Magnetic Effects in Sensing Applications

by
Evangelos Hristoforou
Laboratory of Physical Metallurgy, National Technical University of Athens
Athens 15780, Greece

* Corresponding Author. Address for correspondence:

Laboratory of Physical Metallurgy, National Technical University of Athens, Athens
15780, Greece
Tel: +302107722178
Fax: +302107722119
E-mail: [email protected]

1
Contents
1. Introduction
2. An overview of magnetics
2.1 Magnetic force, field and injuction
2.2 The magnetic dipole and magnetic dipole moment
2.3 Magnetization, magnetic field, magnetic induction
2.4 Susceptibility and permeability
2.5 The electromagnetic induction
2.6 The unit system issue
2.7 Maxwell’s equations
2.8 Magnetic circuits
2.9 Phenomenology of the magnetization process
3. Magnetism and magnetic materials
3.1 The origins of magnetism
3.2 Types of magnetism and magnetic materials
3.3 The magnetization process
3.4 The energy viewpoint and modeling
3.5 Dynamic phenomena
3.6 Superparamagnetism
3.7 Magnetotransport
3.8 Noise
3.9 Hysteresis curve revisited
4. Magnetic effects
4.1 Induction B(H)
4.2 Magnetostrictive effects
4.3 Magnetotransport effects
4.4 Magnetomechanical effects: dependent effects
4.5 Other effects
5. Sensor arrangements
5.1. Field sensors
5.2. Position sensors
5.3. Stress sensors
5.4. Security sensors
5.5. Smart sensors
6. Sensor applications
6.1. Industrial applications
6.2. Biomedical applications
6.3. Military applications
6.4. Transport and Automotive applications
6.5. Environmental applications
6.6. Domestic and information technology applications
6.7. Laboratory sensors
7. Designing the future
Annex A: Developing a sensor
Glossary
References

2
1. Introduction
If magnetism is loosely defined as the macroscopic property of certain materials to
exert an attractive (magnetic) force, being neither gravitational nor electrostatic, then
magnetics can be defined as the science investigating and the technology utilizing the
phenomenon of magnetism to develop systems and products, an example of which are
magnetic sensors.
Sensors and transducers are attracting a renewed interest both among the industrial
and the academic community because of their increasing importance and their
stringent specifications in reliability, sensitivity, size, operating conditions and
interface capability in today's automated applications [1]. All modern vehicles and
transportation means use a vast variety of sensors and transducers. The operation of
all medical instruments is based on sensors. Industry is employing more and more
transducers for the monitoring and control of production lines.
In literature, sensors have been categorized in several ways [2,3]. They can be
classified according to the physical phenomenon they are based on, i.e, chemical,
optical, magnetic, semiconducting etc; or, according to the size they measure, i.e.,
position, pressure, flow, velocity, etc; or, according to the application, i.e., military,
domestic, industrial, transportation, automotive, medical, environmental sensors.
This chapter is focused on magnetic sensors; that is, on sensing devices based on
magnetic effects and materials, which mainly respond to magnetic fields, either
externally applied, or induced through other means, such as pressure, stress etc.
Magnetic sensors are used to measure a wide variety of physical sizes: position,
distance, displacement, mass, velocity, acceleration, stress and of course magnetic
field. Actually, all that magnetic sensors do is detecting and measuring the magnitude,
direction, change or rate of change of magnetic fields or magnetization and relate it, if
needed, to another size [4,5]. A major advantage of magnetic sensors is that they may
allow for contact-less operation with the sensing element reacting to an applied field
in the vicinity.
More specifically, this chapter is mainly about the magnetic effects, the phenomena
and processes, used in various types of sensors and sensing arrangements. The most
often used magnetic phenomena in today’s magnetic sensor technology are the ones
based on magnetotransport properties, like the magneto-resistance [6], the magneto-
impedance [7], and the tunneling effects [8], those based on magnetostrictive effects
[9] and those based on electromagnetic induction effects [10]. Electromagnetic
effects, such as the Hall effect, the quantum Hall effect, the Jossephson effect, and the
Zeeman effect, though outside the scope of this work, are utilized in field detection
and they are briefly presented for the sake of completeness.
Special emphasis is given on the dynamics of magnetic domains and domain walls
involved in the magnetization process since, at the mesoscopic level, it is the main
mechanism responsible for magnetic effects used in sensing applications [3, 11]. Any
possible use of the dynamic response of this mechanism can result in a sensing
element.
The following sections are organized in such a way as to lead the reader, through a
brief overview of magnetics and magnetostatics and the related terminology (section
2), to the understanding of the underlying magnetization processes, in the various
kinds of magnetic materials, at the macroscopic, mesoscopic and atomic level (section
3). Then, the magnetic effects are described in terms of these two preceding sections
and with respect to their applications in sensing (section 4). Finally, an overview of
the existing magnetic sensors with an outlook for the design requirements and
procedures in the future is presented (sections 5-7).

3
2. An overview of magnetics
The following overview comprises the basic notions and the related formulae of the
theory of magnetics. More detailed accounts can be found in any textbook of
magnetism or physics [12-15]. Maxwell equations are also included in this section
because of the close relationship between electrical and magnetic fields [15-16].

2.1 Magnetic force, field and induction

The theory of magnetostatics has been developed in analogy to the theory of
electrostatics with one basic difference: the charges are in motion. The origins of
magnetism lie in the spin and orbital motion of the electrons. The electric force
between two point charges q1 and q2 at rest, separated by distance r, is given by the
Coulomb’s Law:
1 q1q 2
F= rˆ . (2.1)
4πε 0 r 2
C2
ε0 is a constant corresponding to the permittivity of vacuum: ε 0 = 8.85 × 10−12
N ⋅ m2
F
or, as is usually listed, ε 0 = 8.85 × 10−12 .
m
If these two charges move a magnetic force is acting on them, as well, which is given
by a modification of the Coulomb’s law to allow for the respective charge velocities
v1 and v2:
μ qq
F = 0 1 2 2 ( v1 × ( v 2 × rˆ ) ) (2.2)
4π r
μ0 is a constant corresponding to the permeability of vacuum: μ 0 = 4π× 10−7 N s2/C2
H
or, as typically quoted, μ 0 = 4π× 10−7 . The notion of permeability will be discussed
m
later in more detail.
Defining the quantity resulting from the moving charge q at velocity v as a magnetic
pole m = μ0 q v, (2.2) is rewritten in the following form:
1 m1m 2
F= rˆ (2.3)
4πμ 0 r 2
The magnetic force F as given by (2.3) is now in direct analogy to the electric force as
expressed by Coulomb’s law in (2.1).
N ⋅s
The strength of a magnetic pole is measured in Webers: 1Wb = 1 m.
C
The existence of a magnetic force F implies the existence of a magnetic field H acting
as a mediator for the force that spatially distributed magnetic poles exert on one
another. In analogy to the electric field due to the existence of electric charge, H can
be defined by:
H=F/m. (2.4)
N C 1 A
measured in = = .
Wb s m m

In analogy to the electric field, a magnetic field generates magnetic flux Φ whose
lines originate at the positive (South) pole and sink in the negative (North) pole of a
bar magnet. In the case of a current-carrying wire the direction of flux lines is

4
perpendicular to the current direction and follows the right-hand rule. Magnetic flux
Φ, the number of field lines intercepting a surface of area S, corresponds to the
strength of the magnetic poles generating the field and is therefore also measured in
Wb.

(a)

(b)

S N

(c)

Figure 2.1 Magnetic field of (a) a solenoid and (b) a bar magnet and (c) magnetic induction
of a bar magnet

The amount of flux per unit area, dS, is called the magnetic flux density or magnetic
induction B:
Φ = ∫ B ⋅ dS (2.5)

5
 Wb N ⋅ s
The unit of magnetic induction is the Tesla (T), defined as T = = .
m2 C ⋅ m
For weak fields, e.g. the earth field, B is measured in Gauss (cgs): 1Gauss=10-4T.
The magnetic induction B can be implicitly defined by the force F exerted on a charge
q moving at velocity v inside a magnetic field:
F = qv × B (2.6)
From (2.4) and (2.6), it can be deduced that:
B= μ0H (2.7)
Therefore, a magnetic field can be generated either by the flow of electric current, e.g.
in a wire or a solenoid (Fig. 2.1a), or by the existence of magnetic charges (poles)
interacting with each other, e.g. a bar magnet (Fig. 2.1b). In a bar magnet there is a
high concentration of positive and negative charges, which, as will be discussed later,
are ultimately the result of the spin motion of the electrons in the material, i.e.
elementary currents. The field lines of a current-carrying solenoid follow the right-
hand rule while the field lines inside and outside a bar magnet originate at the North
pole and sink in the South pole. The magnetic induction lines of a bar magnet (Fig.
2.1c) are similar to the solenoid field lines.
The magnetic field H as a result of N wires carrying current I is defined by the line
integral known as Ampère’s law:
∫ H ⋅ dl = NI
closed
(2.8)
path

For example, the magnetic field inside an infinite length solenoid of n turns/meter
carrying current I is H=nI A/m.
The magnetic field intensity dH at a distance r from a segment dl of an arbitrarily
shaped conductor carrying current I is calculated by the Biot-Savart law:
1
dH = Idl × rˆ (2.9)
4πr 2
The magnetic field produced by current flow in a conductor is perpendicular to it and
follows the right-hand rule.
The laws of Ampère (2.8) and Biot-Savart (2.9) are equivalent and part of the
Maxwell equations.

2.2 The magnetic dipole and magnetic dipole moment

Following (2.4), the torque L exerted on a bar magnet of length l placed inside a
uniform magnetic field H and lying along the direction of unit vector r̂ is given by:
L = mlrˆ × H (2.10)
The quantity ml r̂ is called the magnetic (dipole) moment, m, measured in Wb m.
Torque can also be exerted on a current loop of area S with a normal unit vector
n̂ placed inside a uniform magnetic field H:
L = ISnˆ × H (2.11)
In that case, the magnetic moment is defined as m = ISn .
ˆ
The stored energy, E, is given by the inner product of the magnetic moment m and the
applied field, H:
E=mH (2.12)
The magnetic dipole is the basic magnetic entity. It can be thought of as a very small
bar magnet with a positive (South) and negative (North) pole. (“Very small” refers to
the distance separating the two poles compared to the distance at which we calculate

6
the magnetic force.) Or, equivalently, as a current loop of very small surface, such as
the spin or orbital motion of the electrons, as will be explained later. It is measured in
emu (cgs system)
The smallest magnetic moment, the magnetic quantum, is called Bohr magneton and
is equal to MB=1.165x10-29 Wb-m. It corresponds to the value by which the magnetic
moment of the orbital motion is allowed to change:
μ he
MB = 0 (2.13)
2m
where ћ is Planck’s constant, e and m the charge and mass of electron respectively.

2.3 Magnetization, magnetic field, magnetic induction

The magnetic moment per unit volume is called the magnetic polarization (in analogy
to electrostatics) or the intensity of magnetization, I, measured in Wb/m2, like the
magnetic induction B. The quantity more often used is, however, the magnetization:
M=I/μ0 (2.14)
measured in A/m like the magnetic field H.
As follows from the definition of magnetization, in the cgs system, the units are
emu/cm3.
The three magnetic quantities, magnetic field H, magnetic induction B and
magnetization M are related by:
B= μ0(H+M)= μ0H+I (2.15)
Equation (2.15) states that, for a given system, e.g. solid, inside a magnetic field, the
magnetic induction, or flux density through the surface of the solid, is the vector sum
of the magnetic field H and the magnetization M, both measured in A/m. The H
contribution is due to the presence of external magnetic fields generated either by
conductors or permanent magnets outside the system and the M contribution is due to
the dipole moments inside the system. In the absence of an external magnetic field,
B= μ0M=I while in the absence of a solid, B=μ0H.

2.4 Susceptibility and permeability

M and B can both be measured for a given system as a function of H.
It can then be found that the magnetic induction B is related to the applied field H
through
B= μH (2.16)
where μ is the magnetic permeability of the system or material while the
magnetization M and the applied field H are related through
M=χH (2.17)
where χ is the magnetic susceptibility of the material. Because χ is a measure of the
material’s “sensitivity” to magnetic fields, it is used as a classification criterion of
magnetic materials.
The permeability μ is measured in H/m like the permeability of free space μ0 while
the susceptibility χ is a dimensionless quantity:
μ = μ0(χ+1) (2.18)
A quantity typically used in the case of ferromagnets is the dimensionless relative
permeability: μr= μ/ μ0.
As discussed in section 2.8, μ and χ are neither constant nor, necessarily, linear
functions of the field Η.

7
2.5 The electromagnetic induction
At this point, it should be clear enough that the variation of electrical fields (or motion
of charges) generates magnetic fields. It so happens, that the variation of magnetic
fields generates electrical fields, called induced electrical fields. The magnetic field
changes when the flux lines change or move. According to Faraday’s law, the induced
voltage is proportional to the rate of change of flux lines:
dΦ
V=− (2.19)
dt
For example, the induced voltage across an N-turn solenoid of cross-section S
subjected to a variable magnetic induction, B, is given by:
dB
V = − NS (2.20)
dt
The minus sign indicates that the sign of the induced voltage will be opposite to the
rate of flux change.
Electromagnetic induction can therefore be used to measure flux changes and is
widely used in sensing circuits and devices as discussed in Section 4.1.

2.6 The unit system issue

At this point, the variety in units related to magnetic phenomena may have already
become obvious to the reader. A brief discussion is deemed appropriate. The sets of
units for magnetic quantities most often encountered in literature are two: the cgs
system and the SI (Sommerfeld) system. Even though the SI is the standard one – also
used in this work – the cgs is still quite popular in the market as well as in the
scientific community. The reasons for this are both historical and practical. The
familiarization with both systems is therefore helpful if not necessary.
In the cgs system, Eq. 2.15 is written as B = H + 4πM from which it can be seen that
the units of magnetic induction (G) and magnetic field (Oe) are equivalent. For
example, the earth’s magnetic field is about 0.7 Oe or 56 A/m and its magnetic
induction is 0.7 G or 0.07 mT. The earth’s field is considered a weak field. On the
contrary, the intrinsic magnetization of the magnetic domains of a ferromagnetic
substance is calculated to be in the order of 10T, while the measured magnetic field
typically generated by a permanent magnet or an electromagnet is of the order of 1T.

2.7 Maxwell’s equations

The theories of magnetostatics and electrostatics were unified by Maxwell for time
varying electric and magnetic fields in his “Dynamical Theory of the Electromagnetic
Field”. In this theory, the motion of charge (electric current) is not steady as is in
magnetostatics and the distribution of charge is not stationary as in electrostatics.
Maxwell showed that in the general case where the electric charge moves in a time
varying way, a new type of field, the electromagnetic field is established. The
electromagnetic field stores energy that can be propagated in time and space in the
form of electromagnetic waves.
The four Maxwell’s equations are given below in integral and point form:
d ∂B
Faraday’s law: Ρ∫c E ⋅ dl = − dt ∫s B ⋅ ds, ∇ × E = − ∂t (2.21a)

d ∂D
Ampère’s law: Ρ∫ H ⋅ dl = ∫ J ⋅ ds + dt ∫ D ⋅ ds,
c s s
∇×H = J +
∂t
(2.21b)

8
Gauss’ laws: Ρ∫ D ⋅ ds = ∫ ρdv,
s v
∇⋅D = ρ (2.21c)

Ρ∫ B ⋅ ds = 0,
s
∇⋅B = 0 (2.21d)

Note that all fields H, B, E, D in Eq. 2.21 are functions of both time and space, i.e.
H(x,y,z,t), E(x,y,z,t), etc…. and they are coupled in pairs and through the constitutive
relations:
D = ε0E + P (2.21e)
and
H = B / μ0 − M (2.21f)
where P is the electric polarization and M is the magnetization.

Faraday’s law (2.20, 2.21a) states that a time varying magnetic flux density B through
a surface S gives rise to a time-varying electric field E along a contour c. Or, the time
derivative of B corresponds to the curl (the net circulation along a contour) of E. In
the case of a steady magnetic flux density, there is no circulation of the electrostatic
field which is in agreement with the conservation property of the electrostatic fields:

c
Ρ∫ E ⋅ dl = 0 .
Ampère’s law (2.8) states that the current passing through a surface S gives rise to a
magnetic field H along a contour c. Maxwell’s version (2.21b) states that, in the case
of time varying fields, the total current consists of two contributions: J is the
conduction or convection current density, I c = ∫ J ⋅ ds , and D is the time varying
s

electric flux density through surface S whose time derivative corresponds to the
d
displacement current, I d = ∫ D ⋅ ds . If D is steady there is no displacement current
dt s
and Eq. 2.21b is the Ampère’s law for stationary magnetic fields (2.8). The
displacement current ensures the continuity of the current flow in a circuit when there
is a gap in the conducting elements.
The Gauss’ laws for magnetic and electric fields are the same for both static and time-
varying fields. Eq. (2.21c) states that the net electric flux emanating from a closed
surface S is equivalent to the net volume charge enclosed by the surface. Eq. (2.21d)
states that the net magnetic flux through a closed surface S is zero. This is consistent
with the fact there are no known free magnetic poles.
The force related to the electromagnetic field is the Lorentz force, the vector sum of
the force exerted on a charge moving at velocity v inside a magnetic field B (2.6), and
the force exerted by the electric field E on the same charge q:
F = qE + qv × B (2.22)
The Lorentz force couples electromagnetics to mechanical phenomena and has found
a lot of applications in sensors, actuators, and the study of material properties, e.g.
Lorentz microscopy, Transmission Electron Microscopy (TEM), Magnetic and
Atomic Force Microscopy (AFM/MFM). It is also related to magnetotrasnport
material properties, like the MR effect (3.2.12), the Hall effect et.al.
More on Maxwell equations can be found in any standard electromagnetics textbook
[16].

9
2.8 Magnetic circuits
A circuit can be defined as a closed path in which the flow of some quantity takes
place. In electrical circuits, electrical current flows through the conductors and the
other - passive or active – devices comprising it. Magnetic circuits are the devices
guaranteeing or making use of the flow of magnetic flux, Φ. The purpose of magnetic
circuits is to guide the flux through a desired route in order to generate a magnetic
field of a particular spatial profile or to “sense” an existing magnetic field or link in a
contact-less way two circuits or components.
-V
I

I I
CIRCUIT CIRCUIT
1 2

(a) (b)
Figure 2.2 Magnetic circuits (a) electromagnet (b) transformer

Figure 2.2 shows two common examples of magnetic circuits. The air gap (Fig. 2.2a)
causes the fringing of the flux lines in the way shown. The higher the permeability of
the core (gray) material, the stronger is the flux linkage from one pole of the
electromagnet to the other. With proper design of the electromagnet material, pole and
air gap, the desired field profile can be generated inside the air gap, as in the case of
field electromagnets. Alternatively, the fringing profile is used in the case of
recording heads where the magnetic medium moves underneath the air gap; the
electromagnet write current controls the flux fringing and the recording fields in the
air gap responsible for the magnetization pattern recorded in the medium. On the
other hand, the reproduction process is a “sensing” operation. The flux emanating
from the recorded pattern is linked through the air gap to the coil, now functioning as
a pick-up element, inducing a voltage to it (Faraday’s law).
In the case of a transformer (Fig. 2.2b), circuits 1 and 2 are linked to each other
through the mutual inductance between coils 1 and 2. The mutual inductance is the
flux linkage between the two coils assisted by the high permeability path supplied by
the (iron) yoke. The ability of flux to flow even through air (air has a finite
permeability, μ0) is one of the major advantages of magnetic devices, such as sensors,
since it allows for contact-less operation.
The analysis of magnetic circuits is in many ways similar to that of electric circuits
[11, 12]. A quantity called magnetic resistance or reluctance, Rm, can be defined for a
segment dl of a magnetic circuit or component, of permeability μ and cross section S,
as:
dl
Rm = ∫ (2.23)
μS
The magnetomotive force Vm, in analogy to the emf, for a closed path of the circuit is
given by the Ampère’s law:

10
 ∫ Hdl
Vm = Ρ
c
(2.24)

Vm has the dimensions of current, measured in Ampère-turns, as opposed to the

voltage dimensions of the electromotive force.
The total magnetomotive force Vm at a circuit where flux Φ flows through paths of
different reluctances can be calculated by the second Kirchhoff’s circuital law as:
Vm = Φ ( Rm1 + Rm 2 + ...) (2.25)
Kirchhoff’s first law in magnetic circuits guarantees the preservation of flux whenever
there are shunt branches in a circuit:
∑ Φi = 0 (2.26)
The complication with magnetic circuits lies usually in determining the flux Φ, or
equivalently the magnetic induction or flux density B (Eq. 2.5). The magnetic
induction is given by Eq. 2.15 but the presence of hysteresis in the magnetic materials
comprising the circuits, the highly nonlinear function μ(H), the shape dependent
demagnetizing fields, the elastic properties of magnetostrictive materials are a few of
the factors that make the calculation of B a challenging task. Maxwell’s equations
(see 2.7) coupled with hysteresis compensating models (3.2.8) must then be solved, a
task that, for arbitrary geometries, requires numerical solutions, finite element
calculations or other techniques [17-20].

2.9 Phenomenology of the magnetization process

Figure 2.3 shows the M(H) curve of a ferromagnetic material. Starting from the
demagnetized state, at which the net magnetization is zero, i.e., M|H=0=0, the field is
being gradually increased. The magnetization is then increasing following the path
ABC known as the virgin curve. For every material, there exists a field value beyond
which the magnetization does not increase any further (point C), having reached the
saturation magnetization, Ms. Decreasing the field from saturation, the magnetization
values do not retrace the virgin curve but, exhibiting hysteresis, at zero field (point D)
reach the remanent magnetization value, or remanence, Mr. Decreasing the field
further into the second quadrant, a field value is reached at which the magnetization
becomes zero (point E). This is called the coercive field or coercivity, Hc. A further
decrease in field will eventually lead to negative saturation (point F). If now the field
starts being increased the positive saturation state will eventually be reached through
the right hand side of the curve (FGC).
The curve just described is known as the hysteresis curve or major loop of the
specimen being measured. It delimits the space of all the magnetization states reached
through any possible sequence of fields. The curves originating from the major loop
or other points inside the major loop are called minor loops. Obviously, there is no
unique field sequence leading to a magnetization state and the response of the
specimen to an applied field is a function of past field values as well. In other words, a
material with hysteresis possesses the memory property.
The ratio of remanence to saturation is called the squareness of the loop, S=Mr/Ms.
The slope of the loop at the coercivity is called the coercivity squareness,
dM
S* = .
dH H = Hc
The coercivity Hc is a measure of the magnetic hardness of the material. A high
coercivity value suggests a wide hysteresis loop and higher fields are needed in order
to reverse the magnetization from positive to negative or the opposite. Hard magnetic

11
materials exhibit coercivities in the order of 1 MA/m while the coercivities of soft
magnetic materials can be far below 1 kA/m. Semihard materials lie in between
exhibiting hysteresis in the order of a few tens of kA/m. Obviously, hard and
semihard magnets are the materials of choice in applications where high hysteresis is
desired as in the case of permanent magnets and recording materials, respectively. In
the case of most magnetic sensors where hysteresis should ideally tend to zero, soft
materials are chosen.
M
Ms C

Mr D

K’
K
A B
G H

virgin curve
E F
Hs anhysteretic
Hc,r curve

Hc (a)

(b)

Figure 2.3 (a) Hysteretic and anhysteretic magnetization curve M(H) hysteresis (b) shearing
of the loop due to the demagnetizing fields Hd=NdM

The M(H) curve of an anhysteretic material is shown on Fig. 2.3a. The B(H) curve
can also be obtained by adding H on the M(H) curve. This curve is used in the design
of permanent magnets where the maximum energy product (BH)max, usually measured
at the second quadrant, is of interest, being a figure of merit for the energy storing

12
capacity of the magnet. The shape of the loop also depends on the shape of the
specimen being measured. The ensuing demagnetizing fields (see 3.3.2) tend to shear
the loop and de-shearing needs to be performed to obtain the intrinsic loop (Fig. 2.3b).
From Fig. 2.3, it is obvious that neither the permeability μ nor the susceptibility χ are
constant or even linear functions of H. Quantities of interest are the initial
susceptibility, χini, the reversible and irreversible susceptibility, χrev and χirr, the
differential and total susceptibility, χdiff and χtot: χdiff = χrev + χirr, all obtained from the
virgin or major curve measurement.
The susceptibility in ferromagnets is a highly nonlinear function of the applied field
and previous magnetization history whose measurement is not a trivial task. This is
the reason why there are may different “types” of susceptibility according to the
phenomena they represent or the way they are calculated.
The initial susceptibility χini is measured as the slope of the virgin curve at the origin
(point A on Fig. 2.3). The differential, reversible and irreversible susceptibility can be
measured either on the virgin curve or on the major loop.
In either case, the differential susceptibility is measured as the slope of the curve at
dM
each point: χ diff ( H ) = .
dH
The reversible susceptibility is deduced from the slope of closed minor loops (KK’ on
Fig. 2.3a) originating at the curve.
The irreversible susceptibility is calculated as the difference: χirr(H)=χdiff(H)-χrev(H).
As will be mentioned in 3.3, the reversible and irreversible susceptibilities are a
measure of the reversible and irreversible processes taking place at each portion of the
major loop or virgin curve [21].
Other types of susceptibility measurements include the total susceptibility, measured
as the slope of the straight line connecting the origin with each point on the virgin
curve, or the maximum susceptibility, defined as the maximum of the total
susceptibility.

Table 2.1 The cgs vs the SI system of units in magnetism.

Quantity cgs SI Conversion from cgs to SI
Magnetic Field, H Oe A/m 1Oe = 1/4π kA/m ≈ 80 A/m
Flux , Φ maxwell Wb 1 maxwell= 10-8 Wb
Magnetic Induction (flux density), B G T=Wb/ m2 1G = 1/10 mT
Magnetic (dipole) moment, m emu A m2 1 emu = 10-3 Am2
3
Magnetization, M G=emu/cm A/m 1 emu/cm3 =1 kA/m
Permeability of free space, μ0 1 4π 10-7 H/m -

2.9a The Barkhausen jumps

Even though the magnetization curve appears to be a smooth curve, a closer
examination reveals discontinuous small or large changes called small or large
Barkhausen jumps, after Barkhausen who first discovered and studied the
phenomenon [22, 11,12]. For example, on the descending part of the curve, the small
jumps are observed at the upper part of the second quadrant while the large ones are
associated with the abrupt changes in magnetization observed in the rest of the second
quadrant and the third. The Barkhausen jumps are discontinuous changes in flux
density that are due to irreversible processes, namely, the irreversible translation of
the domain wall and the flipping of the domain magnetization to a new easy axis. The
speed of the Barkhausen jumps is limited by microscopic eddy currents,
magnetomechanical lattice interactions, and spin relaxation.

13
3 Magnetism and magnetic materials

3.1 The origins of magnetism

In the previous section, magnetism has been treated as the result of charge motion. It
should therefore come as no surprise that the fundamental phenomenon in magnetism
is considered to be the spin and the orbital motion of electron, or, to be more concise,
the spin vs. the orbital motion of electron.
The motion of the electron around the nucleus generates a magnetic moment like an
elementary current loop. This is true for any orbit as long as the angular momentum of
the orbit is an integer multiple of Planck’s constant, ћ. The magnetic moment of the
orbital motion changes in units of Bohr magnetons, MB=1.165x10-29 Wb-m (Eq.
2.13). An external magnetic field interacts with the orbital motion of the electron
causing it to change the shape of the orbit in order to accommodate the effect of the
current induced by the magnetic field and align with it.
The spin motion of the electron can be thought of as a precession about its own axis.
This motion also gives rise to a magnetic moment whose change is measured in MB.
The angular momentum of spin is restricted to ±ћ/2. In the presence of an applied
field, spins follow a precession around the field axis tending to align parallel or
antiparallel to the field, depending on which state is more energetically favorable.
Since all atoms have electrons one would expect that all elements should in principle
be magnetic. Indeed, all atoms are more or less susceptible to magnetic fields. The
application of an increasing field, in all cases but diamagnetism, results in an
increasing magnetization. Upon the application of an external field, the randomly
oriented magnetic moments tend to align with it in order to minimize the total energy
of the system and balance out the energy added to it by the applied field. The higher
the field, the more (orbital or spin) magnetic moments align with it and therefore the
higher the net magnetic moment and the resulting magnetization is.
This mechanism however is not enough to explain the very high susceptibilities and
nonzero magnetization in zero field of certain materials like ferromagnets. These
elements have, in principle, an unfilled 3d-shell (transition metals) or a 4f-shell (rare
earths). This suggests that the magnetization is also the result of uncompensated spin
and the resulting interactions. There are three types of interactions between the outer
shell electrons: spin-spin, spin-orbit and orbit-orbit. It turns out that the strongest type
is the spin-spin interaction forcing neighboring spins to align parallel to each other.
This type of interaction is held responsible for the spontaneous magnetization
exhibited by magnetic materials, such as ferromagnets.

3.2 Types of magnetism and magnetic materials

Magnetic materials are classified according to their susceptibility in diamagnetic,
paramagnetic, ferromagnetic, ferrimagnetic, and antiferromagnetic.

3.2.1 Diamagnetism
Diamagnetic materials develop a very weak magnetization varying linearly with but
opposing the applied field (Fig. 3.1a1). The origin of diamagnetism is the interaction
of the electrons’ orbital motion with the applied field. The applied field induces a
current opposing the field that caused it and thus altering the velocity of motion of the
electrons. Diamagnetic materials have no unfilled shells and therefore the magnetic
moment at zero field and ground state is zero.
The susceptibility χ=M/H is therefore negative, constant and in the order of 10-6.

14
 M M

H H

(a1) (b1)
1/χ 1/χ
T
(b3)
T

(a2) (b2)

Figure 3.1 (a) Diamagnets 1. Magnetization vs applied field 2. Inverse susceptibility vs.
temperature (b) Paramagnets 1. Magnetization vs applied field 2. Inverse
susceptibility vs. temperature 3. Random spin orientation due to thermal agitation.

Note that, diamagnetism is independent of temperature, except in the case of

superconductors that will be examined separately (Fig. 3.1a2).
A theoretical treatment of diamagnetism can be found in [13].
All materials exhibit diamagnetism but the resulting magnetization is so weak that its
presence is often masked by either the paramagnetic magnetization (see 3.2.2) or the
much prominent ferromagnetic magnetization (see 3.2.3).
Diamagnetism, except the case of superconducting materials, is not really useful in
engineering applications but its presence cannot be ignored. A typical example is the
case of very thin films grown on diamagnetic substrates, like glass, copper etc, where
diamagnetism may affect the major loop measurement. Because of the small volume
of magnetic material, the magnetic moment (magnetization × volume) of the thin film
may be small enough to be comparable to the diamagnetism of the substrate. In this
case, the loop must be corrected for this linear diamagnetic dependence.

3.2.1a Superconductors: special case of diamagnets.

Superconductors [23,24] are materials whose electrical resistance becomes zero
beyond a certain temperature, Tc and behave as perfect conductors as well as perfect
diamagnets at fields below a certain field, Hc=H0(1-T/Tc). The critical temperature is
usually very low (below 100K) and liquid helium is needed to observe or use the
phenomenon. The transition to zero resistance occurs in a discontinuous way,
suggesting that the material undergoes a kind of electronic phase transition at Tc. The
superconductors, being perfect conductors, are able to support loss-less current flow,
the so-called persistent currents. Since current flow is related to the generation of
magnetic fields, superconductors have found applications in the generation of high
and quite uniform magnetic fields as high as 15 T at low energy cost, as in the case of
accelerator magnets or SQUID magnetometers. For example, a superconducting
magnet of 1.8 T with 4.8 m diameter coils requires 190kW for the liquid Helium
refrigerator while a similar conventional electromagnet would require 10,000kW of
electrical power [15].
In the presence of a magnetic field, superconducting materials exclude the magnetic
flux as long as the applied magnetic field is lower than the critical field, Hc. This is
known as the Meissner effect. Above Hc the superconducting ordering is destroyed.
Also, when a solenoid or toroid undergoes the transition into the superconducting

15
phase while being inside a magnetic field, the flux existing at the moment of the phase
transition remains trapped inside the toroid frozen at a constant value.
Another application of superconductors is magnetic levitation. A permanent magnet in
close proximity to a superconductor induces a persistent current to it, which in turn
induces a magnetic field opposing the permanent’s magnet field (Lenz’ law). This
induced magnetic field is strong enough to overcome gravity and repel the permanent
magnet, resulting in levitation and “floating” of the magnetic object. However, the
magnetic field should not be higher than the critical field value at the operating
temperature or the superconducting order is destroyed.
The really significant application in sensors is the Josephson effect and corresponding
Josephson junction used in high precision measurement set-ups, as in the case of
SQUID magnetometers. This junction is a superconductor-insulator-superconductor
device allowing electron tunneling when a voltage is applied across the junction. In
fact, SQUID magnetometers are the reference standard in magnetic field measurement
because their response changes in a quantized way, depending on the number of
magnetic quanta, i.e.Bohr magnetons, tunneling through the junction. This effect will
be discussed in detail in section 5.

3.2.2 Paramagnetism
Paramagnetism is also a weak form of magnetization varying linearly with and
proportionally to the applied field (Fig. 3.1b1). Some saturation may be observed at
high fields. The susceptibility χ=M/H of paramagnetic materials is constant, positive
and in the order of 10-5 to 10-3.
The origins of paramagnetism lie in the forced alignment in the direction of an
externally applied field of the magnetic moments of isolated atoms or uncompensated
spins. At zero field, these magnetic moments are randomly oriented resulting in zero
net magnetic moment. As the field increases an increasing number of individual
magnetic moments rotates towards the field direction thus increasing the net magnetic
moment.
Paramagnetic materials include several salts, i.e. solids containing magnetic ions like
Cr, Mn, and Fe (transition metals, uncompensated spins in the 3d) or Gd (rare earth,
uncompensated spins in the 4f) isolated from each other by nonmagnetic impurities.
Several metals, e.g. Al, Sn, or Pt are also paramagnetic. Ferromagnetic materials also
become paramagnetic above the Curie temperature, Tc.
In the case of paramagnetic salts, the susceptibility is inversely proportional to
temperature T (Fig. 3.1b2) following the Curie law:
C
χ= (3.1)
T
where C is the Curie constant.
In the case of paramagnetic metals, the susceptibility is independent of temperature.
The Langévin [12, 13] theory is used to explain the temperature dependent
paramagnetism while the Pauli and Van Vleck theories [11, 12] provide better insight
to the paramagnetism observed in metals.
The former considers localized orbital magnetic moments not allowed to interact with
each other because they are separated by nonmagnetic material, as in the case of salts.
The magnetic moments m are therefore randomly oriented (Fig. 3.1b3) because of
thermal agitation.
When an external field H is applied at a given temperature T, the energy added is
E=-mH.

16
The magnetic moment will then tend to minimize its energy by aligning with the field.
But first, it will have to overcome the thermal energy, kBT. Langévin showed that the
net magnetization M is given by:
⎛ 1⎞
M = Nm ⎜ coth α − ⎟ (3.2)
⎝ α⎠
μ mH
where N is the number of magnetic moments per unit volume and α = 0 is the
k BT
ratio of the two energy terms. The term in parenthesis is the Langévin function (a
special case of the Brillouin function). In general, the thermal energy of the electron is
much larger than the magnetic field energy of its isolated magnetic moment, implying
that α = 1 . In this case, the series expansion of (3.2) yields:
Νmα
M= (3.3)
3
This states that the net magnetization in paramagnetic materials is proportional to the
applied field and zero at zero field but inversely proportional to temperature. It can
easily be shown that the resulting susceptibility follows the Curie law (3.1). The
Langévin theory describes accurately the type of paramagnetism observed in solids
like paramagnetic salts.
However, the even weaker and temperature independent paramagnetism observed in
metals is rather described by the Pauli theory. Pauli paramagnetism is due to the spin
moments of the conduction band electrons in metals. At zero field, the number of
spins parallel to the field is equal to the number of spins antiparallel to it. When the
field is increased, the spin moments parallel to the field increase at the expense of the
antiparallel ones. Because of the Pauli exclusion principle, only the electrons whose
total energy exceeds the Fermi level Ef= kBTf will be allowed to rotate and occupy
new electron states, i.e. only the fraction (T/Tf) of the N magnetic moments will
contribute to the net magnetization. The resulting susceptibility is then independent of
temperature T:
3 T 3Nm 2
χ= Nmα = μ 0 (3.4)
2 Tf 2k BTf

3.2.3 Ferromagnetism
In the cases of dia- and paramagnetism, examined so far, a weak form of magnetism
emerges as the result of the isolated spin or orbital magnetic moment interaction with
the applied field. While, in the cases of ferro-, antiferro- and ferrimagnetism, the
magnetization is due to the alignment of spin magnetic moments through interaction
with the field as well as with neighboring spins. Among the three, the simplest to
describe as well as the most widely used are the ferromagnets and will be introduced
first, thus serving themselves as an introduction to the description of the other two
types of magnetic materials.
In the case of ferromagnets, the magnetization can be nonzero at a zero field while
even a small increase in field can produce a high increase in magnetic moment. Their
susceptibility is several orders of magnitude higher than the one observed in
paramagnetic materials ranging from 50 to 10,000 and more. A typical M(H) curve
for a ferromagnet is the hysteresis loop and virgin curve shown in Fig. 3.2a (or Fig.
2.3a) which clearly suggests that the susceptibility is, typically, a highly nonlinear
function of applied field and the underlying magnetization mechanism is far more
complex than the ones described so far.

17
 M
M=0
H=0
H

(a) M>0
H>0
M
(c)
M=Msat
T H=Hsat
Tc
(b)
H
(d)

Figure 3.2 Ferromagnets. (a) Magnetization vs applied field, hysteresis curve (b)
Magnetization vs. temperature – the Curie point, Tc. (c) Self-aligned spin
configuration due to exchange coupling (d) Magnetization process from zero
magnetization (M=0, H=0) to saturation (M=Msat, H=Hsat). The effect of
domains.

The large number of uncoupled 3d- (transition metals) or 4f- (rare earth) spins of a
typical ferromagnet results in strong spin-spin interactions serving as an ordering
mechanism. The spin-spin interaction is called exchange interaction (see 3.3.1) and
forces the spins in ferromagnets to lie parallel to each other (Fig. 3.2c) thus giving rise
to a high net magnetic moment.
The exchange energy is strong enough that it overcomes the thermal agitation of the
spins tending to randomize their magnetic moments. However, there is a critical
temperature, the Curie temperature, beyond which the thermal energy overpowers the
ferromagnetic ordering and the material behaves as a (Langévin) paramagnet (Fig.
3.1b2):
C
χ= (3.5)
T − Tc
Therefore, at a macroscopic level, the susceptibility is high and constant for a wide
range of temperature.
Ideally, in a monocrystalline solid of a given size perfect alignment of all the
uncompensated spins could be reached even at zero field thus resulting in a very high
spontaneous magnetization, the so-called magnetic monodomain structure. However,
in reality, the net magnetization of a ferromagnetic substance at zero field is lower
than the saturation (maximum) value of magnetization. Further more, as suggested by
the hysteresis curve of Fig. 3.2a, an infinite number of magnetization values can be
attained depending on the field sequence followed. Obviously, exchange interactions
and temperature effects are not enough to describe the mechanism of the
magnetization process.
The alignment of spins locally leads to an increase in the magnetostatic energy forcing
the break down of the ferromagnetic ordering into magnetic domains. The
magnetostatic energy is the result of dipole-dipole interactions or long-range
interactions between all the magnetic moments in the sample. It is a function of the

18
shape and the size of the ferromagnet and the existence of impurities or nonmagnetic
inclusions.
The magnetic domain (see 3.3.5) is a finite size region in the magnetic material within
which the spin magnetic moments are aligned in the same direction because of
exchange interactions. Fig. 3.2d depicts schematically the transition of a
ferromagnetic material from the demagnetized to the saturation state as a uniaxial
field, gradually increasing from zero, is applied along one of the seemingly main
magnetization directions. The M vs. H plot of this process would yield the virgin
curve. If one thinks of a magnetic domain as a giant dipole, then a ferromagnetic
material can be thought of as a collection of dipoles, oriented along different
directions. The net magnetic moment of the domains-dipoles along a given direction
is the aggregate magnetization measured along this direction for a given field or
absence of field.
The preferred orientation direction of a domain depends on the anisotropy axis of the
domain and may vary between domains as in polycrystalline materials. The
anisotropy gives rise to the anisotropy energy that competes against the exchange
energy in the formation of the domain walls separating magnetic domains.
Another important energy term involved in the domain formation is the
magnetoelastic energy that pits the elastic bond energy against the crystalline
magnetic order.
Popular ferromagnetic materials include Fe, Co, Ni, rare earths, and a big number of
alloys [25]. Soft ferromagnets with high permeability are used in electromagnets and
electromagnetic relays. If they also have low conductivity, leading to lower eddy
current losses, they are used in transformers. Soft ferromagnets with good elastic
properties are used as magnetostrictive sensors. Hard ferromagnets find uses as
permanent magnets in motors and generators or as recording media [26]. Even though
high permeability and high coercivity are desired in the two latter cases of hard
ferromagnets, a high energy product (BH) is necessary in the first case while high
squareness S and coercivity squareness S* in the latter.
In any case, the tailoring and control of the parameters, properties and magnetization
process of the material of choice for a given application is of paramount importance.
Similarly, a basic understanding of the phenomena described so far in relation to the
material properties is necessary for the development of novel materials, devices and
sensors. Both remarks couldn’t apply more than in the case of sensors as discussed in
section 5.

