 nm
266 nm
X1
Aluminum Breadboard (24 x 36”)
532 nm HR
266 nm HR
Dovetail Optical Rail
Bench Top
Breadboard hangs
over ~4”
Lens
Breadboard Straps
CLBO Crystal
Beam Dump
Cardstock Walls
25 mm Optical Rail (for walls)
Excitation Laser
and Optics
Probe Light
Source
Monochromator
And Detector
Sample
Compartment
a)
b) c)
51

First, the external laser will be discussed, a photograph of the setup is displayed in Figure
2.10. To eliminate confusions, the laser head itself will be referred to as the external laser whereas
the box containing this laser will be referred to as the external laser housing. The external laser is
an Nd:YAG-type, 532 nm output, 0.12 – 20 kHz repetition rate, 400 uJ, < 5 ns duration pulse width
laser. The output beam size is approximately 8 mm in diameter. The output from the external laser
(In) is first sent to a beamsplitter (BS) which directs a small portion to a reference photodiode
(TPD). The remaining transmitted light is then sent to a waveplate-polarizer combination (WP-P)
which is controlled digitally via the Magnitude laser control software to adjust the power delivered
to the instrument. This output is then sent to a transmissive Keplerian telescope (L1a, L2a) to
reduce the beam size of the output for it to pass through a single blade of the edge of the chopper

Figure 2.10 – Photograph of previous external laser housing design for pairing with the
enVISTA. Along the beam propagation the optical components are as follows: In – input from
external laser cavity; BS – beamsplitter for delivering a small fraction to the triggering
photodiode, TPD; WP – mechanical waveplate; P – cube polarizer; BD – beam dump; L1a –
lens for focusing through chopper blades; Ch – chopper; PH – 1 mm pinhole; L2a –
recollimating lens; M1a – mirror for directing output, removed in previous setup; Out – output
to SHG setup and enVISTA.
52

wheel (Ch). The telescope has a magnification ratio of 1 to not expand or reduce the beam size.
This is the first component of this setup and after leaving the external laser housing (Out), we look
at the second component of this setup, the external optics and SHG setup.
Then, the beam exits the external laser housing, and we now refer back to Figure 2.9a.
The external laser output is then steered via M1 and FM1 to enter the enVISTA instrument for 532
nm pump. Alternatively, if 266 nm light was desired, FM1 is flipped out of the path of the beam,
bypassing the enVISTA, instead being sent to M2. M1, M2 and FM1 are all 532 nm 45
o
AOI,
Nd:YAG laser line mirrors (Edmund Optics #38-893). Mirror M2 controls the steering of the 532
nm into the beam reduction telescope for 266 nm generation. Efficient SHG generation requires
high peak power of an input laser beam. This is achieved by focusing or reducing the incoming
laser light into SHG crystal. For this setup, the light from the external laser after exiting the external
laser housing has a spot size of 8 mm.
Central to the SHG setup is the use of a non-linear crystal which upconverts the 532 nm
into 266 nm. In the enVISTA design, and as per recommendation from Eksma, a cesium lithium

Figure 2.11 – Photographs of CLBO crystal immediately after unpacking from Eksma and
housing left) front facing, center) side facing and on the right) in the SHG setup. The small
arrow in the right-hand photo indicates the fused silica window with visible AR coating. This
arrow points in the opposite direction to the laser beam propagation. The CLBO crystal was
placed such that the flattened side with the arrow facing away from the camera.
53

borate (CLBO) crystal (Eksma CLBO-401S) was used, pictured in Figure 2.11. For the ultrafast
experiments, BBO crystals are routinely used for their particularly high non-linear coefficient and
use down to short wavelengths in the UV. However, CLBO, for use in doubling nanosecond pulses
of 532 nm from 266 nm, is a possible option. Despite possessing a smaller non-linear coefficient,
CLBO has a smaller walk-off angle compared to BBO and as such, and this in turn leads to higher
overall doubling efficiency over a long pathlength inside the crystal. The downside of CLBO is
that the crystals are highly hygroscopic. To avoid this, Eksma CLBO crystals are cased within a
mount with AR/AR @ 532 + 266 nm coating UV-fused silica windows which allow for
transmission of visible and UV light. This crystal is also cut along the optimal phase matching
angles for 266 nm generation from 532 nm of Φ = 45° and Θ = 61.5°. The crystal was mounted
in a rotational mount and a kinematic mount to provide full parameter control for phase matching
(see Figure Xb). In order to avoid laser induced damage (LID), the current for the diodes of the
532 nm laser were set low to 80%, the minimum recommended power for good mode quality. The
external waveplate was rotated to 40-44
o
to lower the power as well with full power at 0
o
and
complete attenuation at 45
o
. The laser repetition rate was set to 500 Hz.
To achieve optimal SHG generation from the CLBO crystal, the recommended conditions
are to have a collimated input beam with a waist of ~200 microns. Collimated light is desired to
avoid focusing the beam too tightly and risk damaging the crystal. To achieve this, a telescope
with a magnification of 0.05 was implemented where the region of collimated light after the
telescope is where the CLBO crystal is placed. The first telescope in the SHG setup is for down
collimating the 532 nm and was a Galilean, all-transmissive optics telescope formed by two lenses,
L1 (N-BK7 plano-convex lens, AR coating from 350-700 nm, ThorLabs LA1908-A) and L2
(achromatic doublet lens, Ø ½" with anti-reflection coating from 400 to 700 nm, ThorLabs
54

ACN127-025-A), with focal lengths of 50 cm and -2.5 cm, respectively. Thus, the absolute value
of L2 to L1 is the desired magnification of 0.05. The distance from L1 to L2 to generate collimated
light is set by the sum of the focal lengths, 47.5 cm or 18.7”. L2 was chosen to be an achromat
instead of a singlet lens, an error as achromat was confused with an asphere which produces a
smaller beam diameter at the focal position and removes spherical aberrations that can be present
in singlet lenses.
Upon use of the achromat for ~5 minutes, laser induced damage (LID) occurred due to the
cement between the lens of the achromat being burned. The cement between the two lenses of the
achromat was found to have a lower LID threshold (~10x) than the optical glasses of the two parts
of the achromat. Before several iterations of the L1-L2 telescope were considered, L2 was removed
and the crystal was placed ~5 cm after the now present focus where the spotsize was 350 µm and
450 µm before and after the SHG crystal, respectively. Several iterations upon the setup where
L2 was added back. It was found that L2, regardless of identity, introduces a substantial
contribution of pointing instability that is not present in the external laser itself or when the beam
passes only through L1. Therefore, L2 was removed during the acquisition of data using 266 nm
and the 532 nm beam was focused to ~7 cm before the crystal and allowed to softly diverge
afterwards. Additionally, we have found that the 532 nm does not possess any pointing instability
in the far field (~5 m).
55

While CLBO was recommended over BBO by Eksma Optics engineers, the CLBO crystal
suffered in multiple ways. A gaussian mode can be recaptured at the cost of lower 266 nm power,
only producing ~0.5% conversion efficiency. Unfortunately, even at modest powers (~100 mW)
of 532 nm and spotsizes of 350 µm and 460 µm at the front and back of the crystal faces,
respectively, the generated 266 nm was poor. After exposures of only ~5 minutes to 532 nm, the
power would drop, and the mode would develop a long streak along the vertical axis: the generated
SHG had a time-varying power. Additionally, LID was seen on the CLBO crystal when the power
was raised above 100 mW, leading to a distorted, refracted mode. Even with tight phase angle
control, phase matching conditions did not produce a circular gaussian mode. Instead, a mode of
parallel lines of varying distances between them was produced as a function of phase matching
(Figure 2.12). Upon sending back the instrument to Magnitude Instruments for the upgrade, it
was confirmed that the crystal was being
damaged over time.
After SHG, the fundamental and the
second harmonic light both exit the crystal and
the beam begins to be expanded after
transmitting through a f = -25 mm, UV-fused
silica, Ø 1/2”, AR coated for 245-440 nm lens
(Newport SPC013AR.10). The UV light is then
steered via mirrors M3 and M4 into the
enVISTA instrument. Mirrors M3 and M4 are
Ø 1” 266 nm 45
o
AOI, Nd:YAG laser line
mirrors (Edmund Optics 34-815). The 532 nm

Figure 2.12 – Photograph of 266 nm beam
with the CLBO crystal. Taken after the circled
optic in Figure 2.10c.
56

light is sequentially transmitted through M3 and M4, serving to dump the visible light effectively
to reduce 532 nm contamination in the 266 nm line. The final optic before the enVISTA instrument
is L4, a f = 500 mm, Ø 1”, UV Fused Silica Plano-convex lens, AR coating: 245-500 nm lens.
Lens L4 serves to upcollimate the light back to 4 mm, the same spot size as the 532 and 355 nm
laser spot sizes. The 266 and 532 nm light enters the enVISTA instrument at a port on the left-
hand side where they are directed 90
o
into the sample compartment. All three pump colors have
the option to be sent through a series of neutral density filters, useful for attenuating the pump
power. The pump is then transmitted through the sample and then dumped via a beam dump on
the internal wall of the sample chamber. For final experiments with 266 nm before shipping the
instrument back to be upgraded to the enVISion, L3 was also removed and L4 was used to
recollimated the beam.
The external optics of the SHG setup was encased in black cardboard to solve several
issues. First, any stray 532 or 266 nm light will be blocked by the cardboard and absorbed, assisting
in laser safety. Second, the cardboard, due to being held in a location where movement is common,
the Mark Thompson photophysics room, the possibility of knocking an optic or mount is high.
The cardboard, while thin, would reduce the chance of an optic being moved unintentionally.
Ideally, plexiglass would be used in place of the cardboard.
We then upgraded the enVISTA for an enVISion. Equipped with this knowledge,
Magnitude incorporated these designs elements into the SHG setup with the enVISion.
57

2.2.4. Current Setup in the enVISion
The enVISion is similar in design to the enVISTA with additions that automate several key
features while providing additional features. The probe chamber has a dual lamp setup: in addition
to the xenon lamp (300 to 1050 nm) from the enVISTA, a halogen lamp is also present (visible-
NIR). The monochromator is equipped with three gratings with blaze angles in the NUV (300 nm),
visible (600 nm), and NIR (1600 nm). The single silicon detector reappears from the enVISTA
along with two more detectors, both InGaAs detectors for detection in the NIR. The recommended
wavelength regions for these three components are displayed in Table 2.1. The transmission of
the probe as a function of instrument, grating, lamp, and detector is displayed in Figure 2.13. The
major improvement that allows further extension into the UV is the replacement of probe optics
from borosilicate (enVISTA) to calcium fluoride where the later boasts high transmission past 200
nm. This can be seen by comparing the green and black curves where the enVISTA transmission
extinguishes ~60-80 nm redder than the enVISion transmission. Preliminary results indicate a UV
cutoff of ~340 nm for good signal to noise for a “standard” 2DTA.
The combination of grating 3 and InGaAs detectors provide extension of the probe into the
NIR. The user can go to ~1700 nm with no loss to time resolution but the probe can be even further
extended to ~2500 nm with detector 3 while still achieving a ~14 ns time response. The halogen
lamp provides higher transmission voltages and a flatter spectrum than the xenon lamp after ~1000
nm in the NIR. No work in this thesis has examined the extension into the NIR afforded by the
enVISion aside from the transmission spectra displayed here.
58

From the wide array of options for wavelength probing, the user should select the
combination that is most effective for their experiment, while avoiding detector or grating switches
during a scan if possible to minimize experiment time. Grating and detector switches can take up
to a few minutes with each scan. Additionally, the slits on the monochromator are now automated
with desired resolution set by the user, in comparison to the enVISTA which had static metallic
slits. The software, “enVISion_v4_1”, was in use during installation up to June 2023 and
contained diagnostics and information that may be useful and pertinent to the user. This was
replaced by “enVISion_v4_4” which functions identically save for a pause after shutting down the
instrument to avoid locking a flip mirror in an unspecified state. Further replacement of the
software to streamline the software interface experience is slated to appear.

Table 2.1 – Spectral Range Details and Components for enVISion
Probe Lamp Type Range
1 Xenon
200 to 1000 nm
(UV – Vis)
2 Halogen
450 to 2600 nm
(Vis – NIR)
Gratings Details Range
1
300 nm blaze
1200 lines/mm
200 – 700 nm
2
600 nm blaze
1200 lines/mm
350 – 1500 nm
3
1600 nm blaze
600 lines/mm
950 – 2600 nm
Detectors Material Range
1 Silicon 300 – 1050 nm
2 InGaAs 800 – 1700 nm
3 InGaAs
950 – 2550 nm
(time resolution <14 ns)

59

In addition to the various possible probe spectra, the pump spectra for both the internal 355
nm and external 532 nm laser were taken, displayed in Figure 2.14. Both spectra were recorded
on a USB 650 Red Tide spectrometer, using SpectraSuite as the interfacing software and 100 ms
integration time. The respective laser powers were attenuated so as to not oversaturate the
spectrometer. A Kimtech dry wipe was placed in front of the entrance to the spectrometer to
remove pointing errors as a fiber optic cable was not used. The background counts of the
spectrometer were removed by positioning the spectrometer in place, under identical room light
conditions and subtracting the offset. Post-processing calibration was afforded by comparing
known mercury emission lines to a mercury pencil lamp spectra collected by the spectrometer. An
offset of –2.3 nm was applied to align the known mercury emission lines to the recorded mercury
lines.
Photographs of the beam modes for both the 355 nm and 532 nm laser outputs in the sample
chamber were taken, displayed in Figure 2.15. These photographs were taken on an iPhone 12
mini with the front facing camera in photo mode. As can be seen, both beams are gaussian in shape
and 8 mm in diameter as measured via a gradated business card. The laser power outputs were
attenuated to reduce saturation on the camera. The segment that is clipped on the end of the card
in Figure 2.15, left, is a reflection caused by the OD 1.0 filter used to attenuate the beam’s power.
The black square in the back is the beam dump in the sample chamber.
The power and pulse energies of the internal 355 nm laser for both the enVISTA and
enVISion are documented in Figure 2.16a. Both sets of power values were recorded at 2 kHz
using a Coherent Field Max II controller with a PM-2 head with 100% laser diode current. The
dependence of the power on the percentage diode current is roughly linear, saturating at higher
current. There is a ~15% increase in average 355 nm pump power from the enVISTA to the
60

enVISion. The internal 355 nm and external 532 nm laser setups now contain laser attenuators
which are motorized waveplate-polarizer pairs. These attenuators offer continuous linear control
over the lasers’ power in the form of percent of laser power transmitted. The effect the attenuators
has on the laser outputs is displayed in Figure 2.16b and displays a good linear response across
an order of magnitude and the useful range of pump powers. For both setups, the attenuators are
placed immediately following the laser output, after the beamsplitter which sends light to trigger
the reference photodiode.
61

Figure 2.13 – a) Transmission spectra of the probe for differing instruments (enVISion or
enVISTA), G – grating number, and lamp (Xenon or Halogen). Not every combination was
explored. b) Transmission spectra of the probe in the NIR for the enVISion. D – Detector.

200 300 400 500 600 700 800 900 1000
0
400
800
1200
1600
2000
Probe Transmission (mV)
Wavelength (nm)
enVISion G1 Xenon
enVISion G2 Xenon
enVISion G2 Halogen
enVISTA G2 Xenon
x5
Transmission Spectra of Probe (UV-vis-NIR)
x20
Silicon Detector
(Detector #1)
a)

800 1000 1200 1400 1600 1800
0
200
400
600
800
1000
1200
1400
Probe Transmission (mV)
Wavelength (nm)
Xenon G2D1
Xenon G3D1
Xenon G3D2
Xenon G3D2
Halogen G3D2
Halogen G2D2
Transmission Spectra of Probe (NIR)
b)
62

350 360 530 540
0
20
40
60
80
100
Intensity (arb. u.)
Wavelength (nm)
enVISion Pump Laser Spectra

Figure 2.14 – Laser pump spectra for blue – internal 355 nm and green – external 532 nm laser,
normalized to 100 at peak height.

Figure 2.15 – Photographs of outputs of 532 nm, left and 355 nm, right, taken May 2
nd
, 2023.
63

For the upgrade from the enVISTA, the enVISion instrument has a larger footprint (44” x
24”) to house the larger array of experimental options with the enVISion and external laser pictured
in Figure 2.17. The external laser housing was also redesigned (15” x 24”) and the 266 nm
harmonic setup was moved to the inside of the laser housing. The enVISion and external laser
0 20 40 60 80 100
0
20
40
60
80
100
120
0
20
40
60
80
100
120
Power (mW)
Power Transmission (%)
Laser Rep Rate = 2 kHz
Diode Current = 100%
λ
pump
= 355 nm
Intercept -3 ± 1
Slope 1.03 ± 0.02
Adj. R-Square 0.994
Pulse Energy (µJ)
a)
0 20 40 60 80 100
0
20
40
60
80
100
120
140
Power (mW)
Power Transmission (%)
Laser Rep Rate = 5 kHz
Diode Current = 80%
λ
pump
= 355 nm
Intercept 6.2 ± 0.2
Slope 1.36 ± 0.01
Adj. R-Square 0.999
0
10
20
30
40
50
60
Pulse Energy (µJ)
b)

80 85 90 95 100
0
50
100
150
200
250
300 enVISion enVISTA
OD 0 OD 0
OD 0.3 OD 0.3
OD 1.0 OD 1.0
Power (mW)
Diode Current (%)
Internal Laser, 355 nm Power and Pulse Energies
0
20
40
60
80
100
120
Pulse Energy (µJ)
Rep Rate 5 kHz
c)
0 20 40 60 80 100
0
100
200
300
400
0
100
200
300
400
Pulse Energy (µJ)
Power (mW)
Transmission (%)
Laser Rep Rate = 2 kHz
Diode Current = 100%
λ
pump
= 532 nm
Intercept 6.9 ± 3
Slope 4.1 ± 0.1
Adj. R-Square 0.9971
d)

Figure 2.16 – Laser powers (left axis, mW) and pulse energies (right axis, 𝜇𝜇 J) as a function of
the laser power attenuator transmission for the 355 nm laser at a) 100% diode current, 2 kHz
repetition rate and b) 80% diode current, 5 kHz repetition rate. A linear fit is displayed as well
for each (red line). c) Recorded internal laser power and pulse energies as a function of laser
diode current and OD filter, collected at 5 kHz laser repetition rate. Pulse energies were
obtained through dividing the power by half the laser rep rate. The powers were recorded after
the chopper, at the sample stage. Solid symbols – current enVISion values; open symbols –
previous enVISTA values. d) Recorded laser powers as a function of the laser power attenuator
for the 532 nm laser at 100% diode current, 2 kHz repetition rate. A linear fit is displayed as
well for each (red line). Errors in power are ~0.4 mW but not displayed as they are smaller
than the size of the symbol.
64

housing are contained on the same breadboards as the enVISTA with the breadboard straps moved
away from the edges to accommodate the feet of the enVISion. A beam tube is placed between the
external laser housing and enVISion instrument for a fully encapsulated instrument pair. A metal
6x12” bracket strapped to both housings them together. Magnitude was alerted to our findings and
incorporated the changes; the current setup is displayed in Figure 2.18. There are several key
comparisons of which we will explore via a thorough description of the current layout. The

Figure 2.17 – Photograph of enVISion (right hand side, sample chamber open) and external
laser (left hand side) as of April 2023 in the Mark Thompson lab space.

65

external laser addon electronics and the SHG setup from the enVISTA are combined within the
external laser housing for the enVISion to form a complete encapsulated external laser setup.
2.2.4.1. Current SHG Design
The placement of the SHG setup within the external laser housing removes the possibility
for misalignment, reduces introduction of dust or fine air particles, and provides an environment
for the SHG crystal. The external laser produces excess heat, which when enclosed, raises the

Figure 2.18 – Photograph of external laser and SHG setup as of April 2023. The naming
convention for this setup is adopted from the previous external laser housing and SHG setup.
Modifications: Atten – laser power attenuator; M1-M4 – 532 nm HRs; M5 and M6 – 266 nm
HRs; FM2 – Magnetic kinematic mount with 266 nm HR to send SHG into enVISion. Mounts
are in place for 266 nm generation. FM1 would be placed on the pedestal in front of L1 and
FM2 would be removed to pump 532 nm for the enVISion.
External laser
BS
M2
TPD
Atten
M1
L1a
Ch
PH L2a
FM1
L1
M3
M4
X1
FM2
L4
M5
M6
enVISion
Out
66

internal temperature of the external housing, driving off excess water from the BBO crystal,
possibly leading to a longer crystal lifetime.
The 532 nm laser from the enVISTA setup reappears in the enVISion setup and is indicated
by External Laser in Figure 2.18. The output of the external laser is sent into a beam splitter (BS)
with the reflected beam directed into a timing silicon photodiode (TPD), synchronized to the
timing electronics of the instrument. The transmitted beam is sent into a motorized waveplate-
polarizer combo, housed and referred to as a laser power attenuator (Atten, Eksma Optics, 990-
0075-532M, Com port 5). The transmission on a percent scale, is controlled via the laser tool
software. The beam is then directed by 532 nm HRs (M1 and M2) into a lens (L1a, f = ~3 cm),
focusing the beam to decrease the spot size enough to pass through the 100-blade chopper (Ch,
ThorLabs). The beam diverges and passes through a pinhole (PH, 1 mm) to clean the mode and
enters a recollimating lens (L2a, f = 3 cm). Before the SHG setup, a 532 nm HR, upon a removable
kinematic magnetic mount can be placed here to direct the 532 nm into the instrument to use as a
pump (FM1). This mount was found to produce far more reproducible beam pointing when
compared to a flip mount. The beam is then sent through a f = 500 mm plano-convex lens (L1) to
begin focusing the beam into the SHG crystal (X1). The diverging lens, L2 was removed due to
pointing instability produced by the high curvature and small focal length. The 532 nm is further
directed by two 532 nm HRs (M3 and M4) into the SHG crystal at the front face of the crystal.
The focus of the 532 nm is set to be ~54 cm after L1, ~4 cm after the crystal. In the previous
iteration with the enVISTA, a CLBO crystal was used as X1, the SHG crystal, described vide ante.
Magnitude Instruments discovered similar burning and power instability issues with the
CLBO crystal when the crystal was shipped back for the enVISTA to enVISion upgrade. It is
currently unclear as to whether the fused silica windows or the CLBO crystal are suffering laser
67

damage. Magnitude achieved a conversion efficiency of >20% at 80% diode current with a
spotsize of 200 microns; however, the 266 nm was short lived and degraded within a few minutes
along with degradation from the upconverted laser mode. Magnitude also attempted conversion
at 400 microns and achieved ~15% conversion efficiency with reasonable mode quality; but this
output too degraded, this time within 15-20 minutes.

As an effective replacement for the CLBO as the SHG crystal, a BBO (Eksma, 4HG,
unmounted, 6x6x6 mm) was used. For the enVISTA and enVISion, the external laser generated
and continues to generate pulse energies of ~400 µJ, independent to laser repetition rate. At a laser
rep rate of 2 kHz, 100% diode current, and 100% transmission from the laser power attenuator,
~110 mW or 110 µJ, (PM-2 head, Coherent Laser Mate II controller) of 266 nm was generated for
a conversion efficiency of 28%. Magnitude reported similar conversion efficiencies in house. A

Figure 2.19 – Photograph of 266 nm laser mode on a ruled card in the sample chamber in the
left) horizontal and right) vertical directions. The laser power was attenuated to ~1% to avoid
camera saturation.
68

kinematic mount to control crystal translation, focusing conditions, and the phase matching angle
(vertical actuation) on a 45 cm long dovetail rail (Thorlabs RLA1800) was installed. The back
edge of the crystal rail carrier (Thorlabs RC2) was at 362 mm.
The residual 532 nm and generated 266 nm are then directed by 266 nm HRs (M5 and M6),
highly transmissive for 532 nm, into a f = 500 mm lens (L4), serving to recollimate the 266 nm.
Similar to the focusing of the 532 nm, the 266 nm forgoes the recollimating L3 lens. A second
removable magnetic mount containing a 266 nm mirror is placed upon the second post to direct
266 nm into the instrument (FM2). It is this final optic that is adjusted for overlap with the probe
in the sample chamber. The flip mount inside the enVISion instrument is flipped down for nsTA
excitation by 266 nm (and 532 nm). The beam then hits a fused silica, right angle prism which
directs the beam into the sample chamber.
There were concerns of burning the BBO crystal via the high pump fluence. To
compensate, the BBO crystal was translated along the dovetail rail by 19 mm to end at 381 mm.
It was found that the crystal was not burned. The upconverted power dropped from 100 µJ to 70
µJ which is still sufficient for performing nsTA experiments with 266 nm. Adjustments made to
either of the phase matching angles does not drastically improve the SHG power. Previous
experiments indicate the 266 nm pump power leads to high sample degradation as well so lower
powers may improve this degradation issue.
Similar to the 355 and 532 nm outputs, the 266 nm output, power, spectrum and spotsize
was characterized. At the sample stage, the 266 nm beam is ~5 x 1.5 mm, width x height. A
photograph is displayed in Figure 2.19. Unlike the 532 and 355 nm outputs, the 266 nm beam is
not gaussian and is longer in the horizontal direction. When compared to the 532 nm laser spot, it
is smaller. Efforts to increase the spotsize such as the removal of L4 from the beam path did not
69

increase the 266 nm spotsize in the sample chamber greatly, only to ~5-6 mm in the long
dimension.

The spectrum of 266 nm was recorded by a USB 2000 spectrometer (Ocean Optics now
Ocean Insight), using SpectraSuite as the interfacing software and 100 ms integration time,
displayed in Figure 2.20. The respective laser power was attenuated to not oversaturate the
spectrometer. A Kimtech dry wipe was placed in front of the entrance to the spectrometer to
remove pointing errors as a fiber optic cable was not used. The background counts of the
spectrometer were removed by positioning the spectrometer in place, under identical room light
conditions and subtracting the offset. Post-processing calibration was afforded by comparing
260 270 280
0
20
40
60
80
100
Intensity (arb. u.)
Wavelength (nm)
enVISion SHG Pump Laser Spectra
*

Figure 2.20 – Laser pump spectrum 266 nm laser output, normalized to 100 at peak height.
The ‘ * ’ above the peak at 270 are an artifact from the spectrometer, confirmed by comparing
the mercury spectrum on both the USB 650 and USB 2000 spectrometers where every peak in
the USB 2000 possessed a ~1-10% shoulder, ~4 nm red of the main feature. The USB 650 did
not and the artifacts in the USB 2000 were not known mercury lines.
70

known mercury emission lines to a mercury pencil lamp spectra collected by the spectrometer. An
offset of +2.7 nm was applied to align the known mercury emission lines to the recorded mercury
lines.

The 266 nm laser power output was recorded as a function of transmission (%) from the
laser power attenuator (Figure 2.21). The 266 nm was recorded in the same way the 532 nm was
recorded: the power meter head (PM-2) was placed after the external laser output after removing
the beam tube between the external laser and enVISion. The same power meter (Coherent
FieldMaxII) was used, with the setting on 266 nm. Unlike the 532 and 355 nm outputs, the power
dependence quadratically depends on the attenuator transmission. This is expected as SHG is an
0 20 40 60 80 100
0
20
40
60
80
0
20
40
60
80
Pulse Energy (µJ)
Power (mW)
Transmission (%)
Laser Rep Rate = 2 kHz
Diode Current = 100%
λ
pump
= 266 nm
Equation y=A*x^2+C
A 0.0072 ± 0.0001
C 1.5 ± 0.4
Adj. R-Square 0.9974

Figure 2.21 – Laser powers (left axis, mW) and pulse energies (right axis, 𝜇𝜇 J) as a function of
the laser power attenuator transmission for the 266 nm laser at a) 100% diode current, 2 kHz
repetition rate. A quadratic fit is displayed as well for each (red line).

71

I
2
process and the laser power attenuator controls the power of the input 532 nm to the SHG, not
the 266 nm power directly.
2.2.4.2. Comparison of Data Obtained from enVISTA and enVISion SHG Setups
In order to assess the working conditions of the instrument, a molecule with set conditions
was chosen with the intention of providing a metric for later generations. Due to the strong DUV
absorption ( 𝜖𝜖 ~10
4
M
-1
cm
-1
) in the UV and a bright, long-lived, triplet transition in the visible
region, benzophenone (BP, SigmaAldrich, ReagentPlus, 25G, B9300-25G-A) was used as a metric
for signal levels in the nsTA with 266 nm pump. Due to the presence of a nπ
*
transition near 355
nm, BP can also be excited at 355 nm for a complimentary study; however, a 355 nm experiment
for BP was not performed here. BP undergoes ultrafast intersystem crossing (10 ps) to the triplet
state. So, for the nsTA experiment, the entire population is in the triplet state, a state quenched
diffusively by oxygen. Therefore, lifetimes will vary depending on purging, an effect we have
seen in lab as well. Benzophenone in ethanol (EtOH) was run in both the enVISTA and enVISion
at the time when the respective 266 nm setups were in place (Figure 2.22a,b). Figure 2.22a
appears to have a broader signal due to the lower resolution (9.6 nm) when compared to Figure
2.22b (6.9 nm) as well as larger wavelength step size (25 nm vs 10 nm). The pump energies pump
pulse energy in the enVISion setup is ~100x larger than in the enVISion and a 100x increase in
signal would be expected but the signal level in Figure 2.22b is even smaller than Figure 2.22a.
The exact time between the introduction of the laser to the sample and the initialization of
the run is not known, i.e., time for alignment, pre-2DTA scans, possible degradation during scans,
etc. It is therefore possible that the lower signal in the enVISion data is a result of overexposure
to the laser, resulting in degradation. It was found that BP undergoes acidic proton abstraction
from alcohols to form a ketyl radical which leads to BP loss and diminishment of TA signal.
9

72

Therefore, CHX was substituted for EtOH as the solvent in Figure 2.22c,d. Here, better signal
levels were achieved with reduced degradation over time. To reiterate, the lifetime is strongly
dependent on purging adeptness, not because of degradation. No care was taken to dry the solvent
so the reduction in signal levels is attributed to water/ethanol impurities. With benzophenone in
cyclohexane, we achieved a signal level of Δ 𝐴𝐴 𝑏𝑏𝑠𝑠 ~700 µOD at 525 nm for ~100 µJ of 266 nm.
This data was taken before the BBO was translated away from the 532 nm focus. The table of
parameters for the experiment with the enVISion, Figure 2.22c,d is tabulated in Table 2.2. We
obtained 2DTA of BP in cyclohexane to demonstrate signal levels that can be achieved and to
serve as metric for future experiments.

Figure 2.22 – nsTA data obtained on benzophenone pumped via 266 nm. Comparison of BP
in EtOH data obtained from a) the enVISTA and b) the enVISion. See text for experimental
differences. d) A SWTA of BP in CHX and d) a 2DTA of the same sample.

400 500 600 700 800
0
0.05
0.1
0.15
∆Abs (mOD)
Wavelength (nm)
20 ns
50 ns
80 ns
100 ns
500 ns
1000 ns
Benzophenone in EtOH, enVISion b)

400 500 600 700 800
0
0.05
0.1
0.15
∆Abs (mOD)
Wavelength (nm)
20 ns
50 ns
80 ns
100 ns
500 ns
1000 ns
Benzophenone in EtOH, enVISTA
a)
0 1000 2000 3000 4000 5000
Time (ns)
0
0.2
0.4
0.6
0.8
Abs (mOD)
Benzophenone, CHX, 266nm, 525 nm, enVISion
16 scans
3 minutes
700 uOD initially
c)

400 500 600 700 800
-0.2
0
0.2
0.4
0.6
∆Abs (mOD)
Wavelength (nm)
10 ns 200 ns
50 ns 400 ns
100 ns 800 ns
Benzophenone in CHX, enVISion d)
73

Table 2.2 – Experimental Parameters of enVISion with Benzophenone in Cyclohexane
Sample Parameters
Sample
Benzophenone in cyclohexane, 80 µM, 30 mL total volume; Purged with
house N2 (SGM) for 10 minutes,
Sample Cell Custom flow quartz, 1 x 0.5 cm (probe x pump)
Flow Cell
Masterflex Gear pump, 1 setting on “Alice” controller (~5 mL/sec), Viton
tubing
Instrument Parameters
Pump
𝜆𝜆 𝑒𝑒𝑒𝑒
= 266 nm, 2kHz rep rate, 2k shots, 100% power (~100 uJ), 3 x 1.5
mm spot size
Probe Xenon lamp
Detection Grating 2, Detector 1, 350 to 800 nm, 10 nm increment, 9.6 nm resolution
Time Settings 8192 ns, 8 bit (left hand side), 14 bit (right hand side)
PL Subtraction None
Scan 2DTA, 6 Scans, 35 minutes
Workup
Workup Displayed as is; no smoothing or averaging

74

2.3. References
(1) Korovina, N. V.; Joy, J.; Feng, X.; Feltenberger, C.; Krylov, A. I.; Bradforth, S. E.;
Thompson, M. E. Linker-Dependent Singlet Fission in Tetracene Dimers. J Am Chem Soc 2018,
140 (32), 10179–10190. https://doi.org/10.1021/jacs.8b04401.
(2) Das, S.; Thornbury, W. G.; Bartynski, A. N.; Thompson, M. E.; Bradforth, S. E.
Manipulating Triplet Yield through Control of Symmetry-Breaking Charge Transfer. J Phys
Chem Lett 2018, 9 (12), 3264–3270. https://doi.org/10.1021/acs.jpclett.8b01237.
(3) Kennehan, E. R.; Munson, K. T.; Grieco, C.; Doucette, G. S.; Marshall, A. R.; Beard, M. C.;
Asbury, J. B. Influence of Ligand Structure on Excited State Surface Chemistry of Lead Sulfide
Quantum Dots. J Am Chem Soc 2021, 143 (34), 13824–13834.
https://doi.org/10.1021/jacs.1c06248.
(4) Kennehan, E. R.; Munson, K. T.; Grieco, C.; Doucette, G. S.; Marshall, A. R.; Beard, M. C.;
Asbury, J. B. Exciton–Phonon Coupling and Carrier Relaxation in PbS Quantum Dots: The Case
of Carboxylate Ligands. J Phys Chem C 2021, 125 (41), 22622–22629.
https://doi.org/10.1021/acs.jpcc.1c05803.
(5) Miyashita, T.; Jaimes, P.; Lian, T.; Tang, M. L.; Xu, Z. Quantifying the Ligand-Induced
Triplet Energy Transfer Barrier in a Quantum Dot-Based Upconversion System. J Phys Chem
Lett 2022, 13 (13), 3002–3007. https://doi.org/10.1021/acs.jpclett.2c00514.
(6) Rigsby, E. M.; Miyashita, T.; Jaimes, P.; Fishman, D. A.; Tang, M. L. On the Size-
Dependence of CdSe Nanocrystals for Photon Upconversion with Anthracene. J Chem Phys
2020, 153 (11), 114702. https://doi.org/10.1063/5.0017585.
(7) Rimshaw, A.; Grieco, C.; Kennehan, E.; Asbury, J. B. Short Pulsewidth Repetition Rate
Nanosecond Transient Absorption Spectrometer, November 29, 2018.
https://app.dimensions.ai/details/patent/WO-2018217997-A1 (accessed 2023-03-13).
(8) Roy, A. Following Redox Chemistry and Excited State Dynamics in Solution Using Liquid
Jet Photoelectron Spectroscopy, University of Southern California, Los Angeles, CA, 2015.
(9) Pitts, J. N.; Letsinger, R. L.; Taylor, R. P.; Patterson, J. M.; Recktenwald, G.; Martin, R. B.
Photochemical Reactions of Benzophenone in Alcohols 1. J Am Chem Soc 1959, 81 (5), 1068–
1077. https://doi.org/10.1021/ja01514a014.

75

2.4. Appendix
2.4.1. Data Structure of Magnitude Instruments Raw Data File (MATLAB MAT-file)
All the data obtained from Magnitude Instruments instruments in this thesis were in the
format of a MATLAB MAT-file, extension “.mat” whether from the enVISTA at USC or the
enVISions at USC, UCR or Penn State. MAT-files are binary MATLAB files that store workplace
variables and are compatible with all versions of MATLAB published since 2006b. This file
format makes loading data files into MATLAB quite simple; simply set a variable equal to loading
a .mat file and it will be stored in the workplace. See Appendix B, section 3 for a single line for
loading the raw .mat file. Magnitude also offers a .mat to .txt file converter on all their instruments
but this was not used in this thesis but rather a code like mentioned. The MATLAB Import Wizard
was not used but displayed here in the screenshots for ease of variable description.
Every raw data .mat file from Magnitude Instruments contains at least three key
components. The first and most important is the “data” matrix. There are a variety of experimental
setups (Steady state UV-vis; TA and PL experiments with both single wavelength and 2D) which
lead to a variety of collected data and dimensions. These data are stored as matrices in the raw
data .mat file with dimensions that match the number of wavelengths and time points collected
under the matrix name “data”. For example, a steady state absorption spectrum, with a blank
taken, will contain a “data” matrix with three columns: wavelength (nm), diode
intensity/transmission (mV), and absorption (OD) with equal number of rows, with the number
depending on the probe axis defined during the experiment (Figure S2.1a). A blank steady state
spectrum will only contain the first two columns as an absorption spectrum requires an initial
transmission curve to perform the absorption calculation. A single wavelength TA or PL (SWTA
and SWPL, respectively) measurement will contain a data matrix with two columns: time (ns) and
76

Δ 𝐴𝐴 (OD) or diode intensity/PL intensity (mV), respectively (Figure S2.1b). For a PL measurement
only (single wavelength or 2D), will be referred to as “data” in the .mat file. Key point, when PL
subtraction is included and the PL data is recorded, a second matrix is included for the PL, referred
to as “dataPL”. This only occurs when the TA and PL are collected simultaneously.
As the collection of transmission during a SWTA, the intensity after the sample in an
absorption measurement is given for free. For a SWTA, the transmission at the probe wavelength
is recorded as “dataT” (Figure S2.1c). PL subtraction was selected on for this instrument and so
the PL is recorded and stored as “dataPL” ( 𝑘𝑘 𝑃𝑃 𝑃𝑃 ( 𝑘𝑘 , 𝜆𝜆 )). For a 2D measurement, the data is structured
with the time axis as the row header, the wavelengths as the column headers (Figure S2.1d). For
a 2DTA data set, the row 2 is the averaged transmission of the probe at each of the probe
wavelengths, 𝑇𝑇 ( 𝜆𝜆 ) (the row leading with the second 0 in Figure S2.1d). Unlike the SWTA, the
transmission is stored within the data matrix for 2DTA.
77

With every raw .mat file from Magnitude, two other MATLAB structs are stored along
with “data”. The first struct is “fileInfo” and contains minor information pertaining to the specifics
of the instrument used, the software version and the type of scan used (Figure S2.2a). The second
struct is “p”, a construction that contains a list of the parameters used to design the specific
experiment (Figure S2.2b). While this list is nearly exhaustive, it is recommended to the user to
continue to keep a complete list of their parameters used. Notably, the instrument will only save
the parameters inherent to its design and does not record sample name, solvent, conditions, etc.
We encourage the use to utilize notes field on the experimental window.
It was discussed with Magnitude that the .mat files may be replaced with a file type more
transferrable such as a .txt or .csv file extension. Additionally, it is hoped that scans in a data file

Figure S2.1 – Screenshots of the MATLAB 2020b Import Wizard importing various Magnitude
Instruments data files with the “data” matrix selected: a) Steady state/UV-vis; b) SWPL; c)
SWTA; d) 2DTA.
a) Steady State
b) SWPL
c) SWTA d) 2DTA
78

are saved individually and not as a running average as they are currently during thesis submission.
Both of these changes may have taken place.

Figure S2.2 – Screenshots of a Magnitude .mat file with the a) fileInfo struct and b) p struct
selected.
a) fileInfo b) p
79

2.4.2. Dual Phase-Locked Choppers Scheme

Figure S2.3– Transient pump-probe detection scheme using two choppers. The dark shades of
the chopped signals denote blocked pulses. See section 2.2.1.2 for details. Adapted from
Anirban Roy PhD dissertation.(Roy 2015)
80

2.4.3. Subtraction of Two Exponentials
The kinetics of a single state species can be described with an exponential function in the
following equation:
𝑏𝑏 = 𝐴𝐴 ∙ 𝑏𝑏 𝑡𝑡 𝜏𝜏 �
Equation S2.1
Where 𝐴𝐴 is a prefactor, the intensity of the signal at time zero; 𝑏𝑏 is Euler’s Constant; 𝑘𝑘 is time; and
𝜏𝜏 is the lifetime of the species. Two exponentials with the same lifetime can be subtracted from
each other according to the following. We define a pair of exponentials:
𝑏𝑏 1
= 𝐴𝐴 1
∙ 𝑏𝑏 𝑡𝑡 𝜏𝜏 �
; 𝑏𝑏 2
= 𝐴𝐴 2
∙ 𝑏𝑏 𝑡𝑡 𝜏𝜏 �
; Equation S2.2
Then we simply subtract:
𝑏𝑏 3
= 𝑏𝑏 1
− 𝑏𝑏 2
Equation S2.3
𝑏𝑏 3
= 𝐴𝐴 1
∙ 𝑏𝑏 𝑡𝑡 𝜏𝜏 �
− 𝐴𝐴 2
∙ 𝑏𝑏 𝑡𝑡 𝜏𝜏 �
Equation S2.4
And because 𝑏𝑏 𝑡𝑡 𝜏𝜏 �
is a common factor, we factor out the 𝑏𝑏 𝑡𝑡 𝜏𝜏 �
term:
𝑏𝑏 3
= ( 𝐴𝐴 1
− 𝐴𝐴 2
) ∙ 𝑏𝑏 𝑡𝑡 𝜏𝜏 �
Equation S2.6
𝑏𝑏 3
= 𝐴𝐴 3
∙ 𝑏𝑏 𝑡𝑡 𝜏𝜏 �
Equation S2.7
𝐴𝐴 3
= 𝐴𝐴 1
− 𝐴𝐴 2
Equation S2.8
The subtraction of two exponentials with identical lifetimes should lead to an exponential
with the same lifetime but with a prefactor that is the difference between the two prefactors. This
case does not hold if the exponentials have different lifetimes.

81

2.4.4. Subtraction of a Raw Dataset from a Recreated Dataset

Figure S2.4 – Surface plot of a subtraction between a raw TA dataset and the exact same dataset
but recreated according to Equations 2.3 and 2.4. Z-axis units are in OD. The colorbar carries
a 10
-17
exponent.
82

2.4.5. Instrument Response Function of a Magnitude enVISTA
To independently confirm the IRF reported by Magnitude Instruments ( Δ 𝑘𝑘 ~ 5ns) and to
characterize the diode ringing near 𝑘𝑘 0
, a series of IRFs were taken with the enVISTA. All data
presented was pumped with 355 nm internal laser and taken in PL operation mode. The beam was
strongly attenuated in the pump scatter case (OD 2 filter in place, 80% diode current on pump
laser). The natural fluorescence lifetime of fluorescein is 4 ns (section 3.7.3.2). With NaI, a
quencher, it is now ~700 ps, within the IRF width, hence its similarity to the pump scatter profile.
We reconfirm the FWHM or IRF width of the enVISTA to be ~5 ns and expect this to be the case
with the enVISion as the electronics, pump laser, and timing circuits are similar between the two
instruments. Additionally, while the main feature of the IRF is extinguished by 5 ns, the diode
ringing occurs out to ~22 ns in this case. The ringing extends further with higher light intensities
striking the photodiode.
-10 -5 0 5 10 15 20 25 30
-0.2
0
0.2
0.4
0.6
0.8
1
Detector Intensity (norm.)
Time (ns)
Fluorescein
Fluorescein + 0.5 M NaI
Pump Laser Scatter
∆t = 5 ns

Figure S2.5 – Intensity time profile of IRF in the enVISTA with 355 nm excitation in “PL”
mode detection. Black circle) unquenched fluorescein with 0.1 M NaOH in water. Red square)
fluorescein with 0.1 M NaOH and 0.5 M NaI in water. Blue triangle) Collected pump scatter
The detection wavelength on the monochromator was 510 nm for the two fluorescein
experiments and 355 nm for the pump scatter experiment.
83

Chapter 3. Symmetry Breaking Charge Transfer as a Means to Study Electron Transfer
with No Driving Force
1‡

3.1. Abstract
Symmetry breaking charge transfer (SBCT) is a process where a symmetrically disposed
pair of identical chromophores forms a charge transfer excited state with the hole and electron on
different chromophores (chr), i.e., chr–chr + hν / chr
+
–chr
-
. Herein we explore this process in two
dipyrrin-based bichromophoric systems. One of these bisdipyrrins involved a pair of
BODIPY chromophores linked by a single bond at their meso-positions (compound 1) and
the other involved two dipyrrin ligands coordinated in a tetrahedral geometry at the Zn
2+

ion (compound 2). Both compounds show rapid SBCT in polar solvents and only dipyrrin
based emission in nonpolar solvents, the latter arising from a dipyrrin localized excited
sate (LE). By “tuning” the solvent polarity the equilibrium between the LE and SBCT states
can be shifted to favor either state. Ultrafast transient absorption spectroscopy (TA) was
used to probe the kinetics of the charge transfer for 2 in solvents where the electron
transfer is endergonic, exergonic and has a Δ 𝐺𝐺 close to zero. Our TA derived rates were
used to predict fluorescence efficiencies in each of the different solvent systems and
showed a good correspondence to measured values. Detailed density functional theory
(DFT) and time dependent DFT were used to model the ground states as well as the LE
and SBCT states of 1 and 2, in both polar and nonpolar media. The ground and LE
excited states show small dipole moments, while the SBCT states show dipole moments
of 16.4 and 20.3 D for 1 and 2, respectively.

1‡
The contents of this chapter were adapted from Kellogg, M, et al., Faraday Discuss., 2019, 216, 379.
84

3.2. Introduction
The simplest way to form a charge separated excited state in a molecular chromophore is
to link an electron donor to an acceptor so that an intramolecular charge transfer (ICT) state is
formed upon irradiation. Strong coupling of D–A groups gives a good oscillator strength for the
CT transition but lowers the energy of the ICT state relative to the singlet energy of either the
donor or acceptor. This limits the oxidization and reduction potential available from the ICT state
to be well below that of the oxidized donor or reduced acceptor alone, respectively. Linking two
dye molecules into a non-polar symmetric pair that are only weakly coupled is an alternative
approach to promote charge separation (Figure 3.1). An example of this occurs in the bacterial
photosynthetic reaction centre, where two bacteriochlorophylls are spatially close, but weakly
interacting, to form the “special pair”. After bacteriochlorophylls are optically excited, ultrafast
formation of an intradimer CT state is observed.
1
This approach leads to rapid (i.e. ps time scale)
charge separation with an energy loss between the exciton and charge separated state of <100 mV.
2

This rapid formation of the SBCT state is especially useful for red and NIR excited states, such as
those of bacteriochlorophylls, since this will efficiently outcompete fluorescence or nonradiative
decay from the singlet excited state.
The small energy loss in charge separation in SBCT is in contrast to the typical
offset of 500 mV or more between the singlet energy of the donor or acceptor and
D+/A, used to promote rapid intramolecular charge transfer. As previously noted by Rettig,
3

symmetry-breaking CT states facilitate charge separation with minimal energy loss, and
simultaneously slow recombination rates for systems with orthogonal chromophores, both of
which are beneficial in a wide range of applications, from photoelectrocatalysis and photovoltaics
to photodynamic therapy. We have recently shown that SBCT can be used to significantly enhance
85

the open circuit voltage of organic solar cells,
4
by increasing the rate of charge transfer between
donor and acceptors layers in the devices. Symmetry-breaking charge transfer is proposed to occur
in any symmetric molecular dyad provided: (1) the energy of the singlet state for the monomeric

Figure 3.1 - (top) A schematic representation of the decay process for SBCT materials in polar
and non-polar media. Two identical chromophores are represented by the open rectangles and
the asterisk indicates an excited chromophore (singlet or triplet). (bottom) A kinetic scheme is
shown for the SBCT process. The energy of the CT state is strongly solvent independent, while
that of the chromophore localized singlet (labelled LE) is largely solvent independent.
86

component of the dyad is nearly degenerate with or greater than that of redox gap ( 𝛥𝛥 𝐸𝐸 𝑟𝑟 𝑒𝑒 𝑟𝑟 𝑜𝑜 𝑒𝑒 =
𝐸𝐸 1/ 2
𝑜𝑜 𝑒𝑒 − 𝐸𝐸 1/ 2
𝑟𝑟 𝑒𝑒 𝑟𝑟 ), (2) the chromophores are spatially oriented to minimize both orbital and exciton
coupling, and (3) the chromophores are close enough to undergo rapid electron transfer. Until
recently, model compounds capable of SBCT have been largely confined to bi-acenes that absorb
light at predominantly ultraviolet wavelengths.
3,5–7
The 9,9’-bianthryl molecule is the best studied
system of this sort, and the emissive nature of its excited state provides a useful probe to study the
nature of SBCT.
8–10

The approach used to generate chromophore dimers suitable for SBCT has involved
coupling two planar chromophores such that steric constraints force them to be orthogonal to each
other. This geometry guarantees minimal overlap of their wavefunctions in the ground state or of
the hole and electron formed in SBCT. The orthogonal geometry leads to a very small exchange
energy for the singlet and triplet states and thus acts as an efficient intermediate state for the
formation of the chromophore localized triplet (Figure 3.1).
10,11
This efficient triplet formation
has been used to efficiently sensitize singlet oxygen, i.e. chr–chr + hν →
1
chr–chr →
1,3
(chr
+
–chr
-
) →
3
chr–chr +
1
O2 → chr–chr +
3
O2 (chr = chromophore) and been proposed as a heavy metal
free photodynamic therapy agent.
12–14
It is important to note that the SBCT state lives for
nanoseconds before decaying to the chromophore localized triplet in the absence of oxygen and
can be used to drive redox reactions at oxidation and reduction potentials close to the potentials of
the isolated chr + and chr ions, respectively.
87

Cyanine-type dyes such as difluoroboron dipyrrins (BODIPY) have the appropriate
energetics to satisfy requirement (1) above for SBCT and absorb strongly at visible wavelengths.
15

The HOMO and LUMO orbitals are shown in Figure 3.2 for a metal coordinated dipyrrin. The
principal absorption band for these complexes involves a HOMO–LUMO transition, leading to
strong absorption in the 500 nm range for a simple dialkyl-dipyrrin of the type pictured here (ε =
10
5
M
-1
cm
-1
). We have recently reported a meso-bridged BODIPY dimer dyad (1 in Figure 3.2)

Figure 3.2 – (a) Dipyrrin structures used to study symmetry breaking charge transfer in this
paper. Ar = 1-mesityl. (b) The HOMO and LUMO orbitals of BODIPY are shown. The surfaces
are the same for Zn coordinated to the dipyrrin.
88

that undergoes SBCT upon excitation.
16
Excitation of 1 at 500 nm leads to rapid formation of the
SBCT state (<1 ps) in polar solvents such as dichloromethane or acetonitrile. While we have not
been able to get X-ray quality crystals of this compound, modeling predicts that the two dipyrrins
are close to orthogonal (98°). A second class of molecules that have the characteristics to exhibit
SBCT are bis-dipyrrinato (dipy) zinc complexes such as 2.
17,18
Colloquially, 2 has also been
referred to as zDIP2 and are identical molecules. In these complexes the tetrahedral coordination
geometry ensures an orthogonal relationship of the two dipyrrins, minimizing the orbital overlap
and exciton coupling between the two ligands. The Zn complexes are readily synthesized, absorb
strongly in the visible spectrum and do in fact undergo rapid charge separation (<5 ps), even in
solvents of low polarity such as toluene.
17
Interestingly, SBCT occurs despite the fact that Δ 𝐸𝐸 𝑟𝑟 𝑒𝑒 𝑟𝑟 𝑜𝑜 𝑒𝑒
is greater than the singlet energy in these complexes, seemingly counter to the thermodynamic
requirement (1) given above. However, charge separation is favored in this case due to the close
proximity of the chromophores and the additional stabilization from the screened interaction of the
charge separated pair over the cation or anion by the solvent dielectric.
The 9,9’-bianthryl molecule is probably the best studied system exhibiting
SBCT, and the emissive nature of its excited state provides a useful probe to study the nature of
SBCT.
8–10
9,9’-Bianthryl has been studied through time resolved microwave conductivity,
19
dc
photocurrent measurements,
20
and transient absorption measurements.
21–23
The combined studies
infer that the charge transfer is occurring by a through-bond mechanism where the anthracene
chromophores are initially coupled by activation of a torsional mode that has the chromophores
oscillating between 60° and 120° with respect to one another.
22
This torsion results in an initial
partial charge transfer in all solvents, largely at the meso-carbons joining the two anthryl units,
followed by complete interchromophore charge transfer in polar solvents, where the CT state is
89

further stabilized. Herein we will compare the rates of SBCT between the meso-bridged BODIPY
dimer and an organometallic dimer system where the through-bond mechanism is inhibited.
The energy difference between the locally excited state on the chromophore and charge
transfer state is strongly solvent dependent. Both 1 and 2 give high fluorescence efficiencies in
nonpolar solvents, with emission spectra nearly identical to the isolated chromophore, consistent
with the SBCT state lying well above the LE state. As the polarity of the solvent is increased, the
SBCT state is stabilized to the point where it is the lowest energy excited state, and its red-shifted
and inefficient emission takes over (Figure 3.1). Using the polarity of the solvent, we can create a
situation where the LE and SBCT states become degenerate, so the free energy for the charge
transfer reaction is zero. In this paper we discuss these experiments, where solvent mixtures were
used to “fine tune” the solvent polarity. We have measured the kinetic parameters for all of the
steps outlined in the scheme in Fig. 1 as a function of the solvent dielectric and will discuss them
in the context of the charge transfer process in the absence of a driving force, i.e., Δ 𝐺𝐺 𝑜𝑜 = 0. We
will also discuss our molecular modeling of the LE and SBCT states for 1 and 2, to better
understand the structural and electronic changes that take place in this electron transfer process.
3.3. Experimental
3.3.1. Synthetic methods
Both 1 and 2 were prepared in the Mark Thompson group via literature procedures.
16,17

Cong Trinh synthesized 2.
3.3.2. Quantum Yield Measurements
All absorption measurements were taken on a Cary 50 UV-VIS. Steady state Φ
𝑜𝑜 𝑓𝑓
measurements for 2 (zDIP2) were performed using a fluorescence Φ
𝑜𝑜 𝑓𝑓 standard, fluorescein in
aqueous 0.1 M NaOH. The Φ
𝑜𝑜 𝑓𝑓 for each of the pure solvents and solvent mixtures were obtained.
90

First, 2 was dissolved in its respective solvent or solvent mixture to obtain an absorbance of near
or below 0.1 OD at 496 nm, the excitation wavelength used here (Figures S3.1a - S3.4a). This
wavelength was chosen because it was the wavelength used for fluorescein, the quantum yield
reference, as well as being close to the maximum absorption of 2. A quartz cuvette with a
pathlength of 4 mm was used for all Φ
𝑜𝑜 𝑓𝑓 measurements. The emission spectra were then taken
(Figures S3.1b - S3.4b). Fluorescence spectra were obtained on a Horiba Jobin Yvon FluoroMax-
3 fluorometer using DataMax v2.2 software. The parameters used for fluorescence measurements
were 0.25 nm interval, 1 nm spectral resolution, and 100 ms integration time. The reference
fluorescein was run under the same absorption and emission conditions as previously described.
20

The total area of the fluorescence spectrum was then calculated after subtracting off the dark
background. The Φ
𝑜𝑜 𝑓𝑓 in reference to fluorescein was calculated using Equation 3.1:
Φ
𝑜𝑜 𝑓𝑓 ,𝑆𝑆 =
𝐸𝐸 𝑆𝑆 𝑜𝑜 𝑅𝑅 ( 𝑛𝑛 𝑆𝑆 )
2
𝐸𝐸 𝑅𝑅 𝑜𝑜 𝑆𝑆 ( 𝑛𝑛 𝑅𝑅 )
2
∙ Φ
𝑜𝑜 𝑓𝑓 ,𝑅𝑅 Equation 3.1
Here, 𝑆𝑆 and 𝑅𝑅 subscripts refer to the sample and reference respectively; 𝐸𝐸 𝑋𝑋 is the area under
the emission curve; 𝑀𝑀 𝑋𝑋 is the refractive index of the solvent or mixed solvent; Φ
𝑜𝑜 𝑓𝑓 , 𝑅𝑅 is the reference
Φ
𝑜𝑜 𝑓𝑓 of fluorescein (95%)
5
; and 𝑓𝑓 𝑋𝑋 is the absorption correction factor where 𝑓𝑓 𝑒𝑒 = 1 − 10
− 𝑀𝑀 𝑥𝑥
where 𝐴𝐴 𝑒𝑒 is the absorbance at the excitation wavelength. The refractive indices of the pure solvents
were taken from the literature and the solvent mixture refractive indices were calculated as a
weighted sum of the pure solvents, according to the Arago–Biot equation.
24,25
This is not the ideal
case for a binary mixture but the other errors in the Φ
𝑜𝑜 𝑓𝑓 measurement due to absorbance and
fluorescence are assumed to be larger.

91

3.3.3. Transient Absorption
The TA experiments were done after taking the output of a Ti:sapphire regenerative
amplifier (Coherent Legend Elite, 1 kHz, 3.8 mJ, 35 fs) with both single-color pumping and
broadband probing the sample. For the pure solvents and the CHX–THF mixtures, the excitation
pulses were generated from a type-II OPA (Spectra Physics OPA-800CF), centered around 500
nm with a spectral bandwidth of 10–13 nm. For the TOL–THF and CHX–TOL mixtures, a
homebuilt noncollinear optical parametric amplifier (NOPA) was used to pump the sample instead
of the commercial OPA mentioned previously. Here the residual 800 nm pump from the amplifier
was sent through a type I BBO to frequency double the pump which was then combined with a
white light continuum generated by a sapphire disk to generate 500 nm. The signal output was then
sent through a double prism compressor to temporally shorten the pulses.
The probe pulses were generated after taking the residual 800 nm amplifier output, passing
it through a l/2 waveplate, and focusing it onto a rotating 2 mm CaF2 window. This white light
supercontinuum (320 nm to 950 nm) was then collimated and focused at the sample using a pair
off-axis parabolic mirrors. After passing through the sample, the supercontinuum was then
collimated and focused at the slit of a Czerny–Turner monochromator. The probe was then
dispersed onto a 256-pixel silicon diode array (Hamamatsu) for multiplexed detection of the probe.
The probe polarization was set at magic angle (54.7°) with respect to the pump to avoid any
contribution to the observed signal from orientational dynamics.
After passing through a chopper tuned to 500 Hz, the pump was focused before the sample
using a 25 cm lens. The cross-correlation of the pump and probe pulses through a 1 mm quartz
cuvette was determined to have a <150 fs instrument response for the OPA and 80 fs for the NOPA.
The pump was blocked after exciting the sample.
92

3.3.4. Time-Correlated Single Photon Counting
The solvent dependent TCSPC of 2 was performed with the TCSPC apparatus discussed
in full in section 4.3.4. The TCSPC experiments for SBCT were performed first; however, the
experimental setup for the TCSPC of the cMa compounds remains virtually the same.
3.4. Results and Discussion
3.4.1. Electron Transfer at Zero Driving Force
By adjusting the polarity of the solvent, we can shift the energy of the SBCT states to fall
above or below the LE state of 2. The same situation exists for 1, but we will focus solely on 2 in
this section. First, we will consider pure solvents: cyclohexane (CHX), toluene (TOL)
tetrahydrofuran (THF) and acetonitrile (ACN). With cyclohexane as the solvent, the SBCT state
is sufficiently high in energy that it is not populated in the lifetime of the LE state, so the
fluorescence efficiency Φ
𝑜𝑜 𝑓𝑓 is high ( Φ
𝑜𝑜 𝑓𝑓 ~ 0.9). In Figure 3.3 we show the measured Φ
𝑜𝑜 𝑓𝑓 values
for 2 as a function of the solvent dielectric, for both pure and mixed solvents (the mixed solvent
dielectric is taken as the weighted average of the dielectric constants of the two pure solvents).

Figure 3.3 – Photoluminescence quantum yields in a range of solvents and solvent mixtures.
Both measured (open symbols) and values calculated from the TA measurements using
Equation 1.2 (closed symbols) are shown. The Φ
𝑜𝑜 𝑓𝑓 values are plotted versus both solvent
dielectric (left) and solvent polarity, using the ET(30) scale (right).
93

Shifting to toluene decreases the Φ
𝑜𝑜 𝑓𝑓 to ~0.2. This reduction in Φ
𝑜𝑜 𝑓𝑓 is due to the non-emissive
SBCT state dropping below the LE state and rate of populating the SBCT state outcompeting
fluorescence.
17
Shifting from toluene to more polar solvents further degrades Φ
𝑜𝑜 𝑓𝑓 , such that
fluorescence is barely measurable in THF and is not observed in acetonitrile. In order to look at
more subtle effects of the solvent polarity on the photophysical properties we have investigated
Φ
𝑜𝑜 𝑓𝑓 of 2 in solvent mixtures of TOL–CHX, TOL–THF and CHX–THF. The Φ
𝑜𝑜 𝑓𝑓 values in solvent
mixtures fall monotonically with the increasing dielectric constant of the mixture. The Φ
𝑜𝑜 𝑓𝑓 for the
TOL–THF and CHX–THF mixtures seem contradictory. While the solvent ratios chosen here
(volume ratios were TOL–THF: 70/30–85/15 and CHX–THF: 65/35–80/20) have similar
dielectric values, the measured Φ
𝑜𝑜 𝑓𝑓 values differ markedly. It appears that the solvent dielectric
does not accurately represent the solvent effects in these two solvent mixtures. An alternative way
to treat the solvent effects is to consider the polarity of the solvent, using the ET(30) scale.
26
This
solvent polarity scale is based on the absorption spectrum of a charge transfer dye (Reichardt’s
dye: 2,6-diphenyl-4-(2,4,6-triphenylpyridin-1-ium-1-yl) phenolate) dissolved in the solvent or a
solvent mixture. Polar solvents hypsochromically shift the absorbance and less polar solvents lead
to bathochromic shifts. Reichardt used this method to develop a polarity scale for common organic
solvents,
27
which was later extended to solvent mixtures.
28
The absorption band used in deriving
the ET(30) values for each solvent or mixture involved a phenoxide to pyridinium charge transfer,
so many of the same solvent effects captured in the ET(30) scale should be similar for the electronic
transitions being studied here. When the Φ
𝑜𝑜 𝑓𝑓 values are replotted against the ET(30), the values of
the solvent mixtures overlap less (Figure 3.3). While there are still some contradictions at the
points where the TOL–THF and CHX–THF mixtures have similar ET(30) values, the ET(30) scale
94

captures the solvent effects better than the solvent dielectric, so we will use ET(30) in our
discussion hereafter.
The Φ
𝑜𝑜 𝑓𝑓 values in TOL–CHX mixtures are perhaps the most interesting. In this case, the
non-emissive SBCT appears to smoothly transition from not being significantly populated in the
excited state to being the dominant excited state. Assuming that cyclohexane gives an LE state and
toluene gives a predominantly SBCT state, this mixed solvent system allows us to find the
conditions where Δ 𝐺𝐺 𝑜𝑜 = 0 for the LE → SBCT transition and study the rates of charge transfer
and recombination as the system flips from endergonic to exergonic. In order to probe this solvent
region, the rates of forward and backward charge transfer were experimentally determined using
femtosecond transient absorption spectroscopy (TA). The TA is displayed in the appendix to this
chapter, section 3.7.2. The TA data was fit to the three-state model proposed in Figure 3.1, bottom
to determine the rate constants, with methodology described in section 3.7.4.1. To determine the
best fit for each of the samples, the fit for the bands at 380 nm and 538 nm, the absorption maxima
assigned to the CT state, were monitored. Assigning error bars for the fast rate constants, 𝑘𝑘 𝑃𝑃 𝑒𝑒 𝑡𝑡 and
𝑘𝑘 𝑐𝑐 𝑡𝑡 , was achieved by incrementing one value by hand, until the quality of the fit in the time and
spectral domains was obvious as a systematic deviation in the residuals. As the time-range probed
in the TA experiment is not sensitive to ( 𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 ), this value was fixed throughout to the inverse
lifetime measured for the complex in cyclohexane. With the values of 𝑘𝑘 𝑟𝑟 and 𝑘𝑘 𝑛𝑛 𝑟𝑟 from the
cyclohexane solution in hand we can use the lifetime and quantum efficiencies in other solvent
systems to solve for 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 (see section 3.7.4.1, Equations S3.3 and S3.4). For the pure solvents,
THF and acetonitrile, we find that the TA data does not allow robust assignment of 𝑘𝑘 𝑐𝑐 𝑡𝑡 and 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
.
Even after assigning rate constants ~5 times the best fit value, no deviation in any of the time traces
was detected. The ratio of these values is better pinned by the quantum yield measurement. All
95

derived rate parameters from the TA analysis were used to construct an LE state Φ
𝑜𝑜 𝑓𝑓 utilizing
Equation 3.2, and were found to show a good correlation with the experimentally measured values
(Figure 3.3).
29,30

Φ
𝑜𝑜 𝑓𝑓 =
𝑘𝑘 𝑟𝑟 ( 𝑘𝑘 𝑃𝑃 𝑏𝑏𝑏𝑏 + 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 )
( 𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑜𝑜𝑟𝑟 )( 𝑘𝑘 𝑃𝑃 𝑏𝑏𝑏𝑏 + 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 ) + 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 Equation 3.2
Figure 3.4 shows the rate constants as a function of solvent polar for pure solvents. In a
highly polar medium, the charge transfer state is greatly stabilized and as such, the SBCT state is
populated faster than for a nonpolar medium ( 𝑘𝑘 𝑐𝑐 𝑡𝑡 > 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
). For the solvents acetonitrile and THF,
𝑘𝑘 𝑐𝑐 𝑡𝑡 is much faster than 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
, so much so that the TA data can be fit even by simplifying to 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
=
0. This indicates that for these solvents, the equilibrium established is so far to the side of charge
separation that the TA experiment no longer detects the small LE population. On the other hand,
the SBCT state in cyclohexane is too destabilized to support any form of SBCT so the rates for the
forward reaction is assumed to be zero and the rate constants involved on the SBCT side of Figure
3.4 are now undefined. The mixed solvent systems allow us to probe media with polarities between

Figure 3.4 – (left) Rate constants from TA experiments are shown for 2 in a range of solvents,
from nonpolar to polar. (right) Rate constants from TA experiments are shown for both pure
solvent sand solvent mixtures, focusing on the less polar solvents and mixtures. Filled symbols
are 𝑘𝑘 𝑐𝑐 𝑡𝑡 and open symbols are 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
(see Figure 3.1).
96

toluene, where 𝑘𝑘 𝑐𝑐 𝑡𝑡 > 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
and cyclohexane, where 𝑘𝑘 𝑐𝑐 𝑡𝑡 < 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
, showing a clear crossing over of
the rates of forward and backward charge transfer. For a nonpolar medium, the backward charge
transfer rate outcompetes the forward rate. Using the values for 𝑘𝑘 𝑐𝑐 𝑡𝑡 and 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
, the equilibrium
constant and then Δ 𝐺𝐺 of formation of the SBCT state from the LE state was calculated at room
temperature (Figure 3.5). From the calculation of Δ 𝐺𝐺 , we see that all mixtures of CHX and TOL
lead to a positive free energy change. Thus, for a solvent less polar or polarizable than toluene, the
CT state is destabilized but not enough to eliminate the population of this pathway during the
lifetime of fluorescence. In addition to the TOL/CHX mixtures, the 20/80 THF–CHX mixture
gives a Δ 𝐺𝐺 value close to zero. The ET(30) solvent parameters for the CHX–THF mixed solvent
that are >60% CHX may be impacted by the low solubility of Reichardt’s dye in CHX. We are
exploring other solvent models for the mixed solvent systems used here to find a better match to
the experimental trends.

