SYNTHETIC MODIFICATION OF SPIN DENSITY IN PHENANTHRENESEMIQUINONES,
ZN(II)-PHENANTHRENESEMIQUINONE AND NI(II)-PHENANTHRENESEMIQUINONE
COMPLEXES
By
Larry D. Morris III

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Chemistry—Doctor of Philosophy
2017

ABSTRACT
SYNTHETIC MODIFICATION OF SPIN DENSITY IN PHENANTHRENESEMIQUINONES,
ZN(II)-PHENANTHRENESEMIQUINONE AND NI(II)-PHENANTHRENESEMIQUINONE
COMPLEXES
By
Larry D. Morris III
Contained within this body of work is a study that aims to quantify spin density delocalization
in 3,6-R2-phenanthrenesemiquinone (3,6-R2-PSQ, where R = OMe and H) radicals and their metal
complexes (where metals=ZnII, GaIII and NiII), of the form [M(tren)(PSQ)](BPh4)n, where M is a
metal that is listed above, tren = tris(2-aminoethyl)amine, PSQ = phenanthrenesemiquinone and
BPh4 = tetraphenylborate. Spin delocalization will be scrutinized by analyzing carbonyl carbon
hyperfine coupling constants by continuous wave electron paramagnetic resonance (CW-EPR)
spectroscopy, electron spin echo envelope modulation (ESEEM) spectroscopy, and hyperfine
sublevel correlation (HYSCORE) spectroscopy. These examinations will reveal the influence
molecular structure has on the hyperfine coupling at the carbonyl carbon nuclei.

Spin

delocalization has been quantified by isotopically labeling the carbonyl carbon nuclei with carbon13 nuclei.

13

C-labelled quinones were synthesized from a sequence of reactions, beginning with

C-α-N,N’-dimethylformamide and 2,2’-dibromobiphenyl to form a diformyl-biphenyl. The

13

carbonyl carbons were then coupled with hydrazine via a reductive cyclisation reaction, followed
by oxidation with CrO3 to afford the quinone. CW-EPR of the 12C isotopologues revealed 1H fine
structure in the organic radicals and the ZnII-PSQ complex.

Inspection of the

13

C-labeled

compounds provided isotropic hyperfine coupling constants for all compounds studied except for
NiII complexes; any fine structure from NiII CW-EPR spectra was unable to be resolved. The
magnitude of all isotropic coupling constants fell within the range of 0 – 3 MHz. ESEEM and

HYSCORE spectroscopies allowed for scrupulous investigation of the dipolar coupling. Carbonyl
carbon dipolar couplings varied across the spectrum of compounds examined. This study shows
promise for the use of ESEEM and especially HYSCORE spectroscopies for examining the spin
properties of molecules in the field of molecular magnetic materials.

Copyright by
LARRY D. MORRIS III
2017

This dissertation is dedicated to my wife, Melissa, my daughter Arlyn Mae, and my son, Ronan.

v

ACKNOWLEDGEMENTS

First and foremost, I would like to acknowledge and thank my advisor, Professor James K.
McCusker, without whom, none of the research in this dissertation would have been accomplished.
I initially made my decision to apply to graduate school to Michigan State intending to work for
Jim; from the beginning I knew that I wanted to work on inorganic systems that exhibited
interesting magnetic properties. I have learned much through Jim’s example. There is absolutely
no other person in the scientific community that I have met that is able to recall the amount of
information about any given subject and/or person within chemistry or physics in the same manner
in which Jim is able. Due to this, I take great joy in reading scientific literature, not only for the
scientific content, but to appreciate what others are able to do in science and to give them credit
when talking about their scientific accomplishments. He has also become somewhat of a scientific
“Jiminy Cricket” to me; if I ever have an inquiry as to what should be done or what should be
considered, I need not wonder long before I know what Jim’s answer would be. I have also learned
(and this is also an ongoing process) how to be simple and precise when talking about science to
people, and this is a critical skill for understanding anything. Jim is charismatic, charming and
simply put, an entertaining person to pass the time with. What else can I say but, “SCIENCE!”
Secondly, I would like to thank Professor John McCracken. I have spent so many hours
learning about EPR and how to perform EPR experiments with John. Although I cannot speak to
all the players in the EPR community, I could not have picked a better person to learn from.
Working and learning from John has been one of the greatest serendipitous occurrences in my life.
John delights in teaching, and no matter the ability of the student, he will go great lengths to play
his part in the student’s learning process.

vi

Next, I would like to thank the many members of the McCusker group and chemistry
department who have contributed to my development as a scientist. First of which, I would like
to thank Dr. John “Drew” Kouzelos (or “Johnny OC”, whichever you prefer). I have gained a
great knowledge about magnetism, exchange coupling, DFT, performing calculations, and
anything electrical or mechanical in nature from Drew. Also, there is no one that can have a better
nonsensical argument better than Drew, it can be frustrating at times, but this stems from his thirst
for knowledge and intrigue for how people think. Thank you to Dr. Eileen (Dixon) Foszcz for
being a dear friend to my wife and me, and for putting up with questions about chirp correction
and the other cornucopia (there you go Johnny!) of laser questions I’ve ever had and will ever
have. Big thanks are indebted to Dr. Dong Guo and Dr. Lindsey Jamula Kvolchik for teaching me
sound chemistry practices, habits and lab safety. Thank you to Dr. Lisa Harlow for being a great
sounding board and for your blunt honesty, it is always appreciated! Thank you to Dr. Kate
McCusker for showing me what a good teacher looks like; anytime I am teaching, I always
consider, “how would Kate approach this?” I would like to thank Dani and Selene for being a
great example of working your butt off, all of the time, and just being a constant force in the lab,
both scientific and comedic. And of course I would like to thank the LASER ladies! I will always
hold Monica’s knowledge of cutting edge of music higher than mine and continue to pick her brain
for new music ideas. Jenny Miller, the heartbeat of the McCusker group, thank you for being
awesome, a great example of a good, wholesome person, and a careful scientist. Good luck to the
new students, Jon, Sara and Brian (and all the others) and thanks for bringing some new livelihood
into the lab; you all have fun and work hard, and that is always appreciated. Finally, (for this
section) thanks to Dr. Joel Schrauben and Dr. Rick Fehir. Uncle Joel gave me an example of what

vii

a great graduate student looks like and without Uncle Rick, I wouldn’t have had such a cool science
project.
A special thank you is in order to Professor Malcolm Forbes (Bowling Green State University).
Thank you for picking me up when I was down. And whenever you need picked up from the
airport, I will be there.
Thank you to my family for the constant support and love. I honestly don’t know the kind of
place I would have been in without them. And finally (seriously), thank you to my wife, Melissa,
my daughter, Arlyn Mae, and my son, Ronan. Melissa, you have given me so much support, I can
say that without your support and confidence instilling, there would be no dissertation, there would
be no research, and there would not have been graduate school. The sacrifices you have made for
me to be here and do this, show your devotion and respect for our family is of the highest standard.
I love you and thank you for always telling me I’m a great chemist, because whether or not it’s
true, it’s always nice that someone thinks it’s true (ha ha). When I entered graduate school, I
thought I was doing something for myself and Melissa, but I was not entirely correct, I was also
doing this for our children. You are the best of what life has to offer and are the driving force for
everything your mother and I do.

viii

TABLE OF CONTENTS

LIST OF TABLES……………………………………………………………………………….. x
LIST OF FIGURES……………………………………………………………………………... xi
Chapter 1.

Introduction to Spin Exchange and 13C-Labelling of Phenanthrenequinones……. 1
1.1 Background of Research……………………………………………………… 1
1.2 Introduction to spin delocalization and 13C as an isotopic label for measuring
spin delocalization………………………………………………………………... 9
1.3 Contents of Dissertation……………………………………………………... 11
REFERENCES………………………………………………………………….. 15

Chapter 2.

Quantifying Spin Density Delocalization in Phenanthrenesemiquinones…….… 17
2.1 Introduction…………...…………………………………………………….. 17
2.2 Experimental.……………………………………………………………….. 21
2.3 Results and Discussion.……………………………………………………... 29
2.4 Conclusions and Future Work……………………………………………… 101
APPENDIX….….....…………………………………………………………... 104
REFERENCES..……………………………………………………………….. 128

Chapter 3.

The Influence of Lewis Acidity on Spin Density Delocalization……………… 133
3.1 Introduction…………...…………………………………………………… 133
3.2 Experimental.……………………………………………………………… 137
3.3 Results and Discussion.……………………………………………………. 143
3.4 Conclusions and Future Work……………………………………………… 171
APPENDIX….….....…………………………………………………………... 173
REFERENCES..……………………………………………………………….. 182

Chapter 4.

Quantifying Spin Density Delocalization in NiII-Phenanthrenesemiquinone
Complexes……………………………………………………………………... 186
4.1 Introduction…………...…………………………………………………… 186
4.2 Experimental.……………………………………………………………… 190
4.3 Results and Discussion.……………………………………………………. 194
4.4 Conclusions and Future Work……………………………………………… 222
APPENDIX….….....…………………………………………………………... 224
REFERENCES..……………………………………………………………….. 232

Chapter 5.

Conclusions and Future Directions…………………………………………..… 235
5.1 Concluding Remarks….…………………………………………………… 235
5.2 Future Directions.………………………………………………………….. 241
5.3 Final Remarks………..………………………………………………….…. 245
REFERENCES..……………………………………………………………….. 247

ix

LIST OF TABLES

Table 2-1.

Selected bond lengths and angles for compound 13……………………………... 38

Table 2-2.

Crystal structure parameters for compound 13…………………………………... 38

Table 2-3.

1

Table 2-4.

Calculated spin density values from equations 2.8a and 2.8b for NaPSQ complex
(data and from Fermi analysis). Spin density values are unit-less……………….. 57

Table 2-5.

1

Table 2-6.

Calculated spin density values from equations 2.8a and 2.8b for Na(3,6-(OMe)2PSQ)…………………………………………………………………………….... 63

Table 2-7.

Hyperfine coupling parameters for NaPSQ dimer and molecule. All values in
MHz…………………………………………………………………………….... 77

Table 2-8.

Hyperfine coupling parameters for Na(3,6-(OMe)2-PSQ) homoleptic complex,
dimer and molecule. All values in MHz…………………………………………. 84

Table 2-9.

Comparison of hyperfine coupling parameters between NaPSQ components and
Na(3,6-(OMe)2-PSQ) components……………………………………………….. 85

Table 3-1.

Average carbonyl carbon and oxygen spin density in Zn-PSQ and PSQ systems as
calculated from NPA analysis…………………………………………………... 146

Table 3-2.

Carbon and Oxygen spin density in 3,6-R2-PSQ radical anions, [Zn(tren)(PSQ)]+
and [Ga(tren)(PSQ)]2+ complexes………………………………………………. 149

Table 3-3.

Calculated Fermi contact integrals and 1H-aiso values of 2 obtained from simulations
of CW spectra, in MHz…………………………………………………………. 158

Table 3-4.

Calculated spin density values at 1H and 13C nuclei for 2 and 3………………… 160

Table 4-1.

Hyperfine coupling parameters obtained from Ni-[13C2]PSQ/NiPSQ ESEEM…. 210

Table 5-1.

13

H, 13C and 23Na aiso and DFT calculated values (Fermi contact integral). All values
in MHz…………………………………………………………………..……….. 54

H, 13C and 23Na aiso coupling constants, in MHz, for Na(3,6-(OMe)2-PSQ)……. 62

C Hyperfine parameters, in MHz, for NaPSQ Ion-Pair, Na(OMe)2PSQ Ion-Pair
Zn-PSQ and Ni-PSQ……………………………………………………………. 236

x

LIST OF FIGURES

Figure 1-1.

Left: Interaction between two S = ½ centers where the ground state is the singlet.
Right: Interaction where the triplet is the ground state……………………………. 2

Figure 1-2.

Energy diagram for the interaction of a NiII ion and semiquinone………………… 4

Figure 1-3.

Average oxygen atom spin density versus J in [Ni(tren)(3,6-R2-PSQ)]+
complexes…………………………………………………………………………. 5

Figure 1-4.

Molecular structure of [Ni(tren)(PSQ)]+…………………………………………... 7

Figure 1-5.

Oxidation states of phenanthrenequinone………………………………………..... 7

Figure 2-1.

Structure of 3,6-R2-PSQ radical anions with proton numbering scheme………… 18

Figure 2-2.

Average carbonyl carbon versus average carbonyl oxygen spin density in 3,6-R2PSQs………………………………………....………………………………...… 30

Figure 2-3.

Average oxygen spin density versus total carbonyl spin density…………….…... 31

Figure 2-4.

Average carbonyl spin density versus average carbonyl carbon spin density…… 32

Figure 2-5.

Retrosynthetic analysis for synthesizing a di-13C-labeled phenanthrenequinone… 34

Figure 2-6.

General synthetic scheme for [13C2]phenanthrenequinone………………………. 35

Figure 2-7.

Proposed intermediate formation and expulsion of N2 for synthesizing phenanthrene
from 2,2’-diformyl biphenyl. The figure was reproduced from Reference
22....……………………………………..……………………………………….. 36

Figure 2-8.

Crystal structure of [9,10-13C2]3,6-(OMe)2-PhenQ (13). The structure is displayed
with thermal ellipsoids at 50% probability………………………………………. 37

Figure 2-9.

Resonance structure of 3,6-(OMe)2-PhenQ……………………………………… 38

Figure 2-10.

1

Figure 2-11.

1

H-NMR spectra of (CHO)2-Ph2 (bottom) and (13CHO)2-Ph2 (top). Bold numbers
are relative integrals with respect to the aldehyde proton(s)……………………... 40
H-NMR spectra of (CHO)2-(OMe)2-Ph2 (bottom) and (13CHO)2-(OMe)2-Ph2
(top)……………………………………………………………………………… 41

Figure 2-12. Splitting of α and β electron orbitals in an applied magnetic field………………. 43
xi

Figure 2-13. Electronic and nuclear Zeeman effects. Allowed EPR transitions are highlighted by
double-headed arrows in the MI manifold……………………………………….. 45
Figure 2-14. EPR spectrum of PSQ. Reproduced from Reference 52…………………………. 46
Figure 2-15. Room temperature EPR spectrum (black) and simulation (red) of PSQ radical anion
at 20 mG modulation amplitude. Simulation parameters: g = 2.00754, lwpp = 0.0005
mT, 0.0191 mT, aiso values: 23Na = 1.32 MHz, 1H4,5 = 0.77 MHz, 1H2,7 = 1.19 MHz,
1
H1,8 = -4.06 MHz and 1H3,6 = -4.75 MHz. Experimental parameters: 9.85 GHz ÎźW
frequency, 0.50 mW power, 3508 G center field, 20 G sweep width, 20 mG
modulation amplitude, 1024 points. χ2 = 0.39…………………………………... 48
Figure 2-16. Possible Na+ coordination to PSQ-∙………………………………………………. 50
Figure 2-17. PSQ (black) and [13C2]PSQ (red) overlaid CW spectra………………………….. 50
Figure 2-18. Room temperature EPR spectrum (black) and simulation (red) of 13C2PSQ radical
anion at 20 mG modulation amplitude. Simulation parameters: g = 2.00755, lwpp =
0.0001 mT, 0.0222 mT, 13C-aiso values: 1.57 MHz. Experimental parameters: 9.85
GHz ÎźW frequency, 0.50 mW power, 3508 G center field, 20 G sweep width, 20
mG modulation amplitude, 1024 points. χ2 = 0.85................................................ 52
Figure 2-19. Room temperature EPR spectrum (black) and simulation (red) of 13C2PSQ radical
anion at 20 mG modulation amplitude. Simulation parameters: g = 2.00755, lwpp =
0.0001 mT, 0.0255 mT, 13C-aiso values: 2.17 MHz. Experimental parameters: 9.85
GHz ÎźW frequency, 0.50 mW power, 3508 G center field, 20 G sweep width, 20
mG modulation amplitude, 1024 points. χ2 = 1.02…………………………….... 53
Figure 2-20. Visual representation of Table 3. The aiso values listed are located at each of their
respective nuclear positions…………………………………………………….... 54
Figure 2-21. Room temperature EPR spectrum (black) and simulation (red) of Na(3,6-(OMe)2PSQ) radical anion at 20 mG modulation amplitude. Simulation parameters: g =
2.00746, lwpp = 0.0015 mT, 0.0135 mT, aiso values: 23Na = 1.17 MHz, 1HMe-a = 0.56,
1
HMe-b = 0.41 1H4,5 = 0.96 MHz, 1H2,7 = 1.01 MHz and 1H1,8 = 3.84, MHz.
Experimental parameters: 9.85 GHz ÎźW frequency, 0.5 mW power, 3508 G center
field, 20 G sweep width, 20 mG modulation amplitude, 1024 points. χ2 = 0.52... 58
Figure 2-22. [13C2]3,6-(OMe)2-PSQ (red) versus 3,6-(OMe)2-PSQ (black) spectra…………... 60
Figure 2-23. Room temperature EPR spectrum (black) and simulation (red) of Na[13C2]3,6(OMe)2-PSQ at 20 mG modulation amplitude. Simulation parameters: g =
2.007443, lwpp = 0.0015 mT, 0.0145 mT, 13C-aiso value: 0.41 MHz. Experimental
parameters: 9.85 GHz ÎźW frequency, 0.5 mW power, 3508 G center field, 20 G
sweep width, 20 mG modulation amplitude, 1024 points. χ2 = 0.50……………. 61
xii

Figure 2-24. Visual representation of Table 4. The aiso values listed are located at each of their
respective nuclear positions…………………………………………………….... 63
Figure 2-25. Energy level diagram of an S = ½ I = ½ system…………………………………. 66
Figure 2-26. Vector model of a two pulse ESEEM experiment……………………………….. 68
Figure 2-27. Timing diagram of a two pulse ESEEM sequence……………………………….. 68
Figure 2-28. Timing diagram of a three pulse ESEEM sequence……………………………... 69
Figure 2-29. [13C2]PSQ (red) and PSQ (black) ESEEM frequency spectra. Experimental
conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns time
increment, 136 ns tau, 40 ns starting T value. Spectra were normalized post-Fourier
transform to illustrate the differences of the ……………………………………. 72
Figure 2-30. [13C2]PSQ ESEEM frequency spectrum and fit for dimer. Simulation parameters: g
= 2.00575, 2 13C nuclei, aiso = 2.1713 MHz, T = 2.7757 MHz, ρ = 0 MHz.
Experimental conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns
time increment, 136 ns tau, 40 ns starting T value………………………………. 73
Figure 2-31. [13C2]PSQ ESEEM time domain spectrum and fit for dimer. Simulation parameters:
g = 2.00575, 2 13C nuclei, aiso = 2.1713 MHz, T = 2.7757 MHz, ρ = 0 MHz.
Experimental conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns
time increment, 136 ns tau, 40 ns starting T value. χ2 = 0.67…………………... 74
Figure 2-32. [13C2]PSQ ESEEM frequency spectrum and fit for multicomponent system.
Simulation parameters: g = 2.00575, Dimer: 4 13C nuclei, aiso = 2.1713 MHz, T =
2.7757 MHz, ρ = 0 MHz; ion-pair: 2 13C nuclei, aiso = 1.5741 MHz, T = 13.5674
MHz, ρ = 0.44 MHz. Experimental conditions: 345.5 mT magnetic field, 3 pulse
ESEEM sequence, 12 ns time increment, 136 ns tau, 40 ns starting T value……. 75
Figure 2-33. [13C2]PSQ ESEEM time domain spectrum and fit for multicomponent system.
Simulation parameters: g = 2.00575, Dimer: 4 13C nuclei, aiso = 2.1713 MHz, T =
2.7757 MHz, ρ = 0 MHz; ion-pair: 2 13C nuclei, aiso = 1.5741 MHz, T = 13.5674
MHz, ρ = 0.44 MHz. Experimental conditions: 345.5 mT magnetic field, 3 pulse
ESEEM sequence, 12 ns time increment, 136 ns tau, 40 ns starting T value. χ2 =
0.22………………………………………………………………………………. 76
Figure 2-34. Cartoon depiction of the ion-pair and dimer of the NaPSQ multicomponent
system……………………………………………………………………………. 77
Figure 2-35. [13C2]3,6-(OMe)2-PSQ ESEEM frequency spectrum and fit for the homoleptic
complex. Simulation parameters: g = 2.00507, 6 13C nuclei, aiso = 0.9758 MHz, T =
1.15 MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic field, 3 pulse
ESEEM sequence, 12 ns time increment, 136 ns tau, 40 ns starting T value……. 79
xiii

Figure 2-36. [13C2]3,6-(OMe)2-PSQ ESEEM time domain spectrum and fit for the homoleptic
complex. Simulation parameters: g = 2.00507, 6 13C nuclei, aiso = 0.9758 MHz, T =
1.15 MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic field, 3 pulse
ESEEM sequence, 12 ns time increment, 136 ns tau, 40 ns starting T value. χ2 =
0.94………………………………………………………………………………. 80
Figure 2-37. [13C2] 3,6-(OMe)2-PSQ ESEEM frequency spectrum and fit for multicomponent
system. Simulation parameters: g = 2.00507, homoleptic complex: 6 13C nuclei, aiso
= 0.9758 MHz, T = 1.15 MHz, ρ = 0 MHz (4 of these); dimer: 4 13C nuclei, aiso =
0.3849 MHz, T = 7.9563 MHz, ρ = 0 MHz; ion-pair: 2 13C nuclei, aiso = -0.4155
MHz, T = 16.1325 MHz, ρ = 0.63 MHz. Experimental conditions: 345.5 mT
magnetic field, 3 pulse ESEEM sequence, 12 ns time increment, 136 ns tau, 40 ns
starting T value…………………………………………………………………... 81
Figure 2-38. [13C2] 3,6-(OMe)2-PSQ ESEEM time domain spectrum and fit for multicomponent
system. Simulation parameters: g = 2.00507, homoleptic complex: 6 13C nuclei, aiso
= 0.9758 MHz, T = 1.15 MHz, ρ = 0 MHz (4 of these); dimer: 4 13C nuclei, aiso =
0.3849 MHz, T = 7.9563 MHz, ρ = 0 MHz; ion-pair: 2 13C nuclei, aiso = -0.4155
MHz, T = 16.1325 MHz, ρ = 0.63 MHz. Experimental conditions: 345.5 mT
magnetic field, 3 pulse ESEEM sequence, 12 ns time increment, 136 ns tau, 40 ns
starting T value. χ2 = 0.35……………………………………………………….. 82
Figure 2-39. Cartoon depiction of the Na(3,6-(OMe)2-PSQ) multicomponent system………... 83
Figure 2-40. Resonance structures of 3,6-(OMe)2-PSQ-∙ illustrating the pi-donating character of
the methoxy group……………………………………………………………….. 86
Figure 2-41. Four pulse HYSCORE sequence…………………………………………………. 87
Figure 2-42. HYSCORE spectrum of [13C2]PSQ. Experimental settings: 345.5 mT Field, 16 ns
time increment, 136 ns tau, 36 ns starting t1 and t2 time…………………………. 89
Figure 2-43. HYSCORE spectrum (black) and simulation (red) of the [13C2]PSQ dimer.
Simulation parameters: g = 2.00575, 4 13C nuclei, aiso = 2.1713 MHz, T = 2.7757
MHz, ρ = 0 MHz. Experimental conditions: 346 mT magnetic field, 4 pulse
HYSCORE sequence, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2
values…………………………………………………………………………….. 90
Figure 2-44. HYSCORE spectrum (black) and simulation (red) of the [13C2]PSQ ion-pair.
Simulation parameters: g = 2.00575, 2 13C nuclei, aiso = 1.5741 MHz, T = 13.5674
MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic field, 4 pulse
HYSCORE sequence, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2
values…………………………………………………………………………….. 91
Figure 2-45. HYSCORE spectrum (black) and simulation (red) of the [13C2]PSQ multicomponent system. Simulation parameters: System 1, g = 2.00575, 4 13C nuclei, aiso
xiv

= 2.1713 MHz, T = 2.7757 MHz, ρ = 0 MHz; System 2, 2 13C nuclei, aiso = 1.5741
MHz, T = 13.5674 MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic
field, 4 pulse HYSCORE sequence, 16 ns time increment, 136 ns tau, 36 ns starting
t1 and t2 values……………………………………………………………………. 93
Figure 2-46. HYSCORE spectrum of [13C2]3,6-(OMe)2-PSQ. Experimental settings: 345.5 mT
Field, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2 time……………. 95
Figure 2-47. HYSCORE spectrum (black) and simulation (red) of the [13C2]3,6-(OMe)2-PSQ
homoleptic complex. Simulation parameters: g = 2.0057, 6 13C nuclei, aiso = 0.9758
MHz, T = 1.3064 MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic
field, 4 pulse HYSCORE sequence, 16 ns time increment, 136 ns tau, 36 ns starting
t1 and t2 values……………………………………………………………………. 96
Figure 2-48. HYSCORE spectrum (black) and simulation (red) of the [13C2]3,6-(OMe)2-PSQ
dimer. Simulation parameters: g = 2.0057, 4 13C nuclei, aiso = 0.3849 MHz, T =
7.9653 MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic field, 4
pulse HYSCORE sequence, 16 ns time increment, 136 ns tau, 36 ns starting t1 and
t2 values…………………………………………………………………………... 97
Figure 2-49. HYSCORE spectrum (black) and simulation (red) of the [13C2]3,6-(OMe)2-PSQ ionpair in Quadrant II. Simulation parameters: g = 2.0057, 2 13C nuclei, aiso = -0.4155
MHz, T = 16.1325 MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic
field, 4 pulse HYSCORE sequence, 16 ns time increment, 136 ns tau, 36 ns starting
t1 and t2 values……………………………………………………………………. 98
Figure 2-50. HYSCORE spectrum (black) and simulation (red) of the [13C2]3,6-(OMe)2-PSQ
multi-component system in Quadrant I. Simulation parameters: g = 2.0057, 6 13C
nuclei, aiso = 0.9758 MHz, T = 1.3064 MHz, ρ = 0 MHz, 4 13C nuclei, aiso = 0.3849
MHz, T = 7.9653 MHz, ρ = 0 MHz, and 2 13C nuclei, aiso = -0.4155 MHz, T =
16.1325 MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic field, 4
pulse HYSCORE sequence, 16 ns time increment, 136 ns tau, 36 ns starting t1 and
t2 values…………………………………………………………………………... 99
Figure 2-51. HYSCORE spectrum (black) and simulation (red) of the [13C2]3,6-(OMe)2-PSQ
multi-component system in Quadrant II. Simulation parameters: g = 2.0057, 6 13C
nuclei, aiso = 0.9758 MHz, T = 1.3064 MHz, ρ = 0 MHz, 4 13C nuclei, aiso = 0.3849
MHz, T = 7.9653 MHz, ρ = 0 MHz, and 2 13C nuclei, aiso = -0.4155 MHz, T =
16.1325 MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic field, 4
pulse HYSCORE sequence, 16 ns time increment, 136 ns tau, 36 ns starting t1 and
t2 values…………………………………………………………………………. 100
Figure 3-1.

Molecular structure of [Zn(tren)(3,6-R2-PSQ)]+………………………………... 134

Figure 3-2.

Average Carbonyl carbon spin density versus average oxygen for [Zn(tren)(3,6-R2PSQ)]+ Complexes…………………………………………………………….... 146
xv

Figure 3-3.

Average Carbonyl carbon spin density versus average oxygen for [Ga(tren)(3,6-R2PSQ)]2+ Complexes……………………………………………………………... 149

Figure 3-4.

Oxidation states of phenanthrenequinone………………………………………. 150

Figure 3-5.

10 K CW spectrum (black) and simulation (red) of 5, [Ga(tren)(PSQ)](BPh4)2.
Simulation parameters: g = 2.00535, lwpp = 0.71 mT, 0.28 mT, aiso values: 69,71Ga
= 19.73 MHz. Experimental parameters: 9.72 GHz ÎźW frequency, 0.63 mW power,
3460 G center field, 200 G sweep width, 1 G modulation amplitude, 512 points. χ2
= 0.23…………………………………………………………………………... 155

Figure 3-6.

10 K CW spectrum (black) and simulation (red) of 6, [Ga(tren)([13C2]PSQ)](BPh4)2.
Simulation parameters: g = 2.00532, lwpp = 1.09 mT Lorentzian + 0.57 mT
Gaussian, aiso values: 69,71Ga = 19.73 MHz, 13C-aiso values: 27.77 MHz, 32.10 MHz.
Experimental parameters: 9.72 GHz ÎźW frequency, 0.016 mW power, 3460 G
center field, 200 G sweep width, 1 G modulation amplitude, 512 points. χ2 =
0.22……………………………………………………………………………... 156

Figure 3-7.

Room temperature CW spectrum (black) and simulation (red) of 2,
[Zn(tren)(PSQ)](BPh4) at 20 mG modulation amplitude. Simulation parameters: g
= 2.0071; aiso values: H3,6 = 5.90, H1,8 = 4.01, H2,7 = 1.96, H4,5 = 1.01; Peak to peak
line width: 0.0136 mT Gaussian 0.0107 mT Lorentzian broadening. Experimental
parameters: 9.42 GHz microwave frequency, 0.80 ÎźW microwave power, 3353 G
centerfield, 50 G sweep width, 20 mG modulation amplitude, 1024 points. χ2 =
0.45……………………………………………………………………………... 157

Figure 3-8.

Room temperature CW spectrum (black) and simulation (red) of 3,
[Zn(tren)([13C2]PSQ)](BPh4) at 20 mG modulation amplitude.
Simulation
13
parameters: g = 2.007107; C-aiso values: 1.08 MHz and 2.78 MHz; Peak to peak
line width: 0.0246 mT Gaussian plus 0.0247 mT Lorentzian broadening.
Experimental parameters: 9.39 GHz microwave frequency, 0.80 mW microwave
power, 3320 G centerfield, 50 G sweep width, 20 mG modulation amplitude, 1024
points. χ2 = 0.93………………………………………………………………... 159

Figure 3-9.

13

Figure 3-10.

13

C/12C Zn-PSQ ESEEM frequency domain spectrum and fit. Simulation
parameters: C1: aiso = 1.08 MHz, T = 2.84 MHz ρ = 0; C2: aiso = 2.78, T = 2.46, ρ =
0. Experimental parameters: Field = 345.5 mT, 3 pulse ESEEM sequence, 12 ns
time increment, 136 ns τ and 40 ns starting time value…………………………. 163
C/12C Zn-PSQ ESEEM time domain spectrum and fit. Simulation parameters: C1:
aiso = 1.08 MHz, T = 2.84 MHz ρ = 0; C2: aiso = 2.78, T = 2.46, ρ = 0. Experimental
parameters: Field = 345.5 mT, 3 pulse ESEEM sequence, 12 ns time increment, 136
ns τ and 40 ns starting time value………………………………………………. 165

Figure 3-11. Comparison of 2 (black) and 3 (red) ESEEM spectra. Fourier transforms were
normalized to compare features from each spectrum…………………………... 166
xvi

Figure 3-12. HYSCORE spectrum of 3, Zn-13C2PSQ. Experimental conditions: 345.5 mT
magnetic field, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2 time
values…………………………………………………………………………… 167
Figure 3-13. HYSCORE spectrum of 3, Zn-13C2PSQ. Experimental conditions: 345.5 mT
magnetic field, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2 time values.
A lower z-threshold was used to accentuate weak cross-couplings in quadrant
two…………………………………………………………………….………... 168
Figure 3-14. HYSCORE simulation (red) and spectrum (black) of 3, Zn-13C2PSQ. Experimental
conditions: 345.5 mT magnetic field, 16 ns time increment, 136 ns tau, 36 ns starting
t1 and t2 time values. Simulation parameters: g = 2.0033; 13C1, aiso = 1.08 MHz, T =
2.84 MHz; 13C2, aiso = 2.78 MHz, T = 2.46 MHz; HYSCORE pulse sequence…. 169
Figure 3-15. HYSCORE simulation (red) and spectrum (black) of 3, Zn-13C2PSQ in quadrant one.
Experimental conditions: 345.5 mT magnetic field, 16 ns time increment, 136 ns
tau, 36 ns starting t1 and t2 time values. Simulation parameters: g = 2.0033; 13C1,
aiso = 1.08 MHz, T = 2.84 MHz; 13C2, aiso = 2.78 MHz, T = 2.46 MHz; HYSCORE
pulse sequence………………………………………………………………….. 170
Figure 4-1.

Average oxygen spin density versus J, reproduced from Reference 1…………. 186

Figure 4-2.

Molecular structure of [Ni(tren)(3,6-R2-PSQ)]+ ( where tren = tris(2amineoethyl)amine and 3,6-R2-PSQ = 3,6-R2-Phenanthrenesemiquinone)
complexes………………………………………………………………………. 187

Figure 4-3.

Average carbonyl carbon spin density versus J………………………………… 189

Figure 4-4.

Synthetic scheme for synthesizing [Ni(tren)(PSQ)](BPh4) complexes…………. 194

Figure 4-5.

Zero field splitting diagram for a quartet ground state………………………….. 197

Figure 4-6.

7 K CW spectrum and simulation of 2, [Ni(tren)(PSQ)](BPh4). Simulation
parameters: S = [1,1/2], gxx = gyy = 2.08, gzz = 2.44, D = 49,000 MHz (1.63 cm-1), E
= 6200 MHz (0.21 cm-1), electron exchange = -6.1 cm-1 (-182,647.9 MHz), HStrain
= (2200,1100,90) MHz, DStrain = 38,000 MHz, 2,900 MHz. Experimental
conditions: 4 K, 9.41 GHz microwave operating frequency, 0.8 mT modulation
amplitude, 4000 G sweep, 260 mT center field and 2048 points………………. 198

Figure 4-7.

7 K CW spectra of 2 (black), [Ni(tren)(PSQ)](BPh4), and 3 (blue),
[Ni(tren)([13C2]PSQ)](BPh4)………………………………………………….... 199

Figure 4-8.

EPR spectra of [Ni(tren)(3,6-R2-PSQ)](BPh4) complexes where R = H (red, bottom),
NH2 (blue, middle) and NO2 (green, top). Spectra taken from Reference 1…… 203

xvii

Figure 4-9.

Comparison of various electron exchange (exchange coupling) values for Ni-PSQ
based on simulation parameters for Ni-PSQ…………………………………… 204

Figure 4-10.

13

Figure 4-11.

13

Figure 4-12.

13

Figure 4-13.

13

C/12C NiPSQ ESEEM Fourier transform (black) and fit (red). Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1:
aiso = -2.11 MHz, T = 3.29 MHz ρ = 0.433 MHz; C2: aiso = 1.58 MHz, T = 6.32 MHz,
ρ = 0.03 MHz. Experimental conditions: 334 mT magnetic field, 9.71 GHz
microwave frequency, 4 K, 3 pulse ESEEM sequence, 140 ns τ, 40 ns acquisition
trigger and 12 ns increments……………………………………………………. 207
C/12C NiPSQ ESEEM Fourier transform (black) and fit (red). Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1:
aiso = -2.11 MHz, T = 3.29 MHz ρ = 0.433 MHz; C2: aiso = 1.58 MHz, T = 6.32 MHz,
ρ = 0.03 MHz. Experimental conditions: 334 mT magnetic field, 9.71 GHz
microwave frequency, 4 K, 3 pulse ESEEM sequence, 140 ns τ, 40 ns acquisition
trigger and 12 ns increments……………………………………………………. 208
C/12C NiPSQ ESEEM Fourier transform (black) and fit (red). Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1:
aiso = -4.09 MHz, T = 1.535 MHz ρ = 0.0 MHz; C2: aiso = 1.07 MHz, T = 6.36 MHz,
ρ = 0.19 MHz. Experimental conditions: 334 mT magnetic field, 9.71 GHz
microwave frequency, 4 K, 3 pulse ESEEM sequence, 140 ns τ, 40 ns acquisition
trigger and 12 ns increments……………………………………………………. 211
C/12C NiPSQ ESEEM Fourier transform (black) and fit (red). Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1:
aiso = -4.09 MHz, T = 1.535 MHz ρ = 0.0 MHz; C2: aiso = 1.07 MHz, T = 6.36 MHz,
ρ = 0.19 MHz. Experimental conditions: 334 mT magnetic field, 9.71 GHz
microwave frequency, 4 K, 3 pulse ESEEM sequence, 140 ns τ, 40 ns acquisition
trigger and 12 ns increments……………………………………………………. 212

Figure 4-14. Overlaid ESEEM spectra of 13C/12C Ni-PSQ: 334 mT (black), 260 mT (red) and 180
mT (blue)……………………………………………………………………….. 213
Figure 4-15.

13

C/12C NiPSQ ESEEM Fourier transform (black) and fit (red). Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1:
aiso = -2.11 MHz, T = 3.29 MHz ρ = 0.433 MHz; C2: aiso = 1.58 MHz, T = 6.32 MHz,
ρ = 0.03 MHz. Experimental conditions: 260 mT magnetic field, 9.71 GHz
microwave frequency, 4 K, 3 pulse ESEEM sequence, 92 ns τ, 40 ns acquisition
trigger and 12 ns increments……………………………………………………. 214

Figure 4-16. HYSCORE 2D-FT spectrum of 3 acquired at 4 K. Experimental conditions: 334
mT magnetic field, 9.71 GHz microwave frequency, 4 K, 4 pulse HYSCORE
sequence, 140 ns τ, 40 ns acquisition trigger (both dimensions) 16 ns increments and
256 x 256 points……………………………………………………………….... 216

xviii

Figure 4-17. HYSCORE 2D-FT spectrum of 2 acquired at 4 K. Experimental conditions: 334
mT magnetic field, 9.71 GHz microwave frequency, 4 K, 4 pulse HYSCORE
sequence, 140 ns τ, 40 ns acquisition trigger (both dimensions) 16 ns increments and
256 x 256 points……………………………………………………………….... 217
Figure 4-18. HYSCORE 2D-FT simulation (red) and spectrum of 3 acquired at 4 K. Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1:
aiso = -2.11 MHz, T = 3.29 MHz ρ = 0.433 MHz; C2: aiso = 1.58 MHz, T = 6.32 MHz,
ρ = 0.03 MHz. Experimental conditions: 334 mT magnetic field, 9.71 GHz
microwave frequency, 4 K, 4 pulse HYSCORE sequence, 140 ns τ, 40 ns acquisition
trigger (both dimensions) 16 ns increments and 256 x 256 points……………... 218
Figure 4-19. HYSCORE 2D-FT spectrum of 3 acquired at 4 K. Experimental conditions: 260
mT magnetic field, 9.71 GHz microwave frequency, 4 K, 4 pulse HYSCORE
sequence, 92 ns τ, 40 ns acquisition trigger (both dimensions) 16 ns increments and
256 x 256 points……………………………………………………………….... 220
Figure 4-20. HYSCORE 2D-FT spectrum of 2 acquired at 4 K. Experimental conditions: 260
mT magnetic field, 9.71 GHz microwave frequency, 4 K, 4 pulse HYSCORE
sequence, 92 ns τ, 40 ns acquisition trigger (both dimensions) 16 ns increments and
256 x 256 points……………………………………………………………….... 221
Figure 5-1.

Molecular structures of [Ni(tren)(ASQ)]+ (top) and [Ni(tren)(PazSQ)]+
(bottom)……………………………………………………………………….... 241

Figure 5-2.

Average oxygen spin density versus J for [Ni(tren)(4,5-R2-ASQ)]+ complexes... 243

Figure 5-3.

Average oxygen spin density versus J for [Ni(tren)(4,5-R2-PazSQ)]+
complexes………………………………………………………………………. 244

xix

Chapter 1. Introduction to Spin Exchange and 13C-Labelling of Phenanthrenequinones

1.1 Background of Research
The very cornerstone of the research presented within this dissertation is based on results
from DFT (Density Functional Theory) calculations on [Ni(tren)(3,6-R2-PSQ)]+ (where R is 17
various substituents varying in electron-withdrawing/-donating character) complexes1. These
complexes exhibit a particular intramolecular property known as exchange coupling. Exchange
coupling is a physical phenomenon in which two or more unpaired electrons on separate nuclei
communicate electronically. This is a stabilization effect and can be as large as hundreds of
wavenumbers (cm-1). This communication can occur by various methods: 1. “direct exchange”,
in which two spin centers are directly bonded to each other; 2. “super exchange”, communication
between the spin centers occurs through a diamagnetic mediator; 3. “double exchange”, which is
a complicated “hopping” type mechanism between paramagnetic centers and a diamagnetic
bridge. Only direct exchange will be addressed in the context of the compounds studied in this
dissertation.
In a direct exchange mechanism, two different forms of exchange coupling can occur,
depending on the nature of the magnetic orbitals involved in the exchange. If the magnetic
orbitals are aligned parallel to each other, then the coupling is said to be antiferromagnetic. If
the magnetic orbitals are orthogonal (in other words, no orbital overlap) then the interaction is
ferromagnetic. To describe this physical interaction mathematically, a Hamiltonian of the form
̂ = 𝐽𝑆̂1 ∙ 𝑆̂2
𝐻

(1.1)

is employed. Specifically, the Heisenberg Dirac van Vleck (HDvV) Hamiltonian above is one
expression of the exchange interactions and will be the only expression used in this dissertation.
1

J is the exchange coupling constant and is directly related to the magnitude of the interaction.
The sign of J is also sensitive to the kind of interaction; when J > 0, antiferromagnetic coupling
is the mechanism and when J < 0, ferromagnetic coupling is present. This is represented
pictorially in Figure 1.

Figure 1-1. Left: Interaction between two S = ½ centers where the ground state is the singlet.
Right: Interaction where the triplet is the ground state.
The exchange coupling constant, J, is directly proportional to the product of the spin densities of
the spin centers J2,3, (see equation 2 below). Thus, the spin density is the progenitor of the
magnitude of the exchange coupling.
𝐽 ≈ 𝜌(𝑆1 )𝜌(𝑆2 )

2

(1.2)

The energy splitting separating the singlet (S = 0) and triplet (S = 1) states in Figure 1 can be
calculated with respect to J. Using the Kambe approximation4, which assumes 𝐒T = 𝐒1 + 𝐒2 ,
and squaring both sides of the equation and solving for S1∙S2, one can obtain the following
relationship
𝑺𝟏 ∙ 𝑺𝟐 =

𝑺2𝑻 − 𝑺2𝟏 − 𝑺2𝟐
2

(1.3)

This can then be inserted into equation 1, to afford the following equation
𝐇=

1
∙ 𝐽12 ∙ (𝐒T2 − 𝐒12 − 𝐒22 )
2

(1.4)

It is also known from quantum mechanics that when a squared operator operates on a particular
wave function, the eigenvalue can be obtained from the following equation:
𝐒𝐓𝟐 𝚿 = S(S + 1)𝚿

(1.5)

When this relationship is combined with the operator equivalent form of (1.3), the following
equation is the result:
1

E = 2 ∙ 𝐽12 ∙ [ST (ST + 1) + S1 (S1 + 1) + S2 (S2 + 1)]

(1.6)

From this equation, one can calculate the relative energies of the individual spin states of a given
system, and thus the energy difference(s) that separate them. For example, if there was a two
spin system where S1 = 1 for a high spin NiII ion in an octahedral environment and S2 = ½ for an
organic radical that was directly bound to the NiII nucleus, then the following relative energies
and energy splittings could be calculated for the two spin states:
ES= 3 =

1
15 8 3
𝐽12
∙ 𝐽12 ∙ [ − − ] =
2
4 4 4
2

ES= 1 =

1
3 8 3
∙ 𝐽12 ∙ [ − − ] = −𝐽12
2
4 4 4

2

2

3

3
|ES= 3 − ES= 1 | = 𝐽12
2
2
2
Figure 2 depicts a splitting diagram of this interaction. Now, if J ˃˃ kBT, then there will be little
probability of populating the doublet state and only an isolated, quartet ground state will exist at
room temperature. This is the exact situation that has been observed experimentally, both by
variable temperature magnetic susceptibility, as well as variable temperature CW-EPR
measurements.

Figure 1-2. Energy diagram for the interaction of a NiII ion and semiquinone.
As it is depicted in Figure 2, the exchange interaction is ferromagnetic for these Ni-SQ
complexes and as it was stated earlier, this is a result of the orthogonality of the magnetic
orbitals. The unpaired electrons reside in eg orbitals on nickel, specifically the dx2-y2 and dz2
orbitals. The unpaired electron on the semiquinone resides in a p – orbital. These electrons are
exchangeable and obey Hund’s rule of maximum multiplicity, therefore the ground state is a
quartet.
The calculations mentioned in the beginning of this chapter were carried out to assess how
the exchange coupling constant and average oxygen spin density varies with choice of
substituent5, and are pictured in Figure 3. These calculations were a part of an effort to uncover
the mechanism of spin polarization by choice of substituent. The mechanism was not uncovered,
4

but many consequences of spin polarization in exchange coupled molecules were observed.
-280
SO2CN
-270

NO2
CN
H

-260

IF4
CF3
C(CF3)3

Br

J/cm-1

-250

Me

-240

-230
OMe

Cl
F
TMS
t-Bu

Vinyl

-220
-210
-200
0.21

R² = 0.9941

NH2
NMe2

0.22

0.23

0.24

0.25

0.26

0.27

0.28

Carbonyl-O (ρ)

Figure 1-3. Average oxygen atom spin density versus J in [Ni(tren)(3,6-R2-PSQ)]+ complexes.
The calculations clearly show a linear trend in choice of substituent when comparing
exchange coupling constant, J, and average oxygen spin density. This linear trend was also
observed in CrIII analogues, where the coupling was antiferromagnetic, due to the overlap of the
π-type orbitals of the t2g orbitals and the semiquinone π radical. The inherent observation in
these results is that when R is an electron withdrawing group, J is large and when R is electron
donating J is small, relative to R = H.

These calculations, that examine several substituents

based on exchange coupling, are unparalleled in the literature. Venegas-Yazigi and coworkers
produced a study on bis-phenoxo bridged dicopper(II) complexes where they examined the

5

effects of several structural parameters on exchange coupling5.