3.2.4. Antiferromagnetism
The antiferromagnets can be thought of as consisting of two sublattices with opposite
magnetizations. Each one of the sublattices is ferromagnetically ordered with a
positive exchange constant (Eq. 3.6) while the exchange coupling between the two
lattices is negative (J<0) thus favoring the antiparallel alignment (see 3.3.1).
Therefore, the net magnetic moment is zero at zero field. Inside a magnetic field, the
magnetic moments of each sublattice rotate as to align with the applied field thus
leading to an increase of the magnetization of one lattice at the expense of the other
and therefore to a nonzero weak net magnetization. This M-H dependence
macroscopically resembles that of a paramagnet and it is the reason why
antiferromagnets cannot be detected macroscopically. The antiferromagnetic ordering
is revealed only through neutron diffraction or Mössbauer spectroscopy [13].
Indeed, the susceptibility of antiferromagnetic materials is very low as in
paramagnetic materials but the temperature dependence of the susceptibility is

19
characteristically different (Fig. 3.3a). An antiferromagnet undergoes a phase
transition at the Néel temperature, TN, above which it behaves paramagnetically
following Eq. 3.5. The linear extrapolation of the paramagnetic susceptibility yields a
negative Curie temperature, TC.

1/χ
1/χ//

1/χ⊥

T
TC TN
(a)

(b)

FM1

NM

FM2

(c)

Figure 3.3 Antiferromagnets. (a) Inverse susceptibility vs. temperature – the Néel
temperature, TN, and the predicted Curie temperature, TC, the paramagnetic region
(T>TN) and the antiferromagentic ordering region (T<TN) with parallel and
perpendicular components of the susceptibility in a cubic crystal (b)
Antiferromagnetically ordered cubic crystal, magnetic sublattice A (solid dots),
non-magnetic sublattice B (open dots) (c) Antiferromagnetically coupled
recording thin film structure consisting of a hard (FM1) and a soft (FM2)
ferromagnet coupled through a thin nonmagnetic layer (NM).

The antiferromagnetic ordering takes place below TN. For a cubic crystal with
uniaxial anisotropy, if the field is applied parallel to the easy axis, the material tries to
maintain the antiferromagnetic arrangement and the zero net magnetic moment. It
therefore resists the applied field resulting in lower susceptibility at lower
temperatures (χ//) where the thermal energy assistance is lower. For fields
perpendicular to the easy axis, the measured magnetization is the result of rotations
and is, therefore, independent of temperature (χ⊥). In polycrystalline materials, the
susceptibility below TN is a mixture of the two components.
Antiferromagnetic materials typically consist of transition metals with an unfilled 3d-
shell and nonmagnetic elements with an unfilled p-orbit, e.g., oxides and chlorides of
Fe, Ni, Co, Cr, Mn (fig. 3.3b).

20
The antiferromagnetic coupling mechanism is called superexchange and is used to
describe the coupling between the spin moments of the transition metals (sublattice A)
through the p-electrons of sublattice B [12, 13].
The antiferromagnetic coupling mechanism is increasingly exploited in other thin film
structures with enhanced material properties. For example, antiferromagnetically
coupled recording media (Fig. 3.3c), consisting of one hard (FM1) and one soft (FM2)
magnetic structure antiferromagnetically coupled through a nonmagnetic layer (NM),
offer higher recording densities [27-31]. Or, spin valves, with an enhanced GMR of
tens of percent, are sandwich structures of a hard and a soft ferromagnetic film
antiferromagnetically coupled through a very thin non-magnetic tunneling layer [32-
36].

3.2.5 Ferrimagnetism
The structure and ordering mechanism of ferrimagnets is similar to antiferromagnets,
i.e., they consist of two sublattices antiferromagnetically coupled through
superexchange. However, the net magnetic moment is nonzero and can attain high
values because the magnetizations of the two sublattices are not equal to each other
[12]. As a result, ferrimagnetic materials behave like ferromagnets on a macroscopic
scale; they exhibit high magnetic moments, hysteresis and saturation below the Curie
temperature, TC, and become paramagnetic above it.
However, they have a much more complex structure than ferromagnets which makes
their susceptibility dependence on temperature more complicated as well and their
properties more interesting. The magnetization may vanish not only at TC but also at a
lower temperature known as the compensation point depending on the relative
magnetization and TC values of the two sublattices. For example, let’s take the case of
a structure where one sublattice exhibits a high spontaneous magnetization and a low
TC with respect to the other. At low temperatures, the net magnetization will be
parallel to that of the dominant first sublattice and the material will behave as a
ferromagnet. As the temperature approaches the TC of the first sublattice the
ferromagnetic ordering in the first sublattice will start breaking down and its
magnetization will be decreasing until it vanishes and the sublattice starts behaving as
a paramagnet. At this point, the other sublattice, still ferromagnetically ordered,
prevails and the magnetization is of opposite sign and smaller magnitude than that at
low temperatures. The temperature where the crossing from positive to negative
magnetization occurs is the compensation temperature [13]. Therefore, other
structures may be designed and developed with respect to desirable magnetic and
temperature properties through the tailoring of their spontaneous magnetization and
Curie temperature points.
Ferrimagnets, according to their crystal structures, are classified into ferrites, other
metal oxides, perovskites, and garnets [12]. Typical crystal structures of ferrimagnets
include the spinel and inverse spinel, or intertwined cubic, hexagonal and
rhombohedral structures. It is, however, because of this very variety and complexity
in crystal structure that they cover a wide spectrum of electrical, elastic, magnetic and
optical properties making them appropriate for several applications.
Ferrites are metal oxides of the form MO-Fe2O3. Because they are oxides, they have a
higher resistivity than metallic magnetic materials, i.e. ferromagnets, and therefore
lower eddy current losses. This makes them appropriate candidates for high frequency
applications. If M is a transition metal the compound is magnetically soft, ideal for
use as a soft core in electromagnets and transformers. M can also be Ba, Sr or Pb in
which case the ferrite is hard and can be used as permanent magnet or in magnetic

21
recording (barium ferrite) [11]. Ferrites also exhibit interesting magnetostrictive
properties. Garnets are minerals or possibly technical materials of the form 3M2O3-
5Fe2O3 with very good optical and high frequency properties [12].
Other types of magnetism also include metamagnetism and parasitic magnetism [12].
The transition of some antiferromagnetic materials to a ferromagnetic ordering
because of an applied field or temperature is referred to as metamagnetism. Parasitic
ferromagnetism refers to certain antiferromagnetic materials like α-Fe2O3 that exhibit
some weak ferromagnetic ordering and spontaneous magnetization below the Néel
temperature.

3.3 The magnetization process

In this section, the magnetic ordering mechanisms and the corresponding

magnetization processes are overviewed. The discussion is focusing on the
ferromagnetic substances, although it may apply to other types of magnetic materials
exhibiting magnetic ordering, such as ferrimagnets. The primary mechanism of
magnetization reversal in granular media is the domain and domain wall dynamics. In
general, a ferromagnetic substance is made up of several domains, regions of uniform
magnetization that can be considered as giant dipoles. The magnetic dipoles that make
up a domain are aligned because of the short exchange interactions along a specific
direction imposed by the anisotropy. The size of a domain depends on the competition
between the exchange and the internal magnetostatic energy that builds up because of
the long-range dipole interactions. Domain walls, regions within which the
magnetization changes orientation, separate the domains. The size of the domain walls
depends on the competition between the anisotropy, tending to keep the spins as close
as possible to a local easy axis and the exchange energy tending to keep neighboring
spins aligned.

3.3.1 Exchange interactions

Exchange interaction is the term used to describe the interaction between neighboring
spins. It is also called short-range interaction because it mainly involves nearest
neighbors. The number of nearest neighbors depends on the kind of crystal lattice
involved, e.g. there are 8 nearest neighbors in the case of a BCC crystal (Fig. 3.2c).
As implied by the exchange energy involved in the mean field approximation of the
Weiss theory of ferromagnetism [11-12],
E ex = −μ 0mJ ′ ∑ m j , J’>0 (3.6)
nearest
neighbors

or the exchange Hamiltonian involved in the Heisenberg model [11-13],

H = −2 J ∑ ∑ sis j , J>0 (3.7)
nearest
neighbors

the most favorable arrangement, minimizing the exchange energy of the solid, is the
parallel alignment of spins.
(J’ and J are the mean field and exchange constants respectively).
The positive exchange constant refers to ferromagnets.
In antiferromagnets and ferrimagnets the exchange constant is negative, thus favoring
the antiparallel alignment of spins.

22
3.3.2 Magnetostatic interactions and demagnetizing field
Magnetostatic interactions, also known as long-range or dipole-dipole interactions,
give rise to an energy term called the (self)-magnetostatic energy density
μ
Em = − 0 M ⋅ H m (3.8)
2
This energy is due to the interaction of each dipole with the field Hm created by all
other dipoles in the medium. The ½ factor us used to avoid double counting. Because
dipole interactions are essentially Coulomb interactions (Eq. 2.3) they tend to keep
dipoles as far apart form each other as possible. In this sense, the magnetostatic field
is opposing the effective field of anisotropy and exchange that keep the spins aligned
along a given direction and tends to demagnetize (randomize) the medium. Therefore
the resulting field is also called demagnetizing field Hd. The magnetostatic energy is
clearly non-local, as opposed to anisotropy and exchange, and its calculation is not a
straightforward procedure.
The calculation of Hm involves integration of the magnetization contributions M
throughout the medium. This results in a double integral for the energy density or a
six-fold integral for the volume energy. Self-consistent and finite element calculations
are usually used to compute the self- magnetostatic energy [37-49].
Since the magnetostatic interactions tend to keep the magnetic dipoles apart, surface
charge is established on the surface of the medium or, in the microscopic case, of the
domain. The demagnetizing field is then clearly a function of shape of the domain or
the multi-domain solid and it can be expressed as:
Hd=Nd(M)M
Nd is the shape dependent demagnetizing factor. If Nd is known the total field H’
acting on the medium is
H’=Ha - Hd
where Ha is the applied external field. The demagnetizing field lines inside a bar
magnet are shown on Fig. 2.1b. Even though the calculation of the demagnetizing
factors is clearly not a trivial task, in the case of simple or symmetrical shapes, Nd can
be calculated exactly or with adequate precision. For example, the demagnetizing
factor of a homogeneously magnetized sphere is 1/3. Or, in the case of ellipsoids of
revolution with axes, x, y, z, it can be shown that: Nx+Ny+Nz=1 and Ny=Nz if x is the
long axis of the ellipse. Or, toroids and infinite cylinders have zero demagnetizing
field.
Therefore, in the case of simple geometries the micromagnetic calculation (see 3.2.8)
is greatly simplified to allow for and analytical solutions.

3.3.3 Anisotropy
The term anisotropy is used to explain any preference a material’s property may
exhibit, macroscopically, for a given direction or directions. In the case of magnetic
materials, anisotropy reflects the tendency of the magnetization vector to lie along one
specific direction. The preferred axes are easy axes of magnetization as opposed to the
rest, which are the hard axes. From a thermodynamic point of view, any system or
material tends to minimize its internal energy. Hence, the easy axis is the direction of
the magnetization along which the magnetic material is minimizing is energy.
Depending on the origin, anisotropy can be classified as magnetocrystalline,
magnetostrictive, shape and induced anisotropy.
Magnetocrystalline anisotropy refers to a preference induced by the crystalline
anisotropy of the material. It follows the crystalline anisotropy along the major
crystallographic axes, e.g. the (100) in the BCC Fe-based or the (111) in the Ni-based

23
ferromagnets, or the uniaxial anisotropy with the easy axis along the only existing
anisotropy axis, e.g. the c axis in the hexagonal Co structure.
The origins of magnetocrystalline anisotropy lie mainly in the spin-orbit interactions
in a crystal. Because the electron orbits are bound to the crystal structure, the electron
spins prefer to align, through the exchange coupling mechanism, along a specific
crystallographic direction, thus minimizing the energy of the crystal. Anisotropy is
therefore responsible for the choice of the orientation direction while exchange is
responsible for the aligning itself [12-13].
The anisotropy energy density (energy per unit volume) is of the form:
E a = K1 sin 2 φ + K 2 sin 4 φ + ... (3.9)
φ is the angle between the magnetization direction and the easy axis. The anisotropy
energy density term is minimized (zero) when the two coincide while it increases with
an increase in φ reaching a maximum when the magnetization is aligned with a hard
axis.
K1 and K2, etc… are the anisotropy constants deduced from experiments or calculated
from first order principles. They are a function of temperature and are related to the
crystal structure. K1 is the most important constant, K2 is smaller or zero, K3 is related
to the azimuthal and is negligible, while the higher order terms are of no practical
importance in the case of magnetic materials. The sign of the anisotropy constants can
be either positive or negative and their magnitude ranges from several hundreds of
J/m3 to several hundreds of kJ/m3.
Cubic crystals possess cubic anisotropy with the easy axis lying along some principal
crystallographic direction (100) or (111) etc [13]. Using the direction cosines, α1, α2,
α3 to express the magnetization direction with respect to the principal crystallographic
axes, the energy density is then of the form:
E a = K1 ( α12 α 22 + α 22 α 32 + α 32α12 ) + K 2α12α 22α 32 (3.10)
The M(H) curve in anisotropic materials depends on the direction of the applied field.
For example, in many cases, when the field is applied along the easy axis, the loop is
hysteretic while along the hard axis, the loop appears to be anhysteretic (Fig. 2.3a).
For other directions, the loop is between these two limiting cases.
Anisotropy is usually measured through torque curves. Torque meters apply a large
rotating field of constant magnitude to the specimen, forcing the magnetization vector
to align with it, and measure the resulting torque or rotate the under measurement
sample detecting the softer or easier anisotropy axis.
In the simple case, that the specimen possesses uniaxial anisotropy with energy
density:
E a = K1 sin 2 φ (3.11)
the torque can be calculated by
∂E
L = − a = 2K1 sin φ cos φ = K1 sin 2φ (3.12)
∂φ
and from the torque measurement, the first anisotropy constant can be deduced.
φ is the angle between the easy axis and the magnetization vector. In this case,
because the applied field is large enough, the magnetization vector follows the applied
field without hysteresis and the field and magnetization vectors are co-linear. Note
that as expected, the torque is maximized when the field is perpendicular to the easy
axis. In the case that several forms of anisotropy are present the torque curve has
higher harmonics. Then, Fourier analysis may be used to estimate the various
components.

24
The torque can be also calculated as the cross product between the magnetization and
the applied field (Eqs. 2.10, 2.11). In general, the magnetization lags the applied field
when applied at an angle θ to the easy axis (Fig. 3.4). If the associated anisotropy
energy density is given by Eq. 3.10 and the field is assumed large enough so that θ= φ
and M=Ms, then Eq. 3.12 yields:
L = M × H = MH sin φ = 2K1 sin φ cos φ (3.13)
H 2K1
cos φ = where HK = (3.14)
HK Ms

hard
axis
H

Ms
θ easy
φ
axis

Figure 3.4 The magnetization lags the applied field vector when applied at an angle to the
easy axis.

HK is called the effective anisotropy field and is a measure of the anisotropy strength
of a material and the field that needs to be used in order to overcome the lag due to the
anisotropy. In the case of ideal uniaxial single-domain ellipsoidal particles of
revolution, the anisotropy field corresponds to the coercivity [50] but in realistic
media the relationship between the two is much more complex and beyond the scope
of this paper.
The magnetostrictive anisotropy is associated to the effect of magnetostriction (see
4.2). Certain materials tend to elongate or shrink along the direction of the
magnetization or, inversely, the magnetization tends to align itself with an applied
stress or perpendicular to it. The stress can be external or internal inducing a global or
local magnetostrictive anisotropy axis. In this case, the form of the anisotropy energy
density is similar to Eq. 3.9. Actually, the effect of magnetostrictive anisotropy is
incorporated in the measurements of the anisotropy constants and cannot be
decoupled from the magnetocrystalline anisotropy unless calculated from first order
principles [13] or possibly deduced through Fourier analysis of torque curves [12].
A material can also become anisotropic along some direction after appropriate
treatment, called annealing. The resulting uniaxial anisotropy is called induced
anisotropy and can be realized through a variety of annealing methods. A classical
technique is magnetic annealing, which consists in the heat treatment of a ferromagnet
inside a magnetic field inside an inert atmosphere. The magnetic field is applied along
a desired direction. The resulting anisotropy depends on the magnitudes of
temperature and field. If weak induced anisotropy is desired the temperature is set
around the 1/3 of the melting point of the substance. Accordingly, the magnetic field
could be well below the desired anisotropy field. If strong induced anisotropy is
desired a temperature in the order of 2/3 of the melting point and a field well above
the anisotropy field are chosen, provided that the combination of temperature and
field does not cause the material to undergo a phase transformation.

25
Recently, the stress-current annealing is being widely used, especially in industrial
environments because it does not need an inert atmosphere. Electric dc or ac current
flows through the magnetic substance causing both a magnetic field perpendicular to
the direction of the current flow as well as a temperature increase. Additionally,
tensile or compressive stresses may also be applied acting as effective field.
Alternatively, stress may be applied by itself in repetitive cycles to induce special
kinds of anisotropy, like the creep-induced anisotropy. Cold-rolled or cold-worked
ferromagnets may exhibit, the so-called, roll magnetic anisotropy.
A phase transition in a crystal may also result in induced uniaxial anisotropy, e.g. a
crystal undergoing a transformation from cubic to some other type with lower
symmetry, like hexagonal or orthorhombic. The theories about the mechanism
through which the induced anisotropy is realized vary depending on the method and
the type of magnetic material used [12]. Annealing is widely used in ferromagnets
while in ferrimagnets it is effective mainly in the case of oxides. If two-phases of
anisotropy are desired, chemical processes and/or combination of the above
mentioned techniques may be used.
Shape anisotropy is related to the geometry of the magnetic substance and can be
thought of as the energy required in establishing a magnetization along a direction
least favored by the shape of the material. For example, in an ellipsoidal specimen, the
easy axis will most likely lie along the long axis of the ellipse. This comes as a result
of both the crystallographic order and the interplay of magnetostatic and exchange
interactions in the domain formation.
The surface anisotropy depends on the shape of surface. It is observed in
inhomogeneous structures such as thin films deposited on magnetic or nonmagnetic
substrates. Because of the accumulation of magnetic charge and the increased
magnetostatic energy at the surface (or interface), an easy axis other than that in the
bulk structure may be observed at the surface.
The anisotropy energy is temperature dependent. As temperature increases, the
magnetic moments diverge from the easy axis orientation over a wider distribution of
angles. For small angle φ,
M (T ) : M ( 0 ) cos ϕ (3.15)
In the case of uniaxial anisotropy,
3
⎛ M (T ) ⎞
EK (T ) ≈ EK ( 0 ) ⎜⎜ ⎟⎟ (3.16)
⎝ M ( 0 ) ⎠
The exponent is even higher for other types or higher orders of anisotropy.

3.3.4 Magnetoelastic energy

Another energy term, which is usually grouped with the anisotropy term, is the
magnetoelastic energy. It is mainly associated with the effect of magnetostriction. The
Coulomb interaction (Eq. 2.3) between two dipoles induces a strain in the bond length
resulting in the deformation (elongation or shrinking) of the crystal along the direction
of the magnetization. This is a short-range interaction involving nearest neighbors
mainly, resembling in form the anisotropy energy of Eq. 3.10. It is a function of lattice
strain and domain magnetization orientation. The magnetoelastic constants involved
depend on the bond length and the crystal structure. The magnetoelastic energy is
counterbalanced by the elastic bond energy opposing changes in the bond length and
ensuring that the equilibrium is reached at a finite strain. Therefore, the resulting
magnetostrictive constants λ of a material are a function of the elastic moduli of the
crystal and the magnetoelastic constants [12]. Even though this is a local effect, the

26
deformation of the crystal leads to a rearrangement and deformation of the domains
resulting in global strain along the magnetization direction [9, 51-53]. The
deformation may be caused by either a simple strain in the bond length as in typical
magnetostrictive materials or by the phase transformation of the crystal as in magnetic
shape memory alloys.

3.3.5 Domains and domain walls

In order to control the magnetization process and the related effects and macroscopic
properties, a basic understanding of the domain formation and wall dynamics is
needed. The domain and domain wall dynamics are primarily responsible for most of
the magnetic effects used in sensing applications.
Atomic spins in ferromagnetic materials tend to align because of the short-range
exchange interactions. The extent of this local effect depends on the strength of the
exchange. The direction of alignment is decided by the anisotropy of the material. The
increasing number of aligned spins along a certain direction results in an increase in
the non-local magnetostatic interactions giving rise to demagnetizing fields opposing
the process of alignment. In order for the material to accommodate this increase in
energy, it breaks down to regions uniformly magnetized along directions antiparallel,
perpendicular or at other angles to each other (Fig. 3.2d). This process can be
macroscopically understood by observing the behaviour of several bar magnets placed
parallel next to each other: they will tend to rotate either antiparallel or at angle to
each other.
The shape, the size and the number of the domains depend on the atomic structure and
ensuing magnetic properties (susceptibility, anisotropy and exchange constants, etc)
of the elements involved, the built-in strains because of the elements involved or the
preparation procedure (giving rise to local anisotropy axes), the inhomogeneities or
voids (resulting in local strains and stray fields), the shape of the bulk material and
other parameters.
When a particle of a magnetic substance is small enough so that long-range
magnetostatic interactions cannot overcome the short-range exchange interactions, the
particle can support one domain only and is referred to as single domain particle or
grain. The maximum size below which single domain particles exist is a rather
complicated function of the material properties. In some cases the properties of single
domain devices may be useful for sensing applications, e.g. as in the yttrium garnet
based fluxgates (see 5.1) but care should be taken that their size is not reduced beyond
the superparamagnetic limit (see 3.6).
There is a big variety in the domain configurations. The configuration shown in Fig.
3.2d is the simplest one but typical in ferromagnetic thin films. The vertically oriented
domains are called closure domains offering a return (closure) path for the flux of the
magnetostatic fields.
Domains are separated by domain walls. In the demagnetized state, the domains’
magnetizations, oriented in various directions, cancel out. In the saturation state, the
applied field helps them overcome the demagnetizing field and it looks like there is
only one domain oriented along the applied field direction or that there are still some
separate domains left but they are all aligned. Therefore, during the magnetization
process, the domain walls move changing the shape, the size and the number of the
domains.
A domain wall is a transition region of finite thickness inside which the magnetization
changes its orientation from one domain to the other. This reorientation can be seen as
a continuous rotation of the magnetization vector inside the domain wall. The domain

27
walls are 180o walls when they separate regions of antiparallel orientation. In the case
of closure domains, non-180o walls are observed. Usually, they appear in 90o, 71o or
109o depending on the crystalline anisotropy axes.
Depending on the plane of rotation of the magnetization from one domain to the other,
the walls are classified into Bloch-type and Néel-type walls (Fig. 3.5a-b). i.e. in the
case of an 180o wall, with the magnetization lying along the y axis and its rotation in
the zy plane which is normal to the wall plane (xy), a Bloch wall occurs (Fig. 3.5a);
while when the rotation is confined in the domain wall plane, a Néel wall emerges
(Fig. 3.5b). The out-of-plane component in Bloch walls, results in the accumulation of
surface charge which, in the case of thin films where the surface effects are more
prominent than volume effects, gives rise to a high demagnetizing energy forbidding
the rotation of the magnetization vector and Néel walls are preferred. The in-plane
rotation of Néel walls results in the accumulation of volume charge which is
unfavorable in thicker films or bulk materials.
y

x
(a) (b)
z

(c) (d)

(e) (f)
Figure 3.5 Types of domain walls (a) Bloch wall vs. (b) Néel wall (c) Zigzag walls (d) fir
tree pattern for 180o basic domain walls and (e) for 90o basic domain walls (f)
cross-tie walls

The thickness of the domain wall is the result of the equilibrium between the
competing anisotropy and exchange energies. The exchange, in ferromagnets, keeps
the spins aligned tending to form wide walls. The anisotropy opposes the spins
orienting at an angle to the easy axis tending to minimize the wall width.

28
 easy
axis

easy
axis

H=0

easy
axis
(a)

easy
axis

easy
axis

H=0

easy
axis
(b)
Figure 3.6 Single domain particle (a) reversible rotation (b) irreversible switching

Indicatively, for a crystal with uniaxial anisotropy, the thickness can be expressed as
A
δ=π (3.15)
Ku
and the surface wall energy (per unit area) as
γ = 4 AK u (3.16)
Ku is the uniaxial anisotropy constant and A is the exchange energy constant related to
J of Eq. (3.6) and the crystal lattice configuration and properties.
When a magnetic field is applied, the domain walls move as to allow for the
alignment of the domains’ magnetizations along the direction of the field. The

29
alignment process includes domain wall motion dynamics and modes as well as
domain rotation.

3.3.5a The effect of magnetic field on the domain walls

The domain wall motion includes both reversible and irreversible bowing and
translation of the wall (Fig. 3.7). The presence of an applied field H is “sensed” by the
wall as pressure, P, (force per unit area) on the wall, forcing to move:
P==2μ0MsHcosθ (3.17)
θ is the angle between the domain easy axis and the field. This pressure is equivalent
to the wall energy density.

Initial domain wall Propagating domain wall

(a)

Initial domain wall Propagating domain wall

(b)

Figure 3.7 Domain wall (a) bowing (b) translation

Ideally, in uniform materials with constant wall energy throughout the surface, i.e.
pinning effects are absent, in the presence of a field, the domain walls should move
linearly and reversibly parallel to the wall surfaces. In practice, however, an expansive
motion, known as domain wall “bowing”, is observed. This suggests that the wall
energy is not spatially constant, e.g., there are pinning sites holding the edges or
certain points of the wall “pinned”. In the absence of defects or voids these pinning
sites still exist as the edges of the material function as boundary conditions. If the
field is not strong enough to overcome the pinning effect, the wall bows in the
direction of the field and returns to its original position when the field is removed. If
the field increases beyond a certain magnitude the reversible bowing is followed by an

30
abrupt and irreversible displacement of the wall along the direction of the field, which
can be considered as a type of Barkhausen jump (see 2.9a). If, however, the wall
expansion exceeds a certain curvature it is possible that the wall does not return to its
original position upon removal of the field and stays bowed. The critical field for
irreversible expansion is a function of the wall surface energy, γ and the distance
between two pinning sites, l:
γ 2γ
H0 = for a 180o wall or H 0 = for a 90o wall (3.18)
M s l cos θ M s l cos θ
The pinning sites that inhibit the motion of domain walls can be explained as high-
energy regions in the material because of increased internal stresses, impurities and
voids. Because the wall energy depends on the microstructure of the material, the
domain wall motion, irreversible or not, is very sensitive in changes in microstructure
effected through intentional treatment or other exposure to fields, stresses or
temperature fluctuations. In general, it can be argued that bowing occurs mainly in the
low energy walls encountered in the soft materials while the parallel motion in hard
materials.
The reversibility of the domain wall propagation partially determines whether a
sensing element will exhibit hysteresis. Hysteresis is desirable in storage applications,
e.g. security sensors (see 5.4), but undesirable in other sensing applications, e.g.
position sensors, where the presence of hysteresis compromises the uncertainty levels
and reliability of the sensors. Since reversible processes are related to a low number of
pinning sites or defects in the material, hysteresis can be controlled through proper
tailoring of the material [54-58]. To avoid hysteresis, one must choose a soft material
that can be prepared into the desired form with as little defects as possible and submit
it to the appropriate annealing procedure to relieve it of internal stresses that may act
as pinning centres. If hysteresis is still a problem, it can be compensated with
modeling, (see 3.4). In the case that hysteresis is desirable, hard materials or
composites of soft and hard materials are used, where defects can be purposefully
introduced in a controllable way [26, 40, 44, 61, 62].

3.3.5b The effect of magnetic field on the domains

As with domain wall propagation, the rotation processes can be either reversible or
irreversible. A helpful analogy to the behaviour of the magnetization inside the
domain is maybe that of a single domain particle with anisotropy, acting as a giant
dipole (Figs. 3.6, 3.8). When subjected to a magnetic field H, the domain
magnetization rotates as to align with it. The rotation, for small angles, is reversible
until the increasing field strength causes the magnetization to snap irreversibly from
the initial easy axis A to a new easy axis B closer to the direction of H (Fig. 3.8a).
This irreversible rotation of the magnetization is the second type of Barkhausen jump,
also associated with the hysteresis observed in magnetic sensors. More often than not,
the new easy axis B is still not in line with the axis of H, so the magnetic dipoles will
try to further align with H through reversible rotation, as shown in (Fig. 3.8b), until
saturation is reached. Upon the removal of the external field, the domain
magnetization falls back onto the latest easy axis B. In general, magnetic domains do
not return back to their initial easy axis A.
The irreversible domain wall motion and domain rotation are the two sources of
magnetic noise in magnetic sensors. So, if a sensor with negligible hysteresis and
noise is desired, it should be operated along the reversible part of the hysteresis curve.
This can be obtained by proper choice of the magnetic material as well as through

31
proper tailoring to increase the reversible part of the response. This way the operating
field range of the device can be controlled.

Easy axis A Easy axis A

(a)

H
(b)

Figure 3.8 (a) Hysteretic domain rotation (b) Anhysteretic domain rotation

32
Annealing techniques can also be used, not only to eliminate the defects of the
material, like in domain wall dynamics, but also to re-orientate the magnetic easy axis
in such a direction as to eliminate the irreversible swift process of the domain rotation.
Elimination of the irreversible process takes place simply due to the absence of easy
axis direction close to the external field H, e.g. the magnetic field used in field
annealing should be perpendicular to the easy axis of the material.
In case that one of the two types of Barkhausen jumps is utilized in a given sensing
mechanism, then the other one can be eliminated through proper material choice
followed by proper annealing to reduce the noise levels.
Both reversible and irreversible processes are associated not only with the
magnetization process but also with the, much used, effect of magnetostriction (see
4.2). The dynamic behavior of these processes can result in elastic waves, propagating
along the magnetic substance [63-65, 3]. Domain rotation has found applications in
the field of mechanical sensors [66].

3.3.5c Types of domain walls

The domains are characterized by the mode of magnetization rotation into Bloch and
Néel walls (Fig. 3.5a-b). Bloch walls are favored in bulk structures where the build up
of surface charge is preferable to that of volume charges. On the contrary, Néel walls
are preferred in thin film structures where surface charges result in very strong
demagnetizing fields. Recently, the geometrically constrained wall, related to the
ballistic magnetoresistance (see 4.3.1), is reported. It forms inside a constriction
separating two wider regions and its properties and structure are different than the
unconstrained Bloch or Néel walls [67].
Another way to categorize domain walls is with respect to their configurations
observed with different types of microscopy as those based on the Bitter, Kerr and
Faraday effects, TEM (Transmission Electron Microscopy), MFM (Magnetic Force
Microscopy) or neutron and X-ray diffraction methods [68]. The variety of
configurations, which depends on the crystal structure as well as on the preparation
process and subsequent treatment, reflects the complexity of microstructures
encountered in magnetic materials.
As a rule, domain walls tend to minimize their energy by opting for the wall
configuration with the smallest surface. Therefore, the ideal low energy wall is the
180o separating regions of opposite magnetization. 180o walls are observed in soft
materials with cubic anisotropic, the easy axis parallel to one of the cube’s edges and
low magnetostriction. When the magnetostatic self-energy dictates the formation of
closure domains (Fig. 3.2d), 90o walls are formed to separate domains with
magnetizations perpendicular to each other. 90o walls are related to enhanced
magnetostrictive properties. If the easy axis is along (111), i.e. K1 < 0, 180o walls are
also formed but the corresponding closure domains form 72o and 108o walls as the
magnetization is forced to rotate along another easy axis.
In cases where the material is forced to change its magnetization state through a high
energy transition, e.g. the magnetization has to rotate through a hard axis, the area
minimization is not the only criterion deciding the domain wall configuration. Zigzag
walls (Fig. 3.5c) are a typical example of such a wall. The increased surface actually
leads to a lower energy state even though it also adds elastic stresses to the crystal.
Fir tree patterns (Fig 3.5d-e) emerge when the crystal is slightly misoriented with
respect to the closest easy axis. They correspond to shallow surface domains that
minimize the effective wall width at the surface. Their role is to collect the net flux
escaping from the surface and direct it to domains of reverse polarity [68].

33
In defect-free thin films, the domain configuration is dominated by 90o and 180o Néel
type walls. Again, the system is trying to minimize its energy, so the domain wall
length is minimized. In thin films or ribbons with low anisotropy, the stripe domain
configuration is very common. However, there are cases where wall configurations of
increased wall length are preferred as in the case of 90o cross-tie (Fig. 3.5f)
established in the place of 180o walls.
Other configurations include the maze or the more “neat” bubble structure observed
on the surface of films with perpendicular anisotropy; the bamboo structure observed
in wires with an inner core and a shell with either circumferential or perpendicular
(radial) permeability; the 360o walls in a zigzag shape, observed in thin films and
generated by the swinging of a 180o wall over a defect; or, spike domains, similar to
the fir configuration, in the neighborhood of defects [68, 12].

3.4 The energy viewpoint and modeling

Brown [69] was the first one to treat magnetism as a system in thermodynamic
equilibrium introducing the term micromagnetics. As Aharoni [13] points out, the
term is misleading since the medium is treated as a continuum and the microstructure
is ignored. However, it serves as a sound basis for calculations used by hundreds of
researchers in the sixty years following its introduction.
According to Brown’s micromagnetics, the free energy of a magnetic system consists
of the three main energy density terms (energy per unit volume), i.e. the exchange E ex ,
the anisotropy E k (Eqs. 3.9, 3.10), and the self-magnetostatic energy Em (Eq. 3.8) to
which the Zeeman energy density of the material’s interaction with the applied field
EH (Eq. 3.19) is added:
E = ∫ { Eex + Ea − Em − EH }dv (3.20)
In the case of magnetostrictive materials the magnetoelastic energy term can be added
in the form of an additional anisotropy term. A surface anisotropy term can also be
added if deemed necessary, e.g. as in the case of thin film interfaces.
Note that the Zeeman and magnetostatic energy counterbalance the exchange and
anisotropy energy. It has been mentioned that, in zero field, the alignment because of
exchange and anisotropy leads to a build up of magnetic charge. This results in a
demagnetizing field that opposes the established magnetization.
Eq. 3.20 is an equation based on classical mechanics and thermodynamics. On the
other hand, the exchange interactions are calculated based on quantum mechanics
principles. Therefore, in order for the exchange energy term to be meaningful, Eq. 3.6
must be rewritten in a “classical” way under the assumptions that the temperature is
not near the Curie point and that the lattice constant is too small compared to other
lengths of importance in the calculation so that the “continuum” formulation is
reasonably valid:
Eex = A ( ∇ ⋅m )
2
(3.21)
A is the so-called exchange constant whose value depends on the crystal structure and
dimensions.
All terms in 3.20 are functions of a position vector, r. Even the anisotropy and
exchange “constants” can be assumed to have a spatial profile [44] if this is to make
the model more realistic. The micromagnetic calculation can start from a well-defined
state such as saturation or a randomized (0,0) state. The energy E at some position r in
the material is calculated as the sum of the contributing energies at point r integrated
over the specimen volume v. If surface terms are included, a surface integral is also

34
calculated and added to the volume term. The stable state for a value of the applied
field Ha is the magnetization vector configuration for which E is minimized. The
resulting minimized energy equation is usually a complicated and nonlinear
differential equation. This equation can be linearized for certain geometries and
simplified under certain assumptions enough to be solved analytically [44, 45, 61]. In
the general case, however, an analytical solution is not possible and finite element or
other numerical calculation [38-41, 43] has to be carried out in order to compute the
magnetization direction at each point in the discretized space.
It must be noted that the micromagnetics treatment is quasistatic, i.e. the time interval
between two successive applied field values is assumed long enough to allow for the
transients to die out. However, it is possible to combine this type of calculation with
the LLG equation (see 3.2.9a) describing the dynamic behavior of magnetization and
carry out calculations for the effect of (high) frequency or temperature.
The major drawback of micromagnetic calculations is that they are time-consuming
and computer-intensive. Simpler modeling approaches have been proposed yielding
reasonably good results for specific cases of materials or applications. A different
treatment yielding an easier to implement model is the Energetic Model (EM)
developed by Hauser [70-71]. Instead of using one energy term for each contribution,
the internal energy of the system is constructed by only three energy terms: the
reversible energy, the irreversible energy and the stray field energy. The Zeeman
energy is then added to obtain the Gibbs energy of the system.
Other approaches are focused on developing phenomenological hysteresis models.
The Stoner-Wohlfarth model describes the magnetization reversal mechanism for an
isolated, single-domain ellipsoidal particle [50]. It has been widely used to study
systems of particles and in conjunction with self-consistent calculations in order to
model more realistic media [72-73]. The Jiles-Atherton model treats hysteresis as the
result of impediments to the wall motion. Three fitting parameters are used but it is
not clear how they can be identified with material parameters [74-76]. Preisach
models have widely been used to model hysteresis in recording media or hysteresis
losses in transformers [77-79] but they also lack a sound identification method
relating their fitting parameters to structural parameters of the magnetic media.
Furthermore, the incorporation of frequency, temperature or even stress effects is still
more or less an open question.
In conclusion, there is no universal model in magnetism that would allow the
modeling of the magnetization behavior based on the structure and microstructure of
any material regardless of its grain or particle shape and orientation, the type of
domain walls forming or whether the material is magnetically soft or hard.

3.5 Dynamic phenomena

Major loop measurements, B(H) curves, of a specimen at constant stress and
temperature but variable field rate, show that the loop becomes wider and shorter as
the cycling frequency of the field increases (Fig. 3.9). This suggests that the
magnetization vector does not have enough time to settle to its steady state value each
time the field changes.

3.5.1 The LLG equation

So far, the magnetization process has been assumed to be quasistatic, i.e. the applied
field is changing at a rate slow enough to allow for the magnetization to reach steady

35
state. When this assumption is no longer valid the dynamic behavior of the
magnetization must be examined.

Figure 3.9 B(H) curves at variable frequency: ω1< ω2

The dynamic treatment of the magnetization is based on the quantum-mechanical

gyromagnetic effect, i.e. the atomic magnetic moment, whether it is due to spin or
orbital motion, has an angular momentum which, when inside an applied field, sets
the magnetic moment to precession until it is aligned with the field the same way that
a spinning top with damping aligns with a gravitational field.
The time dependent behavior of the unit magnetization vector, m=M/Ms is described
by the Landau-Lifshitz-Gilbert equation (LLG) [12-14, 80-82]:
∂m ⎛ ∂m ⎞
= γm × ⎜ H eff − α ⎟ (3.22)
∂t ⎝ ∂t ⎠
γ is called the gyromagnetic constant, and corresponds to the ratio of the magnetic
moment M produced by the spin or orbital motion of an electron over the angular
moment of this motion:
M μe
γ=− = −g 0 (3.23)
P 2me
where e and me are the charge and mass of the electron and g is the Landé factor
assuming the value 1 for orbital and 2 for spin motion.
The effective field Heff corresponds to the vector sum of the anisotropy, exchange, and
magnetostatic fields plus the external applied field as derived by Brown’s energy
equation (3.20). Therefore, the first part of the Eq. 3.22 is the equation of precession
of m under the influence of field Heff .
The second part of the equation describes the damping part of the motion due to the
crystal lattice restoring forces. α is a phenomenological damping parameter that
ensures the eventual alignment of the magnetization vector with field. If α→0 the
equation reduces to the gyroscopic equation and the m vector is simply set to
precession around the field without approaching equilibrium. When the damping is
∂m
strong, α→∞, then the dynamic limit → 0 is reached.
∂t

36
Eddy current effects (see 3.2.11) can also be included in Eq. 3.22 since the time
∂m
varying magnetization vector will induce a current, which in turn generates a field
∂t
Heddy that can be added to Heff. [14].

