You are on page 1of 34

4.

Magnetic ordering
We have seen in previous chapters that atoms, molecules and other fundamental units of
matter can have a magnetic moment. Such units are often just called spins, because, in order to
have a dipolar moment, they must also have an angular momentum, or spin. Since spin is a
much compacter expression, we will from now on use the term spin to refer elemental dipole mo-
ments in matter. This can be an atom, a molecule, or valence electrons in metals, which are not
bonded to individual atoms.
The sum of all microscopic magnetic moments in a given volume defines its magnetization,
which can be understood as a net magnetic dipole moment per unit volume (see chapter 2). De-
pending on whether a substance contains uncompensated spins, and on how these spins interact
with each other, distinct magnetic properties will be observed. According to these properties, mat-
ter is divided into four main categories (Fig. 4.1):
1) Diamagnetic materials. Diamagnetic materials do not contain free spins, because all spin and
orbital moments within individual atoms or molecules are cancelling each other. An external
field, however, modifies the electron orbitals, inducing small magnetic moments in the oppo-
site direction. The resulting magnetization is antiparallel to the applied field. It is extremely
small and can only be measured with sensitive instruments.
2) Paramagnetic materials. Paramagnetic substances contain free spins which do not interact with
each other. In absence of external fields, they are completely randomized by thermal fluctua-
tions. An external field aligns the spins to a degree that depends on temperature. At room
temperature, the alignment is extremely small and the resulting magnetization is small, but
generally larger than that of diamagnetic materials. This magnetization disappears as soon as
the field is removed.
3) Ferromagnetic materials. Ferromagnetic matter also contains free spins; however, unlike the
case of paramagnetism, the spins interact strongly with each other in a way that makes them
become spontaneously aligned with each other, even in the absence of external fields. The
degree of alignment depends on temperature, and interactions between spins are lost above
the Curie temperature . At much lower temperatures, the alignment is almost perfect, and
the material is strongly magnetic. The resulting magnetization is called spontaneous magnetiza-
tion, , because it does not require any external influences to exist. Ferromagnetic mate-
rials, whose name comes form iron (lat. ferrum), have enormous technological and scientific
relevance. Iron, Cobalt, and Nickel are the most well known ferromagnetic elements.
C
T
s
M
4) Antiferromagnetic materials. Antiferromagnetic materials also contain strongly interacting free
spins. Unlike the case of ferromagnetism, the spins are arranged into two or more so-called
sublattices that are magnetized in exact antiparallel directions. The net magnetization in
absence of external fields is zero. An external field, however, alters this alignment, inducing a
weak magnetization. Spin coupling within and between the sublattices disappears above the
47
Nel temperature . Despite strong spin coupling below , antiferromagnetic materials do
not possess a spontaneous magnetization.
N
T
N
T
(a)
H
M
(b)
H
M
diamagnetism paramagnetism
(c) (d) ferromagnetism antiferromagnetism
M
s
M
s
= 0

Fig. 4.1: Four main types of magnetic ordering, as described in the text. (a) Diamagnetic materials do not contain
spins, but an applied field induces a precession of the electrons, which in turn generates magnetic dipole moments
(arrows) that point to the opposite direction. (b) Paramagnetic materials contain spins that are not coupled with each
other and whose direction is randomized by thermal fluctuations (arrows). The resulting magnetization is zero, unless
a field is applied, making the probability of orientations parallel to the field more likely. (c) Ferromagnetic materials
contain strongly coupled spins that generate a spontaneous magnetization
s
M without the need of an external field.
(d) Antiferromagnetic materials contain sublattices (two in the figure) of strongly coupled spins. The sublattice spins
each other out, and the resulting magnetization is zero if the antiparallel arrangement is not perturbed by external
fields.
Other types of magnetic materials are described by more complicated spin configurations,
which can be expressed as combinations of ferromagnetic and antiferromagnetic spin coupling.
These are (Fig. 4.2):
1) Ferrimagnetic materials. Ferrimagnetic substances possess two spin sublattices, like antiferro-
magnets; however, the two sublattices are unbalanced, because one is carrying a larger mag-
netic moment than the other. This results in a spontaneous magnetization that disappears at
the Curie temperature, like in ferromagnetic materials. Because of the existence of sublattices,
however, ferrimagnetic materials are often more complicated than ferromagnets, especially
48
with respect to the temperature dependence of various magnetic properties. The name ferri-
magnetism originates from ferrites, which are a special class of metallic oxides with spinel
crystal structure. Ferrites have important technological applications and are relevant as paleo-
magnetic carriers (magnetite, , is a type of ferrite).
3 4
Fe O
2) Canted antiferromagnetic materials. Canted antiferromagnetism refers to antiferromagnetic
materials where the spins in different sublattices are not perfectly antiparallel, but rather at a
generally small angle called canting angle. Spin canting generates a net magnetization that is
perpendicular to the mean direction of the spins. Canted antiferromagnets have a relatively
small spontaneous magnetization that disappears at the Nel temperature. The second most
important paleomagnetic carrier, hematite ( ), is a canted antiferromagnet.
2 3
-Fe O
3) Defect antiferromagnetic materials. Defect antiferromagnetism is another case where the mag-
netizations of the two sublattices are not perfectly compensated. This is due to the preferential
occurrence of crystal defects (vacancies, cation substitutions) within one sublattice, whose
magnetization is consequently smaller. Defect antiferromagnets have a weak spontaneous
magnetization that depends on the defect concentration. Goethite (-FeOOH) is a natural
antiferromagnetic iron hydroxide whose weak magnetization is produced by defects.
4) Spin glasses. Spin glasses are materials that contain ferro- and antiferromagnetically coupled
spins in a frustrated state. Spin frustration denotes cases where the coupling between neighbor
spins is contradictory. This means that different orientations would be needed for a given
spin to satisfy all couplings with his neighbors, whereby an optimal coupling configuration
does not exist. Frustration occurs in particular lattices, such as the example in Fig. 4.2d, but
also in cases where the spins are arranged randomly, as atoms in glassy structures. Materials
that contain frustrated spins can be strongly magnetic, even without an external field being
applied; however, their magnetization is metastable and subjected to ageing effects. The Curie
temperature is replaced by a so-called transition, or critical temperature above which spin cou-
pling is broken by the randomizing effect of thermal energy. Spin glasses have recently gained
broad interest and found some technological applications. Spin frustration is relevant for
understanding certain magnetic properties of very fine and/or partially oxidized magnetic par-
ticles in rocks.
49
(a) (b) ferrimagnetism canted antiferromagnetism
(c) (d) defect antiferromagnetism spin frustration
M
s
= ?
M
s
M
s
M
s
?

Fig. 4.2: Additional types of magnetic ordering, as derived from Fig. 4.1. (a) The net spontaneous magnetization of
ferrimagnetic materials originates from two antiferromagnetically coupled sublattices with different magnetizations.
(b) The spins in the sublattices of canted antiferromagnetic materials are not perfectly antiparallel, and thus generate
a small spontaneous magnetization perpendicular to them. (c) One sublattice in defect antiferromagnets contains a
larger number of defects (vacancies, cation substitutions) with respect to the other, leading to an imperfect compen-
sation of the spins. The resulting small spontaneous magnetization depends on defect concentration. (d) Some
lattices can produce a so-called frustrated state, as shown here on the example of tree antiferromagnetically coupled
spins. The third spin should be oriented downward to accommodate the coupling with the spin to the left, and
upward to accommodate the coupling with the third spin of the dashed triangle. The resulting spin configuration is
metastable and changes spontaneously with time.
In the following chapters we will learn more on each of these classes of magnetic materials,
starting from the simplest cases of diamagnetism and paramagnetism. Tab. 4.1-2 provide a
summary of the magnetic properties of some relevant materials.
50
Tab. 4.1: Magnetic properties of some dia-, para- and ferromagnetic materials. k is the (mass normalized) magnetic
susceptibility, the (volume normalized) spontaneous magnetization, and T is the Curie temperature.
s
M
C
Material Occurrence/
application
Magnetic
property
ordering
temperature
Diamagnetic
C (graphite) levitation demonstration