In particular, J versus the

Hammett σ parameter where the Cu-O-Cu angle is 105° and the Cu-O-C (phenoxo) angle is 0°
(in plane) and 50° (out of plane). In both instances of Cu-O-C angle, a linear dependence on J
and σ was observed. A similar trend in electronegativity was also observed, only for a small
series of substituents (R = CH3, H, CONH2, CO2H, CN, NO2). These complexes, however, have
an inherent problem in that the exchange coupling is too large to be measured by variable
temperature magnetic susceptibility. Therefore, this work examines spin density fluctuations
near the exchange coupling site.
The system of study is [Ni(tren)(PSQ)]+ The design of this system is not our own, but is
rather based on a molecule synthesized by Andrea Dei and Alessandro Bencini6. The system has
been

slightly

modified

in

that

the

dl-5,7,7,12,14,14-

hexamethyl-1,4,8,11-

tetraazacyclotetradecane (CTH) ancillary ligand has been replaced by tris(2-aminoethyl)amine
(tren). The reason for this switch is because CTH is chiral and has the ability to form isomers,
whereas tren does not form isomers. A secondary purpose of tren is to ensure an octahedral
geometry for NiII by occupying coordination sites, not occupied by the semiquinone ligand. A
monosemiquinone complex was chosen to study one exchange interaction. Again, specifically a
direct exchange interaction was desired to examine the effects of spin density at the exchange
center on spin density.

6

Figure 1-4. Molecular structure of [Ni(tren)(PSQ)]+.
A quinone was chosen because of its redox active nature and wealth of known information.
The three oxidation states of quinones allow for them to be manipulated synthetically, which is
advantageous for synthesizing metal complexes. Quinones are the fully oxidized version of the
molecule and are not very strong Lewis bases/nucleophiles; however, quinone-transition metal
complexes are known7. The one-electron reduced form, the semiquinone, is a radical containing
molecule where most of the spin density of the ligand resides on the oxygen atom. These
radicals are generally not air-stable and therefore must be handled in air-free atmospheres.
Although these radicals are not air-stable, when bound to a metal ion, as a solid the compounds
can

be

stable

from

days

to

months.

Figure 1-5. Oxidation states of phenanthrenequinone.
7

In

air,

solutions

of

semiquinone-metal complexes can easily be oxidized, so care must be taken when handling these
types of compounds. The semiquinone form, although having an extra electron compared to the
quinone form, is still a relatively poor nucleophile. The two-electron reduced form, the
catecholate, is a strong nucleophile/Lewis base and is desirable for complexation reactions to
metals. Complexes within this body of research were prepared by reducing a quinone to the
catecholate, followed by addition of the catecholate solution to a metal precursor complex, and
finally oxidizing the metal-catecholate complex with a chemical oxidant to produce the desired
metal-semiquinone. Lastly, phenanthrenequinone was chosen so as to delocalize the spin density
to a greater extent than benzosemiquinones, and create a greater likelihood for exchange
coupling constants to be directly measured by variable temperature magnetic susceptibility
measurements. Unfortunately, this was not realized when three NiII complexes were made
(where R = NH2, H and NO2), and their magnetic susceptibilities showed no temperature
dependence. This being the case, it was decided to examine spin density at, or near, the
exchange site directly.
In order to examine spin density fluctuations a few items must be taken into consideration. If
our intent is to gain knowledge about spin density then we must examine a property that is
directly tied to spin density, so we will be examining hyperfine coupling constants. Specifically,
a hyperfine interaction is the coupling of the radical’s spin (S) to the spin of a nucleus (I). Thus
hyperfine coupling necessitates both an open shell molecule and a nucleus with spin. Since we
are dealing in the nature of radical containing compounds our method of detection is electron
paramagnetic resonance (EPR).
Due to our interest in the magnetic exchange, it would be desirable to know the magnitude of
spin density at the center of the exchange interaction, since it is near-impossible to measure J in

8

NiII-PSQ+ complexes in a typical manner (other methods to measure exchange coupling
constants exist but are costly and cumbersome). However, there are some obstacles to overcome
when considering measuring hyperfine coupling constants at oxygen atoms. First,
nuclear spin and therefore an isotope must be used in its place,

17

16

O has no

O, which implies isotopic

labeling. Fortunately for quinones it is quite simple to exchange oxygen atoms with 17OH2, and a
mild acid in a non-aqueous environment8. The cost of 17OH2 is high (~ $2,000/gram at the outset
of this research project), however, and enrichment methods are effective but not a scale that is
feasible for synthesizing an amount for physical characterization, synthesizing metal complexes
and EPR experiments9.

1.2 Introduction to spin delocalization and

13C

as an isotopic label for measuring spin

delocalization
Seeing as there are certain obstacles to overcome with 17O labeling, we chose to use 13C as a
spin label. By measuring the hyperfine coupling at the carbonyl carbon in free radicals, d10semiquinone complexes and NiII-semiquinone complexes, a single study has come about that
tracks spin density (via hyperfine coupling) fluctuations in molecules in a manner not previously
studied.

13

C-labelling of the carbonyl carbon nuclei will also provide some ease in interpretation

due to the I = ½ nucleus; however, this does not come without obstacles to overcome. Thus, the
effect of spin density delocalization on semiquinones, MII-semiquinone complexes, as well as
NiII-semiquinone complexes, has been examined.
Spin density delocalization and spin polarization are terms that may be difficult to
understand in terms of their context. In the context of this work spin density delocalization is a
manifestation of where the radical wave function has been dispersed, and spin polarization is the
9

perturbation of the wave function by some stimulus. Within this context, one can refer to
substituent effects, from complex to complex, as a means of polarizing spin density, as in the
computational study mentioned at the beginning of this chapter. The bulk of the material in this
dissertation will focus on the delocalization of the unpaired spin density on the PSQ moiety.
Room temperature, CW-EPR experiments have the resolution necessary to detect the spin
density at the nucleus. This is often referred to as the scalar coupling, isotropic coupling or the
Fermi isotropic coupling. Specifically, this is the amount of spin density in the s-orbitals. In
aromatic compounds this is on the order of 0 – 20 MHz10, given that aromatic radicals generally
reside in p-orbitals. The spacing between lines in the EPR signal are due to the isotropic
hyperfine coupling, which is tied directly to the sigma bonding environment of the nucleus.
Isotropic hyperfine coupling constants, or aiso, are also directly observable from ENDOR spectra
and can be extrapolated from ESEEM spectra.
Electron Spin Echo Envelope Modulation, ESEEM, is a pulsed technique performed at fixed
magnetic field values that measures the decay of the amplitude of a spin echo as a function of
time. This time trace is then background corrected and Fourier transformed and results in a
frequency spectrum. The frequency spectrum is an expression of the unpaired spin density and
the nuclei on which it resides. The frequency at which a resonance occurs coincides with the
Larmor frequency of a specified nucleus. The Larmor frequency is the frequency at which a
particular nucleus precesses in a given magnetic field and is represented mathematically in the
equation below,
𝜔 = −𝛾𝛽

(1.7)

where ω is the Larmor frequency, γ is the gyromagnetic ratio of the nucleus in a magnetic field
and β is the magnetic field. In a typical ESEEM experiment there is a preparatory pulse followed

10

by a second pulse at a specified time length, τ, after the first, and finally a third pulse creates a
stimulated echo and this third pulse is applied at varying times after the second pulse to record
echo amplitude as a function of time. ESEEM can also be a two pulse experiment, but only three
pulse experiments were used in the assessment of the molecules studied in this dissertation.
HYSCORE is a two-dimensional EPR technique that teases apart the strong and weak
hyperfine couplings, relative to the Larmor frequencies of the nuclei involved.

A two-

dimensional technique necessitates a second observable, to obtain this a fourth pulse is placed
between the second and third pulses of the ESEEM pulse sequence. This allows for two
dimensions of time to be monitored.

The data is worked-up similarly to that of a one-

dimensional experiment, but the main difference is that a two dimensional Fourier transform is
applied to the background corrected spectrum and the data is presented as a two-dimensional
contour plot, separated into quadrants. Only quadrants one and two are displayed; quadrant one
displays weak hyperfine couplings (A ˂ 2ν) and strong couplings (A ˃ 2ν) are found in quadrant
two. As in two-dimensional NMR experiments (like COSY), the on-diagonal elements are a
consequence of the one dimensional spectral features mapping onto each other and the offdiagonal elements are due to the so-called hyperfine sublevel correlations.
The details and necessities of the experiments and analysis thereof will be explained in the
forthcoming chapters.

1.3 Contents of Dissertation

The information, data, results and analysis in this dissertation can be grouped into three
sections: (1) spin density delocalization in PSQ radical anions; (2) the effect of Lewis acidity on

11

spin density delocalization, utilizing [Zn(tren)(13C2PSQ)](BPh4) and [Cd(tren)(PSQ)](BPh4)
complexes; (3) and finally the effect of exchange coupling on spin density delocalization in
[Ni(tren)(13C2PSQ)](BPh4) and [Ni(tren)(3,6-(OMe)2-13C2PSQ)](BPh4).
Chapter 2 examines spin density distribution in PSQ and 3,6-(OMe)2-PSQ radicals. The
study begins with a DFT analysis of the carbonyl carbon and oxygen spin density distribution. It
was found that the majority of the alpha spin density is contained within the carbonyl units of the
ligands and is split, symmetrically between the two units. Average spin densities are presented
due to the symmetry of the molecules. No linear trend was found in comparing carbon and
oxygen spin density. The chapter then switches to mainly experimental results and analysis,
beginning with synthesis of the

13

C-labeled phenanthrenequinones and their physical

characterization. Proceeding synthesis are CW-EPR, ESEEM and HYSCORE results from the
four radical compounds, paired with simulation results and DFT predictions of hyperfine
coupling constants.
Moving forward, Chapter 3 discusses spin polarization dependence on Lewis acidity.
Specifically, PSQ was bound to transition metals with full d shells, Zn II and GaIII. The choice of
metals is based on the NiII complexes. ZnII was chosen because its charge to size ratio is similar
to that of NiII’s, and it is diamagnetic and thus limits our study to focus on the semiquinone
ligand.

To restate, the purpose of using

13

C-labeling was to measure hyperfine coupling

constants near the exchange center in the NiII complexes, thus we desire to know the spin density
near the exchange center in the absence of unpaired electrons at the metal center. This system,
therefore, is absent of exchangeable electrons and becomes an excellent background for the
exchange coupled complexes. GaIII was originally chosen as a backdrop for NiII and CrIII-PSQ
complexes because the synthesis was more straight-forward and a crystal structure of the

12

catecholate complex was known. Additionally, it was desired to know if any hyperfine coupling
was present at the metal center, and Ga allows such measurements due to the I = 3/2 nuclei of
69

Ga and

71

Ga. This chapter is analogous to Chapter 2 in that DFT results of spin density

fluctuations in the ligand carbonyl unit are examined, synthesis of the complexes is explained
and a full EPR treatment is presented.
Following Chapter 3 adds the final layer of spin distortion based on the physical surrounding
of the PSQ radical. The effect of exchange coupling on spin density is examined by studying
[Ni(tren)(13C2PSQ)](BPh4) and its

12

C isotopologue. As in the previous chapters, the hyperfine

coupling of the 13C atoms is examined with CW and pulsed techniques. The CW-EPR line shape
is completely dominated by isotropic exchange coupling and zero-field splitting. Only a small
line shape change is observed between the 12C and 13C spectra in the g-2 region; unfortunately,
this difference in line shape could not be resolved by simulations. ESEEM and HYSCORE were
employed to study the hyperfine coupling at various magnetic field values, based on the CWEPR spectrum, so as to fully characterize the EPR line shape. It was found that in going from
high field to low field, the effect of the 13C hyperfine disappears. This is most likely an effect of
moving from g-2 to g-4, because the radical wave function is mostly associated with the PSQ
ligand. Also presented in Chapter 4 is a comparison of the DFT results with the experimental
hyperfine coupling constants.
The finale of this dissertation culminates all the conclusions and short-comings of this study
and lays out the landscape of the future of the project in light of the short computational study
completed for this closing chapter. First, Chapter 5 compares exchange coupling constants in
[Ni(tren)(SQ)]+ complexes. Various semiquinone and Îą-diimine ligands are examined to assess
the following postulate: is it possible to construct a NiII-SQ based system in which the exchange

13

coupling constant is small enough, in magnitude, and the spin density at the exchange center can
be monitored by EPR techniques, whether it be by CW-EPR, ENDOR, ESEEM and/or
HYSCORE? It is noted that, this research group may be the only research group attempting to
decrease the amount of exchange coupling. Groups are typically interested with increasing J so
as to increase the likelihood that the molecule/material will exhibit magnetic properties that
would be amenable to making usable molecular magnetic materials, such as single molecule
magnets11. Regardless of the progress of other groups, we aspire to gain knowledge of the spin
density at the exchange center concomitantly whilst knowing J in a series of derivatives with
symmetrically placed substituents on the SQ backbone; this study would produce the knowledge
necessary to construct molecular magnetic materials with a bottom-up type approach. In order to
reach such goals, the results in Chapter 5 point towards using

17

O as a spin label. The results

also show that for each SQ ligand, there is a dependence of J based on choice of substituent.
Additionally, this final chapter brings together the results of this dissertation and places them
all under a single light. The final chapter explains that it was the hope of this study, as a whole,
to assess if there would be appreciable differences in

13

C hyperfine coupling constants to

determine an experimental substituent effect. This would then be correlated with theoretical
results of hyperfine coupling constants and J values. And if there was a correlation between
hyperfine interactions and J what would that perpetuate for constructing molecular magnetic
materials?

That would be the full realization of this project, creating molecular magnetic

materials base on a “bottom up” approach.

14

REFERENCES

15

REFERENCES

(1)

Fehir, R. J. Differential Polarization of Spin and Charge Density in Phenoxy Radicals,
Semiquinone Radicals, and Ni(II) and Cr(III) Phenanthrenesemiquinone Complexes,
Michigan State University, 2009.

(2)

Shultz, D. A.; Sloop, J. C.; Washington, G. Jounal Org. Chem. 2006, 71, 9104–9113.

(3)

Kahn, O. Molecular Magnetism; VCH Publishers: New York, 1993.

(4)

Kambe, K. J. Phys. Soc. Jpn 1950, 5, 48–51.

(5)

Venegas-Yazigi, D.; Aravena, D.; Spodine, E.; Ruiz, E.; Alvarez, S. Coord. Chem. Rev.
2010, 254 (17–18), 2086–2095.

(6)

Bencini, A.; Carbonera, C.; Vaz, M. G. F. Dalt. Trans. 2003, 1701–1706.

(7)

Pierpont, C. Coord. Chem. Rev. 1981, 38 (1), 45–87.

(8)

Broze, M. J. Chem. Phys. 1967, 46 (12), 4891.

(9)

Prasad, B.; Lewis, A. R.; Plettner, E. Anal. Chem. 2011, 83 (1), 231–239.

(10)

Stegmann, H. B.; Dao-Ba, H.; Mäurer, M.; Scheffler, K.; Buchner, H.; Hartmann, E.;
Mannschreck, A. Magn. Reson. Chem. 1988, 26 (March), 547–551.

(11)

Hoeke, V.; Heidemeier, M.; Krickemeyer, E.; Stammler, A.; BĂśgge, H.; Schnack, J.;
Glaser, T. Dalt. Trans. 2012, 41, 12942.

16

Chapter 2. Quantifying Spin Density Delocalization in Phenanthrenesemiquinones

2.1 Introduction

The scope of this research is to theoretically and experimentally determine if unpaired electron
density can be displaced, delocalized or polarized by means of a substituent effect. To an extent,
this is the most rudimentary form of experiment for the chemist; create a molecule, take a
measurement, modify the molecule, take a measurement, etc. It is a simple means of altering the
physical properties and/or reactivity of molecules. Increasing the nucleophilicity of an organic
molecule in substitution reactions, changing the steric bulk on a ligand system to prevent
unwanted excited state geometries, these are a couple of examples where modifying chemical
structure changes both the physical properties and/or the chemical reactivity of a molecule. In
the interest of this type of study, the goal of this project is to learn how to better construct
molecular magnetic materials. The method for this goal is taking a simple redox active ligand
and attempting to manipulate its unpaired electron distribution by modifying the molecular
structure.
The system for studying spin properties affected by substituent choice are semiquinones.
Semiquinones have been vastly studied for several decades.

The richness of their redox

chemistry lends them to several fields of study including, but not limited to, valence
tautomerism1–4, electron transfer processes5–7, synthetic oxidants8,9 and molecular magnetic
materials10–13. The eventuality of this project points towards studying the magnetic properties of
exchange coupled, semiquinone-paramagnetic transition metal complexes (Chapter 4). Prior to
studying said exchange coupled molecules, we wish to learn more about spin density

17

delocalization within the ligand. In order to assess this inquiry, a combined computational and
experimental study has been undertaken concerning 3,6-R2-phenanthrenesemiquinones (3,6-R2PSQs).

Figure 2-1. Structure of 3,6-R2-PSQ radical anions with proton numbering scheme.
Another facet of this study is to elucidate spin density parameters from hyperfine coupling
constants and compare these findings to ZnII-PSQ, GaIII-PSQ and NiII-PSQ complexes. As of
recently, there has not been such a study undertaken in the literature. Labeling semiquinones to
extract spin density information through EPR techniques is not a new experimental method.
There have been limited examples in the literature that study the carbonyl oxygen spin density by
substituting
atoms14.

17

O for

16

O in para-semiquinones to measure the hyperfine coupling at the oxygen

Time-resolved EPR15 experiments as well as simple continuous wave-electron

paramagnetic resonance (CW-EPR) experiments have been undertaken to measure hyperfine
coupling constants in complex biological environments and simple organic molecules,
respectively. Time resolved studies of 1,4-naphthoquinone in Photosystem II revealed that
oxygen atoms were non-equivalent due to strong hydrogen bonding; and as a consequence to the
hydrogen bonding, the anisotropic coupling (T) was unable to be determined. In other studies,
typical CW-EPR experiments were conducted on mostly para-semiquionones16,17,18. The oxygen
hyperfine coupling was found to be of similar magnitude (7-10 G, 19-29 MHz) in each case of
18

the organic molecules; however, in molecules containing more oxygen atoms than the two
semiquinone groups, the hyperfine coupling was smaller. Attempts to measure anisotropic
contributions to the hyperfine coupling in chloranil radical anion were successful, but showed
variance in the hyperfine coupling in using various methods to produce the semiquinone from the
parent quinone17. This is an important note from the literature because it is similar to the
findings of this work. The method of reduction has an influence on the spin density map of the
molecule, due to the physical nature of the method used; e.g. using sodium versus potassium to
reduce the Q to the SQ. To generalize this statement, the method by which the radical is
produced has an influence on the spin density.
As it was stated in the introductory chapter, the method for measuring differences in
hyperfine coupling amongst semiquinones is isotopic labeling.
carbons will be exchanged for

13

Specifically, the carbonyl

C nuclei. There have been experimental studies using

study spin density distribution in semiquinones. Recently,

13

13

C to

C-labeling in quinones has been

most prominent in biological chemistry. Labeling carbonyl carbons19 as well as ring20 and
substituent carbons21 in ubisemiquinone-10 (a para-quinone) has proved to be a powerful
technique in uncovering structural information regarding the chemical environment; specifically
hydrogen bonding networks and the effect of the hydrogen bonding on the hyperfine coupling at
the carbonyl carbons. It was found that the carbonyl carbons were non-equivalent due to strong
hydrogen-bonding at one of the oxygen atoms.
In addition to magnetic resonance, density functional theory (DFT) calculations provide for a
theoretical basis of where the unpaired spin density resides in the open shell molecules. There
have been numerous DFT studies of semiquinones22–25.

DFT results allow for a deep

comparison of structural parameters, electronic structure, magnetic properties and other

19

quantities not easily obtained from experiment. It should be noted that although experimental
spin density values are compared to theoretical results, the series of molecules studied by DFT
were intended to uncover trends and not calculate the absolute magnitudes of the values.
Experimentally, there is a wealth of knowledge to be learned from conventional EPR. Fine
structure in CW spectra can be used to measure isotropic hyperfine, aiso, couplings. This can be
powerful information for determining wave function contribution across a molecule, electronic
structure, or even structural parameters. Indeed, the difference in line shape alone can indicate
qualitatively if a substituent effect is operative. The semiquinones and their

13

C isotopologues

studied in this chapter have been designed to determine, quantitatively, the effect of substituent
on spin delocalization/polarization in phenanthrenesemiquinones. Therefore, a combination of
CW-EPR, ESEEM and HYSCORE spectroscopic techniques have been undertaken to determine
both isotropic, aiso, and anisotropic, or dipolar, T, carbonyl carbon hyperfine splitting constants.
As detailed below, in the CW-EPR section, the 13C-aiso only accounts for the spin density at
the nucleus, and does not include the spin density residing in p-orbitals. The spin density
residing in p-orbitals is called the anisotropic hyperfine coupling. So, along these lines, ESEEM
and HYSCORE experiments have been carried out to determine the anisotropic portion of the
13

C -hyperfine coupling. ESEEM has been largely successful for our system due to the large

anisotropic portion of the hyperfine coupling. To add a second dimension to this study we have
also included HYSCORE results, which directly reveal frequency correlations arising from the
same hyperfine coupling. Although there may be small portions of anisotropy on the protons
from spin polarization of the radical, for the purpose of this study, the focus has been placed on
the 13C nuclei and the anisotropy on the protons have been ignored.

20

2.2 Experimental
General Methods. Reagents were purchased from commercially available sources and used
without further purification unless otherwise stated. [Carbonyl-13C]N,N’-dimethylformamide
was purchased from Cambridge Isotope Labs with 99%

13

C enrichment in 1 mL ampoules and

used as received. Stabilizer-free tetrahydrofuran was purchased from Fisher Scientific and either
distilled over sodium metal and benzophenone and freeze, pump, thawed to remove O2, or taken
from a dry still with a solvent bomb. Elemental analyses and electro-spray ionization ESI mass
spectra were obtained through the analytical facilities at Michigan State University. GC mass
spectra were obtained on a Hewlett Packard G1800D GCD system equipped with an electron
ionization detector or acquired by the analytical facilities at Michigan State University. NMR
spectra were collected on Agilent DDR2 500 MHz Spectrometers equipped with 7600A
autosamplers at the Max T. Rogers NMR Facility at Michigan State University.
Synthesis. 2,2’-Diformylbiphenyl (1). The procedure for the preparation of this molecule is
based on a literature procedure for a similar molecule26.

624 mg (2.0 mmol) g of 2,2’-

dibromobiphenyl was placed in a round bottom flask, evacuated with nitrogen and equipped with
a gas-flow adapter and dry still adapter. A minimal amount of dry THF was added to the flask
from a dry still and the flask was placed in a 0 °C bath. 5.0 mL (8.0 mmol) n-BuLi was added
drop wise, via syringe, over the course of five minutes; this changed the reaction mixture from
clear/colorless to cloudy/light yellow. After the addition the solution was stirred for 20 minutes
at 0°C. 0.93 mL (12.0 mmol) of Anhydrous DMF was then added drop wise, by syringe; this
changed the reaction mixture clear/colorless. This was then removed from the ice bath and
stirred for one hour. The reaction was then quenched with 30 mL of 1 M HCl (aq). The product
was extracted with dichloromethane (3 x 50 mL) and the combined organic layers were washed

21

with water and dried over MgSO4. The MgSO4 was filtered off and the CH2Cl2 was evaporated
to dryness to produce a yellow oil. This product was used without further purification. Yield:
(>100%). MS [GC m/z (rel. int.)]: C14H10O2+ 210.1 (10). 1H-NMR (500 MHz, CDCl3, δ): 9.82
(d, 2H, CHO), 8.05 (dd, 2H, 3 and 3’), 7.65 (td, 2H, 5 and 5’), 7.58 (tt, 2H, 4 and 4’), 7.34 (dd,
2H, 6 and 6’).
[1,1-Carbonyl-13C2]2,2’-Diformylbiphenyl (2). Same procedure used to synthesize 2,2’diformylbiphenyl. Yield: (>100%). MS [APCI+ (TOF) m/z (rel. int.)]: (C1213C2H10O4+H)+
213.1 (100 %). 1H-NMR (500 MHz, CDCl3, δ): 10.02 (s, 1H, CHO), 9.7 (s, 1H, CHO), 8.1 (m,
2H, 3 and 3’), 7.7 (td, 2H, 5 and 5’), 7.64 (tt, 2H, 4 and 4’), 7.39 (d, 2H, 6 and 6’).
Phenanthrene (3). The preparation for this molecule is based on a literature procedure27.
423 mg (2.01 mmol) of 2,2’-diformylbiphenyl was placed in a flask fitted with a reflux
condenser and taken up in acetic acid. The reaction flask was fitted with a gas-flow adaptor and
placed under N2. The reaction flask was placed in an oil bath and heated to reflux. In a separate
vial, a 0.6 M solution of N2H4∙H2O in Acetic acid was prepared by adding 4.1 mL AcOH to
0.123 mL (2.5 mmol) N2H4∙H2O. This solution was added to the refluxing mixture over the
course of one hour, by pipette. The reaction turned dark red-orange as the N2H4 solution was
added. The reaction was then refluxed for an additional two hours. The reaction vessel was then
removed from the oil bath to cool down to ambient temperature and the solvent was evaporated
to dryness under a stream of N2 overnight. The light brown residue was taken up in C6H6 and
passed through an alumina (neutral) column. The solution was evaporated to dryness and a light
yellow powder was obtained. Yield: (81.4 %). MS [GC (EI) m/z (rel. int.)]; C14H10+ 178.23
(100). 1H-NMR (500 MHz, CDCl3, δ): 8.69 (d, 2H, 4 and 5), 7.88 (dd, 2H, 1 and 8), 7.73 (s, 2H,
9 and 10), 7.64 (td, 2H, 3 and 6), 7.58 (td, 2H, 2 and 7).

22

[9,10-Carbonyl-13C2]Phenanthrene (4). Same procedure used to synthesize phenanthrene.
Yield: (64%) MS [GC (EI) m/z (rel. int.)]; C1213C2H10+ 180.1 (100 %).

1

H-NMR (500 MHz,

CDCl3, δ): 8.68 (d, 2H, 4 and 5), 7.98 (s, H, 9), 7.88 (m, 2H, 1 and 8), 7.68-7.55 (m, 4H, 3 and 6,
2 and 7), 7.46 (s, H, 10).
9,10-Phenanthrenequinone (5). The preparation for this molecule is based on a literature
procedure28. 382 mg (3.802 mmol) of CrO3 was dissolved in AcOH and the flask was placed in
an ice bath. In a separate flask, 292 mg (1.6 mmol) phenanthrene was taken up in AcOH, and
then added in portions over the course of 2.5 hours, at 0°C. After the addition was complete, the
dark orange reaction mixture was warmed to room temperature and then stirred until a dark
green reaction mixture persisted (usually 2-3 days). The reaction was quenched by pouring into
~100 mL of H2O; this produces a yellow voluminous precipitate.

The product was then

extracted with EtOAc (3 x 50 mL), washed with ~10% NaHCO3 (aq) and H2O, and the combined
organic layers were dried over MgSO4. The MgSO4 was filtered off and the EtOAc solution was
evaporated to dryness under a stream of N2. The yellow residue was recrystallized from hot
MeOH to produce orange-yellow plates. Yield: (45%) MS [APCI+ (TOF) m/z (rel. int.)];
(C14H8O2+H)+ 209.1 (100). 1H-NMR (500 MHz, CDCl3, δ): 8.18 (dd, 2H, 4 and 5), 8.01 (d, 2H,
1 and 8), 7.71 (td, 2H, 3 and 6), 7.46 (td, 2H, 2 and 7).
[9,10-Carbonyl-13C2]9,10-Phenanthrenequinone (6). This molecule was prepared with the
same procedure as 9,10-phenanthrenequinone. Yield: (43%) MS [APCI+ m/z (rel. int.)];
(C1213C2H8O2+H)+ 211.1 (100). 1H-NMR (500 MHz, CDCl3, δ): 8.18 (d, 2H, 4 and 5), 8.01 (d,
2H, 1 and 8), 7.71 (t, 2H, 3 and 6), 7.46 (t, 2H, 2 and 7).
2,2’-Dibromo-5,5’-dimethoxybiphenyl (7).

The preparation of this molecule has been

previously published29. 2.5 g (11.65 mmol) 3,3’-Dimethoxybiphenyl was dissolved in AcOH

23

and 2.0 mL (39.02 mmol) Br2 was added drop-wise. Almost immediately a yellow precipitate
forms. The reaction was stirred at room temperature for 2 hours and then the yellow solid is
filtered off and recrystallized from hot EtOH to afford colorless crystals. MS [GC m/z (rel. int.)];
(C14H4O2)+ 371.9207 (100%). 1H-NMR (500 MHz, CDCl3, δ): 7.54 (d, 2H, 6 and 6’), 6.83 (dd,
2H, 4 and 4’), 6.80 (d, 2H, 5 and 5’), 3.82 (s, 6H, OMe).
2,2’-Diformyl-5,5’-dimethoxybiphenyl (8). The procedure for this molecule is based on a
literature procedure26. 2.14 g (5.5 mmol) of 2,2’-dibromo-5,5’-dimethoxybiphenyl was placed in
a round bottom flask evacuated with nitrogen and equipped with a gas-flow adapter and dry still
adapter. A minimal amount of dry THF was added to the flask from a dry still and the flask was
placed in a -77°C bath (crushed CO2). 17.1 mL (27.4 mmol) n-BuLi was added drop wise, via
syringe, over the course of five minutes; this changed the reaction mixture from clear/colorless to
cloudy/yellow. After the addition the solution stirred for 20 minutes at 0°C. 2.85 mL (37 mmol)
of anhydrous DMF was then added drop wise, by syringe; this changed the reaction mixture to
clear/colorless. This was then removed from the bed of crushed CO2 and stirred for one hour.
The reaction was then quenched with 30 mL of 1 M HCl (aq). The product was extracted with
dichloromethane (3 x 50 mL) and the combined organic layers were washed with H2O and dried
over MgSO4. The MgSO4 was filtered off and the CH2Cl2 solution was evaporated to dryness to
produce a yellow oil. This product was used without further purification. Yield: (72%).

1

H-

NMR (500 MHz, CDCl3, δ): 9.69 (s, 2H, CHO), 8.04 (d, 2H, 3 and 3’), 7.08 (dd, 2H, 4 and 4’),
6.8 (d,2H, 6 and 6’), 3.89 (s, 6H, OMe).
[2,2’-Carbonyl-13C2]2,2’-Diformyl-5,5’-dimethoxybiphenyl (9).

This molecule was

prepared in the same way as 2,2’-Diformyl-5,5’-dimethoxybiphenyl. Yield: (>100%). 1H-NMR

24

(500 MHz, CDCl3, δ): 9.85 (s, H, CHO), 9.8 (s, H, CHO), 8.01 (m, 2H, 3 and 3’), 7.08 (dd, 2H, 4
and 4’) 6.8 (d, 2H 6 and 6’), 3.89 (s, 6H, OMe).
3,6-Dimethoxyphenanthrene (10). This molecule was prepared in the same fashion as
phenanthrene (3). Yield: (58%). MS [GC (EI) m/z (rel. int.)]: C16H10O4+ 238 (100%). 1H-NMR
(500 MHz, CDCl3, δ): 7.93 (d, 2H, 4 and 5), 7.78 (d, 2H, 1 and 8), 7.54 (s, 2H, 1 and 9), 7.22
(dd, 2H, 2 and 7), 4.02 (s, 6H, OMe).
[9,10-Carbonyl-13C2]3,6-Dimethoxyphenanthrene (11). This molecule was prepared in the
same way as phenanthrene (3). Yield: (52%). MS [GC (EI) m/z (rel. int.)]: C16H10O4+ 240.1059
(100%). 1H-NMR (500 MHz, CDCl3, δ): 7.93 (d, 2H, 4 and 5), 7.78 (m, 2H, 1 and 8), 7.7-7.62
(distorted m, H, 9), 7.45-7.36 (distorted m, H, 10) 7.22 (dd, 2H, 2 and 7), 4.02 (s, 6H, OMe).
3,6-Dimethoxyphenanthrenequinone (12).

This molecule was prepared in the same

fashion as 9,10-phenanthrenequinone (5). Recrystallized Yield: (44%). MS [GC (EI) m/z (rel.
int.)]: C16H10O4+ 268 (35%). 1H-NMR (500 MHz, CDCl3, δ): 8.19 (dt, 2H, 2 and 7), 7.37 (d, 2H,
4 and 5), 6.96 (dd, 2H, 1 and 8), 3.75 (s, 6H, OMe).
[9,10-Carbonyl-13C2]3,6-Dimethoxyphenanthrene-9,10-quinone (13). This molecule was
prepared with the same procedure as 9,10-phenanthrenequinone (5).

Recrystallized Yield:

(59%). MS [GC (EI) m/z (rel. int.)]: C16H10O4+ (). 1H-NMR (500 MHz, CDCl3, δ): 8.19 (dt, 2H,
2 and 7), 7.37 (d, 2H, 4 and 5), 6.96 (dd, 2H, 1 and 8), 3.75 (s, 6H, OMe).
Physical Measurements. X-ray Crystallography. Single crystal x-ray diffraction data was
collected for compound 13 on a Bruker APEX-II CCD area detector single crystal diffractometer
at the MSU Center for Crystallographic research. The crystals were irradiated with graphitemonochromatic copper Kα radiation (λ = 1.5418 Å). Data were collected at 173 K with an
Oxford Cryosystems low temperature device. Unit cell parameters were determined from least

25

squares analysis and integrated with the program SAINT v8.34A. The structures were solved
using direct methods in the program Olex2 and refined with XL. Anisotropic thermal parameters
were refined for all non-hydrogen atoms while hydrogen atom positions were calculated using
the riding model.
EPR Spectroscopy. All EPR data were collected on a Bruker Elexsys E-680X spectrometer,
operating at X-band frequency (9.42-9.71 GHz). Low temperature experiments were performed
between 4-30 K using an Oxford Instruments liquid helium flow system equipped with a CF-935
cryostat and an ITC-503 temperature controller. Quinones were reduced in a glove box with
excess sodium metal in THF. The resulting semiquinone solutions were then filtered through a
glass frit. The filtered solutions were placed in modified Wilmad Glass SQ-706 EPR tubes with
3.8 mm diameter. Prior to sample preparation, the tubes were modified in the glass blowing lab
at Michigan State University to contain a grated seal and vacuum adapter port. A vacuum was
pulled on the sample tubes and then the vacuum ports were blown off at the grated seal by
hydrogen torch. Prior to collecting CW-EPR data, microwave power-dependence experiments
were conducted to ensure non-saturation of the EPR-signal. Continuous wave spectra were
simulated using the garlic module of Easy Spin 4.530–32.
All pulsed experiments were conducted at X-band frequency and on the same spectrometer
as stated above. ESEEM spectra were obtained using a three pulse sequence of 90°-τ-90°-T-90°
with 90° pulse lengths of 16 ns. Tau values were chosen to suppress hyperfine contributions
from weakly coupled protons. Echo amplitudes were integrated at FWHM and the T values were
scanned from 40 ns to 6 Îźs with a 12 ns time increment. A four step phase-cycling procedure
was used to eliminate unwanted 2-pulse echoes33. Raw data were processed by phase shifting
the spectra to remove the dispersion component from the time domain data. The in-phase

26

portion of the data were normalized by dividing by a biexponential background decay function.
These normalized data were then further processed by subtracting a second degree polynomial
(DC-component of the signal), followed by application of a Hamming window. The data set was
then zero-filled from 512 points to 1024 points and Fourier transformed. Frequency spectra are
displayed as an absolute value of the transformed time domain data.

13

C and

12

C spectra are

presented as one divided spectrum at each magnetic field value. After normalizing each time
domain spectrum individually, the

13

C spectrum was divided by the

12

C spectrum and the same

work up procedure described above was applied34.
HYSCORE experiments were carried in a similar manner to the ESEEM experiments but
with a four-pulse sequence of 90°-τ-90°-t1-180°-t2-90° with a 180° pulse length of 16 ns. Echo
amplitudes were measured at FWHM and τ values were chosen as stated above. T1 and t2 values
were scanned from 40 ns to 2 Îźs with 16 ns time increments. Data sets were 128 points in both
dimensions. A 4-step phase cycling program was used to filter out unwanted 2-pulse echoes35.
The real, or in-phase portion of the HYSCORE data were processed, in both dimensions, by first
subtracting a 2° polynomial, then tapering the data with a Hamming window and zero-filling to
256 points. A two-dimensional fast Fourier transform was applied and absolute values taken.
Data are presented as a two-dimensional contour plot. Each contour plot is presented at a
threshold of 5-20% of the maximum peak amplitude (z-direction).
Spectra were fit using customized MATLAB scripts where EasySpin was used to perform
spectral simulations and the fminsearch function in Matlab was used to optimize the calculation
parameters. Fminsearch is a function that finds the minimum of an unconstrained multivariable
function using a derivative-free method36. Residuals were calculated by squaring the difference

27

of the experimental and simulated amplitudes. Goodness of fit for each spectral fit was assessed
by calculating a χ2 value with the following equation:

𝜒𝑛2 =

𝑒𝑥𝑝
𝑁 (𝑦𝑖
𝛴𝑖=1

− 𝑦𝑖𝑐𝑎𝑙𝑐 )
𝜎2
𝑁−𝐿

2

(2.1)

where the sum is taken over N points, yiexp and yicalc are the experimental and calculated y-axis
values, σ is the standard deviation fixed at 2% of the maximum spectral amplitude and L is the
number of adjustable variables.
Calculations. Calculations were performed with the Gaussian0337 computational chemistry
package on the supercomputing cluster HPCC (High Performance Computing Center) at
Michigan State University. Structures were drawn in GaussView438 and were optimized on
phenanthrenesemiquinones (S=1/2) using the unrestricted formalism with the B3LYP39–41
functional and 6-311g(d,p)42–45,46 basis set. Frequency calculations were carried out to ensure
optimizations reached global minima. Single point calculations were performed at the same
level of theory as geometry optimizations. In addition the “pop=(Full,NPA)” keyword command
was used to calculate all molecular orbital contributions (‘Full’) from the atoms and spin
densities computed by natural population analyses47 (NPA) from the NBO (Natural Bond Order)
package.

28

2.3 Results and Discussion.
Density Functional Theory. Density functional theory was employed to assess the spin
density distribution based on the substituent effect of 3,6-R2-PSQs (where R = NH2, OMe, H, Br,
CN, CF3, NO2).
method.

The unrestricted formalism was used as opposed to the restricted open shell

The unrestricted approach polarizes all molecular orbitals into singly occupied

molecular orbitals (SOMOs) for all electrons whereas restricted open shell only places the
radical in a SOMO, all other electrons are paired in traditional molecular orbitals. Geometry
optimizations were based on the crystal structure of phenanthrenequinone and derivatives were
drawn by modifying phenanthrenequinone in GaussView4. Geometries were optimized and
single point calculations were calculated at the UB3LYP/6-311g(d,p) level of theory.

All

derivatives were optimized to global minima and contained no imaginary frequencies. Spin
density was calculated by both Mulikan and NPA analysis within the Gaussian03 software
program. The NPA spin densities were calculated by subtracting the beta spin values from the
alpha spin values. The spin densities presented are an average of each specific atom pair
assessed; e.g. carbonyl carbons were averaged and presented as a single value.
The primary focus of this study was to compare carbonyl carbon spin density to oxygen
spin density in PSQs. Although the majority of the unpaired spin density resides on the oxygen
atoms, our interest is in determining whether labeling the carbonyl carbons with carbon-13 atoms
is a “good” spin label for semiquinones. This method, as stated previously, has some precedence
in the literature but for our interests it would be advantageous to use the carbonyl carbon as a
marker for spin density variances in semiquinones with similar back bone structure. So by
investigating variances in spin density through a theoretical lens, it will be apparent as to whether

29

or not a carbonyl carbon spin label has the potential to be effective in measuring hyperfine
coupling constants effectively.
0.26
H
0.25

CF3

Average O (ρ)

0.24

Br

OMe

NH2

CN

0.23

0.22

0.21
NO2
0.2
0.04

0.05

0.06

0.07

0.08

0.09

0.1

Average Carbonyl C (ρ)
Figure 2-2. Average carbonyl carbon versus average carbonyl oxygen spin density in 3,6-R2PSQs.

Depicted in Figure 2 is a graph that compares the carbonyl carbon spin density to oxygen
spin density. There was no linear correlation, but a general trend seems to be operative in that as
average carbonyl carbon (ρ) increases, electron donating character increases. These results
suggest that changes in carbonyl carbon spin density can be related to a substituent effect. The
average oxygen (ρ) also increases with electron donating character, however, to a lesser extent,
relative to carbon. There is somewhat of a trend from -NO2 to -Br, however, once an actual
electron donating substituent is considered, the trend devolves into clutter. Indeed, this poses

30

grounds for using

13

C instead of

17

O. To restate, the purpose of this exercise is to assess the

ability of the carbonyl carbons to be a physical marker for tracking spin density, not necessarily
to find a direct relationship between the spin density on the carbonyl carbon atoms and oxygen
atoms.
The majority of the spin density, in the series of compounds resides on the oxygen atoms,
holding a maximum of 83% (NO2) of the spin density and minimum of 74% (NH2), between the
two oxygen atoms and two carbon atoms. The carbonyl carbons contribute 26% (NH2) at the
most and 17% (NO2) at the least; figures 3 and 4 show this trend quite vividly.
0.34
NH2
OMe

0.33

Br

0.32

CF3

0.31

Total C=O (ρ)

H

0.30

CN
0.29
0.28
0.27
0.26
0.25

NO2

0.24
0.2

0.21

0.22

0.23
Average O (ρ)

0.24

Figure 2-3. Average oxygen spin density versus total carbonyl spin density.

31

0.25

0.26

The most notable observation here is that carbonyl carbon spin density (Figure 3) tracks with
total carbonyl spin density more than oxygen spin density (Figure 4). Although there is no linear
0.34

H
OMe NH2

0.33
Br

0.32
CF3

Total C=O (ρ)

0.31
0.30
CN
0.29
0.28
0.27
0.26
0.25
0.24
0.04

NO2
0.05

0.06
0.07
Average C (ρ)

0.08

Figure 2-4. Average carbonyl spin density versus average carbonyl carbon spin density.

32

0.09

correlation in Figure 4, the increasing amount of spin density on the carbonyl carbons (with
respect to electronegativity of the substituent) suggests that the carbonyl carbon spin density has
a greater dependence on the electronegativity of the substituent than the oxygen atoms. It is
important to note that electronegativity is a property of all the electrons in the system and the
spin is based solely on the electron occupying the SOMO. The carbonyl group contributes 38%
to the composition of the SOMO, with 18% from oxygen and 10% from carbon (when R = H).
From this short analysis, it can be concluded that using 13C as a means of tracking changes in
spin density could be a useful method. The changes in spin density are more drastic than 17O
and even have a general trend of increasing with electron donating ability.
Synthesis.

Recent EPR studies of labeled semiquinones and the synthesis of such

compounds has been largely dominated by the area of bioorganic chemistry. Specifically,
studying the spin density distribution of ubisemiquinone has been an area of interest due to its
involvement in electron transfer chemistry in oxidases. This para-quinone is a six membered
quinone with methyl and methoxy groups nestled between the carbonyl groups, with an
additional polymeric side chain protruding from the opposite side of the ring. Similar to the
study presented here, the carbonyl carbons have been replaced by 13C atoms in favor of tracking
spin density changes19,48. The labeled molecules were achieved through several synthetic steps
and required several

13

C sources. In regards to this field of study, the ring carbons20 as well as

the methyl and methoxy substituents21 have also been

13

C-labeled and studied with EPR

methods. Along the lines of isotopic labeling of quinones and investigating their spin properties,
the other investigations found in the literature revolve around switching the carbonyl
for 17O atoms15–17.

33

16

O atoms

The syntheses of the organic molecules were based on procedures previously developed in
the literature. The “total synthesis” of this model was realized through trial and error of
retrosynthetic analysis. The procedure that was arrived at can be seen in Figure 5. Reducing the
parent quinone and cleaving the carbon-oxygen double bonds affords phenanthrene.
Phenanthrene opens up possibilities for placing carbon-13 atoms in the 9 and 10 positions of
phenanthrenequinone. Splitting the 9,10 carbon-carbon double bond into two vinyl groups
allows for the possibility of creating a ring closing metathesis reaction where the olefin groups of
2,2’-divinylbiphenyl would be clipped together by a first or second generation Grubbs’
catalyst26. However, obtaining the divinyl-biphenyl species would require a step where an
aldehyde is transformed into an olefin (perhaps by a Wittig reaction). In the interest of keeping
the number of reactions to a minimum, it was decided to pursue a coupling method for
aldehydes, to produce a double bond from the aldehyde carbons. The final retrosynthetic step
involves

removing the

aldehyde

group

and

replacing

it

with

a

leaving

group.

Figure 2-5. Retrosynthetic analysis for synthesizing a di-13C-labeled phenanthrenequinone.

