Upload
savash84
View
267
Download
5
Tags:
Embed Size (px)
Citation preview
CHARACTERIZATION OF STEEL MICROSTRUCTURES BY
MAGNETIC BARKHAUSEN NOISE TECHNIQUE
A THESIS SUBMITTED TO THE GRADUATE SCHOOL OF NATURAL AND APPLIED SCIENCES
OF MIDDLE EAST TECHNICAL UNIVERSITY
BY
KEMAL DAVUT
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR
THE DEGREE OF MASTER OF SCIENCE IN
METALLURGICAL AND MATERIALS ENGINEERING
DECEMBER 2006
Approval of the Graduate School of Natural and Applied Sciences Prof. Dr. Canan Özgen Director
I certify that this thesis satisfies all the requirements as a thesis for the degree of Master of Science. Prof. Dr. Tayfur Öztürk Head of Department
This is to certify that we have read this thesis and that in our opinion it is fully adequate, in scope and quality, as a thesis for the degree of Master of Science. Assoc. Prof. Dr. C. Hakan Gür Supervisor Examining Committee Members Prof. Dr. Tayfur Öztürk (METU, METE) Prof. Dr. Şakir Bor (METU, METE) Prof. Dr. Macit Özenbaş (METU, METE) Assoc. Prof. Dr. C. Hakan Gür (METU, METE) Dr. İbrahim Çam (METU, Central Lab.)
iii
I hereby declare that all information in this document has been obtained and presented in accordance with academic rules and ethical conduct. I also declare that, as required by these rules and conduct, I have fully cited and referenced all material and results that are not original to this work. Name, Last name : Kemal DAVUT
Signature :
iv
ABSTRACT
CHARACTERIZATION OF STEEL MICROSTRUCTURES BY
MAGNETIC BARKHAUSEN NOISE TECHNIQUE
DAVUT, Kemal
M.S., Department of Metallurgical and Materials Engineering
Supervisor: Assoc. Prof. Dr. C. Hakan Gür
December 2006, 80 pages
This aim of this thesis is to examine the possibility of using Magnetic
Barkhausen Noise (MBN) technique in characterizing the microstructures of
quenched and tempered low alloy steels as well as annealed low carbon
steels. To determine the average grain size by MBN, SAE 1010 steel
consisting of dominantly ferrite was used. The specimens were slowly cooled
in the furnace after austenitizing at different time and temperature variations.
By metallographic examination the average ferrite grain size of specimens was
determined. The magnetic parameters were measured by a commercial MBN
system. With increasing ferrite grain size, the magnetic Barkhausen jumps
caused by the microstructure were decreased due to the reduction in grain
boundary density per unit volume. A clear relationship has been observed
between average grain size and the magnetic Barkhausen noise signals. SAE
4140, 5140 and 1040 steels were used to characterize the microstructures of
quenched and tempered specimens. After austenitizing and quenching
identically, the specimens were tempered at various temperatures between
200oC and 600oC. Formation of the desired microstructures was ensured by
metallographic examinations and hardness measurements. The results show
that as tempering temperature increases the Barkhausen activity increases
due to the enhancement of domain wall displacement with softening of the
martensite. It has been shown that MBN is a powerful tool for evaluating the
microstructures of martensitic and annealed steels.
Keywords: Magnetic Barkhausen Noise, Microstructure, Grain Size,
Quenching, Tempering
v
ÖZ
ÇELİK İÇYAPILARININ MANYETİK BARKHAUSEN GÜRÜLTÜSÜ YÖNTEMİ
ile KARATERİZASYONU
Davut, Kemal
Yüksek Lisans, Metalurji ve Malzeme Mühendisliği Bölümü
Tez Yöneticisi: Doç. Dr. C. Hakan Gür
Aralık 2006, 80 Sayfa
Bu tezin amacı Manyetik Barkhausen Gürültüsü (MBG) tekniğinin su verilmiş
ve temperlenmiş düşük alaşımlı çeliklerin ve de tavlanmış düşük karbonlu
çeliklerin iç yapılarının karakterizasyonu için kullanılabilme olasılığını
incelemektir. Çeliklerde tane boyutunun MBG yöntemi ile tayini için ferrit içeriği
çok yüksek SAE 1010 çeliği kullanılmıştır. Numuneler farklı sıcaklık ve zaman
varyasyonlarında östenitlendikten sonra fırında soğutulmuştur. Metalografik
inceleme ile numunelerde ortalama ferrit tane boyutu ölçülmüştür. Numunelerin
manyetik parametreleri ticari bir MBN sistemi ile ölçülmüştür. İçyapıdaki ferrit
taneleri irileştikçe Barkhausen zıplamaları birim hacimdeki tane sınırı
yoğunluğuna bağlı olarak azalmakta ve zayıf manyetik gürültü sinyallerine
neden olmaktadır. Ortalama tane büyüklüğü ile MBN sinyalleri arasında
belirgin bir ilişki gözlemlenmiştir. Su verilmiş ve temperlenmiş yapıları
karakterize etmek için SAE 4140, 5140 ve 1040 çelikleri kullanılmıştır. Benzer
şekilde östenitlenip su verilen numuneler 200oC ve 600oC aralığında
temperlenmiştir. İç yapı metalografik inceleme ve sertlik ölçümleriyle
karakterize edilmiştir. Temperleme sıcaklığı arttıkça Barkhausen aktivitesi,
yumuşayan martensitik yapıda domen duvarlarının daha kolay hareket
edebilmesi sayesinde artmaktadır. Bu çalışma MBG tekniğinin martensitik ve
tavlanmış çeliklerin içyapılarını değerlendirmek için uygun bir yöntem olduğunu
göstermiştir.
Anahtar Kelimeler: Manyetik Barkhausen Gürültüsü, İçyapı, Tane büyüklüğü,
Su Verme, Temperleme
vi
To my family
vii
ACKNOWLEDGEMENTS The author wishes to express his deepest gratitude to his supervisor
Assoc. Prof. Dr. C. Hakan Gür for his guidance, understanding and
continuous support throughout the study.
The author would also like to thank METU-Central Laboratory for the
MBN measurements and would like to express his gratitude to Dr.
İbrahim Çam for introducing him with the concept of MBN and helping
with the measurements.
The author is deeply grateful to Cengiz Tan for his help with the SEM
appointments.
The author would also like to thank research assistant Caner Şimşir for
all his help throughout the study.
The technical assistance of Mr. Hüseyin Çolak and Mr. Özdemir Dinç in
the heat treatment and metallography laboratories are gratefully
acknowledged.
The author wishes to express his gratitude to his parents Lale and Haluk
Davut for always supporting him and encouraging him to continue.
