Electron Spin Resonance
Analysis and Interpretation
Electron Spin Resonance
Analysis and Interpretation
Philip H Rieger
Brown University, Providence, RI, USA
ISBN13: 9780854043552
A catalogue record for this book is available from the British Library
r The Royal Society of Chemistry 2007
All rights reserved
Apart from fair dealing for the purposes of research for noncommercial purposes or
private study, criticism or review, as permitted under the Copyright, Designs and Patents
Act 1988 and the Copyright and Related Rights Regulations 2003, this publication may
not be reproduced, stored or transmitted, in any form or by any means, without the prior
permission in writing of The Royal Society of Chemistry, or in the case of reproduction
in accordance with the terms of the licences issued by the Copyright Licensing Agency
in the UK, or in accordance with the terms of the licences issued by the appropriate
Reproduction Rights Organization outside the UK. Enquiries concerning reproduction
outside the terms stated here should be sent to The Royal Society of Chemistry at the
address printed on this page.
Published by The Royal Society of Chemistry,
Thomas Graham House, Science Park, Milton Road,
Cambridge CB4 0WF, UK
Registered Charity Number 207890
For further information see our web site at www.rsc.org
Preface
At the time of his death in April, 2004, Professor Philip H. Rieger had nearly
ﬁnished the manuscript of this book. His intention was to present a monograph
summarizing his approach to the ﬁeld of Electron Spin Resonance using
examples and other explanatory material developed during the course of a
40year career of research and teaching at Brown University. Although
the presentation was intended to be at the beginning graduate level, it could
also serve as an introduction to the fundamentals of ESR for working re
search scientists in organometallic chemistry or other areas from which Phil
attracted his many research collaborators. It gives the reader a thorough
introduction to the analysis and interpretation of CW ESR spectra at Xband
(9.5 GHz.) as applied to paramagnetic organic, inorganic and organometallic
molecules.
When Professor Rieger ﬁrst became interested in ESR, commercial instru
ments were not available. His introduction to the ﬁeld, as a graduate student
with George Fraenkel at Columbia University, took place in one of the
few laboratories in the world at the time where ESR equipment had been
built. Upon arriving at Brown his ﬁrst item of business was to design and
construct a spectrometer. The instrument was eventually retired once reliable,
sensitive commercial instruments became available. Nevertheless, that ﬁrst
spectrometer enabled one of us (ALR) to begin a scientiﬁc collaboration that
lasted the rest of Phil’s life, and the other (RGL) to get his own career started at
Brown.
In a preliminary draft of this book Phil wrote the following paragraph
explaining its origin:
This book has been many years in the making. It began with an
invitation from Prof. William Trogler to write a chapter on ESR appli
cations for the book Organometallic Radical Processes that was published
in 1990. There are some strong resemblances to Chapter 4 of the present
book. The writing was extended to a handout in 1991 when I was invited
to spend a sabbatical year at the University of Bristol. It was extended a
bit further when I was invited to give a series of lectures on ESR at the
University of Oviedo, still further when I was invited to give some lectures
at the University of Edinburgh, and still further in 1999 for a lecture series
v
at the University of Otago. Meanwhile, I had given a short series of
lectures on ESR at Brown University most years as part of a graduate
course in Physical Inorganic Chemistry.
In completing Phil’s book we have retained the set of examples and expla
nations, and occasional commentary, as he had intended it. It has, however,
been some time since a book on ESR for the nonexpert has appeared. We have
therefore supplemented his original manuscript in two ways. At the end of
Chapter 1 is added an uptodate list of texts and monographs on ESR which
should serve the interested reader as a source of additional treatments of the
subject. Secondly, in Appendix 2 we have referenced and given brief descrip
tions of some advanced ESR methods that have been developed in recent years
and applied in various ﬁelds, including biochemistry. The modern ESR spec
troscopist is now as likely to need an understanding of these techniques as of
the classic Xband methods described here. We hope that this book will provide
a basis for study of the newer methods.
One of Philip Rieger’s most important contributions to the ﬁeld of ESR, and
the motivation for much of his collaborative research, was his instinctive
understanding of how to analyze the powder patterns of paramagnetic inor
ganic complexes, often using programs for simulating and analyzing such
spectra that he had developed over the years. A summary of his work in this
area may be found in a review titled ‘‘Simulation and analysis of ESR powder
patterns’’ published in the Specialist Periodical Report Electron Spin Reso
nance, Royal Society of Chemistry, Cambridge, 1993, vol 13B, ch. 4, pp.
178–199. Specism and other ESR tools written by Prof. Rieger are available
from the Manchester University website at www.epr.chemistry.manchester.
ac.uk. When you go to this site you will ﬁnd a menu on the left with a
‘Software’ button which will give you access to this material.
We are particularly grateful to Elsevier Publishing Company for allowing the
use, and modiﬁcation, of material that originally appeared in Organometallic
Radical Processes, Journal of Organometallic Chemistry Library, ed. W. C.
Trogler, Elsevier, Amsterdam, 1990, vol 22. Most of Chapter 4, and also
substantial parts of Chapter 3, were ﬁrst published in this review. Other reviews
by Professor Rieger on topics covered in this book include ‘‘Electron
paramagnetic resonance studies of lowspin d
5
metal complexes’’, Coord.
Chem. Rev., 1994, 135, 203; ‘‘Chemical insights from EPR spectra of organo
metallic radicals and radical ions’’ (with Dr. Anne L. Rieger), Organometallics,
2004, 23, 154; and ‘‘Electron spin resonance’’, in Physical Methods of Chem
istry, ed. A. Weissberger and B. W. Rossiter, John Wiley and Sons, Inc., New
York, 1972, Part IIIA, ch.VI. pp 499–598.
Anne L. Rieger (Mrs. Philip Rieger) is grateful to Brown University for
providing her with the facilities to make the completion of this project possible:
oﬃce space, computers, the use of libraries, and a high speed internet connec
tion to Professor Lawler in the mountains of New Hampshire. For many years
Professor Rieger was associated with the ESR group of the Royal Society of
Chemistry. The support and encouragement of many members of the group has
vi Preface
served as an impetus to see this project to completion and is gratefully
acknowledged. Professor Neil Connelly of Bristol University, a long time
collaborator and friend, has also contributed to the completion of the project
and his input is very much appreciated. Finally, Ron Lawler is grateful to
Margaret Merritt for her advice and support during this occasionally frustrat
ing, but always interesting, introduction to the world of desktop publishing.
We hope that our friend, companion and colleague Phil would be pleased with
the result.
Anne L. Rieger
Ronald G. Lawler
Pawtucket, RI, USA
Center Sandwich, NH, USA
vii Preface
Contents
Chapter 1 Introduction
1.1 What is ESR Spectroscopy? 1
1.2 The ESR Experiment 3
1.2.1 Sensitivity 4
1.2.2 Saturation 5
1.2.3 Nuclear Hyperﬁne Interaction 5
1.3 Operation of an ESR Spectrometer 7
1.4 Optimization of Operating Parameters 11
1.4.1 Microwave Frequency 11
1.4.2 Microwave Power 12
1.4.3 Center Field, Sweep Width and Field Oﬀset 13
1.4.4 Sweep Time 14
1.4.5 Modulation Frequency 14
1.4.6 Second Harmonic Detection 15
1.4.7 Modulation Amplitude 15
1.4.8 Modulation Phase 15
1.4.9 Signal Gain 16
1.4.10 Filter Time Constant 16
1.5 Applications of ESR Spectroscopy 17
1.5.1 Electronic Structure Determination 17
1.5.2 Analytical Applications 17
1.5.3 Determination of Rates 18
References 18
Chapter 2 Isotropic ESR Spectra of Organic Radicals
2.1 Isotropic ESR Spectra 21
2.1.1 Line Positions in Isotropic Spectra 21
2.1.2 Hyperﬁne Coupling Patterns 22
2.1.3 Secondorder Splittings 25
2.1.4 Spin Hamiltonian Parameters from Spectra 26
viii
2.2 Interpretation of Isotropic Parameters 27
2.3 Line Widths in Isotropic Spectra 29
2.3.1 Incomplete Averaging of Anisotropies 29
2.3.2 Rates of Fluxionality from Line Widths 30
2.4 Organic Radical Reactions 32
2.5 Analysis of Isotropic ESR Spectra 32
2.5.1 Preliminary Examination of the Spectrum 33
2.5.2 What do you Expect to See? 33
2.5.3 Are the Gross Features of the Spectrum
Consistent with the Model? 34
2.5.4 An Example 34
2.5.5 Detailed Analysis for Determination of
Parameters 36
2.5.6 Computation of Multiplet Intensity Ratios 37
2.5.7 Multiplet Patterns due to Isotopomers 38
2.5.8 Secondorder Shifts in Line Positions 39
2.6 Related Techniques (ENDOR) 41
References 42
Chapter 3 Isotropic Spectra of Organometallic Radicals
3.1 Secondorder Eﬀects on Line Positions 44
3.2 Understanding the Variation in Line Widths 47
3.3 Puzzling Line Shapes 48
3.4 Use of ESR Spectra to Determine Formation
Constants 49
References 51
Chapter 4 Anisotropic ESR Spectra
4.1 Introduction 52
4.2 Solidstate ESR Spectra 53
4.2.1 Spectra of Dilute Single Crystals 54
4.2.2 Analysis of Frozen Solution Spectra 55
4.3 Interpretation of the gMatrix 59
4.4 Interpretation of the Hyperﬁne Matrix 60
4.5 Organometallic Examples 63
4.5.1 A Lowspin Manganese(II) Complex 63
4.5.2 Some Cobalt(0) Radical Anions 66
4.6 Organic Examples of Solidstate ESR Spectra 69
4.6.1 Irradiated Single Crystal of Glycylglycine 69
4.6.2 Xirradiated Single Crystal of Methylene
Diphosphonic Acid 70
4.7 Noncoincident Matrix Axes 71
ix Contents
4.7.1 Symmetry Considerations 71
4.7.2 Experimental Determination of Matrix Axis
Orientations 72
4.8 Organometallic Examples of Noncoincident Matrix
Axes 73
4.8.1 A Chromium Nitrosyl Complex 73
4.8.2 Iron Pentacarbonyl Ions 74
4.8.3 Another Lowspin Manganese(II) Complex 76
4.8.4 Chromium(I) Pianostool Complex 77
4.8.5 [(RCCR
0
)Co
2
(CO)
6
]
À
and [SFeCo
2
(CO)
9
]
À
79
4.8.6 (oXylylene)Mn(dmpe)
2
81
4.8.7 Cobalt Dithiolene Complexes 86
4.9 ‘‘gStrain’’ 87
References 89
Chapter 5 ESR Kinetic Studies
5.1 Bloch’s Phenomenological Model 92
5.1.1 Derivation of the Bloch Equations 94
5.1.2 Steadystate Solution 95
5.2 Chemical Exchange  The Modiﬁed Bloch Equations 98
5.3 Further Discussion of Line Shapes 102
5.4 Applications of the Modiﬁed Bloch Equations 102
5.5 Alternating Line Width Eﬀects 107
5.6 Spin Labels 108
References 110
Chapter 6 ESR Spectra of Biradicals, Triplet States, and other S>1/2
Systems
6.1 Biradicals 112
6.1.1 Exchange Coupling 113
6.2 Organic Triplet State Molecules and the Dipolar
Interaction 117
6.2.1 Organic Triplet State Molecules 122
6.3 Transition Metal Complexes with S41/2 122
6.3.1 Spin–Orbit Coupling 122
6.3.2 Highspin Transition Metal Ions 126
6.3.3 Examples: K
3
Cr(CN)
6
and K
4
V(CN)
6
130
References 132
Chapter 7 Perturbation Theory Calculations
7.1 Secondorder Perturbation Theory Treatment of Spin
Hamiltonian with Noncoincident g and Aaxes 133
7.1.1 The Electron Zeeman Term 133
x Contents
7.1.2 Nuclear Hyperﬁne Interaction 135
7.1.3 Perturbation Theory Treatment of Hyperﬁne
Term 138
7.1.4 Example Application of these Results 144
7.2 Quadrupole Coupling 145
7.2.1 Perturbation Theory Treatment of Quadrupole
Term 146
7.2.2 Example Application of Analysis Quadrupole
Eﬀects 149
References 152
Appendix 1 Physical Constants, Conversion Factors, and Properties of
Nuclei (Tables A1.1–A1.4)
Example 155
References 157
Appendix 2 Advanced ESR Methods
A2.1 High Frequency ESR 159
A2.2 Double Resonance 161
A2.3 Pulsed Methods 162
References 164
Subject Index 169
xi Contents
CHAPTER 1
Introduction
1.1 What is ESR Spectroscopy?
Electron spin resonance spectroscopy (ESR), also known as electron paramag
netic resonance (EPR) or electron magnetic resonance (EMR), was invented by
the Russian physicist Zavoisky
1
in 1945. It was extended by a group of
physicists at Oxford University in the next decade. Reviews of the Oxford
group’s successes are available
2
and books by Abragam and Bleaney and by
Abragam
3
cover the major points discovered by the Oxford group. In the
present book, we focus on the spectra of organic and organotransition metal
radicals and coordination complexes. Although ESR spectroscopy is supposed
to be a mature ﬁeld with a fully developed theory,
4
there have been some
surprises as organometallic problems have explored new domains in ESR
parameter space. We will start in this chapter with a synopsis of the funda
mentals of ESR spectroscopy. For further details on the theory and practice of
ESR spectroscopy, the reader is referred to one of the excellent texts and
monographs on ESR spectroscopy.
3,5–36
Sources of data and a guide to the
literature of ESR up to about 1990 can be found in ref. 16a. The history of ESR
has also been described by many of those involved in the founding and
development of the ﬁeld.
37
The electron spin resonance spectrum of a free radical or coordination
complex with one unpaired electron is the simplest of all forms of spectroscopy.
The degeneracy of the electron spin states characterized by the quantum
number, m
S
¼ Æ1/2, is lifted by the application of a magnetic ﬁeld, and
transitions between the spin levels are induced by radiation of the appropriate
frequency (Figure 1.1). If unpaired electrons in radicals were indistinguishable
from free electrons, the only information content of an ESR spectrum would be
the integrated intensity, proportional to the radical concentration. Fortunately,
an unpaired electron interacts with its environment, and the details of ESR
spectra depend on the nature of those interactions. The arrow in Figure 1.1
shows the transitions induced by 0.315 cm
À1
radiation.
Two kinds of environmental interactions are commonly important in the
ESR spectrum of a free radical: (i) To the extent that the unpaired electron has
residual, or unquenched, orbital angular momentum, the total magnetic
moment is diﬀerent from the spinonly moment (either larger or smaller,
1
depending on how the angular momentum vectors couple). It is customary to
lump the orbital and spin angular momenta together in an eﬀective spin and to
treat the eﬀect as a shift in the energy of the spin transition. (ii) The electron
spin energy levels are split by interaction with nuclear magnetic moments – the
nuclear hyperﬁne interaction. Each nucleus of spin I splits the electron spin
levels into (2I þ 1) sublevels. Since transitions are observed between sublevels
with the same values of m
I
, nuclear spin splitting of energy levels is mirrored by
splitting of the resonance line (Figure 1.2).
0.3
0.2
0.1
0
0.1
0.2
0.3
E
/
c
m

1
0 1000 2000 3000 4000 5000
B/Gauss
Figure 1.1 Energy levels of an electron placed in a magnetic ﬁeld.
0.3
0.2
0.1
0
0.1
0.2
0.3
E
/
c
m

1
0 1000 2000 3000 4000 5000
B/Gauss
Figure 1.2 Energy levels of an unpaired electron in a magnetic ﬁeld interacting with a
spin1/2 nucleus. The arrows show two allowed transitions.
2 Chapter 1
1.2 The ESR Experiment
When an electron is placed in a magnetic ﬁeld, the degeneracy of the electron
spin energy levels is lifted
w
as shown in Figure 1.1 and as described by the spin
Hamiltonian:
^
H
s
¼ gm
B
B
^
S
z
ð1:1Þ
In eqn (1.1), g is called the gvalue (or gfactor), (g
e
¼ 2.00232 for a free
electron), m
B
is the Bohr magneton (9.274 Â 10
À28
J G
À1
), B is the magnetic
ﬁeld strength in Gauss
z
, and S
z
is the zcomponent of the spin angular
momentum operator (the magnetic ﬁeld deﬁnes the zdirection). The electron
spin energy levels are easily found by application of
^
H
s
to the electron spin
eigenfunctions corresponding to m
S
¼ Æ1/2:
^
H
s
Æ
1
2
¼ Æ
1
2
gm
B
B Æ
1
2
¼ E
Æ
Æ
1
2
Thus
E
Æ
¼ Æ
1
2
À Á
gm
B
B ð1:2Þ
The diﬀerence in energy between the two levels:
DE ¼ E
þ
À E
À
¼ gm
B
B ¼ hn ð1:3Þ
corresponds to the energy, hn, of a photon required to cause a transition; or in
wavenumbers by eqn (1.4), where g
e
m
B
/hc ¼ 0.9348 Â 10
À4
cm
À1
G
À1
:
~ n ¼ l
À1
¼ n=c ¼ ðgm
B
=hcÞB ð1:4Þ
Since the gvalues of organic and organometallic free radicals are usually in
the range 1.9–2.1, the free electron value is a good starting point for describing
the experiment.
Magnetic ﬁelds of up to ca. 15000 G are easily obtained with an ironcore
electromagnet; thus we could use radiation with ~ n up to 1.4 cm
À1
(n o42 GHz
or l 4 0.71 cm). Radiation with this kind of wavelength is in the microwave
region. Microwaves are normally handled using waveguides designed to trans
mit radiation over a relatively narrow frequency range. Waveguides look like
rectangular crosssection pipes with dimensions on the order of the wavelength
w
Energy level splitting in a magnetic ﬁeld is called the Zeeman eﬀect, and the Hamiltonian of eqn
(1.1) is sometimes referred to as the electron Zeeman Hamiltonian. Technically, the energy of a
magnetic dipole in a magnetic ﬁeld is the negative of that shown in eqn (1.1). For electron spins,
however, the parameter g is negative, i.e., the magnetic moment and spin angular momentum
vectors are antiparallel, and both negative signs may be combined to give the formula as written.
This has the advantage of allowing gvalues to be tabulated as positive numbers.
z
Using the symbol ‘‘B’’ for the magnetic ﬁeld technically implies we are using the MKS system of
units in which B is given in Tesla (T). Many spectroscopists still express the magnetic ﬁeld in Gauss
(G) ¼ 10
À4
T, however, and we will continue that practice here.
3 Introduction
to be transmitted. As a practical matter for ESR, waveguides can not be too big
or too small À1 cm is a bit small and 10 cm a bit large; the most common
choice, called Xband microwaves, has l in the range 3.0–3.3 cm (n E 9–10
GHz); in the middle of Xband, the free electron resonance is found at 3390 G.
Although Xband is by far the most common, ESR spectrometers are
available commercially or have been custom built in several frequency ranges
(Table 1.1).
1.2.1 Sensitivity
As for any quantum mechanical system interacting with electromagnetic radi
ation, a photon can induce either absorption or emission. The experiment
detects net absorption, i.e., the diﬀerence between the number of photons
absorbed and the number emitted. Since absorption is proportional to the
number of spins in the lower level and emission is proportional to the number
of spins in the upper level, net absorption, i.e., absorption intensity, is propor
tional to the diﬀerence:
Net absorption / N
À
À N
þ
The ratio of populations at equilibrium is given by the Boltzmann distribution:
N
þ
=N
À
¼ expðÀDE=kTÞ ¼ expðÀgm
B
B=kTÞ ð1:5Þ
For ordinary temperatures and ordinary magnetic ﬁelds, the exponent is very
small and the exponential can be accurately approximated by the expansion,
e
Àx
E 1 À x. Thus
N
þ
=N
À
% 1 À gm
B
B=kT
Since N
À
E N
1
E N/2, the population diﬀerence can be written:
N
À
À N
þ
¼ N
À
½1 À ð1 À gm
B
B=kTÞ ¼ Ngm
B
B=2kT ð1:6Þ
This expression tells us that ESR sensitivity (net absorption) increases with
the total number of spins, N, with decreasing temperature and with increasing
magnetic ﬁeld strength. Since the ﬁeld at which absorption occurs is propor
tional to microwave frequency, in principle sensitivity should be greater for
Table 1.1 Common frequencies used for ESR
Designation n (GHz) l (cm) B (electron) (G)
L 1.1 27 390
S 3.0 10 1070
X 9.5 3.2 3400
K 24 1.2 8600
Q 35 0.85 12500
W 95 0.31 34000
– 360 0.083 128000
4 Chapter 1
higher frequency K or Qband spectrometers than for Xband. However, the
K or Qband waveguides are smaller, so samples are also necessarily smaller
and for the same concentration contain fewer spins. This usually more than
cancels the advantage of a more favorable Boltzmann factor for samples of
unlimited size or ﬁxed concentration.
Under ideal conditions, a commercial Xband spectrometer can detect about
10
12
spins (ca. 10
À12
moles) at room temperature. This number of spins in a 1
cm
3
sample corresponds to a concentration of about 10
À9
M. By ideal condi
tions, we mean a single line, on the order of 0.1 G wide, with sensitivity going
down roughly as the reciprocal square of the line width. When the resonance is
split into two or more hyperﬁne lines, sensitivity decreases still further. None
theless, ESR is a remarkably sensitive technique, especially compared with
NMR.
1.2.2 Saturation
Because the two spin levels are aﬀected primarily by magnetic forces, which are
weaker than the electric forces responsible for most other types of spectroscopy,
once the populations are disturbed by radiation it takes longer for equilibrium
population diﬀerences to be established. Therefore an intense radiation ﬁeld,
which tends to equalize the populations, leads to a decrease in net absorption
which is not instantly restored once the radiation is removed. This eﬀect is
called ‘‘saturation’’. The return of the spin system to thermal equilibrium, via
energy transfer to the surroundings, is a rate process called spin–lattice relax
ation, with a characteristic time (T
1
), the spin–lattice relaxation time (relaxation
rate constant ¼ 1/T
1
). Systems with a long T
1
(i.e., spin systems weakly coupled
to the surroundings) will be easily saturated; those with shorter T
1
will be more
diﬃcult to saturate. Since spin–orbit coupling provides an important energy
transfer mechanism, we usually ﬁnd that oddelectron species with light atoms
(e.g., organic radicals) have long T
1
s, those with heavier atoms (e.g., organo
transition metal radicals) have shorter T
1
s. The eﬀect of saturation is con
sidered in more detail in Chapter 5, where the phenomenological Bloch
equations are introduced.
1.2.3 Nuclear Hyperﬁne Interaction
When one or more magnetic nuclei interact with the unpaired electron, we have
another perturbation of the electron energy, i.e., another term in the spin
Hamiltonian:
^
H
s
¼ gm
B
B
^
S
z
þ A
^
I Á
^
S ð1:7Þ
where A is the hyperﬁne coupling parameter in energy units (joule). Strictly
speaking we should include the nuclear Zeeman interaction, gBI
z
. However, in
most cases the energy contributions are negligible on the ESR energy scale, and,
5 Introduction
since observed transitions are between levels with the same values of m
I
, the
nuclear Zeeman energies cancel in computing ESR transition energies.
y
The eigenfunctions of the spin Hamiltonian [eqn (1.7)] are expressed in terms
of an electron and nuclearspin basis set m
S
, m
I
i, corresponding to the
electron and nuclear spin quantum numbers m
S
and m
I
, respectively. The
energy eigenvalues of eqn (1.7) are:
E
1
2
;
1
2
À Á
¼
1
2
gm
B
B þ
1
4
A ð1:8AÞ
E À
1
2
; À
1
2
À Á
¼ À
1
2
gm
B
B þ
1
4
A ð1:8BÞ
E
þ
¼
1
2
gm
B
B½1 þ ðA=gm
B
BÞ
2
1
2
À
1
4
A ð1:8CÞ
E
À
¼ À
1
2
gm
B
B½1 þ ðA=gm
B
BÞ
2
1
2
À
1
4
A ð1:8DÞ
The eigenfunctions corresponding the E
1
and E
À
are mixtures of j
1
2
; À
1
2
i
and j À
1
2
;
1
2
i.
If the hyperﬁne coupling is suﬃciently small, A oogm
B
B, the second term in
brackets in eqns (1.8C) and (1.8D) are negligible, which corresponds to ﬁrst
order in perturbation theory, and the energies become:
E ¼ Æ
1
2
gm
B
B Æ
1
2
A ð1:9Þ
These are the energy levels shown in Figure 1.2. The exact energies in eqn.
(1.8), which were ﬁrst derived by Breit and Rabi,
38
are plotted as functions of B
in Figure 1.3 for g ¼ 2.00, A/hc ¼ 0.1 cm
À1
. Notice that, at zero ﬁeld, there are
two levels corresponding to a spin singlet (E ¼ À3A/4hc) and a triplet (E ¼ þA/
4hc). At high ﬁeld, the four levels divide into two higher levels (m
S
¼ þ1/2) and
two lower levels (m
S
¼ À1/2) and approach Figure 1.2, the ﬁrstorder result,
eqn. (1.9) (the ﬁrstorder solution is called the highﬁeld approximation). To
conserve angular momentum, transitions among these levels can involve
changes in angular momentum of only one unit. At high ﬁelds this corresponds
to ﬂipping only one spin at a time; in other words, the selection rules are Dm
S
¼
Æ1, Dm
I
¼ 0 (ESR transitions) or Dm
S
¼ 0, Dm
I
¼ Æ1 (NMR transitions). The
latter involves much lower energy photons, and, in practice, only the Dm
S
¼ Æ1
transitions are observed in an ESR spectrometer. At lower ﬁelds, or when A
y
An exception to this rule arises in the ESR spectra of radicals with small hyperﬁne parameters in
solids. In that case the interplay between the Zeeman and anisotropic hyperﬁne interaction may
give rise to satellite peaks for some radical orientations (S. M. Blinder, J. Chem. Phys., 1960, 33,
748; H. Sternlicht, J. Chem. Phys., 1960, 33, 1128). Such eﬀects have been observed in organic free
radicals (H. M. McConnell, C. Heller, T. Cole and R. W. Fessenden, J. Am. Chem. Soc., 1959, 82,
766) but are assumed to be negligible for the analysis of powder spectra (see Chapter 4) where A is
often large or the resolution is insuﬃcient to reveal subtle spectral features. The nuclear Zeeman
interaction does, however, play a central role in electronnuclear double resonance experiments
and related methods [Appendix 2 and Section 2.6 (Chapter 2)].
6 Chapter 1
becomes comparable in magnitude to gm
B
B, the transitions may involve
simultaneous ﬂipping of electron and nuclear spins. This gives rise to second
order shifts in ESR spectra (see Chapters 2 and 3).
1.3 Operation of an ESR Spectrometer
Although many spectrometer designs have been produced over the years, the
vast majority of laboratory instruments are based on the simpliﬁed block
diagram shown in Figure 1.4. Planepolarized microwaves are generated by the
klystron tube and the power level adjusted with the Attenuator. The Circulator
Klystron
Attenuator
Load
Circulator
Magnet
Cavity
µAmmeter
Diode
Detector
Figure 1.4 Block diagram of an ESR spectrometer.
0.3
0.2
0.1
0
0.1
0.2
0.3
E
/
c
m

1
0 1000 2000 3000 4000 5000
B/Gauss
Figure 1.3 Energy levels for an electron interacting with a spin1/2 nucleus with A/hc ¼
0.1 cm
À1
. The arrows show the transitions induced by 0.315 cm
À1
radiation.
7 Introduction
behaves like a traﬃc circle: microwaves entering from the klystron are routed
toward the Cavity where the sample is mounted. Microwaves reﬂected back
from the Cavity (which is reduced when power is being absorbed) are routed to
the Diode Detector, and any power reﬂected from the diode is absorbed
completely by the Load. The diode is mounted along the Evector of the
planepolarized microwaves and thus produces a current proportional to the
microwave power reﬂected from the cavity. Thus, in principle, the absorption
of microwaves by the sample could be detected by noting a decrease in current
in the Microammeter. In practice, of course, such a measurement would detect
noise at all frequencies as well as signal and have a far too low signaltonoise
ratio to be useful.
The solution to the signaltonoise problem is to introduce small amplitude
ﬁeld modulation. An oscillating magnetic ﬁeld is superimposed on the dc ﬁeld
by means of small coils, usually built into the cavity walls. When the ﬁeld is in
the vicinity of a resonance line, it is swept back and forth through part of the
line, leading to an a.c. component in the diode current. This a.c. component is
ampliﬁed using a frequency selective ampliﬁer tuned to the modulation fre
quency, thus eliminating a great deal of noise. The modulation amplitude is
normally less than the line width. Thus the detected a.c. signal is proportional
to the change in sample absorption as the ﬁeld is swept. As shown in Figure 1.5,
this amounts to detection of the ﬁrst derivative of the absorption curve.
It takes a little practice to get used to looking at ﬁrstderivative spectra, but
there is a distinct advantage: ﬁrstderivative spectra have much better apparent
resolution than do absorption spectra. Indeed, secondderivative spectra are
even better resolved (though the signaltonoise ratio decreases on further
diﬀerentiation). Figure 1.6 shows the eﬀect of higher derivatives on the reso
lution of a 1:2:1 triplet arising from the interaction of an electron with two
equivalent I ¼ 1/2 nuclei.
The microwavegenerating klystron tube
z
requires explanation. A schematic
drawing of the klystron is shown in Figure 1.7. There are three electrodes:
a heated cathode from which electrons are emitted, an anode to collect the
electrons, and a highly negative reﬂector electrode that sends those electrons
which pass through a hole in the anode back to the anode. The motion of
the charged electrons from the hole in the anode to the reﬂector and back to the
anode generates an oscillating electric ﬁeld and thus electromagnetic radiation.
The transit time from the hole to the reﬂector and back again corresponds to
the period of oscillation (n). Thus the microwave frequency can be tuned
(over a small range) by adjusting the physical distance between the anode and
z
Some modern spectrometers, especially those employing pulsed excitation (see Appendix 2), use a
solidstate Gunn diode microwave source. This device makes use of the Gunn Eﬀect, discovered by
J. B. Gunn of IBM in 1963, in which electrons become ‘‘trapped’’ in potential wells within a solid
and oscillate at a frequency that may be varied by changing the applied voltage. As with a
klystron, tuning is done by varying the voltage. Gunn diode oscillators have better frequency
stability and longer lifetimes than klystrons and are becoming the preferred microwave source for
ESR spectrometers.
8 Chapter 1
A
b
s
o
r
p
t
i
o
n
2
n
d
d
e
r
i
v
a
t
i
v
e
Magnetic Field Strength
1
s
t

d
e
r
i
v
a
t
i
v
e
Figure 1.6 Firstderivative curves show better apparent resolution than do absorption
curves – and secondderivatives curves are still better.
D
e
t
e
c
t
e
d
S
i
g
n
a
l
Magnetic Field
A
b
s
o
r
p
t
i
o
n
Mod.
Ampl.
Figure 1.5 Smallamplitude ﬁeld modulation converts the absorption curve into a ﬁrst
derivative.
9 Introduction
the reﬂector or by adjusting the reﬂector voltage. In practice, both methods
are used: the metal tube is distorted mechanically to adjust the distance
(a coarse frequency adjustment) and the reﬂector voltage is adjusted as a
ﬁne control.
The sample is mounted in the microwave cavity (Figure 1.8). The cavity is a
rectangular metal box, exactly one wavelength long. An Xband cavity has
dimensions of about 1 Â 2 Â 3 cm. The electric and magnetic ﬁelds of the
standing wave are shown in the ﬁgure. Note that the sample is mounted in the
electric ﬁeld nodal plane, but at a maximum in the magnetic ﬁeld. The static
ﬁeld, B, is perpendicular to the sample port.
The cavity length is not adjustable, but it must be exactly one wavelength.
Thus the spectrometer must be tuned such that the klystron frequency is equal
to the cavity resonant frequency. The tuneup procedure usually includes
observing the klystron power mode. That is, the frequency is swept across a
range that includes the cavity resonance by sweeping the klystron reﬂector
voltage, and the diode detector current is plotted on an oscilloscope or other
device. When the klystron frequency is close to the cavity resonance, microwave
energy is absorbed by the cavity and the power reﬂected from the cavity to the
diode is minimized, resulting in a dip in the power mode (Figure 1.9). The
‘‘cavity dip’’ is centered on the power mode using the coarse mechanical
frequency adjustment, while the reﬂector voltage is used to ﬁne tune the
frequency.
Electric Field
Magnetic Field
Sample Port
Iris
Figure 1.8 Microwave cavity.
Electron
pathway
Reflector
electrode
Anode
Heated
filament
cathode
Figure 1.7 Schematic drawing of a microwavegenerating klystron tube.
10 Chapter 1
1.4 Optimization of Operating Parameters
8
Determining the ESR spectrum of a sample using a typical CW spectrometer of
the sort outlined in Figure 1.4, which is still the most common commercially
available type of instrument, involves adjustment of the set of operating
parameters described below. In the early days of ESR these adjustments would
be carried out using control knobs on a console. Nowadays, of course, the
settings are carried out under control of a computer interface. The purpose of
these parameters and typical values, however, has remained unchanged. Such
spectrometers are quite adequate for studying relatively stable samples. Char
acterizing transient species by ESR, however, requires substantial modiﬁcation
of commercial instruments or the use of a pulsed spectrometer.
There are twelve parameters that must be set or known in recording an ESR
spectrum (Table 1.2). Brieﬂy, below, each parameter is discussed and the means
used to optimize or measure the parameter described.
1.4.1 Microwave Frequency
The resonant microwave frequency reaching the sample is determined by the
eﬀective length of the microwave cavity. The actual length is somewhat
Reflector Voltage
D
i
o
d
e
C
u
r
r
e
n
t
Figure 1.9 Klystron mode and cavity dip.
Table 1.2 Parameters involved in the recording of an ESR spectrum
Microwave frequency Center ﬁeld Modulation frequency Modulation phase
Microwave power Sweep width 1st or 2nd harmonic Signal gain
Sweep time Field oﬀset Modulation amplitude Filter time constant
8
A comprehensive discussion and illustrations of the eﬀects of spectrometer operating parameters
on ESR spectra are given by the author in Electron spin resonance, in Physical Methods of
Chemistry, ed. A. Weissberger and B. W. Rossiter, John Wiley and Sons, Inc., New York, 1972,
part IIIA, ch. VI,.
11 Introduction
modiﬁed by the inﬂuence of the sample tube and the Dewar insert (if controlled
temperature operation is required) so that the microwave frequency varies by a
few percent even for the same cavity. Since the klystron frequency is locked to
the cavity resonant frequency by a suitable feedback circuit, this is not an
adjustable parameter. However, to compute a gvalue from a spectrum, the
frequency must be known quite accurately. This is normally done using a
microwave frequency counter installed somewhere in the waveguide circuit.
1.4.2 Microwave Power
According to the solution of the Bloch equations (Chapter 5), the magnetic
resonance absorption, sometimes called the ‘‘vmode signal’’, v, is given by eqn
(1.10).
v ¼
gB
1
M
0
T
2
1 þ T
2
2
ðo
0
À oÞ
2
þ g
2
B
2
1
T
1
T
2
ð1:10Þ
where T
1
and T
2
are the spin–lattice and transverse relaxation times, respec
tively, M
0
is the bulk magnetization, g is the magnetogyric ratio (gm
B
in ESR
terms), o
0
is the resonant frequency (proportional to B
0
), o is the angular
microwave frequency (n), and B
1
is the amplitude of the oscillating ﬁeld. The
microwave power is proportional to B
1
2
. For small B
1
, the absorption signal
increases linearly with increasing B
1
. However, when the third term in
the denominator becomes important, v goes through a maximum when
g
2
B
1
2
T
1
T
2
¼ 1, and begins to decrease with increasing B
1
– the resonance
begins to ‘‘saturate’’. At maximum v, however, the saturation term leads to an
increase in apparent line width by a factor of O2. This eﬀect is illustrated in
Figure 1.10.
To simultaneously maximize signaltonoise ratio (S/N) and minimize dis
tortion, it is best to adjust the microwave power, as measured by the power
Figure 1.10 Eﬀect of increasing microwave power on intensity and shape of an ESR
line. Power increases from top to bottom of the ﬁgure. Units are relative
values of B
1
2
.
12 Chapter 1
meter, down by a factor or 4 or 5 (6–7 db) from the maximum amplitude power.
This will result in a decrease of less than 2 in signal amplitude. In practice,
following this procedure would require ﬁnding the maximum amplitude by trial
and error and then turning the power level down by 6–7 db. This eﬀect is
illustrated in Figure 1.11.
In general it is unnecessary to spend much time adjusting the power level.
The general rule is to adjust the power to about 10 db attenuation for
organic radicals and to use full power for transition metal complexes and
those organometallics where the unpaired electron is primarily located on
the metal.
1.4.3 Center Field, Sweep Width and Field Oﬀset
Once you know, or can guess, the ﬁeld limits of your spectrum, setting the
center ﬁeld and sweep width values is not very diﬃcult. The center ﬁeld
corresponds to the middle of the spectrum and a suﬃciently large sweep width
chosen so that all of the spectrum is recorded.
If you do not know the ﬁeld range occupied by your spectrum in advance, the
center ﬁeld must be chosen by educated guess; set the sweep width 2–4Â greater
than the expected width. Hopefully, you will see at least a piece of your
spectrum and can make appropriate adjustments to zero in on the correct
settings.
Most spectrometers measure the magnetic ﬁeld by a Hall eﬀect probe
consisting of a sensor mounted on one of the pole faces of the magnet.
However, such estimates of the value of B inside the cavity are not suﬃciently
accurate to be used for gfactor determinations. There are two ways around this
problem: (i) measure the spectrum of a solid free radical such as dip
henylpicrylhydrazyl (DPPH), which has a wellknown gvalue (2.0028), at least
once during acquisition of the desired spectrum; or (ii) use of an NMR
gaussmeter probe inserted in or near the cavity several times during the
collection of the spectrum.
Figure 1.11 Eﬀect of microwave power on relative width and amplitude of a line.
13 Introduction
1.4.4 Sweep Time
In general, the longer the sweep time the better the sensitivity since the ﬁlter
time constant parameter can be set longer with consequent improvement in
signaltonoise ratio. In practice, however, sweep times are usually set in
accordance with the expected lifetime of the radical species, the stability of
the instrument, and the patience of the operator. Decay of the radical or drift of
the spectrometer during a scan is clearly undesirable. The sweep time is most
commonly set in the range 4–10 min.
1.4.5 Modulation Frequency
With most spectrometers, you have a choice of either 100 kHz or a lower
frequency of ﬁeld modulation. The higher frequency generally gives better S/N,
but if the lines are unusually sharp (o0.08 Gauss), 100 kHz modulation leads
to ‘‘side bands’’, lumps in the line shape that confuse the interpretation of the
spectrum. This eﬀect is illustrated in Figure 1.12. Under such circumstances,
use the lower frequency for which the sidebands are closer together and thus
less likely to be a problem.
Figure 1.12 Eﬀect of 100 kHz modulation on an ESR line of decreasing width. The
xaxis is magnetic ﬁeld, in Gauss, relative to the center of the line.
14 Chapter 1
1.4.6 Second Harmonic Detection
In most cases, you will use the ﬁrst harmonic and the normal ﬁrstderivative of
the absorption spectrum will be presented. If your spectrum has very good S/N
and has some regions where you would like better resolution, a second
derivative presentation may help. However, second derivatives from second
harmonic detection are very costly in terms of S/N ratio and so you really do
have to have a strong signal!
1.4.7 Modulation Amplitude
Since the absorption signal is usually detected by imposing a 100 kHz ﬁeld
modulation on the static ﬁeld, the signal disappears when the modulation
amplitude is turned to zero. In general, the signal increases more or less in
proportion to the modulation amplitude, but eventually the detected lines begin
to broaden and then the signal amplitude decays as well. This is illustrated in
Figure 1.13.
Depending on what you want to optimize, here are some rules:
For optimum S/N ratio, but decreased resolution: Modulation amplitude ¼ 2 Â
linewidth.
For accurate line width measurements: Modulation amplitude ¼ linewidth/10.
For most practical work: Modulation amplitude ¼ linewidth/3.
1.4.8 Modulation Phase
To improve the S/N ratio, the modulation signal is processed by ampliﬁcation
with a tuned ampliﬁer using phasesensitive detection. This means that the
detected signal must not only be at the modulation frequency, but must also be
in phase with the modulation. Since the ampliﬁer itself can introduce a bit of
phase shift, there is a phase control which, in principle, should be adjusted to
maximize the signal amplitude. In practice, this control needs to be adjusted
only rarely and in most cases the best approach is to leave it alone.
Figure 1.13 Eﬀect of ﬁeld changing (modulation/line width) on relative amplitude
(lefthand axis) and observed width (righthand axis) of an ESR line.
15 Introduction
1.4.9 Signal Gain
Adjustment of the signal gain is straightforward. Set the gain suﬃciently high
that the recorded spectrum is nearly fullscale on the computer displays or
recorder. If you have no idea of the strength of your expected signal, a good
starting point would be 1 Â 10
4
. Too high a gain can result in ampliﬁer
overload.
1.4.10 Filter Time Constant
The output of the signal ampliﬁer is ﬁltered using a circuit designed to pass low
frequency, but not high frequency, signals. The cutoﬀ frequency is the recip
rocal of the time constant setting. Thus, a time constant of 0.1 s would lead to
signals with frequencies greater than about 10 Hz being attenuated. The best
time constant setting thus depends critically on the sweep time. If you sweep
through a line rapidly using a long time constant, you may ﬁlter out the signal
altogether! With a shorter, but still too long, time constant, the line shape will
be distorted, the apparent line center will shift, and the apparent line width will
increase. These eﬀects are illustrated in Figure 1.14. The best ruleofthumb is
to set the time constant to 10% of the time required to pass through the
narrowest line of your spectrum.
Figure 1.14 Eﬀect of time constant/time to sweep through the line (xaxis) on the
relative amplitude and signaltonoise ratio (lefthand axis) and relative
width and shift (righthand axis) of an ESR line.
16 Chapter 1
For example, suppose that the line width is 0.25 G, that you are scanning
through 10 G in 4 min. The scan rate then is 10 G/4 min ¼ 2.5 G min
À1
. The
time required to pass through a line is 0.25 G/2.5 G min
À1
¼ 0.1 min or 6 s. The
time constant should be set close to 0.6 s.
1.5 Applications of ESR Spectroscopy
1.5.1 Electronic Structure Determination
Most commonly, ESR spectroscopy is used to obtain information pertaining to
the electronic structure of the species being studied. We will explore these kinds
of applications extensively in subsequent chapters.
The magnetic ﬁeld values at which microwaves are absorbed to produce the
ESR spectrum of an isotropic sample, e.g., the spectrum of a freely tumbling
radical in liquid solution, can contain two principal types of chemically useful
information: (i) The hyperﬁne coupling pattern provides information on the
numbers and kinds of magnetic nuclei with which the unpaired electron
interacts. (ii) The spacing of the lines and the center of gravity of the spectrum
yield the hyperﬁne coupling constants, A
i
, and gvalue, g, which are related to
the way in which the unpaired electron spin density is distributed in the
molecule. In isotropic spectra the observed parameters are averaged over
rotation of the molecules. In solids the parameters may also depend on
molecular orientation relative to the magnetic ﬁeld, B.
Structural applications range from organic, inorganic and organometallic
radicals to coordination complexes and biological macromolecules containing a
paramagnetic center.
Even more information is available from the spectrum of a solidstate
sample, either a dilute single crystal or a frozen solution. We will discuss solid
state samples later, beginning in Chapter 4.
1.5.2 Analytical Applications
Like all forms of spectroscopy, ESR intensities, through the dependence on the
number of spins, N, can be used analytically to determine the concentration of a
paramagnetic species. Such analytical applications usually require a standard
solution to establish a calibrated intensity scale. Computer software is required
in most cases to doubly integrate the spectrum (two integrations are required to
convert the derivative spectrum into absorption and then to a number repre
senting the total concentration of the species being studied). The interested
reader is referred to Wertz and Bolton,
12a
who discuss the technique, point out
the variables that must be controlled, and suggest intensity standards. The
degree to which ESR intensities have been exploited varies widely. Experimen
talists often make qualitative observations relating ‘‘strong’’ or ‘‘weak’’ spectra
to chemical circumstances, but quantitative applications of integrated intensi
ties are rare in ESR studies.
17 Introduction
‘‘Spin traps’’ which scavenge a reactive radical to give a more persistent
radical may be used to detect the intermediacy of such transient radicals in
sorting out a reaction mechanism.
39
1.5.3 Determination of Rates
In some cases, ESR spectra can be used to determine the rate of a chemical
reaction or the rate of a conformation change. Such applications are discussed
in more detail in Chapter 5.
The spectral line widths are related to the rate of the rotational motions,
which average anisotropies in the g and hyperﬁne matrices (Chapter 5), and to
the rates of ﬂuxional processes, which average nuclear positions in a radical.
The saturation behavior of a spectrum – the variation of integrated intensity
with microwave power – is related to the spin–lattice relaxation time, a measure
of the rate of energy transfer between the electron spin and its surroundings.
Saturation often depends on the same structural and dynamic properties as line
widths.
Largely because spin–orbit coupling results in spin state admixture, elec
tronic spin–lattice relaxation times are normally short for species containing
heavy atoms such as transition metals. This has three consequences. Short
relaxation times mean that saturation problems, which plague ESR spec
troscopists studying organic radicals and NMR spectroscopists in general,
are largely absent in organometallic ESR studies. Thus, spectra ordinarily can
be recorded at full microwave power with salutary consequences for sensitivity.
However, relaxation times are most easily determined by measuring spectral
intensity as a function of microwave power in the saturation region. If relax
ation times are short, very high power is required, which is out of the range of
operation of most continuous wave spectrometers. Similarly, short relaxation
times imply broad lines and reduced sensitivity. The advent of modern high
power pulsed spectrometers
32,34,36
has made it possible to overcome some of the
restrictions arising from relaxation behavior.
Some workers have used ‘‘spin labels’’ attached to a membrane or biolog
ical macromolecule to study the motion of these components of living cells
(Chapter 5).
References
1. E. Zavoisky, J. Phys. USSR, 1945, 9, 211, 245.
2. (a) B. Bleaney and K.W.H. Stevens, Rep. Progr. Phys., 1953, 16, 108; (b)
K.D. Bowers and J. Owen, Rep. Progr. Phys., 1955, 18, 304; (c) D.M.S.
Bagguley and J. Owen, Rep. Progr. Phys., 1957, 20, 304.
3. (a) A. Abragam and B. Bleaney, Electron Paramagnetic Resonance of
Transition Ions, Clarendon Press, Oxford, 1970; (b) A. Abragam, The
Principles of Nuclear Magnetism, Clarendon Press, Oxford, 1961.
18 Chapter 1
4. A. Hudson, in Electron Spin Resonance, Specialist Periodical Report,
ed. M.C.R. Symons, Royal Chemical Society, London, 1988, 11A, 55.
Early monographs
5. D.J.E. Ingram, Free Radicals as Studied by Electron Spin Resonance,
Butterworth, London, 1958.
6. G.E. Pake, Paramagnetic Resonance: An Introductory Monograph,
W. A. Benjamin, New York, 1962.
7. S.A. Altshuler and B.M. Kozyrev, Electron Paramagnetic Resonance,
Academic Press, New York, 1964.
8. M. Berson and J.C. Baird, An Introduction to Electron Paramagnetic
Resonance, W. A. Benjamin, New York, 1966.
Standard texts and monographs
9. P.B. Ayscough, Electron Spin Resonance in Chemistry, Methuen, London, 1967.
10. H.M. Assenheim, Introduction to Electron Spin Resonance, Plenum Press,
New York, 1967.
11. R.S. Alger, Electron Paramagnetic Resonance, Techniques and Applications,
Interscience, New York, 1968.
12. (a) J.E. Wertz and J.R. Bolton, Electron Spin Resonance, McGrawHill,
New York, 1972; (b) J.A. Weil, J.R. Bolton and J.E. Wertz, Electron
Paramagnetic Resonance: Elementary Theory and Practical Applications,
WileyInterscience, New York, 1994.
13. (a) N.M. Atherton, Electron Spin Resonance: Theory and Applications, Ellis
Horwood, Chicester, 1973; (b) Principles of Electron Spin Resonance, Ellis
Horwood, Chicester, 1993.
14. L.A. Blumenfeld, V.V. Voevodski and A.G. Semenov, Monographs on
Electron Spin Resonance, Halsted Press, New York, 1974.
15. W. Gordy, Theory and Applications of Electron Spin Resonance, Wiley,
New York, 1980.
16. Handbook of Electron Spin Resonance, ed. C.P. Poole, Jr. and H.A. Farach,
2 vols., AIP Press, New York, (a) Vol. 1, 1994, (b) Vol. 2, 1999.
Instrumentation
17. T.H. Wilmshurst, Electron Spin Resonance Spectrometers, Plenum Press,
New York, 1967.
18. C.P. Poole, Jr., Electron Spin Resonance: A Comprehensive Treatise on
Experimental Techniques, Wiley, New York, 1983.
Biological applications
19. D.J.E. Ingram, Biological and Biochemical Applications of ESR, Adam
Hilger, London, 1969.
20. G. Feher, Electron Paramagnetic Resonance with Applications to Selected
Problems in Biology, Gordon and Breach, New York, 1970.
21. Biological Applications of Electron Spin Resonance, ed. H.M. Schwartz,
J.R. Bolton and D.C. Borg, WileyInterscience, New York, 1972.
19 Introduction
22. M.C.R. Symons, Chemical and Biochemical Aspects of Electron Spin
Resonance, John Wiley and Sons, New York, 1978.
23. Paramagnetic Resonance of Metallobiomolecules, ed. J. Telser, ACS Sym
posium Series, 858, Oxford Unversity Press, Oxford, 2003.
24. Biomedical EPR, ed. S.S. Eaton, G.R. Eaton and L.J. Berliner, Kluwer
Academic/Plenum Publishers, New York, 2004.
25. L.J. Berliner, In vivo EPR (ESR): Theory and Applications, Baker and
Taylor, Charlotte, NC, 2004.
Inorganic and organometallic applications
26. P.W. Atkins and M.C.R. Symons, The Structure of Inorganic Radicals,
Elsevier, Amsterdam, 1967.
27. J.R. Pilbrow, Transition Ion Electron Paramagnetic Resonance, Clarendon
Press, Oxford, 1990.
28. F.E. Mabbs and D. Collison, Electron Paramagnetic Resonance of d
Transition Metal Compounds, Elsevier, Amsterdam, 1992.
Organic free radicals
29. F. Gerson, High Resolution E.S.R. Spectroscopy, John Wiley and Sons
Ltd., London, 1970.
30. F. Gerson and W. Huber, Electron Spin Resonance Spectroscopy of Organic
Radicals, WileyVCH, Weinheim, 2003.
Advanced methods
31. L. Kevan and L.D. Kispert, Electron Spin Double Resonance Spectroscopy,
Wiley, New York, 1976.
32. Time Domain Electron Spin Resonance, ed. L. Kevan and R.N. Schwartz,
WileyInterscience, New York, 1979.
33. Advanced EPR – Applications in Biology and Biochemistry, ed. A.J. Hoﬀ,
Elsevier, Amsterdam, 1989.
34. Modern Pulsed and ContinueousWave Electron Spin Resonance, ed.
L. Kevan and M.K. Bowman, John Wiley, New York, 1990.
35. M. Ikeya, New Applications of Electron Spin Resonance: ESR Dating,
Dosimetry and Microscopy, World Scientiﬁc Publishing Co., London, 1993.
36. A. Schweiger and G. Jeschke, Principles of Pulse Electron Paramagnetic
Resonance, Oxford University Press, Oxford, 2001.
37. Foundations of Modern EPR, ed. G.R. Eaton, S.R. Eaton and
K.M. Salikhov, World Scientiﬁc, Singapore, 1998.
38. G. Breit and I.I. Rabi, Phys. Rev., 1931, 38, 2082.
39. D. Rehorek, Chem. Soc. Rev., 1991, 20, 341.
20 Chapter 1
CHAPTER 2
Isotropic ESR Spectra of
Organic Radicals
2.1 Isotropic ESR Spectra
2.1.1 Line Positions in Isotropic Spectra
As introduced in Chapter 1, ESR spectra of radicals in liquid solution are
usually interpreted in terms of a spin Hamiltonian:
^
H
S
¼ m
B
gB
^
S
z
þ
X
i
A
i
^
I
i
Á
^
S ð2:1Þ
where we have generalized eqn (1.7) to include the possibility of many nuclei,
each with an isotropic hyperﬁne coupling constant A
i
. The spectral infor
mation is contained in the parameters, g, the isotropic gfactor, and the set of
values of A
i
. Using spin functions based on the quantum numbers m
S
and m
I
,
eqn (2.1) can be used to compute energy levels. Equating energy diﬀer
ences for the allowed transitions (Dm
S
¼ Æ1, Dm
Ii
¼ 0 for all i) with the
microwave photon energy (hn) [eqn (2.2)] the resonant magnetic ﬁelds can be
predicted.
Eðm
s
¼ 1=2; m
Ii
Þ ÀEðm
s
¼ À1=2; m
Ii
Þ ¼ hn ð2:2Þ
As will be derived in Chapter 3, to ﬁrstorder in perturbation theory, the
resonant ﬁelds are:
B ¼ B
0
À
X
i
a
i
m
Ii
ð2:3Þ
where B
0
¼ hn/gm
B
represents the center of the spectrum and a
i
¼ A
i
/gm
B
is the
hyperﬁne coupling constant in ﬁeld units.
The coupling constants in eqn (2.1) have energy units, but the energies are
very small so that frequency (MHz) or wavenumber (10
À4
cm
À1
) units are more
commonly used. Even more often, however, isotropic coupling constants are
given in units of magnetic ﬁeld strength, since they may be derived directly from
line positions in magnetic ﬁeld units. The unit usually employed for a is gauss
(G) (or oersteds), although SI purists sometimes use millitesla (1 mT ¼ 10 G).
21
Conversions from units of gauss into frequency or wavenumber units involves
the gvalue:
A
i
ðMHzÞ ¼ 2:8025ðg=g
e
Þa
i
ðGÞ ð2:4aÞ
A
i
ð10
À4
cm
À1
Þ ¼ 0:93480 ðg=g
e
Þa
i
ðGÞ ð2:4bÞ
Note that coupling constants in 10
À4
cm
À1
are comparable in magnitude to
those expressed in gauss. Although the units used for isotropic hyperﬁne
coupling constants is largely a matter of taste, the components of an aniso
tropic hyperﬁne coupling matrix (Chapter 4) should be given in frequency or
wavenumber units unless the gmatrix is virtually isotropic.
2.1.2 Hyperﬁne Coupling Patterns
Nuclear hyperﬁne coupling results in a multiline ESR spectrum, analogous to
the spin–spin coupling multiplets of NMR spectra. ESR spectra are simpler to
understand than NMR spectra in that secondorder eﬀects normally do not
alter the intensities of components; on the other hand, ESR multiplets can be
much more complex when the electron interacts with several highspin nuclei,
and, as we will see in Chapter 3, there can also be considerable variation in line
width within a spectrum.
When several magnetically equivalent nuclei are present in a radical, some
of the multiplet lines appear at exactly the same ﬁeld position, i.e., are
‘‘degenerate’’, resulting in variations in component intensity. Equivalent spin
1/2 nuclei such as
1
H,
19
F, or
31
P result in multiplets with intensities given by
binomial coeﬃcients (1 : 1 for one nucleus, 1 : 2 : 1 for two, 1 : 3 : 3 : 1 for three,
1 : 4 : 6 : 4 : 1 for four, etc.). One of the ﬁrst aromatic organic radical anions
studied by ESR spectroscopy was the naphthalene anion radical,
1
the spectrum
of which is shown in Figure 2.2. The spectrum consists of 25 lines, a quintet of
quintets as expected for hyperﬁne coupling to two sets of four equivalent
protons.
Just as in NMR, a multiplet pattern gives an important clue to the identity of
a radical. For example, in the naphthalene anion radical, there are four a
(positions 1, 4, 5, 8) and four b protons (positions 2, 3, 6, 7). Each proton splits
the electronic energy levels in two. Since the a protons are equivalent, for
example, the splitting is the same for each proton. Thus, as shown on the right
side of Figure 2.1, ﬁve equally spaced energy level values result.
The degeneracies of the levels shown in Figure 2.1 can be obtained by the
following line of reasoning: Assuming that the probability that any one proton
is spin up is identical to the probability that it is spin down, the probability that
all four protons are spin up or spin down is (1/2)
4
¼ 1/16. Similarly, the
probability that three of the four protons are spin up or spin down is 4/16 and
the probability that two of the four protons are spin up or spin down is 6/16. If
this pattern sounds familiar, it should be since this is just the set of binomial
coeﬃcients for (a þ b)
4
¼ a
4
þ 4a
3
b þ 6a
2
b
2
þ 4ab
3
þ b
4
. We can generate these
22 Chapter 2
coeﬃcients for other numbers of equivalent protons by the familiar Pascal’s
triangle (Figure 2.11), which is similar in structure to Figure 2.1.
Naphthalene and other aromatic hydrocarbons can be reduced by one
electron to produce the anion radical. The reduction can be carried out with
sodium in an ether solvent or electrochemically in a polar aprotic solvent.
The naphthalene anion radical spectrum (Figure 2.2) provided several sur
prises when Samuel Weissman and his associates
1
ﬁrst obtained it in the early
1950s at Washington University in St. Louis. It was a surprise that such an
oddelectron species would be stable, but in the absence of air or other
oxidants, [C
10
H
8
]
À
is stable virtually indeﬁnitely. A second surprise was the
appearance of hyperﬁne coupling to the two sets of four equivalent protons.
The odd electron was presumed (correctly) to occupy a p* molecular orbital
Figure 2.1 Splitting pattern for one of the electron spin energy levels coupled to four
equivalent protons. Note that the degeneracies of the levels are not shown.
See Section 2.5 for details.
3340 3345 3350 3355 3360
Magnetic Field/Gauss
α
β
Figure 2.2 ESR spectrum of the naphthalene anion radical;
1
simulated using hyperﬁne
couplings given in Table 2.1.
23 Isotropic ESR Spectra of Organic Radicals
(MO) with the protons in the nodal plane. The mechanism of coupling
(discussed below) requires ‘‘contact’’ between the unpaired electron and the
proton, an apparent impossibility for p electrons that have a nodal plane at
the position of an attached proton. A third, pleasant, surprise was the ratio of
the magnitudes of the two couplings, 5.01 G/1.79 G ¼ 2.80. This ratio is
remarkably close to the ratio of spin densities at the a and b positions, 2.62,
predicted by simple Hu¨ ckel MO theory for an electron placed in the lowest
unoccupied MO (LUMO) of naphthalene (see Table 2.1). This result led to
Hu¨ ckel MO theory being used extensively in the semiquantitative interpre
tation of ESR spectra of aromatic hydrocarbon anion and cation radicals.
Things get a little more complicated when a spin 1 nucleus like
14
N is added
to the picture, but the same technique works again for the determination of
the relative intensities of the ESR lines. Consider, for example, the relative
intensities of the hyperﬁne lines arising from the pyrazine anion radical, whose
spectrum is shown in Figure 2.3. Like that of the naphthalene anion radical, the
spectrum observed for the pyrazine anion radical
2
consists of 25 wellresolved
Table 2.1 Hyperﬁne parameters and spin densities for aromatic radical
anions. (Data from ref. 11.)
Compound a
H
(G) r
p
Compound a
H
(G) r
p
Benzene (B) 3.75 0.167 Pyrene (Py) 4.75 0.136
C
8
H
8
(C) 3.21 0.125 1.09 0.000
Naphthalene (N) 5.01 0.181 2.08 0.087
1.79 0.069 Perylene (P) 3.09 0.083
Anthracene (A) 2.74 0.096 0.46 0.013
1.57 0.047 3.53 0.108
5.56 0.192 Biphenylene (Bi) 0.00 0.027
2.75 0.087
3325 3330 3335 3340 3345 3350 3355 3360 3365
Magnetic Field/Gauss
N
N
Figure 2.3 ESR spectrum of the potassium salt of pyrazine radical anion; simulated
using hyperﬁne couplings from ref. 2.
24 Chapter 2
lines. In fact, the spectra in Figures 2.2 and 2.3 appear almost identical and are
only distinguishable by careful examination of the intensities of the lines.
The intensities for pyrazine are derived in Figure 2.4, assuming that the ratio
of hyperﬁne couplings, a
N
/a
H
, is nearly the same as a
a
/a
b
for naphthalene anion
radical. The method used is the same as implied by the energy level splittings
shown in Figure 2.1. The pattern arising from the two equivalent
14
N nuclei
that have the larger hyperﬁne splitting are ﬁrst derived, giving ﬁve lines with
relative intensities of 1 : 2 : 3 : 2 : 1. Each of these is then split into ﬁve lines
with relative intensities of 1 : 4 : 6 : 4 : 1, corresponding to the four equivalent
protons. The relative intensities of all 25 lines are shown at the bottom of the
ﬁgure in the same order in which they are appear in the observed spectrum in
Figure 2.2. Section 2.5 gives a fuller explanation of this procedure.
2.1.3 Secondorder Splittings
Equation (2.3) describes line positions correctly for spectra with small hyperﬁne
coupling to two or more nuclei provided that the nuclei are not magnetically
equivalent. When two or more nuclei are completely equivalent, i.e., both
instantaneously equivalent and equivalent over a time average, then the nuclear
spins should be described in terms of the total nuclear spin quantum numbers
I and m
I
rather than the individual I
i
and m
Ii
. In this ‘‘coupled representation’’,
the degeneracies of some multiplet lines are lifted when secondorder shifts are
included. This can lead to extra lines and/or asymmetric line shapes. The eﬀect
was ﬁrst observed in the spectrum of the methyl radical, CH
3
d
, produced by
N
N
H
H
H
H

one spin1 nitrogen: 1 1 1
1 1 1
1 1 1
1 1 1
two spin1 nitrogens: 1 2 3 2 1
1 4 6 4 1
2 8 12 8 2
plus four spin1/2 protons: 3 12 18 12 3
2 8 12 8 2
1 4 6 4 1
Relative intensities: 1 4 6 2 4 8 1 12 3 8 12 2 18 2 12 8 3 12 1 8 4 2 6 4 1
Figure 2.4 Derivation of the relative intensities of the 25 ESR lines of the pyrazine
anion radical. See Section 2.5 for details of the method.
25 Isotropic ESR Spectra of Organic Radicals
highenergy electron irradiation of liquid methane, by Fessenden and Schuler.
3
The three equivalent protons lead to a nondegenerate nuclear spin state
with I ¼ 3/2 (m
I
¼ Æ3/2, Æ1/2) and a twofold degenerate state with I ¼ 1/2
(m
I
¼ Æ1/2). Thus, six hyperﬁne lines are observed under conditions of high
resolution (Figure 2.5).
2.1.4 Spin Hamiltonian Parameters from Spectra
Once a hyperﬁne pattern has been recognized, the line position information can
be summarized by the spin Hamiltonian parameters, g and a
i
. These parameters
can be extracted from spectra by a linear leastsquares ﬁt of experimental line
positions to eqn (2.3). However, for highspin nuclei and/or large couplings,
one soon ﬁnds that the lines are not evenly spaced as predicted by eqn (2.3)
and secondorder corrections must be made. Solving the spin Hamiltonian,
eqn (2.1), to second order in perturbation theory, eqn (2.3) becomes:
4
B ¼ B
0
À
X
i
a
i
m
I
i
þ a
2
i
=2B
Â Ã
I
i
ðI
i
þ1Þ Àm
2
I
i
h i
þ. . .
n o
ð2:5Þ
The magnitude of the secondorder corrections is often quite signiﬁcant. For
example, Figure 2.5 shows the ESR spectrum of methyl radical, CH
3
d
. Notice
that all lines are shifted to low ﬁeld, relative to the ﬁrstorder spectrum, but, as
expected from eqn (2.5), the shift depends on m
I
. Failure to account for the
secondorder terms in an analysis results in a signiﬁcant error in both the
gvalue and in the hyperﬁne coupling constants For large I
i
and a
i
, well
resolved spectra may warrant inclusion of third or higherorder corrections.
Since the thirdorder corrections involve cross terms among coupling constants,
in principle they permit determination of the relative signs of the coupling
constants.
5
However, in the example of Figure 2.5, the thirdorder corrections
3356 3357 3358
Magnetic Field/Gauss
3333 3334 3335 3379 3380 3381
3402 3403 3404
Figure 2.5 ESR spectrum of the methyl radical, CH
3
d
(note discontinuities in magnetic
ﬁeld axis). Simulated using hyperﬁne splitting from ref. 3 and eqn (2.5).
26 Chapter 2
amount to ca. 1 mG. Secondorder corrections are also required for accurate
determinations of gfactors of radicals with very narrow lines
6
or in properly
accounting for line positions in radicals exhibiting chemical exchange eﬀects.
7
Secondorder eﬀects on hyperﬁne structure in organometallic compounds are
discussed in Chapter 3.
2.2 Interpretation of Isotropic Parameters
When ESR spectra were obtained for the benzene anion radical, [C
6
H
6
]
À
d
, and
the methyl radical, CH
3
d
, the proton hyperﬁne coupling constants were found
to be 3.75 and 23.0 G, respectively, i.e. they diﬀer by about a factor of 6. Since
the carbon atom of CH
3
has a spin density corresponding to one unpaired
electron and the benzene anion carries an electron spin density of 1/6, the two
results suggest that the proton coupling to an electron in a porbital is
proportional to the spin density on the adjacent carbon atom:
a
H
¼ Q
H
CH
r
p
C
ð2:6Þ
where the parameter Q
H
CH
¼ 23.0 G (based on CH
3
), 22.5 G (based on [C
6
H
6
]
À
),
or À23.7 G based on a valencebond theoretical calculation by McConnell.
8
Karplus and Fraenkel went on to analyze the origins of
13
C couplings in
aromatic radicals.
9
Their results are summarized by eqn (2.7):
a
C
¼ S
C
þ
X
i
Q
C
CX
i
!
r
p
C
þ
X
i
Q
C
X
i
C
r
p
X
i
ð2:7Þ
where S
c
¼ À12.7 G, Q
C
CC
0
¼ þ14:4 G, Q
C
CH
¼ þ19:5 G, and Q
C
C
0
C
¼ À13:9 G.
The ﬁrst term refers to the spin density in the porbital of the carbon in question
and the second term from porbital spin density on atoms attached to that
carbon.
An isotropic hyperﬁne coupling to H can only arise through the socalled
Fermi contact interaction of the unpaired electron at the position of the
H nucleus; but this is apparently symmetry forbidden for organic pradicals
where the H nuclei lie in the nodal plane. The interaction arises in a slightly
more complicated way: ‘‘spin polarization’’. As shown in Figure 2.6, the C 2p
z
orbital has zero probability at the H nucleus, but there is signiﬁcant overlap of
the C 2p
z
and C–H s bond orbitals. The H 1s orbital is part of the sbond
orbital and the C 2p
z
part of the singlyoccupied p* MO (SOMO). In the
overlap region of these two MOs, there is a tendency for the unpaired spin in
the SOMO to polarize the pair of electrons in the bonding MO such that the
spins in the overlap region are parallel, necessarily leaving an oppositely
oriented spin near the H nucleus.
For aromatic hydrocarbon radical anions, this approach works pretty well.
Figure 2.7 shows a correlation plot of observed hyperﬁne splitting versus the
spin density calculated from Hu¨ ckel MO theory. It also correctly predicts the
negative sign of a
H
for protons attached to p systems.
27 Isotropic ESR Spectra of Organic Radicals
Examination of Figure 2.7 shows that the largest fractional errors are for the
smallest spin densities. The reason for this is related to the mechanism for
aromatic proton coupling. Positions with very small spin density, according to
Hu¨ ckel MO theory, tend to be polarized such that a negative spin density is
found at that site. Thus the zero spin density predicted for one position in
pyrene leads to a coupling of a bit more than 1 G as a result of this polarization
mechanism. Note that the sign of a coupling constant is not an experimental
result in ordinary ESR. The alternation of the sign of a
H
between positions was
H
C
π*orbital
σorbital
Figure 2.6 Schematic representation of spin polarization of a C–H sorbital by
electron spin in a p* orbital. Note that the polarization eﬀect is far from
complete; only a tiny fraction of the electron density near the H nucleus is
excess spindown.
0
1
2
3
4
5
6
a
H
/
G
a
u
s
s
0 0.05 0.1 0.15 0.2
Hückel MO spin density
A
A
Py
Py
N
N
B
Bi
A
P
P
P
C
Py
Bi
Figure 2.7 Correlation plot of observed
1
H coupling constant vs. computed spin
density from Hu¨ ckel MO theory. See Table 2.1 for identiﬁcation of points.
28 Chapter 2
conﬁrmed by performing NMR measurements of the Knight shift of pyrene
anion radical in the solid state.
10
It is sometimes assumed that there is a relation analogous to eqn (2.6) for metal
or ligand hyperﬁne couplings in spectra of organometallic radicals. Such an
assumption is usually unwarranted. An isotropic hyperﬁne coupling has three
contributions: (i) Fermi contact interaction between the nuclear spin and electron
spin density in the valenceshell sorbital; (ii) Fermi contact interaction between
the nuclear spin and spin density in innershell sorbitals arising from spin
polarization by unpaired electron density in valenceshell p or dorbitals; and
(iii) a contribution from spin–orbit coupling. The ﬁrst contribution is positive
(for a positive nuclear magnetic moment), the second is normally negative, and
the third may be of either sign. Because direct involvement of hydrogen 1s
character in the SOMO of an organic pradical is symmetryforbidden and spin–
orbit coupling is negligible in carbonbased radicals, proton couplings in such
radicals result solely from spin polarization and thus are proportional to the
polarizing spin density. In contrast, all three contributions are usually signiﬁcant
for organometallic radicals. Although there are a few cases where polarization
constants, analogous to Q
H
CH
, have been estimated, they are of use only in a more
complete analysis based on the results of a solidstate ESR study.
The situation with regard to protons that are attached to atoms adjacent to
p centers, such as the bprotons (methyl group) of the ethyl radical, CH
3
CH
2
d
, is
on somewhat ﬁrmer ground. In this case a
H
varies with the dihedral angle between
the C–H bond and the adjacent porbital containing the unpaired electron. This
was interpreted early on as evidence for ‘‘hyperconjugation’’ which delocalizes the
unpaired electron onto the adjacent proton. The dependence of a
H
on geometry
had also proven to be a very useful tool for studying the conformations of radicals.
Most ESR studies of organic radicals were carried out in the 1950s and
1960s. They provided important tests of early developments in valence theory.
The results of these early studies are nicely summarized in a review by
Bowers.
11
Applications of hyperﬁne splittings to structure determination are
discussed in many of the texts and monographs referenced in Chapter 1.
As we will see in Chapter 4, gmatrices are often diﬃcult to interpret reliably.
The interpretation of isotropic gvalues is even less useful and subject to
misinterpretation. Thus isotropic ESR spectra should be used to characterize
a radical by means of the hyperﬁne coupling pattern, to study its dynamical
properties through line width eﬀects, or to measure its concentration by
integration of the spectrum and comparison with an appropriate standard;
but considerable caution should be exercised in interpreting the gvalue or
nuclear hyperﬁne coupling constants.
2.3 Line Widths in Isotropic Spectra
2.3.1 Incomplete Averaging of Anisotropies
Careful examination of the isotropic ESR spectra of organic radicals may reveal
variations in line widths from one hyperﬁne component to another. Such eﬀects
29 Isotropic ESR Spectra of Organic Radicals
are enhanced in solvents of high viscosity or at low temperatures. The widths
can often be expressed as a power series in the nuclear spin quantum numbers:
Width ¼ a þ
X
i
b
i
m
i
þg
i
m
2
i
þ. . .
À Á
ð2:8Þ
Much of the width arises from incomplete averaging of anisotropies in the g
and hyperﬁne matrices (Chapter 3). For radicals with axial symmetry the
parameters of eqn (2.8) depend on Dg ¼ g

– g
>
, DA
i
¼ A
i,
– A
i,>
and t
R
,
the rotational correlation time:
12
a / a
0
þðDggBÞ
2
t
R
ð2:9aÞ
b / DggBDAt
R
ð2:9bÞ
g / ðDAÞ
2
t
R
ð2:9cÞ
Since these terms are proportional to t
R
, they increase with decreasing temper
ature.
w
There are several linewidth contributions, included in a
0
, which do not
depend on m
i
. These include magnetic ﬁeld inhomogeneity and the spin–rotation
interaction, the latter increasing with 1/t
R
and thus with increasing temperature.
These and other linewidth eﬀects have been studied in some detail and are
discussed elsewhere.
13
If the g and hyperﬁne anisotropies are known from analysis of a solidstate
spectrum, the linewidth parameters b
i
and g
i
can be used to compute the
rotational correlation time, t
R
, a useful measure of freedom of motion. Line
widths in ESR spectra of nitroxide spin labels, for example, have been used to
probe the motional freedom of biological macromolecules.
14
Since t
R
is related
to the molecular hydrodynamic volume, V
h
, and the solution viscosity, Z, by a
relationship introduced by Debye:
15
t
R
¼ 4pr
3
Z=3kT ¼ V
h
Z=kT ð2:10Þ
ESR line widths can provide a measure of the eﬀective size of a paramagnetic
molecule or ion – useful information when there is a suspicion that a radical has
polymerized.
2.3.2 Rates of Fluxionality from Line Widths
ESR line widths are also sensitive to processes that modulate the gvalue or
hyperﬁne coupling constants or limit the lifetime of the electron spin state. The
eﬀects are closely analogous to those observed in NMR spectra of dynamical
systems. However, since ESR line widths are typically on the order of 0.1–10 G
w
Equations (2.9a–c) assume that the product t
R
o
0
41, where o
0
is the (angular) frequency of the
microwaves. For Xband microwaves o
0
¼ 6Â10
10
rad s
À1
, so these relationships hold for
t
R
41.6Â10
À11
s. Using eqn (2.10) this corresponds to the tumbling time for a molecule with a
molecular weight of about 40 in water at room temperature. Equations (2.9a–c) should therefore
apply to all but the smallest, i.e., most rapidly tumbling, radicals in media of normal viscosity.
30 Chapter 2
3360 3365 3370 3375 3380 3385
3360 3365 3370 3375 3380 3385
Magnetic Field/Gauss
F CN
e

