Chapter 1

Consequences of Magnetic Interaction

Phenomena in Granular Systems

Leandro M. Socolovsky and Oscar Moscoso Londoño

1.1 Introduction

Magnetic granular materials are composites in which a matrix is filled with magnetic
particles that have sizes in the nanometer range. Examples of such materials are
those prepared by sputtering, where two or more elements that do not easily form
compounds are deposited onto a substrate; an incomplete multilayer deposition; a
heterogeneous alloy; nanoparticles incorporated into a matrix by physical methods;
or even agglomerations of nanoparticles covered with another substance. Figure 1.1
shows two such granular materials. For several decades nanostructures have been
produced by physical and chemical methods. Enhanced techniques for the manip-
ulation of matter at the nanoscale have enabled the preparation of better controlled
nanostructures. Such magnetic granular materials have interesting physical and
chemical properties that are also of technological interest. Some of those physical
properties are related to magnetism, like superparamagnetism, or, in connection with
magnetotransport, giant magnetoresistance (GMR), tunneling magnetoresistance
(TMR), and the giant Hall effect (GHE). Some of these are currently being exploited
in computer hard disks, perpendicular recording of magnetic bits, and in biomedical
applications such as drug delivery and magnetohyperthermia, among others. The
read heads of modern hard disks, which use the GMR effect, represent the first

L.M. Socolovsky ()

LSA, INTECIN, Facultad de Ingeniería, Universidad de Buenos Aires-CONICET, Av. Paseo
Colón 850, C1063ACV, Buenos Aires, Argentina
e-mail: lsocolovsky@fi.uba.ar
O. Moscoso Londoño
LMBT, Instituto de Física “Gleb Wataghin”, Universidade Estadual de Campinas,
Campinas, SP, Brazil
Universidad Autónoma de Manizales, Antigua Estación del Ferrocarril, Manizales, Colombia
e-mail: omoscoso@ifi.unicamp.br; oscar.moscosol@autonoma.edu.co

© Springer International Publishing AG 2017 1

S.K. Sharma (ed.), Complex Magnetic Nanostructures,
DOI 10.1007/978-3-319-52087-2_1
2 L.M. Socolovsky and O. Moscoso Londoño

Fig. 1.1 Above: A granular material can be formed by the dispersion of nanoparticles in a matrix
(left) or by the clustering of grafted nanoparticles (right). Below: (left) cobalt nanoparticles in a
silica matrix; (right) magnetite nanoparticles grafted with silica (Image courtesy Dr. D. Muraca)

practical realization of spintronics [1]. Spintronics is considered a new technology

that differs from electronics in that it can be used to control the spins of electron
instead of their electrical charge, opening up new possibilities for devices that will
cause them to run faster and with less thermal energy losses.
Our focus in this chapter is on phenomena that occur in magnetic granular
materials, considering the broad spectrum of interest they have attracted in very
disparate areas of research, for example, chemistry, physics, biology, medicine,
and various branches of engineering. For most workers in these fields, we believe,
an article that presents a first approach to experimental results found using the
most common measurement techniques would be of great practical value. We
expressly omit any discussions of in-depth theory and less common measurement
techniques. Our purpose is to review briefly some important concepts, recent
results, and discussions and to highlight the connections between magnetic and
magnetotransport phenomena, stressing experimental results and the best known
analyses. We present results obtained from various experiments performed in the
course of our research.
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 3

Fig. 1.2 Transmission

electron microscopy image of
a magnetite nanoparticle.
Atoms are seen as dots, and
the crystalline arrangement is
evident. Surface atoms are a
significant fraction

1.1.1 Superparamagnetism

A single magnetic nanoparticle of a few nanometers (Fig. 1.2) can be composed of

a magnetic material like iron, cobalt, nickel, gadolinium, or their alloys, and some
oxides of iron, cobalt, and manganese have a magnetic order of their own given
by interchange or super-interchange interactions, depending on the material [2, 3].
Magnetic materials tend to minimize energy by breaking down their magnetic order
into domains, which are regions that are magnetized in different orientations. Fron-
tiers between domains are called domain walls, which require the creation of extra
energy. If a magnetic material is small enough, that extra energy is costly compared
with that needed to remain a unique magnetic domain. In that case, to keep energy
in a minimum, the magnetic order inside a nanoparticle will not break down into
domains, so it will remain a magnetic monodomain. In a first approximation, all
atoms in the nanoparticle will add their magnetic moments to form a supermoment,
which represents nanoparticle magnetization as a whole. In other magnetic materials
that are antiferromagnetic (AFM) or ferrimagnetic (FiM), the net magnetic moment
of the atomic arrangement will represent the nanoparticle’s whole magnetization.
This supermoment also indicates that all atomic magnetic moments will behave in
unison in response to external magnetic fields [4].
Considering that competition over the energy needed to create a domain wall and
to minimize energy use, all magnetic materials will have a critical size for a tiny
magnetic particle to become a monodomain. Theoretically, for a uniaxial spherical
crystal with a large anisotropy constant K, the critical radius rc is

.AK/1=2
rc  9 ; (1.1)
0 Ms2
4 L.M. Socolovsky and O. Moscoso Londoño

where A is the exchange constant, 0 is the constant of magnetic permeability,

and Ms is the saturation magnetization. Using the preceding expression, rc is 3 nm
for Fe and 4–8 nm for Co [3, 5]. This critical size must be taken as a rough first
approximation owing to the simplifications performed in that calculation. More
accurate models give larger critical radii [3]. Critical sizes are below 100 nm
for most magnetic materials [5, 6]. Most experiments conducted with magnetic
nanoparticles use magnetite, maghemite, or cobalt ferrite, which have approximate
critical sizes of 64, 45, and 50 nm [5], respectively.
Because of the crystalline order in nanoparticles, supermoments tend to align
with some specific crystallographic orientations. Those orientations are known as
easy axes. Thermal energy causes supermoments to flip among easy axes. At a high
enough temperature and for a given period of observation time (measurement time),
this flipping causes the average magnetic moment of nanoparticle supermoments
to be zero. In other words, a magnetic material, if small enough, will appear
as a nonmagnetic material at, for example, room temperature. We can measure
a nonzero magnetization if our measurement device is fast enough at a given
temperature. As a consequence, depending on the technique used for measuring,
we can “see” the magnetic moment of a nanoparticle if our instrument is fast
enough to measure some supermoment at a given orientation. If the device takes
measurements too slowly, the supermoment will flip back and forth between easy
axis orientations, so the measured average moment will be zero. This property is
called superparamagnetism (SPM), and it is considered a type of magnetic order in
materials [2, 6–10]. For details on this mechanism and its formulation, see Knobel
et al. [4]. Early formulations of superparamagnetic theory started by considering a
set of noninteracting magnetic nanoparticles. The magnetization, M, of these kinds
of magnetic systems can be described by a Langevin function L:
   
M H kB T H
D coth  DL ; (1.2)
MS kB T H kB T

where MS is the saturation magnetization, H the magnetic field, T the absolute

temperature,  the magnetic moment (supermoment), and kB the Boltzmann
constant.
In these schemes, all atoms behave in similar way, regardless of whether they
are in the core of the nanoparticle or on the surface. It is also assumed that there
is only one easy axis. These descriptions apply in the study of granular systems
in first approximations. Of course, these are naïve descriptions that do not take
into account surface effects, more than one easy axis, close nanoparticles, or
distributed properties such as size, moment, or blocking temperature distributions,
for example [4].
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 5

1.1.2 Anisotropies

The energies of interactions are manifested through a quantity called anisotropy, K.

Anisotropy derives from two basic origins: spin-orbit and magnetostatic inter-
actions. We briefly enumerate those that are usually found in nanoparticles:
magnetocrystalline, surface, shape, and strain anisotropies.
1. Magnetocrystalline anisotropy: this is a manifestation of the internal energy’s
dependence on the direction of spontaneous magnetization. It forces a magnetic
moment to stay in certain preferred orientations that coincide with crystal-
lographic axes because electron orbits are linked to them. Although magne-
tocrystalline anisotropy changes with temperature, in most analyses in SPM
experimental studies, this is not considered explicitly [3, 9]. It is important to
stress that it is this anisotropy that orients the magnetization of a nanoparticle.
2. Surface anisotropy: atoms located at surfaces experience a break in symmetry
in their interactions because the near-neighbor population is different from those
atoms located at the core. In nanoparticles, the surface-to-volume ratio is large
because of the small dimensions, so this effect can be important [11]. As is made
clear in Fig. 1.2, which shows a 7 nm nanoparticle, several atoms are located on
or close to the surface. The smaller the particle, the more relevant this anisotropy
becomes, eventually surpassing the effects of magnetocrystalline anisotropies.
The roughness of the interface is also another source of the anisotropy.
3. Shape anisotropy: this type of anisotropy depends on the differences in the
shape of the magnetic specimen, which gives rise to different demagnetization
factors in different orientations. If it is spherical, there is no such contribution.
4. Strain anisotropy: this comes from mechanical stresses induced by the defor-
mation of the crystalline structure.
Detailed discussions on anisotropies can be found in the excellent and enjoyable
book of Aharoni and in the very useful reviews of Bedanta [5, 9, 10] and Majetich
[12].
Magnetostriction plays an important role in interactions with temperature by
shrinking or expanding nanocrystals, which can change their shape and size. Also,
dependences with temperature of magnetocrystalline anisotropy factors contribution
to modifications of the energy barrier [9]. In real systems all those anisotropies are
considered, through their effects on the measured magnetization, as an effective
value, Kef . This effective value is calculated using
EB
 D  0 e kB T ; (1.3)

in which EB D Kef V, where  is the relaxation time, which is spent by the

supermoment to jump between easy axes. The constant  0 is the attempt time, in
which the magnetization vector “tries” to jump between easy axis directions. It
is accepted that  0 has values in a range of 109 –1011 s. The reciprocal of this
parameter has dimensions of frequency and is denoted by f0 [4, 5, 9–12].
6 L.M. Socolovsky and O. Moscoso Londoño

A systematic study on well-prepared cobalt ferrite nanoparticles showed a

consistency with this usage of an effective anisotropy with an explicit dependence on
temperature and a phenomenological model that works with that material [13, 14].
In brief, it is through anisotropies that specific characteristics of nanoparticles and
granular materials will be reflected in magnetic properties.