Bi (Bismuth) levitation demonstration
Cu (Copper) electric conductor
H
2
O (water) everywhere
rocks, sediments CaCO
3
(calcite)
rocks, sediments SiO
2
(quartz)
sample holders Plastic (acrylic glass)
Paramagnetic
electric conductor Al
calibration, NMR Dy
2
O
3
rocks Biotite
rocks Chlorite
rocks Muscovite
sediments Montmorillonite
sediments Ferruginous smectite
rocks Fayalite (Fe
2
SiO
4
)
rocks, sediments Pyrite (FeS
2
)
Ferromagnetic
electromagnets Iron (-Fe)
soft cores Nickel (Ni)
electromagnets Cobalt (Co)
Fe
65
Co
35
Mumetal (76%Ni+Fe)
Dysprosium (Dy)
Gadolinium (Gd)
AlNiCo
Sm
2
Co
17
Nd
2
Fe
14
B
Awaruite (Ni
3
Fe)
Wairauite (CoFe)
electromagnets
magnetic shields
highest M
s
T
C
at 19C
hard magnets
hard magnets
hard magnets
meteorites
meteorites
k , m
3
/kg, 20C


7
2.94 10 ( c)


8
2.64 10 ( c)


8
1.7 10


9
1.1 10


9
9.05 10


9
4.8 10


9
6.2 10


9
7.6 10
k , m
3
/kg, 20C

+
7
2.07 10

+
6
3.1 10

+
7
4.0 10

+
8
8.5 10

+
8
8.9 10

+
8
(2 4) 10

+
7
3.1 10

+
6
1.26 10

+
7
3.0 10
s
M , kA/m, 0K
1740
510
1446
1950
636
2920
2060
250
1025
1280
960
1955



















C
T , K
1043
627
1403
1223
673
88
292
1133
1203
588
893
1256
51
Tab. 4.2: Magnetic properties of some ferri- and antiferromagnetic materials. is the (mass normalized) spon-
taneous magnetization, is the Curie temperature, is the Nel temperature, is a critical ordering tempe-
rature.
s
M
C
T
N
T
crit
T
Material Occurrence/
application
Magnetic property ordering
temperature
Ferrimagnetic
Magnetite (Fe
3
O
4
) rocks, sediments
Maghemite (-Fe
2
O
3
) sediments, tapes
Greigite (Fe
3
S
4
) sediments
Chromium Oxide (CrO
2
) tapes
Jacobsite (MnFe
2
O
4
) ores
Trevorite (NiFe
2
O
4
) ores
ores

rocks
rocks
rocks, sediments
sediments
sediments
rocks, sediments
rocks, sediments
sediments
synthetic
sediments

electronics
s
M , Am
2
/kg, 20C
92
73
59
105
77
51
21
s
M , Am
2
/kg, 0 K
0
0
0.48 (0.1-canted)
0.47 (defect)
0.50 (defect)
20
0
0.76 (defect)
12 (6-canted)
0.93 (defect)
s
M , kA/m, 20C
796
C
T , K
853
873
623
386
673
858
713
N
T , K
40
120
948
393
52
593
578
38
18.1
32
crit
T , K
758
Mg-ferrite (MgFe
2
O
4
)
Antiferromagnetic
Ilmenite (FeTiO
3
)
Ulvospinel (Fe
2
TiO
2
)
Hematite (-Fe
2
O
3
)
Goethite (-FeOOH)
Lepidocrocite (-FeOOH)
Pyrrhotite (monoclinic, Fe
7
S
8
)
Pyrrhotite (hexagonal, Fe
9
S
10
)
Siderite (FeCO
3
)
CoCO
3
Rhodocrosite (MnCO
3
)
Spin glass
FeCo (amorph)

52
5. Diamagnetism
Diamagnetism is the manifestation of the tendency of electrical charges to partially shield the
interior of a body from external magnetic fields. The physical reason is the same as for Lenzs law:
when the magnetic flux through an electrical circuit is changed over time, the electric current in
the circuit adjusts to compensate the flux change. Since all materials contain electric charges,
diamagnetism occurs in all solid, liquid or gaseous substances. However, this phenomenon is very
weak and is usually masked by stronger magnetic properties such as paramagnetism or ferromag-
netism. Therefore, diamagnetism is observable only in materials which do not contain atoms or
molecules with a net magnetic moment: these materials are called diamagnetic, even though dia-
magnetism is present in other substances as well.
In the following, we discuss the simplest physical model of diamagnetism, called Langevin
diamagnetism. This model is based on a classical treatment of electron orbitals in isolated atoms.
As explained in chapter 1, all atoms have spin and orbital angular momentums which, in case of a
diamagnetic substance, are perfectly compensated. Let us consider the orbital angular momentum
of one electron and describe it from the point of view of classical mechanics (Fig 5.1). Accor-
ding to eq. (1.12), the magnetic moment associated to is:
L
L

B
e
2
Le L
m
= = . (5.1)
This equation does not need to be corrected for quantum mechanical effects, because the g-factor
of the orbital magnetic moment is unity.

r
B
L

L
e

R
T
e

L
dL/dt





Fig. 5.1: Classical model for the Larmor precession of an
orbiting electron in a magnetic field B . is the angular
momentum of the electron mass rotating on an orbit of
radius with angular velocity . A torque exerted by
on the orbital magnetic moment produces a pre-
cession with angular velocity . Projection of this preces-
sion onto a plane perpendicular to B (shaded) is described
by a charge moving on along a circumference of radius
with angular velocity . This is equivalent to a cir-
cular current loop that adds magnetic dipole moment
opposed to .
L
R T
B

L
e
r
L

B
53
A magnetic field is now applied at a random direction. As seen in chapter 3, a
torque (eq. 3.2) tries to align the magnetic moment with the field. As a result of
Newtons law for the angular momentum ( ), there will be a precession of the angular
momentum, which means that L rotates around B with an angular velocity . This situation
is similar to the case of a gyroscope subjected to gravity: as soon as the gyros rotation axis is not
perfectly vertical, the combined action of the Earths gravity and the corresponding reaction force
of the pivot on which the system is resting generates a torque that brings the gyroscopes rotation
axis into a circular motion.
0
H = B
= T B
T

|
d dt / = L
L

The classical model of the electron orbital consists of a point mass orbiting with angular
velocity on a circle of radius (Fig. 5.1). The torque exerted by B is , and the
orbital precession can be written as , with being the angle between L and
. Equating the expressions for T and we obtain the angular velocity of precession:
R sin T B =
L
d d sin t L | / | = L
B d dt | / L
L
e
2
B eB
L m

= = , (5.2)
which is independent of and is called Larmor precession. A similar precession phenomenon
occurs for hydrogen nuclei (protons), which also possess a magnetic moment, and is exploited in
nuclear magnetic resonance (NMR) and for field intensity measurements with the proton magneto-
meter (see Box 5.1).
If we project the precessing electron orbital onto a plane perpendicular to B, we see that the
electrons angular velocity is sped up by . This is equivalent to an additional circular current
produced by a charge making turns per second. The surface enclo-
sed by the loop (a circle of radius r ) is , and the resulting additional magnetic moment is
. The negative sign accounts for the fact that is antiparallel to the magnetic
field. Using eq. (5.2) for we obtain:
L

L
(2 ) I e = / e
L
(2 ) /
2
r
2
L
2 e r = /
L

2 2
e
4
e r
B
m
= . (5.3)
Our result is valid for one electron, but can be extended to all electrons of an atom with atomic
number . Each electron will produce a magnetic moment according to (5.3) and the sum is
given by
Z
2 2
atom
e
4
Ze r
B
m


= , (5.4)
where is the average squared radius of the electron orbitals projected on a plane perpen-
dicular to the applied field (shaded area in Fig. 5.1). If the orbitals are spherically symmetric, their
radius is given by , with all three components being equal on average:
. Then, . If there are atoms per unit volume
2
r
2 2 2
R x y z = + +
2
2 2 2
x y z = =
2 2 2 2
r x y R = + =2 /3 N
54
or per unit mass, will be the induced magnetization. This magnetization is
proportional to and can be expressed as : the negative proportionality constant
dia
M NZ =
B
dia dia
M = H
2
2
dia 0
e
6
e
NZ R
m
= (5.5)
is the diamagnetic susceptibility (Fig. 5.2). Its S.I. unit is a pure number if is the number of
atoms per unit volume, or if is the number of atoms per unit mass.
N
3
m kg / N

