University of Southern California Dissertations and Theses

Excited state proton transfer in quinoline photobases

(USC Thesis Other)

Excited state proton transfer in quinoline photobases

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Excited State Proton Transfer in Quinoline Photobases

by

Jonathan Ryan Hunt

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

December 2021

Copyright 2021 Jonathan Ryan Hunt
ii
TABLE OF CONTENTS
List of Figures iv
List of Tables xii
Abstract xiii
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Alcohol Deprotonation with a Quinoline Photobase . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1 Thermodynamics from Steady State Experiments . . . . . . . . . . . . . . . . . . . . . 11
2.2 Kinetics from Transient Absorption Experiments . . . . . . . . . . . . . . . . . . . . . . 15
3 Excited State Proton Capture as a Function of Donor Concentration . . . . . . . . . . . 32
3.1 Absorption and Emission Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2 Transient Absorption and TCSPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4 Integration of a Photobase Molecule into an Iridium Catalyst . . . . . . . . . . . . . . . . 61
5 Excited State Electronic Structure Calculations of Photobase Molecules . . . . . . . . 67
6 Other Literature and Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .83
6.1 Works Inspired by My Own and An Exciting New Photobase . . . . . . . . . . . . 83
6.2 Future Research Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.2.1 Applications where Proton Donors are Dilute . . . . . . . . . . . . . . . . . . 89
6.2.2 Deprotonation of C-H Bonds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
6.2.3 Excited State Lewis Acid Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.2.4 Putting Quinoline’s Triplet States to Good Use . . . . . . . . . . . . . . . . . 100
6.2.5 Electrode-Tethered Photobases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
A Förster Cycle Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

iii
B Transient Absorption in the Dawlaty Lab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
B.1 What the Heck is Balanced Detection Pump Probe? . . . . . . . . . . . . . . . . . . . . . 117
B.2 Preparing to Run an Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
B.3 Preparing Your Pump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
B.4 How to Generate Nice White Light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
B.5 The Programs and How They Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
B.6 Alignment of White Light into Camera . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
B.7 The New Delay Stage and the Importance of Pump Alignment . . . . . . . . . . . . 148
B.8 The Phase of the Probe Beam with Respect to the Phase of the Pump Beam . 151
B.9 The Phase of the Probe Beams in the Transient Absorption Math . . . . . . . . . . 155
B.10 Arduino Box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
C Global Fitting of Transient Absorption Data In MATLAB . . . . . . . . . . . . . . . . . . . . . 162
C.1 General Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
C.2 Fitting 5-Methoxyquinoline in Methanol TA Data . . . . . . . . . . . . . . . . . . . . . . . . 164
D Electron Density Difference Plots and Löwdin Analysis in Q-Chem . . . . . . . . . . . . 166
E Marcus Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
References 189

v
List of Figures
1.1

Demonstrations of photoacidity in 5-R-naphthol and
photobasicity in 5-R-quinoline

2
1.2

Energy level diagram for quinoline 3
1.3

Absorption and emission spectra of the base and acid forms of
5-bromo-quinoline

4
1.4

Ground and excited state 𝑝𝐾
!
trends for 5-R-substituted
quinolines, calculated using the Förster cycle, as a function of
substituent Hammett parameter

5
2.1

Left: 5-methoxyquinoline’s structure. Right: emission from
5-methoxyquinoline’s unprotonated and protonated forms

12
2.2

Absorption and emission of 5-methoxyquinoline in various
solvents as a function of solvent pKa. The absorption indicates
that 5-methoxyquinoline is unprotonated in the ground state in
each solvent. Emission from the protonated form (around 500
nm) indicates excited state proton transfer occurs.

13
2.3

Transient absorption spectra of 5-methoxyquinoline in A) 2,2,2-
trifluoroethanol and B) isopropanol.

16
2.4

Transient absorption spectra of 5-methoxyquinoline in
methanol

18
2.5

a) Transient absorption basis spectra for 5-methoxyquinoline in
TFE. b) Transient absorption spectra in TFE. c) Global fit of
transient absorption data in TFE using basis spectra from (a) and
the kinetic model discussed in the text.

22
2.6

Demonstration of excited state absorption blue-shifts and
stimulated emission red-shifts following solvent reorganization
in the electronic excited state.

23
vi
2.7

a) Transient absorption basis spectra for 5-methoxyquinoline in
methanol. b) Transient absorption spectra in methanol. c)
Global fit of transient absorption data in TFE using basis spectra
from (a) and the kinetic model discussed in the text.

24
2.8

Free energy relationship between the rate constant for ESPT in
5-methoxyquinoline and the pKa of the proton donor.

26
2.9

The ESPT rate constant of 5-methoxyquinoline in water fits on a
photoacid-photobase ESPT free energy relationship from the
literature. Recreated from Cation-Enhanced Deprotonation of
Water by a Strong Photobase
1
and subsequently modified.

29
3.1

Can 5-methoxyquinoline ESPT occur in a dilute solution of the
proton donor?

34
3.2

a) Absorption and b) emission of 5-methoxyquinoline as a
function of TFE concentration in a background solvent of DCM

36
3.3

Fraction of 5-methoxyquinoline molecules that are hydrogen
bonded in the ground state (blue curve) and fraction that
undergo ESPT in the excited state (red curve) as a function of
TFE concentration in a background solvent of DCM.

38
3.4

Proposed mechanism for solvation of ESPT products of 5-
methoxyquinoline and TFE with a second TFE molecule.

41
3.5

Transient absorption spectra for 5-methoxyquinoline in a) pure
TFE, b) 5.2 M TFE, c) 0.52 M TFE, and d) 0.05 M TFE.

44
3.6

a) Transient absorption basis spectra for 5-methoxyquinoline in
8.5 M TFE. b) Transient absorption spectra in 8.5 M TFE. c)
Global fit of transient absorption data in 8.5 M TFE using basis
spectra from (a) and the kinetic model discussed in the text.

46
3.7

a) Transient absorption data of 5-methoxyquinoline in 1.04 M
TFE. b) Single wavelength kinetic analysis of 5-methoxyquinoline
in 1.04 M TFE.

47
3.8

a) TCSPC decay curve of 5-methoxyquinoline’s unprotonated
form in 0.52 M TFE with instrument response function and
biexponential fit. b) TCSPC decay curve of 5-methoxyquinoline’s
protonated form in 0.52 M TFE with instrument response
function and biexponential fit.
50
vii
3.9

Log-log plot comparing the proton transfer rate constant
between 5-methoxyquinoline and TFE as a function of TFE
concentration, with fits for the high-concentration non-diffusive
regime (blue) and low-concentration diffusive regime (orange).
The diffusive model discussed in the text is shown in purple.

54
4.1

The synthesized iridium complex with pendant quinoline
photobase. The photobasic action of the pendant quinoline is
shown.

62
4.2

A,B) Emission spectra of quinoline in various proton-donor
solvents. C,D) Emission spectra of iridium complex with pendant
quinoline in various proton-donor solvents.

64
4.3

Absorption and emission spectra of iridium complex with
pendant quinoline.

66
5.1

Ground and excited state pKa trends for 5-R-substituted
quinolines

68
5.2

Q-Chem outputs describing the singlet L A and L B states in terms
of molecular orbital transitions.

71
5.3

Leading configuration analysis and electron density difference
analysis for the L A state of quinoline.

72
5.4

Electron density difference maps for 5-R-quinolines as a
function of substituent directing effect.

74
5.5

Trends for experimentally determined versus theoretical excess
excited state charge density on the N heteroatom for TDDFT and
SOS-CIS(D) excited state electronic structure methods.

76
5.6

Calculated excess excited state charge density on N-heteroatom
versus computed excitation energy for a variety of substituted
aromatic N-heterocycles.

80
6.1

The structure of FR0-SB, a very strong photobase that uses a
Schiff’s base motif.

89
6.2

Presence of a second proton-donor stabilizes the products of
the 5-methoxyquinoline ESPT reaction and therefore reduces
the activation energy barrier.

92
viii
6.3

Can a high dielectric solvent stabilize the products of the 5-
methoxyquinoline ESPT reaction enough to significantly reduce
the activation energy barrier?

94
6.4

Some molecules with acidic C-H bonds. The acidic C-H is
denoted with a red shape

95
6.5

Some interesting Lewis acids and Lewis bases.

98
6.6

Can 5-methoxyquinoline act as a Lewis photobase?

100
6.7

Approximate singlet (blue) and triplet (red) state energies for
the base and acid forms of quinoline showing the possibility for
ESPT in the triplet manifold.

102
6.8

Relationship between applied electric field at an interface and
the Hammett parameter of the substituent that would have
similar effects

106
6.9

Concept for a surface-tethered quinoline experiment with
applied potentials.

108
A.1

Energy level diagram for the Förster cycle analysis of a
photobase molecule.

112
A.2

Potential energy surfaces and nuclear wavefunctions of ground
and excited states, demonstrating the overestimation and
underestimation of the 0-0 vibronic energy gap by absorption
and emission spectroscopy, respectively.

117
B.1

A pulse diagram for a “normal” transient absorption
experiment.

118
B.2

Two slightly different white light spectra.

120
B.3

False signal from poor white light subtraction and how it
compares to a reasonably intense transient signal.

121
B.4

A pulse diagram for balanced detection pump probe.

122
B.5

Coherent artefact caused by the overlap of the pump and probe
in pure isopropanol

128
ix
B.6

LEFT: TA spectra of 5-methoxyquinoline in 2,2-dichloroethanol 2
ps after excitation as a function of pulse energy. RIGHT: Signal
strength as a function of pulse energy with fit.

129
B.7

A cartoon diagram of the white light generation stage in the
Dawlaty lab transient absorption apparatus.

132
B.8

Stable white light immediately after the calcium fluoride crystal
in the Dawlaty lab white light generation stage.

135
B.9

A demonstration of robust white light subtraction.

136
B.10

The BalancedPPP Labview program used for running transient
absorption experiments with important inputs labeled.

139
B.11

Important icons and files in the Lightfield program.

141
B.12

The front panel of the ArduinoSendEdges Labview program.

142
B.13

The USBSpectrometerControl program used to control the
transient absorption spectrometer.

143
B.14

Cartoon diagram of the readout of the Pixis camera in a
balanced detection transient absorption experiment.

144
B.15

Readout of Lightfield in “1 kHz” mode, where individual white
light pulses are shown as a function of time.

146
B.16

Readout of Lightfield in “Image View” mode, where the entire
CCD array is displayed at once.

146
B.17

Mirror pair used for individual control of the sample and
reference probe alignment into the spectrometer.

147
B.18 The 4 ns delay state with a retroreflector.

149
B.19

A cartoon diagram showing the correct phase for the pump and
probe pulses in a balanced detection pump probe experiment
and the incorrect phase.

152
B.20

A cartoon diagram of an oscilloscope readout of the pump
(purple) and the probe (green) with (top) correct phase and
(bottom) incorrect phase. The vertical grey lines indicate 1 ms
spacing.
153
x

B.21

800 nm amplifier output immediately following the chopper
when (left) it is chopped correctly and when (right) it is chopped
incorrectly.

155
B.22

The correct pulse order for balanced detection transient
absorption in the Dawlaty lab setup and the subsequent math.

157
B.23

The pulse order for balanced detection transient absorption
when the phase of the pulses is off by 1 and the subsequent
math.

158
B.24

The pulse order for balanced detection transient absorption
when the phase of the pulses is off by 2 and the subsequent
math.

159
B.25

A picture of the Arduino box with labeled inputs and outputs.

160
B.26

Cartoon diagram of the operation of the Arduino box.

161
C.1

Unprotonated and protonated basis spectra for 5-
methoxyquinoline in TFE and their analytical gaussian fits used
in the kinetic model.

162
C.2

(Left) Transient absorption spectra for 5-methoxyquinoline in
TFE and (Right) the global fit of the data.

164
D.1

A Q-Chem input file for requesting excited state single point
energy calculations of quinoline and the excited state Löwdin
charges.

166
D.2

A Q-Chem input file for requesting electron density difference
maps of quinoline, output as cube files via the $plots section of
Q-Chem.

168
E.1

Solvation of (left) a neutral species and (right) a negatively
charged species in a dielectric medium.

171
E.2

Electronic motion and nuclear motion usually do not occur on
the same timescales.

173
E.3

Electron transfer at equilibrium solvent configurations works
from an electronic-nuclear timescale separation perspective,
but may violate the conservation of energy.
174
xi

E.4

Fluctuation of dipole arrangements in dielectric media can
create arrangements where the energies before and after
electron transfer are equal and the electron transfer happens
readily.

175
E.5

Marcus parabolas for two electronic states. State a is shown in
the blue box at its equilibrium solvent geometry; state b is
shown in the red box at its equilibrium solvent geometry; and
the transition state solvent geometry is shown in the black box.

176
E.6

Electronic distributions and nuclear polarizations of states a and
b.

180
E.7

Electronic distribution and nuclear polarization of state t.

181
E.8

Electronic distribution and nuclear polarization of state 𝜃.

182
E.9

The states of interest in Marcus Theory and the mathematical
relationship between them.

183

xii
List of Tables
1.1

Ultrafast dynamics of 5-R-quinolines in their base and acid forms as
a function of substituent Hammett parameter

6
3.1

Summary of TCSPC data for the unprotonated form of 5-
methoxyquinoline

52
5.1

Theoretical excitation energies and oscillator strengths for the L A
and L B states of 5-R-quinolines

70
5.2

Comparison of errors between experiment and theory for TDDFT
and SOS-CIS(D) excited state electronic structure theory methods
for 5-R-quinoline L A excitation energies

78
B.1

Correct order of probe pulses in our balanced detection transient
absorption experiment

145

xiii
Abstract
Photobases are molecules that convert light to proton transfer drive and therefore have
potential applications in many areas of chemistry. Quinoline and its substituted analogs are
examples of photobase molecules. I discuss the ability of 5-methoxyquinoline to deprotonate a
series of alcohols upon excitation by light. I report both the thermodynamic limits and the
relevant kinetics for this process using steady state and transient absorption spectroscopies,
respectively. A correlation is shown between the thermodynamic drive for the reaction and the
proton transfer timescale. Next, I investigate the effect of proton donor concentration on
ground-state hydrogen bonding and the excited-state proton transfer reaction of 5-
methoxyquinoline. It is shown that a large excess of proton donor is needed to see ground state
hydrogen bonding and an even larger excess is necessary to see excited state proton transfer.
The mechanism for this concentration-dependence is investigated with kinetic studies using
transient absorption spectroscopy and TCSPC. The results suggest that multiple proton donor
molecules are necessary to solvate the excited state proton transfer reaction. I then discuss an
iridium complex with a covalently attached quinoline that shows the ability to capture protons
independent of the electronics of the complex. Finally, I investigate the electronic origins of the
quinoline photobase phenomenon using electronic structure calculations. A convincing
correlation between changes in electron density on the nitrogen heteroatom and
experimentally determined changes in pKa is seen. I use this correlation to predict the
photobasicity of a variety of heteroatomic aromatic molecules. These results are necessary
fundamental steps towards applying photobases in potential applications, such as
xiv
deprotonation of alcohols for catalytic and synthetic purposes, optical regulation of pH, and
transfer of protons in redox reactions.
1
Chapter 1
Introduction
Proton transfer is widely important in chemistry. Redox reactions often require the
transfer of protons in addition to the transfer of electrons. For processes in many living
organisms, proton gradients are the driving force. The most notable example is the synthesis of
ATP. Most proposals involving the storage of solar energy in chemical bonds require driving
multi-electron and multi-proton reactions. If one could control proton transfer with some
external handle, it would potentially provide them with the ability to control these types of
reactions. But what are the chemical tools that we have to control proton transfer in this way?
There exist two main molecular tools that can be used to drive excited state proton
transfer (ESPT) reactions upon the absorption of a photon. One of these tools is called a
photoacid. Photoacids become more acidic upon photoexcitation and then donate a proton to
their surroundings. Some common photoacids include substituted naphthols
2–4
and pyrenols
5–7
.
A demonstration of photoacidity is shown in figure 1.1 for 5-R-naphthol. The thermodynamics
and kinetics of photoacid molecules have been studied somewhat extensively in the literature
8–
13
, with known applications ranging from controlling the conductivity of a polymer electrolyte
14
,
generating protonic potential differences
15
, triggering acid-initiated protein folding
16,17
, and
switching on acid-catalyzed synthetic
18–20
and enzymatic reactions
21,22
. The other type of ESPT
tool is called a photobase. Photobases become more basic upon photoexcitation and then
capture a proton from their surroundings. The most common structural motif for a photobase
molecule is an aromatic N-heterocycle, like pyridines
23
, quinolines
1,24–28
, and acridines
29–32
,
2
although other types of photobases are known
33–35
. A demonstration of photobasicity is shown
in figure 1.1 for 5-R-quinoline. Though photobases offer similarly rich prospects for applications
– one could imagine, for example, their use in speeding up catalytic reactions that are rate-
limited by the sluggish removal of protons from the reaction center – they have not found the
same widespread study and use as their photoacid counterparts.

Figure 1.1: Demonstrations of photoacidity in 5-R-naphthol and photobasicity in 5-R-quinoline.

This underutilization and promise for exciting applications are what encouraged the
Dawlaty lab to begin their investigations into photobase molecules. Eric Driscoll, my previous
research mentor, was inspired by an ESIPT (excited state intramolecular proton transfer)
project
36
. He noticed that one part of the ESIPT molecule became more acidic upon excitation,
while another became more basic. Intrigued, he investigated these concepts and was surprised
to learn how little work had been done on individual molecular units that become more basic
upon excitation. He set out to fill in some of the fundamental gaps in the understanding of
photobase molecules.
N
R
O
H
R
O
H
R
Photoacid Photobase
O
H
H
O
H
O
H H
H
O
H
H
!"
+
+
-
-
!"
N
H
N
R
5-R-Naphthol 5-R-Quinoline
3
First, he chose to study a series of 5-substituted quinoline molecules
27
. He chose this
system because it was the obvious photobase analog to 5-substituted naphthol, the
photoacidity of which had been studied systematically
2
. Eric measured the thermodynamic
drive for proton transfer in both the ground and excited states of the 5-substituted quinoline
molecules using steady state spectroscopy (absorption and emission) and Förster cycle analysis.
In short, one can calculate the excess drive for proton transfer in the excited state by knowing
the ground state drive for proton transfer and the S0 -> S1 energy gaps, obtained via absorption
and emission spectroscopy. Absorption and emission spectra for protonated and unprotonated
5-bromo-quinoline are shown in figure 1.3. Förster cycle analysis is discussed in-depth in
appendix A.

Figure 1.2: Energy level diagram for quinoline
!"
!
∗
= $.&'( ∆*
∗
Absorption
!"
!
= $.&'( ∆*
Emission
4

Figure 1.3: Absorption and emission spectra of the base and acid forms of 5-bromo-
quinoline

He observed – for the first time in the literature – that the thermodynamics of ESPT in
photobase molecules responds linearly to the polarization induced by the Hammett parameter.
That is, electron-donating groups make the molecule more basic in the excited state; electron-
withdrawing groups make the molecule less basic in the excited state; and the trend is well-
defined. Eric’s results are shown in figure 1.4. Depending on your perspective, this might seem
like a trivial result. Of course the excited state will be polarized in a predictable way by
substituents, you might say. Didn’t we learn this in organic chemistry? But one can never be too
confident when predicting the wild and wooly behavior of excited states – as Eric found out
when he attempted to study the kinetics of the excited state proton transfer of these same
molecules!
5

Figure 1.4: Ground and excited state 𝑝𝐾
!
trends for 5-R-substituted quinolines, calculated using
the Förster cycle, as a function of substituent Hammett parameter.

Eric studied the kinetics of excited state proton transfer in 5-substituted quinoline
molecules directly using UV-pump Vis-probe transient absorption spectroscopy
28
. Eric expected
each quinoline molecule to undergo ESPT and expected the kinetics to perhaps be dictated by
the strength of the drive for proton transfer in the excited state. What he found was far more
complicated. Many of the quinolines had a tendency for intersystem crossing (ISC). In some
cases, the ISC shut down the proton transfer altogether. In other cases, the proton transfer
seemed to happen in the triplet manifold, or ISC occurred after the proton transfer was
complete. Only one molecule in the 5-substituted quinoline series behaved as predicted, with a
clean proton transfer from unprotonated S1 to protonated S1: 5-methoxyquinoline (5-MeOQ).
6

Table 1.1: Ultrafast dynamics of 5-R-quinolines in their base and acid forms as a function of
substituent Hammett parameter

It was around this time that the quinoline project became mine upon Eric’s graduation. I
soon took the mantle of filling in the fundamental gaps in the photobase literature with the
help of my well-behaved friend 5-MeOQ. This project was the most exciting, fulfilling, and
fruitful research project of my PhD, and it comprises the entirety of the work presented here.
Because it was impossible to see a trend in the kinetics of the substituted quinolines due
to varying kinetic pathways, I was inspired to focus on one photobase molecule (5-MeOQ) and
instead change the thermodynamic drive by changing the pKa of the proton donor
37
. I could
then study the resulting kinetic trend. The results – that the kinetics followed a linear free
energy relationship with respect to proton donor pKa – were a testament to the well-behaved
nature of the photobase proton capture exhibited by 5-MeOQ. It also gave us insight into the
mechanism of the ESPT reaction. This was also the first time such a trend was shown in the
literature for photobase molecules. This work can be found in the chapter titled “Alcohol
Deprotonation with a Quinoline Photobase”.
7
In the experiment described above, the proton capture reactions were studied in pure
solvent where the solvent was the proton-donor. Because many practical applications of
photobases likely involve solutions where the proton donor is in low concentration, I sought to
study the impact of proton donor concentration on the excited state proton transfer
experiment
38,39
. I discovered that a large excess of proton donor was necessary to achieve
ground state preassociation with the photobase. Such preassociation is necessary due to the
short excited state lifetime (around 1 ns) and thus short-lived photobasicity of the molecules. I
also discovered that – at least for our current experimental conditions – two proton donor
molecules must reside in the solvation sphere to solvate the excited state proton transfer
reaction. This requires high concentrations of proton donor, which demonstrates one potential
hurdle in the chemical application of quinoline photobases as a useful chemical tool. This work
can be found in the chapter titled “Excited State Proton Capture as a Function of Donor
Concentration”.
I collaborated on a project with Ivan Demianets, a student of Travis Williams, where
quinoline was tethered to a well-known iridium catalyst with the intent to use the photobase to
control the kinetics of proton transfer steps in the catalytic cycle
40
. We demonstrated that the
quinoline molecule remains an active photobase, even while covalently linked to the ligand of a
heavy-metal catalyst, and that the excitation of the photobase and subsequent proton capture
are orthogonal to excitation of the catalyst itself. This was the first experiment of its kind. This
work can be found in the chapter “Integration of a Photobase Molecule into an Iridium
Catalyst”.
8
And because the molecular origins of the photobase process are not well understood, I
used DFT calculations to probe the changes in the electronic structure of quinoline photobases
upon excitation and found an interesting trend between the experimental excited state
thermodynamic drive and the electron density on the photobasic site
38
. I used similar
calculations to study other substituted N-heterocycles to better understand the tradeoffs
between excitation energy and thermodynamic drive in typical photobase molecules. This work
can be found in the chapter titled “Excited State Electronic Structure Calculations of Photobase
Molecules”.
Finally, I conclude by waxing philosophically about the past, present, and future of the
photobase field in the chapter “Other Literature and Future Directions”. I try to provide as
many potential experiments as possible for researchers who are interested in the field of
photobases. Check it out – some of the ideas are quite good!
I hope whoever reads this work feels inspired by the strange world of photobases and
sees their limitless potential for just about every area of chemistry. I know I do!

9
Chapter 2
Alcohol Deprotonation with a Quinoline Photobase
Most of the work described in this chapter originally appeared in the article
“Photodriven Deprotonation of Alcohols by a Quinoline Photobase” by Jonathan Ryan Hunt and
Jahan M. Dawlaty in the Journal of Physical Chermistry A
37
.
As mentioned in the introduction, I was interested in systematically exploring the
excited state kinetics of quinoline photobases as a function of proton transfer drive. Eric
Driscoll’s attempts to do so by changing the substituent (and thus the excited state pKa) were
somewhat unsuccessful due to the variety of excited state kinetic pathways: straightforward
singlet state proton transfer; proton transfer followed by intersystem crossing; rapid
intersystem crossing with subsequent deactivation of the unprotonated quinoline; and even
triplet state proton transfer
28
! While the experiments provided a lot of interesting information
(and led us to 5-MeOQ, the photobase studied throughout this dissertation), the results did not
tell a clean story of the relation between ESPT kinetics and thermodynamic drive.
I was inspired to study 5-MeOQ’s ability to deprotonate alcohols by an emission
experiment Eric completed before he graduated: 5-MeOQ in methanol showed emission from
both unprotonated and protonated forms, even though the absorption spectrum made it clear
that quinoline was not protonated in the ground state. An emission spectrum demonstrating
this behavior may be found later in the chapter. While we had seen ESPT in water, which was
10
readily deprotonated in the excited state by 5-MeOQ, the prospect of using a photobase to
deprotonate alcohols was enticing for a few reasons.
First, as mentioned above, it opened a new avenue for investigating the kinetics-
thermodynamics relationship for ESPT in quinoline photobases. If alcohols could be
deprotonated, it wouldn’t be necessary to change the excited state pKa of the quinoline itself,
as Eric had tried. Instead, you can change the pKa of the proton donor and create a
thermodynamic trend that way. If you chose a photobase molecule you knew to behave well,
such as 5-MeOQ, and used that molecule to deprotonate a series of alcohols, you had less risk
of fundamentally changing the excited state landscape and thus the photophysics – as had been
the case with the 5-substituted quinolines. It was a fun example of a small change in
perspective yielding a considerably easier and more well-designed experiment.
Second, controlling the deprotonation of alcohols with light using 5-MeOQ as a chemical
tool is a very appealing application of a photobase molecule. Transformation of alcohols is
important from a synthetic perspective, since alcohols serve as the feedstock for a wide range
of reactions. Furthermore, alcohols are important as a source of chemical energy, as in direct
fuel cells
41
. Deprotonation, often concurrent with oxidation, is required to activate an
alcohol
42,43
. This usually necessitates using strong bases or oxidative chemistry at electrodes.
The ability to use light and a photobase to activate an alkoxide instead is a powerful one – this
gives one greater temporal and spatial control over the activation process. The study of alcohol
deprotonation using photobase molecules, then, provides not only a wealth of fundamental
understanding but also a useful illustration of the potential utility of the class of molecules.
11
2.1 Thermodynamics from Steady State Experiments
We first studied the deprotonation of alcohols as a function of pKa with 5-MeOQ using
steady state absorption and emission spectroscopy. Spectra were taken using the instruments
in the USC Chemistry Instrument Room (Absorption: Cary 50 UV/Vis; Emission: Jobin-Yvon
Fluoromax 3). 5-MeOQ solutions of 2x10
-5
M in pure solvents of the proton donor were studied
in 1 cm fused quartz fluorescence cuvettes. A group of alcohols with a wide range of pKa values
were chosen. These included, in order of increasing pKa, 2,2,2-trifluoroethanol (TFE), 2,2-
dichloroethanol (DCE), water, methanol, ethanol, isopropanol, and tert-butanol. Their pKa
values, as defined in water, are listed figure 2.2.
5-MeOQ has distinct emission spectra for unprotonated and protonated forms. Emission
from the unprotonated has a 𝜆
"!#
of around 400 nm, while the protonated emission has a
𝜆
"!#
of around 500 nm. In figure 2.1 I show emission from 5-MeOQ in dichloromethane and in
acidified water. In dichloromethane, there are no labile protons, so we can be certain we are
seeing emission from the unprotonated form. Because the pH of the acidified water is well
below the pKa of 5-MeOQ, we should expect all MeOQ molecules to be protonated in the
ground state. Therefore, we can be confident we are seeing emission from the protonated
form.
12

Figure 2.1: Left: 5-methoxyquinoline’s structure. Right: emission from 5-methoxyquinoline’s
unprotonated and protonated forms

In all alcohols studied, 5-MeOQ was shown to be unprotonated in the ground state – as
expected, based on the pKa of the donor molecules – using absorption spectroscopy. In our
experiments, then, if emission from the protonated state of 5-MeOQ is observed, this is
evidence of successful excited state proton transfer. These experiments served a dual purpose:
to show proof that the deprotonation occurrs as expected while also giving a direct estimation
of the excited state pKa of MeOQ based on its deprotonating abilities.
These results are summarized in the figure below. In low pKa alcohols, we see emission
exclusively from the protonated form of the photobase, implying that most 5-MeOQ molecules
were able to capture a proton in the excited state before emitting. In high pKa alcohols, we see
emission exclusively from the unprotonated form, implying that 5-MeOQ was unable to
deprotonate the alcohol. Interestingly, in methanol, one observes emission from both
unprotonated and protonated forms, as mentioned earlier. These results demonstrate the well-
pKa = 4.9
pKa* = 15.1
(as predicted via
Forster cycle)
N
O
13
behaved nature of 5-MeOQ as a photobase: it can deprotonate more acidic alcohols, it can’t
deprotonate less acidic alcohols, and there is a clear trend demonstrating this tendency.
Methanol is the tipping point.

Figure 2.2: Absorption and emission of 5-methoxyquinoline in various solvents as a function of
solvent pKa. The absorption indicates that 5-methoxyquinoline is unprotonated in the ground
state in each solvent. Emission from the protonated form (around 500 nm) indicates excited
state proton transfer occurs.
300 400 500 600
0
1
300 400 500 600
0
1
300 400 500 600
0
1
300 400 500 600
0
1
300 400 500 600
0
1
300 400 500 600
0
1
300 400 500 600
0
1
Wavelength (nm)
Increasing pKa of proton donor
12.5
12.9
14
15.5
15.9
16.5
16.5
Absorption and Emission (normalized)
Absorption
Emission
(protonated)
Emission
(unprotonated)
14
Why do we see emission from both the protonated and unprotonated forms when 5-
MeOQ is excited in methanol? There are two possible reasons. First, perhaps the timescale of
proton transfer in methanol is very similar to the timescale for emission of unprotonated 5-
MeOQ. Such a competition would result in approximately half of the 5-MeOQ molecules
emitting from the unprotonated form and half emitting from the protonated form. We can call
this the “kinetic hypothesis”. Second, maybe there is an excited state equilibrium somewhat
rapidly established in the excited state, such that some 5-MeOQ molecules are protonated and
some are unprotonated in a fixed ratio based on the thermodynamic drive. In this scenario, the
rate of the forward and backward proton transfer is equal after a certain amount of ESPT. Such
an equilibrium would need to be formed well before the timescale for emission from the
unprotonated photobase. We can call this the “thermodynamic hypothesis”. The truth could
also lie somewhere in-between these two hypotheses if the time required to establish dynamic
equilibrium is long enough compared to the excited state lifetime of the photobase.
Based on the kinetic data reported later in this chapter, the “thermodynamic
hypothesis” is the more likely. We will discuss the evidence for this later, but for now we will
assume that the thermodynamic hypothesis is true. Since 5-MeOQ establishes an equilibrium
between protonated and unprotonated forms when MeOH is used as a proton donor, the pKa*
of 5-MeOQ should be close to the pKa of MeOH, which is 15.5. Note that the pKa of MeOH is
15.5 in water, but that this experiment was carried out in pure MeOH so solvation conditions
will be different. Therefore, one must be aware that 15.5 cannot be used as the exact pKa of
the proton donor for this experiment, but only as an estimate. As will be discussed in greater
detail below, using aqueous pKa values in non-aqueous environments seems less egregious
15
when you consider that the solvation of the photobase is similarly perturbed in MeOH with
respect to water. In Eric’s first paper on 5-susbtituted quinoline molecules, he used Förster
Cycle Analysis to estimate that 5-MeOQ’s pKa* was approximately 15.1
27
. The agreement
between these two experiments, which estimate pKa in distinctly different ways, is a
reassurance of the robust nature of our experimental and theoretical methods.
2.2 Kinetics from Transient Absorption Experiments
As mentioned in the intro to this chapter, one of the main appeals of studying alcohol
deprotonation was the ability to change the pKa of the proton donor and study the resulting
change in the kinetics in a systematic way. I accomplished this by studying the proton capture
reaction of 5-MeOQ in the low pKa solvents – where proton transfer is thermodynamically
possible - using transient absorption spectroscopy.
I used the Dawlaty Lab balanced detection transient absorption apparatus for the
following experiments. 3mM solutions of 5-MeOQ in pure solvents of the proton donors were
flowed through a fused quartz cell with 0.5 mm path length. The samples were pumped at 315
nm (near the absorption maximum) and studied with a broadband visible probe (400 – 650
nm). Pump pulse energies of 50 – 100 nJ were used. The pump and probe were focused to 140
um and 180 um, respectively, determined using the edge method. The time resolution of the
experiment was approximately 300 fs. Additional details about the balanced detection transient
absorption apparatus can be found in appendix B.
Transient absorption experiments show that the excited state proton transfer kinetics of
5-MeOQ are well-behaved. In low pKa proton donors, where proton transfer occurs, a decay in
16
the unprotonated 5-MeOQ TA spectrum occurs at the same rate the protonated 5-MeOQ TA
spectrum grows in. This is shown in figure 2.3A. The most obvious indicator of this
transformation is the growth of the stimulated emission of protonated 5-MeOQ at around 550
nm. Seemingly little else – aside from population loss (which occurs on longer timescales) and
solvent relaxation effects (to be described later) – happens. In high pKa proton donors, where
proton transfer does not occur, we see only population loss and solvent relaxation effects. This
is shown in figure 2.3B.