Figure 3.5 – ∆G for the equilibrium LE ⇄ SBCT for 2 as a function of solvent ET(30). A line
has been added at Δ 𝐺𝐺 = 0 as a guide.
97

3.4.2. Modeling LE and SBCT states
The molecular modeling, geometrical, and theoretical calculations of 1 and 2 were carried
by Daniel Sylvinson M. R. and Ali Akil. The first step in the modeling studies was to optimize the
geometries of the ground states for 1 and 2. Both structures were optimized using Density
Functional Theory (DFT) with the B3LYP functional and 6-31+G* basis set. A crystal structure
for 1 is not available for comparison, but the geometry optimized structure of 2 matches the
experimental one closely with no more than 0.03Å difference in corresponding C–C or C–N bond
lengths between the two structures. The bond lengths and angles of the dipyrrin ligands of both 1
and 2 are very similar to other BODIPY and zinc-dipyrrin compounds, respectively. The geometry
optimized structure for the meso-bridged 1 has a dihedral angle of 81° while the geometry
optimized structure of the zinc-based compound 2 is 90°. The dipole moments for the two
compounds are very similar and quite low, <0.2 D. The DFT derived MOs of 1 and 2 (Figure 3.6)
show the top two filled orbitals in both compounds are dipyrrin p-orbitals, and the bottom two
unfilled orbitals are comprised of the dipyrrin π*-orbitals, (see Figure 3.2b for π and π* orbitals).
The LUMO and LUMO+1 orbitals of 1 also show substantial overlap of the pz orbitals of the two

Figure 3.6 – Molecular orbitals for 1 and 2 (DFT: B3LYP functional, 6-31+G* basis).
98

meso-carbons, enabled by the dihedral angle of 81°. The key to the photophysics of these
molecules is their solvent dependence. For this reason, we carried out time dependent DFT
(TDDFT) calculations both in a vacuum and with an implicit solvent continuum to probe the
excited states. The B3LYP functional, 6-31G basis set was used for geometry optimization of
excited states and 6-31+G* was used for electronic structure calculations with the optimized
structures. These TDDFT calculations were carried out with a nonequilibrium implicit continuum
solvation model. More specifically, the conductor-like screening model, COSMO,
31
is used to

Figure 3.7 – SBCT and LE natural transition orbitals (NTOs) for 1 and 2. The hole orbitals are
in blue, and the electron orbitals are in green.
99

qualitatively model the character of the excited states in polar and nonpolar environments.
COSMO is a self-consistent reaction field method. In COSMO, the solute cavity takes the shape
of the molecule and is constructed using atom-centered spheres. The exact electron density of the
solute is used to polarize the cavity surface into discrete point charges. COSMO also includes a
correction for outlying charges, which is the part of the solute electron density that extends beyond
the cavity. It should be noted, however, that COSMO is better suited for modeling high dielectric
solvents. Thus, there is a sacrifice of quantitative accuracy in favor of a qualitative description.
For these COSMO modeling studies, we chose the dielectric of cyclohexane, a solvent
where we see only the LE state, and acetonitrile, where the SBCT state dominates. At both a low
and high dielectric we see the lowest four excited states being composed of two LE states and two
SBCT states. The dihedral angles between the dipyrrin planes of 1 and 2 in both the LE and SBCT
states are within 1° of their ground state values, i.e., 81° and 90°, respectively. The natural
transition orbitals (NTOs) for the LE and SBCT states for 1 and 2 are illustrated in Figure 3.7.
The two SBCT states have a hole on one dipyrrin and an electron on the other ligand, i.e.
(dpy
+
)Zn(dpy
-
) and (dpy-)Zn(dpy
+
). The LE state consists of linear combinations of the dipyrrin
localized sates on the two dipyrrins. The LE states for 1 shown in Figure 3.7 has NTO
contributions (hole + electron) that are 70% from one dipyrrin and 30% from the other. The other
LE state (not shown) has the opposite composition. The LE state does not lead to any charge build
up in the molecule, and as such the LE excited states have dipole moments close to those of the
ground state, 0.6 D for both 1 and 2, with a calculated hole–electron separation of <0.1 Å. The low
ground state and LE dipole moments are consistent with the low degree of solvatochromism for
the absorption and emission spectra for S0 → S1/LE. It is noteworthy that the electron NTO for the
LE state of 1 has character at both meso-carbons, consistent with the non-orthogonal orientation
100

of the two dipyrrins. No such “bridging” character is seen for 2. In contrast to the LE states, the
SBCT states for 1 and 2 give very large dipole moments of 16.4 and 20.3 D and hole–electron
separations of 3.4 and 4.4 Å, respectively, consistent with the high degree of charge separation in
the SBCT state. The SBCT state of 2 shows a significant distortion of the coordination sphere
around the Zn
2+
ion (Figure 3.8). The ground state structure of 2 has the zinc ion equally spaced
between the two dipyrrin ligands, as judged by the zinc to meso-carbon distance (a and b in Figure
3.8). The Zn
2+
ion in the SBCT structure shifts to be closer to the reduced dipyrrin and further from
the oxidized one. This shift in the position of the Zn
2+
ion is dependent on the dielectric of the
continuum, with a nonpolar solvent (low dielectric) giving a 0.24 Å difference in the Zn to meso-
carbon distances, while the shift in a high dielectric medium is 0.16 Å. This is consistent with more
efficient screening of charge in the high dielectric medium. A related, albeit much smaller shift is
observed for 1. In the SBCT state for 1 the B atom is closer to the meso-carbon of the dipyrrin

Figure 3.8 – Geometry optimized excited state structures of 2 differ in media with different
dielectric constants. The structural change here is largely a displacement of the Zn
2+
ion along
the axis containing the two meso-carbons and the Zn
2+
ion, given by a and b in the image. The
effect is marked in the SBCT state, where the Zn
2+
ion moves closer to the dipyrrin carrying
the electron. The effect is most pronounced in nonpolar media.
101

with the electron than it is to the meso-carbon of the dipyrrin with the hole, but the difference in
distances is only 0.06 Å.
A problem with the dielectric continuum model presented above is that while the orbital
make-up and transitions are consistent with the experimental data, the energies of the states are
not well matched to experiments. In both high and low dielectric media the SBCT state falls below
the LE state in energy. This is also true in the vacuum calculations of both 1 and 2. In order to
better model the system, a system with explicit solvent molecules may be needed. Moreover, it has
been found that using common hybrid functionals, such as B3LYP, severely underestimate the
energies of charge transfer states,
32–34
whereas long-range corrected functionals, such as CAM-
B3LYP, ωPBE and ωB97xD, better match the observed energies,
35–38
and will be used in future
studies of 1 and 2. To incorporate an explicit solvent model a hybrid multiscale approach that
employs classical Molecular Dynamics (MD) simulations in conjunction with the aforementioned
higher level TDDFT method is being explored. In this hybrid scheme, the formal atomic charge
forcefield parameters of the chromophore are replaced by the CHELPG (Charges from
Electrostatic Potentials using a Grid-based method) charges calculated for the state of interest
using DFT. MD simulations are then performed with these modified parameters. Single-point
TDDFT calculations are performed on snapshots of the MD equilibrated cell with the solvent
atoms replaced by their corresponding CHELPG atomic charges to serve as a polarizing influence
on the chromophore. The results of these studies will be reported in due course.
3.4.3. Comparing the SBCT process in meso-BODIPY (1) and Zn(dpy)2 (2)
It is instructive to compare the CT and charge recombination rates for compounds 1 and 2,
shown in Table 3.1. The two compounds have very different connections between the two dipyrrin
units, which lead to very different CT rates. The meso-bridged dimer shows a CT rate that is nearly
102

an order of magnitude faster than the zinc dipyrrin complex. The two dipyrrins are orthogonal to
each other in both structures, but the connections between them in 1 and 2 are very different. The
closest contact between the two dipyrrins in 1 is 1.49 Å (C–C bond), while the closest contact
between the two dipyrrins in 2 is 3.39 Å (interligand N–N contact). It is also important to stress
that the differences go beyond just distance. The closest contact for 1 is at the meso-carbon of each
dipyrrin, which is a node in the HOMO and has substantial orbital density in the LUMO (Figure
3.2b). In contrast, the closest dipyrrin–dipyrrin contacts in 2 are the nitrogen atoms of the ligand.
Both the HOMO and LUMO orbitals have substantial nitrogen character, but the symmetries of
the two are opposite. Twisting of the ligands away from an orthogonal relationship leads to a very
weak but positive overlap for the LUMOs of the two ligands and no constructive overlap for the
HOMOs.
Table 3.1 – Kinetic parameters for compounds 1 and 2
Solvent 𝜏𝜏 𝑐𝑐 𝑡𝑡 (ps) 𝜏𝜏 𝑟𝑟 𝑒𝑒 𝑐𝑐 (ns) 𝜙𝜙 𝑜𝑜 𝑓𝑓
1
ACN 0.18 0.65 <0.01
CHX -- -- 0.8
2
ACN 1.1 0.9 <0.01
CHX -- -- 0.9

Looking closely at the NTOs for the LE states of 1 and 2, one can see that there is no
interaction between the hole levels of the dipyrrin ligands for either compound. The same is seen
for the electron levels for 2, where the NTOs remain fully localized on single dipyrrins. However,
the electron levels for 1 show a weak but positive interaction between the two dipyrrin ligands at
the bridging meso-carbons. This is enabled by the 80
o
dihedral angle observed for 1. This meso–
meso orbital overlap is also observed in the LUMO orbitals of the ground state of 1. A similar
orbital overlap is seen for bianthryl and has been suggested as a bridge state for the LE → SBCT
103

transition in that system.
21
In the case of 1 and 2 we postulate that it is this LE bridge state in 1
that increases the rate of charge transfer to the SBCT state by an order of magnitude, relative to
charge transfer rate for 2. In the SBCT states of both 1 and 2 no such bridge state exists; the hole
and electron NTOs are fully localized in single dipyrrin ligands. Thus, it is not surprising that the
recombination rates ( 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 ) are nearly the same for both 1 and 2, with lifetimes of 0.65 and 0.9 ns.
Another interesting comparison is that the emissive state in polar solvents of 1 is from the
CT state, like 9,9’-bianthryl, meaning that radiative recombination to the ground state is weakly
allowed from the SBCT state,
16
whereas 2 displays only emission from the S1 of the BODIPY
chromophore.
17
In order for recombination from the SBCT state to radiatively relax back to the
ground, sufficient orbital overlap and dipolar coupling are necessary. The bridging group in 1
allows for sufficient torsion where the BODIPY chromophores can become less orthogonal,
increasing orbital overlap and dipole coupling.
SBCT in 1 displays charge transfer rates like that of 9,9’-bianthryl (hundreds of
femtoseconds) while in 2 the CT rate is an order of magnitude slower. Like 9,9’-bianthryl, 1 shows
an electron density localized on the bridge. This implies that the BODIPY dimer complex transfers
charge through-bond and may also require activation of the torsion mode as noted for bianthryl,
allowing for charge transfer to occur rapidly, relative to 2. An interesting question is what moves
in the LE → SBCT transition, the hole or the electron? Either carrier could be transferred to form
the SBCT state. While the coupling of the two meso-carbons in the LUMO and electron NTO
suggest it is the electron that will be transferred, the data we have presented here does not allow
us to state which carrier moves unequivocally. We are currently exploring the use of ultrafast
transient absorption anisotropy measurements to answer this question.
104

3.5. Conclusion
Marcus theory relates the rate of electron transfer to the electronic coupling between the
initial and final states ( 𝐻𝐻 𝑀𝑀𝐵𝐵 ), the solvent and molecular reorganization energies ( 𝜆𝜆 ) and the free
energy of the electron transfer process ( Δ 𝐺𝐺 ).
39,40
For a case where Δ 𝐺𝐺 = 0, the forward and
backward rates are determined by only 𝐻𝐻 𝑀𝑀𝐵𝐵 , l and the temperature. The system reported here allows
us to potentially determine the electronic coupling directly if the reorganization energy can be
estimated. Experimental approaches to determining l using the Stokes shift
41–43
are not useful since
the SBCT state does not emit. We could alternatively estimate l theoretically, however, the
dielectric continuum calculations discussed above are not accurate enough for this. We are
currently carrying out QMMM modeling studies with explicit solvent molecules, at a markedly
high level of quantum mechanical theory for the excited chromophores than used above. Deriving
a reliable λ value in mixed solvent systems with this QMMM approach may be problematic but
should give us good values in pure solvents. Thus, we will use QMMM derived l and our
experimental values for Δ 𝐺𝐺 and 𝑘𝑘 𝑐𝑐 𝑡𝑡 in toluene, THF and acetonitrile to measure the electron
coupling in each of these solvents. Solvent dependent structural differences between the LE and
SBCT structures will affect l and may impact the electronic coupling as well.

105

3.6. References
(1) Norris, J. R.; Druyan, M. E.; Katz, J. J. Electron Nuclear Double Resonance of
Bacteriochlorophyll Free Radical in Vitro and in Vivo. J Am Chem Soc 1973, 95 (5), 1680–1682.
https://doi.org/10.1021/ja00786a066.
(2) Parson, W. W.; Scherz, A.; Warshel, A. Antennas and Reaction Centers of Photosynthetic
Bacteria, Structure, Interactions and Dynamics. Springer Series Chem 1985, 122–130.
https://doi.org/10.1007/978-3-642-82688-7_20.
(3) Rettig, W. Charge Separation in Excited States of Decoupled Systems—TICT Compounds
and Implications Regarding the Development of New Laser Dyes and the Primary Process of
Vision and Photosynthesis. Angewandte Chemie Int Ed Engl 1986, 25 (11), 971–988.
https://doi.org/10.1002/anie.198609711.
(4) Bartynski, A. N.; Gruber, M.; Das, S.; Rangan, S.; Mollinger, S.; Trinh, C.; Bradforth, S. E.;
Vandewal, K.; Salleo, A.; Bartynski, R. A.; Bruetting, W.; Thompson, M. E. Symmetry-
Breaking Charge Transfer in a Zinc Chlorodipyrrin Acceptor for High Open Circuit Voltage
Organic Photovoltaics. J Am Chem Soc 2015, 137 (16), 5397–5405.
https://doi.org/10.1021/jacs.5b00146.
(5) Grabowski, Z. R.; Rotkiewicz, K.; Rettig, W. Structural Changes Accompanying
Intramolecular Electron Transfer: Focus on Twisted Intramolecular Charge-Transfer States and
Structures. Chem Rev 2003, 103 (10), 3899–4032. https://doi.org/10.1021/cr940745l.
(6) Giaimo, J. M.; Gusev, A. V.; Wasielewski, M. R. Excited-State Symmetry Breaking in
Cofacial and Linear Dimers of a Green Perylenediimide Chlorophyll Analogue Leading to
Ultrafast Charge Separation. J Am Chem Soc 2002, 124 (29), 8530–8531.
https://doi.org/10.1021/ja026422l.
(7) Giaimo, J. M.; Lockard, J. V.; Sinks, L. E.; Scott, A. M.; Wilson, T. M.; Wasielewski, M. R.
Excited Singlet States of Covalently Bound, Cofacial Dimers and Trimers of Perylene-3,4:9,10-
Bis(Dicarboximide)s. J Phys Chem 2008, 112 (11), 2322–2330.
https://doi.org/10.1021/jp710847q.
(8) Schneider, F.; Lippert, E. Elektronenspektren Und Elektronenstruktur von 9,9’‐Dianthryl.
Berichte Der Bunsengesellschaft Für Physikalische Chemie 1968, 72 (9‐10), 1155–1160.
https://doi.org/10.1002/bbpc.19680720917.
(9) Schneider, F.; Lippert, E. Molekülrechnungen Zur ‐Elektronenstruktur von 9,9′‐Dianthryl.
Berichte Der Bunsengesellschaft Für Physikalische Chemie 1970, 74 (7), 624–630.
https://doi.org/10.1002/bbpc.19700740705.
(10) Grozema, F. C.; Swart, M.; Zijlstra, R. W. J.; Piet, J. J.; Siebbeles, L. D. A.; Duijnen, P. Th.
van. QM/MM Study of the Role of the Solvent in the Formation of the Charge Separated Excited
106

State in 9,9‘-Bianthryl. J Am Chem Soc 2005, 127 (31), 11019–11028.
https://doi.org/10.1021/ja051729g.
(11) Das, S.; Thornbury, W. G.; Bartynski, A. N.; Thompson, M. E.; Bradforth, S. E.
Manipulating Triplet Yield through Control of Symmetry-Breaking Charge Transfer. J Phys
Chem Lett 2018, 9 (12), 3264–3270. https://doi.org/10.1021/acs.jpclett.8b01237.
(12) Duman, S.; Cakmak, Y.; Kolemen, S.; Akkaya, E. U.; Dede, Y. Heavy Atom Free Singlet
Oxygen Generation: Doubly Substituted Configurations Dominate S1 States of Bis-BODIPYs. J
Org Chem 2012, 77 (10), 4516–4527. https://doi.org/10.1021/jo300051v.
(13) Wang, L.; Cao, J.; Wang, J.; Chen, Q.; Cui, A.; He, M. Facile Synthesis of Dimeric
BODIPY and Its Catalytic Activity for Sulfide Oxidation under Visible Light. Rsc Adv 2014, 4
(28), 14786–14790. https://doi.org/10.1039/c4ra01501k.
(14) Cakmak, Y.; Kolemen, S.; Duman, S.; Dede, Y.; Dolen, Y.; Kilic, B.; Kostereli, Z.;
Yildirim, L. T.; Dogan, A. L.; Guc, D.; Akkaya, E. U. Designing Excited States: Theory‐Guided
Access to Efficient Photosensitizers for Photodynamic Action. Angew. Chem. Int. Ed. 2011, 50
(50), 11937–11941. https://doi.org/10.1002/anie.201105736.
(15) Nepomnyashchii, A. B.; Bard, A. J. Electrochemistry and Electrogenerated
Chemiluminescence of BODIPY Dyes. Accounts Chem Res 2012, 45 (11), 1844–1853.
https://doi.org/10.1021/ar200278b.
(16) Whited, M. T.; Patel, N. M.; Roberts, S. T.; Allen, K.; Djurovich, P. I.; Bradforth, S. E.;
Thompson, M. E. Symmetry-Breaking Intramolecular Charge Transfer in the Excited State of
Meso -Linked BODIPY Dyads. Chem Commun 2011, 48 (2), 284–286.
https://doi.org/10.1039/c1cc12260f.
(17) Trinh, C.; Kirlikovali, K.; Das, S.; Ener, M. E.; Gray, H. B.; Djurovich, P.; Bradforth, S. E.;
Thompson, M. E. Symmetry-Breaking Charge Transfer of Visible Light Absorbing Systems:
Zinc Dipyrrins. J Phys Chem C 2014, 118 (38), 21834–21845. https://doi.org/10.1021/jp506855t.
(18) Sazanovich, I. V.; Kirmaier, C.; Hindin, E.; Yu, L.; Bocian, D. F.; Lindsey, J. S.; Holten, D.
Structural Control of the Excited-State Dynamics of Bis(Dipyrrinato)Zinc Complexes: Self-
Assembling Chromophores for Light-Harvesting Architectures. J Am Chem Soc 2004, 126 (9),
2664–2665. https://doi.org/10.1021/ja038763k.
(19) Piet, J. J.; Schuddeboom, W.; Wegewijs, B. R.; Grozema, F. C.; Warman, J. M. Symmetry
Breaking in the Relaxed S1 Excited State of Bianthryl Derivatives in Weakly Polar Solvents. J
Am Chem Soc 2001, 123 (22), 5337–5347. https://doi.org/10.1021/ja004341o.
(20) Smirnov, S. N.; Braun, C. L. Advances in the Transient Dc Photocurrent Technique for
Excited State Dipole Moment Measurements. Rev Sci Instrum 1998, 69 (8), 2875–2887.
https://doi.org/10.1063/1.1149028.
107

(21) Takaya, T.; Hamaguchi, H.; Iwata, K. Femtosecond Time-Resolved Absorption Anisotropy
Spectroscopy on 9,9′-Bianthryl: Detection of Partial Intramolecular Charge Transfer in Polar and
Nonpolar Solvents. J Chem Phys 1998, 130 (1), 014501. https://doi.org/10.1063/1.3043368.
(22) Hashimoto, S.; Yabushita, A.; Kobayashi, T.; Okamura, K.; Iwakura, I. Direct Observation
of the Change in Transient Molecular Structure of 9,9′-Bianthryl Using a 10 fs Pulse UV Laser.
Chem Phys 2018, 512, 128–134. https://doi.org/10.1016/j.chemphys.2017.12.016.
(23) Nishiyama, K.; Honda, T.; Okada, T. Formation of a Complete Electron Transfer State …
— Library of Science. Acta Physica Polonica 1998, 94 (5–6).
(24) Bhatia, S. C.; Tripathi, N.; Dubey, G. P. Refractive Indices of Binary Liquid Mixtures of
(Decane + Benzene) and (Hexadecane + Benzene, or + Hexane) at 303.15, 308.15 and 313.15 K.
Indian Journal of Chemistry 2002, 41A, 266–269.
(25) O., I. S.; B., O. E.; O., A. O.; A., A. S. Estimation of the Refractive Indices of Some Binary
Mixtures. Afr J Pure Appl Chem 2015, 10 (4), 58–64. https://doi.org/10.5897/ajpac2015.0613.
(26) Reichardt, C. Solvatochromic Dyes as Solvent Polarity Indicators. Chem Rev 1994, 94 (8),
2319–2358. https://doi.org/10.1021/cr00032a005.
(27) Reichardt, C. Empirical Parameters of Solvent Polarity as Linear Free‐Energy
Relationships. Angewandte Chemie Int Ed Engl 1979, 18 (2), 98–110.
https://doi.org/10.1002/anie.197900981.
(28) Mancini, P. M. E.; Terenzani, A.; Gasparri, M. G.; Vottero, L. R. Determination of the
Empirical Polarity Parameter ET(30) for Binary Solvent Mixtures. J. Phys. Org. Chem. 1995, 8
(9), 617–625. https://doi.org/10.1002/poc.610080908.
(29) Veldman, D.; Chopin, S. M. A.; Meskers, S. C. J.; Janssen, R. A. J. Enhanced Intersystem
Crossing via a High Energy Charge Transfer State in a Perylenediimide−Perylenemonoimide
Dyad. J Phys Chem 2008, 112 (37), 8617–8632. https://doi.org/10.1021/jp805949r.
(30) Ware, W. R.; Watt, D.; Holmes, J. D. Exciple Photophysics. I. .Alpha.-Cyanonaphthalene-
Olefin System. J Am Chem Soc 1974, 96 (26), 7853–7860. https://doi.org/10.1021/ja00833a002.
(31) Shao, Y.; Gan, Z.; Epifanovsky, E.; Gilbert, A. T. B.; Wormit, M.; Kussmann, J.; Lange, A.
W.; Behn, A.; Deng, J.; Feng, X.; Ghosh, D.; Goldey, M.; Horn, P. R.; Jacobson, L. D.; Kaliman,
I.; Khaliullin, R. Z.; Kuś, T.; Landau, A.; Liu, J.; Proynov, E. I.; Rhee, Y. M.; Richard, R. M.;
Rohrdanz, M. A.; Steele, R. P.; Sundstrom, E. J.; Woodcock, H. L.; Zimmerman, P. M.; Zuev,
D.; Albrecht, B.; Alguire, E.; Austin, B.; Beran, G. J. O.; Bernard, Y. A.; Berquist, E.;
Brandhorst, K.; Bravaya, K. B.; Brown, S. T.; Casanova, D.; Chang, C.-M.; Chen, Y.; Chien, S.
H.; Closser, K. D.; Crittenden, D. L.; Diedenhofen, M.; DiStasio, R. A.; Do, H.; Dutoi, A. D.;
Edgar, R. G.; Fatehi, S.; Fusti-Molnar, L.; Ghysels, A.; Golubeva-Zadorozhnaya, A.; Gomes, J.;
Hanson-Heine, M. W. D.; Harbach, P. H. P.; Hauser, A. W.; Hohenstein, E. G.; Holden, Z. C.;
Jagau, T.-C.; Ji, H.; Kaduk, B.; Khistyaev, K.; Kim, J.; Kim, J.; King, R. A.; Klunzinger, P.;
108

Kosenkov, D.; Kowalczyk, T.; Krauter, C. M.; Lao, K. U.; Laurent, A. D.; Lawler, K. V.;
Levchenko, S. V.; Lin, C. Y.; Liu, F.; Livshits, E.; Lochan, R. C.; Luenser, A.; Manohar, P.;
Manzer, S. F.; Mao, S.-P.; Mardirossian, N.; Marenich, A. V.; Maurer, S. A.; Mayhall, N. J.;
Neuscamman, E.; Oana, C. M.; Olivares-Amaya, R.; O’Neill, D. P.; Parkhill, J. A.; Perrine, T.
M.; Peverati, R.; Prociuk, A.; Rehn, D. R.; Rosta, E.; Russ, N. J.; Sharada, S. M.; Sharma, S.;
Small, D. W.; Sodt, A.; Stein, T.; Stück, D.; Su, Y.-C.; Thom, A. J. W.; Tsuchimochi, T.;
Vanovschi, V.; Vogt, L.; Vydrov, O.; Wang, T.; Watson, M. A.; Wenzel, J.; White, A.;
Williams, C. F.; Yang, J.; Yeganeh, S.; Yost, S. R.; You, Z.-Q.; Zhang, I. Y.; Zhang, X.; Zhao,
Y.; Brooks, B. R.; Chan, G. K. L.; Chipman, D. M.; Cramer, C. J.; Goddard, W. A.; Gordon, M.
S.; Hehre, W. J.; Klamt, A.; Schaefer, H. F.; Schmidt, M. W.; Sherrill, C. D.; Truhlar, D. G.;
Warshel, A.; Xu, X.; Aspuru-Guzik, A.; Baer, R.; Bell, A. T.; Besley, N. A.; Chai, J.-D.; Dreuw,
A.; Dunietz, B. D.; Furlani, T. R.; Gwaltney, S. R.; Hsu, C.-P.; Jung, Y.; Kong, J.; Lambrecht, D.
S.; Liang, W.; Ochsenfeld, C.; Rassolov, V. A.; Slipchenko, L. V.; Subotnik, J. E.; Voorhis, T.
V.; Herbert, J. M.; Krylov, A. I.; Gill, P. M. W.; Head-Gordon, M. Advances in Molecular
Quantum Chemistry Contained in the Q-Chem 4 Program Package. Mol Phys 2015, 113 (2),
184–215. https://doi.org/10.1080/00268976.2014.952696.
(32) Dreuw, A.; Weisman, J. L.; Head-Gordon, M. Long-Range Charge-Transfer Excited States
in Time-Dependent Density Functional Theory Require Non-Local Exchange. J Chem Phys
2003, 119 (6), 2943–2946. https://doi.org/10.1063/1.1590951.
(33) Dreuw, A.; Head-Gordon, M. Failure of Time-Dependent Density Functional Theory for
Long-Range Charge-Transfer Excited States: The Zincbacteriochlorin−Bacteriochlorin and
Bacteriochlorophyll−Spheroidene Complexes. J Am Chem Soc 2004, 126 (12), 4007–4016.
https://doi.org/10.1021/ja039556n.
(34) Tozer, D. J.; Amos, R. D.; Handy, N. C.; Roos, B. O.; Serrano-ANDRES, L. Does Density
Functional Theory Contribute to the Understanding of Excited States of Unsaturated Organic
Compounds? Mol Phys 1999, 97 (7), 859–868. https://doi.org/10.1080/00268979909482888.
(35) Chai, J.-D.; Head-Gordon, M. Systematic Optimization of Long-Range Corrected Hybrid
Density Functionals. J Chem Phys 2008, 128 (8), 084106. https://doi.org/10.1063/1.2834918.
(36) Jacquemin, D.; Perpète, E. A.; Scuseria, G. E.; Ciofini, I.; Adamo, C. TD-DFT Performance
for the Visible Absorption Spectra of Organic Dyes: Conventional versus Long-Range Hybrids.
J Chem Theory Comput 2008, 4 (1), 123–135. https://doi.org/10.1021/ct700187z.
(37) Jacquemin, D.; Wathelet, V.; Perpète, E. A.; Adamo, C. Extensive TD-DFT Benchmark:
Singlet-Excited States of Organic Molecules. J Chem Theory Comput 2009, 5 (9), 2420–2435.
https://doi.org/10.1021/ct900298e.
(38) Tsuneda, T.; Hirao, K. Long‐range Correction for Density Functional Theory. Wiley
Interdiscip Rev Comput Mol Sci 2014, 4 (4), 375–390. https://doi.org/10.1002/wcms.1178.
109

(39) Marcus, R. A. Electron Transfer Reactions in Chemistry: Theory and Experiment (Nobel
Lecture). Angewandte Chemie Int Ed Engl 1993, 32 (8), 1111–1121.
https://doi.org/10.1002/anie.199311113.
(40) Sutin, N. Progress in Inorganic Chemistry. Prog Inorg Chem 2010, 441–498.
https://doi.org/10.1002/9780470166314.ch9.
(41) Mertz, E. L.; Tikhomirov, V. A.; Krishtalik, L. I. Stokes Shift as a Tool for Probing the
Solvent Reorganization Energy. J Phys Chem 1997, 101 (19), 3433–3442.
https://doi.org/10.1021/jp963042b.
(42) Halder, M.; Datta, S.; Bolel, P.; Mahapatra, N.; Panja, S.; Vardhan, H.; Kayal, S.; Khatua,
D. K.; Das, I. Reorganization Energy and Stokes Shift Calculations from Spectral Data as New
Efficient Approaches in Distinguishing the End Point of Micellization/Aggregation. Anal
Methods-uk 2016, 8 (13), 2805–2811. https://doi.org/10.1039/c5ay02982a.
(43) Parson, W. W. Modern Optical Spectroscopy With Exercises and Examples from Biophysics
and Biochemistry Second Edition; Springer Heidelberg, New York, 2015.
https://doi.org/10.1007/978-3-662-46777-0.

110

3.7. Supplement and Appendix
3.7.1. Steady State Absorption and Emission
3.7.1.1. Pure Solvents

3.7.1.2. CHX/THF Mixtures

300 350 400 450 500 550
0
0.05
0.1
0.15
Absorption (OD)
Wavelength (nm)
CHX
TOL
THF
ACN
FLUOR
Pure Solvents
a)
450 500 550 600 650
0
1
2
3
4
5
6
Emission (Counts x10
7
)
Wavelength (nm)
CHX
TOL
THF
ACN
FLUOR
Pure Solvents
b)

Figure S3.1 – Graphs of the steady state (a) absorption and (b) emission spectra for the pure
solvents used in the PL measurements for 2. In all solvents, 2 has a sharp max at ~490-495 nm
with a small red shoulder (470 nm) with some structure in the UV, below 400 nm. The emission
for 2 is mostly broad and featureless. The molecule exhibits very little solvatochromism save
for solvation by acetonitrile in which the absorption maximum blue shifts to 490 nm from 494
nm. The absorption and emission spectra for fluorescein in 0.1M NaOH is also displayed.
300 350 400 450 500 550
0
0.05
0.1
0.15
Absorption (OD)
Wavelength (nm)
65/35
70/30
75/25
80/20
CHX/THF Mixtures
a)

450 500 550 600 650
0
1
2
3
4
Emission (Counts x10
7
)
Wavelength (nm)
65/35
70/30
75/25
80/20
CHX/THF Mixtures
b)

Figure S3.2 – Graphs of the steady state (a) absorption and (b) emission spectra for the
CHX/THF Mixtures used in the PL measurements for 2.

111

3.7.1.3. TOL/THF Mixtures

3.7.1.4. TOL/CHX Mixtures

300 350 400 450 500 550
0
0.05
0.1
0.15
Absorption (OD)
Wavelength (nm)
70/30
75/25
80/20
85/15
TOL/THF Mixtures
a)
450 500 550 600 650
0
0.2
0.4
0.6
0.8
1
Emission (Counts x10
7
)
Wavelength (nm)
70/30
75/25
80/20
85/15
TOL/THF Mixtures
b)

Figure S3.3 – Graphs of the steady state (a) absorption and (b) emission spectra for the
TOL/THF Mixtures used in the PL measurements for 2.

300 350 400 450 500 550
0
0.05
0.1
0.15
Absorption (OD)
Wavelength (nm)
25/75
50/50
75/25
a)
TOL/CHX Mixtures
450 500 550 600 650
0
1
2
3
4
5
Emission (Counts x10
7
)
Wavelength (nm)
25/75
50/50
75/25
TOL/CHX Mixtures
b)

Figure S3.4 – Graphs of the steady state (a) absorption and (b) emission spectra for the
TOL/CHX Mixtures used in the PL measurements for 2.

112

3.7.2. Transient Absorption
3.7.2.1. Pure Solvents
Displayed here are the TA spectra of 2 in various solvents and solvent mixtures. We will
now assign the TA. The spectral features are as assigned as: LE excited state absorption (ESA) at
354nm, SBCT ESA at both 380 nm and 538 nm, ground state bleach (GSB) from 450 to 500 nm
and stimulated emission (SE) from the LE state at 500 nm to 565 nm. As the excited state evolves,
in all solvents except for CHX, the bands at 380 and 538 nm grow in as the SE decreases. For
CHX, there is no obvious spectral evolution forming within the excited state lifetime. There
appears to be no evidence for SBCT formation. The ACN spectrum is seen to decay fastest of all
solvents. Interesting to note for each of the solvents where SBCT does occur, the GSB increases
over time, indicating that another identical chromophore is being bleached due to its involvement
in the SBCT process. This is most obvious in THF and ACN plots (10 ps traces) shown here.

113

350 400 450 500 550 600 650
-30
-20
-10
0
10
0.3 ps
1 ps
10 ps
100 ps
500 ps
900 ps

∆Abs (mOD)
Wavelength (nm)
CHX
Absorption
Emission
ESA of LE
SE
GSB
350 400 450 500 550 600 650
-40
-30
-20
-10
0
10
0.3 ps
1 ps
10 ps
100 ps
500 ps
900 ps
∆Abs (mOD)

Wavelength (nm)
TOL
SE loss
Absorption
Emission
ESA of SBCT
ESA of SBCT
350 400 450 500 550 600 650
-30
-20
-10
0
10
∆Abs (mOD)

Wavelength (nm)
THF
0.3 ps
1 ps
10 ps
100 ps
500 ps
900 ps
Absorption
Emission
GSB growth
350 400 450 500 550 600 650
-30
-20
-10
0
10
∆Abs (mOD)

Wavelength (nm)
ACN
0.3 ps
1 ps
10 ps
100 ps
500 ps
900 ps
Absorption
Emission
Short
Lifetime

Figure S3.5 – Transient absorption spectra of 2 in varying pure solvents with respective solvent
listed above each figure.
114

3.7.2.2. CHX/THF Mixtures

350 400 450 500 550 600 650
-16
-12
-8
-4
0
4

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
900 ps
65/35 CHX/THF
350 400 450 500 550 600 650
-20
-10
0

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
900 ps
70/30 CHX/THF
350 400 450 500 550 600 650
-20
-10
0

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
900 ps
75/25 CHX/THF
350 400 450 500 550 600 650
-12
-8
-4
0

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
900 ps
80/20 CHX/THF

Figure S3.6 – Transient absorption spectra of 2 in varying ratios of CHX and THF. A slight
increase in ESA of the SBCT state is seen as the solvent polarity increases (decreasing ratio of
CHX compared to THF).
115

3.7.2.3. TOL/THF Mixtures

350 400 450 500 550 600 650
-8
-6
-4
-2
0
2

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
1000 ps
70/30 TOL/THF
350 400 450 500 550 600 650
-4
-3
-2
-1
0
1

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
1000 ps
75/25 TOL/THF
350 400 450 500 550 600 650
-6
-4
-2
0
2

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
1000 ps
80/20 TOL/THF
350 400 450 500 550 600 650
-8
-6
-4
-2
0
2

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
1000 ps
85/15 TOL/THF

Figure S3.7 – Transient absorption spectra of 2 in varying ratios of TOL and THF. A slight
increase in ESA of the SBCT state is seen as the solvent polarity increases (decreasing ratio of
TOL compared to THF).
116

3.7.2.4. TOL/CHX Mixtures

350 400 450 500 550 600 650
-10
-5
0

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
1000 ps
25/75 TOL/CHX
350 400 450 500 550 600 650
-15
-10
-5
0

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
1000 ps
50/50 TOL/CHX
350 400 450 500 550 600 650
-5
-4
-3
-2
-1
0
1

∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
1000 ps
75/25 TOL/CHX

Figure S3.8 – Transient absorption spectra of 2 in varying ratios of TOL and CHX. In
transitioning from 25% TOL to 50% TOL, the SBCT bands go from unnoticeable to dominant
(most prominent for the 380 nm band).
117

3.7.3. Time-Correlated Single Photon Counting
In order to determine long time constants outside of the experimental window of the TA,
TCSPC measurements were performed on 2 in each of the pure solvents. Experiments were
performed on 2 with 400 nm pump and detection at 550 nm, emission bandwidth of 4 nm,
collection time of 20 minutes, a t0 offset of 5.2 ns post-processing, and an IRF width of 22 ps. The
longtime fitting constants are assigned as 𝜏𝜏 1
and short time constants are assigned as 𝜏𝜏 2
. The
TCSPC will be more sensitive to the long 𝜏𝜏 1
values. Pump powers are given in the figures.
3.7.3.1. Pure Solvents

Figure S3.9 – TCSPC measurement of 2 in a) CHX, b) TOL, c) THF, and d) ACN. Pump
powers at 400 nm are displayed below the legend.

0 5 10 15 20 25 30 35 40
g(×)10
0
g(×)10
1
g(×)10
2
g(×)10
3
g(×)10
4
Time (ns)
Decay
IRF
Fit
Emission (Counts)
Cyclohexane
a)
P
400
= 9 µW

0 5 10 15 20 25 30 35 40
g(×)10
0
g(×)10
1
g(×)10
2
g(×)10
3
g(×)10
4
Time (ns)
Decay
IRF
Fit
Emission (Counts)
Toluene
b)
P
400
= 4 µW

0 5 10 15 20 25 30 35 40
g(×)10
0
g(×)10
1
g(×)10
2
g(×)10
3
g(×)10
4
Time (ns)
Decay
IRF
Fit
Emission (Counts)
THF
c)
P
400
= 93 µW

0 5 10 15 20 25 30 35 40
g(×)10
0
g(×)10
1
g(×)10
2
g(×)10
3
g(×)10
4
Time (ns)
Decay
IRF
Fit
Emission (Counts)
Acetonitrile
d)
P
400
= 142 µW
118

3.7.3.2. Fluorescein

3.7.3.3. Table of Fitting Values
Solvent 𝜏𝜏 1
(ns) 𝜏𝜏 2
(ns) 𝜏𝜏 3
(ns) 𝐴𝐴 1
(Counts) 𝐴𝐴 2
(Counts)
𝐴𝐴 3

(Counts)
CHX 4.40 ± 0.02 7740 ± 50
TOL 3.87 ± 0.01 0.07 ± 0.02 8740 ± 50 2900 ± 600
THF 2.97 ± 0.04 0.039 ± 0.006 880 ± 10 2400 ± 300
ACN 0.80 ± 0.07 0.025 ± 0.002 4.6 ± 0.1 280 ± 20 4800 ± 500 150 ± 5
Reference -- -- -- -- -- --
Fluor. 4.00 ± 0.02 11150 ± 70

Table S3.1 – Summary of Fitting Parameters for TCSPC of 2 in pure solvents and the
fluorescence standard, fluorescein in 0.1 M NaOH.
Amplitudes of each exponential are assigned as 𝐴𝐴 𝑒𝑒 .
CHX – One exponential was fit as only the LE is populated.
TOL – Two exponentials were used the data where there is now a fast component due to
SBCT.
THF – Two exponentials were used as in toluene.
ACN – Three exponentials were used to fit the data; the 3rd exponential is assigned as an
impurity.

Figure S3.10 – TCSPC measurement of fluorescein in 0.1M NaOH water. One exponential was
fit to the data. This experiment was performed in order to determine the amount of possible
quenching by oxygen by comparing conditions under which the PL measurements were
performed to the literature value. Our value of 4.00 ns compares well with reported value of
4.1 ± 0.1 ns.

0 5 10 15 20 25 30 35 40
g(×)10
0
g(×)10
1
g(×)10
2
g(×)10
3
g(×)10
4
Time (ns)
Decay
IRF
Fit
Emission (Counts)
Fluorescein
P
400
= 140 µW
119

3.7.4. Summary of Rate Constants
3.7.4.1. TA Fitting Protocol – Abstraction of Rate Constants
Displayed below is the summary of the rate constants used to fit the TA, TCSPC and the
PL measurements. The TA was performed in order to determine the fast rates, 𝑘𝑘 𝑐𝑐 𝑡𝑡 and 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
, and
determine 𝛥𝛥 𝐺𝐺 . The TA measurements has sensitivity for processes that occur on the order of
hundreds of femtoseconds to hundreds of picoseconds. However, processes like fluorescence and
charge recombination to the ground state from the SBCT state occurs on the order of nanoseconds,
times that are too long for our TA setup, limited to a max delay of ~1 ns. However, TCSPC can
be used, a technique that is sensitive in the nanosecond regime.
In CHX, SBCT is absent and only fluorescence ( 𝑘𝑘 𝑟𝑟 ) and internal conversion from the LE
( 𝑘𝑘 𝑛𝑛 𝑟𝑟 ) is present. From the QY ( Φ
𝑜𝑜 𝑓𝑓 ) and TCSPC ( 𝜏𝜏 ) of 2 in pure CHX, we pull both 𝑘𝑘 𝑟𝑟 and 𝑘𝑘 𝑛𝑛 𝑟𝑟 ,
according to Equations S3.1 and S3.2.
Φ
𝑜𝑜 𝑓𝑓 𝐵𝐵 𝐶𝐶 𝑋𝑋 =
𝑘𝑘 𝑟𝑟 𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑜𝑜𝑟𝑟 Equation S3.1
𝜏𝜏 𝐵𝐵 𝐶𝐶 𝑋𝑋 =
1
𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑜𝑜𝑟𝑟 Equation S3.2
The values for 𝑘𝑘 𝑟𝑟 and 𝑘𝑘 𝑛𝑛 𝑟𝑟 are now fixed for all solvent systems. From the TCSPC, the
value for 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 can now be pulled from 𝜏𝜏 2
according to Equation S3.4, a simplification of Equation
S3.3 since 𝑘𝑘 𝑐𝑐 𝑡𝑡 , 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
>> 𝑘𝑘 𝑟𝑟 , 𝑘𝑘 𝑛𝑛 𝑟𝑟 , 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 .
29

1
𝜏𝜏 1, 2
=
1
2
� 𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 + 𝑘𝑘 𝑐𝑐 𝑡𝑡 + 𝑘𝑘 𝑃𝑃 𝑒𝑒 𝑡𝑡 + 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 ∓ �( 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
+ 𝑘𝑘 𝑟𝑟 𝑒𝑒𝑐𝑐
− 𝑘𝑘 𝑟𝑟 − 𝑘𝑘 𝑛𝑛 𝑟𝑟 − 𝑘𝑘 𝑐𝑐 𝑡𝑡 )
2
+ 4 𝑘𝑘 𝑐𝑐 𝑡𝑡 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
� Equation S3.3

1
𝜏𝜏 1
=
1
2
( 𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 + 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 ) Equation S3.4
Now, the TA data can be fit. The longtime rate constants, ( 𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 ) and 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 are fixed
and the fast time constants are varied in the TA fitting program. Then 𝑘𝑘 𝑐𝑐 𝑡𝑡 and 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
are determined.
These results are listed in Table S3.2. The values for these are then plugged into Equation 3.1 and
120

the PL quantum yield is back calculated from the kinetic parameters determined from the time-
resolved data.
3.7.4.2. Summary of Rate Constants
We now provide a complete summary of the kinetic rate constants used. Column 2, Percent
refers to the relative amount of the first listed solvent compared to the second one. The SBCT is
not seen in pure CHX but is seen in mixtures with CHX. Therefore a 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 for CHX mixtures exists.
To determine the 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 that CHX would have if SBCT was accessed, a straight line was taken
between the 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 values for TOL and THF and plotted as a function of 𝐸𝐸 𝑇𝑇 (30). From this, a value
for a pseudo- 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 for CHX was imputed to be 1.2 ∙ 10
− 4
𝑝𝑝 𝑠𝑠 − 1
. From this value, 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 for solvent
mixtures containing CHX was calculated as a weighted average of the 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 from CHX and the
other solvent, either TOL or THF. These values are listed below in Table S3.2.

Table S3.2 – Summary of Rate Constants for 2 in various solvents and solvent mixtures. All
values are in inverse picoseconds ( 𝒑𝒑 𝒔𝒔 − 𝟏𝟏 ).
Solvent Percent 𝑘𝑘 𝑟𝑟 (10
− 4
) 𝑘𝑘 𝑛𝑛 𝑟𝑟 (10
− 5
) 𝑘𝑘 𝑐𝑐 𝑡𝑡 𝑘𝑘 𝑃𝑃𝑒𝑒𝑡𝑡
𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑐𝑐 (10
− 4
)
CHX
pure
2.1 2.3 0 0 1.2
TOL 2.1 2.3 0.15 0.057 2.9
THF 2.1 2.3 0.52 0.037 4.5
ACN 2.1 2.3 0.93 0.047 23
CHX/THF
65 2.1 2.3 0.14 0.10 2.4
70 2.1 2.3 0.17 0.11 2.2
75 2.1 2.3 0.15 0.11 2.0
80 2.1 2.3 0.11 0.11 1.9
TOL/THF
70 2.1 2.3 0.28 0.014 3.4
75 2.1 2.3 0.28 0.022 3.3
80 2.1 2.3 0.24 0.036 3.2
85 2.1 2.3 0.24 0.029 3.1
CHX/TOL
25 2.1 2.3 0.20 0.22 2.5
50 2.1 2.3 0.090 0.15 2.1
75 2.1 2.3 0.050 0.20 1.6

121

Chapter 4. Intra- and Inter-Molecular Charge Transfer Dynamics of Carbene-Metal-Amide
Photosensitizers
2‡

4.1. Abstract
A series of steady state and time-resolved spectroscopies were performed on a set of
carbene-metal-amide (cMa) complexes, where M = Cu and Au, that could be used as
photosensitizers for photosensitized electrocatalytic reactions. Using ps-to-ns and ns-to-ms
transient absorption spectroscopies (psTA and nsTA, respectively), the excited state kinetics from
light absorption, intersystem crossing, and eventually intermolecular charge transfer were
thoroughly characterized. Ultrafast intersystem crossing (ISC) rates for these compounds were
obtained from time correlated single photon counting (TCSPC) experiments utilizing a thermally
activated delayed fluorescence (TADF) model, leading to ~3 − 20 ∙ 10
9
𝑠𝑠 − 1
rate constants for
ISC ( 𝑆𝑆 1
→ 𝑇𝑇 1
). These rates were corroborated with psTA, while also confirming previously
instrument limited ISC rates for gold complexes (80 − 130 ∙ 10
9
𝑠𝑠 − 1
). The psTA additionally
abstracted an early time (0.2 − 0.8 ∙ 10
1 2
𝑠𝑠 − 1
) relaxation rate attributed to solvent relaxation and
vibrational cooling. The nsTA experiments for a gold-based cMa complex demonstrated efficient
intermolecular charge transfer from the excited cMa to either an electron acceptor or donor.
Spectroelectrochemical experiments allow us to identify products observed in the nsTA as the
formation of the oxidized and reduced forms of the cMa sensitizer, respectively.

2‡
The contents of this chapter were adapted from Kellogg, M, et al. Intra- and Inter-Molecular Charge Transfer
Dynamics of Carbene-Metal-Amide Photosensitizers. ChemRxiv. Cambridge: Cambridge Open Engage; 2023; This
content is a preprint and has not been peer-reviewed.
122

4.2. Introduction
The need for the replacement of fossil fuels with renewable sources becomes more severe each
day. While many renewable energy sources have been identified, solar energy is the most readily
available source across the earth.
1
While the combination of photovoltaic (PV) panels and batteries
are an efficient means to collect and store solar energy for use at a later time,
2
storing the solar
energy in the form of liquid or gaseous fuels would be advantages from an energy density

Figure 4.1 – Schematic representation of a carbene metal amide (cMa, 𝐴𝐴 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒
𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 ) being used
as a sensitizer for a photoelectrocatalytic reduction reaction. The cycle begins with light
excitation of the 𝐴𝐴 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒
𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 to form the 𝐴𝐴 ( 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒 )
+
( 𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃 𝑒𝑒𝑛𝑛𝑒𝑒 )
−

excited state. The excited state then either
donates an electron to a catalyst (CAT) (right path) and a reductant (RED) returns 𝐴𝐴 ( 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒 )
+
𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 to
the ground state. In alternate path, (left path) the reductant captures the 𝐴𝐴 ( 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒 )
+
( 𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃 𝑒𝑒𝑛𝑛𝑒𝑒 )
−
excited
state, forming 𝐴𝐴 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒
( 𝑐𝑐 𝑎𝑎𝑟𝑟 𝑃𝑃 𝑒𝑒 𝑛𝑛 𝑒𝑒 )
−
, which has sufficient reducing potential to reduce the catalyst. The
reductant can be either an electrode or a chemical reductant.

N
X
M N
+
-
hν
CAT
CAT
-
RED
RED
+
RED
RED
+
CAT
CAT
-
N
X
M N
-
N
X
M N
N
X
M N
+
123

standpoint, especially for use in the transportation sector.
3,4
Utilizing solar energy to drive
electrocatalytic transformation of abundant feedstocks such as water or CO2 to H2, CO, or
methanol is an actively investigated approach to generating solar fuels.
A common approach to producing these solar fuels is to couple a solar photosensitizer with
an electrocatalyst. Upon absorbing light, the photosensitizer (PS) is promoted to its excited state.
The excited PS (PS*) is simultaneously a more potent oxidizing agent and a more potent reducing
agent than the PS.
5
Thus, the PS* can have sufficient chemical potential to oxidize or reduce an
electrocatalyst (EC), driving the production of the fuel. The oxidized PS can then recover an
electron from an electrode or sacrificial reductant (SAC) and repeat the photocatalytic cycle. The
PS/EC/SAC cycle is summarized in Figure 4.1. A similar scheme can be constructed to describe
a photo-oxidative process. This approach shows great promise as a homogeneous system for
generating solar fuels from sunlight without the use of a PV to power the electrocatalysts.
For many years, the most prominent photosensitizers have been based on metal complexes
involving noble metals, such as Ru and Ir.
6
However, the low abundance and high expense of
these elements makes these heavy metal-based sensitizers an unsustainable solution. While
monovalent copper complexes have also been shown to be good photosensitizers, the reported
complexes are four-coordinate, which suffer from deactivation via large excited state
reorganizations limiting their excited state lifetime. In considering new and effective PS
chromophores to study, a number of features are important. The potential PS must have 1) a strong,
broad, and tunable absorption, 2) a long-excited state lifetime to allow for diffusion of PS* to the
EC in solution, 3) photo- and electro-chemical stability, and 4) high excited state electrochemical
potential for oxidation and/or reduction. Additionally, a PS is more sustainable if it is composed
of only earth abundant elements.
124

Recently, our group as well as other have reported photophysical and electroluminescent
properties of linear, two-coordinate carbene-metal-amide (cMa) compounds where M = Cu, Ag,
and Au with long lifetimes (0.2 – 3 µs), and strong, tunable absorption in the UV-visible region
(ε >10
3
M
-1
cm
-1
, λmax = 350 – 600 nm).
7–11
Herein we report a study of the photophysics and
photochemistry of a set of cMa complexes as representative of a larger cMa family previously
described. The nature of the lowest energy excited state in these compounds is an interligand
charge transfer (ICT) state, where the carbene acts as an acceptor and the amide a donor to form
the (c
-
)-M-(a
+
) ICT state. The metal center remains monovalent in the excited state, so geometric
distortions associated with moving from d
10
to d
9
are avoided. The energy of the ICT state can be
tuned by careful selection of each ligand’s electrochemical potential. These compounds emit from
a thermally assisted delayed fluorescence (TADF) process leading to the long excited state
lifetimes.
12,13
The ICT nature of the excited state separates hole and electron instantaneously upon
absorption. Utilizing an ICT state helps predispose the excited state to intermolecular charge
transfer by adopting a molecular geometry akin to the cationic or anionic PS geometry, thus
lowering the internal reorganization energy associated with charge transfer.
14

In this paper we investigate the photophysical preparation (psTA), intermolecular charge
transfer dynamics and electrochemical stability of several cMa complexes (nsTA). The goal is to
demonstrate the efficacy of this class of compounds for use in a photoelectrocatalytic cycle as a
photosensitizer. Using a range of spectroscopic techniques, i.e. pico- →nano-second and
nano- →milli-second transient absorption (psTA and nsTA, respectively), intersystem crossing
(ISC) rates and lifetimes have been determined through time correlated single photon counting
(TCSPC), and characterization of PS
+
and PS
-
analyzed by spectroelectrochemistry (SEC), we
have generated a full picture of the charge transfer dynamics of these cMa complexes.
8,14–16
We
125

also report a nsTA study of intermolecular electron and hole transfer from the excited cMa to
oxidants and reductants in solution, respectively.

4.3. Experimental
4.3.1. Bulk Electrolysis
Bulk Electrolysis experiments were performed by Nina Baluyot-Reyes at NREL. A silver
wire pseudo-reference electrode and gold honeycomb cell card (Pine Research Instrumentation,
NC, USA) which contains the working and counter electrodes were used for bulk electrolysis (BE).
The gold working electrode component of the cell card features honeycomb-shaped perforations
which allow the passage of light. An Ocean FX miniature spectrometer and HL-2000-HP-FHSA
light source (Ocean Insight, FL, USA) were used to obtain absorption spectra during bulk
electrolysis. The absorption spectra were referenced to solvent. Electrode potential was controlled
with an SP-300 potentiostat (Bio-Logic, TN, USA). DPV was performed to determine the potential
at the working electrode and the desired voltage required to reduce/oxidize cMa at the perforation
surface. Chronoamperometry was then performed to determine when equilibration of the sample
had occurred, ~2 min after initialization. Light from the halogen lamp passes through the working
electrode and to be collected and detected by the spectrometer. All electrochemistry was performed
in an argon glovebox and 0.1 M TBA(PF6) in THF.
4.3.2. Pulse Radiolysis
Pulse radiolysis (PR) was used to measure the molar absorptivities of the oxidized and
reduced forms of the cMa photosensitizers studied here, as well as the absorption spectra of the
same complexes in their triplet excited states. PR experiments were conducted by Austin Mencke
and Matthew Bird at the 9 MeV Linear Electron Accelerator Facility (LEAF) at Brookhaven
126

Nation Laboratory (BNL),
17
using pulses less than 50 ps in duration. The optical detection path
consisted of a pulsed xenon arc lamp, a 0.5 cm pathlength quartz optical cuvette fitted with an
airtight Teflon valve, a selectable band pass interference filter (~10 nm) and either a silicon (400-
1000 nm) or a germanium (1000-1500 nm) photodiode (2-3 ns response time). Optical
measurements were collected orthogonal to the direction of the electron pulse through the sample
cuvette.
Cations of the cMa complexes were generated by irradiating aerobic o-xylene or
benzonitrile with β
-
radiation to generate solvent cations, solvated electrons, and solvent excited
states. The dissolved oxygen readily quenches solvent/solute excited states and solvated electrons,
while the solvent cations can be utilized to sensitize the cMa solute. Molar absorptivities were
determined via an internal standard method with triphenylamine, whose cation molar absorptivity
spectrum is known.
Anions of each complex were generated by irradiating THF with β
-
radiation to generate
solvent cations and solvated electrons. Solvent excited states are not appreciably generated in THF
via pulse radiolysis. The THF cation readily decomposes before it has time to pass charge to a
solute, leaving only solvated electrons available for sensitization. The lifetime of the solvated
electron is stabilized by the presence of the electrolyte tetrabutylammonium hexafluorophosphate,
TBAPF, in order to ensure diffusion and reduction of analyte. Molar absorbtivities were measured
via the internal standard method using biphenyl, whose anion molar absorptivity spectrum is
known.
Triplet sensitization experiments were conducted in one of two methods. In the first method
o-xylene degassed with argon was irradiated with β
-
radiation to generate solvent cations, solvated
electrons, and solvent excited states. The excited state singlets decay many times faster than the
127

rate of diffusion in the solvent, such that only triplet state solvent molecules are available for
sensitization experiments. Since the population of the solvent triplet states is many times larger
than the solvent cation or solvated anion, it is assumed that the after sensitization of the solute, the
resulting spectrum is that of the triplet state of the analyte of interest. In the second method, the
solution of o-xylene was kept aerobic. Here the spectra were recorded in the first 10 ns, before
sufficient quenching from O2 has occurred to reduce the population of the excited state. A
comparison of the results of the two methods is made in the main body of the paper.
4.3.3. Sample Preparation for Optical Measurements
The samples were prepared in house-dry toluene or THF. The concentration was set to have
an optical density of 0.1-0.3OD at the pump wavelengths. For the TCSPC and nsTA
measurements, an in-house designed cuvette, ~1 cm pathlength, made from borosilicate glass was
used, utilizing a Schlenk tap to achieve air exclusion. The samples were bubbled with dry nitrogen
for 15 minutes prior to optical work.
For the psTA experiments, a quartz, 1 mm pathlength screw cap cuvette was used. First,
the cMa compound was prepared in dry solvent (toluene or THF) at the appropriate optical density
at 405 nm. This solution was the then bubbled with dry house nitrogen. The deareated solution
was quickly transferred to the cuvette and capped with a septum. These solutions were all used
within a few hours of preparation as the atmosphere can leak even after a day.
For flow cell experiments, a custom made 1 cm glass pathlength cuvette with two 3 mm
outer diameter sidearm inlet/outlet was used. The inlet sidearm was attached via a FEP lined Tygon
tube (ID 1/8” OD 1/4”) threaded through a 14/20 joint compression clamp thermometer adapter
into a 50 mL three-neck round bottom flask (RBF). The outlet sidearm was attached via the FEP
line Tygon tubing to a gear pump (Cole-Parmer No. 7144-05), which also fed back to the
128

three-neck RBF. The third neck of the RBF was fitted with an appropriately sized rubber septa that
was used to bubble degas the reservoir in operando. The solvent was allowed to circulate through
the system for five minutes while being bubble degassed before spectra were collected. The final
reservoir volume accounting for volume of the lines, cuvette, and pump amounts to ~100 mL and
the system has a flow rate of 40 mL/min.

4.3.4. Time-Correlated Single Photon Counting
The TCSPC experiments in this section were performed by Fabiola Cardoso-Delgado and
Thabassum Natti-Kallungal in the Bradforth group. To determine the ISC rates of the cMa
complexes, the prompt and delayed emission lifetimes and amplitudes were determined by
employing time-correlated single photon counting technique. The samples were excited by 400
nm laser pulses produced by frequency-doubling the output from a Ti:sapphire regenerative
amplifier (Coherent RegA 9050, 800 nm) operating at a 100 kHz repetition rate. The 400 nm
excitation pulse was focused on the sample with a focusing lens of 10 cm. Emission was collected
with a collection lens kept perpendicular to the excitation beam. For collecting the emission, a
Digikröm CM112 double monochromator was used, equipped with a slit width of 0.6 mm and a
grating of 1200 lines/ mm, achieving 4.5 nm spectral resolution. The detected emission wavelength
varied per compound (492 to 620 nm). The Hamamatsu R3809U-50 PMT attached at the exit slit
of the monochromator operating at 3 kV provided an instrument response of 22 ps. The signals are
then amplified and directed toward the Becker and Hickl SPC-630 photon-counting board. To
avoid pulse pile-up, all the experiments were performed by keeping the photon counting rate <2%
of the repetition rate of the laser.

129

4.3.5. Picosecond Transient Absorption
Cardoso-Delgado and Kellogg contributed equally to the acquisition of psTA data. The
picosecond transient absorption (psTA) setup has been described previously in literature
14
and in
section 3.3.3 of this thesis but will be described in full here. The probe line is fundamentally the
same whereas the pump line is different so emphasis here will be placed on the pump line. Pump
probe experiments were performed using the output of a Ti:sapphire regenerative amplifier
(Coherent Legend Elite, λ = 810 nm, ∆λ = 30 nm, 1 kHz, 2.9 mJ, 35 fs) employing a single
wavelength excitation pulse and broadband probing with a white light supercontinuum pulse. The
excitation pulses (λ = 405 nm, ∆λ = 6.5 nm) were generated by directing a portion of the amplifier
output into a 2 m lens which softly focused the 810 nm into a Type I BBO (Red Optronics, 500
µm). Approximately, 45 mW of 405 nm is generated from ~250 mW of 810 nm, leading to 17%
conversion. An uncoated fused silica window with a 3.3° wedge (CVI Laser, 3 mm thickness) used
in reflection geometry is used to attenuate the 405 nm by a factor of ~20 to 2 µJ. The wedge was
placed so that the beam reflected off the back surface was dumped onto the surface of an iris,
allowing the front-reflected beam to pass through cleanly. The 405 nm is steered by a subsequent
pair of dichroic mirrors, dumping the residual 810 nm. The beam is sent through a synchronous
optical chopper (ThorLabs, MC1F10, 10 blades) set at half the laser repetition rate. The beam is
then directed to a 25 cm, CaF2 focusing lens which focuses the beam ~5 cm before the sample.
After sample excitation, the beam is dumped. The pump spotsize was 450 µm for most of the
samples with ~2 µJ at target, achieving a fluence of ~1200 µJ/cm
2
. 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 was pumped with 700
nJ and a 250 µm spotsize.
The probe pulses were generated after taking 10% of the 810 nm amplifier output and
passing it through a λ/2 waveplate (λc = 810 nm, Ø1/2”). The probe polarization was set at magic
130

angle (54.7
o
) with respect to the pump to avoid any contribution to the observed signal from
orientational dynamics. The 810 nm was focused by a 10 cm onto a circularly rastering 2 mm CaF2
window (Koch Crystal, Ø1”, 111 crystal surface). The white light supercontinuum (320 nm to 950
nm) was then collimated and focused on the sample using a pair off-axis aluminum parabolic
mirrors (Janos Technology, focal lengths 5 and 25 cm respectively) to a spotsize of 40-70 µm
depending on the color and placement of the cuvette along the beam propagation axis. After
passing through the sample and before entering the spectrometer, the probe was sent through a 800
nm HR to reject the residual 810 nm pump. Then, the supercontinuum was then collimated and
focused on the slit of a Czerny–Turner monochromator. The probe was then dispersed by a
diffraction grating (Newport, 500 nm blaze, 150 lines/mm, 2.2
o
nominal angle) onto a 256-pixel
silicon diode array (Hamamatsu) for multiplexed detection of the probe. A 150 lines/mm grating
(~2 nm resolution) used in the psTA experiments except for the 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 experiment, a 300
lines/mm grating was used that obtains twice the spectral resolution (1 nm) but a ~1/2x reduced
spectral range for the dataset. The cross-correlation of the pump and probe pulses through a 1 mm
quartz cuvette was used to determine a ≤ 340 fs instrument response, which is entirely limited by
signal from the cuvette windows. Similar pump fluence (~0.2 mJ/cm
2
) and sample ground state
absorption (0.15-0.25 OD at 405 nm) were maintained experiment to experiment.
4.3.6. Nanosecond Transient Absorption
Every author from Kellogg and Mencke et al contributed in some form to the acquisition
of nsTA data. The nsTA experiments were performed with a series of instruments from Magnitude,
described previously
18
and far more extensively in chapter 2. Most experiments were performed
on an enVISTA instrument at USC unless otherwise noted. The pump beam was generated from
the third harmonic output of a pulsed Nd:YAG laser at 355 nm housed within the overall
131

instrument. The pump laser was set to a 5 kHz repetition rate with routine pulse energies of 40-75
µJ, achieved by varying the laser diode current (80-100%). The pump beam was ~8 mm Ø,
achieving pump fluences of 80-150 µJ/cm
2
. Here, a xenon lamp output acting as the probe was
collected and focused into the sample chamber at the sample position. The diverging probe light
was recollimated and focused into the monochromator and onto a fast photodiode. The
monochromator was equipped with slits of 1.2 mm, affording 6.3 nm resolution. The oscilloscope
voltage resolution was set to 8 bit for a 2 ns step size.
Several of the neat cMa samples and the quenching experiments of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with MePI in
toluene were performed at Magnitude Instruments facility (Pennsylvania State University, State
College, PA) with an enVISion instrument where the pump beam was generated with an external
355 nm laser. Finally, the high quencher concentration (30 mM) experiments with MePI and the
BIH experiments for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF were performed at the University of California, Riverside
(UCR) in the Chris Bardeen lab with a second enVISion instrument. The pump beam was
generated by a 50 Hz repetition rate optical parametric oscillator (OPO), identity unknown, tuned
to 450 nm (420 nm for the MePI experiment). The pump beam spotsize was set to ~8 mm Ø with
a pulse energy of ~1.3 mJ, achieving a pump fluence of ~2.6 mJ/cm
2
. The higher pump fluence
leads to higher signal at the cost of signal to noise due to reduced averaging from the low laser
repetition rate for similar experimental acquisition times. The NIR probe region was collected
where the probe light was taken from a halogen lamp and detected via an InGaAs photodiode. The
response on the NIR photodiode afforded a ~10 ns temporal resolution.