H
M

dia
M(H)







Fig. 5.2: The magnetization M of a diamag-
netic material in a field . The slope of H
M H ( ) is the diamagnetic susceptibility .
dia


The diamagnetic susceptibility predicted by the Langevin model is identical to the result ob-
tained from quantum mechanical calculations, and thus represents a correct description of the
magnetic properties of isolated atoms with no net magnetic moment [Morrish, 1965]. It is im-
portant to observe that, unlike other magnetic phenomena, mass-normalized Langevin diamagne-
tism is temperature independent as long as ionization does not occur. Typical mass-normalized
diamagnetic susceptibilities are in the order of (Tab. 4.1). This is at least one order
of magnitude less than the susceptibility of paramagnetic materials and several orders of mag-
nitude smaller than the susceptibility of ferro- and ferrimagnetic materials. Therefore, diamagne-
tism is only observed in very pure diamagnetic materials. Minute ferromagnetic contaminations
are sufficient to mask diamagnetism (see exercise 5.2).
9 3
10 m kg

/
Exercise 5.1: Calculate the diamagnetic susceptibility of helium (atomic radius: 50 pm, molar mass: ),
and compare it with the mass-normalized measured value of .
4 g/mol
A
m =
9 3
5.9 10 m kg

/
3 3
Exercise 5.2: Calculate the concentration of magnetite impurities sufficient to make the susceptibility of a limestone
(calcite CaCO
3
, see Tab. 4.1) become positive. Assume that the mass normalized susceptibility of magnetite is
. 2 10 m kg /
55
A notable exception to the typical range of diamagnetic susceptibilities is represented by gra-
phite. Graphite is made of stacked planes of hexagonal carbon lattices (Fig. 5.3). Within these
planes, valence electrons are not bound to individual atoms and can move freely in what is called
a two dimensional free-electron gas. These electrons are responsible for the strong anisotropy of
graphites physical properties. Electrical and thermal conductivities, for example, are much higher
in any direction parallel to the planes than perpendicular to them.
a = 0.246 nm
c = 0.671 nm

||
c
a






Fig. 5.3: The atomic structure of graphite,
shown for two consecutive planes of the hexa-
gonal lattice.

Graphites magnetic susceptibility is highly anisotropic as well. Its value along any direction
parallel to the planes conforms the free-atom Langevin model. On the other hand, the perpendi-
cular susceptibility is about one order of magnitude larger, and is explained by a quantum
mechanical model of a two-dimensional free electron gas. Unlike Langevin diamagnetism,
depends on the temperature and reaches its maximum below 100 K [Ganguli and Krishnan,
1941]. The large diamagnetic susceptibility of graphite, combined with its relatively low density
( ) is exploited in a simple demonstration of diamagnetism with a graphite plate-
let and strong NdFeB permanent magnets (Fig. 5.4). This demonstration shows the so-called
diamagnetic levitation, which is an example of magnetic force experienced by a dipole in an inho-
mogeneous field (chapter 3). A piece of graphite with volume V , when immersed in a field, will
acquire a magnetic moment that is antiparallel to the field: with . Field
inhomogeneities produce a net force (eq. 3.4) on the induced magnetic moment of
graphite. If the field gradient is one-dimensional, for instance along the z-axis, the force is given
by:

3
2100 kg m = /
dia
H =
dia
0 <
m
( F = ) B
2
m 0 dia 0 dia
( ) ( ) 2
H
F B H H
x x


= = =
x

(5.6)
Because is negative, is pointing toward the direction of maximum decrease of the mag-
netic field. This means that diamagnetic materials are pushed away from strong field sources. For
example, they are repelled by a permanent magnet, since the strongest field occurs right on the
magnets surface. Indeed, the repulsion force exerted by a strong magnet on a graphite platelet is
large enough to exceed the platelets own weight within few mm from the magnets surface (Fig.
dia

m
F
56
5.4). This demonstration of magnetic levitation has become popular, and commercial levitation
kits composed of NdFeB permanent magnets and a small piece of pyrolytic graphite are available
(pyrolytic graphite is ultrapure, crystallographically oriented graphite obtained by heating hydro-
carbon films just below the decomposition temperature).


Fig. 5.4: A pyrolytic graphite platelet levitating
over four cubic NdFeB permanent magnets. The
magnet polarity sequence, seen from the top, is
N-S for the first row and S-N for the second row.
This arrangement creates a strong field gradient,
which also stabilizes the horizontal position of
the platelet right above the centre of the four
magnets. A single magnet, although producing a
field of the same strength, would not sustain the
platelet, because the gradient is not large enough.

Typical diamagnetic substances occurring in rocks and sediments are calcite ( ), quartz
( ), and water ( ). Calcite and quartz are sometimes pure enough to have negative su-
sceptibilities. More commonly however, minute ferromagnetic inclusions are sufficient to mask
this property. Most organic substances and plastics are also diamagnetic. Biological tissues, being
made of organic substances and water, are diamagnetic. A spectacular demonstration of biolo-
gical diamagnetism has been obtained in the high field magnetic laboratory of the Radboud Uni-
versity Nijmengen (Netherlands), by levitating a living frog over a 16 T bitter magnet (see exer-
cise 5.3 and Fig. 5.5). Diamagnetic levitation creates a zero-gravity environment, since gravity is
exactly compensated by magnetic forces on each atom. This is the subject of serious scientific
investigation beyond science-fiction [Simon and Geim, 2000].
3
CaCO
2
SiO
2
H O




Fig. 5.5: Diamagnetic levitation of a living frog
over a bitter magnet at the high field magnetic
laboratory of the Radboud University Nijmengen
(Netherlands). The video is available on the inter-
net. The field produced by the magnet was 16 T.
The frog survived the experiment with no damage
but some confusion. In order to levitate a human
being, a 40 T field over a large volume is needed,
a condition still beyond the present (2011) techno-
logy.
57
Exercise 5.3: The magnetic field produced by a circular current loop of radius at a height over the center of the
loop is given by:
R z
2 2
2
z
I R
H
R z
=
+

Calculate the minimum current that is necessary to make a frog ( , density ) levitate over
a loop with . How strong is the field seen by the frog? Compare your result with the 16 T needed in the
real experiment.
5
10

3
10 kg m /
3
15 cm R =

58
Box 5.1. The proton precession magnetometer
The magnetic moment of a proton ( ) is expressed in units of the so called nuclear magneton
. By analogy with Bohrs magneton (chapter 1), is defined as the magnetic moment of a
particle in classical mechanics having the same properties of a proton; i.e., electric charge e , spin
, and mass :
p
+
N

1 2 s = /
p
m

27 2
N
p
5.051 10 Am
2
e
m


= =
The measured proton magnetic moment is with the g-factor accoun-
ting for quantum mechanical effects. It is 320 times smaller than , which explains why the
magnetic properties of matter are controlled by the electron shell, and not by atomic nuclei. The
precession of protons that are forced to align in an external field can be calculated in analogy to
the Larmor precession (eq. 5.2), obtaining:
p p
g =
N p
5.5857 g =
B

N
p
2
g ge
B B
m

= =
The measuring principle of the proton precession magnetometer is based on the proportionality
between and , which is expressed by the constant . This con-
stant is the precession frequency in a 1 T field. The magnetometer construction is very simple: it
consists of a tank containing a hydrogen-rich substance (e.g. water, kerosene), whose protons are
responsible for the signal, wrapped with a coil. The coil is used to produce a strong magnetic
field pulse that aligns all protons in the tank. After the pulse, protons will relax back to the
direction of the ambient field (which needs to point at a different direction), through precession
with angular velocity . Because all protons are subjected to the same field, precession is equi-
valent to a rotating magnetization, which, by Faradays law (eq. 2.8), induces a sinusoidal voltage
with frequency at the coil terminals. The frequency of the induced voltage is pro-
portional to the intensity of the ambient field. Typical values of the Earths field ( )
produce a 2 kHz signal. Notice that the magnetometer measures the field intensity, providing
no information about the field direction. The sensitivity of the instrument is limited by the
lowest frequency that can be measured before the precession effect is extinguished. It corresponds
to 0.04 Hz, or 1 nT.
B ) 42.5775 MHz T B /(2 = /

2 f = /( )
50 T B =



Construction principle of a proton precession
magnetometer. Protons are aligned with the
field produced by a coil wrapped around the
tank (top). A switch (middle) disconnects the
coil from the battery and connects it to a sig-
nal amplifier. Because the current in the coil
is interrupted, protons are now able to precess
around the ambient field (bottom). The
precession induces an AC voltage that is am-
plified and send to a circuit that measures the
signal frequency, which is proportional to .
B
B







59
Literature:
Ganguli, N., and K.S. Krishnan (1941). The magnetic and other properties of the free electrons in graphite,
Proceedings of the Royal Society of London, A177, 168182.
Morrish, A.P. (1965). The physical principles of magnetism, Wiley series on the science and technology of materials,
Wiley, New York.
Simon, M.D., and A.K. Geim (2000). Diamagnetic levitation: flying frogs and floating magnets (invited), Journal of
Applied Physics, 87, 62006204.