1.2 Interactions Among Nanoparticles

In real systems, nanoparticles are not very far from each other, so in principle, the
magnetism of one nanoparticle should affect others. Principally, two long-range
interactions, known as dipolar and RKKY, must be considered in granular materials.
A dipolar interaction is described by
 
ED
0 
!
m !  3 
m ! !
m r C  r  ;
! !
m (1.4)
1 2 1 2
4r3 r2

where m1 and m2 are the magnetization vectors of entities like nanoparticles,

separated by a distance r [6]. Since a nanoparticle can have hundreds or thousands
of atoms, a typical value of m can be on the order of 103 –105 Bohr magnetons. At
close distances, this energy term can be relevant. In any case, the dipolar energy
calculated using this expression should be taken as an upper limit.
As can be seen, dipolar energy depends on nanoparticles’ magnetic moment
separation and their mutual alignment [5–7]. Because real nanoparticle systems
contain between at least 1010 and 1020 nanoparticles/cm3 , use of the preceding
equation makes it rather difficult to quantify the dipolar effects in a nanostructured
system. In this situation, Hansen and Mørup show that for a set of randomly
distributed magnetic nanoparticles, the dipolar energy can be calculated using

0 m2
Ed  : (1.5)
4 r3
In this formulation, m and r are assumed to be the averages of the magnetic
moment and the nanoparticle separation, respectively [14].
In a granular material, another important type of long-range interaction is
the well-known RKKY interaction. Such interactions, formulated by Ruderman,
Kittel, Kasuya, and Yoshida in the mid-1950s, appears in materials with itinerant
electrons. In the particular case of nanoparticles supported in a metallic matrix, the
RKKY interaction does not involve a direct coupling between magnetic nanoparticle
moments but is mediated by itinerant electrons, which are subject to a localized
magnetic moment that polarizes them [8], that is, the mediated electrons are affected
by a spin-dependent local potential. This interaction is present in granular materials
where the matrix is metallic. RKKY interactions are oscillatory and fall with the
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 7

cubic power of distance. At large distances the term that regulates the strength of
this interaction, the exchange parameter J, is
 
cos .2kF r C '/
J D J0 ; (1.6)
.2kF r/3

where J0 is a constant, kF is the Fermi momentum, and ® is a phase that depends on

the matrix and the magnetic entity. The parameter J can have positive or negative
values, depending on the distance. Details on this expression are not important for
our study and can be found in [8, 9, 15, 16].
Besides when magnetic nanoparticles are close enough or in contact, a direct
exchange, or a Dzyaloshinskii–Moriya or superexchange, interaction is established
among surface magnetic ions with spins Si and Sj , which belong to different
nanoparticles:

 !
!  ! 
HDM D dij  Si  Sj : (1.7)

When nanoparticles are very close to each other, or even in contact, a dipolar
interaction is not the predominant one. Direct interaction between magnetic ions
located on different nanoparticles can be common, so a new magnetic state emerges.
This state is usually called superspin glass (SSG) [5]. This name might no seem
appropriate at first sight because the origin of the new state does not lie in the
competing RKKY interaction, as is the case for spin glasses [17]. When percolation
is established, then we have a state called superferromagnetism (SFM) [5, 18]. In
the authors’ opinion, more appropriate names could be used, such as interacting
superparamagnetism (I-SPM) instead of SSG or percolated superparamagnetism
(P-SPM) instead of SFM. A qualitative magnetic phase diagram is shown in Fig. 1.3.
We can think of these magnetic behaviors in granular systems as balls of the same
size that fill a large glass box (Fig. 1.4). Let us imagine black balls as magnetic
nanoparticles and the white matter that surrounds them as a nonmagnetic matrix.
We begin by considering the box filled exclusively by white matter: our system
is nonmagnetic. If we randomly add more balls, which may be far apart, we will
have a canonical superparamagnetic system. With the addition of more balls, the
system more closely resembles a real superparamagnet. We can easily see that as
our magnetic nanoparticle concentration increases (with the addition of more balls),
some of the balls will come into close contact with each other, sometimes with more
than two balls, and we see SSG or I-SPM states. If we continue to fill the receptacle,
we will see more and more chains of balls, until we see a pathway spanning the
whole box. This point is called the percolation limit, and its magnetic manifestation
is the SFM or P-SPM state. When this percolated state is reached, there will still be
isolated balls, so we must expect a complex magnetic behavior that will show both
SPM and ferromagnetic features. By adding more balls, we finally obtain a fully
ferromagnetic state.
Percolation is a geometrical concept [19]. The situation presented here just shows
contacting particles. Because we are talking about magnetism, which is force that
8 L.M. Socolovsky and O. Moscoso Londoño

Fig. 1.3 Changing nanoparticle concentration, in three phases. Less-concentrated granular mate-
rials behave as predicted by theory. The intermediate range (I-SPM or SSG) can be described,
at lower concentrations, by Dormann–Bessais–Fiorani, Mørup, or T* models [4]. At higher
concentrations, only a T* model can describe magnetic behavior. Close to the percolation limit
all models fail. At that percolation limit, a long-range magnetic order, based on direct interchange
between atoms, starts to become established until a full ferromagnetic-like (e.g., FM or AFM)
order sets in

Fig. 1.4 Black balls represent magnetic nanoparticles. As the receptacle is filled with magnetic
nanoparticles, the system evolves from a noninteracting system to a fully percolated one. Pictures
(a–c) represents granular materials above percolation threshold, and (d–f) represent percolated
systems
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 9

Fig. 1.5 In real samples

there are larger and smaller
nanoparticles, each with a
magnetic moment of different
magnitude

acts at a distance, it is not strictly necessary that the nanoparticles be touching

each other. Direct exchange can be sizable at distances of up to a few nanometers,
so a coupled magnetic ordering can be established, even without direct contact
between nanoparticles [20]. A question arises: when can a granular material be
considered to be diluted or noninteracting? As we saw from the earlier equations,
the strength of forces among nanoparticles depends not only on distances but also
on the supermoment and size [21]. As a rule of thumb, a dispersion of 0.01 vol.%
or less can be considered as being close to a noninteracting system [22].
Also, to depict a real system, we must consider the size distribution, so in our
imaginary experiment, we must use larger and smaller balls (Fig. 1.5). Clearly,
the supermoments of larger nanoparticles exert a stronger influence on their
neighborhoods compared to smaller nanoparticles, as was pointed in the work of
Allia et al. [23].
Initially, we considered a nonmagnetic matrix. This matrix can, in principle, be
a conductive, semiconductive, or dielectric material. It can also have a different
magnetic order, like an antiferromagnet, that can lead to the emergence of very
interesting properties, like the shift of the blocked regime of the superparamagnetic
state to higher temperatures. Such a granular system can enable denser recording
on hard disk drives. In a disk, a bit is recorded by orienting a magnetic sector on
the surface in one direction or another, representing a 1 or a 0, the basis for binary
code. Because this recording sector must be small, to allow for dense recording, its
size approaches the superparamagnetic limit. If that size is small enough to be in
the superparamagnetic state at room temperature or slightly higher, it is not possible
to keep the magnetization as at the original level. To avoid that superparamagnetic
limit, it was demonstrated that it is possible to produce a granular material with the
desired effect [24, 25]. A granular structure like that described in the cited articles
is composed of superparamagnetic cobalt nanoparticles in an antiferromagnetic
matrix. The matrix, made up of cobalt oxide, helps to retain the magnetization by
shifting the blocking temperature to higher temperatures than that of the original
nanoparticle, as would be the case in a nanosized sector alone. This architecture
makes it possible to overcome the size limitation, thereby enabling dense recording.
10 L.M. Socolovsky and O. Moscoso Londoño

1.2.1 Footprints of a Granular System

When we need to know whether a material is composed of superparamagnetic

particles and how they are dispersed in the matrix, we can use magnetic measure-
ments. Two easy ones are magnetization M versus temperature T (M vs. T) and
magnetization versus the applied magnetic field H (M vs. H).

1.2.1.1 Magnetization vs. Temperature Measurements

Magnetization temperature dependence is commonly used to investigate the mag-

netic behavior of nanoparticle systems. This is usually done in two modes, first
by freezing the sample without an applied field up to a low enough temperature
(usually a few Kelvin), then applying a very small magnetic field, followed by
measuring the magnetization at various increasing temperatures. This mode is called
zero field cooling (ZFC). Using the same procedure but this time cooling the
sample in the same applied field will give a slightly different curve, called field
cooling (FC). The reason for using such modes can be understood if we consider
that when our sample is cooled without a magnetic field (ZFC), nanoparticle’s
supermoments are frozen in random orientations. When a sample reaches the desired
temperature, a magnetic field of a few oersted is set. This magnetic field establishes
an orientation for those supermoments to follow. Because all supermoments are
frozen in fixed positions, the measured magnetization will be almost zero. Raising
the temperature will add thermal energy to our sample, so supermoments will have
more freedom to move, to follow the magnetic field orientation. In this way, the
measured magnetization will increase as more and more supermoments become
unfrozen in a process known as unblocking. This will occur until a majority of
the supermoments become unblocked. The temperature at which this happens is
called the blocking temperature, TB . Thermal energy introduces disorder because
the supermoments gain energy to flip between easy axes. The disordering of
more and more supermoments takes place when the temperature is raised, so the
measured magnetization decreases, following a Curie-type law. Although this type
of magnetism type is referred to as superparamagnetic, it is common to use the
term blocked state when the system is at temperatures below TB and to call it a
superparamagnetic state when it is at temperatures above TB . The functional form
for a system consisting of noninteracting nanoparticles is
  Z Z 1 
MS2 H m 1 T
MZFC .T/ D ln TB f .TB / dT B C f .TB / dT B : (1.8)
3Kef  T 0 T

Note that in the preceding function, MZFC (T) is dependent on the magnetization
saturation MS , applied magnetic field H, effective anisotropy Kef , the measurement
time  m , and the attempt time  0 , which is assumed to be on the order of
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 11

109 –1010 s, as mentioned earlier. Also, such a dependence was weighted using
a blocking temperature distribution, which of course is directly related to the size
distribution.
All real systems contain relatively larger and smaller particles, regardless of the
method used to fabricate them. This distribution of sizes usually follows a lognormal
or Gaussian function. This means that the exponential term in Eq. 1.3 is different,
not only because of the volume but because of the anisotropy. Most nanoparticle
systems have a lognormal size distribution. Because they are nonsymmetric, the
mean value is different from the median and mode values. Knobel et al. present an
interesting discussion on this point [4].
On the other hand, a large number of magnetic nanostructured systems have
been reported as superparamagnetic and studied in the framework of this theory.
Nevertheless, a lack of agreement between experimental data and standard super-
paramagnetic theory, as well as the obtained spurious parameters, is commonly
observed. Anisotropy constant K is generally calculated using the approximate
equation