132

4.4. Results and Discussion
4.4.1. Ground and Excited State Spectra of cMa Complexes
The general structure for the cMa complexes discussed in this study are displayed in Figure
4.2. The cMa compounds presented here are a class of luminescent, linear, two-coordinate metal
complexes composed of a carbene (blue) and a carbazole (red) bonded to a central Cu
+
or Au
+

atom. The photophysical properties of the cMa complexes have been previously reported.
8,9
The
steady state absorption and emission spectra in THF or 2-MeTHF are displayed in Figure 4.3.
Each complex features structured absorptions from 300-380 nm attributed to carbazole,
9
as well
as a relatively strong (ε = 5000-9000 M
-1
cm
-1
), broad CT transition between 375-550 nm. The
energy of the ICT band can be easily controlled via careful selection of the carbene and carbazole,
with the ICT transition having lower energy with increasing carbene electrophilicity
(DAC > MAC > CAAC) as well as increasing carbazole nucleophilicity
(Cz > PhCz > BCz >> CNCz). The emission spectra all present as structureless bands indicative
N N
N
O
Dipp Dipp
Dipp
iPr
N N
O
Dipp Dipp
X
M
N
N
R R
Dipp
CAAC
MAC
DAC
O
R = H
R = CN
R = t-Bu
R = Ph
M = Au
+
or Cu
+
*
N
O
O
N
N
Ph
MePI
BIH

Figure 4.2 – Left) General structure of the cMa compounds in this article. The cMa on the left
is displayed in its excited ICT state. The carbene ligand is displayed in blue, the amide is
displayed in red. Center) R-group substitutions on carbazole. Right) Carbene ligands used in
this study. Bottom) N-methylphthalimide (MePI) and dihydrobenzimidazole (BIH) used in
electron and hole transfer studies, respectively.
133

of ICT states, with λmax values spanning a 220 nm range. The Stokes shifts in THF or 2-MeTHF
range from 550 to 900 meV, indicating large excited state relaxation. Each complex has a long
lifetime in polar solvents (τ = 0.08-2 µs), comparable to the lifetimes of 3- and 4- coordinate copper
photosensitizers.
6

Figure 4.3 – (a) Steady state molar absorptivity spectra in THF for 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 and 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (top) and
𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 (bottom). 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 is in 2-MeTHF. The pump wavelengths used in the
laser experiments are indicated as dashed lines: nsTA at 355 nm, TCSPC at 400 nm and psTA
at 405 nm. (b) Steady state emission spectra of 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 and 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in 2-MeTHF (top) and 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 ,
𝐴𝐴𝐴𝐴
𝑃𝑃 ℎ 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in 2-MeTHF. 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 is in THF.

0
2
4
6
8
10
Cu
CAAC
Cz
Au
CAAC
Cz
Cu
MAC
Cz
Au
MAC
Cz
(a)

0
0.2
0.4
0.6
0.8
1
Cu
CAAC
Cz
Au
CAAC
Cz
Cu
MAC
Cz
Au
MAC
Cz
(b)

300 350 400 450 500 550
0
4
8
12
Wavelength (nm)
Cu
MAC
BCz
Au
MAC
BCz
Cu
MAC
PhCz
Cu
DAC
CNCz

400 500 600 700 800
0
0.2
0.4
0.6
0.8
1
Wavelength (nm)
Cu
MAC
BCz
Au
MAC
BCz
Cu
MAC
PhCz
Cu
DAC
CNCz
Molar Absorptivity (10
3
M
-1
cm
-1
)
Emission Intensity (norm.)
134

4.4.2. Spectroelectrochemistry
Transient absorption spectroscopy (TA) is a powerful tool for determining the mechanism
of excited state dynamics; however, analysis can prove complicated due to overlapping features.
The features of transient absorption can be divided into three categories: ground state bleach
(GSB), stimulated emission (SE), and excited state absorption (ESA). The GSB and SE can be
identified and assigned using the ground state absorption spectrum and fluorescence/spontaneous
emission spectrum, respectively. However, the ESA can prove to be a more difficult feature to
assign unless prior spectroscopic knowledge of the excited states formed transiently is available
or assumptions are made about these spectra. Recently, McCusker et al. has reported that the TA
spectra of the lowest MLCT state of a family of group 8 bis-terpyridine compounds can be
approximated as the sum of the cation and anion spectra with the ground state spectra subtracted.
19

In this light, we first present a series of pulse radiolysis (PR) experiments to measure triplet excited
state absorption spectra, then spectroelectrochemical (SEC) measurements, using both bulk
electrolysis (BE) and PR, of the cation and anion absorption spectra of the cMa compounds. Both
the triplet and cation/anion spectra will be used to assign the TA spectra.
Pulse radiolysis is used to measure the triplet state absorption spectra of compounds by
utilizing β
-
radiation to generate solvent triplet states which are then used to sensitize the analyte.
20

All five Cu-based compounds’ triplet ESA were measured in aerobic o-xylene and recorded in the
early time frame (<5 ns) before competition by cation sensitization and oxygen quenching of the
excited state could occur (Figure 4.4a and Figure S4.3). The validity of this approach was
confirmed by measuring the spectra of three of the compounds in an anerobic environment where
only triplet sensitization occurs (Figure 4.4b). The resulting spectra are in good agreement (Figure
S4.3). For each of the cMa complexes studied here, the excited state spectra have a peak between
135

600-800 nm. In addition, the excited state spectrum of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 displays a NIR absorption located
near 950 nm.

For compounds with fully reversible electrochemical oxidation and/or reduction, the
absorption spectra of their oxidized (cation) or reduced (anion) forms can be readily obtained by
measuring the spectra of the compound in solution as it is oxidized or reduced during bulk
electrolysis (BE).
19
Figure 4.5 displays the BE absorption spectra for 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 at
positive, negative, and neutral applied voltages. We assign the spectra at positive applied voltages
to the cation (denoted 𝐴𝐴 ( 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒 )
+
𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 ) and negative applied voltages to the anion (denoted
𝐴𝐴 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒
( 𝑐𝑐 𝑎𝑎 𝑟𝑟𝑃𝑃 𝑒𝑒 𝑛𝑛𝑒𝑒 )
−
). For 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
𝑀𝑀𝑀𝑀 𝐵𝐵 , the spectrum shows a vibronically structured absorption with the
principal 𝑆𝑆 1
(0-0) peak maximum occurring at 720 nm. The absorption spectra bears striking
similarities to free carbazolium
7,9,21,22
which is consistent with oxidation being a carbazole
centered event.
7–9
For 𝐴𝐴𝐴𝐴
( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
𝑀𝑀𝑀𝑀 𝐵𝐵 , the spectrum displays a 10 nm red shift compared to 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
𝑀𝑀𝑀𝑀 𝐵𝐵 ,
500 600 700 800 900 1000
0
0.2
0.4
0.6
0.8
1
Wavelength (nm)
Cu
MAC
Cz
Cu
MAC
BCz
Cu
MAC
PhCz
∆Abs (norm.)
(a)
500 600 700 800 900 1000
0
0.2
0.4
0.6
0.8
1
Wavelength (nm)
Cu
MAC
Cz
Cu
CAAC
Cz
Cu
DAC
CNCz
∆Abs (norm.)
(b)

Figure 4.4 – (a) Excited state absorptivity spectra of 10 mM 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , and 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in a
solution of 20 mM triphenylamine in o-xylene 5 ns after electron pulse. (b) Excited state
absorptivity spectra of 10 mM 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 , and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in o-xylene 10 ms after electron
pulse. Minor differences between the two experimental conditions are observed in the spectra
for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 . Data collected is displayed as symbols and the lines are interpolations between data.
136

but has the same shape. The spectra recorded under reducing conditions for both compounds
present as a broad featureless transition from 400 to 900 nm. Due to the timeframe of collecting a
spectrum by bulk electrolysis (minutes), it is unclear whether these spectra represent the reduced
molecular species or the formation of aggregates or colloids at the electrode surface.

For compounds with non-reversible redox events, monitoring the absorption spectra during
a BE experiment is problematic due to competing rates of ion degradation. In this case, PR is useful
for measuring the spectra of ions before they decay (<10 µs). An additional benefit of PR over BE
is that the former gives the molar absorptivity spectra of the charged species. The molar
absorptivity spectra of various Cu-based cMa ions are represented in Figure 4.6 and Figure S4.1.

Figure 4.5 – The bulk electrolysis spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF. Neutral (0 V) – black,
Cation (+0.9 V) – Red, Anion (-2.1 V) – Blue. The voltages were referenced to a silver pseudo-
reference electrode. The peak centers of the cations are also indicated by arrows. A 7-point
smooth was applied to the data to remove high noise in the < 500 nm region.

400 500 600 700 800 900
0
0.2
0.4
0.6
0.8
1
Absorption (OD)
Wavelength (nm)
V = 0
V = +
V = -
605 nm
660 nm
730 nm
Au
MAC
BCz

400 500 600 700 800 900
0
0.5
1
1.5
2
2.5
3
Absorption (OD)
Wavelength (nm)
V = 0
V = +
V = -
720 nm
Cu
MAC
BCz
655 nm
595 nm
137

Irradiating non-polar solvents generates a large amount of excited solvent molecules and a
relatively small number of solvent cations and solvated electrons. However, by keeping the solvent
aerated, the excited states and solvated electrons can be quenched out, leaving only the solvent
cations to sensitize the analyte. For each of the Cu cMa cations in aerobic o-xylene, the molar
absorptivity spectra feature a vibronically structured band, located between 600-800 nm in 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 ,
𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , and 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , and between 800-1000 nm for 𝐴𝐴𝐴𝐴
𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 . Incomplete oxidation in o-xylene is
observed for 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 , likely due to charge delocalization across multiple solvent molecules
stabilizing the solvent cation, requiring a switch to benzonitrile. In benzonitrile, 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 presents
as a structureless band between 800-1000 nm (Figure S4.2). The shift to a more polar solvent could
be the cause for the loss of structured absorption.
In order to measure anion spectra, THF is irritated by β
-
radiation yielding solvated
electrons and solvent cations. The solvent cations of THF rapidly decompose, leaving only the
solvated electrons to sensitize analyte. Due to reduction of the various Cu cMa compounds being

Figure 4.6 – (a) Cation molar absorptivity spectra of 10 mM 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (blue), 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (red), and
𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (black) in a 20 mM solution of triphenylamine in o-xylene. (b) Anion molar absorptivity
spectra of 10 mM 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (blue), 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 (green), and 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 (orange) in 10 mM TADPF in
THF. Data collected is displayed as symbols and the lines are interpolations between data.

500 600 700 800 900
0
2
4
6
8
10
Cations
Wavelength (nm)
Cu
MAC
Cz
Cu
MAC
BCz
Cu
MAC
PhCz
Molar Absorptivity (10
3
M
-1
cm
-1
)
(a)
600 800 1000 1200 1400 1600
0
1
2
3
Anions
Wavelength (nm)
Molar Absorptivity (10
3
M
-1
cm
-1
)
Cu
MAC
Cz
Cu
CAAC
Cz
Cu
DAC
CNCz
(b)
138

carbene centered,
7–9
𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 , and 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 each possess unique anion molar absorptivity
spectra. 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 possess broad featureless absorptions extending from <500 to 1000 nm
and from <500 to 700 nm, respectively. 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 displays a large featureless absorption from
600-1000 nm, as well as a large, well-defined peak centered at ~1300 nm. Wavelengths below
500 nm cannot be measured for these compounds due to the presence of the substantial ground
state absorption limiting the detection of the charged species. Since the reduced species are present
in a relatively low concentration, being generated uniformly through the cuvette, and spectra are
collected in less than 10 µs, the formation of colloids can be excluded. The similarity of the anionic
𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 spectra generated via BE to the anionic spectra generated via PR suggests the formation of
aggregates or colloids in the BE experiment does not take place. Both the cationic and anionic
absorption spectra are useful for providing basis spectra for analyzing intra- and inter-molecular
charge transfer events via TA.
4.4.3. ISC Rates Determined by TCSPC
Previously, we have estimated the rates of ISC for 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 and 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 compounds, where
M = Cu, Ag, Au, using TCSPC.
8
Here, we extend these measurements to include a more accurate
picture of the ISC process, with differing carbazole ligands and solvents, as well additional carbene
ligands. We also use a more complete model than the one used in our previous study, including
both the 𝑇𝑇 1
→ 𝑆𝑆 1
and 𝑆𝑆 1
→ 𝑇𝑇 1
ISC rates. This allows for an estimate of Δ 𝐸𝐸 𝑆𝑆 𝑇𝑇 for all compounds
so far considered.
The fluorescence decay curves obtained for the cMa compounds possess bimodal
fluorescence decay curves with two distinct decay phases: prompt (picosecond, time constant 𝜏𝜏 𝑝𝑝 )
and delayed fluorescence (decaying up to a microsecond, time constant 𝜏𝜏 𝑇𝑇 𝑀𝑀𝐷𝐷 𝑇𝑇 ). The prompt
fluorescence is emission directly from the singlet state before ISC occurs, while the delayed
139

fluorescence is a result of TADF, which results from a rapid equilibration of the 𝑆𝑆 1
and 𝑇𝑇 1
excited
states, before ultimate emission from the singlet state. When time resolved, the majority emission
% amplitude (Ap) is prompt, while around 1-2% amplitude describes the delayed component
(ATADF = 1 – Ap); this relative fraction can be captured accurately because of the high dynamic
range when photon counting. (The delayed fraction has the higher overall integrated area).
Previous work simply assumed the ISC rate could be set exactly to 𝜏𝜏 𝑝𝑝 − 1
. This approach is
oversimplified when we consider the thorough TADF kinetics when the singlet-triplet gap is only
a few 𝑘𝑘 𝐵𝐵 𝑇𝑇 .
8
To more accurately determine the ISC rates in the presence of a fast singlet-triplet
equilibrium, we employed the kinetic solution reported by Adachi et al.
23
The exact equation for
the emission decay proposed by Adachi et al. is based on a three state model. The three states
model comprises 𝑆𝑆 1
and 𝑇𝑇 1
as excited states and 𝑆𝑆 0
as ground state (Figure 4.7). The rate equations
for the 𝑆𝑆 1
and 𝑇𝑇 1
population can be written as:

𝑟𝑟 𝑟𝑟𝑡𝑡
�
[𝑆𝑆 1
]
[𝑇𝑇 1
]
�= �
−( 𝑘𝑘 𝑟𝑟 𝑃𝑃 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 𝑃𝑃 + 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 ) 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 − � 𝑘𝑘 𝑟𝑟 𝑇𝑇 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 𝑇𝑇 + 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 �
� �
[ 𝑆𝑆 1
]
[ 𝑇𝑇 1
]
� Equation 4.1
The emission decay directly mirrors the time-dependent population of the 𝑆𝑆 1
state.
24
The
intersystem crossing rates reported here are obtained by fitting the emission decays to the solutions
to the above rate equation with the assumption of 𝑘𝑘 𝑟𝑟 𝑇𝑇 = 0 and 𝑘𝑘 𝑛𝑛 𝑟𝑟 𝑇𝑇 = 0, reasonable based on low
temperature phosphorescence lifetimes determined previously, can be further simplified as
follows. Because 𝑘𝑘 𝑟𝑟 𝑆𝑆 and 𝑘𝑘 𝑛𝑛 𝑟𝑟 𝑆𝑆 (together defining 𝜏𝜏 𝑇𝑇 𝑀𝑀𝐷𝐷 𝑇𝑇 − 1
) are small compared to both 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 and
𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 , a very simple relationship exists to extract both intersystem crossing rates:
𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 = 𝐴𝐴 𝑝𝑝 1
𝜏𝜏 𝑝𝑝 Equation 4.2
𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 = � 1 − 𝐴𝐴 𝑝𝑝 �
1
𝜏𝜏 𝑝𝑝 Equation 4.3
Note that:
140

𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 � = �
1 − 𝑀𝑀 𝑝𝑝 𝑀𝑀 𝑝𝑝 � Equation 4.4
If we assume the entropy change on a spin flip is negligible then statistical mechanics tells
us that
1
3
exp � −
∆ 𝐸𝐸 𝑆𝑆𝑆𝑆
𝑘𝑘 𝐵𝐵 𝑇𝑇 �= �
1 − 𝑀𝑀 𝑝𝑝 𝑀𝑀 𝑝𝑝 � . If the relative fluorescence component amplitudes can be
captured accurately, then the singlet-triplet energy gap is accessible to us without temperature
dependent measurements. The emission decay curve was fit to a bi-exponential decay to obtain the
amplitudes and lifetimes of the prompt and delayed fluorescence. The prompt and delayed
fluorescence decay constants and amplitudes fitted to TCSPC for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , 𝐴𝐴𝐴𝐴
𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 ,
𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 and 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 in both toluene and THF are tabulated in Table 4.1 (All data and fits are

Figure 4.7 – Kinetic model of the cMa compounds. Rates: 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑓𝑓 – solvent and intramolecular
vibrational relaxation to 𝑆𝑆 1
minimum, 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒 𝑒𝑒 𝑟𝑟 – exothermic intersystem crossing from 𝑆𝑆 1
to 𝑇𝑇 1
,
𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 – endothermic intersystem crossing from 𝑇𝑇 1
to 𝑆𝑆 1
, 𝑘𝑘 𝑟𝑟 𝑆𝑆 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 𝑆𝑆 / 𝑘𝑘 𝑟𝑟 𝑇𝑇 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 𝑇𝑇 – the sum of
radiative and non-radiative relaxation rates from 𝑆𝑆 1
/ 𝑇𝑇 1
. Additionally, 𝜏𝜏 𝑇𝑇 𝑀𝑀𝐷𝐷 𝑇𝑇 =
1
𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 � .
.
141

shown in section 4.7.2). ISC rates and the ∆ 𝐸𝐸 𝑆𝑆 𝑇𝑇 are also shown, having been determined by using
equations above.
Unlike in our earlier work
8
where we imputed the reverse intersystem crossing rate, here
we determine it directly from �
1 − 𝑀𝑀 𝑝𝑝 𝑀𝑀 𝑝𝑝 � . We confirm 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 ~ 10
7
– 10
8
s
-1
. Comparable 𝜏𝜏 𝑝𝑝 and Αp,
within error limits, in both THF and toluene suggests ISC is independent of the nature of the
solvent. The fact that
𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 � , as reflected by the relative amplitude of long-time
fluorescence component, is approximately constant, even though 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑒𝑒𝑒𝑒 𝑟𝑟 varies by over an order of
magnitude in the compounds studied, reflects that there is only a small variation in ∆ 𝐸𝐸 𝑆𝑆 𝑇𝑇 , all about
80-90 meV for Cu compounds, a value consistent with temperature dependent studies reported
earlier.
8,25
Of course, for a given compound, the same spin-orbit coupling constant mediates both
forward and backward processes. On the other hand, we see a smaller but now solvent-dependent
variations in the 𝜏𝜏 𝑇𝑇 𝑀𝑀𝐷𝐷 𝑇𝑇 , all close to 1 µs (except Ag cMa compounds),
8
suggesting radiative and
non-radiative relaxation of the singlet are sensitive, as expected, to different factors than the ISC
processes.
Looking now at the specific effects of ligand identity, for the series of MAC compounds,
the choice of carbazole has a small effect on ISC rate. Regardless of R group choice on the
carbazole in the 𝐴𝐴 𝐴𝐴 𝑀𝑀𝑀𝑀 𝐵𝐵 systems, the ISC rates change is less than 40%. Changing the carbene
however has a more substantial effect on ISC. For example, changing MAC to CAAC in the 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵

systems, the ISC rate speeds four-fold from 4.5·10
9
s
-1
to 20.2·10
9
s
-1
. Unfortunately, a meaningful
comparison of 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 is hard to make since the identity of both ligands are changing.
The insensitivity to the carbazole and the effect of the carbene on ISC makes sense when
we consider the unusual electronic structure of a singlet carbene. The largest contributor to rates
142

of ISC should stem from spin orbit coupling (SOC), which will change with the prominence of the
empty p-orbital on the carbene as well as a function of metal identity.
8
We can see in the data
(Figures S4.4 - S4.9) the dramatic effect on substituting Au for Cu with the same ligands; all three
Au compounds show instrument limited τp, meaning that the ISC rate is accelerated by at least a
factor of 10 with MAC and 2.5 with CAAC. With the instrument limit on 𝜏𝜏 𝑝𝑝 of 22 ps we cannot
directly determine either ISC rate constant directly. To determine the Au cMa ISC rates, we
therefore implemented a higher time resolution technique, psTA with a ~350 fs time resolution
applying to both the Au and Cu compounds for a consistent treatment. The psTA and TCSPC
experiments are fit with the same fast equilibrium kinetic model, with the addition of a solvent
relaxation captured by the psTA experiment but too fast for the TCSPC as will be discussed below.
So that they can be combined with the amplitudes in the TCSPC to complete the analysis for ∆ 𝐸𝐸 𝑆𝑆 𝑇𝑇 ,
the psTA derived 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 for the gold compounds are included in Table 4.1 (subscript d).
143

Table 4.1 – Summary of measured TCSPC fluorescence decays in toluene
and derived ISC rates and singlet-triplet energy gap
𝐴𝐴 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟 𝑒𝑒 𝑐𝑐 𝑎𝑎 𝑟𝑟𝑃𝑃 𝑒𝑒𝑛𝑛𝑒𝑒 𝐴𝐴 𝑝𝑝
(%) 𝐴𝐴 𝑇𝑇 𝑀𝑀𝐷𝐷 𝑇𝑇
(%) 𝜏𝜏 𝑝𝑝 (ps)
𝜏𝜏 𝑇𝑇 𝑀𝑀𝐷𝐷 𝑇𝑇
( 𝜇𝜇 s)
𝑘𝑘 𝑀𝑀 𝑏𝑏𝑓𝑓

(10
12
s
-1
)
𝑘𝑘 𝐼𝐼 𝑆𝑆𝐵𝐵
𝑒𝑒 𝑒𝑒 𝑒𝑒 𝑟𝑟
(10
9
s
-1
)
𝑘𝑘 𝐼𝐼 𝑆𝑆𝐵𝐵
𝑒𝑒 𝑛𝑛 𝑟𝑟 𝑒𝑒 𝑟𝑟
(10
9
s
-1
)
∆ 𝐸𝐸 𝑆𝑆 𝑇𝑇
b

(meV)
𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵
𝑀𝑀𝑀𝑀 𝐵𝐵 d

98.69
0.07
1.31
0.07

220
8
a

1.32
0.02
0.20
0.2
4.5
0.2
0.060
0.005
83
3

𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵
98.56
0.07
a

98.70
0.05
a

1.44
0.07
a

1.30
0.05
a

300
10
a

350
10
a

0.73
0.02
a

0.345
0.008
a

0.20
0.04

g

3.3
0.1
a

2.8
0.1
a

0.050
0.004
c

0.040
0.003
a,c

80
2
a

83
4
a

𝐴𝐴 𝐴𝐴 𝑃𝑃 h 𝐵𝐵𝐵𝐵
𝑀𝑀𝑀𝑀 𝐵𝐵
98.1
0.1
a

98.1
0.1
a

1.9
0.1
a

1.9
0.1
a

280
10
a

340
10
a

0.727
0.003
a

0.45
0.02
a

0.20
0.04

g

3.5
0.1
a

2.8
0.3
a

0.069
0.007
a

0.058
0.005
a,c

73
3
a

73
2
a

𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵
𝐵𝐵 𝑀𝑀𝑀𝑀𝐵𝐵 e

99.19
0.04
0.81
0.04
49
2

2.0
0.2
a

0.2
0.1
20.2
0.8
0.162
0.006
96
1

𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵𝐵𝐵
𝐷𝐷 𝑀𝑀𝐵𝐵

98.28
0.09
a
99.03
0.03
a

1.72
0.09
a
0.97
0.03
a

127
5
a

141
3
a

0.68
0.01
a

0.048
0.002
a

0.20
0.04

g

7.7
0.3
a,c

7.0
0.2
a,c

0.140
0.005
a,c

0.090
0.04
a,c

75
1
a

84
1
a

A 𝐴𝐴 𝐵𝐵𝐵𝐵
𝑀𝑀𝑀𝑀 𝐵𝐵 d

99.62
0.03
0.38
0.03

8
4
d
0.935
0.005
a

0.3
0.1
120
40
0.5
0.2
110
20

A 𝐴𝐴 B 𝐵𝐵𝐵𝐵
𝑀𝑀𝑀𝑀 𝐵𝐵
99.52
0.05
a

99.59
0.05
a

0.48
0.05
a

0.41
0.05
a

12
1
d

12
1
a,d

0.462
0.004
a

0.248
0.003
a

0.3
0.1
a

0.8
0.1
a

83
6
a

77
6
a

0.40
0.07
a

0.35
0.06
a

109
6
a

113
5
a

A 𝐴𝐴 𝐵𝐵𝐵𝐵
𝐵𝐵 𝑀𝑀𝑀𝑀𝐵𝐵

99.45
0.03
a

f

0.55
0.03

f

9
3
d

11
2

a,d

1.15
0.05

0.83
0.08
a,e

0.5
0.1
a

0.8
0.3
a

110
30
a

90
20
a

0.6
0.2

f

105
10

f

Uncertainty values are displayed in subscript as ± Δ 𝑘𝑘 , i.e. 2208 = 220 ± 8 ps
a
Solvent is THF
b
Calculated from − 𝑘𝑘 𝐵𝐵 𝑇𝑇 𝑓𝑓 𝑀𝑀 � 3 ∙
𝑘𝑘 𝐼𝐼 𝑆𝑆𝐵𝐵
𝑒𝑒 𝑛𝑛 𝑟𝑟 𝑒𝑒 𝑟𝑟 𝑘𝑘 𝐼𝐼 𝑆𝑆𝐵𝐵
𝑒𝑒𝑒𝑒 𝑒𝑒 𝑟𝑟 � � (see text).
c
𝑘𝑘 𝐼𝐼 𝑆𝑆𝐵𝐵
𝑒𝑒𝑒𝑒 𝑒𝑒 𝑟𝑟 or 𝑘𝑘 𝐼𝐼 𝑆𝑆𝐵𝐵
𝑒𝑒 𝑛𝑛 𝑟𝑟 𝑒𝑒 𝑟𝑟 were calculated for these cases with short 𝜏𝜏 𝑇𝑇 𝑀𝑀𝐷𝐷 𝑇𝑇 values using the full solution by Adachi et
al.
21

d
TCSPC instrument limited; ISC rate values and recovered ∆ 𝐸𝐸 𝑆𝑆 𝑇𝑇 from psTA
e
Data in 2-MeTHF presented in ref. 8.
f
A TCSPC experiment in THF is unavailable, therefore these values are undetermined.
g
A psTA experiment in THF is unavailable, therefore these values are undetermined.

144

4.4.4. Spectrally Resolving ISC with psTA
With the basis spectra signatures for cMa radical anion and radical cation obtained from
SEC, we now look more carefully at the spectroscopy and early time dynamics for the cMa
compounds using psTA. We present the psTA data for cMa compounds beginning with the
example of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in Figure 4.8. The psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 were collected in both toluene and
THF with excitation at 405 nm, corresponding to the ICT absorption band of the cMa complexes;
the full dataset (rendered as a contour plots) can be found in the SI (Figures S4.17 - S4.27). The
psTA spectrum has three prominent features. A GSB is assigned based on the good match with the
inverted ground state absorption spectrum from 400 to 450 nm for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (dashed blue). Second,
a dip in the ESA spanning ~500 to 620 nm is assigned to SE because of a good match in shape
with the inverted steady state photoluminescence (PL) emission spectrum (dashed red). The final
assignment is the overlying ESA spectrum, spanning the entire probe window. The largest and
sharpest feature, at earliest delay times (black trace), in the ESA is at 690 nm with a shoulder at
620 nm with 60% intensity, which bears a strong resemblance to the cation feature in the bulk
electrolysis spectrum (Figure 4.5) but is blue shifted ~30 nm. The absorption intensity of the ESA
outcompetes the absorption intensity of the SE at all probe wavelengths and delays for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in
both solvents.

145

Having assigned the psTA features, we can analyze the kinetics. Upon light absorption, the
photoexcited state formed is a singlet. In toluene, the GSB of 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−
does not shift or lose
intensity, indicating the ground state population does not recover much over the full 1.5 ns
timescale of the measurement. This is expected as this compound has an overall excited state
lifetime of 720 ns in toluene. Over the course of the first 4 ps (black to orange traces), the structured
ESA of the singlet state at 690 nm redshifts by 10 nm (Figure 4.8, arrow 1). Then, from 5 to 100
ps (orange to blue traces), the initial structured ESA is replaced by a new structured ESA band
with ~2x broader features and with a new center at 650 nm (arrow 2) to the blue of the initial peak.
These changes happen on a timescale comparable to the evolution of the SE feature: this is first
characterized by the rapid loss of prominence along with a 20-25 nm redshift, on the order of 4 ps.
After this, the SE band vanishes, almost completely absent at 10 ps while the ESA reaches its
maximum overall intensity at 100 ps (blue trace; arrow 3); subsequent to this the ESA does not
evolve any further. We can now understand that the ESA broadening seen at 100 ps on the bluer

Figure 4.8 – The psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene (a) and THF (b) with 405 nm pump. The
temporal evolution is indicated by rainbow color coded spectral traces. The inverted steady
state absorption (blue dash) and emission spectra (red dash) are displayed. The regions of 400
and 800 nm are removed due to large pump scatter. The major spectral evolutions are depicted
with arrows: Initial ultrafast ESA redshift (1), subsequent ESA blueshift (2), and simultaneous
SE recovery (3).

400 500 600 700 800 900
-10
-5
0
5
10
Wavelength (nm)
∆Abs (mOD)
0.3 ps 10 ps
1 ps 100 ps
4 ps 1500 ps
Absorption Emission
(a)
1
2
3

400 500 600 700 800 900
-6
-4
-2
0
2
4
6
∆Abs (mOD)
Wavelength (nm)
0.3 ps 10 ps
1 ps 100 ps
4 ps 1500 ps
Absorption Emission
(b)
3
1
2
146

side can in part be attributed to loss of the SE. Recall, the time-resolved fluorescence, reporting
just on the singlet population, decays on the same timescale as the loss of the SE and the loss of
the sharper ESA and the simultaneous changeover to the blue ESA band (Figure S4.20). The state
formed after SE loss with its major peak at 650 nm is therefore assigned to the triplet and the rise
of this band is identified with 𝑆𝑆 1
→ 𝑇𝑇 1
conversion. Interestingly, aside from the SE and a modest
blueshift, the singlet and triplet bands bear a qualitative resemblance to one another.
The psTA experiment in THF qualitatively resembles the situation in toluene. However,
the red shift in SE is faster and more extensive over the first 1-4 ps (see also Figure S4.21); the
larger dynamic Stokes shift mirrors the change in steady state emission spectra (dashed red line)
when changing from toluene to polar THF. This earliest phase corresponds to solvent
rearrangement around a very different charge distribution, dependent on the solvent polarity. On
the other hand, we find by target analysis fitting (see section 4.7.3.4) that the rate of conversion of
𝑆𝑆 1
to 𝑇𝑇 1
is the same in both toluene and THF, consistent with the weak dependence on solvent seen
for the copper cMa revealed directly by fluorescence lifetime. We can readily determine the
forward ISC rate for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in both solvents as 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 = 80 ∙ 10
9
𝑠𝑠 − 1
(Table 4.1), a result consistent
with hitting the 22 ps instrument limit for the TCSPC measurement.
Previous temperature dependent TCSPC studies indicate that 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 ≫ 𝑘𝑘 𝑟𝑟 𝑇𝑇 and 𝑘𝑘 𝑛𝑛 𝑟𝑟 𝑇𝑇 .
Depopulation of the triplet is assumed to be solely from 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 and therefore, the emission is
approximated to only occur from 𝑆𝑆 1
. This special case is possible due to the small 𝛥𝛥 𝐸𝐸 𝑆𝑆 𝑇𝑇 where
thermal energy allows for intersystem crossing from the longer-lived triplet population pool giving
rise to further delayed emission from the singlet. Due to the small energy gap between the singlet
and triplet states, depletion of the singlet occurs but is not complete. While psTA is effective at
determining 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 due to the large contribution of 𝑇𝑇 1
to the excited state spectrum, 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 is
147

indeterminate in psTA due this technique’s more limited dynamic range and the only ~1-2%
equilibrium 𝑆𝑆 1
population. Complementarily, TCSPC only reports on the singlet population with
high dynamic range but is blind to the triplet population which is fully revealed in the psTA.
Therefore, if possible, values of 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 and 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 were taken from the TCSPC
measurements and treated as fixed. Because the fastest decay for the gold complexes is below the
time response of the TCSPC apparatus, the values for 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 for these complexes were fit from
psTA data such that 𝜏𝜏 𝑝𝑝 = ( 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑒𝑒𝑒𝑒 𝑟𝑟 )
− 1
and this parameter fixed. Then from a fit to the TCSPC data
of the gold compounds that includes convolution with the measured IRF, the 𝐴𝐴 𝑝𝑝 and � 1 − 𝐴𝐴 𝑝𝑝 �
were extracted allowing 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛 𝑟𝑟𝑜𝑜
and Δ 𝐸𝐸 𝑆𝑆 𝑇𝑇 to be subsequently obtained. For all the Au containing
cMa compounds, this approach is used to determine 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 , estimate 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 and thus Δ 𝐸𝐸 𝑆𝑆 𝑇𝑇 to
complete Table 4.1. Through this analysis, we find that 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 has Δ 𝐸𝐸 𝑆𝑆 𝑇𝑇 ~30-40% higher than
the analogous copper complex, 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 .
We are now ready to move on to the psTA spectroscopy when substituting Cu for Au and
explore the CAAC carbene ligand for both metals, (Figure 4.9) and cMa compounds where the
carbazole ligand is substituted (Section 4.7.3.1). For all four 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 datasets in toluene
(Figure 4.9), there are strong qualitative similarities to the discussion above for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and all
datasets can be fit with the same model. Namely, for each data set there is a distinct negative GSB
at the blue end, with the remaining spectrum dominated by a sharper absorption in the region of
600-725 nm for the ESA overlapped by a time-evolving SE band to its blue side, which captures
first the ultrafast solvation of the singlet and then its demise with a parallel time evolution for the
148

red end of the ESA. Additional analysis of the spectral changes follows in the next section.
However, because of the broad similarities in kinetics and spectral detail, the spectral assignments
for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 are applied to the remaining cMa compounds. A quantitative analysis is applied to the
full 2D spectral-temporal data set of each compound using a target analysis with the kinetic model
described in the previous TCSPC section and Figure 4.7. Fitting details are provided in section
4.7.3.4. The same set of parameters are used simultaneously to fit psTA data and TCSPC for all
compounds, supporting the assignment of the ISC rate constants. A full summary of kinetic

Wavelength (nm)
Figure 4.9 – The psTA spectra of 𝐴𝐴 𝐵𝐵𝐵𝐵 class of cMa compounds with excitation at 405 nm in
toluene with Cz kept constant as the amide ligand. The gold and copper complexes are the top
and bottom rows. 𝐴𝐴 𝑀𝑀𝑀𝑀 𝐵𝐵 and 𝐴𝐴 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 compounds are the left and right columns, respectively.

-4
0
4
8
0.3 ps 7 ps
1 ps 100 ps
3 ps 1500 ps
Absorption Emission
Au
MAC
Cz

-10
0
10
20
0.3 ps 7 ps
1 ps 100 ps
3 ps 1500 ps
Absorption Emission
Au
CAAC
Cz

400 500 600 700 800 900
-2
-1
0
1
2
0.3 ps 100 ps
1 ps 1000 ps
10 ps 1500 ps
Absorption Emission
Cu
MAC
Cz

400 500 600 700 800 900
-4
-2
0
2
4
0.3 ps 50 ps
1 ps 1000 ps
10 ps 1500 ps
Absorption Emission
Cu
CAAC
Cz
ΔAbs (mOD)
149

parameters resulting from this target analysis fitting for all cMa compounds measured is organized
in Table 4.1.
To briefly summarize an extensive dataset, the initial singlet population experiences
solvent configurational relaxation labeled as 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑓𝑓 where 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑓𝑓 ~ (0.2 − 0.8) ∙ 10
1 2
𝑠𝑠 − 1
for all cMa
compounds. Meech et al
26
have published a recent study of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in a more polar solvent,
chlorobenzene. They assign the early time to solvation, however there are also contributions from
vibrational relaxation, namely along the central cMa bond, determined from femtosecond
stimulated Raman spectroscopy. Fitting reveals 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑓𝑓 in THF is a factor of 2 faster than in toluene
but independent of carbene and amide identity. After initial stabilization, the cMa exists in a 𝑆𝑆 1
⇄
𝑇𝑇 1
dynamic equilibrium for the lifetime of the excited state. We confirm the large impact on spin-
orbit coupling (SOC) effect with a change in metal; psTA studies reveal that a gold atom leads to
a ~25x fold increase in 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 due to its higher SOC strength compared to copper. The psTA
experiments allow us to reveal for gold the same behavior already noted for copper. Namely,
for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 , 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 is moderately independent of solvent or the identity of the amide
with 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 experiencing a ~1.2 increase compared to the other gold compounds. And our
analysis suggests that all the gold containing complexes have somewhat higher Δ 𝐸𝐸 𝑆𝑆 𝑇𝑇 than the
copper complexes, consistent with theoretical calculations for the gap by Muniz, et al.
25

4.4.5. Excited State Spectra Recreation
Considering the spectral variation exhibited in Figure 4.9, the largest peaks at 700 nm in
the singlet and 650 nm in the triplet for 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 are sharper than for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 . The valley at early
times carved out by the stimulated emission in the gold cMa complexes is also further enhanced,
evidenced by the shoulder at 470 nm in 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 rising sharper than 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 .
150

To more fully analyze the character of the excited state, we have compared the excited state
spectra (originating from both singlet and triplet 𝐴𝐴 ( 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒 )
+
( 𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒 𝑛𝑛𝑒𝑒 )
−
) for several of the complexes to the
sums of 𝐴𝐴 ( 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒 )
+
𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 and 𝐴𝐴 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒
( 𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 )
−
spectra measured from SEC (Figure 4.10, Section 4.7.6).
19

The two features present in the simulated 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 (Figure 4.10a) spectrum at ~ 1 eV and ~ 1.8
eV, arising from 𝐴𝐴 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒
( 𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 )
−
and 𝐴𝐴 ( 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒 )
+
𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 respectively (from PR and BE SEC) also appear in
the measured 𝑇𝑇 1
spectra for this compound, but a constant 0.3 eV blue shift needs to be applied to
align the peak positions. However, only the higher energy (cation) feature of the simulated spectra
seems to be present in the 𝑆𝑆 1
ESA and is better energy-aligned without the need for a blue shift.
Similarly, the simulated spectra for the rest of the copper cMa complexes (Figure S4.60 - S4.61)
are largely comprised of the carbazolium feature between 1.5-1.7 eV and match well to the
respective 𝑆𝑆 1
and 𝑇𝑇 1
ESA which are dominated by a feature between 1.6-1.9 eV. Despite the choice
of carbazole varying the oxidation potential of the cMa widely, its appearance in the ESA is
qualitatively similar. In general, the width of this feature for the singlet and the cation/anion
composite is narrower, and the triplet somewhat broader. In the case of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , the
relative intensity of the absorption peaks and the peak vibronic structure seem well matched
between the simulated and real spectra while this is not the case for the other compounds.
151

The ESA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 can also be reconstructed by summing the cation and anion,
but this time from BE spectra (Figure 4.10b). The singlet and triplet are represented by the species
associated decay spectra (SADS) obtained from fitting the psTA and nsTA data in THF
respectively. The summation of these BE basis spectra is similarly effective in reconstructing the
ESA as the use of PR spectra was for the copper complexes, however, PR data and therefore
spectra on an absolute molar absorptivity scale for gold cMa basis spectra is not currently available.
The spectral sum was terminated > 3.3 eV due to high spectral uncertainty from saturation for the
unoxidized (unreduced) 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in the cation (anion) spectra. The emission and absorption spectra
are included to indicate the strongest deviations from the sum and the singlet/triplet spectra are
due to the GS and SE bleaches. The same procedure with BE was carried out for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and
reveals a similar trend (Figure S4.62).
The most obvious feature in common from these two different analyses is the transition
energy mismatch when comparing the reconstructed anion/cation summed spectra and the triplet.

Figure 4.10 – Simulation of the ESA by summation of cation and anion basis spectra (black)
of (a) C 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 using PR data and (b) 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 using BE data with comparisons made to the triplet
(red) and singlet (blue) SADS from psTA. The PR triplet data for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 was used in place of
the triplet spectra. The inverted absorption (dashed purple) and emission spectra (dashed
orange) are displayed.

1 1.5 2 2.5 3
-1
0
1
2
3
4
5
6
Energy (eV)
Cat + An Emission
S
1
(norm) T
1
(norm)
Molar Absorptivity (10
3
M
-1
cm
-1
)
Cu
CAAC
Cz
(a)
500 600 800 1100 1600
Wavelength (nm)

1.5 2 2.5 3 3.5
-1
0
1
2
Absorption (norm.)
Energy (eV)
Cat + An
S
1
(norm) T
1
(norm)
Absorption Emission
(b)
Au
MAC
BCz
900 700 600 500 400
Wavelength (nm)
152

With the exception of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (Figure S4.60, left) where there seems to be good alignment, the
peaks originating from both the anion and cation components of the 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 simulated spectra are
~300 meV lower in energy then their 𝑇𝑇 1
ESA counterparts, while the carbazolium features in the
ESA of the MAC family are somewhere between 0-200 meV. The need for an energy shift is not
observed in the treatment reported by McCusker et al. for octahedral, tris-bisimine metal
complexes.
19
This energy mismatch could potentially be due to a larger Coulombic interaction in
the cMa ICT state, or different interactions in the higher lying 𝑆𝑆 𝑛𝑛 and 𝑇𝑇 𝑛𝑛 states to which the ESA
transitions originating in the 𝑆𝑆 1
and 𝑇𝑇 1
states connect.
PR is a useful technique for determining absolute molar absorptivities of cations and
anions. However, PR is not a readily accessible technique to most synthetic-photophysical groups,
requiring one-of-a-kind radiolysis facility and expert staff to take measurements. BE, on the other
hand, trades absolute absorptivities for ease of use and higher availability. Here, we provided
evidence that both SEC techniques provide the ability to deconstruct and simulate ICT excited
state spectra. With a fuller understanding of the intramolecular spectra and dynamics, we shift
focus to intermolecular charge transfer dynamics of the cMa with an electron acceptor or donor.
4.4.6. Observing Intermolecular CT via nsTA
The psTA experiments provide an excellent picture of the intramolecular charge transfer
dynamics of the excited state, but the 1.5 ns time limitation of the psTA restricts our ability to
monitor diffusion controlled intermolecular charge transfer reactions. Diffusion of the
photosensitizer and electron or hole acceptor close enough for charge transfer to take place is on
the order of nanoseconds for moderate concentrations (1-50 mM). To demonstrate this, psTA was
used to follow the reaction of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with the oxidizing agent MePI to form 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene.
However, the psTA instrumental time range limitation of ~1 ns captures only the onset of the
153

intermolecular electron transfer in this case, complicated by concomitant ISC (4.7.3.2.1). Hole
transfer from 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 to BIH to form 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 ( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−
bears a similar story (4.7.3.2.2). Therefore, we
have used nsTA, pumped by a nanosecond laser, to capture long-time diffusive charge transfer
from cMas to both an electron and hole acceptor.
For all 5 cMa complexes in THF and toluene, the 355 nm pumped nsTA spectra display no
noticeable changes from 10 ns to 2 µs, when the traces return to baseline (Figure 4.11a-b, Figures
S4.35 – S4.42). However, during the experimental run time, we find that in THF solution both
𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀 𝐵𝐵 and 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 decompose while 𝐴𝐴𝐴𝐴
𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 decomposes on irradiation in both solvents. On the
other hand, degradation is not observed when these complexes are excited with either the 400 or
405 nm sources described earlier, nor during the collection of photophysical data reported
previously.
7–9
We therefore do not suspect inherent instability in the 𝑆𝑆 1
and 𝑇𝑇 1
ICT states, but
rather because 355 nm leads to excitation into a higher lying ligand field state which may be
unstable. While we are currently limited to 355 nm laser as our excitation source, future work will
substitute direct excitation into the ICT state for these nsTA studies.
154

Figure 4.11 – 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 nsTA spectra as neat solutions (top row), with MePI as an electron
acceptor (middle row), and BIH as a hole acceptor (bottom row) in THF (left column) and
toluene (right column, except (f) which is in THF). Plot (f) is generated from the experiment in
(e) after averaging together the long time spectra (>100 ns) for the spectra with (black) and
without BIH (red). All spectra were collected on a Magnitude enVISion except for (c) which
was performed on an enVISTA (see Experimental). All data was pumped with 355 nm except
for (e) and (f) which were pumped with 450 nm.

0
1
2
10 ns 200 ns
50 ns 300 ns
100 ns 500 ns
Absorption
(a)

-2
-1
0
1
2
3
4
5
10 ns 500 ns
100 ns 1000 ns
250 ns 2000 ns
Absorption
(b)

-1
0
1
2
3
10 ns 100 ns
20 ns 250 ns
50 ns 500 ns
Absorption BE Cation
(c)

-1
0
1
2
3
4 ns 100 ns
20 ns 150 ns
50 ns 300 ns
Absorption BE Cation
(d)

400 500 600 700 800 90
-4
0
4
8
10 ns 80 ns
20 ns 120 ns
50 ns 200 ns
Absorption
(e)

00 500 600 700 800 900
-2
0
2
4
x10
x40
Long time With BIH
Long time No BIH
BE Anion
in THF
(f)

Wavelength (nm)
ΔAbs (mOD)
No Quencher
With MePI With BIH
Toluene TH

155

In order to study charge transfer dynamics from PS* to an electron acceptor or donor, the
excited states of various cMa complexes were used to reduce N-methylphthalimide, MePI, or
oxidize N-dimethyl-dihydrophenylbenzimidazole, BIH. To avoid photo-degradation issues, we
chose to use gold based cMa, i.e., 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , which is stable to 355 nm excitation. The addition of
7 mM MePI to 46 µM 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF results in a decreased lifetime of 240 ns, attributed to the
transfer of either charge or energy to MePI. However, despite the decreased lifetime, no new
features are observed in the spectra (Figure S4.43). Recording the ground state absorption of the
sample cuvette pre- and post-experiment reveals a large amount of sample degradation post-
quenching. This degradation is unsurprising as the cyclic voltammogram (CV) trace reveals a
non-reversible oxidation wave for the compound,
9
leading to a lifetime too short for the cation to
recombine with MePI
-
to form the stable neutral state. As the oxidative non-reversibility in 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵
is attributed to polymerization of the unsubstituted 3-, and 6- positions in the carbazole ligand,
consistent with irreversible oxidation of bare Cz,
9
substitutions at these locations with tert-butyl
groups in 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 should impart reversibility, which is observed in its CV.
25,27
Still some
degradation is observed upon adding MePI to the solution. We believe this instability is due again
to excitation of a ligand field transition, this time of the cation, upon irradiation with 355 nm light.
To circumvent the problems associated with 355 nm excitation in the nsTA system, a flow
cell apparatus with a reservoir volume of 100 mL was used to mitigate the slow degradation of the
complex. Upon the addition of 30 mM MePI to 75 µM 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF, a new feature matching
the cation peak is seen at ~720 nm which persists for ~110 µs (Figures S4.44 - S4.45). The PL is
extinguished with a 50 ns lifetime, indicating charge transfer with no recombination to the excited
state (Figure S4.45a). The 110 µs lifetime is more than sufficient for a sacrificial reductant to
regenerate the neutral 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 . Reducing the concentration of MePI to 6 mM in THF allows us to
156

clearly observe signal from the triplet alone at early time slices which slowly evolves into the
cation spectra over ~250 ns (Figure 4.11c). The quenching rate constant, kq, of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with MePI
of (1.4 ± 0.1) · 10
10
M
-1
s
-1
is in agreement with that obtained previously.
25

Unlike electron transfer to MePI, using 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

to transfer a hole to 100 mM BIH in
THF (i.e. oxidize BIH) results in no degradation of the cMa over the course of the ns TA
experiment, indicating that 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 ( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

is markedly more stable to 355 nm irradiation. However,
BIH absorbs 355 nm radiation (Figure S4.15) and generates the BIH triplet (Figure S4.46);
therefore, the BIH experiments were pumped with 450 nm. A nsTA experiment of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with
100 mM BIH in THF results in excited state quenching and reduced lifetime (Figure 4.11e, 240
to 28 ns). We report the quenching rate of 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

with BIH to be (3.2 ± 0.9) ·10
8
M
-1
cm
-1
(Figure S4.63, left) with 450 nm excitation. Independently, a kq of (3.6 ± 0.1) ∙ 10
8
M
-1
s
-1
has
been established from a Stern-Volmer analysis using TCSPC (Figure S4.63, right) with 405 nm
excitation, using the same methodology as Muniz et al.
25
In the nsTA, at concentrations of 100
mM, we expect formation of the anion around ~30 ns. However, no appreciable formation of
𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 ( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

is seen at 30 ns or longer times. If the average at long delay times of the spectra with
BIH (100 ns to 13 µs) and without BIH (100 ns to 700 ns) is taken, a small but distinctly positive
signal reveals itself (Figure 4.11f). Due to the low SNR and the similarity of the spectra of
𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 ( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

, it is difficult to assign hole transfer to BIH from 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 . Furthermore,
from PR, the molar absorptivity of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 ( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−
is known and should not be ~100x weaker for
𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 ( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

. Further work was performed with other hole acceptors and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵
but a spectral
signature of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 ( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

was not isolated (Figure S4.47). Additionally, triplet transfer from
𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

to BIH is also possible as the remaining spectrum at long times bears resemblance to
157

BIH triplet. A third hypothesis is that the carbene donates charge to the BIH but the protons on
BIH form hydrogen with the donated electron from the cMa, leading to no steady state production
of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 ( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

. Whether the anion is not formed due to triplet transfer, residual absorption to BIH,
or formation of hydrogen, the same conclusion follows: no large amount of anion was isolated as
was for the case with the cation.
Electron transfer from 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

to MePI was reexamined in toluene (Figure 4.11d).
Using 280 mM MePI, results in the formation of 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
𝑀𝑀𝑀𝑀 𝐵𝐵
at the earliest experimental time delays
resolvable. Throughout the course of the experiment, the ratio between the triplet and cation peaks
remains unchanged and all kinetic traces have a lifetime of 120 ns, less than the lifetime of the
complex alone. Photoluminescence is also detected until the TA spectra returns to baseline. This
result suggests that while an electron is transferred from 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 to MePI in toluene, as expected,
the ion pair of 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
𝑀𝑀𝑀𝑀 𝐵𝐵 and MePI
-
is not able to escape from the nonpolar solvent cage. Inside the
cage, the charges then collide many times, recombining to either the ground or excited state of
𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 . Attempts to repeat this experiment at a lower concentration of 6 mM MePI resulted in no
cation peaks being observed, mostly likely attributed to the increased driving force required to
transfer charge in non-polar solvents.
5

4.5. Conclusion
The cMa class of complexes presented here are strong candidates as photosensitizers in the
production of energy dense solar fuels from a common feedstock such as hydrogen from water,
due to their strong, tunable absorptions, large excited state redox potentials, long lifetimes, and
robust electrochemical stability. The long lifetimes have been achieved via a small Δ 𝐸𝐸 𝑆𝑆 𝑇𝑇 , allowing
for rapid ISC into the 𝑇𝑇 1
state even without the use of a heavy atom, by invoking a TADF regime.
By applying a model laid out by Adachi et al.
23
values of 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 , 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 and Δ 𝐸𝐸 𝑆𝑆 𝑇𝑇 have been
158

determined through fluorescence lifetimes measured by TCSPC, without the need to rely on a
thermal fitting procedure as done in our prior report.
8

Fluorescence lifetime measurements indicated the strongest change in the ISC rates occurs
when changing the metal from Cu to Au, consistent with higher SOC and leading to an increase in
ISC by ~25 times. The identity of the carbene and amide ligands also play a role in moving between
the two states, affecting the 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 by a factor ~4.5 and ~1.3, respectively. We attribute the larger
effect of carbene changes on the ISC rate to the carbene ligand having a greater effect on the SOC
from the metal ion on the ISC rate than the carbazole. Fundamentally, we have demonstrated the
ability to independently have knowledge of the spin-orbit coupling and the Δ 𝐸𝐸 𝑆𝑆 𝑇𝑇 . This is made
possible by our knowledge of both the intersystem crossing rates from singlet to triplet and triplet
to singlet separately. These cMa complexes able to achieve fast rates of ISC and long lifetimes
while utilizing an abundant first row transition metal are promising for low-cost sensitizers for
photo-electrocatalytic processes, such as the production of solar fuels.
The excited state dynamics of the complexes were further probed by psTA, which revealed
a further time regime: a fast component (~1.5-5 ps) contributed to solvent relaxation and
vibrational relaxation of the excited state. The spectral evolution of 𝑆𝑆 1
to 𝑇𝑇 1
ISC was revealed with
dynamics mirroring fluorescence lifetime as well as elucidating ISC rates for the gold atom
complexes otherwise beyond the instrument limit of TCSPC. The psTA triplet spectra matches
well to the long time spectra obtained by PR. By summing together the molar absorption spectra
of the cation and anion from PR, the ESA spectrum of the 𝑆𝑆 1
and 𝑇𝑇 1
states of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 can be
approximated with good matching of the relative intensities of the transition, but with an energy
discrepancy between the simulated and real ESA of 0-400 meV. We recognize that access to PR
facilities is limited; however, BE exists as an equivalently viable technique. As such the ESA of
159

𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 was also simulated by summing the spectra for the cationic and anionic absorptivity spectra
collected via BE, which again was in good agreement with the real ESA.
In the ns regime, TA studies of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 were used to spectroscopically capture the transfer
of either a hole to BIH or an electron to MePI in THF. Here, quenching produces the respective
cation of the 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with a diffusion limited charge transfer rate, with spectral signatures in good
agreement with the ionic spectra collected via SEC. The anion of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 is less definitive; but
quenching does indeed take place. The cation of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 possesses a 110 µs lifetime before either
charge recombination or degradation. This lifetime is more than sufficient for reduction by
separate source to recover the ground state for the photocatalytic cycle to continue.
Moving from THF to toluene results in an equilibrium seen in the earliest time slices,
between ( 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 )* and the solvent caged 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
𝑀𝑀𝑀𝑀 𝐵𝐵 and MePI
-
. The same equilibrium behavior is
seen for 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 . Such a cage mimics covalently linking the PS to an EC, removing the kinetic rate
limit stemming from the two diffusing together during the production of solar fuels.
28
By using
an inert covalent linker, this intermolecular charge separate ion pair can be rapidly realized,
increasing catalytic rates. While degradation was observed in nsTA experiments for some of the
other copper cMa complexes using a high energy excitation source (355 nm), no such degradation
was observed when the same molecules were irradiated by lower energy sources at 400/405 nm in
both TCSPC and psTA, indicating that the PS family presented here will be stable to illumination
by sunlight. Further work to establish these materials as possible PS compounds will investigate
the transfer of charges to electrocatalysts of interest for the reduction of either CO2 or water
splitting. Finally, work is underway on the cMa class of complexes will explore covalently linking
a cMa to a desired electrocatalyst to remove the need for long time diffusion, facilitating the
photocatalytic cycle.
160

4.6. References
(1) Inganäs, O.; Sundström, V. Solar Energy for Electricity and Fuels. Ambio 2016, 45 (Suppl 1),
15–23. https://doi.org/10.1007/s13280-015-0729-6.
(2) Service, R. F. Solar plus Batteries Is Now Cheaper than Fossil Power. Science 2019, 365
(6449), 108–108. https://doi.org/10.1126/science.365.6449.108.
(3) Newman, J.; Hoertz, P. G.; Bonino, C. A.; Trainham, J. A. Review: An Economic
Perspective on Liquid Solar Fuels. J Electrochem Soc 2012, 159 (10), A1722–A1729.
https://doi.org/10.1149/2.046210jes.
(4) Pellow, M. A.; Emmott, C. J. M.; Barnhart, C. J.; Benson, S. M. Hydrogen or Batteries for
Grid Storage? A Net Energy Analysis. Energ Environ Sci 2015, 8 (7), 1938–1952.
https://doi.org/10.1039/c4ee04041d.
(5) Rehm, D.; Weller, A. Kinetik Und Mechanismus Der Elektronübertragung Bei Der
Fluoreszenzlöschung in Acetonitril. Berichte Der Bunsengesellschaft Für Physikalische Chemie
2010, 73 (8‐9), 834–839. https://doi.org/10.1002/bbpc.19690730818.
(6) Hockin, B. M.; Li, C.; Robertson, N.; Zysman-Colman, E. Photoredox Catalysts Based on
Earth-Abundant Metal Complexes. Catal Sci Technol 2019, 9 (4), 889–915.
https://doi.org/10.1039/c8cy02336k.
(7) Hamze, R.; Peltier, J. L.; Sylvinson, D.; Jung, M.; Cardenas, J.; Haiges, R.; Soleilhavoup, M.;
Jazzar, R.; Djurovich, P. I.; Bertrand, G.; Thompson, M. E. Eliminating Nonradiative Decay in
Cu(I) Emitters: >99% Quantum Efficiency and Microsecond Lifetime. Science 2019, 363 (6427),
601–606. https://doi.org/10.1126/science.aav2865.
(8) Hamze, R.; Shi, S.; Kapper, S. C.; Ravinson, D. S. M.; Estergreen, L.; Jung, M.-C.; Tadle, A.
C.; Haiges, R.; Djurovich, P. I.; Peltier, J. L.; Jazzar, R.; Bertrand, G.; Bradforth, S. E.;
Thompson, M. E. “Quick-Silver” from a Systematic Study of Highly Luminescent, Two-
Coordinate, D10 Coinage Metal Complexes. J Am Chem Soc 2019, 141 (21), 8616–8626.
https://doi.org/10.1021/jacs.9b03657.
(9) Shi, S.; Jung, M. C.; Coburn, C.; Tadle, A.; R., D. S. M.; Djurovich, P. I.; Forrest, S. R.;
Thompson, M. E. Highly Efficient Photo- and Electroluminescence from Two-Coordinate Cu(I)
Complexes Featuring Nonconventional N-Heterocyclic Carbenes. J Am Chem Soc 2019, 141 (8),
3576–3588. https://doi.org/10.1021/jacs.8b12397.
(10) Romanov, A. S.; Bochmann, M. Synthesis, Structures and Photoluminescence Properties of
Silver Complexes of Cyclic (Alkyl)(Amino)Carbenes. J Organomet Chem 2017, 847, 114–120.
https://doi.org/10.1016/j.jorganchem.2017.02.045.
161

(11) Feng, J.; Reponen, A. M.; Romanov, A. S.; Linnolahti, M.; Bochmann, M.; Greenham, N.
C.; Penfold, T.; Credgington, D. Influence of Heavy Atom Effect on the Photophysics of
Coinage Metal Carbene‐Metal‐Amide Emitters. Adv. Funct. Mater. 2021, 31 (1), 2005438.
https://doi.org/10.1002/adfm.202005438.
(12) Ravinson, D. S. M.; Thompson, M. E. Thermally Assisted Delayed Fluorescence (TADF):
Fluorescence Delayed Is Fluorescence Denied. Mater Horizons 2020, 7 (5), 1210–1217.
https://doi.org/10.1039/d0mh00276c.
(13) Turro, N. J. Modern Molecular Photochemistry; University Science Books: Sausalito,
California, 1991.
(14) Kellogg, M.; Akil, A.; Ravinson, D. S. M.; Estergreen, L.; Bradforth, S. E.; Thompson, M.
E. Symmetry Breaking Charge Transfer as a Means to Study Electron Transfer with No Driving
Force. Faraday Discuss 2019, 216 (0), 379–394. https://doi.org/10.1039/c8fd00201k.
(15) Kennehan, E. R.; Munson, K. T.; Grieco, C.; Doucette, G. S.; Marshall, A. R.; Beard, M. C.;
Asbury, J. B. Exciton–Phonon Coupling and Carrier Relaxation in PbS Quantum Dots: The Case
of Carboxylate Ligands. J Phys Chem C 2021, 125 (41), 22622–22629.
https://doi.org/10.1021/acs.jpcc.1c05803.
(16) Kennehan, E. R.; Munson, K. T.; Doucette, G. S.; Marshall, A. R.; Beard, M. C.; Asbury, J.
B. Dynamic Ligand Surface Chemistry of Excited PbS Quantum Dots. J Phys Chem Lett 2020,
11 (6), 2291–2297. https://doi.org/10.1021/acs.jpclett.0c00539.
(17) Wishart, J. F.; Cook, A. R.; Miller, J. R. The LEAF Picosecond Pulse Radiolysis Facility at
Brookhaven National Laboratory. Rev Sci Instrum 2004, 75 (11), 4359–4366.
https://doi.org/10.1063/1.1807004.
(18) Oshima, T.; Nishioka, S.; Kikuchi, Y.; Hirai, S.; Yanagisawa, K.; Eguchi, M.; Miseki, Y.;
Yokoi, T.; Yui, T.; Kimoto, K.; Sayama, K.; Ishitani, O.; Mallouk, T. E.; Maeda, K. An Artificial
Z-Scheme Constructed from Dye-Sensitized Metal Oxide Nanosheets for Visible Light-Driven
Overall Water Splitting. J Am Chem Soc 2020, 142 (18), 8412–8420.
https://doi.org/10.1021/jacs.0c02053.
(19) Brown, A. M.; McCusker, C. E.; McCusker, J. K. Spectroelectrochemical Identification of
Charge-Transfer Excited States in Transition Metal-Based Polypyridyl Complexes. Dalton T
2014, 43 (47), 17635–17646. https://doi.org/10.1039/c4dt02849j.
(20) Busby, E.; Xia, J.; Wu, Q.; Low, J. Z.; Song, R.; Miller, J. R.; Zhu, X.-Y.; Campos, L. M.;
Sfeir, M. Y. A Design Strategy for Intramolecular Singlet Fission Mediated by Charge-Transfer
States in Donor–Acceptor Organic Materials. Nat Mater 2015, 14 (4), 426–433.
https://doi.org/10.1038/nmat4175.
(21) Kaafarani, B. R.; Risko, C.; El-Assaad, T. H.; El-Ballouli, A. O.; Marder, S. R.; Barlow, S.
Mixed-Valence Cations of Di(Carbazol-9-Yl) Biphenyl, Tetrahydropyrene, and Pyrene
162

Derivatives. J Phys Chem C 2016, 120 (6), 3156–3166.
https://doi.org/10.1021/acs.jpcc.5b11061.
(22) Chiu, S.; Chung, Y.; Liou, G.; Su, Y. O. Electrochemical and Spectral Characterizations of
9‐Phenylcarbazoles. J Chin Chem Soc-taip 2012, 59 (3), 331–337.
https://doi.org/10.1002/jccs.201100601.
(23) Tsuchiya, Y.; Diesing, S.; Bencheikh, F.; Wada, Y.; Santos, P. L. dos; Kaji, H.; Zysman-
Colman, E.; Samuel, I. D. W.; Adachi, C. Exact Solution of Kinetic Analysis for Thermally
Activated Delayed Fluorescence Materials. J Phys Chem 2021, 125 (36), 8074–8089.
https://doi.org/10.1021/acs.jpca.1c04056.
(24) Haase, N.; Danos, A.; Pflumm, C.; Morherr, A.; Stachelek, P.; Mekic, A.; Brütting, W.;
Monkman, A. P. Kinetic Modeling of Transient Photoluminescence from Thermally Activated
Delayed Fluorescence. J Phys Chem C 2018, 122 (51), 29173–29179.
https://doi.org/10.1021/acs.jpcc.8b11020.
(25) Muniz, C.; Archer, C.; Applebaum, J.; Schaab, J.; Alagaratnam, A.; Djurovich, P.;
Thompson, M. Two-Coordinate Coinage Metal Complexes as Solar Photosensitizers. 2023.
https://doi.org/10.26434/chemrxiv-2023-fkkfz-v2.
(26) Hall, C. R.; Romanov, A. S.; Bochmann, M.; Meech, S. R. Ultrafast Structure and
Dynamics in the Thermally Activated Delayed Fluorescence of a Carbene–Metal–Amide. J Phys
Chem Lett 2018, 9 (19), 5873–5876. https://doi.org/10.1021/acs.jpclett.8b02797.
(27) Muniz, C. N.; Schaab, J.; Razgoniaev, A.; Djurovich, P. I.; Thompson, M. E. π-Extended
Ligands in Two-Coordinate Coinage Metal Complexes. J Am Chem Soc 2022.
https://doi.org/10.1021/jacs.2c06948.
(28) Church, T. L.; Getzler, Y. D. Y. L.; Coates, G. W. The Mechanism of Epoxide
Carbonylation by [Lewis Acid]+[Co(CO)4]- Catalysts. J Am Chem Soc 2006, 128 (31), 10125–
10133. https://doi.org/10.1021/ja061503t.
(29) Ekvall, K.; Meulen, P. van der; Dhollande, C.; Berg, L.-E.; Pommeret, S.; Naskrecki, R.;
Mialocq, J.-C. Cross Phase Modulation Artifact in Liquid Phase Transient Absorption
Spectroscopy. J Appl Phys 2000, 87 (5), 2340–2352. https://doi.org/10.1063/1.372185.
(30) Lorenc, M.; Ziolek, M.; Naskrecki, R.; Karolczak, J.; Kubicki, J.; Maciejewski, A. Artifacts
in Femtosecond Transient Absorption Spectroscopy. Appl Phys B 2002, 74 (1), 19–27.
https://doi.org/10.1007/s003400100750.
(31) Chen, X.; Larsen, D. S.; Bradforth, S. E.; Stokkum, I. H. M. van. Broadband Spectral
Probing Revealing Ultrafast Photochemical Branching after Ultraviolet Excitation of the
Aqueous Phenolate Anion. J Phys Chem 2011, 115 (16), 3807–3819.
https://doi.org/10.1021/jp107935f.
163

4.7. Appendix
4.7.1. Absorptivity Spectra of cMa ions from Pulse Radiolysis
Austin Mencke and Matt Bird collected the PR data in this section at Linear Electron
Accelerator Facility (LEAF) at Brookhaven Nation Laboratory (BNL).