Figure 2.3: Transient absorption spectra of 5-methoxyquinoline in A) 2,2,2-trifluoroethanol and
B) isopropanol
400 450 500 550 600 650
2
3
4
5
6
7
8
9
Wavelength / nm
A / mOD

0.3ps
4ps
12ps
35ps
105ps
203ps
Wavelength / nm
Time / ps

400 450 500 550 600 650
2
4
6
8
-0.5
0
0.5
1
1.5
2
2.5
3
Emergence of
protonated form
Wavelength / nm
Time / ps

400 450 500 550 600 650
50
100
150
200
2
3
4
5
6
7
8
400 450 500 550 600 650
0
1
2
3
Wavelength / nm
A / mOD

0.3ps
1ps
3ps
5ps
7ps
9ps
pKa
12.5
pKa
16.5
A
B
17
Inspection of the transient absorption data and comparison with known spectra of the
acid form of the molecule suggests that, in all these solvents studied here except methanol, all
or most of the excited MeOQ molecules undergo proton transfer. That is, every single (or
almost every single) excited molecule becomes protonated after a certain amount of time. The
kinetics are qualitatively different in methanol, however. The transient absorption data for 5-
MeOQ in methanol are shown in figure 2.4 below. The kinetics at early times appear like those
in the other, lower pKa solvents. But at longer times, the kinetics stop. What remains is a
spectrum that looks like a linear combination of the base spectrum and the acid spectrum that
undergoes no further change. This can be interpreted as excited state proton transfer by a
portion of the excited state MeOQ molecules, followed by the attainment of excited state
quasi-equilibrium between protonated and unprotonated MeOQ molecules. This result is not
surprising given the competitive pKa values of MeOQH+* and methanol, and the comparative
lifetimes of proton transfer (around 50 ps) and the excited state of the molecule (1 ns). There is
plenty of time for the equilibrium to be established before most of the molecules return to the
ground state.
This mechanism – the establishment of an excited state equilibrium – is likely pervasive
in all the proton transfer reactions studied here. The existence of a common mechanism is
supported by the trend in proton transfer rates shown below. In the case of the other solvents,
however, due to the larger thermodynamic drives for protonation, the equilibrium is skewed
heavily towards the protonated form of the molecule and we can view those reactions as going
“to completion”.

18

Figure 2.4: Transient absorption spectra of 5-methoxyquinoline in methanol.
Proton transfer timescales were extracted for the four proton donors which showed
evidence of ESPT in the previous section. These donors are methanol, water, 2,2-
dichloroethanol (DCE), and 2,2,2-trifluoroethanol (TFE). To fit the kinetics, we first needed a
model. Based on the seemingly straightforward kinetics of the proton transfer reaction in all
solvents except methanol (the kinetic modeling of which will be discussed later), a
correspondingly straightforward kinetic model was developed:
𝑑
𝑑𝑡
[𝐵
∗
] = −𝑘
%
[𝐵
∗
]
𝑑
𝑑𝑡
[𝐴
∗
] = 𝑘
%
[𝐵
∗
]
where [B*] is the proportion of the excited base form of 5-MeOQ, [A*] is the proportion of the
excited acid form of 5-MeOQ, and k p is the proton transfer rate constant. In layman’s terms, the
rate at which the base form converts into the acid form is dictated by the concentration of the
base form. The magnitude of the rate is given by the rate constant, 𝑘
%
. We assume that the
Wavelength / nm
A / mOD

400 450 500 550 600 650
20
40
60
80
2
3
4
5
6
7
8
9
400 450 500 550 600 650
1
2
3
4
5
6
7
8
Wavelength / nm
A / mOD

5ps
10ps
25ps
50ps
75ps
100ps
pKa
15.5
19
encounter rate of the photobase and proton donor is much faster than the proton transfer rate
since these experiments are carried out in pure solvents. We therefore ignore it in the model.
TCSPC measurements (discussed in chapter 3) show that the excited state lifetime of MeOQ in
hydrogen-bonding solvents is longer than 1 ns. Our longest proton-transfer timescale is around
50 ps. We therefore ignore population decay in our model, since it happens on much longer
timescales than the delays studied in our experiments and can thus be shown to affect these
dynamics only very slightly.
Upon excitation, all 5-MeOQ molecules should be in the [B*] form. If we normalize with
respect to the initial concentration of the base form, the initial conditions of this experiment
are [B*](t=0) = 1 and [A*](t=0) = 0. In our model, the loss of one B* molecule results in the gain
of one A* molecule, so any decrease in the concentration of [B*] should result in an equal
magnitude increase in [A*]. After proton transfer is complete (assuming the ESPT reaction goes
to completion), all 5-MeOQ molecules should be in the [A*] form and the final conditions of this
experiment are [B*] = 0 and [A*] = 1. Based on this behavior, we can alternatively represent
[B*] as (1 – [A*]). This allows us to solve the differential equations in a straightforward way.
Inserting [B*] = 1 – [A*] into the second equation gives us
𝑑
𝑑𝑡
[𝐵
∗
] = −𝑘
%
[𝐵
∗
]
𝑑
𝑑𝑡
[𝐴
∗
] = 𝑘
%
(1−[𝐴
∗
])
The solutions to these differential equations are well-known. Solving gives us exponential
equations describing the behavior of [A*] and [B*]:
20
[𝐵
∗
] = 𝑒
'(
!
)
+𝐶
[𝐴
∗
] = 𝐷− 𝑒
'(
!
)

Given our initial conditions above ([B*](t=0) = 1 and [A*](t=0) = 0), we know that C = 0
and D = 1. Our final equations are therefore
[𝐵
∗
] = 𝑒
'(
!
)

[𝐴
∗
] = 1− 𝑒
'(
!
)

These are the equations that we use to fit the data in all solvents except methanol.
Because ESPT in methanol does not go to completion, we must modify the above equations to
include static amounts of 𝐵
∗
and 𝐴
∗
at long times. The easiest way to achieve this is by
introducing a constant, n, describing the proportion of B* molecules that participate in the ESPT
reaction before equilibrium is achieved:
[𝐵
∗
] = 1−𝑛(1− 𝑒
'(
!
)
)
[𝐴
∗
] = 𝑛∗(1−𝑒
'(
!
)
)
Note that [B*] decreases exponentially from 1 at t = 0 to (1 – n) at long times; and that [𝐴
∗
]
increases exponentially from 0 at t = 0 to n at long times. This is the expected behavior.
To perform a global fit of the transient data using the above kinetic model, we must
choose suitable basis functions for [B*] and [A*]. The basis function for [B*] is obtained from
early time measurements, where proton transfer has not occurred in any notable quantity. The
transient spectrum for [A*] can be obtained either from long time measurements (sufficiently
long such that all proton transfer has completed) or by protonation of 5-MeOQ using a strong
21
acid. I prefer the first method, since ground state protonation can lead to additional
complications, like the addition of protonated spontaneous emission and ground state bleach
features. However, the protonation method was necessary for kinetic modeling of 5-MeOQ in
methanol since the proton transfer reaction did not go to completion. Using these basis spectra
along with the kinetic model discussed above, we were able to fit the proton transfer in
trifluoroethanol, dichloroethanol, and water quite well. A representative fit – for MeOQ in TFE
– is shown below in figure 2.5. The specifics of how the kinetic model and basis spectra were
applied in MATLAB to fit the transient absorption data and extract time constants are discussed
in appendix C.
There is one additional complication associated with fitting transient absorption spectra:
the reorganization of solvent leads to time-dependent changes in the basis spectra. This was
observed above for 5-MeOQ in isopropanol. Although there is no proton transfer in this solvent
due to its high pKa, there are changes in the transient absorption spectra – namely, the blue
shifting of the excited state absorption features and changes in their intensity. Let’s briefly
discuss the origins of this phenomenon. Immediately after excitation of the photobase
molecule, surrounding solvent molecules are out of equilibrium with the new electronic
structure. It takes the solvent molecules a certain amount of time to reach a new equilibrium.
The amount of time it takes the solvent molecules to find the new equilibrium solvation shell
will depend largely on the viscosity of the solvent, since much of this new equilibrium will be
established via translation and rotation of the solvent.

22

Figure 2.5: a) Transient absorption basis spectra for 5-methoxyquinoline in TFE. b) Transient
absorption spectra in TFE. c) Global fit of transient absorption data in TFE using basis spectra
from (a) and the kinetic model discussed in the text.

The blue shifting of excited state absorption in the isopropanol data is caused by the
molecular system moving from a non-equilibrium configuration to an equilibrium configuration.
As the system approaches equilibrium in the excited state, its energy will decrease. This will
cause the excitation energy necessary for excited state absorption to increase, thus resulting in
a blue shift, as shown in figure 2.6 below. If the feature of interest was instead stimulated
emission, we should expect such a change to result in a red shift, since stabilization of the
excited state decreases the energy gap between excited and ground states. The changes in the
intensity of the excited state absorption are the result of changes in Franck-Condon overlap
400 450 500 550 600 650
0
1
2
3
Wavelength / nm
A / mOD

B* spectrum (0.3ps)
A* spectrum (300ps)
400 450 500 550 600 650
0
1
2
3
Wavelength / nm
A / mOD

0.3ps
1ps
3ps
5ps
7ps
9ps
400 450 500 550 600 650
0
1
2
3
Wavelength / nm
A / mOD

0.3ps
1ps
3ps
5ps
7ps
9ps
b. TA Data c. Fit
τ = 2.3ps
a. Basis spectra
23
between the S1 and higher excited states caused by the changes in the nuclear geometry of the
photobase molecule. Whether the intensities increase or decrease will be determined by the
geometry changes and Franck-Condon overlaps of the particular molecule of interest.

Figure 2.6: Demonstration of excited state absorption blue-shifts and stimulated emission red-
shifts following solvent reorganization in the electronic excited state.

We can largely ignore these internal and solvent nuclear reorganization changes in our
kinetic modeling. In water, such changes are typically very fast (sub ps)
44
, so they will not affect
the dynamics in our study, where the earliest time studied is 1 ps. The timescales for
reorganization should be longer in dichloroethanol and trifluoroethanol considering typical
reorganization timescales for other alcohols
45
, but since the proton transfer reactions occur
quite rapidly in these solvents, we assume that the effect on the base spectrum and thus on the
overall kinetic fit is minimal. Note that, because specific solvation requirements must be
S0
S1
S2
Solvent
Reorganization
Excited
State
Absorption
Stimulated
Emission
Reorganization
Energy
24
satisfied for the proton transfer reaction to occur, we don’t expect a large degree of nuclear
reorganization following the ESPT reaction. We therefore approximate that the acid spectrum
does not show reorganizational changes.

Figure 2.7: a) Transient absorption basis spectra for 5-methoxyquinoline in methanol. b)
Transient absorption spectra in methanol. c) Global fit of transient absorption data in TFE using
basis spectra from (a) and the kinetic model discussed in the text.

We cannot ignore the effects of solvation on the MeOQ proton transfer reaction in
methanol. Because a steady state equilibrium is reached, the base spectrum never really goes
away – so any changes that occur to the base spectrum must be included in the kinetic model
to achieve a reasonable fit. We were therefore required to modify our kinetic model to include
time-dependent changes in the basis spectrum of the unprotonated form. The resulting fit is
shown in figure 2.7. The changes we made to the kinetic model are quite complicated and are
therefore beyond the scope of this section. We include the specifics in appendix C. However,
400 450 500 550 600 650
1
2
3
4
5
6
7
8
Wavelength / nm
A / mOD

5ps
10ps
25ps
50ps
75ps
100ps
400 450 500 550 600 650
1
2
3
4
5
6
7
8
Wavelength / nm
A / mOD

5ps
10ps
25ps
50ps
75ps
100ps
400 450 500 550 600 650
0
2
4
6
8
Wavelength / nm
A / mOD

B* spectrum (1ps)
A* spectrum (MeOH + HCl)
b. TA Data
c. Fit
τ 54ps
a. Basis spectra
25
we can comment on the reasonableness of our fit of the reorganization dynamics. Our global
fitting procedure retrieved a value of about 9 ps for these dynamics, reasonably consistent with
literature values that range between 5 ps
45
and 9 ps
6,46
.
The thermodynamic drive for a proton transfer reaction scales with the difference in
pKa between the two protonated species – in this case, the difference in pKa (ΔpKa) between
the excited conjugate acid of the photobase and the solvent proton donor. If we assume that
the pKa* of the conjugate photobase is constant in each solvent, then the pKa of the proton
donor molecule is directly related to the thermodynamic drive of the excited state proton
transfer reaction. If the pKa is lower, the donor is more acidic and the ΔpKa is higher; there
should therefore be a greater thermodynamic drive for proton transfer. Using either Arrhenius-
or Marcus-type theory, it is predicted that – within the same family of reactions, assuming the
same mechanism prevails – the speed of a reaction should increase as the thermodynamic drive
increases
47
. This relationship is referred to as a Free Energy Relation (FER).
In figure 2.8, I plot the natural log of the rate constant for each proton transfer reaction
as a function of the pKa of the solvent. A clear trend presents itself: more acidic proton donors
tend to have faster proton transfer rates, as expected from a reaction where there is an FER.
The trend should not be expected to be quantitative, since each solvent has a different
dielectric constant and a different viscosity and thus will have different rates of solvation. We
anticipate only a qualitative trend, and that is what is observed. It is also important to note that
water can form long-range hydrogen bonds, unlike the alcohols, which may contribute to its
faster proton transfer rate.
26

Figure 2.8: Free energy relationship between the rate constant for ESPT in 5-methoxyquinoline
and the pKa of the proton donor

A brief aside: why is it, exactly, that we (and many others) plot the log of the rate as a
function of thermodynamic drive? Where did that log come from? For the simplest answer we
can return to our kinetics lessons from introductory chemistry. Specifically, we turn to the
Arrhenius equation:
𝑘 = 𝐴 𝑒
'*
"
+,

12 13 14 15 16
-4
-3
-2
-1
pKa
ln(k [1/ps])
Increasing thermodynamic drive
Increasing Rate
24 ps
54 ps
28 ps
2.3 ps
27
where k is the rate of the reaction, A is an exponential pre-factor that determines the
magnitude of the reaction rate, and E a is the activation energy of the reaction. This equation
explains how, typically, the rate of the reaction changes as a function of the activation energy. If
we take the natural log of both sides of this equation, we get the following:
ln𝑘 =
−𝐸
!
𝑅𝑇
+ln𝐴
This equation implies that, within Arrhenius behavior, the natural log of the rate of a reaction is
proportional to the activation energy barrier for that reaction.
The ΔpKa is proportional to the ΔG of the proton transfer reaction. If we plot the natural
log of k as a function of ΔG, as we have done above, we are effectively probing the following
question: how does the activation energy of the reaction scale with the thermodynamic drive of
the reaction? This is the fundamental question of a free energy relationship, and it explains why
we take the log of the rate in our plot.
Note that there is an important caveat with regards to the interpretation of this kinetic
trend. These reactions are not carried out in water. The literature values of pKa used here,
which are determined in water, are therefore not strictly correct to describe the
thermodynamic properties of the proton donors in our experiments. Furthermore, the pKa* of
5-MeOQ is likely to change slightly from solvent to solvent due to changes in solvation. Thus,
our assumption that the thermodynamic drive of the reaction is entirely decided by the pKa of
the solvent molecule is flawed. For these reasons and others mentioned above, we are unable
to explain the observed trend in kinetics quantitatively.
28
However, the trend is still qualitatively useful and thermodynamically reasonable. pKa
values change from one solvent to another due to changes in solvation, which can come from
both bulk dielectric effects and specific interactions. The use of aqueous pKa values in non-
aqueous environments seems less egregious when one considers that the photobase is
experiencing the same solvation conditions as the proton donor. Any changes in solvation that
would result in an increased pKa for the proton donor should probably also result in a similarly
increased pKa for the photobase. The ΔpKa between the proton donor and the excited
photobase will therefore be perturbed minimally. Since the ΔpKa is the relevant quantity to
describe the thermodynamics of the ESPT reaction, the use of aqueous pKa values is more
reasonable than it might appear at first glance.
While we were unable to numerically fit the trend to a single model due to uncertainty
about thermodynamic drives and differences in solvent viscosity and dielectric, we show below
that our time constant for proton transfer in water fits comfortably on a Marcus Theory curve
for excited state proton transfer in photoacids and photobases that was found in the
literature
1
. This confirms that the pKa* value for MeOQ determined from Förster Cycle
calculations is robust, as it correctly predicted the thermodynamics and thus the kinetics of the
proton transfer process in water, and that our transient absorption experiments and data fitting
are reliable.
We also studied proton transfer from deuterated forms of water and methanol. Because
a deuteron is heavier than a proton and thus has a more localized wavefunction, there tends to
be less through-barrier overlap of deuteron wavefunctions than in proton wavefunctions. This
means that, if there is any tunneling involved in the proton transfer process, the deuteron
29
transfer should be impeded relative to the proton transfer and should therefore be slower. The
kinetic isotope effect (KIE), reported as a ratio of deuteron capture time to proton capture time,
quantifies this effect. A number larger than 1, then, typically signifies the importance of
tunneling in the studied reaction.

Figure 2.9. The ESPT rate constant of 5-methoxyquinoline in water fits on a photoacid-
photobase ESPT free energy relationship from the literature. Recreated from Cation-Enhanced
Deprotonation of Water by a Strong Photobase
1
and subsequently modified.

In heavy water, a deuteron capture time of 22 ± 5 ps was extracted, yielding a KIE of
0.9 ± 0.3. Since a KIE below 1 is non-physical, we can assume that the KIE is 1 within error. This
implies that, in water, there is no tunneling effect on the proton transfer reaction. In heavy
methanol, a deuteron capture time of 133 ± 30 ps was extracted, indicating a KIE of 2.5 ± 0.7.
This implies that there is a tunneling effect on the proton transfer reaction in methanol. The
30
existence of two seemingly different mechanisms for the proton capture by 5-MeOQ in water
and methanol is an open question. One theory is that, because of the larger barrier for proton
transfer and subsequently slow dynamics in methanol, tunneling becomes a viable pathway;
whereas in water, the proton transfer happens with sufficient ease and speed due to the
smaller activation barrier that tunneling need not contribute.
AN IMPORTANT NOTE: In my original publication, I claimed that the observed trend in
kinetics suggested that solvation was not the rate-limiting step. My argument was that, if
solvation were truly rate-limiting, then water should have the fastest proton transfer reaction
since solvation in water is by far the fastest. Similarly, I argued that TFE (which has the fastest
proton transfer rate by far) would have the slowest proton transfer reaction since it is the
bulkiest solvent. This was wrong, written due to a fundamental misunderstanding of solvation
kinetics and what role they play in condensed phase reactions.
Now that I understand concepts such as Marcus Theory more fully, I can revise my
stance on solvation dynamics. The rate at which a solvent solvates a species will no doubt play a
role in the kinetics associated with that species, as will other considerations like the solvent
dielectric. This is part of the difficulty of coming up with a single model that numerically
describes the trend in in proton transfer amongst different solvents! However, it is not
solvation of the species that’s relevant here – it is the fluctuation of the solvent molecules
around the equilibrium solvation configuration that ultimately give rise to conditions where
proton transfer is possible. This will be related to solvent viscosity, solvent dielectric, specific
interactions, etc. but ultimately the most relevant quantity is the size of the activation energy
barrier, which is correlated with the thermodynamic drive, as discussed above. It is the size of
31
this activation barrier that dictates how large solvent fluctuations around the equilibrium
configuration must be to allow the reaction, and thus how likely the reaction is to occur in a
given timeframe.
This work was, to the best of my knowledge, the first study where the thermodynamics
of the proton donor was systematically altered to study the kinetics of proton transfer in a
photobase molecule. It demonstrated the well-behaved nature of 5-MeOQ as a photobase and
suggested that the scope of photobase activity might be larger than previously thought. It is
work I am proud of, especially since this work resulted in my first first-author paper (and a work
produced solely by me and my PI, Jahan Dawlaty).

32
Chapter 3
Excited State Proton Capture as a Function of Donor Concentration
Most of the work described in this chapter originally appeared in the articles “Donor-
Acceptor Preassociation, Excited State Solvation Threshold, and Optical Energy Cost as
Challenges in Chemical Applications of Photobases” by Jonathan Ryan Hunt, Cindy Tseng, and
Jahan M. Dawlaty in Faraday Discussions
38
; and “Kinetic Evidence for the Necessity of Two
Proton Donor Molecules for Successful Excited State Proton Transfer by a Photobase” by
Jonathan Ryan Hunt and Jahan M. Dawlaty in The Journal of Physical Chemistry A
39
.
Up until this point, I had only considered pure solvents for hydrogen bond-donating
systems. These solvent systems are relatively simple: the photobase molecule is always
surrounded by an excess of donor molecules, meaning that diffusional motion and long-range
solvation changes are unnecessary for the proton transfer reaction to proceed. But in areas of
chemistry where quinoline photobases might be helpful tools – for fine-control over PT in
catalytic reactions, for example – it is unlikely that the solvent would be the molecule one
hopes to deprotonate. This made me wonder: what would happen to the behavior of the
photobase if the proton donor were no longer the solvent, but was instead in a dilute solution?
Would ESPT occur as straightforwardly? Would it be easy for the photobase to hydrogen bond
in the ground state? And would hydrogen bonding be enough for the proton transfer reaction
to proceed following excitation of the photobase? These are questions that had not been
investigated for photobase reactions.
33
Even though the excited states of quinoline photobases likely have larger drives for
complexation than their ground states, it is preferential for hydrogen bond association to occur
in the ground state. This is because the excited state lifetime of quinoline photobases is short –
around 1 ns. If the photobase and proton donor are not pre-associated in the ground state, and
the concentration of the proton donor is low, it may take longer than this for the proton donor
and photobase to find each other via diffusional motion. The ESPT reaction will therefore not
occur during the excited state lifetime of the molecule.
Although there is an enthalpic drive for hydrogen bond association, there is also an
entropic drive for the photobase and the proton donor to be solvated separately. In other
words, even if it is energetically favorable for the photobase to hydrogen bond with a proton
donor in solution, it may not do so at concentrations where the Gibbs free energy associated
with the entropy of mixing between the proton donor and the solvent is larger in magnitude. To
enforce a significant population of hydrogen bonded complexes, one may need to increase the
concentration of the proton donor well beyond a 1:1 ratio of photobase to proton donor. One
big question we wanted to answer was this: exactly how much excess proton donor is
necessary to force a significant population of the hydrogen-bonded species?
One other important aspect of the investigation into the proton donor-dependence of
the photobase ESPT reaction that may not be immediately obvious is this: will hydrogen
bonding with a single proton donor molecule result in proton transfer following excitation of
the photobase? Based on previous works, there was no reason to suspect this was not the case.
However, as we will see later, there may be additional, specific solvation requirements that are
not met by the presence of a single proton-donor molecule.
34
I decided to study mixed solvents – that is, solvents where the proton donor and
photobase were mixed with some other, non-proton-donating solvent in various ratios. I
continued to use 5-methoxyquinoline (5-MeOQ) as my photobase molecule. In my work on the
deprotonation of alcohols with 5-MeOQ (pKa* = 15.5), I found that 2,2,2-trifluoroethanol (TFE,
pKa = 12.4) was an easy solvent to work with and that TFE was readily deprotonated by 5-
MeOQ with fast kinetics (𝜏
-,
= 2 ps for the pure solvent)
37
. It is well-known from the literature
that TFE is a strong hydrogen-bond donor
48
, which makes it uniquely well-suited for this
investigation. As such, I chose TFE as the proton-donor in my mixed solvent studies. I chose
DCM as my non-proton-donating solvent since both TFE and 5-MeOQ readily dissolve in DCM,
and DCM had no possible specific-interactions that would complicate the analysis. All chemicals
were purchased from Sigma-Aldrich and used without further purification. The concentration of
MeOQ was kept at 5 x 10^-5 M for the following steady state measurements, while the
concentration of TFE was varied from 0 M to 13.9 M, the concentration of the pure solvent. The
specific concentrations of TFE used may be observed in figures 3.2 and 3.3.

Figure 3.1: Can 5-methoxyquinoline ESPT occur in a dilute solution of the proton donor?
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
N
O
O
F
F
F
H
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
N
O
O
F
F
F
H
+
-
ℎ"
35
3.1 Absorption and Emission Spectroscopy
Absorption and emission spectra (figure 3.2) for each solution were obtained to
investigate the ground state hydrogen-bond complexation and excited state proton transfer,
respectively. Absorption spectra were obtained using a Cary 50 UV/Vis spectrophotometer.
Measurements were made in a 1 cm fused quartz cuvette. Absorption spectra were background
subtracted and normalized for analysis. These corrected absorption spectra are shown in figure
3.2a.
Ground-state hydrogen-bonding results in changes to the electronic structure of the
photobase that are indicated by a red shift of the absorption spectrum. Note that most of the
shift occurs at low (non-bulk) TFE concentrations. This indicates that the shift arises due to
specific interactions – hydrogen bonding – rather than through modification of the solvent
dielectric or other bulk characteristics. We quantified the degree of hydrogen bonding in each
solution by considering the red shift in each solvent and comparing it to the boundary
conditions: low TFE concentration, before the red shift begins; and high TFE concentration,
once the red shift has completed. The percent red shift from the low concentration TFE solution
to the high concentration TFE solution corresponds to the percent of the MeOQ species that
are hydrogen bonded. The percent red shift was quantified by fitting the endpoints to a sum of
Gaussians and then fitting the intermediate spectra as a linear combination of those Gaussian
fits. The corresponding curve, showing percent of hydrogen-bonded MeOQ molecules as a
function of TFE concentration, can be found in blue in figure 3.3.

36

Figure 3.2: a) Absorption and b) emission of 5-methoxyquinoline as a function of TFE
concentration in a background solvent of DCM.

Remembering that the concentration of MeOQ in these samples is 5𝑥10
'.
M, I’ll briefly
summarize the important findings from this curve. Appreciable hydrogen bonding does not
begin until there are about 200 TFE molecules for each MeOQ molecule; a ratio of about 2000:1
TFE to MeOQ molecules is necessary for approximately half of the photobase molecules to
(a)
More
hydrogen
bonding
Absorption
TFE
Concentration
(b)
Emission
37
hydrogen bond; and it takes a ratio of around 50000:1 for virtually all photobase molecules to
be hydrogen bonded. Even though there is an enthalpic drive for hydrogen bonding in the
ground state, dissociation is favored at low concentrations of the proton donor. Since pre-
association is preferable for photobase reactions due to the limited excited state lifetime, these
findings may have an impact on the application of photobases in systems where the donor
molecule is dilute. It’s important to note that we have intentionally chosen a strong hydrogen-
bond donor in our work; this problem may be even more pronounced in other systems.
Next, we turned our attention to the excited state proton transfer reaction. We were
particularly interested in whether the ESPT reaction would follow the same concentration-
dependence as the ground-state hydrogen bonding. That is, we were curious if hydrogen
bonding in the ground state was enough to guarantee a successful ESPT reaction.
To analyze the ESPT reaction, we studied the same solution mixtures using emission
spectroscopy, as seen in figure 3.2b. Emission spectra were collected on a Jobin-Yvon
Fluoromax 3 fluorometer. Measurements were made in a 1 cm fused quartz cuvette. All
samples were excited at 310 nm. As mentioned in the previous chapters, the protonated and
unprotonated forms of the photobase have dramatically different emission characteristics.
There should be no protonation in the ground state in these solutions, so the emission can be
used to study whether the excited state proton transfer reaction is successful. That is, if
emission from the protonated form is observed, it must come from the excited state proton
transfer reaction. As the concentration of TFE increases, the emission of the unprotonated form
(around 400 nm) starts to decrease and the emission of the protonated form (around 500 nm)
38
starts to increase, indicating that the ESPT reaction becomes more favorable at higher TFE
concentrations.

Figure 3.3: Fraction of 5-methoxyquinoline molecules that are hydrogen bonded in the ground
state (blue curve) and fraction that undergo ESPT in the excited state (red curve) as a function
of TFE concentration in a background solvent of DCM.

We can monitor the rise of the protonated spectrum to determine the concentration-
dependence of the ESPT reaction. The percentage of protonated emission vs the protonated
emission in pure TFE (where protonated MeOQ should be generated with unity efficiency) can
be used to determine the percentage of successful ESPT reactions. For example, if half of the
No Significant
Encounters
Onset of
Hydrogen
Bonding
Onset of Excited State
Proton transfer
Fraction of Population
Hydrogen bonded
in the ground state
Protonated in the
Excited State
TFE Concentration
(M)
5 x 10
-4
5 x 10
-3
5 x 10
-2
5 x 10
-1
5.2
39
maximum emission intensity is present, then half of the MeOQ molecules have successfully
participated in the ESPT reaction. We plotted the emission intensities at 575 nm as a function of
TFE concentration, normalized with respect to minimum and maximum intensities, to produce
the percentage of protonated MeOQ molecules. The resulting curve from this analysis can be
seen in red in figure 3.3.
It’s important to note that there are changes in the emission spectrum of the
unprotonated form caused by the addition of the TFE that occur before emission from the
protonated form is observed. Namely, there is some decrease in the emission intensity and
some red shifting. We attribute this to changes in the electronic structure of MeOQ caused by
hydrogen bonding, since the onset of these changes occurs at the same TFE concentrations
where hydrogen bonding occurs and because there is no rise in the protonated emission. By
using the rise of the protonated form to study the efficiency of the ESPT reaction, we are
effectively avoiding these complications, since the emission from the protonated form should
not be affected in a similar way.
After the onset of emission from the protonated form of MeOQ, a clear trend with
respect to growth of emission from the protonated form was observed and most of the
emission spectra shared an isosbestic point at approximately 450 nm. Two of the collected
spectra were offset in the overall emission intensity, but not in the relative peak ratios, likely
due to scattering at the sample or small differences in MeOQ concentration. To correct for this
deviation, we multiplied each spectrum by a constant to enforce the isosbestic point at 450 nm.
This choice was imperfect, since continued hydrogen bonding after the onset of ESPT will lead
to small changes in the profile of the unprotonated spectrum which will also lead to small
40
deviations from a clean isosbestic point. However, these deviations are likely to be small since
most MeOQ molecules (around 80%) are already hydrogen bonded at the onset of the ESPT
reaction. We therefore make this imperfect approximation, assuming that its positives
outweigh the negatives.
It is clearly observed that a higher TFE concentration is necessary for the ESPT reaction
than for ground state hydrogen bonding. While the onset of hydrogen bonding was observed
via absorption spectroscopy at a ratio of 200:1 TFE to MeOQ, the onset of protonated emission
was not observed until a ratio of approximately 5000:1, where around 75% of the MeOQ
molecules were already hydrogen bonded in the ground state. There is a greater than order of
magnitude lag in the concentration between the onset of hydrogen bonding and the onset of
ESPT! What could give rise to this phenomenon?
These results show that hydrogen-bonding to the proton donor is NOT enough to ensure
successful ESPT – and point towards the importance of proper solvation of the photogenerated
products following ESPT. The results suggest that multiple proton donors are necessary to
solvate the ESPT process. This makes sense because, if a small cluster of alcohol molecules were
present around the hydrogen bonded complex, there could be solvation of the conjugate base
product via an additional hydrogen bonding molecule – as shown in figure 3.4 below. This type
of behavior has previously been shown in the photoacid literature
49
.
The steady state data provides some clues about the specific nature of these alcohol
clusters. Emission from the protonated form of MeOQ begins while there is still an appreciable
population of non-hydrogen-bonded MeOQ in the ground state – a quick look at the curves in
41
figure 3 indicates that ESPT is first observed when about 75% of the molecules are hydrogen
bonded in the ground state. If multiple proton donors are necessary for ESPT, it makes little
sense that there would be a large alcohol cluster around an MeOQ molecule while other MeOQ
molecules aren’t even hydrogen bonded – especially since, while there is an enthalpic drive for
hydrogen bonding, there is likely no enthalpic drive for cluster formation in the ground state.
Therefore, it is most likely that the onset of ESPT occurs when there is one additional TFE
molecule in addition to the hydrogen bonded one – when there is a total of two TFE molecules
present. The previous photoacid literature which showed similar behavior used the kinetics of
photoacid dissociation as a function of alcohol concentration to determine that two alcohol
molecules were capable of solvating the photoacid ESPT products
49
. Therefore, a two-molecule
alcohol cluster for solvating ESPT products is not unprecedented.