Kef D 25kB TB =V; (1.9)

which derives from Eq. 1.3, in which are applied logarithms evaluated at the
blocking temperature TB . It is considered that a characteristic measurement time
for DC magnetometry lies around 100 s [4]. The first challenge, which involves
the use of Eq. 1.9, is to correctly determine the blocking temperature, for which
there is no established protocol, meaning some research groups use the maximum
of the ZFC curve, whereas others, knowing that the sample has distributed properties
(e.g., distribution of sizes, blocking temperatures, anisotropies), first determine the
blocking temperature distribution, generally through

d
f .TB / D .MZFC  MFC / ; (1.10)
dT

from which the mean blocking temperature hTB i can be extracted [26]. According
to the Ref. [27], the derivative method is the best for obtaining the blocking
temperature distribution [25].
We depict this process to illustrate the following remarks. Figure 1.6 shows
six different types of ZFC curves commonly seen in real granular materials.
Each curve corresponds to the physical situation described in Fig. 1.4. Here,
a, b, and c correspond to nonpercolated systems. The first one, a, is typical of a
noninteracting system. Characteristic features we can observe in such curves include
(1) a concavity at low temperatures, (2) a rounded peak, and, at T  Tpeak , (3)
a decrease in the magnetization with temperature, approximately following a 1/T
dependence.
For other interacting systems, the peak shifts to higher temperatures and the
high-temperature side of the curve departs from a Curie-type law at temperatures
close to that of the maximum, as seen in b. The third one, c, corresponds to a
12 L.M. Socolovsky and O. Moscoso Londoño

Fig. 1.6 The maximum of a ZFC is called the peak temperature, Tpeak . The irreversibility
temperature, Tirr , is also marked. (a) Magnetite nanoparticles dispersed in paraffin at very low
concentration. (b) Magnetite nanoparticles grafted with SiO2 ; dipolar interaction plays a role. (c)
Magnetite nanoparticles grafted with acetic acid, compressed in pellets. (d) Fe–Cu alloy, with
iron concentration slightly above the percolation limit. (e) Fe–Cu alloy, percolated; its blocking
temperature is above 300 K. FC was not recorded. (f) Fully percolated Fe–Cu alloy. A TB above
550 K is evident. The FC curve seems to show that a structural change could happen at those
high temperatures. Each panel can be viewed as corresponding to the situations in Fig. 1.4.
Measurements were done in a field of 20 Oe

strongly interacting system, but still nonpercolated. It consists of a broad, almost

rounded peak, with its maximum close to room temperature. Here the d, e, and
f plots correspond to increasingly percolated systems. Three features are seen on
these curves: (1) a concavity or an almost straight increase in the magnetization
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 13

until a plateau is reached, yielding to a characteristic “shoulder”, (2) an acute peak

at higher temperatures, and (3) a fall in magnetization at higher temperatures that do
not closely follow a Curie law. The shoulder actually does not have a straightforward
interpretation. Other examples of these situations can be seen in Fig. 1.8 of this
chapter or in Fig. 1.3 of Ref. [28].
An FC measurement is the complementary curve of a ZFC one. When we
measure in FC mode, because we freeze our sample with an applied field, a
nanoparticle’s supermoments tend to align with the lines of the external magnetic
field, which is easier when the interaction strength is weak. Thus, when we start
measuring at low temperatures, a nonzero magnetization is measured. It is observed
that when interactions are relatively stronger, the low temperature part of the FC
curve has magnetization values closer to that of the maximum in the ZFC curve. For
noninteracting systems the functional form for FC curves is
  Z T Z 1 
MS2 H m 1
MFC .T/ D ln TB f .TB / dT B C f .TB / dT B : (1.11)
3Kef 0 T 0 T

For systems close to or at percolation there are no mathematical expressions to

depict magnetic vs. temperature curves. Some attempts have been made, but no
mathematical models have been developed for such systems. Two studies published
by Tournus et al. demonstrate the impact of the size distribution and sweeping rate
of temperatures on ZFC FC curves and attempt to obtain a better description, despite
the curious crossover of the FC curve below ZFC, close to TB , which had not been
seen in prior experiments [29, 30].
Moreover, because of the size distribution, FC and ZFC curves diverge at tem-
peratures above TB , called the irreversibility temperature Tirr , if the size distribution
is very small, blocking temperature TB is almost equal to irreversibility temperature
Tirr , as was observed in a carefully prepared and measured system [31].
Magnetization is usually measured using coils, as in vibrating sample magne-
tometers, which use the movement of a magnetic sample to induce an electromotive
force (emf) on the detecting coils. These devices are able to measure with confidence
up to 105 emu. A useful, but expensive, device is a SQUID magnetometer, which
uses a Josephson junction to measure magnetization at the highest sensitivity. In
practice, it is possible to accurately measure magnetizations of 107 emu or even
less [32].

1.2.1.2 Curie-Type Behavior

At temperatures above TB , in which the magnetic moments are in the superparam-

agnetic regime, magnetization falls with increasing temperature. In a noninteracting
system, the magnetization follows a Curie law:

N2 H M 2 VH H
M .H; T/ D D s DC ; (1.12)
3kB T 3kB T T
14 L.M. Socolovsky and O. Moscoso Londoño

where H is the applied magnetic field, N D 1/V is the density of nanoparticles

per volume unit,  D Ms V is the sample magnetic moment, Ms is the saturation
magnetization, kB is the Boltzmann constant, and T is the temperature [4]. Dividing
M by H, we obtain the magnetic susceptibility . Because it is calculated from a
measured magnetization in a fixed field, it is called DC or static susceptibility (DC ).
Equation 1.12 can be rewritten as

M C
DC D D : (1.13)
H T
We can use Eqs. 1.12 and 1.13 to calculate the mean magnetic moment of each
particle, if we have their volume, by plotting the magnetization vs. the inverse
temperature. Alternatively, we can plot 1/M or 1/ vs. T to obtain a more viewable
graph (Fig. 1.7). In this case, the slope of the high-temperature straight part is

1 3kB
D 2 : (1.14)
C Ms V

For interacting systems, we can consider using a Curie–Weiss law,

M C
DC D D ; (1.15)
H T 

Fig. 1.7 Plot of inverse susceptibility vs. temperature of iron nanoparticles dispersed in paraffin;
the high-temperature part can be fitted to a Curie-type law
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 15

where  is a phenomenological parameter that takes into account the interaction

strength. In this case,  cannot be associated with the interpretation given for
canonical ferromagnets or antiferromagnets [32].
It is seen that a curve like that in Fig. 1.7 departs from a straight line near
the blocking temperature, that is, from a Curie behavior (“paramagnetic”). Such
a departure is more pronounced when interactions are stronger or there is a
considerable size distribution.
An alternative way to visualize these behaviors is to plot M T/H as a function of
temperature:

MT=H  C .1 C TCF =T/ : (1.16)

Magnetic behavior at high temperatures must be a constant (if TCF D 0) or a

monotonically decreasing function of T (if TCF ¤ 0). Here, C is the aforementioned
Curie constant and TCF is the ordering temperature of the ferromagnetic phase.
Such measurements must be taken using very small applied fields, lower than
the anisotropy field, to obtain the details of the curve [4]. Usually, researchers use
a magnetic field of 20 Oe. In higher applied fields, the peaks become wider and the
details are lost. Recall M vs. H curves. Our analysis is valid if we are in the straight
part of the initial curve, where it has a linear dependence on the applied field.

1.2.1.3 Susceptometry

A useful alternative to characterize magnetic granular materials is through a suscep-

tometer device. It makes it possible to measure the first derivative of magnetization
with respect to the applied magnetic field dM/dH. This quantity is the so-called
magnetic susceptibility, , also known as dynamical or AC susceptibility (AC ).
In such a measurement, in-phase (0 ) and out-of-phase (00) susceptibilities are
measured by applying an oscillating magnetic field H of frequency ! and recording
an induced voltage in pick-up coils proportional to  through a lock-in technique,
together with the dephasing angle ®.1 This is a measure of the ability of the system
magnetization to follow the oscillating magnetic field H. A larger 0 means that the
magnetization can respond to H. At the highest frequencies it becomes difficult for
the system to follow H. The dragging effect is reflected by 00.
The functional forms of these recorded quantities are as follows:

0 D  cos ';
(1.17)
00 D  sin ':

In a susceptibility vs. temperature measurement of a magnetic granular system,

0 has a shape closely resembling that of a ZFC curve, so our earlier depiction of the

1
A more advanced version of this technique is used in modern SQUID devices.
16 L.M. Socolovsky and O. Moscoso Londoño

Fig. 1.8 In-phase 0 (dots) and out-of-phase 00 (triangles) AC susceptibilities of a lightly
interacting system composed of packed silica-covered Fe3 O4 nanoparticles (upper left), a strongly
interacting system composed of Fe14 Cu86 (upper right), an almost percolating interacting system
composed of Fe17 Cu83 (lower left), and a percolated system composed of Fe30 Au70 (lower right).
High-frequency measurements are indicated by open symbols, low-frequency by black symbols

shapes of the curves can also be used (Fig. 1.8). The 00 curve shows an interesting
feature. Because this phase is representative of the energy dissipation in the system,
it gives an additional view on it. The maximum in 0 or 00 can be used as the
blocking temperature, but the blocking temperature is best represented by the peak
of 00 [33]. This peak of the 00 curve usually occurs at lower temperatures than the
peak of the 0 measurement. In a normal susceptometer, 00 is small compared with
0 . In adapted SQUID devices, this 00 signal is bigger. Based on the relationships
of both susceptibilities Cole–Cole plots, information about the frequency attempt  0
and energy barrier distribution can be obtained. A useful article on the underlying
physics is that of García Palacios [34]. A measurement of 00 vs. temperature in
strongly interacting systems, even percolated ones, displays a peak that more or less
still corresponds to the effect of individual nanoparticles.
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 17

1.2.1.4 Néel–Arrhenius and Vogel–Fulcher Laws

Using Eq. 1.3 we show how energy barriers affect relaxation times. This equation is
known as the Néel–Arrhenius (N–A) law [35]. The relaxation time  for a system of
noninteracting superparamagnetic particles follows that law, which we repeat here
to compare with the Vogel–Fulcher (V–F) law in this section:
 
Kef V
 D  0 exp : (1.18)
kB T

As explained earlier, Kef is the effective anisotropy constant, V is the volume

of the nanoparticle, and  0 must take values between 109 s and 1011 s for
superparamagnetic systems. However, only in a few systems, which are closest
to the ideal, the N–A model works in an appropriate way, for example, systems
reported by Torres et al. [13]. In this report, the authors present the validity of
the N–A model for highly anisotropic Cox Fe3  x O4 nanoparticles, from which are
obtained realistic values for the response time. Nevertheless, in most published
papers, when the N–A law is applied on a nanoparticle system, nonphysical values
of  0 are obtained. This behavior is commonly attributed to nanoparticle interaction
effects. One of the first works to address this challenge was that of Shtrikmann and
Wohlfarth [35], who used a mean field approximation to propose a V–F law. Such
model establishes that the magnetization processes are perturbed by a dipole field
emerged from the interactions among particles. This perturbation is immersed in a
fictitious temperature T0 and expressed as
 
