Michigan State
University

This is to certify that the
dissertation entitled

DIFFERENTIAL POLARIZATION OF SPIN AND CHARGE
DENSITY IN PHENOXY RADICALS, SEMIQUINONE
RADICALS, AND NICKEL(|I) AND CHROMIUM(III)
PHENANTHRENESEMIQUINONE COMPLEXES

presented by
Richard J. Fehir Jr.

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Chemistry

Quiz vim/x

fl Major Professor’s Signature

MSU is an Affirmative Action/Equal Opportunity Employer I

 

 

 

i PLACE IN RETURN BOX to remove this checkout from your record.
. TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Acc&Pres/ClRC/DateOueindd

 

DIFFERENTIAL POLARIZATION OF SPIN AND CHARGE DENSITY IN
PHENOXY RADICALS, SEMIQUINONE RADICALS, AND NICKEL(II) AND
CHROMIUMUII) PHENANTHRENESEMIQUINONE COMPLEXES

By

Richard J. Fehir Jr.

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Chemistry

2009

ABSTRACT

DIFFERENTIAL POLARIZATION OF SPIN AND CHARGE DENSITY IN
PHENOXY RADICALS, SEMIQUINONE RADICALS, AND NICKEL(II) AND
CHROMIUMOII) PHENANTHRENESEMIQUINONE COMPLEXES
By
Richard J. Fehir Jr.

A density functional theory (DFT) investigation has been conducted on para-
substituted phenoxy radicals in an effort to understand the factors influencing spin and
charge density distribution in open shell systems. The charge density variation correlated
well with the Hammett-Brown (of) constants for each substituent, indicating that
inductive and resonance effects are responsible for charge density variations within the
molecule. However, the calculations revealed that the distribution of spin and charge are
not correlated: cases were found for which spin and charge move in concert, whereas for
other substituents the two quantities exhibit different polarizations. We uncovered an
empirical relationship between spin density polarization and the a—HOMO-a-LUMO gap
of substituted phenoxy radicals. This led to the development of a model based on a three-
electron, two-orbital scheme in which mixing between the HOMO of the substituent and
the SOMO of the phenoxy radical largely dictates the nature of the unpaired spin
distribution throughout the molecule. DF T calculations were also carried out on 3,6- and
4,5-R-o-1,2-semiquinones and 2,7- and 3,6-R-o-9,10-phenanthrenesemiquinones and
similar trends have been observed.

The effect of coordination upon the spin density distribution in nitro-substituted

phenanthrenesemiquinone (PSQ) radicals was examined in gallium(III) 3,6- and 2,7-di-

5‘,

A"

NOz-PSQ compounds by electron paramagnetic resonance spectroscopy (EPR) and
electron spin echo envelope modulation spectroscopy (ESEEM). The 14N hyperfine
coupling constants (HF C) determined from ESEEM spectroscopy indicated a small
increase in the spin density polarization in the Ga—3,6-di-NOz-PSQ complex compared to
the free radical ligand, while the Ga-2,7-di-N02-PSQ complex exhibited a much larger
l4N HF C. The larger MN for Ga—2,7-di-NOz-PSQ was determined to be due to direct spin
density delocalization onto the nitrogen nucleus, whereas the nitrogens in Ga—3,6-di-N02-
PSQ attain spin population from only the spin polarization mechanism.

The effect of spin density delocalization on the magnitude of the Heisenberg
exchange coupling between a paramagnetic transition metal and the PSQ radical was
elucidated from both a DFT and experimental magnetic susceptibility examination of Ni”
and Crm-PSQ complexes. DFT calculations have shown that electron donors increase the
exchange coupling constant (J) and decrease the spin density at the oxygen atoms of the
PSQ ligand, while electron acceptors exhibit the opposite trend. Analogous to the
phenoxy radicals, the HOMO-LUMO gap was observed to be the marker for spin
delocalization in both ferromagnetically and antiferromagnetically exchange coupled
Nin-PSQ and Crm-PSQ complexes, respectively. Experimental magnetic measurements
confirmed that an electron donor decreased the J value in the [Ni(tren)(3,6-di-NH2-
PSQ)](BPh4) (where tren == tris-Z-aminoethylamine) complex, since in all other cases J is
too strong to be determined experimentally. These results could have implications in the
field of molecular magnetism by providing a mechanism for synthetically tuning the

magnitude of intramolecular exchange interactions in a rational and predictable fashion.

Copyright by
RICHARD J. FEHIR JR.

2009

This thesis is dedicated to Richard Sr., MaryAnn, Sarah, and Margaret Fehir, Mary Knop,

the entire Fehir Family, and for Rachel.

ACKNOWLEDGEMENTS

First of all I would like to thank my advisor Jim McCusker for all the help and
guidance he has provided me throughout my graduate school career. Jim always had a
way to “fire me up” about science and to focus my thoughts on drinking deeply about my
project. I thank the entire McCusker group for all their aid during my group meetings,
seminar preparations, and for the great moments at the whiteboard. A few members must
be singled out, and I thank Joel Schrauben and Dong Guo for their kindness in critically
evaluating this thesis. I also thank Dong for bestowing upon me his expert synthetic
advice and training in air-sensitive chemistry and for solving all my X-ray structures. I
have also enjoyed the times Spent with my EPR collaborators Matt Krzyaniak, Joel, and
Prof. John McCracken with the “boat anchor” and collecting some great ESEEM data. I
also thank Dr. Reza Lolee for his constant assistance with the SQUID magnetometer.

I would also like to thank my family for their unending support during my tenure
at Michigan State, and of course Rachel for keeping me sane during the PhD process. I
am also grateful to my friends Chris Boy, T-Roy, DB, Joel, Johnny Bravo, the Captain,
Mikey, Speedo, Berry, Tana, Kate, Johnny OC, and Blondie for all the great times here at
State, and I know we will have many more! I will miss all of our outings to the Dairy

Store, Big B, the RIV, and 40’s at group meeting. Thanks for making my time here some

of the best years of my life.

vi

TABLE OF CONTENTS

List Of Tables ................................................................................................................... vii
List Of Figures ................................................................................................................. viii
Chapter 1. Introduction to Heisenberg Exchange Coupling and the Importance of Spin
Density Polarization and Delocalization ......................................................... 1

1.1 Introduction to Heisenberg Exchange Coupling ...................................... 1

1.2 Importance of Spin Polarization and Delocalization ............................... 8

1.3 Contents of Dissertation ........................................................................... 9

1.4 References .............................................................................................. 15

Chapter 2. Differential Polarization of Spin and Charge Density in Substituted Phenoxy
Radicals ......................................................................................................... 18

2.1 Introduction ............................................................................................ 18

2.2 Experimental Section ............................................................................. 22

2.3 Results and Discussion .......................................................................... 24

2.4 Concluding Comments ........................................................................... 47

2.5 References and Notes ............................................................................. 50

Chapter 3. Differential Polarization of Spin and Charge Density in ortho-semiquinones
Radicals ......................................................................................................... 57

3.1 Introduction ............................................................................................ 57

3.2 Theoretical Calculations ........................................................................ 59

3.3 Results and Discussion .......................................................................... 59

3.3.1 Substituted o-1,2-semiquinones .................................................. 60

3.3.2 Substituted 9,10-PhenanthrenesemiquinoneAnion Radicals ...... 78

3.4 Concluding Comments ........................................................................... 89

3.5 References .............................................................................................. 92

Chapter 4. Density Functional Theory and Electron Paramagnetic Resonance

Spectroscopy of Substituted Phenanthrenesemiquinones and Gallium(III)

Phenanthrenesemiquinone Complexes ......................................................... 94
4.1 Introduction ............................................................................................ 94
4.1.1 Electron Paramagnetic Resonance Spectroscopy ......................... 98
4.1.2 Electron Spin Echo Envelope Modulation Spectroscopy ........... 102
4.2 Experimental ........................................................................................ 104
4.3 Results and Discussion ........................................................................ 115
4.3.1 Hyperfine Coupling in 2,7- and 3,6-Phenanthrenesemiquinone
Ligands ...................................................................................... 115

4.3.2 ESEEM Spectroscopy of 2,7- and 3,6-N02
phenanthrenesemiquinones ....................................................... 1 23

Vii

4.3.3 EPR and ESEEM Spectroscopy of Gallium 3,6-
and 2,7-N02-PSQ ...................................................................... 125
4.3.4 Potassium 2,7-NOz-PSQ Contact Radical Ion Pair ................... 130
4.3.5 Density Functional Theory of Gallium
Phenanthrenesemiquinones ....................................................... 136
4.4 Concluding Comments ......................................................................... 142
4.5 References ............................................................................................ 145

Chapter 5. The Substituent Effect in the Modulation of the Heisenberg Exchange
Coupling Interaction in Ni(II) and Cr(III) Phenanthrenesemiquinones: A

Combined Experimental and Theoretical Examination .............................. 150
5.1 Introduction .......................................................................................... 150
5.2 Experimental ........................................................................................ 153
5.3 Results and Discussion ........................................................................ 168
5.3.1 Synthesis and Characterization of Nickel(II)
2,7-Phenanthrenesemiquinone Complexes ............................... 169
5.3.2 Synthesis and Characterization of [Ni(tren)(3,6-R-PSQ)](BPh4)
Complexes ................................................................................. 198
5.3.3 Density Functional Theory Study of Substituted
[N i(tren)(3,6-R—PSQ)]+ ............................................................. 212
5.3.4 Antiferromagnetic Coupling in [Cr(tren)(3,6-R-PSQ)]2+
Complexes ................................................................................. 222
5.3.5 Synthesis and Characterization of Cr-3,6-R-PSQ complexes...231
5.4 Concluding Comments ......................................................................... 237
5.5 References ............................................................................................ 241
Chapter 6. Concluding Comments and Future Directions ............................................... 247
6.1 Concluding Comments ......................................................................... 247
6.2 Future directions .................................................................................. 253
6.3 References ............................................................................................ 260

viii

Table 2-1.

Table 3-1.

Table 3-2.

Table 4-1.

Table 4-2.

Table 4-3.

Table 5-1.

Table 5-2.

Table 5-3.

Table 5-4.
Table 5-5.

Table 5-6.

Table 5-7.

LIST OF TABLES

Percent atomic contributions for oxygen (%0) and the substituent (%R)
group in the a—HOMO and or-LUMO Kohn—Sham MO’s for a series of

para-Phenoxy radicals ............................................................................... 42
Percent composition of 3,6-R-o-l,2-semiquinone frontier orbitals ........... 71
Percent contribution to the frontier orbitals of 2,7-
phenanthrenesemiquinones ........................................................................ 85

Experimental and DF T calculated isotropic hyperfine coupling constants

for 2,7-N02-PSQ and 3,6-N02-PSQ ........................................................ 119
Experimental and calculated spin densities in 2,7 and

3,6-N02-PSQ radicals .............................................................................. 121
DFT calculated gallium HFC’s and spin densities .................................. 136

Selected bond distances and angles for [Ni(tren)(PSQ)]+ (1), [Ni(tren)2,7-
NOz-PSQ)](BPh4) (2), and [Ni(tren)(2,7-NH2-PSQ)]+ (4) ...................... 171

Crystallographic data for [Ni(tren)(PSQ)]+ (l), [Ni(tren)2,7—NOz-
PSQ)](BPh4) (2), and [Ni(tren)(2,7-NH2-PSQ)]+ (4) ............................... 172

Comparison of electronic absorption transitions in [Ni(tren)(PSQ)](BPh4)
(1), [Ni(CTH)(PSQ)l(PF6), Wi(tren)(2,7-di-N02-PSQ)I(BPh4) (2),
[Ni(tren)(2,7-di-NH2-PSQ)](BPh4) (4), and in the last column the
diamagnetic [Zn(CTH)(PSQ)](BPh4), and [Ga(tren)(PSQ)](BPh4) (12)
analogues, where MLCT = metal-to-ligand-charge transfer and IL =
intraligand ................................................................................................ 174

Crystallographic data for [Ga(tren)(PCAT)](BPh4) (11) ......................... 175
Selected bond distances and angles for [Ga(tren)(PCAT)](BPh4) (11) 176
The NPA UB3LYP/6—311g** spin density distribution in the Ni- 2 ,-7 R-
PSQ complexes, where the spin density was summed for all atoms in the
fragment ................................................................................................... 193
DFT calculations at the UB3LYP/6-311g** level for exchange coupling

constants of Ni-2,7-R-PSQ complexes and the spin densities at Ni and O
in the quartet state .................................................................................... 193

ix

Table 5-8.

Table 5-9.

Table 5-10.

Table 5-11.

Table 5-12.

Table 5-13.

Table 5-14.

Table 5-15.

Table 5-16.

Table 6-1.

DFT calculations at the UB3LYP/6-311g** level for exchange coupling
constants of the Ni-3,6-R-PSQ complexes and the spin densities at Ni and
O in the quartet state ................................................................................ 197

Crystallographic data for [Ni(tren)(3,6-NH2-PSQ)](BPh4) (6) and
[Ni(tren)(3,6-N02—PSQ)](BPh4) (5) ......................................................... 200

Selected bond distances and angles for [Ni(tren)(3,6-NH2-PSQ)](BPh4) (6)
and [Ni(tren)(3,6-N02-PSQ)](BPh4) (5) .................................................. 201

Transitions in [Ni(tren)(3,6-NOz-PSQ)](BPh4) (5) and [Ni(tren)(3,6-NH2-
PSQ)](BPh4) (6), where MLCT == metal—to-ligand-charge transfer and IL =
intraligand ................................................................................................ 203

The % nickel (% Ni) and average % oxygen (% O) in the highest occupied
a-HOMO (ct-110), along with the average spin density at the oxygen

atoms(p(O)) .............................................................................................. 220

The % chromium (% Cr) and average % oxygen (% O) in the highest
occupied B-HOMO, of substituted [Cr(tren)(3,6-R-PSQ)]2+ complexes and
the average spin density at the oxygen atoms (p(O)) .............................. 230

Crystallographic data for [Cr(tren)(PCAT)](BPh4) (7) and
[Cr(tren)(PSQ)] (BPh4)2 (8) ...................................................................... 233

Selected bond distances and angles for [Cr(tren)(PCAT)](BPh4) (7) and
[Cr(tren)(PSQ)](BPh4)2 (8) ...................................................................... 234

Comparison of the absorption features in [Cr(tren)(PSQ)](BPh4)2 (8),
[Cr(tren)(3,6-di-NOz-PSQ)](BF4)2 (10), and [Ga(tren)(PSQ)](BPh4)2
(blue) (12), where LMCT == ligand-to-metal charge transfer ................... 236

NPA spin density comparison between the phenanthrenesemiquinone
(PSQ), coronenesemiquinone (coronene-SQ), 2,3-anthrasemiquinone
(ASQ), and phenazinesemiquinone (phenazine-SQ) ............................... 257

Figure 1-1.

Figure 1-2.

Figure 1-3.

Figure 2-1.

Figure 2-2.

Figure 2-3.

Figure 2-4.

Figure 2-5.

Figure 2-6.

LIST OF FIGURES

Antiferromagnetic exchange coupling interaction between two S = 1/2 spin

centers leading to a singlet ground state. States are labeled as IST), where
ST is the total spin ......................................................................................... 3

Ferromagnetic exchange coupling interaction between two S = 1/2 centers
leading to a triplet ground state .................................................................... 4

Ferromagnetic coupling between the S = 3/2 and S = 1/2 of [Ni(CTH)(3,5-
di-t-butylsemiquinone)](PFG). The energetic results of zero field splitting
of the S = 3/2 state is also depicted, where D is much smaller in
comparison to J ............................................................................................ 5

Shifts in charge (blue) and unpaired spin (red) density at the oxygen atom
for a series of para-substituted phenoxy radicals. The % change in both
parameters is referenced to values obtained for the unsubstituted
compound (i.e. R = H). A positive change indicates an increase in the
magnitude of negative charge density or an increase in spin density ........ 26

Plot of the calculated change in charge density at the oxygen atom as a
function of the OJ coefficient of the R-group in para-substituted phenoxy
radicals. The value given is relative to that obtained for the unsubstituted
radical (i.e. R = H); a positive number indicates an increase in negative

charge density at the oxygen atom. The of values were taken from the
review by Hansch, Leo, and Taft (cf. 4). See text for funher details ........ 29

Plots of the change in spin density at the oxygen atom of para-substituted
phenoxy radicals as a function of the AD (A), 00" (B), 06 (C), and 013' (D)
Hammett-type spin delocalization parameters ........................................... 34

Comparison of the ct-HOMO, B-LUMO, and oc-LUMO Kohn—Sham
molecular orbitals of the p-NHz-phenoxy radical. The orbitals were
generated at the UB3LYP/6-311G** level of theory and visualized using
GaussView with an isosurface value of 0.020 ........................................... 36

A plot of the calculated changes in spin density at the oxygen atom versus
the oc-HOMO-oc-LUMO gaps for a series of para-substituted phenoxy
radicals calculated at the UB3LYP/6—31 16’” level of theory .................. 37

Molecular orbital-based description of the spin delocalization mechanism
proposed for rr-donor-containing phenoxy radicals. The interaction of the
dyadono, orbital of the substituent with the phenoxy ¢phenoxy-SOM0 results in
a net one-electron stabilization. The degree of interaction between these
two orbitals dictates the oc-HOMO-oc-LUMO energy gap, which in turn

xi

Figure 2-7.

Figure 2—8.

Figure 2-9.

Figure 3-1.

Figure 3-2.

Figure 3-3.
Figure 3-4.

Figure 3-5.

Figure 3-6.

Figure 3-7.

Figure 3-8.

Figure 3-9.

Figure 3-10.

Figure 3-11.

alters the composition of the or-HOMO and thus the spatial distribution of
unpaired spin density ................................................................................. 39

Comparison between the Kohn-Sham (X-HOMO molecular orbitals of p-
CF3- and p-NHz-substituted phenoxy radicals............; .............................. 40

Comparison of the a-LUMO orbitals for phenoxyl and p—NHz—phenoxyl
radicals with the LUMO and LUMO+1 orbitals of p-NOz-phenoxyl ....... 43

Proposed molecular orbital diagram for p-NOz-phenoxyl. The presence of
a low—energy empty if“ orbital on the N02 group leads to a change in the
nature of the LUMO and is responsible for the deviation exhibited by p-
NOz-phenoxyl from the correlation illustrated in Figure 5 ........................ 44

Spin density (red) and charge density (blue) polarization in 3,6-R-0-1,2-
semiquinones .............................................................................................. 61

Charge density vs. additive Hammett parameter for 3,6-R-o-1,2-

semiquinones .............................................................................................. 63
Charge density vs. 60 for 3,6-R—o-1,2-semiquinones ................................. 64
Charge density vs. cm for 3,6-R—o-1,2-semiquinones ................................ 64

Spin resonance interactions of a B—spin donor with the semiquinone radical
(left) and a-Spin acceptor (right). The red arrows represent the donation or
withdrawal of charge density from the negatively charged semiquinone .65
Spin density distribution in 3,6-N02- and 3,6-NH2-0-1,2-semiquinones ..67

Spin density correlation with the oc' substituent constant in 3,6-R-o-1,2-
semiquinones .............................................................................................. 68

Spin density correlation with the 055' substituent constant in 3,6-R-o-1,2-

semiquinones .............................................................................................. 69
Charge density vs. Hammett op' correlation for 2,7-R-
phenanthrenesemiquinones ........................................................................ 69

Spin density correlation with the additive (ADm + ADO) substituent
constant in 3,6-R-o-1,2-semiquinones ....................................................... 7O

Spin density correlation with ct-HOMO-a-LUMO gap in 3,6-R-o-1,2-
semiquinones .............................................................................................. 71

xii

Figure 3-12.

Figure 3-13.

Figure 3-14.
Figure 3-15.

Figure 3-16.

Figure 3-17.

Figure 3-18.

Figure 3-19.

Figure 3-20.

Figure 3-21.

Figure 3-22.

Figure 3-23.

Figure 3-24.

Figure 3-25.

Figure 3-26.

Figure 3-27.

Spin density (red) and charge density (blue) polarization in 4,5-R-0-1,2-
semiquinones .............................................................................................. 73

Charge density vs. additive Hammett parameter for 4,5-R—o-1,2-

semiquinones .............................................................................................. 74
Charge density vs. 6,, for 4,5-R-o-l ,2-semiquinones ................................. 75
Charge density vs. cm for 4,5-R-o-l ,2-semiquinones ................................ 75

Correlation of spin density with the a-HOMO-a-LUMO gap in 4,5-R-0-
1,2-semiquinones ....................................................................................... 77

DFT optimized structures depicting the steric interactions in 3,6-N02-o-

1,2-semiquinone (lefi) and 4,5-N02-o-1,2-semiquinone (right) ................ 77
Spin (red) and charge (blue) polarization for 2,7-R-
phenanthrenesemiquinones ........................................................................ 80
Charge density vs. Hammett om. correlation for 2,7-R-
phenanthrenesemiquinones ........................................................................ 81
Charge density vs. Hammett op correlation for 2,7-R-
phenanthreneserniquinones ........................................................................ 81
Charge density vs. Hammett op’ correlation for 2,7-R-
phenanthrenesemiquinones ........................................................................ 82
Spin resonance structures of 2,7-R-phenanthrenesemiquinone substituted

with a n-donor (left) and n-acceptor (right). The red arrows represent

charge donation or withdrawal from the PSQ anion .................................. 83
Spin density delocalization in 2,7-NH? and 2,7-N02-
phenanthrenesemiquinone .......................................................................... 84
Change in spin density vs. u-HOMO-u-LUMO gap for 2,7-
phenanthrenesemiquinones ........................................................................ 84
Spin density (red) and charge density (blue) polarization in 3,6-R-

phenanthrenesemiquinones ........................................................................ 86
Spin density delocalization in 3,6-NHz-PSQ and 3,6-N02-PSQ ............... 87
Charge density polarization in 3,6-phenanthanrenesemiquinones ............. 88

xiii

Figure 3-28.

Figure 4-1.

Figure 4-2.

Figure 4—3.

Figure 4-4.

Figure 4-5.

Figure 4-6.

Figure 4-7.

Figure 4-8.

Figure 4-9.

Figure 4-10.

Change in spin density vs. a-HOMO—a-LUMO gap for 3,6-
phenanthrenesemiquinones ........................................................................ 89

General spin density distribution in a metal-ligand complex .................... 94

Interaction of the at and B'spin states of the electron with an applied
magnetic field ............................................................................................. 99

Electron-nuclear spin states of the hydrogen atom .................................. 101

Energy level diagram for an S = 1/2, == 1 system near exact cancellation
conditions ................................................................................................. 103

The experimental setup for the 36 hour photolysis reaction. The
equipment pictured is a quartz Schlenk flask, quartz water bath, and HgH
lamp. The reaction is done in a fume hood, with the sashes covered to
prevent accidental UV-light exposure ..................................................... 105

Generic 2,7- and 3,6-R-PSQ radical labeling .......................................... 116

Experimental X-Band EPR spectra of 2,7-di—N02-PSQ (1) in black (top)
acquired in THF at room temperature, and simulation in red (bottom). The
simulation parameters'are: g = 2.01023, A(H1,3) = 0.936 G, A(H3,(,) =
1.952 G, A(H4,5) = 0.393 G, and linewidth = 0.15 G. Experimental
Conditions: frequency = 9.4479, modulation amplitude = 0.1 G, and
power = 1.25 mW .................................................................................... 117

Experimental X-Band EPR spectra of 3,6-di-NOz-PSQ (2) in black (top)
acquired in THF, and simulation in red (bottom). The simulation
parameters are: g = 2.0106, A(N3,6) = 0.406 G, A(Ht,s) = 0.413 G, A(H2,7)
= 1.197, A(H4,5) = 0.407, and linewidth = 0.262 G ................................. 118

Spin density distribution in 1 (left) and 2 (right), where the blue color
represents excess a-spin density and the green represents excess B-spin
density ...................................................................................................... 120

Stimulated echo detected ESEEM spectra of 3,6-di-NOz-PSQ (2) (black)
at 50 K with frequency = 9.682 GHz; center field = 3445 G; t = 136 ns;
starting T = 40 ns; time increment = 16 ns. Simulation (red) parameters: g
= isotropic, A = (1.3, 1.1, 0.9) MHz, equ = 1.2 MHz, Euler = (0.0, 0.4,
0.0), and n = 0.8. Spectra were obtained by normalizing the raw data and
subtracting the bi-exponential decay followed by application of a
Hamming window and Fourier transformation. Absolute value spectra are
shown. Recall, 1.0 G = 0.1 mT = 2.8025 MHz ....................................... 123

xiv

Figure 4-11.

Figure 4-12.

Figure 4-13.

Figure 4-14.

Figure 4-15.

Figure 4-16.

Figure 4-17.

The room temperature X-band EPR spectrum of [Ga(tren)(2,7-di-N02-
PSQ)](Cl)(BF4) (3) in acetonitrile. Experimental conditions: frequency =
9.772 GHz, power = 5 mW, modulation frequency = 100 G, modulation
amplitude = 0.2 G. See text for further details ........................................ 125

The room temperature X-band EPR spectrum (top) and simulation
(bottom) of [Ga(tren)(3,6-di-N02-PSQ)](BF4)2 (4) in acetonitrile.
Simulation parameters: g = 2.00979, A(Ga) = 6.20 G, A(H) = 0.90 G, line
width = 2.90 G. Experimental conditions: frequency = 9.672 GHz, power
= 5 mW, modulation frequency == 100 G, modulation amplitude = 0.2 G.
See text for further details ........................................................................ 126

Stimulated echo detected ESEEM spectra of [Ga(tren)(3,6-di-N02-
PSQ)](BF4)2 (4) (black) at 50 K in acetonitrile with frequency = 9.694
GHz (black); center field = 3450 G; t = 136 ns; starting T = 40 ns; time
increment == 16 ns. Simulation (red) parameters: g = isotropic, A = (2.0,
1.3, 0.8) MHz, eZqQ = 2.5 MHz, 11 = 0.60, and Euler angles = (0.0,0.9,0.0).
Spectra were obtained by normalizing the raw data and subtracting the bi-
exponential decay followed by application of a Hamming window and
Fourier transformation. Absolute value spectra are shown .................... 128

Stimulated echo detected ESEEM of [Ga(tren)(2,7-di-N02-
PSQ)](C1)(BF4) (3) (black) at 50 K in acetonitrile with frequency = 9.676
GHz; center field = 3450 G; 1: == 136 ns; starting T = 40 ns; time increment
= 16 ns. Simulation (red) parameters: g = isotropic, A = (-2.1, -2.1, 21)
MHz, eZqQ = 1.6 MHz, 11 = 0.60, and Euler angles = (0.0, 0.9, 0.0).
Spectra were obtained by normalizing the raw data and subtracting the bi-
exponential decay followed by application of a Hamming window and
Fourier transformation. Absolute value spectra are shown .................... 129

A representation of the spin delocalization mechanism onto the nitrogen of
the N02 group in the [Ga(tren)(2,7-di-N02-PSQ)](C1)(BF4) (3)
complex .................................................................................................... 130

X-Band CW spectrum in ethanol at room temperature of potassium 2,7-di-
NOz-PSQ contact radical ion pair (top) and simulation (bottom).
Simulation parameters: g = 2.01030, A(N) = 1.5545 G, A(H) = 0.398 G,
A(H) = 0.383 G, and linewidth = 0.215 G. Experimental conditions:
frequency = 9.734 GHz, modulation amplitude = 0.1 G, and power =
lmW. See text for further details ............................................................ 131

Stimulated echo detected ESEEM spectra of 2,7-di-NOz-PSQ (1) (black)
at 50 K in ethanol with frequency = 9.685 GHz; center field = 3450 G; r =
136 ns; starting T = 40 ns; time increment = 16 ns. Simulation (red)
parameters: g = isotropic, A = (0.65, 0.35, 0.05) MHz, e2qQ = 1.1 MHz, 11

XV

Figure 4-18.

Figure 4-19.

Figure 4-20.

Figure 4-21.

Figure 4-22.

Figure 5-1.

Figure 5-2.

Figure 5-3.

Figure 5-4.

Figure 5-5.

Figure 5-6.

= 0.80, and Euler angles = (0.0, 0.2, 0.0). Spectra were obtained by
normalizing the raw data and subtracting the bi-exponential decay
followed by application of a Hamming window and Fourier
transformation. Absolute value spectra are shown. See text for further

details ....................................................................................................... 133
Interaction between the filled Ga(III) 3dxz and the

2,7-NOz-PSQ SOMO ............................................................................... 135
Total spin density distribution in 3 (left) and 4 (right) ............................ 137

The polarization of n—spin density is depicted as it is predicted by DFT
calculations for the Ga-2,7- and Ga-3,6-carbon atoms and N02 groups. In
order for a simple comparison, the PSQ ligands are represented by a
benzene ring ............................................................................................. 138

tit-HOMO of2 (left)and 4 (right) .............................................................. 139

The results of a UB3LYP/6-31g* total spin density field calculation for
2,7-N02-PSQ (left) along with the result from the same calculation
without the field keyword (right). The Gaussian 03 field = y + 200
keyword was used in the field calculation ............................................... 142

Drawings of the cations of [Ni(tren)(PSQ)]+ (1) (top), [Ni(tren)(2,7-N02-
PSQ)]+ (2) (bottom left), and [Ni(tren)2,7-NH2-PSQ)](BPh4) (4) (bottom
right) obtained from X-ray structure determinations. Atoms are
represented as 50 % probability ellipsoids ............ v ................................... 170

Electronic absorption spectra of [Ni(tren)(PSQ)](BPh4) (1) (red),
[Ni(tren)(2,7-di-NOz-PSQ)](BPh4) (2) (green), and [Ni(tren)(2,7-di-NH2-
PSQ)](BPh4) (4) 031116) in ACN at 25 °C ................................................. 173

X-ray crystal structure of [Ga(tren)(PCAT)](BPh4) (11) ......................... 175
Plots of the effective magnetic moment versus temperature for
[Ni(tren)(PSQ)](BPh4) (1) (red circles), [Ni(tren)(2,7-di-NOz-PSQ)](BPh4)
(2) (green squares), and [Ni(tren)(2,7-di-NH2-PSQ)](BPh4) (4) (blue
triangles) .................................................................................................. 178

The exchange coupling interaction of the Ni“ S = 1 and PSQ
S = 1/2 spins .............................................................................................. 179

TGA plots of [Ni(tren)(PSQ)](BPh4) (1) (red), [Ni(tren)(2,7-di-N02-
PSQ)](BPh4) (2) (green), and [Ni(tren)(2,7-di-NH2-PSQ)](BPh4) (4) ....I79

xvi

Figure 5-7.

Figure 5—8.

Figure 5-9.

Figure 5-10.

Figure 5-11.

Figure 5-12.

Figure 5-13.

Figure 5-14.

Figure 5-15.

Figure 5-16.

Figure 5-17.

Figure 5-18.

Figure 5-19.

Plot of effective magnetic moment versus temperature for a high
temperature (300-450 K) SQUID experiment of [Ni(tren)(PSQ)](BPh4)
(1). The low temperature data (5350 K) was done under normal

conditions ................................................................................................. 180
The packing diagram for [Ni(tren)2,7-N02-PSQ)](BPh4) (2) ................. 182
Packing diagram for the [Ni(tren)(PSQ)](BPh4) complex (1) ................. 183

Variable temperature Evans NMR of [Ni(tren)2,7—NOz-PSQ)](BPhu) (2) in
d7 DMF. Experimental conditions were: 500 MHz NMR spectrometer
with 2 % t-butanol used as the reference solvent ..................................... 184

Liquid phase SQUID susceptibility of Cu(NO3)2-3H20 in water at 5.5 T, at
1.0 M (red square), 2.0 M (green circle), 4.0 M (blue triangle), and 5.6 M
(brown diamond) concentrations ............................................................. 186

The 0.83 M Ni-PSQ solution in DMF liquid phase SQUID data (blue
circles), and the solid state data (red square) is also shown for
comparison .............................................................................................. 187

The energy splitting caused by zero field splitting for an S = 3/2 ground
state with C2,, symmetry .......................................................................... 187

SQUID susceptibility of a 0.41 M dichloromethane solution of
[Ni(tren)(2,7-NOz-PSQ)](BArf) (red circles) compared with the solid BPh4
salt (green squares) .................................................................................. 189

Magnetic orbitals for the [Ni(tren)(PSQ)]+ complex ............................... 191

The UB3LYP/6-3llg** spin density plots for the [Ni(tren)(2,7-N02-
PSQ)l(BPh4), [Ni(tren)(2r7-NH2-PSQ)](BPh4), and Ni(tren)(PSQ)l(BPh4)
complexes ................................................................................................ 196

The UB3LYP/6-311g** spin density plots for the [Ni(tren)(3,6-N02-
PSQ)I(BPh4), [Ni(trenX3,6-NH2-PSQ)](BPh4), and Ni(tren)(PSQ)l(BPh4)
complexes ................................................................................................ 198

Drawings of the cations of [Ni(tren)(3,6-NHz-PSQ)]Jr (6) (left) and
[Ni(tren)(3,6-NOz-PSQ)]+ (5) (right) obtained from single crystal X-ray
structure determinations. Atoms are represented as 50 % probability
thermal ellipsoids ..................................................................................... 199

Electronic absorption spectra of [Ni(tren)(PSQ)](BPh4) (1) (red),

[Ni(tren)(3,6-di—N02-PSQ)](BPh4) (5) (green), and [Ni(tren)(3,6-di-NH2-
PSQ)](BPh4) (6) (blue) in acetonitrile ..................................................... 202

xvii

Figure 5.20.

Figure 5-21.

Figure 5-22.

Figure 5-23.

Figure 5-24.

Figure 5-25.

Figure 5-26.

Figure 5-27.

Figure 5-28.

Figure 5-29.

Figure 5-30.

Cyclic voltarnmograms of the [Ni(tren)(3,6-di-NH2-PSQ)](BPh4) (6) (top),
Ni(tren)(PSQ)](BPh4) (1) (middle), and [Ni(tren)(3,6-di-NOz-PSQ)](BPh4)
(5) (bottom). All data were acquired in degassed acetonitrile containing
0.1 M NBuuPF6 at a scan rate of 100 mV s'l. All potentials are referenced
to the ferrocenium/ferrocene couple as an internal standard ................... 203

The X-band EPR spectra of the [Ni(tren)(3,6-di-NOz-PSQ)](BPh4) (5)
(top), [Ni(tren)(3,6-di-NH2-PSQ)](BPh4) (6) (middle), and
Ni(tren)(PSQ)](BPh4) (1) (bottom) complexes. The spectra were acquired
at 4 K in a 9:2 propionitrile to butyronitrile glass at 9.465 GHz with a 10
G modulation amplitude .......................................................................... 206

Temperature dependence of a powder sample of [Ni(tren)(PSQ)](BPh4) at ,
331.2 GHz ................................................................................................ 207

The effective magnetic moment vs. temperature for the
Ni(tren)(PSQ)](BPh4) (1) (red circle), [Ni(tren)(3,6-di-N02-PSQ)](BPh4)
(5) (green square), and [Ni(tren)(3,6—di-NH2-PSQ)](BPh4) (6) (blue
triangle) complexes .................................................................................. 209

The effective magnetic moment of [Ni(tren)(3,6-NH2-PSQ)](BPh4) (6)
(blue triangle), and the solid line represents a fit to the data using
parameters described in the text ............................................................... 210

Correlation between the UB3 LYP/6-31 1g** calculated average (it-spin
density (NPA) at the PSQ oxygen atoms and the J value calculated with
Equation 3 ................................................................................................ 213

The correlation between J calculated with Equation 3 and the product of
the NPA Ni" and average oxygen (it-spin densities, calculated at the
UB3 LYP/6-3 1 1g** level ......................................................................... 214

The correlation between the UB3LYP/6-311g** J value calculated with

Equation 3 and the Hammett op+ substituent constant. The substituent
constants were taken from Hansch and coworkers .................................. 215

The correlation between the average u-Spln density at oxygen with the 01-
HOMO-a-LUMO gap, calculated at the UB3LYP/6-311g** level ......... 216

Weak bonding interaction between the filled Nin-dxz and two
energetically different PSQ-SOMO orbitals, where the u-HOMO-u-
LUMO gap is labeled in blue ................................................................... 217

The four-orbital, three-electron interaction of the outliers in the Ni-3,6-R-
PSQ complexes ........................................................................................ 219

xviii

Figure 5-31.

Figure 5-32.

Figure 5-33.

Figure 5—34.

Figure 5-35.

Figure 5-36.

Figure 5-37.

Figure 5-38.

Figure 5-39.

Figure 5-40.

Figure 5-41.

Figure 6-1.

Figure 6-2.

The antiferromagnetic coupling interaction in Cr-3,6-R-PSQ complexes.

States are labeled as IST), where ST is the total spin ................................ 223
The UB3LYP/6-311g** calculated B-HOMO orbital of [Cr(tren)(3,6-
PSQ)]2+ of the triplet ground state ........................................................... 223

J (calculated with Equation 3) vs. the average B-spin at the oxygen atoms
for [Cr(tren)(3,6-R-PSQ)]2+ complexes calculated at the UB3LYP/6-
311g“ level ............................................................................................. 225

J (calculated with Equation 3) vs. the product of the average B-spin at the
oxygen atoms and Cr"I centers for [Cr(tren)(3,6-R-PSQ)]2+ complexes
calculated at the UB3LYP/6-311g** level .............................................. 225

The correlation between the average P-spin and charge density at the
oxygen center in [Cr(tren)(3,6-R-PSQ)] + compounds ................................. 226

The correlation between the average B-spin density at the oxygen atoms
with the B-HOMO-B-LUMO gap for the [Cr(tren)(PSQ)]2+ complexes..227

The changes in the B-HOMO-B-LUMO gap for the 3,6—H-PSQ vs. 3,6-
NOz-PSQ substituted Cr-PSQ’s ............................................................... 228

The changes in the B-HOMO-B-LUMO gap for the 3,6-H-PSQ vs. 3,6-
NHz-PSQ substituted Cr-PSQ’s ............................................................... 228

Drawings of the cations of [Cr(tren)(PCAT)]+ (7) (left) and
[Cr(tren)(PSQ)]2+ (8) (right) obtained from single crystal X-ray structure
determinations. Atoms are represented as 50 % probability thermal
ellipsoids .................................................................................................. 232

The normalized absorption spectra of [Cr(tren)(PSQ)](BPh4)2 (8) (red),

[Cr(tren)(3,6-di-NOz-PSQ)](BF4)2 (10) (green), and
[Ga(tren)(PSQ)](BPh4)2 (blue) (12) ......................................................... 235

The temperature dependent SQUID susceptibility of
[Cr(tren)(PSQ)](BPh4)2 (8) (red square) and [Cr(tren)(3,6-di-N02-
PSQ)](BF4)2 (10) (green triangle) ............................................................ 237
Phenazine ligand ...................................................................................... 257

Calculated structure (top) and total spin density plot (bottom) for
[Ni(tren)(ASQ)]+ ...................................................................................... 258

xix

Images in this dissertation are presented in color.

XX

Chapter 1. Introduction to Heisenberg Exchange Coupling and the Importance of

Spin Density Polarization and Delocalization.

1.1 Introduction to Heisenberg Exchange Coupling

Interest in Heisenberg exchange coupling is expanding from its origins in
filndamental electronic structure research to encompass a variety of disciplines from
biology to materials science.“ For example, it has been suggested that exchange
coupling plays a role in the oxidation of alcohols to aldehydes coupled with the reduction
of oxygen to peroxide in the enzyme galactose oxidase.7’8 The ability to modulate the
exchange coupling constant can also have far reaching applicability to aid chemists in
designing molecular magnetic materialsg’M A number of workers have suggested that
increasing the strength of the exchange coupling constant in molecular magnetic
materials can increase the critical temperature at which these materials become

12,15

magnets. It is known that the exchange coupling constant can be modulated by

”’18 and by substituents”20 for organic

molecular confirmationf”l6 electron delocalization,
systems. Shultz et al. have also shown that the exchange coupling constant can be
modulated by substituents in Cu"- and Mnu-S-aryl-substituted semiquinones,” but the
mechanism for the changes was not firmly developed. Several workers have suggested
that the magnitude of the Heisenberg exchange coupling interaction is proportional to the
spin densities at the interacting atoms.”26 Along these lines, spin polarization in radical
ligands must be fully understood to aid in the intelligent design of molecular magnetic

materials and to properly predict the exchange coupling constants. The overall goal of

this thesis is to further our understanding of the relationship between spin density and the

exchange coupling constant. In this regard, the cause and effect relationship of
substituent perturbations to the spin density population at a paramagnetic center is critical
for the understanding of the mechanism to modulate the exchange coupling interaction.
Exchange coupling is an electrostatic interaction that arises whenever two or more
paramagnetic centers are in close proximity.27 In order for exchange coupling to occur,
three general criteria must be met: (1) each constituent undergoing the exchange
interaction must have unpaired electrons; (2) there must be an orbital pathway (usually
orthogonal or non-orthogonal) for the unpaired spins to interact; and (3) there must be a
reasonable energetic match between the orbitals involved in the exchange pathway.28
These interactions can be present in the excited state as well as the ground state, and spin
exchange effects can cause large perturbations to both the ground and excited state
electronic structure of molecules. The most common type of exchange coupling is
intramolecular exchange, where either direct exchange or superexchange interactions can
occur. Direct exchange is the through-bond overlap between two spin centers and has
been observed in several transition metal semiquinone,29 verdazyl,3O and nitroxide“
radical complexes. Superexchange coupling occurs through a diamagnetic bridge and has
been observed in several transition metal dimers, with those of copper(II) ions being the
most thoroughly examinedmz'34 Intermolecular or through-space exchange interactions
between different molecules can also occur, but are usually found to be very weak in
magnitude compared to intramolecular exchange coupling.35 Additionally, both intra-

and intermolecular exchange can be operative in transition metal complexes and each

interaction can be important in determining the ground state magnetic properties.1

The Heisenberg Hamiltonian (H) for the interaction between two spin operators
Si and S2 on different centers is depicted in Equation 1, where J12 is the exchange integral
that measures the magnitude of the electronic coupling between spin centers Si and S2.

H =JIZSi-Sz (1)

The Hamiltonian in Equation 1 is also represented by using different terms for the
exchange coupling constant, i.e. 2.112, -J12, and ~2Jiz. The form used in Equation 1
indicates that if J > 0 ferromagnetic exchange coupling ensues, and if J < 0 there will be
antiferromagnetic exchange coupling. The most common types of exchange coupling,
ferromagnetic and antiferromagnetic exchange, are described below. For two S = 1/2 spin
centers undergoing an antiferromagnetic exchange coupling interaction the following
energy level diagram for the coupled spin states can be drawn according to Figure 1-1. In
the direct exchange regime an antiferromagnetic interaction can occur if there is overlap
of the magnetic orbitals between the unpaired spin centers, and following Hund’s rules

the spins will pair and the singlet state will be lower in energy.

Antiferromagnetic Interaction

 

>~ I
@ _,_—" I )
g S} “ l/z ____}_,a "
IL] 522% AE=J
\ IO) Non-zero overlap betucen orbitals
Non-Interacti ng (J = 0) Anti ferromagnetic
Exchange (J > 0)

Figure 1-1. Antiferromagnetic exchange coupling interaction between two S = l/z spin centers leading to a
Singlet ground state. States are labeled as IST), where ST is the total spin.

The opposite situation occurs for a ferromagnetic interaction between two S = '/2 spin
states and the S = 1 state is lower in energy as seen in Figure 1-2. The magnetic orbitals
of the two spin centers must be strictly orthogonal for a ferromagnetic interaction to
operate, and following Hund’s rule the direction of the spin vectors is the same leading to

an S = 1 ground state as depicted in Figure 1-2.

 

, |O) l-‘crromagnclic Interaction
>9 ,. '
9.” x’
q) S = l/ [1’ AE - J
t: l 2 __
m 52:1/2 —' \‘ ‘~\~

I ll) Orthogonal orbitals

Non-Interacting (J = 0) Ferromagnetic Exchange
(J < 0)

. . . . . _ I . .
F lgure 1-2. Ferromagnetlc exchange coupllng lnteractlon between two S — /2 centers leadlng to a trlplet

ground state.

An example in the literature where the above rules can be applied is in the case of

[Ni(CTH)(3,5-di-t—butylsemiquinone)](PFg) or Ni”-SQ (CTH = d1-5,7,7,12,14,14-

hexamethyl-1,4,8,1l-tetraazacyclotetradecane), where strong ferromagnetic coupling

between an S = 1 Niu center and a S = V2 semiquinone radical has been experimentally
observed.36 In the case of the Niu-SQ complex, the dx2.Y2 and dz2 magnetic orbitals on

the Ni" have c*-symmetry and the unpaired electron in the PSQ radical is in a it“-

SJimmetry orbital. These orbitals are strictly orthogonal and Hund’s rules require that the

unpaired electrons on the Nill (S = l) and the PSQ (S = 1/2) undergo ferromagnetic

exchange coupling, giving rise to a S = 3/2 ground state and a S = V2 excited state, as seen

 

 

in Figure 1-3.
I 02>

xii

>‘ ’1’ :3

an I S _ 1/2 1" AB /2 J

g f 3:3/ ‘ ,I’I§‘Ims=i3/2

m I \\\\ ””/
‘L: AE = 2 D

I3/2> “

“I‘lms =i v,

 

 

 

Non-Interacting J > 0
I (J = 0)

Figure 1-3. Ferromagnetic coupling between the S = 3/2 and S — /2 of [Ni(CTH)(3, 5-di-t-

butylsemiquinone)](PF6). The ener rgetic results of zero field splitting of the S — 3/2 state is also depicted,
where D 15 much smaller ln comparison to J.

An operator-equalivant form of Eq 1 can be generated by applying the Kambe37 method
of a vector coupling of spin operators, where ST = S i + $2, to yield Equation 2
H: /2 Jl2'2(Sr ‘Slz '322) (2)
and recall from quantum mechanics that when a squared operator (Opz) operates on the
wavefunction the eigenvalue can be determined by Equation 3
Opz‘l’ = op(op + 1))11 (3 )
and using Equation 3 and the operator-equalivant Hamiltonian the eigenvalues can be

determined according to Equation 4

E = '/2- J12- (sT(sr +1 ) - Si(Si+1) - 32(Sz+1)) (4)

 

We can utilize Equation 4 to calculate the eigenvalues for the Niu-SQ spin system as seen

1 15 8 3 1
E =3 :‘EJIZIT—Z‘Z) ‘5le
1 3 8 3
E ——_J -———~— —
E E 3 (5)
-3“ _l =‘ 12
3‘5 7 2

 

 

where the energy difference between the S = 3/2 and S = 1/2 sates is 3/2 J12. At room
temperature, if the magnitude of J >> kBT then an isolated ground state is observed with
no population of the excited state. The lack of thermal population of the S = 1/2 state
occurs since the energy gap is much larger than thermal energy available at room
temperature (~ 208 cm"). The magnitude of J in the Niu-SQ complex has been estimated
to be greater than 400 cm" indicating very strong ferromagnetic coupling and confirming
an isolated quartet ground state.

The exchange coupling interaction occurs in zero applied magnetic field, and
other interactions may also present that perturb the electronic structure of this Niu-SQ
complex. One possible interaction is called zero field splitting. The magnitude of the
Splitting is equal to the zero field splitting constant, D. Zero field splitting is one
mechanism that lifts the ground state degeneracy and results in the splitting of the S = 3/2
into the ms = i 3/2 and i 1/2 states as depicted in Figure 1-3. The contribution from zero
field splitting to lift the degeneracy of the S = 3/2 ground state of the Nin-SQ cation is a

.11
Symntetry effect. This state is 4-fold degenerate (2S + 1) and the symmetry of the N1 -

SQ complex can be approximated by the sz point group which can only support l-fold

‘m-t-afl

 

degeneracy, thus the ground state must split according to symmetry rules. The zero field
energy gap can be derived in an analogous manner to the exchange coupling constant.
The Hamiltonian for zero field splitting is shown in Equation 6, where 822 is the 2-

component of the spin angular momentum and ST is the total spin.
Has = 0'522 - '/3-sr2 (6)
Utilizing Equation 4 and recalling that SZZ‘P = mgz‘lJ38 the eigenvalues can be determined
by Equation 7,
Ezfs=D'(m52)—1/3'ST‘IST‘H» (7)

and the energy gap caused by zero field splitting will then be:

= 2D (8)

 

 

In Figure 1-3, D is assumed to be positive giving the ms = d: '/2 states as the lowest in

energy, but the sign of the D value must be experimentally determined from EPR
spectroscopy. An appreciable zero field splitting can decrease the observed magnitude of
the magnetic susceptibility at low temperatures. This is due to non-zero population of the
S = i 1/2 states in Figure 1-3 when compared to only population of the S = 3/2 state.
When the thermal energy exceeds the zero field splitting then the susceptibility will
correspond to the expected S = 3/2 state. At low temperatures lower susceptibility values

. -1
are experimentally observed for the Niu-SQ complex, Wthh has D = 2.54 cm .

For molecular systems the magnitude of J is usually determined from measuring
the macroscopic molar magnetization (M) of a sample as a fimction of temperature. The
macroscopic magnetization is a sum of the microscopic magnetizations of all spin states
weighted according to the Boltzmann distribution.1 The magnetization is related to the
magnetic susceptibility by the following, where x is the molar magnetic susceptibility and
B is the magnetic field.

3.4.
B

x = (9)

The magnetic susceptibility can also be described by the following general expression1
for the Van Vleck Equation in Eq 10,
—E(S)/kT
S S+1 ZS+1 e
NAgzuf, ; < x )

x = (10)
3kT gas +1)e-E(S)/kT

 

where NA is Avogadro’s ntunber, k is Boltzmann’s constant, g is the gyromagnetic factor
of the electron, in; is the Bohr Magneton, T is the temperature, S is the total spin quantum
number of a given spin state, and E(S) are the eigenvalues from the Heisenberg
Hamiltonian (Eq 1) that depend on J. Equation 10 is then compared with the measured
values of x as a function of temperature and the exchange coupling constant(s) can be

determined by a least-squares fitting.

1.2 Importance of Spin Polarization and Delocalization
The focus of this dissertation will be on transition metal complexes that are directly
coordinated to organic radicals, thus a major factor to control spin density at the

paramagnetic centers will be spin delocalization or localization in the organic radicals.

The organic radicals of choice are the o—semiquinones (SQ) shown in Scheme H, where
the molecule as a whole is non-innocent due to its redox reactivity and can also exist in
the quinone and catechol redox states. Transition metal complexes with the semiquinone
ligands are well known and have been studied extensivelyf'g“43 The magnetic properties
of transition metal semiquinones has also been reviewed in complexes containing one to
three coordinated semiquinone ligands.29 In order to focus on the effects of spin
polarization, this thesis will only focus on transition metals coordinated to one

semiquinone radical.

o o' o9
+e' +e'
fl ...,___--—--—_______"
-e‘ -e'
e
o 0 0e
Quinone Semi uinone C atechol
Cl

Scheme 1-1. The redox states of the o-quinone ligands.

The substituent effects of radical spin delocalization and polarization will be important to
develop a prescription to predictably change the spin density population in radicals so
that the J value can be modulated in transition metal organic radical complexes. If a
basic theory that relates the spin density in the radical to J can be found, the intelligent
design of exchange coupled molecular magnetic materials can be achieved. Efforts to
understand and characterize the model that can predictably control the spin density at a

magnetic center and the J value in an exchange coupled system is the main focus of this

work.

1.3 Contents of Dissertation

The heart of this research can be split up into three main fronts; (1) computational
studies of spin and charge polarization in phenoxy and semiquinone radicals; (2) electron
paramagnetic resonance spectroscopy of semiquinones and gallium complexes; and (3)
the synthesis and magnetic characterization of Ni"- and Crm-semiquinone complexes
combined with a DF T study of exchange coupling interactions. The literature is lacking
in the deep understanding of the intimate movement of spin and charge density in
radicals, and coupled with this is a need to develop a more in-depth knowledge of the
substituent effect on the exchange coupling constant. The contributions of this thesis
along these lines is highlighted below.

Chapter 2 begins with the DFT examination of a series of para-substituted
phenoxy radicals in an effort to understand the factors influencing spin and charge
density distribution in open-shell systems. The calculations reveal that the distribution of
spin and charge are not correlated: cases were found for which spin and charge move
together, whereas for other substituents the two quantities exhibit spatially distinct
intramolecular polarizations. Charge density variations across the series were found to
correlate well with both the Hammett (op) and Hammett-Brown (op+) constants for each
substituent, indicating that inductive and/or resonance effects are primarily responsible
for the polarization of charge within the molecule. In contrast, the distribution of
unpaired spin density could not be adequately accounted for using any of the typical
Hammett-type spin delocalization constants cited in the literature. We uncovered an
empirical correlation between the polarization of spin density and the or-HOMO-ot-

LUMO gap of the substituted phenoxy radicals: this led to the development of a simple

10

model based on a three-electron, two-orbital bonding scheme in which mixing between
the HOMO of the substituent and the SOMO of the phenoxy moiety serves to define the
nature and extent of unpaired spin polarization throughout the molecule. This analysis
yielded a correlation coefficient of r > 0.97 for the 15 substituents examined in the study;
spin polarization effects in compounds that exhibited the greatest deviation from this
correlation could also be readily explained within the context of the model. The
underlying reason for the ability to differentially polarize spin and charge likely stems
from the fact that net unpaired spin density is completely carried by the unpaired electron
(and thus the spatial characteristics of the SOMO), whereas charge density reflects the
behavior of all of the electrons of the system. This research laid down a basic
understanding of how to manipulate spin and charge density in a paramagnetic radical
ligand. In order to apply this knowledge to transition metal based systems, ligands that
can easily bind to a metal center were needed. This led to the examination of substituted
ortho-semiquinone complexes since transition metal complexes with these ligands are
well-known.29’39‘“’44

Chapter 3 uses DFT to examine the spin and charge polarization in 3,6-o-l,2-
semiquinones, 4,5-0-1,2-semiquinones, and o-9,10-phenanthrenesemiquinones. Spin and
charge density were found to differentially polarize. In all cases the Hammett substituent
constants correlated with the change in charge density at the oxygen atoms. The spin
density mechanism was found to be more complicated than in the phenoxy radical case,
and unfavorable charge-dipole interactions increased spin density at the oxygen atoms.
Substituents that are strong electron donors were found to destabilize the negative charge

within the ring, and spin density was localized on the oxygen atoms. Electron acceptors

11

were found to stabilize the negative charge and decrease the spin density at the oxygen
centers. The spin density was again found to correlate with the a-HOMO-a—LUMO gap,
but the correlation could not be as easily accounted for using MO theory as the phenoxy
radical case. Even though the mechanism of spin delocalization could not be firmly
established, it was noted that the trend in the u-HOMO-a-LUMO gap with the spin
density is a marker for the spin delocalization. Proper choice of substituents in the
semiquinone radicals can lead to increases or decreases in the spin density at the oxygen
atoms, and this prescription will be used to see if the same trends are observed in
transition metal systems.