3.5.2 Magnetic viscosity and aftereffects

The phenomenon of gradual approach of the material to its thermodynamic
equilibrium is called magnetic viscosity. Magnetic viscosity can be observed through
a simple experiment: a field high enough to saturate the medium is applied and then it
is decreased to some lower value and kept there constant. The magnetization will keep
changing for some time after the field stops changing. Magnetic viscosity is also
known as magnetic aftereffect and they are in general highly temperature dependent.
Various models have been employed to explain the phenomenon of magnetic
aftereffect and several types of aftereffect can be distinguished with respect to their
origin or modeling approach [83,84]. In any case, the aftereffect refers to a gradual
change in the state of the medium after the external stimulus has stopped changing.
One way to think of it is as damped oscillations of the magnetization vector towards
the thermodynamic equilibrium. Because of the complex energy landscape of a
magnetic substance, once the magnetization vector finds itself in the vicinity of a
stable solution, it starts moving from one energy minimum to another until the lowest
and most stable is found.
This effect can also be partially explained in terms of domain walls. In section 3.3.5,
we saw that the domain walls under the influence of an applied field, and depending
on the existence of microstrains, local anisotropy and exchange, as well as stray
fields, move to a new position through bowing, pinning, and translation. This is not an
instantaneous effect. So if we think of the domain wall as a membrane being exposed
to competing forces, we can probably think of the aftereffect as vibrations of the
membrane as it relaxes to a new position.
One type of aftereffect is the dissipation aftereffect. It can be explained by the
diffusion process of impurities in the magnetic crystal. During the diffusion process,
the crystal is strained, the magnetoelastic energy changes and strain induced
anisotropy is gradually established forcing the magnetization to align. This is in
general a reversible process allowing good repeatability of a measurement or
experiment. There are ways to counterbalance the diffusion process and return to the
order existing before the aftereffect. When this happens over a long period of time
where irreversible changes in the microstructure are established, the aftereffect is no
longer reversible and is referred to as magnetic aging.
Another type of aftereffect may be attributed to the presence of eddy currents
impeding the change in magnetization under the application of an external field. But
because this is an electromagnetic phenomenon it is not considered part of the
magnetic aftereffect.
The discussion of aftereffect, however, is mostly focused on the thermal aftereffect,
which is due to the thermal fluctuations of the magnetization. The latter may assist the
reversal of unstable domains or particles and impede the magnetization’s course to
equilibrium. The thermal aftereffect is an irreversible process and cannot be
compensated by simply reversing the external stimulus.
Since thermal fluctuations assist the magnetization reversal, the effect of temperature
can be treated as the source of an effective “thermal” field Hth. Indeed, the thermally
activated magnetization reversal can be treated micromagnetically by adding an extra
term Hth to the effective field Heff in the LLG equation (3.22) [83].

37
The magnetic viscosity is phenomenologically described by:
−t
M (t ) = M ( 0) e τ
(3.24)
τ is the thermal relaxation time, defined as the average time needed for the energy of a
magnetic material to attain a new minimum:
K uV
τ = τ0 e kT (3.25)
where kT is the thermal energy barrier at temperature T (k is the Boltzmann constant)
and KuV is the energy of a magnetic particle of uniaxial anisotropy Ku and volume V.
Eq. 3.25 is phenomenological and does not take into account the temperature
dependence of Ku or τ0. However, it is a rough estimate of the relaxation times at
room temperature measurements. If the time-scale of the experiment is much larger
than τ no aftereffect is going to be observed and the magnetic substance is considered
to be thermally stable. In the opposite case, the magnetic material is considered to be
thermally unstable.

3.5.3 Accommodation
The irreversible magnetic aftereffect is related to the phenomenon of accommodation
or reptation that is responsible for the repeatability levels in sensors. When a magnetic
material is cycled between two field values for a large number of times, the minor
loop traced slowly drifts downwards (reptation) or changes slope (tilting) [84-85].
Relaxation is again responsible for accommodation. During the repeated cycling of
the field, the magnetization sinks deeper into lower energy states, as some electrons
migrate to neighboring, more energetically preferable, sites [86]. The multiple local
minima existing in the energy landscape of magnetic substances are strongly related
to the irreversible processes and the hysteresis in ferromagnets. Accommodation is
therefore more prominent in the case of hard rather than the soft magnets used in
sensing.

3.5.4 Ferromagnetic resonance

Ferromagnetic resonance is the phenomenon observed at some value of a large DC
magnetic field HDC applied along some axis of a ferromagnet excited by a transverse
alternating field HAC of frequency ω0 in the microwave range [87](Fig. 3.10). The
atomic moments inside the ferromagnet tend to align with HDC while HAC sets them to
precession motion. The precession angle θ << π because the ac field is much weaker
than the dc field and the atomic moment barely deviates from equilibrium to start the
precession. Its value, i.e. the radius of precession, depends on the strength of the ac
field. The angular velocity of the motion, frequency ω, depends on the strength of the
static field HDC: ω ∝ HDC, the stronger is the field the faster the precession will be.
Resonance occurs at that field value HDC,0 where ω=ω0. At resonance, the real part of
the susceptibility, resembling the frequency dependence of the voltage across LC
tank, becomes zero while the imaginary part, resembling the frequency dependence of
the voltage across a dissipative element R, is maximized. The width of the resonance
curves is proportional to the damping coefficient α (Eq. 3.22) related to dissipative
mechanisms. These include eddy currents induced by the precession motion causing
its damping, or electron hopping for conducting materials, or spin-wave scattering
interacting with the lattice.
Last but not least, the skin effect is observed in FMR. The effect of the microwave
field HAC is restricted at a very thin surface layer (Eq. 3.27). This is important in the
case of GMI applications because it sets a limit to how thin a sensor may become.

38
 z
HAC
φ

HDC
θ
x

y
Figure 3.10 Ferromagnetic resonance

3.5.5 Eddy current losses and skin depth

Eddy currents are currents induced by a time varying magnetization, according to
dM
Faraday’s law (3.22a), whose magnitude is proportional to . They, in turn,
dt
generate a magnetic field opposing any change in magnetization, which if comparable
in magnitude to the applied field gives rise to a loss term that cannot be ignored. The
temporal variation of magnetization may be due to a time varying external applied
field, to the movement of a domain wall or to thermal fluctuations. In the first case,
eddy currents are treated in a macroscopic way by Maxwell’s equations for a
homogeneously magnetized medium. In the second case, eddy currents depend on the
velocity of the domain wall propagation and must be treated in a microscopic,
intermediate scale, a usually demanding task. In the third case, the field generated by
the eddy currents is opposing the magnetization’s alignment with the applied field. It
can therefore be treated either as part of the damping mechanism or as a field
opposing the field acting on the magnetization’s precession (Eq. 3.22) [14].
Macroscopically, this latter type of eddy current losses is related to or masked by the
dissipation mechanism of hysteresis. Eddy currents can also account for the widening
of the hysteresis loop at high frequencies.
In the case of macroscopic eddy currents, the loss is proportional to the square of
dM
and inversely proportional to the resistivity of the material. Therefore, in time-
dt
varying applications, like transformers or electrical machines, ferromagnetic metals
are much more susceptible to eddy current losses as opposed to ferrimagnetic oxides.
Depending on the operating frequency, one can select ferrimagnets instead of
ferromagnets.
Of practical interest is the quantity of skin depth, δ,
2ρ
δ= (3.27)
μω
When an electromagnetic wave is traveling through a lossy conducting medium of
resistivity ρ and permeability μ, the related field magnitude cycled at a frequency ω is
−z
attenuated as e δ where z is the distance traveled and δ is the quantity given by Eq.
3.27. Therefore, the skin depth corresponds to that distance from the surface of the
medium at which the magnetization has decreased by 1/e. The dependence of the
impedance and the permeability of a metallic magnetic material due to the eddy
currents losses is called the skin effect and is proportional to frequency and inversely

39
proportional to the skin depth. If a material is much thinner than δ complete
penetration of the magnetic flux is guaranteed and the skin effect is avoided. Such an
example are the thin laminated sheets of transformers.

3.6 Superparamagnetism
From Eq. 3.25 it is clear that the thermal relaxation time depends on the volume of a
particle or a domain. When a particle is small enough so that τ is very small compared
to the time-scale of the process to which it is exposed, superparamagnetism is
observed. In this case, the magnetization cannot be pinned by the anisotropy energy
along some direction and yield the spontaneous magnetization behavior, typical in
ferromagnets. Instead, the magnetization vector oscillates back and forth along all
possible easy axes under the effect of thermal energy, and the particle or domain is
considered thermally unstable. Because of the resulting low effective anisotropy field,
the behavior of the particles is paramagnetic following the Langévin model (Eq. 3.2).
The difference with paramagnets is that superparamagnetic particles can align and
reach saturation at relatively low fields.
The minimum magnetic volume below which ferromagnetic ordering is destroyed is
known as the superparamagnetic limit:
kT
V0 = (3.26)
2 Ku
Eq. 3.26 is derived from the anisotropic and thermal energy equations expressing he
competition between the thermal energy, kT, and the anisotropy energy, Ku.

3.7 Magnetotransport
Magnetotransport or spin-polarized transport [8] is the result of strong coupling
among the spin, charge, lattice and orbital degrees of freedom leading to spin
dependent scattering, spin tunneling and spin relaxation. Magnetotransport properties
refer to the magnetic field dependence of the conductivity of ferromagnets.
Magnetotransport phenomena are a direct consequence of the quantum mechanical
nature of ferromagnetism.
In section 3.1, magnetism in ferromagnets was interpreted as the result of interacting
uncompensated spin. In order to minimize the energy of the high density of states at
the Fermi level, there is an unbalanced splitting between up (majority carriers) and
down (minority carriers) spin of the d-electron states. The difference between the two
spin bands is proportional to the magnetic moment of an element. Inside a magnetic
field, some of them act as carriers, the so-called itinerant electrons, tunneling through
to neighboring atoms and giving rise to a spin polarized current. In classical physics,
for an electron to appear on the other side of a barrier it must have energy enough to
overcome the barrier. In quantum mechanics, if the barrier is very thin, i.e. d is small,
electrons may tunnel through, provided there is a large number of electrons available
for tunneling and a large number of empty states at the same energy level at the other
side (Fig. 3.11). The transmission and reflection of the carriers in metallic
ferromagnets depends on their spin polarization and the polarization of the states at
the other side of the barrier.
One way to understand the magnetotransport properties at the atomic level is by
studying the case of ferromagnetic sandwiches, also known as spin valves. These
materials consist of two different ferromagnetic layers coupled through a nonmagnetic
layer, acting as a tunneling barrier. In this configuration, carriers of the first layer
tunnel through to occupy unfilled states in the second. When the two layers are

40
magnetized parallel to each other, the tunneling is easier and the conductivity higher
because the spin polarized carriers easily find their way through to empty states of the
same spin. In the case of antiparallel alignment, the majority carriers will attempt to
occupy minority states and vice versa and the probability of scattering is much higher.
Therefore the conductivity of the ferromagnetic structure is maximum when the two
layers are aligned and minimum when they are antiparallel to each other because of
increased scattering.
Potential
barrier
filled empty
states states

electron
Figure 3.11 Tunneling effect

In principle, all ferromagnetic metals should exhibit magnetotransport properties and

the scattering of the itinerant electrons should give rise to a field dependent change in
resistivity. However, magnetotransport phenomena are more prominent when the
mean free path of conduction electrons is much larger than the corresponding Fermi
wavelength as well as larger than other relevant lengths such as the lattice constant,
the grain size, or the interface thickness. Then, the electrons may travel through
several atomic sites, layers, or grains without losing their spin polarization in spite of
any scattering. This has as a result the enhancement of the magnetotransport
properties as the number of scatterers, e.g. layers and interfaces, increases.
Overall, it can be said that at the atomic level, this spin-dependent scattering
mechanism is responsible for the change of resistance in a ferromagnet as a function
of an applied field. Possible scatterers include domain walls, magnetic dipoles, grain
boundaries, imperfections, interfaces, impurities and phonons.
The application of an external field, at the microscopic level, reorients and reshapes
the magnetic domains through rotation and domain wall motion. When current is sent
through a ferromagnet, the measured resistivity is a function of this field because the
scattering centers change as function of the magnetic configuration of the material and
hence, the applied field. This is observed in ordinary magnetoresistive films where the
resistivity of the film is maximized when the magnetization and the current density
are perpendicular to one another and minimum when they are collinear (Fig. 4.7).
In the case of GMR multilayers, structures consisting of alternating magnetic and
nonmagnetic films, most of the scattering takes place at the interface between the
magnetic and the nonmagnetic layer where they dissolve into one another [89-91].
Structures where spin-dependent scattering is of importance include the GMR
multilayers, the spin valves, and the granular media. In the latter, the scattering takes
place at the grain boundaries in a way similar to the multilayers [92-94].
Spin dependent scattering is related to spin dependent tunneling (SDT), the
mechanism used in spin valves. SDT allows current to flow through when adjacent
layers are collinear and stops or severely inhibits conduction when the alignment is
antiparallel. The related resistivity is called tunneling magnetoresistance and has a
bistable characteristic.

41
Magnetotransport is also at the basis of spintronics, the emerging field of the
electronics of spin. These devices are based on the conduction properties of electrons
with energy much higher than Fermi energy levels, the so-called hot electrons that
undergo scattering. When the electrons are transported adiabatically without
scattering, are called ballistic electrons.
The spin relaxation process [89] is related to the magnetoimpedance (MI), i.e. the
dependence of the inductive impedance of a medium on an AC magnetic field and is
related to the real and imaginary parts of the rotational susceptibility as the spin
approaches its equilibrium state through precession under the influence of a cycling
magnetic field.
Spin relaxation is also responsible for the Extraordinary Hall Effect (EHE).
All these effects are to be analyzed in section 4.3.

3.8 Noise
Noise, in general, refers to the output signal instabilities under steady operating
conditions. In sensing devices, noise can have contributions from the electronics, the
packaging and any mechanical moving parts, and what is of interest here, the
magnetic material itself. Because the signals in sensors are usually weak, noise may
seriously limit the resolution of the sensor.
The noise attributed to the magnetic material is made up of three different sources:
thermal noise, Barkhausen noise or magnetic noise.
Thermal noise is due to the thermal fluctuations of magnetic moments. This decreases
with decreasing temperature or increasing magnetic volume.
The Barkhausen noise is the noise related to Barkhausen jumps, the sudden,
discontinuous changes in magnetization. Contrary to thermal noise, it increases with
magnetic volume. Therefore, a compromise must be made with respect to the grain
size. This magnetic noise is also related to the grain boundary imperfections affecting
the domain wall propagation mode. Proper treatment may provide finer grain
boundaries as well as a wider angular distribution of anisotropies thus optimising the
magnetic noise levels. The Barkhausen noise is, in general, undesirable in sensing
applications but the related magnetoacoustic effect (see 4.4.5) can be used in non-
destructive testing.
In the special case of magnetic recording, the grain size and boundary affect the shape
and the position (jitter) of the recorded transitions. This noise usually decreases with
finer grains and weaker coupling between them. However, in ultra high-density
recording, a DC noise contribution is also being observed that is related to the
nonmagnetic material between grains [95]. Because the magnetic noise is dependent
on the microstructure, it is repeatable and deterministic [96].
The 1/f noise dependence, according to which the noise level decreases linearly with
increasing frequency, is observed in several sensor arrangements. In fluxgates, the 1/f
noise is a low frequency noise extending up to several decades of Hz. It is related to
the domain wall movement and to the material’s microstructure. Single domain
materials or amorphous cores with almost zero magnetostriction significantly reduce
the noise level [97, 98]. In spin valves, it is a function of the spin transfer mechanism
and the resulting magnetization fluctuations reaching a maximum when the sensing
and the pinned layer are antiparallel to each other.

3.9 Hysteresis curve revisited

A material is said to have reached positive (negative) saturation (fig. 2.3a, point C)
when no further increase (decrease) in magnetization can be observed by further

42
increasing (decreasing) the applied field. At this state, the material is well ordered
with the magnetic moments of all the domains or particles all aligned along the
direction of the field. Because a domain behaves macroscopically like a “giant
dipole”, the term dipole will be used to describe both particle and domain behavior.
The perfect alignment of dipoles at the saturation state [Hsat,Msat] does not mean that
all dipoles have their easy axes parallel to the applied field. It simply means that the
applied field is large enough to overcome, among others, the anisotropy energy and
force the dipoles to rotate away from their easy axes.
Decreasing the field from saturation, the dipoles gradually relax down to their easy
axes rotating reversibly. Therefore, the curve all the way down to remanence [0, Mr],
generally, consists of reversible contributions only: χdiff =χrev , i.e. the curve is retraced
if we increase the field to Hsat . In the second quadrant, the field becomes negative
thus favoring the switching of some domains or particles towards the direction of the
field. The onset of irreversible processes is also referred to as nucleation and the
corresponding field as the nucleation field. Then the domain walls start propagating
through bowing and translation. At this point (upper second quadrant) a lot of
reversible rotation and bowing is still taking place. As the field further decreases the
Barkhausen jumps become larger. This implies that an increasing number of domains
or particles flip irreversibly because of both short- and long-range interactions.
The slope of the curve at coercivity is an indicator of the interactions battle during the
magnetization reversal. The exchange interactions tend to keep the dipoles aligned
while the magnetostatic interactions act in the opposite direction. Thus, a very steep
loop suggests strong exchange and a large number of simultaneous Barkhausen jumps
while a smoother curve suggests stronger long-range interactions. For example, thin
films, where the exchange coupling is strong, have very square in-plane loops while
bulk materials, where demagnetizing and stray field effects are prominent because of
the non-uniformity and relatively big volume, have in general smoother, more canted,
loops. Susceptibility measurements show typically a maximum in χirr around the
coercivity [99].
Coercivity Hc has so far been defined as the field at which the magnetization becomes
zero. However, increasing the field from negative coercivity to zero will yield a
magnetization value higher than zero. The field at which zero magnetization is
reached upon its removal is called the recoil field, H r > H c , and is of practical
importance, not signifying anything for the microstructure of the material.
As the field approaches the negative saturation levels, the Barkhausen jumps continue
until practically all domains or particles have switched into the opposite direction of
the respective easy axes. At that point, the applied field forces the domain
magnetizations to snap to another easy axis closer to the applied field (if such an easy
axis exists). When all irreversible switching has taken place, reversible rotations take
over again further decreasing the magnetization to [-Hsat, -Msat].
A similar description can apply in the case of the virgin curve, starting at [0,0] and
increasing the field up to saturation. One thing though must be kept in mind. The [0,0]
state is not uniquely defined, as are the ordered saturation states. This means that there
are more than one stable domain configurations yielding [0,0] depending on the
demagnetizing procedure followed. Given that the virgin curve’s shape and properties
depend on the initial state, one should not expect the same virgin curve every time the
measurement is repeated. The most popular demagnetizing procedure is the ac
demagnetization, i.e. a decreasing in magnitude field is cycled down to zero (Fig.
3.12). The resulting state depends on the operating temperature and frequency of the
field.

43
The onset of irreversible processes signifies also the onset of hysteresis. When the
field is increased from any point in the second and third quadrants, the descending
curve cannot be retraced and a different ascending path, called a first order minor
ascending curve, is taken. If the field stops increasing before saturation and starts
decreasing then again a different descending trajectory is taken, called second order
descending curve. The number of combinations of minor trajectories that need to be
taken to reach a specific state [H,M] is infinite. The path followed depends every time
on the history of the material. The history can be wiped out totally only at the
saturation (ordered) states or partially for local field maxima. Hysteresis and the
energy dissipative irreversible processes are therefore responsible for the memory
property of ferromagnets or shape memory alloys. They are also responsible for their
highly nonlinear response that enhances the uncertainty levels in sensors. Hysteresis is
therefore highly undesirable in sensing applications unless the sensor is operated as a
switch between plus and minus levels, between the well-defined saturation states. This
technique may be used in all magnetic sensing devices to erase the memory of the
previous field sequences. Characteristically, it is widely used in the case of fluxgate
sensors [98].
500
M(G)

0
-2 0 -1 0 0 10 20

-5 0 0
H (k O e )

Figure 3.12 AC demagnetization curve

Hysteresis can be eliminated by the right choice of (soft) materials, preparation

method and appropriate annealing to relieve the material from internal stresses and
pinning sites and achieve a homogeneous and uniformly magnetizable specimen. In
the case of an anhysteretic curve like the one shown in Fig. 2.3a, the M(H) loop and
the virgin curve coincide. Domain walls bow reversibly back and forth.
Operating the sensor along its hard axis also ensures operation with little or no
hysteresis, the only disadvantage being the high-energy consumption required.

44
4. Magnetic effects
In this section, we present the major sensing effects based on magnetic properties
and/or on the magnetization process. In general, magnetic sensors sense or respond to
the presence of a magnetic field allowing contact-free operation, one of their major
advantages. They respond to the magnitude of the field H and to its temporal/spatial
variation, changing the spatial distribution of magnetization ∇ ⋅ M and its dynamic
dM
response (Fig. 4.1).
dt

∂
H(x,y,z,t) ∇⋅ M(x,y,z,t)
∂t

M(H,f,T,σ)

R(H,f,T,σ)

λ(H,f,T,σ)

Z(H,f,T,σ)

Figure 4.1 Magnetic effects in sensing

Macroscopically, the effect of the magnetic field is perceived as a change in the

magnetization M(H), in the magnetostriction λ(H), in the magnetoresistance, R(H), or
in the magnetoimpedance, Z(H) of the magnetic element. The field dependence of
these macroscopic properties leads to their use as major sensing effects.
At the microscopic level, these effects are all related to the magnetic dipole rotation,
which is mostly attributed to domain rotation or domain wall dynamics (see 3.3.5).
The applied field H causes the rearrangement and reorientation of the domains,
through long-range magnetostatic interactions. The domains rotate and their walls
move as to accommodate the energy added by the applied field and reach a new
energetically favorable and stable configuration. There are two modes in domain wall
propagation: reversible and irreversible bowing and translation (Fig. 3.7). There are
also two types of domain rotation: irreversible and reversible (see 3.2.7). The
reversible characteristics are preferred in most sensing applications whereas

45
irreversible processes, related to hysteresis, are chosen in the case of storage like in
security sensors.
The domain processes except from being the cause for the observed changes in M(H),
λ(H), R(H), and Z(H) can themselves be used as sensing effects. As an example,
domain rotation can be used in mechanical sensors using positive or negative
magnetostrictive materials [100]. Domain wall dynamics are used for small field or
mechanical sensors based on small field measurements [101,102,54]. The dynamic
domain rotation resulting in the propagation of elastic waves in magnetostrictive
materials has been used in the MDL principle [3, 103-105].
At the atomic level, the domain processes are the result of spin-spin interactions, spin-
orbit interactions, spin dependent scattering and tunneling, and spin relaxation. The
applied field adds energy to the system through interacting with the atomic moments.
This affects the spin and orbital motions of electrons and, as a result, the interactions
between them, yielding the M(H) dependence, and with the elastic properties of the
bonds connecting the atoms, yielding the λ(H) dependence. Because magnetic
materials can be metallic, the magnetic field also affects their transport characteristics
through interacting with the conduction band electrons, yielding R(H) while the time
dependence of the applied field affects the spin relaxation modes and Z(H).
For practical purposes, the above phenomena, processes or effects, are often treated as
being decoupled or independent from each other and the ‘ceteris paribus’ principle,
i.e. all other equal, is used. This can be appropriate for certain configurations,
arrangements, devices, or excitation modes. One should bear in mind, though, that
magnetic phenomena are all coupled to each other, at all levels. The magnetic field
H(t), the magnetic induction B(t) and the magnetization M(t) all have a temporal-
spatial dependence, coupled through Maxwell’s equations (see 2.7) with time-varying
electric fields E(t). For example, a DC field H(t)=HDC applied along one dimension,
causes a change in M(t) in all three dimensions and evokes a transient response to the
steady state. Therefore, if the field is switched off or changed before steady state is
reached a quasistatic solution or model will lead to wrong or inaccurate conclusions or
predictions. Or, if the magnetic body is a wire and the field is applied along its axis,
an 1D treatment may be adequate, otherwise it will fail. Or, if the magnetic body is
metallic, its resistivity will also be affected. Or, if the wire is compressed along its
length, its magnetization will be affected.
In order to be able to use magnetic materials as sensors, it is therefore important to
comprehend the processes involved and the coupling between them in order to control
them and elicit the desired response out of them.
In the above discussion, there is a hidden factor or player, which has silently been
agreed to as remaining constant throughout a process: temperature. All the above
properties, M, λ, R, or Z, change with temperature. The spin agitation changes with
temperature. This means that the exchange and anisotropy change, and therefore the
domain wall thickness and mobility, and therefore the domain wall motion and
magnetostatic interactions change, and finally the macroscopic properties change. So,
unless temperature is indeed kept constant, it must be taken into account and the
thermal stability of the sensing elements must be investigated. Besides its effect on
the magnetic response of materials, the temperature dependence of magnetic materials
has also found applications as a sensing effect in itself.
In effect, all that magnetic sensors sense, directly or indirectly, is magnetic field. The
measured field quantity is then mapped onto some other size, like stress,
displacement, flow etc. There are magnetic sensors performing the same task, e.g.
position measurements, that are based on different sensing effects. By the same token,

46
the same sensing principle can be used to develop different sensors through a different
arrangement, i.e. the magnetostriction effect can be used to measure field or position
or velocity etc. In any case, it is useful to study or obtain the behavior of the M(H),
λ(H), R(H), and Z(H) characteristics as a function of the control parameters,
frequency f, temperature T, stress σ, or at least bear in mind that the above
characteristics are not one curve each but families of curves.
In the following paragraphs of this section, the magnetic effects will first be examined
and described in relation to the theory exposed in sections 2 and 3. An attempt will be
made to relate all the different types of sensors to each effect. The different types of
sensors and the arrangements used for each type of sensor are examined in sections 5
and 6.

4.1 Induction B(H)

One of the oldest and standard sensing principles is the electromagnetic induction
based on Faraday’s law (see 2.5), i.e. a time-varying flux, Φ, induces a voltage
proportional to its rate of change. Inductive sensing elements are therefore based on
the temporal variation of B(t) (Eq. 2.5) and therefore of the magnetization M(t) (Eq.
2.15).
This property has found applications in the pick up or sensing coils that are in the
heart of any inductive sensor arrangement. A coil of N turns and area S, is wound
around the specimen or device whose flux change is to be measured. Then the
induced voltage is given by:
dB
V = − NS (4.1)
dt
Inductive sensors have found applications in field measurements, position
measurements and magnetic alloy detection [106]. Typical arrangements include the
Linear Variable Displacement Transformer (LVDT) for position measurements and
fluxgates [98, 107] for field measurements. Ring and pole reproduction heads in
magnetic recording are also typical inductive elements.
The inductive sensors are relatively simple, cheap and reliable but their future is
compromised by size limitations, thus increasing the labor and the cost of the sensor.
They always include a coil or coils that cannot be made small enough to be integrated
on a chip. Currently, hybrid solutions are sought to compete with other technologies
such as MR-based sensors.

4.1.1 Parametric control of B(H) and M(H) loop

The performance of inductive sensors is controlled through the sensing element’s
B(H) or M(H) characteristic. The operating temperature, the applied field’s cycling,
frequency and the applied stress affect the shape of the major loop, i.e. alter the
coercivity, saturation levels and squareness of the loop.
The loop becomes wider and shorter with increasing frequency (Fig. 3.9) because the
magnetic dipole moments of the material are not allowed enough time to relax at
lower energy local minima. In metallic ferromagnets, eddy current losses are also
added. The parametric characterization of a material with respect to frequency is
therefore important if it is going to be used in an environment with alternating
magnetic fields.
The operating temperature determines the thermal energy of the ferromagnet that is
responsible for the magnetic stability of the response. Higher temperatures cause more
fluctuations of the magnetic moments and delay the relaxation process towards stable

47
low energy states, altering the macroscopic major loop characteristic (Fig. 4.2).
Furthermore, the Curie or Néel point for a magnetic material must be known in order
to ensure that the sensor will not become paramagnetic.

200

180 Heating (5°C/min)

Cooling (5°C/min)
160

140
Magnetization at 1T [Am 2/kg]

120

100

80

60

40

20

0
0 100 200 300 400 500 600 700 800 900 1000
Temperature [°C]

Figure 4.2 Temperature dependence of magnetization M(T) of nanocrystalline Fe73Nb7Si5B15

Finally, the applied load changes the magnetoelastic energy of the material, again
altering the energy landscape and therefore the hysteresis loop. The effect of applied
stress is of special interest in the case of magnetostrictive materials.
In order to improve the linearity and the uncertainty of the sensor’s response,
hysteresis must be eliminated or otherwise compensated in most sensor arrangements.
Proper annealing improves the microstructure of the magnetic materials, relieving
them of internal stresses and making the reversible domain rotation a more favourable
mechanism for the magnetization reversal [55-58, 64, 107-110]. The above discussion
holds in the case of anhysteretic magnetization curves as well since the temperature,
frequency and stress affect the saturation levels and the slope of the curve or may
cause hysteresis to reappear.

48
4.1.2 Reentrant flux reversal
The phenomenon of re-entrant flux reversal, also known as large Barkhausen effect or
magnetic bistability, is observed in FeSiB, CoFeSib or even NiFe amorphous wires
with a core domain and a bamboo shaped shell domain structure [111-115]. This
structure is associated to the distribution of magnetoelastic anisotropy obtained as the
result of coupling between internal stresses frozen-in during the preparation (ultra
rapid quenching) and magnetostriction. Upon the application of an AC axial field,
these wires exhibit square-shaped hysteresis loops characterized by one single large
Barkhausen jump at a certain switching field value. The bistable re-entrant
characteristic is caused by the depinning of a nucleated closure domain at one end of
the wire and its propagation to the other end, hence the single Barkhausen jump. The
closure domains are due to demagnetising fields at the ends of the wire. The damping
mechanism of the motion is due primarily to eddy currents as well as spin relaxation
[116-117]. The velocity of the wall has an axial and a normal component resulting in
quasiplanar motion and a wall length of several cm. In order for the phenomenon to
take place the wire must be longer than 10-20 cm so that closure domains are
established and initiate the nucleation and the single domain wall propagation. [118-
120]. The sign of magnetostriction is also important. Wires richer in Fe exhibit
positive magnetostriction, related to a radial easy axis in the outer shell. On the
contrary, wires richer in Co exhibit negative magnetostriction and circumferential
anisotropy at the outer layer [54]. The velocity and the length, in general, increase
with the applied field but they can be tailored as to remain practically constant over an
operating range of values. Then, the effect can be used either in tachometers where
the frequency changes or security sensors where the drive field changes [102]. It can
also be used in distance [101] sensors that can be as long as 1m, stress sensors or in
the study of magnetic microstructural characteristics.

4.2 Magnetostrictive effects

Magnetostriction, the deformation of a magnetic body during its magnetization, is a
fundamental sensing effect with a variety of configurations and applications. The
MDL set-up and the Villari effect are some of the applications of magnetostriction.
Magnetostrictive sensors cover a wide spectrum of applications; they are used to
detect and measure stress, microstrains, field, position, acceleration. The sensing
elements come in the form of amorphous or crystalline ribbons, wires, or thin films.

4.2.1 Magnetostriction
Magnetostriction is defined as the phenomenon where a magnetic body shrinks or
expands in the direction of the magnetization as a function of an applied magnetic
δl
field. It is measured as the % strain, λ = where l0 is the length of the material along
l0
a given direction at zero field and δl=l(H)-l0. λ can be positive or negative depending
on whether the effect is expansive or compressive. The magnetostrictive characteristic
of a material with positive magnetostriction is the λ(H) curve depicted on Fig. 4.3a
[121-123]. The response can be hysteretic (solid line) or anhysteretic (dashed line).
Magnetostrictive hysteresis is related to the hysteresis observed in the magnetization
curve M(H) and therefore to irreversible processes taking place during the
magnetization process, as explained in section 3.4. In this case, l0 is the minimum
observed length of the material, occurring at the coercivity Hc. A material can be
magnetostrictively saturated in the same sense that it can be magnetically saturated,

49
i.e. beyond a certain field value, Hs, the strain stops changing with the field. This is
called saturation magnetostriction, λs and is a characteristic phenomenological
property of a material, related to elastic Young’s modulus, showing the strength of the
effect. Fig. 4.3b depicts minor curves inside the λ(H) characteristic. Similar to the
M(H) characteristics, all minor trajectories are confined within the major curve. In the
hysteretic case, which is also the most common one, the virgin state is not uniquely or
at all defined. Upon the preparation of a material, and before the exposition to
magnetic fields (even though the earth’s field may be strong enough for some
materials) there is already an internal stress and magnetization configuration, which
does not correspond to the (0,0) state. However, if this is taken to be the virgin state, it
cannot be recovered exactly through any demagnetization process once the material is
exposed to magnetic fields. The initial configuration is permanently altered as new
local energy minima have been established in the medium.

λ
λs
λs

anhysteretic hysteretic
curve curve

Η
-Hc Hc
-H s +Hs

(a)

Η
-Hc Hc

(b)

Figure 4.3 Magnetostrictive curves (a) major λ(H) curve with hysteresis (solid) and without
hysteresis (dashed) (b) set of minor curves in a hysteretic major curve.

50
The deformation of a material due to the magnetization process also results in the
propagation of magnetoelastic waves throughout the medium [124-128], a
phenomenon utilized in the MDL arrangement (see 4.4.1) [3].
The magnetostriction can be confined along a single direction in which case it is
called line magnetostriction. It may also be planar (2D) or take place throughout the
volume of a ferromagnetic body.
The volume magnetostriction, the change in volume as a function of the applied field,
is due to a change in intensity of the spontaneous magnetization and not just rotation.
It has three almost distinct cases: the form effect, the crystal effect and the forced
magnetostriction. The magnetization is always the result of an effective field, the
vector sum of the applied field and the internal demagnetising field, which is in turn a
function of the magnetization state. This field and the related demagnetizing factor N
is shape dependent. As the magnetization M increases from zero to saturation, the
volume V of the ferromagnetic body with elastic modulus c changes as:
δV NM 2
= μ0 (4.4)
V 2c
This is called the form effect; it occurs from zero field up to almost saturation and
reflects the quadratic dependence of the relative volume change on the applied field.
The form effect, is also related to the shape anisotropy and, for negligible values of
volume magnetostriction, it gives rise to linear magnetostriction. As discussed in 3.4,
when the field increases further to reach saturation, the domain magnetizations snap to
new easy axes closer to the direction of the field (small Barkhausen jumps). This
portion of the magnetization curve is associated with the crystal effect, a negative
volume magnetostrictive effect observed near saturation, giving rise to hysteretic
magnetostriction. Finally, for fields beyond saturation, the forced magnetostriction, a
linear function of the field, is observed [12], giving rise to anhysteretic
magnetostriction.

4.2.2 Origin
Magnetostriction [9] is inextricably related to the magnetization process with origins
similar to those of the anisotropy (section 3.2.3). It is the result of interactions
between atomic magnetic moments in a crystal lattice and the interplay with the
elastic bond length. In the paramagnetic regime where magnetic atoms are dispersed
in a nonmagnetic environment with no interaction between them, their magnetic
moments have no effect on the crystal bond shape or length. In the ferromagnetic
regime, under the influence of the exchange, the magnetic moments align with each
other along the direction dictated by the local anisotropy. The ordering increases the
magnetostatic energy, i.e. aligned dipole moments have a higher magnetostatic
energy, which after a certain volume of aligned moments overtakes the combined
effect of exchange and anisotropy and breaks down the ordering forming the domains
(spontaneous magnetization). During the alignment process within a domain, the
magnetic moments rotate as to align with the local easy axis. The local field h acting
as to align the moments m along this direction is the result of the orbital and spin
motion of the 3d- (or 4f- ) electrons taking place in the plane defined by either the
short or the long axis of the atomic ellipse (Fig. 4.4). Because the magnetic moment
m wants to remain normal to the plane of motion, the rotation of m towards h causes
the rotation of the ellipse as well. The new aligned configuration strains the
interatomic bonds in an effort to keep in balance the magnetostatic (attractive) and
electrostatic (repulsive) forces. In the case that the resultant spin and orbital motion is
in the plane of the long axis, the magnetic moment is along the short axis and the

51
alignment of m with h results in the compression of the space occupied by the
participating atoms (Fig. 4.4a); hence, the magnetostriction is negative. In the case of
positive magnetostriction, the plane of motion is defined by the short axis, the
magnetic moment is along the long axis and the alignment of m with h results in the
expansion of the space occupied by the participating atoms (Fig. 4.4b). The crystal
deformation due to ferromagnetic ordering is known as spontaneous magnetostriction.
It corresponds to the built-in strain existing in the demagnetised state where the net
magnetization is zero.
Length in the Length in the magnetized state
magnetized
state

m
m

Length in the H
non-magnetized Length in the non-magnetized state
state

(a) (b)

Figure 4.4 (a) Negative and (b) positive magnetostriction as a result of interaction between
magnetic moments.

This argument can be extended to domains, if one accepts the convention that the
latter behave as giant magnetic dipoles in quasistatic equilibrium. In section 3.2.5,
domains were considered the result of the balance between exchange and anisotropy
energy, aligning the atomic moments inside them along a given direction, and the
magnetostatic energy deciding their size. In view of magnetostriction, the domain
shape, size and orientation is decided by the magnetoelastic energy as well acting as
an additional anisotropy term [129-130]. The externally applied field causes the
dipoles-domains to align with it through rotation of their magnetization and/or domain
wall movement. This results in the deformation of the whole volume containing the
domains, i.e. of the bulk magnetic body.
One can think of the energy added to the system because of an externally applied field
as being counterbalanced by the change in the magnetic configuration in cooperation
with the elastic bond energy. The coupling between the elastic strain and the
magnetization orientation gives rise to an anisotropy energy term, the magnetoelastic
energy, which is dependent on the crystal structure. The magnetoelastic energy is a
linear function of strain favoring the crystal deformation. It is counterbalanced by the
elastic energy, a quadratic function of strain. The equilibrium state reached determines
the linear deformation, λ, observed along some direction.
As mentioned already, the saturation magnetostriction λs is dependent on the magnetic
and crystal configuration and its elastic properties. There is no general form for it but
it can be expressed as a function of the saturation magnetostriction measured along
100 and 111:
λ s = f ( λ100 , λ111 ) (4.2)
For example, in the case of a single domain, single crystal cubic material Eq. 4.2 is:

52
 3 ⎛ 1⎞
λ s = λ100 ⎜ α12β12 + α 22β22 + α 32β32 − ⎟ + 3λ111 ( α1α 2β1β 2 + α1α 3β1β3 + α 2α 3β 2β3 ) (4.3)
2 ⎝ 3⎠
where α are the magnetization direction cosines with the applied field and β are the
respective magnetostriction cosines.

4.2.3 More on the phenomenology of magnetostriction

In the case of ordinary magnetostriction, described so far, the local magnetization, at a
micromagnetic scale, is considered uniform and only varying in direction. Therefore,
the strain is a result of rotation only, and no change in volume takes place.
The ordinary magnetostriction effect, λ(H), in terms of domain wall motion, reflects
the displacement of 90o walls rather than of 180o walls during the magnetization
process. Therefore, the domain rotation and domain wall propagation play an
important role in the magnetostrictive behaviour of the sensing element. Reversible
rotation of the domain magnetization and/or bowing of the domain walls leads to low
or zero hysteresis response, and therefore higher linearity and lower uncertainty. The
reversibility can be obtained if operating the sensor in the linear (reversible) part of
the curve. Therefore, the designer should expose the magnetic material to such a
treatment (stress-current annealing, creep induced anisotropy etc) in order to minimize
the amplitude of the external field responsible for the irreversible rotation process and
maximize the one of reversible rotation [55-58, 109].
The magnetostriction effect is strongly related to the magnetization process described
in section 3. It is the alignment with the applied field (magnetization process) that
causes the crystal deformations resulting in the macroscopic magnetostriction
characteristic of Fig. 4.3. Phenomenologically, the minimum in the λ(H) curve of
positive magnetostriction occurs at the coercive field Hc where M(Hc)=0, and the
butterfly shape of λ(H) could be proportional to the derivative of M(H). So, the
discussion on the properties and parameters affecting the M(H) characteristic applies
in the case of λ(H), as well [114].