Kef V
 D  0 exp (1.19)
kB .T  T0 /

Despite the fact that the V–F model does not take into account the surface effects
or polydisperse properties present in all real nanoparticle systems, the use of this
model provides a more physically relevant result. Table 1.1 presents the values of  0
obtained for some systems composed of magnetic nanoparticles by means of N–A
and V–F laws.
As can be seen in Table 1.1 and in most cases where nanoparticle interactions
are relevant, the response time takes nonphysical values when the N–A law is
used. Despite the fact that these values are physically incorrect, it is interesting
to observe that in interacting or weakly interacting nanoparticle systems, there is
a clear tendency to underestimate these parameters. However, such values can be
corrected by adopting the Vogel–Fulcher expression from which values ranging
between 109 and 1011 s are obtained.
18 L.M. Socolovsky and O. Moscoso Londoño

Table 1.1 Values of  0 obtained by applying both Néel–Arrhenius and Vogel–Fulcher models
on granular materials composed of monodomain magnetic nanoparticles
Sample Mean NP size (nm)  0 using N–A law (s)  0 using V–F law (s) Reference
”-F2 O3 4.9 – 2.5  1011 [36]
Fe3 O4 3.8 4.9  1036 4.7  1010 [37]
4.5 1.6  1016 3.2  1010
Fe3 O4 5.5 1032 1013 [38]
”-F2 O3 (powder) 4 3  1018 5.2  107 [39]
”-F2 O3 (compacted) 1.6  1042 9.7  108
Ni (1.9 wt%)-SiO2 4.3 3.4  1010 1.7  109 [40]
Ni (2.7 wt%)-SiO2 5 2.9  1011 2.1  1010
Ni (4 wt%)-SiO2 4.9 4.2  1011 1.4  1010
Ni (7.9 wt%)-SiO2 5.3 1.3  1012 1.2  1010
Ni (12.8 wt%)-SiO2 5.5 5.8  1015 6.3  1010
Fe3 O4 4.9 – 6  106 [41]
CoO-Pt core-shell 4 6  1019 2.4  1011 [42]
Fe3 O4 7.5 9.5  1010 9.8  1010 [43]
Au-Fe3 O4 dimer 11 1.8  1016 1.4  109
Au-Fe3 O4 core-shell 15 5.9  1014 8.9  109
CoFe2 O4 4.5 6.8  1027 – [44]
6.3 1.5  1032 –

1.2.1.5 Nonlinear Susceptibility

A special analysis of nonlinear susceptibilities sheds light on systems that can

be considered to be of the spin-glass type. To do that, AC susceptibility must be
measured at different driving fields. Each measurement must be fitted with a poly-
nomial expression to extract linear and nonlinear terms. In this case, magnetization
is written

M D M0 C 1 H C 2 H 2 C 3 H 3 : (1.20)

In this kind of study, higher-order terms of susceptibility are analyzed. The peak
in 3 corresponds to a blocking temperature or spin-glass transition temperature
[12, 32].
Another alternative that allows for a different way to study magnetization in a
sample is provided by Mössbauer spectroscopy, which is based on the resonant
absorption on certain isotopes. The effect is quite remarkable on iron, so for many
magnetic materials it is a very useful technique. One feature that a Mössbauer
experiment can measure is the hyperfine magnetic field, Bhf , which is a measure
of the internal fields sensed at the nuclear level. Bhf is seen as a sextet of peaks in
an absorption vs. energy graph. It has been observed in granular systems that such
a sextet collapses to a singlet or doublet when the temperature is raised. Because
the dominant frequency corresponds to a Larmor precession, which in this case
corresponds to that of the nuclear magnetic moment with the surrounding magnetic
field, we can make a distinct measure of a sample’s magnetization. By measuring
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 19

changes in absorption at an energy corresponding to one of those peaks, we see the

transition from blocked to superparamagnetic states as a function of temperature.
Because the Larmor frequency for the iron isotope used in this spectroscopy, Fe57 ,
is   1010 Hz, an additional frequency is available to study the dependence
of the blocking temperature, as used in a pioneer work by Dormann et al. [45].
This technique is called Mössbauer thermal scanning (MTS) [46, 47]. Without
appropriate equipment, this measurement can be made simply by taking Mössbauer
spectra at different temperatures. An extended criterion says that when half of the
sextet area turns into a singlet or doublet, this is the blocking temperature [48, 49].
Studies on blocking temperatures can be used to plot a magnetic phase diagram,
which can help to understand magnetic alloys [49, 50].

1.2.1.6 Magnetization vs. Magnetic Field Measurements

Magnetization M vs. applied magnetic field H is a customary measurement for

characterizing magnetic materials. In granular systems, this measurement shows
a characteristic S-shaped curve. Depending on the temperature at which this
measurement is taken, it either displays hysteresis (at T < TB ) or it does not (at
T > TB ). For noninteracting systems, M vs. H curves will collapse into a single
curve if the magnetization is plotted against H/T [4, 51]. Conversely, if they do
not collapse into the same curve, the system has nonnegligible interactions. On
a real granular system, we must take into account the magnetic response of the
matrix. High field values follow an asymptotic curve if the magnetic response of
the matrix is not important. The nonsaturation effect is due to uncompensated spins
on the surfaces of nanoparticles [11, 52]. If the magnetic response of the matrix is
important, if it is paramagnetic or diamagnetic, the effect on the M vs. H curve is
evident at high fields, so to analyze separately the contributions of nanoparticles, we
must subtract a straight line using the extreme parts of the curve at the highest fields.
A positive slope that changes with temperature indicates a paramagnetic matrix. A
negative slope, which does not change with temperature, indicates a diamagnetic
matrix. We must also take into account that some matrices can be composed of
mixtures of magnetic materials, so mixed behaviors should be expected. In addition,
recall that all materials have a magnetic response (Fig. 1.9) [2, 3, 7, 53].

1.2.2 Models Commonly Used to Describe Interactions in Sets

of Nanoparticles

There are commonly two schools of thought when it comes to analyzing magnetic
interactions in systems of nanoparticles. One of those schools is based on the
Dormann–Bessais–Fiorani [54] model, which uses energy barriers to describe the
effect on a nanoparticle of a full ensemble; the other school uses the Mørup
20 L.M. Socolovsky and O. Moscoso Londoño

Fig. 1.9 M vs. H curves measured at 5 and 350 K of a sample composed of iron nanoparticles in
a SiO2 matrix. Both measurements (which are S-shaped) were taken at temperatures far from the
blocking temperature (TB  35 K). The 5 K curve (up black triangle), which is in the blocked
state, shows an important coercive field, while at 350 K (down red triangle), which is in the
superparamagnetic state, it shows little coercivity. Saturation magnetization is lower in the high-
temperature measurement, as it should be

model [55], in which a mean field approximation is proposed. Both models are
based on approximations commonly used in physics and are discussed in detail
in Ref. [56]. The first one seems to represent experimental results better than the
second. But in more interacting systems, a third approach, developed by Allia et al.,
the so-called T* model, has shown better depictions of real systems [23, 57].
This model, sometimes called the T-star model or interacting superparamagnetic
model (ISP), was proposed as a phenomenological approach that treats nanoparticle
magnetic moments as interacting through dipolar-type long-range fields, and the
overall effect can be modeled by a fictitious temperature, T*, which should be added
to the real temperature in the denominator of the Langevin function argument. The
new temperature, called the apparent temperature, has the effect of slowing the
approach to saturation and can be written TA D T C T*, where T* is related to
the dipolar interaction energy through "D D kB T*. Then the modified Langevin
function has the form
Z 1  
CO H
M .H; T/ D nCO CO L f .CO / dCO ; (1.21)
0 kB .T C T  /
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 21

where the suffix CO refers to corrected values. To solve the modified Langevin
equation, it is necessary first to determine T*, which can be accomplished by starting
with the low-field inverse susceptibility, given by
 
 T
D 3kB NCO C ˇ; (1.22)
 .ıM s /2

where NCO is the corrected number of magnetic moments of superparamagnetic

particles per unit volume, which is related to the number of magnetic moments of
superparamagnetic particles per mass unit through NRE D nRE ı, with ı being the
bulk density of the magnetic nanoparticle.  is a parameter introduced to consider
the nanoparticle size distribution, which can be determined from the relationship
among average values:
˝ 2 ˛ ˝ 2 ˛
CO AP
D D ; (1.23)
hCO i2 hAP i2

where AP is the magnetic moment magnitude obtained from the standard Langevin
equation, called apparent because its values are screened by the nanoparticle
interaction effects. The susceptibility  is obtained from the low-field region of M
vs. H curves, and  is estimated from the lognormal distribution obtained from the
parameters and hAP i. By plotting / vs. T/(ıMs )2 and following the relationship
between apparent and real values shown in Ref. [23], one obtains ˇ and, therefore,
T* [23].
But even this model fails when systems are close to the percolation limit [57, 58].
As discussed earlier, in almost and fully percolating systems, magnetic behavior
is complex. Particularly since the publication of the aforementioned works of
Dormann–Bessais–Fiorani and Mørup, many experiments have been conducted and
theoretical developments made on mildly interacting systems, but few advances
have been made in theoretical models that can accurately describe systems that are
close to percolation or fully percolating.

