Electron Paramagnetic Resonance (EPR) Spectroscopy

tulipx.caltech.edu/groups/ch6/manual/Ch6-7EPR.doc

OVERVIEW

Electron paramagnetic resonance (EPR) or electron spin resonance (ESR) is a technique used to study the
spectroscopy of paramagnetic (species with nonzero electron spin s) atoms and molecules. The spectra
provide a measure of the magnetic environment of the unpaired spin(s). Among the most commonly
measured parameters are the g factor, the hyperfine splittings (interaction of the electron spin s with the
nuclear spins I), and the relaxation (dephasing) time T2. These physical observables tell us about the
structure of the molecule and the nature of the orbital occupied by the radical electrons.

I. INTRODUCTION
In most ions and molecules, electrons are paired so that the spin angular momenta s and orbital
angular momenta l cancel; the substance is then diamagnetic. If the substance contains unpaired electrons,
it is said to be paramagnetic. Some examples of paramagnetic species are:
a. Atoms for which a shell inside the valence shell is not filled, so that unpaired electrons are
present (by Hund's rules), e.g. transition elements. The first transition group is the iron group in which
electrons start filling the 3d shell after the 4s valence shell is filled. Another transition group is the rare
earth group in which the 4f shell is unfilled.
b. Molecules containing unpaired electrons. Examples are O2, NO, NO2, peroxylaminedisulfonate

SO3 —

O=N NO2

SO3 — N N NO2

NO2

and dipenylpicrylhydrazyl (DPPH)

1
Other examples are molecules in which bonds have been broken by radiation damage, or by one-electron
oxidation or reduction.

We can detect and study such paramagnetic atoms and molecules by the technique of Electron
paramagnetic resonance (EPR) or electron spin resonance (ESR) spectroscopy. This method is used
extensively throughout chemistry, in physical chemistry as a tool for understanding physical properties of
radicals, in organic chemistry to study reactive radical intermediates, in inorganic chemistry to study
transition metal complexes, and in biochemistry to study radicals or metal centers in large biological
molecules.

This technique is analogous to NMR, in that the sample is placed in a large external magnetic field, causing
Zeeman splitting of the spin orientations. Low energy radiation (microwaves, in this case) excites
transitions among different magnetic sub-levels. The effective g-factor (analogous to chemical shift) tells
us about the average magnetic environment of the unpaired electron or electrons. The EPR transition
sometimes exhibits hyperfine splittings, arising from interactions with nearby nuclear spins I. Both of
these effects tell us something about the molecular orbital in which the unpaired electron resides. Finally,
the linewidths can provide information on the relaxation of the spin s by the environment.

Figure 1. Schematic of EPR experiment Detector

E
Basic EPR Apparatus
MW
In the most conventional form (which we will use here), sometimes called cw-EPR (for continuous wave
oscillator
EPR), the sample is located in the poles of a high field magnet. It is irradiated by microwave radiation of
fixed frequency, enhanced by having the sample at the center of a resonant, hi-Q microwave cavity. The
hv
magnetic field is slowly scanned, and absorption occurs when the microwave MW photons are resonant with the
Cavity
split Zeeman levels. (fixed)
II. THEORY
H0

Single Electron in a Magnetic Field

An electron has associated with it an intrinsic spin and hence a spin angular momentum s, with
quantum number s = ½. In an external field, say in the z direction, the projection of the spin angular

Magnetic Field H0 2 Sample

momentum in the direction of the field is quantized and has magnitude msħ where ms = ± 1 2 , the
eigenvalue of the z component of s, the operator sz . When a paramagnetic sample is in zero magnetic field
in an isotropic sample (e.g. a liquid), the different orientations of the spins have equal energy.