4.2.3a Parametric control of the λ(H) characteristic

Similar to the M(H) characteristic, the λ(H) curve depends on the frequency of the
applied field, the operating temperature and the applied stress.
Increasing the frequency f of the cycling field, results in the widening of the loop (Fig.
4.5b). The magnetic moments do not have enough time to relax along the minimum
energy direction but reach a higher energy minimum first. This higher energy
minimum corresponds to an effective smaller field in magnitude or to “hardening” of
the material, i.e. at high frequencies, the magnetization, for a given field value H(t)
reaches a state that corresponds to a smaller field |Heff(t)|<|H(t)| than the one applied.
Therefore, the frequency acts as an effective field offset, positive for the increasing
fields and negative for decreasing fields.
Temperature T is also related to spin relaxation dynamics. As discussed in section
3.5.2, the increasing thermal agitation of spin moments reduces the exchange
correlation length favouring magnetostatic contributions. As the temperature
increases, the material breaks down into smaller domains and the local anisotropy
axes get more dispersed. As a result, the hysteresis loop becomes less square and the
magnetization values for a given field decrease. Beyond the Curie temperature, all
ordering breaks down and paramagnetism sets in. The magnetostriction follows
closely this behaviour.
Stress σ finally acts as an effective field along the direction it is applied (Fig. 4.5a).
When a tensile stress is applied to a material with positive magnetostriction, it forces

53
or helps the domain magnetizations to rotate or the domain walls to propagate in the
direction of the applied stress. In the first case, the direction of the applied stress acts
as an easy axis while in the second case it acts as a force putting pressure on the
domain wall and forcing it to move, as also does a magnetic field (section 3.2.5). A
similar analysis can be carried out for the case of compressive stress in materials with
negative magnetostriction or torsional stresses in the case of helicoidal anisotropy.
The opposite happens in the case of positive magnetostrictive materials being
compressed and negative magnetostrictive materials being stressed.
Because a λ(H) or an M(H) curve is performed under given conditions of temperature,
loading or rate of applied field, taking into account the parametric effect of f, T, and
σ on the measurements is crucial in interpreting and using the data correctly.

σ1

σ2

f1
f2

Figure 4.5 λ(H) curves for (a) applied stress σ1< σ2 (b) cycling field frequency f1<f2

4.2.4 Inverse magnetostriction

The inverse magnetostriction effect is the magnetization dependence on the applied
stress [131, 132]. An applied stress induces an easy axis forcing the magnetization of

54
the domains to align with it in much the same way as a magnetic field would. The
energy due to the applied stress σ is anisotropic as can be seen for the simplified case
of isotropic magnetostriction:
3
Eσ = − λσ cos 2 φ (4.5)
2
where φ is the angle between the applied stress and the magnetization direction.
The effect is based in domain rotation and is used in the development of mechanical
sensors.
The inverse magnetostriction effect plays also a role in the magnetoresistive response
of a material since it competes with the uniaxial anisotropy causing irreversible
processes and hysteresis to appear when it increases.

4.2.5 Magnetostriction and phase transformation

Big and repeatable (reversible) deformations have recently been observed in a new
class of materials, the (ferro)magnetic shape memory alloys (MSMAs or FSMAs)
(Fig. 4.6). The underlying mechanism in this case is quite different than that of
classical magnetostriction, described in 4.2.2 and closer to that observed in classical
shape memory alloys (SMAs) [133].
relief surface

H at T< Tm
Figure 4.6 Magnetic shape memory alloys

SMAs are alloys (TiNi, TiPd) undergoing a phase transformation from cubic
(austenitic phase) to orthorhombic (martensitic phase) structure as temperature T
decreases (and the reverse when T increases) which results in an elongation in the
order of 5-10% [134-139]. The distortion can take place in four equivalent directions
leading to the so-called twinned microstructure, i.e. regions with the same structure
but distortion in different directions, called twin variants, are separated by a clearly
defined boundary. This boundary can be moved by an externally applied stimulus
resulting to large deformations. The original structure in SMAs is recovered when
heating the alloy at a temperature above the austenite-martensite phase
transformation.
In MSMAs, additionally to the structural domains, magnetic domains also exist [138].
Each twin has its own magnetic easy axis and magnetic domains. In zero applied
field, the magnetic moments within the magnetic domains of a twin lie along the easy
axis of the given twin. Hence, there is a distribution of easy axes in the material.
When an external field is applied two things may occur: 1) either the magnetic
moments rotate within the twins or 2) the twin boundaries themselves move to allow
the alignment of the magnetization directions with the applied field (Fig. 1). For the
latter to take place, the magnetocystalline easy axis coincides with the structural c axis
of the orthorhombic phase.

55
MSMAs can undergo a phase transformation either because of field or because of
temperature and in both cases the response is hysteretic. The field actuation has the
advantage of fast (high frequency) response and, in the case that the c axis is also the
anisotropy axis, low field values are needed. The magnetic field acts as an equivalent
stress yielding an irreversible giant strain through the displacement of the twin
boundaries, the mechanical analogous to the 90o domain walls. The effect is also
known as colossal magnetostriction [139].
The magnetic moment in MSMAs stems mainly from the Mn (or Fe) atoms of the
alloys and the interdependence of structural and magnetic domains is due to the high
magnetocrystalline anisotropy of the martensite. Going from the cubic structure to the
orthorhombic, with a shorter c axis, an aspect ratio of c/a is established. A typical
value for c/a is 0.94; this may account for the 6% deformation usually reported for
such alloys [136]. It has been found that there is a correlation between the (s+d)
electron/atom ratio (e/a) and the instabilities leading to the phase transformation.
More specifically, materials making good candidates for MSMA applications present
an e/a ratio of 7.4 at the onset of the instability [135].

4.2.6 Materials
All magnetic materials are essentially magnetostrictive with either a low or high
magnetostrictive effect. In general, if the magnetization process is due mainly to the
propagation of 180o domain walls then magnetostriction is negligible while if it is due
mainly to domain rotation then the magnetostriction effect is greatly enhanced. The
magnetostrictive materials are distinguished according to their magnetostrictive
coefficient λ or according to the mechanism leading to large deformations.

4.2.6a Classical magnetostriction

Classical 1D, 2D, and 3D magnetostrictive materials have λ values up to a few
decades ppm. The magnetization processes are dominated by the propagation of 90o,
108o, and 60o domain walls and irreversible and reversible domain rotation. The latter
mechanism leads to higher magnetostriction values; the (irreversible) Barkhausen
jumps within the domains where the magnetization flips to a new easy axis closer to
the applied field are followed by the (reversible) small angle rotation (SMAR) of the
magnetic moments towards the field. This leads to a large expansion (or
compression) of the magnetic domains as the dipoles rotate and strain the crystal
lattice. The non-180o domain walls inhibit the propagation in favor of the rotation
magnetization reversal mechanism. The magnetostriction values, for a given material,
can be positive or negative depending on the axis of measurement (100 or 111), e.g. in
the case of Fe, λ100 is positive up to 20-30 ppm while λ111 is negative around 10 ppm.
In classical magnetostriction materials, like Ni, Fe, or NiFe alloys, the hysteretic part
of λ(Η) associated with Barkhausen jumps dominates the rotation process. This is
undesirable in sensing because of the increased uncertainty and non-linearity levels so
amorphous phases are preferred instead. In these materials, the distribution of local
anisotropy fields is very wide and the application of a field causes many irreversible
rotations instead of a few large ones so that the macroscopic response is closer to
anhysteretic rather than to hysteretic response. Furthermore, in amorphous materials
the magnetization reversal takes place mostly through domain rotation, which results
in domain expansion of the large number of crystallites and high λ. A typical
amorphous magnetostriction material is FeSiB in ribbon or wire form with
magnetostriction up to 100 ppm. Nanocrystalline materials grown out of an
amorphous phase, through the addition of Al, Cu and Nb, in most cases, have

56
magnetostriction values that can decrease down to 0 and negative values as the
content of nanoislands in the amorphous phase increases. They are purposefully
designed with high remanence, and low coercivity and magnetostriction, in order to
be used in electromagnets. In order to use these materials in magnetostriction
applications they have to be designed so that the magnetostriction of the
nanocrystalline phase is added (rather than subtracted) to the amorphous one.

4.2.6b Small magnetostriction

These are materials with magnetostriction values lower than 1 ppm. The main
magnetization mechanism is the domain wall propagation with 180o walls dominating
the process. Single element ferromagnetic materials cannot really be considered as
small magnetostriction materials. In order to develop small magnetostriction
materials, ferromagnetic alloys are preferred that either allow for (mainly) domain
wall propagation like in NiFe alloys or where macroscopic cancellation of positive
and negative micro strain dependence is observed as in the case of FeCo families of
amorphous and crystalline materials.

4.2.6c Large magnetostriction

A typical example of this family of materials is terfenol (TbFe) yielding values of
magnetostriction as high as 1000-2000ppm. The big localized magnetic moments of
the 4f element occupy sites of Fe forcing zero domain wall propagation and
Barkhausen jumps at high H values leading to high rotation (big magnetic volumes
expanding inside the domains). This is the reason why any treatment of the material
cannot significantly affect the hysteresis in magnetostriction behavior.

4.2.6d Colossal Magnetostriction (MSMAs)

MSMAs are ferromagnetic alloys actuated by magnetic fields and yielding
deformations up to 10%. Made of relatively low cost (compared to rare-earth alloys)
materials, they combine the high material strength and elasticity of classical SMAs
with the contact-less and high frequency and efficiency control allowed by the
magnetic field actuation found in magnetostrictive elements. They are relatively new
materials with possible application more in the field of actuation rather than sensing.
The most popular MSMA is the Heusler ferromagnetic alloy Ni2MnGa followed by
FePd, Fe3Pt, CuAlMn et. al.. The impressive advantage of colossal magnetostriction
is coupled with the room temperature operating temperature (martensitic
transformation temperature is in the range of 265 K and Curie temperature is about
350 K) and the relatively low magnetic fields required (a few kOe).
In general, the criteria for colossal magnetostrictive materials are the thermoelastic
transition to the martensitic phase (as in SMA), a ferromagnetic martensitic phase and
high anisotropy in the martensitic phase (about two orders of magnitude higher than
that of the austenite) [136].

4.2.6e Composites & Synthetic materials

The last two categories of magnetostrictive materials are the, rather recently
introduced, composite and synthetic materials.
Composite materials consist of magnetic particles with shape anisotropy embedded in
an elastomer matrix. The magnetostrictive effect in these materials does not depend
on the mechanism of magnetostriction but on the magnetic moments of the inclusions
as well as their interactions. The inclusions may be permanently magnetized made of
hard magnetic materials, such as Terfenol-D or SmCo, dispersed in an elastomer

57
matrix of low Young’s modulus, such as silicon. The above-mentioned inclusions
may also be made of soft magnetic alloys, such as Fe-based, Ni-Fe-based and Co-Fe
based alloys. The application of an applied field stretches or shrinks the elastic
material through the change in orientation of the magnetic particles (inverse
elastomagnetic effect). Alternatively, the application of stress alters the magnetization
of the composites through the change in interactions between the particles (direct
elastomagnetic effect). The latter may find applications in stress sensing while the
inverse effect in actuating [140-143].
Synthetic magnetostrictive materials are heterostructures consisting of rare
earth/transition metal sandwiches, e.g. Y-Co/Gd-Co/Y-Co or Tb-Co/Nd-Co/Tb-Co.
The layers of about 100 nm each are antiferromagnetically coupled. The
magnetization processes are dominated by the extended domain wall mechanism
[144-145].

4.3 Magnetotransport effects

Magnetotransport effects (see 3.7) is a term used to describe the interplay between
spin, orbital, lattice and charge dynamics, in the presence of magnetic fields, resulting
in spin tunnelling, scattering and damping. The effects used in sensing include the
magnetoresistance (MR) and giant magnetoresistance (GMR), the magnetoimpedance
(MI), spin dependent tunnelling (SDT), and the extraordinary Hall effect (EHE). The
most important and popular effect used mainly in field sensors and recording is the
giant magnetoresistance (GMR) effect while recently colossal MR (CMR) values
have been reported but in cryogenic environments. The giant magnetoimpedance
effect (GMI) as well as its asymmetric counterpart (AMI) is currently attracting the
interest of several sensor designers, while EHE is still under investigation for potential
uses.

4.3.1 Magnetoresistance (MR)

The magnetoresistance (MR) effect is the % change in the resistance of a ferromagnet
depending on the magnitude and direction of an applied field [6]. It is observed
mainly in magnetic thin films and the dc electrical resistance of the film may change
up to 1-2% with the externally applied magnetic field.

M
θ
J
H

Figure 4.7 MR effect

The operating principle is depicted on Fig. 4.3, i.e. the change in resistivity ρ of the
film occurs when the current density J flowing through it changes from being parallel
to its magnetization M to being perpendicular to it. The applied field H forces M to
rotate away from its original direction (easy axis) and hence changes ρ. Therefore ρ
depends on the angle θ between the direction of M and that of the current J (Fig. 3.7):
ρ : ρ0 + Δρ cos 2 θ (4.2)
where ρ0 is the base resistivity at H=0.
Typical MR materials are NiFe (permalloy) and NiCo films with typical MR values of
2%-4% at fields of a few Oe.

58
The mechanism of magnetotransport at the atomic level is described in section 3.2.12.
At the domain wall level, the effect is similar to ordinary magnetostriction, in the
sense that it is due mainly to domain rotation and in some cases also to domain wall
nucleation. The magnetization process through the domain rotation changes the
amount of scattering of the conduction electrons by magnetic dipoles and domain
walls. This change depends on the spin polarization and resulting magnetization
orientation following the Lorentz law (Eq. 3.22).
In principle, any metallic ferromagnet can exhibit magnetotransport properties but the
strength of the effect depends on the amount of scattering of the conduction electrons.
This in turn depends on the thickness of the domain wall, acting as a tunnelling barrier
or as a scatterer. The thinner the domain walls are the higher the MR. In typical
ferromagnets, the domain wall thickness is usually much larger than the Fermi
wavelength so that the exchange field allows for very little scattering of the
conduction electrons. The resistance decreases with the field because the
magnetization aligns with it and scattering of the conduction electrons against the
dipole moments of the 3d- electrons is reduced.
Magnetoresistive materials are characterized by the MR (%) ratio, defined as ΔR/R0
where R0 is the base resistivity at zero field and ΔR=R(H)-R0. The MR characteristic
is depicted in Fig. 4.4. The % change ΔR/R0, is plotted against the applied field. The
peak of the R(H) curve occurs around the coercivity of the magnetic medium or at 0
field for non-hysteretic media. In the case of a “virgin” hysteretic film, a film that has
never been exposed to magnetic fields, the MR value at 0 field is almost double the
value attained at the coercivity. The “virgin” state cannot be recovered again through
demagnetisation. The magnetotransport properties depend on the magnetization
reversal mechanisms and the domain wall rotation and dynamics. Therefore, the MR
effect, as a spin-dependent scattering mechanism, is strongly dependent on the
magnetic configuration of the medium. Once a magnetic medium is magnetized,
permanent changes in its magnetic configuration take place that cannot be reversed or
erased.
MR sensors sense the angle of the field to the magnetization and can be used in field
measurements having the advantage that they are independent of the frequency of the
field being measured. In that sense, they are superior to inductive field sensors.
Because they are grown in thin film structures, they are inherently compatible with IC
techniques, following closely the trends in today’s micro- and nano- systems, e.g. the
state-of the-art in the recording industry are MR reading heads [146].
Because the MR effect is dependent on domain wall scattering, magnetoresistive
elements, e.g. Ni wires, may be used to study domain wall dynamics and
magnetization reversal mechanisms. The MR curves can prove a helpful tool in the
study of the magnetic microstructure of a material since the large irreversible jumps
often observed in MR curves are attributed to spin accumulation within the domain
walls and suggest the presence of inhomogeneities or constrictions.
The MR characteristic is also dependent on the applied stress through the inverse
magnetostrictive effect. The applied stress induces an anisotropy axis that competes
with the easy axis of the material. When it becomes large enough, hysteresis related to
Barkhausen jumps appears in the MR curve.
In general, MR sensors should exhibit zero magnetostriction.
Because MR sensors are thin films, another advantage of MR sensors is their inherent
compatibility with thin film production methods, resulting in lower production costs,
and integrated circuit technology.

59
4.3.2 Giant magnetoresistance
The most significant magneto-resistive effect is the giant magnetoresistance (GMR)
with resistance changes higher than 10% at room temperature. GMR structures
include multilayers, sandwich structures and granular films [88-94, 147-156].
Unlike the ordinary magnetoresistance that depends on the angle between the
magnetization and the current flow, the GMR effect depends on the angle between the
magnetization in neighboring layers; it is therefore independent of the direction of the
sense current.
The multilayers consist of several identical ferromagnetic layers separated by
nonmagnetic but conducting layers. At zero field, the magnetic films are
antiferromagnetically coupled to each other through the diamagnetic or paramagnetic
spacer and can be forced to parallel alignment through the application of a magnetic
field perpendicular to the film plane [88]. This magnetization rotation is accompanied
by a big change in resistivity exhibiting a peak around the coercivity. Because the
thicknesses of the spacers as well as of the magnetic layers are small compared to the
Fermi wavelength and the mean free path wavelength, surface anisotropy dominates
the bulk anisotropy and most of the scattering takes place at the interface rather than
against scatterers inside the bulk structure. Because the process is dominated by the
interface anisotropy, the magnetization is forced to rotate perpendicular to the film
plane. Also, as the number of layers increases, the GMR effect increases as well since
this scattering mechanism has an aggregate effect. The conduction in multilayers can
be either perpendicular to (CPP) or in plane (CIP). CPP sensors offer higher ΔR
values but low resistivity and at relatively low temperatures. These drawbacks are
partially remedied by using very thick structures of several dozens of layers of very
small surface [147].
Typical multilayers are CoCu or FeCr multilayers [149, 88]. In FeCr multilayers, the
antiferromagnetic coupling decreases with decreasing Cr thickness, in favour of
ferromagnetic coupling, and so does the GMR effect. When a field higher than
saturation is applied forcing the magnetizations in neighboring Fe layers to align
parallel to each other, there is high transmission of the conduction electrons and the
scattering at the interface is low and similar to scattering due to Cr impurities inside
the bulk Fe. At zero field, the thin Cr layer couples antiferromagnetically the Fe
layers. The thinner the Cr layer is, the stronger the antiferromagnetic coupling. The
scattering in this case is definitely stronger since Fe atoms in the antiparallel layer as
well as Cr atoms act as scatterers. Typical field values required for saturation are in
the order of a few kOe at room temperature and for GMR values a couple of tens of
percent.
A similar GMR structure is the spin valve sandwich consisting of two different
ferromagnetic layers separated by a nonmagnetic layer (Fig. 4.8). The top layer is
magnetically hard (e.g. Co) and pinned along the longitudinal or perpendicular
direction and the bottom layer is magnetically soft (e.g. NiFe) and free to rotate
according to the sense current. The interlayer can be either conducting (Cu) or
insulating (Al2O3). In both cases, it acts as a tunneling barrier controlling the amount
of scattering and conduction. In the case that an insulator is used, the R(H)
characteristic may be bistable and appropriate for memory applications. Besides the
trilayer structure just described, sandwich symmetrical structures of the type
NiFe/Co/Cu/Co/NiFe are also used [148].
Similar to the multilayers, the resistivity is lowest when the magnetizations of the
magnetic layers are parallel to each other and highest when they are antiparallel but
spin valves are more sensitive at low fields [32,35]. In general, the fields required for

60
saturation are higher than in ordinary MR films but quite lower than in GMR
multilayers.

Pinning layer

FM2 FM1 FM2

free layer
Conducting
nonmagnetic spacer

Pinning layer

Free layer

Figure 4.8 Spin-valve structure

The change in resistivity, ΔR, is proportional to the cosine of the angle between the
magnetization of the two ferromagnetic layers. If the magnetization of the hard layer
is pinned in the perpendicular direction, c is proportional to sinθ1, where θ1 is the
angle between the magnetization and the in-plane easy axis. The resulting sensing
element has then a quite linear characteristic, one of the advantages of spin valve
sensors.
The spin valve structure can be used as a RAM cell [150] at low fields that are not
strong enough to switch the hard layer but can be sensed by the free layer. Then the
R(H) characteristic is bistable with the highest value occurring when the two layers
are antiparallel to each other. When the stripe width is narrower than the magnetic
domain wall width (<1μm), the free layer switches as single domain through rotation
and the characteristic is square.
Currently, spin valve structures of the type F1/F2/S with S being a superconductor are
also being examined [91].
Note that in both multilayers and spin valves, the ordinary MR effect is also present
but it is small compared to the GMR effect.
A more recent GMR structure are the granular materials consisting of magnetic
granules dispersed inside a nonmagnetic matrix like in the case of FeCu, CoAu, or
CoCu. The measured ΔR is at least of an order of magnitude higher than in other
GMR structures, so the effect is known as “colossal” magnetoresistance (CMR).
However, these high values of MR are observed in cryogenic environments while at
room temperature, they reduce down to a few percent. As long as the mean free path
wavelength is bigger than both the grain size and the intergranular distance, the tunnel

61
barrier is thin enough to allow for a significant amount of scattering [93]. The grain
size must therefore be kept small, also to avoid strong intergranular interactions, often
at the superparamagnetic limit. Inhomogeneities or constrictions lead to increased spin
scattering during the domain wall motion, resulting to higher MR with respect to
homogeneous or single domain materials. The coupling mechanism between the
grains is similar to that of GMR films.
CMR has also been observed in intrinsic materials, such as manganate perovskites
(LaMnO3) [157-158]. CMR materials haven’t found applications yet, because of their
temperature sensitivity, but they are interesting structures to study and understand
magnetic phenomena.

4.3.2a Anisotropic magnetoresistance (AMR)

The anisotropic (or asymmetric) magnetoresistance (AMR) effect is based on
anisotropic magnetotransport, generated by the so-called barber pole structure,
inducing a shape anisotropy in the whole structure [159-160]. The AMR effect is
around 1-5%. In spite of the low values, microelectronic vector sensors based on the
AMR phenomenon have been constructed with precision of the order of 50 nT. The
biggest disadvantage of the AMRs, as field sensors, is their field anisotropy values
and their high levels of magnetic noise, this limiting their uncertainty levels. AMR
field sensors can be manufactured in large quantities by using thin film technologies,
but they have lower levels of accuracy in comparison with the more traditional
fluxgate sensors [161].

4.3.2c Ballistic magnetoresistance (BMR)

Extremely high values of magnetoresistance (>10000%) have been observed in
electrodeposited Ni nanocontacts at room temperature even at a few hundred Oe [162-
163]. This is referred to as ballistic magnetoresistance (BMR). BMR is attributed to
spin-dependent electron adiabatic transport across nanometer sharp domain walls
within the nanocontacts. These geometrically constrained walls are formed inside
nanoconstrictions and can be extremely narrow if the constriction’s characteristic
length is small, as in the case of nanocontacts [67]. The constrictions are the result of
inhomogeneities or voids inserted during the preparation process.
However, the effect of BMR is still a rather unexplored and unexploited field with
several researchers arguing about the very existence of it as being confused with other
magnetostatic, magnetostrictive or magnetomechanical effects [164]. In any case, the
resistance ratio change is so high, that it may be usable for sensitive field sensor
development. Understanding of the mechanism may lead to ultra high density, i.e.
1012 bits/in2, magnetic storage devices [165].

4.3.2d Spintronics
Spintronics or magnetoelectronics is an emerging field of technology utilizing the
“electronics of spin” [167-170]. Spin-dependent tunnelling is used in magnetoresistive
devices where the resulting scattering is of interest (spin valves), in spin tunnelling
microscopy (STM) that allows the manipulation of atoms and the study of the
magnetic properties at the atomic level, or in magnetic tunnel junctions (MTJ) and
Magnetic Tunnel Transistors (MTT) or Spin Valve Transistors (SVT) [168].
MTJs are sandwich structures consisting of several layers of ferromagnets (CoFe,
CoFeb, NiFe) and antiferromagnets (MnIr, CrMnPt) separated by an oxide layer
(usually AlO3) thinner than 20 Å. They are potential candidates as memory cells in
MRAM applications and as read heads.

62
MTTs and SVTs are three terminal multilayer devices consisting of alternating layers
of ferromagnetic and metal materials sandwiched between Si (or other semiconductor)
layers. The SVT is based on spin-dependent transport of hot electrons through a
metallic spin valve (Base) sandwiched between semiconductors (Emitter and
Collector).
The MTT is a similar device with higher energy hot electrons and fewer layers.

4.3.2e MRAM technologies

Lately, the efforts of many research groups in the areas of magnetic recording,
magnetic material and sensor development have been directed towards the
development of magnetic random access memories (MRAM). MRAMs are intended
as universal memories, e.g. nonvolatile, high-density storage devices with fast access,
combining the accessibility characteristics of the electronic RAMs and the
nonvolatility of the hard disks. The speed and accessibility of the hard disks are
limited by their moving mechanical parts and the complex system of sensors and
actuators involved in the process, while electronic RAMS cannot seem to go further
than 1 Gb while in the case of nonvolatility this number is at the moment seriously
decreased (flash memories). The goal of the MRAMs is therefore to achieve
accessibility rates of the order of 100 ns with permanent storage capabilities at least as
high as 1ΤΒit/inch2.
The proposed MRAM structures, so far, are based on spin valve structures of a hard
(Co) and a soft (NiFe) magnetic layer allowing for spin tunneling through the
intermediate oxide layer (Al2O3). The parallel and antiparallel alignment between the
two layers corresponds to the binary bit states and the hysteresis observed guarantees
the storage capabilities of the structure [171]. The compatibility with switching times
of transistors and CMOS technologies as well as their thermal stability is still under
investigation [171-172].

4.3.3 Extraordinary Hall Effect

According to the Hall effect (4.4.8) the Hall voltage is proportional to the magnetic
induction B. For the case of a ferromagnetic thin film, this means that it is a function
of both the magnetic field H and the magnetization M (Eq. 2.15):
VHall=f(H, M)
The field dependence of VHall is the well-known Hall effect while the magnetization
dependence of VHall in ferromagnetic thin films is the extraordinary Hall effect (EHE).
The EHE at room temperature is due mainly to the interaction of conduction electrons
with the 3d moments and therefore its magnitude depends on the scattering
mechanism and the state of the magnetic scatterers of a given system. The EHE
coefficient is proportional to the square of resistivity ρ as opposed to the linear
relationship of the ordinary Hall effect with ρ.
It is observed in granular media, e.g. Cu-Co, FeNi - SiO2, or ultra-thin film structures,
e.g., Pt/Co/Pt and lately it has been attracting a lot of attention due its potential for
applications in data storage and magnetic sensors.

4.3.4 Magnetoimpedance Z(H), MI & GMI

Magnetoimpedance is the dependence of the impedance Z=R+jX of a magnetic or
non-magnetic conductor carrying an ac current IAC on an axial DC magnetic field,
ΔZ Z ( H DC ) − Z 0
HDC. The magnetoimpedance ratio is defined as = where
Z0 Z0

63
Z 0 = R 2 + ( 2πfL )
2
is the magnitude of the complex impedance at saturation field
HDC,sat. This is the typical working definition of the GMI ratio even though more
information is included in the ratio, ΔZ/R0. The magnitude and phase of the
impedance are a function of the field dependence of the transverse (circumferential in
the case of wires) permeability μt and the corresponding skin effect. So, in order to
have a significant MI effect, μt must be large enough and strongly dependent on the
DC magnetic field [173-185].
The impedance of a conductor of length L and perimeter l can be calculated as [173]:
L⎛ h ⎞
Z = ⎜ ζ zz − ζ zφ z ⎟ (4.3)
l ⎜⎝ hφ ⎟⎠
where ζzz and ζzφ are the diagonal and off-diagonal elements of the impedance tensor
and hz and hφ are the axial and circumferential field components respectively. ζzz
corresponds to the circumferential impedance measured in symmetric GMI materials
when the off-diagonal component or the axial field is zero. In the absence of
symmetry when ζzφ is not zero or in the presence of an axial field component hz the
Z(H) becomes asymmetric with respect to the field origin and the AMI phenomenon
is observed (see 4.2.2.a).
Magnetoimpedance is not quite the same as the impedance in nonmagnetic conductors
because the circumferential magnetic field generated by the AC current magnetizes
the conductor and the interplay between the two magnetic fields – the axial DC and
the circumferential AC – generates harmonics in the very current that generated one
of them. These harmonics cause a change in the skin depth, thus modifying the
impedance. The skin effect in magnetic conductors is observed at much lower
frequencies than the nonmagnetic ones.

3
MI (%)

0
-15 -10 -5 0 5 10 15
B (mT)

Figure 4.9 MI curves at various frequencies

GMI has been observed and measured in a variety of structures, ribbons, wires,
nanotubes and thin films, in crystalline, polycrystalline or amorphous form. A typical
MI curve for a Fe/Co/Fe trilayer structure is shown on Fig. 4.9. The peak observed in
the MI curve occurs at a field value equal to the anisotropy field of the material. The
most attractive behaviour has been encountered in amorphous wires of zero or
negative magnetostriction and it is on them that most of the following discussion is

64
focused. Interesting GMI behaviour has recently been observed in nanotube
structures. The preparation methods, treatment and resulting properties are currently
attracting a lot of interest [175-179]. GMI has also been reported in nanocomposites
consisting of magnetic particles embedded in a dielectric nonmagnetic matrix, e.g. Fe-
C, Co-C etc. In that case, the GMI effect is due to the interaction between the nano-
magnetic phase and the dielectric matrix [173].
The GMI behaviour in wires, or structures with cylindrical symmetry like nanotubes,
is much easier to control since the there are no free poles at the ends and closure
domains to complicate the magnetization process. The domain structure in these
materials consists of a core of axial domains surrounded by circumferential surface
domains [173].
In the other cases, the MI effect strongly depends on the growth process and shape
effects, i.e. in thin film structures because of the shape anisotropy and free poles and
consequent closure domain at the edges, a sizeable MI effect can be observed but at
very high frequencies only - in the range of GHz [183].
The zero or negative magnetostriction suggests magnetically softer materials in the
transverse direction and therefore a higher dependence of the transverse permeability
on the axial DC field. The MI ratio curve then has a single peak form, similar to MR
curves, and the GMI ratio can be as high as 400%. Applying stress along the axis of
the wire, “hardens” the wire in the axial direction and induces a longitudinal
anisotropy axis. The GMI ratio is reduced but more interestingly the MI curve now
shows two peaks symmetrically around the zero field at a field value comparable to
the stress-induced anisotropy field [173-175]. The presence of longitudinal anisotropy
affects the transverse permeability leading to lower and lower GMI ratios. Materials
with positive magnetostriction also exhibit the MI effect but it is much weaker.
The transverse permeability has contributions from both the relaxation of the domain
wall motion and the rotation of magnetization as HDC increases or decreases from 0. It
is therefore a function of the excitation frequency of IAC affecting the magnetization
reversal mechanism and the strength of the eddy current damping on the domain
walls.
At low frequencies, up to a few kHz, the magnetoimpedance is also called
magnetoinductance because the skin effect is very low and only the imaginary part
X=2πfL varies with HDC. The Z(H) response is highly nonlinear and the sharp voltage
peaks are due to irreversible Brakhausen jumps or domain wall motion, as the
transverse magnetization flips in the opposite direction [181-183]. At intermediate
frequencies, up to a few hundred MHz, the skin effect increases, the transverse
permeability becomes increasingly dependent on the axial DC field. The surface
domain walls move under the influence of both the DC axial and the AC circular field
reaching a maximum at the anisotropy field of the surface domains. But, as the
frequency increases, their inertia inhibits their motion and therefore prohibits their
contribution to GMI above a certain frequency threshold. The eddy currents induced
and the relevant relaxation mechanisms are primarily responsible for the effect. At
higher field values, contributions from the rotation of magnetization are added as
well. As the frequency increases, the damping because of eddy currents becomes
strong and the domain wall relaxation is still no longer a mechanism of MI as opposed
to the magnetization rotation that takes over. At very high frequencies – GHz – the
effect is due to FMR.
It has been shown that the single peak structure evolves to a double peak structure
with an increasing number of pinning sites or as the magnetization process changes
from domain wall motion to magnetization rotation. The transition from the double

65
peak to single peak structure can be realized with increasing IAC or its frequency, or
annealing of the material to relieve it of pinning sites.
An extensive discussion of the GMI phenomenon, modeling, prospects and material
design is given in Ref [173].

4.3.4a GMI applications

The high GMI values, the low power consumption, the speed of the response, the
sensitivity and the controllability of the effect through proper choice of post-
preparation treatment of the material, or of the IAC magnitude and frequency make
GMI devices a very attractive option of field sensing, e.g., performance as high as
GMI ratios of up to 600% at 1Mhz and 10kA/m, and sensitivity of 1-2% A/m even at
a few decades of A/m has been reported [176, 181-183].
GMI field sensors are relatively low cost 2D or 3D field sensors and due to their
appealing room temperature characteristics, they are to compete with the fluxgate and
SQUID technology rather than with the GMR or TMR devices. Compared to
fluxgates they have about 30 times smaller sensing heads, they are 1000 times faster,
and they consume 1000 times less power [183].
A major drawback is the lack of linearity at very low frequencies or near zero field. In
such cases, asymmetrical GMI (see section 4.3.4c) sensors are preferred. Other
drawbacks include hysteresis effects due to irreversible magnetization reversal
processes and magnetic aftereffect but they both can be treated through proper
thermal treatment [175-176]. Their miniaturization capabilities based on hybrid
CMOS compatible structures using magnetic wires are still under investigation. GMI
sensors currently in the market are mainly microwires driven by CMOS circuitry but
the skin effect poses a limit in wire dimensions, especially at low frequency operation.
Thin film GMI structures would allow their integration with semiconductor
technologies and possibly open the way for many applications including high-density
recording. As field sensors they can also be used for 3D position sensing, current
sensing or NDT applications.
Because the GMI effect is strongly dependent on magnetization reversal processes,
domain wall formation and motion mechanisms, local anisotropy distribution, and the
origins of the transverse permeability, it can be used in applications other than
sensing, such as in the study of critical phenomena and phase transitions in magnetic
alloys, of the temperature dependence of magnetic properties or the measurement of
transverse permeability [179].

4.3.4b Stress impedance (SI) and sensors

The effect of stress-impedance, where the GMI ratio varies with the applied stress at a
given field value (or zero field), is also significant enough to be used in sensing
applications (Fig. 4.10). The applied stress can be either tensile or torsion and the
relevant applications are in sensitive tensile stress or torsion sensors.
Appropriate materials mainly include, as in the previous case, soft amorphous
(FeCoSiB) wires with small or negative magnetostriction. The application of applied
stress in wires with positive magnetostriction helps the formation of longitudinal
domains and the GMI ratio decreases continuously with the applied DC field. In wires
with negative magnetostriction, the tensile stress induces a hard axis in the
longitudinal direction, along which the DC field is also applied. The bamboo-like
surface domains are then forced to rotate towards it, giving rise to the symmetrical
two-peak structure with the peaks occurring around the stress-induced transverse
anisotropy field.

66
 (a)

(b)

Figure 4.10 Stress impedance response

The torsional stress induces a diagonal (i.e. 45o) magnetoelastic anisotropy with the
maximum value at the surface of the wire. The resulting torsion-impedance can be
used in torque sensors.

4.3.4c Asymmetric Magnetoimpedance (AMI)

In the case of asymmetric magnetoimpedance (AMI) the MI curve is asymmetric with
respect to the field origin with one peak smaller than the other depending on the
direction of the applied longitudinal field (Fig. 4.11). The AMI effect can be observed
when a DC bias current resulting in a DC circular field is applied to the conductor.
Another way to bias the AMI element is by placing it inside an AC longitudinal field
produced by a solenoid placed around the sensing element. Depending on the

67
excitation IAC frequency, the peak value can be as high a few hundred %. A
combination of two AGMI sensors can be used to form one GMI element with more
linear response at very weak fields [186-189].
Ha
Ib

Ia

Figure 4.11 AMI biasing scheme

The preferred magnetic configuration for the observation of the AMI effect is that of a
wire with helicoidal anisotropy that helps establish a sizeable circumferential
magnetization component. The asymmetry, when a DC bias field is applied, can then
be explained by the dependence of the magnetization reversal on the direction of the
applied field. This is also referred to as static AMI effect [186].
In the case of the AC field bias scheme, the mechanism responsible for the AMI effect
is the movement of the helical domain walls and the resulting change in the
circumferential magnetization (Fig. 4.10). It is also referred as the dynamic AMI
effect and resembles the mechanism in the high frequency Matteucci effect [173,
186], so such AMI sensors can also be considered as Matteucci sensors.
This mechanism depends on the cross-magnetization component, as suggested by Eq.
4.3, i.e., AMI can be observed when a nonzero off-diagonal component ζzφ of the
impedance tensor exists along with a nonzero axial field hz. This can be
accomplished by biasing schemes that affect the domain formation in a way as to give
rise to a cross-magnetization component as well as to the circumferential one. The
helicoidal anisotropy often induced during the preparation process of wires enhances
the cross magnetization component (ζzφ, hz) and therefore the AMI effect.
Another bias mechanism utilising the effect of the cross-magnetization component,
has also been proposed. A FeCoSiB ribbon composed of an amorphous bulk and a
crystallized surface structure may lead to a strong exchange interaction at the interface
inducing unidirectional surface anisotropy that acts as bias [188]. The major
advantage of this technique – still under investigation – is that no additional biasing is
required and therefore the power consumption is kept low. The asymmetry evolves
out of the symmetric response (at HDC=0) as HDC increases. At low frequencies
(100kHz) the asymmetry has a step-like response and the configuration has been
called a GMI-valve. At higher frequencies, the magnetization rotation takes over the
domain wall propagation and the asymmetry decreases [187]. This mechanism can be
generalised by using a cylindrical or planar ferromagnetic body, which has undergone
proper treatment in order to achieve a two-phase surface structure.

4.4 Magnetomechanical effects: dependent effects

The magnetostrictive effect is related to several effects known as magnetomechanical.
Some of the most important ones include the magnetostrictive delay lines, and the
Villari, Matteucci and Weidemann effects. The Wiedemann and Matteucci effects and
the AMI effect are all somehow related to the large Barkhausen jump and the flux re-
entrant mechanism observed in bistable wires.