Figure 3.4: Proposed mechanism for solvation of ESPT products of 5-methoxyquinoline and TFE
with a second TFE molecule.
+
-
42
We can explore the necessary number of proton donors more quantitatively by
assuming that the probability of n number of TFE molecules residing in the solvation sphere
follows a Poisson distribution, the form of which is shown below:
𝑃
/
=
〈𝑛〉
/
𝑛!
𝑒
'〈/〉

where n is the number of TFE molecules and 〈𝑛〉 is the average number of TFE molecules in the
solvation sphere. 〈𝑛〉 can be calculated in the following way:
〈𝑛〉 = 𝑉
2345
𝑁
6
[𝑇𝐹𝐸]
where 𝑉
2345
is the volume of the solvation sphere in L, 𝑁
6
is Avogadro’s number, and [TFE] is
the concentration of TFE in M. We can estimate the 𝑉
2345
by assuming a reasonable value for
the radius of the solvation sphere. If we take the radius to be 1 nm, the calculated 𝑉
2345
is about
4 𝑥 10
'78
L.
Let’s calculate the Poisson probabilities for a few representative TFE concentrations. At
5 𝑥 10
'7
M, we see no emission from the protonated form. 〈𝑛〉 for this concentration is 0.12.
The probability of 2 or more TFE molecules in the solvation sphere can be calculated by
subtracting the probabilities of zero or one TFE molecules from unity:
𝑃
97
= 1−𝑃
:
−𝑃
;

𝑃
97
= 1−
0.12
:
0!
𝑒
':.;7
−
0.12
;
1!
𝑒
':.;7

𝑃
97
= 7 𝑥 10
'=

43
According to the Poisson distribution, the chance of having two or more TFE molecules
in the solvation sphere at this concentration is less than 1%. This is consistent with the fact that
we see no ESPT at this concentration. Now let’s consider a concentration of 0.5 M, where about
20% of the MeOQ molecules undergo ESPT. At this concentration, 〈𝑛〉 = 1.2. We once again
calculate the probability of 2 or more TFE molecules in the solvation sphere:
𝑃
97
= 1−
1.2
:
0!
𝑒
';.7
−
1.2
;
1!
𝑒
';.7

𝑃
97
= 0.34
The chance of having two or more molecules in the solvation sphere at this
concentration is about 34% - which correlates with our experimental results quite well! Finally,
let’s consider a concentration of 5 M, where all of the molecules participate in the ESPT
reaction. At this concentration, 〈𝑛〉 = 12:
𝑃
97
= 1−
12
:
0!
𝑒
';7
−
12
;
1!
𝑒
';7

𝑃
97
≈ 1
Every photobase molecule should have at least two TFE molecules in its solvation sphere
at this concentration, which correlates with the unity ESPT we see at this concentration! The
neat correspondence between the predictions of the Poisson distribution and the experimental
data lends additional credence to the idea that two TFE molecules are needed in the solvation
sphere for the ESPT process to proceed. Note that we have ignored the enthalpic drive for
hydrogen bonding in these calculations. Since the TFE hydrogen-bonded to MeOQ has a
greater-than-random chance of being found in the solvation sphere, its probability of being
44
found there should not be well-described by the Poisson distribution. We have also ignored the
possibility of diffusional motion into the solvation sphere. This exercise still provides additional
weight to our analysis of the steady state data.
3.2 Transient Absorption and TCSPC

Figure 3.5: Transient absorption spectra for 5-methoxyquinoline in a) pure TFE, b) 5.2 M TFE, c)
0.52 M TFE, and d) 0.05 M TFE.

Inspired by literature on a photoacid molecule where the authors used kinetic data to
pinpoint the exact number of alcohol molecules necessary for ESPT
49
, I began studying the
kinetics of the MeOQ photobase reaction as a function of proton donor concentration. I used
a.
13.9 M (pure TFE)
𝝉 𝑷𝑻
𝟐 . 𝟎 . 𝒑𝒔
b.
5.2 M TFE
𝝉 𝑷𝑻
. 𝟎 𝟐 . 𝟎 𝒑𝒔
c.
0.52 M TFE
𝝉 𝑷𝑻
𝟎 𝟎 𝟎 𝒑𝒔
d.
0.05 M TFE
Negligible PT
45
transient absorption spectroscopy and time-correlated single-photon counting (TCSPC). The
details of the transient absorption apparatus are discussed in chapter 2 and in appendix B. The
TCSPC apparatus is discussed in more detail below.
Transient absorption data for several TFE concentrations can be seen in figure 3.5. The
behavior and analysis of the transient absorption data is nearly identical to that described in
chapter 2: at early times, we see a spectrum that we assign to the unprotonated form; as the
unprotonated form decays, we see a corresponding rise in the spectrum of the protonated
form. The most obvious indicator of this process is an increase in the negative stimulated
emission from the protonated form around 535 nm. We associate these kinetics with the ESPT
reaction, so we fit them to obtain the proton transfer timescale, 𝜏
-,
.
The major difference between the kinetics we study here and those in chapter 2 is that,
as the TFE concentration drops, it eventually becomes limited by long-range translational
motion. Therefore, as the TFE concentration decreases, we see longer and longer ESPT
timescales. In the previous chapter, we did not explicitly consider the excited state lifetime of
the photobase in our kinetic model because the ESPT timescales were fast compared to
population decay from S1 to S0. This is still the case for most of the solutions studied here.
However, at low TFE concentrations, we must include the excited state lifetime in our model:
𝑑[𝐵
∗
]
𝑑𝑡
= −
[𝐵
∗
]
𝜏
-,
−
[𝐵
∗
]
𝜏
4>?@)>"@,B
∗

𝑑[𝐴
∗
]
𝑑𝑡
=
[𝐵
∗
]
𝜏
-,
−
[𝐴
∗
]
𝜏
4>?@)>"@,6
∗

46
We obtain these excited state lifetimes from TCSPC, discussed below. The method used for
global fits of the transient absorption data is discussed at length in chapter 2 and in appendix C.
An example is shown below for an 8.5 M solution of TFE.

Figure 3.6: a) Transient absorption basis spectra for 5-methoxyquinoline in 8.5 M TFE. b)
Transient absorption spectra in 8.5 M TFE. c) Global fit of transient absorption data in 8.5 M TFE
using basis spectra from (a) and the kinetic model discussed in the text. The background solvent
is DCM.

Because we were limited to delays of 600 ps in our TA setup, good basis spectra for the
acid form were unobtainable in solutions where the proton transfer timescale was longer than
100 ps. We were therefore not confident in our ability to perform a robust global fit for the two
lowest concentrations of TFE (0.52 M and 1.04 M) in which we saw ESPT in our transient
absorption studies. We were forced to use a single wavelength kinetic analysis to retrieve
a. Basis Spectra
b. TA Data c. Fit to Model
𝝉 𝑷𝑻
𝟑 . 𝟎 . 𝒑𝒔
47
proton transfer timescales. In short, we averaged several kinetic traces in the spectral range
where the stimulated emission feature grows (approximately 535 nm) and fit the average time
traces with our kinetic model. An example of this type of fitting is shown in figure 3.7. Although
single wavelength analysis is less robust than a global fit, we were confident in our obtained
time constants because they were in reasonable agreement with the proton transfer timescales
observed using TCSPC, as discussed below.

Figure 3.7: a) Transient absorption data of 5-methoxyquinoline in 1.04 M TFE. b) Single
wavelength kinetic analysis of 5-methoxyquinoline in 1.04 M TFE.
Now that we have discussed how to extract the proton transfer timescales, we can
discuss the behavior that we see as a function of TFE concentration. Figure 3.5a above shows
the TA of MeOQ in pure TFE. Although this data was collected at a different time, the extracted
timescale is consistent with that reported in the previous chapter for the same experiment.
Figure 3.5b shows the TA of MeOQ in a solution that is approximately half TFE in terms of
solvent mole fraction. It is observed that the proton transfer timescale has already more than
doubled, even when over half of the solvent molecules are TFE! Figure 3.5c shows this effect
much more dramatically. At a TFE concentration an order of magnitude lower than in 3.5b, the
a b
𝝉 𝑷𝑻
𝟐𝟎𝟎 𝟐𝟓𝒑 𝒔
48
proton transfer timescale is drastically longer (400 ps). As shown in 3.5d, at concentrations of
TFE lower than 0.5 M, the proton transfer timescale becomes too long to be studied via our
transient absorption apparatus, which only had delays up to 600 ps at the time. However, as
discussed below, proton transfer was seen at these TFE concentrations using TCSPC, which can
detect longer timescales. The trend seen here – longer timescales at lower TFE concentrations –
will be explored quantitatively below.
We used TCSPC to find the excited state lifetime of MeOQ in various solutions. At
concentrations around and below 1 M TFE, we could also use TCSPC to find ESPT timescales.
TCSPC is a time-resolved fluorescence technique. The sample is excited on resonance by a
pulsed, low power source, so that few molecules are excited and at most one emitted photon
reaches the detector per pulse. The time that occurs between the emission of the photon from
the source and the photon reaching the detector is monitored using a charging capacitor. The
charging begins when the excitation source emits and stops when the photon reaches the
detector. The time required for the photon to reach the detector is then calculated using the
well-calibrated charge-to-time ratio of the capacitor. This experiment is repeated, typically
thousands of times, until enough photons have been collected to achieve a statistically
representative decay curve for the fluorescence. The times are corrected – for proton travel
from the excitation source to the sample, for proton travel from the sample to the detector,
and for any other delays introduced by the instrumentation – using an instrument response
function. The corrected time reveals how long it took the sample to emit the photon following
excitation. This decay curve can then be studied to extract timescales for excited state lifetimes
and other excited state dynamics.
49
The ESPT reaction diminishes the concentration of the unprotonated form while
concurrently increasing the concentration of the protonated form. In principle, we can study
both occurrences using TCSPC. The loss of unprotonated form via ESPT will be competitive with
fluorescence, and thus we will observe a decay in fluorescence associated with the ESPT
reaction. As discussed above, all MeOQ molecules in these solutions are unprotonated in the
ground state. Therefore, if emission from the protonated form occurs, we can be certain this is
a result of ESPT. Any rise-time associated with the fluorescence post-excitation will therefore
also provide the ESPT timescale. All of this makes the MeOQ ESPT reaction well-suited to be
studied using a time-resolved fluorescence technique like TCPSC. We are, however, limited by
the time resolution of the apparatus, which is around 200 ps (or 1/10 of the FWHM of the
instrument response function) if careful deconvolution with the instrument response function is
achieved
50
.
A Horiba Fluorohub apparatus, equipped with a TBX picosecond photon detection
module, was used to collect the TCSPC data in this manuscript. The 𝜆
"!#
for absorption of
MeOQ is around 310 nm in these solvents, but a 280 nm pulsed LED (Horiba N-280, FWHM = 10
nm, 1 – 2 pJ/pulse) was used as the excitation source because it was the closest available.
Although 280 nm is energetically off from the absorption maximum, it is still resonant with the
broad S1 <- S0 transition. Fluorescence was collected at 390 nm and 550 nm to independently
monitor the excited state emission from unprotonated MeOQ and protonated MeOQ,
respectively. The collection monochromator was operated at various slit widths to achieve an
appropriate collection efficiency. Instrument response functions were collected using the same
50
excitation and collection conditions but by replacing the sample with an empty, frosted 1 cm
quartz cuvette.

Figure 3.8: a) TCSPC decay curve of 5-methoxyquinoline’s unprotonated form in 0.52 M TFE
with instrument response function and biexponential fit. b) TCSPC decay curve of 5-
methoxyquinoline’s protonated form in 0.52 M TFE with instrument response function and
biexponential fit.

The TCSPC data was deconvoluted with the instrument response function (IRF) and fit
using the DAS6 fitting software. Representative TCSPC spectra – at a TFE concentration of 0.52
M so that the results can be compared to those shown for transient absorption above – are
shown in figure 3.8. Figure 3.8a shows the fluorescence decay at 390 nm, where the
unprotonated form emits; figure 3.8b shows the fluorescence decay at 550 nm, where the
protonated form emits. The decay of the unprotonated form is visibly biexponential. The
resulting fit gives one long decay timescale of around 2.8 ns, which is assigned to population
a
𝝀 𝒆𝒎
𝒏𝒎 (unprotonated form)
𝝉 𝒍𝒊 𝒇𝒆𝒕 𝒊𝒎 𝒆 . 𝒏𝒔 , 𝝉 𝑷𝑻
𝒑𝒔 b
𝝀 𝒆𝒎
𝒏𝒎 (protonated form)
𝝉 𝒍𝒊 𝒇𝒆𝒕 𝒊𝒎 𝒆 . 𝒏𝒔 , 𝝉 𝑷𝑻
𝒑𝒔 𝒓 𝒊 𝒔𝒆 TCSPC of MeOQ in 0.52M TFE
51
decay from the excited state to the ground state, and one shorter decay of around 480 ps,
which is assigned to the excited state proton transfer reaction. We can confidently assign the
shorter decay to ESPT because it is consistent with the 400 ps ESPT timescale extracted from
our transient absorption data. Furthermore, as shown in the table below, 92% of the overall
decay of the unprotonated excited state of MeOQ comes from this 480ps decay channel. This is
consistent with the near-complete proton transfer observed in the transient absorption data.
We gain further confidence when we observe the fluorescence decay of the protonated form.
There is an exponential rise with a timescale of approximately 530 ps! This shows rather
unambiguously that the short decay timescale seen in the unprotonated form causes a rise in
the protonated form – that we are observing the timescale of the ESPT reaction. Note that,
since these ESPT timescales are shorter than the FWHM of the IRF, they will be very sensitive to
the quality of the IRF and of the statistical sampling
50
. The variability in the proton transfer
timescales (transient absorption vs. 390 nm TCSPC vs 550 nm TCSPC) is likely due to these
sensitivities.
The ESPT timescales for the other solutions studied using this method can be seen in
table 3.1, where fits of the decays of the unprotonated form at 390 nm are reported. At
concentrations of 5.2 x 10^-3 M TEA and above, we see fluorescence decay due to both ESPT
and population decay. Concentrations above 1.04 M are not shown because the ESPT
timescales are too short to be resolved using TCSPC. The ESPT reaction is so near complete and
fast that virtually no fluorescence from the unprotonated form can be seen within the time
resolution of the apparatus, and the rise of the protonated form is too quick to resolve.
52

Table 3.1: Summary of TCSPC Data for the unprotonated form of 5-methoxyquinoline.
For both kinetic techniques, I was required to increase the concentration of MeOQ in
my solutions – from 5 𝑥 10
'.
M in steady state experiments to 3 𝑥 10
'=
M – to see a
reasonable amount of signal. I studied the same concentrations of TFE in these higher [MeOQ]
solutions as I did in my steady state experiments. Although the mole ratios of the solutions
used for kinetic analysis are no longer identical to those used for steady state analysis, we
approximate that the local solvation in the vicinity of a given MeOQ molecule will be the same
in both cases. That is, the local environment of a MeOQ molecule, whether in a 5 𝑥 10
'.
M or
3 𝑥 10
'=
M solution, should be largely determined by the concentration of TFE. Exceptions
arise when the concentration of MeOQ becomes sufficiently large that the molecules begin to
interact with each other. This is likely not the case at either 5 𝑥 10
'.
M or 3 𝑥 10
'=
M
concentrations. Therefore, we feel comfortable comparing solutions with the same TFE
concentrations even when the concentration of MeOQ has been altered.
The goal of these kinetics experiments was to determine the number of TFE molecules
necessary to solvate the products of the ESPT reaction, since our steady state analyses
suggested it was more than one TFE molecule. With ESPT timescales for a variety of TFE
concentrations at hand, we can now investigate the mechanism. In a classic rate equation, the
53
rate of a single step reaction depends on the concentration of the reactants raised to the power
of the molecularity of the reactant (the number of reactant molecules that participate in that
reaction), which can, in principle, be estimated from measuring reaction rates as a function of
concentration. The rate constant of the ESPT reaction as a function of concentration can be
written as:
𝑘
-,
= 𝑘
3
[𝑇𝐹𝐸]
/

where 𝑘
-,
is the proton transfer rate constant, ko is a proportionality constant, [TFE] is the
molar concentration of TFE, and n is the molecularity of TFE. Taking the log of both sides yields
log
;:
𝑘
-,
= 𝑛 log
;:
[𝑇𝐹𝐸]+ log
;:
𝑘
3

Therefore, the molecularity of the proton donor is the slope of log
;:
𝑘
-,
plotted as a
function of log
;:
[𝑇𝐹𝐸]. A log-log plot of our data can be seen in figure 3.9. Two regimes are
obvious – one at high concentration and one at low concentration. First, we will focus on the
high concentration regime, fit to a blue line.
We know from the steady state analysis presented above that virtually all MeOQ
molecules are hydrogen bonded with TFE in the ground state at these concentrations.
Furthermore, because of the high concentrations of TFE used in this regime, we assume that, on
average, at least one TFE molecule exists in the solvation shell in addition to the TFE molecule
that is hydrogen bonded to MeOQ. In the 3.7 M TFE solution, the lowest concentration of this
regime, approximately 1/4 of all solvent molecules are TFE, so it is reasonable to think that at
least one TFE molecule will be in the vicinity of the hydrogen-bonded photobase reaction
center. Therefore, we refer to this as the "non-diffusive regime", implying that only short-range
54
reorientation of nearby TFE molecules is important for any necessary solvation of the ESPT
products.
Figure 3.9: Log-log plot comparing the proton transfer rate constant between 5-
methoxyquinoline and TFE as a function of TFE concentration, with fits for the high
concentration non-diffusive regime (blue) and low-concentration diffusive regime (orange). The
diffusive model discussed in the text is shown in purple.

The proton transfer timescales of these concentrated TFE solutions form a linear trend
with a slope of approximately 1.2. Based on the simple argument made above, where I
discussed the reactant molecularity, it is tempting to conclude that this implies the involvement
of only one TFE molecule. However, the concentration of the hydrogen-bonded MeOQ species
changes negligibly throughout this region, since virtually all of the MeOQ molecules are
hydrogen bonded to a donor molecule prior to excitation, as mentioned above. Therefore, the
loe 2/3
loe 1
55
proton transfer timescale in this region is only sensitive to changes in the composition of the
solvent near the hydrogen bonded photobase. For this reason, a slope of 1 in this region
strongly suggests the necessity for two TFE molecules - one pre-hydrogen bonded to MeOQ to
donate the proton to the photobase, and one in the surrounding solvent to stabilize the
resulting TFE anion. Put another way: if all MeOQ molecules are already hydrogen bonded at
these concentrations and only one TFE molecule is neccessary for ESPT, there should be no
kinetic dependence on [TFE] in this region. The fact that there is a kinetic dependence at all,
while all MeOQ are pre-hydrogen bonded in the ground state, implies the involvement of more
than one TFE molecule. The fact that the slope is approximately 1 implies that it is likely one TFE
molecule required in addition to the hydrogen-bonded one.
I have one important correction from my initial publication of this kinetic analysis. In my
original paper I explain that, because DCM and TFE have very similar dielectric constants, the
changes seen at high concentrations of TFE must be due to “explicit proton transfer, hydrogen
bonding, and local solvation” and that “bulk dielectric effects contribute minimally” as a
result
39
. There is one problem: DCM and TFE do NOT have very similar dielectric constants!
Website after website reports the dielectric constant of TFE as 8.55. This is close to the
dielectric constant of DCM (8.93)
51
. However, I have since discovered that the dielectric
constant of TFE is actually around 27
52
! It is possible that some portion of the change in ESPT
rate constant in the high concentration regime comes from the changing bulk dielectric
stabilizing the ESPT products. Perhaps this is why the slope in this region is above 1. However,
as we have already seen and will continue to see below, there is still plenty of strong evidence
of the need for two proton donors.
56
At lower concentrations of TEA, we no longer see a trend consistent with the
molecularity of the reaction. This is because diffusion becomes the rate limiting step, so the
kinetic trend scales with concentration in a way that reflects diffusional motion. This “diffusive
regime” is fit with an orange line in figure 3.9.
At these concentrations, based on the steady state analysis above, many of the
photobase molecules are hydrogen bonded to a TFE molecule in the ground state. At the lowest
concentration in figure 22, we estimate that about 30% of the MeOQ molecules are hydrogen
bonded in the ground state, while at all higher concentrations more than 80% of the
photobases are hydrogen bonded to the donor. We propose that on average there are no
additional donor molecules in the solvation shell. This makes sense: there is almost certainly no
enthalpic drive for a second proton donor molecule to be nearby the already hydrogen bonded
photobase, and TFE makes up around only 20% of the solution by mole at the highest
concentrations in this regime. The vast majority of photobase molecules, then, should have at
most one MeOQ molecule in their immediate solvation shell. If two or more TFE molecules
must be present in the solvation shell (one hydrogen bonded, one or more for stabilization of
the products), then one or more must diffuse to the reaction center, as mentioned above.
The behavior of this regime can be readily explained using a simple diffusion model. The
average diffusion length (L) for two molecules to find one another in solution is given by
𝐿 = √𝐷 𝜏
where D is the mutual diffusion coefficient (
C"
$
2
) for the two molecules and 𝜏 is the diffusion
time (s), the amount of time it takes for the two molecules to find one another. If we assume
57
that the reaction occurs successfully whenever the hydrogen-bonded photobase finds another
TFE molecule, the reaction timescale will be described by 𝜏. Given the strong thermodynamic
drive for the ESPT process between MeOQ and TFE, this is probably a reasonable assumption.
We know that the average distance d (cm) between TFE molecules in solution is related
to the concentration c (
"342
C"
%
) through the following equation:
𝑑 = (𝑁
6
𝑐)
'
;
=

where NA is Avogadro’s number. Because the concentration of TFE is significantly higher (at
least 1 order of magnitude) than the concentration of MeOQ in all of the relevant solutions, our
diffusion length will largely be determined by the concentration of TFE molecules and we can
assume that d is equivalent to L. We can then combine these two equations to predict the
timescale of diffusion of TFE to the reaction center (and thus the reaction timescale):
𝜏 =
(𝑁
6
𝑐)
'
7
=
𝐷

By reciprocating and taking the log of both sides, we get a linear version of this equation
that shows the rate of the ESPT reaction as a function of concentration for a diffusive scheme:
log
;:
𝑘 =
2
3
log
;:
𝑐+U
2
3
log
;:
𝑁
6
1000
+log
;:
𝐷V
Note that the parentheses contain the y-intercept. The factor of
;
;:::
comes from
converting concentration from
"342
C"
%
in the diffusion equations above to
"342
D
so that the
58
equation can be plotted on the same axis as the non-diffusive regime in figure 3.9. The model is
shown in figure 3.9 as a purple line, and it agrees with the linear fit to the data very well.
The slope of this linear diffusion equation is 2/3. The linear fit of the log-log diffusion
regime in figure 22 has a slope of 0.69. The similarity of the values provides evidence that the
model accurately describes the ESPT dynamics at these TFE concentrations. We can obtain
additional confirmation by using an estimate of D to calculate the y-intercept and comparing it
with the y-intercept from the fit. If we assume that TFE, MeOQ, and DCM are all around the
same size, then we can assume that they will all have approximately the same diffusion
coefficient in the solution. Since the bulk of the solution is made up of DCM, we will estimate
this value using the self-diffusion coefficient of DCM – the rate at which a DCM molecule
diffuses in pure DCM – which is reported as 4.0 𝑥 10
'.

C"
$
2
at room temperature
53
. We are
ultimately interested in the mutual diffusion coefficient of MeOQ and TFE – the value
describing the diffusion of MeOQ and TFE towards one another – which should be the sum of
their individual diffusion coefficients. Therefore, the expected value of D, the mutual diffusion
coefficient, is about 8.0 𝑥 10
'.

C"
$
2
. Calculating the expected y-intercept using this value gives
us a unitless value of 9.8. Within significant digits, this is the exact same value we extract for
the y-intercept in our linear fit of the log-log data in the diffusive regime. Our linear diffusion
model is very simple, and we did not anticipate the match to be exact!
That both the slope and the mutual diffusion coefficient are within the same range as
the values predicted by our diffusion model strongly supports the diffusive mechanism for this
TFE concentration range. That diffusion of a proton donor to an already hydrogen-bonded
59
photobase is necessary confirms the importance of extra donor molecules for stabilization of
the ESPT products! Furthermore, if diffusion of more than one TFE molecule was necessary for
ESPT, we would expect the relevant diffusion timescales to be much longer than the ones we
observed experimentally – or for the reaction to not proceed at all during the photobase’s
excited state lifetime!
These results reveal the challenges of pre-association and solvation of photogenerated
products in the practical applications of photobases. Even if there is a large thermodynamic
drive for deprotonation of a proton donor (as there is in this scenario), dissociation entropy
must be overcome with a large excess of the donor before the necessary pre-association is
achieved. Furthermore, solvation of the photogenerated products must be ensured for excited
state proton transfer, and this requires an even larger excess of proton donor, based on our
findings!
While similar results have already been shown for photoacid ESPT reactions with
similarly high thermodynamic drives
49
, it’s important to note that it’s possible that these results
may not be entirely generalizable. For instance: if the drive for ground state hydrogen bonding
becomes much larger, will such a large excess for ground state complexation be necessary? If
the thermodynamic drive for the ESPT reaction becomes even larger, will the stabilization of
the products still be as important for the reaction to proceed? Could the reaction proceed
without two proton donors in that case? Could the ESPT products be screened by a high
dielectric solvent instead of specific hydrogen-bonding interactions? These questions represent
exciting directions for continued study.
60
In the meantime, we can work to generate strategies for overcoming the hydrogen-
bonding and ESPT solvation difficulties we have uncovered. One strategy is directly tethering
the photobase to a location where it is needed, thereby avoiding the necessity for pre-
association. For example, a photobase can be covalently bound to the ligand periphery of an
organometallic catalyst as a pendant moiety. The center of action for deprotonation is the
catalytic metal.
When a substrate binds to the metal and the reaction progresses to the stage that it
becomes limited by proton transfer, excitation of the photobase can facilitate proton removal
from the substrate. Work by us and our collaborators has resulted in the synthesis of a model
iridium complex with a pendant photobase that intends to utilize this strategy
40
. Even though
the catalytic properties of the complex is still unclear, this is a first step toward rationally
incorporating photobases in catalytic centers. This system will be discussed in the next chapter.

61
Chapter 4
Integration of a Photobase Molecule into an Iridium Catalyst
Most of the work described in this chapter originally appeared in the article “Optical pKa
Control in a Bifunctional Iridium Complex” by Ivan Demianets, Jonathan R. Hunt, Jahan M.
Dawlaty, and Travis J Williams in Organometallics
40
.
We discovered in the last chapter some of the challenges associated with the
application of photobases for chemical problems. Namely, some photobase molecules require a
large excess of proton donor molecules for ground-state preassociation and an even larger
excess for solvation of the ESPT products. If one hopes to take advantage of photobases’ ability
to finely control the transfer of protons using light, one must find ways to overcome these
challenges.
As mentioned at the end of the previous chapter, one strategy for overcoming the
issues associated with pre-association is to tether the photobase near a reaction site where a
proton donor molecule will reside. For example, a photobase can be covalently bound to the
ligand periphery of an organometallic catalyst as a pendant moiety. When a substrate binds to
the metal and the reaction progresses to the stage that it becomes limited by proton transfer,
excitation of the photobase can facilitate proton removal from the substrate. In collaboration
with the organometallic group of Travis Williams and his student Ivan Demianets, we have
studied such a complex, shown below. Because I was not involved in the synthesis of this
complex, I will only be discussing our photophysical investigations into the complex in this
62
chapter. The details of the synthesis and characterization of this complex can be found in the
original publication
40
.

Figure 4.1: The synthesized iridium complex with pendant quinoline photobase. The photobasic
action of the pendant quinoline is shown.

There have been several inorganic photoacids and photobases described in the
literature, mostly involving ruthenium and functionalized polypyridine ligands
54–60
, but there
are also examples involving transition metals Re(I)
61
, Fe(II)
62
, Os(II)
63,64
, Pt(II)
65
, Ir(III)
66
, and
Rh(III)
67
, as well as some bimetallic
67
and trimetallic
62
complexes. In all these examples, the
metal is part of the photoacid or photobase chromophore. Our complex is different: the
pendant quinoline can be excited and behave like a photobase without necessarily also exciting
the metal center. This is the first example – to the best of our knowledge – of an organometallic
complex with a covalently tethered photobase that can operate orthogonally to the complex
itself, allowing for control of the ESPT reaction separate from the possible catalytic operation of
the complex. Evidence for this orthogonality will be discussed below.
Ivan and I set out to characterize the photobasicity of the quinoline in this complex. We
wanted to show that ESPT still occurs in the covalently linked quinoline and estimate its excited
state pKa. We worried that the covalent linkage to an organometallic complex could
63
significantly affect the electronics of the molecule, and perhaps perturb the photobasic
behavior of the quinoline. Since we could not take for granted that the molecule still exhibits
photobasic behavior attached to the complex, we needed to prove it. For the same reasons we
could not assume that the excited state pKa was unchanged and so needed to characterize it.
It would be nice to use the Förster thermodynamic cycle to measure the excited state
pK a* value of the pendant quinoline. However, Förster analysis of the pendant quinoline of the
organometallic complex is frustrated by overlap of broad iridium absorption features with those
of the quinoline chromophore and by the instability of the complex in acidic aqueous solution.
We were therefore unable to perform a clean Förster cycle analysis. In chapter 2 I used the
emission of 5-methoxyquinoline in organic solvents of varying pK a values to bracket its excited
state pK a* as an alternative to Förster cycle analysis. We can use the same logic here. However,
because the pendant quinoline molecule of the complex is more closely related to
unsubstituted quinoline, we must first familiarize ourselves with the kinetics and emission
behavior of the unsubstituted quinoline molecule before we can understand the behavior of
the complex.
It is well-known that unsubstituted quinoline undergoes rapid (sub ps) intersystem
crossing following photoexcitation
68–71
. This converts excited quinoline into a triplet state, so
under normal conditions there will be very little emission from the excited quinoline molecule.
However, the same behavior is not observed for protonated quinoline – it emits with
reasonable efficiency. Previous studies of quinoline have used Förster cycle analysis to estimate
that the excited state pKa of quinoline is around 11.5. Therefore, if the unprotonated quinoline
molecule can grab a proton from a donor before the occurrence of ISC, it will avoid ISC
64
altogether and can go on to emit. In summary: if the ESPT process of quinoline is unsuccessful,
we should expect to see virtually no emission; if the ESPT process is successful and very fast, we
will see emission from the protonated form. As before, in all cases, we assume that quinoline is
unprotonated in the ground state such that all emission from the protonated form arises due to
ESPT.

Figure 4.2: A,B) Emission spectra of quinoline in various proton-donor solvents. C,D) Emission
spectra of iridium complex with pendant quinoline in various proton-donor solvents.

Emission from quinoline displaying these characteristics is shown on the left in figure
4.2. In hexafluoroisopropanol (HFIPA), which has a pKa of around 9.3
72
, we see prominent
emission from the protonated form of quinoline, indicating that there is successful, fast ESPT.
Iridium
Complex
w/ Quinoline Quinoline
65
We also see protonated emission in 2,2,2-trifluoroethanol (TFE), which has a pKa of 12.5
52
, but
far less than is seen in HFIPA. In higher pKa solvents (ethanol, isopropanol) and aprotic solvents
(DCM), no protonated emission is seen. All of these observations are consistent with the
observations made above and the 11.5 pKa* of quinoline.
Now we may study the emission from the complex in similar solvents to study the ESPT
reaction of the pendant quinoline molecule – namely, we ask, is quinoline still a photobase, and
is it still as strong as in its free molecular form? This emission may be seen on the right side of
figure 4.2. Once again, emission from the protonated form is observed in HFIPA. This indicates
that the quinoline moiety can still complete the ESPT reaction. Notice that there is little or no
emission from the protonated form in TFE for this version of quinoline. This suggests that
covalent attachment to the organometallic complex did affect the electronic structure of the
molecule: with respect to the molecular version of quinoline, the tethered quinoline’s pKa*
seems to be lower. We can therefore estimate the pKa* of the pendant quinoline in the
following way: on the low end, it’s around 9.3, the pKa of HFIPA; and on the high end, it’s less
than 11.5, the pKa* of the unsubstituted molecular quinoline.
The absorption and emission of the complex in HFIPA is shown in figure 4.3. Although
this data is quite simple, it contains insight into the orthogonality of the complex – the ability of
the quinoline molecule to be excited and exhibit photobase behavior separate from that of the
complex itself. The two lowest energy absorption bands of the complex can be assigned to 𝑑 →
𝜎 MLCT (472 nm) and 𝑑 → 𝜋
∗
MLCT to the N-heterocycle (530 nm), respectively, by analogy to
well-characterized systems in the literature
73
. The emission we observe for complex 2 in HFIPA
is blue shifted with respect to both MLCT absorption bands. Since emission cannot be higher in
66
energy than its corresponding absorption, the emission we see at 440 nm must be from direct
excitation of the quinoline moiety rather than from excitation of the complex followed by
energy transfer.

Figure 4.3: Absorption and emission spectra of iridium complex with pendant quinoline.
As mentioned above, this is the first example of an orthogonal photobase molecule
attached to an organometallic catalyst. Application of this idea to catalytic systems where
proton removal from the reaction site is the rate limiting step is a thrilling concept. That work is
ongoing in the lab of Travis Williams.