34

Figure 2-6. General synthetic scheme for [13C2]phenanthrenequinone.
The general synthetic route for the 13C-labeled compounds can be seen in Figure 6. The 13Clabeled diformyl-biphenyl compounds (both -H, compound 2, and -OMe, compound 9) were
achieved by first adding nBuLi to 2,2’-dibromobiphenyl at low temperature, under N2, during
which lithium-bromide exchange produces a stable di-aryl-carbanion (provided the cryogenic
temperature is held constant). The carbanion is then quenched by adding [ 13C]DMF to the
reaction; the carbanion attacks the carbonyl carbon forming a new carbon-carbon single bond.
The carbonyl oxygen is stabilized by the lithium cation until the reaction is quenched by aqueous
hydrochloric acid; it has been hypothesized26 that by protonating the nitrogen, the carbonyl pi
bond reforms and releases dimethylamine gas to complete the formylation. Observations are
consistent with this hypothesis as upon addition of acid to the reaction mixture, a colorless gas is
formed. This gas was subsequently blown off with N2.
The

13

C-labeled diformyl biphenyl species were then subjected to a reductive

homocoupling. The carbonyl groups of diformyl biphenyl were activated by dissolving the
compound in AcOH. Then N2H4 was added to the diformyl-biphenyl compounds, at reflux, in
portions over the course of two hours. After each addition, the reaction mixture became darker
and more orange in color. The authors have hypothesized27 that the intermediate for this reaction
is an eight-membered ring

35

Figure 2-7. Proposed intermediate formation and expulsion of N2 for synthesizing phenanthrene
from 2,2’-diformyl biphenyl. The figure was reproduced from Reference 22.
presumably produced by a typical Schiff base reaction. The authors found that when the
aldehyde proton was replaced with an alkyl or aryl group, the stable diazocine was obtained, but
the reaction involving the aldehyde groups expelled N2 and left phenanthrene as the sole
product49.

After releasing N2, the imine carbons then clip together to produce [9,10-

13

C2]phenanthrene (Figure 7). Finally, phenanthrene was oxidized by a modified Jones oxidation

with CrO3 and AcOH to afford the [9,10-carbonyl-13C2]phenanthrenequinone.
The synthesis for 3,6-(OMe)2-phenanthrenequinone was performed in the same fashion as
phenanthrenequinone.

Ground State Characterization. X-ray Crystallography. X-ray quality crystals of 13
were grown from slow evaporation of the CDCl3 solvent in an NMR tube. The structure was
found to be in the P21/c space group with a residual of 0.038.

36

Figure 2-8. Crystal structure of [9,10-13C2]3,6-(OMe)2-PhenQ (13). The structure is displayed
with thermal ellipsoids at 50% probability.
Table 1 shows the selected bond lengths and angles for 13. The carbonyl bonds were found to be
symmetrical, and of typical carbon-oxygen double bond length. The carbonyl carbon-oxygen
bond lengths are similar to the carbonyl carbon-oxygen bond lengths for 9,10phenanthrenequinone, 1.215 Å (average between 2 crystal morphologies), as well as the intradiol
distance (C-C bond connecting carbonyl carbons), 1.52 Å (average between two crystal
morphologies)50. The carbon-carbon bond lengths of the central ring were found to be longer
than typical aromatic carbon-carbon bond lengths, which is due to the cross-conjugated nature of
the quinone group and the phenanthrene fused ring system. The C3-C8 and C9-C14 bond
lengths are ~0.6-0.7 Å shorter than the C2-C3 and C14-C1 bond lengths, respectively, giving
credit to the resonance structure pictured below.

37

Figure 2-9. Resonance structure of 3,6-(OMe)2-PhenQ.
Table 2-1. Selected bond lengths and angles for compound 13.
Bond Distances (Å)

Bond Angles (Degrees)

O1-C1

1.2176 (15)

O1-C1-C2

118.76 (11)

O2-C2

1.2160 (15)

O1-C1-C14

123.34 (12)

C1-C2

1.5421 (17)

C14-C1-C2

117.90 (10)

C2-C3

1.4720 (16)

O2-C2-C1

119.21 (10)

C3-C8

1.4152 (16)

O2-C2-C3

123.25 (11)

C8-C9

1.4851 (16)

C3-C2-C1

117.54 (10)

C9-C14

1.4109 (16)

C14-C1

1.4697 (16)

Table 2-2. Crystal structure parameters for compound 13.
Formula

C16H12O4

Molecular Weight

268.26

Crystal System

Monoclinic

Space Group

P21/c

a/Å

6.8451 (1)

38

Table 2-2. (cont’d)
b/Å

21.5856 (3)

c/Å

8.0850 (1)

β/°

69.4644 (7)

Volume/Å3

1187.01 (3)

Z

4

Density

1.501 Mg∙m-3

Temperature

173 K

Reflections Measured

8695

Independent Reflections

2284

Observed Reflections with I>2σ(I)

2137

Îź (Cu KÎą)/mm-1

0.90

R1[F2>2 σ(F2)]

0.038

wR(F2)

0.106

NMR Spectroscopy.

1

H-NMR spectra of the labeled and non-labeled molecules show

splitting in the NMR spectra that correspond to 13C-labeling. Due to the I = ½ nature of the 1H
and 13C nuclei, the coupling of the two nuclei should produce a splitting of the aldehyde proton
in the 1H-NMR spectrum of the labeled compound relative to the non-labeled molecule. This is
precisely what is observed; the aldehyde protons correspond to one resonance at a chemical shift
of 9.83 ppm in the

12

C-diformylbiphenyl and integrate to two protons. When compared to the

13

C-diformylbiphenyl, the single resonance splits into two separate resonances (a rather strongly

coupled doublet, one at 10.0 ppm and a second at 9.65 ppm). A secondary advantage of

39

13

C-

labeling is proton assignment. The doublet at 8.05 in the
doublet of doublets in the
that resonance.

13

12

C spectrum is further split into a

C spectrum, which supports the assignment of the 3 and 3’ 1H’s at

The aromatic and aldehyde regions of the 1H-NMR of (CHO)2-Ph2 and

(13CHO)2-Ph2 spectra are seen in Figure 10.

Figure 2-10. 1H-NMR spectra of (CHO)2-Ph2 (bottom) and (13CHO)2-Ph2 (top). Bold numbers
are relative integrals with respect to the aldehyde proton(s).
This splitting of the aldehyde protons is also observed in the

12

C and 13C dimethoxy derivatives

of diformyl-biphenyl (8 and 9). As described above, the aldehyde protons are split by the

13

C-

carbonyl carbons, and produce two singlets at 9.69 and 10.05 ppm, as well as incurring a new
splitting pattern in protons 3 and 3’ (8.02 ppm) by going from a doublet (12C spectrum) to a
doublet of doublets (13C spectrum).

40

Figure 2-11. 1H-NMR spectra of (CHO)2-(OMe)2-Ph2 (bottom) and (13CHO)2-(OMe)2-Ph2 (top).
This effect can also be seen in the spectra of the labeled and non-labeled phenanthrene. The
9 and 10 protons have been assigned to the resonance at 8.03 ppm in the

12

C-phenanthrene

spectrum (Figure 11); however, in [13C2]phenanthrene this single resonance also splits into two
prominent resonances, 7.99 and 7.46. This spectrum is not as easily understood as the diformylbiphenyl spectrum; the two singlets only integrate to 0.5 1H each, leaving the other half grouped
into resonances (multiplets) at 7.6 and 7.88. The dimethoxy derivatives of diformyl-biphenyl
and phenanthrene exhibit this same effect.

CW-EPR. In an EPR experiment, electron spin states are probes in the presence of an
applied magnetic field.

The electron Zeeman effect is the main interaction which is a

41

consequence of the interaction of the unpaired electron’s magnetic moment and the applied
magnetic field as described by:
̂ = −𝝁 ∙ 𝑩 =
𝐻

𝑔𝛽𝑒
̂∙𝑩
𝑺
ℏ

(2.2)

where g for a free electron is2.0023, B is the applied magnetic field, β is the electron Bohr
magneton (9.274e-24 J∙T-1) and ̂
𝑺 is the electron spin operator. Then, if the z-axis is coincident
with the axis of the laboratory field, then 𝐵𝑜 ∥ 𝑧, the dot product becomes
̂=
𝐻

𝑔𝛽𝑒 𝐵𝑜
𝑆̂𝑧
ℏ

(2.3)

For an S = ½ spin system, the stabilized microstate is 𝑚𝑠 = − 1⁄2 (β), and is aligned parallel
with the applied magnetic field; the destabilized state is then 𝑚𝑠 = 1⁄2 (α) and is aligned
antiparallel with the magnetic field. These spin states are eigenfunctions of 𝑆̂𝑧 , so for spin states
−ℏ
ℏ
|𝛽⟩ and |𝛼⟩, 𝑆̂𝑧 |𝛽⟩ = |𝛽⟩ and 𝑆̂𝑧 |𝛼⟩ = |𝛼⟩. Then, the time-independent Schrödinger equation
2
2

can be applied and the energies can be obtained from the energy expression below:
𝐸 = 𝑚𝑠 𝑔𝛽𝐵

(2.4)

In a conventional EPR experiment, continuous microwave radiation, of a fixed frequency,
is applied to the sample and the magnetic field, B, is scanned to achieve resonance. The resultant
EPR absorption spectrum is displayed as a first derivative, as a result of field modulation and
phase-sensitive detection.

Additionally, these conditions provide for accurate g-value

measurement, and enhanced sensitivity. Figure 12, below, shows this effect pictorially:

42

Figure 2-12. Splitting of ι and β electron orbitals in an applied magnetic field.
The above case is for a free radical, that is not coupled to any nuclei.

When the unpaired

electron spin is coupled to a nuclear spin, the nuclear Zeeman and hyperfine terms are introduced
to the Hamiltonian.
̂ ∙ 𝑨 ∙ 𝑰̂
̂ = 𝑔𝑒 𝛽𝑒 𝐵𝑆̂𝑧 − 𝑔𝑁 𝛽𝑁 𝐵𝐼̂𝑧 + 𝑺
𝐻

(2.5)

The second term of the Hamiltonian is the nuclear Zeeman where 𝐼̂ is the nuclear spin angular
momentum operator.

The third term is the hyperfine coupling parameter where A is the

hyperfine coupling tensor, which is a 3x3 matrix that describes the coupling between the electron
and nuclear magnetic moments. If the sample (paramagnetic molecule) is tumbling fast through
solution, relative to the timescale of the EPR experiment, then the hyperfine coupling tensor

43

becomes Aiso∙1. The bold 1 is the unit matrix; with this replacement, the Hamiltonian can be
written as follows:
̂ = 𝜔𝑆 𝑆̂𝑧 − 𝜔𝐼 𝐼̂𝑧 + 𝐴𝑖𝑠𝑜 (𝐼̂𝑥 𝑆̂𝑥 + 𝐼̂𝑦 𝑆̂𝑦 + 𝐼̂𝑧 𝑆̂𝑧 )
𝐻
where ωS and ωI are simply compact ways of writing the Zeeman pre-factors (e.g., 𝜔𝑆 =

(2.6)
𝑔𝛽𝑒 𝐵∘
ℏ

).

These are also known as the electron and nuclear Larmor frequencies in angular units. The xand y- spin operator terms can be ignored from equation 2.6 if Aiso << ℏωS, and this yields
̂ = 𝜔𝑆 𝑆̂𝑧 − 𝜔𝐼 𝐼̂𝑧 +
𝐻

2𝜋𝑎𝑖𝑠𝑜
𝑆̂𝑧 𝐼̂𝑧
ℏ

(2.7)

and aiso is the isotropic hyperfine coupling with units of MHz. Using the above form of the
Hamiltonian to operate on the spin product states |𝑚𝑆 𝑚𝐼 ⟩, yields the following energy states:
𝐸1 = −

ℏ𝜔𝑆 ℏ𝜔𝐼 ℎ𝑎𝑖𝑠𝑜
−
−
2
2
4

𝐸2 = −

ℏ𝜔𝑆 ℏ𝜔𝐼 ℎ𝑎𝑖𝑠𝑜
+
+
2
2
4

𝐸3 =

ℏ𝜔𝑆 ℏ𝜔𝐼 ℎ𝑎𝑖𝑠𝑜
−
+
2
2
4

𝐸4 =

ℏ𝜔𝑆 ℏ𝜔𝐼 ℎ𝑎𝑖𝑠𝑜
+
−
2
2
4

E1 and E2 correspond to the |𝛽𝛼𝑁 ⟩ and |𝛽𝛽𝑁 ⟩ states, respectively, in the lower spin manifold, and
E3 and E4 correspond to |𝛼𝛼𝑁 ⟩ and |𝛼𝛽𝑁 ⟩ states, respectively. The allowed transitions are
governed by the selection rules ΔmS = ± 1 and ΔmI = 0; thus the transitions are E1 → E3 and E2
→ E4. The change in energy, with respect to these transitions is then
∆𝐸 = ℏ𝜔𝑆 ±

ℎ𝑎𝑖𝑠𝑜
2

and the splitting between the two nuclear states within each electron spin state are

44

(2.8)

∆𝐸𝑛𝑢𝑐 = ℏ𝜔𝐼 ±

ℎ𝑎𝑖𝑠𝑜
2

(2.9)

These transitions and their associated energies are reflected below, in Figure 13.

Figure 2-13. Electronic and nuclear Zeeman effects. Allowed EPR transitions are highlighted
by double-headed arrows in the MI manifold.
Continuous Wave EPR spectra were collected at various temperatures and field modulation
values. It has been documented that large field modulation values can affect lineshape, so as to
cause line distortions to resemble unresolved hyperfine splitting51. In light of these observations,
room temperature CW spectra were collected at 20 mG, 500 mG and 1 G modulation amplitudes.
The 20 mG spectra have been simulated and fit, and are discussed in the body of this chapter
while the 500 mG and 1 G spectra are located in the Chapter 2 Appendix.

45

The 20 mG, room temperature CW spectra were fit with the following parameters: g-factor,
Voigtian (Gaussian + Lorentzian) peak-to-peak line-broadening (abbreviated lwpp), and
isotropic hyperfine coupling parameters. The g-factors were taken from Îť-max (inflection point)
of the 1 G spectra. Initial guesses for 1H aiso values came from previous works and Gaussian03
calculations.

1

H aiso values were varied until the width of the simulated spectrum matched the

spectral width of the experimental spectrum. Then, line width parameters were adjusted to
accommodate both Gaussian and Lorentzian parameters of the lineshape. Once a decent fit was
achieved, all simulation parameters were fit with the fminsearch tool in Matlab. This was
repeated until the residual value reached a minimum.

Figure 2-14. EPR spectrum of PSQ. Reproduced from Reference 52.
The

EPR

spectrum

of

PSQ

has

previously

been

recorded52

(Figure

14).

Phenanthrenequinone was reduced with zinc metal in alkaline solution, and the spectrum was
found to have five groups of five lines, which corresponds to a system where there are two
groups of four equivalent protons. It was not indicated which sets of protons were considered to
be equivalent and only 26 lines were observed, possibly due to overlapping lines and low
resolution.
For a molecule with an equivalent pair of protons the number of lines is predicted by the
formula 2nI + 1, where I is the nuclear spin and n is the number of equivalent nuclei. For
example, four lines are predicted for the methyl radical and four lines are observed, in a 1:3:3:1
46

peak ratio53,54; I = ½ for protons and n = 3. However, for a system where there are several sets of
equivalent nuclei, the equation becomes a product of the prediction for the individual sets. For
PSQ with four sets of two equivalent protons each, the number of lines expected would be
represented by
# 𝑜𝑓 𝐿𝑖𝑛𝑒𝑠 = [(2𝑛𝐼 + 1)(2𝑚𝐼 + 1)(2𝑝𝐼 + 1)(2𝑞𝐼 + 1)]

(2.10)

where n = m = p = q = 2; the result being 81 lines. As stated above, only 26 lines were observed,
which leads to the conclusion that there are unresolved hyperfine couplings and/or overlapping
lines.
Phenanthrenesemiquinone has also been studied by Kazuhiro in the 1960s.55 Kazuhiro
studied the EPR dependence on alkali and alkaline earth “ketyls” (semiquinone). Kazuhiro
found that by reducing phenanthrenequinone into its various alkali and alkaline ketyls, the
hyperfine splittings and spectral width were dependent on the reducing metal; although,
Kazuhiro conceded no reason as to why this phenomenon occurred. Kazuhiro also provided
hyperfine splittings for a free radical phenanthrenequinone, reduced with potassium in liquid
ammonia. The main difference between Kazuhiro’s work and that of Sands’, aside from the
reduction methods, is that the protons are grouped differently.

Sands shows two sets of

hyperfine splittings, 4.65 MHz and 1.07 MHz, for two sets of four protons, while Kazuhiro has
three groups of protons for the free radical, 4.37 MHz (four equivalent protons), 1.23 MHz (two
protons) and 1.01 MHz (two protons).

47

Figure 2-15. Room temperature EPR spectrum (black) and simulation (red) of PSQ radical
anion at 20 mG modulation amplitude. Simulation parameters: g = 2.00754, lwpp = 0.0005 mT,
0.0191 mT, aiso values: 23Na = 1.32 MHz, 1H4,5 = 0.77 MHz, 1H2,7 = 1.19 MHz, 1H1,8 = -4.06
MHz and 1H3,6 = -4.75 MHz. Experimental parameters: 9.85 GHz ÎźW frequency, 0.50 mW
power, 3508 G center field, 20 G sweep width, 20 mG modulation amplitude, 1024 points. χ2 =
0.39.

The room temperature, CW spectrum of the PSQ radical anion at 20 mG modulation amplitude
is shown in Figure 15 (see above). These values correlate well with what has been found in the
literature for PSQ (in MHz): 4.68, 3.78, 1.18, 0.5656.

The number of lines present in the

spectrum is approximately 38, which does not agree with the theoretical number of lines. The
approach to fit this spectrum was to include only the protons; however, this path produced a
simulation that required one set of two protons to be inequivalent. After further investigation, it
48

was found that in order to properly model the spectra above, a sodium ion was needed in the
simulation. With a sodium ion included, the total expected number of lines is 324. It is perhaps
the case that all the lines are not visible due to the limits of the spectrometer.
Figure 16 depicts a cartoon that includes a sodium ion as part of the model system.
Initially it was not our impression that sodium was participating in the hyperfine coupling; again,
upon analyzing the spectrum, it became clear that the four groups of two protons were not the
only active nuclei. Using only four groups of two protons produced line shapes that were not
consistent with the experimental spectrum. As previously stated, making one set of protons
inequivalent allowed for a proper line shape; however, this approach was not logical as the
molecule has C2v symmetry and the proton hyperfine coupling constants will be the same for
corresponding protons across the vertical mirror plane. Thus, a single sodium ion was added to
the simulation, in addition to the four groups of two protons; this addition produced a line shape
similar to that of the experimental spectrum. It is not known that the sodium ion is interacting
with the oxygen atoms directly, because there is no crystal structure of this species, but due to
the electrostatic nature of the two components, this was the most logical place for the Na+. In the
literature, this is referred to as ion pairing and is consistent with Kazuhiro’s analysis. 57 Upon
addition of the Na+, the spectrum was able to be fit with a respectable χ2 value of 0.39.

49

Figure 2-16. Possible Na+ coordination to PSQ-∙.

Figure 2-17. PSQ (black) and [13C2]PSQ (red) overlaid CW spectra.

50

Figure 17 shows the spectra of PSQ (black) versus [13C2]PSQ (red).

13

C-labelling appears to

have an overall line-broadening effect, relative to the 1H-coupling in PSQ. Figure 18 depicts the
13

C2PSQ spectrum (black) and simulated spectrum (red). The [13C2]PSQ CW data was simulated

by first holding the PSQ 1H-aiso constant, as well as the linewidth parameters and g-factor. Then,
13

C-aiso values were added and varied until a minimal residual value was obtained; the linewidth

parameters and g-factor were adjusted for an optimal χ2 value. The data were found to be fit with
C-aiso values in the range of 1.5 – 2.2 MHz (absolute values). These values were obtained by

13

varying carbonyl carbon hyperfine coupling and linewidth parameters. It will become more clear
in the ESEEM section as to why the spectrum was fit with multiple

13

C-aiso values, but to

summarize shortly, ESEEM data suggest the presence of multiple species. Figure 18 shows the
[13C2]PSQ spectrum and spectral fit with

13

C-aiso = 1.57 MHz. CW simulations were not

dependent on the sign of the hyperfine coupling.

51

Figure 2-18. Room temperature EPR spectrum (black) and simulation (red) of 13C2PSQ radical
anion at 20 mG modulation amplitude. Simulation parameters: g = 2.00755, lwpp = 0.0001 mT,
0.0222 mT, 13C-aiso values: 1.57 MHz. Experimental parameters: 9.85 GHz ÎźW frequency, 0.50
mW power, 3508 G center field, 20 G sweep width, 20 mG modulation amplitude, 1024 points.
χ2 = 0.85.

52

Figure 2-19. Room temperature EPR spectrum (black) and simulation (red) of 13C2PSQ radical
anion at 20 mG modulation amplitude. Simulation parameters: g = 2.00755, lwpp = 0.0001 mT,
0.0255 mT, 13C-aiso values: 2.17 MHz. Experimental parameters: 9.85 GHz ÎźW frequency, 0.50
mW power, 3508 G center field, 20 G sweep width, 20 mG modulation amplitude, 1024 points.
χ2 = 1.02.
Table 3 summarizes the results from the simulations/fits and DFT calculations found for PSQ
and [13C2]PSQ aiso values. The first row contains the results from simulating the experimental
spectrum of the NaPSQ complex, the second row contains the Fermi analysis from PSQ-∙ and the
third row contains the Fermi analysis from NaPSQ. After analyzing the CW data, further
calculations were undertaken to properly compare to the NaPSQ complex. The DFT module
used for calculating isotropic hyperfine coupling constants is the Fermi contact integral. With
respect to 1H-aiso values, the predicted values resemble the calculated values quite well. Most
importantly, the order of ascending/descending magnitude of hyperfine coupling agrees between
53

the fits and DFT results. These values do resemble what has been seen in the literature for PSQ.
In all cases the H3,6 values are in agreement. There is some discrepancy between the remaining
1

H values; the NaPSQ DFT model seems to model the 1H-aiso values to a greater degree while

the PSQ model accurately predicts the H2,7 and H4,5 values. Two other reports have found two
sets of four equivalent protons with aiso = 4.65 and 1.065 MHz57, and three sets of protons (one
group of four protons and 2 groups of 2 protons each) with aiso = 4.372, 1.233 and 1.01 MHz52,
respectively.
Table 2-3. 1H, 13C and 23Na aiso and DFT calculated values (Fermi contact integral). All values
in MHz.
H3,6

H1,8

H2,7

H4,5

23

Data - aiso

4.75

4.06

0.77

1.19

1.32

1.57,2.17

PSQ - Fermi

-4.59

-2.38

0.43

1.33

-

-10.25

NaPSQ - Fermi

-4.69

-4.41

1.14

1.37

-7.06

-1.84

Na

13

CC=O

Figure 2-20. Visual representation of Table 3. The aiso values listed are located at each of their
respective nuclear positions.
The greatest disagreement in values concerns the carbonyl carbon and

23

Na aiso values.

Undeniably, the 23Na-aiso value found by the simulation completely disagrees with the calculated
value. Previously, the sodium ion contribution to the hyperfine coupling was unrecorded and/or,
54

not measured57. Hyperfine coupling constants for metals in a series of compounds of Group III
metal halides chelated to PSQs were measured by Klimov, et al. In the closest comparison,
aluminum was found to have a aiso = 5.88 MHz in [Al(PSQ)Cl2]. Though, no other studies were
found to quantitatively accommodate Group I or Group II metal contribution to the hyperfine.
The first of these studies have begun here, with sodium.
Including Na+ in the experiment was both favorable and detrimental to the experiment. In
favor of using Na as a reductant, it showed that a significant portion of spin delocalization is
possible, across the carbonyl bridge. As a detriment to the experiment, the Na+ adds to the
complexity of the spectrum, as will also be evidenced in the ESEEM section below. Ideally, an
isolated picture of PSQ’s spin distribution would have been advantageous, rather than a NaPSQ
complex. An unobstructed view of the hyperfine interaction could be achieved by reducing
phenanthrenequinone by electrical ionization, or even using a different reducing metal. For
example,

39

K has an I = 3/2 nucleus, but the Larmor frequency would be quite small,

approximately 0.7 MHz at 3500 G (g 2). In addition, the hyperfine coupling to

39

K would be

reduced by a factor of 5.6, relative to 23Na; this factor comes is a result of comparing the nuclear
g-values.
The carbonyl carbon values were correctly predicted by the NaPSQ model calculation, rather
than PSQ-∙, which does not include sodium contribution.

Figure 16 depicts the Na+ ion

coordinated between the two carbonyl oxygen atoms, which is supported by the difference in
calculated hyperfine coupling of the PSQ and NaPSQ carbonyl carbons. Comparatively, there
have not been other studies where carbon-13 was used as an isotopic label at the carbonyl carbon
in phenanthrenequinone; there have been some accounts in other ortho-semiquinones, however.
The carbonyl carbons of ortho-benzosemiquinone have been found to have an aiso = 5.66 MHz58,

55

which is several times than what was found for carbonyl carbons in this work. This could be
attributed to the greater degree of delocalization in PSQ, due to the extended aromatic network.
The sign of the hyperfine is not usually obtainable from conventional CW experiments, but
rather a culmination of several techniques. One of which includes a double ENDOR technique
that probes the ENDOR transition of one type of nuclei followed by a second ENDOR excitation
of another type of nucleus. The sign of the hyperfine is then determined by the change of
intensity from the first ENDOR signal to second ENDOR signal59,60.
The isotropic hyperfine coupling can be related to the spin density through the following
equation:
𝑎𝑖𝑠𝑜 (𝑁) =

4𝜋
𝑔 𝑔 𝛽 𝛽 〈𝑆 〉−1 𝜌(𝑁)
3 𝑒 𝑁 𝑒 𝑁 𝑧

(2.11)

where ge is the electron g-factor, gn is the nuclear g-factor, βe is the electron Bohr magneton, βN
is the nuclear Bohr magneton, Sz is the spin operator and ρ(N) is the Fermi contact integral at
nucleus N, or the spin density at nucleus N. Nguyen and coworkers have summarized the
collection of physical constants for individual nuclei with the following singular constants; aiso in
MHz and spin density is therefore unitless61:
𝑎𝑖𝑠𝑜 ( 1𝐻 ) = 4469.7𝜌( 1𝐻 )

(2.12a)

𝑎𝑖𝑠𝑜 ( 13𝐶 ) = 1124.1𝜌( 13𝐶 )

(2.12b)

If then, these equations are applied to the aiso values obtained in Table 1, the following spin
densities are obtained for PSQ, in Table 4.

56

Table 2-4. Calculated spin density values from equations 2.8a and 2.8b for NaPSQ complex
(data and from Fermi analysis). Spin density values are unit-less.
23

Na

13

H3,6

H1,8

H2,7

H4,5

CC=O

ρ(N)Data (x10-3)

1.06

0.91

0.17

0.26

-

1.39,1.93

ρ(N)Fermi (x10-3)

1.05

0.99

0.25

0.30

-

1.6

The second radical system studied was 3,6-(OMe)2-PSQ radical anion. Unlike the PSQ
system, this radical anion has not been investigated previously by EPR. The addition of methoxy
groups introduces π - donating effects to the phenanthrene ring system, as well as σ - accepting.
In the DFT study it was found that, relative to PSQ, the methoxy substituents polarized spin
away from the oxygen atoms; the decrease in oxygen spin density is ~2.5%, relative to PSQ.
However, this EPR study is concerned with difference in carbonyl carbon hyperfine coupling
between the two compounds, which is related to spin density, but is not the same as spin density.
Regardless of the differences of the two quantities, the hyperfine interaction will be used to
calculate the isotropic contribution to the spin density, just as with PSQ.
The CW EPR spectrum of Na(3,6-(OMe)2-PSQ) can be seen in Figure 26. The number
of lines present is approximately 32, which like NaPSQ, does not agree with the number of
theoretical lines. The number of lines that should be present, according to equation 2.5, is 1,620,
which is inconsistent with the number of lines present. This is consistent, however, with the
NaPSQ complex.

57

Figure 2-21. Room temperature EPR spectrum (black) and simulation (red) of Na(3,6-(OMe)2PSQ) radical anion at 20 mG modulation amplitude. Simulation parameters: g = 2.00746, lwpp
= 0.0015 mT, 0.0135 mT, aiso values: 23Na = 1.17 MHz, 1HMe-a = 0.56, 1HMe-b = 0.41 1H4,5 = 0.96
MHz, 1H2,7 = 1.01 MHz and 1H1,8 = 3.84, MHz. Experimental parameters: 9.85 GHz ÎźW
frequency, 0.5 mW power, 3508 G center field, 20 G sweep width, 20 mG modulation
amplitude, 1024 points. χ2 = 0.52.
Similarly, to NaPSQ, the final values of the simulation were found by holding the gfactor and linewidth parameters constant, and varying the hyperfine parameters until the width of
the simulation matched the spectral width of the experimental spectrum.

The linewidth

parameters were adjusted to account for the Gaussian and Lorentzian lineshape and finally all the
parameters were then minimized with fminsearch in Matlab, until a global minimum was
reached. Methoxy protons were not originally considered in the simulations but were added to
account for the high degree of splitting in the spectrum.

58

The hyperfine coupling constants are comparatively less than the hyperfine couplings from
NaPSQ (see Table 5 below). By replacing the 3,6-protons with methoxy groups, the hyperfine
coupling is now spread to the methoxy groups as well. In addition to being a π-donating group,
methoxy substituents also double as sigma-acceptors, so it is sensible that the proton couplings
are smaller in magnitude, relative to NaPSQ, because the isotropic contribution to the hyperfine
is directly related to the s – electron density. Additionally, the majority of the DFT predicted 1H
aiso values agree with the experimental results. The one discrepancy being that DFT predicted
that methoxy protons have a larger coupling than the 4,5-protons on the ring.

59

Figure 2-22. [13C2]3,6-(OMe)2-PSQ (red) versus 3,6-(OMe)2-PSQ (black) spectra.
The room temperature, CW spectrum of [13C2]3,6-(OMe)2-PSQ versus 3,6-(OMe)2-PSQ can
be seen above in Figure 22. The effect of

13

C addition is apparently rather small and the

resulting EPR spectrum resembles the 12C spectrum almost perfectly. The spectrum (black) and
fit (red) of [13C2]3,6-(OMe)2-PSQ is shown in Figure 23. The spectrum was fit with

13

C aiso =

0.41. This spectrum was fit with multiple 13C aiso values, similarly to [13C2]PSQ; the additional
13

C aiso values are 0.98 MHz and 0.16 MHz. The reason for multiple

60

13

C aiso values is due to

ESEEM data confirming multiple species in solution. The other spectral fits are shown in the
Chapter 2 Appendix.

Figure 2-23. Room temperature EPR spectrum (black) and simulation (red) of Na[13C2]3,6(OMe)2-PSQ at 20 mG modulation amplitude. Simulation parameters: g = 2.007443, lwpp =
0.0015 mT, 0.0145 mT, 13C-aiso value: 0.41 MHz. Experimental parameters: 9.85 GHz ÎźW
frequency, 0.5 mW power, 3508 G center field, 20 G sweep width, 20 mG modulation
amplitude, 1024 points. χ2 = 0.50.
In terms of a substituent effect, there does seem to be a difference between PSQ and 3,6(OMe)2-PSQ. Considering 1H aiso values alone, 1H1,8 and 1H2,7 decrease from PSQ to 3,6(OMe)2-PSQ.

Moving to carbonyl carbons, a much more dramatic effect is operative.

[13C2]PSQ has aiso values in the range of 1.5 – 2.2 MHz while [13C2]3,6-(OMe)2-PSQ was found

61

to have aiso values 0.15 – 1 MHz. Table 5 tabulates the isotropic coupling pathways for Na(3,6(OMe)2-PSQ).
Table 2-5. 1H, 13C and 23Na aiso coupling constants, in MHz, for Na(3,6-(OMe)2-PSQ).
23

H2,7

H4,5

HMe-a

HMe-b

Data - aiso

3.84

1.01

0.96

0.56

0.41

1.17

0.41,0.16,0.98

OMePSQFermi
NaOMePSQ Fermi

-1.40

-0.54

0.43

0.23

-0.17

-

-9.85

-3.71

1.41

0.42

0.71

-0.14

-6.98

-1.25

62

Na

13

H1,8

CC=O

Figure 2-24. Visual representation of Table 4. The aiso values listed are located at each of their
respective nuclear positions.
Table 2-6. Calculated spin density values from equations 2.8a and 2.8b for Na(3,6-(OMe)2PSQ).
13

H1,8

H2,7

H4,5

HMe-a

HMe-b

ρ(N)Data (x10-4)

8.59

2.26

2.14

1.25

0.91

3.66,1.42,8.72

ρ(N)Fermi

8.30

3.15

0.94

1.59

0.31

11.1

(x10-4)

63

CC=O

Table 6 compares the calculated spin density at the nucleus to the observed (fit) values from
the Nguyen constants. The spin density reflects the same trend as the hyperfine coupling
constants, and in most cases is comparable to the spin density calculated from the Fermi contact
integrals. The spin density is largest for H1,8 and the largest carbonyl carbon value (8.59 and
8.72, respectively), while unquestionably the carbonyl carbon has the largest spin density from
the Fermi analysis. Although these values seem large, they are an order of magnitude smaller
than the NaPSQ values. With the addition of a Na+, it appears that the experimental model is not
in agreement with the theoretical model discussed earlier in this chapter. The carbonyl carbon
spin density value calculated from the Fermi contact integral for NaPSQ (1.6) falls within the
range of spin density values obtained from fitting the experimental data (1.39-1.93); however,
this is not true for Na(3,6-(OMe)2-PSQ). The spin density of the carbonyl carbons obtained from
conversion of the Fermi contact integral from DFT calculations (11.1) exceeds the range of
values obtained for the experimental set of data (1.42-8.72).
This is an interesting result, as it was also observed in the [Ni(tren)(3,6-R2-PSQ)]+ DFT
calculations. The OMe derivative had similar, but less spin density at the carbonyl carbon atoms,
relative to the proteo complex. It is interesting that amongst the additional layers of complexity,
covalent bonding to a metal (in this case experimentally observed by x-ray diffraction), electron
spin exchange, spin polarization effects from unpaired electrons on the metal, etc., there remains
a common thread between the radical compounds and NiII complexes.
ESEEM Spectroscopy. ESEEM (Electron Spin Echo Envelope Modulation) spectroscopy
was employed to further understand the complexities of the NaPSQ and Na(3,6-(OMe)2-PSQ)
CW-EPR spectra. CW-EPR was performed under ambient conditions and the tumbling of the
molecules in solution nulls any anisotropic, or dipolar (T), contribution to the hyperfine coupling.

64

ESEEM spectroscopy indirectly measures anisotropic contributions to the hyperfine coupling by
measuring interferences between the different EPR transition frequencies (refer to Figure 13
above) that are involved in creating an electron spin echo. This technique will be quite useful in
unraveling the complications that arose in the CW spectra, by shedding a light on the dipolar
coupling. The spin echo is a physical observable that occurs as a result of spins refocusing from
a pulse of [microwave] radiation.
To move forward, however, the energy splitting diagram in Figure 13 must be amended to
include nuclear effects. The spin Hamiltonian for an S = ½ I = ½ system is now of the form:
̂ 𝑔𝑒 𝛽𝑒 𝐵
𝐻
𝑔𝑁 𝛽𝑁 𝐵
=
𝑆̂𝑍 + 𝐴𝑍𝑍 𝑆̂𝑍 𝐼̂𝑍 + 𝐴𝑋𝑍 𝑆̂𝑍 𝐼̂𝑋 −
𝐼̂𝑍
ℏ
ℏ
ℏ

(2.13)

Where the first and last terms are the electronic and nuclear Zeeman interactions, and the middle
terms are the electron-nuclear hyperfine coupling terms as described previously in the CW
section. 𝐴𝑍𝑍 = 𝐴 = 𝐴∥ 𝑐𝑜𝑠 2 𝜃 + 𝐴⊥ 𝑠𝑖𝑛2 𝜃 and 𝐴𝑋𝑍 = 𝐵 = (𝐴∥ − 𝐴⊥ )𝑐𝑜𝑠𝜃𝑠𝑖𝑛𝜃 where 𝐴⊥ and 𝐴∥
are the principal values in an axially symmetric hyperfine coupling tensor:
𝐴𝑖𝑠𝑜 − 𝑇
[ ⋯
⋯

⋯
𝐴𝑖𝑠𝑜 − 𝑇
⋯

…
⋯
]
𝐴𝑖𝑠𝑜 + 2𝑇

The angle θ describes the angle between the principal axis of the hyperfine coupling tensor and
the applied magnetic field, B0 and 𝑇 =

𝑔𝑒 𝑔𝑁 𝛽𝑒 𝛽𝑁
𝑟3

. Using a basis set of |𝑚𝑆 ,𝑚𝐼 ⟩, the Hamiltonian

matrix can be diagonalized resulting in eigenvalues of the four energy levels produced by an S =
½ I = ½ system, as seen in Figure 25 below.

65

Figure 2-25. Energy level diagram of an S = ½ I = ½ system.
Figure 25 shows the splitting of the β and ι electron spin manifolds due to an I = ½
nucleus. The ordering of the states is the same as that of Figure 13, where |4⟩ = |𝛽𝛼𝑁 ⟩, |3⟩ =
|𝛽𝛽𝑁 ⟩, |2⟩ = |𝛼𝛽𝑁 ⟩ and |1⟩ = |𝛼𝛼𝑁 ⟩. The angles 𝜙𝛽 and 𝜙𝛼 determine the direction of the
nuclear spin frequency, and 𝜔𝛽 and 𝜔𝛼 are the angular hyperfine frequencies and are defined by
the equations:
2

2

(2.14)

2

2

(2.15)

𝜔𝛼 = √(𝜔𝐼 − 𝐴⁄2) + 𝐵 ⁄4
𝜔𝛽 = √(𝜔𝐼 + 𝐴⁄2) + 𝐵 ⁄4

In measuring EPR spectra under ambient conditions, only the isotropic portion of the
hyperfine coupling is measurable; any anisotropic effects are canceled-out due to tumbling in
66

solution. Freezing the sample removes any tumbling motion and anisotropic effects can arise.
The nuclear states are now mixed in each electron spin manifold, as in Figure 25, and therefore
the spin states, |1⟩, |2⟩, |3⟩ and |4⟩ are mixed and all transitions are allowed. The EPR transitions
in Figure 25 marked |𝑢| and |𝑣| are the normalized probability amplitudes and correspond to
former (in CW-EPR) allowed and forbidden EPR transitions, respectively. Application of a
pulse of microwave radiation excites all four transitions simultaneously. Multiple pulses can
cause molecules to branch from one transition to another during the pulse sequence.
Branching establishes interferences on the echo amplitude that occur at frequency differences
between the transitions. For example, the result of taking the difference between the frequency
of |1⟩ → |3⟩ and |1⟩ → |4⟩ is a frequency corresponding to

|4⟩−|3⟩
ℎ

. Keeping in mind that the

sample is frozen, that frequency has both isotropic and anisotropic hyperfine components.
Determining isotropic hyperfine coupling from CW-EPR is important for untangling these
dipolar hyperfine components observed by ESEEM.
Figures 26 and 27 show a vector picture and timing diagram, respectively, for a two-pulse
⃗⃗ , is aligned with the external, fixed,
ESEEM experiment. Initially, the bulk magnetization, 𝑀
magnetic field along the z axis. A 90° pulse is applied and this places the magnetization
perpendicular to B0, in the xy plane. The magnetization then begins to fan out in the xy plane
into spin packets during a duration of time period τ. Some spin packets experience stronger local
magnetic field effects than others and develop a positive phase while the spin packets that
experience weaker local magnetic field effects have a negative phase. After the time period τ, a
180° microwave pulse is applied that turns the spins about the x axis; the spin packets then begin
to rephase and an echo is observed after a time period of 2τ. The amplitude of the echo decay is
then measured as a function of τ.62
67

Figure 2-26. Vector model of a two pulse ESEEM experiment.

Figure 2-27. Timing diagram of a two pulse ESEEM sequence.
In two-pulse ESEEM, modulation occur at the fundamental hyperfine frequencies, as well as
at sum and difference combination frequencies. The resolution is also poor due to fast spin-spin
relaxation times. Inclusion of a third microwave pulse increases sensitivity by eliminating the
sum and combination frequencies and only detecting the hyperfine frequencies.

The echo

amplitude decay is now monitored with respect to a τ + T time period and allows for a much
longer decay, resulting in higher hyperfine frequency resolution. Additionally, the time period τ
can be varied and is most often set at integer values of a particular Larmor frequency. This
allows for the suppression of modulations from weakly coupled nuclei; this is known as the tau

68

suppression effect. For the purpose of this study, τ was set at various proton Larmor frequencies,
in order to enhance the hyperfine frequencies from the 13C-labelled carbonyl carbons.

Figure 2-28. Timing diagram of a three pulse ESEEM sequence.
ESEEM spectra were collected at a magnetic field value of 3455 Gauss (g 2), at 4 K or 10 K.
Tau (τ) values were chosen to suppress modulations from weakly coupled protons. Time domain
spectra were normalized by division of a background decay function, then application of a
Hamming window, followed by a zero filling to double the data set length and Fourier
transformation. Spectra are displayed as absolute values of the Fourier transform. For analysis
of 13C couplings the time domain data for
13

C spectra were divided by

12

12

C and

13

C isotopologues were normalized and then

C spectra to isolate the effect of the

13

C hyperfine interaction.

Divided time domain data were then worked up as stated above.
Time and frequency domain data were simulated by the Saffron module within Easyspin.
The spectra were fit with the fminsearch function in Matlab. The best fits were obtained from
simulating the time domain data and their Fourier transforms, separately; it should be noted that
this process of simulating both the time and frequency domain spectra is essential for obtaining
all information from the modulation decay of the time trace, and the resonances of the frequency
spectra. Initially, the carbonyl carbons were treated as an equivalent set and the aiso values were
fixed to the values obtained from the CW fits and only the dipolar couplings were varied. It
became quite apparent, while attempting to fit the data, that a single set of values obtained would
69

not be ideal for fitting the data. The time and frequency domain spectra were then simulated and
fit to accommodate for multicomponent systems. The proton hyperfine couplings were left out
of the simulations because the labeled spectra were divided by the

12

C spectra and thus the 1H

aiso values have been divided out. Hyperfine coupling parameters were simulated by taking the
anisotropic hyperfine coupling tensor from the DFT results and transforming them to their
spherical hyperfine components, a (isotropic), T (dipolar), and ρ (rhombicity), as seen in
equation 2.13.
𝑨𝑑𝑖𝑎𝑔

𝑎𝑖𝑠𝑜
= 𝑎𝑖𝑠𝑜 𝑰 + 𝑻 = ( 0
0

0
𝑎𝑖𝑠𝑜
0

0
−(1 − 𝜌)
0 )+𝑇(
0
𝑎𝑖𝑠𝑜
0

0
−(1 + 𝜌)
0

0
0)
2

(2.16)

These parameters were used as initial guesses and were subjected to simulations and fits until
values were achieved that gave minimum χ2 values. Only time domain fits were able to be
achieved with residual values of 10 % or less; further work is being done to optimize the
frequency domain parameters and reach an agreement between time and frequency domain
spectra.
Figure 28 shows the frequency domain ESEEM
shows the normalized time domain

13

13

C/12C spectrum and fit, and Figure 29

C/12C spectrum and fit. Peaks in the ESEEM frequency

spectrum arise from Larmor frequencies of nuclei that have unpaired spin density. When a
particle with a magnetic moment, and subsequently angular momentum, is placed in an external
magnetic field, the magnetic field will exert a torque on the magnetic moment of the particle:
𝝉 = 𝝁 × 𝑩 = 𝛾𝑱 × 𝑩

70

(2.17)

where τ is the torque, μ is the magnetic moment, B is the external magnetic field, J is the angular
momentum operator and Îł is the g-factor of the electron. The Larmor frequency is the angular
frequency at which the magnetic moment precesses about the external magnetic field axis
𝜔 = −𝛾𝐵
where ω is the Larmor frequency. For

13

(2.18)

C, Îł = 10.705 MHz/T, and at a fixed field of 3455 G

(0.3455 T) ω = 3.69 MHz. Indeed, the first peak of the frequency spectrum occurs at 3.66 MHz,
corresponding to significant spin density located on the carbonyl carbons. Initially, it was
thought that the features centered at 7.8 MHz and 11.1 MHz were likely to be second and third
harmonics from negative frequencies. To further drive the point that the spectrum is completely
dominated by 13C hyperfine effects, γH = 42.576 MHz∙T-1 and at B = 3455 G, ω = 14.71 MHz.
Therefore, there is no evidence of 1H hyperfine in the divided spectrum. This can be seen in
Figure 29, where the ESEEM frequency spectra of [13C2]PSQ and PSQ are compared. The
resonance at 14.7 MHz from the PSQ spectrum is a result of dipolar couplings from 1H on the
PSQ ring. The resonances at 3.66 MHz from both spectra are from 13C dipolar couplings, but the
difference is that in the PSQ spectrum, the couplings are arising due to naturally occurring

13

C

and the [13C2]PSQ is from enrichment of the 9 and 10 carbonyl carbons. The spectra have been
normalized to show resonances from the PSQ (12C) sample.

71

Figure 2-29. [13C2]PSQ (red) and PSQ (black) ESEEM frequency spectra. Experimental
conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns time increment, 136 ns
tau, 40 ns starting T value. Spectra were normalized post-Fourier transform to illustrate the
differences of the

72

Figure 2-30. [13C2]PSQ ESEEM frequency spectrum and fit for dimer. Simulation parameters:
g = 2.00575, 2 13C nuclei, aiso = 2.1713 MHz, T = 2.7757 MHz, ρ = 0 MHz. Experimental
conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns time increment, 136 ns
tau, 40 ns starting T value.

73

Figure 2-31. [13C2]PSQ ESEEM time domain spectrum and fit for dimer. Simulation
parameters: g = 2.00575, 2 13C nuclei, aiso = 2.1713 MHz, T = 2.7757 MHz, ρ = 0 MHz.
Experimental conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns time
increment, 136 ns tau, 40 ns starting T value. χ2 = 0.67.
Initially, the data was fit with one set of two equivalent 13C nuclei. The peak position of the
largest resonance in the frequency spectrum can be simulated with a single pair of

13

C nuclei

where aiso = 2.17 MHz, T = 2.78 MHz. Clearly, though, this model of a single NaPSQ ion-pair
cannot be the most accurate portrayal of the unpaired spin density.