In an effort to uncover any changes in spin delocalization/polarization when
semiquinone radical ligands are coordinated to a metal center, Gam-nitro-
phenanthrenesemiquinone complexes were examined by DFT and electron paramagnetic
resonance (EPR) spectroscopy, which is described in Chapter 4. The nitrated radical
ligands 3,6-di-N02—phenanthrenesemiquinone (3,6-di-N02-PSQ) and 2,7-di-N02—
phenanthrenesemiquinone (2,7-di-N02—PSQ) were perturbed with X-band EPR and
electron spin echo envelope modulation spectroscopy (ESEEM) to experimentally
determine the spin density population on the nitro-substituents by observing the 14N
hyperfine coupling constants (HFC). The 3,6- and 2,7-di-N02 radicals were both found
to have weak HFC of 14N = 0.35 and 0.15 G, respectively. The 14N HFC is due to spin
polarization, which agreed very well with the predicted l4N HFC from DFT calculations.
Upon complexation to the Gal" the X-band data are dominated by gallium HFC, and the
1”N could not be determined in either complex. ESEEM spectroscopy was successful in

determining the IN for the Ga"I complexes, and Ga—3,6-di-N02-PSQ 14N = 0.453 G and

12

Ga-2,7-di-N02-PSQ 14N = 2 G. The ESEEM data indicated that there is a large
difference in the magnitude of the 14N HFC and the spin density that is delocalized onto
the nitrogens. The ESSEM data suggest that direct spin delocalization onto the nitrogen
nuclei is operating in the Ga-2,7-di-N02-PSQ complex45 and spin polarization effects are
responsible for the smaller HF C in the Ga-3,6-di-N02-PSQ system. The DF T results
were noted to predict the incorrect spin distribution in the Ga-2,7-di-N02-PSQ complex,
and the HF C was calculated to be a factor of seven times lower in magnitude. Thus large
differences are observed in the spin distribution upon complexation to a metal center, and
DFT may not adequately describe the spin density distribution in the Ga—2,7-di—N02-PSQ
complex.
Chapter 5 pools the collected knowledge of the previous chapters to evaluate the
effects of substituents on the spin density distribution and J value in Niu and Cr"I
phenanthrenesemiquinone systems. A seminal paper by Andrea Dei and coworkers
showed that a Niu-PSQ complex had a decreased J value than the comparable 3,6-1-
butylsemiquinone complex.46 Consequently substituted PSQ ligands were chosen for our
study. The exchange coupling constants were calculated according to the approximate
spin projection technique of Yamaguchi et a1.47 by DF T methods. The DFT results for
the ferromagnetically coupled Nine-PSQ complexes indicated that the spin density at the
oxygen atoms correlated with the magnitude of the exchange coupling constant, and
substituents that increased spin density increased J and groups that decreased the spin
density decreased J. A three-electron, two-orbital MO scheme was used to describe the
spin delocalization in these systems. Experimental SQUID data also confirmed that

electron donors decrease the J value and electron acceptors increase J. Analogous results

13

were also obtained for the antiferromagnetically coupled Crm-PSQ complexes. These
results are significant, since control of the exchange coupling constant with substituents
can lead to clever design of molecular magnetic materials. Concluding comments and
future directions are included in Chapter 6, where the implications of the previous

chapters are reviewed and possible future directions for this research project are outlined.

14

1.4 References

l

2

10

ll

l2

l3

14

15

l6

17

18

19

Kahn, 0. Molecular Magnetism; VCH Publishers: New York, 1993.
Kahn, O. Acc. Chem. Res. 2000, 33, 647.

Beltran, L. M. C.; Long, J. R. Acc. Chem. Res. 2005, 38, 325, and references
therein.

Holm, R. H.; Kennepohl, P.; Solomon, E. 1. Chem. Rev. 1996, 96, 2239.
Yoon, S.; Lippard, S. J. J. Am. Chem. Soc. 2005, 127, 8386.

Shultz, D. A.; Fico, R. M., Jr.; Bodnar, S. H.; Kumar, R. K.; Vostrikova, K. E.;
Kampf, J. W.; Boyle, P. D. J. Am. Chem. Soc. 2003, 125, 11761.

Itoh, S.; Taki, M.; F ukuzumi, S. Coord. Chem. Rev. 2000, 198, 3.

Whittaker, M. M.; Ekberg, C. A.; Peterson, J.; Sendova, M. 8.; Day, E. P.;
Whittaker, J. W. J Mol. Catal. B. : Enzym. 2000, 8, 3.

Luneau, D. Curr. Opin. Solid State Mater. Sci. 2001, 5, 123.

Ohrstrom, L. C. R. Chimie 2005, 8, 1374.

Shultz, D. A. Polyhedron 2001, 20, 1627.

Caneschi, A.; Gatteschi, D.; Sessoli, R.; Rey, P. Acc. Chem. Res. 1989, 22, 392.
Vostrikova, K. E. Coord. Chem. Rev. 2008, 252, 1409.

Kirk, M. L.; Shultz, D. A.; Depperrnan, E. C. Polyhedron 2005, 24, 2880.
Kahn, 0.; Martinez, C. J. Science 1998, 279, 44.

Shultz, D. A.; Fico, R. M., Jr.; Lee, H.; Kampf, J. W.; Kirschbaum, K.; Pinkerton,
A. A.; Boyle, P. D. J. Am. Chem. Soc. 2003, 125, 15426.

Shultz, D. A.; Kumar, R. K. J. Am. Chem. Soc. 2001, 123, 6431.
Franzen, S.; Shultz, D. A. J. Phys. Chem. A 2003, 107, 4292.

Shultz, D. A.; Bodnar, S. H.; Lee, H.; Kampf, J. W.; Incarvito, C. D.; Rheingold,
A. L. J. Am. Chem. Soc. 2002, 124, 10054.

15

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

 

West, A. P., Jr.; Silverman, S. K.; Dougherty, D. A. J. Am. Chem. Soc. 1996, 118,
1452.

Shultz, D. A.; Sloop, J. C.; Coote, T.-A.; Beikmohmadi, M.; Kampf, J .; Boyle,
P. D. Inorg. Chem. 2007, 46, 273.

Sugiura, K.-I.; Arif, A. M.; Rittenberg, D. K.; Schweizer, J .; Ohrstrom, L.;
Epstein, A. J.; Miller, J. S. Chem. Eur. J. 1997, 3, 138.

Osanai, K.; Okazawa, A.; Nogami, T.; Ishida, T. J. Am. Chem. Soc. 2006, 128,
14008.

McConnell, H., M, J. Chem. Phys. 1963, 39, 1910.

Hicks, R. G.; Lemaire, M. T.; Thompson, L. K.; Barclay, T. M. J. Am. Chem. Soc.
2000, 122, 8077.

Shultz, D. A.; Sloop, J. C.; Washington, G. J. Org. Chem. 2006, 71, 9104.
Ginsberg, A. P. Inorg. Chim. Acta. Rev. 1971, 5, 45.

Wheeler, D. E.; McCusker, J. K. Inorg. Chem. 1998, 37, 2296-2307.
Pierpont, C. G.; Attia, A. S. Collect. Czech. Chem. Commun. 2001, 66, 33.
Hicks, R. G. Aust. J. Chem. 2001, 54, 597.

Eaton, S. S.; Eaton, G. R. Coord. Chem. Rev. 1988, 83, 29.

Hay, P. J.; Thibeault, J. C.; Hoffmann, R. J. Am. Chem. Soc. 1975, 97, 4884.
Kahn, 0.; Chariot, M. F. Nouv. J. Chim. 1980, 4, 567.

Giidel, H. U.; Stebler, A.; Furrer, A. Inorg. Chem. 1979, 18, 1021.

Schultz, D.; Weyhermiiller, T.; Wieghardt, K.; Butzlaff, C.; Trautwein, A. X.
Inorg. Chim. Acta. 1996, 246, 387.

Benelli, C.; Dei, A.; Gatteschi, D.; Pardi, L. Inorg. Chem. 1988, 27, 2831.
Kambe, K. J. Phys. Soc. Japan 1950, 5, 48.

Palmer, G. In Physical Methods in Bioinorganic Chemistry: Spectroscopy and
Magnetism; Que, J ., L., Ed.; University Science Books: Sausallto, 2000.

Pierpont, C. G.; Buchanan, R. M. Coord. Chem. Rev. 1981, 38, 45.

16

40

41

42

43

44

45

46

47

Pierpont, C. G.; Lange, C. W. Prog. Inorg. Chem. 1994, 41, 331.

Zanello, P.; Corsini, M. Coord. Chem. Rev. 2006, 250, 2000, and references
therein.

Pierpont, C. G. Coord. Chem. Rev. 2001, 216-21 7, 99.
Kaim, W. Coord. Chem. Rev. 1987, 76, 187.
Picrpont, C. G. Coord. Chem. Rev. 2001, 219, 415.

McCracken, J.; Peisach, J .; Cote, C. E.; McGuirl, M. A.; Dooley, D. M. J. Am.
Chem. Soc. 1992, 114, 3715.

Bencini, A.; Carbonera, C.; Dei, A.; Vaz, M. G. F. Dalton Trans. 2003, 1701.

Yamaguchi, K.; Jensen, F .; Dorigo, A.; Honk, K. Chem Phys. Lett. 1988, 149,
537.

17

Chapter 2. Differential Polarization of Spin and Charge Density in Substituted

Phenoxy Radicals

2.1 Introduction

Spin and charge represent two of the most fundamental physical properties in
nature. Indeed, the role of these quantities in defining the electronic structures of atoms
and molecules serves to underscore their importance in virtually all areas of molecular
science. From relatively simple ideas such as nucleophilicity to more advanced concepts
involving the emerging field of spintronics,“3 the interplay between these two related but
distinctiaspects of electronic structure represents an intriguing aspect of the chemistry of
open-shell systems.

Many of our expectations concerning charge distribution in molecular systems are
based on the concept of electronegativity. This simple yet fundamental idea provides the
foundation for envisioning the polarization of charge in response to compositional
changes in a molecule. The correlation of experimental observables with the various
Hammett-type relations that have been developed over the years has helped to put these
ideas on a more quantitative (though still largely empirical) basis.“'8 In so doing, the
physical-organic community in particular has been able to develop tremendous physical
and mechanistic insights into the cause-and-effect relationship between charge
polarization and chemical reactivity.9'l3

A comparable level of understanding of the effect of spin on chemical reactivity
suffers in part from the fact that there does not exist a similarly intuitive counterpart to

electronegativity for spin polarization. Experimental techniques for measuring the

18

distribution of unpaired spin density —— most notably EPR and ENDOR spectroscopies -
have contributed greatly toward describing spin polarization in a wide range of chemical
systems. The development of spin-unrestricted formalisms in density functional theory
has also emerged as a powerful computational tool, particularly with regard to large
molecules.”’16 Such calculations can provide a description of unpaired Spin polarization

and provide a theoretical basis for correlating quantum-mechanical descriptions of

17,18 19-21

systems with experimental data that directly or indirectly probe the spatial
distribution of unpaired spin. The question of local spin and spin distribution in
molecules, has been considered in detail by Reiher.22 More recently, Lecomte and co-
workers have pioneered the use of high-resolution electron density maps to probe spin-
spin interaction mechanisms in a variety of systems based on a detailed analysis of
electron density distributions.23 Both of these efforts represent significant first steps
toward making the connection between charge and spin density and assessing its impact
on the physical and chemical properties of molecules.

In the field of magnetism, from which our interest in this issue originates, spin
and charge play very different roles. For example, in inorganic-organic hybrid systems
(e.g., metal-semiquinone complexes)?"27 large formation constants will generally be
favored when the donor atom from the ligand possesses negative charge density sufficient
for it to act as a strong Lewis base to the metal ion. In this context, the charge density
dictates, to a large extent, the stability of the adduct being formed. The magnitude of the
exchange interaction between the paramagnetic constituents will reflect several variables,
but significant among these is the spatial distribution of unpaired spin density. All other

2
1,8

factors being equa polarization of unpaired spin density toward (or away) from the

19

ligating atoms will increase (or decrease) the magnitude of electronic coupling. The
notion of altering the magnetic properties of molecules through chemical means has been
addressed both computationallyzg’32 and experimentally.”'35 Recent work by Shultz and
co—workers, in particular, has clearly demonstrated that spin delocalization contributes to
the attenuation of exchange interactions in bridged polynuclear systems.”37 In related
work, this team has also provided examples of metal-semiquinone complexes in which
changes in ring substituents give rise to variations in the magnitude of intramolecular
spin exchange.38 Clearly, the ability to independently manipulate charge and spin density
within the molecule has significant potential in terms of tailoring the thermodynamic as
well as electronic properties of chemical systems.

At first glance, it would seem that spin and charge density should move in
concert: since electrons carry both charge and spin, it seems reasonable that their
polarizations should track each other. A case in point is the phenoxy radical, several

resonance structures for which are shown in Scheme 1.

4‘46

Scheme l-l. Phenoxy radical resonance structures.

 

We can see that the unpaired electron can be localized on the oxygen atom as well as the
ortho- and para—positions on the ring. EPR data have confirmed that a significant portion
of the spin density is in fact localized at the ortho- and para—carbon atoms in the ring: 170

hyperfine coupling constants of 2,4,6-tri-t—butylphenoxy and 2,4,6-triphenylphenoxy

20

radicals indicate that the spin density at the oxygen atom is approximately equal to the
spin density at the ortho-carbons, a result that is wholly consistent with this picture.”42
Previous computational studies by various DFT methods support these findings in that a
substantial fraction of the spin density was found to be associated with the oxygen
atom.43’44 Not surprisingly, these calculations also revealed that the oxygen carries most
of the negative charge density. The combined experimental and theoretical data therefore
show that the spin and charge density in the (unsubstituted) phenoxy radical are both
polarized in the same direction, i.e. toward the oxygen atom. This case thus conforms to
our initial expectations.

Though reasonable and in fact correct for some systems, there are indications that
spin and charge polarization may not necessarily track one another. Contrast the above
observations with what has been described for the ketyl radical. The resonance structures
in Scheme 2 show that the spin or charge density can be accommodated at either the
carbonyl carbon or the oxygen atom. DFT calculations have shown that the charge

density is largely localized on the oxygen atom,45 which is again consistent with our

expectations based on the electronegativity of oxygen relative to carbon.

 

 

 

09 :5:
‘ 1)
/C\ /Ce\
R R R R

Scheme 1-2. Resonance structures of the ketyl radical.

The EPR spectrum of the benzophenone ketyl radical exhibits 17O and 13C hyperfine

coupling constants consistent with the spin density on both atoms of the carbonyl unit,

21

however, the spin density was shown to be more highly localized at the carbon atom.4648

DFT calculations yielded similar results.

These results for two relatively simple chemical systems provide a convenient
backdrop for the central question of this investigation: is it possible to design molecules
in which spin and charge density can be independently manipulated? To examine this
issue, we have carried out a computational study of substituted phenoxy radicals. This
system has been examined in some detail by other workers, however, the question of spin
versus charge polarization has not been explicitly addressed. In the course of analyzing
our results, we have discovered a previously unrecognized correlation between spin
polarization and electronic structure which we believe not only supports the notion of
differential polarization of spin and charge, but suggests that the potential exists for the
independent manipulation of these two variables through chemical means as a

mechanism for tailoring the physical properties of molecules.

2.2 Experimental Section

General Methods. All self-consistent field Hartree-Fock and density fimctional
calculations were carried out using Gaussian 98.49 Both the BLYP and 133va
functionals were used in the calculations. The BLYP functional contains the gradient
corrected Becke (B)50 exchange functional with the gradient corrected Lee, Yang, and
Pan (LYP)5' correlation functional, whereas the hybrid 33va functional contains the
three parameter exchange of Becke (BB)52 with the LYP correlation functional. All

calculations were performed using tight convergence criteria.53 Analysis of the atomic

22

 

“(‘5

charges and spin densities were performed using the natural population analysis (NPA)
framework developed by Weinhold et al.54 Calculations were carried out using the
Michigan State University Chemistry Supercomputer Facility.

Geometry Optimizations. The initial geometries of the molecules were
generated using SPARTAN 5 5 or GaussView“ and subsequently optimized in three steps.
The first optimization was performed using restricted open-shell Hartree—Fock (ROI-IF)
theory with the STO-3G** basis set. These structures were then subjected to a second
optimization using the restricted open-shell density functional ROBLYP with the 3-
21G" basis set, followed by a third optimization using the ROB3LYP functional with
the 6-31G* basis set. Frequency calculations at the ROB3LYP/6-31G* level of theory
were performed on the final optimized structures to ensure that these geometries
corresponded to global minima. In several cases (R = NH;, NMez, OMe, and CEO,
imaginary frequencies were obtained. For these systems, the molecules were distorted
along those coordinates in order to find the lowest energy structure. The structure was
then reoptimized at the ROB3LYP/6-31G* level and subjected to a subsequent frequency
calculation. This procedure was repeated until no imaginary frequencies were found. In
all cases, the final optimized structures were used in the single point calculations
described below.

Single Point Calculations. Single point calculations were performed using the
unrestricted density functional UB3LYP with the 6-311G** basis set assuming a, doublet
ground state and a molecular charge of zero. The maximum error for the expectation
value of the spin operator (82) differed from the theoretical value for a pure spin state of

S(S + 1) (where S is the total spin) by an average error of 4.3%, and in no case exceeded

23

6%, indicating the degree of spin contamination in the Kohn-Sham Slater determinant is
negligible for the molecules investigated.'5’57’58 The UBLYP functional with the 6—
311G” basis was also used on a subset of substituents (H, Cl, OMe, NHz, N02, and CF 3)
to determine whether any calculated trends in charge and/or spin density distribution
were functional dependent. Although specific values for charge and spin density were
different, trends in structure, spin and charge density across the series were qualitatively
similar. To simplify the discussion, the results reported here derive from calculations at
the UB3LYP/6-31 1G** level.

The substituents examined in these calculations were selected to cover a wide
range of possible interactions with the phenoxyl ring and included c-donors (Me, t-butyl,
and TMS), a o-acceptor (CF3), n-donors (OH, OMe, OEt, NHZ, and NMez), it—acceptors
(N02 and CN), rt-donor/o-acceptors (F, Cl, and Br), and a it-acceptor/o-donor (vinyl). In

all cases the unsubstituted phenoxy radical (i.e., R = H) was used as a reference point for

comparisons.

2.3 Results and Discussion

Geometries of para-Substituted Phenoxy Radicals. A crystal structure for the
phenoxy radical has not been reported, however, it has been studied extensively using a
wide range of computational methods.43 Based on comparisons to the unsubstituted
compound, our results indicate that the geometry of the phenoxy radical in terms of both
bond distances and bond angles is largely insensitive to changes in the para-substituent

on the ring. The most significant perturbation arises in the C—0 distance for the vinyl

24

 

l“

derivative, where a distance of 1.247 A (versus 1.257 A for R = H) was noted. A slight
increase in double bond character was also observed for the N02- and CN-substituted
compounds, though to a lesser extent. The resonance structures in Scheme 1 illustrate that
the origin of double-bond character in the C—0 bond arises from delocalization of lone-
pair electrons on the oxygen into the ring. The slight increases in bond order observed for
the three n-accepting substituents in this study can therefore be viewed as a reflection of
this resonance effect. The influence of electron delocalization will be expanded upon
later in the discussion in the context of spin polarization, but in terms of geometry this
appears to be the only significant structural perturbation across this series.
Spin and Charge Density Movement in Para-substituted Phenoxy Radicals.
A. Overview. This work was originally motivated by an interest in magnetic exchange
interactions between transition metals and bound organic radicals such as
orthosemiquinones.59'6' For such systems, significant negative charge density at the
oxygen atoms coupled with varying degrees of unpaired spin density would provide an
entry into a series of compounds having tunable magnetic properties. Our initial
computational efforts with this class of compounds were inconclusive due to the presence
of competing effects from substituents that simultaneously have different regiospecific
interactions with the two ligating oxygen atoms. We therefore sought to simplify the
problem sufficiently to extract the basic phenomenological effects of compositional
changes on spin and charge polarization that we could then use as a guide for an analysis
of more complex systems.
A graphical depiction of our computational results is shown in Figure 2-1.

Inspection of this plot provides an immediate answer to one of the central questions of

25

 

F7

dt‘

our study: differential polarization of spin and charge appears to be possible, even facile,

given the proper choice of a substituent.

 

10t-

=H)

 

 

 

TMS CF3

— 1

Cl 3. N02 Mgbuty
CN

-10 .. OH

OM
. eOEt

% Change (Relative to R

. NH2
NMez

 

 

 

 

 

 

 

 

ti)
o
r

vinyl

 

it-donors n—donors rt-acceptors n-acceptor o-donors o-acceptor
o-acceptors o-donor

Figure 2-1. Shifts in charge (blue) and unpaired spin (red) density at the oxygen atom for a series of para-
substituted phenoxy radicals. The % change in both parameters is referenced to values obtained for the

unsubstituted compound (i.e. R = H, where spin = 0.43937 (it-electrons and charge = -O.48359 electrons). A
positive change indicates an increase in the magnitude of negative charge or an increase in spin density.

In fact, there are far more examples evident where shifts in spin and charge density
oppose each other rather than track together. In principle, there are four regimes of spin
and charge polarization that could be sampled corresponding to an increase and/or
decrease of spin and/or charge at the oxygen atom. Figure 2-1 illustrates that three of
these four regimes were realized, whereas none of the substituents we chose gave rise to a

simultaneous increase in both spin and charge density at the oxygen atom. The degree of

26

polarization varies considerably depending on the specific identity of the substituent,
although the most significant effects appear to be associated with substituents that can
couple strongly to the n-system of the molecule. The magnitude of the shifts in spin and
charge density in some cases are substantial: dimethylarnine, for example, causes a nearly
20% reduction in spin density at the oxygen atom as compared to the unsubstituted
phenoxy radical, while at the same time serves to increase the negative charge density at
oxygen by over 10%. The vinyl group appears to induce a similar effect on spin, though
its influence on charge density is less pronounced. Even seemingly innocuous
substituents such as methyl and t-butyl are predicted to differentially polarize spin and
charge. It is interesting to note that, in general, the substituent effect on spin polarization
appears to be larger in a relative sense than the corresponding shift in charge, though
exceptions to this trend are apparent (e.g., R = N02).

Clearly, spin and charge are responding differently to the perturbations introduced
by the various substituents. As mentioned above, this simple observation addresses the
question of whether spin and charge can be differentially polarized, however, in order for
this to be of use the mechanism(s) of the effect must be understood.

B. Substituent Effects on Charge Density Distribution. Computational
methods have been employed by many different groups in an effort to describe charge
distributions in molecules.62'64 Of greatest relevance to our work is a study by Wu and

Lai, who employed density functional theory to examine the effect of para-substituents
on charge density polarization in phenoxy radicals.65 These authors demonstrated that the
change in charge density at the oxygen atom correlates well with the Hammett-Brown“66

polar substituent constant (of). The of constants derive from data on the solvolysis of t-

27

cumyl chlorides, an SNl reaction whose rate depends on the electrophilicity of the tertiary
carbocation associated with the transition state. Thus, in the most common
interpretation,67 the of constant reflects the ability of a substituent to donate or remove
charge density; based on its original definition,66 negative values of of correspond to
charge-donating substituents whereas positive values are considered electron-
withdrawing (all relative to hydrogen). The 0'er constants broadly encompass both
inductive (tr—donating or o-accepting) and resonance (it-accepting or n-donating) effects
associated with the substituent.

The results of our calculations fully support the work of Wu and Lai; a plot of the
change in charge density at the oxygen atom as a function of the of coefficient for each
substituent that we examined is shown in Figure 2-2. The correlation in Figure 2-2
indicates that the individual resonance and inductive effects of the substituents are
effective in modulating the charge density at the oxygen atom, with electron-donating
and withdrawing substituents increasing and decreasing, respectively, the negative charge
density at the oxygen atom. Overall, the results in Figure 2—2 conform to our qualitative
expectations given the electronegativity of oxygen and the tendency of n-donors to
increase electron density within an aromatic ring system. On a more quantitative level,
the linearity of the correlation is consistent with what one expects from a Hammett-type
plot, though the underlying reason for the empirical success of such plots is the subject of
continued debate.68 Experimental evidence indicating a correlation between (I; and the
reduction potentials for phenoxy radicals (i.e., PhO' + e' —> PhO’) supports the notion
that electron deficient radicals are stabilized by the transfer of negative charge density

from the substituent to the oxygen radical.69 The correlation in Figure 2-2 coupled with

28

the linear free energy relationship upon which the of constants are defined can therefore

be interpreted as reflecting the ability of the substituent to inductively and/or resonantly

stabilize the oxygen radical Via the modulation of charge density within the molecule.68

 

 

 

 

0.08
E \
u \ \
ad 0.06 - ' \
8 NMCZ \ INHZ
2 \
E 0.04 - \ \ OMe
a.) OE
5 OH I‘ t
\ ' l I
.«E‘ 0.02 - 3"”
g CH; t-butyl
g, 0.00 - F - {Cl
a H \
A: r- \
0 \..'CN
.5 -0.02 r- \
00 + \
g NO2
6 _0_04 . 1 n l . l r l . L .
-2.0 -l.5 -l.0 -0.5 0.0 0.5 1.0
oP+

Figure 2-2. Plot of the calculated change in charge density at the oxygen atom as a function of the Gp+
coefficient of the R-group in para—substituted phenoxy radicals. The value given is relative to that obtained
for the unsubstituted radical (i.e. R = H); a positive number indicates an increase in negative charge density

at the oxygen atom. The O'p+ values were taken from the review by Hansch, Leo, and Taft (cf. 4). See text
for further details.

C. Substituent Effects on Spin Polarization. The ability of substituents to
influence the distribution of spin density within an aromatic system can be understood
qualitatively by extending the resonance picture for the phenoxy radical as shown in
Scheme 2-3. In the case of n-donors (A), adherence to the Pauli principle requires that the
interaction between the lone-pair of the donor and the aromatic ring involve the effective

transfer of an electron whose polarization is opposite to that associated with the carbon at

29

the para position: arbitrarily assigning the spin density within the ring as or, substituents

with lone pairs having n-symmetry can therefore be viewed as fi-spin donors.

 

Scheme 2-3. Spin delocalization mechanisms in phenoxy radicals: (A) B-spin donor (n-donors); (B)
conjugative a~spin acceptor (Tr-acceptors); (C) hyperconjugative B—spin donors (CS-donors).

30

The consequence of this interaction is the creation of residual or-spin on the R group and
a net redistribution of spin density out of the aromatic ring.70 Taken to the extreme,
complete transfer of the electron from the donor gives rise to the zwitterion depicted on
the right side of Scheme 2-3 A, a description reminiscent of the three-electron bonding
schemes often invoked to describe such systems.71 Groups that can act as n-acceptors
(e.g., vinyl) can be described in a similar manner (B). Their ability to conjugate to the
ring lends to their depiction as ot-spin acceptors, however, the net effect in terms of the
redistribution of spin polarization to the periphery of the substituent is largely the same as
seen for n-donors.

The mechanism by which o-donors/acceptors influence spin polarization is not as
straightforward but can still be viewed in terms of a direct orbital interaction between the
substituent and the radical-containing n-system of the ring through hyperconjugation (C).
In this context, aliphatic groups such as CH3 can be considered B-spin donors wherein a o
orbital of the OH fragment plays a role similar to that depicted for the it-donors. An
important difference is that the net transfer of l3-spin density from the C—H fragment to
the para-carbon will serve to destabilize the bonding within the substituent by virtue of
delocalization of the bonding pair of electrons to the phenoxy p-orbital: this could
explain, in part, the generally smaller polarization effect observed for the 0'-
donor/acceptor-substituents relative to their it-based counterparts.

Common to all three of the spin polarization mechanisms illustrated in Scheme 2-
3 is the ability of the substituent to couple into the 1: system of the phenoxyl ring to
redistribute spin density out of the aromatic ring (and, given the resonance structures in

Scheme 1, away from the oxygen atom). Although useful from a qualitative perspective,

31

the scenarios presented above still do not provide sufficient insight as to the relative
effectiveness of one substituent versus another at polarizing spin density, nor do they
allow us to understand why spin and charge behave independently in terms of the
directionality of their polarizations.

Previous Studies. A number of groups have considered substituent effects on spin
polarization by correlating spin density distribution with various physical properties (both
calculated and experimental). Both Wu and Lai44 and Brinck et al.72 have examined the
link between spin density polarization and thermodynamics by calculating the O—H bond
dissociation energy (BDE) of substituted phenols. A correlation between decreasing BDE
and the presence of it-donors was noted suggesting stabilization of the radical is
associated with increased spin delocalization of the odd electron.72 The influence of para
substituents on the phenoxy radical has also been investigated in an AM] study by G. P.
Bean,73 who likewise focused on bond dissociation energies of substituted phenols. Bean
concluded that the relative stability of the radicals could be understood in the context of

74'” a concept first introduced by Dewar78 in 1952 in which a

the captodative effect,
combination of donors and acceptors within the same compound led to a “push-pull”
effect that ultimately stabilized the radical. Viewing the oxygen radical as a strong
electron-withdrawing group, this model holds that a strong electron-donating group in the
para-position of the ring leads to stabilization of the radical (conversely, an electron-
withdrawing group would destabilize the radical via an anti-captodative effect). The
validity of this idea garners support from the results of Wu and Lai and Brinck et a1.“72 —

and to some extent our own — insofar as strong donors are observed to promote

redistribution of spin polarization away from the oxygen atom and toward the substituent.

32

Correlations with bond dissociation energies represent one of many ways in
which workers have sought to systematize the effects of substituents on spin polarization.
These efforts have resulted in numerous Hammett-type radical scales: among the more

793 . .
’0 Wthh 15 based on

commonly employed are (l) the 0'“. scale of Jiang and Ji,
cycloaddition reactions of or,B,B-trifluorstyrenes, (2) the oa' scale of Arnold,7°’81 which
relates spin density to hyperfine interactions in substituted benzyl radicals, (3) 06, which
Creary used to describe the rearrangement of 2-aryl-3,3-

82“” (4) Jackson’s of parameter, which was derived

dimethylmethylenecyclopropanes,
from the thermal decomposition of dibenzyl mercurials,84'85 (5) Fisher’s of parameter
based on the bromination of 4-substituted 3-cyanotoluenes by NBS,“87 and (6) the AD
scale of Adam,66 which correlates spin density with zero-field splitting in substituted 1,3-
diarylcyclopentane triplet diradicals. The success of these various parameters at
describing substituent effects on spin polarization in a broader context has been mixed.
For example, Wu and Lai modeled the change in spin density (AS) at the oxygen atom in
their study using 611'; the authors acknowledged that the correlation they reported was not
as compelling as what they observed between opi and charge density, although it was
clear that spin delocalization was maximized with n-donor ligands. It should be noted
that these authors made a point of noting that spin delocalization effects in the phenoxy
system could be quite different than in the benzyl radicals upon which the 611' constants
are based, implying that a correlation between spin density and 611' perhaps should not be
expected in the case of substituted phenoxy radicals. In Figure 2-3 are shown our

calculated results for the 15 substituents examined in this study plotted against two of the

33

Hammett parameters most commonly invoked to describe substituent effects on spin

distribution.

=H)

Spin Density at Oxygen (Relative to R

=H)

Spin Density at Oxygen (Relative to R

Figure 2-3. Plots of the change in spin density at the oxygen atom of para-substituted phenoxy radicals as

 

 

 

 

 

 

 

 

0.02
\ C53
0.00 - \IH
\
I \ t-butyl NO
F x I I I 2
-0.02 — Me' \ \TMS
Cl Br \ ‘CN
-0.04 - . ‘ \
, OMe \ \
\
-0.06 — \
\
\
\
-0.08 - - ‘
r NMe2
_0_10 I I I I L I I
-0.2 0.0 0.2 0.4 0.6 0.8 1.0
6.1.1.
H CF
0.00 e . ' 3
F. No2
CH3 I
I
-0.02 — _ BrIICl
L ‘ “ ~ ~ ~ ~ CN
OH 7 ‘ " ~ -.
‘0.04 " .- ~ ~ ~ .4
OMe
-0.06 —
NHZI
-0.08 e I
NMe2
J I I I I 4 A I_ L
-0.2 0.0 0.2 0.4 0.6 0.8 1.0
AD

21 function of the 0’“. (top) and AD (bottom) Hammett-type spin delocalization parameters.

34

It can be seen upon inspection that neither of these parameters (as well as 06 and 00", not

shown) yields a particularly good correlation. Indeed, the AD parameter, which we
initially believed held the greatest promise due to the fact that zero-field splitting in
diradicals should correlate quite well with unpaired spin density, actually does the worst
in terms of describing the polarization of spin away from the oxygen atom in the phenoxy
radical. The correlation appears to be strongest with the 611' parameter —— similar to what
was reported by Wu and Lai —— though in our case this observation is clearly dependent
upon inclusion of the data on the dimethylamino derivative. Figure 2-3 reinforces the
suggestion by Wu and Lai that while these various spin delocalization scales may work
quite well for their parent systems, they do not appear to be transferable between
different systems to any significant degree.72 We therefore looked to our calculations to
see if there was another, more robust marker for spin polarization that could provide us
with greater insight into the origin of these substituent effects.

HOMO-L UMO Gap Correlation. After an extensive analysis of our data, we
uncovered an empirical relationship between the polarization of spin density within the
compounds being examined and the energy gap between their highesteoccupied and
lowest-unoccupied molecular orbitals. Before we describe these results, we must first
define what we mean by the LUMO of a molecule within the context of this study.

As mentioned in the Experimental section, we carried out our calculations using a
spin-unrestricted formalism. This approach is necessary in order to allow the spatial
characteristics of each spin polarization to behave independently, but a drawback of this
method is that the orbitals produced within this formalism are one-electron constructs. As

a result, each one-electron orbital of the molecule is a distinct eigenvector with its own

35

eigenvalue. The challenge is therefore to correlate these one-electron orbitals with a two-
electron orbital picture of the kind that chemists typically invoke to describe bonding
interactions in molecules. Amos and Hall have argued that this can be done in a
straightforward manner provided that the spatial overlap of a given pair of or- and B-spin
orbitals approaches unity.”90 In this circumstance, the a- and B-spin one-electron
orbitals calculated within the spin-unrestricted formalism can be thought of as describing
essentially the same molecular orbital (i.e., the corresponding two-electron orbital),
differing only in the polarization of the electron.

This idea can be understood by examining the relevant wavefunctions of one of
the compounds fiom our study. Energetically, the true HOMO-LUMO gap in an S = '12
system within a spin-unrestricted formalism is typically the a-HOMO-B-LUMO gap: this
was found to be the case in all of the molecules whose properties are summarized in
Figure 2-4. In Figure 2-4 are shown the oc-HOMO, B-LUMO, and ct-LUMO of the p-

NHz-phenoxyl radical.

      

J
(it-HOMO B-LUMO or-LUMO

Figure 2-4. Comparison of the OL-HOMO, B-LUMO, and OL-LUMO Kohn-Sham molecular orbitals of the

p-NH2-phenoxy radical. The orbitals were generated at the UB3LYP/6—311G** level of theory and
visualized using GaussView with an isosurface value of 0.020.

36

It can be seen that the a-HOMO and B-LUMO of p-NHz-phenoxyl radical are essentially
identical in terms of their spatial properties,“ whereas the or-LUMO reflects completely
distinct bonding characteristics. Within the context of Amos and Hall’s corresponding
orbital transformation,”'90 the a-HOMO and B-LUMO, despite having different
eigenvalues in the spin-unrestricted calculation, can be viewed as representing the same
molecular orbital when projected onto a two-electron picture. An analogous result was
found for all 15 compounds examined in this study. We therefore believe that the OL-
LUMO represents the physically significant orbital for defining the HOMO-LUMO gaps

of these open—shell systems.”96

A plot of the change in spin density at the oxygen atom versus the HOMO-LUMO

gap of all 15 of the p-substituted phenoxy radicals we examined is shown in Figure 2—5.

 

 

 

 

 

 

E ,’
ll

8 I /

O) I

.5. v NO F , ’ IrMs

% -0.02 - 2 lat-butyl

E5, ‘ ,BIFCI e

E / r

SE 0” ,’/ 'CN

c -0.04 - OMe: ,’

9,9,, OEt ’

>,. /

X x

‘3 , ’

5 ~0.06 — / , ’

t6 , ,

3;: * I ’ I NH2

3 -0.08 —/ INMe2

a .

'5. I . 4"?“le . L . l . I r l . I
m 4.4 4.6 4.8 5.0 5.2 5.4 5.6 5.8

a—HOMO-a—LUMO Gap (eV)

Figure 2-5. A plot of the calculated changes in Spin density at the oxygen atom versus the a-HOMO-OL-
t’hUMO gaps for a series of para-substituted phenoxy radicals calculated at the UB3LYP/6-311G** level of
eory.

37

Our results clearly show a strong correlation between these two variables.97 While this
correlation is not as strong as was observed between charge density and (SW (Figure 2—2),
a comparison of these results with Figure 2-3 illustrates that the HOMO-LUMO gap is a
significantly better marker for changes in spin density across this series than any of the
commonly invoked Hammett-type radical constants that have been described in the
literature.98 Molecules with smaller HOMO-LUMO gaps are seen to exhibit a greater
degree of spin polarization away fiom the oxygen atom: closer inspection reveals that this
region is dominated by n-donating substituents, whereas larger HOMO-LUMO gaps are

associated with substituents whose primary influence is o-based. It is interesting to note

that, although the general grouping of substituents (e.g., it-donors, o-acceptors, etc.) is
similar to what was observed for charge polarization effects, the specific ordering within
these subgroups is different, a point that underscores the general result from Figure 2-1
concerning the distinct nature of charge and spin polarization in this system. The
correlation is clearly not perfect: N02 and vinyl derivatives all fall significantly off the
line, the former having a smaller gap than expected given the spin density at oxygen
calculated for p-NOz-phenoxyl, while the vinyl-containing compound exhibits an
anomalously large splitting. Nevertheless, the overall trend evident in Figure 2-5
suggests that the factors influencing spin polarization and the HOMO-LUMO gap in this
series of compounds are related.
D. Mechanism of Spin Polarization. The overall correlation illustrated in Figure
2-5 can be explained in the context of a simple molecular orbital scheme (Figure 2-6). In

accord with the resonance picture illustrated in Scheme 2-3, the dominant interaction is

38

proposed to occur between the lone pair associated with the rt-donor and the n-orbital(s)

containing the unpaired electron of the phenoxy radical.“99

—--'- \PQ-LUMO __-___-, llICI-LUMO
‘fphenoxy-LUMO ‘i’phenoxy-LUMO AEZ LL
AEI \‘xWOL-HOMO
. .‘ Wet-HOMO " “

Figure 2-6. Molecular orbital-based description of the spin delocalization mechanism proposed for n-
donor-containing phenoxy radicals. The interaction of the (bu—donor orbital of the substituent with the

phenoxy 9phenoxy-SOMO results in a net one-electron stabilization. The degree of interaction between
these two orbitals dictates the OL—HOMO-Ot-LUMO energy gap, which in turn alters the composition of the
Ot-HOMO and thus the spatial distribution of unpaired spin density.

Assuming that the degree of spatial overlap between the donor and the n—system of the
ring is roughly invariant, the extent of destabilization of the HOMO will be dictated
primarily by the relative energies of the (imam, and ¢phen0xy-SOMo orbitals: as these two
orbitals become more energetically equivalent the degree of this destabilization will
increase. This will have two immediate consequences: ( 1) The wavefunction
corresponding to the SOMO of the adduct will exhibit an enhanced contribution from the
substituent, the net effect of which will be to redistribute the electron density associated

' with the unpaired electron onto the substituent and away from the oxygen atom; and (2)

39

The HOMO-LUMO gap will decrease provided the a-LUMO of the phenoxy radical is
essentially non-bonding with respect to the substituent.

The results from the calculations we have carried out provide considerable
support for this picture. For example, it can be seen in Figure 2-5 that decreasing the
electronegativity of the donor atom from oxygen in R = OMe to nitrogen in R = NMez
leads to a smaller HOMO-LUMO gap, consistent with the notion that the lone pair of
nitrogen in dimethylamine should present a better energetic match with the aromatic ring

than does the methoxy group. More compelling is the fact that the redistribution of
electron density due to mixing between the ‘l’phenoxy-SOMO and ¢n-donor orbitals
suggested in Figure 2-6 is reflected in the spatial characteristics of the wavefunctions. In
Figure 2-7, the Wet-HOMO Kohn-Sham molecular orbitals of both p—NI—lz- and p-CF3-

phenoxyl are shown as an example underscoring the significant contribution of the NH;
group to the overall composition of the SOMO orbital as compared to the (relatively)

small perturbation induced by the CF; substituent.

   

p-CF3-phenoxyl p-NHz-phenoxyl

Figure 2-7. Comparison between the Kohn-Sham Ot-HOMO molecular orbitals of p-CF3- and p‘NHZ'
substituted phenoxy radicals.

40

 

CL‘

-1:

It:

Quantitative support for this model is afforded by examining the composition of the

‘I’a-H0M0 orbitals explicitly. According to our model, this wavefunction is
won-HOMO = C1°‘i’phenoxy-SOMO + CZ‘¢n-donor (1)

where of + of = 1. As the 07,40,101. orbital increases in energy (i.e. shifting from left to

right in Figure 2-6), more extensive orbital mixing should result in an increase in c; at the
expense of c]. The data in Table 2-1 confirms these expectations: as the HOMO-LUMO

gap decreases, the wavefunction corresponding to the HOMO exhibits a systematic
increase in the percent contribution from the substituent (%R) and a corresponding
decrease in oxygen character (%O). Since the unpaired electron carries all of the excess
spin density of the molecule, the delocalization of the SOMO onto the substituent implied
by its increased contribution to the wavefunction has the net effect of shifting
polarization of spin away from the oxygen atom. This simple MO explanation, which

75.77

essentially expands upon the concept of the captodative effect and three-electron

bonding,”100 thus accounts in a general way for the correlation evident in Figure 2-5.
Deviations from the HOMO-L UMO Gap Correlation. Implicit in the model just

described —— and in particular the use of the HOMO-LUMO gap as a marker for spin

polarization —— is the notion that the Wa-LUMO is not perturbed upon incorporation of the
substituent. Defining the or-LUMO as,
Wa-LUMO : c3'(i’phenoxy-LUMO + C4'¢7t*-substituent (2)

we can see from Table 2-1 that c4 is, in fact, either zero or relatively small (i.e., < 10%)

for most of the substituents included in our study. Two glaring exceptions to this are N02

41

and vinyl: these groups contribute 69% and 48% to their respective LUMOs. We now
consider each of these cases to determine whether their properties can be understood

within the context of the model we have developed.

Table 2-1. Percent atomic contributions for oxygen (%0) and the substituent (%R) group in the (it-HOMO
and a-LUMO Kohn-Sham MO’s for a series of para-Phenoxy radicals.

R °/o O % R % O % R AS

 

 

or-HOMO (X-HOMO ct-LUMO or-LUMO
Reference
H 27.28 0.03 0 0 0
it-donor
OH 22.23 15.48 0 0 -0.03 82
OMe 21.31 19.05 0 2.99 -0.041 3
OEt 21.1 19.86 0 3.41 -0.0427
NH2 19.95 22.61 0 0 -0.071 1
NMez 18.01 31.9 0 7.91 -0.0789
n-donor, o-acceptor
F 24.51 9.24 0 0 -0.0098
Cl 21.06 18.01 0 0 -0.0222
Br 20.07 21.78 0 0 -0.0243
n—acceptor
N02 24.98 8.37 4.58 69.04 -0.01 31
CN 23.42 11.97 10.58 15.95 -0.0321
n-acceptor, o-donor
vinyl 20.17 21.31 7.25 47.9 -0.0863
o-donor
Me 24.72 5.84 0 6.24 -0.0184
t-butyl 24.07 9.33 0 14.94 -0.0169
TMS 24.52 9.27 0 10.7 -0.0145
o-acceptor
CF3 26.22 4.97 0 0.45 0.0031

 

aw 2 2 .
[Zn a.o. / 21121119.] * 100 = % contribution. in ac. is the sum of the squares of the atomic orbital

. . 2 . .
coefficrents of the atom or group of interest and Zn mo. 18 the sum of the squares of all atomic orbital

coefficients in a specific molecular orbital.

42

The results we obtained for p—NOz-phenoxyl constitute the most significant
deviation from the HOMO-LUMO gap correlation illustrated in Figure 2-5. Specifically,
our calculations indicate that the HOMO-LUMO gap for this compound is substantially
smaller than expected based on the calculated redistribution of spin density away from
the oxygen atom. The origin of this result can be understood quite easily by examining
the frontier orbitals of this compound. Figure 2-8 presents a comparison of the LUMOs

for phenoxyl and p—NHz-phenoxyl with the LUMO and LUMO+1 of p-NOz-phenoxyl.

      

J
J

phenoxyl para-NHz-phenoxyl

 

para-NOz—phenoxyl:

or-LUMO a-LUMO + 1

Figure 2-8. Comparison of the a-LUMO orbitals for phenoxyl and p-NHz-phenoxyl radicals with the
LUMO and LUMO+1 orbitals of p-NOz-phenoxyl.

These plots clearly Show that the character of the LUMO changes upon incorporation of
the N02 group, wherein the orbital corresponding to the LUMO in most of the other
compounds in our study is now the next-highest energy orbital (LUMO+1) in this system.
This situation arises due to the presence of a low-lying 1t* acceptor orbital associated
with the N02 group, allowing for a new interaction between the substituent and the
LUMO+1 of the free phenoxyl moiety and the stabilization of that orbital in p-NOz—

phenoxyl; a qualitative molecular orbital diagram depicting this situation is shown in

Figure 2-9.

 

_L ””””” Tot-HOMO

(i’phenoxy-SOMO “‘ “

“‘\ ’ \J—j-i— Til-N02

Figure 2-9. Proposed molecular orbital diagram for p-NOz-phenoxyl. The presence of a low-energy
empty 7t* orbital on the N02 group leads to a change in the nature of the LUMO and is responsible for the
deviation exhibited by p-NOz-phenoxyl from the correlation illustrated in Figure 5.

It should be noted that the polarization of spin density is still driven by the character of

the HOMO as described above: Table 1 shows that the contribution of N02 to the HOMO

44

is comparable to other substituents exhibiting similar degrees of spin polarization. In this
regard, we believe that the model we have developed remains conceptually robust, but
the HOMO-LUMO gap correlation breaks down in the case of N02 due to the presence
of a new interaction not present in the other compounds and thus not accounted for in the
simple two-orbital, three-electron scheme depicted in Figure 6.

The vinyl group also shows significant contributions to the LUMO as well, as
exhibiting the same LUMO/LUMO+1 inversion as seen for p-NOz-phenoxyl. However,
the situation for p-vinyl-phenoxyl is fimdamentally different: whereas the N02 adduct
displays less spin polarization than the HOMO-LUMO gap would suggest, the vinyl
group appears to be just the opposite. In fact, the results from Figure 1 show that the
vinyl group is the most effective at polarizing unpaired spin away from the oxygen atom
of any substituent we examined. This enhanced delocalizing ability of the vinyl group has

101 The contrast

also been noted by Radom et al. in a study of substituted methyl radicals.
between N02 and vinyl implies that a-spin resonance delocalization coupled with B-spin
donation is the best mechanism to delocalize spin, but the ability of N02 to delocalize
spin is attenuated compared to the vinyl group. This attenuation is most likely due to the
inductive electron withdrawing ability of these substituents: it is known in the
literaturemz'lO7 that an electron-withdrawing group usually plays a duel role in radical
stability. A withdrawing group can stabilize a radical by resonance delocalization (1t-
withdrawing) and at the same time destabilize the radical by electron induction (0-
withdrawing). Destabilization by electron induction is not surprising due to the electron
deficient nature of the oxygen radical center, and can be thought of as an unfavorable

108

“pull-pull” dipole-dipole interaction. Therefore, we believe that the enhanced spin

45

delocalization in the vinyl group can be linked to two contributing factors: (1) o donation
by the vinyl group in the absence of inductive destabilization, the combination of which
increases B‘spin density at the oxygen center, and (2) stabilization of the radical by
resonance delocalization of a spin density. The latter appears to be a particularly
effective mechanism for delocalizing the spin in these systems.

Using the vinyl substituent as a starting point, we sought even stronger spin
delocalizing substituents. To this end, there are substituents known as “super radical
stabilizers”, such as p-N=N(O)-‘Bu, p-N=N-Ph, and p-CH=N(O)-‘Bu.'09‘l '0 Incorporation
of the CH=N(O)-‘Bu group at the para position phenoxy radical decreased unpaired spin
density at the oxygen atom by 60% in part due to an increase in B-spin density at the
oxygen atom, in line with our expectations based on the model just described.

o-donors and d-acceptors. The o-donor substituents CH3, t-butyl, and TMS
decrease the spin density at the oxygen atom, but to a lesser extent than that of the T!-
donors. This trend in spin delocalization is also mirrored in the larger HOMO-LUMO
gaps for these substituents as compared to the n—donors. The delocalization of spin in
these substituents can be accounted for by the hyperconjugation B-spin donation
mechanism already discussed above and depicted in Scheme 3. A three-electron, two-
orbital interaction can again be invoked to describe the spin delocalization in these
substituents, where the interaction between a filled o-orbital with the partially filled 1t-
orbital of the phenoxy radical leads to a stabilizing interaction.99 The degree of
stabilization between the substituents is small, and literature reports suggest that
branching of the alkyl chain will lead to less effective hyperconjugation.70 This is exactly

what is observed in the phenoxy radical complexes: the composition of the relevant

46

molecular orbitals indicates delocalization of the wavefunction onto the substituent via

mixing between the ‘l’phenoxy-SOMO and a filled c-orbital of the cr-donor substituents.

With regard to o-acceptor groups, studies by Brinck et a1.72 as well as Bean73

indicate that CF3 is a radical destabilizing substituent which in the present context
corresponds to increasing the spin density at the oxygen atom. Bean suggests this can be
attributed to an anti-captodative effect arising from a “pull-pull” effect in the system due

to decreased conjugation. Brinck et a1. points out that this substituent has been shown to

be destabilizing in other radical delocalization scales such as do: and 611.. A similar

mechanism is likely operative in p-F-phenoxyl, since 011', 6a., and 66 have shown p—F
to be destabilizing. It should be noted that, in the case of CF3, we observed a functional
dependence as to whether CF3 increased or decreased spin density at the oxygen:
calculations at the UBLYP/63110” level gave AS = -0.002, whereas a value of AS =
0.003 was obtained with the hybrid UB3LYP functional. Given these differences, we are

reluctant to draw any distinctive conclusions concerning the role of o—acceptors in this

system.

2.4 Concluding Comments

The goal of this study was to examine whether spin and charge density could be
independently manipulated in a simple molecular system, and to understand the
underlying mechanism for such a process. Using para—substituted phenoxy radicals as a

test case, we carried out calculations using a wide range of substituents. A strong

47

correlation between the Hammett a; polar substituent constants and changes in charge
density at the oxygen atom support previous work suggesting that resonance and
inductive effects are largely responsible for charge polarization. Simple electronegativity
arguments can be applied, wherein electron acceptors withdraw charge density and
electron donors increase the charge density at the oxygen atom in the substituted phenoxy
radicals. In contrast, the spin density at the oxygen atom behaved in a manner quite
distinct from the charge; none of the commonly invoked spin polarization parameters
were adequate for understanding our results. A correlation was established between the
a-HOMO-a—LUMO gap and the spin density at the oxygen centers whereby
increasing/decreasing spin delocalization coincided with a decrease/increase in the
a-HOMO-or-LUMO gap. A simple three-electron, two-orbital model was used to explain
the correlation between the changes in spin density at the oxygen atom with the

a-HOMO-a-LUMO gap for the n-donor para-substituted phenoxy radicals. The amount

of mixing between the ¢phenoxy_SOMO and the ¢1t~d0n0r was identified as the dominating

mechanism that controls both the HOMO-LUMO gap and the spin delocalization in these
systems. n—acceptor substituents were found to deviate from the correlation due to

additional interactions from unoccupied 1t* orbitals of the substituent with the

‘l’phenoxy-LUMOH, giving rise to a change in the character of the LUMO relative to other

members of the series. Substituents that can serve as both a n-acceptor and a cr-donor
(e.g. vinyl) were found to have the greatest propensity for redistributing spin density

within the molecule.

48

Our results suggest that it should be possible to differentially polarize spin and
charge through synthetic means by judicious choice of substituents. This could be of
considerable use in the field of molecular magnetism, for example, whereby control of
spin density at the radical centers could be used to manipulate intramolecular exchange
interactions. One could also envision use of this approach in the design of catalysts for
radical-radical coupling reactions, in which spatially distinct polarization of or and [3 spin
density could facilitate the formation of 0 bonds. Efforts to implement some of the ideas

generated from the results described in this report are on going.

49

2.5 References and Notes

10

ll

12

13

14

15

16

17

Wolf, S. A.; Awschalom, D. D.; Buhrman, R. A.; Daughton, J. M.; von Molnar,
S.; Roukes, M. L.; Chtchelkanova, A. Y.; Treger, D. M. Science 2001, 294, 1488.

Fert, A. Rev. Mod. Phys. 2008, 80, 1517.
Bogani, L.; Wemsdorfer, W. Nature Mater. 2008, 7, 179.
Hansch, C.; Leo, A.; Taft, R. W. Chem. Rev. 1991, 91, 165.

Ewing, D. F. In Correlation Analysis in Chemistry: Recent Advances; Chapman,
N. B., Shorter, J ., Eds; Plenum Press: New York, 1978.

Yang, J. S.; Liu, K. T.; Su, Y. J. Phys. Org. Chem. 1990, 3, 723.

Pratt, D. A.; Dilabio, G. A.; Mulder, P.; Ingold, K. U. Acc. Chem. Res. 2004, 37,
334.

Sekigawa, K. T etrahedron 1972, 28, 505.
Fernandez, 1.; Frenking, G. J. Org. Chem. 2006, 71, 2251.
Wiberg, K. B. J. Org. Chem. 2002, 67, 4787.

Liu, L.; Fu, Y.; Liu, R.; Li, R. Q.; Guo, Q. X. J. Chem. Inf Comput. Sci. 2004, 44,
652.

Krygowski, T. G.; Stepien, B. T. 2005, 105, 3482.
Genix, P.; Jullien, H.; Le Goas, R. J. Chemometrics 1996, 10, 631.
Head-Gordon, M. J. Phys. Chem. A 1996, 100, 13213.

Koch, W.; Holthausen, M. C. A Chemist’s Guide to Density Functional Theory;
Weinheim; Chichester: Wiley-VCH, 2000.

Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and Molecules;
Clarendon: Oxford, 1989.

Claiser, N.; Souhassou, M.; Lecomte, C.; Gillon, B.; Carbonera, C.; Caneschi, A.;

Dei, A.; Gatteschi, D.; Bencini, A.; Pontillon, Y.; Leliévre-Bema J. Phys. Chem.
B 2005, 109, 2723.

50

 

 

4

‘J;

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

Zheludev, A.; Barone, V.; Bonnet, M.; Delley, B.; Grand, A.; Ressouche, E.; Rey,
P.; Subra, R.; Schweizer, J. .1. Am. Chem. Soc. 1994, 116, 2019.

McConnell, H., M, J. Chem. Phys. 1956, 24, 764.
Remenyi, C.; Reviakine, R.; Kaupp, M. J. Phys. Chem. B 2007, 111, 8290.

Nguyen, M. T.; Creve, S.; Vanquickenbome, L. G. J. Phys. Chem. A 1997, 101,
3174.

Reiher, M. Faraday Discuss. 2007, 135, 97.

Pillet, S.; Legrand, V.; Weber, H. P.; Souhassou, M.; Letard, J. F.; Guionneau, P.;
Lecomte, C. Zeitschrift Fur Kristallographie 2008, 223, 235.

Pierpont, C. G. Coord. Chem. Rev. 2001, 216-217, 99.

Pierpont, C. G.; Buchanan, R. M. Coord. Chem. Rev. 1981, 38, 45.

Pierpont, C. G.; Lange, C. W. Prog. Inorg. Chem. 1994, 41, 331.

Kaim, W. Coord. Chem. Rev. 1987, 76, 187.

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

Adam, W.; Borden, W. T.; Burda, C.; Foster, H.; Heidenfelder, T.; Heubes, M.;
Hzl‘gyeslt9,3D.; Kita, F.; Lewis, S.; Scheutzow, D.; Wirz, J. J Am. Chem. Soc. 1998,

West, A. P., Jr.; Silverman, S. K.; Dougherty, D. A. J. Am. Chem. Soc. 1996, 118,
1452.

Geise, C. M.; Hadad, C. M. J. Org. Chem. 2000, 65, 8348.

Mitani, M.; Yamaki, D.; Takano, Y.; Kitagawa, Y.; Yoshioka, Y.; Yamaguchi, K.
J. Chem. Phys. 2000, 113, 10486.

Baton, G. R.; Eaton, S. S. Acc. Chem. Res. 1988, 21, 107.
Eaton, S. S.; Eaton, G. R. Coord. Chem. Rev. 1988, 83, 29.

Shultz, D. A.; Bodnar, S. H.; Lee, H.; Kampf, J. W.; Incarvito, C. D.; Rheingold,
A. L. J. Am. Chem. Soc. 2002, 124, 10054.