The author would also like to thank Oya Araslı, Naciye Tezel, Serhat
Önol Şakar, Selen Gürbüz, Gül Çevik for giving him the strength to finish
his degree by sharing all the good and the bad times.
viii
TABLE OF CONTENTS PLAGIARISM.....................................................................................................iii ABSTRACT...................................................................................................... iv ÖZ.......................................................................................................................v ACKNOWLEDGMENTS...................................................................................vii TABLE OF CONTENTS..................................................................................viii LIST OF TABLES..............................................................................................xi LIST OF FIGURES.............................................................................................x Chapter 1 – Introduction ..................................................................................1
1.1 Theory of Magnetic Barkhausen Noise.....................................................1
1.2 Literature Survey.....................................................................................16
1.2.1 Determination of Grain Size of Steels by MBN ................................16
1.2.2 Characterization of Quenched and Tempered Steel Microstructures
by MBN .....................................................................................................17
1.3 The Aim of the study ...............................................................................18
Chapter 2 - Experimental................................................................................20 2.1 Experimental Procedure for Grain Size Determination ...........................20
2.2 Experimental Procedure for Investigation of Quenched and Tempered
Samples ........................................................................................................21
Chapter 3 Results ...........................................................................................22 3.1 Grain Size ...............................................................................................22
3.1.1 Microstructure ..................................................................................22
3.1.2 MBN Measurements ........................................................................29
3.2 Quenched and Tempered Microstructures .............................................34
3.2.1 Microstructure and hardness ...........................................................34
3.2.2 MBN measurements ........................................................................42
3.2.3 Detection of Faulty Quenching and Tempering Treatment by MBN 56
Chapter 4 Conclusions...................................................................................63 REFERENCES ...................................................................................................65
ix
LIST OF TABLES
TABLES
Table 2.1 Chemical composition of the SAE 1010 steel used (wt %)………..19
Table 2.2 Chemical composition of the steels used for quenching and
tempering (wt%)…………………………………………………………………….19
Table 3.1 Details of heat treatments applied to SAE 1010 specimens...........22
Table 3.2 Hardness values and MBN parameters of SAE 4140 specimens...47
Table 3.3 Hardness values and MBN parameters of SAE 4140 specimens...47
x
LIST OF FIGURES
FIGURES
Figure 1.1 Domains in a ferromagnetic material……….………………………...1
Figure 1.2 Structure of a domain wall...............................................................2
Figure 1.3 Colloid patterns collected over domain walls of a polycrystalline
silicon iron…………………………………………………………………………….4
Figure 1.4 Colloid patterns on cobalt with magnetic field applied normal to
surface (a) -130 oersteds; (b) +130 oersteds………………………………….....4
Figure 1.5 Domain structures on barium ferrite observed by the Kerr effect (a)
H = 2700 oersteds; (b) H = 3000 oersteds…………………………….………….5
Figure 1.6 Domains in gadolinium garnet with a field of varying strength
observed by the Faraday effect……………………………………….……………5
Figure 1.7 Series showing the process of magnetization reversal in a bloomed
film of Ni-Fe as observed by the Kerr effect......................................................6
Figure 1.8 Magnetization steps.........................................................................7
Figure 1.9 Barkhausen jumps during magnetization………..............................9
Figure 1.10 A typical MBN system configuration……………………………….9
Figure 1.11 Passage of a domain wall through an inclusion………………….10
xi
Figure 1.12 Variation of the probability of nucleation, annihilation, growth of
domains and domain wall density as a function of H…………………………11
Figure 1.13 MBN effect during hysteresis ……………………………………..12
Figure 1.14 Barkhausen event at various depths of specimen ………………13
Figure 1.15 The raw MBN data (a), MBN fingerprint (b), frequency spectrum
(c) and pulse height distribution (d).................................................................15
Figure 3.1 Optical micrographs and corresponding grain size distribution
histograms of group 1 specimens; (a) 1100A, (b) 1200A, (c) 1300A ………23
Figure 3.2 Optical micrographs and corresponding grain size distribution
histograms of group 2 specimens; (a) 1100B, (b) 1200B, (c) 1300B …...……24
Figure 3.3 Optical micrographs and corresponding grain size distribution
histograms of group 3 specimens; (a) 700A-1, (b) 700A-2, (c) 700A-3 ……...25
Figure 3.4 Effect of annealing temperature on grain configuration at various
stages of annealing…………………………………………………………………27
Figure 3.5 SEM micrograph (a) and corresponding EDX analysis taken from
(b) grain boundary and (c) grain interior of 700A-1 specimen…………………28
Figure 3.6 MBN profiles of group 1 specimens……………………………….30
Figure 3.7 MBN profiles of group 2 specimens……………………………….30
Figure 3.8 MBN profiles of group 3 specimens……………………………….30
Figure 3.9 Optical (a) and SEM micrographs (b); corresponding EDX analysis
taken from grain boundary (c) and grain interior (d) of 1300B specimen…….31
xii
Figure 3.10 The relation between RMS and AGS-0.5 …………………………. 33
Figure 3.11 SEM micrographs of quenched and tempered SAE 4140
specimens.......................................................................................................35
Figure 3.12 SEM micrographs of quenched and tempered SAE 5140
specimens.......................................................................................................36
Figure 3.13 Continuous cooling transformation (CCT) diagram and
hardenability curve of SAE 4140 steel……………………………………….…..37
Figure 3.14 Continuous cooling transformation (CCT) diagram and
hardenability curve of SAE 5140 steel……………………………………….…..38
Figure 3.15 MBN profiles of SAE 4140 specimens……………………………44
Figure 3.16 MBN profiles of SAE 5140 specimens……………………………44
Figure 3.17 The SEM micrographs of as quenched SAE 4140 (a) and SAE
5140 (b) steels and schematic representation of magnetic microstructure in
martensite........................................................................................................45
Figure 3.18 The SEM micrographs of SAE 4140 (a) and SAE 5140 (b) steels
tempered at 600oC and schematic representation of magnetic microstructure in
tempered martensite....................................................................................46
Figure 3.19 Frequency spectrums of SAE 4140 specimens…………….........48
Figure 3.20 Frequency spectrums of SAE 5140 specimens…………….........48
Figure 3.21 Pulse height distributions of SAE 4140 specimens………….…..50
Figure 3.22 Pulse height distributions of SAE 5140 specimens………….…..50
xiii
Figure 3.23 Coercivity (a), remanence (b), permeability (c) values of SAE 4140
specimens ………………………………………..…………………………..52
Figure 3.24 Coercivity (a), remanence (b), permeability (c) values of SAE 5140
specimens………………………………………….…………………………53
Figure 3.25 Hardness correlation for SAE 4140 specimens…………………..55
Figure 3.26 Hardness correlation for SAE 5140 specimens…………………..55
Figure 3.27 Continuous cooling transformation diagram of SAE 1040 steel..56
Figure 3.28 SEM micrographs taken from edge (a) and centre (b); MBN
profiles (c) and frequency spectrums (d) of as-quenched SAE 1040
specimen…………………………………………………………………………….58
Figure 3.29 SEM micrographs taken from edge (a) and centre (b); MBN
profiles (c) and frequency spectrums (d) SAE 1040 specimen tempered at
300oC……………………………………………………………..………………….59
Figure 3.30 SEM micrographs taken from edge (a) and centre (b); MBN
profiles (c) and frequency spectrums (d) SAE 1040 specimen tempered at
500oC……………………………………………………………..………………….60
Figure 3.31 SEM micrographs taken from edge (a) and centre (b); MBN
profiles (c) and frequency spectrums (d) SAE 1040 specimen tempered at
600oC……………………………………………………………..………………….61
Figure 3.32 Schematic drawing of edge and centre regions of SAE 1040
specimens………………………………………………………..………………….62
Figure 3.33 MBN profiles of edge regions of SAE 1040 specimens……….62
1
CHAPTER 1
INTRODUCTION
1.1 Theory of Magnetic Barkhausen Noise
An electrical charge in motion creates a magnetic field. Since electrons are electrical
charges their motion can produce magnetic moments in two ways.
i) Their orbital motion around nucleus generates a small magnetic field and they have
magnetic moment along their axis of rotation.
ii) The spinning of electrons also cause a magnetic moment directed along spin axis.
This moment can only be in up or down direction.
In materials having completely filled levels contribute zero magnetic moment, because pairs
of electrons in each level have opposite spin and cancel each other. Ferromagnetic materials
like iron, cobalt and nickel have occupied 4s levels leaving vacant orbits in 3d shell.