68
4.4.1 Magnetostrictive Delay Lines (MDLs)
Magnetostrictive delay lines (MDLs) have been applied to a variety of physical size
measurements such as force, position, stress, torque, field, acceleration [3]. The basic
delay line arrangement is given in Fig. 4.12. A short coil and a search or detecting coil
is placed around each one of the two ends of the MDL terminated using latex adhesive
to eliminate acoustic reflections [3, 103-105, 190-197].
Magnetostrictive
Excitation coil line

sensing coil

Ie
Vo

Figure 4.12 MDL set-up

As described in 4.2.2, magnetostrictive materials tend to orient the magnetic dipoles

of the individual domains towards the direction of the externally applied field. Thus,
applying external bias or pulsed field along the MDL axis results initially in
Barkhausen jumps, which contribute to the hysteretic and irreversible part of the λ(H)
function and consequently in small angle rotation which is the anhysteretic and
reversible part of the said λ(H) function.
Therefore, polarizing the MDL with a dc bias field Hdcx, results in an elongation of the
material following the λ(H) function (Fig. 4.3a). When a pulsed field He(t) is
additionally applied at the region where the bias field has been applied, a similar but
dynamic elongation δλ(t) occurs, resulting in an elastic wave propagating along the
MDL, following the classical wave equation.
In classical magnetostrictive materials, the optimum pulsed field width is in the order
of μs. Thus, the wavelength of the propagating elastic wave is in the order of several
mm. Therefore, in the most common MDL elements, where the MDL cross section is
a tenth of mm2, a Lamb wave is propagating. Using materials with higher frequency
response or larger cross section can result in surface acoustic wave propagation. Skin
effect plays an important role in modeling and tailoring the behavior of the
microstrain generation and propagation.
The pulsed field along the MDL, responsible for the elastic wave generation presents
a decaying profile extending from the fully magnetized central region to a limit which
is practically of the order of the excitation coil diameter, indicating the active region
of the magnetostrictive material involved in the microstrain generation.
This elastic wave propagates along the length of the MDL, mainly as a longitudinal
elastic wave, because of the shape of the acoustic wave guide: the short cross section
with respect to the wavelength and the dimensions of the MDL eliminate any
transverse and quasi-transverse waves.
The propagating elastic wave, in its course, changes the local magnetization
component along the MDL axis, provided that the MDL is locally magnetized. The
total, macroscopic change of the magnetic flux along the axis of the wire is the result
of the statistical sum of local infinitesimal changes in the orientation of magnetic
dipoles, in the course of the propagating elastic wave. Thus, the magnitude of the
biasing field determines the change of the local magnetization component along the

69
MDL axis. This is actually the inverse magnetostriction effect. In some materials, the
earth’s field can be enough to polarize and consequently cause the presence of such
effect.
50

40

Voltage output Vo (mV)

A morphous Fibe r

30 A morphous ribbon

(a)
20

10 G la ss c ove re d fibe r

0
0 0,5 1 1,5 2 2,5 3
Pu ls ed excitation field He (kA /m)

35
A mo rp h o u s rib b o n
G las s co v ered fib er
30
Voltage output Vo (mV)

A mo rp h o u s fib er

25

20 (b)
15

10

0
-1,5 -1 -0,5 0 0,5 1 1,5
Bias field (kA /m)

50
Amorphous fiber

40
Voltage output Vo (mV)

Amorphous ribbon
(c)
30

Glass covered fiber

20

10

0
0 50 100 150 200 250 300
T ravel of acoustic wave (mm)

80
70
Amo rp h o u s fib er
Voltage output Vo (mV)

60
50
Amo rp h o u s rib b o n (d)
40
30
Glass co v ered fib er
20
10

0
0 5 10 15 20 25 30
Applied force (N)

60
Voltage output Vo (mV)

50 Glass covered fiber

Amorphous fiber

40
(e)
30

20

10

0
0 0,5 1 1,5 2 2,5 3
Applied torsion (turns/meter)

Figure 4.13 MDL output curves

Thus, if an inductive means, like a search coil, is set around the MDL, a pulsed
voltage proportional to the first derivative of the flux is induced across its ends (Fig.

70
4.14). The search coil should be set at a distance x from the elastic wave point of
origin (PO), which ought to be small enough to cause negligible attenuation and large
enough to avoid electromagnetic coupling between excitation and detection means.
Iac
~

torsion

Vo

Figure 4.14 The Matteucci effect

The parameters affecting the MDL operation are the pulsed magnetic field He, the
biasing fields at the point of origin and the search point and the mechanical properties
of the magnetostrictive material.
Changing one and only one of the above mentioned parameters by a given physical
size can result in a modification of the amplitude of the received pulsed voltage output
and result in a different sensor.
All kinds of classical magnetostrictive materials, like ribbons, wires, glass-covered
wires, thin films and composites can be used for all these MDL arrangements.
Attention must be paid to the sign of magnetostriction. Positive magnetostrictive
elements result in a decrease of the magneto-mechanical coupling factor or the λ(H)
function due to the applied stress, where negative magnetostrictive elements exhibit
the opposite. The sensor designer should select the magnetostrictive element mainly
with respect to this principle.

4.4.2 Villari effect

The Villari effect is closely related to the inverse magnetostriction effect. The
magnetic state of the material changes as a function of an external mechanical stress
[198-200]. The induction B for a given field H changes with stress σ, yielding the
characteristic curve B ( σ ) H that goes through an extremum at the Villari point,
⎛ ∂B ⎞
⎜ ⎟ = 0 (Fig. 4.15). More on this effect can be found in Section 4.2, where the
⎝ ∂σ ⎠ H
volume magnetostriction effect is analytically discussed. The Villari effect is mostly
used in stress and torsion sensors (see section 5.3.2).

4.4.3 Matteucci effect

The Matteucci effect is the generation of a voltage pulse at the ends of a
magnetostrictive wire, preferably but not necessarily with helical domain structure,
being magnetized by an axial ac field while torsion is being applied (Fig. 4.14). The
application of torsion couples the longitudinal flux to the circumferential flux Φc
inducing a voltage drop Vm at its ends:

71
Figure 4.15 Villari effect: B(H) loops at several stress levels

d Φc
Vm = (4.4)
dt
The helicoidal anisotropy is induced during the preparation method and can be
enhanced by annealing or realized by applying torsion to the wire [201]. The
Matteucci effect, even though it is closely related to the inverse magnetostriction,
AMI, and the Wiedemann effects, it is very convenient because it does not require
pick up coils to measure the flux changes and therefore the sensor design is greatly
simplified [102, 201-203].

4.4.4 Wiedemann effect

The Wiedemann effect is depicted on Fig. 4.16. A wire inside an axial magnetic field
Hz carries and ac current I that generates a circular field Hφ. The result is then a
helicoidal technical magnetization, with both a longitudinal and a transverse
component, which simulates the effect of torsion to the wire. Such magnetization can
be depicted as voltage output at the ends of a detecting coil set around the wire.
The inverse Wiedemann effect (IWE) (Fig. 4.16), implemented in several sensing
arrangements [204] consists in the appearance of a longitudinal component of
magnetization in a twisted wire that is being subjected to a circular magnetic field
generated by an ac current I flowing through it. The applied torsion produces an
effective helicoidal anisotropy field at the surface of the wire. The longitudinal
component of the magnetization is then the result of the propagation of a domain wall
along the core under the combined influence of an axial and a circumferential
effective field generated by the ac field and the applied torsion [204-206].

4.4.5 Magnetoacoustic emission

The magnetoacoustic emission is also called the Barkhausen effect because it is
related to the Barkhausen jumps occurring during the magnetization processes in a
magnetostrictive medium (see 2.9a). The domain wall displacements, causing the
Barkhausen jumps, generate local strains that propagate as acoustic waves inside the
medium. The presence of magnetostriction enhances the effect enough to be detected
by a microphone. The Barkhausen effect is used in non-destructive testing (NDT) for
the detection of cracks and defects in ferromagnetic substances [207].

4.5 Other effects

Other effects related to the magnetic state, properties, or processes in a medium
include the magnetothermal, Bitter, Kerr, faraday, and Hall effects reviewed below.
The Josephson effect (section 3.2.1a) based on superconducting materials and the

72
Zeeman effect based on the split energy levels of alkalic gases are also used in
magnetic field sensing. Finally, the Suhl and Nernst effects are reported with respect
to semiconductors but are beyond the scope of this paper. The Suhl effect, also known
as the magnetoelectric effect, is about the conductivity dependence on an applied
magnetic field, while the Nernst or photomagnetoelectric effect, concerns the
temperature, magnetization and electric field dependence on the magnetic field.

Figure 4.16 (a) Experimental setup used for IWE experiments. 1.Glass covered amorphous
wire. 2. Ceramic tube. 3. Pick-up coil. 4,5. Electrical contact. 6. Torsion device. 7.
Stress device. 8. Fixing support. 9. Frequency generator. 10. Power amplifier. 11.
ac amplifier. 12. Filter. 13. Voltage detector. (b) The IWE voltage response for a
FeSiB glass covered amorphous wires at 0mT, 0.025 mT and 0.04 mT (top to
bottom)

73
Figure 4.17 Barkhausen effect used in NDE of cracks in a medium magnetized by a sinusoidal
excitation

4.5.2 Bitter effect

The Bitter effect uses the stray fields generated by the magnetization divergence
( ∇ ⋅ M ) of a ferromagnet to record the domain microstructure. It consists in applying
a magnetic colloid on top of the pattern of interest. The flux emanating from the
domain walls where the flux divergence is highest magnetizes the particles of the
colloid orienting them along the flux lines. Hence, magnetic particles are concentrated
at the domain wall areas.

4.5.3 Magnetoopics – the Kerr and Faraday effects

Plane polarized light interacts with the magnetic moments inside a magnetized body
resulting in the rotation of the polarization of the transmitted or reflected ray.
Classically, magneto-optical (MO) phenomena are described by the Lorentz force
(Eq. 2.22). When the magnetization M is perpendicular to the surface of incidence, the
Lorentz force causes an oscillatory motion perpendicular to the M and to the plane of
polarization. This motion causes a rotation in the plane of polarization of the reflected
as well as the transmitted beam. The first one is the Kerr rotation and the second one
is the Faraday rotation. They are both dependent on the magnetization state and
therefore, they can be used to observe domain wall structure, motion and dynamics.
The Kerr effect (Fig. 4.18) can be used with any light absorbing magnetic material,
but it is very weak and image-processing techniques need to be employed. The
Faraday effect (Fig. 4.19) can be applied only on transparent media, like garnets, and
the effect is strong enough to be used without further processing. The Kerr and
Faraday effects can also be used as field measurement techniques, since the rotation is
proportional to the field H (or magnetization M). The MO techniques are appropriate
for the observation of the dynamic behavior of domain walls contrary to the “slow”
Bitter effect. Sensing applications are illustrated in section 5 and 6.

Figure 4.18 Kerr effects

4.5.4 Hall Effect

The Hall effect is based on the transverse Lorentz force (Eq. 2.22) experienced by
charge carriers inside a magnetic field, B. When a current carrying conductor is
placed inside a magnetic field it experiences a Lorentz force perpendicular to the

74
plane defined by the field and the current. When a magnetic field Bz is applied along
the z-direction of a magnetic solid (Fig. 4.20), i.e. a thin semiconductor film, and the
current density flows through the sheet along the y-direction, a voltage VH appears
between the electrodes along the x-direction. The current density of the sheet is
J=I/wt, where w and t are the width and the thickness of the wafer and the electric
field in the x direction is Ex=VH/w. The magnetic field Bz forces the carriers of current
I to deflect towards the electrodes thus establishing the electric field Ex. The force
exerted by this field called the Hall field, EHall is equivalent to the Lorentz deflection.
In the steady-state, this gives:
qEx = − q ( v y × Bz ) (4.4)

Figure 4.19 Faraday effect

I w
VH
t

y
x

Figure 4.20 Hall effect

75
In general,
E Hall = v × B (4.5)
The corresponding voltage VH is the Hall electromotive force.
When the current I is constant the carrier drift velocity vy is also constant:
vy=μcEy= Jy/nq= JyRH
where μc is the carrier mobility, n the number of carriers per unit volume, and
RH=1/nq is the Hall coefficient RH. The ratio
Ex
= −μ c Bz (4.6)
Ey
is called the Hall angle and is a measure of the Hall probe sensitivity.
Therefore, at constant current, the Hall voltage VH is proportional to the induction Bz
and dependent on the Hall sensor geometrical and conducting characteristics:
t VH
Bz = − (4.7)
RH I
The field dependence of the Hall voltage in semiconducting materials is the ordinary
Hall effect used in field sensing while its magnetization dependence in ferromagnetic
substances is the extraordinary Hall effect (see 4.3.2).
Hall probes are usually made of made of InSb because of its high mobility and high
MR. Bismuth nano-Hall sensors have been used in Scanning Hall Probe Microscopy.
Hall magnetometers are probably the most popular field sensors because of the wide
range of fields they measure and their compatibility with VLSI (Very Large Scale
Integration) circuits. However, the precision of these instruments, in the order of 100
μT, is not adequate for the measurement of very weak magnetic fields, i.e. around or
less than the order of the earth’s field.

76
5. Sensor arrangements

The various magnetic effects and materials described in the previous chapter have
been used as the core of many sensing elements. These sensing elements are divided
into five distinct families of physical sensors and are presented next. As it will be
shown, most of them are based on field dependence devices. We start the presentation
with the field sensors for two reasons: the one is that they are responsible for a
number of other sensors based on magnetic effects and materials. The other is that
they occupy a significant share in the global market, larger than the mechanical sensor
one, since they are also used in recording systems. The second family comprises the
position sensors and the third one the stress sensors. The fourth family of sensors is
the security sensors, which represent a special market niche with increasing and large
total budget. Smart sensing systems are the fifth family of the presented sensors, with
an emphasis to multi-parametric sensing elements.

5.1. Field sensors

This is probably the largest percentage in the magnetic sensors market. The most
frequently used field sensors are the ones used for small field variations or field
gradient or magnetic anomaly detection (MAD) [208]. The most important techniques
and effects used to develop such field sensors based on magnetic materials are the
magneto-resistance effect, the magneto-impedance effect and inductive techniques
like fluxgates [5]. Magnetostriction has also been used in field sensing but only in
special applications, as it will be seen later.

5.1.1. Primary Standards

The primary standard in magnetic field sensors is the superconducting quantum
interference device (SQUID) based on the Josephson junction effect [23, 24]. As
known, the tendency in sensors and primary standards is that primary standard
measuring techniques should be considered not only just the most sensitive sensing
devices, but also the ones able to measure by definition a quantum entity, like field,
current etc. As described in Chapter 4, this sensor is made of a closed loop of
superconducting material separated by a thin insulating layer. Such a thin insulator
may operate as single electron transport tunnel when it is properly biased. Indeed,
transmitting the smallest possible magnetic quantity, the Bohr’s magneton along the
Josephson junction, results in a pulsed voltage output increment across the two ends
of the superconducting materials. Characterization of SQUID sensors has exhibited
the lowest possible noise level in field sensors. The lowest one has been found to be
0.1 fT @ 1 Hz, concerning high purity single superconducting crystals of low Tc
(critical transition temperature), tested in very well magnetically isolated champers.
Such sensitivity is achieved only with liquid helium superconducting in magnetically
shielded rooms of 115 db and better. The SQUID noise level increases with the
critical transition temperature of the superconducting element as well as the magnetic
shielding ability of the testing room. Concerning high Tc SQUID sensing elements
(liquid nitrogen superconductors), the noise level cannot be better than 1 fT @ 1 Hz,
while most of the measurements report noise levels of the order of 10 fT @ 1 Hz. The
main disadvantage of these sensors is that they require helium or nitrogen cooling
systems to operate. Therefore, their cost and cost of operation are quite high, of the
order of 10 kEuros to 100 kEuros per sensing element, thus allowing their use in

77
special cases only, like primary level calibration facilities and biomedical
applications.
Before the appearance of SQUID sensors, the standard in magnetic field
measurements used to be the proton and gas field sensors [209]. These sensors, which
currently serve as very sensitive magnetometers mainly dedicated for military sensing
applications and mapping, are based on the Zeeman effect. According to this effect,
when a specific gas of a low electron element is excited, its electrons are moved to a
higher energy level. When de-energized, electrons drop down to its previous state, but
the new “steady state” level is one of the two hyperfine states, separating the spin up
and spin down stage. So, the energy state level of the electron depends on the ambient
magnetic field. The noise level of this sensing element depends on the purity of the
sensing gas and the conditions of measurement operation. The lowest ever measured
noise level has been 1 fT @ 1 Hz, while common measurements are of the order of 10
fT to 100 fT @ 1 Hz (0.1 pT). They are less noisy in the presence of ambient field
compared to SQUID devices. They also cost a good amount of tens of thousands of
Euros but they are less demanding devices compared to SQUID sensors. However,
they are still huge devices with respect to other field sensors.

5.1.2. Second(ary standard) bests

The second best sensors in terms of sensitivity and noise level are the fluxgate sensors
[107, 210]. The main principles of this type of sensors have been illustrated in
Chapter 4. The evolution of soft magnetic materials in terms of manufacturing and
post-cast treatment, allowed B-H loops with negligible hysteresis, exhibiting either
sharp and bistable behavior or linear suddenly saturated B-H loop, as illustrated in
Fig. 5.1. These materials, may exhibit magnetic hysteresis as low as a few A/m with
very low Barkhausen noise. Hence, the sensitivity of these elements has been
determined to be in the order of 10 pT @ 1 Hz. These devices are not as spacious as
the proton – gas sensors and do not require any special systems to operate, like the
cryogenic demands of the superconducting SQUIDs. But they are still far from
meeting the current demand for miniaturization. A fluxgate having the above
mentioned noise sensitivity, either in the form of toroid, or in the form of race track or
even in the form of strip element, has dimensions of several cm. Any attempt to
decrease their size has resulted in a decrease of the noise sensitivity. For fluxgate
sensors in the size of several mm, the noise level increases up to 100 pT-1 nT @ 1 Hz.
M

Figure 5.1 Bistable B-H loop.

78
Further decrease of the size of the fluxgate down to several μm, results in noise levels
in the order of 100 nT @ 1 Hz. The reason for this is the magnetic response of the
sensing core: hysteresis becomes significant and Barkhausen jumps are present. The
smaller the size of the magnetic element, the wider the B-H loop and the more
prominent the Barkhausen jump.
In order to overcome the problem of the size, the recently developed GMI technique
has been employed in field sensing development [173, 174, 178, 181]. Sensitivity
levels in the order of 100 or even 10 pT @ 1 Hz have been reported in some special
magnetic core materials. In the same way, a combination of the fluxgate and the
magneto-inductive principle has been employed to realize smaller and low noise field
sensors [211]. Other techniques have also been proposed, employing magneto-
transport properties, all targeting the low noise response at low frequencies. As an
example, a pseudo-superparamagnetic film has been reported, according to which the
magnetic noise due to Barkhausen jumps can be overriden, by using rotating in plane
field with magnitude much larger than the anisotropy field of the film [212]. At this
moment, it can be said that there still is a need for further research and development
of magnetic field sensors concerning the realization of miniaturized field sensors
integrated with CMOS technology if possible to offer sensitivity of the order of 0.1 pT
to 1 pT @ 1 Hz.
Up to this point all the magnetic field sensors described refer to dc or quasistatic field
detection. If high frequency fields are to be measured, coil sensors or antennas are
used, their response extracted from the Maxwell equations. Usually, they are used
empty without magnetic sensing core inside. This way their sensitivity is by definition
smaller, but noise level is negligible due to the 1/f law. If higher sensitivity is
required, high frequency soft ferrites are use. There is also an open question at this
point, namely being the development of linear, unhysteretic and low Barkhausen noise
magnetic sensing cores.

5.1.3. Working standard field sensors

Magneto-transport sensors are very common in the sensor market. The most well
known elements are the classic magnetoresistive or anisotropic magnetoresistive
heads [6]. These devices exhibit in plane anisotropy and despite the fact that the MR
ratio is of a few %, their noise resistivity is of the order of 10 nT @ 1 Hz. Their cost is
between 10-20 Euros, depending of the post-cast treatment of the magnetic core.
Another type of magneto-transport field sensor is devices based on GMR multilayers,
a stack of repeated bilayers made of a magnetic and a non-magnetic conducting film
[88, 148]. In this case, the resistance is measured across the multilayer and the effect
of the Lorentz force altering the path of conducting electrons due to the perpendicular
magnetic field is multiplied due to the several MR layers. Special structures of GMR
elements utilize multiple layers of a basic tri-layer cell made of a hard magnetic film,
a soft magnetic film and a non-magnetic film all of them being conducting [90]. Spin
valves, spin tunneling devices (like TMR sensors) have also been developed for field
sensing [150] based on the tunneling principles described in chapter 4. They have
been developed targeting vast numbers of production and low cost. The sensitivity of
these elements is very similar if not worse than the sensitivity of the AMR films,
despite the fact that the MR ratio can be of the order of tens percent. The reason that
they have been developed and produced is that they are very sensitive in sudden and
sharp changes of magnetic field. This way, they may be manufactured in vast
quantities following microelectronic techniques like sputtering deposition, wet or dry
etching lithography and wire bonding, offering not very good quality of field sensing,

79
but in the same time offering low cost and reliable operation. Therefore, they can be
used as recording heads in hard disks with a lower cost. These sensors (GMR and
STD sensors with low quality) are not analyzed very much in this paper, because the
subject of recording process is out of the scope of the paper. GMR and STD devices
have also been developed for field sensing. Their cost is of the order of 10 – 100
Euros offering sensitivity of the order of 1 nT to 10 nT @ 1 Hz.
The magneto-impedance (MI) effect in amorphous and nanocrystalline wires [176]
allows for much better levels of sensitivity and uncertainty, in the order of 1-10 pT
and 100 pT respectively @ 1 Hz. Unfortunately, this claimed sensitivity is not very
repeatable and co-tracable. A probable reason for this is the temporal dependence of
the MI sensing cores. The cost of such devices is in the order of 100 – 300
Euro/sensor. These attractive features allow the use of MI sensors in industrial and
automotive applications, although their cost is relatively high in comparison to
magneto-resistance sensors. A couple of research teams have been trying to develop
durable MI sensors on thin films for mass production [173, 183]. Generally, the
problem in many magnetic thin film sensors is that the sensor characteristics become
less attractive in comparison to the bulk sensing elements. Recently, an interesting
combination of the MI and fluxgate effect in thin film structures has shown promising
results in two dimensional field measurements, by using amplitude and pulse width
modulation techniques, offering sensitivity in the level of 10 pT [213].
Another magnetic technique takes advantage of the magnetostriction effect and the
generated magneto-elastic waves in order to realize a distribution field sensor for non
destructive evaluation of magnetic surfaces [214] offering noise level of ~ 10 nT @ 1
Hz.
The most traditional magnetic field sensors, the Hall effect devices [215], are still in
the market despite the poor noise level characteristics but thanks to their low cost,
miniaturization capabilities and inherent compatibility with CMOS technologies.
Their noise level is of the order of 0.1 to 1 mT @ 1 Hz and their cost ranges from 2 to
20 Euros per sensor concerning switches (poor quality sensors) and magnetometers
respectively.

5.2. Position sensors

Position sensors have a relatively large percentage in the global sensor market,
probably the second best after the magnetic field sensors. We group position sensors
into three main categories. The first category refers to devices which are defined as
primary or secondary standards, i.e. devices with well established and recognized
properties being as close as possible to the reference measurement. The second
category refers to sensors able to detect absolute or differential static position. The
third and final category refers to sensors able to detect dynamic displacement and
derivative entities. Typical schematics of position sensors are illustrated in Fig. 5.2.

5.2.1. Primary and secondary standards

Since when the speed of the light in vacuum was established as a universal constant
and equal to 2.99792458×108 m/s, the primary standard of measuring the meter has
been defined as the path of light in vacuum, which is covered in (1/2.99792458)×10-8
seconds. The realization of such an experiment requires the use of vacuum and of
course the use of a second measuring instrument, the clock, which makes the above
mentioned definition of primary standard dependent not only on universal constants
but also in measurands.

80
 (a)

1 2 3

(b)

3
2
1

(c)

2 3 4
1

(d)

1 2

(e)

2
1

Figure 5.2 Position sensors based on magnetic effects

(a) Magnetostrictive delay line (MDL) set-up. (1) Movable excitation coil, (2)
MDL, (3) Movable search coil.
(b) Linear variable differential transformer (LVDT) set-up. (1) Soft magnetic
material, (2) Excitation coil, (3) In series opposition search coils.
(c) Inductive set-up. (1) Search coil, (2) Excitation coil, (3) Soft magnetic
material, (4) Movable magnetic element, closing the magnetic circuit.
(d) Permanent magnet type. (1) Permanent magnetic thin films, (2) Searching
magnetic head.
(e) Angular positioning sensor. (1) Radial permanent magnets in tooth
arrangement, (2) Searching field sensor.

But, this is really considered as a minor problem from all Metrological Laboratories,
after the development of atomic clocks offering a time uncertainty of the order of 1 fs.

81
Furthermore, the discovery and development of frozen atomic clocks resulted in a
further improvement of the time measurement uncertainty, at a level of 10-3 fs! So, the
error in measurement caused by the time measurement is really small. Instead, the
stability of the light source and moreover the level of vacuum are far more important
parameters introducing higher orders of error. The stability of the light source is
mainly considered as frequency or wavelength stability. Therefore, single wavelength
laser sources with low distortion in time are rather preferred. (Conventional light
sources are not used for many reasons, the most important being the inclusion of
many wavelengths, the unpredictable distortion in time and the multi-phase waves.)
Inter-laboratory comparison tests provide useful information in maintaining the
Gaussian distribution of the light source wavelength as sharp and steady as possible
with respect to wavelength and phase. Apart from that, vacuum conditions also
determine the uncertainty of measurement. Ultra-high vacuum of the order of 10-10
mbars is used for the best experimental set-ups, while molecular pump high vacuum
of the order of 10-9 mbars is also acceptable. Of course, the whole experimental set-up
has to be maintained in an environmentally steady area. The most important
parametric effects are E/M field, temperature and relative humidity. An acceptable
level of electromagnetic field shielding is 115 db with dc shielding ability.
Additionally, temperature and humidity levels must also be maintained constant
(typically 20oC +0.1oC and RH 50% +2%). From all these parameters it can be seen
that the maintainable level of uncertainty in measuring the meter depends on the
experimental conditions of the experiment. Single light path measurements in the
highest possible specification level set-ups guarantee an uncertainty of +1nm in 10
meters (0.1 ppm uncertainty), while multiple light path measurements can improve
this uncertainty. Usual set-ups in Metrological Laboratories offer a maintainable
uncertainty of 1 ppm. The cost of such an experimental set-up starts from several
hundreds of thousands of Euros for an uncertainty of 1 ppm up to a couple of millions
of Euros for a 0.1 ppm uncertainty.
The above described reference position measurement cannot be used for real field
measurements. Therefore, second best sensors, the so called secondary standards are
used as reference sensors in position measurements. They are differential position
sensors, able to measure the interferometric response of a reflected, single
wavelength, polarized laser light. In optimized conditions they are able to measure
distances as low as the ¼ of the transmitted wavelength, i.e. the distance
corresponding to the first peak or valley of the standing light wave; hence,
considering that the modern lasers with relatively stable and sharp Gaussian
distribution of wavelength are of the order of 300 nm, these sensors, when used in
well defined environments are sensitive to differential position in the order of 75 nm.
In harsh environments, the sensitivity levels decrease down to 100 nm. Some of these
devices are equipped with electronics able to determine the analog output of the
search photodiode corresponding to the intermediate position between a peak and a
valley, thus reaching a sensitivity of 3-10 nm. The cost of a complete and certified
laser interferometer set-up is of the order of 100 kEuros. The problem with this device
is that it is sensitive to temperature, humidity and dust conditions, so that it cannot
really be used in harsh environments like production lines etc. Of course, laboratories
performing any kind of position detection, requiring sensitivity measurement of the
order of nm, ought to be equipped with such an instrument. It is worth mentioning that
an interesting research target is the development of a position sensor able to meet the
specifications of the laser interferometer, having the additional ability of been used in

82
harsh environments as well as of measuring absolute position and not only
differential.

5.2.2. Absolute position sensors

Absolute position sensors are able to detect the absolute distance between two points.
These two points are usually the exciting and detecting means. Examples of absolute
position sensors are the magnetostrictive delay line (MDL) set-up [216], which
combines time delay detection and voltage modification, the linear variable
differential transformer (LVDT) [217], which, despites its name is able to detect
absolute position and a linear inductive sensor using closed magnetic paths [218], as
shown in Fig. 5.2 (a), (b) and (c) respectively. Additionally, field sensors can be
converted to position sensors by means of using a well placed permanent magnet
displaced on top of the field sensor. Their sensitivity and uncertainty can reach 10 μm
and 100 μm/m respectively. Their cost is of the order of 10 – 100 Euro/sensor. All of
them can be cordless devices. Their competitor is the ultrasonic position sensor which
is less expensive but less accurate too.

5.2.2.1. MDL position sensors.

MDL position sensors have been thoroughly investigated in the past, with interesting
results in the field of position sensing.

Measuring tapes
Measuring tapes is the existing off-the-shelf product based on the MDL technique. All
these sensors existing in the market are based on the measurement of the delay time
between two signals, the one being the reference signal and steady in position, while
the other is dependent on the position of the sensing core. A typical example of such a
sensing instrument can be realized by using the classical MDL set-up: an excitation
coil is used for the generation of the elastic pulse and a search coil is used for the
detection of the propagating elastic pulse. Considering that one of the two coils is
steady in position and the other is the moving core of the sensor, the output of the
device is the delay time between these two signals. In such an arrangement, the
sensitivity and uncertainty of the sensor mainly depends on the sound velocity and the
sound velocity uniformity of the MDL, which is of the order of 5 km/s, as well as on
the geometrical instability of the moving coil [219]. Experimental results have
illustrated that the sensitivity of this sensor in well-prepared magnetostrictive wires
can be from 0.2 mm up to 0.1 mm and the uncertainty ~0.3 mm/meter or 300 ppm.
Alternatively, use of ac excitation field results in similar response [220, 221]. The
best-developed example of such a sensor is illustrated in Fig. 5.3 [222]. The sensor
basically consists of a two-part cylinder, with a conducting inner core, preferably
made of copper or aluminum and an outer magnetic thin layer, deposited on the
conducting inner core. The outer magnetic thin layer is tailored to have
circumferential magnetic and magnetoelastic anisotropy. Therefore, passing pulsed
current through the conducting element generates circumferential pulsed magnetic
field on the magnetic thin layer. Such a field is able to generate transverse elastic
waves along the whole surface of the magnetic layer due to its circumferential
magnetoelastic anisotropy. Provided that the magnetic layer is magnetoelastically
uniform and that there are no local magnetic anomalies in the ambient field, the local
transverse elastic waves cancel each other, resulting in no elastic wave propagation
except the elastic waves originated from the very ends of the two-part cylinder.

83
 Magnetostrictive outer cell

Conducting inner cell

(a)

R Ie=0

VR

R Ie=Id

VR

R Ie~Is

VR
(b)

Figure 5.3 The new magneto-elastic device. (a) The device. (b) The operation of the device.
Transmitting no current through the conductor results in no net macroscopic
magnetization. Current Id, larger than a critical value able to cause domain walls
motion, results in magnetization of the wire. Transmitting pulsed current close to the
saturation levels results in micro-strain generation.

84
Due to the bias field effect, using a moving ring magnet around the cylinder, results in
a local break of the symmetry of the elastic waves, which in turn results in a
propagating elastic wave. The elastic wave is then detected using a search coil, fixed
at the one end of the two-part cylinder. The main advantage of this principle is that the
transverse sound velocity is about half the longitudinal one. Therefore the obtained
sensitivity and uncertainty can be 0.1 mm up to 50 μm and ~150-200 ppm
respectively. Another advantage is that this sensor is cordless, meaning that the
moving coil does not involve any movement of electric wires, which is important
from many points of view, like noise inducing, compatibility with industrial
environments etc.
In order to keep the advantage of the cordless operation and improve the sensitivity
and uncertainty of measuring tapes based on the MDL technique, a sensor illustrated
in Fig. 5.4, may be used. The sensing element (1) is a magnetostrictive delay line
(MDL) in the form of ribbon or fiber. A short excitation coil (2) is set around the one
end of the MDL and an array of short, single layer coils (3), connected in series and
named hereinafter “search coils”, is spread around the MDL along its length, being
used as the sensor output. A moving hard magnet (4), able to be displaced parallel to
the sensing material, is the active core of the sensor. Without any loss of the
generality, in a specific application, the moving magnet was set at the end part of a
hydraulic piston. Details of such arrangement can be found in [216]. The sensor
operates as follows: Pulsed current is transmitted through the excitation coil,
generating an elastic pulse, which propagates along the MDL length. The propagating
elastic pulse induces a pulsed voltage train in the serially connected search coils, with
pulse intervals corresponding to the distance between consequent coils. These voltage
pulses are proportional to the ambient field along the axis of the search coils. In the
absence of the moving magnet and low ambient field along the array of the short
search coils, these voltage pulses are small in amplitude. In the presence of the
moving magnet, the voltage pulses of the neighboring coils become larger. The closer
the moving magnet is to a coil, the larger the corresponding voltage pulse, following
the classical dependence of magnetostriction and inverse magnetostriction on the
ambient field. Thus, if the moving magnet core is approaching three consequent
search coils, the voltage output of these three coils overcomes a preset threshold and
indicates that the magnet approaches the vicinity of these coils. Having tailored the
magnetostrictive element in a way to retain a uniform and monotonic output response
with respect to the ambient field along its length, the amplitude of the voltage pulses
can be the indication of the distance of the magnet with respect to the three coils. An
algorithm, which can be used for the determination of the absolute position of the
moving magnet, includes three steps. At the first step, a voltage threshold comparator
determines the absolute position of the three neighboring coils, due to the delay time
of the pulsed outputs. At the second, a fast analog to digital converter detects the
amplitude of the three neighboring pulses. Finally, at the third step, a look-up table
using real time software indicates the absolute position of the magnet, using
correlation techniques. Thus, the position of the magnet with respect to the excitation
coil can be determined. This measurement is also the means for the auto-calibration
procedure.
The above described position sensor has been realized using FeSiB amorphous
ribbons and fibers. The dependence of the pulsed voltage output of a single search coil
on the displacement of the moving magnet, which corresponds directly to the position
of the moving piston, is illustrated in Fig. 5.5 (a). The uncertainty of measurement
concerning all search coils is illustrated in Fig. 5.5 (b).

85
 MDL Horizontally moving permanent Nd-Fe-B magnet

Ie
Pulse voltage output train of the in series connected search coils

1 2 3 4

6 7

Display

Step 1:
Three
neighboring
coils

Step 2:
Auto-
Counting the
calibration
delay times
(if required)
of the three

Step 3:
ADC and
determinatio ADC
n of magnet

5 9 8

Figure 5.4 The schematic of the position sensor. (1) Delay line, (2) Excitation coil, (3)
Search coil, (4) Moving magnet, (5) Electronic conditioning circuit, (6) Pulsed
current generator, (7) Signal amplifier, (8) Electronic counter, (9) ADC for
position tracking control.

86
Sensitivity of 1 μm and uncertainty of 10 μm/m or 10 ppm were determined due to
these measurements for the case of a carefully annealed amorphous fiber. These
ribbons and fibers after careful treatment offer a uniform, sensitive and anhysteretic
magnetoelastic performance along the length of the MDL, thus allowing the use of a
unique look up table for all search coils. Without such a unique look up table the
sensor could not measure position and displacement in real time conditions. In order
to industrialize such a sensor, one should be careful about the control of the λ(H)
function and the λ(H) function uniformity. Experimental results also show that the
minimum distance between two consequent MDLs, can be 5 cm, to avoid magnetic
interference for two-dimensional arrays for magnetically mapped plane surfaces,
laboratory tables and production lines dependent on terrestrial field.

100
Magnetoelastic response (mV)

90
80
70
60
50
40
30
20
10
0
0 5 10 15 20

Magnet-piston position (mm)

(a)
ribbon wire

10
8
Uncertainty (microns)

6
4
2
0
-2
-4
-6
-8
-10
0 5 10 15 20

Measuring length (mm)

(b)
ribbon wire

Figure 5.5 Response of the position sensor of Figure 5.4. (a) Output dependence on moving
magnet position, (b) Uncertainty of measurements.

Displacement sensors
There are many occasions in which large arrays of sensors are needed in order to
measure the distribution of displacement of objects along a surface. There are also
cases where the absolute steady state position of an object or the dynamic change of
the position of an object needs to be measured. In all these cases, one could use the

87
principles of the measuring tape technique presented above, concerning single point
output or array output response, where the dominating principle is the modification of
the bias field around the excitation or the detecting point. Displacement sensors based
on the MDL technique have been presented in the past [223]. Families of single point
and distribution sensors based on this principle have also been developed [224]. These
sensors can operate by displacing the moving magnet at either the point of elastic
wave generation or detection as illustrated in Fig. 5.6. A typical response of such a
sensor is given in Fig. 5.7. The limitation of this sensor is the non-linear response and
the uncertainty introduced by the moving magnet.

4
1

Ie Vo

(a)

3 4

Ie

(b)

Figure 5.6 Displacement sensors based on the moving magnet technique. (a) The magnet is
displaced along the excitation coil, (b) The magnet is displaced along the search
coli. (1) Pulsed current conductor, (2) Moving magnet, (3)Excitation coil, (4).
Search coil

Targeting a linear displacement sensor, a new sensing element based on a

combination of eddy currents and the MDL technique has been developed [225]. The
arrangement of this sensor is shown in Fig. 5.8 (a). A detecting coil is set around the
MDL at the one end of it. MDL is well terminated by latex adhesive. A small field

88
magnet bar is set close to the MDL in order to maximize the output of the detecting
coil. A pulsed current conductor is set parallel to the MDL, so that in the absence of
the movable disk, the output of the detecting coil equals zero.

80
70
60

Vo (mVolts)
50
40
30
20 (a)
10
0
0 5 10 15 20
Magnet displacement (mm)

80
70
60
Vo (mVolts)

50
40
30
20 (b)
10
0
0 5 10 15 20
Magnet displacement (mm)

Figure 5.7 Typical responses of the sensors of Figure 5.6. (a) Response of sensor of Figure
5.6(a), (b) Response of sensor of Figure 5.6(b).

Approaching the conducting disk close to the MDL-conductor arrangement, eddy

currents are induced in the disk. Keeping unchanged the pulsed current and the
biasing field along the MDL, the amplitude of the pulsed eddy current in the disk
increases as the distance between disk and MDL decreases. Such an eddy current
causes a pulsed magnetic field along the length of the MDL, at a region of the MDL
called sensing point, which in turn causes a pulsed voltage output, induced in the
detecting coil. The amplitude of this pulsed voltage is the output of the sensor. The
delay time between voltage output and exciting pulse defines the position of the
conducting disk. The higher the eddy current, the higher the magnitude of the
detected voltage output. Displacing the conducting disk along the MDL, results in a
change of the delay time between exciting pulse and detected voltage.
The experimental set-up has been illustrated in [225]. A typical response of this
sensor using stress-current annealed MDL is given in Fig. 5.8 (b), maintaining the
same distance between sensing point and detecting coil. It can be seen that the
illustrated response is a fairly linear function of the displacement of the disk.
Additionally, these results have been obtained for increasing and decreasing the
distance between conducting disk and MDL. The resulting curves are identical,

89
showing absence of hysteresis within the accuracy of the experimental apparatus
used. From these results, it can be concluded that such a sensor can detect
displacement from 0 to 5 cm, with sensitivity dependent on the measuring electronic
system. Use of a commercially available 14-bit A/D converter resulted in 1 μm
sensitivity and 100 ppm uncertainty. From the experimental results, it can be seen that
using stress-current annealed amorphous wire MDL one can obtain a more sensitive
response of the sensor, due to the increase of the magnetomechanical coupling factor
of the amorphous wire. A serious advantage of this sensor with respect to the state of
the art is that its response is quite linear in a significant range of measured
displacement.

1 2 3 4

(a)

45
40
(b)
35
30
Vo (mVolts)

25
20
15
10
5
0
0 10 20 30 40
Disk displacement (mm)

Figure 5.8 (a) The sensor based on eddy currents generated on a disk on top of the MDL. (1)
Pulsed current conductor, (2) Delay line, (3) Conducting disk, (4) Search coil.
(b) Typical response of the sensor.

Another application of such a set-up is an eddy current sensor for quality tests.
Having fixed the position of a conducting disk, the output of the voltage output
defines the amount of eddy current in the disk. This is an indirect source of

90
information of the resistivity of the disk. Tests among various diameters and kinds of
coins are in good agreement with respect to their resistivity. The applications of this
arrangement can be numerous, an important one being the security systems. It can
also be seen that this type of sensor can be used in measuring tape applications, but
with a limited sensitivity and uncertainty, since the principle of measurement is the
delay time based on the longitudinal sound velocity of the magnetostrictive element.

3 4

(a)

2 4

(b)

2 4

(c)

2 4

(d)

Figure 5.9 Dilatomeres based on MDLs. (1) Magnetic core S, (2) Delay line, (3) Pulsed
current conductor, (4) Search coil.