Shultz, D. A.; Fico, R. M., Jr.; Bodnar, S. H.; Kumar, R. K.; Vostrikova, K. E.;
Kampf, J. W.; Boyle, P. D. J. Am. Chem. Soc. 2003, 125, 11761.

51

37

38

39

4O

41

42

43

44

45

46

47

48

49

50

51

52

Shultz, D. A.; Fico, R. M., Jr.; Lee, H.; Kampf, J. W.; Kirschbaum, K.; Pinkerton,
A. A.; Boyle, P. D. J. Am. Chem. Soc. 2003, 125, 15426.

Shultz, D. A.; Sloop, J. C.; Coote, T.-A.; Beikmohammadi, M.; Kampf, J .; Boyle,
P. D. Inorg. Chem. 2007, 46, 273.

Stone, T. J .; Waters, W. A. J. Chem. Soc. 1964, 213.

Waters, W. A.; Stone, T. J. Proc. Chem. Soc. London 1962, 253.

Dixon, W. T.; Norman, R. O. C. J. Chem. Soc. 1964, 4587.

Scheffler, K.; Stegmann, H. B. Elektronenspinresonanz; Springer: Berlin, 1970.
Qin, Y.; Wheeler, R. A. J. Chem. Phys. 1995, 22, 1689.

Lai, D. K. W.; Wu, Y. J. Org. Chem. 1996, 61, 7904.

Kawai, A.; Hirakawa, M.; Abe, T.; Obi, K.; Shibuya, K. J. Phys. Chem. A 2001,
105, 9628.

Broze, M.; Luz, Z. J. Chem. Phys. 1969, 51, 738.
Broze, M.; Luz, Z. J Chem. Phys. 1969, 51, 749.
Hirota, N. J. Chem. Phys. 1962, 3 7, 1884.

Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann, R. E.;
Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain,
M. C.; Farkas, 0.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.;
Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.;
Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.;
Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko,
A.; Piskorz, P.; Komaromi, 1.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.;
Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.;
Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.;
Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.4,
Gaussian, Inc.: Pittsburgh, PA, 1998.

Becke, A. D. Phys. Rev. A 1988, 38, 3098.
Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 3 7, 785.

Becke, A. D. J Chem. Phys. 1993, 98, 5648.

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

Frisch, M. J .; Frisch, A. Gaussian 98 User 's Reference; Gaussian, Inc.: Pittsburgh,
1998.

Glendening, E. D.; Reed, A. E.; Carpenter, J. E.; Weinhold, F. NBO 3.1,
Theoretical Chemistry Institute, University of Wisconsin: Madison, WI, 1996.

SPARTAN, 5.0 ed.; Wavefunction Inc.: Irvine, CA, 1997.

Gauss View, 2.1 ed.; Gaussian, Inc.: Pittsburgh, PA, 2001.

Baker, J .; Scheiner, A.; Andzelm, J. Chem. Phys. Lett. 1993, 216, 380.
Laming, G. J.; Handy, N. C.; Amos, R. D. Mol. Phys. 1993, 80, 1121.
Pierpont, C. G.; Attia, A. S. Collect. Czech. Chem. Commun. 2001, 66, 33.
Pierpont, C. G. Coord. Chem. Rev. 2001, 219, 415.

Poddel'ski, A.; Cherkasov, V. K.; Abakumov, G. A. Coord. Chem. Rev. 2009,
253, 291 .

Merino, G.; Vela, A.; Heine, T. Chem. Rev. 2005, 105, 3812.
Baerends, E. J.; Ellis, D. E.; Ros, P. Chem. Phys. 1973, 2, 41.
Wiberg, K. B.; Rablen, P. R. J Comp. Chem. 1993, 14, 1504.
Wu, Y.; Lai, D. K. W. J Org. Chem. 1996, 61 , 7904.

Brown, H. C.; Okamoto, Y. J Am. Chem. Soc. 1958, 80, 4979.

Correlation Analysis in Chemistry: Recent Advances; Plenum Press: New York
and London, 1978.

Shorter, J. Correlation analysis in oganic chemistry: an introduction to linear
flee-energy relationships; Oxford University Press: Oxford, 1973.

Lind, J .; Shen, X.; Eriksen, T. E.; Mere'nyi, G. J. Am. Chem. Soc. 1990, 112, 479.
Wayner, D. D. M.; Arnold, D. R. Can. J Chem. 1985, 63, 2378.

Bauld, N. L. Radicals, Ion Radicals, and Triplets: The Spin-Bearing
Intermediates of Organic Chemistry; Wiley-VHC, Inc., 1997.

Brinck, T.; Haeberlein, M.; Jonsson, M. J Am. Chem. Soc. 1997, 119, 4239.

53

73

74

75

76

77

78

79

8O

81

82

83

84

85

86

87

88

89

90

91

92

Bean, G. P. T ertahedron 2002, 58, 9941.
Viehe, H. G.; Janousek, Z.; Mere’nyi, R.; Stella, L. Acc. Chem. Res. 1985, 18, 148.

Viehe, H. G.; Mere'nyi, R.; Stella, L.; Janousek, Z. Agnew. Chem. Int. Ed. Engl.
1979, 18, 917.

Klessinger, M. Agnew. Chem. Int. Ed. Engl. 1980, 19, 908.
Sustrnann, R.; Korth, H.-G. Adv. Phy. Org. Chem. 1990, 26, 131.
Dewar, M. J. S. J Am. Chem. Soc. 1952, 74, 3353.

Jiang, X.-K. Acc. Chem. Res. 1997, 30, 283.

Jiang, X.; Ji, G. J Org. Chem. 1992, 57, 6051.

Dust, J. M.; Arnold, D. R. J Am. Chem. Soc. 1983, 105, 1221.
Creary, X. J Org. Chem. 1980, 45, 280.

Creary, X.; Mehrsheikh-Mohammadi, M. E.; McDonald, S. J Org. Chem. 1987,
52, 3254.

Dincturk, S.; Jackson, R. A.; Townson, H.; Agribas, H.; Bellingham, N. C.;
March, G. J. J. Chem. Soc. Perkin Trans. 2 1981, 1121.

Jackson, R. A.; Dincturk, S. J Chem. Soc. Perkin Trans. 2 1981, 1127.
Fisher, T. H.; Meierhoefer, A. W. J Org. Chem. 1978, 43, 221.

Fisher, T. H.; Meierhoefer, A. W. J Org. Chem. 1978, 43, 224.

Amos, A. T.; Hall, G. G. Proc. R. Soc. Ser. A. 1961, 263, 483.

King, H. F.; Stanton, R. E.; Kim, H.; Wyatt, R. E.; Parr, R. G. J Chem. Phys.
1967, 47, 1936.

Neese, F. J Phy. Chem. Solids 2004, 65, 781.
Ohta, K. J Mol. Struct. T heochem 2002, 587, 33.
There has been some discussion about the possibility and usefulness of the Kohn-

Sham orbitals,”97 but good approximations to experimental excitation energies
can be made from them. 8

54

93

94

95

96

97

98

99

100

101

102

103

104

105

106

107

108

109

Stowasser, R.; Hoffmann, R. J Am. Chem. Soc. 1999, 121, 3414.

Bickelhaupt, F. M.; Baerends, E. J. In Rev. Comput. Chem; Lipkowitz, K. B.,
Boyd, D. B., Eds.; Wiley-VCH: New York, 2000; Vol. 15, p 1-86.

Chong, D. P.; Gritsenko, O. V.; Baerends, E. J. J Chem. Phys. 2002, 116, 1760.
Savin, A.; Umrigar, C. J .; Gonze, X. Chem. Phys. Lett. 1998, 288, 391.

Hartee Fock calculations at the UHF/6-311g** level have also been done, and the
same correlation is observed between the a-HOMO-u-LUMO gap and the spin
density at the oxygen atom. The overall shapes of the (l-HOMO and a-LUMO
wavefimctions are also qualitatively similar.

We stress that the form of Figure 5 is not meant to imply that spin density is a
linear function of the HOMO-LUMO gap, merely that there appears to be a

correlation between the two variables.

Bemardi, F.; Epiotis, N. D.; Cherry, W.; Schlegel, H. B.; Whangbo, M.; Wolfe, S.
J Am. Chem. Soc. 1976, 98, 469.

Clark, T. In Substituent Effects in Radical Chemistry; Reidel: Dordrecht, 1986, p
49.

Henry, D. J.; Parkinson, C. J.; Mayer, P. M.; Radom, L. J Phys. Chem. A 2001,
105, 6750.

Cheng, J.; Liu, B.; Zhao, Y.; Wen, Z.; Sun, Y. J Am. Chem. Soc. 2000, 122,
9987.

Cheng, J .; Zhao, Y. T ertahedron 1993, 49, 5267.

Pasto, D. J .; Krasnansky, R.; Zercher, C. J Org. Chem. 1987, 52, 3062.

Fu, Y.; Liu, R.; Guo, Q.-X. J Phys. Org. Chem. 2005, 18, 529.

Bordwell, F. G.; Harrelson, J. A., Jr.; Zhang, X. J Org. Chem. 1991, 56, 4448.
Zipse, H. T 0p. Curr. Chem. 2006, 263, 163.

Bordwell, F. G.; Zhang, X.-M.; Satish, A. V.; Chen, J .-P. J Am. Chem. Soc. 1994,
116, 6605.

Creary, X. Acc. Chem. Res. 2006, 39, 761.

55

110 Creary, X.; Engel, P. S.; Kavaluskas, N.; Pan, L.; Wolf, A. J Org. Chem. 1999,
64, 5634.

56

Chapter 3. Differential Polarization of Spin and Charge Density in ortho-

semiquinone Radicals

3.1 Introduction

The results of our previous density functional theory (DF T) examination of
phenoxy radicals indicated that spin and charge density can be differentially polarized.1
The proper choice of substituent can lead to an increase or decrease in the spin population
at a paramagnetic center, and workers in the field of magnetism have suggested that the
spin density at two paramagnetic centers will be proportional to the magnitude of the
exchange coupling between those centersz'6 Extending the knowledge learned in the
phenoxy radical systems to design transition metal radical systems that exhibit tunable
electron exchange coupling is the overall goal of this project. Transition metal
complexes with phenoxy radicals are known, but they usually incorporate a macrocyclic
phenoxy moiety due to the inherent weaker donating ability of the neutral radical
ligand-F9 Incorporation of substituents in these phenoxy ligand systems may be difficult,
and other bidentate radical ligands were sought out. One requirement for synthesizing
transition metal radical complexes is that the radical ligands act as strong Lewis bases.
Increasing the Lewis basicity will allow for a stronger tendency for complexation of the

10 A type of ligand that suits

radicals to the metal by increasing the formation constant.
this requirement quite nicely is the o-catechol, which is the fully reduced member of the
redox series shown in Scheme 3-1. The o-quinone ligands can exist as o-benzoquinones

(Q), o-semiquinones (SQ) and o-catecholates (CAT), which are all related by the addition

or removal of an electron as seen in Scheme 3-1.

57

 

 

O O O
.____i’f_'__.. *6 _
-e‘ ‘ -6-
o 06 09
Quinone Semiquinone Catechol

Scheme 3-1. Redox chemistry of the quinone ligands.

The o-catechol type ligands are dianions, and a significant amount of negative charge

density is localized on the oxygen atoms.11

The large excess negative charge density at
the oxygen centers of these ligands makes them strong Lewis bases. Once a metal-
catechol complex is synthesized it can be easily oxidized to the semiquinone form, which
can effectively turn on the Heisenberg exchange coupling interaction when the radical is
bound to a paramagnetic metal. Transition metal catechol, semiquinone, and quinone
complexes are well known, and have been studied extensively in the literature for the past
40 years.”17 It has been discussed in this dissertation that Spin and charge density can be
controlled by substituents in the phenoxy radical. The next logical step is to examine the
effects of spin polarization in substituted o-semiquinone ligands and look for similar
trends as those found in the substituted phenoxy radicals.

The first part of this investigation will focus on substituted o-l,2-semiquinones
(SQ), and the second part will focus on substituted o-9,10-phenanthrenesemiquinones
(PSQ). These semiquinone ligands are synthetically available in the literature, and
transition metal complexes with these ligands can be made.18 Therefore, they will be an

excellent starting point to examine which substituents increase or decrease the spin

density at the oxygen centers of the semiquinone ligand. If a prescription for the

58

modulation of the spin density in semiquinone ligands can be found with our theoretical
examination, then it might be possible to implement these ideas in terms of ligand design.
The effects of substituents could then be experimentally determined by measuring spin

exchange constants or spin densities in transition metal systems.

3.2 Theoretical Calculations

All DFT and Hartree-Fock geometry optimizations, frequency, and single point
calculations were carried out with the same procedures as was described for the phenoxy
radicals in Chapter 2 of this thesis. Substituents were placed at the positions ortho and
meta (3,6) as well as para and meta (4,5) to the oxygens of the o-1,2-semiquinone
radicals (Scheme 3-2) and at the 2,7- and 3,6-positions for the 0-9,10-

phenanthrenesemiquinones.

R
0' R \ 0'
09 R 09
R
3,6-R-o-1 ,2-semiquinone 4,5-R-o-1 ,2—semiquinone

Scheme 3-2. Substituted o-l ,2-semiquinones.

In all cases spin and charge density populations were calculated at the oxygen atoms and
reported as the average value. The calculations were carried out assuming a molecular

charge of -l and a doublet spin state for all substituted semiquinones. Spin

59

contamination of the doublet state was found to be negligible in all cases, and the largest
deviation from the theoretical value of (Sz) was found to be 3 %.

The substituents examined in these calculations were selected to cover a wide

range of possible interactions with the semiquinone ring and included o-donors (Me), a

o-acceptor (CF 3), tr-donors (OH, OMe, and NHz), tr-acceptors (N02 and CN), n-donor/o-

acceptors (F, Cl, and Br), and a n-acceptor/o-donor (vinyl). In all cases the unsubstituted

semiquinone radicals (i.e., R = H) were used as a reference point for comparisons.

3.3 Results and Discussion

3.3.1 Substituted o-lJ-semiquinones

Geometry. The crystal structure for the 0-1,2-quinone19 ligand has been reported,
and Wheeler et al.11 have shown that the UB3LYP/6-31g* level yielded structures that
were nearly identical to the experimental X-ray data. This gives confidence that the o-
semiquinone structures calculated at the same level of theory will be accurate
representations of their true geometries. Based on comparisons to the unsubstituted
compound, our results indicate that the geometry of the o-semiquinone radicals in terms
of both bond distances and bond angles is I largely insensitive to changes in the
substituents on the ring. The most significant perturbation arises in the C-0 distance for
the 3,6-R—o-1,2-semiquinones and the geometries had variations in the range of i 0.03 A,
while smaller changes of i 0.01 A were observed for the 4,5-R-o-l,2-semiquinones.
Additionally, the calculated structures with the 3,6-0Me, -N02, and -NH2 substituents

were found to deviate from the semiquinone plane, and the 4,5-N02, OMe, vinyl, and

60

NHz 0-1,2-semiquinones substituents were also canted from the plane of the semiquinone
ring.

Spin and Charge Density Movement of Substituted 3,6-R-o—1,2-
semiquinones. A graphical depiction of our computational results concerning the change
in charge and spin density at the oxygen atom for all 3,6-R-o-1,2-semiquinones (3,6—SQ)
is shown in Figure 3-1. A positive inflection in Figure 3-1 indicates an increase in
spin/charge density at the oxygen center relative to R = H (spin = 0.26345 a-electrons

and charge == -O.6345 electrons), and a negative inflection indicates the opposite trend.

 

 

 

 

 

 

 

 

 

 

 

10 -
a O r-l 4
u H on OMe NH2 CH3
or P F
8 '10 ‘- Cl Br
0% P CF3
E
0) - r-
a 20
o ,
if
g '30 b
D
°\°
.40 _
NO2
Tt-dOHOFS, “tr-donors n-acceptors o-donor o-acceptor
o-acceptors

Figure 3—1. Spin density (red) and charge density (blue) polarization in 3,6-R-0-1 ,2-semiquinones.

In principle, there are four regimes of spin and charge polarization that could be sampled

corresponding to an increase and/or decrease of spin and/or charge at the oxygen atom.

61

Figure 3-1 illustrates that three of these four regimes were realized, whereas none of the
substituents we chose gave rise to an increase in charge density and decrease in spin
density at the oxygen atoms. The degree of polarization varies considerably depending
on the specific identity of the substituent, although the more significant effects appear to
be with the rc-acceptor substituents. The most important qualitative conclusion we can
immediately draw from the results shown in Figure 3-1 is that differential polarization of
spin and charge is not as pronounced as was seen in the phenoxy case. Except for the
fluoro substituent, spin and charge density actually track together. This indicates that the
mechanism for the modulation of spin and charge density could be different in these
radicals. In comparison to the phenoxy radical, the overall charge polarizations are the
same, except for vinyl substituent which withdraws charge density from the oxygens.
Additionally, we observe the regime where spin and charge density are both increased,
but in the substituted phenoxy radicals this regime was not seen. Most interesting of all
is the observed shift from spin delocalization in the phenoxy radicals to spin localization
at the oxygens in the 3,6-R-o-l,2-semiquinones by the n-donor substituents, F, and Me
(to a smaller degree). Possible mechanisms for these deviations from the phenoxy
systems will be elaborated upon in the following sections.

Charge Density Polarization in 3,6-R-0-1,2-semiquinones. In general, the 3,6-
SQ results were similar to what was seen in the substituted phenoxy radicals, in that the
electron withdrawing substituents decreased the negative charge density at the oxygen
atom and electron donating substituents increased the charge density at the oxygens. The
mechanism of charge density polarization is slightly more complicated in the case of the

substituted 3,6-R—o-l,2-semiquinones, since the substituents are both ortho and meta to

62

the oxygen atoms. Hammett constants, op', have been defined for electron rich systems

from the dissociation of phenols in water.20 These are the most logical choice to correlate
charge in an electron rich system, but these constants have not been experimentally
determined for meta substituents. In order to account for both the ortho and meta effects

an additive Hammett parameter that encompasses both the ortho and meta substituent
constants, i.e. o = om + do will be used.21 This additive effect explains why the vinyl
substituent is electron withdrawing in the o-semiquinone case and not for the vinyl
substituted phenoxy radical. In the 3,6-SQ case, the vinyl om Hammett parameter is

dominating, where am (vinyl) = 0.06 and 00 (vinyl) = -0.0264, leading to an overall

positive (oviny, = 0.034, electron withdrawing) Hammett parameter.22 The charge density
polarization correlates (r = 0.95) rather well with the sum of the ortho and meta Hammett
substituent constants as shown in Figure 3-2. The same correlations with do and cm gave

r = 0.94 and 0.92, respectively, and are depicted in Figure 3-3 and Figure 3—4.

 

 

 

 

0.04 ~ ‘ \

b l- \ \ .

'52; 0.02 4 NH2 \ 0H

5 ’ \I\ .OMC

o 0 00 - Me‘ a H

DD _ x

E -0.02 - I. '~ \

U . Vinyl I F

D \

"a ‘0.04 " ‘\ I

:3: '- Cl B \ \ CF3

0 -0.06 +- r ‘

M \ \ I

.= ‘ . CN

3 -0.08 - . \

a - \

g -o.10 -

U - I

-o.12 - N02
1 r 1_ 4 l _4 1 A
-0.5 0.0 0.5 1.0 1.5
cm + 00

Figure 3-2. Charge density vs. additive Hammett parameter for 3,6-R-0-1,2-semiquinones.

63

 

 

 

ii

0.04
0.02
0.00
-0.02
-0.04
-0.06
-0.08
-0.10

Change in Relative Charge Density

—0.12

—0.6

 

r1

r1

)-

 

J

L J

L l L L 4 _l

 

 

-o.4

J
—0.2

0.0

0.0

Figure 3-3. Charge density vs. ('50 Hammett parameter for 3,6-R-0-1,2-semiquinones.

0.04
0.02
0.00
-0.02
-0.04
-0.06
-0.08
-0.10

Change in Relative Charge Density

-0.12

 

V T T r T j V— I W

iyrfii

 

 

 

-0.2

0.0

Figure 3-4. Charge density vs. O'm Hammett parameter for 3,6-R-0-1,2-semiquinones.

64

0.8

 

8 ~.- .—_

 

F I

The increased correlation with 00 indicates that n-effects are likely more dominant than
the inductive effects of the substituents, since the cm constants reflect stronger inductive

effects (0;) than resonance/n-effects (0R), i.e. cm = o. + 0.33 0R compared to so = c; +

GR.” Overall the combined substituent constant gives the best correlation, and the

Hammett constants reflect the charge distribution in the 3,6—R-o-1,2-semiquinones.

Spin Density Polarization in 3,6-R—o-1,2-semiquinones. Examination of Figure
3-1 reveals that the n-donors (NHZ, OH, and OMe) and F increase spin density at the
oxygen atom. The Me group also increases the spin at the oxygen center, but to a much
lesser extent. These results are in contrast with the data fiom the phenoxy radical study,
and indicates that the mechanism for the spin delocalization is inherently different in
these anion radicals. The n-donors are again B-spin donors,24 and will produce a charge
separated resonance structure as shown in Figure 3-5 (left) with (it-spin delocalized onto

the substituent.

.0 Ce

\/
N

'69

No
0% \Oe

 

Figure 3-5. Spin resonance interactions of a B-spin donor with the semiquinone radical (lefi) and a-spin
acceptor (right). The red arrows represent the donation or withdrawal of charge density from the negatively
charged semiquinone.

65

A possible explanation for these substituents increasing spin density at oxygen is that the
charge separated resonance structure is destabilizing to the radical system. Unfavorable
electrostatic repulsion will occur between the negative charge density delocalized onto
the semiquinone ring (the SQ ligand is negatively charged) and the negative charge
created resulting from the B-spin donation of the n-donor ligand (Figure 3-5 left). This
effect can be visualized as occurring through a charge-dipole repulsion type interaction25
represented by the red arrow in Figure 35, where the negative charge density in the ring
system is destabilized by the donation of electron density from the n-donor ligands and
ultimately leads to spin localization at the oxygen centers. The magnitude of the
destabilization is related to the DFT calculated strength of the electron donation, and the
strongest donor of charge density (OH in Figure 3-1) exhibits the largest spin localization
at the oxygens. Similar dipole-dipole interactions have also been argued by Bordwell et.
a126 and Clark and Arnold25 to be responsible for the increase in bond dissociation energy
of phenols substituted with electron acceptors, since the ground state energy can be
destabilized due to the substituent. This argument can be equated to the inductive
electron withdrawing destabilizing effects observed for other systems in the literature.
As discussed in Chapter 1, increases in the bond dissociation energy are usually
correlated with spin localization. This electrostatic repulsion is likely to be responsible
for the increase in spin density at the oxygen atom for these ligands, since a trend is
observed between the calculated donor strength and spin localization at oxygen.
There is a lack of an opposing charge-dipole interaction in the n-acceptors (Figure
3-5 right), since these ligands are a-spin acceptors24 and the negative charge density at

the semiquinone oxygens will be stabilized by electron withdrawal. These substituents

66

were all observed to decrease the spin density at the oxygen atoms, analogous to the
phenoxy radicals. In fact, the stabilization of the charge density in the 3,6-SQ by the
nitro group is one factor that leads to the larger spin delocalizing ability of the 7:—
acceptors. The spin delocalization in the 3,6—R-o-1,2-semiquinones is depicted in Figure
3-6 and for both the u-spin acceptor and B-spin donor spin density is delocalized onto the

substituents.

 

J

Figure 3-6. Spin density distribution in 3,6-N02— and 3,6-NH2-0-1,2-semiquinones.

Although both a-spin acceptor and B—spin donor substituents delocalize spin density onto
the substituent, the unfavorable charge-dipole interaction of the n—donors causes
increased spin localization at the oxygens to avoid participation of this resonance
interaction. The DF T results for the movement of spin and charge density support these
arguments, and substituents that participate in charge separated resonance structures (1r-
donors) localize spin, and groups where this is less likely (tr-acceptors) will delocalize
spin density from oxygen. Thus, the spin density is modulated to a large extent by
charge-dipole interactions in these systems.

Fluorine is also in the group of substituents that increase the spin density at the

Oxygen atom. The F -substituent has been shown to be destabilizing in the cm', 60', and

67

oc’ Hammett spin delocalization scales. The expected strong inductive stabilization from
the fluorine substituent should aid in the spin delocalization, but this is not observed and

another mechanism must be responsible. The Hammett polar substituent constants for an
electron rich system, (Sp-{20’ 22 indicate that F is slightly electron donating and this could
be a possible cause for the observed deviation since the calculations indicate that donors

increase the spin density at the oxygen atoms.

The change in spin density was also correlated with the Hammett oc', ou', and the
ADp spin delocalization constants,”29 and the results were generally inconsistent
(Figures 3-7 to 3-9) with r = -0.87, r = -0.63, and r = 088, respectively. The best

correlation (r = -0.94) was observed for an additive type AD parameter29 (AD = ADm +

ADO), where the ADD value was approximated by 0.66-ADp, shown in Figure 3-10.30

 

 

 

 

0.04 \

l' \\
Q“ 0.02 - F ‘ \\
m \
G b \ CF I
5 000 - . x .3 OMe
.E b H C‘l‘\
(2" -0.02 P Bl‘ \\ \
<1) i- . \
...>.. CH \\
+55 -0.04 - 3 \\
7.3 - x .
a: _ __ \ CN
.5 0.06 \\\
Q) \
on ‘0.08 " \ \
g n \\
u: \
U "O.10 " N02

' I
-O.12 L-
L 1 4 L 4 l 4 _L _L L_n__l g I
-0 l 00 0 l 02 0.3 0.4 0.5 06
oc°

Figure 3-7. Spin density correlation with the 0C. constant in 3,6-R-0-l ,2-semiquinones.

68

 

 

in

Figu re 3-8.

Figure 3-9.

'ty

Change in Relative Spin Densr

   

-0.1 0.0 0.1 0.2 0.3 0.4 0.5
on

Spin density correlation with the 0'“. substituent constant in 3,6-R-0-1,2-semiquinones.

Change in Relative Spin Density
.5 .5 .<'= .6 :5
—- O O O O
O 00 O\ A N

-0.12

 

-0.2 I 0.0 0.2 0.4 0.6 0.8 1.0

P

Spin density correlation with the ADp substituent constant in 3,6-R-0-1,2-semiquinones.

69

0.04 ~

 

 

 

 

0.02 4 F' \ NH2'

.5 r OMe I \ \

g 0.00 4 CH m. u H

O " 3 Cl. \- 31'

3’0 -0.02 ~ \

g . I \ \

o —0.04 _ CF3 \

3 L ‘-\

:3. -0.06 r CN \\

é ' ‘

.5 -0.08 , \\

Q) J \

g) .010 4 . \

.c: t I \

U -012 4 N02 \

L l _. L L L . J L 1 4
-04 -02 0.0 0.2 0.4 0.6 0.8

ADm + ADO

Figure 3—10. Spin density correlation with the additive (ADm + ADO) substituent constant in 3,6-R-0- l ,2-
semiquinones.

Given the known non-transferability3 1 of the Hammett type spin delocalization constants
for different systems and the observed inconsistency other correlations were sought. As
was done for the phenoxy radicals, the change in spin density was correlated with the a-
HOMO-a-LUMO gap. Figure 3-11 indicates that as the a-HOMO-a-LUMO gap is
decreased the spin density at the oxygen atom decreases. This appears to be a marker for
identifying the magnitude of spin delocalization in the neutral and anion radicals studied
thus far. The correlation is strong (r = 0.96) with the vinyl substituent is the only outlier
(the correlation coefficient including the 'vinyl group is r = 0.92). We checked the
compositions of the u-HOMO and a-LUMO wavefunctions to uncover any orbital
interactions that would give evidence to support the removal of the vinyl substituent as

was done for the phenoxy systems. The results of this analysis are shown in Table 3-1.

70

 

0.04 /

 

 

 

A ‘ NH I OH

f 0.02 — 2 ' ’ '

Dd . OMe I /

8 0.00 - AB

.3 ' BrI ///Me-Cl

2 0.02 — /

é . , i

c -004 - , CF3

8’6 - . 1' . ’

>4 vrny CN /

5 -0.06 /

‘26 ’ J ’

g» -0.08 — I I

'a , ,

5 1

Q -0.10 - I ’

E r / ’ IN02

mo“ _0_12 L L 1 1 1 . 1 . 1 .
3.0 3.5 4.0 4.5 5.0 5.5

a—HOMO-a-LUMO Gap (eV)

Figure 3-11. Spin density correlation with (x-HOMO-Ot-LUMO gap in 3,6-R—0-1,2-semiquinones.

Table 3-1. Percent composition of 3,6-R-o-l ,2-semiquinone frontier orbitals.

 

 

% 0 %R %0 %R a
Substituent (Jr-HOMO or-HOMO or-LUMO a-LUMO Spin Gap (eV)
H 19.85 0.007 5.28 0.03 0.26345 4.69128
F 21.45 1.68 5.77 0.29 0.28741 4.79033
c1 17.64 2.82 0.35 12.04 0.25621 4.73727
Br 17.72 2.78 0.42 18.25 0.25448 4.22297
Mb 21.63 3.19 1.22 27.60 0.28231 4.64420
0H 23.77 1.804 5.24 0.50 0.28786 5.14381
OMe 18.99 2.929 0.64 37.66 0.27281 4.71251
CH3 19.01 1.91 4.09 14.38 0.26368 4.64067
CF: 16.19 1.46 7.37 10.38 0.23504 4.20501
CN 14.75 3.34 9.46 14.52 0.21184 3.72173
N02 10.51 8.30 9.34 29.32 0.15423 3.02375
Vinyl 16.51 4.66 6.62 28.96 0.21843 3.29750

 

1::
Average spin at oxygen.

71

Recall that a large % R contribution to the a—LUMO molecular orbital was evidence for
the deviation of the outlying substituents in the mechanism put forth for the phenoxy
radical correlation of spin density and a-HOMO-u-LUMO gap. Upon examination of
Table 3-1, we can see that the Cl, Br, NHz, OMe, CF3, CH3, N02, and vinyl substituents
have a moderate to significant contribution to the a-LUMO. Multiple substituents having
a contribution to the a—LUMO in the 3,6-R—o-1,2-semiquinones indicates that the
mechanism controlling the spin polarization is more complicated for these systems when
compared to the phenoxy radicals. Thus, we can not use the orbital composition data for
support of the removal of the vinyl substituent from the a-HOMO-a-LUMO gap
correlation. Another possible explanation for the vinyl group straying from the
correlation is the competing ortho and meta electronic effects in the 3,6—R—o-1,2-
semiquinones. Literature reports have shown that meta-electron withdrawing substituents
in general are radical destabilizing.32 In the case of the 3,6-vinyl-o-1,2-semiquinone, the
spin localizing o-effects can outweigh the ortho (tr-effect), and this could be responsible
for the observed deviation.

Due to the large % R in the a-LUMO orbital in several of the substituents it is
hypothesized that a four-orbital (or possibly more) interaction is the likely mechanism
modulating the HOMO-LUMO gap. Interactions with higher lying orbitals were also
suggested in the phenoxy radical case to explain larger % R values in the a—LUMO
orbital. We were unable to firmly establish this hypothesis using the data in Table 3—1
due to the higher complexity of the multiple orbital interactions, but it is evident that the
correlation between the u—HOMO-u—LUMO gap and the spin density at oxygen is a

marker to quantify the spin delocalization in these anion radicals.

72

Spin and Charge Density Movement of Substituted 4,5-R-o-1,2-
semiquinones. A graphical depiction of our computational results concerning the change
in charge and spin density at the oxygen atom for all 4,5—R—0-l,2-semiquinones is shown
in Figure 3—12. As with Figure 3-1, a positive inflection in Figure 3-5 indicates an
increase in spin/charge density at the oxygen center, and a negative inflection indicates
the opposite trend. Only three of the 4 possible regimes for spin and charge density
variation were observed, including: (1) decrease spin density and increase negative
charge density at the oxygen atom (OH, NH2, and Me); (2) decrease spin density and
decrease negative charge density at oxygen (halogens, tc-acceptors, and OMe); and (3)

increase spin density and decrease negative charge density at the oxygen center (CF3).

 

101-

 

=H)

"T—T—T f—
- F OMe
Cl Br
CF3

N02 vrnyl

% Change (Relative to R

 

 

 

 

 

 

 

 

 

n-donors, tt-donors n-acceptors c-donor G-acceptor

G-acceptors

Figure 3-12. Spin density (red) and charge density (blue) polarization in 4,5-R-0- l ,2-semiquinones.

In general, all spin and charge density changes are smaller when compared to the 3,6-

isomers. Two notable differences in spin and charge polarization between the isomers

73

were observed. First, the CF 3 group is the only substituent that increases the spin density
in the 4,5-R—0-1,2-semiquinones, and the increase spin, increase charge regime is not
observed with these isomers. Second, the OMe substituent decreases charge density,
which is the opposite trend for the other n-donors in this system and from the 3,6-
isomers. Possible mechanisms for the deviations between the 4,5- and 3,6-isomers will
be elaborated upon in the following sections.

Charge Density Polarization in 4,5-R-o-1,2-semiquinones. The changes in
charge density are similar to the 3,6—R-o-1,2-semiquinone isomers except for the
methoxy substituent which decreases charge density at the oxygen atoms. The optimized
structure of the 4,5-OMe-o-l,2-semiquinone reveals that the OMe group is not planar
with the semiquinone ring, and this is a possible cause for the deviation in charge density.
The Hammett parameters were again added together to take into consideration both the

para and meta polar effects of the substituents, as shown in Figure 3-13.

 

 

 

 

0.02 l- ; \ \
\ I
g . NH2 \ \ OH
:3 0.00 4 Me‘ . n
a) I \ H
g) , OMe ‘ , . F
'5 \ ‘
~0.02 1- I \ \
E Vinyl \IICl
Tc: P Br \ \
m '004 r: \ Q CF3
.S \ \
& L ‘ x
g CN' \
4: -0.06 r- '\
O NO \ t
.1 _L L n J n L rm L _. 2|
-1.0 -0.5 0.0 0.5 1.0 1.5
(Sm + (5p

Figure 3-13. Charge density vs. additive Hammett parameter for 4,5-R-0-1,2-semiquinones.

74

 

 

 

 

0.02 i- . ‘ \ \ 0H

£9 NH2 ‘~ .-

8 L \ \ CH3

5 0.00 1— \f

8. ' \ F

a , OMe ‘ t '

6 \ ‘

Q) ‘0.02 r" . . \ \ Cl

S vrnyl ‘5

.53 Br \ x

Oct.) \ ‘ I

.5 -0.04 l- \ \ \CF3

a . ‘ .

C I \

«3 CN ‘ \

4: -O.

U 06 i N02- 1

#1 g4 4 _L J L _1 L , J L i L J
-0.8 -0.6 -O.4 -O.2 0.0 0.2 0.4 0.6 0.8

“9

Figure 3-14. Charge density vs. 6p Hammett parameter for 4,5-R-0-l ,2-semiquinones.

 

 

 

 

0.02 ‘— \ x!
3.3: NH2 ‘ . 'OH
58 ' ‘ .
O \
D 0.00 L CH3 I ‘ \
& H " OMe
:3 r \ ‘ F '
6 ‘x
,, -0.02 _ . ‘ , c1
> Vinyl \ \ I
'43 L \I\ Br
0 ‘ x
m _ P I \
.S 0.04 CF3 \ \
& L \ \
E CN' \ .
O -0.06 b N02. \ ‘
J .I_ l L 1 _L J 1% _L _1
~02 0.0 0.2 0.4 0.6 0.8
6m

Figure 3-15. Charge density vs. O'm Hammett parameter for 4,5-R-0-l ,2-semiquinones.

The change in charge density correlates rather nicely with the additive Hammett

parameter with r = 0.96. We also attempted the same correlation with only the GP and om

75

substituent constants, and the correlation resulted in r = 0.96 (Figure 3-14) and r = 0.90

(Figure 3-15), respectively. As before in the 3,6-SQ isomers the better match with the op

constants indicates that n-effects are more dominate than the depolar effects of the 4,5-
isomers.

Spin Density Polarization in 4,5-R-o-1,2-semiquinones. Major differences are
noted in the spin density distribution in the 4,5-R—o-l,2-semiquinones when compared to
the 3,6-isomers. First, the n-donors, fluorine, and methyl (weakly) decreased the spin
density in the 4,5-isomers, and in the 3,6-isomers the opposite trend was seen. Secondly,
in the 4,5-SQ series the CF 3 substituent is the only substituent to increase the spin density
at the oxygen atom. Third, the magnitude of the changes in spin density are attenuated in
the 4,5-R-o-l,2-semiquinones. To explain these differences and find the mechanism for
the spin delocalization we searched for correlations between the changes in spin density
with the Hammett spin delocalization constants, the C—0 bond distance, and the a-
HOMO-u-LUMO gap. All the correlations with the Hammett parameters, C-O bond
distance, and even the u-HOMO-a-LUMO gap gave poor correlation constants, and the
points were highly scattered. As an example, the poor correlation with the a-HOMO-a—
LUMO gap can be seen in Figure 3-16. The differences in the spin density distribution
and the lack of any correlation with the a-HOMO-u—LUMO gap are most likely due to
steric factors between the substituents. The competing meta and para electronic factors
could also be a cause of the deviations, but the correlation of the 3,6-isomers with the a-
HOMO-a-LUMO gap was decent even when all substituents were used. There are less
steric repulsive interactions in the 3,6—SQ isomers since the substituents are not ortho to

one another as seen in the comparative N02 example in Figure 3-17. Even though both

76

isomers deviate from planarity with the SQ ring, the 4,5-isomer canting is much more

pronounced due to the proximity of both nitro groups.

 

H)

0.010 —

0.005 1
0.000 .- Me IH
—0.005 4 Br . C,
-0.010 .- ' F
0015 .-

-0.020 b

 

 

I
I I
0.025 9 vinyl NH2 0H
. J l l

2.0 2.5 3.0 3.5 4.0 4.5 5.0
a-HOMO-a-LUMO gap (eV)

L .1 a l

 

Change in Charge Density (Relative to R

Figure 3—16. Correlation of spin density with the a-HOMO-a—LUMO gap in 4,5-R-0-l,2-semiquinones.

 

Figure 3-17. DFT optimized structures depicting the steric interactions in 3,6—N02-o~1,2-semiquinone
(left) and 4,5—N02-0-l ,2-semiquinone (right).

The results of the 3,6-R-o-l,2-semiquinones then suggest that geometric fluctuations in
the 4,5-R-o-l,2-semiquinones are likely the major factor causing the deviations in the

Spin and charge distribution. Recall that the geometry optimizations indicated that the

77

N02, vinyl, OMe, and NH2 substituents are not planar with the semiquinone ring. Due to
the non-planarity and steric crowding between the substituents the spin distribution
mechanism will be difficult to interpret, and therefore the spin delocalization mechanism

can not be determined in this case.

3.3.2 Substituted 9,10-Phenanthrenesemiquinone Anion Radicals

Geometry Optimizations of Substituted 9,10-Phenanthrenesemiquinones.
The crystal structure for the 9,10vphenanthrenequinone ligand has been reported,33 and
Praseuth34 has shown that calculations at the UB3LYP/6-31g* level yielded structures
that were nearly identical to the experimental X-ray data. Analogous to the case of the o-
l,2~semiquinones, the DFT optimized structures are then good estimates of the true
geometries. The optimized geometries of the 2,7-Phenanthrenesemiquinones (2,7-PSQ)
and 3,6-phenanthrenesemiquinones (3,6-PSQ) had no large structural variations, and the
largest C—O bond distance variations were found to be :1: 0.019 A. The only substituents
that were found to deviate from planarity were the 2,7—NMe2 and 3,6-NMe2 substituted
radicals. All optimized structures corresponded to global minima verified by frequency
calculations.

The phenanthrenesemiquinone ligands are polycyclic ring systems, and the
numbering scheme of the carbon atoms is indicated in Scheme 3-3. Virtually all
positions of the phenanthrenesemiquinones can be substituted, but for our purposes we
wanted to have symmetric substitution within the ring system to facilitate a more
straightforward analysis of the charge and spin distributions. To this end the 2,7—(meta)

and 3,6-(para) substituted radicals were studied. Substitution at these positions will not

78

be susceptible to steric crowding as was the case for the substituted o—l ,2-semiquinones.
We shall look at each set of phenanthrenesemiquinone isomers individually, and look for
similar trends in spin and charge density polarization as we observed for the o—l,2-

semiquinone and phenoxy radical systems.

3/2\1 3/2\1 3/2\1
y u e n e
/ 10/0 4 / /o 4 / /o
I ———-— ‘i .__:... ii
9 -e- 9 T 9
i \ P" i \ \°°" in \ \°e
6\,/B 6\7/8 6\,//a

Scheme 3-3. Redox properties of the phenanthrenesemiquinones.

Spin and charge density polarization in 2,7-phenanthrenesemiquinones. At
the oxygens, the unsubstituted PSQ has spin = 0.25703 (rt-electrons and charge = -0.6329
electrons. Of the four possible regimes for the increase/decrease of spin and charge
density only the increase charge, decrease Spin was not observed (Figure 3-18). The spin
and charge polarization in the 2,7—phenanthrenesemiquinones is nearly identical to the
3,6-R-o-1,2-semiquinones, except for the OH and CH3 substituents. In these radicals the
OH substituent actually decreases the charge at the oxygen atom, which contrasts with
the results from the 3,6-R-o-l,2-semiquinones. Additionally, the methyl group decreases
spin density with essentially no change in charge density, while the opposite trend was
observed for the 3,6-R-o—1,2-semiquinones. The underlying trends and possible

mechanisms for spin and charge delocalization will be discussed below.

79

U'

 

 

 

 

 

 

 

 

 

 

 

 

10
L

A o k Ja—L—L ..
a: H T T NMe2 0H! CH3]
n F Cl Br NHZ OMe
M CF3

_ L.
3 10 vinyl
8; L
*3
—-t -20 L
a)
86.

L.

8’0
:5 -30 +-
.2
O )-
°\°

-40 *- N02

fl-donor, 7t-donors n-acceptors G-donor G-acceptor

O-acceptors

Figure 3-18. Spin (red) and charge (blue) polarization for 2,7-R-phenanthrenesemiquinones.

Charge Density Polarization in 2,7-phenanthrenesemiquinones. The charge

density was correlated with the Hammett parameters (om, op, and GP") and the best
resulting correlation was with the op'. The correlation constants for the varying Hammett
parameters are r(crm) = 0.88 (Figure 3-19), r(op) = 0.86 (Figure 3-20), and r(op') = 0.94
(Figure 3-21). The op" Hammett constant is usually invoked in an electron rich system,

and it is not surprising then that the op' parameter has the best correlation with the change
in charge density, since the semiquinone ring is an anion.” The correlation constant
should increase if cm- type-constants were available (2,7 is formally a meta isomer), but

no comprehensive list of these constants is reported in the literature. The overall trend in

80

 

 

F:

this system is comparable to what we have observed in the other semiquinone anion

radicals.

Figure 3-19. Charge density vs. Hammett 6m correlation for 2,7 R f

Figure 3-20. Charge density vs. Hammett Op correlation for 2,7 R r‘

Change in Relative Charge Density

Change in Relative Charge Density

0.02
0.01
0.00

-0.0l

-0.02

-0.03

-0.04

-0.05

-0.06

~0.07

0.02

 

\\ NH

s.
\

r.-

 

L I -

\

H

OMe

\.\
3Ii‘\{

I
vinyl

OH F
\ I

x
\

\SH'IBr
\\ I
\\§F3

 

 

—0.2

0.0

0.2

cym

 

0.01

0.00 —

-0.0l

-0.02

-0.03

-0.04

-0.05

-0.06

-0.07

.
.. I-
\

I
N M62

L

.—

)—

 

1 I 4
-l.0 -0.8

\‘etNHz

OH

x

x
x
\
I \

OMe

L4.

I I
.Hs

\

.'F

I \
vinyl

Cl
\ \{Br I
\ \ \ CF3
‘qu

\

NO
I2

L LmL#4l 1 l l

\

 

 

-0.6

-0.4

81

-0.2 0.0 0.2 0.4

6p

 

 

 

 

0.02
" , NH
f 0.01 \ 2-
o: - 1 e2
3 0.00 L OMe If H
I \

.QZ) - OH CH3 I\‘\
E ‘0.01 "' F \
E - I\.\\
v -0.02 - CIBr \ .
.g ' \\ CF3
5 -o.03 — _
Q - ‘\ CN
‘9 -0.04 - \ \
OD \
D '0.05 - \\
.5 ~ ‘\
(D - >-
g) 0.06 - 'NOZ
.5410.|.ihLJm.i.i.i.|.i.i

-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Up.

 

Figure 3-21. Charge density vs. Hammett O'p- correlation for 2,7 R ,‘

 

Spin Density Distribution in 2,7 _ L " ° l ' The calculations
have shown that the strong n-donors and fluorine increase the spin density at the oxygen
atoms, while all other substituents decrease the spin at the oxygen centers. It should be

noted that the increase in spin at the oxygens is much weaker in comparison to the 3,6-

 

SQ radicals. The substituents in the 2,7 [‘ “ ' . ' um and the 3,6-R-0-l,2-
semiquinones have similar spin polarization regimes, and it is likely then that the 2,7-
. L “ ' . ' mm will have comparable polarization mechanisms. Analogous
resonance interactions can occur, and the rt-donors will participate in charge separated

resonance structures (Figure 3-22 left). A much weaker charge-dipole type-repulsion is

 

exhibited in the calculations of the 2,7 . L “ ', ' mr- systems, where the

82

negative charge density in the ring system is destabilized by the donation of electron

density from the n-donor ligands.

 

Figure 3-22. Spin resonance structures of 2,7-R-phenanthrenesemiquinone substituted with a n—donor
(left) and n-acceptor (right). The red arrows represent charge donation or withdrawal from the PSQ anion.
The magnitude of the spin localization was found to be related to the calculated donating
strength of the substituents, and in the 2,7-PSQ radicals the calculated n-donation is
weaker and the corresponding spin localization is decreased.

Again analogous to the 3,6-R—0-l,2-semiquinones, there are no charge
destabilizing interactions in the a-spin acceptors (Figure 3-22 right) and these
substituents were found to increase spin delocalization in the radicals. The spin
polarization mechanism can also be qualitatively verified by examining the spin density
distributions in 2,7-NH2- and 2,7-NO, ,' “ ' , ' mes Figure 3-23 indicates
that the spin density is easily delocalized onto the N02 substituent, and there is a lack of
spin population on the NH; substituents. The spin delocalization onto the nitro
substituents is consistent with no destabilizing charge-dipole type interactions being
present. Very little spin density is delocalized onto the NH2 substituent, since formation
of the charge separated resonance structure is minimized by weak B-spin donation. The

Weaker B-spin donation accounts for the attenuated spin localization in the 2,7-PSQ

83

compared to the 3,6-SQ radicals. Therefore, the spin distribution in the 2,7-

] Al

I 1 supports a weaker charge repulsron mechanism.

 

Figure 3-23. Spin density delocalization in 2,7-NH2- and 2,7-N02-phenanthrenesemiquinone.

The spin density was again correlated with the a-HOMO-u-LUMO gap shown in
Figure 3-24. The N02 substituent is an obvious outlier in the correlation with r = 0.957.
Just as in the case of the 3,6-R—0-1,2-semiquinones, the percent contribution of the

substituents to the frontier orbitals was examined, and is tabulated in Table 3-2.

 

_o
8

OH
9.“ F!—:’ , - 'NHZ
Br , , I”
C123, - - - ’ Me NMez OMe
-0.02 — r ’ I
-0.04 -
-0.06 -
.008 -

-0.10 - Noz'

 

 

_0.12 I III, I I I I I
1.4 2.0 2.2 2.4 2.6 2.8 3.0

a-HOMO-a—LUMO Gap (eV)

 

Change in Spin Density (Relative to R

Figure 3-24. Change in spin density vs. a-HOMO-a-LUMO gap for 2,7-phenanthrenesemiquinones.

84

Table 3—2. Percent contribution to the frontier orbitals of 2,7-phenanthrenesemiquinones.

 

 

% R % o % R % 0 Spin a b
Substituent (X-HOMO a—HOMO a-LUMO ct-LUMO Density Gap
H 0.00 18.23 0.00 0.40 0.2570 2.515
F 0.20 18.40 1.34 0.06 0.2616 2.675
c1 0.36 17.82 1.82 0.32 0.2569 2.480
Br 0.35 17.77 1.65 0.36 0.2560 2.462
OH 0.34 18.58 1.89 0.00 0.2626 2.715
OMe 0.68 18.43 2.83 0.05 0.2603 2.861
NH; 0.70 18.29 4.65 0.00 0.2632 2.854
NMez 1.27 17.04 13.51 0.16 0.2579 2.715
CH3 0.94 17.87 3.24 0.39 0.2555 2.595
CF3 0.57 16.70 3.70 1.31 0.2440 2.254
CN 1.25 15.60 6.88 2.54 0.2286 1.930
N02 8.49 10.52 20.60 6.81 0.1595 1.430
vinyl 2.46 16.13 17.91 2.54 0.2328 1.968

 

a 5
The average spin density at oxygen. The a-HOMO-a—LUMO gap is measured in electron volts.

The N02 substituent does have a 20.6 % contribution to the a-LUMO, but the vinyl and
NMez substituents both also have a large contribution to the ot-LUMO orbital. The nitro
substituent can’t be excluded based on the %R data in the a—LUMO, since other
substituents fall nicely on the line with elevated %R contributions. A dominating
electron withdrawing o-effect argument similar to the exclusion of the vinyl substituent
in the 3,6-R-o-l,2-semiquinones can be suggested for the N02 substituent here.
Literature reports have indicated that the electron withdrawing o—effects of N02 at the

243235 Further evidence for the dominating

meta-position is destabilizing in nature.
electron withdrawing o-effects causing the N02 deviation will be revealed in the

examination of the 3,6-isomers (vide infra). The correlation of the spin density with the

85

a—HOMO-a-LUMO gap and the %R data in Table 3-2 supports invoking a similar four-

orbital mechanism that was used for the 3,6-R-0-l,2-semiquinones.

Spin and Charge Density Polarization in 3,6-phenanthrenesemiquinones.
Only two regimes for spin and charge density variation are observed for the 3,6-
phenanthrenesemiquinones compared to the 2,7-isomers, and these changes are shown in
Figure 3-25. All substituents decrease spin density at the oxygen atoms. The substituents
that decrease spin density and decrease negative charge density at oxygen are N02, CN,
vinyl, CF3, F, Cl, and Br, and the substituents that decrease spin density and increase

negative charge density at the oxygen center are NHz, NMez, OMe, OH, and CH3.

 

T

CF 3

=H)

 

% Change (Relative to R

 

 

 

1&32

n-acceptors U‘dODOI’ o-acceptor

-20 _

 

 

 

 

 

 

n-donor, n-donors
o-acceptors

Figure 3-25. Spin density (red) and charge density (blue) polarization in 3,6-R-

phenanthrenesemiquinones.

86

The changes in charge density are similar in the two sets of , L ° , '

isomers, but there is a difference in the spin polarization of the n—donors. The changes in
spin density for the n-donors are small, and difficult to quantify. A reasonable
assumption for the shift in spin delocalizing ability of the n-donors is that the charge-
dipole repulsion interaction is decreased or inoperable compared to the 2,7-PSQ. In other
words, there is a larger donor destabilizing effect in the 2,7-PSQ ligands. The n-donors
decrease spin density at the oxygen atoms and a-spin density is delocalized onto the
substituent, thus the theoretical spin density data for the 3,6-PSQ complexes supports the
lack of charge-dipole destabilizations. The opposite was true for the 2,7-PSQ isomers,
and spin density was increased at the oxygen centers. The spin density distribution for
the 3,6-N02 and 3,6—NH2-PSQ’s can be seen in Figure 3-26, and the spin distribution

supports the proposed mechanism for decreased charge dipole repulsion.

 

Figure 3-26. Spin density delocalization in 3.6-NH2-PSQ and 3,6-N02-PSQ.

 

Charge Density Polarization in 3,5 _ L " '1 ‘ The charge

density distribution for the 3,6—pbenanfln ', ' mes was correlated with the

Hammett op. substituent constants, and the correlation is shown in Figure 3-27. The

87

correlation coefficient, r = 0.955, is strong and the substituents follow typical
electronegativity arguments as was seen in the other semiquinones studied above.

Therefore, in all anion radicals examined here the Hammett correlations hold.

 

 

 

 

E t

u 0.02 \ INHZ

Dd ‘ \

8 0.01 F" OH “ \

d; r I .\ \ CH3

'5: 0.00 r- OMe -' 3. H

N

’5 g NMCZ F \ \

6 -0.01 4 9.,

b . Cl ‘ x

'5‘) Br \ \ CF

5 ~0.02 1- \ \- 3

Q l- \ \

0) \

on -0.03 r- \ I

H ‘ \ CN

2 - .\

O -0.04 - .

.E _ ‘ x

Q) I \

go 005 - N02

3 . 1 . 1 J J _1 1 4 1 , 1 4 1 4 1 4 J I 1

U -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Up-

Figure 3—27. Charge density polarization in 3,6-phenanthanrenesemiquinones.

Spin Density Polarization in 3,6-R-phenanthrenesemiquinones. The spin
density was again found to correlate with the a-HOMO-a-LUMO gap, as shown in
Figure 3-28. No substituents were removed from the regression line to achieve a
correlation constant of r = 0.948, but the correlation does rely heavily on the nitro-
substituent to achieve a correlation coefficient close to one. Interestingly, upon
examination of the percent atomic contributions to the a—LUMO orbitals of the N02, CN,
and vinyl substituents, they were found to have large % R character. This data supports
our hypothesis that a complicated four—orbital (or higher) interaction is the mechanism

that could modulate the spin density in the semiquinone radicals. The lack of any firm

88

correlations in the orbital data preclude us from quantifying this interaction. Although
the a-HOMO-a-LUMO gap correlation with spin density is present, due to the heavy
dependence on N02 the reliability of the correlation remains unclear. It is noteworthy
that all radicals examined did exhibit the correlation with the a-HOMO-a-LUMO gap,

and the HOMO-LUMO gap suggests the degree of spin delocalizing ability of the

 

 

 

 

substituents.
E 0.00 , F ,i/H
u OMe: 3“ C".
‘3‘ OH ' E s CH3 'CF
9 -0 01 — u , NM 3
a) ' NH2 , ’ 62
'E L // I
t3
_ . ’ CN
‘9 ~002- vmll,’
E‘, y, ,
r: ,
Q) /
fl -0 03 - , ’
X /
O L. I ’
*5 z
b '0.04 - / I
a t' / /
D /
e -005 9’ .N02
08" I . i 1 I I l l l 1 I 1 l
m

1.4 1.6 1.8 2.0 2.2 2.4 2.6
oc-HOMO-oc-LUMO Gap (eV)

Figure 3—28. Change in spin density vs. a-HOMO-a—LUMO gap for 3,6-phenanthrenesemiquinones.

3.4 Concluding Comments.

The investigations of the spin and charge density polarization for o-1,2-

semiquinones and o-9,10-phenanthrenesemiquinones have revealed that the negative

89

charge density is well represented by the Hammett substituent constants, resulting in
strong correlations with the different substituent constants (op, om, op', etc. ). In the 3,6-

SQ and 2,7-PSQ radicals the interaction between the electron donor and acceptor
substituents with the delocalized negative charge density in the SQ ring impacted the spin
delocalizing ability of the substituents. The dipole resulting from the charge separated
resonance structure of a n—donor interacts unfavorably with the negative charge in the
ring resulting in spin density localization at the oxygen centers. This localization is a
direct result of the energetically unfavorable charge-dipole repulsion interaction in the
semiquinone ring. This accounts for the change is spin delocalizing ability observed
between the n-donors in the phenoxy radical and 3,6-R-SQ and 2,7-PSQ radicals, since
the phenoxy radical is a neutral radical the destabilizing charge repulsion interaction is
not likely. The electron acceptors stabilize the negative charge and were all found to
increase spin delocalization in the o-semiquinones.