Therefore in ferromagnetic materials, instead of canceling each other, moments line up by
an exchange force of quantum mechanical nature [1]. Actually no single comprehensive
theory exists for explaining ferromagnetic behavior. Two distinct theories, band theory and
molecular field theory, try to explain the phenomenon on the basis of lowering exchange
energy. The exchange energy refers to the part of electrostatic energy of a system of
electrons which depends on the spin states of neighboring electrons. In case of
ferromagnetic materials this exchange energy is lowest when the spins of the 3d electrons in
adjacent atoms are aligned parallel [2]. Because of this ferromagnetic materials exhibit a
large spontaneous moment due to the cooperative alignment of unpaired electron spins
along a certain direction.
Figure 1. Domains in a ferromagnetic material [2]
2
Ferromagnetic materials have regions called domains where all magnetic moments are
aligned in the same direction as seen in Figure 1.1. The existence of domains is a
consequence of energy minimization. A single domain specimen would have large
magnetostatic energy due to Coulomb interaction between magnetic free poles. The breakup
of magnetization into domains provides flux closure at the ends of specimen and hence
reduces the magnetostatic energy which represents the total free energy of the domain
structure. In each domain the magnetization is equal to saturation magnetization and the net
magnetization of a material is the vector sum of the magnetization within all domains. The
direction of alignment varies randomly from domain to domain, although certain
crystallographic axes, easy directions of magnetization, are preferred especially in the
presence of a magnetic field. The internal magnetization is stable when pointing parallel to
one of these easy directions. There are transition layers in which the magnetic moments
realign between the domains and therefore belong to neither domain as seen in Figure 1.2.
These transition layers are called domain walls or sometimes referred as Bloch walls. The
interactions between the magnetic moments and the crystal lattice cause magnetization to lie
in easy directions of magnetization. The magnetocrystalline anisotropy energy is the energy
related, the dependence of internal energy on direction of inner magnetization, which is
minimum for domains located parallel to easy directions of magnetization [1, 3]. The
magnetocrystalline anisotropy energy tends to make the domain walls thinner in order to
increase the number of spins pointing in easy directions of magnetization. The exchange
energy tends to make the walls thicker since it is minimized when neighboring moments are
aligned parallel. As a result of this competition the domain wall has a certain finite width and
a certain structure [4, 5].
Figure 1.2 Structure of a domain wall [5]
3
Presence of domains has been proven by innumerable experimental observations, both
direct and indirect. The first conformation was the indirect detection of domains by the
Barkhausen effect in which the reorientation of domains caused discrete changes in
magnetic induction within a ferromagnet. Domains are normally so small that one must use a
microscope to see them directly [5]. What one sees then depends on the technique involved.
The most popular direct observation techniques fall in two groups:
1. Those which disclose domain walls (Bitter method, electron microscope). The
individual domains, whatever their direction of magnetization, look more or less the
same, but the domain walls are delineated. The Bitter or powder method involves
the application of an aqueous suspension of extremely fine (colloidal) particles of
magnetite Fe3O4 to the polished surface of the specimen and it can detect moving,
as well as stationary domain walls as shown in Figures 1.3 and 1.4. Transmission
electron microscopy, which is often called Lorentz microscopy, can disclose domain
walls in specimens thin enough to transmit electrons. The electrons passing through
the specimen will be deviated due to different orientations of local magnetization
vector across a domain wall. This method has the advantage of high resolution,
which allows the examination of the fine detail of domain structure. It also permits
the direct observation of interactions between domains, crystal imperfections and
grain boundaries.
2. Those which disclose domains (optical methods involving the Kerr or Faraday
effects). Here domains are magnetized in different directions appear as areas of
different color, and the domain wall separating them appears merely as a line of
demarcation where one color changes to the other. The Kerr effect is a rotation of
the plane of polarization of a light beam during reflection from a magnetized
specimen. The Kerr method is ideal for studies of domain walls in motion (Figures 1.5 and 1.7) and has largely supplanted the Bitter method for such studies. The
Faraday effect is a rotation of the plane of polarization of a light beam as it is
transmitted through a magnetized specimen (Figure 1.6). This method is similar to
Kerr effect but it is limited to specimens thin enough to transmit light [6].
Changes in domain structure can occur by two principal means. Either the magnetization
within each domain can coherently rotate to a direction parallel to the applied field or the
boundary between two domains can move causing the entire magnetization change to be
localized at the domain boundary. Thus magnetization changes as a result of both of domain
wall motion and domain rotation. At lower applied magnetic fields the domain walls are
stretching so that they return to the non-magnetized state on removal of the applied field. Up
4
Figure 1.3 Colloid patterns collected over domain walls of a polycrystalline silicon-iron [6]
Figure 1.4 Colloid patterns on cobalt with magnetic field applied normal to surface.
(a) -130 oersteds, (b) +130 oersteds [6]
5
Figure 1.5 Domain structures on barium ferrite observed by the Kerr effect
(a) H = 2700 oersteds, (b) H = 3000 oersteds [6]
Figure 1.6 Domains in gadolinium iron garnet with a field of varying strength observed by the
Faraday effect [5]
6
Figure 1.7 Series showing the process of magnetization reversal in a bloomed film of Ni-Fe
as observed by the Kerr effect [6]
7
to that point the wall motion is reversible. When the applied field is increased further, the
process occurring is irreversible wall motion, in other words the growth of domains which are
aligned closely parallel to applied magnetic field (H) grow at the expense of others until the
magnetic structure becomes a single domain pointing in one of easy directions of
magnetization. Still greater H makes domain rotation predominant and in this region work
must be done against anisotropy forces; a rather large increase in H is required to produce a
relatively small increase in magnetization [5, 7]. Figure 1.8 shows these steps of
magnetization.
Figure 1.8 Magnetization steps [7]
During the magnetization process imperfection like dislocations or impurity elements in the
metal cause an increase in the energy lost in the form of a kind of internal friction. These
imperfections give rise to hysteresis. Also higher magnetocrystalline anisotropy gives rise to
hysteresis. Many ferromagnetic materials can be characterized from parameters obtained
from hysteresis curves. These parameters are:
8
• Permeability (µ): Measure of the degree to which the material can be magnetized.
The relation between magnetic induction (B) and applied magnetic field strength (H)
is as follows:
HB .µ= (1.1)
• Remanence: When the field is reduced to zero after magnetizing a magnetic
material the remaining magnetic induction is called remanent induction (BR) and the
remaining magnetization is called the remanent magnetization (MR).These are
related by the following formula:
RoR MB .µ= (1.2)
where µ0 is the permeability of vacuum. The remanence is used to describe the
value of either remaining induction or magnetization when the field has been
removed after the material has been magnetized to saturation. The remanent
induction or magnetization is used to describe the remaining induction or
magnetization when the field has been removed after magnetizing to an arbitrary
level.
• Coercivity: By applying a reverse magnetic field strength of Hc the magnetic
induction declines to zero. The coercive field is the magnetic field needed to reduce
the magnetization to zero from an arbitrary level whereas coercivity is the magnetic
field required to reduce magnetization to zero from saturation.
• Saturation magnetization (Ms): As H is increased the magnetization finally reaches
saturation at value designated Ms. At saturation all the magnetic moments are
aligned in the direction of applied magnetic field.
All ferromagnets when heated to sufficiently high temperatures become paramagnetic due to
the increased randomness of atomic moment. This transition temperature, above which the
thermal energy overcomes the exchange forces, is called the Curie temperature. At this
temperature the permeability of the material drops suddenly and both coercivity and
remanence become zero. The Curie temperature of iron is 770oC.