Figure S4.1 – Cation molar absorptivity spectra of 10 mM 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in 20 mM solution of
triphenylamine in o-xylene. The circles are data points and the line is the interpolation between
the symbols.

500 550 600 650 700 750 800
0
1
2
3
Cu
CAAC
Cz
in o-xylene
with 20 mM triphenylamine
Wavelength (nm)
Molar Absorptivity (10
3
M
-1

cm
-1
)

Figure S4.2 – Cation absorptivity of 10 mM 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 in benzonitirile. The circles are data points
and the line is the interpolation between the symbols.triphenylamine in o-xylene. The circles
are data points and the line is the interpolation between the symbols.

400 600 800 1000 1200 1400 1600
0.5
1
1.5
2
2.5
3
Wavelength (nm)
Absorptivity (mO.D.)
10 mM Cu
DAC
CNCz

in benzonitrile
164

Figure S4.3 – Triplet absorption spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 (red) and 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 (black) in aerated o-
xylene, before O2 quenching of the triplet state and cation sensitization can occur (<5ns). The
negative feature observed in the 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 spectra is attributed to Cherenkov radiation from the
electrons slowing down as it passes into solution media. The circles are data points and the line
is the interpolation between the symbols.

400 500 600 700 800 900 1000
-0.4
0
0.4
0.8
1.2
Cu
CAAC
Cz
Cu
DAC
CNCz
Wavelength (nm)
∆Absoprtion (norm.)
165

4.7.2. TCSPC Curves
Depicted below are the TCSPC curves for the cMa compounds in both toluene and THF
pumped at 400 nm. The values for 𝜏𝜏 𝑃𝑃 , 𝐴𝐴 𝑃𝑃 , and � 1 − 𝐴𝐴 𝑝𝑝 � , are depicted as well. The longer time
range data, i.e. lifetimes, are displayed as the inset. For most compounds, TCSPC was collected
with an instrument described in the main text with an IRF ~ 22ps. However, for 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 ,
lifetimes were measured with a Horiba Fluorohub A+ equipped with a 405 nm laser source (< 300
ps IRF). Lifetime measurements were acquired from solutions at maximum optical densities
between 0.1 and 0.2 to minimize effects of solute-solute interactions.

Figure S4.4 – Emission decay trace for C 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and A 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (right) in toluene.

0 10 20 30 40
0.001
0.01
0.1
1
0 2 4 6 8 10 12
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Cu
MAC
Cz
in
Toluene
τ
p
= 220 ps
A
p
= 98.69%
1-A
p
= 1.31%

0 10 20 30 40
0.001
0.01
0.1
1
0 2 4 6 8
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Au
MAC
Cz
in
Toluene
τ
p
= <22 ps
A
p
= 99.62 %
1-A
p
= 0.38 %

Figure S4.5 – Emission decay traces for 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene (left) and in THF (right).

0 10 20 30 40
0.001
0.01
0.1
1
0 0.5 1 1.5
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
IRF
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Cu
MAC
BCz
in
Toluene
τ
p
= 298 ps
A
p
= 98.56%
1-A
p
= 1.44%

0 10 20 30 40
0.001
0.01
0.1
1
0 0.5 1 1.5
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
IRF
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Cu
MAC
BCz
in
THF
τ
p
= 347 ps
A
p
= 98.7 %
1-A
p
= 1.30 %
166

Figure S4.7 – Emission decay traces for 𝐴𝐴𝐴𝐴
𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene (left) and in THF (right).

0 10 20 30 40
0.001
0.01
0.1
1
0 0.5 1 1.5
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
IRF
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Cu
MAC
PhCz
in
Toluene
τ
p
= 284 ps
A
p
= 98.07%
1-A
p
= 1.93%

0 10 20 30 40
0.001
0.01
0.1
1
0 0.5 1 1.5
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
IRF
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Cu
MAC
PhCz
in
THF
τ
p
= 336 ps
A
p
= 98.13 %
1-A
p
= 1.87 %

Figure S4.8 – Emission decay traces for 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 in toluene (left) and in THF (right).

0 10 20 30 40
0.001
0.01
0.1
1
0 0.5 1 1.5
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
IRF
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Cu
DAC
CNCz
in
Toluene
τ
p
= 127 ps
A
p
= 98.28 %
1-A
p
= 1.72 %

0 10 20 30 40
0.001
0.01
0.1
1
0 0.5 1 1.5
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
IRF
Fit
A
p
= 99.03 %
1-A
p
= 0.97 % Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Cu
DAC
CNCz

in THF
τ
p
= 141 ps

Figure S4.6 – Emission decay traces for 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene (left) and in THF (right).

0 10 20 30 40
0.001
0.01
0.1
1
0 0.5 1 1.5
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
IRF
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Au
MAC
BCz
in
Toluene
τ
p
= <22 ps
A
p
= 99.52 %
1-A
p
= 0.48 %

0 10 20 30 40
0.001
0.01
0.1
1
0 0.5 1 1.5
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
IRF
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Au
MAC
BCz
in
THF
τ
p
= < 22 ps
A
p
= 99.59 %
1-A
p
= 0.41 %
167

Figure S4.9 – Emission decay trace for 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 (left) and 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 (right) in toluene

0 10 20 30 40
0.001
0.01
0.1
1
0 5 10 15
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Cu
CAAC
Cz
in
Toluene
τ
p
= 49 ps
A
p
= 99.19 %
1-A
p
= 0.81 %

0 10 20 30 40
0.001
0.01
0.1
1
0 2 4 6 8 10
0.001
0.01
0.1
1
Emission Intensity (norm.)
Time(µs)
Raw
Fit
Emission Intensity (norm.)
Time(ns)
Raw
IRF
Fit
Au
CAAC
Cz
in
Toluene
τ
p
= <22 ps
A
p
= 99.45 %
1-A
p
= 0.55 %
168

4.7.3. Picosecond Transient Absorption
4.7.3.1. psTA Data on Changing the Carbazole Unit and Solvent
In this section, we provide psTA datasets for the full set of compounds studied. The first
datasets are the remaining cMa compounds in toluene: 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , and 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 . Here, we
discuss the effect the carbazole has on the ESA of three cMa compounds in the 𝐴𝐴 𝐴𝐴 𝑀𝑀𝑀𝑀 𝐵𝐵 class with
reference to 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 from the previous results and discussion. The carbazole is the largest
contributor to the ESA line shape but substitutions to the carbazole unit leads to the smallest
spectral changes. This can be seen by comparing the relative peak heights of the ESA of the
𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 , and 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 at 1 ns for the triplet which have a relative ratio of ~1.2 (Figure S4.10,
Figure 4.9). The most significant change is the positioning of the 𝑆𝑆 1
690 nm peak in 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 is
now redshifted to 730 nm in 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 . 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 remains unshifted compared to 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 . Due to the
redshift of the ESA of the PhCz ligand, the SE carves out more of the ESA in 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 . The 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵
peak is obscured due to the weak probe intensity in the region near 800 nm. 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 is also
displayed in Figure S4.11. We note that two ligand substitutions have been made however we can
see clear evidence of the evolution of the SE “dip” toward the steady state fluorescence spectra
and in the ESA, the strongest ESA signature between 500 – 700 nm also seen in the bulk
electrolysis to form anions and cations from 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 as shown in Figure S4.2 and Figure S4.3.
The reduced spectral range is due to the use of a less dispersive grating for this particular
experiment.
169

In the main text, we showed that a change of solvent around 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 led to more dramatic
spectral evolution in the more polar THF, but little change in the important ISC rate constants.
Here we show another example, the psTA spectra for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in THF. Again, a significant red
shifting of the SE is seen. Once again, if we compare to 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in toluene, target analysis fitting
reveals once again little change in 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑒𝑒𝑒𝑒 𝑟𝑟 (within error) on change of solvent polarity. We have
used the 1.5 ns trace recorded in this experiment to properly correct the PL subtraction of the
corresponding nsTA experiment (see section 2.2.2).

Figure S4.10 – psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (right) in toluene.

400 500 600 700 800 900
-5
0
5
10
Wavelength (nm)
∆Abs (mOD)
0.3 ps 100 ps
1 ps 1000 ps
10 ps 1250 ps
Absorption Emission
Cu
MAC
BCz

400 500 600 700 800 900
-12
-6
0
6
12
18
∆Abs (mOD)
Wavelength (nm)
0.3 ps 100 ps
1 ps 1000 ps
10 ps 1500 ps
Absorption Emission
Cu
MAC
PhCz

Figure S4.11 – psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 in toluene.

400 500 600 700 800 900
-2
-1
0
1
2
3
Wavelength (nm)
∆Abs (mOD)
0.4 ps 100 ps
1 ps 500 ps
10 ps 1200 ps
Absorption Emission
Cu
DAC
CNCz
170

4.7.3.2. Charge Transfer Within a 1 ns Window with Concentrated Quencher
4.7.3.2.1. Electron Transfer to MePI
In this section by showing a dataset which shows effective and rapid quenching of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵
with 280 mM MePI in toluene to obtain 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
𝑀𝑀𝑀𝑀 𝐵𝐵 with the spectrum displayed in Figure S4.13.
Following the analysis performed in the main article on the compounds without quencher, we
assign similar features to 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with 280 mM MePI as we did with 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (Figure S4.10). The
early time feature at 300 fs is assigned to 𝑆𝑆 1
∗
. The relaxation of 𝑆𝑆 1
∗
to 𝑆𝑆 1
occurs on the order of 5
ps, consistent with the neat experiment. The spectrum at 10 ps corresponds to the 𝑆𝑆 1
state. The
apparent full SE loss from 𝑆𝑆 1
∗
to 𝑆𝑆 1
is likely because some quenching has begun to occur to bring
the entire spectrum closer to baseline producing an apparent SE loss. The region of SE has flattened
by ~2x as the delay advances from 10 to 100 ps. We assign the new positive feature at λ > 700
nm to 𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
𝑀𝑀𝑀𝑀 𝐵𝐵 based on good agreement with SEC. Importantly, this indicates that quenching of
𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene with MePI occurs on a 100s ps timescale and provides a useful contrast to the
slower quenching in the nsTA experiment of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene with a lower concentration of MePI
(Figure 4.11d).

Figure S4.12 – The psTA spectrum of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in THF.

400 500 600 700 800 900
-6
-4
-2
0
2
4
6
∆Abs (mOD)
Wavelength (nm)
0.3 ps 10 ps
1 ps 100 ps
3 ps 1500 ps
Absorption Emission
Au
CAAC
Cz
171

Determination of 𝑘𝑘 𝑞𝑞 was obtained using a fitting model that was an extension of Figure
4.7, displayed here as Figure S4.14. A fourth state (green) is added as the trapped pair,
𝐴𝐴 𝐴𝐴 ( 𝐵𝐵𝐵𝐵𝐵𝐵 )
+
𝑀𝑀𝑀𝑀 𝐵𝐵 /MePI
-
. Here, we have to assume quenching from both the singlet and the triplet, but we
assign the same rate constant (kq) as the excited state reduction potential only varies by a small
Δ 𝐸𝐸 𝑆𝑆 𝑇𝑇 .
25
Without singlet transfer, the SADS for the triplet are contaminated by the presence of
cation. If we do not include singlet transfer, we find physically unreasonable SADS. The other
kinetic parameters (grey arrows) are fixed from the neat 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 experiment. No decay pathway
from the trapped pair was considered as the trapped pair for a similar system, 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with MePI
in toluene, has a lifetime 110 ns and so the psTA will not have decayed appreciably by 1 ns, the
total length of this psTA experiment.

A good fit was found for kq∙[MePI] = 2.9∙10
9
M
-1
s
-1
and by dividing by concentration, 280
mM, 𝑘𝑘 𝑞𝑞 =(10 ± 8) ∙10
9
M
-1
s
-1
. Error bar values were obtained by iterating 𝑘𝑘 𝑞𝑞 with goodness of fit
based on a physically meaningful triplet spectrum from the SADS (this is displayed in the SADS
of psTA section, section 4.7.3.5). Deviations of the fitted triplet spectrum from the neat experiment
were rejected. This value varies from the values reported by Muniz et al.
25
which have a kq ~5x
400 450 500 550 600 650 700
-5
0
5
10
15
20
25
∆Abs (mOD)
Wavelength (nm)
0.3 ps
1 ps
10 ps
100 ps
500 ps
1000 ps
Cu
MAC
BCz
in toluene with
280 mM MePI

Figure S4.13 – The psTA spectrum of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene with 280 mM MePI.
172

smaller than the value reported here; however, we are using two orders of magnitude higher
concentration of quencher. Whether linearity over this wide quencher concentration regime is
reasonable is yet to be established.

Figure S4.14 – Kinetic model used for fitting the quenching of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with MePI. The left-
hand side is the same model as the psTA fitting of the neat compounds. The grey arrows are
fixed parameters from the neat experiment and the black arrows are allowed to vary but are
equal.
173

4.7.3.2.2. Hole Transfer to BIH
To compliment the electron transfer experiments at high quencher concentration with
psTA, we extend a similar scenario to hole transfer to BIH. Previously, we had performed nsTA
experiments with BIH at lower concentrations, but BIH absorbs the pump wavelength, 355 nm
(Figure S4.15). While BIH has low absorptivity at 355 nm, 30 M
-1
cm
-1
, at concentrations of 100
mM, the absorption of BIH in a 1 cm cuvette would be ~0.3 OD, the same order as the cMa. A
previous experiment of BIH only at 355 nm produced the BIH triplet (Figure S4.46, right) and
excitation at 355 nm of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 produced a spectra contaminated with BIH triplet singnal (Figure
S4.46, left). Therefore, in this section, we use the advantage that the psTA pump is 405 nm and
we excite the cMa solely. Additionally, these results were used to corroborate the nsTA
experiments at 450 nm (Figure 4.10).

We perform psTA quenching experiments of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF with BIH at BIH
concentrations of 0.25 and 0.75 M (Figure S4.16, left and right respectively). Due to photoproduct
buildup on the walls of the cuvette, the cuvette in the 0.75 M experiment was rastered by translation
perpendicular to the laser beams between each scan. The initial time slices (0.35 and 1 ps) mirror
300 350 400 450 500
0
1
2
3
4
5
6
7
Extinction (∙10
3
M
-1
cm
-1
)
Wavelength (nm)
ε
355
= 30 M
-1
cm
-1
ε
450
~ 0 M
-1
cm
-1
ε
405
~ 0 M
-1
cm
-1

Figure S4.15 – Molar absorptivity spectrum of BIH in THF with absorptivities at the pump
wavelengths indicated.
174

the neat psTA case indicating no quenching before these times. And in the case of the 0.25 M
spectra, it bears strong resemblance to the 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 only case. When we analyze the 0.75 M case,
we see strong diminishment of the 690 nm feature, assigned to loss the cation of the ICT state. A
strong trend reveals when we compare the neat to the two quenched cases at long times (Figure
S4.16, bottom). The ratio of the 550 to the 690 nm feature increases as a function of quencher
concentration which is expected as the cation of ( 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 )
∗
has larger contribution at 690 nm and
less at 550 nm like the anion. Unlike the electron transfer case, hole transfer produces 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 ( 𝑀𝑀𝑀𝑀 𝐵𝐵 )
−

which has a broad, featureless absorption spectrum. The hole on 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 is transferred to the BIH
cation, removing the cation of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 .

Figure S4.16 – psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF with BIH in concentrations of left) 0.25 M
and right) 0.75M. Bottom) comparison of the neat, 0.25 M and 0.75 M psTA spectra at long
times. Blue arrow → growth in the 550 nm region; black arrow → loss of 690 nm signature.
175

4.7.3.3. Contour Plots of psTA spectra
Here we display the contour plots of the psTA spectra of the cMa compounds pumped at
405 nm. Similar conditions were run for each sample, detailed in section 4.3.5. For the contour
plots. the x-axis is time in ps, plotted as linear-logarithmic axis with a break at 5 ps. The delay
times before 5 ps are plotted linearly and times after 5 ps are plotted logarithmically. The y-axis is
wavelength in nm, and the z-axis is 𝛥𝛥 𝐴𝐴𝑏𝑏𝑠𝑠 values in mOD. Positive 𝛥𝛥 𝐴𝐴𝑏𝑏𝑠𝑠 values tend red, negative
values tend blue and values near zero are white. The colormap code used here is detailed in
Appendix B, section 1. Please note the range on the colorbar changes with each dataset. The z-axis
values were truncated so the max value on the colorbar is taken from t > 340 fs to avoid
emphasizing the coherent response of the solvent. A coherent artifact is generated at time zero due
to two-photon absorption (2PA) from toluene and cross-phase modulation (XPM) from the cuvette
walls.
29,30
The coherent response appears as a red feature in the region of t = ± 300 fs and λ < 450
nm.
The psTA data were corrected for white light chirp both before displaying the present
contour plots and fitting the datasets. In addition to the IRF signal at λ < 420 nm and time zero,
another similar feature appears from -1 to -0.4 ps and 725 to 775 nm. This second feature reappears
at twice the wavelength due to the 2
nd
order diffraction in the spectrometer. Before white light
chirp correction, this feature appeared at the same time as the 2PA in the 400 to 450 nm region
and after correction, this feature appears at negative time. The pump-probe signal in the λ > 700
nm range is also contaminated. By taking the ratio of the peaks of the contamination at 750 nm to
the inherent coherent artifact at 370 nm, we find the contamination is 5-20%, varying between
compounds. The GSB region occurs λ
1
= 350- 450 nm, and the 2
nd
order diffraction would then
176

be λ
2
= 700 – 900 nm. Therefore, the ESA in the 700-900 nm region is dampened by 5-20%. This
effect will be revisited in the 𝑇𝑇 1
SADS comparison between psTA and nsTA, section 4.7.5.
The white streaks along λ = 400 and 800 nm are due to pump scatter removal. These
regions differ in exclusion size due to pump scatter intensity varying between each experiment.
The region of 725 to 850 nm (specific wavelengths varying for each experiment) was averaged
with a 5 pt smoothing. An exception is made with 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene where the 800 nm region is
smoothed by a 50 pt smoothing mandated by a weaker signal to noise ratio compared to the other
spectra. The smoothed data was used for fitting the data to obtain 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 for the gold complexes
and to obtain the species associated decay spectra (SADS) for every compound.

177

Figure S4.17 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene

Figure S4.18 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene

Figure S4.19 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene
178

Figure S4.20 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene

Figure S4.21 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF

Figure S4.22 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene

179

Figure S4.23 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 in toluene

Figure S4.24 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in toluene

Figure S4.25 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in toluene

180

Figure S4.26 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in THF

Figure S4.27 – Contour plot of psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene with 280 mM MePI

181

4.7.3.4. Operational Details for psTA and Target Analysis TA Fitting
In this section, we describe the use of target analysis to provide a full spectral and kinetic
fit of each TA dataset.
31
Target analysis is defined by photophysical and photochemical
parameters which extract meaningful information and provide a useful description of the transient
2D data. The fitted dataset is composed of a user-given number of compartments each further
composed of a time-independent spectrum and time-resolved population. To obtain the ultrafast
solvent and vibrational relaxation, the rate constants for the establishment of the ISC equilibrium
(psTA), lifetimes, quenching rates (nsTA), and the species associated decay spectra (SADS), a
target analysis and fitting program was used, referred here as PDP (developed by Mikas Vengris,
Laser Research Facility, Physics Dept, Vilnius University). The PDP code uses a master solution
to the kinetic equations based on a target model provided by the user. The scheme in Figure 4.7
was used as the target model for each cMa psTA experiment. First, we will discuss the fitting
procedure of the psTA, then the nsTA. Further description of other, non-cMa psTA data, can be
found in section 3.7.4.1.
For each experimental run of a day, one or several cMa compounds would be run along
with a matching solvent only spectrum. The solvent-only spectrum was used to determine several
key items. First, the IRF width of the experiment can be determined by plotting the wavelength of
the strongest signal of the coherent artifact as a function of time, focusing on time zero region.
The IRF containing only positive 2PA signal is at 350 nm and was fit to a single gaussian. The full
width half maximum (FWHM) of the gaussian was taken as the IRF width. The IRF width varied
from 300 to 340 fs from experiment to experiment. Second, proper excitation conditions are
determined by performing a solvent scan. With high pump fluence, the solvent can undergo two-
photon absorption (2PA) and generate a solvent excited state. This can remove excitation energy
182

from the cMa, producing undesired artifacts. An example of this is displayed in Figure S4.28,
where the pump fluence was too high and a subtraction of the toluene only signal from the 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵
spectrum was required. This was the only spectra this subtraction was needed for and applied to.
The time zero artifacts of cross-phase modulation and 2PA are significant and can skew
fitting to bias these features. To avoid biasing these features, the fitting around time zero was
ignored by adding a weighting function with the center at time zero, a width of ~300 fs, with zero
weight in this region. Time zero was allowed to float but fixed within ± 50 fs due to prior chirp
correction.
A good fit was judged with the following procedure in mind. Above all, a fit for a given
cMa compound must be physically and chemically sensible. As an example, if an excited state
absorption spectrum is known, from other measurements like SEC, for which an ESA is always
positive, a negative component in the SADS is discarded. The PDP fitting program was tailored
and appropriately tuned so that artifacts like these are rejected. With meaningful SADS, the time
trace at every probe wavelength, and the spectrum at every time delay was reviewed, and the
goodness of fit was judged by eye. A plot of fit residual is also used to determine systematic
divergence between model and data.
To determine 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑓𝑓 , the time traces of several wavelengths were monitored for the goodness
of fit at early times (< 10 ps) and iterated by 1 ps till the fit was best. For copper cMa complexes,
the values of 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 and 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 were obtained and fixed from TCSPC due to higher sensitivity to
these parameters in TCSPC. The values of 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 for the gold cMa complexes were obtained in a
similar manner as 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑓𝑓 . The sensitive range for 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑒𝑒𝑒𝑒 𝑟𝑟 was 10 to 100 ps. The values for 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 were
input and fixed in fluorescence fitting, which includes convolution with the measured IRF, to more
accurately determine 𝐴𝐴 𝑝𝑝 and thus provide values for 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 . The errors in the 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑓𝑓 and 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟 values
183

for the gold complexes were obtained via iteration and determination of quality of fit by hand
therefore establishing an estimate for confidence limits.

Figure S4.28 – psTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in toluene. (a) 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in toluene along with the
corresponding toluene only signal at the same time slices. (b) Spectrum recreated after
subtraction of toluene only signal from left experiment.

400 500 600 700 800 900
-10
0
10
20
30
∆Abs (mOD)
Wavelength (nm)
0.3 ps 100 ps
1 ps 1000 ps
10 ps 1500 ps
Solid - Au
CAAC
Cz
in Toluene
Dashed - Toluene Only
(a)

400 500 600 700 800 900
-10
0
10
20
30
∆Abs (mOD)
Wavelength (nm)
0.3 ps 100 ps
1 ps 1000 ps
10 ps 1500 ps
(b)
184

4.7.3.5. Species Associated Decay Spectra from psTA Datasets
In this section, we present the species associated decay spectra (SADS) obtained from
fitting the psTA data. When these basis spectra are multiplied by the time dependent
concentrations of each state, we recover the complete fitted psTA data. The SADS are the transient
signatures associated with each distinct kinetic state displayed in the model of Figure 4.7. The
fitting procedure is discussed in the previous section, 4.7.3.4.

400 500 600 700 800 900
-2
0
2
4
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
Cu
CAAC
Cz
400 500 600 700 800 900
-1
0
1
2
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
Cu
MAC
Cz

Figure S4.29 – SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (right) in toluene.
400 500 600 700 800 900
-10
0
10
20
30
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
Au
CAAC
Cz
400 500 600 700 800 900
-4
-2
0
2
4
6
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
Au
MAC
Cz

Figure S4.30 – SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (right) in toluene.

185

Figure S4.31 – SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (right) in toluene.

400 500 600 700 800 900
-3
0
3
6
9
12
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
Cu
MAC
BCz

400 500 600 700 800 900
-3
0
3
6
9
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
Au
MAC
BCz
400 500 600 700 800 900
-4
0
4
8
12
16
20
24
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
Cu
MAC
PhCz
400 500 600 700 800 900
0
1
2
3
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
Cu
DAC
CNCz

Figure S4.32 – SADS of 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 (right) in toluene.

400 500 600 700 800 900
-2
0
2
4
6
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
Au
MAC
BCz
400 500 600 700 800 900
-4
-2
0
2
4
6
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
Au
CAAC
Cz

Figure S4.33 – SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 (right) in THF

186

400 450 500 550 600 650 700
-10
0
10
20
30
∆Abs (mOD)
Wavelength (nm)
S
*
1
S
1
T
1
T
1
+ (Cu
BCz
MAC
)
+
Cu
MAC
BCz
in toluene
with 280 mM MePI

Figure S4.34 – SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with 280 mM MePI in toluene.

187

4.7.4. Nanosecond Transient Absorption
4.7.4.1. nsTA Spectra
The nsTA data displayed in this section were recorded in the same manner as described in
section 4.3.6 and additionally modified in the following way. First, the spectral traces were an
average of three consecutive time points, i.e. 10 ns is an average of the 8, 10, and 12 spectral traces.
Then, the averaged spectral traces were smoothed via a three-point algorithm. The data needed to
be corrected for a PL artifact, detailed in section 2.2.2.

Figure S4.35 – nsTA spectra of 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF (left) and toluene (right).

400 500 600 700 800 900
-1
0
1
2
∆Abs (mOD)
Wavelength (nm)
10 ns 500 ns
50 ns 1000 ns
200 ns 2000 ns

400 500 600 700 800 900
-1
0
1
2
∆Abs (mOD)
Wavelength (nm)
20 ns 1000 ns
200 ns 1500 ns
500 ns 2500 ns

Figure S4.36 – nsTA spectra of 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF (left) and toluene (right).

400 500 600 700 800 900
-0.5
0
0.5
1
∆Abs (mOD)
Wavelength (nm)
20 ns 600 ns
100 ns 1000 ns
300 ns 2000 ns

400 500 600 700 800 900
-1
0
1
2
∆Abs (mOD)
Wavelength (nm)
10 ns 500 ns
100 ns 1000 ns
250 ns 2000 ns
188

Figure S4.37 – nsTA spectra of 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF (left) and toluene (right).

400 500 600 700 800 900
-1
0
1
2
∆Abs (mOD)
Wavelength (nm)
10 ns 500 ns
50 ns 1000 ns
200 ns 2000 ns

400 500 600 700 800 900
-1
0
1
2
∆Abs (mOD)
Wavelength (nm)
20 ns 1000 ns
200 ns 1500 ns
500 ns 2500 ns

Figure S4.38 – nsTA spectra of 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF with excitation at (left) 410 nm and (right) 450
nm. This data was obtained on a Magnitude enVISion at UCR with excitation afforded by an
OPO.

400 500 600 700 800 900
-5
0
5
10
∆Abs (mOD)
Wavelength (nm)
10 ns 200 ns
50 ns 300 ns
100 ns 500 ns

400 500 600 700 800 900
-8
-4
0
4
8
12
16
20
∆Abs (mOD)
Wavelength (nm)
10 ns 500 ns
100 ns 1000 ns
250 ns 2000 ns

Figure S4.39 – nsTA spectra of 𝐴𝐴𝐴𝐴
𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF (left) and toluene (right).

400 500 600 700 800 900
-1
0
1
2
3
4
∆Abs (mOD)
Wavelength (nm)
20 ns 300 ns
100 ns 500 ns
200 ns 1000 ns

400 500 600 700 800 900
-1
0
1
2
∆Abs (mOD)
Wavelength (nm)
10 ns 500 ns
50 ns 1000 ns
200 ns 1600 ns
189

Figure S4.40 – nsTA spectra of 𝐴𝐴𝐴𝐴
𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 in THF (left) and toluene (right). Degradation of
𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 occurs with illumination in THF.

400 500 600 700 800 900
-0.2
0
0.2
0.4
Wavelength (nm)
10 ns 74 ns
24 ns 100 ns
50 ns 150 ns
∆Abs (mOD)

400 500 600 700 800 900
-0.5
0
0.5
1
∆Abs (mOD)
Wavelength (nm)
10 ns 500 ns
100 ns 1000 ns
250 ns 2000 ns

Figure S4.41 – nsTA spectra of 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in THF.

400 500 600 700 800 900
-1
0
1
2
3
∆Abs (mOD)
Wavelength (nm)
100 ns 2000 ns
500 ns 4000 ns
1000 ns 7000 ns

Figure S4.42 – nsTA spectra of 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 in THF (left) and toluene (right).

400 500 600 700 800 900
-1
0
1
2
3
∆Abs (mOD)
Wavelength (nm)
20 ns 300 ns
100 ns 500 ns
200 ns 1000 ns

400 500 600 700 800 900
-1
0
1
2
3
∆Abs (mOD)
Wavelength (nm)
10 ns 500 ns
100 ns 1000 ns
250 ns 2000 ns
190

Figure S4.43 – nsTA spectra of 46 µM 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 7 mM MePI in THF.

600 650 700 750
-0.1
0
0.1
0.2
0.3
Wavelength (nm)
10 ns 500 ns
50 ns 1000 ns
100 ns 2000 ns
∆Abs (mOD)

Figure S4.44 – nsTA spectra of 75 µM 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 30 mM MePI in THF, excitation at 420 nm
(see section 4.3.6). A peak attributed to the 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 cation is observed in the earliest time traces.

400 500 600 700 800 900
-4
-2
0
2
4
6
8
Wavelength (nm)
∆Abs (mOD)
6 ns 100 ns
12 ns 200 ns
50 ns 2000 ns
Absorption PR Cation
191

Figure S4.45 – A set of normalized nsTA decay traces of (a) 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 30 mM MePI (b)
𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and BIH in THF and. The TA trace of 750 nm – black and PL trace of 650 nm – red.
The time axis is lin-log with the break at 0.1 µs. The inset is the same traces displayed on a log-
log plot, demonstrating the PL intensity decays far below the TA. The apparent t0 offset of the
PL from the TA is due to the cation feature reaching its maximum ~400 ns after t0.

-0.1 0 0.1 1 10 100 1000
0
0.2
0.4
0.6
0.8
1
0 0.1 1 10 100 1000
1⋅10
-5
1⋅10
-4
1⋅10
-3
1⋅10
-2
1⋅10
-1
1⋅10
0
Time (µs)
∆Abs/PL Intensity (arb. u.)
10
3
10
2
10
1
10
0
10
-1
0
∆Abs/PL Intensity (arb. u.)
Time (µs)
TA
PL
(a)

0.05 0.1 1 10 100 1000
0
0.2
0.4
0.6
0.8
1
0 0.1 1 10 100 1000
1x10
-5
1x10
-4
1x10
-3
1x10
-2
1x10
-1
1x10
0
Time (µs)
∆Abs/PL Intensity (arb. u.)
10
3
10
2
10
1
10
0
10
-1
0
∆Abs/PL Intensity (arb. U.)
Time (µs)
TA
PL
(b)

Figure S4.46 – left) 64 mM BIH only in THF. Right) nsTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with 800 𝜇𝜇 M BIH
in THF. Excitation for both spectra is at 355 nm.

400 500 600 700 800 900
-0.1
0
0.1
0.2
∆Abs (mOD)
Wavelength (nm)
10 ns 1000 ns
100 ns 10000 ns
500 ns 60000 ns

400 500 600 700 800 900
-1
0
1
2
Wavelength (nm)
∆Abs (mOD)
20 ns 200 ns
50 ns 400 ns
100 ns 800 ns
192

Figure S4.47 – Left) Molecular structure of TMPD, another hole acceptor used in this chapter.
Right) nsTA spectra of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 and 25 mM TMPD in THF, excitation at 450 nm.

N
N
Tetramethyl-phenylenediamine
(TMPD)

400 500 600 700 800 900
-2
0
2
4
6
∆Abs (mOD)
Wavelength (nm)
10 ns
16 ns
24 ns
36 ns
50 ns
100 ns
193

4.7.4.2. Fitting Schemes for Quenching nsTA
For the nsTA data, the PDP fitting program was used again and where the fitting process
was simpler as there is less overall spectral evolution. The IRF was determined by collecting pump
light off a scattering solution and then fitting the resultant decay trace at the pump wavelength
with a gaussian. The IRF, 4 ns is consistent with manufacturer specifications. The nsTA is
therefore not sensitive to 𝑘𝑘 𝑟𝑟 𝑒𝑒 𝑓𝑓 , or to the ISC rates, 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑒𝑒𝑒𝑒 𝑟𝑟 and 𝑘𝑘 𝐼𝐼 𝑆𝑆 𝐵𝐵 𝑒𝑒 𝑛𝑛𝑟𝑟 𝑒𝑒 𝑟𝑟 , meaning the scheme in Figure
4.7 was simplified to only contain 𝑇𝑇 1
for the unquenched studies, with only one decay pathway,
𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 , displayed in Figure S4.48. A single compartment decay provides a strong constraint for
the nsTA lifetime. The nsTA lifetimes obtained from fitting the unquenched data has good
agreement with TCSPC measurements. For the quenched studies, a second compartment was
added to represent the quenched species, 𝐴𝐴 ( 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒 )
+
𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 or 𝐴𝐴 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒
( 𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒 𝑛𝑛𝑒𝑒 )
−
, with 2 decay pathways: 𝑘𝑘 𝑟𝑟 +
𝑘𝑘 𝑛𝑛 𝑟𝑟 and 𝑘𝑘 𝑞𝑞 [𝑄𝑄 ]. The value for 𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑛𝑛 𝑟𝑟 was fixed from the unquenched nsTA data.
The 𝑆𝑆 1
→ 𝑇𝑇 1
ISC occurs within the instrument response for the nsTA with nearly complete
extinguishing of the 𝑆𝑆 1
. Therefore, the 𝑇𝑇 1
state is considered the only active excited state from the
cMa when fitting the nsTA with low concentration of quencher, where quenching is expected to
occur solely from the 𝑇𝑇 1
. We can simplify Figure S4.14 to generate Figure S4.48.
194

Figure S4.48 – Simplified kinetic model used for fitting the nsTA data, adapted from Figure
S4.14, where states and processes are ignored to which the nsTA is insensitive. There are only
two active compartments: 𝑇𝑇 1
and 𝐴𝐴 ( 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒 )
+
𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 or 𝐴𝐴 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒
( 𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 )
−
. For the unquenched experiments,
only the 𝑇𝑇 1
compartment (red), is considered, while the 𝐴𝐴 ( 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒 )
+
𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒𝑛𝑛𝑒𝑒 or 𝐴𝐴 𝑎𝑎 𝑝𝑝 𝑎𝑎 𝑟𝑟𝑒𝑒
( 𝑐𝑐 𝑎𝑎 𝑟𝑟 𝑃𝑃𝑒𝑒 𝑛𝑛𝑒𝑒 )
−
compartment
(green) is added for the quenching studies.

𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑛𝑛𝑟𝑟

𝑘𝑘 𝑞𝑞 [ 𝑄𝑄 ]
𝑇𝑇 1

𝑆𝑆 0

𝐴𝐴 𝑎𝑎𝑝𝑝 𝑎𝑎 𝑟𝑟 𝑒𝑒 ( 𝑐𝑐𝑎𝑎𝑟𝑟𝑃𝑃𝑒𝑒𝑛𝑛 𝑒𝑒 )
−

Or
𝐴𝐴 ( 𝑎𝑎𝑝𝑝𝑎𝑎𝑟𝑟 𝑒𝑒 )
+
𝑐𝑐𝑎𝑎𝑟𝑟𝑃𝑃𝑒𝑒𝑛𝑛 𝑒𝑒
195

4.7.4.3. Species Associated Decay Spectra (SADS)/Triplet Spectra of cMa

400 500 600 700 800 900
-0.5
0
0.5
1
1.5
2
∆Abs (mOD)
Wavelength (nm)
T
1
Cu
MAC
Cz
in
toluene
400 500 600 700 800 900
-0.5
0
0.5
1
1.5
2
2.5
∆Abs (mOD)
Wavelength (nm)
T
1
Cu
MAC
BCz
in
toluene

Figure S4.49 – SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (right) in toluene.
400 500 600 700 800 900
-0.5
0
0.5
1
1.5
2
2.5
∆Abs (mOD)
Wavelength (nm)
T
1
Cu
MAC
PhCz
in
toluene
400 500 600 700 800 900
0
0.5
1
∆Abs (mOD)
Wavelength (nm)
T
1
Cu
DAC
CNCz
in
toluene

Figure S4.50 – SADS of 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 (right) in toluene.
400 500 600 700 800 900
-0.5
0
0.5
1
1.5
2
∆Abs (mOD)
Wavelength (nm)
T
1
Au
MAC
Cz
in
toluene
400 500 600 700 800 900
-1
0
1
2
∆Abs (mOD)
Wavelength (nm)
T
1
Au
CAAC
Cz
in
toluene

Figure S4.51 – SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 (right) in toluene.
196

4.7.4.4. SADS of nsTA Quenching Experiments

400 500 600 700 800 900
0
0.5
1
1.5
∆Abs (mOD)
Wavelength (nm)
T
1
Au
MAC
BCz
in
toluene
400 500 600 700 800 900
-0.5
0
0.5
1
1.5
∆Abs (mOD)
Wavelength (nm)
T
1
Au
MAC
BCz
in
THF

Figure S4.52 – SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene (left) and in THF (right).

400 500 600 700 800 900
-1
0
1
2
3
∆Abs (mOD)
Wavelength (nm)
T
1
Au
MAC
(BCz)
+
Au
MAC
BCz
in THF
6 mM MePI
(a)
400 500 600 700 800 900
-3
0
3
6
9
∆Abs (mOD)
Wavelength (nm)
T
1
Au
MAC
(BCz)
+
Au
MAC
BCz
in THF
30 mM MePI
(b)

Figure S4.53 – nsTA SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF with MePI concentrations of (a) 6 mM and (b)
30 mM.

400 500 600 700 800 900
-1
0
1
2
3
4
∆Abs (mOD)
Wavelength (nm)
T
1
/Au
MAC
(BCz)
+
Residual
Au
MAC
BCz
in Toluene
280 mM MePI

Figure S4.54 – nsTA SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with 280 mM MePI in toluene; the fit was unable to
separate the two compartments.
197

Figure S4.55 – nsTA SADS of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with 100 mM BIH in THF. The right SADS is a zoom
in on the small signal component of the left SADS.
198

4.7.5. Comparison of Triplet Spectra from psTA and nsTA
In this section, the normalized SADS from the triplet spectrum for psTA (black) and nsTA
(red) are depicted. The triplet spectra show good agreement. Deviations occur most strongly in
the region of strongest PL and the 800 nm region. Deviations the PL region in the nsTA
demonstrate imperfect PL subtraction. The region around 800 nm is more poorly determined in
the psTA due to the overlap with the white light continuum driving wavelength and imperfect
filtering in this spectral region.

400 500 600 700 800 900
-0.5
0
0.5
1
∆Abs (norm.)
Wavelength (nm)
psTA T
1
nsTA T
1
Cu
MAC
Cz
in
toluene
400 500 600 700 800 900
-0.5
0
0.5
1
∆Abs (norm.)
Wavelength (nm)
psTA T
1
nsTA T
1
Cu
MAC
BCz
in
toluene

Figure S4.56 – SADS from psTA and nsTA of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (right) in toluene.
400 500 600 700 800 900
-0.5
0
0.5
1
∆Abs (norm.)
Wavelength (nm)
psTA T
1
nsTA T
1
Cu
MAC
PhCz
in
toluene

400 500 600 700 800 900
-0.5
0
0.5
1
∆Abs (norm.)
Wavelength (nm)
psTA T
1
nsTA T
1
Cu
DAC
CNCz
in
toluene

Figure S4.57 – SADS from psTA and nsTA of 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵 (right) in toluene.
199

400 500 600 700 800 900
-0.5
0
0.5
1
∆Abs (norm.)
Wavelength (nm)
psTA T
1
nsTA T
1
Au
MAC
Cz
in
toluene
400 500 600 700 800 900
-0.5
0
0.5
1
∆Abs (mOD)
Wavelength (nm)
psTA T
1
nsTA T
1
Au
CAAC
Cz
in
toluene

Figure S4.58 – SADS from psTA and nsTA of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 (right) in toluene.

400 500 600 700 800 900
-0.5
0
0.5
1
∆Abs (norm.)
Wavelength (nm)
psTA T
1
nsTA T
1
Au
MAC
BCz
in
toluene
400 500 600 700 800 900
-0.5
0
0.5
1
∆Abs (mOD)
Wavelength (nm)
nsTA T
1
psTA T
1
Au
MAC
BCz
in
THF

Figure S4.59 – SADS from psTA and nsTA of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in toluene (left) and in THF (right).

200

4.7.6. Simulation of the 𝑆𝑆 1
and 𝑇𝑇 1
ESA utilizing Pulse Radiolysis Spectra
The spectra measured via PR can be utilized to simulate both the 𝑆𝑆 1
and 𝑇𝑇 1
ESA absorption
spectra. We extend this procedure for each of the copper complexes below examined by PR below.
The two ionic molar absorptivity spectra are added together in a 1:1 ratio, with the cation spectra
collected in o-xylene and the anion spectra collected in THF. The 𝑆𝑆 1
traces are taken from the
psTA SADS, while the 𝑇𝑇 1
spectra are taken from the SADS from the nsTA, with the exception of
𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝐵𝐵𝑀𝑀𝑀𝑀𝐵𝐵 which is taken from the PR data, due to sample instabilities when excited with the 355 nm
excitation source in our nsTA setup. Both ESA spectra have been normalized to the highest molar
absorptivity value of the black trace. The absorption and emission spectra of each compound in
toluene is reflected below the zero-line, normalized to an arbitrary 1000 M
-1
cm
-1
, in order to
highlight bleaches in the ESA spectra.

Figure S4.60 – The sum of the PR molar absorptivity plots (black) of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵
(right) compared to the 𝑆𝑆 1
state (blue) and the 𝑇𝑇 1
state (red) from SADS analysis.

1 1.5 2 2.5 3
-2
0
2
4
Energy (eV)
Cat + An (PR)
S
1
(norm) T
1
(norm)
Abs Norm Em Norm
Molar Absorptivity (10
3
M
-1
cm
-1
)
Cu
MAC
Cz
500 600 800 1100 1600
Wavelength (nm)
1 1.5 2 2.5 3
-2
0
2
4
6
8
Energy (eV)
Cat + An (PR)
S
1
(norm)
T
1
(norm)
Absorption
Emission
Molar Absorptivity (10
3
M
-1
cm
-1
)
Cu
MAC
BCz
500 600 800 1100 1600
Wavelength (nm)
201

Figure S4.61 – The sum of the PR molar absorptivity plots (black) of 𝐴𝐴 𝐴𝐴 𝑃𝑃 ℎ 𝐵𝐵 𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (left) and 𝐴𝐴 𝐴𝐴 𝐵𝐵 𝐶𝐶 𝐵𝐵 𝐵𝐵 𝐷𝐷 𝑀𝑀𝐵𝐵
(right) compared to the 𝑆𝑆 1
state (blue) and the 𝑇𝑇 1
state (red) from SADS analysis.

1 1.5 2 2.5 3
-2
0
2
4
6
8
Energy (eV)
Cat + An (PR)
S
1
(norm) Absorption
T
1
(norm) Emission
Molar Absorptivity (10
3
M
-1
cm
-1
)
Cu
MAC
PhCz
500 600 800 1100 1600
Wavelength (nm)
1 1.5 2 2.5 3
-2
0
2
4
Energy (eV)
Cat + An (PR)
S
1
(norm) Absoprtion
T
1
(norm) Emission
Molar Absorptivity (10
3
M
-1
cm
-1
)
500 600 800 1100 1600
Wavelength (nm)
Cu
DAC
CNCz
202

4.7.7. Simulation of the 𝑆𝑆 1
and 𝑇𝑇 1
ESA utilizing Bulk Electrolysis Spectra
In this section, in a similar method to sections 4.4.5 and 4.7.6, we utilize
spectroelectrochemistry data to simulate what we might expect the ICT spectrum of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 to see
if it could be well approximated as a sum of absorption spectra of the oxidized and of the reduced
complex. We start with the two BE spectra obtained for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF (Figure 4.5, right) and
then compare their sum with the 𝑆𝑆 1
and 𝑇𝑇 1
spectra derived from the psTA and nsTA SADS,
respectively (Figure S4.31, left and Figure S4.56, right). We are forced to use TA data from 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵
in toluene to show derived ICT spectra as TA data for 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 in THF was not collected. The
emission and absorption spectra shown are also collected in toluene. While the absolute molar
absorptivity of the 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 cation and anion is known from PR, the concentrations of the BE cation
and anion are not known. Therefore, the anion and cation are scaled independently and then added
together. The anion appears scaled by 4x compared to the cation – this best qualitatively matches
the 500 to 700 nm region and the region >800 nm.
We can see that the 𝑆𝑆 1
SADS matches quite closely the sharpest peak, and vibrational
sideband, in the ICT simulation based on the individual anion and cation absorption spectra. In
the 𝑆𝑆 1
SADS, these two resolved peaks are shifted by about 30 nm to the blue, presumably by
Coulombic interaction with the anion on the carbene not present in the electrochemically oxidized
complex. Progressing to higher transition energies, the contribution of the stimulated emission in
the 𝑆𝑆 1
SADS (and not to the 𝑇𝑇 1
SADS) explains the greater divergence to the simulated ICT,
although it is not as dramatic as in 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 (Figure 4.10). The 𝑇𝑇 1
SADS also exhibits peaks that
can be associated with the carbazole cation, but they are shifted by ~ 80 nm to the blue and
somewhat broadened, as found for 𝐴𝐴𝐴𝐴
𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 . Both SADS have contributions from the bleach that
make comparisons with the BE spectra harder in the 400 – 500 nm region.
203

1.5 2 2.5 3 3.5
0
1
Absorption (norm.)
Energy (eV)
Cat + An
S
1
(norm) T
1
(norm)
Absorption Emission
Cu
MAC
BCz
900 700 600 500 400
Wavelength (nm)

Figure S4.62 – The sum of the BE spectra (black) of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 compared to the 𝑆𝑆 1
state (blue) and
the 𝑇𝑇 1
state (red) from SADS analysis.
204

4.7.8. Stern-Volmer plots of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with BIH in THF

Figure S4.63 – Left) Stern-Volmer plot of 𝐴𝐴 𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵 𝑀𝑀𝑀𝑀 𝐵𝐵 with BIH in THF generated from TCSPC
left) at 405 nm with the Horiba Fluorohub A+ instrument and right) at 450 nm with the
enVISion at UCR. Note the 450 nm data is extrapolated to 0.1 M, ~10x higher than the 405 nm
data.
205

Chapter 5. Role of the Perfluoro Effect in the Selective Photochemical Isomerization of
Hexafluorobenzene
3‡

5.1. Abstract
Hexafluorobenzene and many of its derivatives exhibit a chemoselective photochemical
isomerization, resulting in highly strained, Dewar-type bicyclohexenes. While the changes in
absorption and emission associated with benzene hexafluorination have been attributed to the so-
called “perfluoro effect”, the resulting electronic structure and photochemical reactivity of
hexafluorobenzene is still unclear. We now use a combination of ultrafast time-resolved
spectroscopy, multiconfigurational computations, and non-adiabatic dynamics simulations to
develop a holistic description of the absorption, emission, and photochemical dynamics of the 4π-
electrocyclic ring-closing of hexafluorobenzene and the fluorination effect along the reaction
coordinate. Our calculations suggest that the electron-withdrawing fluorine substituents induce a
vibronic coupling between the lowest-energy
1
B2u ( ππ*) and
1
E1g ( πσ*) excited states by selectively
stabilizing the σ-type states. The vibronic coupling occurs along vibrational modes of e2u symmetry
which distorts the excited-state minimum geometry resulting in the experimentally broad,
featureless absorption bands, and a ∼100 nm Stokes shift in fluorescence in stark contrast to
benzene. Finally, the vibronic coupling is shown to simultaneously destabilize the reaction
pathway toward hexafluoro-benzvalene and promote molecular vibrations along the 4π ring-
closing pathway, resulting in the chemoselectivity for hexafluoro-Dewar-benzene.

3‡
The contents of this chapter were adapted from Cox et, al, J. Am. Chem. Soc. 2021, 143,
7002-7012.

206

5.2. Introduction
The synthesis of polyacetylene sparked the field of organic electronics, despite difficulties
in processing polyacetylene and its high propensity to be oxidized when exposed to air.
1,2
As early
as 1979, polymer chemists have sought to circumvent these challenges, especially its instability,
with a strategy of fluorination. Theoretical calculations predict that the electron-withdrawing
fluorines significantly reduce the oxidation rate by lowering the frontier molecular orbitals of
polyacetylene and minimizing overlap with those of molecular oxygen. However, synthesizing
fluorinated polyacetylenes has proven difficult by the traditional means of polymerizing
fluorinated acetylene.
3
Not only is difluoroacetylene an explosive and pyrophoric gas, but
polymerization results in a highly irregular material that contains a mixture of −CF, −CF2, and
−CF3 functional groups. Burns and co-workers
4
recently achieved this elusive synthesis by
leveraging the mechanochemical “unzipping” of highly strained ladderene polymers to form
partially fluorinated polyacetylenes. The key step in their synthetic scheme is the formation of a
fluorinated ladderene, achieved through a photochemical cascade reaction, involving a
hexafluorobenzene (HFB) [2 + 2]-cycloaddition, followed by a disrotatory 4π-electrocyclic ring-
closing of the hexafluorinated diene intermediate (Figure 5.1b). This fluorinated ladderene is then
polymerized and mechanochemically unzipped to yield fluorinated polyacetylene (Figure 5.1c).
207

The 4π electrocyclic ring-closing reaction is known for many highly fluorinated cyclic
polyenes, many of which result from cycloadditions with HFB.
5–8
HFB itself undergoes the same
ring-closing reaction,
9
photochemically isomerizing to hexafluoro-Dewar-benzene (Dewar-HFB)
chemoselectively. The hexafluoro-benzvalene or -fulvene isomers are not observed, but are the
major products of benzene photochemistry (Figure 5.1a).
10,11
HFB also shows significantly
different photophysical behavior than benzene; its absorption spectrum has a broad, featureless
absorption band and a substantial (100 nm) Stokes shift in the emission band, even in the gas
phase.
12
These dramatic changes in photophysics have been attributed to the “perfluoro effect”.
13

The stark contrast in the photochemical outcomes of HFB and benzene photochemistries suggests
that the perfluoro effect also plays an important role as HFB isomerizes to Dewar HFB. The
chemoselective isomerization of HFB was first reported in 1966 when Haller and Camaggi
9,14

Figure 5.1 – (a) Photochemical isomerization reactions of benzene and HFB yielding different
product distributions.
5
(b) Photochemical cascade reaction of Burns and co-workers to form a
fluorinated ladderene.
4
(c) Subsequent polymerization and mechanochemical unzipping of the
fluorinated ladderene to form fluorinated polyacetylene.
208

independently reported the reaction. Further work by Haller determined that the quantum yield of
the gas-phase isomerization was ≤3%, under pressures ranging from 44 to 1052 Torr.
15
The
quantum yield was unchanged in the presence of a triplet sensitizer, O2. This suggests that the
reaction proceeds entirely in the singlet manifold. Picosecond transient absorption (TA)
spectroscopy indicates the possibility of diradical or charge-transfer intermediates, though the role
of these species in the photochemistry is still unknown.
16
This quantum yield is low compared to
other examples of 4π electrocyclic ring-closing reactions where quantum yields can reach 17%.
17

Laser-induced fluorescence measurements by Zgierski et al. show that benzene
perfluorination substantially lowers the energies of the C−F σ and σ* orbitals.
12
In fact, Zgierski
proposes that in HFB, the σ-type orbital stabilization is sufficient to reverse the ππ* and πσ* state-
ordering relative to benzene and that the S1 state in HFB is of πσ* character.
12
The πσ* S1
assignment has been used by Zgierski and Temps
12,18
to explain the large Stokes shift by assigning
the gap between the absorption and emission onsets to an optically dark 𝑆𝑆 0
→ 𝑆𝑆 1
transition of πσ*
character. This idea has been used to explain the low fluorescence yield and short fluorescence
lifetime of HFB. Philis et al., Motch et al., and Holland et al. instead agree that the S1 and S2 states
are ππ* and πσ*, respectively, based on photoabsorption and photoelectron spectroscopy
measurements.
19–21

Mondal et al. pioneered theoretical work on HFB dynamics, describing the initial ultrafast
𝑆𝑆 𝐶𝐶 → 𝑆𝑆 1
relaxation processes, and used equation-of-motion coupled-cluster (EOM-CCSD)
calculations to support the S1 assignment of Philis, Motch, and Holland.
22
However, a model of
the photophysics and photochemistry of HFB which unifies the current reported observations does
not yet exist. Developing this unified description would lay the foundation for design rules of
highly strained compounds that leverage the chemoselective photoisomerization of HFB and its
209

derivatives. This work utilizes ultrafast time-resolved spectroscopic studies, multireference
calculations, and nonadiabatic molecular dynamics simulations to clarify the photophysics and
origin of the low yielding, chemoselective isomerization to Dewar-HFB. While this photochemical
reaction is synthetically useful,
4
this work shows why this photochemical reaction is inefficient
and emphasizes the difficulty in tracking such reaction pathways.
5.3. Experimental
5.3.1. Computational Methods
A full description of the computational methods is described in Cox et al.
3‡

5.3.2. Sample Preparation and Purification
Hexafluorobenzene (Sigma-Aldrich, >99% purity) was used without further purification
for the transient absorption and TCSPC measurements. For fluorescence and UV−vis experiments,
HFB was purified by fractional distillation at 120 °C to remove fluorescent impurities. All HFB
solutions were made with ethanol (VWR, 200 proof). Distillation was performed using a fractional
distillation column and a water jacketed condenser. The HFB was boiled in a round-bottom flask
using mineral oil to a bath temperature of 120 °C. The first few drops of the distillate, the head,
was removed from the collection flask and discarded. The first distillate was then collected and
redistilled again. The HFB used in long-wavelength UV−vis experiments, both distilled and
nondistilled, was then diluted via serial dilution and used for the experiment. Tryptophan (Sigma-
Aldrich, 97%) was used as the fluorescence quantum yield standard without further purification.
5.3.3. Absorption and Fluorescence Measurements
All absorption measurements were performed on a Cary 60 UV−vis spectrometer with
ethanol in a quartz 1 cm cuvette used to set the baseline. The fluorescence measurements were
carried out on a Horiba Jobin Yvon FluoroMax-3 fluorometer. A Starna Cells spectrosil 1 cm path
210

length cuvette was used for all measurements. The emission and excitation spectra were obtained
during the same experiment. The excitation was fixed at a given wavelength as the emission
spectrum was scanned between 230 and 500 nm. The excitation was stepped in 2 nm increments
between 200 and 300 nm and again the emission spectrum was collected. The integration time was
100 ms with bandwidths for the excitation and emission set at 5 nm. To create the excitation
spectrum for Figure S8, the entire 2D array was integrated along the emission axis to collapse it
down to a 1D array with excitation wavelength as the only variable. In the quantum yield
measurements, the absorbances of the HFB and tryptophan/buffer solutions were 0.034 OD and
0.049 OD, respectively, at the excitation wavelength of 270 nm. The buffer solution, used as the
solvent for tryptophan, produced no fluorescence at 270 nm excitation. For the QY fluorescence
measurements, the bandwidths of both the excitation and emission monochromators were set to 2
nm and scanned in 0.25 nm increments with each point integrated for 100 ms. The fluorescence
from the pure ethanol was subtracted off the HFB intensity. The wavelength-dependent
fluorescence quantum yield (Figure S8) was calculated by normalizing the HFB absorption and
integrated excitation to 1 at 270 nm. The QY was calculated by dividing the integrated excitation
spectrum by the absorption spectrum and scaling by 0.029−the fluorescence quantum yield of
tryptophan at 270 nm.
5.3.4. Temperature-Dependent Fluorescence
Temperature-dependent fluorescence was recorded using a QuantaMaster Xe arc lamp
spectrofluorometer equipped with an R1527 PMT detector loaded with a 71 μM solution of HFB
in ethanol in a 1 cm screw-top quartz Starna Cells cell. The excitation monochromator was
centered at 255 nm, while the emission monochromator was stepped in 1 nm intervals from 300 to
500 nm. The integration time was set to 1 s, and the excitation and emission monochromator
211

bandwidth slits were set to 5 nm. Scans were averaged twice. The sample cell is immersed in a
temperature-controlled flowing water bath which, was varied between 23 and 56 °C.
5.3.5. Femtosecond-nanosecond Tranisent Absorption
The 35 fs, 800 nm output of a Coherent Legend amplifier was taken to generate both the
pump and probe beams. About 10% was sent to generate the white light continuum probe. The
time delay between the pump and probe was generated by sending the probe down a translating
gold-plated cube corner retroreflector. The white light was generated by focusing the fundamental
onto a 2 mm rotating CaF2 disk. The supercontinuum was collimated and then focused on the
sample by two off-axis parabolic mirrors. The transmitted probe beam was collimated and then
focused onto a 120 μm slit of a grating Czerny-Turner monochromator. The light was then
dispersed across a 256-pixel array to achieve a detected continuum spanning from 345 to 650 nm.
The deep UV (DUV) pump beam was generated in a home-built noncolinear optical
parametric amplifier (NOPA) pumped with 60 mW of 800 nm (35 fs). About 10% of the
fundamental was focused on a sapphire disk to generate a white light seed. The remaining
fundamental was directed and focused on an SHG BBO to generate the 400 nm pump. The signal
was generated by combining the weak white light seed onto a thin BBO crystal with the 400 nm
pump. The signal generated, 510 nm, is then passed through a prism compressor to compensate
for spectral chirp of the visible beam and increase DUV SHG efficiency. The signal is then sent
through another BBO crystal to generate the second harmonic of 510 nm, 255 nm. The DUV beam
was sent through two calcium fluoride prisms to compress the UV pulses. The DUV pulse was
focused to a spot size of 360 μm with a pulse energy of 580 nJ, 5 cm, in front of the gravity jet
sample.
23
A gravity jet was used in place of a sample flow cell to reduce nonlinear effects such as
cross-phase modulation and possible signal from the glass and reduce temporal walk off of the
212

pump and probe pulses due to spectral chirp, optimizing the time resolution. For this setup, the
sample was contained within an upper reservoir that flowed downward, forming a thin sheet across
a loop of 0.005 in. tungsten wire, forming a jet with a thickness of 130 μm at the pump−probe
overlap. The pump and probe were overlapped at the gravity jet sheet in the center of the loop to
minimize lensing effects. The reservoir height was maintained at a constant height of ∼40 cm to
ensure a consistent sheet thickness throughout the experiment. To compensate for the low signal-
to-noise ratio, the spectra were averaged in delay time using a 10-point moving mean algorithm.
This improves the signal-to-noise ratio from 16 to 87 when assessed at the same 300 fs, 550 nm
point.
5.3.6. Time-Correlated Single Photon Counting
The fluorescence lifetime of HFB was measured using Time-Correlated Single Photon
Counting (TCSPC). The TCSPC data presented in this chapter is contained two sections, with
pump frequencies produced by two different methods. The method used in Lopez et al and section
5.7.3 is detailed first. 532 nm, produced as the second harmonic of an Alphalas PULSELAS-A-
266-300-SP Nd:YAG laser operating at 10 kHz, was focused into a fourth harmonic generation
(FHG) crystal to yield 266 nm, which was used to pump the sample. The sync trigger is provided
by the residual 532 nm passing through the 266 nm pick off the high reflector. This residue was
used to trigger a Becker & Hickl GmbH PHD-400-N fast photodiode sync. The bandwidth of the
laser is 1.9 nm at 266 nm. The experiment was done in a 1 cm quartz cuvette with a sample
absorption of 0.043 OD at 266 nm. The total time acquisition of the TCSPC measurements was 60
min with an intensity of 150 mW/cm
2
.
The wavelength dependent lifetimes, results detailed in section 5.7.4, were obtained in the
following way. The pump is driven by a 100 kHz pulse train of 50 fs 6 mJ pulses produced by a
213

Coherent RegA 9050 seeded by a Coherent Mira Seed, centered at 800 nm. About 25% of the
fundamental is directed to the probe path of a separate TA experiment, leaving 75% to pump the
UV TCSPC experiment.
The residual 800 nm light is fed into a Coherent 9450 Optical Parametric Amplifier (OPA)
tuned to produce 150 nJ pulses at 520 nm with a bandwidth of 10 nm FWHM. These visible pulses
are fed into a prism dispersion compensation system consisting of a pair of UV-Fused Silica
(UVFS) prisms. The pulses are focused using a 10 cm focal length UVFS lens into a 0.1 mm BBO
crystal cut for SHG in the visible. The prism pair are optimized to maximize the conversion
efficiency into the UV rather than for shortest visible pulse length. This yields pulses which are
recollimated using a 10 cm focal length VUV grade CaF2 lens. The UV pulses are then directed to
the TCSPC setup and were focused at the interaction region using a 10 cm focal length VUV grade
CaF2. First, 266 nm was generated and used with the remaining UV wavelengths (247, 253 and
273 nm) were generated in a similar way. Acquisition time for this experiment was set to 20
minutes.
For both sets of experiments, the power was set such that the number of detected photons
was lower than ∼1% of the laser repetition rate. This was to reduce the probability of two detected
events per laser shot, which can distort the timed-binned histogram. The monochromator used was
a CVI Digikrom CM-112 1/8 Monochromator. The monochromator was set at 370 nm, the height
of the HFB emission spectrum. The slit widths were set at 16 nm by using static metal slits. A
large bandwidth was required due to low signal level.
5.4. Results and Discussion
The density of frontier excited states and the highly symmetric D6h structure of HFB has
generated substantial debate in the literature about the nature of the light-induced electronic
214

transitions. The ππ* or πσ* character of these states dictates the shape of the potential energy
surface which directs the photochemistry. Without a full understanding of the order and interplay
between the molecular excited states of HFB, a clear understanding of the inefficient yet
chemoselective reactivity has remained elusive. Figure 5.2 shows the overlay of the experimental
absorption spectrum for HFB solvated in ethanol and the simulated gas- phase absorption
spectrum, with the contributions of each of the four lowest-energy electronic states to the simulated
spectrum. The steady-state solution-phase UV−vis absorption spectrum, shown by the dashed gray
trace, was measured for a 75 μM solution of HFB in ethanol. The molar absorptivities in the
measured transitions lie in the 200−600 M
−1
cm
−1
range, indicating that HFB is a relatively weak
absorber for these lowest energy transitions – an intuitively undesirable quality in a photochemical
reagent. The simulated gas-phase absorption spectrum, the solid black trace, was computed (by J.
Cox) and at the extended multistate XMS- CASPT2(6,7)/aug-cc-pVDZ level of theory based on
164 HFB geometries randomly sampled from a 1 ps SA-10-CAS(6,7)/6- 31G molecular dynamics
trajectory propagated in the ground state at 300 K in vacuum. As the ground state equilibrium HFB
structure has no net dipole moment, precluding solvent dipole interactions, no polarizable
continuum models were included in the calculation. More sophisticated explicit solvent models
were deemed too computationally expensive at this time. The absorption line-spectrum computed
at the XMS-CASPT2-(6,7)/aug-cc-pVDZ level for the D6h-symmetric minimum geometry predicts
absorption oscillator strengths of 1.7 × 10
−5
and 7.6 × 10
−5
for the
1
B2u and
1
B1u ( ππ*) states,
respectively, and 1.9 × 10
−9
for the
1
E1g ( πσ*) state. The computed absorption spectrum in Figure
5.2 encompasses geometries that break the D6h symmetry leading to substantially higher average
oscillator strengths of 6.7 × 10
−3
and 2.4 × 10
−3
for the two
1
E1g states, which is comparable to the
1
B2u and
1
B1u states at 3.2 × 10
−3
and 9.0 × 10
−2
, respectively.
215

5.4.1. Absorption Spectrum
The calculations in this section were performed by J. Cox. The colored curves in Figure
5.2 show the contributions to the total gas-phase absorption spectrum computed for each of the
four lowest-energy electronically excited states. The calculated absorption spectrum suggests that
the lowest-energy excited state is the
1
B2u ( ππ*) state and that the absorption peaks at 230 and 220
nm arise from splitting of the
1
E1g ( πσ*) state. This assignment matches the previous experimental
work of Philis, Motch, and Holland and the theoretical work of Mondal. Mondal demonstrated that
the splitting of the
1
E1g state resulted from the Jahn−Teller effect but disagrees with Temps and
coworkers who assign the
1
E1g state as 𝑆𝑆 1
. The Temps assignment is bolstered by providing an
intuitive explanation of the large gap between the absorption and emission onsets of 100 nm by
suggesting that this gap arises instead from emission from a low-lying
1
E1g ( πσ*) dark state
centered at around 280 nm. Therefore, for the assignment of 𝑆𝑆 1
as ππ* to be correct, an unusually
large Stokes shift would require an 𝑆𝑆 1
potential energy surface where the excited state minimum
geometry is significantly different from the ground state minimum – a potentially important
observation in understanding the photochemistry.
To determine the origins of the 100 nm Stokes shift observed in the absorption/fluorescence
spectra of HFB, we present the absorption and fluorescence emission spectra of a 75 μM solution
of HFB in ethanol with a solid red line (absorption) and red filled trace (fluorescence) in the main
panel of Figure 5.3.
216

Figure 5.2 – Theoretically computed gas-phase absorption of HFB in black and the measured
UV−vis absorption spectrum of HFB solvated in ethanol in dashed gray. The transition-specific
computed spectra are rendered below in solid-colored lines with upper state symmetry
assignments shown in the legend. The inset panel shows the vertical excitation energies for the
four lowest excited states of HFB. All calculations used the aug-cc-pVDZ basis set, CAS and
(XMS-CASPT2) calculations included a (6, 7) active space described in the Experimental
Section in Lopez et al.
2‡

Figure 5.3 – Absorption and emission spectra for HFB (red, main panel) and benzene (blue,
inset panel). Absorption spectra are rendered in solid lines and emission spectra in filled traces.
Both absorption and emission spectra for HFB are measured using a 75 μM solution in ethanol,
while the analogous benzene spectra are adapted from Du et al.
24
and the extinction values
calibrated by Berlman.
25
The black stick spectra represent vibrationally resolved absorption and
emission spectra computed from Franck−Condon factors of the ground- and
1
B2u excited state
gas phase minimum structures. Franck−Condon factors are computed using the parallel mode
approximation at 0 K.
217

5.4.2. Vibrationally Resolved Electronic Spectra
Beneath these experimental traces is a black stick spectrum showing the computed
vibrationally resolved absorption and emission spectra between the
1
A1g and
1
B2u ( ππ*) states of
the isolated molecule. The inset panel shows the equivalent spectra for benzene in blue, where the
absorption and fluorescence were adapted from Du et al.
24
and the absorption scaled by the
Berlman extinction coefficient.
25
The vibrationally resolved electronic spectra were simulated by
computing the Franck− Condon factors for the overlap between the
1
A1g and
1
B2u state vibrational
wavefunctions using the harmonic approximation as implemented in ezSpectrum.
26
In detail, these
computed spectra consider the 13 modes with most significant Franck−Condon activity (vide infra)
and includes up to 21 and 30 vibrational quanta in these modes for the absorption and emission
spectra, respectively. Figure 5.3 shows the overlay of the experimental and simulated vibrationally
resolved electronic spectra. The Franck−Condon overlap between the
1
A1g and
1
B2u states suffices
to explain the origin of the large Stokes shift of 1.3 eV (260−360 nm) between the absorption and
emission spectra. This computed value slightly underestimates the experimentally measured
Stokes shift of 1.56 eV. This conclusively demonstrates that the Stokes shift is an inherent feature
of the
1
B2u state, rather than an artifact from a dark emissive state. The large value of the Stokes
shift is equivalent to recognizing there is a large displacement along one or more vibrational
modes. Our analysis of the vibrationally resolved electronic spectra, shows the most intense
absorption and emission features deposit 10 and 14 vibrational quanta in the excited-state and
ground-state, respectively. These vibrational quanta are distributed through several combination
progressions and predominantly activate the ν18 and ν19 modes.
27
These modes have e2u
symmetry in D6h and correspond to out- of-plane bending of the para-oriented carbon and fluorine
218

atoms, respectively. The selective population of the e2u vibrational modes suggests a symmetry-
dependent coupling dominates the absorption and emission spectra.
5.4.3. Electronic State Coupling
The electronic structures and gas-phase minimum energy geometries of the ground and
1
B2u states and the motions connecting them to understand the origin of the Stokes shift and
vibrationally resolved electronic spectra were examined here and greater in detail in Lopez et al.
3‡

The left-hand side of Figure 5.4 shows the completely planar D6h geometry of the ground state
minimum and the minimum-energy geometry of the
1
B2u state. The excited state minimum is
significantly distorted compared to the high-symmetry ground state geometry. Two of the carbon
atoms, with a 1,4 relationship, are visibly distorted out- of-plane, and the fluorine atoms attached
to the 1,4-carbons are bent out-of-plane by 18°. This results in a C2v symmetric geometry. This
contrasts with calculations of the optimized excited state minima of benzene, where the
1
B2u
minimum- energy geometry is D6h symmetric.
28
This significant geometric distortion from 𝑆𝑆 0
to
𝑆𝑆 1
results in the observed Stokes shift by reducing the 𝑆𝑆 1
/ 𝑆𝑆 0
gap by 1.53 eV.