60
6. Paramagnetism
Paramagnetism occurs in all materials containing magnetic dipoles that do not interact with
each other [Kittel, 2005]. This is the case for:
1) Atoms, molecules, and lattice defects possessing an odd number of electrons, as here the total
spin of the system cannot be zero. Some examples are free sodium atoms, nitric oxide gas
(NO, as a molecule), organic free radicals (i.e. molecules with unpaired electrons), ionic sub-
stances such as pyrite ( , as a crystal), and color centers (i.e. electron-filled vacancies in a
crystal lattice).
2
FeS
2) Free atoms and ions with a partially filled inner electron shell, such as transition and rare
earth elements (e.g. , , , ). Paramagnetism is preserved in most cases,
but not always, when ions are incorporated into solids. For example, iron chloride ( )
and manganese fluoride ( ) are paramagnetic, but magnetite, which also contains
ions, is ferrimagnetic.
2
Fe
+ 2
Mn
+ 3
Gd
+ 4
U
+
2
FeCl
3
MnF
2
Fe
+
3) A few compounds with an even number of electrons, such as molecular oxygen ( ).
2
O
The quantum theory of paramagnetism is simple enough to be explained here, and gives us the
occasion to introduce fundamental concepts which are used later to explain ferromagnetism. The
theory describes an isolated spin (i.e. non-interacting with other spins) with angular momentum
J , which is immersed in a homogeneous magnetic field B. As we have learned in chapter 1
(eq. 1.8), any arbitrary component of the magnetic moment associated to the spin, for example the
component parallel to , is quantized, and can be written as B
B J
m g = , (6.1)
where
J
m is the azimuthal quantum number. There are 2 1 J + possible values of
J
m , given by
, where , 1, , 1,
J
m J J J = + J J is the principal quantum number. We recall that J is a
half-integer (i.e. an integer or an integer plus 1/2). The g-factor is a correction constant that
accounts for the difference between the classical and the quantum mechanical theory of the an-
gular momentum. It is specified by Lands equation (eq. 1.16) and depends on how the atomic
angular momentum is distributed between electron spins ( ) and orbitals ( ). It is also
influenced by the coupling between electron spins and orbital momentums. The important
concept to retain from eq. (6.1) is that any measurable component of the magnetic moment has
only 2
2 g = 1 g =
1 J + discrete values, like the angular momentum.
We have seen in chapter 3 how magnetic moments experience a torque that tends to align
them with (eq. 3.2). The alignment can be expressed in terms of the potential energy
. Recalling that the scalar product is the orthogonal projection of onto B,
we can write , where is the component of parallel to the field. Like any other
component, is specified by equation (6.1), and we obtain following expression for the poten-
tial energy of a free spin in a magnetic field:
B
E = B B
E = B

61
B J
E m g = B . (6.2)
E has 2 1 J + discrete values dictated by
J
m , which represents the allowed energy levels of the
spin. The energy levels depend on the applied field and collapse to the same value in the
zero field. It is said that the energy level of one atom with non-zero spin is split by a magnetic
field (Fig. 6.1). This important phenomenon is called Zeeman splitting.
0 E =
(a)
B

B
E
m
J
= +3/2
m
J
= +1/2
m
J
= 3/2
m
J
= 1/2
0
(b)
0


Fig. 6.1: (a) The magnetic moment of an atom with spin number in a magnetic field . The com-
ponent along the direction of B can only have discrete values given by the azimuthal spin number
(dashed lines). (b) The potential energy of the magnetic moment (energy level) is a function of the magnetic
field . The energy level is split in levels determined by .
3 2 J = / B
2 1 J + =4
4
J
= k
J
m E
B 2 1 J + =
J
m
The lowest energy level is given by , which also represents the maximum possible
alignment with the magnetic field. In the absence of external perturbations, the lowest energy le-
vel is occupied, such that the magnetic moment is maximally aligned with the field. The weakest
field would be sufficient to align all spins, giving raise to a large magnetization that is never obser-
ved in paramagnetic minerals at room temperature. What prevents full alignment? At absolute
temperatures , atoms are subjected to thermal agitation, which can be understood as a con-
tinuous series of random perturbations acting on every degree of freedom, such as for example the
three coordinates in space, or as of interest in our case the spin vector. The typical energy asso-
ciated with the perturbation of one degree of freedom is , where is the Boltzmann con-
stant (Tab. 0.2), and T the temperature in K. The nature of this energy depends on the kind of
perturbations we are looking at: for example, perturbation of an atoms position in a crystal lattice
requires some kinetic energy. Perturbations of , on the other hand, are accompanied by a chan-
ge of the dipole potential energy in a field. A simple calculation using eq. (6.2) gives us an idea
of how large is thermal energy compared to the difference between consecutive
energy levels. Using , we obtain in a reasonably strong field of 1 T (the
maximum field of most electromagnets is 1-2 T). On the other hand, the thermal energy at room
temperature ( ) is , a value that is much larger than and sufficient to
induce multiple transitions between different energy levels. If , the rate at which such
transitions occur is extremely high: typically of the order of magnitude of events/s. This rate
can be considered as the intrinsic frequency of thermal perturbations.
J
m =+
0 T >
B
k T
B
k

E
B
E g B =
2 g
B
1.3 E
300 K T =
B
300k E
B
k T E
10
10
62
Because transitions between energy levels are made possible by thermal perturbations, each
energy level is populated with a probability that is determined by the so-called Boltzmann distri-
bution. Accordingly, the occupation probability for the i-th energy level is given by:
i
E
1
i
j
i
N
j
e
P
e

=
=

(6.3)
where is the Boltzmann factor. The larger the energy level is with respect to ,
the larger is the Boltzmann factor, and the smaller is the probability that this energy level is
occupied. In simple words, high-energy levels are rarely populated, in favor of low-energy levels.
Combining eq. (6.2) and (6.3) we obtain the probability to observe a specific value of the mag-
netic moment given by
B
E k T = /( )
B
k T
J
m . This probability is:
( )
J
m x
J
J
j x
j J
e
P m
e
=
=

(6.4)
with . One can easily verify that the sum of over all values of
B B
x g B k T =( )/( ) P
J
m is 1, as
expected for the sum of probabilities for all possible states of a system. As long as , the
probabilities of all energy levels are very similar. This means that the thermal randomizing effect
is much larger than the aligning effect of the magnetic field (Fig. 6.2).
1 x
B
E
m
J
= +3/2
m
J
= +1/2
m
J
= 3/2
m
J
= 1/2
0
0
//
1 T
0 0.1 0.2 0 0.2 0.4
T = 300 K T = 3 K
P P

Fig. 6.2: Boltzmann distribution of the energy level probability for a spin at room temperature
( ), and at a temperature of 3 K, which is close to the lowest limit that can be reached with a conventional
P 3 2 J = /
300 K T =
4
He cryostat. The dashed line represents the uniform Boltzmann distribution of a completely randomized spin.
A typical macroscopic sample contains atoms, and each atom is subjected to
thermal impulses during a typical measurement time of 0.1 s. This means that a single measure-
ment averages over different spin configurations. Such a large number of configurations
21
10
9
10

30
10
63
allows us to replace
J
m for each atom with the expected value , which is calculated on the
basis of the probability distribution given by eq. (6.4). Using the definition of expected value for
a statistical variable with discrete probability distribution P we obtain:
J
m
J
m
all 's
J
J J
m
m m P = ( )