67
Chapter 5
Excited State Electronic Structure Calculations of Photobase Molecules
Much of the work described in this chapter originally appeared in the article “Donor-
Acceptor Preassociation, Excited State Solvation Threshold, and Optical Energy Cost as
Challenges in Chemical Applications of Photobases” by Jonathan Ryan Hunt, Cindy Tseng, and
Jahan M. Dawlaty in Faraday Discussions
38
.
One good question that arises when studying photobase molecules is: “what is the
electronic origin of the photobase effect, and why does the effect show a thermodynamic trend
with respect to electron-donating capabilities of substituents?”
The hand-wavy explanation we use to answer such questions is this: because excited
states are more easily polarized, electrons are preferentially drawn towards electronegative
atoms in the excited state. The most electronegative atom in quinoline is nitrogen, so electronic
charge piles up on this atom upon excitation. This preferentially attracts protons to the basic
site in the excited state, thus making the molecule behave like a photobase. Being in a more
easily polarized excited state only amplifies the effects of the substituent: if a substituent was
electron donating in the ground state, it will be even more so in the excited state. This will
make the molecule more photobasic, since it will result in even more charge building up on the
electronegative nitrogen atom. This explanation has an elegance to it akin to much that is
learned in an organic chemistry class. However, the explanation needed to be proven!
68
This question seemed like a great candidate to be explored using quantum chemistry
calculations. If excitation of quinoline did indeed lead to a buildup of electron density on the
heterocyclic nitrogen, this should be seen easily enough in electronic structure calculations.
Even more exciting was the possibility of seeing a trend in the electronic structure that would
correlate with the experimental thermodynamic trend as a function of substituent, shown in
figure 5.1. It would maybe even be possible to find some theoretical observable that could
correlate with experimental values and be used in the future to predict the photobasicity of
other molecules.

Figure 5.1: Ground and excited state pKa trends for 5-R-substituted quinolines

I studied the excited state electronic structure of the 5-substituted quinoline molecules
from Eric’s paper so that my theoretical results could be correlated with his experimental ones.
69
All electronic structure calculations were carried out using Q-Chem software. Geometries of
quinoline and its 5-substituted forms were optimized in the ground state via DFT using the
omega-B97X-D/6-31+G* level of theory. The excited state analysis was then carried out at two
different levels of theory for comparison’s sake: TDDFT/omega-B97X-D/6-31+G* and SOS-
CIS(D)/6-31+G*/RIMP2-cc-PVDZ. These two levels of theory have similar computational cost,
and we would like to know which has better correlation with experimental data. For each level
of theory, electron density difference maps were generated for excited states of interest via the
$plots section of Q-chem and were then visualized using VMD. Löwdin population analysis was
carried out as a numerical analog to the electron density difference maps. Specific details about
the calculations, how I calculated electron density differences, and how I visualized those
differences is provided in appendix D.
The excited state calculations provided several possible excited states of interest.
Specifically, two states of similar excitation energy and of notable oscillator strength were
typically observed as the lowest energy excited states. These represented the L A state, which
has a transition dipole moment along the short axis of the molecule and, in these calculations,
has a leading configuration of HOMO -> LUMO; and the L B state, which has a transition dipole
moment along the long axis of the molecule and has a leading configuration of (HOMO – 1) ->
LUMO. The L A state tends to pile up charge density on atoms – including the heterocyclic
nitrogen atom, the site of proton transfer in our systems – and was selected as the state of
interest even though the L B state was often lower in energy. This choice was justified because
the L A state was experimentally observed by Eric Driscoll to be the lowest lying excited state in
70
all of the molecules studied
27
. It is also worth mentioning that the L A state consistently had the
higher oscillator strength in the calculations.
R = NH 2
L A L B
Excitation energy (eV) 4.08 4.52
Oscillator strength 0.112162 0.017869

R = OCH 3
L A L B
Excitation energy (eV) 4.63 4.74
Oscillator strength 0.112587 0.004506

R = H
L A L B
Excitation energy (eV) 5.02 4.79
Oscillator strength 0.070637 0.023979

R = Br
L A L B
Excitation energy (eV) 4.80 4.73
Oscillator strength 0.121854 0.032286

R = Cl
L A L B
Excitation energy (eV) 4.84 4.74
Oscillator strength 0.111615 0.022430

R = CN
L A L B
Excitation energy (eV) 4.89 4.70
Oscillator strength 0.183791 0.059636

Table 5.1: Theoretical excitation energies and oscillator strengths for the L A and L B states of 5-R-
quinolines calculated at the TDDFT/omega-B97X-D/6-31+G* level of theory.

Here I will provide a brief aside regarding the concept of “leading configurations”, which
have been mentioned several times. Occupied and unoccupied molecular orbitals can be
calculated for a given molecule in the ground state. These molecular orbitals act as a basis set
71
for explaining the electronic structure of the molecule of interest. The most commonly
discussed molecular orbitals are the highest occupied molecular orbital (HOMO) and the lowest
unoccupied molecular orbital (LUMO). When discussing the lowest energy excited state of a
molecule, it is common to think of the change in electronic structure as a transition of an
electron from the HOMO to the LUMO. Similarly, higher lying electronic states are often
thought of in terms of electronic transitions from the HOMO to the LUMO+1, and so on.

Figure 5.2: Q-Chem outputs describing the singlet L A and L B states in terms of molecular orbital
transitions.

Because these molecular orbitals are calculated for the ground state’s electronic
structure, though, they are often imperfect at describing excited state electronic structures. To
fully capture the true nature of an excited state, it is often necessary to represent an electronic
excitation as a sum of multiple orbital transitions. In figure 5.2, one can see the TDDFT output
for the two lowest singlet excited states of unsubstituted quinoline. These are the L B state
72
(labeled “excited state 6”) and the L A state (labeled “excited state 7”). One can see that the L A
state is largely explained by HOMO -> LUMO (or D(34) -> V(1)), but also has some contribution
from HOMO-1 -> LUMO+1 (or D(33) -> V(2)). The orbital change that describes the largest
percentage of the electronic transition – in this case HOMO -> LUMO – is called the “leading
configuration”. Note that the percentage contribution of a given orbital transition can be
obtained by squaring their amplitudes.

Figure 5.3: Leading configuration analysis and electron density difference analysis for the L A
state of quinoline.

Electron density difference (EDD) maps, which show differences in electron density
between the excited state of interest and the ground electronic state, were calculated and
Electron
Density
Difference
Analysis
Leading Configuration Analysis
!"
HOMO LUMO
#
!
73
plotted for the L A states of these molecules. These maps are helpful because they give an easy-
to-understand picture of the changes in the electron distribution upon photoexcitation. They
also show changes that might be ignored by a traditional analysis of leading configurations,
since they include contributions from non-leading configurations as well. In figure 5.3, I show
the leading configuration analysis and the EDD map for the L A transition of unsubstituted
quinoline. Aside from the benefits discussed above, EDD maps only require one image to show
the effects of the transition, where traditional leading configuration analysis requires two. The
electron density difference map immediately provides qualitative evidence for the
photobasicity of the molecule. An increase in electron density on the nitrogen heteroatom is
apparent, and an increase of electron density on the basic site means that the molecule is more
likely to attract protons in the excited state. That means it’s photobasic according to our
proposed mechanism.
But wait! The results from the EDD maps are even more exciting than this! Comparison
of EDD maps for differently substituted quinoline shows a clear trend: electron donating groups
lead to a greater change in electron density on the heterocyclic nitrogen, while electron
withdrawing groups lead to a smaller change in electron density on the heterocyclic nitrogen.
This qualitative trend is shown in figure 5.4 below for select substituted quinolines. In terms of
our hypothesis on the origins of photobasicity, this would imply that electron donating groups
lead to greater photobasicity, while electron withdrawing groups lead to weaker photobasicity.
74

Figure 5.4: Electron density difference maps for 5-R-quinolines as a function of substituent
directing effect.
R = NH
2
(donating)
R = CN (withdrawing)
R = H
ElectronDensity Difference on N Decreases
75
To quantify this trend, we used a population analysis tool called Löwdin analysis. This
analysis essentially assigns a certain number of electrons to each atom in the molecule based
on electron density in the vicinity of that atom – it answers, for example, “how many electrons
in this molecule belong to the nitrogen atom?”. The value that is reported is the excess (or
deficient) number of electrons associated with that atom. A “neutral” atom, where the number
of electrons exactly matches the number of protons, will have a Löwdin charge of 0. A negative
number means the atom has more electrons than protons, and a positive number means it has
more protons than electrons.
We plotted the experimentally determined change in pKa upon photoexcitation (ΔpKa)
as a function of the change in number of electrons (Δq) on the nitrogen - that is, the Löwdin
charge in the excited state minus the Löwdin charge in the ground state. The results are
strikingly linear, as shown below in figure 5.5. The trendline for the TDDFT analysis suggests
that one additional electron charge added to the nitrogen heteroatom should result in a pKa
increase of 65.1 units. In more reasonable terms, the trend predicts that the addition of 0.015
electrons onto the heterocyclic nitrogen should correspond to a 1 unit increase in the pKa.
76

Figure 5.5: Trends for experimentally determined Δ𝑝𝐾
!
versus theoretical excess excited state
charge density on the N heteroatom for TDDFT and SOS-CIS(D) excited state electronic
structure methods.

-0.06 -0.04 -0.02 0 0.02
Excess Excited State Charge Density on N(e)
2
4
6
8
10
12
Experimental pK
a
OCH3
NH2
CN
Br
H
Cl
Acridine
-0.16 -0.14 -0.12 -0.1 -0.08 -0.06 -0.04 -0.02
Excess Excited State Charge Density on N(e)
2
4
6
8
10
12
Experimental pK
a
Acridine
NH2
OCH3
Cl
H
Br
CN
-0.16 -0.14 -0.12 -0.1 -0.08 -0.06 -0.04 -0.02
Excess Excited State Charge Density on N(e)
2
4
6
8
10
12
Experimental pK
a
Acridine
NH2
OCH3
Cl
H
Br
CN
!"#
!
=−&'.)*
"#$"%%
+).)
!"#
!
=−),-*
"#$"%%
+'..
TDDFT
SOS-CIS(D)
77
I must mention one fundamental approximation in my electronic structure calculations.
Immediately after excitation, due to separation of electronic and nuclear timescales of motion,
a molecule will have a new electronic structure but the exact same nuclear structure as in its
ground state. It will take a short time, often on the order of picoseconds, for the molecule to
find a nuclear structure at equilibrium with its new electronic structure. The fully relaxed
molecule is often the species of interest in the ESPT reaction and should therefore be, in an
ideal case, the molecule studied via electronic structure calculations. However, we do not study
the molecules following excited state nuclear relaxation; we study our molecules fixed at the
ground state geometry with the new excited state electron density before that relaxation
occurs.
The relaxation following vertical excitation likely has an impact on the observables we
hope to correlate with excited state proton transfer. For example, the geometry optimization in
the excited state may result in even more electron density on the heterocyclic nitrogen.
However, including this relaxation by performing excited state geometry optimizations turned
out to be quite challenging. As such, we made the approximation that the vertically excited
states are similar enough to their relaxed counterparts that they will provide us with
meaningful information. Seemingly, this turned out to be a quite good approximation based on
the correlation with experiment.
As mentioned earlier, I used two different electronic structure methods – TDDFT and
SOS-CIS(D) – to study the excited states and the trends associated with the excess electronic
charge on the heteroatom. While both methods show an obviously linear trend between
ΔpKa and the change in Löwdin charge, TDDFT seems to capture the electron density changes
78
more accurately. While the linear correlation between experimentally determined ΔpKa and
theoretically determined Löwdin Δq’s using TDDFT yielded an R^2 value of 0.9757, the R^2 for
the correlation when SOS-CIS(D) delta-q’s are used is 0.7801.
TDDFT excitation
energy (eV)
SOS-CIS(D)
excitation energy
(eV)
Experimental
excitation energy (eV)
-NH2 4.08 4.79 3.94
-OCH3 4.63 5.19 3.96
-H 5.02 5.30 3.96
-Br 4.80 5.20 3.91
-Cl 4.84 5.21 3.91
-CN 4.89 5.14 3.88

Average error 0.78 1.21
Standard deviation
of error
0.34 0.18

Table 5.2: Comparison of errors between experiment and theory for TDDFT and SOS-CIS(D)
excited state electronic structure theory methods for 5-R-quinoline L A excitation energies.

TDDFT also gave more accurate excitation energies for the excited states of interest. The
calculated excitation energies for the La states of each 5-substituted quinoline are shown in
table 5.2 for both TDDFT and SOS-CIS(D) methods and are compared to the excitation energies
obtained via experiment. The average error of the TDDFT excitation energies (0.7826 eV) is
considerably smaller than the average error of the SOS-CIS(D) excitation energies (1.2109 eV). It
is interesting to note, however, that the standard deviation of the errors was smaller in SOS-
CIS(D) (0.1824 eV) than for TDDFT (0.3419 eV). This means that, although the excitation values
for TDDFT are on average closer to the experimental excitation energies, the SOS-CIS(D)
excitation energies tend to be off from the experimental values in a more systematic way. It
79
would therefore perhaps be easier to come up with a correction factor to account for the
differences between experiment and theory for SOS-CIS(D) values than for TDDFT values. It
would be interesting to study the basis-set dependence of the excitation energies to see
whether SOS-CIS(D) could outperform TDDFT in both precision and accuracy with a larger basis
set. For now, we choose to use TDDFT in all future calculations because of its superior
performance at this level of theory and basis set size.
The success of this correlation suggested that our trend might be applied to predict the
ΔpKa of similar photobase systems through the calculation of the change in electron density via
Löwdin analysis. Additional confidence was provided by the fact that another photobasic n-
heterocycle, acridine, fit decently well onto the curve, as shown in figure 5.5. We decided to
study a series of n-heterocycles in this way to probe the relationship between substituent
effect, conjugation length, excitation energy, and change in pKa upon photoexcitation. We
expected that electron-donating substituents should increase the 𝑝𝐾
!
∗
and that conjugation
length should decrease the excitation energy, but we didn’t know how these effects should
intertwine. The ideal photobase would have a small excitation energy and a very large ΔpKa.
What is the best way to attain such a system?
In theory, we can apply our electron-density picture to any potential aromatic N-
heteroatomic photobase molecule. We therefore chose to study quinolines, acridines, and
benzacridines, which have two, three, and four conjugated rings, respectively. We performed
calculations on these molecules with a range of substituents of varying electron-withdrawing
strengths. The substituent positions on these molecules were all the same and are shown in
figure 5.6 below. Our experimental pKa* data is from 5-substituted quinolines, so this
80
substituent position was a clear choice. We chose the analogous positions on acridine and
benzacridine. The substituents studied, in order of increasing electron withdrawing power, are
-NH2, -OCH3, -Cl, -Br, and -CN, in addition to the unsubstituted molecules.

Figure 5.6: Calculated excess excited state charge density on N-heteroatom versus computed
excitation energy for a variety of substituted aromatic N-heterocycles. Arrows point in the
direction of increased electron-withdrawing capabilities of the substituent.

As expected, the energy required for excitation decreases for the systems with more
conjugation. However, this decrease in excitation energy coincides with a reduction of the
excess electron density on N, which corresponds to a reduction of the ΔpKa. This observation is
logical, because the excitation will be more delocalized over the molecule and there should
3 3.5 4 4.5 5
-0.18
-0.16
-0.14
-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
Computed Excitation Energy
Lowdin q
Increasing Photobasicity
(eV)
(e) Excess Excited State Charge Density on N (e)
81
therefore there should be less charge buildup in one spot, like the heteroatom. The size of the
conjugated system is intimately coupled with the strength of the photobasicity – one can’t
arbitrarily make a photobase system larger to red shift the absorption without sacrificing some
of the proton-grabbing power.
Next, we draw our attention to the effect of the substituent on the photobasicity of the
system. The substituent makes a larger difference for smaller conjugated systems. For the
largest molecules studied here (benzacridines), the substituents barely change the excitation
gap and the excess charge on nitrogen. For smaller molecules, however, the effect is more
pronounced. This is reasonable, since a single substituent should have a smaller perturbing
force on a larger electronic system. From this observation, we learn that – even though it is
possible to tune the excitation gap and the photobasic drive with substituents – that effect is
much weaker on large systems.
As the electron-withdrawing strength of the substituent is increased (following the
arrows in figure 5.6), the change in charge density on N decreases. It is expected that increasing
the withdrawing strength should decrease the electron density on N in both the ground and
excited states. However, the fact that the difference between these values also decreases as a
function of increasing withdrawing strength implies that the excited state charge density is
more sensitive to the substituent than the ground state charge density. This makes sense
because excited states are more polarizable and thus more subject to substituent effects. The
suggestion is clear: if a substituent makes the molecule a better base in the ground state, it will
make it a better photobase in the excited state; if a substituent makes the molecule a worse
base in the ground state, it will make it a worse photobase in the excited state.
82
It is important to interpret these results with caution. These calculations use the lowest
optically active excited state of each molecule to predict a ΔpKa value. However, these
calculations do not consider any other possible states or excited state kinetic pathways that
may contribute to the behavior of the excited molecule. For instance, it is possible that the
molecules could undergo intersystem crossing that’s kinetically faster than or competitive with
the proton-transfer process. This could prevent excited state proton transfer, even if a large
thermodynamic drive is present. These issues are discussed in my work with Eric Driscoll on the
kinetics of excited 5-R quinoline molecules
28
. While it is possible for triplet states to be
photobasic, the drive for proton transfer is likely smaller due to the lowered energies of the
triplet manifold relative to the singlet manifold. Therefore, the values presented in the above
figure are only relevant if the proton capture reaction is otherwise kinetically feasible.
We have found some design trends for tuning the optical gap for photobases, with the
intention of finding molecules that will achieve photobasicity at lower excitation energies. We
have realized that this design space, at least in the dimensions studied here, is rather
constrained. Increasing the conjugation size lowers the optical gap, but also adversely affects
photobasicity. Tuning the gap via substituents is more efficient for smaller conjugated systems.
Our study lays the foundation for exploring this design space further. For example, multiple
substituents, multiple heterocyclic nitrogens or other heteroatoms in the conjugated system,
exploring non-benzene rings, and identifying the influence of triplet states on the photobase
kinetics are all potentially fruitful directions of future work.

83
Chapter 6
Other Literature and Future Directions
I am proud of my work on quinoline photobases. These publications may someday prove
important for the development of fine control over ESPT reactions in a variety of chemical
scenarios. Even when my work showed the limitations of photobases, I was happy to
communicate an accurate picture of these systems so that researchers in the future are better
prepared for the reality of their application.
ESPT in quinoline photobases – and other photobases – is still an active and fruitful area
of research. Below, I provide a brief glimpse into recent works that were inspired by my own
and into a new and exciting photobase molecule. I then list what I believe are the most
interesting and potentially fruitful research questions related to quinoline photobases.
6.1 Works Inspired by My Own and An Exciting New Photobase
Following publication of the works described throughout this manuscript, there have
been additional important publications in the field of photobase chemistry. I will briefly discuss
three of them below. Two were directly inspired by my own publications. I will mention these
first. The last is a work about an exciting new photobase molecule, published around the same
time as my own works on 5-MeOQ.
In 2019, an article titled “5-Methoxyquinoline Photobasicity is Mediated by Water
Oxidation” by Roy et al. was published in the Journal of Physical Chemistry A
74
. This paper, a
direct response to the work I had done on the 5-MeOQ photobase, made this extraordinary
84
claim: following excitation of 5-MeOQ, water donates an electron to the photogenerated
HOMO hole. Now that 5-MeOQ has been reduced, it can readily attract a proton from a nearby
proton-donor molecule. The photobasicity, according to the authors, is not due to an electronic
rearrangement of the singlet state, placing additional electron density on the heteroatomic
nitrogen, that results in the ability of 5-MeOQ to deprotonate proton donors more readily in
the excited state. Instead, the molecule is reduced! The authors therefore claim that Förster
cycle analysis is ill-suited for studying the proton transfer thermodynamics of the system. In the
TOC graphic for the paper, the authors rather creatively write “FÖRSTER CYCLE” with two red
lines crossing out the words, to imply how silly it is to use the Förster cycle analysis on this
system.
These conclusions are drawn from molecular dynamics simulations. In short, excited 5-
MeOQ is placed in a small cluster of water molecules and the resulting dynamics are observed.
In 95 out of the 100 simulations, the 5-MeOQ molecule is reduced by the surrounding water
molecules. The reduced MeOQ then undergoes rapid proton capture due to the Coulombic
attraction, and the species rapidly returns to the ground state to produce unexcited MeOQH+
and OH-. The authors note that it is impossible for charge reorganization within the molecule to
be responsible for the photobasic behavior because the states with significant charge transfer
character to the N heteroatom are significantly higher in energy than the excitation energies
used in our experiments.
The issue regarding the hypothesis of reduction being responsible for MeOQ
photobasicity is that it is completely inconsistent with the work I’ve presented throughout this
manuscript. For a work that makes such extraordinary claims and challenges years of my own
85
work, there is a surprising lack of correspondence between what is claimed in this paper and
what I observe in my experiments. For example, if MeOQ were reduced and then underwent
proton transfer, how could there be emission from the protonated form? A reduced molecule is
no longer emissive, because there is no empty HOMO orbital for an electron to return to. But in
all solvents I’ve studied, including water, there is efficient emission from the protonated form
of MeOQ following ESPT that matches the emission seen when ground state protonated MeOQ
is excited. This is simply irreconcilable with the concept of MeOQ being reduced prior to proton
transfer. This is perhaps why the authors never mentioned my emission studies.
Their hypothesis is also easily invalidated by my transient absorption and TCSPC work
into 5-MeOQ’s ESPT reaction. We see a clean transition from the base form of the molecule to
the acid form of the molecule. These assignments are made unambiguously, as is discussed at
great length in this manuscript and in my publications. Furthermore, the authors claim that
following proton transfer to the reduced MeOQ molecule, there is rapid deactivation. Our
transient absorption data shows that the product of the reaction is long-lived
37
; TCSPC reports
that the excited state lifetime of the protonated species is multiple ns
39
. It is of course
important to note, once again, that the system could not be studied via TCSPC if MeOQ had
been reduced, since the molecule would no longer be emissive. The authors say, in their
conclusions, that “the excited state lifetime, determined experimentally and theoretically to be
on the order of 10 ps, is too short for acid-base equilibration…”. This is a blatant
misrepresentation of my works, seemingly made for the reason of strengthening their own
argument.
86
I believe this work can act as a cautionary tale. There are times when theoretical
calculations may predict extraordinary behavior, as is the case here. However, these predictions
are only as good as their correlation with experiment. Ignoring or misrepresenting experimental
evidence that contradicts those theoretical claims is scientifically indefensible - especially when
those contradictions are as obvious as the ones discussed above.
Another work directly inspired by my own is “Structure-Photochemical Function
Relationships in Nitrogen-Containing Heterocyclic Aromatic Photobases Derived from
Quinoline” by Alamudun et. al. in JPCA
75
. The authors were inspired by my paper, presented in
chapter 5, where I used theoretical calculations for the electron density change on the nitrogen
heteroatom to predict the ΔpKa for a variety of singly substituted N-heterocycles. In the work,
Alamudun et. al. expand my work to include many more molecules, including substituted
isoquinolines and azaanthracenes in addition to the quinolines and acridines. All substituent
positions were investigated, whereas my study only included the 5-position. Most interestingly,
they studied multi-substituted versions of these molecules.
There are a few other improvements from my work made by these authors. First, they
included excited state nuclear relaxation in their theoretical calculations. In my work, I used the
electron density change on the N-heteroatom following vertical excitation of the relevant
molecules. I did not include any nuclear reorganization that may have occurred following the
excitation of the molecule. A more rigorous approach would be to include those excited state
geometry changes and then calculate the change in charge density. This is exactly what the
authors have done by performing excited state geometry optimization calculations. In my
defense, I was aware that this would be a more rigorous way to obtain these values. I just didn’t
87
have the expertise necessary to perform the relevant excited state geometry optimizations! I
am glad that someone with such expertise stumbled upon my work.
Another improvement is the inclusion of solvation effects. In my work, all calculations
were performed in the gas phase. In experiment, solvation of the photobase molecule will
influence the nuclear and electronic structure, resulting in different Δq’s than the ones I
calculated. The authors used a PCM model to mimic the bulk effects of water, thereby closer
approximating experimental conditions. The inclusion of solvent effects will also improve the
calculation of excitation energies, which are important to understand since, experimentally,
appropriate photon energies must be supplied to excite these molecules. Even these
theoretical conditions are imperfect: specific interactions, namely hydrogen bonding to the
heteroatom, are likely important to get a full picture of the electronic structure of N-
heterocycles. Regardless, this was a great improvement over my own work and I am grateful
that it was performed.
There were several interesting revelations from this work. First, the authors studied the
effect of substituent position on the pKa* of quinoline. It was clear substituents in the 5
position had the greatest impact. The most interesting discovery was that, while adding
electron-donating substituents to all other positions positively impacted the photobasicity to
varying degrees, adding electron-donating substituents to the 4 position negatively impacted
the photobasicity of the molecule! Clearly, the directing effects taught in organic chemistry are
still relevant to excited states. Our previous claim that “the substituent polarizes the excited
state” does not tell the whole story!
88
Another important discovery is that the substituent effect is often additive. By adding
several electron-donating substituents to the same molecule, provided that they are in optimal
positions, one can further increase the pKa* of the photobase molecule. One important
consequence of this fact is that, by adding multiple substituents to larger N-containing
aromatics like acridine, one could create systems with much lower excitation energies than
quinoline while still reaching decently high pKa* values. For example, the authors identified
several disubstituted acridine and 1-azaanthracene molecules with pKa* values higher than 14
than could be excited with visible photons! Although these observations are not yet
experimentally confirmed, this is the first evidence that multiple substituents may be used to
further increase the photobasic capability of a molecule. These are exciting results, because
they may open the door for the synthesis of more useful photobase molecules in the future.
Finally, I will mention an exciting new photobase molecule that is pushing the
boundaries of the field. In 2018, a paper was published that described a Schiff’s base that acted
as a photobase
35
. This molecule, FR0-SB, is shown below with its photobasic atom highlighted.
Although I fundamentally disagree with the authors’ invention and definition of the term “super
photobase”, the work is interesting for a few reasons. Although the excitation wavelength of
this molecule is about 372 nm – and, because of the broadness of the absorption spectrum, it
can be excited in the near-visible – it is powerful enough upon photoexcitation to capture a
proton from ethanol and isopropanol! This makes FR0-SB the strongest photobase molecule
ever observed in the literature, to the best of my knowledge. Clearly N-heteroatomic aromatics,
like MeOQ, are not the only photobases out there. They might not even be the best!
89

Figure 6.1: The structure of FR0-SB, a very strong photobase that uses a Schiff’s base motif. The
photobasic nitrogen is indicated with a red shape.

A few follow-up works have shown that FR0-SB can deprotonate a wide series of
alcohols in a systematic way
76
; have claimed that the ESPT process is somewhat dictated by the
sterics of those alcohol molecules
77
; and have claimed that the ESPT reaction can be enhanced
via two-photon excitation of the FR0-SB molecule
78
. Although I am somewhat less enthused
about these follow-up works, they demonstrate that the photobase field is still alive and
thriving. If you’re interested in contributing to this living and thriving field, peek through the
Future Research Questions below for inspiration!
6.2 Future Research Directions
6.2.1 Applications where Proton Donors are Dilute
In chapter 2, I discussed at length the necessary solvation requirements for the ESPT
reaction in MeOQ. Namely, there must be a sufficiently high proton donor concentration such
that ground state preassociation via hydrogen bonding is achieved and the solvation of the
90
ESPT products is possible – namely through a second hydrogen bond to the photogenerated
alkoxide. The large excess of proton donors required for this reaction affects the possible
application of these molecules in a significant way. Imagine that the molecule you wish to
deprotonate in a catalytic reaction or synthetic reaction is necessarily quite low in
concentration. Arbitrarily increasing the concentration of the proton donor might have
unintended consequences or might prevent the reaction altogether. To avoid such
complications, one must figure out ways to deprotonate proton donors, even when they are in
low concentrations!
We will first discuss the problem of preassociation in the ground state. As discussed
before, the photobase and the proton donor will hopefully already be interacting in the ground
state. If they aren’t, the photobase might not have enough time in its excited state to locate the
proton donor molecule, and there will be no ESPT reaction. There are several strategies for
overcoming the need for a great excess of proton donor to achieve such interaction. One
strategy is to covalently link the photobase to a reaction center, much like was done in chapter
3. In this case, the proton donor of interest will be directly in front of the photobase molecule
and will readily hydrogen bond. One could also imagine tethering the molecule to the surface of
an electrode, where it can participate by interacting with the intermediates or products being
generated there. This concept is discussed in greater detail below. These methods are
particularly exciting because, in addition to making the association likely at the proton donor
concentrations of interest, they place the photobase molecule exactly where they are needed
for application to interesting chemistry.
91
Another, perhaps more mundane, way to potentially overcome this issue is by
increasing the enthalpic drive for preassociation such that it is more favorable at lower
concentrations of proton donor. This would allow the photobase reaction to occur at
reasonable proton donor concentration in mixed phase without special synthetic requirements.
Finding proton donors that are stronger hydrogen bonders is an option, but one would like the
solution to this problem to be more general. Perhaps a better option is to find photobase
molecules with higher ground state pKa values and thus greater drives to hydrogen bond.
Luckily, our work has shown that electron-donating substituents, which should increase the pKa
of the photobase in the ground state, also increase the pKa* of the molecule.
Now, we can discuss strategies for overcoming our observed necessity of two proton
donor molecules for successful ESPT. The reason for this requirement is the stabilization of the
ESPT products. Without a second proton donor to directly stabilize the resulting alkoxide, the
products cannot be stabilized enough to allow the reaction to proceed. The best way to think
about this is from a Marcus Theory perspective: hydrogen bonding by a second proton donor to
the photogenerated alkoxide greatly reduces the energy of the ESPT products. This effectively
makes the thermodynamic drive larger and correspondingly makes the activation energy barrier
small enough to overcome. This is demonstrated in figure 33 below.
92

Figure 6.2: Presence of a second proton-donor stabilizes the products of the 5-
methoxyquinoline ESPT reaction and therefore reduces the activation energy barrier.

However, there may be other ways to increase the thermodynamic drive of the reaction
that accomplish something similar. If the difference in pKa between the excited photobase and
the proton donor is sufficiently large, there may be enough drive to encourage proton transfer
even in the absence of a second proton donor. One could perhaps show a proof of this concept
by using a very low pKa proton donor in combination with an already well-characterized
photobase, like 5-MeOQ. The more exciting idea, though, is to find a photobase with a high
enough pKa* that it can readily overcome these energy barriers, even for common proton
donors. We have done some work on optimizing the excited state pKa of N-heteroatomics (see
Energy
Nuclear Coordinate
!
!
!
!
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
N
O
O
F
F
F
H
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl
H
H
Cl
Cl N
O
O
F
F
F
H
O
F
F F
H
-
+
93
chapter 4), as have other groups
75
. There has also been the discovery of high pKa* photobases
that are non-N-heteroatomics, like the Schiff base discovered by Sheng et. al.
35
There is still
work to be done on this front.
Instead of a second proton donor, one could stabilize the ESPT products via the use of
high dielectric solvents, like propylene carbonate or DMSO. It’s currently unclear the extent to
which a high dielectric solvent could mediate an ESPT reaction – especially compared to a
specific hydrogen bonding interaction - and this strategy would likely be less applicable for less
thermodynamically favorable ESPT reactions. Another problem with this idea is that many high
dielectric solvents, including the ones just listed, are hydrogen bond acceptors or donors and
may interfere with the interaction between the photobase and the proton donor. A similar, but
perhaps more versatile, option is to introduce salts into the solution of interest. At high
concentrations, the salts could produce a screening effect like a high dielectric solvent’s and
could therefore help reduce the reorganization energy of the system. Salts typically have no
specific interactions, so one would not have to worry about interference with the hydrogen
bonding between the photobase and the proton donor like one might with solvents like
propylene carbonate or DMSO.
It is likely that the best solution to the excited state solvation problem will be a
combination of all these parameters: a reaction with a high thermodynamic drive in a high
dielectric solvent with some salt added for additional screening. One can imagine an exciting
study where all three concepts are studied systematically: the hydrogen bond donor or
photobase is varied to modify the thermodynamic drive of the reaction; the solvent is varied as
a function of dielectric constant; and the identity and concentration of the salt is varied as well.
94
This type of study could provide an incredible wealth of information about the fundamentals of
the ESPT transfer in photobase molecules and about the application of these molecules to
dilute applications.

Figure 6.3: Can a high dielectric solvent stabilize the products of the 5-methoxyquinoline ESPT
reaction enough to significantly reduce the activation energy barrier?

6.2.2 Deprotonation of C-H Bonds
All works in this manuscript discuss the deprotonation of water or alcohols – that is, O-H
bonds. While this is the most obvious form of deprotonation, it is not the only kind. One
potential application of photobases is the deprotonation of C-H bonds to produce carbanions.
To the best of my knowledge, this type of experiment has never been performed. C-H bond
deprotonation is far less common than O-H deprotonation but far more useful synthetically.
Energy
Nuclear Coordinate
!
!
!
!
N
O
O
F
F
F
H
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
N
O
F
F
F
H
O
"=$
"=%&
-
+
95
The facile generation of a reactive carbon is a synthetic chemist’s dream. If one were able to
show a proof of concept for the deprotonation of a C-H bond using a photobase, this would be
a high impact work.