F CN CN
F
CN 2 CN NC
CN NC CN NC
e

Figure 2.8 ESR spectra resulting from the reduction of PhCN (bottom) and pFPhCN
(top). The top spectrum is identical to that of the 4,4
0
dicyanobiphenyl
anion radical. (Spectra were simulated using hyperﬁne couplings with
permission from ref. 16, copyright (1963) American Chemical Society.)
31 Isotropic ESR Spectra of Organic Radicals
(0.3–30 MHz), rate processes that give observable increases in line widths must
be fast. Such processes are discussed in detail in Chapter 5.
2.4 Organic Radical Reactions
Sometimes the ESR spectrum obtained when an organic molecule undergoes
oneelectron oxidation or reduction is not of the expected cation or anion radical.
Figure 2.8 shows spectra that result when benzonitrile and 4ﬂuorobenzonitrile
are reduced electrochemically.
16
The spectrum from the reduction of benzonitrile
is indeed that of [C
6
H
5
CN]
À
d
, but the spectrum resulting from the reduction of
FC
6
H
4
CN is not that of the expected radical anion. Careful examination of the
spectrum shows 1 : 4 : 6 : 4 : 1 quintets at either end of the spectrum, indicating
the presence of four equivalent protons. This suggests that F
À
is lost from the
initially formed radical anion and that the resulting neutral radical dimerizes. In
fact, the observed spectrum is identical to that obtained by oneelectron reduc
tion of 4,4
0
dicyanobiphenyl, the product of radical dimerization.
Another example from the same paper
16
involves the cyanocarbon anion,
1,1,2,3,3pentacyanopropene. Oneelectron reduction produces the expected
dianion radical, the spectrum of which is shown in Figure 2.9. However, further
reduction leads to a very diﬀerent spectrum, which results from loss of CN
À
from position 2, and attachment of a proton (presumably from the solvent
N,Ndimethylformamide) to produce the 1,1,3,3,tetracyanopropene dianion
radical the spectrum of which is also shown in Figure 2.9.
2.5 Analysis of Isotropic ESR Spectra
The analysis of an isotropic ESR spectrum is relatively straightforward if a
systematic approach is used, but it can be a frustrating experience for a
3340 3342 3344 3346 3348 3350 3352 3354 3356 3358 3360
Magnetic Field/Gauss
Figure 2.9 (Top) ESR spectrum of 1,1,2,3,3pentacyanopropene dianion radical;
(bottom) ESR spectrum of the dianion radical of 1,1,3,3tetracyanopro
pene. Spectra were simulated using hyperﬁne couplings with permission
from ref. 16, copyright (1963) American Chemical Society.
32 Chapter 2
beginner. Given below is a series of steps which, if followed with care, will result
in a successful interpretation for most reasonably wellresolved spectra.
2.5.1 Preliminary Examination of the Spectrum
(A) Check to see if the spectrum is symmetric in line positions and relative
intensities. If it is not, then most likely there are two or more radical species.
Variation of line widths with m
I
may, in principle, cause the spectrum to appear
unsymmetric, but in such a case line positions would still be at least approx
imately symmetrically distributed about the center.
(B) Is there a central line? If there is no central line, then there must be
an even number of lines, which suggests an odd number of halfinteger nuclei
(i.e., I ¼ 1/2, 3/2, etc.), which would cause splitting of any center line arising
from an even number of halfintegral nuclei or any number of nuclei with
integer spin (I ¼ 1, 2, etc.).
(C) Are the outermost lines visible above the noise? Observation of the lines
in the ‘‘wings’’ of the spectrum is often crucial to successful interpretation of
ESR spectra since less overlap of lines may occur there. In noisy spectra,
however, it may be diﬃcult to account for all the expected lines in the wings.
(D) How many lines are there? Count them carefully, watching for evidence
of unresolved features (shoulders or bumps).
(E) What is the ratio of the amplitudes of the most intense to least intense
lines in the spectrum?
2.5.2 What do you Expect to See?
(A) If the radical was produced by a chemically straightforward procedure,
you usually have some idea of the identity of the radical. How many and what
kinds of magnetic nuclei, i.e. nuclei with spins, should be present if your guess is
correct?
(B) How many lines are expected from this model? The total number of
nuclear spin states is (2I
1
þ 1) Â (2I
2
þ 1) Â (2I
3
þ 1) . . . . Thus, if the model
structure has six protons (I ¼ 1/2), there should be (2 Â 1/2 þ 1)
6
¼ 2
6
¼ 64
nuclear spin states. If some of the nuclei are expected to be equivalent, then the
number of lines will be less than the number of spin states, i.e., some of the spin
states will be degenerate (to ﬁrstorder in perturbation theory). Thus, if the six
protons are in three groups of two, it is as if you had three spin1 nuclei and you
expect (2 Â 1 þ 1)
3
¼ 3
3
¼ 27 distinct lines. If there is one group of four
equivalent protons and another group of two, then it is as if you had one spin2
nucleus and one spin1 nucleus and you expect (2 Â 2 þ 1)(2 Â 1þ1) ¼ 15 lines.
If all the nuclei are nonequivalent, then all lines should have the same
intensity (barring accidental overlap). If there are equivalent sets of nuclei, then
the expected ratio is the product of the largesttosmallest ratios of the various
multiplets. Thus, for the above examples, three groups of two spin1/2 nuclei
gives a 1 : 2 : 1 triplet of 1 : 2 : 1 triplets of 1 : 2 : 1 triplets; thus the maximum
33 Isotropic ESR Spectra of Organic Radicals
intensity ratio would be 2 Â 2 Â 2 ¼ 8. If we have a 1 : 4 : 6 : 4 : 1 quintet of
1 : 2 : 1 triplets, then the maximum intensity ratio will be 6 Â 2 ¼ 12.
2.5.3 Are the Gross Features of the Spectrum Consistent with the
Model?
(A) Compare the number of observed lines with the number expected. If
there are more lines than expected, either the model is wrong or there is more
than one radical contributing to the spectrum. If the expected and observed
numbers are equal, you are in luck – the analysis should be easy. If you see
fewer lines than expected (the most common case!), there may be accidental
superpositions, small amplitude lines buried under large ones, or just poor
resolution. The bigger the discrepancy between expected and observed numbers
of lines, the less deﬁnitive the analysis will be.
(B) Compare the expected and observed largesttosmallest intensity ratios.
Accidental degeneracies (or a m
I
2
linewidth dependence) may cause the
observed ratio to be bigger than expected. It is rarer (but not unknown) to
ﬁnd a smaller ratio than expected.
(C) Is the presence or absence of a central line consistent with the odd or even
number of expected lines?
(D) If the model still seems plausible after these tests, go on to a detailed
analysis. If there are gross inconsistencies, the model is probably wrong.
2.5.4 An Example
We can apply these rules to the three spectra shown in Figure 2.10. The radicals
in the ﬁgure are anion radicals of the three isomers of xylene (dimethylben
zene).
17
Let us see if we can ﬁgure out which is which.
The pxylene anion radical is easiest. We expect to have four equivalent ring
protons and six equivalent methyl protons. Thus we expect to see a 1 : 4 : 6 : 4 : 1
quintet of 1 : 6 : 15 : 20 : 15 : 6 : 1 septets. Spectrum (b) shows the quintet clearly,
but the CH
3
proton coupling is poorly resolved. No other isomer has a quintet,
so we can assign (b) to the pxylene radical anion.
The oxylene anion radical should have two pairs of ring protons and,
again, six equivalent methyl protons. Spectrum (a) shows a 1 : 2 : 1 triplet of
multiplets, which could well be the expected quintets. The outermost lines have
intensity ratios: 1 : 2 : 6 : 1 : 12 : 12À, suggesting that the methyl and ring proton
couplings are comparable. Apparently only one pair of ring protons has a
signiﬁcant coupling, but, with this assumption, spectrum (a) is consistent with
oxylene.
That leaves spectrum (c) for mxylene. We expect two unique protons, one
pair of equivalent ring protons, and six equivalent methyl protons. When we
examine the wings of the spectrum, we see a 1 : 6 : 15 : 20 : 15 : 6 set of doublets.
Table 2.2 gives the coupling constants for the three xylene anion radical
isomers.
34 Chapter 2
CH
3
H
3
C
H
H
H
CH
3
H
H
H
3
C
H H
H
CH
3
CH
3
H
H
oxylene mxylene pxylene
3330 3335 3340 3345 3350 3355
Magnetic Field/Gauss
(a)
(b)
(c)
H
H
Figure 2.10 ESR spectra of o, m, and pxylene radical anions (see text for assignment
of spectra). Spectrum (a) was simulated with permission using hyperﬁne
parameters from Ref. 17b, copyright (1964) American Institute of Physics;
spectra (b) and (c) were simulated with permission using hyperﬁne pa
rameters from ref. 17a, copyright (1961) Taylor and Francis (www.tandk.
co.uk).
Table 2.2 Hyperﬁne parameters for xylene radical anions
17
oXylene mXylene pXylene
Number a
H
(G) Number a
H
(G) Number a
H
(G)
2 ring H 6.93 1 ring H 6.85 4 ring H 5.34
2 ring H 1.81 1 ring H 7.72 6 Me H 0.10
6 Me H 2.00 2 ring H 1.46
6 Me H 2.26
35 Isotropic ESR Spectra of Organic Radicals
2.5.5 Detailed Analysis for Determination of Parameters
(A) Measure the positions and amplitudes of all the lines in the spectrum and
list them in order in a table (a spreadsheet program is convenient for this
purpose). A welldeﬁned measure of ‘‘position’’ in a complex spectrum is the x
axis point halfway between the maximum and minimum of the ﬁrstderivative
line. The ‘‘amplitude’’ is the diﬀerence in height between the maximum and
minimum. If convenient, measure the line positions in gauss; if this is inconven
ient, use arbitrary units such as inches, centimeters, or recorder chart boxes
measured from an arbitrary zero. In your table, also provide headings for the
quantum numbers (m
1
, m
2
, etc.) for each of the line positions, for the coupling
constants (a
1
, a
2
, etc.), and for the theoretical intensity (degeneracy) of each peak.
(B) The highest and lowest ﬁeld lines will always have theoretical intensities
of 1 and will have maximum positive and negative values for all the quantum
numbers. It does not really matter whether you start with positive quantum
numbers at the highﬁeld or lowﬁeld end of the spectrum – the signs of the
coupling constants are indeterminant – but, for consistency, let us agree to
assign positive quantum numbers to the lowest ﬁeld line (i.e., pretend that the
coupling constants are positive). Enter these assignments in the table. From this
point on, you can work from either end of the spectrum or even from both at
once. In the following, we assume that we are working from the lowﬁeld end.
(C) The spacing between the ﬁrst and second lines will be the smallest
coupling constant, a
1
. The intensity ratio of these two lines will usually indicate
the multiplet to which the coupling constant corresponds. Assign quantum
numbers to the second line, compute a
1
and enter these numbers in the table. If
you have started into a multiplet, you can then predict the positions and
intensities of the remaining lines of the multiplet. Find them and enter the
quantum numbers and new estimates of a
1
in the table.
(D) After ﬁnding all the lines of the ﬁrst a
1
multiplet, ﬁnd the ﬁrst unassigned
line as you move upﬁeld. This will correspond to the next larger coupling
constant a
2
. From the ratio of its amplitude to that of the end line, you can
usually ﬁgure out which multiplet this coupling constant corresponds to.
Assign the quantum numbers and compute the coupling constant a
2
. Enter
in the table. This line will also correspond to the ﬁrst line in a second a
1
multiplet, and, knowing a
1
, you can predict the positions and intensities of the
remaining lines of this multiplet. Find them and enter the quantum number
assignments and new estimates of a
1
and a
2
in the table.
(E) If a
2
corresponds to a multiplet, the positions of the remaining lines can
be predicted and located. Continue in this way through the spectrum. Small
discrepancies may arise through measurement errors or because of overlapped
lines. A position discrepancy larger than your estimated measurement accu
racy, however, may signal a misassignment or inconsistency with the model.
Notice that the spacing between the lowest and highest ﬁeld lines is equal to the
sum of the coupling constants. Thus, if you have found two coupling constants
and know the total extent of the spectrum, you can compute the sum of the
remaining couplings.
36 Chapter 2
(F) When all the lines have been assigned, average all the measured values
of each coupling constant and compute the standard deviation. If an arbitrary
position scale was used, convert the coupling constants into gauss. Least
squares ﬁtting of line positions to quantum numbers gives better statistics
than this method, but it is impossibly tedious to do by hand in all but the
simplest cases. Fortunately, computer programs and spreadsheet templates
for carrying out a multiparameter leastsquares analysis are now readily
available.
2.5.6 Computation of Multiplet Intensity Ratios
For sets of spin
1
2
nuclei, the multiplet intensity ratios are simply the binomial
coeﬃcients found most easily from Pascal’s triangle (Figure 2.11).
For higher spin nuclei, one can construct a splitting diagram. For example,
for two spin1 nuclei:
This procedure quickly gets out of hand, however, and a more compact
notation is preferable. Thus if we think of each of the three lines resulting from
coupling to the ﬁrst spin1 nucleus split into a 1 : 1 : 1 triplet, we have:
1 1 1
1 1 1
1 1 1
Add: 1 2 3 2 1
This procedure can be extended to three spin1 nuclei by thinking of each line
of a 1 : 1 : 1 triplet split into a 1 : 2 : 3 : 2 : 1 quintet:
1 2 3 2 1
1 2 3 2 1
1 2 3 2 1
Add: 1 3 6 7 6 3 1
For really complicated situations, this method can be used as the basis for a
computer algorithm.
37 Isotropic ESR Spectra of Organic Radicals
2.5.7 Multiplet Patterns due to Isotopomers
We often encounter cases where an element has one or more isotopes with a
nuclear spin and one or more isotopes with zero spin. The most common
example is that of carbon where 1.1% of naturally occurring carbon is
13
C with
I ¼ 1/2 and 98.9% is
12
C with I ¼ 0. Thus, in a carboncontaining radical, the
probability that any given carbon is
13
C is 0.011. Now suppose that there are
n equivalent carbon atoms in the radical. The probability that all n will be
12
C is
(0.989)
n
. The probability that one speciﬁc carbon is
13
C and the other nÀ1
12
C
will be (0.011)(0.989)
nÀ1
; but, since any one of the n carbons could be
13
C, we
see that the total probability of ﬁnding one
13
C is n(0.011)(0.989)
nÀ1
. Similarly,
the probability that two speciﬁc carbons are
13
C is (0.011)
2
(0.989)
nÀ2
, but any
one of the n carbons could be
13
C and any of the nÀ1 remaining carbons
could also be
13
C, so that the total probability of having two
13
Cs is
n(nÀ1)(0.011)
2
(0.989)
nÀ2
. Since the two
13
Cs are indistinguishable, however,
the coeﬃcient must be divided by 2 to avoid double counting. On reﬂection, we
see that the coeﬃcients 1, n, n(nÀ1)/2, are just binomial coeﬃcients and that
the probabilities of ﬁnding 0, 1, 2, etc.,
13
Cs out of n equivalent carbon atoms
are just the terms in the expansion of the expression:
ð0:989 þ0:011Þ
n
¼ð0:989Þ
n
þnð0:989Þ
nÀ1
ð0:011Þ
þnðn À1Þ=2ð0:989Þ
nÀ2
ð0:011Þ
2
þÁ Á Á
For example, in a radical with six equivalent carbon atoms (e.g., benzene
anion radical), the probabilities are: P
0
¼ 0.936, P
1
¼ 0.062, P
2
¼ 0.0017,
P
3
¼ 0.00003, etc. In practice, we would probably see only the central line with
intensity 0.936 and a pair of satellites with intensity 0.031, corresponding to
splitting of the center line by a single
13
C (the intensity is distributed between
the two resonances).
More complex patterns can arise from atoms with a higher fraction of
magnetic isotopes. Consider the case of a radical species with three equivalent
1
1 1
1 2 1
1 3 3 1
1 4 6 4 1
1 5 10 10 5 1
1 6 15 20 15 6 1
1 7 21 35 35 21 7 1
1 8 28 56 70 56 28 8 1
Figure 2.11 Pascal’s triangle for the determination of binomial coeﬃcients.
38 Chapter 2
platinum atoms.
195
Pt has I ¼ 1/2 and is 33.8% in natural abundance. Other
Pt isotopes have I ¼ 0. Thus the probabilities are: P
0
¼ 0.290, P
1
¼ 0.444,
P
2
¼ 0.227, and P
3
¼ 0.039. These four isotopomers will result, respectively, in
a singlet, a doublet, a 1 : 2 : 1 triplet, and a 1 : 3 : 3 : 1 quartet. However, since the
platinum atoms are assumed to be equivalent, the coupling constant will be
the same in each pattern and there will be some superpositions. Consequently,
seven hyperﬁne components are expected with positions and intensities
(Figure 2.12).
The relative intensities of the sevenline pattern then is expected to be
approximately: 1 : 12 : 49 : 84 : 49 : 12 : 1. With luck, all seven lines might be
observable. Patterns like this are more complicated to analyze, but they also
provide a ﬁngerprint, identifying the number of equivalent nuclei involved in
the multiplet pattern.
2.5.8 Secondorder Shifts in Line Positions
Our analysis thus far has assumed that solution of the spin Hamiltonian to ﬁrst
order in perturbation theory will suﬃce. This is often adequate, especially for
spectra of organic radicals, but when coupling constants are large (greater than
about 20 gauss) or when line widths are small (so that line positions can be very
accurately measured) secondorder eﬀects become important. As we see from
.290
0
.444/2
1 a
Pt
.227/2
2 .227/4
.039x3/8
3 .039/8
.404
SUM .237
.057
.005
Figure 2.12 Predicted hyperﬁne pattern for a radical with three equivalent Pt atoms.
Numbers on the left refer to the number of
195
Pt atoms in the radical. The
intensity labels indicate the normalized probabilities of the corresponding
lines which are derived from the isotope probabilities.
39 Isotropic ESR Spectra of Organic Radicals
eqn (2.5) for a single nucleus with nuclear spin I, the hyperﬁne line positions are
given to second order in perturbation theory by:
B ¼ B
0
Àam
I
À
a
2
2B
0
IðI þ1Þ Àm
2
I
Â Ã
ð2:11Þ
Thus, for example, a spin1 nucleus with a ¼ 20 G and B
0
¼ 3200 G will have
lines at:
B ¼ 3200 À20ðþ1Þ Àð0:0625Þð2 À1Þ ¼ 3179:94 G
B ¼ 3200 À20ð0Þ Àð0:0625Þð2 À0Þ ¼ 3199:88 G
B ¼ 3200 À20ðÀ1Þ Àð0:0625Þð2 À1Þ ¼ 3219:94 G
so that the line spacings are, respectively, 19.94 and 20.06 G, and the center line
is shifted to low ﬁeld (relative to the ﬁrstorder result) by 0.12 G. This is a rather
small eﬀect and would require careful measurements to notice. When a is
larger, the shifts are much more noticeable. Thus, for example, when a ¼100 G,
the line spacings become 98.44 and 101.56 G, a much more easily noticeable
discrepancy.
When a spectrum results from coupling to several equivalent nuclei with
couplings large enough to warrant secondorder corrections, the situation
becomes somewhat more complicated since we must then think of the quantum
numbers I and m
I
corresponding to the sum of the quantum numbers for a full
set of equivalent nuclei. The secondorder corrections then depend on the total
I rather than the individual I
i
s. Thus, for example, two equivalent spin1 nuclei
will give rise to an I ¼ 2 state with m
I
¼ Æ2, Æ1, and 0. There is only one way of
getting m
I
¼ Æ2 (each nucleus has m
Ii
¼ Æ1), but there are two ways of getting
m
I
¼ Æ1 (either of the nuclei can have m
Ii
¼ 0 and the other m
Ii
¼ Æ1), so there
must also be a I ¼ 1 state with m
I
¼ Æ1 and 0. There are three ways of getting a
m
I
¼ 0 state (either of the nuclei can have m
Ii
¼ Æ1 and the other m
Ii
¼ Æ1
or both nuclei can have m
Ii
¼ 0); thus there must be a I ¼ 0 state with m
I
¼ 0.
Substituting into eqn (2.11), to secondorder, the degeneracies of the
1 : 2 : 3 : 2 : 1 multiplet pattern are lifted and a total of (3)
2
¼ 9 individual
lines would be observed. Summarizing these results in Table 2.3, we have, for
B
0
¼ 3200 G, a ¼ 20 G:
Table 2.3 Secondorder shifts for two I ¼ 1 nuclei
m
I
I [I(I þ 1) – m
I
2
]/2 B (G)
2 2 1 3159.88
1 2 5/2 3179.69
1 1 1/2 3179.94
0 2 3 3199.62
0 1 1 3199.88
0 0 0 3200.00
–1 2 5/2 3219.69
–1 1 1/2 3219.94
–2 2 1 3239.88
40 Chapter 2
2.6 Related Techniques (ENDOR)
The instrumental method described in Chapter 1 and illustrated in this chapter
proves to be adequate for studying many of the samples to which chemists and
biologists wish to apply ESR. Indeed, even if more advanced techniques turn
out to be required, CW, ﬁeldswept Xband ESR still provides the most
convenient, and commonly used, method for preliminary examination of
samples that are known, or suspected, to possess interesting paramagnetic
properties. Nevertheless, for those who may need to extend their studies to
more advanced methods Appendix 2 lists several of the most useful techniques
and gives references to recent reviews and relevant papers that should serve as
an entry into the still developing literature on advanced ESR.
This chapter concludes with a brief description of one advanced technique,
Electron Nuclear Double Resonance (ENDOR), the capabilities for which,
unlike pulsed methods, may be added as a relatively minor modiﬁcation to
commercial CW ESR spectrometers.
ENDOR was ﬁrst developed in the mid1950s
18
by George Feher as a means
of determining unresolved nuclear hyperﬁne coupling parameters in solids. It
has since found many other applications, including simplifying the hyperﬁne
patterns of complex radicals in solution. The technique works by partially
saturating an ESR transition at a ﬁxed ﬁeld while simultaneously sweeping the
NMR frequency through nuclear spin transitions. The result, which is illus
trated in Figure 2.13 for a single spin1/2 nucleus, is an increase in absorption at
the microwave frequency when an appropriate NMR transition is irradiated. In
Figure 2.13 the vertical arrow on the left refers to the orientation of the electron
spin and the one on the right the nuclear spin. At equilibrium, the lower and
upper pairs of levels have relative populations indicated by 1þd and 1Àd,
respectively [see Section 1.2 (Chapter 1)]. In the example, saturation of one of
the ESR transitions initially equalizes the populations of the km and mm levels,
leading to a decrease in intensity of that transition. If the NMR transition from
Equilibrium
populations
1  δ
1  δ
1 + δ
1 + δ
Saturate one
ESR transition
1  δ
1 + δ
Saturated
populations
1  δ/2
1
Perturbed
populations
Microwave absorption spectrum
ν
N
– A/2
ν
N
+ A/2 ν
N
1
1 + δ
1
1  δ/2
↑↑
↑ ↑
↑↓
↓↑
Figure 2.13 Example of an ENDOR experiment for a single spin1/2 nucleus. See text
for explanation.
41 Isotropic ESR Spectra of Organic Radicals
mm to mk is then saturated by irradiation at the frequency of the corresponding
NMR transition at, in this example, n
N
þ A/2, the resulting population change
will lead to an increase in the population diﬀerence between the km and mm
levels and increased microwave absorption at the ESR frequency. Irradiation of
the other NMR transition, km to kk, at frequency n
N
– A/2, will also lead to an
increase in intensity of the same saturated ESR transition. The ENDOR
spectrum is generated by varying the NMR frequency and observing the
corresponding increase in intensity of the saturated ESR transition. Notably,
the example given here corresponds to the case where the hyperﬁne frequency,
A ¼ A/h, is less than the nuclear Zeeman frequency, n
N
. For the case where
A/2 4 n
N
, interpretation of the ENDOR frequencies is somewhat modiﬁed.
The interested reader should consult the relevant references in Appendix 2 or
one of the more general books on ESR.
An early example of an application of ENDOR to extract hyperﬁne couplings
from a very complex, poorly resolved ESR spectrum is that of Allendoefer and
Englemann
19
who studied a solution of bis(pmethoxyphenyl)nitroxide and
obtained the results shown in Table 2.4. The ﬁgures in parentheses indicate
the level of precision.
References
1. S.I. Weissman, J. Townsend, G.E. Paul and G.E. Pake, J. Chem. Phys.,
1953, 21, 2227.
2. E.W. Stone and A.H. Maki, J. Chem. Phys., 1963, 39, 1635.
3. R.W. Fessenden and R.H. Schuler, J. Chem. Phys., 1963, 39, 2147.
4. R.W. Fessenden, J. Chem. Phys., 1962, 37, 747.
5. R.W. Fessenden and R.H. Schuler, J. Chem. Phys., 1965, 43, 2704.
6. R.G. Lawler, J.R. Bolton, M. Karplus and G.K. Fraenkel, J. Chem. Phys.,
1967, 47, 2149.
7. (a) G.K. Fraenkel, J. Chem. Phys., 1965, 42, 4275; (b) R.D. Allendoerfer
and P.H. Rieger, J. Chem. Phys., 1967, 46, 3410.
8. H.M. McConnell, J. Chem. Phys., 1956, 24, 633.
9. M. Karplus and G.K. Fraenkel, J. Chem. Phys., 1961, 35, 1312.
10. M.E. Anderson, P.J. Zandstra and T.R. Tuttle, J. Chem. Phys., 1960
33, 1591.
11. K.W. Bowers, in Advances in Magnetic Resonance, ed. J. S. Waugh, 1965,
1, 317.
Table 2.4 Hyperﬁne parameters for bis( pmethoxyphenyl)nitroxide
19
Position N oH mH Me
a (G) 9.99(1) 1.962(2) 0.764(2) 0.249(2)
42 Chapter 2
12. (a) D. Kivelson, J. Chem. Phys., 1960, 33, 1094; (b) R. Wilson and
D. Kivelson, J. Chem. Phys., 1966, 44, 154; (c) P.W. Atkins and
D. Kivelson, J. Chem. Phys., 1966, 44, 169.
13. Electron Spin Relaxation in Liquids, ed. L.T. Muus and P.W. Atkins,
Plenum Press, New York, 1972.
14. H.M. McConnell, in Foundations of Modern EPR, ed. G.R. Eaton,
S.S. Eaton and K.M. Salikhov, World Scientiﬁc, Singapore, 1998, ch. G.5.
15. See, for example, N. Bloembergen, E.M. Purcell and R.V. Pound, Phys.
Rev., 1948, 73, 579.
16. P.H. Rieger, I. Bernal, W.H. Reinmuth and G.K. Fraenkel, J. Am. Chem.
Soc., 1963, 85, 683.
17. (a) J.R. Bolton and A. Carrington, Mol. Phys., 1961, 4, 497; (b) J.R. Bolton,
J. Chem. Phys., 1964, 41, 2455.
18. G. Feher, Phys. Rev., 1959, 114, 1219.
19. R.D. Allendoerfer and J.H. Englemann, Mol. Phys., 1971, 20, 569.
43 Isotropic ESR Spectra of Organic Radicals
CHAPTER 3
Isotropic Spectra of
Organometallic Radicals
Most of the information content from ESR spectra of organometallic radicals and
coordination complexes comes from dilute singlecrystal spectra or frozen solu
tion spectra. Nonetheless, there are some bits of information and applications that
come uniquely from isotropic spectra, and we discuss those aspects in this chapter.
3.1 Secondorder Eﬀects on Line Positions
As illustrated in Chapter 2, ESR spectra of radicals in liquid solution can
usually be interpreted in terms of the spin Hamiltonian:
^
H
s
¼ m
B
gB Á
^
S þ
¸
i
A
i
^
I
i
Á
^
S ð3:1Þ
This is a simpliﬁed Hamiltonian that ignores the direct interaction of any
nuclear spins with the applied ﬁeld, B. Because of the larger coupling, A
i
, to
most transition metal nuclei, however, it is often necessary to use secondorder
perturbation theory to accurately determine the isotropic parameters g and A.
Consider, for example, the ESR spectrum of vanadium(IV) in acidic aqueous
solution (Figure 3.1), where the species is [VO(H
2
O)
5
]
21
.
Clearly, the eight hyperﬁne lines (I ¼ 7/2 for
51
V) have diﬀerent widths; but
careful examination also shows that the line spacing varies, increasing with increas
ing B. To understand the origin of this eﬀect we must take a closer look at the
solutions to Eqn. (3.1) for the case of an unpaired electron interacting with a single
nucleus. This will lead us to a derivation of eqns (2.5) and (2.11) of Chapter 2.
Given the Hamiltonian eqn (3.1), it is reasonable to express the eigen
functions in terms of the electron and nuclear spin quantum numbers: m
S
,m
I
i.
Applying to this function only the two terms in the Hamiltonian operator that
involve the zdirection of the ﬁeld B we get:
^
H
s
m
s
; m
I
j i ¼ gm
B
B
^
S
z
þA
^
S
z
^
I
z
m
s
; m
I
j i
¼ m
s
gm
B
B þAm
s
m
I
ð Þ m
s
; m
I
j i
ð3:2Þ
44
Thus the zerothorder, or unperturbed, energy is:
E
ð0Þ
¼ m
s
gm
B
B þm
s
m
I
A
or, since m
S
¼ Æ1/2:
E
ð0Þ
¼ Æ
1
2
gm
B
B Æ
1
2
m
I
A
With I ¼ 7/2, m
I
¼ Æ7/2, Æ5/2, Æ3/2, Æ1/2. This ‘‘ﬁrstorder’’ solution
w
would
of course lead to eight lines, equally spaced.
Now consider the application of secondorder perturbation theory. First, we
need to represent the xÀy components of the spins, S
x
, S
y
, I
x
and I
y
, in terms of
their raising and lowering combinations:
S
Æ
¼ S
x
ÆiS
y
I
Æ
¼ I
x
ÆiI
y
S
x
¼
1
2
S
þ
þS
À
ð Þ I
x
¼
1
2
I
þ
þI
À
ð Þ
S
y
¼ À
i
2
S
þ
þS
À
ð Þ I
y
¼ À
i
2
I
þ
þI
À
ð Þ
3000 3200 3400 3600 3800
v = 9.5 GHz
B/Gauss
Figure 3.1 Isotropic ESR spectrum of 0.005 M vanadium(IV) in 1 M NaClO
4,
pH 2.0.
(Reproduced with permission from ref. 3, copyright (1975) American
Chemical Society.)
w
By convention, the hyperﬁne energy is usually referred to as a ‘‘ﬁrstorder’’ correction to the
Zeeman energy. This nomenclature will also be used here, despite the fact that the notation E
(0)
properly acknowledges that both terms are included in the ‘‘zerothorder’’ energy. Strictly
speaking, in this case there is no ﬁrstorder perturbation.
45 Isotropic Spectra of Organometallic Radicals
Thus the hyperﬁne term of the spin Hamiltonian becomes:
AS Á I ¼ A S
x
I
x
þS
y
I
y
þS
z
I
z
¼ AS
z
I
z
þ
1
2
A S
þ
I
À
þS
À
I
þ
ð Þ
Operating on the spin functions with the extra hyperﬁne operator then gives:
1
2
A
^
S
þ
^
I
À
þ
^
S
À
^
I
þ
1
2
; m
I
¼
A
2
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
IðI þ1Þ Àm
I
m
I
þ1 ð Þ
À
1
2
; m
I
þ1
1
2
A
^
S
þ
^
I
À
þ
^
S
À
^
I
þ
À
1
2
; m
I
¼
A
2
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
IðI þ1Þ Àm
I
m
I
À1 ð Þ
1
2
; m
I
À1
Thus the matrix elements are:
À
1
2
; m
I
þ1
^
H þ
1
2
; m
I
¼
A
2
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
IðI þ1Þ Àm
I
m
I
þ1 ð Þ
þ
1
2
; m
I
À1
^
H À
1
2
; m
I
¼
A
2
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
IðI þ1Þ Àm
I
m
I
À1 ð Þ
and the secondorder energy corrections are:
E
ð2Þ 1
2
; m
I
¼
A
2
IðI þ1Þ Àm
I
m
I
þ1 ð Þ ½
4 E
ð0Þ
þ
1
2
; m
I
ÀE
ð0Þ
À
1
2
; m
I
þ1
E
ð2Þ
À
1
2
; m
I
¼
A
2
IðI þ1Þ Àm
I
m
I
À1 ð Þ ½
4 E
ð0Þ
À
1
2
; m
I
ÀE
ð0Þ
þ
1
2
; m
I
À1
or since the zerothorder energy diﬀerences are:
E
ð0Þ
þ
1
2
; m
I
ÀE
ð0Þ
À
1
2
; m
I
þ1
¼ gm
B
B þ
1
2
A 1 þ2m
I
ð Þ
E
ð0Þ
À
1
2
; m
I
ÀE
ð0Þ
þ
1
2
; m
I
À1
¼ Àgm
B
B þ
1
2
A 1 À2m
I
ð Þ
E
ð2Þ 1
2
; m
I
¼
A
2
IðI þ1Þ Àm
I
m
I
þ1 ð Þ ½
4gm
B
B þ
1
2
A 1 þ2m
I
ð Þ
E
ð2Þ
À
1
2
; m
I
¼
A
2
IðI þ1Þ Àm
I
m
I
À1 ð Þ ½
4gm
B
B þ
1
2
A 1 À2m
I
ð Þ
We can neglect the hyperﬁne terms in the denominators without serious error
since 4gm
B
B )
1
2
Að1 Æ2m
I
Þ. Including them would lead to energies that are
thirdorder or higher in powers of A, making it necessary to use higher order
perturbation theory.
1
46 Chapter 3
The transition energies, correct to second order in A, are:
DE ¼ gm
B
B þm
I
A
þ
A
2
IðI þ1Þ Àm
I
m
I
þ1 ð Þ ½
4gm
B
B
À À
A
2
IðI þ1Þ Àm
I
m
I
À1 ð Þ ½
4gm
B
B
¼ gm
B
B þm
I
A þ
A
2
IðI þ1Þ Àm
2
I
2gm
B
B
and the resonant ﬁelds are:
B ¼ B
0
Àa m
I
À
a
2
IðI þ1Þ Àm
2
I
2B
ð3:3Þ
where
B
0
¼
hn
gm
B
a ¼
A
gm
B
For [VO(H
2
O)
5
]
21
(Fig. 3.1), g ¼ 1.964, a ¼ 116.4 G, and v ¼ 9.5000 GHz. The
predicted line positions (in Gauss) are given in Table 3.1.
3.2 Understanding the Variation in Line Widths
The widths of the lines in Fig. 3.1 vary because the anisotropies of g and A (to
be discussed in Chapter 4) are not completely averaged out when the molecule
or ion tumbles in solution. This issue was implicit in the classic work of
Bloembergen, Purcell and Pound
2a
on nuclear spin relaxation and was for
mulated in a useful way for EPR by Daniel Kivelson and coworkers.
2b–d
As
described in Chapter 2, they showed that the widths, in units of Hz, can often
be written as a power series in m
I
with terms up to secondorder (a thirdorder
term is sometimes signiﬁcant):
width ðm
I
Þ ¼ a þbm
I
þgm
2
I
ð3:4Þ
Table 3.1 Higher order eﬀects on line positions in [VO(H
2
O)
5
]
21
m
I
Firstorder
positions
Secondorder
positions
Exact
positions
a
Secondorder
spacing
b
7/2 3048.6 3040.8 3041.6 103.9
5/2 3165.0 3144.7 3146.2 108.8
3/2 3281.4 3253.5 3254.8 113.4
1/2 3397.8 3366.9 3367.2 117.4
À1/2 3514.2 3484.3 3483.7 121.1
À3/2 3630.6 3605.4 3604.1 124.4
À5/2 3747.0 3729.8 3728.4 127.5
À7/2 3863.4 3857.3 3856.6
a
Calculated to nearest 0.1 G using methods given in ref. 1 by iterative solution of the BreitRabi
equation.
b
1
st
order spacing of lines would be 116.4 G.
47 Isotropic Spectra of Organometallic Radicals
where
a / a
0
þðDggBÞ
2
t
R
ð3:5aÞ
b / DggBDAt
R
ð3:5bÞ
g / ðDAÞ
2
t
R
ð3:5cÞ
and t
R
is the rotational correlation time, usually expressed in an equation ﬁrst
introduced by Debye:
2
t
R
¼
4pr
3
k
3
Z
kT
ð3:6Þ
where Z is the viscosity of the solution, r is the radius of the (spherical) particle, and
k is a correction factor to account for deviations from sphericity. The rotational
correlation time therefore provides a measure of the ‘‘hydrodynamic volume’’ of
the tumbling molecule or ion. If values of Dg and DA can be determined
independently, e.g., from solid state spectra (Chapter 4), line width measurements
may be used to evaluate t
R
as a function of Z and T, allowing r to be estimated.
For example, dilute aqueous solutions of vanadium(IV) show an eightline
ESR spectrum in both acid (Fig. 3.1) and strong base, suggesting a monomeric
unit under all conditions. (
51
V has I ¼ 7/2 so more than eight lines would be
observed if more than one
51
V was coupled to the unpaired electron.) A likely
structure in base would be [VO(OH)
3
(H
2
O)
2
]
À
, formed by removal of three
protons from the species whose spectrum is shown in Fig. 3.1.
3
Previous work
on the system, however, was also consistent with the formation of polymers
such as V
3
O
7
2À
or V
4
O
9
2À
or even V
18
O
42
12À
at high pH. Comparison of values
of t
R
from the basic solution line widths with those obtained from spectra
of acidic solutions containing VO(H
2
O)
5
21
showed that the hydrodynamic
volume of the aquo cation is actually about twice that of the basic solution
species, eﬀectively ruling out the presence of ESRactive polymers in solution
(Figure 3.2).
4
The slopes correspond to r values of 3.4 A
˚
in acid solution and 2.7
A
˚
in basic solution. The conclusion is that vanadium(IV) is monomeric in dilute
basic solutions.
3.3 Puzzling Line Shapes
An example of secondorder splitting, in addition to the secondorder shifts
described above for a single nucleus, and related to the spectrum of methyl
radical discussed in Chapter 2, is seen in Fig. 3.3 for the isotropic ESR spectrum
of [PhCCo
3
(CO)
9
]
À
(ref. 5). The situation is complicated by the fact that
magnetically equivalent nuclei with I 41/2 give less familiar multiplet intensity
ratios. Thus, for example, three equivalent
59
Co nuclei (I ¼ 7/2) give (to ﬁrst
order) 22 lines with intensity ratios 1:3:6:10:15:21:28:36:42:46:48:48:46 . . . ,
(Figure 3.3a). The experimental spectrum of [PhCCo
3
(CO)
9
]
À
(Figure 3.3c)
48 Chapter 3
does indeed exhibit a spectrum with about the expected relative intensities, and
all but the weakest outermost pair of lines is visible. The observed line shapes,
however, are very unsymmetrical and correspond to a marked deviation from a
Lorentzian shape. Incorporating secondorder splittings into the spectrum as
follows solves the puzzle.
The combined spin angular momentum of three equivalent spin7/2
59
Co
nuclei is properly described in terms of 11 Jstates with J ranging from 21/2 to
1/2. The line positions through secondorder are then calculated using
Eqn. (3.3) substituting J for I. The m
J
¼ 17/2 feature, for example, has three
components with J ¼ 21/2, 19/2, and 17/2, degeneracies of 1, 2, and 3, and
secondorder shifts of 97a
2
/4B, 55a
2
/4B, and 17a
2
/4B, respectively. The shifts
are too small compared with the line width to be resolved, but they lead to
an asymmetric absorption line envelope with apparent broadening on the
lowﬁeld side, as shown in Figure 3.3(b) and in the experimental spectrum
(Figure 3.3c).
5
3.4 Use of ESR Spectra to Determine Formation
Constants
Reeder and Rieger
6
used ESR spectra to identify complex ions and to estimate
formation constants for aqueous oxovanadium(IV) complexes with lactic acid,
thiolactic acid, glycolic acid, and thioglycolic acid. Through the use of second
harmonic detection, which produces secondderivative spectra, the resolution
was good enough that several of the individual species could be separately
0
10
20
30
40
50
60
70
80
1 2 3 4 5 6 7 8 9
VO
2+
in 1.0 M NaClO
4
V(IV) in 1.0 M NaOH
η/T(10
−6
kg/msK)
τ
R
/
p
s
Figure 3.2 Plots of t
R
, determined from ESR line widths, vs. Z/T for [VO(H
2
O)
5
]
21
and for [VO(OH)
3
(H
2
O)
2
]
À
. Data from ref. 4.
49 Isotropic Spectra of Organometallic Radicals
detectable. The key requirement was that the
51
V coupling constant varied with
the nature of the ligands, the coupling constant decreasing with each additional
ligand coordinated. For each acid, H
2
A, the equilibria were:
VO
2þ
þHA
À
!
VOðHAÞ
þ
VOðHAÞ
þ
þHA
À
!VOðHAÞ
2
VOðHAÞ
þ
!
VOðAÞ þH
þ
VOðHAÞ
2
!VOðAÞðHAÞ
À
þH
þ
VOðAÞðHAÞ
À
!
VOðAÞ
2À
2
þH
þ
where HA
À
indicates ionization of the carboxyl group but retention of the
hydroxyl or sulfhydryl proton. For example, for VO
21
/lactic acid mixtures at
Figure 3.3 ‘‘Stick spectrum’’ showing hyperﬁne pattern for coupling to three equiva
lent
59
Co nuclei (I ¼ 7/2) computed to (a) ﬁrstorder and (b) secondorder
in perturbation theory. (Adapted from ref. 7.) (c) Isotropic ESR spectrum
of [PhCCo
3
(CO)
9
]
À
in THF solution at 401C.
(Experimental spectrum, reprinted from ref. 5 with permission, copyright
(1979) American Chemical Society.)
50 Chapter 3
pH 3.5, the lowﬁeld hyperﬁne line (in secondderivative presentation) appears
as shown in Figure 3.4.
Peak I was assigned to the superposition of peaks resulting from
VO(H
2
O)
5
21
and VO(HA)(H
2
O)
3
1
; peak II was assigned to VO(HA)
2
(H
2
O),
and peak III to the superposition of VO(A)(H
2
O)
3
and VO(A)(HA)(H
2
O)
–
.
From the variation in peak heights with pH, a reasonable set of formation
constants was deduced for each of the complexes.
References
1. N.M. Atherton, Electron Spin Resonance: Theory and Practice, John Wiley
and Sons, Inc., New York, 1973, section 3.12. For solutions through
fourthorder see also: R.W. Fessenden and R.H. Schuler, J. Chem. Phys.,
1965, 43, 2704.
2. (a) N. Bloembergen, E.M. Purcell and R.V. Pound, Phys. Rev., 1948
73, 679; (b) D. Kivelson, J. Chem. Phys., 1960, 33, 1094; (c) R. Wilson
and D. Kivelson, J. Chem. Phys., 1966, 44, 154; (d) P. W. Atkins and
D. Kivelson, J. Chem. Phys., 1966, 44, 169.
3. M.M. Iannuzzi and P.H. Rieger, Inorg. Chem., 1975, 14, 2895.
4. M.M. Iannuzzi, C.P. Kubiak and P.H. Rieger, J. Phys. Chem., 1976, 80,
541.
5. B.M. Peake, P.H. Rieger, B.H. Robinson and J. Simpson, Inorg. Chem.,
1979, 18, 1000.
6. R.R. Reeder and P.H. Rieger, Inorg. Chem., 1971, 10, 1258.
7. P.H. Rieger, Journal of Organometallic Chemistry Library, ed.
W.C. Trogler, Elsevier, Amsterdam, 1990, vol. 22, ch. 8.
I
II
III
Figure 3.4 Secondderivative ESR spectrum of the lowﬁeld line of the VO
21
octet for
a 1:4 mole ratio mixture of VO
21
and lactic acid at pH 3.5 and 251C. The
separation between peaks is approximately 25 Gauss.
(Reprinted with permission from ref. 6, copyright (1971) American Che
mical Society.)
51 Isotropic Spectra of Organometallic Radicals
CHAPTER 4
w
Anisotropic ESR Spectra
4.1 Introduction
The anisotropies that lead to line broadening in isotropic ESR spectra inﬂuence
solidstate spectra more directly. Accordingly a more complex spin Hamiltonian
is required to interpret such spectra:
^
H
s
¼ m
B
~
B Á g Á
~
S þ
¸
i
~
I
i
Á A
i
Á
~
S ð4:1Þ
In eqn (4.1), g and A
i
are 3 Â 3 matrices representing the anisotropic Zeeman
and nuclear hyperﬁne interactions. In general, a coordinate system can be found
– the gmatrix principal axes – in which g is diagonal. If g and A
i
are diagonal in
the same coordinate system, we say that their principal axes are coincident.
In species with two or more unpaired electrons, a ﬁne structure term must be
added to the spin Hamiltonian to represent electron spin–spin interactions. We
conﬁne our attention here to radicals with one unpaired electron (S ¼ 1/2) but
will address the S 4 1/2 problem in Chapter 6.
Nuclear quadrupole interactions introduce line shifts and forbidden transi
tions in spectra of radicals with nuclei having I 4 1/2. In practice, quadrupolar
eﬀects are observable only in very wellresolved spectra or in spectra of radicals
with nuclei having small magnetic moments and large quadrupole moments.
The most extreme case of a small magnetic moment to quadrupole moment
ratio is that of
191
Ir and
193
Ir. The spectra of [Ir(CN)
6
]
3À
(ref. 1), [Ir(CN)
5
Cl]
4À
and [Ir(CN)
4
Cl
2
]
4À
(ref. 2), and [Ir
2
(CO)
2
(PPh
3
)
2
(mRNNNR)
2
]
1
, R ¼ ptolyl
(ref. 3), show easily recognizable quadrupolar eﬀects. Other nuclei for which
quadrupolar eﬀects might be expected include
151
Eu/
153
Eu,
155
Gd and
157
Gd,
175
Lu,
181
Ta,
189
Os, and
197
Au. When quadrupolar eﬀects are important, it is
usually necessary to take account of the nuclear Zeeman interaction as well.
The nuclear quadrupole and nuclear Zeeman interactions add two more terms
w
This chapter has been taken in part from material ﬁrst published by Elsevier Publishing in
The Journal of Organometallic Chemistry Library Series, Organometallic Radical Processes, ed.
D. Trogler, 1990, ch. 8. The material from that chapter has been somewhat modiﬁed and is
republished here with permission from the publisher. Figures reprinted from the chapter are
designated by an ‘‘[E]’’.
52
to the spin Hamiltonian. Since these terms considerably complicate an already
complex situation, we will deal with quadrupole eﬀects in Chapter 7 and
conﬁne our attention here to nuclei for which quadrupolar eﬀects can be
neglected.
When a radical is oriented such that the magnetic ﬁeld direction is located by
the polar and azimuthal angles, y and j, relative to the gmatrix principal axes,
the resonant ﬁeld is given, to ﬁrst order in perturbation theory, by:
4
B ¼ B
0
À
¸
i
A
i
m
i
gm
B
ð4:2Þ
where
B
0
¼
hn
gm
B
ð4:3Þ
g
2
¼ g
2
x
cos
2
j þg
2
y
sin
2
j
sin
2
y þg
2
z
cos
2
y ð4:4Þ
A
2
i
¼ A
2
ix
S
2
ix
þA
2
iy
S
2
iy
þA
2
iz
S
2
iz
ð4:5Þ
S
ik
¼
g
x
sin y cos j l
ixk
þg
y
sin y cos j l
iyk
þg
z
cos y l
izk
g
ð4:6Þ
and the l
ijk
are direction cosines indicating the orientation of the kth principal
axis of the ith hyperﬁne matrix relative to the jth gmatrix principal axis. When
the matrix principal axes are coincident, only one of the l
ijk
in eqn (4.6) will be
nonzero. When the hyperﬁne matrix components are large, secondorder terms
4
must be added to eqn (4.2); these result in downﬁeld shifts, proportional to m
i
2
.
4.2 Solidstate ESR Spectra
So long as they are dilute (to avoid line broadening from intermolecular spin
exchange), radicals can be studied in the solid state as solutes in single crystals,
powders, glasses or frozen solutions. Radicals can be produced in situ by UV
or girradiation of a suitable precursor in a crystalline or glassy matrix. While
many organometallic radicals have been studied in this way,
5
it is often easier to
obtain solidstate ESR spectra by freezing the liquid solution in which the
radical is formed. Various techniques then can be used to generate radicals, e.g.,
chemical reactions, electrochemical reduction or oxidation, or photochemical
methods. Furthermore, the radical is studied under conditions more closely
approximating those in which its reaction chemistry is known. Not all solvents
give a satisfactory frozen solution. Highly symmetrical solvent molecules like
benzene tend to freeze with semicrystalline regions, which leads to undesirable
orientation of the solute molecules. The goal is a completely random
53 Anisotropic ESR Spectra
arrangement of solutes. Toluene gives a good random glass, as do mixtures of
CH
2
Cl
2
and C
2
H
4
Cl
2
or of tetrahydrofuran and CH
2
Cl
2
.
4.2.1 Spectra of Dilute Single Crystals
Spectra of radicals in a dilute single crystal are obtained for various orientat
ions, usually with the ﬁeld perpendicular to one of the crystal axes. Each
spectrum usually can be analyzed as if they were isotropic to obtain an eﬀective
gvalue and hyperﬁne coupling constants for that orientation. Since the g and
hyperﬁnematrix principal axes are not necessarily the same as the crystal axes,
the matrices, written in the crystal axis system, usually will have oﬀdiagonal
elements. Thus, for example, if spectra are obtained for various orientations in
the crystal xyplane, the eﬀective gvalue is:
g
2
j
¼ g
xx
cos j þg
yx
sin j
2
þ g
xy
cos j þg
yy
sin j
2
þ g
xz
cos j þg
yz
sin j
2
ð4:7Þ
or
g
2
j
¼ K
1
þK
2
cos 2j þK
3
sin 2j ð4:8Þ
where
K
1
¼
1
2
g
2
xx
þg
2
yy
þg
2
xz
þg
2
yz
þ2g
2
xy
ð4:9aÞ
K
2
¼
1
2
g
2
xx
Àg
2
yy
þg
2
xz
Àg
2
yz
ð4:9bÞ
K
3
¼ g
xy
g
xx
þg
xy
g
yy
þg
xz
g
yz
ð4:9cÞ
A sinusoidal plot of g
j
2
vs. j can be analyzed to determine K
1
, K
2
, and K
3
.
Exploration of another crystal plane gives another set of Ks that depend on
other combinations of the g
ij
; eventually enough data are obtained to determine
the six independent values of g
ij
(g is a symmetric matrix so that g
ij
¼ g
ji
). The
gmatrix is then diagonalized to obtain the principal values and the transforma
tion matrix, elements of which are the direction cosines of the gmatrix
principal axes relative to the crystal axes. An analogous treatment of the
eﬀective hyperﬁne coupling constants leads to the principal values of the A
2

matrix and the orientation of its principal axes in the crystal coordinate system.
In their 1967 book, Atkins and Symons
6a
summarized much of the early ESR
work on small inorganic radicals, many of which were produced by photolysis
or radiolysis of single crystals of the precursor molecules. A good example of
the application of the information that can be obtained from such singlecrystal
studies is the analysis of the spectrum of NO
3
, produced by girradiation of
KNO
3
crystals by Livingston and Zeldes.
6b
Table 4.1 summarizes the results.
54 Chapter 4
The data clearly show that NO
3
in this medium has axial symmetry, most
likely with a threefold axis passing through the nitrogen atom. Furthermore,
the relatively small
14
N hyperﬁne splitting is consistent with spin density
localized to a large degree on the oxygen atoms. While the geometry may in
principle also be obtained from vibrational and rotational spectra, the spin
density distribution would be inferred only indirectly without the ESR data.
Dilute solutions of welloriented paramagnetic molecules can also sometimes
be prepared by cocrystallization with a suitable diamagnetic host. Much of the
early work by Bleaney and coworkers (Chapter 1, ref. 2) on transition metal
ions was carried out this way. In such samples one may often obtain not only
the desired ESR parameters but also determine the orientation of the para
magnetic molecule within the host crystal. A good example of this approach is
the work by Collison and Mabbs
7
who studied [VO(mquin)
2
] (mquin ¼
2methylquinolin8olate) dissolved in a single crystal of [GaCl(mquin)
2
], using
both X and Qband spectrometers. As it happened, [GaCl(mquin)
2
] was not
the perfect host and the ESR matrix axes diﬀered from the crystal axes by 11.71.
Furthermore, the gmatrix axes were not coincident with the Amatrix axes,
being displaced in the xyplane by a ¼ 27.51. The frozen solution spectrum (in
toluene) of [VO(mquin)
2
] and the powder spectrum of [VO(mquin)
2
] in
[GaCl(mquin)
2
] are nearly identical. Simulations of the powder spectra at X
and Qband are shown in Figure 4.1. The spectra of Figure 4.1 illustrate one of
the advantages of higher frequencies over Xband ESR spectra: The various
gcomponents are better separated and the spectrum is more easily analyzed.
The next section deals with the analysis of frozen solution spectra.
4.2.2 Analysis of Frozen Solution Spectra
Since ESR spectra are normally recorded as the ﬁrst derivative of absorption vs.
ﬁeld, observable features in the spectrum of a powder correspond to molecular
orientations for which the derivative is large in magnitude or changes in sign.
For any spin Hamiltonian, there will be minimum and maximum resonant
ﬁelds at which the absorption changes rapidly from zero, leading to a large
value of the derivative and features that resemble positivegoing and negative
going absorption lines. Peaks in the absorption envelope correspond to deriv
ative sign changes and lead to features resembling isotropic derivative lines. The
interpretation of a powder spectrum thus depends on the connection of the
positions of these features to the g and hyperﬁnematrix components.
Frozen solution spectra must be taken in solvents that form good glasses. A
random but homogeneous distribution of the species to be studied is required to
obtain a wellresolved spectrum. Notably, some solvents, such as toluene, can
‘‘crack’’ when frozen, thus giving rise to more features than predicted as spectra
Table 4.1 ESR parameters for NO
3
6b
g
xx
¼ 2.0232 g
yy
¼ 2.0232 g
zz
¼ 2.0031
A
xx
¼ 3.46 G A
yy
¼ 3.46 G A
zz
¼ 4.31 G
55 Anisotropic ESR Spectra
from diﬀerent regions of the sample will be overlapped. Thawing and refreezing
of the sample often corrects the problem; but it is well to keep the predicted
number of features in mind when attempting to interpret the results. Drago has
given a list of good glassforming solvents and mixtures.
8
Early treatments of powder patterns attempted to deal with the spatial
distribution of resonant ﬁelds by analytical mathematics.
9
This approach led to
some valuable insights but the algebra is much too complex when nonaxial
hyperﬁne matrices are involved. Consider the simplest case: a single resonance
line without hyperﬁne structure. The resonant ﬁeld is given by eqn (4.3).
Features in the ﬁrst derivative spectrum correspond to discontinuities or
turning points in the absorption spectrum that arise when qB/qy or qB/qj
are zero:
@B
@y
¼
hn
m
B
g
2
z
Àg
2
?
g
3
sin y cos y ¼ 0 ð4:10aÞ
@B
@f
¼
hn
m
B
g
2
x
Àg
2
y
g
3
sin
2
y sin fcos f ¼ 0 ð4:10bÞ
and
g
2
?
¼ g
2
x
cos
2
j þg
2
y
sin
2
j
12000 12250 12500 12750 13000 13250 13500
1.949
1.988 1.983
2500 2750 3000 3250 3500 3750 4000
Magnetic Field/Gauss
1.949
1.988
1.983
mquin
N
O

CH
3
Figure 4.1 Powder spectrum of [VO(mquin)
2
] in [GaCl(mquin)
2
]: (bottom) 9.25 GHz
(Xband), (top) 34.99 GHz (Q band).
(Reproduced from ref. 7 with permission, copyright (1987) Royal Society
of Chemistry.)
56 Chapter 4
These equations have three solutions: (i) y ¼ 0; (ii) y ¼ 901, j ¼ 0; and (iii) y ¼
j ¼ 901. Since y and j are in the gmatrix axis system, observable features are
expected for those ﬁelds corresponding to orientations along the principal axes
of the gmatrix. This being the case, the principal values of the gmatrix are
obtained from a straightforward application of eqn (4.10).
Powder spectra with hyperﬁne structure often can be interpreted similarly
with spectral features identiﬁed with orientation of the magnetic ﬁeld along one
of the g and hyperﬁnematrix principal axes. However, this simple situation
often breaks down. Using a ﬁrstorder theory and one hyperﬁne coupling,
Ovchinnikov and Konstantinov
10
have shown that eqns (4.10) may have up to
six solutions corresponding to observable spectral features. Three of these
correspond to orientation of B along principal axes, but the ‘‘extra lines’’
correspond to less obvious orientations. Even more extra lines may creep in
when the spin Hamiltonian is treated to secondorder or when there is more
than one hyperﬁne coupling. The problem is illustrated by the resonant ﬁeld
vs. cos y and j surface shown in Figure 4.2, corresponding to m
Cu
¼ À3/2 in
the spectrum of Cu(acac)
2
(g ¼ 2.0527, 2.0570, 2.2514; A
Cu
¼ 27.0, 19.5,
Figure 4.2 [E] Resonant ﬁeld as a function of cos y and j for the m
Cu
¼ À3/2 ‘‘line’’ of
the frozen solution spectrum of Cu(acac)
2
, ESR parameters from ref 10.
(Figure reproduced from ref. 13.)
57 Anisotropic ESR Spectra
193.4 Â 10
À4
cm
À1
).
10
The minimum resonant ﬁeld, B ¼ 3290.7 G, corresponds
to B along the zaxis (cosy ¼ Æ1). With B along the xaxis (cosy ¼ 0, j ¼ 01),
the surface shows a saddle point at 3344.3 G (which corresponds to a maximum
in the absorption spectrum and therefore appears with closely spaced maxi
mum and minimum ﬁrstderivative features), and with B along the yaxis (cosy
¼ 0, j ¼ 901) there is a local minimum at 3325.5 G. In addition, another saddle
point occurs in the yzplane at B ¼ 3371.2 G (cosy ¼ Æ0.482, j ¼ 901); the only
maximum is in the xzplane at B ¼ 3379.0 G (cosy ¼ Æ0.459, j ¼ 01). Thus ﬁve
features are expected and indeed are shown in the computersimulated spec
trum of Cu(acac)
2
shown in Figure 4.3. Interestingly, the two most intense
features, highﬁeld, correspond to oﬀaxis ﬁeld orientations and thus are ‘‘extra
lines’’. The situation is more complex when the g and hyperﬁnematrix
principal axes are noncoincident (see below); in this case, none of the features
need correspond to the orientation of B along a principal axis direction.
Since the analytical approach is so complicated, powder patterns have
usually been analyzed by comparing the experimental spectrum with a com
putersimulated spectrum using estimates of the g and hyperﬁnematrix com
ponents.
11
Parameters are then adjusted and another simulation computed
until the ﬁt is satisfactory (or the experimentalist loses patience with the
problem). The most straightforward computer simulation method
10
involves
computation of the resonant magnetic ﬁeld using eqn (4.2) for ca. 10
5
values of
cosy and j for each set of nuclear spin quantum numbers. The ﬁeld is then
divided into equal increments and the number of times the resonant ﬁeld falls
between B
i
and B
i11
is plotted vs. B to give an approximation to the unbroad
ened absorption spectrum. The absorption spectrum is then broadened by
numerical convolution with a line shape function and diﬀerentiated to give the
desired simulation. Although such a ‘‘cut and try’’ approach to spectrum
Figure 4.3 [E] Computersimulated ﬁrstderivative ESR powder spectrum of Cu(acac)
2
.
(a) Features corresponding to m
Cu
¼ À3/2, (b) the complete spectrum.
58 Chapter 4
analysis works reasonably well when there are small numbers of parameters,
analysis of a complex spectrum is exceedingly tedious.
DeGray and Rieger
12
have developed a computer algorithm to locate powder
pattern features in spectra, given estimates of the spin Hamiltonian parameters.
The method employs a search of the resonant ﬁeld surface in cosy, j space for
maxima, minima, and saddle points. Since the search procedure requires
computation of B for only B100 orientations, the method is much faster than
a complete simulation. The predicted locations of spectral features are then
compared with the experimental values and the parameters are reﬁned using a
nonlinear leastsquares method. Using this method, relatively complex powder
patterns can be analyzed, provided that the spectrum is suﬃciently well
resolved that enough features can be located and identiﬁed to provide statis
tically signiﬁcant estimates of the parameters. Even with leastsquares ﬁtting,
however, comparison of the experimental spectrum with a highresolution
computer simulation is required to check the assignments of spectral features.
4.3 Interpretation of the gMatrix
The gvalue of a free electron is a scalar, g
e
¼ 2.00232. In a radical species, g
becomes a matrix because of the admixture of orbital angular momentum into S
through spin–orbit coupling. The components of the gmatrix thus diﬀer from g
e
to the extent that p, d, or forbital character has been incorporated, and they
diﬀer from one another, depending on which p, d, or forbitals are involved.
In general, the components of the gmatrix are given by:
g
ij
¼ g
e
d
ij
þ2
¸
k
¸
m6¼0
z
k
m l
ki
j j0 h i 0 l
kj
m
E
0
ÀE
m
ð4:11Þ
where the indices i and j refer to molecular coordinate axes (x, y, z), k sums over
atoms with unpaired electron density, and m sums over ﬁlled and empty
molecular orbitals with energy E
m
(E
0
is the energy of the SOMO); z
k
is the
spin–orbit coupling constant for atom k, and l
ki
is the icomponent orbital
angular momentum operator for atom k. The integrals hml
ki
ni are easily
computed for an electron centered on a single atom if the MOs are written as
linear combinations of real p or d atomic orbitals. Table 4.2 shows the results of
operation by l
i
on these functions. Thus, for example, in the usual shorthand
notation for p and d orbitals:
zjl
x
jy h i ¼ i and z
2
jl
y
jxz
¼ À
ﬃﬃﬃ
3
p
i
The orbital angular momentum operations needed to calculate integrals for
other orbitals are summarized in Table 4.2.
Notice that d
z
2 is unique among the dorbitals in that l
z
does not couple it to
any other orbital. Thus if the major metal contribution to the SOMO is d
z
2 , g
z
will be close to the free electron value. Accordingly, when one gmatrix
59 Anisotropic ESR Spectra
component is found close to the free electron value, it is often taken as evidence
for a d
z
2 based SOMO; such reasoning should be applied with caution, how
ever, since cancellation of negative and positive terms in eqn (4.11) could have
the same eﬀect.
Spin–orbit coupling to empty MOs (E
0
ÀE
m
o 0) gives a negative contribu
tion to g
ij
whereas coupling to ﬁlled MOs has the opposite eﬀect. Thus, for
example, ESR spectra of d
1
vanadium(IV) complexes generally have gvalues
less than g
e
(admixture of empty MOs) whereas d
9
copper(II) complexes have
gvalues greater than g
e
(admixture of ﬁlled MOs).
Since the gmatrix has only three principal values and there are almost
always many potentially interacting molecular orbitals, there is rarely suﬃcient
information to interpret a gmatrix with complete conﬁdence. When a well
resolved and reliably assigned optical spectrum is available, the energy diﬀer
ences, E
0
ÀE
m
, are known and can be used in eqn (4.11) to estimate the
contribution of the corresponding MOs to the gmatrix. Extended Hu¨ ckel
MO (EHMO) calculations can be useful (but do not trust EHMO energies!),
but one is most commonly reduced to arguments designed to show that the
observed gmatrix is consistent with the interpretation placed on the hyperﬁne
matrix.
4.4 Interpretation of the Hyperﬁne Matrix
Electron–nuclear hyperﬁne coupling arises mainly through two mechanisms: (i)
The Fermi contact interaction between the nuclear spin and selectron spin
density; this contribution, designated A, is isotropic and has been discussed in
Chapters 2 and 3. (ii) The electron spin–nuclear spin magnetic dipolar interac
tion; this contribution is almost entirely anisotropic, i.e., neglecting spin–orbit
coupling, the average dipolar contribution to the hyperﬁne coupling is zero.
The general form of the dipolar contribution to the hyperﬁne term of the
Hamiltonian is:
^
H
dipolar
¼ g
e
g
N
m
B
m
N
c
SOMO
~
S Á
~
I
r
3
À
3ð
~
S Á~rÞð
~
I Á~rÞ
r
5
c
SOMO
¸
ð4:12Þ
Table 4.2 [E] Angular momentum operations on the real p and d orbitals
l
x
l
y
l
z
xi 0 Àizi iyi
yi izi 0 Àixi
zi Àiyi ixi 0
x
2
Ày
2
i Àiyzi Àixzi 2ixyi
xyi ixzi Àiyzi À2ix
2
Ày
2
i
yzi ijx
2
Ày
2
i þ
ﬃﬃﬃ
3
p
ijz
2
i ixyi Àixzi
xzi Àixyi ijx
2
Ày
2
i À
ﬃﬃﬃ
3
p
ijz
2
i iyzi
z
2
i À
ﬃﬃﬃ
3
p
ijyzi
ﬃﬃﬃ
3
p
ijxzi 0
60 Chapter 4
where g
e
and g
N
are the electron and nuclear gvalues, m
B
and m
N
are the Bohr
and nuclear magnetons, and the matrix element is evaluated by integration over
the spatial coordinates, leaving the spins as operators. Equation (4.12) can then
be written:
^
H
dipolar
¼
~
I Á A
d
Á
~
S ð4:13Þ
where A
d
is the dipolar contribution to the hyperﬁne matrix, and the total
hyperﬁne coupling is:
A ¼ AE þA
d
ð4:14Þ
(E is the unit matrix). In evaluating the matrix element of eqn (4.12), the
integration over the angular variables is quite straightforward.
14
The integral
over r, however, requires a good atomic orbital wavefunction. Ordinarily, the
integral is combined with the constants as a parameter:
P ¼ g
e
g
N
m
B
m
N
r
À3
ð4:15Þ
P has been computed using Hartree–Fock atomic orbital wavefunctions and
can be found in several published tabulations
14–17
and in Appendix 1. Because
of the hr
À3
i dependence of P, dipolar coupling of a nuclear spin with electron
spin density on another atom is usually negligible.
If an atom contributes p
x
, p
y
, and p
z
atomic orbitals to the SOMO:
c
x
jxi þc
y
jyi þc
z
jzi ð4:16Þ
the total porbital spin density is (in the Hu¨ ckel approximation, i.e., neglecting
overlap):
r
p
¼ c
2
x
þc
2
y
þc
2
z
ð4:17Þ
and the dipolar contribution to the hyperﬁne matrix can be written:
ðA
d
Þ
ij
¼
2
5
Pl
ij
ð4:18Þ
where the l
ij
are:
l
xx
¼ 2c
2
x
Àc
2
y
Àc
2
z
ð4:19aÞ
l
yy
¼ 2c
2
y
Àc
2
x
Àc
2
z
ð4:19bÞ
l
zz
¼ 2c
2
z
Àc
2
x
Àc
2
y
ð4:19cÞ
l
ij
¼ À3c
i
c
j
ði 6¼ jÞ ð4:19dÞ
61 Anisotropic ESR Spectra
The factor of 2/5 and the weighting of the orbital coeﬃcients is determined by
the angular factors.
14
Equations (4.17–4.19) can therefore be combined in
matrix notation to write the dipolar hyperﬁne matrix for porbitals as:
A
d
¼ ð2=5ÞPr
p
2 0 0
0 À1 0
0 0 À1
¸
¸
ð4:20Þ
where r
p
, the spin density, is deﬁned by eqn (4.17). The porbital axis corre
sponds to the positive principal value of the matrix. When the porbitals are
written as hybrids, the orbital shape is unchanged, but the principal axes of the
hyperﬁne matrix, which reﬂect the spatial orientation of the hybrid porbital,
diﬀer from those in which the SOMO was formulated. Thus, for example, a p
hybrid with c
x
¼ c
z
and c
y
¼ 0 corresponds to a porbital with the major axis in
the xzplane and halfway between the x and zaxes (Euler angle b ¼ 451).
Similarly, if an atom contributes d atomic orbitals to the SOMO,
c
z
2 jz
2
þc
yz
jyz
þc
xz
jxzi þc
x
2
Ày
2 jx
2
Ày
2
þc
xy
jxy
ð4:21Þ
the total dorbital spin density is (in the Hu¨ ckel approximation):
r
d
¼ c
z
2 ð Þ
2
þ c
yz
2
þ c
xz
ð Þ
2
þ c
x
2
Ày
2
2
þ c
xy
2
ð4:22Þ
and the dipolar contribution to the hyperﬁne matrix is:
18
ðA
d
Þ
ij
¼
2
7
Pl
ij
ð4:23Þ
where P is given by eqn (4.15) and the l
ij
are:
l
xx
¼À c
z
2 ð Þ
2
À2 c
yz
2
þ c
xz
ð Þ
2
þ c
x
2
Ày
2
2
þ c
xy
2
À2
ﬃﬃﬃ
3
p
c
z
2 ð Þ c
x
2
Ày
2
ð4:24aÞ
l
yy
¼À c
z
2 ð Þ
2
À2 c
yz
2
þ c
xz
ð Þ
2
þ c
x
2
Ày
2
2
þ c
xy
2
þ2
ﬃﬃﬃ
3
p
c
z
2 ð Þ c
x
2
Ày
2
ð4:24bÞ
l
zz
¼ 2 c
z
2 ð Þ
2
þ c
yz
2
þ c
xz
ð Þ
2
À2 c
x
2
Ày
2
2
À2 c
xy
2
ð4:24cÞ
l
xy
¼ À2
ﬃﬃﬃ
3
p
c
z
2 ð Þ c
xy
þ3 c
yz
c
xz
ð Þ ð4:24dÞ
l
yz
¼
ﬃﬃﬃ
3
p
c
z
2 ð Þ c
yz
þ3 c
xz
ð Þ c
xy
À3 c
yz
c
x
2
Ày
2
ð4:24eÞ
l
xz
¼
ﬃﬃﬃ
3
p
c
z
2 ð Þ c
xz
ð Þ þ3 c
yz
c
xy
þ3 c
xz
ð Þ c
x
2
Ày
2
ð4:24fÞ
62 Chapter 4
The dipolar contribution to the hyperﬁne matrix for a pure dorbital is:
A
d
¼ ðÆ
2
7
ÞPr
d
2 0 0
0 À1 0
0 0 À1
¸
¸
ð4:25Þ
where the positive sign applies for d
z
2 and the negative sign to the other
four orbitals. Hybrid combinations of d
yz
, d
xz
, and d
xy
or d
x
2
Ày
2 and d
xy
give
a dorbital of the same shape and the same dipolar matrix, though the
principal axes in general are diﬀerent from the axes in which the SOMO was
formulated. Other hybrid orbitals are generally of diﬀerent shape, reﬂected by
diﬀerent principal values of the dipolar matrix, usually with diﬀerent principal
axes.
Spin–orbit coupling perturbs these results, adding terms to the diagonal
matrix components on the order of P(g
i
Àg
e
). These can be neglected only
when the gmatrix anisotropy is small. Calculation of the spin–orbit coupling
corrections is fairly straightforward for mononuclear complexes where the
SOMO is composed mainly of dorbitals from a single metal.
19–21
In radicals
with two or more transition metals, the spin–orbit coupling calculation is
seriously nontrivial. A major part of the problem is that the solution must
be gaugeinvariant, i.e., it must not depend on the choice of coordinate
system. This problem was addressed in the context of spin–orbit coupling
corrections to the gmatrix,
22
with eqn (4.11) as the result, but it has received
only cursory examination with regard to spin–orbit contributions to hyperﬁne
matrices.
23
Fortunately, polynuclear radicals containing ﬁrstrow transition
metals generally have gmatrix components suﬃciently close to g
e
that the
problem can be ignored. As organometallic radicals with second and third
row transition metals appear, the problem will become more urgent; it is to
be hoped that some future theoretician will deem the problem worthy of
attention.
The AO composition of the SOMO can often be deduced from the dipolar
hyperﬁne matrix, particularly when the radical has enough symmetry to restrict
possible hybridization. Thus an axial hyperﬁne matrix can usually be inter
preted in terms of coupling to a SOMO composed of a single p or dorbital. A
departure from axial symmetry may be due to spin–orbit coupling eﬀects, if (for
example) A

¼A
z
and A
x
ÀA
y
EP(g
x
Àg
y
). If the departure from axial symmetry
is larger, it is usually caused by dorbital hybridization. The procedure is best
illustrated by examples.
4.5 Organometallic Examples
4.5.1 A Lowspin Manganese(II) Complex
The spectrum of the lowspin manganese(II) complex, [Mn(dppe)
2
(CO)(CN
Bu)]
21
, (dppe ¼ Ph
2
PCH
2
CH
2
PPh
2
), in a CH
2
Cl
2
/THF glass is shown in
Figure 4.4(a).
24
The spin Hamiltonian parameters, obtained from leastsquares
63 Anisotropic ESR Spectra
ﬁtting of the ﬁeld positions of the spectral features, are given in Table 4.3, and a
simulation based on those parameters is shown in Figure 4.4(b). In this case the
principle axes of the g and hyperﬁnematrices are coincident.
[Mn(dppe)
2
(CO)(CNBu)]
21
has approximate C
2v
symmetry, although the
actual symmetry is reduced to C
2
, depending on the conformation of the
CH
2
CH
2
bridges of the dppe ligands. Since it has a nominal d
5
conﬁguration,
the SOMO is expected to be one of the ‘‘t
2g
’’ orbitals of an idealized octahedral
complex – d
xz
(b
1
), d
yz
(b
2
), or d
x
2
Ày
2 (a
1
), where the representations refer to C
2v
.
The energies of the d
xz
and d
yz
orbitals are expected to be lowered by
Figure 4.4 [E] (a) ESR spectrum of [Mn(dppe)
2
(CO)(CNBu)]
21
in frozen CH
2
Cl
2
/
THF glass.
(Reproduced with permission from ref. 24, copyright (1987) Royal
Society of Chemistry). (b) Computer simulation using the parameters in
Table 4.3.
Table 4.3 [E] ESR parameters for [Mn(dppe)
2
(CO)(CNBu)]
21
g
ii
10
4
A
ii
Mn
(cm
À1
) 10
4
A
ii
P
(cm
À1
)
2.107 30.2 27.2
2.051 20.6 25.3
1.998 146.9 26.4
64 Chapter 4
backdonation into the p* orbitals of the CO and CNBu ligands so that the
SOMO is likely based on d
x
2
Ày
2 :
jSOMOi ¼ ajz
2
þbjx
2
Ày
2
þÁ Á Á ð4:26Þ
Although the isotropic spectrum was not suﬃciently resolved to unambigu
ously determine A
Mn
, other closely related species give isotropic couplings on
the order of 60–70 G;
25
if we assume an isotropic coupling in this range, all
three matrix components must have the same sign. If the isotropic hyperﬁne
coupling is negative, as expected if it arises mostly through polarization of
innershell s orbitals, we have A
Mn
¼ À65.9 Â 10
À4
cm
À1
. Assuming that the
SOMO is mostly d
x
2
Ày
2 , (b 44 a) and including the appropriate spin–orbit
coupling corrections, we have:
21,26
A
z
ÀA
Mn
¼ P
4
7
a
2
Àb
2
À
2
3
Dg
z
À
5
42
Dg
x
þDg
y
¸
ð4:27Þ
With Dg
z
¼ À0.004, (Dg
x
þ Dg
y
) ¼ 0.154, and P ¼ 207.6 Â 10
À4
cm
À1
(ref. 14),
we get (a
2
Àb
2
) ¼ À0.655. The departure from axial symmetry is due to spin–
orbit coupling and/or d
x
2
Ày
2 =d
z
2 hybridization,
A
x
ÀA
y
¼ P À
8
ﬃﬃﬃ
3
p
7
ab þ
17
14
Dg
x
ÀDg
y
¸ ¸
ð4:28Þ
Substituting the parameters, we have ab ¼ Æ0.058. (The upper sign applies if
the components are listed in the order x, y, z in Table 4.3, the lower sign if the
order is y, x, z.) Finally, we get b
2
¼ 0.660, a
2
¼ 0.005. The d
z
2 component is not
really signiﬁcant, given the accuracy of the data and the theory, i.e., most of the
departure from axial symmetry can be explained by the spin–orbit coupling
correction.
Using eqn (4.11), the deviations of the gmatrix components from the free
electron value, g
e
, are found to be:
Dg
xx
¼ 2z
Mn
¸
k
b
2
c
2
yz;k
E
0
ÀE
k
ð4:29aÞ
Dg
yy
¼ 2z
Mn
¸
k
b
2
c
2
xz;k
E
0
ÀE
k
ð4:29bÞ
Dg
zz
¼ 2z
Mn
¸
k
4b
2
c
2
xy;k
E
0
ÀE
k
ð4:29cÞ
If we assume coupling with single pure d
yz
, d
xz
, and d
xy
orbitals, we have
DE
yz
¼ 16z, DE
xz
¼ 19z, DE
xy
¼ –1100z, which is qualitatively consistent with
the expected MO energy level scheme.
65 Anisotropic ESR Spectra
4.5.2 Some Cobalt(0) Radical Anions
ESR spectra of [CpCo(1,3COD)]
À
in frozen THF solution and [CpCo(1,5
COD)]
À
in frozen DMF have been reported by Geiger and coworkers
27
and
are reproduced in Figures 4.5(a) and 4.6(a). These spectra have been

]
2
O) Co(C )
5
Ph
5
C [(

D)] CO  o(1,5 [CpC

] ) D O C  1,3 o( C [Cp
_
_
CO OC
Co
Ph
Ph
Ph
Ph Ph
Co
Co
_
Figure 4.5 [E] ESR spectrum of [CpCo(1,3COD)]
À
: (a) experimental spectrum in
frozen THF solution.
(Reproduced with permission from ref. 27, copyright (1981) American
Chemical Society.) (b) Computer simulation, based on the parameters in
Table 4.4.
66 Chapter 4
reinterpreted to give the parameters shown in Table 4.4; computersimulated
spectra based on these parameters are shown in Figures 4.5(b) and 4.6(b). Also
shown in Table 4.4 are the ESR parameters for [(C
5
Ph
5
)Co(CO)
2
]
À
.
20
The
59
Co hyperﬁne matrix components must have identical signs in order
that the average values match the observed isotropic couplings; we assume the
signs are negative since the isotropic couplings almost certainly arise from
polarization of inner shell s orbitals (see below).
The SOMO in these radicals is expected from extended Hu¨ ckel MO calcu
lations
27–29
to be primarily cobalt 3d
yz
in character. In the C
s
symmetry of the
radicals, d
yz
belongs to the a
00
representation and dhybridization is possible
only with d
xy
. Assuming that such hybridization is negligible, the gmatrix
Figure 4.6 [E] (a) ESR spectrum of [CpCo(1,5COD)]
À
in frozen DMF solution.
(Reproduced with permission from ref. 27, copyright (1981) American
Chemical Society.) (b) Computer simulation, based on the parameters in
Table 4.4.
Table 4.4 [E] ESR parameters for cobalt(0) radical anions
Radical anion g
x
g
y
g
z
10
4
A
x
(cm
À1
) 10
4
A
y
(cm
À1
) 10
4
A
z
(cm
À1
)
[CpCo(1,5COD)]
Àa
2.171 2.027 1.985 158.6 36.7 45.8
[CpCo(1,3COD)]
Àa
2.151 2.027 1.997 139.2 36.4 38.2
[(C
5
Ph
5
)Co(CO)
2
]
Àb
2.018 2.041 1.995 157.9 16.8 44.1
a
From ref. 27.
b
From ref. 20.
67 Anisotropic ESR Spectra
components are given by:
20
Dg
xx
¼ 2z
Co
¸
k
a
2
c
x
2
Ày
2
;k
2
þ3a
2
c
z
2
;k
2
E
0
ÀE
k
ð4:30aÞ
Dg
yy
¼ 2z
Co
¸
k
a
2
c
xy;k
2
E
0
ÀE
k
ð4:30bÞ
Dg
zz
¼ 2z
Co
¸
k
a
2
c
xz;k
2
E
0
ÀE
k
ð4:30cÞ
The dipolar contribution to the hyperﬁne matrix is given by eqn (4.20), but
spin–orbit coupling contributions are signiﬁcant. These often can be expressed
in terms of the gmatrix components, as in the Mn(II) example discussed above,
but here spin–orbit coupling with the four other dorbitals contributes some
what diﬀerently to the gmatrix and to the hyperﬁne matrix. The simplest way
of expressing the hyperﬁne matrix is in terms of the isotropic coupling, the x
component, and the departure from axial symmetry. With the assumed signs of
the hyperﬁne components of Table 4.4, eqn (4.31b) can be used unambiguously
to compute a
2
¼ r
d
with the results shown in Table 4.5.
A ¼ A
s
þ
1
3
P Dg
xx
þDg
yy
þDg
zz
ð4:31aÞ
A
x
ÀA ¼ P À
4
7
a
2
þ
2
3
Dg
xx
À
5
42
Dg
yy
þDg
zz
¸
ð4:31bÞ
A
y
ÀA
z
¼
17P
14
Dg
yy
þDg
zz
þ
6a
2
zP
7
1
DE
x
2
Ày
2
À
1
DE
z
2
ð4:31cÞ
Since 3d
yz
/4s admixture is symmetryforbidden for these radicals, the Fermi
contact contribution to the isotropic coupling, A
s
, must be entirely from spin
polarization,
A
s
¼ Q
d
r
d
ð4:32Þ
Table 4.5 [E] Electron spin densities in cobalt(0) radical anions
Radical anion r
d
A
s
a
A
s
/Q
d
[CpCo(1,5COD)]
Àb
0.681 À97.0 0.740
[CpCo(1,3COD)]
Àb
0.591 À87.2 0.666
[(C
5
Ph
5
)Co(CO)
2
]
Àc
0.540 À77.4 0.591
a
In units of 10
À4
cm
À1
.
b
From ref. 27.
c
From ref. 20.
68 Chapter 4
Thus we can obtain an independent estimate of the delectron spin density from
the values of A
s
, taking Q
d
¼ À131 Â 10
À4
cm
À1
– estimated from the isotropic
Co coupling in [PhCCo
3
(CO)
9
]
À
(ref. 18). The results are shown in the last
column of Table 4.5. The spin densities estimated from the isotropic couplings
are consistently about 10% higher than those from the dipolar coupling matrix,
suggesting a systematic error in one of the parameters, but a reliable ordering of
the spin densities.
The gmatrix presents an interesting problem in these cases. EHMO calcu
lations
27,28,29
suggest that the SOMO is the highestenergy MO, which is
primarily cobalt 3d in character. At lower energy is an orbital with d
xz
character and still lower, but grouped at about the same energy, are MOs with
d
x
2
Ày
2 , d
xy
, and d
z
2 contributions. Equations (4.30) then would suggest that
Dg
xx
/4 EDg
yy
oDg
zz
. With the assignments of Table 4.4, the ﬁrst relationship
is approximately correct for [CpCo(1,3COD)]
À
and [CpCo(1,5COD)]
À
, but
very poor for [(C
5
Ph
5
)Co(CO)
2
]
À
. The second relationship is not found for any
of the anions. Reversing the y and z assignments makes the agreement worse. In
discussing this problem for [(C
5
Ph
5
)Co(CO)
2
]
À
,
20
we postulated admixture of
some cobalt 4p
y
character in the SOMO,
jSOMOi ¼ ajyzi þbjyi þÁ Á Á ð4:33Þ
which would result in additional contributions to g
xx
and g
zz
:
g
xx
ðpÞ ¼ 2z
p
b
2
=DE
z
and g
zz
ðpÞ ¼ 2z
p
b
2
=DE
x
ð4:34Þ
where z
p
is the cobalt 4p spin–orbit coupling parameter (z
p
Ez
d
/3). If MOs with
signiﬁcant p
z
or p
x
character lie just above the SOMO, then g
xx
and g
zz
would be
less positive than expected from eqns (4.30), and possibly even negative. The
component g
xx
is indeed smaller than expected for [(C
5
Ph
5
)Co(CO)
2
]
À
and
EHMO calculations do indeed suggest an MO with signiﬁcant p
z
character just
above the SOMO in energy; this orbital is apparently substantially higher in
energy in [CpCo(1,3COD)]
À
and [CpCo(1,5COD)]
À
. An MO with signiﬁcant
p
x
character, at about the same energy for all three anions, is implied by these
results but is unsubstantiated by the reported EHMO calculations.
4.6 Organic Examples of Solidstate ESR Spectra
Far less eﬀort has been directed at organic solidstate spectra than at organo
metallics and coordination complexes. Much of the work on organic systems
was done by Walter Gordy and his students at Duke University in the 1960s
and 1970s. We review a couple of these experiments here. More information is
available in Gordy’s book.
30
4.6.1 Irradiated Single Crystal of Glycylglycine
Katayama and Gordy
31
studied ESR spectra of girradiated single crystals of
glycylglycine. The data were analyzed as described above to yield the
69 Anisotropic ESR Spectra
parameters shown in Table 4.6. The axes were deﬁned as follows: z is along the
CH bond, y is in the NCHC plane > to CH, and x is > to the NCHC plane.
The conclusion was that the radical produced by girradiation was that
pictured in Table 4.6. This conclusion was based on the larger coupling to H
than to N, but nonetheless there is a signiﬁcant coupling to N.
4.6.2 Xirradiated Single Crystal of Methylene Diphosphonic Acid
Lucken and coworkers
32
subjected a single crystal of methylene diphosphonic
acid to Xirradiation. The ESR spectrum indicated that the radicals produced
were those pictured in Tables 4.7 and 4.8. The spectra were analyzed as
described above and the results are also summarized in the tables. The species
shown in Table 4.7 is the more abundant of the two. The methylene group freely
rotates at room temperature but is stationary at 77 K, where splitting from two
nonequivalent protons is observed for some orientations of the crystal.
Table 4.6 ESR parameters for the glycylglycine radical.
31
(H and N hyperﬁne splittings from nuclei in
structure shown in boldface.)
H
3
N
+
C
C
N
C
CO
2

H O
H
H H
Axis g a
H
(G) a
N
(G)
x 2.0028 19 4
y 2.0035 28 3
z 2.0033 9 2
Table 4.7 ESR parameters for the phosphonylmethyl
radical
32
P
O
CH
2
HO
HO
Axis g a
H
(G) a
P
(G)
x 2.0024 16.3 38.7
y 2.0029 27.2 44.3
z 2.0022 20.4 41.4
70 Chapter 4
4.7 Noncoincident Matrix Axes
In general, the g and nuclear hyperﬁne coupling matrices, g and A
i
, can be
written in diagonal form with three principal values, i.e., g
x
, g
y
, g
z
and A
ix
, A
iy
,
A
iz
. In textbooks on ESR
6a,30,33–35
it is usually assumed that the same set of
principal axes diagonalizes all the relevant matrices. While this is sometimes
true, there are many instances where the principal axes are noncoincident.
36
4.7.1 Symmetry Considerations
Kneubu¨ hl
37,38
has given a detailed group theoretical analysis of symmetry
restrictions on the orientations of g and hyperﬁne matrix principal axes. His
results are summarized in Table 4.9.
For a nucleus sharing all the molecular symmetry elements (e.g., the metal
nucleus in a mononuclear complex), the hyperﬁne matrix is subject to the same
Table 4.8 ESR parameters for the diphosphonyl
methyl radical
32
P
C
P
O
O
HO
HO
OH
OH
H
Axis g a
H
(G) a
P
(G)
x 2.0024 12.7 36.5
y 2.0029 30.0 38.1
z 2.0022 18.5 40.4
Table 4.9 [E] Symmetry restrictions on gmatrix components
Symmetry Triclinic Monoclinic Orthorhombic Axial
Point groups C
1
, C
i
C
2
, C
s
, C
2h
C
2v
, D
2
, D
2h
C
n
, C
nv
, C
nh
, D
n
,
D
nd
, D
nh
, n 4 2
Restrictions on
diagonal
elements
None None None g
xx
¼ g
yy
Restrictions on
oﬀdiagonal
elements
None g
xz
¼g
yz
¼0 g
xz
¼ g
yz
¼ g
xy
¼ 0 g
xz
¼ g
yz
¼ g
xy
¼ 0
Required
matrix axes
None z x,y,z x,y,z
71 Anisotropic ESR Spectra
restrictions as the gmatrix. In orthorhombic or axial symmetry, such nuclear
hyperﬁne matrices necessarily share principal axes with the gmatrix. In
monoclinic symmetry, one hyperﬁne axis is also a gmatrix axis, but the other
two may be diﬀerent. In triclinic symmetry (C
1
or C
i
), none of the three
principal axes need be shared by the gmatrix and hyperﬁne matrix. The
hyperﬁne matrix for a ligand atom (or for a metal in polynuclear complexes)
is constrained only by the symmetry elements that the nucleus shares with the
molecule.
Although symmetry considerations often permit g and hyperﬁne matrix
principal axes to be noncoincident, there are relatively few cases of such non
coincidence reported in the literature. Most of the examples discussed by
Pilbrow and Lowrey in their 1980 review
36
cite cases of transition metal ions
doped into a host lattice at sites of low symmetry. This is not to say that matrix
axis noncoincidence is rare but that the eﬀects have only rarely been recognized.
4.7.2 Experimental Determination of Matrix Axis Orientations
We have seen that spectra of dilute single crystals are analyzed in a way that
gives the orientations of the g and hyperﬁnematrix principal axes relative to
the crystal axes. Historically, most of the information on noncoincident matrix
axes is derived from such studies.
At ﬁrst glance, it would appear that all orientation dependence should be lost
in the spectrum of a randomly oriented sample and that location of the g and
hyperﬁnematrix principal axes would be impossible. While it is true that there
is no way of obtaining matrix axes relative to molecular axes from a powder
pattern, it is frequently possible to ﬁnd the orientation of a set of matrix axes
relative to those of another matrix.
The observable eﬀects of matrix axis noncoincidence on powder patterns
range from blatantly obvious to negligible. In general, the eﬀects of axis non
coincidence will be more noticeable if two (or more) matrices have large
anisotropies that are comparable in magnitude, e.g., Dgm
B
B EDA. This follows
from the fact that minimum and maximum resonant ﬁelds are determined by a
competition between extrema in the angledependent values of g and A.
Consider the case of noncoincident g and hyperﬁnematrix axes. For large
values of m
I
, the ﬁeld extrema will be determined largely by the extrema in the
eﬀective hyperﬁne coupling and will occur at angles close to the hyperﬁne
matrix axes, but for small m
I
 the extrema will be determined by extrema in the
eﬀective gvalue and will correspond to angles close to the gmatrix axes. The
result of such a competition is that a series of features that would be equally
spaced (to ﬁrstorder) acquires markedly uneven spacings.
Two corollaries stem from this generalization. Since a spin1/2 nucleus gives
only two hyperﬁne lines, there can be no variation in spacings. Thus powder
spectra cannot be analyzed to extract the orientations of hyperﬁne matrix axes
for such important nuclei as
1
H,
13
C,
19
F,
31
P,
57
Fe, and
103
Rh. Secondly, since
the observable eﬀects in powder spectra depend on the magnitude of the matrix
72 Chapter 4
anisotropies, the principal axes of the hyperﬁne matrix for a nucleus with small
hyperﬁne coupling generally cannot be located from a powder spectrum, even
though the relative anisotropy may be large.
4.8 Organometallic Examples of Noncoincident Matrix
Axes
4.8.1 A Chromium Nitrosyl Complex
A good example of the eﬀect of g and hyperﬁne matrix axis noncoincidence is
the ESR spectrum of [CpCr(CO)
2
NO]
À
, studied by Geiger and coworkers;
39
a
simulation is shown in Figure 4.7.
Cr
N
O
C
C
O
O

3340 3360 3380 3400 3420 3440
Magnetic Field/Gauss
(a)
(b)
Figure 4.7 (a) Computer simulation of an absorption spectrum, zero linewidth,
showing positions m
A
¼ À1, 0, þ1. (b) Firstderivative spectrum in frozen
DMF.
(Reproduced with permission from ref. 39, copyright (1984) American
Chemical Society.)
73 Anisotropic ESR Spectra
The g and
14
N hyperﬁne matrices are approximately axial for this radical,
but the g

axis lies close to the perpendicular plane of the hyperﬁne matrix. If
the g

axis was exactly in the A
>
plane, the three negativegoing g

, A
>
features, corresponding to resonant ﬁeld maxima, would be evenly spaced. In
fact, the spacings are very uneven – far more so than can be explained by
secondorder shifts. The eﬀect can be understood, and the spectrum simulated
virtually exactly, if the g

axis is about 151 out of the A
>
plane.
4.8.2 Iron Pentacarbonyl Ions
Two particularly interesting organometallic examples have been reported by
Morton, Preston and coworkers.
40,41
Spectra of single crystals of Cr(CO)
6
,
doped with
13
C or
57
Feenriched Fe(CO)
5
and girradiated at 77 K, showed
two diﬀerent radicals. One species, identiﬁed as Fe(CO)
5
1
, showed coupling to
57
Fe and to a unique
13
C nucleus with axial hyperﬁne matrices sharing principal
axes with the gmatrix.
40
Coupling was also observed to four other
13
C nuclei
with identical coupling matrices but with the major axis approximately along
the gmatrix xaxis for two nuclei and along the gmatrix yaxis for the other
two. Table 4.10 lists the parameters.
If the radical is square pyramidal (C
4v
) Fe(CO)
5
1
(1), the principal axes of
the gmatrix must be the molecular axes (the C
4
axis and normals to the
reﬂection planes). The iron atom and the carbon of the axial CO group would
have the full symmetry of the group and so these hyperﬁne matrices would
share principal axes with the gmatrix. The four equatorial carbonyl carbons,
on the other hand, lie in reﬂection planes, but not on the C
4
axis and so are
symmetryrequired to share only one principal axis with the gmatrix. In fact,
the major matrix axes for the equatorial carbons are tilted slightly in the –z
direction from the ideal locations along the Æx and Æy axes. The gmatrix
suggests that the metal contribution is d
z
2 and the iron hyperﬁne matrix then
can be used to estimate about 55% iron 3d and 34% axial carbon 2p
z
spin
density. The spin density on the equatorial carbons then is mostly negative and
due to spin polarization.
The other species observed in irradiated Fe(CO)
5
doped crystals of Cr(CO)
6
also showed coupling to
57
Fe, to a unique
13
C, and to four other carbons.
However, in this case g, A
Fe
, and A
C1
have only one matrix axis in common
(that corresponding to the third component of each matrix listed in Table 4.10).
Table 4.10 [E] ESR parameters for Fe(CO)
5
1
and Fe(CO)
5
Àa
Fe(CO)
þ
5
g

¼ 2.001 A

Fe
¼ 9.4 A

C1
¼ 19.6 A

C2C5
¼(þ)6.4
g
>
¼ 2.081 A
>
Fe
¼ 6.2 A
>
C1
¼ 17.6 A
>
C2C5
¼(þ)8.6
Fe(CO)
À
5
g
1
¼ 1.989 A
1
Fe
¼ 6.7 A
1
C1
¼ 87.4
g
2
¼ 2.003 A
2
Fe
¼ 4.5 A
2
C1
¼ 70.7
g
3
¼ 2.010 A
3
Fe
¼ 3.2 A
3
C1
¼ 65.7
a
Coupling constants in units of 10
À4
cm
À1
. Data from refs. 40 and 41.
74 Chapter 4
The other
57
Fe hyperﬁne axes are rotated by about 271 and those of the
13
C
hyperﬁne matrix by about 481 relative to the gmatrix axes. Insuﬃcient data
were accumulated to determine the complete hyperﬁne matrices for the other
four carbons, but the components are considerably smaller (4–15 Â10
À4
cm
À1
).
The hyperﬁne matrices suggest about 38% iron 3d
z
2 , 18% carbon 2p, and 6%
carbon 2s spin densities. Using detailed arguments regarding the orientation of
the gmatrix axes relative to the crystal axes, the authors conclude that the
carbon 2p axis is oriented at about 1061 relative to the Fe–C bond axis and that
the Fe–C–O bond angle is about 1191.
Fe
O
C
C
O
O
C
C
O
O
Fe
O
C
C
O
C
O
C
CO
O
C
+
x
106°
27°
y
z(gmatrix)
z
1 2
−
The most striking feature of these results is the orientation of the unique
13
C
hyperﬁne matrix axes, relative to those of the
57
Fe hyperﬁne axes. This
orientation led Fairhurst et al.
41
to assign the spectrum to [Fe(CO)
5
]
À
(2)
and to describe the species as a substituted acyl radical. However, these authors
did not discuss the orientation of the gmatrix axes. The yaxis, normal to the
reﬂection plane, is common to all three matrices. The x and zaxes of the
gmatrix, however, are oriented about 271 away from the corresponding
57
Fe
hyperﬁne matrix axes. Since the iron dorbital contribution to the SOMO
appears to be nearly pure d
z
2 , the
57
Fe hyperﬁne matrix major axis must
correspond to the local zaxis, assumed to be essentially the Fe–C bond. Thus
we must ask: Why are the gmatrix axes diﬀerent? The SOMO can be written:
jSOMOi ¼ ajz
2
; Fe
þb
x
jx; Ci þb
z
jz; Ci ð4:35Þ
where a ¼ 0.62, b
x
¼ À0.41, and b
z
¼ 0.12. Spin–orbit coupling will mix the
SOMO with MOs having iron d
yz
or d
xz
character, but d
yz
is involved in the p
orbitals of the C ¼ O group:
jpi ¼ c
yz
jyz; Fe
þc
y
jy; C
ð4:36Þ
Assuming that there is only one p orbital close enough in energy to couple
signiﬁcantly, eqn (4.11) gives the gmatrix components:
Dg
xx
¼ 2z
Fe
3a
2
c
2
yz
þ
ﬃﬃﬃ
3
p
ab
z
c
yz
c
y
DE
p
ð4:37aÞ
Dg
yy
¼ 2z
Fe
3a
2
c
2
xz
DE
xz
ð4:37bÞ
75 Anisotropic ESR Spectra
Dg
zz
¼ 0 ð4:37cÞ
g
xz
¼ À2z
Fe
ﬃﬃﬃ
3
p
ab
x
c
yz
c
y
DE
p
ð4:37dÞ
The gmatrix can be diagonalized by rotation about the yaxis by the angle b:
tan 2b ¼
2g
xz
g
xx
Àg
zz
¼
À2
ﬃﬃﬃ
3
p
ab
x
3a
2
c
yz
=c
y
þ
ﬃﬃﬃ
3
p
ab
z
ð4:38Þ
With b ¼ 271, this expression gives c
yz
/c
y
E 0.5, a reasonable result.
This may be a rather general eﬀect; if the unpaired electron in a radical is
delocalized asymmetrically, and other MOs are similarly delocalized, the g
matrix will have oﬀdiagonal elements that may be large enough to shift the
principal axes away from the molecular coordinate system.
4.8.3 Another Lowspin Manganese(II) Complex
The lowspin manganese(II) complex [Mn(dppe)
2
(CO)(CNBu)]
21
gave us a
textbook example of a wellbehaved ESR spectrum characterized by coincident
g and hyperﬁnematrix principal axes. The nearly identical complex
[Mn(dppm)
2
(CO)(CN)]
1
, (dppm ¼ Ph
2
PCH
2
PPh
2
) (ref. 25) provides us with
a good example of noncoincident principal axes. The frozen solution spectrum
(Figure 4.8) shows that the ‘‘parallel’’ features are not evenly spaced.
3000 3200 3400 3600 3800
Magnetic Field/Gauss
Figure 4.8 ESR spectrum of the Mn(II) complex [Mn(dppm)
2
(CO)(CN)]
1
in CH
2
Cl
2
/
C
2
H
4
Cl
2
at 90 K.
(Reproduced with permission from ref. 25, copyright (1993) Royal Society
of Chemistry.)
76 Chapter 4
The spectrum can be understood if the zaxes of the g and Amatrices are
displaced by b ¼ 19.61. This, of course, tells us that the molecule does not have
C
2v
symmetry, and that, unlike the dppe analog, d
x
2
Ày
2 is not the only Mn
contribution to the SOMO. One way of interpreting the results is that the Mn
contribution to the SOMO incorporates a small amount of d
xz
character. The
consequence of this hybridization would be to tilt the ‘‘x’’ lobes of d
x
2
Ày
2 up
and down, i.e., rotation about the yaxis. The reason for this hybridization is
not diﬃcult to discover. If the CH
2
groups of the dppm ligands were coplanar
with Mn, the ‘‘x’’ lobes of the d
x
2
Ày
2 SOMO would be directed toward the
carbon atoms and an antibonding interaction would result. To avoid this
interaction, two things happen: (1) incorporation of d
xz
character tilts the lobes
up and down, away from the C atoms. At the same time, the Xray structure of
the Mn(II) cation (Figure 4.9) shows that the CH
2
groups tilt down and up,
further decreasing the antibonding interaction. With only one electron in the
SOMO, the antibonding interaction is strong enough to produce this distor
tion. With two electrons, the neutral parent Mn(I) complex would be expected
to be even more distorted, and an Xray structure shows that the up and down
displacements of the CH
2
groups is approximately doubled.
4.8.4 Chromium(I) Pianostool Complex
Ordinarily, there is no way of extracting the orientation of the principal axes of
the gmatrix from a powder or frozen solution ESR spectrum. However, there
are exceptional circumstances in which nature is kind to the experimentalist!
Figure 4.9 Xray structure showing methylene groups tilted up and down to avoid
antibonding interaction with the Mn d
x
2
Ày
2 SOMO in the complex
[Mn(dppm)
2
(CO)(CN)]
1
.
(Reproduced with permission from ref. 25, copyright (1993) Royal Society
of Chemistry.)
77 Anisotropic ESR Spectra
Spectra of the lowspin d
5
Cr(I) complex [(C
5
Ph
5
)Cr(CO)
2
PMe
3
]
28
are shown
at 125 and 200 K in Figure 4.10. The lowtemperature spectrum shows three
sets of doublets, corresponding to the three gcomponents (2.104, 2.013, 1.994),
each a doublet due to hyperﬁne coupling to
31
P. As the sample is warmed to
200 K, above the freezing point of toluene, the spectrum still appears as a
‘‘powder pattern’’, but the lowﬁeld ‘‘perpendicular’’ features are nearly
averaged and the ‘‘parallel’’ features have shifted slightly upﬁeld (g