. (6.5)
Using eq. (6.1) and (6.4) we the expected value of the magnetic moment:
B B
J
j x
j J
J
J
j x
j J
je
m g g
e

=
=
= =

. (6.6)
The rather complicated expression in eq. (6.6) can be simplified by taking advantage of some
mathematical manipulations [Morrish, 1965], as we show in the following. Our goal is to elimi-
nate summations. To do so, we first observe that the numerator is the derivative of the denomi-
nator with respect to . From calculus we know that x ( ) ( ) [ln( )] x f x x

/ = f , so that we can
rewrite eq. (6.6) as:
B
d
ln
d
J
j x
j J
g
x

=
=

e . (6.7)
The summation in eq. (6.7) is a geometric progression with following solution:

2 1
2 2
1
1
1
J J x
j x Jx x x Jx Jx
x
j J
e
e e e e e e
e
( + )

=

= [ + + + + ] =

. (6.8)
Remembering that sinh ( ) 2
x x
x e e

= / , eq. (6.8) can be rewritten as:



2 2 2
2 2 2
1 2 1 2
2 2
1
sinh 2 1 2
.
sinh 2
J Jx Jx x x Jx x Jx x
jx
x x x x
j J
J x J x
x x
e e e e e e e e
e
e e e e
e e J x
x e e
/ /

/ / /
=
( + / ) ( + / )
/ /
(
= =
( )
[( +
= =
( / )

/
)
) / ]
(6.9)
We can now replace the sum in eq. (6.7) with the result of eq. (6.9), obtaining:

B
d sinh 2 1 2
ln
d sinh 2
J x
g
x x

[( + ) / ]
=
( / )
. (6.10)
Next, we calculate the derivative:



B
B
sinh 2 d sinh 2 1 2
sinh 2 1 2 d sinh 2
2 1 1
coth 2 1 2 coth 2 .
2 2
x J x
g
J x x x
J
g J J x x
J J

( / ) [( + ) / ]
=
[( + ) / ] ( / )
+
= [( + ) / ]


( / )
(6.11)
64
If a paramagnetic material contains N spins per unit volume, the resulting magnetization is
, and we finally obtain: M N =
B
0 B
B
2 1 2 1 1
coth coth
2 2 2
J
J
2
M y Ng JB y
J J
B y y
y
J J J
g J H
y
k T

+ +
J
( ) = ( )

( ) =



=

1
. (6.12)
The odd function is called Brillouin function [Brillouin, 1927]. It represents the
transition from a fully randomized spin configuration, defined by , to full alignment in
an external field, as given by the limit case of . The Brillouin function is
linear for small values of
1
J
B < <+
(0) 0
J
B =
( ) 1
J
B y =
y , and the following approximation:
1
0
3
J
J
B y y
J
+
( ) = (6.13)
is valid for the linear range. Transition from the linear range to saturation occurs around
3 (1 ) y J = / + J , and is most sharply pronounced for (Fig. 6.3). The cases correspon-
ding to and to
1 2 J = /
1 2 J = / J are of special interest for the description of magnetic states at
thermal equilibrium. In those cases, the Brillouin function is given by:
L
1 2
tanh
coth 1 ,
B y y
B y y y y
/

( ) =
( ) = ( ) = /
(6.14)
where is the so-called Langevin function. L
2 4 6 8 10
0.0
1.0
2 4 6 8 10
J = 1/2
7/2

J = 1/2
0 0
0.5
0.0
1.0
0.5
y
(a) (b)

1 3 y J J ( + )/( )
J
B
y
(
)
J
B
y
(
)

Fig. 6.3: (a) Brillouin function for and , ,
J
B y ( ) 0 y 1 2 J = / 7 2 J = / J . (b) The argument is scaled
so that all three examples have the same initial slope.
y
The initial slope of the magnetization curve in eq. (6.12) can be calculated using eq. (6.13),
and is called paramagnetic susceptibility :
p

65
2 2
B
p 0
B
3
p
N
k T

= , (6.15)
with (1 ) p g J J = + being the so-called effective number of Bohr magnetons. Some values of p
for common ions are listed in Tab. 6.1. The paramagnetic susceptibility is inversely proportional
to the absolute temperature, since it can be written as
p
2 2
B
0
B
.
3
C
T
p
C N
k

=
=
(6.16)
Equation (6.16) is known as the Curie law of paramagnetism, where C is the Curie constant. The
quantum theory of paramagnetism for free ions has been precisely validated for a wide range of
temperatures and applied fields, as shown in Fig. 6.4. The Curie law is best verified by plotting
the inverse paramagnetic susceptibility, , vs. temperature (Fig. 6.4b).
p
1 /
Tab. 6.1: Effective number of Bohr magnetons for some free ions, as obtained experimentally from the mea-
surement of various salts.
eff
p
Iron group, other p
eff
Pd/Pt group p
eff
Lanthanides p
eff
K
+
, Ca
2+
,Na
+
, Ca
2+
Ti
3+
, V
4+
V
3+
V
2+
, Cr
3+
, Mn
4+
Cr
2+
, Mn
3+
Mn
2+
, Fe
3+
(J = 2.5)
Fe
2+
(J = 2)
Co
2+
Ni
2+
Cu
2+
Cu
2+
, Zn
2+
0
1.7 (*)
2.8 (*)
3.8 (*)
4.9 (*)
5.9 (*)
5.4 (*)
4.8
3.2
1.9
0
Mo
4+
Mo
3+
Ru
4+
Ru
3+
Rh
3+
Pd
2+
Ir
4+
Os
2+
Ir
3+
Pt
2+

0.2
3.6
0.6
2.1
0.06
0.01
1.9
0.3
0.04
0
Ce
3+
Pr
3+
, Nd
3+
Sm
3+
Eu
3+
Gd
3+
(J = 3.5)
Tb
3+
Dy
3+
(J = 7.5)

Ho
3+
Er
3+
Tm
3+
Yb
3+
2.4
3.5
1.5
3.4
8.0
9.5
10.6
10.4
9.5
7.3
4.5
(*) J = S
66

Fig. 6.4: (a) Magnetic moment vs. for following salts: KCr(SO B T /
4
)
2
12H
2
O (I), FeNH
4
(SO
4
)
2
12H
2
O (II), and
Gd
2
(SO
4
)
3
8H
2
O (III) [Henry, 1952]. Points are measurements preformed in fields up to 5 T at temperatures indi-
cated in the plot, lines are the Brillouin functions for J S = and . The c.g.s. unit of converts to
. (b) Validation of the Curie law (solid line) with measurements of Gd(C
2 g = B T /
1
10 T K

/
2
H
3
SO
4
)
3
9H
2
O (dots) [Jackson
and Kamerlingh Onnes, 1923].
The magnetization curve of all paramagnetic substances, even those with the largest number
of effective Bohr magnetons, is perfectly linear at room temperature in fields that can be produ-
ced with a normal electromagnet. Linearity is often reported as a characteristic property of para-
magnetism. This is not completely correct, since saturation occurs at low temperatures, as shown
in Fig. 6.4, and event at room temperature in sufficiently large fields. Some lanthanide ions, such
as gadolinium ( ), and dysprosium ( ) have unusually large values of
3
Gd
+ 3
Dy
+
p (Tab. 6.1), and
their salts are the strongest paramagnetic compounds known. For example, dysprosium oxide
( ) is used for calibration purposes, and its magnetization curve in fields is clearly
non-linear below 100 K (Fig. 6.5).
2 3
Dy O 6 T <
The saturation magnetization of paramagnetic materials is
s B
M Ng J = (eq. 6.12). This
magnetization is quite large and comparable with that of ferromagnetic minerals. For example,
for dysprosium oxide: for comparison, magnetite has a saturation magnetiza-
tion of . At room temperature, very high fields in the order of are needed to
saturate a typical paramagnetic substance, as you will calculate in the following exercise.
2
s
298 Am kg M = /
2
90 Am kg /
2
10 T
Exercise 6.1. Calculate the magnetic field necessary to exceed the linear region of the magnetization curve of a Fe
3+

paramagnetic salt at room temperature (use Tab. 6.1 and use Fig. 6.3b to guess the range where the Brillouin
function is reasonably linear).
67
+2 +4 +6
6 4 2
+100
+200
+300
300
200
100
T = 3 K
T = 20 K
T = 100 K
B [T]
M [Am
2
/kg]








Fig. 6.5: Magnetization curves of
dysprosium oxide (Dy
2
O
3
), calcula-
ted for three different temperatures.
Dashed lines show the saturation
magnetization.