When a ferromagnetic material is magnetized, changes in physical dimensions, in general
occur. These changes could be longitudinal, transverse, or volumetric and are known as
magnetostriction. The dimensional change occurring along the direction of induced magnetic
field is called Joule effect magnetostriction and the converse is known as the Villari effect. [7]
9
Figure 1.9 Barkhausen jumps during magnetization [4]
High resolution examination of magnetization cycles of ferromagnetic materials reveals
discontinuous flux changes as shown in Figure 1.9. This effect is discovered by Prof.
Barkhausen in 1919 and named Barkhausen effect. During magnetization if a search coil is
placed close to the surface of the specimen and connected to an oscilloscope or computer
as shown in Figure 1.10, voltage spikes can be observed. These voltage spikes are known
as Magnetic Barkhausen Noise (MBN). At first these discontinuities in magnetization was
attributed to sudden discontinuous rotation of domain but now it is known that discontinuous
domain wall motion is the most significant factor. In fact both of these mechanisms occur and
contribute to MBN [8].
Figure 1.10 A typical MBN system configuration [9]
High Resolution Examination
10
When a domain wall bisects an inclusion, magnetostatic energy can be reduced to zero, at
the cost of a little wall energy, if closure domains form as shown in figure 1.11(a).
Observation of moving domain walls in crystals has shown that domain walls are pinned by
interaction of the moving wall with the spike domains attached to inclusion rather than by interaction of the inclusions themselves. Figure 1.11(a-d) shows the passage of a domain
wall through an inclusion. In response to an upward applied field, the wall in (a) moves to
right, as seen in (b) dragging out the closure domains into the form of tubes and creating a
new domain just to the right of the inclusion. Further motion of the main wall lengthens the
tubular domains as in (c). The change from (a) to (b) to (c) is reversible and the domain
arrangement (a) can be regained if the field is reduced. But if the field is increased further
the tubular domains do not continue to lengthen infinitely because their increasing surface
area adds two much wall energy to the system. A point is reached when the wall suddenly
snaps of the tubular domains irreversibly and jumps a distance to the right, leaving two spike domains attached to the inclusion as in (d) [5]. This is a Barkhausen jump and detected in
the form of voltage pulses induced in a search coil positioned close to the specimen surface.
Not only inclusions but also other microstructural features such as dislocations and grain
boundaries can cause Barkhausen jumps by pinning domain walls.
Figure 1.11 Passage of a domain wall through an inclusion [5]
During MBN measurements representative magnetic hysteresis loop is induced in the small
volume due to the energy loss with the irreversible process of magnetization. This
irreversible process mentioned above is strongly related to the dynamic behavior of domains,
i.e., nucleation, annihilation and growth of domains. Grain/lathe boundaries, dislocations and
precipitates affect this dynamic behavior. A schematical illustration in Figure 1.12 shows the
variation of the probability of the nucleation, annihilation and growth of domains as well as
the domain wall density as a function of H. The given magnetization loop is divided into two
parts, namely Region 1 and 2 as seen in Figure 1.12(a).
11
Figure 1.12 Variation of the probability of nucleation, annihilation, growth of domains and
domain wall density as a function of H [10].
When existing field strength is reduced from saturation magnetization (Region 1), new
domains are nucleated close to precipitates and grain or lathe boundaries where magnetic poles are accumulated during the spin rotation. The domain nucleation can proceed when
the reduction in the magnetostatic energy associated with the poles during the formation of
new domains is greater than the work required to form domains. The probability of the
domain nucleation, Pn, increases with decreasing H in order to reduce the magnetostatic
energy associated with the poles as shown in Figure 1.12(b). Thus, the density of domain
walls, Fd, increases with decreasing H as Figure 1.12(c) shows. The driving force for the
growth of domains arises from the difference between the applied field energy and the
domain wall energy. Thus, the probability for the growth of domains, Pg, decreases with
decreasing H as the wall energy of the domain preferentially oriented to the H direction decrease. When a magnetic field in the reverse direction increases beyond Hc (Region 2),
domains formed in region 1 grow at the expense domains aligned unfavorably. This growth
causes annihilation of unfavorably aligned domains and results in a decrease of Fd. Thus,
the domain wall density, Fd, is highest around H=0. As the driving force for the domain wall
propagation increases with increasing H, Pg as well as the probability of domain annihilation,
Pa, increases as shown in Figure 1.12(b) [10].
12
The total angular displacement across a domain wall is often 180o or 90o, particularly in cubic
materials because of the anisotropy and the change in direction of moments take place
gradually over many atomic planes. MBN is principally caused by the motion of 180o walls.
At the same time non-180o domain wall motion will have to occur due to presence of closure
domains. However, contributions of the non-180o domain walls to MBN are smaller than
those of the 180o walls for two reasons:
i) The average velocity of 180o walls are larger than that of the non180o walls,
ii) The volume swept out by 180o walls is larger than that by non-180o walls [10]
The Barkhausen effect is strongest on the steepest part of the magnetization curve as shown
in Figure 1.13. MBN is sensitive to changes in mechanical stress, composition and
microstructural features such as grain size, inclusions, precipitates, dislocations. Because of
the number of influential variables, for material characterization this technique makes relative
comparisons between material states.
Figure 1.13 MBN effect during hysteresis [11]
Eddy currents arise in any conducting material in which magnetization is changing and flow
in such a direction as to produce magnetic fields opposing the change [12]. Eddy currents
become more significant as magnetizing frequency is increased. The MBN signal is
13
attenuated by this eddy current opposition (Figure 1.14). Thus we have a limited
measurement depth for MBN measurements given by the formula 1.3:
0....1
µµσπδ
rf= (1.3)
where
σ = conductivity
f = frequency content of MBN
µr = relative permeability
µo = permeability of vacuum
The MBN signal has frequency contents up to 2 MHz and measurement depths for practical
applications vary between 0.01 to 1.5 mm.
Figure 1.14 Barkhausen event at various depths of specimen [13]
For MBN method typical measurement parameters are as follows:
• Magnetizing frequency (Hz): Adjusts the frequency of magnetizing field applied to
the specimen. When magnetizing frequency is increased and magnetizing voltage
kept constant, magnetization level will decrease. On the other hand the magnetic
noise level, including peak heights and root-mean-square (RMS) voltages, will
increase due the fact that much more domain walls are participated during
magnetization at higher frequencies.
14
• Waveform: Different waveforms have been used for the alternating excitation field to
obtain MBN signals. Triangular and sinusoidal waves are used commonly whereas
square waves rarely. Similar MBN signals obtained when triangular or sinusoidal
waveforms are used.
• Magnetizing voltage (V): Adjusts the magnitude of magnetizing field applied to the
specimen. The actual level of magnetization depends on the sensor used and can
be checked. Increasing magnetizing voltage increases the level of magnetization.
• Magnetizing field strength (H): The field strength can be varied by changing the
magnetizing frequency and voltage simultaneously. In some sources the field
strength is preferred instead of magnetizing voltage for MBN signal analysis. The
signal level increases to a maximum at an intermediate field strength and then
decreased at higher fields. The increase was attributed to greater capacity for
overcoming pinning obstacles when the field is getting larger, and the decrease to
the predominance of domain rotation over domain wall motion at very high fields.
• Number of bursts: Determines how many magnetizing half cycles or Barkhausen
Noise bursts will be stored for signal analysis. Increasing the number of bursts
makes results more reliable whereas increases data analysis time.