The spin density at the oxygen atoms in the o-semiquinones can be correlated
with the a-HOMO-a-LUMO gap (except for the 4,5-o—l,2-semiquinones), and this
correlation seems to be a general result for both neutral and anion radicals. In contrast to
the phenoxy radicals, a firm mechanism to describe the spin delocalization using
molecular orbital theory was not determined. It is hypothesized that a complicated four-
orbital (or higher) interaction could be modulating the spin density in these systems, but
this could not be confirmed based on the DF T data. Choices of the prOper substituent for
the synthesis of transition metal semiquinone complexes to modulate exchange
interactions based on these DFT results will then be of a more empirical basis. More

work is needed to quantify the spin polarization mechanisms in order to intelligently

90

design exchange coupled complexes. Experimental work along these lines using electron
paramagnetic resonance spectroscopy to determine the substituents effect on the spin
delocalization/polarization mechanisms of substituted PSQ radicals and gallium(III)—PSQ

radical complexes will be examined, and is the topic of Chapter 4.

91

 

3.5 References

1.

2.

10.

11.

12.

13.

14.

15.

16.

17.

18.

Fehir Jr., R. J .; McCusker, J. K. J. Am. Chem. Soc. 2009, submitted.

Sugiura, K.-I.; Arif, A. M.; Rittenberg, D. K.; Schweizer, J.; Ohrstrom, L.;
Epstein, A. J.; Miller, J. S. Chem. Eur. J. 1997, 3, 138.

Osanai, K.; Okazawa, A.; Nogami, T.; Ishida, T. J. Am. Chem. Soc. 2006, 128,
14008.

McConnell, H., M, J. Chem. Phys. 1963, 39, 1910.

Hicks, R. G.; Lemaire, M. T.; Thompson, L. K.; Barclay, T. M. J. Am. Chem. Soc.
2000, 122, 8077.

Shultz, D. A.; Sloop, J. C.; Washington, G. J. Org. Chem. 2006, 71, 9104.
Chaudhuri, P.; Wieghardt, K. Prog. Inorg. Chem. 2001, 50, 151.

Miiller, J .; Kikuchi, A.; Bill, E.; Weyhermiiller, T.; Hildebrandt, P.; Ould-Moussa,
L.; Wieghardt, K. Inorg. Chim. Acta. 2000, 297, 265.

Rotthaus, 0.; Thomas, F.; Jarjayes, 0.; Philouze, C.; Saint-Adam, E.; Pierre, J .-L.
Chem. Eur. J. 2006, 12, 6953.

Mishustin, A. 1. Russ. J Inorg. Chem. 2008, 53, 1376.

Wheeler, D. E.; Rodriguez, J. H.; McCusker, J. K. J. Phys. Chem. A 1999, 103,
4101.

Pierpont, C. G.; Buchanan, R. M. Coord. Chem. Rev. 1981, 38, 45.
Pierpont, C. G.; Lange, C. W. Prog. Inorg. Chem. 1994, 41, 331.
Pierpont, C. G. Coord. Chem. Rev. 2001, 216-217, 99.

Kaim, W. Coord. Chem. Rev. 1987, 76, 187.

Zanello, P.; Corsini, M. Coord. Chem. Rev. 2006, 250, 2000, and references
therein.

Poddel'ski, A.; Cherkasov, V. K.; Abakumov, G. A. Coord. Chem. Rev. 2009,
253, 291.

Pierpont, C. G.; Attia, A. S. Collect. Czech. Chem. Commun. 2001, 66, 33.

92

 

(.4)

tJ-l

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

Macdonald, A. L.; Trotter, J. J. Chem. Soc. Perkin Trans. 2 1973, 473.

Correlation Analysis in Chemistry: Recent Advances. Plenum Press: New York
and London, 1978.

Exner, 0. Correlation Analysis of Chemical Data. Plenum Publishing
Corporation: New York, NY, 1988.

Hansch, C.; Leo, A.; Taft, R. W. Chem. Rev. 1991, 91, 165.

Shorter, J. Correlation analysis in organic chemistry: an introduction to linear
flee-energy relationships. Oxford University Press: Oxford, 1973.

Wayner, D. D. M.; Arnold, D. R. Can. J. Chem. 1985, 63, 2378.
Clark, K. B.; Wayner, D. D. M. J. Am. Chem. Soc. 1991, 113, 9363.

Bordwell, F. G.; Zhang, X.-M.; Satish, A. V.; Chen, J .-P. J Am. Chem. Soc. 1994,
116, 6605.

Creary, X. J. Org. Chem. 1980, 45, 280.
Jiang, X.-K. Acc. Chem. Res. 1997, 30, 283.

Adam, W.; Harrer, H. M.; Kita, F.; Korth, H.; Nau, W. M. J. Org. Chem. 1997,
62, 1419.

Korth, H.-G.; de Heer, M. 1.; Mulder, P. J. Phys. Chem. A 2002, 106, 8779.
Brinck, T.; Haeberlein, M.; Jonsson, M. J. Am. Chem. Soc. 1997, 119, 4239.
Fisher, T. H.; Dershem, S. M.; Prewitt, M. L. J. Org. Chem. 1990, 55, 1040.

Matsuzaki, S. Y.; Gotoh, M.; Kuboyama, A. Mol. Cryst. Liq. Cryst. 1987, 142,
127.

Praseuth, R. T. Michigan State University, 2002.

Creary, X.; Mehrsheikh-Mohammadi, M. E.; McDonald, S. J. Org. Chem. 1987,
52, 3254.

93

Chapter 4. Density Functional Theory and Electron Paramagnetic Resonance
Spectroscopy of Dinitophenanthrenesemiquinone Radicals and Their Gallium(III)

Complexes

4.1 Introduction

After the prescription for the movement of spin and charge density was realized
for the phenoxy radicals and substituted semiquinone complexes, a natural extension of
that work is to see if the same rules that govern spin density distribution can also apply in
transition metal radical complexes. It is well known in the literature that transition metal
complexes have a combination of spin polarization and spin delocalization mechanisms
that describe the total spin density distribution within a molecule."2 The examples cited
by Alvarez and coworkers” mostly consist of paramagnetic transition metals ligated with

diamagnetic donor ligands. The general picture of spin density distribution in such a

‘ +i__1

”"”\ /i

Figure 4-1. General spin density distribution in a metal-ligand complex.

system can be seen in Figure 4-1 below.

In Figure 4-1, the nitrogen donor atom carries positive spin density (upward arrow),
because spin delocalization is the dominant mechanism since it is directly bonded to the
paramagnetic metal, and the rest of the atomic n-orbitals alternate spin density due to spin
polarization being the dominant mechanism. Remenyi and Kaupp3 have shown that spin

polarization is important in the unrestricted Kohn-Sham description of the electronic

94

structure of open-shell systems. There are also several examples in the literature of
paramagnetic transition metals with organic radicals such as semiquinonesf’5
nitroxidesf’7 and verdazyls8 where spin density delocalization and polarization can play
an important role in creating new molecular magnetic materials. The impact that bonding
of the semiquinone radicals with different diamagnetic metals has on the spin density
distribution has also been examined by Cherkasov and coworkers.9 They determined that
the magnitudes of the hyperfine coupling constants vary when semiquinones are bound to
a metal, but there is still a significant amount of spin density that is delocalized in the
semiquinone ring and usually < 1 % of the spin is delocalized onto the metal.

We will examine the changes in spin density distribution in the semiquinone
radical upon complexation to a gallium diamagnetic metal center. A report from Oshio et
a1.10 compares the spin density distribution of a nitroxide radical to transition metal
complexes of the same radical, with the authors noting that upon coordination the spin
density within the radical moiety changes only slightly. It should be noted that the metal
complexes in Oshio’s work were not directly coordinated to the atoms with the highest

“"2 and theoretically[3 known to

spin density, whereas semiquinones are experimentally
have the highest a-spin population at the oxygen atoms which coordinate directly to the
metal center. Rockenbauer et al.'4 have examined tetrahedral copper(I)
phenanthrenesemiquinone complexes, and have shown that spin density is transferred to
carbons closer to the semiquinone oxygens upon coordination based on changes in the 1H
hyperfine coupling constants in the semiquinone ring. Mixing between the radical

SOMO and a filled 3d-orbital on the copper(I) center was put forth to account for the spin

transfer.14 The Rockenbauer et al. report also gives experimental evidence to support that

95

changes in spin distribution occur when a semiquinone radical is coordinated to a metal
center. To further examine the coordination effects of diamagnetic metals on the spin
density distribution, gallium(III)-phenanthrenesemiquinone complexes will be examined.
Both EPR experiments and DF T methods were employed to probe any changes in spin
density distribution. Thus, a theoretical and experimental picture of the coordination
effects will help uncover any mechanisms that describe the effects of complexation on
the spin density population.

DF T calculations on the 2,7-dinitrophenanthrenesemiquinone (2,7-di-NOz-PSQ)
(1) and 3,6-dinitrophenanthrenesemiquinone (3,6-di—NOz-PSQ) (2) radicals and their
corresponding gallium(III) complexes, [Ga(tren)(2,7—dinitrophenanthrenesemiquinone)]2+
(Ga-2,7-di-NOz-PSQ) (3) and [Ga(tren)(3,6-dinitrophenanthrenesemiquinone)]2+ (Ga-3,6-
di-NOz-PSQ) (4), where tren = tris-2-aminoethylamine, were undertaken to see if the spin
density distribution in the radicals changed upon coordination to a metal center (Chart 1).
Specifically, we wish to uncover any changes coordination has on the a-spin density
delocalization onto the nitrogen and hydrogen atoms. This will manifest itself by the
magnitudes of the 1H and 14N hyperfine coupling constants (HFC). In order to prove
experimentally what the DF T calculations predict, EPR was also carried out on the free
ligands and gallium complexes. Gallium(III) is ideal for this EPR experiment, since it is
a d10 metal with no Heisenberg magnetic exchange interactions to complicate the EPR
spectrum. The EPR experiments will determine the experimental 1H and 14N hyperfine
coupling constants in each radical and gallium complex. The importance of the accurate
determination of the 14N HF C’s stems from our goal to properly examine the effects of

spin delocalization and polarization on the nitro-substituent itself, and compare the results

96

when these same radical ligands are bound to gallium. These values can then be
compared with those calculated by density functional theory methods to interpret the
overall mechanism of spin ‘ ' "mtinn/pularization. If this mechanism can be found,
then the rational design of transition metal organic-radical molecular magnetic materials

can be accomplished.

   

 

 

2,7-di-N02-PSQ (1): R1 = N02 and R2 = H Ga-2,7-di-N02-PSQ (3); Rl = no2 and R2 = H

3,6-di-N02-PSQ (2): R1 = H and R2 = N02 Ga-3,6-di-N02-PSQ (4): R1 = H and R2 = N02
Chart 4‘]. Free di-NOz-PSQ radicals (left) and gallium(lll)-di-NOz-PSQ complexes.

The spin density at the carbon atoms around the semiquinone ring can be
approximated experimentally from the 1H hyperfine coupling constants by using the well
known McConnell relationship,15 seen here in Equation 1.

A(‘H) = QCHHpc (1)
In Eq 1, A(lH) is the experimentally determined proton hyperfine coupling constant (in

Gauss), QCHH is a constant which describes the G-TC interaction between carbon and

97

hydrogen (for the hydrogen nucleus), and pc is the n—spin density at the carbon atom that

the hydrogen is directly bonded to.15

The experimentally determined spin density can
then be compared to what has been calculated by DF T. The isotropic HFC associated

with a given nucleus N can be calculated using Equation 2.
4 -1
Aisom) = 3’“ gegnBeBn<Sz> 9(N) (2)

where g6 and gn are the electronic and nuclear g-factors, Be and [3,, are the electronic and
nuclear Bohr magneton constants, and p(N) is the Fermi contact integral which

corresponds to the spin density at N. By substituting the appropriate constants into

Equation 2, Nguyen et al. '6 derived the following conversion factors:

Aiso (’H) = 159496011) (3a)

Arso<”N)= 576136418) (3b)

and from which one can extract the total s-orbital spin densities for 1H and '4N given the

experimental number for AiSOCN) (in Gauss).

4.1.1 Electron Paramagnetic Resonance Spectroscopy

Electron paramagnetic resonance spectroscopy (EPR), also referred in the
literature as electron spin resonance spectroscopy (ESR), is a technique that probes the
environment of a paramagnetic species by characterizing the interaction of that species
with an applied magnetic field. This method can provide detailed information about the
spin density distribution within a paramagnetic species if there is spin density at a nucleus
with I at 0, where I is the nuclear spin. The basis for EPR is the existence of the two

different spin states of the electron, which become distinguishable in an applied magnetic

98

field. The interaction of the a and B spin states of the electron with an applied magnetic

field is depicted in Figure 4-2, and this is called the electronic Zeeman interaction.

.x—rt— 0‘9

<" AB : gBB

Energy
:2
(P
u
(D

 

 

\JL Be

i Electronic
Zeeman
Interaction

3"

 

Figure 4-2. Interaction of the a and [3 spin states of the electron with an applied magnetic field.

The Zeeman Hamiltonian (H) for the interaction of the spin of the electron and the
magnetic field is shown in Equation 4, where Sz is the spin operator, B is the external
magnetic field applied along the z-axis, g is the g-factor, and B is the electronic Bohr
magneton.

H = 8131332 (4)
When this Hamiltonian operates on the electron spin functions, a and [3, the eigenvalues
produced are m,.'7 The energy of the electron spin states then become

E = msgBB (5)
The electron has ms = 1/2 (the one spin state) and 1115 = -'/2 (the Be spin state), and when we

solve Equation 5 for the energy the ms = -'/2 state is lower in energy (due to the magnetic

moment of the electron). An EPR transition occurs when microwave radiation that

99

equals the separation between the two spin states is absorbed, i.e. when the microwave
radiation is at resonance with AB = gBB.

The fine structure usually seen in EPR experiments arises from electron-nuclear
hyperfine coupling, which is the interaction between the magnetic moment of the electron
with that of magnetic nuclei in the molecule. The magnitude of this interaction is
represented by the hyperfine coupling constant (HFC), A, and can easily be visualized by
examining the hydrogen atom. Both g and A are usually represented as 3 x 3 tensors; but
in the hydrogen atom the spherical symmetry eliminates all anisotropic interactions and
we can assume colinearity of these tensors with projections along x, y, and z. B is again
applied along the z-axis. The corresponding effective spin Hamiltonian for hydrogen
must now take into consideration the interaction of the nuclear spin magnetic moment

with the magnetic field and the hyperfine interaction, and this is shown in Equation 6.
H = gBBSz - ganBIz + AS] (6)

The eigenvalues resulting from solving the Schrodinger equation with this Hamiltonian

are given by Equation 7.'7
E(mSml) = msgBB ' mlganB + Amsml (7)

The first term in Equation 6 is the electronic Zeeman interaction as already discussed
above. The second term is the interaction between the nuclear spin‘magnetic moment (I)
and the applied field, and it is called the nuclear Zeeman interaction. The last term is the
electron-nuclear hyperfine interaction, representing the coupling of the nuclear and
electron magnetic moments. The sum of these interactions can be visualized in the EPR
energy level diagram of the hydrogen atom. The electron-nuclear spin states of the

hydrogen atom are depicted in Figure 4-3. The electronic Zeeman interaction gives rise

100

to the splitting of the B and 01 spin states of the electron. Hydrogen has a nuclear spin of I
= 1/2, and the nuclear Zeeman interaction splits the m: 1/2 (01p) and mi: -‘/2 (Bp) proton
spin states. Due to the charge of the proton the magnetic moment vector is directed

opposite to that of the electron, resulting in the m1 = V; (01p) state being lower in energy.

 

 

oreBp
+ I" \‘ ’ i (leap
" \“ ’I
0‘e ,I’ x 1’
’l-l——“ I“
”I s“ I” \‘
" ‘\ C1. Cl. ,’ U aCBp
I’ ‘\ e p I
l”
613
LII “C’Be ’I’
8 I
LIJ ‘~,
‘\ Befip “a“ BeBp
\\ ' '¢
\\\‘ "’
\“ I”
s B e :
. ‘~“ Beap
Electric ‘h———.,
Zeeman “NJ.—
Nuclear 15er
Zeeman Hyperfine
Coupling
B j)-

 

Figure 4-3. Electron-nuclear spin states of the hydrogen atom.

According to Equation 4, the energetic consequence of proton-electron hyperfine
coupling is Amsml, In the lower spin manifold the 13er (ms = -'/2 and m; = 1/2) state
will be lower in energy, and in the upper spin manifold, the aeBp (mg = V2 and m; = ~1/2)

state will be stabilized. If we assume a large A then the states in the upper spin manifold
can cross due to the increased stabilization as depicted in Figure 4-3. There are two EPR

transitions observed for the hydrogen atom, shown with arrows in Figure 4-3. These

101

transitions are “spin flips”, and the absorptions occur when the microwave radiation

equals the energy difference between the Amg = :1: V2 states. The selection rules for EPR

transitions are Ams == :1: 1 and Am; = 0 for a typical microwave experiment.

4.1.2 Electron Spin Echo Envelope Modulation Spectroscopy (ESEEM)

The ESEEM experiment monitors NMR transitions indirectly, via mixing of the
frequencies of the semi-forbidden and allowed EPR transitions which have been
coherently excited using short, intense microwave pulses.‘8 The resonant microwave
pulses create electron spin echoes, and the spin echo decay envelop is then measured by
plotting the integrated intensity of these echoes as a function of the times between
pulses.19 The result shows an overall decay of the electron spin magnetization that is
modulated by electron-nuclear hyperfine Couplings.19 The common three-pulse ESEEM
experimental spectral lines correspond to nuclear frequencies, which simplifies the
identification of nuclei in the data. The 14N nitrogen nucleus we wish to examine is a
quadrupole with a nuclear spin of I = 1, therefore the nuclear-quadrupole interaction must
be taken into consideration. The spin Hamiltonian for an S = 1A», and I = 1 system is

shown in Equation 8.
H = gBS~B - ganI-B + A8] + I-Q'I (8)

The first three terms were already discussed above, and the fourth term is the interaction
of the nuclear spin with the quadrupole of the nuclei. Q is the nuclear quadrupole tensor
Which is traceless and is only non-zero for I > 1/2. Five parameters effectively describe
the nuclear quadrupole interaction (N QI): equ, the quadrupole coupling constant; 1], the

asymmetry parameter, and or, B, and y, the Euler angles that describe the orientation of the

102

equ principal axis system with respect to the principal axis system of the hyperfine

20

tensor. The energy level diagram for a rhombic quadrupole interaction (n ¢ 0) with

both A and g assumed to be isotropic is shown in Figure 4-4.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ll "‘1 " ' 1
’I I ‘~‘\\“ 1” l 6
m5 = l/z I," m] 0 \‘\____..r’/
,_—_—.—————_____. ........ r 5
”’ \\\ ’Ilm‘u“\ ‘
x “xv v
’1’ \\\ ml :2 1’1” 4
:13 ms =il/2 ’
2 1:1. ,_ 3
LI-J ,,,,,,,,,, 4|
\‘ mI :: _ 1 ,,,,,
\‘ ”______,,
\ . ." m1 ‘- 0
~————-<-———————— ........ 1 2
m5 = -l/2 ‘ g
‘\ m1: 1
“———“ ...... i 1 1
Electronic Nuclear Hyperfine Electric ‘
Zeeman Zeeman Coupl mg Quadrupole
B :

Figure 4-4. Energy level diagram for an S = l/z, l = 1 system near exact cancellation21 conditions.

The energies of the states are obtained by solving the 3 x 3 secular determinant in the

nuclear basis functions lmr) (m1: 0, i 1) for each his = i '/2 state, where the parameter K

= equ/4h and the nuclear Zeeman, on = ganB/h. Six transitions are allowed2

A 2
V56=3K+ {—5—an +n2K2 (9a)

A 2
v45=—3K+ [E-Vn] +1121<2 (9b)

 

 

103

 

A 2
v46 =2 {-5—an +6218 (9c)

 

 

 

2
v23 =3K+\[(—§—+vn] +112K2 (10a)
2
v12 =_3K+\/[-’§-+vn) +nzr<2 (10b)
2
VB =2\/[-’3‘—+vn] “fix2 (100)

where, Equations 9 a—c correspond to the upper m5 = 1/2 state, and 10 a-c give the

energies of the ms = -‘/2 spin manifold. A total of six allowed transitions are possible in
the experiment, but the number observed depends on the mixing between the various
states. The extent of mixing will depend upon the magnitudes of the quadrupole and

hyperfine terms in the Hamiltonian.

4.2 Experimental

General. All reagents were obtained from commercial sources and used without
further purification. HPLC grade methanol was distilled from Mg and 12, and degassed
15 times by gas exchange with N2. Elemental analyses, mass spectra, and ESI-MS were
obtained through the analytical facilities at Michigan State University. ' H NMR spectra
were recorded either on a Varian INOVA 300 MHz or a Varian Utility—Plus 500 MHz
spectrometer.

2,7-dinitrophenanthrenequinone. The synthesis of this ligand has been

modified from literature procedures.”24 A flask was charged with 3.00 g (14.4 mmol) of

104

phenanthrenequinone and then 40 ml of fuming nitric acid (D = 1.5) and 5 m1 of sulfuric
acid was added. The reaction mixture was refluxed for 30 minutes, and then allowed to
cool to room temperature. The mother liquor was then added to 300 ml of distilled water
and a yellow voluminous precipitate was immediately formed. The precipitate was
collected on a glass frit and washed with distilled water until the washings were neutral.
Once neutral, the precipitate was further washed with ethanol (20 ml x 3) and ether (20
ml x 3) and dried in vacuo. Recrystallization from 300 ml of hot glacial acetic acid
yielded 1.92 g (45 %) of yellow needles. Anal. Calcd for C14H6N206: C, 56.38; H, 2.03;

N, 9.39; Found: C, 55.83; H, 1.91; N, 9.14. mp = 299-300 °C. 1H NMR (tfashake-
DMSO-d6)25 8 8.71 (d, JHZ = 4.15), 8.69 (d, JHZ = 1.95), 8.56 (dd, JHZ = 2.20, 6.59).

2,7-dinitrophenanthrenesemiquinone (1). The generation of this radical was
modified from published procedures.26 In an oxygen free drybox, an excess of K metal
was added to a 1 mmol/L solution of 2,7-di-NOz-phenanthrenequinone in THF. The
solution was then stirred for 3-5 minutes, and during this time the color of the solution
turns pink in color. The solution was filtered, and then used immediately for EPR
experiments. Upon freezing the solution in liquid nitrogen the color changes to green,
which has been attributed to diamagnetic dimer formation.27

If the solution is stirred for an additional 30-40 minutes in the presence of excess
K metal a precipitate falls out of solution. Based on the EPR spectrum (vide infra), this
precipitate is paramagnetic and can be attributed to formation of a contact radical ion pair
where potassium is coordinated to the 2,7-dinitrophenanthrnesemiquinone radical.

2,7-dinitrophenanthrenecatechol. The synthesis of this ligand has been slightly

modified from literature procedures.23 A suspension of 1.00 g (3.35 mmol) of 2,7-di-

105

NOz-phenanthrenequinone in 35 ml ethanol was heated to 55-60 °C, then a mixture of 2.5
ml 80 % glacial acetic acid and 0.355 g (3.28 mmol) of phenylhydrazine was added drop
wise while stirring. The solution slowly becomes clear and red in color upon addition of
the phenylhydrazine/acetic acid mixture. The solution was stirred for 5 minutes and
filtered with a glass frit, and upon addition of 150 ml of distilled water a brick red
precipitate was immediately formed. The brick red precipitate was collected on a glass
frit and dried in vacuo. Yield 0.958 g (96 %). The complex should be stored under
nitrogen to prevent oxidation and solutions of the compound are very air sensitive. Anal.
Calcd for C14H3N206-0.5H20: C, 54.38; H, 2.93; N, 9.06; Found: C, 54.36; H, 2.71; N,
8.76.
Phenanthrenecatechol-SOZ. The synthesis of this complex has been modified
from literature procedures.28 Caution: when using any U V-light source, proper safety
glasses and protective clothing must be used to protect the eyes and skin. Additionally,
due to the production of ozone by the lamp the reaction must be carried out in a fume
hood. The experimental setup is depicted in Figure 4-5, and during the reaction the glass
plates of the fume hood are covered to avoid accidental UV-light exposure. A quartz
schlenk flask was charged with 1.15 g (5.52 mmol) of phenanthrenequinone and 40 ml of
benzene. S02 (g) was bubbled through the stirring suspension for 20-25 minutes, and the
schlenk flask was capped and placed under positive pressure of 802. The quartz schlenk
flask was then placed into a quartz water bath directly in the path of a 450 W HgH lamp
as depicted in Figure 4-5. The suspension was irradiated with the HgH lamp for 36
hours, and the schlenk flask was rotated by 120° every 12 hours to ensure maximum

penetration of light (some blackening occurs on the flask in the light path).

106

1".

ti
3

    
 

'Siiill
t't GQOZIVZ/Z

“it“

Figure 4-5. The experimental setup for the 36 hour photolysis reaction. The equipment pictured is a
quartz Schlenk flask, quartz water bath, and 450W HgH lamp.

A white precipitate forms after approximately 12 hours of irradiation, but the reaction is
not yet complete. After 36 hours the fine gray-white precipitate was collected on a glass
frit and washed with ether (10 ml x 3) and dried in vacuo to give 1.10 g of product (77 %
yield). The complex can be recrystallized from hot chloroform to give white needles.
Anal. Calcd for C14H804S: C, 61.76; H, 2.96; Found: C, 61.62; H, 2.65. mp = 200-202
°C. 1H NMR (CDCl3) 5 8.692 (m), 7.956 (m), 7.73 (m). IR (KBr, cm") 1669 s, 1603 s,

1452 3,1412 5, 1348 s, 1214 s, 858 s, 808 s, 750 s, 718 s, 699 s.

 

3,6 diult. _ ' “ ‘L L ‘ $02. The synthesis of this complex has been

modified from literature procedures.28 3.0 g (11 mmol) of phenanthrenecatechol-SOZ was

added to 100 ml of ACS grade nitric acid (D = 1.4) at 110 °C, the solution is then quickly

107

heated to boiling and stirred for 3-4 minutes. The dissolved phenthrenecatechol-S02
gives a red solution and red-brown gas is evolved, after brief stirring a yellow precipitate
falls out of the hot solution. The reaction mixture was allowed to cool to room
temperature and then cooled in an ice bath to ensure all the precipitate falls out of
solution. The precipitate was collected and washed with 30 ml of methanol and 30 ml of
ether and dried in vacuo. The complex is then dissolved in 170 ml of simmering acetic
anhydride, and left out to crystallize. The flask was then placed in a freezer overnight to
complete precipitatation. The precipitate was collected on a glass fi'it and washed with
methanol (30 ml x 4) and ether (30 ml x 4) and dried in vacuo to give 1.3 g (33 %) of
yellow needles. The precipitate was reported to be light sensitive and should be protected
from right.28 Anal. Calcd for C14H6Nzogst c, 46.42; H, 1.67; N, 7.73; Found: c, 46.60;
H, 1.27; N, 8.07. mp = 240-241 °C.
3,6-dinitrophenanthrenequinone. The synthesis of this compound was based on
methods already developed in the literature.28 1.3 g of 3,6-di-NO2-phenanthrenecatechol-
S02 was heated in a sand bath at 250 °C until the evolution of S02 (g) ceases (typically
30-45 minutes). The complex is then recrystallized from a minimum amount of hot
acetic anhydride to give orange-yellow plates. Yield 0.78 g (73 %). Anal. Calcd for
C14H6N206: C, 56.39; H, 2.03; N, 9.39; Found: C, 56.80; H, 1.61; N, 9.39. mp = 292-295
°C. ‘H NMR (CD3CN) 8 9.32 (m, 2H), 8.35 (d, 4H).
3,6-dinitrophenanthrenesemiquinone (2). The generation of this radical was
done using the same procedure for the 2,7-isomer. The color of the radical solution in

THF is violet/purple, and upon freezing the color of the solution turns green again most

108

likely due to diamagnetic dimer formation.27 The solution was immediately used for EPR
experiments.

3,6-dinitrophenanthrenecatechol. The synthesis of this ligand has been slightly
modified from literature procedures for the 2,7-di-N02 isomer.23 A suspension of 0.41 5 g
(1.39 mmol) of 3,6—N02-phenanthrenequinone in 20 ml ethanol was heated to 55-60 °C,
then a mixture of 1.00 ml 80% glacial acetic acid and 0.137 ml (1.36 mmol) of
phenylhydrazine was added drop wise while stirring. The solution slowly becomes clear
and red in color upon addition of the phenylhydrazine/acetic acid mixture. The solution
was stirred for 15 minutes and filtered with a glass frit, and upon addition of 150 ml of
distilled water a red-orange precipitate was immediately formed. The red-orange
precipitate was collected on a glass frit and dried in vacuo. The complex should be
stored under nitrogen to prevent oxidation, and solutions of the compound are very air
sensitive. Yield 0.384 g (92 %). Anal. Calcd for C14H3N206: C, 56.01; H, 2.69; N, 9.33;
Found: C, 56.62; H, 2.67; N, 9.10.

Acetyl-ferrocenium Tetrafluoroborate. A literature procedure for this complex
has been published,29 but another synthetic preparation was adapted that yielded better
results.30 In a schlenk flask, a filtered solution of 60 ml ether and 0.21 g (1.9 mmol) of p-
quinone was degassed for 5 minutes by bubbling N2 through the solution, then the flask is
placed under positive pressure of N2. 0.63 ml of HBF4-0Et2 was then added to the
ethereal solution of p-quinone. Immediately after the HBF4-0Et2 addition, a filtered
solution of 0.22 g (0.96 mmol) acetyl-ferrocene in CHC13 was then added dropwise to the
stirring solution under N2. The solution immediately turns deep blue in color, and after

stirring for 10 minutes the solution is filtered and the blue precipitate is washed with

109

ether (30 ml x 4) and dried in vacuo. The complex is stored under nitrogen to prevent
oxidation, and solutions of the compound are very air sensitive. Yield: 0.24 g (81 %).
Anal. Calcd for C12H12FeOBF4: C, 45.77; H, 3.84; Found: C, 45.09; H, 3.95.
[Ga(tren)C12]Cl. In the dry box, 0.858 g (4.87 mmol) of anhydrous GaC13 was
dissolved in EtOH and added slowly in a drop wise fashion to a stirring solution of 0.784
g (5.36 mmol) of tris~2~aminoethylamine in EtOH, and during the addition a white
precipitate begins to form. The mixture was stirred for an additional 15 minutes, and
then the precipitate was filtered off and washed with EtOH (20 ml x 3) and ether (30 ml x
4) and dried in vacuo. The complex should be stored and used in the drybox, because it is
very hygroscopic. Yield: 1.30 g (83 %). Anal. Calcd for C6H13N4GaCl3: C, 22.36; H,
5.63; N, 17.38; Found: C, 22.43; H, 6.16; N, 17.11.
[Ga(tren)(2,7-di-NO2-PCAT)]Cl. In the dry box, a flask was charged with 0.068
g of 2,7-NO2-phenanthrenecatechol (0.23 mmol) and was partially dissolved with 10 ml
of degassed MeOH, and then 0.019 g NaOH (0.48 mmol) in 10 ml of methanol was
added to this solution. The solution changes in color from red to blue-purple. The
solution was then stirred for 5 minutes to ensure that all solids were dissolved. In a
separate flask, 0.068 g [Ga(tren)C12]Cl (0.21 mmol) was dissolved in 20-30 ml of MeOH,
and then filtered. The [Ga(tren)Cl2]Cl MeOH solution was then added drop wise to a
stirring solution of the deprotonated 2,7-di-N02-phenanthrenecatechol, and during the
addition a dark blue-purple precipitate forms. The solution was then stirred for 45
minutes in the dry box. The dark blue-purple precipitate was filtered off and washed with

MeOH (5 ml x 3) and ether (5 m1 x 3). The complex was used without firrther

110

purification. Yield: 0.052 g (45 %). MS [ESL m/z (rel. int.)]: [Ga(tren)(2,7-di-NO2-
PCAT)'2NaOH]+ 593.2 (100) [Ga(tren)(2,7-di-NO2-PCAT)]+ 513.2 (10).

[Ga(tren)(2,7—di-NO2-PSQ)](Cl)(BF4) (3). In the dry box, a vial was charged
with 0.024 g of [Ga(tren)(2,7-di-NO2-PCAT)]Cl (0.044 mmol) and 10 m1 acetonitrile.
The stirring suspension was oxidized with 0.014 g acetyl-ferrocenium tetrafluoroborate
(0.044 mmol) in 2 ml of acetonitrile, and the color immediately changed from blue-
purple to dark green. Also, upon the addition of the oxidant the suspension goes into
solution. The solution was stirred for 5 minutes and filtered. Ether was added until the
solution turned cloudy, and brown microcrystals formed upon standing. The complex
was washed with CH2C12 (5 ml x 3) to remove any unreacted oxidant. The complex can
be recrystallized from acetonitrile/ether mixtures. Yield: 0.010 g (37 %) Anal. Calcd for
C20H24N606GaC1BF4'0.5CH2C12: C, 36.83; H, 3.68; N, 12.27; Found: C, 37.39; H, 3.36;
N, 12.59.

[Ga(tren)(3,6-di-N02-PCAT)]Cl. In the dry box, a flask was charged with 0.060
g of 3,6-dinitrophenanthrenecatechol (0.20 mmol) and was partially dissolved with 10 ml
of degassed MeOH, and then 0.016 g NaOH (0.40 mmol) in 10 ml of methanol was
added to this solution. The solution changes in color from red to red-purple. The
solution was then stirred for 5 minutes to ensure that all solids were dissolved. In a
separate flask, 0.058 g [Ga(tren)Cl2]C1 (0.18 mmol) was dissolved in 20-30 ml of MeOH,
and then filtered. The [Ga(tren)Cl2]Cl MeOH solution was then added drop wise to a
stirring solution of the deprotonated 3,6-di-N02-phenanthrenecatechol, and during the
addition a dark—purple precipitate forms. The solution was then stirred overnight in the

dry box. The dark-purple precipitate was filtered off and washed with MeOH (5 m] x 3),

111

EtOH (5 ml x 3), and ether (5 ml x 3). The complex was used without further
purification. Yield: 0.080 g (82 %). MS [ESL m/z (rel. int.)]: [Ga(tren)(3,6-di-NO2-
PCAT)]+ 513.2 (20).

[Ga(tren)(3,6-di-NO2-PSQ)](BF4)2 (4). This gallium semiquinone radical was
generated by first adding 0.024 g of [Ga(tren)(3,6—di-N02-PCAT)]C1 (0.044 mmol) and

10 ml of acetonitrile to a flask. This suspension was stirred and 0.0090 g of AgBF4
(0.046 mmol) was added in 1 ml of acetonitrile to exchange the CT for BF4'. The

solution was stirred for an additional 15 minutes and then filtered through celite. The
purple solution was then oxidized by adding 0.014 g of acetyl-ferroceniurn
tetrafluoroborate (0.044 mmol) in 2-3 ml of acetonitrile, and the color immediately
changed to red-brown. Only freshly prepared solutions were used for EPR experiments,
and the solution is very air sensitive.
EPR Spectroscopy. In general, X-band EPR spectra were acquired on samples
(~ 10 mmol/L) and prepared in a dry, inert atmosphere (N2). The details of each sample
preparation are described below. Room temperature continuous-wave (cw) X-band EPR
spectra for all samples were obtained with a Bruker ESP300E spectrometer. Spectra of
radicals l and 2 were acquired with a perpendicular-mode cavity (Bruker ST4102/ 8943)
with resonances around 9.46 GHz, while complexes 3 and 4 required an EPR flat cell for
acquisition at room temperature with a room temperature cavity (Bruker 4103TM) that
has resonances around 9.67 GHz. The power dependence of all signals was checked to
ensure that data acquisition was carried out below saturation. All simulations of X-band

EPR spectra were carried out using the Bruker XSOPHE program.3 "33

112

Electron Spin Echo Envelop Modulation (ESEEM) spectra for 1, 2, 3, and 4 were
acquired using a Bruker E-680X pulsed X-band spectrometer with a model ER4118-MD—
X-5W1 probe that employs a 5 mm dielectric resonator. The sample temperature was
maintained at 50 K using an Oxford Instruments liquid helium flow system equipped
with a CF -935 cryostat and a ITC-503 temperature controller. ESEEM data were
collected using a three-pulse, stimulated echo sequence (90°—t—90°—T—90°) with 90°
microwave pulse widths of 16 ns (full width at half maximum) and peak powers of 250
W. A four-step phase cycling sequence, (+x, +x, +x), (-—x, +x, +x), (+x, —x, +x), (-—x, ——x,
+x), together with the appropriate addition and subtraction of the integrated spin echo
intensities served to actively remove the contributions of two-pulse echoes and baseline
offsets from the data.34 An integration window of 24 ns was used to acquire spin echo
amplitudes, and data set lengths were 512 points. ESEEM spectra were tapered with a
Hamming window and Fourier transformed.35 ESEEM spectra were obtained by taking
the absolute value of the Fourier transformations.

ESEEM simulations of the I = 1 14N hyperfine coupling constants were
accomplished by using software written in MATLAB. Calculations used the density
matrix formalism of Mirns to simulate the time domain ESEEM data.”37 MATLAB
scripts were then used to assemble the simulations according to the product rule and to
complete the Fourier analysis to obtain simulated ESEEM spectra. The processing and
Fourier transformation procedure was identical to that outlined above for the Bruker
software. The spherical model approximation to the product rule was used for

simulations of ESEEM patterns of multiple nuclei.38

113

Calculations. Previous DF T calculations on semiquinone radicals have shown
that hybrid functionals such as B3LYP are suitable for calculating hyperfine coupling
constants and spin densities.39 The DFT calculation of '4N hyperfine coupling constants
was shown to be basis set dependent with the 6-31g* basis set and the B3LYP functional
yielding the best results with experimental data.40 It has also been shown that solvent has
a strong influence on the hyperfine coupling constants of semiquinone radicals.41 Taking
previous work into consideration, the starting structures for the radical ligands were
generated using GaussView,42 and then optimized using Gaussian 0343 with the UB3LYP
firnctional and the 6-31 g* basis set. Further single point gas phase calculations were then
done using the UB3LYP, UB3PW91, and UPBElPBE functionals with the 6-31g* basis
set, and the hyperfine coupling constants were calculated by the Fermi analysis method in
the Gaussian 03 program. The spin density was examined by the natural population
analysis (NPA)44 method and by the Mulliken population analysis scheme incorporated in
Gaussian 03. The influence of solvent on the hyperfine coupling constants was
represented by the static isodensity surface polarized continuum model (IPCM).45 The
hyperfine coupling constants were also calculated in tetrahydrofuran using the UB3LYP,
UB3PW91, and UPBElPBE functionals with the 6-31g* basis set including the scrf =
IPCM keyword with the isodensity value of 0.0004 (default) and a dielectric constant of
7.58. All single point calculations were done using tight convergence criteria,46 and the
calculations assumed a charge of ——l and a doublet spin state.

The starting structures for the gallium complexes (3 and 4) were adapted from the
crystal structure of an analogous [Ni(tren)(PSQ)]+ complex (replacing NiII with Gam)

using GaussView, then optimized with Gaussian 03 using the UB3LYP functional with

114

the LanL2DZ basis set. Previous calculations of gallium(lll) semiquinone dimer systems
have shown that hybrid functionals give good estimates of the corresponding
experimental hyperfine coupling constants and spin densities.47 Subsequent single point
calculations were then carried out using the UB3LYP, UB3PW91, and UPBElPBE
functionals with the 6-31g* basis set, and the isotropic hyperfine coupling constants were
calculated by means of the Fermi analysis method in the Gaussian 03 program. The spin
density was examined by the natural population analysis (NPA)44 method and by the
Mulliken population analysis scheme incorporated in Gaussian 03. The effect of solvent
on the hyperfine coupling constants was also explored by means of the polarizable

continuum model (PCM)48'5 ]

and IPCM models. All single point calculations were done
using tight convergence criteria, and the calculations assumed a charge of +2 and a

doublet spin state.

4.3 Results and Discussion

4.3.1 Hyperfine Coupling in 2,7- and 3,6-dinitrophenanthrenesemiquinone Radicals
There are 3 sets and 1 set of equalivant protons and nitrogens, respectively, in
complexes 1 and 2 (Figure 4-6). In Figure 4-6, the protons are labeled by the carbon
atom that they are bonded to, for example the equivalent protons bonded at positions 1
and 8 are labeled as 1H“. The room temperature EPR Spectrum of 1 was recorded

4,26’27 and the hyperfine coupling constants were

previously by Maruyama in 196
determined to be 71041427) = 0.0 G, 710143,): 1.94 G, A(‘HL3) = 0.97 G, and A(1H45) =

0.44 G.

115

(N2 0' H2) (H7 0' N7)

(N3 or H3) H4 H5 (H6 or N6)

Figure 4-6. Generic proton labeling in the 2,7-di-N02—PSQ (1) and 3,6-di-N02-PSQ (2) radicals.

To our knowledge there is no previous EPR spectrum of the corresponding 3,6-di-N02-
phenanthrenesemiquinone (complex 2). In our efforts to determine the spin distribution
in these complexes, an analysis of the room temperature X-band EPR spectra of both of
the radicals was carried out. Specifically, the 14N HFC’s are important in this
examination, since the HFC constant suggests how much spin is delocalized onto the
nitrogen nucleus of the N02 substituent. In essence, we can experimentally approximate
the spin delocalizing effect of the nitro-substituent in the PSQ ring. EPR spectra of l and
2 generated by reduction of the quinones with excess K metal in THF solution under a
nitrogen atmosphere are depicted in Figures 4-7 and 4-8, respectively. The simulation of
1 does not include a contribution from the two equalivant l4N27 nuclei of the nitro groups
consistent with what was observed by Maruyama. The simulation of the data is good and
all lines are accounted for except for the peaks on the ends of the spectrum. Although, if
the simulation of the spectrum included two 14N nuclei the two side peaks could be
simulated, but the overall results were not as good as depicted in Figure 4-7. Thus, it
appears necessary that l4N HF C is needed to accurately simulate the side peaks in the
Spectrum, but the coupling is probably very weak. The proton hyperfine coupling

constants agree quite nicely with the reported data as seen in Table 4-1. DFT calculations

116

at the UB3LYP/6-31g* level were used to aid in the assignment of the proton HFC’s, and
the agreement between experiment and theory is good with the deviations being less than
0.35 Gauss. These deviations are within the range of errors of 0.10-0.90 G observed by

O’Malley in the DFT calculated lH HFC of semquinones.39’52

11.1.1.1.1.1.1.1.1.
3350 3352 3354 3356 3358 3360 3362 3364 3366

B (G)

Figure 4-7. Experimental X-Band EPR spectra of 2,7-di-N02-PSQ (l) in black (top) acquired in THF at
room temperature, and simulation in red (bottom). The simulation parameters are: g = 2.01023, A(HLg) =

0.936 G, A(H3,6) = 1.952 G, A(H4‘5) = 0.393 G, and linewidth = 0.15 G. Experimental Conditions:
frequency = 9.4479, modulation amplitude = 0.1 G, and power = 1.25 mW.

The EPR spectrum and simulation of complex 2 are shown in Figure 4-8. The
simulation in this case did include a contribution fi'om the MN nuclei in the radical, and
the hyperfine coupling constants are compared with results fi'om DFT calculations shown
in Table 4-1. The assignments of the 1H isotropic HFC’S for 2 was again aided by DFT

calculations at the UB3LYP/6-31g* level and by comparison with complex 1.

117

 

 

 

 

 

 

 

L_L4LLI 1'1_I L'AJ LJ LJ

3352 3354 3356 3358 3360 3362 3364 3366 3368
B(G)

Figure 4-8. Experimental X-Band EPR spectra of 3,6-di-N02-PSQ (2) in black (top) acquired in THF, and
simulation in red (bottom). The simulation parameters are: g = 20106., A(N3,6) = 0.406 G, A(Hl,8) =
0.413 G, A(H2,7) = 1.197, A(H4,5) = 0.407, and linewidth = 0.262 G.

The overall agreement between the UB3LYP/6-31g* calculated and experimental lH
HFC’s is good with a maximum deviation of approximately 0.20 Gauss. The calculation
of the MN HFC’s at the UB3LYP level gave a reasonable comparison with what was
experimentally determined with a difference of 0.28 G. This is in excellent agreement
with literature calculations, since typical deviations between calculated and experimental
l4N HFC’s in the literature were on the order of 1.67 G for a large series of nitrogen
centered radicals.40 The 14N of 1 can be estimated to be smaller than 0.413 G based on
the DFT and the data for 2. These small HFC constants imply that spin polarization is
largely responsible for the spin density on the nitrogen nucleus, since nitroaromatic
radicals in the literature exhibit much larger 14N HFC when spin is directly delocalized

onto the nitrogen atom.”'56

118

Table 4.1. Experimental and DFT isotropic HFC’s for 2,7-di-N02-PSQ (1 ) and 3,6-di-NOz-PSQ (2).

 

b DFT;
Radical Exp. DFT-gas IPCM '6
HFC (G): HFC (G): HFC (G):

B3LYP B3PW91PBE1PBE B3LYP B3PW91PBE1PBE

'11,,8 0.94(0.97)a 0.788 0.87701 1.03855 0.64384 0.69379 0.79198
”N27 0.0(0.0)" 0.151 0.10046 0.04056 0.28668 0.22278 0.15991
‘11,,6 1.95 (1.94)“ -231 -249943 25529 -2.57818 0.79495 -2.87572
‘11.,5 0.39(0.44)" 0.117 0.15014 0.20617 0.20052 0.24994 0.33512

2

‘11”; 0.413 -0.498 0.53699 0.5224 -0.5416 0.63843 0.70544
‘Hz,7 1.197 -1.018 -l.0966 0.12191 0.90553 0.928 0.88268
”N3; 0.406 0.171 0.11801 0.05853 0.55098 0.41096 0.2925
‘11,, 0.407 0.477 0.53473 0.67852 0.42527 0.51362 0.69666

 

a Ref. [29]. 7’ Using the 6-3 lg“ basis set. 0 Dielectric constant = 7.58, and isodensity = 0.0004 for 2,7-di-
NOz-PSQ and isodensity = 0.003 for 3,6-di-N02-PSQ.
We also used the B3PW91 and PBElPBE functionals with the 6-31g* basis set to
see if we could get better agreement/estimates of the MN and 1H HF C for both 1 and 2.
In some cases the HFC’s improved, but in others there were larger deviations observed in
the HFC’s, as seen by examining Table 4-1. The effect of solvent on the 14N HF CS was
also examined with the IPCM solvent model as suggested by Langgard and Spanget-
Larsen“, and we did exploratory calculations using isodensity values of 0.0004, 0.001,
and 0.003 for both radicals, with the results generally having greater agreement with the
experimental values. The isodensity value is an empirical parameter that is usually
adjusted for differences in protic and aprotic solvents, and increasing this parameter
effectively increases the strength of the solvent reaction field.41 Using the IPCM model
with THF as the solvent, an isovalue of 0.0004 was empirically found to give reasonable

HFC’s for 1. 1n the case of 2 also in THF, an isodensity parameter of 0.003 improved the

119

magnitude of the 14N HFC. It is interesting that the isomers needed different isodensity
values to better compare with experiments, and it is not clear why this is the case. We
also did the calculations using an isodensity value of 0.0004 for 2 which resulted in
increased errors compared to the gas phase calculations. Thus the improvements were
not consistent. Although there were improvements seen in the HN HFC’s by including
solvent for both radicals, there are greater errors in some of the 1H HF C’s, most notably
the 1H”, HFC. The larger errors in 1H HFC’s with the solvent models and the fact that
the isodensity values were different for the two isomers prompted us to use the gas phase
calculations for our comparisons with experimental values. The UB3LYP results gave us
the best overall comparisons with experiments, therefore these results will be used
throughout the rest of this work.

The DFT calculated lH HF C’s for l and 2 revealed both shifts in magnitude and
sign of the HF C between the two complexes. In order to better visualize and interpret the

results, the total spin density surfaces for both radicals are shown in Figure 4-9.

 

Figure 4-9. Spin density distribution in I (left) and 2 (right), where the blue color represents excess a-spin
density and the green represents excess B-spin density.

The spin density distribution calculated by the NPA analysis and Mulliken are

qualitatively similar, and we shall focus on the Mulliken population analysis for the

120

discussion of the results. The til-spin densities for the carbon, hydrogen, and nitrogen

atoms have been tabulated in Table 4-2.

Table 4-2. Experimental and calculated spin densities in 2,7-di-N02-PSQ (l) and 3,6-di-N02-PSQ (2).

Radical Experimental

s-orbital

p-orbital

Calculated s-orbital p-orbital

 

111‘ch p(Hor N)“ p(C)b HFC(G) p(HorN) p(C)c
1
'H,,, 0.936 0.000584 0.03928 0.788 0.001066 0.04003
”N27 0.0 0.0 0.151 0.00033 0.04056
‘11,,6 1.952 0.001236 0.083206 —2.31 0.004435 0.07921
‘Hr5 0.393 0.000263 0.017679 0.117 0.000351 0.01903
2
1111,, 0.413 0.000259 0.017426 0.498 0.001385 0.01075
'H2,7 1.197 0.000751 0.050506 -1.02 0.001933 0.03245
”143,6 0.406 0.007049 0.171 0.00037 0.04427
1114,, 0.407 0.000255 0.017172 0.477 0.000871 0.02546

 

a 6
Calculated using Equation 3a or 3b. The carbon p-Tc-spin density calculated using Equation 1 and

c
using the experimental IHM, HF C and QCHH = 23.7.57 Average total p-orbital spin density on the carbon
atoms bonded to the proton or nitrogen nucleus.

According to the DFT calculated HF C’s and from visualizing the spin density plots for 2,
the 1H”; and 1H27 HFC’s are negative in sign due to B-spin polarization of the s-orbital
spin density by the excess a-rt-spin density at the 1, 8, 2, and 7-carbon atoms.58 The 11145
HFC is positive due to s-orbital a-spin polarization from the B-spin density at the 4,5-
carbons. Similar arguments can also be made for the signs of the 2,7-di-N02-PSQ
HF CS. The agreement between DFT and the experimentally determined s-orbital spin
densities for the protons and the n—spin densities at the carbons is good using Equation 3a
and 3b. For isotropic hyperfine coupling (Fermi contact) to occur there must be excess

spin population at the nucleus, therefore the important variable is the s-orbital spin

density at the nucleus. The s-orbital spin density at a nucleus is directly proportional to

121

the magnitude of the isotropic HF C. For example, in 2 there is more calculated Mulliken
S-orbital Spin density at the 2,7-hydrogens (0.000751) then at the 1,8-protons (0.000259),
and the corresponding HFC’S at the 2,7-protons are larger. The magnitudes of all the
experimental HF C’s agree within the range of calculated HFC’S. The largest shifts
between the two radicals occurs at the 1HM; HFC. Figure 4-8 (and Table 4-2) shows that
the spin density in 1 is more centralized on the 2,7-carbons, whereas increased a-spin
population is present on the 2,7- and 1,8-carbons of 2, which is likely a result of
increased spin delocalization effects. Additionally, the s-orbital spin density polarization
onto the nitrogen atom is the same for both radicals as predicted by DFT: we can say to a
first approximation that the spin delocalizing ability onto the nitro group at the 2,7- and
3,6-positions is equivalent, but further experiments will be required to quantify

experimental 14N values for the radicals.

Now that l and 2 have been examined by X-band EPR spectroscopy, we will
further quantify the experimental l4N HF C’s by using the Electron Spin Echo Envelope
Modulation (ESEEM) technique. ESEEM is a pulsed EPR experiment that is often used
to elucidate weak hyperfine couplings that are sometimes masked by inhomogeneous
broadening of the EPR lineshape.19 In our case, we wish to complement the experimental
determination of the 14N HFC’S that was suggested in the room temperature X-Band EPR
experiments of the radical PSQ ligands with ESEEM data. Additionally, gallium(lll)
may broaden the EPR lineshape as already seen in the literature,47 requiring us to use
ESEEM Spectroscopy for comparisons of 14N between the radicals and the gallium
complexes (vide infra). This broadening is usually due to the larger contribution of

gallium HFC compared to hydrogen or nitrogen HF C.

122

4.3.2 ESEEM Spectroscopy of 2,7- and 3,6-di-N02-phenanthrenesemiquinones
The experimental conditions were identical for generating 2 in the X-band EPR
experiments, and the ESEEM data were collected in frozen THF solution at 50 K. The

resulting data and simulation are shown in Figure 4-10.

 

 

 

 

 

 

 

 

41-
L.
3t
b
'52-
1:
£3
E 1-
11-
L
()1. MN
I J n i L 1 L L m
0 2 4 6 8 10
MHz

Figure 4—10. Stimulated echo detected ESEEM spectra of 3,6-di-N02-PSQ (2) (black) at 50 K with
frequency = 9.682 GHz; center field = 3445 G; 1: = 136 ns; starting T = 40 ns; time increment = 16 ns.

Simulation (red) parameters: g = isotropic, A = (1.3, 1.1, 0.9) MHz, e2qQ = 1.2 MHz, Euler = (0.0, 0.4,

0.0), and n = 0.8. Spectra were obtained by normalizing the raw data and subtracting the bi-exponential
decay followed by application of a Hamming window and Fourier transformation. Absolute value spectra
are shown. Recall, 1.0 G = 0.1 ml“ = 2.8025 MHz.

It should be noted that simulation of ESEEM spectra is much more complicated than X-
band EPR Spectra, and all ESEEM simulations presented in this chapter are within 10-20
% of the experimental data. Work is currently ongoing to improve all simulations, and
the best results are reported henceforth. The critical step in all ESEEM data simulation is
the accurate simulation of the experimental time domain data as well as the fiequency

domain spectra. For 2 the experimental l4N ESEEM modulations are observed at 0.82,

123

1.47, and 3.18 MHz. The simulation of the 14N ESEEM spectra in Figure 4-10 were done
numerically and employed a spin Hamiltonian that included the nuclear Zeeman,
electron-nuclear hyperfine, and nuclear quadrupole interactions. The simulation
parameters were identical for three 1: values used to acquire the experimental spectra: g =
isotropic, A = (1.3, 1.1, 0.9) MHz for (Axx, AW, A22), 11 = 0.80, and equ = 1.2 MHz.
The isotropic l4N HFC is equal to 1/3 the trace of the anisotropic HFC tensor, or 0.39 G
(1.0 G = 2.8025 MHz), and this is in reasonable agreement with 14N = 0.43 G estimated
from simulating the room temperature X-band spectra.

One regime for observing 14N ESEEM frequencies is a circumstance near exact

21 ,59,60

cancellation conditions, where the nuclear Zeeman frequency, on, is equal to one-

half the hyperfine coupling constant A (i.e. on = A/2). These terms cancel making the
quadrupole interaction the dominant factor in the upper Spin manifold transitions (Figure
4-10). Three nuclear quadrupole transitions 1156, D45, and 046 in the upper m5 = 1/2 spin
manifold and a higher frequency double quantum transition udq (Aml = 2) in the lower

manifold (Figure 4-9) are observed and given below.

v56 =Vo = 2K’q
V45=v-=K(3-n) (11)
V46 =v+ =K(3+n)

A 2 2
Vl3=qu=2[[Vn+‘2—] +K (3+n )] (12)

The first two transitions 056 and D45 essentially add up to 046, and the vdq transition is

usually observed at 4-6 MHz for exact cancellation conditions. In this case the data

124

indicate conditions near cancellation, where four frequency components are also usually
resolved.“ The first peak in Figure 4-10 at 0.82 MHz is most likely the combination of

two frequencies which add to the peak observed at 1.47 MHz. Additionally, the peak
observed at 3.18 MHz can be assigned to the udq transition.

ESEEM was also attempted using the same experimental conditions as above for
1. Deep modulations were not observed in the THF data, but the characteristic nitrogen
lines near exact cancellation were observed. The signal to noise was quite high and a
quantitative l4N could not be determined, but it can be approximated that the 14N HFC is

quite small and less than 0.39 G observed for the 3,6-di-NO2-PSQ radical.

4.3.3 EPR and ESEEM Spectroscopy of Gallium 3,6- and 2,7-di-N02-PSQ

The room temperature X-band EPR data for 3 and 4 are shown in Figure 4-11 and

4-12 below, respectively.