¼ 2.090,
g
>
¼ 2.012).
The exceptionally bulky C
5
Ph
5
ligand apparently moves very slowly at 200 K
but on the ESR time scale the Cr(CO)
2
PMe
3
moiety rotates quickly, producing
a spectrum averaged about the Cr–C
5
axis. With this interpretation, we can
assign this axis as the principal axis for g

measured at 200 K and, noting that
the g

axis diﬀers from the g
z
axis by the angle b and that g
y
and g
x
are averaged
according to:
2g
2
jj
¼ g
2
z
þg
2
x
þðg
2
z
þg
2
x
Þ cos
2
2b
Substitution of the gcomponents gives b ¼ 151.
The symmetry of the static complex is at most C
s
, requiring one of the
gmatrix principal axes to be normal to the plane of symmetry – this is assumed
3200 3250 3300 3350 3400 3450
Magnetic Field/Gauss
125 K
200 K
×4
Figure 4.10 Spectra of [(ZC
5
Ph
5
)Cr(CO)
2
PMe
3
] in toluene at 125 and 200 K.
(Reproduced with permission from ref. 28, copyright (1996) American
Chemical Society.)
78 Chapter 4
to be xz. pBackbonding to the CO ligands is expected to lead to a hybrid
SOMO. If, as suggested by extended Hu¨ ckel MO calculations, the SOMO and
ﬁrst HOMO are of a
0
symmetry, and the second HOMO of a
00
symmetry:
jSOMOi ¼ a
1
jx
2
Ày
2
þa
2
jz
2
þa
3
jxzi
jHOMO
1
i ¼ b
1
jx
2
Ày
2
þb
2
jz
2
þb
3
jxzi
jHOMO
2
i ¼ c
1
jxyi þc
2
jyzi
the gmatrix components can be computed:
42
g
xx
¼ g
e
þ
z
Cr
a
1
c
2
þ
ﬃﬃﬃ
3
p
a
2
c
2
þa
3
c
1
2
E
0
ÀE
2
g
zz
¼ g
e
þ
z
Cr
2a
1
c
1
þa
3
c
2
ð Þ
2
E
0
ÀE
2
g
yy
¼ g
e
þ
z
Cr
a
1
b
3
Àa
3
b
1
þ
ﬃﬃﬃ
3
p
a
3
b
2
þa
2
b
3
ð Þ
2
E
0
ÀE
1
g
xz
¼ À
z
Cr
2a
1
c
1
þa
3
c
2
ð Þ a
1
c
2
þ
ﬃﬃﬃ
3
p
a
2
c
2
þa
3
c
1
E
0
ÀE
2
Rotation about the yaxis by b diagonalizes the matrix, and we ﬁnd:
tan 2b ¼ À
2g
xz
g
zz
Àg
xx
The single experimental observable, b, is hardly enough to evaluate the LCAO
coeﬃcients for the SOMO and second HOMO, but we can compare the results
of an extended Hu¨ ckel MO calculation. (Since z
Cr
and E
0
ÀE
2
cancel in the
calculation of tan 2b, the EHMO calculation could come close.) The results are:
a
1
¼ 0.538, a
2
¼ 0.216, a
3
¼ À0.194, c
1
¼ 0.582, c
2
¼ À0.061, b ¼ 14.81.
Nevertheless, if we were to substitute these LCAO coeﬃcients, but also use the
EHMO estimate of the energy diﬀerence, into the expressions for g
xx
, g
yy
, g
zz
,
and g
xz
, the results would be in very poor agreement with experiment; the
moral here is that EHMO calculations, lacking charge selfconsistency, usually
have large errors in the energies, but the MO wavefunctions are often fairly
accurate.
4.8.5 [(RCCR
0
)Co
2
(CO)
6
]
À
and [SFeCo
2
(CO)
9
]
À
Noncoincident matrix axis eﬀects are seen in the frozen solution spectra of
[(RCCR
0
)Co
2
(CO)
6
]
À
(3)
18
and [SFeCo
2
(CO)
9
]
À
(4),
43
but the eﬀects are rather
more subtle than those discussed above.
79 Anisotropic ESR Spectra
3 4
C C
Ph
C
O
C
O
C
Co Co
Fe S
CO
O O
OC
C
OC
C
Co Co
C C
CO
O
Ph
O
O
O
C
C
O
OC
Co Co
C
y
β
−
z(g)
z(A)
O
−
In these cases, the gmatrix is nearly isotropic, but the principal axes of the
two
59
Co hyperﬁne matrices are noncoincident. The largest hyperﬁne matrix
component (a
y
¼ 66.0 G in the case of the CoCoFeS cluster) results in 15
features, evenly spaced (apart from small secondorder shifts). Another series
of features, less widely spaced, shows some variation in spacing and, in a few
cases, resolution into components. This behavior can be understood as follows:
Suppose that the hyperﬁne matrix yaxes are coincident and consider molecular
orientations with the magnetic ﬁeld in the xzplane. To ﬁrst order, the resonant
ﬁeld then is:
B ¼ B
o
Àm
1
a
þ
Àm
2
a
À
ð4:39Þ
where B
o
¼ hn/gm
B
and:
a
2
Æ
¼ a
2
z
cos
2
ðy ÆbÞ þa
2
x
sin
2
ðy ÆbÞ ð4:40Þ
where b describes the orientation of the hyperﬁne matrix zaxes relative to the
gmatrix zaxis. Since g is nearly isotropic, the extrema in B are determined
mostly by the hyperﬁne term. When m
1
¼ m
2
, a
1
and a
À
are equally weighted
and the extrema occur at y ¼ 0 or 901, but when m
1
a m
2
, the extrema
correspond to other angles. Consider, for example, the ﬁve components of the
m ¼ m
1
þ m
2
¼ þ3 feature. With a
z
¼ 53.6 G, a
x
¼ 15.5 G, b ¼ 181, the
hyperﬁne contributions to the ﬁeld extrema and the corresponding values of y
are given in Table 4.11. In the experimental spectrum of the CoCoFeS
cluster, two resolved ﬁeld maximum features were seen, corresponding to the
ﬁrst two and the last three of the above components. Since the resolution is
sensitive to the noncoincidence, it was possible to ﬁt the experimental spectrum
to obtain b quite accurately.
Table 4.11 [E] Splitting of m ¼ 3 features in the [SFeCo
2
(CO)
9
]
À
spectrum
43
(m
1
, m
2
) y
min
(1) (BÀB
0
)
min
(G) y
max
(1) (BÀB
0
)
max
(G)
þ7/2,À1/2 À20 37.0 þ67 166.1
À1/2,þ7/2 þ20 37.0 À67 166.1
þ5/2,þ1/2 À15 55.3 þ78 156.9
þ1/2,þ5/2 þ15 55.3 À78 156.9
þ3/2,þ3/2 0 66.5 90 153.6
80 Chapter 4
It is relatively easy to understand the signiﬁcance of the noncoincident
matrix axes in these cases. For the Co
2
C
2
cluster, the C
2v
molecular symmetry
permits a speciﬁc prediction of the possible matrix axis orientations. The g
matrix principal axes must be coincident with the molecular symmetry axes.
The two cobalt nuclei are located in a reﬂection plane (which we label xz) so
that symmetry requires the yaxis to be a principal axis for all three matrices.
The other two axes may be rotated, relative to the molecular x and zaxes,
by Æb. (Since the two nuclei are symmetrically equivalent, the rotations must be
equal and opposite.)
Since the magnitudes and probable signs of the cobalt hyperﬁne matrices
suggest a SOMO predominantly d
z
2 in character, the major axes of the
hyperﬁne matrices approximate the local zaxes at the cobalt atoms and the
angular displacement indicates a bent Co–Co antibonding interaction, as
shown in the structure, where the C–C or Fe–S bond axis (the molecular y
axis) is perpendicular to the plane of the page. Comparison with the crystal
structure of neutral (Ph
2
C
2
)Co
2
(CO)
6
44
shows that these local axes are roughly
in the direction of the axial carbonyl ligands (the Co–CO bond is tilted 281
away from the Co–Co vector). Thus it seems reasonable to say that the local
axes on a metal are determined primarily by the more strongly interacting
ligands and that bonds to other atoms can be described as bent.
4.8.6 (oXylylene)Mn(dmpe)
2
As part of a study of Mn(II) dialkyls, Wilkinson, Hursthouse, and coworkers
45
reported the ESR spectrum of the approximately octahedral (oxylyl
ene)Mn(dmpe)
2
(Figure 4.11a) (dmpe ¼ Me
2
PCH
2
CH
2
PMe
2
).
The spectrum was interpreted assuming coincident g and hyperﬁnematrix
axes, but a simulation based on the reported parameters gave a very poor ﬁt to
the published spectrum. On closer examination, it was realized that this is a
rather extreme example of a spectrum inﬂuenced by noncoincident g and
hyperﬁne matrix principal axes. The clue evident in the spectrum is the large
gap between the m
Mn
¼ À1/2 and þ1/2 ‘‘parallel’’ features, suggesting one or
more extra features. Figure 4.12 shows a set of simulated spectra for a
hypothetical lowspin Mn(II) species; all seven spectra correspond to the same
g and hyperﬁnematrices, but the angle b, between the g and hyperﬁne matrix
zaxes, varies from 0 to 901. As shown in Figure 4.12, it is possible to obtain
spectra with more than six resolved ‘‘parallel’’ features. Indeed, the spectrum is
suﬃciently sensitive to the angle that b can be evaluated quite precisely. The
ﬁnal parameters, based on leastsquares ﬁtting of the positions of the resolved
features and the isotropic parameters, are given in Table 4.12; a computer
simulation using these parameters is shown in Figure 4.11(b).
[(oXylylene)Mn(dmpe)
2
] has approximate C
2v
symmetry, but the ethylene
bridges of the dmpe ligands destroy the reﬂection planes; the Xray structure
45
shows a small fold at the methylene groups of the oxylylene ligand, which
destroys the C
2
axis. Thus the molecule can be regarded, with increasing
81 Anisotropic ESR Spectra
accuracy, as C
2v
, Cs, C
2
, or C
1
. To explain the matrix axis noncoincidence, the
metal contribution to the SOMO must be a dhybrid. Since it is a lowspin d
5
Mn(II) species, the SOMO is expected to be dominated by one of the orbitals of
the octahedral t
2g
set – d
xz
, d
yz
, or d
x
2
Ày
2 . This is consistent with a clue
contained in the ESR parameters. The four
31
P couplings are apparently
equivalent, and all are relatively small. Thus no lobe of the SOMO can be
directed toward a phosphorus atom, and major d
xy
or d
z
2 contributions to the
SOMO can be ruled out. Consider the twelve binary hybrids based on these
orbitals, which are listed in Table 4.13.
Since we know that the hyperﬁne matrix major axis is not a gmatrix
principal axis, we can immediately reject hybrids 1, 3, 4, 7, 8, and 10 for which
the major axis is a common axis. Hybrids 2 and 6, 5 and 9, and 11 and 12 diﬀer
Mn
C
C
P
P
P
P
x
z
y
Figure 4.11 [E] (a) ESR spectrum of (oxylylene)Mn(dmpe)
2
in frozen toluene
solution.
(Reproduced with permission from ref. 45, copyright (1983) Royal
Society of Chemistry.) (b) Computer simulation of spectrum using
parameters of Table 4.12.
82 Chapter 4
only in the x and ylabels and are essentially indistinguishable, so that there are
only three cases to consider in detail.
Hybrid 6 can be written:
jSOMOi ¼ ajx
2
Ày
2
þbjxzi ð4:41Þ
Figure 4.12 [E] Computersimulated ESR spectra for a hypothetical lowspin Mn(II)
radical with g ¼ (2.100, 2.050, 2.000), A
Mn
¼ (150, 25, 25) Â 10
À4
cm
À1
,
for various values of b, the Euler angle between the gmatrix and hyper
ﬁne matrix zaxes.
Table 4.12 [E] ESR parameters for (oxylylene)Mn(dmpe)
2
(b ¼ 411)
g 10
4
A
Mn
(cm
À1
) 10
4
A
P
(cm
À1
)
2.110 27 24.5
2.035 27 24.5
2.000 125 24.5
83 Anisotropic ESR Spectra
Straightforward application of eqns (4.24) yields a hyperﬁne matrix that can be
diagonalized by rotation about the yaxis by the angle b:
tan 2b ¼ À2b=a ð4:42Þ
The gmatrix can be written in relatively simple form if we assume that the only
MOs close enough in energy to contribute signiﬁcantly are the ﬁlled MOs,
primarily d
xz
and d
yz
in character. With this assumption, we obtain a matrix
which can be diagonalized by rotation about the yaxis by b
0
,
tan 2b
0
¼ À
2b=a
1 À b=a ð Þ
2
ð4:43Þ
For reasonable values of the hybridization ratio, b/a up to about 0.5, b and b
0
can diﬀer by only a few degrees and this kind of hybrid cannot explain the
matrix axis noncoincidence. Following the same reasoning with x and y
interchanged, exactly the same conclusions can be reached for hybrid 2.
Hybrids 11 and 12 are:
jSOMOi ¼ ajxzi þbjyzi ð4:44Þ
In this case, the hyperﬁne matrix remains axial, independent of the hybridiza
tion ratio, although the principal axes are rotated in the xyplane by an angle a
equal to angle b of eqn (4.43). Assuming that only MOs with predominantly
d
xz
, d
yz
, or d
x
2
Ày
2 character contribute, a gmatrix is found that can be
diagonalized by rotation in the xyplane by the angle a
0
, also given by eqn
(4.43). Thus this hybrid gives identical g and hyperﬁne matrix principal axes
for all hybridization ratios.
Finally, hybrid 9:
jSOMOi ¼ ajyzi þbjxyi ð4:45Þ
Table 4.13 [E] SOMO candidates: binary dhybrids
#
Representation Major Minor Approx. major Common
C
s
C
2
dAO dAO HF axis axis
1 a
0
a d
x
2
Ày
2 d
z
2 z x,y,z
2 a
0
– d
x
2
Ày
2 d
yz
z x
3 a
0
– d
yz
d
x
2
Ày
2 x x
4 a
0
– d
yz
d
z
2 x x
5 a
00
– d
xz
d
xy
y x
6 – a d
x
2
Ày
2 d
xz
z y
7 – a d
xz
d
x
2
Ày
2 y y
8 – a d
xz
d
z
2 y y
9 – b d
yz
d
xy
x y
10 – – d
x
2
Ày
2 d
xy
z z
11 – – d
xz
d
yz
y z
12 – – d
yz
d
xz
x z
84 Chapter 4
gives an axial hyperﬁne matrix with principal axes rotated in the xzplane by an
angle b, given by eqn (4.43). The gmatrix is somewhat more complicated:
Dg
2z
¼
ac
x
2
Ày
2
2
DE
x
2
Ày
2
þ
bc
xz
ð Þ
2
DE
xz
0 À
2ab c
x
2
Ày
2
2
DE
x
2
Ày
2
À
ab c
xz
ð Þ
2
DE
xz
0
bc
yz ð Þ
2
DE
yz
0
À
2ab c
x
2
Ày
2
2
DE
x
2
Ày
2
À
ab c
xz
ð Þ
2
DE
xz
0
2bc
x
2
Ày
2
2
DE
x
2
Ày
2
þ
ac
xz
ð Þ
2
DE
xz
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
ð4:46Þ
Diagonalization requires rotation about the yaxis by the angle b
0
,
tan 2b
0
¼ À
2b
a
2Qþ1
QÀ1 Àðb=aÞ
2
ð4QÀ1Þ
¸ ¸
ð4:47Þ
where
Q ¼
c
x
2
Ày
2
2
DE
x
2
Ày
2
c
xz
ð Þ
2
DE
xz
ð4:48Þ
Since the energy diﬀerences, DE
x
2
Ày
2 and DE
xz
are expected to be comparable,
the parameter Q is probably not far from unity. For Q ¼ 1, eqn (4.47) has a
particularly simple form, tan 2b
0
¼ þ2a/b so that, for small b/a, we expect b E
0 and b
0
E451, which is entirely consistent with experiment. The axial hyperﬁne
matrix is in agreement with experiment, and the principal values of the gmatrix
can also be rationalized with reasonable values of Q and b/a. A small rotation
of d
yz
about the yaxis might reﬂect the small displacements of the phosphorus
atoms from the idealized octahedral positions.
An extended Hu¨ ckel MO calculation supports the assumptions made in the
above analysis in that the three ‘‘t
2g
’’ orbitals are indeed close together in
energy and remain nearly nonbonding metalbased dorbitals. The detailed
agreement is less satisfactory: the SOMO is predicted to be primarily d
x
2
Ày
2
with a small d
xz
admixture (hybrid 6 of Table 4.13), a result that can be ruled
out from our analysis of the ESR results. The EHMO overlap matrix based on
the Xray structure suggests that the molecule is much closer to C
2
symmetry
than to Cs. If we accept that conclusion, then d
xz
/d
xy
hybridization is less likely
than d
yz
/d
xy
, as we tacitly assumed above.
Several alternative explanations were considered for the matrix axis non
coincidence. In particular, it seemed possible that delocalization of spin density
into the oxylylene ligand, either through the sbonds or into the psystem,
might lead to signiﬁcant contributions to the oﬀdiagonal terms of the gmatrix.
While the EHMO calculations suggest that the MOs containing d
x
2
Ày
2 and d
yz
do have contributions from the carbon atoms of the oxylylene group, the
amount is far too small to rationalize rotation of the gmatrix axes by 411;
indeed, to explain the eﬀect in this way would require each of several carbon
85 Anisotropic ESR Spectra
atoms to contribute 5% or more to the MOs with d
x
2
Ày
2 or d
yz
character, which
is unreasonably large considering the poor overlap of these metal dorbitals
with the relevant carbon orbitals.
4.8.7 Cobalt Dithiolene Complexes
In the 1960s and 1970s, dithiolene complexes were very popular subjects for
investigation by inorganic chemists. An ESR spectrum was reported
46
for one
such complex, the structure of which is shown in Figure 4.13, coordinated with
an axial PPh
3
ligand. A simulation based on a redetermination of the ESR
parameters
47
is also shown in Figure 4.13.
Careful examination of the spacings of the
59
Co hyperﬁne lines in the
spectrum reveals that the g and Aaxes are noncoincident. This tells us that
the symmetry cannot be as high as C
2v
. This is surprising since several crystal
structures obtained for iron dithiolene complexes are textbook examples of
square pyramidal complexes of C
2v
symmetry. Accordingly, Carpenter, et al.
47
determined the structure of the above complex and obtained ESR spectra for
various complexes with diﬀerent steric requirements. The noncoincidence
angle a is given in Table 4.14.
Co
S S
S S
CF
3
CF
3
F
3
C
F
3
C
P(Ph)
3
3100 3200 3300 3400 3500 3600
Magnetic Field/Gauss
Figure 4.13 Simulation of ESR spectrum of the above cobalt dithiolene complex in
frozen toluene at 77 K.
(Reproduced with permission from ref. 47, copyright (1994) Royal Soci
ety of Chemistry.)
86 Chapter 4
Clearly, from inspection of Table 4.14, there is a good correlation between
the steric bulk of R and L and the noncoincidence angle a. Furthermore,
analysis of the hyperﬁne parameters leads to the conclusion that only about
25% of the electron spin resides in Co orbitals (mainly d
xz
), and crystal
structures of the R ¼ CF
3
, L ¼ PPh
3
and P(OPh)
3
complexes do indeed show
distortions. The diﬀerence between iron and cobalt is just one electron, but this
electron occupies a dithiolene p* orbital, which makes the cobalt complexes
much more easily distorted.
4.9 ‘‘gStrain’’
ESR spectral lines in a frozen solution or powder spectrum are sometimes
peculiarly broad. The phenomenon is sometimes called ‘‘gstrain’’. In most
cases, the matter is dropped at that point and no attempt to explain further is
made. A successful interpretation of the eﬀect has, however, been made for
some organometallic radicals.
48
Figure 4.14 shows the ESR spectrum of [Mn(CO)(dmpe)Cp]
1
. Clearly, the
‘‘parallel’’ features are rather variable in line width, with the lowﬁeld lines
broad and the highﬁeld lines narrow. The ESR parameters for this spectrum
49
are given in Table 4.15 where b is the angle between the gmatrix zaxis and the
hyperﬁne matrix zaxis.
Because of the noncoincidence of the g and Amatrix principal axes, the
various ‘‘parallel’’ features correspond to diﬀerent orientations of the mag
netic ﬁeld in the gmatrix principal axis system. These orientations are given in
Table 4.16.
If the excess width is associated with orientation along the gmatrix zaxis, we
expect all widths to increase in the order m
I
¼ À1/2 oÀ3/2 oÀ5/2 ooþ5/2 o
þ3/2 o þ1/2, and this is observed. The ﬁeld position of a spectral feature can
be written as:
B ¼
hn Àkm
I
g
eff
m
B
ð4:49Þ
Table 4.14 Noncoincidence angles for [Co{S
2
C
2
R
2
}
2
L]. (Data reproduced
with permission from ref. 47, copyright (1994) Royal Society of
Chemistry.)
R L a (1)
CN Pet
3
2 Æ 2
CF
3
P(OPh)
3
11 Æ 5
CF
3
PPh
3
16 Æ 1
Ph P(OPh)
3
11 Æ 5
Ph PPh
3
24 Æ 1
4MePh Pet
3
10 Æ 2
4MePh PPh
3
24 Æ 1
4MeOPh PPh
3
31 Æ 2
87 Anisotropic ESR Spectra
where k is the angledependent hyperﬁne coupling and g
eﬀ
is given by eqn
(4.50):
g
eff
¼
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
g
2
z
cos
2
y þ g
2
x
cos
2
j þg
2
y
sin
2
j
sin
2
y
ð4:50Þ
Diﬀerentiating B with respect to g
z
, we see that the line width of a feature depends
not only on the orientation in the ﬁeld but also on the magnitude of the ﬁeld:
@B
@g
z
¼ À
Bg
z
g
2
eff
cos
2
y ð4:51Þ
Values of qB/qg
z
 are given in Table 4.16, as are values of w
excess
[computed from
eqn (4.52)].
w
excess
¼ w
g
@B=@g
z
j j ¼ w
g
B=g
z
ð4:52Þ
Figure 4.14 ESR spectrum of [Mn(CO)(dmpe)Cp]
1
in 1:1 CH
2
Cl
2
:C
2
H
4
Cl
2
.
(Reproduced from ref. 48, with permission.) (a) Experimental spectrum at
120 K, (b) simulation with parameters of Table 4.15 and constant 4.1 G
Gaussian line widths, (c) simulation with constant 4.1 G line widths, w
g
¼
0.0049. (Reproduced with permission from ref. 48, copyright (1997)
American Chemical Society.)
Table 4.15 ESR parameters for [Mn(CO)(dmpe)Cp]
148
g
1
g
2
g
3
10
4
A
1
(cm
À1
) 10
4
A
2
(cm
À1
) 10
4
A
3
(cm
À1
) b (1)
2.000 2.021 2.187 35.6 ca. 0 115.3 45.4
88 Chapter 4
The experimental value of w
excess
is deﬁned by w
excess
¼
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
w
2
low
Àw
2
high
where
w
high
¼ 4.1 G (the À3/2 line was used since the À1/2 line was subject to
destructive interferences by a divergence feature). The simulation in Figure
4.14(c) is based on w
g
¼ 0.0049 G. A very similar explanation deals with the
broader lowﬁeld features in the ESR spectrum of [Cr(CO)
2
(PMe
3
)(ZC
5
Ph
5
)]
(ref. 28) (Figure 4.10).
References
1. N.V. Vugman, A.O. Caride and J. Danon, J. Chem. Phys., 1973, 59, 4418.
2. N.V. Vugman and N.M. Pinhal, Mol. Phys., 1983, 49, 1315.
3. J.A. DeGray, P.H. Rieger, N.G. Connelly and G. Garcia Herbosa,
J. Magn. Reson., 1990, 88, 376.
4. P.H. Rieger, J. Magn. Reson., 1982, 50, 485.
5. M.C.R. Symons, J.R. Morton and K.F. Preston, in HighEnergy Processes
in Organometallic Chemistry, ed. K.S. Suslick, American Chemical Society,
Washington, DC, 1987, p. 169.
6. (a) P.W. Atkins and M.C.R. Symons, The Structure of Inorganic Radicals,
Elsevier, Amsterdam, 1967; (b) R. Livingston and H. Zeldes, J. Chem.
Phys., 1964, 41, 4011.
7. D. Collison, B. Gahan and F.E. Mabbs, J. Chem. Soc., Dalton Trans.,
1987, 111.
8. R.S. Drago, Physical Methods in Chemistry, Saunders, New York, 1977,
p. 318.
9. P.C. Taylor, J.F. Baugher and H.M. Kriz, Chem. Rev., 1975, 75, 203.
10. I.V. Ovchinnikov and V.N. Konstantinov, J. Magn. Reson., 1978, 32, 179.
11. P.C. Taylor and P.J. Bray, J. Magn. Reson., 1970, 2, 305.
12. J.A. DeGray and P.H. Rieger, Bull. Magn. Reson., 1987, 8, 95.
13. P.H. Rieger, in Organometallic Processes, ed. W.C. Trogler, 1990, Ch. 8.
14. J.R. Morton and K.F. Preston, J. Magn. Reson., 1978, 30, 577.
15. B.A. Goodman and J.B. Raynor, Adv. Inorg. Chem. Radiochem., 1971
13, 136.
Table 4.16 Analysis of widths of ‘‘parallel’’ features of the spectrum of
[Mn(CO)(dmpe)Cp]
1
.
48
(Data from ref. 48 with permission,
copyright (1997) American Chemical Society.)
m
I
5/2 3/2 1/2 À1/2 À3/2 À5/2
B

(G) 2865 2962 3053 3432 3532 3636
y

(1) 14.5 9.4 3.3 86.4 81.1 78.6
qB/qg
z
 (G) 1240 1324 1392 8 46 105
w (G) 7.2 7.5 8.2 3.1 4.1 5.4
w
excess
(G) 5.9 6.2 7.2 0
w
g
0.0048 0.0047 0.0051
89 Anisotropic ESR Spectra
16. A.K. Koh and D.J. Miller, Atom. Data Nucl. Data Tab., 1985, 33, 235.
17. P.H. Rieger, J. Magn. Reson., 1997, 124, 140.
18. B.M. Peake, P.H. Rieger, B.H. Robinson and J. Simpson, J. Am. Chem.
Soc., 1980, 102, 156.
19. N.M. Atherton, Electron Spin Resonance, Ellis Horwood, Chicester, 1973.
20. N.G. Connelly, W.E. Geiger, G.A. Lane, S.J. Raven and P.H. Rieger,
J. Am. Chem. Soc., 1986, 108, 6219.
21. B.R. McGarvey, in Electron Spin Resonance of Metal Complexes, ed.
T.F. Yen, Plenum Press, New York, 1969, p. 1.
22. A.J. Stone, Proc. Roy. Soc. (London), 1963, A271, 424.
23. T. Kawamura, K. Fukamachi, T. Sowa, S. Hayashida and T. Yonezawa,
J. Am. Chem. Soc., 1981, 103, 364.
24. G.A. Carriedo, V. Riera, N.G. Connelly and S.J. Raven, J. Chem. Soc.,
Dalton Trans., 1987, 1769.
25. G.A. Carriedo, N.G. Connelly, E. PerezCarreno, A.G. Orpen,
A.L. Rieger, P.H. Rieger, V. Riera and G.M. Rosair, J. Chem. Soc.,
Dalton Trans., 1993, 3103.
26. J.A. DeGray, Q. Meng and P.H. Rieger, J. Chem. Soc., Faraday I, 1987, 83,
3565.
27. T.A. Albright, W.E. Geiger, J. Moraczweski and B. Tulyathan, J. Am.
Chem. Soc., 1981, 103, 4787.
28. D.J. Hammack, M.M. Dillard, M.P. Castellani, A.L. Rheingold,
A.L. Rieger and P.H. Rieger, Organometallics, 1996, 15, 4791.
29. T.A. Albright and R. Hoﬀmann, Chem. Ber., 1978, 111, 1390.
30. W. Gordy, Theory and Applications of Electron Spin Resonance, John
Wiley, New York, 1980.
31. M. Katayama and W. Gordy, J. Chem. Phys., 1961, 35, 117.
32. M. Geoﬀroy, L. Ginet and E.A.C. Lucken, Mol. Phys., 1974, 28, 1289.
33. P.B. Ayscough, Electron Spin Resonance in Chemistry, Methuen, London,
1967.
34. A. Abragam and B. Bleaney, Electron Paramagnetic Resonance of Transi
tion Ions, Clarendon Press, Oxford, 1970.
35. J.E. Wertz and J.E. Bolton, Electron Spin Resonance, McGrawHill,
New York, 1972.
36. (a) J.R. Pilbrow and M.R. Lowrey, Rep. Prog. Phys., 1980, 43, 433;
(b) A recent example that illustrates the intricacies involved in extracting
information about axis orientations may be found in C.J. Adams,
I.M. Bartlett, S. Boonyuen, N.G. Connelly, D.J. Harding, O.D. Hayward,
E.J.L. McInnes, A.G. Orpen, M.J. Quayle and P.H. Rieger, Dalton Trans.,
2006, 3466.
37. F.K. Kneubu¨ hl, Phys. Kondens. Mat., 1963, 1, 410; 1965, 4, 50.
38. For a summary of Kneubu¨ hl’s methods, see ref. 36.
39. W.E. Geiger, P.H. Rieger, B. Tulyathan and M.C. Rausch, J. Am. Chem.
Soc., 1984, 106, 7000.
40. T. Lionel, J.R. Morton and K.F. Preston, J. Chem. Phys., 1982, 76, 234.
90 Chapter 4
41. S.A. Fairhurst, J.R. Morton and K.F. Preston, J. Chem. Phys., 1982
77, 5872.
42. P.H. Rieger, Coord. Chem. Rev., 1994, 135, 203.
43. B.M. Peake, P.H. Rieger, B.H. Robinson and J. Simpson, Inorg. Chem.,
1981, 20, 2540.
44. F.A. Cotton, J.D. Jamerson and B.R. Stults, J. Am. Chem. Soc., 1976, 98,
1774.
45. C.G. Howard, G.S. Girolami, G. Wilkinson, M. ThorntonPett and
M.B. Hursthouse, J. Chem. Soc., Dalton Trans., 1983, 2631.
46. A.L. Balch, Inorg. Chem., 1967, 6, 2158.
47. G.B. Carpenter, G.S. Clark, A.l. Rieger, P.H. Rieger and D.A. Sweigart,
J. Chem. Soc., Dalton Trans., 1994, 2903.
48. M.P. Castellani, N.G. Connelly, R.D. Pike, A.L. Rieger and P.H. Rieger,
Organometallics, 1997, 16, 4369.
49. R.D. Pike, A.L. Rieger and P.H. Rieger, J. Chem. Soc., Faraday Trans. I,
1989, 85, 3913.
91 Anisotropic ESR Spectra
CHAPTER 5
ESR Kinetic Studies
One of the ﬁrst uses of ESR spectra to measure the rate of a chemical reaction
was by Ward and Weissman in the early 1950s.
1
They made use of a form of the
Heisenberg uncertainty principle (eqn 5.1) to relate the lifetime of a spin state to
the uncertainty in the energy of the state.
dEdt ! h ð5:1Þ
Thus, if the lifetime of a spin state is dt, the energy level is broadened by an
amount h/dt, with consequences for ESR line widths. Ward and Weissman
1
added some unreduced naphthalene to a solution of the radical anion, and,
from the observed broadening, computed dt, and from dt the rate constant for
the electron transfer reaction:
C
10
H
8
þ½C
10
H
8
À
!½C
10
H
8
À
þC
10
H
8
The result is very rapid electron exchange, as expected, with k ¼ 1 Â 10
6
L mol
À1
s
À1
.
This approach works well for electron transfer reactions where the rate is
simply related to the broadening, but to proceed further in kinetic applications
of ESR spectroscopy we must deal with the Bloch equations and modiﬁed
Bloch equations.
The phenomenological equations proposed by Felix Bloch in 1946
2
have had
a profound eﬀect on the development of magnetic resonance, both ESR and
NMR, on the ways in which the experiments are described (particularly in
NMR), and on the analysis of line widths and saturation behavior. Here we will
describe the phenomenological model, derive the Bloch equations and solve
them for steadystate conditions. We will also show how the Bloch equations
can be extended to treat inter and intramolecular exchange phenomena and
give examples of applications.
5.1 Bloch’s Phenomenological Model
When a magnetic ﬁeld is applied to an electron or nuclear spin, the spin
quantization axis is deﬁned by the ﬁeld direction. Spin magnetic moments
92
aligned with the ﬁeld are only slightly lower in energy than those aligned
opposed to the ﬁeld. If we consider an ensemble of spins, the vector sum of all
the spin magnetic moments will be a nonzero net magnetic moment or
macroscopic magnetization:
~
M ¼ S
i
~m
i
ð5:2Þ
At equilibrium
~
M is in the direction of the ﬁeld
~
B. If somehow
~
M is tilted
away from
~
B there will be a torque that causes
~
M to precess about
~
B with the
equation of motion:
d
~
M
dt
¼ g
~
B Â
~
M ð5:3Þ
where g ¼ 2pgm
B
/h (or gm
B
/h). In addition to the precessional motion, there are
two relaxation eﬀects.
If M
0
is the equilibrium magnetization along
~
B and M
z
is the zcomponent
under nonequilibrium conditions, then we assume that M
z
approaches M
0
with ﬁrstorder kinetics:
dM
Z
dt
¼ À
M
Z
ÀM
o
T
1
ð5:4Þ
where T
1
is the characteristic time for approach to equilibrium (the reciprocal
of the rate constant). Since this process involves transfer of energy from the spin
system to the surroundings (conventionally called the ‘‘lattice’’), T
1
is called the
spin–lattice relaxation time. Since electrons are much more strongly coupled to
molecular interactions than are nuclei (which are buried in a sea of innershell
electrons), it is not surprising that T
1
for electrons is usually much shorter (on the
order of microseconds) than are nuclear T
1
s (on the order of seconds).
There is a second kind of relaxation process that is at least as important for
magnetic resonance as the T
1
process. Suppose that
~
M is somehow tilted down
from the zaxis toward the xaxis and the precessional motion is started. Each
individual magnetic moment undergoes this precessional motion, but the
individual spins may precess at slightly diﬀerent rates. Local shielding may
cause small variations in
~
B or the eﬀective gfactor may vary slightly through
the sample. Thus an ensemble of spins that all start out in phase will gradually
lose phase coherence – the individual spins will get out of step. We assume that
there is a characteristic time for this process, called the transverse relaxation
time, T
2
, and that the transverse magnetization components decay to the
equilibrium value of zero accordingly:
dM
x
dt
¼ À
M
x
T
2
dM
y
dt
¼ À
M
y
T
2
ð5:5Þ
Notice that dephasing of the transverse magnetization does not aﬀect M
z
; a T
2
process involves no energy transfer but, being a spontaneous process, does
involve an increase in the entropy of the spin system.
93 ESR Kinetic Studies
It should be emphasized that the approach to equilibrium by a T
1
process, in
which M
z
approaches M
0
, also causes M
x
and M
y
to approach zero. Thus, the
T
2
of eqn (5.5) must include both the eﬀects of spin–lattice relaxation as well as
the dephasing of the transverse magnetization. Transverse relaxation is often
much faster than spin–lattice relaxation and T
2
is then determined mostly by
spin dephasing. In general, however, we should write:
1
T
2
¼
1
T
1
þ
1
T
2
0
ð5:6Þ
where T
2
0
is the spin dephasing relaxation time, and T
2
is the observed
transverse relaxation time.
In ESR, it is also customary to classify relaxation processes by their eﬀects on
electron and nuclear spins. A process that involves an electron spin ﬂip
necessarily involves energy transfer to or from the lattice and is therefore a
contribution to T
1
; we call such a process nonsecular. A process that involves
no spin ﬂips, but which results in loss of phase coherence, is termed secular.
Processes that involve nuclear spin ﬂips but not electron spin ﬂips are, from the
point of view of the electron spins, nonsecular, but because the energy trans
ferred is so small (compared with electron spin ﬂips) these processes are termed
pseudosecular.
5.1.1 Derivation of the Bloch Equations
Combining eqns (5.3)–(5.5), we have:
d
~
M
dt
¼ g
~
B Â
~
M À
^
i
M
x
T
2
À
^
j
M
y
T
2
À
^
k
M
z
ÀM
0
T
1
ð5:7Þ
In a magnetic resonance experiment, we apply not only a static ﬁeld B
0
in the
zdirection but an oscillating radiation ﬁeld B
1
in the xyplane, so that the total
ﬁeld is:
~
B ¼
^
i B
1
cos ot þ
^
j B
1
sin ot þ
^
kB
0
ð5:8Þ
Note that there are other possible ways to impose a timedependent B
1
. The one
described in eqn (5.8) corresponds to a circularly polarized ﬁeld initially aligned
along the xaxis and rotating about the zaxis in a counterclockwise direction.
The vector product of eqn (5.7) then becomes:
~
B Â
~
M ¼À
^
i B
0
M
y
ÀB
1
M
z
sin ot
Â Ã
þ
^
j B
0
M
x
ÀB
1
M
z
cos ot ½
þ
^
k B
1
M
y
cos ot ÀM
x
sin ot
Â Ã
Inserting this expression in eqn (5.7) and separating it into components, we get:
dM
x
dt
¼ ÀgB
0
M
y
þgB
1
M
z
sin ot À
M
x
T
2
ð5:9aÞ
94 Chapter 5
dM
y
dt
¼ gB
0
M
x
ÀgB
1
M
z
cos ot À
M
y
T
2
ð5:9bÞ
dM
z
=dt ¼ gB
1
M
y
cos ot ÀM
x
sin ot
Â Ã
À M
z
ÀM
0
ð Þ=T
1
ð5:9cÞ
It is convenient to write Mx and My as:
M
x
¼ u cos ot þv sin ot ð5:10aÞ
M
y
¼ u sin ot Àv cos ot ð5:10bÞ
or
u ¼ M
x
cos ot þM
y
sin ot ð5:11aÞ
v ¼ M
x
sin ot ÀM
y
cos ot ð5:11bÞ
This is equivalent to transformation into a coordinate system that rotates
with the oscillating ﬁeld; u is that part of M
x
which is inphase with B
1
and v is
the part which is 901 out of phase. Diﬀerentiating eqn (5.11a) and substituting
eqns (5.9a) and (5.9b), we get:
du
dt
¼
dM
x
dt
cos ot ÀM
x
osin ot þ
dM
y
dt
sin ot þM
y
ocos ot
¼ gB
0
Ào ½ M
x
sin ot ÀM
y
cos ot
Â Ã
À
M
x
cos ot þM
y
sin ot
T
2
and substituting from eqns (5.11):
du
dt
¼ À o ÀgB
0
½ v À
u
T
2
ð5:12aÞ
Similarly, we obtain:
dv
dt
¼ o ÀgB
0
½ u À
v
T
2
þgB
1
M
z
ð5:12bÞ
dM
z
dt
¼ ÀgB
1
v À
M
z
ÀM
0
T
1
ð5:12cÞ
Equations (5.12a–c) are the Bloch equations in the rotating coordinate frame.
5.1.2 Steadystate Solution
In a continuous wave (CW) magnetic resonance experiment, the radiation ﬁeld
B
1
is continuous and B
0
is changed only slowly compared with the relaxation
rates (socalled slow passage conditions). Thus a steadystate solution to eqns
95 ESR Kinetic Studies
(5.12) is appropriate. Setting the derivatives to zero and solving the three
simultaneous equations, we get:
u ¼
gB
1
M
0
ðo
0
ÀoÞT
2
2
1 þT
2
2
ðo
0
ÀoÞ
2
þg
2
B
2
1
T
1
T
2
ð5:13aÞ
v ¼
gB
1
M
0
T
2
1 þT
2
2
ðo
0
ÀoÞ
2
þg
2
B
2
1
T
1
T
2
ð5:13bÞ
M
z
¼
M
0
1 þT
2
2
ðo
0
ÀoÞ
2
h i
1 þT
2
2
ðo
0
ÀoÞ
2
þg
2
B
2
1
T
1
T
2
ð5:13cÞ
where o
0
¼ gB
0
is called the Larmor frequency and corresponds, in a quantum
mechanical description of the experiment, to the (angular) frequency of the
energy level transition.
Notice that as B
1
approaches zero, u and v go to zero and M
z
approaches
M
0
, as expected. That is, it is the transverse oscillating ﬁeld that causes
the magnetization to have a nonequilibrium value. On the other hand, as
B
1
increases, M
z
decreases (moves away from equilibrium); u and v at ﬁrst
increase with increasing B
1
, but eventually they decrease as the third term in the
denominator begins to dominate.
Recall that u is the transverse magnetization component inphase with the
driving ﬁeld B
1
. In general a response that is exactly in phase with a driving
signal does not absorb power from the signal source and in spectroscopy
corresponds to dispersion – in optical spectroscopy dispersion results from a
small reduction of the speed of light as it traverses the medium; but the problem
can also be formulated so that optical dispersion is described in a way closely
analogous to the Bloch equations description. An outofphase response, in
contrast, corresponds to absorption. In magnetic resonance, it is usually the
absorption, or vmode, that is detected and so we conﬁne our attention to
eqn (5.13b) in what follows.
When the microwave or radiofrequency power, proportional to B
1
2
, is small
so that g
2
B
1
2
T
1
T
2
oo 1, eqn (5.13b) becomes:
v ¼
gB
1
M
0
T
2
1 þT
2
2
ðo
0
ÀoÞ
2
ð5:14Þ
A plot of v vs. T
2
(o
0
– o) is shown in Figure 5.1. Equation (5.14) corresponds
to the classical Lorentzian line shape function and the absorption curve of
Figure 5.1 is a Lorentzian ‘‘line’’. The halfwidth at halfheight is easily found
to be:
Do ¼
1
T
2
or Dn ¼
1
2pT
2
or DB ¼
h
gm
B
T
2
where the last form is appropriate when (as in CW ESR) B
0
is changed while
keeping o constant.
96 Chapter 5
When the absorption is detected via small amplitude ﬁeld modulation, the
signal is proportional to the ﬁrst derivative of absorption:
dv
do
¼
2gB
1
M
0
T
3
2
ðo
0
ÀoÞ
1 þT
2
2
ðo
0
ÀoÞ
2
h i
2
ð5:15Þ
Figure 5.2 shows a Lorentzian derivative line.
In ﬁrstderivative spectra, it is most convenient to describe the line width as
the separation between derivative extrema. This width may be computed by
taking the second derivative and ﬁnding the zeros, obtaining:
Do ¼
2
ﬃﬃﬃ
3
p
T
2
or Dv ¼
1
ﬃﬃﬃ
3
p
pT
2
or DB ¼
h
ﬃﬃﬃ
3
p
pgm
B
T
2
The absorption derivative amplitude is proportional to T
2
2
whereas the width is
proportional to T
2
À1
. In other words, the derivative amplitude is inversely
proportional to the square of the line width. Furthermore, the product of the
amplitude and the square of the width is independent of T
2
and is sometimes
taken as a measure of the intensity of the line, i.e., proportional to M
0
.
Notice that in the limit of small B
1
[where eqns (5.14) and (5.15) are valid] the
derivative amplitude increases linearly with B
1
, that the width is independent of
B
1
and that neither width nor amplitude depends on T
1
. At higher power,
g
2
B
1
2
T
1
T
2
cannot be neglected in eqn (5.13b), the amplitude is no longer linear
4 2 0 2 4
T
2
(ω  ω
0
)
Figure 5.1 A Lorentzian absorption line.
97 ESR Kinetic Studies
in B
1
and both amplitude and width depend on T
1
. Eventually, the amplitude
begins to decrease with increasing B
1
and we say that the resonance is saturated.
In quantum mechanical language this corresponds to equalization of the energy
level populations and reduction of net absorption.
5.2 Chemical Exchange – The Modiﬁed Bloch Equations
Suppose we have a system in which a spin can exist in either of two diﬀerent
sites, A or B, and that these are distinguished by diﬀerent resonant frequencies,
o
A
and o
B
, and/or by diﬀerent relaxation times, T
2A
and T
2B
. If there is no
exchange between sites, site A spins and site B spins can be described separately
and independently by sets of Bloch equations. When exchange takes place,
however, additional rate terms – completely analogous to terms in chemical
rate equations – must be added to the Bloch equations.
The algebra we are about to get into can be compacted somewhat by
introducing the complex magnetization, G ¼ u þ iv, so that eqns (5.12a) and
(5.12b) can be combined to obtain:
dG
dt
¼
du
dt
þi
dv
dt
¼ À
G
T2
þiðo
0
ÀoÞG þigB
1
M
0
ð5:16Þ
In eqn (5.16), we have ignored the diﬀerence between M
0
and M
z
and so have
assumed that B
1
is small.
4 2 0 2 4
T
2
(ω  ω
0
)
Figure 5.2 A Lorentzian ﬁrstderivative line.
98 Chapter 5
Spins at site A will have magnetization G
A
and those at site B will have
magnetization G
B
. We now assume that A and B are interconverted by ﬁrstorder
kinetics with an A B rate constant t
A
À1
and a B A rate constant t
B
À1
. The
site A magnetization thus decreases with a rate term –G
A
/t
A
and increases with
rate G
B
/t
B
. Combining these terms with eqn (5.16) for site A:
dG
A
dt
¼ À
G
A
T
2A
À
G
A
t
A
þ
G
B
t
B
þiðo
A
ÀoÞG
A
þigB
1
M
0A
ð5:17aÞ
and a similar expression for site B:
dG
B
dt
¼ À
G
B
T
2B
þ
G
A
t
A
À
G
B
t
B
þiðo
B
ÀoÞG
B
þigB
1
M
0B
ð5:17bÞ
The rather fearsome algebra can be somewhat simpliﬁed by deﬁning:
a
A
¼ T
À1
2A
Àiðo
A
ÀoÞ; a
B
¼ T
À1
2B
Àiðo
B
ÀoÞ
and noting that, at chemical equilibrium:
p
A
p
B
¼
t
A
t
B
and p
A
þp
B
¼ 1
where p
A
and p
B
are the fractions of the population at sites A and B. Deﬁning a
mean lifetime:
t ¼
t
A
t
B
t
A
þt
B
several useful relations result:
p
A
¼
t
A
t
A
þt
B
p
B
¼
t
B
t
A
þt
B
t ¼ p
A
t
B
¼ p
B
t
A
Also, if M
0
is the total equilibrium magnetization, we can write:
M
0A
¼ p
A
M
0
M
0B
¼ p
B
M
0
With these relations, it is possible to ﬁnd the steadystate solution to eqns (5.17):
G ¼ G
A
þG
B
¼ igB
1
M
0
t
A
þt
B
þt
A
t
B
ðp
A
a
A
þp
B
a
B
Þ
1 þt
A
a
A
ð Þ 1 þt
B
a
B
ð Þ À1
ð5:18Þ
The absorption signal, of course, is the imaginary part of eqn (5.18); the
equation is too horrible to contemplate, but computersimulations, such as those
shown in Figures 5.3 and 5.4, are relatively easy to produce. There are two
limiting cases where the equations are easier to understand. In the slow exchange
limit, where t
A
À1
and t
B
À1
are both small compared with o
A
– o
B
, the
99 ESR Kinetic Studies
absorption, v ¼ Im(G), is:
v ¼
gB
1
M
0
p
A
T
À1
2A
þt
À1
A
À Á
T
À1
2A
þt
À1
A
À Á
2
þ o
A
Ào ð Þ
2
þ
gB
1
M
0
p
B
T
À1
2B
þt
À1
B
À Á
T
À1
2B
þt
À1
B
À Á
2
þ o
B
Ào ð Þ
2
ð5:19Þ
which corresponds to two Lorentzian lines centered at o
A
and o
B
and with
widths ðT
2A
À1 þt
A
À1 Þ and ðT
2B
À1 þt
B
À1 Þ. In other words, the lines are
unshifted but are broadened by an amount proportional to the reciprocal of
the lifetimes.
In the fast exchange limit, where t
A
À1
and t
B
À1
are both large compared with
o
A
– o
B
, the absorption is:
v ¼
gB
1
M
0
T
À1
2
T
À2
2
þðo
0
ÀoÞ
2
ð5:20Þ
5 0 5 5 0 5 5 0 5
(ω  ω
0
)
τ = 0.001 τ = 0.01 τ = 0.02
τ = 0.05 τ = 0.1 τ = 0.2
τ = 0.5 τ = 1.0 τ = 10
Figure 5.3 Absorption curves, computed using eqn (5.18) for various values of t, and
o
A
¼ o
0
À 5, o
B
¼ o
0
þ5, T
2A
À1
¼ T
2B
À1
¼ 0.5, P
A
¼ P
B
¼ 0.5.
100 Chapter 5
where
o
0
¼ p
A
o
A
þp
B
o
B
1
T
2
¼
p
A
T
2A
þ
p
B
T
2B
þp
A
p
B
ðo
A
Ào
B
Þ
2
t
Thus a single Lorentzian line is obtained that is centered at a weighted average
resonant frequency and has a width proportional to a weighted average
T
À1
2
plus a term proportional to the average lifetime and the square of the
separation of the slow exchange resonances.
In the socalled intermediate exchange region, eqn (5.18) is not easily tractable
and recourse is usually made to computer simulations. Qualitatively, however,
it is clear that as the rate increases, the separate resonances of the slow
exchange limit broaden, shift together, coalesce and then begin to sharpen
into the single line of the fast exchange limit.
5 0 5 5 0 5 5 0 5
(ω  ω
0
)
x2/3
τ = 0.001
x4
τ = 0.05
x20
τ = 0.1
x20
τ = 0.2
x1
τ = 10
x4
τ = 1.0
x10
τ = 0.5
x1
τ = 0.01
x2
τ = 0.02
Figure 5.4 Firstderivative curves, computed using eqn (5.18) for various values of t,
and o
A
¼o
0
À5, T
2A
À1 ¼ T
2B
À1 ¼ 0:5, p
A
¼p
B
¼0.5; note that the vertical
scale diﬀers – the plots are magniﬁed by the factors shown.
101 ESR Kinetic Studies
Simulations spanning all three exchange regions are shown in Figures 5.3
and 5.4.
5.3 Further Discussion of Line Shapes
Lorentzian line shapes are expected in magnetic resonance spectra whenever the
Bloch phenomenological model is applicable, i.e., when the loss of magnetiza
tion phase coherence in the xyplane is a ﬁrstorder process. As we have seen, a
chemical reaction meets this criterion, but so do several other line broadening
mechanisms such as averaging of the g and hyperﬁne matrix anisotropies
through molecular tumbling (rotational diﬀusion) in solution.
Some sources of line broadening, however, cannot be thought of as ﬁrstorder
rate processes. For example, when the magnetic ﬁeld is inhomogeneous and
varies over the sample, not all the molecules are at resonance at the same nominal
ﬁeld. It is usually reasonable to guess that the ﬁeld has a Gaussian distribution
(i.e., a normal error distribution), so that the resonance line will also have a
Gaussian shape. Various other eﬀects can lead to such inhomogeneous broadening
and thus to Gaussian line shapes. For example, the instantaneous chemical
environment may vary from one radical to another because of diﬀerent degrees
of solvation or ion pairing. Probably for such reasons, Gaussian line shapes are
commonly observed in frozen solution ESR spectra. Unresolved hyperﬁne
couplings also often give a resonance line with an approximately Gaussian shape.
The Gaussian line shape function can be written:
SðoÞ ¼ e
ÀðoÀo
0
Þ
2
=2d
2
ð5:21Þ
so that the halfwidth at halfheight is
Half width ¼
ﬃﬃﬃ
2
p
ln 2 d ¼ 0:980 d
and the derivative width (between extrema) is exactly 2d.
A Gaussian line and its ﬁrst derivative are shown in Figures 5.5 and 5.6.
Comparison with Figures 5.1 and 5.2 shows that the Gaussian line is somewhat
fatter near the middle but lacks the broad wings of the Lorentzian line.
5.4 Applications of the Modiﬁed Bloch Equations
ESR spectra of the radical anion of bis(diphenylphosphino)maleic anhydride
(BMA),
3
are shown in Figure 5.7. These spectra provide a good example of
dynamical linewidth eﬀects. This radical shows a 1:2:1 triplet spectrum at
higher temperatures, a
P
¼ 3.52 G at 240 K, but the phosphorus coupling is
strongly temperature dependent and extra lines appear at 200 K and below,
indicating a dynamical equilibrium between two isomeric forms of the radical,
one having two equivalent P nuclei (1:2:1 triplet spectrum), the other two non
equivalent P nuclei (doublet of doublets spectrum). The lines broaden with
increasing temperature and coalesce at about 220 K.
102 Chapter 5
These data can be understood in terms of two rotational isomers of BMA
À
(1 and 2).
C C
Ph
2
P
PPh
2
C
O
O O
C C
Ph
2
P
C
O
O O
1 2
C
Ph
2
P
C
4 2 0 2 4
(ω−ω
0
)/δ
Figure 5.5 Gaussian absorption line.
4 2 0 2 4
(ω−ω
0
)/δ
Figure 5.6 Firstderivative Gaussian line.
103 ESR Kinetic Studies
Detailed analysis led to the thermodynamic and kinetic parameters: DH1 ¼
0.8 Æ 0.2 kJ mol
À1
, DS1 ¼ –4 Æ 1 J mol
À1
K
À1
(K
160
¼ 3.0); DH
w
¼ 18.2 Æ 0.4
kJ mol
À1
, DS
w
¼ À30 Æ 2 J mol
À1
K
À1
(k
200
¼ 1.9 Â 10
6
s
À1
).
A very similar application of the modiﬁed Bloch equations was based
in the work of Adams and Connelly.
4
ESR spectra (Figure 5.8) of
[Mo{P(OMe)
3
}
2
(MeCRCMe)Cp] show the expected triplet (two equivalent
31
P nuclei) at 280 K, but only a doublet at 160 K. At intermediate temperatures,
the lines broaden. The interpretation is that the alkyne undergoes a pendulum
oscillation, which in the extrema diverts spin density from one or the other
phosphite. Interestingly, the diamagnetic cation undergoes a similar motion on
the NMR time scale, but then the alkyne undergoes a complete rotation. Thus,
analysis of the eﬀect leads to a measure of the rate of the oscillation. The
3365 3370 3375 3380
Magnetic Field/Gauss
280 K
260 K
240 K
220 K
200 K
180 K
160 K
Figure 5.7 Isotropic ESR spectra of the BMA radical anion in THF solution at various
temperatures.
(Reproduced with permission from ref. 3, copyright (1998) American
Chemical Society.)
104 Chapter 5
resulting kinetic parameters are DH
w
¼ 13.2 Æ 0.3 kJ mol
À1
, DS
w
¼ À14 Æ
2 J mol
À1
K
À1
, k
298
¼ 1.0 Â 10
9
s
À1
.
Casagrande et al.
5
have used linewidth eﬀects to study the rate of ﬂuxiona
lity in (Ph
2
C
2
)Co(CO)[P(OMe)
3
]
2
. The experimental spectrum (Figure 5.9a),
3340 3360
240 K
3340 3360
Magnetic Field/Gauss
220 K
3340 3360 3340 3360
220 K
180 K
Figure 5.8 ESR spectra of [Mo{P(OMe)
3
}
2
(MeCRCMe)Cp] in 2:1 THF–CH
2
Cl
2
.
(Reproduced with permission from ref. 4, copyright (2001) Royal Society
of Chemistry.)
B/Gauss
3500
(c)
(b)
(a)
3000
Figure 5.9 (a) Experimental spectrum of (Ph
2
C
2
)Co(CO)[P(OMe)
3
]
2
in THF solution
at 270 K; (b and c) Computersimulated spectra: (b) the m
Co
and m
P
line
width dependence and (c) the m
Co
line width dependence only.
(Adapted from ref. 5 with permission, copyright (1984) American Chemical
Society.)
105 ESR Kinetic Studies
can be described as a 1:2:1 triplet of octets (I ¼ 7/2 for
59
Co); the spectrum
is complicated by a large line width dependence on m
Co
, but, as demonstrated
in Figure 5.9(b) and 5.9(c), the central lines of the triplets are much broader
than the outer lines. This radical has a distorted tetrahedral structure with
the singly occupied molecular orbital (SOMO) largely cobalt 3 d
z
2 in character.
6
Thus the ligand sites can be described as axial or equatorial relative to the
unique zaxis. Several isomers are possible, but the
31
P couplings distinguish
between the isomer with an axial phosphite (ax,eq) and those with either CO
or the acetylene axial and both phosphites equatorial (eq,eq). The rate of
interconversion between (eq,eq) and (ax,eq) isomers (3 and 4) was estimated
from the relative widths of the m
P
¼ Æ 1 and 0 lines, given the isotropic
coupling constants for the various
31
P nuclei (which were determined from
the frozen solution spectrum
5
). The average rate was found to be approxi
mately 2 Â 10
10
s
À1
(E
a
¼ 17 Æ 2 kJ mol
À1
) at 298 K.
3 4
Co
P(OMe)
3
P(OMe)
3
C
O
Co
C
C
Ph
Ph
O
C
P(OMe)
3
P(OMe)
3
C
C
Ph
Ph
z
z
(ax,eq) (eq,eq)
F. A. Walker et al. studied the rate of base exchange with VO(acac)
2
in
benzene solution:
7
VOðacacÞ
2
þBase
!
VOðacacÞ
2
Á Base
where Base ¼ pyridine, 3picoline, 4picoline, 3,4lutidine, 3,5lutidine, and
piperidine, by using ESR line broadening. Line widths were ﬁtted to a power
series expression:
T
À1
2
¼ a þbm
I
þgm
2
I
þdm
3
I
Table 5.1 shows the spin Hamiltonian parameters for some of the complexes.
The equilibrium constants and rate constants are given in Table 5.2.
Table 5.1 ESR parameters for VO(acac)
2
Á Base
7
g 10
4
A
V
(cm
À1
)
VO(acac)
2
pyridine 1.9693 99.5
VO(acac)
2
2picoline 1.9697 96.7
VO(acac)
2
piperidine 1.9690 96.9
106 Chapter 5
5.5 Alternating Line Width Eﬀects
Several examples were found during the 1950s and 1960s of spectra of organic
radical ions in which the line widths were alternatively sharp and broad. One
example of this is the ESR spectrum of dinitrodurene (5) (Figure 5.10).
NO
2
H
3
C
O
2
N
CH
3
CH
3
CH
3
5
Table 5.2 Equilibrium and rate constants for VO(acac)
2
.
Base exchange.
(Reproduced from ref. 7a with permission, copyright (1966)
American Institute of Physics.)
Base T (1C) K (M
À1
)
a
k
r
(s
À1
) k
r
(M
À1
s
À1
)
Pyridine 18 78 1.3 Â 10
7
1.0 Â 10
9
2Picoline 16 0.78
b
7.2 Â10
7
5.0 Â 10
6
Piperidine 17 1400 o0.2 Â 10
7
o2.8 Â 10
9
a
Ref. 7b.
b
Using estimated DH1.
τ = 2 × 10
11
s
τ = 2 × 10
10
s
τ = 2 × 10
9
s
50 40 30 20 10 0 10 20 30 40 50
(B – B
0
)/Gauss
Figure 5.10 Simulations of the ESR spectrum of the dinitrodurene radical anion.
Parameters taken from ref. 8.
107 ESR Kinetic Studies
This spectrum was at ﬁrst quite puzzling, but, on reﬂection, the explanation
became clear.
8
The unpaired electron is mostly on the electronwithdrawing nitro
groups in the pdinitrobenzene radical anion but, in the present case, the methyl
groups do not allow the two nitro groups to be coplanar with the ring simul
taneously. Thus the unpaired electron hops back and forth between the two nitro
groups, with the one having the unpaired electron coplanar with the ring. The
simulations in Figure 5.10 show the eﬀect of rate on the shape of the spectrum.
Another example of alternating line width eﬀect was found in the spectra of
durosemiquinone (6),
9,10
where the eﬀect is due to alkali metal ions hopping
back and forth from one oxygen atom to the other. The rates depend on the
alkali metal as shown in Table 5.3.
O
CH
3
H
3
C
O
CH
3
CH
3
6
Another example is the mdinitrobenzene anion radical in aqueous solution,
11,12
where the eﬀect is due to asymmetric solvation (one nitro group solvated, the
other not), an eﬀect very similar to that with dinitrodurene anions. In this case the
mean lifetime of one solvation state was 0.8 ms at 291 K and 4.5 ms at 282 K. Still
more examples are mentioned in the reviews by Atkins in the early 1970s.
13
5.6 Spin Labels
One of the many advances in ESR spectroscopy introduced by Harden
McConnell is the idea of labeling a biological membrane or macromolecule
with a stable free radical (typically a nitroxide).
14
Since in a nitroxide the spin is
primarily in a nitrogen 2p orbital, A

aA
>
, and the isotropic line widths give a
good indication of the freedom of motion of the nitroxide. Moro and Freed
15
developed an algorithm for the analysis of ESR spectra to give motional
Table 5.3 Eﬀect of cation and solvent on hopping rate
in durosemiquinone
Cation/solvent
a
Log A E
a
(kJ mol
À1
)
K
1
/THF 10.7 24
Rb
1
/THF 10.9 22
Cs
1
/THF 11.1 20
Na
1
/DME 12
K
1
/DME 28
a
THF ¼ tetrahydrofuran (ref. 9); DME ¼ dimethoxyethane (ref. 10).
108 Chapter 5
lifetimes. The subject has been reviewed nearly annually in the Specialist
Periodical Report, Electron Spin Resonance and other monographs.
16
The
eﬀects of decrease in the rate of rotation motion on the appearance of nitroxide
ESR is shown qualitatively in Figure 5.11.
Applications of spin labels to problems in structural biology have continued
to grow over the four decades since McConnell’s original proposal. We
mention here only two examples, which provided early support for the method.
Morrisett, et al.
17
studied phase transitions in E. coli membranes using three
diﬀerent spin labels, including 5doxyl stearate (5DS) (7).
7
O
N O
(CH
2
)
3
COOH H
3
C(H
2
C)
12
They observed abrupt changes in the slope of Arrhenius plots for reactions
catalyzed by NADH oxidase and plactate oxidase that correlate well with
phase transitions detected by the ESR spectra of the nitroxide spin labels bound
covalently to the enzymes (Table 5.4).
Another example comes from the work of Johnson, et al.
18
These workers
studied spin labels dissolved in lipid bilayer dispersions of dipalmitoylphos
phatidylcholine and cholesterol (9:1 by weight) in the hope that anisotropic
rotational diﬀusion of the spin label would mimic the motion of the bilayer
components. In addition to 5DS, which is sensitive to rotational motion about
the NO bond, they used the steroidal nitroxide 8, which tends to rotate about
an axis perpendicular to the N–O bond. ESR measurements were carried out at
both 9 and 35 GHz and at temperatures ranging from –30 to 30 1C. Rather
diﬀerent results were obtained with the two spin labels, largely as a result of the
diﬀerent axes of rotation. Because the rotation rates were very slow, ESR
spectra appeared as powder patterns rather than isotropic spectra and special
methods were needed to extract the motional data.
8
N
O
H
C
6
H
13
O
109 ESR Kinetic Studies
References
1. R.L. Ward and S.I. Weissman, J. Am. Chem. Soc., 1954, 76, 3612.
2. F. Bloch, Phys. Rev., 1946, 70, 460.
3. N.W. Duﬀy, R.R. Nelson, M.G. Richmond, A.L. Rieger, P.H. Rieger,
B.H. Robinson, D.R. Tyler, J.C. Wang and K. Yang, Inorg. Chem., 1998,
37, 4849.
4. C.J. Adams, N.G. Connelly and P.H. Rieger, Chem. Commun., 2001, 2458.
5. L.V. Casagrande, T. Chen, P.H. Rieger, B.H. Robinson, J. Simpson and
S.J. Visco, Inorg. Chem., 1984, 23, 2019.
3330 3335 3340 3345 3350 3355
Magnetic Field/Gauss
fast
slower
Figure 5.11 Eﬀect of a slowing of the rate of rotational motion on the simulated ESR
spectrum of a typical nitroxide spin label. Broadening becomes even more
pronounced and nonuniform as the rate is further decreased.
Table 5.4 Transition temperatures for enzyme activity and nitroxide mo
tion
17
Activity Activity ESR
NADH oxidase pLactate oxidase Phase trans. temp (1C)
T
2b
(1C)
a
27 29.0
T
2m
(1C)
a
32 31 31.0
T
2e
(1C)
a
36 33.0
a
T
2b
¼ transition beginning; T
2m
¼ transition midpoint; T
2e
¼ transition endpoint.
110 Chapter 5
6. J.A. DeGray, Q. Meng and P.H. Rieger, J. Chem. Soc., Faraday Trans. I,
1987, 83, 3565.
7. (a) F.A. Walker, R.L. Carlin and P.H. Rieger, J. Chem. Phys., 1966
45, 4181; (b) R.L. Carlin and F.A. Walker, J. Am. Chem. Soc., 1965
87, 2128.
8. J.H. Freed and G.K. Fraenkel, J. Chem. Phys., 1962, 37, 1156.
9. L. Pasimeni, M. Brustoon and C. Corvaia, J. Chem. Soc., Faraday Trans.
II, 1972, 223.
10. P.S. Gill and T.E. Gough, Trans. Faraday Soc., 1968, 64, 1997.
11. D. Jones and M.C.R. Symons, Trans. Faraday Soc., 1971, 67, 961.
12. J.H. Freed, P.H. Rieger and G.K. Fraenkel, J. Chem. Phys., 1962, 37, 1881.
13. P. W. Atkins, Electron Spin Resonance, Specialist Periodical Report, Royal
Society of Chemistry, London, 1973, 1, 47; 1974, 2, 52; 1976, 3, 35; 1977, 4, 1.
14. T.J. Stone, T. Buckman, P.L. Nordio and H.M. McConnell, Proc. Nat.
Acad. Sci. U.S.A., 1965, 54, 1010.
15. G. Moro and J.H. Freed, J. Chem. Phys., 1981, 74, 3757, 75, 3157.
16. T. Lund and J. B. Raynor, Electron Spin Resonance, Specialist Periodical
Report, Royal Society of Chemistry, London, 1974, 2, 295; J. B. Raynor,
ibid. 1976, 3, 256; P. F. Knowles and B. Peake, ibid. 1977, 4, 212;
B. M Peake, ibid. 1978, 5, 256; B. M. Peake, ibid. 1980, 6, 233; B. H.
Robinson, ibid. 1981, 7, 293; B. H. Robinson and A. H. Beth, ibid. 1983, 8,
346; C.S. Lai, ibid. 1983, 8, 378; C.S. Lai, ibid. 1984, 9, 246; C.S. Lai, ibid.
1987, 10A, 116; C.S. Lai, ibid. 1991, 12A, 74; D. Marsh, ibid. 1993, 14, 166;
P. Tordo, ibid. 1997, 16, 116; Spin Labeling Theory and Applications, ed.
L. J. Berliner, Academic Press, New York, 1976; M. J. Davies, Electron
Paramagnetic Resonance 18, Royal Society of Chemistry, Cambridge, UK,
2003, ch 2, pp. 47–75.
17. J.D. Morrisett, H.J. Pownall, R.T. Plumlee, L.C. Smith, Z.E. Zehner,
M. Eseahani and S.J. Wakil, J. Biol. Chem., 1975, 250, 6969.
18. M.E. Johnson, L. Lee and L.W.M. Fung, Biochemistry, 1982, 21, 4459.
111 ESR Kinetic Studies
CHAPTER 6
ESR Spectra of Biradicals,
Triplet States, and other S41/2
Systems
6.1 Biradicals
Molecules with two or more unpaired electrons may be divided into two classes:
by far the most common examples are molecules where the unpaired electrons
are contained in a set of degenerate atomic or molecular orbitals with quali
tatively similar spatial distributions, e.g., an octahedral Cr(III) (
4
A
2g
) or Ni(II)
(
3
A
2g
) complex, a ground state triplet molecule like O
2
, or the excited triplet
states of naphthalene or benzophenone.
A second class of molecules with two unpaired electrons has the two
electrons localized, to a ﬁrst approximation, in diﬀerent parts of the molecule.
We refer to such molecules as biradicals (or, with three or more unpaired
electrons, triradicals, etc.). Examples are the dinitroxides (e.g. 1), and certain
binuclear vanadium(IV) and copper(II) complexes.
N O N O
1
From the point of view of ESR spectroscopy, the distinction between molecules
with one unpaired electron and those with more than one lies in the fact that
electrons interact with one another; these interactions lead to additional terms
in the spin Hamiltonian and additional features in the ESR spectrum. The most
important electron–electron interaction is coulombic repulsion; with two un
paired electrons, repulsion leads to the singlet–triplet splitting. As we will see,
this eﬀect can be modeled by adding a term, J
~
S
1
Á
~
S
2
, to the spin Hamiltonian,
112
where J is called the exchange coupling constant and turns out to be equal to
energy diﬀerence between the singlet and triplet states, i.e. the singlet–triplet
splitting.
In many cases, the singlet–triplet splitting is large compared with kT, as well
as any other term in the spin Hamiltonian, and we can safely ignore the singlet
state, focusing entirely on the triplet. The simpliﬁcation of being able to ignore
the singlet state is more than compensated for by the introduction of a ﬁne
structure term into the spin Hamiltonian.
This term, which has the form shown in eqn (6.1) introduces considerable
complication into the shape and interpretation of ESR spectra.
D S
2
Z
À
1
3
SðS þ1Þ
þE S
2
x
ÀS
2
y
ð6:1Þ
Further complicating the situation is the fact that the same term can arise
from two quite diﬀerent physical eﬀects: electron–electron dipolar interaction
and spin–orbit coupling.
The distinction between a biradical and an ordinary triplet state molecule is
often somewhat fuzzy. For our purposes, we consider a molecule a biradical
if the exchange interaction between the two electrons is relatively weak –
comparable in energy to the electron–nuclear hyperﬁne interaction. When the
exchange interaction is weak, the singlet–triplet splitting is small and (as we will
see) singlet–triplet mixing has an eﬀect on the ESR spectrum. J falls oﬀ with
electron–electron distance as 1/r, whereas the dipolar coupling constant D falls
oﬀ as 1/r
3
. Thus when J is small, D is negligible (though spin–orbit eﬀects may
contribute).
6.1.1 Exchange Coupling
In this section we consider the spin Hamiltonian appropriate to a biradical with
weak dipolar coupling and see how ESR spectra of such species should appear.
Obviously, it is possible to ﬁnd triradicals, tetraradicals, etc.; treatment of such
species is similar, though of course somewhat more complicated.
The spin Hamiltonian for a biradical consists of terms representing the
electron Zeeman interaction, the exchange coupling of the two electron spins,
and hyperﬁne interaction of each electron with the nuclear spins. We assume
that there are two equivalent nuclei, each strongly coupled to one electron and
essentially uncoupled to the other. The spin Hamiltonian is:
H
s
¼ gm
B
BðS
1z
þS
2z
Þ þAð
~
S
1
Á
~
I
1
þ
~
S
2
Á
~
I
2
Þ þJ
~
S
1
Á
~
S
2
ð6:2Þ
where J is the exchange coupling constant. Notice that we have also assumed that
the gvalues for the two electrons are the same. To simplify matters, we will
assume that A {gm
B
B so that a ﬁrstorder treatment of the hyperﬁne term will
suﬃce. We choose as basis functions the singlet and triplet electron spin functions:
S
0
j i ¼
1
ﬃﬃ
2
p
1
2
; À
1
2
À À
1
2
;
1
2
ð6:3aÞ
113 ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
T
0
j i ¼
1
ﬃﬃ
2
p
1
2
; À
1
2
þ À
1
2
;
1
2
ð6:3bÞ
T
1
j i ¼
1
2
;
1
2
ð6:3cÞ
T
À1
j i ¼ À
1
2
; À
1
2
ð6:3dÞ
The singlet function corresponds to zero total electron spin angular momen
tum, S¼0; the triplet functions correspond to S¼1. Operating on these
functions with the spin Hamiltonian, we get:
^
H
s
T
1
j i ¼ gm
B
B þ
1
4
J þ
1
2
Aðm
1
þm
2
Þ
T
1
j i
^
H
s
T
À1
j i ¼ Àgm
B
B þ
1
4
J À
1
2
Aðm
1
þm
2
Þ
T
À1
j i
^
H
s
T
0
j i ¼
1
4
J T
0
j i þ
1
2
Aðm
1
Àm
2
Þ S
0
j i
^
H
s
S
0
j i ¼ À
3
4
J S
0
j i þ
1
2
Aðm
1
Àm
2
Þ T
0
j i
Thus T
1
i and T
À1
i are eigenfunctions of
^
H
s
, but T
0
i and S
0
i are mixed.
(Notice, however, that if there were no hyperﬁne coupling, A¼0, then T
0
i and
S
0
i would be eigenfunctions as well.) In the absence of a hyperﬁne interaction,
the triplet energy is J/4 and the singlet energy is À3J/4; J is normally negative so
that the triplet lies lower in energy.
To get the eigenvalues resulting from the admixture of T
0
i and S
0
i, we solve
the secular equation:
J
4
ÀE
A
2
ðm
1
Àm
2
Þ
A
2
ðm
1
Àm
2
Þ À
3J
4
ÀE
¼ E
2
þ
1
2
JE À
3
16
J
2
À
1
4
A
2
ðm
1
Àm
2
Þ
2
¼ 0
The roots are:
E ¼ À
1
4
J Æ
1
2
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
J
2
þA
2
ðm
1
Àm
2
Þ
2
or, deﬁning:
R ¼
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
J
2
þA
2
ðm
1
Àm
2
Þ
2
we have:
E ¼ À
J
4
Æ
R
2
114 Chapter 6
The eigenfunctions corresponding to these energies may be found by inserting a
value of E into one of the linear equations that lead to the secular equation:
1
4
J ÀE
c
T
þ
1
2
Aðm
1
Àm
2
Þ c
S
¼ 0
1
2
Aðm
1
Àm
2
Þ c
T
þ À
3
4
J ÀE
c
S
¼ 0
where c
T
and c
S
are the coeﬃcients of T
0
i and S
0
i in the eigenfunc
tion corresponding to E(c
2
T
þ c
2
S
¼1). The resulting eigenfunctions and
energies are:
E
1
¼ gm
B
B þ
J
4
þ
A
2
ðm
1
þm
2
Þ 1 j i ¼ T
1
j i
E
2
¼ À
J
4
þ
R
2
2 j i ¼
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
R þJ
2R
T
0
j i þ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
R ÀJ
2R
S
0
j i
E
3
¼ À
J
4
À
R
2
3 j i ¼
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
R ÀJ
2R
T
0
j i À
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
R þJ
2R
S
0
j i
E
4
¼ Àgm
B
B þ
J
4
À
A
2
ðm
1
þm
2
Þ 4 j i ¼ T
À1
j i
Remembering that each of these is further split by the hyperﬁne interaction,
there are obviously several possible transitions among these four energy levels.
To ﬁnd out which are important, we must evaluate the transition dipole
moment matrix elements, hiS
x
ji, since the absorption intensity is proportional
to the square of these matrix elements. The operator S
x
can be written:
^
S
x
¼
^
S
1x
þ
^
S
2x
¼
1
2
^
S
1þ
þ
^
S
1À
þ
^
S
2þ
þ
^
S
2À
Applying S
x
to 1i and 4i, we have:
^
S
x
1 j i ¼
1
2
^
S
1þ
þ
^
S
1À
þ
^
S
2þ
þ
^
S
2À
1
2
;
1
2
¼
1
2
À
1
2
;
1
2
þ
1
2
; À
1
2
¼
ﬃﬃﬃ
2
p
T
0
j i
^
S
x
4 j i ¼
1
2
^
S
1þ
þ
^
S
1À
þ
^
S
2þ
þ
^
S
2À
À
1
2
; À
1
2
¼
1
2
1
2
; À
1
2
þ À
1
2
;
1
2
¼
ﬃﬃﬃ
2
p
T
0
j i
Thus the matrix elements are:
1
^
S
x
1
¼ 4
^
S
x
4
¼ 1
^
S
x
4
¼ 4
^
S
x
1
¼ 0
2
^
S
x
1
¼ 2
^
S
x
4
¼ 1
^
S
x
2
¼ 4
^
S
x
2
¼
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
R þJ
R
3
^
S
x
1
¼ 3
^
S
x
4
¼ 1
^
S
x
3
¼ 4
^
S
x
3
¼
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
R ÀJ
R
115 ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
Since the relative intensity of a transition is proportional to the square of the
corresponding matrix element of
^
S
x
, we see that there are four allowed
transitions:
DE
42
¼ gm
B
B þ
1
2
ðR ÀJÞ þ
1
2
Aðm
1
þm
2
Þ
Relative intensity ¼
R þJ
R
DE
43
¼ gm
B
B À
1
2
ðR þJÞ þ
1
2
Aðm
1
þm
2
Þ
Relative intensity ¼
R ÀJ
R
DE
31
¼ gm
B
B þ
1
2
ðR þJÞ þ
1
2
Aðm
1
þm
2
Þ
Relative intensity ¼
R ÀJ
R
DE
21
¼ gm
B
B À
1
2
ðR ÀJÞ þ
1
2
Aðm
1
þm
2
Þ
Relative intensity ¼
R þJ
R
Consider now the limiting case of strong exchange coupling. When J 44A,
R EJ, and the 43 and 31 transitions are forbidden. The 42 and 21 transitions
are at equal energy and so we have only:
DE ¼ gm
B
B þ
1
2
Aðm
1
þm
2
Þ ð6:4Þ
Thus in the limit of strong exchange interaction, the resulting spectrum is
identical to that which would be observed if one electron interacted with two
equivalent nuclei with coupling constant A/2.
In the limiting case of weak coupling, J { A, R E A(m
1
À m
2
), and all
four transitions have equal relative intensities; the transition energies then are:
DE
42
¼DE
31
¼gm
B
B þ
A
2
ðm
1
Àm
2
Þ þ
A
2
ðm
1
þm
2
Þ
¼gm
B
B þAm
1
ð6:5aÞ
DE
43
¼DE
21
¼gm
B
B À
A
2
ðm
1
Àm
2
Þ þ
A
2
ðm
1
þm
2
Þ
¼gm
B
B þAm
2
ð6:5bÞ
Thus in the limit of negligible exchange interaction, we expect a spectrum
identical to that observed for two independent radicals.
116 Chapter 6
Now consider a concrete example. Suppose we have a nitroxide biradical
with a
N
¼13 G. In the strong exchange limit, we expect a ﬁveline spectrum
with a spacing of 6.5 G and the usual 1:2:3:2:1 intensity ratios for two
equivalent spin1 nuclei. In the weak exchange limit, we expect a threeline
spectrum with a spacing of 13 G and intensity ratios 1:1:1. In intermediate
cases, up to 15 lines are expected, as shown in Figure 6.1.
In a classic early study of biradicals, Glarum and Marshall
1
were able to vary
the exchange coupling between two nitroxide radicals by changing the temper
ature, solvent, and the number of atoms linking the two groups. Most of their
spectra corresponded to the intermediate exchange region where J E A.
6.2 Organic Triplet State Molecules and the Dipolar
Interaction
The Hamiltonian term for the electron–electron dipolar interaction is:
^
H
d
¼ g
2
m
2
B
~s
1
Á ~s
2
r
3
À3
~s
1
Á ~r ð Þ ~s
2
Á ~r ð Þ
r
5
¸
ð6:6Þ
Figure 6.1 Stick spectra for a dinitroxide biradical with a ¼13 G for various values of
the exchange coupling constant J. (Several very small, widely spaced
resonances have been omitted for J ¼2A and J ¼5A.)
117 ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
where r is the vector pointing from electron 1 to electron 2. We have used a
lowercase s for the oneelectron spin operators, reserving uppercase S for the
total electron spin operators. The dot products can be expanded to give:
^
H
d
¼g
2
m
2
B
r
2
À3x
2
r
5
^ s
1x
^s
2x
þ
r
2
À3y
2
r
5
^ s
1y
^ s
2y
¸
þ
r
2
À3z
2
r
5
^s
1z
^s
2z
À
3xy
r
5
^ s
1x
^s
2y
þ ^ s
2y
^ s
1x
À
3yz
r
5
^ s
1y
^ s
2z
þ^ s
1z
^s
2y
À
3zx
r
5
^s
1z
^s
2x
þ^s
1x
^ s
2z
ð Þ
Our next goal is to transform this expression into one based on the total
electron spin operator, S¼s
1
þ s
2
. The ﬁrst three terms can be simpliﬁed by
making use of the identity (derived using raising and lowering operators):
2s
1i
s
2i
¼ S
2
i
À
1
2
ði ¼ x; y; zÞ
When these are substituted in the above expression, the terms arising from –1/2
cancel since x
2
þ y
2
þ z
2
¼r
2
. Transformation of the last three terms makes use
of the identities:
2ðs
1i
s
2j
þs
1j
s
2i
Þ ¼ S
i
S
j
þS
j
S
i
The transformed Hamiltonian then is:
^
H
d
¼
1
2
g
2
m
2
B
r
2
À3x
2
r
5
¸
^
S
2
x
þ
r
2
À3y
2
r
5
^
S
2
z
þ
r
2
À3z
2
r
5
^
S
2
z
À
3xy
r
5
^
S
x
^
S
y
þ
^
S
y
^
S
x
À
3yz
r
5
^
S
y
^
S
z
þ
^
S
z
^
S
y
À
3zx
r
5
^
S
z
^
S
x
þ
^
S
x
^
S
z
The coeﬃcients of the spin operators must be evaluated using the electron wave
function, an operation that is not usually possible in practice. However, we can
parameterize the problem, deﬁning the matrix D with elements:
D
ij
¼
1
2
g
2
m
2
B
r
2
d
ij
À3ij
r
5
ð6:7Þ
where the angle brackets indicate averaging over the spatial coordinates of the
wave function.
It is possible to choose an axis system in which the Dmatrix is diagonal – the
principal axes. In many cases, these axes will also be the principal axes of the
gmatrix and we will so assume in the following.
Since the trace of D (D
xx
þ D
yy
þ D
zz
) is zero, there are really only two
independent parameters. The conventional choice of these parameters is:
D ¼
3
2
D
zz
E ¼
1
2
D
xx
ÀD
yy
118 Chapter 6
or:
D
zz
¼
2
3
D D
xx
¼ À
1
3
DþE D
yy
¼ À
1
3
DÀE
The Hamiltonian then becomes:
^
H
d
¼ D
xx
^
S
2
x
þD
yy
^
S
2
y
þD
zz
^
S
2
z
ð6:8aÞ
or:
^
H
d
¼ D
^
S
2
z
À
1
3
^
S
2
x
þ
^
S
2
y
þ
^
S
2
z
þ
1
2
E
^
S
2
x
þ
^
S
2
y
ð6:8bÞ
We can somewhat simplify the Hamiltonian by noting that S
x
2
þ S
y
2
þ
S
z
2
¼S
2
and the eigenvalue of S
2
is S(S þ 1), and that S
x
and S
y
can be written
in terms of the raising and lowering operators:
^
H
d
¼ D
^
S
2
z
À
1
3
SðS þ1Þ
þE
^
S
2
þ
þ
^
S
2
À
ð6:9Þ
We will see that a Hamiltonian term identical in form also arises from spin–
orbit coupling, but ﬁrst we will pause to see the eﬀect of this Hamiltonian on
the energy levels and ESR spectrum of a tripletstate molecule. The spin triplet
wave functions can be written in the notation S,m
S
i:
^
H
d
1; 1 j i ¼ D 1 À
2
3
1; 1 j i þE 1; À1 j i
^
H
d
1; 0 j i ¼ D 0 À
2
3
1; 0 j i
^
H
d
1; À1 j i ¼ D 1 À
2
3
1; À1 j i þE 1; 1 j i
The Hamiltonian matrix then is:
1
3
D 0 E
0 À
2
3
D 0
E 0
1
3
D
¸
¸
Solution of the corresponding secular equation leads to energy levels, À2D/3
and D/3 Æ E. The splitting pattern is shown in Figure 6.2.
Notice that if the molecule has axial symmetry, D
xx
¼D
yy
so that E¼0. If the
molecule has octahedral symmetry, D
xx
¼D
yy
¼D
zz
so that D¼E¼0. Thus the
appearance of a zeroﬁeld splitting into two or three levels tells the spectroscopist
something about the symmetry of the molecule. It is possible, of course, to do
spectroscopy on these energy levels at zero magnetic ﬁeld. Our concern here is the
eﬀect of zeroﬁeld splitting on the ESR spectrum where a magnetic ﬁeld is applied.
When we include the Zeeman interaction term, gm
B
BÁ S, in the spin
Hamiltonian a complication arises. We have been accustomed to evaluating
the dot product by simply taking the direction of the magnetic ﬁeld to deﬁne the
zaxis (the axis of quantization). When we have a strong dipolar interaction, the
119 ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
molecule deﬁnes a quantization axis for itself. Thus, in general, the BÁ S term
has three components. It is possible to deal with the general case, but the
algebra is very messy and not very enlightening. Instead we will assume that
the triplet molecule is in a dilute single crystal and that we can orient the crystal
in the ﬁeld with B along one of the internal coordinate axes.
Suppose that we orient the crystal with B in the zdirection. The spin
Hamiltonian then is:
^
H
s
¼ gm
B
B
^
S
z
þD
^
S
2
z
À
1
3
SðS þ1Þ
þ
1
2
E
^
S
2
þ
þ
^
S
2
À
Operating on the triplet wave functions as before, we get the Hamiltonian
matrix:
gm
B
B þ
1
3
D 0 E
0 À
2
3
D 0
E 0 Àgm
B
B þ
1
3
D
¸
¸
Solution of the secular equation leads to the energies:
À
2
3
D;
1
3
D Æ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
g
2
m
2
B
B
2
þE
2
These are plotted vs. magnetic ﬁeld in Figure 6.3, using D¼0.1003 cm
À1
,
E¼À0.0137 cm
À1
– parameters appropriate to the excited triplet of
naphthalene.
2
The energies of the allowed transitions are:
DE ¼
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
g
2
m
2
B
B
2
þE
2
ÆD
Figure 6.2 Zeroﬁeld splitting of a triplet state.
120 Chapter 6
At constant frequency, n
0
¼DE/h, the resonant ﬁelds are:
B ¼
1
gm
B
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
hn
0
ÆD ð Þ
2
ÀE
2
With the magnetic ﬁeld oriented along the xaxis, the Hamiltonian is:
^
H
s
¼ gm
B
B
^
S
x
þD
^
S
2
z
À
1
3
SðS þ1Þ
þ
1
2
E
^
S
2
þ
þ
^
S
2
À
The Hamiltonian matrix is:
1
3
D
1
ﬃﬃ
2
p
gm
B
B E
1
ﬃﬃ
2
p
gm
B
B À
2
3
D
1
ﬃﬃ
2
p
gm
B
B
E
1
ﬃﬃ
2
p
gm
B
B
1
3
D
¸
¸
¸
The cubic secular equation factors; the resulting energies are:
D
3
ÀE; À
DÀ3E
6
Æ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
g
2
m
2
B
B
2
þ
1
4
ðDþEÞ
2
These energies are also plotted in Figure 6.3; energies of the allowed transitions
are:
DE ¼
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
g
2
m
2
B
B
2
þ
1
4
ðDþEÞ
2
Æ
1
2
ðDÀ3EÞ
The resonant ﬁelds then are:
B ¼
1
gm
B
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
hn
0
Æ
1
2
ðDÀ3EÞ
2
À
1
4
ðDþEÞ
2
Figure 6.3 Energy level diagram for the triplet state of naphthalene (D¼0.1003 cm
À1
,
E¼À0.0137 cm
À1
, g ¼2.003). Solid lines correspond to orientation of
the magnetic ﬁeld along the zaxis, dashed lines for orientation along the
xaxis. Arrows show the allowed transitions for 9.50 GHz microwave
radiation.
121 ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
6.2.1 Organic Triplet State Molecules
The phosphorescent triplet state of naphthalene, for which the energy levels are
shown in Figure 6.3, is produced by irradiation of naphthalene doped into a
single crystal of durene. This prevents relaxation eﬀects due to rapid transfer of
energy between triplet and ground state naphthalene molecules. The triplet
state is longlived at 77 K. The energylevel diagram (Figure 6.3) predicts two
transitions: at 2315 and 4465 G when the ﬁeld is oriented along the zaxis, and
at 2595 and 4125 G for orientation along the xaxis (n
0
¼9.50 GHz).
The experimental determination of D and E for a dilute single crystal is not
trivial, even when the crystal axes are known. Durene, for example, has two
molecules per unit cell with diﬀerent orientations of the molecular plane. Thus
for any orientation there are four resonances, two from each type of site.
Sorting out the data is a challenging exercise.
2
Triplet state powder spectra (or frozen solution glasses) are generally easier to
interpret and much easier to get experimentally than dilute single crystal spectra.
The features of the derivative spectrum correspond to orientations along the
principal axis directions. Thus, six features can be found in the spectrum of
naphthalene in glassy THF solution at 77 K (after irradiation). The problem, as
is usual with powder spectra, is that there is no way to assign the features to
molecular axes; recourse must be made to theoretical considerations or
to analogy with a related system studied in a dilute single crystal. Since the
resonances are spread over a large ﬁeld range and are very orientation depend
ent, there is little hope of detecting the resonance of a triplet state molecule in
liquid solution, even if the triplet state lifetime could be made long enough.
What do we make of the parameters D and E once we have extracted them
from a spectrum? Seven examples are given in Table 6.1.
There are some qualitative trends that make some sense. Since D is a measure
of the dipolar interaction of the two unpaired electrons, we might expect that
D would be large when both electrons are forced to be close together, and
naphthalene and the related nitrogen heterocyclics do indeed give the largest
values of D. When the psystem remains about the same size but heteroatoms
are substituted, we might expect that D would not change by much, and we see
that naphthalene and related nitrogen heterocyclics do indeed have nearly the
same values of D. When the psystem gets bigger, we might expect D to
decrease, and anthracene and pyrene fulﬁll this expectation, but phenanthrene
seems anomalous. To go beyond a qualitative explanation of D or to explain
E at all requires rather sophisticated valence theory calculations. Indeed
electron dipolar interaction parameters provide one of the more challenging
tests of valence theory.
6.3 Transition Metal Complexes with S41/2
6.3.1 Spin–Orbit Coupling
We now will show that spin–orbit coupling can give a spin Hamiltonian term
identical to that we obtained from the electron dipolar interaction. Consider the
122 Chapter 6
spin Hamiltonian including orbital angular momentum and the usual spin–
orbit coupling term:
^
H ¼ m
B
~
B Á
~
L þg
e
~
S
þl
~
L Á
~
S ð6:10Þ
where l is the spin–orbit coupling constant. The zeroorder ground state wave
function will be characterized by the quantum numbers L, m
L
, S, m
S
, and we
Table 6.1 ESR data for some organic tripletstate molecules
Molecule Structure D (cm
À1
) E (cm
À1
) Ref
Naphthalene (in durene) 0.1003 –0.0137 2
Quinoline (in durene)
N
0.1030 –0.0162 3
Isoquinoline (in durene)
N
0.1004 –0.0117 3
Quinoxaline (in durene)
N
N
0.1007 –0.0182 4
Anthracene (in biphenyl) 0.0716 –0.0084 5
Phenanthrene (in biphenyl) 0.1004 –0.0466 6
Pyrene (in ﬂuorene) 0.0678 –0.0314 7
123 ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
assume that the wave function is orbitally nondegenerate, i.e., m
L
¼0. Thus we
write the ground state wave function as m
L
,m
S
i ¼0,m
S
i. With B deﬁning the
zaxis, the energy is easily found to ﬁrst order in perturbation theory:
E
ð1Þ
¼ gm
B
Bm
s
ð6:11Þ
since L
z
0,m
S
i ¼0. The secondorder energy is not so simple since the excited
states in general have nonzero m
L
. Thus we get secondorder contributions
from matrix 0,m
S
i elements connecting with excited state functions:
E
ð2Þ
¼ À
¸
i
0; m
S;0
m
B
~
B þl
~
S
Á
~
L þg
e
m
B
~
B Á
~
S m
L;i
; m
S;i
2
E
ð1Þ
i
ÀE
ð1Þ
0
ð6:12Þ
The matrix element can be expanded and written as:
0; m
S;0
m
B
~
B þl
~
S
m
S;i
Á 0
~
L
m
L;i
þg
e
m
B
~
B Á m
S;0
~
S
m
S;i
0jm
L;i
Since the orbital functions, h0 and m
L,i
i are orthogonal, the second term
vanishes. The absolute value square of the matrix element of a Hermitean
operator can be written as:
ijOpjj h i j j
2
¼ ijOpjj h i jjOpji h i
Thus we can write E
(2)
as:
E
ð2Þ
¼
¸
1
m
S
0
m
S;0
m
B
~
B þl
~
S
m
S
0
Á
¸
i
0j
~
Ljm
L
m
L
j
~
Lj0
E
ð1Þ
i
ÀE
ð1Þ
0
¸
¸
Á m
S
0
jm
B
~
B þl
~
Sjm
S;0
¸
Deﬁning:
L
$
¼
¸
i
0j
~
Ljm
L
m
L
j
~
Lj0
E
ð1Þ
i
ÀE
ð1Þ
0
ð6:13Þ
we can write E
(2)
as:
E
ð2Þ
¼
¸
1
m
S
0
m
2
B
m
S
j
~
B Á L
$
Á
~
Bjm
S
0
þl
2
m
S
j
~
S Á L
$
Á
~
Sjm
S
0
þ2lm
B
m
S
j
~
B Á L
$
Á
~
Sjm
S
0
¸
ð6:14Þ
124 Chapter 6
We now notice that we could write a Hamiltonian operator that would give
the same matrix elements we have here, but as a ﬁrstorder result. Including the
electron Zeeman interaction term, we have the resulting spin Hamiltonian:
^
H
s
¼ g
e
m
B
~
B Á
~
S þm
2
B
~
B Á L
$
Á
~
B þ2lm
B
~
B Á L
$
Á
~
S þl
2
~
S Á L
$
Á
~
S ð6:15Þ
The m
2
B
~
B Á L
$
~
B term is independent of spin state and so changes all levels by
the same amount. Although the term would be important to the thermo
dynamic properties of the system, it is uninteresting to spectroscopists and we
will ignore it. The ﬁrst and third terms can be combined to obtain the gtensor:
g ¼ g
e
E þ2lL
$
ð6:16Þ
where E is the unit matrix. We can also deﬁne the ﬁne structure tensor D as:
D ¼ l
2
L
$
ð6:17Þ
so that the spin Hamiltonian reduces to:
H
s
¼ m
B
~
B Á g Á
~
S þ
~
S Á DÁ
~
S ð6:18Þ
Notice that the ﬁne structure term found here has the same form (and the
tensor is given the same symbol) as that obtained from the electron dipolar
interaction. Unlike the dipolar Dtensor, however, the spin–orbit coupling
Dtensor in general does not have zero trace. Nonetheless, we introduce
analogous parameters:
D ¼ D
zz
À
1
2
D
xx
ÀD
yy
E ¼
1
2
D
xx
ÀD
yy
In the coordinate system that diagonalizes g, the related Dtensor is also
diagonal. Expanding the ﬁne structure term in the principal axis system, we have:
~
S Á DÁ
~
S ¼ D
xx
S
2
x
þD
yy
S
2
y
þD
zz
S
2
z
and substituting:
~
S Á DÁ
~
S ¼D S
2
z
À
1
3
S
2
x
þS
2
y
þS
2
z
þE S
2
x
ÀS
2
y
þ
1
3
D
xx
þD
yy
þD
zz
S
2
x
þS
2
y
þS
2
z
Since S
2
¼S
2
x
þ S
2
y
þ S
2
z
and the eigenvalue of S
2
is S(S þ 1) we have:
~
S Á DÁ
~
S ¼D S
2
z
À
1
3
SðS þ1Þ
þE S
2
x
ÀS
2
y
þ
1
3
D
xx
þD
yy
þD
zz
SðS þ1Þ
125 ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
The last term (which would be zero if D came from the dipolar interaction and
thus had zero trace) raises all levels equally and so has no eﬀect on spectroscopy
and can be dropped. Thus, again, only two parameters, D and E, are required to
completely specify the ﬁne structure interaction.
Although it is unfortunate that spin–orbit coupling and the electron dipolar
interaction give ﬁne structure terms of the same form, it is possible to separate
the eﬀects. Since the spin–orbit contribution to D is related to the gtensor:
D
so
¼
l
2
g Àg
e
E ð Þ
the parameters D
so
and E
so
can be computed:
D
so
¼
l
2
g
zz
À
1
2
g
xx
þg
yy
¸
E
so
¼
l
4
g
xx
Àg
yy
The diﬀerence between the ﬁne structure parameters computed from the
experimental gtensor and those measured from the spectrum are presumed
to be the electron dipolar contributions.
In the above derivation, we have made no explicit assumption about the total
electron spin quantum number S so that the results should be correct for S¼1/2
as well as higher values. However, the ﬁne structure term is not usually included
in spin Hamiltonians for S¼1/2 systems. The ﬁne structure term can be ignored
since in that case the results of operating on a spin1/2 wave function is always
zero:
D S
2
z
À
1
3
S S þ1 ð Þ
þ
1
2
E S
2
þ
þS
2
À
¸ ¸
1
2
; Æ
1
2
¼ D
1
4
À
1
3
Â
1
2
Â
3
2
þ
1
2
E Â0
¸ ¸
1
2
; Æ
1
2
¼ 0
6.3.2 Highspin Transition Metal Ions
For axially symmetric complexes, the parameter E is zero, and the spin
functions S,m
S
i are eigenfunctions of the spin Hamiltonian:
H
s
¼ m
B
~
B Á g Á
~
S þD S
2
z
À
1
3
SðS þ1Þ
For example, consider a d
3
Cr(III) complex in an axial ligand ﬁeld with g ¼1.98,
D¼0.0455 cm
À1
, E¼0. For the magnetic ﬁeld along the molecular zaxis, the
energies are:
E Æ
3
2
¼Æ
3
2
g
jj
þD
E Æ
1
2
¼Æ
1
2
g
jj
ÀD
These energies are plotted vs. magnetic ﬁeld in Figure 6.4.
126 Chapter 6
Transitions among these levels have intensities proportional to the square of
the matrix element of S
x
. These are easily found to be:
3
2
; Æ
3
2
S
x
3
2
; Æ
1
2
¼
ﬃﬃ
3
p
2
3
2
;
1
2
S
x
3
2
; À
1
2
¼ 1
Thus the À3/2 À1/2 and 1/2 3/2 transitions, at B¼hn
0
(1 Æ 2D)/gm
B
, have
relative intensities of 3/4 the intensity of the À1/2 1/2 transition, at hn
0
/gm
B
.
When the magnetic ﬁeld is oriented along the xaxis, the Hamiltonian matrix is:
D
ﬃﬃ
3
p
2
g
?
m
B
B 0 0
ﬃﬃ
3
p
2
g
?
m
B
B ÀD g
?
m
B
B 0
0 g
?
m
B
B ÀD
ﬃﬃ
3
p
2
g
?
m
B
B
0 0
ﬃﬃ
3
p
2
g
?
m
B
B D
¸
¸
¸
¸
¸
¸
Unfortunately, the secular equation doesn’t factor and the energies must be
computed numerically. A plot of the computed energies is shown in Figure 6.4 as
a function of magnetic ﬁeld.
As we might expect from the nondiagonal Hamiltonian matrix, the spin
functions are thoroughly mixed when the ﬁeld is in the xdirection. The
immediate consequence of this mixing is that the selection rules are compli
cated, and the transition from the lowest level to the highest level becomes
allowed. (For the ﬁeld along the zaxis, this would be a forbidden twoquantum
transition.) When B¼1000 G, corresponding to the lowesttohighest transi
tion, the wave functions are:
E ¼ À0:179 cm
À1
; 1 j i ¼À0:255
3
2
À À
3
2
þ0:660
1
2
À À
1
2
1
0.5
0
0.5
1
E
n
e
r
g
y
/
c
m
–
1
0 1000 2000 3000 4000
Magnetic Field/Gauss
B  z
0 1000 2000 3000 4000 5000
Magnetic Field/Gauss
B  x
Figure 6.4 Energy levels and allowed transitions for a Cr(III) complex with g ¼1.98,
D¼0.0455 cm
À1
, for the magnetic ﬁeld along the x and zaxes.
127 ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
E ¼ À0:037 cm
À1
; 2 j i ¼À0:513
3
2
þ À
3
2
þ0:487
1
2
þ À
1
2
E ¼ þ0:079 cm
À1
; 3 j i ¼0:660
3
2
À À
3
2
þ0:255
1
2
À À
1
2
E ¼ þ0:137 cm
À1
; 4 j i ¼0:487
3
2
þ À
3
2
þ0:513
1
2
þ À
1
2
When the ﬁeld is along the zaxis, transition intensities are proportional to the
square of the S
z
2
matrix element. The S
z
2
matrix for B¼1000 G is:
0 1:197 0 0:0011
1:197 0 0:794 0
0 0:794 0 0:508
0:0011 0 0:508 0
¸
¸
¸
¸
The 1 4 transition is only weakly allowed compared with the 1 2, 2 3,
and 3  4 transitions; however, it is often observed, particularly in powder
spectra since it tends to be considerably sharper than the other transitions.
Notice that the 1 3 and 2 4 transitions are still forbidden. Since the wave
functions are ﬁelddependent, the S
z
matrix elements also depend on the ﬁeld.
Thus the observed 1 2, 2 3, and 3 4 transitions would be diﬀerent than
predicted from the S
z
2
matrix at 1000 G.
For Cr(III) complexes, D is relatively small (comparable to the Xband
microwave quantum, 0.317 cm
À1
) and all three ﬁne structure lines are observ
able. This is not always the case. Consider highspin Fe(III) in an axial ligand
ﬁeld with D 44 hn
0
, E¼0. With the same Hamiltonian as above and the
magnetic ﬁeld along the zaxis, the energies are:
E Æ
5
2
¼Æ
5
2
g
k
m
B
B þ
10
8
D
E Æ
3
2
¼Æ
3
2
g
k
m
B
B À
2
3
D
E Æ
1
2
¼Æ
1
2
g
k
m
B
B À
8
3
D
These are plotted vs. B in Figure 6.5.
The transition energies are:
E
5
2
ÀE
3
2
¼ g
k
m
B
B þ4D
E
3
2
ÀE
1
2
¼ g
k
m
B
B þ2D
E
1
2
ÀE À
1
2
¼ g
k
m
B
B
128 Chapter 6
E À
1
2
ÀE À
3
2
¼ g
k
m
B
B À2D
E À
3
2
ÀE À
5
2
¼ g
k
m
B
B À4D
However, if D 4 hn
o
, only the À1/2 1/2 transition will be observable. The
ﬁrst two transitions are always higher in energy then hn
o
and it is usually not
possible to make B large enough to bring the last two transitions into
resonance.
Now consider what happens when the ﬁeld is applied perpendicular to the
symmetry axis. The large value of D ensures that z will continue to be the
quantization axis. We ought to solve a 6 Â 6 secular equation, but we can get a
reasonable approximation more easily. Since D is big, the Æ5/2 and Æ3/2 levels
are well separated from the Æ1/2 levels before application of the magnetic ﬁeld.
Thus mixing of 3/2i with 1/2i will be much less important than mixing of 1/2i
with À1/2i. Thus we can solve just the middle 2 Â 2 block for the energies of
m
S
¼Æ1/2 levels:
À
8
3
D
3
2
g
?
m
B
B
3
2
g
?
m
B
B À
8
3
D
The expanded block gives:
À
8
3
DÀE
2
À
3
2
g
?
m
B
B
2
¼ 0
so that the energies are:
E ¼ À
8
3
DÆ
3
2
g
?
m
B
B
2
1
0
1
2
3
E
n
e
r
g
y
/
c
m
–
1
0 1000 2000 3000 4000
Magnetic Field/Gauss
5/2
–5/2
3/2
–3/2
1/2
–1/2
Figure 6.5 Energies of an S¼5/2 spin system with D¼0.5 cm
À1
for B along the zaxis.
129 ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
and the energy diﬀerence is:
E
1
2
ÀE À
1
2
¼ 3g
?
m
B
B
The apparent gvalue for the transition is 3g
>
, g
eﬀ
¼6 if g
>
¼2.
Thus, for example, highspin d
5
Fe(III) in an axial ligand ﬁeld should show
a resonance around g ¼2 and another resonance near g ¼6 when B is perpen
dicular to the symmetry axis. In solution, where the complex tumbles rapidly
and averages the gvalues, the resonance is expected to be so broad as to be
undetectable, but, in frozen solution, both resonances should be observable.
6.3.3 Examples: K
3
Cr(CN)
6
and K
4
V(CN)
6
These two salts were studied by the Oxford physics group in the early days of
ESR spectroscopy.
8
The Cr(III) and V(II) salts were doped into diamagnetic
single crystals of K
3
Co(CN)
6
and K
4
Fe(CN)
6
, respectively. The spin–orbit
coupling parameters found from the spectroscopic study are listed, together
with the gcomponents, in Table 6.2.
Let us see what the energy levels look like for these two systems and try to
understand how Baker, Bleaney, and Bowers determined the values given in
Table 6.2. The spin Hamiltonian is:
^
H ¼ m
B
~
S Á g Á
~
B þD S
2
z
À
1
3
SðS þ1Þ
þE S
2
x
ÀS
2
y
þ
~
I Á A Á
~
S
with eigenfunctions 3/2i, 1/2i, À1/2i, and À3/2i for Cr, and 1i, 0i, and À1i
for V (the
51
V hyperﬁne coupling is essentially isotropic and equal to À55.5 Â
10
À4
cm
À1
). With the ﬁeld along the zaxis, using the Hamiltonian on the
eigenfunctions gives for Cr(CN)
6
3À
:
^
H
3
2
¼
3
2
m
B
g
z
B þD
3
2
^
H
1
2
¼
1
2
m
B
g
z
B ÀD
1
2
^
H À
1
2
¼ À
1
2
m
B
g
z
B ÀD
À
1
2
^
H À
3
2
¼ À
3
2
m
B
g
z
B þD
À
3
2
and for V(CN)
6
4À
:
^
Hj1i ¼ m
B
g
z
B þ
1
3
D
j1i
^
Hj0i ¼ À
2
3
D
j0i
^
Hj À1i ¼ Àm
B
B þ
1
3
D
j À1i
130 Chapter 6
Substituting the values from Table 6.2 into these equations, we get the energy
levels shown in Figure 6.6.
Given the measured positions of the transitions, the values of g
z
and D can be
determined. Orienting the ﬁeld along x and y gives g
x
, g
y
, and E. Ligandﬁeld
theory was still in its infancy in 1956, so one of the goals of Baker, Bleaney, and
Bowers
8
was a test of the theory – of course it passed with ﬂying colors. In the
same paper, K
3
Fe(CN)
6
and K
4
Mn(CN)
6
were also studied. With the CN
À
ligand, which is high in the spectrochemical series, it was expected that both
complexes would be low spin and, sure enough, they were.
Because the d
5
conﬁguration is spherically symmetric, highspin Mn(II) and
Fe(III) usually have nearly isotropic gmatrices and Mn(II) usually has a nearly
isotropic Amatrix. This means that there usually is not much information in
the ESR spectrum of these highspin species. Indeed, highspin Mn(II) is usually
an unwanted interference for those interested in lowspin Mn(II); the ESR
spectrum is very characteristic with six hyperﬁne lines with a coupling constant
of 80–100 G. Because the g and Amatrices are nearly isotropic, the sixline
spectrum persists in frozen solutions.
In part because the highspin Mn(II) signal is so easily detected, several
workers have doped Mn(II) into crystals for which a phase transition was
indicated. Thus Das and Pal
9
have doped CoSiF
6
Á 6H
2
O and Co
1Àx
Zn
x