• Sampling frequency: Determines how many samples per second are stored for
signal analysis. The sampling frequency is adjusted by regarding the magnetizing
and analyzing frequencies. The sampling frequency should be at least twice the
maximum analyzing frequency for consistency.
The raw Magnetic Barkhausen Noise data consists of series of voltage pulses and their
associated applied field values obtained as a function of time shown in Figure 1.15(a). The
raw MBN signal may be amplified and filtered for detailed analyses. By filtering the noise
with a band-pass filter the background noise can be eliminated and measurement depth can
be varied according to the formula 1.1. After being amplified and filtered the signal is ready
to be analyzed by the following methods and parameters:
• MBN fingerprints: Using a definite sampling frequency, a local root-mean-square
(RMS) value is calculated and plotted against applied field strength. This
instantaneous local value is averaged over several field cycles whose number is
determined by the number of bursts. Also a smoothing algorithm may be applied and
a fingerprint as shown schematically in Figure 1.15(b) is obtained. The fingerprints,
sometimes referred as MBN envelopes or profiles, are characterized by the
maximum noise amplitude referred as MBN peak height and the corresponding
magnetic field referred as peak position.
15
Figure 1.15 The raw MBN data (a), MBN fingerprint (b), frequency spectrum (c) and pulse
height distribution (d).
• Frequency spectrum: The frequency content of the noise can be obtained by using
Fourier analysis. In the literature a plot such as Figure 1.15(c) is referred as
frequency spectrum where y-axis is the amplitude or noise power (V2).
• Pulse height distribution: Another plot frequently encountered in literature is the
pulse height distribution which gives the size distribution of pulses as shown in
Figure 1.15(d) schematically.
• Representative B-H curves: The Barkhausen Noise is derived from magnetization
cycles. Thus the sum integral of rectified bursts gives a simulation of the hysteresis
loop. From that simulation coercivity, remanence and permeability can be calculated.
It should be stated that all these calculated parameters from the simulated
hysteresis can not be used as true values because the saturation magnetization of
the specimens are far beyond the actual level of magnetization reached locally
during MBN measurements.
In addition, single parameters such as total number of pulses, maximum pulse size and root
mean square (RMS) of all signal amplitudes can be used to characterize the noise signal.
16
1.2 Literature Survey Magnetic Barkhausen Noise (MBN) technique has proved its viability for characterization of
microstructures and it is considered as a valuable non-destructive evaluation (NDE)
technique for microstructural characterization of ferromagnetic materials. The dual sensitivity
of the phenomenon to stresses and to microstructure on which this study is focused, gives a
wide range of potential applications to the technique including determination of grain size of
steels.
1.2.1 Determination of Grain Size of Steels by MBN
S. Titto et al. studied non-destructive measurement of grain sizes of 500 samples of 0.17 –
0.37mm low alloy low carbon steel in soft annealed and temper annealed condition. Grain
sizes in the range of 5 – 25 µm were studied and a linear relation was found between grain
size and the amplitude of the MBN signal [14].
In a study of R. Ranjan et al. the peak heights and RMS values of the Barkhausen signal
was found to be increased with increasing grain size. Specimens of decarburized steel were
undergone the following sequence of hot rolling, cold rolling, annealing, temper rolling and
decarburization annealing (%C < 0.0005 wt%) and grain sizes of 70 – 120 µm were obtained
[15].
Another study investigated the influence of applied tensile stress and grain size on MBN in
SAE 1005 steels. Specimens having ferrite grain sizes of 20 – 45 µm were prepared by
applying variations of furnace and air cooling. The MBN peak heights decreased as ferrite
grains get coarser [16].
C. Gatelier-Rothea et al. made a similar study using ultrahigh purity iron with less than 20ppm impurities. By annealing the samples at 450oC, 600oC and 750oC for 3 hours
equiaxial grains of 50 – 300 µm were obtained. The results were similar to [16]; peak heights
decreased when grain size increased [17].
The effects of grain size and grain boundary misorientation were studied in pure iron
prepared by vacuum melting. Samples containing 50 -180 µm grains were prepared by
annealing at various temperatures between 1193 and 1393 K. Hall-Petch like relationship
was found between the grain size and RMS values. Moreover the frequency spectrums of
the MBN signals analyzed and the grain size was related to the ratio of two definite
frequency components. The RMS values and the ratio between the definite frequency
components decrease as grain size increases [18].
17
R. Ranjan used pure nickel samples, whose impurity content is less than 75ppm, instead of
iron or steel to investigate the effect of grain size on MBN. Samples composed of 20 – 240
µm grains were obtained by annealing at different temperatures between 500oC – 800oC for
three hours. The MBN fingerprints showed two peak behavior and the ratio between the
second and first peak heights decreased as grains became coarser [10].
Sakamoto et al. analyzed MBN signals theoretically and relate the RMS values of MBN to
grain size as follows:
2/1. −= dCRMS g (1.4)
In this Petch like relation “d” is the average grain size; Cg is independent of the grain size
and given by:
( ) ( )2
4/1
4/5
max
3.9.
28/
⎟⎟⎠
⎞⎜⎜⎝
⎛∆Φ=
KItC
NHdtdHC
S
totalvg
γπ
(1.5)
where H is the magnetic field strength, N is the total number of Gaussian pulses in the cross-
sectional unit area of a specimen, ∆Φ is the quantity of magnetic flux change in a
microregion, Cv is a constant, ttotal is the total time of generation of Gaussian pulses, γ is the
wall energy per unit area, Is is the saturation magnetization and K is the magnetic anisotropy
constant. In this study voltage pulses in each micro-region, that MBN is composed of, are
approximated by Gaussian pulses in order to facilitate the mathematical treatment [19].
1.2.2 Characterization of Quenched and Tempered Steel Microstructures by MBN
Blaow et al. used Ovako 677 steel to study effects of tempering on MBN. The specimens
were austenitised at 950oC for 1hour, followed by air cooling. The martensite structure
produced was tempered at 180oC and 400oC. The profiles obtained from MBN
measurements were characterized by peak height, peak position and area under the profile.
The differences between profiles of tempered structures were not significant [11].
J. Kameda used AISI 4340 steel austenitised at 850oC for 1 hour, oil quenched and
tempered at various temperatures between 100oC and 500oC for 1 hour. The peak height of
the MBN signal was sensitive to microstructure change induced during tempering [20].
18
O. Saquet et al. studied tempering induced changes on MBN, using water quenched XC 55
steel austenitised at 875oC and tempered at various temperatures ranging from 100oC to
600oC. The MBN fingerprints showed that peak heights increased and the peak positions
shifted to lower fields as tempering temperature increased. In addition a simple model was
proposed for the source mechanism of MBN signals [21].
Moorthy et al. used MBN to characterize the microstructures of water quenched 0.2%C steel
solutionised at 950oC for 1 hour and tempered at 600oC for 0.5 – 100 hour. Peak heights and
positions of MBN fingerprints were used to differentiate various carbide size distributions. A
single peak was observed in samples tempered for 0.5 and 1 hour whereas after 5 hour
tempering MBN fingerprints showed a clear two peak behavior [22].
The effect of tempering was also studied for case carburized EN 36 steel in another study by
Moorthy et al. The specimens were case carburized to produce a surface carbon content of
0.85%C and a case depth of 1mm followed by oil quenching from 900oC. The specimens
were tempered at 192oC for 2 hour and at 250oC for 4 hour. Effect of tempering was studied
using a range of magnetizing frequencies and a number of analyzing frequency ranges for
characterizing the MBN signal. A correlation between hardness depth profile and peak height
of the MBN fingerprint was found [23].