I l U j

I fl 1
3440 3460 3480

B (G)

l ' 1
3400 3420

Figure 4-11. The room temperature X-band EPR spectrum of [Ga(tren)(2,7-di-N02-PSQ)](Cl)(BF4) (3) in
acetonitrile. Experimental conditions: frequency = 9.772 GHz, power = 5 mW, modulation frequency =
100 G, modulation amplitude = 0.2 G. See text for further details.

125

 

L L L L I r _l_ m .1

3400 3420 3440 3460 3480
B (G)
Figure 4-12. The room temperature X—band EPR Spectrum (top) and simulation (bottom) of [Ga(tren)(3,6-
di-N02-PSQ)](BF4)2 (4) in acetonitrile. Simulation parameters: g = 2.00979, A(Ga) = 6.20 G, A(H) = 0.90

G, line width = 2.90 G. Experimental conditions: frequency = 9.672 0111, power = 5 mW, modulation
frequency = 100 G, modulation amplitude = 0.2 G. See text for further details.

Figure 4-12 depicts the expected 4 line pattern (21 +1) from a Single Ga (1 = 3/2) nucleus,
and the signal is relatively broad and featureless analogous to other Ga—SQ complexes
reported in the literature.47 The experimental spectrum for 4 could be fit reasonably well
using only Ga HFC, yielding g = 2.00987, A(Ga) = 6.74 G, and a line width = 3.4 G (not
Shown), but Figure 4-11 shows an improved simulation by including pair of equivalent
protons.

The room temperature spectrum of 3 is different from 4 but does resembles a
similar gallium(III) o-iminobenzosemiquinonate radical dimer from the work of
Wieghardt and coworkers,62 where an identical line shape and pattern due to Gam
hyperfine coupling (1 = 3/2) is observed. From the isotropic A(Ga) = 4.6 G in the
Wieghardt and coworkers complex, and from the peak positions of the Ga—2,7-di-NO2-

PSQ spectrum we estimate a gallium HFC of about 6.2 G. We were unable to simulate

126

the spectrum by using Ga"I alone, and several features point to the fact that other
hyperfine interactions are contributing significantly to this spectrum. The broad,

unresolved Gal"

HF C and considerable fine structure on top of the peaks (the second
derivative spectrum reveals much of this structure, but we have been unable to simulate it
at this point) points to contributions from other nuclei (”N and/or lH). This fine structure
was not observed in complex 4, and suggests that the MN and/or lH HFC’s are larger in
magnitude in 3 due to the additional fine structure that is observed. The magnitude
changes in the MN HF C will also be examined by ESEEM Spectroscopy below, which
may confirm our suspicions for a larger magnitude l4N coupling constant in 3.

The ESEEM spectrum and simulation of 4 are shown in Figure 4-13 below. The
data reveal l4N modulations at 0.73, 1.52, and 2.30 MHz which are typical of transitions
near exact cancellation conditions for the upper spin manifold, and the double quantum
transition of the lower spin manifold at 3.39 MHz is also observed. The transition at 4.62
MHz is due to a combination of the 0.73, 1.52, and 2.30 MHz frequency components, and
this peak usually arises when more than one nucleus with the same magnetic coupling
gives rise to the ESEEM.‘53 The experimentally determined A = (2.0, 1.3, 0.8) MHz gives
an isotropic 14N of 0.487 G, which is increased in comparison with the free ligand
experimental data. We note that the simulation iS not as good as in complex 2, and
within the 10-20 % conservative error bars, the 14N for the free ligand and gallium
complex are qualitatively the same. The absence of any large anisotropy in the 14N HFC

tensor indicates that the spin density at the nitrogen nucleus is mainly due to spin

polarization. Thus, only a small change if any is observed in the MN HFC upon the

127

binding of the 3,6-di-N02-PSQ ligand to gallium(III). More accurate simulations are

required to quantitatively compare these complexes.

 

 

 

 

6..
4-
.53
m
a .
.93
"9' U

 

 

 

 

0 r
J _1 I 4 J _L L L J 4
0 2 4 6 8 10 12
MHz

Figure 4-13. Stimulated echo detected ESEEM spectra of [Ga(tren)(3,6-di—N02-PSQ)](BF4)2 (4) (black)
at 50 K in acetonitrile with frequency = 9.694 GHz (black); center field = 3450 G; starting T = 40 ns; t =

136 ns; time increment = 16 ns. Simulation (red) parameters: g = isotropic, A = (2.0, 1.3, 0.8) MHz, equ

= 2.5 MHz, ‘1 = 0.60, and Euler angles = (0.0,0.9,0.0). Spectra were obtained by normalizing the raw data
and subtracting the bi-exponential decay followed by application of a Hamming window and Fourier
transformation. Absolute value spectra are shown.

The MN ESSEEM spectra for 3 (Figure 4-14) depicts a broad feature centered at
2.9 MHz, which differs from exact or near exact cancellation conditions. This feature
results from hyperfine splittings that are dominated by electron-nuclear hyperfine
coupling where both the isotropic and anisotropic contributions are large.61 This broad
peak has been assigned to a “AM; = 1” transition, and its assignment is discussed in detail
by McCracken et al.61 and will not be further elaborated upon here. The large anisotropic

HFC of the 1"'N in this complex is evident in the spectral simulation, where

128

”N = (-2,1, -2.1, 21) MHz for (Axx, Ayy, A22) and the isotropic HFC is also quite large

with ”N = 2.0 G.

3.5

 

3.0 l-

2.5 >-

2.0 r'

Intensity

0.5 -

 

 

 

MHz

Figure 4—14. Stimulated echo detected ESEEM of [Ga(tren)(ZJ-di—NOg-PSQ)](C1)(BF 4) (3) (black) at 50
K in acetonitrile with frequency = 9.676 GHz; center field = 3450 G; t = 136 ns; starting T = 40 ns; time
increment = 16 ns. Simulation (red) parameters: g = isotropic, A = (-2.1, -2.1, 21) MHz, e qQ = 1.6 MHz,
11 = 0.60, and Euler angles = (0.0, 0.9, 0.0). Spectra were obtained by normalizing the raw data and
subtracting the bi-exponential decay followed by application of a Hamming window and Fourier
transformation. Absolute value spectra are shown.

If the unpaired spin density is largely localized in the pz-n-orbital of the radical, then the

1 In the

hyperfine coupling would be expected to have a large anisotropic contribution.6
present case, this implies that Spin density is most likely directly delocalized into the
nitrogen atom’s pZ orbital by the spin delocalization mechanism. The resonance structure
on the right shown in Figure 4-15 is then deemed to be highly important for this complex.

The large increase in 14N from 1 to 3 is most likely real, since even with a possible error

of 20 % there is Still a large increase in the magnitude of the hyperfine coupling constant.

129

/1“e\
O 03 a

Figure 4-15. A representation of the spin delocalization mechanism onto the nitrogen of the N02 group in
the [Ga(tren)(2,7-di-N02-PSQ)](Cl)(BF4) (3) complex.

The room temperature X-band data also support a larger contribution from the ”N
HF C due to the observed fine structure on top of the Ga"I HF C. Thus, the X-band EPR
data and more conclusively the ESEEM of complex 3 indicate that the spin density is
more highly localized at the nitrogen than for the corresponding Ga-3,6-di-NO2-PSQ
isomer. The direct spin delocalization onto the nitrogen atom is most likely responsible
for the large experimental ”N HFC, and the smaller HF C for 4 is more in line with a
mechanism like spin polarization dominating in the population of unpaired spin density
on the nitrogen atoms. This experimental examination then indicates that there is a
greater population of spin density at the 2,7-positions than the 3,6-positions in the PSQ
. ring, and substituents in the 2,7—positons will most likely delocalize spin density to a

greater extent than at the 3,6-positions.

4.3.4 Potassium 2,7-di-N02-PSQ Contact Radical Ion Pair

In an effort to obtain better ESEEM data for the 2,7-di-NO2-PSQ radical, a
fortuitous observation was noted. During the reaction to generate the anion radical it was
observed that a dark precipitate fell out of the THF solution. The precipitate was

dissolved in ethanol and the EPR spectrum was collected using a flat cell (Figure 4-16).

130

It should be noted that only very pure 2,7-di-NOz-PQ ligand, potassium metal, and THF
were used during the reaction where this precipitate was formed. Interestingly, the
observed spectrum is quite different from the radical species (complex 1) that is soluble

in THF (Figure 4-6). The likely species in solution during the reduction reaction are 2,7-
di-NOz-PQ, 2,7-di-N02-PCAT, 2,7-di-N02-PSQ, K+, K0, and THF. It is highly unlikely
then that the spectrum in Figure 4—16 is due to any other species except the 2,7-di-N02-

PSQ radical, since this is the only paramagnetic species generated.

 

 

 

 

 

 

 

 

L l_ L L n 1 m L _. L
3450 3455 3460 3465 3470
B (G)

Figure 4-16. X-Band CW spectrum in ethanol at room temperature of potassium 2,7-di-N02-PSQ contact
radical ion pair (top) and simulation (bottom). Simulation parameters: g = 2.01030, A(N) = 1.5545 G,
A(H) = 0.398 G, A(H) = 0.383 G, and linewidth = 0.215 G. Experimental conditions: frequency = 9.734
GHZ, modulation amplitude = 0.1 G, and power = lmW. See text for further details.

It is hypothesized then that the precipitate is a potassium radical anion salt, where
potassium is directly coordinated to the semiquinone oxygens. This would explain the
greater solubility of the precipitate in more polar solvents. These species are known in

the literature as contact radical ion pairs, and their X-ray and EPR spectra have been

131

reported.64 The species reduced in THF (Figure 4-6) can then be described as a solvent
separated radical ion pair, where the potassium is not directly coordinated to the radical.
Additionally, the lineshape of the K-2,7-di-N02-PSQ radical is analogous to the 3-bis(2-
pyridyl)quinoxaline potassium contact radical ion pair,65 which suggests that the 2,7-di-
NOz-PSQ may be coordinated to potassium in this case. The spectrum of both contact
radical ion pairs is dominated by the interaction of the unpaired electron with the nitrogen
nucleus producing five hyperfine splitting lines (2nI + 1). Contributions from the
remaining nuclei are also superimposed on the nitrogen lines in both cases. The observed
fine structure corroborates with the structure observed in the X-band CW spectrum of 3.
The simulation of the K-2,7-di-NOz-PSQ data yields a l4N = 1.55 G, in qualitative
agreement with the 14N = 1.23 G of a dinitrobenzene contact radical ion pair()5 and with
the A(MN) determined from ESEEM for 3. Therefore, this data support both the ESEEM
and X-band EPR data for 3, where the 14N was found to be larger when compared to
complex 4.

ESEEM was also collected on the K-2,7-di-N‘Oz-PSQ complex, and upon freezing
the solution in ethanol a color change from brown to violet-red was observed. Recall that
the solvent separated potassium radical is pink in solution. The color change upon
freezing then suggests that the K-2,7-di-N02-PSQ is dissociating upon cooling to give a
more concentrated solution of solvent separated anion radicals. This hypothesis is
supported by literature precedence, since the dissociation of contact radical ion pairs can
result from favorable counterion solvation at low temperatureéi66 The resulting ESEEM
spectrum is shown in Figure 4-17. The spectrum resembles conditions away from exact

cancellation, and the signal is very weak with the isotropic 14N = 0.15 G.

132

 

0.7
0.6 l
0.5 l
0.4 —

0.3 -

Intensity

0.2 -
0.] -

0.0 —

 

 

 

Figure 4-17. Stimulated echo detected ESEEM spectra of 2,7—di-NOz-PSQ (1) (black) at 50 K in ethanol
with frequency = 9.685 Gl-lz; center field = 3450 G; t = 136 ns; starting T = 40 ns; time increment = 16 ns.
Simulation (red) parameters: g = isotropic, A = (0.65, 0.35, 0.05) MHz, equ = 1.1 MHz, 1’] = 0.80, and
Euler angles = (0.0, 0.2, 0.0). Spectra were obtained by normalizing the raw data and subtracting the bi-
exponential decay followed by application of a Hamming window and Fourier transformation. Absolute
value spectra are shown. See text for further details.

The additional two experimental peaks that were not simulated could be do to another
species that is present in the system due to non-isotropic conditions. Although, the 14N
value is identical to the DFT calculated MN at the UB3LYP/6-31g* level for 1, and lends
additional support for the simulated ESEEM data. This ESEEM data corresponds to a
spin polarization mechanism for populating spin density at the nitrogen nuclei and the
HFC is in the range of what has been observed for 2. Thus, the ESEEM data indicate that
the species in frozen solution is most likely the solvent separated 2,7-di-N02-PSQ
radical, and the spin density at the nitrogen nucleus is very small and agrees with the
prediction from the THF experimental ESEEM data. Therefore, the combined data for

the contact radical ion pair and the dissociated ligand contribute additional strong

133

experimental support for the spin density being directly delocalized onto the nitrogen
atom in 3, and for spin polarization being the dominate factor for 4.

A possible mechanism for the different spin delocalization/polarization in the
gallium complexes is based upon experimental data in the literature from the work of
Rockenbauer et al.14 where the EPR data of a tetrahedral Cul-2,7-di-N02-PSQ analogue
determined that the [Hm > 1H16 in contrast with what has been observed in the 2,7-di—
NOz-PSQ free ligand examination, where 1H”, > 1H”. Additionally, they observed that
2,7-di-t-butyl-PSQ had a more evenly distributed spin delocalization, with the [H13 and
1H345 HFC being close in magnitude. The mechanism used to explain the spin transfer
was an interaction between a filled orbital on the copper(I) metal center with the SOMO
of the radical. They also proposed that the charge density changes in a direction opposite

to the spin density. An identical situation is possible with the Ga—2,7-di-NOz-PSQ
complex (and 3,6-isomer). Since the molecule has C2v symmetry the Gam-3dx2 orbital

has the correct symmetry to interact with the 2,7-di—N02-PSQ SOMO orbital of b.

symmetry67 (Figure 4-18). The spin delocalization in the SOMO molecular orbital of the
metal complex is dependent upon the degree of orbital mixing between the Gam-3dXZ and

2,7-di-N02-PSQ SOMO. Recall from MO theory, that when two orbitals are more
energetically matched a stronger interaction occurs between the orbitals. Based on
Rockenbauer’s EPR data, the larger lHl,3 HFC for the 2,7-di-N02-PSQ radical could be
due to the fact that the N02 group is more electron withdrawing than t-butyl, so that and

the NOz-SOMO will be lower in energy than the 2,7-di-t-butyl-PSQ SOMO. The spin
will then localize more at the oxygen atoms due to the decreased mixing between the 3dxz

and 2,7-di-N02-PSQ SOMO compared to the higher energy 2,7-di-t-butyl-PSQ SOMO,

134

which will display increased mixing between the orbitals. The increased spin or
decreased spin delocalization from the oxygen toward the PSQ ring will then account for

the observed spin density changes in the carbon atoms.

Ga I L I” “\
3dxz ‘~ I“ 2,7-di-N02-PSQ
‘ ,/ SOMO

Figure 4—18. Interaction between the filled Ga(lll) 3dxz and the 2,7-di-N02-PSQ SOMO.

It is known that 2 is more electron withdrawing than 1, based on their
corresponding half-wave reduction potentials.68 The above mechanism then implies the
spin density should be more localized at the oxygen atoms in 2 resulting in less overall
spin density delocalization in the PSQ ring. This would predict larger MN HFC in 4 if
spin delocalization is the dominating mechanism, but this is not observed experimentally.
It has been hypothesized that the main mechanism for spin p0pulation at the nitrogen is
spin polarization in complex 4 and this is known to have a decreased effect at transferring
spin density compared to spin delocalization.58 It is unclear then if this mechanism can
be used to account for the different spin delocalization/polarization in the gallium
complexes compared to the free ligands. To determine why such a dramatic change in

spin population occurs for one isomer and not the other, a full DF T examination will be

135

undertaken to see if a model can be elucidated to explain the experimental observations in
better detail. Additionally, Chapter 5 of this thesis will. examine the mechanisms of spin

delocalization more intensively with a variety of substituted Ni-PSQ complexes.

4.3.4 Density Functional Theory of Gallium Phenanthrenesemiquinones
The gas phase DFT calculations were done using the UB3LYP, UB3PW91, and
the UPBElPBE functionals using the 6-31 g* basis set. Two key points regarding the

DF T calculations can be made by examining the results in Table 4-3.

Table 4—3. DFT calculated HFC’s and spin densities for [Ga(tren)(2,7-di-N02-PSQ)]+ (3) and
[Ga(tren)(3,6-di-N02-PSQ)]+ (4).

 

Compound Experimental Calculated s-orbital p-n b
HFC (G): HFC (G): p (H or N)“ p (C)
B3LYP B3PW91PBE1PBE

3
‘Hig -29723 -3.2629 -3.4804 0.00499 0.10837
”N2; 0.1944 0.2211 0.2791 0.00085 0.05038
1H3,6 -3.0670 -3.3630 -3.5590 0.00516 0.11193
‘an5 0.4392 0.5736 0.7721 0.000850 0.02545

4
‘ng -3.0741 0.3703 0.5994 0.00564 0.11538
‘HZ,7 1.0199 1.1820 1.3893 0.00176 0.0477
”1513,, 0.5681 0.5958 0.6610 0.00233 0.10276
1H4,5 0.6691 0.8087 1.0244 0.00121 0.03399

 

a F
Mulliken spin densities form the UB3LYP/6-31g* level of theory. Average total p-orbital spin density
on the carbon atoms bonded to the proton or nitrogen nucleus as indicated in the left column.

First, the DFT calculated l4N hyperfine coupling constant for 3 is 1 order of magnitude
too small, while 4 has an l4N HF C that is in excellent agreement with experiment. The
DFT calculated trend in the 14N HF C is not consistent with the experimental data with 4

exhibiting the largest l4N HFC. Secondly, the sign of the 14N HFC’s for the complexes

136

changes upon coordination (not determined in the EPR and ESEEM experiments) and
contrasts the free radical DFT data. Interestingly, the spin density at the 2,7- and 3,6-
carbon atoms in the gallium complexes exhibits opposite polarization, which also
contrasts the free ligand calculations. After the above discrepancies were discovered,
calculations were also done using the PCM and IPCM solvent continuum models in an
attempt to calculate 14N values closer to the experimental numbers. Unfortunately, the
results of the solvent calculations predicted I4N HFC’s that were smaller in magnitude
than the gas phase calculations.

A complete examination of the spin density distribution may possibly elaborate
upon the disagreement between the experimental and calculated l4N for 3, and this is

depicted in the total spin density plots for both complexes in Figure 4-19.

 

Figure 4—19. Total spin density distribution in 3 (left) and 4 (right).

In both gallium radical complexes, the polarization of spin density is identical, and the
large (it-spin density on the C—0 group of the PSQ radicals causes the alternating B-spin

and (it-spin density polarization within the n-structure in the ring. The DFT spin density

137

data can be interpreted to predict that strong spin delocalization is operative at the 3,6-
carbon atoms (large (it-spin density), while spin polarization at the 2,7-carbons is the
dominant mechanism (smaller B—spin density). This is not surprising since spin
polarization is known to cause negative spin densities in similar ruthenium o-
semiquinone type complexes.3 The difference in sign for the 14N in 3 and 4 is reflected in
the opposing spin polarization mechanisms due to the different carbon atom spin
populations. The Mulliken total pz-n-spin density populations for each fragment is

depicted in Figure 4-20.

Ga-3,6-di-N02-PSQ Ga-2,7-di-NOZ-PSQ

0.002 O -0.006 -0. 003 O O® 003

0\0
®/O -.0 003 O>Oz

0.021 0

Figure 4-20. The polarization of 1t-spin density is depicted as it is predicted by DFT calculations for the

Ga-2,7- and Ga-3,6-carbon atoms and N02 groups. In order for a simple comparison, the PSQ ligands are
represented by a benzene ring.

The polarization of pZ-n-spin density is represented by an upward arrow for (ll—spin or a
downward arrow for excess B-spin. It is evident in Figure 4-20 that the magnitude of spin
density at the 3,6—carbons is a factor of 2.5 times greater than in the 2,7-isomer, and the N
spin density is also 2 times greater (but negative). In all cases spin polarization is
predicted to place spin density on the nitrogen atoms according to the DFT calculations.
If spin delocalization were operative, the sign of the spin density on the carbon and

nitrogen atoms would be the same to indicate the delocalization.

138

Based on the experimental results an incorrect physical picture is then predicted
by the DFT calculations for complex 3 and no large excess (l-Spln or B-spin density is
delocalized onto the 2,7-carbon atoms. Caution must be used since the ESEEM
experiments have not determined the sign of the 14N HF C only the magnitude. Overall,
the DFT calculated spin density distribution supports the DFT calculated HF Cs and spin
densities, but the opposite trend is observed in the experimental ESEEM data.
Interestingly, the free ligand calculations of the spin density distribution in the two
radicals is consistent with the spin density being delocalized onto the nitrogen center of
the N02 group in both complexes. A comparison between the a-HOMO’s of 2 and 4 are

depicted in Figure 4-21 to further investigate the Ga-DFT data.

 

Figure 4-21. u-HOMO of 2 (left) and 4 (right).

The results shown in Figure 4-21 are for the 3,6-isomer, but identical conclusions can
also be drawn for the 2,7-analogue since the coordination effects yield the same general
results. There is a noticeable shifl in the radical wavefunction upon coordination to

gallium, with the wavefunction delocalizing away from the 2,7-positions in the gallium

139

complex. The depletion of the SOMO wavefunction away from the 2,7-carbons is one
cause for the lack of spin density at these centers. It is also recognized that there is no
large amplitude of the wavefunction on the nitrogen atoms in either Ga-complex. A
simple explanation for the change in the a-HOMO upon complexation for the DFT data
is that gallium(IlI) is a strong Lewis acid and it draws electron density away from the
PSQ ring toward the carbonyl groups. This general idea supports the experimental data
from Rockenbauer and the ”N ESEEM data in the gallium complexes, since a shift in
spin density toward the metal centers is observed. DFT also supports the role of Lewis
acidity in increasing wavefunction localization at the carbonyl carbon atoms. The
calculated carbon coefficient in the a-HOMO wavefunction more than doubles from 7 %
for 2 to 15.5 % carbon for 4, and there is a factor of 7 increase in the a-Spin density at the
carbonyl carbon atoms. Similar increases are also observed from 1 to 3. The large
localization of spin at the carbonyl carbons, causes the unpaired Spin to be polarized in
the alternating B-spin, ot—spin, pattern (Figure 4-19). Therefore, the Lewis acid effect
could be responsible for the DFT pattern of spin polarization in the gallium complexes,
and may be the reason why we do not see it in the free ligands. The disagreement
between the experimental and theoretical data likely stems from the DF T overestimation

"I Lewis acid.

of the spin polarizing effect of the Ga
The above arguments explain the trend in the DFT results, but do not explain why
there is an observed deviation from the experimental values. In the literature it is known

69’” and errors in A(Ga) as

that spin polarization effects are difficult to model accurately;
much as 22 G have been reported for the neutral gallium boraarnidinate radicals.71 Spin

contamination of the ground state wavefunction was also noted to overestimate spin

140

polarization due to large UHF contributions in hybrid functionals,3 but the (82); values
for the gallium radicals indicated negligible contamination ((852 = 0.75 for a pure spin
state), and the maximum value found with all hybrid filnctionals was ($2); = 0.77.
Functionals such as BLYP and SVWN, which have no Hartree-Fock exchange included
in their description, actually had larger deviations in magnitude from the experimental
l4N HF C and predicted the same spin density distribution. Increasing the size of the basis
set from 6-31g* to 6-31lg** for all atoms or using the cc-pVQZ basis set for N atoms
also did not show any improvement in the calculation of the 14N HF C. A survey of the
literature did find another possible explanation of the incorrect spin distribution, due to
large self interaction errors in DFT.72 Piquemal et al. have shown that self interaction
error is largely responsible for incorrect spin delocalization in tyrosinase models such as
[Cu202]2+ and ZnH-imidazole complexesn’74 The overpolarization of spin density in the
complexes is explained by the presence of a strong electric field generated by the
transition metals.”‘74 The gallium electric field could then possibly “overpolarize” the
spin density distribution in 3 and 4. To test this hypothesis a DFT calculation in the
presence of an electric dipole field applied along the y-axis (the y-axis goes through the
oxygen atoms in the PSQ plane) using the Gaussian 03 keyword field = y+200 was done,
and the results are depicted in Figure 4-22. The field value of y+200 indicates that a field
of 0.02 au is applied in the y-direction based upon the standard orientation of the
molecule. The use of the electric dipole field applied along the y-axis polarized the spin
density, and negative spin density is observed on the 2,7-carbon atoms identical to that
observed in complexes 3 and 4 (and K-coordinated complex, not shown). In fact, the

entire spin density distribution seen in 3 and 4 is modeled by using an electric dipole field

141

 

 

ill l’.$ :-

applied along the y-axis for the 2,7-di-N02-PSQ ligand. This attests to the fact that the

I" complexes might be overpolarized by the gallium

calculated spin density in the Ga
electric field, resulting in the absence of spin delocalization to the nitro group in complex
3. Higher level calculations are likely required to get an accurate description of the spin

density distribution in these systems, and will be pursed in the future.

 

Figure 4-22. The results of a UB3LYP/661g“ total spin density field calculation for 2,7-di-N02-PSQ
(lett) along with the result from the same calculation without the field keyword (right). The Gaussian 03
field = y+200 keyword was used in the field calculation. Identical spin polarization results were also

obtained for the 3,6-di-N02-PSQ complex with the same electric field (not shown).

4.4 Concluding Comments
Several salient results concerning the experimental and DFT spin density
distribution and hyperfine coupling constants in the 2,7—di-NOz-PSQ (l) and 3,6-di-N02-
PSQ (2) radical ligands and the corresponding Ga-2,7-di-N02-PSQ (3) and Ga-3,6-di-
NOz-PSQ (4) complexes have been realized. The experimental X-band EPR and DFT
calculated HFC and spin densities for the radical ligands are in excellent agreement. The
magnitudes of the MN HFC for the radical ligands are in the same range, and indicate the
spin density on each nitrogen atoms results from spin polarization. Upon coordination to

gallium, the l"’N HFC was found to be a factor of 4 times greater in 3 then in 4.

142

Although, the ESEEM simulations may have errors as great as 20 % for the 14N HF C, the
trend in the data is real. The increased HF C in 3 is due to direct spin delocalization onto
the nitrogen atom, and the smaller HF C for 4 is likely the result of spin polarization. The
experiments then indicate that substituents at the 2,7-positions will delocalize more spin
density away from the oxygen atoms. DFT calculated 14N HFC for 4 was consistent with
the value determined from ESEEM spectroscopy, but the MN for 3 was found to deviate
by a factor of 7. The DFT deviations could be the result of self-interaction errors
overpolarizing the spin density distribution in the gallium complexes, whereby spin
delocalization is masked. More work is needed to verify a model to describe the spin
delocalization/polarization mechanism both experimentally and theoretically. It must be
stressed that only the relative magnitudes of the 14N HFC have been determined, and the
signs of the HF C are needed to unambiguously experimentally determine the spin
delocalization model in these systems.

The above model for the shifting of spin density towards the gallium center is still
possibly valid, but DFT calculations overestimated the polarization of the spin density of
the n—system in complex 3. In order to determine a bulletproof experimental mechanism,
the determination of all the signs of the HFC is needed. This can be accomplished by
using a combination of EPR, ENDOR, and TRIPLE spectroscopies.75’76 The sign
determination is of course predicated upon observing the ENDOR and TRIPLE EPR
Signals, but if they can be determined then it can be deduced where 01- and B-spin density
resides in the complexes, enabling determination of a correct spin delocalization
mechanism. Another experimental Option to model this effect is 17O. This would be

useful to determine the shifts in magnitude of the spin density nearest to the gallium. ”N

143

labeling of the tren capping ligand would also be highly desired to inconclusively prove
the 14N HFC is from the dinitro radical ligands. Additionally, the effect of n-donor
ligands such as di—NHz-PSQ along with the n—acceptor di-NOz—PSQ radicals on the
magnitude of 14N and 170 HF C would give strong evidence for the support or failure of

the spin density shifting model. Future work along these lines will be undertaken.

144

4.5 References

1

2

10

ll

12

13

14

15

16,

17

18

19

Ruiz, E.; Cirera, J.; Alvarez, S. Coord. Chem. Rev. 2005, 249, 2649.

Cano, J .; Ruiz, E.; Alvarez, S.; Verdaguer, M. Comments. Inorg. Chem. 1998, 20,
27.

Reymenyi, C.; Kaupp, M. J. Am. Chem. Soc. 2005, 127, 11399.

Dei, A.; Gatteschi, D.; Sangregorio, C.; Sorace, L. Acc. Chem. Res. 2004, 37, 827.
Vostrikova, K. E. Coord. Chem. Rev. 2008, 252, 1409.

Lin, H.-H.; Wei, H.-H.; Lee, G.-H.; Wang, Y. Polyhedron 2001, 20, 3057.
Caneschi, A.; Gatteschi, D.; Sessoli, R.; Rey, P. Acc. Chem. Res. 1989, 22, 392.
Hicks, R. G. Aust. J. Chem. 2001, 54, 597.

Abakumov, G. A.; Cherkasov, V. K.; Nevodchikov, V. 1.; Druzhkov, N. 0.;
Kuropatov, V. A. Russ. Chem. Bull. 1998, 4 7, 875.

Oshio, H.; Yamamoto, M.; Ito, T.; Kawauchi, H.; Koga, N.; Ikoma, T.; Tero-
Kubota, S. Inorg. Chem. 2001, 40, 5518.

Broze, M.; Luz, 2.; Silver, B. L. J. Chem. Phys. 1967, 46, 4891.
Broze, M.; Luz, Z. J. Chem. Phys. 1969, 51, 738.

Wheeler, D. E.; Rodriguez, .1. H.; McCusker, J. K. J. Phys. Chem. A 1999, 103,
4101.

Rockenbauer, A.; Gyor, M.; Speier, G.; Tyeklar, Z. Inorg. Chem. 1987, 26, 3293.
McConnell, H., M, J. Chem. Phys. 1956, 24, 764.

Nguyen, M. T.; Creve, S.; Vanquickenbome, L. G. J. Phys. Chem. A 1997, 101,
3174.

Palmer, G. In Physical Methods in Bioinorganic Chemistry: Spectroscopy and
Magnetism; Que, J ., L., Ed.; University Science Books: Sausalito, 2000.

Deligiarmakis, Y.; Louloudi, M.; Hadjiliadis, N. Coord. Chem. Rev. 2000, 204, l.

McCracken, J. L. In Handbook of Electron Spin Resonance; Poole, J., C. P.,
Farach, H. A., Eds.; Springer-Verlag: New York, 1999; Vol. 2, p 69.

145

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

McCracken, J.; Vassiliev, I. R.; Yang, E.-C.; Range, K.; Barry, B. A. J. Phys.
Chem. B 2007, 111, 6586.

Jiang, F .; McCracken, J.; Peisach, .I. J. Am. Chem. Soc. 1990, 112, 9035.
Chasteen, N. D.; Snetsinger, P. A. In Physical Methods in Bioinorganic
Chemistry: Spectroscopy and Magnetism; University Science Books: Sausalito,
CA, 2000, p 187.

Schmidt, J.; Kampf, A. Chem. Ber. 1902, 35, 3117.

Schmidt, J .; Kampf, A. Chem. Ber. 1903, 36, 3738.

Urbanek, R. A.; Suchard, S. J .; Steelman, G. B.; Knappenberger, K. S.; Sygowski,
L. A.; Veale, C. A.; Chapdelaine, M. J. J. Med. Chem. 2001, 44, 1777.

Maruyama, K.; Otsuki, T. Bull. Chem. Soc. Japan 1968, 41, 444.

Maruyama, K. Bull. Chem. Soc. Japan 1964, 37, 553.

Schenck, G. 0.; Schmidt-Thombée, G. A. Ann. Chem. 1953, 584, 199.

Connolly, N. G.; Geiger, W. E. Chem. Rev. 1996, 96, 877.

Vierling, P.; Guillon, C. J. Organomet. Chem. 1994, 464, C42.

XSOPHE, Computer Simulation Software Suite 1.1.4; Center for Magnetic
Resonance and Department of Mathmatics, The University of Queensland
(Brisbane, Queensland, Australia 4072) and Bruker Biospin GMBH (formerly
Buruker Analytik; Rheinstetten, Germany): 1993-2004.

Wang, D.; Hanson, G. J. Mag. Reson. Ser. A 1995, 117, 1.

Hanson, G.; Gates, K. E.; Noble, C. J .; Griffen, M.; Mitchell, A.; Benson, S. J.
Inorg. Biochem. 2004, 98, 903.

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

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

Mims, W. B. Phys. Rev. B: Condens Matter 1972, 5, 2409.

Mims, W. B. Phys. Rev. B: Condens Matter 1972, 6, 3543.

146

38

39

40

41

42

43

44

45

46

47

48

49

50

Dikanov, S. A.; Tsvetkov, Y. D. In Electron Spin Echo Envelope Modulation
Spectroscopy; CRC Press: Boca Raton, F1, 1992, p 17.

O'Malley, P. J. Antiox. Redox. Sig. 2001, 3, 825.

Hermosilla, L.; Calle, P.; Garcia de la Vega, J. M.; Sieiro, C. J. Phys. Chem. A
2006, 110, 13600.

Langgard, M.; Spanget-Larsen, J. J. Mol. Struct. THEOCHEM 1998, 431, 173.
Gauss View, 2.1 ed.; Gaussian, Inc.: Pittsburgh, PA, 2001.

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

, Tomasi; V. Barone, B. M., M. Cossi, G. Scalmarri, N. Rega; G. A. Petersson, H.

N., M. Hada, M. Ehara, K. Toyota; R. Fukuda, J. H., M. Ishida, T. Nakajima, Y.
Honda, 0. Kitao; H. Nakai, M. K., X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross;
V. Bakken, C. A., J. Jaramillo, R. Gomperts, R. E. Stratrnann; O. Yazyev, A. J.
A., R. Carnmi, C. Pomelli, J. W. Ochterski; P. Y. Ayala, K. M., G. A. Voth, P.
Salvador, J. J. Dannenberg; V. G. Zakrzewski, S. D., A. D. Daniels, M. C. Strain;
0. Farkas, D. K. M., A. D. Rabuck, K. Raghavachari; J. B. Foresman, J. V. 0., Q.
Cui, A. G. Baboul, S. Clifford; J. Cioslowski, B. B. S., G. Liu, A. Liashenko, P.
Piskorz; I. Komaromi, R. L. M., D. J. Fox, T. Keith, M. A. Al-Laham; C. Y. Peng,
A. N., M. Challacombe, P. M. W. Gill; B. Johnson, W. C., M. W. Wong, C.
Gonzalez, and J. A. Pople, Gaussian 03, Revision D.01; Gaussian, Inc.:
Wallingford CT, 2004.

Glendening, E. D.; Reed, A. E.; Carpenter, J. E.; Weinhold, F. NBO 3.1,
Theoretical Chemistry Institute, University of Wisconsin: Madison, WI, 1996.

Foresman, J. B.; Keith, T. A.; Wiberg, K. B.; Snoonian, J .; Frisch, M. J. J. Phys.
Chem. 1996, 100, 16098.

Frisch, M. J .; F risch, A. Gaussian 98 User’s Reference; Gaussian, Inc.: Pittsburgh,
1998.

Schrauben, J. N.; Guo, D.; McCracken, J. L.; McCusker, J. K. Inorg. Chim. Acta.
2008, 361, 3539.

Canoes, M. T.; Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 107, 3032.

Cossi, M.; Barone, V.; Mennucci, B.; Tomasi, J. Chem. Phys. Lett. 1998, 286,
253.

Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 106, 5151.

147

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

Cossi, M.; Scalmani, G.; Rega, N.; Barone, V. J. Chem. Phys. 2002, 11 7, 43.
O'Malley, P. J .; Collins, S. J. Chem. Phys. Lett. 1996, 259, 296.
Rieger, P. H.; F raenkel, G. K. J. Chem. Phys. 1963, 39, 609.

Corrie, J. E. T.; Gilbert, B. C.; Munasinghe, V. R. N.; Whitwood, A. C. J. Chem.
Soc. Perkin Trans. 2 2000, 2483.

Morales-Morales, J. A.; Frontana, C.; Aguilar-Martinez, M.; Bautista—Martinez,
A.; Gonzalez, F. J.; Gonzalez, 1. J. Phys. Chem. A 2007, 111, 8993.

Kazakova, V. M.; Pesterev, P. A.; Minina, N. E. Rus. J. Gen. Chem. 2003, 73,
1114.

Dehl, R.; Fraenkel, G. K. J. Chem. Phys. 1963, 39, 1793.

Adamo, C.; Barone, V.; Subra, R. Theor. Chem. Acc. 2000, 104, 207.
Flanagan, H. L.; Singel, D. J. J. Chem. Phys. 1987, 87, 5606.

Mims, W. B.; Peisach, J. J. Chem. Phys. 1978, 69, 4921.

McCracken, J.; Peisach, J.; Cote, C. E.; McGuirl, M. A.; Dooley, D. M. J. Am.
Chem. Soc. 1992, 114, 3715.

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

McCracken, J.; Pember, S.; Benkovic, S. J.; Villafranca, J. J.; Miller, R. J.;
Peisach, J. J. Am. Chem. Soc. 1988, 110, 1069.

Lil, J. M.; Rosokha, S. V.; Neretin, I. S.; Kochi, J. K. J. Am. Chem. Soc. 2006,
128, 16708.

Lit, J. M.; Rosokha, S. V.; Lindeman, S. V.; Neretin, I. S.; Kochi, J. K. J. Am.
Chem. Soc. 2005, 127, 1797.

Hirota, N. J. Phys. Chem. 1967, 71 , 127.
Gordon, D. J .; Fenske, R. F. Inorg. Chem. 1982, 21, 2907.

Mukherjee, T. K. J. Phys. Chem. 1967, 71, 2277.

148

69

70

71

72

73

74

75

76

Munzarova, M. L. In Calculation of NMR and EPR parameters: Theory and
Applications; Kaupp, M., 811th, M., Malkin, V. G., Eds.; VCH, Weinheim: 2004,
p 463-482.

Kaupp, M.; Munzarova, M. L. J. Phys. Chem. A 1999, 103, 9966.

Chivers, T.; Eisler, D. J .; Fedorchuk, C.; Schatte, G.; Tuononen, H. M.; Boere', R.
T. Chem. Commun. 2005, 3930.

Zhang, Y.; Yang, W. J. Chem. Phys. 1998, 109, 2604.
Piquemal, J. P.; Pilme’, J. J. Mol. Struct. 2006, 764, 77.

Piquemal, J. P.; Marquez, A.; Parisel, 0.; Giessner-Prettre, C. J. Comp. Chem.
2005, 26, 1052.

Stegmann, H. B.; Dao-Ba, H.; Maurer, M.; Scheffler, K.; Buchner, H.; Hartmann,
E.; Mannschreck, A. Mag. Res. Chem. 1988, 26, 547.

Kay, C. W. M.; Feicht, R.; Schulz, K.; Sadewater, P.; Sancar, A.; Bacher, A.;
Mobius, K.; Richter, G.; Weber, S. Biochemistry 1999, 38, 16740.

149

Chapter 5. The Substituent Effect in the Modulation of the Heisenberg Exchange
Coupling Interaction in Ni(II) and Cr(III) Phenanthrenesemiquinones: A Combined

Experimental and Theoretical Examination.

5.1 Introduction.

The applications for manipulating exchange coupling in transition metals could be
applied to many disciplines from magnetism to biology. For example, if the mechanism
for spin density delocalization/polarization can be determined using both experimental
and theoretical tools, then advances in the design of transition metal-radical molecular
magnetic materials could be accomplished."6 Additionally, the control of spin density at
a radical center could be utilized in the design of catalysts that participate in o-bond
forming reactions. Thus it is a great importance to many scientists to elucidate the
mechanism of spin delocalization/polarization in transition metal radical complexes.

The previous results from the examination of the polarization of spin and charge
density in neutral and anion radicals have revealed that both spin and charge density can
be differentially polarized. This work will be extended to utilize the knowledge of spin
and charge polarization to design paramagnetic transition metal radical complexes that
could exhibit tunable ferromagnetic and antiferromagnetic Heisenberg exchange coupling
interactions. The charge density polarization information will aid in the synthesis of the
semiquinone complexes, since increasing the negative charge density at the oxygen
centers in the semiquinone (SQ) radicals will increase the Lewis basicity of these ligands
and the formation constant.7 The strength of the Heisenberg exchange coupling

interaction has been shown to be proportional to the spin densities at the interacting

150

atoms,8'12 and control of the spin density polarization with substituents can then be used
to increase or decrease the strength of the metal-radical exchange coupling interaction.
Previous work from Shultz et al.13 has shown that substituents can modulate the exchange
coupling interactions in Mn(II) and Cu(II) 5-aryl-substituted semiquinones, but no
mechanisms concerning spin density distribution have been put forth. It is the goal of
this work to examine the effects of substituents on the spin density distribution, and
determine any mechanisms that account for the modulations in the magnitude of
ferromagnetic and antiferromagnetic Heisenberg exchange coupling constants.

It is well known that the unpaired electron in the SQ radical is housed in an orbital
of rt-symmetry.l4 In order to examine the effects of substituents on ferromagnetic
exchange coupling nickel(II) will be used, and in order to simplify the exchange coupling
interaction we will focus on six coordinate complexes that contain only one semiquinone

radical. Six coordinate nicke1(II) is a d8 transition metal and its unpaired electrons are in
the dx2-Y2 and dz2 orbitals which have o—symmetry. The oeorbitals of NiII and the 7t* SQ

orbital are strictly orthogonal, and Hund’s rule will apply and the electrons will
experience a ferromagnetic interactionw’16 Conversely, chromium(IlI) is a d3 transition
metal having unpaired electrons in the dxz, dyz, and dxy orbitals of n-symmetry. Due to
the n-orbital overlap, Cr(III) will undergo an antiferromagnetic interaction with the SQ

1,1546 Nickelal)m8 and Chromium(III)19 semiquinone compounds with the

radica
tetradentate CTH counterligand (CTH = dl-5,7,7,12,14,14-hexamethyl-1,4,8,1l-
tetrazacyclotetradecane) are experimental examples of the ferromagnetic and

antiferromagnetic exchange coupling interaction found in these types of transition metal

radical complexes. It must be noted that no temperature dependence was seen in the

151

experimental magnetic susceptibility data for either the Cr(lll) and Ni(II) complexes,
barring any firm determination of J and only allowing a lower limit for J to be
established. For experimental J values to be determined, thermal population of the
doublet state is required to determine the energy gap and a lack of a temperature
dependence in the Ni(II) complex indicates an isolated quartet ground state with no
thermal population of the antiferromagnetic doublet state. Based on the magnetic data,
experimental estimates for the exchange coupling constants in these systems have been in
excess of 400 cm’1 in both cases.

The broken symmetry formalismzo'23 has been applied to calculate J in several

24-28 -

transition metal organic radical complexes including semiquinone compounds.”33 A

recent report from Bencini et al.“ for the [N i(CTH)(PSQ)]2+ and [Ni(CTH)(3,5-
DTBSQ)]2+ (PSQ = 9,IO-phenanthrenesemiquinone and 3,5-DTBSQ = 3,5-di-tert-
butylsemiquinone) complexes has calculated J = 285 and 356 cm}, respectively, from
broken symmetry density functional theory (DFT). For our purposes we wish to be able
to experimentally determine J, and in order to do so we need J values that will be small
enough to allow for temperature dependence in the magnetic susceptibility
measurements. Therefore, for these reasons we will prepare Nickel(II) and
Chromium(III) complexes of substituted PSQ’s, since the Bencini DFT calculations
predicted that the PSQ ligand should lead to decreased exchange coupling.

For the transition metal complexes studied here, the tetradentate tren (tren = tris-
2-aminoethylamine) ligand will be used as the counter ligand, since only one isomer can

be formed when bound to a metal (the CTH ligand can exist in both meso and racemic

forms). The [M(tren)(2,7—R-PSQ)]"+ and [M(tren)(3,6-R-PSQ)]n+ (where M = Cr(III)

152

 

 

 

 

.‘l 4.-

and Ni(II), 2,7-R1 = N02, H, and NH; and 3,6—R2 = N02, H, and NH; , and for Cr(III)
n = 2 and for Ni(II) n = 1) depicted in Chart 5-1 will be examined using a combination of
DFT and magnetic susceptibility measurements in an effort to determine the effects of
substituents on the spin density distribution and on the corresponding J values. The 2,7-
R-PSQ ligands are easiest to prepare synthetically and the free ligands showed variable

spin polarization at the oxygen atoms, thus this will be the starting point of this work.

 

 

 

Chart 5-1. General formula for the metal-radical complexes examined.

5.2 Experimental.

General Methods. All reagents were obtained from commercial sources and
used without further purification. HPLC grade methanol was distilled from Mg and 12,
and degassed 15 times by gas exchange with N2. Elemental analyses and mass spectra
were obtained through the analytical facilities at Michigan State University, ESI-MS

experiments used standard conditions and all spectra agreed with appropriate simulations.

153

34

The 3,6-dinitrophenanthrenequinone, 3,6t-dinitrophenanthrenecatechol,34 2,7-

34 2,7--dinitrophenanthrenecatechol,34 acetyl-

dinitrophenanthrenesemiquinone,
ferroceniurn tetraphenylborate,34 and [Cr(tren)Cl2]Cl35 compounds have been prepared as
previously described.

Phenanthrenecatechol. The procedure for the preparation of this ligand has been
modified from a published procedure.36 To a schlenk flask containing 2.5 g (12 mmol) of
phenanthrenequinone in 125 ml of degassed ethanol, 2.6 g (48 mmol) of solid KBH4 was
added over a period of three hours. The addition of KBH4 to the reaction mixture under
N2 was done every 20 minutes, and upon the addition of KBH4 the orange suspension
gradually goes into solution and the color changes to yellow. The solution was stirred
under nitrogen for 48 hours, then cooled in an ice bath. Dilute aqueous HCl was then
added to deactivate the excess potassium borohydride. When the deactivation with HCI
was complete (no H2 (g) evolution), the ethanol solution was poured over 500 ml of an
ice-water mixture and a white precipitate was immediately formed. The product was
collected on a glass frit washed with water and dried in vacuo. The ligand is stored under
nitrogen to prevent oxidation, and solutions of the compound are very air sensitive.
Yield: 2.0 g (80 %) Anal. Calcd for C14H1002-0.5H2O: C, 76.70; H, 5.06; Found: C,
77.05; H, 4.57.

2,7-diaminophenanthrenecatechol. The synthesis of this compound was based

on methods already developed in the literature.37

A flask was charged with 0.36 g (1.2
mmol) of 2,7-dinitr0phenanthrenequinone and 0.85 g (7.1 mmol) Sn0 and then 100 ml of
12 M HCl was added to the flask, and the solution was refluxed until it turns clear and

becomes yellow in color. Not all the solids dissolved, so approximately 20 ml of distilled

154

water was added to the reaction flask to help dissolve all remaining solids. The reaction
mixture is then filtered hot and upon slow cooling white crystals precipitate out of
solution. The complex was recrystallized by boiling in a 12 M HCl solution and allowing
the flask to slowly cool to room temperature. The white crystals were collected on a
glass frit and dried in vacuo. Yield: 0.27 g (92 %). The complex is stored under nitrogen
to prevent oxidation and solutions of the compound are very air sensitive. Anal. Calcd
for C14H3N206-0.5H20'2HC1: C, 52.19; H, 4.69; N, 8.69; Found: C, 52.09; H, 5.09; N,
8.67.

3,6-diaminophenanthrenecatechol. The synthesis of this compound was based
on methods already developed in the literature and is similar to the 2,7-di-NH2-isomer
above.37 In a roundbottom flask 0.25 g (0.82 mmol) 3,6-dinitrophenanthrenequinone and
0.59 g Sn0 (5.0 mmol) and 150 ml of 12 M HCl was refluxed for 45 minutes or until the
solution turns yellow and slightly cloudy. 20 ml of water was then added to the hot
solution to aid in the solubility of the product. The reaction mixture is then filtered hot
and left out to crystallize. The flask was then placed in the fridge overnight to ensure all
the precipitate fell out of solution. The precipitate was washed with cold ACN and ether
and dried in vacuo. Yield: yellow-white crystalline solid 0.17 g (84 %). The complex is
stored under nitrogen to prevent oxidation, and solutions of the compound are very air
sensitive. Anal. Calcd for C14H12N2O2~0.5H2O'2HC12 C, 52.19; H, 4.69; N, 8.69; Found:
C, 52.48; H, 4.27; N, 8.71.

Ferrocenium Tetraphenylborate. The synthesis of this compound was based on
methods already developed in the literature.38 To a stirring solution of ferrocene 0.50 g

(2.7 mmol) in a solution of 20 ml of water and 20 ml of acetone was added 0.61 g

155

anhydrous F eC13 (3.8 mmol) under N2. After stirring for 15 minutes the deep blue
solution was filtered and 1.3 g (3.8 mmol) of NaBPh4 was added, resulting in the
formation of a fine dark blue precipitate. Tetrahydrofuran (ca. 100 ml) was then added to
the reaction flask until all solids dissolved. The solution was filtered and concentrated
under a stream of nitrogen and deep blue crystals fell out of solution. The crystals were
washed with acetone and ether, and the complex is immediately stored in a dry box. The
sample should be used the same day it is synthesized, since it is unstable even under an
atmosphere of dry nitrogen. Anal. Calcd for C34H30BFe-2OC(CH3)2: C, 77.31; H, 6.81;
Found: C, 77.38; H, 6.53.

Ferrocenium Chloride. Although a literature procedure using C12 (g) as a
reagent for this complex has been reported,39 a new synthetic procedure based upon the
synthesis of a similar ferrocenium salt was adapted in order to exclude the use of the
hazardous C12 (g).40 4 ml of 2.0 M hydrogen chloride in ether was added to a to a
degassed filtered solution of 0.33 g (3.0 mmol) 1,4—quinone under nitrogen. Then 0.37 g
ferrocene (2.0 mmol) was dissolved in ether and filtered, and added to the stirring
solution under N2. The solution darkens in color and was stirred for 15 minutes under N2,
during which time a blue-green precipitate falls out of solution. The precipitate was
filtered off and washed with ether (30 ml x 3) and dried in vacuo. Yield: 0.20 g (44 %).

Anal. Calcd for C toHloFeCl'OSHCl: C, 51.68; H, 4.47; Found: C, 51.49; H, 4.19.
TlBArf. Caution: thallium salts are known to be highly toxic and care should be

taken during all manipulations using any thallium reagent. The synthesis of this

complex has been published}1 but a much less synthetically intensive procedure based on

the silver analogue42 was found to give similar results. 0.40 g of NaBArf (0.45 mmol)

156

was dissolved in 15 ml of ether then filtered into a separatory funnel. 0.24 g (0.90 mmol)
of TIN03 was dissolved in 10 ml of water and added to the funnel. The funnel was then
shaken for 5 minutes (periodically releasing the pressure), and then the layers were
allowed to separate. The ether layer was evaporated to dryness with a nitrogen stream,
and then the complex was dried in vacuo. The complex was then washed with water (3 x
30 ml) and hexanes (3 x 30 ml), and further dried in vacuo. Yield: 0.37 g (77 %). Anal.
Calcd for C32H12T1BF24-0.5CH3CH2OCH3CH2: C, 36.97; H, 1.55; Found: C, 36.75; H,
1.37.

Ni(tren)Cl2. The synthesis of this compound was based on methods already
developed in the literature.43 1.46 g (10.0 mmol) of tren (tris-2—aminoethy1amine) was
dissolved in ethanol and 2.38 g (10.0 mmol) of NiCl2-6H2O was added to the stirring
solution. The solution was stirred for one hour during which time a blue precipitate
forms. The precipitate is filtered off and washed with ethanol, and then the product was
recrystallized from hot ethanol to give dark blue crystals. The complex is stored in a
desiccator since its highly hydros00pic. Recrystallized yield: 0.963 g (35 %) Anal. Calcd
for C6H13C12N4Ni-H2O: C, 24.53; H, 6.86; N, 19.07; Found: C, 25.37; H, 6.79; N, 18.71.

[Ni(tren)(PSQ)](BPh4) (1). The synthesis of this compound was based on
methods already developed in the literature.3| Under N2, 0.24 g (0.86 mmol) of
Ni(tren)Cl2 was dissolved in 20 ml methanol, and to this solution was added a 20 ml
methanol solution of 0.20 g (0.96 mmol) phenanthrenecatechol and 0.076 g (1.9 mmol)
NaOH. The gold colored reaction mixture was stirred for 30 minutes and then oxidized
with 0.32 g (0.96 mmol) of [FeCp2](PF6) under N2, and the color changed to brown-

black. The solution was stirred for an additional 15 minutes after adding the oxidant, and

157

was then filtered under nitrogen. A concentrated methanol solution of 3.3 g (9.6 mmol)
NaBPh4 was slowly added to the filtered reaction mixture and it was left out to crystallize
in the dry box. A gold-brown microcrystalline product was recovered and 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 under N2. Crystals suitable for X-ray diffraction were obtained by carefully
layering a NaBPh4/methanol solution under the filtered reaction mixture in the dry box.
Yield: 0.42 g (67 %). Anal. Calcd for C44H47BN4NiO2: C, 72.16; H, 6.33; N, 7.65;
Found: C, 71.96; H, 6.72; N, 7.72. MS [ESL m/z (rel. int.)]: [Ni(tren)(PSQ)]+ 412.1 (80).
[Ni(tren)(2,7-di-NO2-PSQ)l(BPh4) (2). Under a N2 atmosphere, 0.083 g (0.30
mmol) Ni(tren)Cl2 was dissolved in 30 ml of methanol, and to this solution was added a
30 ml methanol solution of 0.10 g (0.33 mmol) 2,7-dinitrophenanthrenecatechol and
0.027 g (0.68 mmol) NaOH. The solution was stirred for 1 hour and then oxidized with
0.099 g (0.30 mmol) [FeCp2](PF6) and the solution turns red-black in color and was
stirred for an additional 15 minutes. The reaction mixture is then filtered under nitrogen
and a concentrated methanol solution of 1.0 g (3.0 mmol) NaBPh4 was slowly added. A
red-brown precipitate was recovered and washed with methanol (10 ml x 3). Yield: 0.16
g (64 %). The product is recrystallized from hot acetonitrile with a 41 % yield. X-ray
quality crystals were grown by slow evaporation of concentrated ACN solutions. Anal.
Calcd for C44H44BN6Ni06: C, 64.26; H, 5.39; N, 10.22; Found: C, 64.15; H, 5.74; N,
10.46. MS [ESL m/z (rel. int.)]: [Ni(tren)(2,7-di-NO2-PSQ)]+ 502 (100).
[Ni(tren)(2,7-di-N02-PSQ)](BArf) (3). This salt was prepared by first making

the [N i(tren)(2,7-di-NO2-PSQ)]C1 intermediate. The nickel chloride salt was synthesized

158

under N2 by dissolving 0.058 g (0.21 mmol) Ni(tren)Cl2 in 30 ml of methanol, and to this
solution was added a 20 ml methanol solution of 0.070 g (0.23 mmol) 2,7-
dinitrophenanthrenecatechol and 0.027 g (0.47 mmol) NaOH. The dark-blue solution
was stirred for 1 hour and then oxidized with 0.052 g (0.24 mmol) of [FeCp2]Cl, and the
solution turns black in color and was stirred for an additional 15 minutes. The reaction
mixture was then filtered, and the filtrate was evaporated to dryness in the dry box. The
recovered dark-brown/black precipitate was washed with water (30 ml x 3), isopropanol
(30 ml x 3), and ether (30 ml x 3) and dried in vacuo. Additional washing with CH2C12
(5 ml x 3) was done under N2 in the drybox. Yield: 0.096 g (84 %). This complex was
then used without further purification.