1.2.3 Experiments on Granular Materials with Controlled

Interactions
1.2.3.1 Dispersions of Nanoparticles in Polymers

These are a class of magnetic granular systems made of magnetic nanostructures,

such as nanoparticles, embedded in a nonmagnetic and viscous matrix. Given
that in this kind of nanomaterial it is relatively simple to adjust the nanoparticle
concentration, and therefore interactions between them, these materials are excellent
candidates for experimentally studying some fundamental aspects of magnetism at
the nanoscale.
22 L.M. Socolovsky and O. Moscoso Londoño

1.2.3.2 Experimental Evidence of Dipolar Interaction Influence

on ZFC-FC Curves

In this section, we discuss how the strengthening of dipolar interactions among

magnetic nanoparticles significantly affects the shape, and therefore the obtained
parameters, of ZFC-FC curves. To this end, we have prepared two sets of magnetic
nanocomposites based on citric acid–coated magnetite (Fe3 O4 ) nanoparticles, which
are distributed on a nonmagnetic matrix of polyvinyl alcohol (PVA). For the first
system (labeled A), we used nanoparticles with 7 nm mean diameter, while in
the second one we used nanoparticles of 10 nm (labeled B). For the details on
nanoparticle synthesis, please see Refs. [59, 60]. In both systems, the PVA was
loaded with 0.5 wt.%, 3 wt.%, 15 wt.%, and 30 wt.% of citric acid–coated magnetite
nanoparticles (magnetic nanocomposite preparation details are available in Ref.
[60, 61]).
As explained earlier and shown in the examples of Fig. 1.6, we see a similar
behavior in this experiment (Fig. 1.8). It is possible to observe that by increasing
the concentration of magnetic nanoparticles, and thereby strengthening the dipolar
interactions among magnetic nanoparticles, the blocking temperature shifts to
higher temperatures. Often this type of shift is used to compare the magnetic
properties of nanostructured systems composed of nanoparticles of different sizes
or phases, and the observed differences, such as the aforementioned blocking
temperature displacements or broadening of the ZFC curve, are associated with the
intrinsic nanoparticle size distribution, surface effects, or anisotropies of diverse
origins, to mention just a few. However, in nanostructured systems based on the
same magnetic constituent (the same magnetic nanoparticles in our experiment),
the experimental dissimilarities must be based on other phenomena, like dipolar
interactions or clustering effects. It is important to mention that proximity effects
can alter the anisotropy of individual nanoparticles or even form smaller regions of
different anisotropies [61] (Fig. 1.10).
We used a derivative method to obtain the blocking temperature distributions
of each sample to get a mean value of TB . We found that for both systems,
each synthesized using the same batch of magnetic nanoparticles, different mean
blocking temperature values were obtained (Fig. 1.11).
Using Eq. 1.9, allows us to obtain the effective anisotropy constant K. Then,
using 7 and 10 nm as mean diameters to find the nanoparticle volume and using the
mean blocking temperature hTB i, different values of K are obtained for each sample
(Table 1.2). For example, we find for system A, K D 9.03  104 J/m3 for a 0.5 wt.%
sample or K D 2.6  105 J/m3 for 30 wt.% sample, while for system B, loaded with
10 nm-Fe3 O4 nanoparticles, values of K D 3.22  104 J/m3 for a 0.5 wt.% sample
or K D 5.6  104 J/m3 . Differences between the anisotropy values of a particular
system should not exist because anisotropy is an intrinsic property that must remain
unaltered for a non-interacting nanoparticulated system. However, as can be noted
for the PVA loaded with 7 nm Fe3 O4 , differences in almost one order of magnitude
are observed. Regardless of whether the ZFC maximum is used as the blocking
temperature, the dissimilarities among the obtained K values are larger (Table 1.2).
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 23

Fig. 1.10 Zero field cooling and FC curves obtained for both PVA loaded with (left) 7 nm and
(right) 10 nm magnetite nanoparticles. The effect of interactions can be seen as the nanoparticle
concentration increases. Powder samples are composed of single nanoparticles in close contact,
with only the coating of citric acid separating them

Now, the question arises as to what the correct anisotropy value is, and what is its
real dependence with magnetic strenght. We have magnetite nanoparticles in which
the effective anisotropy’s largest contribution comes from the magnetocrystalline
anisotropy (due to the spin–orbit interaction). In addition, recall that in the presented
systems the magnetic nanoparticles are not percolated. We believe that the value
that is closest to that of a nanoparticle alone should come from a system in which
the magnetite nanoparticles are in the lower interacting state. This means that
interactions modify the anisotropy constant, as was observed by Ferrari et al. [62].
24 L.M. Socolovsky and O. Moscoso Londoño

Fig. 1.11 Blocking temperature distribution obtained for both PVA loaded with (left) 7 nm and
(right) 10 nm magnetite nanoparticles. Symbols: obtained from derivative method. Straight line:
lognormal fit from where the mean blocking temperature TB and the lognormal standard
deviation were determined

Note that when anisotropy constants are studied or compared, careful attention
should be paid specifically to the dilution conditions, besides the size and grafting
of the nanoparticle.

1.2.4 Comment on Spin-Glass Systems

A different way of building systems similar to those we have reviewed so far is by

using a spin-glass system. A spin glass is formed when magnetic ions are randomly
dispersed in a nonmagnetic matrix. At very low concentrations of magnetic ions
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 25

Table 1.2 Anisotropy values obtained for both systems using hTB i through the derivative method,
KhTBi (column 4), maximum of ZFC curve, KT-ZFC (last column)
NP mean size wt. % of NP hTB i (K) KhTBi (104 J/m3 ) TZFC (K) KT-ZFC (104 J/m3 )
7 nm (system A) 0:5 47 9:03 80 15.3
3 48 9:22 92 17.6
15 49 9:41 115 22.1
30 135 26:0 148 28.4
10 nm (system B) 0:5 49 3:22 125 8.24
3 66 4:35 150 9.88
15 71 4:68 236 15.5
30 86 5:66 240 15.8

(less than 1 at.%) some features appear that resemble those of superparamagnetic
systems, like a cusp in a magnetization or susceptibility vs. temperature measure-
ment, or irreversibilities in a ZFC-FC curve. The origin of that magnetic order
is frustration because of the competing interactions among magnetic ions, like
RKKY and others mentioned earlier, which makes it different from the thermally
activated process that is SPM [8, 16, 63]. When the concentration of magnetic atoms
is increased, some ions will be close enough to establish direct interactions like
direct interchange, super interchange, or Dzyalonskii–Moriya interchange. Those
interactions are stronger than RKKY, so the magnetic landscape becomes more
complicated. If we look at the conformation of the sample, at a concentration of
atomic ions of a few atomic percent, we see that those that are in close contact can
be regarded as nanoparticles, despite their origin. Those regions in close contact
have interactions like direct interchange or super interchange, as in a regular
nanoparticle. This is called cluster glass (CG) [64, 65]. For face-centered-cubic (fcc)
arrangements, percolation takes place at concentrations of approximately 17 at.%
for ferromagnetic ordering and 45 at.% for antiferromagnetic ordering. Chains of
interconnected magnetic ions are distinguishable, which produces regions of fer-
romagnetic (or antiferromagnetic) order. This state is called quasi-ferromagnetism
(QFM). They used to be called mictomagnetic (mixed magnetism) because it was
believed that they were mixtures of antiferro and ferromagnetic interactions [64,
65]. In addition, the term reentrant magnetism is also used. Earlier, discussions took
place about the physical situation for this state, but now the state is interpreted as a
transverse component freezing of the magnetization vector [66].
Because in some systems it can be difficult to adequately characterize an
unknown material and determine whether it is a spin-glass or superparamagnetic
system, we can use a criterion that takes into account blocking (or glassy) tempera-
tures and frequencies, obtained in susceptibility vs. temperature measurements [16].
Such a criterion is known as the Mydosh parameter ˚ [16] and can be calculated
from the maximum of the imaginary part of AC susceptibility (TM ), according to
the equation
26 L.M. Socolovsky and O. Moscoso Londoño

Fig. 1.12 Magnetic phases when magnetic atom concentration is increased. At very low concen-
trations of less than 1 at.%, we have canonical spin glasses. Increasing the concentration produces
a phase where spin-glass structures coexist with short-range magnetic order (SRO), such as direct
interchanges between neighboring magnetic atoms. Above 10 at.%, bigger SRO regions appear,
forming clusters that can be viewed as nanoparticles. Crossing the percolation limit, we encounter
a long-range magnetic order (LRO) that causes ferromagnetic features in the magnetic behavior

TM
ˆD ; (1.24)
TM log .!/

where TM is the frequency shift for the temperature of that maximum, and ! is
the frequency of the measurement [16]. Values of ˚ < 0.018 are found for canonical
spin glasses. Higher values, on the order of 101 , are typical of superparamagnetic
systems. In almost and fully percolated systems, ˚ values are small, independently
of whether the system is a spin glass or SPM in origin (Fig. 1.12).
These systems have been studied systematically since the 1960s. Interestingly,
one of the most studied systems is Au: Fe, both of whose elements are metallic.
Since they do not form alloys, they are prone to segregation, so a tendency to form
aggregates of pure iron is expected. The enthalpy of formation is positive [67]. At
room temperature and concentrations above 1 at % it is not difficult to imagine that
nanoparticles of iron can form from the migration of isolated atoms. Experiments
on such a system started as early as the 1930s, but the term spin glass was coined
only in the early 1970s. The subject attracted the attention of theoreticians who
developed models to describe these systems [12].
An interesting experiment on a special type of nanoparticle was performed
by Nunes et al. [68]. They prepared by chemical synthesis nickel nanoparticles
with an almost amorphous internal atomic structure. Those nanoparticles have an
approximate diameter of 12 nm, but inside, the atoms are highly disordered, forming
small clusters of just a few atoms. This leads to a magnetic behavior that is closer
to a cluster-glass system, despite the fact that it is formed by very well-defined
nanoparticles. In this experiment, ZFC-FC measurements showed a maximum close
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 27

to 20 K that did not change if higher fields were used, at least up to 1000 Oe, a
characteristic of spin glasses. AC susceptibility measurements showed, by applying
the Mydosh criterion, that the system could be treated as a spin-glass one ( D 0.01),
so it must be regarded as a cluster glass. The same researchers also investigated
smaller Ni nanoparticles, approximately 5 nm, prepared using a similar synthesis
route [69]. They found that SPM order no longer existed to give rise to a mixture
of paramagnetic and ferromagnetic orders, again despite the existence of very
well-defined nanoparticles. An important increase in surface anisotropy was also
registered.

1.3 “Giant” Magnetotransport Properties

Magnetic granular materials display interesting magnetotransport properties: giant

magnetoresistance (GMR), tunneling magnetoresistance (TMR), and the giant Hall
effect. All three effects arise as a consequence of the interaction between electrons
and a nanoparticle’s magnetic moment.