Consider now a paramagnetic molecule in a time-independent, uniform magnetic field H0. The
Hamiltonian for the energy of interaction of the electronic magnetic dipole moment vector µ with the
external field is:

Ĥ = −μ× H 0 (1)

This part of the complete Hamiltonian is responsible for the magnetic field splittings of atomic energy
levels discovered by Zeeman. In the iron group transition elements, and in practically all polyatomic free
radicals, the orbital contribution to the magnetic moment is small. The orbital momentum is said to be
“quenched”. µ can then be written in the form of:

μ = − βe g ⋅ s (2a)

in which the spectroscopic splitting factor g is a symmetric second-rank tensor, s is the effective spin
vector, and β e is the Bohr magneton

β e = eh 4 me c
By definition, the magnitude of the magnetic dipole moment is

µ ≡ s, ms = s µz s, ms = s = ge βe s
If the ion or molecule is rapidly tumbling in solution, the anisotropy in g is averaged away and equation (2)
can be written as:

Hˆ = g e β e Ho ⋅ s (3)

in which ge is one third of the sum of the principle values of the 2nd rank tensor g. In general, ge is
approximately equal to 2.0023, the free electron value. For a magnetic field applied in the z direction, (3)
may be written as:

Hˆ = g e β e H 0 ⋅ sz (4)

From the Hamiltonian (4), the two stationary-state energy levels for spin s = 1/2 are given by:

E = ± 1 2 ge βe H 0 (5)

Magnetic dipole transitions between these two levels may be produced by the application of an oscillating
magnetic field perpendicular to H0 at a frequency ν given by:

ge β e H 0
ν= (6)
h
For fields of 3000 Gauss, this frequency is in the X-band range of the microwave region of the spectrum
(about 9.5 GHz).

3
Population Changes upon Excitation by the Radiation
In the presence of only the static field H0, the electrons are distributed between the two spin states
(α and β ) with a Boltzmann distribution

N α = N β exp ( g e β e H 0 kBT ) (8)

where kB is the Boltzmann constant. Thus, there is an initial population difference between the two states,

∆N = N α − N β (9)

At 300 K and typical field strengths, the population difference is quite small (see Prelab question).

Excitation by the oscillating field induces transitions between the energy levels. If we denote the
transition probability as p and the rate of transition as dp/dt, then it can be shown that this population
difference and the rate of energy absorption by the sample dE/dt follow:

∆N = ∆N 0 exp ( −2( dp dt )t ) (10)

dE dp
= ∆N ∆E (11)
dt dt
Thus, according to Equation (10), the population difference exponentially decays until the states are evenly
populated. Equation (11) indicated that when ∆ N = 0 there will be no power absorption. Thus on this
basis, the absorption of power should exponentially decay to zero. In fact, this is not seen. There is a
further relaxation mechanism termed spin-lattice relaxation. This consists of nonradiative transitions
between states caused by interactions with the lattice. Thus, as long as the microwave power level is not
too strong, there will be fast relaxation of the population. Relaxation will be dealt with more fully in a later
discussion.

Hyperfine Interaction

Spin and Magnetic Moment of the Nucleus.

Nuclei have an intrinsic spin characterized by a nuclear spin quantum number I , e.g.

I = 0 C12, O16, S32

I=½ C13, F19, P31
I=1 H2, N14
I = 3/2 Cr53, Cl35, Cl37
I = 5/2 Mo95,97
I = 7/2 V51, Co59

This nuclear spin gives rise to a physicallly obervable magnetic moment, given by an expression analogous
to (2a),

μ N = + gN βN I

where gN is the nuclear g-factor, which is dependent on the nucleus, and β N is the nuclear Bohr magneton:

β N = eh 2 Mc

where M is the mass of the proton. For the hydrogen atom, gN = 5.586.

4
In a paramagnetic atom or molecule, if the orbit of the unpaired electron overlaps with or approaches a
nucleus which has a nuclear spin I > 0, then there will be a magnetic interaction between this nucleus and
the electron. The energy levels of the electron will be split by a small amount. This small splitting arising
out of electron-nuclear interactions is called the hyperfine splitting (hfs).

Hyperfine Interactions
In a paramagnetic species, the magnetic dipole moment of the electron µ will interact with the
nuclear magnetic moment µ N, if the nucleus has nonzero I. This gives rise to small splitting(s) of the EPR
resonance line, called the hyperfine splitting. These splittings give us information on the nuclei that
interact with the unpaired electron.