Figure 5.4 – (Left) Side and top elevations of the minimum energy geometries for the 𝑆𝑆 0
(
1
A1g)
and 𝑆𝑆 1
(
1
B2u) states. (Right) Cuts through the ground,
1
A1g (black), and first four excited,
1
B2u
(orange),
1
E1g (green),
1
E1g (blue), and
1
B1u (red), potential energy surfaces, along the ν18 e2u
motion corresponding to opposing carbons moving symmetrically out of plane.
219

The minimum energy path (MEP) from the Franck− Condon point on the
1
B2u surface was
computed to investigate the evolution of the electronic structure of HFB along this geometric
distortion. The wavefunction along the MEP is entirely comprised of ππ* configurations while the
molecule is D6h symmetric; symmetry breaking along the MEP induces πσ * mixing. It was found
that the ππ* and πσ* configuration mixing with the symmetry-breaking distortion suggests that the
1
B2u state is vibronically coupled with the energetically nearby
1
E1g ( πσ*) state.
As this coupling should occur via e2u vibrational modes, an energy plot of the ground-state
and the first four excited-states with respect to e2u normal mode displacement is shown in the right
of Figure 5.4. The orange trace shows the double-well stabilization of the
1
B2u state when
displaced along the e2u symmetric modes leading to the distorted minimum energy geometry. The
blue and green traces represent the splitting of the two components of the
1
E1g state. Given this
observation, one would have expected the harmonic approximation used above to perform poorly
for the excited state double well potential, so the good agreement between the computed vibronic
absorption band and the first experimental absorption band rendered in Figure 5.3 is somewhat
surprising. Previous work has shown that the vibronic coupling constant between the
1
B2u and
1
E1g
states is 0.1548 and 0.0619 eV for coupling through modes ν18 and ν19, respectively.
22
The cross-
section of the potential energy surface (PES) in Figure 5.4 shows that the combination of the
vibronic coupling constant and the low energy gap is sufficient to induce the instability of the D6h
geometry of the HFB
1
B2u state with respect to distortions along the e2u modes. Lopez et al
concluded that the small energy gap (0.42 eV) between the
1
B2u and
1
E1g states in the FC region
results from the perfluoro effect, implying that the Stokes shift, results from the perfluoro effect
via a pseudo-Jahn−Teller distortion in the
1
B2u (ππ*) state.
3‡
Further motion along this e2u mode
220

from the
1
B2u minimum geometry leads to Dewar-HFB, implying that this coupling encourages
Dewar-HFB formation.

5.4.4. Transient Absorption Spectroscopy
With the multiconfigurational calculations reproducing the experimental photophysical
processes in HFB, the photochemical isomerization of HFB to Dewar-HFB is now considered.
Figure 5.5 shows the experimental transient absorption spectrum of a 200 mM solution of HFB in
ethanol recorded following excitation with a 255 nm UV laser pulse (580 nJ). This wavelength
was selected to match the 255 nm lamps employed in gram scale reactors for these syntheses.
4

The transient absorption spectrum is rendered at a series of pump−probe delays denoted by the
color-coordinated legend. The stick spectrum represents the computed gas-phase excited state
absorption spectrum at the 𝑆𝑆 1
minimum energy geometry. Each absorption peak in the stick
spectrum was convoluted with a Gaussian function with an (arbitrary) full-width half-max of 0.2

Figure 5.5 – Transient absorption spectra for 200 mM HFB in ethanol. The main panel shows
transient absorption spectra at pump−probe delays indicated by the color-coded legend (in
picoseconds) following photoexcitation at 255 nm (4.86 eV). The inset shows the time
evolution on a linear scale from −1 to 1 ps pump−probe delay and a logarithmic scale from 1
to 1000 ps delays. For sticks and simulated bands see the text.
221

eV, which yielded the simulated absorption spectrum shown by the red trace. The corresponding
contour plot and the full range of pump−probe delays are rendered in Figure S5.4 and Table S5.1.
By analyzing the ground state absorption spectrum (Figure S5.12 and Table S5.4),
approximately 3% of the molecules excited at 255 nm are estimated to be promoted to the 𝑆𝑆 2

1
E1g
state and 97% promoted to the 𝑆𝑆 1

1
B2u state. The transient spectrum contains two excited-state
absorption features longer than 400 nm, overlapping a (negative) stimulated emission band
peaking at 375 nm (compare to fluorescence band in Figure 5.3). The excited-state absorption
from 𝑆𝑆 1
features a broad absorption peak centered at 550 nm that immediately rises and a sharper,
more intense absorption centered at 440 nm, which rises to 90% of its maximum intensity
instantaneously and then subsequently rises to 100% over ∼10 ps. A similar spectrum is afforded
by excitation to 266 nm (Figure S5.3) which has 100% contribution to 𝑆𝑆 1
.
A broad 550 nm absorption has previously been assigned to a biradical form of HFB based
upon then-current CNDO/S calculations.
16
We prefer the assignment to an excited state 𝑆𝑆 1
because
stimulated emission is simultaneously observed. Our new multiconfigurational calculations predict
an excited-state spectrum based on the S1 minimum geometry. This spectrum shows two dominant
absorption peaks at 540 and 590 nm. These peaks correspond, in C2V symmetry, to the 𝑆𝑆 1
(
1
B1) →
𝑆𝑆 3
(
1
B1) and 𝑆𝑆 1
(
1
B1) → 𝑆𝑆 4
(
1
A1) transitions, respectively. These two transitions from the 𝑆𝑆 1

minimum overlap to form one peak and better explain the broad absorption feature at 550 nm in
the experimental spectrum. On the basis of the conditions in our experiment, we calculate ∼0.04%
of HFB molecules are photoexcited (Appendix A.3). The XMS-CASPT2/aug-cc-pVDZ
calculations predicted oscillator strengths of 0.0014 and 0.0057 for the 𝑆𝑆 1
(
1
B1) → 𝑆𝑆 3
(
1
B1) and 𝑆𝑆 1

(
1
B1) → 𝑆𝑆 4
(
1
A1), respectively. From these values, and the estimated concentration of excited- state
molecules, an excited state molar absorptivity of 500, and in turn, an excited state optical density
222

of 0.5 mOD is predicted. Both values agree well with the recorded transient absorption
measurement. This calculation can be reproduced from Equation A2.4. However, an assignment
for the 440 nm peak is more difficult because it is absent from the computed spectrum.
Turning to the 440 nm spectral feature, a rising absorption could be assigned to a new
absorbing species formed by a bimolecular process. (Figure S5.1 shows that the sharpening of the
440 nm feature cannot be explained simply as a weakening subtractive effect from the nearby
stimulated emission peak narrowing following the cooling of the solvent environment.) A rising
band centered at this same wavelength has previously been assigned by Varma et al. to both cation
and anion absorption bands using CNDO/S calculations.
16
Varma thus speculated that charge
transfer occurs between HFB pairs during the 10 ps rise-time, consistent with the diffusion time
scale of charge transfer partners finding one another. Matt Bain from our group performed EOM-
IP- CCSD/aug-cc-pVDZ and EOM-EA-CCSD/aug-cc-pVDZ calculations of the isolated cation
and anion absorption bands, respectively, which predict that within the 350−650 nm spectral range
of the TA experiment, the cation has a single strong absorption centered at 385 nm (3.23 eV) with
an oscillator strength of 0.13. In contrast, the anion has three strong absorption bands at 483 nm
(2.57 eV), 445 nm (2.78 eV), and 368 nm (3.37 eV) with oscillator strengths of 0.19, 0.14 and
0.077, respectively. These oscillator strengths are ∼10
4
stronger than the excited state absorption
bands. If correct, they must belie a correspondingly lower concentration of charge transfer contact
pairs to yield optical densities on the same order as the 𝑆𝑆 1
excited state absorption. The anion band
predicted at 445 nm provides an explanation for the 440 nm band in the TA spectrum. However,
the conspicuous lack of a corresponding cation peak at 386 nm – which would also need to show
the same 10 ps rise time – brings this assignment into question. Moreover, transient absorption
data recorded in nonpolar cyclohexane, shown in Figure S5.2 and Figure S5.5, shows no
223

diminishment or shifting of the 440 nm band, as one would expect when a polar environment does
not stabilize charged species.
Excimer formation may explain the time-delayed 440 nm absorption band.
29
Such
rearrangement from a ground-state T–shape to an excited-state slip-stacked arrangement of two
aromatic rings occurs on a 10−18 ps time scale in preformed dispersion-bound dimers of benzene
in the gas phase.
30
However, in dilute solutions, like those studied here, the dimer formation time
scale should increase because of the additional diffusion time required for ring association. The
formation time should also scale with HFB concentration; the cyclohexane data in Figure S5.2 and
Figure S5.5 rises on a comparable time scale despite HFB being 4 times more dilute (Figure S5.6).
Concentration dependent TA spectra reveal no diminishing of the 440 nm despite a ~40x decrease
in concentration (Figure S5.8) as one would expect if this were a dimer band. The assignment of
this peak, therefore, remains somewhat elusive.
The inset panel in Figure 5.5 shows the spectral intensity at two wavelengths, 440 nm
(blue) and 550 nm (red), as a function of pump−probe time delay. The time axis is split with the
−1 to 1 ps region rendered linearly, while the 1 to 1000 ps region is plotted logarithmically in time.
The majority of the excited-state signal is formed within the experimental instrument response
function (IRF) and persists for a lifetime significantly greater than the window of the experiment
(1 ns). Although the intensities of the features fall over this time scale, the general spectral shape
does not change indicating that all the dynamics observed within the TA window occur on the
same singlet state.
We noticed that rapidly damped oscillations appear at both probe wavelengths; these
oscillations in the ESA intensity over the first picosecond are fit to a convolution of a damped sine
wave and an exponential decay. A rendering of this fit and the resultant constants can be found in
224

Figure S5.6 and Table S5.2, respectively, along with the model’s fully functional form. The
oscillation frequency is ∼110 cm
−1
, which most closely matches the lowest frequency normal
mode ν19 (computed as 157 cm
−1
in the ground state) – one of the two vibrational modes implicated
in the vibrationally resolved absorption to the
1
B2u state. 93 cm
−1
oscillations with a longer
damping time ( ∼1 ps compared to ∼360 fs seen here) were also observed in the ion-detected
pump−probe experiments of Temps for gasphase HFB and assigned to the same motion.
18

There is no signal in the TA spectra that can be assigned to the Dewar-HFB photoproduct.
We would expect Dewar-HFB product to absorb wavelengths much shorter than 350 nm, the edge
of the experimental probe window. The maximum transient absorption signal of 0.5 mOD is weak
relative to typical molecules measured with this technique. We ascribe this to the weak excited
state absorption strengths (1000 M
−1
cm
−1
), possibly compounded with the low excited state
population.

Figure 5.6 – Transient absorption spectra of benzene (green) and HFB (blue) taken 1 ps after
excitation with 266 nm. The concentrations of benzene and HFB were ∼4 M and 400 mM,
respectively. These spectra were measured at 1 ps to avoid the rise time of the benzene excimer
and intermolecular dynamics.
225

To cross check that a significant fraction of excited state population has not branched at
earlier delays than captured in the current ∼50 fs resolution study, we compared it to the excited
state evolution of benzene. The excited state of benzene has no ultrafast deactivation pathways,
and the quantum yield of the photoproduct channel is <1%.
31
Thus, picosecond-scale excited state
signals in benzene should be attributable to the full excited state population. Using the same
computational method as applied to HFB, the calculated excited-state molar absorptivity of
benzene is 200 M
−1
cm
−1
. This is comparable with HFB at 500 M
−1
cm
−1
allowing estimation of
the HFB excited state population if the data is recorded under identical conditions. A comparison
of the TA spectra of benzene (green) and HFB (blue) recorded back-to- back in a single
experimental run is shown in Figure 5.6. Both of the transient spectra shown were recorded at a
pump−probe delay time of 1 ps to avoid interference either by solvent two- photon absorption at
early times or benzene excimer formation at later times. The HFB and benzene solutions were
prepared with identical ground state optical densities to ensure the same number of excited state
molecules were formed. We measured very similar transient absorbances. This indicates that,
similar to benzene, the TA signal captures the majority of the HFB molecules excited and rules
out the possibility of a significant bifurcated flux back to the ground state, or to Dewar-HFB
formation, at early times.

226

5.4.5. Potential Energy Surface
To rationalize the long excited state lifetime, the low quantum yield, and chemoselectivity
of the HFB 4π-electrocyclic isomerization, Lopez et al
3‡
computed the potential energy surface
along two independent degrees of freedom toward Dewar-HFB (Figure 5.7). The reaction
coordinates are the out-of-plane angles of two carbon atoms that form the new carbon−carbon σ
bond in Dewar-HFB. The resulting PESs of the ground and
1
B2u excited states is shown in Figure
5.7, along with the geometries of several critical points in the reaction pathway.

Figure 5.7 – A 2D cut through the 𝑆𝑆 0

1
A1g (blue) and 𝑆𝑆 1

1
B2u (red) gas-phase potential energy
surfaces. The two coordinates correspond to the angle of the two para carbons, indicated by the
orange and green arrows, rising out of the plane defined by the other four carbons. Motion
diagonally across the PES from the D6h minimum in the left-hand corner (A) to the right-hand
corner corresponds to both carbons rising symmetrically out of the plane to form the bridgehead
bond and yield the Dewar-HFB isomer (B). Here both carbons are out of plane by 76°. Partial
motion along the diagonal, where carbons are out-of-plane by 8° and corresponds to the
minimum energy geometry of the
1
B2u state (C). Motion along only one axis, where only one
carbon rises out of plane, leads to the MECI at 65° (D). From this MECI, internal conversion
to the ground state leads exclusively to the D6h minimum (A). However, from the MECI motion
of the other carbon is not degeneracy breaking until 40°. Molecules which pass through the CI
at the upper portion of the seam are on a portion of the ground state surface which bifurcates.
One trajectory leading back to (A) while the other trajectory leads to the Dewar form (B).
227

The PES shows that both the 𝑆𝑆 0
and 𝑆𝑆 1
surfaces feature a broad, energetic minimum
centered at 0° and 8°, respectively, though the S1 surface is significantly flatter. The wide range of
available motion in this flat region around the 𝑆𝑆 1
minimum likely accounts for the fast 360 fs
dampening of the oscillations in the transient absorption data. The surfaces also show that there is
an 𝑆𝑆 0
/𝑆𝑆 1
crossing seam, which lies ≥1.4 eV above the 𝑆𝑆 1
minimum geometry (structure C). This
seam corresponds to one out-of-plane angle of >65°, and the 𝑆𝑆 0
/𝑆𝑆 1
degeneracy persists for
geometries where the opposing out-of-plane angle is <40°. The geometry of the minimum-energy
crossing point (MECP) on this crossing seam is shown as structure D. The two out-of-plane angles
in the MECP geometry are 76° and 0°, analogous to the 𝑆𝑆 0
/𝑆𝑆 1
MECP in benzene.
28

Nonadiabatic relaxation from the 𝑆𝑆 1
state requires at least 1.4 eV more energy than the 𝑆𝑆 1

minimum, suggesting that this process would proceed slowly – if at all. This is in line with the TA
measurements that indicate a long excited-state lifetime. Time Correlated Single Photon Counting
(TCSPC) measurements at 255 nm, shown in Figure S5.11 and Table S5, confirm this lifetime
approaches 2.29 ns. In fact, the lifetime does not change with photon energy (Figure S5.11). The
ground-state surface near the crossing seam also shows that relaxation via geometries near the
MECP results in a steep path back to the HFB reactant rather than Dewar-HFB. Thus, the crossing
seam’s productive region lies at higher energies, where the crossing geometry more closely
resembles the photoproduct. This region of the seam is up to 0.3 eV higher than the MECP,
indicating this process is less favorable than relaxation back to the reactant explaining the low
quantum yield for the photoisomerization. It is not however immediately evident from the potential
energy surface why hexafluoro-benzvalene is not an observed HFB photoproduct after passing
through the MECP in HFB as it is in the analogous benzene. The branching spaces which define
the MECPs in HFB and benzene were analyzed using the first-order approximation of Galván et
228

al by J. Cox and M. Bain.
32
The reaction pathway to benzvalene from the MECP lies along a
vector in the branching space, labeled ĝ. The relative slope of the PES along +ĝ and −ĝ is shown
in Figure 5.8, along with the MECP geometry and atom-wise contributions to the ĝ vector. The
ground state reaction pathway that leads away from the MECP toward −ĝ leads to a reformation
of the D6h minimum benzene geometry, while +ĝ leads to the formation of benzvalene.

In benzene’s branching space, shown by the dashed lines in Figure 5.8, the excited state
surface slopes downward toward the MECP from both + ĝ and −ĝ directions, forming an energetic
funnel. The ground state surface shows that relaxation away from the MECP along both +ĝ and
−ĝ directions is also downhill, although −ĝ is overwhelmingly preferred. These surfaces suggest
that benzene molecules nonadiabatically relaxing through this MECP will primarily reform the
reactant benzene, with a small benzvalene yield. This analysis agrees well with the observed
benzvalene isomerization quantum yield of 1%.
10
The branching space of HFB is shifted
compared to benzene, such that both of the interacting surfaces increase in energy from −ĝ to +ĝ.
This subtle difference has profound effects on the photochemistry through this branching space.
The excited-state surface no longer forms a funnel. Instead, a downhill pathway exists in the

Figure 5.8 – (Left) Minimum Energy CI geometry in HFB with atomwise contributions to the
ĝ vector. (Right) Cut of the ground (blue) and first excited (red) states along the branching space
ĝ vector for HFB (solid) and benzene (dashed).
229

excited state along the −ĝ. The ground state surface of HFB is sloped away from +ĝ, which further
disfavors the formation of hexafluoro-benzvalene. The shift in the PES slope is a product of the
PJT coupling between the
1
B2u and
1
E1g states, which lowers the energy of the
1
B2u minimum,
thereby increasing the curvature of the PES. Thus, while benzvalene is an observed photochemical
product for benzene, that reaction pathway is destabilized in HFB, effectively shutting down this
reaction channel and enforcing the observed chemoselectivity in the (overall inefficient)
photoisomerization.
5.4.6. Simulated Photochemical Dynamics
Static calculations of the PES provide important mechanistic insight but lack information
on the dynamic molecular evolution or the relative rates of competing excited-state processes. To
understand the time-resolved excited state structural dynamics of HFB, an ensemble of gas-phase
nonadiabatic dynamics trajectories were computed using the surface-hopping including arbitrary
coupling method as implemented in the SHARC suite of programs in Cox et al.
3‡, 33–35
Initial
conditions for these trajectories were sampled from the Wigner distribution and initialized in the
𝑆𝑆 1
and 𝑆𝑆 2
electronic states. For the trajectories initialized on the 𝑆𝑆 2
state, the
1
E1g degeneracy was
broken by the Wigner sampling of nonplanar geometries and the initial state was taken as the lower
of the previously degenerate pair. Each trajectory was propagated for 1 ps with a 0.5 fs time step.
A set of 100 trajectories were propagated from the 𝑆𝑆 1
state and 400 trajectories from the 𝑆𝑆 2
state.
For the 𝑆𝑆 1
trajectories, only three reach the ground state in the simulation time, making a
statistical analysis of the rate constant unreliable. However, all three of these trajectories reform
the HFB reactant upon reaching the ground state. These observations are again in line with the
experimentally observed long-lived excited state and low reaction quantum yield.
230

Considering the behavior of HFB molecules excited to the 𝑆𝑆 2
(
1
E1g) is informative. The
time evolution of the state populations for the ensemble of 𝑆𝑆 2
trajectories is shown in the top panel
in Figure 5.9. 397 (99.25%) of these trajectories nonadiabatically relaxed to the 𝑆𝑆 1
state, and 189
(47.25%) went on to reach the ground state within the 1 ps simulation window. This is due to the
additional kinetic energy provided by initial excitation to a higher-energy electronic state. Kinetic
fitting of these populations yielded two time constants. The 𝑆𝑆 2
− 𝑆𝑆 1
relaxation process occured with
a time constant of 129 ± 5 fs, and the 𝑆𝑆 1
− 𝑆𝑆 0
relaxation fitted to a time constant of 1180 ± 90 fs.
The initial 𝑆𝑆 2
− 𝑆𝑆 1
relaxation agreed well with the previously determined values 0.10−0.54 ps based
on femtosecond time-resolved ion yield
measurements for gas- phase HFB.
18

Trajectories deposited on the 𝑆𝑆 1

surface at the 𝑆𝑆 2
/ 𝑆𝑆 1
conical intersection
were deposited near the 𝑆𝑆 1
minimum but
with more kinetic energy than those
prepared in the FC region and
consequently relaxed faster than the 2.29
ns time constant observed in TCSPC
following excitation to 𝑆𝑆 1
. This result is
consistent with the decreasing 𝑆𝑆 2

fluorescence quantum yield observed in
Figure S5.9.
The bottom of Figure 5.9 shows
the geometric evolution of the 400

Figure 5.9 – (Top) Time evolution of the
population of each electronic state. Green trace
shows S2, orange trace shows S1, and black trace
shows S0. (Bottom) Time evolution of reactive
C−C bond distance for full ensemble of 400
trajectories.
231

trajectories initialized on 𝑆𝑆 2
throughout the simulation time. Of the 189 trajectories which reach
the ground state surface in the simulation time, 96.8% (183) of these followed a reaction channel
that reverts to the HFB reactant, similar to the 𝑆𝑆 1
-initialized trajectories. However, the remaining
6 trajectories followed a different reaction channel, which did not reform HFB and instead formed
the experimentally observed photoproduct, Dewar-HFB. These trajectories appeared in the figure
at a reactive bond length of <2 Å. While statistical analysis of such a small number of trajectories
is unreliable, these data do show that the isomerization reaction can occur directly from
nonadiabatic relaxation to the ground state through higher-energy geometries in the 𝑆𝑆 1
/𝑆𝑆 0
crossing
seam.
Further, none of the photochemically productive trajectories pass through a ground state
biradical intermediate, as was proposed by Varma et al.
16
Instead, the softening of the
1
B2u surface
along the e2u vibrational modes promotes molecular vibrations that resemble the Dewar-HFB
photoproduct while simultaneously rendering the competing photochemical path- way
unfavorable. This is sufficient to produce a small yield of Dewar-HFB from an ensemble of
photoexcited HFB molecules, on the order of 1%.
It would appear from our analysis that under the conditions typically used in synthesis (254
nm, 80 W, 8 h), the rather small fraction of molecules born on 𝑆𝑆 2
may be responsible for the scant
photoproduct formed. Why this photochemical preparation works at all under synthesis conditions
would seem to come down to the product accumulating slowly over long irradiation times due to
the kinetic stability of Dewar-HFB and its lack of absorption in the near-UV region. Irradiation
studies of HFB and HFB with norbornene (an olefin) at 266 nm was performed and will be
discussed in the following chapter. TCSPC Stern-Volmer analysis reveal quenching with
norbornene at 266 nm, leading to a reaction quantum yield of ~7% where photoproduct formation
232

is not sensitive to fluorescence; therefore it was coupled with
19
F NMR. Irradiation revealed a
wealth of photoproducts from the reaction of HFB with norbornene to yield a set of polymeric
photoproducts.
It is thus clear that the strength of hexafluorobenzene derivatives as synthetic reagents, in
contrast to hydrogenated analogs, is not just the presence of a pathway toward Dewar- HFB on S1
but a strict chemoselective photoproduct formation. In the demonstration of this reaction by Burns
et al., the ratio of photons to molecular reagents was 168:1 and achieved a 41% synthetic yield.
4

Interpreting this through the lens of these results shows that although most absorption events result
in the reformation of the D6h HFB, provided there is a small but exclusive pathway leading to
Dewar-HFB, constant upcycling of the HFB by using a high molar excess of photons still yields
an effective reaction.
5.5. Conclusion
HFB and its derivatives are photochemically reactive moieties employed synthetically for
their photochemical chemoselective isomerization to Dewar-HFB and its analogs. Ultrafast
transient absorption spectroscopy, multireference calculations, and nonadiabatic molecular
dynamics simulations have been employed to understand the chemoselectivity, which makes this
highly inefficient reaction still synthetically useful. Experimental and simulated absorption spectra
show that the perfluoro effect is responsible for lowering the energy of the
1
E1g ( πσ*) state to S2,
0.42 eV above the
1
B2u ( ππ*) state. As a result, these states are then vibronically coupled along e2u
vibrational modes, distorting the excited state minimum geometry toward the Dewar-HFB product.
This perturbation of the potential energy surface creates a barrier toward hexafluoro-benzvalene,
which enforces the Dewar-HFB chemo- selectivity despite it still being a disfavored channel when
compared with reformation of the D6h reactant. These results demonstrate that the internal 4π
233

electrocyclic ring closing is promoted by the electronegative perfluoro effect. Molecular dynamics
simulations of photoexcitation to higher electronic states suggest that this would further increase
the reaction rate toward Dewar-HFB formation, but this would require experimental confirmation
to determine the magnitude of yield enhancement. This work serves as a foundation for
understanding the photochemistry of highly fluorinated systems, paving the way for the targeted
synthesis of high- value strained materials.

234

5.6. References
(1) Watson, W. H.; McMordie, W. C.; Lands, L. G. Polymerization of Alkynes by Ziegler‐type
Catalyst. J Polym Sci 1961, 55 (161), 137–144. https://doi.org/10.1002/pol.1961.1205516114.
(2) Shirakawa, H.; Louis, E. J.; MacDiarmid, A. G.; Chiang, C. K.; Heeger, A. J. Synthesis of
Electrically Conducting Organic Polymers: Halogen Derivatives of Polyacetylene, (CH) x. J
Chem Soc Chem Commun 1977, 0 (16), 578–580. https://doi.org/10.1039/c39770000578.
(3) Gould, G. L.; Eswara, V.; Trifu, R. M.; Castner, D. G. Polydifluoroacetylene,
Polychlorofluoroacetylene, and Polydichloroacetylene. J Am Chem Soc 1999, 121 (15), 3781–
3782. https://doi.org/10.1021/ja983840a.
(4) Boswell, B. R.; Mansson, C. M. F.; Cox, J. M.; Jin, Z.; Romaniuk, J. A. H.; Lindquist, K. P.;
Cegelski, L.; Xia, Y.; Lopez, S. A.; Burns, N. Z. Mechanochemical Synthesis of an Elusive
Fluorinated Polyacetylene. Nat Chem 2021, 13 (1), 41–46. https://doi.org/10.1038/s41557-020-
00608-8.
(5) Bryce-Smith, D.; Gilbert, A.; Orger, B. H. Photoaddition of Cis -Cyclo-Octene to
Hexafluorobenzene. J Chem Soc D Chem Commun 1969, 0 (14), 800b–8802.
https://doi.org/10.1039/c2969000800b.
(6) Šket, B.; Zupan, M. Stereochemistry of 2 + 2 Photoaddition of Cyclopentene to
Hexafluorobenzene. J Chem Soc Chem Commun 1977, 0 (11), 365–366.
https://doi.org/10.1039/c39770000365.
(7) Zupan, M.; Šket, B. Photochemistry of Fluorosubstituted Aromatic and Heteroaromatic
Molecules. Israel J Chem 1978, 17 (1‐2), 92–99. https://doi.org/10.1002/ijch.197800012.
(8) Lemal, D. M. Hexafluorobenzene Photochemistry: Wellspring of Fluorocarbon Structures.
Accounts Chem Res 2001, 34 (8), 662–671. https://doi.org/10.1021/ar960057j.
(9) Haller, I. Photoisomerization of Hexafluorobenzene. J Am Chem Soc 1966, 88 (9), 2070–
2071. https://doi.org/10.1021/ja00961a055.
(10) Wilzbach, K. E.; Ritscher, J. S.; Kaplan, L. Benzvalene, the Tricyclic Valence Isomer of
Benzene. J Am Chem Soc 1967, 89 (4), 1031–1032. https://doi.org/10.1021/ja00980a053.
(11) Proceedings of the Chemical Society. October 1957. Proc Chem Soc 1957, 0 (October),
273–300. https://doi.org/10.1039/ps9570000273.
(12) Zgierski, M. Z.; Fujiwara, T.; Lim, E. C. Photophysics of Aromatic Molecules with Low-
Lying Πσ* States: Fluorinated Benzenes. J Chem Phys 2005, 122 (14), 144312.
https://doi.org/10.1063/1.1873752.
235

(13) Clark, I. D.; Frost, D. C. A Study of the Energy Levels in Benzene and Some
Fluorobenzenes by Photoelectron Spectroscopy. J Am Chem Soc 1967, 89 (2), 244–247.
https://doi.org/10.1021/ja00978a011.
(14) Camaggi, G.; Gozzo, F.; Cevidalli, G. Para -Bonded Isomers of Fluoroaromatic
Compounds. Chem Commun Lond 1966, 0 (10), 313–314.
https://doi.org/10.1039/c19660000313.
(15) Haller, I. Kinetics and Mechanism of the Photochemical Valence Tautomerization of
Hexafluorobenzene. J Chem Phys 1967, 47 (3), 1117–1125. https://doi.org/10.1063/1.1711996.
(16) L., S., J.; H., H., A.; O., V., C. A. G. Picosecond Spectroscopy in Study of the Photoinduced
Isomerization of Hexafluorobenzene. Laser Chem 1986, 6 (5), 333–347.
https://doi.org/10.1155/lc.6.333.
(17) Cordes, T.; Herzog, T. T.; Malkmus, S.; Draxler, S.; Brust, T.; DiGirolamo, J. A.; Lees, W.
J.; Braun, M. Wavelength and Solvent Independent Photochemistry: The Electrocyclic Ring-
Closure of Indolylfulgides. Photochem Photobiol Sci 2009, 8 (4), 528–534.
https://doi.org/10.1039/b817627b.
(18) Studzinski, H.; Zhang, S.; Wang, Y.; Temps, F. Ultrafast Nonradiative Dynamics in
Electronically Excited Hexafluorobenzene by Femtosecond Time-Resolved Mass Spectrometry.
J Chem Phys 2008, 128 (16), 164314. https://doi.org/10.1063/1.2907859.
(19) Motch, C.; Giuliani, A.; Delwiche, J.; Limão-Vieira, P.; Mason, N. J.; Hoffmann, S. V.;
Hubin-Franskin, M.-J. Electronic Structure of Hexafluorobenzene by High-Resolution Vacuum
Ultraviolet Photo-Absorption and He(I) Photoelectron Spectroscopy. Chem Phys 2006, 328 (1–
3), 183–189. https://doi.org/10.1016/j.chemphys.2006.05.032.
(20) Holland, D. M. P.; Shaw, D. A.; Stener, M.; Decleva, P. A Study of the Valence Shell
Electronic Structure of Hexafluorobenzene Using Photoabsorption and Photoelectron
Spectroscopy, and TDDFT Calculations. J Phys B Atomic Mol Opt Phys 2009, 42 (24), 245201.
https://doi.org/10.1088/0953-4075/42/24/245201.
(21) Philis, J.; Bolovinos, A.; Andritsopoulos, G.; Pantos, E.; Tsekeris, P. A Comparison of the
Absorption Spectra of the Fluorobenzenes and Benzene in the Region 4.5-9.5 EV. J Phys B
Atomic Mol Phys 1999, 14 (19), 3621. https://doi.org/10.1088/0022-3700/14/19/013.
(22) Mondal, T.; Reddy, S. R.; Mahapatra, S. Photophysics of Fluorinated Benzene. III.
Hexafluorobenzene. J Chem Phys 2012, 137 (5), 054311. https://doi.org/10.1063/1.4739502.
(23) Tauber, M. J.; Mathies, R. A.; Chen, X.; Bradforth, S. E. Flowing Liquid Sample Jet for
Resonance Raman and Ultrafast Optical Spectroscopy. Rev Sci Instrum 2003, 74 (11), 4958–
4960. https://doi.org/10.1063/1.1614874.
236

(24) Taniguchi, M.; Lindsey, J. S. Database of Absorption and Fluorescence Spectra of >300
Common Compounds for Use in PhotochemCAD. Photochem Photobiol 2018, 94 (2), 290–327.
https://doi.org/10.1111/php.12860.
(25) Berlman, I. B. Handbook of Fluorescence Spectra of Aromatic Molecules; Academic Press;
Academic Press, 1971.
(26) Gozem, S.; Krylov, A. I. The EzSpectra Suite: An Easy‐to‐use Toolkit for Spectroscopy
Modeling. Wiley Interdiscip Rev Comput Mol Sci 2022, 12 (2).
https://doi.org/10.1002/wcms.1546.
(27) Goodman, L.; Ozkabak, A. G.; Thakur, S. N. A Benchmark Vibrational Potential Surface:
Ground-State Benzene. J Phys Chem 1991, 95 (23), 9044–9058.
https://doi.org/10.1021/j100176a008.
(28) Palmer, I. J.; Ragazos, I. N.; Bernardi, F.; Olivucci, M.; Robb, M. A. An MC-SCF Study of
the S1 and S2 Photochemical Reactions of Benzene. J Am Chem Soc 1993, 115 (2), 673–682.
https://doi.org/10.1021/ja00055a042.
(29) Nakashima, N.; Sumitani, M.; Ohmine, I.; Yoshihara, K. Nanosecond Laser Photolysis of
the Benzene Monomer and Eximer. J Chem Phys 1980, 72 (4), 2226–2230.
https://doi.org/10.1063/1.439465.
(30) Miyazaki, M.; Fujii, M. Real Time Observation of the Excimer Formation Dynamics of a
Gas Phase Benzene Dimer by Picosecond Pump–Probe Spectroscopy. Phys Chem Chem Phys
2015, 17 (39), 25989–25997. https://doi.org/10.1039/c5cp03010b.
(31) Thompson, A. L.; Martínez, T. J. Time-Resolved Photoelectron Spectroscopy from First
Principles: Excited State Dynamics of Benzene. Faraday Discuss 2011, 150 (0), 293–311.
https://doi.org/10.1039/c1fd00003a.
(32) Galván, I. Fdez.; Delcey, M. G.; Pedersen, T. B.; Aquilante, F.; Lindh, R. Analytical State-
Average Complete-Active-Space Self-Consistent Field Nonadiabatic Coupling Vectors:
Implementation with Density-Fitted Two-Electron Integrals and Application to Conical
Intersections. J Chem Theory Comput 2016, 12 (8), 3636–3653.
https://doi.org/10.1021/acs.jctc.6b00384.
(33) Mai, S.; Marquetand, P.; González, L. Nonadiabatic Dynamics: The SHARC Approach.
Wiley Interdiscip Rev Comput Mol Sci 2018, 8 (6), e1370. https://doi.org/10.1002/wcms.1370.
(34) Richter, M.; Marquetand, P.; González-Vázquez, J.; Sola, I.; González, L. SHARC: Ab
Initio Molecular Dynamics with Surface Hopping in the Adiabatic Representation Including
Arbitrary Couplings. J Chem Theory Comput 2011, 7 (5), 1253–1258.
https://doi.org/10.1021/ct1007394.
237

(35) Mai, S.; Marquetand, P.; González, L.; Richter, M.; Hendl, M. Surface Hopping Including
Arbitrary Couplings - Program Package for Non-Adiabatic Dynamics. 2021.

238

5.7. Supplement and Appendix
5.7.1. Temperature Dependent Fluorescence

Figure S5.1 – Temperature dependent fluorescence of 71 μM HFB in ethanol. At each
temperature, the solvent fluorescence was also taken and subtracted from the HFB spectrum at
the same temperature. Both were excited at 255 nm. The impurity fluorescence is mainly
present in the ~300nm region which can introduce subtraction errors. No narrowing of the
emission band as a function of temperature is observed. We would therefore expect no
narrowing of the stimulated emission band to be observed in transient absorption experiments
as a result of cooling of the solvent local to the non-equilibrium HFB chromophore.

300 350 400 450 500
0
5000
10000
15000
FL Intensity (Counts/Sec)
Wavelength (nm)
23
o
C
30
o
C
40
o
C
47
o
C
56
o
C
239

5.7.2. psTA Data
5.7.2.1. HFB psTA in Cyclohexane

Figure S5.2 – Transient absorption spectrum for 255 nm photoexcited HFB in cyclohexane
(CHX) at a concentration of 50 mM. The main panel shows transient absorption spectra of HFB
following at pump-probe delays indicated by the color-coded legend following photoexcitation
at 255 nm. The inset panel shows the time evolution on a linear scale from -1 to 1 ps pump-
probe delays and then a logarithmic scale from 1 to 1000 ps delay for the probe wavelength
cuts at 435 nm and 550 nm. This experiment, in contrast to the experiment in Figure 5.5, was
performed in a 1 mm quartz flow cell therefore yielding a higher signal to noise ratio than
gravity jet experiments shown in Figure S5.3 and the main paper (using a ~100-micron path
length) at a cost of lower time resolution. The stimulated emission band (centered at 380 nm)
grows more negative over the course of the experimental time window.
240

5.7.2.2. HFB psTA in Ethanol, excitation at 266 nm

Figure S5.3 – TA spectra of 250 mM HFB in ethanol with excitation at 266 nm for two different
experimental runs and solutions. These spectra bear strong resemblance to the spectrum at 255
nm, Figure 5.5. This is expected as both pump wavelengths launch HFB into the 𝑆𝑆 1
state almost
entirely. (3% and 0% population at 255 and 266 nm excitation, respectively). Similar to 255
nm, the stimulated emission grows over the pump probe delay times. The band at 440 nm is the
strongest feature in the left hand spectrum but diminished in the right hand spectrum.

350 400 450 500 550 600 650
0
0.1
0.2
0.3
0.4
0.5
0
0.5
1
∆Abs (mOD)
Time (ps)
435 nm
550 nm
-1 0 1 10
1
10
2
10
3
∆Abs (mOD)
Wavelength (nm)
0.3 ps 100 ps
1 ps 500 ps
10 ps 930 ps

350 400 450 500 550 600 650
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0
0.5
1
1.5
∆Abs (mOD)
Time (ps)
435 nm
551 nm
10
3
10
2
1 0 -1
∆Abs (mOD)
Wavelength (nm)
0.4 ps 100 ps
1 ps 500 ps
10 ps 950 ps
241

5.7.2.3. Contour Plots of HFB

Figure S5.4 – TA contour plot (units mOD) of a 200 mM solution of HFB in ethanol excited
with 255 nm. The large optical density near time zero (t < 0.1 ps) is attributed to the 2PA of the
solvent, ethanol. Time points were binned in a similar way as described in section 5.3.5. Each
time point is a moving average of an 11-point mean.
242

5.7.2.4. psTA Data Time Delays

Figure S5.5 – TA contour plot (units mOD) of a 200 mM solution of HFB in CHX excited with
255 nm in a flow cell. Due to the large coherent signal from the solvent at zero time-delay,
made broader by the walk off introduced in the 1mm path length, pump-probe data before 350
fs is not plotted.
Table S5.1 – The delay points
used in transient absorption
experiment.
Start Stop Step
-2.5 -0.8 0.1
-0.75 -0.55 0.05
-0.5 -0.1 0.02
-0.09 0.75 0.01
0.8 2.5 0.05
2.75 10 0.25
12 100 2
110 300 10
330 1200 30

The first number listed, denoted Start, is the most negative number in the time file regime and
the second listed number Stop, is the most positive in that region. The third number is the step
size, Step. All values are in picoseconds.
243

5.7.2.5. Integrated Time Traces of psTA

𝜏𝜏𝑀𝑀𝑘𝑘 = 𝐴𝐴 𝑇𝑇 𝑀𝑀 ∙ � exp �
𝜎𝜎 2
2 𝜏𝜏 𝑇𝑇 𝑀𝑀 2
―
𝑘𝑘 ― 𝑘𝑘 0
𝜏𝜏 𝑇𝑇 𝑀𝑀 � ∙ � 1 ― 𝑏𝑏 𝑀𝑀 𝑓𝑓 �
𝜎𝜎 2
― 𝜏𝜏 𝑇𝑇 𝑀𝑀 ( 𝑘𝑘 ― 𝑘𝑘 0
)
√2 𝜎𝜎 𝜏𝜏 𝑇𝑇 𝑀𝑀 � � �
+ 𝐴𝐴 𝑜𝑜 𝑃𝑃𝑐𝑐
∙ sin(2 𝜋𝜋 ( 𝑘𝑘 ― 𝑘𝑘 0
) 𝑓𝑓 + 𝜙𝜙 ) ∙ exp � ―
𝑘𝑘 ― 𝑘𝑘 0
𝜏𝜏 𝑟𝑟𝑎𝑎 𝑝𝑝 𝑝𝑝 �
+ 𝐴𝐴 2𝑃𝑃 𝑀𝑀 ∙ exp � ―
( 𝑡𝑡 ― 𝑡𝑡 0
)
2
2 𝜎𝜎 2
� Equation S5.1

The 2PA signal was modelled with a gaussian with the width of 𝜎𝜎 . The time constant for
the entire excited state lifetime, 𝜏𝜏 𝑇𝑇 𝑀𝑀 , was fixed at its value from the 266nm pumped TCSPC. The

Figure S5.6 – HFB transient absorption signal recorded with a 255 nm excitation pulse,
integrated across all probe wavelengths. The early time oscillations are accounted for in the
temporal fit. Each of the troughs and valleys for the oscillations line up across all probe colors
indicating there is no wavelength dependent phase shift. The large spike at zero delay is
attributed to the large coherent 2PA artefact from the solvent, ethanol. The three contributing
signals are the pump- probe signal, the early time oscillations and the 2PA coherent artefact.
The time slice is fit with Equation S5.1 with the parameters shown in Table S5.2:
Table S5.2 – Fitting parameters for the TA time slices.
𝐴𝐴 𝑇𝑇 𝑀𝑀
( 𝑀𝑀 𝑚𝑚 𝐶𝐶 )
𝐴𝐴 2 𝑃𝑃 𝑀𝑀
( 𝑀𝑀 𝑚𝑚 𝐶𝐶 )
𝐴𝐴 𝑜𝑜 𝑃𝑃 𝑐𝑐
(mOD)
𝜏𝜏 𝑇𝑇 𝑀𝑀
( 𝑀𝑀𝑠𝑠 )
𝜏𝜏 𝑟𝑟𝑎𝑎𝑝𝑝𝑝𝑝

(𝑓𝑓 𝑠𝑠 )
𝑓𝑓
( 𝑝𝑝 𝑠𝑠 − 1
)
𝜙𝜙
( 𝑀𝑀 𝑀𝑀𝑀𝑀 )
𝑘𝑘 0

(𝑓𝑓 𝑠𝑠 )
𝜎𝜎
(𝑓𝑓 𝑠𝑠 )
𝑓𝑓 𝑀𝑀 𝑏𝑏 𝑓𝑓 𝑜𝑜𝑓𝑓 𝑜𝑜 𝑠𝑠 𝑘𝑘
( 𝑘𝑘 𝑀𝑀 ―1
)
0.178 0.96 0.1 2.29 360 3.1 1.4 50 46 96

244

variables, 𝜙𝜙 , 𝑘𝑘 0
, and 𝜎𝜎 , are closely correlated but 𝑘𝑘 0
and 𝜎𝜎 were determined from a separate solvent-
only scan where no 𝜙𝜙 dependence exists.
245

Figure S5.7 – Time slices of HFB at the probe wavelengths of 435 nm, bottom panel and 550
nm, top panel, respectively. In green and red are the raw data for ethanol and CHX, respectively.
The fits are in green and blue for ethanol and CHX, respectively. The data and fits were
manually offset to separate them from each other. The y-values for the transient signal are
arbitrary. Data was plotted after 1 ps to remove any contributions from the large IRF
contributions (CHX and EtOH) or any oscillatory motion in the transient absorption intensity
(EtOH).
The top panel with probe wavelength at 550 nm, is assigned to be the ESA of the
1
B2u so its
rise time is assumed to be on the order of the IRF, instantaneous. These curves were fit with a
single exponential decay, fixed to be 2.29 ns, obtained from the TCSPC measurement. The
bottom panel shows evidence of possible bimolecular excimer formation. There is a rise time
associated which can be seen in both of the contour plots for the peak rising at 435 nm. The
435 nm slices were fit with the same decay as the 550 nm bands, but now with the inclusion of
a rise time.
𝛥𝛥 𝐴𝐴 4 3 5 𝑛𝑛 𝑝𝑝 = 𝐴𝐴 𝑟𝑟𝑒𝑒 𝑐𝑐 𝑏𝑏 𝑘𝑘𝑝𝑝 � ―
𝑡𝑡 𝜏𝜏 𝑑𝑑 𝑏𝑏𝑟𝑟 𝑑𝑑 𝑑𝑑 � + 𝐴𝐴 𝑟𝑟 𝑎𝑎 𝑃𝑃𝑒𝑒
� 1 ― exp � ―
𝑡𝑡 𝜏𝜏 𝑟𝑟𝑟𝑟 𝑃𝑃 𝑏𝑏 �� Equation S5.2

For 550 nm, 𝐴𝐴 𝑟𝑟 𝑎𝑎 𝑃𝑃𝑒𝑒
= 0. The ethanol experiment was fit with a 𝜏𝜏 𝑟𝑟 𝑎𝑎 𝑃𝑃𝑒𝑒
of 2.2 ps and the CHX
experiment was fit with a rise of 𝜏𝜏 𝑟𝑟 𝑎𝑎 𝑃𝑃𝑒𝑒
of 2.7 ps. In other words, the ESA rise in ethanol is 1.2
times faster than in CHX. But this is not a factor of four which would be expected the four-fold
greater concentration (EtOH – 200 mM and CHX – 50 mM). If this band represents excimers,
there must be dimers or larger clusters of HFB pre-existing in the solution, and this timescale
represents geometric rearrangement of the rings toward a sandwich geometry.
246

5.7.2.6. Concentration Dependent Transient Absorption

Figure S5.8 – TA of HFB in ethanol with excitation at 253 nm for left) 30 mM and right) 5.6
mM. The 30 mM experiment was performed in a variable pathlength flow cell with the
pathlength set to 1 mm. No diminishing of the 440 nm band is observed as one would expect
for an excimer or dimer band for a ~10x decrease in concentration. The 5.6 mM experiment
was performed in a rectangular flow cell with pathlength of 6 mm. In fact, in the 5.6 mM, only
3% of the 200 mM HFB presented in section 5.4.4, experiences an increase in the 440 nm
relative to the 550 nm band. Both experiments were performed with a pump spotsize of ~200
µm. The probe had a spotsize of 40-80 µm, depending on probe color. Therefore, it is possible
that the overlap of pump and probe was not sufficient to ensure no spectral distortion.

Essentially, the 440 nm does not disappear in concentrations that are ~40x smaller than the TA
presented in the main discussion in 5.4.4 and therefore we do not assign the 440 nm feature to
an excimer or dimer absorption.

350 400 450 500 550 600
0
1
2
3
4
5
∆Abs (mOD)
Wavelength (nm)
0.51 ps
1 ps
10 ps
100 ps
990 ps
30 mM

350 400 450 500 550 600 650
0
0.2
0.4
0.6
0.8
1
∆Abs (mOD)
Wavelength (nm)
1 ps
2.5 ps
10 ps
100 ps
1000 ps
5.6 mM
247

5.7.3. Time-Correlated Single Photon Counting

𝑘𝑘 ( 𝑘𝑘 ) = ∫ 𝑘𝑘𝑅𝑅 𝜏𝜏 ( 𝑘𝑘 ′
) ∙ 𝐴𝐴 𝑏𝑏 𝑘𝑘𝑝𝑝 � ―
𝑡𝑡 ― 𝑡𝑡 ′
𝜏𝜏 𝑑𝑑 𝑏𝑏𝑟𝑟 𝑑𝑑 𝑑𝑑 � 𝑀𝑀 𝑘𝑘 ′
𝑡𝑡
―∞
+ 𝑘𝑘 0
Equation S5.3
The decay was fit to Equation S5.2 with the parameters from Table S5.5. The fit is a
numerical convolution of the measured IRF with a single exponential decay and as such, does
not use a Gaussian IRF.
To determine the lifetime of HFB, TCSPC experiments were performed as the TA is limited to
delays of ~1.5 ns. The emission wavelength detection was set at 370 nm with a bandwidth of 16
nm. The instrument response of the experiment using scattered pump light (green) and the
convolution of the IRF with a single exponential fit (red). The black dotted lines are the limits of
fitting. The variable, 𝜏𝜏 𝑟𝑟𝑒𝑒 𝑐𝑐𝑎𝑎 𝑑𝑑 is the fluorescence lifetime used in the exponential fit. The small
departures from the fit in the photon counts between ~1 and 8 ns are artifacts of the instrumental
system.

Figure S5.9 – Fluorescence lifetime of HFB (blue) measured by time correlated single photon
counting (TCSPC) following excitation at 266 nm.
248

5.7.4. Wavelength Dependence of HFB Fluorescence Lifetime
In section 5.4.6, we saw evidence of excitation wavelength dependent photochemical
dynamics where high energy photons put HFB into its higher energy states and lead to a dramatic
decrease in lifetime. A possible start of this exploration of this phenomenon begins analyzing the
fluorescence quantum yield (flQY, Φ
𝑜𝑜 𝑓𝑓 ) of HFB as a function of wavelength denoted Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 𝑒𝑒𝑒𝑒
).
Three key pieces of information are required to obtain Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 𝑒𝑒𝑒𝑒
): absorption spectrum � 𝐴𝐴 𝑏𝑏𝑠𝑠 ( 𝜆𝜆 ) � ,
a fluorescence excitation spectrum � 𝐸𝐸 𝑘𝑘 ( 𝜆𝜆 ) � , and the fluorescence quantum yield at a particular
wavelength � Φ( 𝜆𝜆 𝑒𝑒𝑒𝑒
= 𝜆𝜆 𝑟𝑟 𝑒𝑒 𝑜𝑜 ) � . The absorption spectrum determines the amount of light absorbed
per unit wavelength. The excitation spectrum determines the amount of light emitted per unit
wavelength. The ratio of these two spectra is then arbitrary. To remedy this, the known
fluorescence quantum yield of HFB was used to fix the relative ratio of the absorption and
excitation spectra. The absorption and excitation spectra were normalized at 𝜆𝜆 𝑒𝑒𝑒𝑒
= 𝜆𝜆 𝑟𝑟 𝑒𝑒𝑜𝑜
. Then, the
excitation spectrum was divided, point by point after interpolation to a common wavelength axis,
by the absorption spectrum and scaled to Φ( 𝜆𝜆 𝑒𝑒𝑒𝑒
= 𝜆𝜆 𝑟𝑟 𝑒𝑒 𝑜𝑜 ) via Equation S5.4.
Table S5.3 – Fitting parameters for the TCSPC decay.
Parameter Value Confidence
𝐴𝐴 ( 𝐴𝐴 𝑜𝑜 𝐴𝐴𝑀𝑀𝑘𝑘 𝑠𝑠 ) 1510 ± 40
𝜏𝜏 𝑟𝑟𝑒𝑒 𝑐𝑐𝑎𝑎 𝑑𝑑 ( 𝑀𝑀𝑠𝑠 ) 2.29 ± 0.04
𝑘𝑘 0
( 𝐶𝐶𝑏𝑏 𝑘𝑘𝑀𝑀 𝑏𝑏 ) ( 𝐴𝐴 𝑜𝑜 𝐴𝐴𝑀𝑀𝑘𝑘 𝑠𝑠 ) 1.72 ± 1
𝐴𝐴 𝑀𝑀 𝑘𝑘𝑘𝑘 𝐵𝐵 𝑀𝑀 𝑜𝑜 𝐴𝐴𝑀𝑀𝑀𝑀 ( 𝑘𝑘𝑅𝑅 𝜏𝜏 ) ( 𝐴𝐴 𝑜𝑜 𝐴𝐴𝑀𝑀𝑘𝑘 𝑠𝑠 ) 0.5 ± 0 (fixed)

249

Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 ) =
𝐸𝐸 𝑒𝑒 ( 𝜆𝜆 )
𝐸𝐸 𝑒𝑒 � 𝜆𝜆 𝑏𝑏𝑥𝑥
= 𝜆𝜆 𝑟𝑟 𝑏𝑏𝑜𝑜
�
∙
𝑀𝑀𝑃𝑃 𝑃𝑃 � 𝜆𝜆 𝑏𝑏𝑥𝑥
= 𝜆𝜆 𝑟𝑟 𝑏𝑏𝑜𝑜
�
𝑀𝑀𝑃𝑃 𝑃𝑃 ( 𝜆𝜆 )
∙ Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 𝑒𝑒𝑒𝑒 = 𝜆𝜆 𝑟𝑟 𝑒𝑒𝑜𝑜 ) Equation S5.4
The first and second terms are the normalized excitation and absorption spectra. Equation S5.4
gives the ratio of light emitted as a function of light absorbed as a function of excitation
wavelength, similar to Equation 3.1, now with the Φ
𝑜𝑜 𝑓𝑓 being wavelength resolved.
To provide further understanding, we turn to HFB in ethanol for which the same process
was repeated. Obtaining absorption and excitation spectra were straightforward. The Φ
𝑜𝑜 𝑓𝑓 obtained
from a previous Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 𝑏𝑏 𝑘𝑘 = 𝜆𝜆 𝑀𝑀 𝑏𝑏𝑓𝑓
), referenced to the value at 270 nm, with a value of 3.4%. There is
a marked 10-fold increase in fluorescence ranging from the shortest to the longest wavelengths of
the absorption spectrum.
The concentration of the sample was set to be 0.1 OD at 270 nm when performing these
experiments to match the tryptophan value. However, due to the shape of the absorption spectrum,
the optical density rises far above 0.1 OD for 𝜆𝜆 < 270 nm, which can lead to fluorescence
quenching in the excitation spectrum from inner filtering. To eliminate inner filtering effects, a
series of qualitative dilutions were performed on a solution. Here, a solution of HFB was prepared
at 1 mM and an excitation spectrum was taken. Then, roughly half the solution in the cuvette was

Figure S5.10 – Excitation spectra of HFB during qualitative half dilutions of HFB (a) raw and
(b) normalized to 230 nm.

200 220 240 260 280 300
0
1
2
3
4
5
6
Fl. Intensity (x10
7
counts)
Wavelength (nm)
Dilution #
1
2
3
4
5
6
7
8
9
increasing
dilution
(a)

200 220 240 260 280 300
0
0.2
0.4
0.6
0.8
1
Emission Intensity (norm.)
Wavelength (nm)
Dilution #
1 4 7
2 5 8
3 6 9
(b)
250

discarded and refilled with blank solvent to create a ~2x dilution. The excitation spectrum was
then taken. This was repeated 8 times total. The scans were stopped being taken at the 9
th
dilution
as the large excitation peak at 260 nm is due to solvent impurity emission and the spectrum
becomes distorted. The results of this are depicted in Figure S5.10. The error with respect to
absolute height is attributed to the non-analytical dilution technique, hence normalizing the
spectra. The normalized spectra indicate a relative drop of the 255 nm band with respect to the 230
nm band, reversely indicating a repression of the 230 nm band at high concentrations. As such, we
can say the higher energy was being attenuated due to inner filtering effects. External filtering
effects from reabsorption are considered to be minor due to the large Stokes shift in HFB. We will
explore this result here.
The molar absorptivity of HFB at 266 nm is 100 M
-1
cm
-1
and 700 M
-1
cm
-1
at 230 nm, a
1/7 ratio. The sample was prepared at 1 mM in a cuvette with 1 cm pathlength which gives 0.1 OD
at 266 nm. Avoiding inner filtering effects can be achieved by reducing absorption at a given
excitation wavelength to be less than 0.1 OD. Reducing the 230 nm band to have absorption at
or below 0.1 OD would remove the inner filtering effects. Hence why the 5
th
dilution (~1/16x)
spectrum is the first traces to indicate no longer loss of the 255 nm with respect to the 230 nm

Figure S5.11 – Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 𝑒𝑒𝑒𝑒
) of HFB in ethanol obtained from (a) current measurements and (b)
adapted from Lopez et al.
3‡
The ethanol excitation spectrum (dashed red) is overlayed in (a).

200 220 240 260 280
0
5
10
15
Absorption/Emission (arb. u.)
Wavelength (nm)
Absorption
HFB Excitation
EtOH Excitation
(a)
0
0.05
0.1
0.15
Φ
fl
(λ
ex
)
φ
fl
(λ
ex
)

200 220 240 260 280
0
5
10
15
Absorption/Emission (arb. u.)
Wavelength (nm)
Absorption
Excitation
0
0.05
0.1
0.15
Φ
fl
(λ)
φ
fl
(λ
ex
)
(b)
251

band. The 4
th
dilution (~1/8x) spectrum is
still showing incomplete repression, but
this can be attributed to the fact that the
actual dilution factor isn’t 1/8 but likely
less. We can see that the 2
nd
dilution band
is not actually fifty percent of the unity
band and is ~60%. For these reasons, the
1/16 dilution excitation spectrum was used
in the Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 𝑒𝑒𝑒𝑒
= 𝜆𝜆 ) calculation, divided
by the absorption spectrum and the ratio of
the two spectra normalized to Φ
𝑜𝑜 𝑓𝑓 -
(270 𝑀𝑀𝑀𝑀 ). The fluorescence quantum yield of HFB is displayed in Figure S5.11. The result is a
continuously increasing curve: at Φ( 𝜆𝜆 = 200 𝑀𝑀𝑀𝑀 )

~ 0.2% which rises to 7.5% out to 280 nm. The
result obtained currently matches the curve previously obtained from Lopez et al. The wavelength
region less than 210 nm was removed as the absorption spectrum strongly rises and the inner
filtering effect takes over, leading to a possible loss of emission and as such is removed from
calculating Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 ) for HFB.
This result informs us that the excited state is overwhelming deactivated by non-radiative
processes. Probing fluorescent lifetimes at varying wavelengths could further help explain this
trend. A Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 ) is defined as the ratio of the radiative rate compared to the sum of the radiative
and non-radiative rates. The first possibility is that the fluorescence lifetime changes as a function
of excitation wavelength. If the quantum yield drops as the excitation wavelength moves blue, the
fluorescence lifetime could also experience a similar drop. The higher energy photon should

Figure S5.12 – Normalized absorption spectrum
of HFB in ethanol (black) decomposed into a sum
(orange) of three gaussians (dashed). The
experimental pump spectra for the TCSPC
experiments are plotted as well. The 266 nm
spectrum was not recorded and so is represented
by a simulated Gaussian.

210 220 230 240 250 260 270 280
0
0.2
0.4
0.6
0.8
1
Absorption (norm.)
Wavelength (nm)
Absorption
S
1
S
2
S
3
Total
252

provide enough energy possibly to overcome the large barrier in the excited state onto a pathway
of photoproduct formation, as discussed in the main body of this chapter. This internal conversion
could be either to the HFB or Dewar-HFB ground states.

To further explore Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 ), fluorescence lifetimes were taken at four UV wavelengths,
spanning the region of 273 to 247 nm. Here, the quantum yield drops by a factor of ½ from 4% to
2%. If the fluorescence lifetime changes by the same 0.5 factor, this would be visible in TCSPC.
An absorption spectrum of HFB fit to a series of four gaussians representing the 𝑆𝑆 0
→ ( 𝑆𝑆 1
− 𝑆𝑆 3
)
transitions is displayed in Figure S5.12. Here, we disentangle each states contribution to the
absorption of HFB. For quantitative pump contributions, Table S5.4 can used. For wavelengths
273 and 266 nm, there is no appreciable contribution from any higher lying excited state.
Therefore, the experimental lifetimes from these two excitation wavelengths should be identical.
The Φ( 𝜆𝜆 ) drops from 4% to 3% from 266 nm to 273 nm. The quantum yield roughly doubles when
the pump wavelength is changed from 247 to 272 nm. A pump wavelength of 247 nm commits
48% of its excitation energy into 𝑆𝑆 2
(Table S5.4). The lower SNR of the 247 nm data is attributed

Figure S5.13 – TCPSC decay curves of HFB in ethanol at varying pump wavelengths with
detection at 370 nm for (a) the full trace and (b) early time region. The IRFs were taken for
266 nm and 247 nm.

253

to lower UV pump power (40 µW vs >300 µW for other pump frequencies) due to reduced
efficiency from the OPA when compared to the redder wavelengths.