The shape may be

reminiscent of the resonance at 3.66 MHz, but the intensity is not reproduced by this simple
model. The corresponding fit for the time trace does bare some semblance to the data; however,
after ~ 1.8 us the beating is no longer in accord with the data and the simulation does not track
with the data. This data can also be modeled with four equivalent carbon nuclei, as a dimer, with
the same isotropic and dipolar values. This is some evidence that this peak at 3.66 MHz is from
a dimer or similar species in solution. This model does not account for the entire data set,
though. The humps at 8.0 and 11.8 MHz are unaccounted for in this model.
74

Figure 2-32. [13C2]PSQ ESEEM frequency spectrum and fit for multicomponent system.
Simulation parameters: g = 2.00575, Dimer: 4 13C nuclei, aiso = 2.1713 MHz, T = 2.7757 MHz, ρ
= 0 MHz; ion-pair: 2 13C nuclei, aiso = 1.5741 MHz, T = 13.5674 MHz, ρ = 0.44 MHz.
Experimental conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns time
increment, 136 ns tau, 40 ns starting T value.

75

Figure 2-33. [13C2]PSQ ESEEM time domain spectrum and fit for multicomponent system.
Simulation parameters: g = 2.00575, Dimer: 4 13C nuclei, aiso = 2.1713 MHz, T = 2.7757 MHz, ρ
= 0 MHz; ion-pair: 2 13C nuclei, aiso = 1.5741 MHz, T = 13.5674 MHz, ρ = 0.44 MHz.
Experimental conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns time
increment, 136 ns tau, 40 ns starting T value. χ2 = 0.22.
The total frequency spectrum (see above) can be modeled by a multiple component system.
The system contains a dimer of two phenanthrenequinone moieties bridged by one sodium
cation, and a single NaPSQ ion-pair. The dimer has four equivalent

13

C nuclei with aiso = 2.17

MHz, T = 2.78 MHz and ρ = 0 MHz. The second species in this system consists of two
equivalent

13

C nuclei with aiso = 1.57 MHz, T = 13.56 MHz and ρ = 0.44 MHz. Inclusion of a

second species improves the fit for both the time domain and the Fourier transform. The biggest
improvement being that the second and third humps at 8 MHz and 11 MHz, respectively, have
been fit. Therefore, it can be concluded that since the peak at 3.66 MHz can be fit with two pairs
of equivalent carbons, that the resonances at 8 MHz and 11 MHz are either solely due to the ion-

76

pair or from the combination of the two species. It is noted that these humps cannot be fit by
only the ion-pair, in the absence of the dimer.

Figure 2-34. Cartoon depiction of the ion-pair and dimer of the NaPSQ multicomponent system.
Table 2-7. Hyperfine coupling parameters for NaPSQ dimer and molecule. All values in MHz.
aiso

T

ρ

NaPSQ Dimer

2.17

2.77

0

NaPSQ Ion-Pair

-1.57

13.57

0.44

Pictured above are cartoons of the proposed structures of the dimer and molecule that make
up the multicomponent system of the NaPSQ complex. It is a bit easier to see how the unpaired
electron density can be delocalized, to a greater extant, in the dimer system. Simply, the dipolar
coupling is smaller in the dimer, relative to the NaPSQ molecule, due to spin delocalization
across the Na+ bridge, to a second set of carbonyl units. The dimer has been drawn with a single
radical, as opposed to two radicals on each quinone moiety. It has been drawn as such because
the system has been modeled with Na aiso coupling and there is, therefore, potential
communication between the two quinones. Additionally, CW measurements did not indicate any
triplet state formation.

77

The rhombicity component has been introduced because without it, the intensity of the
resonances at 8 MHz and 11.8 MHz would be quite large. The inclusion of rhombicity for the
molecule decreases the intensity of the simulated peak at 11.8 MHz and leaves all other peaks
unaltered. In terms of physical effects, an S = ½ system that exhibits rhombicity is more of a
second order effect. As indicated by equation 2.9, the rhombicity is a perturbation on T, the
dipolar coupling.
Pictured below is the ESEEM spectrum and fit of [13C2]3,6-(OMe)2-PSQ (Figures 35 and
36). This particular fit, though, is only of the first peak. Similar to [ 13C2]PSQ, the first peak
(3.66 MHz) can be fit with a single set of parameters. In contrast to [13C2]PSQ, the first peak is
not modeled by four carbons, but six; in other words, not a dimer, but a possible homoleptic
complex, with a single sodium ion bonded to three quinone ligands with aiso = 0.98 MHz and T =
1.15 MHz. Three pairs of carbon atoms were used in order to accurately model the intensity of
the peak at 3.66 MHz.

78

Figure 2-35. [13C2]3,6-(OMe)2-PSQ ESEEM frequency spectrum and fit for the homoleptic
complex. Simulation parameters: g = 2.00507, 6 13C nuclei, aiso = 0.9758 MHz, T = 1.15 MHz, ρ
= 0 MHz. Experimental conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns
time increment, 136 ns tau, 40 ns starting T value.

79

Figure 2-36. [13C2]3,6-(OMe)2-PSQ ESEEM time domain spectrum and fit for the homoleptic
complex. Simulation parameters: g = 2.00507, 6 13C nuclei, aiso = 0.9758 MHz, T = 1.15 MHz, ρ
= 0 MHz. Experimental conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns
time increment, 136 ns tau, 40 ns starting T value. χ2 = 0.94.
The process of fitting this spectrum was much more difficult than that of [13C2]PSQ. This
was also a multicomponent system, but not two components were needed to properly simulate
the spectrum, but rather three: a homoleptic complex, a dimer and an ion-pair.

Also

differentiating from the analysis of [13C2]PSQ, the components are not 1:1, but rather 4:1:1,
homoleptic complex : dimer : ion-pair, respectively (Figures 37, frequency, and 38, time
domain).

80

Figure 2-37. [13C2] 3,6-(OMe)2-PSQ ESEEM frequency spectrum and fit for multicomponent
system. Simulation parameters: g = 2.00507, homoleptic complex: 6 13C nuclei, aiso = 0.9758
MHz, T = 1.15 MHz, ρ = 0 MHz (4 of these); dimer: 4 13C nuclei, aiso = 0.3849 MHz, T = 7.9563
MHz, ρ = 0 MHz; ion-pair: 2 13C nuclei, aiso = -0.4155 MHz, T = 16.1325 MHz, ρ = 0.63 MHz.
Experimental conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns time
increment, 136 ns tau, 40 ns starting T value.

81

Figure 2-38. [13C2] 3,6-(OMe)2-PSQ ESEEM time domain spectrum and fit for multicomponent
system. Simulation parameters: g = 2.00507, homoleptic complex: 6 13C nuclei, aiso = 0.9758
MHz, T = 1.15 MHz, ρ = 0 MHz (4 of these); dimer: 4 13C nuclei, aiso = 0.3849 MHz, T = 7.9563
MHz, ρ = 0 MHz; ion-pair: 2 13C nuclei, aiso = -0.4155 MHz, T = 16.1325 MHz, ρ = 0.63 MHz.
Experimental conditions: 345.5 mT magnetic field, 3 pulse ESEEM sequence, 12 ns time
increment, 136 ns tau, 40 ns starting T value. χ2 = 0.35.
After fitting the peak at 3.66 MHz with a single, homoleptic complex, a dimer and ion-pair
were added to the simulation to account for the remaining resonances in the ESEEM spectrum.
The structures that were added after several attempts with varying aiso values; indeed, even after
fitting the resonance at 3.66 MHz, the aiso values were exchanged between homoleptic complex,
dimer and ion-pair, in order to arrive at the best fit for the data. After addition of the dimer and
ion-pair, the intensity of the peak at 3.66 MHz decreased; however, if the number of homoleptic
complexes was increased, the intensity was restored.

Thus, it was decided that multiple

homoleptic complexes would be required for the fitting of this spectrum. It should be noted that
upon addition of additional homoleptic complexes, no additional features were produced, only
the intensity was affected.
82

Figure 2-39. Cartoon depiction of the Na(3,6-(OMe)2-PSQ) multicomponent system.
Table 2-8. Hyperfine coupling parameters for Na(3,6-(OMe)2-PSQ) homoleptic complex, dimer
and molecule. All values in MHz.
aiso

T

ρ

Na(OMe2PSQ)3

0.98

1.30

0

NaOMe2PSQ Dimer

0.38

7.96

0

NaOMe2PSQ Ion-Pair

-0.41

16.13

0.63

83

When considering that the CW spectra can be fit with various

13

C aiso values, and the fact

that the ESEEM spectra cannot be fit with a single pair of carbonyl carbon atoms, it becomes
clear that the spectra are sums of various combinations of Na+ and PSQ-∙/phenanthrenequinone.
This is a theme within the EPR and ESEEM spectra of the molecules studied within this chapter.
It is the belief of the author that when reducing the quinones with excess sodium metal, the
sodium metal is a surface for reactions to occur and there could be numerous products from this
chemistry. The outcome of such is evidenced by the findings of fitting the ESEEM spectra.
Several attempts were made to understand the ESEEM data in the case of one pair of carbonyl
carbon atoms, but as seen above, this was not the case. There do exist possibilities where the
carbonyl carbon atoms are inequivalent, that is to say, the hyperfine coupling differs between
two carbonyl carbon atoms located on the same semiquinone molecule. These possible instances
have been studied by DFT and take shape of [Na(PSQ)(THF)n] + complexes (where n = 1, 2, 3,
4), where the sodium ion is bonded to a single semiquinone, while other coordination sites are
occupied by various numbers of THF solvent molecules. The results of this study are not
included in this work.
Lastly, when considering a substituent effect, the results of this ESEEM study are a bit
chaotic. Only the dimer and ion-pair from each multicomponent system can be compared. Table
9 tabulates the isotropic, dipolar and rhombic components of the hyperfine coupling found for
the dimers and ion-pairs. Considering the dimer first, the aiso values follow the trend of the CW
spectra, where NaPSQ has a larger value then Na(3,6-(OMe)2-PSQ). This can be explained by
the sigma-accepting nature of the methoxy groups. The oxygen atoms in the methoxy groups

84

have a higher electronegativity that protons; this shifts, or polarizes, the unpaired spin density
away from the carbonyl carbons and results in a lower isotropic hyperfine coupling constant.
Table 2-9. Comparison of hyperfine coupling parameters between NaPSQ components and
Na(3,6-(OMe)2-PSQ) components.
aiso

T

ρ

NaPSQ Dimer

2.17

2.77

0

NaOMe2PSQ Dimer

0.38

7.96

0

NaPSQ Ion-Pair

-1.57

13.57

0.44

NaOMe2PSQ Ion-Pair

-0.41

16.13

0.63

Conversely, when comparing dipolar values, the situation is reversed; dipolar coupling
values are larger for Na(3,6-(OMe)2-PSQ) than for NaPSQ. This can be accounted for by
considering the pi-donating ability of the oxygen atoms of the methoxy groups. The radical
allows for the methoxy oxygen atoms to participate in the resonance of the fused ring system
through the pi electrons, specifically the unshared electrons on the oxygen atoms. An illustration
of this resonance effect is seen in Figure 40 below.

Figure 2-40. Resonance structures of 3,6-(OMe)2-PSQ-∙ illustrating the pi-donating character of
the methoxy group.
While this finding is interesting and agrees with the DFT study, these results are due to a
mixture of species produced by reaction with sodium metal. So it would be advantageous to
synthesize a single ion pair, a dimer and perhaps a homoleptic complex and compare their
85

spectra, as well as mixtures to assess the validity of the analysis of this study. These experiments
will be performed elsewhere as they aren’t the focus of this study.

HYSCORE
HYSCORE (Hyperfine Sublevel CORrElation) spectra were taken at the same magnetic
field value as ESEEM spectra, 3455 G (345.5 mT), with the same tau value, τ = 136 ns, and the
same temperature, 4 K. HYSCORE spectra were not fit, but only simulated. The simulations
presented are simply overlaid onto the spectra themselves. Before diving any further into the
actual results, a qualitative description of HYSCORE spectroscopy is presented.
Concisely, HYSCORE can be described as a modified, two-dimensional, four-pulse
ESEEM experiment. Two-dimensional, three pulse ESEEM experiments do exist, however they
are usually foregone for the HYSCORE experiment, due to low resolution and other
complications63. The pulse sequence for HYSCORE can be seen below; the experiment is a
modified form of three pulse ESEEM with the addition of a π-pulse in between the second and
third π/2 pulses. This additional pulse not only bisects the T dimension into t1 and t2 but creates
nuclear coherence transfer echoes. The second π pulse creates a nuclear coherence evolution,
then the π pulse flips these nuclear coherences between the α and β spin manifolds. This creation
of two, individual nuclear coherence evolution times, allows for Fourier transform in two
dimensions and production of cross peaks at (ω12,ω34) and (ω34,ω12) that correlate to the nuclear
transitions in the ι and β electron spin manifolds. Additionally, due to the nuclear coherence
transfer, the two dimensional patterns produced are not the same in every quadrant; when ω12
and ω34 have the same sign and the coupling is weak (A/2 ˂ ωI) the peaks appear in the first and

86

third quadrant while for the strong coupling case (A/2 ˃ ωI) ω12 and ω34 have different signs and
the peaks are in the second and fourth quadrants63.

Figure 2-41. Four pulse HYSCORE sequence.
The time domain data are worked up similarly to ESEEM time domain data, the greatest
difference being that each time domain is worked up separately until a two-dimensional Fourier
transform is applied. Following FT work-up, the data is presented as a contour plot. In this
section the 13C data have not been divided by their

12

C counterparts, as in the previous ESEEM

section.
Pictured in Figure 42 is the contour plot of [13C2]PSQ. The most noticeable features of
this spectrum is that there are on-diagonal peaks in both the first (weak coupling) and second
quadrant (strong coupling) centered at the

13

C Larmor frequency (3.7 MHz). To compare these

two features, the more intense peak is located in the first quadrant, indicating that there is a
component that has dipolar coupling less than the Larmor frequency, and there exists a less
intense component that has dipolar coupling that is larger than the Larmor frequency. This
notion is supported by the ESEEM analysis. It should also be noted that these on-diagonal peaks
are possibly due to non-ideal pulses and/or superimposed echoes on the HYSCORE modulations.
87

Cross peaks are observed in the first quadrant at (11.23,3.74) and (3.74,11.23) and in the
second quadrant at (-11.23,3.74) and (-3.74,11.23). These peaks are centered about the
Larmor frequency, 3.70 MHz, and thus are the result of

13

13

C

C hyperfine coupling. The shapes of

the peaks in the first quadrant follow that peaks perpendicular to the diagonal are a feature of
weakly coupled nuclei. This is evidence for T ˂ 2νI; a dipolar coupling corresponding to the
proposed dimer structure. In quadrant two, the shapes of the peaks run parallel to the diagonal,
indicating strong coupling of the magnitude T ˃ 2νI. These peaks would then correspond to the
carbonyl carbons from the ion-pair.

88

Figure 2-42. HYSCORE spectrum of [13C2]PSQ. Experimental settings: 345.5 mT Field, 16 ns
time increment, 136 ns tau, 36 ns starting t1 and t2 time.
The [13C2]PSQ HYSCORE spectrum has been simulated in three various fashions: dimer
only, ion-pair only, and one simulation containing both the dimer and the ion-pair. It was done
this way in order to identify the strong and weak components, separately, and then compare them
to the simulation of the entire spectrum (quadrants one and two).

89

Figure 2-43. HYSCORE spectrum (black) and simulation (red) of the [13C2]PSQ dimer.
Simulation parameters: g = 2.00575, 4 13C nuclei, aiso = 2.1713 MHz, T = 2.7757 MHz, ρ = 0
MHz. Experimental conditions: 346 mT magnetic field, 4 pulse HYSCORE sequence, 16 ns
time increment, 136 ns tau, 36 ns starting t1 and t2 values.

The HYSCORE simulation of the dimer, is pictured above in Figure 43. The same
parameters for the ESEEM simulation of the dimer were used here, except for some
experimental parameters. The position of the peak has been accurately modeled, even though the
shape has not. The cross peaks produced by the simulation are merely shy of the observed cross
peaks of the experimental spectrum, and are indicative that the dimer is the source of the weakly
90

coupled carbonyl carbons.

The second quadrant shows no resonances produced by this

simulation and thus indicates no strong coupling from this species.
In contrast to the dimer simulation, the ion-pair simulation (Figure 44, below) has produced
resonances in the first and second quadrants, as well as cross peaks. The ends of the cross peaks
in the second quadrant nearly correspond to the positions of the observed cross peaks. This
shows more evidence that it is possible for a second, strongly coupled component to be present
in solution.

Figure 2-44. HYSCORE spectrum (black) and simulation (red) of the [13C2]PSQ ion-pair.
Simulation parameters: g = 2.00575, 2 13C nuclei, aiso = 1.5741 MHz, T = 13.5674 MHz, ρ = 0
MHz. Experimental conditions: 345.5 mT magnetic field, 4 pulse HYSCORE sequence, 16 ns
time increment, 136 ns tau, 36 ns starting t1 and t2 values.
91

Lastly, the combined simulation of the ion-pair and dimer produces a simulation that is
reminiscent of the simulation of the dimer (Figure 45). It is possible that the intensity of the
dimer is much greater than that of the ion-pair and therefore only the dimer is visible. Similarly,
to constructing two dimensional spectra of HYSCORE data, the simulations also have a Zthreshold that must be varied. This threshold is typically varied until only the resonances are
visible and the noise (baseline) has been eliminated from the spectrum.

That being said, no

other resonances were found in the first or second quadrants by varying the Z-threshold
parameters. Conclusively though, the presence of cross peaks and the ability of the simulations
to produce cross peaks in the first and second quadrants is more evidence to the conclusions of
the ESEEM section.

92

Figure 2-45. HYSCORE spectrum (black) and simulation (red) of the [13C2]PSQ multicomponent system. Simulation parameters: System 1, g = 2.00575, 4 13C nuclei, aiso = 2.1713
MHz, T = 2.7757 MHz, ρ = 0 MHz; System 2, 2 13C nuclei, aiso = 1.5741 MHz, T = 13.5674
MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic field, 4 pulse HYSCORE
sequence, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2 values.
The HYSCORE spectra of [13C2]3,6-(OMe)2-PSQ are presented below. The spectra and
simulation results are quite analogous to those of [13C2]PSQ, and were simulated similarly. Each
species of the multi-component system was simulated separately and the final simulation
includes all the components of the system. As can be seen below in the HYSCORE spectrum of
[13C2]3,6-(OMe)2-PSQ, resonances appear at (3.9,3.9) and (7.55,7.55) in the first quadrant and at

93

(-2.3,2.3) and (-6.3,6.3) in the second quadrant; the only resonance not visible from the
[13C2]PSQ HYSCORE being at (7.55,7.55). It is not a great surprise that these spectra are similar
to those of [13C2]PSQ for several reasons; however, the primary reason being that the ESEEM
spectra produced by each molecule are similar in nature; each have large resonances at or near
the 13C – Larmor frequency, followed by two smaller humps near 8 MHz and 11 MHz.
Another similar facet of this HYSCORE are the prominent features and their respective
resonances. The dominant feature in the [13C2]PSQ HYSCORE (and ESEEM) spectrum, at 3.66
MHz, was the result of dimer formation between two PSQ units and one bridging Na+ ion. This
dimer was the most complex species from that reaction. Moving to [13C2]3,6-(OMe)2-PSQ, the
most complex species was a homoleptic complex, where Na+ was complexed to three PSQ
moieties. Therefore, in each spectrum, the most complex species has the largest resonance at or
near the

13

C – Larmor frequency. This observation is highly dependent on the way in which

these spectra were simulated; however, the final simulations were arrived at after long and
rigorous trial and error periods.

94

Figure 2-46. HYSCORE spectrum of [13C2]3,6-(OMe)2-PSQ. Experimental settings: 345.5 mT
Field, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2 time.

95

Figure 2-47. HYSCORE spectrum (black) and simulation (red) of the [13C2]3,6-(OMe)2-PSQ
homoleptic complex. Simulation parameters: g = 2.0057, 6 13C nuclei, aiso = 0.9758 MHz, T =
1.3064 MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic field, 4 pulse
HYSCORE sequence, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2 values.
The next species that was examined was the dimer. The simulation produced an ondiagonal peak at (6.71,6.71); however, the simulation also produced cross-peak arcs that radiate
from the on-diagonal peak at (6.71,6.71). They range from (4.51,11.72) to (6.35,7.69) and
(7.69,6.35) to (4.51,11.72). These peaks do not appear in the experimental spectrum. In contrast
to these arcs are the arc centered at (10.62,10.62). This arc is not present in the HYSCORE, but
it is close to the feature centered at 11.07 MHz in the ESEEM spectrum, and since the simulation
96

for HYSCORE is based on the simulation for ESEEM, this is most likely showing a true. With
regards to the second quadrant, cross peaks have also been produced by the simulation,
indicating strong coupling being generated from the dimer. The dimer, having T = 7.96 MHz, is
just over the weak coupling limit.

Figure 2-48. HYSCORE spectrum (black) and simulation (red) of the [13C2]3,6-(OMe)2-PSQ
dimer. Simulation parameters: g = 2.0057, 4 13C nuclei, aiso = 0.3849 MHz, T = 7.9653 MHz, ρ
= 0 MHz. Experimental conditions: 345.5 mT magnetic field, 4 pulse HYSCORE sequence, 16
ns time increment, 136 ns tau, 36 ns starting t1 and t2 values.

97

Figure 2-49. HYSCORE spectrum (black) and simulation (red) of the [13C2]3,6-(OMe)2-PSQ
ion-pair in Quadrant II. Simulation parameters: g = 2.0057, 2 13C nuclei, aiso = -0.4155 MHz, T
= 16.1325 MHz, ρ = 0 MHz. Experimental conditions: 345.5 mT magnetic field, 4 pulse
HYSCORE sequence, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2 values.
Lastly, the simulations for the ion-pair and the entire multiple components system. The
simulation for the ion-pair produced features only present in the second quadrant, indicating
coupling larger than twice the Larmor frequency; this is upheld by T = 16.13 MHz for the
molecule. Compare this to the simulation of the entire system and everything becomes irrelevant
except for the homoleptic complex, in Quadrant I. This is analogous to [13C2]PSQ, as the
HYSCORE simulation for the entire system resembled only the dimer system, rather than both
98

components. Again, this could be for a variety of reasons but it seems most likely that the
homoleptic complex may be the most prominent component of the mixture. Additionally,
though, an on-diagonal peak is produced in Quadrant II, precisely at the 13C – Larmor frequency
and frequency of the most intense resonance. It nearly maps onto the existing diagonal peak
from the spectrum.

Figure 2-50. HYSCORE spectrum (black) and simulation (red) of the [13C2]3,6-(OMe)2-PSQ
multi-component system in Quadrant I. Simulation parameters: g = 2.0057, 6 13C nuclei, aiso =
0.9758 MHz, T = 1.3064 MHz, ρ = 0 MHz, 4 13C nuclei, aiso = 0.3849 MHz, T = 7.9653 MHz, ρ
= 0 MHz, and 2 13C nuclei, aiso = -0.4155 MHz, T = 16.1325 MHz, ρ = 0 MHz. Experimental
conditions: 345.5 mT magnetic field, 4 pulse HYSCORE sequence, 16 ns time increment, 136 ns
tau, 36 ns starting t1 and t2 values.
99

Figure 2-51. HYSCORE spectrum (black) and simulation (red) of the [13C2]3,6-(OMe)2-PSQ
multi-component system in Quadrant II. Simulation parameters: g = 2.0057, 6 13C nuclei, aiso =
0.9758 MHz, T = 1.3064 MHz, ρ = 0 MHz, 4 13C nuclei, aiso = 0.3849 MHz, T = 7.9653 MHz, ρ
= 0 MHz, and 2 13C nuclei, aiso = -0.4155 MHz, T = 16.1325 MHz, ρ = 0 MHz. Experimental
conditions: 345.5 mT magnetic field, 4 pulse HYSCORE sequence, 16 ns time increment, 136 ns
tau, 36 ns starting t1 and t2 values.
To conclude the results of the HYSCORE analysis and simulations, the inclusion of these
results may not be as pertinent for the analysis of this chapter, but rather Chapter 4, where
HYSCORE analysis of exchange-coupled Ni-PSQ systems was valuable in characterizing the
100

magnetic structure. The HYSCORE analysis of this chapter would be compared to that of
Chapter 4 to draw conclusions on the presence and absence of a paramagnetic Lewis acid bonded
directly to the semiquinone moiety.
For the purposes of this chapter, however, the HYSCORE analysis bolstered the
conclusions of the ESEEM section for [13C2]PSQ, but not for [13C2](3,6-(OMe)2-PSQ). The
presence of cross peaks in the [13C2]PSQ spectra gives credence to the hypothesis that multiple
species are present in solution. However, the dipolar coupling constants for [13C2]PSQ were
smaller than [13C2](3,6-(OMe)2-PSQ) leading one to believe that cross peaks would be present in
these HYSCORE spectra but this was not the case. Perhaps this was due to un-optimized
experimental conditions; the [13C2](3,6-(OMe)2-PSQ) spectra were taken at 345.5 mT whereas
the [13C2]PSQ spectra were taken at 346 mT. Whatever the cause, the experimental conditions
for both samples would be optimized if the various components were individually synthesized
and examined.

2.4 Conclusions and Future Work
The purpose of the work of this chapter was to uncover an effect, if any, the choice of
substituent has on spin density delocalization within a semiquinone. Though, there were also
minor themes on display through out this investigation: the validity of

13

C as a monitor of spin

density fluctuation, the influence of counter ion on both species formation and spin density
delocalization, and the limits of EPR, ESEEM and HYSCORE to study these semiquinones.
The over-arching theme to this entire work, present chapter included, is the influence a
substituent has on the spin properties of phenanthrenesemiquinones. Within the molecules
studied, it is evident that a substituent effect is active. It can be seen from merely comparing the

101

CW spectra by eye, that choice of substituent has an overall effect on line shape; which is
indicative of the spin properties of the molecule. A dependence on substituent choice was also
observed quantitatively, in comparing hyperfine coupling parameters obtained by EPR and
ESEEM.
It was observed that

13

C aiso values were larger for PSQ than 3,6-(OMe)2-PSQ, which was

predicted by DFT, even without inclusion of Na+. Although DFT incorrectly predicted the 23Na
aiso, this same calculation very closely predicted the isotropic hyperfine coupling constants for
both

13

C and 1H nuclei. On the other hand, when considering the ESEEM data, dipolar values

showed the opposite trend; for comparative species (i.e. molecule versus molecule and dimer
versus dimer),

13

C T values were slightly larger for 3,6-(OMe)2-PSQ than PSQ. This was an

interesting revelation considering DFT predicted that the total spin density for PSQ would be
larger than 3,6-(OMe)2-PSQ. This being said, using 13C substitution at the carbonyl carbon as a
means for tracking variances in hyperfine coupling, seems to be a sound method. CW EPR and
ESEEM also both proved to be powerful techniques in studying these molecules. The only
shortfall in the results was from HYSCORE. At the present time it does not seem that much can
be done to improve upon the lack of cross peaks in the spectra.
These results can be improved upon. The largest issue within this study lies with the method
of reduction. Quinones were reduced with excess Na metal chunks in THF under anaerobic
conditions. The Na metal were cut up into small cubes in order to reduce the surface chemistry;
however, as witnessed by the ESEEM data and simulations, this may have been the largest road
block in determining the make-up of the multi-component system. It was originally thought that
only the PSQ-∙ would be produced and Na would not even participate in the hyperfine interaction.
This hypothesis was undercut immediately, when CW data clearly showed that the 1H nuclei of

102

the PSQ moiety were not the only active nuclei present.

This sodium issue also led to

convoluted ESEEM and HYSCORE data as it seemed that the surface of the Na chunks had
given way to various reactions, producing several different species in solution. This is also
evidenced with the range of

13

C aiso values obtained from CW EPR measurements.

An

alternative hypothesis is that during freezing (ESEEM and HYSCORE experiments conducted
between 4 -10 K), small amounts of the various species precipitated out of solution, causing
convoluted spectra. This sodium issue could be remedied with alternative reduction techniques.
In the following chapter, the techniques and methods used in this chapter will be applied to
GaIII and ZnII complexes of PSQ and [13C2]PSQ. These results will be directly compared to the
results of this chapter to deduce the effects of Lewis acidity on spin delocalization.

103

APPENDIX

104

APPENDIX: Simulation/Fitting Scripts
CW-EPR
function res = PSQcwRTblc_fit(hpar)
% PSQ Room Temp McCracken Re-Take and Spectral Fit
%==============================================================
% Load the experimental spectrum
[B,y,params] = eprload('PSQcwRTblc.DTA'); %Loads the spectrum
B=B/10; %Gauss to mT
Edata = y/(max(y)-min(y)); % Normalizes the data
lmin = 185; % min point to start w/in the data set
lmax = 800; % max point to stop w/in data set
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.00754;
Sys0.lwpp = [0.0005 0.0191]; %Gaussian & Lorentzian Linewidth, mT
Sys0.Nucs = '23Na,1H,1H,1H,1H'; % Treating the nuclei separately
Sys0.n = [1 2 2 2 2]; % 1 23Na, 4 classes of 2, 1H nuclei each
Sys0.A = [1.3231 0.7723 1.1914 -4.0643 -4.7515];
% Load the experimental Parameters
Exp.mwFreq = params.MWFQ/1.00e+09;
Exp.CenterSweep = [params.A1CT*1000,params.A1SW*1000];
Exp.ModAmp = 0.002; %mT
Exp.nPoints = params.XPTS;
% Call the simulation program
[B,sim] = garlic(Sys0,Exp);
Cdata = sim/(max(sim)-min(sim)); % Normalizes the simulation
%offset = Edata(lmin); % corrects for baseline offset
%Cdata = Cdata + offset; % Corrected baseline
% Calculation of residual -> value that determines difference between sim
% and spectrum.
res = 0.0;
for ipt = lmin:lmax,
res = res + (Cdata(ipt) - Edata(ipt))^2;
end;
plot(B(lmin:lmax),Cdata(lmin:lmax),' r',B(lmin:lmax),Edata(lmin:lmax),' k');
return;

105

function res = ldm1152cwRTblc_fit(hpar)
% LDM1152, [13C2]PSQ, Room Temp McCracken Re-Take and Spectral Fit
%==========================================================================
% Load the experimental spectrum
[B,spc,params] = eprload('ldm1152cwRTblc.DTA'); %Loads the spectrum
B=B/10; %Gauss to mT
Edata = spc/(max(spc)-min(spc)); % Normalizes the data
lmin = 150; % min point to start w/in the data
lmax = 850; % max point to stop w/in data set
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.00755;
Sys0.lwpp = [0.0001 0.0222]; %mT
Sys0.Nucs = '23Na,1H,1H,1H,1H,13C'; % Treating the nuclei separately
Sys0.n = [1 2 2 2 2 2]; % 1 23Na, 4 classes of 1H, 2 13C nuclei
Sys0.A = [1.3231 0.7723 1.1914 -4.0643 -4.7515 1.57];
% Load the experimental Parameters
Exp.mwFreq = params.MWFQ/1.00e+09;
Exp.CenterSweep = [params.A1CT*1000,params.A1SW*1000];
Exp.ModAmp = 0.002; % mT
Exp.nPoints = params.XPTS;
% Call the simulation program
[B,sim] = garlic(Sys0,Exp);
Cdata = sim/(max(sim)-min(sim)); % Normalizes the simulation
offset = Edata(lmin); % corrects for baseline offset
Cdata = Cdata + offset; % Corrected baseline
% Calculation of residual -> value that determines difference between sim
% and spectrum.
res = 0.0;
for ipt = lmin:lmax,
res = res + (Cdata(ipt) - Edata(ipt))^2;
end;
plot(B(lmin:lmax),Cdata(lmin:lmax),' r',B(lmin:lmax),Edata(lmin:lmax),' k');
return;

106

function res = ldm2140cwRTblc_fit(hpar)
% LDM2140, Na(3,6-(OMe)2-PSQ), Room Temp McCracken Re-Take and Spectral Fit
%==========================================================================
% Load the experimental spectrum
[B,y,params] = eprload('ldm2140cwRTblc.DTA'); %Loads the spectrum
B=B/10; %Gauss to mT
Edata = y/(max(y)-min(y)); % Normalizes the data
lmin = 220; % min point to start
lmax = 790; % max point to stop
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.00746;
Sys0.lwpp = [0.0015 0.0135];
Sys0.Nucs = '23Na,1H,1H,1H,1H,1H'; % Treating the nuclei separately
Sys0.n = [1 2 4 2 2 2]; % 1 23Na, 5 classes of 1H nuclei
Sys0.A = [-1.1677 -0.4142 0.5611 1.0046 0.9607 -3.8385];
% Load the experimental Parameters
Exp.mwFreq = params.MWFQ/1.00e+09;
Exp.CenterSweep = [params.A1CT*1000,params.A1SW*1000];
Exp.ModAmp = 0.002;
Exp.nPoints = params.XPTS;
% Call the simulation program
[B,sim] = garlic(Sys0,Exp);
Cdata = sim/(max(sim)-min(sim)); % Normalizes the simulation
%offset = Edata(lmin); % corrects for baseline offset
%Cdata = Cdata + offset; % Corrected baseline
% Calculation of residual -> value that determines difference between sim
% and spectrum.
res = 0.0;
for ipt = lmin:lmax,
res = res + (Cdata(ipt) - Edata(ipt))^2;
end;
plot(B(lmin:lmax),Cdata(lmin:lmax),' r',B(lmin:lmax),Edata(lmin:lmax),' k');
return;

107

function res = ldm2134cwRTblc_fit(hpar)
% LDM2134, [13C2](3,6-(OMe)2-PSQ) Room Temp 20mG Sim/Fit McCracken Retake
%==========================================================================
% Load the experimental spectrum
[B,spc,params] = eprload('ldm2134cwRTblc.DTA'); %Loads the spectrum
B=B/10; %Gauss to mT
Edata = spc/(max(spc)-min(spc)); % Normalizes the data
lmin = 220; % min point to start
lmax = 790; % max point to stop
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.007443;
Sys0.lwpp = [0.0015 0.0145];
Sys0.Nucs = '23Na,1H,1H,1H,1H,1H,13C'; % Treating the nuclei separately
Sys0.n = [1 2 4 2 2 2 2]; % 1 23Na, 5 classes of 1H, 2 13C nuclei
Sys0.A = [-1.1677 -0.4142 0.5611 1.0046 0.9607 -3.8385 -0.4155]; % MHz
% Load the experimental Parameters
Exp.mwFreq = params.MWFQ/1.00e+09;
Exp.CenterSweep = [params.A1CT*1000,params.A1SW*1000];
Exp.ModAmp = 0.002;
Exp.nPoints = params.XPTS;
% Call the simulation program
[B,sim] = garlic(Sys0,Exp);
Cdata = sim/(max(sim)-min(sim)); % Normalizes the simulation
%offset = Edata(lmin); % corrects for baseline offset
%Cdata = Cdata + offset; % Corrected baseline
% Calculation of residual -> value that determines difference between sim
% and spectrum.
res = 0.0;
for ipt = lmin:lmax,
res = res + (Cdata(ipt) - Edata(ipt))^2;
end;
plot(B(lmin:lmax),Cdata(lmin:lmax),' r',B(lmin:lmax),Edata(lmin:lmax),' k');
return;

108

ESEEM
function res = LDM1152_ESEEM_FFT_1peak_fit(hpar)
% LDM1152, [13C2]PSQ Frequency Domain Sim/Fit
%==========================================================================
% Load the experimental spectrum
[freq,intens] = eprload('LDM1152divPSQnormSP.DTA');
freq = freq*1000; %GHz to MHz
lmin = 1;
lmax = 256;
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.00575;
Sys0.Nucs = '13C,13C';
Sys0.n = [1 1]; % 2 classes of 13C nuclei
Sys0.A_ = [2.1713 2.7757 0; 2.1713 2.7757 0]; % [a-iso T rhombicity]
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012; % Time Increment/microseconds
Exp.tau = 0.136; % Tau/microseconds
Exp.T = 0.04; % Starting T value/microseconds
Opt.nKnots = 90;
[x,y] = saffron(Sys0,Exp,Opt);
ESnormDAL; %Normalizes the simulation
res = 0;
for ipt = (lmin+1):lmax,
res = res + (spec(ipt) - intens(ipt))^2;
end;
plot(f(lmin:lmax),spec(lmin:lmax),' r',freq(lmin:lmax),intens(lmin:lmax),'
k');
return;

109

function res = LDM1152_ESEEM_time_1peak_fit(hpar)
% LDM1152, [13C2]PSQ Time Domain Sim/Fit
%==========================================================================
% Load the experimental spectrum
[time,amp] = eprload('LDM1152divPSQnorm.DTA');
Amp = real(amp)/max(real(amp)); % Normalizes the data
time = (time/1000)+0.136;
lmin = 1;
lmax = 512;
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.00575;
Sys0.Nucs = '13C,13C';
Sys0.n = [1 1]; % 2 classes of 13C nuclei
Sys0.A_ = [2.1713 2.7757 0; 2.1713 2.7757 0]; % [a-iso T rhombicity]
% Load the experimental Parameters
Exp.Field = 345.5; % Magnetic Field/millitesla
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012; % Time Increment/microseconds
Exp.tau = 0.136; % Tau/microseconds
Exp.T = 0.04; % Starting T value/microseconds
Opt.nKnots = 90;
Opt.TimeDomain = 1;
[x,y] = saffron(Sys0,Exp,Opt);
Y = real(y)/max(real(y)); % Normalizes the simulation
res = 0;
for ipt = lmin:lmax,
res = res + (Y(ipt) - Amp(ipt))^2;
end;
plot(x(lmin:lmax),Y(lmin:lmax),' r',time(lmin:lmax),Amp(lmin:lmax),' k');
return;

110

function res = LDM1152_ESEEM_FFT_mccracke_fitC(hpar)
% LDM1152, [13C2]PSQ Multicomponent Frequency Domain Sim/Fit
%==========================================================================
% Load the experimental spectrum
[freq,intens] = eprload('LDM1152divPSQnormSP.DTA');
freq = freq*1000;
lmin = 1;
lmax = 256;
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.00575;
Sys0.Nucs = '13C,13C,13C,13C';
Sys0.A_ = [2.1713 2.7757 0;2.1713 2.7757 0;2.1713 2.7757 0;2.1713 2.7757 0];
% Species 2
Sys1.S = 1/2;
Sys1.g = 2.00575;
Sys1.Nucs = '13C,13C';
Sys1.A_ = [-1.5741 13.5674 0.44;-1.5741 13.5674 0.44]; % [a-iso T rhombicity]
% Load the experimental Parameters
Exp.Field = 345.5; % Magnetic Field/millitesla
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012; % Time Increment/microseconds
Exp.tau = 0.136; % Tau/microseconds
Exp.T = 0.04; % Starting T Value/microseconds
Opt.nKnots = 180;
[x0,y0] = saffron(Sys0,Exp,Opt);
[x,y1] = saffron(Sys1,Exp,Opt);
y = y0 + y1; % combine time domain traces and process
ESnormDAL; %Normalizes the simulation
res = 0;
for ipt = (lmin+1):lmax,
res = res + (spec(ipt) - intens(ipt))^2;
end;
plot(f(lmin:lmax),spec(lmin:lmax),' r',freq(lmin:lmax),intens(lmin:lmax),'
k');
return;

111

function res = LDM1152_ESEEM_time_fit(hpar)
% LDM1152 [13C2]PSQ Multicomponent Time Domain Sim/Fit
%==========================================================================
% Load the experimental spectrum
[time,amp] = eprload('LDM1152divPSQnorm.DTA');
Amp = real(amp)/max(real(amp)); % Normalizes the data
time = (time/1000)+0.136;
lmin = 1;
lmax = 512;
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.00575;
Sys0.Nucs = '13C,13C,13C,13C';
Sys0.A_ = [2.1713 2.7757 0;2.1713 2.7757 0;2.1713 2.7757 0;2.1713 2.7757 0];
% Species 2
Sys1.S = 1/2;
Sys1.g = 2.00575;
Sys1.Nucs = '13C,13C';
Sys1.A_ = [-1.5741 13.5674 0.44;-1.5741 13.5674 0.44]; % [a-iso T rhombicity]
% Experimental Parameters
Exp.Field = 345.5; % Magnetic Field/millitesla
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012; % Time Increment/microseconds
Exp.tau = 0.136; % Tau/microseconds
Exp.T = 0.04; % Starting T Value/microseconds
Opt.nKnots = 90;
[x0,y0] = saffron(Sys0,Exp,Opt);
[x,y1] = saffron(Sys1,Exp,Opt);
y = y0 + y1;
Y = real(y)/max(real(y)); % Normalizes the simulation
res = 0;
for ipt = (lmin+2):lmax,
res = res + (Y(ipt) - Amp(ipt))^2;
end;
plot(x(lmin:lmax),Y(lmin:lmax),' r',time(lmin:lmax),Amp(lmin:lmax),' k');
return;

112

function res = LDM2134div2140ESEEM_1peak_fit(hpar)
% LDM2134/LDM2140 [13C2](3,6-(OMe)2-PSQ) Frequency Domain Sim/Fit
%==========================================================================
% Load the experimental spectrum
[freq,intens] = eprload('ldm2134div2140_136sp.DTA');
freq = freq*1000;
lmin = 1;
lmax = 256;
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.0057;
Sys0.Nucs = '13C,13C,13C,13C,13C,13C';
Sys0.n = [1 1 1 1 1 1]; % 6 classes of 13C nuclei
Sys0.A_ = [0.9758 1.15 0;0.9758 1.15 0;0.9758 1.15 0;0.9758 1.15 0;0.9758
1.15 0;0.9758 1.15 0];
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012;
Exp.tau = 0.136;
Exp.T = 0.04;
Opt.nKnots = 120;
[x,y] = saffron(Sys0,Exp,Opt);
ESnormDAL; % Normalizes the simulation
res = 0;
for ipt = (lmin):lmax,
res = res + (spec(ipt) - intens(ipt))^2;
end;
plot(f(lmin:lmax),spec(lmin:lmax),' r',freq(lmin:lmax),intens(lmin:lmax),'
k');
return;

113

function res = LDM2134div2140ESEEMtime_1peak_fit(hpar)
% LDM2134/LDM2140 [13C2](3,6-(OMe)2-PSQ) Time Domain Sim/Fit
%==========================================================================
% Load the experimental spectrum
[time,amp] = eprload('ldm2134div2140_136.DTA');
Amp = real(amp)/max(real(amp)); % Normalizes the data
time = (time/1000)+0.136;
lmin = 1;
lmax = 512;
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.00507;
Sys0.Nucs = '13C,13C,13C,13C,13C,13C';
Sys0.n = [1 1 1 1 1 1]; % 6 classes of 13C nuclei
Sys0.A_ = [0.9758 1.15 0;0.9758 1.15 0;0.9758 1.15 0;0.9758 1.15 0;0.9758
1.15 0;0.9758 1.15 0];
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012;
Exp.tau = 0.136;
Exp.T = 0.04;
Opt.nKnots = 90;
[x,y] = saffron(Sys0,Exp,Opt);
Y = real(y)/max(real(y)); % Normalizes the simulation
res = 0;
for ipt = (lmin):lmax,
res = res + (Y(ipt) - Amp(ipt))^2;
end;
plot(x(lmin:lmax),Y(lmin:lmax),' r',time(lmin:lmax),Amp(lmin:lmax),' k');
return;

114

function res = LDM2134div2140_ESEEM_FFT_fitC(hpar)
% LDM2134/LDM2140 [13C2](3,6-(OMe)2-PSQ) Multicomponent Frequency Domain
Sim/Fit
%==========================================================================
% Load the experimental spectrum
[freq,intens] = eprload('ldm2134div2140_136sp.DTA');
freq = freq*1000;
lmin = 1;
lmax = 256;
% Species 1
Sys1.S = 1/2;
Sys1.g = 2.0057;
Sys1.Nucs = '13C,13C,13C,13C,13C,13C';
Sys1.A_ = [0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064
0;0.9758 1.3064 0;0.9758 1.3064 0];
% Species 1a
Sys1a.S = 1/2;
Sys1a.g = 2.0057;
Sys1a.Nucs = '13C,13C,13C,13C,13C,13C';
Sys1a.A_ = [0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064
0;0.9758 1.3064 0;0.9758 1.3064 0];
% Species 1b
Sys1b.S = 1/2;
Sys1b.g = 2.0057;
Sys1b.Nucs = '13C,13C,13C,13C,13C,13C';
Sys1b.A_ = [0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064
0;0.9758 1.3064 0;0.9758 1.3064 0];
% Species 1c
Sys1c.S = 1/2;
Sys1c.g = 2.0057;
Sys1c.Nucs = '13C,13C,13C,13C,13C,13C';
Sys1c.A_ = [0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064
0;0.9758 1.3064 0;0.9758 1.3064 0];
% Species 2
Sys2.S = 1/2;
Sys2.g = 2.0057;
Sys2.Nucs = '13C,13C';
Sys2.A_ = [-0.4155 16.1325 0.63;-0.4155 16.1325 0.63];
% Species 3
Sys3.S = 1/2;
Sys3.g = 2.0057;
Sys3.Nucs = '13C,13C,13C,13C';