Figure 6.4: Some molecules with acidic C-H bonds. The acidic C-H is denoted with a red shape.
I have personally worked on such a project with no avail, and I can provide some insight
into why this concept has not yet been published. Most C-H bonds have very low acidity, so in
general C-H bonds are not thermodynamically deprotonatable by today’s well-characterized
photobases. For comparison, 5-MeOQ has an excited state pKa of around 15.5, while an alkane
usually has a pKa of around 50. The alkane is more than 10
30
times less acidic than 5-MeOQ!
There are, however, more acidic C-H bonds - you just have to know where to look! For example,
the alpha carbon of a ketone tends to have a pKa around 20. The presence of multiple electron
withdrawing groups can even lower the pKa such that it’s theoretically deprotonatable by 5-
MeOQ! A couple of examples are shown in figure 6.4
79,80
.
N
N
O O
Malononitrile Acetylacetone
pKa = 11.1 pKa = 9.0
96
Even when I’ve studied C-H bonds with appropriate thermodynamic drives, though, I
was unable to see successful proton transfer. I have a few theories for why that is. In chapter 3,
I talked extensively about and showed evidence for the necessary requirements for a successful
ESPT reaction in MeOQ. Namely, an excess of proton donor is necessary for ground state
hydrogen bonding, which is important to achieve so that we need to worry less about
diffusional kinetics in the excited state. When one learns about hydrogen bonding as a high
school or undergraduate student, the possibility of a C-H bond acting as a hydrogen bond donor
is never introduced. It is not a common idea, and that is probably because C-H bonds do not
have the proper orbital arrangement or the proper enthalpic drive to form hydrogen bonding
complexes.
If we cannot form a hydrogen bond with an acidic C-H bond, what are the hopes for
deprotonating it? This question raises another question: if we can’t even hydrogen bond in the
ground state, how will we stabilize the ESPT products in a way analogous to that shown in
chapter 2, where an additional hydrogen bond directly interacts with and stabilizes the
deprotonated species? Work is required to answer these questions. Doing so will likely not be
straightforward. But the rewards are vast: the deprotonation of a C-H bond with a photobase
would open a lot of synthetic and chemical doors. I will list my suggestions for overcoming
these problems.
If one could tether the photobase directly in front of an acidic C-H bond, perhaps
covalently, then perhaps “hydrogen bonding” in the traditional sense would be unnecessary.
There is no way to guarantee that this would be enough to convince the photobase and the C-H
bond to interact – but it’s an exciting idea! Another possibility is to dissolve the photobase in a
97
solvent that itself contains the acidic C-H bond. That way, there will always be a C-H proton
donor molecule directly in front of the photobase. All acidic C-H molecules I have studied have
been solid and have thus required a co-solvent. Maybe the right solvent just hasn’t been found.
Maybe one could melt the solid and then use it as the solvent!
The issue of solvation of the ESPT products remains. I have already discussed the
potential ways to get around the necessity of specific interactions from a second hydrogen
bond donor. These include things like the use of a high dielectric solvent to increase the
screening of the ESPT ion products, or the use of salts in the solvent to accomplish the same
thing. It is also possible that the importance of the solvation could be overcome if the drive for
electron transfer became sufficiently large. A combination of these three ideas is likely to be
the most fruitful. Of course, these need to be applied in tandem with the methods to achieve
ground-state preassociation. One just needs to find the combination of compatible ideas that
works!
6.2.3 Excited State Lewis Acid Transfer
It is clear from our work and many others than photobases and photoacids are robust
for ESPT reactions. The transfer of protons places the demonstrated utility of these molecules
firmly within the Bronsted-Lowry definition of basicity and acidity. This raises an important
question: will these molecules also show activity as electron transfer reagents in Lewis acid
reactions? The Lewis acid formalism of acids and bases is more general and does not necessarily
involve the transfer of protons. Instead, it is viewed through the lens of electron donating
(Lewis bases) and accepting (Lewis acids). A proton is an electron acceptor – so in addition to
98
being a Bronsted acid, it is also Lewis acid. A hydroxide molecule is an electron donor and a
proton acceptor – so it is both a Bronsted acid and a Lewis acid. Imagine an electron deficient
molecule like BF3, though: it wants an electron, so it is a Lewis acid, but it is not a proton, so it
is not a Bronsted acid.

Figure 6.5: Some interesting Lewis acids and Lewis bases.
We know that MeOQ and other photobases have an increase in Bronsted basicity upon
photoexcitation. We also know that they act as Lewis bases in the ground state. What is unclear
is whether their Lewis basicity will be enhanced in the excited state. There is some reason to
believe that it might. Remember from chapter 5 the discussion of the mechanism of
photobasicity in quinoline molecules that was uncovered from electronic structure theory
calculations. Upon photoexcitation, there is an increase in the electron density on the nitrogen
heteroatom. This results in attracting protons more readily. Shouldn’t this increase in electron
density also attract electron-deficient Lewis acids?
Some Interesting Lewis Acids
Some Interesting Lewis Bases
N
O
. . N
. .
N
. .
N
H H
H
. .
B F
F
F
Al F
F
F
!"
!"
#
"
99
To my knowledge, no example of Lewis photobasicity in this sense has been shown in
the literature. One of my colleagues, Matt Voegtle, is working on this idea at the time of writing
this dissertation. If it is shown that Bronsted photobases also act as Lewis photobases, the
application space is multiplied significantly.
There are some interesting questions here, beyond the use of photobases to attract
Lewis acids. For instance, is there something resembling a hydrogen bond in Lewis acid-base
reactions? Is there some mechanism of ground state preassociation? One can imagine that, if a
Lewis acid is already in a stable complex with a Lewis base, it might be able to form a weak
interaction with a separate Lewis base. This would be analogous to protons already in a stable
interaction with a Bronsted base forming hydrogen bonds with other Bronsted bases. Such an
interaction is likely necessary to establish an appropriate geometry for Lewis acid transfer,
much like how ground state hydrogen bonding is necessary for proper Bronsted photobasicity.
One should also consider the Lewis acid version of specific solvation of the photobase
products. Interestingly, the product of an excited state Lewis acid transfer should not be
charged. Instead, it will be a new Lewis acid-base complex. Although there might be partial
charge transfer and a resultingly high dipole moment for this photogenerated product, there
will not be an ion pair that requires solvation. These photogenerated products should be much
easier to solvate than their Bronsted counterparts! In this respect, the Lewis photobase could
be an easier concept to apply to a variety of chemical systems since less stringent solvation
requirements may be necessary.
100

Figure 6.6: Can 5-methoxyquinoline act as a Lewis photobase?

6.2.4 Putting Quinoline’s Triplet States to Good Use
The work here and in all my publications discusses ESPT in the singlet manifold of the
excited state photobase molecule. However, it is well-known that many N-heteroatomics
(including unsubstituted quinoline) undergo rapid (sub ps) ISC
68–71
. Usually, it is assumed that
following ISC the molecules are no longer useful as ESPT agents. There are a few reasons for
this. The first reason is practical: with the molecule now in the triplet manifold, it becomes
much less straightforward to study the ESPT reaction. One cannot observe fluorescence, so one
cannot use it to figure out whether the ESPT reaction was successful, as has been done in most
of my works. One may not use TCSPC for similar reasons. One could possibly study these
systems using phosphorescence studies, but typically one does not see phosphorescence under
ambient conditions.
There is another reason why many assume ISC is the death of the ESPT reaction. It is a
well-known fact that triplet states are always lower in energy than their corresponding singlet
!" N
O
B
N
F
F F
N
O
B
F
F
F
N
N
O
B
F
F
F
N
Pre-association?
101
states. Because of this, the triplet manifold exists over a smaller energy range than the singlet
manifold. Therefore, there is likely to be much less drive for proton transfer in the triplet
manifold than in the singlet manifold.
However, if a photobase molecule could still undergo ESPT following ISC, there are some
advantages to this process as opposed to singlet ESPT. Triplet states are typically much longer
lived than singlet states. We have talked extensively about the necessity for pre-association
with the proton donor in the ground state due to the relatively short lifetime of the excited
singlet state of the photobase molecule. What if this were no longer a concern? If it were a
triplet state that was initiating ESPT, pre-association would be far less necessary since the
molecule would have a significantly longer excited state lifetime during which to find the
proton donor.
Additionally, the separation of the proton from its donor would occur over a longer
amount of time. Once the photobase has returned to the ground state, there is no longer a
drive for the photobase molecule to hold the proton it captured in the excited state. Reverse
proton transfer will certainly occur following deexcitation of the photobase. If the excited state
lifetime is longer, as it would be following ESPT in the triplet manifold, then the
photogenerated products can exist for a longer amount of time, thus giving the products a
greater probability to do interesting chemistry.

102

Figure 6.7: Approximate singlet (blue) and triplet (red) state energies for the base and acid
forms of quinoline showing the feasibility of ESPT in the triplet manifold.

There has been some preliminary work done by Eric Driscoll and me into ESPT in the
triplet manifold of quinoline molecules
28
. Eric saw what he believed to be proton transfer
following ISC in both unsubstituted and cyano-substituted quinoline. He did some work
quantifying the electronic structure of the triplet manifold of unsubstituted quinoline to explain
the rapid ISC and subsequent ESPT. His proposed energy diagram for unsubstituted quinoline is
shown in figure 6.7. However, this work is inconclusive and far from telling a complete story.
For instance, Eric’s triplet state energy diagram is generated using many assumptions and
approximations, and the pKa* of the proton transfer in the triplet manifolds has never been
estimated. This would usually be done with the Förster cycle, like Eric Driscoll did in his original
paper on quinoline photobases. However, this process is much more complicated for triplet
103
states than it is for singlet states. For one, estimation of the 0-0 gap for triplet states is
considerably more conceptually challenging than it is for singlet states. In appendix A, I go into
detail about the Förster cycle, why the 0-0 gap is needed, and how it is typically estimated. One
usually needs both absorption and emission spectra to estimate the gap. Typically, one cannot
directly excite to the triplet state, so the estimation of the 0-0 gap falls entirely onto the
phosphorescence. Estimation of the 0-0 gap using solely phosphorescence could be robust
depending on the particular system and the degree of nuclear reorganization in the excited
state. If it’s not robust, though, it will have a significant impact on the estimated pKa* value.
One must also consider the difficulties associated with obtaining phosphorescence
spectra. Because it takes triplet states a while to emit (often 𝜇s – ms), their emission is typically
out-competed by non-radiative channels under ambient conditions. To see the
phosphorescence, then, one must design their system to avoid non-radiative decay channels.
Usually this is accomplished by freezing the solvent (such as in an ethanol glass) or by
incorporating the molecules of interest into a polymer matrix. These strategies are much more
complicated than the solution-phase spectroscopy required of fluorescence! If one could
successfully obtain phosphorescence data for a series of photobase molecules and use this to
estimate the triplet pKa*, this could be an exciting and potentially high impact research project.
It would also be possible to study the proton transfer reaction directly using time-
resolved techniques like electronic or vibrational transient absorption spectroscopy. There is
already some preliminary data in a work by Eric Driscoll and me suggesting that triplet manifold
proton transfer is possible. I am personally skeptical of these results and their interpretation. I
think additional work is necessary, such as verification that the states of interest are triplet
104
states and that the dynamics of interest are truly proton transfer related. Such work should be
relatively straightforward: do these excited states live as long as one expects a triplet state to
live? Can we see their phosphorescence somehow? Are they quenched by oxygen or other
known triplet quenchers? Are the dynamics affected by the kinetic isotope effect? That is, do
the observed kinetics slow down when deuterated water is used instead of regular water?
Careful, convincing proof that proton transfer can occur in the triplet manifold would certainly
be publishable.
At conferences, when I talk about photobase molecules, catalytic scientists always want
to know: “how long does the excited state of the photobase live? How long do I have to do
catalysis?” They are always disappointed by the answer (ns), because they would like to have
enough time to do interesting chemistry during the excited state of the photobase. The use of
the triplet manifold could provide that extra time.
6.2.5 Electrode-Tethered Photobases
Eric’s first work on quinoline photobases demonstrated the power of the substituent
Hammett parameter to modify the photobasicity of the molecule. He showed that electron
withdrawing substituents tended to decrease both the ground and excited state basicity, while
electron-donating substituents tended to increase both. In chapter 5, we used electronic
structure calculations to study how changing the Hammet parameter could be used to find an
“optimized” photobase.
This strategy is limited by the availability of substituents and additional complications
that may be caused by them. In addition to the synthetic reality of substituting photobase
105
molecules, which can be non-straightforward, the fine-tuning of the substituent Hammett
parameter is limited by the fact that there are only so many reasonable substituents to choose
from. Perhaps more importantly, the largest electron-donating capabilities available to us are
limited by the most electron-donating substituents we can reasonably use as substituents for
our photobase molecules. Throughout this work, I have studied 5-MeOQ. This is in large part
because the methoxy substituent is the best electron-donating group we can use without
significantly changing the properties of the molecule. Take the more electron-donating
substituent amino substituent for example. It should donate more electrons to the quinoline
system and increase electron density on the heteroatomic N, thus making the molecule more
basic. But the presence of an amine group, which can donate hydrogen bonds to the
surroundings, altogether changes the way this molecule operates. Specifically, it is well-known
that an amine group attached to an aromatic system can behave as a photoacid. Clearly it is not
only the Hammett parameter we must be worried about – we also must be worried about the
“inertness” of the substituent if we don’t wish to perturb our system too much.
There is a recent work by Sohini Sarkar in my group that investigated electrode-tethered
molecules and how they respond to applied potential at the electrode
81
. Sohini showed
convincingly that the response of the molecule to the potential can be correlated to a similar
response due to varying the substituent’s Hammett parameter. Basically, she showed that
pushing or pulling electrons into or out of the molecule attached at the electrode has the same
effect as pushing or pulling electrons using a traditional substituent. She provided a guide,
showing what strength of Hammett parameter corresponds to what applied electric field at the
surface of the electrode. This is reproduced below in figure 6.8 with the permission of Sohini.
106

Figure 6.8: Relationship between applied electric field at an interface and the Hammett
parameter of the substituent that would have approximately the same effects. Recreated from
Sarkar et. al.
81
with permission.

Taking these two ideas – the modification of photobasicity using substituents, and the
ability to mimic substituents using applied potential for electrode-tethered molecules – and
combining them suggests a very exciting idea: what if we could tether the photobase molecule
to the surface of an electrode and could then finely tune the photobasic behavior with applied
potential? Figure 6.9 below provides a visual explanation for this concept, where quinoline
molecules are tethered to a gold surface via gold-thiol bonds. This idea is tantalizing for several
reasons. First, this could theoretically allow one to tune the strength of the photobase to any
value allowed by the voltage range of the electrode. This control would be very fine, such that if
a specific pKa* was required for a given application, that would be easily and readily
Hammett Electron
Withdrawing
Parameter
Electric Field
𝐹 𝐴 𝜎 𝐵 𝜎 𝐶 𝝈 𝒑 Field
Hammett Parameter
107
accomplished. Second, if a suitable system was found, this could allow for the application of
larger “electron-donating” capabilities than is possible with traditional substituents. Want to
increase the photobasicity of your molecule? Just apply a bigger negative potential and don’t
worry about complications due to synthesis of new molecules or the impact of the substituent
on the basic functioning of your molecule!
It is necessary to mention an important caveat to this last point. One cannot arbitrarily
apply large negative potentials and hope to only affect the polarization of your surface-tethered
molecule. In an electrochemical experiment at high potentials, there are other things to worry
about: Will side reactions begin to happen? Instead of simply polarizing the molecule of
interest, will you accidentally reduce it? Will these potentials cause the covalent linkage
between the electrode and the tethered molecule to break down? With careful consideration
regarding solvent environment, electrode identity, and tethering mechanism, one can access
higher potentials while avoiding these pitfalls, but the system must be designed carefully and
there will always be an upper limit to the potential that can be applied without issue.
There is another reason why tethering photobase molecules at the surface is so exciting:
surface reactions. A lot of interesting and important chemistry occurs at surfaces, particularly at
the surface of electrodes. Consider the electrochemical hydrogen evolution reaction, one
possible source of hydrogen to fill the fuel cells of tomorrow. Consider solid-state photovoltaic
devices, which harvest energy from the sun in a clean way. Consider the multitude of industrial
electrochemical syntheses that generate the products we use every day. One can imagine many
scenarios where having fine control over proton concentration and transfer at these surfaces
can lead to fine control over the reaction itself. Having an electrode-tethered photobase is
108
having a tunable and externally controllable chemical tool right at the reaction center! It could
be a game changer.

Figure 6.9: Concept for a surface-tethered quinoline experiment with applied potentials.
Positive potentials act like electron-withdrawing substituents, while negative potentials act like
electron-donating substituents.
Tethering molecules to electrode surfaces and applying potentials are not my specialty,
but they are well-studied in the Dawlaty group and are therefore minor experimental hurdles.
N
S
N
S
N
S
V
Counter
electrode
Working
electrode
N
S
N
S
N
S
- - - - - - - - -
Electron
Donating
N
S
N
S
N
S
Electron
Withdrawing
+ + + + + + + +
109
However, once you’ve attached the photobase molecule, the study and characterization of your
surface monolayer will present unique challenges. For instance: how does one estimate the
excited state pKa of this system? As mentioned before, this would typically be accomplished
using a Förster cycle analysis, requiring knowledge of the ground state pKa and the 0-0 energy
gaps between S0 and S1 for both the unprotonated and protonated forms of the photobase.
Using this knowledge, the pKa* can be estimated. The ground state pKa at the surface can be
obtained experimentally by changing the pH of the solution at the interface and using a surface-
specific vibrational spectroscopy (such as ATR or SERS) to monitor the N-H stretch that indicates
protonation of the molecule. However, how does one obtain the 0-0 energy gap for a surface-
tethered monolayer? The concentration is too low for a traditional absorption measurement,
and emission experiments will be difficult for a similar reason. One may be able to find a way
around these difficulties, perhaps by careful long-time integration of emission data in a special
geometry for surfaces.
There are other methods for characterizing these surface-tethered monolayers and the
impact of the applied potential on their photobasicity. If one already has a well-defined
relationship between substituent Hammett parameter and excited state basicity in the complex
of interest, one can assume that that they know the pKa* based on the relationship between
applied potential and Hammett parameter discovered by Sohini, as discussed before. This is
one of the marvels of a surface-tethered photobase: one can tune to a new potential with the
touch of a button and know, based on this correlation, the behavior of the molecule! This
assumes that the correlation does not change as a function of system, which may be an
imperfect assumption when one intends to finely tune the electronics of the monolayer.
110
Furthermore, if one wishes to tune the system beyond the known Hammett parameter values –
to make a stronger photobase than has been seen using traditional substituents by applying a
large negative potential, for example, there is additional uncertainty. Therefore, even though
we have a known correlation, it is still probably best to directly characterize the surface-
tethered system, if only to confirm the accuracy of Sohini’s correlation for the system of
interest.
It is possible to use a time-resolved surface-specific technique to directly study the
deprotonation of proton donors by the surface-tethered photobase. The technique would
require sub-nanosecond time resolution to see the proton transfer process. One candidate
technique for such an experiment is time-resolved sum frequency generation (TR-SFG). This
technique would allow for ps time resolution of vibrations and is highly surface-specific, so the
low concentration of molecules would not be an issue. One could watch the rise of the N-H
stretch vibration as a direct sign of proton transfer. This is a complicated experiment that is not
for the faint of heart! However, the Dawlaty lab is equipped for such an experiment if the right
student were to come along.

111
Appendix A
Förster Cycle Analysis
How does one experimentally obtain the pKa* of a photobase molecule – or any other
molecule, for that matter? In the ground state, one adds a titrant (usually either a strong acid or
strong base) with a well-known concentration in well-defined quantities to generate a titration
curve. The pKa will correspond to the pH at the midpoint of the titration where there is an
equal concentration of acid and conjugate base. This is clear from the Henderson-Hasselbach
equation:
𝑝𝐻 = 𝑝𝐾
!
+log
[𝐴
'
]
[𝐻𝐴]

When [𝐴
'
] = [𝐻𝐴], then we have log1 = 0 and are left with 𝑝𝐻 = 𝑝𝐾
!
.
We cannot use this strategy to study the excited state for a few reasons. First, the
concentration of excited molecules is typically far too low to affect the pH of the solution. This
makes it impossible to directly construct a titration curve for the excited state acid-base
reaction. Second, the excited state lifetime of many organic chromophores is too short for
proper acid-base equilibrium to be established throughout the solution. One may be tempted
to generate a fluorescence titration curve based on whether the emission from the molecule of
interest is from the unprotonated or protonated forms as a function of solution pH. However,
interpretation of such results can be very complicated, since one must worry about the
competitive kinetics of ESPT and excited state decay. This is especially an issue at pH values well
112
below the equivalence point, where reaching excited state equilibrium will likely be slower than
or competitive with deactivation of the excited state.
Instead of measuring the excited state pKa in a way that’s analogous to the
determination of the ground state pKa, an indirect method called the Förster Cycle is often
used
8
. This method uses a thermodynamic cycle in conjunction with steady state spectroscopic
measurements (UV/Vis absorption and fluorescence). Figure A.1 shows the basic idea behind
the Förster cycle: that the excited state drive for protonation or deprotonation – and, by
extension, the pKa* – can be estimated if one knows (1) the ground state enthalpic drive for
protonation, (2) the 0-0 vibronic gap between the S0 and S1 states for the unprotonated form,
and (3) the 0-0 vibronic gap between the S0 and S1 states for the protonated form.

Figure A.1: Energy level diagram for the Förster cycle analysis of a photobase molecule.

!"
!
( $−$)
!"
!"
!( $−$)
'(
#
'(
#∗
!"
!∗
!"
!
!
!
∗
113
In equation form, we can represent this cycle as
𝑁
!
ℎ𝑣
B
+Δ𝐻
:∗
= 𝑁
!
ℎ𝑣
BE
& +Δ𝐻
:

If the thermodynamic cycle above is true, then the equation follows. Remember that, in
the above equation, Δ𝐻
:∗
will be negative because there is a drive for that reaction. Realizing
Δ𝐻
:∗
is negative makes the above equation much easier to grasp when comparing it with the
figure. Note that I have generated the above figure with MeOQ in mind: at neutral pH, it tends
to be unprotonated, and in the excited state it has a large drive for protonation. We can
reorganize the equation in the following way:
Δ𝐻
:∗
− Δ𝐻
:
= 𝑁
!
(ℎ𝑣
BE
& − ℎ𝑣
B
)
It would be much more convenient if the quantities on the left-hand side of the above
equation were Gibbs free energy values (Δ𝐺
:
), since this is the quantity we can extract from the
ground state pKa:
Δ𝐺
:
= 2.3 𝑅𝑇 𝑝𝐾
!

and is the quantity we need to calculate the excited state pKa by the same logic. However, the
Förster thermodynamic cycle explicitly involves enthalpy only, since excitation energies (ℎ𝑣
B
)
do not contain contributions from entropy in any meaningful way. However, if we assume that
the ground state and excited state proton transfer reactions have the same change in entropy,
we can make the following mathematical argument. Since Gibbs energy is defined as
∆𝐺 = ∆𝐻−𝑇∆𝑆
if the two reactions have the same entropy, we can say
114
Δ𝐺
:∗
− Δ𝐺
:
= Δ𝐻
:∗
−Δ𝐻
:
−(𝑇Δ𝑆
:∗
−𝑇Δ𝑆
:
) = Δ𝐻
:∗
−Δ𝐻
:

If the difference in Gibbs free energy between the excited state and ground state is
equivalent to the difference in enthalpy between the excited state and ground state, we can
therefore write
Δ𝐺
:∗
− Δ𝐺
:
= 𝑁
!
(ℎ𝑣
BE
& − ℎ𝑣
B
)
which gives us the Förster cycle in terms of more useful quantities.
It’s important to justify why the change in entropy for these two reactions should be the
same. The entropy change for the system itself will be similar because, in both ground and
excited states, we start and end with the same number of species. That is, we start with a
photobase and a proton donor, and we end with a protonated photobase and an alkoxide. The
largest contribution to entropy of the proton transfer reaction, then, will be the necessary
change in the solvent structure following the generation of charged products. The stabilization
of charged species requires the nearby solvent molecules to become much more structured,
reorienting in a specific way such that their dipoles cancel out the resulting electric field. In
both the ground and excited states, there is a generation of charged species, so the
restructuring of the solvent will be similar in both cases. Therefore, assuming that the entropy
change in both reactions is the same is mostly justified.
We can combine two of the above equations to give us
𝑝𝐾
!
∗
− 𝑝𝐾
!
= 2.1 𝑥 10
'=
(𝑣
BE
& − 𝑣
B
)
115
where the frequencies are now given in wavenumbers. It is also convenient to formulate this
equation in terms of nm, since these are the typical units for absorption and emission
experiments:
𝑝𝐾
!
∗
− 𝑝𝐾
!
= 2.1 𝑥 10
8
(
1
𝜆
::,BE
&
−
1
𝜆
::,B
)
Now, in principle, if one knows the ground state pKa and the 0-0 vibronic energy gaps
between the S0 and S1 states of the protonated and unprotonated forms of the molecule, they
can calculate the excited state pKa.
It is important to know the 0-0 energy gaps for the transition of interest because,
according to Kasha’s rule, there will be rapid relaxation to the ground vibrational level of the
excited state. The ground vibrational level of the excited state is therefore the most relevant
since it is where the acid-base reaction will mostly occur. Most of the acid-base equilibrium in
the ground electronic state will similarly involve molecules in their ground vibrational state. If
both reactions are likely to occur in the 0
th
vibrational energy level, it follows that we should
know the 0-0 gap. It is, however, often nontrivial to find the 0-0 vibronic energy gap for the
transitions of interest.
If the vibronic structure of the absorption and emission for the molecule are well-
defined, it may be possible to directly estimate the 0-0 gap from the spectra. However, this is
often not the case. For example, 5-MeOQ has very broad and unstructured absorption and
emission bands. Due to the nature of nuclear rearrangement in the excited state, the
absorption maximum will generally be an overestimation of the 0-0 energy gap, while the
116
emission maximum will generally be an underestimation. This is demonstrated below in figure
A.2. The 0-0 energy gap is therefore often estimated as an average of these two maxima:
𝑣
::
=
𝑣
!F2
"!#
+𝑣
@"
"!#
2

Another common method is normalizing the absorption and emission spectra and
finding the intersection point. This was the method used in Förster’s laboratory. When the
absorption and emission spectra have the exact same linewidth, this should give the same
result as the averaging method mentioned above. But when the absorption and emission
spectra have distinct linewidths, this method will give a different value. Those distinct
linewidths are often related to unique curvatures of the ground and excited state potential
energy surfaces, and the intersection method can help account for this in the estimation of the
0-0 energy gap.
It is important to note that the estimation of the 0-0 gap is usually the biggest source of
inaccuracy in the calculation of pKa* using the Förster cycle analysis. An error of about 4 nm
from the true 0-0 gap when absorption and emission are around 400 nm can result in a 1 unit
deviation in the pKa*!
117
z
Figure A.2: Potential energy surfaces and nuclear wavefunctions of ground and excited states,
demonstrating the overestimation and underestimation of the 0-0 vibronic energy gap by
absorption and emission spectroscopy, respectively.

!
!"!
Nuclear Coordinate
Energy
Absorption
Emission
118
Appendix B
Transient Absorption in the Dawlaty Lab
Below, I will discuss the Dawlaty Lab transient absorption apparatus in detail. I will
include common issues I have experienced and how I dealt with them. There will be several
subsections for the convenience of the reader.
B.1 What the Heck is Balanced Detection Pump Probe?
In our lab, we used balanced detection pump probe. To understand this technique, one
must understand “normal” pump probe first. In the “normal” experiment, you use a high
intensity pulse of light that is resonant with your chromophore – called the pump – to create an
excited population. Then, you use a broadband (“white light”) pulse – called the probe – at
various delays with respect to the pump to monitor the time-dependent activity of the sample.
The delay is typically introduced via a delay stage, discussed in detail in the subsection called
“The New Delay Stage and the Importance of Pump Alignment”.

Figure B.1: A pulse diagram for a “normal” transient absorption experiment.
119
In this experiment, you must collect and keep track of two different types of probe pulse
intensities: 𝐼
%G"% 3/
, which goes through the sample with the pump and therefore contains
transient signal in addition to steady state signal, and 𝐼
%G"% 3??
, which does not go through
with the pump and therefore only contains steady state signal. The difference between these
two pulses gives just the transient signal, which is what we are interested in.
In a typical absorption experiment, we calculate the absorption using the following
equation:
𝐴 = −log(𝑇)= −logU
𝐼
)H!/2">))@I
𝐼
>/C>I@/)
V
where T is the transmittance. I mentioned above that we measure the transmitted light when
the pump is on (𝐼
%G"% 3/
) and the transmitted light when the pump is off (𝐼
%G"% 3??
). However,
we never measure 𝐼
>/C>I@/)
. How, then, can we calculate the absorption? We use the following
mathematical trick:
∆A = 𝐴
%G"% 3/
−𝐴
%G"% 3??
= −logU
𝐼
%G"% 3/
𝐼
>/C>I@/)
V−g−logU
𝐼
%G"% 3??
𝐼
>/C>I@/)
Vh
∆𝐴 = −logU
𝐼
%G"% 3/
𝐼
>/C>I@/)
V−ilogj
𝐼
>/C>I@/)
𝐼
%G"% 3??
kl
∆𝐴 = −logj
𝐼
%G"% 3/
𝐼
>/C>I@/)
𝐼
>/C>I@/)
𝐼
%G"% 3??
k
If we assume that the I incident is the same for both pump on and pump off pulses, we can cancel
them out in the above equation and are left with:
120
∆𝐴 = −logj
𝐼
)H!/2,%G"% 3/
𝐼
)H!/2,%G"% 3??
k
where we have calculated the change in absorption using only the two measured quantities.
However, in order for our assumption (the consistency of I incident ) to be true, we must have very
stable white light probe spectra that are the same from pulse to pulse. This “normal” way of
doing pump probe is therefore extremely sensitive to white light stability.
Small fluctuations in white light intensity can lead to quite big “signal”. An example is
shown below. On the left in figure B.2, the white light spectra look identical. The difference isn’t
noticed until we zoom in, shown on the right.

Figure B.2: Two slightly different white light spectra.
What will the “signal” arising from these small white light fluctuations look like? It turns
out it’s quite large, as shown in figure B.3. This false signal is shown on the left, calculated using
the equation ∆𝐴 = −logm
J
'()
J
'($
n. On the right, I’ve shown this false signal added to some “real
signal” of reasonable magnitude to show how large of an impact this white light fluctuation can
have on your data.
121

Figure B.3: False signal from poor white light subtraction and how it compares to a reasonably
intense transient signal.