In another study 12% Cr Mo V stainless steel, which was solutionised in the range of 950oC
– 1150oC for 1 hour, was used to investigate the effects of tempering on MBN. The
solutionised samples were then air cooled and tempered in the range of 200oC -800oC for 1
hour. The noise energy and number of MBN pulses were the parameters used for signal
analysis. No significant change was observed in MBN signal due to tempering up to 500oC;
whereas increasing tempering temperature further caused a rapid increase in both of the
parameters [24].
1.3 The Aim of the study
Steel is the most important and widely used industrial commodity. Microstructure affects the
properties, especially mechanical properties, of steels. New designs require steels providing
longer service life with higher performance which makes quality control essential. Increasing
demand brings a growing need for non-destructive inspection of steel components. Magnetic
Barkhausen Noise (MBN) measurement provides a good alternative to traditional methods in
terms of fastness and accuracy. Nevertheless the MBN method has not yet been fully exploited
when compared with some other non-destructive methods ultrasound. The aim of this study is
non-destructive characterization of steel microstructures by Magnetic Barkhausen Noise method.
19
The aim of the first part of this study is determination of grain size of steels by MBN method.
Average ferrite grain size is one of the most important microstructural features that affect the
properties of steel components. Annealing is a widely used industrial process in order to
refine the grain, induce softness, improve electrical and magnetic properties, and improve
machinability as well. Annealing consists in heating the steel to the proper temperature and
then cooling slowly through the transformation range in the furnace up to low temperatures.
Annealing may be divided into 3 stages: recovery, recrystallization and grain growth. Various
grain sizes and distributions may be obtained by annealing due to differences in degree of
prior deformation, impurity content of steel, annealing time and temperature. Traditional
metallographic and mechanic methods, that involve taking representative specimens, can
not allow 100% inspection of steel work pieces. Besides being destructive and time
consuming makes these traditional methods too slow for present production rates. Instead of
traditional methods, MBN measurements may be used to determine grain size quickly, easily
and without damaging the material.
The aim of the second part of this study is characterizing the microstructures of quenched
and tempered steels by MBN method and correlating the hardness of these steels with MBN
findings. Steels are widely utilised in different industries, usually in the form of quenched and
tempered components. Any temperature up to the lower critical may be used for tempering;
thus an extremely wide variation in properties and microstructure ranging from those of as
quenched martensite to spheroidized carbides in ferrite can be produced by tempering.
Ultimately it is the balance of hardness (or strength) and toughness required in service that
determines the conditions of tempering for a given application [25]. If the principal desired
property is hardness or wear resistance, the part is tempered at about 200oC; if the primary
requirement is toughness, the part is then tempered above 400oC. Residual stresses are
relieved almost completely when the tempering temperature reaches 500oC. For consistency
and less dependence on time, quenched steel components generally tempered for 1 to 2
hours [26]. The industry has been searching for methods capable of characterizing material
properties accurately, quickly and without damaging the material. The MBN technique is
considered here a candidate method for such mechanical and microstructural
characterizations.
20
CHAPTER 2
EXPERIMENTAL
2.1 Experimental Procedure for Grain Size Determination 8 mm specimens were cut from cold drawn 50 mm diameter SAE 1010 steel. Table 2.1
shows the composition of the steel used. All the machining operations were done before the
heat treatments in order to avoid surface residual stresses.
Table 2.1 Chemical composition of the SAE 1010 steel used (wt%)
Steel Type C Cr Mo Mn Si P S Fe
SAE 1010 0.113 0.073 0.039 0.503 0.216 0.006 <0.001 Bal.
One group of specimens was annealed at 700oC for 2, 6, 16 and 24 hours. Another group
was austenitized at 900oC, 1000oC, 1100oC, 1200oC and 1300oC for 30 and 90 minutes
followed by furnace cooling.
For metallographic investigation the samples were finely ground and polished with diamond
paste. In order to reveal ferrite grains with enhanced contrast, color etching was used. After
polishing, the specimens were etched first by 5% Nital followed by bisulfate. The surfaces of
specimens were examined under optical microscope and scanning electron microscope.
Average ferrite grain size and grain size distributions for each specimen were obtained by
analyzing the photographs of examined surfaces using Clemex Image Analyzer software.
MBN measurements were performed using a commercial system (Rollscan, µscan 500-2). The
sensor S1-138-13-01 was used for the MBN measurements. A sinusoidal cyclic magnetic field
with an excitation frequency of 10 Hz was induced in a small volume of the specimen via a
ferrite core C-coil. The Barkhausen signals were filtered with a wide band-pass filter (1-200
kHz), amplified with a gain of 50 dB, and then, analyzed using the Rollscan-software. The peak
magnetizing voltage was 15V and sampling frequency was 2 MHz.
21
Before MBN measurements, in order to eliminate the effects of remanent magnetization on
results, all the specimens were passed through demagnetization tunnel. Zero remanent
magnetization for all specimens was ensured by Gauss-meter measurements.
2.2 Experimental Procedure for Investigation of Quenched
and Tempered Samples The specimens of 7 mm-thick and 22 mm diameter were prepared from the hot rolled SAE
4140 bar and 15 x 15 x 7 mm specimens from SAE 5140 bar. In order to investigate the
effects of low hardenability, specimens of 15 x 15 x 7 mm were prepared from SAE 1040
bar. Table 2.2 gives the chemical composition of the steels used. All the cutting and grinding
operations were done prior to the heat treatments in order to avoid surface machining
residual stresses. Austenitization was done under controlled atmosphere to avoid oxidation
and decarburization. All specimens were quenched in water after austenitization at 860oC for
30 minutes. Then, specimens were separately tempered at 200oC, 300oC, 400oC, 500oC and
600oC for 90 minutes. One specimen from each type of steel was left as-quenched.
Table 2.2 Chemical compositions of the steels used for quenching and tempering (wt%)
Steel Type C Cr Mo Mn Si P S Fe
SAE 1040 0.416 0.233 0.047 0.800 0.423 0.020 <0.001 Bal.
SAE 4140 0.475 0.942 0.224 0.840 0.202 0.023 0.015 Bal.
SAE 5140 0.491 1.143 0.053 0.730 0.312 0.016 0.047 Bal. Before metallographic investigation, the samples were finely ground, polished with diamond
paste and etched with 2% Nital. The through-thickness sections of the specimens were
examined using optical microscope and scanning electron microscope. For each specimen
an average hardness value was determined by measuring Vickers hardness at different
locations.
During MBN measurements a sinusoidal cyclic magnetic field with an excitation frequency of
125 Hz was induced in a small volume of the specimen via a ferrite core C-coil. The
Barkhausen signals were filtered with a wide band-pass filter (0.1-1000 kHz), amplified with
a gain of 20 dB, and then, analyzed using the Rollscan-software. The peak magnetizing
voltage was 10V and sampling frequency was 2 MHz. During the analyses average of 154
bursts were used to obtain Barkhausen parameters for each specimen where each burst
represents one half of the magnetization cycle.
22
CHAPTER 3
RESULTS
3.1 Grain Size
3.1.1 Microstructure
Figure 3.1, 3.2 and 3.3 show the optical micrographs and related grain size distribution
histograms of annealed SAE 1010 specimens. In order to enhance the contrast between
ferrite grains color metallographic techniques were used. Average grain sizes (AGS) and
size distribution histograms were obtained from direct measurement of about 100 grains per
specimen. The histograms indicate that the maximum distribution percentages are around
the average grain size value. Table 3.1 shows the details of heat treatments applied and the
corresponding average grain sizes (AGS) obtained for every SAE 1010 specimen. By
changing the time or temperature of annealing treatment, specimens composed of various
grain sizes are obtained.