SiF
6
Á 6H
2
O with Mn(II), replacing some of the Co(II). For CoSiF
6
Á 6H
2
O,
Table 6.2 ESR parameters for two highspin complexes
8
Complex D (cm
À1
) E (cm
À1
) g
x
g
y
g
z
Cr(CN)
6
3À
0.083 0.011 1.993 1.991 1.991
V(CN)
6
4À
–0.0264 –0.0972 1.9919 1.9920 1.9920
1.5
1
0.5
0
0.5
1
1.5
E
/
c
m

1
0 2000 4000 6000 8000
Magnetic Field/Gauss
3/2
1/2
1/2
3/2
0.5
0.25
0
0.25
0.5
0 1000 2000 3000 4000
1
0
Cr(CN)
6
3
V(CN)
6
4
1
Figure 6.6 Energy levels for Cr(CN)
6
3À
and V(CN)
6
4À
for B along the zaxis.
8
131 ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
following the ESR spectrum as a function of temperature, they found a change
in the dipolar parameter D fromÀ176 to À274 Â10
À4
cm
À1
while the hyperﬁne
coupling changed from 93 to 96 Â 10
À4
cm
À1
. These changes correlate with the
phase transition. The temperature of the phase transition depended on the
Mn(II) content; for X
Mn
¼0.082 and 0.182, the transitions occurred at 205, 218
K and 175, 185 K, respectively.
Hirota
10
reported a very similar experiment in which PbCl
2
was doped with
Mn(II). In this experiment, the goal was to study Mn(II) in an ionic medium.
The results are summarized in Table 6.3.
References
1. S.H. Glarum and J.H. Marshall, J. Chem. Phys., 1967, 47, 1374.
2. C.A. Hutchison and C.W. Mangum, J. Chem. Phys., 1958, 29, 952; 1961,
34, 908.
3. J.S. Vincent and A.H. Maki, J. Chem. Phys., 1965, 42, 865.
4. J.S. Vincent and A.H. Maki, J. Chem. Phys., 1964, 39, 3088.
5. J. Grivet, Chem. Phys. Lett., 1969, 4, 104.
6. R.W. Brandon, R.E. Gerkin and C. A. Hutchison, J. Chem. Phys., 1964,
41, 3717.
7. O.H. Griﬃth, J. Phys. Chem., 1965, 69, 1429.
8. J.M. Baker, B. Bleaney and K.D. Bowers, Proc. Phys. Soc., 1956, B69,
1205.
9. M. Das and A.K. Pal, J. Phys. Chem. Solids, 1987, 48, 903.
10. T. Hirota, Jpn. J. Appl. Phys., 1988, 27, 493.
Table 6.3 ESR parameters for Mn(II) in PbCl
2
10
g A
a
D
a
E
a
x 1.9996(2) 79.7(5) 463.8(1) 127.9(1)
y 2.0004(2) 79.6(2)
z 2.003(1) 80.9(2)
a
A, D, and E in units of 10
À4
cm
À1
. Estimated errors in last signiﬁcant
ﬁgure shown in parentheses.
132 Chapter 6
CHAPTER 7
Perturbation Theory
Calculations
7.1 Secondorder Perturbation Theory Treatment
of Spin Hamiltonian with Noncoincident
g and Aaxes
In Chapter 4 (Sections 4.7 and 4.8) several examples were presented to illustrate
the eﬀects of noncoincident g and Amatrices on the ESR of transition metal
complexes. Analysis of such spectra requires the introduction of a set of
Eulerian angles, a, b, and g, relating the orientations of the two coordinate
systems. Here is presented a detailed description of how the spin Hamiltonian is
modiﬁed, to secondorder in perturbation theory, to incorporate these new
parameters in a systematic way. Most of the calculations in this chapter were
ﬁrst executed by Janice DeGray.
1
Some of the details, in the notation used here,
have also been published in ref. 8.
The key to success turns out to be taking careful account of the various axis
systems (Table 7.1).
7.1.1 The Electron Zeeman Term
H ¼ m
B
~
S Á g Á
~
B
~
S is quantized along g Á
~
B i.e.:
~
S Á g Á
~
B ¼ gBS
z
0
Table 7.1 Various axis systems
Coordinate system Coordinates
gMatrix principal axes x, y, z
AMatrix principal axes X, Y, Z
Electron spin quantization axes x
0
, y
0
, z
0
Nuclear spin quantization axes x
00
, y
00
, z
00
133
Since S ¼ (S
x
0
, S
y
0
, S
z
0
), it is necessary that:
g Á
~
B ¼
0
0
gB
¸
¸
The x,y,z coordinate system is deﬁned by the gmatrix principal axes:
g ¼
g
x
0 0
0 g
y
0
0 0 g
z
¸
¸
The magnetic ﬁeld vector in that coordinate system is:
~
B ¼
Bsin y cos f
Bsin y sin f
Bcos y
¸
¸
where y and j are the usual polar angles.
We therefore have:
g Á
~
B ¼
g
x
Bsin y cos f
g
y
Bsin y sin f
g
z
Bcos y
¸
¸
The transformation from the x,y,z axes to x
0
,y
0
,z
0
(the axis system in which
~
S is
quantized) must then take this form to:
g Á
~
B ¼
g
x
Bsin y cos f
g
y
Bsin y sin f
g
z
Bcos y
¸
¸
)
0
0
gB
¸
¸
Q Á g Á
~
B ¼ Q Á
g
x
Bsin y cos f
g
y
Bsin y sin f
g
z
Bcos y
¸
¸
¼
0
0
gB
¸
¸
where Q is a transformation matrix corresponding to rotations of the axes
by the Euler angles z (rotation about yaxis), x (rotation about new zaxis),
w (rotation about new yaxis) (Whittaker deﬁnitions
2,3
):
Q ¼
Q
11
Q
12
Q
13
Q
21
Q
22
Q
23
Q
31
Q
32
Q
33
¸
¸
where:
Q
11
¼ cos z cos x cos w À sin z sin w
Q
12
¼ sin z cos x cos w þ cos z sin w
134 Chapter 7
Q
13
¼ Àsin x cos w
Q
21
¼ Àcos z cos x sin w À sin z cos w
Q
22
¼ Àsin z cos x sin w þ cos z cos w
Q
23
¼ sin x sin w Q
31
¼ cos z sin x
Q
32
¼ sin z sin x Q
33
¼ cos x
Thus:
Q
11
g
x
sin y cos j þ Q
12
g
y
sin y sin j þ Q
13
g
z
cos y ¼ 0
Q
21
g
x
sin y cos j þ Q
22
g
y
sin y sin j þ Q
23
g
z
cos y ¼ 0
Q
31
g
x
sin y cos j þ Q
32
g
y
sin y sin j þ Q
33
g
z
cos y ¼ g
These equations are satisﬁed with:
g
2
¼ g
2
?
sin
2
y þ g
2
z
cos
2
y
g
2
?
¼ g
2
x
cos
2
f þ g
2
y
sin
2
f
if:
cos z ¼
g
x
g
?
cos f sin z ¼
g
y
g
?
sin f
cos x ¼
g
z
g
cos y sin x ¼
g
?
g
sin y
Note that the angle w is left indeterminant by this transformation. This amounts
to saying that S
x
and S
y
are not ﬁxed in space by the quantization of
~
S along
g
~
B. The above result was used in deriving eqn (4.4) in Chapter 4.
7.1.2 Nuclear Hyperﬁne Interaction
H
hf
¼
~
S Á A Á
~
I
where the principal axes of A are (X,Y,Z), which are related to the principal
axes of g by the transformation matrix R , which depends in the Euler angles
a, b, and g in the same way that Q depends on z, x, and w. Following Blinder,
4
we assume that
~
I is quantized along the eﬀective or resultant ﬁeld due to the
applied ﬁeld B and the hyperﬁne ﬁeld due to the unpaired electron:
~
B
hf
¼
~
S Á A
To a ﬁrst approximation, we neglect
~
B compared with
~
B
hf
. For B E3000 G, i.e.
Xband ESR, this will be a good approximation for most metal nuclei and for
protons or ﬂuorines with hyperﬁne couplings greater than ca. 20 G.
We label the nuclear spin quantization axes (x
00
,y
00
,z
00
) and require that the S
z
0
term has the form KS
z
0 I
z
.
135 Perturbation Theory Calculations
The hyperﬁne term may be written in the (x
0
,y
0
,z
0
) system:
H
hf
¼
~
S Á Z Á
~
I
where:
Z ¼ S
À1
Á A Á S S ¼ Q Á R
That is, matrix R transforms from (X,Y,Z) to (x,y,z) and matrix Q transforms
from (x,y,z) to (x
0
,y
0
,z
0
), so that matrix S transforms from (X,Y,Z) to (x
0
,y
0
,z
0
).
In the (x
0
,y
0
,z
0
) coordinate system:
S
!
Á Z Á I
!
¼S
x
0 Z
11
I
x
0 þ Z
12
I
y
0 þ Z
13
I
z
0
þ S
y
0 Z
21
I
x
0 þ Z
22
I
y
0 þ Z
23
I
z
0
þ S
z
0 Z
31
I
x
0 þ Z
32
I
y
0 þ Z
33
I
z
0
In the nuclear spin quantization axis system, the last term has the form, KS
z
0 I
z
00
where K is the eﬀective hyperﬁne coupling for the particular orientation. Thus:
I
z
00 ¼
1
K
Z
31
I
x
0 þ Z
32
I
y
0 þ Z
33
I
z
0
where:
K
2
¼ Z
2
31
þ Z
2
32
þ Z
2
33
As it happens, the only components of the Smatrix we will need to know (for
the perturbation theory treatment) are S
31
, S
32
and S
33
:
S
31
¼Q
31
R
11
þ Q
32
R
21
þ Q
33
R
31
¼
g
x
g
sin y cos f cos a cos bcos g À sin a sin g ð Þ
þ
g
y
g
sin y sin f Àcos a cos b sin g À sin a cos g ð Þ
þ
g
z
g
cos y cos a sin b
ð7:1aÞ
S
32
¼Q
31
R
12
þ Q
32
R
22
þ Q
33
R
32
¼
g
x
g
sin y cos f sin a cos b cos g þ cos a sin g ð Þ
þ
g
y
g
sin y sin f Àsin a cos b sin g þ cos a cos g ð Þ
þ
g
z
g
cos y sin a sin b
ð7:1bÞ
S
33
¼Q
31
R
13
þ Q
32
R
23
þ Q
33
R
33
¼ À
g
x
g
sin y cos fsin b cos g
þ
g
y
g
sin y sin f sin b sin g þ
g
z
g
cos y cos b
ð7:1cÞ
136 Chapter 7
Consider the matrix which transforms the
~
I vector from the nuclearspin into
the electronspin quantization axes:
Q
I
Á
I
x
00
I
y
00
I
z
00
¸
¸
¼
I
x
0
I
y
0
I
z
0
¸
¸
where Q
I
can be written in terms of Euler angles a, b and g. Since the hyperﬁne
term can be written:
S
x
0
S
y
0
S
z
0
¸
¸
¸
Á
Z
11
Z
12
Z
13
Z
21
Z
22
Z
23
Z
31
Z
32
Z
33
¸
¸
¸
Á
I
x
0
I
y
0
I
z
0
¸
¸
¸
¼
S
x
0
S
y
0
S
z
0
¸
¸
¸
Á
Z
11
Z
12
Z
13
Z
21
Z
22
Z
23
Z
31
Z
32
Z
33
¸
¸
¸
Á Q
I
Á
I
x
00
I
y
00
I
z
00
¸
¸
¸
we also have:
I
x
0 ¼ cos a cos bcos g À sin a sin g ð ÞI
x
00
þ sin a cos b cos g þ cos a sin g ð ÞI
y
00 À sin bcos gI
z
00
I
y
0 ¼ Àcos a cos bsin g À sin a cos g ð ÞI
x
00
þ Àsin a cos bsin g þ cos a cos g ð ÞI
y
00 þ sin bsin gI
z
00
I
z
0 ¼ cos a sin bI
x
00 þ sin a sin bI
y
00 þ cos bI
z
00
where:
cos a sin b ¼ Z
31
=K
sin a sin b ¼ Z
32
=K
cos b ¼ Z
33
=K
Thus the angles a and b are determined, but g (like w) remains indeterminant.
However, in this case we have a means of approximating g. If Q
I
were used to
transform Z into the nuclear spin quantization axis system, the trace of the
matrix would remain constant. We are multiplying Z from one side only, so
that the trace is not necessarily invariant. However, we can write:
Z
11
cos a cos b cos g À sin a sin g ð Þ
þZ
12
Àcos a cos b sin g À sin a cos g ð Þ þ Z
13
cos a sin b
þZ
21
sin a cos bcos g þ cos a sin g ð Þ
þZ
22
Àsin a cos bsin g þ cos a cos g ð Þ þ Z
23
sin a sin b
% Z
11
þ Z
22
þ Z
33
À K
137 Perturbation Theory Calculations
Collecting terms in cos g and sin g:
Acos g þ Bsin g þ Z
13
cos a sin b þ Z
23
sin a sin b
% Z
11
þ Z
22
þ Z
33
À K
where:
A ¼Z
11
cos a cos b
À Z
12
sin a þ Z
21
sin a cos b þ Z
22
cos a
B ¼ À Z
11
sin a À Z
12
cos a cos b
þ Z
21
cos a À Z
22
sin a cos b
If this were not an approximation, we could evaluate the components of the
Zmatrix, the angles a and b, and solve for g. The nature of the approximation
causes this approach to fail, but we can still look for the value of g that brings
the function closest to a solution. Diﬀerentiating with respect to g, we have:
ÀAsin g þ Bcos g ¼ 0
or:
tan g ¼
B
A
In practice, we must check this result to make sure that we have a minimum and
not a maximum. Now assume that the hyperﬁne term can be written in the
form:
~
S Á Z Á
~
I ¼ S
x
0 S
y
0 S
z
0
D
x
E
x
F
x
D
y
E
y
F
y
0 0 K
¸
¸
I
x
00
I
y
00
I
z
00
¸
¸
~
S Á Z Á
~
I ¼S
x
0 D
x
I
x
00 þ E
x
I
y
00 þ F
x
I
z
00
þ S
y
0 D
y
I
x
00 þ E
y
I
y
00 þ F
y
I
z
00
þ KS
z
0 I
z
00
Given values of a, b and g, calculation of the new parameters D
x
, D
y
, E
x
, E
y
, F
x
and F
y
is straightforward using perturbation theory.
7.1.3 Perturbation Theory Treatment of Hyperﬁne Term
If we expand the hyperﬁne term of the spin Hamiltonian and write the
operators in terms of raising and lowering operators:
S
þ
À
¼ S
x
0 þ
À
iS
y
0 I
þ
À
¼ I
x
00 þ
À
iI
y
00
138 Chapter 7
H
hf
¼ KS
z
0 I
z
00 þ
1
2
F
x
À iF
y
S
þ
I
z
00 þ
1
2
F
x
þ iF
y
S
À
I
z
00
þ
1
4
D
x
À E
y
À i D
y
þ E
x
S
þ
I
þ
þ
1
4
D
x
À E
y
þ i D
y
þ E
x
S
À
I
À
þ
1
4
D
x
þ E
y
À i D
y
À E
x
S
þ
I
À
þ
1
4
D
x
þ E
y
þ i D
y
À E
x
S
À
I
þ
The unperturbed and perturbation Hamiltonians then are:
H
0
¼ gm
B
BS
z
0 þ KS
z
0 I
z
00
H
1
¼
1
2
F
x
À iF
y
S
þ
I
z
00 þ
1
2
F
x
þ iF
y
S
À
I
z
00
þ
1
4
D
x
À E
y
À i D
y
þ E
x
S
þ
I
þ
þ
1
4
D
x
À E
y
þ i D
y
þ E
x
S
À
I
À
þ
1
4
D
x
þ E
y
À i D
y
À E
x
S
þ
I
À
þ
1
4
D
x
þ E
y
þ i D
y
À E
x
S
À
I
þ
Operating on the zeroorder wavefunctions:
1
2
; m
s
; I; m
I
we have
H
0
1
2
; m
s
; I; m
I
¼ gm
B
BS
z
0 þ KS
z
0 I
z
00 ð Þ
1
2
; m
s
; I; m
I
¼ m
s
gm
B
B þ m
s
m
I
K ð Þ
1
2
; m
s
; I; m
I
H
1
1
2
; m
s
; I; m
I
¼
1
2
F
x
À iF
y
S
þ
I
z
00 þ
1
2
F
x
þ iF
y
S
À
I
z
00
1
2
; m
s
; I; m
I
þ
1
4
D
x
À E
y
À i D
y
þ E
x
S
þ
I
þ
1
2
; m
s
; I; m
I
þ
1
4
D
x
À E
y
þ i D
y
þ E
x
S
À
I
À
1
2
; m
s
; I; m
I
þ
1
4
D
x
þ E
y
þ i D
y
À E
x
S
À
I
þ
1
2
; m
s
; I; m
I
þ
1
4
D
x
þ E
y
À i D
y
À E
x
S
þ
I
À
1
2
; m
s
; I; m
I
H
1
1
2
;
1
2
; I; m
I
¼
m
I
2
F
x
þ iF
y
1
2
; À
1
2
; I; m
I
þ
1
4
D
x
À E
y
þ i D
y
þ E
x
Â I I þ 1 ð Þ À m
I
m
I
À 1 ð Þ ½
1
2
1
2
; À
1
2
; I; m
I
À 1
þ
1
4
D
x
þ E
y
þ i D
y
À E
x
Â I I þ 1 ð Þ À m
I
m
I
þ 1 ð Þ ½
1
2
1
2
; À
1
2
; I; m
I
þ 1
139 Perturbation Theory Calculations
H
1
1
2
; À
1
2
; I; m
I
¼
m
I
2
F
x
À iF
y
1
2
;
1
2
; I; m
I
þ
1
4
D
x
À E
y
À i D
y
þ E
x
Â I I þ 1 ð Þ À m
I
m
I
þ 1 ð Þ ½
1
2
1
2
;
1
2
; I; m
I
þ 1
þ
1
4
D
x
þ E
y
À i D
y
À E
x
Â I I þ 1 ð Þ À m
I
m
I
À 1 ð Þ ½
1
2
1
2
;
1
2
; I; m
I
À 1
Thus the signiﬁcant matrix elements are:
1
2
; À
1
2
; I; m
I
H
1
1
2
;
1
2
; I; m
I
¼
m
I
2
F
x
þ iF
y
1
2
; À
1
2
; I; m
I
À 1
H
1
1
2
;
1
2
; I; m
I
¼
1
4
D
x
À E
y
þ i D
y
þ E
x
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m
I
m
I
À 1 ð Þ
1
2
; À
1
2
; I; m
I
þ 1
H
1
1
2
;
1
2
; I; m
I
¼
1
4
D
x
þ E
y
þ i D
y
À E
x
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m
I
m
I
þ 1 ð Þ
1
2
;
1
2
; I; m
I
H
1
1
2
; À
1
2
; I; m
I
¼
m
I
2
F
x
À iF
y
1
2
;
1
2
; I; m
I
þ 1
H
1
1
2
; À
1
2
; I; m
I
¼
1
4
D
x
À E
y
À i D
y
þ E
x
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m
I
m
I
þ 1 ð Þ
1
2
;
1
2
; I; m
I
À 1
H
1
1
2
; À
1
2
; I; m
I
¼
1
4
D
x
þ E
y
À i D
y
À E
x
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m
I
m
I
À 1 ð Þ
Thus the zero, ﬁrst, and secondorder energies are:
E
ð0Þ
þ
À
1=2
¼ þ
À
1
2
gm
B
B þ
À
1
2
Km
I
E
ð1Þ
þ
À
1=2
¼ 0
E
ð2Þ
1=2
¼
m
2
I
4hv
F
2
x
þ F
2
y
þ
1
16hv
D
x
À E
y
2
þ D
y
þ E
x
2
I I þ 1 ð Þ À m
I
m
I
À 1 ð Þ ½
þ
1
16hv
D
x
þ E
y
2
þ D
y
À E
x
2
I I þ 1 ð Þ À m
I
m
I
þ 1 ð Þ ½
140 Chapter 7
E
ð2Þ
À1=2
¼ À
m
2
I
4hv
F
2
x
þ F
2
y
À
1
16hv
D
x
À E
y
2
þ D
y
þ E
x
2
I I þ 1 ð Þ À m
I
m
I
þ 1 ð Þ ½
À
1
16hv
D
x
þ E
y
2
þ D
y
À E
x
2
I I þ 1 ð Þ À m
I
m
I
À 1 ð Þ ½
For the ‘‘allowed’’ transitions, Dm
S
¼ þ 1, Dm
I
¼ 0, the energy is:
DE ¼ hv ¼gm
B
B þ Km
I
þ
m
2
I
2hv
F
2
x
þ F
2
y
þ
m
I
4hv
D
2
x
þ E
2
y
þ D
2
y
þ E
2
x
I I þ 1 ð Þ À m
2
I
ð7:2Þ
Deﬁning:
F
1
¼ F
2
x
þ F
2
y
F
2
¼ D
2
x
þ D
2
y
þ E
2
x
þ E
2
y
the transition energy can be written:
DE ¼ hv ¼ gm
B
B þ Km
I
þ
2F
1
À F
2
ð Þm
2
I
4hv
þ
F
2
I I þ 1 ð Þ
4hv
ð7:3Þ
and the ﬁeld:
B ¼
hv
gm
B
À
Km
I
gm
B
À
2F
1
À F
2
ð Þm
2
I
þ F
2
I I þ 1 ð Þ
4gm
B
hv
ð7:4Þ
Thus we must express F
1
and F
2
in parameters of the original spin
Hamiltonian.
Starting with the two ways of expressing the hyperﬁne Hamiltonian term, we
equate the coeﬃcients of S
x
0
, S
y
0
and S
z
0
:
Z
11
I
x
0 þ Z
12
I
y
0 þ Z
13
I
z
0 ¼ D
x
I
x
00 þ E
x
I
y
00 þ F
x
I
z
00
Z
21
I
x
0 þ Z
22
I
y
0 þ Z
23
I
z
0 ¼ D
y
I
x
00 þ E
y
I
y
00 þ F
y
I
z
00
Z
31
I
x
0 þ Z
32
I
y
0 þ Z
33
I
z
0 ¼ K
or:
Z
11
I
x
0 þ Z
12
I
y
0 þ Z
13
I
z
0 ¼D
x
a
1
I
x
0 þ b
1
I
y
0 þ c
1
I
z
0
þ E
x
a
2
I
x
0 þ b
2
I
y
0 þ c
2
I
z
0
þ F
x
a
3
I
x
0 þ b
3
I
y
0 þ c
3
I
z
0
Z
21
I
x
0 þ Z
22
I
y
0 þ Z
23
I
z
0 ¼D
y
a
1
I
x
0 þ b
1
I
y
0 þ c
1
I
z
0
þ E
y
a
2
I
x
0 þ b
2
I
y
0 þ c
2
I
z
0
þ F
y
a
3
I
x
0 þ b
3
I
y
0 þ c
3
I
z
0
141 Perturbation Theory Calculations
Equating coeﬃcients of I
x
0 , I
y
0 and I
z
0 , we have:
Z
11
¼ D
x
a
1
þ E
x
a
2
þ F
x
a
3
Z
12
¼ D
x
b
1
þ E
x
b
2
þ F
x
b
3
Z
13
¼ D
x
c
1
þ E
x
c
2
þ F
x
c
3
Squaring and summing:
Z
2
11
þ Z
2
12
þ Z
2
13
¼D
2
x
a
2
1
þ b
2
1
þ c
2
1
þ E
2
x
a
2
2
þ b
2
2
þ c
2
2
þ F
2
x
a
2
3
þ b
2
3
þ c
2
3
þ 2D
x
E
x
a
1
a
2
þ b
1
b
2
þ c
1
c
2
ð Þ
þ 2D
x
F
x
a
1
a
3
þ b
1
b
3
þ c
1
c
3
ð Þ
þ 2E
x
F
x
a
2
a
3
þ b
2
b
3
þ c
2
c
3
ð Þ
or:
Z
2
11
þ Z
2
12
þ Z
2
13
¼ D
2
x
þ E
2
x
þ F
2
x
Similarly:
Z
2
21
þ Z
2
22
þ Z
2
23
¼ D
2
y
þ E
2
y
þ F
2
y
Since:
a
1
a
3
þ b
1
b
3
þ c
1
c
3
¼ 0; c
1
¼ À
a
1
a
3
þ b
1
b
3
c
3
Thus:
Z
11
¼D
x
a
1
þ E
x
a
2
þ F
x
a
3
Z
12
¼D
x
b
1
þ E
x
b
2
þ F
x
b
3
Z
13
¼ À D
x
a
1
a
3
þ b
1
b
3
c
3
À E
x
a
2
a
3
þ b
2
b
3
c
3
þ F
x
c
3
or:
a
3
Z
11
¼D
x
a
1
a
3
þ E
x
a
2
a
3
þ F
x
a
2
3
b
3
Z
12
¼D
x
b
1
b
3
þ E
x
b
2
b
3
þ F
x
b
2
3
c
3
Z
13
¼ À D
x
a
1
a
3
þ b
1
b
3
ð Þ À E
x
a
2
a
3
þ b
2
b
3
ð Þ þ F
x
c
2
3
Summing, we have:
F
x
¼ a
3
Z
11
þ b
3
Z
12
þ c
3
Z
13
or:
F
x
¼
1
K
Z
31
Z
11
þ Z
32
Z
12
þ Z
33
Z
13
ð Þ
142 Chapter 7
Similarly:
F
y
¼
1
K
Z
31
Z
21
þ Z
32
Z
22
þ Z
33
Z
23
ð Þ
Expanding the Z matrix in terms of A
x
, A
y
and A
z
and the components of the
Smatrix, and simplifying making use of the orthonormality of rows of the
Smatrix, we have:
Z
33
Z
13
þ Z
32
Z
12
þ Z
31
Z
11
¼ A
2
x
S
31
S
11
þ A
2
y
S
32
S
12
þ A
2
z
S
33
S
13
Z
33
Z
23
þ Z
32
Z
22
þ Z
31
Z
21
¼ A
2
x
S
31
S
21
þ A
2
y
S
32
S
22
þ A
2
z
S
33
S
23
Thus:
F
1
¼ F
2
x
þ F
2
y
¼
1
K
2
A
4
x
S
2
31
þ A
4
y
S
2
32
þ A
4
z
S
2
33
À K
4
ð7:5Þ
K
2
¼ Z
2
11
þ Z
2
12
þ Z
2
13
¼ A
2
z
S
2
13
þ A
2
y
S
2
12
þ A
2
x
S
2
11
D
2
x
þ E
2
x
¼ Z
2
11
þ Z
2
12
þ Z
2
13
À F
2
x
K
2
Z
2
11
þ Z
2
12
þ Z
2
13
À K
2
F
2
x
¼ S
33
S
2
À S
32
S
3
ð Þ
2
A
2
y
A
2
z
þ S
33
S
1
À S
31
S
3
ð Þ
2
A
2
x
A
2
z
þ S
32
S
1
À S
31
S
2
ð Þ
2
A
2
x
A
2
y
D
2
y
þ E
2
y
¼ Z
2
21
þ Z
2
22
þ Z
2
23
À F
2
y
Z
2
21
þ Z
2
22
þ Z
2
23
¼ A
2
z
S
2
23
þ A
2
y
S
2
22
þ A
2
x
S
2
21
K
2
Z
2
21
þ Z
2
22
þ Z
2
23
À K
2
F
2
y
¼ S
33
S
22
À S
32
S
23
ð Þ
2
A
2
y
A
2
z
þ S
33
S
21
À S
31
S
23
ð Þ
2
A
2
x
A
2
z
þ S
32
S
21
À S
31
S
22
ð Þ
2
A
2
x
A
2
y
F
2
¼
1
K
2
A
2
y
A
2
z
S
2
33
þ S
2
32
þA
2
x
A
2
z
S
2
33
þ S
2
31
þ A
2
x
A
2
y
S
2
32
þ S
2
31
ð7:6Þ
143 Perturbation Theory Calculations
7.1.4 Example Application of these Results
Robert Pike has prepared various lowspin Mn(II) complexes. Figure 7.1 shows
the ESR spectrum of [Mn(CO)
2
(PPh
3
)(C
5
H
5
)]
1
in frozen CH
2
Cl
2
/C
2
H
4
Cl
2
(see
also Tables 7.2 and 7.3). Some of Pike’s spectra, and an analysis of the
spectrum simulated in Figure 7.1, have been published.
5
We will now see how to apply the equations derived above. As a ﬁrst step, we
notice that the
31
P coupling is virtually isotropic. Thus, things can be simpliﬁed
by taking the average of the 5/2,1/2 and 5/2,À1/2 assignments, etc.:
The next thing to notice is that the widely spaced features are very approx
imately equally spaced. This suggests that the g and Amatrix principal axes
are noncoincident. You might think that a simple application of the above
equations would suﬃce for a complete analysis. It is not quite so simple, and a
nonlinear leastsquares program is required.
1,6
Table 7.4 shows the ﬁtted
parameters.
2800 2900 3000 3100 3200 3300 3400 3500
Magnetic Field/Gauss
Figure 7.1 Frozen solution ESR spectrum (n ¼ 9.45 GHz) of [Mn(CO)
2
(PPh
3
)(C
5
H
5
)]
1
.
Simulated using parameters in ref. 5.
Table 7.2 Measured line positions (m
Mn
,m
P
) for [Mn(CO)
2
(PPh
3
)(C
5
H
5
)]
1a
5/2,1/2 3/2,1/2 1/2,1/2 À1/2,1/2 À3/2,1/2 À5/2,1/2
Min 2836.3 2930.4 3024.1 3116.1 3207.4 3295.3
Sp 3215.1 3246.5 3279.5 3314.0 3350.4 3388.3*
Max 3309.1* 3320.2 3340.6 3372.4 3417.1 3473.6
5/2,À1/2 3/2,À1/2 1/2,À1/2 À1/2,À1/2 À3/2,À1/2 À5/2,À1/2
Min 2865.5 2959.6 3053.4 3145.9 3236.6 3324.6
Sp 3246.5 3277.8 3310.8 3245.5 3381.8 3419.7*
Max 3340.8* 3352.0 3372.4* 3404.3 3448.8 3505.1
a
Min ¼ minimum, Sp ¼ saddle point, and Max ¼ maximum; those indicated with an asterisk (*)
are poorly resolved.
144 Chapter 7
7.2 Quadrupole Coupling
The spectra discussed in Chapter 4 were analyzed by neglecting the eﬀects of
nuclear quadrupole coupling on the nuclear hyperﬁne structure. Presented here
is the way such eﬀects may be incorporated into the spectra using perturbation
theory.
The quadrupole coupling term in the spin Hamiltonian can be written as:
H
Q
¼
~
I Á P Á
~
I
where P is the quadrupole coupling matrix:
P ¼
ÀPð1 À ZÞ 0 0
0 ÀPð1 þ ZÞ 0
0 0 2P
¸
¸
Assuming identical principal axes for A and P, the Hamiltonian term would
have the form in the (x
00
,y
00
,z
00
) coordinate system:
~
I Á P Á
~
I ¼ I
x
00 I
y
00 I
z
00
P
11
P
12
P
13
P
12
P
22
P
23
P
13
P
23
P
33
¸
¸
¸
I
x
00
I
y
00
I
z
00
¸
¸
¸
¼P
33
I
2
z
00
À
1
4
P
33
I
þ
I
À
þ I
À
I
þ
ð Þ þ
1
2
P
13
þ iP
23
ð Þ I
À
I
z
þ I
z
I
À
ð Þ
þ
1
2
P
13
À iP
23
ð Þ I
þ
I
z
þ I
z
I
þ
ð Þ À
i
2
P
12
I
2
þ
À I
2
À
þ
1
4
P
11
À P
22
ð Þ I
2
þ
þ I
2
À
Table 7.3 Line positions for [Mn(CO)
2
(PPh
3
)(C
5
H
5
)]
1
. Averaged over m
P
5
m
Mn
Min SP Max
5/2 2850.94 3230.77 3324.94
3/2 2945.04 3262.15 3336.12
1/2 3038.68 3295.16 3356.51
À1/2 3131.28 3379.81 3388.34
À3/2 3222.05 3366.09 3432.99
À5/2 3309.96 3404.00 3489.34
Table 7.4 ESR parameters for [Mn(CO)
2
(PPh
3
)(C
5
H
5
)]
15
x y z
g 2.188 2.034 2.002
A
Mn
ca. 10 32.9 98.4
A
P
29.8 29.8 29.8
Euler angles (1) a ¼ 0 b ¼ 73.91 g ¼ 0
145 Perturbation Theory Calculations
Note that transformation by a general coordinate transformation matrix
leaves the quadrupole matrix symmetrical, i.e., P
ij
¼ P
ji
and with zero trace,
P
11
þ P
22
¼ ÀP
33
.
7.2.1 Perturbation Theory Treatment of Quadrupole Term
Operating on the spin functions, we have:
H
Q
I; m j i ¼
1
2
P
33
3m
2
À I I þ 1 ð Þ
I; m j i
Â
1
2
P
13
þ iP
23
ð Þ 2m À 1 ð Þ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m m À 1 ð Þ
I; m À 1 j i
þ
1
2
P
13
À iP
23
ð Þ 2m þ 1 ð Þ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m m þ 1 ð Þ
I; m þ 1 j i
þ
1
4
P
11
À P
22
À 2iP
12
ð Þ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m þ 1 ð Þ
2
2
À m þ 1 ð Þ
2
I; m þ 2 j i
þ
1
4
P
11
À P
22
þ 2iP
12
ð Þ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m À 1 ð Þ
2
2
À m À 1 ð Þ
2
I; m À 2 j i
which leads to matrix elements:
I; m h jH
Q
I; m j i ¼
1
2
P
33
3m
2
À I I þ 1 ð Þ
I; m À 1 h jH
Q
I; m j i ¼
1
2
P
13
þ iP
23
ð Þ 2m À 1 ð Þ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m m À 1 ð Þ
I; m þ 1 h jH
Q
I; m j i ¼
1
2
P
13
À iP
23
ð Þ 2m þ 1 ð Þ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m m þ 1 ð Þ
I; m þ 2 h jH
Q
I; m j i
¼
1
4
P
11
À P
22
À 2iP
12
ð Þ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m þ 1 ð Þ
2
2
À m þ 1 ð Þ
2
I; m À 2 h jH
Q
I; m j i
¼
1
4
P
11
À P
22
þ 2iP
12
ð Þ
ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
I I þ 1 ð Þ À m À 1 ð Þ
2
2
À m À 1 ð Þ
2
First and secondorder energy corrections then are:
E
ð1Þ
¼
1
2
P
33
3m
2
À I I þ 1 ð Þ
E
ð2Þ
¼ þ
À
4m
K
P
2
13
þ P
2
23
I I þ 1 ð Þ À 2m
2
þ
1
4
À
þ
m
2K
P
11
À P
22
ð Þ
2
þ4P
2
12
I I þ 1 ð Þ À m
2
þ
1
2
where K is the angledependent hyperﬁne coupling deﬁned in the previous
section and the upper sign corresponds to m
s
¼ þ1/2, the lower to À 1/2.
146 Chapter 7
In the electronspin quantization system (x
0
,y
0
,z
0
), the quadrupole interaction
matrix is:
S
11
S
12
S
13
S
21
S
22
S
23
S
31
S
32
S
33
¸
¸
ÀPð1 À ZÞ 0 0
0 ÀPð1 þ ZÞ 0
0 0 2P
¸
¸
S
11
S
21
S
31
S
12
S
22
S
32
S
13
S
23
S
33
¸
¸
Since we can transform from the (x
0
,y
0
,z
0
) coordinate system into (x
00
,y
00
,z
00
) by:
I
x
00
I
y
00
I
z
00
¸
¸
¼
a
1
b
1
c
1
a
2
b
2
c
2
a
3
b
3
c
3
¸
¸
I
x
0
I
y
0
I
z
0
¸
¸
the reverse transformation must be:
I
x
0
I
y
0
I
z
0
¸
¸
¼
a
1
a
2
a
3
b
1
b
2
b
3
c
1
c
2
c
3
¸
¸
I
x
00
I
y
00
I
z
00
¸
¸
We next transform the P matrix, using the a,b,cmatrix and compute P
33
,
F
3
¼ P
13
2
þ P
23
2
, and F
4
¼ (P
11
– P
22
)
2
þ 4P
12
2
, simplifying the expressions
using the relations:
a
1
b
1
þ a
2
b
2
þ a
3
b
3
¼ 0 a
1
c
1
þ a
2
c
2
þ a
3
c
3
¼ 0
b
1
c
1
þ b
2
c
2
þ b
3
c
3
¼ 0 a
2
1
þ a
2
2
þ a
2
3
¼ 1
b
2
1
þ b
2
2
þ b
2
3
¼ 1 c
2
1
þ c
2
2
þ c
2
3
¼ 1
a
1
b
2
À a
2
b
1
¼ c
3
a
1
c
2
À a
2
c
1
¼ Àb
3
b
1
c
2
À b
2
c
1
¼ a
3
with similar expressions relating components of the Smatrices:
P
33
¼3 c
3
S
33
þ b
3
S
23
þ a
3
S
13
ð Þ
2
P À P
þ c
3
S
31
þ b
3
S
21
þ a
3
S
11
ð Þ
2
PZ
À c
3
S
32
þ b
3
S
22
þ a
3
S
12
ð Þ
2
PZ
F
3
¼ 9F
3a
P
2
þ F
3b
P
2
Z þ F
3c
P
2
Z
2
F
3a
¼ c
3
S
33
þ b
3
S
23
þ a
3
S
13
ð Þ
2
À c
3
S
33
þ b
3
S
23
þ a
3
S
13
ð Þ
4
F
3b
¼ 2 c
3
S
33
þ b
3
S
23
þ a
3
S
13
ð Þ
2
Â c
3
S
32
þ b
3
S
22
þ a
3
S
12
ð Þ
2
À c
3
S
31
þ b
3
S
21
þ a
3
S
11
ð Þ
2
147 Perturbation Theory Calculations
F
3c
¼ c
3
S
32
þ b
3
S
22
þ a
3
S
12
ð Þ
2
þ c
3
S
31
þ b
3
S
21
þ a
3
S
11
ð Þ
2
À c
3
S
32
þ b
3
S
22
þ a
3
S
12
ð Þ
2
À c
3
S
31
þ b
3
S
21
þ a
3
S
11
ð Þ
2
2
F
4
¼ 9F
4a
P
2
þ F
4b
PZ þ F
4c
P
2
Z
2
F
4a
¼ 1 À 2 c
3
S
33
þ b
3
S
23
þ a
3
S
13
ð Þ
2
þ c
3
S
33
þ b
3
S
23
þ a
3
S
13
ð Þ
4
F
4b
¼ 2 a
3
S
13
þ b
3
S
23
þ c
3
S
33
ð Þ
2
þ1
F
4c
¼ 4 a
3
S
13
þ b
3
S
23
þ c
3
S
33
ð Þ
2
þ a
3
S
11
þ b
3
S
21
þ c
3
S
31
ð Þ
2
À a
3
S
12
þ b
3
S
22
þ c
3
S
32
ð Þ
2
2
but:
c
3
S
31
þ b
3
S
21
þ a
3
S
11
¼
A
x
S
31
K
c
3
S
32
þ b
3
S
22
þ a
3
S
12
¼
A
y
S
32
K
c
3
S
33
þ b
3
S
23
þ a
3
S
13
¼
A
z
S
33
K
Finally:
F
i
¼ P
2
9F
ia
þ 3ZF
ib
þ Z
2
F
ic
P
33
¼
P
K
2
2A
2
z
S
2
33
À A
2
?
1 À S
2
33
þ Z A
2
x
S
2
31
À A
2
y
S
2
32
F
3
¼
P
2
K
4
9 þ Z
2
A
2
?
A
2
z
S
2
33
1 À S
2
33
þ6ZA
2
z
S
2
33
A
2
y
S
2
32
À A
2
x
S
2
31
þ 4Z
2
A
2
x
S
2
31
A
2
y
S
2
32
F
4
¼
P
2
K
4
9A
4
?
1 À S
2
33
2
þ6Z A
2
x
S
2
31
À A
2
y
S
2
32
K
2
þ A
2
z
S
2
33
þZ
2
A
2
x
S
2
31
À A
2
y
S
2
32
2
þ4Z
2
K
2
A
2
z
S
2
33
Putting together all the bits and pieces of the above perturbation theory
treatment of quadrupole coupling eﬀects, we arrive at the following equations:
B ¼B
0
À
mK
gm
B
À
F
1
m
2
2g
2
m
2
B
B
À
F
2
4g
2
m
2
B
B
I I þ 1 ð Þ À m
2
þ
8mF
3
gm
B
K
I I þ 1 ð Þ À 2m
2
þ
1
4
À
mF
4
gm
B
K
I I þ 1 ð Þ À m
2
þ
1
2
148 Chapter 7
where:
K
2
¼
1
g
2
A
2
zz
g
2
z
cos
2
y þ A
2
xx
g
2
x
cos
2
j þ A
2
yy
g
2
y
sin
2
j
sin
2
y
F
1
ðxÞ ¼
g
2
x
A
4
x
À K
4
g
2
K
2
; F
1
ðyÞ ¼
g
2
y
A
4
y
À K
4
g
2
K
2
; F
1
ðzÞ ¼
g
2
z
A
4
z
À K
4
g
2
K
2
F
2
ðxÞ ¼
g
2
x
A
4
x
A
2
y
þ A
2
z
g
2
K
2
F
2
ðyÞ ¼
g
2
y
A
4
y
A
2
x
þ A
2
z
g
2
K
2
F
2
ðzÞ ¼
g
2
z
A
4
z
A
2
x
þ A
2
y
g
2
K
2
F
3
ðxÞ ¼ F
3
ðyÞ ¼ P
2
Z
2
1 À cos
2
2a
; F
3
ðzÞ ¼ 0
F
4
ðxÞ ¼ P
2
9 þ 6Z cos 2a þ Z
2
cos
2
2a
F
4
ðyÞ ¼ P
2
9 À 6Z cos 2a þ Z
2
cos
2
2a
F
4
ðzÞ ¼ 4P
2
Z
2
7.2.2 Example Application of Analysis of Quadrupole Eﬀects
The ESR spectrum of a RhIr complex
7,8
is shown in Figure 7.2, together with a
very satisfactory simulation.
Since
103
Rh has I ¼ 1/2 and both isotopes of Ir have I ¼ 3/2, we might have
expected each gcomponent to be a quartet of doublets. The Rh coupling is
barely visible on the highﬁeld features, but there is no trace on the other two
gcomponents; indeed, the lowﬁeld and midﬁeld components are more like
1:2:1 triplets than the expected 1:1:1:1 quartet.
We must ﬁrst examine the predicted spacing of the expected quartet. The
outer and inner spacings are given by:
DB
outer
¼
3K
gm
B
þ
24F
3
gm
B
K
þ
3F
4
gm
B
K
DB
inner
¼
K
gm
B
À
24F
3
gm
B
K
þ
3F
4
gm
B
K
Assuming that the quadrupole matrix axes diﬀer from the gmatrix axes by
a ¼ 451, F
3
and F
4
simplify to:
F
3
ðxÞ ¼F
3
ðyÞ ¼ P
2
Z
2
; F
3
ðzÞ ¼ 0
F
4
ðxÞ ¼F
4
ðyÞ ¼ 9P
2
; F
4
ðzÞ ¼ 4p
2
Z
2
Under these circumstances, the inner and outer quartet spacings are:
DB
outer
ðx; yÞ ¼
3K
gm
B
þ
24P
2
Z
2
gm
B
K
þ
27P
2
gm
B
K
DB
inner
ðx; yÞ ¼
K
gm
B
À
24P
2
Z
2
gm
B
K
þ
27P
2
gm
B
K
149 Perturbation Theory Calculations
DB
outer
ðzÞ ¼
3K
gm
B
þ
12P
2
Z
2
gm
B
K
DB
inner
ðzÞ ¼
K
gm
B
þ
12P
2
Z
2
gm
B
K
Examination of these equations shows that the spacing of the zcomponent is
expected to be a little bigger for the inner (1/2, À1/2) spacing than 1/3 the outer
spacing. In contrast, the inner spacing of the x and y components is expected to be
less than 1/3 the outer spacing. This is entirely consistent with the experimental
spectrum of Figure 7.2. Unfortunately, the eﬀect is too big for perturbation theory
to be entirely reliable, and it was necessary to perform a matrix diagonalization to
get the exact solution of the spin Hamiltonian. For further details, see ref. 8.
One more item needs to be discussed in reference to this spectrum. Why did it
turn out that a ¼ 451? Consider the coordination geometry of the Ir atom (1).
Ir
N N
Ph
3
P
y
x
1
CO
2800 2900 3000 3200 3300 3400 3500
Simulation
Exptl. spectrum
Magnetic Field/Gauss
Figure 7.2 ESR spectrum of [RhLL
0
{m(tol)NNN(tol)}
2
IrLL
0
]
1
PF
6
À
(tol ¼pMeC
6
H
4
,
L ¼ CO, L
0
¼ PPh
3
).
(Reproduced with permission from ref. 7, copyright (2000) Royal Society of
Chemistry. Simulation using methods of ref. 8.)
150 Chapter 7
The Rh–Ir vector deﬁnes the zaxis, and the NN bridges deﬁne the x and
yaxes. The N and P atoms are solely electron donors, but the CO ligand
involves a lot of pbackbonding. Thus the quadrupole matrix, which has large
components if there is an electric ﬁeld gradient at the Ir nucleus, is rotated 451
compared with the gmatrix principal axes.
References
1. J.A. DeGray, PhD Thesis, Brown University, 1989.
2. H. Watanabe, Operator Methods in Ligand Field Theory, PrenticeHall,
Englewood Cliﬀs, 1966, p. 147.
3. E.B. Wilson, J.C. Decius and P.C. Cross, Molecular Vibrations, McGraw
Hill, New York, 1955, p. 286.
4. S.M. Blinder, J. Chem. Phys., 1960, 33, 748.
5. R.D. Pike, A.L. Rieger and P.H. Rieger, J. Chem. Soc., Faraday Trans. I,
1989, 85, 3913.
6. (a) P.H. Rieger, Specialist Periodical Reports, Electron Spin Resonance,
Royal Society of Chemistry, Cambridge, 1993, vol. 13b, ch. 4, p. 178;
(b) J.A. DeGray and P.H. Rieger, Bull. Mag. Reson., 1987, 8, 95.
7. N.G. Connelly, O.D. Hayward, P. Klangsinsirikul, A.G. Orpen and
P.H. Rieger, Chem. Commun., 2000, 963.
8. N.G. Connelly, D.J.H. Emslie, P. Klangsinsirikul and P.H. Rieger, J. Phys.
Chem. A, 2002, 106, 12214.
151 Perturbation Theory Calculations
APPENDIX 1
Physical Constants, Conversion
Factors, and Properties of
Nuclei (Tables A1.1–A1.4)
Table A1. 1 Physical constants
Name Symbol Value
Planck constant h 6.62607 Â 10
À34
J s
1
Speed of light c 2.99792 Â 10
8
m s
À1
Avogadro’s number N
A
6.02214 Â 10
23
mol
À1
Boltzmann constant k 1.38065 Â 10
À23
J K
À1
Bohr magneton m
B
9.27401 Â 10
À28
J G
À1
Nuclear magneton m
N
5.05078 Â 10
À31
J G
À1
Free electron gvalue g
e
2.002319304
Gas constant R 8.31447 J mol
À1
K
À1
Electronic charge e À1.60218 Â 10
À19
C
Electron mass m
e
5.486 Â 10
À4
amu
Table A1. 2 Conversion factors
1 amu 1.66 Â 10
À24
g
1 Tesla 10 000 Gauss
10
À4
cm
À1
1.9864 Â 10
À27
J
A/MHz 2.8025 (g/g
e
) a/G
A/10
À4
cm
À1
0.93481 (g/g
e
) a/G
152
Table A1. 3 Some properties of stable magnetic nuclei
Nucleus I % Abund.
a
10
4
A
s
(cm
À1
)
b
10
4
P (cm
À1
)
b
m (nm)
a
10
28
Q (m
2
)
a
1
H 1/2 99.985 473.8 2.7928 0
2
H 1 0.012 0.8574 0.00286
6
Li 1 7.5 30.7 0.8220 À0.0008
7
Li 3/2 92.4 121.7 3.2564 À0.041
9
Be 3/2 100.0 À150.6 À1.1776 0.0529
10
B 3 19.9 569 35.1 1.8006 0.085
11
B 3/2 80.1 850 52.4 2.6886 0.0406
13
C 1/2 1.11 1260 89.6 0.7024 0
14
N 1 99.63 604 46.3 0.40376 0.0200
15
N 1/2 0.37 À424 À32.5 À0.28319 0
17
O 5/2 0.04 À1755 À140.4 À1.8938 À0.026
19
F 1/2 100.0 2066 109.4 2.6289 0
23
Na 3/2 100.0 309.2 2.2175 0.104
25
Mg 5/2 10.0 À162.1 À0.8554 0.199
27
Al 5/2 100.0 1304 69.28 3.6415 0.1540
29
Si 1/2 4.683 À1532 À95.23 À0.5553 0
31
P 1/2 100.0 4438 305.9 1.1316 0
33
S 3/2 0.76 1155 83.82 0.6438 0.068
35
Cl 3/2 75.78 1909 146.4 0.82187 À0.083
37
Cl 3/2 24.22 1589 121.9 0.68412 À0.014
39
K 3/2 93.258 76.2 0.39146 0.049
41
K 3/2 7.730 41.8 0.21487 0.060
43
Ca 7/2 0.135 À213.7 À1.3173 À0.05
45
Sc 7/2 100.0 941.6 80.2 4.75649 À0.220
47
Ti 5/2 7.44 À260.8 À24.56 À0.78848 0.30
49
Ti 7/2 5.41 À365.2 À34.40 À1.10417 0.24
50
V 6 0.25 1120 117.7 3.3457
51
V 7/2 99.75 1389 146.0 5.148706 À0.04
53
Cr 3/2 9.50 À249.6 À34.4 0.47464 À0.15
55
Mn 5/2 100.0 1680 207.5 3.4687 0.32
57
Fe 1/2 2.12 249.2 32.61 0.0906 0
59
Co 7/2 100.0 1984 282.0 4.63 0.41
61
Ni 3/2 1.140 À834 À125.2 À0.75002 0.16
63
Cu 3/2 69.17 2000 399.0 2.2233 À0.211
65
Cu 3/2 30.83 2142 427.0 2.3817 À0.195
67
Zn 5/2 4.1 696 117.3 0.8755 0.15
69
Ga 3/2 60.11 4073 170.0 2.01639 0.17
71
Ga 3/2 39.89 5176 216.0 2.56227 0.11
73
Ge 9/2 7.73 À788 À40.1 À0.87947 À0.17
75
As 3/2 100.0 4890 278.2 1.43947 0.31
77
Se 1/2 7.63 6711 410.0 0.53506 0
79
Br 3/2 50.69 10697 682.0 2.1064 0.331
81
Br 3/2 49.31 11529 735.0 2.2703 0.276
85
Rb 5/2 72.17 346 1.352 0.23
87
Rb 3/2 27.83 704 2.7512 0.15
87
Sr 9/2 7.00 À584.7 1.0936 0.34
89
Y 1/2 100.0 À417 À20.76 À0.13742 0
91
Zr 5/2 11.22 À91.8 À51.9 À1.30362 À0.21
93
Nb 9/2 100.0 2198 152.5 6.1705 À0.32
95
Mo 5/2 15.92 À662 À50.3 À0.9142 À0.02
97
Mo 5/2 9.55 À676 À51.3 À0.9335 0.26
99
Ru 5/2 12.76 À525 À47.5 À0.6413 0.079
101
Ru 5/2 17.06 À588 À53.3 À0.7188 0.46
103
Rh 1/2 100.0 À410 À40.4 À0.0884 0
105
Pd 5/2 22.33 À62.7 À0.642 0.66
153 Physical Constants, Conversion Factors, and Properties of Nuclei
Table A1. 3 (continued )
Nucleus I % Abund.
a
10
4
A
s
(cm
À1
)
b
10
4
P (cm
À1
)
b
m (nm)
a
10
28
Q (m
2
)
a
107
Ag 1/2 51.84 À611 À68.3 À0.11357 0
109
Ag 1/2 48.16 À703 À78.6 À0.13069 0
111
Cd 1/2 12.80 À4553 À430.0 À0.59489 0
113
Cd 1/2 12.22 À4763 À450.0 À0.62230 0
113
In 9/2 4.29 6731 237.1 5.529 0.80
115
In 9/2 95.71 6746 236.6 5.541 0.81
117
Sn 1/2 7.68 À14002 À584.0 À1.0010 0
119
Sn 1/2 8.59 À14650 À611.0 À1.0473 0
121
Sb 5/2 57.21 11708 524.0 3.363 À0.4
123
Sb 7/2 42.79 8878 397.0 2.550 À0.5
125
Te 1/2 7.07 À18542 À875.0 À0.8885 0
127
I 5/2 100.0 13876 677.0 2.8133 À0.79
133
Cs 7/2 100.0 823 2.582 À0.004
135
Ba 3/2 6.59 1220 0.838 0.16
137
Ba 3/2 11.23 1324 0.9374 0.245
139
La 7/2 99.91 2004 79.1 2.7830 0.20
141
Pr 5/2 100.0 4166 295.1 4.275 À0.08
143
Nd 7/2 12.18 À793 À60.0 À0.67 À0.60
145
Nd 7/2 8.3 À481 À59.0 À0.66 À0.31
147
Sm 7/2 14.99 À672 À55.8 À0.815 À0.26
149
Sm 7/2 13.82 À554 À46.0 À0.672 0.075
151
Eu 5/2 47.81 1086 91.6 À0.872 0.90
153
Eu 5/2 52.19 1909 161.0 1.533 2.41
155
Gd 3/2 14.80 À647 À21.9 À2.59 1.30
157
Gd 3/2 15.65 À849 À28.78 À3.40 1.36
159
Tb 3/2 100.0 4546 417.0 2.014 1.43
161
Dy 5/2 18.9 À705 À65.1 À0.480 2.51
163
Dy 5/2 24.9 988 91.3 0.673 2.65
165
Ho 7/2 100.0 4523 429.0 4.17 3.49
167
Er 7/2 22.9 À645 À63.2 À0.5639 3.57
169
Tm 1/2 100.0 À1946 À195.3 À0.232 0
171
Yb 1/2 14.8 889 89.0 0.4937 0
173
Yb 5/2 16.1 À1224 À122.6 À0.6799 2.80
175
Lu 7/2 97.41 3546 93.2 2.2327 3.49
177
Hf 7/2 18.60 1471 41.4 0.7935 0.337
179
Hf 9/2 13.62 À1188 À33.4 À0.641 3.79
181
Ta 7/2 99.99 5010 148.6 2.370 3.3
183
W 1/2 14.31 1927 60.9 0.11778 0
185
Re 5/2 37.40 11718 382.0 3.1871 2.18
187
Re 5/2 62.60 11838 386.0 3.2197 2.07
187
Os 1/2 1.96 431.4 14.74 0.064652 0
189
Os 3/2 16.15 4403 150.4 0.65993 0.86
191
Ir 3/2 37.3 1072 37.8 0.151 0.82
193
Ir 3/2 62.7 1165 41.0 0.164 0.75
195
Pt 1/2 33.83 11478 492.0 0.6095 0
197
Au 3/2 100.0 959 44.0 0.14575 0.55
199
Hg 1/2 16.87 13969 537.4 0.50588 0
201
Hg 3/2 13.18 À15470 À595 À0.56023 0.39
203
Tl 1/2 29.52 60711 710 1.62226 0
205
Tl 1/2 70.48 61308 717.0 1.63822 0
207
Pb 1/2 22.1 27188 542.0 0.59258 0
209
Bi 9/2 100.0 25860 553.0 4.111 À0.37
a
Ref. 1.
b
Ref. 2.
154 Appendix 1
Dipolar hyperﬁne coupling parameters for the transition metals and (n þ 1)p
orbitals were computed from SCF Hartree–Fock–Slater atomic orbitals [3].
The parameter P is given by eqn (5) below, and hr
À3
i
nd
and hr
À3
i
(n11)p
are given
by eqns (1) and (3). The sorbital contribution, responsible for the isotropic
coupling, is given by eqn (3); Dn
d
is the diﬀerence between the number of
delectrons present in the ion of interest and the number present in the neutral
metal. Using these equations and the parameters given in Table A1.3, the P and
isotropic coupling parameters can be computed for each of the transition
metals (Table A1.4).
r
À3
nd
¼ A þ BDn
d
þ Cn
s
þ Dn
p
ð1Þ
8p
3
c
2
ð0Þ
Â Ã
ðnþ1Þs
¼ A þ BDn
d
þ Cn
s
þ Dn
p
ð2Þ
r
À3
ðnþ1Þp
¼ A þ BDn
d
þ Cn
s
ð3Þ
F ¼
g
e
g
N
m
0
m
B
m
N
4ph
6:74834 Â 10
30
au
3
m
À3
À Á
10
À6
MHz s
À1
À Á
ð4Þ
P ¼ g
e
g
N
m
B
m
N
r
À3
ð5Þ
P ¼ 4:180hr
À3
i Â 10
À4
cm
À1
ð5aÞ
Example
For Cr(II), 3d
0.9
4s
0.1
:
r
À3
3d
¼ 3:99 À 0:517 ðÀ2:1Þ À 0:062 ð0:1Þ À 0:017 ð0:0Þ ¼ 5:07
c
2
ð0Þ
Â Ã
4s
¼ 47:8 À 16:0 ðÀ2:1Þ À 6:9 ð0:1Þ À 3:2 ð0:0Þ ¼ 80:7
r
À3
4p
¼ 2:69 À 1:70 ðÀ2:1Þ À 0:97 ð0:1Þ ¼ 6:16
Using eqn (5a):
P
3d
¼ 21:2 Â 10
À4
cm
À1
P
4p
¼ 25:7 Â 10
À4
cm
À1
155 Physical Constants, Conversion Factors, and Properties of Nuclei
Table A1. 4 EPR hyperﬁne coupling parameters for the transition metals
Isotope I F
a
A
b
ÀB
b
ÀC
b
ÀD
b
47
Ti 5/2 À30.13 hr
À3
i
3d
2.554 0.444 0.057 0.009
49
Ti 7/2 À30.14 [c
2
(0)]
4s
37.1 12.8 5.5(4) 2.7
hr
À3
i
4p
2.15 1.37 0.76 À
50
V 6 53.26 hr
À3
i
3d
3.229 0.479 0.060 0.013
51
V 7/2 140.52 [c
2
(0)]
4s
42.6 14.3 6.1 2.9
hr
À3
i
4p
2.42 1.53 0.86 À
53
Cr 3/2 À30.22 hr
À3
i
3d
3.990 0.517 0.062 0.017
[c
2
(0)]
4s
47.8 16.0 6.9 3.2
hr
À3
i
4p
2.69 1.70 0.97 À
55
Mn 5/2 132.53 hr
À3
i
3d
4.841 0.559 0.064 0.020
[c
2
(0)]
4s
53.6 17.8 7.6 3.4
hr
À3
i
4p
2.96 1.88 1.08 À
57
Fe 1/2 17.31 hr
À3
i
3d
5.789 0.604 0.066 0.021
[c
2
(0)]
4s
59.8 9.7 8.5 3.7
hr
À3
i
4p
3.23 2.07 1.19 À
59
Co 7/2 126.4 hr
À3
i
3d
6.840 0.650 0.067 0.024
[c
2
(0)]
4s
66.2 21.7 9.4 4.0
hr
À3
i
4p
3.50 2.26 1.32 À
61
Ni 3/2 À47.76 hr
À3
i
3d
7.997 0.700 0.070 0.024
[c
2
(0)]
4s
73.1 23.7 0.3 4.3
hr
À3
i
4p
3.77 2.45 1.44 À
63
Cu 3/2 141.58 hr
À3
i
3d
9.270 0.750 0.072 0.025
65
Cu 3/2 151.67 [c
2
(0)]
4s
80.3 25.8 11.3 4.6
hr
À3
i
4p
4.05 2.65 1.57 À
91
Zr 5/2 À49.81 hr
À3
i
4d
3.407 0.542 0.144 0.036
[c
2
(0)]
5s
75.7 21.3 10.3 4.8
hr
À3
i
5p
3.70 2.03 1.20 À
93
Nb 9/2 130.98 hr
À3
i
4d
4.197 0.565 0.142 0.045
[c
2
(0)]
5s
86.1 24.2 11.5 5.4
hr
À3
i
5p
4.14 2.27 1.35 À
95
Mo 5/2 À34.93 hr
À3
i
4d
5.04 0.591 0.140 0.050
97
Mo 5/2 À35.67 [c
2
(0)]
5s
96.0 27.0 12.5 5.6
hr
À3
i
5p
4.55 2.51 1.49 À
99
Tc 9/2 120.67 hr
À3
i
4d
5.951 0.619 0.137 0.053
[c
2
(0)]
5s
106.9 À30.4 14.2 6.5
hr
À3
i
5p
4.93 2.76 1.64 À
99
Ru 5/2 À24.50 hr
À3
i
4d
6.929 0.648 0.134 0.054
101
Ru 5/2 À27.46 [c
2
(0)]
5s
117.6 33.7 15.6 7.0
hr
À3
i
5p
5.29 3.00 1.78 À
103
Rh 1/2 À16.89 hr
À3
i
4d
7.980 0.678 0.132 0.055
[c
2
(0)]
5s
128.5 37.2 17.0 7.6
hr
À3
i
5p
5.64 3.24 1.93 À
105
Pd 5/2 À24.53 hr
À3
i
4d
9.107 À0.710 À0.129 À0.055
[c
2
(0)]
5s
139.7 À40.9 À18.6 À8.1
hr
À3
i
5p
5.98 À3.49 À2.08 À
107
Ag 1/2 À21.70 hr
À3
i
4d
10.31 À0.744 À0.127 À0.054
109
Ag 1/2 À24.95 [c
2
(0)]
5s
151.2 À44.8 À20.2 À8.7
hr
À3
i
5p
6.30 À3.73 À2.22 À
177
Hf 7/2 21.66 hr
À3
i
5d
6.34 À0.99 À0.30 À0.07
179
Hf 9/2 À13.60 [c
2
(0)]
6s
274 À73 À36 À16
hr
À3
i
6p
6.80 À3.61 À2.21 À
156 Appendix 1
References
1. Handbook of Chemistry and Physics, ed. D. R. Lide, 83rd edn, CRC Press,
Boca Raton, FL, 2002.
2. J.R. Morton and K.F. Preston, J. Magn. Reson., 1978, 30, 577.
3. P.H. Rieger, J. Magn. Reson., 1997, 124, 140.
Table A1. 4 (continued )
Isotope I F
a
A
b
ÀB
b
ÀC
b
ÀD
b
181
Ta 7/2 64.68 hr
À3
i
5d
7.51 À0.99 À0.29 À0.09
[c
2
(0)]
6s
310 À81 À40 À18
hr
À3
i
6p
7.51 À3.96 À2.42 À
183
W 1/2 22.50 hr
À3
i
5d
8.72 À1.00 À0.28 À0.10
[c
2
(0)]
6s
345 À90 À44 À20
hr
À3
i
6p
8.16 À4.30 À2.63 À
185
Re 5/2 121.77 hr
À3
i
5d
9.98 À1.02 À0.27 À0.11
187
Re 5/2 123.02 [c
2
(0)]
6s
387 À103 À52 À25
hr
À3
i
6p
8.77 À4.65 À2.84 À
187
Os 1/2 12.34 hr
À3
i
5d
11.30 À1.03 À0.26 À0.11
189
Os 3/2 42.02 [c
2
(0)]
6s
417 À110 À53 À24
hr
À3
i
6p
9.33 À4.98 À3.05 À
191
Ir 3/2 9.62 hr
À3
i
5d
12.68 À1.05 À0.26 À0.11
193
Ir 3/2 10.44 [c
2
(0)]
6s
454 À120 À58 À26
hr
À3
i
6p
9.85 À5.32 À3.26 À
195
Pt 1/2 116.44 hr
À3
i
5d
14.12 À1.08 À0.25 À0.11
[c
2
(0)]
6s
491 À131 À63 À27
hr
À3
i
6p
10.34 À5.64 À3.45 À
197
Au 3/2 9.285 hr
À3
i
5d
15.64 À1.10 À0.24 À0.10
[c
2
(0)]
6s
530 À143 À68 À29
hr
À3
i
6p
10.80 À5.97 À3.66 À
a
Factor deﬁned by eqn (4).
b
Parameters A, B, C, D refer to eqns (1)–(3).
157 Physical Constants, Conversion Factors, and Properties of Nuclei
APPENDIX 2
Advanced ESR Methods
For most of the 60 years since ESR was discovered, the vast majority of
experiments have been carried out using CW spectrometers operating at the
Xband frequency and employing conventional electromagnets. Indeed, most
commercially available instruments are still of the general type described in
Chapter 1, albeit with enhancements made possible by advancements in elec
tronics, computers and, to some extent, microwave technology. The last two
decades, however, have seen ever increasing development and applications of
other ESR techniques. Most of these advanced methods fall into one of three
broad categories:
High ﬁeld/multifrequency ESR
Double resonance
Pulsed methods
Although several of these methods were ﬁrst applied in the 1950s and 1960s,
none would have been possible at their current technical level using the post
World War II technology that gave rise to Xband ESR. In particular:
High frequency ESR has required not only the use of superconducting magnets,
developed primarily for NMR, but even more importantly employs stable
frequency sources and sensitive detectors in the millimeter and submillimeter
range that have only recently become available.
Double resonance techniques, on the other hand, of which the earliest was
ENDOR (described in Chapter 1), have greatly beneﬁted from advances in
signal processing technology, of the sort now employed, for example, in
wireless communication systems.
Pulsed methods in ESR, which have by now taken over NMR instrumentation,
have required the development of high power ampliﬁers and fast switches for
microwave and higher frequency radiation.
One group of reviewers
1
has actually attributed much of the rapid develop
ment in all three categories to declassiﬁcation of research in high frequency and
communications technology that was an outgrowth of the end of the Cold War!
The ready availability of this technology is also reﬂected in the fact that at least
one supplier of commercial ESR equipment now oﬀers a high ﬁeld, pulsed,
158
double resonance spectrometer, i.e., an instrument that incorporates advanced
technology in all three categories.
The impact of these new methods has been felt particularly strongly in two
areas:
(i) Improvements in eﬀective resolution using high ﬁelds, double resonance,
and pulsed methods have made possible the analysis and interpretation
of spectra from increasingly complex paramagnetic materials, especially
those of biological and solid state relevance.
(ii) Nanosecond pulsed methods have made possible the detection and
characterization of various transient paramagnetic intermediates that
could be previously studied by optical methods but were inaccessible to
ESR. The latter application, which has led to advances in the under
standing of reactive intermediates in photochemistry, radiation chemis
try and other areas of mechanistic chemistry, is beyond the scope of this
book. Reviews of these and other recent chemical applications of ESR
have been written by Christopher Rhodes.
6,13
Presented below are brief descriptions of some of the applications to structure
analysis to which each of the three advanced methods are making important
contributions. For the reader who wishes to learn more about these
methods and applications a list of recent reviews and other leading references
to these applications is also included. The titles of the papers that we have
referenced will make it clear that we have tried to include applications of
relevance not only to organometallic chemistry but also to biochemistry and
related ﬁelds.
A2.1 High Frequency ESR
The two most obvious advantages of high frequency/high ﬁeld spectrometers,
better sensitivity and improved resolution of ﬁelddependent features, which
has revolutionized the ﬁeld of highresolution NMR, have proved to have
less impact on ESR. For example, the smaller sample sizes that must be used
when higher frequency, smaller wavelength, radiation is used have made it
necessary in many cases to work with smaller numbers of spins and therefore
not fully exploit the sensitivity advantage inherent in use of the higher
frequency. Improved resolution with increasing frequency is also not a forgone
conclusion, especially with randomly oriented samples. While it is certainly
possible, for example, to detect smaller gfactor anisotropies at high ﬁelds,
the ﬁrst indication of such eﬀects often appears as poorer resolution of
individual features because of increased apparent line width as the anisotropy
is revealed.
Nevertheless many recent studies simply could not have been carried out
without the use of high frequency EPR. One of these is described below. In this,
159 Advanced ESR Methods
and most other studies to date, custombuilt equipment was necessary because
no commercial instruments are available yet at very high frequencies.
A recent example of an application of high ﬁeld ESR (‘‘HFEPR’’) is shown
in Figure A2.1. The sample studied is a frozen aqueous solution of Cr
21
sulfate
(3d
4
, S ¼ 2), cooled to 10 K.
This corresponds to an ‘‘EPRsilent’’ sample that gives no detectable ESR
spectrum at Xband frequencies because it possesses a zeroﬁeld splitting larger
than the Zeeman interaction (see Chapter 6), and the energy spacing between
the two lowest levels is too large to be spanned by a microwave quantum at
Xband. Nevertheless, higher frequencies are able to induce transitions. Since
Figure A2.1 Microwave frequency dependent HFEPR spectra of aqueous Cr
21
(0.1–
0.2 M), sulfate counterion. Experimental conditions: temperature 10 K;
microwave frequency as indicated. In the spectrum taken at 329 GHz a
sharp signal from aqueous Cr
31
impurity at g ¼ 2 is indicated and the
resonances due to Cr
21
are labeled (Figure A2.2).
(Reprinted with permission from ref. 25, copyright (1998) American
Chemical Society.)
160 Appendix 2
the energy level scheme is quite complex, as shown in Figure A2.1 the spectrum
varies markedly with frequency. Figure A2.2 shows the frequency dependence
of the various spectral features.
Analysis of the spectra at diﬀerent frequencies yielded the parameters
D¼À2.20(5) cm
À1
, E¼0.0(1) cm
À1
, and a nearly isotropic gfactor,
g ¼ 1.98(2), none of which could have been determined at Xband. Analysis
was aided by the observation of diﬀerent slopes of the B vs. n plots for Dm
S
41
and Dm
S
¼ 1 transitions. A review of advanced methods, including highﬁeld
EPR, is given in ref. 11. Various recent applications of high ﬁeld and multi
frequency EPR are described in refs 19–31.
A2.2 Double Resonance
In Chapter 2, ENDOR (electron–nuclear double resonance) was brieﬂy
described. To perform an ENDOR experiment it is necessary to apply both a
radiofrequency and a microwave frequency, eﬀectively performing simultane
ous NMR and ESR, respectively, on the sample. The experiment is performed
at a ﬁxed magnetic ﬁeld, with the ESR saturating frequency centered on a
Figure A2.2 Plot of resonance ﬁeld versus microwave frequency for data from
Figure A2.1. Least squares ﬁts are given for each Cr
21
resonance (dotted
line) and for the Cr
31
impurity (solid line). C and E correspond to
Dm
S
4 1 while the rest correspond to Dm
S
¼ 1.
(Reprinted with permission from ref. 25, copyright (1998) Amer
ican Chemical society.)
161 Advanced ESR Methods
spectral feature usually identiﬁed from a CWESR experiment. The RF ﬁeld
at the NMR frequency is then varied. When the RF frequency matches the
appropriate NMR transition frequency the ESR absorption reappears. The
number of observed lines is reduced, making it easier to detect small hyperﬁne
splittings, even though the number of nuclei giving rise to the coupling is not
revealed. Furthermore, one immediately knows from the NMR frequency what
type of nucleus is involved. We saw, for example, in Chapter 2 that hyperﬁne
patterns from
14
N and
1
H splittings can be quite similar and proper analysis
of the spectrum requires recognition of subtle diﬀerences in peak intensities.
This in turn demands resolution that is almost never available in randomly
oriented samples. In the case of such samples, ENDOR is playing an increas
ingly important role since careful selection of the saturated features in the ESR
spectrum leads to excitation of molecules over only a narrow range of orientat
ions. This makes it possible to extract anisotropic spectral parameters that must
otherwise be obtained using single crystals or laborious simulation of the
full spectrum.
32–34,39,42,43
The relatively old technique of Dynamic Nuclear
Polarization (DNP), originally proposed by Overhauser, in which the NMR
spectrum is observed during irradiation of ESR transitions, is being rejuve
nated as a method of enhancing the NMR spectra of solids.
35
As discussed in Chapter 6, in systems with more than one unpaired electron
the ESR spectrum contains features that involve electron–electron coupling
parameters analogous to the nuclear hyperﬁne parameters. In those types of
samples the advantages of double resonance are carried out by employing the
use of two diﬀerent microwave frequencies, one ﬁxed and saturating, and one
variable frequency that searches for transitions. This technique is known as
ELDOR (electron–electron double resonance).
38,40,41,44
It has been used much
less than ENDOR and usually requires custombuilt equipment.
A major limitation of CW double resonance methods is the sensitivity of the
intensities of the transitions to the relative rates of spin relaxation processes.
For that reason the peak intensities often convey little quantitative information
about the numbers of spins involved and, in extreme cases, may be undetect
able. This limitation can be especially severe for liquid samples where several
relaxation pathways may have about the same rates. The situation is somewhat
better in solids, especially at low temperatures, where some pathways are
eﬀectively frozen out. Fortunately, fewer limitations occur when pulsed radio
and microwave ﬁelds are employed. In that case one can better adapt the
excitation and detection timing to the rates of relaxation that are intrinsic to
the sample.
50
There are now several versions of pulsed ENDOR and other
double resonance methods. Some of these methods also make it possible to
separate in the time domain overlapping transitions that have diﬀerent relax
ation behavior, thereby improving the resolution of the spectrum.
A2.3 Pulsed Methods
The recent advent of the ability to apply short and very intense microwave
pulses to samples and detect the fast response to the excitation has made it
162 Appendix 2
possible to collect an ESR freeinduction decay in the time domain and to
Fourier transform the result and view the data in the frequency domain. Such
experiments are an outgrowth of the pulsed methods commonly used to
determine relaxation times. They mimic the ubiquitous pulsed methods that
have revolutionized NMR. It has taken the ability to do this much longer to
develop for ESR than for NMR, however, due both to the much shorter ESR
relaxation times and to the much wider extent of ESR spectra in the frequency
domain, both of which require shorter, more intense pulses and much faster
detector response
8,16,18
than for pulsed NMR. Applications of Fourier
transform ESR, and related multiplepulse twodimensional techniques
now common in NMR, are still in their infancy and require custombuilt
equipment.
47,48,54,55
Nevertheless, there is one type of pulsed ESR measurement, the phenomenon
of Electron Spin Echo Envelope Modulation (ESEEM) that is growing very
rapidly and is in some instances replacing ENDOR as a way of extracting
parameters from randomly oriented samples, since it can also be used to study
weakly coupled nuclei, i.e., cases where A o 10 MHz. Interestingly, the most
robust version of the method works best in randomly oriented solids with
overlapping lines. In its simplest form, the phenomenon appears as modulation
of the intensity of the ‘‘echo’’ detected after a sequence of two pulses separated
by a time t, on the order of the reciprocal of a hyperﬁne coupling in frequency
units, typically micro to nanoseconds. This ‘‘envelope’’ has in it two sorts of
information, an approximately exponential decay corresponding to the phase
memory time, T
2
, as well as information about the amplitude modulation by
nuclear spin ﬂips under the inﬂuence of A. For an ESEEM experiment to be
successful the applied microwave pulse must be such that both allowed and
semiforbidden, i.e., double quantum, transitions are excited. There must also
be enough nuclear modulations within the time before the echo decays to deﬁne
the observable values of A. ESEEM experiments are usually run on frozen
samples at low temperatures both to lengthen decay times and to achieve the
anisotropic eﬀects that make the desired transitions possible. It is also necessary
for the deadtime, the time needed between pulses before the detector can record
new information, to be short compared with the time over which the modu
lations are observable. The echo intensity as a function of t may then be
Fourier transformed to yield a frequency domain ‘‘spectrum’’ that looks a lot
like ENDOR. ESEEM experiments can also be performed using more than two
pulses, giving somewhat greater ﬂexibility in the range of relaxation times over
which modulations can be observed. As with ENDOR, however, ESEEM
spectra can be quite sensitive to relaxation behavior so that failure to observe a
modulation does not prove the absence of a coupled nucleus. Several recent
reviews of the technique are available.
2,9,12,14
ENDOR and ESEEM are also
sometimes employed together.
46,49
The combination of higher ﬁelds and pulsed, double resonance methods is
now making it possible to use ESR as a tool to determine distances within
macromolecules. This is a valuable supplement to the very widespread use of
multidimensional NMR in structural biology.
33
163 Advanced ESR Methods
References
In developing this short bibliography of advanced ESR methods we have tried
to select references primarily from widely available journals, review series and
books published after 1990. We have therefore avoided references to meeting
abstracts or highly specialized journals with low circulation or earlier publica
tions where the methods were ﬁrst demonstrated. Citations to these pioneering
papers are found in the references.
For ease of recognition of papers relevant to a particular method or
application we have included the title of the review or paper within the citation.
The references are grouped by topic and listed in reverse chronological order
within each topic. They are numbered consecutively.
Books and Reviews
1. Highﬁeld EPR, Magn. Reson. Chem., Special Issue, ed. W. Lubitz,
K. Mobius and K.P. Dinse, 2005, 43, S1–S266.
2. G.R. Eaton and S.S. Eaton, Electronnuclear double resonance spectro
scopy and electron spin echo envelope modulation spectroscopy, Compre
hensive Coordination Chemistry II, Elsevier, Boston, 2004, 49.
3. C. Coulon and R. Clerac, Electron spin resonance: A major probe of
molecular conductors, Chem. Rev., 2004, 104, 5655.
4. Biomedical EPR, ed. S.S. Eaton, G.R. Eaton and L.J. Berliner, Kluwer
Academic/Plenum Publishers, New York, 2004.
5. L.J. Berliner, ed., In vivo EPR (ESR): Theory and Applications, Kluwer
Academic/Plenum Publishers, New York, 2004.
6. C.J. Rhodes, Electron spin resonance (some applications for the biolog
ical and environmental sciences), Annu. Rep. Prog. Chem., Sec. C, 2004
100, 149.
7. Paramagnetic Resonance of Metallobiomolecules, ed. J. Telser, ACS
Symposium Series, 858, American Chemical Society, 2003.
8. T. Prisner, M. Rohrer and F. MacMillan, Pulsed EPR spectroscopy:
Biological applications, Annu. Rev. Phys. Chem., 2001, 52, 279.
9. N.D. Chasteen and P.A. Snetsinger, ESEEM and ENDOR spectro
scopy, in Physical Methods in Bioinorganic Chemistry, Spectroscopy
and Magnetism, ed. L. Que, Jr, University Science Books, Sausalito,
CA, 2000.
10. G. Palmer, Electron paramagnetic resonance of metalloproteins, in Phys
ical Methods in Bioinorganic Chemistry, Spectroscopy and Magnetism,
ed. L. Que, Jr, University Science Books, Sausalito, CA, 2000.
11. J.H. Freed, New technologies in electron spin resonance, Annu. Rev. Phys.
Chem., 2000, 51, 655.
12. Y. Deligiannakis, M. Louloudi and N. Hadjiliadis, Electron spin echo enve
lope modulation (ESEEM) spectroscopy as a tool to investigate the coordi
nation environment of metal centers, Coord. Chem. Rev., 2000, 204, 1.
13. C.J. Rhodes, Electron spin resonance, Annu. Rep. Prog. Chem., Sect.
C, 1999, 95, 199.
164 Appendix 2
14. J. McCracken, Electron spin echo modulation, in Handbook of Electron
Spin Resonance, ed. C.P. Poole and H.A. Farach, SpringerVerlag,
New York, 1999, vol. 2.
15. M. Ikeya, New Applications of Electron Spin Resonance: ESR Dating,
Dosimetry and Microscopy, World Scientiﬁc Publishing Co., Singapore,
1993.
16. A. Schweiger, Pulsed electron spin resonance spectroscopy: Basic princi
ples, techniques, and examples of applications, Angew. Chem. Int. Ed.
Engl., 1991, 30, 265.
17. L.R. Dalton, A. Bain and C.J. Young, Recent advances in electron
paramagnetic resonance, Annu. Rev. Phys. Chem., 1990, 41, 389.
18. Modern Pulsed and ContinuousWave Electron Spin Resonance, ed.
L. Kevan and M.K. Bowman, John Wiley and Sons, New York, 1990.
Selected Papers from the Recent Literature
High Field/Multifrequency EPR:
19. K. Mo¨ bius, A. Savitsky, A. Schnegg, M. Plato and M. Fuchs, Highﬁeld
EPR spectroscopy applied to biological systems: Characterization of
molecular switches for electron and ion transfer, Phys. Chem. Chem. Phys.,
2005, 7, 19.
20. T.A. Konovalova, L.D. Kispert, J. van Tol and L.C. Brunel, Multifre
quency highﬁeld electron paramagnetic resonance characterization of the
peroxyl radical location in horse heart myoglobin oxidized by H
2
O
2
,
J. Phys. Chem. B, 2004, 108, 11820.
21. J. Krzystek, S.A. Zvyagin, A. Ozarowski, A.T. Fiedler, T.C. Brunold and
J. Telser, Deﬁnitive spectroscopic determination of zeroﬁeld splitting in
highspin colbalt(II), J. Am. Chem. Soc., 2004, 126, 2148.
22. K.K. Andersson, P.P. Schmidt, B. Katterle, K.R. Strand, A.E. Palmer,
S.K. Lee, E.I. Solomon, A. Gra¨ slund and A.L. Barra, Examples of high
frequency EPR studies in bioinorganic chemistry, J. Biol. Inorg. Chem.,
2003, 8, 235.
23. J. van Slageren, R. Sessoli, D. Gatteschi, A.A. Smith, M. Helliwell,
R.E.P. Winpenny, A. Cornia, A.L. Barra, A.G.M. Jansen, E. Rentschler
and G.A. Timco, Magnetic anisotropy of the antiferromagnetic ring
[Cr
8
F
8
Piv
16
], Chem. Eur. J., 2002, 8, 277.
24. Z. Liang and J.H. Freed, An assessment of the applicability of multifre
quency ESR to study the complex dynamics of biomolecules, J. Phys.
Chem. B, 1999, 103, 6384.
25. J. Telser, L.A. Pardi, J. Krzystek and L.C. Brunel, EPR spectra from
EPRsilent species: Highﬁeld spectroscopy of aqueous chromium(II),
Inorg. Chem, 1998, 37, 5769.
26. D. Collison, M. Helliwell, V.M. Jones, F.E. Mabbs, E.J.L. McInnes,
P.C. Riedi, G.M. Smith, R.G. Pritchard and W.I. Cross, Single and double
quantum transitions in the multifrequency continuous wave electron par
amagnetic resonance (cwEPR) of three sixcoordinate nickel(II) complexes:
165 Advanced ESR Methods
[Ni(EtL)
2
(Me
5
dien)] and [Ni(5methylpyrazole)
6
]X
2
, X ¼ (ClO
4
)
À
or (BF
4
)
À
.
The single crystal Xray structure at room temperature of [Ni(5methyl
pyrazole)
6
]ClO
4
)
2
, J. Chem. Soc., Faraday Trans., 1998, 94, 3019.
27. P.J.M. van Kan, E. van der Horst, E.J. Reijerse, P.J.M. van Bentum and
W.R. Hagen, Multifrequency EPR spectroscopy of myoglobin. Spectral
eﬀects for highspin iron(III) at high magnetic ﬁelds, J. Chem. Soc.,
Faraday Trans., 1998, 94, 2975.
28. E.J.L. McInnes, F.E. Mabbs, S.M. Harben, P.D. Smith, D. Collison,
C.D. Garner, G.M. Sith and P.C. Riedi, Singlecrystal and multifrequency
EPR studies on chemical analogues of Amavadin
00
V(IV)doped Ca[Ti
(hida)
2
].6H
2
O, and Mo(V)doped [PPh
4
][Nb(hida)
2
] and [NEt
4
][Ta
(R,Rhidpa)
2
] [H
3
hida ¼2,2
0
(hydroxyimino)diacetic acid, H
3
hidpa ¼
2,2
0
(hydroxyimino)dipropionic acid], J. Chem. Soc., Faraday Trans.,
1998, 94, 3013.
29. A. Klein, E.J.L. McInnes, T. Scheiring and S. Zalis, Electronic structure of
radical anionic binuclear organoplatinum complexes. A multiple frequency
EPR investigation, J. Chem. Soc., Faraday Trans., 1998, 94, 2979.
30. B. Cage, A.K. Hassan, L. Pard, J. Krzystek, L.C. Brunel and N.S. Dalal,
375 GHz EPR measurements on undiluted Cr(V) salts. The role of
exchange eﬀects and gstrain broadening in determining resolution in
highﬁeld EPR spectroscopy of S ¼ 1/2 paramagnets, J. Magn. Reson.,
1997, 124, 495.
31. W.B. Lynch, R.S. Boorse and J.H. Freed, A 250GHz ESR study of highly
distorted manganese complexes, J. Am. Chem. Soc., 1993, 115, 10909.
Double Resonance:
32. J.C. Wilson, G. Wu, A.I. Tsai and G.J. Gerfen, Determination of
the structural environment of the tyrosyl radical in prostaglandin H
2
synt
hase1: A high frequency ENDOR/EPR study, J. Am. Chem. Soc., 2005,
127, 1618.
33. (a) D. Goldfarb and D. Arieli, Spin distribution and the location of
protons in paramagnetic proteins, Annu. Rev. Biophys. Biomol. Struct.,
2004, 33, 441; (b) S.S. Eaton and G.R. Eaton, Measurement of interspin
distances by EPR, Electron Paramagnetic Reson., 2004, 19, 318.
34. O.G. Poluektov, L.M. Utschig, A.A. Dubinskij and M. Thurnauer,
ENDOR of spincorrelated radical pairs in photosynthesis at high mag
netic ﬁeld: A tool for mapping electron transfer pathways, J. Am. Chem.
Soc., 2004, 126, 1644.
35. K.N. Hu, H.h. Yu, T.M. Swager and R.G. Griﬃn, Dynamic nuclear
polarization with biradicals, J. Am. Chem. Soc., 2004, 126, 10844.
36. S. Sinnecker, E. Reijerse, F. Neese and W. Lubitz, Hydrogen bond
geometries from paramagnetic resonance and electronnuclear double
resonance parameters: Density functional study of quinone radical
anionsolvent interactions, J. Am. Chem. Soc., 2004, 126, 3280.
37. B.M. Hoﬀman, (a) Electronnuclear double resonance spectroscopy (and
electronspinecho envelope modulation spectroscopy) in bioinorganic
166 Appendix 2
chemistry, Proc. Natl. Acad. Sci. U.S.A., 2003, 100, 3575; (b) ENDOR of
metalloenzymes, Acc. Chem. Res., 2003, 36, 522.
38. M. Bennati, A. Weber, J. Antonic, D.L. Perlstein, J. Robblee and
J. Stubbe, Pulsed ELDOR spectroscopy measures the distance between
the two tyrosyl radicals in the R2 subunit of the E. coli ribonucleotide
reductase, J. Am. Chem. Soc., 2003, 125, 14988.
39. J. Kang, S. Tokdemir, J. Shao and W.H. Nelson, Electronic gfactor
measurement from ENDORinduced EPR patterns: Malonic acid and
guanine dihydrate, J. Magn. Reson., 2003, 165, 128.
40. H. Mino and T. Ono, Applications of pulsed ELDORdetected NMR
measurements to studies of photosystem II: Magnetic characterization of
Y
D
tyrosine radical and Mn
21
bound to the highaﬃnity site, Appl. Magn.
Reson., 2003, 23, 571.
41. C. Elsasser, M. Brecht and R. Bittl, Pulsed electronelectron double reso
nance on multinuclear metal clusters: Assignment of spin projection factors
bsed on the dipolar interaction, J. Am. Chem. Soc., 2002, 124, 12606.
42. S.K. Smoukov, J. Telser, B.A. Bernat, C.L. Rife, R.N. Armstrong and
B.M. Hoﬀman, EPR study of substrate binding to the Mn(II) active site of
the bacterial antibiotic resistance enzyme FosA: A better way to examine
Mn(II), J. Am. Chem. Soc., 2002, 124, 2318.
43. A.R. Raitsimring and F.A. Walker, Porphyrin and ligand protons
as internal labels for determination of ligand orientations in ESEEMS
of lowspin d
5
complexes in glassy media: ESEEM studies of the orientation
of the g tensor with respect to the planes of axial ligands and porphyrin
nitrogens of lowspin ferriheme systems, J. Am. Chem. Soc., 1998, 120, 991.
Pulsed Methods:
44. C.W.M. Kay, C. Elsa¨ sser, R. Bittl, S.R. Farrell and C. Thorpe, Determi
nation of the distance between the two neutral ﬂavin radicals in augmenter
of liver regeneration by pulsed ELDOR, J. Am. Chem. Soc., 2006, 128, 76.
45. H.I. Lee, R.Y. Igarashi, M. Laryukhin, P.E. Doan, P.C. Dos Santos,
D.R. Dean, L. C. Seefeldt and B.M. Hoﬀman, An organometallic inter
mediate during alkyne reduction by nitrogenase, J. Am. Chem. Soc., 2004,
126, 9563.
46. C. Finazzo, J. Harmer, B. Jaun, E.C. Duin, F. Mahlert, R.K. Thauer,
S. Van Doorslaer and A. Schweiger, Characterization of the MCR
red2
form
of methylcoenzyme M reductase: A pulse EPR and ENDOR study,
J. Biol. Inorg. Chem., 2003, 8, 586.
47. J.M. Lu, J. Geimer, S. Naumov and D. Beckert, A Fourier transform EPR
study of uracil and thymine radical anions in aqueous solution, Phys.
Chem. Chem. Phys., 2001, 3, 952.
48. B. Epel and D. Goldfarb, Twodimensional pulsed TRIPLE at 95 GHz,
J. Magn. Reson., 2000, 146, 196.
49. S. Van Doorslaer, R. Bachmann and A. Schweiger, A pulse EPR and
ENDOR investigation of the electronic and geometric structure of cobalt
ous tetraphenylporphyrin(pyridine), J. Phys. Chem. A, 1999, 103, 5446.
167 Advanced ESR Methods
50. M. Bennati, C.T. Farrar, J.A. Bryant, S.J. Inati, V. Weis, G.J. Gerfen,
P. RiggsGelasco, J. Stubbe and R.G. Griﬃn, Pulsed electronnuclear
double resonance (ENDOR) at 140 GHz, J. Magn. Reson., 1999, 138, 232.
51. K. Fukui, Y. Fujisawa, H. OhyaNishiguchi, H. Kamada and H. Sakurai,
In vivo coordination structural changes of a potent insulinmimetic agent,
bis(picolinato)oxovanadium(IV), studied by electron spinecho modulation
spectroscopy, J. Inorg. Biochem., 1999, 77, 215.
52. W.R. Hagen, W.A.M. van den Berg, W.M.A.M. van Dongen, E.J. Reijerse
and P.J.M. van Kan, EPR spectroscopy of biological ironsulfur clusters
with spinadmixed S ¼ 3/2 ground states, J. Chem. Soc., Faraday Trans.,
1998, 94, 2969.
53. A.M. Raitsimring, P. Borbat, T. Kh. Shokjireva and F.A. Walker,
Magnetic ﬁeld (gvalue) dependence of proton hyperﬁne couplings obrai
ned from ESEEM measurements: Determination of the orientation of the
magnetic axes of model heme complexes in glassy media, J. Phys. Chem.,
1996, 100, 5235.
54. B.R. Patyal, R.H. Crepeau, D. Gamliel and J.H. Freed, Twodimensional
Fourier transform ESR in the slowmotional and rigid limits: SECSY
ESR, Chem. Phys. Lett., 1990, 175, 445.
55. B.R. Patyal, R.H. Crepeau, D. Gamliel and J.H. Freed, Twodimensional
Fourier transform ESR in the slowmotional and rigid limits: 2DELDOR,
Chem. Phys. Lett., 1990, 175, 453.
168 Appendix 2
Subject Index
Organometallic complexes have been indexed under their molecular formulae.
alkyne motion in [Mo{P(OMe)
3
}
2
(MeC≡CMe)Cp] 104–5
alternating linewidth effects 107–8
analysis
frozen solution spectra 55–9
isotropic spectra 32–40
anisotropic spectra
gmatrix Interpretation 59–60
“gstrain” 87–9
hyperfine matrix interpretation 60–3
noncoincident matrix axes 71–89
organic compounds 69–71
organometallic complexes 63–9, 83–9
solidstate spectra 53–9
spin Hamiltonian 52–3
applications
analytical 17–18
electronic structure determination 17
modified Bloch equations 102–7
rate determination 18
aromatic proton coupling, interpretation
27–9
aromatic radical anions 24
aromatic triplet state molecules 123–4
benzene anion radical spectrum 27
biological applications 17–18
biradicals
classification 112–13
exchange coupling 113–17
spin Hamiltonian 113–14
bis(diphenylphosphino)maleic
anhydride (BMA), motion 102–4
Bloch’s phenomenological model 92–4
derivation of equations 94–5
modified Bloch equations (chemical
exchange) 98–102
steadystate solution 95–8
Bohr magneton 3
cavity dip 10–11
[CH
2
{PO(OH)
2
}
2
] irrad. single crystal
spectrum 70–1
chromium nitrosyl frozen sol
n
.
spectrum 73–4
chromium(I) pianostool complex 77–9
cobalt dithiolene complexes 86–7
cobalt(0) radical anions 66–9
conversion factors 152
[Co{SC(CF
3
)C(CF
3
)S}
2
P(OPh)
3
]
frozen sol
n.
spectrum 86–7
[Co{S
2
C
2
R
2
}
2
L], noncoincidence
angles 87
[CpCo(1,3COD)]
–
frozen sol
n
.
spectrum 66, 68
[CpCo(1,5COD)]
–
frozen sol
n
.
spectrum 66, 67, 68
[CpCr(CO)
2
{NO}]
–
frozen sol
n
.
spectrum 73–4
[(C
5
Ph
5
)Co(CO)
2
]
–
frozen soln.
spectrum 67, 68
[(C
5
Ph
5
)Cr(CO)
2
{PMe
3
}] frozen sol
n
.
spectrum 77–9
170 Subject Index
Cr
2+
in aqueous solution 73–4
Cu(acac)
2
frozen solution spectrum
57–8
4,4'dicyanobiphenyl anion radical
spectrum 31
m, and pdinitrobenzene radical anion
108
dinitrodurene radical anion spectrum
107–8
dinitroxide biradicals 112, 117
bis(diphenylphosphino)maleic
anhydride (BMA), motion 102–4
diphenylpicrylhydrazyl (DPPH) 13
diphosphonylmethyl radical,
parameters 70
dipolar coupling parameter, P 155
double resonance 158, 161–2
5doxyl stearate (5DS), spin label 109
duroquinone radical anion spectrum
108
durosemiquinone, hopping rate 108
Dynamic Nuclear Polarization (DNP)
162
electron magnetic resonance (EMR)
see electron spin resonance
spectroscopy
Electron Nuclear Double Resonance
spectroscopy see ENDOR
spectroscopy
electron paramagnetic resonance (EPR)
see electron spin resonance
spectroscopy
Electron Spin Echo Envelope
Modulation, ESEEM 163
electron spin resonance spectroscopy
(ESR)
applications 17–18
background texts 1, 54, 69
definition 1
double resonance 158, 161–2
ESR experiment 3–7
frequencies 4, 11–12
high field/frequency ESR (HFEPR)
158, 159–61
history 1
nuclear hyperfine interaction 5–7
power 12–13
pulsed ESR 158–9, 162–3
saturation 5
sensitivity 4–5
electron transitions 1–2, 7, 23
ENDOR spectroscopy 41–2, 158, 161–
2
“EPRsilent” species 160–1
extended Hückel MO (EHMO)
calculations 60
fast exchange limit 100–1
[Fe(CO)
5
]
+
dilute single crystal
spectrum 74–6
[Fe(CO)
5
]
–
dilute single crystal
spectrum 74–6
Fermi contact interaction 27, 29
field modulation 8–9, 13–14
fluxionality in
[Co(Ph
2
C
2
)(CO){P(OMe)
3
}
2
] 105
fluxionality rates 30–2
forbidden transitions 116, 127–8
formation constant determination 49–
51
pFPhCN spectrum 31
frozen solution spectra 55–9
GaCl(mquin)
2
dilute single crystal
spectrum 55, 56
Gaussian line shapes 102
glycylglycine irrad. single crystal
spectrum 69–70
gmatrix Interpretation 59–60
“gstrain” 87–9
guidelines, interpretation of isotropic
spectra 27–9
Gunn diode 8
high field/frequency ESR (HFEPR)
158, 159–61
highspin transition metal ions 126–30
Hückel MO theory 27–9
extended Hückel MO (EHMO)
calculations 60
Subject Index 171
hyperfine matrix interpretation 60–3
intermediate exchange region 101
iron pentacarbonyl dilute single crystal
spectra 74–6
[IrRh(CO)
2
(PPh
3
)
2
(tolNNNtol)
2
]
+
,
frozen solution spectrum 150
isotopomer, multiplet patterns 38–9
isotropic spectra
guidelines for interpretation 27–9
isotropic spectra, organic radicals
analysis 32–40
hyperfine coupling patterns 22–5
interpretation 27–9
line positions 21–2
line widths 29–32
secondorder shifts 39–40
secondorder splittings 25–6
spin Hamiltonian parameters 26–7
isotropic spectra, organometallic
radicals
formation constant determination
49–51
line width variation 47–8
puzzling line shapes! 48–9
secondorder effects on line positions
44–7
spin Hamiltonian parameters 44
K
3
Cr(CN)
6
, spinorbit coupling
parameters 130–2
kinetic studies
alternating linewidth effects 107–8
Bloch’s phenomenological model
92–8
Gaussian line shapes 102
Lorentzian line shapes 96–8, 102
modified Bloch equations (chemical
exchange) 98–102
spin labels, rate of motion from 108–
10
klystron 7–8, 10–12
K
4
V(CN)
6
, spinorbit coupling
parameters 130–2
Larmor frequency 96
line positions, isotropic spectra 21–2
line width variations 29–32
Lorentzian line shapes 96–8, 102
lowspin Mn(II) spectra 63–5
bis( pmethoxyphenyl)nitroxide,
hyperfine parameters 42
methyl radical spectrum 26, 27, 48
methylene diphosphonic acid, irrad.
single crystal spectrum 70–1
2methylquinolin8olate (mquin)
single crystal spectra 55
[Mn(CO)(dmpe)Cp]
+
frozen sol
n
.
spectrum 88–9
[Mn(CO)
2
(PPh
3
)(C
5
H
5
)]
+
frozen sol
n
.
spectrum 144–5
[Mn(dppe)
2
(CO)(CNBu)]
2+
frozen sol
n
.
spectrum 63–5
[Mn(dppm)
2
(CO)(CN)]
+
frozen sol
n
.
spectrum 76–7
[Mn(dppm)
2
(CO)(CNBu)]
2+
frozen
sol
n
. spectrum 63–5
Mn(II) doped into PbCl
2
, spinorbit
coupling 131–2
modified Bloch equations (chemical
exchange) 98–102
applications 102–7
modulation
amplitude 15
frequency 11, 14
phase 15–16
small amplitude field modulation 8–
9
[Mo{P(OMe)
3
}
2
(MeC≡CMe)Cp] 104–
5
multiplet intensity ratios 37–8
multiplet patterns, isotopomer 38–9
naphthalene
anion radical spectrum 23–4
triplet state 121, 122
nitrogen base exchange rate, VO(acac)
2
106
nitroxide
dinitroxide biradicals 112, 117
spin labels 109–10
172 Subject Index
NO
3
dilute single crystal parameters
54–5
noncoincident matrix axes
experimental determination 72–3
organometallic complexes 73–87
perturbation theory 133–51
symmetry requirements 71–2
nuclear hyperfine interaction 5–7, 135–
43
nuclear magnetic resonance (NMR)
spectra 22
operating parameters, optimization 11
orbital angular momentum operations 60
organic radical anions, reactions 22–5
organic systems
anisotropic spectra 69–71
isotropic spectra 32
origins of ESR 1
Pascal triangle 23, 37–8
1,1,2,3,3pentacyanopropene dianion
radical spectrum 32
perturbation theory, noncoincident
matrix axes
electron Zeeman term 133–5
nuclear hyperfine interaction 135–43
quadrupole coupling 146–9
[PhCCo
3
(CO)
9
]
–
spectrum 48–9, 50
phosphonylmethyl radical, parameters
70
physical constants 152
platinum radical, hyperfine pattern 39
powder or frozen solution spectra 55–9
pulsed ESR 158–9, 162–3
pyrazine radical anion spectrum 24–5
quadrupole coupling 145–51
perturbation theory 146–9
quantization axes 92–3
[(RCCR')Co
2
(CO)
6
]
–
frozen sol
n
.
spectrum 79–81
resonance field vs microwave
frequency 161
RhIr complex 149–51
[RhLL'{μ(tol)NNN(tol)}
2
IrLL']
+
PF
6
–
spectrum 150
rotational correlation time 48
saturation, resonance 5
secondorder effects
shifts in line positions 39–40, 44–7
splittings 25–6
sensitivity, ESR spectrometer 4–5
[SFeCo
2
(CO)
9
]
–
frozen sol
n
. spectrum
79–81
signaltonoise ratio 8, 12, 16
single crystal spectra 54–5
singlyoccupied π* MO (SOMO) 27,
29, 59–63
slow passage conditions 95
solidstate spectra 53–9
spectrometer
filter time constant 16–17
frequencies 4, 11–12
modulation 8–9, 13–16
operation parameters 7–11
power 12–13
sensitivity 4–5
signal gain 16
structure 7
sweep time 14
spin Hamiltonian 3
see also specific spectral types
spin labels, rate of motion from 108–10
spin polarization 27, 28
spin traps 18
spinorbit coupling (S>1/2) 122–6
stable magnetic nuclei, properties 152–5
1,1,3,3tetracyanopropene dianion
radical spectrum 32
transition metal complexes
highspin transition metals 126–30
hyperfine coupling parameters 155–7
spin–orbit coupling 122–6
triplet state
organic molecules 122, 123
spin Hamiltonian 117–21
transition metal complexes 122–32
zerofield splitting 120
Subject Index 173
units, Tesla (T) vs Gauss (G) 3, 21
vanadium (IV) 44–51
in acidic aqueous solution 44
in basic aqueous solution 48
VO(acac)
2
106–7
nitrogen base exchange rate 106
VO(mquin)
2
dilute single crystal
spectrum 55, 56
waveguides 3–5, 10
o, m, and pxylene radical anion
spectra 35
[(oxylylene)Mn(dmpe)
2
] frozen sol
n
.
spectrum 81–6
Zavoisky, E. K. 1
Zeeman effect
electron 3, 45, 133–5
nuclear 5–6
Zeeman Hamiltonian 3, 133–5
zerofield splitting, triplet state 120
Electron Spin Resonance
Analysis and Interpretation
RI. USA
.Electron Spin Resonance
Analysis and Interpretation
Philip H Rieger Brown University. Providence.
Enquiries concerning reproduction outside the terms stated here should be sent to The Royal Society of Chemistry at the address printed on this page. without the prior permission in writing of The Royal Society of Chemistry. Designs and Patents Act 1988 and the Copyright and Related Rights Regulations 2003. in any form or by any means. criticism or review.rsc. Published by The Royal Society of Chemistry. Science Park. Cambridge CB4 0WF. Milton Road.ISBN13: 9780854043552 A catalogue record for this book is available from the British Library r The Royal Society of Chemistry 2007 All rights reserved Apart from fair dealing for the purposes of research for noncommercial purposes or private study. this publication may not be reproduced. or in accordance with the terms of the licences issued by the appropriate Reproduction Rights Organization outside the UK. Thomas Graham House. UK Registered Charity Number 207890 For further information see our web site at www. or in the case of reproduction in accordance with the terms of the licences issued by the Copyright Licensing Agency in the UK. as permitted under the Copyright. stored or transmitted.org
.
took place in one of the few laboratories in the world at the time where ESR equipment had been built. it could also serve as an introduction to the fundamentals of ESR for working research scientists in organometallic chemistry or other areas from which Phil attracted his many research collaborators. sensitive commercial instruments became available. Professor Philip H. Upon arriving at Brown his ﬁrst item of business was to design and construct a spectrometer. The writing was extended to a handout in 1991 when I was invited to spend a sabbatical year at the University of Bristol. It gives the reader a thorough introduction to the analysis and interpretation of CW ESR spectra at Xband (9. Rieger had nearly ﬁnished the manuscript of this book. and still further in 1999 for a lecture series
v
. His introduction to the ﬁeld.5 GHz. commercial instruments were not available. It began with an invitation from Prof.) as applied to paramagnetic organic.Preface
At the time of his death in April. Nevertheless. It was extended a bit further when I was invited to give a series of lectures on ESR at the University of Oviedo. William Trogler to write a chapter on ESR applications for the book Organometallic Radical Processes that was published in 1990. 2004. His intention was to present a monograph summarizing his approach to the ﬁeld of Electron Spin Resonance using examples and other explanatory material developed during the course of a 40year career of research and teaching at Brown University. There are some strong resemblances to Chapter 4 of the present book. Although the presentation was intended to be at the beginning graduate level. still further when I was invited to give some lectures at the University of Edinburgh. inorganic and organometallic molecules. When Professor Rieger ﬁrst became interested in ESR. that ﬁrst spectrometer enabled one of us (ALR) to begin a scientiﬁc collaboration that lasted the rest of Phil’s life. The instrument was eventually retired once reliable. and the other (RGL) to get his own career started at Brown. In a preliminary draft of this book Phil wrote the following paragraph explaining its origin: This book has been many years in the making. as a graduate student with George Fraenkel at Columbia University.
ch. vol 13B. 1972. 203. Journal of Organometallic Chemistry Library.. Coord. Meanwhile. pp. W. The support and encouragement of many members of the group has
. and ‘‘Electron spin resonance’’. Royal Society of Chemistry. Amsterdam.VI. computers. At the end of Chapter 1 is added an uptodate list of texts and monographs on ESR which should serve the interested reader as a source of additional treatments of the subject. Secondly. and a high speed internet connection to Professor Lawler in the mountains of New Hampshire. New York. Part IIIA. 2004. and the motivation for much of his collaborative research. W. Most of Chapter 4. Rieger are available from the Manchester University website at www. the use of libraries. as he had intended it. 135. I had given a short series of lectures on ESR at Brown University most years as part of a graduate course in Physical Inorganic Chemistry.chemistry. Rieger). C. 1994. in Appendix 2 we have referenced and given brief descriptions of some advanced ESR methods that have been developed in recent years and applied in various ﬁelds. and occasional commentary. Anne L. Philip Rieger) is grateful to Brown University for providing her with the facilities to make the completion of this project possible: oﬃce space. Specism and other ESR tools written by Prof. and also substantial parts of Chapter 3. including biochemistry. 154. Chem. 4. ed. Anne L. 178–199. Rieger (Mrs. Elsevier. Weissberger and B. was his instinctive understanding of how to analyze the powder patterns of paramagnetic inorganic complexes. Rev. ed. The modern ESR spectroscopist is now as likely to need an understanding of these techniques as of the classic Xband methods described here. pp 499–598. Other reviews by Professor Rieger on topics covered in this book include ‘‘Electron paramagnetic resonance studies of lowspin d5 metal complexes’’. Trogler. 1990. Rossiter. ac.epr. John Wiley and Sons. in Physical Methods of Chemistry.. 23. One of Philip Rieger’s most important contributions to the ﬁeld of ESR. A summary of his work in this area may be found in a review titled ‘‘Simulation and analysis of ESR powder patterns’’ published in the Specialist Periodical Report Electron Spin Resonance. Organometallics. 1993. ‘‘Chemical insights from EPR spectra of organometallic radicals and radical ions’’ (with Dr. Inc. ch.manchester.uk. been some time since a book on ESR for the nonexpert has appeared. For many years Professor Rieger was associated with the ESR group of the Royal Society of Chemistry. often using programs for simulating and analyzing such spectra that he had developed over the years. were ﬁrst published in this review. and modiﬁcation. of material that originally appeared in Organometallic Radical Processes. In completing Phil’s book we have retained the set of examples and explanations. Cambridge. When you go to this site you will ﬁnd a menu on the left with a ‘Software’ button which will give you access to this material. We hope that this book will provide a basis for study of the newer methods. A.vi