1.3.1 Giant Magnetoresistance

Magnetoresistance is the effect of changing electrical resistance when a magnetic

field is applied. Electrical resistance has different origins. One of them is the
scattering of electrons due to magnetism. It is an effect with wide technological
applications, from magnetic wires (used in black boxes as a rough way to record
crew conversations) and flexible ribbons (such as those used in cassette tapes)
to hard disk heads. When produced by spin-dependent scattering on nanostruc-
tured magnetic entities, like coupled multiple layers of nanometric thicknesses or
nanoparticles in a matrix, the change is negative and it is called GMR. A negative
change means that the electrical resistance falls when a magnetic field is applied.
In experiments, this change is usually calculated through

ŒR .Hmax /  R.0/
GMR D  ; (1.25)
R.0/

where R(Hmax ) and R(0) are the electrical resistances measured at the maximum
magnetic field Hmax and at the zero field.
Alternatively, we can use R(Hc ) instead of R(0) if the coercive field is well
marked or if it is large, as is the case for multilayered materials. The minus sign
means only that the change is negative, and sometimes it is calculated without it, as
shown in Fig. 1.13.
Giant magnetoresistance is produced by differential spin-dependent scattering
due to the relative orientation of moments of magnetic entities, as in sandwiched
28 L.M. Socolovsky and O. Moscoso Londoño

Fig. 1.13 Nanoparticles with

a size of D, at a distance d.
This distance must be equal
to or less than the electron
correlation length,

multiple layers [70] or granular materials [71]. In granular materials it is isotropic,

and in multiple layers it depends on the relative orientation of the sandwiched layers
to the electrical current. An electrical current can be considered as being divided into
two subbands, each corresponding to a spin orientation, parallel or antiparallel to
the local magnetization of the entity (magnetic nanoparticle or layer). If two entities
that are close enough to allow electrons to retain their coherence (spin orientation)
have their moments aligned in an antiparallel orientation, the subband of electrons
with spins in the same orientation of the magnetization of the first entity (“spin-
up”) will not experience magnetic scattering when passing through the first one.
When a current that does not lose its spin coherence encounters a second entity,
it will experience magnetic scattering because their moments are not parallel. For
another subband composed of electrons with spins in an antiparallel orientation
(“spin-down”), the situation will be the opposite: it will suffer magnetic scattering
when passing through the first entity, but not the second one, which is parallel to it.
If we apply an external magnetic field, the entities will tend to align, increasingly
so if the strength of the field is raised. With this magnetic landscape, a spin-up
current will not suffer scattering in the first or second entity. A spin-down current
will experience scattering of magnetic origin in both entities. We can see that the first
configuration (parallel–antiparallel entities) has a higher resistance than the second
case (parallel–parallel) and intermediate resistances with partially oriented entities.
Usually, both situations are compared by solving a simple circuit by Kirchhoff laws.
The effect on the measurement is depicted in Fig. 1.12.
The GMR effect was first identified and explained in multiple layers, so the
theory differentiated two geometrical situations, current in plane (CIP) and current
parallel to the plane (CPP), depending on the relative orientations of the current
and the plane of the layers. In granular materials there is no such geometrical
situation. GMR in granular materials was modeled theoretically in Ref. [72]. The
basic condition for GMR is that the electron correlation length, , should be at least
equal to or larger than the interparticle distance, d. Polarization of the spin current
is kept through jumps, as explained earlier. This condition of  d or larger says
that the effect will be observable in enough concentrated granular materials. More
in-depth models were discussed in the 1990s.
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 29

GMR was observed and interpreted first in multiple Fe-Cr layers in two
contemporary experiments done almost simultaneously by the groups of Albert Fert
and Peter Grünberg [73, 74]. Fert had already studied theoretically and developed
a two-subband model for the resistivity in ferromagnetic metals [75]. Experimental
realization of those sandwiches with thicknesses of just a few nanometers was only
possible when MBE machines were available, in the late 1980s. GMR in granular
materials was first reported by Chien in 1992 [76], but it had been previously
observed by other researchers, as noted even in that article. Systems that are
regarded as canonical spin glasses, like Au-Fe, show negative magnetoresistance.
Nigam and Majumdar, in an article published in 1983, analyzed magnetoresistance
measured in several samples of canonical spin glasses, in concentrations of magnetic
solute that reached approximately 10 at.%, which cannot be considered a diluted
magnetic sample (see our earlier discussion on the magnetism of spin glasses) [77].
Some results on magnetoresistance on diluted alloys that form spin glasses were
collected in Refs. [78, 79], and in Ref. [80]. The first report on a so-called negative
magnetoresistance on a spin glass–like system, at least from what we investigated,
was reported by Nakhimovich in 1941, in a Soviet physics journal [81].
In granular systems, GMR depends on the relative orientation of nanoparti-
cles’ supermoments (Figs. 1.13 and 1.14). When a magnetic field is applied,
supermoments align. The greater the field strength, the greater the alignment of
supermoments and, hence, the lower the electrical resistance [72, 82]. An interesting
theory of GMR in granular materials was presented in an article by Ferrari et al.,
where spin-dependent scattering was taken into account within nanoparticles and
their boundaries (Fig. 1.15) [83].
At low nanoparticle concentrations, GMR scales with the square of magneti-
zation. In concentrated systems, as an effect of increased magnetic interaction,
this relationship no longer exists. In such systems it is seen, in low fields, that
magnetoresistance vs. magnetization parabola flattens [84]. This is clearly seen in
the article of Liu et al., where 8–9 nm magnetite nanoparticles were compressed
into pellets, to show the GMR. This effect does not scale with M2 , as is clearly seen
in Fig. 1.15. It is also seen in Refs. [84–85].
The effect is greater when NP concentration is close to the percolation limit.
When this limit is crossed, mixed effects take place. A lower GMR effect can
be recorded, owing to the remaining nanoparticles that are still relatively free to
orientate with the field and the chains of nanoparticles that have ferromagnetic
interactions among them [86].

1.3.2 Tunneling Magnetoresistance

TMR is a cousin effect of GMR. It also undergoes a negative change in electrical

resistance when a magnetic field is applied. In granular materials it appears when
the matrix is a dielectric, preventing electron transfer between nanoparticles, so
no exchange forces are present. There are only magnetostatic forces between
30 L.M. Socolovsky and O. Moscoso Londoño

Fig. 1.14 Alignment of a nanoparticle’s supermoment with an external magnetic field causes
electrical resistance to drop

nanoparticles that act at distance. Instead of an electrical current formed by freely

moving electrons, the formation of current depends on the electrons’ tunneling
between adjacent magnetic entities (again, sandwiched layers or nanoparticles),
which changes the relative orientation of their magnetic moments by the application
of an external magnetic field. The population of subbands at the Fermi level of each
magnetic entity changes.
Jullière discovered this effect in multiple Fe-Ge layers in 1975, years before
GMR was discovered [87]. The electrical properties of granular materials with
dielectric matrices were extensively studied by Abeles et al. and reported in a
seminal article in 1975 [88]. Interesting theoretical analyses have been reported
since the development of granular materials with nonmagnetic matrices [89, 90].
TMR is usually calculated as

ŒR .Hs /  R .Hc /
TMR D  ; (1.26)
R .Hc /

where R(Hs ) and R(Hc ) are respectively the resistances measured at the magnetic
saturation Hs and coercive field Hc . Alternatively, we can use R(0) instead of R(Hc )
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 31

0
Fe10Cu90
MR [%]

-1 0 1
H [T]

3
0
Fe19Cu81
2

MR [%]
MR [%]

1 0

-1 0 1 0 10 20 30 40 50
H [T] Fe concentration [at %]

Fig. 1.15 Magnetoresistance of a Fe-Cu alloy prepared by mechanical alloying, measured at 77 K.

(Left, upper panel): Fe10 Cu90 ; (left, lower panel): Fe19 Cu81 , note the rounded curve at low fields.
Right panel: GMR increases with increasing concentration of the magnetic material, up to the
percolation limit, around 20 at.% in this case [51]

if the coercive field is not well defined or is small, and both cases are common in
granular materials at room temperature or with a wide size distribution, as is the
case with GMR presented earlier.
In Jullière’s model, TMR is related to the polarization, P, of spins of the tunneling
electrons by

P2
TMR D  : (1.27)
1CP2

Similarly to what happened with GMR, TMR was measured first in multilayered
systems, followed by granular materials. Impressive values have been recorded,
sometimes greater than their multilayered counterparts [91]. Some groups have
measured large TMRs at room temperature in granular samples of Fe3 O4 in
polymeric matrices [92, 93].
Magnetoresistance has been found in nanoparticles grafted with oleic acid, a
system that can also be considered granular. Those experiments were based on
the fact that grafting prevents direct contact between nanoparticles and retains a
distance of around 2–4 nm. By thermal annealing or chemical treatment, the distance
between nanoparticles can be reduced to less than 1 nm [94–97]. Unfortunately, in
those functionalized nanoparticles grafting seems to be different if nanoparticles are
close enough, so in concentrated samples nanoparticles have a tendency to form
chains, as was observed experimentally [98].
32 L.M. Socolovsky and O. Moscoso Londoño

1.3.3 Giant Hall Effect

In 1995 Pakhomov et al. discovered a large extraordinary Hall effect in granular

materials [99]. A Hall effect is measured using the common Hall setup. It is
recorded by measuring the voltage perpendicular to the direction (Hall voltage) of
an injected electrical current, with an applied magnetic field H perpendicular to the
plane defined by both quantities. The initial curve resembles that of a hysteresis.
Hall resistivity H is related to H (through magnetic induction B) and sample
magnetization M through the empirical expression

H D R0 B C REHE 0 M; (1.28)

where R0 is the ordinary Hall coefficient, and REHE is the extraordinary one [100].
The first term is the usual effect that appears as a consequence of the Lorentz force
that is exerted by the external magnetic field on charge carriers. The second term is
the extraordinary part, which is due to the scattering of the polarized conduction d
electrons and in granular systems can be very large (Fig. 1.16).
The sources of this effect are not clear. The effect is associated with spin-
dependent scattering at nanoparticle surfaces. This breaks the spatial symmetry on
the trajectories of charge carriers and produces two effects, skew scattering and a
side jump [101]. Skew scattering is a constant deviation of the trajectories of charge
carriers [102]. It is usually described by

REHE D A C B2 : (1.29)

Fig. 1.16 A typical Hall resistivity curve. The low-field part is almost straight, in which the
contribution of the first term in Eq. 1.28, which is the ordinary part, contributes slightly to the Hall
resistivity. Thus, the slope of this part is almost the extraordinary contribution, and REHE can be
extracted from this value. The high-field part, where magnetization is reversible, is almost straight,
so a second slope can be calculated. From that, the ordinary Hall resistivity can be calculated
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 33

Frequently the second term is dropped, so it is accepted that REHE has a linear
dependence on . At low temperatures or in metals, where resistivity is small, skew
scattering is usually considered the main mechanism for this extraordinary Hall
resistivity [101].
The second mechanism is the side jump, a lateral quantum displacement of
charges, which results in a quadratic dependence on  of REHE [101, 103]. This
mechanism is usually considered in concentrated granular materials, and it is
important at high temperatures.
Thus, it is considered that a measurement of the extraordinary Hall resistivity
should scale with ordinary resistance to a power of 1 or 2, depending on the
mechanism. A shorter expression is used:

REHE D ˛n : (1.30)

For several systems is measured an intermediate power law, which is indicative of

a mixed mechanism. But in other systems, a power law with n > 2 arises, indicating
that a third mechanism for the extraordinary Hall resistance is possible [101]. The
GHE seems to depend also on solitary atoms that might exist between nanoparticles
or nanoparticle surface roughnesses [101, 104, 105].
By annealing granular samples of metallic nanoparticles embedded in a silica
matrix, detectable changes in the extraordinary Hall resistivity have been measured.
It was observed that the GHE diminishes for annealed samples, which leads to a
connection of this effect with structural characteristics in which, because of thermal
treatment, surface roughness smoothes and possible isolated atoms have enough
energy to migrate. In this way, possible sources for the large values of this parameter
have been considered, such as metallic atoms forming chains or roughness in
nanoparticle surfaces [104, 105]. In granular materials close to percolation there
are regions or large chains of nanoparticles that short-circuit the passage of charges,
and they do not contribute to spin-dependent scattering [106].
The GHE has been reported in granular materials produced by magnetron
sputtering [105–107]. It is also seen in magnetoresistance measurements as an
imbalance between negative and positive magnetic fields’ semi-axis.
All these so-called giant effects are stronger when the granular system is close
to the percolation limit. After this point, because of increasingly ferromagnetic-like
interactions among nanoparticles, those effects diminish [28, 86, 106–108].