The term annealing has been used in its broadest sense to refer to any heat treatment that
has as its objective the development of a nonmartensitic microstructure of low hardness and
high ductility. This understanding of annealing is much too broad, however, and a number of
Table 3.1 Details of heat treatments applied to SAE 1010 specimens
Specimen Code
Annealing Temperature
Annealing time
AGS (µm)
1100A 1100oC 30 min 40,2 ± 22
1200A 1200oC 30 min 54,1 ± 27
Gro
up 1
1300A 1300oC 30 min 58,2 ± 32
1100B 1100oC 90 min 40 ± 20
1200B 1200oC 90 min 47,5 ± 26
Gro
up 2
1300B 1300oC 90 min 63 ± 32
700A-1 700oC 2 hour 22 ± 10
700A-2 700oC 6 hour 24,1 ± 10
Gro
up 3
700A-3 700oC 24 hour 26,7 ± 11
23
Group 1 Specimens
0
5
10
15
20
25
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
Grain Size (mm)
Perc
enta
ge (%
)
0
5
10
15
20
25
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
Grain Size (mm)
Perc
enta
ge (%
)
0
5
10
15
20
25
0 10 20 30 40 50 60 70 80 90 100
110
120
130
140
150
160
170
180
Grain Size (mm)
Perc
enta
ge (%
)
Figure 3.1 Optical micrographs and corresponding grain size distribution histograms of group 1 specimens, (a) 1100A, (b) 1200A, (c) 1300A.
c) 1300A; AGS = 58,2 µm
b) 1200A; AGS = 54,1 µm
a) 1100A; AGS = 40,2 µm
24
Group 2 Specimens
0
5
10
15
20
25
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Grain Size (mm)
Perc
enta
ge (%
)
0
5
10
15
20
25
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Grain Size (mm)
Perc
enta
ge (%
)
0
5
10
15
20
25
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Grain Size
Perc
enta
ge (%
)
Figure 3.2 Optical micrographs and corresponding grain size distribution histograms of group 2 specimens, (a) 1100B, (b) 1200B, (c) 1300B.
c) 1300B; AGS = 63 µm
b) 1200B; AGS = 47,5 µm
a) 1100B; AGS = 40 µm
25
Group 3 Specimens
0
5
10
15
20
25
30
0 5 10 15 20 25 30 35 40 45 50 55 60
Grain Size (mm)
Perc
enta
ge (%
)
0
5
10
15
20
25
30
0 5 10 15 20 25 30 35 40 45 50 55 60
Grain Size (mm)
Perc
enta
ge (%
)
0
5
10
15
20
25
30
0 5 10 15 20 25 30 35 40 45 50 55 60Grain Size (mm)
Perc
enta
ge (%
)
Figure 3.3 Optical micrographs and corresponding grain size distribution histograms of group 3 specimens, (a) 700A-1, (b) 700A-2, (c) 700A-3.
c) 700A-3; AGS = 26,7 µm
b) 700A-2; AGS = 24,1 µm
a) 700A-1; AGS = 22 µm
26
more specific annealing heat treatments have been developed and defined. Full annealing is
a heat treatment accomplished by heating steels into a single-phase austenite field and
slowly cooling, usually in a furnace, through the critical transformation ranges [25]. Group 1
and 2 specimens were full annealed at various temperatures as shown in Table 3.1.
The optical micrographs of Group 1 (Figure 3.1) and Group 2 (Figure 3.2) indicates that
specimens are composed of equiaxed ferrite grains with various AGS. The formation of pro
eutectoid products like proeutectoid ferrite, has all the manifestations of a process of
nucleation and growth; the nuclei appear at the austenite grain boundaries and grow until a
continuous layer is formed along the grain boundary. Further growth merely thickens the
grain boundary and at very low contents, this continues until all the austenite is transformed.
The transformation of austenite into proeutectic ferrite is a good example of heterogeneous
nucleation and occurs preferentially at grain boundaries. Any steel with fine austenite grains
has greater grain boundary surface area, consequently favors nucleation and reduces
incubation period [27]. Thus, with increasing austenite grain size parallel to the annealing
temperature, average grain size of proeutectic ferrite increases. The increase in austenite
grain size, which in turn affects ferrite grain size, is controlled by a time and temperature
dependent growth mechanism. The comparison of Group 1 and 2 specimens show that the
effect of temperature on growth of austenite is more dominant as expected.
Group 3 specimens were annealed below the eutectoid temperature; a treatment named as
process and recrystallization annealing. Prior to heat treatments the samples were in cold
deformed condition and this type of annealing is usually applied to soften and restore
ductility of cold worked steel products. Most of the energy expended in cold work is released
as heat during deformation. However, a small percent is stored by dislocations and some
other crystal imperfections introduced during deformation. The stored energy is the driving
force for the changes during annealing. On heating, high strain energy of the deformed
ferrite first drives recovery, a mechanism by which some of the crystal imperfections are
eliminated or rearranged into new configurations. Then ferrite grains having high dislocation
density are replaced by new grains having much lower dislocation density. The decrease of
energy associated with dislocations is the driving force for recrystallization. Eventually
recrystallized grains grow at the expense of other recrystallized grains. The driving force for
this grain growth is the reduction of energy associated with grain boundaries [28]. Various
grain sizes and distributions may be obtained by annealing due to differences in degree of
prior deformation, impurity content of steel, annealing time and temperature. The effects of
prior deformation and impurities were eliminated by using the samples from the same cold
deformed SAE 1010 steel bar. Figure 3.4 shows the effect of temperature on the grain
configuration at different stages of annealing for constant specified time period. The
comparison of Group 3 optical micrographs with Figure 3.4 indicates that the temperature
27
chosen was enough to recrystallize and grow all the grains. By changing the annealing time
differences in growth of domain and hence samples composed of various AGS were obtained.
However, impurity segregation at grain boundaries pinned the growing grains. Figure 3.5 shows the SEM micrograph and corresponding EDX analysis. The white regions in the SEM
micrograph are impurities and corresponding EDX analysis shows that sulphur (S)
concentration at grain boundary is much higher than grain interior. Thus, the difference
between AGS of specimens was not significant.
Figure 3.4 Effect of annealing temperature on grain configuration at various stages of annealing [26] In order to lessen the effects of pearlite, SAE 1010 low carbon steel was used. Pearlite
amount calculated from equilibrium iron-cementite phase diagram is about 12%. All the heat
treatments were done in a decarburizing atmosphere; thus much less pearlite is present in
the resulting microstructures. All the specimens are composed of nearly 100% ferrite so the
effects of pearlite on MBN signals are neglected.
The hardness measurements indicate that all specimens have close hardness values about
100 HV. Previous studies and second part of this study report that MBN signals are sensitive to
hardness changes. On the other hand the hardness difference between SAE 1010 steels in
this study is small enough to be negligible. In addition all specimens were cooled in furnace
which eliminates the effects of residual stresses. Regarding these, the differences between
MBN signals is due to the variations in AGS of ferrite.
28
b) Grain Boundary EDX-Analysis = 1.44% S + 98.56% Fe (wt.%)
c) Grain interior EDX-Analysis = 100% Fe (wt.%)
Figure 3.5. SEM micrograph (a) and corresponding EDX analysis taken from (b) grain boundary and (c) grain interior of 700A-1 specimen
a)
29
3.1.2 MBN Measurements Figure 3.6, 3.7 and 3.8 show the MBN profiles of group 1, 2 and 3 specimens respectively.
The peak heights of group 1 profiles show a clear difference between the specimens.