Exercise 6.2. Calculate the susceptibility of biotite K(Mg,Fe
2+
)
2
(AlSi
3
O
10
)(F,OH)
2
at room temperature and compare
your result with the measured value reported in Tab. 4.1 (use Tab. 6.1, knowing that Fe
2+
is the only paramagnetic
ion. Biotite molar mass: ). 433.5 g mol /
For completeness, we briefly mention that atomic nuclei also possess a spin with a corres-
ponding magnetic moment
N I
m g = (6.17)
where is the so-called nuclear magneton (see Box 5.1), is the nuclear
g-factor, and is the nuclear counterpart of
27 2
N
5.051 10 Am

= g
I
m
J
m . Because is 2000 times smaller than
Bohrs magneton, the contribution of atomic nuclei to the magnetization of matter is negligible,
even in the case of diamagnetic substances.
N

We conclude this chapter with a brief overview of phenomena and applications related to
paramagnetism. An interesting technological application is the so-called magnetocaloric effect. This
effect arises from the change in potential energy that occurs when atomic spins in a paramagnetic
material are aligned by an external field. Because potential energy decreases with increasing
alignment, conservation laws require that an equivalent amount of energy is released. Indeed, heat
is released when a paramagnetic material is magnetized. On the other hand, removing a magnetic
field allows the magnetic moments to relax into randomized states with higher energy levels. The
additional energy required to occupy these levels is taken from the material in form of heat, and a
temperature decrease is observed. The magnetocaloric effect is particularly effective when it is
possible to saturate a paramagnetic material. Saturation is easier to obtain with strong paramag-
netic substances at very low temperatures. This principle is used in the adiabatic demagnetization
refrigerator (ADR), which can reach temperatures of 1 mK. The working principle of an ADR is
very simple. A strongly paramagnetic salt is refrigerated to 1.4 K using a conventional
4
He
cryostat. A field of 1 T, sufficient for saturation at these temperatures, is then applied while the
salt is in thermal contact with the cryostat. The cryostat removes the heat produced while the
68
field is ramped up. When temperature is stabilized, the thermal contact with the cryostat is inter-
rupted and the magnetic field is decreased to zero. Field removal produces adiabatic cooling to a
final temperature that depends on the energy stored by the magnetic dipoles and the heat capacity
of the salt.
Paramagnetic minerals are obviously of little interest in paleomagnetism, because they cannot
carry a stable magnetization. The paramagnetic susceptibility is also a rather unspecific parameter
that cannot be used as a discriminator to distinguish among different minerals or mineral classes.
Nevertheless, an important application of paramagnetic susceptibility measurements is found in
structural geology. It is based on the intrinsic magnetic anisotropy of many rock-forming mine-
rals. Paramagnetic susceptibility anisotropy, defined as the dependence of on the direction
along which it is measured, arises from weak interactions between atomic spins and crystal struc-
ture. Typical anisotropies of paramagnetic minerals are in the order of 1-10%. Phyllosilicates
represents a particularly interesting case, because of their highly anisotropic crystal shapes. This
property makes them a perfect rock fabric marker, which is used to reconstruct tectonic defor-
mations and the original sedimentary deposition environment. The anisotropy of any magnetic
property is called magnetic fabric. Because of the simple and low-cost experimental setup required
to measure magnetic susceptibility in small fields, the anisotropy of magnetic susceptibility, AMS, is
one of the most widely used magnetic fabric parameters (see Box 6.1 and 6.2). One problem of
low-field magnetic susceptibility measurements is that they are responsive to all types of minerals,
including ferromagnetic ones. Phyllosilicates almost always contain inclusions of ferrimagnetic
minerals, whose strong magnetism easily overshadows the properties of the hosting mineral. A
method to isolate is shown in Fig. 6.6.
p

1.0 +1.0

0
H [T]
0
+1.5
1.5
M

[
1
0

2

A
m
2
/
k
g
]
0

p

Fig. 6.6: Magnetization of a red bed sample from the Sevier thrust belt (N of Salt Lake City, U.S.A.), measured in a
field that is cycled between +1 and 1 T (arrows). The magnetization is mainly carried by paramagnetic phyllosilica-
tes with minor contributions from very fine hematite (Fe
2
O
3
). Hematite is responsible for the difference between
measurements obtained in increasing or decreasing fields (arrows) and for the small magnetization remaining in zero
field. Because hematite saturates in large fields, can be estimated from the slope of the magnetization curve in 1
T [adapted from Weil and Yonkee, 2010].
p

69
Phyllosilicates contain structural and exchangeable cations: the most important are ,
, , ,
K
+
Na
+ 2
Ca
+ 2
Mg
+ 3
Al
+
, , and . Among these elements, and are the
only ions that possess a magnetic moment. Therefore, iron is a key element that controls the mag-
netic properties of these minerals, although a minor role is attributable to the lattice arrangement
of the other ions. Variable Fe substitution in the structure is responsible for the large scattering of
magnetic susceptibility data for the same mineral, depending on its provenance. The paramag-
netic susceptibility is found to be proportional to the total Fe content determined by chemical
methods (Fig. 6.7).
2
Fe
+ 3
Fe
+ 2
Fe
+ 3
Fe
+













Fig. 6.7: Magnetic susceptibility of micas vs.
total iron content as determined by chemical
analysis [from Hood and Custer, 1967].
70
Box 6.1: Mathematical description of anisotropy: second order tensors
Most crystalline substances are anisotropic: their vector properties depend on the direction along
which they are measured with respect to the crystallographic axes. The mathematical description
of anisotropy is summarized here using the example of magnetic susceptibility.
If a (small) magnetic field H is applied along an arbitrary direction to an isotropic paramagnetic
material, the resulting magnetization is parallel to H, with the scalar k being the
susceptibility (here we use k instead of to denote a mass-normalized susceptibility, which is
most used in practice). If the material is anisotropic, the induced magnetization tends to align
with preferred orientations, and M is no longer parallel to H. In this case, k is replaced by the
susceptibility tensor k , which, in a Cartesian coordinate system, has the form of a 33 matrix:
p
k = M H
p
p p

p
p
11 12 13
p 21 22 23
31 32 33
k k k
k k k
k k k



=



k
The susceptibility tensor is symmetric, which means that
ji ij
k for any i j : the matrix
has therefore only six independent elements (e.g. k , , , , k , ). The magnetization
is then given by , with both the amplitude and the direction of depending on the
direction of with respect to a coordinate system that is fixed to the material being measured.
k = =
22
k
, 1, 2,3
11 12
k
13
k
23 33
k
p
= M k H M
H
A fundamental property of (and all tensors of this type) is that there are three particular
directions, called principal axes, where remains parallel to : the values of along these
directions are the eigenvalues of the susceptibility tensor. The tensor can be visualized by drawing
the surface described by the tip of the vector when is a unit vector with variable orienta-
tion. The resulting surface is an ellipsoid whose half-axes are vectors with directions and amplitu-
des coinciding with the principal directions and eigenvalues of the tensor, respectively. This re-
presentation of a tensor follows from the so-called eigendecomposition:
p
k
M H
p
k
p
k r r
min
p min int max int min int max
max
0 0
0 0
0 0
k
k
k
1



=[ ] [ ]



k r r r r r r
with , , and being the minimum, intermediate, and maximum eigenvalues and
, , and r the unit column vectors (so-called eigenvectors) indicating the direction of
the corresponding principal axes. An isotropic material is characterized by : the
tensor is represented by a sphere and the principal axes are undetermined.
min
k
int
k
max
k
min
r
int
r
max
min int max
k k k = =