The detection wavelength was set to 370 nm with 6 mm slit widths, corresponding to 4.5
nm bandwidth. Except for 266 nm, each decay trace was reconvoluted with the 247 nm IRF. Due
to UV absorption by the recollimation lens of the setup, the recollimation lens was removed for
each of the IRF decays. The normalized TCSPC decay traces for HFB are displayed in Figure
S5.13. The lifetime of HFB is 2.24 ± 0.01 𝑀𝑀𝑠𝑠 , regardless of pump wavelength. The second
scatter peak at ~ 1 ns in the 247 nm IRF decay was removed from fitting due to no visible
replication in the HFB decay curves. This artifact was treated as a relic of solely the IRF and
excluded from fitting the data.
Present in the data is the unique shape of the decay that occurs in the early times, before 1
ns. There is an apparent rise for every wavelength, but the rise is strongest in the 247 nm case
where the other cases have identical rise profiles. The 𝑆𝑆 1
state experiences significant relaxation
from the FC region down to the 𝑆𝑆 1
minimum, a discussion probed in the main section of the chapter
and Lopez et al.
3‡
The relaxation time of 𝑆𝑆 1
, as indicated by the TA of 255 and 266 nm, is roughly
10 ps, which is instrument limited for the TCSPC which has a 45 ps measured IRF here. The time
constant obtained for the decay time, 𝜏𝜏 2
, is on the order of 100 – 200 ps, an order of magnitude
different. The uncertainty on the amplitude fit parameter is the same as the value itself which
Table S5.4 – Absorption contributions to 𝑆𝑆 𝑛𝑛 states for HFB with varying pump wavelengths
λex (nm) 𝑆𝑆 1
𝑆𝑆 2
𝑆𝑆 3

273 1 0 0
266 0.998 0.002 0
252 0.84 0.16 0
247 0.52 0.48 0

254

leaves little confidence in this value, indicating the value can be zero. The 𝝉𝝉 𝟐𝟐 value is not a decay
value and should not be treated as such when it is in fact a rise time.
Therefore, instead of fitting these decay curves as a sum of decay exponentials, we apply
a sequential model with excitation leading to a hot state which decays to a relaxed excited state
which can then emit or non-radiatively decay. The model used is summarized in the following
sequential scheme.
𝑆𝑆 2
→ 𝑆𝑆 1
→ 𝑆𝑆 0

For the new fitting set, the HFB traces were fitted using a two-state model with PDP, the
homebuilt TA fitting software. Time zero was subtracted and the datasets were normalized. The
amplitudes were restricted to be always positive. The rise in the HFB TCSPC with subsequent
decay was fit to a sequential two compartment model. The first compartment is populated with
100% population with decay out of the second compartment. The 247 nm pump TCSPC reveals a
larger contribution from the 𝑆𝑆 1
∗
compartment, matching the larger pump donation of excited state
energy to the higher lying 𝑆𝑆 2
state. The 252 nm pump experiment does not show an increased
contribution of S. Even for 247 nm with a 48% population of 𝑆𝑆 2
, there was no visibly observed
other features when compared to 273 nm. The summary of the fitting is displayed in Table S5.5.
Here, the fit abstracted an early/fast rise time of ~180 ps and then the long time fluorescence decay
of 2.24 ns, regardless of excitation wavelength.
In applying this model to the TCSPC data, the result is displayed in Table S5.5. We find
that the major fluorescence decay time constant is insensitive to excitation energy changes over
the measured 0.5 eV. Therefore, the Φ
𝑜𝑜 𝑓𝑓 ( 𝜆𝜆 𝑒𝑒𝑒𝑒
) does not increase as a function wavelength due to
variations in fluorescence lifetimes. However, the TCSPC has an instrument response of ~22 ps.
It is therefore ultrafast decay pathway exists faster than TCSPC can detect but this is speculation.
255

Referring to the TA of HFB at 266/255 nm, we see a loss of SE from 10 to 100 ps, matching the
same timescale presented by the TCSPC. A corollary TA experiment with a faster IRF of ~100 fs
at the shorter wavelengths than 255 nm are not currently available and therefore it is
experimentally unknown if the full excited state is represented in the TCSPC. Excited state theory
trajectories launched on the 𝑆𝑆 2
surface demonstrate a 𝑆𝑆 2
→ 𝑆𝑆 1
relaxation of 130 fs with a 𝑆𝑆 1
→ 𝑆𝑆 1

lifetime of 1.2 ps (section 5.4.6). If the lifetime of the HFB molecules launched onto the 𝑆𝑆 1
surface
is greatly reduced, it is believed that the 𝑆𝑆 2
lifetime is within the IRF of the TCSPC and therefore
obscured. With 247 nm excitation, we are only capturing the 𝑆𝑆 1
population which would have a
lifetime similar at the longer wavelengths.

Table S5.5 – Summary of rate constants and amplitudes from sequential
model of HFB in ethanol at various pump wavelengths
𝜆𝜆 𝑝𝑝 (nm) 𝜏𝜏 𝑟𝑟 𝑎𝑎𝑃𝑃𝑒𝑒
(ns) 𝜏𝜏 𝑜𝑜 𝑓𝑓 (ns) A1 (%) A2 (%)
247 0.20 2.24 32 68
253 0.16 2.23 43 57
266 0.19 2.24 43 57
273 0.18 2.22 44 56

256

Chapter 6. Photochemistry of Hexafluorobenzene and Norbornene
6.1. Abstract
The [2+2] photocycloaddition of hexafluorobenzene (HFB) with norbornene (NB) was
explored via quenching from steady state fluorescence, fluorescence lifetime quenching via time-
correlated single photon counting (TCSPC) and irradiation studies with product analysis by
fluorine NMR. Theoretical work has suggested this reaction is concerted and enhanced by pre-
formed π interactions in the ground state. Lifetime quenching experiments revealed no early time
decay component (<22 ps) which would indicate static quenching, i.e., no equilibrium significantly
favoring pre-formed dimer for concentrations less than 500 mM. We therefore assign the reaction
of HFB with an olefin to be strictly diffusive and with at most an on-contact reaction efficiency of
~7%. Irradiation of solutions of HFB with NB, characterized by
19
F NMR revealed a myriad of
fluorinated photoproducts produced with the major photoproduct produced only ~2% of the
original HFB. No HFB-NB photoproduct is produced but Dewar-HFB formation occurs with a
~0.04% reaction quantum yield.
6.2. Introduction
Conjugated polymers have drawn significant attention as an attractive possibility for
organic electronics.
1,2
One simple possibility is polyacetylene (PA): the continuous chain of
alternating single and double bonds offers the possibility for essentially organic conducting wires
as an alternative to metals. Unfortunately, PA is exceedingly unstable, even under the mildest of
conditions and polymers of PA rapidly combust in open air and is moisture sensitive.
3,4
However,
upon fluorination to make polyfluoroacetylene (PFA), the stability dramatically increases.
5,6

Synthesizing PFA from difluoroacetylene is not viable due to reactivity of difluoroacetylene to
air and combustibility in addition to reduced chemoselectivity producing multiple fluorocarbons.
257

Burns and coworkers have found a synthetic method in which it is possible to generate polymers
of PFA from hexafluorobenzene (HFB) and Dewar-benzene after exposure to UV light and ring-
opening metathesis polymerization.
7
Here, it is detailed in Figure 6.1a.

F
F
F
F
F
F
+
F
F
F
F
F
F
h
ν
F
F F
F
F
F
h
ν
F F
F
F F
F
F F
F
F F
F
F
F
F
F F
F
F
F
F
F
F
F
+
F
F
F
F
F
F
h
ν
Norbornene
a)
b)
ROMP
force
h
ν
F
F F
F
F
F
HFB Dewar-benzene
1. 2. 3.
4.
5.
6.
3. 2. 1.
PFA
P1 P2
h
ν
F
F
F
F
F
F
Dewar-HFB
x
x
(NB)

Figure 6.1 – a) Reaction scheme leading to PFA. Intermolecular [2+2] photocycloaddition
between HFB and Dewar benzene followed by an additional intramolecular [2+2]
photocycloaddition to form the fluorinated ladderane. Then a ROMP is performed, and
mechanical force opens the ladderane into a polymer, PFA. b) Similar reaction scheme of HFB
with norbornene as to b). The full saturation of the backbone of the norbornene unit prevents
the photoproduct from undergoing ROMP.
258

Here, an equimolar solution of hexafluorobenzene (HFB) and Dewar benzene is mixed
together (1.). Upon light exposure from the 254 nm atomic line from a high power mercury lamp,
the HFB harvests the UV photons to produce a 𝑆𝑆 1
excited state. The excited state HFB then
encounters the olefin and the two can undergo a [2+2] photocycloaddition forming a single product
(2.). Further light excitation drives a similar intramolecular 4π-electrocyclic ring closing to form
a fluorinated ladderane (3./4.). Next, a ring-opening metathesis polymerization (ROMP) is
performed (5.) to generate a polymer with ladderane subunits. The final step involves applying
mechanical force to the ladderane set in which ultrasonic waves are delivered and the vibrations
cause the unstable ladderane to unzip down the rungs of the ladderane to generate a single polymer
chain of alternating PFA-PA units (6.). The HFB-olefin photocycloaddition achieves a reaction
yield of 41% but requires ~8 hours of irradiation with 20 W mercury lamps. In this chapter, we
provide a reaction quantum yield of HFB with an olefin, a metric for reaction efficiency.
Theoretical calculations using CASPT2 from Boswell et al depict HFB and Dewar-benzene
as existing in a preformed ground state geometry as a dispersion complex.
7
In the CASSCF
wavefunction, Boswell et al utilize D3 empirical dispersion correction of Grimme with Becke-
Johnson damping (D3BJ). Weak dynamic correlation from CASSCF required the use of D3BJ to
accurately produce the dispersion complex and improve the optimized geometries. This D3BJ
correction is required for the complex to exist by providing van der Waals and dispersion
interactions. In this chapter, we utilize experimental lifetime measurements to probe these
theoretical calculations.
As we saw from Cox et al
8
and in chapter 5, HFB also undergoes a chemoselective
unimolecular 4π-electrocyclic ring closing to Dewar-HFB (Figure 6.1b, lower pathway) but with
exceedingly poor reaction quantum yield ( Φ
𝑟𝑟 𝑒𝑒 𝑛𝑛 < 1%), requiring hours of irradiation from eight
259

20 W mercury lamps exciting each HFB tens of times to achieve appreciable synthetic yield. In
the 𝑆𝑆 1
excited state, the barrier to formation of Dewar-HFB is high and instead return from the S1
HFB back to ground state is preferred.
In this chapter, we examine a similar [2+2] photocycloaddition of HFB, this time with
norbornene (NB), Figure 6.1b. As this is an excited state reaction that occurs only in the 𝑆𝑆 1
state
with a fluorescent chromophore,
9
albeit weak ( Φ
𝑜𝑜 𝑓𝑓 ~3%, section 5.7.4), we can utilize Stern-
Volmer (SV) analysis as a metric for the reaction efficiency. Photoproduct formation or non-
radiative deactivation of HFB leads to a reduction in fluorescence intensity – so called quenching.
Fluorescence affords high sensitivity to minor changes in the spectral intensity which can be
probed, and fluorescence quenching analysis is routinely performed utilizing steady state
fluorescence. Here, several solutions are made with equivalent concentrations of fluorophores and
varying concentrations of quencher. SV analysis was performed for HFB with NB as a quencher
to determine the effectiveness of reaction. To circumvent HFB concentration errors, lifetime
quenching measurements of HFB with NB via time-correlated single photon counting (TCSPC)
were run as well. And we provide full error analysis to errors in quencher concentrations which
cannot be removed by doing lifetime measurements. Additionally, we provide spectroscopic
evidence in relation to the preformed dimer suggested by Boswell et al and the nature of ground
state complex of HFB and NB. Finally, we perform an irradiation test of HFB-only and HFB with
NB in conjunction with
19
F NMR to determine reaction yield as well as assign yielded
photoproducts.

260

6.3. Experimental
6.3.1. Sample Preparation for Stern-Volmer Experiments
The solutions of HFB (Fisher Scientific Acros, HEXAFLUOROBENZENE, 99%, 100 mL,
AC1205410) and NB (Sigma-Aldrich, BICYCLO(2.2.1)HEPT-2-ENE, 99%, N32407-500G)
prepared and used in the steady state fluorescence and lifetime were identical samples. To extract
the NB from the bottle from the vendor, the NB bottle was heated in a water bath to just above the
melting point of NB. This was then poured into smaller containers which was then used to obtain
NB solid after it had cooled and solidified. To determine the mass of the NB for the stock solutions,
the NB was put into a pre-weighed vial and the mass was determined via subtraction after weighing
the filled vial. Two stock solutions of 0.5 and 1 M were prepared. Analytical dilutions from the
0.5 M stock solution were used to prepare the majority of NB concentrations except for the 0.3
and 0.5 M NB solutions which used the 1 M NB stock solution. A 10 mM stock solution of HFB
was prepared wherein 2 mL aliquots of this solution was added to each solution of NB, except for
0.3 and 0.5M. In the 0.3 and 0.5 M NB solutions, the HFB was added via a 10 µL syringe
(Hamilton, gastight syringe, 1700 series). The small HFB volume that was used lead to a large
fluorescence intensity error bar in the corresponding steady state SV quenching experiment as the
fluorescence intensity is directly proportional to HFB concentration.
6.3.2. Sample Preparation for Irradiation and NMR Studies
HFB (Thermo Scientific Chemicals, Hexafluorobenzene, 99%, AC120541000, 100 mL)
and NB (TCI America, 2-Norbornene 99.0+%, N016625G, 25 g) used in the irradiation and NMR
studies were obtained from the vendor and used without further purification. The HFB and NB
used here were from different stock bottles than the chemicals used in the SV analysis. Deuterated
methanol (CD3OD, Cambridge Isotope Laboratories, Inc, DLM-24-10X1, 10x1g) was used as the
261

solvent for both the irradiation and NMR studies without further purification. To prepare the two
samples, 1. HFB and CD3OD and 2. HFB, NB and CD3OD, 2 g of CD3OD was used as solvent.
The amount of the HFB and NB was measured out gravimetrically using the difference between
an empty capped vial and the capped vial with the sample inside. Irradiation studies were
performed in Wilmad quartz NMR tubes. No photoproduct isolation or purification was
performed. Both
19
F and
1
H NMR of the samples before and after irradiation were performed on
a Varian 500S NMR with assistance from Shawn Wagner. Post processing included automatic
phase correction and baseline oscillations were corrected by using Bernstein polynomials, afforded
by the Mnova software.

6.3.3. Prediction of NMR of Irradiated Samples
Display of NMR structure and prediction of
19
F NMR was afforded by Mnova software by
Mestrelab and NMRPredict Desktop v1.16, respectively. This software utilizes prediction
algorithms, known as Ensemble NMR Prediction,
10
in combination with machine learning,
Increments and previous HOSE-code algorithms,
11
developed by Modgraph Consultants
12
to
produce the depicted spectra.
6.3.4. Steady State Fluorescence Measurements
All steady state fluorescence experiments were performed on a Horiba Jobin Yvon
FluoroMax-3 fluorometer using DataMax v2.2 software at 266 nm excitation with 6 nm slit widths
for both excitation and emission monochromators at 100 ms integration time. The spectra were
collected from 250 to 600 nm in 1 nm increments. The blue edge of the spectra was collected to
record the Rayleigh scatter as a check on the fluorimeter wavelength calibration. All spectra were
collected in a square 1 x 1 cm quartz cuvette (Starna Cells). The neat solvent fluorescence
262

spectrum was also taken which did not show appreciable emission in the collected region at 266
nm excitation. While NB does not absorb at 266 nm (Figure S6.1), samples available obtained
commercially have some absorption at 266 nm from impurities which leads to fluorescence
contamination. Therefore, NB-only sample fluorescence was taken for the two stock solutions of
0.5 and 1 M. The fluorescence of NB sample contamination was strongest < 350 nm. The spectra
for HFB-NB were integrated from 350 to 500 nm to avoid contributions from the impurities in the
NB sample.
6.3.5. HFB Lifetime Quenching Measurements via TCSPC
The Stern-Volmer lifetime quenching experiments were performed in the following way.
The UV excitation beam is driven by a 100 kHz pulse train of 50 fs 6 mJ pulses produced by a
Coherent RegA 9050 seeded by a Coherent Mira Seed, centered at 800 nm. About 25% of the
fundamental is directed to the probe path of a separate TA experiment, leaving 75% to pump the
UV TCSPC experiment.
The residual 800 nm light is directed into a Coherent 9450 Optical Parametric Amplifier
(OPA) tuned to produce 150 nJ pulses at 532 nm. These visible pulses are fed into a prism
dispersion compensation system consisting of a pair of UV-Fused Silica (UVFS) prisms. The
pulses are focused using a 10 cm focal length UVFS lens into a 0.1 mm BBO crystal cut for SHG
in the visible to generate 266 nm. The prism pair are optimized to maximize the conversion
efficiency into the UV rather than for shortest visible pulse length. This yielded 266 nm pulses
which were recollimated using a 10 cm focal length VUV grade CaF2 lens. The UV pulses were
then directed to the TCSPC setup and were focused at the interaction region using a 10 cm focal
length VUV grade CaF2.
263

The power was set such that the count rate of detected photons was lower than ∼1% of the
laser repetition rate. This was to reduce the Poissonian probability of two detected events per laser
shot, which can distort the timed-binned histogram. The monochromator used was a CVI Digikrom
CM-112 1/8 Monochromator. The monochromator was set at 370 nm, the peak of the HFB
emission spectrum. The slit widths were set at 16 nm by using static metal slits. Acquisition time
for this experiment was set to 20 minutes and a time range of 50 ns. The 266 nm power was set to
60 µW for HFB only sample and the power was ~2-3x higher power for the HFB-NB (110-180
µW).
6.3.6. Irradiation Setup of HFB with and without NB
The 266 nm used for irradiation of the samples was generated in the following way. The
laser used here is part of a larger setup in which the laser is used to pump a Magnitude Instruments
enVISion (State College, PA). An OEM Nd:YAG, 400 µJ energy per pulse laser, operating at 2
kHz laser and 532 nm from a Magnitude enVISion setup was employed. The output of the external
laser is sent into a beam splitter with the reflected beam directed into a timing silicon photodiode,
synchronized to the timing electronics of the instrument. The transmitted beam is sent into a
motorized waveplate-polarizer combo, housed in a single unit to form a laser power attenuator
(Atten, Eksma Optics, 990-0075-532M), transmission percentage controlled by the user, set to
100% for these experiments. The beam is then directed by 532 nm HRs into a lens (f = ~3 cm),
focusing the beam to decrease the spot size enough to pass through the 100-blade chopper
(ThorLabs). The beam diverges and passes through a pinhole (1 mm) and enters a recollimating
lens (f = ~3 cm). The beam is then sent through a f = 500 mm plano-convex lens (L1) to begin
focusing the beam, The 532 nm is then directed into a BBO crystal (Eksma, BBO-700 4HG@1064
264

nm, Type 1, unmounted, 7x7x6 mm, Θ = 47.6, Φ = 90 P/P coating @532/266 nm). The focus of
the 532 nm is set at ~4 cm after the BBO crystal to minimize crystal damage.
Powers of 266 nm of ~40-60 mW were generated for a conversion efficiency of 10-15%
from 400 mW of 532 nm as measured by a PM-2 head, Coherent Laser Mate II controller. The
residual 532 nm and generated 266 nm are then directed by two 266 nm dichroic mirrors, rejecting
the 532 nm, further directing into a f = 500 mm lens (L2), serving to recollimated the 266 nm. The
beam is then directed by a third 266 nm dichroic mirror into the enVISion instrument, into a fused
silica, right angle prism which directs the beam into the sample chamber.
At the sample stage, the 266 nm beam is ~5 x 1.5 mm, width x height. The beam was of
the same width as the NMR tube, with the assumption that all the 266 nm light transferred through
the NMR tube. For the HFB only sample, the tube was irradiated for 2 hours continuously at
powers of 266 of 41 mW. The HFB-NB sample was irradiated a total of four times, at intervals of
~40 minutes each. Due to solid photoproduct formation building up, the tube needed to be
translated vertically between each irradiation period. No visible photoproduct formation was seen
with the HFB only sample. A similar experiment was carried out on a sample with only NB at a
similar concentration and a neat CD3OD sample. Neither of these samples produced any visible
photoproduct either. The total number of photons of 266 nm impingent on the HFB only solution
was ~4.0·10
20
or ~0.66 millimoles of photons (or 0.6 mEinsteins). The total number of 266 nm
photons impingent on the HFB-NB solution was ~6.4·10
20
or ~1.1 millimoles of photons (or 1.1
mEinsteins).

265

6.4. Results and Discussion
6.4.1. Reaction Efficiency Through Steady State Measurements

The Stern-Volmer analysis of HFB with NB using steady state fluorescence is displayed in
Figure 6.2. As can be seen from Figure 6.2a, the presence of NB leads to a moderate decrease in
HFB fluorescence only at high concentrations of NB. The concentration of HFB in these
experiments was 1 mM and a 1:1 molar ratio of HFB to NB leads to a minor decrease in HFB
fluorescence. Even a 500:1 molar ratio leads to a drop of HFB fluorescence of only ~1.8x. The
emission of HFB does not monotonically decrease with increase in NB concentration as would be
expected for an SV plot but is rather scattered.
The dashed lines in Figure 6.2a correspond to emission from solutions with NB and
without any HFB. We can see it is from a different molecule from HFB due to a completely
different emission spectrum. NB, as a one-double bond cycloalkene,
13,14
does not absorb 266 nm
250 300 350 400 450 500
0
1
2
3
4
Fl. Intensity (Counts x 10
7
)
Wavelength (nm)
[NB] (M)
0 0.05 0.3
0.001 0.1 0.4
0.01 0.2 0.5
0.5 (no HFB)
1.0 (no HFB)
a)

0 0.1 0.2 0.3 0.4 0.5
0
1
2
3
4
I
0/
I
[NB] (M)
b)
Intercept 1 (fixed)
Slope

1.5 ± 0.4
Adj. R
2
0.9758

Figure 6.2 – a) Steady state fluorescence spectra of HFB at varying norbornene concentrations.
The semi-transparent region in grey was removed from integration due to emission from NB
impurities. The unscaled emission from the NB stock solutions is displayed as dashed lines. b)
corresponding Stern-Volmer plot. An inset of the low [NB] is displayed as well. The error bars
for all [NB] are present but only easily visible in the inset and for the maximum [NB], 0.5 M
NB, where they are ± 0.006 M. Error bars for b) are discussed in section 6.7.4. A best fit line
is applied in red with fitting parameters displayed in the legend.
266

as the absorption onset is < 220 nm (Figure S6.1a). Therefore, emission is not attributed to NB
itself. Additionally, emission from the solvent, ethanol, or impurities within the solvent are not
detected under similar conditions (Figure S6.4). We therefore assign the dashed lines in Figure
6.2a to impurities in the NB stock sample solution. The only fluorescent impurities encountered
in this chapter will be from the NB samples. To reduce the effect of impurity emission from the
NB samples, the emission integration to collect 𝑘𝑘 0
and 𝑘𝑘 was performed from 350 to 500 nm to
avoid the blue emission of the impurities from the NB samples (non-gray area in Figure 6.2a). A
scaled subtraction of the fluorescence arising from NB impurities does not lead to improvement in
the SV values.
Stern-Volmer theory is used in fluorescence quenching to obtain the quenching rate, 𝑘𝑘 𝑞𝑞 , a
measure of a quencher’s potency for fluorescence quenching. The quenching rate can be
decomposed into two parameters via Equation 6.1:
𝑘𝑘 𝑞𝑞 = 𝑓𝑓 𝑄𝑄 𝑘𝑘 0
Equation 6.1
where 𝑓𝑓 𝑄𝑄 and 𝑘𝑘 0
are the quenching efficiency and mutual diffusion rate , respectively. This
decomposition is performed as we can determine the amount of possible photoproduct formed on
each collision via 𝑓𝑓 𝑄𝑄 . While fluorescence quenching leads to a non-radiative relaxation to a ground
state species, SV cannot distinguish between quenching that returns the fluorophore to the initial
ground state or the quenching leads to photoproduct formation. Therefore, 𝑓𝑓 𝑄𝑄 assigns a maximum
value to the photoproduct on-contact quenching efficiency.
The mutual diffusion rate was calculated from the Smoluchowski equation:
13

𝑘𝑘 0
=
4 𝜋𝜋 𝐶𝐶 𝐴𝐴 1 0 0 0
� 𝑅𝑅 𝑇𝑇 + 𝑅𝑅 𝑄𝑄 � � 𝐶𝐶 𝑇𝑇 + 𝐶𝐶 𝑄𝑄 � Equation 6.2
where 𝐶𝐶 𝑀𝑀 is the Avogadro constant; 𝑅𝑅 𝑇𝑇 and 𝑅𝑅 𝑄𝑄 are the molecular radii (in cm) of the fluorophore
and quencher respectively; 𝐶𝐶 𝑇𝑇 and 𝐶𝐶 𝑄𝑄 are the diffusion coefficients (in cm
2
·s
-1
) for the fluorophore
267

and quencher, respectively. The factor of 1000 converts cm
3
to liters. This 𝑘𝑘 0
value used the
diffusion coefficient of benzene in ethanol ( 𝐶𝐶 𝑇𝑇 = 𝐶𝐶 𝑄𝑄 = 1.8 ∙ 10
− 5
𝑘𝑘 𝑀𝑀 2
𝑠𝑠 − 1
) and molecular size
of benzene ( 𝑅𝑅 𝑇𝑇 = 𝑅𝑅 𝑄𝑄 = 2 ∙ 10
− 8
𝑘𝑘𝑀𝑀 ) with the assumption that both HFB and NB can be treated
as benzene-like.
15,16
Here, we obtain a value of 𝑘𝑘 0
= 1.1 ∙ 10
1 0
𝐴𝐴 − 1
𝑠𝑠 − 1
. We turn to the Stern-
Volmer equation now:

𝑘𝑘 0
𝑘𝑘 � = 1 + 𝜏𝜏 0
∙ 𝑘𝑘 𝑞𝑞 [ 𝑄𝑄 ] = 1 + 𝜏𝜏 0
∙ 𝑓𝑓 𝑄𝑄 𝑘𝑘 0
[𝑄𝑄 ] Equation 6.3
where 𝜏𝜏 0
is the fluorescence lifetime of the fluorophore and [ 𝑄𝑄 ] is the quencher concentration.
From a linear fit of the SV plot, we can abstract the value for 𝑘𝑘 𝑞𝑞 . From Figure 6.2b, the slope
= 𝜏𝜏 0
∙ 𝑓𝑓 𝑄𝑄 𝑘𝑘 0
= 1.5 ± 0.4 𝐴𝐴 − 1
. The fluorescence lifetime of HFB in ethanol is known, 𝜏𝜏 0
=
2.24 ± 0.01 ns.
8
We now extract 𝑓𝑓 𝑄𝑄 and obtain 𝑓𝑓 𝑄𝑄 = 6 ± 2%. Further discussion of the
quenching rate dissection is in 6.7.3. Unfortunately, the SV plot using steady state fluorescence
leads to large error and ambiguity for a value that is already quite small.
A fluorophore and a quencher can exist in a ground state complex equilibrium wherein the
two are close in proximity and form a close contact pair via the following:
17

𝜏𝜏 + 𝑄𝑄 ⇌ ( 𝜏𝜏 ⋯ 𝑄𝑄 ) Equation 6.4
After excitation, the quencher deactivates the fluorophore instantly. This is referred to as static
quenching. The degree of ground state attraction can be described via Equation 6.5:
𝐾𝐾 𝑆𝑆 =
[ 𝑇𝑇 ⋯ 𝑄𝑄 ]
[ 𝑇𝑇 ][ 𝑄𝑄 ]
Equation 6.5
where 𝐾𝐾 𝑆𝑆 (subscript 𝑆𝑆 for static) is the equilibrium constant between 𝜏𝜏 , the fluorophore and 𝑄𝑄 , the
quencher. The amount of quenching due to static quenching then depends on the degree of
preformed complex. The remaining uncomplexed species are therefore fluorescent and defined
by the ratio, 𝑅𝑅 :
268

𝑅𝑅 =
[ 𝜏𝜏 ]
[𝜏𝜏 ]
0
� Equation 6.6
The total amount of fluorophore present is given by:
[ 𝜏𝜏 ]
0
= [ 𝜏𝜏 ] + [𝜏𝜏 − 𝑄𝑄 ] Equation 6.7
If we substitute into Equation 6.5, we obtain:
𝐾𝐾 𝑆𝑆 =
[ 𝑇𝑇 ]
0
−[ 𝑇𝑇 ]
[ 𝑇𝑇 ][ 𝑄𝑄 ]
=
[ 𝑇𝑇 ]
0
[ 𝑇𝑇 ][ 𝑄𝑄 ]
−
1
[ 𝑄𝑄 ]
Equation 6.8
We understand that the fluorophore concentrations are proportional to the fluorescence intensities;
rearrangement of Equation 6.8 achieves Equation 6.9:

𝑘𝑘 0
𝑘𝑘 � = 1 + 𝐾𝐾 𝑆𝑆 [ 𝑄𝑄 ] Equation 6.9
Wherein we achieve a similar SV equation, Equation 6.3, from dynamic quenching but with static
quenching replacing 𝑘𝑘 𝑞𝑞 with 𝐾𝐾 𝑆𝑆 . Upon excitation of the fluorophore, due to the proximity of the
quencher, the excited state fluorescence is instantly extinguished by the quencher without emission
from the fluorophore.
It is possible for a fluorophore-quencher system to undergo both static and dynamic
quenching at the same time. In these systems, some fraction of the fluorophores forms a ground
state complex, dictated by 𝐾𝐾 𝑆𝑆 . The fractional fluorescence remaining is given by the product of the
fluorophores not statically quenched and the fluorophores not dynamically quenched. Namely:
17

𝜏𝜏 𝜏𝜏 0
� = 𝑅𝑅 ∙
( 𝜏𝜏 0
)
− 1
( 𝜏𝜏 0
)
− 1
+ 𝑘𝑘 𝑞𝑞 [ 𝑄𝑄 ]
Equation 6.10
From Equation 6.6 and Equation 6.9, we see 𝑅𝑅 − 1
= 1 + 𝐾𝐾 𝑆𝑆 [ 𝑄𝑄 ] and substitution into Equation 6.10
with rearrangement, we achieve:

𝜏𝜏 0
𝜏𝜏 � = � 1 + 𝜏𝜏 0
∙ 𝑓𝑓 𝑄𝑄 𝑘𝑘 0
[𝑄𝑄 ]� ∙ (1 + 𝐾𝐾 𝑆𝑆 [ 𝑄𝑄 ]) Equation 6.11
Which is second order in [𝑄𝑄 ] which would manifest as an upward turning parabola in the SV plot.
7

269

The use of steady state fluorescence to determine if quenching is static, diffusive or both
can lead to ambiguities as a single SV plot can be described by a quadratic equation, requiring two
parameters to describe, 𝐾𝐾 𝑆𝑆 and 𝑘𝑘 𝑞𝑞 . Without prior knowledge and due to the symmetry of Equation
6.11, the values for 𝐾𝐾 𝑆𝑆 and 𝑓𝑓 𝑄𝑄 𝑘𝑘 0
can be interchanged and are ambiguous. Steady state SV is not
only quencher concentration dependent, but also fluorophore concentration dependent, which can
vary due to sampling error as each sample cuvette is made up. Without prior knowledge, these
ambiguities are further exacerbated if there are impurities.
Diffusion requires 1-10 ns, but static quenching can occur at a much faster rate. Due to
vastly different time scales between static and dynamic quenching, we can implement lifetime
measurements as these can definitively assign the difference between static and dynamic
quenching. Specific to the topic of static quenching, we wish to determine the degree of ground
state complex formation between HFB and NB as suggested by Boswell et al
7
which is possible
with lifetime measurements. Time-resolved fluorescence, in our case carried out by TCSPC, can
better inform us about the preformed dimer that theoretical conditions suggest may exists for this
reaction.
6.4.2. Reaction Efficiency Through Lifetime Quenching
In addition to steady state fluorescence, SV experiments can also be carried out by
observing an equivalent decrease in the fluorophore’s lifetime. Fluorescence quenching can be
quantified via lifetime measurements, where the error is far less sensitive to fluorophore
concentrations (for dilute solutions), and only sensitive to quencher concentration. We therefore
turn to SV analysis using fluorescence lifetimes for determination of HFB-olefin reaction yield.
270

First, a TCSPC experiment of 1 mM HFB (same concentration of HFB as it is in the HFB-
NB mixtures) in ethanol was run to determine 𝜏𝜏 0
for the SV analysis, obtaining 𝜏𝜏 0
= 2.24 ±
0.01 𝑀𝑀𝑠𝑠 (Figure 6.3a). This 𝜏𝜏 0
value agrees with what was obtained previously
8
(additionally,
section 5.7.4). Additionally, a TCSPC experiment of just the 0.5 M stock NB solution was run to
isolate the impurity lifetime(s) and the contributions (Figure 6.3b). A reminder – the NB itself is
not absorbing the 266 nm pump (Figure S6.1) – these TCSPC traces do not indicate the lifetime
of NB, rather the observed decay is from impurities in the NB stock solution. The identities of the
impurities are unknown, but this is of little relevance. We only need to know the lifetimes of the
impurities so as to isolate them from the HFB fluorescence. The trace was fit to three exponentials.
The values for the time constants were then fixed when fitting the mixture data, only floating the
amplitudes of the impurities.

Figure 6.3 – TCSPC trace of a) 1 mM HFB and b) 500 mM NB stock solutions in ethanol with
excitation at 266 nm under identical conditions. The time constants for the decays are displayed
on the figure with the amplitude percentages of the decays included after the semicolon.

0 10 20 30 40
g(×)10
0
g(×)10
1
g(×)10
2
g(×)10
3
g(×)10
4
Fl. Intensity (Counts)
Time (ns)
Decay
IRF
Fit
1 mM HFB a)
τ
0
= 2.24 ± 0.01 ns; 100%

0 10 20 30 40
g(×)10
0
g(×)10
1
g(×)10
2
g(×)10
3
g(×)10
4
Time (ns)
Decay
IRF
Fit
Fl. Intensity (Counts)
τ
1
= 0.15 ± 0.01 ns; 90%
τ
2
= 6.7 ± 0.2 ns; 5%
τ
3
= 1.6 ± 0.2 ns; 5%
0.500 M NB b)
271

The TCSPC was then performed on the same HFB-NB solutions that the steady state
measurements were performed, displayed in Figure 6.4a. The HFB lifetime trends downward
monotonically as the quencher concentration is increased. The largest contribution to the signal is
the HFB with smaller, long-lived components at later times attributed to NB sample impurities,
the same impurities as in the steady state measurements. The lifetimes of the NB impurities were
fixed during fitting ( 𝜏𝜏 1
, 𝜏𝜏 2
, 𝜏𝜏 3
in Figure 6.1b) while the relative amplitudes of the impurities,
lifetime (tabulated in Table 6.1) and amplitudes of the HFB component were left floating. Ideally,
the amplitudes of the NB impurities should have been set with the amplitude ratios set by the NB
stock solution experiment and the contributions of the impurities scaled by a single value. Fitting
the amplitudes independently was found to not have a large effect on the obtained HFB lifetime.
A linear trend is visible in the lifetimes as would also be expected for a purely diffusive quenching.

Figure 6.4 – a) The set of TCSPC decay traces of HFB with varying concentrations of
norbornene in ethanol. The [NB] of 0.2 to 0.5 are displayed with open symbols to improve
visibility with t > 15 ns. The HFB with [NB] = 0 M trace is the same trace from Figure 6.3a.
b) corresponding SV plot using lifetimes from a) where the fitting parameters for the linear fit
are displayed. The error bars for all [NB] are present but only easily visible in the inset and for
the maximum [NB], 0.5 M NB, where they are ± 0.006 M. Error bars for b) are discussed in
section 6.7.4. A best fit line is applied in red with fitting parameters displayed in the legend.

0 10 20 30 40
g(×)10
-4
g(×)10
-3
g(×)10
-2
g(×)10
-1
g(×)10
0
Time (ns)
[NB] (M)
0 0.2
0.01 0.3
0.05 0.4
0.1 0.5
Fluorescence Intensity (norm.)
Increasing
[NB]
a)
Impurity Fl.

0 0.1 0.2 0.3 0.4 0.5
1
1.5
2
τ
0
/
τ
[NB] (M)
b)
Intercept 1 (fixed)
Slope

2.53 ± 0.09
Adj. R
2
0.9989
272

We can also extract an additional piece of information
from the raw data. The TCSPC also disambiguates the
relative amount of static and dynamic quenching. Static
quenching occurs on a much faster time scale compared to
dynamic quenching as diffusion, by definition, is not
required. As evident in the TCSPC data, static quenching
would manifest as a sharp decrease in lifetime at early
time, with the ratio of static to dynamic diffusive
quenching dependent on the equilibrium constant for the ground state complex (Equation 6.10).
However, no such feature is observed and therefore, we state that there is little to no static
quenching in this concentration regime of the HFB-NB reaction. These NB concentrations mirror
the synthetic reaction conditions of the Burns group ([NB]~100 mM). However, it may be possible
that at high concentrations of NB or in solution where the olefin is the solvent (i.e., the liquids
cyclohexene or cyclopentene whereas NB is solid at room temperature), static quenching
dominates being forced by weak association of the olefin with HFB. An equivalent TCSPC
experiment with an olefin solvent should in the future be conducted to test this hypothesis.
It should be noted that although the fluorescent impurities of the NB were accounted for
during fitting to isolate the HFB lifetime, it is possible that the impurities are not entirely benign.
The unknown impurities could be leading to quenching of the HFB excited state but that is
currently not believed to be possible as the lifetimes of the NB impurities do not change as a
function of NB concentration as would be expected for diffusive quenching. A similar TCSPC set
of experiments after purification of each starting material and solvent should be performed to
eliminate this possibility.
Table 6.1 – HFB Lifetimes
from TCSPC of HFB-NB
mixtures
[NB] (M) 𝜏𝜏 (ns)
0.0 2.24 ± 0.01
0.01 2.17 ± 0.01
0.05 1.96 ± 0.01
0.1 1.75 ± 0.01
0.2 1.47 ± 0.01
0.3 1.18 ± 0.01
0.4 1.13 ± 0.01
0.5 1.01 ± 0.01

273

From slope of Figure 6.1b, 𝜏𝜏 0
, and 𝑘𝑘 0
, we abstract a value of 𝑓𝑓 𝑞𝑞 = 0.10 ± 0.01. These
results of 𝑘𝑘 𝑞𝑞 are in good agreement with the steady state fluorescence but with higher confidence
(10% vs 33% error). The 𝑓𝑓 𝑄𝑄 value obtained here distinguishes between the reaction efficiency
between quenched and unquenched fluorophores. SV does not distinguish between fluorescence
quenched via successful photoproduct formation and fluorescence quenched but no photoproduct
formation. The quenched species via photoproduct formation and quenched species via failed
reaction both return to the ground state without emission. To determine the actual reaction yield,
we perform high UV light fluence irradiation studies with photoproduct amount determined by
NMR.
6.4.3. Irradiation of HFB to Afford Dewar-HFB
HFB is an unusual organic compound due to the absence of any hydrogen atoms.
Therefore, standard
1
H NMR to determine structure is not applicable. In photochemistry involving
HFB, both the starting material and photoproducts only contain fluorine and carbon atoms, and
fluorine has only one nuclear spin state, which suggest the use of
19
F NMR as the best tool to
determine photoproduct formation. The reactant’s six fluorine atoms around the aromatic ring each
experience an identical chemical environment.
19
F NMR should therefore be selective and
sensitive to any addition or isomerization photoreaction that leads to a loss in symmetry around
the initial aromatic ring.
13
C NMR was performed on the same HFB irradiated solutions but was
not sensitive to new signals although a strong signal was associated with unreacted HFB from
19
F
NMR.
Irradiation studies of HFB begin with irradiating HFB in CD3OD in the absence of NB as
a control study to determine photoreactions that occur with only HFB. Additionally, the quantum
yield of HFB to Dewar-HFB can be determined. The results from the irradiation of an HFB-only
274

solution are shown in Figure 6.5a. The concentration of 40 mM was chosen to have >1.0 OD
absorption of HFB in a ~2.5 mm pathlength (circular NMR tube), meaning >90% of the light is
absorbed. No photoproduct isolation as a neat/solid compound was anticipated, so the irradiation
reactions were run in pure CD3OD to allow for direct comparison of before and after irradiated
samples. For a volume of 2 mL and 40 mM HFB, ~0.08 millimoles of HFB are present, which for
a one photon, one HFB molecule process such as the introduced photoreaction, ~0.08 millimoles
of photons are also required to turn every possible HFB molecule into a photoproduct. As
calculated in section 6.3.6, ~1.1 millimoles of photons were impingent on the sample, meaning
~14 photons per HFB molecule. If the reaction quantum yield is significant, the photoreaction
should be driven to completion as we have a large excess of photons.
275

Now, we look at the
19
F NMR spectra before and after irradiation of the HFB only solution,
displayed in Figure 6.5. Most notably, the spectrum remains largely unchanged with a single
peak, assigned to HFB, staying as the dominant feature (Figure 6.5a). We obtain a chemical shift
value of -165.38 ppm and -163.39 ppm for HFB in CD3OD for the before and after irradiation
spectra, respectively. The reported literature value for HFB in CD3OD is -165.37,
18
in excellent
agreement with our experimental value for HFB. What is most evident is that the spectra between
the before and after spectra do not vary upon first inspection. The ratio of starting HFB to all

Figure 6.5 –
19
F NMR of 40 mM HFB in CD3OD displayed as the a) full spectrum and b) focus
on <0.5% intensity of a) with central HFB peak removed for visibility. c) Focus on the region
of Dewar-HFB. Peak areas are given by 𝐴𝐴 𝑇𝑇 1, 2
. Integration limits are given by the dotted black
lines. Red – before irradiation, blue – after irradiation. All spectra were normalized at the HFB
peak maximum in their respective spectra.

-120 -140 -160 -180 -200
0
0.2
0.4
0.6
0.8
1
Intensity (norm.)
Chemical Shift (ppm)
After Irradiation
Before Irradiation
-165.4 ppm
19
F NMR of HFB + NB in CD
3
OD
a)
F
F
F
F
F
F
A
HFB
= 5.5·10
-3

-120 -140 -160 -180 -200
-0.001
0
0.001
0.002
0.003
0.004
0.005
Intensity (norm.)
Chemical Shift (ppm)
After Irradiation
Before Irradiation
19
F NMR of HFB in CD
3
OD
b)

-123.2 -123.6-190.4 -190.8
0
0.0005
0.001
0.0015
0.002
F
2
F
1
F
1
F
2
F
1
F
1
Intensity (norm.)
Chemical Shift (ppm)
After Irradiation
Before Irradiation
19
F NMR of HFB in CD
3
OD
c)
A
F
1
= 1.6·10
-5
A
F
2
= 0.85·10
-5
276

formed photoproducts is nearly ~100%, indicating the low reaction yield of Dewar-HFB or any
photoproducts from HFB provided all fluorines are still seen and there exists no photoproducts
outside this spectral region.
A more interesting story presents itself when we focus on the region <0.5% of the NMR
signal intensity, in Figure 6.5b. Approximately 11 peaks become visible. Amongst these peaks
are two peaks associated with Dewar-HFB. Haller et al assigned the Dewar-HFB photoproduct to
having two
19
F NMR peaks with chemical shifts at -121.2 and -190.1 ppm with an intensity ratio
of 2:1.
19
The peaks most similar to these literature values are focused on in Figure 6.5c where we
obtain values of -123.4 ppm and -190.6 ppm with an area ratio of 1.9. The difference between the
ratios can be explained by a secondary, larger peak at -190.8 ppm contributing some intensity to
the -190.6 ppm peak, reducing the ratio. The literature spectra were taken with CFCl3 as a
reference standard. In CFCl3, HFB possesses an NMR shift of -164.9 ppm. Therefore, we apply a
-0.4 ppm shift to our peaks wherein the -190.6 ppm now matches the peak in the literature spectra.
However, the peak at -123.0 ppm does not match the literature value for the chemical shift. We
currently do not understand why this is the case.
Haller reported formation of no other photoproducts aside from Dewar-HFB in the gas
phase.
20
However, in the solution phase (methylcyclohexane, cyclohexane or ether) at varying
wavelengths (254-303 nm) multiple photoproducts were formed, the identities of, were not
revealed by Haller. However, it is known that they possess higher molecular weights, indicating
some form of polymerization from HFB. These results mirror our results presented here as Dewar-
HFB is not the only photoproduct present here. An equivalent gas phase photolysis and
isomerization of HFB was not performed here.
277

We now determine the amount of Dewar-HFB that is formed within this irradiation study.
Integration of the large HFB peak achieves an area of 4.3 ∙ 10
− 3
or 0.72 ∙ 10
− 3
area per fluorine.
Division of the area corresponding to the Dewar-HFB peak at -190.6 ppm per fluorine by the area
of the neat HFB peak at -165.4 ppm per fluorine we achieve 4.4 ∙ 10
− 3
or:
𝑅𝑅 𝑏𝑏 𝑀𝑀𝑘𝑘𝑘𝑘 𝑀𝑀𝑜𝑜 𝑀𝑀 𝑌𝑌 𝑀𝑀 𝑏𝑏𝑓𝑓𝑀𝑀 =
𝐴𝐴 𝑀𝑀 𝑏𝑏 𝑀𝑀 𝐷𝐷 𝑒𝑒 𝑒𝑒 𝑎𝑎𝑟𝑟
𝐴𝐴 𝑀𝑀 𝑏𝑏 𝑀𝑀 𝐶𝐶 𝑇𝑇𝐵𝐵
� Equation 6.12
We can translate this into a concentration of Dewar-HFB as we know the total
concentration of HFB to be 40 mM and the intensity of the HFB peak is normalized to 1. Therefore,
we obtain 170 µM Dewar-HFB. We can convert this into a total amount of molecules via
multiplication of the volume of the solution, 2 mL, and Avogadro’s constant to obtain the total
number of Dewar-HFB molecules. We then take the value for the number of photons from section
6.3.6. Division of the number of Dewar-HFB by the number of photons gives use the reaction
quantum yield, Φ
𝑟𝑟 𝑒𝑒 𝑛𝑛 . Together:
Φ
𝑟𝑟 𝑒𝑒 𝑛𝑛 =
𝐶𝐶 𝑃𝑃 𝐶𝐶 𝑑𝑑 𝑃𝑃𝑃𝑃 =
( 𝑅𝑅 𝑒𝑒 𝑛𝑛 𝑌𝑌 𝑎𝑎 𝑒𝑒𝑓𝑓 𝑟𝑟 ) ∙ 𝐶𝐶 𝑃𝑃 𝑏𝑏 𝑑𝑑 𝑟𝑟𝑏𝑏 ( 1 − 1 0
− 𝐴𝐴 ) ∙ 𝐶𝐶 𝑝𝑝 ℎ 𝑜𝑜 𝑏𝑏 𝑜𝑜𝑜𝑜 𝑃𝑃 𝑏𝑏𝑜𝑜 𝑏𝑏𝑑𝑑 𝑡𝑡 Equation 6.13
Φ
𝑟𝑟 𝑒𝑒 𝑛𝑛 =
�
𝐴𝐴 𝑟𝑟 𝑏𝑏 𝑑𝑑 𝑃𝑃 𝑃𝑃 𝐴𝐴 𝑟𝑟 𝑏𝑏 𝑑𝑑 0
� ∙ 𝐵𝐵 ∙ 𝑉𝑉 � 1 − 1 0
− 𝐴𝐴 � ∙
𝑃𝑃 ∙ 𝑏𝑏 ℎ ∙ 𝑟𝑟 ∙ 𝑁𝑁 𝐴𝐴 ∙ 𝜆𝜆 𝑏𝑏𝑥𝑥
Equation 6.14
where 𝐶𝐶 𝑃𝑃 is the number of photoproducts produced; 𝐶𝐶 𝑎𝑎 𝑃𝑃 𝑃𝑃 is the number of absorbed photons; 𝐴𝐴 𝑀𝑀𝑏𝑏𝑀𝑀
is the integrated area under the NMR peak assigned to the photoproduct after normalizing to the
reactant molecule’s NMR peak; 𝐴𝐴 is the concentration of the starting material; 𝑉𝑉 is the volume of
the NMR tube or sample tube; 𝐴𝐴 is the absorption of the sample at the irradiation wavelength;
𝐶𝐶 𝑝𝑝 ℎ 𝑜𝑜 𝑡𝑡 𝑜𝑜 𝑛𝑛 𝑃𝑃 is the number of photons impingent on the sample; 𝐶𝐶 is the power of the irradiating light;
𝑘𝑘 is the length of irradiation in time; ℎ is Planck’s constant; 𝑘𝑘 is the speed of light; 𝐶𝐶 𝑀𝑀 is Avagadro’s
number; and 𝜆𝜆 𝑒𝑒𝑒𝑒
is the excitation wavelength. Using Equation 6.14 reveals Φ
𝑟𝑟 𝑒𝑒 𝑛𝑛 = 0.06% for
278

Dewar-HFB formation from HFB at 266 nm excitation. This value is in good agreement with
previously reported <1% values for the Dewar-HFB formation.
8,19

While assignment of the remaining photoproducts is not performed, extraction of the total
amount of photoproducts formed in relation to the Dewar-HFB can be determined. First, we
integrate each of the remaining peaks in the post-irradiation
19
F NMR spectrum. We achieve an
area for the remaining photoproducts of 3.98 ∙10
-4
which translates into ~93% of the photoproducts
formed being unknown, indicating that the majority photoproducts formed in solution is not
Dewar-HFB.
6.4.4. Irradiation of HFB and NB to Afford Photoproducts
Now, we turn to an HFB-NB irradiated solution to determine the reaction quantum yield
of HFB with NB. Irradiation studies were performed on a solution of 40 mM HFB and 250 mM
NB in deuterated methanol (CD3OD) in a quartz NMR tube in a similar way to the HFB-only
solution. In a similar fashion to the HFB-only solution, the majority of HFB molecules appear to
not have been turned over (Figure 6.6a). It is only when features less than 2% intensity in the
19
F
NMR response are focused on (Figure 6.6b) that more is revealed about the photochemistry. We
now see that the irradiation did produce photoproducts but did not produce a single fluorinated
photoproduct but rather a myriad of differing photoproducts, the amount and identity of each is
not readily apparent without careful analysis. The production of photoproducts reveals a larger
amount of photoproducts present compared to the HFB-only irradiated sample upon the
introduction of another reagent, NB (Figure 6.6c).
Boswell et al obtained
19
F NMR of a similar compound to P2 (Figure 6.1b) in CDCl3
wherein the NB backbone is replaced with three fused biyclobutene rings (compound 13).
7
The
chemical environments of each fluorine do not vary between 13 from Boswell et al and P2 and we
279

therefore would expect a similar photoproduct in this experiment. For this compound, they
obtained chemical shift values of -119.5, -165.8, and -188.2 ppm. While there are two possible
peaks in the after irradiation spectra, the peak at -188.2 ppm is absent. Therefore, we state that we
do not observe any P2 photoproduct. Currently, this is the only known spectra of a fluorinated
ladderane. Therefore, to assign possible photoproducts, we must introduce a new assignment
method.

Figure 6.6 –
19
F NMR of 40 mM HFB and 250 mM NB in CD3OD displayed as the a) full
spectrum and b) focus on <2% intensity of a). Red – before irradiation, blue – after irradiation.
Both spectra were normalized to 1 at the HFB peak maximum of the before experiment. c)
19
F
NMR spectra of HFB-only (blue) and HFB with NB (red). The central HFB peak in panels b)
and c) was removed for visibility.

-120 -140 -160 -180 -200
0
0.2
0.4
0.6
0.8
1
Intensity (norm.)
Chemical Shift (ppm)
After Irradiation
Before Irradiation
19
F NMR of HFB + NB in CD
3
OD
a)

-120 -140 -160 -180 -200
0
0.005
0.01
0.015
0.02
Intensity (norm.)
Chemical Shift (ppm)
After Irradiation
Before Irradiation
19
F NMR of HFB + Norbornene in d-MeOH
b)

-120 -140 -160 -180 -200
-0.005
0
0.005
0.01
0.015
0.02
Intensity (norm.)
Chemical Shift (ppm)
HFB only
HFB + NB
c)
280

We now turn to the prediction software of Mnova, NMRPredict Desktop v1.16 and
Mestrelab Predictor to suggest
19
F NMR spectra for possible photoproducts, details in the
experimental section, section 6.3.3. For calibration processes, we begin by predicting the
19
F NMR
spectrum for pure HFB. We obtain a predicted value of -162.55 ppm for HFB in d-chloroform, in
comparison to an experimentally obtained -165.33 ppm in CD3OD, 3.2 ppm downfield from
experimental HFB. However, the solvent assumed in the prediction is d-chloroform rather than
CD3OD which is the experimental solvent. CD3OD shifts
19
F peaks 1-2 ppm upfield from CDCl3.
18

The Ensemble prediction software by Mnova for a set of 13 randomly assigned molecules has an
average error of 1.3 ppm for
13
C spectra.
10
Here, we therefore assign an error value to the
prediction algorithm of ±1 ppm for the
19
F NMR spectra here. Experimental NMR spectral peaks
different from predicted values by 1 ppm are therefore rejected as a possible assignment.

Figure 6.7 –
19
F NMR of HFB and norbornene in CD3OD after irradiation (blue) with predicted
photoproducts. The predicted
19
F NMR spectral lines for photoproducts are displayed as drop
lines with colors (black squares for P1, red circles for P2 and green triangles for P3), normalized
to 0.01 intensity. The corresponding fluorines are displayed above their predicted positions.

-120 -130 -140 -150 -160 -170 -180 -190 -200
0
0.005
0.01
0.015
0.02
Intensity (norm.)
Chemical Shift (ppm)
P1 P2 P3
F
1
F
2
F
3
F
4
F
5
F
6
F
1
F
2
F
3
F
4
F
5
F
6
F
1
F
2
F
3
F
4
F
5
F
6
F
1,6
F
1,6
F
1,6
F
2,5
F
2,5
F
2-5
F
3,4
281

We now utilize Ensemble to assign the after-irradiation spectra; first we attempt to assign
the most desired photoproducts (Figure 6.7). The first photoproduct, P1, is the product of the first
step in the published mechanism: an initial [2+2] photocycloaddition of HFB with NB. But the
failure of any of the NMR peaks to match predicted lines suggests that P1 is not present in the
irradiated reaction tubes. A possible explanation is that all P1 has converted to P2 after formation
and continued irradiation. However, P2 also does not appear to be observed, expected as stated
previously from experimental results from Boswell et al.
7
A third photoproduct suggested by Ben
Boswell from the Noah Burns group, P3, is also displayed. He observed this after irradiation of
HFB with NB (full kinetic scheme is displayed in Figure S6.15) which underwent a 6 𝜋𝜋
rearrangement. From the lack of matching NMR peaks, photoproduct P3 also is not observed.
It should be noted that the NMR spectra may not capturing the major photoproduct. During
the irradiation, a spot built up after ~40 minutes on the front surface of the NMR tube, preventing
any further light penetration into the NMR tube. The tube was moved vertically three more times
and the same shape of insoluble photoproduct spot, albeit in differing amount of buildup,
reappeared each time (Figure S6.16). The spot was yellow, darker yellow in certain spots and the
shape matched the profile of the laser. The solid photoproduct did not fuse with the glass and
detached from the walls after only minor agitation. This photoproduct was insoluble in CD3OD as
it crashed out and did not redissolve and therefore was probably not seen in the NMR – solid state
signals are substantially broadened. Haller experienced a similar photoproduct formation in the
gas phase under a mercury lamp irradiation for HFB only.
19

These results point to formation of an insoluble polymer that forms upon irradiation due to
the absence of any of the predicted photoproducts as well as the formation of the yellow product.
We believe that equivalent molar ratios of HFB and NB are consumed to form P1 at a given rate.
282

The formation of P1 is slower than the formation of P2, leading to no steady state formation of P2.
P2 then undergoes fast conversion to a polymeric form of P2.

We continue to use the prediction software by Mnova to assign possible photoproducts to
the peaks in the NMR. As the
19
F NMR is only sensitive to fluorine atoms, the photoproducts must
contain a fluorine atom. The spectra of possible fluorinated photoproducts in Figure 6.8 were
predicted via the Ensemble software. Out of all the possible photoproducts, only P6, P11, P13, and
P14 are candidates via their predicted proximity to the experimental spectrum. However, the
experimental splitting of P13 and P14 do not match what is predicted. Theory predicts singlets as
F
1
F
2
F
3
F
2
F
1
O
D
F O
D
D D
F
1
F
3
F
2
F
3
F
1
D
F
F
F
F
F
2
F
1
F
1
F
2
F
3
F
3
F
3
F
3
F
2
F
1
F
1
F
2
F
1
F
2
F
3
F
2
F
1
F
1
F
2
F
3
F
2
F
1
F
2
F
1
F
1
F
2
F
1
F
1
-153.41, -158.75, -160.73 -138.86, -153.79, -162.17
-122.24, -162.78
-124.28, -150.51, -198.20 -139.26, -150.17, -152.70
-110.19 -122.22 -190.50 -173.20 -157.15
F
3
F
3
F
1
F
2
F
2
F
1
-122.70
-131.31
-159.78
P4 P5 P6
P7 P9 P8
P10 P11 P12 P13 P14

Figure 6.8 – Possible fluorinated photoproducts in the irradiation study. The predicted chemical
shifts are displayed below each compound ordered by the indicated fluorine numerical label,
e.g., for P4, F1 is -153.41, F2 is -158.75, and F3 is -160.73.
283

the structure contains only a single fluorine atom with no fluorine neighbors. However, P13 and
P14 are multiplets and we therefore do not assign P13 or P14 as possible photoproducts.
We see that if we use Ensemble to predict the
19
F resonance, the results are close to
experiment. However, if we use Ensemble to predict Dewar-HFB, the frequencies do not match.
Haller and this current work obtained -120 to -122 ppm and -190 ppm. Ensemble predicts a value
for the bridgehead fluorines (F2) to be -163 ppm, a value of +30 ppm different. While the
prediction of
1
H and
13
C appear to have an error of 1 ppm, the situation may be different for
19
F
NMR as it is not indicated in accompanying literature. As Ensemble operates with machine
learning and a vast library of NMR spectra for
1
H and
13
C exists, it is far larger than the equivalent
19
F NMR database. A check of literature reveals no reported value for Dewar-HFB. It is possible
the error in assignment for the Dewar-HFB peaks is due to lack of supporting
19
F NMR spectra.
HFB itself is a well-known standard in
19
F NMR with literature sources, which increase the
accuracy of HFB prediction. This leaves little confidence in the predicted
19
F NMR spectra and
therefore, experimental isolation of the photoproducts would be ideal.
A similar analysis to determine the percentage of photoproducts formed reveals 49% of the
integrated signals are formed photoproducts. This number is lower if the solid photoproduct
polymer that is formed is taken into account. No further work to identify the array of photoproducts
was performed. It is likely that these photoproducts are a combination of degradation
photoproducts, polymers, and possibly photoproducts with reagents with HFB. Fluorinated
contaminants are possible but not determined. Isolation of the polymer formed should be carried
out to determine the identity of it, structure, and molecular weight.
284

6.5. Conclusion
We performed Stern-Volmer quenching via steady state fluorescence and lifetime
measurements from TCSPC and irradiation experiments. These experiments produced a metric
on how effective the HFB-NB reaction is and the associated photochemical reaction quantum
yield. The TCSPC demonstrates that the lifetime of HFB decreases with increasing NB
concentration, with no early time decay, indicating entirely diffusive quenching in the
concentration regime <0.5 M. SV analysis reveals a quenching coefficient, 𝑓𝑓 𝑄𝑄 ~ 6%. This value
indicates a maximum reaction efficiency; no solid photoproducts are seen in the fluorescence
experiments due to the much lower number of photons delivered compared to the long-time
irradiation experiments. Irradiation studies probed by
19
F NMR reveal that the reaction yield of
the first product of a [2+2] photocycloaddition of HFB with NB is not seen although a significant
solid residue is observed that could amount to up to a significant fraction of photoproduct yield.
The formation of photoproduct indicates complete transformation from HFB and NB to polymer
with no large steady state production of an intermediate photoproduct as the polymerization is
driven to completion. To compare the SV results with the irradiation study reaction quantum yield,
more separation of photoproduct needs to be done to complete the story. In addition, remaining in
solution, we see the Dewar-HFB with Φ
𝑟𝑟 𝑒𝑒 𝑛𝑛 ~ 0.06%.
The lack of desired photoproduct formation is concerning but is not entirely defeating.
Boswell have reported the existence of these photoproducts after [2+2] photocycloadditions in
gram scale production but only after carefully adjusting the experimental conditions to afford
effective photoproduct formation.
7
The largest single improvement we suggest is the change of the
sample delivery system to a flow cell. This will reduce the rate of photoproduct deposition,
allowing for more pump photons to reach the HFB chromophore; additionally, the amount of
285

photoproduct will increase as more HFB molecules experience light. Additionally, more
purification work on the NB sample, such as sublimation, is suggested to be performed to remove
the possible fluorinated contaminants. To remove the issue of all the light being absorbed by the
first few layers of solution, a thin flow cell, pathlength can be used. For example, for an absorption
of 1.0 OD for a 0.2 M HFB solution, a ~500 µm pathlength cell can be used. More effective
calculation of the photons absorbed would be achieved with a flat cell.
Photochemistry provides the possibility of producing not only highly strained molecules
that are not accessible by thermal chemistry, but instead can produce a wide variety of
photoproducts. While HFB itself undergoes a single, chemoselective reaction to Dewar-HFB in
the gas phase,
8
we have shown here that the presence of solvent
20
and olefins (as well as possible
impurities) indeed lead to other reactions. Further work is required to control the photoproduct
formation and more quantitatively assess the product quantum yields.

286

6.6. References
(1) Guo, X.; Baumgarten, M.; Müllen, K. Designing π-Conjugated Polymers for Organic
Electronics. Prog Polym Sci 2013, 38 (12), 1832–1908.
https://doi.org/10.1016/j.progpolymsci.2013.09.005.
(2) Agrawal, A. K.; Jenekhe, S. A. Synthesis and Processing of Heterocyclic Polymers as
Electronic, Optoelectronic, and Nonlinear Optical Materials. 3. New Conjugated Polyquinolines
with Electron-Donor or -Acceptor Side Groups. Chem Mater 1993, 5 (5), 633–640.
https://doi.org/10.1021/cm00029a010.
(3) Shirakawa, H.; Louis, E. J.; MacDiarmid, A. G.; Chiang, C. K.; Heeger, A. J. Synthesis of
Electrically Conducting Organic Polymers: Halogen Derivatives of Polyacetylene, (CH) x. J
Chem Soc Chem Commun 1977, 0 (16), 578–580. https://doi.org/10.1039/c39770000578.
(4) Watson, W. H.; McMordie, W. C.; Lands, L. G. Polymerization of Alkynes by Ziegler‐type
Catalyst. J Polym Sci 1961, 55 (161), 137–144. https://doi.org/10.1002/pol.1961.1205516114.
(5) Okano, T.; Ito, K.; Ueda, T.; Muramatsu, H. Generation and Thermal Polymerization of 1-
Fluoro-2-Phenylacetylene. J Fluorine Chem 1986, 32 (4), 377–388.
https://doi.org/10.1016/s0022-1139(00)81946-6.
(6) Goto, Y.; Shiosaki, M.; Hanamoto, T.; Yoshida, M.; Sawada, H. Synthesis and Properties of
Polyfluoro(Silyl)Acetylene Nanoparticles by Reaction of Fluoro(Silyl)Acetylenes with
Triethylamine. Colloid Polymer Sci 2013, 291 (5), 1211–1217. https://doi.org/10.1007/s00396-
012-2851-3.
(7) Boswell, B. R.; Mansson, C. M. F.; Cox, J. M.; Jin, Z.; Romaniuk, J. A. H.; Lindquist, K. P.;
Cegelski, L.; Xia, Y.; Lopez, S. A.; Burns, N. Z. Mechanochemical Synthesis of an Elusive
Fluorinated Polyacetylene. Nat Chem 2021, 13 (1), 41–46. https://doi.org/10.1038/s41557-020-
00608-8.
(8) Cox, J. M.; Bain, M.; Kellogg, M.; Bradforth, S. E.; Lopez, S. A. Role of the Perfluoro Effect
in the Selective Photochemical Isomerization of Hexafluorobenzene. J Am Chem Soc 2021, 143
(18), 7002–7012. https://doi.org/10.1021/jacs.1c01506.
(9) Haller, I. Kinetics and Mechanism of the Photochemical Valence Tautomerization of
Hexafluorobenzene. J Chem Phys 1967, 47 (3), 1117–1125. https://doi.org/10.1063/1.1711996.
(10) Cobas, C. Ensemble NMR Prediction. https://resources.mestrelab.com/ensemble-nmr-
prediction/.
(11) Bremser, W. HOSE - A Novel Substructure Code. Anal Chim Acta 1978, 103 (4), 355–365.
https://doi.org/10.1016/s0003-2670(01)83100-7.
(12) Davies, A. N. Who has the best proton NMR crystal ball?
287

(13) Pickett, L. W.; Muntz, M.; McPherson, E. M. Vacuum Ultraviolet Absorption Spectra of
Cyclic Compounds. I. Cyclohexane, Cyclohexene, Cyclopentane, Cyclopentene and Benzene 1.
J Am Chem Soc 1951, 73 (10), 4862–4865. https://doi.org/10.1021/ja01154a116.
(14) Doner, A. C.; Christianson, M. G.; Davis, J. C.; Koritzke, A. L.; Larsson, A.; Frandsen, K.;
Rotavera, B. Vacuum-Ultraviolet Absorption Cross-Sections of Functionalized Cyclic
Hydrocarbons: Six-Membered Rings. J Quantitative Spectrosc Radiat Transf 2019, 236, 106603.
https://doi.org/10.1016/j.jqsrt.2019.106603.
(15) Hills, E. E.; Abraham, M. H.; Hersey, A.; Bevan, C. D. Diffusion Coefficients in Ethanol
and in Water at 298K: Linear Free Energy Relationships. Fluid Phase Equilibr 2011, 303 (1),
45–55. https://doi.org/10.1016/j.fluid.2011.01.002.
(16) Narten, A. H. Diffraction Pattern and Structure of Liquid Benzene. J Chem Phys 1968, 48
(4), 1630–1634. https://doi.org/10.1063/1.1668888.
(17) Lakowicz, J. R. Principles of Fluorescence Spectroscopy, Third Edition, Third.; Springer,
2006.
(18) Rosenau, C. P.; Jelier, B. J.; Gossert, A. D.; Togni, A. Exposing the Origins of
Irreproducibility in Fluorine NMR Spectroscopy. Angew. Chem. Int. Ed. 2018, 57 (30), 9528–
9533. https://doi.org/10.1002/anie.201802620.
(19) Haller, I. Photoisomerization of Hexafluorobenzene. J Am Chem Soc 1966, 88 (9), 2070–
2071. https://doi.org/10.1021/ja00961a055.
(20) Haller, I. Photoisomerization of Hexafluorobenzene. J Am Chem Soc 1966, 88 (9), 2070–
2071. https://doi.org/10.1021/ja00961a055.
(21) Berlman, I. B. Handbook of Fluorescence Spectra of Aromatic Molecules; Academic Press;
Academic Press, 1971.

288

6.7. Supplement and Appendix
6.7.1. Absorption Spectra

In this section we present the absorption spectra of the samples used in the experiments,
both steady state and TCSPC and a basis spectrum. First, we present the absorption spectra of NB
in methanol, Figure S6.1, a structureless absorption in the DUV which contains no absorption

Figure S6.1 – Absorption spectra of NB in methanol in a 1 mm cuvette with an unknown
concentration of NB. Sample 1is the starting solution. Sample 2 is dilution of sample 1 and
sample 3 is a dilution of sample 2. Note even at saturation (samples 1 and 2) (A > 2 OD), NB
only absorbs 𝜆𝜆 < 225 nm.