115

Sys3.A_ = [0.3849 7.9653 0;0.3849 7.9653 0;0.3849 7.9653 0;0.3849 7.9653 0];
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012;
Exp.tau = 0.136;
Exp.T = 0.04;
Opt.nKnots = 180;
[x1,y1] = saffron(Sys1,Exp,Opt);
[x1a,y1a] = saffron(Sys1a,Exp,Opt);
[x1b,y1b] = saffron(Sys1b,Exp,Opt);
[x1c,y1c] = saffron(Sys1c,Exp,Opt);
[x2,y2] = saffron(Sys2,Exp,Opt);
[x,y3] = saffron(Sys3,Exp,Opt);
% combine time domain traces and process
y = y1 + y1a + y1b + y1c + y2 + y3;
ESnormDAL; %Normalizes the simulation
res = 0;
for ipt = (lmin+1):lmax,
res = res + (spec(ipt) - intens(ipt))^2;
end;
plot(f(lmin:lmax),spec(lmin:lmax),' r',freq(lmin:lmax),intens(lmin:lmax),'
k');
return;

116

function res = LDM2134div2140ESEEMtime_fit(hpar)
% LDM2134/LDM2140 [13C2](3,6-(OMe)2-PSQ) Multicomponent Time Domain Sim/Fit
%==========================================================================
% Load the experimental spectrum
[time,amp] = eprload('ldm2134div2140_136.DTA');
Amp = real(amp)/max(real(amp)); % Normalizes the data
time = (time/1000)+0.136;
lmin = 1;
lmax = 512;
% Species 1
Sys1.S = 1/2;
Sys1.g = 2.0057;
Sys1.Nucs = '13C,13C,13C,13C,13C,13C';
Sys1.A_ = [0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064
0;0.9758 1.3064 0;0.9758 1.3064 0];
% Species 1a
Sys1a.S = 1/2;
Sys1a.g = 2.0057;
Sys1a.Nucs = '13C,13C,13C,13C,13C,13C';
Sys1a.A_ = [0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064
0;0.9758 1.3064 0;0.9758 1.3064 0];
% Species 1b
Sys1b.S = 1/2;
Sys1b.g = 2.0057;
Sys1b.Nucs = '13C,13C,13C,13C,13C,13C';
Sys1b.A_ = [0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064
0;0.9758 1.3064 0;0.9758 1.3064 0];
% Species 1c
Sys1c.S = 1/2;
Sys1c.g = 2.0057;
Sys1c.Nucs = '13C,13C,13C,13C,13C,13C';
Sys1c.A_ = [0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064
0;0.9758 1.3064 0;0.9758 1.3064 0];
% Species 2
Sys2.S = 1/2;
Sys2.g = 2.0057;
Sys2.Nucs = '13C,13C';
Sys2.A_ = [-0.4155 16.1325 0.63;-0.4155 16.1325 0.63];
% Species 3
Sys3.S = 1/2;
Sys3.g = 2.0057;
Sys3.Nucs = '13C,13C,13C,13C';
Sys3.A_ = [0.3849 7.9653 0;0.3849 7.9653 0;0.3849 7.9653 0;0.3849 7.9653 0];

117

% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012;
Exp.tau = 0.136;
Exp.T = 0.04;
Opt.nKnots = 180;
[x1,y1] = saffron(Sys1,Exp,Opt);
[x1a,y1a] = saffron(Sys1a,Exp,Opt);
[x1b,y1b] = saffron(Sys1b,Exp,Opt);
[x1c,y1c] = saffron(Sys1c,Exp,Opt);
[x2,y2] = saffron(Sys2,Exp,Opt);
[x,y3] = saffron(Sys3,Exp,Opt);
% combine time domain traces and process
y = y1 + y1a + y1b + y1c + y2 + y3;
Y = real(y)/max(real(y)); % Normalizes the simulation
res = 0;
for ipt = (lmin):lmax,
res = res + (Y(ipt) - Amp(ipt))^2;
end;
plot(x(lmin:lmax),Y(lmin:lmax),' r',time(lmin:lmax),Amp(lmin:lmax),' k');
return;

118

HYSCORE
% LDM1152 [13C2]PSQ Dimer
%==========================================================================
clear;
Sys0.S = 1/2;
Sys0.g = 2.00575;
Sys0.Nucs = '13C,13C,13C,13C'; % Treating the nuclei differently
Sys0.A_ = [2.1713 2.7757 0;2.1713 2.7757 0;2.1713 2.7757 0;2.1713 2.7757 0];
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = 'HYSCORE';
Exp.dt = 0.016;
Exp.tau = 0.136;
Exp.t1 = 0.036;
Exp.t2 = 0.036;
Exp.nPoints = [256 256];
Opt.nKnots = 181;
[x1,x2,y1,p1] = saffron(Sys0,Exp,Opt);
% Plot the data
f1 = p1.f1;
f2 = p1.f2;
Z = (real(p1.fd));
% set threshhold for contour at 10% max amplitude
Zthresh = 0.1*max(max(Z));
for iy=1:512,
for ix = 1:512,
if Z(ix,iy) < Zthresh,
ZR(ix,iy) = Zthresh;
else
ZR(ix,iy) = Z(ix,iy);
end
end;
end;
%
f1min=0.0;
f1max=7.0;
f2min=0.0;
f2max=7.0;
%
frinc=f1(2)-f1(1);
%
xmin=floor((f1min-f1(1))/frinc)+1;
xmax=floor((f1max-f1(1))/frinc)+1;
ymin=floor((f2min-f2(1))/frinc)+1;
ymax=floor((f2max-f2(1))/frinc)+1;

119

contour(f1(xmin:xmax),f2(ymin:ymax),ZR(ymin:ymax,xmin:xmax),10,'r');
xlabel('f1 (MHz)');
ylabel('f2 (MHz)');
grid;
% LDM1152 HYSCORE Ion-Pair Simulation
%==========================================================================
clear;
Sys0.S = 1/2;
Sys0.g = 2.00575;
Sys0.Nucs = '13C,13C';
Sys0.A_ = [-1.5741 13.5674 0.44;-1.5741 13.5674 0.44];
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = 'HYSCORE';
Exp.dt = 0.016;
Exp.tau = 0.136;
Exp.t1 = 0.036;
Exp.t2 = 0.036;
Exp.nPoints = [256 256];
Opt.nKnots = 181;
[x1,x2,y1,p1] = saffron(Sys0,Exp,Opt);
% Plot the data
f1 = p1.f1;
f2 = p1.f2;
Z = (real(p1.fd));
% set threshhold for contour at 10% max amplitude
Zthresh = 0.02*max(max(Z));
for iy=1:512,
for ix = 1:512,
if Z(ix,iy) < Zthresh,
ZR(ix,iy) = Zthresh;
else
ZR(ix,iy) = Z(ix,iy);
end
end;
end;
%
f1min=-15.0;
f1max=15.0;
f2min=0.0;
f2max=15.0;
%
frinc=f1(2)-f1(1);
%
xmin=floor((f1min-f1(1))/frinc)+1;

120

xmax=floor((f1max-f1(1))/frinc)+1;
ymin=floor((f2min-f2(1))/frinc)+1;
ymax=floor((f2max-f2(1))/frinc)+1;
contour(f1(xmin:xmax),f2(ymin:ymax),ZR(ymin:ymax,xmin:xmax),10,'r');
xlabel('f1 (MHz)');
ylabel('f2 (MHz)');
grid;

%LDM1152 [13C2]PSQ HYSCORE Simulation Dimer & Ion Pair
%==========================================================================
clear;
Sys0.S = 1/2;
Sys0.g = 2.00575;
Sys0.Nucs = '13C,13C,13C,13C';
Sys0.A_ = [2.1713 2.7757 0;2.1713 2.7757 0;2.1713 2.7757 0;2.1713 2.7757 0];
Sys1.S = 1/2;
Sys1.g = 2.00575;
Sys1.Nucs = '13C,13C';
Sys1.A_ = [-1.5741 13.5674 0.44;-1.5741 13.5674 0.44];
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = 'HYSCORE';
Exp.dt = 0.016;
Exp.tau = 0.136;
Exp.t1 = 0.036;
Exp.t2 = 0.036;
Exp.nPoints = [256 256];
Opt.nKnots = 181;
[x1,x2,y1,p1] = saffron(Sys0,Exp,Opt);
[x1,x2,y2,p2] = saffron(Sys1,Exp,Opt);
% Plot the data
f1 = p1.f1;
f2 = p1.f2;
Z = (real(p1.fd)) + (real(p2.fd));
ZR = zeros(512,512);
% Set threshhold for contour at 10% max amplitude
Zthresh = 0.02*max(max(Z));
for iy=1:512,
for ix = 1:512,
if Z(ix,iy) < Zthresh,
ZR(ix,iy) = Zthresh;
else
ZR(ix,iy) = Z(ix,iy);

121

end
end;
end;
f1min=-15.0;
f1max=15.0;
f2min=0.0;
f2max=15.0;
frinc=f1(2)-f1(1);
xmin=floor((f1min-f1(1))/frinc)+1;
xmax=floor((f1max-f1(1))/frinc)+1;
ymin=floor((f2min-f2(1))/frinc)+1;
ymax=floor((f2max-f2(1))/frinc)+1;
contour(f1(xmin:xmax),f2(ymin:ymax),ZR(ymin:ymax,xmin:xmax),20,'r');
xlabel('f1 (MHz)');
ylabel('f2 (MHz)');
grid;
return;

% LDM2134 [13C2](3,6-(OMe)2-PSQ) HYSCORE Tris Complex
%==========================================================================
clear;
Sys0.S = 1/2;
Sys0.g = 2.0057;
Sys0.Nucs = '13C,13C,13C,13C,13C,13C';
Sys0.A_ = [0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064
0;0.9758 1.3064 0;0.9758 1.3064 0];
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = 'HYSCORE';
Exp.dt = 0.016;
Exp.tau = 0.136;
Exp.t1 = 0.036;
Exp.t2 = 0.036;
Exp.nPoints = [256 256];
Opt.nKnots = 181;
[x1,x2,y1,p1] = saffron(Sys0,Exp,Opt);
% Plot the data
f1 = p1.f1;
f2 = p1.f2;
Z = (real(p1.fd));
% set threshhold for contour at 10% max amplitude

122

Zthresh = 0.01*max(max(Z));
for iy=1:512,
for ix = 1:512,
if Z(ix,iy) < Zthresh,
ZR(ix,iy) = Zthresh;
else
ZR(ix,iy) = Z(ix,iy);
end
end;
end;
%
f1min=-7.0;
f1max=7.0;
f2min=0.0;
f2max=7.0;
%
frinc=f1(2)-f1(1);
%
xmin=floor((f1min-f1(1))/frinc)+1;
xmax=floor((f1max-f1(1))/frinc)+1;
ymin=floor((f2min-f2(1))/frinc)+1;
ymax=floor((f2max-f2(1))/frinc)+1;
contour(f1(xmin:xmax),f2(ymin:ymax),ZR(ymin:ymax,xmin:xmax),10,'r');
xlabel('f1 (MHz)');
ylabel('f2 (MHz)');
grid;

% LDM2134 [13C2](3,6-(OMe)2-PSQ) HYSCORE Dimer Simulation
%==========================================================================
clear;
Sys0.S = 1/2;
Sys0.g = 2.0057;
Sys0.Nucs = '13C,13C,13C,13C';
Sys0.A_ = [0.3849 7.9653 0;0.3849 7.9653 0;0.3849 7.9653 0;0.3849 7.9653 0];
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = 'HYSCORE';
Exp.dt = 0.016;
Exp.tau = 0.136;
Exp.t1 = 0.036;
Exp.t2 = 0.036;
Exp.nPoints = [256 256];
Opt.nKnots = 181;

123

[x1,x2,y1,p1] = saffron(Sys0,Exp,Opt);
% Plot the data
f1 = p1.f1;
f2 = p1.f2;
Z = (real(p1.fd));
% set threshhold for contour at 10% max amplitude
Zthresh = 0.05*max(max(Z));
for iy=1:512,
for ix = 1:512,
if Z(ix,iy) < Zthresh,
ZR(ix,iy) = Zthresh;
else
ZR(ix,iy) = Z(ix,iy);
end
end;
end;
%
f1min=-15.0;
f1max=15.0;
f2min=0.0;
f2max=15.0;
%
frinc=f1(2)-f1(1);
%
xmin=floor((f1min-f1(1))/frinc)+1;
xmax=floor((f1max-f1(1))/frinc)+1;
ymin=floor((f2min-f2(1))/frinc)+1;
ymax=floor((f2max-f2(1))/frinc)+1;
contour(f1(xmin:xmax),f2(ymin:ymax),ZR(ymin:ymax,xmin:xmax),10,'r');
xlabel('f1 (MHz)');
ylabel('f2 (MHz)');
grid;

%LDM2134 [13C2](3,6-(OMe)2-PSQ) HYSCORE Simulation Molecule
%==========================================================================
clear;
Sys0.S = 1/2;
Sys0.g = 2.00575;
Sys0.Nucs = '13C,13C';
Sys0.A_ = [-0.4155 16.1325 0.63;-0.4155 16.1325 0.63];
% Load the experimental Parameters

124

Exp.Field = 345.5;
Exp.Sequence = 'HYSCORE';
Exp.dt = 0.016;
Exp.tau = 0.136;
Exp.t1 = 0.036;
Exp.t2 = 0.036;
Exp.nPoints = [256 256];
Opt.nKnots = 181;
[x1,x2,y1,p1] = saffron(Sys0,Exp,Opt);
% Plot the data
f1 = p1.f1;
f2 = p1.f2;
Z = (real(p1.fd));
% set threshhold for contour at 10% max amplitude
Zthresh = 0.05*max(max(Z));
for iy=1:512,
for ix = 1:512,
if Z(ix,iy) < Zthresh,
ZR(ix,iy) = Zthresh;
else
ZR(ix,iy) = Z(ix,iy);
end
end;
end;
%
f1min=-25.0;
f1max=25.0;
f2min=0.0;
f2max=25.0;
%
frinc=f1(2)-f1(1);
%
xmin=floor((f1min-f1(1))/frinc)+1;
xmax=floor((f1max-f1(1))/frinc)+1;
ymin=floor((f2min-f2(1))/frinc)+1;
ymax=floor((f2max-f2(1))/frinc)+1;
contour(f1(xmin:xmax),f2(ymin:ymax),ZR(ymin:ymax,xmin:xmax),10,'r');
xlabel('f1 (MHz)');
ylabel('f2 (MHz)');
grid;

125

%LDM2134 [13C2](3,6-(OMe)2-PSQ) HYSCORE Simulation All Components
%==========================================================================
clear;
% Species 1
Sys1.S = 1/2;
Sys1.g = 2.0057;
Sys1.Nucs = '13C,13C,13C,13C,13C,13C'; % Treating the nuclei seperately
Sys1.A_ = [0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064 0;0.9758 1.3064
0;0.9758 1.3064 0;0.9758 1.3064 0];
% Species 2
Sys2.S = 1/2;
Sys2.g = 2.0057;
Sys2.Nucs = '13C,13C';
Sys2.A_ = [-0.4155 16.1325 0.63;-0.4155 16.1325 0.63];
% Species 3
Sys3.S = 1/2;
Sys3.g = 2.0057;
Sys3.Nucs = '13C,13C,13C,13C';
Sys3.A_ = [0.3849 7.9653 0;0.3849 7.9653 0;0.3849 7.9653 0;0.3849 7.9653 0];
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = 'HYSCORE';
Exp.dt = 0.016;
Exp.tau = 0.136;
Exp.t1 = 0.036;
Exp.t2 = 0.036;
Exp.nPoints = [256 256];
Opt.nKnots = 181;
[x1,x2,y1,p1] = saffron(Sys1,Exp,Opt);
[x1,x2,y2,p2] = saffron(Sys2,Exp,Opt);
[x1,x2,y3,p3] = saffron(Sys3,Exp,Opt);
% Plot the data
f1 = p1.f1;
f2 = p1.f2;
Z = (real(p1.fd)) + (real(p2.fd)) + (real(p3.fd));
% set threshhold for contour at 10% max amplitude
Zthresh = 0.01*max(max(Z));
for iy=1:512,
for ix = 1:512,
if Z(ix,iy) < Zthresh,
ZR(ix,iy) = Zthresh;
else

126

ZR(ix,iy) = Z(ix,iy);
end
end;
end;
%
f1min=-25.0;
f1max=0.0;
f2min=0.0;
f2max=25.0;
%
frinc=f1(2)-f1(1);
%
xmin=floor((f1min-f1(1))/frinc)+1;
xmax=floor((f1max-f1(1))/frinc)+1;
ymin=floor((f2min-f2(1))/frinc)+1;
ymax=floor((f2max-f2(1))/frinc)+1;
contour(f1(xmin:xmax),f2(ymin:ymax),ZR(ymin:ymax,xmin:xmax),15,'r');
xlabel('f1 (MHz)');
ylabel('f2 (MHz)');
grid;

127

REFERENCES

128

REFERENCES

(1)

Chang, H. C.; Kiriya, D. Eur. J. Inorg. Chem. 2013, No. 5–6, 642–652.

(2)

Poddel’sky, A. I.; Cherkasov, V. K.; Abakumov, G. a. Coord. Chem. Rev. 2009, 253, 291–
324.

(3)

Evangelio, E.; Ruiz-Molina, D. Eur. J. Inorg. Chem. 2005, 2005 (15), 2957–2971.

(4)

Pierpont, C. G.; Hendrickson, D. N. In Topics in Current Chemistry; Goodwin, H. A., Ed.;
Springer: Berlin, Heidleberg, 2004.

(5)

Almeida, W. B. De; Taguchi, A. T.; Dikanov, S. a; Wraight, C. a; Malley, P. J. O. J. Phys.
Chem. Lett. 2014, 5, 2506–2509.

(6)

Zhang, C.; Boussac, A.; Rutherford, a W. Biochemistry 2004, 43 (43), 13787–13795.

(7)

Ghosh, M. C.; Reed, J. W.; Bose, R. N.; Gould, E. S. Inorg. Chem. 1994, 33, 73–78.

(8)

Wang, R.; Yang, W.; Yue, L.; Pan, W.; Zeng, H. Synlett 2012, 23 (11), 1643–1648.

(9)

Bose, D. S.; Idrees, M.; Srikanth, B. Synthesis (Stuttg). 2007, 0819–0823.

(10)

Kirk, M. L.; Shultz, D. a. Coord. Chem. Rev. 2013, 257 (1), 218–233.

(11)

Guo, D.; McCusker, J. K. Inorg. Chem. 2007, 46 (8), 3257–3274.

(12)

Mitsumi, M.; Nishitani, T.; Yamasaki, S.; Shimada, N.; Komatsu, Y.; Toriumi, K.;
Kitagawa, Y.; Okumura, M.; Miyazaki, Y.; GĂłrska, N.; Inaba, A.; Kanda, A.; Hanasaki,
N. J. Am. Chem. Soc. 2014, 136 (I), 7026–7037.

(13)

Wheeler, D. E.; McCusker, J. K. Inorg. Chem. 1998, 37 (9), 2296–2307.

(14)

Melamud, E.; Silver, B. J. Phys. Chem. 1974, 78 (21), 2140–2144.

(15)

Pushkar, Y. N.; Ayzatulin, O.; Stehlik, D. Appl. Magn. Reson. 2005, 28, 195–211.

(16)

Stevenson, G. R.; Wang, Z. Y.; Reiter, R. C. Jounal Org. Chem. 1990, 55 (760 mm), 78–
81.

(17)

Melamud, E.; Silver, B. J. Phys. Chem. 1974, 78 (21), 2140–2144.

(18)

Broze, M. J. Chem. Phys. 1967, 46 (12), 4891.

129

(19)

Grimaldi, S.; Ostermann, T.; Weiden, N.; Mogi, T.; Miyoshi, H.; Ludwig, B.; Michel, H.;
Prisner, T. F.; MacMillan, F. Biochemistry 2003, 42 (19), 5632–5639.

(20)

Samoilova, R. I.; Taguchi, A. T.; O’Malley, P. J.; Dikanov, S. a.; Lugtenburg, J. Appl.
Magn. Reson. 2014, 45 (9), 941–953.

(21)

Lin, M. T.; Shubin, A. a.; Samoilova, R. I.; Narasimhulu, K. V.; Baldansuren, A.; Gennis,
R. B.; Dikanov, S. a. J. Biol. Chem. 2011, 286 (12), 10105–10114.

(22)

Rodriguez, J. H.; Wheeler, D. E.; McCusker, J. K. J. Am. Chem. Soc. 1998, 120 (46),
12051–12068.

(23)

Wheeler, D. E.; Rodriguez, J. H.; McCusker, J. K. J. Phys. Chem. A 1999, 103 (20), 4101–
4112.

(24)

Bencini, A.; Carbonera, C.; Totti, F. Chemistry 2004, 10 (6), 1472–1480.

(25)

Chaudhary, A.; Patra, R.; Rath, S. P. Eur. J. Inorg. Chem. 2010, 2010 (33), 5211–5221.

(26)

Iuliano, A.; Piccioli, P.; Fabbri, D. Org. Lett. 2004, 6 (21), 3711–3714.

(27)

Bacon, R.; Lindsay, W. J. Chem. Soc. 1958, 1375–1381.

(28)

Alarcon, S. H.; Olivieri, A. C.; Labadie, G. R.; Cravero, R. M.; Gonz, M. Tetrahedron
1995, 51 (16), 4619–4626.

(29)

Gmbh, C. W. V.; Kgaa, C.; Kalkeren, H. A. Van; Leenders, S. H. A. M.; Hommersom, C.
R. A.; Rutjes, F. P. J. T.; Delft, F. L. Van. Chem. - A Eur. J. 2011, 17 (40), 11290–11295.

(30)

Stoll, S.; Schweiger, A. J. Magn. Reson. 2006, 178 (1), 42–55.

(31)

Stoll, S.; Schweiger, A. 2006.

(32)

Stoll, S.; Britt, R. D. Phys. Chem. Chem. Phys. 2009, 11 (31), 6614.

(33)

Fauth, J. J. Magn. Reson. 1986, 66 (1), 74–85.

(34)

Warncke, K.; McCracken, J. J. Chem. Phys. 1994, 101 (3), 1832.

(35)

Gemperle, C.; Aebli, G.; Schweiger, A.; Ernst, R. R. J. Magn. Reson. 1990, 88, 241–256.

(36)

https://www.mathworks.com/help/matlab/ref/fminsearch.html?s_tid=gn_loc_drop
https://www.mathworks.com/help/matlab/ref/fminsearch.html?s_tid=gn_loc_drop.

(37)

Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J.
R.; Montgomery, Jr., J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.;
130

Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.;
Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa,
J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J.
E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.;
Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.;
Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V.
G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;
Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.;
Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin,
R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.;
Pople, J. A. Gaussian, Inc.: Wallingford, CT 2004.
(38)

Denningham II, R. D.; Keith, T. A.; Millan, J. Semichem, Inc.: Shawnee Mission, KS
2003.

(39)

Becke, A. D. J. Chem. Phys. 1993, 98 (7), 5648.

(40)

Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37 (2), 785–789.

(41)

Stephens, P.; Devlin, F. J. … 1994, 98 (45), 11623–11627.

(42)

Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72 (1), 650–
654.

(43)

McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72 (10), 5639.

(44)

Wachters, A. J. H. J. Chem. Phys. 1970, 52 (3), 1033.

(45)

Hay, P. J. J. Chem. Phys. 1977, 66 (10), 4377.

(46)

Raghavachari, K.; Trucks, G. W. J. Chem. Phys. 1989, 91 (2), 1062.

(47)

Reed, A. E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys. F. Full J. TitleJournal Chem.
Phys. 1985, 83 (2), 735–746.

(48)

J. Lugtenburg, R. Esmeijer, W. F. Steggerada, W. B. S. van L. Recl. des Trav. Chim. des
Pays-Bas 1994, 113, 153–161.

(49)

Bacon, R.; Lindsay, W. S. J. Chem. Soc. 1958, 1382–1389.

(50)

Matsuzaki, S. Y.; Gotoh, M.; Kuboyama, A. Mol. Cryst. Liq. Cryst. 1987, 142 (1–4), 127–
139.

(51)

Jones, M. J. Magn. Reson. 1973, 11, 207–216.

131

(52)

Adams, M.; Blois, M. S.; Sands, R. H. J. Chem. Phys. 1958, 28 (5), 774.

(53)

Smaller, B.; Matheson, M. S. J. Chem. Phys. 1958, 28 (6), 1169–1178.

(54)

Fessenden, R. W.; Schuler, R. H. J. Chem. Phys. 1963, 39 (9), 2147–2195.

(55)

Kazuhiro, B. Bull. Chem. Soc. Jpn. 1964, 37 (1963), 553–558.

(56)

Blackhurst, A. J. J. Chem. Soc. Faraday Trans. 2 Mol. Chem. Phys. 1971, 68, 470–475.

(57)

Kazuhiro, B. Bull. Chem. Soc. Jpn. 1964, 37, 553–558.

(58)

Pasimeni, L.; Brustolon, M.; Corvaja, C. Chem. Phys. Lett. 1975, 30 (2), 249–252.

(59)

Coope, J. A. R.; Dalal, N. S.; McDowell, C. A.; Srinivasan, R. Mol. Phys. 1972, 24 (2),
403–415.

(60)

Cook, R. J.; Whiffen, D. H. Proc. Phys. Soc. 1964, 84 (6), 845–848.

(61)

Nguyen, M. J. Phys. Chem. A 1997, 5639 (96), 3174–3181.

(62)

McCracken, J. L. Handbook of Electron Spin Resonance; Springer-Verlag New York,
Inc., 1999; Vol. 2, pp 69–83.

(63)

Jeschke, G.; Schweiger, A. Principles of Pulse Electron Paramagnetic Resonance; Oxford
University Press: New York, 2001.

132

Chapter 3. The Influence of Lewis Acidity on Spin Density Delocalization

3.1 Introduction

In the previous chapter, it was shown that the wave function of the unpaired spin density can
be, and is, influenced by a variety of factors. DFT results predicted an overall decrease in
average carbonyl carbon unpaired spin density when examining a family of 3,6-R2-PSQ
molecules moving from electron-donating groups to electron-withdrawing groups. This showed
a spin polarization effect based on substituent choice. When examining NaPSQ, Na-[13C2]PSQ,
Na(3,6-(OMe)2-PSQ) and Na([13C2]3,6-(OMe)2-PSQ) by EPR and ESEEM, however, the trend
predicted by DFT was also true for unpaired spin density at the nucleus (isotropic coupling), but
not unpaired spin density in p-orbitals (dipolar coupling).
(OMe)2-PSQ than PSQ, as well as

13

1

H-aiso values were smaller for 3,6-

C-aiso values. ESEEM results contrasted with these as

dipolar values were found with the opposite trend. For both dimer and ion-pair molecules,
[13C2]3,6-(OMe)2-PSQ was found to have larger T values than [13C2]PSQ. In this chapter Na+ is
replaced with a Zn(tren)2+ unit and examined is the effect(s) Lewis acidity has on unpaired spin
density distribution.
The

purpose

of

this

chapter

is

to

explore

spin

density

delocalization

in

phenanthrenesemiquinone when covalently bound to a d10 metal, specifically ZnII. ZnII was
chosen to mimic the ionic radius of NiII, due to the interest in Ni-SQ complexes. NiII has an
ionic radius of 83 pm (octahedral, high spin,) and ZnII has an ionic radius of 88 pm1, making zinc
an appropriate d10 model for NiII complexes. The importance of having a d10 model is to study
what effect an MII ion has on spin density in a bound, organic radical ligand, in the absence of
exchange coupling. Since our ultimate goal is to understand the effects of exchange coupling on
133

spin density delocalization/polarization, we wish to first know what would happen when the
semiquinone is bound to a non-paramagnetic metal, in a similar chemical environment to
[Ni(tren)(PSQ)](BPh4).

In addition, the hyperfine coupling constants of NaPSQ will be

compared to those of Zn-PSQ, to assess the effect of an MII ion on spin density delocalization.
The system of study is seen below in Figure 1 and is molecularly analogous to the
aforementioned NiII complex. The system contains tris(2-aminoethyl)amine (tren), an ancillary
ligand used to occupy coordination sites on zinc, and a single PSQ moiety. This system has been
previously studied2, albeit with a tetraazzmacrocyle in the place of tren, but only as a
spectroscopic baseline for its NiII analogue. Steady-state electronic absorption features were
compared and assigned based on the ZnII compound and its absence of any d-d transitions in the
visible region of the spectrum. No synthetic or physical characterization was provided by the
study. Other studies that targeted Zn II-SQ species concentrated on bis-semiquinone systems that
exhibited triplet ground states3–5. These studies were focused on super exchange interactions
between the ligands as well as interligand electron transfer and valence tautomerism.

Figure 3-1. Molecular structure of [Zn(tren)(3,6-R2-PSQ)]+.
Additionally, GaIII was chosen to assess spin density at the metal center. Zn has an I = 0
nucleus and thus no hyperfine would be seen from EPR experiments, whereas Ga has two I = 3/2
134

nuclei, 69Ga and 71Ga, and therefore the hyperfine coupling can be measured. These compounds
will be essential to the analysis of spin density delocalization as it provides hyperfine
information at the metal center, and hyperfine information at the carbonyl carbons for a M III ion.
Although there is no hyperfine coupling available for M II ions in this study, it will still be
advantageous to compare MI (Na) and MIII ions hyperfine interactions, both at the metal and the
carbonyl carbons. This will give a much fuller picture of the interplay between ligand and metal,
when an unpaired electron is involved.
Of course this is not the first study to employ GaIII as a diamagnetic stand-in for a
paramagnetic metal. Fehir6 has studied [Ga(tren)(3,6-NO2-PSQ)]2+ as well as [Ga(tren)(2,7NO2-PSQ)]2+ complexes. This study was conducted to uncover spin density delocalization as a
mechanism of substituent placement. It was found that the 2,7 isomer shifted the spin density
onto the N atoms of the NO2 groups to a greater extent than in the 3,6 isomer. There is, however,
a slight possibility that some of the

14

N hyperfine interaction could be a result of N ligating

atoms from the tren capping ligand. Hyperfine coupling at the Gallium nucleus was said to be
consistent with a GaIII2 iminosemiquinone complex synthesized and studied by Wieghardt and
coworkers.7
McCusker and coworkers have studied [Ga2(tren)2(CA)]3+ complexes in which two Ga(tren)3+
units are binding to a tetraoxolene (CA) ligand where one side possesses semiquinone character,
and the other side, catecholate, giving an S = ½ system. This study found the two gallium nuclei
had comparable hyperfine parameters and yet smaller than the hyperfine coupling constants of
the carbonyl carbons of the free CA ligand. This system also served as a template for studying
analogous

CrIII complexes8.

DFT

and

CW-EPR

have

been

carried

out

on

a

[Ga(tren)(PSQ)](BPh4) complex and its’ isotopologue, [Ga(tren)([9,10-13C2]PSQ)](BPh4), in an

135

attempt to measure spin density at the gallium nucleus; the ability to explore hyperfine
interactions at the ligating metal are made possible by the I = 3/2 gallium nucleus. Attempts
were made to isolate and study a CdII-PSQ complex to elucidate hyperfine coupling at a M II
center, and compare to ZnII and NiII; however, the complex was unable to be isolated. In lieu of
obtaining said CdII complex, the information obtained from the GaIII complex aided in drawing
conclusions regarding spin density delocalization across the semiquinone ligand to the metal
center.
A density functional theory (DFT) has been conducted in addition to the EPR study to examine
the possibilities of a substituent effect. ZnII- and GaIII-3,6-R2-PSQ complexes were examined,
and only in the case of GaIII complexes was a trend found comparing average oxygen spin
density and average carbon spin density. In both cases, however, the magnitude of spin density
on carbonyl carbon atoms was greater than on oxygen atoms.
Continuous Wave EPR (cw-EPR) experiments of [Zn(tren)(PSQ)](BPh4) and its
isotopologue as well as [Ga(tren)(PSQ)](BPh4)2 and its

13

13

C

C isotopologue were conducted. For

[Zn(tren)(PSQ)](BPh4) and [Zn(tren)(13C2PSQ)](BPh4) the treatment was similar to that of the
free ligands; room temperature CW spectra were taken at varying field modulation values to
uncover proton aiso values, ESEEM experiments to uncover axial (anisotropic) and rhombic
contributions to the hyperfine interaction. Results of HYSCORE experiments are also presented
to aid in the interpretation of the ESEEM data.

136

3.2 Experimental

General Methods. Reagents were purchased from commercially available sources and used
without further purification unless otherwise stated.

Stabilizer-free tetrahydrofuran was

purchased from Fisher Scientific and either distilled over sodium metal and benzophenone and
freeze, pump, thawed to remove O2, or taken from a dry still with a solvent bomb. Elemental
analyses and electro-spray ionization ESI mass spectra were obtained through the analytical
facilities at Michigan State University. GC mass spectra were obtained on a Hewlett Packard
G1800D GCD system equipped with an electron ionization detector.

NMR spectra were

collected on Agilent DDR2 500 MHz Spectrometers equipped with 7600A autosamplers at the
Max T. Rogers NMR Facility at Michigan State University.
Synthesis. [Zn(tren)(OH2)2](ClO4)2 (1). The synthesis of this compound is based on an
unpublished procedure. 2.55 g (6.8 mmol) Zn(ClO4)2∙6H2O was added to a stirring ethanolic
solution of 1 g (6.8 mmol) of tren and immediately a colorless precipitate crashed out of solution.
The reaction was stirred for an additional hour to ensure complete complexation. The colorless
solid was then filtered, washed with EtOH and dried on an aspirator. Yield: (>100%). MS [GC
m/z (rel. int.)]: C6H21N4O2Zn+ 245.07 (100%), ClO4- 98.95 (100%).
[Zn(tren)(PSQ)](BPh4) (2). The synthesis of this compound is based on an unpublished
procedure. In a dry box, a THF solution of 297 mg (1.41 mmol) 9,10-phenanthrenequinone was
made and an excess of sodium metal (5x5x5 mm3 chunks) were added and the reaction was
stirred overnight. The clear/deep red catecholate solution was filtered to remove the unreacted
sodium. In a separate flask, 564 mg (1.26 mmol) [Zn(tren)(OH2)](ClO4)2 was suspended in
MeOH and the catecholate solution was added to this, drop-wise. As the catecholate solution

137

was added, the reaction mixture changed from colorless/cloudy to yellow/clear. This was stirred
overnight and the following day 483 mg (1.41 mmol) [Fc](PF6) in MeOH was added to produce
a dark green reaction mixture. The mixture was then filtered to remove the ferrocene. A 20-fold
excess of Na(BPh4) (8.6 g, 25.2 mmol) was added to the filtrate and the reaction mixture was
then transferred to a round bottom flask equipped with a gas flow adapter and pumped out of the
dry box. The solution was then dried under a stream of N2, and subsequently dried under
vacuum. The resultant green powder was then taken up in dry, degassed CH2Cl2 (dried via dry
still), by way of cannula transfer. In a secondary flask, hexanes was sparged with N2 for 15-30
minutes and then cannula transferred to the CH2Cl2 flask to crash out a pale green solid. The
solid was then filtered on a glass frit, washed with Et2O and dried on an aspirator for a minimum
amount of time and then pumped into a dry box. Yield: (18%). MS [ESI m/z (rel. int.)]:
C21H31N4O3Zn+ (Complex ∙ MeOH) 451.12 (10%).

Analytical composition calculated for

C44H46BN4O2Zn∙0.5 CH2Cl2∙0.05 CH3OH: C, 68.33; H, 6.07; N, 7.15; Found: C, 68.52; H, 6.00;
N, 6.94.
[Zn(tren)(13C2-PSQ)](BPh4) (3). This complex was synthesized analogously to complex 2.
In a dry box, a THF solution of 56 mg (0.266 mmol) 9,10-phenanthrenequinone was made and
an excess of sodium metal (5x5x5 mm3 chunks) were added and the reaction was stirred
overnight. The clear/deep red catecholate solution was filtered to remove the unreacted sodium.
In a separate flask, 106 mg (0.238 mmol) [Zn(tren)(OH2)](ClO4)2 was suspended in MeOH and
the catecholate solution was added to this, drop-wise. As the catecholate solution was added the
reaction mixture changed from cloudy/colorless to yellow/clear. This was then stirred overnight
and the following day 88 mg (0.266 mmol) [Fc](PF6) in MeOH was added to produce a dark
green reaction mixture. The mixture was then filtered to remove the ferrocene. A 20-fold excess

138

of Na(BPh4) (1.82 g, 5.32 mmol) was added to the filtrate and the reaction mixture was then
transferred to a round bottom flask equipped with a gas flow adapter and pumped out of the dry
box. The solution was then dried under a stream of N2, and subsequently dried under vacuum.
The resultant green powder was then taken up in dry, degassed CH2Cl2 (dried via drystill), by
way of cannula transfer. In a secondary flask, hexanes was sparged with N2 for 15-30 minutes
and then cannula transferred to the CH2Cl2 flask to crash out a pale green solid. The solid was
then filtered on a glass frit, washed with Et2O and dried in vacuo for a minimum amount of time
and then pumped into a dry box. Yield: (47%). MS [ESI m/z (rel. int.)]: C1913C2H31N4O3Zn+
(Complex

∙

MeOH)

453.18

(50%).

Analytical

composition

calculated

for

C2C43H46BN4O2Zn∙H2O∙CH3OH: C, 68.57; H, 6.63; N, 7.08; Found: C, 68.64; H, 6.43; N, 7.35.

13

[Fe(C5H5)2](BPh4) (4). The synthesis of this complex was based on methods developed in
the literature9. A suspension was made of 1.0 g (5.4 mmol) of ferrocene in 40 mL of Acetone
and 40 mL H2O was made. This suspension was placed under N2 and 1.22 g (7.6 mmol)
anhydrous FeCl3 was added and the orange suspension changed to deep blue. This was stirred
for 20 minutes and 2.6 g (7.6 mmol) of Na(BPh4) was added, producing a dark blue precipitate.
THF was added (ca. 200 mL) until all solids dissolved and this produced a light blue/milky white
solution. The solution was filtered and the filtrate concentrated under a stream of N2. A dark
blue, fine crystalline powder resulted from concentrating the solvent. This solid was filtered,
dried in vacuo and immediately stored in the glove box. The compound naturally decomposes
(even under N2) after several days. 1.2 grams of the dark blue, crystalline powder was recovered.
Yield: (44%). MS [ESI (Hi-Res) m/z (rel. int.)]: C10H10Fe+ 186.0 (100%), C24H20B- 319.2
(100%).

139

[Ga(tren)(PSQ)](BPh4) (5).

The synthesis of this compound is based on methods

previously developed10. In a glove box, 99.9 mg (0.48 mmol) of PhenQ was taken up in THF,
and 5x5x5 mm3 Na cubes were added to the stirring solution. After stirring overnight, a blood
red solution persisted and was filtered. In a separate flask, 190 mg (0.43mmol) of GaI3 was
dissolved in 5 mL of MeOH and 0.06 mL (0.43 mmol) tren was added to the solution to produce
a cloudy/yellow reaction mixture.

The filtered PCat2- solution was then added to the

clear/colorless [Ga(tren)I2]I solution to produce a yellow/clear reaction mixture and was stirred
for 2 hours. The reaction mixture was then filtered and a concentrated MeOH solution of a 10fold excess of NaBPh4 was added. The solvent volume was reduced by vacuum and the flask
was covered and placed in the refrigerator (glove box). After several days a yellow crystalline
precipitate had fallen out of solution to yield 200 mg (54 % yield) of Ga-PCat. The Ga-PCat
complex was then dissolved in MeCN and 136 mg (0.269 mmol) of [Fc](BPh4) was added to the
stirring solution and immediately the color changed from bright yellow to brown-gold. This was
stirred for 1 hour, after which the solution was filtered and an excess of NaBPh4 was added and
the flask was swirled until all solids dissolved. 300 mL of degassed Et2O was added and the
solution became cloudy. The flask was covered and placed in the fridge. A brown-gold
crystalline solid had crashed out of solution after several days; this was filtered and washed with
Et2O. MS [ESI (Hi-Res) m/z (rel. int.)]: C20H26GaN4O2+ 423.13 (100%), C24H20B- 319.2 (100%).
[Ga(tren)([9,10-13C2]PSQ)](BPh4) (6). This compound was prepared in an analogous
manner to 4. MS [ESI (Hi-Res) m/z (rel. int.)]: C1813C2H26GaN4O2+ 425.13 (32%), C24H20B319.2 (100%).
Physical Measurements. EPR Spectroscopy. All EPR data were collected on a Bruker
Elexsys E-680X spectrometer, operating at X-band frequency (9.42-9.71 GHz).

140

Low

temperature experiments were performed between 4-30 K using an Oxford Instruments liquid
helium flow system equipped with a CF-935 cryostat and an ITC-503 temperature controller.
Samples were prepared in a dry box in modified Wilmad Glass SQ-706 EPR tubes with 3.8 mm
diameter. Prior to sample preparation, tubes were modified in the glass blowing lab at Michigan
State University to contain a grated seal and vacuum adapter port. A vacuum was pulled on the
sample tubes and then the vacuum ports were blown off at the grated seal, by hydrogen torch.
Prior to collecting data, power-dependence experiments were conducted to ensure non-saturation
of the detector. Continuous wave spectra were simulated using the garlic module of Easy Spin
4.511,12,13.
All pulsed experiments were conducted at X-band microwave frequency. ESEEM spectra
were obtained using a three pulse sequence of 90°-τ-90°-T-90° with 90° pulse lengths of 16 ns.
Tau values were chosen to suppress hyperfine contributions from weakly coupled protons. Echo
amplitudes were integrated at FWHM and the T values were scanned from 40 ns to 6 Îźs with a
12 ns time increment. A four step phase-cycling procedure was used to eliminate unwanted 2pulse echoes.14 Raw data were processed by phase shifting the spectra to remove the dispersion
component from the time domain data. The in-phase portion of the data were normalized by
dividing by a biexponential background decay function. These normalized data were then
further processed by subtracting a second degree polynomial (DC-component of the signal),
followed by application of a Hamming window. The data set was then zero-filled from 512
points to 1024 points and Fourier transformed. Frequency spectra are displayed as an absolute
value of the transformed time domain data.

13

spectrum at each magnetic field value.

After normalizing each time domain spectrum

C and

141

12

C spectra are presented as one divided

individually, the 13C spectrum was divided by the 12C spectrum and the same work up procedure
described above was applied15.
HYSCORE experiments were carried in a similar manner to the ESEEM experiments but
with a four-pulse sequence of 90°-τ-90°-t1-180°-t2-90° with a 180° pulse length of 16 ns. Echo
amplitudes were measured at FWHM and τ values were chosen as stated above. T1 and t2 values
were scanned from 40 ns to 2 Îźs with 16 ns time increments. Data sets were 128 points in both
dimensions. A 4-step phase cycling program was used to filter out unwanted 2-pulse echoes.16
The real, or in-phase portion of the HYSCORE data were processed, in both dimensions, by first
subtracting a 2° polynomial, then tapering the data with a Hamming window and zero-filling to
256 points. A two-dimensional fast Fourier transform was applied and absolute values taken.
Data are presented as a two-dimensional contour plot. Each contour plot is presented at a
threshold of 5-20% of the maximum peak amplitude (z-direction).
Spectra were fit using customized MATLAB scripts where EasySpin was used to perform
spectral simulations and the fminsearch function in Matlab was used to optimize the calculation
parameters. Fminsearch is a function that finds the minimum of an unconstrained multivariable
function using a derivative-free method17. Residuals were calculated by squaring the difference
of the experimental and simulated amplitudes. Goodness of fit for each spectral fit was assessed
by calculating a χ2 value with the following equation:

𝜒𝑛2 =

𝑒𝑥𝑝
𝑁 (𝑦𝑖
𝛴𝑖=1

− 𝑦𝑖𝑐𝑎𝑙𝑐 )
𝜎2
𝑁−𝐿

142

2

(3.1)

where the sum is taken over N points, yiexp and yicalc are the experimental and calculated y-axis
values, σ is the standard deviation fixed at 2% of the maximum spectral amplitude and L is the
number of adjustable variables.
Calculations. Calculations were performed with the Gaussian0318 computational chemistry
package on the supercomputing cluster HPCC (High Performance Computing Center) at
Michigan State University. Structures were drawn in GaussView4 and were optimized on
phenanthrenesemiquinones (S=1/2) using the unrestricted formalism with the B3LYP19–21
functional. Molecular geometries were optimized with the 6-31g(d,p)22–32 basis set and single
point calculations were carried out with the 6-311g(d,p)33–36 basis set. Frequencies calculations
were carried out to ensure optimizations reached global minima. Single point calculations were
performed at the same level of theory as geometry optimizations. In addition the “pop=NPA”
keyword command was used to calculate spin densities by natural population analyses (NPA)

3.3 Results and Discussion

Density Functional Theory. DFT analysis was carried out on the series of [Zn(tren)(3,6-R2PSQ)]+ complexes, where R = NH2, OMe, H, Br, CF3, CN and NO2. Initial structures were
based on the crystal structure of [Ni(tren)(PSQ)](BPh4),10 as well as the [Ni(tren)(3,6-R2-PSQ)]+
complexes studied within Reference 10. The geometries were successfully optimized with the
unrestricted form of the B3LYP functional and the 6-31g(d,p) basis set. Frequencies calculations
were undertaken to ensure that a global minimum had been reached and no imaginary
frequencies were found. Single point calculations and population analyses were then carried out
with the same functional and the 6-311g(d,p) basis set. Spin density values were calculated by

143

subtracting the beta spin density from the alpha spin density for each atom and then averaging
each value.
Examining unpaired spin density delocalization for complexes of this type is unprecedented in
the literature.