Triaxial susceptibility tensor ellipsoid (shaded surface)
with eigenvalues k , , and k along the prin-
cipal axes defined by the eigenvectors , , and
. The magnetization induced by a field H along
an arbitrary direction defined by the unit vector is
.
min int
k
max
min
r
int
r
max
r
r
p
( )H = M k r




r
max
r
min
r
int
k
max
k
min
k
int
r
k

r
71
Box 6.2: Rock fabrics and anisotropy of magnetic susceptibility
Rock fabric is often correlated with the anisotropy of magnetic susceptibility (AMS), especially if
magnetic susceptibility is carried by minerals that have a highly anisotropic relationship between
shape and directions of minimum and maximum susceptibility. This is for example the case of
phyllosilicates, whose crystalline structure consist of stacked tetrahedral layers and
octahedral or layers. These minerals tend to occur in form of platelets (e.g.
micas). Magnetic susceptibility is described by an oblate (i.e. disk-shaped) tensor with per-
pendicular to the silicate layers, and in the plane defined by the layers (Table).
4
Si, Al O ( )
6
Fe, Mg O ( )
6
AlO ( )
min
k
int max
k k
AMS of selected phyllosilicates after Martn-Hernandez and Hirt [2003]: is the bulk susceptibility.
bulk
k
mineral
bulk
k ,
8 3
10 m /kg
min bulk
/ k k
int bulk
/ k k
max bulk
/ k k
Biotite
Muscovite
Chlorite
40


2
8.9


2
8.5


1

0.85


0.02
0.92


0.03
0.92


0.02
1.07


0.01
1.03


0.01
1.04


0.01
1.08


0.01
1.04


0.01
1.04


0.01
The sedimentary fabric of phyllosilicates-bearing rocks is controlled by the horizontal deposition
of platy minerals. The typical AMS signature is an oblate tensor with minimum susceptibility
perpendicular to the bedding plane (Fig. A). Post-sedimentary tectonic activity can deform sedi-
mentary rocks, superimposing a new fabric controlled by changes in the orientation of existing
minerals or by the oriented growth of new minerals. For instance, compression in the bedding
plane (Fig. B), will cause a rotation of the phyllosilicates around an axis that is parallel to the bed-
ding plane and perpendicular to the compression axis. In the most extreme case, rotations by any
angle between 0 and 90 are possible. The resulting AMS tensor is the sum of individual tensors
of the rotated minerals. It is a prolate (i.e. cigar-shaped) tensor with maximum susceptibility
along the rotation axis, and along any direction per-
pendicular to the rotation axis.
k

k
max max
k k

=
min int min max
2 k k k k

= =( + )/

(A) Sedimentary fabric of
red beds from the Sevier
thrust belt (N of Salt Lake
City, U.S.A.), and (B) tec-
tonic fabric overprint re-
sulting from compression
in the bedding plane. (C,
D) show the correspon-
ding AMS fabric with
principal axis represented
on a stereographic dia-
gram. Small symbols are
individual sample measu-
rements of , ;
large symbols represent
averages with numbers
min
k
max
k











indicating tthe eigenvalues normalized to [modified from Weil and Yonkee, 2010].
bulk
1 k =
Sedimentary deposition & compaction Intermediate tectonic fabric
A) B)
compression
compaction
oblate
k
max
k
min
detrital mica hematite clays neocrystallized mica
0.992
0.995
1.013
C) D)
k
int
Tensor principal values & axes:
prolate
k
min
k
max
0.990
1.004
1.006
72
Literature:
Brillouin, M.L. (1927). Les moments de rotation et le magntisme dans la mcanique ondulatoire, Journal de
Physique et le Radium, 8, 7484.
Henry, W.E. (1952). Spin paramagnetism of Cr
3+
, Fe
3+
, and Gd
3+
at liquid helium temperatures and in strong
magnetic fields, Physical Review, 88, 559562.
Hood, W.C., and R.L.P. Curster (1967). Mass magnetic susceptibilities of some trioctahedral micas, American
Mineralogist, 52, 16431648.
Jackson, L.C., and H. Kamerlingh Onnes (1923). Les proprits magntiques de lethylsulfate de gadolinium aux
basses tempratures, Comptes Rendus hebdomadaires des sances de lAcadmie des sciences, T177, 154158.
Kittel, C. (2005). Introduction to solid state physics, John Wiley & Sons, Inc.
Martn-Hernandez, F. and A.M. Hirt (2003). The anisotropy of magnetic susceptibility in biotite, muscovite, and
chlorite single crystals, Tectonophysics, 367, 1328.
Morrish, A.P. (1965). The physical principles of magnetism, Wiley series on the science and technology of materials,
Wiley, New York.
Weil, A.B., and A. Yonkee (2010). Anisotropy of magnetic susceptibility in weakly deformed red beds from the
Wyoming salient, Sevier thrust belt: Relations to layer-parallel shortening and orogenic curvature, Lithosphere, 1,
235256.
73

74
7. Magnetic resonances
In this chapter we briefly introduce the concept of magnetic resonance, a phenomenon that
occurs as consequence of the discrete energy levels of atomic spins. Here we discuss the simplest
form of resonance, which is called electron paramagnetic resonance (EPR). We have seen in chapter
6 that the application of a magnetic field to a spin with quantum number J produces a Zeeman
splitting of the spin energy into 2 1 J + equally spaced levels. The difference between conse-
cutive energy levels is easily derived from eq. (6.2), and is given by:
E
eff B
E g = B
H g
(7.1)
with . In case of free (isolated) ions, is a factor given by Lands equation (eq.
1.16). In crystalline solids, might differ from the Lands value, due to lattice interactions
that we are not discussing here. We have also seen that transitions between different energy levels
are made possible by thermal perturbations with energy . Another possibility to stimulate a
transition is provided by the absorption (for a transition to the next higher level) or emission (for
a transition to the next lower level) of a photon ( ) with energy (Fig. 7.1a). Interactions
between atoms and electromagnetic waves of frequency occur through light quanta (photons)
of energy h , with being Plancks constant (Tab. 0.2). Comparison of the photons energy
with eq. (7.1) gives the frequency of electromagnetic waves that are absorbed or emitted by
magnetic energy level transitions:
0
B =
eff
g =
eff
g
B
k T
E
h
eff B
g
B
h

= (7.2)
Using , eq. (7.2) can be written as , where we see that, for ordinary
magnetic fields of 0.1-1 T, the corresponding frequency is in the microwave range of the electro-
magnetic spectrum. Because lower energy levels are more populated than higher energy levels,
there is a higher probability for a photon to be absorbed by an atom, rather than emitted. There-
fore, paramagnetic materials absorb electromagnetic waves when (7.2) is satisfied. If the applied
field is constant, absorption occurs for a very narrow range of frequencies, and this phenomenon
is called electron paramagnetic resonance. This expression comes from the fact that the magnetic
moment of atoms is mostly produced by unpaired electrons, and is used to distinguish it from nu-
clear magnetic resonance (NMR) occurring inside atomic nuclei.
eff
2 g [GHz] 28 [T] B
Paramagnetic resonance can also be explained in the framework of classical physics by the fact
that an atomic magnetic moment is always produced by an angular momentum . An
external field applied along an arbitrary direction generates a torque T (eq. 3.2), with
being the component of perpendicular to B and . This torque causes the magnetic
moment to precess with an angular velocity , which is the Larmor precession we have
seen with Langevins diamagnetism (chapter 5). Since is a component of the full moment
vector, it can be written as (eq. 6.1). Similarly, is the component of
that produces , and the Larmor precession is given by:
L
B B
L

L
L
T = /

B J
m g =
J
L m

= L

75

B
L
B g
B
L

= = (7.3)
Using and , equation (7.2) is obtained again. Therefore, Larmors preces-
sion can be stimulated by a magnetic field of the same frequency, obtaining resonance (Fig. 7.1b).
Such magnetic field can be seen as the magnetic component of an electromagnetic wave, explai-
ning the absorption of microwaves.
2 = 2 h = /( )
B
E
0
0
//
E
B

B(t)
(a) (b)


Fig. 7.1: (a) Quantum mechanical EPR model: a photon produces a transition between two energy levels of an
atomic magnetic moment in a field, provided that its energy matches the gap between consecutive levels. (b)
Classical model of EPR, where the atomic magnetic moment precesses around an applied field . The precession
is stimulated by a rotating field perpendicular to B . Any arbitrary component of is equivalent to an oscillating
(electromagnetic) field whose frequency matches the Larmor precession frequency.