200 220 240 260 280 300
0
1
2
3
Absorption (OD)
Wavelength (nm)
1
2
3
NB in Methanol

Figure S6.2 – a) Absorption spectra of the HFB and the two norbornene stock solutions in
ethanol. A benzene absorption spectrum, scaled by 0.001 is displayed for reference.
20
b)
Absorption spectra of samples of the HFB-NB mixtures used in SV.

200 250 300 350 400
0
0.2
0.4
0.6
0.8
1
1.2
Absorption (OD)
Wavelength (nm)
HFB (1 mM)
NB (1.0 M)
NB (0.5 M)
Benzene (x0.001)
266 nm
a)

200 250 300 350 400
0
0.5
1
Absorption (OD)
Wavelength (nm)
[NB] (M)
0 0.05 0.3
0.001 0.1 0.4
0.01 0.2 0.5
b)
266 nm
289

above 225 nm. While this spectrum is recorded in methanol, the SV experiments were performed
in ethanol and there is no reason to believe the absorption spectra of a simple alkene would shift
~40 nm red for such similar solvents.
The HFB absorption spectrum is very similar to the literature spectrum.
8
What is striking
is the amount of absorptive impurities in the NB stock spectra (Figure S6.2) when compared to
NB without impurities (Figure S6.1). An overlaid, scaled molar absorptivity spectrum of benzene
is included, matching the small peaks near 250 nm. One of the impurities in the SV solutions is
therefore attributed to benzene at about ~0.2 mM concentration in the 0.5 M NB solution This
calculation is based on known 𝜖𝜖 2 5 4
= 220 𝐴𝐴 − 1
𝑘𝑘 𝑀𝑀 − 1
,
21
a 1 cm pathlength and an absorption of
~0.05 OD. Determining other impurities leading to the residual absorption features was not
pursued.
The absorption spectra of the HFB-NB SV mixtures demonstrate the inconsistent HFB
concentrations (Figure S6.2b). Attempts to fit each of the HFB-NB absorption spectra as a sum
of the HFB and NB spectra was not successful.

290

6.7.2. Fluorescence Spectra of Norbornene Samples

Figure S6.3 – Fluorescence spectra of the two norbornene stock solutions. The peak at 266 nm
is Rayleigh scatter. The peak at ~270 nm, stronger in the 0.5 M solution, is assigned to the
Raman band of ethanol. An emission spectrum benzene is added with arbitrary scaling as well.

250 300 350 400 450 500
0
0.5
1
1.5
Emission Intensity (Counts x10
7
)
Wavelength (nm)
Norbornene (0.5 M)
Norbornene (1 M)
Benzene (arb.)

Figure S6.4 – Fluorescence spectra of red – 1 mM solution of HFB in ethanol; and blue – neat
ethanol. Similar to Figure S6.3, Rayleigh and Raman bands possess similar wavelength
positions.

250 300 350 400 450 500
0
1
2
3
4
Fl. Intensity (Counts x 10
7
)
Wavelength (nm)
Neat Ethanol
HFB (1mM)
291

6.7.3. Stern-Volmer Theory and Simulation
Photoluminescent processes like fluorescence and phosphorescence are deactivation
pathways for an excited state species which are inherent and serve as the photophysical, molecular
clock for a molecule. Hence, a molecule’s natural lifetime, in the absence of other deactivating
pathways, is simply the inverse of the radiative rate. However, other pathways can deactivate a
molecular excited state as well, leading to a reduction in the radiative lifetime. Deactivation
pathways include but are not limited to internal conversion, intersystem crossing, transitions
through conical intersections, and for the purposes of this section, photochemical reactions.
Unless a reaction is chemiluminescent, i.e., producing an excited state product, a
photochemical reaction produces no light when complete. As we will see, HFB and the molecules
in which it reacts with do not produce emission after the photochemistry. While HFB is a weak
fluorophore, ( Φ
𝑜𝑜 𝑓𝑓 = 2 − 4%), the fluorescence of HFB can be used as a possible marker for
reactions while in the excited state. We are concerned specifically with HFB but for the
pedagogical purposes here, we will start with a hypothetical fluorophore, referred to as 𝜏𝜏 and a
quencher, 𝑄𝑄 . This photochemical reaction can be explored via Stern-Volmer theory.
Stern-Volmer (SV) theory or SV relationship, is named eponymously after its formulators,
German American Nobel laureate in physics Otto Stern and German physical chemist Max
Volmer. Stern-Volmer theory states that there is a linear relationship between the fluorescent
lifetime of a given fluorophore in the presence of a fluorescent quencher and the concentration of
the quencher. This relationship is described by the following equation:

𝐼𝐼 0
𝐼𝐼 = 1 + 𝜏𝜏 0
𝑘𝑘 𝑞𝑞 [ 𝑄𝑄 ] Equation S6.1
where the following symbols represent:
𝑘𝑘 0
– Fluorescence intensity in the absence of quencher with units of counts, current, etc
292

𝑘𝑘 – Fluorescence intensity in the presence of quencher with units the same as 𝑘𝑘 0

𝜏𝜏 0
– Natural fluorescence lifetime of molecule of interest with units of seconds.
𝑘𝑘 𝑞𝑞 – Quenching rate constant with units of 𝐴𝐴 − 1
𝑠𝑠 − 1

[ 𝑄𝑄 ] – Quencher concentration with units of molar.
To begin SV analysis, the
fluorescence intensity of the unquenched
species is collected and can be recorded with
a standard, steady state fluorimeter. The
area/integral underneath the fluorescence
band is calculated to produce 𝑘𝑘 0
. Then, a
series of fluorescence spectra is taken as a
function of quencher concentration. The area
underneath each of these spectra is taken and
each one is a 𝑘𝑘 ([ 𝑄𝑄 ]) recorded value. Then, 𝑘𝑘 0

is divided by each of the 𝑘𝑘 and these ratios
are plotted as a function of quencher
concentration. The value for 𝜏𝜏 0
is known
from literature or can be obtained via a
fluorescence lifetime measurement. The value for 𝑘𝑘 𝑞𝑞 can be extracted using Equation S6.1.
The above process assumes a quencher and fluorophore are separated by some distance
and are not in a ground state equilibrium. After mutual diffusion, the fluorophore encounters the
quencher, and the possibility of quenching can occur. The potency of the quencher is judged by
the value of 𝑘𝑘 𝑞𝑞 where here, the quenching rate will be expressed as a product of two terms:

Figure S6.5 – Cartoon depicting 𝑘𝑘 0
and 𝑓𝑓 𝑄𝑄 . Red
circles with Q are the quencher and blue circles
with F are the fluorophore. Glowing circles
indicate excited state fluorophores. After a
successful quencher encounter, are deactivated,
turning into either F ground state or
photoproduct formation, P.
293

𝑘𝑘 𝑞𝑞 = 𝑓𝑓 𝑄𝑄 ∙ 𝑘𝑘 0
Equation S6.2
where, 𝑓𝑓 𝑄𝑄 is the quenching efficiency and in terms of photochemical reactions, the reaction
likelihood. We can think of 𝑓𝑓 𝑄𝑄 as corresponding to the quenching quantum yield where the value
ranges from 0 for a completely non-quenching reaction to 1 for unity quenching yield. The variable
𝑘𝑘 0
is the mutual diffusion rate of the quencher and fluorophore usually defined in each solvent.
These values can be obtained from the Smoluchowski equation:
17

𝑘𝑘 0
=
4 𝜋𝜋 𝐶𝐶 𝐴𝐴 1 0 0 0
� 𝑅𝑅 𝑇𝑇 + 𝑅𝑅 𝑄𝑄 � � 𝐶𝐶 𝑇𝑇 + 𝐶𝐶 𝑄𝑄 � Equation S6.3
where 𝐶𝐶 𝑀𝑀 is Avogadro’s number, 𝑅𝑅 𝑇𝑇 and 𝑅𝑅 𝑄𝑄 are the molecular radii of the fluorophore and
quencher, respectively and 𝐶𝐶 𝑇𝑇 and 𝐶𝐶 𝑄𝑄 are the diffusion constants of fluorophore and quencher
respectively. Roughly, for small molecules, 𝑅𝑅 𝑇𝑇 = 𝑅𝑅 𝑄𝑄 ≅ 5Å and for most solvent/analytes, 𝐶𝐶 𝑇𝑇 =
𝐶𝐶 𝑄𝑄 = 10
− 5
𝑘𝑘 𝑀𝑀 2
𝑠𝑠 − 1
. These numbers were estimated based on assuming benzene as both the
fluorophore and quencher diffusing in ethanol due to the molecular size/shape similarities to HFB
and NB. And so, with these values, 𝑘𝑘 0
= 1.5 ∙ 10
1 0
𝐴𝐴 − 1
𝑠𝑠 − 1
.
15
We treat this 𝑘𝑘 0
as a constant for
this chapter and has not been independently determined by our group. Going back to equation
S6.2, this leaves only 𝑓𝑓 𝑄𝑄 to be the variable we wish to know the value for. Refer to Figure S6.5 for
a visual depiction. This is useful to help determine if a quenching rate is diffusion limited or there
is some form of a barrier which slows down the fluorescence quenching. In that case, 𝑓𝑓 𝑄𝑄 will
depend on temperature, but we have not attempted to characterize a temperature dependence.
To fully explore the usefulness of SV theory, we will simulate various possibilities with an
arbitrary fluorophore for which we chose its properties. In our dilute solution, we have a
fluorophore and a quencher. This imaginary fluorophore has a fluorescence lifetime of 10 ns and
emits at 450 nm. This quencher has no fluorescence quenching capability. Quantitatively, 𝑓𝑓 𝑞𝑞 =
294

0. At all concentrations, the “quencher”, Q, performs no quenching. And so, the fluorescence
intensity is not extinguished and remains constant as a function of Q concentration. The
fluorescence spectra and the corresponding SV plot are displayed in Figure S6.8. In this parameter
space, the fluorescence spectrum is independent of [Q] as Q does not interact or quench the
fluorophore’s fluorescence. While this chemical situation is somewhat unhelpful, it serves as a
useful beginning.

We will now substitute the previous quencher with an effective quencher that can remove
fluorescence from the fluorophore. Here, 𝑓𝑓 𝑄𝑄 = 1. Instead of a set of identical fluorescence spectra,
the intensity now drops linearly (Figure S6.7a) and the SV plot reflects this (Figure S6.4b). It is
assumed that the quencher does not perturb the shape or structure of the fluorescence spectra.

Figure S6.6 – Simulated a) fluorescence spectra and b) corresponding SV plot, depicting a set
of experiments where 𝑓𝑓 𝑄𝑄 = 0. Quencher concentrations range from 0 to 1 M. All the
fluorescence spectra are identical in this case.

350 400 450 500 550 600 650
0
0.2
0.4
0.6
0.8
1
Fluorescence Intensity
Wavelength (nm)
Quencher
Concentration (M)
0 0.57
0.14 0.71
0.29 0.86
0.43 1.00
f
Q
= 0
a)

0 0.2 0.4 0.6 0.8 1
0
0.5
1
1.5
2
I
0
/
I
Quencher Concentration (M)
b)
295

To capture the full breadth of possibilities that SV plots can be used for, the fundamental
ground state thermodynamics of the system needs to be reconsidered. Diffusional quenching, while
a common phenomenon leading to fluorescence quenching, it is not the only possible mechanism
for quenching. A ground state complex, F-Q can form and will be in equilibrium with the
fluorophore and quencher. Upon
excitation of the complex, due to
close proximity, the quencher can
instantly deactivate the fluorophore
without the need for diffusion. This
in turn, leads to a drop in
fluorescence as diffusive quenching
does. In fact, complex quenching
leads to the same linear drop that
diffusional quenching does.

Figure S6.7 – Simulated a) fluorescence spectra and b) corresponding SV plot, depicting a set
of experiments where 𝑓𝑓 𝑄𝑄 = 1. Quencher concentrations range from 0 to 1 M.

350 400 450 500 550 600 650
0
0.2
0.4
0.6
0.8
1
Fluorescence Intensity
Wavelength (nm)
Quencher
Concentration (M)
0 0.57
0.14 0.71
0.29 0.86
0.43 1.00
f = 1
a)

0 0.2 0.4 0.6 0.8 1
0
2
4
6
8
10
12
I
0
/
I
Quencher Concentration (M)

Figure S6.8 – Reaction equations depicting the possible
routes for quenching via equilibrium quenching. Each
reaction/arrow is numbered. 1. Ground state equilibrium
reaction. 2. Absorption of light from the isolated
fluorophore. 3. Natural fluorophore deactivation via
radiative and non-radiative pathways. 4. Excited state
complex equilibrium 5. Absorption of light directly into
the complex. 6. Photoproduct formation. 7. Fluorophore
quenching via complex formation.
296

We can setup this system via a set of reactions. If we take equation S6.4, we can expand it
to include the various possible pathways, pictured in Figure S6.4. The relevant equations are 1.,
6., and 7. All equations are required to fully describe this general system, but these are the
equations we will pay most attention to in order to understand this system. Revisiting SV theory,
we can take the equilibrium constant of equation and introduce it into the SV equation to obtain:

𝐼𝐼 0
𝐼𝐼 = 1 + 𝐾𝐾 𝑆𝑆 𝑉𝑉 [ 𝑄𝑄 ] Equation S6.4
where instead of a quenching rate constant, the equilibrium constant, 𝐾𝐾 𝑆𝑆 𝑉𝑉 , takes
place where the remaining terms stay identical. And if one were to carry out a similar experiment
to the one depicted in Figures S6.2 and S6.3, a very similar outcome would result: a linear SV plot
with a slope that is the product of the equilibrium constant and the fluorescence lifetime.
Unfortunately, if one does not have prior knowledge of the system under study, using the
linear drop in fluorescence quantum yield with [Q] as a metric for quenching is not sufficient. Not
sufficient in the sense it is not possible to determine if the reaction is diffusional quenching or
complex formation quenching, or both.
Several assumptions have been made here which can be tested. First, we assume that the
formation of the complex does not greatly inhibit absorption of light, nor does it distort the
fluorophore’s absorption spectrum substantially. We are able to check this by taking an absorption
spectrum of the fluorophore and quencher and comparing to the quencher-less absorption spectrum
of the fluorophore. If these match, the absorption of the fluorophore is not distorted.

Equation 6.15
297

6.7.4. Assignment of Error Bars in SV plots
Analytical techniques such as SV analysis are only as useful as the accuracy to which we
are able to derive values from them. Proper error analysis not only gives confidence but also belies
any useful interpretation. Therefore, it becomes evident that one must correctly perform error
propagation to assign significance to deviations from the SV model – or rather, any experimental
data. More work must be done to carry out and properly assign error that accurately represents the
degree to which an experimental value is known.
6.7.4.1. Error Bar Assignment from Instrument Uncertainty Only
For pedagogical reasons, we will start by assigning errors with the most optimistic
beginning values to emphasize the problems with simply reporting the error from an instrument as
the experimental error. The same applies for the manufacturer’s tolerance of a given tool. This
sort of analysis greatly underestimates the error, often by orders of magnitude, leading to
undeserved confidence.
The SV quenching experiment solutions of HFB with NB began by preparing two stock
solutions of HFB and NB. We will start with NB and the dilutions of the NB stock solution
required to make the mixture solutions. Then we will concentrate on how the HFB was added to
these solutions. The use of which error propagation equations will be described along with the
error analysis.
First, the mass of the NB solid was weighed out in a capped vial. This was done by
measuring the empty vial, cap included, then adding the NB in the fume hood to avoid NB odors.
The masses of the filled and empty vials were then subtracted to find the remaining mass of the
NB. So,
𝑀𝑀 𝐶𝐶𝐵𝐵
= 𝑀𝑀 𝑣𝑣 𝑎𝑎 𝑎𝑎𝑓𝑓 𝑜𝑜 𝑝𝑝 𝑓𝑓𝑓𝑓
− 𝑀𝑀 𝑣𝑣 𝑎𝑎 𝑎𝑎𝑓𝑓𝑒𝑒 𝑝𝑝 𝑝𝑝 𝑡𝑡 𝑑𝑑 = 19.3697 − 14.6679 𝐵𝐵 = 4.7018 𝐵𝐵 Equation S6.5
298

The number of significant figures used here are due to the tolerance of the balance being 0.1 mg.
Since the mass is found via subtraction of two values, the error is calculated using the square root
of the sum of the squares:
𝑏𝑏 𝐶𝐶𝐵𝐵
𝑝𝑝𝑎𝑎𝑃𝑃𝑃𝑃
=
�
� 𝑏𝑏 𝑣𝑣 𝑎𝑎 𝑎𝑎𝑓𝑓 𝑜𝑜 𝑝𝑝 𝑓𝑓𝑓𝑓
𝑝𝑝𝑎𝑎𝑃𝑃𝑃𝑃
�
2
+ � 𝑏𝑏 𝑣𝑣 𝑎𝑎 𝑎𝑎𝑓𝑓 𝑒𝑒 𝑝𝑝𝑝𝑝 𝑡𝑡 𝑑𝑑 𝑝𝑝𝑎𝑎𝑃𝑃𝑃𝑃
�
2
= √2 ∙ 0.0001 = 1 ∙ 10
− 4
𝐵𝐵 (0.003%) Equation S6.6
and because the error in the balance is constant, the equation reduces to √2 times the error in the
balance. Relative errors are reported in parentheses. All errors are reported to one significant
figure.
Then, the NB stock solution was prepared by dissolving the solid NB in ethanol in a 50 mL
volumetric flask. Volumetric flask tolerances are reported to be 0.1% accurate with respect to the
indicated volume. The concentration of NB is found using the following:
[ 𝐶𝐶𝐴𝐴 ]
𝑃𝑃𝑡𝑡 𝑜𝑜𝑐𝑐𝑘𝑘 =
𝑝𝑝 𝑁𝑁 𝐵𝐵 𝑀𝑀 𝑀𝑀 𝑁𝑁 𝐵𝐵 ∙ 𝑣𝑣 𝑜𝑜 𝑓𝑓 5 0 𝑝𝑝𝑃𝑃
= 0.999 𝐴𝐴 Equation S6.7
where 𝐴𝐴 𝐴𝐴 𝐶𝐶𝐵𝐵
is the molar mass of NB. The error in the concentration of NB is found via Equation
6.8. Here, and for any error propagation of equations with division and multiplication, the error
of a given quantity is found by taking the square root of the sum of the relative errors. The error
in the mass and the mass itself is carried forward. And the 50 mL volumetric flask error is 0.1%
or 0.05 mL. The error in the molar mass is zero, removing it from the error equation.
𝑏𝑏 [ 𝐶𝐶𝐵𝐵 ]
𝑃𝑃 𝑏𝑏 𝑜𝑜 𝑟𝑟 𝑡𝑡 = [ 𝐶𝐶𝐴𝐴 ]
𝑃𝑃𝑡𝑡 𝑜𝑜𝑐𝑐𝑘𝑘 ∙
�
�
𝑒𝑒 𝑁𝑁 𝐵𝐵 𝑝𝑝𝑑𝑑𝑃𝑃 𝑃𝑃 𝑝𝑝 𝑁𝑁 𝐵𝐵 �
2
+ �
𝑒𝑒 𝑣𝑣 𝑜𝑜 𝑡𝑡𝑝𝑝𝑝𝑝𝑏𝑏 𝑏𝑏𝑟𝑟𝑟𝑟 𝑟𝑟 5 0 𝑝𝑝𝑃𝑃
𝑣𝑣 𝑜𝑜 𝑓𝑓 5 0 𝑝𝑝𝑃𝑃
�
2
= 1 ∙ 10
− 3
𝐴𝐴 (0.1%) Equation S6.8
Now, we calculate the error during the dilution step of NB to mix with HFB to make the quenching
solutions. These were not prepared via serial dilution as removing volume from one solution
would change the stock solutions volume, leading to uncertainty in the HFB volume required. As
such, each NB solution was prepared from the NB stock solution created earlier. A given amount
299

of volume was transferred from the NB stock solution and added to a 25 mL volumetric flask. This
flask would already have HFB solution inside. Finally, the solution was finished by diluting with
ethanol for the remaining volume.
Due to the wide range of concentrations required, several volumetric instruments were used
to transfer volumes. A 100 µL syringe, a 10 mL serological pipette, and a 5 mL volumetric flask
were used to carry out these dilutions. The syringe used had a reported tolerance of 1% of the
nominal volume so the absolute reported error of it would be 1 µL. The serological pipette has a
tolerance of 0.15 mL and volumes of 2 mL were used so an error of 7.5% is reported. And finally,
like the 50 mL and 25 mL volumetric flasks, the error is 0.05 mL or 0.01%. Not every error
calculation will be included for each mixture, but one representative will be.
For example, for a low concentration measurement, a syringe and pipette were used with
several uses from the syringe. Here, [total volume of transfer] was used, with [# of uses with
syringe] and [# of uses from pipette]. Since these volumes are all additive, we use the square root
of the sum of the squares of the errors. And since the error with the syringe is fixed here, we can
pull out the number of times the syringe or pipette was used like Equation 6.8.
𝑏𝑏 𝑣𝑣 𝑜𝑜 𝑓𝑓 𝑏𝑏 𝑟𝑟𝑑𝑑𝑜𝑜𝑃𝑃 𝑜𝑜𝑏𝑏 𝑟𝑟 ( 𝑓𝑓 𝑜𝑜 𝑙𝑙 𝑘𝑘 𝑜𝑜𝑀𝑀 𝑘𝑘 ) = � � 𝑀𝑀 𝑃𝑃𝑑𝑑 𝑟𝑟 𝑝𝑝𝑃𝑃 𝑏𝑏𝑃𝑃
∙ 𝑏𝑏 𝑃𝑃𝑑𝑑𝑟𝑟 2
� + � 𝑀𝑀 𝑝𝑝 𝑎𝑎 𝑝𝑝 𝑃𝑃𝑑𝑑𝑟𝑟 ∙ 𝑏𝑏 𝑝𝑝 𝑎𝑎 𝑝𝑝 2
�= 1 ∙ 10
− 6
𝐿𝐿 Equation S6.9
For the larger concentrations, the syringe is replaced by the 5 mL volumetric flask. The final NB
error concentration is calculated from the dilution equation of 𝐴𝐴 1
𝑉𝑉 1
= 𝐴𝐴 2
𝑉𝑉 2
where solution 1 is
the stock concentration and solution 2 is the mixture.
[ 𝐶𝐶𝐴𝐴 ]
𝑝𝑝𝑎𝑎𝑒𝑒𝑡𝑡 𝑝𝑝𝑟𝑟 𝑒𝑒 =
[ 𝐶𝐶𝐵𝐵 ]
𝑃𝑃 𝑏𝑏 𝑜𝑜 𝑟𝑟𝑡𝑡 𝑉𝑉 𝑏𝑏 𝑟𝑟𝑑𝑑𝑜𝑜 𝑃𝑃 𝑜𝑜𝑏𝑏 𝑟𝑟 𝑉𝑉 𝑝𝑝𝑟𝑟 𝑥𝑥 𝑏𝑏 𝑝𝑝𝑟𝑟𝑏𝑏 = 0.001 𝑘𝑘 𝑜𝑜 0.5 𝐴𝐴 Equation S6.10
The volume of each of the mixture solutions is 25 mL. The error for [ 𝐶𝐶𝐴𝐴 ]
𝑝𝑝𝑎𝑎𝑒𝑒𝑡𝑡 𝑝𝑝𝑟𝑟 𝑒𝑒 is then carried
forward in Equation S6.11.
300

𝑏𝑏 [ 𝐶𝐶𝐵𝐵 ]
𝑑𝑑 𝑟𝑟 𝑡𝑡 𝑝𝑝𝑏𝑏 𝑟𝑟 𝑜𝑜 𝑜𝑜𝑃𝑃 = [ 𝐶𝐶𝐴𝐴 ]
𝑟𝑟 𝑎𝑎 𝑓𝑓𝑝𝑝𝑡𝑡 𝑎𝑎 𝑜𝑜 𝑛𝑛𝑃𝑃
∙
�
�
𝑒𝑒 [ 𝑁𝑁 𝐵𝐵 ]
𝑃𝑃 𝑏𝑏 𝑜𝑜 𝑟𝑟𝑡𝑡 [ 𝐶𝐶𝐵𝐵 ]
𝑃𝑃 𝑏𝑏 𝑜𝑜 𝑟𝑟𝑡𝑡 �
2
+ �
𝑒𝑒 𝑣𝑣 𝑜𝑜 𝑡𝑡 2 5 𝑝𝑝𝑃𝑃
𝑣𝑣 𝑜𝑜 𝑓𝑓 2 5 𝑝𝑝𝑃𝑃
�
2
+ �
𝑒𝑒 𝑣𝑣 𝑜𝑜 𝑡𝑡 𝑏𝑏 𝑟𝑟𝑑𝑑𝑜𝑜𝑃𝑃 𝑜𝑜𝑏𝑏 𝑟𝑟 𝑣𝑣 𝑜𝑜 𝑓𝑓 𝑏𝑏 𝑟𝑟𝑑𝑑𝑜𝑜𝑃𝑃 𝑜𝑜𝑏𝑏 𝑟𝑟 �
2
Equation S6.11
Now, we begin propagation of the HFB concentrations. The HFB stock solution was
prepared by taking a syringe needle amount of neat HFB and adding it to a 25 mL volumetric flask.
This was diluted then by a factor of ten. The HFB stock solution calculation is in Equation S6.12.
[ 𝐻𝐻 𝜏𝜏𝐴𝐴 ]
𝑃𝑃𝑡𝑡 𝑜𝑜𝑐𝑐𝑘𝑘 =
𝑣𝑣 𝑜𝑜 𝑓𝑓 𝐻𝐻𝐻𝐻 𝐵𝐵 𝑃𝑃 𝑑𝑑𝑟𝑟
∙ 𝜌𝜌 𝐻𝐻𝐻𝐻 𝐵𝐵 𝑀𝑀 𝑀𝑀 𝐻𝐻𝐻𝐻 𝐵𝐵 ∙ 𝑉𝑉 𝑜𝑜 𝑓𝑓 𝑃𝑃 𝑏𝑏 𝑜𝑜 𝑟𝑟𝑡𝑡 = 0.01 𝐴𝐴 𝑜𝑜 𝑀𝑀 10 𝑀𝑀𝐴𝐴 Equation S6.12
The volume of the HFB stock solution was 25 mL. Like for the NB dilutions, the HFB mixture
concentrations and the errors of those concentrations are calculated in Equations S6.13 and S6.14.
[ 𝐻𝐻 𝜏𝜏𝐴𝐴 ]
𝑝𝑝𝑎𝑎𝑒𝑒𝑡𝑡 𝑝𝑝𝑟𝑟 𝑒𝑒 =
𝑣𝑣 𝑜𝑜 𝑓𝑓 𝐻𝐻𝐻𝐻 𝐵𝐵 2 𝑝𝑝 𝑃𝑃𝑝𝑝 𝑟𝑟 𝑝𝑝 [ 𝐶𝐶 𝑇𝑇𝐵𝐵 ]
𝑃𝑃 𝑏𝑏 𝑜𝑜 𝑟𝑟𝑡𝑡 𝑣𝑣 𝑜𝑜 𝑓𝑓 𝐻𝐻𝐻𝐻 𝐵𝐵 2 5 𝑝𝑝𝑃𝑃
Equation S6.13
𝑏𝑏 [ 𝐶𝐶 𝑇𝑇𝐵𝐵 ]
𝑝𝑝𝑎𝑎 𝑒𝑒𝑡𝑡 𝑝𝑝𝑟𝑟 𝑒𝑒 = [ 𝐻𝐻 𝜏𝜏𝐴𝐴 ]
𝑝𝑝 𝑎𝑎𝑒𝑒𝑡𝑡𝑝𝑝𝑟𝑟𝑒𝑒
∙ � �
𝑒𝑒 𝑝𝑝 𝑟𝑟 𝑝𝑝 𝑏𝑏 𝑏𝑏 𝑏𝑏 𝑏𝑏 𝑣𝑣 𝑜𝑜 𝑓𝑓 𝐻𝐻𝐻𝐻 𝐵𝐵 2 𝑝𝑝𝑃𝑃 𝑝𝑝 𝑟𝑟 𝑝𝑝 �
2
+ �
𝑒𝑒 [ 𝐻𝐻𝐻𝐻 𝐵𝐵 ]
𝑃𝑃 𝑏𝑏 𝑜𝑜 𝑟𝑟𝑡𝑡 [ 𝐶𝐶 𝑇𝑇𝐵𝐵 ]
𝑃𝑃 𝑏𝑏 𝑜𝑜 𝑟𝑟𝑡𝑡 �
2
+ �
𝑒𝑒 𝑣𝑣 𝑜𝑜 𝑡𝑡 2 5 𝑝𝑝𝑃𝑃
𝑣𝑣 𝑜𝑜 𝑓𝑓 2 5 𝑝𝑝𝑃𝑃
�
2
Equation S6.14
Now that the error in the concentration
of NB has been calculated, the error in the
fluorescence intensities must be accounted for
to provide a complete picture. Naively, because
the fluorescence measurement is a counting
experiment, the most optimistic view of error
would be to simply use the error of a Poisson
distribution: the square root of the number of
recorded events. The intensities recorded here
are found by integrating the fluorescence band
of HFB. So, the error of a given SV intensity

Figure S6.9 – Stern-Volmer plot of HFB
fluorescence as a function of NB
concentration. The red line is a line of best fit
with a y-intercept of zero. The error bars are
the result of the previous error propagation
measurements shown previously in text.
Similar to Figure 6.1, error bars are smaller
than the dots presented.

301

ratio would be to take the square root of the intensities. The error on the data points here for the y-
axis are calculated in Equation 6.15.
𝑏𝑏 �
𝐼𝐼 0
𝐼𝐼 �= �
𝐼𝐼 0
𝐼𝐼 � ∙
�
�
𝑒𝑒 𝐼𝐼 0
𝐼𝐼 0
�
2
+ �
𝑒𝑒 𝐼𝐼 𝐼𝐼 �
2
= 4% Equation 6.15
The SV graph of HFB fluorescence intensity as a function of NB concentration is plotted
in Figure S6.9. While the general trend of scatter follows the fit line, the spread of points greatly
exceeds the error bars. Variation on this order is non-physical so the error bars are not properly
describing the spread of data points. A more careful analysis of possible errors is required and is
demonstrated further in the next section.
6.7.4.2. Error Bar Assignment Incorporating Human Error
Error analysis is effective when all sources of error and their relative magnitudes are
properly accounted for an experiment. This error includes not only the tolerance and inherent error
in a tool, instrument, or piece of glassware, but also human error. Each glassware used can be
accounted for with rigorous repetition of a given measurement. This repetition was done here.
First, we must account for each specific glassware used that were “direct” additions in
comparison to dilutions. These include the pipette, syringe and 5 mL volumetric flask used for the
previous experiments. To assign error bars to the human use of these instruments and to find the
error in each tool, the method by determining the error must be less than the experimental error
when using the tool. A very accurate instrument available to nearly all chemistry laboratories is
the analytical balance which was used here. A volume in a tool was measured in a similar manner
to the SV analysis and then weighed against a reference. The error was calculated as the standard
deviation of these measurements. This was done for a variety of pieces of volumetric tools.
Absorption spectroscopy cannot be used in the case of absolute measurements. The
absolute mass amount of neat liquid chromophore; like HFB, would require a very thin pathlength
302

cell as concentrations of pure liquids are ~20-40 M. The pathlength and concentration are the
variables sensitive in absorption spectroscopy. Additionally, the geometry of UV-vis spectroscopy
is not ideal for such low volumes unless the user possesses a low volume cuvette. However,
absorption would be ideal for measuring dilutions which will be done in the next section. Here, we
focus on measuring the error in direct measurements of liquid using gravimetric analysis.
The method for determining human error in the glassware is as follows. An empty vial,
cleaned and dried previously, was weighed, including the stopper. Every weighing measurement
was performed three to five times by removing the volumetric flask and placing it back on the
balance. Variations for successive weighings were seen on the final digit, 0.1 mg. For the pipette
and syringe, an amount similar to the SV was measured out. The pipette was 2 mL and the syringe,
29 µL. For the pipette and flask, the liquid used was 200 Proof VWR ethanol and HFB was used
for the syringe. The liquid would be taken into the glassware, measured, then dispensed into the
volumetric flask. The flask was capped and then weighed three to five times. This amount was
measured repeatedly with each addition being added to the flask, and not emptying the flask.
Emptying the flask would leave residual liquid behind and lead to error that was not present in the
SV analysis. For a given volume dispensed, the average amount was taken. The differences in the
mass measured was calculated between runs and the mean and standard deviation of this was taken.
The volumetric flask was poured out each time, with only the mass with the ethanol measured.
303

Displayed in Table S6.1 is the summary of a series of measurements performed using these
pieces of glassware. The final column displays the percent error for each piece of glassware. The
10 µL syringe is the least accurate by an order of magnitude. The 10 µL syringe was also used
without a needle as the needle dead volume would greatly exceed the volume of the syringe. The
volume of 2.9 µL is the size of a small droplet which can evaporate a larger portion of its volume
compared to a large droplet. There was also difficulty in removing the droplet from the syringe
barrel head and getting it off required tapping against the walls of the vial or volumetric flask.
Next, the 100 µL syringe was more accurate but still almost 6% error. The 100 µL syringe was
used with a needle which can contain dead volume. The barrel of the syringe will extract air
bubbles, and these are only removed through proper air purging by priming the syringe. The
pipette is the next most error prone due to the need to measure twice for a single measurement: the
initial and final values. The pipette, due to the wider barrel compared to the volumetric, has a
wider meniscus which leads to larger error.
Table S6.1 – Error Determination of Various Glassware

Volume
Meas.
# of
Trials
m0 (g) mf (g) ∆avg (g) σ (g)
Percent Error
(%)
Syringe
(100 µL)
29 µL 9 26.0367 26.6444 0.0760 0.0057 7.5
Syringe
(10 µL)
2.9 µL 10 14.4495 14.4963 0.0047 0.0031 66
Pipette
(10 mL)
2 mL 11 21.3315 38.9800 1.0644 0.0884 5.5
Vol. Flask
(5 mL)
5 mL 9 16.6578 -- 3.9338 0.0209 0.53
m0 – initial mass of volumetric flask
mf – Final mass of volumetric flask
∆avg – Average difference between trials for syringe and pipette/Average mass of volumetric
flask with ethanol
σ - standard deviation of ∆avg

304

Importantly, when calculating the error bars for the fluorescence yield, the vertical axis
errors in the SV plot, the concentration of the HFB is important. For generally low absorbing
solutions (ODpump < 0.1 OD), the fluorescence yield increases as the concentration increases. Inner
and outer filtering effects are removed by having such low concentration solutions. Ideally, each
solution of an SV series should have an identical fluorophore concentration, but this is not always
the case. And it is possible that these concentrations can vary. To determine the effect of the error
of HFB concentration on the SV series, error analysis was carried out for the HFB concentration.
The HFB added to each mixture was from a stock solution of 10 mM using the
aforementioned pipette. Here, 2 mL of HFB stock was added to the mixture solutions, then the
appropriate amount of quencher from a stock solution and the rest was diluted to the mark on the
volumetric flask. The errors for all these transfers were carried forward, leading to a percent error
in HFB concentration of 26%, a substantial error. The fraction of light absorbed by a fluorophore
with a given absorption value/concentration is given in Equation S6.16:
𝜏𝜏 = 1 − 10
− 𝑀𝑀 = 1 − 10
− 𝜖𝜖 ∙ 𝑃𝑃 ∙ 𝑐𝑐 𝑜𝑜 𝑛𝑛 𝑐𝑐 Equation S6.16
where 𝜖𝜖 is the molar absorptivity (M
-1
cm
-1
) of the fluorophore at the excitation wavelength, 𝑏𝑏 is
the pathlength (cm) of the cuvette and 𝑘𝑘𝑜𝑜 𝑀𝑀𝑘𝑘 is the concentration (M) of the fluorophore. This can
then be translated to the error of the absorption of HFB with Equation S6.17:
𝑏𝑏 ( 𝜏𝜏 ) = 10
− 𝜖𝜖 ∙ 𝑃𝑃 ∙ 𝑐𝑐 𝑜𝑜 𝑛𝑛 𝑐𝑐 ∙ ln(10) ∙ 𝑏𝑏 ( 𝐴𝐴 ) Equation S6.17.
To find the error in absorption, we must find the error in 𝜖𝜖 , 𝑏𝑏 , and 𝑘𝑘𝑜𝑜 𝑀𝑀𝑘𝑘 . Here, however, we assume
the pathlength and absorptivity are known absolutely and possess no error. This is an assumption
but the error in concentration will outweigh both errors in pathlength and molar absorpvitiy.
Therefore, the error in the absorption of light absorbed is simply:
𝑏𝑏 ( 𝐴𝐴 ) = 𝐴𝐴 ∙
�
�
𝑒𝑒 ( 𝑐𝑐 𝑜𝑜 𝑛𝑛 𝑐𝑐 )
𝑐𝑐 𝑜𝑜 𝑛𝑛 𝑐𝑐 �
2
=
𝑀𝑀 𝑐𝑐 𝑜𝑜 𝑛𝑛 𝑐𝑐 ∙ 𝑏𝑏 ( 𝑘𝑘𝑜𝑜 𝑀𝑀𝑘𝑘 ) Equation S6.18
305

The error of the absorption from Equation S6.17 is substituted into Equation S6.18. The error that
extends to the fluorescence measurement will be explored after starting with possible errors in the
spectroscopic measurements.
To isolate the error in the dilutions, absorption, and fluorescence measurements, a set of
different approaches is taken. Here, the error in these measurements is determined by the standard
deviation of a series of absorption and emission spectra. To do this, dilutions were performed from
a stock solution of HFB and then the absorption spectra were taken. The instrument used was an
Agilent Cary 60. The scan rate was the “fast” setting with a 1200 nm/min scan rate, 1 nm divisions
and 12.5 ms integration time. The following measurements give insight into how much a scan can
vary over experimental times.
The first possible error in an absorption measurement is the scan error. This is the error
that can occur between runs of an identical solution with no change to the cuvette or instrument
whatsoever. To perform this, a baseline with ethanol was taken repeatedly with no adjustments
done to the cuvette, simply initiating several back-to-back runs. The stock solution which was
~10x more concentrated than the dilutions, must be taken in a 1 mm cuvette to avoid extreme
absorption leading to poor determination of the transmitted beam intensity in the UV-vis. The
results are displayed in Figure S6.10. The variations between each scan are minor and the
instrumental error is 0.2%.
306

The next possible source of error can be found when placing a cuvette in and outside of a
cuvette holder. The cuvette holder used has two stipulations to note. One, it has large degrees of
horizontal freedom. The cuvette can “rock” back and forth when placed in the holder. Secondly,
the next error more minor and not systematic. The white light for the absorption measurement is
brought to a focus within the sample chamber of the Cary 60. However, the placement of the
cuvette holder for the 1 mm homebuilt and 1 cm provided by Agilent is different. The 1 cm has
the white light focus to the center of the cuvette whereas the 1 mm cell holder is stationed so the
focus is ~5 cm in front of the cuvette. This placement exacerbates the differences when placing
the cuvette inside the instrument.
The error for the placement of the cuvette is detailed in Figure S6.11. The error with
respect to cuvette placement is 8%, much larger the scan error. Also of note is the two sets of
baselines (Trials 2, 10, and 11 being offset by +10 mOD). The side of the cuvette that faced the
beam was not accounted for and is attributed to the difference between the high and low baselines.
Trial 11 was taken as the baseline for the next set of 1 mm cuvette absorption measurements.

Figure S6.10 – Absorption baseline of ethanol in a 1 mm cuvette. Left: full wavelength range.
Right: inset.

200 300 400 500 600 700 800
0
0.1
0.2
0.3
0.4
Absorption (OD)
Wavelength (nm)
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5

230 235 240 245 250
0.15
0.155
0.16
0.165
0.17
Absorption (OD)
Wavelength (nm)
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5
307

A cuvettes outer walls should be cleaned after each removal and addition of a liquid. Failure
to do so could lead to residual material on the walls which can affect the measurement. Dust and
debris can lead to scatter, solvent can lead to a reduction of signal due to scatter and solution can
lead to an increase in absorption due to an increased effective pathlength. Therefore, the cleaning
procedure used for the 1 mm cuvette was tested. Here, the cuvette was taken out, washed without
any solution removal or addition, and then placed back in for another absorption measurement.
The results of which are displayed in Figure S6.12. The error here is 3%, which is surprising
considering the simple act of removing and placing of the cuvette is 8%. However, the overall
signal of HFB compared to blank ethanol is much higher and the error should stay the same,
leading to an overall drop in percent error; the source of the cleaning error being larger than the
placement error is currently unknown.

Figure S6.11 – Absorption baseline of ethanol in a 1 mm cuvette with repeated placement of
the cuvette in and out of the cuvette holder. Left: full wavelength range. Right: inset.

200 300 400 500 600 700 800
0
0.1
0.2
0.3
0.4
Absorption (OD)
Wavelength (nm)
Trial 1 Trial 7
Trial 2 Trial 8
Trial 3 Trial 9
Trial 4 Trial 10
Trial 5 Trial 11
Trial 6

230 235 240 245 250
0.145
0.15
0.155
0.16
0.165
0.17
Absorption (OD)
Wavelength (nm)
Trial 1 Trial 5 Trial 9
Trial 2 Trial 6 Trial 10
Trial 3 Trial 7 Trial 11
Trial 4 Trial 8
308

Since the dilutions drop the signal level by a factor of 13x, the 1 mm cuvette was replaced
with a 1 cm cuvette. Baselines of the 1 cm cuvette when placing in and out of the spectrometer
produced a 0.7% error. This is understandable as the 1 cm cuvette holder is far more rigid than the
1 mm cuvette holder. Here, the cell was placed in a consistent orientation with respect to the beam
to remove face dependent error.

The sample was then diluted in the same manner as the SV analysis: 2 mL measured from
the stock solution using a serological pipette, added to a 25 mL volumetric flask, and filled to the

Figure S6.12 – Absorption spectrum of 10 mM HFB in ethanol in a 1 mm cuvette with repeated
external washings. Left: full wavelength range. Right: inset.

200 220 240 260 280 300
0
0.5
1
1.5
Absorption (OD)
Wavelength (nm)
Trial 2.1
Trial 2.2
Trial 2.3
Trial 2.4

265 266 267 268 269 270
0.12
0.14
0.16
0.18
0.2
0.22
0.24
Absorption (OD)
Wavelength (nm)
Trial 2.1
Trial 2.2
Trial 2.3
Trial 2.4

Figure S6.13 – Absorption spectrum of 1 mM HFB in ethanol in a 1 cm cuvette with repeated
washings. Right: full wavelength range. Left: inset.

200 220 240 260 280 300
0
0.2
0.4
0.6
Absorption (OD)
Wavelength (nm)
Aliquot 1
Aliquot 2
Aliquot 3
Aliquot 4

220 225 230 235 240
0.4
0.45
0.5
0.55
0.6
Absorption (OD)
Wavelength (nm)
Aliquot 1
Aliquot 2
Aliquot 3
Aliquot 4
309

mark with blank ethanol. This cuvette was then placed inside the sample chamber and removed.
This process was repeated several times to obtain a standard deviation. Cuvette insertion error was
3%.
Another possible error can be the transferring of solution to a cuvette for a spectroscopic
measurement. The simple procedure of pouring out a solution and adding another one can
introduce issues like dilution if the previous liquid was just solvent, evaporation of solvent leading
to increased concentration, contamination, etc. Often, the first of these issues is accounted for by
repeated washings of the cuvette with the solution of interest. Therefore, a series of experiments
where an HFB solution was put into a cuvette multiple times with an absorption experiment
conducted each time. The results of this experiment are displayed in Figure S6.13, producing a
1.1% error at 230 nm.

The final source of error possible in these measurements would be the error of dilutions.
The error of dilutions was found by preparing a 10 mM stock solution of HFB, and then creating
a series of repeated dilutions. A UV-vis spectrum is taken for each dilution. The results of this

Figure S6.14 – Absorption spectrum of 1 mM HFB in ethanol in a 1 cm cuvette with repeated
dilutions. Left: full wavelength range. Right: inset.

200 220 240 260 280 300
0
0.2
0.4
0.6
Absorption (OD)
Wavelength (nm)
Dilution 1
Dilution 2
Dilution 3
Dilution 4

220 225 230 235 240
0.4
0.44
0.48
0.52
0.56
0.6
Absorption (OD)
Wavelength (nm)
Dilution 1
Dilution 2
Dilution 3
Dilution 4
310

series are displayed in Figure S6.14. An error of 1% was obtained. This concludes the errors
associated with HFB concentration determination and the absorption measurements.

The fluorescence experiment also contains error in scan error, cuvette positioning, and
reproducibility. Determining fluorescence error in experiment was performed similarly to the
absorption measurements. Here, an empty 1 cm cuvette was placed in and out of the fluorimeter
(Horiba Instruments, Spex) several times with a fluorescence spectrum taken each time displayed
in Figure S6.13. The wavelength increment was 2.5 nm, 5 nm for the slit width and 200 ms
integration time. The error was found by calculating the average and standard deviation of the
Rayleigh scatter of the excitation beam. The error for the fluorescence scan is 0.7%, the same as
for the absorption measurement, understandably as both instruments are equipped with rigid
cuvette holders. For both instruments, lamp fluctuations of several minutes do not seem to
contribute strongly.
With the error in a given fluorescence measurement identified, it’s important to revisit how
the concentration plays into effect with the overall emission intensity. The emission intensity of a
given species in solution is defined by Equation 6.19:

Figure S6.15 – Emission/Scatter spectrum of a blank 1 cm cuvette, focusing on the Rayleigh
scatter region. Right: full wavelength range. Left: inset on peak of Rayleigh scatter.

250 260 270 280 290
0
4
8
12
16
Fl. Intensity (Counts x10
6
)
Wavelength (nm)
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5
Trial 6
Trial 7
Trial 8

266.5 267 267.5 268 268.5 269
15.5
16
16.5
17
17.5
Fl. Intensity (Counts x10
6
)
Wavelength (nm)
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5
Trial 6
Trial 7
Trial 8
311

𝑘𝑘 = 𝛼𝛼 𝑘𝑘 𝑓𝑓𝑎𝑎𝑝𝑝𝑝𝑝
∙ (1 − 10
− 𝑀𝑀 ) ∙ 𝜙𝜙 𝑜𝑜 𝑓𝑓 = 𝛼𝛼 𝑘𝑘 𝑓𝑓𝑎𝑎𝑝𝑝𝑝𝑝
∙ 𝜏𝜏 ∙ 𝜙𝜙 𝑜𝑜 𝑓𝑓 Equation 6.19
where 𝛼𝛼 is a constant corresponding to the geometrical layout of the fluorimeter, the relative
reflectivities of the mirrors, the detection efficiency, etc; 𝑘𝑘 𝑓𝑓𝑎𝑎𝑝𝑝𝑝𝑝
is the light intensity of the lamp;
and 𝜙𝜙 𝑜𝑜 𝑓𝑓 is the fluorescence quantum yield of the fluorophore. The geometrical constant 𝛼𝛼 is always
constant over the period of the experiments and is taken to be error-free. The ratio of
𝐼𝐼 𝑡𝑡𝑑𝑑𝑝𝑝𝑝𝑝
0
𝐼𝐼 𝑡𝑡𝑑𝑑𝑝𝑝𝑝𝑝
is taken
to be one when SV ratio,
𝐼𝐼 0
𝐼𝐼 is taken. The error in
𝐼𝐼 𝑡𝑡𝑑𝑑𝑝𝑝𝑝𝑝
0
𝐼𝐼 𝑡𝑡𝑑𝑑𝑝𝑝𝑝𝑝
is taken to be 0.7% from the previous error
analysis. The error in the fluorescence measurement is then taken as a sum of the percent errors
from the results of Figures S6.10-S6.13 as well as the error in the absorption of HFB.
With each of the errors for the instruments, glassware, concentrations, and experimental
procedures accounted for, it is now possible to revisit the SV plot of HFB with concentration
dependence. We apply each of the errors in the
manner described in the previous section but
replacing the instrumental/tolerance errors
with the errors obtained here. The SV plot with
these errors is displayed in Figure S6.14.
There are several takeaways of note
here that are important to address. First, and
most importantly, the large deviations are due
to errors in measurements and the error bars
capture this deviation. Secondly, the errors in
the methods taken by the scientist, unless analytical methods are adopted, lead to far stronger
deviations rather than the error simply associated with a piece of glassware. Third, the error in the

Figure S6.16 – SV plot of HFB with
norbornene with updated error bars using the
error propagation method described in the
preceding text.
0 0.1 0.2 0.3 0.4 0.5
0
0.5
1
1.5
2
2.5
3
I
0
/
I
[Norbornene] (M)
312

HFB concentration leads to a larger error than the concentration of the quencher, a possible
overlook when performing SV analysis. And finally, the largest y-error bars are for the solutions
with 0.3 and 0.5 M, solutions that were prepared with a 10 µL syringe, a piece of glassware that is
not suitable for analytical quality methods.
The error in the concentration of a fluorophore can be minimized by resorting to a
technique that is less susceptible to concentration effects. Therefore, when performing quenching
studies, we turn to lifetime measurements. Lifetime measurements are less affected by
concentration error of the fluorophore due to independent lifetimes with respect to fluorophore
concentrations.

313

6.7.5. HFB and Norbornene Results from Burns Group

Figure S6.17 – Kinetic scheme of HFB with norbornene in pentane with irradiation at 254 nm
as produced by B. Boswell.
314

6.7.6. Photograph of Irradiated Samples

Figure S6.18 – Photographs of NMR tube with HFB and norbornene during irradiation (left)
and after irradiation (right). For left, the beam is coming from the top of the picture and the
blue color is the fluorescence of HFB.
315

Appendix A: Derivations and Calculations
1. Calculated QY from Rate Constants (Equation 3.2)
The model to describe the two-system in SBCT used to fit the data comes from quencher
kinetics (chapter 3). In our case, the quencher concentration is time dependent (SBCT). As such,
the Φ
𝑜𝑜 𝑓𝑓 used to describe the system is also derived from quencher kinetics.
1,2
Here, we will back
calculate the provided equation to what matches in quencher kinetics. Equation 3.2 is displayed
below.
Φ
𝑜𝑜 𝑓𝑓 =
𝑘𝑘 𝑟𝑟 ( 𝑘𝑘 𝑃𝑃 𝑏𝑏𝑏𝑏
+ 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 )
( 𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑜𝑜 𝑟𝑟 )( 𝑘𝑘 𝑃𝑃 𝑏𝑏𝑏𝑏
+ 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 ) + 𝑘𝑘 𝑟𝑟𝑏𝑏 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 Equation 3.2
First, we will invert equation 3.2 and separate to a common denominator.

1
Φ
𝑜𝑜𝑡𝑡
=
( 𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑜𝑜 𝑟𝑟 )( 𝑘𝑘 𝑃𝑃 𝑏𝑏𝑏𝑏
+ 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 )
𝑘𝑘 𝑟𝑟 ( 𝑘𝑘 𝑃𝑃 𝑏𝑏𝑏𝑏
+ 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 )
+
𝑘𝑘 𝑟𝑟𝑏𝑏 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 𝑘𝑘 𝑟𝑟 ( 𝑘𝑘 𝑃𝑃 𝑏𝑏𝑏𝑏
+ 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 )
Equation A1.1
The like terms for the first fraction cancel.

1
Φ
𝑜𝑜𝑡𝑡
=
𝑘𝑘 𝑟𝑟 + 𝑘𝑘 𝑜𝑜 𝑟𝑟 𝑘𝑘 𝑟𝑟 +
𝑘𝑘 𝑟𝑟𝑏𝑏 𝑘𝑘 𝑟𝑟 𝑏𝑏𝑟𝑟
𝑘𝑘 𝑟𝑟 ( 𝑘𝑘 𝑃𝑃 𝑏𝑏𝑏𝑏
+ 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 )
Equation A1.2
Here we see that the first term is just the reciprocal of the common quantum yield equation, the
expected fluorescence from the LE state. The second term is ratio of the exciplex emission to the
fluorescence emission.

1
Φ
𝑜𝑜𝑡𝑡
=
1
Φ
0
𝐻𝐻 +
Φ
𝑜𝑜𝑡𝑡
𝐸𝐸 Φ
0
𝐻𝐻 Equation A1.3
Φ
𝑜𝑜 𝑓𝑓 𝐸𝐸 =
𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 𝑘𝑘 𝑃𝑃 𝑏𝑏𝑏𝑏
+ 𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑟𝑟 Equation A1.4
Φ
𝑜𝑜 𝑓𝑓 𝐸𝐸 =
𝑘𝑘 𝑟𝑟 𝑏𝑏 𝑘𝑘 𝑟𝑟 Equation A1.5
These equations relate back to what is found in quencher kinetics.

316

2. Measured QY from Steady State Measurements (Equation 3.1)
In chapter 3, the Φ
𝑜𝑜 𝑓𝑓 of 2 in various solvents was measured via the relative fluorescence
quantum yield method. In this method, a fluorophore’s unknown Φ
𝑜𝑜 𝑓𝑓 is referenced to a standard,
whose quantum yield is known and characterized. We derive that equation, Equation 3.1, here. To
start, for an optically dilute solution of a fluorophore, the total fluorescence intensity, 𝐸𝐸 , is
proportional to the intensity of amount of light absorbed and the fluorescence quantum yield
(Equation A2.1).
𝐸𝐸 = 𝑘𝑘 0
(1 − 10
− 𝜖𝜖 ∙ 𝑓𝑓 ∙ 𝑐𝑐 ) Φ
𝑜𝑜 𝑓𝑓 Equation A2.1
where:
𝑓𝑓 = 1 − 10
− 𝜖𝜖 ∙ 𝑓𝑓 ∙ 𝑐𝑐 = 1 − 10
− 𝑀𝑀 Equation A2.2
so:
𝐸𝐸 = 𝑘𝑘 0
𝑓𝑓 Φ
𝑜𝑜 𝑓𝑓 Equation A2.3
Here, 𝜖𝜖 is the molar absorptivity, 𝑓𝑓 is the optical pathlength, 𝑘𝑘 is the concentration of
fluorophore, 𝐴𝐴 is the absorption value of the fluorophore at the excitation wavelength, 𝑓𝑓 is the
fraction of light absorbed by the fluorophore and Φ
𝑜𝑜 𝑓𝑓 is the sample’s fluorescence quantum yield.
Usually, the sample will be contained in a cuvette in a solvent which has a differing refractive
index, 𝑀𝑀 𝑎𝑎 compared to the refractive index which contains the detector, 𝑀𝑀 0
, often air. When light
passes from a higher refractive index to a lower refractive index (since 𝑀𝑀 𝑎𝑎 > 𝑀𝑀 0
most often), the
light undergoes refraction as displayed in Figure A0.1a. To correct this, consider a point source,
S, from which light is emitting with a steady state intensity of 𝑘𝑘 (photons/sec).
In order to determine the distribution of 𝑘𝑘 over a conical region, we introduce Figure A0.1b.
Skip to Equation A2.15 for the result after geometrical corrections. A steradian is defined as the
317

solid angle subtended at the center of a unit sphere by a circular unit area on its surface. Given a
sphere with radius 𝑀𝑀 a solid angle Ω, the area subtended by one steradian is:
Ω =
𝑀𝑀 𝑟𝑟 2
=
2 𝜋𝜋 ℎ
𝑟𝑟 Equation A2.4
where also:
Ω = 2 𝜋𝜋 (1 − cos 𝜃𝜃 𝑎𝑎 ) = 4 𝜋𝜋 sin
2
𝜃𝜃 𝑟𝑟 2
Equation A2.5
Which can be found from computing the double integral using the unit surface element in spherical
coordinates:
∫ ∫ sin 𝜃𝜃 ′ 𝑀𝑀 𝜃𝜃 ′
𝑀𝑀𝜙𝜙 = ∫ 𝑀𝑀𝜙𝜙 2 𝜋𝜋 0
∫ sin 𝜃𝜃 ′
𝑀𝑀 𝜃𝜃 ′
𝜃𝜃 𝑟𝑟 0
𝜃𝜃 𝑟𝑟 0
2 𝜋𝜋 0
Equation A2.6
= 2 𝜋𝜋 ∫ sin 𝜃𝜃 ′
𝑀𝑀 𝜃𝜃 ′
𝜃𝜃 𝑟𝑟 0
Equation A2.7
= 2 𝜋𝜋 [ − cos 𝜃𝜃 ′ ]
0
𝜃𝜃 𝑟𝑟 Equation A2.8
= 2 𝜋𝜋 (1 − cos 𝜃𝜃 𝑎𝑎 ) Equation A2.9
For small 𝜃𝜃 𝑎𝑎 , we introduce a truncated Taylor series for cos 𝜃𝜃 𝑎𝑎 :
cos 𝜃𝜃 𝑎𝑎 ≅ 1 −
𝜃𝜃 𝑟𝑟 2
2
Equation A2.10
Therefore
Ω = 2 𝜋𝜋 � 1 − � 1 −
𝜃𝜃 𝑟𝑟 2
2
� � = 𝜋𝜋 𝜃𝜃 𝑎𝑎 2
Equation A2.11
The intensity, 𝑘𝑘 𝑎𝑎 , emitted into the cone with the full angle, 2 𝜃𝜃 𝑎𝑎 , is given by:
𝑘𝑘 𝑎𝑎 ≅
𝐼𝐼 𝜋𝜋 𝜃𝜃 𝑟𝑟 2
Equation A2.12
318

We assume there is no loss of photons due to absorption or reflection off the glass or
scattering when light passes from first medium into the second medium. So, the same number of
photons are present; they are now spread over a larger solid angle. Like Equation A2.12, the
intensity of light measured by the detector is given by:
𝑘𝑘 0
≅
𝐼𝐼 𝜋𝜋 𝜃𝜃 0
2
Equation A2.13
Solving for 𝑘𝑘 in Equation A2.13 and substituting it into Equation A2.12, we get
𝑘𝑘 𝑎𝑎 =
𝐼𝐼 0
𝜃𝜃 0
2
𝜃𝜃 𝑟𝑟 2
Equation A2.14
Then, we apply Snell’s to find the effect of the refractive indices to get Equation A2.15:
𝑘𝑘 𝑎𝑎 = 𝑘𝑘 0
�
𝜃𝜃 0
2
𝜃𝜃 𝑟𝑟 2
�= 𝑘𝑘 0
�
𝑛𝑛 𝑟𝑟 2
𝑛𝑛 0
2
� Equation A2.15
Now that the refractive indices have been corrected for, they can be plugged into Equation A2.3.

Figure A0.1 – a) Refraction effect of light passing through cuvette after fluorescence. The red
circle is the fluorescence. Blue arrow is excitation light, and black arrows depict fluorescence
the subtended by the angle 𝜃𝜃 𝑎𝑎 . S is the point source. The internal angles and refractive indices
are indicated as well. Angles and distances are exaggerated. b) Depiction of a 1 steradian, Ω in
a circle with a radius, 𝑀𝑀 . Adopted from the Wikipedia article on Solid Angle.
Detector
a) b)
319

𝐸𝐸 = 𝑘𝑘 0
𝑓𝑓 Φ
𝑜𝑜 𝑓𝑓 �
𝑛𝑛 0
2
𝑛𝑛 𝑟𝑟 2
� Equation A2.16
Solving for Φ
𝑜𝑜 𝑓𝑓 we get:
Φ
𝑜𝑜 𝑓𝑓 =
𝐸𝐸 𝐼𝐼 0
𝑜𝑜 ∙ �
𝑛𝑛 0
2
𝑛𝑛 𝑟𝑟 2
� Equation A2.17
In order to get the final equation, Equation 3.1, used in chapter 3, we must divide Equation
A2.17 by itself where one equation is for the sample and the other is the reference.

Φ
𝑜𝑜𝑡𝑡 , 𝑆𝑆 Φ
𝑜𝑜𝑡𝑡 , 𝑅𝑅 =
𝐸𝐸 𝑆𝑆 𝑜𝑜 𝑅𝑅 𝑛𝑛 𝑆𝑆 2
𝐸𝐸 𝑅𝑅 𝑜𝑜 𝑆𝑆 𝑛𝑛 𝑅𝑅 2
Equation A2.10 (Equation 3.1)
The incoming light intensity, 𝑘𝑘 0
, and refractive index outside the cuvette, 𝑀𝑀 0
, will cancel. These
light intensities only cancel if excitation is at the same wavelength due to inherent variations in the
fluorimeter lamp intensity. Since the fluorescence light is isotropic, the total emission light
intensity and the amount measured by the detector are proportional to each other. This
proportionality factor is the same for both reference and sample due to same distance from detector,
so this factor cancels.
For much more detailed description of the method and of various approximations that are
currently made and the errors than can arise from them, we direct the reader to the review by
Demas and Crosby.
3
Figure A0.1 was adapted from this review.

320

3. Excited State Population Calculation from Experimental Parameters
The expected transient absorption signal is calculated by first calculating the number of
molecules which absorb a photon ( 𝐶𝐶 𝑎𝑎 𝑃𝑃 𝑃𝑃 ) by using the Beer Lambert law:

𝐼𝐼 𝑏𝑏 𝑟𝑟 𝐼𝐼 0
= 10
𝜀𝜀 ∙ 𝑓𝑓 ∙ 𝑐𝑐 = 10
− 𝑀𝑀 𝐺𝐺 𝑆𝑆 Equation A3.1

𝐼𝐼 𝐴𝐴 𝑃𝑃𝑃𝑃
𝐼𝐼 0
= 1 ―
𝐼𝐼 𝑏𝑏 𝑟𝑟 𝐼𝐼 0
= 1 ― 10
𝜀𝜀 𝐺𝐺 𝑆𝑆 ∙ 𝑓𝑓 ∙ 𝑐𝑐 𝐺𝐺 𝑆𝑆 Equation A3.2

𝐶𝐶 𝐴𝐴 𝑃𝑃𝑃𝑃
𝐶𝐶 0
=
𝐼𝐼 𝐴𝐴 𝑃𝑃𝑃𝑃
𝐼𝐼 0
Equation A3.3
where, 𝑘𝑘 𝑡𝑡 𝑟𝑟 is the intensity of the transmitted light, 𝑘𝑘 0
is the intensity of the incident light, 𝜀𝜀 𝐺𝐺 𝑆𝑆 is the
ground state molar absorptivity, in units of M
-1
cm
-1
, 𝑓𝑓 is the path length in units of cm, 𝑘𝑘 𝐺𝐺 𝑆𝑆 is the
concentration in units of M, 𝐴𝐴 𝐺𝐺 𝑆𝑆 is the ground state absorption, obtainable from a standard UV-vis
absorption experiment, 𝑘𝑘 𝑎𝑎 𝑃𝑃 𝑃𝑃 is the intensity of light absorbed. 𝐶𝐶 𝑎𝑎 𝑃𝑃 𝑃𝑃 and 𝐶𝐶 0
are the molecular
equivalents of 𝑘𝑘 0
and 𝑘𝑘 𝑎𝑎 𝑃𝑃 𝑃𝑃 in that they lay in the path of the light and absorbed a photon,
respectively. This assumes only 1 photon processes and that the molecule of interest is the only
light absorbing portion, neglecting solvent or sample cell absorption. The concentration of excited
state molecules 𝑘𝑘 𝐸𝐸 𝑆𝑆 is then:
𝑘𝑘 𝐸𝐸 𝑆𝑆 =
𝐶𝐶 𝐴𝐴 𝑃𝑃𝑃𝑃
𝑉𝑉 𝑝𝑝 =
𝐶𝐶 𝐴𝐴 𝑃𝑃𝑃𝑃
𝜋𝜋 ∙ 𝑟𝑟 2
∙ 𝑓𝑓 Equation A3.4
where 𝑉𝑉 𝑝𝑝 , the volume probed by the light beam and is assumed to be a cylinder of radius 𝑀𝑀 and
length 𝑓𝑓 . This approximation holds well for gaussian laser beams. The excited state optical density
can then be calculated as
𝐸𝐸 𝑘𝑘 𝑘𝑘𝑀𝑀 𝑘𝑘 𝑏𝑏 𝑀𝑀 𝑆𝑆 𝑘𝑘 𝑀𝑀 𝑘𝑘 𝑏𝑏 𝑚𝑚 𝑝𝑝 𝑘𝑘 𝑀𝑀𝑘𝑘𝑀𝑀𝑓𝑓 𝐶𝐶 𝑏𝑏 𝑀𝑀𝑠𝑠 𝑀𝑀 𝑘𝑘 𝑏𝑏 = 𝜀𝜀 𝐸𝐸 𝑆𝑆 ∙ 𝑓𝑓 ∙ 𝑘𝑘 𝐸𝐸 𝑆𝑆 Equation A3.6
where 𝜀𝜀 𝐸𝐸 𝑆𝑆 , is the excited state molar absorptivity – computed via quantum chemistry calculations.