To avoid these kinds of effects, we must either make sure our white light is very stable
or use balanced detection pump probe. We choose the section option. In balanced detection
pump probe, we split the probe beam in half using a beam splitter and then send both through
the sample to the detector. One of the probe beams, which we call the “sample” beam, is
coincident with the pump through the sample. Therefore, when the pump is on, the “sample”
pulse will contain the transient signal. The other probe beam, which we call the “reference”
beam, is not coincident with the pump through the sample. Even when the pump is on, this
pulse will contain no transient signal. The job of a reference pulse is to be a copy of the
respective sample pulse that keeps track of white light fluctuations from pulse to pulse. This
pulse arrangement and alignment is shown in figure B.4.
122

Figure B.4: A pulse diagram for balanced detection pump probe.
We must now keep track of four different types of pulse intensities: 𝐼
2!"%4@,%G"% 3/
,
𝐼
H@?@H@/C@,%G"% 3/
, 𝐼
2!"%4@,%G"% 3??
, and 𝐼
H@?@H@/C@,%G"% 3??
. Note that the time between probe
pulses (prior to splitting) and pump pulses has been doubled in the balanced detection
experiment. This is because our camera can only read out 1 pulse per millisecond. If the probe
pulses still came every millisecond and we split them, we would then have two pulses every
millisecond and the camera would not have enough time to read out both sample and
reference pulses. Therefore, we chop the repetition rate of both the probe and pump beams in
half to solve this readout problem. Note that, since the sample probe and its corresponding
reference probe should hit the camera at approximately the same time (within a few ns, since
the reference pathlength is slightly longer) and we need to read out both, we must direct the
beams to two separate parts of the camera that we read out separately. This is discussed in
greater detail in the subsection titled “Alignment of White Light into Camera”.
When we collect all four probe beams, we can correct for any white light fluctuations.
We first calculate the “corrected” absorption, shown below for a pump-on pulse pair:
123
∆𝐴
%G"% 3/ (C3HH@C)@I)
= 𝐴
%G"% 3/ (2!"%4@)
−𝐴
%G"% 3/ (H@?@H@/C@)

where 𝐴
%G"% 3/ (2!"%4@)
contains the transient signal and 𝐴
%G"% 3/ (H@?@H@/C@)
acts as a copy
that doesn’t contain the transient signal. We can rewrite the equation in the following way:
∆𝐴
%G"% 3/ (C3HH@C)@I)
= −𝑙𝑜𝑔j
𝐼
%G"% 3/, 2!"%4@
𝐼
>/C>I@/),%G"% 3/,2!"%4@
k−𝑙𝑜𝑔j
𝐼
>/C>I@/),%G"% 3/,H@?@H@/C@
𝐼
%G"% 3/, H@?@H@/C@
k
If the 𝐼
>/C>I@/)
for both pulses were identical, they would cancel out and
∆𝐴
%G"% 3/ (C3HH@C)@I)
would be equivalent to the transient absorption signal. It would be as if
you were using the [pump on, sample] pulse as the [pump on] pulse and the [pump on,
reference] pulse as the [pump off] pulse in a “normal” transient absorption experiment. This
makes conceptual sense, since the sample pulse contains both steady state and transient signal
since it transmits coincident with the pump, while the reference pulse only contains steady
state signal, so subtracting the two should result in only the transient signal if the incident
spectra are identical.
We can’t assume that the 𝑰
𝒊𝒏𝒄𝒊𝒅𝒆𝒏𝒕
in these two pulses is the same, though – the sample
and reference beams have different paths and encounter different numbers of optics, so it’s
unlikely that they will be identical. In practice, they are usually quite different. We do,
however, assume that the sample and reference are consistently different from one another
from pulse pair to pulse pair. This is a reasonable assumption, since the difference is dependent
on the beam path, which is unlikely to change much as a function of time. We represent this
wavelength-dependent difference as a(𝜆) and rewrite the above equation as:
124
𝐴
%G"% 3/, C3HH@C)@I
= −𝑙𝑜𝑔U
𝐼
%G"% 3/, 2!"%4@
𝐼
>/C>I@/)
V−𝑙𝑜𝑔j
𝑎∗𝐼
>/C>I@/)
𝐼
%G"% 3/, H@?@H@/C@
k
Now we can cancel out 𝐼
>/C>I@/)
, but we are left with this a thing:
𝐴
%G"% 3/, C3HH@C)@I
= − 𝑙𝑜𝑔j
𝐼
)H!/2,%G"% 3/, 2!"%4@
𝐼
)H!/2,%G"% 3/, H@?@H@/C@
k−𝑙𝑜𝑔(𝑎)
As can be inferred from this equation, the a value will result in a (possibly wavelength-
dependent) vertical offset in the absorption.
One could characterize the difference between the sample and reference pulses to
calculate this a. Luckily, you never need to. When you do the same math on the pump off pulse
pair – provided that the splitting between the sample and reference is truly consistent – you
will get
𝐴
%G"% 3??,C3HH@C)@I
= −𝑙𝑜𝑔j
𝐼
)H!/2,%G"% 3??, 2!"%4@
𝐼
)H!/2,%G"% 3??, H@?@H@/C@
k−𝑙𝑜𝑔(𝑎)
The a shows up again! Then when we do the final step in the balanced detection math
𝐴
C3HH@C)@I
= 𝐴
%G"% 3/,C3HH@C)@I
− 𝐴
%G"% 3??,C3HH@C)@I

The 𝑙𝑜𝑔(𝑎) in each term cancels out and we are left with
∆𝐴
C3HH@C)@I
= −𝑙𝑜𝑔j
𝐼
)H!/2,%G"% 3/,2!"%4@
𝐼
)H!/2,%G"% 3/, H@?@H@/C@
k+𝑙𝑜𝑔j
𝐼
)H!/2,%G"% 3??,2!"%4@
𝐼
)H!/2,%G"% 3??, H@?@H@/C@
k
This is the transient absorption signal, corrected for white light fluctuations, in terms of the four
pulses we collect in our experiment. The only assumption we had to make was that relationship
between the sample and reference pulses doesn’t change as a function of time.
125
Balanced detection is a great technique that allows for robust data collection without
too much fretting over white light stability. However, it’s important to mention the drawbacks
associated with balanced detection as well. Because you must now read out 4 pulses for each
experiment instead of 2 in “normal” transient absorption, the data acquisition time is doubled.
The electronics and alignment are also considerably more complex. See the “Arduino Box” and
“Alignment of White Light into Camera” subsections for more in-depth discussions of these
complications.
B.2 Preparing to Run an Experiment
Here, I will briefly describe the general procedure for preparing to run a transient
absorption experiment in the Dawlaty Lab. Detailed explanations of many of the steps are
available in separate subsections.
The first step will typically be tuning the OPA to the wavelength where the sample
absorbs. The appropriate wavelength should be determined by steady state absorption
spectroscopy. In addition to having the appropriate wavelength for the experiment of interest,
one should also have appropriate stability for transient absorption measurements. The specifics
of tuning the OPA and achieving the appropriate stability are discussed in the subsection
“Prepare Your Pump”.
You will need an ND filter after the OPA to attenuate the pump beam for transient
absorption experiments. This is usually placed directly after the OPA. Make sure the ND filter is
in place before any pump alignment is performed. Adding it in after the alignment is complete
may affect the beam path and require realignment. After one has produced a stable pump
126
output and ensured that the ND filter is in place, they should use the two pump mirrors directly
before the retroreflector to align through the retroreflector irises. This will provide decent
alignment into the retroreflector. One doesn’t need to be too careful during this step, since the
pump alignment at the sample will be checked more carefully later. More details about pump
alignment can be found in the subsection “The New Delay Stage and the Importance of Pump
Alignment”.
Next, one should generate stable white light and align both the sample and reference
probe beams into the camera. Generation of stable white light is discussed in detail in
subsection “How To Generate Nice White Light (And Make Sure It’s Good Enough For
Experiments)”, while alignment into the camera is discussed in detail in subsection “Alignment
of White Light into Camera”.
Next, we must spatially overlap the pump and the probe. The easiest way to do this is to
use a pinhole in the same place the sample will be. I typically use a 100 𝜇m pinhole, as the
pump at the probe are usually around 150 𝜇m at their focal points. Using the monochromators
on the sample holder, move the pinhole so that the sample probe beam transmits through it at
its focal point. It is important to remember that your probe beam is already aligned to the
camera, so you do not want to move it! Now you can align the pump through the same hole
using the two pump mirrors before the pinhole. The pump just needs to overlap with the probe
in the sample – its alignment afterwards doesn’t matter – so we can adjust its pointing as much
as we want. Don’t adjust the parabolic mirror directly before the sample, though, since moving
it will change the pointing of both the pump and the probes.
127
Once they are both aligned through the pinhole, it’s a good time to double check the
alignment into the retroreflector by moving the delay stage. We want the pump to go through
the pinhole the same at all delays. Refer to subsection “The New Delay Stage and the
Importance of Pump Alignment” to see the details of this method.
Now, we must find signal. Replace the pinhole with a sample, preferably something that
is known to produce large transient absorption signal, and adjust the sample so the pump and
the probe beam transmit through it. For this step, one can use higher concentrations and
greater pump power than one would use in an actual experiment to ensure that signal is
obvious. If one doesn’t see signal, there is likely one of two reasons: the overlap of the pump
and the probe is not inside the sample, or the probe is hitting the sample before the pump
(a.k.a. “you are before time zero”). One can fix the first issue by moving the sample with
monochromators to place the overlap spot inside the sample. Move it until you find the
maximum amount of signal. One can fix the second issue by changing the delay of the pump
with the delay stage. If time zero does not seem to be on the range of your stage (if the probe
hits the sample before the pump at all possible delays), you may need to move the optics in
your setup to change the overall pathlengths of your pump and/or probe.
Once you’ve found signal, you can (optionally) further optimize it by changing the pump
pointing into the sample by small amounts to increase the signal intensity. Then use the ND
filter after the OPA to decrease the intensity of the pump until you have at most 10-20 mOD of
signal. Decreasing the pump power will help avoid complications like nonlinear effects,
ionization of solvent molecules, sample degradation, and pump scatter. If you’ve tried all the
steps above and still cannot find signal, consider investigating the phase between the pump and
128
probe choppers. See subection “The phase of the probe beam with respect to the phase of the
pump beam” for more discussion on this topic.

Figure B.5: Coherent artefact caused by the overlap of the pump and probe in pure isopropanol
Now that you’ve found signal, it’s important to identify time zero – the delay when the
pump and the probe are overlapped in the sample. Before time zero, there should be no signal,
and at time zero, there should be something called the “coherent artefact”, even in pure
solvents that ordinarily produce no signal. You will sometimes hear this called the “cross-phase
modulation”. An example is shown in figure B.5. If you move to the back of the stage – where
the pump delay is the longest – and still see signal, then time zero is not obtainable from your
current transient absorption setup. You must either make the probe pathlength shorter – so
129
the probe reaches your sample before the pump at the attainable delays - or make the pump
pathlength longer (for the same reason) by moving one of the optics in the beam paths.
When studying a new system for the first time, it’s useful to perform a power-
dependence measurement around the pump power you’d like to use in your experiment. An
example of a power-dependence measurement is shown below in figure B.6 As you change the
pump power, the signal intensity should respond linearly. If it does not, you should probably
attenuate your pump and/or decrease the concentration of your chromophore.

Figure B.6: LEFT: TA spectra of 5-methoxyquinoline in 2,2-dichloroethanol 2 ps after excitation
as a function of pulse energy. RIGHT: Signal strength as a function of pulse energy with fit.

Congratulations – you are now ready to perform a transient absorption experiment on
your chemical system! It is important to note, however, that the transient absorption apparatus
is a complex beast that can misalign itself. If you’ll be collecting data over a long span of time,
it’s important to occasionally re-check the apparatus. For instance, you’ll want to occasionally
check the white light subtraction to ensure that it’s still robust.

130
B.3 Preparing Your Pump
The UV-Visible OPA, the one typically used for transient absorption experiments, is (at
the time of writing) decently well-calibrated. This means that, by typing the appropriate
wavelength into WinTOPAS, the motors will automatically move to the approximate right
location for the generation of that wavelength. If this is not the case, one can directly access
the motors via the WinTOPAS program and switch the wavelength manually. If the alignment is
completely off and adjusting the motors won’t cut it, you may have to open the OPA and play
with some optics. The Coherent OPerA manual is a thorough and helpful resource for
understanding the alignment of the OPA and for general troubleshooting. Additional details,
such as mirror changes necessary for different OPA processes (sum frequency generation vs
fourth harmonic generation vs second harmonic of the sum frequency generation, for
example), can also be found in the manual.
The stability of the pump is important for obtaining good transient absorption data.
Because the transient absorption signal should be linear with respect to the pump intensity, any
fluctuations in the pump will also affect the signal linearly. My threshold for pump stability is
2% standard deviation after monitoring for 60 seconds since I am unconcerned about 2%
fluctuations in the transient absorption signal. I also look at the time trace for the laser power
to ensure that there are no sudden, large changes.
There are three main ways to improve the stability of the OPA output. First, one can
directly access the motors controlling the OPA optics and systematically change them. I typically
monitor the power and attempt to optimize it with these controls, since a higher power usually
131
correlates with higher stability. Note that moving certain optics (most noticeably Crystal 2) can
change the wavelength of the light, so one must be careful when increasing power in this way
to not change the wavelength too dramatically. Second, one can change the compression of the
amplifier output. A tighter compression typically leads to a higher OPA output and greater
stability because the nonlinear processes in the OPA are more efficient at higher photon
concentrations. Third, one can open the OPA and stabilize the white light generated in the pre-
amplification stage. Because this white light is later amplified to generate the wavelengths of
interest, its stability is important to the eventual OPA output. Adjustment of the rotating ND
filter and the pre-WLG iris are the main tools for this. Unless there is a fundamental issue with
the amplifier output or the OPA alignment, one can usually iterate these three steps until the
desired stability is achieved. Note that changing the compression will also impact the white
light stability, so one should always check the white light following compression changes.
B.4 How to Generate Nice White Light
Because we use a balanced detection apparatus, the requirement for the quality of our
white light is much lower than many other transient absorption setups. It is still important to
know how to generate stable white light, because not even the balanced detection apparatus
can handle highly unstable white light. Figure B.7 below shows a diagram of our white light
generation (WLG) stage. It will be referenced throughout this section.
The WLG process will be optimized when the compression of the 800 nm beam is tight. I
recommend optimizing the power of your OPA output using the 800 nm compression before
beginning the white light optimization process. The same compression conditions that generate
132
high OPA output powers seem to also generate stable white light. This makes sense, since the
first step in the OPA is the generation of white light to be amplified!

Figure B.7: A cartoon diagram of the white light generation stage in the Dawlaty lab transient
absorption apparatus.

In the Dawlaty lab, we generate the white light continuum using the 800 nm amplifier
output that is sent to the laser table. The white light generation process is highly nonlinear. As
such, it requires a high concentration of photons. However, the amplifier output has far too
much power for white light generation – if you focus this into your crystal, you will likely
133
destroy it! One should be careful to always attenuate the 800 nm, most likely with an ND filter,
before focusing it into a WLG crystal. If in doubt, start at a point you believe is over attenuated
and then gradually increase the power to avoid damage to your crystal.
In the current Dawlaty lab WLG stage, we use 2 ND filters to achieve an attenuation of
2.5 OD. These are shown above as “non-adjustable ND filter”. These place the power of the
beam near the optimal value. The 800 nm beam then goes through a variable ND filter so that it
can be attenuated more finely. The ability to fine-tune your WLG beam intensity is a powerful
tool in finding stable white light. The variable ND filter is therefore one of the adjustable
parameters you will frequently use when stabilizing the white light.
The 800 nm beam is aligned into our WLG stage with two irises. It is usually easiest to
perform this alignment step after the removal of the non-adjustable ND filters so that the beam
is much brighter. Just make sure to protect the calcium fluoride crystal so that it doesn’t get
burned by the full power of the 800 nm beam! Once the beam is aligned through the irises,
open the first iris all the way – it should always be fully open during WLG. The second iris, on
the other hand, is used to create the largest possible symmetric, clean beam profile. Open the
second iris until the beam leaving it is large but free of any structure, aberrations, or
asymmetry. It should look like a red circle that tapers off in brightness equally in all directions.
We want this type of beam because it will focus the best and give us the cleanest WLG possible.
Remember to replace the ND filters following alignment and beam shaping. Unless there
is an alignment change in your system, you should not need to adjust the irises any further. In
many labs, the pre-WLG iris is treated as an adjustable parameter when finding stable white
134
light. It used to be the case in our own lab. But experience over the past few years has taught
me that the above method – of finding the optimal iris size for a beautiful beam profile and
then not touching it – produces consistently good white light while reducing the degrees of
freedom in the WLG stage. Of course, don’t be afraid to adjust the iris size if you believe it will
help you in some way. You can always undo it!
Now that you hopefully have an attenuated beam with a nice profile, you can generate
some white light! The beam will be focused into a WLG crystal – in our lab, we use calcium
fluoride – to accomplish this. The last major adjustable parameter for generating nice white
light is the focus of the 800 nm with respect to the WLG crystal. You’ll want the tightest part of
your focus inside the crystal. Our WLG crystal, which is moved perpendicular to the beam path
by a motorized stage to avoid burning, sits on a monochromator that moves it parallel to the
beam path and can therefore be used for fine adjustment of the focusing conditions inside the
crystal. The easiest way to make sure that the focusing conditions are optimized is to generate
some white light and then move the crystal back and forth with the monochromator. Where
the white light is brightest should be the correct spot.
If your iris and your focusing conditions have been optimized, your main adjustable
parameter is the variable ND filter. If you carefully optimize these first two conditions, you
therefore only have one real degree of freedom. This simplifies the WLG process considerably.
Stable white light generated using these methods is shown in figure B.8 directly after the
calcium fluoride crystal.
135

Figure B.8: Stable white light immediately after the calcium fluoride crystal in the Dawlaty lab
white light generation stage.

There are a few other issues that could occur. If the crystal is burned or damaged in the
locations where white light is being generated, or if it’s not clean in those spots, WLG will be
negatively affected. These effects can often be seen as flickering or transient aberrations of the
white light beam. To ensure that this is not occurring, place a card shortly after the generation
of the white light and observe the beam for a while. The motor should be moving the crystal,
but the beam profile should not change. If you do see these issues, move the WLG crystal so
that the 800 nm is pumping a different spot and then check the white light beam again. Repeat
until you’ve found a good spot. If you cannot find a good spot, it may be necessary to replace
the crystal altogether. We can avoid burn spots on the crystal by always making sure that the
136
beam is attenuated and that the motorized state is always running when the crystal is exposed
to the beam.

Figure B.9: A demonstration of robust white light subtraction.
Assuming your two probe beams are properly aligned into the spectrometer (see
subsection “Alignment of White Light into Camera”), it is time to check the white light
subtraction using the BalancedPPP Labview program, which is the program we use to run
transient absorption experiments. This program is discussed in detail in the subsection “The
Programs and How They Work”. With your pump beam turned off, run the program. The delays
you choose won’t matter, since you are uninterested in signal at this point and should have the
137
pump beam blocked. If the white light is stable enough for experiments, your final spectrum
should be a flat line right around 0 mOD. An example is shown in figure B.9. Small deviations
from perfect subtraction are often fine, especially if your anticipated signals are large.
If your white light subtraction is unsuccessful, there are several things you can try. First,
go back to the WLG stage and make sure your white light beam is visibly stable. It should not be
changing as a function of the movement of the WLG crystal. If it does change, make sure the
800 nm is not encountering any burn spots or smudges on the crystal, as mentioned above.
Additionally, you want to make sure that the crystal is perpendicular to the beam path of the
800 nm, such that the angle of the crystal does not change with respect to the 800 nm beam
throughout the motion of the motor. If it does, this can result in WLG efficiency changes during
the motion and in small changes in alignment post-crystal that will affect white light
subtraction.
If your white light beams (sample and reference) are passing through a cuvette (or some
other type of sample) while you are studying the white light subtraction, make sure that they
are hitting similar spots. If one is hitting a dirty spot that is causing some scattering, this can
negatively impact your white light subtraction. This issue is particularly bad if you are rastering
your sample, such that the scattering conditions of your two probe beams can change as a
function of time. Make sure that your cuvettes are clean and that your solutions are free of
scattering particles. It’s also important that the two beams are passing as perpendicularly
through the cuvette as possible, so that there are no diffraction effects caused by the motion of
the cuvette that can negatively impact the alignment post-cuvette.
138
Careful alignment into the spectrometer also helps, perhaps by mitigating some of the
time-dependent misalignment effects described above. Ensure that the beams are horizontally
centered on the vertical spectrometer slit by optimizing the intensity that reaches the camera.
Optimization of alignment at the camera is discussed in greater detail in the subsection
“Alignment of White Light into Camera”. Because moving the WLG crystal and the sample
cuvette can result in small changes in alignment, be sure you always re-check the alignment
following any motion of these two objects. If you remove the sample cuvette to change the
solution, always check the alignment again before continuing experiments. You’d be surprised
how much it can change!
Finally, you can readjust any of the three parameters we discussed initially: the
attenuation, the iris size, and the focusing conditions. Attenuation is the easiest, so I would
recommend adjusting only this parameter of the three unless optimal white light conditions
absolutely cannot be achieved otherwise. Iterate all the processes discussed above until the
white light subtraction is robust enough such that you are comfortable using it for experiment.
B.5 The Programs and How They Work
In this section, I will discuss the programs used for transient absorption experiments in
the Dawlaty lab, what they do, and what the necessary inputs are.
The main program used for running transient absorption experiments is a Labview
program called “BalancedPPP”, where the PPP stands for Pump Probe Pixis (Pixis is the
transient absorption camera). When you ready to run an experiment or study white light
139
subtraction, this is the program you should use. The front panel of the program is displayed
below in figure B.10 with important inputs labeled with colored boxes.

Figure B.10: The BalancedPPP Labview program used for running transient absorption
experiments with important inputs labeled.

In red, you can see the inputs for the directory the data will be saved in and the
requested experiment title. Make sure your experiment title is acceptable as a file name, or
else the data will not be saved properly. In gold, you can see the requested delays for the
transient absorption experiment: the initial delay, the final delay, and the step size. Note that
these values are all input as femtoseconds. Also note that “0’ delay corresponds to the back of
Range of
pixels for
white light
subtraction
figure
Range of
wavelengths
for data
figure
Center
wavelength
(used for
wavelength
calibration)
Number of spectra to
throw out before
data collection
GPIB info for delay
stage controller
Delay range and
step size
Name of
data file and
directory
to save it to
140
the delay stage – not necessarily to the actual time zero of your experiment. In blue, you see
the number of averages per time point. Since each balanced detection experiment requires four
pulses to be recorded, the number of pulses read out per time point will be four times this
amount. I recommend 2000 averages per time point when collecting data for publication. It is
convenient to use fewer when performing quality checks or exploratory experiments.
In grey, you see the input called “Throw out”, which is the number of spectra that are
thrown out before data collection begins. The utility and importance of this input are discussed
in detail in the subsection “The phase of the probe beams in the camera readout”. In black are
the delay stage control inputs. The axis value must match the location of the delay stage’s
connection to the controller, and the GPIB value must match the setting for the delay stage on
the controller. Otherwise, the program cannot communicate with the delay stage to provide
the requested delays.
In orange, you see the center wavelength input. This input is used for spectrometer
calibration when generating the wavelength axis of the experiment by assigning the center pixel
of the camera to a wavelength. This value should generally be 599 nm (along with the setting
on the IHR 320 spectrometer controller, discussed below), although it is important to
occasionally calibrate the spectrometer to ensure that the value remains consistent. In green,
you see the pixel range that should be displayed in the white light figure. See figure 51 above
for an example of where this pixel range is used. In the current calibration of the spectrometer,
pixel 200 is around 400 nm and pixel 800 is around 650 nm, so this is typically the range I use. In
purple, you see the range of wavelengths that should be used when displaying the transient
absorption data. Note that, regardless of which input range you use, the transient absorption
141
data that is saved includes the entire range of the camera. This selection is for your
convenience during the experiment. Because I am typically only interested in 400 nm to 650
nm, I choose these as my wavelength range.

Figure B.11: Important icons and files in the Lightfield program.
Next, I mention Lightfield, the program used to communicate with the Pixis camera.
Lightfield can display both the individual white light pulses, read out at a rate of 1 kHz, and an
image of the entire camera showing the location and orientation of the white light beams
dispersed onto the camera. Images of these two modes of readout are shown and discussed
below in the subsection “Alignment of White Light into Camera”. One can choose between
these two modes by clicking on the folder icon (shown in the green box in figure B.11) and
choosing “1kHz” (shown in red) for the individual pulse readout or “Image View” (shown in
blue) for the camera image readout. Readout can be started by hitting the “Run” button (shown
in yellow) and stopped with the “Stop” button (shown in violet).
142
Note that the readout will not occur until the camera is being triggered, which won’t
occur until the ArduinoSendEdges Labview program is used to toggle the Arduino box on. The
arrow in the green box in figure B.12 should be pressed to turn the Arduino box on and off.
What the Arduino box is and the essentials of how it works are discussed in the subsection
“Arduino Box”. Be sure to turn the Arduino box off again when you are done using Lightfield. If
it is left on, the BalancedPPP program will not work.

Figure B.12: The front panel of the ArduinoSendEdges Labview program.
Finally, I will mention the program used to control the spectrometer, called
USBSpectrometerControl, which is shown in figure B.13. The purple box is where the correct
spectrometer is chosen from a dropdown list. Ours is called “IHR 320 (yes)” for whatever
reason. The wavelength of the camera’s center pixel is dictated by the position of the grating in
the spectrometer, which is controlled with the “Position, nm” input in the green box. Note that
143
the value input here should match the value input for “Center Wavelength” in the BalancedPPP
Labview program, discussed above. The appropriate grating for the experiment of interest is
chosen under “Grating”, shown in the red box. I always use the grating “150, 500nm” for my
experiments. Finally, the entrance slit of the spectrometer can be controlled via the input
shown in the orange box.

Figure B.13: The USBSpectrometerControl program used to control the transient absorption
spectrometer with important inputs labeled.

144
B.6 Alignment of White Light into Camera
In balanced detection transient absorption, we must record sample and reference pulse
pairs. Because they are generated at the same time, the pulse pairs will also be incident on the
camera at roughly the same time. To distinguish the two pulses, we must direct them to
different parts of the camera and read the two parts of the camera separately.

Figure B.14: Cartoon diagram of the readout of the Pixis camera in a balanced detection
transient absorption experiment.

To accomplish this, we split the CCD array into two readout “zones”: the top and the
bottom of the camera, which are each read out in 1 ms. The bottom of the camera is read out
before the top, so we typically align the sample pulse onto the bottom half so it is read out
145
before the corresponding reference pulse. This is necessary because the proper pulse order for
our math is as shown in table B.1.
Pulse Order 1 2 3 4
Pulse Identity Sample
Pump On
Reference
Pump On
Sample
Pump Off
Reference
Pump Off

Table B.1: Correct order of probe pulses in our balanced detection transient absorption
experiment.

Lightfield is a program we use to study the readout of our camera. One can choose to
look at the readouts of the top and the bottom of the camera as a function of time using the 1
kHz experiment. This shows the individual white light spectra read out at a rate of 1 kHz. One of
these white light spectra are shown in Lightfield in figure B.15 below. Use of this method
requires the Arduino box to be turned on, which can be accomplished via the use of the
ArduinoSendEdges Labview program, as discussed in subsection “The programs and how they
work”. Always be sure to turn off the Arduino Box (using the same program) when you are done
using it. If the Arduino Box is left on, the transient absorption program will not run correctly.
One can also use the experiment “Image view” which shows the readout of the entire
CCD array. This view allows one to easily see where the two probe pulses are hitting the camera
and adjust them accordingly. In figure B.16 below, you can see the dispersed reference probe
hitting the top of the camera and the dispersed sample probe hitting the bottom of the camera.
The red color shows saturation of the camera.

146

Figure B.15: Readout of Lightfield in “1 kHz” mode, where individual white light pulses are
shown as a function of time.

Figure B.16: Readout of Lightfield in “Image View” mode, where the entire CCD array is
displayed at once.

147

Figure B.17: Mirror pair used for individual control of the sample and reference probe
alignment into the spectrometer.

Most optics in the probe beam path control the alignment of both the sample and
reference pulses. There is one mirror pair that allows for individual control of the sample and
Sample
Probe Mirror
Reference
Probe Mirror
148
reference probe pulses. A picture of this mirror pair is shown below in figure B.17. The top
mirror controls the sample pulse, and the bottom controls the reference pulse. These should be
adjusted while viewing the CCD array to ensure the proper locations on the camera are being
hit. Don’t be confused by the fact that the top mirror controls the sample probe, which is
supposed to hit the bottom of the camera!
Once the probe beams are properly aligned into the spectrometer, one should switch
back to the 1 kHz readout on Lightfield to view the intensity of the individual pulses. If it is too
low, you may open the slit on the spectrometer to increase the intensity that reaches the
camera. I typically open the slit until the white light is saturating above pixel 800 but not below.
It is good practice to check the horizontal alignment of the probe beams after adjusting the slit
width to ensure that you are properly centered.
Always be sure to close Lightfield when you are done using it. If Lightfield is left on, the
transient absorption program will not run correctly.
B.7 The New Delay Stage and the Importance of Pump Alignment
In mid-2021, I introduced a new delay stage to the Dawlaty Lab transient absorption
apparatus. The stage increased our possible delay range from 600 ps to 4 ns. This is a significant
increase, especially when studying organic chromophores, which often have excited state
lifetimes on the order of single nanoseconds. To properly use these long timescales, the
alignment of the pump into the delay stage’s retroreflector must be robust.
Delay stages work using a retroreflector, a set of mirrors which reflect light at the same
angle it enters. If light enters a retroreflector at a 90-degree angle, it will also exit that
149
retroreflector at a 90-degree angle. This means that, regardless of the retroreflector’s position
in space, provided it moves parallel to the beam path, the direction of the beam will be
unaltered. We can, in principle, change the delay of the beam with no changes in the beam
alignment. This is important in transient absorption spectroscopy, because we need the overlap
between the pump and the probe at the sample to be identical regardless of the time delay.

Figure B.18: The 4 ns delay state with a retroreflector.
When introducing this new stage, the first job I had was to find the proper beam path
into the retroreflector such that the alignment of the pump doesn’t change when we move the
delay stage. We initially found the proper beam path into the retroreflector by sending the
pump beam a long distance post-retroreflector, so that any small changes in alignment would
be amplified and thus easy to identify. By using two mirrors placed before the delay stage to
150
ensure that the beam hit the exact same spot at both ends of the delay stage, we found the
appropriate beam path. In short, we moved the delay stage from the front (where the delay is
shortest) to the back (where the delay is longest) and then used the second pre-retroreflector
mirror to correct for any beam path changes. We moved back to the front of the delay stage
and used the first pre-retroreflector mirror to correct again. We iterated this process until there
was no longer a change in alignment at the back and front of the delay stage. Once we found
the correct alignment, we placed irises around the beam for future convenience. This strategy
can be used whenever careful alignment is needed in the laser lab.
Those irises are still in the transient absorption apparatus. Pointing the pump beam
through the irises will provide decent alignment at the sample, regardless of delay stage
position, and should be completed each day you use the transient absorption apparatus. The
first mirror pre-retroreflector should be used to align into the first iris; the second mirror should
be used to align into the second iris; and these two steps should be iterated until the pump
goes through both irises cleanly.
However, it’s also important to directly check whether the movement of the delay stage
affects the overlap of the pump and the probe, since this is ultimately the thing we care about.
Shortly before collecting data, after white light has been aligned and stabilized, put the pump
and the probe through a pinhole. Move the delay stage from one end to the other and observe
whether any changes in the alignment of the pump occur. It is likely that they will because
alignment through the pinhole at the focus is much more sensitive than alignment of the
unfocused pump beam through the irises. One can correct for this by using the first pre-
retroreflector mirror to repoint the pump through pinhole at the front of the delay stage, and
151
the second pre-reflector mirror to repoint the pump through the pinhole at the back of the
delay stage. Iterate these steps until the pointing of the pump through the pinhole doesn’t
change across the entire delay stage. When this is the case – congratulations! Your system is
now well-aligned, and you can take full advantage of the entire delay stage without worrying
about changes in alignment.
B.8 The Phase of the Probe Beam with Respect to the Phase of the Pump Beam
The balanced detection apparatus can be confusing! The need to chop both the probe
pulses (from 1 kHz to 500 Hz) and the pump pulses (from 1 kHz to 500 Hz, then from 500 Hz to
250 Hz) can occasionally result in a problem where the “phases” of your probe and pump
beams do not match, and you won’t see any signal – even though the beams are overlapped in
space and you’re supposed to be shortly after time zero!
The “correct” phase, required for seeing signal and collecting data, is shown at the top of
figure B.19. There is a probe beam every 2 ms and a pump beam every 4 ms. They overlap in
the sample. Life is good. The “incorrect” phase, which will prevent you from seeing signal and
collecting data, is shown on the bottom of figure B.19. Once again, there is a probe beam every
2 ms and a pump beam every 4 ms. You’d think you were doing fine, but you’d be thinking
incorrectly. The pump and the probe are NOT overlapping in the sample – at least, not
anywhere near each other in time. In fact, after the pump hits the sample, the probe doesn’t
get there until 1 ms later! If excited molecules were somehow still excited, they’ll have diffused
away from the overlap position. You won’t see any signal.

152

Figure B.19: A cartoon diagram showing the correct phase for the pump and probe pulses in our
balanced detection pump probe experiment versus the incorrect phase.

How does this occur? It boils down to the phase of your choppers being incompatible
with one another. If all the chopper boxes are turned on prior to the SDG Elite (the clock of the
laser lab] and are already in phase, this problem normally shouldn’t occur. Sometimes laser
gnomes turn off the probe chopper box at night and you don’t notice until the laser is already
warmed up. When you start the probe chopper after the pump choppers are already running,
Pump and probe
temporally
coincident in sample
Pump and probe
NOT temporally
coincident in
sample
Correct
Phase
Incorrect
Phase
153
you have a 50/50 chance that its phase will be correct (compatible with the pump choppers).
It’s good to get in the habit of always ensuring that the chopper boxes are on prior to starting
the SDG Elite to prevent this problem. There have also been times when this phase issue has
occurred seemingly out of nowhere for seemingly no reason, even with all the chopper boxes
turned on, so keep that in mind as well.

Figure B.20: A cartoon diagram of an oscilloscope readout of the pump (purple) and the probe
(green) with (top) correct phase and (bottom) incorrect phase. The vertical grey lines indicate 1
ms spacing.

Do you want to know if you have a phase problem? If you’re worried this might be
happening – perhaps because you’re not seeing signal, even though you really think you should
be! – there is an easy diagnostic strategy. Direct the pump and probe into two different
photodetectors and view the output on an oscilloscope. Make sure you study the beams after
Correct
Phase
Incorrect
Phase
154
their respective choppers, or they’ll both just be coming every ms. If you trigger with respect to
the pump pulses, you’ll be able to see whether the pump and probe are in phase. If they are in
phase, you’ll see something like the top of figure B.20. The probe pulses, shown in green, come
every 2 ms. The pump pulses, shown in purple, come every 4 ms. They hit their relative
photodectors at the same time, so your pulses are in phase! If the phase is incorrect, you’ll see
something like the bottom of figure B.20. The probe pulses come every 2 ms; the pump pulses
come every 4 ms; but they hit their relative photodectors 1 ms apart! You’ll never see signal if
they’re hitting the sample like that.
How do you fix this problem? The easiest solution is to turn the chopper box for the
probe on and off. The phase of the probe will be effectively randomized with respect to the
pump pulses, and you’ll have a 50% chance of fixing the issue. Repeat until the problem is fixed.
You know the problem is fixed if you see the appropriate oscilloscope result, shown above, or if
signal appears. If this phase problem occurred even though all the chopper boxes were turned
on before the SDG Elite and there was no human interference with the phase of the choppers,
turning the chopper box on and off will only fix it for that day. To fix it for good, go into the
chopper box and adjust the phase until you achieve the proper oscilloscope readout. Again:
only do this if you’re having the phase issue and you’re certain all the boxes were turned on
before the SDG Elite! Otherwise, you’re causing a potential headache for future you. When you
adjust the phase directly, be sure to check the 800 nm on a card shortly after the chopper to
make sure that the center of the beam if not being clipped by the chopper. On the left in figure
B.21 you see what the 800 nm beam should look like following the chopper if it is being
155
chopped correctly. The center of the beam is not cut by the edge of the chopper. On the right,
you see what the 800 nm beam will look like if it is not being chopped correctly.