1.4 Conclusions

Granular materials show features that are interesting for both academic studies
and technological applications. A basic set of characterizations can be made using
equipment available in several laboratories. Zero field cooling and field cooling
magnetization vs. temperature measurements provide important information, in
addition to unambiguously showing the existence of nanoparticles and an approx-
34 L.M. Socolovsky and O. Moscoso Londoño

imate idea of their sizes. In particular, susceptibility measurements are highly

valuable for studying granular materials. It should be clear that to determine the
properties of individual nanoparticles, they must be measured in dilutions. Spin-
glass systems in medium and high concentrations share characteristics with granular
materials. We propose the use of the term interacting superparamagnetism (I-SPM)
instead of SSG and percolated superparamagnetism (P-SPM) instead of SFM.
There exists an intimate relationship between the superparamagnetic characteristics
of granular materials with so-called giant magnetotransport properties, known as
giant magnetoresistance, tunneling magnetoresistance, and the giant Hall effect.
Actual knowledge of the magnetic and magnetotransport properties points to the
need for updated experiments with much better controlled structures. Studies on
magnetic granular systems must be conducted with an eye toward their structural
properties owing to their effect on magnetic and magnetotransport properties.

Acknowledgements LMS wants to acknowledge ANPCyT and CONICET (Argentina), CNPq,

and FAPESP (Brazil). OML acknowledges FAPESP Grant 2014/26672-8 and COLCIENCIAS for
the fellowship awarded from 2010 to 2014. LNNano-CNPEM (Campinas, Brazil) is acknowledged
for the transmission electron microscopy images taken there with a JEOL 3010 URP and a 2100
MSC TEM microscopes.

References

1. Fert A (2008) The present and the future of spintronics. Thin Solids Films 517(1):2–5
2. Martin DH (1967) Magnetism in solids. Iliffe Books Ltd, Londres
3. O’Handley RC (2000) Modern magnetic materials: principles and application. Wiley, New
York
4. Knobel M, Nunes WC, Socolovsky LM, De Biasi E, Vargas JM, Denardin JC (2008)
Superparamagnetism and other magnetic features in granular materials: a review on ideal
and real systems. J Nanosci Nanotech 8:2836
5. Bedanta S, Petracic O, Kleemann W (2015) Supermagnetism. In: Buschow KHJ (ed)
Handbook of magnetic materials. Elsevier B.V., Amsterdam
6. Guimarães AP (2009) Principles of nanomagnetism. Springer, Berlin
7. Blundell S (2001) Magnetism in Condensed Matter. Oxford University Press, London
8. Hurd CM (1982) Varieties of magnetic order in solids. Contemp Phys 23(5):469
9. Aharoni A (1996) Introduction to the theory of ferromagnetism. Clarendon Press, Oxford
10. Bedanta S, Kleemann W (2009) Supermagnetism. J Phys D: Appl Phys 42:013001
11. Battle X, Labarta A (2002) Finite-size effects in fine particles: magnetic and transport
properties. J Phys D: Appl Phys 35:R15–R42
12. Majetich SA, Sachan M (2006) Magnetostatic interactions in magnetic nanoparticle assem-
blies: energy, time and length scales. J Phys D: Appl Phys 39:R407–R422
13. Zener C (1954) Classical theory of the temperature dependence of magnetic anisotropy
energy. Phys Rev 96(5):1335
14. Torres TE, Lima E Jr, Mayoral A, Ibarra A, Marquina C, Ibarra MR, Goya GF (2015) Validity
of the Néel-Arrhenius model for highly anisotropic Cox Fe3x O4 nanoparticles. J Appl Phys
118:183902
15. Skomski R (1999) RKKY interactions between nanomagnets of arbitrary shape. Europhys
Lett 48(4):455–460
16. Mydosh JA (1993) Spin Glasses. Taylor & Francis, Washington, DC
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 35

17. Skomski R (2001) Are there superspin glasses? J Appl Phys 109:07E149
18. Mørup S (1994) Superferromagnetic nanostructures. Hyp Int 9:171–185
19. Essam JW (1980) Percolation theory. Reports on Progress in Physics 43(7):833–912
20. Bakuzis AP, Morais PC (2004) Dilution effect upon the superferromagnetic ordering of two-
dimensional magnetic nanodots. Phys. Stat. Sol. (C) 1(12):3332–3335
21. Yang HT, Liu HL, Song NN, Du HF, Zhang XQ, Cheng ZH, Shen J, Li LF (2011)
Determination of the critical interspacing for the noninteracting magnetic nanoparticle
system. Appl Phys Lett 98:153112
22. Vargas JM, Nunes WC, Socolovsky LM, Knobel M, Zanchet D (2005) Phys Rev B 72:184428
23. Allia P, Coisson M, Tiberto P, Vinai F, Knobel M, Novak MA, Nunes WC (2001) Granular
Cu-Co alloys as interacting superparamagnets. Phys Rev B 64:144420
24. Skumryev V, Stoyanov S, Zhang Y, Hadjipanayis G, Givord D, Nogués J (2003) Beating the
superparamagnetic limit with exchange bias. Nature 423:850–853
25. Gamino M, Michea S, Denardin JC, Schelp LF, Dorneles LS. Exchange-biased extraordinary
Hall effect in SiO2 /Co/CoO multilayers (unpublished results)
26. Luis F, Petroff F, Torres JM, García LM, Bartolomé J, Carrey J, Vaurès A (2002) Magnetic
relaxation of interacting Co clusters: crossover from two- to three-dimensional lattices. Phys
Rev Lett 88:217205
27. Bruvera IJ, Zélis PM, Calatayud MP, Goya GF, Sánchez FH (2016) Determination of the
blocking temperature of magnetic nanoparticles: the good, the bad, and the ugly. J Appl Phys
118(18):184304
28. Socolovsky LM, Denardin JC, Brandl AL, Knobel M (2003) Magnetotransport, magnetic, and
structural properties of TM–SiO2 (TMDFe, Co, Ni) granular alloys. Mat Char 50:117–121
29. Tournus F, Bonet E (2011) J Mag Magn Mat. 323:1109
30. Tournus F, Tamion A (2011) J Mag Magn Mat 323:1118
31. Knobel M, Nunes WC, Winnischofer H, Rocha TCR, Socolovsky LM, Mayorga CL,
Zanchet D (2007) Effects of magnetic interparticle coupling on the blocking temperature of
ferromagnetic nanoparticle arrays. M. Journal of Non-Crystalline Solids 353:743–747
32. Socolovsky LM, Sánchez FH, Shingu PH (2002) Physica B: Phys Cond Matter 3201-4:149–
152
33. Fernández van Raap MB, Sánchez FH, Rodríguez Torres CE, Casas LI, Roig A, Molins E
(2005) J Phys: Cond Matter 17:6519
34. García-Palacios JL (2000) On the statics and dynamics of magneto-anisotropic nanoparticles.
Adv Chem Phys 112:1–210
35. Shtrikmann S, Wohlfarth EP (1981) The theory of the Vogel-Fulcher law of spin glasses. Phys
Lett A 85:467
36. Bittova B, Poltierova-Vejpravova J, Roca AG, Morales MP, Tyrpekl V (2010) Relaxation
phenomena in ensembles of CoFe2 O4 nanoparticles. J Phys: Conference Series 200:072012
37. Fang M, Strom V, Olsson RT, Belova L, Rao KV (2012) Particle size and magnetic properties
dependence on growth temperature for rapid mixed co-precipitated magnetite nanoparticles.
Nanotechnology 23:145601
38. Barbeta VB, Jardim RF, Kiyohara PK, Effenberger FB, Rossi LM (2010) Magnetic properties
of Fe3 O4 nanoparticles coated with oleic and dodecanoic acids. J Appl Phys 107:073913
39. Nadeem K, Krenn H, Traussnig T, Wurschum R, Szabo DV, Letofsky-Papst I (2011) Effect of
dipolar and exchange interactions on magnetic blocking of maghemite nanoparticles. J Mag
Magn Mat 323:1998–2004
40. Masunaga SH, Jardim RF, Fichtner PFP, Rivas J (2009) Role of dipolar interactions in a
system of Ni nanoparticles studied by magnetic susceptibility measurements. Phys Rev B
80:184428
41. Parekh K, Upadhyay RV, Aswal VK (2009) Monodispersed superparamagnetic Fe3 O4
nanoparticles: synthesis and characterization. J Nanosci Nanotech 9:2104–2110
42. Zelenaková A, Zelenak V, Michalík S, Kovác J, Meisel MW (2014) Structural and magnetic
properties of CoO-Pt core-shell nanoparticles. Phys Rev B 89:104417
36 L.M. Socolovsky and O. Moscoso Londoño