91
Dilatometers
The operation of such a sensor is based on the change of the magnetic circuit at the
point of origin (PO). It has been mentioned that the amplitude of the acoustic pulse is
qualitatively related to the change of the magnetic flux density of the cross section
area of the delay line, at the point of excitation. Regarding the arrangement of Fig. 5.9
(a), approaching a soft magnetic material, called hereinafter core S, close to the MDL-
conductor basic arrangement, results in a decrease of the magnetic flux in the MDL
and in a corresponding decrease of the amplitude of the generated elastic wave. In
Fig. 5.9 (b), the opposite effect takes place. Regarding the arrangement of Fig. 5.9 (c)
the balanced conductor structure results in the complementary use of the two above-
mentioned principles. These conductors are joined together at the ends, so that the
amplitude and direction of the pulsed current in both of them is the same. In the
absence of the core S there is no elastic pulse. Approaching the core S closer than 1
mm the flux in the MDL starts increasing, thus resulting in an elastic pulse.
Another displacement sensor is shown in Fig. 5.9 (d). In this case, the two active cores
S1 and S2 are above and below the MDL respectively and are joined together in order
to keep their distance fixed. When no displacement is applied, cores S1 and S2 are at
their maximum and minimum distance from the MDL respectively. Both of these
magnetic cores are involved in altering the resulting Vo. As the balanced structure is
also used here, Vo becomes zero when the distance between cores and delay line is
equal. One would expect that the gap between the two cores should affect the
performance of the sensors. A full explanation of the operation of this sensor as well
as the arrangement of the experimental set-up has been illustrated in [226].
The arrangement described above has been used to obtain the typical dependence of
Vo on the distance a of the core from the delay line assembly, illustrated in Figs. 5.10
(a), 5.10 (b), 5.10 (c), concerning the sensors of Figs. 5.9 (a), 5.9 (c) and 5.9 (d).
Experimental results have been taken by decreasing and subsequently increasing the
displacement a. The reported output is the same for both cases, i.e. the hysteresis
appears to be small. The response of this family of sensors is fairly linear between 0.5
and 1.5 mm.

Digitizers
Two and three-dimensional digitizers have been proposed in the past based on the
MDL technique. The very first digitizer has been proposed by Murakami et al [227],
using cordless moving magnetic stylus. Such a digitizer utilizes a short coil for
generating the elastic pulse in the MDL and a long coil as the receiving means. In the
absence of any magnetic material or local magnetic anomaly along the long coil, and
provided that the magnetostrictive ribbon or wire is magnetoelastically uniform, the
output signal along the long coil is zero except at the very end of it, where two small
in amplitude voltage pulses can be detected. Approaching the moving magnetic stylus
along the long coil, the magnetic symmetry in the corresponding region of the MDL is
broken, thus resulting in a pulsed voltage output, in a time position corresponding to
the distance between the moving stylus and the exciting coil. The obtained resolution
was 0.1 mm, while the maximum measurable distance between moving stylus and
long coil was 12 mm. These long coils can be arranged in parallel arrays, and vertical
arrays, thus allowing two-dimensional digitization. Furthermore, Meydan [228] has
developed a similar digitizer with improved resolution (85 μm) and improved
maximum measurable distance between moving stylus and long coil (15 mm). Apart
from that a careful study of the magnetic stylus was realized, thus finding the
optimum material, geometry and inclination during use.

92
 120

100

80

Vo (mVolts)
60

40

20 (a)
0
0 0,5 1 1,5 2
Core displacement (mm)

60

50

40
Vo (mVolts)

(b)
30

20

10

0
0 0,5 1 1,5 2
Core displacement (mm)

60

50
Vo (mVolts)

40

30

20

10 (c)
0
0 0,5 1 1,5 2
Core displacement (mm)

Figure 5.10 Typical responses of the sensors of Figure 5.9.

Other digitizers using the MDL principle have also been developed aiming to improve
the given state of the art. A force digitizer able to detect the position and the
magnitude of a force applied along a magnetostrictive element has been proposed
[229]. This device is based on changes of the pulsed elastic waveform due to a
mechanical action, applied on the MDL. For this operation, the delay line is just
regarded as an acoustic waveguide. The sensing principle is based on the generation

93
of a reflection of the propagating elastic pulse at a point of the delay line, when force
is applied on it.

Vo

MDL
Ie

(a)

70
60
50
Vo (mVolts)

40
30
20
10 (b)
0
0 0,2 0,4 0,6

Applied force (N)

Figure 5.11 (a) The principle of operation of a force digitizer based on reflections in MDLs.
(b) Typical response of the digitizer of Figure 5.11(a).

Hence, the magnitude and the time delay of the reflection with respect to the main
acoustic pulse define the magnitude and the position of the applied pressure
respectively. One can measure a pressure applied on a point of a surface. A valid
condition for this arrangement is that the pulsed current conductor is in between the
receiving coil and any possible sensing point. The principal idea of such a force
digitizer is shown in Fig. 5.11 (a). Details for the sensor description and operation can
be found in [230]. The dependence of Vo on the applied force at a point of the MDL
out of the point of origin (PO) and receiving coil region is illustrated in Fig. 5.11 (b).
For the given arrangements the smallest detectable force (manual observations) was
10 mN. The experiment was carried out by increasing and decreasing the magnitude
of the applied force F. The resulting output was the same for both ways of getting the
response of the sensor, showing that the hysteresis of the sensor is small (undetectable
under 8-bit digital readings). The advantage of this digitizer with respect to the above-
mentioned ones is that it can detect the magnitude and the position of the applied
force at the same time.
Fig. 5.12 illustrates a three-dimensional digitizer based on the principle of the
measuring tapes. In this case, an array of n parallel pulsed current conductors is set
orthogonal to an array of m MDL, thus generating a matrix of n x m points of origin of
elastic pulses. An array of m search coils is set at the one end of the MDL for

94
detecting purposes. A cordless stylus having a permanent magnet at the end of it is
used as the active core of the digitizer. Provided that MDL have been tailored
properly, the elastic waves in the absence of the moving magnet have a minimum
magnitude. In the presence of the moving magnet of the stylus at the neighborhood of
any rectangle of four points of origin, the corresponding pulsed output signals
increase with respect to the amount of field at each one of them. Comparison of the
four pulsed outputs with look up tables gives the absolute position of the stylus with
respect to the points of origin. An advantage of this digitizer is its ability to read in the
third dimension up to 50 mm. A certain disadvantage is the possible uncertainties
induced by the arbitrary inclination of the moving cordless stylus.

Active core

Pulsed current MDL Search coil

conductor

Figure 5.12 Principle of operation of a digitizer based on the moving magnet technique.

The two-dimensional digitizer, illustrated in Fig. 5.13, utilizes the principle of the
previously presented dilatometer. Following the balanced structure of Fig. 5.9 (c), two
arrays of n parallel pulsed current conductors are set orthogonal above and below an
array of m MDL, thus generating a matrix of n x m points of origin of elastic pulses.
An array of m search coils is set at the one end of the MDL for detecting purposes. A
cordless stylus having a soft magnetic material at the end of it is used as the active
core of the digitizer. In the absence of the stylus, the balance structure prohibits the
generation of elastic pulses. Approaching the soft magnetic material of the active core
close to the neighborhood of any rectangle of four points of origin, the corresponding
pulsed output signals increase. As in the previous case look up tables give the absolute
position of the stylus with the respect to the points of origin. This digitizer is a two –
dimensional device, since the ability in 3-d measurements is limited to 1 mm
maximum vertical measurement. An advantage of this digitizer with respect to the
previous one is the absence of induced uncertainties due to the difference in
inclination of the active core.

95
 MDL

Pulsed current

Search coil

Figure 5.13 Principle of operation of a digitizer based on the dilatometers of Figure 5.12.

The sensor of Fig. 5.8 (a) has also been designed as a two-dimensional digitizer. Such
a sensor could detect the position of the movable, conducting disk in three
dimensions. Experimental results show that it has the advantage of isotropic behavior
in three dimensions, thus competing with the digitizers using hard magnetic material
as stylus, where magnetic field anisotropy generates uncertainties in position
measurement. But, a disadvantage is that this digitizer can’t have sensitivity better
than 0.1mm, since it is based on the longitudinal sound velocity of iron based alloys.
All these cordless digitizers based on the MDL technique can be developed in sizes of
the order of square meters due to the ability of amorphous magnetostrictive ribbons
and wires to exhibit negligible acoustic attenuation factor.

5.2.2.2. LVDT sensors

The linear variable differential transformer (LVDT) is well known for its applications
in displacement measurements. The working principle is based on a differential
transformer with variable coupling between the primary and secondary coils. It
actually follows the basic set-up of the fluxgate element. The magnetic coupling
between the primary coil and the secondary coils depends on the type of magnetic
material and the position of a movable magnetic core with respect to the secondary
coils. Fig. 5.2 (b) shows the basic arrangement of the LVDT.
Improvement of LVDT sensors has been based on the use of amorphous wires and
glass covered wires. Conventional amorphous wires with diameters between 80 and
300 μm prepared by rapid quenching from the melt present a good sensitivity in the

96
as-cast form, of the order of 50 μm, while improved sensitivity of the order of 10 μm
appears after field annealing. The relatively recently developed glass covered
amorphous magnetic wires can replace successfully, in various applications, the
conventional amorphous wires produced by the in-rotating-water technique. These
materials consist of a cylindrical metallic core with diameter of 2–40 μm, covered by
a glass insulation with a thickness of 1–20 μm, thus allowing a by design corrosion
protection and packaging. Additionally, their mechanical flexibility can be used in
derivative devices where this property is desirable. Their sensitivity is of the same
order compared with conventional amorphous wires in the as-cast and after annealing
states.

5.2.3. Differential position sensors

Differential position sensors are able to detect the traveling distance of the sensing
head and not the absolute position. The most important differential position sensor
based on magnetic effects materials is the magnetic tape (Fig. 5.2 (d)). A magnetic
head reads the flux as it passes on top of a tape [231]. The reading means can be either
a field sensor like low cost MR or GMR devices or magneto-optic devices like laser
Kerr detectors. Therefore, this tape is a corded device and is constructed from a series
of hard magnets arranged in up and down orientation of magnetization. The sensitivity
and uncertainty of these sensors is of the order of 1 μm and 10 μm/m respectively.
Their cost is of the order of 1 kEuro/meter. Their competitor is the optic tape and has
similar properties, but it cannot be used in harsh environments like the magnetic tape.
A special type of LVDT arrangement, using no excitation circuitry has also been
employed for dynamic measurements of displacement, mainly dedicated for
dilatometric measurements.
Angular sensors can be either absolute or differential. The most classic sensor is the
rotating tooth (Fig. 5.2 (e)), which is a differential sensor, used in many important
applications like ABS in cars. As the disk holding the magnetic teeth is rotating, a
magnetic head counts them by means of a series of pulses. Interesting angular sensors
based on inductive and magnetic techniques, have also been proposed in the past
[232].
Velocity sensors and accelerometers are based either on calculation of the response of
absolute or differential position sensors, or detect directly velocity or acceleration.
Although most of the velocity sensors or accelerometers are based on calculation
process, the development of accelerometers is of interest. A simple velocity meter or
accelerometer can be an inductive sensor using an inertia magnetic mass, moving on
top of the coil. The motion of the magnetic mass introduces a change in the magnetic
flux of the coil.
The terminating switches are simply low accuracy field sensors with a threshold
checker, operating like on-off switch [6]. Their vast majority is magneto-resistive
sensors, since the cost of production of poor quality, mass produced elements is small.
Their main competitor is the capacitive switch based on capacitance change.
Meydan et al. [132] have developed accelerometers based on the inverse
magnetostriction effect, offering good stability and hysteresis characteristics.
Additionally, MDL set-up has been employed for accelerometer device development.
The schematic of an acceleration sensor based on the MDL technique, requiring no
excitation current for the generation of elastomagnetic waves, is illustrated in Fig.
5.14 (a). A long magnetostrictive ribbon, acting as MDL, is set on a glass substrate. A
short multi-turn search coil is set at the one end of the MDL. A permanent magnet is
set on top of the MDL having a fixed pole orientation with respect to the MDL and

97
being magnetically isolated from the receiving coil. An electronic integrator detects
and integrates the signals of the search coil. Dynamic displacement of the permanent
magnet results in a change of the magnetic flux at the corresponding intersection of
the MDL. Such dynamic flux change is translated into an elastic pulse propagating
along the length of the material and being detected as a single pulse shot at the search
coil. Provided that such a microstrain change has a bandwidth response corresponding
to the MDL bandwidth, the level of the integrated output is proportional to the first
derivative of magnet’s displacement. The frequency response of the accelerometer
corresponds to the MDL bandwidth, which is from 10 kHz up to 10 MHz. The
experimental set-up to determine the output of the signal has been analyzed in [233].
The sensor response is shown in Fig. 5.14 (b). It can be observed that the response is
monotonic for a large scale of acceleration.

3 4

(a)

35
30
25
Vo (mVolts)

20
15
10 (b)
5
0
0 10 20 30 40 50

Acceleration (m/s2)

Figure 5.14 (a) The schematic of the arrangement of the self excited accelerometer. (1) Delay
line, (2) Search coil, (3) Moving magnet, (4) Shielding material, (5) Amplifier and
integrator.
(b) Typical response of the sensor of Figure 5.14(a).

5.3. Stress sensors

Magnetostriction is a rather undesirable property in field and position sensors, with
the exception of the MDL technique where it can be treated in such a way that it

98
exhibits zero hysteresis, thus allowing could levels of uncertainty. On the contrary, in
the case of stress sensors, it is the magnetostriction effect that allows magnetic
materials to play an important role in the global market. This is due to the fact that a
magnetostrictive material exhibits a large variation in the magneto-mechanical
coupling factor under stress, which can be used for much higher output with respect to
the rather classical conducting strain gages.
Because the approved universal standard in mass measurements is still an artifact,
maintained at the International Bureau of Standards, there are no standards in all kinds
of stress sensors. The only parameter of importance is the uncertainty of stress sensing
elements, expressed in ppms. It is worth mentioning that the currently under
development standard of the Ampère and Kilogram realization, based on universal
constant experiment is using magnetic effects and materials, namely the Quantum
Hall effect and the Lorentz/Lenz laws. Therefore, the official acceptance of this
experimental will allow a new boost in the development and categorization of these
sensing elements.
Stress sensors are divided into three main categories: the load cells, the pressure
sensors and the torque meters, having the flow meters and mass flow meters as a
derivative sensing application. All these devices can be based on position or strain
sensors detecting indirectly the applied stress. Fig. 5.15 illustrates the most significant
types of stress sensors based on magnetic effects.

5.3.1. Magnetoelastic stress sensors and related devices

Stress sensors measuring directly tensile stress are mainly based on inductive
arrangements using as ferromagnetic core a material sensitive to tensile stress [234,
198-200], as shown in Fig. 5.15 (a). Such a core is usually a positive magnetostrictive
material. The permeability decreases dramatically with stress, so that the output of the
coil decreases correspondingly. The sensitivity of these devices is much larger than
the sensitivity of strain gages even without any signal conditioning attempt. In case of
negative magnetostrictive materials the permeability tends to increase with a lower
ratio compared to positive magnetostrictive materials, but it offers a more repeatable
response after packaging, i.e. coating of the magnetostrictive element with an inactive
element. Typical values of sensitivity and uncertainty of these devices are 10-100
ppm, with a cost of 1 kEuro/sensor, including the electronics and the communication
protocols.
Using the very same principle, load cells and torque meters have also been developed.
Load cells are translating the load applied on a beam to tensile or compressive stress
to the magnetoelastic sensing element, while torque sensors utilize the Shevron type
design [235]. According to this set-up, torque on a shaft results in tensile and
compressive stresses on vertically arranged magnetoelastic strips, thus allowing
torque and torque direction determination.
Recently, other types of torque sensors have also been used, taking advantage of the
dependence of magnetization of specially annealed plane or circumferential films or
even as-cast ferromagnetic substances on the applied torsion on them. Garshelis, has
developed a family of torque sensors based on the inductive and cordless response of
a sample under torsion [236]. According to the main principle of operation of this
device, two magnetic field sensors are connected in series opposition, thus detecting
the gradient of magnetization change due to the torsion applied on a ferromagnetic-
magnetoelastic shaft or a non-ferromagnetic shaft covered with a ferromagnetic-
magnetoelastic film. Sensitivity and repeatability of this device seems to be
remarkably good, allowing the use of this torque sensor in automotive industry. Apart

99
from that, trials to develop very sensitive magnetoelastic films on top of cylindrical
shafts [237] have resulted in an increase of sensitivity to 50-100 ppms, also leading to
remarkable repeatability. Cost of these devices is of the same order like load cells,
including electronic signal conditioning and communication protocols.

F F
1 2

F F

1 2 3
(b)

MDL set-up

1 2 3

4 5 MI set-up

(c)

Figure 5.15 Stress sensors based on magnetic effects.

(a) Inductive sensors. (1) Soft magnetic element, (2) Inductive excitation coil.
(b) MDL sensors. (1) Movable excitation coil, (2) MDL, (3) Movable search
coil.
(c) MDL and MI torque sensors. (1) Movable excitation coil, (2) MDL, (3)
Movable search coil, (4) Sinusoidal excitation circuit, (5) MI element.

Flow sensors based on electromagnetic techniques are well known in industry.

Recently, flow meters have also been proposed, using the effect of bending stress on

100
an amorphous wire or the displacement due to flow of amorphous wires and glass
covered wires [238].

5.3.2. Magnetostrictive delay lines in stress sensing applications

The inherent property of magnetoelastic materials to change their magnetization with
stress induced anisotropy resulted in the development of many stress devices. The
four most important families of such sensors, namely load cells, torque meters,
pressure gages and tensile stress sensors as well as a miscellaneous application for
thin film thickness sensors, are presented. The sign of magnetostriction in this group
of sensors determines the sensor response. The indicative response of all sensors
presented in this chapter refers to positive magnetostrictive elements.

Load Cells
Load cells based on the MDL principle have been developed in the past [239]. The
basic operation of such a sensor is illustrated in Fig. 5.16 (a). In this case, the MDL
itself is subject to load. Stressing the MDL results in a distortion of the λ(H) function
and consequently in a decrease of the pulsed voltage output. A typical response is
illustrated in Fig. 5.16 (b), demonstrating a monotonic and very sensitive response.
Prestressing the MDL before setting it in position, allows the measurement of
compressive and tensile loads on the load cell. Details on the description and
evaluation of the sensor can be found in [230]. Absence of hysteresis suggests good
levels of uncertainty, which promises that this type of load cells can be marketable.

MDL

Ie
Vo

(a)

80
70
60 (b)
Vo (mVolts)

50
40
30
20
10
0
0 0,2 0,4 0,6 0,8 1

Applied force (N)

Figure 5.16 Load cells based on MDLs. (a) The principle of the sensor. (b) Typical response
of the sensor.

101
Torque meters
Torque meters have also been developed based on the MDL principle, as illustrated in
Fig. 5.17 (a). In this case, torsion is applied on the MDL, resulting in a change of the
λ(H) function and consequently in a change of the pulsed voltage output. Proper
tailoring of the magnetostrictive material can result in monotonic decrease of this
function, allowing monotonic response of the sensor. A typical response is illustrated
in Fig. 5.17 (b), demonstrating a sensitive and monotonic response, while absence of
hysteresis was also observed within the limits of our experimental set-up. Details on
the description and evaluation of the sensor can be found in [239].

MDL

Ie
Vo

(a)

50

40
Voltage output (mV)

30

20 Torsion

Force (b)
10

0
0 0,4 0,8 1,2 1,6
Applied torsion (turns/m) compared with applied force (N)

Figure 5.17 Torque meters based on MDLs. (a) The principle of the sensor. (b) Typical
response of the sensor.

Pressure gages
Pressure gages have been developed based on the distortion of the propagating elastic
waveform, when pressure is applied on a point of the delay line, so that the detected
peak voltage Vo, determines the magnitude of the applied pressure. The principal idea
of such a pressure/force element is shown in Fig. 5.18 (a), where the pressure is
applied between excitation and detection points. The experimental set-ups to obtain

102
the response of these sensors can be found in [240]. A typical dependence of Vo on
the applied pressure is illustrated in Fig. 5.18(b). For the given arrangements, the
smallest detectable pressure was 10 Pa. The experiment was carried out by increasing
and decreasing the magnitude of the applied pressure. The resulting output was the
same for both ways of getting the response of the sensor, showing that the hysteresis
of the sensor is small (undetectable under oscilloscope observations).

Vo
MDL

Ie
(a)

80
70
60 (b)
Vo (mVolts)

50
40
30
20
10
0
0 20 40 60
Applied pressure (kPa)

Figure 5.18 Pressure gages based on MDLs. (a) The principle of the sensor. (b) Typical
response of the sensor.

Thin film miniaturized arrangements based on the MDL technique have also been
developed, able to be used in pressure gage applications. In order to obtain a high-
performance miniaturized magnetostrictive delay line arrangement, Fe70B20Si6C4
amorphous thin films are used because of their remarkable magnetoelastic
characteristics. A special technological solution to improve the performance and to
simplify the fabrication process of the miniaturized magnetostrictive delay line
(MDL) obtained by thin film technology has also been proposed. Details of thin film
fabrication can be found in [241].
The schematic arrangement of the miniaturized magnetostrictive delay line fabricated
on a silicon wafer using a multilayer-like structure X/Y/X, where X is a multilayer
structure of SiO2/Cu/SiO2 type and Y is the magnetostrictive material used as delay
medium, is presented in Fig. 5.19. The miniaturized MDL arrangement using
Fe70B20Si6C4 amorphous thin films 9500 Å in thickness as sensing element is produced
on silicon substrates using electron gun or sputtering process. The deposited
amorphous alloy, after mask formation in the shape of a 0. 5 × 33 mm narrow
element, is the delay line medium. The excitation element is a copper or aluminum

103
straight line orthogonal to the amorphous alloy, set at one end of it separated by a
silicon dioxide layer placed under the whole deposited amorphous alloy.

Figure 5.19 The miniaturized arrangement of MDL fabricated by thin film technology: (a) top
view of the complete packaged thin film magnetostrictive delay line; (b) cross
sectional view A - A: 1, thin film wafer; 2, insulating layer; 3, first layer of Cu; 4,
insulating layer; 5, Fe70B20Si6C4 amorphous thin films operating as MDL medium;
6, insulating layer; 7, last layer of Cu; (c) cross sectional view B - B: 8, pads of
exciting conductor and receiving coil.

Figure 5.20 The miniaturized arrangement of MDL fabricated using a multilayer-like

structure: (a) top view of the complete multilayer-like structure; (b) cross
sectional view A - A: 1 - thin film wafer; 2 - insulating layer; 3 - Fe70B15Si15
amorphous thin film; 4 - Cu thin film; 5 - insulating layer; 6 - three-layered MR
sensing structure; 7 - Fe70B15Si15 amorphous thin film operating as MDL medium.

Additionally, the receiving element can be realized arranging two arrays of

conductors situated above and below Fe70B20Si6C4 alloy, in order to form a receiving

104
coil surrounding it. The insulating layers and amorphous thin films are deposited by
the r.f. sputtering method. The lithography process used to form the exciting and
receiving coils is based on the ordinary positive photoresist application, exposure by
ultraviolet contact photolithography and development. The pattern is defined by wet
chemical etching. An alternative set-up is illustrated in Fig. 5.20 [242]. According to
this, a Ni81Fe19/SiO2/Ni81Fe19 three-layered MR structure is arranged bellow the
Fe70Si15B15 thin film, for flux detection. These thin film arrangements are promising
sensing elements for various magnetomechanical microsensors in terms of sensitivity.

Tensile stress sensors

This group of sensors is based on the elastomagnetic properties of the
magnetostrictive metallic glasses [243]. In these alloys, made by the rapid
solidification process, their relative permeability is dependent on the tensile stress,
which is applied on their surface. A similar effect is also observed for compressing the
material in the same direction. This property is used as the fundamental idea behind
all the proposed sensors at this group.
In this family of sensors, the MDL has to remain free of stresses, under any
circumstances. The first type is shown in Fig. 5.21 (a). The active core S, is made of
metallic glass and is placed at a fixed distance between the pulsed current conductor
and the line. Transmitting pulsed current through the pulsed current conductor, results
in an amount of pulsed magnetic field in the MDL, while a certain amount of
magnetic flux is trapped into the active core S, due to its high magnetic permeability.
Applying tensile stress on the core S, the relative permeability decreases and more
flux is allowed to enter the MDL. So, the elastic pulse and the MDL pulsed voltage
output are larger. Fig. 5.21 (b) shows another type of sensor. In this case, the active
core S operates in a slightly different way. In the presence of it, the magnetic circuit
and consequently the magnetic flux at the sensing point changes because the core S
attracts some of the magnetic lines, although the MDL operates now as a magnetic
screen. So, coupling between core S and MDL results in an increase of the magnetic
flux in the delay line with the increase of the tensile stress along the active core,
causing thus another type of tensile stress sensor.
Fig. 5.21 (c) shows a balanced structure as in Fig. 5.9 (c) and is a combination of the
sensors of Figs. 5.21 (a) and (b). As pulsed current Ie is transmitted in the same
direction in the two conductors, then, in the absence of the core S, there is no
magnetic flux in the delay line and consequently zero pulsed voltage output is
detected. When core S is positioned unstressed and close to the MDL, the pulsed
voltage output is maximized. Applying stress on the core S, results in decreasing the
relative permeability down to one and therefore the signal decays to zero, making the
core S magnetically transparent. Sensor of Fig. 5.21 (d) also uses the balance structure
set-up. Two cores S1 and S2 above and below the delay line respectively are fixed in
equal distances from the MDL. Applying load on such a beam structure, the one core
(S1) is stressed and the other (S2) is compressed. Therefore, a large amount of
magnetic flux is added to the MDL due to the tensile stress of the core S1 and a
relatively small amount of flux in MDL is subtracted due to the compressive stress of
the core S2. One can find an analytical explanation of the operation of these sensors in
[243].
A typical response of the sensors of Figs. 5.21 (a), (c) and (d) is illustrated in Figs.
5.22 (a), (b) and (c) respectively. All these results are taken by increasing and
decreasing the applied tensile stress and the reported output is the same for both cases,
so that the resulting hysteresis appears to be small.

105
2 1

3 4

(a)

2 4

(b)

3 4

(c)

3 1

2 1 4

(d)

Figure 5.21 The group of tensile stress sensors. (1) Magnetic core S, (2) Pulsed current
conductor, (3) Delay line, (4) Search coil.

The sensors of Figs. 5.21 (a) and (b) could be regarded as similar to the simple strain
gage, while the rest of the sensors as bridge structures. It is observed that for the
function shown in Fig. 5.22 (a), the gain of full-to-zero applied tensile stress ranges
from 2.5 to 1. This fact makes this type of sensor more attractive than strain gages in

106
terms of sensitivity (strain gage gain about 0.1 - 0.3 %). The response of the sensors
of Figs. 5.22 (c) and (d) exhibits very good sensitivity, thus allowing various kinds of
applications.

100

80

Vo (mVolts)
60

40

20
(a)
0
0 50 100 150
Tensile stress (M Pa)

70
60
50
Vo (mVolts)

40
30
(b)
20
10
0
0 50 100 150
Tensile stress (M Pa)

60
50

40
Vo (mVolts)

30
20

10 (c)
0
0 50 100 150
Tensile stress (M Pa)

Figure 5.22 Typical responses of the sensors of Figure 5.21.

This type of sensor is ideal for civil and mining engineering applications, where the
sensor core –preferably in the form of ribbon - can be connected on the under
inspection element, in order to perform contactless stress monitoring. Taking into
account that such a sensing core does not require any surrounding coil, it can be glued
on the edge of the concrete or the metallic structure under test. Approaching an MDL

107
intersection of pulsed current conductors and magnetostrictive element close to the
sensing core, the pulsed voltage output is monotonically dependent on the applied
stress on the sensor core.

Thin film thickness sensors

Measurements of thin film thickness during manufacturing are important for many
reasons and applications. The state of the art concerning in situ thin film thickness
sensors is the oscillating quartz technology. The frequency of an oscillating quartz is
decreased as the thin film thickness grows.
The MDL technique can be used to develop a thin film thickness sensor with a more
sensitive dependence in the time domain and a small dependence on temperature
[244]. The pressure or force applied on an MDL during the thin film deposition can
result in attenuation of the propagating elastic signal, so that the detected signal output
decreases with respect to the applied pressure. In particular, deposition of atoms on the
magnetostrictive surface results in applying static force and consequently static
distortion on the propagating elastic wave. Such distortion results in a decrease of the
MDL pulsed voltage output. The amount of atoms and force is proportional to the
thickness layer deposited on the magnetostrictive element. Therefore, such an
arrangement can result in a thin film thickness sensor, detecting film thickness during
deposition.
In order to make such sensing principle industrially applicable in terms or durability
and reproducibility, the MDL arrangement, illustrated in Fig. 5.23 (a), may be used. In
this set-up two rectangular magnetostrictive elements (1) are set at the two ends of a
glass substrate (2), which has two long parallel cuts, thus being able to act as long
acoustic waveguide with rectangular profile. Two pairs of copper ribbons (3) are
connected by silver paint (4) aside the magnetostrictive elements. Supplying pulsed
voltage at the ends of the one pair of copper ribbons and through the magnetostrictive
material, results in pulsed field and consequent elastic pulse generation. This elastic
pulse is coupled with the glass substrate, propagates through it and is coupled again
with the second magnetostrictive element, inducing a pulsed voltage output due to the
inverse magnetostriction effect. When the procedure of the thin film development
starts, atoms of the target hit the deposition substrate as well as the glass substrate of
the sensor, while beating the magnetostrictive elements is prohibited by using a metal
mask. Details of the description and operation of this arrangement as well as
experimental details can be found in [244].
The dependence of the amplitude of the sensor voltage output on the exciting pulsed
current input, and on the thickness of the deposited film is given in Fig. 5.23 (b) and
(c) respectively. The “near true value” of the thickness is the response of the quartz
oscillator. These results indicate a good agreement with the prediction of the
exponential response. But, differences in the response of different deposited materials
may be observed, which necessitates the use of look-up tables when accurate
measurements are required. Experimental results indicate a different bandwidth
response concerning field and stress effect on the whole arrangement. Using a fast
Fourier transform (FFT) procedure, the separation of the two measurands could be
obtained, thus allowing the manipulation of the undesirable effect of ambient fields at
the vicinity of the sensor. Considering the ambient field shielding process, the use of a
permanent magnet ribbon below the magnetostrictive element could strongly bias the
MDL set-up and therefore minimize the effect of ambient field interference.

108
 3 1 2 5
4

6
7
8

(a)

10 10
Output Voltage (mV)

8 8
Output (mV)

6 6

4 4

2 2

0 0
10 30 50 70 90 110 130 0 2 4 6 8 10 12 14 16 18 20
Pulse d curre nt (mA) Thickne ss (X100nm)

(b) (c)

Figure 5.23 (a) Thickness sensor arrangement. (1) Rectangular magnetostrictive element, (2)
Glass substrate, (3) Copper ribbon, (4) Silver paint connection, (5) Sensor
support, (6) Pulsed voltage generator, (7) Pulsed voltage detector, (8) Control PC.
(b) Output voltage dependence on pulsed current.
(c) Output voltage dependence on thin film thickness.

5.3.3. Dynamic stress sensors

One type of dynamic stress sensors can be based on displacement sensors, which
additionally employ spring means for translation of stress to displacement. Usually,
low compliance properties of the dynamic stress sensor prohibit the use of this type of
sensors.
Classic stress sensors can also be used for dynamic stress monitoring. This is realized
by continuous measurement of stress and then calculating the dynamic stress by using
the signal conditioning and microprocessor based control circuitry and software.
The most advanced type of dynamic stress sensors is the one employing no excitation
means. This way, the variation in magnetization due to the magnetostriction or the
inverse magnetostriction effect can be detected inductively, conductively or magneto-
optically.

109
5.4. Security sensors

Security sensors are based on the combination of different magnetic properties, for
example different B-H loops of materials, in order to generate a code based on a series
of magnetic signatures, which allows the recognition of an object without direct
optical observation. This family of sensors can be used in applications where optical
bar coding is impossible for practical reasons.
Various magnetic techniques have been employed for magnetic security systems or
magnetic signature systems [245]. The most well known technique is the one
combining the properties of very soft magnetic materials in the form of rapidly
quenched ribbons and the eddy current principle. A coil or a pair of coils connected in
series opposition is used as excitation means, while search coil or pair of coils is used
for signal detection. The coil is supplied with ac current which results in ac field
vertical to the coil level. The search coil is set in such a way that it detects the
response signal corresponding to the area where ac magnetic field is transmitted. In
the absence of any magnetic material in front of the excitation coil or between the two
excitation coils, the voltage output of the search coil is below a certain threshold (or
zero in the case of the pair coils). If a magnetic material used after a special tailoring
treatment process is between the coil or the pair of coils, a pulsed voltage is induced
in the search coil. This system has been widely used in security systems in
commercial shops. A large panel hosting the excitation and search means or a pair of
large panels hosting the pair of excitation and search means is usually set at the exit of
a shop. All under inspection products are equipped with a magnetic label. This label is
usually a zero magnetostriction soft magnetic ribbon, used after heat and magnetic
annealing treatment, thus having a very narrow and quite high B-H loop. If this label
is still attached on the magnetic product at the vicinity of the shop exit, a sharp pulsed
voltage is induced in the search coil(s), thus informing that a product is most probably
taken out of the shop without check-out.
Another type of security sensors is the magnetic tapes for cards or tickets etc. A
special magnetic material, most probably a semi-hard magnetic material in the form of
particulate medium, is deposited on the plastic card or on the plastic coated paper
surface. Then, a magnetic code is written on this surface: magnetic heads transmit or
not magnetic field on top of the recording magnetic surface, thus generating a
sequence or magnetic “zeros” and “ones”. Then, a recording head, like a GMR or
AMR or STD or magneto-optic sensor can read the information stored on this surface.
In fact, due to the parametric dependence of this writing and read process on some
parameters like temperature, ambient fields or so, the recently developed spin
tunneling devices or hard films have started to be used. Writing on such a
magnetically hard surface is realized by “hot” writing, i.e. heating the surface
approaching the Curie point of the hard material, “write” the zero or the one with a
relatively low field and then let it cool down.
Research and development efforts have also been directed towards the development of
magnetic bar codes, which ought to be used in environments where optic codes cannot
last for long or cannot be read. For this reason, two types of magnetic bars have been
developed so far. The (historically) first is the use of soft and zero magnetostrictive
amorphous wires of different lengths, which some times also have different
compositions, used after magnetic annealing. Thus, the B-H loop of each wire differs
from each other. The wires are pre-magnetized at a given remanence, in the positive

110
or negative direction. The transmission of an opposing ac field may saturate some, but
not all of them. Then, a reading head (like AMR, GMR etc sensor) reads a sequence
of positive and negative signals corresponding to “ones” and “zeros” respectively.
Thus, the identity of the marked product can be determined. A rather more
sophisticated technique utilizes either a recording medium like in the case of magnetic
cards or tickets or a spin valve or GMR array. Proper polarization of this medium can
be read as “one” or “zero” resulting again in recognition. This technique can result in
a mass production of magnetic bars in comparison to the wire write and read process,
thus allowing cheaper magnetic bar code systems.
Finally, another interesting security system is the magnetic vehicle signature. Up to
this moment this technique is widely used in military applications especially in ship
recognition, but efforts are currently being made in order to use a kind of such a
system in domestic applications like automotive and railway applications. According
to this technique, a large excitation coil is set at the bottom of a ship surrounding it.
Provided that the ship has a metallic (soft steel) backbone, then transmitting ac current
in the excitation coil, the ferromagnetic “backbone” is magnetized. Transmitting a
special code through the exciting coil, a field sensor out of the ship can read the
corresponding magnetization process of the ferromagnetic backbone of the ship which
is her magnetic signature. Therefore, having set the magnetic field sensor at the areas
where the control of the passing ships is desired, the sensor can read the signal of the
ship. Using a particular code of sequences, the magnetic signature of friendly ships
can be recognized. There are two problems of such a technique. The first is that in old
or large ships, such magnetic signature ought to be large in magnitude, thus been
detectable from other very sensitive sensors. The second problem is that metallic ships
utilizing the magnetic signature procedure ought to be demagnetized from time to
time in the so called demagnetizing stations. At these stations they are demagnetized
by cyclic and descending magnetic loops (see Fig. 3.12). In order to avoid both
problems as well as to use magnetic signature in non-ferromagnetic vehicles, other
magnetic signature techniques can be developed, utilizing a magnetic bar code
system. Trials on this idea have been initiated towards traffic control and car
recognition.
It is worth mentioning that the field of security sensor systems is still quite open in
terms of being able to absorb new ideas and principles, allowing less expensive and
more accurate systems.

5.5. Smart sensors

Another important family of sensors also exists in all kinds of electronic materials,
employed for sensing applications including magnetic materials and effects. It
includes three different subcategories, the multi-parameter sensors, the self-learning
sensors and the sensors in self-decision making and reacting systems.

Multi-parameter sensors
The multi-parameter sensors are able to detect more than one physical size. They can
employ one or more sensing elements. In many applications, accurate multi-parameter
sensors have low uncertainty if the same core is used for all under measurement
parameters. An example is a magneto-elastic arrangement, based on negative
magnetostrictive ribbons, which is able to detect field, stress and temperature at the
same time [246]. The sensor is based on three different magnetic effects or operating

111
modes, using the same sensor topology, which consists of a magnetic wire as sensing
core, two coils as excitation or search means and two electric contacts at the ends of
the magnetic wire. The involved magnetic effects are magnetostriction, magneto-
impedance and re-entrant flux reversal. Operating the sensor in these three different
modes separately and sequentially, one can obtain the response of the sensor related to
the three different physical quantities, namely stress, temperature and field. It has
been experimentally observed that the total output of the sensor in each one of the
three different modes is equal to the product of each corresponding physical quantity
function concerned. Therefore, the three unknown parameters of stress, temperature
and field can be determined from a 3X3 matrix equation. Therefore, having specified
the formalism of the above mentioned unknown parameters of stress, temperature and
field, one can determine their magnitude using numerical analysis techniques.
Experimental results also indicate that other effects, like the Wiedeman effect, can
also be employed to measure more physical quantities. This occurs because the
amplitude of the pulsed output of the sensor is dependent only on the ambient field,
while the width of the pulsed output is modulated only by the tensile stress.

Self-learning sensors
The self-learning sensors are based on their ability of self-calibration and the auto-
scaling of the range of measurement. Generally, microprocessor based techniques are
used in such sensors, usually employing a look-up table in the memory of the system.
An example for such a sensor family is an auto-scaling field sensor based on the
magnetostrictive delay line (MDL) technique. The principle of operation is based on
the variation of the magneto-elastic response of an array of MDLs with respect to the
applied input, which depends on the composition and the annealing history of the
material. The motivation for such a sensor is the combination of good sensitivity of
the order of 100 pT in a range of measurement up to 0.1 T. This was realized by using
three Fe78Si7B15 magnetostrictive wires with different annealing [247]. All of them
were annealed under magnetic field, resulting in magneto-elastic saturation fields of
10, 100 and 1000 A/m respectively, named as W1, W2 and W3 wires respectively.

120

100
MDL output (mVolts)

80

60

40

20

0
0 0,2 0,4 0,6 0,8 1 1,2
Applied field (mT)

Figure 5.24 Illustrative response of the MDL field sensor using the wire W1.

Their response under ambient dc field was monotonic. A representative response of

one of the wires is illustrated in Fig. 5.24. The sensitivity levels of the three different

112
wires W1, W2 and W3 are illustrated in Table 5.1. Such response suggests that
simultaneous use of these three wires can detect with remarkable accuracy magnetic
fields. One way to use this result is to arrange them parallel to each other in the
classical MDL arrangement, using one excitation and one search coil per delay line. A
microprocessor based circuit controls the excitation current and the readouts of the
search coils. The arrangement is subject to the ambient field. The electronic circuit
selects the largest response in amplitude corresponding to the closest range of
measurement and consequently determines the applied biasing field along the wire.

Table 5.1 Levels of sensitivity of the three different wires.

Sensitivity Uncertainty Range
Wire W1 100 pT 100 ppm 1 mT
Wire W2 1 nT 80 ppm 10 mT
Wire W3 10 nT 150 ppm 100 mT

Such an idea can also be used in mechanical sensors like position sensors. Using wires
under different annealing conditions one can realize position sensors, which can
detect different ranges of measurement. We are in the process of applying such an
idea in position and stress sensors.
It is straightforward that a number of self-learning sensors can be developed based on
various magnetic effects and materials, provided that the sensing cores are properly
tailored, so that the range of measurement corresponds to the desired one.