The interaction energy must go as µ ⋅ µ N , but through two possible mechanisms: dipole-dipole
coupling through space or via direct overlap.

a. The dipolar interaction results from the long range force between the two magnetic dipole
moments separated by a distance r. The Hamiltonian for the interaction of the electron spin with the ith
nucleus is:

H dd,i = ge β e g N ,i β N { ( Ii ⋅ S ) rie −3 − 3 ( Ii ⋅ rie ) ( S ⋅ rie ) rie −5 } (12)

This must be averaged over the orbital of the wavefunction of the unpaired electron,so that the operator rie (
separation between the unpaired electron and a specific nucleus i) becomes 〈rie〉 . This term is anisotropic:
It results in a term which is orientation-dependent. If the molecule is tumbling rapidly, or if the electron is
in a spherically symmetric orbital about that nucleus (i.e. an s orbital), then Hdd,i= 0.

b. The Fermi contact interaction arises when there is a finite probability that the electron is at the
specific nucleus. This results in an isotropic hyperfine splitting of the resonance and gives rise to the
“Fermi contact” terms:

H C = Σi Αi Ii ⋅ s (13)

Ii is the spin and Ai is a scalar hyperfine constant for the ith nucleus. For example, for the hydrogen atom in
its ground 1s state, the hyperfine splitting is given by
2
∆EH-atom = g e β e gN β N Ψ1s (0)
Note that the Fermi contact term is proportional to the magnitudes of the two dipole moments and the
probability density of the electron at the nucleus.

If the radical is rapidly tumbling, then the nuclear-electronic dipole interactions (Equation (12))
average to zero and we have only the Fermi contact term (13). However, if the nuclear-electronic dipole
interaction must be included, e.g. in a solid, then the interaction is given by

H c = Σ i I i ⋅A i ⋅s (14)

where Ai is a tensor.

Paramagnetic Species in an External Field

From equations (3) and (13) the complete Hamiltonian is:

H = ge βe H o ⋅s + Σ i Ai Ii ⋅s (15)

5
In a strong applied field in the z direction, to first order in the hyperfine interactions, the Hamiltonian is:

H = ge βe Hosz + Σ i Ai Iiz sz (16)

The energy levels as associated with (16) are given by:

E = g e β e H o ms + Σi Ai mIi ms

where ms is the electronic quantum number with ms = −s, −s+1, … , +s, and mIi is the magnetic quantum
number of the ith nucleus, with mIi = −Ii, −Ii+1, …, +Ii.

The transitions obey the selection rules

∆ m s = ±1
and
∆mIi = 0 .

In general, (2I+1) lines are observed from splittings due to one nucleus. Figure 2 shows an energy
level diagram for one electron interaction with a nucleus of spin 1/2.

Interpretation of the EPR Spectra of Known Species

In interpreting the first order isotropic hyperfine splittings of EPR spectra one should use the rules
outlined before for sets of equivalent nuclei to determine hyperfine splitting constants from the observed
spectra. Completely resolved spectra of a known radical are then easily analyzed with some initial
guesswork concerning the values of the hfs constants. However, difficulties arise due to several causes
such as: 1) accidental degeneracy of hyperfine splitting constants (value of one hfs constant a multiple of
another value); 2) overlapping of lines (and hence distortion of the intensities); and 3) effects of broadening
of the lines which will be discussed later. In some circumstances, spectra simulated by a computer are

6
quite useful in hyperfine structure analysis. A few important remarks that help the interpretation of EPR
isotropic hyperfine structure are the following:
i) The spectra must by symmetric about a center point; asymmetry may be caused by superposition
of two spectra, due to differences in g-factors. In our investigations, we will not consider any such
superposition of two spectra.
ii) A spectrum with no center line indicates an odd number of equivalent nuclei of spin 1/2 n (n
odd integer). The presence of a center line does not exclude the presence of an odd number of nuclei.
iii) The total sum of the hyperfine splitting constants (absolute values) must equal the separation in
gauss between outside lines (which may be very weak).
iv) Some of the interacting nuclei may have more than one isotope in natural abundance. For example,
Carbon has two isotopes C12 (I=0, 99%) and C13 (I=1/2, 1%) in natural abundance. In some cases of EPR
measurements of carbon-containing compounds C13 nuclear-electron interaction may show up as small
satellites. Most transition metal nuclei have different isotopes. Chromium, for example, has two isotopes
in natural abundance, Cr52 (I=0, 90%) and Cr53 (I=3/2, 10%). If there is electron-nuclear interaction in any
Chromium compound with S=1/2, there will be one very intense line at the center, flanked by a total of 4
lines as small satellites with equal spacings due to the electronic-nuclear (Cr53) interactions.
v) Due to the second-order effects, particularly in the EPR spectra of transition metal ions, there
will be inequalities in the spacings of successive lines and hyperfine components may differ significantly.
This is manifested by differences in amplitudes.