Under N2, 0.0670 g (0.124 mmol) of [Ni(tren)(2,7-di-NO2-PSQ)]CI was
suspended in 20 m1 CH2Cl2, and 0.133 g (0.124 mmol) of TlBArf in 10 ml of CH2C12 was
added with stirring. Care was taken to ensure that equal molar amounts of each reactant
are used. Upon the addition of the TlBArf the suspension goes into solution with the
formation of a white precipitate (TlCl). The solution was stirred for 1 day under
nitrogen, and then filtered through celite under N2. The filtrate was evaporated to dryness
in the drybox yielding a brown precipitate, and dried in vacuo. Yield: 0.142 g (83 %).
Anal. Calcd for C52H36BF24N6Ni06: C, 45.71; H, 2.66; N, 6.15; Found: C, 45.61; H, 2.87;
N, 5.40.

[Ni(tren)(2,7-di-NH2-PSQ)](BPh4) (4). Under a N2 atmosphere, 0.103 g (0.375
mmol) of Ni(tren)Cl2 was dissolved in 20 ml of methanol, and to this solution was added
a 10 ml methanol solution of 0.100 g (0.416 mmol) of 2,7-diaminophenanthrenecathechol

and 0.0600 g (1.50 mmol) NaOH. Excess base is required due to the presence of 2

159

equivalents of HCl in the elemental analysis of the ligand. The golden-yellow solution
was stirred for 30 minutes, then the solution was oxidized with 0.124 g (0.375 mmol) of
[FeCp2](PF6), which caused the color to change to dark-brown. The solution was stirred
for an additional 15 minutes and was filtered in the dry box. A saturated methanol
solution of 1.28 g (3.75 mmol) of NaBPh4 was layered under the reaction mixture to give
a brown crystalline solid. The product was washed with methanol (5 ml x 3) and ether
(10 ml x 3) and dried in vacuo. Yield: 0.154 g (54 %). X-ray quality crystals were
grown by layering a NaBPha/methanol solution under the reaction mixture in the dry box.
Anal. Calcd for C44HagBN6NiO2-MeOH: C, 68.03; H, 6.60; N, 10.58; Found: C, 68.18; H,
6.65; N, 10.84. MS [ESL m/z (rel. int.)]: [Ni(tren)(2,7-di-NH2-PSQ)]+ 442.1 (100).
[Ni(tren)(3,6-di—NO2-PSQ)](BPh4) (5). Under a N2 atmosphere, 0.054 g (0.19
mmol) Ni(tren)Cl2 was dissolved in 30 ml of methanol, and to this solution was added a
30 ml methanol solution of 0.070 g (0.23 mmol) 3,6-dinitrophenanthrenecatechol and
0.027 g (0.47 mmol) NaOH. The dark purple-red solution was stirred for 1 hour and then
oxidized with 0.064 g (0.23 mmol) [FeCp2](BF4) and the solution turns blue-black in
color and was stirred for an additional 15 minutes. The mixture is then filtered and a
concentrated methanol solution of 1.2 g (3.5 mmol) NaBPh4 was slowly added. Upon
sitting under nitrogen, deep-blue needles precipitate out of solution. The product was
then collected on a glass frit and washed with methanol (10 ml x 3) and ether (10 ml x 3)
and dried in vacuo. The product can be recrystallized by adding ether to a concentrated
acetonitrile solution of the complex. X-ray quality crystals were obtained by slow ether

diffusion into a ACN solution of the complex. Yield: 0.069 g (43 %). Anal. Calcd for

160

CMHMBNsNi06-05MeOH: C, 63.75; H, 5.53; N, 10.02; Found: C, 63.59; H, 5.49; N,
10.18. MS [ESL m/z (rel. int.)]: [Ni(tren)(3,6-di-NO2-PSQ)]+ 502.1 (100).
[Ni(tren)(3,6-di-NH2-PSQ)](BPha) (6). Under a N2 atmosphere, 0.066 g (0.24
mmol) of Ni(tren)Cl2 was dissolved in 10 ml of methanol, and to this solution was added
a 10 ml methanol solution of 0.063 g (0.26 mmol) of 3,6-diaminophenanthrenecathechol
and 0.042 g (1.1 mmol) NaOH. Excess base is required due to the presence of 2
equivalents of HCl in the elemental analysis of the ligand. The golden-yellow solution
was stirred for 30 min, then the solution was oxidized with 0.079 g (0.24 mmol) of
[FeCp2](PF6), which caused the color to change to deep red-black. The solution was
stirred for an additional 15 minutes and was filtered in the dry box. A saturated methanol
solution of 1.35 g (3.9 mmol) of NaBPh4 was slowly added to the reaction mixture and
upon standing overnight red-black crystals were formed. The product was washed with
methanol (10 ml x 3) and ether (10 ml x 3) and dried in vacuo. Yield: 0.055 g (32 %).
X—ray quality crystals were grown by layering a NaBPh4/methanol solution under a
concentrated solution of the reaction mixture in the dry box. Anal. Calcd for
C44PI433N6NiO2-H2O: C, 67.72; H, 6.46; N, 10.77; Found: C, 67.54; H, 6.51; N, 10.78.
MS [ESL m/z (rel. int.)]: [Ni(tren)(3,6-di-NH2-PSQ)-H]+ 443.2 (100).
[Cr(tren)(PCAT)](BPh4) (7). A schlenk flask was charged with 0.13 g (0.43
mmol) of [Cr(tren)Cl2]Cl that was dissolved in 10 ml of water and 50 ml of methanol,
then this solution was degassed for 30 minutes by bubbling N2 through the stirring
solution. The flask was connected to a reflux condenser and placed under N2, then a
oxygen free solution of 0.10 g (0.48 mmol) of phenanthrenecatechol and 0.038 g (0.95

mmol) NaOH in 20 ml methanol was added to the [Cr(tren)Cl2]Cl solution using a

161

cannula. The solution was then refluxed for 24 hours, and filtered under N2 into a flask
containing 1.6 g (4.7 mmol) of NaBPha. As the solution cools, a brown microcrystalline
product forms and was collected on a glass frit and washed with water (20 ml x 3) and
dried in vacuo.

A second method of preparation involves adding the deprotonated ligand to a
stirring suspension of [Cr(tren)Cl2]Cl in methanol. This suspension is then stirred for 24
hours in the dry box, and 1.2 equivalents of [F eCp2](BPh4) was added. After stirring for
2 hours the solution is filtered and a 10 fold excess of NaBPh4 is added to precipitate out
brown crystals. X-ray quality crystals were obtained by slow evaporation of the reaction
mixture in the dry box. Yield: 0.16 g (51 %). Anal. Calcd for C44H46BN4CrO2: C, 72.82;
H, 6.39; N, 7.72; Found: C, 72.27; H, 6.53; N, 7.76. MS [ESL m/z (rel. int.)]:

[Cr(tren)(PCAT)]+ 406.1 (100).

[Cr(tren)(PSQ)](BPh4)2 (8). Utilizing the same reaction conditions for the
second method of preparation of the [Cr(tren)(PCAT)]+ complex, except for stirring for
48 hours, the reaction mixture was oxidized with 0.22 g (0.43 mmol) of freshly prepared
[FeCp2](BPh4). The reaction was stirred for 2 hours with the oxidant and then filtered,
and a concentrated methanol solution of 0.92 g (2.7 mmol) of NaBPha was added and
black crystals fell out of solution. X-ray quality crystals were obtained by slow
evaporation of the reaction mixture in the dry box. Yield 0.13 g (43 %) Anal. Calcd for
C44H46BN4CrO2'MeOH: C, 76.95; H, 6.55; N, 5.20; Found: C, 76.57; H, 6.85; N, 5.48.
MS [ESL m/z (rel. int.)]: {[Cr(tren)(PSQ)]2+-CH3CN} 244.1 (100).

[Cr(tren)(3,6-di-NO2-PCAT)](BFa) (9). In a schlenk flask fitted with a reflux

condenser, 0.059 g (0.19 mmol) of [Cr(tren)Cl2]Cl was dissolved in a mixture of 30 m1

162

H20 and 45 ml of methanol, and this solution was degassed by bubbling N2 through the
stirring solution for 30 minutes. Under N2, 0.070 g (0.23 mmol) of 3,6-di-NO2-PCATH2
was dissolved in 10 ml of methanol and deprotonated with 0.019 g of NaOH (0.47 mmol)
in 10 ml of methanol. The deprotonated catechol ligand was then added via cannula to
the stirring [Cr(tren)Cl2]Cl solution under N2. The mixture is then refluxed for 16 hours,
and filtered hot under N2 into a Schlenk flask containing 0.38 g (15 fold molar excess) of
NaBF4 in a minimum amount of water. The solution was then allowed to cool to room
temperature and concentrated until a purple-brown precipitate forms. The precipitate is
then filtered off and washed with H20 (20 ml x 3) and ether (20 ml x 3). The complex
can be recrystallized from ACN/ether mixtures. Yield: 0.055 g (48 %). An analogous
procedure can also be used for the BPh4 salt (56 % Yield). Anal. Calcd for
C44H44BN6Cr06~2H2O: C, 62.05; H, 5.68; N, 9.87; Found: C, 61.8; H, 5.51; N, 9.99. MS
[ESL m/z (rel. int.)]: [Cr(tren)(3,6—di-NO2-PCAT)]+ 496.2 (20).
[Cr(tren)(3,6-di-NO2-PSQ)](BF4)2 (10). 0.031 g (0.047 mmol) of [Cr(tren)(3,6-
di-NO2-PCAT)](BF4) was dissolved in 5 ml of acetonitrile and 0.015 g (0.047 mmol) of
acetylferrocenium tetrafluoroborate in 2 m1 of acetonitrile was added to a stirring solution
of the catechol, and the color immediately changes from purple to dark brown. The
solution was stirred for 5 minutes then filtered, and ether was added until the solution
turned cloudy. A brown precipitate was recovered and washed with dichloromethane (5
ml x 3) and ether (5 ml x 3). Yield: 0.028 g (78 %). Anal. Calcd for
C20H24B2F3N6Cr06°H2O: C, 34.91; H, 3.81; N, 12.21; Found: C, 34.84; H, 3.84; N, 12.39.

MS [ESL m/z (rel. int.)]: [Cr(tren)(3,6-di-NO2-PSQ)]+ 496 (100).

163

[Ga(tren)(PCAT)](BPh4) (11). Under a dry atmosphere of N2, a methanol
solution of 0.10 g (0.48 mmol) phenanthrenecatechol and 0.04 g (0.95 mmol) of NaOH
was added to a stirring methanol solution of 0.19 g (0.43 mmol) of Gal3, and the cloudy
yellow solution was stirred for 1 hour. A solution of 0.060 g (0.43 mmol) of tren in 5 ml
of methanol was then added resulting in a clear yellow solution, and the reaction was
stirred under N2 overnight. The solution was filtered and a concentrated methanol
solution of 1.6 g (4.7 mmol) NaBPh4 was added, and upon standing a yellow
microcrystalline precipitate was recovered. X-ray quality crystals were grown from slow
evaporation of methanol solutions of the complex. Yield: 0.19 g (59 %). Anal. Calcd for
C44H46BGaN4O2: C, 71.09; H, 6.24; N, 7.54; Found: C, 70.61; H, 6.38; N, 7.48.

[Ga(tren)(PSQ)](BPh4)2 (12). To a concentrated stirring solution of
[Ga(tren)(PCAT)](BPh4) in acetonitrile, 1.1 equivalents of [FeCp2](BF4) in 2 ml of
acetonitrile was added changing the yellow color to brown. The solution was filtered and
added to 15 equivalents of NaBPha (no solvent) and then 300 ml of ether was added to
precipitate out brown microcrystals. Anal. Calcd for C63H66B2GaNaO2-MeOH: C, 75.71;
H, 6.45; N, 5.12; Found: c, 75.60; H, 6.29; N, 5.45.

Physical Measurements: X-ray Crystallography. Single-crystal X-ray
diffraction for l, 2, 4, 5, 6, 7, 8, and 11 were acquired at the X-ray facility at Michigan
State University. A total of four single crystals of 4 were used to collect the X-ray data,
and only one data set could be solved to yield a reasonable R1 value. Diffraction data
were collected on a Siemens SMART diffractometer with graphite-monochromatic Mo
K01 radiation (A = 0.71073 A). Data were collected at —100 °C by using an Oxford

Cryosystems low temperature device. Lattice parameters were obtained from least-

164

squares analyses and data were integrated with the program SAINT.44 The integration
method employed a three-dimensional profiling algorithm and all data were corrected for
Lorentz and polarization factors, as well as for crystal decay effects. The absorption
correction program SADABS was employed to correct the data for absorption effects.45
The structures were solved by direct methods and expanded Fourier techniques. All
structure calculations were performed with the SHELXTL 6.12 software package.46
Anisotropic thermal parameters were refined for all non-hydrogen atoms. Hydrogen
atoms were localized in their calculation positions and refined by using the riding model.
Further details concerning the structure determinations may be found in the text.

Cyclic Voltammetry. Electrochemical measurements were carried out in a Ar
filled dry box using a BAS CV-SOW electrochemical analyzer. A standard three-
electrode configuration was employed consisting of a Pt working electrode, a graphite
counter electrode, and Ag/AgNO3 reference electrode. Compounds were dissolved in
CH3CN that was 0.10 M in NBuaPF6; the CH3CN has been run through a dry still and
degassed by the freeze-pump-thaw method. Data were corrected for ferrocene as an
internal standard, and run at a scan rate of 100 mV-s".

Electronic Absorption Spectroscopy. Extinction coefficients were determined
using a Hewlett-Packard 8452A diode-array spectrometer in CH3CN. Data were obtained
on samples dissolved in CH3CN that had been run through a dry still and oxygen gas was
removed by gas exchange with nitrogen (15 cycles), and stored under an inert
atmosphere. All solutions were prepared under an inert atmosphere in 1 cm pathlength

sealed Optical cells.

165

Magnetic Susceptibility. Solution susceptibility was estimated based on the
chemical shifts of t-butanol using variable temperature Evans NMR.47 In the N2
atmosphere dry box, the paramagnetic sample was dissolved in d7-DMF containing 2-3 %
t-butanol and placed into a melting point tube. The open end of the tube was sealed with
epoxy, and then placed into a standard NMR tube containing d7-DMF and 2-3 % t-
butanol. Typical paramagnetic solution concentration’s were 3.0 x 10’2 M and the
temperature of the 500 MHz NMR spectrometer was calibrated with standard techniques
using CH3OH. The solution susceptibility of a known compound with an S = 3’/2 ground
state, Cr(acac)3 (acac = acetylacetonate), was run prior to the Evans to ensure the
accuracy of the experiments.

Magnetic susceptibility data were also collected using a Quantum Design MPMS
SQUID magnetometer interfaced with a IBM PC. The solid samples were ground with a
mortar and pestle and placed in a plastic freezer bag and rolled into a drinking straw. The
bag was placed 50 mm from the bottom of the straw, and the size of the sample in the
straw was less than 10 mm in height to ensure the samples proper placement in the
SQUID coils. NOTE: if the height of the sample in the straw was greater than 10 m,
we experienced reduced magnetization due to the sample’s position outside the SQUID
measurement range. Data were typically collected in an applied field of 1 T in a
temperature range of 5-350 K. Following each temperature change, the system was kept
at the new temperature for an additional 5 minutes before data collection to ensure
thermal equilibrium of the sample. Data were corrected for diamagnetism of the sample
using Pascal’s constants and for the measured susceptibility of the sample holder and bag

and are reported herein as effective magnetic moments. Solution SQUID susceptibility

166

4 49
measurements 8’

were collected in a laboratory fabricated quartz sample tube (Chart 5-
2) with a 5 mm ID and 7 mm OD that was equipped with a Pyrex vacuum adapter. The
samme holder had a quartz plug 50 mm from the bottom of the tube, so that sample could
be placed into the tube. An additional 70 mm (or more) of quartz tube was above the
plug to ensure that the environment above and below the sample was identical. To
collect good liquid phase data the sample needed to be as concentrated as possible
without powdering out of solution when frozen. Again, the height of the frozen solution
was kept below 10 mm to ensure the correct SQUID signal measurement. Typically
0.075 ml of an approximately 1.0 M solution (for S = 3/2 systems) was used in the
experiment, and the diamagnetism of the solvent and sample holder were corrected for by
subtracting their magnetic susceptibility determined in a separate experiment. The liquid

sample was then degassed by the freeze-pump-thaw technique at least 5 times and flame

sealed under vacuum.

Screw Cap Vacuum Adapter

in

T : Flame Seal

 

Flame Sealed Tube

70 mm or
longer

 

Sample

 

l H Quartz Plug

T
50 mm

1L. -

Chart 5-2. Liquid phase SQUID sample holder.

 

I67

Calculations. Previous calculations have shown that the B3LYP functional gives

reasonable results in the calculation of J values.“

The starting geometries for all
complexes were adapted from the crystal structures of either the [Ni(tren)(PSQ)]+ or
[Cr(tren)(PSQ)]2+ complexes using GaussView,50 and the quartet spin state of the Ni-PSQ
complexes and the triplet spin state of Cr-PSQ analogues were then optimized using
either Gaussian 985 I or 0352 with the UB3LYP functional with the LANL2DZ basis set.
The calculated energy differences between the Gaussian 98 and 03 results were
indistinguishable (identical J values resulted), as tested for the [Ni(tren)(PSQ)]+ and
[Cr(tren)(PSQ)]2+ complexes. For both the nickel and chromium compounds further
single point calculations for the doublet and quartet, and triplet and quintet spin states,
respectively, were done using the optimized geometries with the UB3LYP functional
with the 6-311g** basis set. In some cases an energetically higher antiferromagnetic
electronic state was calculated as the solution to the Kohn-Sham equations, and in these
cases the guess = mix keyword was applied in an additional calculation at the same level
of theory in an attempt to calculate a lower energy antiferromagnetic state. The guess =
mix calculations all corresponded to lower energy electronic states when compared to the
initial results. The spin densities for all molecules were calculated using the natural
population analysis developed by Weinhold and coworkers.53 All calculations were done
using tight convergence criteria,54 with a molecular charge of +1 for the nickel and +2 for

the chromium complexes. The point groups of all molecules were determined to be C1 in

the Gaussian software.

5.3 Results and Discussion

168

5.3.1 Synthesis and Characterization of NickeI(II) 2,7-di-R-
Phenanthrenesemiquinone Complexes.

Synthesis. The general synthetic procedure for the Ni-2,7-di-R-PSQ complexes
was adapted from a procedure by Bencini et al.31 and is depicted in Chart 5-3. The
deprotonated ligands are very air-sensitive in solution, and care must be taken to handle
solutions of these compounds only under inert atmospheres. The methanol solvent used
in the reaction has to be thoroughly degassed to facilitate clean reactions. The Ni-PSQ
complexes were made in reasonable yield by oxidation of the Niu-phenanthrenecatechol
species with ferrocenium hexafluorophosphate, followed by precipitation with excess
NaBPh4 salt out of a methanol solution. The phenanthrenecatechol (PCAT) complexes
were made initially due to the large amount of negative charge density at the oxygen
atoms in these dianions, making these ligands stronger Lewis bases than the semiquinone
analogues. The isolated solids were stable in air for magnetic measurement purposes, but
are highly air sensitive in solution. Nevertheless, the solid compounds were kept under

vacuum or inert atmosphere to ensure no decomposition during storage.

 

MeOH, 2NaOH
Ni(tren)Cl2 + 2,7-di-R-PCATH2 » [Ni(tren)(2,7-di-R-PCAT)]
N2 (g)
N2 (g) [FeCp2](PF6)
N
MeOH 10 NaBPh4
NH2 M12
tren = . .
HZN [N1(tren)(2,7-dl-R-PSQ)](BPh4)

Chart 5-3. General synthetic procedure for the [Ni(tren)(2,7-di-R-PSQ)](BPh4) complexes.

169

Ground State Characterization: X-ray Crystallography. ORTEP drawings of
the crystal structures of complexes 1 (Ni-PSQ), 2 (Ni-2,7-di-NO2-PSQ), and 4 (Ni-2,7-di-

NH2-PSQ) are shown in Figure 5-1.

 

Figure 5-1. Drawings of the cations of [Ni(tren)(PSQ)]+ (1) (top), [Ni(tren)(2,7-di-NOz-PSQ)]+ (2)

(bottom left), and [Ni(tren)(2,7-di-NH2-PSQ)]+ (4) (bottom right) obtained from X-ray structure
determinations. Atoms are represented as 50 % probability ellipsoids.

The selected bond distances and angles for the compounds are listed in Table 5-1, and

crystallographic data are reported in Table 5-2.

170

Table 5-1. Selected bond distances and angles for 1, 2, and 4.

 

 

1 2 4
Bond Distances (A)
Ni(1)—O(l) 2.0561(10) 2.055(4) 2.078(7)
Ni(1)—O(2) 2.0685(10) 2.061(5) 2.138(6)
Ni(1)—N(1) 2.1413(12) 2.113(5) 2.159(10)
Ni(1)—N(2) 2.0857(1 1) 2.061(5) 2.086(9)
Ni(1)—N(3) 2.1381(12) 2.135(5) 2.063(9)
Ni(1)—N(4) 2.0786(1 1) 2.073(4) 2.093(9)
C(10)-O(1) 1.2829(15) 1.283(6) 1.336(7)
C(9)—O(2) 1.2797(16 1.268(7) 1.268(8)
C(9)—C(10) 1.4398(18) 1.441(8) 1.39“
C(9)—C(14) 1.4502(18) 1.468(8) 1.39“
C(14)—C(13) 1.4111(19) 1.415(7) 1.39“
C(13)—C(12) 1.4731(19) 1.455(8) 1.39“
C(12)—C(1 1) 1.4144(18) 1.421(8) 1.39"
C(11)—C(10) 1.4560(18) 1.460(8) 1.39“
Bond Angles (deg)
O(1)—Ni(1)—N(1) 9427(4) 103.33(18) 102.5(4)
O(1)—Ni(1)—N(2) 103.41(4) 100.71(18) 98.3(3)
O(1)—Ni(1)—N(3) 9849(4) 89.10(17) 92.3(4)
O(1)—Ni(l)-N(4) 172.10(4) 170.27(18) 174.5(4)
O(1)—Ni(1)—O(2) 8054(4) 80.05(16) 78.5(3)
O(2)—Ni(1)—N(1) 8644(5) 84.38(17) 82.9(4)
O(2)—Ni(l)—N(2) l75.83(4) 179.19(18) 175.5(3)
O(2)—Ni(1)—N(3) 85.71(4) 85.84(17) 89.3(4)
O(2)—Ni(1)—N(4) 9191(4) 95.20(16) 99.4(3)
N(l)—Ni(1)—N(2) 9450(5) 95.2(2) 94.7(4)
N(l)—Ni(l)-N(3) 163.71(5) 162.58(19) 161.5(4)
N(1)—Ni(1)—N(4) 8286(4) 84.52(18) 82.2(4)
N(2)—Ni(1)—N‘(3) 9236(5) 94.44(19) 94.1(4)
N(2)—Ni(1)—N(4) 8418(4) 84.09(18) 84.0(4)
N(3)—Ni(1)—N(4) 8316(4) 82.05(17) 82.6(4)
C(9)—O(2)—Ni(1) 1 10.77(8) 109.8(4) 1 10.7(4)
C(10)—O(l)—Ni(l) 1 1098(8) 110.7(4) 1 10.4(5)
O(1)—C(10)-C(9) 118.64(1 1) 118.4(5) 119.1(5)
_ 0(2)-C(9)—C(10) 118.55(11) 119.2(6) 118.1(5)

 

a
Larger errors due to crystal quality.

171

Table 5-2. Crystallographic data for [Ni(tren)(PSQ)](BPh4) (1), [Ni(tren)(2,7-di-NOz-PSQ)](BPh4) (2),
and [Ni(tren)(2,7-di-NH2-PSQ)](BPh4) (4).

 

 

l 2 4
formula C44H46BN4Ni02 C44HMBN6NIO6 C44H43BN6N102
MW 732.37 822.37 762.4
cryst syst Monoclinic Triclinic Monoclinic
space group P2(1)/c P-l P2(1)/c
a/A 19.766 (5) 10.7720 (13) 41.270 (6)
b/ A 9.143 (2) 10.9028 (13) 10.1298 (14)
c/ A 20.466 (5) 19.510 (3) 18.190 (3)
[3/ ° 100.779 (4) 87.794 (3) 102.700 (3)
V/ A3 3633.4 (15) 1925.7 (4) 7418.5 (18)
Z 4 2 8
T/K 173(2) 173(2) 173(2)
DJ g cm'3 1.339 1.418 1.365
0max 28.3 22.5 22.5
reflns measured 41984 15583 44927
independent reflns 8750 5030 8830
observed reflns [I > 20(1)] 7175 2933 2269
u(Mo K01)/mm_1 0.58 0.56 0.57
Rim 0.04 0.098 0.098
R1“ 0.029 0.059 0.073
wR2b 0.083 0.146 0.206
GDP 1 .06 0.97 1.00

 

" R1 = Z||Fo| — chII/ 21130). b wR2 = [Zw(Fo2 — Fc2)/Zw(Fo2)2]"2. w = 1/[02(F02) + (4P)2
+ bP], where P = (F02 + 2Fc2)/3.

The X-ray structures show distorted cis-octahedral symmetry around the NilI ion, and the
Ni—O bond distances for the complexes fall within the range observed for analogous Ni"-
SQ complexes.'7‘3l The axial Ni—N distances are significantly longer than the equatorial
Ni—N bonds in each complex analogous to what was found for [Ni(CTH)(PSQ)](PF6-),

31 One way to confirm the oxidation

indicating an elongated octahedron type geometry.
state of the semiquinone ligands is to examine the C—0 and the intradiol (C(9)-C(10))

bonds. Based on the X-ray data of several semiquinone complexes, the distances of C-0

5 1.29 A and C(10)-C(9) E 1.44 A indicate the semiquinone oxidation state.55 The

172

average C—O bonds (Table 5-2) for 1, 2, and 4 of 1.28, 1.28, and 1.30 A and the intradiol
distances of 1.44, 1.44, and 1.39 A, respectively, for the complexes agree with the
literature values, except a larger deviation is seen for the intradiol distance in 4. This is
most likely due to the quality of the single crystal. Despite this the X-ray data does
provide experimental evidence for the identification of the Ni-2,7-di-NH2-PSQ (4)
complex.

Electronic Absorption Spectra. The electronic absorption spectra of acetonitrile

solutions of the Nin-PSQ compounds are shown in Figure 5-2.

 

 

 

 

————Ni-PSQ
8000 ————Ni-2,7-di-N02-PSQ
TE t -——-Ni-2,7-di-NI-12-PSQ
_Q
g 6000 5
b L
IE
-e 4000 r
8
.0
<
E
O
2 2000 t
t
0 a l r I r I
35000 30000 25000 20000 15000

Wavenurnber (cm!)

Figure 5-2. Electronic absorption spectra of [N i(tren)(PSQ)](BPh4) (1) (red), [Ni(tren)(2,7-di-N02-
PSQ)](BPh4) (2) (green), and [Ni(tren)(2,7-di-NH2-PSQ)](BPh4) (4) (blue), in ACN at 25 °C.

The electronic absorption spectrum of [Ni(CTH)(PSQ)](PF6) has been recorded
previously by Bencini et al.,31 and the observed transitions qualitatively match what is
observed in the [Ni(tren)(PSQ)](BPh4) spectrtun in Figure 5-2. Slight differences in the

energy of the transitions could be attributed to different solvent and capping ligand used

173

in the experiments. A comparison of the data is shown in Table 5-3. Bencini et al.
determined that the transitions observed at 20400, 21600, and 29200 cm’1 are MLCT in

nature and the transitions at 25800 and 31400 cm'1 are intraligand in nature based on
comparisons with a Znn analogue. Several attempts to synthesize the [Zn(tren)(PSQ)]+

complex were unsuccessful, but Gallium(III)-semiquinone molecules have been
reported.56 [Ga(tren)(PSQ)](BPh4) (12) or Ga-PSQ was then synthesized to help assign
the electronic absorption spectrum. The X-ray crystal structure of the Ga—PCAT
precursor, [Ga(tren)(PCAT)](BPh4) (11), is shown in Figure 5-3, but unfortunately single
crystals of the semiquinone analogue could not be obtained. The crystallographic data is

listed in Table 5-4, and the selected bond distances and angles are also listed in Table 5-5.

Table 5-3. Comparison of electronic absorption transitions in [Ni(tren)(PSQ)](BPh4) (l),
[Ni(CTH)(PSQ)](PF6), [Ni(tren)(2,7-di-N02-PSQ)](BPh4) (2), [Ni(tren)(2,7-di-NHz-PSQ)](BPh4) (4), and

in the last column the diamagnetic [Zn(CTH)(PSQ)](BPh4), and [Ga(tren)(PSQ)](BPh4) (12) analogues,
where MLCT = metal-to-ligand-charge transfer and IL = intraligand.

 

Electronic Absorptions (cm'l):
[Ni(CTH)(PSQ)](PF6)a’b [Ni(tren)(2,7-di-NO2-PSQ)](BPh4)c [Zn(CTH)(PSQ)](ClO4)a’7)—

 

20400 (MLCT) 19800 (MLCT) 19800 (IL)
21600 (MLCT) 21000 (MLCT) 24700 (IL)
25800 (1L) 31000 (IL)
29200 (MLCT)
31400 (IL)
[Ni(tren)(PSQ)](BPh4)c [Ni(tren)<2.7-di-NH2-Pso)l(131)th [Ga(tren)(PsoiltBPhih"
20000 (MLCT) 19800 (MLCT) 21000 (IL)
21400 (MLCT) 21400 (MLCT) 22400 (IL)
25400 (IL) 23800 (IL) 24100 (IL)
29000 (MLCT) 23 800 (sh) 29500 (IL)
31100 (IL) 31800 (IL)
33200 (IL)

 

 

a b c
Ref 13. In CH3CH2C12. In CH3CN.

174

 

Figure 5-3. X-ray crystal structure of [Ga(tren)(PCAT)](BPh4) (11).

The Ga-PCAT Ga—O and Ga—N bond lengths are in the range of the reported values of a
similar Ga-catechol complex?6 The C—0 and intradiol (C(10)—C(9)) bond lengths are
one diagnostic measure of the oxidation state of the catechol ligand, and based on the X-
ray studies of multiple transition metal catechol complexes the distances should be, C—O
z 1.35 A and C(10)—C(9) z 1.39 A. The Ga C-O = 1.36 A and C(10)—C(9) = 1.37 A

bond distances agree well with this formalism.

Table 5-4. Crystallographic data for [Ga(tren)(PCAT)](BPh4) (l l).

 

 

IQaWGIIXPCADKBPm)
formula C44H46BGaN402 T/K 1 73 (2)
M., 743.38 DJ g cm'3 1.363
cryst syst Orthorhombic Omax 25
space group Pbca reflns measured 67369
a/A 10.194 (2) independent reflns 6352
b/ A 17.807 (4) bbserved reflns [I > 26(1)] 4098
c/ A 39.916 (9) p(Mo K01)/ mm‘1 0.8
B/ ° 90 Rim 0.138
V/ A3 7246 R1“ 0.062
Z 8 MRZb 0.164
OF 1.07

 

 

5 .
" R1 = leFol — chll/ leol. wR2 = [ZW(F02 - Fc2)/2W(F02)2]m, w = 1/[02(F02) + (an
+ bP], where P = (1702 + 2Fc2)/3.

175

Table 5—5. Selected bond distances and angles for [Ga(tren)(PCAT)](BPh4) (11).

 

Ga(1)-O(1)
Ga(1)-0(2)
Ga(1)-N(1)
Ga(1)—N(2)
Ga(1)-N(3)
Ga(l)-N(4)
C(10)—O(1)
C(9)-0(2)
C(9)-C(10)
C(9)-—C(14)
C(14)—C(13)
C(13)—C(12)
C(12)—C(1 1)
C(1 1)—C(10)

Bond Distances (A)
1 395(3)
1 933(3)
2.193(4)
2.076(4)
2.100(4)
2.108(3)
1 356(5)
1.358(5)
1.369(6)
1.434(6)
1.423(6)
1 .442(6)

1.422(6
1 .43 8(6)

O(l)—Ga(1)—N(1)
O(1)—Ga(1)-—N(2)
O(1)—Ga(1)—N(3)
O(1)-Ga(1)—N(4)
O(1)—Ga(1)—O(2)
O(2)—Ga(1)—N(l)
O(2)—Ga(1)—N(2)
O(2)—Ga(1)—N(3)
O(2)-Ga(1)‘N(4)
N(l)—Ga(1)—N(2)
N(1)-Ga(1)—N(3)
N(l)—Ga(l)—N(4)
N(2)—Ga(1)—N(3)
N(2)—Ga(1)—N(4)
N(3)—Ga(1)—N(4)
C(9)—O(2)-—Ga(1)
C(10)—O(1)—Ga(1)
O(1)—C(10)—C(9)

Bond Angles (deg)

9711(14)
92.62(13)
100.78(14)
174.46(13)
88.43(12)
84.47(14)
175.74(14)
9074(14)
9593(13)
9130(15)
161.3505)
79.96(15)
93.11(15)
82.80(14)
82.63(14)
106.8(3)
107.4(3)
119.2(4)

 

The Ga—PSQ absorptions may be shifted when compared to the Znu-PSQ due to

the different solvent dielectric media of CH3CH2CI2 and CH3CN.

An additional

transition at 29000 cm'1 can also be assigned as intraligand due to a similar peak seen in

the Ga-PSQ spectrum at 29500 cm'1 (Table 5-3); Bencini et al. did not report this peak.

The transitions at 20000 and 21400 cm'1 are then assigned as MLCT bands. Complex 2

has two main features at 19800 and 21000 cm'1 and these transitions are assigned as

MLCT in nature based upon comparisons with the Ni-PSQ complex. Compound 4 has

three main features at 19800, 21400, and 23 800 cm", with a shoulder at 25200 cm". The

two transitions at 19800 and 21400 cm" are assigned as MLCT, and the band at 23800

Cm'l is assigned as intraligand in nature based upon comparisons with the Ni-PSQ

complex.

Magnetic Susceptibility. The effect of electron exchange on the ground-state
configuration of these molecules can be probed by variable temperature magnetic
susceptibility measurements. The exchange coupling in these molecules can be described
by a simple Heisenberg exchange Hamiltonian, shown in Equation 1,

H =J.as,-s2 (1)
where S] and S2 are spin operators on the two spin centers, the Ni" ion and the PSQ
radical, and J is the electron exchange integral that gauges the magnitude of the
electronic coupling between the two spin centers. Other formalisms of Equation 1 with
—J, -2.], and 2.] are also encountered in the literature. In the representation of Equation 1,
when J is negative a ferromagnetic interaction will occur or if J is positive an
antiferromagnetic coupling will operate.

Plots of the effective magnetic moment versus temperature in the range of 5—350
K for solid samples of l, 2, and 4 are shown in Figure 5-4. Complexes 1 and 4 will be
discussed first, and the Ni-2,7-di-NO2-PSQ data will be discussed below. The value of
Herr = 3.92 :1: 0.1 ug (up, = BM = Bohr Magneton) for 1 and 4 are in excellent agreement
with the spin-only value of 3.87 .313 expected for a S = 3/2 ground state and is consistent
with ferromagnetic coupling between the Nill and PSQ unpaired spins. The temperature

profile of both complexes is essentially flat from 75-350 K, with a slight decrease in the
low temperature range. The experimental data is in agreement with the lieff = 4.11 113

reported for the [Ni(CTH)(PSQ)](PF6) analogue, and the slight drop at low temperature
was tentatively assigned to zero field splitting.l7’3' The lack of temperature dependence
in the susceptibility data for 1 and 4 indicate that only an isolated ground quartet state is

present, and no population of the doublet state is seen up to 350 K (Figure 5-5) indicating

177

the 3/2 J >> kBT. Due to the lack of population in the doublet state, the data in Figure 5-4
could not be fit to extract a J value, and this agrees with what Bencini et al.“ had

determined.

 

 

 

H “a“.‘aooooaeaannn

8 . "(LT-”>15? I I I I I

E I . I '

E I I -

.2 I..-

w—a I

(D

{in

as

E

d)

.2

‘8

2::

LL]
I r I r I r l¥ I r I 1
50 100 150 200 250 300 350

 

Temperature (K)

Figure 5-4. Plots of the effective magnetic moment versus temperature for [Ni(tren)(PSQ)](BPh4) (1) (red

circles), [Ni(tren)(2,7-di-N02-PSQ)](BPh4) (2) (green squares), and [Ni(tren)(2,7-di-NH2-PSQ)](BPh4) (4)
(blue triangles).

Higher temperature SQUID measurements were undertaken to see if increasing the
thermal energy of the systems would lead to an observed temperature dependence in the
effective magnetic moment. Before these measurements could be taken, thermal
gravimetric analysis (TGA) of the samples was needed to determine the decomposition
temperature of the complexes. TGA is a simple analytical technique that measures

weight loss of a sample as a function of temperature. The resulting TGA plots for all the

Ni-PSQ complexes are depicted in Figure 5-6. Complex 1 starts to loose mass at 490 K,

178

and compounds 2 and 4 loose mass at 510 K. The Ni-PSQ complex was chosen to be the

test molecule to examine if increasing the temperature would lead to an observed

temperature dependence in the susceptibility.

Energy

 

 

 

W2)
IIIl AB = 3/2 J
S = 1 }
S : l/z \
l3/2>
Non-Interacting Ferromagnetic Coupling
J < 0

Figure 5-5. The exchange coupling interaction between the Ni” S = land PSQ S =l/2 spins.

.—
—

 

 

O

t
L
1'

r—

1..

Mass (mg)

l.

— Ni-PSQ
— Ni-2,7-di-N02-PSQ

-— Ni-2,7-di-NH2-PSQ

 

 

r—thJ-D-MONNOOO

 

 

I X

 

300

Figure 5—6. TGA plots of [Ni(tren)(PSQ)](BPh4) (1) (red), [Ni(tren)(2,7-di-N02-PSQ)](BPh4) (2) (green),

L I r
400 450 500 550

Temperature (K)

350 600

and [Ni(tren)(2,7-di-NH2-PSQ)](BPh4) (4) (blue).

179

To ensure that the mass of the sample remains unchanged during the high temperature
SQUID experiment, the experiment was run from 300-450 K. A high temperature quartz
cell had to be fabricated for use in the SQUID oven. The quartz cell had the following
dimensions: 3 mm OD, 2 mm ID with 45 mm of open tube to the quartz plug (approx. 5
m), then an additional 130 mm of open ended tube above the plug. The tube had holes
drilled just below the top to allow for attachment to the SQUID sample holder. The plot
of the high temperature data for the Ni-PSQ complex is shown in Figure 5-7. The
experimental conditions of loading the sample in the high temperature tube left part of the

powder on the sides on the tube, and not all the sample was in the sampling region.

4.5

 

4.0

Effective Magnetic Moment
N
O

. 5"
LII
° '—lfl"1 "l” l Ff'xfif

 

 

 

1.5
1.0
0.5
0.0 I i I n_ I 4“ I_
100 200 300 400
Temperature (K)

Figure 5-7. Plot of effective magnetic moment versus temperature for a high temperature (300-450 K)

SQUID experiment of [Ni(tren)(PSQ)](BPh4) (I). The low temperature data (5-350 K) was done under
normal conditions.

Since all the powder was not in the sampling area, decreased magnetization was observed

compared to the data run under normal conditions. The decreased magnetization was

180

accounted for by normalizing the high temperature data to the data collected under
normal conditions. The normalization was achieved by decreasing the measured mass of
0.0108 g to 0.00978 g used in the calculation of the effective magnetic moment to
account for the loss of the sample on the sides of the tube. The high temperature
experiment shows that only the quartet state is populated up to 450 K, with no population
of the doublet state (Figure 5-5). The thermal energy available at 450 K is approx. 320
cm], and since no population of the doublet state could be detected at this temperature,

1. The high temperature experiments agree

the energy gap must be larger than 320 cm'
with Bencini et al. and the J value is too strong to be determined, and Bencini et al. was
able set a lower limit of 400 cm'1 to the true value of J.”‘3 l DFT calculations may reveal

additional information to explain the large exchange coupling constants in these

molecules (vide infra).
Interestingly, the susceptibility data for complex 2 has a maximum value of ueff =

3.67 113 at 350 K, and the overall temperature profile displays decreased effective
magnetic moment values compared to complexes 1 and 4. This anomalous data has not
been observed in any other Nin-semiquinone complex reported in the literature, and could
possibly be explained by strong intermolecular interactions in the solid compound. In
order to determine if intermolecular interactions are present, the crystal packing diagram
showing the ligand stacking is depicted in Figure 5-8. It is evident that there is strong 7:-
stacking between the 2,7-di-N02-PSQ rings, with the shortest center-to-center distance
equal to 3.57 A. This n-stacking between the 2,7-di-N02-PSQ rings can lead to an
antiferromagnetic interaction between the S = V; radicals, which would yield decreased

effective magnetic moments across the entire temperature range. The packing diagram of

181

complex 1 is shown in Figure 5-9 for comparison purposes, where the shortest center-to-

center distance is 6.28 A.

 

 

 

 

 

labels Symmetry Plane-to-plane Center-to-center Shortest atom-to-
Angle (degree) distances (A) atom distances (A)

A -x, 41» 1-2 0 3.57 3.30 (C7...C13A)

B -1-x, 4-y, 1-2 0 8.20 3.67 (C3...C4B)

 

 

 

 

 

 

Figure 5-8. The packing diagram for [Ni(tren)(2,7-di-NOz-PSQ)](BPh4) (2).

Although the center-to-center distance is longer in 1 compared to 2 weaker
intermolecular interactions in 1 can not be ruled out in the low temperature data. It is
highly likely then that an intermolecular n-stacking antiferromagnetic interaction between
the 2,7-di-N02-PSQ radicals is responsible for the observed lower effective magnetic
moment of 2. In order to prove this experimentally, Evans NMR and liquid phase
SQUID experiments will be undertaken. A liquid phase experiment is necessary to
experimentally show that there are no intermolecular interactions, since in solution a
completely random and dilute orientation of the complex will occur that should

significantly reduce n-n interactions.

182

 

 

 

 

 

 

 

labels Symmetry Plane-to-plane Center-to-center Shortest atom-to-
Argle (degree) distances (A) atom distances (A)

A l-x, l-y, -z 0 6.69 3.39 (C5...C6A)

B l-x, -O.5+j, 0.5-z 32.4 6.28 3.57 (C10...C14B)

C l-x, 0.5+y, 0.5-z 32.4 6.28 3.48 (C11...C7C)

D X, 1+y, z 0 9.14 3.57 (C12...C20D)

 

 

 

 

 

Figure 5-9. Packing diagram for the [N i(tren)(PSQ)](BPh4) (1) complex.

NMR method.57 The Evans method uses the measured difference in the chemical shift of
a signal from a reference material in the absence and presence of a paramagnetic solute.
The simplest way to accomplish this is by placing a spiked solution of the paramagnetic
material with a standard in a sealed melting point tube inside of an NMR tube.47 The

susceptibility can then be measured by the following equation:

_ 5v
Vosf [C]

X—

183

 

58

Solution susceptibility measurements have been done since 1959 by the Evans

(2)

 

where, x is the mass susceptibility (ml/g) of the dissolved substance, 81) is the shifi in
frequency (Hz) from the value found for the pure solvent, [C] is the concentration of the
solute (g/ml), Sf is the shape factor of the magnet (41t/3 for a superconducting magnet),
and no is the operating frequency of the NMR spectrometer. The results of the variable

temperature Evans NMR experiment for 2 are plotted in Figure 5-10.

5.0

 

4.5

4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5

Effective Magnetic Moment (BM)

0
O
Vfi— I 1—' I‘V” ' fi' ‘—'_ V 1— ' jfi I r

 

 

l n l A J n l n I n l l l 1 l A l

220 230 240 250 260 270 280 290 300 310

 

Temperature (K)

Figure 5-10. Variable temperature Evans NMR of [Ni(tren)(2,7-di-N02-PSQ)](BPh4) (2) in d7 DMF.
The experimental was conducted using a 500 MHz NMR spectrometer with 2 % t-butanol used as the
reference solvent.

The ueff = 4.2 i 0.1 ug observed across the 225-300 K temperature range is in agreement

with the previous measurements form Bencini et al. for Ni-SQ’s and the solid powder
SQUID measurements for complexes 1 and 4. Although the experiment was limited by
the freezing point of the NMR solvent, the data do indicate that intermolecular
interactions are responsible for the lower effective magnetic moment observed for the

225-300 K temperature range. It can be hypothesized that the lower temperature regime

184

also has a decreased effective magnetic moment due to intermolecular interactions. A
firll temperature range from 5-350 K in solution is needed to experimentally prove this,
so liquid phase SQUID experiments were attempted.

Liquid phase SQUID experiments have been done previously by Hendrickson and
coworkers in valence tautomeric cobalt-quinone complexes; the quartz sample holder
used in our experiments was based on their designs.“49 Hendrickson and coworkers
used toluene-d3 to dissolve their sample to a concentration of approximately 1.0 x 10‘3 M,
and operated at a 1.0 T magnetic field over a temperature range of 2-400 K. Several
attempts at using the paramagnetic copper standard tetramethylethylenediammonium
tetrachlorocuprate(II),59 [(CH3)2NHCH2CH2NH(CH3)2]CuCl4, for susceptibility
measurements at 1.0 x 10'3 M and either 1.0 T or 5.5 T gave very poor results in CD3CN,
CH3CN, H20, and D20. Increasing the concentration of the sample gave somewhat
better results. To determine the proper concentration of samples appropriate for our
laboratory fabricated quartz cell, a concentration study was then done with
Cu(NO3)2°3H20 due its high solubility in water. The experiments were also run at 5.5 T
to get an increased signal response, since the magnetization (M) is directly proportional
to the magnetic field (B) multiplied by the magnetic susceptibility (x), i.e. M = xB. The
results of the concentration study with Cu(N03)2-3H20 are shown in Figure 5-11. The

literature reported effective magnetic moment for Cu(NO3)2-3H20 is um = 1.94 it}; at

300 K60 The data in Figure 5-11 clearly show for the 1.0 M, 2.0 M, and 4.0 M

concentrations the effective magnetic moment deviates to a higher value. For the 5.6 M
water solution, ueff = 1.97 113 at 300 K and matches the known experimental data quite

nicely. The less concentrated solutions show the greatest deviations from the known

185

experimental values, and this could be due to dominating background signal from the
quartz cell. Thus for this particular experimental setup to work, the highest concentration
possible must be used to observe good data. Complex 1 will be the first to be examined

in liquid phase.

 

i"
u-

2.0

0.5

Effective Magnetic Moment (BM)

 

 

l r I

l n l n l n n l r l
0 50 100 150 200 250 300 350

.0
o

Temperature (K)

Figure 5-11. Liquid phase SQUID susceptibility of Cu(NO3)2-3H20 in water at 5.5 T, at 1.0 M (red
square), 2.0 M (green circle), 4.0 M (blue triangle), and 5.6 M (brown diamond) concentrations.

The plot of the effective magnetic moment for l in DMF is depicted in Figure 5—
12. The experimental conditions for loading the liquid SQUID sample lefi some solution
on the sides of the tube, and again the data had sub-normal moments as a result. To
correct this problem we applied the same normalization condition that was used for the
high temperature data by subtracting mass from the sample (0.0062 g), and the corrected
data are shown in Figure 5-12. The data agrees with the high temperature side quite well
and asymptotes to 3.9 rig, and an additional drop is observed in the low temperature

regime. The drop starting at 40 K can be assigned to zero field splitting in the Ni-PSQ

186

complex, since intermolecular interactions should not be operative in solution. However,
to inconclusively determine the magnitude of the zero field splitting present in the low
temperature data high field EPR experiments will be required (vide infra). The zero field

splitting energy diagram can be seen in Figure 5-13.

 

 

 

 

 

5.0
4.5L
4'0L I I I I I I
E .....IO'OIO 0.0 O O O 0 Cl.-
0 i
E 3.5
O
2 3.0
.g F
v 2.5 -
En
«3 L . w.
2 2-0 - SolrdNr-PSQ
4; 1.5L 0 0.83MDMF solution ofNi-PSQ
'3 t
8: 1.0
L” t
0.5
0.0 1 1 a l .L l L l 4 l r l h
0 so 100 150 200 250 300 350
Temperature (K)

Figure 5-12. The 0.83 M DMF solution of '1 (blue circles), and the solid state data (red square) is also
shown for comparison.

56 /’/[ lmS=i 3/2
33 13%) —*:l AE=2D
Ci ‘\
m “\
\[ lms=i l/2

 

 

Figure S-13. The energy splitting caused by zero field splitting for an S = 3/2 ground state.

187

Zero field splitting is one possible mechanism that lifis the ground state degeneracy and
results in the splitting of the S = 3/2 into the mg = :1: 3/2 and i 1/2 states as depicted in
Figure 5-13. The contribution from zero field splitting to lift the degeneracy of the S = 3/2
ground state of 1 is a symmetry effect. This state is 4-fold degenerate (ZS +1) and the
symmetry of the complex can be approximated by the C2,, point group which can only
support 1-fold degeneracy, thus the ground state must split according to symmetry rules.

The data indicate that there is population of the ms = :t 1/2 states at low temperature, since
year = 2.96 at 5 K and this value reflects contributions from the partially filled S = 1/2 (ueflc
= 1.73 113 ) and S = 3/2 (pet? = 3.87 pa) states. Then as the thermal energy increases to 40

K, the population is localized in the S = 3/2 state and the data asymptote to neff = 3.9 143.

The reason for the larger dip in the low temperature data for the Ni-PSQ liquid phase data
could be due to weak ferromagnetic intermolecular interactions in the solid state, which is
not observed in solution. The interactions would be weak, since the packing diagram in
Figure 5-9 indicates long center-to-center distances between the PSQ rings. Thus, the
liquid phase data for complex 1 has experimentally indicated that the drop below 40 K is
due to ZFS, and weak ferromagnetic interactions are likely in the solid state.

Several attempts at collecting liquid phase data on [N(tren)(2,7-di-N02-

PSQ)](BPh4) (Complex 2) were unsuccessful due to solubility problems in all laboratory

available solvents examined. The [Ni(tren)(2,7-di-N02~PSQ)](BArf) (3) complex, where
BArf is the tetrakis{3,5-bis-(trifluoromethyl)phenyl}borate anion, was synthesized to

help increase the solubility, since salts with the BArf counterion are well known to

increase solubility in organic solvents.“ Increased solubility was observed in lower

188

,ar w,

 

 

it ’1‘}! .u

dielectric constant solvents such as dichloromethane, but no large improvement in the
solubility of 3 in DMF was observed. Solution SQUID data for 3 was then collected in
dichloromethane and the results are depicted in Figure 5-14. The solution phase data
depict an overall elevated effective magnetic moment compared to the solid sample,
which is consistent with intermolecular interactions in the solid state suppressing the
effective magnetic moment. The dip in the low temperature side of the data at 25 K can
be assigned to zero field splitting, and at higher temperatures the data displays a positive
slope from 50 — 280 K. The positive slope in the solution data implies that the
concentration of the sample is not high enough, since the Cu(N03)2-3H20 standard also
displays a positive slope in less concentrated samples. Overall, the solution phase
SQUID and the Evans NMR data support that intermolecular interactions are responsible

for the unique solid state magnetic data for the Ni-2,7-di-NOz-PSQ complex.

 

:5
KI!
fi f-fifiv

 

 

 

E
0)
3.5 - I
g L! I I I I I
. I
‘2 3.0 _‘ I I _ .
H I
g 2.5 ‘— _,-'
‘99 I I
g 2.0 r .I f
g 15 +_ I 0 CH2CI2 solution (BAr Salt)
"3 ' [I - Solid phase (BPh4 Salt)
g 1.0
m
0.5
0.0 I l L n J 4; L A L
0 so 100 150 200 250

Temperature (K)

Figure 5-14. SQUID susceptibility of a 0.41 M dichloromethane solution of [Ni(tren)(2,7-di-N02-
PSQ)](BArf) (3) (red circles) compared with the solid BPh4 (2) salt (green squares).

189

The above data does indicate that the exchange coupling between the PSQ
radicals and the Ni(II) center is strong and ferromagnetic, but unfortunately the J values
could not be experimentally determined. As discussed in Chapter 3, the spin density at
the oxygen atoms was found to differentially polarize at the oxygen atoms for the N02
and NH2 substituents, and it was hypothesized that the J value should be modulated by
the changes in spin at oxygen. The experimental results disagree with the free ligand spin
density hypothesis, and this it is not too surprising since in Chapter 4 it was found that the
spin density distribution shifts upon binding to a metal center. In an attempt to explain
the deviations, broken symmetry DFT calculations will be used to calculate the J values
and the electronic structures of the Ni-PSQ, Ni-2,7-di-NH2-PSQ, and the Ni-2,7-di-N02-
PSQ complexes.

Density Functional Theory. The singly occupied orbitals (SOMOs),
corresponding to the unpaired electrons in the Ni-PSQ cation, have been computed with
the UB3LYP functional and are shown pictorially in Figure 5-15 for the quartet state.
Only the Ni-PSQ complex’s orbitals are shown, since the orbitals of Ni-2,7-di-NH2-PSQ
and Ni-2,7-di-NOz-PSQ are qualitatively similar. The orbital descriptions are in
agreement with what has been found by Bencini et al. for [Ni(CTH)(PSQ)]+. The PSQ
radical SOMO is a-110 and it is delocalized over the PSQ ring with a large contribution

from the C—0 1r* molecular orbital, and this orbital is weakly interacting with the 3d Ni"-
dxz orbital. The a-108 (dzz-like) and (ll-109 (dx2.y2 like) orbitals are the two o*-type

orbitals of Ni11 that interact with the oxygen atoms of the PSQ. The observed
ferromagnetism in the temperature dependent SQUID measurements for l, 2, and 4 can

be qualitatively explained by examining the magnetic orbitals in Figure 5-14. The Ni-

190

PSQ complexes have sz symmetry and the a-108 and (ll-109 orbitals transform as the a1

irreducible representation and the o-llO orbital spans b1, and an overall ferromagnetic

interaction will then operate due to the orthogonality of the magnetic orbitals.

 

(1-110

 

(1—108

 

0t—106

Figure 5.15. Magnetic orbitals for the [Ni(tren)(PSQ)]+ complex.

One of the most theoretically sound DFT approaches to calculate the exchange

coupling constant is the approximate spin projection method of Yamaguchiém3 The

191

 

 

lanky“! J 3'.

equation used is shown in Equation 3 below, and the derivation has been reported by

Rodriguez and McCusker.64

E(HS)-E(BS)

2 _ 2 (3)
<8 > HS <8 >BS

where E(HS) is the energy of the high spin (HS) state, E(BS) is the energy of the low spin

J12=2

 

or broken symmetry (BS) state, (82mg and (82);;3 are the spin expectation values of the
HS state and BS state, respectively. The broken symmetry state is a single determinant

wavefunction of the type,28’65‘66

'33) = ITaTbi (4)
where Ta and i, are the BS spin-orbitals that are essentially localized on sites a and b. In

the case of the Ni-PSQ complexes, we then expect a net B-spin density on the oxygen
atoms of the PSQ radical and a net a—spin density on the Ni'I center of the
antiferromagnetic or BS doublet state. As for the HS state we expect net a-spin density at A
both the NiII and PSQ radical centers. The spin density distributions in the Ni-2,7-di-R-
PSQ complexes are listed in Table 5-6, and the or and B spin densities match what is
predicted for both the HS wavefunction and the BS state. The spin density at the Ni”
center is essentially identical in all three substituted Ni-2,7-di-R—PSQ’S, and the largest
localization of spin density on the PSQ rings is found at the oxygen atoms. The results of
the calculations of the exchange coupling constants at the B3LYP/6-311g** level are
shown in Table 5-7. The average Ni—O DF T bond lengths are within 0.02 A of the X-ray
values, indicating that DFT accurately models the distance between the magnetic centers.

The calculation of the J values in Equation 3 were done with (82 HS = 3.75, since the

unrestricted Kohn-Sham wave functions for the quartet state are nearly pure (i.e. (SZ)H3 =

192

 

-' 1" 11"

'.3

S(S + 1)). The (82); indicates considerable spin contamination of the broken symmetry

state, and (82); = 1.76 implies nearly a linear combination of a quartet and doublet state.