Figure B.21: 800 nm amplifier output immediately following the chopper when (left) it is
chopped correctly and when (right) it is chopped incorrectly.

B.9 The Phase of the Probe Beams in the Transient Absorption Math
One of the inputs of the BalancedPPP Labview program used to run transient absorption
experiments in the Dawlaty Lab is an integer called “Throw Out”. Essentially, this tells the
program how many camera readouts to throw away before data analysis begins. This input has
two important aspects: first, by selecting a large enough Throw Out value (usually around 20),
we can effectively “clean” the pixels, so that they no longer show the effects of the pulses that
hit the camera while it was not being read out and probably caused the camera to saturate.
800nmafter
chopperwhen
choppedcorrectly
800nmafter
chopperwhen
choppedincorrectly
156
Second, it determines the “phase” of the probe beams in the experiment math. This will be
explained below.
Each readout from the TA camera corresponds to one probe spectrum. The order in
which the probe spectra are read out is dictated by where they hit the camera, since the
bottom of the camera is read out first, followed by the top. See the subsection “Alignment of
White Light into Camera” for more details. The order of the pulses is always the same: Sample
Probe, Pump On; Reference Probe, Pump on; Sample Probe, Pump Off; and Reference Probe,
Pump Off. The pulses then repeat. However, the “phase” of these pulses – that is, which of
these pulses will be read out first by the camera in an experiment - isn’t obvious. As discussed
in the “Arduino box” section of this appendix, the 250 Hz output of the second chopper box is
used to ensure that the identity of the first probe beam on the camera when the experiment
starts is consistent. For example, if the Reference Probe, Pump Off pulse is read out first by the
camera in an experiment, it will also be read out first in the next experiment unless
fundamental changes are made to the system. The phase is consistent, but ill-defined, because
we have no good way of figuring out the identity of that first pulse.
The MATLAB code that does the transient absorption math requires the first pulse that’s
recorded to be Sample Probe, Pump On. This is demonstrated by the table and math below.
We can make sure this is the case by choosing the correct “Throw Away” value. The Throw
Away value will determine the identity of the first pulse that’s recorded and used for analysis by
throwing away the correct number of pulses before it. Of course, this still requires a way of
figuring out when Sample Probe, Pump On is indeed being recorded first.
157

Figure B.22: The correct pulse order for balanced detection transient absorption in the Dawlaty
lab setup and the subsequent math.

There are several clues. For example, in our experiments, the sample probe is typically
more intense than the reference probe at the camera when they are both optimally aligned. If
the less intense spectrum is assigned the “sample” probe, this is probably incorrect. If one turns
up the pump to a high power when there is overlap in the sample, one might see the transient
signal (or pump scatter) in the white light spectrum. The white light spectrum that shows such
signal should be labeled as the Sample Probe, Pump On.
If the white light is subtracting poorly, and it seems to be because the sample and probe
difference spectra have different signs, then the “Throw Away” value is off by either 1 or 3. This
is caused by the reference spectrum being labeled as a sample spectrum and vice versa. An
example of this phenomenon is shown below in figure B.23. Notice that in one case we have
Pulse Order #1 #2 #3 #4
Pulse Identity
Sample
Pump On
Reference
Pump On
Sample
Pump Off
Reference
Pump Off
Sample Difference Spectrum = −log
#"
##
=−log
$
!"#$%&, $(#$)*
$
!"#$%&, $(#$)++
Reference Difference Spectrum =−log
#%
#&
=−log
$
,&+&,&*-&, $(#$)*
$
,&+&,&*-&, $(#$)++
158
pump on divided by pump off, while in the other we have pump off divided by pump on. This
results in the difference spectra having different signs and subtracting incorrectly.

Figure B.23: The pulse order for balanced detection transient absorption when the phase of the
pulses is off by 1 and the subsequent math.

The final clue for finding the correct phase is the sign of the transient absorption signals.
If the white light is subtracting correctly but the sign of your transient signal is opposite what
you expect – for example, if a feature you believe to be excited state absorption shows a
negative ∆A instead of a positive one – then your Throw Away value is likely off by 2. The math
in this case is shown below in figure B.24. Because both fractions are now inverted with respect
to the expected math, both difference spectra have the wrong signs. Your transient absorption
data will also be the wrong sign.

Pulse Order #1 #2 #3 #4
Pulse Identity
Reference
Pump On
Sample
Pump Off
Reference
Pump Off
Sample
Pump On
Sample Difference Spectrum = −log
#"
##
=−log
$
!"#"!"$%", '()'*$
$
!"#"!"$%", '()'*##
Reference Difference Spectrum =−log
#%
#&
=−log
$
+,)'-", '()'*##
$
+,)'-", '()'*$
159

Figure B.24 The pulse order for balanced detection transient absorption when the phase of the
pulses is off by 2 and the subsequent math.

B.10 Arduino Box
Balanced detection pump probe is complicated. To do the appropriate math to calculate
the resulting transient absorption signal, we have to keep track of four different probe pulses.
As discussed at length in the section “The phase of the probe beams in the transient absorption
math”, if the identities of these pulses are confused, signs can be switched and white light
subtraction can become erratic.
We must know, when the transient absorption program starts, that the first pulse we
read out will be consistent from experiment to experiment. Otherwise, we have no hope of
labeling each subsequent pulse correctly. If the computer began collecting data from the
camera the second you clicked “start”, the first pulse would be random – whatever had just hit
Pulse Order #1 #2 #3 #4
Pulse Identity
Sample
Pump Off
Reference
Pump Off
Sample
Pump On
Reference
Pump On
Sample Difference Spectrum = −log
#"
##
=−log
$
!"#$%&, $(#$)**
$
!"#$%&, $(#$)+
Reference Difference Spectrum =−log
#%
#&
=−log
$
,&*&,&+-&, $(#$)**
$
,&*&,&+-&, $(#$)+
160
the camera would be read out first and it would change from experiment to experiment! We
need something that keeps track of the identity of the pulses.
The Arduino box is the tool that performs this feat. It is called the “Arduino box”
because it contains an Arduino, in addition to some circuitry. The exterior of the box, including
its inputs and outputs, is shown below in figure B.25. The Arduino box takes in a 1kHz electronic
pulse train from the SDG Elite (blue) and a 250 Hz electronic pulse train from the output of the
second pump chopper (purple) and is linked via a USB connection (red) with the transient
absorption computer. A Labview program called ArduinoSendEdges toggles the Arduino on and
off. When it’s toggled on, a bright green light will appear inside the Arduino box. After it’s
turned on, the Arduino waits to receive both the 1 kHz pulse train and the 250 Hz pulse train.
When it has received them, the Arduino box outputs a 1 kHz pulse train that is sent to the
camera (green).

Figure B.25: A picture of the Arduino box with labeled inputs and outputs.
161
This 1 kHz pulse train triggers the acquisition of the probe spectra. The fact that the kHz
train isn’t sent until the 250 Hz train is received is the mechanism for keeping track of the
identity of the probe pulses. The 250 Hz train has a pulse every 4 ms. Since there are 4 different
types of probe pulses that each need to be read out in 1 ms, the 250 Hz pulse always
corresponds to the readout of the same probe pulse every time. It ensures that the first probe
pulse read out in each experiment is consistent.

Figure B.26: Cartoon diagram of the operation of the Arduino box.
If you are interested in learning about the circuitry involved in the Arduino box, see the
appendix of Eric Driscoll’s dissertation. He developed this technology!

162
Appendix C
Global Fitting of Transient Absorption Data In MATLAB
Here I will discuss my general procedure for performing global fitting of transient
absorption data in MATLAB. I will also mention the specifics of how I fit the 5-methoxyquinoline
in methanol data, accounting for solvent reorganizational effects.
C.1 General Procedure

Figure C.1: Unprotonated and protonated basis spectra for 5-methoxyquinoline in TFE and their
analytical gaussian fits used in the kinetic model.

I will walk through my global fit of 5-methoxyquinoline in pure TFE to demonstrate my
general procedure. I discuss my method for finding the basis spectra used for global fitting in
Chapter 1. Once those basis spectra have been found, we must generate an analytical
expression for them to use in our kinetic fit. This can be accomplished by fitting them as a sum
of Gaussians using the “Curve Fitting” app in MATLAB. This is demonstrated in figure C.1. Note
400 450 500 550 600 650
-0.5
0
0.5
1
1.5
2
Wavelength / nm
D
A / mOD

Experimental protonated spectrum
Gaussian fit
400 450 500 550 600 650
0.5
1
1.5
2
2.5
3
3.5
Wavelength / nm
D
A / mOD

Experimental unprotonated spectrum
Gaussian Fit
163
that we are interested in getting a robust fit, not in extracting physical significance, so you may
use an arbitrarily large number of Gaussians for this step.
You’ll want to keep track of these fits for later. For demonstration purposes, I will show
the three Gaussian fit of the unprotonated spectrum below:
𝑎
;
∗𝑒
'S
#'F
)
C
)
T
$
+𝑎
7
∗𝑒
'S
#'F
$
C
$
T
$
+𝑎
=
∗𝑒
'(
#'F
%
C
%
)
%

where
𝑎
;
= 2.21 𝑚𝑂𝐷 ,𝑏
;
= 410.3 𝑛𝑚,𝑐
;
= 16.3 𝑛𝑚
𝑎
7
= 1.384 𝑚𝑂𝐷,𝑏
7
= 425.8 𝑛𝑚,𝑐
=
= 25.43 𝑛𝑚
𝑎
=
= 0.7714 𝑚𝑂𝐷,𝑏
=
= 587.4 𝑛𝑚,𝑐
=
= 121.1 𝑛𝑚
𝑅
7
= 0.9915
Now that we have analytical fits for the basis spectra, we can use the “Curve Fitting”
tool in MATLAB to fit the entire 2D transient absorption dataset. Open the wavelength and time
axes as either the x and y data, respectively, depending on the arrangement of the transient
absorption data, and open the transient absorption data as the z data. The data should now be
shown in a 3D plot.
You should choose “Custom Equation” from the dropdown menu. We will put our
kinetic model, along with our basis spectra, in the input space that appears. The model for this
system (as derived in chapter 1) is shown below. This is what will be used as the fitting input in
MATLAB.
𝑒
'
)
U
*+
∗(𝑢𝑛𝑝𝑟𝑜𝑡𝑜𝑛𝑎𝑡𝑒𝑑 𝑠𝑝𝑒𝑐𝑡𝑟𝑢𝑚)+U1− 𝑒
'
)
U
*+
V∗(𝑝𝑟𝑜𝑡𝑜𝑛𝑎𝑡𝑒𝑑 𝑠𝑝𝑒𝑐𝑡𝑟𝑢𝑚)
164
where the analytical Gaussian forms of the basis spectra are used in place of “unprotonated
spectrum” and “protonated spectrum”. The entire transient absorption data set can now be fit
with the model with just one adjustable parameter: 𝜏
-,
. My fit of the transient absorption data
of 5-methoxyquinoline in TFE is shown in figure C.2. The fit has an 𝑅
7
of 0.9651.

Figure C.2: (Left) Transient absorption spectra for 5-methoxyquinoline in TFE and (Right) the
global fit of the data.

This same general procedure can be used for more complicated kinetic models or for
fits with more adjustable parameters. We will see an example of the latter below.

C.2 Fitting 5-Methoxyquinoline in Methanol TA Data
As discussed in chapter 1, the effects of solvent reorganization could not be reasonably
ignored for 5-methoxyquinoline in methanol due to the excited state equilibrium reaction and
the resulting long lifetime of the base form of MeOQ. We therefore had to allow the basis
spectrum for the base form of MeOQ in methanol to relax according to some solvent
reorganization timescale, 𝜏
2345@/)
.
400 450 500 550 600 650
0
1
2
3
Wavelength / nm
D
A / mOD

0.3ps
1ps
3ps
5ps
7ps
9ps
400 450 500 550 600 650
0
1
2
3
Wavelength / nm
D
A / mOD

0.3ps
1ps
3ps
5ps
7ps
9ps
τ = 2.3 ± 0.6ps
165
In the analytical form for the base basis spectrum, we introduced three new adjustable
parameters (in addition to 𝜏
2345@/)
) for each Gaussian – one to capture changes in magnitude,
one to capture changes in center wavelength, and one to capture changes in linewidth – to
capture these reorganization effects. Each of these changes should occur at the same timescale
(𝜏
2345@/)
) but need not occur with the same magnitude, so we need a separate parameter for
each. Because we had three Gaussians in our basis spectrum fit, we introduced nine new
parameters in total. An example expression for one Gaussian is shown below, where o, p, and q
are the new adjustable parameters.
}𝑎
;
+𝑜∗U𝑒
'
)
U
,-./012
−1V~∗𝑒
'
⎝
⎜
⎜
⎜
⎛
#'YF
)
Z%∗[@
3
2
4
,-./012';\]
C
)
Z^∗[@
3
2
4
,-./012';\
⎠
⎟
⎟
⎟
⎞
$

The recurring motif of 𝑒
'
2
4
,-./012
−1 works in the following way: at t = 0, this term is 0
and so the initial basis spectrum is unperturbed. As time increases, the term becomes more
negative and thus the initial basis spectrum becomes more and more perturbed as a rate
according to the time constant 𝜏
2345@/)
. The fact that this term becomes negative is
inconsequential, since the negative sign can be arbitrarily absorbed into the constants o, p, and
q, which determine the magnitude of the basis spectrum perturbation.
Once the basis spectra were formulated in the above way, the data was globally fit with
the appropriate kinetic model (discussed in chapter 1) similarly to the TFE data shown above –
only with lots more adjustable parameters!

166
Appendix D
Excited State Löwdin Analysis and Electron Density Difference Maps in Q-Chem
In this section, I will briefly describe how to perform some of the calculations I
presented in chapter 4. Namely, I will discuss how to calculate Löwdin charges for excited
states, how to output electron density difference files, and how to plot those files.

Figure D.1: A Q-Chem input file for requesting excited state single point energy calculations of
quinoline and the excited state Löwdin charges.
$molecule
1 1
C -1.2144507 -1.3903482 0.0000000
C -2.4378524 -0.7208944 0.0000000
C -0.0216151 -0.6702257 0.0000000
C -2.4685998 0.6732045 0.0000000
C -1.2795677 1.3937101 0.0000000
C -0.0282523 0.7515405 0.0000000
N 1.2077435 -1.3205953 0.0000000
C 2.4146463 -0.6440310 0.0000000
C 2.4250500 0.7270707 0.0000000
C 1.2231757 1.4515668 0.0000000
H -1.1876335 -2.4780772 0.0000000
H -3.3636834 -1.2883755 0.0000000
H -3.4216628 1.1934260 0.0000000
H -1.3008106 2.4807119 0.0000000
H 3.3065446 -1.2572378 0.0000000
H 3.3824407 1.2371040 0.0000000
H 1.2261195 2.5364135 0.0000000
H 1.2100829 -2.3286704 0.0000000
$end

$rem
BASIS = 6-31+G*
CIS_MULLIKEN = 1
CIS_N_ROOTS = 4
CIS_TRIPLETS = 0
EXCHANGE = omegaB97X-D
GUI = 2
LOWDIN_POPULATION = 1
$end

167
First, I will discuss the excited state Löwdin analysis. To perform excited state electronic
structure calculations, you must first optimize the geometry of the ground state. Then you can
use that optimized geometry to perform an excited state single point energy calculation. It’s
best to include several excited states if you don’t already know what the excited state of
interest is. You can request the Löwdin charges and Löwdin charge differences for those excited
states straightforwardly. An example input file is shown in figure D.1 for quinoline.
The command ‘CIS_N_ROOTS’ instructs Q-Chem to calculate excited state energies using
TDDFT, and the number which follows – in this case, 4 – tells Q-Chem how many excited states
to calculate, starting with the lowest energy state and moving up. ‘CIS_TRIPLETS = 0’ tells Q-
Chem to ignore triplet states, which is useful if you know your states of interest should be
singlet states. ‘CIS_MULLIKEN = 1’ and ‘LÖWDIN_POPULATION = 1’ tell Q-Chem to calculate the
Mulliken and Löwdin charges, respectively, for the excited states. Both Mulliken and Löwdin
charges are a form of population analysis – that is, assigning quantities of electronic charge in a
molecule to individual atoms. I used Löwdin charges in my analysis because they are known to
have less basis-set dependence than Mulliken charges, but both can be valuable tools.
Next, I will discuss the generation of electron density difference files. You will calculate
excited states in a similar way as above, but with an added command (‘MAKE_CUBE_FILES’) and
an added section (‘plots’) in the input file. An example is shown in figure D.2.
168

Figure D.2: A Q-Chem input file for requesting electron density difference maps of quinoline,
output as cube files via the $plots section of Q-Chem.

1 $molecule
2 0 1
3 C -1.2053753 -1.3970755 0.0000000
4 C -2.3938012 -0.7117606 0.0000000
5 C 0.0287126 -0.6957660 0.0000000
6 C -2.4081752 0.7054933 0.0000000
7 C -1.2302112 1.4084512 0.0000000
8 C 0.0148054 0.7266696 0.0000000
9 N 1.1857427 -1.4183570 0.0000000
10 C 2.3201918 -0.7587036 0.0000000
11 C 2.4192968 0.6550989 0.0000000
12 C 1.2654736 1.3932346 0.0000000
13 H -1.1718317 -2.4821302 0.0000000
14 H -3.3362368 -1.2516626 0.0000000
15 H -3.3597642 1.2286190 0.0000000
16 H -1.2340713 2.4959617 0.0000000
17 H 3.2278351 -1.3602361 0.0000000
18 H 3.3977744 1.1241055 0.0000000
19 H 1.2921004 2.4805026 0.0000000
20 $end

21 $rem
22 BASIS = 6-31+G*
23 CIS_N_ROOTS = 3
24 CIS_TRIPLETS = 0
25 EXCHANGE = omegaB97X-D
26 GUI = 2
27 MAKE_CUBE_FILES = TRUE
28 $end

29 $plots
30 Make Quinoline EDD maps
31 100 -5.0 5.0
32 100 -5.0 5.0
33 40 -2.0 2.0
34 0 4 0 0
35 0 1 2 3
36 $end

169
You want to request only enough excited states to reach the excited state of interest.
Line 27 turns on the generation of cube files, which is how the electron density difference maps
will be output. We add the ‘plots’ section to specify the details of these cube files. Line 30 is a
comment for your own reference. Lines 31 – 33 specify the number of points and the range of
values calculated for the electron density difference maps in the x, y, and z directions,
respectively. It’s best to look at the geometry input of your molecule to figure out the
appropriate values. You’ll want to expand the range 1-2 units beyond the structure of your
molecule, since the density difference maps will be larger than the molecule. You’ll also want to
choose a generous number of points for the generation high resolution images. Line 34
specifies the type and number of cube files. In this case, it specifies that 4 electron density
difference files will be generated. All other numbers on this line can be left as zero. Line 35
indicates the states for which electron density differences should be output. Unfortunately, it
seems mandatory to calculate all electron density difference maps up until your excited state of
interest. For example, if you’d like to calculate the electron density difference map for excited
state 2, you must calculate three total maps: the ground state (0), the first excited state (1), and
the second excited state (2).
Note that the cube files are stored in the scratch directory generated for the calculation.
This directory is typically deleted automatically after the calculation is complete. You’ll likely
need to use a special run command to save the scratch directory.
Finally, you’ll need to visualize the electron density difference maps. For the pictures
presented in chapter 4, I used a free software called VMD, which stands for Visual Molecular
Dynamics. The software is user-friendly and allows for customization of the molecular
170
representation and of the electron density difference map. I encourage you to play around with
the program to figure out what customization is possible and use this to generate your ideal
images.
The trickiest part of generating my electron density difference maps in Chapter 4 was
realizing that I needed to plot both positive and negative isosurfaces. The negative isosurface
shows the areas where there is an increase in electron density upon photoexcitation, while the
positive isosurface will show areas where there is a decrease in electron density upon
photoexcitation. Under “Graphics” on VMD, you should generate three “Representations”. The
first will be your molecular structure. I chose to represent my molecule using the ‘CPK’ format.
The other representations will be the two isosurfaces, which can be selected in the drawing
methods for the representation. You’ll need to play around with the isosurface value until you
find a map that fits your liking. Then you should use the same isosurface value, but different
colors, for both the positive and negative surfaces.

171
Appendix E
Marcus Theory
Marcus Theory is a way to describe electron transfer in dielectric media that is based on
simple electrostatics arguments. Understanding Marcus Theory is key for understanding charge
transfer reactions in condensed phases. Although Marcus Theory explicitly deals with electron
transfer, understanding the concepts behind the theory will help with understanding proton
transfer as well. There are extensions of Marcus Theory that describe proton transfer, but the
reader is encouraged to explore those extensions on their own.

Figure E.1. Solvation of (left) a neutral species and (right) a negatively charged species in a
dielectric medium.

Dielectric media contain dipoles that can orient themselves to counteract electric fields
produced by charged species. In figure E.1 below, the arrangement of the solvent dipoles is
largely unstructured in the presence of a neutral species, as there is little or no electric field to
q = 0 q = -1
Red = negatively charge
Blue = positively charged
172
stabilize. On the right, there is a negative ion in solution. The dipoles orient their positive ends
(blue) towards the charge to counteract the electric field lines. Mathematically, one can think of
the dielectric solvent attenuating the electric potential according to the following equation:
𝛷 =
𝑞
𝜀 𝑟

where 𝜀 is the dielectric constant of the solvent.
Many redox reactions occur spontaneously in solution. This is, in a way, quite surprising!
In vacuum, the energy required to ionize a molecule is typically (100−400) 𝑘
B
𝑇. Since an
electron must leave a molecule before it can go to another, one can think of this as the
activation energy barrier of the vacuum electron transfer reaction. For most redox reactions to
occur spontaneously in vacuum, the temperature would need to be 100x the typical room
temperature – it would need to be 30,000 K or higher – to overcome the activation energy
barrier! This is 5x higher than the surface temperature of the sun. It is the solvation provided by
the dipoles in dielectric media – and, more specifically, the fluctuations in the arrangements of
those dipoles – that make redox reactions possible at room temperature in solution.
Since Marcus Theory exists at the intersection of solvent fluctuations (nuclear motions)
and electron transfer, it is important to understand the relative timescales of these two
phenomena. According to the Born-Oppenheimer Approximation, nuclei move much slower
than electrons, since nuclei are so much heavier than electrons. Therefore, we assume that
electrons are always at equilibrium with a given nuclear configuration. This logic of timescale
separation also gives rise to the Franck-Condon Principle, which states that, because electrons
move so much faster, the nuclear configuration will not change during an electron transfer.
173
So how does an electron transfer reaction in condensed phase occur? The mechanism in
figure E.2 won’t work, because the electrons and the nuclei move on the same timescale, which
is unrealistic according to the discussion above.

Figure E.2. Electronic motion and nuclear motion usually do not occur on the same timescales.
The next guess, then, might be that the electron transfer happens at the initial solvent
structure, and then the solvent slowly stabilizes around the new electron configuration. This is
shown in figure E.3. While this idea is almost correct, if the electron transfer happens while the
nuclear structure is in equilibrium with the initial electron configuration, there will be a
violation of the conservation of energy! In the second picture, the negative parts of the dipole
are pointing at a neutral species and the positive charge isn’t solvated at all – the energy will be
much higher than in the first image!
q = 0 q = -1
q = 0
q = -1
174

Figure E.3. Electron transfer at equilibrium solvent configurations works from an electronic-
nuclear timescale separation perspective, but may violate the conservation of energy.

The trick is the electron transfer must occur at a fixed nuclear configuration where the
energy is about the same before and after the electron transfer. This occurs when the nuclei
fluctuate about their equilibrium positions. In the case where the electron transfer is occurring
between two identical species, the electron transfer will occur when the solvation looks the
same for both the neutral species and the charged species – then the electron won’t have any
preference, since it will be solvated equivalently in either spot! This is shown in figure E.4.
q = 0 q = -1
q = -1
q = 0
q = -1
q = 0
175

Figure E.4. Fluctuation of dipole arrangements in dielectric media can create arrangements
where the energies before and after electron transfer are equal and the electron transfer
happens readily.

These nuclear fluctuations are at the heart of Marcus Theory. The Marcus parabolas are
shown in figure E.5.. Each parabola represents an electronic state. The bottom of each parabola
represents the relevant electronic state when the solvent nuclear geometry is at equilibrium
with it. The parabolas themselves represent changes in energy associated with fluctuations
from that equilibrium. The x-axis of the parabolas is terms of X, a vague solvent coordinate that
describes those fluctuations. The nature of this solvent coordinates will be elucidated in greater
detail when the derive the theory later.
q = 0 q = -1 q = 0
q = -1
q = -1
q = 0
q = 0
q = -1
2
1
3
176

Figure E.5. Marcus parabolas for two electronic states. State a is shown in the blue box at its
equilibrium solvent geometry; state b is shown in the red box at its equilibrium solvent
geometry; and the transition state solvent geometry is shown in the black box.

We can represent the parabolas with the following equations:
𝑊
!
= 𝐸
!
+
1
2
𝐾(𝑋−𝑋
!
)
7

𝑊
F
= 𝐸
F
+
1
2
𝐾(𝑋−𝑋
F
)
7

where 𝑋
!
is the equilibrium solvent coordinate for state a and 𝑋
F
is the equilibrium solvent
coordinate for state b. These functional forms – namely, the quadratic nature of the energy as a
function of solvent configuration and the fact that both parabolas have the same curvature (K)
– will fall out of Marcus Theory when we derive it later.
State a at !
!"
Energy
!
#
!
$
!
!"
State a
State b
State a at !
#
State b at !
$
"
#
"
$
"
!"
"
"
"
%
177
One important thing to point out is the intersection point of the two parabolas. Because
this represents where the two electronic states have the same energy as a function of the
solvent coordinate, this is the transition state where the electron transfer will occur. We refer
to the solvent configuration at this point 𝑋
)H
. To find a functional form for 𝑋
)H
, one can set the
equations for the two parabolas equal and solve for X. This is left to the reader.
One other important concept, which will show up several times, is the reorganization
energy (𝐸
+
) of the electron transfer reaction. This is the energy released by the system if there
was to be a vertical electron transfer (transition from state a to state b at 𝑋
!
) and a subsequent
relaxation of the solvent coordinate. The black arrow in the above figure represents this energy.
We can calculate the reorganization energy by subtracting equilibrium energy of state b from
the energy of state b at 𝑋
!
:
𝐸
+
= 𝑊
F
(𝑋
!
)−𝐸
F

which gives
𝐸
+
=
;
7
𝐾 (𝑋
!
−𝑋
F
)
7

The reorganization energy shows up when we calculate the activation energy (𝐸
6
), the
quantity that will ultimately dictate the rate of the reaction. Based on the parabolas, the
activation energy should be given by the energy at the intersection point minus the equilibrium
energy for state a:
𝐸
6
= 𝑊
!
(𝑋
)H
)− 𝐸
!

The calculation of 𝑋
)H
was mentioned above. Substitution of 𝑋
)H
into this equation yields
178
𝐸
6
=
[(𝐸
!
−𝐸
F
)−𝐸
+
]
7
4𝐸
+

This result is interesting not just because 𝐸
6
is the value we’re after in a kinetic model. It’s also
interesting because, aside from the energy difference of the equilibrium states (𝐸
!
−𝐸
F
), it’s
entirely dependent on the reorganization energy! There is seemingly something fundamentally
important about the solvation energy following a vertical excitation.
Now we should derive the Marcus parabolas and obtain an expression for the
reorganization energy in terms of known quantities. Marcus theory is based on a linear
dielectric continuum model. “Continuum” means that no solvent structure is considered in the
math. We will view the solvent like a goo that can reduce electric potential. Its ability to reduce
the potential is related linearly to the dielectric constant of the solvent – thus, the theory is
“linear”. The electric potential generated by a charged species in the dielectric continuum can
be expressed with the following equation:
Φ =
𝑞
𝜀 𝑟

We make an important assumption about the dielectric response of the solvent in order
to derive Marcus Theory. We assume that the response can be separated into two timescales:
the fast response (𝜺
𝒆
, also called the electronic response or the optical response) and the slow
response (𝜺
𝒔
, also called the static response). The fast component is the result of electrons in
the solvent polarizing in response to the system’s electronic structure change. It is assumed to
be instantaneous. Assume an ion is somehow spontaneously generated in solution. The electric
179
potential immediately following ion generation is attenuated, but only by the electronic
response:
𝛷 =
𝑞
𝜀
@
𝑟

The slow component (which includes the fast component!) is the result of all possible
solvent relaxation mechanisms, but most notably the reorganization of the nuclear coordinates
of the solvent. Remember our earlier discussion about the Born-Oppenheimer approximation
and the Franck-Condon principle: electrons are light and can move very fast, while nuclei are
much heavier and thus move much more slowly. Once enough time has passed, the nuclear
solvation will reach equilibrium and the potential will be fully attenuated by the static response:
𝛷 =
𝑞
𝜀
2
𝑟

It is worth repeating: 𝜺
𝒔
describes the dielectric response at equilibrium.
With these tools in hand, let’s derive Marcus Theory for a simplified system: the
spontaneous generation of an ion in solution. Then, by extension, we will find derive the theory
for electron transfer from one species to another. The equilibrium states of interest are shown
in figure E.6. State a has electronic distribution 𝑞
!
and an equilibrium solvent nuclear
polarization 𝑃
/!
. State b has electronic distribution 𝑞
F
and has an equilibrium solvent nuclear
polarization 𝑃
/F
. We hope to find the activation energy barrier associated with this reaction.
180

Figure E.6: Electronic distributions and nuclear polarizations of states a and b.

As discussed earlier, it is the fluctuations in the nuclear polarization around electronic
state 𝑞
!
that will ultimately determine the activation energy barrier. It is therefore useful to
define a non-equilibrium state, which we will call t(θ), which is a function of the nuclear
polarization coordinate θ. This state has the same electronic distribution as state a (𝑞
!
) but has
a variable, non-equilibrium nuclear polarization (𝑃
/c
). Because the Marcus parabolas represent
the change in energy of the electronic state as a function of nuclear polarization fluctuations,
the energy difference between state a and state t(θ) (∆𝐺
!→)
) should represent the Marcus
parabola for state a.
q = 0
q = -1
State a State b
Nuclear Polarization: !
!"
Electronic distribution: "
#
Nuclear Polarization: !
!#
Electronic distribution: "
"
181

Figure E.7: Electronic distribution and nuclear polarization of state t.

There is one lingering question: what on earth does the “solvent coordinate” keep track
of physically? What does it mean? How do we make sure the solvent coordinate is moving us
towards the transition state, where the electron transfer process can happen? To get around
this confusion, we perform a magic trick: we invent an imaginary state to keep track of solvent
fluctuations from an equilibrium perspective. We call this state θ. It is a state that has a nuclear
polarization (𝑃
/c
) in equilibrium with an imaginary electronic distribution 𝑞
c
that is defined as
𝑞
c
= 𝑞
!
+𝜃(𝑞
F
−𝑞
!
)
As the solvent parameter θ varies from 0 to 1, 𝑞
c
varies linearly between electronic
distributions 𝑞
!
and 𝑞
F
. The corresponding nuclear solvent configurations are shown in figure
E.8.
State t
Electronic distribution: !
!
Nuclear Polarization: "
"#
q = 0
182

Figure E.8: Electronic distribution and nuclear polarization of state 𝜃.