Preface
at the University of Otago. We have therefore supplemented his original manuscript in two ways. We are particularly grateful to Elsevier Publishing Company for allowing the use. vol 22. It has. however.
NH. a long time collaborator and friend. companion and colleague Phil would be pleased with the result. USA
. Rieger Ronald G. Anne L. but always interesting. USA Center Sandwich. Lawler Pawtucket. introduction to the world of desktop publishing. We hope that our friend. RI. Ron Lawler is grateful to Margaret Merritt for her advice and support during this occasionally frustrating. Finally.Preface
vii
served as an impetus to see this project to completion and is gratefully acknowledged. has also contributed to the completion of the project and his input is very much appreciated. Professor Neil Connelly of Bristol University.
1 Microwave Frequency 1.3 Center Field.1.2 Saturation 1.2 1 3 4 5 5 7 11 11 12 13 14 14 15 15 15 16 16 17 17 17 18 18
viii
.2.4.5 Modulation Frequency 1.1.3 Secondorder Splittings 2.5.4 Sweep Time 1.1 Line Positions in Isotropic Spectra 2.4.4.4.4.2.1 1.4.1.5 Applications of ESR Spectroscopy 1.2 Microwave Power 1.4.10 Filter Time Constant 1.4 Spin Hamiltonian Parameters from Spectra 21 21 22 25 26 1.6 Second Harmonic Detection 1.4.9 Signal Gain 1.2 Hyperﬁne Coupling Patterns 2.3 Determination of Rates References Chapter 2 Isotropic ESR Spectra of Organic Radicals 2.5.7 Modulation Amplitude 1.Contents
Chapter 1 Introduction What is ESR Spectroscopy? The ESR Experiment 1.8 Modulation Phase 1.4.2 Analytical Applications 1.1 Sensitivity 1.2.4.1 Isotropic ESR Spectra 2.1 Electronic Structure Determination 1.1.3 Nuclear Hyperﬁne Interaction 1.4 Optimization of Operating Parameters 1.3 Operation of an ESR Spectrometer 1.5. Sweep Width and Field Oﬀset 1.
5.6
52 53 54 55 59 60 63 63 66 69 69 70 71
4.4 An Example 2.4 Secondorder Eﬀects on Line Positions Understanding the Variation in Line Widths Puzzling Line Shapes Use of ESR Spectra to Determine Formation Constants References Chapter 4 Anisotropic ESR Spectra 4.5.2 What do you Expect to See? 2.2 3.5.3.2 Some Cobalt(0) Radical Anions Organic Examples of Solidstate ESR Spectra 4.5.2 Analysis of Frozen Solution Spectra Interpretation of the gMatrix Interpretation of the Hyperﬁne Matrix Organometallic Examples 4.2.5.1 Irradiated Single Crystal of Glycylglycine 4.6.5.5.2 Xirradiated Single Crystal of Methylene Diphosphonic Acid Noncoincident Matrix Axes
27 29 29 30 32 32 33 33 34 34 36 37 38 39 41 42
44 47 48 49 51
4.7 Multiplet Patterns due to Isotopomers 2.1 3.5
4.1 4.1 A Lowspin Manganese(II) Complex 4.2 2.6 Computation of Multiplet Intensity Ratios 2.3 3.3.5 Detailed Analysis for Determination of Parameters 2.1 Preliminary Examination of the Spectrum 2.3 4.4 4.7
.3
Interpretation of Isotropic Parameters Line Widths in Isotropic Spectra 2.5.2 Introduction Solidstate ESR Spectra 4.6.5.8 Secondorder Shifts in Line Positions 2.2 Rates of Fluxionality from Line Widths 2.3 Are the Gross Features of the Spectrum Consistent with the Model? 2.5.4 Organic Radical Reactions 2.1 Spectra of Dilute Single Crystals 4.6 Related Techniques (ENDOR) References Chapter 3 Isotropic Spectra of Organometallic Radicals 3.1 Incomplete Averaging of Anisotropies 2.5 Analysis of Isotropic ESR Spectra 2.Contents
ix
2.2.
1 The Electron Zeeman Term 5.2 Organic Triplet State Molecules and the Dipolar Interaction 6.2
71 72 73 73 74 76 77 79 81 86 87 89
92 94 95 98 102 102 107 108 110
112 113 117 122 122 122 126 130 132
133 133
.4 Applications of the Modiﬁed Bloch Equations 5.and Aaxes 7.1.3 Another Lowspin Manganese(II) Complex 4.The Modiﬁed Bloch Equations 5.1 Spin–Orbit Coupling 6.3.3. Triplet States.3 Examples: K3Cr(CN)6 and K4V(CN)6 References Chapter 7 Perturbation Theory Calculations 7.2.8.7. and other S>1/2 Systems 6.2 Iron Pentacarbonyl Ions 4.1 4.8.8.2 Steadystate Solution 5.8.3.1 Secondorder Perturbation Theory Treatment of Spin Hamiltonian with Noncoincident g.3 Further Discussion of Line Shapes 5.1.1 Biradicals 6.7.1 Derivation of the Bloch Equations 5.8.x
Contents
Symmetry Considerations Experimental Determination of Matrix Axis Orientations 4.1.8.4 Chromium(I) Pianostool Complex 4.1 A Chromium Nitrosyl Complex 4.8 Organometallic Examples of Noncoincident Matrix Axes 4.3 Transition Metal Complexes with S41/2 6.2 Highspin Transition Metal Ions 6.2 Chemical Exchange .6 (oXylylene)Mn(dmpe)2 4.1 Exchange Coupling 6.1
4.5 Alternating Line Width Eﬀects 5.7 Cobalt Dithiolene Complexes 4.1.5 [(RCCR 0 )Co2(CO)6]À and [SFeCo2(CO)9]À 4.6 Spin Labels References Chapter 6 ESR Spectra of Biradicals.9 ‘‘gStrain’’ References Chapter 5 ESR Kinetic Studies Bloch’s Phenomenological Model 5.1 Organic Triplet State Molecules 6.8.
2 Example Application of Analysis Quadrupole Eﬀects References Appendix 1 Physical Constants.2.1.4) Example References Advanced ESR Methods A2.1 Perturbation Theory Treatment of Quadrupole Term 7.2 Quadrupole Coupling 7.2 7.1.1–A1.2.1 High Frequency ESR A2. Conversion Factors.2 Double Resonance A2.3 Pulsed Methods References
7. and Properties of Nuclei (Tables A1.Contents
xi
Nuclear Hyperﬁne Interaction Perturbation Theory Treatment of Hyperﬁne Term 7.1.3
135 138 144 145 146 149 152
155 157
Appendix 2
159 161 162 164 169
Subject Index
.4 Example Application of these Results 7.
.
We will start in this chapter with a synopsis of the fundamentals of ESR spectroscopy. 16a. The history of ESR has also been described by many of those involved in the founding and development of the ﬁeld.5–36 Sources of data and a guide to the literature of ESR up to about 1990 can be found in ref. Reviews of the Oxford group’s successes are available2 and books by Abragam and Bleaney and by Abragam3 cover the major points discovered by the Oxford group.CHAPTER 1
Introduction
1. The arrow in Figure 1. also known as electron paramagnetic resonance (EPR) or electron magnetic resonance (EMR). an unpaired electron interacts with its environment. The degeneracy of the electron spin states characterized by the quantum number. or unquenched. Two kinds of environmental interactions are commonly important in the ESR spectrum of a free radical: (i) To the extent that the unpaired electron has residual.37 The electron spin resonance spectrum of a free radical or coordination complex with one unpaired electron is the simplest of all forms of spectroscopy. and the details of ESR spectra depend on the nature of those interactions. It was extended by a group of physicists at Oxford University in the next decade. proportional to the radical concentration. the total magnetic moment is diﬀerent from the spinonly moment (either larger or smaller.1 shows the transitions induced by 0.3. we focus on the spectra of organic and organotransition metal radicals and coordination complexes.
1
. the reader is referred to one of the excellent texts and monographs on ESR spectroscopy. For further details on the theory and practice of ESR spectroscopy. is lifted by the application of a magnetic ﬁeld.1 What is ESR Spectroscopy?
Electron spin resonance spectroscopy (ESR). Although ESR spectroscopy is supposed to be a mature ﬁeld with a fully developed theory. Fortunately. If unpaired electrons in radicals were indistinguishable from free electrons. mS ¼ Æ1/2. orbital angular momentum. was invented by the Russian physicist Zavoisky1 in 1945. and transitions between the spin levels are induced by radiation of the appropriate frequency (Figure 1. the only information content of an ESR spectrum would be the integrated intensity. In the present book.4 there have been some surprises as organometallic problems have explored new domains in ESR parameter space.1).315 cmÀ1 radiation.
1 0. It is customary to lump the orbital and spin angular momenta together in an eﬀective spin and to treat the eﬀect as a shift in the energy of the spin transition. The arrows show two allowed transitions.1 E/cm1 0 0.
0.2 0.2 0.
.3 0 1000 2000 3000 4000 5000 B/Gauss
Chapter 1
Figure 1.1 Energy levels of an electron placed in a magnetic ﬁeld. Since transitions are observed between sublevels with the same values of mI.2).3
0
1000
2000 3000 B/Gauss
4000
5000
Figure 1.1 E/cm1 0 0.1 0.2 0.2 0.3 0.3 0. Each nucleus of spin I splits the electron spin levels into (2I þ 1) sublevels.
depending on how the angular momentum vectors couple).2 Energy levels of an unpaired electron in a magnetic ﬁeld interacting with a spin1/2 nucleus. (ii) The electron spin energy levels are split by interaction with nuclear magnetic moments – the nuclear hyperﬁne interaction.2
0. nuclear spin splitting of energy levels is mirrored by splitting of the resonance line (Figure 1.
the parameter g is negative. where gemB/hc ¼ 0. For electron spins. and both negative signs may be combined to give the formula as written. Magnetic ﬁelds of up to ca. mB is the Bohr magneton (9. This has the advantage of allowing gvalues to be tabulated as positive numbers. Waveguides look like rectangular crosssection pipes with dimensions on the order of the wavelength
w
Energy level splitting in a magnetic ﬁeld is called the Zeeman eﬀect. and the Hamiltonian of eqn (1.9348 Â 10À4 cmÀ1 GÀ1: ~ n ¼ lÀ1 ¼ n=c ¼ ðgmB =hcÞB ð1:4Þ
Since the gvalues of organic and organometallic free radicals are usually in the range 1.274 Â 10À28 J GÀ1).9–2. however. the energy of a magnetic dipole in a magnetic ﬁeld is the negative of that shown in eqn (1. (ge ¼ 2.1). Technically. i.1 and as described by the spin Hamiltonian: ^ ^ H s ¼ gmB BS z ð1:1Þ
In eqn (1.
. B is the magnetic ﬁeld strength in Gaussz.1) is sometimes referred to as the electron Zeeman Hamiltonian.1. Many spectroscopists still express the magnetic ﬁeld in Gauss (G) ¼ 10À4 T. The electron ^ spin energy levels are easily found by application of H s to the electron spin eigenfunctions corresponding to mS ¼ Æ1/2: ^ H s Æ1 ¼ Æ1 gmB BÆ1 ¼ EÆ Æ1 2 2 2 2 Thus EÆ ¼ Æ À1Á
2
gmB B
ð1:2Þ
The diﬀerence in energy between the two levels: DE ¼ Eþ À EÀ ¼ gmB B ¼ hn ð1:3Þ
corresponds to the energy. thus we could use radiation with n up to 1. 15000 G are easily obtained with an ironcore ~ electromagnet.4 cmÀ1 (n o 42 GHz or l 4 0. Microwaves are normally handled using waveguides designed to transmit radiation over a relatively narrow frequency range.71 cm).4).Introduction
3
1. hn.2 The ESR Experiment
When an electron is placed in a magnetic ﬁeld. Radiation with this kind of wavelength is in the microwave region. the free electron value is a good starting point for describing the experiment. the magnetic moment and spin angular momentum vectors are antiparallel. z Using the symbol ‘‘B’’ for the magnetic ﬁeld technically implies we are using the MKS system of units in which B is given in Tesla (T). however. or in wavenumbers by eqn (1.e. the degeneracy of the electron spin energy levels is liftedw as shown in Figure 1. g is called the gvalue (or gfactor).00232 for a free electron). of a photon required to cause a transition.. and Sz is the zcomponent of the spin angular momentum operator (the magnetic ﬁeld deﬁnes the zdirection). and we will continue that practice here.1).
2 0. the free electron resonance is found at 3390 G. a photon can induce either absorption or emission. Since the ﬁeld at which absorption occurs is proportional to microwave frequency. ESR spectrometers are available commercially or have been custom built in several frequency ranges (Table 1.2.
1.5 24 35 95 360 l (cm) 27 10 3.3 cm (n E 9–10 GHz). is proportional to the diﬀerence: Net absorption / NÀ À Nþ The ratio of populations at equilibrium is given by the Boltzmann distribution: Nþ =NÀ ¼ expðÀDE=kTÞ ¼ expðÀgmB B=kTÞ ð1:5Þ
For ordinary temperatures and ordinary magnetic ﬁelds.0 9. the most common choice. Since absorption is proportional to the number of spins in the lower level and emission is proportional to the number of spins in the upper level.2 1.e.083 B (electron) (G) 390 1070 3400 8600 12500 34000 128000
to be transmitted.1
Designation L S X K Q W –
Common frequencies used for ESR
n (GHz) 1. the population diﬀerence can be written: NÀ À Nþ ¼ NÀ ½1 À ð1 À gmB B=kTÞ ¼ NgmB B=2kT ð1:6Þ
This expression tells us that ESR sensitivity (net absorption) increases with the total number of spins. N..1
Sensitivity
As for any quantum mechanical system interacting with electromagnetic radiation. eÀx E 1 À x. Although Xband is by far the most common. absorption intensity.e. in principle sensitivity should be greater for
.0–3. called Xband microwaves. i. the diﬀerence between the number of photons absorbed and the number emitted.31 0..4
Chapter 1
Table 1. has l in the range 3.1 3. i. with decreasing temperature and with increasing magnetic ﬁeld strength. the exponent is very small and the exponential can be accurately approximated by the expansion. As a practical matter for ESR. waveguides can not be too big or too small À1 cm is a bit small and 10 cm a bit large. net absorption. Thus Nþ =NÀ % 1 À gmB B=kT Since NÀ E N1 E N/2.85 0.1). The experiment detects net absorption. in the middle of Xband.
By ideal conditions. gBIz. Therefore an intense radiation ﬁeld.. the K. Since spin–orbit coupling provides an important energy transfer mechanism. those with heavier atoms (e.3
Nuclear Hyperﬁne Interaction
When one or more magnetic nuclei interact with the unpaired electron.
1. When the resonance is split into two or more hyperﬁne lines. which tends to equalize the populations. those with shorter T1 will be more diﬃcult to saturate. with sensitivity going down roughly as the reciprocal square of the line width. However. which are weaker than the electric forces responsible for most other types of spectroscopy. organic radicals) have long T1s. The eﬀect of saturation is considered in more detail in Chapter 5..2. on the order of 0. This eﬀect is called ‘‘saturation’’.
1.2. is a rate process called spin–lattice relaxation.e. and. However.1 G wide. sensitivity decreases still further. another term in the spin Hamiltonian: ^ ^ ^ ^ H s ¼ gmB BS z þ AI Á S ð1:7Þ
where A is the hyperﬁne coupling parameter in energy units (joule)..2
Saturation
Because the two spin levels are aﬀected primarily by magnetic forces.g.g. via energy transfer to the surroundings.. in most cases the energy contributions are negligible on the ESR energy scale. spin systems weakly coupled to the surroundings) will be easily saturated.e. a commercial Xband spectrometer can detect about 1012 spins (ca.
. i. we usually ﬁnd that oddelectron species with light atoms (e. Nonetheless. with a characteristic time (T1). so samples are also necessarily smaller and for the same concentration contain fewer spins. This usually more than cancels the advantage of a more favorable Boltzmann factor for samples of unlimited size or ﬁxed concentration. 10À12 moles) at room temperature. once the populations are disturbed by radiation it takes longer for equilibrium population diﬀerences to be established. organotransition metal radicals) have shorter T1s.Introduction
5
higher frequency K. Strictly speaking we should include the nuclear Zeeman interaction.or Qband waveguides are smaller. the spin–lattice relaxation time (relaxation rate constant ¼ 1/T1). ESR is a remarkably sensitive technique. Systems with a long T1 (i. we have another perturbation of the electron energy. This number of spins in a 1 cm3 sample corresponds to a concentration of about 10À9 M. where the phenomenological Bloch equations are introduced. leads to a decrease in net absorption which is not instantly restored once the radiation is removed. Under ideal conditions.or Qband spectrometers than for Xband. we mean a single line. especially compared with NMR. The return of the spin system to thermal equilibrium.
M. there are two levels corresponding to a spin singlet (E ¼ À3A/4hc) and a triplet (E ¼ þA/ 4hc).y The eigenfunctions of the spin Hamiltonian [eqn (1. eqn. M. 766) but are assumed to be negligible for the analysis of powder spectra (see Chapter 4) where A is often large or the resolution is insuﬃcient to reveal subtle spectral features.8D) are negligible. J. which corresponds to ﬁrstorder in perturbation theory. Notice that. H. 1 ¼ 1gmB B þ 1A ð1:8AÞ 2 2 2 4 À Á E À1.. DmI ¼ Æ1 (NMR transitions). the selection rules are DmS ¼ Æ1. A/hc ¼ 0.. 2 2 If the hyperﬁne coupling is suﬃciently small. 1959.9) (the ﬁrstorder solution is called the highﬁeld approximation). however. Phys. J.7) are: À Á E 1. T.00.2.7)] are expressed in terms of an electron. 1960.. corresponding to the electron and nuclear spin quantum numbers mS and mI. À1i 2 2 and j À 1. 748. mIi. and. Heller. The nuclear Zeeman interaction does. 1960. The latter involves much lower energy photons.
. Blinder. J. the second term in brackets in eqns (1. Chem. 1i. Phys. At lower ﬁelds. or when A
y
An exception to this rule arises in the ESR spectra of radicals with small hyperﬁne parameters in solids. transitions among these levels can involve changes in angular momentum of only one unit. Chem.38 are plotted as functions of B in Figure 1. Fessenden. which were ﬁrst derived by Breit and Rabi. 82. 33. only the DmS ¼ Æ1 transitions are observed in an ESR spectrometer.6
Chapter 1
since observed transitions are between levels with the same values of mI. McConnell. Soc. In that case the interplay between the Zeeman and anisotropic hyperﬁne interaction may give rise to satellite peaks for some radical orientations (S. at zero ﬁeld. the four levels divide into two higher levels (mS ¼ þ1/2) and two lower levels (mS ¼ À1/2) and approach Figure 1.1 cmÀ1. À1 ¼ À1gmB B þ 1A 2 2 2 4 Eþ ¼ 1gmB B ½1 þ ðA=gmB BÞ2 2 À 1 A 2 4 EÀ ¼ À1gmB B ½1 þ ðA=gmB BÞ2 2 À 1 A 4 2
1 1
ð1:8BÞ ð1:8CÞ ð1:8DÞ
The eigenfunctions corresponding the E1 and EÀ are mixtures of j1. A oo gmBB. (1. play a central role in electronnuclear double resonance experiments and related methods [Appendix 2 and Section 2.8). DmI ¼ 0 (ESR transitions) or DmS ¼ 0. At high ﬁeld. the ﬁrstorder result. W. in other words. C. Cole and R. 33. To conserve angular momentum.2. respectively. Chem. the nuclear Zeeman energies cancel in computing ESR transition energies. The energy eigenvalues of eqn (1.6 (Chapter 2)].8C) and (1. in practice. The exact energies in eqn. Sternlicht.3 for g ¼ 2. Am. and the energies become: E ¼ Æ1gmB B Æ 1 A 2 2 ð1:9Þ
These are the energy levels shown in Figure 1. (1. 1128). Such eﬀects have been observed in organic free radicals (H.and nuclearspin basis set mS. At high ﬁelds this corresponds to ﬂipping only one spin at a time.
3 0 1000 2000 3000 B/Gauss 4000 5000
7
Figure 1.2 0.
1.
Load
Diode Detector
Klystron
Attenuator
Circulator µAmmeter Magnet Cavity
Figure 1.3 0.4 Block diagram of an ESR spectrometer.4. the transitions may involve simultaneous ﬂipping of electron and nuclear spins.1 E/cm1 0 0. Planepolarized microwaves are generated by the klystron tube and the power level adjusted with the Attenuator. The arrows show the transitions induced by 0. The Circulator
.315 cmÀ1 radiation. the vast majority of laboratory instruments are based on the simpliﬁed block diagram shown in Figure 1.Introduction
0.3 Energy levels for an electron interacting with a spin1/2 nucleus with A/hc ¼ 0.3 Operation of an ESR Spectrometer
Although many spectrometer designs have been produced over the years. This gives rise to secondorder shifts in ESR spectra (see Chapters 2 and 3).2 0.
becomes comparable in magnitude to gmBB.1 0.1 cmÀ1.
component in the diode current. Figure 1. component is ampliﬁed using a frequency selective ampliﬁer tuned to the modulation frequency. A schematic drawing of the klystron is shown in Figure 1. an anode to collect the electrons. signal is proportional to the change in sample absorption as the ﬁeld is swept. and a highly negative reﬂector electrode that sends those electrons which pass through a hole in the anode back to the anode. Thus the detected a. of course. As with a klystron. It takes a little practice to get used to looking at ﬁrstderivative spectra. such a measurement would detect noise at all frequencies as well as signal and have a far too low signaltonoise ratio to be useful. Microwaves reﬂected back from the Cavity (which is reduced when power is being absorbed) are routed to the Diode Detector.
. especially those employing pulsed excitation (see Appendix 2). but there is a distinct advantage: ﬁrstderivative spectra have much better apparent resolution than do absorption spectra. An oscillating magnetic ﬁeld is superimposed on the dc ﬁeld by means of small coils. thus eliminating a great deal of noise. this amounts to detection of the ﬁrst derivative of the absorption curve. and any power reﬂected from the diode is absorbed completely by the Load. The motion of the charged electrons from the hole in the anode to the reﬂector and back to the anode generates an oscillating electric ﬁeld and thus electromagnetic radiation. B. This a. This device makes use of the Gunn Eﬀect. in which electrons become ‘‘trapped’’ in potential wells within a solid and oscillate at a frequency that may be varied by changing the applied voltage. tuning is done by varying the voltage. Indeed.8
Chapter 1
behaves like a traﬃc circle: microwaves entering from the klystron are routed toward the Cavity where the sample is mounted. leading to an a.6 shows the eﬀect of higher derivatives on the resolution of a 1:2:1 triplet arising from the interaction of an electron with two equivalent I ¼ 1/2 nuclei. The transit time from the hole to the reﬂector and back again corresponds to the period of oscillation (n). Thus. As shown in Figure 1. The microwavegenerating klystron tubez requires explanation.7. The solution to the signaltonoise problem is to introduce small amplitude ﬁeld modulation.c. The diode is mounted along the Evector of the planepolarized microwaves and thus produces a current proportional to the microwave power reﬂected from the cavity. usually built into the cavity walls. discovered by J.c.c. Gunn of IBM in 1963. it is swept back and forth through part of the line. Thus the microwave frequency can be tuned (over a small range) by adjusting the physical distance between the anode and
z
Some modern spectrometers.5. In practice. use a solidstate Gunn diode microwave source. When the ﬁeld is in the vicinity of a resonance line. the absorption of microwaves by the sample could be detected by noting a decrease in current in the Microammeter. There are three electrodes: a heated cathode from which electrons are emitted. Gunn diode oscillators have better frequency stability and longer lifetimes than klystrons and are becoming the preferred microwave source for ESR spectrometers. secondderivative spectra are even better resolved (though the signaltonoise ratio decreases on further diﬀerentiation). The modulation amplitude is normally less than the line width. in principle.
Introduction
9
Absorption Detected Signal
Mod. Ampl.
Magnetic Field
Figure 1.5
Smallamplitude ﬁeld modulation converts the absorption curve into a ﬁrstderivative.
2nd derivative
1stderivative
Absorption
Magnetic Field Strength
Figure 1.
.6 Firstderivative curves show better apparent resolution than do absorption curves – and secondderivatives curves are still better.
10
Electron pathway Heated filament cathode Reflector electrode
Chapter 1
Anode
Figure 1. The tuneup procedure usually includes observing the klystron power mode. the frequency is swept across a range that includes the cavity resonance by sweeping the klystron reﬂector voltage. while the reﬂector voltage is used to ﬁne tune the frequency.9). Thus the spectrometer must be tuned such that the klystron frequency is equal to the cavity resonant frequency.
Electric Field Sample Port
Iris
Magnetic Field
Figure 1. B. Note that the sample is mounted in the electric ﬁeld nodal plane. The ‘‘cavity dip’’ is centered on the power mode using the coarse mechanical frequency adjustment.8 Microwave cavity.7 Schematic drawing of a microwavegenerating klystron tube.8). is perpendicular to the sample port. The electric and magnetic ﬁelds of the standing wave are shown in the ﬁgure. but it must be exactly one wavelength. The static ﬁeld. exactly one wavelength long. microwave energy is absorbed by the cavity and the power reﬂected from the cavity to the diode is minimized. The cavity length is not adjustable. When the klystron frequency is close to the cavity resonance. An Xband cavity has dimensions of about 1 Â 2 Â 3 cm. The cavity is a rectangular metal box. both methods are used: the metal tube is distorted mechanically to adjust the distance (a coarse frequency adjustment) and the reﬂector voltage is adjusted as a ﬁne control. but at a maximum in the magnetic ﬁeld. In practice. The sample is mounted in the microwave cavity (Figure 1.
the reﬂector or by adjusting the reﬂector voltage. and the diode detector current is plotted on an oscilloscope or other device. resulting in a dip in the power mode (Figure 1.
. That is.
Brieﬂy. of course. The purpose of these parameters and typical values. In the early days of ESR these adjustments would be carried out using control knobs on a console. A.
1. involves adjustment of the set of operating parameters described below. however.
.1
Microwave Frequency
The resonant microwave frequency reaching the sample is determined by the eﬀective length of the microwave cavity. Weissberger and B. which is still the most common commercially available type of instrument. VI. Inc. New York. however. in Physical Methods of Chemistry.2
Parameters involved in the recording of an ESR spectrum
Center ﬁeld Sweep width Field oﬀset Modulation frequency 1st or 2nd harmonic Modulation amplitude Modulation phase Signal gain Filter time constant
Microwave frequency Microwave power Sweep time
1. ed. part IIIA. Such spectrometers are quite adequate for studying relatively stable samples.
Table 1.Introduction
11
Diode Current
Reflector Voltage
Figure 1.. has remained unchanged. the settings are carried out under control of a computer interface. Characterizing transient species by ESR.4 Optimization of Operating Parameters8
Determining the ESR spectrum of a sample using a typical CW spectrometer of the sort outlined in Figure 1. Rossiter.4. below. ch. Nowadays. John Wiley and Sons. each parameter is discussed and the means used to optimize or measure the parameter described. W. There are twelve parameters that must be set or known in recording an ESR spectrum (Table 1. The actual length is somewhat
8
A comprehensive discussion and illustrations of the eﬀects of spectrometer operating parameters on ESR spectra are given by the author in Electron spin resonance. 1972.2).9 Klystron mode and cavity dip.4. requires substantial modiﬁcation of commercial instruments or the use of a pulsed spectrometer..
is given by eqn (1. Units are relative values of B12. v¼ gB1 M0 T2 1þ
2 T2 ðo0
À oÞ2 þ g2 B2 T1 T2 1
ð1:10Þ
where T1 and T2 are the spin–lattice and transverse relaxation times. However. Since the klystron frequency is locked to the cavity resonant frequency by a suitable feedback circuit. For small B1.10).
1. v goes through a maximum when g2B12T1T2 ¼ 1. the absorption signal increases linearly with increasing B1.10 Eﬀect of increasing microwave power on intensity and shape of an ESR line. v. At maximum v. the saturation term leads to an increase in apparent line width by a factor of O2.
. and B1 is the amplitude of the oscillating ﬁeld. Power increases from top to bottom of the ﬁgure. g is the magnetogyric ratio (gmB in ESR terms). the frequency must be known quite accurately.2
Microwave Power
According to the solution of the Bloch equations (Chapter 5). sometimes called the ‘‘vmode signal’’. respectively.10. however. To simultaneously maximize signaltonoise ratio (S/N) and minimize distortion. and begins to decrease with increasing B1 – the resonance begins to ‘‘saturate’’. o0 is the resonant frequency (proportional to B0). it is best to adjust the microwave power.4. This is normally done using a microwave frequency counter installed somewhere in the waveguide circuit. as measured by the power
Figure 1. this is not an adjustable parameter. However. This eﬀect is illustrated in Figure 1. The microwave power is proportional to B12. to compute a gvalue from a spectrum.12
Chapter 1
modiﬁed by the inﬂuence of the sample tube and the Dewar insert (if controlled temperature operation is required) so that the microwave frequency varies by a few percent even for the same cavity. M0 is the bulk magnetization. o is the angular microwave frequency (n). the magnetic resonance absorption. when the third term in the denominator becomes important.
This will result in a decrease of less than 2 in signal amplitude. or can guess.0028). This eﬀect is illustrated in Figure 1. down by a factor or 4 or 5 (6–7 db) from the maximum amplitude power. The center ﬁeld corresponds to the middle of the spectrum and a suﬃciently large sweep width chosen so that all of the spectrum is recorded. setting the center ﬁeld and sweep width values is not very diﬃcult.Introduction
13
Figure 1. Hopefully. following this procedure would require ﬁnding the maximum amplitude by trial and error and then turning the power level down by 6–7 db. Sweep Width and Field Oﬀset
Once you know. which has a wellknown gvalue (2. such estimates of the value of B inside the cavity are not suﬃciently accurate to be used for gfactor determinations. or (ii) use of an NMR gaussmeter probe inserted in or near the cavity several times during the collection of the spectrum. the ﬁeld limits of your spectrum. Most spectrometers measure the magnetic ﬁeld by a Hall eﬀect probe consisting of a sensor mounted on one of the pole faces of the magnet. There are two ways around this problem: (i) measure the spectrum of a solid free radical such as diphenylpicrylhydrazyl (DPPH).11 Eﬀect of microwave power on relative width and amplitude of a line.
1.
.11.3
Center Field. you will see at least a piece of your spectrum and can make appropriate adjustments to zero in on the correct settings. If you do not know the ﬁeld range occupied by your spectrum in advance. at least once during acquisition of the desired spectrum. However. In practice.
meter. the center ﬁeld must be chosen by educated guess. set the sweep width 2–4Â greater than the expected width. In general it is unnecessary to spend much time adjusting the power level.4. The general rule is to adjust the power to about 10 db attenuation for organic radicals and to use full power for transition metal complexes and those organometallics where the unpaired electron is primarily located on the metal.
4. The higher frequency generally gives better S/N.12
Eﬀect of 100 kHz modulation on an ESR line of decreasing width. The xaxis is magnetic ﬁeld.
1. and the patience of the operator. but if the lines are unusually sharp (o0. you have a choice of either 100 kHz or a lower frequency of ﬁeld modulation.
. 100 kHz modulation leads to ‘‘side bands’’. This eﬀect is illustrated in Figure 1. the longer the sweep time the better the sensitivity since the ﬁlter time constant parameter can be set longer with consequent improvement in signaltonoise ratio. use the lower frequency for which the sidebands are closer together and thus less likely to be a problem.08 Gauss).4. sweep times are usually set in accordance with the expected lifetime of the radical species. Under such circumstances.12. In practice. the stability of the instrument. lumps in the line shape that confuse the interpretation of the spectrum. relative to the center of the line.5
Modulation Frequency
With most spectrometers.14
Chapter 1
1. however.
Figure 1.4
Sweep Time
In general. in Gauss. Decay of the radical or drift of the spectrometer during a scan is clearly undesirable. The sweep time is most commonly set in the range 4–10 min.
13 Eﬀect of ﬁeld changing (modulation/line width) on relative amplitude (lefthand axis) and observed width (righthand axis) of an ESR line.Introduction
15
1. here are some rules: For optimum S/N ratio. This is illustrated in Figure 1. there is a phase control which. This means that the detected signal must not only be at the modulation frequency.
. However. this control needs to be adjusted only rarely and in most cases the best approach is to leave it alone. For most practical work: Modulation amplitude ¼ linewidth/3.6
Second Harmonic Detection
In most cases.
1. Since the ampliﬁer itself can introduce a bit of phase shift. Depending on what you want to optimize. If your spectrum has very good S/N and has some regions where you would like better resolution. the modulation signal is processed by ampliﬁcation with a tuned ampliﬁer using phasesensitive detection. In general.8
Modulation Phase
To improve the S/N ratio. a secondderivative presentation may help. second derivatives from second harmonic detection are very costly in terms of S/N ratio and so you really do have to have a strong signal!
1. should be adjusted to maximize the signal amplitude. the signal disappears when the modulation amplitude is turned to zero. but must also be in phase with the modulation. For accurate line width measurements: Modulation amplitude ¼ linewidth/10.4.4. you will use the ﬁrst harmonic and the normal ﬁrstderivative of the absorption spectrum will be presented.4. In practice. in principle.
Figure 1. the signal increases more or less in proportion to the modulation amplitude. but eventually the detected lines begin to broaden and then the signal amplitude decays as well. but decreased resolution: Modulation amplitude ¼ 2 Â linewidth.13.7
Modulation Amplitude
Since the absorption signal is usually detected by imposing a 100 kHz ﬁeld modulation on the static ﬁeld.
but still too long.14 Eﬀect of time constant/time to sweep through the line (xaxis) on the relative amplitude and signaltonoise ratio (lefthand axis) and relative width and shift (righthand axis) of an ESR line. These eﬀects are illustrated in Figure 1.16
Chapter 1
1. Too high a gain can result in ampliﬁer overload. time constant. Set the gain suﬃciently high that the recorded spectrum is nearly fullscale on the computer displays or recorder. The best time constant setting thus depends critically on the sweep time. If you have no idea of the strength of your expected signal.
Figure 1. and the apparent line width will increase. you may ﬁlter out the signal altogether! With a shorter. the line shape will be distorted.9
Signal Gain
Adjustment of the signal gain is straightforward.4. a good starting point would be 1 Â 104. Thus.4.14. the apparent line center will shift. but not high frequency. If you sweep through a line rapidly using a long time constant.
. The cutoﬀ frequency is the reciprocal of the time constant setting. signals. The best ruleofthumb is to set the time constant to 10% of the time required to pass through the narrowest line of your spectrum.10
Filter Time Constant
The output of the signal ampliﬁer is ﬁltered using a circuit designed to pass lowfrequency.1 s would lead to signals with frequencies greater than about 10 Hz being attenuated.
1. a time constant of 0.
Ai.
1. We will discuss solid state samples later.6 s. can be used analytically to determine the concentration of a paramagnetic species.25 G/2.
1. the spectrum of a freely tumbling radical in liquid solution. which are related to the way in which the unpaired electron spin density is distributed in the molecule. B. The time required to pass through a line is 0. either a dilute single crystal or a frozen solution.5. Experimentalists often make qualitative observations relating ‘‘strong’’ or ‘‘weak’’ spectra to chemical circumstances. can contain two principal types of chemically useful information: (i) The hyperﬁne coupling pattern provides information on the numbers and kinds of magnetic nuclei with which the unpaired electron interacts. point out the variables that must be controlled.5 G minÀ1 ¼ 0.. The degree to which ESR intensities have been exploited varies widely.2
Analytical Applications
Like all forms of spectroscopy.
. and gvalue. inorganic and organometallic radicals to coordination complexes and biological macromolecules containing a paramagnetic center. beginning in Chapter 4. The time constant should be set close to 0. ESR intensities. The magnetic ﬁeld values at which microwaves are absorbed to produce the ESR spectrum of an isotropic sample. Such analytical applications usually require a standard solution to establish a calibrated intensity scale. e. suppose that the line width is 0. We will explore these kinds of applications extensively in subsequent chapters.1 Electronic Structure Determination
Most commonly.12a who discuss the technique. N. The scan rate then is 10 G/4 min ¼ 2. The interested reader is referred to Wertz and Bolton. and suggest intensity standards. ESR spectroscopy is used to obtain information pertaining to the electronic structure of the species being studied.Introduction
17
For example.5 G minÀ1. Computer software is required in most cases to doubly integrate the spectrum (two integrations are required to convert the derivative spectrum into absorption and then to a number representing the total concentration of the species being studied). In solids the parameters may also depend on molecular orientation relative to the magnetic ﬁeld. through the dependence on the number of spins.1 min or 6 s. g.g. that you are scanning through 10 G in 4 min. but quantitative applications of integrated intensities are rare in ESR studies. Even more information is available from the spectrum of a solidstate sample.25 G. Structural applications range from organic.5.5 Applications of ESR Spectroscopy
1. In isotropic spectra the observed parameters are averaged over rotation of the molecules. (ii) The spacing of the lines and the center of gravity of the spectrum yield the hyperﬁne coupling constants.
34. 1957. 1961. which average anisotropies in the g.
. (c) D. 1945.. Such applications are discussed in more detail in Chapter 5. 108. which plague ESR spectroscopists studying organic radicals and NMR spectroscopists in general. Bleaney. Rep. Rep. which is out of the range of operation of most continuous wave spectrometers. a measure of the rate of energy transfer between the electron spin and its surroundings. short relaxation times imply broad lines and reduced sensitivity. 9.5.W. The advent of modern high power pulsed spectrometers32. Short relaxation times mean that saturation problems. Abragam and B. Phys. 245. E. which average nuclear positions in a radical. 304. (a) A. Similarly.. Bleaney and K. Owen. J. Zavoisky. 2. The Principles of Nuclear Magnetism. Electron Paramagnetic Resonance of Transition Ions. Largely because spin–orbit coupling results in spin state admixture. Abragam. The spectral line widths are related to the rate of the rotational motions. Phys. Stevens. 16. Progr. 20.39
1. Rep. Bagguley and J. However. Bowers and J. 211. Saturation often depends on the same structural and dynamic properties as line widths. Oxford. The saturation behavior of a spectrum – the variation of integrated intensity with microwave power – is related to the spin–lattice relaxation time. are largely absent in organometallic ESR studies. (b) K. Owen. This has three consequences. ESR spectra can be used to determine the rate of a chemical reaction or the rate of a conformation change.18
Chapter 1
‘‘Spin traps’’ which scavenge a reactive radical to give a more persistent radical may be used to detect the intermediacy of such transient radicals in sorting out a reaction mechanism. spectra ordinarily can be recorded at full microwave power with salutary consequences for sensitivity.D.S. electronic spin–lattice relaxation times are normally short for species containing heavy atoms such as transition metals. Phys. 3.
References
1. 1970.. and to the rates of ﬂuxional processes. USSR. (b) A. 1953.36 has made it possible to overcome some of the restrictions arising from relaxation behavior. Progr. 304. 1955. very high power is required. Progr. Thus.3
Determination of Rates
In some cases. If relaxation times are short. 18. Some workers have used ‘‘spin labels’’ attached to a membrane or biological macromolecule to study the motion of these components of living cells (Chapter 5). Clarendon Press.H. Clarendon Press. relaxation times are most easily determined by measuring spectral intensity as a function of microwave power in the saturation region.and hyperﬁne matrices (Chapter 5). Oxford. (a) B. Phys.M.
Schwartz. Wiley. Techniques and Applications. 1967. Electron Paramagnetic Resonance.H. Electron Paramagnetic Resonance: Elementary Theory and Practical Applications. New York. to Electron Paramagnetic
Standard texts and monographs
9. 21. 1999. 11. Hudson.V.M. Specialist Periodical Report. New York.
. T. Pake. G.J.M. Biological Applications of Electron Spin Resonance. W. J. 1983.R. Biological and Biochemical Applications of ESR. P.A. Wiley. Symons. Kozyrev. 1968. 1974. New York. Baird. Electron Spin Resonance. 55.A. Poole. L. 14. J. Blumenfeld.P. 7. 1972. New York. Bolton and J.G. Wertz and J. WileyInterscience.C. An Introduction Resonance. 11A. Ingram. 16. 1970.
Instrumentation
17. (b) Vol. Paramagnetic Resonance: An W. Semenov.Introduction
19
4. A. AIP Press. Paramagnetic Resonance. New York. C. Adam Hilger. Alger. Electron Academic Press. Gordon and Breach. New York. Wilmshurst.R. Jr.R.A. Chicester. Jr. 20. 18.
Early monographs
5. Gordy.P. 1967. McGrawHill. Wertz. London. Electron Spin Resonance: Theory and Applications.E. 1964. 1966. London.E. Electron Spin Resonance Spectrometers. Farach. Ingram. Electron Spin Resonance.M.C. Halsted Press. (b) J. Plenum Press. 6. New York. 1988. A. Ellis Horwood. M. Bolton and D. 15. ed. New York. 1994. Assenheim. D. Benjamin. Poole. (a) J. 1962. Electron Spin Resonance in Chemistry. 2. Bolton.M..E. S. WileyInterscience. H. New York. G. Ellis Horwood. Monographs on Electron Spin Resonance. Feher. 1980. Voevodski and A. 1958. Chicester. V. W. 1994. 1.E. Interscience. 1969. 12.
Biological applications
19. Theory and Applications of Electron Spin Resonance. Berson and J. A. 13. Borg. Plenum Press. Weil. London. 2 vols. (a) Vol. ed. (b) Principles of Electron Spin Resonance. Atherton.J. Benjamin. 1972. 1973. New York. Free Radicals as Studied by Butterworth.. D.C. New York. New York. Methuen. Electron Paramagnetic Resonance with Applications to Selected Problems in Biology. London. 8. Electron Spin Resonance: A Comprehensive Treatise on Experimental Techniques. Ayscough.A. and H. ed. Introduction to Electron Spin Resonance. R. Altshuler and B.R. in Electron Spin Resonance. New York. C. 1993.E. Introductory Monograph. Royal Chemical Society.S. 1967. 10. (a) N. Handbook of Electron Spin Resonance. H. M.B. New York.
1990.. J. M. High Resolution E. WileyVCH. Kluwer Academic/Plenum Publishers. Electron Spin Double Resonance Spectroscopy. Kevan and M. Hoﬀ. 2004. Schwartz. Bowman. D.S.R. 2082. L.
Organic free radicals
29. Symons. Electron Paramagnetic Resonance of d Transition Metal Compounds. 38. 2003. Eaton and L.C. 32. 36. John Wiley.D. Dosimetry and Microscopy. Schweiger and G. Rehorek. L. Mabbs and D. G. Eaton. Transition Ion Electron Paramagnetic Resonance. Gerson. Wiley.
Advanced methods
31. Amsterdam.J. 38. World Scientiﬁc.E. Singapore. ed.J. Oxford. Telser. 1992.20
Chapter 1
22. F. 341.C. Berliner. 2003. A.
. Pilbrow.M. 37. Soc.. Rabi. M. 1967. 1976.W. 20. S. 1993. Oxford. Paramagnetic Resonance of Metallobiomolecules. Ikeya. WileyInterscience. Eaton.R. ed. 27. 23. Clarendon Press. Modern Pulsed and ContinueousWave Electron Spin Resonance. 1990. Collison. Oxford. New York. 39.J.N. New York. Principles of Pulse Electron Paramagnetic Resonance. John Wiley and Sons Ltd. L. Breit and I. 30. Weinheim.S. 28. Rev. Chem. 33. Huber. Salikhov. Kispert. 24. ed. A. Baker and Taylor. 25. 1989. NC. Time Domain Electron Spin Resonance. 1931. 1979. Foundations of Modern EPR. Gerson and W. New York. Biomedical EPR. In vivo EPR (ESR): Theory and Applications. Amsterdam. Kevan and L. L. Oxford University Press. Jeschke. Elsevier. 2004. 2001.R. Electron Spin Resonance Spectroscopy of Organic Radicals. 1978. Elsevier. J. New York. Spectroscopy. Kevan and R. G. New Applications of Electron Spin Resonance: ESR Dating. G.. P. Atkins and M. 858. London. Phys. New York. F.R. Symons.R. S. 34. ACS Symposium Series. Eaton and K. World Scientiﬁc Publishing Co. London.
Inorganic and organometallic applications
26. ed. F. The Structure of Inorganic Radicals. 1970.R. Charlotte.. Berliner. Elsevier. ed.I.K. 35. Chemical and Biochemical Aspects of Electron Spin Resonance. Amsterdam. Oxford Unversity Press. Advanced EPR – Applications in Biology and Biochemistry. 1998. ed.R. Rev. 1991. John Wiley and Sons.
7) to include the possibility of many nuclei. Even more often.1. Using spin functions based on the quantum numbers mS and mI.1 Line Positions in Isotropic Spectra
As introduced in Chapter 1. the resonant ﬁelds are: X B ¼ B0 À ai mIi ð2:3Þ
i
where B0 ¼ hn/gmB represents the center of the spectrum and ai ¼ Ai/gmB is the hyperﬁne coupling constant in ﬁeld units.1 Isotropic ESR Spectra
2.
21
. DmIi ¼ 0 for all i) with the microwave photon energy (hn) [eqn (2. each with an isotropic hyperﬁne coupling constant Ai. however. since they may be derived directly from line positions in magnetic ﬁeld units. to ﬁrstorder in perturbation theory. eqn (2.1) can be used to compute energy levels. but the energies are very small so that frequency (MHz) or wavenumber (10À4 cmÀ1) units are more commonly used.1) have energy units. Equating energy diﬀerences for the allowed transitions (DmS ¼ Æ1. The spectral information is contained in the parameters. isotropic coupling constants are given in units of magnetic ﬁeld strength.CHAPTER 2
Isotropic ESR Spectra of Organic Radicals
2. The unit usually employed for a is gauss (G) (or oersteds). although SI purists sometimes use millitesla (1 mT ¼ 10 G). mIi Þ ¼ hn ð2:2Þ
As will be derived in Chapter 3.2)] the resonant magnetic ﬁelds can be predicted. g. mIi Þ À Eðms ¼ À1=2. the isotropic gfactor. ESR spectra of radicals in liquid solution are usually interpreted in terms of a spin Hamiltonian: X ^ ^ ^ ^ Ai I i Á S ð2:1Þ H S ¼ mB gBS z þ
i
where we have generalized eqn (1. and the set of values of Ai. The coupling constants in eqn (2. Eðms ¼ 1=2.
1 the spectrum of which is shown in Figure 2. a quintet of quintets as expected for hyperﬁne coupling to two sets of four equivalent protons. The spectrum consists of 25 lines. as we will see in Chapter 3. 7). Each proton splits the electronic energy levels in two. some of the multiplet lines appear at exactly the same ﬁeld position.22
Chapter 2
Conversions from units of gauss into frequency or wavenumber units involves the gvalue: Ai ðMHzÞ ¼ 2:8025ðg=ge Þai ðGÞ Ai ð10À4 cmÀ1 Þ ¼ 0:93480 ðg=ge Þai ðGÞ ð2:4aÞ ð2:4bÞ
Note that coupling constants in 10À4 cmÀ1 are comparable in magnitude to those expressed in gauss. the components of an anisotropic hyperﬁne coupling matrix (Chapter 4) should be given in frequency or wavenumber units unless the gmatrix is virtually isotropic. a multiplet pattern gives an important clue to the identity of a radical. etc. 4. Just as in NMR. the probability that three of the four protons are spin up or spin down is 4/16 and the probability that two of the four protons are spin up or spin down is 6/16. The degeneracies of the levels shown in Figure 2. it should be since this is just the set of binomial coeﬃcients for (a þ b)4 ¼ a4 þ 4a3b þ 6a2b2 þ 4ab3 þ b4. 3. in the naphthalene anion radical.1. there can also be considerable variation in line width within a spectrum.
2. 6. analogous to the spin–spin coupling multiplets of NMR spectra. Equivalent spin1/2 nuclei such as 1H. resulting in variations in component intensity. Although the units used for isotropic hyperﬁne coupling constants is largely a matter of taste. For example. 19F. i.. When several magnetically equivalent nuclei are present in a radical. the splitting is the same for each proton. are ‘‘degenerate’’. 1 : 3 : 3 : 1 for three. 1 : 2 : 1 for two. If this pattern sounds familiar. ESR multiplets can be much more complex when the electron interacts with several highspin nuclei. 5. as shown on the right side of Figure 2. Thus. for example. ESR spectra are simpler to understand than NMR spectra in that secondorder eﬀects normally do not alter the intensities of components. or 31P result in multiplets with intensities given by binomial coeﬃcients (1 : 1 for one nucleus. We can generate these
.). 8) and four b protons (positions 2. the probability that all four protons are spin up or spin down is (1/2)4 ¼ 1/16. 1 : 4 : 6 : 4 : 1 for four. ﬁve equally spaced energy level values result.2
Hyperﬁne Coupling Patterns
Nuclear hyperﬁne coupling results in a multiline ESR spectrum.1 can be obtained by the following line of reasoning: Assuming that the probability that any one proton is spin up is identical to the probability that it is spin down. on the other hand. Since the a protons are equivalent.2.e.1. and. One of the ﬁrst aromatic organic radical anions studied by ESR spectroscopy was the naphthalene anion radical. there are four a (positions 1. Similarly.
1 simulated using hyperﬁne couplings given in Table 2. A second surprise was the appearance of hyperﬁne coupling to the two sets of four equivalent protons. It was a surprise that such an oddelectron species would be stable.
coeﬃcients for other numbers of equivalent protons by the familiar Pascal’s triangle (Figure 2. See Section 2. Note that the degeneracies of the levels are not shown. Naphthalene and other aromatic hydrocarbons can be reduced by one electron to produce the anion radical. The reduction can be carried out with sodium in an ether solvent or electrochemically in a polar aprotic solvent.1. Louis. which is similar in structure to Figure 2.1 Splitting pattern for one of the electron spin energy levels coupled to four equivalent protons.
α
β
3340
3345
3350 3355 Magnetic Field/Gauss
3360
Figure 2. but in the absence of air or other oxidants. [C10H8]À is stable virtually indeﬁnitely.1. The odd electron was presumed (correctly) to occupy a p* molecular orbital
.2 ESR spectrum of the naphthalene anion radical.2) provided several surprises when Samuel Weissman and his associates1 ﬁrst obtained it in the early 1950s at Washington University in St.5 for details. The naphthalene anion radical spectrum (Figure 2.Isotropic ESR Spectra of Organic Radicals
23
Figure 2.11).
This ratio is remarkably close to the ratio of spin densities at the a and b positions.)
aH (G) 3. Things get a little more complicated when a spin 1 nucleus like 14N is added to the picture.087 0.46 3.08 3.00 2.56 rp 0.75 1.
.09 2. surprise was the ratio of the magnitudes of the two couplings.75 rp 0. 2.79 2.096 0.3 ESR spectrum of the potassium salt of pyrazine radical anion. simulated using hyperﬁne couplings from ref. whose spectrum is shown in Figure 2. Like that of the naphthalene anion radical. A third.62.136 0.069 0.80.74 1. 2.192 Compound Pyrene (Py) Perylene (P) Biphenylene (Bi) aH (G) 4.21 5. This result led to Huckel MO theory being used extensively in the semiquantitative interpre¨ tation of ESR spectra of aromatic hydrocarbon anion and cation radicals.57 5.167 0. the relative intensities of the hyperﬁne lines arising from the pyrazine anion radical.027 0. an apparent impossibility for p electrons that have a nodal plane at the position of an attached proton.013 0.3. for example.79 G ¼ 2.087
Compound Benzene (B) C8H8 (C) Naphthalene (N) Anthracene (A)
N
N
3325
3330
3335 3340 3345 3350 Magnetic Field/Gauss
3355
3360
3365
Figure 2.125 0. 11. Consider.53 0.01 1. pleasant. 5. The mechanism of coupling (discussed below) requires ‘‘contact’’ between the unpaired electron and the proton.083 0.75 3.000 0. the spectrum observed for the pyrazine anion radical2 consists of 25 wellresolved Table 2.24
Chapter 2
(MO) with the protons in the nodal plane.01 G/1.047 0. (Data from ref.108 0.181 0.1 Hyperﬁne parameters and spin densities for aromatic radical anions. but the same technique works again for the determination of the relative intensities of the ESR lines.09 0.1). predicted by simple Huckel MO theory for an electron placed in the lowest ¨ unoccupied MO (LUMO) of naphthalene (see Table 2.
The relative intensities of all 25 lines are shown at the bottom of the ﬁgure in the same order in which they are appear in the observed spectrum in Figure 2. the degeneracies of some multiplet lines are lifted when secondorder shifts are included. Section 2. giving ﬁve lines with relative intensities of 1 : 2 : 3 : 2 : 1.3 appear almost identical and are only distinguishable by careful examination of the intensities of the lines.4. The intensities for pyrazine are derived in Figure 2.5 for details of the method. i.e. The pattern arising from the two equivalent 14N nuclei that have the larger hyperﬁne splitting are ﬁrst derived.Isotropic ESR Spectra of Organic Radicals
one spin1 nitrogen: 1 1 1 1 1 1 1 1 3 1
25
1 1 2
1 1
two spin1 nitrogens:
1
2
H
N
H
H
N
H
1 plus four spin1/2 protons:
4
6 2
4 8
1 12 3
8 12
2 18 2 12 8 3 12 1 8 4 2 6 4 4 1 1
Relative intensities:
1
4
6 2 4 8 1 12 3 8 12 2 18 2 12 8 3 12 1 8 4 2 6
Figure 2. In this ‘‘coupled representation’’. aN/aH. the spectra in Figures 2. In fact.3) describes line positions correctly for spectra with small hyperﬁne coupling to two or more nuclei provided that the nuclei are not magnetically equivalent. assuming that the ratio of hyperﬁne couplings. then the nuclear spins should be described in terms of the total nuclear spin quantum numbers I and mI rather than the individual Ii and mIi. When two or more nuclei are completely equivalent.
lines. The method used is the same as implied by the energy level splittings shown in Figure 2.4
Derivation of the relative intensities of the 25 ESR lines of the pyrazine anion radical.2 and 2. See Section 2.. corresponding to the four equivalent protons. Each of these is then split into ﬁve lines with relative intensities of 1 : 4 : 6 : 4 : 1. is nearly the same as aa/ab for naphthalene anion radical.3
Secondorder Splittings
Equation (2. This can lead to extra lines and/or asymmetric line shapes.2.
2. produced by
. CH3d. both instantaneously equivalent and equivalent over a time average.1.5 gives a fuller explanation of this procedure. The eﬀect was ﬁrst observed in the spectrum of the methyl radical.1.
eqn (2. Figure 2.3) becomes:4 i o Xn Â Ãh B ¼ B0 À ai mIi þ a2 =2B Ii ðIi þ 1Þ À m2i þ . 3 and eqn (2. the line position information can be summarized by the spin Hamiltonian parameters. Since the thirdorder corrections involve cross terms among coupling constants. g and ai.3) and secondorder corrections must be made.
2. . However. as expected from eqn (2. six hyperﬁne lines are observed under conditions of high resolution (Figure 2. to second order in perturbation theory. For example. in principle they permit determination of the relative signs of the coupling constants. by Fessenden and Schuler. These parameters can be extracted from spectra by a linear leastsquares ﬁt of experimental line positions to eqn (2. Solving the spin Hamiltonian. in the example of Figure 2. the shift depends on mI.5). wellresolved spectra may warrant inclusion of third.5 However. eqn (2.5. Thus.5).
highenergy electron irradiation of liquid methane. Simulated using hyperﬁne splitting from ref.5). CH3d.5 ESR spectrum of the methyl radical.or higherorder corrections.3). CH3d (note discontinuities in magnetic ﬁeld axis).5 shows the ESR spectrum of methyl radical.3 The three equivalent protons lead to a nondegenerate nuclear spin state with I ¼ 3/2 (mI ¼ Æ3/2.26
Chapter 2
3333 3334 3335
3356 3357 3358
3379 3380 3381
3402 3403 3404
Magnetic Field/Gauss
Figure 2. but. Notice that all lines are shifted to low ﬁeld.4
Spin Hamiltonian Parameters from Spectra
Once a hyperﬁne pattern has been recognized. . ð2:5Þ i I
i
The magnitude of the secondorder corrections is often quite signiﬁcant. relative to the ﬁrstorder spectrum. for highspin nuclei and/or large couplings.1). Æ1/2) and a twofold degenerate state with I ¼ 1/2 (mI ¼ Æ1/2). Failure to account for the secondorder terms in an analysis results in a signiﬁcant error in both the gvalue and in the hyperﬁne coupling constants For large Ii and ai. one soon ﬁnds that the lines are not evenly spaced as predicted by eqn (2. the thirdorder corrections
.1.
0 G (based on CH3). CH3d.7 shows a correlation plot of observed hyperﬁne splitting versus the spin density calculated from Huckel MO theory. and QC0 C ¼ À13:9 G.7 Secondorder eﬀects on hyperﬁne structure in organometallic compounds are discussed in Chapter 3. and the methyl radical. but this is apparently symmetry forbidden for organic pradicals where the H nuclei lie in the nodal plane. CH CC C The ﬁrst term refers to the spin density in the porbital of the carbon in question and the second term from porbital spin density on atoms attached to that carbon. they diﬀer by about a factor of 6. the proton hyperﬁne coupling constants were found to be 3. The H 1s orbital is part of the sbond orbital and the C 2pz part of the singlyoccupied p* MO (SOMO).5 G (based on [C6H6]À).7 G based on a valencebond theoretical calculation by McConnell. As shown in Figure 2. It also correctly predicts the ¨ negative sign of aH for protons attached to p systems. the C 2pz orbital has zero probability at the H nucleus. QC ¼ þ19:5 G. The interaction arises in a slightly more complicated way: ‘‘spin polarization’’. necessarily leaving an oppositely oriented spin near the H nucleus. this approach works pretty well.
.0 G. there is a tendency for the unpaired spin in the SOMO to polarize the pair of electrons in the bonding MO such that the spins in the overlap region are parallel.6. In the overlap region of these two MOs.9 Their results are summarized by eqn (2. CH or À23.8 Karplus and Fraenkel went on to analyze the origins of 13C couplings in aromatic radicals. An isotropic hyperﬁne coupling to H can only arise through the socalled Fermi contact interaction of the unpaired electron at the position of the H nucleus.7 G.
2. Since the carbon atom of CH3 has a spin density corresponding to one unpaired electron and the benzene anion carries an electron spin density of 1/6.75 and 23. 1 mG. respectively. but there is signiﬁcant overlap of the C 2pz and C–H s bond orbitals.e. the two results suggest that the proton coupling to an electron in a porbital is proportional to the spin density on the adjacent carbon atom: aH ¼ QH rp CH C ð2:6Þ
where the parameter QH ¼ 23. 22. Figure 2. For aromatic hydrocarbon radical anions. [C6H6]Àd.7): ! X X C C a ¼ SC þ QCXi rp þ QC i C rp i ð2:7Þ C X X
i i
where Sc ¼ À12. i.2 Interpretation of Isotropic Parameters
When ESR spectra were obtained for the benzene anion radical.Isotropic ESR Spectra of Organic Radicals
27
amount to ca. Secondorder corrections are also required for accurate determinations of gfactors of radicals with very narrow lines6 or in properly accounting for line positions in radicals exhibiting chemical exchange eﬀects. QC 0 ¼ þ14:4 G.
computed spin density from Huckel MO theory.7
Correlation plot of observed 1H coupling constant vs.1 0. Positions with very small spin density. according to Huckel MO theory.05 0.1 for identiﬁcation of points.
6 A 5 4 aH/Gauss P 3 Bi 2 1 0 Py P Bi 0 0. tend to be polarized such that a negative spin density is ¨ found at that site. The reason for this is related to the mechanism for aromatic proton coupling.28
Chapter 2
σorbital H C
π*orbital
Figure
2.2 A N Py A Py B C N
P
Figure 2. See Table 2. only a tiny fraction of the electron density near the H nucleus is excess spindown.15 Hückel MO spin density 0. The alternation of the sign of aH between positions was
. ¨
Examination of Figure 2.6 Schematic representation of spin polarization of a C–H sorbital by electron spin in a p* orbital.7 shows that the largest fractional errors are for the smallest spin densities. Thus the zero spin density predicted for one position in pyrene leads to a coupling of a bit more than 1 G as a result of this polarization mechanism. Note that the sign of a coupling constant is not an experimental result in ordinary ESR. Note that the polarization eﬀect is far from complete.
but considerable caution should be exercised in interpreting the gvalue or nuclear hyperﬁne coupling constants. such as the bprotons (methyl group) of the ethyl radical. Most ESR studies of organic radicals were carried out in the 1950s and 1960s. have been estimated. They provided important tests of early developments in valence theory. or to measure its concentration by integration of the spectrum and comparison with an appropriate standard. This was interpreted early on as evidence for ‘‘hyperconjugation’’ which delocalizes the unpaired electron onto the adjacent proton.11 Applications of hyperﬁne splittings to structure determination are discussed in many of the texts and monographs referenced in Chapter 1. and (iii) a contribution from spin–orbit coupling. is on somewhat ﬁrmer ground. the second is normally negative. The ﬁrst contribution is positive (for a positive nuclear magnetic moment). analogous to QH . The results of these early studies are nicely summarized in a review by Bowers. Thus isotropic ESR spectra should be used to characterize a radical by means of the hyperﬁne coupling pattern. An isotropic hyperﬁne coupling has three contributions: (i) Fermi contact interaction between the nuclear spin and electron spin density in the valenceshell sorbital.3 Line Widths in Isotropic Spectra
2.
2. The dependence of aH on geometry had also proven to be a very useful tool for studying the conformations of radicals. (ii) Fermi contact interaction between the nuclear spin and spin density in innershell sorbitals arising from spin polarization by unpaired electron density in valenceshell p. In contrast. The interpretation of isotropic gvalues is even less useful and subject to misinterpretation. In this case aH varies with the dihedral angle between the C–H bond and the adjacent porbital containing the unpaired electron. proton couplings in such radicals result solely from spin polarization and thus are proportional to the polarizing spin density. all three contributions are usually signiﬁcant for organometallic radicals. they are of use only in a more CH complete analysis based on the results of a solidstate ESR study. Such an assumption is usually unwarranted. and the third may be of either sign.1 Incomplete Averaging of Anisotropies
Careful examination of the isotropic ESR spectra of organic radicals may reveal variations in line widths from one hyperﬁne component to another.6) for metal or ligand hyperﬁne couplings in spectra of organometallic radicals. As we will see in Chapter 4. The situation with regard to protons that are attached to atoms adjacent to p centers. Because direct involvement of hydrogen 1s character in the SOMO of an organic pradical is symmetryforbidden and spin– orbit coupling is negligible in carbonbased radicals. gmatrices are often diﬃcult to interpret reliably. Although there are a few cases where polarization constants.10 It is sometimes assumed that there is a relation analogous to eqn (2.Isotropic ESR Spectra of Organic Radicals
29
conﬁrmed by performing NMR measurements of the Knight shift of pyrene anion radical in the solid state. to study its dynamical properties through line width eﬀects.3. Such eﬀects
.or dorbitals. CH3CH2d.
ð2:8Þ i
i
Much of the width arises from incomplete averaging of anisotropies in the gand hyperﬁne matrices (Chapter 3). the latter increasing with 1/tR and thus with increasing temperature.> and tR. where o0 is the (angular) frequency of the microwaves. they increase with decreasing temperature. for example. by a relationship introduced by Debye:15 tR ¼ 4pr3 Z=3kT ¼ Vh Z=kT ð2:10Þ
ESR line widths can provide a measure of the eﬀective size of a paramagnetic molecule or ion – useful information when there is a suspicion that a radical has polymerized.13 If the g. the linewidth parameters bi and gi can be used to compute the rotational correlation time. For Xband microwaves o0 ¼ 6Â1010 rad sÀ1.w There are several linewidth contributions. have been used to probe the motional freedom of biological macromolecules. which do not depend on mi. included in a0. For radicals with axial symmetry the parameters of eqn (2.1–10 G
w
Equations (2.8) depend on Dg ¼ g – g>.
2.2
Rates of Fluxionality from Line Widths
ESR line widths are also sensitive to processes that modulate the gvalue or hyperﬁne coupling constants or limit the lifetime of the electron spin state. a useful measure of freedom of motion.. . i. The widths can often be expressed as a power series in the nuclear spin quantum numbers: XÀ Á Width ¼ a þ bi mi þ gi m2 þ . Vh. DAi ¼ Ai. .and hyperﬁne anisotropies are known from analysis of a solidstate spectrum. most rapidly tumbling.9a–c) assume that the product tRo041.
.9a–c) should therefore apply to all but the smallest.10) this corresponds to the tumbling time for a molecule with a molecular weight of about 40 in water at room temperature. so these relationships hold for tR41. since ESR line widths are typically on the order of 0. Equations (2. radicals in media of normal viscosity. Z.6Â10À11 s. tR. These include magnetic ﬁeld inhomogeneity and the spin–rotation interaction.e. and the solution viscosity. However. – Ai.30
Chapter 2
are enhanced in solvents of high viscosity or at low temperatures. the rotational correlation time:12 a / a0 þ ðDggBÞ2 tR b / DggBDAtR g / ðDAÞ2 tR ð2:9aÞ ð2:9bÞ ð2:9cÞ
Since these terms are proportional to tR. Using eqn (2. The eﬀects are closely analogous to those observed in NMR spectra of dynamical systems.3. Line widths in ESR spectra of nitroxide spin labels.14 Since tR is related to the molecular hydrodynamic volume. These and other linewidth eﬀects have been studied in some detail and are discussed elsewhere.
The top spectrum is identical to that of the 4.8 ESR spectra resulting from the reduction of PhCN (bottom) and pFPhCN (top). 16.)
. (Spectra were simulated using hyperﬁne couplings with permission from ref.4 0 dicyanobiphenyl anion radical. copyright (1963) American Chemical Society.Isotropic ESR Spectra of Organic Radicals
F
31
eF CN F CN CN
2
CN
NC
CN
eNC CN NC CN
3360
3365
3370
3375
3380
3385
3360
3365
3370
3375
3380
3385
Magnetic Field/Gauss
Figure 2.
9.2.8 shows spectra that result when benzonitrile and 4ﬂuorobenzonitrile are reduced electrochemically.1. and attachment of a proton (presumably from the solvent N. In fact. indicating the presence of four equivalent protons.3pentacyanopropene dianion radical.9 (Top) ESR spectrum of 1.3. This suggests that FÀ is lost from the initially formed radical anion and that the resulting neutral radical dimerizes.
2. 1. Such processes are discussed in detail in Chapter 5.4 0 dicyanobiphenyl.3.5 Analysis of Isotropic ESR Spectra
The analysis of an isotropic ESR spectrum is relatively straightforward if a systematic approach is used. 16. (bottom) ESR spectrum of the dianion radical of 1.3tetracyanopropene. but it can be a frustrating experience for a
3340 3342 3344 3346 3348 3350 3352 3354 3356 3358 3360 Magnetic Field/Gauss
Figure
2. Oneelectron reduction produces the expected dianion radical.2. which results from loss of CNÀ from position 2. the spectrum of which is shown in Figure 2. but the spectrum resulting from the reduction of FC6H4CN is not that of the expected radical anion. the product of radical dimerization.1. Careful examination of the spectrum shows 1 : 4 : 6 : 4 : 1 quintets at either end of the spectrum.9. Figure 2. Another example from the same paper16 involves the cyanocarbon anion.1.3pentacyanopropene.32
Chapter 2
(0.
2.4 Organic Radical Reactions
Sometimes the ESR spectrum obtained when an organic molecule undergoes oneelectron oxidation or reduction is not of the expected cation or anion radical.16 The spectrum from the reduction of benzonitrile is indeed that of [C6H5CN]Àd. Spectra were simulated using hyperﬁne couplings with permission from ref.3–30 MHz).3.3. However.
. rate processes that give observable increases in line widths must be fast.3.1.Ndimethylformamide) to produce the 1.tetracyanopropene dianion radical the spectrum of which is also shown in Figure 2. the observed spectrum is identical to that obtained by oneelectron reduction of 4. further reduction leads to a very diﬀerent spectrum. copyright (1963) American Chemical Society.
there should be (2 Â 1/2 þ 1)6 ¼ 26 ¼ 64 nuclear spin states. If some of the nuclei are expected to be equivalent. then there must be an even number of lines.e. however..2
What do you Expect to See?
(A) If the radical was produced by a chemically straightforward procedure. In noisy spectra. I ¼ 1/2. then all lines should have the same intensity (barring accidental overlap). If there are equivalent sets of nuclei.e. If there is one group of four equivalent protons and another group of two. 3/2. if the model structure has six protons (I ¼ 1/2).).e. (D) How many lines are there? Count them carefully. Thus.). . then it is as if you had one spin2 nucleus and one spin1 nucleus and you expect (2 Â 2 þ 1)(2 Â 1þ1) ¼ 15 lines. if followed with care. in principle.5. Given below is a series of steps which. (B) Is there a central line? If there is no central line. . you usually have some idea of the identity of the radical. some of the spin states will be degenerate (to ﬁrstorder in perturbation theory). If all the nuclei are nonequivalent. cause the spectrum to appear unsymmetric. for the above examples.Isotropic ESR Spectra of Organic Radicals
33
beginner. 2.1
Preliminary Examination of the Spectrum
(A) Check to see if the spectrum is symmetric in line positions and relative intensities. should be present if your guess is correct? (B) How many lines are expected from this model? The total number of nuclear spin states is (2I1 þ 1) Â (2I2 þ 1) Â (2I3 þ 1) . then the number of lines will be less than the number of spin states.5. if the six protons are in three groups of two. then the expected ratio is the product of the largesttosmallest ratios of the various multiplets. etc. but in such a case line positions would still be at least approximately symmetrically distributed about the center. then most likely there are two or more radical species. watching for evidence of unresolved features (shoulders or bumps). How many and what kinds of magnetic nuclei. which suggests an odd number of halfinteger nuclei (i.. three groups of two spin1/2 nuclei gives a 1 : 2 : 1 triplet of 1 : 2 : 1 triplets of 1 : 2 : 1 triplets. i. Thus. i. nuclei with spins.
2. Thus. . etc. Variation of line widths with mI may. (E) What is the ratio of the amplitudes of the most intense to least intense lines in the spectrum?
2. it is as if you had three spin1 nuclei and you expect (2 Â 1 þ 1)3 ¼ 33 ¼ 27 distinct lines. If it is not. which would cause splitting of any center line arising from an even number of halfintegral nuclei or any number of nuclei with integer spin (I ¼ 1. (C) Are the outermost lines visible above the noise? Observation of the lines in the ‘‘wings’’ of the spectrum is often crucial to successful interpretation of ESR spectra since less overlap of lines may occur there. thus the maximum
. it may be diﬃcult to account for all the expected lines in the wings. will result in a successful interpretation for most reasonably wellresolved spectra.
either the model is wrong or there is more than one radical contributing to the spectrum.17 Let us see if we can ﬁgure out which is which. When we examine the wings of the spectrum.5. one pair of equivalent ring protons. (B) Compare the expected and observed largesttosmallest intensity ratios. Accidental degeneracies (or a mI2 linewidth dependence) may cause the observed ratio to be bigger than expected. with this assumption. The pxylene anion radical is easiest. we see a 1 : 6 : 15 : 20 : 15 : 6 set of doublets. If the expected and observed numbers are equal. That leaves spectrum (c) for mxylene.4
An Example
We can apply these rules to the three spectra shown in Figure 2. Spectrum (a) shows a 1 : 2 : 1 triplet of multiplets. No other isomer has a quintet. The outermost lines have intensity ratios: 1 : 2 : 6 : 1 : 12 : 12À. If there are more lines than expected. you are in luck – the analysis should be easy.
2. Thus we expect to see a 1 : 4 : 6 : 4 : 1 quintet of 1 : 6 : 15 : 20 : 15 : 6 : 1 septets.
. so we can assign (b) to the pxylene radical anion. The bigger the discrepancy between expected and observed numbers of lines.2 gives the coupling constants for the three xylene anion radical isomers. We expect two unique protons. or just poor resolution. but. Spectrum (b) shows the quintet clearly. Apparently only one pair of ring protons has a signiﬁcant coupling. six equivalent methyl protons. spectrum (a) is consistent with oxylene. the model is probably wrong. there may be accidental superpositions. The radicals in the ﬁgure are anion radicals of the three isomers of xylene (dimethylbenzene). again. The oxylene anion radical should have two pairs of ring protons and. the less deﬁnitive the analysis will be. then the maximum intensity ratio will be 6 Â 2 ¼ 12. go on to a detailed analysis. which could well be the expected quintets. and six equivalent methyl protons. suggesting that the methyl and ring proton couplings are comparable.34
Chapter 2
intensity ratio would be 2 Â 2 Â 2 ¼ 8. It is rarer (but not unknown) to ﬁnd a smaller ratio than expected. If you see fewer lines than expected (the most common case!). Table 2. If we have a 1 : 4 : 6 : 4 : 1 quintet of 1 : 2 : 1 triplets.
2.5. If there are gross inconsistencies. but the CH3 proton coupling is poorly resolved. We expect to have four equivalent ring protons and six equivalent methyl protons.10. small amplitude lines buried under large ones. (C) Is the presence or absence of a central line consistent with the odd or even number of expected lines? (D) If the model still seems plausible after these tests.3
Are the Gross Features of the Spectrum Consistent with the Model?
(A) Compare the number of observed lines with the number expected.
26
H
Number 1 ring H 1 ring H 2 ring H 6 Me H
Number 4 ring H 6 Me H
aH (G) 5. Spectrum (a) was simulated with permission using hyperﬁne parameters from Ref.34 0. 17a.85 7. 17b.uk). m. copyright (1961) Taylor and Francis (www.93 1. and pxylene radical anions (see text for assignment of spectra).81 2.tandk.Isotropic ESR Spectra of Organic Radicals
H3C CH3 H3C H H H H oxylene H H mxylene H H CH 3 pxylene H H CH3 H CH3 H
35
(a)
(b)
(c)
3330
3335
3340 3345 Magnetic Field/Gauss
3350
3355
Figure 2.46 2.10
.
Table 2. copyright (1964) American Institute of Physics.72 1.00
H
pXylene a (G) 6. spectra (b) and (c) were simulated with permission using hyperﬁne parameters from ref.2
oXylene Number 2 ring H 2 ring H 6 Me H
Hyperﬁne parameters for xylene radical anions17
mXylene a (G) 6.10 ESR spectra of o. co.
and. for the coupling constants (a1. etc. Assign quantum numbers to the second line. centimeters. If convenient.36
Chapter 2
2. (B) The highest and lowest ﬁeld lines will always have theoretical intensities of 1 and will have maximum positive and negative values for all the quantum numbers. The ‘‘amplitude’’ is the diﬀerence in height between the maximum and minimum.
. It does not really matter whether you start with positive quantum numbers at the highﬁeld or lowﬁeld end of the spectrum – the signs of the coupling constants are indeterminant – but. Continue in this way through the spectrum. If you have started into a multiplet. you can then predict the positions and intensities of the remaining lines of the multiplet. a1. you can work from either end of the spectrum or even from both at once.5
Detailed Analysis for Determination of Parameters