Table 5-6. The NPA UB3LYP/6-31 lg“ spin density distribution in the Ni-2,7-di-R-PSQ complexes,
where the spin density was summed for all atoms in the fragment.

 

 

 

San Density @artet:
Ni Oxygen (avg) PSQ tren

[Ni(tren)(PSQ)]+ 1.68176 0.25357 0.59828 0.2128
[Ni(tren)(2,7-di-N02-PSQ)]+ 1.68537 0.26344 0.5641 1 0.22362
[N i(tren)(2,7-di-NH2-PSQ)]+ 1.68005 0.25259 0.60654 0.20827
Spin Density Doublet:

[Ni(tren)(PSQ)]+ 1.67995 -0.13841 -0.60655 0.20338
[Ni(tren)(2,7-di-NO;;-PSQ)]+ 1.6819 -0. 16135 -0.5 7265 0.21344
[Ni(tren)(2,7-di-NH2-PSQ)]+ 1.6788 -0.1316 -0.61447 0.19892

 

Table 5-7. DFT calculations of J at the UB3LYP/6-3l lg" level for the Ni-2,7-di-R-PSQ complexes
(using Equation 3), and the corresponding spin densities at the Ni and O atoms in the quartet state.

 

 

Ni-O

b
Complex p (Ni) p (0)“ <Sz>2 <S”'>4 (A) J(cm")
[Ni(tren)(PSQ)]+ 1.682 0.254 1.76 3.76 2.068 -258

[Ni(tren)(2,7-di-NH2-PSQ)]+ 1.680 0.253 1.76 3.76 2.059 -258
[Ni(tren)(2,7-di-NOz-PSQ)]+ 1.685 0.263 1.76 3.76 2.088 -268
a

b . .
Average spin density at oxygen. Average Ni—O bond length (wrthln 0.02 A of the X-ray values).

The assumption of admixture observed in the BS state calculations agrees with previous

. . . . 31.32 - ~
reports 1n N1" and Crm-semlqulnone complexes. Accordlngly, we can wrlte

(I32 z CZCDZ + C4434 (5)

193

And then we apply the normalization condition that follows from Equation 5.
c3 + C3, :1 (6)
The spin expectation values for the contaminated state (Sz)mix, the doublet state in this
case, can be calculated by Equation 7.
2 2 2
(8 )mix z C282(82 +1)+c,s4(s4 +1) (7)
Using Equations 6 and 7 the coefficients can be determined

<82>2 —s4(s4+1)

2
= 8

 

_ 82(82 +1)—(s2)2
4 _SZ(SZ+1)—S4(S4+1)

 

(9)

For a perfect admixture of quartet and doublet spin states the spin expectation value is
(S2>mix = 1.75, which indicates the calculated antiferromagnetic state ((82); = 1.76) for the
Ni-PSQ’S is a mixture of quartet and doublet states.

The calculated J value for the Ni-PSQ complex is in good agreement with the
calculated value of ——285 cm'1 for the [N i(CTH)(PSQ)](PF6) complex.31 The calculated J
values for Ni-PSQ and Ni-2,7-di-NH2-PSQ are identical, and there is a slight increase of
10 cm'1 in the coupling constant of the Ni-2,7-di-N02-PSQ complex. The strength of the
ferromagnetic interaction between two centers is given by the magnitude of the exchange
integral between the magnetic orbitals, or it can be described as the self repulsion of the

overlap density p(i), where J is proportional to the two electron exchange integral K.'6’67

PG) = (DNi (DCDPSQ (i) (10)

194

 

J. ==K 2 [1 meaning) (11)

space

Since the spin density is the square of the square of the SOMO wavefunction, and the
product of the spin densities is proportional to the overlap integral of the natural orbitals,

and therefore to J, [2’68

JM_SQ z pMpSQ (12)
where pM and pSQ are the spin densities on the metal and semiquinone radical,
respectively. As long as the spin densities are a contributing factor to the J value the
approximation in Equation 12 should hold. The calculated spin densities at the Nin
centers are nearly equivalent, and we have hypothesized that the other variable
responsible for the changes in the magnitude of J is the spin density at oxygen. There is
an observed correlation (a small change) with the spin density at oxygen and the coupling
constant, and if the spin density is increased the J value increases. The other observable
in Table 5-7 is that if the spin density at oxygen stays the same, then the magnitude of J
remains unchanged. The observed changes agree with Equation 12, even though the
observed differences are small. The calculations of the free ligands indicated more spin
delocalization than was seen in the metal complexes. Upon examination of the 01-
SOMOs in Figure 5-15 and the spin density distributions in Figure 5-16, there is no
excess a-spin density at the 2,7-carbon atoms. The lack of spin density at the 2,7-carbon
atoms is the likely reason responsible for the small observed changes in J, since no spin
density can be delocalized to the substituent atoms. Another observation in Figure 5-16
is that there is net (ll-spin density at the 3,6-carbon atoms. Possibly placing the

substituents there would increase spin delocalization and reduce the J value.

195

 

Figure 5-16. The UB3LYP/6-311g" spin density plots for the [Ni(tren)(2,7-di-N02-PSQ)]+,
[Ni(tren)(2,7-di-NH2-PSQ)]+, and Ni(tren)(PSQ)]+ complexes.

The 3,6-di-NOz-PQ ligand (from which the 3,6-di-R-PCAT ligands are made) can be
synthesized, but its preparation is much more intensive compared to the 2,7-di-NOz-PQ
ligand. Before the preparation of the complexes, we decided to run DFT calculations at
the same level of theory on [Ni(tren)(3,6-di-NOz-PSQ)]+ (Ni-3,6-di-NOz-PSQ) and
[Ni(tren)(3,6-di-NH2-PSQ)]+ (Ni-3,6-di-NH2-PSQ) to see if there were any noticeable
changes in the spin density and/or J. The results of the UB3LYP/6-311g** calculations
using Equation 3 are shown in Table 5-8, and the most striking result (golden row) is that
the J value for the Ni-3,6-di-NH2-PSQ complex is approximately 50 cm‘l less than the
Ni-PSQ complex. Upon examination of the spin densities at the Ni" center, they were
found to be almost identical as was seen in the Ni-2,7-di-R-PSQ complexes. The average
spin density at oxygen for Ni-3,6-di-NH2-PSQ does show a larger decrease compared to
the 2,7-isomer (Table 5-3), and as a result the magnitude of the J value is also reduced.
Interestingly, the Ni-3,6-di-N02-PSQ isomer has nearly the exact calculated J value as

the 2,7-cation and the spin densities at oxygen are almost identical.

196

Table 5-8. DFT calculations at the UB3LYP/6-311g“ level for exchange coupling constants of the Ni-
3,6-R-PSQ complexes and the spin densities (p) at Ni and O in the quartet state.

 

 

Ni-O
b _
Complex Law p (O)a <Sl>2 <Sz>4 (A) J (cm l)c
[Ni(tren)(PSQ)]+ 1.682 0.254 1.76 3.76 2.068 -258

[Ni(tren)(3,6-di-NH2-PSQ)]+ 1.680 0.220 1.76 3.76 2.057 —210

[Ni(tren)(3,6-di-NOz-PSQ)]+ 1.685 0.261 1.76 3.76 2.085 -267

 

a b
Average spin density at oxygen. The average Ni—O bond length is within 0.005 A of the X-ray data,

mdlcatlng an accurate model of the distances between the magnetic centers. Calculated usmg Equation 3.

The DFT calculated J values and spin densities support the proportionality in Equation
12, and in these systems the spin density at oxygen seems to be modulating the J value.
An additional factor that could have an effect on the J value is the bond distance between
the two centers. Intuitively one would guess that the shorter the distance between the
spins, the stronger the exchange coupling. Table 5-3 also lists the average bond distances
for the complexes, and it can be seen that even though the bond length is shortened Ni-
3,6-di-NH2-PSQ compared to the Ni-PSQ the J value is decreased. The opposite trend is
seen for the Ni-3,6-di-N02-PSQ complex, and J increases despite the Ni—O bond distance
elongating. The combined structural evidence also supports the observation that the spin
density at oxygen is a major variable modulating the J value. The spin density
distribution in the Ni-3,6-di-NOz-PSQ complexes is depicted in Figure 5-17, and it can be
seen that the spin density is delocalized onto the substituents groups. As a consequence
of this the spin density is more delocalized away from the oxygen atoms in the
complexes. Interestingly, even though spin is delocalized onto the oxygen atoms of the

N02 group in the Ni-3,6-di-NOz-PSQ complex, there is a small net increase in spin

197

population at the PSQ oxygen atoms. Strong electron withdrawing substituents have

been shown to be destabilizing in radical systems due to their o-withdrawing

69,70

character, and this is one possibility for the predicted increase in spin density at

oxygen.

 

Figure 5-17. The UB3LYP/6-311g** spin density plots for the [Ni(tren)(3,6-di-N02-PSQ)]+,
[Ni(tren)(3,6-di-NI-12-PSQ)]+, and l\1i(tren)(PSQ)]+ complexes.

A larger set of substituents will need to be examined to determine if the withdrawing
ability plays a role in destabilizing the radical (vide infra). Previous literature
examinations have shown that the J value has been correlated with the spin density at a
transition metal center in metal cluster compounds.”73 To our knowledge, this is the
first instance of correlating the calculated spin density in a series of transition metal
organic radicals with the J value. In order to prove the DFT results, the [Ni(tren)(3,6-di-
NOz-PSQ)](BPh4) and [Ni(tren)(3,6-di-NH2-PSQ)](BPh4) complexes will be prepared

and analyzed by SQUID magnetometry.

5.3.2 Synthesis and Characterization of [Ni(tren)(3,6-di-R—PSQ)](BPh4) Complexes
Synthesis of [Ni(tren)(3,6-di-R-PSQ)](BPh4) complexes. The synthesis of the
3,6-di-NOz-PSQ ligand involved the photochemical reaction of 9,10-

phenanthrenequinone with SO; (g) to make the PCAT-cyclic sulfate complex, followed

198

by its nitration, and subsequent pyrolysis to remove 802 (g).74 The yield of the overall
reaction from PCAT-SOz to the 3,6-di-NOz-PQ was z 33 %, which is lower than the 45
% yield for the 2,7-di-NOz-PSQ isomer. The total reaction time and workup was also 3-4
days longer (mostly due to the PCAT-SOz synthesis time) to prepare the 3,6-di-N02-PQ
compared to the 1 day prep of the 2,7-di-NOz-PQ. Once the 3,6-di-N02-PQ ligand was
in hand, the same synthetic procedures for the 2,7-di-R-PCAT and Ni-2,7-di-R-PSQ
complexes were utilized to make the corresponding 3,6-di-R-PCAT ligands and the metal
complexes in reasonable yields. Solutions of the complexes were found to be very air
sensitive, especially the Ni-3,6-di-NH2-PSQ complex, and the Nill complexes were stored
under inert atmosphere to circumvent any decomposition.

Ground State Characterization: X-ray Crystallography. The Ni-3,6-di-R-PSQ
complexes are isostructural with the Ni-2,7-di-R—PSQ analogues, and ORTEP drawings
of the [Ni(tren)(3,6—di-N02-PSQ)]+ (5) and [Ni(tren)(3,6-di-NH2-PSQ)]+ (6) compounds
are depicted in Figure 5-18. Selected bond distances and angles are listed in Table 5-9

and the crystallographic data is listed in Table 5-10

 

Figure 5-18. Drawings of the cations of [Ni(tren)(3,6-di~NOz-PSQ)]+ (5) (lefi) and [Ni(tren)(3,6-di-Nl-I2-
PSQ)]+ (6) (right) cations obtained from single crystal X~ray structure determinations. Atoms are
represented as 50 % probability thermal ellipsoids.

199

Table 5-9. Selected bond distances and angles for [Ni(tren)(3,6-di-N02-PSQ)](BPh4) (5) and
[N i(tren)(3,6-di-NH2-PSQ)](BPh4) (6).

 

 

5 6
Bond Distances (A)
Ni(1)—O(1) 2.0352(15) 2.0372(16)
Ni(l)—O(2) 2.1357(16) 2.0877(16)
Ni(1)—N(l) 2.1093(18) 2.105(2)
Ni(1)-N(2) 2.0634(19) 2.097(2)
Ni(1)—N(3) 2.122(2) 2.145(2)
Ni(1)—N(4) 2.0993(18) 2.074(2)
C(10)-O(1) 1.270(3) 1.287(3)
C(9)—O(2) 1.263(3) 1.285(3)
C(9)—C(10) 1.451(3) 1.441(3)
C(9)—C(14) 1.455(3) 1.451(3)
C(14)—C(13) 1.413(3) 1.415(3)
C(13)—C(12) 1.461(4) 1.465(3)
C(12)—C(1 1) 1.417(3) 1.420(3)
C(1 1)-C(10) 1.449(3) 1.446(3)
Bond Angles (deg)

O(1)—Ni(1)—N(l) 101.86(7) 94.61(8)
O(1)—Ni(1)—N(2) 92.97(7) 100.69(7)
O(1)—Ni(1)—N(3) 9477(7) 9831(8)
O(1)—Ni(1)—N(4) 175 .04(7) 175 .45(8)
O(l )—Ni(1)—O(2) 7924(6) 81 .10(6)
O(2)—Ni(1 )—N( 1) 84.19(7) 87. 14(8)
O(2)—Ni(1)—N(2) 172. 16(7) 178.13(8)
O(2)—Ni( 1 )—N(3) 8357(8) 8474(8)
O(2)—Ni(1)—N(4) 103 .94(7) 94.62(7)
N(1)—Ni(1)—N(2) 9673(8) 93 23(9)
N(1)—Ni(1)—N(3) 157.09(8) 163.48(8)
N(1)-Ni(1)—N(4) 8235(7) 8357(8)
N(2)—Ni(1)—N(3) 9803(9) 9444(9)
N(2)—Ni(1)—N(4) 83 .90(8) 83 .60(8)
N(3)—Ni(1)-N(4) 81 .91(8) 82.77(8)
C(9)—O(2)—Ni( 1) 1 10. 1 6(14) 109.67(14)
C(10)—O(1)—Ni(1) 113.31(14) 111.29(l4)
g1)—c(10)_c(9) 1 18.4(2) 1 18.6(2)

 

The C—Oavg = 1.29 A and C(10)—C(9)) = 1.45 A for 6 and C—Oavg = 1.27 A and

C(10)—C(9)) = 1.44 A for 5 agree well with the ranges reported in the literature for

200

semiquinone ligands. Another structural feature of the semiquinone form of the ligand is
the bond lengths in the carbon ring. Specifically, the semiquinone should exhibit
alternating short and long C—C bond distances because of the more localized nature of the
double bonds in the rings, and this is clearly seen for the Ni-3,6—R-PSQ complexes
depicted in Table 5-10.75‘77 The axial Ni—N distances are significantly longer than the
equatorial Ni—N bonds in each complex analogous to what was found for
[Ni(CTH)(PSQ)](PF6), indicating an elongated octahedron-type geometry.“

Additionally, the Ni—O and Ni—N bond lengths are in agreement with the Ni-2,7-di-R-

 

 

PSQ complexes and similar Ni-SQ compounds in the literature.”3 '
Table 5-10. Crystallographic data for 5 and 6.

5 6
formula C45H47BN6NIO650 C45H5zBN6NlO3
Mw 845.41 794.45
cryst syst Triclinic Orthorhombic
space group P-l Pna2(l)
a/A 10.26900 (10) 18.434 (2)
b/ A 19.8542 (3) 10.3031 (13)
c/ A 20.1780 (3) 20.582 (3)
B/ ° 91.0240 (10) 90
V/ A3 4111.65 (10) 3909.0 (8)
Z 4 4
T/K 1 7 3(2) 173(2)
Dc/ g cm'3 1.366 1.35
0max 25 27.5
reflns measured 52040 44033
independent reflns 14433 8941
observed reflns [I > 26(1)] 12442 3263
p(Mo K01)/ mm" 0.53 0.55
Rim 0.021 0.048
R1 0.038“ 0.041
wR2 0.113” 0.084
@F 1.09 1.08

 

T
“ R1 = ZIIFoI — 1Fc11/ ZIFol- wR2 = [ZW(F02 — Fcziiwcozfl”, w = 1/162(Foz) +01”)2
+ bP], where P = (F02 + 2Fc2)/3.

201

 

 

l—Iazz ' -_'- L’lmr—i

C .“I

Electronic Absorption Spectroscopy. The electronic absorption spectra of an
acetonitrile solution of compounds 1, 5, and 6 are shown in Figure 5-19, and the main
transitions are listed in Table 5-11. The absorption spectra of the Ni-3,6-di-R-PSQ
isomers are quite different compared to the Ni-2,7-di-R-PSQ complexes. Complex 6 has
two main features at 20700 and 28700 cm'1 with shoulders at 15000, 23200, and 24600
cm". Based on comparisons with the assignments of complex 1 (Table 5-3), the
transition at 20700 cm'1 is MLCT and the band at 28700 cm'1 is intraligand in nature.
Complex 5 has 4 main features at 16200, 21500, 24600, and 31000 cm", with a shoulder
at 26000 cm". Again using the transitions of 1 as a guide, the band at 31000 cm'1 is
assigned as MLCT and all other transitions are intraligand in nature. It should be noted

that these are tentative assignments, and gallium analogues will be needed to properly

assign the intraligand bands.

 

10000 L

7500 -

\

5000

L

L__ i A J A

30000 25000 20000 1 5000 10000

f

2500

Molar Absorbtivity (M- 1cm- I)

 

 

Wavenumber (cm-1)

Figure 5-19. Electronic absorption spectra of [Ni(tren)(PSQ)](BPh4) (1) (red), [Ni(tren)(3,6-di-N02-
PSQ)](BPh4) (5) (green), and [Ni(tren)(3,6-di-NH2-PSQ)](BPh4) (6) (blue) in acetonitrile.

202

Table 5-11. Transitions in [Ni(tren)(3,6-di-NOz-PSQ)](BPh4) (5) and [Ni(tren)(3,6-NH2-PSQ)](BPh4)
(6), where MLCT = metal-to-ligand-charge transfer and IL = intraligand.

 

 

[Ni(tren)(3,6-di-N02-PSQ)](BPh4) [Ni(tren)(3,6-di-NH2-PSQ)](BPh4)
16200 (MLCT) 20700 (MLCT)
21500 (IL) 28700 (IL)
24600 (IL)
. 31000 (IL)

 

Electrochemistry. Metal complexes containing quinoidal ligands typically
exhibit rich electrochemistry due to the multiple oxidation states that can be accessed by

the ligand. The plots of the voltammograrns for 1, 5, and 6 are shown in Figure 5-20.

7uA

 

17 r—r ' 1 ' I ' fi fi 1
500 0 -500 — 1000 — l 500 -2000
Potential (mV)

Figure 5-20. Cyclic voltammograms of the [Ni(tren)(3,6-di-NHZ-PSQ)](BPh4) (6) (top),
Ni(tren)(PSQ)](BPh4) (1) (middle), and [Ni(tren)(3,6-di-N02-PSQ)](BPh4) (5) (bottom). All data were

acqulred 1n degassed acetonitrlle contalnlng 0.1 M NBU4PF6 at a scan rate of 100 mV 5 . All potentlals
are referenced to the ferrocenium/ferrocene couple as an internal standard.

Bencini et al. have reported the electrochemistry of [Ni(CTH)(PSQ)](PF6),3| and

observed a reversible process at -1.10 V vs. Fc+/Fc which was attributed to the

203

semiquinone/catechol couple, and an additional irreversible process observed at -0.18 V
vs. Fc+/Fc, which was attributed to the semiquinone/quinone couple. The assignment of

these processes to the ligand are straightforward, due to oxidation and reduction
processes involving Ni" being at significantly more positive and negative potentials,
respectivelyn’m’79 For 1 the redox couples at -—l.09 V (semiquinone/catechol) and at —
0.36 V (semiquinone/quinone) are in excellent agreement with the previous work.

Complex 6 exhibits a irreversible redox couple at -0.772 V (semiquinone/quinone) and a
reversible couple at -1.37 V (semiquinone/catechol) vs. Fc+/Fc. Analogously, complex 5

has an irreversible redox couple at 0.03 V (semiquinone/quinone) and a reversible couple
at -0.620 V (semiquinone/catechol). The two additional couples for 5 at —l.71 V and
—1.51 V are most likely due to the reduction of the N02 to NH; group. The E172 values
for the semiquinone/catechol redox couple for the series of compounds corresponds to the
electron accepting character of the ligands, and follows the order Ni-3,6-di-NOZ > Ni-

PSQ > Ni-3,6-di-NH2-PSQ. The redox potentials were also found to correlate with the

Hammett op parameter yielding a correlation coefficient of 0.999, suggesting that the Ni-

PSQ complexes are stabilized by transfer of negative charge from the substituents to the
radical centers.”82 This idea is also supported by the DF T data, since the strong n-donor
NH2 delocalizes the spin to the greatest extent in the Ni-3,6-di-R-PSQ complexes.

EPR Spectroscopy. The EPR spectra of l, 5, and 6 can be considered in the
strong exchange limit, where exchange coupling is much greater than the other spin
parameters, such as the Zeeman interaction or zero field splitting effects, based upon the

experimental and theoretically determined J values. In the strong exchange limit, the

204

 

 

I‘D

t. I.

effects of other operators acts as a perturbation of the isotropic exchange Hamiltonian

(Eq 1), and the following linear relationships hold for the various spin exchange levels.

gS =c1gA +c2gB (13)

DS =dlDA +d2DB+d12DAB (14)
where gs and D3 are the g and zero-field splitting tensors, respectively, and g and g; are
the g tensors of the individual spins centers. DA and D3 are the zero field splitting tensors
of the individual spin centers, while DAB is the asymmetric exchange tensor, and arises
via the exchange interaction. Bencini and Gatteschi83 provide the scalar values c1, 02, d1,
d2, and dlz for various spin pairs, and for a S = 1/2 and S = 1 spin pair the resulting

equations for the ST = 3/2 manifold are:

l 2
gs =§gA+§gB (15)
l 1
DS =0DA+§DB+§DAB (16)

where the coefficient of DA is zero, because there can be no zero field splitting in an S =
1/2 system. Using the above equations and the standard g-values for a phenoxy radical (g
= 2.00) and for Ni(II) (g = 2.2), the equation predicts a g-value for the exchange coupled
system of g = 2.13. The frozen solution EPR spectra of complexes 1, 5, and 6 are shown
in Figure 5-21. The EPR spectra were simulated with XSOPHE84 (not shown) to yield
the following isotropic g-values: l: g = 2.11, 5: g = 2.12, and 6: g = 2.12. Due to the
breadth of the EPR transitions the accuracy in the g values is not high, but due agree with
the calculations. These g values are lower than what has been observed for other Ni-SQ

complexes (g = 2.2),'7 but are in relative agreement with a similar Ni(II)-phenoxy radical

205

 

complex from Weighardt and coworkers,85 where they determined the isotropic g = 2.11.

Recall, for an EPR transition to occur,

AB = ho = geffsB (17)
ht)

where h is Planck’s constant, 1) is the frequency, 13 is the Bohn magneton, B is the

magnetic field, and geff is the effective g value.

7.,»—

Ii ' F - 1 - fl ' fl mi 1 ' fl
0 l 000 2000 3000 4000 5000 6000
B (G)

 

 

Figure 5-21. The X-band EPR spectra of the [Ni(tren)(3,6-di~N02-PSQ)](BPh4) (5) (top), [Ni(tren)(3,6-

di-NHz-PSQ)](BPh4) (6) (middle), and Ni(tren)(PSQ)](BPh4) (1) (bottom) complexes. The spectra were
acquired at 4 K in a 9:2 propionitrile to butyronitrile glass at 9.465 GHZ with a 10 G modulation amplitude.

The transition observed at gm = 2 (3340 G) is due to transitions between the ms = i ll;
manifold, and the absorptions at geff E 4 (E 1700 G) are due to transitions between the ms
= i 3/2 manifold (Figure 5-12) and are typical of a Kramers doublet.l7 The transitions in

the ms = i 3/2 manifold are formally forbidden due to the EPR selection rule (Ams = .+_

1), but due to mixing of the ms = 1/2 and ms = 3/2 states the transitions can become

206

allowed. In the spectra, no transitions were observed that could be formally assigned as
transitions between the ms = i 3/2 and 1115 = i- 1/2 states (Figure 5-12). The zero field

splitting constant (D) could not be accurately determined, since 2D (Figure 5-12) is larger
than the microwave quantum available for the spectrometer used in the experiment (0.3
cm"). Temperature dependent (4-75 K) EPR data were also collected to see if increasing

the thermal energy would lead to transitions observed between the ms = i 3/2 and m5 = i

'/2 states, but these transitions were not observed. High frequency EPR was needed to
accurately determine the zero field splitting constants in these samples.

Along these lines, complex 1 was sent to Dr. Naresh Dalal at the National High
Magnetic Field Laboratory in Tallahassee to determine the D value. The results of the
temperature dependent high frequency (up to 406 GHz) measurements on a powder

sample of l are shown in Figure 5-22.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

9 1 10 11 12 13
Field (T)

Figure 5-22. Temperature dependence of a powder sample of [Ni(tren)(PSQ)](BPh4) (l) at 33 '1 .2 GHz.

207

 

-v-cw"

__. _

 

[I ‘1'.

As the temperature increases in the spectrum, the side peaks become weaker showing that
the spectra originate from the S = 3/2 ground state. The spectrum at 10 K and 331.2 GHz
was simulated to yield gx = 2.127, gy = 2.127, gz = 2.13, D = 0.651 cm", and E = 0.064
cm". The isotropic g value is in good agreement with the value derived from X-band
measurements and calculated by Equation 15, and the D value is in the range of the lower
limit estimate of what has been determined by Weighardt and coworkers for the Ni-
phenoxy radical system.85

Interestingly, a radical spin delocalization constant is based upon the zero field
splitting constant of 1,3-diaryl-substituted triplet diradicals, where the magnitude of the D

parameter depends upon the magnetic dipole interaction of the two uncoupled spins,”87

D=CpApB

 

(19)
die

where pA and p3 are the n-spin densities on the two radical centers and d AB is the distance
between the uncoupled radical centers and is constant. The D value was found to provide
a reliable measure of radical stabilization due to a strong correlation with the D parameter
found from experiments and the calculated spin densities at the same radical center using
semiempirical methods.88 It is hypothesized that the spin delocalization predicted by
DFT in the Ni-PSQ systems may correlate with the magnitude of the zero field splitting
parameter, in essence an experimentally determined transition metal-radical stabilization
scale. It must be noted that the physical origin of the zero field splitting in the Ni-PSQ
complexes arises from symmetry arguments. The sz symmetry of the complexes
dictates that there can only be 1-fold degeneracy, and as a result the 4-fold degenerate (ZS
+1) S = 3/2 ground state must split into Kramers doublets in zero magnetic field. Since

the zero field interaction in the biradical has a different physical origin from the Ni-PSQ

208

complexes it is unclear if a correlation will be found. Future work along these lines with
the other members of the Ni-3,6-di-R-PSQ series will be conducted to ascertain if this
hypothesis can be justified.

Magnetic Susceptibility. The magnetic susceptibility plots for complexes 1, 5,

and 6 are shown in Figure 5-23.

 

4wean.......t.m-o-~»~--

53—91

o Ni-3,6-l—l-PSQ
* A Ni-3,6-di-NH2-PSQ

]_ - Ni-3,6-di-N02-PSQ

Effective Magnetic Moment
N
f

 

 

 

0 . l . l I l - l 1 l r I . l
0 50 100 150 200 250 300 350
Temperature (K)

Figure 5-23. The effective magnetic moment vs. temperature for the Ni(tren)(PSQ)](BPh4) ( 1) (red circle),
[Ni(tren)(3,6-di-N02-PSQ)](BPh4) (5) (green square), and [Ni(tren)(3,6-di-NI—12-PSQ)](BPh4) (6) (blue

triangle) complexes.

The data show a slight rise in the effective magnetic moment at low temperature to 50 K,
and then the moment flattens out to neff = 3.9 BM. The data are in agreement with what
was observed for the Ni-2,7-di-R~PSQ isomers, and an S = 3/2 ground state is again
experimentally confirmed. The rise at low temperature can be assigned to zero field
splitting, based on the liquid phase Ni-PSQ data (vide supra). The DFT calculations

predicted that the Ni-3,6-di-NH2-PSQ complex would have a reduced J value in

209

comparison to the Ni-PSQ complex, due to decreased spin density at the oxygen atoms.
Upon closer examination of Figure 5-23, the data for 6 do show temperature dependence

over the observed range as seen in Figure 5-24.

 

 

 

 

4.2
4.1 i
E
Q)
E
o
2 4 0 p H
.2
H
G)
t:
a” 3 9 l
2 - _ _,
0 J— -119 cm
>
g g = 2.08
o — u
3:, 3'8 TIP = .00022 cm3mol ‘
3.7 a l 4 l L l 4 l A J a l L I
0 50 100 150 200 250 300 350

Temperature (K)

Figure 5-24. The effective magnetic moment of [Ni(tren)(3,6-di-NH2-PSQ)](BPh4) (6) (blue triangle), and
the solid line represents a fit to the data using parameters described in the text.

The value for peer at 50 K is 4.04 up for 6, which well represents the S = 3/2 ground state;

however, upon increasing the temperature to 350 K um = 3.92 113. The observed

temperature dependence was quantitatively fit with MAGFIT89 by using a form of the

Van Vleck equation,68 shown in Equation 20

s3+1 mac-E(S)“
NAgzplz3 ;( )( )

 

x = (20)

3kT 2 (ZS +1)e-E(S)/kT
S

210

 

where NA is Avogadro’s number, g is the gyromagnetic factor of the electron, up is the
Bohr Magneton, T is the temperature, S is the total spin quantum number of a given spin
state, and E(S) are the eigenvalues from the Heisenberg Hamiltonian (Equation 1) that
depend on J63 Equation 20 is then compared with the measured values of x and the
exchange coupling constants can be determined by a least-squares fitting. The J and g
values were the only two variables allowed to vary and the temperature independent
paramagnetism (TIP) was fixed at 0.00022 cm3mol" analogous to a Niu-verdazyl radical
complex.90 The fit to the experimental data gave J = -119 cm"1 and g = 2.08. It must be
noted that the MAGFIT program utilizes H = 2.11;St'S; for the data analysis, where the

energy difference between the spin states is then 3J with this Hamiltonian. Using our
experimental and calculated J values, the energy gap between the spin states is AEexp =

357 cm'l and AEDFT = 330 cm". The experimental and theoretical quartet-doublet energy
gaps agree satisfactorily, but remember that our goal is to find a trend in J not exact J
values. The g value also agrees well with the above EPR measurements. Unfortunately,
complexes 1 and 5 did not show any temperature dependence, and experimental J values
could not be determined. High temperature SQUID data was attempted for complex 5
(TGA decomposition temperature is 480 K), but no temperature dependence was
observed. Three important conclusions arise out of the above examination. First, the J
value was modulated by the NH; substituent to lead to an observable decrease in the
effective magnetic moment vs. temperature. Secondly, the modulation in J is most likely
due to the spin delocalizing effect of the 3,6-di-NH; substituents as predicted by the DFT
calculations. Thirdly, the J values for 1 and 5 are too large to be determined

experimentally, confirming an isolated S = 3/; ground state. DFT did predict that the spin

211

density is more localized at the oxygens in these complexes, and this suggests that we
should have a large J value. Experimentally, J does decrease as predicted for 6 and
remains strong for both 1 and 5, and as a result we only have two pieces of experimental
data to confirm the relationship in Equation 12. A general trend of experimentally
determined J values between at least three complexes is needed to soundly prove the
hypothesis. To investigate the presence of a firm correlation between the spin density at
oxygen and the J value according to Equation 12, a larger set of substituents will be

examined with broken symmetry DF T calculations to see if the trend is indeed general.

5.3.3 Density Functional Theory Study of Substituted [Ni(tren)(3,6-di-R-PSQ)]+.
The above DFT results for the three Ni-3,6-di-R-PSQ complexes indicated a trend
for modulating the J value, and to examine if this trend is indeed general a set of 17
substituents including n—donors (N H;, NMe;, and OMe), n—acceptors (N 0;, CN, SO;CN,
and vinyl), o—donors (CH3, C(CH3)3, and Si(CH3)3), o-acceptors (CF3, IF4, and C(CF3)3),
and n-donors, o-acceptors (F, Cl, and Br) were examined, where H was used as the
reference point. Equation 3 was used to calculate all J values, and the results indicated a
strong correlation of J with the average spin density at the oxygen atom, with r = 0.998,
shown in Figure 5-25. Figure 5-25 indicates that as the magnitude of J is increased
(becomes more negative), the spin density at the oxygen atoms also increases. The
opposite trend is also true, when the spin decreases at the oxygen atoms, the J value also
decreases. The substituents responsible for increasing J are strongly electron
withdrawing, and those responsible for decreasing J are more electron donating. The

high correlation coefficient and number of substituents used indicate that the trend is

212

general for the ferromagnetic exchange interaction between Ni“ and an organic radical.
The Nill a-unpaired spin density remains largely unchanged throughout the entire series,
and according to Equation 12 a correlation should also be seen between the product of
spin densities at the nickel and oxygen centers. The correlation between the product of

the spin densities is depicted in Figure 5-26.

 

-280 -

~260 -

‘T -240 —

J(cm )

-220 —

 

 

 

-200 -
. NIVIICZ

l r l

I 1 l . 1
0.21 0.22 0.23 0.24 0.25 0.26 0.27

 

Average Spin Density at PSQ Oxygens

Figure 5-25. Correlation between the UB3LYP/6-311g** calculated average a-spin density (NPA) at the
PSQ oxygen atoms and the J value calculated with Equation 3.

The correlation coefficient in Figure 5-26 is strong with r = 0.998, and it is identical to
the correlation coefficient seen in Figure 5-25 for the J vs. average spin density at
oxygen. This attests to the unchanging spin density population at the nickel center, and
indicates that the spin at oxygen is the main factor modulating the J value in these

systems. The correlation in Figure 5-26 also supports the proportionality between the

213

 

.

two electron exchange integral and the product of the spin densities suggested in

Equation 12, and gives theoretical evidence for the applicability of the equation.

 

 

 

 

 

-280

-260

“7224

E 0

8

H

-220

-200 NMe2
I 1 l 1 l 1 14 1
0.36 0.38 0.40 0.42 0.44 0.46

p(Ni)p(O)

Figure 5-26. The correlation between J calculated with Equation 3 and the product of the NPA Nill and
average oxygen a-spin densities, calculated at the UB3 LYP/6-3l 1 g“ level.

The electron withdrawing nature of the PSQ ligands seemed to be playing a role

in the spin delocalization as seen in Figures 5-25 and 5-26, and another correlation that
was identified is between the J value and the Hammett op+ substituent constant,9| and this
is shown in Figure 5-27. Not all substituents could be used, since the op+ parameters for

SO;CN, IF4, and C(CF3)3 have not been experimentally determined. The correlation in
Figure 5-26 is strong with r = 0.97, and indicates that the strongly electron withdrawing

substituents (large positive op+) increase the J value, while the strong electron donating

groups (large negative op+) decrease the J value.

214

 

-280

r

-270
-260
-250

-240

J (cm'l)

-230
-220
-210

-200

-2.0 -1.5 -l.0 -0.5 0.0 0.5 1.0
0' +

 

 

 

 

Figure 5-27. The correlation between the UB3LYP/6-3l 1g** J value calculated with Equation 3 and the
Hammett op+ substituent constant. The substituent constants were taken from Hansch and coworkers.92

The above series of correlations imply yet another correlation: that as the spin density at
oxygen decreases, the charge density increases at the oxygen atoms. The opposite trend
is also observed, and if the spin is increased at the oxygen atom, then the charge density
decreases. This correlation (r = 0.97) is different from the phenoxy and free semiquinone
radicals studied thus far, because in each substituent spin and charge density at the
oxygen centers move with opposite polarization. In order to try to develop a model for
the movement of spin density in these systems, and then try to relate it to the spin and
charge density correlations, the molecular orbitals were examined for any similar trends
that were developed for the phenoxy and semiquinones radical systems.

Similar to the phenoxy radical case, a correlation was discovered between the 01-

HOMO-a-LUMO gap and the spin density, depicted in Figure 5-28.

215

 

 

 

_. _ “(J-Ff

 

0.270 - ‘SOZCN
o
m - A1174
5 - A CN 3 3 /

B Cl / H
E30252— FILL"
E - TMSZ'; IIEH
c6 / -
3, 0.243 t on, , t "‘y 3
g . A vinyl ' /
o I
o 0.234 - IfOMe
1: /
a / OEt
U) /
a) 0.225 - , ,
§ l/{NHZ
2 0.216 ~ , INM62
l 1 141—L 1 l 1 l #1 1 J_1_l 1 J_L l

 

 

 

2.9 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4.0
a-HOMO-or-LUMO gap (eV)

Figure 5-28. The correlation between the average a-spin density at oxygen with the (I-HOMO-(l-LUMO
gap, calculated at the UB3LYP/6-3l 1g** level.

A general conclusion can be drawn from Figure 5-28, and as the u-HOMO-u-LUMO gap
decreases, the spin density at the oxygen atoms also decreases. This is in agreement with
the previous data for substituted phenoxy radicals and semiquinones: the a-HOMO-a-
LUMO gap is a marker which describes the spin delocalization in these systems.
Another striking piece of data in Figure 5-28 is the lack of a correlation for the strong
electron withdrawing groups. These substituents all have an artificially small (ll-HOMO-
a-LUMO gap based upon their calculated spin delocalizing ability. Strong electron
withdrawing groups were also found to be outliers in the phenoxy radical case, and the
smaller a-HOMO-a-LUMO gaps were accounted for by a simple four-orbital, three-

electron molecular orbital interaction.34 Analogous MO interactions were sought after to

216

explain the outliers and correlation in Figure 5-28. First the general correlation with the

a-HOMO-a-LUMO will be examined, then the outliers will be explained using molecular

orbital models.
A weak interaction can occur between the filled dxz orbital of Ni11 and the PSQ

SOMO orbital, since they both span the b1 irreducible representation of the C;v point

group (Figure 5-29).
PSQ-
LUMO_ ________ __ LUMO

— ------- —— PSQ-LUMO 0" 11
I 11 ot-HOMo-ot-LUMO

a-LUMO _“
(l-HOMO-(l-LU MO gap (ix-HOMO! '\ gap

(at-HOMO ,_l_li,
at. 1.1...-
' - ‘ SOMO

 

fl

 

Energy

(l-HOMO - l
a-HOMO - 1

 

Figure 5-29. Weak bonding interaction between the filled Ni'l-dxz and two energetically different PSQ-
SOMO orbitals, where the a-HOMO-a-LUMO gap is labeled in blue.

The highest occupied a-SOMO of the Ni"-PSQ complex (01-110 in Figure 5-15) depicts
the delocalization of the SOMO over the PSQ ligand, and the weak interaction with the

Nill dxz orbital. Recall that the general molecular orbital model used to describe the spin

delocalization in the phenoxy radical was a three-electron, two-orbital interaction, where

217

 

a filled lone pair orbital of a donor group interacted with the phenoxy radical SOMO. In
the case of the Ni-3,6-di-R-PSQ complexes, a three-electron, two-orbital interaction can
also be invoked. This MO model assumes that the energy of the PSQ-LUMO and Ni"-
dxz orbitals remain unchanged, while the energy of the PSQ-SOMO will depend on the
nature of the substituents. If the substituents are strongly electron withdrawing, the
energy of the PSQ-SOMO will decrease (Figure 5-29 left) and less mixing between the
PSQ—SOMO and the Niu-dxz orbitals will occur, lowering the energy of the a-HOMO,
thereby decreasing the a-HOMO-a-LUMO gap. Increased spin localization at the
oxygen atoms will also occur as a consequence of the decreased orbital mixing, since less
spin can be delocalized away from the oxygen atoms of the PSQ radical. If a substituent
is strongly electron donating, then the energy of the PSQ-SOMO will increase (Figure 5-
29 right) and increased orbital mixing with the Ni"-dXZ orbitals will occur, and this has
two consequences: (1) the spin density will be more delocalized, and less spin population
will be observed at the oxygen atoms, and (2) the energy of the a-HOMO will increase
resulting in the a-HOMO-a-LUMO gap decreasing. These observations are all
predicated upon the energy of the u-LUMO remaining unchanged. The MO model above
supports the observed decreased Spin density at the oxygen atoms for the donors and
weak acceptor substituents, but does not account for the deviations of the strong electron
withdrawing groups.

Using the molecular orbital model for the outliers of the phenoxy radical case as a
guide, an analogous four-orbital, three-electron interaction can be invoked for the Ni-
PSQ complexes and is depicted in Figure 5-30. The proposed MO diagram to describe

the decreased a-HOMO-a-LUMO gap involves the interaction between a 71:*-PSQ orbital

218

 

that is low enough in energy due to the electron withdrawing nature of the 3,6-acceptor-
PSQ radicals and can interact with the LUMO orbital of the Ni11 cation. The three-
electron, two-orbital model above assumed no interaction with the unoccupied 1t*-PSQ
orbital. If this orbital is interacting the a-HOMO-a-LUMO gap will exhibit artificially
decreased values compared to cases where this interaction is not operative (Figure 5-29).
Examination of the percent contributions to each orbital may possibly give additional
theoretical evidence for suggesting the molecular orbital diagrams above. The data for
the analysis of the molecular orbitals for the all complexes along with the protio reference

is given in Table 5-12.

 

LUMO—{I \.
‘x :1—— 1t * -PSQ
OL-LUMO ‘vfi "
E ot-HOMo-ot-LUMO gap
:5 OL-HOMOI ‘L‘
La :\
Nin-dxzt \

I

pi— PSQ~SOMO

a-HOMO - l

 

Figure 5-30. The four-orbital, three-electron interaction of the outliers in the Ni-3,6-di-R-PSQ complexes.

219

'll

”fwfimlmlnprpnhlnihriil

 

Table 5-12. The % nickel (% Ni) and average % oxygen (% O) in the highest occupied 1).-HOMO (01-
110), along with the average a-spin density at the oxygen atoms (p(O) in a-electron units).

 

% Nia

%o"

 

 

Complex 9(0)

[Ni(tren)(3,6-di-NMe;-PSQ)]+ 0.267 9.44 0.2161
[Ni(tren)(3,6-di-NH;-PSQ)]+ 0.308 10.69 0.2200
[Ni(tren)(3,6-di-OEt-PSQ)]+ 0.364 12.12 0.2330
[Ni(tren)(3,6-di-OMe-PSQ)]+ 0.370 12.28 0.2338
[Ni(tren)(3,6-di-vinyl-PSQ)]+ 0.41 1 12.96 0.2390
[Ni(tren)(3,6-di-OH-PSQ)]+ 0.415 13.25 0.2404
Ni(tren)(3,6-di-t-butyl-PSQ)]+ 0.466 14.31 0.2473
[Ni(tren)(3,6-di-Me-PSQ)]+ 0.479 14.63 0.2481
[Ni(tren)(3,6-di-TMS-PSQ)]+ 0.491 14.78 0.2507
[Ni(tren)(3,6-di-F-PSQ)]+ 0.497 15.02 0.2514
[Ni(tren)(3,6-di-Br-PSQ)]+ 0.463 13.96 0.2518
[Ni(tren)(3,6-di-Cl-PSQ)]+ 0.481 14.39 0.2527
[Ni(tren)(PSQ)]+ 0.518 15.68 0.2540
[Ni(tren)(3,6-di-CN-PSQ)]+ 0.545 15.54 0.2574
[Ni(tren)(3,6-di-C(CF3)3-PSQ)]+ 0.564 16.03 0.2619
[Ni(tren)(3,6-di-CF3-PSQ)]+ 0.566 16.05 0.2622
[Ni(tren)(3,6-di-NO;-PSQ)]+ 0.589 16.24 0.2610
[Ni(tren)(3,6-di-IF4-PSQ)]+ 0.608 16.70 0.2658
@(trenx3,6-di-SO;CN—PSQ)]+ 0.637 16.96 0.2698

 

a
[Enzao / EnzmoJ :1: 100 = % contribution. 271230. is the sum of the squares of the atomic orbital

. , 2 . . .
coefficrents of the atom or group of interest, and Zn mo. 15 the sum of the squares of all atomic orbital
coefficients in a specific molecular orbital.

Table 5-12 shows strong support for the modulation of the energy of the a-HOMO as
Shown in Figures 5-29 and 5-30. As the energy of the (ll-HOMO orbital is raised by a
strong electron donating group like NH; there is more mixing between the orbitals, and as
a result the % Ni and % 0 should decrease compared to the protio case. This is exactly
what is observed in the data in Table 5-12. Additionally, the energy of the 111-HOMO will
be decreased by a strong electron withdrawing group like NO;, and as a result there is
less mixing between the orbitals and the % Ni and °/o 0 will increase compared to

hydrogen. This increase in % Ni and % O is also observed in Table 5-12. The data also

220

support the correlation observed for the opposite polarization of spin and charge density
at the oxygen atoms of the PSQ radical. As the electron withdrawing character of the
PSQ ligand is increased (charge density at oxygen decreases), the spin density is more
localized due to less mixing. The opposite trend was also found to be true of an electron
donor increasing the charge density at the oxygen atoms, but the Spin is decreased due to
more orbital mixing. The % Ni and % O values then give reasonable evidence for the
variation in the relative energy of the a-HOMO orbital.

The four-orbital, three-electron MO diagram is also supported by the % Ni and
% 0 data. However, the energy of the a-LUMO is also proposed to be effected by an
interaction with higher lying orbitals. As before in the phenoxy case, a large
% substituent in the u-LUMO was indicative of the four-orbital, three-electron
interaction, but in the Ni-PSQ complexes the a—LUMO is mostly dominated by the PSQ
contribution. AS a result of this the changes in % substituent are difficult to interpret.
Certainly the possibility of multiple orbitals interacting with the LUMO of Ni11 to
decrease the HOMO-LUMO gap may also be important in these molecules. Indirect
evidence of this possibility is given in Figure 5-23, where a large random range of
deviations is seen. Since the deviations are like “shot gun scatter” no clear trend can be
determined, and the mechanism for spin polarization can’t be firmly established.

The DFT data for the Ni-3,6-di-R-PSQ complexes yields strong support for the
proportionality in Equation 12, where the spin density at the oxygen magnetic center was
found to be directly proportional to the exchange coupling constant J. The spin density
was found to vary with opposite polarization as the charge density at the oxygen atoms,

in contrast to the PSQ and phenoxy systems. The a-HOMO-a-LUMO gap was again

221

 

found to be a marker for the spin delocalization of the substituents, as increasing the gap
increases the a-spin population and decreasing the gap decreases the spin density at the
oxygen centers. The mechanism responsible for the spin delocalization is the three-
electron, two-orbital interaction, and this MO description is comparable to that proposed
for the radical ligands. The strongly electron withdrawing substituents were again
outliers in the a-HOMO-a-LUMO gap correlation, and interactions with higher lying
orbitals are the likely cause. In order to see if the a-HOMO-a-LUMO gap correlation is
general for all exchange coupling interactions, antiferromagnetic coupling in Cr-PSQ
complexes will be examined to see if any of the same trends are observed.

It should also be noted that another functional, BLYP, was also used to ensure
that all of the above correlations and results were not functional dependent, and although
the numbers for the variables were different the exact same correlations are observed.
The effect of bond distance on the J value was also examined, and the same set of
calculations at the UB3LYP/6-311g** level were done using identical Ni—O bond
distances (2.07 A) in all cases. The results of fixing the bond distance again produced
different numbers, but in all cases the correlations held. This data lends additional
support for the spin density at the oxygen atom being the major variable responsible for

the modulation of the J value in these systems.

5.3.4 Antiferromagnetic Coupling in [Cr(tren)(3,6-di-R-PSQ)]2+ Complexes.
Crm was chosen to study antiferromagnetic exchange coupling in the
[Cr(tren)(3,6-di-R-PSQ)]2+ (Cr-PSQ) complexes. Cr"' is a d3 transition metal with its

111

unpaired electrons in the dxz, dxv, and dyz orbitals. These orbitals in Cr have the same

222

n—symmetry as the PSQ 7:*-SOMO which results in orbital overlap and an
antiferromagnetic interaction. The energy level diagram for this interaction is shown in

Figure 5—31, where the energy separation between the S = 1 and S = 2 states is 2.].

 

 

L
,. 12>
g S1 = 3/2 ](,"
L5 52 = 1/2 \\\ A E = 2.]
‘ 11>
Non-interacting Antiferromagnetic Coupling
J > 0

Figure 5—31. The antiferromagnetic coupling interaction in Cr-3,6-di-R-PSQ complexes. States are
labeled as |ST), where ST is the total spin.

 

Figure 5-32. The UB3LYP/6-3l 1g** calculated B-HOMO orbital of [Cr(tren)(3,6-di-R-PSQ)]2+ of the
triplet ground state.

In the case of the Cr-PSQ complexes, the B-HOMO is now important because it is the B-

111

spin on the PSQ ligands that communicates electronically with the Cr a-Spin density to

yield the Weak bonding-type interaction. The weak bonding-type antiferromagnetic

223

interaction between the Crm-dxz and the PSQ 11:*-SOMO is depicted in the [El-HOMO
wavefunction in Figure 5-32 above. The strength of the antiferromagnetic coupling will

depend upon the overlap integral between the magnetic orbitals, i.e.l6

S12 = (W(l)|‘P(2)) (21)
where S); is the overlap integral and J is proportional to the absolute value of 512?”.l6’93
Due to the success of the proportionality between the product of the spins at the two
paramagnetic centers and the ferromagnetic J in Ni-PSQ, it is hypothesized that an
analogous relationship to Equation 12 can be suggested for the antiferromagnetic
coupling constant in the Cr-PSQ complexes, i.e.

JCr—PSQ “ pCerSQ (22)

where pcr is the a-spin density on Cr!" and pPSQ is the average B-spin density at the
oxygen centers of the PSQ radical. In order to test this hypothesis, DFT calculations
were undertaken on substituted Cr-3,6-di-R-PSQ complexes using the same 17
substituents as the Ni-3,6-di-R—PSQ complexes.

The J vs. spin density and J vs. pc,ppsQ results for the 17 Cr-3,6-di-R-PSQ
complexes are depicted below in Figures 5-33 and 5-34, respectively. Recall that the J
value should be positive in the Cr-PSQ complexes (according to Equation 1), and since
an antiferromagnetic interaction is operating the spin density at the oxygen atoms should
be negative. The correlations are quite strong with r = 0.99 in both cases. The proposed
proportionality in Equation 21 seems to be valid for antiferromagnetic coupling, and
applies for both ferromagnetism and antiferromagnetism for metal semiquinone
complexes. The spin density at the oxygen centers is again the major variable

responsible for the changes in the J value, since the spin at Cr"l remains constant.

224

750

 

700
650 h
600 h

"A 550 i

g 500 -

..,

450

400

 

350 1-

 

NMe2'

% 4; I P I | l

00 1 - .
-0.16 -0. 14 -0.12 -0.10 -0.08 -0.06 -0.04
Average Spin Density at Oxygen in PSQ

 

 

Figure 5-33. J (calculated with Equation 3) vs. the average B-spin at the oxygen atoms for [Cr(tren)(3,6-
di-R-PSQ)]2+ complexes calculated at the UB3LYP/6-3l 1 g“ level.

750

 

700
650
600

550

 

-l
J(cm )
(I!
O
O
I ' I ' I ' I ' I ' I ' I ' I

 

 

 

450
400
350
. NMe;
300 l_1_ I 1 I 4 l 14 1 I 1 l 1 I
-0.45 -0.40 -O.35 -0.30 —0.25 -0.20 -0.15 ~0.10
9(Cr)p(PSQ)

Figure 5-34. J (calculated with Equation 3) vs. the product of the average B—spin at the oxygen atoms and
Cr'" centers for [Cr(tren)(3,6-di-R-PSQ)]2+ complexes calculated at the UB3LYP/6-31 1 g” level.

225

r
bf

 

sf:—

-O.72

 

 

 

 

a -O.73

cu

01)

32‘

O -0'74F

‘66

b

'5 -0.75

I:

a)

Q

30 -0.76

:a

.2

0 -0.77

a)

DD .

Q2; \\
> -0.78 NMe
< l 1 L 1 1 I 1 I 1 2!

 

-0.16 -0.14 -0.12 -0.10 -0.08 -0.06 -0.04
Average Spin Density at Oxygen

Figure 5-35. The correlation between the average B-spin and charge density at the oxygen center in
[Cr(tren)(3,6-di-R-PSQ)]2+ compounds.

The spin population is controlled by the choice of substituents, and the electron
withdrawing groups increase J by increasing spin localization, while the donating groups
decrease J by increasing spin delocalization. The charge density was again found to
correlate very strongly with the spin density at the oxygen centers, and this is depicted in
Figure 5-35. The correlation coefficient of r = 0.999 is very strong, and as the charge
density increases/decreases at the oxygen atoms, the spin density decreases/increases.
The exact same correlations are seen in both the Ni-PSQ and Cr-PSQ complexes. In
order to identify the mechanism for the spin delocalization in these systems we once
again identified a correlation with the HOMO-LUMO gap. In this case the spin density
correlates strongly with the B-HOMO-B-LUMO gap as shown in Figure 5-36. The most

salient feature of Figure 5—36 is that the average spin density at the oxygen centers for

226

every substituent correlates with the B-HOMO-B-LUMO gap. Recall, that in the Ni-PSQ

complexes all strongly electron withdrawing substituents deviated from the correlation.

 

 

 

 

\ NMCZ
o -0.04 - s!
‘53 ‘ , II NH2
2: s
a -0.06 - \ \ \
8:13 OMe .\
X -0.08 _ Vinyl \ \
e ‘ s \ t-butyl
‘3 ' CH
3. TMS-II.\ 3
3 '0.10 - Br I \ F
a _ C1 l
0.) \I\
.E -0.12 L C(CF3); i' [F4
9. ‘ I
m 1 N02 ‘ I
go -014 f SOZCN ‘ s
e s
q, .
2 -0.16 1 . 1 . 1 . 1 1 1
2.6 2.8 3.0 3.2 3.4

B-HOMO-B-LUMO gap (eV)

Figure 5—36. The correlation between the average B-spin density at the oxygen atoms with the B-HOMO-
B-LUMO gap for the [Cr(tren)(3,6-di-R-PSQ)]2+ complexes.

The difference in the two systems can be explained with a simple molecular orbital

diagram similar to Figure 5-29 where the Cr”I

-3dxz has the proper symmetry to interact
with the u—PSQ-SOMO orbital. The MO comparison between the unsubstituted complex
and the electron withdrawing NO;-PSQ ligand is depicted in Figure 5-37, and the
comparison with the donor NH;-PSQ is shown in Figure 5-38. The molecular orbital

model proposed is a two-orbital, two-electron interaction, where there is a net bonding-

type stabilization of two electrons. In this case there is no involvement with energetically

227

higher lying orbitals, as only the Crm-3dxz and PSQ-SOMO orbitals determine the B-

HOMO-B-LUMO gap. The lack of any interactions with energetically higher orbitals is

likely the reason why there are no outliers in the substituted Cr-PSQ B-HOMO-B-LUMO
gap correlation.

 

I, \
I, ‘
I,’ 1‘ I it
II, \‘ ’I ‘1
II 1‘ [’1 \“
II “ ’x \‘
a 1 1
\ \i 1 ‘ 1
1 1 1
Cr-3dxz 1‘ Cr-3dxz .
\\ ‘1 \ i
\‘ “ '1 AI: 2 ‘\
“ 1“ \|‘ ‘|
1 1
‘\ , TIE-PSQ ‘1 ‘1
I \ 1
i, 1 ,r’ SOMO ‘. x
\ ’ ‘ \
‘1 I ’l ‘ ‘1
J’ ‘
1
‘. n-NOZ
‘ I
‘1 I, PSQ
1 I
1 /
\

 

Figure 5—37. The changes in the B-HOMO-B-LUMO gap for the 3,6—di-H-PSQ vs. 3,6-di-N02-PSQ
substituted Cr-PSQ’S.