Relaxation Effects and Linewidths

The finite line widths observed are due to limitations imposed by the Heisenberg uncertainty principle:

∆v ∆ t = 1 (18)

The spin states in the sample have a finite lifetime ∆t which implies that their energies have spread h∆ν .
The average lifetime of a spin state is determined by two relaxation time T1 and T2. The spin-lattice
relaxation time T1 is a measure of the rate at which a spin reverses direction and exchanges a quantum of
energy with the lattice. The term lattice refers to all degrees of freedom outside the spin system (e.g.,
rotations, vibrations, etc.). The transverse relaxation time T2 is the time for processes which vary the
relative energies of the spin levels.
Transverse relaxation occurs via some time-dependent interactions between the spins and
their environment. (A static interaction adds to the Hamiltonian and thus changes the position and
intensity of the line, but not its width.)
One such interaction arises from the dipole-dipole interaction between two paramagnetic
molecules. As the concentration of paramagnetic molecules in solution increases, the interactions between
the magnetic dipoles increases and T2 decreases. This is called dipole broadening of the absorption line.
A compensating effect is exchange narrowing which also occurs at high concentrations. This is a
result of hJs1s2 forces. During collisions, the electronic wavefunctions of paramagnetic molecules may
overlap, possibly resulting in an exchange of spins. Rapid spin exchange tends to average out dipole-dipole
interactions narrowing the line width.
Anisotropies in the g factor and hyperfine constant A, due to spin-orbit coupling and nuclear-
electronic dipole interactions, cause changes in line width. These effects were neglected in deriving the
isotropic Hamiltonian (13). As the molecule tumbles in solution, the Hamiltonian will be time dependent
since the z components of g and A will vary. This is treated by separating the Hamiltonian into a time-
independent (19) and a time-dependent (20) equation.

H o = geβe Hosz + AI zsz (19)

V(t ) = β Ho ⋅g' (t )⋅S + S⋅ T' (t) ⋅I (20)

A and T' are respectively the isotropic and anisotropic portions of the hyperfine interaction. The
perturbation V(t) is expanded as:

7
 V(t ) = fx s− x + fys− y + fzs− z (21)

where fx, fy and fz are time dependent:

fx(t) = β Hog'zx + m1Tzx

fy(t) = β Hog'zy + m1Tzy (22)
fx(t) = β Hog'zz + m1Tzz

The relaxation rate depends on the mean squared values of fx, fy, and fz. One such contribution is:
2
|fz(t)|2 = β Η ο (g'zz)2 + 2β Η ο g'zz T'zzm1 + (T'zz)2 m1 (23)

The combined factors yield (24) for

(T2 )−1 = A + BmI + CmI 2 (24)

Line Shapes
There are two fundamental line shapes found in electron spin resonance (or indeed most single
line) spectra: Lorentzian and Gaussian.

Lorentzian lineshapes are usually observed in solution spectra although if the line width is
determined by inhomogeneous broadening due to unresolved hyperfine components, the line shapes will be
Gaussian. For solid concentrated free radicals where the electron exchange with neighbors narrows the
resonance lines, Lorentzian lines are observed. For single crystals containing paramagnetic centers, the line
shapes will vary from roughly Gaussian at high concentration to approximately Lorentzian at low
concentrations.
The more common Lorentzian line is described by:
−1 −1
g(ω ) = T2 π [1+ T22(ω −ω ο )2] (25)

The distribution g is usually given in terms of ω ; g(ω ) = g (υ) 2π . ω o is the center resonance
frequency.
Gaussian lines are described by:

g(ω ) = T2(2π )-½ exp[-0.5 T22(ω − ω ο ) ] (26)

Figure 5 illustrates the differences between the two line shapes and also and example of the Lorentzian
derivative shape.