(A) Measure the positions and amplitudes of all the lines in the spectrum and list them in order in a table (a spreadsheet program is convenient for this purpose). you can compute the sum of the remaining couplings. Enter these assignments in the table. etc. you can predict the positions and intensities of the remaining lines of this multiplet. A position discrepancy larger than your estimated measurement accuracy. and for the theoretical intensity (degeneracy) of each peak. (C) The spacing between the ﬁrst and second lines will be the smallest coupling constant. Find them and enter the quantum number assignments and new estimates of a1 and a2 in the table. The intensity ratio of these two lines will usually indicate the multiplet to which the coupling constant corresponds. (E) If a2 corresponds to a multiplet. Small discrepancies may arise through measurement errors or because of overlapped lines. if you have found two coupling constants and know the total extent of the spectrum. From the ratio of its amplitude to that of the end line. if this is inconvenient. we assume that we are working from the lowﬁeld end. you can usually ﬁgure out which multiplet this coupling constant corresponds to.e. This will correspond to the next larger coupling constant a2.. m2. ﬁnd the ﬁrst unassigned line as you move upﬁeld.). Enter in the table. Thus. This line will also correspond to the ﬁrst line in a second a1 multiplet. pretend that the coupling constants are positive). A welldeﬁned measure of ‘‘position’’ in a complex spectrum is the xaxis point halfway between the maximum and minimum of the ﬁrstderivative line. knowing a1. or recorder chart boxes measured from an arbitrary zero. Assign the quantum numbers and compute the coupling constant a2. may signal a misassignment or inconsistency with the model. the positions of the remaining lines can be predicted and located. use arbitrary units such as inches.) for each of the line positions. for consistency.5. a2. also provide headings for the quantum numbers (m1. In the following. Notice that the spacing between the lowest and highest ﬁeld lines is equal to the sum of the coupling constants. let us agree to assign positive quantum numbers to the lowest ﬁeld line (i. In your table. compute a1 and enter these numbers in the table. Find them and enter the quantum numbers and new estimates of a1 in the table. measure the line positions in gauss. From this point on. however. (D) After ﬁnding all the lines of the ﬁrst a1 multiplet.
Isotropic ESR Spectra of Organic Radicals
37
(F) When all the lines have been assigned. the multiplet intensity ratios are simply the binomial 2 coeﬃcients found most easily from Pascal’s triangle (Figure 2. If an arbitrary position scale was used. For example.6
Computation of Multiplet Intensity Ratios
For sets of spin1 nuclei. however. and a more compact notation is preferable.5. Fortunately. this method can be used as the basis for a computer algorithm. one can construct a splitting diagram. For higher spin nuclei. but it is impossibly tedious to do by hand in all but the simplest cases.
2. computer programs and spreadsheet templates for carrying out a multiparameter leastsquares analysis are now readily available. we have:
1 1 1 1 1 1 Add: 1 2 3 1 1 2 1 1
This procedure can be extended to three spin1 nuclei by thinking of each line of a 1 : 1 : 1 triplet split into a 1 : 2 : 3 : 2 : 1 quintet:
1 2 1 3 2 1 Add: 1 3 6 2 3 2 7 1 2 3 6 1 2 3 1 1
For really complicated situations. for two spin1 nuclei:
This procedure quickly gets out of hand. Leastsquares ﬁtting of line positions to quantum numbers gives better statistics than this method.
. Thus if we think of each of the three lines resulting from coupling to the ﬁrst spin1 nucleus split into a 1 : 1 : 1 triplet. average all the measured values of each coupling constant and compute the standard deviation. convert the coupling constants into gauss.11).
P1 ¼ 0.5. The most common example is that of carbon where 1.989)nÀ1. 1.1% of naturally occurring carbon is 13C with I ¼ 1/2 and 98.011)2(0. so that the total probability of having two 13Cs is n(nÀ1)(0.989)n.031. Similarly. The probability that one speciﬁc carbon is 13C and the other nÀ1 12C will be (0. P3 ¼ 0. The probability that all n will be 12C is (0. benzene anion radical).00003. In practice.936.011. n.011)2(0.
2.0017..g. 13Cs out of n equivalent carbon atoms are just the terms in the expansion of the expression: ð0:989 þ 0:011Þn ¼ð0:989Þn þ nð0:989ÞnÀ1 ð0:011Þ þ nðn À 1Þ=2ð0:989ÞnÀ2 ð0:011Þ2 þ Á Á Á For example.989)nÀ2. the probabilities are: P0 ¼ 0. corresponding to splitting of the center line by a single 13C (the intensity is distributed between the two resonances). etc. the probability that two speciﬁc carbons are 13C is (0.062. Since the two 13Cs are indistinguishable.38
1 1 1 1 1 1 1 1 1 8 7 28 6 5 15 21 56 4 10 20 35 70 3 6 10 15 35 56 21 28 2 3 4 5 6 7 8 1 1 1 1 1 1 1 1
Chapter 2
Figure 2.011)(0. More complex patterns can arise from atoms with a higher fraction of magnetic isotopes. the coeﬃcient must be divided by 2 to avoid double counting.936 and a pair of satellites with intensity 0. we would probably see only the central line with intensity 0. Thus. in a radical with six equivalent carbon atoms (e. etc. Now suppose that there are n equivalent carbon atoms in the radical.989)nÀ1.989)nÀ2. On reﬂection. the probability that any given carbon is 13C is 0. 2. however.7
Multiplet Patterns due to Isotopomers
We often encounter cases where an element has one or more isotopes with a nuclear spin and one or more isotopes with zero spin. Consider the case of a radical species with three equivalent
. are just binomial coeﬃcients and that the probabilities of ﬁnding 0.9% is 12C with I ¼ 0. n(nÀ1)/2. we see that the coeﬃcients 1. we see that the total probability of ﬁnding one 13C is n(0.11 Pascal’s triangle for the determination of binomial coeﬃcients. P2 ¼ 0. but. since any one of the n carbons could be 13C..011)(0. in a carboncontaining radical. but any one of the n carbons could be 13C and any of the nÀ1 remaining carbons could also be 13C.
This is often adequate.12). P1 ¼ 0.227/2 2 .
platinum atoms.8% in natural abundance. seven hyperﬁne components are expected with positions and intensities (Figure 2. and a 1 : 3 : 3 : 1 quartet. Other Pt isotopes have I ¼ 0.237 . all seven lines might be observable.Isotropic ESR Spectra of Organic Radicals
. in a singlet.
2.444/2 1 a Pt
. The relative intensities of the sevenline pattern then is expected to be approximately: 1 : 12 : 49 : 84 : 49 : 12 : 1. a 1 : 2 : 1 triplet.12 Predicted hyperﬁne pattern for a radical with three equivalent Pt atoms. The intensity labels indicate the normalized probabilities of the corresponding lines which are derived from the isotope probabilities. but when coupling constants are large (greater than about 20 gauss) or when line widths are small (so that line positions can be very accurately measured) secondorder eﬀects become important. 195Pt has I ¼ 1/2 and is 33. With luck.444. but they also provide a ﬁngerprint. P2 ¼ 0.290
39
0
.227/4
.5. Patterns like this are more complicated to analyze. a doublet.290.039x3/8 3 . the coupling constant will be the same in each pattern and there will be some superpositions. identifying the number of equivalent nuclei involved in the multiplet pattern. As we see from
. These four isotopomers will result. However. respectively.057 . Consequently. since the platinum atoms are assumed to be equivalent. Numbers on the left refer to the number of 195Pt atoms in the radical. and P3 ¼ 0.039. Thus the probabilities are: P0 ¼ 0.005 . especially for spectra of organic radicals.404 SUM .8
Secondorder Shifts in Line Positions
Our analysis thus far has assumed that solution of the spin Hamiltonian to ﬁrst order in perturbation theory will suﬃce.227.039/8
Figure 2.
the shifts are much more noticeable. when a ¼ 100 G. for B0 ¼ 3200 G. we have.3
mI 2 1 1 0 0 0 –1 –1 –2 I 2 2 1 2 1 0 2 1 2
Secondorder shifts for two I ¼ 1 nuclei
[I(I þ 1) – mI2]/2 1 5/2 1/2 3 1 0 5/2 1/2 1 B (G) 3159. When a is larger. but there are two ways of getting mI ¼ Æ1 (either of the nuclei can have mIi ¼ 0 and the other mIi ¼ Æ1). thus there must be a I ¼ 0 state with mI ¼ 0. When a spectrum results from coupling to several equivalent nuclei with couplings large enough to warrant secondorder corrections.62 3199. Thus. The secondorder corrections then depend on the total I rather than the individual Iis. for example. Thus. a ¼ 20 G: Table 2.44 and 101. 19. There is only one way of getting mI ¼ Æ2 (each nucleus has mIi ¼ Æ1).94 3199. This is a rather small eﬀect and would require careful measurements to notice. for example. a spin1 nucleus with a ¼ 20 G and B0 ¼ 3200 G will have lines at: B ¼ 3200 À 20ðþ1Þ À ð0:0625Þð2 À 1Þ ¼ 3179:94 G B ¼ 3200 À 20ð0Þ À ð0:0625Þð2 À 0Þ ¼ 3199:88 G B ¼ 3200 À 20ðÀ1Þ À ð0:0625Þð2 À 1Þ ¼ 3219:94 G so that the line spacings are.00 3219. a much more easily noticeable discrepancy.06 G. Æ1.5) for a single nucleus with nuclear spin I. Substituting into eqn (2.94 and 20. the situation becomes somewhat more complicated since we must then think of the quantum numbers I and mI corresponding to the sum of the quantum numbers for a full set of equivalent nuclei.69 3219. so there must also be a I ¼ 1 state with mI ¼ Æ1 and 0.12 G. the degeneracies of the 1 : 2 : 3 : 2 : 1 multiplet pattern are lifted and a total of (3)2 ¼ 9 individual lines would be observed.94 3239. and 0. the line spacings become 98. and the center line is shifted to low ﬁeld (relative to the ﬁrstorder result) by 0. to secondorder. for example. Summarizing these results in Table 2.69 3179. the hyperﬁne line positions are given to second order in perturbation theory by: B ¼ B0 À amI À Ã a2 Â IðI þ 1Þ À m2 I 2B0 ð2:11Þ
Thus.40
Chapter 2
eqn (2.88
.3. There are three ways of getting a mI ¼ 0 state (either of the nuclei can have mIi ¼ Æ1 and the other mIi ¼ Æ1 or both nuclei can have mIi ¼ 0).56 G. two equivalent spin1 nuclei will give rise to an I ¼ 2 state with mI ¼ Æ2.88 3179.11). respectively.88 3200.
Electron Nuclear Double Resonance (ENDOR). See text for explanation. At equilibrium. respectively [see Section 1. Indeed.6 Related Techniques (ENDOR)
The instrumental method described in Chapter 1 and illustrated in this chapter proves to be adequate for studying many of the samples to which chemists and biologists wish to apply ESR.δ/2 1 .13 Example of an ENDOR experiment for a single spin1/2 nucleus. the lower and upper pairs of levels have relative populations indicated by 1þd and 1Àd. the capabilities for which. In Figure 2. saturation of one of the ESR transitions initially equalizes the populations of the km and mm levels. ﬁeldswept Xband ESR still provides the most convenient. for those who may need to extend their studies to more advanced methods Appendix 2 lists several of the most useful techniques and gives references to recent reviews and relevant papers that should serve as an entry into the still developing literature on advanced ESR. CW.13 the vertical arrow on the left refers to the orientation of the electron spin and the one on the right the nuclear spin. The technique works by partially saturating an ESR transition at a ﬁxed ﬁeld while simultaneously sweeping the NMR frequency through nuclear spin transitions.
↑↑
1+δ 1+δ
1+δ 1
1+δ 1
Microwave absorption spectrum
νN – A/2
νN
νN + A/2
. and commonly used. In the example. including simplifying the hyperﬁne patterns of complex radicals in solution. leading to a decrease in intensity of that transition. or suspected. ENDOR was ﬁrst developed in the mid1950s18 by George Feher as a means of determining unresolved nuclear hyperﬁne coupling parameters in solids.13 for a single spin1/2 nucleus. is an increase in absorption at the microwave frequency when an appropriate NMR transition is irradiated.δ/2
↑↓ ↑↑
1δ 1δ Saturate one ESR transition
↓↑
Figure 2. even if more advanced techniques turn out to be required. may be added as a relatively minor modiﬁcation to commercial CW ESR spectrometers. The result. which is illustrated in Figure 2.Isotropic ESR Spectra of Organic Radicals
41
2. method for preliminary examination of samples that are known. It has since found many other applications. to possess interesting paramagnetic properties.2 (Chapter 1)]. This chapter concludes with a brief description of one advanced technique. Nevertheless. If the NMR transition from
Equilibrium populations Saturated populations 1δ 1
Perturbed populations
1 . unlike pulsed methods.
J. Pake. Fessenden and R. 633. the example given here corresponds to the case where the hyperﬁne frequency.42
Chapter 2
Table 2. Zandstra and T. Karplus and G. A ¼ A/h.D. 5. J. An early example of an application of ENDOR to extract hyperﬁne couplings from a very complex. G. J. K.. 4275. Chem.249(2)
mm to mk is then saturated by irradiation at the frequency of the corresponding NMR transition at. (b) R. 1967. 1965. Weissman. 1956. 2704. 10. The interested reader should consult the relevant references in Appendix 2 or one of the more general books on ESR.M.R. 4. 1961. Stone and A.H. Phys. S. Phys. (a) G. Chem. Schuler. 2149.K. Maki.W. J.W. M. 3410. Chem. Chem. 43. Chem.E..G. 1962. R.. Bolton. J.. Phys. 35. R. J. Schuler. J. Chem. Phys. 2227.4. Allendoerfer and P. is less than the nuclear Zeeman frequency... E. the resulting population change will lead to an increase in the population diﬀerence between the km and mm levels and increased microwave absorption at the ESR frequency.K. Chem. km to kk. Fraenkel. 1. M.962(2) mH 0. at frequency nN – A/2.I. Notably.W. J. 37. Chem. Karplus and G. 6. 1965. 1635. Townsend.99(1) oH 1. R. Paul and G.
. J. J.H. Fessenden and R.E. 9. P. ed.W. 1960 33. J. H.H. J. Phys. Phys.764(2) Me 0. 8.J.. 21.. Waugh. 317. Chem. nN. For the case where A/2 4 nN. McConnell. will also lead to an increase in intensity of the same saturated ESR transition.4
Position a (G)
Hyperﬁne parameters for bis( pmethoxyphenyl)nitroxide19
N 9. 2147. J. R. Phys. Bowers. 7. 24. Fessenden. 46. 2. 47. in this example. 1953.R. 1963. 42. 39.. Rieger. Anderson. 1967.W. Tuttle. 1591.E. J. 3. in Advances in Magnetic Resonance. Phys. poorly resolved ESR spectrum is that of Allendoefer and Englemann19 who studied a solution of bis(pmethoxyphenyl)nitroxide and obtained the results shown in Table 2.H. interpretation of the ENDOR frequencies is somewhat modiﬁed. Lawler. 747. Phys. Chem. 39. nN þ A/2. S. The ENDOR spectrum is generated by varying the NMR frequency and observing the corresponding increase in intensity of the saturated ESR transition. 1312. The ﬁgures in parentheses indicate the level of precision. 1965. 11.
References
1. Phys. Fraenkel. 1963. Irradiation of the other NMR transition. Phys.K.. Chem. Fraenkel.. M.
114. H. Plenum Press..R. 1972.H. Salikhov. 17. 16. N. ch. ed. (b) R. Chem. Eaton and K. J. 13. J. Kivelson. 1971. 1094. Kivelson. Bloembergen. W. Soc.Isotropic ESR Spectra of Organic Radicals
43
12.. 4. 20. E.H. 2455.K. 1948. Electron Spin Relaxation in Liquids.
. Am. (a) D. J.M. Rev. Carrington. Pound. 169.. 1963. McConnell. G. Phys.S.. 579.T. (b) J. Rieger.5. Mol. Singapore. Phys.R. Bernal. J. Atkins and D.. Rev. Chem.V.W. 683. Chem. Chem. 1961. 15.. Phys. L. J. G. 85. Fraenkel. 1219. Mol. Phys. 44. S. Phys. 18. 1964. 19. (a) J. for example. 154. in Foundations of Modern EPR. R. 1960. 44. Phys. Purcell and R. Kivelson. 73.M. ed. Chem.D. Feher. 497. Reinmuth and G. See. 33. Bolton and A. G. Phys. Phys.. 1966. 1959. I. Englemann.R.M.. Wilson and D. 41. Allendoerfer and J. 14. World Scientiﬁc. New York. P. 1966.H. Eaton. Bolton. 1998. Muus and P. Atkins. 569.. (c) P.W.
1). mI i ¼ gmB B Sz þ A Sz Iz jms . increasing with increasing B. Clearly. Ai. To understand the origin of this eﬀect we must take a closer look at the solutions to Eqn. the eight hyperﬁne lines (I ¼ 7/2 for 51V) have diﬀerent widths. but careful examination also shows that the line spacing varies. Nonetheless. there are some bits of information and applications that come uniquely from isotropic spectra. however. it is reasonable to express the eigenfunctions in terms of the electron and nuclear spin quantum numbers: mS.1) for the case of an unpaired electron interacting with a single nucleus. B. Consider. (3. Applying to this function only the two terms in the Hamiltonian operator that involve the zdirection of the ﬁeld B we get: À Á ^ ^^ ^ Hs jms . and we discuss those aspects in this chapter. the ESR spectrum of vanadium(IV) in acidic aqueous solution (Figure 3. Given the Hamiltonian eqn (3.CHAPTER 3
Isotropic Spectra of Organometallic Radicals
Most of the information content from ESR spectra of organometallic radicals and coordination complexes comes from dilute singlecrystal spectra or frozen solution spectra.1 Secondorder Eﬀects on Line Positions
As illustrated in Chapter 2. ESR spectra of radicals in liquid solution can usually be interpreted in terms of the spin Hamiltonian: X ^ ^ Hs ¼ mB gB Á S þ Ai^i Á S ð3:1Þ I ^
i
This is a simpliﬁed Hamiltonian that ignores the direct interaction of any nuclear spins with the applied ﬁeld. where the species is [VO(H2O)5]21. for example. mI i ð3:2Þ ¼ðms gmB B þ Ams mI Þjms . to most transition metal nuclei.5) and (2. Because of the larger coupling. This will lead us to a derivation of eqns (2.11) of Chapter 2. it is often necessary to use secondorder perturbation theory to accurately determine the isotropic parameters g and A.1).
3. mI i
44
.mIi.
equally spaced. energy is: E ð0Þ ¼ ms gmB B þ ms mI A or. Æ1/2. Sy. in this case there is no ﬁrstorder perturbation. First. mI ¼ Æ7/2.005 M vanadium(IV) in 1 M NaClO4.
. despite the fact that the notation E(0) properly acknowledges that both terms are included in the ‘‘zerothorder’’ energy. This nomenclature will also be used here. Æ5/2.)
Thus the zerothorder. Strictly speaking. Now consider the application of secondorder perturbation theory. Æ3/2.5 GHz
45
3000
3200
3400
3600
3800
B/Gauss
Figure 3.Isotropic Spectra of Organometallic Radicals
v = 9. pH 2. or unperturbed. copyright (1975) American Chemical Society. (Reproduced with permission from ref. the hyperﬁne energy is usually referred to as a ‘‘ﬁrstorder’’ correction to the Zeeman energy. in terms of their raising and lowering combinations: SÆ ¼ Sx Æ iSy 1 S x ¼ ðS þ þ S À Þ 2 i Sy ¼ À ðSþ þ SÀ Þ 2 IÆ ¼ Ix Æ iIy 1 Ix ¼ ðIþ þ IÀ Þ 2 i Iy ¼ À ðIþ þ IÀ Þ 2
w
By convention. Ix and Iy. we need to represent the xÀy components of the spins. since mS ¼ Æ1/2: E ð0Þ ¼ Æ1 gmB B Æ 1 mI A 2 2 With I ¼ 7/2.1 Isotropic ESR spectrum of 0. Sx. This ‘‘ﬁrstorder’’ solutionw would of course lead to eight lines. 3.0.
Including them would lead to energies that are 2 thirdorder or higher in powers of A. mI ¼ 2 2 and the secondorder energy corrections are: À Á A2 ½IðI þ 1Þ À mI ðmI þ 1Þ Á À ÁÁ E ð2Þ 1. mI À 1 ¼ 2 Thus the matrix elements are: A pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ^ À1. mI Apﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 1 IðI þ 1Þ À mI ðmI þ 1ÞÀ2.46
Chapter 3
Thus the hyperﬁne term of the spin Hamiltonian becomes: À Á AS Á I ¼ A Sx Ix þ Sy Iy þ Sz Iz ¼ ASz Iz þ 1AðSþ IÀ þ SÀ Iþ Þ 2 Operating on the spin functions with the extra hyperﬁne operator then gives: 1 ^ ^ ^ ^ 1 2A Sþ IÀ þ SÀ Iþ 2. mI À E ð0Þ þ1. mI ¼ 2 4gmB B þ 1Að1 þ 2mI Þ 2
À Á A2 ½IðI þ 1Þ À mI ðmI À 1Þ E ð2Þ À1. mI À E ð0Þ À1. mI þ 1 ¼ gmB B þ 1Að1 þ 2mI Þ 2 2 2 À Á À Á E ð0Þ À1. mI 2 2 Apﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ1 IðI þ 1Þ À mI ðmI À 1Þ2. mI ¼ 2 4gmB B þ 1Að1 À 2mI Þ 2 We can neglect the hyperﬁne terms in the denominators without serious error since 4gmB B ) 1Að1 Æ 2mI Þ. mI À 1 ¼ ÀgmB B þ 1Að1 À 2mI Þ 2 2 2 À Á A2 ½IðI þ 1Þ À mI ðmI þ 1Þ E ð2Þ 1. mI À E ð0Þ À1. mI ¼ À ð0Þ À 1 2 4 E þ2. mI þ 1 2 À Á A2 ½IðI þ 1Þ À mI ðmI À 1Þ Á À ÁÁ E ð2Þ À1. making it necessary to use higher order perturbation theory. mI À 1HÀ1. mI À E ð0Þ þ1. mI ¼ À ð0Þ À 1 2 4 E À2. mI þ 1Hþ1. mI À 1 2 or since the zerothorder energy diﬀerences are: À Á À Á E ð0Þ þ1. mI ¼ IðI þ 1Þ À mI ðmI þ 1Þ 2 2 2 pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 1 A ^ IðI þ 1Þ À mI ðmI À 1Þ þ2.1
. mI þ 1 ¼ 2 1 ^ I ^ I A Sþ^À þ SÀ^þ À1.
correct to second order in A.3 Exact positionsa 3041. g ¼ 1. and v ¼ 9.1).8 3144.0 3863.2b–d As described in Chapter 2. 1storder spacing of lines would be 116.0 3281. 3.964.7 3253.8 3514.8 113.1 124.7 3604.Isotropic Spectra of Organometallic Radicals
47
21
Table 3. Purcell and Pound2a on nuclear spin relaxation and was formulated in a useful way for EPR by Daniel Kivelson and coworkers.4 3397.1.8 3857.9 3484.4 3729.
The transition energies.
3. they showed that the widths. are: DE ¼ gmB B þ mI A & ' A ½IðI þ 1Þ À mI ðmI þ 1Þ A2 ½IðI þ 1Þ À mI ðmI À 1Þ þ À À 4gmB B 4gmB B Â Ã 2 A IðI þ 1Þ À m2 I ¼ gmB B þ mI A þ 2gmB B &
2
'
and the resonant ﬁelds are:
Â Ã a2 IðI þ 1Þ À m2 I B ¼ B 0 À a mI À 2B hn gmB A gmB
ð3:3Þ
where B0 ¼ a¼
For [VO(H2O)5]21 (Fig.6 3747.1 G using methods given in ref.4 117. in units of Hz.4 127.2 3483. The predicted line positions (in Gauss) are given in Table 3.8 3367.1
mI 7/2 5/2 3/2 1/2 À1/2 À3/2 À5/2 À7/2
a b
Higher order eﬀects on line positions in [VO(H2O)5]
Firstorder positions 3048.4 Secondorder positions 3040. This issue was implicit in the classic work of Bloembergen.9 108.2 Understanding the Variation in Line Widths
The widths of the lines in Fig.6 3146.6 3165.1 vary because the anisotropies of g and A (to be discussed in Chapter 4) are not completely averaged out when the molecule or ion tumbles in solution.1 3728.3 3605. 1 by iterative solution of the BreitRabi equation.6
Secondorder spacingb 103.2 3254.2 3630.5000 GHz.4 121.4 G.4 3856. 3. a ¼ 116.5
Calculated to nearest 0. can often be written as a power series in mI with terms up to secondorder (a thirdorder term is sometimes signiﬁcant): width ðmI Þ ¼ a þ bmI þ gm2 I ð3:4Þ
.4 G.5 3366.
however.3c)
. for example. is seen in Fig. eﬀectively ruling out the presence of ESRactive polymers in solution ˚ (Figure 3.4 A in acid solution and 2. For example. .7 ˚ in basic solution.3 Previous work on the system. line width measurements may be used to evaluate tR as a function of Z and T.4 The slopes correspond to r values of 3.2). .1) and strong base. 3. allowing r to be estimated.. 5).g. The rotational correlation time therefore provides a measure of the ‘‘hydrodynamic volume’’ of the tumbling molecule or ion. The conclusion is that vanadium(IV) is monomeric in dilute A basic solutions. If values of Dg and DA can be determined independently. and related to the spectrum of methyl radical discussed in Chapter 2.1. dilute aqueous solutions of vanadium(IV) show an eightline ESR spectrum in both acid (Fig. and k is a correction factor to account for deviations from sphericity. was also consistent with the formation of polymers such as V3O72À or V4O92À or even V18O4212À at high pH.48
Chapter 3
where a / a0 þ ðDggBÞ2 tR b / DggBDA tR g / ðDAÞ2 tR ð3:5aÞ ð3:5bÞ ð3:5cÞ
and tR is the rotational correlation time. The experimental spectrum of [PhCCo3(CO)9]À (Figure 3. formed by removal of three protons from the species whose spectrum is shown in Fig. Comparison of values of tR from the basic solution line widths with those obtained from spectra of acidic solutions containing VO(H2O)521 showed that the hydrodynamic volume of the aquo cation is actually about twice that of the basic solution species.
3.3a).3 for the isotropic ESR spectrum of [PhCCo3(CO)9]À (ref.) A likely structure in base would be [VO(OH)3(H2O)2]À. Thus. The situation is complicated by the fact that magnetically equivalent nuclei with I 4 1/2 give less familiar multiplet intensity ratios.3 Puzzling Line Shapes
An example of secondorder splitting. 3. usually expressed in an equation ﬁrst introduced by Debye:2 tR ¼ 4pr3 k Z 3 kT ð3:6Þ
where Z is the viscosity of the solution. suggesting a monomeric unit under all conditions. r is the radius of the (spherical) particle. 3. (51V has I ¼ 7/2 so more than eight lines would be observed if more than one 51V was coupled to the unpaired electron. from solid state spectra (Chapter 4). . e. in addition to the secondorder shifts described above for a single nucleus. three equivalent 59Co nuclei (I ¼ 7/2) give (to ﬁrstorder) 22 lines with intensity ratios 1:3:6:10:15:21:28:36:42:46:48:48:46 . (Figure 3.
and secondorder shifts of 97a2/4B. The observed line shapes. and 17a2/4B. respectively.Isotropic Spectra of Organometallic Radicals
80 70 60 50
τR/ps
49
VO2+ in 1.0 M NaOH 10 0 1 2 3 4 5 6 7 8 9
η/T(10−6 kg /msK)
Figure 3. vs. and all but the weakest outermost pair of lines is visible. and 3. and 17/2. degeneracies of 1. The combined spin angular momentum of three equivalent spin7/2 59Co nuclei is properly described in terms of 11 Jstates with J ranging from 21/2 to 1/2. The shifts are too small compared with the line width to be resolved. 4. for example. 55a2/4B. as shown in Figure 3. which produces secondderivative spectra.
does indeed exhibit a spectrum with about the expected relative intensities.2 Plots of tR. (3.3) substituting J for I. are very unsymmetrical and correspond to a marked deviation from a Lorentzian shape. The line positions through secondorder are then calculated using Eqn. The mJ ¼ 17/2 feature.0 M NaClO4
40 30 20 V(IV) in 1. 19/2. but they lead to an asymmetric absorption line envelope with apparent broadening on the lowﬁeld side. the resolution was good enough that several of the individual species could be separately
. glycolic acid. Data from ref.3(b) and in the experimental spectrum (Figure 3. Incorporating secondorder splittings into the spectrum as follows solves the puzzle. has three components with J ¼ 21/2. thiolactic acid.4 Use of ESR Spectra to Determine Formation Constants
Reeder and Rieger6 used ESR spectra to identify complex ions and to estimate formation constants for aqueous oxovanadium(IV) complexes with lactic acid.5
3. determined from ESR line widths. 2. Z/T for [VO(H2O)5]21 and for [VO(OH)3(H2O)2]À. however. Through the use of secondharmonic detection.3c). and thioglycolic acid.
reprinted from ref. for VO21/lactic acid mixtures at
. H2A. 7.) (c) Isotropic ESR spectrum of [PhCCo3(CO)9]À in THF solution at 401C.50
Chapter 3
Figure 3. For example. the coupling constant decreasing with each additional ligand coordinated. The key requirement was that the 51V coupling constant varied with the nature of the ligands. copyright (1979) American Chemical Society. the equilibria were: VO2þ þ HAÀ ! VOðHAÞþ VOðHAÞþ þ HAÀ ! VOðHAÞ2 VOðHAÞþ ! VOðAÞ þ Hþ VOðHAÞ2 ! VOðAÞðHAÞÀ þ Hþ VOðAÞðHAÞÀ ! VOðAÞ2À þ Hþ 2 where HAÀ indicates ionization of the carboxyl group but retention of the hydroxyl or sulfhydryl proton. (Experimental spectrum.)
detectable.3
‘‘Stick spectrum’’ showing hyperﬁne pattern for coupling to three equivalent 59Co nuclei (I ¼ 7/2) computed to (a) ﬁrstorder and (b) secondorder in perturbation theory. For each acid. 5 with permission. (Adapted from ref.
. 1966. (b) D.
References
1. Chem. 1966. Amsterdam. 2895. 1971. Peake. Phys. Robinson and J. 2704. 6. Bloembergen. Schuler.4 Secondderivative ESR spectrum of the lowﬁeld line of the VO21 octet for a 1:4 mole ratio mixture of VO21 and lactic acid at pH 3. ch. 1965. Phys..H. 1973..R. vol. Electron Spin Resonance: Theory and Practice.. 1979.12. R. Chem. 1000.. Inorg. Chem. (d) P.. (a) N. Chem. M. J.. 3. 44. J. 22. Phys.H.M. Iannuzzi. Reeder and P. Peak I was assigned to the superposition of peaks resulting from VO(H2O)521 and VO(HA)(H2O)31.. ed. 1976.H. Rieger. (c) R. (Reprinted with permission from ref. Phys. Trogler. E. 8.5 and 251C. Rieger.H.M. Pound. Journal of Organometallic Chemistry Library. Inc. J. and peak III to the superposition of VO(A)(H2O)3 and VO(A)(HA)(H2O)–. Chem. 4. 1990. 14. B. Atkins and D. Rev. the lowﬁeld hyperﬁne line (in secondderivative presentation) appears as shown in Figure 3. Purcell and R. 2.C. B.5. New York. 169.)
pH 3. W. section 3. 1094. For solutions through fourthorder see also: R.
. 679. Inorg. Phys. 43.M. M. peak II was assigned to VO(HA)2(H2O). Fessenden and R. Rieger. Kivelson. J. Simpson. 5. Rieger. Kubiak and P.V. Chem. 7.H. Chem. Phys. a reasonable set of formation constants was deduced for each of the complexes. N. J. Chem. Kivelson. 541. Atherton.P. Kivelson. copyright (1971) American Chemical Society. Wilson and D. Inorg. P. 1960.4. Elsevier. Rieger.M. 33. C. 1948 73. W. P.. Iannuzzi and P. 18. 154.Isotropic Spectra of Organometallic Radicals
51
I
II
III
Figure 3.H. 6. 80.H. John Wiley and Sons. From the variation in peak heights with pH.M. 10. 44. The separation between peaks is approximately 25 Gauss..W. 1258. 1975.
52
. 3). R ¼ ptolyl (ref. The nuclear quadrupole and nuclear Zeeman interactions add two more terms
w
This chapter has been taken in part from material ﬁrst published by Elsevier Publishing in The Journal of Organometallic Chemistry Library Series. 1). ch. In general. When quadrupolar eﬀects are important.CHAPTER 4w
Anisotropic ESR Spectra
4. In species with two or more unpaired electrons.1 Introduction
The anisotropies that lead to line broadening in isotropic ESR spectra inﬂuence solidstate spectra more directly. quadrupolar eﬀects are observable only in very wellresolved spectra or in spectra of radicals with nuclei having small magnetic moments and large quadrupole moments. Organometallic Radical Processes. In practice. Trogler. Nuclear quadrupole interactions introduce line shifts and forbidden transitions in spectra of radicals with nuclei having I 4 1/2. Other nuclei for which quadrupolar eﬀects might be expected include 151Eu/153Eu. The spectra of [Ir(CN)6]3À (ref. it is usually necessary to take account of the nuclear Zeeman interaction as well. a coordinate system can be found – the gmatrix principal axes – in which g is diagonal. 175 Lu. 189Os. [Ir(CN)5Cl]4À and [Ir(CN)4Cl2]4À (ref. We conﬁne our attention here to radicals with one unpaired electron (S ¼ 1/2) but will address the S 4 1/2 problem in Chapter 6.1). ed. 1990. and [Ir2(CO)2(PPh3)2(mRNNNR)2]1. g and Ai are 3 Â 3 matrices representing the anisotropic Zeeman and nuclear hyperﬁne interactions. we say that their principal axes are coincident. D. If g and Ai are diagonal in the same coordinate system. The material from that chapter has been somewhat modiﬁed and is republished here with permission from the publisher. 8. and 197Au. Figures reprinted from the chapter are designated by an ‘‘[E]’’. 155Gd and 157Gd. 181Ta. 2). The most extreme case of a small magnetic moment to quadrupole moment ratio is that of 191Ir and 193Ir. show easily recognizable quadrupolar eﬀects. a ﬁne structure term must be added to the spin Hamiltonian to represent electron spin–spin interactions. Accordingly a more complex spin Hamiltonian is required to interpret such spectra: X ~ ~ ^ ~ H s ¼ mB~ Á g Á S þ ð4:1Þ B Ii Á Ai Á S
i
In eqn (4.
Furthermore. When a radical is oriented such that the magnetic ﬁeld direction is located by the polar and azimuthal angles. When the hyperﬁne matrix components are large. relative to the gmatrix principal axes. Highly symmetrical solvent molecules like benzene tend to freeze with semicrystalline regions. When the matrix principal axes are coincident. electrochemical reduction or oxidation. or photochemical methods.Anisotropic ESR Spectra
53
to the spin Hamiltonian. chemical reactions.
4. Various techniques then can be used to generate radicals. The goal is a completely random
. we will deal with quadrupole eﬀects in Chapter 7 and conﬁne our attention here to nuclei for which quadrupolar eﬀects can be neglected. y and j. only one of the lijk in eqn (4. these result in downﬁeld shifts. While many organometallic radicals have been studied in this way. secondorder terms4 must be added to eqn (4. to ﬁrst order in perturbation theory. the resonant ﬁeld is given. Radicals can be produced in situ by UVor girradiation of a suitable precursor in a crystalline or glassy matrix.5 it is often easier to obtain solidstate ESR spectra by freezing the liquid solution in which the radical is formed. powders. the radical is studied under conditions more closely approximating those in which its reaction chemistry is known.2 Solidstate ESR Spectra
So long as they are dilute (to avoid line broadening from intermolecular spin exchange). by:4 B ¼ B0 À where B0 ¼ hn gmB ð4:3Þ X Ai mi
i
gmB
ð4:2Þ
g2 ¼ g2 cos2 j þ g2 sin2 j sin2 y þ g2 cos2 y x y z
2 2 2 A2 ¼ A2 Six þ A2 Siy þ A2 Siz i ix iy iz
ð4:4Þ ð4:5Þ Ã ð4:6Þ
gx sin y cos j lixk þ gy sin y cos j liyk þ gz cos y lizk Sik ¼ g
Â
and the lijk are direction cosines indicating the orientation of the kth principal axis of the ith hyperﬁne matrix relative to the jth gmatrix principal axis. glasses or frozen solutions.2). e. proportional to mi2. Not all solvents give a satisfactory frozen solution. Since these terms considerably complicate an already complex situation.g. which leads to undesirable orientation of the solute molecules..6) will be nonzero. radicals can be studied in the solid state as solutes in single crystals.
many of which were produced by photolysis or radiolysis of single crystals of the precursor molecules. K2. A good example of the application of the information that can be obtained from such singlecrystal studies is the analysis of the spectrum of NO3.1
Spectra of Dilute Single Crystals
Spectra of radicals in a dilute single crystal are obtained for various orientations. usually with the ﬁeld perpendicular to one of the crystal axes. In their 1967 book. elements of which are the direction cosines of the gmatrix principal axes relative to the crystal axes. The gmatrix is then diagonalized to obtain the principal values and the transformation matrix.and hyperﬁnematrix principal axes are not necessarily the same as the crystal axes. if spectra are obtained for various orientations in the crystal xyplane. eventually enough data are obtained to determine the six independent values of gij (g is a symmetric matrix so that gij ¼ gji). for example.
4. Exploration of another crystal plane gives another set of Ks that depend on other combinations of the gij. Atkins and Symons6a summarized much of the early ESR work on small inorganic radicals. Toluene gives a good random glass. An analogous treatment of the eﬀective hyperﬁne coupling constants leads to the principal values of the A2matrix and the orientation of its principal axes in the crystal coordinate system. usually will have oﬀdiagonal elements.2. j can be analyzed to determine K1. as do mixtures of CH2Cl2 and C2H4Cl2 or of tetrahydrofuran and CH2Cl2. Since the g. Thus.1 summarizes the results. written in the crystal axis system.54
Chapter 4
arrangement of solutes.6b Table 4. and K3. the matrices.
. Each spectrum usually can be analyzed as if they were isotropic to obtain an eﬀective gvalue and hyperﬁne coupling constants for that orientation. produced by girradiation of KNO3 crystals by Livingston and Zeldes. the eﬀective gvalue is: À Á2 À Á2 g2 ¼ gxx cos j þ gyx sin j þ gxy cos j þ gyy sin j j ð4:7Þ À Á2 þ gxz cos j þ gyz sin j or g2 ¼ K1 þ K2 cos 2j þ K3 sin 2j j where K1 ¼ 1 2 gxx þ g2 þ g2 þ g2 þ 2g2 yy xz yz xy 2 1 2 gxx À g2 þ g2 À g2 yy xz yz 2 ð4:9aÞ ð4:8Þ
K2 ¼
ð4:9bÞ ð4:9cÞ
K3 ¼ gxy gxx þ gxy gyy þ gxz gyz
A sinusoidal plot of gj2 vs.
some solvents. the relatively small 14N hyperﬁne splitting is consistent with spin density localized to a large degree on the oxygen atoms.1 illustrate one of the advantages of higher frequencies over Xband ESR spectra: The various gcomponents are better separated and the spectrum is more easily analyzed. The spectra of Figure 4. Notably. ref. The interpretation of a powder spectrum thus depends on the connection of the positions of these features to the g.31 G
The data clearly show that NO3 in this medium has axial symmetry.0232 Axx ¼ 3. such as toluene. Frozen solution spectra must be taken in solvents that form good glasses. thus giving rise to more features than predicted as spectra
.46 G
6b
gzz ¼ 2. the spin density distribution would be inferred only indirectly without the ESR data.1. leading to a large value of the derivative and features that resemble positivegoing and negativegoing absorption lines. [GaCl(mquin)2] was not the perfect host and the ESR matrix axes diﬀered from the crystal axes by 11. In such samples one may often obtain not only the desired ESR parameters but also determine the orientation of the paramagnetic molecule within the host crystal. the gmatrix axes were not coincident with the Amatrix axes.71. there will be minimum and maximum resonant ﬁelds at which the absorption changes rapidly from zero. As it happened.Anisotropic ESR Spectra
55
Table 4. observable features in the spectrum of a powder correspond to molecular orientations for which the derivative is large in magnitude or changes in sign. The frozen solution spectrum (in toluene) of [VO(mquin)2] and the powder spectrum of [VO(mquin)2] in [GaCl(mquin)2] are nearly identical. being displaced in the xyplane by a ¼ 27. 2) on transition metal ions was carried out this way.
4.and Qband spectrometers.51. Dilute solutions of welloriented paramagnetic molecules can also sometimes be prepared by cocrystallization with a suitable diamagnetic host. Simulations of the powder spectra at Xand Qband are shown in Figure 4.2
Analysis of Frozen Solution Spectra
Since ESR spectra are normally recorded as the ﬁrst derivative of absorption vs. A random but homogeneous distribution of the species to be studied is required to obtain a wellresolved spectrum. While the geometry may in principle also be obtained from vibrational and rotational spectra. A good example of this approach is the work by Collison and Mabbs7 who studied [VO(mquin)2] (mquin ¼ 2methylquinolin8olate) dissolved in a single crystal of [GaCl(mquin)2]. Much of the early work by Bleaney and coworkers (Chapter 1. For any spin Hamiltonian.46 G
ESR parameters for NO3
gyy ¼ 2. Furthermore.1
gxx ¼ 2. can ‘‘crack’’ when frozen.2. using both X. Furthermore. ﬁeld. most likely with a threefold axis passing through the nitrogen atom.0031 Azz ¼ 4.and hyperﬁnematrix components. Peaks in the absorption envelope correspond to derivative sign changes and lead to features resembling isotropic derivative lines. The next section deals with the analysis of frozen solution spectra.0232 Ayy ¼ 3.
(top) 34. but it is well to keep the predicted number of features in mind when attempting to interpret the results.988
1.988
O.25 GHz (Xband). Consider the simplest case: a single resonance line without hyperﬁne structure.983 1.9 This approach led to some valuable insights but the algebra is much too complex when nonaxial hyperﬁne matrices are involved. Thawing and refreezing of the sample often corrects the problem.949 1.1
Powder spectrum of [VO(mquin)2] in [GaCl(mquin)2]: (bottom) 9.949 13250 13500
Chapter 4
1.8 Early treatments of powder patterns attempted to deal with the spatial distribution of resonant ﬁelds by analytical mathematics. 7 with permission.99 GHz (Q band).3). The resonant ﬁeld is given by eqn (4. Drago has given a list of good glassforming solvents and mixtures.56
12000 12250 12500 12750 13000 1. Features in the ﬁrst derivative spectrum correspond to discontinuities or turning points in the absorption spectrum that arise when qB/qy or qB/qj are zero: @B hn g2 À g2 z ? ¼ sin y cos y ¼ 0 @y mB g3 @B hn g2 À g2 2 x y ¼ sin y sin f cos f ¼ 0 @f mB g3 and g2 ¼ g2 cos2 j þ g2 sin2 j ? x y ð4:10aÞ
ð4:10bÞ
. (Reproduced from ref.mquin N
CH3
2500
2750
3000 3250 3500 Magnetic Field/Gauss
3750
4000
Figure 4.)
from diﬀerent regions of the sample will be overlapped.983 1. copyright (1987) Royal Society of Chemistry.
ESR parameters from ref 10. Since y and j are in the gmatrix axis system.0.0570. but the ‘‘extra lines’’ correspond to less obvious orientations.Anisotropic ESR Spectra
57
These equations have three solutions: (i) y ¼ 0. the principal values of the gmatrix are obtained from a straightforward application of eqn (4. Powder spectra with hyperﬁne structure often can be interpreted similarly with spectral features identiﬁed with orientation of the magnetic ﬁeld along one of the g. Using a ﬁrstorder theory and one hyperﬁne coupling. Three of these correspond to orientation of B along principal axes. The problem is illustrated by the resonant ﬁeld vs.2 [E] Resonant ﬁeld as a function of cos y and j for the mCu ¼ À3/2 ‘‘line’’ of the frozen solution spectrum of Cu(acac)2. (ii) y ¼ 901. 2.)
. j ¼ 0. However. and (iii) y ¼ j ¼ 901.5.0527. cos y and j surface shown in Figure 4. 19. 2. Even more extra lines may creep in when the spin Hamiltonian is treated to secondorder or when there is more than one hyperﬁne coupling. this simple situation often breaks down.10) may have up to six solutions corresponding to observable spectral features. corresponding to mCu ¼ À3/2 in the spectrum of Cu(acac)2 (g ¼ 2.10). ACu ¼ 27. This being the case. Ovchinnikov and Konstantinov10 have shown that eqns (4. (Figure reproduced from ref.2514. 13.2.and hyperﬁnematrix principal axes. observable features are expected for those ﬁelds corresponding to orientations along the principal axes of the gmatrix.
Figure 4.
11 Parameters are then adjusted and another simulation computed until the ﬁt is satisfactory (or the experimentalist loses patience with the problem).
. In addition.0 G (cosy ¼ Æ 0. j ¼ 901). Although such a ‘‘cut and try’’ approach to spectrum
Figure 4. another saddle point occurs in the yzplane at B ¼ 3371. With B along the xaxis (cosy ¼ 0. and with B along the yaxis (cosy ¼ 0. B to give an approximation to the unbroadened absorption spectrum. in this case. (a) Features corresponding to mCu ¼ À3/2.4 Â 10 cm ).5 G. correspond to oﬀaxis ﬁeld orientations and thus are ‘‘extra lines’’. The minimum resonant ﬁeld. Thus ﬁve features are expected and indeed are shown in the computersimulated spectrum of Cu(acac)2 shown in Figure 4. j ¼ 01). The most straightforward computer simulation method10 involves computation of the resonant magnetic ﬁeld using eqn (4. The ﬁeld is then divided into equal increments and the number of times the resonant ﬁeld falls between Bi and Bi11 is plotted vs. B ¼ 3290.and hyperﬁnematrix components. the two most intense features. j ¼ 901) there is a local minimum at 3325. corresponds to B along the zaxis (cosy ¼ Æ1). j ¼ 01).58
À4 À1 10
Chapter 4
193.3 [E] Computersimulated ﬁrstderivative ESR powder spectrum of Cu(acac)2. the only maximum is in the xzplane at B ¼ 3379. Since the analytical approach is so complicated. none of the features need correspond to the orientation of B along a principal axis direction. The absorption spectrum is then broadened by numerical convolution with a line shape function and diﬀerentiated to give the desired simulation. highﬁeld.459.7 G. (b) the complete spectrum. powder patterns have usually been analyzed by comparing the experimental spectrum with a computersimulated spectrum using estimates of the g. 105 values of cosy and j for each set of nuclear spin quantum numbers. The situation is more complex when the g.2) for ca.482.and hyperﬁnematrix principal axes are noncoincident (see below).2 G (cosy ¼ Æ0. the surface shows a saddle point at 3344. Interestingly.3 G (which corresponds to a maximum in the absorption spectrum and therefore appears with closely spaced maximum and minimum ﬁrstderivative features).3.
Thus. The components of the gmatrix thus diﬀer from ge to the extent that p. and saddle points. The method employs a search of the resonant ﬁeld surface in cosy. provided that the spectrum is suﬃciently wellresolved that enough features can be located and identiﬁed to provide statistically signiﬁcant estimates of the parameters. In a radical species. analysis of a complex spectrum is exceedingly tedious. ge ¼ 2. or forbitals are involved. depending on which p. Since the search procedure requires computation of B for only B100 orientations. gz will be close to the free electron value. the method is much faster than a complete simulation. The predicted locations of spectral features are then compared with the experimental values and the parameters are reﬁned using a nonlinear leastsquares method.2 shows the results of operation by li on these functions. Thus if the major metal contribution to the SOMO is dz2 .3 Interpretation of the gMatrix
The gvalue of a free electron is a scalar. zk is the spin–orbit coupling constant for atom k. k sums over atoms with unpaired electron density. when one gmatrix
.00232. z). in the usual shorthand notation for p and d orbitals: pﬃﬃﬃ hzjl x jyi ¼ i and z2 jl y jxz ¼ À 3i The orbital angular momentum operations needed to calculate integrals for other orbitals are summarized in Table 4.2. y. d. minima. relatively complex powder patterns can be analyzed. j space for maxima. The integrals hmlkini are easily computed for an electron centered on a single atom if the MOs are written as linear combinations of real p or d atomic orbitals. given estimates of the spin Hamiltonian parameters. Even with leastsquares ﬁtting. and m sums over ﬁlled and empty molecular orbitals with energy Em (E0 is the energy of the SOMO). DeGray and Rieger12 have developed a computer algorithm to locate powder pattern features in spectra. d. Using this method.Anisotropic ESR Spectra
59
analysis works reasonably well when there are small numbers of parameters. and they diﬀer from one another. for example. Accordingly.
4. however. g becomes a matrix because of the admixture of orbital angular momentum into S through spin–orbit coupling. In general. or forbital character has been incorporated. Notice that dz2 is unique among the dorbitals in that lz does not couple it to any other orbital. Table 4. the components of the gmatrix are given by: X X zk hmjl ki j0i 0l kj m gij ¼ ge dij þ 2 ð4:11Þ E0 À Em k m6¼0 where the indices i and j refer to molecular coordinate axes (x. comparison of the experimental spectrum with a highresolution computer simulation is required to check the assignments of spectral features. and lki is the icomponent orbital angular momentum operator for atom k.
4.. this contribution. the average dipolar contribution to the hyperﬁne coupling is zero. E0ÀEm. designated A. there is rarely suﬃcient information to interpret a gmatrix with complete conﬁdence. Extended Huckel ¨ MO (EHMO) calculations can be useful (but do not trust EHMO energies!). however. but one is most commonly reduced to arguments designed to show that the observed gmatrix is consistent with the interpretation placed on the hyperﬁne matrix. the energy diﬀerences.2
[E] Angular momentum operations on the real p and d orbitals
lx ly Àizi 0 ixi Àixzi Àiyzi ixyi pﬃﬃﬃ 2 2 ijx2 pﬃﬃﬃ À y i À 3ijz i 3ijxzi lz iyi Àixi 0 2ixyi À2ix2Ày2i Àixzi iyzi 0
xi yi zi x2Ày2i xyi yzi xzi z2i
0 izi Àiyi Àiyzi ixzi pﬃﬃﬃ ijx2 À y2 i þ 3ijz2 i Àixyi pﬃﬃﬃ À 3ijyzi
component is found close to the free electron value. is isotropic and has been discussed in Chapters 2 and 3. i.60
Chapter 4
Table 4. When a wellresolved and reliably assigned optical spectrum is available. it is often taken as evidence for a dz2 based SOMO.11) to estimate the contribution of the corresponding MOs to the gmatrix. Thus. for example. are known and can be used in eqn (4. since cancellation of negative and positive terms in eqn (4.e.11) could have the same eﬀect. ESR spectra of d1 vanadium(IV) complexes generally have gvalues less than ge (admixture of empty MOs) whereas d9 copper(II) complexes have gvalues greater than ge (admixture of ﬁlled MOs). (ii) The electron spin–nuclear spin magnetic dipolar interaction. Spin–orbit coupling to empty MOs (E0ÀEm o 0) gives a negative contribution to gij whereas coupling to ﬁlled MOs has the opposite eﬀect. such reasoning should be applied with caution. this contribution is almost entirely anisotropic. neglecting spin–orbit coupling. The general form of the dipolar contribution to the hyperﬁne term of the Hamiltonian is: * + S Á I 3ðS Á~ I Á~ ~ rÞð~ rÞ ~ ~ ^ H dipolar ¼ ge gN mB mN cSOMO 3 À ð4:12Þ cSOMO r r5
.4 Interpretation of the Hyperﬁne Matrix
Electron–nuclear hyperﬁne coupling arises mainly through two mechanisms: (i) The Fermi contact interaction between the nuclear spin and selectron spin density. Since the gmatrix has only three principal values and there are almost always many potentially interacting molecular orbitals.
If an atom contributes px. mB and mN are the Bohr and nuclear magnetons. Equation (4. and pz atomic orbitals to the SOMO: cx jxi þ cy jyi þ cz jzi ð4:16Þ
the total porbital spin density is (in the Huckel approximation.e.12). leaving the spins as operators.12) can then be written: ~ ^ H dipolar ¼ ~Á Ad Á S I ð4:13Þ
where Ad is the dipolar contribution to the hyperﬁne matrix. Ordinarily. the integral is combined with the constants as a parameter: P ¼ ge gN mB mN rÀ3 ð4:15Þ
P has been computed using Hartree–Fock atomic orbital wavefunctions and can be found in several published tabulations14–17 and in Appendix 1.. i. the integration over the angular variables is quite straightforward.14 The integral over r. dipolar coupling of a nuclear spin with electron spin density on another atom is usually negligible. and the total hyperﬁne coupling is: A ¼ AE þ Ad ð4:14Þ
(E is the unit matrix). In evaluating the matrix element of eqn (4. neglecting ¨ overlap): rp ¼ c 2 þ c 2 þ c 2 x y z and the dipolar contribution to the hyperﬁne matrix can be written: 2 ðAd Þij ¼ Plij 5 where the lij are: lxx ¼ 2c2 À c2 À c2 x y z lyy ¼ 2c2 À c2 À c2 y x z lzz ¼ 2c2 À c2 À c2 z x y lij ¼ À3ci cj ði 6¼ jÞ ð4:19aÞ ð4:19bÞ ð4:19cÞ ð4:19dÞ ð4:18Þ ð4:17Þ
. however. and the matrix element is evaluated by integration over the spatial coordinates.Anisotropic ESR Spectra
61
where ge and gN are the electron and nuclear gvalues. requires a good atomic orbital wavefunction. Because of the hrÀ3i dependence of P. py.
the orbital shape is unchanged. if an atom contributes d atomic orbitals to the SOMO. When the porbitals are written as hybrids.19) can therefore be combined in matrix notation to write the dipolar hyperﬁne matrix for porbitals as: 0 1 2 0 0 ð4:20Þ Ad ¼ ð2=5ÞPrp @ 0 À1 0 A 0 0 À1 where rp.14 Equations (4. which reﬂect the spatial orientation of the hybrid porbital. ð4:21Þ cz2 jz2 þ cyz jyz þ cxz jxzi þ cx2 Ày2 jx2 À y2 þ cxy jxy the total dorbital spin density is (in the Huckel approximation): ¨ À Á2 À Á2 À Á2 rd ¼ ðcz2 Þ2 þ cyz þðcxz Þ2 þ cx2 Ày2 þ cxy and the dipolar contribution to the hyperﬁne matrix is:18 2 ðAd Þij ¼ Plij 7 where P is given by eqn (4. the spin density.17–4.and zaxes (Euler angle b ¼ 451). for example. Similarly. a phybrid with cx ¼ cz and cy ¼ 0 corresponds to a porbital with the major axis in the xzplane and halfway between the x. Thus. but the principal axes of the hyperﬁne matrix. diﬀer from those in which the SOMO was formulated. is deﬁned by eqn (4. The porbital axis corresponds to the positive principal value of the matrix.15) and the lij are: À Á2 lxx ¼ À ðcz2 Þ2 À2 cyz þðcxz Þ2 À Á2 À Á2 pﬃﬃﬃ À Á þ cx2 Ày2 þ cxy À2 3ðcz2 Þ cx2 Ày2 À Á2 lyy ¼ À ðcz2 Þ2 À2 cyz þðcxz Þ2 À Á2 À Á2 pﬃﬃﬃ À Á þ cx2 Ày2 þ cxy þ2 3ðcz2 Þ cx2 Ày2 À Á2 À Á2 À Á2 lzz ¼ 2ðcz2 Þ2 þ cyz þðcxz Þ2 À2 cx2 Ày2 À2 cxy pﬃﬃﬃ À Á À Á lxy ¼ À2 3ðcz2 Þ cxy þ 3 cyz ðcxz Þ lyz ¼ lxz ¼ pﬃﬃﬃ À Á À Á À ÁÀ Á 3ðcz2 Þ cyz þ 3ðcxz Þ cxy À 3 cyz cx2 Ày2 pﬃﬃﬃ À ÁÀ Á À Á 3ðcz2 Þðcxz Þ þ 3 cyz cxy þ 3ðcxz Þ cx2 Ày2 ð4:23Þ
ð4:22Þ
ð4:24aÞ
ð4:24bÞ
ð4:24cÞ ð4:24dÞ ð4:24eÞ ð4:24fÞ
.17).62
Chapter 4
The factor of 2/5 and the weighting of the orbital coeﬃcients is determined by the angular factors.
particularly when the radical has enough symmetry to restrict possible hybridization.22 with eqn (4. it must not depend on the choice of coordinate system. Hybrid combinations of dyz. if (for example) A ¼ Az and AxÀAyEP(gxÀgy).or dorbital.and thirdrow transition metals appear. A departure from axial symmetry may be due to spin–orbit coupling eﬀects.23 Fortunately. it is usually caused by dorbital hybridization. (dppe ¼ Ph2PCH2CH2PPh2).19–21 In radicals with two or more transition metals. i. A major part of the problem is that the solution must be gaugeinvariant.Anisotropic ESR Spectra
63
The dipolar contribution to the hyperﬁne matrix for a pure dorbital is: 0 1 2 0 0 ð4:25Þ Ad ¼ ðÆ2ÞPrd @ 0 À1 0 A 7 0 0 À1 where the positive sign applies for dz2 and the negative sign to the other four orbitals. the problem will become more urgent. dxz. but it has received only cursory examination with regard to spin–orbit contributions to hyperﬁne matrices.11) as the result. it is to be hoped that some future theoretician will deem the problem worthy of attention. usually with diﬀerent principal axes. [Mn(dppe)2(CO)(CNBu)]21. and dxy or dx2 Ày2 and dxy give a dorbital of the same shape and the same dipolar matrix. in a CH2Cl2/THF glass is shown in Figure 4. The procedure is best illustrated by examples. If the departure from axial symmetry is larger. Other hybrid orbitals are generally of diﬀerent shape. though the principal axes in general are diﬀerent from the axes in which the SOMO was formulated. As organometallic radicals with second.1 A Lowspin Manganese(II) Complex
The spectrum of the lowspin manganese(II) complex.
4. Spin–orbit coupling perturbs these results. The AO composition of the SOMO can often be deduced from the dipolar hyperﬁne matrix.e. This problem was addressed in the context of spin–orbit coupling corrections to the gmatrix. obtained from leastsquares
.5 Organometallic Examples
4. reﬂected by diﬀerent principal values of the dipolar matrix. Calculation of the spin–orbit coupling corrections is fairly straightforward for mononuclear complexes where the SOMO is composed mainly of dorbitals from a single metal. polynuclear radicals containing ﬁrstrow transition metals generally have gmatrix components suﬃciently close to ge that the problem can be ignored. Thus an axial hyperﬁne matrix can usually be interpreted in terms of coupling to a SOMO composed of a single p.4(a)..5.24 The spin Hamiltonian parameters. These can be neglected only when the gmatrix anisotropy is small. the spin–orbit coupling calculation is seriously nontrivial. adding terms to the diagonal matrix components on the order of P(giÀge).
3 26. copyright (1987) Royal Society of Chemistry).2 25. dyz (b2). where the representations refer to C2v. the SOMO is expected to be one of the ‘‘t2g’’ orbitals of an idealized octahedral complex – dxz (b1).and hyperﬁnematrices are coincident. depending on the conformation of the CH2CH2 bridges of the dppe ligands.4
Table 4.
. (b) Computer simulation using the parameters in Table 4.64
Chapter 4
ﬁtting of the ﬁeld positions of the spectral features. 24. Since it has a nominal d5 conﬁguration.998
Figure 4.4 [E] (a) ESR spectrum of [Mn(dppe)2(CO)(CNBu)]21 in frozen CH2Cl2/ THF glass. The energies of the dxz and dyz orbitals are expected to be lowered by [E] ESR parameters for [Mn(dppe)2(CO)(CNBu)]21
104AiiMn (cmÀ1) 30. are given in Table 4.3. In this case the principle axes of the g. or dx2 Ày2 (a1). and a simulation based on those parameters is shown in Figure 4.3
gii 2.3.2 20.4(b). (Reproduced with permission from ref.6 146.107 2.9 104AiiP (cmÀ1) 27. [Mn(dppe)2(CO)(CNBu)]21 has approximate C2v symmetry.051 1. although the actual symmetry is reduced to C2.
as expected if it arises mostly through polarization of innershell s orbitals. other closely related species give isotropic couplings on the order of 60–70 G. a2 ¼ 0.k E0 À Ek k X 4b2 c2 xy. we have:21. DExy ¼ –1100z.660. " pﬃﬃﬃ # Á 8 3 17 À Dgx À Dgy ab þ Ax À Ay ¼ P À ð4:28Þ 7 14 Substituting the parameters. we get (a2Àb2) ¼ À0. Using eqn (4. we have DEyz ¼ 16z.004. which is qualitatively consistent with the expected MO energy level scheme.005. y.154. z in Table 4.Anisotropic ESR Spectra
65
backdonation into the p* orbitals of the CO and CNBu ligands so that the SOMO is likely based on dx2 Ày2 : ð4:26Þ jSOMOi ¼ ajz2 þ bjx2 À y2 þ Á Á Á Although the isotropic spectrum was not suﬃciently resolved to unambiguously determine AMn.11)..
. dxz. and P ¼ 207. most of the departure from axial symmetry can be explained by the spin–orbit coupling correction. The departure from axial symmetry is due to spin– orbit coupling and/or dx2 Ày2 =dz2 hybridization. the deviations of the gmatrix components from the freeelectron value. (Dgx þ Dgy) ¼ 0.e. DExz ¼ 19z. (The upper sign applies if the components are listed in the order x.3.25 if we assume an isotropic coupling in this range. and dxy orbitals. Assuming that the SOMO is mostly dx2 Ày2 . we get b2 ¼ 0. all three matrix components must have the same sign.) Finally.655. we have ab ¼ Æ0. the lower sign if the order is y.k
k
ð4:29bÞ
Dgzz ¼ 2zMn
E0 À Ek
ð4:29cÞ
If we assume coupling with single pure dyz. are found to be: Dgxx ¼ 2zMn X b2 c 2 yz. The dz2 component is not really signiﬁcant. i. we have AMn ¼ À65. z.26 ! Á 2 Á 4À 2 5 À a À b2 À Dgz À Dgx þ Dgy Az À AMn ¼ P ð4:27Þ 7 3 42 With Dgz ¼ À0. If the isotropic hyperﬁne coupling is negative. ge. x. (b 44 a) and including the appropriate spin–orbit coupling corrections.058. given the accuracy of the data and the theory.9 Â 10À4 cmÀ1.6 Â 10À4 cmÀ1 (ref. 14).k
k
E0 À Ek
ð4:29aÞ
Dgyy ¼ 2zMn
X b2 c 2 xz.
copyright (1981) American Chemical Society.3COD)]À: (a) experimental spectrum in frozen THF solution.
.2
Some Cobalt(0) Radical Anions
ESR spectra of [CpCo(1.3COD)]
[CpCo(1.4. based on the parameters in Table 4.3COD)]À in frozen THF solution and [CpCo(1.66
Chapter 4
4. (Reproduced with permission from ref. 27. These spectra have been
_ _
_
OC Co Co
Ph Ph Ph
Co
CO
Ph Ph
[CpCo(1.) (b) Computer simulation.5(a) and 4.5 [E] ESR spectrum of [CpCo(1.6(a).5COD)]À in frozen DMF have been reported by Geiger and coworkers27 and are reproduced in Figures 4.5.5COD)]
[(C5Ph5)Co(CO)2]
Figure 4.
171 2.) (b) Computer simulation.6 [CpCo(1. 27.018 2. 20.4 16.2 [(C5Ph5)Co(CO)2]Àb 2. Also shown in Table 4.997 139.6(b).151 2.8 45.5COD)]À in frozen DMF solution.3COD)]Àa 2. computersimulated spectra based on these parameters are shown in Figures 4.6 [E] (a) ESR spectrum of [CpCo(1.2 44.4
Radical anion
[E] ESR parameters for cobalt(0) radical anions
gx gy gz 104Ax(cmÀ1) 104Ay (cmÀ1) 104Az (cmÀ1) 36. (Reproduced with permission from ref.1
[CpCo(1.041 1. copyright (1981) American Chemical Society. the gmatrix
.5COD)]Àa 2.027 1.027 1.7 36. 27.5(b) and 4. Assuming that such hybridization is negligible.8 38.
Table 4.4.995 157.9
a b
From ref. From ref. The SOMO in these radicals is expected from extended Huckel MO calcu¨ lations27–29 to be primarily cobalt 3dyz in character. we assume the signs are negative since the isotropic couplings almost certainly arise from polarization of inner shell s orbitals (see below). In the Cs symmetry of the radicals.4 are the ESR parameters for [(C5Ph5)Co(CO)2]À.4.20 The 59Co hyperﬁne matrix components must have identical signs in order that the average values match the observed isotropic couplings.
reinterpreted to give the parameters shown in Table 4.Anisotropic ESR Spectra
67
Figure 4. based on the parameters in Table 4. dyz belongs to the a00 representation and dhybridization is possible only with dxy.985 158.
k 2 E0 À Ek k ð4:30aÞ
Dgyy
ð4:30bÞ
Dgzz ¼ 2zCo
ð4:30cÞ
The dipolar contribution to the hyperﬁne matrix is given by eqn (4. the xcomponent.68
Chapter 4
Table 4. 27.740 0.k 2 þ3a2 cz2 .5
Radical anion
[E] Electron spin densities in cobalt(0) radical anions
rd
Àb
Asa À97. but here spin–orbit coupling with the four other dorbitals contributes somewhat diﬀerently to the gmatrix and to the hyperﬁne matrix. eqn (4.591
[CpCo(1.k 2 ¼ 2zCo E0 À Ek k À Á X a2 cxy. the Fermi contact contribution to the isotropic coupling. The simplest way of expressing the hyperﬁne matrix is in terms of the isotropic coupling.666 0. and the departure from axial symmetry. As.681 0.5COD)] [CpCo(1.4.5.591 0. These often can be expressed in terms of the gmatrix components.3COD)]Àb [(C5Ph5)Co(CO)2]Àc
a b
0.0 À87.c From ref.2 À77. as in the Mn(II) example discussed above. 20. With the assumed signs of the hyperﬁne components of Table 4. but spin–orbit coupling contributions are signiﬁcant.
components are given by:20 Dgxx À Á À Á X a2 cx2 Ày2 . From ref.20). Á 1 À A ¼ As þ P Dgxx þ Dgyy þ Dgzz 3 Á 4 2 5 À Ax À A ¼ P À a2 þ Dgxx À Dgyy þ Dgzz 7 3 42 Á 6a2 zP 17P À 1 1 Dgyy þ Dgzz þ À Ay À Az ¼ 14 7 DEx2 Ày2 DEz2 ! ð4:31bÞ ! ð4:31cÞ ð4:31aÞ
Since 3dyz/4s admixture is symmetryforbidden for these radicals.4
As/Qd 0.31b) can be used unambiguously to compute a2 ¼ rd with the results shown in Table 4. As ¼ Qd rd ð4:32Þ
.540
In units of 10À4 cmÀ1. must be entirely from spin polarization.k 2 ¼ 2zCo E0 À Ek k À Á X a2 cxz.
The component gxx is indeed smaller than expected for [(C5Ph5)Co(CO)2]À and EHMO calculations do indeed suggest an MO with signiﬁcant pz character just above the SOMO in energy. dxy. which is primarily cobalt 3d in character. but grouped at about the same energy. at about the same energy for all three anions. More information is available in Gordy’s book.30) then would suggest that Dgxx/4 E Dgyy o Dgzz. With the assignments of Table 4. 18). is implied by these results but is unsubstantiated by the reported EHMO calculations. Equations (4. The results are shown in the last column of Table 4. If MOs with signiﬁcant pz or px character lie just above the SOMO. taking Qd ¼ À131 Â 10À4 cmÀ1 – estimated from the isotropic Co coupling in [PhCCo3(CO)9]À (ref. Much of the work on organic systems was done by Walter Gordy and his students at Duke University in the 1960s and 1970s.29 suggest that the SOMO is the highestenergy MO.20 we postulated admixture of some cobalt 4py character in the SOMO. The second relationship is not found for any of the anions.3COD)]À and [CpCo(1.30
4. EHMO calculations27. and possibly even negative.5COD)]À. We review a couple of these experiments here. and dz2 contributions. but a reliable ordering of the spin densities.30). then gxx and gzz would be less positive than expected from eqns (4. The data were analyzed as described above to yield the
. An MO with signiﬁcant px character. At lower energy is an orbital with dxz character and still lower.3COD)]À and [CpCo(1. this orbital is apparently substantially higher in energy in [CpCo(1. The gmatrix presents an interesting problem in these cases.Anisotropic ESR Spectra
69
Thus we can obtain an independent estimate of the delectron spin density from the values of As.6 Organic Examples of Solidstate ESR Spectra
Far less eﬀort has been directed at organic solidstate spectra than at organometallics and coordination complexes.5.28. jSOMOi ¼ ajyzi þ bjyi þ Á Á Á which would result in additional contributions to gxx and gzz: gxx ðpÞ ¼ 2zp b2 =DEz and gzz ðpÞ ¼ 2zp b2 =DEx ð4:34Þ ð4:33Þ
where zp is the cobalt 4p spin–orbit coupling parameter (zp E zd/3). are MOs with dx2 Ày2 .5COD)]À. suggesting a systematic error in one of the parameters.4.1
Irradiated Single Crystal of Glycylglycine
Katayama and Gordy31 studied ESR spectra of girradiated single crystals of glycylglycine. Reversing the y and z assignments makes the agreement worse. the ﬁrst relationship is approximately correct for [CpCo(1. In discussing this problem for [(C5Ph5)Co(CO)2]À. but very poor for [(C5Ph5)Co(CO)2]À. The spin densities estimated from the isotropic couplings are consistently about 10% higher than those from the dipolar coupling matrix.6.
4.
2 20.6.7 and 4. The axes were deﬁned as follows: z is along the CH bond. The methylene group freely rotates at room temperature but is stationary at 77 K.0029 2.3 27.0033
aH(G) 19 28 9
aN(G) 4 3 2
Table 4. The ESR spectrum indicated that the radicals produced were those pictured in Tables 4. and x is > to the NCHC plane. y is in the NCHC plane > to CH.0035 2.4
aP(G) 38.6 ESR parameters for the glycylglycine radical.4
.7
ESR parameters for the phosphonylmethyl radical32
O
HO HO
P
CH2
Axis x y z
g 2.7 is the more abundant of the two.31 (H and N hyperﬁne splittings from nuclei in structure shown in boldface.7 44. Table 4.8.0024 2. The conclusion was that the radical produced by girradiation was that pictured in Table 4.6.0028 2.
4.70
Chapter 4
parameters shown in Table 4. This conclusion was based on the larger coupling to H than to N.)
O H3N + C H H C N H C CO2H
Axis x y z
g 2.0022
aH(G) 16. The spectra were analyzed as described above and the results are also summarized in the tables.2
Xirradiated Single Crystal of Methylene Diphosphonic Acid
Lucken and coworkers32 subjected a single crystal of methylene diphosphonic acid to Xirradiation. but nonetheless there is a signiﬁcant coupling to N.3 41. where splitting from two nonequivalent protons is observed for some orientations of the crystal.6. The species shown in Table 4.
Dnd.. D2.z Axial Cn. i.g. Cs. the metal nucleus in a mononuclear complex).8
ESR parameters for the diphosphonylmethyl radical32
O O
HO HO
P C
P OH OH
H
Axis x y z
g 2. the hyperﬁne matrix is subject to the same
.5 38. D2h None gxz ¼ gyz ¼ gxy ¼ 0 x.0029 2.and hyperﬁne matrix principal axes.Anisotropic ESR Spectra
71
Table
4.y. While this is sometimes true. In textbooks on ESR6a.e.30. For a nucleus sharing all the molecular symmetry elements (e.5
aP(G) 36.4
Table 4.0024 2.36
4.7 Noncoincident Matrix Axes
In general.38 has given a detailed group theoretical analysis of symmetry ¨ restrictions on the orientations of g.9
Symmetry Point groups
[E] Symmetry restrictions on gmatrix components
Triclinic C1. g and Ai. Dn. gx..1
Symmetry Considerations
Kneubuhl37. Cnv.0 18. Cnh. n 4 2 gxx ¼ gyy gxz ¼ gyz ¼ gxy ¼ 0 x.1 40. Ci None None None Monoclinic C2.7. Dnh.9. gz and Aix. gy.and nuclear hyperﬁne coupling matrices. the g. Aiy. there are many instances where the principal axes are noncoincident.z
Restrictions on diagonal elements Restrictions on oﬀdiagonal elements Required matrix axes
4. C2h None gxz ¼ gyz ¼ 0 z Orthorhombic C2v.y. Aiz. can be written in diagonal form with three principal values.7 30. His results are summarized in Table 4.0022
aH(G) 12.33–35 it is usually assumed that the same set of principal axes diagonalizes all the relevant matrices.
13C. 57Fe. Secondly. but the other two may be diﬀerent. The result of such a competition is that a series of features that would be equally spaced (to ﬁrstorder) acquires markedly uneven spacings. but for small mI the extrema will be determined by extrema in the eﬀective gvalue and will correspond to angles close to the gmatrix axes. In general. one hyperﬁne axis is also a gmatrix axis. At ﬁrst glance. most of the information on noncoincident matrix axes is derived from such studies. In triclinic symmetry (C1 or Ci).g.and hyperﬁnematrix axes. there are relatively few cases of such noncoincidence reported in the literature. it is frequently possible to ﬁnd the orientation of a set of matrix axes relative to those of another matrix. the ﬁeld extrema will be determined largely by the extrema in the eﬀective hyperﬁne coupling and will occur at angles close to the hyperﬁne matrix axes. it would appear that all orientation dependence should be lost in the spectrum of a randomly oriented sample and that location of the g.
4. This follows from the fact that minimum and maximum resonant ﬁelds are determined by a competition between extrema in the angledependent values of g and A. since the observable eﬀects in powder spectra depend on the magnitude of the matrix
. Since a spin1/2 nucleus gives only two hyperﬁne lines. Most of the examples discussed by Pilbrow and Lowrey in their 1980 review36 cite cases of transition metal ions doped into a host lattice at sites of low symmetry. In monoclinic symmetry. the eﬀects of axis noncoincidence will be more noticeable if two (or more) matrices have large anisotropies that are comparable in magnitude. DgmBB E DA.and hyperﬁne matrix principal axes to be noncoincident. e. This is not to say that matrix axis noncoincidence is rare but that the eﬀects have only rarely been recognized.. Thus powder spectra cannot be analyzed to extract the orientations of hyperﬁne matrix axes for such important nuclei as 1H.7. Consider the case of noncoincident g.and hyperﬁnematrix principal axes would be impossible. such nuclear hyperﬁne matrices necessarily share principal axes with the gmatrix. For large values of mI. 31P. While it is true that there is no way of obtaining matrix axes relative to molecular axes from a powder pattern. there can be no variation in spacings. The observable eﬀects of matrix axis noncoincidence on powder patterns range from blatantly obvious to negligible. and 103Rh.2
Experimental Determination of Matrix Axis Orientations
We have seen that spectra of dilute single crystals are analyzed in a way that gives the orientations of the g.72
Chapter 4
restrictions as the gmatrix. Historically. The hyperﬁne matrix for a ligand atom (or for a metal in polynuclear complexes) is constrained only by the symmetry elements that the nucleus shares with the molecule.and hyperﬁnematrix principal axes relative to the crystal axes. none of the three principal axes need be shared by the gmatrix and hyperﬁne matrix. Although symmetry considerations often permit g. Two corollaries stem from this generalization. 19F. In orthorhombic or axial symmetry.
showing positions mA ¼ À1. even though the relative anisotropy may be large.and hyperﬁne matrix axis noncoincidence is the ESR spectrum of [CpCr(CO)2NO]À. 39.8 Organometallic Examples of Noncoincident Matrix Axes
4. zero linewidth.1 A Chromium Nitrosyl Complex
A good example of the eﬀect of g. þ1.39 a simulation is shown in Figure 4. the principal axes of the hyperﬁne matrix for a nucleus with small hyperﬁne coupling generally cannot be located from a powder spectrum. (b) Firstderivative spectrum in frozen DMF.
Cr O C N O C O
(a)
(b)
3340
3360
3380
3400
3420
3440
Magnetic Field/Gauss
Figure
4.7 (a) Computer simulation of an absorption spectrum. 0. studied by Geiger and coworkers. copyright (1984) American Chemical Society.8.7.)
.Anisotropic ESR Spectra
73
anisotropies.
4. (Reproduced with permission from ref.
showed coupling to 57 Fe and to a unique 13C nucleus with axial hyperﬁne matrices sharing principal axes with the gmatrix.
.010 AFe ¼ 9. identiﬁed as Fe(CO)51.6 C1 A> ¼ 17.10
Fe(CO)þ 5 Fe(CO)À 5
a
Coupling constants in units of 10À4 cmÀ1. and the spectrum simulated virtually exactly.or 57Feenriched Fe(CO)5 and girradiated at 77 K. lie in reﬂection planes. 40 and 41. One species. would be evenly spaced. In fact. The iron atom and the carbon of the axial CO group would have the full symmetry of the group and so these hyperﬁne matrices would share principal axes with the gmatrix. The eﬀect can be understood.10 lists the parameters.7 A2Fe ¼ 4.74
14
Chapter 4
The g.001 g>¼ 2.10).40.
4. AFe.7 AC2C5 ¼ (þ)6. the principal axes of the gmatrix must be the molecular axes (the C4 axis and normals to the reﬂection planes). If the g axis was exactly in the A> plane. the three negativegoing g. Preston and coworkers. and AC1 have only one matrix axis in common (that corresponding to the third component of each matrix listed in Table 4. If the radical is square pyramidal (C4v) Fe(CO)51 (1). the major matrix axes for the equatorial carbons are tilted slightly in the –z direction from the ideal locations along the Æx and Æy axes. [E] ESR parameters for Fe(CO)51 and Fe(CO)5Àa
g ¼ 2.6 A1C1 ¼ 87. A> features.2 AC1 ¼ 19.8. if the g axis is about 151 out of the A> plane. The gmatrix suggests that the metal contribution is dz2 and the iron hyperﬁne matrix then can be used to estimate about 55% iron 3d and 34% axial carbon 2pz spin density.4 Fe A> ¼ 6.989 g2 ¼ 2.and N hyperﬁne matrices are approximately axial for this radical.2 A1Fe ¼ 6.4 A>C2C5 ¼ (þ)8.7 A3C1 ¼ 65. the spacings are very uneven – far more so than can be explained by secondorder shifts. on the other hand. but the g axis lies close to the perpendicular plane of the hyperﬁne matrix.5 A3Fe ¼ 3.4 A2C1 ¼ 70. In fact.2
Iron Pentacarbonyl Ions
Two particularly interesting organometallic examples have been reported by Morton. The four equatorial carbonyl carbons.081 g1 ¼ 1.40 Coupling was also observed to four other 13C nuclei with identical coupling matrices but with the major axis approximately along the gmatrix xaxis for two nuclei and along the gmatrix yaxis for the other two. corresponding to resonant ﬁeld maxima. The spin density on the equatorial carbons then is mostly negative and due to spin polarization. and to four other carbons. Data from refs. However. but not on the C4axis and so are symmetryrequired to share only one principal axis with the gmatrix. in this case g. The other species observed in irradiated Fe(CO)5doped crystals of Cr(CO)6 also showed coupling to 57Fe. doped with 13C. Table 4. to a unique 13C.41 Spectra of single crystals of Cr(CO)6. showed two diﬀerent radicals.6
Table 4.003 g3 ¼ 2.
bx ¼ À0. but dyz is involved in the p orbitals of the C ¼ O group: jpi ¼ cyz jyz. The hyperﬁne matrices suggest about 38% iron 3dz2 . Fe þ cy jy. are oriented about 271 away from the corresponding 57Fe hyperﬁne matrix axes. assumed to be essentially the Fe–C bond. The yaxis. relative to those of the 57Fe hyperﬁne axes. normal to the reﬂection plane. Spin–orbit coupling will mix the SOMO with MOs having iron dyz or dxz character. but the components are considerably smaller (4–15 Â 10À4 cmÀ1).11) gives the gmatrix components: pﬃﬃﬃ 3a2 c2 þ 3abz cyz cy yz Dgxx ¼ 2zFe ð4:37aÞ DEp Dgyy ¼ 2zFe 3a2 c2 xz DExz ð4:37bÞ
.and zaxes of the gmatrix. is common to all three matrices.
O O C C Fe C C O O 1 CO + O O C C Fe C − O z(gmatrix)
27°
x z y
106° C C O O 2
The most striking feature of these results is the orientation of the unique 13C hyperﬁne matrix axes.12. however. This orientation led Fairhurst et al. However. 18% carbon 2p. the authors conclude that the carbon 2p axis is oriented at about 1061 relative to the Fe–C bond axis and that the Fe–C–O bond angle is about 1191.Anisotropic ESR Spectra
57
75
13
The other Fe hyperﬁne axes are rotated by about 271 and those of the C hyperﬁne matrix by about 481 relative to the gmatrix axes. Insuﬃcient data were accumulated to determine the complete hyperﬁne matrices for the other four carbons. and bz ¼ 0. eqn (4. Ci ð4:35Þ where a ¼ 0. Using detailed arguments regarding the orientation of the gmatrix axes relative to the crystal axes.41. Thus we must ask: Why are the gmatrix axes diﬀerent? The SOMO can be written: jSOMOi ¼ ajz2 .62. and 6% carbon 2s spin densities. the 57Fe hyperﬁne matrix major axis must correspond to the local zaxis. Since the iron dorbital contribution to the SOMO appears to be nearly pure dz2 . these authors did not discuss the orientation of the gmatrix axes.41 to assign the spectrum to [Fe(CO)5]À (2) and to describe the species as a substituted acyl radical. C ð4:36Þ Assuming that there is only one p orbital close enough in energy to couple signiﬁcantly. The x. Ci þ bz jz. Fe þ bx jx.
(Reproduced with permission from ref.
4. the gmatrix will have oﬀdiagonal elements that may be large enough to shift the principal axes away from the molecular coordinate system.8.and hyperﬁnematrix principal axes.)
.8) shows that the ‘‘parallel’’ features are not evenly spaced. The frozen solution spectrum (Figure 4. and other MOs are similarly delocalized.
3000
3200
3400 3600 Magnetic Field/Gauss
3800
Figure 4.8 ESR spectrum of the Mn(II) complex [Mn(dppm)2(CO)(CN)]1 in CH2Cl2/ C2H4Cl2 at 90 K. (dppm ¼ Ph2PCH2PPh2) (ref. if the unpaired electron in a radical is delocalized asymmetrically. The nearly identical complex [Mn(dppm)2(CO)(CN)]1. This may be a rather general eﬀect. 25.3
Another Lowspin Manganese(II) Complex
The lowspin manganese(II) complex [Mn(dppe)2(CO)(CNBu)]21 gave us a textbook example of a wellbehaved ESR spectrum characterized by coincident g. this expression gives cyz/cy E 0. a reasonable result.76
Chapter 4
Dgzz ¼ 0 pﬃﬃﬃ 3abx cyz cy DEp
ð4:37cÞ
gxz ¼ À2zFe
ð4:37dÞ
The gmatrix can be diagonalized by rotation about the yaxis by the angle b: pﬃﬃﬃ 2gxz À2 3abx tan 2b ¼ ¼ À Á pﬃﬃﬃ ð4:38Þ gxx À gzz 3a2 cyz =cy þ 3abz With b ¼ 271. 25) provides us with a good example of noncoincident principal axes. copyright (1993) Royal Society of Chemistry.5.
9) shows that the CH2 groups tilt down and up. If the CH2 groups of the dppm ligands were coplanar with Mn. (Reproduced with permission from ref. This. there are exceptional circumstances in which nature is kind to the experimentalist!
Figure 4. To avoid this interaction.61. of course. With two electrons. and that. dx2 Ày2 is not the only Mn contribution to the SOMO. i.9
Xray structure showing methylene groups tilted up and down to avoid antibonding interaction with the Mn dx2 Ày2 SOMO in the complex [Mn(dppm)2(CO)(CN)]1. further decreasing the antibonding interaction. tells us that the molecule does not have C2v symmetry. rotation about the yaxis.and Amatrices are displaced by b ¼ 19.
4. However. two things happen: (1) incorporation of dxz character tilts the lobes up and down.8. unlike the dppe analog. One way of interpreting the results is that the Mn contribution to the SOMO incorporates a small amount of dxz character. the ‘‘x’’ lobes of the dx2 Ày2 SOMO would be directed toward the carbon atoms and an antibonding interaction would result. there is no way of extracting the orientation of the principal axes of the gmatrix from a powder or frozen solution ESR spectrum. The reason for this hybridization is not diﬃcult to discover.Anisotropic ESR Spectra
77
The spectrum can be understood if the zaxes of the g. the neutral parent Mn(I) complex would be expected to be even more distorted.e. the Xray structure of the Mn(II) cation (Figure 4.. copyright (1993) Royal Society of Chemistry.4
Chromium(I) Pianostool Complex
Ordinarily. The consequence of this hybridization would be to tilt the ‘‘x’’ lobes of dx2 Ày2 up and down.)
. 25. With only one electron in the SOMO. away from the C atoms. At the same time. and an Xray structure shows that the up and down displacements of the CH2 groups is approximately doubled. the antibonding interaction is strong enough to produce this distortion.
)
Spectra of the lowspin d5 Cr(I) complex [(C5Ph5)Cr(CO)2PMe3]28 are shown at 125 and 200 K in Figure 4. we can assign this axis as the principal axis for g measured at 200 K and. copyright (1996) American Chemical Society.10 Spectra of [(ZC5Ph5)Cr(CO)2PMe3] in toluene at 125 and 200 K. the spectrum still appears as a ‘‘powder pattern’’.994).012). 28.013. g> ¼ 2.104. above the freezing point of toluene. noting that the g axis diﬀers from the gz axis by the angle b and that gy and gx are averaged according to: 2g2 ¼ g2 þ g2 þ ðg2 þ g2 Þ cos2 2b jj z x z x Substitution of the gcomponents gives b ¼ 151.10. With this interpretation. 1. corresponding to the three gcomponents (2. each a doublet due to hyperﬁne coupling to 31P. (Reproduced with permission from ref. The symmetry of the static complex is at most Cs.78
Chapter 4
×4 125 K
200 K
3200
3250
3300 3350 Magnetic Field/Gauss
3400
3450
Figure
4. As the sample is warmed to 200 K. producing a spectrum averaged about the Cr–C5 axis. 2. but the lowﬁeld ‘‘perpendicular’’ features are nearly averaged and the ‘‘parallel’’ features have shifted slightly upﬁeld (g ¼ 2. The lowtemperature spectrum shows three sets of doublets. requiring one of the gmatrix principal axes to be normal to the plane of symmetry – this is assumed
.090. The exceptionally bulky C5Ph5 ligand apparently moves very slowly at 200 K but on the ESR time scale the Cr(CO)2PMe3 moiety rotates quickly.