"3 ‘1 ‘L AE3 J— 7t-NH -PS
Cr-3dxz ’ 2 Q

SOMO

 

 

Figure 5-38. The changes in the B-HOMO-B-LUMO gap for the 3,6-di-H-PSQ vs. 3,6-di-NH2-PSQ
substituted Cr-PSQ’s.

228

 

”-7." .a- 4 '-
-1

In the case of the Ni"-PSQ and phenoxy radicals, it was proposed that there were
interactions with higher lying orbitals that produced outliers in the correlations with the
o—HOMO-a-LUMO gap. By comparison, the lack of outliers in the Cr-PSQ case gives
evidence to support the previous arguments that higher lying orbitals must have been
involved in the deviations in the Ni-PSQ and phenoxy complexes, since all substituents
correlate very strongly in the Cr-PSQ case.

The correlations with the B-HOMO-B—LUMO are analogous with the Ni-PSQ
case: electron withdrawing groups increase the B-HOMO-B-LUMO gap, while electron
donors decrease the B-HOMO-B-LUMO gap. In comparison to Cr-PSQ, a strong
acceptor electron like NO; in Cr-3,6-di—NO;—PSQ (Figure 5-3 7) will lower the energy of
the PSQ-SOMO, thereby increasing the B-HOMO-B-LUMO gap due to decreased orbital
mixing. An electron donor like NH; will have the opposite effect (Figure 5-38) and a
better energetic match between the PSQ-SOMO and the CrI"-3dxz orbitals will occur,
with the B-HOMO-B-LUMO gap decreasing. Normally one would expect that a stronger
orbital interaction would decrease the gap, but due to the energetic ordering of the SOMO
orbitals this effect is not observed in these systems. Specifically, the B-HOMO
(-l 1.03 eV) of Cr-PSQ is much lower in energy than the B-HOMO of Cr-3,6-di-NH;-PSQ
(-9.73 eV). This enlarges the gap due to the weaker orbital interactions more than the
increased mixing seen in Cr-3,6-di-NH;-PSQ.

As before in the Ni-PSQ complexes, the molecular orbital contributions of the
Crm-center and the average oxygen contributions were determined for all complexes, and
are depicted in Table 5-13. The average % 0 again correlates quite strongly with the spin

density delocalization (r = 0.96), and the donors have smaller % 0 contributions to the B-

229

HOMO which reflect the increasing mixing between the PSQ-SOMO and the Cr—3dxz

orbitals.

Table 5-13. The % chromium (% Cr) and average % oxygen (% O) in the highest occupied B-HOMO, of

substituted [Cr(n'en)(3,6-di-R-PSQ)]2+ complexes and the average B-spin density at the oxygen atoms
(p(O) in B-electron units).

 

 

Complex % Cra % 0a p (O)

[Cr(tren)(3,6-di-NMe;-PSQ)]2+ 2.34 4.12 -00404
[Cr(tren)(3,6-di-NH;-PSQ)]2+ 2.52 4.87 -00530
[Cr(tren)(3,6-di-0Me--PSQ)]2+ 2.54 5.60 -0.0690
[Cr(tren)(3,6-di-vinyl-PSQ)]2+ 2.31 5.72 -00743
[Cr(tren)(3,6-di-13r-PSQ)]2+ 2.07 5.94 -0.0971
[Cr(tren)(3,6-di—r-butyl-PSQ)]2+ 2.75 7.19 -00903
[Cr(tren)(3,6-di-TMs-PSQ)]2+ 2.73 7.28 -00932
[Cr(tren)(3,6-di-Cl-PSQ)]2+ 2.26 7.35 -0.1015
[Cr(tren)(3,6-di-CH3-PSQ)]2+ 2.77 7.59 -0.0983
[Cr(tren)(3,6-di-F-1>SQ)]2+ 2.62 8.04 -0.1043
[Cr(tren)(3,6-PSQ)]2+ 2.85 8.79 -0.1060
[Cr(tren)(3,6-di-CN-PSQ)]2+ 2.35 8.13 -0.1131
[Cr(tren)(3,6-di-C(CF3)3-PSQ)]2+ 2.60 9.06 -01195
[Cr(tren)(3,6-di-CF3-PSQ)]2+ 2.64 9.53 -01225
[Cr(tren)(3,6-di-NO;-PSQ)]2+ 2.51 9.61 -0.1260
[Cr(tren)(3,6-di-IF4-PSQ)]2+ 2.56 9.74 -0.1272
[Cr(tren)(3,6-di-SO;CN-PSQ)]2+ 2.40 9.86 -01341

 

a 2 . . .
[Enzaa / Enzmflj * 100 = % contribution. in a.o. 18 the sum of the squares of the atomic orbltal

. . 2 . . .
coefficrents of the atom or group of 1nterest, and Zn mo. 15 the sum of the squares of all atomic orbltal
coefficients in a specific molecular orbital.

The electron acceptors increase the % 0 character in the B-HOMO, since these
substituents increase the spin density at the oxygen atoms. Thus the % 0 data supports
the MO diagrams that have been proposed in Figures 5-37 and 5-38. The % Cr data is
more difficult to interpret and the changes are quite small, but a subtle correlation can be
suggested. If we consider the N02, H, and NH; substituted complexes for our

comparisons, and by examining Figures 5-37 and 5-38 we can expect the % Cr for the

230

NO; and NH; complexes to decrease in comparison with the H substituent. For the NO;
complex the decreased % Cr is due to a decreased energetic match between the B-HOMO
and Cr—3dxz orbitals, because of the strongly electron withdrawing ability of NO;. The
NH; complex will have increased mixing between the B-HOMO and Cr-3dxz and as a
result the % Cr will decrease.

The Cr—PSQ data and correlations were also tested for functional dependence.
The UBLYP results indicated identical correlations but different numbers. The Cr—O
bond length was also fixed (1.94 A) to examine any effects of the bond distance on the

exchange coupling constant, and although the numbers were different the exact same
trends were seen. Overall, the DFT data for the [Cr(tren)(3,6-R-PSQ)]2+ complexes have

corroborated our hypothesis, and Equation 22 seems to be valid in these systems. The J
value can be modulated with substituents for both the ferromagnetic and
antiferromagnetic exchange coupling regimes, and the spin density at the oxygen atoms
of the PSQ ligand is the main variable responsible for the changes in J. In an attempt to
see if any experimental J values could be determined, the synthesis of Cr-3,6-di-NO;-
PSQ, Cr-3,6-di-NH;-PSQ, and Cr-PSQ complexes will be pursued and if we are

successful SQUID measurements will been taken.

5.3.5 Synthesis and Characterization of Cr-3,6-di-R—PSQ complexes
Synthesis. The synthesis of the Cr-PSQ complexes was done in a Similar fashion

as the Ni-PSQ complexes, and based on other literature reports for the synthesis of
[Cr(tren)(3,6-di-t4butylsemiquinone)]2+ and [Cr(tren)(3,6--di-t-butylcatechol)]2+ refluxing

conditions are needed in the reactions.94 In a typical reaction, the 3,6-di-R-PCATH;

231

 

ligands were deprotonated in the dry box with NaOH and then cannula transferred to a
flask equipped with a reflux condenser containing degassed [Cr(tren)C1;]Cl under
nitrogen. The solution was then refluxed for 24 hours under N2, and subsequently filtered
under N; into a flask containing excess salt (NaBPh4, NaBF4, etc.) to precipitate out the
desired Cr-PCAT product. The Cr-PCAT compounds were then oxidized under N; with
acetylferrocenium tetrafluoroborate to the corresponding semiquinone complexes. The
compounds should be stored under vacuum or nitrogen until needed. Unfortunately, the
[Cr(tren)(3,6-di-NH;-PCAT)](BPh4) complex could not be made successfully. Part of the
problem with the synthesis was the formation of Cr(3,6-di-NH;-PSQ)3, which was
confn'med by elemental analysis. The other two phenanthrenesemiquinone complexes
were successfully prepared, and their properties will be discussed below.

Ground State Characterization: X-ray Crystallography. X-ray structures

were obtained for the [Cr(tren)(PCAT)]+ (7) and [Cr(tren)(PSQ)]2+ (8) cations and

ORTEP drawing are depicted in Figure 5-39.

 

Figure 5-39. Drawings of the cations of [Cr(tren)(PCAT)]+ (7) (left) and [Cr(tren)(PSQ)]2+ (8) (right)
obtained from single crystal X-ray structure determinations. Atoms are represented as 50 % probability
thermal ellipsoids.

232

The crystallographic data are listed in Table 5-14 and selected bond distances and angles

are listed in Table 5-15. The C—0 and intradiol (C(10)—C(9)) bond lengths are again
diagnostic of the semiquinone oxidation state, where the Cr-PSQ C--Oavg = 1.31 A and
C(10)—C(9)) = 1.40 A bond distances agree well with literature values.55 The Cr-PCAT
C--Oavg = 1.36 A and C(10)—C(9)) = 1.36 A bond distances are also diagnostic of the

catechol oxidation state of the ligand.55 The semiquinone exhibits altemating short and
long C—C distances because of the more localized nature of the double bonds in the rings

and the catechol exhibits more delocalized C—C bond distances due to its aromaticity.75'77

Table 5-14. Crystallographic data for [Cr(tren)(PCAT)](BPh4) (7) and [Cr(tren)(PSQ)](BPh4)2 (8).

 

 

 

7 8
formula C44H46BCI‘N402 C69H70B2C1‘N4O3
MW 725.66 1076.91
cryst syst Orthorhombic Orthorhombic
space group Pbca Pna2(1)
a/A 10.2323 (12) 31.165 (3)
b/ A 17.838 (2) 10.6549 (12)
c/ A 39.772 (5) 17.2766 (19)
B/ ° 90 90
V/ A3 7259.4 (15) 5736.8 (11)
Z 8 4
T/K 173(2) 173(2)
DJ g cm'3 1.328 1.247
Omax 25 25
reflns measured 68045 54464
independent reflns 63 84 10085
observed reflns [I > 20(1)] 4159 8397
11(Mo K01)/ mm‘1 0.36 0.25
Rim 0.095 0.065
R1 0.049“ 0.047
wR2 0.13” 0.1
GOF 0.99 1.02

 

F
“ R1 = ZIIFoI — 113.3117 zlrol. wR2 = [Zw(l=o2 — Fc2)/ZW(F02)2] ”2, w = 1/102(F02) + (an
+ bP], where P = (F02 + 2Fc2)/3.

233

Table 5-15. Selected bond distances and angles for [Cr(tren)(PCAT)](BPh4) (7) and

[Cr(tren)(PSQ)l(BPh4)2 (8)-

 

 

7 8
Bond Distances (A)
Cr(1)-O(1) 1.914(2) 1.9223(18)
Cr(1)-O(2) 1 936(2) 1 .961 (2)
Cr(1)—N(1) 2.122(3) 2.075(2)
Cr(1)—N(2) 2.085(3) 2.065(3)
Cr(1)—N(3) 2.071(3) 2.074(2)
Cr(1)—N(4) 2.079(3) 2.057(2)
C(10)—O(1) 1.349(4) 1.303(3)
C(9)—O(2) 1.366(3) 1.31 1(3)
C(9)—C(10) 1.359(4) 1 .400(4)
C(9)—C(l4) l .43 6(4) 1 .429(4)
C(14)—C(13) 1.429(4) 1.424(4)
C(13)—C(12) 1.449(4) 1.450(5)
C(12)—C(1 1) 1.425(4) 1.41 1(4)
C(1 1)_c(10) 1.435(4) 1.438(4)
Bond Angles (deg)

O(1)—Cr(1)—N(1) 97.74(10) 100.16(9)
O(1)—Cr(1)—N(2) 9540(9) 9690(9)
0(1)—Cr(1)—N(3) 97.97(10) 9385(9)
O(1)—Cr(1)—N(4) 177.44(10) 176.14(9)
O(1)—Cr(1)—O(2) 8605(9) 8224(8)
O(2)—Cr(1)-N( 1) 85.03(10) 87.51(10)
O(2)—Cr(1)—N(2) 177.85(10) 178.10(10)
O(2)—Cr(1)—N(3) 89.26(10) 88.29( 10)
O(2)—Cr(1)—N(4) 9638(9) 97.19(9)
N(1)—Cr(1)—N(2) 93. 19(1 1) 90.98(1 l)
N(1)—Cr(1)—N(3) 162.86(1 1) 164.69(1 1)
N(1)—Cr(1)—N(4) 81 .70(1 1) 83.61(10)
N(2)-Cr(1)—N(3) 92.1 1(1 1) 93.46(1 1)
N(2)—Cr(1)—N(4) 82.15(10) 8378(9)
N(3)—Cr(1)—N(4) 82.90(10) 82.31(10)
C(9)—O(2)—Cr(1) 108.72(17) l 1 1 .84(18)
C(10)—O(1)—Cr(1) 109.43(18) 113.27(17)
Q(1)—C(10)—C(9) 118.5(3) 116.3(3)

 

Electronic Absorption Spectroscopy. The normalized absorption spectra of the

Cr-PSQ (8), Cr-3,6-di-NO;-PSQ (10), and Ga-PSQ (12) taken in acetonitrile are shown in

234

Figure 5-40 below. The chromium spectra are quite rich, and the absorption features are
listed in Table 5-15 and compared to 12 to aid in the assignments. The transitions at
22300, 23300, 28300, and 30100 cm'l for 8 are due to intraligand transitions upon
comparison to the Ga-PSQ analogue. Analogously, the absorptions at 23300, 24900, and
26950 cm'1 for 10 are also intraligand transitions. The transitions at 21000 and (18300
and 21150) cm'1 are assigned to charge transfer transitions for 8 and 10, respectively,
since these features are not observed in the Ga-PSQ complex for which charge transfer
transitions are highly unlikely. The sharp lower energy bands in both chromium

complexes (15000-16500 cm") are characteristic of intraconfigurational transitions, and

their location in the spectrum is consistent with transitions having 4A;g —> 2Eg and/or 2T1g

 

 

 

 

 

 

parentage.
1.4
1.2 1-
f
8 1 0 _
g o
e - .
g 0.8 -
< .
“U
fé 0.6 -
7g 1
o 0.4 1-
Z .
0.2 -
0.0 f .
35000 25000 20000 15000

Wavenumber (cm' )

Figure 5-40. The normalized absorption spectra of [Cr(tren)(PSQ)](BPh4)2 (8) (red), [Cr(tren)(3,6-di-
N02-PSQ)1(BF4)2 (10) (green), and [Ga(tren)(PSQ)l(BPh4)2 (blue) (12)-

235

 

Table 5-15. Comparison of the absorption features in [Cr(tren)(PSQ)](BPh4)2 (8), [Cr(tren)(3,6—di-N02-

PSQ)](BF4)2 (10), and [Ga(tren)(PSQ)](BPh4)2 (blue) (12), where LMCT = ligand-to-metal charge transfer
transition.

LGa(tren)QSQ)l@Ph4)2 [Cr(tren)(PSQ)](BPhth [Cr(tren)(3.6-di-N02-PSQ)](BPh4)2

 

 

15000 (d-d) 15150 (d-d)
15600 (d-d) 15700 (d-d)
16400 (d-d) 16450 (d—d)

21000 (LMCT) 18300 (LMCT)

22400 (IL) 22300 (IL) 21150 (LMCT)
24100 (IL) 23900 (IL) 23300 (IL)
29500 (IL) 28300 (IL) 24900 (IL)
31800 (IL) 30100 (IL) 26950 (IL)

 

Their origin is due to exchange-enhanced ligand-field absorptions of the Cr"I ion.'9’56’94

Heisenberg exchange coupling in the excited state manifold of these complexes gives rise
to the formation of new electronic states with different spin, and as a result formally spin
forbidden transitions become spin allowed transitions when exchange coupling is
operative. McCusker and coworker356’94 have laid out a detailed analysis of the effect of
Heisenberg exchange coupling on the electronic structure of these molecules, and this
will not be further elaborated upon here.

Magnetic Susceptibility. Plots of the temperature dependent magnetic

susceptibility for 8 and 110 are depicted in figure 5-41 below. The data in Figure 5-41

indicate that lief-f = 2.81 113 for 8 and neff = 2.98 113 for 10 at 300 K, which agrees with the

spin only value for an S = 1 spin state of 11m = 2.87 113, thus confirming the ground state
of the molecule is S = 1. There is a slight rise in the data over the temperature range and
according to the similar [Cr(tren)(3,5-di-t-butylsemiquinone)](PF6); complex in the
literature,94 this can be assigned to temperature independent paramagnetism in the

sample, which is manifested as a T”2 dependence in the data. Just as in the Ni-PSQ

236

complexes, the exchange coupling is strong and J can not be determined. This iS not too
surprising, since the DFT calculations actually predict that the coupling constants should
be at least 2 times greater in magnitude then that of the Ni-PSQ systems. The predicted
increase is not surprising either, since the magnetic orbitals actually overlap in the case of
the Cr-PSQ complexes and should yield a greater interaction than the repulsion-type
interaction observed for ferromagnetism. It is evident then that to observe temperature
dependent magnetic data in a series of these types of complexes a new radical ligand

system must be designed that greatly reduces the spin density at the oxygen centers.

4.0

 

3.5 -

 

1.0 - I [Cr(tren)(PSQ)](BPh4)2
- 1 [Cr(tren)(3,6-N02-PSQ)(BF4)2

Effective Magnetic Moment (BM)

 

 

 

00 r L 1 L 1‘1 1 I 1444+L1

0 50 100 150 200 250 300 350
Temperature (K)

Figure 5-41. The temperature dependent SQUID susceptibility of [Cr(tren)(PSQ)](BPh4)2 (8) (red square)
and [Cr(tren)(3,6-di-N02-PSQ)](BF4)2 (10) (green triangle).

5.3.6 Concluding Comments

237

 

“ «er 1':

Several salient relationships have been developed in this chapter, and most
importantly the proper choice of substituents can modulate the J value for both
ferromagnetic and antiferromagnetic complexes. Electron donors and acceptors were
found to decrease and increase the J value, respectively. Linear correlations were found
between J and the average spin density at the oxygen centers in both the Cr'"- and Ni"-
PSQ complexes, and the spin population at oxygen dictates the magnitude of the
Heisenberg exchange coupling constant. The proportionality, J ~ pMpSQ, has been found
to be valid for both ferromagnetic and antiferromagnetic exchange coupling. The spin

density also varies with opposite polarization to the charge density for all substituents
examined, and the Hammett constants GP and op+ correlate linearly with the exchange

coupling constant. The a-HOMO-a-LUMO gap in the NiH-PSQ’S and the B-HOMO-B-
LUMO gap in the Crm-PSQ complexes has been identified as a marker for the spin
delocalization in the molecules. Where decreases in the HOMO-LUMO gap mirror
increases in spin delocalization, increases in the HOMO-LUMO gap will reflect
increased spin localization at the oxygen atoms. The amount of mixing between the M-
3dxz orbitals and the PSQ-SOMO is the main factor that controls the HOMO-LUMO
gap. Two main factors have been identified to describe the mixing. First, electron
acceptors will lower the energy of the PSQ-SOMO, thereby causing a decreased
energetic match between the M-3dxz and the PSQ-SOMO orbitals, resulting in less spin
delocalization. Secondly, electron donors will raise the energy of the PSQ-SOMO, which
will result in greater spin delocalization due to the better energetic match between the M-
3dxz and the PSQ-SOMO orbitals. There were outliers observed in the Nin-PSQ spin

density vs. a-HOMO-a-LUMO gap correlation, and interactions with energetically higher

238

 

 

"."Ll'y.£..

lying orbitals were suggested as being the cause of the deviations. This is similar to what
has been observed in the phenoxy radical complexes. The Crm-PSQ spin density
correlation with the B-HOMO-B-LUMO gap does not exhibit any deviations, because
only two orbitals are interacting to produce the B-HOMO-B-LUMO gap. This piece of
evidence gives strong support for the removal of the outliers in the Nin-PSQ and phenoxy
radical a-HOMO-a-LUMO gap correlations, since the substituents correlate with the
HOMO-LUMO gap if no more than two orbitals are involved in the orbital interaction.
Unfortunately, experimental evidence for an entire series of complexes could not
be found in this examination, which is a direct result of the strong ferromagnetic and
antiferromagnetic exchange coupling constants precluding us from seeing temperature
dependence in the magnetic susceptibility data. Temperature dependence was observed
in the [Ni(tren)(3,6-di-NH;-PSQ)]+ (6) complex, and J was determined to be —1 19 cm".
DFT had predicted that the spin density at oxygen and thus the J value should be smaller
in the case of the 6 versus the [Ni(tren)(PSQ)]+ (1) or [Ni(tren)(3,6-di-NO;-PSQ)]+ (5)
complexes. The DFT data is supported in our SQUID data because we were able to fit
the susceptibility data to extract a J value in 6, but 1 and 5 have too strong a coupling
constant (not allowing experimental determination of J). These are the only two
experimental points that could be determined from the NiII-PSQ examinations, and to
properly describe the correlation between J and the spin density at oxygen at least three
complexes with experimentally determined J values are required. The experimental
SQUID data for the Crm-PSQ complexes also indicated that J is too strong to be
experimentally determined. Therefore due to the lack of experimental data to confirm the

relationship between the magnitude of J and the spin density, a new easily substituted

239

radical ligand system must be designed to decrease the spin density at the oxygen atoms
even greater than in the PSQ radicals. Possible pathways to continue this work will be

explored in the future work section of Chapter 6.

240

5.5 References

l

2

10

11

12

13

14

15

16

17

18

19

2O

Luneau, D. Curr. Opin. Solid State Mater. Sci. 2001, 5, 123.

Ohrstrom, L. C. R. Chimie 2005, 8, 1374.

Shultz, D. A. Polyhedron 2001, 20, 1627.

Caneschi, A.; Gatteschi, D.; Sessoli, R.; Rey, P. Acc. Chem. Res. 1989, 22, 392.
Vostrikova, K. E. Coord. Chem. Rev. 2008, 252, 1409.

Kahn, C.; Martinez, C. J. Science 1998, 279, 44.

Mishustin, A. 1. Russ. J. Inorg. Chem. 2008, 53, 1376.

Osanai, K.; Okazawa, A.; Nogami, T.; Ishida, T. J. Am. Chem. Soc. 2006, 128,
14008.

McConnell, H., M, J. Chem. Phys. 1963, 39, 1910.

Hicks, R. G.; Lemaire, M. T.; Thompson, L. K.; Barclay, T. M. J. Am. Chem. Soc.
2000, 122, 8077.

Sugiura, K.-I.; Arif, A. M.; Rittenberg, D. K.; Schweizer, J.; Ohrstrom, L.,
Epstein, A. J.; Miller, J. S. Chem. Eur. J. 1997, 3, 138.

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

Shultz, D. A.; Sloop, J. C.; Coote, T.—A.; Beikmohammadi, M.; Kampf, J .; Boyle,
P. D. Inorg. Chem. 2007, 46, 273.

Gordon, D. J.; Fenske, R. F. Inorg. Chem. 1982, 21, 2907.

Ginsberg, A. P. Inorg. Chim. Acta. Rev. 1971, 5, 45.

Kahn, 0.; Charlot, M. F. Nouv. J. Chim. 1980, 4, 567.

Benelli, C.; Dei, A.; Gatteschi, D.; Pardi, L. Inorg. Chem. 1988, 2 7, 2831.
Benelli, C.; Dei, A.; Gatteschi, D.; Pardi, L. Inorg. Chem. 1989, 28, 1476.

Benelli, C.; Dei, A.; Gatteschi, D.; Giidel, H. U.; Pardi, L. Inorg. Chem. 1989, 28,
3091.

Ciofini, 1.; Daul, C. A. Coord. Chem. Rev. 2003, 238-239, 187.

241

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

Noodleman, L. J. Chem. Phys. 1981, 74, 5737.
Noodleman, L.; Case, D. A. Adv. Inorg. Chem. 1992, 38, 423.
Noodleman, L.; Norman Jr., J. G. J. Chem. Phys. 1979, 70, 4903.

Chlopek, K.; Muresan, N.; Neese, F.; Wieghardt, K. Chem. Eur. J. 2007, I3,
8390.

Chlopek, K.; Bothe, E.; Neese, F .; Weyherml'iller, T.; Wieghardt, K. Inorg. Chem.
2006, 45, 6298.

Ghosh, P.; Bill, E.; Weyermfieller, T.; Neese, F .; Wieghardt, K. J. Am. Chem. Soc.
2003, 125, 1293.

Blanchard, S.; Neese, F.; Bothe, E.; Bill, E.; Weyermiieller, T.; Wieghardt, K.
Inorg. Chem. 2005, 44, 3636.

Herebian, D.; Wieghardt, K. E.; Neese, F. J. Am. Chem. Soc. 2003, 125, 10997.

Bencini, A.; Daul, C. A.; Dei, A.; Mariotti, F.; Lee, H.; Shultz, D. A.; Sorace, L.
Inorg. Chem. 2001, 40, 1582.

Ovcharenko, V. I.; Gorelik, E. V.; Fokin, S. V.; Romanenko, G. V.; Ikorskii, V.
N.; Krashilina, A. V.; Cherkasov, V. K.; Abakumov, G. A. J. Am. Chem. Soc.
2007, 129, 10512.

Bencini, A.; Carbonera, C.; Dei, A.; Vaz, M. G. F. Dalton Trans. 2003, 1701.

Rodriguez, J. H.; Wheeler, D. E.; McCusker, J. K. J. Am. Chem. Soc. 1998, 120,
12051.

Kapre, R. R.; Bothe, E.; Weyermiieller, T.; George, S. D.; Muresan, N.;
Wieghardt, K. Inorg. Chem. 2007, 46, 7827.

F ehir Jr., R. J. PhD, Michigan State University, 2009. ‘
Zipp, S. G.; Madan, S. K. Inorg. Chem. 1976, 15, 587.

Allcock, H. R.; Primrose, A. P.; Silverberg, E. N.; Visscher, K. B.; Rheingold, A.
L.; Guzei, I. A.; Parvez, M. Chem. Mater. 2000, 12, 2530.

Anschfitz, R.; Meyer, P. Chem. Ber. 1885, 18, 1942.

Aggarwal, R. P.; Connelly, N. G.; Crespo, M. C.; Dunne, B. J.; Hopkins, P. M.;
Orpen, A. G. J. Chem. Soc. Dalton Trans. 1992, 655.

242

 

39

4O

41

42

43

44

45

46

47

48

49

50

51

52

Cotton, F. A.; Jordan IV, G. T.; Murillo, C. A.; Su, J. Polyhedron 1997, 16, 1831.
Vierling, P.; Guillon, C. J. Organomet. Chem. 1994, 464, C42.

Hughes, R. P.; Lindner, D. C.; Rheingold, A. L.; Yap, G. P. A. Inorg. Chem.
1997, 36, 1726.

Hayashi, Y.; Rohde, J. J.; Corey, E. J. J. Am. Chem. Soc. 1996, 118, 5502.
Perkins, C. M. PhD Thesis, University of Washington, 1980.

SAINT, ver 6.02a; Bruker AXS, Inc.: Madison, WI, 2000.

Sheldrick, G. M.SADABS, ver 2.03; Bruker AXS, Inc.: Madison, WI, 2000.
Sheldrick, G. M.SHELXT L, ver 6.12; Bruker AXS, Inc.: Madison, WI, 2001.
Lbliger, J .; Scheffold, R. J. Chem. Educ. 1972, 49, 646.

Adams, D. M.; Hendrickson, D. N. J. Am. Chem. Soc. 1996, 118, 11515.

Adams, D. M.; Noodleman, L.; Hendrickson, D. N. Inorg. Chem. 1997, 36, 3966.
Gauss View, 2.1 ed.; Gaussian, Inc.: Pittsburgh, PA, 2001.

Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann, R. E.;
Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain,
M. C.; Farkas, 0.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.;
Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J .; Petersson, G. A.; Ayala, P. Y.;
Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.;
Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko,
A.; Piskorz, P.; Komaromi, 1.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.;
Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.;
Gill, P. M. W.; Johnson, 8.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.;
Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.4,
Gaussian, Inc.: Pittsburgh, PA, 1998.

M. J. Frisch, G. W. T., H. B. Schlegel, G. E. Scuseria; M. A. Robb, J. R. C., J. A.
Montgomery, Jr., T. Vreven; K. N. Kudin, J. C. B., J. M. Millam, S. S. Iyengar, J.
Tomasi; V. Barone, B. M., M. Cossi, G. Scalmani, N. Rega; G. A. Petersson, H.
N., M. Hada, M. Ehara, K. Toyota; R. Fukuda, J. H., M. Ishida, T. Nakajima, Y.
Honda, 0. Kitao; H. Nakai, M. K., X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross;
V. Bakken, C. A., J. Jaramillo, R. Gomperts, R. E. Stratmann; O. Yazyev, A. J.
A., R. Cammi, C. Pomelli, J. W. Ochterski; P. Y. Ayala, K. M., G. A. Voth, P.
Salvador, J. J. Dannenberg; V. G. Zakrzewski, S. D., A. D. Daniels, M. C. Strain;

243

 

 

‘1‘

,_'___._ f

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

O. Farkas, D. K. M., A. D. Rabuck, K. Raghavachari; J. B. Foresman, J. V. 0., Q.
Cui, A. G. Baboul, S. Clifford; J. Cioslowski, B. B. S., G. Liu, A. Liashenko, P.
Piskorz; I. Komaromi, R. L. M., D. J. Fox, T. Keith, M. A. Al-Laham; C. Y. Peng,
A. N., M. Challacombe, P. M. W. Gill; B. Johnson, W. C., M. W. Wong, C.
Gonzalez, and J. A. Pople, Gaussian 03, Revision D.01; Gaussian, Inc.:
Wallingford CT, 2004.

Glendening, E. D.; Reed, A. E.; Carpenter, J. E.; Weinhold, F. N80 3.],
Theoretical Chemistry Institute, University of Wisconsin: Madison, WI, 1996.

Frisch, M. J .; Frisch, A. Gaussian 98 User 's Reference; Gaussian, Inc.: Pittsburgh,
1998.

Zanello, P.; Corsini, M. Coord. Chem. Rev. 2006, 250, 2000, and references
therein.

Guo, D.; McCusker, J. K. Inorg. Chem. 2007, 46, 3257.
Evans, D. F. J. Chem. Soc. 1959, 2003.
Grant, D. H. J. Chem. Educ. 1995, 72, 39.

Brown, D. 3.; Crawford, V. H.; Hall, J. W.; Hatfield, W. E. J. Phys. Chem. 1977,
81, 1303.

CRC Handbook of Chemistry and Physics; 89 ed.; CRC Press: Boca Raton, FL.,
2009.

Yamaguchi, K.; Jensen, F.; Dorigo, A.; Houk, K. Chem Phys. Lett. 1988, 149,
537.

Soda, T.; Kitagawa, Y.; Onishi, T.; Takano, Y.; Shigeta, Y.; Nagao, H.; Yoshioka,
Y.; Yamaguchi, K. Chem Phys. Lett. 2000, 319, 223, and references therein.

Bencini, A. Inorg. Chim. Acta. 2008, 361, 3820.
Rodriguez, J. H.; McCusker, J. K. J. Chem. Phys. 2002, 116, 6253.
Ruiz, E.; Cano, J .; Alvarez, S.; Alemany, P. J. Comp. Chem. 1999, 20, 1391.

Caballol, R.; Castell, 0.; Illas, F.; de P. R. Moreira, 1.; Malrieu, J. P. J. Phys.
Chem. A 1997, 101, 7860.

Kahn, 0.; Prins, R.; Reedijk, J.; Thompson, J. S. Inorg. Chem. 1987, 26, 3557.

Kahn, 0. Molecular Magnetism; VCH Publishers: New York, 1993.

244

 

69

70

71

72

73

74

75

76

77

78

79

8O

81

82

83

84

85

86

87

88

Krenske, E. H.; Coote, M. L. J. Phys. Chem. A 2007, 111, 8229.

Creary, X. J. Org. Chem. 1980, 45, 280. '

Zhang, Y. Q.; Luo, C. L. J. Mater. Chem. 2006, 16,4657.

Zhang, Y. Q.; Luo, C. L. Eur. J. Inorg. Chem. 2006, 11, 2292.

Zhang, Y. Q.; Luo, C. L. J. Phys. Chem. A 2006, 110, 5096.

Schenck, G. 0.; Schmidt-Thombe'e, G. A. Ann. Chem. 1953, 584, 199.
Pierpont, C. G.; Buchanan, R. M. Coord. Chem. Rev. 1981, 38, 45.
Pierpont, C. G.; Attia, A. S. Collect. Czech. Chem. Commun. 2001, 66, 33.
Pierpont, C. G.; Lange, C. W. Prog. Inorg. Chem. 1994, 41, 331.

Olson, D. C.; Vasilevskij, J. Inorg. Chem. 1969, 8, 1611.

Nag, K.; Chakravorty, A. Coord. Chem. Rev. 1980, 33, 87.

Jovanovic, S. V. J. Phys. Chem. 1991, 95, 10824.

Bordwell, F. G.; Cheng, J.-P. J. Am. Chem. Soc. 1991, 113, 1736.

Lind, J .; Shen, X.; Eriksen, T. E.; Mere'nyi, G. J. Am. Chem. Soc. 1990, 112, 479.

Bencini, A.; Gatteschi, D. EPR of Exchange Coupled Systems; Springer-Verlag:
Berlin, Heidelberg, 1990.

XSOPHE, Computer Simulation Software Suite 1.1.4; Center for Magnetic
Resonance and Department of Mathmatics, The University of Queensland
(Brisbane, Queensland, Australia 4072) and Bruker Biospin GMBH (formerly
Buruker Analytik; Rheinstetten, Germany): 1993-2004.

Mueller, J .; Kikuchi, A.; Bill, E.; Weyermueller, T.; Hildebrandt, P.; Ould-
Moussa, L.; Wieghardt, K. Inorganica Chimica Acta 2000, 297, 265.

Adam, W.; Barneveld, C. V. J. Am. Chem. Soc. 1999, 121, 10820.

Adam, W.; Kita, F.; Harrer, H. M.; Nau, W. M.; Zipf, R. J. Org. Chem. 1996, 61,
7056.

Adam, W.; Schulte, C. M. O. J. Org. Chem. 2002, 67, 4569.

245

89

90

91

92

93

94

Schmitt, E. A., Department Of Chemistry, University of Illinois, 1996.

Barclay, T. M.; Hicks, R. G.; Lemaire, M. T.; Thompson, L. K. Chem. Commun.
2000, 2141.

Brown, H. C.; Okamoto, Y. J. Am. Chem. Soc. 1958, 80, 4979.
Hansch, C.; Leo, A.; Taft, R. W. Chem. Rev. 1991, 91, 165.
Liu, C.; Hu, H.; Yang, X. Chem. Phys. Lett. 2001, 349, 89.

Wheeler, D. E.; McCusker, J. K. Inorg. Chem. 1998, 37, 2296.

246

Chapter 6. Concluding Comments and Future Directions

6.1 Concluding Comments

All of the research in this dissertation has been focused on using substituents to
develop a prescription for the movement of spin and charge density in radicals and
transition metal radical complexes. Based on the reasonable assumption that one of the
major variables responsible for the modulation of J was the spin density population at the
oxygen magnetic centers, it was the overall goal of this work to develop a theoretical
model for the substituent modulation of the exchange coupling constant based on DFT
calculations. Then experimental magnetic susceptibility measurements of transition
metal radical complexes would be done to verify this theory. The majority of the work
was split up into two fronts, one computational and one experimental; the major
successes of each realm of the project will be elaborated upon below.

To begin this examination, it was necessary to understand the fundamental
polarization of spin and charge density in simple systems. It is known that spin and
charge can be differentially polarized, but little is known about the mechanism which
controls spin and charge density in a molecule. The system chosen was the phenoxy
radical, since it is the simplest oxygen center radical that could be used to simulate the
semiquinone radicals that we eventually wanted to study. Correlations of the charge
density with the Hammett substituent constants were discovered, and easily rationalized
based upon simple electronegativity arguments. Similar correlations between the spin
density and the Hammett Spin delocalization substituent constants produced poor results,

so therefore other correlations were sought after. A completely empirical correlation

247

between the a-HOMO-a-LUMO gap and the change in spin density at the oxygen atoms
was found. This correlation can be considered to be a marker which describes the spin
delocalization in these systems. A three-electron, two-orbital model was evoked to
explain the a-HOMO-a-LUMO gap correlation, where the energy of the a-LUMO orbital
was assumed to be largely unchanged in each complex. The energy of the (ll-HOMO
orbital was then influenced by the type of para-substituent in the phenoxy radical, and
electron-donating substituents increased the energy of the (ll-HOMO and electron
acceptors lowered the energy of the 111-HOMO. The mixing between a n—donor lone pair
orbital with the phenoxy SOMO was found to influence the energy of the (ll—HOMO. As
the donating character of a substituent was increased, a better energetic match between
the n—donor lone pair orbital and the phenoxy SOMO would occur. This has two
important consequences. First, the energy of the (ll-HOMO would increase due to an
increased mixing interaction between these orbitals, and the HOMO-LUMO gap would
decrease due to the higher energy (ll-HOMO (assuming the energy of the a-LUMO
remains constant). Secondly, the increased mixing between the orbitals also increases the
spin delocalization from the oxygen radical center to the substituent. The strong 71:-
accepting substituents were found to deviate from the a—HOMO-u-LUMO gap
correlation, and interactions with higher lying orbitals were shown to be the reason for
the outliers. Overall, the correlation is quite strong and can be used to predict which
substituents will increase or decrease spin density at a radical center. To my knowledge,
this was the first instance of a correlation between spin density at a radical center with the

a-HOMO-a-LUMO gap. The predictable ability to differentially polarize spin and

248

A“: a. _. M; was.

 

 

charge density in the phenoxy radical led us to see if the same correlations would hold for
substituted o-semiquinones.

The simplest o-l,2-semiquinones to examine are the 4,5-di-R-o-l,2-semiquinones
and the 3,6—di-R—o-l,2-semiquinones, and these systems were examined by the same DFT
methods as the phenoxy radicals. It was found that spin and charge density can be
differentially polarized, but to a lesser extent than the phenoxy radical case. Similar
correlations with the Hammett substituents constants were found, and the charge density
at the semiquinone oxygen atoms can be reliably estimated from these constants. In the
o-semiquinone complexes, it was found that electron acceptors decreased the spin density
and strong n-electron donors increased the spin density at the oxygen atoms. A paradigm
shift for the n—donor substituents was observed, and now they are spin localizing instead
of strongly delocalizing as in the case of the phenoxy radicals. Unfavorable resonance
interactions between the charge separated resonance structures and the delocalized
negative charge can be cited as the possible cause for this change. For the 3,6-di-R-o-
1,2-semiquinones the spin density at the oxygen atom correlates with the a-HOMO-a-
LUMO gap, and this correlation seems to be present for both neutral and anion radicals.
The 9,10-o-phenanthrenesemiquinones substituted at the 3,6- and 2,7-positions were also
examined, since their more delocalized structures may better shift the spin density away
from the oxygen atoms. This was indeed confirmed and the (ll-spin population at the
oxygen atoms in the unsubstituted o-l,2-semiquinone is 0.2634 electrons and in
phenanthrenesemiquinone it is 0.2570 electrons. Analogous to the other semiquinones
examined the phenanthrenesemiquinones charge density correlated with the Hammett

parameters, and the spin density correlated with the a-HOMO-a-LUMO gap. There were

249

some deviations in the HOMO-LUMO gap correlations just as in the phenoxy case, but
unfortunately the calculated molecular orbital information was inconclusive in aiding in
the determination of the mechanism for the outliers. Based upon the phenoxy radical
data, we can say to a first approximation that one or more higher lying orbitals are likely
involved in causing the observed deviations. The underlying message of the semiquinone
work was that the HOMO-LUMO gap was still a marker for the spin delocalization in
radicals, but more work is needed to understand the mechanism for spin polarization.
Once we understood the theoretical prescription for the spin density variations, we
wanted to see if this could be proved experimentally with EPR measurements.
Additionally, we were also interested in the effects on the spin density distribution upon
coordination of the radicals to a diamagnetic metal center. To this end, 2,7-di-NO; and

[II-complexes with these same radical

3,6-di-NO;-phenanthrenesemiquinones and Ga
ligands were synthesized and characterized by X-Band EPR and ESEEM spectroscopy as
well as DFT calculations. The 14N hyperfine coupling constants (HFC) gave us
experimental evidence for the amount of spin density delocalized onto the nitrogen atoms
of the NO; groups, and thus the spin delocalizing ability of these substituents at these
positions was known. The X—band EPR data and the DFT calculations of the 1H and 14N
hyperfine coupling constants for the two radical ligands were in excellent agreement, and
both predicted small spin density polarizations at the nitrogen atoms. The GaIII
complexes exhibited very different spin populations at nitrogen as determined by 14N
ESEEM spectroscopy. The small magnitude 14N HFC for the Ga-3,6-di-NO;-PSQ

complex agreed quite well with the DFT data, and also correlated with the free ligand

value. The magnitude of the MN HFC for Ga—2,7-di-NO;-PSQ was much greater than the

250

3,6-isomer. The ESEEM data suggested that the spin density at the nitrogen in the Ga-
2,7-di-NO;-PSQ complex was due to direct spin delocalization onto the nitrogen atom,
and the Ga—3,6—di-NO;-PSQ complexes spin population at nitrogen is due to spin
polarization interactions. DFT l4N HF C values were a factor of 7 times too small for the
Ga-2,7-di-NO;-PSQ complex. The failure of DF T in this case has been suggested to be
due to large self-interaction errors, and the transition metal electric field causes an

"1 complexes. This was

“overpolarization” of the spin density distribution in the Ga
confirmed by executing a DFT calculation of the free ligands in an electric field applied
along the axis of the oxygen atoms, and an identical spin polarization to the gallium cases
was observed. Two main points can be made about the above EPR data. First, spin
delocalization onto the nitrogen atoms at the 2,7-positons is more significant than the spin
polarization onto the 3,6-nitrogen atoms, and substituents at these positions will most
likely delocalize spin density to a greater extent. Secondly, DFT may give incorrect
predictions for the spin delocalization onto substituents at the 2,7-positions in these
systems. A firm mechanism for the spin delocalization/polarization mechanism could not
be determined from the experiments and more work is needed to hash it out. Thus the
EPR part of the project has opened the door to many possible new experiments that
would be required to accurately predict the Spin delocalization mechanism in these
systems, and this will be elaborated upon in the future work section below.

The final aspect of this thesis work involved applying the knowledge of spin
polarization that was learned in the above examinations in an attempt to predict how to

increase or decrease the ferromagnetic or antiferromagnetic Heisenberg exchange

coupling constants in transition metal radical complexes. The assumption that was made

251

was that the spin density at the oxygen atoms can be modulated with substituents to
control the magnitude of the exchange coupling constants. This idea is not novel and
Shultz and coworkers1 have predicted that J ~ pMng, but no complete examination of the
mechanism to vary J by changing the spin density with substituents in transition metal
semiquinones has been accomplished. The Ni"-2,7-di-R-PSQ and Ni"-3,6-di-R-PSQ
complexes were examined for substituents effects in the ferromagnetic Heisenberg
exchange coupling, and DF T has shown that the spin density correlated strongly with the
J value. Additionally, J was found to correlate with the product of the spin density on the
NiII center and the average spin density at the oxygen atoms, which fully supports the
proportionality suggested by Shultz. It is also noteworthy that the polarization of the spin
and charge density are completely differential in this case. The correlation with the 01-
HOMO-u-LUMO gap was again observed in the NiH-PSQ complexes, but several
strongly electron—withdrawing substituents deviated from the correlation. Again it was
suggested that participation of higher lying orbitals was responsible for the marked
deviations, but this could not be proved conclusively using only the Ni11 MO data. The
Crm-PSQ complexes were examined to see if the same correlations could be observed for
antiferromagnetic coupling, and indeed the J value correlated strongly with average spin
density at the oxygen centers and the product of the two spin densities as well. It seems
that the Shultz approximation is general for both ferromagnetic and antiferromagnetic
coupling, and the main variable that modulates J in these complexes is the Spin density at
the oxygen atoms. Interestingly, the spin density correlated very strongly with the [3-
HOMO-B-LUMO gap in this case, since the antiferromagnetic state requires B-spin

density on the oxygen atoms, the B-HOMO is the magnetic orbital that properly describes

252

"I-PSQ case, every Single substituent correlated with the

the exchange coupling. In the Cr
B-HOMO-B-LUMO gap and there were no outliers. A two-orbital, two-electron
interaction was proposed to describe the spin delocalization in these complexes, where
the CruI-3dxz and PSQ-SOMO magnetic orbitals underwent a bonding-type interaction.
This model only uses two molecular orbitals to generate the B-HOMO-B-LUMO gap, and
there can be no interaction with other orbitals that influences this orbital mixing. This
point gives additional evidence for the removal of the substituents that deviated in the
Nin-PSQ, o-semiquinone, and phenoxy radical complexes, since all substituents correlate
strongly in the Crm-PSQ case, and any other orbital interaction with the LUMO orbital is
the likely reason for the outlying substituents. Thus, the Crm-PSQ and NiH-PSQ DFT
data has indicated that the exchange coupling constant can be varied in a predictable
fashion by using the correct substituents.

The ability to predict spin polarization and to modulate the exchange coupling
constant can have far reaching applicability to aid chemists in designing molecular
magnetic materialsf“S and increasing the J value in single molecule magnets can increase
the temperature when the complex becomes a magnet.7 Additionally, the control of the
exchange coupling constant and spin polarization can also be important in the
development of O—O bond catalysts, since it is known that exchange coupling plays a
role the oxidation of alcohols to aldehydes coupled with the reduction of oxygen to
peroxide in the enzyme galactose oxidase.9’lO It is hoped that the basic research in this

thesis will be a stepping stone for the design of new promising molecular magnetic

materials and 0—0 bond forming catalysts.

253

6.2 Future directions

This thesis has examined the spin and charge density distribution in neutral and
anion radicals, continuing this trend it may be of fundamental interest to see how the spin
and charge density distributions are effected in cation radicals. Cation radicals are known
to be important in biological electron transfer reactions at the photosynthetic reaction

“’12 and for use in molecular magnetic conductors and superconductors.” It also

center
has been found that the proton dissociation enthalpies of thiophenoll4and phenol15 cation
radicals depends on the spin density population at the sulfur and oxygen centers,
respectively. It would then be advantageous to determine if the spin density
delocalization mechanism involves the a-HOMO-u—LUMO gap as was observed in the
phenoxy and semiquinone radicals to see if the observed correlations are general for
neutral, anion, and cation radicals.

Several EPR experiments should also be conducted in the dinitro-substituted Ga-PSQ
complexes and radical ligands to aid in the determination of the spin
delocalization/polarization mechanism. EPR or ESEEM of 17O (l = S/;) labeled PSQ
ligands would identify the spin density at the oxygen nucleus. If additional Gam-
complexes/radicals such as the unsubstituted and 3,6- and 2,7-di-NH;-PSQ were also

prepared, then the spin density could then be monitored for a substituent effect. Synthetic

procedures for 17O-labeled semiquinones have been published, where exchange with
H;l70 at 60 °C for 3 days produced the desired compounds.l6 If this simple procedure

does not work, an alternate would be an epoxidation reaction of phenanthrene in H;17O
followed by reaction with perchloric acid in l7O-labeled water to yield the

phenanthrenecatechol ligand (Scheme 6-1).”’18 Selective deuteration around the PSQ

254

ring would also be useful, since 2H (I = 1) can be identified in ESEEM experiments and
in determining HF C in X—band ENDOR19 (electron double nuclear resonance
spectroscopy) experiments. W-band (~95 GHz) EPR experiments may also prove to be

useful, depending on the results of the above X-band experiments.

 

 

 

 

[— '—1
170”
Ag20/12
THF/”OH;
_ J
17
”OH ”OH O
* hes/nae...) A

Scheme 6-1. Synthesis of '70 labeled phenanthrenecatechol.

The determination of the signs of all HFC would be highly desired to make a
convincing argument for the spin delocalization/polarization mechanism in these systems.
A combination of EPR, ENDOR, and TRIPLE20 (electron nuclear triple resonance)
spectroscopy is required to experimentally determine the signs of the HFC. In general,
TRIPLE experiments monitor the effect of simultaneous excitation of two nuclear spin
transitions on the level of the EPR absorption.2| From the characteristic intensity changes

of the high-frequency and low-frequency signals compared to those of the ENDOR

255

signals, the relative signs of the HF C’s can be determined?“21 The signs of the HFC’S
could then be compared to the DFT calculations and the delocalization/polarization
mechanism could be firmly explained.

DFT calculations in the Gaul-radicals have provided inconclusive results between
the different radical isomers, suggesting DFT’S inability to model the spin density
distribution in this particular system. Higher level calculations such as complete active
space (CASSCF) or coupled cluster methods CCSD(T) have been shown to qualitatively
determine the spin density for polyene radical chains.22 Additionally, the Pariser-Parr-
Pople semiempirical model yielded the correct spin distribution in the chain.22 Higher
level calculations or semiempirical methods such as these may be needed to accurately

am-PSQ complexes.

model the spin delocalization/polarization in these G

The NilI and Crm-PSQ complexes did suggest that spin density delocalization by
electron donors reduced the magnitude of the exchange coupling constant. Although,
more experimental points are needed to unequivocally prove the DFT results. The first
experiment to try would be 170 EPR of the series of Ni-PSQ compounds since the ”0
HF C would give an experimental spin density at oxygen that could be compared with
DFT calculations. To experimentally determine the exchange coupling constants a new
ligand system is required that can delocalize spin to a greater degree so that several J
values can be determined by SQUID magnetometry. A search of the literature to find
more extended n—systems to reduce the magnitude of the spin density at the oxygen atoms

was done, and it was found that when heterocycles are conjugated to phenoxy radicals23

or methylenecyclopropanes24 they can greatly modulate the spin density. An example of

256

a possible ligand that utilizes both an extended u—system and a heterocyclic ring is the

phenazine ligand,25 which is a catechol-type ligand that is depicted in Figure 6-1.

R1 N CH

\

/

R; N OH
Figure 6-1. Phenazine ligand.
Table 6-1. NPA spin density comparison between the phenanthrenesemiquinone (PSQ),

coronenesemiquinone (coronene-SQ), 2,3-anthrasemiquinone (ASQ), and phenazinesemiquinone
(phenazine-SQ).

 

Complex Avg. spin at O

 

PSQ 0.25703
coronene-SQ 0.21968
ASQ 0.20605

phenazine-SQ 0.17522

 

Preliminary DFT studies indicate that the spin density at the oxygen atoms in phenazine-
semiquinone is greatly decreased compared to the PSQ ligand as Shown in Table 6-1.
Additionally, coronenesemiquinone26 and 2,3-anthrasemiquinone (ASQ) also decrease

Spin density at the oxygen centers, but to a lesser extent. A broken symmetry DFT
calculation was done with [Ni(tren)(ASQ)]+ (Figure 6-2), to ascertain if the above

changes in spin density would be reflected in the exchange coupling constant. Using the
ASQ ligand (which is an isomer of phenanthrenesemiquinone) the resulting J value was
calculated to be —109 cm'1 compared to the —258 cm'1 value for [N i(tren)(PSQ)]+
complex, using the same UB3LYP functional and 6-311g** basis set. This decrease in J

corresponds to a reduction of 48 % in the magnitude compared to the Ni-PSQ complex.

Based upon the DFT data either the ASQ or phenazine nickel complexes would be

257

excellent candidates for further research. Synthetic procedures for the 2,3-

127’28 and phenazine25 ligands are known, and similar reactions used to make

anthracatecho
the Ni-PSQ complex could then be applied to make the Ni-ASQ or Ni-phenazine-SQ

compounds.

 

Figure 6-2. Calculated structure (top) and total spin density plot (bottom) for [Ni(tren)(ASQ)]+.

If measurable temperature dependence in the magnetic susceptibility of either
unsubstituted nickel complex is observed in the SQUID experiments, then the J value
could possibly be experimentally determined. Subsequent substituted Ni-ASQ or Ni-
phenazine-SQ complexes can then be synthesized to examine any substituent effects on
the magnitude of the exchange coupling constant. If the DF T predictions hold, then we

should be able to experimentally determine the J value for a series of complexes and

verify or disprove our hypothesis for the spin density at the paramagnetic centers
modulating the exchange coupling constant. Therefore, there are many fruitful avenues
in theoretical chemistry, EPR spectroscopy, and synthetic chemistry to pursue in the

future work of this project.

259

6.3 References

1.

2.

10.

11.

12.

l3.

14.

15.

l6.

l7.

l8.

l9.

Shultz, D. A.; Sloop, J. C.; Washington, G. J. Org. Chem. 2006, 71, 9104.
Luneau, D. Curr. Opin. Solid State Mater. Sci. 2001, 5, 123.

Ohrstrom, L. C. R. Chimie 2005, 8, 1374.

Shultz, D. A. Polyhedron 2001, 20, 1627.

Caneschi, A.; Gatteschi, D.; Sessoli, R.; Rey, P. A cc. Chem. Res. 1989, 22, 392.
Vostrikova, K. E. Coord. Chem. Rev. 2008, 252, 1409.

Stumpf, H. O.; Ouahab, L.; Pei, Y.; Grandjean, D.; Kahn, 0. Science 1993, 261,
447.

Hoganson, C. W.; Babcock, G. T. Science 1997, 277, 1953.
Itoh, S.; Taki, M.; Fukuzumi, S. Coord. Chem. Rev. 2000, 198, 3.

Whittaker, M. M.; Ekberg, C. A.; Peterson, J.; Sendova, M. S.; Day, E. P.;
Whittaker, J. W. J. Mol. Catal. B. : Enzym. 2000, 8, 3.

Robotham, B.; O'Malley, P. J. Biochemistry 2008, 47, 13261.

Lendzian, F .; Huber, M.; Isaacson, R. A.; Endeward, B.; Plato, M.; Bdnigk, B.;
K., M.; Lubitz, W.; Feher, G. Biochimica et Biophysica Acta 1993, 1183, 139.

Prokhorova, T. G.; Simonov, S. V.; Khasanov, S. S.; Zorina, L. V.; Buravov, L.
1.; Shibaeva, R. P.; Yagubskii, E. B.; Morgunov, R. B.; F oltynowicz, D.; Swietlik,
R. Synth. Met. 2008, 158, 749, and references therein.

Zhang, H. Y.; Ji, H. F. J. Mol. Struct. THEOCHEM 2003, 663, 167.

Zhang, H. Y.; Wang, X. L.; Sun, Y. M. J. Mol. Des. 2003, 2, 262.

Broze, M.; Luz, Z. J. Chem. Phys. 1969, 51, 738.

Borhan, B.; Nazarian, S.; Stocking, E. M.; Hammock, B. D.; Kurth, M. J. J. Org.
Chem. 1994, 59, 4316.

Kol, M.; Rozen, S. J. Org. Chem. 1993, 58, 1593.

Piekara-Sady, L.; Kispert, L. D. Handbook of Electron Spin Resonance. AIP
Press: New York, 1994; Vol. 1, p 311-357.

260

   

20.

21.

22.

23.

24.

25.

26.

27.

28.

Mobius, K.; Plato, M.; Lubitz, W. Phys. Rep. 1982, 87, 17].

Kay, C. W. M.; Feicht, R.; Schulz, K.; Sadewater, P.; Sancar, A.; Bacher, A.;
Mobius, K.; Richter, G.; Weber, S. Biochemistry 1999, 38, 16740.

Ma, H.; Cai, F .; Liu, C.; Jiang, Y. J. Chem. Phys. 2005, 122, 104909.

Xie, C.; Lahti, P. M.; George, C. Org. Lett. 2000, 2, 3417.

Creary, X. Acc. Chem. Res. 2006, 39, 761.

Seillan, C.; Brisset, H.; Siri, O. Org. Lett. 2008, 10, 4013.

Rieger, R.; Kastler, M.; Enkelmann, V.; Mfillen, K. Chem. Eur. J. 2008, 14, 6322.
Green, A. J. Chem. Soc. 1927, 554.

P0220, J. L.; Clavier, G. M.; Colomes, M.; Bouas-Laurent, H. Tetrahedron 1997,
53, 6377.

261

 

       

       

                  

 

 

ERSITV L a

llmmmn will?“ -

MICHIGAN SJATE U
lll‘llll‘lllllllll

I: la .. n,

3 129 03062 9186

3