8
For the derivative spectrum, it is useful to note:

Distance between peaks = 2(T2 3½)-1 (27)

9
II. THE EPR SPECTROMETER

You will be using a Bruker EMX Spectrometer.

Figure 2 Schematic of Bruker EPR Instrument. Right: Closeup of probe region.

Specifications include:

 Magnet System and Magnetic Field Controller:

Water cooled magnets with pole diameters from 6" to 18", air gaps from 62 mm to 160 mm and power
supplies from l kW to 40 kW are available. A high stability, digitally controlled Hall field controller allows
you to sweep the magnetic field over 13 kG (1.3 T) and set field values with a resolution of under 1 mG
(0.1 µT).

 Cavity:
The standard cavity on an EMX is the ER 4119HS high-sensitivity cavity . This X-Band resonator has an
unloaded Q of more than 12000. It features high modulation amplitudes of up to 50 G. All standard
temperature control units can be used.
The WEAK PITCH sensitivity now routinely achieved is: 1500 : 1

 Microwave Bridge:
A fully computer controlled, high sensitivity X-Band Gunn-Oscillator bridge with automatic tuning
capability, Q-factor and power measurement.

 Signal Channels:
The standard EMX Signal Channel can be operated at any modulation frequency between 6 kHz and 100
kHz. It comes with unsurpassed phase resolution and stability.
The new design incorporates automatic digital tuning of any cavity to any modulation frequency, thus
doing away with the classical "tuning-box" required with each cavity.

10
III. PRE-LAB QUESTIONS

1. Experimental considerations
Consider a free electron system (ge).
a. Calculate an empirical factor dv/dH giving the relation between frequency (in GHz) and magnetic field
(in Gauss).
b. For the microwave X-band frequency of 9.25 GHz, what should the magnitude of the magnetic field be?
c. What is the fractional population difference ∆ N/N between the two m Zeeman levels in this field, at 300
K?

2. Hydrogen atom hyperfine splitting

For the H atom in its ground 1s state, only the Fermi contact term leads to hyperfine splittings.
a. Compute the magnitude of the splitting. (hint – you need to know |Ψ 1s(0)|2, the density of the hydrogen
1s orbital at the proton center).

b. What is the splitting that you will observe (in Gauss)? How does this compare to your estimated Zeeman
splitting from (1)? What fractional resolution would you need to observe this splitting? Sketch the
expected spectrum.

3. Hyperfine Splittings
For the following atoms, give S and I, and sketch the expected spectrum.
He
N2
O2
Mn+2
Fe+2 (high spin)
Fe+2 (low spin)

Examples

1 An Organic Radical, C6H6-˙.

The benzene radical anion, C6H6-˙, is generated by reacting benzene with an alkali
metal. In the ESR spectrum it has a symmetrical seven line pattern, as shown in the
figure (right). This implies coupling to all six hydrogen nuclei (2nI+1 lines, n=6,
I=½), which in turn implies that the unpaired electron is delocalised over all six
carbon atoms of the molecule. Given that the LUMO of benzene is a π-orbital, this is
not unreasonable.

2 A Main group radical, (MeO)3PBH2˙ .

11
 The ESR spectrum of (MeO)3PBH2˙ is shown in
the figure (left), together with a coupling tree,
which helps to explain the various couplings
which occur. The largest coupling is to phosphorus
(31P, I= ½), which gives a doublet. This doublet is
further split by the two borohydride hydrogen
nuclei (I = ½) to give triplets of intensity 1:2:1.
Finally, we have to consider the splitting by the
boron nuclei. Boron has two isotopes, 10B (20%, I
= 3) and 11B (80%, I = 3/2). The major influence
on the spectrum is the 11B, which splits each line
into a 1:1:1:1 quartet. Note however, that the
intensities of the splitting by H (1:2:1) are carried
through, so the quartets deriving from the central
line of each triplet have twice the intensity of those
deriving from the outer lines of the triplets. It can
be seen that the experimental spectrum matches up
very well to the theoretical splitting diagram. Also
note that the very weak bands in the baseline arise
from species with 10B present.