Notice that, since this is an equilibrium state, the nuclear polarization “follows” the
change in charge, with the dipoles gradually becoming more oriented as the charge gets larger.
The equilibrium nuclear polarizations of this varying electronic state essentially keep track of
nuclear fluctuations in the “right direction” for the electron transfer, since the nuclear
polarization at the transition state is expected to be somewhere in-between the equilibrium
nuclear polarizations for states a and b. Now we have a much better handle on the solvent
coordinate.
We can now calculate the relevant quantity (∆𝐺
!→)
) according to the diagram in figure
E.9.
q = 0
q = -1 q = -0.5
!=# !=#.% !=&
183

Figure E.9: The states of interest in Marcus Theory and the mathematical relationship between
them.

q = 0
q = -0.5 q = 0
State !
State "
State #
!"
!→#
=!"
$→#
−!"
$→!
184
∆𝐺
c→)
describes the change in energy due to only the change in electronic state, while keeping
the nuclear polarization th e same (this is the nuclear polarization of state t). ∆𝐺
c→!

describes the change in energy due to both changing the electronic state and changing the
nuclear polarization. Naturally, the difference in ∆𝐺
c→)
and ∆𝐺
c→!
should give only the change
in energy due to changing the nuclear polarization. This difference is therefore equivalent to
∆𝐺
!→)
, the change in energy due to only the nuclear polarization fluctuations around the
equilibrium of state a. This is our Marcus parabola for state a!
A brief note: it might be confusing why the correct equation is ∆𝐺
!→)
= ∆𝐺
c→)
- ∆𝐺
c→!

rather than ∆𝐺
!→)
= ∆𝐺
c→!
- ∆𝐺
c→)
. It is a sign issue that is hopefully clarified by the pictures
and arrows above.
We calculate ∆𝐺
c→)
by integrating the energy change associated with slowly changing
the charge distribution from 𝑞
c
to 𝑞
!
with fixed nuclear coordinates. Our initial electric
potential is given by
𝛷 =
𝑞
!
𝜀
"
𝑎

The value “a” is the radius of the species that is ionized. The use of this value is, admittedly,
conceptually confusing. Since the electron is just “arriving” in our spontaneous ion example and
not going anywhere in particular, the relevant location for the calculation of the potential is at
the surface of the ion. Do not overthink this: this is a non-physical example we’re using to make
the derivation easier, so the math might seem a little non-physical too!
185
The use of 𝜀
2
implies that the nuclear polarization is at equilibrium with 𝑞
c
. We will
change from 𝑞
c
to 𝑞
!
using the following expression:
𝛷(𝜉) =
𝑞
c
𝜀
2
𝑎
+
𝜉
𝜀
@
𝑎

where 𝜉 represents the change in the electronic structure. Notice that the second expression
uses the fast response 𝜀
@
. Because we wish for the nuclear polarization to remain static during
this electronic change, we will only consider the electronic response of the solvation to the
changing electronic structure.
If we integrate from 𝜉 = 0 to 𝜉 = 𝑞
!
−𝑞
c
we accomplish the necessary electronic
structure change. We therefore calculate ∆𝐺
c→)
with the following integral:
∆𝐺
c→)
= 𝑑𝜉 𝛷(𝜉)
^
"
'^
5
:
=
𝑞
c
(𝑞
!
−𝑞
c
)
𝜀
2
𝑎
+
(𝑞
!
−𝑞
c
)
7
2𝜀
@
𝑎

Note that electric potential can be expressed in joules/coulomb. Since 𝜉 has units of coulomb,
the integrand has units of joules and so should our final answer. We can substitute in our
earlier expression (𝑞
c
= 𝑞
!
+𝜃(𝑞
F
−𝑞
!
)) to get
∆𝐺
c→)
=
𝑞
!
(𝑞
!
−𝑞
F
)
𝜀
2
𝑎
𝜃+
(𝑞
!
−𝑞
F
)
7
𝑎
(
1
2𝜀
@
−
1
𝜀
2
) 𝜃
7

Next we will calculate ∆𝐺
c→!
. This is actually easier to calculate since both the initial
and final states are at equilibrium. We simply integrate the electric potential
𝛷 =
𝑞
c
e
𝜀
2
𝑎

186
with respect to 𝑞
c
e
from 𝑞
c
to 𝑞
!
. We use 𝜀
2
only because the nuclear polarization should be in
equilibrium with the charge as it slowly changes. The integral looks like this:
∆𝐺
c→!
= 𝑑𝑞
c
e
𝛷
^
"
^
5
=
𝑞
!
7
2𝜀
2
𝑎
−
𝑞
c
7
2𝜀
2
𝑎

Substituting in 𝑞
c
= 𝑞
!
+𝜃(𝑞
F
−𝑞
!
) gives us
∆𝐺
c→!
=
^
"
(^
"
'^
6
)
f
,
!
𝜃−
(^
"
'^
6
)
$
7f
,
!
𝜃
7

Now that we have successfully calculated both ∆𝐺
c→)
and ∆𝐺
c→!
, we can calculate
∆𝐺
!→)
. After simplifying, we get
∆𝐺
!→)
= 𝐸
!
+
(𝑞
!
−𝑞
F
)
7
2𝑎
U
1
𝜀
@
−
1
𝜀
2
V𝜃
7

We know that the lowest energy (when 𝜃 = 0) should be the equilibrium energy of state a (𝐸
!
),
so we add it here as a vertical offset. This is mathematically sound because the integration
steps above should have introduced constants (which we ignored for simplicity).
Voila! It’s the Marcus Theory parabola for state a in all of its glory. It’s worth
remembering that 𝜃 goes from 0 to 1 and walks us linearly from the equilibrium nuclear
polarization of state a to the nuclear polarization of state b. That means if we wanted to
describe the fluctuations in nuclear polarization around state b, we could just walk 𝜃 from 1 to
0 instead! We should also switch the sign of 𝑞
!
−𝑞
F
, since we’re now moving the electronic
structure in the opposite direction - but since the 𝑞
!
−𝑞
F
is squared, we don’t need to worry
about it. Therefore, the Marcus Theory parabola for state b is
187
∆𝐺
F→)
= 𝐸
F
+
(𝑞
!
−𝑞
F
)
7
2𝑎
(
1
𝜀
@
−
1
𝜀
2
)(1−𝜃)
7

where the vertical offset is now the equilibrium energy of state b (𝐸
F
). I mentioned early on in
this appendix that the quadratic dependence on the nuclear configuration coordinate would fall
out of the theory, as would the fact that both parabolas have the same curvature. You can see
now that they do.
A more general expression of the Marcus Theory parabola – for transfer of any quantity
of charge between two species of any size and at any distance – can be developed using the
exact same logic we applied above (only the math is a bit more unwieldy!):
∆𝐺
!→)
= 𝐸
!
+(
1
𝜀
@
−
1
𝜀
2
)(
1
2𝑅
!
+
1
2𝑅
F
−
1
𝑅
!F
)𝛥𝑞
7
𝜃
7

∆𝐺
F→)
= 𝐸
F
+(
1
𝜀
@
−
1
𝜀
2
)(
1
2𝑅
!
+
1
2𝑅
F
−
1
𝑅
!F
)𝛥𝑞
7
(1−𝜃)
7

I won’t go into the details of the derivation, as it would be highly redundant, but we can briefly
discuss the differences. Because we now have two species to consider, we need to include both
of their radiuses. Hence the inclusion of both 𝑅
!
and 𝑅
F
. The distance between the species will
naturally play a role, since the electron transfer should be easier when the two species are
closer. Therefore, 𝑅
!F
is introduced. We change 𝑞
!
−𝑞
F
to 𝛥𝑞 to allow for the possibility of a
multi-electron transfer.
Don’t forget! We still have one more step! We must now use the parabola to calculate
the activation energy. That’s the value that we can use to predict the electron transfer kinetics,
188
which is the point of all of this! We can easily obtain it through analogy to the equations we
derived earlier from the Marcus parabola. Our original equation for the Marcus parabola was
𝑊
g
= E
g
+
1
2
K(X−X
g
)
7

and the equation for the reorganization energy was
𝐸
+
=
;
7
𝐾 (𝑋
!
−𝑋
F
)
7

Comparing these to our new expressions, and remembering that 𝜃 goes from 0 in state a to 1 in
state b, we get
𝐸
+
= U
1
𝜀
@
−
1
𝜀
2
VU
1
2𝑅
!
+
1
2𝑅
F
−
1
𝑅
!F
V∆𝑞
7
(0−1)
7

𝐸
+
= U
1
𝜀
@
−
1
𝜀
2
VU
1
2𝑅
!
+
1
2𝑅
F
−
1
𝑅
!F
V∆𝑞
7

Now, you’re officially ready. If you know the equilibrium energies of states a and b (which you
obtain from experiment or quantum chemistry calculations) and you’ve used the above
equation to calculate the reorganization energy, just plug those values into
𝐸
6
=
[(𝐸
!
−𝐸
F
)−𝐸
+
]
7
4𝐸
+

and you’re good to go!
The rate can be calculated with 𝐸
6
using transition state theory
𝑘
@)
= 𝐴 𝑒
'
*
7
(
8
,

where the form of A will depend on how strong the coupling is between your electronic states.
189
References

(1) Munitz, N.; Avital, Y.; Pines, D.; Nibbering, E. T. J.; Pines, E. Cation-Enhanced
Deprotonation of Water by a Strong Photobase. Isr J Chem 2009, 49, 261.

(2) Prémont-Schwarz, M.; Barak, T.; Pines, D.; Nibbering, E. T. J.; Pines, E. Ultrafast Excited-
State Proton-Transfer Reaction of 1-Naphthol-3,6-Disulfonate and Several 5-Substituted 1-
Naphthol Derivatives. J. Phys. Chem. B 2013, 117 (16), 4594–4603.
https://doi.org/10.1021/jp308746x.

(3) Agmon, N. Elementary Steps in Excited-State Proton Transfer. J. Phys. Chem. A 2005, 109
(1), 13–35. https://doi.org/10.1021/jp047465m.

(4) Gutman, M.; Huppert, D.; Pines, E. The PH Jump: A Rapid Modulation of PH of Aqueous
Solutions by a Laser Pulse. J. Am. Chem. Soc. 1981, 103 (13), 3709–3713.
https://doi.org/10.1021/ja00403a016.

(5) Pines, E.; Huppert, D. PH Jump: A Relaxational Approach. J. Phys. Chem. 1983, 87 (22),
4471–4478. https://doi.org/10.1021/j100245a029.

(6) Spry, D. B.; Goun, A.; Fayer, M. D. Deprotonation Dynamics and Stokes Shift of Pyranine
(HPTS). J. Phys. Chem. A 2007, 111 (2), 230–237. https://doi.org/10.1021/jp066041k.

(7) Spry, D. B.; Fayer, M. D. Charge Redistribution and Photoacidity: Neutral versus Cationic
Photoacids. J. Chem. Phys. 2008, 128 (8), 084508. https://doi.org/10.1063/1.2825297.

(8) Förster, Th. Elektrolytische Dissoziation Angeregter Moleküle. Z. Für Elektrochem. Angew.
Phys. Chem. 1950, 54 (1), 42–46. https://doi.org/10.1002/bbpc.19500540111.

(9) Weller, A. Fast Reactions of Excited Molecules. Prog React Kinet Mech 1961, 1, 187.

(10) Tolbert, L. M.; Solntsev, K. M. Excited-State Proton Transfer: From Constrained Systems to
“Super” Photoacids to Superfast Proton Transfer. Acc Chem Res 2002, 35 (1), 19.

(11) Simkovitch, R.; Akulov, K.; Shomer, S.; Roth, M. E.; Shabat, D.; Schwartz, T.; Huppert, D.
Comprehensive Study of Ultrafast Excited-State Proton Transfer in Water and D2O
Providing the Missing RO(−)·H(+) Ion-Pair Fingerprint. J Phys Chem A 2014, 118, 4425.

190
(12) Simkovitch, R.; Shomer, S.; Gepshtein, R.; Huppert, D. How Fast Can a Proton-Transfer
Reaction Be beyond the Solvent-Control Limit? J Phys Chem B 2015, 119, 2253.

(13) Pines, D.; Nibbering, E. T. J.; Pines, E. Monitoring the Microscopic Molecular Mechanisms
of Proton Transfer in Acid-Base Reactions in Aqueous Solutions. Isr J Chem 2015, 55, 1240.

(14) Haghighat, S.; Ostresh, S.; Dawlaty, J. M. Controlling Proton Conductivity with Light: A
Scheme Based on Photoacid Doping of Materials. J. Phys. Chem. B 2016, 120 (5), 1002–
1007. https://doi.org/10.1021/acs.jpcb.6b00370.

(15) White, W.; Sanborn, C. D.; Reiter, R. S.; Fabian, D. M.; Ardo, S. Observation of Photovoltaic
Action from Photoacid-Modified Nafion Due to Light-Driven Ion Transport. J. Am. Chem.
Soc. 2017, 139 (34), 11726–11733. https://doi.org/10.1021/jacs.7b00974.

(16) Abbruzzetti, S.; Crema, E.; Masino, L.; Vecli, A.; Viappiani, C.; Small, J. R.; Libertini, L. J.;
Small, E. W. Fast Events in Protein Folding: Structural Volume Changes Accompanying the
Early Events in the N→I Transition of Apomyoglobin Induced by Ultrafast PH Jump.
Biophys. J. 2000, 78 (1), 405–415. https://doi.org/10.1016/S0006-3495(00)76603-3.

(17) Causgrove, T. P.; Dyer, R. B. Nonequilibrium Protein Folding Dynamics: Laser-Induced PH-
Jump Studies of the Helix–Coil Transition. Chem. Phys. 2006, 323 (1), 2–10.
https://doi.org/10.1016/j.chemphys.2005.08.032.

(18) Dempsey, J. L.; Winkler, J. R.; Gray, H. B. Mechanism of H2 Evolution from a
Photogenerated Hydridocobaloxime. J. Am. Chem. Soc. 2010, 132 (47), 16774–16776.
https://doi.org/10.1021/ja109351h.

(19) Das, A.; Ayad, S.; Hanson, K. Enantioselective Protonation of Silyl Enol Ether Using Excited
State Proton Transfer Dyes. Org. Lett. 2016, 18 (20), 5416–5419.
https://doi.org/10.1021/acs.orglett.6b02820.

(20) Keitz, B. K.; Grubbs, R. H. A Tandem Approach to Photoactivated Olefin Metathesis:
Combining a Photoacid Generator with an Acid Activated Catalyst. J. Am. Chem. Soc. 2009,
131 (6), 2038–2039. https://doi.org/10.1021/ja807187u.

(21) Kohse, S.; Neubauer, A.; Pazidis, A.; Lochbrunner, S.; Kragl, U. Photoswitching of Enzyme
Activity by Laser-Induced PH-Jump. J. Am. Chem. Soc. 2013, 135 (25), 9407–9411.
https://doi.org/10.1021/ja400700x.

191
(22) Peretz-Soroka, H.; Pevzner, A.; Davidi, G.; Naddaka, V.; Kwiat, M.; Huppert, D.; Patolsky, F.
Manipulating and Monitoring On-Surface Biological Reactions by Light-Triggered Local PH
Alterations. Nano Lett. 2015, 15 (7), 4758–4768.
https://doi.org/10.1021/acs.nanolett.5b01578.

(23) Favaro, G.; Mazzucato, U.; Masetti, F. Excited State Reactivity of Aza Aromatics. I. Basicity
of 3-Styrylpyridines in the First Excited Singlet State. J Phys Chem 1973, 77, 601.

(24) Mataga, N.; Kaifu, Y.; Koizumi, M. On the Base Strength of Some Nitrogen Heterocycles in
the Excited State. Bull Chem Soc Jpn 1956, 29, 373.

(25) Pines, E.; Huppert, D.; Gutman, M.; Nachliel, N.; Fishman, M. The POH Jump:
Determination of Deprotonation Rates of Water by 6-Methoxyquinoline and Acridine. J
Phys Chem 1986, 90, 6366.

(26) Poizat, O.; Bardez, E.; Buntinx, G.; Alain, V. Picosecond Dynamics of the Photoexcited 6-
Methoxyquinoline and 6-Hydroxyquinoline Molecules in Solution. J Phys Chem A 2004,
108, 1873.

(27) Driscoll, E. W.; Hunt, J. R.; Dawlaty, J. M. Photobasicity in Quinolines: Origin and Tunability
via the Substituents’ Hammett Parameters. J Phys Chem Lett 2016, 7, 2093.

(28) Driscoll, E. W.; Hunt, J. R.; Dawlaty, J. M. Proton Capture Dynamics in Quinoline
Photobases: Substituent Effect and Involvement of Triplet States. J Phys Chem A 2017,
121, 7099.

(29) Kellmann, A.; Lion, Y. Acid-Base Equilibria of the Excited Singlet and Triplet States and the
Semi-Reduced Form of Acridine Orange. Photochem Photobiol 1979, 29, 217.

(30) Nachliel, E.; Ophir, Z.; Gutman, M. Kinetic Analysis of Fast Alkalinization Transient by
Photoexcited Heterocyclic Compounds: POH Jump. J Am Chem Soc 1987, 109, 1342.

(31) Ryan, E. T.; Xiang, T.; Johnston, K. P.; Fox, M. A. Absorption and Fluorescence Studies of
Acridine in Subcritical and Supercritical Water. J Phys Chem A 1997, 101, 1827.

(32) Eisenhart, T. T.; Dempsey, J. L. Photo-Induced Proton-Coupled Electron Transfer Reactions
of Acridine Orange: Comprehensive Spectral and Kinetics Analysis I. Experimental
Methods. J Am Chem Soc 2014, 136, 12221.
192

(33) Vogt, S.; Schulman, S. G. Reversible Proton Transfer in Photoexcited Xanthone. Chem Phys
Lett 1983, 97, 450.

(34) Akulov, K.; Simkovitch, R.; Erez, Y.; Gepshtein, R.; Schwartz, T.; Huppert, D. Acid Effect on
Photobase Properties of Curcumin. J Phys Chem A 2014, 118, 2470.

(35) Sheng, W.; Nairat, M.; Pawlaczyk, P. D.; Mroczka, E.; Farris, B.; Pines, E.; Geiger, J. H.;
Borhan, B.; Dantus, M. Ultrafast Dynamics of a ”super” Photobase. Angew Chem Int Ed
2018, 57, 14742.

(36) Driscoll, E.; Sorenson, S.; Dawlaty, J. M. Ultrafast Intramolecular Electron and Proton
Transfer in Bis(Imino)Isoindole Derivatives. J. Phys. Chem. A 2015, 119 (22), 5618–5625.
https://doi.org/10.1021/acs.jpca.5b02889.

(37) Hunt, J. R.; Dawlaty, J. M. Photodriven Deprotonation of Alcohols by a Quinoline
Photobase. J. Phys. Chem. A 2018, 122 (40), 7931–7940.
https://doi.org/10.1021/acs.jpca.8b06152.

(38) Hunt, J. R.; Tseng, C.; Dawlaty, J. M. Donor–Acceptor Preassociation, Excited State
Solvation Threshold, and Optical Energy Cost as Challenges in Chemical Applications of
Photobases. Faraday Discuss. 2019, 216 (0), 252–268.
https://doi.org/10.1039/C8FD00215K.

(39) Hunt, J. R.; Dawlaty, J. M. Kinetic Evidence for the Necessity of Two Proton Donor
Molecules for Successful Excited State Proton Transfer by a Photobase. J. Phys. Chem. A
2019, 123 (48), 10372–10380. https://doi.org/10.1021/acs.jpca.9b08970.

(40) Demianets, I.; Hunt, J. R.; Dawlaty, J. M.; Williams, T. J. Optical PKa Control in a
Bifunctional Iridium Complex. Organometallics 2019, 38, 200.

(41) Olah, G. A.; Goeppert, A.; Prakash, G. S. Beyond Oil and Gas: The Methanol Economy;
2007.

(42) Rafiee, M.; Miles, K. C.; Stahl, S. S. Electrocatalytic Alcohol Oxidation with TEMPO and
Bicyclic Nitroxyl Derivatives: Driving Force Trumps Steric Effects. J Am Chem Soc 2015, 137,
14751.

193
(43) Waldie, K. M.; Flajslik, K. R.; McLoughlin, E.; Chidsey, C. E. D.; Waymouth, R. M.
Electrocatalytic Alcohol Oxidation with Ruthenium Transfer Hydrogenation Catalysts. J Am
Chem Soc 2017, 139, 738.

(44) Vajda, Š.; Jimenez, R.; Rosenthal, S. J.; Fidler, V.; Fleming, G. R.; Castner, E. W.
Femtosecond to Nanosecond Solvation Dynamics in Pure Water and inside the γ-
Cyclodextrin Cavity. J Chem Soc Faraday Trans 1995, 91, 867.

(45) Horng, M. L.; Gardecki, J. A.; Papazyan, A.; Maroncelli, M. Subpicosecond Measurements
of Polar Solvation Dynamics: Coumarin 153 Revisited. J Phys Chem 1995, 99, 17311.

(46) Tominaga, K.; Walker, G. C. Femtosecond Experiments on Solvation Dynamics of an
Anionic Probe Molecule in Methanol. J Photochem Photobiol A 1995, 87, 127.

(47) Anslyn, E. V.; Dougherty, D. A. Modern Physical Organic Chemistry; 2006.

(48) Roccatano, D.; Colombo, G.; Fioroni, M.; Mark, A. E. Mechanism by Which 2, 2, 2-
Trifluoroethanol/Water Mixtures Stabilize Secondary-Structure Formation in Peptides: A
Molecular Dynamics Study. Proc Natl Acad Sci U A 2002, 99, 12179.

(49) Park, S.-Y.; Lee, Y. M.; Kwac, K.; Jung, Y.; Kwon, O.-H. Alcohol Dimer Is Requisite to Form an
Alkyl Oxonium Ion in the Proton Transfer of a Strong (Photo) Acid to Alcohol. Chem - Eur J
2016, 22, 4340.

(50) Wahl, M. Time-Correlated Single Photon Counting; 2014.

(51) Furniss, B. S. Vogel’s Textbook of Practical Organic Chemistry; 1989.

(52) Chitra, R.; Smith, P. E. Properties of 2,2,2-Trifluoroethanol and Water Mixtures. J. Chem.
Phys. 2001, 114 (1), 426–435. https://doi.org/10.1063/1.1330577.

(53) O’Reilly, D. E.; Peterson, E. M.; Yasaitis, E. L. Self-Diffusion Coefficients and Rotational
Correlation Times in Polar Liquids. IV. Dichloromethane and Pyridine. J Chem Phys 1972,
57, 890.

(54) Vos, J. G. Excited-State Acid-Base Properties of Inorganic Compounds. Polyhedron 1992, 11
(18), 2285.

194
(55) Peterson, S. H.; Demas, J. N. Excited State Acid-Base Reactions of Transition Metal
Complexes: Dicyanobis(2,2ʹ-Bipyridine) Ruthenium(III) in Aqueous Acid. J Am Chem Soc
1976, 98, 7880.

(56) Cargill Thompson, A. M. W.; Smailes, M. C. C.; Je, J. C.; Ward, M. D. Ruthenium Tris-
(Bipyridyl) Complexes with Pendant Protonable and Deprotonable Moieties: PH Sensitivity
of Electronic Spectral and Luminescence Properties. J Chem Soc Dalton Trans 1997, 5, 737.

(57) Lay, P. A.; Sasse, W. H. F. Proton Transfer in the Excited State of Carboxylic Acid Derivatives
of Tris(2,2’-Bipyridine-N,N’)Ruthenium(II). Inorg Chem 1984, 23 (12), 4123.

(58) Nazeeruddin, M. K.; Kalyanasundaram, K. Acid-Base Behavior in the Ground and Excited
States of Ruthenium(II) Complexes Containing Tetraimines or Dicarboxybipyridines as
Protonatable Ligands. Inorg Chem 1989, 28 (23), 4251.

(59) Giordano, P. J.; Bock, O.; Wrighton, M. S. Excited State Proton Transfer of Ruthenium(II)
Complexes of 4,7-Dihydroxy- 1,10-Phenanthroline. Increased Acidity in the Excited State. J
Am Chem Soc 1978, 100 (22), 6960.

(60) O’Donnell, R. M.; Sampaio, R. N.; Li, G.; Johansson, P. G.; Ward, C. L.; Meyer, G. J.
Photoacidic and Photobasic Behavior of Transition Metal Compounds with Carboxylic Acid
Group(S). J Am Chem Soc 2016, 138 (11), 3891.

(61) Leasure, R. M.; Sacksteder, L. A.; Reitz, G. A.; Demas, J. N.; Nesselrodt, D.; DeGraff, B. A.
Excited-State Acid-Base Chemistry of (α-Diimine)Cyanotricarbonyl Rhenium(I) Complexes.
Inorg Chem 1991, 30 (19), 3722.

(62) Maity, D.; Mardanya, S.; Karmakar, S.; Baitalik, S. PH-Induced Processes in Wire-like
Multichromophoric Homo- and Heterotrimetallic Complexes of Fe(Ii), Ru(Ii), and Os(Ii).
Dalt Trans 2015, 44 (21), 10048.

(63) Browne, W. R.; O’Connor, C. M.; Hughes, H. P.; Hage, R.; Walter, O.; Doering, M.;
Gallagher, J. F.; Vos, J. G. Ruthenium(Ii) and Osmium(Ii) Polypyridyl Complexes of an
Asymmetric Pyrazinyl- and Pyridinyl-Containing 1,2,4-Triazole Based Ligand. Connectivity
and Physical Properties of Mononuclear Complexes. J Chem Soc Dalt Trans 2002, 0 (21),
4048.

195
(64) Das, S.; Saha, D.; Mardanya, S.; Baitalik, S. A Combined Experimental and DFT/TDDFT
Investigation of Structural, Electronic, and PH-Induced Tuning of Photophysical and Redox
Properties of Osmium(Ii) Mixed-Chelates Derived from Imidazole-4,5-Dicarboxylic Acid and
2,2ʹ-Bipyridine. Dalt Trans 2012, 41 (39), 12296.

(65) Cummings, S. D.; Eisenberg, R. Acid-Base Behavior of the Ground and Excited States of
Platinum(II) Complexes of Quinoxaline-2,3-Dithiolate. Inorg Chem 1995, 34 (13), 3396.

(66) Leavens, B. B. H.; Trindle, C. O.; Sabat, M.; Altun, Z.; Demas, J. N.; DeGraff, B. A.
Photophysical and Analyte Sensing Properties of Cyclometalated Ir(III) Complexes. J
Fluoresc 2012, 22 (1), 163.

(67) Maity, D.; Bhaumik, C.; Karmakar, S.; Baitalik, S. Photoinduced Electron and Energy
Transfer and PH-Induced Modulation of the Photophysical Properties in Homo- and
Heterobimetallic Complexes of Ruthenium(II) and Rhodium(III) Based on a Heteroditopic
Phenanthroline–Terpyridine Bridge. Inorg Chem 2013, 52 (14), 7933.

(68) Kasha, M. Characterization of Electronic Transitions in Complex Molecules. Discuss
Faraday Soc 1950, 9, 14.

(69) El-Sayed, M. A. Spin-Orbit Coupling and the Radiationless Processes in Nitrogen
Heterocyclics. J Chem Phys 1963, 38, 2834.

(70) Hadley, S. G. Direct Determination of Singlet → Triplet Intersystem Crossing Quantum
Yield. II. Quinoline, Isoquinoline, and Quinoxaline. J Phys Chem 1971, 75, 2083.

(71) Komura, A.; Uchida, K.; Yagi, M.; Higuchi, J. Electron Spin Resonance and Phosphorescence
of Quinoline, Isoquinoline and Their Protonated Cations in the Phosphorescent Triplet
States. J Photochem Photobiol A 1988, 42, 293.

(72) Middleton, W. J.; Llndsey, R. V. Hydrogen Bonding in Fluoro Alcohols. J Am Chem Soc 1964,
86 (22), 4948.

(73) Fordyce, W. A.; Crosby, G. A. Electronic Spectroscopy of N-Heterocyclic Complexes of
Rhodium(I) and Iridium(I). Inorg Chem 1982, 21 (3), 1023.

196
(74) Roy, S.; Ardo, S.; Furche, F. 5-Methoxyquinoline Photobasicity Is Mediated by Water
Oxidation. J. Phys. Chem. A 2019, 123 (31), 6645–6651.
https://doi.org/10.1021/acs.jpca.9b05341.

(75) Alamudun, S. F.; Tanovitz, K.; Espinosa, L.; Fajardo, A.; Galvan, J.; Petit, A. S. Structure–
Photochemical Function Relationships in the Photobasicity of Aromatic Heterocycles
Containing Multiple Ring Nitrogen Atoms. J. Phys. Chem. A 2021, 125 (1), 13–24.
https://doi.org/10.1021/acs.jpca.0c07013.

(76) Lahiri, J.; Moemeni, M.; Kline, J.; Borhan, B.; Magoulas, I.; Yuwono, S. H.; Piecuch, P.;
Jackson, J. E.; Dantus, M.; Blanchard, G. J. Proton Abstraction Mediates Interactions
between the Super Photobase FR0-SB and Surrounding Alcohol Solvent. J. Phys. Chem. B
2019, 123 (40), 8448–8456. https://doi.org/10.1021/acs.jpcb.9b06580.

(77) Lahiri, J.; Moemeni, M.; Magoulas, I.; Yuwono, S. H.; Kline, J.; Borhan, B.; Piecuch, P.;
Jackson, J. E.; Blanchard, G. J.; Dantus, M. Steric Effects in Light-Induced Solvent Proton
Abstraction. Phys. Chem. Chem. Phys. 2020, 22 (35), 19613–19622.
https://doi.org/10.1039/D0CP03037F.

(78) Lahiri, J.; Moemeni, M.; Kline, J.; Magoulas, I.; Yuwono, S. H.; Laboe, M.; Shen, J.; Borhan,
B.; Piecuch, P.; Jackson, J. E.; Blanchard, G. J.; Dantus, M. Isoenergetic Two-Photon
Excitation Enhances Solvent-to-Solute Excited-State Proton Transfer. J. Chem. Phys. 2020,
153 (22), 224301. https://doi.org/10.1063/5.0020282.

(79) Raulin, F.; Bloch, S.; Toupance, G. Addition Reactions of Malonic Nitriles with Alkanethiol in
Aqueous Solution. Orig. Life 1977, 8 (3), 247–257. https://doi.org/10.1007/BF00930686.

(80) Bernasconi, C. F.; Bunnell, R. D. Free Energies of Transfer of Carbon Acids and Their
Conjugate Carbanions from Water to Dimethyl Sulfoxide-Water Mixtures. J. Am. Chem.
Soc. 1988, 110 (9), 2900–2903. https://doi.org/10.1021/ja00217a034.

(81) Sarkar, S.; Maitra, A.; Banerjee, S.; Thoi, V. S.; Dawlaty, J. M. Electric Fields at Metal–
Surfactant Interfaces: A Combined Vibrational Spectroscopy and Capacitance Study. J.
Phys. Chem. B 2020, 124 (7), 1311–1321. https://doi.org/10.1021/acs.jpcb.0c00560.

Abstract (if available)

Abstract Photobases are molecules that convert light to proton transfer drive and therefore have potential applications in many areas of chemistry. Quinoline and its substituted analogs are examples of photobase molecules. I discuss the ability of 5-methoxyquinoline to deprotonate a series of alcohols upon excitation by light. I report both the thermodynamic limits and the relevant kinetics for this process using steady state and transient absorption spectroscopies, respectively. A correlation is shown between the thermodynamic drive for the reaction and the proton transfer timescale. Next, I investigate the effect of proton donor concentration on ground-state hydrogen bonding and the excited-state proton transfer reaction of 5-methoxyquinoline. It is shown that a large excess of proton donor is needed to see ground state hydrogen bonding and an even larger excess is necessary to see excited state proton transfer. The mechanism for this concentration-dependence is investigated with kinetic studies using transient absorption spectroscopy and TCSPC. The results suggest that multiple proton donor molecules are necessary to solvate the excited state proton transfer reaction. I then discuss an iridium complex with a covalently attached quinoline that shows the ability to capture protons independent of the electronics of the complex. Finally, I investigate the electronic origins of the quinoline photobase phenomenon using electronic structure calculations. A convincing correlation between changes in electron density on the nitrogen heteroatom and experimentally determined changes in pKa is seen. I use this correlation to predict the photobasicity of a variety of heteroatomic aromatic molecules. These results are necessary fundamental steps towards applying photobases in potential applications, such as deprotonation of alcohols for catalytic and synthetic purposes, optical regulation of pH, and transfer of protons in redox reactions.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

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

PDF

Proton kinetics in electrochemistry: new directions and mechanistic analysis

PDF

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

PDF

Capturing the sun: leveraging excited state dynamics

PDF

The effect of microhydration on ionization energy and proton transfer in nucleobases: analysis and method development

PDF

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

PDF

Unraveling photodissociation pathways in pyruvic acid and the role of methylhydroxycarbene

PDF

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

PDF

Water partitioning between the bulk and an electrode surface

PDF

Computational spectroscopy in gas and condensed phases

PDF

Development of robust ab initio methods for description of excited states and autoionizing resonances

PDF

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

PDF

Photoinduced redox reactions in biologically relevant systems

PDF

Excited state process in perovskites

PDF

Microbial electron transfer and biogenic nanomaterials: experiments, theory, and simulations at the bacterial-inorganic interface

PDF

Otopetrin-1, a proton selective ion channel in taste cells

PDF

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

PDF

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

PDF

Development of new bifunctional iridium complexes for hydrogenation and dehydrogenation reactions

Asset Metadata

Creator Hunt, Jonathan Ryan (author)

Core Title Excited state proton transfer in quinoline photobases

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Degree Conferral Date 2021-12

Publication Date 11/03/2021

Defense Date 09/23/2021

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

Tag absorption,acid,base,DFT,electronic spectroscopy,electronic structure theory,emission,excited state,fluorescence,kinetics,OAI-PMH Harvest,photobase,proton transfer,quinoline,ultrafast spectroscopy,UV-Vis

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Dawlaty, Jahan Monsoor (committee chair), Benderskii, Alexander (committee member), El-Naggar, Moh (committee member)

Creator Email jonathanryanhunt@gmail.com,jonathrh@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC16351277

Unique identifier UC16351277

Legacy Identifier etd-HuntJonath-10199

Document Type Dissertation

Format application/pdf (imt)

Rights Hunt, Jonathan Ryan

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

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

Repository Email cisadmin@lib.usc.edu