the moral here is that EHMO calculations. and the second HOMO of a00 symmetry: jSOMOi ¼ a1 jx2 À y2 a2 jz2 a3 jxzi þ þ jHOMO1 i ¼ b1 jx2 À y2 þ b2 jz2 þ b3 jxzi jHOMO2 i ¼ c1 jxyi þ c2 jyzi the gmatrix components can be computed:42 pﬃﬃﬃ À Á2 zCr a1 c2 þ 3a2 c2 þ a3 c1 gxx ¼ ge þ E0 À E2 zCr ð2a1 c1 þ a3 c2 Þ2 E0 À E2
gzz ¼ ge þ
gyy
pﬃﬃﬃ Â Ã2 zCr a1 b3 À a3 b1 þ 3ða3 b2 þ a2 b3 Þ ¼ ge þ E0 À E1 pﬃﬃﬃ À Á zCr ð2a1 c1 þ a3 c2 Þ a1 c2 þ 3a2 c2 þ a3 c1 ¼À E0 À E2
gxz
Rotation about the yaxis by b diagonalizes the matrix.061.43 but the eﬀects are rather more subtle than those discussed above. c1 ¼ 0. but we can compare the results of an extended Huckel MO calculation.
.538. lacking charge selfconsistency. but the MO wavefunctions are often fairly accurate. If. the results would be in very poor agreement with experiment.81.194.8.Anisotropic ESR Spectra
79
to be xz. b. and gxz. Nevertheless. as suggested by extended Huckel MO calculations.5
[(RCCR 0 )Co2(CO)6]À and [SFeCo2(CO)9]À
Noncoincident matrix axis eﬀects are seen in the frozen solution spectra of [(RCCR 0 )Co2(CO)6]À (3)18 and [SFeCo2(CO)9]À (4).216. gyy. usually have large errors in the energies. is hardly enough to evaluate the LCAO coeﬃcients for the SOMO and second HOMO. and we ﬁnd: tan 2b ¼ À 2gxz gzz À gxx
The single experimental observable. the SOMO and ¨ ﬁrst HOMO are of a 0 symmetry. gzz. a3 ¼ À0. (Since zCr and E0ÀE2 cancel in the ¨ calculation of tan 2b. a2 ¼ 0.
4. if we were to substitute these LCAO coeﬃcients. b ¼ 14. c2 ¼ À0.) The results are: a1 ¼ 0. but also use the EHMO estimate of the energy diﬀerence. the EHMO calculation could come close. into the expressions for gxx. pBackbonding to the CO ligands is expected to lead to a hybrid SOMO.582.
11
(m1. the extrema in B are determined mostly by the hyperﬁne term.6
. To ﬁrst order. ax ¼ 15.3 55. the ﬁve components of the m ¼ m1 þ m2 ¼ þ3 feature. The largest hyperﬁne matrix component (ay ¼ 66.
Table 4.þ7/2 þ5/2. less widely spaced. When m1 ¼ m2.1 166.11. Another series of features. the resonant ﬁeld then is: B ¼ Bo À m1 aþ À m2 aÀ where Bo ¼ hn/gmB and: a2 ¼ a2 cos2 ðy Æ bÞ þ a2 sin2 ðy Æ bÞ Æ z x ð4:40Þ ð4:39Þ
where b describes the orientation of the hyperﬁne matrix zaxes relative to the gmatrix zaxis. evenly spaced (apart from small secondorder shifts). corresponding to the ﬁrst two and the last three of the above components.5 G.0 G in the case of the CoCoFeS cluster) results in 15 features. resolution into components. Since g is nearly isotropic. Since the resolution is sensitive to the noncoincidence. a1 and aÀ are equally weighted and the extrema occur at y ¼ 0 or 901.þ5/2 þ3/2. but the principal axes of the two 59Co hyperﬁne matrices are noncoincident.9 153. With az ¼ 53. In the experimental spectrum of the CoCoFeS cluster.0 37.1 156. it was possible to ﬁt the experimental spectrum to obtain b quite accurately. the hyperﬁne contributions to the ﬁeld extrema and the corresponding values of y are given in Table 4.9 156. b ¼ 181. but when m1 a m2.6 G.3 66.80
− Ph C C Co C O Ph CO C O O O C C OC Fe OC Co C C O O − S Co CO
z(A) y
Chapter 4
OC Co C O C O
β
z(g)
Co
Co
C CO O
3
4
In these cases.þ1/2 þ1/2.þ3/2
[E] Splitting of m ¼ 3 features in the [SFeCo2(CO)9]À spectrum43
ymin (1) À20 þ20 À15 þ15 0 (BÀB0)min (G) 37. in a few cases.À1/2 À1/2.0 55. the gmatrix is nearly isotropic. the extrema correspond to other angles. shows some variation in spacing and. Consider. m2) þ7/2. This behavior can be understood as follows: Suppose that the hyperﬁne matrix yaxes are coincident and consider molecular orientations with the magnetic ﬁeld in the xzplane. two resolved ﬁeld maximum features were seen. for example.5 ymax (1) þ67 À67 þ78 À78 90 (BÀB0)max (G) 166.
and hyperﬁnematrices. relative to the molecular x. are given in Table 4. a computer simulation using these parameters is shown in Figure 4. The other two axes may be rotated. and coworkers45 reported the ESR spectrum of the approximately octahedral (oxylylene)Mn(dmpe)2 (Figure 4. with increasing
. between the g. As shown in Figure 4. The clue evident in the spectrum is the large gap between the mMn ¼ À1/2 and þ1/2 ‘‘parallel’’ features. Hursthouse.and zaxes. (Since the two nuclei are symmetrically equivalent. Thus the molecule can be regarded. the spectrum is suﬃciently sensitive to the angle that b can be evaluated quite precisely.12 shows a set of simulated spectra for a hypothetical lowspin Mn(II) species. but the ethylene bridges of the dmpe ligands destroy the reﬂection planes. but the angle b. the Xray structure45 shows a small fold at the methylene groups of the oxylylene ligand. On closer examination. it is possible to obtain spectra with more than six resolved ‘‘parallel’’ features. Comparison with the crystal structure of neutral (Ph2C2)Co2(CO)644 shows that these local axes are roughly in the direction of the axial carbonyl ligands (the Co–CO bond is tilted 281 away from the Co–Co vector).11a) (dmpe ¼ Me2PCH2CH2PMe2).
4.6
(oXylylene)Mn(dmpe)2
As part of a study of Mn(II) dialkyls. The gmatrix principal axes must be coincident with the molecular symmetry axes. The ﬁnal parameters. the C2v molecular symmetry permits a speciﬁc prediction of the possible matrix axis orientations. The two cobalt nuclei are located in a reﬂection plane (which we label xz) so that symmetry requires the yaxis to be a principal axis for all three matrices. Indeed.Anisotropic ESR Spectra
81
It is relatively easy to understand the signiﬁcance of the noncoincident matrix axes in these cases. [(oXylylene)Mn(dmpe)2] has approximate C2v symmetry. as shown in the structure. The spectrum was interpreted assuming coincident g. it was realized that this is a rather extreme example of a spectrum inﬂuenced by noncoincident g. but a simulation based on the reported parameters gave a very poor ﬁt to the published spectrum.12. Thus it seems reasonable to say that the local axes on a metal are determined primarily by the more strongly interacting ligands and that bonds to other atoms can be described as bent.11(b).and hyperﬁne matrix zaxes.12. varies from 0 to 901. based on leastsquares ﬁtting of the positions of the resolved features and the isotropic parameters. For the Co2C2 cluster. the major axes of the hyperﬁne matrices approximate the local zaxes at the cobalt atoms and the angular displacement indicates a bent Co–Co antibonding interaction. Figure 4.and hyperﬁne matrix principal axes.8. by Æb. all seven spectra correspond to the same g. where the C–C or Fe–S bond axis (the molecular yaxis) is perpendicular to the plane of the page. the rotations must be equal and opposite. Wilkinson.) Since the magnitudes and probable signs of the cobalt hyperﬁne matrices suggest a SOMO predominantly dz2 in character. suggesting one or more extra features.and hyperﬁnematrix axes. which destroys the C2 axis.
Since it is a lowspin d5 Mn(II) species. Since we know that the hyperﬁne matrix major axis is not a gmatrix principal axis. and major dxy or dz2 contributions to the SOMO can be ruled out. or C1.) (b) Computer simulation of spectrum using parameters of Table 4.11 [E] (a) ESR spectrum of (oxylylene)Mn(dmpe)2 in frozen toluene solution. as C2v. the metal contribution to the SOMO must be a dhybrid. The four 31P couplings are apparently equivalent. This is consistent with a clue contained in the ESR parameters. copyright (1983) Royal Society of Chemistry.12. we can immediately reject hybrids 1. which are listed in Table 4.
accuracy. dyz. and 11 and 12 diﬀer
. To explain the matrix axis noncoincidence. 5 and 9. 3. Cs. 4. Consider the twelve binary hybrids based on these orbitals.13. Hybrids 2 and 6. or dx2 Ày2 . 45. and all are relatively small. Thus no lobe of the SOMO can be directed toward a phosphorus atom. the SOMO is expected to be dominated by one of the orbitals of the octahedral t2g set – dxz. C2. and 10 for which the major axis is a common axis. (Reproduced with permission from ref.82
C x z
Chapter 4
P Mn P P
P
C
y
Figure
4. 8. 7.
100. for various values of b. the Euler angle between the gmatrix and hyperﬁne matrix zaxes.000
[E] ESR parameters for (oxylylene)Mn(dmpe)2 (b ¼ 411)
104AMn (cmÀ1) 27 27 125 104AP (cmÀ1) 24.110 2. 2.035 2.000). AMn ¼ (150.5
only in the x.and ylabels and are essentially indistinguishable.12 [E] Computersimulated ESR spectra for a hypothetical lowspin Mn(II) radical with g ¼ (2. 25. so that there are only three cases to consider in detail.5 24.12
g 2. 2. Hybrid 6 can be written: jSOMOi ¼ ajx2 À y2 þ bjxzi ð4:41Þ
.Anisotropic ESR Spectra
83
Figure 4.050. 25) Â 10À4 cmÀ1.
Table 4.5 24.
43).84
Chapter 4
Table 4. major HF axis z z x x y z y y x z y x Common axis x.z x x x x y y y y z z z C2 a – – – – a a a b – – –
Representation
Straightforward application of eqns (4.y.13
# 1 2 3 4 5 6 7 8 9 10 11 12 Cs a0 a0 a0 a0 a00 – – – – – – –
[E] SOMO candidates: binary dhybrids
Major dAO dx2 Ày2 dx2 Ày2 dyz dyz dxz dx2 Ày2 dxz dxz dyz dx2 Ày2 dxz dyz Minor dAO dz2 dyz dx2 Ày2 dz2 dxy dxz dx2 Ày2 dz2 dxy dxy dyz dxz Approx.and hyperﬁne matrix principal axes for all hybridization ratios. dyz. the hyperﬁne matrix remains axial. exactly the same conclusions can be reached for hybrid 2. tan 2b0 ¼ À 2b=a 1 À ðb=aÞ2 ð4:43Þ
For reasonable values of the hybridization ratio. primarily dxz and dyz in character. Following the same reasoning with x and y interchanged. independent of the hybridization ratio. b and b 0 can diﬀer by only a few degrees and this kind of hybrid cannot explain the matrix axis noncoincidence. Hybrids 11 and 12 are: jSOMOi ¼ ajxzi þ bjyzi ð4:44Þ
In this case. although the principal axes are rotated in the xyplane by an angle a equal to angle b of eqn (4.43).24) yields a hyperﬁne matrix that can be diagonalized by rotation about the yaxis by the angle b: tan 2b ¼ À2b=a ð4:42Þ
The gmatrix can be written in relatively simple form if we assume that the only MOs close enough in energy to contribute signiﬁcantly are the ﬁlled MOs. hybrid 9: jSOMOi ¼ ajyzi þ bjxyi ð4:45Þ
. Thus this hybrid gives identical g. With this assumption.5. also given by eqn (4. Assuming that only MOs with predominantly dxz. b/a up to about 0. or dx2 Ày2 character contribute. a gmatrix is found that can be diagonalized by rotation in the xyplane by the angle a 0 . Finally. we obtain a matrix which can be diagonalized by rotation about the yaxis by b 0 .
then dxz/dxy hybridization is less likely than dyz/dxy. DEx2 Ày2 and DExz are expected to be comparable. eqn (4.Anisotropic ESR Spectra
85
gives an axial hyperﬁne matrix with principal axes rotated in the xzplane by an angle b. to explain the eﬀect in this way would require each of several carbon
. given by eqn (4. Several alternative explanations were considered for the matrix axis noncoincidence. and the principal values of the gmatrix can also be rationalized with reasonable values of Q and b/a. which is entirely consistent with experiment. cx2 Ày2 ðcxz Þ2 Q¼ DEx2 Ày2 DExz À
ð4:47Þ
ð4:48Þ
Since the energy diﬀerences. If we accept that conclusion. indeed.47) has a particularly simple form. A small rotation of dyz about the yaxis might reﬂect the small displacements of the phosphorus atoms from the idealized octahedral positions. we expect b E 0 and b 0 E451. The gmatrix is somewhat more complicated: 1 0 2 2 B B B Dg B ¼B 2z B B B @
acx2 Ày2 DEx2 Ày2
þ ðbcxz Þ DExz 0 2
2
0 ðbcyz Þ
DEyz
2
À
À
2ab cx2 Ày2 DEx2 Ày2
À abðcxz Þ DExz
2
0
2 À abðcxz Þ C DExz C C C C 0 C C 2 C A 2bcx2 Ày2 ðacxz Þ2 þ DExz DE 2 2
2ab cx2 Ày2 DEx2 Ày2
ð4:46Þ
x Ày
Diagonalization requires rotation about the yaxis by the angle b 0 . for small b/a.13). " # 2b 2Q þ 1 0 tan 2b ¼ À a Q À 1 À ðb=aÞ2 ð4Q À 1Þ where Á2 . the amount is far too small to rationalize rotation of the gmatrix axes by 411. The EHMO overlap matrix based on the Xray structure suggests that the molecule is much closer to C2 symmetry than to Cs. a result that can be ruled out from our analysis of the ESR results. The axial hyperﬁne matrix is in agreement with experiment. The detailed agreement is less satisfactory: the SOMO is predicted to be primarily dx2 Ày2 with a small dxz admixture (hybrid 6 of Table 4. An extended Huckel MO calculation supports the assumptions made in the ¨ above analysis in that the three ‘‘t2g’’ orbitals are indeed close together in energy and remain nearly nonbonding metalbased dorbitals. as we tacitly assumed above. tan 2b 0 ¼ þ2a/b so that.43). For Q ¼ 1. might lead to signiﬁcant contributions to the oﬀdiagonal terms of the gmatrix. While the EHMO calculations suggest that the MOs containing dx2 Ày2 and dyz do have contributions from the carbon atoms of the oxylylene group. either through the sbonds or into the psystem. it seemed possible that delocalization of spin density into the oxylylene ligand. In particular. the parameter Q is probably not far from unity.
An ESR spectrum was reported46 for one such complex. Careful examination of the spacings of the 59Co hyperﬁne lines in the spectrum reveals that the g and Aaxes are noncoincident. which is unreasonably large considering the poor overlap of these metal dorbitals with the relevant carbon orbitals.8.13
Simulation of ESR spectrum of the above cobalt dithiolene complex in frozen toluene at 77 K. copyright (1994) Royal Society of Chemistry. This tells us that the symmetry cannot be as high as C2v. dithiolene complexes were very popular subjects for investigation by inorganic chemists.)
. The noncoincidence angle a is given in Table 4.7
Cobalt Dithiolene Complexes
In the 1960s and 1970s. et al. the structure of which is shown in Figure 4. Carpenter.
F3C S Co S F3C S CF3 P(Ph)3 S CF3
3100
3200
3300 3400 Magnetic Field/Gauss
3500
3600
Figure 4. coordinated with an axial PPh3 ligand.13. Accordingly.14.13. (Reproduced with permission from ref. 47.47 determined the structure of the above complex and obtained ESR spectra for various complexes with diﬀerent steric requirements.86
Chapter 4
atoms to contribute 5% or more to the MOs with dx2 Ày2 or dyz character. This is surprising since several crystal structures obtained for iron dithiolene complexes are textbook examples of square pyramidal complexes of C2v symmetry.
4. A simulation based on a redetermination of the ESR parameters47 is also shown in Figure 4.
)
L Pet3 P(OPh)3 PPh3 P(OPh)3 PPh3 Pet3 PPh3 PPh3 a (1) 2 11 16 11 24 10 24 31 Æ Æ Æ Æ Æ Æ Æ Æ 2 5 1 5 1 2 1 2
R CN CF3 CF3 Ph Ph 4MePh 4MePh 4MeOPh
Clearly.48 Figure 4. from inspection of Table 4. The phenomenon is sometimes called ‘‘gstrain’’. there is a good correlation between the steric bulk of R and L and the noncoincidence angle a. and crystal structures of the R ¼ CF3. If the excess width is associated with orientation along the gmatrix zaxis. These orientations are given in Table 4.
4. In most cases. The ﬁeld position of a spectral feature can be written as: B¼ hn À kmI geff mB ð4:49Þ
. and this is observed.Anisotropic ESR Spectra
87
Table 4. Furthermore.14. Clearly. however. analysis of the hyperﬁne parameters leads to the conclusion that only about 25% of the electron spin resides in Co orbitals (mainly dxz). A successful interpretation of the eﬀect has.9 ‘‘gStrain’’
ESR spectral lines in a frozen solution or powder spectrum are sometimes peculiarly broad.14
Noncoincidence angles for [Co{S2C2R2}2L]. been made for some organometallic radicals. The ESR parameters for this spectrum49 are given in Table 4. the matter is dropped at that point and no attempt to explain further is made. the ‘‘parallel’’ features are rather variable in line width.15 where b is the angle between the gmatrix zaxis and the hyperﬁne matrix zaxis. with the lowﬁeld lines broad and the highﬁeld lines narrow.14 shows the ESR spectrum of [Mn(CO)(dmpe)Cp]1. Because of the noncoincidence of the g. L ¼ PPh3 and P(OPh)3 complexes do indeed show distortions. which makes the cobalt complexes much more easily distorted. but this electron occupies a dithiolene p* orbital. the various ‘‘parallel’’ features correspond to diﬀerent orientations of the magnetic ﬁeld in the gmatrix principal axis system.16. The diﬀerence between iron and cobalt is just one electron. we expect all widths to increase in the order mI ¼ À1/2 o À3/2 o À5/2 o þ5/2 o o þ3/2 o þ1/2. copyright (1994) Royal Society of Chemistry. (Data reproduced with permission from ref. 47.and Amatrix principal axes.
)
Table 4.6 104A2 (cmÀ1) ca. (Reproduced with permission from ref. (Reproduced from ref. we see that the line width of a feature depends not only on the orientation in the ﬁeld but also on the magnitude of the ﬁeld: @B Bgz ¼ À 2 cos2 y @gz geff ð4:51Þ
Values of qB/qgz are given in Table 4.0049.) (a) Experimental spectrum at 120 K.52)].15
g1 2. as are values of wexcess [computed from eqn (4. (b) simulation with parameters of Table 4. copyright (1997) American Chemical Society.14 ESR spectrum of [Mn(CO)(dmpe)Cp]1 in 1:1 CH2Cl2:C2H4Cl2. 0 104A3 (cmÀ1) 115.187 104A1 (cmÀ1) 35.50): rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ geff ¼ g2 cos2 y þ g2 cos2 j þ g2 sin2 j sin2 y z x y
ð4:50Þ
Diﬀerentiating B with respect to gz. (c) simulation with constant 4.000 g2 2. with permission.1 G Gaussian line widths.88
Chapter 4
Figure 4.16.3 b (1) 45.021
ESR parameters for [Mn(CO)(dmpe)Cp]148
g3 2. wexcess ¼ wg j@B=@gz j ¼ wg B=gz ð4:52Þ
.1 G line widths.15 and constant 4.4
where k is the angledependent hyperﬁne coupling and geﬀ is given by eqn (4. 48. 48. wg ¼ 0.
2. 1987.V. 1990. Preston. 169... J. Symons. Rieger. 1987.1 0 À5/2 3636 78. 1973.S. B. Amsterdam. 2. Drago. 376.A. Rieger. A very similar explanation deals with the broader lowﬁeld features in the ESR spectrum of [Cr(CO)2(PMe3)(ZC5Ph5)] (ref. Preston. W. Kriz. Reson. M. DeGray.10). 577. J. Radiochem. J. 48 with permission. (a) P. 41. 1971 13. 1315. 485. Adv. 8. Baugher and H. 305. Reson. ed. 88. 179.R.F. J.R. 7. P.0047 1/2 3053 3. J..F. New York.S. N. K. 14.4
mI B (G) y (1) qB/qgz (G) w (G) wexcess (G) wg
The experimental value of wexcess is deﬁned by wexcess ¼
qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ w2 À w2 where low high
whigh ¼ 4.C. J. B.H. Magn. I. Ch.)
5/2 2865 14. P. Suslick.N.. 75.. Dalton Trans. 1978. J. 4.2 5.F. 1983. J. J. Chem.. R.4 8 3. 1982. 203. Morton and K. Vugman and N. 1978. Physical Methods in Chemistry. Zeldes. 32. Elsevier. 318. 1977.. Rieger.. Taylor. 1990.Anisotropic ESR Spectra
89
Table
4. Reson.. Chem. P. in HighEnergy Processes in Organometallic Chemistry. D. 12.C.0048 3/2 2962 9.M.3 1392 8. Gahan and F.16
Analysis of widths of ‘‘parallel’’ features of the spectrum of [Mn(CO)(dmpe)Cp]1.5 6. p.0051 À1/2 3432 86.H.A.6 105 5. Goodman and J.J.V. 8. Washington. 10. 111. 11. Rev. Magn.1 À3/2 3532 81.A. Phys. J. Danon. Morton and K. Collison. 28) (Figure 4.R.. A. 6. Inorg. Taylor and P. Livingston and H. Chem. Garcia Herbosa. Atkins and M. Mabbs.. 49. 5.5 1240 7. P. Reson.
. The Structure of Inorganic Radicals. Chem. Vugman.G. 1987. 30. Mol. 3.R. Phys.H.C.1 46 4. 1970. ed. 1975. J. The simulation in Figure 4. DC. Phys.W. J. Magn. Symons.H. Reson. Soc. 59.. 50. 8.2 7. 1964. Reson. Raynor.B.2 0. in Organometallic Processes. (b) R.O. Caride and J. Magn. Ovchinnikov and V. American Chemical Society.C.C. Rieger. 13.1 G (the À3/2 line was used since the À1/2 line was subject to destructive interferences by a divergence feature). N. Trogler. 4418. DeGray and P. Bull. Saunders. P.E. Chem. Konstantinov.14(c) is based on wg ¼ 0. 4011. 1967. 9.9 0. N. p. Magn.V.
References
1.0049 G. Connelly and G.48 (Data from ref.. Pinhal. 15.M. Magn. 95. copyright (1997) American Chemical Society. Bray. J.2 0. 136.4 1324 7.
Kondens. Chem. Rep. 124. Geoﬀroy.J. 4787. M. 1980. Rieger. Mat.M. 4.A. 30.K.H. Preston. Hayward. Fukamachi. 410. Kneubuhl. Phys..L. Rieger. 103. T. New York. Magn. (London).L. Harding. Lionel.A. 15. Electron Spin Resonance in Chemistry. N. J.. A271.F. M. 1980. Organometallics. N. N. A. 234. T.L. T. Electron Spin Resonance. 28. B. Chem. Albright and R. J. 18. 83. B. Tulyathan and M. 19. Quayle and P.J.H.H. 24. Soc. L. New York. London. 37.. Electron Spin Resonance. 43. For a summary of Kneubuhl’s methods. 1980.H. A. J.R.A.. Dillard.J. D. 117. 1963.. Roy. J. Rieger. Rieger.H. Lucken. A. 31. Geiger. Albright. T. Morton and K. Tulyathan. Bolton. Q. 1982. 1974.G. J. Simpson. Soc. 1967. 140. P. 1289.A. Phys. Rieger. Meng and P. I. Plenum Press.. Gordy. W. Oxford.A.R.M. 4791. 364. J. Rieger and P. E. 1973. J.E. Bartlett. John Wiley. 111.H. Soc. S. 1769. 1970. Atherton. Boonyuen. 50. 17. McGarvey. Robinson and J. Soc.R.. 36. Adams. N. Carriedo. 1985.C.K. Ayscough. Chem. 1978. Mol. Clarendon Press. M.H. Data Tab.E. Bleaney. Dalton Trans. in Electron Spin Resonance of Metal Complexes. PerezCarreno. Moraczweski and B.G. ¨ 38. Kawamura. 28. 1963. Ber. 6219. A. McGrawHill. J. Riera and G. Katayama and W. Abragam and B. Phys.. 433. Yen. J. E. Rausch.M. ed. A. 7000. 1. Lowrey. Stone. V. F.90
Chapter 4
16. 3466.
. V. 33. J. Data Nucl.G. ¨ 39. Faraday I. 1981. W. Soc. 1969. (a) J. Ellis Horwood. Theory and Applications of Electron Spin Resonance. Chem. Phys. 76. Electron Paramagnetic Resonance of Transition Ions. Dalton Trans. Chicester. p. J. 32. Soc.F.. P.. 1987. Hoﬀmann. Rosair. Rheingold. 1965. 235.A. 36. Soc. Connelly. 1984. 106. Soc. Am.H. 1981. Am. Miller. K. Chem. McInnes.J. Methuen. W. Rieger. Wertz and J.J. 27. Phys.. Reson. 102. G. Gordy. Geiger. Chem. Carriedo. 26. P.. 1390. New York.B.H. 3565. S. 29.E. 23. (b) A recent example that illustrates the intricacies involved in extracting information about axis orientations may be found in C. Am. 424... A. Chem. 156.. 1972. S. Chem.G.M. G.J. 40. 22. B.G. Lane.C. 108.E. 20.J. DeGray. T. 1987. 34.G. 1993. W. Rieger. Sowa. J. Prog. 3103. N. 25. 2006. Proc. Am. 103.J. Peake. O. M. D. 1. Hammack. Dalton Trans. Raven. B.. Yonezawa.R. M. Koh and D. Atom. Ginet and E.J. Chem. Hayashida and T. Chem. 21.L. 1961.E. Castellani. G. Pilbrow and M.D. T. Chem. Connelly. 35. J. 35.P. J. P.. 1986. Soc. 1997.. Am. 33. Geiger. Connelly and S. J. Rieger. 1996. P. Raven and P. see ref.M. Orpen. A. A. Rieger..A. Orpen. Riera. Connelly.
. Pike. Rieger. 1774. B. Am. J.D.B. Rev. Rieger. 43. 98.A. R.D. A. Soc.G..G.. Rieger. J. Carpenter. Soc. Connelly.M. S. P. 135. Balch. 16. A.R. M.H. 4369.. Chem. 85. G. 6. Morton and K.H. J. Chem. Howard.. 47. Peake. Faraday Trans. 203. P. A. Coord. 1981. Chem. Robinson and J.S. Stults. Dalton Trans. Inorg. 1983.l.S.. 1994. Rieger. Sweigart. 46.H. G. Rieger and D. I. 44. Castellani. 3913. Soc. Cotton.D. B. Soc. M. Chem. Chem. J. 2540.F.B. F. Wilkinson. R. 45. 20.L. Organometallics. Hursthouse. G. Chem. Pike. 2903. Rieger and P.. G. Jamerson and B. Clark. N. 5872.H.. Chem. Girolami. 1982 77. 49.A. Phys. P. Dalton Trans.R.L. J. J. Inorg.H. 2158.L.H. ThorntonPett and M. 1967. 42. A.
. 1994. 1976. Rieger. J. 1989. Chem. 1997.P.A. C.Anisotropic ESR Spectra
91
41. Rieger and P. 48. 2631... Simpson. Preston. Fairhurst.
and intramolecular exchange phenomena and give examples of applications. but to proceed further in kinetic applications of ESR spectroscopy we must deal with the Bloch equations and modiﬁed Bloch equations. Here we will describe the phenomenological model. from the observed broadening. and. dEdt ! h ð5:1Þ
Thus. and on the analysis of line widths and saturation behavior. Ward and Weissman1 added some unreduced naphthalene to a solution of the radical anion. derive the Bloch equations and solve them for steadystate conditions. and from dt the rate constant for the electron transfer reaction: C10 H8 þ ½C10 H8 À ! ½C10 H8 À þ C10 H8 The result is very rapid electron exchange.1) to relate the lifetime of a spin state to the uncertainty in the energy of the state. Spin magnetic moments
92
.1 They made use of a form of the Heisenberg uncertainty principle (eqn 5. The phenomenological equations proposed by Felix Bloch in 19462 have had a profound eﬀect on the development of magnetic resonance. computed dt. We will also show how the Bloch equations can be extended to treat inter.
5. on the ways in which the experiments are described (particularly in NMR).1 Bloch’s Phenomenological Model
When a magnetic ﬁeld is applied to an electron or nuclear spin. with k ¼ 1 Â 106 L molÀ1 sÀ1. the spin quantization axis is deﬁned by the ﬁeld direction. if the lifetime of a spin state is dt. This approach works well for electron transfer reactions where the rate is simply related to the broadening. with consequences for ESR line widths. the energy level is broadened by an amount h/dt. both ESR and NMR. as expected.CHAPTER 5
ESR Kinetic Studies
One of the ﬁrst uses of ESR spectra to measure the rate of a chemical reaction was by Ward and Weissman in the early 1950s.
~ If M0 is the equilibrium magnetization along B and Mz is the zcomponent under nonequilibrium conditions. In addition to the precessional motion. Since this process involves transfer of energy from the spin system to the surroundings (conventionally called the ‘‘lattice’’). does involve an increase in the entropy of the spin system. T1 is called the spin–lattice relaxation time. there are h two relaxation eﬀects. the vector sum of all the spin magnetic moments will be a nonzero net magnetic moment or macroscopic magnetization: ~ M ¼ S ~i m
i
ð5:2Þ
~ ~ ~ At equilibrium M is in the direction of the ﬁeld B. called the transverse relaxation time. If we consider an ensemble of spins. Local shielding may ~ cause small variations in B or the eﬀective gfactor may vary slightly through the sample. then we assume that Mz approaches M0 with ﬁrstorder kinetics: dMZ MZ À Mo ¼À dt T1 ð5:4Þ
where T1 is the characteristic time for approach to equilibrium (the reciprocal of the rate constant). We assume that there is a characteristic time for this process.
. and that the transverse magnetization components decay to the equilibrium value of zero accordingly: dMx Mx ¼À dt T2 dMy My ¼À dt T2 ð5:5Þ
Notice that dephasing of the transverse magnetization does not aﬀect Mz. a T2 process involves no energy transfer but. but the individual spins may precess at slightly diﬀerent rates. Each individual magnetic moment undergoes this precessional motion.ESR Kinetic Studies
93
aligned with the ﬁeld are only slightly lower in energy than those aligned opposed to the ﬁeld. Thus an ensemble of spins that all start out in phase will gradually lose phase coherence – the individual spins will get out of step. being a spontaneous process. Since electrons are much more strongly coupled to molecular interactions than are nuclei (which are buried in a sea of innershell electrons). Suppose that M is somehow tilted down from the zaxis toward the xaxis and the precessional motion is started. If somehow M is tilted ~ there will be a torque that causes M to precess about B with the ~ ~ away from B equation of motion: ~ dM ~ ~ ¼ gB Â M dt ð5:3Þ
where g ¼ 2pgmB/h (or gmB/). T2. it is not surprising that T1 for electrons is usually much shorter (on the order of microseconds) than are nuclear T1s (on the order of seconds). There is a second kind of relaxation process that is at least as important for ~ magnetic resonance as the T1 process.
1
Derivation of the Bloch Equations
~ dM ^ ^ ~ ~ ^Mx À j My À k Mz À M0 ¼ gB Â M À i dt T2 T2 T1
Combining eqns (5.3)–(5. is termed secular.1. Thus. from the point of view of the electron spins. we have: ð5:7Þ
In a magnetic resonance experiment.5) must include both the eﬀects of spin–lattice relaxation as well as the dephasing of the transverse magnetization.5). Processes that involve nuclear spin ﬂips but not electron spin ﬂips are. In general. the T2 of eqn (5. A process that involves an electron spin ﬂip necessarily involves energy transfer to or from the lattice and is therefore a contribution to T1. but which results in loss of phase coherence. Transverse relaxation is often much faster than spin–lattice relaxation and T2 is then determined mostly by spin dephasing. we call such a process nonsecular.7) and separating it into components. but because the energy transferred is so small (compared with electron spin ﬂips) these processes are termed pseudosecular. however. and T2 is the observed transverse relaxation time. In ESR. The vector product of eqn (5. so that the total ﬁeld is: ^ ^ ~ ^ B ¼ i B1 cos ot þ j B1 sin ot þ k B0 ð5:8Þ
Note that there are other possible ways to impose a timedependent B1.7) then becomes: Â Ã ^ ^ ~ ~ B Â M ¼ À i B0 My À B1 Mz sin ot þ j ½B0 Mx À B1 Mz cos ot Â Ã ^ þ k B1 My cos ot À Mx sin ot Inserting this expression in eqn (5. we get: dMx Mx ¼ ÀgB0 My þ gB1 Mz sin ot À dt T2 ð5:9aÞ
. we should write: 1 1 1 ¼ þ T2 T1 T20 ð5:6Þ
where T2 0 is the spin dephasing relaxation time.8) corresponds to a circularly polarized ﬁeld initially aligned along the xaxis and rotating about the zaxis in a counterclockwise direction. it is also customary to classify relaxation processes by their eﬀects on electron and nuclear spins. also causes Mx and My to approach zero. in which Mz approaches M0. The one described in eqn (5.94
Chapter 5
It should be emphasized that the approach to equilibrium by a T1 process.
5. we apply not only a static ﬁeld B0 in the zdirection but an oscillating radiation ﬁeld B1 in the xyplane. nonsecular. A process that involves no spin ﬂips.
9a) and (5.11a) and substituting eqns (5.9b).1.
5.11): du u ¼ À½o À gB0 v À dt T2 Similarly. the radiation ﬁeld B1 is continuous and B0 is changed only slowly compared with the relaxation rates (socalled slow passage conditions). Diﬀerentiating eqn (5.2
Steadystate Solution
In a continuous wave (CW) magnetic resonance experiment. we obtain: dv v ¼ ½o À gB0 u À þ gB1 Mz dt T2 dMz M z À M0 ¼ ÀgB1 v À dt T1 ð5:12bÞ ð5:12aÞ
ð5:12cÞ
Equations (5. Thus a steadystate solution to eqns
. u is that part of Mx which is inphase with B1 and v is the part which is 901 out of phase.ESR Kinetic Studies
95
dMy My ¼ gB0 Mx À gB1 Mz cos ot À dt T2 Â Ã dMz =dt ¼ gB1 My cos ot À Mx sin ot À ðMz À M0 Þ=T1 It is convenient to write Mx and My as: Mx ¼ u cos ot þ v sin ot My ¼ u sin ot À v cos ot or u ¼ Mx cos ot þ My sin ot v ¼ Mx sin ot À My cos ot
ð5:9bÞ ð5:9cÞ
ð5:10aÞ ð5:10bÞ
ð5:11aÞ ð5:11bÞ
This is equivalent to transformation into a coordinate system that rotates with the oscillating ﬁeld.12a–c) are the Bloch equations in the rotating coordinate frame. we get: dMy du dMx ¼ cos ot À Mx o sin ot þ sin ot þ My o cos ot dt dt dt Â Ã Mx cos ot þ My sin ot ¼½gB0 À o Mx sin ot À My cos ot À T2 and substituting from eqns (5.
it is the transverse oscillating ﬁeld that causes the magnetization to have a nonequilibrium value. u and v go to zero and Mz approaches M0. corresponds to absorption. Recall that u is the transverse magnetization component inphase with the driving ﬁeld B1. is small so that g2B12T1T2 oo 1. Mz decreases (moves away from equilibrium). When the microwave or radiofrequency power. in contrast. The halfwidth at halfheight is easily found to be: Do ¼ 1 T2 or Dn ¼ 1 2pT2 or DB ¼ h gmB T2
where the last form is appropriate when (as in CW ESR) B0 is changed while keeping o constant.13b) in what follows. In general a response that is exactly in phase with a driving signal does not absorb power from the signal source and in spectroscopy corresponds to dispersion – in optical spectroscopy dispersion results from a small reduction of the speed of light as it traverses the medium. in a quantum mechanical description of the experiment. proportional to B12. as B1 increases.
. In magnetic resonance. but the problem can also be formulated so that optical dispersion is described in a way closely analogous to the Bloch equations description. or vmode.1 is a Lorentzian ‘‘line’’. Setting the derivatives to zero and solving the three simultaneous equations. u and v at ﬁrst increase with increasing B1. An outofphase response. it is usually the absorption. Notice that as B1 approaches zero.12) is appropriate. Equation (5.1. but eventually they decrease as the third term in the denominator begins to dominate. On the other hand. T2(o0 – o) is shown in Figure 5.14) corresponds to the classical Lorentzian line shape function and the absorption curve of Figure 5. That is. that is detected and so we conﬁne our attention to eqn (5. we get: u¼
2 gB1 M0 ðo0 À oÞT2 2 1 þ T2 ðo0 À oÞ2 þ g2 B2 T1 T2 1
ð5:13aÞ
v¼
gB1 M0 T2 1þ
2 T2 ðo0
À oÞ2 þ g2 B2 T1 T2 1
ð5:13bÞ
Mz ¼
h i 2 M0 1 þ T2 ðo0 À oÞ2
2 1 þ T2 ðo0 À oÞ2 þ g2 B2 T1 T2 1
ð5:13cÞ
where o0 ¼ gB0 is called the Larmor frequency and corresponds.96
Chapter 5
(5. as expected.13b) becomes: v¼ gB1 M0 T2
2 1 þ T2 ðo0 À oÞ2
ð5:14Þ
A plot of v vs. eqn (5. to the (angular) frequency of the energy level transition.
e.2 shows a Lorentzian derivative line.1 A Lorentzian absorption line. In ﬁrstderivative spectra. This width may be computed by taking the second derivative and ﬁnding the zeros. Furthermore.15) are valid] the derivative amplitude increases linearly with B1. proportional to M0. g2B12T1T2 cannot be neglected in eqn (5. i. that the width is independent of B1 and that neither width nor amplitude depends on T1. the amplitude is no longer linear
..14) and (5.ω0)
2
4
Figure 5. the derivative amplitude is inversely proportional to the square of the line width. In other words.
When the absorption is detected via small amplitude ﬁeld modulation. the signal is proportional to the ﬁrst derivative of absorption:
3 dv 2gB1 M0 T2 ðo0 À oÞ ¼h i2 do 2 1 þ T2 ðo0 À oÞ2
ð5:15Þ
Figure 5. obtaining: 2 Do ¼ pﬃﬃﬃ 3T2 1 or Dv ¼ pﬃﬃﬃ 3pT2 or h DB ¼ pﬃﬃﬃ 3pgmB T2
The absorption derivative amplitude is proportional to T22 whereas the width is proportional to T2À1.ESR Kinetic Studies
97
4
2
0 T2(ω . At higher power. Notice that in the limit of small B1 [where eqns (5. the product of the amplitude and the square of the width is independent of T2 and is sometimes taken as a measure of the intensity of the line.13b). it is most convenient to describe the line width as the separation between derivative extrema.
98
Chapter 5
4
2
0 T2(ω  ω0)
2
4
Figure 5.2 A Lorentzian ﬁrstderivative line.
in B1 and both amplitude and width depend on T1. Eventually, the amplitude begins to decrease with increasing B1 and we say that the resonance is saturated. In quantum mechanical language this corresponds to equalization of the energy level populations and reduction of net absorption.
5.2 Chemical Exchange – The Modiﬁed Bloch Equations
Suppose we have a system in which a spin can exist in either of two diﬀerent sites, A or B, and that these are distinguished by diﬀerent resonant frequencies, oA and oB, and/or by diﬀerent relaxation times, T2A and T2B. If there is no exchange between sites, site A spins and site B spins can be described separately and independently by sets of Bloch equations. When exchange takes place, however, additional rate terms – completely analogous to terms in chemical rate equations – must be added to the Bloch equations. The algebra we are about to get into can be compacted somewhat by introducing the complex magnetization, G ¼ u þ iv, so that eqns (5.12a) and (5.12b) can be combined to obtain: dG du dv G ¼ þi ¼À þ iðo0 À oÞG þ igB1 M0 dt dt dt T2 ð5:16Þ
In eqn (5.16), we have ignored the diﬀerence between M0 and Mz and so have assumed that B1 is small.
ESR Kinetic Studies
99
Spins at site A will have magnetization GA and those at site B will have magnetization GB. We now assume that A and B are interconverted by ﬁrstorder kinetics with an A  B rate constant tAÀ1 and a B  A rate constant tBÀ1. The site A magnetization thus decreases with a rate term –GA/tA and increases with rate GB/tB. Combining these terms with eqn (5.16) for site A: dGA GA GA GB ¼À À þ þ iðoA À oÞGA þ igB1 M0A dt T2A tA tB and a similar expression for site B: dGB GB GA GB ¼À þ À þ iðoB À oÞGB þ igB1 M0B dt T2B tA tB ð5:17bÞ ð5:17aÞ
The rather fearsome algebra can be somewhat simpliﬁed by deﬁning:
À1 À1 aA ¼ T2A À iðoA À oÞ; aB ¼ T2B À iðoB À oÞ
and noting that, at chemical equilibrium: pA t A ¼ pB t B and pA þ p B ¼ 1
where pA and pB are the fractions of the population at sites A and B. Deﬁning a mean lifetime: t¼ several useful relations result: pA ¼ tA tA þ tB pB ¼ tB tA þ tB tA tB tA þ tB
t ¼ pA t B ¼ pB t A Also, if M0 is the total equilibrium magnetization, we can write: M0A ¼ pA M0 M0B ¼ pB M0
With these relations, it is possible to ﬁnd the steadystate solution to eqns (5.17): G ¼ GA þ GB ¼ igB1 M0 tA þ tB þ tA tB ðpA aA þ pB aB Þ ð 1 þ t A a A Þ ð 1 þ t B aB Þ À 1 ð5:18Þ
The absorption signal, of course, is the imaginary part of eqn (5.18); the equation is too horrible to contemplate, but computersimulations, such as those shown in Figures 5.3 and 5.4, are relatively easy to produce. There are two limiting cases where the equations are easier to understand. In the slow exchange limit, where tAÀ1 and tBÀ1 are both small compared with oA – oB, the
100
τ = 0.001 τ = 0.01 τ = 0.02
Chapter 5
τ = 0.05
τ = 0.1
τ = 0.2
τ = 0.5
τ = 1.0
τ = 10
5
0
5
5
0 (ω  ω0)
5
5
0
5
Figure 5.3 Absorption curves, computed using eqn (5.18) for various values of t, and oA ¼ o0 À 5, oB ¼ o0þ5, T2AÀ1 ¼ T2BÀ1 ¼ 0.5, PA ¼ PB ¼ 0.5.
absorption, v ¼ Im(G), is: À À1 Á À À1 Á gB1 M0 pA T2A þ tÀ1 gB1 M0 pB T2B þ tÀ1 B A v¼À þÀ Á2 Á2 À1 À1 T2A þ tÀ1 þðoA À oÞ2 T2B þ tÀ1 þðoB À oÞ2 B A
ð5:19Þ
which corresponds to two Lorentzian lines centered at oA and oB and with widths ðT2AÀ1 þ tAÀ1 Þ and ðT2BÀ1 þ tBÀ1 Þ. In other words, the lines are unshifted but are broadened by an amount proportional to the reciprocal of the lifetimes. In the fast exchange limit, where tAÀ1 and tBÀ1 are both large compared with oA – oB, the absorption is: v¼
À1 gB1 M0 T2 À2 T2 þ ðo0 À oÞ2
ð5:20Þ
where o0 ¼ pA oA þ pB oB 1 pA pB ¼ þ þ pA pB ðoA À oB Þ2 t T2 T2A T2B Thus a single Lorentzian line is obtained that is centered at a weighted average resonant frequency and has a width proportional to a weighted average À1 T2 plus a term proportional to the average lifetime and the square of the separation of the slow exchange resonances. T2AÀ1 ¼ T2BÀ1 ¼ 0:5.05
x20 τ = 0.02
101
x4 τ = 0.ESR Kinetic Studies
x2/3 τ = 0. eqn (5. pA ¼ pB ¼ 0. it is clear that as the rate increases. coalesce and then begin to sharpen into the single line of the fast exchange limit.
. note that the vertical scale diﬀers – the plots are magniﬁed by the factors shown.ω 0)
5
5
0
5
Figure 5.001 x1 τ = 0.4 Firstderivative curves.5
x4 τ = 1. and oA ¼ o0 À 5.1
x20 τ = 0. In the socalled intermediate exchange region.0
x1 τ = 10
5
0
5
5
0 (ω .5.2
x10 τ = 0. Qualitatively.01 x2 τ = 0. computed using eqn (5.18) for various values of t. shift together. however.18) is not easily tractable and recourse is usually made to computer simulations. the separate resonances of the slow exchange limit broaden.
2 shows that the Gaussian line is somewhat fatter near the middle but lacks the broad wings of the Lorentzian line. A Gaussian line and its ﬁrst derivative are shown in Figures 5.
5. the instantaneous chemical environment may vary from one radical to another because of diﬀerent degrees of solvation or ion pairing. when the loss of magnetization phase coherence in the xyplane is a ﬁrstorder process.102
Chapter 5
Simulations spanning all three exchange regions are shown in Figures 5. when the magnetic ﬁeld is inhomogeneous and varies over the sample. aP ¼ 3.7. not all the molecules are at resonance at the same nominal ﬁeld.52 G at 240 K.6. For example. however.
.5 and 5. For example. one having two equivalent P nuclei (1:2:1 triplet spectrum).3 and 5. Some sources of line broadening.4 Applications of the Modiﬁed Bloch Equations
ESR spectra of the radical anion of bis(diphenylphosphino)maleic anhydride (BMA). so that the resonance line will also have a Gaussian shape. indicating a dynamical equilibrium between two isomeric forms of the radical. Gaussian line shapes are commonly observed in frozen solution ESR spectra. Unresolved hyperﬁne couplings also often give a resonance line with an approximately Gaussian shape. Comparison with Figures 5.
5.4. It is usually reasonable to guess that the ﬁeld has a Gaussian distribution (i. These spectra provide a good example of dynamical linewidth eﬀects.e.. a chemical reaction meets this criterion. i. a normal error distribution).. This radical shows a 1:2:1 triplet spectrum at higher temperatures. As we have seen. The lines broaden with increasing temperature and coalesce at about 220 K. Probably for such reasons. but the phosphorus coupling is strongly temperature dependent and extra lines appear at 200 K and below.1 and 5.3 Further Discussion of Line Shapes
Lorentzian line shapes are expected in magnetic resonance spectra whenever the Bloch phenomenological model is applicable. cannot be thought of as ﬁrstorder rate processes. The Gaussian line shape function can be written: SðoÞ ¼ eÀðoÀo0 Þ
2
=2d2
ð5:21Þ
so that the halfwidth at halfheight is pﬃﬃﬃ Half width ¼ 2 ln 2 d ¼ 0:980 d and the derivative width (between extrema) is exactly 2d. the other two nonequivalent P nuclei (doublet of doublets spectrum). Various other eﬀects can lead to such inhomogeneous broadening and thus to Gaussian line shapes.and hyperﬁne matrix anisotropies through molecular tumbling (rotational diﬀusion) in solution.e.3 are shown in Figure 5. but so do several other line broadening mechanisms such as averaging of the g.
4
2
0 (ω−ω0)/δ
2
4
Figure 5.ESR Kinetic Studies
103
4
2
0 (ω−ω0)/δ
2
4
Figure 5.6 Firstderivative Gaussian line.
These data can be understood in terms of two rotational isomers of BMAÀ (1 and 2).5 Gaussian absorption line.
O O C C Ph2P
1
C C
O
O C C
O C C Ph2P
2
O
PPh2
Ph2P
.
104
Chapter 5
280 K
260 K
240 K
220 K
200 K
180 K
160 K
3365
3370 3375 Magnetic Field/Gauss
3380
Figure 5.7
Isotropic ESR spectra of the BMA radical anion in THF solution at various temperatures. (Reproduced with permission from ref. 3, copyright (1998) American Chemical Society.)
Detailed analysis led to the thermodynamic and kinetic parameters: DH1 ¼ 0.8 Æ 0.2 kJ molÀ1, DS1 ¼ –4 Æ 1 J molÀ1 KÀ1 (K160 ¼ 3.0); DHw ¼ 18.2 Æ 0.4 kJ molÀ1, DSw ¼ À30 Æ 2 J molÀ1 KÀ1 (k200 ¼ 1.9 Â 106 sÀ1). A very similar application of the modiﬁed Bloch equations was based in the work of Adams and Connelly.4 ESR spectra (Figure 5.8) of [Mo{P(OMe)3}2(MeCRCMe)Cp] show the expected triplet (two equivalent 31 P nuclei) at 280 K, but only a doublet at 160 K. At intermediate temperatures, the lines broaden. The interpretation is that the alkyne undergoes a pendulum oscillation, which in the extrema diverts spin density from one or the other phosphite. Interestingly, the diamagnetic cation undergoes a similar motion on the NMR time scale, but then the alkyne undergoes a complete rotation. Thus, analysis of the eﬀect leads to a measure of the rate of the oscillation. The
ESR Kinetic Studies
240 K 220 K 220 K 180 K
105
3340 3360
3340
3360 3340 3360 Magnetic Field/Gauss
3340
3360
Figure 5.8 ESR spectra of [Mo{P(OMe)3}2(MeCRCMe)Cp] in 2:1 THF–CH2Cl2. (Reproduced with permission from ref. 4, copyright (2001) Royal Society of Chemistry.)
(a)
(b)
(c)
3000 B/Gauss
3500
Figure 5.9 (a) Experimental spectrum of (Ph2C2)Co(CO)[P(OMe)3]2 in THF solution at 270 K; (b and c) Computersimulated spectra: (b) the mCo and mP line width dependence and (c) the mCo line width dependence only. (Adapted from ref. 5 with permission, copyright (1984) American Chemical Society.)
resulting kinetic parameters are DHw ¼ 13.2 Æ 0.3 kJ molÀ1, DSw ¼ À14 Æ 2 J molÀ1 KÀ1, k298 ¼ 1.0 Â 109 sÀ1. Casagrande et al.5 have used linewidth eﬀects to study the rate of ﬂuxionality in (Ph2C2)Co(CO)[P(OMe)3]2. The experimental spectrum (Figure 5.9a),
106
59
Chapter 5
can be described as a 1:2:1 triplet of octets (I ¼ 7/2 for Co); the spectrum is complicated by a large line width dependence on mCo, but, as demonstrated in Figure 5.9(b) and 5.9(c), the central lines of the triplets are much broader than the outer lines. This radical has a distorted tetrahedral structure with the singly occupied molecular orbital (SOMO) largely cobalt 3 dz2 in character.6 Thus the ligand sites can be described as axial or equatorial relative to the unique zaxis. Several isomers are possible, but the 31P couplings distinguish between the isomer with an axial phosphite (ax,eq) and those with either CO or the acetylene axial and both phosphites equatorial (eq,eq). The rate of interconversion between (eq,eq) and (ax,eq) isomers (3 and 4) was estimated from the relative widths of the mP ¼ Æ 1 and 0 lines, given the isotropic coupling constants for the various 31P nuclei (which were determined from the frozen solution spectrum5). The average rate was found to be approximately 2 Â 1010 sÀ1 (Ea ¼ 17 Æ 2 kJ molÀ1) at 298 K.
z Ph C C Ph P(OMe)3 Co P(OMe)3 C O (ax,eq) Ph C C Ph z O C Co P(OMe)3 P(OMe)3 (eq,eq)
3
4
F. A. Walker et al. studied the rate of base exchange with VO(acac)2 in benzene solution:7 VOðacacÞ2 þ Base ! VOðacacÞ2 Á Base where Base ¼ pyridine, 3picoline, 4picoline, 3,4lutidine, 3,5lutidine, and piperidine, by using ESR line broadening. Line widths were ﬁtted to a powerseries expression:
À1 T2 ¼ a þ bmI þ gm2 þ dm3 I I
Table 5.1 shows the spin Hamiltonian parameters for some of the complexes. The equilibrium constants and rate constants are given in Table 5.2.
Table 5.1
ESR parameters for VO(acac)2 Á Base7
g 104AV (cmÀ1) 99.5 96.7 96.9 1.9693 1.9697 1.9690
VO(acac)2pyridine VO(acac)22picoline VO(acac)2piperidine
Parameters taken from ref.78b 1400 kr (sÀ1) 1. 7a with permission.3 Â 107 7.ESR Kinetic Studies
107
Table
5.10). copyright (1966) American Institute of Physics.
τ = 2 × 10 11 s
τ = 2 × 10 10 s
τ = 2 × 10 9 s
50
40
30
20
10
0
10
20
30
40
50
(B – B0)/Gauss
Figure 5.2 Â107 o0.2 Â 107 kr (MÀ1 sÀ1) 1.)
T (1C) 18 16 17 K (MÀ1)a 78 0.5 Alternating Line Width Eﬀects
Several examples were found during the 1950s and 1960s of spectra of organic radical ions in which the line widths were alternatively sharp and broad. One example of this is the ESR spectrum of dinitrodurene (5) (Figure 5.0 Â 109 5. 7b.0 Â 106 o2.
CH3 H3C NO2
O2N CH3 5
CH3
. Using estimated DH1.8 Â 109
Base Pyridine 2Picoline Piperidine
a b
Ref.10 Simulations of the ESR spectrum of the dinitrodurene radical anion.
5. (Reproduced from ref.2
Equilibrium and rate constants for VO(acac)2 . 8. Base exchange.
Moro and Freed15 developed an algorithm for the analysis of ESR spectra to give motional
. Another example of alternating line width eﬀect was found in the spectra of durosemiquinone (6).
This spectrum was at ﬁrst quite puzzling. A a A>.10 show the eﬀect of rate on the shape of the spectrum. Still more examples are mentioned in the reviews by Atkins in the early 1970s.3. The rates depend on the alkali metal as shown in Table 5.11. DME ¼ dimethoxyethane (ref. in the present case.7 10. the other not).12 where the eﬀect is due to asymmetric solvation (one nitro group solvated.
CH3 H3C O
O CH3 6
CH3
Another example is the mdinitrobenzene anion radical in aqueous solution. The simulations in Figure 5. the explanation became clear. but.13
5. on reﬂection.1 Ea (kJ molÀ1) 24 22 20 12 28
Cation/solventa K1/THF Rb1/THF Cs1/THF Na1/DME K1/DME
a
THF ¼ tetrahydrofuran (ref. In this case the mean lifetime of one solvation state was 0.8 The unpaired electron is mostly on the electronwithdrawing nitro groups in the pdinitrobenzene radical anion but.6 Spin Labels
One of the many advances in ESR spectroscopy introduced by Harden McConnell is the idea of labeling a biological membrane or macromolecule with a stable free radical (typically a nitroxide). Thus the unpaired electron hops back and forth between the two nitro groups.3
Eﬀect of cation and solvent on hopping rate in durosemiquinone
Log A 10. the methyl groups do not allow the two nitro groups to be coplanar with the ring simultaneously.8 ms at 291 K and 4. an eﬀect very similar to that with dinitrodurene anions.5 ms at 282 K.14 Since in a nitroxide the spin is primarily in a nitrogen 2p orbital.9.9 11. 9). with the one having the unpaired electron coplanar with the ring. 10). and the isotropic line widths give a good indication of the freedom of motion of the nitroxide.10 where the eﬀect is due to alkali metal ions hopping back and forth from one oxygen atom to the other.108
Chapter 5
Table 5.
ESR Kinetic Studies
109
lifetimes.17 studied phase transitions in E. which provided early support for the method.
O O N
H3C(H2C)12 7
(CH2)3COOH
They observed abrupt changes in the slope of Arrhenius plots for reactions catalyzed by NADH oxidase and plactate oxidase that correlate well with phase transitions detected by the ESR spectra of the nitroxide spin labels bound covalently to the enzymes (Table 5. We mention here only two examples. Another example comes from the work of Johnson.11.4). including 5doxyl stearate (5DS) (7). et al.
O
C6H13 N H
8
O
. which is sensitive to rotational motion about the NO bond. Rather diﬀerent results were obtained with the two spin labels. coli membranes using three diﬀerent spin labels. Morrisett. Electron Spin Resonance and other monographs. ESR spectra appeared as powder patterns rather than isotropic spectra and special methods were needed to extract the motional data. The subject has been reviewed nearly annually in the Specialist Periodical Report. In addition to 5DS.18 These workers studied spin labels dissolved in lipid bilayer dispersions of dipalmitoylphosphatidylcholine and cholesterol (9:1 by weight) in the hope that anisotropic rotational diﬀusion of the spin label would mimic the motion of the bilayer components. largely as a result of the diﬀerent axes of rotation. et al. ESR measurements were carried out at both 9 and 35 GHz and at temperatures ranging from –30 to 30 1C.16 The eﬀects of decrease in the rate of rotation motion on the appearance of nitroxide ESR is shown qualitatively in Figure 5. they used the steroidal nitroxide 8. Because the rotation rates were very slow. which tends to rotate about an axis perpendicular to the N–O bond. Applications of spin labels to problems in structural biology have continued to grow over the four decades since McConnell’s original proposal.
Casagrande. Weissman.0 31. T2e ¼ transition endpoint.W. J.J.G. Wang and K. 2. Chem. L. Chen. Am.. Simpson and S. Rev. Robinson. C.R.H. Soc. Broadening becomes even more pronounced and nonuniform as the rate is further decreased.4
Transition temperatures for enzyme activity and nitroxide motion17
Activity Activity pLactate oxidase 31 36 ESR Phase trans. N. 4849.
Table 5. Rieger.H.0
T2b (1C)a T2m (1C)a T2e (1C)a
a
NADH oxidase 27 32
T2b ¼ transition beginning. 1946.H. 2458. 460.110
Chapter 5
fast
slower
3330
3335
3340
3345
3350
3355
Magnetic Field/Gauss
Figure 5. 70.L. Inorg.J.0 33. B.
References
1. Commun. R. temp (1C) 29. 1984. J. Inorg.
. Nelson. 1998. N. 23. A. Adams. Chem. Robinson. Yang. 4.G. Visco. F.R. P. R.11 Eﬀect of a slowing of the rate of rotational motion on the simulated ESR spectrum of a typical nitroxide spin label. 5.. J.L. 1954.I. Phys. Chem. T2m ¼ transition midpoint. 76. Bloch. Connelly and P. Chem. 3612. Richmond.H. T. Tyler.. Ward and S. P. D.H. 2019. Duﬀy. Rieger. M. Rieger..V. Rieger. 3..C. 2001. B. 37.
J.. Smith. Faraday Soc. 21.S. D. 16. Freed. B. Biochemistry. 1981..D. B. 1965 87. Fraenkel.. J. 9. Atkins. London. Brustoon and C.E.A. Acad. 2003.L. Chem. Peake. 1983. ch 2. Phys.T. ibid.C. 6969. J. 4459. 64. 3757. C. L. 1971.H.
. Tordo. ibid. J. 1976.M. pp. I. 250. 1977. Berliner. Lai. 1987. 2128. 1973. 1983. Buckman. 15. Wakil. 17. McConnell. 3565. 3. Royal Society of Chemistry. Gill and T.L. ibid. DeGray. 4181.H. Symons. Peake. London. Chem. 223. Spin Labeling Theory and Applications. 1975. 2. Lai. 1984. 16. Academic Press.E. J. H. Am. ibid. ibid. 1976.. Soc. Lai. 13. Electron Spin Resonance.A. 1.. New York. (b) R. 293. Rieger. Q. U. Sci. B. H. B. J.. 10A. 47. 1981. Chem. 116. 11. 37. 7.. 74. T. J. Electron Spin Resonance. Chem.L. C. 83. D. Faraday Trans. Nat. Johnson. J. J. Phys. ibid. Walker. L. Plumlee.R.S. ibid. 1962.S. 54. 14. R. 1993. T.J. ibid. Fraenkel. Lai. M. ed. Specialist Periodical Report. 346.K. Soc. 1997.C. 52. 9. Davies. C. M.ESR Kinetic Studies
111
6. Stone. Trans. R. Freed. M. Chem.S. 1962. Knowles and B. Royal Society of Chemistry. Phys. 3. 1976. Chem. Corvaia. ibid. 74. 2.H. ibid. Rieger and G. Raynor. Faraday Soc.W. 256. Robinson and A.S. 1974. 1980. 37.K. Eseahani and S. 1010.. 12. B. P. (a) F. Beth. Carlin and F. B. 378. T. Proc. M Peake. M. J.S. P. 1982. II.M. Z. ibid. Freed and G.E. Lee and L. Rieger. 75.. 8. W. Jones and M. 4. Pownall. 7. 12A. J. Faraday Trans.H. 1974. 1978. Cambridge.J. 3157. Morrisett. 18. Pasimeni. J. 166. Meng and P. 10. 1977. 5.. J. Robinson. L. J. 1881.H. Electron Paramagnetic Resonance 18. Biol. 1966 45. 212. 295. 1972. 4. M. UK. 1987. P. Carlin and P. H. 47–75. 35. Royal Society of Chemistry.J. G. Marsh. 14.H. 233.A. P. Walker. H. 6.A. Gough. 1965. 1. 8. Chem. Trans. Soc. 1156. J. Raynor. F. P. P. 1997. 1968. Fung.. Chem. Phys. L. 67. Moro and J. Specialist Periodical Report. Lund and J. 8. 246. 1991. C. 116. 256. 961. Zehner. ibid. Nordio and H.
repulsion leads to the singlet–triplet splitting. 1). A second class of molecules with two unpaired electrons has the two electrons localized. or the excited triplet states of naphthalene or benzophenone. and certain binuclear vanadium(IV) and copper(II) complexes. etc.1 Biradicals
Molecules with two or more unpaired electrons may be divided into two classes: by far the most common examples are molecules where the unpaired electrons are contained in a set of degenerate atomic or molecular orbitals with qualitatively similar spatial distributions. triradicals. and other S41/2 Systems
6.CHAPTER 6
ESR Spectra of Biradicals.
112
. e. We refer to such molecules as biradicals (or. the distinction between molecules with one unpaired electron and those with more than one lies in the fact that electrons interact with one another. J S1 Á S2 .g. a ground state triplet molecule like O2.. in diﬀerent parts of the molecule. these interactions lead to additional terms in the spin Hamiltonian and additional features in the ESR spectrum. with three or more unpaired electrons. As we will see.g.). Examples are the dinitroxides (e. ~ ~ this eﬀect can be modeled by adding a term. to a ﬁrst approximation. an octahedral Cr(III) (4A2g) or Ni(II) (3A2g) complex. to the spin Hamiltonian. Triplet States. with two unpaired electrons. The most important electron–electron interaction is coulombic repulsion.
O
N
N
O
1
From the point of view of ESR spectroscopy.
The spin Hamiltonian for a biradical consists of terms representing the electron Zeeman interaction. Triplet States. J falls oﬀ with electron–electron distance as 1/r. etc. The spin Hamiltonian is: ~ ~ ~ ~ ~ ~ Hs ¼ gmB B ðS 1z þ S 2z Þ þ A ðS1 Á I1 þ S2 Á I2 Þ þ J S1 Á S2 ð6:2Þ
where J is the exchange coupling constant.ESR Spectra of Biradicals. and hyperﬁne interaction of each electron with the nuclear spins. 1 2 2 2 2 2
. The simpliﬁcation of being able to ignore the singlet state is more than compensated for by the introduction of a ﬁne structure term into the spin Hamiltonian. focusing entirely on the triplet. which has the form shown in eqn (6. the singlet–triplet splitting is small and (as we will see) singlet–triplet mixing has an eﬀect on the ESR spectrum. it is possible to ﬁnd triradicals. Notice that we have also assumed that the gvalues for the two electrons are the same. each strongly coupled to one electron and essentially uncoupled to the other. D is negligible (though spin–orbit eﬀects may contribute). the singlet–triplet splitting is large compared with kT. To simplify matters. In many cases.
6. Obviously. as well as any other term in the spin Hamiltonian. i. The distinction between a biradical and an ordinary triplet state molecule is often somewhat fuzzy. This term. À1 À À1.. We choose as basis functions the singlet and triplet electron spin functions: À Á 1 ð6:3aÞ jS0 i ¼ pﬃﬃ 1. For our purposes. whereas the dipolar coupling constant D falls oﬀ as 1/r3. treatment of such species is similar. When the exchange interaction is weak. though of course somewhat more complicated. Â 2 Ã 2 2 D SZ À 1 SðS þ 1Þ þ E Sx À Sy ð6:1Þ 3 Further complicating the situation is the fact that the same term can arise from two quite diﬀerent physical eﬀects: electron–electron dipolar interaction and spin–orbit coupling. and other S41/2 Systems
113
where J is called the exchange coupling constant and turns out to be equal to energy diﬀerence between the singlet and triplet states. tetraradicals. we consider a molecule a biradical if the exchange interaction between the two electrons is relatively weak – comparable in energy to the electron–nuclear hyperﬁne interaction. the singlet–triplet splitting. Thus when J is small. we will assume that A { gmBB so that a ﬁrstorder treatment of the hyperﬁne term will suﬃce.1
Exchange Coupling
In this section we consider the spin Hamiltonian appropriate to a biradical with weak dipolar coupling and see how ESR spectra of such species should appear.1. We assume that there are two equivalent nuclei.e.1) introduces considerable complication into the shape and interpretation of ESR spectra. the exchange coupling of the two electron spins. and we can safely ignore the singlet state.
deﬁning: R¼ we have: J R E¼À Æ 4 2 qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ J 2 þ A2 ðm1 À m2 Þ2 qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ J 2 þ A2 ðm1 À m2 Þ2
. A ¼ 0. however. then T0i and S0i would be eigenfunctions as well.114
Chapter 6
À Á 1 jT0 i ¼ pﬃﬃ 1. that if there were no hyperﬁne coupling. S ¼ 0. we get: Â Ã ^ H s jT1 i ¼ gmB B þ 1J þ 1A ðm1 þ m2 Þ jT1 i 4 2 Â Ã ^ H s jTÀ1 i ¼ ÀgmB B þ 1J À 1A ðm1 þ m2 Þ jTÀ1 i 4 2 ^ H s jT0 i ¼ 1J jT0 i þ 1A ðm1 À m2 ÞjS0 i 4 2 ^ H s jS0 i ¼ À3J jS0 i þ 1A ðm1 À m2 ÞjT0 i 4 2 ^ Thus T1i and TÀ1i are eigenfunctions of H s . J is normally negative so that the triplet lies lower in energy. 1 2 2 jTÀ1 i ¼ À1. À1 þ À1.) In the absence of a hyperﬁne interaction. the triplet functions correspond to S ¼ 1. (Notice. À1 2 2
ð6:3bÞ ð6:3cÞ ð6:3dÞ
The singlet function corresponds to zero total electron spin angular momentum. 1 2 2 2 2 2 jT1 i ¼ 1. Operating on these functions with the spin Hamiltonian. To get the eigenvalues resulting from the admixture of T0i and S0i. the triplet energy is J/4 and the singlet energy is À3J/4. but T0i and S0i are mixed. we solve the secular equation: J A ÀE 4 2 ðm1 À m2 Þ A ðm1 À m2 Þ À 3J À E 2 4
3 ¼ E 2 þ 1JE À 16J 2 À 1A2 ðm1 À m2 Þ2 ¼ 0 2 4
The roots are: E ¼ À 1J Æ 1 4 2 or.
ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
115
The eigenfunctions corresponding to these energies may be found by inserting a value of E into one of the linear equations that lead to the secular equation: À1 Á 1 4J À E cT þ 2Aðm1 À m2 Þ cS ¼ 0
1 2Aðm1
À Á À m2 Þ cT þ À3J À E cS ¼ 0 4
where cT and cS are the coeﬃcients of T0i and S0i in the eigenfunction corresponding to E(c2 þ c2 ¼ 1). The resulting eigenfunctions and T S energies are: E1 ¼ gmB B þ J R E2 ¼ À þ 4 2 J R E3 ¼ À À 4 2 J A þ ðm1 þ m2 Þ j1i ¼ jT1 i 4 2 ﬃ rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ RþJ RÀJ j 2i ¼ jT0 i þ jS0 i 2R 2R ﬃ rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ RÀJ RþJ j 3i ¼ jT0 i À jS0 i 2R 2R
J A À ðm1 þ m2 Þ j4i ¼ jTÀ1 i 4 2 Remembering that each of these is further split by the hyperﬁne interaction, there are obviously several possible transitions among these four energy levels. To ﬁnd out which are important, we must evaluate the transition dipole moment matrix elements, hiSxji, since the absorption intensity is proportional to the square of these matrix elements. The operator Sx can be written: ^ ^ ^ ^ ^ ^ ^ S x ¼ S 1x þ S 2x ¼ 1 S 1þ þ S 1À þ S 2þ þ S 2À 2 E4 ¼ ÀgmB B þ Applying Sx to 1i and 4i, we have: ^ ^ ^ ^ ^ S x j1i ¼1 S 1þ þ S 1À þ S 2þ þ S 2À 1; 1 2 2 2 À 1 1Á pﬃﬃﬃ 1 1 1 ;À ¼2 À2; 2 þ 2 2 ¼ 2jT0 i ^ ^ ^ ^ ^ S x j4i ¼1 S 1þ þ S 1À þ S 2þ þ S 2À À1; À1 2 2 2 À Á pﬃﬃﬃ ¼1 1; À1 þ À1; 1 ¼ 2jT0 i 2 2 2 2 2 Thus the matrix elements are: D E D E D E D E ^ ^ ^ ^ 1S x 1 ¼ 4S x 4 ¼ 1S x 4 ¼ 4S x 1 ¼ 0 ﬃ D E D E D E D E rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ RþJ ^ ^ ^ ^ 2S x 1 ¼ 2S x 4 ¼ 1S x 2 ¼ 4S x 2 ¼ R ﬃ D E D E D E D E rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ RÀJ ^ ^ ^ ^ 3S x 1 ¼ 3S x 4 ¼ 1S x 3 ¼ 4S x 3 ¼ R
116
Chapter 6
Since the relative intensity of a transition is proportional to the square of the ^ corresponding matrix element of S x , we see that there are four allowed transitions: 1 1 DE42 ¼ gmB B þ ðR À JÞ þ Aðm1 þ m2 Þ 2 2 RþJ Relative intensity ¼ R 1 1 DE43 ¼ gmB B À ðR þ JÞ þ Aðm1 þ m2 Þ 2 2 RÀJ Relative intensity ¼ R 1 1 DE31 ¼ gmB B þ ðR þ JÞ þ Aðm1 þ m2 Þ 2 2 RÀJ Relative intensity ¼ R 1 1 DE21 ¼ gmB B À ðR À JÞ þ Aðm1 þ m2 Þ 2 2 RþJ Relative intensity ¼ R Consider now the limiting case of strong exchange coupling. When J 44 A, R E J, and the 43 and 31 transitions are forbidden. The 42 and 21 transitions are at equal energy and so we have only: DE ¼ gmB B þ 1 Aðm1 þ m2 Þ 2 ð6:4Þ
Thus in the limit of strong exchange interaction, the resulting spectrum is identical to that which would be observed if one electron interacted with two equivalent nuclei with coupling constant A/2. In the limiting case of weak coupling, J { A, R E A(m1 À m2), and all four transitions have equal relative intensities; the transition energies then are: DE42 ¼DE31 ¼gmB B þ A A ðm1 À m2 Þ þ ðm1 þ m2 Þ 2 2 ¼gmB B þ Am1 ð6:5aÞ
DE43 ¼DE21 ¼gmB B À A A ðm1 À m2 Þ þ ðm1 þ m2 Þ 2 2 ¼gmB B þ Am2 ð6:5bÞ
Thus in the limit of negligible exchange interaction, we expect a spectrum identical to that observed for two independent radicals.
ESR Spectra of Biradicals, Triplet States, and other S41/2 Systems
117
Figure 6.1
Stick spectra for a dinitroxide biradical with a ¼ 13 G for various values of the exchange coupling constant J. (Several very small, widely spaced resonances have been omitted for J ¼ 2A and J ¼ 5A.)
Now consider a concrete example. Suppose we have a nitroxide biradical with aN ¼ 13 G. In the strong exchange limit, we expect a ﬁveline spectrum with a spacing of 6.5 G and the usual 1:2:3:2:1 intensity ratios for two equivalent spin1 nuclei. In the weak exchange limit, we expect a threeline spectrum with a spacing of 13 G and intensity ratios 1:1:1. In intermediate cases, up to 15 lines are expected, as shown in Figure 6.1. In a classic early study of biradicals, Glarum and Marshall1 were able to vary the exchange coupling between two nitroxide radicals by changing the temperature, solvent, and the number of atoms linking the two groups. Most of their spectra corresponded to the intermediate exchange region where J E A.
6.2 Organic Triplet State Molecules and the Dipolar Interaction
The Hamiltonian term for the electron–electron dipolar interaction is: ! s1 s2 s1 rÞ s2 rÞ ^ d ¼ g2 m2 ~ Á ~ À 3 ð~ Á ~ ð~ Á ~ H B r5 r3 ð6:6Þ
these axes will also be the principal axes of the gmatrix and we will so assume in the following.118
Chapter 6
where r is the vector pointing from electron 1 to electron 2. deﬁning the matrix D with elements: ( ) 1 2 2 r2 dij À 3ij Dij ¼ g mB ð6:7Þ 2 r5 where the angle brackets indicate averaging over the spatial coordinates of the wave function. y. an operation that is not usually possible in practice. reserving uppercase S for the total electron spin operators. S ¼ s1 þ s2. The dot products can be expanded to give: À2 Á À2 Á 2 2 ^ d ¼g2 m2 r À 3x s1x ^2x þ r À 3y s1y s2y ^ s ^ ^ H B r5 r5 À2 Á Á r À 3z2 3xy À ^ ^ ^1z ^2z À 5 s1x ^2y þ s2y s1x ^ s þ s s 5 r r ! Á 3zx 3yz À ^ s ^ ^ s ^ s s À 5 s1y s2z þ s1z ^2y À 5 ð^1z ^2x þ ^1x s2z Þ r r Our next goal is to transform this expression into one based on the total electron spin operator. there are really only two independent parameters. In many cases. Since the trace of D (Dxx þ Dyy þ Dzz) is zero. However. zÞ
When these are substituted in the above expression. the terms arising from –1/2 cancel since x2 þ y2 þ z2 ¼ r2. It is possible to choose an axis system in which the Dmatrix is diagonal – the principal axes. The ﬁrst three terms can be simpliﬁed by making use of the identity (derived using raising and lowering operators): 2s1i s2i ¼ S 2 À 1 i 2 ði ¼ x. Transformation of the last three terms makes use of the identities: 2ðs1i s2j þ s1j s2i Þ ¼ S i S j þ S j S i The transformed Hamiltonian then is: 1 r2 À 3x2 ^2 r2 À 3y2 ^2 ^ H d ¼ g2 m 2 Sx þ Sz B 2 r5 r5 r2 À 3z2 ^2 3xy ^ ^ ^ ^ þ S z À 5 S xS y þ S yS x 5 r r ! 3yz ^ ^ ^z S y À 3zx S z S x þ S x S z ^ ^ ^ ^ ^ À 5 SySz þ S r r5 The coeﬃcients of the spin operators must be evaluated using the electron wave function. we can parameterize the problem. The conventional choice of these parameters is: À Á D ¼ 3 Dzz E ¼ 1 Dxx À Dyy 2 2
. We have used a lowercase s for the oneelectron spin operators.
Dxx ¼ Dyy ¼ Dzz so that D ¼ E ¼ 0.2. The splitting pattern is shown in Figure 6. 0i 3 À Á ^ H d j1. gmBB Á S. to do spectroscopy on these energy levels at zero magnetic ﬁeld. À1i 3 À Á ^ H d j1. of course.ESR Spectra of Biradicals. but ﬁrst we will pause to see the eﬀect of this Hamiltonian on the energy levels and ESR spectrum of a tripletstate molecule. Notice that if the molecule has axial symmetry. 1i þ E j1. The spin triplet wave functions can be written in the notation S.mSi: À Á ^ H d j1. the
. Thus the appearance of a zeroﬁeld splitting into two or three levels tells the spectroscopist something about the symmetry of the molecule. À2D/3 and D/3 Æ E. À1i ¼ D 1 À 2 j1. It is possible. 1i ¼ D 1 À 2 j1. When we include the Zeeman interaction term. 1i 3 The Hamiltonian matrix then is: 01 3D 0 @0 À2D 3 E 0 1 E 0 A 1 3D
Solution of the corresponding secular equation leads to energy levels. When we have a strong dipolar interaction. Dxx ¼ Dyy so that E ¼ 0. À1i þ E j1. Triplet States. in the spin Hamiltonian a complication arises. If the molecule has octahedral symmetry. Our concern here is the eﬀect of zeroﬁeld splitting on the ESR spectrum where a magnetic ﬁeld is applied. and other S41/2 Systems
119
or: Dzz ¼ 2 D 3 Dxx ¼ À1D þ E 3 Dyy ¼ À1D À E 3
The Hamiltonian then becomes: ^2 ^2 ^2 ^ H d ¼ Dxx S x þ Dyy S y þ Dzz S z or: h 2 2 i 2 ^ ^2 ^2 ^ ^2 ^ ^ H d ¼ D S z À 1 S x þ S y þ S z þ 1E S x þ S y 3 2
2
ð6:8aÞ
ð6:8bÞ
We can somewhat simplify the Hamiltonian by noting that Sx2 þ Sy2 þ Sz ¼ S2 and the eigenvalue of S2 is S(S þ 1). We have been accustomed to evaluating the dot product by simply taking the direction of the magnetic ﬁeld to deﬁne the zaxis (the axis of quantization). 0i ¼ D 0 À 2 j1. and that Sx and Sy can be written in terms of the raising and lowering operators: h 2 i 2 ^ ^ ^2 ^ H d ¼ D S z À 1SðS þ 1Þ þ E S þ þ S À ð6:9Þ 3 We will see that a Hamiltonian term identical in form also arises from spin– orbit coupling.
1003 cmÀ1. but the algebra is very messy and not very enlightening. E ¼ À0.2 The energies of the allowed transitions are: qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ DE ¼ g2 m2 B2 þ E 2 Æ D B
. Thus.0137 cmÀ1 – parameters appropriate to the excited triplet of naphthalene.2 Zeroﬁeld splitting of a triplet state.120
Chapter 6
Figure 6.3. the B Á S term has three components. Suppose that we orient the crystal with B in the zdirection. 1D Æ g2 m2 B2 þ E 2 B 3 3 These are plotted vs. It is possible to deal with the general case. The spin Hamiltonian then is: h 2 i 2 ^ ^ ^ ^2 ^ H s ¼ gmB BS z þ D S z À 1SðS þ 1Þ þ 1E S þ þ S À 3 2 Operating on the triplet wave functions as before. Instead we will assume that the triplet molecule is in a dilute single crystal and that we can orient the crystal in the ﬁeld with B along one of the internal coordinate axes. in general. magnetic ﬁeld in Figure 6. using D ¼ 0.
molecule deﬁnes a quantization axis for itself. we get the Hamiltonian matrix: 0 1 gmB B þ 1D 0 E 3 @ A 0 À2D 0 3 1 E 0 ÀgmB B þ 3D Solution of the secular equation leads to the energies: qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ À2D.
0137 cmÀ1. Triplet States.50 GHz microwave radiation. the resonant ﬁelds are: qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 1 B¼ ðhn 0 Æ DÞ2 ÀE 2 gmB With the magnetic ﬁeld oriented along the xaxis. Solid lines correspond to orientation of the magnetic ﬁeld along the zaxis.
. À Æ g2 m2 B2 þ ðD þ EÞ2 B 3 6 4 These energies are also plotted in Figure 6. the Hamiltonian is: h 2 i 2 ^ ^ ^ ^2 ^ H s ¼ gm BS x þ D S À 1SðS þ 1Þ þ 1E S þ S
B z 3 2 þ À
The Hamiltonian matrix is: 01 3D B pﬃﬃgm B 1 @ 2 B E
1 pﬃﬃgm B B 2 À2D 3 1 pﬃﬃgm B B 2
1 E 1 pﬃﬃgm B C B A 2 1 D 3
The cubic secular equation factors.3 Energy level diagram for the triplet state of naphthalene (D ¼ 0. dashed lines for orientation along the xaxis.ESR Spectra of Biradicals.1003 cmÀ1. E ¼ À0. the resulting energies are: rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ D D À 3E 1 À E.3. n0 ¼ DE/h. Arrows show the allowed transitions for 9.003). and other S41/2 Systems
121
At constant frequency. energies of the allowed transitions are: qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ DE ¼ g2 m2 B2 þ 1ðD þ EÞ2 Æ 1ðD À 3EÞ B 4 2 The resonant ﬁelds then are: sﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 2 1 1 1 hn 0 Æ ðD À 3EÞ À ðD þ EÞ2 B¼ gmB 2 4
Figure 6. g ¼ 2.
1
Organic Triplet State Molecules
The phosphorescent triplet state of naphthalene. Since D is a measure of the dipolar interaction of the two unpaired electrons. When the psystem gets bigger. for example.50 GHz). This prevents relaxation eﬀects due to rapid transfer of energy between triplet and ground state naphthalene molecules. Thus for any orientation there are four resonances. and anthracene and pyrene fulﬁll this expectation. but phenanthrene seems anomalous.
6. is that there is no way to assign the features to molecular axes. and at 2595 and 4125 G for orientation along the xaxis (n0 ¼ 9.122
Chapter 6
6.1 Spin–Orbit Coupling
We now will show that spin–orbit coupling can give a spin Hamiltonian term identical to that we obtained from the electron dipolar interaction. The features of the derivative spectrum correspond to orientations along the principal axis directions. we might expect D to decrease. has two molecules per unit cell with diﬀerent orientations of the molecular plane. Consider the
. Since the resonances are spread over a large ﬁeld range and are very orientation dependent. there is little hope of detecting the resonance of a triplet state molecule in liquid solution. even when the crystal axes are known. The experimental determination of D and E for a dilute single crystal is not trivial. and we see that naphthalene and related nitrogen heterocyclics do indeed have nearly the same values of D. even if the triplet state lifetime could be made long enough. is produced by irradiation of naphthalene doped into a single crystal of durene.1.2. and naphthalene and the related nitrogen heterocyclics do indeed give the largest values of D.2 Triplet state powder spectra (or frozen solution glasses) are generally easier to interpret and much easier to get experimentally than dilute single crystal spectra.3 Transition Metal Complexes with S41/2
6. Indeed electron dipolar interaction parameters provide one of the more challenging tests of valence theory. Sorting out the data is a challenging exercise. What do we make of the parameters D and E once we have extracted them from a spectrum? Seven examples are given in Table 6. recourse must be made to theoretical considerations or to analogy with a related system studied in a dilute single crystal.3. two from each type of site. When the psystem remains about the same size but heteroatoms are substituted. as is usual with powder spectra. The energylevel diagram (Figure 6. The triplet state is longlived at 77 K. for which the energy levels are shown in Figure 6. The problem. we might expect that D would not change by much. To go beyond a qualitative explanation of D or to explain E at all requires rather sophisticated valence theory calculations. Thus. six features can be found in the spectrum of naphthalene in glassy THF solution at 77 K (after irradiation). There are some qualitative trends that make some sense.3. Durene. we might expect that D would be large when both electrons are forced to be close together.3) predicts two transitions: at 2315 and 4465 G when the ﬁeld is oriented along the zaxis.
1003 E (cmÀ1) –0.0466
6
Pyrene (in ﬂuorene)
0. Triplet States. and we
.0716
–0. S.0678
–0.0137 Ref 2
Naphthalene (in durene)
Quinoline (in durene)
N
0.0162
3
Isoquinoline (in durene)
N
0.1007
–0.1
Molecule
ESR data for some organic tripletstate molecules
Structure D (cmÀ1) 0.1004
–0. The zeroorder ground state wave function will be characterized by the quantum numbers L.ESR Spectra of Biradicals.0084
5
Phenanthrene (in biphenyl)
0.0182
4
Anthracene (in biphenyl)
0.0314
7
spin Hamiltonian including orbital angular momentum and the usual spin– orbit coupling term: ^ ~ ~ ~ ~ ~ H ¼ mB B Á L þ g e S þ l L Á S ð6:10Þ where l is the spin–orbit coupling constant. mS.0117
3
Quinoxaline (in durene)
N N
0.1030
–0. and other S41/2 Systems
123
Table 6.1004
–0. mL.
mSi elements connecting with excited state functions: h i2 ~ ~ ~ ~ ~ X 0. i.i ð6:12Þ E ð2Þ ¼ À ð1Þ ð1Þ Ei À E0 i The matrix element can be expanded and written as: E D ~ ~ ~ 0.i Á 0LmL.. mS. the energy is easily found to ﬁrst order in perturbation theory: E ð1Þ ¼ gmB Bms ð6:11Þ
since Lz0.i 0jmL.e.0 mB B þ lS mS 0 mS 0
E1 D ED ~ ~ X 0jLjmL mL jLj0 A Á@ ð1Þ ð1Þ Ei À E0 i D Eo ~ ~ Á mS 0 jm B þ lSjmS. With B deﬁning the zaxis.0 S mS.0 mB B þ lS Á L þ ge mB B Á S mL.i D E ~ ~ þ ge mB B Á mS.mSi ¼ 0. The absolute value square of the matrix element of a Hermitean operator can be written as: jhijOpjj ij2 ¼ hijOpjj ihjjOpjii Thus we can write E(2) as: E ð2Þ ¼
1 E X nD ~ ~ mS. The secondorder energy is not so simple since the excited states in general have nonzero mL.0 0
B
Deﬁning: L¼ we can write E(2) as: E ð2Þ ¼
1 Xn mS 0
$
E D ED ~ ~ X 0jLjmL mL jLj0
i
Ei
ð1Þ
À E0
ð1Þ
ð6:13Þ
D E D E $ $ ~ ~ ~ ~ m2 mS jB Á L Á BjmS 0 þ l2 mS jS Á L Á SjmS 0 B ð6:14Þ
D Eo $ ~ ~ þ2lmB mS jB Á L Á SjmS 0
.mSi. the second term vanishes. Thus we write the ground state wave function as mL. Thus we get secondorder contributions from matrix 0. h0 and mL. mS.i Since the orbital functions.0 mB B þ lS mS.124
Chapter 6
assume that the wave function is orbitally nondegenerate. mL ¼ 0.i . mS.ii are orthogonal.mSi ¼ 0.
Although the term would be important to the thermodynamic properties of the system. Triplet States. The ﬁrst and third terms can be combined to obtain the gtensor: g ¼ ge E þ 2lL
$
ð6:16Þ
where E is the unit matrix. however. the spin–orbit coupling Dtensor in general does not have zero trace. we have:
2 2 2 ~ ~ S Á D Á S ¼ Dxx Sx þ Dyy Sy þ Dzz Sz
and substituting:
h i 2 2 2 2 ~ ~ S Á D Á S ¼D Sz À 1 Sx þ Sy þ Sz 3 2 2 þ E Sx À Sy À Á 2 2 2 þ 1 Dxx þ Dyy þ Dzz Sx þ Sy þ Sz 3
Since S2 ¼ S2 þ S2 þ S2 and the eigenvalue of S2 is S(S þ 1) we have: x y z Â 2 Ã ~ ~ S Á D Á S ¼D Sz À 1SðS þ 1Þ 3 2 2 þ E Sx À Sy À Á þ 1 Dxx þ Dyy þ Dzz SðS þ 1Þ 3
. but as a ﬁrstorder result.ESR Spectra of Biradicals. it is uninteresting to spectroscopists and we will ignore it. the related Dtensor is also diagonal. and other S41/2 Systems
125
We now notice that we could write a Hamiltonian operator that would give the same matrix elements we have here. Including the electron Zeeman interaction term. Expanding the ﬁne structure term in the principal axis system. Unlike the dipolar Dtensor. We can also deﬁne the ﬁne structure tensor D as: D ¼ l2 L so that the spin Hamiltonian reduces to: ~ ~ ~ ~ H s ¼ mB B Á g Á S þ S Á D Á S ð6:18Þ
$
ð6:17Þ
Notice that the ﬁne structure term found here has the same form (and the tensor is given the same symbol) as that obtained from the electron dipolar interaction. we have the resulting spin Hamiltonian: ~ ~ ~ ^ ~ ~ ~ ~ ~ H s ¼ ge mB B Á S þ m2 B Á L Á B þ 2lmB B Á L Á S þ l2 S Á L Á S B
$ $ $ $
ð6:15Þ
~ ~ The m2 B Á LB term is independent of spin state and so changes all levels by B the same amount. Nonetheless. we introduce analogous parameters: À Á D ¼ Dzz À 1 Dxx À Dyy 2 À Á E ¼ 1 Dxx À Dyy 2 In the coordinate system that diagonalizes g.
magnetic ﬁeld in Figure 6. D and E. the ﬁne structure term is not usually included in spin Hamiltonians for S ¼ 1/2 systems. are required to completely specify the ﬁne structure interaction. the energies are: À Á E Æ3 ¼ Æ 3gjj þ D 2 À 2Á E Æ1 ¼ Æ 1gjj À D 2 2 These energies are plotted vs. However. Although it is unfortunate that spin–orbit coupling and the electron dipolar interaction give ﬁne structure terms of the same form. it is possible to separate the eﬀects.0455 cmÀ1. we have made no explicit assumption about the total electron spin quantum number S so that the results should be correct for S ¼ 1/2 as well as higher values. E ¼ 0.2
Highspin Transition Metal Ions
For axially symmetric complexes. and the spin functions S.mSi are eigenfunctions of the spin Hamiltonian: Â 2 Ã ~ ~ H s ¼ mB B Á g Á S þ D Sz À 1SðS þ 1Þ 3 For example. For the magnetic ﬁeld along the molecular zaxis.
. Æ 1 ¼ 0
4 3 2 2 2 2 2
6. In the above derivation.3.126
Chapter 6
The last term (which would be zero if D came from the dipolar interaction and thus had zero trace) raises all levels equally and so has no eﬀect on spectroscopy and can be dropped. Since the spin–orbit contribution to D is related to the gtensor: l Dso ¼ ðg À ge E Þ 2 the parameters Dso and Eso can be computed: Á l 1À Dso ¼ gzz À gxx þ gyy 2 2 Eso ¼ Á lÀ gxx À gyy 4 !
The diﬀerence between the ﬁne structure parameters computed from the experimental gtensor and those measured from the spectrum are presumed to be the electron dipolar contributions. the parameter E is zero. consider a d3 Cr(III) complex in an axial ligand ﬁeld with g ¼ 1. only two parameters.98. D ¼ 0. Æ1 2 2 2 Ã È Â É ¼ D 1 À 1 Â 1 Â 3 þ 1E Â 0 1. The ﬁne structure term can be ignored since in that case the results of operating on a spin1/2 wave function is always zero: Ã È Â 2 1 À 2 ÁÉ 2 D Sz À 3 SðS þ 1Þ þ 1 E Sþ þ SÀ 1. Thus.4. again.
the secular equation doesn’t factor and the energies must be computed numerically.À1/2 and 1/2 . j1i ¼ À 0:255 3 À À3 À 2 2 Á þ 0:660 1 À À1
2 2
1 B  z 0. and the transition from the lowest level to the highest level becomes allowed. corresponding to the lowesttohighest transition. and other S41/2 Systems
127