3 Some Transition Metal Examples

The main use of ESR spectroscopy is in the study

of paramagnetic transition metal complexes. Of
course, in these compounds the use of NMR
spectroscopy is not possible, and so any technique
which will give information on the compound is
useful. In this respect, ESR fits the bill.

Example 1

12
The first example is the ESR spectrum of [V(O)(OH2)5]2- . 51V is 100% abundant 2+
and has a spin, I = 7/2. Vanadium(IV) has a 3d 1 electron configuration and hence O
one unpaired electron, making it ideal for ESR spectroscopy (NB systems with H2 O OH 2
more than one unpaired electron generally give rise to complex spectra which are V
very hard to interpret). H 2O OH 2
The unpaired electron couples only to 51V (16O has I =0), which would be OH 2
expected to give an eight line pattern. However, what is observed is two overlapping eight line patterns,
which arise from the fact that one of the axes in the complex is different to the other two, resulting in a
tetragonal spectrum, with a parallel and perpendicular component. These are labelled in the figure, though
not all of the lines are resolved due to overlaps in the central part of the spectrum. The parallel component
has a larger hyperfine coupling constant than the perpendicular component, and its g-value is at a slightly
higher field.

The next example (below) is the spectrum of [V(O){S2P(Et)(OMe)}2]. Again we see an eight line pattern
due to coupling to vanadium. In this case, however, each line is s plit into

a triplet by coupling to the two phosphorus nuclei (100%, I = ½). Although the phosphorus atom is not

Et

P OMe
S

S
O V
S
S
P
OMe
Et

bonded to the vanadium, the unpaired electron is able to couple (more weakly) to it, probably by some
delocalisation around the VSPS chelate ring. Because the electron is located primarily on the vanadium
centre, the hyperfine coupling constant to phosphorus is smaller than for the coupling to V.

The next example is the ESR spectrum of a solution of [Cu(acac)2]. While in the solid state this has an
aggregated structure, in solution it is probably a simple square planar complex.

O O
Cu
O 13 O
This assignment is supported by the tetragonal spectrum observed, which implies one unique axis, and two
axes the same. In this spectrum we only see coupling to copper, since the most common isotope of oxygen
(16O) has a nuclear spin of zero. Copper has two isotopes, 63Cu and 65Cu, both of which have spin I = 3/2.
The group of more intense peaks at the right of the spectrum are the g ⊥ component (the peak labelled X is a
spin-forbidden band). The less intense peaks at the left are the g| | component. The hyperfine coupling
constant is much larger for g| | than for g⊥ and consequently on the outer line (far left) the coupling due to
the two isotopes of copper is resolved – the splitting “tree” shows one isotope has a smaller splitting than
the other (since the magnetogyric ratio γ is different). The big difference in g-value and coupling constant
for the two components arise from the fact that the unpaired electron is probably in the dz2 orbital which
lies along (parallel to) the unique axis.
The final example concerns the ESR spectrum of a single crystal of a copper(II) complex of a Schiff’s base
ligand, shown on the right. The first spectrum is of the crystal orientation which gives the best peak
separation and hence clearest spectrum.
What we see is coupling of the unpaired electron to the copper nucleus (I = 3/2) to give a four line pattern
Cl

HO Cu
N
N
O OH

Cl

with a large hyperfine coupling constant (ACu on coupling tree on the diagram above). Each line is then
further split into a quintet by interaction with the two nitrogen nuclei ( 14N: I = 1, abundance >99%). There
are 2nI + 1 lines (n = 2, I = 1). The overall spectrum may be described as a quartet of quintets.

Again the splitting due to the metal is large because the electron is located in an orbital which is primarily
of metal character. The splitting due to nitrogen is much smaller because although the orbitals on nitrogen
mix to an extent with those on the metal, the electron spends most of its time near the metal nucleus.

The final
spectrum shows
the effect of
rotating the
crystal through
an angle of 100º.
The hyperfine
coupling to
copper decreases
very
significantly.
This is because

14
the orientation of the orbital containing the unpaired electron changes with respect to the magnetic field
direction hence altering the magnitude of the interaction. The result is overlap of the lines and a spectrum
which is qualitatively more difficult to interpret. However, data of this type, when combined with quantum
chemical calculations is a very powerful technique for understanding the electron distribution in molecules.

15