E
B

B

B
EPR is measured by placing the sample in a so-called microwave cavity and exposed to micro-
waves and constant magnetic field. A microwave antenna inside the cavity measures the electro-
magnetic field, whose intensity decreases as the resonance condition of eq. (7.2) is met. A reso-
nance spectrum is obtained by recording the microwave intensity while continuously changing the
microwave frequency in a constant applied field (frequency sweep), or by continuously changing
the applied field in microwaves of constant frequency (field sweep). The second option is used in
practice, because of the difficulty to generate microwaves with arbitrary frequencies. Most EPR
measurements are performed in the so called X-band at a frequency of 9.3882 GHz. An ideal EPR
spectrum consists of a single absorption peak centered at the resonance field that satisfies eq.
(7.2): this is 0.335 T if X-band microwaves are used (Fig. 7.2). Absorption peak broadening is
caused by paramagnetic relaxation phenomena, which depend on the rate of thermally activated
transitions between energy levels.
EPR would be of little utility if all spectra have the same characteristics plotted in Fig. 7.2.
This is not the case, first of all because of differences in producing multiple absorption peaks
in materials containing different types of paramagnetic ions, or ions with different coordination.
eff
g
76
0.25 0.30 0.35 0.40 0.20
= 9.3882 GHz
a
b
s
o
r
p
t
i
o
n
magnetic eld [T]

Fig. 7.2: Idealized EPR spectrum for an
isolated atom with in the microwa-
ve X-band. The upper curve is microwave
absorption, while the lower curve repre-
sents its first derivative with respect to the
field. The derivative is often plotted as EPR
spectrum, because it is directly measured as
output of a lock-in amplifier used to keep
the microwave power at a constant level
inside the cavity.
2 g =
Furthermore, the amplitude of the gap between magnetic energy levels is finely modulated by
the interaction between electrons and the crystalline electric field, which causes an additional
splitting of the energy levels that is responsible for the so-called fine structure of the spectrum. A
further splitting caused by the interaction of unpaired electron spins with nuclear spins produces
the so-called hyperfine structure. These structures depend on the nature of the material measured
and provide a sort of fingerprint useful for its detection. A main application of EPR is the
detection and identification of free radicals formed during chemical reactions.
EPR is widely used to investigate the origin of paramagnetism in crystalline substances. While
magnetic susceptibility is measuring the total contribution of all paramagnetic ions, EPR can
distinguish the contribution of individual ions on the basis of different values of . For exam-
ple, EPR has been used to understand the origin of magnetic anisotropy in clay minerals (Fig.
7.3), providing a theoretical understanding of magnetic fabrics (see Box 6.1 and 6.2) produced by
these minerals.
eff
g


Fig. 7.3: EPR X-band spectrum of kaolin, measured in a direction parallel () and perpendicular () to the basal
plane. The left plot is a portion of the spectrum showing the contribution of Fe
3+
ions, while the plot on the right
shows minor contributions due to radiation-induced defect centers [from Schreiner et al., 2002].
77
We conclude this section by briefly mentioning two other types of magnetic resonance: ferro-
magnetic resonance (FMR), and nuclear magnetic resonance (NMR). Ferromagnetic resonance is
based on exactly the same principle as EPR, and is measured with the same instruments. The only
difference is that ferromagnetic or ferrimagnetic minerals are measured, instead of paramagnetic
substances. These minerals contain atomic spins, exactly like paramagnetic substances do. There-
fore, the same resonance phenomena appear when transitions between different spin energy levels
are stimulated by microwaves. The only difference with respect to paramagnetism is given by the
presence of an internal field produced by the strong magnetization of ferromagnetic minerals
(see chapter 2 and Fig. 2.6). The internal field adds to the field applied during FMR measure-
ments, changing the resonance condition of eq. (7.2) to
i
H
eff B
0
g
h

= | + H H
i
| . (7.4)
Because is generally inhomogeneous, individual spins are subjected to different resonance
conditions, producing a marked broadening of FMR spectra with respect to EPR. FMR is used to
infer the distribution of internal fields, which depend on the composition and shape of the ferro-
and ferrimagnetic minerals being measured. An interesting FMR application to rock magnetism
has been recently developed for the detection of fossil magnetotactic bacteria remainders in sedi-
ments and sedimentary rocks (see Box 7.1).
i
H
Nuclear magnetic resonance (NMR) is the only popularly known type of magnetic resonance,
because of its medical applications. The resonance condition is given by equation (7.2), with
and being replaced by their nuclear counterparts. Because the nuclear magneton is 2000
times smaller than , the resonance frequency is in the MHz range for applied fields of 1-2 T.
NMR is particularly sensitive to all atoms whose nuclei contain an odd number of nucleons (i.e.
protons and neutrons): for example
eff
g
B

1
H and
13
C. NMR is used as a material characterization tool
in various disciplines such as medicine, chemistry and non-destructive material testing.
Readers interested in learning more about this subject can find a detailed theoretical handling
of magnetic resonance phenomena in Morrish [1965].
78
Box 7.1: Using FMR to detect fossil magnetosomes
The resonance condition for spins contained in ferromagnetic materials depends on the external
field (as in the case of EPR), as well as on magnetic energy contributions arising from (1) crystal
anisotropy, (2) crystal shape, and (3) interactions with other magnetic particles. The only contri-
bution which can be understood directly in terms of an additional (internal) magnetic field co-
mes from crystal shape, leading to the modified resonance condition of eq. (7.4). In order to
include the effect of all energies on resonance, the torque T acting on atomic spins needs to be
expressed as function of the free magnetic energy , which depends on spin orientation,
expressed in spherical coordinates by the polar angle and the azimuthal angle . Due to the
strong coupling between spins, the whole magnetization inside ferromagnetic crystals is subjected
to Larmor precession. Solution of the equation of motion for spins with free magnetic energy
in an applied field gives the resonance condition:
( , ) E

( , , ) E H H

2 B
s
g
E E E
M

=
where g is the Land g-factor,
s
M the spontaneous magnetization, and the subscripts indicate
derivation with respect to or [Morrish, 1965]. The external field at which resonance occurs
in microwaves of angular frequency depends on the orientation of magnetic crystals
with respect to the applied field (Fig. A). A sample containing a large number of identical, ran-
domly oriented magnetic crystals as typically the case of rocks has different resonance condi-
tions for each orientation and produces a broad FMR spectrum. The spectrum shape is control-
led by the angular dependence of . In case of strong uniaxial anisotropy (i.e. the magneti-
zation tends to align with the so-called easy axis), the FMR spectrum measured as first deri-
vative of microwave absorption has the characteristic asymmetric shape shown in Fig. B.
2 =
( , ) E
Magnetite particles derive uniaxial anisotropy from their elongation. Because of the cubic crystal-
line structure, magnetite tents to grow as equidimensional or slightly elongated crystals in most
cases, producing quasi-symmetric FMR spectra. A notable exception is represented by linear
chains of magnetite particles produced by magnetotactic bacteria. Magnetic interactions inside
the chains produce a strongly uniaxial anisotropy that is clearly seen in FMR. This feature has
been used to detect magnetite chains in sediments [e.g. Weiss et al., 2004].







(A) Ferromagnetic resonance field calculated for a chain of magnetite particles, as a function of the chain
axis orientation angles and . (B) FMR spectrum, calculated as the first derivative of microwave
absorption with respect to , for a large number of randomly oriented chains similar to the one shown in
the insert [modified from Charilaou et al., 2011].

H
79
Literature:
Charilaou, M., M. Winklhofer, and A.U. Gehring (2011). Simulation of ferromagnetic resonance spectra of linear
chains of magnetite crystals, Journal of Applied Physics, 109, 093903.
Morrish, A.P. (1965). The physical principles of magnetism, Wiley series on the science and technology of materials,
Wiley, New York.
Schreiner, W.H., K.C. Lombardi, A.J.A. de Oliveira, N. Mattoso, M. Abbate, F. Wypych, and A.S. Mangrich
(2002). Paramagnetic anisotropy of a natural kaolinite and its modification by chemical reduction, Journal of
Magnetism and Magnetic Materials, 241, 422429.
Weiss, B.P., S.S. Kim, J.L. Kirschvink, R.E. Kopp, M. Sankaran, A. Kobayashi, and A. Komeili (2004). Ferromagne-
tic resonance and low-temperature magnetic tests for biogenic magnetite, Earth and Planetary Science Letters,
224, 7389.

80

You might also like