Sensors in reacting systems

The reacting sensors are complete electromechanical systems, which are able to sense
and consequently react with respect to the measurement. An example is a missile
driving sensor, including a precise field navigation sensor that measures the direction
of motion of the missile and a reacting system that changes the direction of the missile
with respect to a pre-loaded order. The sensing part of this system ought to be
accurate within the desired limitations of any given application.

113
6. Sensor applications

The above-described sensing elements are used in a vast variety of applications. We

are briefly introducing the main categories of such applications with respect to the
used sensors. Applications appearing in a given sector and being also applicable in
other(s) will not be repeated.

6.1. Industrial applications

One of the main fields in industrial applications is the non-destructive testing and
evaluation (NDT&E) [248]. There are two distinct cases in magnetic based NDT&E.
The first is the NDT&E using magnetic effects and materials. The second is the
testing of ferromagnetic substances as such. When magnetic materials are used as
sensing elements they are mainly designed in conjunction with the eddy current
technique (ECT) or magnetic anomaly detection (MAD) testing. Both sensing
techniques require the use of small in size field sensors. Thus, the sensitive fluxgate
sensors cannot be used. The most commonly used sensors for this purpose are the Hall
sensors, offering integrated electronics, but, a rather poor sensitivity of the order of
0.1 mT. The giant magneto-resistive (GMR) elements, offer an improved sensitivity
of the order of 1 μT. Furthermore, the recently developed MI elements, offer a
sensitivity of 10 –100 pT. The value of sensitivity reflects the sensor's ability to detect
small field variations, directly related to the size of existing crack or defect. MDL
sensing elements have also been proposed, being able to perform distribution
measurements, despite their current disadvantage of poor sensitivity. Testing
ferromagnetic materials the above mentioned techniques can also be tested, but other
techniques can also be implemented [249]. According to this, the magnetization
process of the ferromagnetic substance, the surface of the substance or part of the
substance is indicative of the status of the device. This is an alternative and indirect
way of detecting not only cracks and defects but also examining the structure and
microstructure of the material. Hence, the measurements of the B-H loop of the
substance, the B-H loop along the substance and the B-H loop of its surface are good
indicators of the structure of the material. As an example, any variation in coercivity
and remanent magnetization is a clear signal of non-uniformly distributed stresses
along the material. Temporal dependence of remanence indicates a change in the
structure of the material. For example, decrease of remanence below a given threshold
indicates the approaching of the life-time of ferrous rods used in mechanical
structures [250]. The cost of such sensing systems is in the order of a few thousand
Euros.
The second more frequent field of industrial applications is the position, velocity and
acceleration/vibration controllers. The vast majority of these sensors are position
switches based on the GMR effect, with a repeatability better than 1012 and cost 10 up
to 100 Euro/sensor. Other kinds of position sensors are also used, depending on the
required sensitivity. For good sensitivity of the order of 1 μm, differential position
sensors are used based on the permanent magnet tape arrangement, with a cost of ~1
kEuro/sensor. As an example, when sensitivity levels in the order of 10 μm are
required, the MDL technique is employed, with a cost 30 Euro/sensor. Recently, a
new MDL arrangement has been proposed [251], according to which the sensitivity is
of the order of 1 μm, offering the additional advantage of performing static position
measurements with a cost of 100 Euro/sensor.
Another important field of industrial applications is the mass sensors, like load cells,
torque meters, pressure gauges and their derivative sensors. Up to this moment the

114
vast majority of such industrial sensors are based on conducting or semiconducting
materials. Strain gages are the most commonly used sensor for load measurement.
Today’s load sensors are based on miniaturized elements fabricated by lithography
techniques, thus allowing a drastic reduction of their cost. But, recently inductive
techniques employing sensors based on magnetoelastic effects or mangetotransport
techniques employing sensors based on the MI-GMI-AMI and MR-AMR-GMR
effects using magnetoelastic materials have been used for some interesting
applications of load and derivative size measurements, with sensitivity better than the
strain gage arrangement, as mentioned in the previous chapter. Their problem at the
moment is their short life-time, which is of the order of 108 repetitions in comparison
to strain gages which is ~1012. In this family of applications, flowmeters either in the
form of magnetic or electromagnetic devices occupy a significant part at this niche,
but significant market.
Finally, security sensors like magnetic bar codes or labels, is another significant
application in production line logistics. All forms of the security sensors described in
chapter 5 may be used for this case.

6.2. Biomedical applications

Another significant application of sensors based on magnetic materials and effects is
the biomedical applications [252]. It is hereby noted that classical testing techniques
like the MRI or NMR techniques they refer to systems rather than to sensors, so they
are outside the scope of this presentation.
The most traditional sensing system based on magnetic effects is the encephalograph.
It consists of field sensor arrays able to detect fields in the range of 0.1 – 1 nT. The
most widely used sensor array is the SQUIDD system, comprising 64 or 128 SQUID
sensing elements, with a sensitivity of 1 – 10 fT at a cost of 1 MEuro for the whole
sensor array. Recently MI and GMI field sensors have started being used in such
applications, having the disadvantage of lower sensitivity, in the order of 10 – 100 pT,
but at a cost of approximately10 kEuro per sensor array.
Apart from that, another family of biomedical sensors based on magnetic effects is a
new type of cardiograph, which is simpler in operation and less expensive than the
classic electrocardiographs, although it does not perform all measurements obtained
with the latter. It requires only two MI sensors set at the two hands of the patient and
costs ~1 kEuro/sensor.
Very recently and due to the evolution in DNA evaluation, some micromachined field
array sensors have been developed [253], having spatial resolution in the order of 1
mm and a cost of approximately 100 Euro/sensor. These devices are mainly based on
the GMR effect, while MI sensors are currently tested for the same application. The
operation of such an arrangement is based on the magnetic separation of magnetic
targets on which the under measurement – counting substance has been attached using
hard magnetic materials and their corresponding counting by using the above
mentioned field sensors. Apart from these sensors another interesting sensing system
appeared referring to blood coagulation. Almost simultaneously, two groups [254,
255] developed two systems using two different magnetoelastic concepts to measure
the time of blood coagulation. The first team, initiated activity in blood coagulation
using resonance frequency techniques. According to this principle, the sensor is able
to detect within a few minutes the time of blood coagulation, the principle of the
method being the induced surface stresses due to the blood coagulation. The second
team developed another sensor based on the MDL technique. The principle of
measurement is based on the distortion of the propagated elastic waves due to the

115
blood coagulation. According to this technique similar time response has been
observed. Further continuation of this work resulted in repeatable results obtain blood
coagulation measurement within 30 seconds, by means of predicting the final
exponential curve of the response. These sensors are still at research level and are
currently tested in hospitals under medical protocols for the statistic evaluation of the
clinical results. This sensor is particularly useful in surgical operations, where the
blood coagulation time is of critical importance. Therefore, the sensor market may
allow a forthcoming selling price in the order of 20-30 Euros/sensor.
Another interesting bio-magnetic application is the detection of stresses in the human
body by using special heterostructures of magnetoelastic elements [256]. According to
this technique, special arrangements of either mechanically connected positive and
negative magnetoelastic ribbons separated by a non-magnetic interlayer, or film
deposited layers of the same configuration, are able to accurately and repeatably
detect stresses induced by the human activity, like lug motion, inspiration activity,
apnea or erection during sleep, mechanical vibrations in the backbone etc. These
sensors are also at research level, performing clinical tests in hospitals under medical
protocol. Provided that they will finally appear in the market, their cost prediction is
estimated around 100 Euros/sensor.
Magnetic targeting is also a part of magnetic effects in bio-sensing applications [257].
A magnetic substance, usually in the form of a very small sphere or powder is hosting
either a drug or a driving-searching fluid, or is covered by a bio-compatible film of a
desired composition or both. Being introduced in the body and driven contactlessly by
a technical field outside the body, they can sense the presence of bodies compatible
with the covering film or the hosting fluid if existing. Then, the concentration of
magnetic powders can tell the concentration of the under investigation substance. The
same procedure has also been proposed by using a microscopic magnetic screw,
which is driven by an external technical field inside the blood arteries. The external
field tries to move the microscopic screw with a constant speed, with a simultaneous
rotation of the screw. Thus, any difficulty in motion corresponds to a possible
presence of undesirable artery closures, which may cause heart attack. Such a
microscopic screw not only senses and informs about the existence of a closure on its
path, but also acts as a small contactless laparoscopic drill able to make small wholes
or enlarge the stenosis in arteries.

6.3. Military applications

Although not very popular, the military applications of sensors including sensors
based on magnetic materials, covers a large sector of the global sensor market, in
terms of budget and importance.
One significant application is the anti-mining control system. Up to now, the low field
or field gradient or magnetic anomaly detection systems are the only existing devices
able to detect mines. The more sophisticated the mines become the less iron they use
and therefore the more sensitive and accurate field sensors are required. Many types
of sensors have been used for this purpose, from fluxgates to MI-GMI and MR-AMR-
GMR sensors. The main problem in such a measurement is the ability to distinguish
the signal from a magnetic object other than the mine, i.e. the static and moving stray
fields, from the magnetic substance existing in the mine. This is a difficult task,
always allowing new ideas and sensing systems. One of the solutions to this problem
is the gradient measurement of field which usually removes the noise from stray
fields, allowing only the possibility of false alarm due to a magnetic element buried in
the earth. Another, additional, possible solution is the use of a look-up table with the

116
magnetic signal of given mines when a sensing system scans their neighborhood with
a given, well determined speed. Such a dynamic displacement of a sensing system
may remove the ambient field/noise by using Fast Fourier Transform or wavelet
techniques. Their sensitivity in today’s sensors is around 1 – 10 nT. It is said that the
budget of the anti-mining control process in the Balkans and the Middle East in the
next decade is going to be in the order of 500 million Euros.
Another military application used also for domestic applications is the magnetic
signature. This is mainly applied to metallic ships. According to this, the military and
other ships are equipped with coils supplied by a given, usually coded, current
waveform. The detection of the field, produced by such a current, results in the
recognition of the given type of vessel, as described in the security sensors of chapter
5, where details of such a system can be found. Applications of this system may also
be realized in other military vehicles like tanks and trucks. Usage of magnetic bar
coding (sometimes they are named magnetic tags) systems may also be useful for
magnetic logistic techniques.
Another application is missile navigation. At this moment, the state of the art of this
application is using gyroscopes based on inertia mass control or the global positioning
systems (GPS). Currently, several research groups in the USA as well as in Japan and
Europe are seeking to employ field sensors and gradient field sensors in such an
activity. The principle of operation is based on the field variation measurement and
corresponding corrective action due to the earth’s field. Fluxgates could be considered
as an ideal sensor for this purpose but their size makes difficult their use in missiles.
Besides, the unavailability of on-board electronics also prohibits their use because of
the slow real time response: for such an application requires not only high sensitivity
but also high speed in the field measurement. Hall effect field sensors could also be
used due to their main advantage of being compatible with CMOS technology, giving
fast results, but their usage is rather undesirable for such an application due to the low
sensitivity and not-acceptable noise level. Magneto-transport sensors like GMR or
GMI devices appear to be desirable for such an application, provided that sensitivity
and time response levels are acceptable. Sensitivity of these devices could be in the
order of 1 nT in production line products, while their time response is a little bit worse
than that of the Hall effect sensors. This is due to the required interface layer between
the CMOS electronics and the magnetic sensing core, in order to avoid mis-operation
problems. It is clear that this field of applications of magnetic sensors is still open for
research and new sensor proposals are welcome for better sensitivity and faster
response.

6.4. Transport and Automotive applications

Automotive applications of sensing elements are more and more significant in our
days probably due to marketing reasons as well as for their significance in the passive
or active safety process. Car or road accidents have unfortunately a high death toll. It
should be noted that in many, especially developing, countries the number of mortal
accidents in the weekends is of the order of 1 every 100,000 citizens. Therefore, the
popularity of and demand for sensors, used in modern cars and transportation means,
is understandable. More than 1,500 different sensors are used in most cars produced
today. Many of them are based on magnetic effects and materials. In the following
lines, some examples of such magnetic sensor applications are given.
Sensors based on magnetic materials govern some of these sensing applications. The
most classical and well-known sensor is probably the angular positioning magnetic
sensor used to activate the ABS systems in the brakes of the cars. Many thousands of

117
human lives have been saved due to this system. A rotating ring with teeth of
permanent magnetic material is rotating with the wheel motion. Any sudden and
unexpected blockage of the wheels during braking energizes the anti-block system
(ABS) thus relieving the braking pressure for some friction of the second. The
durability, reliability and sensitivity of the magnetic sensor in the ABS system are of
the same vital importance as the operation of the signal conditioning circuitry and the
microprocessor-based electronics are. It is worth mentioning that other sensing
systems rather than magnetics have not been applied yet due to their difficulty of
operation in harsh environments.
Another case of applying magnetic effects in vehicles is the torque sensors used in
wheel steering and shaft operation monitoring as well as in other torsion applications
in vehicles. These sensors have been analyzed in chapter 5 and are up to this moment
in laboratory development stage, but it is expected that they will soon be installed in
industrial production. Apart from the development of reliable torque sensors in terms
of being practically insensitive to other ambient parameters, another important
parameter is the signal conditioning circuitry which has to co-operate with the
corresponding micro-processor based system for correct and fast (thus vital) decision
and action, in cases where required.
Very recently, a GMI field sensor has been used in controlling the position of a car
[258]. The sensor is placed at the corners or the front and back middles of the vehicle.
Therefore, the sensor is able to read the position of magnetic tags being prefixed on
the road, where the car/vehicle is expected to pass. Toyota, the Japanese car company,
has successfully employed this field sensor in monitoring the parking process. The
very same sensing principle can be employed in order to develop a relatively new
scientific fiction, which is the driverless car. According to this idea magnetic tags are
periodically set on the road, thus allowing a sensitive field sensor to know the exact
position of the vehicle at any time. Therefore, corresponding actions can be decided
and taken by the microprocessor based system. Such a technique is currently under
development, concerning the automatic control and passage from the national road
tolls. Despite the fact that such positioning sensors take the fun out of driving, it must
be admitted that in certain cases they may be of vital importance to the passengers’
lives as well as to the car condition. Besides all comments, this system may be proven
to be a sufficient system for speed control procedure, by measuring the real ground
speed of the moving vehicle. Such a control may be passive by means of informing
the driver of the speed of the car. Additionally, it may be active by means of
correcting the speed of the vehicle, provided that a co-operation with a GPS system is
used via a microprocessor based circuitry.
Several other force, flow and stress measurements as well as displacement, speed and
acceleration measurements may also be realized using magnetic materials and effects.
As an example, force and pressure sensors based on magnetic materials may be used
in the car tires to measure their pressure and act upon energizing a gas chamber
providing some crucial amount pressure until the car stops. Another example could be
the use of magnetic force and displacement sensors for measuring the forces acting on
the suspension of the car and correspondingly acting by opposing such a pressure or
force for better car handling. Such an idea has been first implemented by Lotus Cars
Ltd in their Elite in early 80's. Another example could also be the use of flowmeters
for the exact detection of fluid in the petrol supply control. A significant sensing idea,
especially applicable in race cars, is the usage of tensile stress sensors to monitor in
real time the stresses applied on the chassis of the vehicle. For example the tensile
stress sensor based on the MDL technique presented in chapter 5, may be used for a

118
contactless stress measurement. Many other examples may also proposed, which can
be implemented and used in car industry applications.

6.5. Environmental applications

In the last decades, the environmental protection becomes more and more vital for the
whole globe. Therefore, the measurement of many parameters affecting the
environmental status is by definition important. Magnetic effects and sensors are
useful mainly for magnetic field monitoring in either dc conditions or in the frequency
domain.
The monitoring of the electromagnetic radiation is a significant part of this process.
The method of detecting such radiation is through field monitoring. The range of
measurement with respect to frequency extends from a few kHz up to 30 GHz fields.
An easily understandable example is the mobile telecommunication electromagnetic
pollution, where the measurement of the radiation in the range of 1-2 GHz determines
the correct control and use of antennas and cellar phones. The sensitivity of all these
measurements ranges from a few pT up to a few mT. These higher frequency
measurements require the use of coils and antennas, rather than the use of dc magnetic
field sensors usable for higher frequencies. These sensing elements ought to be
arranged in arrays to measure the magnetic field at discrete points and then calculate
the field distribution using Maxwell's equations and the law of field continuity. The
electromagnetic compatibility and interference laws suggest that both auto-calibration
procedures and frequent reference calibration procedures are also required. Auto-
calibration may be valid by using magnetic field sources and loads at the field of
measurement, while reference calibration requires calibration of the sensors in
shielded anechoic chambers or in open or semi-open testing areas. Trials for
implementing magnetic field sensors like Hall-effect sensors or other magnetometers
to measure such high frequency fields have resulted in rather disappointing
conclusions. Only magnetotransport sensors seem to give a sensitive enough response
but with poor repeatability. Despite this fact, high frequency magnetic materials like
soft ferrites are used in many cases in such a measuring procedure, especially in the
sensor calibration procedure.
But, since sensors based on magnetic materials govern mainly the low frequency
domain of field measurements, they are the unique method for dc field scanning
processes. This type of measurement is useful for monitoring the earth’s field and its
corresponding changes, which may be the indication of motion of sub-surface levels,
which is the main event before earthquakes and volcano exposure. The most widely
used sensor for this purpose is the proton or gas magnetometer described in chapter 5.
But, the cost of using such magnetometers has prohibited the extensive mapping of
the earth. Recently, fluxgates have been used, at research level, for the monitoring of
earth’s field, while GMI field sensors are currently under test for the same reason. It is
certain that these geophysical studies may be enhanced by the use of sensing systems,
relative inexpensive with respect to proton or gas magnetometers, either fixed on
arrays on the surface of the earth or mapping earth’s field from airplanes. Therefore, it
is believed that such a field of sensing applications is also of economic interest. The
forthcoming results in earthquake prediction and monitoring volcano’s activity are of
vital importance for many reasons, which are out of the scope of this paper.
Another distinct environmental application of dc field sensors is the space field
monitoring. Every satellite is equipped with vector and scalar magnetometers, in order
to perform craft navigation as well as meteorological measurements. The importance
of this measurement becomes evident if one takes into account the conditions of

119
measurement: there are no reference points to be considered as guidelines for field
mapping. These magnetometers are a whole art or science by itself. The geophysical
and electrical engineering research laboratories of Denmark and the Czech Republic
have a long history in the development of these systems [259]. They comprise of a
three co-ordinate field sensors, accompanied by three co-ordinate gradient field
magnetometers as well as by space-monitoring cameras. The most traditional kind of
sensor used in these magnetometers is large toroids or race-track fluxgates equipped
with carefully manufactured and annealed zero magnetostrictive rapidly quenched
ribbons. Due to the labor involved in the manufacturing of these ribbons they are quite
expensive. Recently GMR and GMI field sensors and a combination of fluxgates and
magnetotransport sensors have been employed in order to compete with the
performance, the cost and the operational difficulties rising from the size of fluxgates,
with not promising results up to this moment. Space-monitoring cameras are used to
make pictures of the space in front of the magnetometer. Thus, having had a look-up
table and using imaging processes to provide it with an acceptable reference point of
orientation, the direction of the spacecraft is determined. Accordingly, gradient
magnetometers are used for the navigation of the spacecraft and actual magnetometers
are used for space field mapping. Accurate measurements are important for such
measurements. They are guaranteed by using a three dimensional arrangement of
Helmholtz coils in the form of sphere.
Another environmental application is the counting process in domestic areas.
Counting of vehicles in traffic arteries with corresponding corrective actions in traffic
signaling is an important issue in all large towns. Magnetic field sensors can be used
in order to measure the magnetic signal produced by the steel parts of vehicles. This
may inform the system not only about the number of passing vehicles but also about
their size.

6.6. Domestic and information technology applications

The domestic applications of sensors based on magnetic materials mainly include
navigation sensors, security sensors, as well as reading means of recording media like
CD, DVD hard disks for personal and large computers etc. The recording media and
information storage are a separate engineering sector, different than sensors, and it is
not covered in the current overview.
Field sensors of recording media, as discussed in chapter 5, are field sensors able to
detect “ones” and “zeros” on a magnetic surface. Although they must be robust and
reliable structures with good temporal characteristics in terms of temperature and
humidity, they don’t have to be really precise magnetometers. Of course, one could
say that, an on-off field sensor requires more than a bistable response to correctly
operate in many environmental conditions. Thus, mass production of these sensors
may be allowed provided that mechanical characteristics and parametric properties
control like temperature is sufficiently acceptable. Mass production almost prohibits
the use of devices incorporating coil structures for two reasons: coils require labor;
even if they are made in planar film structure their spatial resolution is not good
enough to allow use in reading a recording medium as well as really mass production.
This is not an issue, for example, in the space field mapping sensors, where sensitivity
and precision is the question and not the vast amount of produced devices. Therefore,
magnetotransport sensors seem to be the ideal solution for this family of sensors.
From the family of magnetotransport sensors, the MI-GMI-AMI types of sensors are
rather excluded due to their need to be post-fabrication attached to the CMOS
electronic circuitry, which costs labor. Therefore, the MR-GMR-STD family of

120
sensors is rather promoted for this purpose. Indeed, these three sensor families, with
the GMR films being the most popular choice, are used for the various reading heads
of domestic recording media. At this point it should be noted that mass production
AMR-GMR-STD sensors are field sensors of rather poor quality, especially in
comparison with purpose annealed and signal conditioned AMR-GMR-STD field
sensors, the main difference being the time and the labor required for post-
manufacturing treatment and calibration. It is evident, that Hall-effect sensors cannot
be used any more in such applications. One reason is the high spatial resolution
required followed by the low sensitivity of these devices: given that in recording
media the spatial resolution is in the order of 250 Gb/in2, reading precisely a threshold
is not easy with Hall sensors. The cost of these low precision and mass production
sensors is in the order of 1-2 Euros per sensor.
The electronic compass is a small field vector sensor, which is the state of the art in
modern navigation aids. Hall-effect field sensors have been initially employed for the
electronic compass development. The realization of the other magnetotransport
sensors allowed the development of more precise electronic compass at the same cost,
which is in the order of 30 Euros offering a few nT sensitivity. This type of compass
is usually offered in cars, good quality watches etc. Lower quality electronic compass
cannot significantly reduce the cost of the device. A figure of merit in low cost
magnetotransport electronic compass is in the order of 20 Euros offering 10 μT
sensitivity. These compasses are more usable in toys rather than in other applications.
In cases where a high sensitivity electronic compass is required, like large vehicle,
boats, even airplanes, fluxgates still have their own market. Their cost, in the order of
100 Euros per sensor, is not high enough to prohibit their purchase in this type of
application.
The security sensor market, as mentioned in the previous sections, attracts significant
attention from research groups. Magnetic security sensors are used when optical bar
coding systems cannot be used due to harsh environmental operation or disability of
optical observation. These sensors consist of a combination of a series of magnetic
substances in a given order, which have distinct characteristic magnetic properties, so
that the non-optical reading of these properties results in a binary or decimal word,
which is the magnetic signature of the subject on which the sensor is attached. But,
the most significant application of security sensors at this moment is the door security
systems at the commercial shops. These systems employ exclusively magnetic
materials. It is estimated that the global market for this kind of sensors is
approximately 10 million systems with a need of more than 103 billion magnetic tags
per year. Although the cost per magnetic tag is not more than 1 cent, it is clear than
the budget is over 10 billion Euros per year.

6.7. Laboratory sensors

In the field of laboratory sensors the science of metrology governs the applications of
sensors and systems based on magnetic materials.
The most widely used application is the field calibration secondary standards based on
accurate field sensors. Fluxgates is the most widely used family of magnetic sensors
for this reason. Such a calibration procedure is based on the precise application of
field using three dimensional Helmholtz coils and precise electronic equipment. But,
the most important problem in implementing calibration of field sensors is the use of
shielded rooms, with proper local control of temperature and humidity in ranges
suggested by the application of the under test sensor. Shielding of these rooms is not a
simple case, since shielding is required not only down to a few tens of Hz but down to

121
dc. Such shielding has to be realized by using soft magnetic materials in orthogonal
arrangement to keep as much of the ambient flux as possible out of the testing facility.
This calibration facility can provide a dc field with a sensitivity of 10-3 fT, making use
of the shielding room and the three dimensional Helmholtz coils which offer
additional compensation to the system. The cost of such a room exceeds 100 kEuros.
Another laboratory application of magnetic sensors and effects is the atomic force
(AFM), magnetic force (MFM) and scanning tunneling (STM) microscopy, according
to which a detailed topography of a flat surface can be determined with atomic
resolution. An interrogating stylus is vibrating on top of the surface being not in touch
with the under test surface generating forces between itself and the above mentioned
surface. Atomic force microscopy (AFM) employs Van der Waals forces and a non
magnetic stylus). If the stylus is magnetic and the under test surface magnetic, the
generated magnetic forces offer additional information with respect to AFM about the
magnetic structure of the surface. This is the magnetic force microscope (MFM). If
the stylus is of a conductive material and the under test surface has electric charge
carriers then electric charges are induced at the stylus offering additional information
about the electrical charge density of the under test surface. This is the scanning
tunneling microscope (STM). The resolution of this microscope depends on the size
of the surface of the stylus approaching the under test surface and is usually in the
order of nm. Recently, a new MFM stylus has been developed having a very narrow
magnetic surface in the order of square submicron. This allowed the realization of
AMF and MFM microscopes offering spatial sensitivity as low as 0.1 nm [260]. The
cost of these microscopes is in the order of 50-100 kEuros.
Laboratory applications of magnetic sensors, also involves the use of high accuracy
field and stress sensors, used for certain experiments. These special sensors can cost
up to 1 kEuro per device due to the specific properties of the used magnetic sensing
core as well as due to the incorporated electronics.
Apart from these sensing instruments, in the metrology related to Current & Mass
primary standard experiments, a new technique based on magnetic materials is
experimentally tested, as already mentioned in chapter 5, aiming to replace the artifact
of the kilogram. Implementing the Lorenz and Ampere’s Laws, according to which
the Ampere and the Kilogram are correlated in one equation, the A and Kg can be
experimentally defined [261]. Such an experimental measure can be realized by the
use of the Quantum Hall Effect of the Volt and Ohm determination. This is due to the
fact that the definition of the Volt and Ohm due to the Quantum Hall effect is based
on universal constants. So, the corresponding realization of Ampere, as it is a
derivative of Volt and Ohm is also based on universal constants. Then, the Kg is
respectively determined by the Lorentz law, according to which a given amount of
current is required to generate a given magnitude of force between two parallel
conductors. Thus, Ampere and Kilogram are determined. This realization involves a
number of sensors. At the moment, three National Metrological Bodies are employed
in such application.

122
7. Designing the future
The successful process in the sensor market depends on the correct prediction for the
near future trends [65]. In this chapter we aim to give some hints for the targets and
future trends in this engineering market.
Miniaturization is the most significant keyword in all research trials supported by
governments and the European Committee. In many cases, miniaturization helps
significantly decrease the production cost due to thin film mass production lines.
Sometimes, the quality of the produced devices increases with the decrease of the size
of the devices. An example for that are the semi-conducting miniatures like VLSI or
ULSI chips. But, sometimes miniaturization can be a disaster in the properties of a
device. For example miniaturized fluxgates lose their competitive sensitivity
advantage over the Hall effect field sensors, or miniaturized MI thin films do not have
the repeatability found in the bulk structures of magnetic wires. In many cases
miniaturization improves the properties of semiconductors but not the properties of
magnetic materials. We believe that the key-point in miniaturization is that it is
absolutely necessary provided that the properties are kept at the required levels. This
means that in designing magnetic materials, it is correct to design atomic scale effects.
Practically, this is in agreement with the recent trends in nanocrystalline magnetic
materials for sensor and actuator applications, where the magnetic properties and
domains are found in the size of nanometers. This way, the sensing devices, which are
at the bottom of the metrological scale, will be able to reach the accuracy and
sensitivity of the secondary standards, i.e. in the order of 1 ppm.
Another important characteristic of sensors based on magnetic materials is the
dependence of their properties on ambient field. It is the main reason why mechanical
sensors based on magnetic materials have not properly found their way to the market
yet. But, it is also one of the major reasons why the vast majority of sensitive field
sensors are based on magnetic effects and materials. Improvement of the magnetic
sensors in the global sensor market strongly depends on their ability in being
independent from the ambient field. One method to do it is the cancellation of the
ambient field by using a kind of in-series-opposition magnetic circuit, wherever this is
applicable. Another way is the development of smart sensors which are to detect the
ambient field as well as the under measurement physical size and then to perform
calculation of the detected size by using microprocessor based embedded circuits.
Integration ability of sensors is another key-point of their successful operation and
properties. If the sensor is integrated with the electronic circuits on board, the
sensitivity and uncertainty of the sensing element is improved. Recent advances in
arranging magnetic materials on integrated electronic circuits indicate that such
integration can be achieved by a metal bonding process.
Concerning some future targets in sensor design, one can conclude that, position
sensors with sub-micron uncertainty are of great interest provided that the problem of
ambient field interference is overcome. This can probably achieved by obtaining
acceptable MI response on thin film structures or MDL response in nanocrystalline
materials. Another significant target is the development of field sensors reaching the
sensitivity of SQUIDs. This could probably be achieved by further tailoring of MI
effect and related materials.
Apart from that, nano-structured materials can be used to develop high resolution
security sensors which can be used in many domestic applications.
Finally, development of sensors based on magnetic materials for harsh environment
use can probably be of advantage in some given applications.

123
Annex A: Developing a sensor
Developing a sensor is a science and an art by itself, combining knowledge of
material science, engineering design and metrology. A relatively generic procedure is
given next, which can be followed to properly develop a sensor. This procedure is
also illustrated in Fig. A.1.

A.1. Facing up to the under measurement size

The very first action is the definition of the under measurement physical size or
property. It includes three certain steps, which must be investigated from the
beginning. Although they do not include much of science and engineering, they do
play a decisive role in the final sensor development.
At first, one should define the under measurement size. This issue can be addressed in
two ways. Either sensor designers do it, through a market research, or manufacturers
or service engineers do it, setting out the specifications for a sensing element as a
solution in a given problem of their production or service. In both cases, first, the
problem that needs to be solved must be precisely delineated and second, the physical
size to be measured must be determined. This can be position, mass, field, time,
electrical sizes, temperature or their derivatives.
Then, the definition of the characteristics/parameters of the under measurement
physical properties takes place, which directly reflects the characteristics of the sensor
under development. The most important properties are the range, the sensitivity, the
uncertainty and the sensor response dependence on parametric effects. The sensitivity
is the ratio of the change of the sensor response under the least possible induced input
over this induced input. The sensitivity reflects how easily a sensor can measure a
physical size. The uncertainty of the sensor is the deviation of the measurement from
the true value of the physical size at the time of measurement. As mentioned in the
Introduction, the uncertainty of sensors based on magnetic materials and effects
includes mainly the hysteresis caused by the magnetic response of the used material.
Usually, sensitivity and uncertainty are determined during the calibration process. The
magnetic sensor response dependence mainly requires ambient field dependence
determination, but also temporal, temperature & humidity dependence.

A.2. Choice of the sensing principle

Having finalized the determination of the required characteristics of the sensor under
development, the next step is the definition of the criterion or criteria for the selection
of the magnetic effect and material to be used. The criteria reflect the importance of
the above-mentioned parameters and their priority in selecting the proper effect and
material. As an example, there are cases where the dependence on ambient field is
more important than the sensor uncertainty or cases where the temperature
dependence does now allow the use of sensors having outstanding uncertainty
performance. In general the sensitivity, uncertainty, ambient field dependence,
temperature & humidity dependence as well as temporal effects are the most
important key-parameters in magnetic effect and material selection for sensor
applications.
Then, a table including the effect characteristics and required properties helps in
deciding on the closest effect to satisfy the set criteria. At this point it is important to
notice that the experience of the sensor designer in material science, in general, is
important in judging the superiority of magnetic effects and materials over other
effects, like conducting, semiconducting, superconducting, optoelectronic etc. Choice

124
of a magnetic material and effect in a given application, where other effects or
materials would perform better, may result in complete project failure.

Determination of the problem and the required properties

Use of magnetic effects

Yes

Determination and development of the sensing element

Characterization and tailoring

• Final sensing element

• Electronic circuitry
• Parametric effect control
• Sensor housing
• Final prototype

Figure A.1 Flow chart of the sensor development process.

125
A.3. Development of the sensing device
Having defined the magnetic effect to be used for the given sensor, the essential
matter is the development and the characterization of the sensing device, as well as
the proper tailoring of the magnetic material.
The development of the sensing device includes the design and manufacturing of the
magnetic material and consequently the development of the device. The design and
manufacturing of the material can be realized following three main procedures,
depending on the kind of the material.
One is the thin film process, realized by thermal evaporation including electron gun
evaporation, dc and ac sputtering, molecular beam epitaxy and chemical vapor
deposition. Thin film manufacturing methods are used in the case of several sensors
using magnetic effects, as is especially the case of GMR multilayers. Apart from that,
magnetostrictive (MX) and magneto-impedance (MI) thin film layers have recently
been developed for miniaturization purposes, with relatively promising results.
The second technique concerns the various rapid quenching techniques for ribbon,
wire and glass covered wires methods of manufacturing. The major application of this
technique concerns MI and MX materials.
Finally, another technique is the thick film manufacturing process mainly dedicated
for magnetic oxide films.
The characterization of the device includes magnetic and structural characterization
and sensing element calibration. Magnetic characterization mainly implies B-H and λ-
H loop determination in one or more axes of anisotropy, but also magneto-optic and
electrical characterization. There are two important parameters, which should be taken
into account when determining those loops. One is the loop dependence on the
frequency and the other is the spatial uniformity of the loops along the length of the
produced material. The frequency dependence is necessary for the operation of many
magnetic sensors while the spatial uniformity of these loops guarantees the
repeatability of the sensor response. Structural characterization mainly implies surface
characterization using scanning microscopes like scanning electron microscopy
(SEM), transmission electron microscopy (TEM) and scanning tunneling techniques
(AFM, MFM & STM), X-ray diffraction and differential temperature analysis (DTA)
for phase transformation analysis. The structural characterization is required in order
to explain and understand the macroscopic magnetic properties of the material as well
as to decide on the tailoring process of the material. Sensing element calibration
implies the use of the classic primary, secondary or working-standard sensor
calibration techniques. The decision concerning the calibration technique to be used
depends on the uncertainty level requirements, set by the previous steps of the sensor
development process.
The tailoring process depends on the results of the characterization procedure. The
most significant tailoring process is the annealing of the material. Annealing is
realized by thermal treatment, with parametric control of field, stress and current
applied during the process. This process mainly controls the microstructure of the
magnetic material allowing proper tailoring of the magnetic domain and magnetic
domain wall structure and dynamic behavior. Another tailoring process is the doping
of the magnetic material with other substances, using diffusion or ion implantation
techniques, which is mainly used for the hardening of the magnetic material by
pinning the magnetic dipoles in given orientations.

126
A.4. Development of the sensor electronics
After the development of the sensing element, which is finalized by the
characterization and the tailoring process, the development of the accompanying
sensor electronics is next, which includes the development of the excitation circuit,
the output signal conditioning circuit and the auto-calibration circuit. These circuits
are developed with respect to the power consumption requirements of the excitation
circuit and the required specifications of the signal conditioning and the auto-
calibration circuit.
The excitation circuit is usually either dc or sinusoidal or triangular or pulsed current
acting either on a coil or on the magnetic element itself. The power consumption of
this circuit is from 1nW to 1mW. MOSFET amplifiers or embedded circuits are the
most usual elements of these circuits.
The signal conditioning and the auto-calibration circuits are usually high frequency
bandwidth amplifiers with microprocessor-controlled analog to digital converters for
the auto-calibration procedure, if it exists on board.
Feeding the excitation circuit as well the signal conditioning and auto-calibration
circuits is realized by either the power supply from the mains if possible, or by
batteries and solar cells if applicable.

A.5. Parametric effects and sensor housing

After developing the sensing element and the sensor electronics, the most important
activity is the determination and action against the parametric effects. As mentioned
already, the main parametric effects are the ambient field interference, the temperature
and the time dependence. The time dependence is usually controlled, implementing
the aging process. This process can be achieved by using combination of oxidation
and acid attack on the magnetic sample itself. The magnetic response of the sensing
element is stabilized after some treatment and is considered as the rather permanent
response of the device. The temperature dependence problem is solved by some
practical negotiations. One has to stabilize the magnetic response of interest, which is
either the magnetization or the magnetostriction, between a minimum and a maximum
temperature limit, the lower and upper thresholds of the thermal behavior of the
sensor. This is achieved by annealing techniques. Usually the lower temperature can
be tailored down to cryogenic temperatures, while the upper temperature limit is set at
some tens of oC lower than the Curie temperature.
The ambient field interference is the most severe problem of parametric effect
dependence. If it is a field sensor that is being developed, then there is no problem for
such parametric control consideration, because it is the physical size under
measurement. If not, then there are two solutions: one is the classical magnetic
shielding to minimize such effect and the other is the cancellation of the ambient field
by in series opposition arrangements, wherever this can be applicable. The housing of
the sensor, except facing-up to the parametric effect dependence, concerns the sensor
protection against environmental conditions. This mainly refers to the IP numbers of
the electrical insulation, as specified by the International Electrotechnical Committee
(IEC). This is the sensor protection against environmental conditions such as rain,
moisture and high temperature. After all these arrangements, including material
development and optimization, parametric effect facing–up and housing, the sensor is
finally calibrated as a complete system and the determination of the final stage
corrective actions remains to be considered. After having performed any necessary
corrective actions, the final design, sensor and its technical envelope are prepared.