Most Zn-SQ complexes studied are bis-o-semiquinone complexes and are

typically ground state triplets. Not unlike in this investigation, they are often studied to escape
the complicated electronic structure issues present with paramagnetic metals; they provide a
means to study the ligand in the same, if not very similar, geometric arrangement. Although
many Zn-SQ complexes are biradicals, [Zn(NH3)4(SQ)]+ has been studied by DFT, by Bencini
and coworkers37, and it does resemble the structure of the Zn-PSQ complex being studied in this
chapter. The tetradentate ligand, tren, has been replaced with four ammonia ligands and the
semiquinone ligand is o-benzosemiquinone. The ammonia ligands allow for a geometry closer
to that of an octahedral complex, being free ligands and not connected to a central nitrogen atom
connected by ethylene bridges, which can constrict the geometry from being completely
octahedral. The other difference of this complex being benzosemiquinone, which does not have
the same spin delocalization capabilities of phenanthrenesemiquinone. These differences are
noted; however, they do not pose large differences with respect to the delocalization of the
unpaired spin density. Bencini and coworkers found that less than 3% of the unpaired spin
density resides on Zinc and the NH3 ligands, while most of the unpaired spin is located on the
SQ ligand. The NiII analogue was also studied and 49% of the unpaired spin is on Ni, 39% on
the semiquinone ligand and 12% on the NH3 ligands. These results are consistent with the
findings of this study (as well as the following chapter); however, this comes with a caveat. The
Natural Population Analysis did not calculate any unpaired spin density on zinc, but the Fermi
analysis found an isotropic hyperfine coupling constant of -6.22 MHz, slightly greater than that

144

of Na (-7.06 MHz). So it should be noted here that the NPA does not first calculate hyperfine
coupling constants and then covert to spin density [in a similar routine that was carried out in
Chapter 2 for the isotropic hyperfine coupling constants].
Figure 2 shows the graph of average carbonyl oxygen spin density versus average carbonyl
carbon spin density. No linear correlation or functional dependence between the two quantities
was found. It is noticeable that the average oxygen spin density has experienced a slight
decrease and the carbonyl carbon spin density has increased by a factor of approximately two,
relative to PSQ. However, the fact that the two quantities do not correlate indicate that the
unpaired spin density found on each atom may be influenced more by its next nearest neighbors;
i.e. unpaired spin density on oxygen may be more influenced by Zn and tren, and likewise,
unpaired spin density on carbon may be more influenced by the phenanthrene ring system.
The average oxygen spin density does increase with increasing electronegativity. This is
consistent with results from Fehir’s calculations on [Ni(tren)(3,6-R2-PSQ)]+ complexes; indeed,
even the ordering of increasing spin density on oxygen is consistent between the two studies.
Compare this with the DFT results of Chapter 2 where there is no trend. The only derivative that
attracts attentions is -NO2, and it has the least amount of spin density at both the carbon and the
oxygen, relative to the other derivatives.

145

0.2
NO2

Average O Spin Density

0.19

CF3

CN

H

Br
0.18

0.17
OMe
0.16

0.15
0.184

NH2
0.186

0.188

0.19
0.192 0.194 0.196
Average C Spin Density

0.198

0.2

0.202

Figure 3-2. Average Carbonyl carbon spin density versus average oxygen for [Zn(tren)(3,6-R2PSQ)]+ Complexes.
Table 3-1. Average carbonyl carbon and oxygen spin density in Zn-PSQ and PSQ systems as
calculated from NPA analysis.
Substituent
Br
CF3
H
NH2
NO2
OMe
CN

PSQ
Avg. C(ρ)
0.073
0.064
0.082
0.0877
0.042
0.081
0.055

Zn-PSQ
Avg. C(ρ)
0.194
0.193
0.2
0.192
0.185
0.199
0.186

PSQ
Avg. O(ρ)
0.25
0.248
0.254
0.247
0.205
0.249
0.239

Zn-PSQ
Avg. O(ρ)
0.183
0.193
0.185
0.152
0.196
0.163
0.192

Relative to PSQ, for R = Br, CN and CF3 oxygen spin density drops by ~20 %, while NO2
remains essentially unaffected, and H, NH2 and OMe drop by ~40 %. There does not seem to be

146

any correlation between change in oxygen spin density and ability to donate/accept π density or
donate/accept σ density. There is a correlation, however, between electron withdrawing and
electron donating groups. The electron donating substituents lost more oxygen spin density than
their electron withdrawing counter parts.
Despite the drastic change in oxygen spin density, the carbonyl carbons experienced an even
greater increase in spin density. A 50-60% increase was observed for each of the substituents
upon complexation. The loss of the spin density at the oxygen does not equal the gain in spin
density at the carbon, however; and in fact, the derivative that gained the greatest amount of
unpaired spin density is –NO2. Compared to all the other derivatives, -NO2 barely lost any
oxygen spin density. Although no correlation was observed between carbon and oxygen spin
density, it was discovered that the carbonyl carbon spin density increases and the oxygen spin
density decreases when bound to MII ion with a full d shell. If the charge density at the metal
was then increased to a MIII ion, what effect would that have on the spin density in the carbonyl
groups?
To answer this inquiry, a similar set of calculations was undertaken for Ga III complexes of the
same type of molecular structure as the ZnII complexes. Displayed below in Figure 3 is a plot of
carbon versus oxygen spin density in [Ga(tren)(3,6-R2-PSQ)]2+ complexes where R = NH2,
OMe, H, Br, CF3, CN and NO2. This is the first instance where a somewhat modest correlation
coefficient has been achieved. The magnitude of unpaired spin density at the oxygen atoms has
decreased dramatically while carbon unpaired spin density has remained in approximately the
same range, with respect to the Zn-PSQ PSQ systems. Table 2 compares the carbonyl spin
density between the free radical, Zn-PSQ and Ga-PSQ compounds. These results illustrate that

147

Lewis acidity and/or metal complexation potentially has an effect on spin density delocalization
in PSQ type systems.
It is observed that in moving from the free radical to the Zn II complex, the spin density at the
oxygen decreases by approximately 20-40%, then another 40-56% decrease in the GaIII
complexes. It has been seen in [Ni(tren)(3,6-R2-PSQ)]+ and [Cr(tren)(3,6-R2-PSQ)]2+ complexes
that a drastic decrease in average oxygen spin density occurs when comparing average oxygen
spin density.
At the carbonyl carbons, however, the effect is not as straight forward. It was previously
discussed that the ZnII polarizes more spin density onto the carbons, but only for R = H and
electron withdrawing derivatives does the carbon spin density decrease in moving from M II to
MIII. On the other hand, for electron donating substituents, the carbonyl spin density decreases.

148

0.13

NO2

0.12

CF3

Average Carbonyl O (ρ)

CN
H

0.11
Br

0.1

R² = 0.9318

0.09
0.08

OMe

0.07

0.06
0.15

NH2

0.16

0.17

0.18
0.19
Average Carbonyl C (ρ)

0.2

0.21

0.22

Figure 3-3. Average Carbonyl carbon spin density versus average oxygen for [Ga(tren)(3,6-R2PSQ)]2+ Complexes.
Table 3-2. Carbon and Oxygen spin density in 3,6-R2-PSQ radical anions, [Zn(tren)(PSQ)]+ and
[Ga(tren)(PSQ)]2+ complexes.
Substituent
Br
CF3
H
NH2
NO2
OMe
CN

PSQ
C (ρ)
0.073
0.064
0.082
0.0877
0.042
0.081
0.055

Zn-PSQ
C (ρ)
0.194
0.193
0.2
0.192
0.185
0.199
0.186

Ga-PSQ
C (ρ)
0.197
0.216
0.214
0.161
0.213
0.18
0.203

PSQ
O (ρ)
0.25
0.248
0.254
0.247
0.205
0.249
0.239

Zn-PSQ
O (ρ)
0.183
0.193
0.185
0.152
0.196
0.163
0.192

Ga-PSQ
O (ρ)
0.102
0.125
0.111
0.066
0.127
0.08
0.115

It is quite apparent from the above results that Lewis acidity perturbs the spin density in the
carbonyl groups. It is not apparent, however, as to why the oxygen spin density decreases by
149

such a magnitude, or why the carbon spin density increases to such a magnitude, in moving from
free radical to MII, to MIII. Following the synthesis section, the carbonyl carbon and proton spin
density values will be assessed by EPR.
Synthesis. The d10-semiquinone complexes were synthesized based on procedures previously
published. Figure 4 shows the several oxidations states of phenanthrenequinone (or any

Figure 3-4. Oxidation states of phenanthrenequinone.
quinone). Typically, quinones and semiquinones are not sought after as nucleophiles. Although
transition metal-quinone complexes have been prepared38,39, quinones have very low binding
affinities, most likely due to the low nucleophilicity of the carbonyl oxygen atoms; their one
electron reduced forms, semiquinones, also suffer from low binding affinities. The low charge
density at the oxygen atoms (one negative charge delocalized across several atoms) is
responsible for the low binding affinity. Additionally, the radical character of the ligand can
lend itself to side reactions and formation of oligomers.

The two-electron reduced form,

catecholates, have high negative charge density on the oxygen atoms and are strong
nucleophiles. Therefore, by exploiting the redox nature of the quinone, one can easily use the
nucleophilicity of the catecholate to bind to a metal and subsequently use chemical oxidants to
produce metal-semiquinone species.

The major stipulation in these reactions is that the
150

semiquinone and catecholate forms are, generally, extremely air and moisture sensitive,
especially in solution. Therefore, these reactions are carried out in dry boxes and on Schlenk
lines.
For both ZnII and GaIII products, phenanthrenequinone was first taken up in dry THF and
reduced with an excess of sodium metal. The filtered catecholate solution was then added to
stirring solutions of metal-tren precursor complexes, either in THF or MeOH and produced
yellow-amber/clear reaction mixtures. Zn-PCat complexes were made with Zn(tren)(ClO4)2,
which was prepared by adding Zn(ClO4)2∙6 H2O to a stirring, EtOH solution of tren. After which
a colorless precipitate crashes out of solution. Ga-PCat complexes, however, were synthesized
by adding tren to a solution of GaI3, producing [Ga(tren)I2]I in solution.
Following complexation of PCat2- to ZnII, the solution was oxidized with a ferrocenium salt to
produce a dark, forest green-colored solution, reminiscent of the NaPSQ ion pair. This was
filtered (to remove any ferrocene that had precipitated out of solution) and the complex was
metathesized with an excess of Na(BPh4) to produce the BPh4 salt. The identity of the complex
was confirmed by ESI mass spectrometry results, as well as elemental analysis. Both the 12C and
13

C complexes were found with a methanol molecule attached to the complex.
The Ga-PCat complexes were also oxidized by a ferrocenium salt, although the resultant

semiquinone was a brown-gold-colored solution. This complex was also metathesized with
Na(BPh4) and finally recrystallized in a solution of MeCN, by adding Et2O and letting it set over
several days. The identity of the

12

C and 13C complexes were confirmed by Hi-Res Mass Spec

results.

151

CW-EPR.
In the previous chapter, CW-EPR was found to be a powerful technique; multiple

13

C-aiso

values were found for both NaPSQ and Na(3,6-(OMe)2-PSQ) compounds. It was shown that
NaPSQ had larger aiso values, in both the ion-pair and dimer compounds; however, when
comparing dipolar values, the situation was reversed. Also, it was observed that the substituent
has an influence on the paired ion. For NaPSQ, 23Na-aiso = 1.32 MHz while for Na(3,6-(OMe)2PSQ) 23Na-aiso = 1.17 MHz. It should also be noted that the Fermi contact integrals for NaPSQ
and Na(3,6-(OMe)2-PSQ) were 7.06 MHz and 6.98 MHz, respectively. Although there is a large
difference in magnitude, both the observed and calculated aiso values conformed to the trend that
NaPSQ had a larger value.
Moving forward, those experimental results and calculations from Chapter 2 will be compared
to ZnII- and GaIII-tren type complexes. Although ZnII has an ionic radius that is similar to NiII,
utilizing zinc does pose a discontinuity to the study, though. Unlike Na and Ga, the hyperfine
coupling at Zn cannot be measured using a standard Zn source.

64

Zn makes of 48% of the

natural abundance of all Zn isotopes and has no nuclear spin. The only Zn isotope with nuclear
spin is

67

Zn with I = 5/2 and its natural abundance is 4.1%, making it difficult to measure

contributions to the hyperfine coupling at X-band microwave frequencies. There was an attempt
to synthesize a CdII analogue which has two isotopes with nuclear spin,

111

Cd and

113

Cd, each

with I = ½; however, the identity of this complex could not be confirmed. So, although there are
computational results from DFT calculations, there will be no hyperfine data for this complex.
A second caveat to this study deals with the Ga-PSQ complex. In addition to CW data,
ESEEM and HYSCORE analyses of the anisotropic contribution to the hyperfine coupling would
round out the study of delocalization of unpaired spin density; however, the physical observable

152

for these experiments is the formation of a spin echo. The ability to produce an echo lies with
the molecule in question, and in the case of this study, no echo was observed for either 5 or 6.
Efforts were made to find an echo by increasing the concentration of the samples and even
increasing the microwave power to the point of saturating the detector, but all of these efforts
were to no avail. It is also possible that this molecule has a spin lattice relaxation time, T2, which
is fast compared to the detector. If this is the case, then it would be possible to observe an echo
at higher operating frequencies. Pulsed Studies of Ga-SQ complexes at W-band, for example,
are known8.
Despite the inability to observe a spin echo, the CW spectra were obtained and hyperfine
coupling for both 69,71Ga and 13C were detected. The CW spectrum (black) and simulation (red)
of [Ga(tren)(PSQ)](BPh4)2 are shown below in Figure 5. Unlike the PSQ ion-pair compounds,
only hyperfine coupling to gallium is observed. The coupling is large enough that no coupling
from ring protons could be detected. Gallium has two isotopes with nuclear spin, 69Ga and 71Ga,
each with I = 3/2. EasySpin computes a single value that is a weighted average of the two
isotopes, based on the formula:
𝐴̅ = 𝐴( 69𝐺𝑎 )𝑤( 69𝐺𝑎 ) + 𝐴( 71𝐺𝑎)𝑤( 71𝐺𝑎)

(3.2)

Only the weighted average will be presented from here on.
The spectrum was fit with aiso = 19.73 MHz. This value is consistent with other Ga-SQ
complexes. For complexes of the form [Ga2(tren)(SQ)]3+, McCusker and coworkers have found
69,71

Ga-aiso = 10.03 MHz and 10.36 MHz, and this is for the case where a single electron is

delocalized across a six-membered ring onto two GaIII ions.8 A pentacoordinated Ga complex of
the

formula

[Ga(3,5-di-tert-butyl-o-iminosemiquinone)(3,5-di-tert-butyl-o-

iminocatecholate)(pyridyl)] exhibited

69,71

Ga-aiso = 40.71 MHz.40 This hyperfine coupling is

153

roughly twice the simulated value of 5, however the iminosemiquinone has a six-membered-ring
and thus less atoms to delocalize the unpaired spin density, relative to PSQ. Lastly, Fehir has
found

69,71

Ga-aiso = 17.37 MHz for [Ga(tren)(3,6-(NO2)2-PSQ)](BF4)2.41 This compound being

the closest, structurally, to the Ga-PSQ complex studied in this chapter.

Although more

compounds are not involved, it may be pertinent to note that GaIII has a smaller isotropic
hyperfine coupling in this compound, relative to 5, leading one to the conclusion that perhaps the
-NO2 groups are polarizing the unpaired spin density away from gallium.
To contrast the experimental observations with theoretical results, there are two points to
69,71

Ga-aiso = 19.73 MHz hyperfine

coupling constant with the calculated value of 20.79 MHz.

For a direct comparison, the

consider. The first deals with comparing the observed

observed constant gives the calculation validation and it seems rather accurate. If the observed
69,71

Ga-aiso = 17.37 MHz hyperfine coupling constant for [Ga(tren)(3,6-(NO2)2-PSQ)](BF4)2 were

compared with the calculated value of 22.25 MHz, then, if only the DFT results were considered,
one could draw the opposite conclusion to the one above. Here it seems that the -NO2 groups
would be polarizing unpaired spin density toward gallium, but this only takes into account DFT
results and the experimental results do not have enough merit to propose a trend from one
derivative to another.

If there were more experimental results to consider from varying

derivatives, then one could propose such a trend.
Secondly, the theoretical results are a contradiction. The fermi contact integral is 20.79 MHz;
however, the calculated spin density at gallium is zero. It should be considered here that in the
above DFT section, the results displayed are from that of the natural population analysis, and in
this section only the Fermi contact integrals are being considered. It was stated previously that
these two entities are calculated both separately, and with differing methods. The Fermi method

154

does predict a hyperfine coupling constant at the gallium nucleus and this agrees with the
observed results, as well as predicting unpaired spin density at gallium. One last caveat to the
Fermi contact integrals: only 69Ga-aiso values are calculated, 71Ga is not considered. Due to this,
the calculated Fermi contact integrals are inherently lower than the weighted average or
The magnetic moment of

71

Ga is larger than the magnetic moment of

69

71

Ga.

Ga, 2.56 versus 2.02,

respectively (where the magnetic moment is a ratio of Îź/ÎźN).

Figure 3-5. 10 K CW spectrum (black) and simulation (red) of 5, [Ga(tren)(PSQ)](BPh4)2.
Simulation parameters: g = 2.00535, lwpp = 0.71 mT, 0.28 mT, aiso values: 69,71Ga = 19.73 MHz.
Experimental parameters: 9.72 GHz ÎźW frequency, 0.63 mW power, 3460 G center field, 200 G
sweep width, 1 G modulation amplitude, 512 points. χ2 = 0.23.

155

Figure 3-6. 10 K CW spectrum (black) and simulation (red) of 6, [Ga(tren)([13C2]PSQ)](BPh4)2.
Simulation parameters: g = 2.00532, lwpp = 1.09 mT Lorentzian + 0.57 mT Gaussian, aiso
values: 69,71Ga = 19.73 MHz, 13C-aiso values: 27.77 MHz, 32.10 MHz. Experimental parameters:
9.72 GHz ÎźW frequency, 0.016 mW power, 3460 G center field, 200 G sweep width, 1 G
modulation amplitude, 512 points. χ2 = 0.22.
Moving on from hyperfine at the metal, hyperfine coupling at the carbonyl carbon atoms are
considered. Pictured above in Figure 6 is the room temperature spectrum CW spectrum of 6,
[Ga(tren)([13C2]PSQ)](BPh4)2 (black) and its simulation (red). The immediate observation from
the simulation of 6, is that the

13

C-aiso is large relative to any other compound in this study,

including the Zn-PSQ complex below. DFT predicts Fermi contact integrals of 7.29 MHz and
7.20 MHz for the two carbonyl carbons, which is high compared to what has been observed thus
far, but 27.77 MHz and 32.10 MHz seems excessive. Unfortunately, these results cannot be
directly compared to any other study in the literature, but for such strong hyperfine coupling at
the carbonyl carbon atoms, they can be compared to the trianion of chloranilic acid. Broze and
156

Luz42 report a

13

C-aiso = 22.9 MHz for the carbonyl carbon atoms. This tetraoxolene ligand

consists of four carbonyl groups total (two o-quinone groups on one six-membered ring), and so
for a coupling this large to be measured on a free ligand, one might expect the coupling decrease
upon complexation. This is yet to be measured, but, 13C-aiso does increase for PSQ when bonded
to Ga; this was both observed experimentally and predicted by DFT.

Figure 3-7. Room temperature CW spectrum (black) and simulation (red) of 2,
[Zn(tren)(PSQ)](BPh4) at 20 mG modulation amplitude. Simulation parameters: g = 2.0071; aiso
values: H3,6 = 5.90, H1,8 = 4.01, H2,7 = 1.96, H4,5 = 1.01; Peak to peak line width: 0.0136 mT
Gaussian 0.0107 mT Lorentzian broadening. Experimental parameters: 9.42 GHz microwave
frequency, 0.80 ÎźW microwave power, 3353 G centerfield, 50 G sweep width, 20 mG
modulation amplitude, 1024 points. χ2 = 0.45.

157

The experimental spectrum and simulation of 2, [Zn(tren)(PSQ)](BPh4), can be seen in Figure
7. The spectrum was fit with four nonequivalent sets of two protons each, line broadening
parameters with Gaussian and Lorentzian (Gaussian plus Lorentzian, or, Voigtian) components,
and an isotropic g value. A total of 81 lines is expected from four nonequivalent groups with 2
protons each; however, only a total of 31 lines were observed in the 20 mG spectrum.
Phenanthrenequinone has been previously studied as a free ion, reduced by zinc metal in alkaline
media. The hyperfine information provided by the analyses from that study merely provided two
hyperfine couplings, produced from two groups of four protons each.43 This arrangement is not
the case as there are four groups of chemically equivalent protons. Table 3 compares 1H-aiso
values obtained from the simulations and those predicted by the Fermi analysis in DFT. The
obtained values from the simulations prove to be in fine agreement with calculated Fermi contact
integrals. DFT values calculated have asymmetry in their values, but these were taken to be
equivalent in order to simplify the simulation.
Table 3-3. Calculated Fermi contact integrals and 1H-aiso values of 2 obtained from simulations
of CW spectra, in MHz.
H3,6

H1,8

H2,7

H4,5

aiso/MHz

5.90

4.01

1.96

1.01

Fermi Contact/MHz

-5.57,-5.29-

5.61,-5.4

1.82,1.67

1.43,1.22

The 1H-aiso values are comparable to those of NaPSQ, albeit with slightly higher hyperfine
coupling constants, therefore no strong conclusions can be drawn from these results alone
relating to the influence of Lewis acidity on unpaired spin density distribution. Though, the line
shape alone of this radical is drastically different than that of NaPSQ, and due to the metal. In
Kazuhiro’s study of ion-pairs of PSQ, one can see the effect of the metal just by line shape alone.
158

So not only has it been seen that substituents have an effect on line shape and coupling, but also
the ion has a profound effect as well. This stems from various physical properties of the nuclei
involved, as well as the local magnetic environment of the molecule. Though, the protons
involved due not experience the greatest effect of the zinc ion, the carbonyl carbons change in
hyperfine is discussed below.

Figure 3-8. Room temperature CW spectrum (black) and simulation (red) of 3,
[Zn(tren)([13C2]PSQ)](BPh4) at 20 mG modulation amplitude. Simulation parameters: g =
2.007107; 13C-aiso values: 1.08 MHz and 2.78 MHz; Peak to peak line width: 0.0246 mT
Gaussian plus 0.0247 mT Lorentzian broadening. Experimental parameters: 9.39 GHz
microwave frequency, 0.80 mW microwave power, 3320 G centerfield, 50 G sweep width, 20
mG modulation amplitude, 1024 points. χ2 = 0.93.

159

Displayed in Figure 8 is the CW spectrum and simulation of 3, [Zn(tren)([13C2]PSQ)](BPh4).
The spectrum resembles that of Na[13C2]PSQ quite well, indicating initially that

13

C hyperfine

coupling had a similar effect on the spectrum. The coupling was within the same range as well;
13

C-aiso = 1.08 MHz and 2.78 MHz for 3, where 13C-aiso = 1.57 MHz for Na[13C2]PSQ ion pair.

This may indicate that an MII ion does not necessarily have an effect on the hyperfine coupling,
per se. However, the effect of the Zn(tren)2+ fragment does have an effect on the hyperfine
coupling and that is manifested as asymmetry in the isotropic coupling of the carbonyl carbon
atoms. The 1H hyperfine couplings were not found to be asymmetrical, however. It is possible
that the perturbations (from ZnII) on the 1H hyperfine couplings of “proton pairs” are small and
therefore do not constitute distinguishing differences between the pairs of protons.
DFT calculations predicted dissimilar spin density values for the carbonyl carbon atoms (0.20
versus 0.19) however, the disparity in the Fermi contact coupling is more telling. DFT predicted
13

C-aiso = 0.71 MHz and 2.17 MHz for the carbonyl carbon atoms; this is some validation for

DFT and the Fermi analysis correctly calculating spin properties for the types of molecules in
this study.

Although the exact magnitudes of the isotropic hyperfine couplings were not

predicted, the disparity of the “delocalization” of the hyperfine coupling was correctly predicted.
Table 3-4. Calculated spin density values at 1H and 13C nuclei for 2 and 3.

ρ(N)obs

H3,6

H1,8

H2,7

H4,5

CC=O

0.0013

0.0009

0.00043

0.00022

0.0024,
0.0009

ρ(N)calc

0.00125,

0.00125,

0.0004,

0.00032,

0.00193,

0.0012

0.0012

0.00037

0.00027

0.00063

160

Table 4 compares the calculated isotropic spin densities with the spin density values obtained
from the fits. As with the hyperfine coupling constants, the calculated values do agree with the
values obtained from fitting the spectra. Additionally, these values are comparable to spin
density values for NaPSQ and Na(3,6-(OMe)2-PSQ).
To summarize the results of the CW experiments, the isotropic hyperfine coupling constants
found for Zn-PSQ are comparable to those found for the ion-pairs of Chapter 2.

And

unfortunately, there does not seem to be a great effect of Lewis acidity on unpaired spin
delocalization. This is unfortunate because as the study moves towards unraveling the intricacies
of the Ni-PSQ system, it would have been advantageous to learn something about the spin
properties of this system in the absence of the paramagnetic Ni II ion. The greatest outlier of this
section are the results of fitting the Ga-PSQ spectra. The

69,71

Ga-aiso and

13

C-aiso hyperfine

coupling constants are much larger than the hyperfine coupling constants of any of the other
semiquinones examined within this study; however, the hyperfine coupling at gallium is in
agreement with other gallium complexes where a paramagnetic ligand is bonded to gallium. So
perhaps Lewis acidity does have an effect on unpaired spin density local, but it is not manifested
in MI or MII type ions but MIII and ions of higher charge.
It must be added, though, that a calculation was performed in which the GaIII ion was replaced
with GeIV, and the flowing Fermi contact integrals were obtained:

73

Ge = 1.99 MHz,

13

CC=O =

6.17 MHz and 5.46 MHz. So while the carbonyl carbon Fermi contact integrals are high, they
are not of the magnitude of the carbonyl carbon atoms of the Ga-PSQ complex, and are of the
same magnitude of the Zn-PSQ complex, as well as the ion-pairs of Chapter 2. It could be that
the hyperfine interactions of Ga-PSQ are truly an anomaly and inherent to Ga-SQ complexes in
general. Additionally, due to complications from ESEEM experiments, a spin echo could not be

161

found originating from Ga-PSQ, therefore the hyperfine coupling within these complexes could
not be further explored.
ESEEM. In Chapter 2 it was shown that the dipolar contribution to the hyperfine coupling
dominated the coupling in NaPSQ and Na(3,6-(OMe)2-PSQ). This was evident for the ion-pair,
dimer and tris complex (Na(3,6-(OMe)2-PSQ) only). For the Zn-PSQ complex studied in this
chapter, however, there is a single species being analyzed. Tris(2-aminoethyl)amine provides for
only a single semiquinone moiety to bond to the metal, so in the future it may be pertinent to use
such a concept for complexes of the type [MI(tren)(SQ)]. It was also concluded from the
previous section that the Lewis acidity provided by ZnII did not have an effect on the isotropic
hyperfine coupling, either for protons or carbonyl carbons.
The frequency domain spectrum and simulation are displayed below in Figure 9, below. The
spin Hamiltonian used to model the Zn-PSQ system has the form:
̂ = 𝑔𝑒 𝛽𝑒 𝐵𝑆̂𝑍 + 𝐴𝑍𝑍 𝑆̂𝑍 𝐼̂𝑍 + 𝐴𝑋𝑍 𝑆̂𝑍 𝐼̂𝑋 − 𝑔𝑁 𝛽𝑁 𝐵𝐼̂𝑍
𝐻

(3.3)

where the first and last terms are the electronic and nuclear Zeeman interactions, respectively,
and the middle terms are the electron-nuclear hyperfine coupling terms.

The hyperfine

parameters were as follows: C1, aiso = 1.08 MHz, T = 2.84 MHz; C2, aiso = 2.78 MHz, T = 2.46
MHz. These hyperfine coupling parameters were arrived upon by holding the isotropic values
constant and varying the dipolar values until a reasonable fit was reached.
The simulation accounts for three out of the four resonances from 0 – 6 MHz. The frequency
domain spectra show peaks at 0.7, 2.17, 3.06, 4.36 MHz and a large hump centered at ~11.8
MHz. None of these resonances correspond to the 13C Larmor frequency of 3.70 MHz at 345.5
mT magnetic field. The large hump at 11.54 MHz is most likely a result of proton anisotropy

162

from 2. This can be seen in the comparison spectrum in Figure 11. These features were
suppressed by using the division technique; however, the origin of this resonance is 2, Zn-PSQ.

Figure 3-9. 13C/12C Zn-PSQ ESEEM frequency domain spectrum and fit. Simulation
parameters: C1: aiso = 1.08 MHz, T = 2.84 MHz ρ = 0; C2: aiso = 2.78, T = 2.46, ρ = 0.
Experimental parameters: Field = 345.5 mT, 3 pulse ESEEM sequence, 12 ns time increment,
136 ns τ and 40 ns starting time value.
Although DFT calculations proved useful when comparing results to theory in the CW section,
the predicted anisotropic couplings do not contrast well with the simulated values.

DFT

calculated the following values: C1, aiso = 0.71 MHz, T = 17.97 MHz; C2, aiso = 2.17 MHz, T =

163

19.14 MHz. These dipolar values are more comparable with dipolar couplings of the NaPSQ ion
pair (aiso = 1.5741 MHz, T = 13.5674 MHz). Isotropic 13C coupling constants of 3 showed to be
comparable to NaPSQ and Na(3,6-(OMe)2-PSQ), which leads to the conclusion that the ZnII ion
has no effect on the hyperfine coupling, but when considering the dipolar coupling, this
conclusion becomes invalid. It is now evident then, that ZnII may not have an influence on the
isotropic coupling, which is a result of unpaired spin density in σ-type orbitals, but rather it has
an effect on the dipolar coupling, which is a result of the unpaired spin density in π-type orbitals.
The case may be that the main effect on the isotropic coupling, simply arises due to the
asymmetry introduced by the [Zn(tren)]2+ unit. This notion remains to be untested, however.
The reduction in dipolar coupling is likely due to the Lewis acidity of the Zn II, but it is unclear,
in a mechanistic sense, how this is occurring. It would be pertinent to the future of this study to
study changes in hyperfine coupling at the metal center.

164

Figure 3-10. 13C/12C Zn-PSQ ESEEM time domain spectrum and fit. Simulation parameters:
C1: aiso = 1.08 MHz, T = 2.84 MHz ρ = 0; C2: aiso = 2.78, T = 2.46, ρ = 0. Experimental
parameters: Field = 345.5 mT, 3 pulse ESEEM sequence, 12 ns time increment, 136 ns τ and 40
ns starting time value.

165

Figure 3-11. Comparison of 2 (blue) and 3 (red) ESEEM spectra. Fourier transforms were
normalized to compare features from each spectrum.
HYSCORE. Consistent with ESEEM of Zn-SQ type species, there have not been explorations
into the anisotropic hyperfine couplings of such complexes, and therefore HYSCORE spectra are
virtually unknown from this field as well. There have been instances of exchanging Fe II with
ZnII in QA and QB sites of bacterial reaction centers that contain ubisemiquinone; however, these
zinc ions do not seem to interact with the oxygen atoms of ubisemiquinone, rather they are part
of the biochemical system at large.44
Figure 12 shows the first and second quadrants of the HYSCORE spectrum of 3 at 345.5 mT.
As a reminder from Chapter 2, cross peaks that occur in the first quadrant signify weak

166

couplings, i.e. couplings where A < 2υLarmor, and cross peaks that occur in the second quadrant
are indicative of strong coupling, i.e., couplings where A ˃ 2υLarmor. On-diagonal peaks are a
consequence of the two-dimensional technique, where resonances from ESEEM are projected
onto each other. Off-diagonal peaks arise due to cross-coupling of ωα and ωβ frequencies
originating from the same hyperfine coupling.

Figure 3-12. HYSCORE spectrum of 3, Zn-13C2PSQ. Experimental conditions: 345.5 mT
magnetic field, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2 time values.
The most intense off-diagonal peaks observed for this spectrum are at (2.9,4.5) and (4.5,2.9)
centered about the 13Cυ frequency (3.69 MHz at 345.5 mT). These off-diagonal peaks are weak,
relative to all of the diagonal peaks except (4.15,4.15), but they are present and it justifies the use
of 13C-labeling. These cross-peaks are not present in the HYSCORE spectrum of 2 (spectrum in
Chapter 3 Appendix). It would have been ideal to detect off diagonal peaks corresponding to
167

13

Cυ frequencies, or even observe resonances in quadrant two, where strong couplings occur.

A

second spectra of 3 can be seen in Figure 13 where the threshold in the z-direction is much
lower, so as to see weak cross coupling in the second quadrant.

Figure 3-13. HYSCORE spectrum of 3, Zn-13C2PSQ. Experimental conditions: 345.5 mT
magnetic field, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2 time values. A lower zthreshold was used to accentuate weak cross-couplings in quadrant two.
These less prominent, off-diagonal peaks were not expected, as the dipolar hyperfine coupling
constants were small and within the same range as the isotropic hyperfine coupling constants.
However, the simulation of quadrants one and two (Figures 14 and 15), based on the ESEEM
simulation parameters, clearly shows off-diagonal resonances in the second quadrant and in the
same vicinity of the observed off-diagonal resonances.

168

Figure 3-14. HYSCORE simulation (red) and spectrum (black) of 3, Zn-13C2PSQ. Experimental
conditions: 345.5 mT magnetic field, 16 ns time increment, 136 ns tau, 36 ns starting t1 and t2
time values. Simulation parameters: g = 2.0033; 13C1, aiso = 1.08 MHz, T = 2.84 MHz; 13C2, aiso
= 2.78 MHz, T = 2.46 MHz; HYSCORE pulse sequence.

169

Figure 3-15. HYSCORE simulation (red) and spectrum (black) of 3, Zn-[13C2]PSQ in quadrant
one. Experimental conditions: 345.5 mT magnetic field, 16 ns time increment, 136 ns tau, 36 ns
starting t1 and t2 time values. Simulation parameters: g = 2.0033; 13C1, aiso = 1.08 MHz, T = 2.84
MHz; 13C2, aiso = 2.78 MHz, T = 2.46 MHz; HYSCORE pulse sequence.
Similarly, the knowledge gained from the HYSCORE spectrum of 3 is likely less important
than comparison to NaPSQ ion pair and dimer, and Ni-PSQ. The HSYCORE spectrum of 3 is
more comparable to that of the NaPSQ dimer, than the ion-pair, in that both spectra show the
effect of delocalization of a single radical across a binding metal (Na and Zn) to an ancillary
ligand (PhenQ and tren). The dimer was simulated to show equal delocalization across the four
carbonyl carbon atoms, which agrees with the spectra, both HYSCORE and ESEEM. 3 was
simulated to have asymmetrical distribution of hyperfine between two carbonyl carbons; it
remains to be observed what is causing this but it could be hypothesized that it is dependent on
the [Zn(tren)] unit. Exploring this possibility may require isotopic labelling of tren nitrogen
170

atoms with

15

N atoms. This addition would be done in the absence of

13

C nuclei, so as to

accentuate 15N hyperfine coupling. Alternatively, tren protons could be replaced with 2H atoms
(I = 1), which would allow faster relaxation of the

14

N nuclei (I = 1) and pronounce the

13

C

effect.

3.4 Conclusions and Future Work

At the outset of this project it was desired to know the effect of Lewis Acidity on spin density
distribution, and to further understand spin density distribution in Mn+-SQ complexes. It was
found for ZnII-PSQ, that upon complexation the dipolar portion collapsed to be within range of
the isotropic coupling. This stands to be the main effect for bonding with a M II ion with an
ancillary ligand. Recall from Chapter 2 that for both ion-pair and dimer molecules, the lowest
dipolar coupling was 2.77 MHz for NaPSQ, but was 7.96 MHz for Na(3,6-(OMe)2-PSQ).
Although, to conclude that carbonyl carbon dipolar hyperfine coupling decreases drastically as a
result of complexation of a semiquinone to a metal would be short-sided. Attempts were made
to study the ESEEM spectra of Ga-PSQ and Ga-13C2PSQ, but no spin echo was located and
therefore no ESEEM spectra were taken; therefore, there is no information on carbonyl carbon
dipolar coupling for MIII-PSQ complexes.
As for the isotropic values of Ga-PSQ and Ga-13C2PSQ, the magnitude increased by more than
a factor of two (relative to all other compounds studied in this chapter and Chapter 2) in the case
of both carbonyl carbon atoms and gallium. It can be hypothesized that this is most likely a
consequence resulting from the bonding of the oxygen atoms to the GaIII ion. The

69,71

Ga-aiso

values were in agreement with similar complexes from literature. It would be advantageous to

171

further this study using CdII in place of ZnII to gain hyperfine information at the metal center for
MII-SQ type complexes. This could then be compared to the hyperfine interactions at gallium
and sodium, and possibly a trend, or more concrete theory could be had.
An interesting observation from the DFT portion of this chapter was the magnitude that the
spin density shifts as it binds to ZnII and GaIII. These findings suggest that the metal causes spin
density to polarize more towards the carbonyl carbon atoms. The carbonyl carbon hyperfine
coupling constants did decrease relative to the free ligands in Chapter 2. It would be helpful for
this study to expand by increasing the number of molecules studied. This would solidify the
findings in this chapter.
In addition to having a Ga-SQ complex that exhibits a spin echo, it would be advantageous to
study the substituent effects predicted by DFT. As was seen by the CW spectrum of 5, the 13C
nuclei have such a line broadening effect that it completely masks all other hyperfine coupling.
Field modulation studies would be useful for molecules of this type once a suitable system is
found. And if a suitable system is found then it may be prudent to pursue 17O labeling in order to
determine if the predicted loss in oxygen spin density is a real trend. A combination of
17

13

C and

O labeling (in separate molecules) could prove and or disprove this prediction.
Now that it has been assessed what happens to hyperfine interactions at the carbonyl carbon

in free radicals and radicals bound by MII and MIII type ions, the next chapter will focus on the
effect of adding an exchange interaction with unpaired spins on the MII ion.

172

APPENDIX

173

APPENDIX
CW-EPR
function res = LDM2120cw1_fit(hpar)
% LDM2120 Ga-PSQ 20K 1 G Mod. Amp. Simulation & Fit
%==========================================================================
% Load the experimental spectrum
[B,spc,params] = eprload('LDM2120cw10.DTA'); %Loads the spectrum
B=B/10; %Gauss to mT
Edata = spc/(max(spc)-min(spc)); % Normalizes the data
lmin = 112; % min point to start w/in the data set
lmax = 412; % max point to stop w/in data set
% Simulation Parameters to give the least-squares fitting program a place
% to start.
Sys0.S = 1/2;
Sys0.g = 2.00535;
Sys0.lwpp = [0.710 0.2815]; % mT
Sys0.Nucs = 'Ga';
Sys0.n = 1;
Sys0.A = 19.7295; % MHz
% Load the experimental Parameters
Exp.mwFreq = params.MWFQ/1.00e+09;
Exp.CenterSweep = [params.A1CT*1000,params.A1SW*1000];
Exp.ModAmp = 0.1; % mT
Exp.nPoints = params.XPTS;
Exp.Temperature = 20;
% Call the simulation program
[B,sim] = pepper(Sys0,Exp);
Cdata = sim/(max(sim)-min(sim)); % Normalizes the simulation
offset = Edata(lmin); % corrects for baseline offset
Cdata = Cdata + offset; % Corrected baseline
% Calculation of residual
res = 0.0;
for ipt = lmin:lmax,
res = res + (Cdata(ipt) - Edata(ipt))^2;
end;
plot(B(lmin:lmax),Cdata(lmin:lmax),' r',B(lmin:lmax),Edata(lmin:lmax),' k');
return;

174

function res = LDM2120_13C_fit(hpar)
% LDM2120_13C Ga-[13C2]PSQ Room Temp. 1 G Mod. Amp. Simulation & Fit
%==========================================================================
% Load the experimental spectrum
[B,spc,params] = eprload('ldm2120_13cwA.DTA'); %Loads the spectrum
B=B/10; %Gauss to mT
Edata = spc/(max(spc)-min(spc)); % Normalizes the data
lmin = 51; % min point to start w/in the data set (depends on # of points in
data set, 512)
lmax = 472; % max point to stop w/in data set
% Simulation Parameters to give the least-squares fitting program a place
% to start.
Sys0.S = 1/2;
Sys0.g = 2.00532;
Sys0.lwpp = [1.0939 0.5693]; % mT
Sys0.Nucs = 'Ga,13C,13C'; %
Sys0.n = [1 1 1]; %
Sys0.A = [19.7295 27.7707 32.1071]; % MHz
% Load the experimental Parameters
Exp.mwFreq = params.MWFQ/1.00e+09;
Exp.CenterSweep = [params.A1CT*1000,params.A1SW*1000];
Exp.ModAmp = 0.1; % mT
Exp.nPoints = params.XPTS;
Exp.Temperature = 20;
% Call the simulation program
[B,sim] = pepper(Sys0,Exp);
Cdata = sim/(max(sim)-min(sim)); % Normalizes the simulation
offset = Edata(lmin); % corrects for baseline offset
Cdata = Cdata + offset; % Corrected baseline
% Calculation of residual
res = 0.0;
for ipt = lmin:lmax,
res = res + (Cdata(ipt) - Edata(ipt))^2;
end;
plot(B(lmin:lmax),Cdata(lmin:lmax),' r',B(lmin:lmax),Edata(lmin:lmax),' k');
return;

175

function res = LDM2163_RT_20mG_fit(hpar)
% LDM2163 Zn-PSQ Room Temp. 20 mG Mod. Amp. Simulation & Fit
%==========================================================================
% Load the experimental spectrum
[B,spc,params] = eprload('LDM2163cw_0.02G_RT.DTA'); %Loads the spectrum
B=B/10; %Gauss to mT
Edata = spc/(max(spc)-min(spc)); % Normalizes the data
lmin = 370; % min point to start w/in the data set
lmax = 638; % max point to stop w/in data set
% Simulation Parameters to give the least-squares fitting program a place
% to start.
Sys0.S = 1/2;
Sys0.g = 2.0071;
Sys0.lwpp = [0.0136 0.0107]; % mT
Sys0.Nucs = '1H,1H,1H,1H'; % Treating the nuclei seperately
Sys0.n = [2 2 2 2]; % 4 classes of 1H nuclei
Sys0.A = [1.0135 1.957 4.0089 5.904]; % MHz
% Load the experimental Parameters
Exp.mwFreq = params.MWFQ/1.00e+09;
Exp.CenterSweep = [params.A1CT*1000,params.A1SW*1000];
Exp.ModAmp = 0.002; % mT
Exp.nPoints = params.XPTS;
% Call the simulation program
[B,sim] = garlic(Sys0,Exp);
Cdata = sim/(max(sim)-min(sim)); % Normalizes the simulation
offset = Edata(lmin); % corrects for baseline offset
Cdata = Cdata + offset; % Corrected baseline
% Calculation of residual
res = 0.0;
for ipt = lmin:lmax,
res = res + (Cdata(ipt) - Edata(ipt))^2;
end;
plot(B(lmin:lmax),Cdata(lmin:lmax),' r',B(lmin:lmax),Edata(lmin:lmax),' k');
return;

176

function res = LDM2189_RT_20mG_fit(hpar)
% LDM2189 Zn-[13C2]PSQ Room Temp. 20 mG Mod. Amp. Simulation & Fit
%==========================================================================
% Load the experimental spectrum
[B,spc,params] = eprload('LDM2189cw_0.02G_RT.DTA'); %Loads the spectrum
B=B/10; %Gauss to mT
Edata = spc/(max(spc)-min(spc)); % Normalizes the data
lmin = 340; % min point to start w/in the data set
lmax = 680; % max point to stop w/in data set
% Simulation Parameters to give the least-squares fitting program a place
% to start.
Sys0.S = 1/2;
Sys0.g = 2.007107;
Sys0.lwpp = [0.0246 0.0247]; % mT
Sys0.Nucs = '1H,1H,1H,1H,13C,13C'; % Treating the nuclei seperately
Sys0.n = [2 2 2 2 1 1]; % 4 1H & 2 13C
Sys0.A = [1.0135 1.957 4.0089 5.904 1.0871 2.7845]; % MHz
% Load the experimental Parameters
Exp.mwFreq = params.MWFQ/1.00e+09;
Exp.CenterSweep = [params.A1CT*1000,params.A1SW*1000];
Exp.ModAmp = 0.002; % mT
Exp.nPoints = params.XPTS;
% Call the simulation program
[B,sim] = garlic(Sys0,Exp);
Cdata = sim/(max(sim)-min(sim)); % Normalizes the simulation
offset = Edata(lmin); % corrects for baseline offset
Cdata = Cdata + offset; % Corrected baseline
% Calculation of residual
res = 0.0;
for ipt = lmin:lmax,
res = res + (Cdata(ipt) - Edata(ipt))^2;
end;
plot(B(lmin:lmax),Cdata(lmin:lmax),' r',B(lmin:lmax),Edata(lmin:lmax),' k');
return;

177

function res = LDM2189ESEEMtime_fit(hpar)
% LDM2189 Time Domain Sim/Fit Using 2189 time domain divided by 2163
% (Zn-[13C2]PSQ)/(Zn-PSQ)
%==========================================================================
% Load the experimental spectrum
[time,amp] = eprload('LDM2189_3p3455G136ERat.DTA');
Amp = real(amp)/(max(real(amp))); % Normalizes the data
time = (time/1000);
lmin = 1;
lmax = 512;
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.0033;
Sys0.Nucs = '13C,13C'; % Treating the nuclei differently
Sys0.n = [1 1]; % 2 classes of 13C nuclei
Sys0.A_ = [1.0871 2.8425 0; 2.7845 2.4614 0]; % [a-iso T rhombicity]
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012;
Exp.tau = 0.136;
Exp.T = 0.04;
Opt.nKnots = 91;
[x,y] = saffron(Sys0,Exp,Opt);
Y = real(y)/max(real(y)); % Normalizes the simulation
res = 0;
for ipt = lmin:lmax,
res = res + (Y(ipt) - Amp(ipt))^2;
end;
plot(x(lmin:lmax),Y(lmin:lmax),' r',time(lmin:lmax),Amp(lmin:lmax),' k');
return;

178

function res = LDM2189ESEEMfit(hpar)
% LDM2189div2136 Frequency Domain Sim/Fit
% (Zn-[13C2]PSQ)/(Zn-PSQ)
%==========================================================================
% Load the experimental spectrum
[larmor,intens] = eprload('LDM2189_3p3455G136spN.DTA');
larmor = larmor*1000; %GHz -> MHz
lmin = 1;
lmax = 186;
% Simulation Parameters
Sys0.S = 1/2;
Sys0.g = 2.0033;
Sys0.Nucs = '13C,13C'; % Treating the nuclei seperately
Sys0.n = [1 1]; % 2 classes of 13C nuclei
Sys0.A_ = [1.0871 2.8425 0; 2.7845 2.4614 0]; % [a-iso T rhombicity]
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012;
Exp.tau = 0.136;
Exp.T = 0.04;
Opt.nKnots = 91;
[x,y] = saffron(Sys0,Exp,Opt);
ESnormDAL; % Normalizes the simulation
res = 0;
for ipt = (lmin):lmax,
res = res + (spec(ipt) - intens(ipt))^2;
end;
plot(larmor(lmin:lmax),intens(lmin:lmax),' k',f(lmin:lmax),spec(lmin:lmax),'
r');
return;

179

% LDM2189 HYSCORE Spectrum Zn-[13C2]PSQ
%==========================================================================
% Load the experimental spectrum
[x,Z,params]=eprload('LDM2189_hysc3455G136sp.DTA');
f1=x{1,1}*1000.;
f2=x{1,2}*1000.;
% Plot the spectrum
contour(f1,f2,Z);
% Edit the plot?
ans=input('Edit Contour plot (y/n)? ','s');
if(ans=='y'),
hyscedit;
end
return;
% LDM2189 HYSCORE Molecule Simulation Zn-[13C2]PSQ
%==========================================================================
clear;
Sys0.S = 1/2;
Sys0.g = 2.0033;
Sys0.Nucs = '13C,13C';
Sys0.A_ = [1.0871 2.8425 0; 2.7845 2.4614 0];
% Load the experimental Parameters
Exp.Field = 345.5;
Exp.Sequence = 'HYSCORE';
Exp.dt = 0.016;
Exp.tau = 0.136;
Exp.t1 = 0.036;
Exp.t2 = 0.036;
Exp.nPoints = [256 256];
Opt.nKnots = 181;
[x1,x2,y1,p1] = saffron(Sys0,Exp,Opt);
% Plot the data
f1 = p1.f1;
f2 = p1.f2;
Z = (real(p1.fd));
% set threshhold for contour at 10% max amplitude
Zthresh = 0.05*max(max(Z));
for iy=1:512,
for ix = 1:512,

180

if Z(ix,iy) < Zthresh,
ZR(ix,iy) = Zthresh;
else
ZR(ix,iy) = Z(ix,iy);
end
end;
end;
%
f1min=-9.0;
f1max=9.0;
f2min=0.0;
f2max=9.0;
%
frinc=f1(2)-f1(1);
%
xmin=floor((f1min-f1(1))/frinc)+1;
xmax=floor((f1max-f1(1))/frinc)+1;
ymin=floor((f2min-f2(1))/frinc)+1;
ymax=floor((f2max-f2(1))/frinc)+1;
contour(f1(xmin:xmax),f2(ymin:ymax),ZR(ymin:ymax,xmin:xmax),10,'r');
xlabel('f1 (MHz)');
ylabel('f2 (MHz)');
grid;

181

REFERENCES

182

REFERENCES

(1)

Shannon, R. D. Acta Crystallogr. 1976, A32, 751–767.