43. Umut E, Pineider F, Arosio P, Sangregorio C, Corti M, Tabak F, Lascialfari A, Ghigna P

(2012) Magnetic, optical and relaxometric properties of organically coated gold–magnetite
(Au–Fe3 O4 ) hybrid nanoparticles for potential use in biomedical applications. J Mag Magn
Mat 324:2373–2379
44. Vázquez-Vázquez C, López-Quintela MA, Buján-Núñez MC, Rivas J (2011) Finite size and
surface effects on the magnetic properties of cobalt ferrite nanoparticles. J Nanopart Res
13:1663–1676
45. Dormann JL, D’Orazio F, Lucari F, Tronc E, Prené P, Jolivet JP, Fiorani D, Cherkaoui R,
Noguès M (1996) Thermal variation of the relaxation time of the magnetic moment of g-
Fe2 O3 nanoparticles with interparticle interactions of various strengths. Phys Rev B 53:14291
46. Socolovsky LM, Sánchez FH (2003) Thermal scanning studies of percolated Fe–Cu granular
alloys. Mat Char 50:123–125
47. Mendoza Zelis P, Pasquevich G, Sanchez FH, Martinez N, Veiga A (2002) Phys Lett A
298:55–59
48. Mørup S, Dumesic JA, Topsoe H (1980) In: Cohen RL (ed) Applications of Mössbauer
spectroscopy, vol 2. Academic Press, New York, p 1
49. Chien CL, Liou SH, Kofalt D, Yu W, Egami T, Watson TJ, McGuire TR (1986) Magnetic
properties of Fex Cu100x solid solutions. Phys Rev B 33:3247
50. Socolovsky LM, Sánchez FH, Shingu PH (2001) Magnetic structure of Fex Cu100–x magne-
toresistive alloys produced by mechanical alloying. Hyp Interact 133:47–52
51. Bean CP, Livingston JD (1959) Superparamagnetism. J Appl Phys 30:120
52. Kodama H, Berkovitz AE, Mc Niff EJ Jr, Foner S (1996) Surface spin disorder in NiFe2 O4
nanoparticles. Phys Rev Lett 77:394
53. Cullity BD (1972) Introduction to magnetic materials. Addison-Wesley, Reading
54. Dormann JL, Bessais L, Fiorani D (1997) A dynamic study of small interacting particles:
superparamagnetic model and spin-glass laws. Adv Chem Phys 98:283
55. Mørup S, Tronc E (1994) Superparamagnetic relaxation of weakly interacting particles. Phys
Rev Lett 72:3278
56. Dormann JL, Fiorani D, Tronc E (1999) On the models for interparticle interactions in
nanoparticle assemblies: comparison with experimental results. J Mag Magn Mater 202:251
57. Brandl AL, Socolovsky LM, Denardin JC, Knobel M (2005) Effects of dipolar interactions
on magnetic properties of granular solids. J Mag Magn Mater 294:127–132
58. Knobel M, Nunes WC, Brandl AL, Vargas JM, Socolovsky LM, Zanchet D (2004) Interaction
effects in magnetic granular systems. Phys B: Phys Cond Matt 354(1–4):80–87
59. Kang YS, Risbud S, Rabolt JF, Stroeve P (1996) Synthesis and characterization of nanometer-
size Fe3 O4 and ”-Fe2 O3 particles. Chem Mater 8:2209–2211
60. Moscoso-Londoño O, Gonzalez JS, Muraca D, Hoppe CE, Alvarez VA, López-Quintela A,
Socolovsky LM, Pirota KR (2013) Structural and magnetic behavior of ferrogels obtained by
freezing thawing of polyvinyl alcohol/poly(acrylic acid) (PAA)-coated iron oxide nanoparti-
cles. Eur Polym J 49:279
61. Moscoso-Londoño O, Muraca D, Oliveira LAS, Pirota KR, Socolovsky LM (2013) The effect
of coated-Fe3 O4 nanoparticles on magnetic properties of ferrogels produced by diffusion
route. IEEE Transactions on Magnetics 49(8):4551–4554
62. Ferrari EF, Brandl AL, Piccin R, Denardin JC, Socolovsky LM, Knobel M. Reduction in the
magnetic anisotropy energy of exchange coupled nanometric particles (unpublished results)
63. Mydosh JA (2015) Rep Prog Phys 78:052501
64. Coles BR, Sarkissian BVB, Taylor RH (1978) The role of finite magnetic clusters in Au-Fe
alloys near the percolation concentration. Phil Mag B 37:489–498
65. Sarkissian BVB (1981) The appearance of critical behaviour at the onset of ferromagnetism
in AuFe alloys. J Phys F Metal Phys 11:2191–2208
66. Campbell IA, Senoussi S (1992) Phil Mag B 65(6):1267
67. Topor L, Kleppa OJ (1984) Thermochemistry of binary liquid gold alloys: the systems Au-Ni,
Au-Co, Au-Fe, and Au-Mn. Metall Trans B 15B:673
1 Consequences of Magnetic Interaction Phenomena in Granular Systems 37

68. Nunes WC, De Biasi E, Meneses CT, Knobel M, Winnischofer H, Rocha TCR, Zanchet D
(2008) Appl Phys Lett 92(18):183113
69. De Biasi E, León-Vanegas A, Nunes WC, Sharma SK, Haddad P, Rocha TCR, Santos Duque
JG, Zanchet D, Knobel M (2008) Eur Phys J B 66(4):503–508
70. Levy PM (1994) Solid State Physics 47:367
71. Chien CL, Xiao JQ, Jiang JS (1993) Giant negative magnetoresistance in granular ferromag-
netic systems. J Appl Phys 73(10):5309
72. Camblong HE, Levy PM, Zhang S (1995) Phys Rev B 51:2216052
73. Baibich MN, Broto JM, Fert A, Van Dau FN, Petroff F, Etienne P, Creuzet G, Friederich A,
Chazelas J (1988) Giant magnetoresistance of (001) Fe/(001) Cr magnetic superlattices. Phys
Rev Lett 61:2472
74. Binasch G, Grünberg P, Saurenbach F, Zinn W (1989) Enhanced magnetoresistance in layered
magnetic structures with antiferromagnetic interlayer exchange. Phys Rev B 39(7):4828(R)
75. Fert A (1969) Two-current conduction in ferromagnetic metals and spin wave-electron
collisions. J Phys C 2(2):1784
76. Xiao JQ, Jiang JS, Chien CL (1992) Giant magnetoresistance in nonmultilayer magnetic
systems. Phys Rev Lett 68(25):3749
77. Nigam AK, Majumdar AK (1979) Anomalous magnetoresistance in AuFe alloys. J App Phys
50:1712
78. Gerritsen AK (1953) The magnetoresistances of alloys of a noble metal and a transition metal
at low temperatures. Physica 19(1–12):61–73
79. Gerritsen AK (1957) Resistance and magneto-resistance of dilute alloys of gold with iron at
low temperatures. Physica 23(6–10):1087–1099
80. Rohrer H (1968) Magnetoresistance of dilute alloys. Phys Rev B 174(2):583
81. Nakhimovich NM (1941) Dependence of the electrical resistance of gold and several of its
alloys on a magnetic field al low temperatures. J Phys V 2–3:141
82. Maekawa S, Inoue J, Itoh C (1996) J Appl Phys 79:8
83. Ferrari EF, da Silva FCS, Knobel M (1999) Phys Rev B 59(13):8412
84. Allia P, Knobel M, Tiberto P, Vinai F (1995) Magnetic properties and giant magnetoresistance
of melt-spun granular Cu100x Cox alloys. Phys Rev B 52:15398–11541
85. Liu K, Zhao L, Klavins P, Osterloh FE, Hiramatsu H (2003) Extrinsic magnetoresistance in
magnetite nanoparticles. J App Phys 93:7951
86. Kechkrakos D, Trohidou KN (2002) Physica B 318:360
87. Julliere M (1975) Phys Lett A 54(3):225
88. Abeles R, Sheng P, Coutts MD, Arie Y(1975) Structural and electrical properties of granular
metal films. Adv Phys 24–3:407–461
89. Kechkrakos D, Trohidou KN (2005) Phys. Rev. B 71:054416
90. Inoue J, Maekawa S (1996) Theory of tunneling magnetoresistance in granular magnetic
films. Phys. Rev. B 53(18):11927–11929
91. Denardin JC, Knobel M, Dorneles LS, Schelp LF (2004) Structural and magnetotransport
properties of discontinuous Co/SiO2 multilayers. Mat Sci Eng B 112(2–3):120–122
92. Varfolomeev AV, Zavyalov AS, Volkov AV, Volkov IA, Moskvina MA, Polyakov SN,
Malashko AP, Bayburtskiy FS, Baldokhin YV, Stepanov GV (2006) Preparation and inves-
tigation of magnetic composite materials properties on the basis of magnetite and polyvinyl
alcohol. Oxidation Comm 29(3):693–697
93. Wang W, Yu M, Chen Y, Tang J (2006) Large room-temperature spin-dependent tunneling
magnetoresistance in a Fe3 O4 -polymer composite system. Phys Rev B 73:134412
94. Black CT, Murray CB, Sandstrom RL, Sun S (2000) Spin-dependent tunneling in self-
assembled cobalt-nanocrystal superlattices. Science 290:1131–1134
95. Taub N, Tsukernik A, Markovich G (2009) J Mag Magn Mater 321:1933
96. Poddar P, Fried T, Markovich G (2002) First-order metal-insulator transition and spin-
polarized tunneling in Fe3 O4 nanocrystals. Phys Rev B 65:172405
97. Markovich G (2013) Magneto-transport and magnetization dynamics in magnetic nanoparti-
cle assemblies. Mat Today 38:939
38 L.M. Socolovsky and O. Moscoso Londoño

98. Bonini M, Fratini E, Baglioni P (2007) SAXS study of chain-like structures formed by
magnetic nanoparticles. Mat Sci Eng C 27:1377–1381
99. Pakhomov AB, Yan X, Zhao B (1995) Appl Phys Lett 67:3497
100. Hurd M (1972) The Hall effect in metals and alloys. Plenum Press, New York
101. Gerber A, Milner A, Finkler A, Karpovski M, Goldsmith L, Tuaillon-Combes J, Boisron O,
Mélinon P, Perez A (2004) Correlation between the extraordinary Hall effect and resistivity.
Phys Rev B 69:224403
102. Karplus R, Luttinger M (1954) Hall effect in ferromagnetics. Phys Rev 95:1154
103. Berger L (1970) Side-jump mechanism for the Hall effect on ferromagnets. Phys Rev
B2:4559
104. Jing XN, Wang N, Pakhomov AB, Fung KK, Yan X (1996) Effect of annealing on the giant
Hall effect. Phys Rev B 53:14032
105. Socolovsky LM, Oliveira CLP, Denardin JC, Knobel M, Torriani IL (2005) Nanostructure of
granular Co-SiO2 thin films modified by thermal treatment and its relationship with giant Hall
effect. Phys Rev B 72:184423
106. Denardin JC, Pakhomov AB, Brandl AL, Socolovsky LM, Knobel M, Zhang XX (2003) Appl
Phys Lett 82(5):763–765
107. Socolovsky LM, Oliveira CLP, Denardin JC, Knobel M, Torriani IL (2006) Nanostructure
and giant Hall effect in TMx (SiO2 )1-x (TM D Co, Fe, Ni) granular system. J Appl Phys
99:08C511
108. Socolovsky LM, Sánchez FH, Shingu PH (2001) Giant magnetoresistive properties of
Fex Au100–x alloys produced by mechanical alloying. J Mag Magn Mat 226–230:736