127
Glossary

Accelerometers Sensors measuring acceleration

Accommodation Change in magnetization that occurs at a constant
reverse field as a result of repetitively cycling the field
between H and a field of lesser magnitude
AFM Atomic Force Microscopy
AMI Asymmetric (anisotropic) magnetoimpedance
AMR Aisotropic magnetoresistance
Anisotropy Preference of the magnetization vector to lie along a
certain (crystalline or other) axis
Antiferromagnet Magnetic medium consisting of two ferromagnetically
ordered sublattices of comparable magnetizations
aligned antiparallel to each other resulting in very low
susceptibility.
Barkhausen effect The effect of Barkhausen jumps; used in
magnetoacoustic detection
Barkhausen jump Irreversible and discontinuous change in magnetization
caused by a magnetic field; related to hysteresis
Barkhausen noise Noise due to Barkhausen jumps
Chemical sensors Sensors measuring chemical sizes
CMR Colossal magnetoresistance
Aoercivity Field at which the body's magnetization is reduced to
zero after being previously saturated
Coercivity squareness The slope of the M(H) loop at the coercivity; a measure
of the strength of exchange interactions
Diamagnet Material with a very weak form of magnetization
opposing the applied field (magnetic isolator)
Digitizers Sensors able to quantize the distribution of a physical
size under measurement
Dilatometers Position sensors detecting very small displacements
Displacement sensors Position sensors measuring displacement
Domain Aligned region of uniform magnetization
Domain wall Transition region of finite thickness between two
domains inside which the magnetization changes its
orientation from one domain to the other.
Eddy currents Currents induced by a time varying magnetization
generating, in turn, a magnetic field opposing the cause
that created them and, therefore, giving rise to losses
EHE Extraordinary Hall effect
Electromagnetic induction Induced voltage at the ends of a coil because of a time-
varying flux inside it
Electronic compass Compass with electronic output
Electronic sensors Sensors with electronic output
EMC-EMI Electromagnetic compatibility- electromagnetic
interference
Exchange interactions Short-range interactions between neighboring spins
forcing their parallel alignment
Ferrimagnet Antiferromagnetically ordered medium with two

128
 sublattices of different magnetizations yielding a net
magnetic moment comparable to ferromagnets
Ferromagnet Magnetically ordered medium with high susceptibility
Field sensors Sensors measuring magnetic fields
Fluxgates Inductive type of field sensor based on the non-
hysteretic and non-linear M(H) loop of very soft
magnetic materials
Force digitizers Digitizers detecting 1D and 2D force distribution
FSMA Ferrromagnetic shape memory alloys (see also MSMA)
Garnet Ferrimagnet with optical properties
Gas field sensors Sensors detecting the presence and/or composition of
gases
GMI Giant magnetoimpedance
GMR Giant magnetoresistance
Hall Effect The voltage measured across a current-carrying
semiconducting or conducting sheet inside a transverse
magnetic field; used in field sensing
Hysteresis Non-linear delay of a system’s output with respect to its
input
Inductive sensor Sensor based on the electromagnetic induction
Interferometric technique Sensing technique based on standing wave interference
IWE Inverse Wiedemann effect
Josephson effect Electron transport through a thin insulating layer
surrounded by superconducting material when
submerged into a magnetic field
Load cells Sensors measuring applied force
LVDT Linear Variable Differential Transformer
MAD Magnetic anomaly detection
Magnetic dipole Basic magnetic entity
Magnetic force Coulomb type force between two magnetic poles
separated by distance r
Magnetic shape memory Alloys undergoing a phase transformation with the
alloys applied field (or temperature) that yields very large
deformations
Magnetic signature Coded magnetic response of a vehicle/vessel generated
by technical magnetization means and detected by field
or gradient field sensors
Magnetic target Biological agent attached to a soft magnetic bead thus
allowing its detection and monitoring through magnetic
field sensing techniques
Magnetic viscosity Change in magnetization that occurs at a constant
reverse field due to thermal fluctuation and is an
increasing function of the total time t for which H is
applied
Magnetoimpedance The % change in the impedance of a ferromagnet
depending on the magnitude and direction of an applied
field
Magnetoresistance The % change in the resistance of a ferromagnet
depending on the magnitude and direction of an applied

129
 field
Magnetostatic interactions Long-range interactions among the magnetic dipoles in
a medium responsible (among others) for the internal
demagnetizing field
Magnetostriction Property of certain ferromagnets that tend to elongate
(positive) or shrink (negative) along the direction of the
magnetization
Magnetostrictive delay line Arrangement comprised of a magnetostrictive element,
in the shape of an acoustic guide, and means of
magnetic excitation and detection offering acoustic
delay time between two points on the line
Matteucci effect Voltage measured across a wire with helical domain
structure under the application of an axial field
MDL Magnetostrictive delay line
Measuring tapes Differential position sensors in the form of tapes with
high sensititvity and accuracy
Metrology The science of measurements
MFM Magnetic force microscopy
MRAM Magnetic random access memory
MSMA Magnetic shape memory alloys
Multi-parametric sensing Sensor detecting more than one physical or chemical
element sizes
NDT&E Non-destructive testing & evaluation
Paramagnetism Weak form of magnetization varying linearly with the
field
Permalloy NiFe ferromagnetic alloy with very soft magnetic
properties and small magnetostriction
Permeability A measure of the strength of the effect a magnetic field
upon a medium or in vacuum; equal to B/H
Physical sensors Sensors detecting physical sizes
Position digitizers Digitizers detecting the position distribution of sensing
cores
Pressure gages Sensors detecting the pressure applied on a surface
Primary standards Experimental arrangements implementing laws of
physics in order to determine a physical quantity using
universal constants
Proton field sensors High sensitivity, low noise field sensors based on the
Zeeman effect
Quantum Hall effect Effect based on superconductors, used to determine the
Ohm and the Volt with respect to universal constants
Recording medium Semi hard magnetic medium used for data storage
Recording sensor Magnetic element used for writing and reading
information on a recording medium
Re-entrant flux reversal Propagation of a single domain wall along the length of
a wire because of a single Barkhausen jump; used as a
sensing effect
Remanence The magnetization of a body at zero field after it has
been previously saturated
Saturation magnetization The maximum magnetization of a body attained at a

130
 field, called the saturation field, beyond which no
further change in magnetization is observed
Secondary standards Experimental arrangements used to detect physical or
chemical sizes with uncertainty levels higher than
primary standards and lower than working standards
Security sensors Sensors detecting a prerecorded code on a substance
Self learning sensors Sensors able to adjust their output with respect to the
history of measurement
Sensing core The material inside a sensor responsible for the
detection of a physical or chemical size
Sensitivity The marginal change in a sensor’s response per unit
input
Sensor housing The arrangement surrounding a sensing element
responsible for its protection against environmental
conditions and other parameters affecting its response;
sometimes, it also refers to the electrical/electronic
signal conditioning circuitry
Sensor packaging Sensor housing
Skin depth Penetration depth of the magnetic flux opposed by eddy
currents
Skin effect Effect related to the eddy currents at the surface of a
metallic ferromagnet, inversely proportional to the skin
depth
Smart sensor Sensor able to perform either as a multi-parametric
sensing element or as a self-learning one
Spin valve Tri-layer GMR structure used in recording heads or as
memory cell
Spintronics Emerging technological field utilizing “the electronics
of spin”.
Squareness The ratio of the remanence to the saturation
magnetization of a body; an indicator of reversible
rotation approaching saturation
SQUID Superconducting quantum interference device
STD Spin Tunneling Device
STM Spin Tunneling Microscopy
Stress sensors Sensors measuring or detecting stress
Superconductor Material behaving as a perfect conductor with zero
electrical resistance below a certain temperature and a
perfect diamagnet below a certain magnetic field
Superparamagnet Small magnetically unstable particle or grain because of
relatively high thermal energy
Susceptibility A measure of the sensitivity of a medium to an applied
field; equal to M/H
Tailoring process Process responsible for adjusting the macroscopic
properties of a material to the desired ones
TMR Tunneling magnetoresistance
Torque meters Sensors detecting torque
Uncertainty The difference between the output of a sensor and the
corresponding true value

131
Villari effect The dependence of the magnetic induction on an applied
stress
Wiedemann effect Torsion induced to a soft magnetic wire when an ac
circumferential field is applied along with a dc axial one
Working standards Sensing elements with uncertainty lower than secondary
standards
Zeeman effect The effect of the magnetic field on the energy level of
electrons

132
References
1. E.O. Doebelin, Measurement Systems: applications and design, fourth edition,
McCraw-Hill (1990)
2. E.E. Herceg, Handbook of measurement and control, rev. ed., Schaevitz
Engineering, Pennsauken, N.J. (1976)
3. E. Hristoforou, Meas. Sci. & Tech., Meas. Sci. Technol. 14 (2), R15-R47
(2003)
4. R. Boll, Soft magnetic materials, Vacuumschmelze handbook, Heyden & Son,
(1977)
5. Magnetic Sensors and Magnetometers, Pavel Ripka (ed), Meas. Sci. Technol.,
13 (2002)
6. J. M. Daughton, J. MAGN. Magn. Mat. 192 (2), 334-342 (1999)
7. K. Mohri, K. Bushida, M. Noda, H. Yoshida, L.V. Panina, T. Uchiyama, IEEE
Trans. Magn., 31, 2455 –2460 (1995)
8. G. A. Prinz, Physics Today, 58-63, April 1995.
9. E. du Tremolet de Lacheisserie, Magnetostriction: Theory and Applications of
Magnetoelasticity, CRC Press, Boca Raton, FL (1994)
10. J.D. Jackson, Electricity and Magnetism, Wiley, NY (1965)
11. D. Jiles, Introduction to magnetism and magnetic materials, Chapman & Hall
(1991)
12. S. Chikazumi, S. H. Charap, Physics of Magnetism, John Wiley & Sons
(1964)
13. A. Aharoni, Introduction to Ferromagnetism, Clarendon Press (1996)
14. G. Bertotti, Hysteresis in Magnetism, Academic Press (1998)
15. H. C. Ohanian, Physics, Norton (1985)
16. C. R. Paul and S. N. Nasar, Introduction to Electromagnetic Fields, Mc Graw
Hill (1987)
17. H. de Gersem, K. Hameyer and T. Weiland, 168 (1-2), 125-133 (2004)
18. H. De Gersem, R. Mertens, D. Lahaye, S. Vandewalle and K. Hameyer,
Solution IEEE Trans. Magn. 36 (4), 1531–1534 (2000)
19. D. Philips, Internat. J. Numer. Methods Eng., 35, 1991–2002 (1992)
20. Van o B. Litovski, eljko Mr arica and Tihomir Ili , Simulation Practice and
Theory, 5(6), 553-570 (1997)
21. R. Cammarano, P.G. McCormick and R. Street, J.Phys.D: Appl. Phys. 29,
2327-2331 (1996).
22. H. Barkhausen, Phys. Z. 20, 401 (1919)
23. M. Tinkham, Introduction to superconductivity, 2nd ed., McGraw-Hill, Inc.
(1996)
24. P. Müller, A. V. Ustinov (Eds), V. V. Schmidt, Springer - Verlag (1997)
25. B. D. Cullity, Introduction to Magnetic Materials, Addison-Wesley Pub. Co,
2nd ed (1972)
26. H. R. Kirchmayr, J. Phys. D, 29, 2763-2778 (1996)
27. E.E. Fullerton, D.T. Margulies, N. Supper, Hoa Do, M. Schabes, A. Berger,
and A. Moser, IEEE Trans. Magn., 39 (2), 639-644 (2003)
28. E.E. Fullerton, D.T. Margulies, M.E. Schabes, M. Carey, B. Gurney, A.
Moser, M. Best, G. Zeltzer, K. Rubin, H. Rosen, Appl. Phys. Lett. 77, 3806
(2000)
29. A.Berger, A. Inomata, J. S. Jiang, J. E. Pearson, and S. D. Bader, Phys. Rev.
Lett. 85, 4176 (2001)
30. A.Berger, A. W. Pang and H. Hopster, Phys. Rev. B 52, 1078 (1995)

133
31. J. P. Wang, Z.S. Shan, S.N. Piramanayagam, and T.C.Chong, IEEE Trans.
Magn., 37(4), 1445-1448 (2001)
32. D.E. Heim, R.E. Fontana, C. Tsang, V.S. Speriosu, B.A. Gurney, M. L.
Williams, IEEE Trans. Magn., 30 (2), 316-321 (1994)
33. P. P. Freitas, F. Silva, N. J. Oliveira, L. V. Melo, L. Costa and N. Almeida,
Sensors and Actuators A: Physical, 81, 2-8 (2000)
34. B. Dieny, J. Magn. Magn. Mat., 136 (3), 335-359 (1994)
35. C. Tsang, R.E. Fontana, T. Lin, D.E. Heim, V.S. Speriosu, B.A. Gurney, M. L.
Williams, IEEE Trans. Magn., 30 (6), 3801-3806 (1994)
36. Zheng-Chuan Wang, Gang Su, Qing-Rong Zheng and Bao-Heng Zhao, Phys.
Lett. A, 277 (1), 47-55 (2000)
37. J-G Zhu and H. N. Bertram, J. Appl. Phys. 63 (8), 3248-3253 (1988)
38. T. Schrefl, J. Fidler, H. Kronmüller, J. Magn. Magn. Mat., 138, 15-30 (1994)
39. H. Kronmüller, T. Schrefl, J. Magn. Magn. Mat., 129, 66-78 (1994)
40. T. Schrefl, R. Fischer, J. Fidler, H. Kronmüller, J. Appl. Phys. 76 (10) 7053-
7058 (1994)
41. T. Schrefl, J. Fidler, H. Kronmüller, Phys. Rev. B, 49 (9), 6100-6110 (1994)
42. R. Fischer and H. Kronmüller, Phys. Rev. B, 54 (10), 7284-7294 (1996)
43. T. Schrefl, J. Fidler, J. Magn. Magn. Mat., 111, 105-114 (1992)
44. H. Kronmüller, Phys. Stat. Sol.(b), 114, 385 (1987)
45. H. Kronmüller, R. Fischer, M. Seeger and A. Zern, J. Phys. D, 29, 2274-2283
(1996)
46. T. R. Koehler, J. Appl. Phys. 55 (6), 2214-2216 (1984)
47. IB. Ortenburger and R. I. Potter, J. Appl. Phys. 50, 2393 (1979)
48. R. I. Potter, and I. A. Beardsley, IEEE Trans. Magn., 16 (5) 967-972 (1980)
49. A. Ktena and S. H. Charap, 29 (6), 3661-3663 (1993).
50. E. C. Stoner and E.P. Wohlfarth, Phil. Trans. Roy. Soc., A240, 599-642
(1948)
51. N. Tsuya, K.I. Arai, T. Oksaka, IEEE Trans. Magn., 14, 946-948 (1978)
52. K. Kakuno, S. Masuda, T. Yamada, C8, 2037 (1988)
53. H. Hauser, E. Hristoforou and A. Ktena, J. Appl. Phys., 93 (10), 8633-8635
(2003)
54. M. Vázquez, Physica B, 299, 302-313 (2002)
55. A. Mitra, M. Vázquez, A. Hernando and C. Gomez-Polo, IEEE Trans. Magn.,
26, 1415-1417 (1990).
56. J. Gonzalez, J.M. Blanco, J.M. Barandiaran, M. Vázquez, A. Hernando, G.
Rivero and D. Niarchos, IEEE Trans. Magn., 26, 1798-1800 (1990)
57. E. Hristoforou and R.E. Reilly, J. Appl. Phys., 69, 5008-5010 (1991)
58. K. Kakuno, S. Masuda, T. Yamada, H. Mochida, IEEE Trans. Magn., 23,
2554-2556 (1987)
59. B. Ahamada, F. Alves and R. Barrué, J. Magn. Magn. Mat., 242-245, 1443-
1445 (2002)
60. J. L. Prieto, P. Sánchez, C. Aroca, E. López, M. C. Sánchez, O. de Abril and
L. Pérez, Sensors and Actuators A: Physical, 84, 338-341 (2000)
61. R. Skomski, J. Appl. Phys. 76 (10), 7059-7064 (1994)
62. R. Skomski and J.M.D. Coey, Phys. Rev. B, 48(21), 15812-15816 (1993)
63. L. Lanotte, V. Iannotti, Intern. J. Appl. Electromagn. Mech., 16, 155-163
(1995)
64. L. Lanotte. Z. Kaczkowski, R. Pasquino, L. Palezzo, V. Tagliaferri, J. Magn.
Magn. Mat., 135, 265-270 (1994)

134
65. E. Hristoforou, J. Opt. Adv. Mat., 4, 245-260 (2002)
66. Y.H. Chen, D.C. Jiles, IEEE Trans. Magn. 37, 3069 –3072 (2001)
67. P. Bruno, Phys. Rev. Lett., 83 (12), 2425-2428 (1999)
68. A.Hubert and R. Schafer, Magnetic Domains: The analysis of magnetic
microstructures, Springer (1998)
69. W. F. Brown Jr., Magnetostatic Principles in Ferromagnetism, North Holland-
Amsterdam (1962)
70. H. Hauser, J. Appl. Phys. 75 (5), 2584-2597 (1994)
71. H. Hauser, J. Appl. Phys. 77 (6), 2625-2633 (1995)
72. A.Lyberatos. J. Magn. Magn. Mater. 202, 239 (1990)
73. A.Thiaville, J. Magn. Magn. Mat., 182, 5-18 (1998)
74. D. C. Jiles and D. L. Atherton, J. Magn. Magn. Mat., 61, 48-60 (1986)
75. D. C. Jiles, J. Magn. Magn. Mat., 242-245, 116-124 (2002)
76. M. J. Sablik and D. C. Jiles, J. Appl. Phys. 64, 5402 (1988)
77. ID. Mayergoyz, Mathematical models of hysteresis, Springer-Verlag, New
York (1991)
78. S. H. Charap and A. Ktena, J. Appl. Phys., 73, 5818-5823 (1993)
79. K. C. Wiesen, S. H. Charap and C.S. Kreafft, J. Appl. Phys., 67, 5367-5369
(1990)
80. T. L. Gilbert, Phys. Rev. 100, 1243 (1955)
81. G. Bertotti, I. D. Mayergoyz, C. Serpico, Physica B, 343, 325-330 (2003)
82. V. Safonov, H. N. Bertram, Phys. Rev. B 63, 0994419 (2001)
83. V. Tsiantos, T. Schrefl, W. Scholz, H, Forster, D. Suess, R. Dittrich. J. Fidler,
Sensors & Actuators, A 106, 134-136 (2003)
84. P. J. Thompson and R. Street, J. Phys. D: Appl. Phys., 29, 2789-2795 (1996)
85. D. C. Crew, P. G. McCormick and R. Street, J. Phys. D: Appl. Phys., 29,
2313-2319 (1996)
86. P. J. Thompson and R. Street, J. Phys. D: Appl. Phys., 30, 1273-1284 (1997)
87. Z. Frait, and D. Fraitovà, “Spin-wave resonance in metals”, in Modern
Problems of Physics, Spin Waves and Magnetic Excitations, Part 2, A.S.
Borovik-Romanov & R. Sinha, eds., Elsevier Science, Amsterdam (1988)
88. B. A. Gurney, D. R. Wilhoit, V. S. Speriosu, I. L. Sanders, IEEE Trans.
Magn., 26 (5), 2747-2749 (1990)
89. Shufeng Zhang, Peter M. Levy, Mat. Sci. Eng. B31, 157-162 (1995)
90. A. Vedyayev, D. Bagrets, A. Bagrets, N. Ryzhanova, N. Strelkov, B. Dieny
and C. Lacroix, J. Magn. Magn. Mat., 242-245, 453-456 (2002)
91. N. Ryzhanova, C. Lacroix, A. Vedyayev, D. Bagrets and B. Dieny, J. Magn.
Magn. Mat., 226-230, 750-751 (2001)
92. H. Itoh, N. Nishimura and J. Inoue, J. Magn. Magn. Mat., 240, 121-123,
(2002)
93. J. M. V. Lopes, J. M. B. Lopes dos Santos and Yu. G. Pogorelov, J. Magn.
Magn. Mat., 242-245, 482-484 (2002)
94. D. Kechrakos and K. N. Trohidou, Physica B: Condensed Matter, 318, 360-
364 (2002)
95. J. E. Wittig, J. F. Al-Sharab, M. Coerner, X. Bian, S. Bentley and N.D. Evans,
48, 943-948 (2003)
96. R.S. Indeck, E. Glavinas, IEEE Trans. Magn., 29 (6), 4095-407 (1993)
97. C. Dolabdjian, A. Qasimi, D. Bloyet and V. Mosser, Physica C:
Superconductivity, 368, 80-84 (2002)
98. P. Ripka, J. Magn. Magn. Mat., 157/158, 424-427 (1996)

135
99. E. H. Feutrill, P. G. McCormick and R. Street, J.Phys.D: Appl. Phys. 29,
2320-2326 (1996)
100. T. Meydan and K. J. Overshott, J. Magn. Magn. Mat., 41, 449-451 (1984)
101. M. Mizutani, H. Katoh, L.V. Panina, K. Mohri, F.B. Humphrey, IEEE Trans.
Magn., 29, 3174 –3176 (1993)
102. Floyd B. Humphrey, Mat. Sci. Eng. A, 179-180, 66-71 (1994)
103. T.B. Thompson and J.A.M. Lyon, Trans. IRE PGUE-4, 8 (1956)
104. S. Masuda, K. Kakuno, IEEE Trans. Magn., 26, 1801-1803 (1990)
105. E. Hristoforou, Sensors and Actuators A, 59, 183 – 191 (1997)
106. K. Diaz de Lezana, A. García-Arribas, J. M. Barandiarán and J. Gutiérrez,
Sensors and Actuators A, 91, 226-229 (2001)
107. P. Ripka, J. Magn. Magn. Mat., 215-216, 735-739 (2003)
108. M. Pasquale, A. Infortuna, L. Martino, C. Sasso, C. Beatrice and S. H. Lim, J.
Magn. Magn. Mat., 215-216, 769-771 (2000)
109. I. Ogasawara and K. Mohri, IEEE Trans. Magn., 26, 1795-1797 (1990)
110. G. Herzer, IEEE Trans. Magn., 26, 1397-1402 (1990)
111. Takajo M, Yamasaki J, Humphrey FB, IEEE Trans. Magn. 35 (5), 3904-3906
(1999)
112. Li L, J. Appl. Phys. 81 (8), 4287-4289 (1997)
113. M. Soeda, M. Takajo, J. Yamasaki, et al., IEEE Trans. Magn. 31 (6), 3877-
3879 (1995)
114. J.L. Costa, Y. Makino, K.V. Rao, IEEE Trans. Magn., 26, 1792 –1794 (1990)
115. J. Yamasaki, M. Takajo, F. B. Humphrey, IEEE Trans. Magn 29 (6), 2545-
2547 (1993)
116. L. V. Panina and K. Mohri, J. Magn. Magn. Mat., 157-158, 137-140 (1996)
117. K. Kawashima, T. Kohzawa, H. Yoshida, K. Mohri, IEEE Trans. Magn., 29,
3168 –3170 (1993)
118. M. Vázquez, C. Gomez-Polo, D.X. Chen, A. Hernando, IEEE Trans. Magn.,
30, 907 –912 (1994)
119. S. Puerta, D. Cortina, H. García-Miquel, D.-X. Chen, M. Vázquez, J. Cryst.
Sol., 287, 370-373 (2001)
120. M. Vázquez and A. Hernando, J. Phys. D: Appl. Phys., 29, 939-949 (1996)
121. L. Kraus, J. Phys. E: Sci. Instrum., 22, 943-947 (1989)
122. M. Kaneko, S. Hashimoto, M. Hayakawa and K. Aso, J. Phys. E: Sci.
Instrum., 21, 487-489 (1988)
123. E. Hristoforou, H. Chiriac and M. Neagu, Sensors and Actuators A, 67, 49-54
(1998)
124. R. Germano, G. Ausanio, V. Iannotti, L. Lanotte and C. Luponio, Sensors and
Actuators A: Physical, 81, 134-136 (2000)
125. R. Germano, L. Lanotte, Sensors and Actuators A, 59, 337-341 (1997)
126. L. Lanotte and R. Germano, Sensors and Actuators A: Physical, 59, 337-341
(1997)
127. H. Chiriac and C. S. Marinescu, Sensors and Actuators A: Physical, 81, 174-
175 (2000)
128. H. Chiriac, M. Pletea and E. Hristoforou, Sensors and Actuators A, 68, 414-
418 (1998)
129. K. Imamura, S. Kwang-Ho, K. Ishiyama, M. Inoue, K.I. Arai, K.I., IEEE
Trans. Magn., 37, 2025 –2027 (2001)
130. A.P. Thomas, M.R.J. Gibbs. and P.T. Squire, IEEE Trans. Magn., 26, 1403-
1405 (1990)

136
131. T. Tatsumi, Y. Tsukamoto, K. Yamada, Y. Motomura, and M. Aoyama, J.
Appl. Phys. 69 (8), 4671-4673 (1991)
132. T. Meydan and P. Choudhary, Sensors and Actuators A: Physical, 59, 51-55
(1997)
133. V. Birman, Appl. Mech. Rev. 50 (11), 629-645 (1997)
134. Qi Pan and R.D.James, J. Appl. Phys. 87(9), 4702-4706 (2000)
135. Manfred Wuttig et. al., J. Appl. Phys., 87(9), 4707-4711 (2000)
136. Alexandre Vassiliev, J. Magn. Magn. Mat., 242-245, 66-67 (2002)
137. Pasquale M, et. al., IEEE Trans. Magn., 36(5), 3263-3265 (2000)
138. O. Heczko, K. Jure, K. Ullakko, Magn. Magn. Mat., 226-230, 996-998 (2001)
139. M. Pasquale, Sensors & Actuators A, 106, 142-148 (2003)
140. L. Lanotte, G. Ausanio, C. Hison, V. Iannotti, C. Luponio, Sensors &
Actuators, A106, 56-60 (2003)
141. S. Bednarek, J. Magn. Magn. Mat., 166, 91-96 (1997)
142. L. Lanotte, G. Ausanio, V. Iannotti, G. Pepe, G. Carotenuto, P. Netti, L.
Nicolais, Phys. Rev. B, 63 (054438) ,1-6, (2001)
143. E. Quandt, 258, 126-132 (1997)
144. S. Wuchner, J. Betz, D. Givord, K. McKay, A. D. Santos, Y. Souche, J.
Voiron, J. Magn. Magn. Mat., 126 (1-3), 352-354 (1993)
145. Givord D, Betz J, Mackay K, Toussaint JC, Voiron J, Wuchner S, J. Magn.
Magn. Mat.,159 (1-2), 71-79 (1996)
146. K. -M. H. Lenssen, H. W. Van Kesteren, J. C. S. Kools, M. C. De Nooijer, R.
Coehoorn, W. Folkerts and Th. G. S. M. Rijks, Sensors and Actuators A:
Physical, 60, 90-97 (1997)
147. J.M. Daughton and Y. J. Chen, IEEE Trans. Magn., 29 (6), 2705-2710 (1993)
148. J. M. Daughton, IEEE Trans. Magn., 30 (2), 364-368 (1994)
149. M. N. Baibich, J. M. Broto, A. Fert, F. Nguyen Van Dau, F. Petroff, Phys.
Rev. Lett., 61 (21), 2472-2475 (1988)
150. D. D. Tang, P. K. Wang, V.S. Speriosu, S. Le, K. K. Kung, IEEE Trans.
Magn., 31 (6), 3206-3208 (1995)
151. S. I. Kasatkin, P. I. Nikitin, A. M. Muravjev, V. V. Lopatin, F. F. Popadinetz,
A. V. Svatkov, A. Yu. Toporov, F. A. Pudonin and A. I. Nikitin, 85, 221-226
(2000)
152. Jagadeesh S. Moodera and George Mathon, J. Magn. Magn. Mat., 200(1-3),
248-273 (1999)
153. S. Yuasa, T. Nagahama, Y. Suzuki, Science, 297, 234-237 (2002)
154. K. Attenborough, H. Boeve, J. de Boeck, G. Borghs and J. -P. Celis, Sensors
and Actuators A: Physical, 81, 9-12 (2000)
155. F. E. Stanley, C. H. Marrows, I. Zoller, E. W. Hill and B. J. Hickey, 81, 32-36
(2000)
156. Y. Kitade, H. Kikuchi, H. Kishi, M. Otagiri, K. Kobayashi, IEEE Trans.
Magn., 31, 2600 –2602 (1995)
157. O. J. Gonzalez, E. Castaño, J. C. Castellano, F. J. Gracia, Sensors & Actuators
A 92, 137-143 (2001)
158. S. Li and M. Greenblatt, J. of Alloys and Compounds, 338, 121-125 (2002)
159. F. Montaigne, S. Mangin and Y. Henry, Phys. Rev. B, 144412 (2003)
160. K.-U. Barholz, R. Mattheis, IEEE Trans. Magn. 38, 2767 (2002)
161. P. Ripka, M. Vopálenský, A. Platil, M. Döscher, K. -M. H. Lenssen and H.
Hauser, J. Magn. Magn. Mat., 254-255, 639-641 (2003)
162. H.D. Chopra, S.Z. Hua, Phys. Rev. B, 66 (2) (2002)

137
163. S.Z. Hua, H.D. Chopra, Phys. Rev. B ,67 (6) (2003)
164. E. B. Svedberg, J. J. Mallett, H. Ettedgui, L. Gan, P.J. Chen, A.J. Shapiro, T.
P. Moffat, W. F. Egelhoff, Appl. Phys. Lett. 84 (2), 236-238 (2004)
165. Yi G, Phys. Rev. B, 69 (13) (2004)
166. S. Das Sarma, Jaroslav Fabian, Xuedong Hu and Igor uti , Superlattices and
Microstructures, 27, 289-295 (2000)
167. P. S. Anil Kumar, R. Jansen, O. M. J. van’t Erve, R. Vlutters, S. D. Kim and J.
C. Lodder, Physica C: Superconductivity, 350, 166-170 (2001)
168. P. S. Anil Kumar, R. Jansen, O. M. J. van’t Erve, R. Vlutters, S. D. Kim and J.
C. Lodder, Physica C: Superconductivity, 350, 166-170 (2001)
169. S. D. Kim, O. M. J. van’t Erve, R. Jansen, P. S. Anil Kumar, R. Vlutters and J.
C. Lodder, Sensors and Actuators A: Physical, 91, 166-168 (2001)
170. Frank (Zhigang) Wang, Lianna He, Desmond J. Mapps and Warwick W.
Clegg, J. Magn. Magn. Mat., 198-199, 125-127 (1999)
171. M. Guth, G. Schmerber and A. Dinia, Mat. Sci. Eng. C 19, 129-133 (2002)
172. J. H. Oh , J. H. Park , H. J. Kim , W. C. Jeong , G. H. Koh , G. T. Jeong , I. J.
Hwang , T. W. Kim , J. E. Lee , H. J. Kim et al., J. Magn. Magn. Mat (2004)
173. M. Knobel, M. Vàzquez, L. Kraus, Giant Magnetoimpedance, in Handbook of
Magnetic Materials, North Holland (2002)
174. L. Kraus, GMI modeling and material optimisation, Sensors & Actuators
A106, 187-194 (2003)
175. M. Vázquez, M. Knobel, M. L. Sánchez, R. Valenzuela and A. P. Zhukov, A:
Physical, 59, 20-29 (1997)
176. K. R. Pirota, L. Kraus, H. Chiriac and M. Knobel, J. Magn. Magn. Mat., 226-
230, 730-732 (2001)
177. T. Yoshinga, S. Furukawa, K. Mohri, IEEE Trans. Magn., 35, 3613 –3615
(1999)
178. L. V.Panina, D.P.Makhnovskiy, K. Mohri, J. Magn. Magn. Mat., 272-276,
1452-1459 (2004)
179. R.Valenzuela, M. Knobel, M. Vázquez and A. Hernando, J. Appl. Phys., 78,
5189 (1995)
180. K. L. García, J.M. García-Beneytez, R. Valenzuela, A. Zhukov, J. González,
M. Vázquez, J. Magn. Magn. Mat., 226-230, 721-723 (2001)
181. K.,Mohri, IEEE Trans. Magn., 28, 3150 (1992)
182. K.,Mohri, Mater. Sci. Eng. A, 185, 141 (1994)
183. L. V. Panina, K. Mohri, Sens. Act. A, 81, 71 (2000)
184. G.V. Kulyandskaya, A. Garcìa-Arribas, J.M. Barandiarán, Sensors &
Actuators A106, 234-239 (2003)
185. M. Tejedor, B. Hernando, M. L. Sánchez, V. M. Prida and P. Gorria, 287, 396-
400 (2001)
186. L. V. Panina, J. Magn. Magn. Mat., 249, 278-287 (2002)
187. K. Kawashima, I. Ogasawara, S. Ueno, K. Mohri, IEEE Trans. Magn., 35,
3610 –3612 (1999)
188. C. G. Kim, K. J. Jang, H. C. Kim and S. S. Yoon, J. Appl. Phys., 85, 5447
(1999)
189. Kwang Seok Byon, Seong-Cho Yu, Cheol Gi Kim, M. Vázquez, J. Non-Cryst.
Sol., 287, 339-343 (2001)
190. E. Hristoforou, J. Magn. Magn. Mater., 249, 387-392 (2002)
191. K.G. Van den Berg, J. Phys. E: Sci. Instrum., 15, 325-327 (1982)
192. M. Inoue, N. Fujita, T. Fujii, IEEE Trans. Magn., 20, 1406-1408 (1984)

138
193. H. Epstein and O.B. Strum, Transactions IRE PGUE-5, 1 (1957)
194. D.W. Forester, C. Vittoria, D.C. Webbs, K.L. Davis, J. Appl. Phys., 49, 1794-
1796 (1978)
195. M. Yamaguchi, K.Y. Hashimoto, H. Kogo, M. Naoe, IEEE Trans. Magn., 16,
916-918 (1980)
196. E. Hristoforou, H. Hauser and D. Niarchos, J. Magn. Magn. Mater., 269-272,
(2002)
197. L. Lanotte. Z. Kaczkowski, L. Maritato, Europhys. Lett., 8, 717-722 (1992)
198. Adam Bie kowski, J. Magn. Magn. Mat., 215-216, 231-234 (2000)
199. Roman Szewczyk and Adam Bie kowski, J. Magn. Magn. Mat., 254-255,
284-286 (2003)
200. I.Mihalca, A. Ercuta and C. Ionascu, Sensors and Actuators A: Physical, 106
(1-3), 61-64 (2003)
201. K. Kimura, M. Kanoh, K. Kawashima, K. Mohri, M. Takagi and L. V. Panina,
IEEE Trans. Magn., 27 (6), 4861-4863 (1991)
202. C. Gómez-Polo, J. Arcas, M. Vázquez and A. Hernando, J. Magn. Magn. Mat.,
160, 194-196 (1996)
203. S. N. Kane, M. Vázquez, A. Hernando and A. Gupta, Mat. Sci. Eng. A, 304-
306, 1055-1057 (2001)
204. H. Chiriac, E. Hristoforou, M. Neagu, F. Barariu, T. A. Ovari, J. Appl. Phys.,
85 (8), 5729-5731 (1999)
205. H. Chiriac, E. Hristoforou, M. Neagu, I. Darie, F. Barariu, IEEE Trans. Magn.,
35(5), 3625-3627 (1999)
206. E. Pulido, R.P. del Real, F. Conde, G. Rivero, M. Vazquez, E. Ascasibar, A.
Hernando, IEEE Trans. Magn., 27, 5241 –5243 (1991)
207. T.K. Worthington and D.A. Thompson, IEEE Trans. Magn., 6, 693-695
(1980)
208. A. Parakka and D. C. Jiles, Magneprobe, J. Magn. Magn. Mat., 140-144,
1841-1842 (1995)
209. J. Skalla, G. Wäckerle, M. Mehring and A. Pines, Physics Letters A, 226, 69-
74 (1997). Also in: J. Avron and I. Herbst and B. Simon, Physics Letters A,
62, 214-216 (1977)
210. O.V. Nielsen, P. Brauer, F. Primdahl, J.L. Jørgensen, C. Boe, T. Risbo, M.
Deyerler and S. Bauereisen, Sensors and Actuators A, 59, 168-176 (1997)
211. I. Sasada and T. Usui, Paper M-I.4, EMSA 2004, Cardiff, UK
212. J. Petrou, P. D. Dimitropoulos, E. Hristoforou, Paper M-O.5, EMSA 2004,
Cardiff, UK
213. P. D. Dimitropoulos, J. N. Avaritsiotis and E. Hristoforou, Sensors and
Actuators, 107, 238-247 (2003)
214. E. Hristoforou, H. Chiriac, M. Neagu, IEEE Trans. Instr. & Meas., 46, 632-
635 (1997)
215. G. Boero, M. Demierre, P. -. A. Besse and R. S. Popovic, Sensors and
Actuators, 106, 314-320, (2003)
216. P. Kemidis, T. Orfanidou and E. Hristoforou, J. Opt. Adv. Mat., 347-352
(2002)
217. H. Chiriac, E. Hristoforou, M. Neagu, M. Pieptanariu, F.G. Castano, J. Appl.
Phys., 87, 5344-5346 (2000)
218. Y. Kano, S. Hasebe, C. Huang, T. Yamada and M. Inubuse, IEEE Trans.
Magn., 26, 2023-2025 (1990)

139
219. E. Hristoforou, H.Chiriac, M.Neagu, V.Karayannis, Sensors and Actuators A,
59, 89 - 93 (1997)
220. L. Lanotte, C. Luponio, A. Annunziata, J. Magn. Magn. Mat., 80, 153-158
(1989)
221. H. Chiriac and C. S. Marinescu, Sensors and Actuators, 81, 174-175 (2000)
222. C. Petridis, P. Dimitropoulos, E. Hristoforou, EMSA 2004, Cardiff, UK
223. E. Hristoforou and D. Niarchos, J. Magn. Magn. Mat., 116, 177-188 (1992)
224. Ε. Hristoforou, H. Chiriac, Proc. 4th International Workshop οn Nano-
Crystalline Solids, 505-509 (World Scientific) Madrid, Spain (1994)
225. E. Hristoforou, R.E. Reilly and D. Niarchos, IEEE Trans. Magn., 29, 3171-
3173 (1993)
226. E. Hristoforou and R.E. Reilly, IEEE Trans. Magn., 30, 2728-2733 (1994)
227. A. Murakami, K. Hosaka, M. Fukushima, M. Maeda, N. Tsuya, IEEE Trans.
Magn., 24, 1758-1760 (1988)
228. T. Meydan, M.S.M. Elshebani, J. Magn. Magn. Mat., 112, 344-346 (1992)
229. E. Hristoforou and R.E. Reilly, J. Appl. Phys., 70, 4577-4580 (1991)
230. Ε. Hristoforou, Αrray Sensors Based οn Amorphous Alloys, PhD Thesis,
University of London (1991)
231. Sony position sensors (1991)
232. T. Meydan, J. Magn. Magn. Mat, 133, 525-532 (1994)
233. E. Hristoforou, M. Neagu, H. Chiriac, IEEE Trans. Magn., 35(5), 3622 – 3624
(1999)
234. D. Son, J. Sievert, IEEE Trans. Magn., 26, 2017 –2019 (1990)
235. I. Sasada, T. Tanaka, M. Baba, K. Harada, IEEE Trans. Magn., 24, 2886 –
2888 (1988)
236. I.J. Garshelis, A torque transducer utilizing a circularly polarized ring, IEEE
Trans. Magn., 28, 2202 –2204 (1992)
237. I. Youroudi, C. Orfanidou and E. Hristoforou, Sensors and Actuators, 106,
179-182 (2003)
238. E. Hristoforou, J.N.Avaritsiotis, H.Chiriac, Sensors and Actuators A, 59, 94-
96 (1997)
239. E. Hristoforou, H. Chiriac, M. Neagu, I. Darie, Sensors and Actuators A, 68,
307-315 (1998)
240. E. Hristoforou and R.E. Reilly, IEEE Trans. Magn., 28, 1974 - 1977 (1992)
241. H.Chiriac, E. Hristoforou, M.Grigorica, A.Moga, Sensors and Actuators A, 59,
280-284 (1997)
242. M. Pletea, H. Chiriac, E. Hristoforou, M. Neagu, Sensors & Actuators A, 92,
115-118 (2001)
243. E. Hristoforou and R.E. Reilly, J. Magn. Magn. Mat., 119, 247-253 (1993)
244. E. Hristoforou, H. Chiriac, and JN Avaritsiotis, Sensors & Actuators A, 76,
156 – 161 (1999)
245. M. Vazquez, C. Gomez-Polo, D.X. Chen, A. Hernando, IEEE Trans. Magn.,
30, 907 –912 (1994)
246. E. Hristoforou and H. Hauser, IEEE Sensors 2004 Conference, Vienna,
Austria.
247. Ε. Hristoforou and K. Pekmestzi, 12th IMEKO, Zagreb, Croatia (2002)
248. Metals Handbook, “Non destructive evaluation and Quality Control”,Vol. 17,
9th Edition, American Society for Metals (1989)
249. B. Zhu, M.J. Johnson, C.C.H. Lo and D.C. Jiles, IEEE Trans. Magn., 37, 1095
– 1099 (2001)

140
250. C.C.H. Lo, J.A. Paulsen, E.R. Kinser and D.C. Jiles, IEEE Trans.
Magn., 40, 2173 – 2175 (2004)
251. E. Hristoforou, D. Niarchos, H. Chiriac, M. Neagu, Sensors & Actuators A,
92, 132-136 (2001)
252. D.L. Graham, H.A. Ferreira and P.P. Freitas, 22, 455-462 (2004)
253. X. Xie, X. Zhang, B. Yu, H. Gao, H. Zhang and W. Fei, J. Magn. Magn. Mat,
280, 164-168 (2004)
254. L.G. Puckett, G. Barrett, D. Kouzoudis, C. Grimes and L.G. Bachas, ,
Biosensors and Bioelectronics, 18, 675-681 (2003)
255. E. Maliaritsi, MSc Dissertation, National Technical University of Athens,
(2002)
256. T. Klinger, H. Pfutzner, P. Schonhuber, K. Hoffmann, N. Bachl, IEEE Trans.
Magn., 28, 2400 –2402 (1992)
257. U. O. Häfeli, Int. J. of Pharmaceutics, 277, 19-24 (2004)
258. Y. Honkura, J. Magn. Magn. Mat, 249, 375-381 (2002)
259. O. L. Sokol-Kutylovskij, Sensors and Actuators, 62, 496-500 (1997)
260. W. Wulfhekel, R. Hertel, H. F. Ding, G. Steierl and J. Kirschner, J. Magn.
Magn. Mat, 249, 368-374 (2002)
261. P.T. Olsen, W.L. Tew, E.R. Williams, R.E. Elmquist, H. Sasaki, IEEE Trans.
Instrum. Meas., 2, 115-120 (1991)

141