(2)

Bencini, A.; Carbonera, C.; Vaz, M. G. F. Dalt. Trans. 2003, 1701–1706.

(3)

Poddel’sky, A. I.; Smolyaninov, I. V.; Skatova, A. A.; Lukoyanov, A. N.; Fukin, G. K.;
Berberova, N. T.; Cherkasov, V. K.; Abakumov, G. A. Zeitschrift fĂźr Anorg. und Allg.
Chemie 2008, 634 (6–7), 1154–1160.

(4)

Shultz, D. a.; Bodnar, S. H. Inorg. Chem. 1999, 38 (3), 591–594.

(5)

Ii, Z.; Iii, C.; Iii, G.; Lange, C. W.; Conklin, B. J.; Pierpont, C. G. Inorg. Chem. 1994, 33,
1276–1283.

(6)

Fehir, R. J. Differential Polarization of Spin & Charge Density in Phenoxy Radicals,
Semiquinone Radicals, and Nickel(II) & Chromium(III) Phenanthrenesemiquinone
Complexes, Michigan State University, 2009, Vol. 113.

(7)

Beckmann, U.; Bill, E.; WeyhermĂźller, T.; Wieghardt, K. Eur. J. Inorg. Chem. 2003,
1768–1777.

(8)

Schrauben, J. N.; Guo, D.; McCracken, J. L.; McCusker, J. K. Inorganica Chim. Acta
2008, 361 (12–13), 3539–3547.

(9)

PauláAggarwal, R. J. Chem. Soc. Dalt. Trans. 1992, 655–662.

(10)

Fehir, R. J. Differential Polarization of Spin and Charge Density in Phenoxy Radicals,
Semiquinone Radicals, and Ni(II) and Cr(III) Phenanthrenesemiquinone Complexes,
Michigan State University, 2009.

(11)

Stoll, S.; Schweiger, A. J. Magn. Reson. 2006, 178 (1), 42–55.

(12)

Stoll, S.; Britt, R. D. Phys. Chem. Chem. Phys. 2009, 11 (31), 6614.

(13)

Stoll, S. In Handbook of Multifrequency Electron Paramagnetic Resonance: Data and
Techniques; Misra, S. K., Ed.; Wiley-VCH, 2014; pp 69–138.

(14)

Fauth, J. J. Magn. Reson. 1986, 66 (1), 74–85.

(15)

Warncke, K.; McCracken, J. J. Chem. Phys. 1994, 101 (3), 1832.

(16)

Gemperle, C.; Aebli, G.; Schweiger, A.; Ernst, R. R. J. Magn. Reson. 1990, 88, 241–256.

(17)

https://www.mathworks.com/help/matlab/ref/fminsearch.html?s_tid=gn_loc_drop
https://www.mathworks.com/help/matlab/ref/fminsearch.html?s_tid=gn_loc_drop.

(18)

Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J.
R.; Montgomery, Jr., J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.;
Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.;
Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa,
J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J.
E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.;
183

Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.;
Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V.
G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;
Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.;
Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin,
R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.;
Pople, J. A. Gaussian, Inc.: Wallingford, CT 2004.
(19)

Becke, A. D. J. Chem. Phys. 1993, 98 (7), 5648.

(20)

Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37 (2), 785–789.

(21)

Stephens, P.; Devlin, F. J. … 1994, 98 (45), 11623–11627.

(22)

Ditchfield, R.; Hehre, W. J.; Pople, J. a. J. Chem. Phys. 1971, 54 (2), 724.

(23)

Hehre, W. J.; Ditchfield, R.; Pople, J. a. J. Chem. Phys. 1972, 56 (5), 2257–2261.

(24)

Hariharan, P. C.; Pople, J. a. Theor. Chim. Acta 1973, 28 (3), 213–222.

(25)

Hariharan, P. C.; Pople, J. A. Mol. Phys. 1974, 27 (1), 209–214.

(26)

Gordon, M. S. Chem. Phys. Lett. 1980, 76 (1), 163–168.

(27)

Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; DeFrees, D. J.;
Pople, J. a. J. Chem. Phys. 1982, 77 (7), 3654–3665.

(28)

Binning, R. C.; Curtiss, L. A. J. Comput. Chem. 1990, 11 (10), 1206–1216.

(29)

Blaudeau, J.-P.; McGrath, M. P.; Curtiss, L. A.; Radom, L. J. Chem. Phys. 1997, 107 (13),
5016.

(30)

Rassolov, V. A.; Pople, J. A.; Ratner, M. A.; Windus, T. L. J. Chem. Phys. 1998, 109 (4),
1223.

(31)

Rassolov, V. A.; Ratner, M. A.; Pople, J. A.; Redfern, P. C.; Curtiss, L. A. J. Comput.
Chem. 2001, 22 (9), 976–984.

(32)

Petersson, G. A. J. Chem. Phys. 1991, 94 (9), 6081–6090.

(33)

Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72 (1), 650–
654.

(34)

McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72 (10), 5639.

(35)

Wachters, A. J. H. J. Chem. Phys. 1970, 52 (3), 1033.

(36)

Raghavachari, K.; Trucks, G. W. J. Chem. Phys. 1989, 91 (2), 1062.

(37)

Bencini, A.; Carbonera, C.; Totti, F. Chemistry 2004, 10 (6), 1472–1480.

(38)

Pierpont, C. Coord. Chem. Rev. 1981, 38 (1), 45–87.

(39)

Lange, C.; Pierpont, C. Inorganica Chim. Acta 1997, 263, 219–224.

(40)

Piskunov, A. V; Maleeva, A. V; Mescheryakova, I. N.; Fukin, G. K. 2012, 4318–4326.
184

(41)

Fehir, R. J. DIFFERENTIAL POLARIZATION OF SPIN AND CHARGE DENSITY IN
PHENOXY RADICALS, SEMIQUINONE RADICALS, AND NICKEL(II) AND
CHROMIUM(III) PHENANTHRENESEMIQUINONE COMPLEXES, 2009.

(42)

Broze, M.; Luz, Z. J. Phys. Chem. 1967, 71 (11), 3690–3693.

(43)

Adams, M.; Blois, M. S.; Sands, R. H. J. Chem. Phys. 1958, 28 (5), 774.

(44)

Martin, E.; Samoilova, R. I.; Narasimhulu, K. V; Lin, T.-J.; O’Malley, P. J.; Wraight, C.
A.; Dikanov, S. A. J Am Chem Soc 2011, 133 (14), 5525–5537

185

Chapter 4. Measuring Spin Density Delocalization in NiII-Phenanthrenesemiquinone
Complexes

4.1 Introduction

This chapter aims to monitor spin density in the presence of an exchange interaction and
assess if the method of using 13C as a spin label correlates with theoretical results. Recall from
Chapter 1 that a DFT study was carried out to study the average oxygen spin density as a
function of calculated exchange coupling constant, J, as seen in Figure 1.

-280

SO2CN
-270

NO2
CN
H

-260

IF4
CF3
C(CF3)3

Br

J/cm-1

-250

Me

-240
-230
OMe

Cl
F
TMS
t-Bu

Vinyl

-220

-210
-200
0.21

R² = 0.9941

NH2
NMe2
0.22

0.23

0.24

0.25

0.26

0.27

Carbonyl-O (ρ)

Figure 4-1. Average oxygen spin density versus J, reproduced from Reference 1.

186

0.28

Figure 4-2.
Molecular structure of [Ni(tren)(3,6-R2-PSQ)]+ ( where tren = tris(2amineoethyl)amine and 3,6-R2-PSQ = 3,6-R2-Phenanthrenesemiquinone) complexes.
The structure of the complexes studied can be seen below in Figure 2. These calculations were
explained in Chapter 1, however, to restate, the importance of these calculations in this body of
work is as a backdrop for why the choice has been made to further study these molecules. It is
our desire to understand these compounds with respect to their spin properties, in light of the fact
that the magnitude of J has not be measured except when R = NH21. The purpose of these
calculations was to uncover any theoretical trend between substituent (derivative of
[Ni(tren)(3,6-R2-PSQ)]+, and the magnitude of exchange coupling, J. A linear trend was found,
correlating J with the average oxygen atom spin density of certain substituents; specifically,
electron donating groups were found to have lower J values and lower average oxygen spin
density values whereas electron withdrawing groups were found to have higher J values and
higher average oxygen spin density values.

The spin density values were obtained by

mathematical manipulation of spin density values that were calculated by DFT with Natural
Population Analysis (NPA) from the Natural Bond Order (NBO) package within Gaussian032–4.

187

What is being correlated here are two values that cannot be measured directly, by
conventional methods.

Exchange coupling constants can be extrapolated from variable

temperature magnetic susceptibility measurements, or directly by inelastic neutron scattering,
while spin populations can be obtained from neutron diffraction experiments.

All three

techniques require a solid, crystalline sample. Regardless of the more unconventional means that
require synchrotron light sources, spin density populations can be calculated from hyperfine
coupling constants5, as it has been accomplished in this work in the preceding chapters; this
formula that has been used only takes into account the isotropic hyperfine coupling. This
coupling is a result of the resident electron density at the nucleus (s-orbital). Therefore, building
on what has been accomplished from DFT calculations, the hyperfine parameters that have been
simulated will be compared to the hyperfine parameters calculated. When referred to in this
chapter, the term spin density will be directly relating to the aforementioned calculations.
It has also been previously stated why we chose to use

13

C-spin labeling instead of

17

O;

however, as a reminder, there is no correlation between carbonyl carbon spin density and J with
respect to these compounds as seen in Figure 3. These are theoretical results and thus are not
necessarily a perfect reflection of what the experimental results will be. Therefore, by using 13C
as an isotopic label at the carbonyl carbon atoms, it is an experiment in and of itself as to whether
or not

13

C labeling of molecules of the type [TM(tren)([13C2]PSQ)]n+ (where TM = transition

metal and n = 0, 1, 2, 3…n) is a sound method for tracking changes in spin density from one
complex to the next within a series of molecules.

188

-280
SO2CN
-270

NO2

IF4
CF3
C(CF3)3

CN

-260

H

Cl

J/cm-1

-250

F
Me
TMS
t-Bu

Br

-240
-230

Vinyl

OMe

-220
-210
-200
0.165

NH2
NMe2
0.17

0.175

0.18
0.185
Carbonyl-C (ρ)

0.19

0.195

Figure 4-3. Average carbonyl carbon spin density versus J.
It has been seen in Chapter 2 that the isotropic hyperfine coupling constants obtained
from the CW experiments create a map of the spin density at each nucleus. Between the two
compounds studied it was suggested that a substituent effect was operative, decreasing the
isotropic spin density at the carbonyl carbons in 3,6-(OMe)2-PSQ relative to PSQ. It was also
apparent in the overall line shape of the CW spectra that installment of methoxy groups to the
phenanthrenequinone backbone created a perturbation in the spin density distribution. However,
when studying the low temperature ESEEM time domain and Fourier transform spectra, it was
observed that the dipolar hyperfine coupling constants (electron density in π – type orbitals),
were larger in Na[13C2]3,6-(OMe)2-PSQ when compared to Na[13C2]PSQ.

189

The isotropic

coupling constants were smaller for [13C2]3,6-(OMe)2-PSQ type radicals, compared to Na[13C2]
PSQ type radicals, as well.
Although ESEEM and HYSCORE have been used for decades to study transition metal
complexes6, there have not been many reports concerning synthetic, non-biologically related,
paramagnetic transition metal complexes bound to organic radical ligands. Indeed, most of the
literature focuses on biological molecules, where it can be difficult to determine structures that
are operative in biological processes. A combination of EPR techniques are often used to help
determine structural parameters from EPR data. Here, however, our intention is to gain carbonyl
carbon hyperfine coupling constants to ascertain as to what, if any effect is to be had on the
coupling from electron exchange.

4.2 Experimental
General Methods. Reagents were purchased from commercially available sources and used
without further purification unless otherwise stated.

Stabilizer-free tetrahydrofuran was

purchased from Fisher Scientific and either distilled over sodium metal and benzophenone and
freeze, pump, thawed to remove O2, or taken from a dry still with a solvent bomb. Elemental
analyses and electro-spray ionization ESI mass spectra were obtained through the analytical
facilities at Michigan State University.
Synthesis. Ni(tren)Cl2 (1). The procedure for the preparation of this compound has
been previously reported.1 To a stirring solution of 2.38 g (10.0 mmol) NiCl2∙6H20 in absolute
ethanol 1.5 mL (10.0 mmol) of tren was added. The solution is stirred for one hour during which
a light purple precipitate forms. The reaction mixture is filtered using a medium frit. The
precipitate is then washed with distilled ethanol, recrystallized from hot ethanol to produce a

190

light blue microcrystalline product. The light blue microcrystalline product can be filtered with a
fine frit into the same filtrate from the first filtration. This was left out overnight to produce dark
blue crystals. The product should be stored in a desiccator or dry box because it is highly
hygroscopic. Recrystallized yield: 37.15%. Anal. calcd. for C6H18Cl2N4Ni∙H2O: C, 24.52; H,
6.86; N, 19.07; Found: C, 24.50; H, 6.74; N, 19.01.
[Ni(tren)(PSQ)](BPh4) (2). The procedure for the preparation of this compound has
been previously reported.1 The entire reaction including recrystallizations were done in a dry
box. 0.24 g (0.86 mmol) of Ni(tren)Cl2 was dissolved in 20 mL methanol. This solution was
added to a 20 mL methanol solution of 0.20 g (0.96 mmol) 9,10-phenanthrenecatechol and 0.076
g (1.9 mmol) NaOH powder. The gold colored reaction mixture was stirred for 30 minutes and
then oxidized with 0.32 g (0.96 mmol) of [FeCp2](PF6), and the color changed to dark brown.
The solution was then stirred for another 15 minutes and filtered using a medium frit. A
concentrated methanol solution of 3.3 g (9.6 mmol) NaBPh4 was added drop wise to the filtered
reaction mixture and it was left out to crystallize. A dark brown microcrystalline product was
recovered. The crystals were washed with methanol (10 mL x 3) and ether (10 mL x 3) and dried
in vacuo. The product can be recrystallized by adding 200 mL of ether to a concentrated
acetonitrile solution of the complex. The dark brown crystals were then filtered off with a fine
frit. The product should be stored under N2. Recrystallized yield: 54.7%. Anal. Calcd. From
C44H47BN4NiO2: C, 72.06; H, 6.46; N, 7.64; found: C, 72.02; H, 6.57; N, 8.00.
[Ni(tren)([9,10-13C2]PSQ)](BPh4) (3). This compound was prepared analogously to
compound 2. Recrystallized yield: 20%. MS [ESI (Hi-Res) m/z (rel. int.)]: C1813C2H26N4NiO2+∙
414.1478 (100%), C24H20B- 319.1663 (100%).

191

Physical Measurements. EPR Spectroscopy. All EPR data were collected on a Bruker
Elexsys E-680X spectrometer, operating at X-band frequency (9.42-9.71 GHz).

Low

temperature experiments were performed between 4-30 K using an Oxford Instruments liquid
helium flow system equipped with a CF-935 cryostat and an ITC-503 temperature controller.
Samples were prepared in a dry box in modified Wilmad Glass SQ-706 EPR tubes with 3.8 mm
diameter. Prior to sample preparation, tubes were modified in the glass blowing lab at Michigan
State University to contain a grated seal and vacuum adapter port. A vacuum was pulled on the
sample tubes and then the vacuum ports were blown off at the grated seal, by hydrogen torch.
Prior to collecting data, power-dependence experiments were conducted to ensure non-saturation
of the detector. Continuous wave spectra were simulated using the garlic module of Easy Spin
4.57,8,9.
All pulsed experiments were conducted at X-band microwave frequency. ESEEM spectra
were obtained using a three pulse sequence of 90°-τ-90°-T-90° with 90° pulse lengths of 16 ns.
Tau values were chosen to suppress hyperfine contributions from weakly coupled protons. Echo
amplitudes were integrated at FWHM and the T values were scanned from 40 ns to 6 Îźs with a
12 ns time increment. A four step phase-cycling procedure was used to eliminate unwanted 2pulse echoes.10 Raw data were processed by phase shifting the spectra to remove the dispersion
component from the time domain data. The in-phase portion of the data were normalized by
dividing by a biexponential background decay function. These normalized data were then
further processed by subtracting a second degree polynomial (DC-component of the signal),
followed by application of a Hamming window. The data set was then zero-filled from 512
points to 1024 points and Fourier transformed. Frequency spectra are displayed as an absolute
value of the transformed time domain data.

13

C and

192

12

C spectra are presented as one divided

spectrum at each magnetic field value.

After normalizing each time domain spectrum

individually, the 13C spectrum was divided by the 12C spectrum and the same work up procedure
described above was applied11.
HYSCORE experiments were carried in a similar manner to the ESEEM experiments but
with a four-pulse sequence of 90°-τ-90°-t1-180°-t2-90° with a 180° pulse length of 16 ns. Echo
amplitudes were measured at FWHM and τ values were chosen as stated above. T1 and t2 values
were scanned from 40 ns to 2 Îźs with 16 ns time increments. Data sets were 128 points in both
dimensions. A 4-step phase cycling program was used to filter out unwanted 2-pulse echoes.12
The real, or in-phase portion of the HYSCORE data were processed, in both dimensions, by first
subtracting a 2° polynomial, then tapering the data with a Hamming window and zero-filling to
256 points. A two-dimensional fast Fourier transform was applied and absolute values taken.
Data are presented as a two-dimensional contour plot. Each contour plot is presented at a
threshold of 5-20% of the maximum peak amplitude (z-direction).
Spectra were fit using customized MATLAB scripts where EasySpin was used to perform
spectral simulations and the fminsearch function in Matlab was used to optimize the calculation
parameters. Fminsearch is a function that finds the minimum of an unconstrained multivariable
function using a derivative-free method13. Residuals were calculated by squaring the difference
of the experimental and simulated amplitudes. Goodness of fit for each spectral fit was assessed
by calculating a χ2 value with the following equation:

𝜒𝑛2 =

𝑒𝑥𝑝
𝑁 (𝑦𝑖
𝛴𝑖=1

− 𝑦𝑖𝑐𝑎𝑙𝑐 )
𝜎2
𝑁−𝐿

193

2

(4.1)

where the sum is taken over N points, yiexp and yicalc are the experimental and calculated y-axis
values, σ is the standard deviation fixed at 2% of the maximum spectral amplitude and L is the
number of adjustable variables. The χ2 values in this chapter are not good signs of merit for the
spectra in this chapter, due to the difficult nature of modeling exchange-coupled molecules
properly.

4.3 Results and Discussion.

Synthesis.

Complexes of this type are typically prepared from a metal precursor

complex that contains an ancillary ligand and good leaving groups.

The ancillary ligand

ultimately serves the purpose of occupying coordination sites while simultaneously discouraging
any spin polarization.

A catecholate is then added to the metal precursor complex and

subsequently oxidized. A general scheme for the synthesis of the compounds studied can be
seen in Figure 4 (both 12C, 2, and 13C, 3, were prepared analogously).

Figure 4-4. Synthetic scheme for synthesizing [Ni(tren)(PSQ)](BPh4) complexes.
The complex precursor, Ni(tren)Cl2 (1), was prepared by adding tren to the NiCl2∙6 H2O salt.
The salt is neon green and upon addition of tren, the H2O molecules are displaced, and the neon
green changes to bright royal blue, as a result of strong absorption in the red14. After stirring for
194

several minutes, a light blue crystalline powder crashes out of solution. This precipitate is then
filtered and washed with cold ethanol. The complex is then recrystallized from hot ethanol,
filtered and stored in a desiccator. Hydrolyzed versions of this complex are light purple.
Complex 2 was made by first placing 9,10-phenanthrenecatechol (PCatH2) (synthesized
by a previous lab mate) in MeOH with 2 equivalents of NaOH. By adding NaOH, this creates an
acid-base reaction, producing H2O as the side product. Following this step, Ni(tren)Cl2 is added
to this stirring PCat2- reaction mixture, producing a gold-brown colored solution. The remaining
steps for synthesizing this compound are outlined below.
The above steps were altered to increase the yield of the 13C-labeled complex. PCatH2 is
synthesized from phenanthrenequinone (PhenQ), and this is one more step than what is necessary
for synthesizing 3. As performed in Chapters 2 and 3 of this dissertation, PhenQ can be reduced
chemically be excess sodium metal. By reducing the number of compounds to be made, this
creates a likelihood for increasing the yield of the final complex. PCatH2 is highly hygroscopic
and must be immediately stored away from sources of hydration.
attempts were made to incorporate

17

In addition to this, earlier

O atoms into the quinone ligand, so as to measure

17

O

hyperfine coupling constants and gain spin density information at the center of the exchange
interaction. By having 17O atoms on the quinone, it would be desirable to have a route in which
16

O sources would not interact with these isotopes. It would be possible that, during the acid-

base reaction, the oxygen atoms from HO- would exchange with the 17O nuclei on the resulting
catecholate. It is noted that this would be a slim chance, but 17O-labeling may be in the future of
this project.
Complexes 2 and 3 were prepared by first performing a two electron reduction with
excess sodium metal. The sodium metal was cut in light mineral oil into 5x5x5 mm3 cubes and

195

prior to pumping into the glove box the oil was removed as much as possible. Before adding the
sodium metal to the THF solution of phenanthrenequinone, the sodium chunks were washed with
dry THF and then added to the reaction flask. Upon addition the solution changes to dark green,
corresponding to the semiquinone oxidation state. After stirring overnight, a blood red solution
persisted for Na2PCat. The catecholate complex was then filtered to remove all unreacted
sodium metal.
After filtration, the catecholate compound were added to Ni(tren)Cl2 and produced dark
brown-gold reflecting complexation of PCat2- to Ni(tren)2+ and formation of NaCl.

This

compound was then stirred overnight, to ensure complexation of the catecholate to Ni II. The
catecholate complexes were then oxidized by one electron with a methanolic solution of
[Fc](PF6) and produced a dark brown solution. After oxidation, the compound was filtered to
remove any ferrocene produced by the reaction. In order to encourage crystal formation, a 20fold excess of NaBPH4 was added, in MeOH solution, to the filtrates of the two complexes.
In the case of complexes 2 and 3, after sitting overnight small brown-black platelets had
crashed out of solution. These were filtered and washed with MeOH and Et2O. The platelettes
were then dissolved in a minimum amount of MeCN and then 500 mL of degassed Et 2O was
added to crash out the product. 495 mG (54.7%) and 150 mg (20%) of 2 and 3 were collected as
dark brown crystals and microcrystals, respectively. Percent yields were not reported for similar
complexes prepared by Gatteschi15 or Bencini and Dei16

CW-EPR. The Ni-PSQ systems are considered to be in the limit of strong exchange, in
which the exchange coupling parameter is much greater than any of the other spin parameters.
This comes about due to the fact that exchange coupling constants have been unable to be

196

measured by typical means, i.e. variable temperature magnetic susceptibility1,15,16, which means
that the energy of the interaction is greater than magnetometers have been able to detect. In
addition, large J values predicted by DFT1. Typically, molecules with a ground state that does
not support the symmetry of the system must be described in such a manner to include effects
that arise from this mismatch. In particular, the ground state configuration of the Ni-SQ systems,
which have an S = 1 spin center interacting with an S = ½ spin center are described with zerofield splitting effects. Because the ground state is a quartet (variable temperature magnetic
susceptibility1), and the symmetry of the system is nearly Cs (distorted by the ethylene arms of
tren), the degeneracy of the S = 3/2 system must be lifted, in which Kramer’s doublets are
produced (Figure 5).

Figure 4-5. Zero field splitting diagram for a quartet ground state.

197

As stated these are zero field effects and arise in the absence of magnetic field. However,
when placed in a magnetic field, the degeneracy of the Kramer’s doublets are lifted and
transitions between the ms = Âą1/2 and ms = Âą3/2 can occur. Figure 7 shows the CW spectra of 2
and 3. The physical origins of this line shape have been described previsously1. The large
absorptive feature near g-4 has been assigned to transitions within the ms = Âą3/2 manifold, which
have been allowed by the breaking of the degeneracy of the Kramer’s doublets. The feature near
g-2 comes from transition in the ms = Âą1/2 manifold.

Figure 4-6. 7 K CW spectrum and simulation of 2, [Ni(tren)(PSQ)](BPh4). Simulation
parameters: S = [1,1/2], gxx = gyy = 2.08, gzz = 2.44, D = 49,000 MHz (1.63 cm-1), E = 6200 MHz
(0.21 cm-1), electron exchange = -6.1 cm-1 (-182,647.9 MHz), HStrain = (2200,1100,90) MHz,
DStrain = 38,000 MHz, 2,900 MHz. Experimental conditions: 4 K, 9.41 GHz microwave
operating frequency, 0.8 mT modulation amplitude, 4000 G sweep, 260 mT center field and
2048 points.
198

This line shape has been observed in other NiII-radical complexes15–18; however, for some of
these citations, simulation parameters have been listed but the line shape has not been produced
(explicitly, no simulated spectrum was shown in the text), most likely due to the level of difficult
of reproducing the spectrum or there was not even an attempt to simulate the experimental
spectrum. A Ni-Phenoxyl system studied by Wieghardt and coworkers, contained a feature
similar to the one produced by the simulation above18; however the line shape was not able to be
reproduced.

Attempts to reproduce the line shape in this study were focused on the

12

C

isotopologue, seeing as the majority of the line shape between the two spectra are the same.

Figure 4-7.
7 K CW spectra of 2 (black), [Ni(tren)(PSQ)](BPh4), and 3 (blue),
13
[Ni(tren)([ C2]PSQ)](BPh4).
The Hamiltonian used to model these complexes is displayed below:
199

̂+̂
̂ = 𝑔𝑒 𝛽𝑒 𝑩𝑺
𝐻
𝑺∙𝑫∙̂
𝑺+𝑱∙̂
𝑺1 ∙ ̂
𝑺2

(4.2)

The first term is the electronic Zeeman where ge gyromagnetic ratio for an electron, which is
2.0023. The second term is the zero field splitting term and describes the energy of the splitting
of the Kramer’s doublets. D is a diagonal tensor and can be explained mathematically, in terms
of its eigenframe:
1

𝐷𝑥
𝑫=(0
0

0
𝐷𝑦
0

0
0)=
𝐷𝑧

−3𝐷 + 𝐸

0

0

0

− 𝐷−𝐸

0

0

0

(

1
3

2
3

𝐷

(4.3)
)

where D = (3/2)Dz and E = (Dx – Dy)/2, and D is the axial parameter and E is the rhombicity
parameter. It is tradition to represent the D and E parameters in terms of an E/D ratio; however,
in this work, in an effort to maintain clarity between the simulation parameters and the
discussion of the data, D and E will be discussed individually. The last term is the exchange
coupling term where J is exchange coupling constant, typically quantified in cm-1, and S1 and S2
are the individual spin centers.

Again, to be transparent between the simulations and the

discussion J will be reported in both cm-1 and MHz.
The simulation presented above in Figure 6 was accomplished with several parameters.
First, a spin system must be defined, and in this situation there is a ferromagnetically coupled
system where the spin centers are S =1 (octahedral NiII) and S =1/2 (PSQ), and in EasySpin this
must be defined explicitly as [1,1/2] for syntax purposes. In each of the subsequent parameters,
values must be defined for both spin centers for values that arise due to S ˃ 1. Next, the g values
are considered; for this system, an isotropic g can be considered, however this is an axial system
and the g-values are gzz = 2.44, gxx = gyy = 2.08 for S = 1 and an isotropic g = 2.00 for S = 1/2.

200

The zero-field splitting parameters are D = 49,000 MHz (1.63 cm-1) and E = 6,200 MHz (0.21
cm-1) for S =1 and there are none for S = ½. The electron exchange (represented as “ee” in
EasySpin) is next considered, and it should be noted that this is isotropic exchange, and can also
be represented as J with other parameters. For this simulation ee = -218,846 MHz (-7.3 cm-1),
this diverges from experimental measurements and will discussed below.

Lastly the line

broadening parameters are considered. The first considered is DStrain, which EasySpin defines
as “Widths (FWHM) in MHz of the Gaussian distributions of the scalar parameters D and E that
specify the D matrix of the zero-field interaction. If FWHM_E is omitted, it defaults to zero”7.
For this complex, the DStrain and EStrain values are 38,000 MHz and 2,900 MHz, respectively
for S = 1 and 0 for S = 1/2. And finally, HStrain, this line-broadening parameter accounts for
unresolved hyperfine couplings. These values are expressed in terms of their contributions along
the x, y and z axes, and EasySpin defines as “Residual line width (full width at half height,
FWHM), in MHz, describing broadening due to unresolved hyperfine couplings. The three
components are the Gaussian line widths in the x, y and z direction of the molecular frame”7.
The values are HStrainx = 2,200 MHz, HStrainy = 1,100 MHz and HStrainz = 90 MHz.
These simulation parameters do not come without critique, and although the major portions
and features of the line shape of 2 have been reproduced, it has not been fit. The major
contributor to this is the exchange coupling parameter ee, or electron exchange. As stated above
this is the isotropic exchange coupling describing the energy of the coupling between the two
spin centers. There was a major effort to use any/all of the EasySpin tools to correctly describing
the exchange interaction of this molecule, but none quite worked properly. The electronic
structure of the interaction is described in the simulation as a combination of zero-field splitting
and isotropic exchange coupling; however, another way to describe this interaction would be to

201

instead use isotropic and anisotropic exchange. This would be a valid interpretation of the
coupling indeed, due to the fact that the organic radical resides largely in a π-type orbital.
EasySpin does accommodate for systems with anisotropic, or dipolar, exchange with the eeD
parameter, but there was simply no effect to the line shape of the simulation when using this
parameter as outlined in the EasySpin documentation.
Even more obviously, the magnitude of the electron exchange here is -7.3 cm-1 and this is not
consistent with variable temperature magnetic susceptibility measurements that the author of this
work and others have measured for this compound17,16 and similar Ni-SQ complexes15. By these
measurements, temperature dependence was only observed for [Ni(tren)(3,6-(NH2)2-PSQ](BPh4)
by Fehir, and a coupling constant of -109 cm-1 was found. At the very least, an idea of what the
exchange coupling could be is several times larger than what is predicted by the simulation.
Additionally, Gatteschi and coworkers estimated that for [Ni(CTH)DTBSQ]PF6 (where CTH =
dl-5,7,7,12,14,14-hexamethyl-1,4,8,1l-tetraazacyclotetrade and is similar in function to tren, and
DTBSQ = 3,5-di-tert-butylsemiquinone) a lower limit on the energy separation between the
quartet ground state and doublet excited state could be 400 cm-1. It is acknowledged that the
electron

exchange

is

most

likely stronger

in

this

complex

when

compared

to

[Ni(tren)(PSQ)](BPh4) due to the DTBSQ ligand having less of an ability to delocalize the
electron density of the radical; however to calculate this lower limit, the data of the two
complexes is similar enough to use this number as a benchmark. To conclude with this line of
critique, the electron exchange parameter of the simulation in Figure 6 is off by at two orders of
magnitude.
The sharp, downward facing feature in the g-2 region could be a qualitative marker for the
magnitude of the exchange coupling. Fehir1 recorded the EPR spectra of [Ni(tren)(3,6-R2-

202

PSQ)](BPh4) complexes (where R = NH2, H and NO2), and these are featured in the figure
below. As it can be seen from the spectra and the aforementioned exchange coupling constant,
the -NH2 derivative exhibits the lowest magnitude of coupling and has a derivative like feature in
contrast to the features of the -H and -NO2 derivatives. It is possible that this is a physical
observable for the magnitude of the exchange coupling. As the exchange coupling becomes
smaller in magnitude, then the shape of the g-2 feature become more derivative in nature. A
simple study with Easy Spin shows that this can be the case.

Figure 4-8. EPR spectra of [Ni(tren)(3,6-R2-PSQ)](BPh4) complexes where R = H (red, bottom),
NH2 (blue, middle) and NO2 (green, top). Spectra taken from Reference 1.
By taking the same script to simulate complex 2 and modifying the electron exchange
parameter (exchange coupling), the simulation reproduces the same observation that was seen in
Figure 8.

When the exchange coupling is sufficiently small, the feature at g-2 becomes

derivative in nature, and as the exchange coupling is increased, this feature decreases in intensity
and moves away from the derivative shape. Although the values being used are inconsistent with
what has been observed experimentally, the simulation shows similar line shapes to those being
produced by actual compounds. If this parameter is improved upon, then exchange coupling
203

constants could be reliably predicted by simulation software for hybrid metal-organic radical
type exchange coupled molecules.

Additionally, this would replace the need for isotopic

labeling and would demonstrate the effect a specific substituent has on spin polarization. That
being said, isotopic labeling has been quite instrumental in this studying for tracking changes in
molecular structure, based on the common them of phenanthrenesemiquinone.

Figure 4-9. Comparison of various electron exchange (exchange coupling) values for Ni-PSQ
based on simulation parameters for Ni-PSQ.
More could be detailed about the inaccuracies of this simulation but the real hurdle that
should be cleared is development of a sound theoretical approach to simulate molecules of this
type. Much has been accomplished in the field of exchange-coupled transition metal-organic
radical systems, there is however, still room for a more precise understanding of the quantum

204

mechanical underpinnings of these systems that accommodate exchangeable electrons and their
energetics.

ESEEM. In an effort to fully characterize the hyperfine coupling as a function of magnetic
field value, ESEEM spectra were taken at various field points along the CW spectrum. It is
practice in the area of ESEEM where S ˃ ½ to study the hyperfine interaction at not just g-2 but
at several points in the spectrum. ESEEM spectra were taken at 334, 260 and 180 mT for both 2
and 3. The presented spectra are divided spectra of the

13

C and

12

C spectra, utilizing the same

Mims’ division technique of Chapters 2 and 3. Time domain spectra were normalized with a biexponential prior to division, and then worked up with the usual procedure as outlined in the
experimental section of this chapter.
In Figure 8 is displayed the ESEEM Fourier transform of Ni-[13C2]PSQ/NiPSQ taken at
334 mT with τ =140 ns and its simulation. The corresponding divided time domain spectrum
and simulation are displayed in Figure 9. It is immediately apparent that there exists some
disagreement in the time domain simulation with the Fourier transform simulation. At this point
it is imperative to explain the approach used to model these complexes.
The EasySpin program used to simulate ESEEM spectra is Saffron and Saffron only
supports systems with one spin center; seeing as there are two spin centers in the complexes
being studied, this is the first fork in the road to be overcome. The most obvious way to continue
forward would be to simply model the system as a quartet, S = 3/2, and move on. This was the
first approach to fitting these spectra. The model spin system would include the same parameters
as the CW model, excluding electron exchange and replacing the spin-coupled system with a
single spin center of S = 3/2. This approach was attempted several times over but a good
agreement amongst the hyperfine parameters could not be accounted for. Resonances could be
205

produced, but the peak placement was askew and this was not the case when S = ½.

Thus, it

was decided to move forward using a spin system consisting of a S = ½ spin center, with the
axial g values and HStrain line broadening parameters of the CW simulation. As it will be seen,
the majority of the discussion on hyperfine parameters stems from analysis of the ESEEM
collected at 334 mT, which is located in the g-2 area of spectrum, where organic radicals are
observed. Since it is the prerogative of this work to study the hyperfine interaction at the
carbonyl carbon nuclei on the organic radical portion of the complex, it was decided that there is
validity to using this approach.
Another layer of complexity was owed from the fact that no hyperfine coupling
parameters were afforded from the CW simulations. In Chapters 2 and 3 the approach for
simulating ESEEM spectra was eased by knowing

13

C-aiso constants from the CW simulations.

This provided a template of only four unknowns, the two dipolar constants and two rhombicity
constants. The rhombicity values were held at zero until acceptable dipolar values were reached,
and lastly the rhombicity values were varied between 0 – 1 until a good fit had been achieved.
With the ESEEM simulations of this chapter, however, the entirety of the hyperfine parameters
was unknown, making this task of simulating the spectra much more arduous than in previous
chapters. The hyperfine parameters were unable to be determined from the CW spectra due to
the exchange coupling and zero field splitting; these values were much greater then menial
hyperfine coupling constants and thus washed out any fine structure that would have been
otherwise observed.

It would be possible to study the CW spectra at higher microwave

frequencies, such as W-band (~94 GHz), which would provide for higher resolution and the
possibility of observing hyperfine coupling. There was also an effort to uncover the isotropic

206

coupling constants using Davies’ ENDOR experiments, but the results of 2 matched those of 3;
the spectra were nearly identical.

Figure 4-10. 13C/12C NiPSQ ESEEM Fourier transform (black) and fit (red). Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1: aiso = -2.11
MHz, T = 3.29 MHz ρ = 0.433 MHz; C2: aiso = 1.58 MHz, T = 6.32 MHz, ρ = 0.03 MHz.
Experimental conditions: 334 mT magnetic field, 9.71 GHz microwave frequency, 4 K, 3 pulse
ESEEM sequence, 140 ns τ, 40 ns acquisition trigger and 12 ns increments.
In continuing with the idea of discrepancy between time and frequency domain spectra, it
was acknowledged early on in the simulation process that this disparity would ultimately exist in
one form or another. And although the time domain data are the raw data and the frequency
domain spectra are but the result of mathematical treatment of the time domain data followed by
207

Fourier transformation, in collaboration with simulating the HYSCORE spectra, it was decided
to use hyperfine values that were more consistent with the Fourier transformations of the time
domain data. Simulations that favor the time domain spectra will also be presented, but only in
brief.

Figure 4-11. 13C/12C NiPSQ ESEEM Fourier transform (black) and fit (red). Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1: aiso = -2.11
MHz, T = 3.29 MHz ρ = 0.433 MHz; C2: aiso = 1.58 MHz, T = 6.32 MHz, ρ = 0.03 MHz.
Experimental conditions: 334 mT magnetic field, 9.71 GHz microwave frequency, 4 K, 3 pulse
ESEEM sequence, 140 ns τ, 40 ns acquisition trigger and 12 ns increments.
The spectrum was simulated by treating the carbonyl carbon nuclei in an asymmetric
manner. Prior to this treatment, the nuclei were treated equivalently, as if the unpaired electron
density were delocalized evenly between the two. As it was stated above, the spectra were also
simulated with the same g and HStrain values as the CW spectrum. The final hyperfine values
208

were achieved by varying the isotropic and dipolar values in both the time and frequency domain
spectra until all the major features were represented. This process was cumbersome and time consuming. The Matlab function fminsearch was then used to find global minima for these
values. Once the isotropic and dipolar constants were settled, the rhombicity values were varied
to achieve the lowest residual value possible.
The values achieved from the simulations are as follows: C1-aiso = -2.11 MHz, C1-T = 3.29
MHz, C1-ρ = 0.433 MHz; C2-aiso = 1.58 MHz, C2-T = 6.32 MHz, C2-ρ = 0.03 MHz. Near g-2,
the simulated spectrum shows a sharp feature at 6.07, followed by small humps at 11.93 and 17.7
MHz, none of which directly correspond to nuclear Larmor frequencies. At 334 mT. 13Cυ = 3.57
MHz, however owing to the fact that this is a divided spectrum, this large feature is most likely
due to a large dipolar coupling. Table 3 (see below) shows the hyperfine parameters for this
spectrum. These hyperfine parameters show that upon complexation to a paramagnetic metal,
the spin density does shift. Recall from Chapter 2 that for the ion-pair, the carbonyl carbon
nuclei had hyperfine values of aiso = -1.57 MHz and T = 13.57 MHz. Comparing these two sets
of results shows that there is little change in isotropic coupling constant, but a large change in
dipolar coupling. Including the results of Chapter 3 where the isotropic and dipolar hyperfine
coupling constants of [Zn(tren)([13C2]PSQ)](BPh4) were in the range of 1 – 3 MHz. Perhaps it is
the situation that the isotropic coupling is rather unperturbed by complexation, and the dipolar
coupling is more delocalized across the carbonyl bond, certainly this is within the realm of
possibilities. It would be beneficial to study more carbon positions within the phenanthrene ring
system as well as the oxygen nuclei, across the series of compounds studied in this dissertation.
Indeed, by theory the majority of the radical spin density resides on the oxygen nuclei.

209

With regards to the theoretical interpretation of the electronic structure, DFT predicts the
following hyperfine parameters: C1-aiso = 1.147 MHz, C1-T = 19.31 MHz, C1-ρ = 0.08 MHz;
C2-aiso = -1.634 MHz, C2-T = 13.96 MHz, C2-ρ = 0.12 MHz. These values relate well with the
hyperfine parameters of the ion-pairs of Chapter 2, but do not predict well the results of the
simulations in this chapter. It is worth noting that the predicted signs of the hyperfine coupling
are in agreement with simulated values. The computational program ORCA (developed by
Frank Neese) was also consulted to attempt to study the spin properties however the results were
similar to those produced by Gaussian03 and Gaussian09.

Table 4-1. Hyperfine coupling parameters obtained from Ni-[13C2]PSQ/NiPSQ ESEEM.
aiso/MHz

T/MHz

ρ/MHz

Carbon 1

-2.11

3.29

0.433

Carbon 2

1.58

6.32

0.03

It was mentioned at the beginning of this section that a secondary set of hyperfine
parameters were found and would be briefly covered. Below, Figures 12 and 13 show the
parameter set for values optimized for the time domain data. At first inspection, these coupling
constants describe the time data more accurately than the values above; however, after
scrutinizing the Fourier transform spectrum, it is noticed that only the resonance at 6.07 MHz is
well represented.

Again, after comparing the two parameter sets with both ESEEM and

HYSCORE experimental spectra, it was determined that the first set was a better representation
of the spectra than the second set.

210

Figure 4-12. 13C/12C NiPSQ ESEEM Fourier transform (black) and fit (red). Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1: aiso = -4.09
MHz, T = 1.535 MHz ρ = 0.0 MHz; C2: aiso = 1.07 MHz, T = 6.36 MHz, ρ = 0.19 MHz.
Experimental conditions: 334 mT magnetic field, 9.71 GHz microwave frequency, 4 K, 3 pulse
ESEEM sequence, 140 ns τ, 40 ns acquisition trigger and 12 ns increments.