321

4. Average Distance Between Molecules Using Concentration
Here, we detail a simple derivation for a calculation that can be used to approximate
distances between non-interacting molecules in solution. To begin, we calculate the number
density, 𝜌𝜌 𝐶𝐶 :
𝜌𝜌 𝐶𝐶 = 𝐶𝐶 𝑀𝑀 ∙ 𝑘𝑘 Equation A4.1
where 𝐶𝐶 𝑀𝑀 is Avagadro’s Number (molecules ∙ mol
-1
) and 𝑘𝑘 is the concentration (M). Then, we must
calculate the number of molecules in unit volume, 𝐶𝐶 𝑉𝑉 (molecules/nm
3
):
𝐶𝐶 𝑉𝑉 = 1000 ∙
𝜌𝜌 𝑁𝑁 ( 1 0
9
)
3
Equation A4.2
where 1000 converts from L to m
3
, 10
9
converts from m to nm, and the cubic factor in (10
9
)
3

converts nm to nm
3
. And finally, to obtain the distance between the molecules, 𝑀𝑀 𝑝𝑝𝑜𝑜𝑓𝑓𝑒𝑒 𝑐𝑐 , we take
the cube root of 𝐶𝐶 𝑉𝑉 :
𝑀𝑀 𝑝𝑝𝑜𝑜𝑓𝑓𝑒𝑒 𝑐𝑐 = ( 𝐶𝐶 𝑉𝑉 )
−
1
3
Equation A4.3
Or all together:
𝑀𝑀 𝑝𝑝𝑜𝑜𝑓𝑓𝑒𝑒 𝑐𝑐 ( 𝑀𝑀𝑀𝑀 ) = � 10
− 2 4
∙ 𝐶𝐶 𝑀𝑀 ∙ 𝑘𝑘 ( 𝐴𝐴 ) �
−
1
3
Equation A4.4
This value can be coupled with the diffusion constant for a molecule in a given solvent ( 𝐶𝐶 ) to
obtain an average time it will take two molecules to find each other, 𝑘𝑘 𝑟𝑟𝑎𝑎𝑜𝑜 𝑜𝑜 :
𝑘𝑘 𝑟𝑟𝑎𝑎𝑜𝑜 𝑜𝑜 =
𝑟𝑟 𝑝𝑝𝑜𝑜 𝑡𝑡𝑏𝑏 𝑟𝑟 2
2 ∙ 𝐷𝐷 Equation A4.5
Most often, literature values for 𝐶𝐶 are given in cm
2
s
-1
and diffusion times are on the order of
naonseconds so Equation AS3.6 is provided to account for the unit conversions:
𝑘𝑘 𝑟𝑟𝑎𝑎𝑜𝑜 𝑜𝑜 ( 𝑀𝑀𝑠𝑠 ) =
𝑟𝑟 𝑝𝑝𝑜𝑜 𝑡𝑡𝑏𝑏 𝑟𝑟 2
( 𝑛𝑛 𝑝𝑝 )
2 ∙ 𝐷𝐷 ( 𝑐𝑐 𝑝𝑝 2
∙ 𝑃𝑃 − 1
) ∙ �
1 0
9
𝑜𝑜𝑝𝑝 1 0
2
𝑟𝑟 𝑝𝑝 �
2
∙ �
1 𝑃𝑃 1 0
9
𝑜𝑜𝑃𝑃 �
= 10
5
∙
𝑟𝑟 𝑝𝑝𝑜𝑜 𝑡𝑡𝑏𝑏 𝑟𝑟 2
( 𝑛𝑛 𝑝𝑝 )
2 ∙ 𝐷𝐷 ( 𝑐𝑐 𝑝𝑝 2
∙ 𝑃𝑃 − 1
)
Equation A4.6

322

Appendix B: Matlab Codes
1. Colormap generation using HSV values for use in contour plots/2DTA
%% Defining and making colormap for use in TA/2D spectra, tested and working in
Matlab R2020b

clc; clear; % not required but I start all my matlab codes with it

x = 1:128; % sets number of colors. Matlab colormaps are defined as 256 colors,
smaller range for coarser colors, larger range for finer colors

size_x = length(x); % Just calculates length of color vector

center_x = 52; % Use this to plot where zero/white is. Useful for TA data with
both positive and negative signals. Plot data and play around with where zero
is. Smaller values correspond to small amounts of negative signal, and vice
versa for larger values. For equal \DeltaAbs values on either side of zero,
set center_x = size_x/2

%% Defining Hue Values
% Hue is the color of the rainbow, starting from and ending at red. Often hue
is defined by values from 1 to 255, a full circle. Going from red, orange,
yellow, green, blue, purple, pink, red.
% Personal bias – I like to define colors using hue, saturation, and value, NOT
red, green, blue (RGB). Easier to define personally. But, I included a line of
code that converts your hsv map to rgb.
% Hue here is defined as a piecewise linear function with center being zero at
center. This is because we want a linear change from one color to another,
usually red to blue

hue_y1 = 0.7; % Bluer if set higher
hue_y2 = 0.5; % Greener if set lower
hue_y3 = 0.0; % Red
hue_y4 = 0.25; % Orange
hue_x1 = x(1);
hue_x2 = x(center_x-1);
hue_x3 = x(end);
hue_x4 = x(center_x);
hue_m1 = (hue_y2 - hue_y1)/(hue_x2 - hue_x1); % calculates slope of hue line
for blue values
hue_m2 = (hue_y4 - hue_y3)/(hue_x4 - hue_x3); % Slope for red values
hue_fit1 = hue_m1*(x(hue_x1:hue_x2) - hue_x1) + hue_y1; % low values, red
hue_fit2 = hue_m2*(x(hue_x4:hue_x3) - hue_x4) + hue_y4;

hue_fit = [hue_fit1 hue_fit2];

%% Defining Saturation Values

% Saturation determines how vibrant or dull/muted a color is. Full saturation,
100 is a rainbow color and no, zero saturation is grey. Light from your laser
is 100% saturation.
% Saturation here is defined as a downside up gaussian (negative) with its min
at the "zero" point, ie in TA data where DelA = 0
323

% We use a gaussian to make sure saturation and value does not exceed 0 or 1
bounds.

sat_a = 1; % amplitude. Set to one so that when negated, values at zero are
grey (white if value = 100).
sat_width = 0.18*size_x; % Determines how wide your saturation gaussian is. The
wider, the more desaturated your colormap values will be
sat_offset = 1; % Brings “bottom” of gaussian to +1, so smallest saturation
value is zero.
sat_center = center_x;
r = 0.62; % skew factor. Values r <1 lead to desaturated reds and values r > 1
lead to desaturated blues. I like to bias against red personally (hence r < 1).
v = r^2*sat_width^2;

% To have an asymmetric gaussian, I split it down the center_x value. Hence the
need for concatenation of the two

sat_fit1 = -sat_a*exp(-(x(1:center_x-1) - sat_center).^2/(2*r^2*sat_width^2))
+ sat_offset;
sat_fit2 = -sat_a*exp(-(x(center_x:end) - sat_center).^2/(2*sat_width^2)) +
sat_offset;
sat_fit = [sat_fit1 sat_fit2];

%% Defining Value Values
% Value is how light or dark a color is. 100 is white, 0 is black
% Value is defined as a right side up Gaussian (positive)
val_a = 0.98; % Defines how high value goes (close to white)
val_cent_skew = 1.5; % shift center max value by certain amount. If < 1 -->
lighter blues
val_center = val_cent_skew*center_x; % Where center is
val_width = size_x*0.8;
val_offset = 0; % value go to zero at ends
val_fit = val_a*exp(-(x - val_center).^2/(2*val_width^2)) + val_offset;
% val_fit = ones(1,256);

close(figure(2));
figure(2);
hold on;
box on;
H1 = plot(hue_x1:hue_x2,hue_fit1,'r','LineWidth',3);
H2 = plot(hue_x4:hue_x3,hue_fit2,'r','LineWidth',3);
S1 = plot(sat_fit,'g','LineWidth',3);
V1 = plot(val_fit,'b','LineWidth',3);
Leg = legend([H1 S1 V1],{'Hue','Saturation','Value'});
set(gca,'FontSize',15,'LineWidth',3);
ylim([0 1]);
xlim([min(x) max(x)]);
xlabel('<-- Blue Red -->','FontSize',15);
hold off

mycolormaphsv = [hue_fit; sat_fit; val_fit]';
mycolormaprgb = hsv2rgb(mycolormaphsv);

save('My_TA_colormap_hsv.txt','mycolormaphsv','-ascii');
324

save('My_TA_colormap_rgb.txt','mycolormaprgb','-ascii');
2. Brewster Angle Calculations
clc; clear;
radtodeg = 180/pi;
FT = 15;

%% For n1 < n2

n1 = 1.0; n2 = 1.5;

phi0 = 0; phi1 = pi/2; dphi = pi/10000;

phi_inc_rad = (phi0:dphi:phi1-dphi)';
phi_inc_deg = radtodeg*phi_inc_rad;

phi_ref_rad = phi_inc_rad;
phi_ref_deg = phi_inc_deg;

phi_trans_rad_in = asin(n1/n2*sin(phi_inc_rad));
phi_trans_deg_in = phi_trans_rad_in*radtodeg;

Rs_in = (abs((n1*cos(phi_inc_rad) - n2*cos(phi_trans_rad_in))./...
(n1*cos(phi_inc_rad) + n2*cos(phi_trans_rad_in)))).^2;

Rp_in = (abs((n1*cos(phi_trans_rad_in) - n2*cos(phi_inc_rad))./...
(n1*cos(phi_trans_rad_in) + n2*cos(phi_inc_rad)))).^2;

Ts_in = 1 - Rs_in;
Tp_in = 1 - Rp_in;

Brewster_i_in = find(Rp_in == min(Rp_in));
Brewster_angle_in = phi_inc_deg(Brewster_i_in);

% Plot angles

close(figure(1));
figure(1);
hold on
ax = gca;
p1 = plot(phi_inc_deg,Rs_in,'b','LineWidth',3);
p2 = plot(phi_inc_deg,Rp_in,'r','LineWidth',3);
p3 = plot(phi_inc_deg,Ts_in,'b--','LineWidth',3);
p4 = plot(phi_inc_deg,Tp_in,'r--','LineWidth',3);
p5 = xline(Brewster_angle_in,'k-.','LineWidth',3);
box on;
ax.LineWidth = 3;
ax.FontSize = FT;
xlabel('Angle of Incidence','FontSize',FT);
ylabel('Reflection/Transmittance','FontSize',FT);
xlim([0 90]);
ylim([-0.01 1.01]);
xtickformat(ax,'degrees');
str1 = {'Brewster Angle =' num2str(round(Brewster_angle_in*100)/100)};
325

dim1 = [0.16 0.4 0.3 0.4];
ann1 = annotation('textbox',dim1,'String',str1,'FitBoxToText','on');
ann1.FontSize = FT;
ann1.LineWidth = 2.5;
str2 = 'degrees';
dim2 = [dim1(1)+0.1, dim1(2) - 0.06, 0.4, 0.4];
ann2 =
annotation('textbox',dim2,'String',str2,'FitBoxToText','on','LineStyle','none
');
ann2.FontSize = FT;
% ann2.LineWidth = 'none';
leg = legend([p1 p2 p3 p4 p5], {'S %R','P %R','S %T','P %T','Brewster
Angle'});
leg.Position = [0.22 0.33 0.2 0.2];
leg.FontSize = FT;
title(['n_{1} = ', num2str(n1), ' n_{2} = ', num2str(n2)]);
hold off

%% For n1 > n2

n1 = 1.5;
n2 = 1;

phi0 = 0; phi1 = pi/2; dphi = pi/10000;
phi_inc_rad = (phi0:dphi:phi1-dphi)';
phi_inc_deg = radtodeg*phi_inc_rad;

Figure B2.1 – Reflection/Transmission curves of light leaving a lower refractive index material
(n
1
= 1) and entering a higher refractive index material, (n
2
= 1.5). Note, only the P
polarized light has reflection of zero at the Brewster angle, denoted by the dashed-dotted dark
grey line.
0° 20° 40° 60° 80°
Angle of Incidence
0
0.2
0.4
0.6
0.8
1
Reflection/Transmittance
n
1
= 1 n
2
= 1.5
S %R
P %R
S %T
P %T
Brewster Angle
Brewster Angle =
56.3
degrees
326

phi_ref_rad = phi_inc_rad;
phi_ref_deg = phi_inc_deg;

phi_trans_rad_out = asin(n1/n2*sin(phi_inc_rad));
phi_trans_deg_out = phi_trans_rad_out*radtodeg;

Rs_out = (abs((n1*cos(phi_inc_rad) - n2*cos(phi_trans_rad_out))./...
(n1*cos(phi_inc_rad) + n2*cos(phi_trans_rad_out)))).^2;

Rp_out = (abs((n1*cos(phi_trans_rad_out) - n2*cos(phi_inc_rad))./...
(n1*cos(phi_trans_rad_out) + n2*cos(phi_inc_rad)))).^2;

Reff_out = 0.5*(Rs_out + Rp_out);

Ts_out = 1 - Rs_out;
Tp_out = 1 - Rp_out;

Teff_out = 0.5*(Ts_out + Tp_out);

Brewster_i_out = find(Rp_out == min(Rp_out));
Brewster_angle_out = phi_inc_deg(Brewster_i_out);

% Plot angles

close(figure(2));
figure(2);
hold on
ax = gca;
p1 = plot(phi_inc_deg,Rs_out,'b','LineWidth',3);
p2 = plot(phi_inc_deg,Rp_out,'r','LineWidth',3);
p3 = plot(phi_inc_deg,Ts_out,'b--','LineWidth',3);
p4 = plot(phi_inc_deg,Tp_out,'r--','LineWidth',3);
p5 = xline(Brewster_angle_out,'k--','LineWidth',3);
box on;
ax.LineWidth = 3;
ax.FontSize = FT;
xlabel('Angle of Incidence','FontSize',FT);
ylabel('Reflection/Transmittance','FontSize',FT);
xlim([0 90]);
ylim([-0.01 1.01]);
xtickformat(ax,'degrees');
leg = legend([p1 p2 p3 p4 p5], {'S %R','P %R','S %T','P %T','Brewster
Angle'},...
'Location', 'east');
leg.FontSize = FT;
hold off

327

%% Brewster Angle as a Function of Refractive index

clc;

% Fused Silica
% https://refractiveindex.info/?shelf=glass&book=fused_silica&page=Malitson
% I. H. Malitson. Interspecimen comparison of the refractive index of fused
silica,
% J. Opt. Soc. Am. 55, 1205-1208 (1965)

Material1 = 'Fused Silica';

w_start = 0.21; w_end = 6.7; w_step = 0.005;

w_um = (w_start:w_step:w_end)'; w_nm = w_um*1000;

B1_FS = 0.6961663; B2_FS = 0.4079426; B3_FS = 0.8974794;
C1_FS = 0.0684043; C2_FS = 0.1162414; C3_FS = 9.896161;

n2 = sqrt(1 + B1_FS./(1-(C1_FS./w_um).^2) + B2_FS./(1-(C2_FS./w_um).^2)...
+ B3_FS./(1-(C3_FS./w_um).^2));

n1 = ones(length(n2),1);

[n1mesh, phi_inc_mesh] = meshgrid(n1,phi_inc_rad);
[n2mesh, phi_inc_mesh] = meshgrid(n2,phi_inc_rad);

Figure B2.2 – Reflection/Transmission curves of light leaving a higher refractive index material
(n
1
= 1.5) and entering a lower refractive index material, (n
2
= 1). Note, only the P polarized
light has reflection of zero at the Brewster angle, denoted by the dashed-dotted dark grey line.
Additionally, we can see the occurrence of total internal reflection (TOI) at an AOI of 41.6
o
.
0° 20° 40° 60° 80°
Angle of Incidence
0
0.2
0.4
0.6
0.8
1
Reflection/Transmittance
n
1
= 1.5 n
2
= 1
S %R
P %R
S %T
P %T
Brewster Angle
328

phi_trans_mesh_in = asin(n1mesh./n2mesh.*sin(phi_inc_rad));
phi_trans_deg_in = phi_trans_rad_in*radtodeg;

close(figure(4));
fig = figure(4);
Allnaxis = axes('Parent', fig,'Position', [0.15, 0.15, 0.75, 0.75]);
hold(Allnaxis, 'on');
p1 = plot(w_nm,n2,'LineWidth',3);
xlabel('Wavelength (nm)','FontSize',FT);
ylabel('Refractive Index','FontSize',FT);
Allnaxis.FontSize = FT;
box on;
Allnaxis.LineWidth = 3;
title(Material1);
xlim([100 7000]);
visWLs = axes('Parent', fig, 'Position', [0.25, 0.27, 0.3, 0.3]);
hold(visWLs, 'on');
box(visWLs, 'on');
q1 = plot(w_nm,n2,'LineWidth',3);
xlim([200 800]);
xlabel('Wavelength (nm)','FontSize',FT-3);
ylabel('Refractive Index','FontSize',FT-3);
visWLs.LineWidth = 2;

%% Wavelength dep n and Brewster Angles

Rs_in_mesh = (abs((n1mesh.*cos(phi_inc_mesh) -
n2mesh.*cos(phi_trans_mesh_in))./...
(n1mesh.*cos(phi_inc_mesh) +
n2mesh.*cos(phi_trans_mesh_in)))).^2;

Rp_in_mesh = (abs((n1mesh.*cos(phi_trans_mesh_in) -
n2mesh.*cos(phi_inc_mesh))./...
(n1mesh.*cos(phi_trans_mesh_in) +
n2mesh.*cos(phi_inc_mesh)))).^2;

Ts_in = 1 - Rs_in_mesh; Tp_in = 1 - Rp_in_mesh;

min_Rp_each_n = min(Rp_in_mesh);
[Brew_each_n, min_ang] = find(Rp_in_mesh == min_Rp_each_n);

w_Brew = phi_inc_deg(Brew_each_n);

close(figure(5));
fig = figure(5);
Allnaxis = axes('Parent', fig,'Position', [0.15, 0.15, 0.72, 0.75]);
hold(Allnaxis, 'on');
yyaxis(Allnaxis,'left');
p1 = plot(w_nm,n2,'LineWidth',3);
xlabel('Wavelength (nm)','FontSize',FT);
ylabel('Refractive Index','FontSize',FT);
Allnaxis.FontSize = FT;
box on;
Allnaxis.LineWidth = 3;
329

title(Material1);
xlim([100 7000]);
yyaxis(Allnaxis,'right');
plot(w_nm,w_Brew,'LineWidth',3);
ylabel('Brewster Angle');
ytickformat(Allnaxis,'degrees');
ylim([48 58.5]);
visWLs = axes('Parent', fig, 'Position', [0.25, 0.27, 0.3, 0.3]);
yyaxis(visWLs,'left');
hold(visWLs, 'on');
box(visWLs, 'on');
q1 = plot(w_nm,n2,'LineWidth',3);
xlim([200 800]);
xlabel('Wavelength (nm)','FontSize',FT-3);
ylabel('Refractive Index','FontSize',FT-3);
visWLs.LineWidth = 2;
yyaxis(visWLs, 'right');
plot(w_nm,w_Brew,'LineWidth',3);
ylabel('Brewster Angle');
ytickformat(Allnaxis,'degrees');
ylim([55.4 57.1]);
ytickformat(visWLs,'degrees');

Figure B2.3 – The refractive index (left y-axis, blue) and Brewster angle (right y-axis, red) of
fused silica as a function of wavelength of radiation. The inset focuses on the DUV-NIR region.
1000 2000 3000 4000 5000 6000 7000
Wavelength (nm)
1.1
1.2
1.3
1.4
1.5
1.6
Refractive Index
48°
50°
52°
54°
56°
58°
Brewster Angle
Fused Silica
200 400 600 800
Wavelength (nm)
1.45
1.5
1.55
Refractive Index
55.5°
56°
56.5°
57°
Brewster Angle
330

3. Photoluminescence Correction of nsTA Data from using psTA data
%% Resubtracting the PL from the TA to correct for PL artifact
% This was run successfully in MATLAB 2023a with the Curve-Fitting Toolbox add-on

clc; clear; close all;

smpt = 3; % Smoothing parameter
t0 = 2; % time zero offset. Close to 0.
LW = 3; % LineWidth variable for Matlab plots
FT = 15; % FontSize variable for Matlab plots

% Just some nice colors I've found
MikeBlack = [0.1, 0.1, 0.1];
MikeRed = [0.9, 0, 0.0];
MikeBlue = [0.2, 0.3, 0.8];
MikeGreen = [0.38, 0.5, 0.25];
MikeOrange = [0.8824, 0.6118, 0.1412];
MikeGold = [0.85, 0.856, 0.07];
MikeViolet = 0.5*(MikeRed + MikeBlue);

colmap_Mike = [MikeBlack; MikeRed; MikeOrange; MikeGold; MikeGreen; MikeBlue;
MikeViolet];
colmap1 = colmap_Mike;

%% Data Loading

% Here, the .mat file produced directly from the nsTA with PL sub should be put in
between the ' '. It is a struct file
nsTA_withPLsub = load('nsTA_data_PLsub.mat');
dataTA = nsTA_withPLsub.data; % the 2D dataset with PL sub.
t = dataTA(3:end,1); % time vector in nanoseconds
w = dataTA(1,2:end); % wavelength vector in nanometer
DeltaAwithPLsub = dataTA(3:end,2:end); % The DeltaA values in OD, not mOD
PLwithPLsub = nsTA_withPLsub.dataPL; % The PL 2D dataset
PL = PLwithPLsub(2:end,2:end); % The isolated intensity values of the PL in mV
T = dataTA(2,2:end); % transmission vector in mV as a function of wavelength
Trans = ones(length(t),1)*T; % transmission vector extended to time so that the
transmission, PL, nsTA, etc have all the same dimensions

% This is the psTA spectral slice at a long delay time obtained from ultrafast data.
% Make sure these are only wavelength-DeltaA pairs. Dont include the time/extra
entries in the data file unless the (1:end) argument is
% changed to reflect only wavelength-DeltaA pairs
fsTA1ns = load('psTA_1ns_spectrum.txt');
fs_w = fsTA1ns(1:end,1); % femto/picosecond wavelength in nm
fs_A = fsTA1ns(1:end,2); % femto/picosecond DeltaA in mOD

% From Equation 2.5
DeltaT = Trans.*(10.^(-DeltaAwithPLsub) - 1) + PL; % Obtain the DeltaT, change in
transmission values from the PL and transmission values

331

%% Plot Time Slices

PLScale = 1.02; % The "s" value detailed in the PL subtraction section, chapter 2.
Please read for full description
fsScale = 1; % a scaling factor to use to match the femtosecond TA slice with the
nsTA. This value is not used in any other calculations and is simply for display
purposes when comparing the psTA to the nsTA
fsScaled = fsScale*fs_A;

% From Equation 2.4
DeltaA_PLScaled = -log10(1 + (DeltaT - PLScale*PL)./Trans)*1000; % Recalculate the
nsTA DeltaA (mOD) values after manually scaling the PL
TA = DeltaA_PLScaled;

% spectral slices chosen at given times to plot Desired time points in ns. Make sure
to match them with exact time points in the time axis
t_p1 = 10;
t_p2 = 50;
t_p3 = 100;
t_p4 = 200;
t_p5 = 500;
t_p6 = 1000;

% Time points vector
t_points = [0 t_p1 t_p2 t_p3 t_p4 t_p5 t_p6];

% finds indices of desired time points
t_i1 = find(t == t_p1);
t_i2 = find(t == t_p2);
t_i3 = find(t == t_p3);
t_i4 = find(t == t_p4);
t_i5 = find(t == t_p5);
t_i6 = find(t == t_p6);

% Finds the spectral traces of the nsTA, then averages three slices together, then
applies a smoothing function, the number of points it smooths over is determined by
smpt. Remove this by setting smpt = 1 or using the next set of traces that are
commented out.
specTA1 = smooth(mean(TA(t_i1-1:t_i1+1,:),1),smpt);
specTA2 = smooth(mean(TA(t_i2-1:t_i2+1,:),1),smpt);
specTA3 = smooth(mean(TA(t_i3-1:t_i3+1,:),1),smpt);
specTA4 = smooth(mean(TA(t_i4-1:t_i4+1,:),1),smpt);
specTA5 = smooth(mean(TA(t_i5-1:t_i5+1,:),1),smpt);
specTA6 = smooth(mean(TA(t_i6-1:t_i6+1,:),1),smpt);

% specTA1 = (mean(TA(t_i1-1:t_i1+1,:),1))';
% specTA2 = (mean(TA(t_i2-1:t_i2+1,:),1))';
% specTA3 = (mean(TA(t_i3-1:t_i3+1,:),1))';
% specTA4 = (mean(TA(t_i4-1:t_i4+1,:),1))';
% specTA5 = (mean(TA(t_i5-1:t_i5+1,:),1))';
% specTA6 = (mean(TA(t_i6-1:t_i6+1,:),1))';

% Sets an easy matrix that can be copied into a spreadsheet program. Used with Origin
2020b to plot the cMa data
specTAall = [specTA1 specTA2 specTA3 specTA4 specTA5 specTA6];
332

a_copy_spec = [t_points; [w' specTAall]];
% Plots spectral traces of a set of spectral slices
close(figure(1));
figure(1);
ax1 = gca;
box on;
p0 = plot(w,w.*0,'k:','linewidth',LW); hold on;
p1 = plot(w,specTA1,'Color',colmap1(1,:),'LineWidth',LW); hold on;
p2 = plot(w,specTA2,'Color',colmap1(2,:),'LineWidth',LW); hold on;
p3 = plot(w,specTA3,'Color',colmap1(3,:),'LineWidth',LW); hold on;
p4 = plot(w,specTA4,'Color',colmap1(4,:),'LineWidth',LW); hold on;
p5 = plot(w,specTA5,'Color',colmap1(5,:),'LineWidth',LW); hold on;
p6 = plot(w,specTA6,'Color',colmap1(6,:),'LineWidth',LW); hold on;
p7 = plot(fs_w,fsScaled,':','Color',colmap1(1,:),'LineWidth',LW*1.2); hold on;
xlabel('Wavelength (nm)');
ylabel('\DeltaAbs (mOD)');
xlim([min(w) max(w)]);
set(ax1,'LineWidth',LW,'FontWeight','bold','FontSize',FT);
L1 = legend([p1 p2 p3 p4 p5 p6],{[num2str(t_p1) ' ns'], [num2str(t_p2) ' ns'],...
[num2str(t_p3) ' ns'], [num2str(t_p4) ' ns'], ...
[num2str(t_p5) ' ns'], [num2str(t_p6) ' ns']},...
'Location','Northeast');
title('Molecule, Solvent, Exp Conditions, etc');

%% Time Slices

% wavelength points chosen to plot the time slices for those wavelengths
w_p1 = 460;
w_p2 = 580;
w_p3 = 680;
w_p4 = 880;

w_points = [0 w_p1 w_p2 w_p3 w_p4];

w_i1 = find(w == w_p1);
w_i2 = find(w == w_p2);
w_i3 = find(w == w_p3);
w_i4 = find(w == w_p4);

kineTA1 = smooth(mean(TA(:,w_i1-1:w_i1+1),2),smpt);
kineTA2 = smooth(mean(TA(:,w_i2-1:w_i2+1),2),smpt);
kineTA3 = smooth(mean(TA(:,w_i3-1:w_i3+1),2),smpt);
kineTA4 = smooth(mean(TA(:,w_i4-1:w_i4+1),2),smpt);

timeall = [kineTA1 kineTA2 kineTA3 kineTA4];
a_copy_time = [w_points; [t timeall]];

close(figure(2));
figure(2);
ax1 = gca;
box on;
p1 = plot(t,kineTA1,'Color',colmap1(1,:),'LineWidth',LW); hold on;
p2 = plot(t,kineTA2,'Color',colmap1(2,:),'LineWidth',LW); hold on;
p3 = plot(t,kineTA3,'Color',colmap1(3,:),'LineWidth',LW); hold on;
p4 = plot(t,kineTA4,'Color',colmap1(4,:),'LineWidth',LW); hold on;
333

xlabel('Time (ns)');
ylabel('\DeltaAbs (mOD)');
xlim([min(t) max(t)]);
ylim([-0.5 1.5]);
set(ax1,'LineWidth',LW,'FontWeight','bold','FontSize',FT);
L1 = legend([p1 p2 p3 p4],{[num2str(w_p1) ' nm'], [num2str(w_p2) ' nm'],...
[num2str(w_p3) ' nm'], [num2str(w_p4) ' nm']},...
'Location','Northeast');
title('Molecule, Solvent, Exp Conditions, etc');

%% PL Spectrum

PLslice1 = PL(t_i1,:);
PLslice2 = PL(t_i2,:);
PLslice3 = PL(t_i3,:);
PLslice4 = PL(t_i4,:);
PLslice5 = PL(t_i5,:);
PLslice6 = PL(t_i6,:);

close(figure(3));
figure(3);
ax1 = gca;
box on;
p0 = plot(w,w.*0,'k:','linewidth',LW); hold on;
p1 = plot(w,PLslice1,'Color',colmap1(1,:),'LineWidth',LW); hold on;
p2 = plot(w,PLslice2,'Color',colmap1(2,:),'LineWidth',LW); hold on;
p3 = plot(w,PLslice3,'Color',colmap1(3,:),'LineWidth',LW); hold on;
p4 = plot(w,PLslice4,'Color',colmap1(4,:),'LineWidth',LW); hold on;
p5 = plot(w,PLslice5,'Color',colmap1(5,:),'LineWidth',LW); hold on;
p6 = plot(w,PLslice6,'Color',colmap1(6,:),'LineWidth',LW); hold on;
xlabel('Wavelength (nm)','FontSize',FT);
ylabel('PL Intensity (mV)','FontSize',FT);
xlim([min(w) max(w)]); % ylim([-0.5 1.5]);
set(ax1,'LineWidth',LW,'FontWeight','bold','FontSize',FT);
L1 = legend([p1 p2 p3 p4 p5 p6],{[num2str(t_p1) ' ns'], [num2str(t_p2) ' ns'],...
[num2str(t_p3) ' ns'], [num2str(t_p4) ' ns'], ...
[num2str(t_p5) ' ns'], [num2str(t_p6) ' ns']},...
'Location','Northeast');
title('Molecule, Solvent, Exp Conditions, etc');

specPLall = [PLslice1; PLslice2; PLslice3; PLslice4; PLslice5; PLslice6]';
a_copy_spec_PL = [t_points; [w' specPLall]];

%% PL Traces/lifetime

kinePL1 = smooth(mean(PL(:,w_i1-1:w_i1+1),2),smpt);
kinePL2 = smooth(mean(PL(:,w_i2-1:w_i2+1),2),smpt);
kinePL3 = smooth(mean(PL(:,w_i3-1:w_i3+1),2),smpt);
kinePL4 = smooth(mean(PL(:,w_i4-1:w_i4+1),2),smpt);

timeall = [kinePL1 kinePL2 kinePL3 kinePL4];
a_copy_time_PL = [w_points; [t timeall]];

close(figure(4));
334

figure(4);
ax1 = gca;
box on;
p1 = plot(t,kinePL1,'Color',colmap1(1,:),'LineWidth',LW); hold on;
p2 = plot(t,kinePL2,'Color',colmap1(2,:),'LineWidth',LW); hold on;
p3 = plot(t,kinePL3,'Color',colmap1(3,:),'LineWidth',LW); hold on;
p4 = plot(t,kinePL4,'Color',colmap1(4,:),'LineWidth',LW); hold on;
xlabel('Time (ns)');
ylabel('PL Intensity (mV)');
xlim([min(t) max(t)]);
% ylim([-0.5 1.5]);
set(ax1,'LineWidth',LW,'FontWeight','bold','FontSize',FT);
set(ax1,'YScale','Log');
L1 = legend([p1 p2 p3 p4],{[num2str(w_p1) ' nm'], [num2str(w_p2) ' nm'],...
[num2str(w_p3) ' nm'], [num2str(w_p4) ' nm']},...
'Location','Northeast');
title('Molecule, Solvent, Exp Conditions, etc');

335

4. Transient Absorption Simulation for a System with Two Excited States
Here, we introduce a MATLAB code that can be used to simulate TA spectra for both one
and two excited state systems. The ground state is not counted as a state here. First, we will explore
a “generic” case with display of the full code. Then we will explore several more case studies, but
only displaying the results from the code. The code will remain identical save for the inputs for
spectral and kinetic parameters and minor formatting changes. Readers are encouraged to follow
along with the provided code as well as further experiment with it. While an attempt to be fully
descriptive has been made, not all features of a psTA spectrum is presented, those like time zero
artifacts, i.e. two-photon absorption, cross-phase modulation, or white light chirp that would be
present in a raw TA dataset.
To begin, this imaginary system has kinetic processes in the ultrafast regime and has
ground and excited state absorption and emission in the visible range, hence the choice for the
wavelength probe and time ranges, respectively. This is purely by choice and can be changed as
desired. This family of systems can be described by the energy diagram in Figure B4.1. This
diagram bears strong resemblance to the SBCT (Figure 3.1, top) and cMa compounds (Figure
4.7), with only the names and values between the species changing with the underlying math being
identical. We can think of 𝑘𝑘 1 1 𝑎𝑎 and 𝑘𝑘 1 1 𝑃𝑃 as the radiative ( 𝑘𝑘 𝑟𝑟 ) and non-radiative ( 𝑘𝑘 𝑛𝑛 𝑟𝑟 ) rates from
State 1, respectively. The rate 𝑘𝑘 1 2
is some forward population transfer rate from State 1 to State 2;
𝑘𝑘 1 2
can be thought of as either a forward charge transfer rate or a singlet to triplet intersystem
crossing rate. The rate 𝑘𝑘 2 1
is the “reverse” of that: back electron transfer to State 1 or triplet to
singlet intersystem crossing. And finally, rate 𝑘𝑘 2 2
is the rate from State 2 back to the ground state.
It can be thought of as charge recombination back to the ground state or triplet
336

emission/phosphorescence. These need not be the specific physical interpretation of these
parameters and a different physical interpretation can be thought of for the specific system.
For the first case, Case #1, we have a system with an emissive first excited state ( Φ
𝑜𝑜 𝑓𝑓 =
0.2 with State 2 not present) with 𝑘𝑘 1 1 𝑎𝑎 = 2.5 ∙ 10
− 4
𝑝𝑝 𝑠𝑠 − 1
and 𝑘𝑘 1 1 𝑃𝑃 = 1.0 ∙ 10
− 4
𝑝𝑝 𝑠𝑠 − 1
. We
establish an excited state equilibrium and transfer excited state concentration between State 1 and
State 2 by 𝑘𝑘 1 2
= 0.1 𝑝𝑝 𝑠𝑠 − 1
and 𝑘𝑘 2 1
= 0.05 𝑝𝑝 𝑠𝑠 − 1
. This equilibrium favors State 2 based on the
faster rate constant for transfer from State 1 to State 2 than the State 2 to State 1 transfer rate, i.e.,
𝑘𝑘 1 2
= 2 ∙ 𝑘𝑘 2 1
. Therefore, the equilibrium constant is 𝐾𝐾 𝑒𝑒 𝑞𝑞 = 2. For our first sanity check, we
should see the excited state concentrations reflect this ratio. Discussion of the results will be done
after the code. Now, let us look at the code in full. Comments are provided to guide use of the
code. This code was tested and working as displayed in MATLAB 2023a and generated the figures
depicted in this section.

Figure B4.1 – State diagram depicting the generic two state system with rate constants defined.
State 1
State 2
337

%% Code Start
%% MATLAB code for creating a simulated TA spectra for two excited states

clear; clc; close all;

LW = 3; % LineWidth
FT = 20; % FonTsize

% Colors used
MikeBlack = [0.1 0.1 0.1];
MikeRed = [0.9,0,0.0];
MikeBlue = [0.2,0.3,0.8];
MikeGreen = [0.38, 0.5, 0.25];
MikeOrange = [0.8824, 0.6118, 0.1412];
MikeGold = [0.85,0.856,0.07];
MikeViolet = 0.5*(MikeRed + MikeBlue);

colmap_Mike = [MikeBlack; MikeRed; MikeOrange; MikeGold; MikeGreen; MikeBlue;
MikeViolet];

%% Defining time and wavelength vectors

% Wavelength - a single linearly spaced array, in the visible, often from
~300 to 700 nm, will vary
w = (300:1:700)'; % Wavelength axis in nm

% Time - as TA experiments often go over multiple decades of time, time can
be defined logarithmically. However, it is more convienent to define it as a
set of time blocks
% tx = start:step:stop

% picosecond range
t1 = -2:0.100:-0.80;
t2 = -0.75:0.050:-0.55;
t3 = -0.5:0.020:-0.1;
t4 = -.09:0.01:0.75;
t5 = 0.8:0.05:2.5;
t6 = 2.75:0.25:10;
t7 = 12:2:100;
t8 = 110:10:300;
t9 = 325:25:1000;
t10 = 1100:100:1900;

% nanosecond range, extend as necessary or desired
t11 = 2000:250:10000;

t = [t1 t2 t3 t4 t5 t6 t7 t8 t9 t10 t11]; % time axis in ps

%% Defining the Amplitudes/Species Associated Decay Spectra for TA signals as
simple gaussians

% Every TA spectrum is composed of these three components:
% Ground state bleach - gsb
% Excited state absorption - esa
338

% Stimulated emission - se

% amplitudes; reminder - esa values are positive and gsb/se values are
negative
% Add as many gaussians as you wish to simulate your desired result, keep
track of indices

% Here, let us say we have a two excited state system that is entirely
fictional. Excitation of this molecule leads to full population of excited
% State 1 with no initial population of state 2. States 1 and 2 are denoted
by a 1 and 2 following each variable, respectivly.
% State 1 has a large gsb with some se and an esa. However, state 2 has a
larger gsb due to some feature of this state depleting more ground states.
This state also does not emit (se2 = 0). It has two excited state absorptions
as well.
% This is a mock for SBCT which displays similar behavior: bright initial
state, non-emissive second state with
% absorptions from both the radical cation and radical anion. Same for cMa
as well

S1_0=1; %initial population of S1 state

% Amplitudes of specific TA features
a_gsb1 = -8;
a_gsb2 = -9;
a_se1 = -4;
a_se2 = 0;
a_esa1 = 5;
a_esa2a = 4;
a_esa2b = 3;

% Center of signal band in wavelength

w0_gsb1 = 500;
w0_gsb2 = 500;
w0_se1 = 530;
w0_se2 = 530;
w0_esa1 = 330;
w0_esa2a = 590;
w0_esa2b = 400;

% Width of signal band in nm; here as full-width half max

FWHM_gsb1 = 70;
FWHM_gsb2 = 90;
FWHM_se1 = 80;
FWHM_se2 = 90;
FWHM_esa1 = 50;
FWHM_esa2a = 70;
FWHM_esa2b = 70;

% Calculating gaussians of each signal

GSB1 = a_gsb1*exp(-0.5*(w - w0_gsb1).^2/(FWHM_gsb1/2.355)^2);
GSB2 = a_gsb2*exp(-0.5*(w - w0_gsb2).^2/(FWHM_gsb2/2.355)^2);
339

SE1 = a_se1 *exp(-0.5*(w - w0_se1 ).^2/(FWHM_se1 /2.355)^2);
SE2 = a_se2 *exp(-0.5*(w - w0_se2 ).^2/(FWHM_se2 /2.355)^2);
ESA1 = a_esa1*exp(-0.5*(w - w0_esa1).^2/(FWHM_esa1/2.355)^2);
ESA2a = a_esa2a*exp(-0.5*(w - w0_esa2a).^2/(FWHM_esa2a/2.355)^2);
ESA2b = a_esa2b*exp(-0.5*(w - w0_esa2b).^2/(FWHM_esa2b/2.355)^2);

% Summing together signals to create SADS

SADS1 = GSB1 + SE1 + ESA1;
SADS2 = GSB2 + SE2 + ESA2a + ESA2b;

% Plots SADS
close(figure(1));
fig1 = figure('Units', 'inches', 'Position', [5, 5, 6.3, 5]);
box on;
hold on
g1 = plot(w,GSB1,'--','Color',MikeBlue,'LineWidth',LW);
g2 = plot(w,GSB2,'--','Color',MikeRed,'LineWidth',LW);
s1 = plot(w,SE1,':','Color',MikeBlue,'LineWidth',LW);
s2 = plot(w,SE2,':','Color',MikeRed,'LineWidth',LW);
e1 = plot(w,ESA1,'-.','Color',MikeBlue,'LineWidth',LW/2);
e2a = plot(w,ESA2a,'-.','Color',MikeRed,'LineWidth',LW/2);
e2b = plot(w,ESA2b,'-.','Color',MikeRed,'LineWidth',LW/2);
p1 = plot(w,SADS1,'Color',MikeBlue,'LineWidth',LW);
p2 = plot(w,SADS2,'Color',MikeRed,'LineWidth',LW);
xlabel('Wavelength (nm)','FontSize',FT); ylabel('\DeltaAbs
(mOD)','FontSize',FT);
xlim([min(w) max(w)]);
ylim([-12 6]);
legend([p1 p2],'State 1', 'State
2','LineWidth',LW,'FontSize',FT,'Location','SouthEast');
set(gca,'FontSize',FT,'LineWidth',LW);
title('SADS');
hold off

%% Kinetic Parameters

% Now we define the kinetic parameters for this two state model. Let us say
we have a fluorescent molecule with some radiative and non-radiative rates.
Then, we have a rate constant that transfers population from state 1 to state
2.
% A lot of molecules establish an excited state equilibrium between two
states, so we also add a "reverse" rate constant.
% Additionally, state 2 can also relax to the ground state so we add a rate
constant for that. Readers are encouraged to change these parameters as
desired.

% We first define the time constants for these rate constants as time
constants are used as the intuitive value to assign to a physical process.
If the reader prefers to use only rate constants, they are encouraged to do
so

%time constants of corresponding lifetimes (ps)
tau11a = 4000; % radiative lifetime (goes from state 1 to ground state)
tau11b = 10000; % non radiative lifetime (goes from state 1 to ground state)
340

tau12 = 10; % state 1 to state 2 lifetime
tau21 = 20; % state 2 "back" to state 2 lifetime
tau22 = 1000; % Recombination/decay from state 2 (seen as SBCT krec or cMa
triplet decay)

%% Kinetics

% Define rate constants

% Rate constants (ps^(-1))
k11a = 1/tau11a; %radiative decay to ground state (state 0)
k11b = 1/tau11b; %non radiative decay to S0
k12 = 1/tau12; % population transfer to state 1
k21 = 1/tau21; %back population transfer to state 0
k22 = 1/tau22; %charge recombination from state 1 to ground state

Keq = k12/k21;

% The following is the kinetic solution to the described above state model.
For further reading, the author directs readers to: Veldman, D., J. Phys.
Chem. A 2008, 112, 37, 8617–8632.
% Essentially, the model has two time constants that define it: a short and
long time component. These components in turn define amplitudes for the state
populations

%time constants
tau1 = 1/(0.5*(k11a+k11b+k12+k21+k22-sqrt((k21+k22-k11a-k11b-
k12)^2+4*k12*k21))); % long time constant
tau2 = 1/(0.5*(k11a+k11b+k12+k21+k22+sqrt((k21+k22-k11a-k11b-
k12)^2+4*k12*k21))); % short time constant

A1=((tau2)^(-1)-k11a-k11b-k12)/((tau2)^(-1)-(tau1)^(-1)); %coefficient long
time exponential
A2=(k11a+k11b+k12-(tau1)^(-1))/((tau2)^(-1)-(tau1)^(-1)); %coefficient short
time exponential
A3=k12/(tau2^(-1)-tau1^(-1)); %coefficient of fitting from ps-PIA

%% Defining Variables of IRF
%Attempting to add convolution/instrument response to data
%Assume Gaussian IR for TA data is sum of three

t0 = 0.0; % Center of IRF (t0)
FWHM = 0.30; % width of Guassian IRF (ps)
a_IR = 0.5; %amplitude of Gaussian IR in time
c_IR = FWHM/2.355;

IR_1 = a_IR*exp((-(t - t0).^2/(2*(c_IR)^2))); %IR as Gaussian - time

%% Table of parameters, summary

Kinetic_Rates = ["k_r";"knr";"k12";"k21";"k22";"K_eq"];
Values = [k11a; k11b; k12; k21; k22; Keq];

% Summary of rate constants
341

table_constants = rows2vars(table(Kinetic_Rates, Values));
table2 = table_constants(:,2:end);

%% Paramaters for analytical convolution

B1 = A1/(A1+A2);
B2 = A2/(A1+A2);
B3 = S1_0*A3;
C1 = a_IR;
sigma = sqrt((c_IR)^2/2);

t_split = 1; % Point where graphs go from linear to logarithmic in time
AmpScale = 1; %Scaling factor for Data Amplitude

%% First state analytical convolution solution population

[tmesh, S1_Ak] = meshgrid(t,SADS1);
[tmesh, S2_Ak] = meshgrid(t,SADS2);
S1_A = S1_Ak;
S2_A = S2_Ak;
S1_a1 = B1*C1*(exp(sigma^2/(2*tau1^2)-((tmesh-t0)/tau1)).*(1-erf((sigma^2-
tau1*(tmesh-t0))/(sqrt(2)*sigma*tau1))));
S1_a2 = B2*C1*(exp(sigma^2/(2*tau2^2)-((tmesh-t0)/tau2)).*(1-erf((sigma^2-
tau2*(tmesh-t0))/(sqrt(2)*sigma*tau2))));
S1_T = S1_0*(S1_a1 + S1_a2);

%% Second state analytical convolution solution population

S2_a1 = B3*C1*(exp(sigma^2/(2*tau1^2)-((tmesh-t0)/tau1)).*(1-erf((sigma^2-
tau1*(tmesh-t0))/(sqrt(2)*sigma*tau1))));
S2_a2 = B3*C1*(exp(sigma^2/(2*tau2^2)-((tmesh-t0)/tau2)).*(1-erf((sigma^2-
tau2*(tmesh-t0))/(sqrt(2)*sigma*tau2))));
S2_T = S1_0*(S2_a1 - S2_a2);

%% Bringing Temporal and Spectral features together

S1Sim = S1_A.*S1_T;
S2Sim = S2_A.*S2_T;
SimTA = AmpScale*(S1Sim + S2Sim);

%% Lin-Log times, simulated and data values

t_s = find(t == t_split); % Don't change
t_Lin = t(1:t_s); t_Log = t(t_s+1:end);
S1_Lin = S1_T(1,1:t_s); S1_Log = S1_T(1,t_s+1:end);
S2_Lin = S2_T(1,1:t_s); S2_Log = S2_T(1,t_s+1:end);

%% Simulated Concentrations

close(figure(2));
fig2 = figure('Units', 'inches', 'Position', [5, 5, 6.3, 5]);
ax1 = subplot(121);
hold on
342

p1 = plot(t_Lin(t_Lin <= t_split),S1_Lin(t_Lin <=
t_split),'LineWidth',LW,'Color',MikeBlue);
p2 = plot(t_Lin(t_Lin <= t_split),S2_Lin(t_Lin <=
t_split),'LineWidth',LW,'Color',MikeRed);
ylabel('Exc State Concentration');
xlabel_pos = get(get(ax1, 'XLabel'), 'Position');
new_xlabel_pos = xlabel_pos + [3.7 -0.08 0];
set(get(ax1, 'XLabel'), 'Position', new_xlabel_pos);
xlabel('Time (ps)', 'Position', new_xlabel_pos);
title('Excited State Concentrations', 'Position', new_xlabel_pos + [0.2 1.14
0]);
ax2 = subplot(122);
hold on
p3 = plot(t_Log(t_Log >= t_split),S1_Log(t_Log >=
t_split),'LineWidth',LW,'Color',MikeBlue);
p4 = plot(t_Log(t_Log >= t_split),S2_Log(t_Log >=
t_split),'LineWidth',LW,'Color',MikeRed);
legend('State 1','State
2','Location','NorthEast','LineWidth',LW/1.5,'FontSize',FT);
ax1pos = [0.15, 0.18, 0.18, 0.72];
ax2pos = [ax1pos(1)+ax1pos(3), ax1pos(2), ax1pos(3)*3.4, ax1pos(4)];
set(ax1,'position',ax1pos,'xlim',[-t_split t_split],'ylim',[0 1]);
set(ax2,'position',ax2pos,'ylim',[0 1]);
set(ax2,'xscale','log','xlim',[t_split t(end)],'yticklabel','');
xticks([10 100 1000 10000]);
xticklabels({'10^{1}','10^{2}','10^{3}','10^{4}'});
set([ax1 ax2],'LineWidth',LW,'FontSize',FT,'box','on');
hold off

%% Spectral Traces

t_p1 = 1; t_i1 = find(t == t_p1);
t_p2 = 10; t_i2 = find(t == t_p2);
t_p3 = 100; t_i3 = find(t == t_p3);
t_p4 = 500; t_i4 = find(t == t_p4);
t_p5 = 1000; t_i5 = find(t == t_p5);
t_p6 = 3000; t_i6 = find(t == t_p6);

spec1 = SimTA(:,t_i1);
spec2 = SimTA(:,t_i2);
spec3 = SimTA(:,t_i3);
spec4 = SimTA(:,t_i4);
spec5 = SimTA(:,t_i5);
spec6 = SimTA(:,t_i6);

close(figure(3));
fig1 = figure('Units', 'inches', 'Position', [5, 5, 6.3, 5]);
box on;
hold on;
p0 = plot(w,w*0,':','Color',MikeBlack,'LineWidth',LW);
p1 = plot(w,spec1,'Color',MikeBlack,'LineWidth',LW);
p2 = plot(w,spec2,'Color',MikeRed,'LineWidth',LW);
p3 = plot(w,spec3,'Color',MikeOrange,'LineWidth',LW);
p4 = plot(w,spec4,'Color',MikeGreen,'LineWidth',LW);
p5 = plot(w,spec5,'Color',MikeBlue,'LineWidth',LW);
p6 = plot(w,spec6,'Color',MikeViolet,'LineWidth',LW);
343

xlabel('Wavelength (nm)','FontSize',FT); ylabel('\DeltaAbs
(mOD)','FontSize',FT);
xlim([min(w) max(w)]);
xlim([min(w) max(w)]); ylim([-12 5.2]);
set(gca,'LineWidth',LW,'FontSize',FT);
L1 = legend([p1 p2 p3 p4 p5 p6],{[num2str(t_p1) ' ps'], [num2str(t_p2) '
ps'],...
[num2str(t_p3) ' ps'], [num2str(t_p4) ' ps'], ...
[num2str(t_p5) ' ps'], [num2str(t_p6) ' ps']},...
'Location','Southwest');
set(gca,'FontSize',FT,'LineWidth',LW);
title('Simulated Spectral Traces','FontWeight','bold');
hold off

%% Defining Colormap for Surface plot
% See Appendix B, section 1

clc;
x_colors = 1:128; % sets number of colors. Matlab colormaps are defined as
256 colors
size_x = length(x_colors); % Just calculates length of color vector
center_x = 88; % Use this to plot where zero/white is. Useful for TA data
with both positive and negative signals

colorsmap = hsv(256);
colors = rgb2hsv(colorsmap);
huenumber = colors(:,1);

% Hue
hue_y1 = 0.7; % Bluer if set higher
hue_y2 = 0.5; % Greener if set lower
hue_y3 = 0.0; % Red
hue_y4 = 0.25; % Orange
hue_x1 = x_colors(1);
hue_x2 = x_colors(center_x-1);
hue_x3 = x_colors(end);
hue_x4 = x_colors(center_x);
hue_m1 = (hue_y2 - hue_y1)/(hue_x2 - hue_x1);
hue_m2 = (hue_y4 - hue_y3)/(hue_x4 - hue_x3);
hue_fit1 = hue_m1*(x_colors(hue_x1:hue_x2) - hue_x1) + hue_y1; % low values,
red
hue_fit2 = hue_m2*(x_colors(hue_x4:hue_x3) - hue_x4) + hue_y4;

hue_fit = [hue_fit1 hue_fit2];

% Saturation
sat_a = 1;
sat_width = 0.18*size_x;
sat_offset = 1;
sat_center = center_x;
r = 0.62;
v = r^2*sat_width^2;

sat_fit1 = -sat_a*exp(-(x_colors(1:center_x-1) -
sat_center).^2/(2*r^2*sat_width^2)) + sat_offset;
344

sat_fit2 = -sat_a*exp(-(x_colors(center_x:end) -
sat_center).^2/(2*sat_width^2)) + sat_offset;
sat_fit = [sat_fit1 sat_fit2];

% Value
val_a = 1.0; % Defines how high value goes (close to white)
val_cent_skew = 1.5; % shift center max value by certain amount. If < 1 -->
lighter blues
val_center = val_cent_skew*center_x; % Where center is
val_width = size_x*0.8;
val_offset = 0; % Does value go to zero at ends
val_fit = val_a*exp(-(x_colors - val_center).^2/(2*val_width^2)) +
val_offset;

mycolormaphsv = [hue_fit; sat_fit; val_fit]';
mycolormaprgb = hsv2rgb(mycolormaphsv);

%% Simulated TA Datacarpet

close(figure(4));
fig4 = figure('Units', 'inches', 'Position', [5, 5, 6.3, 5]);
box on; hold on
surfTA = surf(w,t,SimTA','EdgeColor','none');
xlim([min(w) max(w)]); ylim([min(t) max(t)]);
view(2);
xlabel('Wavelength (nm)','FontSize',FT);
ylabel_pos = get(get(gca, 'YLabel'), 'Position');
new_ylabel_pos = ylabel_pos + [-08 -4980 0];
set(get(ax1, 'YLabel'), 'Position', new_ylabel_pos);
ylabel('Time (ps)', 'Position', new_ylabel_pos,'FontSize',FT);
set(gca,'YScale','log','FontSize',FT,'LineWidth',LW);
ax1pos = [0.14, 0.18, 0.825, 0.72];
ax2pos = [ax1pos(1)+ax1pos(3), ax1pos(2), ax1pos(3)*2.1, ax1pos(4)];
set(gca,'position',ax1pos);
yticks([0.1 1 10 100 1000 10000]);
yticklabels({'10^{-1}','10^{0}','10^{1}','10^{2}','10^{3}','10^{4}'});
colormap(mycolormaprgb);
cb = colorbar;
cb.LineWidth = LW/2;
colorTitleHandle = get(cb,'Title');
cbString = '\DeltaAbs (mOD)';
cbpos = [-20 265 0];
set(colorTitleHandle ,'String',cbString,'FontSize',FT+2,'Position',cbpos);
blackline = [0 0 0 0];
set(gca,'Layer','top')
set(gca,'LineWidth',LW/2);
hold off

%% Code End

345

This code simulated the excited state dynamics of an imaginary system, the results of which
for Case #1 are displayed in Figure B4.2; four plots detailing the dynamics were produced (Figures
1, 2, 3, and 4 in the code). The SADS detail the spectral features associated with each state (top
left). For this example, the SADS were designed to match the LE-SBCT spectral traces of zDIP2
or 2 (chapter 3) which possess distinct features associated with each state. The excited state
concentrations (top right) reflect our expectation that the equilibrium favors State 2 to State 1 in a
2:1 ratio after equilibrium is established (~30 ps). The simulated spectral traces (bottom left)
initially show features associated with State 1; but after ~10 ps, equal population of the states occur
with eventual and mutual decay by ~1 ns. The use of spectral traces is the most common way to
display TA data. The surface plot (bottom right) is another manner to display the TA data without
bias towards selected wavelengths as is done in the spectral trace plot. Here, we clearly see the
domains of time from which State 1 dominates (early time to 10 ps) and State 2 dominates (after
10 ps).

Table B4.1 – Summary of Rate Constants Used in the TA Simulation (all units ps
-1
)
Case # 𝑘𝑘 𝑟𝑟 / 𝑘𝑘 1 1 𝑎𝑎 𝑘𝑘 𝑛𝑛 𝑟𝑟 /𝑘𝑘 1 1 𝑃𝑃 𝑘𝑘 1 2
𝑘𝑘 2 1
𝑘𝑘 2 2
𝐾𝐾 𝑒𝑒 𝑞𝑞
1 2.5E-4 1.0 E-4 0.10 0.050 1.0 E-3 2
2 2.5E-4 1.0 E-4 0 N/A N/A 0
3 2.5E-4 1.0 E-4 0.10 0 1.0 E-3 ∞/Undef
4 9.5E-4 5.0E-5 5.0E-3 5.0E-5 0 100

346

While one case is useful to dissect, it is worth exploring several more examples with
emphasis on drawing general trends. We encourage the reader to follow along with the provided

Figure B4.2 – Simulated TA figures for Case #1 with two excited states and previously defined
parameters.
Top left) Simulated SADS created from the given spectral parameters. Dashed – GSB, dotted
– SE, dash-dot – ESA. Solid lines denote the SADS themselves.
Top right) Simulated excited state concentrations dictated by the kinetics previously described.
Temporal concentrations are displayed on a lin-log time axis with times before 1 ps being linear
and after 1 ps logarithmic.
Bottom left) Simulated spectral traces taken at the indicted times from the full simulated 2D
data set. The powers of 10 for the time slices were chosen simply personal preference. Note the
ESA feature at 330 nm decreasing with time as 400 nm grows in. The ESA feature associated
with the ESA of state 2 grow in at 600 nm in addition to the loss of SE in the same region.
Bottom right) surface plot of simulated TA spectra.
300 400 500 600 700
Wavelength (nm)
-10
-5
0
5
Abs (mOD)
SADS
State 1
State 2
-1 0 1
Time (ps)
0
0.2
0.4
0.6
0.8
1
Exc State Concentration
Excited State Concentrations
10
1
10
2
10
3
10
4
State 1
State 2
300 400 500 600 700
Wavelength (nm)
-10
-5
0
5
Abs (mOD)
Simulated Spectral Traces
1 ps
10 ps
100 ps
500 ps
1000 ps
3000 ps
347

code. See Table B4.1 for a complete summary of the rate constants used in these examples. Again,
there will be some minor changes, mostly in terms of formatting between the code here and the
modified code to produce the figures for cases after Case #1 but the vast majority of the code
remains untouched between cases.
In the simplest case, Case #2, for a TA experiment there is a single excited state system
which is populated initially and then decays via its fluorescence lifetime. State 2 is inactive
completely. We can tune this model to provide such a scenario by simply turning off forward
population transfer and setting 𝑘𝑘 1 2
= 0. Alternatively, 𝜏𝜏 1 2
→ ∞. We leave the previous
parameters unchanged as well as the previously created SADS. The results, all four plots, are
displayed in Figure B4.3. No population of state 2 occurs and the population of state 1 decays at a
rate determined by the sum of 𝑘𝑘 1 1
and 𝑘𝑘 2 2
(Figure B4.3, top right). This simulation case mirrors
the case of zDIP2 in cyclohexane where no SBCT state is populated (section 3.7.2.1).
348

Figure B4.3 – Simulated TA figures for Case #2 with a single excited state and previously
defined parameters ( 𝑘𝑘 1 2
= 0). Top left) Simulated SADS (unchanged). Top right) excited state
concentrations. Bottom left) Simulated spectral traces. Note no growth of features associated
with state 2. Bottom right) surface plot of simulated TA spectra.
300 400 500 600 700
Wavelength (nm)
-10
-5
0
5
Abs (mOD)
SADS
State 1
State 2
-1 0 1
Time (ps)
0
0.2
0.4
0.6
0.8
1
Excited State Concentrations
10
1
10
2
10
3
10
4
State 1
State 2
300 400 500 600 700
Wavelength (nm)
-10
-5
0
5
Abs (mOD)
Simulated Spectral Traces
1 ps
10 ps
500 ps
2000 ps
5000 ps
10000 ps
349

Alternatively, we can also turn off the reverse rate, 𝑘𝑘 2 1
= 0 which we do for Case #3
which may describe a photoreaction in which a bright species is turned over into a non-absorbing
photoproduct. The results of this parameter change are shown in Figure B4.4. Here, the reverse of
the previous scenario occurs wherein after a time of
1
𝑘𝑘 1 2
� or 𝜏𝜏 1 2
, conversion of State 1 to State 2
occurs completely and State 2 dominates entirely. By 100 ps, state 1 is eliminated entirely. Decay
from State 2 is now no longer dictated by the sum of 𝑘𝑘 1 1 𝑎𝑎 and 𝑘𝑘 1 1 𝑃𝑃 but rather, 𝑘𝑘 2 2
and because
𝑘𝑘 2 2
> 𝑘𝑘 1 1 𝑎𝑎 + 𝑘𝑘 1 1 𝑃𝑃 , State 2 decays faster than State 1 in the previous case. Note the apparent shift

Figure B4.4 – Simulated TA figures for Case #3 with two excited states and previously defined
parameters but 𝑘𝑘 2 1
= 0. Top left) Simulated SADS (unchanged). Top right) excited state
concentrations. Bottom left) Simulated spectral traces. Bottom right) simulated TA surface plot.
300 400 500 600 700
Wavelength (nm)
-10
-5
0
5
Abs (mOD)
SADS
State 1
State 2
-1 0 1
Time (ps)
0
0.2
0.4
0.6
0.8
1
Excited State Concentrations
10
1
10
2
10
3
10
4
State 1
State 2
300 400 500 600 700
Wavelength (nm)
-10
-5
0
5
Abs (mOD)
Simulated Spectral Traces
1 ps
10 ps
100 ps
500 ps
1000 ps
3000 ps
350

of the minimum of the GSB appears towards the blue edge of the spectrum. The SADS do not
have a change of wavelength of the minimum of the well (Figure B4.4, top left, dotted traces).
Here, the loss of the SE of state 1 and the growth of the ESA of state 2 work to pull up the red
portion of the bleach, leading to an apparent shift.

Finally, in Case #4, we wish to explore is similar to the cMa/TADF compounds explored
in chapter 5. The SADS were recreated from a sum of gaussians, like Cases 1-3 but now with
differing spectral inputs to achieve a qualitative match to the SADS of the cMa. Here, we set the

Figure B4.5 – Simulated TA figures for Case #4 with two excited states with kinetic parameters
for a generic cMa compound. Top left) Simulated SADS (unchanged). Top right) excited state
concentrations. Bottom left) Simulated spectral traces. Bottom right) simulated TA surface plot.
400 500 600 700 800 900
Wavelength (nm)
-5
0
5
Abs (mOD)
SADS
State 1
State 2
-1 0 1
Time (ps)
0
0.2
0.4
0.6
0.8
1
Excited State Concentrations
10
1
10
2
10
3
10
4
10
5
10
6
State 1
State 2
400 500 600 700 800 900
Wavelength (nm)
-5
0
5
Abs (mOD)
Simulated Spectral Traces
1 ps
10 ps
50 ps
100 ps
200 ps
10000 ps
351

kinetic parameters according to Table B4.1 for Case #4. These values are similar to a copper based
cMa compound. We take the values for a fast ISC from state 1 to state 2, singlet to triplet, 𝑘𝑘 0 1
=
0.1 𝑝𝑝 𝑠𝑠 − 1
and 𝑘𝑘 1 0
= 0.01 ∗ 𝑘𝑘 0 1
, i.e. 𝐾𝐾 𝑒𝑒 𝑞𝑞 = 100. The results of this simulation are shown in Figure
B4.5. Here, in the excited state concentration curve, we see that because 𝑘𝑘 0 1
is 100x greater than
𝑘𝑘 1 0
, the equilibrium for the system greatly favors state 2. Additionally, the system does not decay
by 1 ns as previously stated but by a few hundred nanoseconds.
We hope this foray into simple two state kinetics with probing via TA is useful. Readers
are now encouraged to explore the infinite parameter space afforded by these kinetic parameters
and substitute the system-specific spectra. We challenge the reader to arbitrarily extend the time
axis as a learning exercise to capture the full excited state evolution for a very long lived species.
Further work can be done to provide a more case-specific experiment such as the previously
mentioned time zero artifacts or something more intricate such as a time-dependent spectral
evolution such as the cMa compounds’ SE redshift as a function of time.

352

5. Appendix References
(1) Ware, W. R.; Watt, D.; Holmes, J. D. Exciple Photophysics. I. .Alpha.-Cyanonaphthalene-
Olefin System. J Am Chem Soc 1974, 96 (26), 7853–7860. https://doi.org/10.1021/ja00833a002.
(2) Veldman, D.; Chopin, S. M. A.; Meskers, S. C. J.; Janssen, R. A. J. Enhanced Intersystem
Crossing via a High Energy Charge Transfer State in a Perylenediimide−Perylenemonoimide
Dyad. J Phys Chem 2008, 112 (37), 8617–8632. https://doi.org/10.1021/jp805949r.
(3) Crosby, G. A.; Demas, J. N. Measurement of Photoluminescence Quantum Yields. Review. J
Phys Chem 1971, 75 (8), 991–1024. https://doi.org/10.1021/j100678a001.

Abstract (if available)

Abstract The photophysics, excited state kinetics, charge transfer dynamics, and reaction mechanisms of several photochemically active systems were studied with ultrafast spectroscopy techniques. The process of symmetry breaking charge transfer (SBCT) for a zinc dipyrrin was explored via a solvent dependent study which modulated the intramolecular charge transfer rates to establish an energy regime where the driving force was zero (DeltaG = 0). This was established via femto- to nanosecond transient absorption (psTA) and fluorescence quantum yield. Subsequently, intermolecular charge transfer conditions were mapped out for an entire class of carbene-metal-amide (cMa) compounds for use as solar fuel photosensitizers, assigned by spectral descriptions of the neutral and charged species via pulse radiolysis and bulk electrolysis. The intersystem crossing rates (3 − 20∙109 s-1) were determined by time-correlated single photon counting (TCPSC) and corroborated by psTA while the charge transfer kinetics to a diffusive electron donor or acceptor was mapped with a new nano- to microsecond transient absorption spectrometer, an instrument whose use in the UV is described herein. In a project aimed at strained ring formation via photochemical pathways, the photophysics and ultraviolet photochemistry of hexafluorobenzene (HFB) was explored using absorption, emission, psTA, multiconfigurational computations, and non-adiabatic dynamics simulations to develop a complete description of the dynamics of the 4pi-electrocyclic intramolecular ring-closing of HFB and the per-fluoro effect on the reaction. The intermolecular [2+2] cycloaddition of HFB with an olefin was then explored via fluorescence lifetime quenching and 19F NMR after UV irradiation which successfully produced solid photoproduct.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Singlet fission in covalently-linked tetracenes

Asset Metadata

Creator Kellogg, Michael Sean (author)

Core Title Capturing the sun: leveraging excited state dynamics

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Degree Conferral Date 2023-08

Publication Date 07/11/2023

Defense Date 06/12/2023

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

Tag absorption,charge transfer,cMa,emission,femtosecond transient absorption,fluorescence,hexafluorobenzene,magnitude Instruments,nanosecond transient absorption,OAI-PMH Harvest,organic photovoltaics,photochemistry,photophysics,photosensitizers,time-correlated single photon counting,transient absorption,ultrafast spectroscopy

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Bradforth, Stephen Edmund (committee chair), Kapadia, Rehan (committee member), Thompson, Mark Edward (committee member)

Creator Email michael.kellogg@gmail.com,mskellog@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC113263623

Unique identifier UC113263623

Identifier etd-KelloggMic-12061.pdf (filename)

Legacy Identifier etd-KelloggMic-12061

Document Type Thesis

Format theses (aat)

Rights Kellogg, Michael Sean

Internet Media Type application/pdf

Type texts

Source 20230711-usctheses-batch-1066 (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.

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