Transitions among these levels have intensities proportional to the square of the matrix element of Sx. this would be a forbidden twoquantum transition. As we might expect from the nondiagonal Hamiltonian matrix. A plot of the computed energies is shown in Figure 6. at hn0/gmB.3/2 transitions. have relative intensities of 3/4 the intensity of the À1/2 . D ¼ 0. These are easily found to be: 3 3 3 1 pﬃﬃ 3 2.4 as a function of magnetic ﬁeld.1/2 transition. Æ2 Sx 2.0455 cmÀ1. À2 ¼ 1 Thus the À3/2 . 2 Sx 2.5
1 0 1000 2000 3000 4000 Magnetic Field/Gauss 0 1000 2000 3000 4000 Magnetic Field/Gauss 5000
Figure 6. for the magnetic ﬁeld along the x.5 Energy/cm–1 B  x
0
0. The immediate consequence of this mixing is that the selection rules are complicated. Æ2 ¼ 2 3 1 3 1 2.and zaxes. the spin functions are thoroughly mixed when the ﬁeld is in the xdirection.) When B ¼ 1000 G. the wave functions are: À Á E ¼ À0:179 cmÀ1 . When the magnetic ﬁeld is oriented along the xaxis. (For the ﬁeld along the zaxis. the Hamiltonian matrix is: 1 0 pﬃﬃ 3 D 0 2 g? mB B 0 C B pﬃﬃ C B 3g? mB B ÀD g? mB B 0 C B 2 pﬃﬃ B0 3 g? mB B ÀD g? mB B C A @ 2 pﬃﬃ 3 0 0 g? mB B D 2 Unfortunately.ESR Spectra of Biradicals. Triplet States.4 Energy levels and allowed transitions for a Cr(III) complex with g ¼ 1.98.
. at B ¼ hn0(1 Æ 2D)/gmB.
3.3.4 transitions are still forbidden.
When the ﬁeld is along the zaxis.4 transition is only weakly allowed compared with the 1 . B in Figure 6. The transition energies are: ÀÁ ÀÁ E 5 À E 3 ¼ gk mB B þ 4D 2 2 E À3Á
2
ÀE
À1Á
2
¼ gk mB B þ 2D
E
À1Á
2
À Á À E À 1 ¼ gk m B B 2
. the Sz matrix elements also depend on the ﬁeld. Consider highspin Fe(III) in an axial ligand ﬁeld with D 44 hn0. 2 .3 and 2 . Notice that the 1 . This is not always the case.5. and 3 . 0.4 transitions. particularly in powder spectra since it tends to be considerably sharper than the other transitions. With the same Hamiltonian as above and the magnetic ﬁeld along the zaxis.128
Chapter 6
E ¼ À0:037 cmÀ1 .4 transitions would be diﬀerent than predicted from the Sz2 matrix at 1000 G. Thus the observed 1 .2. D is relatively small (comparable to the Xband microwave quantum.2. however. The Sz2 matrix for B ¼ 1000 G is: 0 1 0 1:197 0 0:0011 B 1:197 C 0 0:794 0 B C @0 0:794 0 0:508 A 0:0011 0 0:508 0 The 1 . 2 . transition intensities are proportional to the square of the Sz2 matrix element. and 3 . For Cr(III) complexes.317 cmÀ1) and all three ﬁne structure lines are observable. Since the wave functions are ﬁelddependent. the energies are: À Á E Æ5 ¼ Æ 5gk mB B þ 10D 2 8 À 2Á E Æ3 ¼ Æ 3gk mB B À 2D 2 3 À 2Á E Æ1 ¼ Æ 1gk mB B À 8D 2 2 3 These are plotted vs.
À Á j3i ¼0:660 3 À À3 2 À 2 Á þ 0:255 1 À À1 2 2 À Á j4i ¼0:487 3 þ À3 2 À 2 Á þ 0:513 1 þ À1 2 2
E ¼ þ0:137 cmÀ1 . E ¼ 0. it is often observed.
À Á j2i ¼ À 0:513 3 þ À3 À12 12 Á þ 0:487 þ À
2 2
E ¼ þ0:079 cmÀ1 .
Triplet States. Thus we can solve just the middle 2 Â 2 block for the energies of mS ¼ Æ1/2 levels: 8 3 À 3D 2g? mB B 3 8 2g? mB B À3D The expanded block gives: À 8 Á2 À Á2 À3D À E À 3g? mB B ¼ 0 2 so that the energies are: E ¼ À 8D Æ 3g? mB B 3 2
. only the À1/2 . We ought to solve a 6 Â 6 secular equation.
À Á À Á E À1 À E À3 ¼ gk mB B À 2D 2 2 À Á À Á E À3 À E À5 ¼ gk mB B À 4D 2 2 However. Since D is big.5 Energies of an S ¼ 5/2 spin system with D ¼ 0.1/2 transition will be observable.ESR Spectra of Biradicals. the Æ5/2 and Æ3/2 levels are well separated from the Æ1/2 levels before application of the magnetic ﬁeld. but we can get a reasonable approximation more easily. The large value of D ensures that z will continue to be the quantization axis. The ﬁrst two transitions are always higher in energy then hno and it is usually not possible to make B large enough to bring the last two transitions into resonance. Thus mixing of 3/2i with 1/2i will be much less important than mixing of 1/2i with À1/2i. if D 4 hno.5 cmÀ1 for B along the zaxis. Now consider what happens when the ﬁeld is applied perpendicular to the symmetry axis. and other S41/2 Systems
3 5/2 2
129
Energy/cm–1
1
–5/2
0
3/2 –3/2
1
1/2 –1/2
2 0 1000 2000 3000 Magnetic Field/Gauss 4000
Figure 6.
0i. for example. Bleaney. highspin d5 Fe(III) in an axial ligand ﬁeld should show a resonance around g ¼ 2 and another resonance near g ¼ 6 when B is perpendicular to the symmetry axis. but. À1/2i.2. geﬀ ¼ 6 if g> ¼ 2. both resonances should be observable.130
Chapter 6
and the energy diﬀerence is: E À1Á
2
À Á À E À1 ¼ 3g? mB B 2
The apparent gvalue for the transition is 3g>.3. With the ﬁeld along the zaxis.2. 1/2i. where the complex tumbles rapidly and averages the gvalues. using the Hamiltonian on the eigenfunctions gives for Cr(CN)63À: Â Ã ^ H 3 ¼ 3mB gz B þ D 3 2 2 2 Â Ã ^ H 1 ¼ 1mB gz B À D 1 2 2 2 Â Ã ^ H À1 ¼ À1mB gz B À D À1 2 2 2 Â Ã ^ H À3 ¼ À3mB gz B þ D À3 2 2 2 and for V(CN)64À: Â Ã ^ Hj1i ¼ mB gz B þ 1D j1i 3 Â Ã ^ Hj0i ¼ À2D j0i 3 Â Ã ^ Hj À 1i ¼ ÀmB B þ 1D j À 1i 3
. and Bowers determined the values given in Table 6. and À1i for V (the 51V hyperﬁne coupling is essentially isotropic and equal to À55. Thus. The spin–orbit coupling parameters found from the spectroscopic study are listed. the resonance is expected to be so broad as to be undetectable. in frozen solution. and À3/2i for Cr. in Table 6. respectively. together with the gcomponents.8 The Cr(III) and V(II) salts were doped into diamagnetic single crystals of K3Co(CN)6 and K4Fe(CN)6. and 1i.
6. Let us see what the energy levels look like for these two systems and try to understand how Baker. In solution.5 Â 10À4 cmÀ1).3
Examples: K3Cr(CN)6 and K4V(CN)6
These two salts were studied by the Oxford physics group in the early days of ESR spectroscopy. The spin Hamiltonian is: Â Ã ~ ~ ^ ~ ~ H ¼ mB S Á g Á B þ D S 2 À 1SðS þ 1Þ þ E S2 À S2 þ I Á A Á S
z 3 x y
with eigenfunctions 3/2i.
Indeed.9919
8
gy 1. With the CNÀ ligand.083 –0.0972 gx 1. so one of the goals of Baker. the values of gz and D can be determined. the sixline spectrum persists in frozen solutions. In the same paper. K3Fe(CN)6 and K4Mn(CN)6 were also studied. several workers have doped Mn(II) into crystals for which a phase transition was indicated. the ESR spectrum is very characteristic with six hyperﬁne lines with a coupling constant of 80–100 G. In part because the highspin Mn(II) signal is so easily detected. Because the d5 conﬁguration is spherically symmetric.25 1/2 0 0
0. Bleaney.9920
1. Orienting the ﬁeld along x and y gives gx. sure enough. Because the g.6 Energy levels for Cr(CN)63À and V(CN)64À for B along the zaxis. Given the measured positions of the transitions.and Amatrices are nearly isotropic. Thus Das and Pal9 have doped CoSiF6 Á 6H2O and Co1ÀxZnxSiF6 Á 6H2O with Mn(II). and Bowers8 was a test of the theory – of course it passed with ﬂying colors.5 1 Cr(CN)633/2 V(CN)64 1
0. For CoSiF6 Á 6H2O. highspin Mn(II) and Fe(III) usually have nearly isotropic gmatrices and Mn(II) usually has a nearly isotropic Amatrix.ESR Spectra of Biradicals.011 –0.993 1.5
0. replacing some of the Co(II).5 3/2 1 1. they were.0264 E (cmÀ1) 0. and E.25
0
2000 4000 6000 8000 0 1000 2000 3000 4000 Magnetic Field/Gauss
0.
.8
Substituting the values from Table 6.5 E/cm1 0 1/2 0. we get the energy levels shown in Figure 6. Ligandﬁeld theory was still in its infancy in 1956.5 1 0. Triplet States. gy.991 1. which is high in the spectrochemical series.2 into these equations.991 1.6.9920
gz 1. it was expected that both complexes would be low spin and. This means that there usually is not much information in the ESR spectrum of these highspin species.5
Figure 6.2
Complex Cr(CN)63À V(CN)64À
ESR parameters for two highspin complexes
D (cmÀ1) 0. and other S41/2 Systems
131
Table 6. highspin Mn(II) is usually an unwanted interference for those interested in lowspin Mn(II).
O.3. Baker. B. Phys. A. Phys. 4. 27.H. they found a change in the dipolar parameter D from À176 to À274 Â 10À4 cmÀ1 while the hyperﬁne coupling changed from 93 to 96 Â 10À4 cmÀ1. 41.H. 7.6(2) 80. Chem. Proc. 42. M.A. 1965.S. Brandon. Estimated errors in last signiﬁcant ﬁgure shown in parentheses. In this experiment. 1969. Appl. Bowers. Glarum and J. B69.9996(2) 2. J..182. 39. 5. 9. Phys. for XMn ¼ 0. Chem. J. the goal was to study Mn(II) in an ionic medium. 3717. The results are summarized in Table 6. 10. 1965.9(1)
1. Maki. and E in units of 10À4 cmÀ1. 34. Hirota. Phys. T. Phys. the transitions occurred at 205. Chem. J. 1964. 185 K. 69. R. Phys.8(1)
10
Ea 127. J. Marshall. Chem. Phys. Phys. 903. 1429. J.. These changes correlate with the phase transition.E. 908..H. 4.. Chem. C.K.H.0004(2) 2. Chem. Mangum. 1967. Hirota10 reported a very similar experiment in which PbCl2 was doped with Mn(II). Vincent and A. The temperature of the phase transition depended on the Mn(II) content. J. 1956. 952. 1205. J. Griﬃth. 218 K and 175. respectively. J.D.M.W.082 and 0. Solids. 8. 1987. 3088. Chem. 1958.9(2) Da 463.W. Phys. 1988. Phys.
following the ESR spectrum as a function of temperature. Jpn.. 493. J. Gerkin and C. 48. S.S. 1964.
. 104. J.. Lett. 1961. Soc. Pal. 47. D. 29. Bleaney and K. Vincent and A. Hutchison. 6. Grivet. J. Das and A.132
Chapter 6
Table 6. R. J. 865.3
g x y z
a
ESR parameters for Mn(II) in PbCl2
Aa 79. Maki. 2.. 1374.H.7(5) 79. Hutchison and C...
References
1. 3. Chem.003(1)
A.
z0 x00 .7 and 4.and Amatrices on the ESR of transition metal complexes. y0. 8.1 Some of the details.
7. z X. and g.8) several examples were presented to illustrate the eﬀects of noncoincident g.: ~ ~ S Á g Á B ¼ gBSz0 Table 7. Most of the calculations in this chapter were ﬁrst executed by Janice DeGray. in the notation used here.CHAPTER 7
Perturbation Theory Calculations
7. to secondorder in perturbation theory. b.1 Secondorder Perturbation Theory Treatment of Spin Hamiltonian with Noncoincident g. relating the orientations of the two coordinate systems. Z x0.e.1
The Electron Zeeman Term
~ ~ H ¼ mB S Á g Á B
~ ~ S is quantized along g Á B i. a. Analysis of such spectra requires the introduction of a set of Eulerian angles.1 Various axis systems
Coordinates x. y. z00
Coordinate system gMatrix principal axes AMatrix principal axes Electron spin quantization axes Nuclear spin quantization axes 133
.1). Here is presented a detailed description of how the spin Hamiltonian is modiﬁed. y00 . to incorporate these new parameters in a systematic way. The key to success turns out to be taking careful account of the various axis systems (Table 7. have also been published in ref. Y.and Aaxes
In Chapter 4 (Sections 4.1.
it is necessary that: 0 1 0 ~ gÁB¼@ 0 A gB The x.z 0 (the axis system in which S is quantized) must then take this form to: 0 1 0 1 gx B sin y cos f 0 ~ g Á B ¼ @ gy B sin y sin f A ) @ 0 A gz B cos y gB 1 0 1 gx B sin y cos f 0 ~ Q Á g Á B ¼ Q Á @ gy B sin y sin f A ¼ @ 0 A gz B cos y gB where Q is a transformation matrix corresponding to rotations of the axes by the Euler angles z (rotation about yaxis). Sz 0 ). Sy 0 . w (rotation about new yaxis) (Whittaker deﬁnitions2. We therefore have: 0 1 gx B sin y cos f ~ g Á B ¼ @ gy B sin y sin f A gz B cos y ~ The transformation from the x.y 0 .3): 0 1 Q11 Q12 Q13 Q ¼ @ Q21 Q22 Q23 A Q31 Q32 Q33 where: Q11 ¼ cos z cos x cos w À sin z sin w Q12 ¼ sin z cos x cos w þ cos z sin w 0
.134
Chapter 7
Since S ¼ (Sx 0 .y.z axes to x 0 .y. x (rotation about new zaxis).z coordinate system is deﬁned by the gmatrix principal axes: 0 1 gx 0 0 g ¼ @ 0 gy 0 A 0 0 gz The magnetic ﬁeld vector in that coordinate system is: 0 1 B sin y cos f ~ B ¼ @ B sin y sin f A B cos y where y and j are the usual polar angles.
Perturbation Theory Calculations
135
Q13 ¼ À sin x cos w Q21 ¼ À cos z cos x sin w À sin z cos w Q22 ¼ À sin z cos x sin w þ cos z cos w Q23 ¼ sin x sin w Q32 ¼ sin z sin x Thus: Q11 gx sin y cos j þ Q12 gy sin y sin j þ Q13 gz cos y ¼ 0 Q21 gx sin y cos j þ Q22 gy sin y sin j þ Q23 gz cos y ¼ 0 Q31 gx sin y cos j þ Q32 gy sin y sin j þ Q33 gz cos y ¼ g These equations are satisﬁed with: g2 ¼ g2 sin2 y þ g2 cos2 y ? z g2 ¼ g2 cos2 f þ g2 sin2 f ? x y if: cos z ¼ gx cos f g? gz cos y g gy sin f g? g? sin y g Q31 ¼ cos z sin x Q33 ¼ cos x
sin z ¼ sin x ¼
cos x ¼
Note that the angle w is left indeterminant by this transformation. The above result was used in deriving eqn (4. and g in the same way that Q depends on z. we neglect B compared with Bhf .4 we assume that ~ is quantized along the eﬀective or resultant ﬁeld due to the I applied ﬁeld B and the hyperﬁne ﬁeld due to the unpaired electron: ~ ~ Bhf ¼ S Á A ~ ~ To a ﬁrst approximation. For B E 3000 G. i.Y. 20 G. We label the nuclear spin quantization axes (x00 . Following Blinder.1. which are related to the principal axes of g by the transformation matrix R .2
Nuclear Hyperﬁne Interaction
~ ~ Hhf ¼ S Á A Á I
where the principal axes of A are (X.
.e. g~
7. Xband ESR.y00 . x.4) in Chapter 4. and w.Z). b.z00 ) and require that the Sz 0 term has the form KSz 0 Iz . this will be a good approximation for most metal nuclei and for protons or ﬂuorines with hyperﬁne couplings greater than ca. which depends in the Euler angles a. This amounts ~ to saying that Sx and Sy are not ﬁxed in space by the quantization of S along B.
z) to (x 0 . S32 and S33: S31 ¼ Q31 R11 þ Q32 R21 þ Q33 R31 gx ¼ sin y cos fðcos a cos b cos g À sin a sin gÞ g gy þ sin y sin fðÀ cos a cos b sin g À sin a cos gÞ g gz þ cos y cos a sin b g S32 ¼ Q31 R12 þ Q32 R22 þ Q33 R32 gx ¼ sin y cos fðsin a cos b cos g þ cos a sin gÞ g gy þ sin y sin fðÀ sin a cos b sin g þ cos a cos gÞ g gz þ cos y sin a sin b g S33 ¼ Q31 R13 þ Q32 R23 þ Q33 R33 gx ¼ À sin y cos f sin b cos g g gy gz þ sin y sin f sin b sin g þ cos y cos b g g
ð7:1aÞ
ð7:1bÞ
ð7:1cÞ
.z) and matrix Q transforms from (x.136
Chapter 7
0 0 0
The hyperﬁne term may be written in the (x .Y. KSz0 Iz00 where K is the eﬀective hyperﬁne coupling for the particular orientation.z 0 ) coordinate system: À Á ! ! S Á Z Á I ¼ Sx0 Z11 Ix0 þ Z12 Iy0 þ Z13 Iz0 À Á þ Sy0 Z21 Ix0 þ Z22 Iy0 þ Z23 Iz0 À Á þ Sz0 Z31 Ix0 þ Z32 Iy0 þ Z33 Iz0 In the nuclear spin quantization axis system. Thus: Iz00 ¼ where:
2 2 2 K 2 ¼ Z31 þ Z32 þ Z33
Á 1À Z31 Ix0 þ Z32 Iy0 þ Z33 Iz0 K
As it happens.Z) to (x.Z) to (x 0 .y 0 . matrix R transforms from (X.y . the only components of the Smatrix we will need to know (for the perturbation theory treatment) are S31.y 0 .y.z 0 ).y.z ) system: ~ I Hhf ¼ S Á Z Á ~ where: Z¼S
À1
ÁAÁS
S ¼QÁR
That is. so that matrix S transforms from (X.Y. In the (x 0 . the last term has the form.y 0 .z 0 ).
we can write: Z11 ðcos a cos b cos g À sin a sin gÞ þ Z12 ðÀ cos a cos b sin g À sin a cos gÞ þ Z13 cos a sin b þZ21 ðsin a cos b cos g þ cos a sin gÞ þ Z22 ðÀ sin a cos b sin g þ cos a cos gÞ þ Z23 sin a sin b % Z11 þ Z22 þ Z33 À K
. b and g. However. in this case we have a means of approximating g. However. We are multiplying Z from one side only. the trace of the matrix would remain constant. so that the trace is not necessarily invariant. Since the hyperﬁne term can be written: 0 1 0 1 0 1 Z11 Z12 Z13 I x0 Sx0 B 0C B C B C Sy A Á @ Z21 Z22 Z23 A Á @ Iy0 A ¼ @ Sz0 Z31 Z32 Z33 Iz0 0 1 0 1 0 1 Z11 Z12 Z13 Sx0 Ix00 B 0C B C B C @ Sy A Á @ Z21 Z22 Z23 A Á QI Á @ Iy00 A Sz0 Z31 Z32 Z33 Iz00 we also have: Ix0 ¼ ðcos a cos b cos g À sin a sin gÞIx00 þ ðsin a cos b cos g þ cos a sin gÞIy00 À sin b cos gIz00 Iy0 ¼ ðÀ cos a cos b sin g À sin a cos gÞIx00 þ ðÀ sin a cos b sin g þ cos a cos gÞIy00 þ sin b sin gIz00 Iz0 ¼ cos a sin bIx00 þ sin a sin bIy00 þ cos bIz00 where: cos a sin b ¼ Z31 =K sin a sin b ¼ Z32 =K cos b ¼ Z33 =K Thus the angles a and b are determined.Perturbation Theory Calculations
137
Consider the matrix which transforms the ~vector from the nuclearspin into I the electronspin quantization axes: 0 1 0 1 I x0 Ix00 QI Á @ Iy00 A ¼ @ Iy0 A Iz00 I z0 where QI can be written in terms of Euler angles a. but g (like w) remains indeterminant. If QI were used to transform Z into the nuclear spin quantization axis system.
The nature of the approximation causes this approach to fail. calculation of the new parameters Dx. Now assume that the hyperﬁne term can be written in the form: 1 0 10 Ix00 À Á Dx Ex Fx ~ S Á Z Á ~ ¼ Sx0 Sy0 Sz0 @ Dy Ey Fy A@ Iy00 A I 0 0 K Iz00 À Á ~ S Á Z Á ~ ¼ Sx0 Dx Ix00 þ Ex Iy00 þ Fx Iz00 I À Á þ Sy0 Dy Ix00 þ Ey Iy00 þ Fy Iz00 þ KSz0 Iz00 Given values of a.
7.3
Perturbation Theory Treatment of Hyperﬁne Term
If we expand the hyperﬁne term of the spin Hamiltonian and write the operators in terms of raising and lowering operators: Sþ ¼ Sx0 þiSy0 À À Iþ ¼ Ix00 þiIy00 À À
. we have: À A sin g þ B cos g ¼ 0 or: tan g ¼ B A
In practice. Fx and Fy is straightforward using perturbation theory. Ey. we must check this result to make sure that we have a minimum and not a maximum. the angles a and b. Dy. and solve for g. Ex. b and g.1.138
Chapter 7
Collecting terms in cos g and sin g: A cos g þ B sin g þ Z13 cos a sin b þ Z23 sin a sin b % Z11 þ Z22 þ Z33 À K where: A ¼ Z11 cos a cos b À Z12 sin a þ Z21 sin a cos b þ Z22 cos a B ¼ À Z11 sin a À Z12 cos a cos b þ Z21 cos a À Z22 sin a cos b If this were not an approximation. but we can still look for the value of g that brings the function closest to a solution. we could evaluate the components of the Zmatrix. Diﬀerentiating with respect to g.
I. I. mI 4 2 hmI À Ái Fx þ iFy 1. mI 2 2 2 ÂÀ Á À ÁÃ þ 1 Dx À Ey À i Dy þ Ex Sþ Iþ 1. mI 2 we have H0 1. I. I. mI ¼ 2 2 2 2 ÂÀ Á2 À ÁÃ þ 1 Dx À Ey þ i Dy þ Ex 4 1 Â ½I ðI þ 1Þ À mI ðmI À 1Þ2 1. I. ms . ms . I. ms . ms . mI 2 2 ¼ ðms gmB B þ ms mI K Þ1.Perturbation Theory Calculations
1 2
Hhf ¼ KSz0 Iz00 þ Fx À iFy Sþ Iz00 þ Fx þ iFy SÀ Iz00 ÂÀ Á À ÁÃ þ 1 Dx À Ey À i Dy þ Ex Sþ Iþ 4 ÂÀ Á À ÁÃ þ 1 Dx À Ey þ i Dy þ Ex SÀ IÀ 4 ÂÀ Á À ÁÃ þ 1 Dx þ Ey À i Dy À Ex Sþ IÀ 4 ÂÀ Á À ÁÃ þ 1 Dx þ Ey þ i Dy À Ex SÀ Iþ 4 The unperturbed and perturbation Hamiltonians then are: H0 ¼ gmB BSz0 þ KSz0 Iz00 À Á À Á H1 ¼ 1 Fx À iFy Sþ Iz00 þ 1 Fx þ iFy SÀ Iz00 2 2 ÂÀ Á À ÁÃ ÂÀ Á À ÁÃ þ 1 Dx À Ey À i Dy þ Ex Sþ Iþ þ 1 Dx À Ey þ i Dy þ Ex SÀ IÀ 4ÂÀ 4ÂÀ Á À ÁÃ Á À ÁÃ þ 1 Dx þ Ey À i Dy À Ex Sþ IÀ þ 1 Dx þ Ey þ i Dy À Ex SÀ Iþ 4 4 Operating on the zeroorder wavefunctions: 1 . ms . I. I. mI
2
À
Á
1 2
À
Á
139
H1 1. mI 4 2 ÂÀ Á À ÁÃ þ 1 Dx À Ey þ i Dy þ Ex SÀ IÀ 1. À1. mI 2 ÂÀ Á À Á Ã ¼ 1 Fx À iFy Sþ Iz00 þ 1 Fx þ iFy SÀ Iz00 1. I. mI 4 2 ÂÀ Á À ÁÃ þ 1 Dx þ Ey À i Dy À Ex Sþ IÀ 1. ms . I. 1. I. ms . I. ms . mI þ 1 2 2
. mI À 1 2 2 ÂÀ Á À ÁÃ þ 1 Dx þ Ey þ i Dy À Ex 4 1 Â ½I ðI þ 1Þ À mI ðmI þ 1Þ2 1. À1. ms . mI H1 1. I. mI 4 2 ÂÀ Á À ÁÃ þ 1 Dx þ Ey þ i Dy À Ex SÀ Iþ 1. À1. ms . mI ¼ ðgmB BSz0 þ KSz0 Iz00 Þ1. I.
mI H1 2. I. 1. 2. and secondorder energies are: Eþ1=2 ¼ þ 1gmB B þ1KmI À2 À2 À Eþ1=2 ¼ 0
À ð1Þ ð0Þ
E1=2 ¼
ð2Þ
m2 2 2 I Fx þ Fy 4hv Á2 À Á2 i 1 hÀ Dx À Ey þ Dy þ Ex ½I ðI þ 1Þ À mI ðmI À 1Þ þ 16hv Á2 À Á2 i 1 hÀ Dx þ Ey þ Dy À Ex ½I ðI þ 1Þ À mI ðmI þ 1Þ þ 16hv
. mI ¼ 2 Fx þ iFy 1
1 1 1 2. 1. mI H1 2. À 2. 2. mI H1 1. mI ¼ 2 Fx À iFy 1
1 1 1 2. À 2. À 2. mI ÂÀ Á À ÁÃpﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ I ðI þ 1Þ À mI ðmI þ 1Þ ¼ 1 Dx À Ey À i Dy þ Ex 4 1 1 1 2. I. I. À 2. I. mI À 1 H1 2. ﬁrst. 1. I. mI þ 1 H1 2. 2.140
hmI À Ái Fx À iFy 1. I. mI À 1 H1 2. I. À1. mI ¼ 2 2 2 2 ÂÀ Á 2À ÁÃ þ 1 Dx À Ey À i Dy þ Ex 4 1 Â ½I ðI þ 1Þ À mI ðmI þ 1Þ2 1. I. 2. I. À 2. 2. I. mI ÂÀ Á À ÁÃpﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ¼ 1 Dx þ Ey þ i Dy À Ex I ðI þ 1Þ À mI ðmI þ 1Þ 4
1
1 1 mI À 1 1 Á 2. mI ÂÀ Á À ÁÃpﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ¼ 1 Dx À Ey þ i Dy þ Ex I ðI þ 1Þ À mI ðmI À 1Þ 4 1 1 1 2. I. 2. À2. I. I. mI ÂÀ Á À ÁÃpﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ¼ 1 Dx þ Ey À i Dy À Ex I ðI þ 1Þ À mI ðmI À 1Þ 4
1
Thus the zero. I. mI À 1 2 2
Chapter 7
Thus the signiﬁcant matrix elements are: 1 1 mI À 1 1 Á 2. mI þ 1 H1 2. mI þ 1 2 2 ÂÀ Á À ÁÃ þ 1 Dx þ Ey À i Dy À Ex 4 1 Â ½I ðI þ 1Þ À mI ðmI À 1Þ2 1. I. I.
the energy is: m2 2 2 DE ¼ hv ¼ gmB B þ KmI þ I Fx þ Fy 2hv iÂ Ã mI h 2 2 2 þ Dx þ Ey þ D2 þ Ex I ðI þ 1Þ À m2 y I 4hv Deﬁning:
2 2 F1 ¼ Fx þ Fy
ð7:2Þ
2 2 F2 ¼ D2 þ D2 þ Ex þ Ey x y
the transition energy can be written: DE ¼ hv ¼ gmB B þ KmI þ and the ﬁeld: B¼ hv KmI ð2F1 À F2 Þm2 þ F2 I ðI þ 1Þ I À À gmB gmB 4gmB hv ð7:4Þ ð2F1 À F2 Þm2 F2 I ðI þ 1Þ I þ 4hv 4hv ð7:3Þ
Thus we must express F1 and F2 in parameters of the original spin Hamiltonian.Perturbation Theory Calculations
141
EÀ1=2 ¼ À
ð2Þ
2 F 2 þ Fy 4hv x Á2 À Á2 i 1 hÀ Dx À Ey þ Dy þ Ex ½I ðI þ 1Þ À mI ðmI þ 1Þ À 16hv Á2 À Á2 i 1 hÀ Dx þ Ey þ Dy À Ex ½I ðI þ 1Þ À mI ðmI À 1Þ À 16hv
m2 I
For the ‘‘allowed’’ transitions. we equate the coeﬃcients of Sx 0 . DmS ¼ þ 1. Starting with the two ways of expressing the hyperﬁne Hamiltonian term. DmI ¼ 0. Sy 0 and Sz 0 : Z11 Ix0 þ Z12 Iy0 þ Z13 Iz0 ¼ Dx Ix00 þ Ex Iy00 þ Fx Iz00 Z21 Ix0 þ Z22 Iy0 þ Z23 Iz0 ¼ Dy Ix00 þ Ey Iy00 þ Fy Iz00 Z31 Ix0 þ Z32 Iy0 þ Z33 Iz0 ¼ K or: À Á Z11 Ix0 þ Z12 Iy0 þ Z13 Iz0 ¼Dx a1 Ix0 þ b1 Iy0 þ c1 Iz0 À Á þ Ex a2 Ix0 þ b2 Iy0 þ c2 Iz0 À Á þ Fx a3 Ix0 þ b3 Iy0 þ c3 Iz0 À Á Z21 Ix0 þ Z22 Iy0 þ Z23 Iz0 ¼Dy a1 Ix0 þ b1 Iy0 þ c1 Iz0 À Á þ E y a 2 I x0 þ b 2 I y 0 þ c 2 I z 0 À Á þ F y a 3 I x0 þ b 3 I y 0 þ c 3 I z 0
.
we have: Z11 ¼ Dx a1 þ Ex a2 þ Fx a3 Z12 ¼ Dx b1 þ Ex b2 þ Fx b3 Z13 ¼ Dx c1 þ Ex c2 þ Fx c3 Squaring and summing: À Á 2 2 2 Z11 þ Z12 þ Z13 ¼ D2 a2 þ b2 þ c2 x 1 1 1 À Á À Á 2 2 þ E x a2 þ b2 þ c 2 þ F x a2 þ b2 þ c 2 2 2 2 3 3 3 þ 2Dx Ex ða1 a2 þ b1 b2 þ c1 c2 Þ þ 2Dx Fx ða1 a3 þ b1 b3 þ c1 c3 Þ þ 2Ex Fx ða2 a3 þ b2 b3 þ c2 c3 Þ or:
2 2 2 2 2 Z11 þ Z12 þ Z13 ¼ D2 þ Ex þ Fx x
Similarly:
2 2 2 2 2 Z21 þ Z22 þ Z23 ¼ D2 þ Ey þ Fy y
Since: a1 a3 þ b1 b3 þ c1 c3 ¼ 0. we have: Fx ¼ a3 Z11 þ b3 Z12 þ c3 Z13 or: Fx ¼ 1 ðZ31 Z11 þ Z32 Z12 þ Z33 Z13 Þ K c1 ¼ À a1 a3 þ b1 b3 c3
.142
Chapter 7
Equating coeﬃcients of Ix 0 . Thus: Z11 ¼ Dx a1 þ Ex a2 þ Fx a3 Z12 ¼ Dx b1 þ Ex b2 þ Fx b3 a1 a3 þ b1 b3 a2 a3 þ b2 b3 Z13 ¼ À Dx À Ex þ F x c3 c3 c3 or: a3 Z11 ¼ Dx a1 a3 þ Ex a2 a3 þ Fx a2 3 b3 Z12 ¼ Dx b1 b3 þ Ex b2 b3 þ Fx b2 3 c3 Z13 ¼ À Dx ða1 a3 þ b1 b3 Þ À Ex ða2 a3 þ b2 b3 Þ þ Fx c2 3 Summing. Iy 0 and Iz 0 .
and simplifying making use of the orthonormality of rows of the Smatrix. we have: Z33 Z13 þ Z32 Z12 þ Z31 Z11 ¼ A2 S31 S11 þ A2 S32 S12 þ A2 S33 S13 x y z Z33 Z23 þ Z32 Z22 þ Z31 Z21 ¼ A2 S31 S21 þ A2 S32 S22 þ A2 S33 S23 x y z Thus:
2 2 F1 ¼ Fx þ Fy ¼
1 4 2 2 2 Ax S31 þ A4 S32 þ A4 S33 À K 4 y z K2
ð7:5Þ
2 2 2 2 2 2 K 2 ¼ Z11 þ Z12 þ Z13 ¼ A2 S13 þ A2 S12 þ A2 S11 z y x 2 2 2 2 2 D2 þ Ex ¼ Z11 þ Z12 þ Z13 À Fx x
À 2 Á 2 2 2 K 2 Z11 þ Z12 þ Z13 À K 2 Fx ¼ ðS33 S2 À S32 S3 Þ2 A2 A2 y z þ ðS33 S1 À S31 S3 Þ2 A2 A2 x z þ ðS32 S1 À S31 S2 Þ2 A2 A2 x y
2 2 2 2 2 D2 þ Ey ¼ Z21 þ Z22 þ Z23 À Fy y 2 2 2 2 2 2 Z21 þ Z22 þ Z23 ¼ A2 S23 þ A2 S22 þ A2 S21 z y x
À 2 Á 2 2 2 K 2 Z21 þ Z22 þ Z23 À K 2 Fy ¼ ðS33 S22 À S32 S23 Þ2 A2 A2 y z þ ðS33 S21 À S31 S23 Þ2 A2 A2 x z þ ðS32 S21 À S31 S22 Þ2 A2 A2 x y F2 ¼ Á 1 h 2 2À 2 2 Ay Az S33 þ S32 2 K À 2 Á À 2 Ái 2 2 þ A2 A2 S33 þ S31 þ A2 A2 S32 þ S31 x z x y
ð7:6Þ
.Perturbation Theory Calculations
143
Similarly: Fy ¼ 1 ðZ31 Z21 þ Z32 Z22 þ Z33 Z23 Þ K
Expanding the Z matrix in terms of Ax. Ay and Az and the components of the Smatrix.
things can be simpliﬁed by taking the average of the 5/2.8 3448.
2800 2900
3000
3100
3200
3300
3400
3500
Magnetic Field/Gauss
Figure 7.6 3381.2 3/2.5 3320.3).À1/2 3236.3 À3/2.7* 3505.3 3215. Figure 7.3 3388.mP) for [Mn(CO)2(PPh3)(C5H5)]1a
3/2.45 GHz) of [Mn(CO)2(PPh3)(C5H5)]1.6 À5/2.1/2 and 5/2.1/2 3295. have been published.1/2 3207.À1/2 3145.6 3419.1 3314.8 3352.5 3246. Sp ¼ saddle point.2
Measured line positions (mMn.4 3350.3* 3473.4 3310.1 Frozen solution ESR spectrum (n ¼ 9. and an analysis of the spectrum simulated in Figure 7.8 À5/2. Some of Pike’s spectra.8 3372.1.0 1/2.4
Example Application of these Results
Robert Pike has prepared various lowspin Mn(II) complexes.1/2 3116.0 3372.À1/2 Min Sp Max
a
2865. 5.1* 5/2.6 3277.À1/2 3053. and Max ¼ maximum.1 3279.5 3340. we notice that the 31P coupling is virtually isotropic.8*
Min ¼ minimum.9 3245.4* À1/2. etc.4 3417.5 3404.
Table 7. Simulated using parameters in ref.À1/2 3324.À1/2 2959.1/2 Min Sp Max 2836.1. and a nonlinear leastsquares program is required.
. those indicated with an asterisk (*) are poorly resolved. Thus. As a ﬁrst step.4 shows the ﬁtted parameters.1/2 3024.6 Table 7.1 3309.1 shows the ESR spectrum of [Mn(CO)2(PPh3)(C5H5)]1 in frozen CH2Cl2/C2H4Cl2 (see also Tables 7. It is not quite so simple.1/2 2930.1
5/2.1.144
Chapter 7
7.5 We will now see how to apply the equations derived above.and Amatrix principal axes are noncoincident.4 À1/2. This suggests that the g.: The next thing to notice is that the widely spaced features are very approximately equally spaced.5 3340. You might think that a simple application of the above equations would suﬃce for a complete analysis.2 and 7.4 3246.1 À3/2.À1/2 assignments.6 1/2.
34 3432.16 3379.z00 ) coordinate system: 0 10 1 P11 P12 P13 Ix00 À ÁB ~Á P Á ~ ¼ Ix00 Iy00 Iz00 @ P12 P22 P23 CB Iy00 C I I A@ A P13 P23 P33 Iz00 1 1 2 ¼P33 Iz00 À P33 ðIþ IÀ þ IÀ Iþ Þ þ ðP13 þ iP23 ÞðIÀ Iz þ Iz IÀ Þ 4 2 À 2 Á 1 i 2 þ ðP13 À iP23 ÞðIþ Iz þ Iz Iþ Þ À P12 Iþ À IÀ 2 2 À 2 Á 1 2 þ ðP11 À P22 Þ Iþ þ IÀ 4
.51 3388.8 g¼0 2.05 3309.04 3038.4 29.Perturbation Theory Calculations
145
1
Table 7.77 3262.91 z 2. 10 29.3
mMn 5/2 3/2 1/2 À1/2 À3/2 À5/2
Line positions for [Mn(CO)2(PPh3)(C5H5)] .99 3489.09 3404.94 3336.94 2945. the Hamiltonian term would have the form in the (x00 .9 29.81 3366.34
Table 7.188 ca.68 3131.8 a¼0
g AMn AP Euler angles (1)
7. The quadrupole coupling term in the spin Hamiltonian can be written as: HQ ¼ ~Á P Á ~ I I where P is the quadrupole coupling matrix: 0 1 ÀPð1 À ZÞ 0 0 P¼@ 0 ÀPð1 þ ZÞ 0 A 0 0 2P Assuming identical principal axes for A and P.00 Max 3324.8 b ¼ 73.002 98.y00 .15 3295.034 32.12 3356. Presented here is the way such eﬀects may be incorporated into the spectra using perturbation theory.2 Quadrupole Coupling
The spectra discussed in Chapter 4 were analyzed by neglecting the eﬀects of nuclear quadrupole coupling on the nuclear hyperﬁne structure. Averaged over mP5
Min 2850.4
ESR parameters for [Mn(CO)2(PPh3)(C5H5)]15
x y 2.28 3222.96 SP 3230.
m þ 1i 2 rh ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ i2 þ 1ðP11 À P22 À 2iP12 Þ I ðI þ 1Þ À ðm þ 1Þ2 Àðm þ 1Þ2 jI.. mi ¼ 1ðP13 þ iP23 Þð2m À 1Þ 2 hI.1
Perturbation Theory Treatment of Quadrupole Term
Operating on the spin functions. m À 2jHQ jI. mi ¼ 1ðP13 À iP23 Þð2m þ 1Þ 2 hI. we have: Â Ã HQ jI. mi ¼ 1P33 3m2 À I ðI þ 1Þ jI. mjHQ jI. mi 2 pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Â 1ðP13 þ iP23 Þð2m À 1Þ I ðI þ 1Þ À mðm À 1ÞjI.2. m À 1jHQ jI. mi ¼
1 4ðP11
pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ I ðI þ 1Þ À mðm À 1Þ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ I ðI þ 1Þ À mðm þ 1Þ
rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ h i2 À P22 À 2iP12 Þ I ðI þ 1Þ À ðm þ 1Þ2 Àðm þ 1Þ2
hI. m À 2i which leads to matrix elements: Â Ã hI. i. mi ¼
1 4ðP11
À P22 þ 2iP12 Þ
rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ h i I ð I þ 1Þ À ð m À 1Þ 2 À ð m À 1Þ 2
2
First. m þ 1jHQ jI. Pij ¼ Pji and with zero trace.146
Chapter 7
Note that transformation by a general coordinate transformation matrix leaves the quadrupole matrix symmetrical. the lower to À 1/2. m þ 2jHQ jI. mi ¼ 1P33 3m2 À I ðI þ 1Þ 2 hI.and secondorder energy corrections then are: Â Ã E ð1Þ ¼ 1P33 3m2 À I ðI þ 1Þ 2 ÁÂ À ÁÃ 4m À 2 P13 þ P2 I ðI þ 1Þ À 2m2 þ 1 E ð2Þ ¼ þ 23 4 À K iÂ À ÁÃ m h ðP11 À P22 Þ2 þ4P2 I ðI þ 1Þ À m2 þ 1 À 12 2 þ 2K where K is the angledependent hyperﬁne coupling deﬁned in the previous section and the upper sign corresponds to ms ¼ þ1/2. m þ 2i 4 rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ h i2 1 þ 4ðP11 À P22 þ 2iP12 Þ I ðI þ 1Þ À ðm À 1Þ2 Àðm À 1Þ2 jI. P11 þ P22 ¼ ÀP33.
.
7.e. m À 1i 2 pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ þ 1ðP13 À iP23 Þð2m þ 1Þ I ðI þ 1Þ À mðm þ 1ÞjI.
y 0 . simplifying the expressions using the relations: a1 b1 þ a2 b2 þ a3 b3 ¼ 0 b2 þ b2 þ b 2 ¼ 1 1 2 3 a1 b2 À a 2 b1 ¼ c 3 b1 c 2 À b2 c 1 ¼ a3 with similar expressions relating components of the Smatrices: P33 ¼ 3ðc3 S33 þ b3 S23 þ a3 S13 Þ2 P À P þ ðc3 S31 þ b3 S21 þ a3 S11 Þ2 PZ À ðc3 S32 þ b3 S22 þ a3 S12 Þ2 PZ F3 ¼ 9F3a P2 þ F3b P2 Z þ F3c P2 Z2 F3a ¼ ðc3 S33 þ b3 S23 þ a3 S13 Þ2 Àðc3 S33 þ b3 S23 þ a3 S13 Þ4 h F3b ¼ 2ðc3 S33 þ b3 S23 þ a3 S13 Þ2 Â ðc3 S32 þ b3 S22 þ a3 S12 Þ2 À ðc3 S31 þ b3 S21 þ a3 S11 Þ2 a1 c 1 þ a2 c 2 þ a3 c 3 ¼ 0
b1 c 1 þ b2 c 2 þ b3 c 3 ¼ 0 a2 þ a 2 þ a2 ¼ 1 1 2 3 c2 þ c2 þ c2 ¼ 1 1 2 3 a1 c2 À a2 c1 ¼ Àb3
i
.y00 . and F4 ¼ (P11 – P22)2 þ 4P122.z ).cmatrix and compute P33.z 0 ) coordinate system into (x00 .y . the quadrupole interaction matrix is: 0 10 10 1 S11 S12 S13 ÀPð1 À ZÞ 0 0 S11 S21 S31 @ S21 S22 S23 A@ 0 ÀPð1 þ ZÞ 0 A@ S12 S22 S32 A 0 0 2P S31 S32 S33 S13 S23 S33 Since we can transform from the (x 0 .z00 ) by: 0 1 0 10 1 Ix00 a1 b1 c 1 Ix0 @ Iy00 A ¼ @ a2 b2 c2 A@ Iy0 A Iz00 a3 b3 c 3 Iz0 the reverse transformation must be: 0 1 0 I x0 a1 @ Iy0 A ¼ @ b1 I z0 c1
a2 b2 c2
10 1 a3 Ix00 b3 A@ Iy00 A c3 Iz00
We next transform the P matrix.Perturbation Theory Calculations
0 0 0
147
In the electronspin quantization system (x . using the a. F3 ¼ P132 þ P232.b.
we arrive at the following equations: B ¼B0 À Ã mK F1 m2 F2 Â À 2 2 À 2 2 I ðI þ 1Þ À m2 gmB 2g mB B 4g mB B À ÁÃ 8mF3 Â þ I ðI þ 1Þ À 2m2 þ 1 4 gmB K À ÁÃ mF4 Â À I ðI þ 1Þ À m2 þ 1 2 gmB K
.148
Chapter 7
F3c ¼ ðc3 S32 þ b3 S22 þ a3 S12 Þ
2
þ ðc3 S31 þ b3 S21 þ a3 S11 Þ2 h À ðc3 S32 þ b3 S22 þ a3 S12 Þ2 i2 Àðc3 S31 þ b3 S21 þ a3 S11 Þ2 F4 ¼ 9F4a P2 þ F4b PZ þ F4c P2 Z2 F4a ¼ 1 À 2ðc3 S33 þ b3 S23 þ a3 S13 Þ2 þðc3 S33 þ b3 S23 þ a3 S13 Þ4 h i F4b ¼ 2 ða3 S13 þ b3 S23 þ c3 S33 Þ2 þ1 F4c ¼ 4ða3 S13 þ b3 S23 þ c3 S33 Þ2 h þ ða3 S11 þ b3 S21 þ c3 S31 Þ2 i2 Àða3 S12 þ b3 S22 þ c3 S32 Þ2 but: c3 S31 þ b3 S21 þ a3 S11 ¼ c3 S32 þ b3 S22 þ a3 S12 c3 S33 þ b3 S23 þ a3 S13 Finally: Ax S31 K Ay S32 ¼ K Az S33 ¼ K
2 2 2 2 2 þ6ZA2 S33 A2 S32 À A2 S31 þ 4Z2 A2 S31 A2 S32 z y x x y À À Á Á P2 h 2 2 2 2 2 F4 ¼ 4 9A4 1 À S33 þ6Z A2 S31 À A2 S32 K 2 þ A2 S33 ? x y z K ! 2 2 2 2 2 2 2 2 2 2 þZ Ax S31 À Ay S32 þ4Z K Az S33
À Á Fi ¼ P2 9Fia þ 3ZFib þ Z2 Fic i À Á P h 2 2 2 2 P33 ¼ 2 2A2 S33 À A2 1 À S33 þ Z A2 S31 À A2 S32 z ? x y K 2 ÂÀ Á À Á P 2 2 F3 ¼ 4 9 þ Z2 A2 A2 S33 1 À S33 ? z K i
Putting together all the bits and pieces of the above perturbation theory treatment of quadrupole coupling eﬀects.
2. F1 ðyÞ ¼ . but there is no trace on the other two gcomponents.Perturbation Theory Calculations
149
where: K2 ¼
i 1h 2 2 Azz gz cos2 y þ A2 g2 cos2 j þ A2 g2 sin2 j sin2 y xx x yy y 2 g
g2 A4 À K 4 g2 A4 À K 4 g2 A 4 À K 4 y y x x . the inner and outer quartet spacings are: DBouter ðx. We must ﬁrst examine the predicted spacing of the expected quartet. together with a very satisfactory simulation.2
Example Application of Analysis of Quadrupole Eﬀects
The ESR spectrum of a RhIr complex7. we might have expected each gcomponent to be a quartet of doublets. Since 103Rh has I ¼ 1/2 and both isotopes of Ir have I ¼ 3/2. yÞ ¼ À gmB gmB K gmB K
. the lowﬁeld and midﬁeld components are more like 1:2:1 triplets than the expected 1:1:1:1 quartet. F1 ðzÞ ¼ z z 2 g2 K 2 g2 K 2 g2 K À 2 Á g2 A4 A2 þ A2 g2 A 4 A x þ A 2 x x y z y y z F2 ðxÞ ¼ F2 ðyÞ ¼ 2K 2 g g2 K 2 g2 A4 A2 þ A2 z z x y F2 ðzÞ ¼ g2 K 2 À Á F3 ðxÞ ¼ F3 ðyÞ ¼ P2 Z2 1 À cos2 2a . F3 ðzÞ ¼ 0 F4 ðxÞ ¼ F4 ðyÞ ¼ 9P2 . F3 and F4 simplify to: F3 ðxÞ ¼ F3 ðyÞ ¼ P2 Z2 . yÞ ¼ 3K 24P2 Z2 27P2 þ þ gmB gmB K gmB K 2 2 K 24P Z 27P2 þ DBinner ðx. indeed. F4 ðzÞ ¼ 4p2 Z2 Under these circumstances.2. F3 ðzÞ ¼ 0 À Á F4 ðxÞ ¼ P2À 9 þ 6Z cos 2a þ Z2 cos2 2aÁ F4 ðyÞ ¼ P2 9 À 6Z cos 2a þ Z2 cos2 2a F4 ðzÞ ¼ 4P2 Z2 F1 ðxÞ ¼
7.8 is shown in Figure 7. The outer and inner spacings are given by: 3K 24F3 3F4 DBouter ¼ þ þ gmB gmB K gmB K K 24F3 3F4 þ À DBinner ¼ gmB gmB K gmB K Assuming that the quadrupole matrix axes diﬀer from the gmatrix axes by a ¼ 451. The Rh coupling is barely visible on the highﬁeld features.
8. see ref. copyright (2000) Royal Society of Chemistry. 7. the eﬀect is too big for perturbation theory to be entirely reliable. L ¼ CO. Why did it turn out that a ¼ 451? Consider the coordination geometry of the Ir atom (1). Unfortunately.2 ESR spectrum of [RhLL 0 {m(tol)NNN(tol)}2IrLL 0 ]1 PF6À (tol ¼ pMeC6H4. In contrast. the inner spacing of the x and y components is expected to be less than 1/3 the outer spacing. À1/2) spacing than 1/3 the outer spacing.
N Ir Ph3P 1 CO
x N y
. This is entirely consistent with the experimental spectrum of Figure 7. For further details. and it was necessary to perform a matrix diagonalization to get the exact solution of the spin Hamiltonian. (Reproduced with permission from ref. One more item needs to be discussed in reference to this spectrum. 8. L 0 ¼ PPh3).150
Chapter 7
Exptl.)
DBouter ðzÞ ¼
3K 12P2 Z2 þ gmB gmB K K 12P2 Z2 DBinner ðzÞ ¼ þ gmB gmB K
Examination of these equations shows that the spacing of the zcomponent is expected to be a little bigger for the inner (1/2. Simulation using methods of ref.2. spectrum
Simulation
2800 2900 3000
3200
3300
3400
3500
Magnetic Field/Gauss
Figure 7.
6.Perturbation Theory Calculations
151
The Rh–Ir vector deﬁnes the zaxis. S. (a) P. Faraday Trans. Phys. 1993. Rieger.J. Specialist Periodical Reports. Wilson. 85.C.and yaxes. 8.D. Emslie. vol.. 748. 2. Orpen and P. P. Phys. 1989. 1987.G.H. Klangsinsirikul. which has large components if there is an electric ﬁeld gradient at the Ir nucleus. Rieger.A. PhD Thesis. N. ch. I. 2002. Klangsinsirikul and P. Brown University. Watanabe. Chem.
References
1. DeGray. Rieger. Royal Society of Chemistry. 4.
. J.. E. but the CO ligand involves a lot of pbackbonding. P. McGrawHill. Molecular Vibrations.C. Connelly. Chem. 3. Chem.D. J. 286. Rieger. D. 963. Connelly. p.L. Decius and P. Operator Methods in Ligand Field Theory.B.A. 1966. O. 1955. Rieger. Hayward.H. Thus the quadrupole matrix. The N and P atoms are solely electron donors. Blinder. Englewood Cliﬀs. p. 5. Bull. and the NN bridges deﬁne the x.H. A. Cross.. J. 178. 1989. 33. (b) J. 106. 3913.H. PrenticeHall. 147. N. A. Electron Spin Resonance. 1960. J. New York. DeGray and P. Reson.M.H. 4. Cambridge. Commun. R. 13b. Soc. Pike. Mag.H. 7. Rieger and P. is rotated 451 compared with the gmatrix principal axes. Chem.G. A. H.. J. 12214. p.G. 95. 2000. 8.
9864 Â 10À27 J 2.002319304 8.99792 Â 108 m sÀ1 6. and Properties of Nuclei (Tables A1.93481 (g/ge) a/G
152
.05078 Â 10À31 J GÀ1 2.02214 Â 1023 molÀ1 1.4)
Table A 1.62607 Â 10À34 J s1 2.APPENDIX 1
Physical Constants. Conversion Factors.38065 Â 10À23 J KÀ1 9.1–A1.27401 Â 10À28 J GÀ1 5.60218 Â 10À19 C 5.66 Â 10À24 g 10 000 Gauss 1. 1
Name
Physical constants
Symbol h c NA k mB mN ge R e me Value 6.8025 (g/ge) a/G 0.31447 J molÀ1 KÀ1 À1.486 Â 10À4 amu
Planck constant Speed of light Avogadro’s number Boltzmann constant Bohr magneton Nuclear magneton Free electron gvalue Gas constant Electronic charge Electron mass
Table A 1. 2
1 amu 1 Tesla 10À4 cmÀ1 A/MHz A/10À4 cmÀ1
Conversion factors
1.
76 17.9 80.00286 À0.6 46.04 À0.8554 3.63 À0.37 0.5 92.75002 2.56 À34.11 À0.89 7.1316 0.9335 À0.17 30.012 7.0 0.7 941.75 9.2233 2.0 100.8220 3.140 69.2 399.0
À20.75649 À0.0906 4.5 32.1540 0 0 0.5 À50.10417 3.7 146.73 100.083 À0.6413 À0.276 0.23 305.15 0.3817 0.0 À34.0529 0.2 410.7512 1.92 9.33 104As (cmÀ1)b 473.40376 À0.0 10.0406 0 0.3 À51.2 À24.43947 0.55 12. 3
Nucleus H H 6 Li 7 Li 9 Be 10 B 11 B 13 C 14 N 15 N 17 O 19 F 23 Na 25 Mg 27 Al 29 Si 31 P 33 S 35 Cl 37 Cl 39 K 41 K 43 Ca 45 Sc 47 Ti 49 Ti 50 V 51 V 53 Cr 55 Mn 57 Fe 59 Co 61 Ni 63 Cu 65 Cu 67 Zn 69 Ga 71 Ga 73 Ge 75 As 77 Se 79 Br 81 Br 85 Rb 87 Rb 87 Sr 89 Y 91 Zr 93 Nb 95 Mo 97 Mo 99 Ru 101 Ru 103 Rh 105 Pd
2 1
Some properties of stable magnetic nuclei
% Abund.085 0.4 100.7
.2 1984 À834 2000 2142 696 4073 5176 À788 4890 6711 10697 11529 346 704 À584.1 278.985 0.0 682.2175 À0.4687 0.6 1680 249.Physical Constants.28319 À1.068 À0.32 À0.16 À0.0 2.41 0.13742 À1.7 121.1 52.4
69.44 5.0 4.56227 À0.0884 À0.66
I 1/2 1 1 3/2 3/2 3 3/2 1/2 1 1/2 5/2 1/2 3/2 5/2 5/2 1/2 1/2 3/2 3/2 3/2 3/2 3/2 7/2 7/2 5/2 7/2 6 7/2 3/2 5/2 1/2 7/2 3/2 3/2 3/2 5/2 3/2 3/2 9/2 3/2 1/2 3/2 3/2 5/2 3/2 9/2 1/2 5/2 9/2 5/2 5/2 5/2 5/2 1/2 5/2
35.68412 0.6289 2.04 100.060 À0.15 0.331 0.11 99.6438 0.6 À260.8 À213.0 15.02 0.9
80.32 0 0.31 72.258 7.83 4.17 0.63 50.2 41.00 100.7024 0.0 11.8 30.0 427.05 À0.352 2.0 735.014 0.53506 2.8574 0.211 À0.4 207.17 27.0 22.8755 2.1064 2.46 0 0.0 117.26 0.31 0 0.8 2198 À662 À676 À525 À588 À410 104P (cmÀ1)b m (nm)a 2.3 À47.079 0.82187 0.7 À417 À91.9 152.6415 À0.0 100.82 146.12 100.2 À162.0 216.8 À365.30 0.1 1.7928 0.9 83.50 100.220 0.2703 1.7 À150.5 À53.6 569 850 1260 604 À424 À1755 2066 309.0 À125.40 117.01639 2.642 1028Q (m2)a 0 0.0200 0 À0.3457 5.6886 0.730 0.199 0.41 0.9142 À0.22 100.0 7.1776 1.0 À40.2 1120 1389 À249.17 0.28 À95.76 75.5 À140.5553 1.22 93.0 7.24 À0. Conversion Factors.3 À40.83 7. and Properties of Nuclei
153
Table A 1.4 À62.78 24.195 0.135 100.026 0 0.8938 2.1 1304 À1532 4438 1155 1909 1589 76.0008 À0.049 0.87947 1.25 99.a 99.0936 À0.63 0.69 49.0 1.041 0.0 19.06 100.4 89.39146 0.21487 À1.61 282.2564 À1.78848 À1.15 0.148706 0.3173 4.23 0.21 À0.30362 6.76 À51.7188 À0.4 109.683 100.11 39.1705 À0.4 121.1 60.104 0.34 0 À0.47464 3.3 À32.8006 2.3 170.
0 537.79 À0.81 0 0 À0.11357 À0.90 2.a 51.151 0.15 37.0 12.11778 3.8 41.13069 À0.65 100.83 100.529 5.111 1028Q (m2)a 0 0 0 0 0.164 0.60 13.2 41.6 60.71 7.4 37.59 À3.79 7.7830 4.6095 0.84 48.275 À0. Ref.74 150.07 100.1871 3.39 0 0 0 À0.9 382.80 15.0 14.14575 0.4 148.67 À0.19 14.16 12.29 95.838 0. 1.2 À195.872 1.0 91.31 37.0473 3.0 À63.23 99.672 À0.541 À1.66 À0.480 0.337 3.86 0.99 13.0 16.1 À60.0 677.87 13.0 100.1 100.81 52.0 À59.1 295.9 24.3 62.550 À0.0010 À1.2197 0.8 À46.1 97.59 57.004 0.4 À33.6 93.79 3.3 À78.5639 À0.57 0 0 2.0 À875.59 11.55 0 0.0 À21.40 2.96 16.7935 À0.154
Appendix 1
Table A1.0 À65.0 À122.21 42.60 1.80 0.0 22.8 16.37
I 1/2 1/2 1/2 1/2 9/2 9/2 1/2 1/2 5/2 7/2 1/2 5/2 7/2 3/2 3/2 7/2 5/2 7/2 7/2 7/2 7/2 5/2 5/2 3/2 3/2 3/2 5/2 5/2 7/2 7/2 1/2 1/2 5/2 7/2 7/2 9/2 7/2 1/2 5/2 5/2 1/2 3/2 3/2 3/2 1/2 3/2 1/2 3/2 1/2 1/2 1/2 9/2
107
79.0 À450.0 237.18 29.31 À0. 3
Nucleus Ag Ag 111 Cd 113 Cd 113 In 115 In 117 Sn 119 Sn 121 Sb 123 Sb 125 Te 127 I 133 Cs 135 Ba 137 Ba 139 La 141 Pr 143 Nd 145 Nd 147 Sm 149 Sm 151 Eu 153 Eu 155 Gd 157 Gd 159 Tb 161 Dy 163 Dy 165 Ho 167 Er 169 Tm 171 Yb 173 Yb 175 Lu 177 Hf 179 Hf 181 Ta 183 W 185 Re 187 Re 187 Os 189 Os 191 Ir 193 Ir 195 Pt 197 Au 199 Hg 201 Hg 203 Tl 205 Tl 207 Pb 209 Bi
109
a b
(continued )
% Abund.78 417.80 12.48 22.
.41 18.7 33.0 À55.533 À2.815 À0.9 100.245 0.52 70.064652 0.4 À0.65993 0.51 2.8133 2.0 542.18 2.5 0 À0.49 3.17 À0.6 À430.4 4403 1072 1165 11478 959 13969 À15470 60711 61308 27188 25860 104P (cmÀ1)b À68.0
Ref.9 À28.014 À0.75 0 0.3 429.62230 5.3 14.3 89.16 0.2327 0.91 100.0 44.07 0 0.26 0.3 0 2.99 14.40 62.65 3.641 2.63822 0.08 À0.36 1.075 0.9 100.43 2.6 À584.0 18.22 4.0 524.0 492.363 2.82 47.20 À0.0 6.582 0.49 0.1 91.0 m (nm)a À0.30 1.80 3.6799 2.4937 À0.6 161.1 236.370 0.56023 1.0 397.60 À0.0 14.50588 À0.59258 4.82 0.68 8.18 8.0 386. 2.4 À595 710 717.0 À611.8885 2.232 0.0 553.62 99.62226 1.59489 À0.0 104As (cmÀ1)b À611 À703 À4553 À4763 6731 6746 À14002 À14650 11708 8878 À18542 13876 823 1220 1324 2004 4166 À793 À481 À672 À554 1086 1909 À647 À849 4546 À705 988 4523 À645 À1946 889 À1224 3546 1471 À1188 5010 1927 11718 11838 431.673 4.41 1.9374 2.
3d0. responsible for the isotropic coupling. and hrÀ3ind and hrÀ3i(n11)p are given by eqns (1) and (3).Physical Constants. Using these equations and the parameters given in Table A1. and Properties of Nuclei
155
Dipolar hyperﬁne coupling parameters for the transition metals and (n þ 1)porbitals were computed from SCF Hartree–Fock–Slater atomic orbitals [3].1: À3 r 3d ¼ 3:99 À 0:517 ðÀ2:1Þ À 0:062 ð0:1Þ À 0:017 ð0:0Þ ¼ 5:07 Â 2 Ã c ð0Þ 4s ¼ 47:8 À 16:0 ðÀ2:1Þ À 6:9 ð0:1Þ À 3:2 ð0:0Þ ¼ 80:7 À3 r 4p ¼ 2:69 À 1:70 ðÀ2:1Þ À 0:97 ð0:1Þ ¼ 6:16 Using eqn (5a): P3d ¼ 21:2 Â 10À4 cmÀ1 P4p ¼ 25:7 Â 10À4 cmÀ1
.3.94s0. is given by eqn (3). the P and isotropic coupling parameters can be computed for each of the transition metals (Table A1.4). The sorbital contribution. The parameter P is given by eqn (5) below. Conversion Factors. Dnd is the diﬀerence between the number of delectrons present in the ion of interest and the number present in the neutral metal. À3 r nd ¼ A þ BD nd þ Cns þ Dnp ð1Þ 8p Â 2 Ã c ð0Þ ðnþ1Þs ¼ A þ BD nd þ Cns þ Dnp 3 À3 r ðnþ1Þp ¼ A þ BD nd þ Cns F¼ ÁÀ Á ge gN m0 mB mN À 6:74834 Â 1030 au3 mÀ3 10À6 MHz sÀ1 4ph P ¼ ge gN mB mN rÀ3 P ¼ 4:180hrÀ3 i Â 10À4 cmÀ1 ð2Þ ð3Þ ð4Þ ð5Þ ð5aÞ
Example
For Cr(II).
107 139.57 0.26 140.8 1.67 À49.648 33.45 0.55 5.840 66.31 126.73 À0.7 3.129 À18.4 À 0.140 12.7 2.08 À0.3 À 0.591 27.3 1.8 2.0 1.067 9.64 0.7 3.980 128.137 14.9 À 0.89 À24.6 2.134 15.65 0.127 À20.1 0.49 0.60 hrÀ3i3d [c2(0)]4s hrÀ3i4p hrÀ3i3d [c2(0)]4s hrÀ3i4p hrÀ3i3d [c2(0)]4s hrÀ3i4p hrÀ3i3d [c2(0)]4s hrÀ3i4p hrÀ3i3d [c2(0)]4s hrÀ3i4p hrÀ3i3d [c2(0)]4s hrÀ3i4p hrÀ3i3d [c2(0)]4s hrÀ3i4p hrÀ3i3d [c2(0)]4s hrÀ3i4p hrÀ3i4d [c2(0)]5s hrÀ3i5p hrÀ3i4d [c2(0)]5s hrÀ3i5p hrÀ3i4d [c2(0)]5s hrÀ3i5p hrÀ3i4d [c2(0)]5s hrÀ3i5p hrÀ3i4d [c2(0)]5s hrÀ3i5p hrÀ3i4d [c2(0)]5s hrÀ3i5p hrÀ3i4d [c2(0)]5s hrÀ3i5p hrÀ3i4d [c2(0)]5s hrÀ3i5p hrÀ3i5d [c2(0)]6s hrÀ3i6p Ab 2.009 2.86 0.4 2.64 9.07 0.26 0.61 ÀCb 0.88 0.3 4.0 À 0.52 À30.32 0.2 3.0 1.559 17.08 0.156
Appendix 1
Table A 1.81 130.700 23.650 21.7 2.050 5.990 47.6 À À0.42 3.024 4.67 À24.060 6.604 9.80 ÀBb 0.3 2.062 6.53 À21.97 0.46 À16.58 151.841 53.444 12.072 11.31 151.70 À24.057 5.29 7.23 6.3 1.054 7.066 8.5(4) 0.44 0.132 17.50 À27.6 1.8 2.045 5.00 0.1 4.34 274 6.710 À40.951 106.2 1.4 À47.7 5.9 4.07 À16 À
I 5/2 7/2 6 7/2 3/2 5/2 1/2 7/2 3/2 3/2 3/2 5/2 9/2 5/2 5/2 9/2 5/2 5/2 1/2 5/2 1/2 1/2 7/2 9/2
Ti Ti V V Cr Mn Fe Co Ni Cu Cu Zr Nb Mo Mo Tc Ru Ru Rh Pd Ag Ag Hf Hf
55
57
59
61
63 65 91
93
95 97 99
99
101 103
105
107 109 177 179
.22 À0.37 0.8 À3.270 80.554 37.479 14.2 3.2 2.5 À 0.013 2.025 4.15 3.67 120.3 1.93 6.53 0.21 ÀDb 0.6 2.7 À 0.407 75.021 3.22 132.30 À36 À2.055 À8.77 9.7 À À0.49 À0.6 À 0.9 À3.93 À0.7 À 0.4 À 0.51 0.565 24.76 0.5 1.6 À 0.0 4.2 6.6 5.69 4.24 À0.4 1.2 À2.5 5.542 21. 4
Isotope
47 49 50 51 53
EPR hyperﬁne coupling parameters for the transition metals
Fa À30.1 2.997 73.66 À13.1 À À0.98 À34.8 À 0.7 2.93 À35.0 À 0.142 11.678 37.144 10.619 À30.024 4.055 7.6 À2.8 3.744 À44.020 3.517 16.96 5.13 À30.53 17.78 0.70 4.9 0.35 0.8 1.99 À73 À3.14 53.95 21.50 7.070 0.3 1.19 0.98 10.76 0.929 117.20 0.6 1.76 141.70 0.017 3.036 4.1 3.04 96.27 0.5 1.229 42.03 0.750 25.064 7.14 5.2 À 0.197 86.5 1.05 3.054 À8.0 2.30 6.789 59.053 6.
11 À26 À À0. P. 30.96 À1. 577. 3.11 À24 À À0.25 À63 À3.28 À44 À2.Physical Constants. Boca Raton.30 À1.H.10 À143 À5..285 hrÀ3i5d [c2(0)]6s hrÀ3i6p hrÀ3i5d [c2(0)]6s hrÀ3i6p hrÀ3i5d [c2(0)]6s hrÀ3i6p hrÀ3i5d [c2(0)]6s hrÀ3i6p hrÀ3i5d [c2(0)]6s hrÀ3i6p hrÀ3i5d [c2(0)]6s hrÀ3i6p hrÀ3i5d [c2(0)]6s hrÀ3i6p Ab 7.16 9.66 ÀDb À0. CRC Press. 140.32 À1.05 À0. D refer to eqns (1)–(3). Morton and K. B. Magn.34 42. 2. 1978. and Properties of Nuclei
157
Table A1. FL. J.84 À0.50 121.09 À18 À À0.30 417 9.24 À68 À3.
References
1.77 11.00 À90 À4.77 123. D.51 310 7. ed.10 À20 À À0.26 À58 À3. Lide.64 530 10.10 À29 À
I 7/2 1/2 5/2 5/2 1/2 3/2 3/2 3/2 1/2 3/2
Ta W Re Re Os Os Ir Ir Pt Au
183
185 187 187 189 191 193 195
197
a b
Factor deﬁned by eqn (4).02 À103 À4.F.62 10. 83rd edn. Reson.02 9.26 À0. Rieger.12 491 10. Reson.03 À110 À4.45 À0.80 ÀBb À0. C.63 À0.11 À27 À À0.64 À1.72 345 8.26 À53 À3.44 116.97 ÀCb À0. 2002.98 À1.27 À52 À2.34 15. R.02 12.99 À81 À3. 1997.42 À0. Handbook of Chemistry and Physics.68 454 9.68 22. 124.11 À25 À À0. Parameters A.85 14.98 387 8.65 À1.44 9. Preston.51 8. Conversion Factors.29 À40 À2.05 À120 À5.33 12. Magn. 4
Isotope
181
(continued )
Fa 64. J.08 À131 À5.
. J.R..
The last two decades. computers and.
158
. for example. which have by now taken over NMR instrumentation. microwave technology. of the sort now employed. Double resonance techniques.APPENDIX 2
Advanced ESR Methods
For most of the 60 years since ESR was discovered. the vast majority of experiments have been carried out using CW spectrometers operating at the Xband frequency and employing conventional electromagnets. Most of these advanced methods fall into one of three broad categories: High ﬁeld/multifrequency ESR Double resonance Pulsed methods Although several of these methods were ﬁrst applied in the 1950s and 1960s. on the other hand. One group of reviewers1 has actually attributed much of the rapid development in all three categories to declassiﬁcation of research in high frequency and communications technology that was an outgrowth of the end of the Cold War! The ready availability of this technology is also reﬂected in the fact that at least one supplier of commercial ESR equipment now oﬀers a high ﬁeld. most commercially available instruments are still of the general type described in Chapter 1. to some extent. albeit with enhancements made possible by advancements in electronics. but even more importantly employs stable frequency sources and sensitive detectors in the millimeter and submillimeter range that have only recently become available. have greatly beneﬁted from advances in signal processing technology. developed primarily for NMR. however. in wireless communication systems. of which the earliest was ENDOR (described in Chapter 1). In particular: High frequency ESR has required not only the use of superconducting magnets. have required the development of high power ampliﬁers and fast switches for microwave and higher frequency radiation. pulsed. Indeed. have seen ever increasing development and applications of other ESR techniques. Pulsed methods in ESR. none would have been possible at their current technical level using the postWorld War II technology that gave rise to Xband ESR.
double resonance. which has revolutionized the ﬁeld of highresolution NMR. The impact of these new methods has been felt particularly strongly in two areas: (i) Improvements in eﬀective resolution using high ﬁelds. is beyond the scope of this book. The titles of the papers that we have referenced will make it clear that we have tried to include applications of relevance not only to organometallic chemistry but also to biochemistry and related ﬁelds.6. radiation chemistry and other areas of mechanistic chemistry.Advanced ESR Methods
159
double resonance spectrometer. In this.
A2. For example. have proved to have less impact on ESR. the ﬁrst indication of such eﬀects often appears as poorer resolution of individual features because of increased apparent line width as the anisotropy is revealed. especially with randomly oriented samples. Reviews of these and other recent chemical applications of ESR have been written by Christopher Rhodes.13 Presented below are brief descriptions of some of the applications to structure analysis to which each of the three advanced methods are making important contributions. which has led to advances in the understanding of reactive intermediates in photochemistry. Improved resolution with increasing frequency is also not a forgone conclusion. While it is certainly possible. (ii) Nanosecond pulsed methods have made possible the detection and characterization of various transient paramagnetic intermediates that could be previously studied by optical methods but were inaccessible to ESR. smaller wavelength. One of these is described below. i.
. better sensitivity and improved resolution of ﬁelddependent features.. and pulsed methods have made possible the analysis and interpretation of spectra from increasingly complex paramagnetic materials. especially those of biological and solid state relevance.e. The latter application. Nevertheless many recent studies simply could not have been carried out without the use of high frequency EPR.1 High Frequency ESR
The two most obvious advantages of high frequency/high ﬁeld spectrometers. an instrument that incorporates advanced technology in all three categories. For the reader who wishes to learn more about these methods and applications a list of recent reviews and other leading references to these applications is also included. the smaller sample sizes that must be used when higher frequency. for example. radiation is used have made it necessary in many cases to work with smaller numbers of spins and therefore not fully exploit the sensitivity advantage inherent in use of the higher frequency. to detect smaller gfactor anisotropies at high ﬁelds.
cooled to 10 K.160
Appendix 2
and most other studies to date. The sample studied is a frozen aqueous solution of Cr21 sulfate (3d4. microwave frequency as indicated. 25. Experimental conditions: temperature 10 K. higher frequencies are able to induce transitions. A recent example of an application of high ﬁeld ESR (‘‘HFEPR’’) is shown in Figure A2.1 Microwave frequency dependent HFEPR spectra of aqueous Cr21 (0. copyright (1998) American Chemical Society. and the energy spacing between the two lowest levels is too large to be spanned by a microwave quantum at Xband.1.)
.1– 0. S ¼ 2). This corresponds to an ‘‘EPRsilent’’ sample that gives no detectable ESR spectrum at Xband frequencies because it possesses a zeroﬁeld splitting larger than the Zeeman interaction (see Chapter 6).2). In the spectrum taken at 329 GHz a sharp signal from aqueous Cr31 impurity at g ¼ 2 is indicated and the resonances due to Cr21 are labeled (Figure A2. Since
Figure A2. sulfate counterion. custombuilt equipment was necessary because no commercial instruments are available yet at very high frequencies. (Reprinted with permission from ref. Nevertheless.2 M).
2
Plot of resonance ﬁeld versus microwave frequency for data from Figure A2.
A2.98(2). copyright (1998) American Chemical society. eﬀectively performing simultaneous NMR and ESR.2 Double Resonance
In Chapter 2. Least squares ﬁts are given for each Cr21resonance (dotted line) and for the Cr31 impurity (solid line).2 shows the frequency dependence of the various spectral features.20(5) cmÀ1. C and E correspond to DmS 4 1 while the rest correspond to DmS ¼ 1. and a nearly isotropic gfactor. ENDOR (electron–nuclear double resonance) was brieﬂy described.1. g ¼ 1. E ¼ 0.0(1) cmÀ1. Various recent applications of high ﬁeld and multifrequency EPR are described in refs 19–31. Analysis was aided by the observation of diﬀerent slopes of the B vs. as shown in Figure A2. none of which could have been determined at Xband. 11.)
the energy level scheme is quite complex. A review of advanced methods. including highﬁeld EPR. with the ESR saturating frequency centered on a
. respectively. To perform an ENDOR experiment it is necessary to apply both a radiofrequency and a microwave frequency. is given in ref.
(Reprinted with permission from ref.Advanced ESR Methods
161
Figure A2. n plots for DmS 4 1 and DmS ¼ 1 transitions. on the sample. The experiment is performed at a ﬁxed magnetic ﬁeld. 25.1 the spectrum varies markedly with frequency. Figure A2. Analysis of the spectra at diﬀerent frequencies yielded the parameters D ¼ À2.
Fortunately. in Chapter 2 that hyperﬁne patterns from 14N and 1H splittings can be quite similar and proper analysis of the spectrum requires recognition of subtle diﬀerences in peak intensities. where some pathways are eﬀectively frozen out. in extreme cases. Some of these methods also make it possible to separate in the time domain overlapping transitions that have diﬀerent relaxation behavior. and one variable frequency that searches for transitions.41. The number of observed lines is reduced. one ﬁxed and saturating. making it easier to detect small hyperﬁne splittings.3 Pulsed Methods
The recent advent of the ability to apply short and very intense microwave pulses to samples and detect the fast response to the excitation has made it
. thereby improving the resolution of the spectrum. may be undetectable. A major limitation of CW double resonance methods is the sensitivity of the intensities of the transitions to the relative rates of spin relaxation processes. For that reason the peak intensities often convey little quantitative information about the numbers of spins involved and.44 It has been used much less than ENDOR and usually requires custombuilt equipment. especially at low temperatures. In that case one can better adapt the excitation and detection timing to the rates of relaxation that are intrinsic to the sample. even though the number of nuclei giving rise to the coupling is not revealed. This makes it possible to extract anisotropic spectral parameters that must otherwise be obtained using single crystals or laborious simulation of the full spectrum. originally proposed by Overhauser. This limitation can be especially severe for liquid samples where several relaxation pathways may have about the same rates.
A2. This technique is known as ELDOR (electron–electron double resonance). fewer limitations occur when pulsed radio and microwave ﬁelds are employed.38.162
Appendix 2
spectral feature usually identiﬁed from a CWESR experiment. The RF ﬁeld at the NMR frequency is then varied.50 There are now several versions of pulsed ENDOR and other double resonance methods.32–34. Furthermore. one immediately knows from the NMR frequency what type of nucleus is involved. We saw. This in turn demands resolution that is almost never available in randomly oriented samples. In those types of samples the advantages of double resonance are carried out by employing the use of two diﬀerent microwave frequencies. ENDOR is playing an increasingly important role since careful selection of the saturated features in the ESR spectrum leads to excitation of molecules over only a narrow range of orientations. In the case of such samples.39.35 As discussed in Chapter 6. The situation is somewhat better in solids. When the RF frequency matches the appropriate NMR transition frequency the ESR absorption reappears.43 The relatively old technique of Dynamic Nuclear Polarization (DNP). for example.40. in which the NMR spectrum is observed during irradiation of ESR transitions. is being rejuvenated as a method of enhancing the NMR spectra of solids. in systems with more than one unpaired electron the ESR spectrum contains features that involve electron–electron coupling parameters analogous to the nuclear hyperﬁne parameters.42.
18 than for pulsed NMR. ESEEM experiments are usually run on frozen samples at low temperatures both to lengthen decay times and to achieve the anisotropic eﬀects that make the desired transitions possible. This is a valuable supplement to the very widespread use of multidimensional NMR in structural biology.12.48.. The echo intensity as a function of t may then be Fourier transformed to yield a frequency domain ‘‘spectrum’’ that looks a lot like ENDOR. on the order of the reciprocal of a hyperﬁne coupling in frequency units. as well as information about the amplitude modulation by nuclear spin ﬂips under the inﬂuence of A.to nanoseconds.55 Nevertheless. i. cases where A o 10 MHz. however. Applications of Fouriertransform ESR.9. They mimic the ubiquitous pulsed methods that have revolutionized NMR.e. transitions are excited. T2.. double quantum. are still in their infancy and require custombuilt equipment.33
.14 ENDOR and ESEEM are also sometimes employed together. since it can also be used to study weakly coupled nuclei. Several recent reviews of the technique are available. and related multiplepulse twodimensional techniques now common in NMR.2. In its simplest form. double resonance methods is now making it possible to use ESR as a tool to determine distances within macromolecules.e.Advanced ESR Methods
163
possible to collect an ESR freeinduction decay in the time domain and to Fourier transform the result and view the data in the frequency domain. As with ENDOR.46. ESEEM spectra can be quite sensitive to relaxation behavior so that failure to observe a modulation does not prove the absence of a coupled nucleus. For an ESEEM experiment to be successful the applied microwave pulse must be such that both allowed and semiforbidden. due both to the much shorter ESR relaxation times and to the much wider extent of ESR spectra in the frequency domain. It is also necessary for the deadtime. both of which require shorter.49 The combination of higher ﬁelds and pulsed. i. This ‘‘envelope’’ has in it two sorts of information. It has taken the ability to do this much longer to develop for ESR than for NMR. the phenomenon appears as modulation of the intensity of the ‘‘echo’’ detected after a sequence of two pulses separated by a time t. the phenomenon of Electron Spin Echo Envelope Modulation (ESEEM) that is growing very rapidly and is in some instances replacing ENDOR as a way of extracting parameters from randomly oriented samples. however. Interestingly. more intense pulses and much faster detector response8.47. to be short compared with the time over which the modulations are observable. ESEEM experiments can also be performed using more than two pulses.16. There must also be enough nuclear modulations within the time before the echo decays to deﬁne the observable values of A. Such experiments are an outgrowth of the pulsed methods commonly used to determine relaxation times. the most robust version of the method works best in randomly oriented solids with overlapping lines. there is one type of pulsed ESR measurement. giving somewhat greater ﬂexibility in the range of relaxation times over which modulations can be observed. typically micro. the time needed between pulses before the detector can record new information.54. an approximately exponential decay corresponding to the phase memory time.
164
Appendix 2
References
In developing this short bibliography of advanced ESR methods we have tried to select references primarily from widely available journals, review series and books published after 1990. We have therefore avoided references to meeting abstracts or highly specialized journals with low circulation or earlier publications where the methods were ﬁrst demonstrated. Citations to these pioneering papers are found in the references. For ease of recognition of papers relevant to a particular method or application we have included the title of the review or paper within the citation. The references are grouped by topic and listed in reverse chronological order within each topic. They are numbered consecutively.
Books and Reviews
1. Highﬁeld EPR, Magn. Reson. Chem., Special Issue, ed. W. Lubitz, K. Mobius and K.P. Dinse, 2005, 43, S1–S266. 2. G.R. Eaton and S.S. Eaton, Electronnuclear double resonance spectroscopy and electron spin echo envelope modulation spectroscopy, Comprehensive Coordination Chemistry II, Elsevier, Boston, 2004, 49. 3. C. Coulon and R. Clerac, Electron spin resonance: A major probe of molecular conductors, Chem. Rev., 2004, 104, 5655. 4. Biomedical EPR, ed. S.S. Eaton, G.R. Eaton and L.J. Berliner, Kluwer Academic/Plenum Publishers, New York, 2004. 5. L.J. Berliner, ed., In vivo EPR (ESR): Theory and Applications, Kluwer Academic/Plenum Publishers, New York, 2004. 6. C.J. Rhodes, Electron spin resonance (some applications for the biological and environmental sciences), Annu. Rep. Prog. Chem., Sec. C, 2004 100, 149. 7. Paramagnetic Resonance of Metallobiomolecules, ed. J. Telser, ACS Symposium Series, 858, American Chemical Society, 2003. 8. T. Prisner, M. Rohrer and F. MacMillan, Pulsed EPR spectroscopy: Biological applications, Annu. Rev. Phys. Chem., 2001, 52, 279. 9. N.D. Chasteen and P.A. Snetsinger, ESEEM and ENDOR spectroscopy, in Physical Methods in Bioinorganic Chemistry, Spectroscopy and Magnetism, ed. L. Que, Jr, University Science Books, Sausalito, CA, 2000. 10. G. Palmer, Electron paramagnetic resonance of metalloproteins, in Physical Methods in Bioinorganic Chemistry, Spectroscopy and Magnetism, ed. L. Que, Jr, University Science Books, Sausalito, CA, 2000. 11. J.H. Freed, New technologies in electron spin resonance, Annu. Rev. Phys. Chem., 2000, 51, 655. 12. Y. Deligiannakis, M. Louloudi and N. Hadjiliadis, Electron spin echo envelope modulation (ESEEM) spectroscopy as a tool to investigate the coordination environment of metal centers, Coord. Chem. Rev., 2000, 204, 1. 13. C.J. Rhodes, Electron spin resonance, Annu. Rep. Prog. Chem., Sect. C, 1999, 95, 199.
Advanced ESR Methods
165
14. J. McCracken, Electron spin echo modulation, in Handbook of Electron Spin Resonance, ed. C.P. Poole and H.A. Farach, SpringerVerlag, New York, 1999, vol. 2. 15. M. Ikeya, New Applications of Electron Spin Resonance: ESR Dating, Dosimetry and Microscopy, World Scientiﬁc Publishing Co., Singapore, 1993. 16. A. Schweiger, Pulsed electron spin resonance spectroscopy: Basic principles, techniques, and examples of applications, Angew. Chem. Int. Ed. Engl., 1991, 30, 265. 17. L.R. Dalton, A. Bain and C.J. Young, Recent advances in electron paramagnetic resonance, Annu. Rev. Phys. Chem., 1990, 41, 389. 18. Modern Pulsed and ContinuousWave Electron Spin Resonance, ed. L. Kevan and M.K. Bowman, John Wiley and Sons, New York, 1990.
Selected Papers from the Recent Literature High Field/Multifrequency EPR:
19. K. Mobius, A. Savitsky, A. Schnegg, M. Plato and M. Fuchs, Highﬁeld ¨ EPR spectroscopy applied to biological systems: Characterization of molecular switches for electron and ion transfer, Phys. Chem. Chem. Phys., 2005, 7, 19. 20. T.A. Konovalova, L.D. Kispert, J. van Tol and L.C. Brunel, Multifrequency highﬁeld electron paramagnetic resonance characterization of the peroxyl radical location in horse heart myoglobin oxidized by H2O2, J. Phys. Chem. B, 2004, 108, 11820. 21. J. Krzystek, S.A. Zvyagin, A. Ozarowski, A.T. Fiedler, T.C. Brunold and J. Telser, Deﬁnitive spectroscopic determination of zeroﬁeld splitting in highspin colbalt(II), J. Am. Chem. Soc., 2004, 126, 2148. 22. K.K. Andersson, P.P. Schmidt, B. Katterle, K.R. Strand, A.E. Palmer, S.K. Lee, E.I. Solomon, A. Graslund and A.L. Barra, Examples of high¨ frequency EPR studies in bioinorganic chemistry, J. Biol. Inorg. Chem., 2003, 8, 235. 23. J. van Slageren, R. Sessoli, D. Gatteschi, A.A. Smith, M. Helliwell, R.E.P. Winpenny, A. Cornia, A.L. Barra, A.G.M. Jansen, E. Rentschler and G.A. Timco, Magnetic anisotropy of the antiferromagnetic ring [Cr8F8Piv16], Chem. Eur. J., 2002, 8, 277. 24. Z. Liang and J.H. Freed, An assessment of the applicability of multifrequency ESR to study the complex dynamics of biomolecules, J. Phys. Chem. B, 1999, 103, 6384. 25. J. Telser, L.A. Pardi, J. Krzystek and L.C. Brunel, EPR spectra from EPRsilent species: Highﬁeld spectroscopy of aqueous chromium(II), Inorg. Chem, 1998, 37, 5769. 26. D. Collison, M. Helliwell, V.M. Jones, F.E. Mabbs, E.J.L. McInnes, P.C. Riedi, G.M. Smith, R.G. Pritchard and W.I. Cross, Single and double quantum transitions in the multifrequency continuous wave electron paramagnetic resonance (cwEPR) of three sixcoordinate nickel(II) complexes:
166
À
Appendix 2
27.
28.
29.
30.
31.
[Ni(EtL)2(Me5dien)] and [Ni(5methylpyrazole)6]X2, X ¼ (ClO4) or (BF4)À. The single crystal Xray structure at room temperature of [Ni(5methylpyrazole)6]ClO4)2, J. Chem. Soc., Faraday Trans., 1998, 94, 3019. P.J.M. van Kan, E. van der Horst, E.J. Reijerse, P.J.M. van Bentum and W.R. Hagen, Multifrequency EPR spectroscopy of myoglobin. Spectral eﬀects for highspin iron(III) at high magnetic ﬁelds, J. Chem. Soc., Faraday Trans., 1998, 94, 2975. E.J.L. McInnes, F.E. Mabbs, S.M. Harben, P.D. Smith, D. Collison, C.D. Garner, G.M. Sith and P.C. Riedi, Singlecrystal and multifrequency EPR studies on chemical analogues of Amavadin00 V(IV)doped Ca[Ti (hida)2].6H2O, and Mo(V)doped [PPh4][Nb(hida)2] and [NEt4][Ta (R,Rhidpa)2] [H3hida ¼ 2,2 0 (hydroxyimino)diacetic acid, H3hidpa ¼ 2,2 0 (hydroxyimino)dipropionic acid], J. Chem. Soc., Faraday Trans., 1998, 94, 3013. A. Klein, E.J.L. McInnes, T. Scheiring and S. Zalis, Electronic structure of radical anionic binuclear organoplatinum complexes. A multiple frequency EPR investigation, J. Chem. Soc., Faraday Trans., 1998, 94, 2979. B. Cage, A.K. Hassan, L. Pard, J. Krzystek, L.C. Brunel and N.S. Dalal, 375 GHz EPR measurements on undiluted Cr(V) salts. The role of exchange eﬀects and gstrain broadening in determining resolution in highﬁeld EPR spectroscopy of S ¼ 1/2 paramagnets, J. Magn. Reson., 1997, 124, 495. W.B. Lynch, R.S. Boorse and J.H. Freed, A 250GHz ESR study of highly distorted manganese complexes, J. Am. Chem. Soc., 1993, 115, 10909.
Double Resonance:
32. J.C. Wilson, G. Wu, A.I. Tsai and G.J. Gerfen, Determination of the structural environment of the tyrosyl radical in prostaglandin H2 synthase1: A high frequency ENDOR/EPR study, J. Am. Chem. Soc., 2005, 127, 1618. 33. (a) D. Goldfarb and D. Arieli, Spin distribution and the location of protons in paramagnetic proteins, Annu. Rev. Biophys. Biomol. Struct., 2004, 33, 441; (b) S.S. Eaton and G.R. Eaton, Measurement of interspin distances by EPR, Electron Paramagnetic Reson., 2004, 19, 318. 34. O.G. Poluektov, L.M. Utschig, A.A. Dubinskij and M. Thurnauer, ENDOR of spincorrelated radical pairs in photosynthesis at high magnetic ﬁeld: A tool for mapping electron transfer pathways, J. Am. Chem. Soc., 2004, 126, 1644. 35. K.N. Hu, H.h. Yu, T.M. Swager and R.G. Griﬃn, Dynamic nuclear polarization with biradicals, J. Am. Chem. Soc., 2004, 126, 10844. 36. S. Sinnecker, E. Reijerse, F. Neese and W. Lubitz, Hydrogen bond geometries from paramagnetic resonance and electronnuclear double resonance parameters: Density functional study of quinone radical anionsolvent interactions, J. Am. Chem. Soc., 2004, 126, 3280. 37. B.M. Hoﬀman, (a) Electronnuclear double resonance spectroscopy (and electronspinecho envelope modulation spectroscopy) in bioinorganic
B.M. Chem. 3575. Hoﬀman.R. J. Chem. Pulsed ELDOR spectroscopy measures the distance between the two tyrosyl radicals in the R2 subunit of the E. Pulsed electronelectron double resonance on multinuclear metal clusters: Assignment of spin projection factors bsed on the dipolar interaction. Brecht and R.
43. Lee. Mino and T.. Bittl. R. J.
41. C.L. Van Doorslaer and A. Soc. Laryukhin.H. Chem. 2003. Chem.E. J. 128. 2003. Soc. Reson.. Ono. Robblee and J. Rife. 2003. J. Goldfarb. Applications of pulsed ELDORdetected NMR measurements to studies of photosystem II: Magnetic characterization of YD tyrosine radical and Mn21 bound to the highaﬃnity site. An organometallic intermediate during alkyne reduction by nitrogenase. J. Jaun. Weber.
chemistry. Acc. Inorg. U. Bernat. Seefeldt and B. Schweiger.K. Thorpe. Naumov and D. Soc. Magn. Soc. 165. Tokdemir. 3.M. J. C. Chem. Porphyrin and ligand protons as internal labels for determination of ligand orientations in ESEEMS of lowspin d5 complexes in glassy media: ESEEM studies of the orientation of the g tensor with respect to the planes of axial ligands and porphyrin nitrogens of lowspin ferriheme systems.A. Magn.
40. 12606. S. Soc. C. Telser.. Armstrong and B.
42. Lu. P. Finazzo. Electronic gfactor measurement from ENDORinduced EPR patterns: Malonic acid and guanine dihydrate. J. J. J. Natl. A. Hoﬀman. 5446. E. 45. Epel and D. A. M. Am. Smoukov. M. Am. J. Chem. D. Determi¨ nation of the distance between the two neutral ﬂavin radicals in augmenter of liver regeneration by pulsed ELDOR. 2003. Chem. Proc. 2002. A Fourier transform EPR study of uracil and thymine radical anions in aqueous solution. EPR study of substrate binding to the Mn(II) active site of the bacterial antibiotic resistance enzyme FosA: A better way to examine Mn(II). Phys. Phys. coli ribonucleotide reductase. Antonic. R.. Chem.S. Elsasser. 2002. Sci. Stubbe..A. Chem. D. (b) ENDOR of metalloenzymes.I. 2006. S. Beckert. Magn. 49. L.R. S. Nelson. 120. Kang.. 103. Igarashi. J. Acad. Phys. C. Reson. Dos Santos. 124. Bittl. Dean.Advanced ESR Methods
167
38. 571. A pulse EPR and ENDOR investigation of the electronic and geometric structure of cobaltous tetraphenylporphyrin(pyridine). Appl... P.. 36. Doan. 124.A. 1999. J. 14988. Characterization of the MCRred2 form of methylcoenzyme M reductase: A pulse EPR and ENDOR study. Walker. 2003.
39. B. J. 1998.
. Reson. Am. R. Am. R. 586. 952.
Pulsed Methods:
44. Kay. C. Res. Harmer. S. 2004. J. Twodimensional pulsed TRIPLE at 95 GHz. Bachmann and A. 46.C. 196. Van Doorslaer. Thauer. 146. C. Farrell and C. Biol. 47.R. 2003. 8. J. 100. 522.K. 2000.. Mahlert. 76.M. M. Duin.L. Am..W. 125. J.C. S. H. R. 2318. Am. Schweiger. Chem. Soc. F. 48. Geimer. Shao and W. 128. 23. Elsasser. 2001. 126. J. Chem. B. 9563.Y. S. Perlstein. 991.M. A.N. Raitsimring and F. J.. Bennati.. H.
A. 138. H.J. Fujisawa. Sakurai.M. Bennati. Faraday Trans. Phys. Chem. van Dongen. Griﬃn. Chem. bis(picolinato)oxovanadium(IV). Fukui.M. Phys. J. J. 1999. D. Twodimensional Fourier transform ESR in the slowmotional and rigid limits: 2DELDOR. Weis. 232..M. Kh. 175. Farrar. 55.J. V. Raitsimring. Lett. H. S.T. Stubbe and R. D.H. Walker. W. Bryant. Gamliel and J.168
Appendix 2
50. van Kan. R. Inati. J. 5235..J.G. 77. 52. Borbat.A.J. Twodimensional Fourier transform ESR in the slowmotional and rigid limits: SECSYESR. Inorg.. Pulsed electronnuclear double resonance (ENDOR) at 140 GHz. Gerfen. 215. Gamliel and J. Freed. P. Biochem. 2969..H. 1990. 1999.R. Y. OhyaNishiguchi. Reson. C. K. Freed.M. 445. Crepeau.R. J. Patyal... E. W.. Shokjireva and F. Chem. B. 1990. Magnetic ﬁeld (gvalue) dependence of proton hyperﬁne couplings obrained from ESEEM measurements: Determination of the orientation of the magnetic axes of model heme complexes in glassy media.
. B. M. P. Kamada and H.H. 51. Reijerse and P. 53.A. Phys.H. R. Soc. Magn. RiggsGelasco. Crepeau. 54. Chem. G.A.R. van den Berg. studied by electron spinecho modulation spectroscopy. 100.A. Patyal. 175. In vivo coordination structural changes of a potent insulinmimetic agent. 94. 1996. EPR spectroscopy of biological ironsulfur clusters with spinadmixed S ¼ 3/2 ground states. T. W. Lett. 453.M. 1998. Hagen. J. J.
68 [(C5Ph5)Cr(CO)2{PMe3}] frozen soln. spectrum 77–9
. spectrum 67. 83–9 solidstate spectra 53–9 spin Hamiltonian 52–3 applications analytical 17–18 electronic structure determination 17 modified Bloch equations 102–7 rate determination 18 aromatic proton coupling. interpretation 27–9 aromatic radical anions 24 aromatic triplet state molecules 123–4 benzene anion radical spectrum 27 biological applications 17–18 biradicals classification 112–13 exchange coupling 113–17 spin Hamiltonian 113–14 bis(diphenylphosphino)maleic anhydride (BMA). spectrum 73–4 chromium(I) pianostool complex 77–9 cobalt dithiolene complexes 86–7 cobalt(0) radical anions 66–9 conversion factors 152 [Co{SC(CF3)C(CF3)S}2P(OPh)3] frozen soln.3COD)]– frozen soln. 67. spectrum 66. 68 [CpCr(CO)2{NO}]– frozen soln. spectrum 73–4 [(C5Ph5)Co(CO)2]– frozen soln. motion 102–4 Bloch’s phenomenological model 92–4 derivation of equations 94–5 modified Bloch equations (chemical exchange) 98–102 steadystate solution 95–8 Bohr magneton 3 cavity dip 10–11 [CH2{PO(OH)2}2] irrad. noncoincidence angles 87 [CpCo(1. spectrum 66.Subject Index
Organometallic complexes have been indexed under their molecular formulae. alkyne motion in [Mo{P(OMe)3}2 (MeC≡CMe)Cp] 104–5 alternating linewidth effects 107–8 analysis frozen solution spectra 55–9 isotropic spectra 32–40 anisotropic spectra gmatrix Interpretation 59–60 “gstrain” 87–9 hyperfine matrix interpretation 60–3 noncoincident matrix axes 71–89 organic compounds 69–71 organometallic complexes 63–9. 68 [CpCo(1. single crystal spectrum 70–1 chromium nitrosyl frozen soln. spectrum 86–7 [Co{S2C2R2}2L].5COD)]– frozen soln.
motion 102–4 diphenylpicrylhydrazyl (DPPH) 13 diphosphonylmethyl radical. 117 bis(diphenylphosphino)maleic anhydride (BMA). 56 Gaussian line shapes 102 glycylglycine irrad. 161–2 ESR experiment 3–7 frequencies 4. 161–2 5doxyl stearate (5DS). 23 ENDOR spectroscopy 41–2. single crystal spectrum 69–70 gmatrix Interpretation 59–60 “gstrain” 87–9 guidelines.4'dicyanobiphenyl anion radical spectrum 31 m. interpretation of isotropic spectra 27–9 Gunn diode 8 high field/frequency ESR (HFEPR) 158. 11–12 high field/frequency ESR (HFEPR) 158. 159–61
2+
history 1 nuclear hyperfine interaction 5–7 power 12–13 pulsed ESR 158–9. 159–61 highspin transition metal ions 126–30 Hückel MO theory 27–9 extended Hückel MO (EHMO) calculations 60
. 127–8 formation constant determination 49– 51 pFPhCN spectrum 31 frozen solution spectra 55–9 GaCl(mquin)2 dilute single crystal spectrum 55. 69 definition 1 double resonance 158.170
Subject Index
Cr in aqueous solution 73–4 Cu(acac)2 frozen solution spectrum 57–8 4. 54. 7. and pdinitrobenzene radical anion 108 dinitrodurene radical anion spectrum 107–8 dinitroxide biradicals 112. P 155 double resonance 158. 161– 2 “EPRsilent” species 160–1 extended Hückel MO (EHMO) calculations 60 fast exchange limit 100–1 [Fe(CO)5]+ dilute single crystal spectrum 74–6 [Fe(CO)5]– dilute single crystal spectrum 74–6 Fermi contact interaction 27. hopping rate 108 Dynamic Nuclear Polarization (DNP) 162 electron magnetic resonance (EMR) see electron spin resonance spectroscopy Electron Nuclear Double Resonance spectroscopy see ENDOR spectroscopy electron paramagnetic resonance (EPR) see electron spin resonance spectroscopy Electron Spin Echo Envelope Modulation. ESEEM 163 electron spin resonance spectroscopy (ESR) applications 17–18 background texts 1. 13–14 fluxionality in [Co(Ph2C2)(CO){P(OMe)3}2] 105 fluxionality rates 30–2 forbidden transitions 116. parameters 70 dipolar coupling parameter. 29 field modulation 8–9. spin label 109 duroquinone radical anion spectrum 108 durosemiquinone. 158. 162–3 saturation 5 sensitivity 4–5 electron transitions 1–2.
27. 14 phase 15–16 small amplitude field modulation 8– 9 [Mo{P(OMe)3}2(MeC≡CMe)Cp] 104– 5 multiplet intensity ratios 37–8 multiplet patterns. spectrum 76–7 [Mn(dppm)2(CO)(CNBu)]2+ frozen soln. multiplet patterns 38–9 isotropic spectra guidelines for interpretation 27–9 isotropic spectra. irrad. organometallic radicals formation constant determination 49–51 line width variation 47–8 puzzling line shapes! 48–9 secondorder effects on line positions 44–7 spin Hamiltonian parameters 44 K3Cr(CN)6. VO(acac)2 106 nitroxide dinitroxide biradicals 112. frozen solution spectrum 150 isotopomer. spinorbit coupling parameters 130–2 Larmor frequency 96
. isotopomer 38–9 naphthalene anion radical spectrum 23–4 triplet state 121. 48 methylene diphosphonic acid.Subject Index
171 line positions. single crystal spectrum 70–1 2methylquinolin8olate (mquin) single crystal spectra 55 [Mn(CO)(dmpe)Cp]+ frozen soln. spinorbit coupling 131–2 modified Bloch equations (chemical exchange) 98–102 applications 102–7 modulation amplitude 15 frequency 11. spectrum 144–5 [Mn(dppe)2(CO)(CNBu)]2+ frozen soln. spectrum 88–9 [Mn(CO)2(PPh3)(C5H5)] + frozen soln. spectrum 63–5 [Mn(dppm)2(CO)(CN)]+ frozen soln. organic radicals analysis 32–40 hyperfine coupling patterns 22–5 interpretation 27–9 line positions 21–2 line widths 29–32 secondorder shifts 39–40 secondorder splittings 25–6 spin Hamiltonian parameters 26–7 isotropic spectra. 102 modified Bloch equations (chemical exchange) 98–102 spin labels. spectrum 63–5 Mn(II) doped into PbCl2. 10–12 K4V(CN)6. 117 spin labels 109–10
hyperfine matrix interpretation 60–3 intermediate exchange region 101 iron pentacarbonyl dilute single crystal spectra 74–6 [IrRh(CO)2(PPh3)2(tolNNNtol)2]+. 102 lowspin Mn(II) spectra 63–5 bis( pmethoxyphenyl)nitroxide. isotropic spectra 21–2 line width variations 29–32 Lorentzian line shapes 96–8. 122 nitrogen base exchange rate. rate of motion from 108– 10 klystron 7–8. hyperfine parameters 42 methyl radical spectrum 26. spinorbit coupling parameters 130–2 kinetic studies alternating linewidth effects 107–8 Bloch’s phenomenological model 92–8 Gaussian line shapes 102 Lorentzian line shapes 96–8.
resonance 5 secondorder effects shifts in line positions 39–40. reactions 22–5 organic systems anisotropic spectra 69–71 isotropic spectra 32 origins of ESR 1 Pascal triangle 23. parameters 70 physical constants 152 platinum radical. 135– 43 nuclear magnetic resonance (NMR) spectra 22 operating parameters. 16 single crystal spectra 54–5 singlyoccupied π* MO (SOMO) 27.1. 28 spin traps 18 spinorbit coupling (S>1/2) 122–6 stable magnetic nuclei. 44–7 splittings 25–6 sensitivity. hyperfine pattern 39 powder or frozen solution spectra 55–9 pulsed ESR 158–9. 29. 11–12 modulation 8–9. rate of motion from 108–10 spin polarization 27. optimization 11 orbital angular momentum operations 60 organic radical anions.2.3. spectrum 79–81 resonance field vs microwave frequency 161 RhIr complex 149–51
[RhLL'{μ(tol)NNN(tol)}2IrLL']+PF6– spectrum 150 rotational correlation time 48 saturation. ESR spectrometer 4–5 [SFeCo2(CO)9]– frozen soln. spectrum 79–81 signaltonoise ratio 8. properties 152–5 1. noncoincident matrix axes electron Zeeman term 133–5 nuclear hyperfine interaction 135–43 quadrupole coupling 146–9 [PhCCo3(CO)9]– spectrum 48–9.3tetracyanopropene dianion radical spectrum 32 transition metal complexes highspin transition metals 126–30 hyperfine coupling parameters 155–7 spin–orbit coupling 122–6 triplet state organic molecules 122. 123 spin Hamiltonian 117–21 transition metal complexes 122–32 zerofield splitting 120
.3pentacyanopropene dianion radical spectrum 32 perturbation theory.172
Subject Index
NO3 dilute single crystal parameters 54–5 noncoincident matrix axes experimental determination 72–3 organometallic complexes 73–87 perturbation theory 133–51 symmetry requirements 71–2 nuclear hyperfine interaction 5–7. 59–63 slow passage conditions 95 solidstate spectra 53–9 spectrometer filter time constant 16–17 frequencies 4. 50 phosphonylmethyl radical. 37–8 1. 13–16 operation parameters 7–11 power 12–13 sensitivity 4–5 signal gain 16 structure 7 sweep time 14 spin Hamiltonian 3 see also specific spectral types spin labels.1. 162–3 pyrazine radical anion spectrum 24–5 quadrupole coupling 145–51 perturbation theory 146–9 quantization axes 92–3 [(RCCR')Co2(CO)6]– frozen soln. 12.3.
Tesla (T) vs Gauss (G) 3. K. 21 vanadium (IV) 44–51 in acidic aqueous solution 44 in basic aqueous solution 48 VO(acac)2 106–7 nitrogen base exchange rate 106 VO(mquin)2 dilute single crystal spectrum 55. 133–5 zerofield splitting. 56 waveguides 3–5. 1 Zeeman effect electron 3. triplet state 120
units. 45. m. E. spectrum 81–6 Zavoisky. and pxylene radical anion spectra 35 [(oxylylene)Mn(dmpe)2] frozen soln. 10
.Subject Index
173 o. 133–5 nuclear 5–6 Zeeman Hamiltonian 3.