211

Figure 4-13. 13C/12C NiPSQ ESEEM Fourier transform (black) and fit (red). Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1: aiso = -4.09
MHz, T = 1.535 MHz ρ = 0.0 MHz; C2: aiso = 1.07 MHz, T = 6.36 MHz, ρ = 0.19 MHz.
Experimental conditions: 334 mT magnetic field, 9.71 GHz microwave frequency, 4 K, 3 pulse
ESEEM sequence, 140 ns τ, 40 ns acquisition trigger and 12 ns increments.
Stacked ESEEM spectra of the

13

C/12C NiPSQ spectra are seen in Figure 14. It is utterly

apparent from this figure that the bulk effect of the carbonyl carbon nuclei occurs at 334 mT,
near g-2 for these molecules. The g-2 region is where organic radicals typically experience
resonance with the scanning magnetic field in a bath of microwave energy. So the fact that the
13

C effect is highest at g-2 is not a surprise, for the portion of the radical wavefunction giving

rise to this resonance is associated, solely, with the semiquinone ligand.
It is noted here that these data are atypical of hyperfine effects when examining the hyperfine
parameters across the spectrum of system with S ˃ 1. It is typical for the hyperfine coupling
212

constants to increase in intensity moving from high magnetic field values to lower magnetic field
values. The opposite trend is observed in the stacked spectrum; EasySpin predicts an increase in
the echo amplitude as well. This can be seen in Figure 13, which is a Fourier transform spectrum
of the data collected at 260 mT.

Figure 4-14. Overlaid ESEEM spectra of
180 mT (blue).

13

C/12C Ni-PSQ: 334 mT (black), 260 mT (red) and

213

Figure 4-15. 13C/12C NiPSQ ESEEM Fourier transform (black) and fit (red). Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1: aiso = -2.11
MHz, T = 3.29 MHz ρ = 0.433 MHz; C2: aiso = 1.58 MHz, T = 6.32 MHz, ρ = 0.03 MHz.
Experimental conditions: 260 mT magnetic field, 9.71 GHz microwave frequency, 4 K, 3 pulse
ESEEM sequence, 92 ns τ, 40 ns acquisition trigger and 12 ns increments.
Before moving on to the HYSCORE section, it is worth mentioning that this type of
study in nearly unprecedented. Hybrid metal-organic radical complexes are not well studied
within the areas of ESEEM and HYSCORE, especially of the type the transition metalsemiquinone type, undergoing exchange coupling.

Additionally, synthetic transition metal

complexes are not well studied either, the ESEEM and HYSCORE techniques have been
traditionally been used to aid in the structural characterization of biological molecules, as well as
biological reactions.
214

HYSCORE. HYSCORE data were collected at 4 K and are presented respective to their
compounds, and not as divided spectra. Spectra presented are from data collected at 334 mT
(140 ns τ) and 260 mT (92 ns τ) for 2 and 3. Two quadrants in each spectra are shown; quadrant
1 (Q1) on the right and quadrant 2 (Q2) on the left. The other two quadrants (not displayed) are
reflections of quadrants 1 and 2 and are a consequence of the two dimensional Fourier
transformation (2D-FT). The z-scale is presented colorimetrically, with yellow being the most
intense and dark blue being the least intense. The spectra were simulated with the same spin
system and hyperfine parameters as the ESEEM spectra above.
The purpose of this study is twofold. Primarily, it would be advantageous to know what
other components of the spin system are operative in the presence and absence of 13C-labelling.
the identity of the peak at 6 MHz in each system would be desirable to know, since it does not
directly correspond to any of the Larmor frequencies available in the molecule. Secondly,
Figures 16 and 17 correspond to 3 and 2 HYSCORE spectra at 334 mT, respectively.
The

13

C spectrum will be assessed first, followed by the

12

C spectra.

The large feature at 6.1

MHz that lies on the diagonal in Q1 is due to the ridges at (1.6,3.1) and (6.1,1.6), centered about
the on diagonal peak 3.7 MHz, which corresponds to the

13

Cυ at 334 mT (3.69 MHz). Also

contributing to the peak at 6.1 MHz would be the smaller peaks in the same ridge at (2.8,3.1) as
well as (5.6,2.0). The side ridges centered about the 1Hυ at 14.2 MHz are most likely from tren
protons and arise as a consequence of spin polarization through the sigma bonding system of the
Ni(tren) fragment. In addition to this, these side ridges centered about 1Hυ are present in both
spectra.

215

Figure 4-16. HYSCORE 2D-FT spectrum of 3 acquired at 4 K. Experimental conditions: 334
mT magnetic field, 9.71 GHz microwave frequency, 4 K, 4 pulse HYSCORE sequence, 140 ns τ,
40 ns acquisition trigger (both dimensions) 16 ns increments and 256 x 256 points.

216

Figure 4-17. HYSCORE 2D-FT spectrum of 2 acquired at 4 K. Experimental conditions: 334
mT magnetic field, 9.71 GHz microwave frequency, 4 K, 4 pulse HYSCORE sequence, 140 ns τ,
40 ns acquisition trigger (both dimensions) 16 ns increments and 256 x 256 points.
If hyperfine coupling from spin polarization to protons is observed, then it is not a stretch
for 14N hyperfine interactions to be present as well. Indeed, in Q2 ridges centered about the third
harmonic of the

14

Nυ frequency (1.02 MHz) at -3.2 MHz could be due to

14

N hyperfine, but in

order to test this 15N-labelling could be employed. By installing I = ½ nuclei for I = 1 nuclei, this
could cause the proton and nitrogen atoms to relax faster by coupling and a decrease in signal
might be observed. It would also be advantageous to create a per-methylated tren analogue of
these compounds to witness a loss in signal from proton couplings (centered about 14 MHz).

217

Within the

12

C spectrum, the features responsible from tren are observed in Q1 around

14.1 MHz. The on-diagonal peak at 3.67 MHz arises from small resonances in the ESEEM 1D
spectrum, most likely from natural abundance of 13C in the molecule. The main purpose of this
spectrum, however, is to serve as a comparison to the

13

C spectrum, and it easily seen that the

effects induced by the 13C spin label are not present and other contributors from both spectra are
13

present and some even more intense than in the

C spectrum. This illustrates the strength of

using 13C as an isotopic label for tracking changes in hyperfine coupling in these molecules.

Figure 4-18. HYSCORE 2D-FT simulation (red) and spectrum of 3 acquired at 4 K. Simulation
parameters: S = ½, gxx = gyy = 2.08, gzz = 2.44, HStrain = (2200,1100,90) MHz, C1: aiso = -2.11
MHz, T = 3.29 MHz ρ = 0.433 MHz; C2: aiso = 1.58 MHz, T = 6.32 MHz, ρ = 0.03 MHz.
Experimental conditions: 334 mT magnetic field, 9.71 GHz microwave frequency, 4 K, 4 pulse
HYSCORE sequence, 140 ns τ, 40 ns acquisition trigger (both dimensions) 16 ns increments and
256 x 256 points.
218

In the above figure, the simulation for the spectrum taken 334 mT is shown in red, plotted
over the experimental 2D-FT. The simulation does quite well in representing most of the
features stemming from the carbon-13 hyperfine, especially in Q1. The major feature in Q1
untouched by the simulation is the on-diagonal peak at 11.47 MHz, which is not even a harmonic
of 13Cυ. In the strong coupling regime, in Q2, the simulation covers the ridges centered about the
peak at (-5.61,5.61). These ridges are a bit atypical in their form in that strong coupling in Q2
usually yields peaks (or ridges) that are parallel to the diagonal line. The Q1 ridges are more
reminiscent of typical HYSCORE spectra where the ridges/resonances are perpendicular to the
diagonal19. The other off-diagonal resonances in Q2 at (-9.52, 5.86) and (-5.86,9.52) are also
present in the NiPSQ 2D-FT spectrum.
Below are the HYSCORE 260 mT spectra, for complexes 3 and 2, respectively. Effects in
Q1, relative to the 334 mT spectrum, that are present, are the tren features, along with baseline
effects likely from the 13C label. The effects from 13C are minimalized, or much less intense, at
this magnetic field value, as was seen in the ESEEM spectrum at the same magnetic field value.
The decrease in echo amplitude (intensity) could be due to the occurrence that the magnetic field
axis is no longer parallel to the magnetic axis of the

13

C coupling at g 2, and therefore can no

longer detect a response from being off resonance with the ligand-centralized radical.
The most prominent new feature in this spectrum, relative to the 334 mT spectrum, are the
more intense features in Q2, centered about -5.2 MHz. Due to the absence of these features from
the

12

C spectrum at 260 mT, it is assumed that this is an effect of the

13

C hyperfine coupling.

Also, in Q1, not only has the intensity (z-axis) dwindled in the 0 – 7 MHz range, but also it
appears that the proton HYSCORE has expanded. As for spectra collected at magnetic field
values lower than 260 mT, the HYSCORE effect becomes dominated by 1H hyperfine, while 13C

219

HYSCORE has completely vanished. It is fair to conclude that the

13

C HYSCORE is most

prominent at g-2 due to the nature of the delocalized electron density on the carbonyl carbon
nuclei.

Figure 4-19. HYSCORE 2D-FT spectrum of 3 acquired at 4 K. Experimental conditions: 260
mT magnetic field, 9.71 GHz microwave frequency, 4 K, 4 pulse HYSCORE sequence, 92 ns τ,
40 ns acquisition trigger (both dimensions) 16 ns increments and 256 x 256 points.

220

Figure 4-20. HYSCORE 2D-FT spectrum of 2 acquired at 4 K. Experimental conditions: 260
mT magnetic field, 9.71 GHz microwave frequency, 4 K, 4 pulse HYSCORE sequence, 92 ns τ,
40 ns acquisition trigger (both dimensions) 16 ns increments and 256 x 256 points.
The HYSCORE experiments found that the

13

C hyperfine coupling was vast and rich in

both the weak coupling (Q1) and strong coupling (Q2) regimes. This notion is also supported by
the ESEEM results; however, due to the second dimension, the HYSCORE spectra not only
bring more information to the table regarding hyperfine sublevel structure, but they also bring an
aesthetic quality that draws attention to these more subversive details.
invaluable in showing the great disparity between 2 and 3 as far as

13

These spectra are

C hyperfine is concerned

and also in displaying the common hyperfine elements active in both complexes.

221

4.4 Conclusions and Future Work
The results and observations produced in this Chapter have possibly raised more
questions than answers with regards to tracking hyperfine coupling in transition metal-organic
radical complexes that experience an electron exchange interaction. The original idea behind
this research was to use an isotopic label to observe changed in hyperfine coupling between
various derivatives of the type [Ni(tren)(3,6-R2-PSQ)]+. There have been other derivatives
synthesized and characterized, and subjected to a treatment of cwEPR, ESEEM and HYSCORE
measurements; however, the difficulty of modeling the spin system of these molecules has
proved to be a task in and of itself. Efforts are ongoing to improve the accuracy of modeling
these types of complexes in various forms.
The CW simulation of 2 does show both promise and concern.

While unable to

accurately represent the correct magnitude of exchange coupling, the line shape was able to be
reproduced with decent accuracy. And, it was shown that the feature at g-2 is a possible marker
for tracking the magnitude of exchange coupling. This will be promising to follow up on in
future studies of molecules of this type, as well as tackle the theoretical efforts to accurately
simulate the magnitude of the exchange coupling.
Interpretation of the ESEEM spectra showed that the data can actually be modeled well
as a doublet system. The hyperfine coupling parameters discovered also showed that, like the
[Zn(tren)([13C2]PSQ)](BPh4), the carbonyl carbon nuclei have a disproportionate amount of
delocalized electron density. This is not surprising given that both complexes are within the C s
point group, and can even be approximated to be C2v. It might be interesting to use a symmetric
ancillary ligand or ligand system to occupy coordination sites and study the hyperfine coupling
parameters to asses as to whether the delocalized electron density on the ligand can be controlled

222

by the ligand, or it is merely a consequence of the bonding in a metal semiquinone complex.
Examining the oxygen-metal bond lengths in such complexes is a means of characterizing the
oxidation state of the semiquinone moiety. Catecholate ligands typically have symmetric bond
lengths while semiquinones have slight discrepancies in the oxygen-metal bond lengths.20
Finally, HYSCORE analysis of 2 and 3 yielded the potential for a great tool for
characterizing differences in electronic structure of metal-radical complexes and their
isotopologues.

Also, HYSCORE could construed to be a more powerful technique for

understanding hyperfine couplings that escape ESEEM analysis.

The extra dimension is

invaluable for studying the fine structure of open shell molecules. To state again, this research
has been unique in that synthetic, exchange-coupled transition-metal complexes and their
isotopologues are not usually studied with this level scrutiny. It is the hope of this author that
HYSCORE and ESEEM analysis would be applied to more types of open-shell inorganic
compounds in order for the field to grow. Currently, a majority of the research in the field
resides in the biological realm, which does contain a good amount of inorganic chemistry, but it
would be desirable for field and researchers if more studies would take advantage of these
methods to further understand delocalization and its effects with regards to chemical reactivity
and potential applications in technology.

223

APPENDIX

224

APPENDIX
CW-EPR
function res = LDM1056cw_fit(hpar)
%LDM1056 Fit - McCracken & Stoll
% Load the Spectrum
[x,Edata,params]=eprload('LDM1052sh2CW.DTA');
Edata=Edata-Edata(1);
Edata = Edata/max(Edata);
Bdata=x/10.0;
lmin = 1;
lmax = 2048;
% Experimental Parameters
Exp.mwFreq = params.MWFQ/1.00e+09;
Exp.nPoints = params.XPTS;
Exp.CenterSweep=[params.A1CT*1000.,params.A1SW*1000.];
Exp.ModAmp = params.B0MA*1000.;
Exp.Temperature = 7.0;
Opt.nKnots =[91 4];
Opt.Output = 'summed';
% Simulation Parameters
Sys.S = [1.0 0.5];
Sys.D = [49000 6200;0 0];
Sys.ee = -6.1*29979;
Sys.g = [2.44 2.08 2.08;2.00 2.00 2.00];
Sys.HStrain = [2200 1100 90];
Sys.DStrain = [38000 2900;0 0];
[B,spec,Trans] = pepper(Sys,Exp,Opt);
spec = spec/max(spec); % Normalize the simulation
res = 0.0;
for ipt = lmin:lmax,
res = res + (spec(ipt) - Edata(ipt))^2;
end;
plot(B(lmin:lmax),spec(lmin:lmax),'r',Bdata(lmin:lmax),Edata(lmin:lmax),'k');
return;

225

ESEEM
function res = LDM1154_ESEEM_334mT_140tau_Doublet_Spec_fit(hpar)
% LDM1154, Ni-[13C2]PSQ ESEEM FFT Sim/Fit
% Doublet - Saffron does NOT support systems w/ more than one spin
%==========================================================================
% Load the experimental spectrum
[freq,intens] = eprload('ldm1154div1052_334mT140sp.DTA');
freq = freq*1000;
lmin = 1;
lmax = 256;
% Simulation Parameters
Sys.S = 1/2; % Actually [1 1/2]
Sys.g = [2.44 2.08 2.08];
Sys.HStrain = [2200 1000 90];
Sys.Nucs = '13C,13C';
Sys.n =[1 1];
Sys.A_ = [-2.11 3.288 0.433; 1.579 6.319 0.03]; % [a-iso T rhombicity]
% Load the experimental Parameters
Exp.Field = 334; % Magnetic Field/millitesla
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012; % Time Increment/microseconds
Exp.tau = 0.14; % Tau/microseconds
Exp.T = 0.04; % Starting T Value/microseconds
Opt.nKnots = 61;
[x,y] = saffron(Sys,Exp,Opt);
ESnormDAL; %Normalizes the simulation
res = 0;
for ipt = lmin:lmax,
res = res + (spec(ipt) - intens(ipt))^2;
end;
plot(f(lmin:lmax),spec(lmin:lmax),' r',freq(lmin:lmax),intens(lmin:lmax),'
k');
return;

226

function res = LDM1154_ESEEM_334mT_140tau_Doublet_Time_fit(hpar)
% LDM1154, Ni-[13C2]PSQ ESEEM Time Sim/Fit
% Quartet - Saffron does NOT support systems w/ more than one spin
%==========================================================================
% Load the experimental spectrum
[time,amp] = eprload('ldm1154div1052_334mT140.DTA');
Amp = real(amp)/max(real(amp)); % Normalizes the data
time = (time/1000)+0.14;
lmin = 1;
lmax = 256;
% Simulation Parameters
Sys.S = 1/2; % Actually [1 1/2]
Sys.g = [2.44 2.08 2.08];
Sys.HStrain = [2200 1000 90];
Sys.Nucs = '13C,13C';
Sys.n =[1 1];
Sys.A_ = [-2.11 3.288 0.433; 1.579 6.319 0.03]; % [a-iso T rhombicity]
% Load the experimental Parameters
Exp.Field = 334; % Magnetic Field/millitesla
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012; % Time Increment/microseconds
Exp.tau = 0.14; % Tau/microseconds
Exp.T = 0.04; % Starting T Value/microseconds
[x,y] = saffron(Sys,Exp);
Y = real(y)/max(real(y)); % Normalizes the simulation
res = 0;
for ipt = lmin:lmax,
res = res + (Y(ipt) - Amp(ipt))^2;
end;
plot(x(lmin:lmax),Y(lmin:lmax),' r',time(lmin:lmax),Amp(lmin:lmax),' k');
return;

227

function res = LDM1154_ESEEM_260mT_92tau_Doublet_Spec_fit(hpar)
% LDM1154, Ni-[13C2]PSQ ESEEM Time Sim/Fit 260 mT, 92 ns tau
% Doublet - Saffron does NOT support systems w/ more than one spin
%==========================================================================
% Load the experimental spectrum
[freq,intens] = eprload('ldm1154div1052_260mT92sp.DTA');
freq = freq*1000;
lmin = 11;
lmax = 256;
% Simulation Parameters
Sys.S = 1/2; % Actually [1 1/2]
Sys.g = [2.44 2.08 2.08];
Sys.HStrain = [2200 1000 90];
Sys.Nucs = '13C,13C';
Sys.n =[1 1];
Sys.A_ = [-2.11 3.288 0.433; 1.579 6.319 0.03]; % [a-iso T rhombicity]
% Load the experimental Parameters
Exp.Field = 260; % Magnetic Field/millitesla
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012; % Time Increment/microseconds
Exp.tau = 0.092; % Tau/microseconds
Exp.T = 0.04; % Starting T Value/microseconds
Opt.nKnots = 51;
[x,y] = saffron(Sys,Exp,Opt);
ESnormDAL; %Normalizes the simulation
res = 0;
for ipt = lmin:lmax,
res = res + (spec(ipt) - intens(ipt))^2;
end;
plot(f(lmin:lmax),spec(lmin:lmax),' r',freq(lmin:lmax),intens(lmin:lmax),'
k');
return;

228

function res = LDM1154_ESEEM_260mT_92tau_Doublet_Time_fit(hpar)
% LDM1154, Ni-[13C2]PSQ ESEEM Time Sim/Fit 260 mT, 92 ns tau
% Doublet - Saffron does NOT support systems w/ more than one spin
%==========================================================================
% Load the experimental spectrum
[time,amp] = eprload('ldm1154div1052_260mT92.DTA');
Amp = real(amp)/max(real(amp)); % Normalizes the data
time = (time/1000) + 0.092;
lmin = 1;
lmax = 256;
% Simulation Parameters
Sys.S = 1/2; % Actually [1 1/2]
Sys.g = [2.44 2.08 2.08];
Sys.HStrain = [2200 1000 90];
Sys.Nucs = '13C,13C';
Sys.n =[1 1];
Sys.A_ = [-2.11 3.288 0.433; 1.579 6.319 0.03]; % [a-iso T rhombicity]
% Load the experimental Parameters
Exp.Field = 260; % Magnetic Field/millitesla
Exp.Sequence = '3pESEEM';
Exp.dt = 0.012; % Time Increment/microseconds
Exp.tau = 0.092; % Tau/microseconds
Exp.T = 0.04; % Starting T Value/microseconds
Opt.nKnots = 51;
[x,y] = saffron(Sys,Exp,Opt);
Y = real(y)/max(real(y)); % Normalizes the simulation
res = 0;
for ipt = lmin:lmax,
res = res + (Y(ipt) - Amp(ipt))^2;
end;
plot(x(lmin:lmax),Y(lmin:lmax),' r',time(lmin:lmax),Amp(lmin:lmax),' k');
return;

229

HYSCORE
% LDM1154, Ni-[13C2]PSQ HYSCORE 2D-FT Simulation
% Quartet - Saffron does NOT support systems w/ more than one spin
%==========================================================================
clear;
Sys.S = 1/2; % Actually [1 1/2]
Sys.g = [2.44 2.08 2.08];
Sys.HStrain = [2200 1000 90];
Sys.Nucs = '13C,13C';
Sys.n =[1 1];
Sys.A_ = [-2.11 3.288 0.433; 1.579 6.319 0.03];% [a-iso T rhombicity]
% Load the experimental Parameters
Exp.Field = 334; % Magnetic Field/millitesla
Exp.Sequence = 'HYSCORE';
Exp.dt = 0.016; % Time Increment/microseconds
Exp.tau = 0.14; % Tau/microseconds
Exp.t1 = 0.04; % Starting T Value/microseconds
Exp.t2 = 0.04;
Exp.nPoints = [256 256];
Opt.nKnots = 181;
[x1,x2,y,p1] = saffron(Sys,Exp,Opt);
% Plot the data
f1 = p1.f1;
f2 = p1.f2;
Z = real(p1.fd);
ZR = zeros(512,512);
% Set threshhold for contour at 10% max amplitude
Zthresh = 0.01*max(max(Z));
for iy=1:512,
for ix = 1:512,
if Z(ix,iy) < Zthresh,
ZR(ix,iy) = Zthresh;
else
ZR(ix,iy) = Z(ix,iy);
end
end;
end;
f1min=-15.0;
f1max=15.0;
f2min=0.0;
f2max=15.0;
frinc=f1(2)-f1(1);

230

xmin=floor((f1min-f1(1))/frinc)+1;
xmax=floor((f1max-f1(1))/frinc)+1;
ymin=floor((f2min-f2(1))/frinc)+1;
ymax=floor((f2max-f2(1))/frinc)+1;
contour(f1(xmin:xmax),f2(ymin:ymax),ZR(ymin:ymax,xmin:xmax),20,'r');
xlabel('f1 (MHz)');
ylabel('f2 (MHz)');
grid;
return;

231

REFERENCES

232

REFERENCES

(1)

Fehir, R. J. Differential Polarization of Spin and Charge Density in Phenoxy Radicals,
Semiquinone Radicals, and Ni(II) and Cr(III) Phenanthrenesemiquinone Complexes,
Michigan State University, 2009.

(2)

Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J.
R.; Montgomery, Jr., J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.;
Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.;
Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa,
J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J.
E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.;
Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.;
Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V.
G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;
Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.;
Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin,
R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.;
Pople, J. A. Gaussian, Inc.: Wallingford, CT 2004.

(3)

2003, 2003.

(4)

Reed, A. E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys. 1985, 83 (2), 735.

(5)

Nguyen, M. J. Phys. Chem. A 1997, 5639 (96), 3174–3181.

(6)

Deligiannakis, Y. Coord. Chem. Rev. 2000, 204 (1), 1–112.

(7)

Stoll, S.; Schweiger, A. J. Magn. Reson. 2006, 178 (1), 42–55.

(8)

Stoll, S.; Britt, R. D. Phys. Chem. Chem. Phys. 2009, 11 (31), 6614.

(9)

Stoll, S. In Handbook of Multifrequency Electron Paramagnetic Resonance: Data and
Techniques; Misra, S. K., Ed.; Wiley-VCH, 2014; pp 69–138.

(10)

Fauth, J. J. Magn. Reson. 1986, 66 (1), 74–85.

(11)

Warncke, K.; McCracken, J. J. Chem. Phys. 1994, 101 (3), 1832.

(12)

Gemperle, C.; Aebli, G.; Schweiger, A.; Ernst, R. R. J. Magn. Reson. 1990, 88, 241–256.

(13)

https://www.mathworks.com/help/matlab/ref/fminsearch.html?s_tid=gn_loc_drop
https://www.mathworks.com/help/matlab/ref/fminsearch.html?s_tid=gn_loc_drop.

(14)

Ochs, C.; Hahn, F. E.; Lügger, T. Eur. J. Inorg. Chem. 2001, 2001 (5), 1279–1285.

(15)

Benelli, C.; Dei, A.; Gatteschi, D.; Pardi, L. Inorg. Chem. 1988, 27 (16), 2831–2836.

(16)

Bencini, A.; Carbonera, C.; Vaz, M. G. F. Dalt. Trans. 2003, 1701–1706.

(17)

Fehir, R. J. DIFFERENTIAL POLARIZATION OF SPIN AND CHARGE DENSITY IN
233

PHENOXY RADICALS, SEMIQUINONE RADICALS, AND NICKEL(II) AND
CHROMIUM(III) PHENANTHRENESEMIQUINONE COMPLEXES, 2009.
(18)

MĂźller, J.; Kikuchi, A.; Bill, E.; WeyhermĂźller, T.; Hildebrandt, P.; Ould-Moussa, L.;
Wieghardt, K. Inorganica Chim. Acta 2000, 297 (1–2), 265–277.

(19)

Mccracken, J. In Encyclopedia of Inorganic Chemistry; 1994.

(20)

Wheeler, D. E.; McCusker, J. K. Inorg. Chem. 1998, 37 (9), 2296–2307.

234

Chapter 5. Conclusions and Future Directions

5.1 Concluding Remarks

The goals of this research project entailed assessing whether or not a substituent effect
was active in derivatives of 3,6-R2-PSQs by investigating carbonyl carbon hyperfine interactions
with CW and pulsed EPR techniques. In short, a substituent effect was observed, but only where
R = H and OMe and only for organic radicals. The underpinnings of this research project
necessitate for many derivatives to be synthesized, and not just organic radicals, but also for the
Zn, Ga, and Ni complexes. Although this original goal was not achieved on a grand scale, what
was achieved was an understanding of hyperfine fluctuations at the carbonyl carbon nuclei in
PSQs when bound to Na+, Zn(tren)2+, Ga(tren)2+ and Ni(tren)2+. This does not qualify as a
substituent effect, but what it does qualify as, is a study that shows the influence of Lewis acidity
and exchange coupling on hyperfine coupling parameters.
Building on this notion, Table 1 shows the hyperfine parameters of all monosemiquinone complexes, excluding [Ga(tren)(PSQ)](BPh4)2 (no spin echo was observed). As an
overarching theme, all isotropic coupling constants were on the scale of 0-3 MHz, regardless of
sign. It is most likely the case that these constants are small because the radicals being studied
are of aromatic nature and therefore the majority of the unpaired spin density is located in π –
type orbitals. These findings are actually consistent with what has been found for the carbonyl
carbons in ubisemiquinol oxidase. MacMillan and coworkers found carbonyl carbon aiso in the
range of 0-1 MHz1.

These were organic radicals and the carbonyl oxygen nuclei were

undergoing hydrogen bonding with nearby amino acid residues.

235

For the metal complexes

however, there aren’t many examples of this kind of study in the literature and therefore real life
comparisons are not easily found. Although, these results could be compared with theoretical
results ad nauseam; however, this was not the point of this study.
Additionally, a comment is to be made for the low magnitude of these isotropic hyperfine
interactions. It is thought that a majority of the spin density is contained within the carbonyl
oxygen nuclei of semiquinones, and thus it is not a gross estimation that there could be a high
concentration of spin density at the carbonyl carbon nuclei as well. This can be a decent
estimation, but one should be careful in estimation; the isotropic values fond in this work were
less than some 1H-aiso values. Again, due to the nature of these radicals, residing in π – type
orbitals, a discrepancy should be noted that the isotropic and dipolar coupling constants do not
necessarily track each other in magnitude. The results of the NaPSQ and Na(OMe) 2PSQ make
this quite obvious.
Table 5-1. 13C Hyperfine parameters, in MHz, for NaPSQ Ion-Pair, Na(OMe)2PSQ Ion-Pair ZnPSQ and Ni-PSQ.
NaPSQ

Na(OMe)2PSQ

Zn-PSQ

Ni-PSQ

aiso, T, ρ

aiso, T, ρ

aiso, T, ρ

aiso, T, ρ

Carbon 1

-1.57, 13.57, 0.44

-0.41, 16.13, 0.63

1.08, 2.84, 0

-2.11, 3.29, 0.43

Carbon 2

-1.57, 13.57, 0.44

-0.41, 16.13, 0.63

2.78, 2.46, 0

1.58, 6.32, 0.03

What is more interesting though, is the change in magnitude of the dipolar coupling across
the series. The dipolar values are highest for the organic radicals, by an order of magnitude. The
hyperfine interaction is quite contained within the semiquinone moiety, and has “nowhere to go”
in comparison to the metal complexes. The dramatic decrease in magnitude of dipolar coupling
from organic radical to metal complex could be explained by a spin delocalization mechanism,
236

however these arguments would only be skeptical as there is no experimental data for the
magnitude of hyperfine coupling that the metal/metal-tren system is experiencing. There is one
exception of note, and that is the Ga-PSQ system, and that will be discussed below.
The study of organic radicals in Chapter 2 saw an interesting twist in that the simulations
showed evidence of multiple species in solutions, rather than one ion pair; an ion-pair and dimer
were found for NaPSQ and an ion-pair, dimer, and a “tris-like” complex for Na(OMe)2PSQ. It is
possible that in using sodium metal as a reductant, not only was the quinone reduced to a
semiquinone but sodium also provided a surface for unwanted reactions. This situation made for
a more serendipitous occasion than at first realization of the active multiple species.
The extra species in solution allowed for a more comprehensive study of radical
delocalization. Although these complexes were not physically characterized, it was shown that it
was not possible to properly simulate all features of the ESEEM spectrum without the additional
species. For NaPSQ, the dimer showed that when an additional ligand is bound to an ion (or
within close proximity), the unpaired spin density will be delocalized across the Na+ bridge, and
onto the second ligand. This is a comparable effect to the delocalization in the Zn-PSQ system
of Chapter 3; indeed, the isotropic and dipolar couplings are nearly identical. It would be
interesting to synthesize each of these complexes (both isotopologues) and analyze their EPR,
ESEEM and HYSCORE spectra and compare them to the spectra presented in this work. Lastly,
if these individual complexes were realized, then to ultimately determine the validity of the
analysis of Chapter 2, mixtures of the complexes could be made and then analyzed by CW and
pulsed techniques.
Chapter 2 also afforded an interesting observation in that, in comparing the hyperfine
parameters of NaPSQ and Na(OMe)2PSQ, the isotropic couplings were larger from NaPSQ,

237

while dipolar couplings were larger for Na(OMe)2PSQ. This was true for both ion-pair and
dimer compounds. The increase in dipolar coupling can be attributed to the methoxy groups;
these are electron-donating groups and in this case the oxygen atoms donate into the π system of
the phenanthrene ring.
Analysis of the d10-semiquinone complexes revealed some expected results and some
unusual results. After analyzing the NaPSQ and Na(OMe)2PSQ dimer systems, it was expected
that the carbonyl carbon dipolar coupling may be on the order of 0 – 10 MHz for the Zn-PSQ
and Ga-PSQ complexes. The hyperfine parameters for Zn-PSQ are listed in Table 1 and they are
nearly identical to the parameters for the NaPSQ dimer (aiso = 2.17 MHz, T = 2.77 MHz). It is
possible that unpaired spin density of the radical has been delocalized across Zn and onto the
amino groups. It would be interesting then to isotopically label the amino nitrogen nuclei with
15

N; 14N ESEEM is a prominent field, but by using

15

N the Mims’ division technique could be

used and only the tren nitrogen nuclei would be analyzed for their hyperfine parameters.
Additionally, the amino protons could be labeled as 2H nuclei, this would give rise to 2H ESEEM
and further analyze the delocalization as a function of 2H dipolar coupling.
Now, for the curious results of the Ga-PSQ complex. As a reminder, only isotropic coupling
constants were found and the values are: 69,71Ga-aiso = 19.73 MHz, C1-aiso = 27.77 MHz and C2aiso = 32.10 MHz. The isotropic hyperfine interaction at the gallium nucleus is consistent with
other Ga-SQ complexes2.

The carbonyl carbon isotropic coupling constants were the real

surprise, especially after analyzing all the complexes of this work; these values are much greater
than any of the other isotropic coupling constants simulated. The objective of Chapter 3 was to
determine if there was an effect of Lewis acidity on spin delocalization. It appeared that when
comparing Zn-PSQ and [NaPSQ(PhenQ)], it was possible that the ancillary ligand had more of

238

an effect on spin delocalization than the ion itself.

Including Ga-PSQ into that equation

introduces yet another layer of complexity. The isotropic values are large enough to suggest that
the dipolar values are minimal, but without ESEEM data, this is only a speculation. This gallium
complex was to aid in answering the inquiry into spin delocalization, but is has created more
questions than answers. A CdII-PSQ complex should be further pursued to examine hyperfine
coupling at the metal center; 111Cd and 113Cd have I = ½ and their relative natural abundances are
12.8% and 12.22%, respectively.

This would be an inexpensive means of studying spin

delocalization in a d10, MII-PSQ complex; this complex was pursued and even EPR, ESEEM and
HYSCORE data were recorded, but the identity of the complex was unable to be determined by
characterization techniques.
Finally, the Ni-PSQ study showed that there are still complications to be overcome in
scrutinizing the spin delocalization mechanisms active in an exchange-coupled complex.
Although the line shape of the Ni-PSQ complex was able to be recreated (to a certain degree),
the model was inaccurate in its estimation of exchange coupling. The value in the model was
much less than estimations for that specific complex and what has been found for similar
complexes3,4. Additionally, the simulated zero-field splitting parameters were also larger than
what had been previously measured3. Although the magnitude was misrepresentative of the
magnitude of the exchange coupling, varying this magnitude showed changes in the feature at g2 and this may be indicative of a substituent effect. This should be pursued by examining the
previously synthesized complexes, as well as by improving the simulation software to account
for exchange coupling values that are more representative of the experimentally observed values.
The magnitude of the carbonyl carbon isotropic coupling constants fell within the range of
what had been observed for the other compounds examined (excluding Ga-PSQ) and what had

239

been predicted by DFT.

The dipolar coupling constants were of a similar magnitude to

[NaPSQ(PhenQ)] dimer and Zn-PSQ, but were a mere fraction of what had been calculated by
DFT. It was not surprising that dipolar coupling predictions were contrasting to the values
measured; this was the case for Zn-PSQ as well, large dipolar values were calculated and values
simulated were vastly lower in magnitude.
Despite these short-comings, the HYSCORE spectra and analysis afforded a decent
understanding of the hyperfine interaction at the carbonyl carbon nuclei. The variance between
Ni-PSQ and Ni-[13C2]PSQ was stark in that the hyperfine coupling was obvious and had rich fine
structure that could not be observed from ESEEM and especially the CW spectrum, where any
hyperfine coupling was completely unresolved and swamped out by the large exchange
interaction. The ESEEM simulations and spin system proved to be a good match for simulating
the HYSCORE spectra as well. Recall that the ESEEM and HYSCORE spectra were modeled as
a doublet, instead of a quartet. The simulation achieved accurate simulation of both strong and
weak hyperfine couplings.
What remains to be accomplished by these simulations was an accurate portrayal of the echo
amplitude (or intensity) of the resonances in both ESEEM and HYSCORE simulations.
Specifically, the height of the peaks typically increases when investigating lower magnetic field
values compared to higher magnetic field values. This effect was unable to be reproduced by
either a double or quartet system; the doublet systems were pursued due to accuracy of peak
position prediction. Again, this could be due to the origin of the radical being housed in orbitals
on the PSQ moiety.

240

5.2 Future Directions

The goal of this project was to measure hyperfine interactions and the heart of a direct
exchange interaction despite not having a measure J. It would be advantageous to devise a
system where J values are easily measured by variable temperature magnetic susceptibility and
spin density information could be obtained by means of EPR. Such a system would therefore
need a nucleus at which spin density could be measured. This would be the most advantageous
situation to learn about the abilities of substituents and metal centers to polarize/delocalize spin
density.
In the course of looking for a system in which this could be done, two systems have been
examined that follow the same structural motif as the metal complexes studied within this work,
phenazinesemiquinone (PazSQ) and anthracenesemiquinone (ASQ). Both of which can be seen
below, in Figure 1.

Figure 5-1. Molecular structures of [Ni(tren)(ASQ)]+ (top) and [Ni(tren)(PazSQ)]+ (bottom).

241

A DFT evaluation was undertaken to examine the ability of these systems to polarize spin
density with various substituents and determine their corresponding exchange coupling
constants.

The initial geometries were created in GaussView45 from the [Ni(tren)(PSQ)]+

structure. Calculations were performed in Guassian036 for [Ni(tren)(4,5-R2-PazSQ)]+ and in
Gaussian096 for [Ni(tren)(4,5-R2-ASQ)]. Calculations were performed with the unrestricted
formalism and the B3LYP7–9 functional (UB3LYP is the appropriate syntax for input files).
Geometries were optimized with the 6-31G(d,p)10–20 basis set and single point calculations were
carried out with the 6-311g(d,p)21–24 basis set. Frequencies were checked so as to ensure that no
imaginary frequencies existed in the course of optimizing the geometry.
The broken symmetry approach was used to model these compounds. The Hamiltonian
̂ = 𝐽𝑆̂1 ∙ 𝑆̂2 , where S1 and S2 are the spin
used to describe these molecules is of the form 𝐻
operators for the individual spin centers and J is the exchange coupling constant. Complexes
were optimized as quartet states and single point calculations were done for the quartet and
doublet states with the quartet optimized geometry. J values were obtained by the following
equation developed by Yamaguchi and coworkers25:
𝐽=2

𝐸𝐻𝑆 − 𝐸𝐵𝑆
2
𝐻𝑆 − 〈𝑆 〉𝐵𝑆

(1)

〈𝑆 2 〉

Where EHS is the energy of the high spin state (quartet) and EBS (doublet) is the energy of the
broken symmetry state, and <S2>HS and <S2>BS are the spin expectation values for the high spin
and broken symmetry states, respectively. Spin density values were calculated as stated in
previous chapters: beta spin densities were subtracted from alpha spin densities for individual
atoms and then the values were averaged over either the two oxygen atoms, or nitrogen atoms.

242

Displayed below in Figure 2 are the results for Ni-ASQ, where the substituents are in the
4,5 positions, on the back of the anthracene ring.

[Ni(tren)(4,5-R2-AnthraSQ)]+ J vs. Avg. Oxygen Spin Density
-200
-190
NO2

-180
-170

J/cm-1

-160
-150

CF3

-140
OMe
NH2

-130
Br

-120
-110
-100
0.110

H

R² = 0.9426
0.120

0.130

0.140

0.150

0.160

0.170

Average Oxygen Spin Density
Figure 5-2. Average oxygen spin density versus J for [Ni(tren)(4,5-R2-ASQ)]+ complexes.
A linear trend between the average oxygen spin density and J is observed, just as in the case of
the 3,6-R2-PSQ series. The main difference here is that the trend is not necessarily associated
with electron withdrawing/donating groups. The two major observations from the Ni-PSQ series
was a linear relationship, as well as electron donating groups were associated with lower J values
and electron withdrawing groups were associated with larger J values. This of course makes it
all the more interesting to study both series of compounds experimentally.

Although the

advantage with the ASQ series is that the J values are approximately half of the calculated values

243

compare to the NiPSQ series. J was able to be experimentally determined for [Ni(tren)(3,6-NH2PSQ)](BPh4) and it was found to -119 cm-1, while the calculated value was -209 cm-1. If this
overestimation of the J value is true for the Ni-ASQ complexes then these would be prime
targets to be pursued for experimental studies.
Displayed below is J versus oxygen spin density for [Ni(tren)(4,5-R2-PazSQ)]+
complexes.

-350
CN

-330

Cl

-310

H
Me

-290

J/cm-1

NO2
CF3

OMe

-270
NH2

-250
NMe2
-230

R² = 0.9753

-210
0.2

0.21

0.22

0.23

0.24

0.25

0.26

Average Oxygen (ρ)
Figure 5-3. Average oxygen spin density versus J for [Ni(tren)(4,5-R2-PazSQ)]+ complexes.

A linear trend is also observed but here the J values are much greater than the values calculated
for the Ni-PSQ series, which nearly solidifies the inevitability that J values for these compounds
will not be able to be determined experimentally; however, these compounds have a N atom in
244

between the exchange center and the substituents, making it convenient to study hyperfine
coupling parameters by 14N-ESEEM and HYSCORE.
These are two possible avenues to pursue when considering future directions for projects
in the same vein of this work. Additional studies were undertaken to calculate J for Îą-diimine
complexes; these are ligands where the carbonyl oxygen nuclei have been replaced with nitrogen
nuclei. The idea here is that

14

N is the most abundant form of nitrogen and it has I = 1, which

means that the hyperfine coupling could be measured by ESEEM. These calculations are not
presented here, but in each complex that was examined, the calculated J values were greater than
those of Ni-PSQ, thus it would be unlikely that J values could be experimentally determined.
One last consideration for future studies would be to add a bridge between the metal and SQ so
as to have a super-exchanged complex. The exchange coupling may then be a more suitable
magnitude to be measured by variable temperature magnetic susceptibility.

5.3 Final Remarks

EPR is a powerful tool for investigating the hyperfine structure of various types of
systems. CW-EPR measurements will often provide isotropic couplings, ESEEM measurements
measure dipolar couplings and HYSCORE spectra show the landscape of the hyperfine coupling
throughout the molecule. It has been established in this work and elsewhere26,27 that this can be a
strong method for investigating the spin properties of exchange-coupled inorganic complexes as
well as organic radicals and organic-based radical-metal complexes.
13

C-labelling of the quinone moiety has shown to be an interesting means of studying the

hyperfine coupling in the compounds studied here. While not providing for a clear correlation of

245

hyperfine coupling and delocalization between PSQs, Zn-PSQ and Ni-PSQ, it was shown that
the isotropic coupling is rather consistent among all these types of compounds, and the main
entity to examine is the dipolar coupling.

The dipolar coupling showed obvious changes

depending on the type of molecule being examined. This study will provide a backdrop for a
more extensive study including multiple derivatives, as well as lay the foundation for more
ESEEM and HYSCORE studies of synthesized inorganic complexes experiencing some form of
exchange coupling.

246

REFERENCES

247

REFERENCES

(1)

Grimaldi, S.; Ostermann, T.; Weiden, N.; Mogi, T.; Miyoshi, H.; Ludwig, B.; Michel, H.;
Prisner, T. F.; MacMillan, F. Biochemistry 2003, 42 (19), 5632–5639.

(2)

Schrauben, J. N.; Guo, D.; McCracken, J. L.; McCusker, J. K. Inorganica Chim. Acta
2008, 361 (12–13), 3539–3547.

(3)

Fehir, R. J. Differential Polarization of Spin and Charge Density in Phenoxy Radicals,
Semiquinone Radicals, and Ni(II) and Cr(III) Phenanthrenesemiquinone Complexes,
Michigan State University, 2009.

(4)

Benelli, C.; Dei, A.; Gatteschi, D.; Pardi, L. Inorg. Chem. 1988, 27 (16), 2831–2836.

(5)

Denningham II, R. D.; Keith, T. A.; Millan, J. Semichem, Inc.: Shawnee Mission, KS
2003.

(6)

M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
G. Scalmani, V. Barone, G. A. Petersson, H. Nakatsuji, X. Li, M. Caricato, A. Marenich,
J. Bloino, B. G. Janesko, R. Gomperts, B. Mennucci, H. P. Hratchian, J. V. Ort, D. J. F.
Gaussian, Inc.: Wallingford, CT 2013.

(7)

Becke, A. D. J. Chem. Phys. 1993, 98 (2), 1372.

(8)

Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37 (2), 785–789.

(9)

Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J. J. Phys. Chem. 1994, 98
(45), 11623–11627.

(10)

Ditchfield, R.; Hehre, W. J.; Pople, J. a. J. Chem. Phys. 1971, 54 (2), 724.

(11)

Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972, 56 (5), 2257–2261.

(12)

Hariharan, P. C.; Pople, J. a. Theor. Chim. Acta 1973, 28 (3), 213–222.

(13)

Hariharan, P. C.; Pople, J. A. Mol. Phys. 1974, 27 (1), 209–214.

(14)

Gordon, M. S. Chem. Phys. Lett. 1980, 76 (1), 163–168.

(15)

Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; DeFrees, D. J.;
Pople, J. a. J. Chem. Phys. 1982, 77 (7), 3654–3665.

(16)

Binning, R. C.; Curtiss, L. A. J. Comput. Chem. 1990, 11 (10), 1206–1216.

(17)

Blaudeau, J.-P.; McGrath, M. P.; Curtiss, L. A.; Radom, L. J. Chem. Phys. 1997, 107 (13),
5016.
248

(18)

Rassolov, V. A.; Pople, J. A.; Ratner, M. A.; Windus, T. L. J. Chem. Phys. 1998, 109 (4),
1223–1229.

(19)

Rassolov, V. A.; Ratner, M. A.; Pople, J. A.; Redfern, P. C.; Curtiss, L. A. J. Comput.
Chem. 2001, 22 (9), 976–984.

(20)

Petersson, G. A. J. Chem. Phys. 1991, 94 (9), 6081–6090.

(21)

Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72 (1), 650–
654.

(22)

McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72 (10), 5639.

(23)

Wachters, A. J. H. J. Chem. Phys. 1970, 52 (3), 1033.

(24)

Raghavachari, K.; Trucks, G. W. J. Chem. Phys. 1989, 91 (2), 1062.

(25)

Yamaguchi, K.; Tsunekawa, T.; Toyoda, Y.; Fueno, T. Chem. Phys. Lett. 1988, 143 (4),
371–376.

(26)

McCracken, J.; Eser, B. E.; Mannikko, D.; Krzyaniak, M. D.; Fitzpatrick, P. F.
Biochemistry 2015, 54 (24), 3759–3771.

(27)

McCracken, J.; Cappillino, P. J.; McNally, J. S.; Krzyaniak, M. D.; Howart, M.; Tarves, P.
C.; Caradonna, J. P. Inorg. Chem. 2015, 54 (13), 6486–6497.

249