However such clear difference does not exist for the group 2 specimens. Group 3 specimens
have overlapping profiles since their grain size difference is not so significant. For all groups
of specimens differences in grain size does not affect the peak positions of profiles.
In the introduction part it was mentioned that the MBN signal is generated due to sudden
changes in magnetization and the irreversible motion of 180o domain walls is the main
contribution. Thus the MBN signal can be written as:
dHBHVHcMBN oo .).().(. 180180
∆= ∫ ρ (3.1)
where “c” is a constant which depends on the search coil, the permeability and conductivity
of the sample; “ )(180
Hhρ ” is the density of 180o domain walls at a field H; “ )(180 HV o ” is the
average critical velocity of a 180o domain wall when it is released from pinning sites; and
“ B∆ ” is the average change in the local magnetism due to unit displacement per unit area of
domain walls. As the grain size increases in a ferromagnetic material, the domains get larger
[29]; thus )(180
Hhρ decreases. Moreover, with increasing grain size, the mean free path of
domain wall motion increases. Consequently, the incremental field (∆H) required for bulging
the domain walls before they are unpinned also increases. Thus )(180 HV o decreases with
increasing grain size. As a result of this it is expected that the amplitude of MBN signals
decrease with increasing grain size [10].
The theory above may not be true in the presence of grain boundary precipitates which
increase the net magnetostatic energy and alter the domain structure. The precipitates may
affect the MBN signals in two ways:
i) They can act as nucleation sites when domains are nucleated during magnetization. As
the density of precipitates increases )(180
Hhρ will increase too.
ii) When the density of precipitates increases, the spacing between neighboring
precipitates will decrease, and hence )(180 HV o will increase.
With the exception of 1300B specimen of group 2, the MBN response of specimens is
consistent with theory described above. For all specimens peak positions of profiles are
30
0
2
4
6
8
10
12
14
-100 -50 0 50 100
Magnetic Field Strength (% of max.)
Avg
. MB
N le
vel (
% o
f 5V)
1100 A1200A1300A
Figure 3.6 MBN profiles of group 1 specimens
0
2
4
6
8
10
12
14
-100 -50 0 50 100
Magnetic Field Strength (% of max.)
Avg
. MB
N le
vel (
% o
f 5V)
1100B1200B1300B
Figure 3.7 MBN profiles of group 2 specimens
0
2
4
6
8
10
12
14
-100 -50 0 50 100
Magnetic Field Strength (% of max.)
Avg
. MB
N le
vel (
% o
f 5V)
700A-1700A-2700A-3
Figure 3.8 MBN profiles of group 3 specimens
31
c) Grain Boundary EDX-Analysis = 1.37% S + 98.63% Fe (wt.%)
d) Grain interior EDX-Analysis = 100% Fe (wt.%)
Figure 3.9. Optical (a) and SEM micrographs (b); corresponding EDX analysis taken from grain boundary (c) and grain interior (d) of 1300B specimen
a) Optical Microscope b) SEM
32
nearly same and are around zero magnetic field strength indicating MBN activity occurs at
very early stages of magnetization. During these very early stages domain nucleation is the
dominant mechanism. Grain boundaries have high internal energy due to unsatisfied atomic
bonding, which makes them preferential sites for domain nucleation. Grain boundary area
per unit volume decreases as grain size increases, which in turn reduces the potential
nucleation sites and causes difficulty in formation of new grains. During later stages of
magnetization the domains grow and eventually rotate as the material reaches magnetic
saturation.
The specimens contain practically 100% ferrite and in such ferritic structures domains can
move freely in ferrite grains until they face a grain boundary. These pinned domain walls at
grain boundary can continue their motion by Barkhausen jumps which in turn generates
magnetic noise signals. As grains coarsen Barkhausen jumps lessen due to reduced grain
boundary density. As a result, difficulties in domain nucleation, reduced number of
Barkhausen jumps and domain density cause low MBN activity in coarse grained structures.
The specimen 1300B of group 2, although coarse grained, exhibits unexpectedly high MBN
activity. Careful examination of optical and SEM micrographs revealed segregation of
impurities at grain boundaries (Figure 3.9). These impurities alter the magnetic structure and
cause an increase in domain wall density and in critical velocity of a domain wall as
explained above. Hence the MBN activity of this sample is unexpectedly high.
Another parameter that reflects the MBN behavior and influenced in the same way as peak
height of profile, is the root-mean-square (RMS) of noise signals. A theoretical study found a
Petch like relation between RMS and AGS:
5.0).( −= AGSkRMS (3.2)
where “k” is a constant and does not depend on AGS [19]. The details of this equation can
be found in “Literature Survey part”. Figure 3.10 compares the theoretical and experimental
results of group 1. The experimental results are consistent with the theoretical result which
suggests a linear relation between RMS and square root of AGS. The correlations of group 2
and 3 specimens are not shown since the magnetic structures of those specimens are
altered by grain boundary segregations as shown in Figures 3.5 and 3.9. In literature good
correlations between MBN and AGS was found in purer steels. Better correlations may be
obtained by using steels with lower impurity concentrations and hence preventing grain
boundary segregations.
33
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0.1 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.2
AGS-0.5
RM
S (V
)
Theoretical [19]Experimental
Figure 3.10 The relation between RMS and AGS-0.5
34
3.2 Quenched and Tempered Microstructures
3.2.1 Microstructure and hardness Representative SEM micrographs (Figure 3.11and 3.12) and hardness values (Table 3.2
and Table 3.3) of the SAE 4140 and SAE 5140 show that typical martensitic and tempered
martensitic structures were successfully obtained.
The as-quenched specimens of both steels are the hardest of all specimens due to the fact
that their martensitic structure has a tetragonal lattice with interstitial carbon in solid solution
and high dislocation density formed by shear. Unlike ferrite or pearlite martensite forms by a
sudden shear process in the austenite lattice which is not accompanied by atomic diffusion.
Ideally the martensite reaction is a diffusionless shear transformation, highly crystallographic
in character, which leads to a characteristic lath or lenticular microstructure. The shear
involved in martensitic transformation cause severe elastic strains in both the martensite and
surrounding austenite matrix. The volume of elastically strained material is greatly reduced if
the shape of the martensite is lath or lenticular [28]. Also should be stated that the terms lath
and plate refer to the three dimensional shapes of individual martensite crystals. During
metallographic examinations the cross sections will appear to be needlelike or acicular [25].
Martensite is a supersaturated solid solution of carbon in ferritic iron. The carbon atoms tend
to order in such a way that the crystal structure changes from body-centered cubic to body
centered tetragonal. The tetragonality is measured by the ratio between the axes, c/a, and
increases with carbon content. The tetragonality of martensite arises as a direct result of
interstitial solution of carbon atoms in the bcc lattice, together with the preference for a
particular type of octahedral site imposed by the diffusionless character of the reaction [30]. In short the combined effects of solid solution and dislocation strengthening, lattice distortion
due to internal strains, fine particle size of martensite crystals make as quenched specimens
hardest [27]. The as quenched hardness values of both steels are expected to be high and equal since
they contain same amount of carbon. Figure 3.13 and 3.14 show the continuous cooling
transformation diagrams and hardenability curves of both steels. These steels have close
critical cooling rates; the critical cooling rate of SAE 4140 is slightly slower. Both steels are
austenitized and quenched identically so any ferrite or pearlite product would probably seen
35
Figure 3.11 SEM micrographs of quenched and tempered SAE 4140 specimens
a) As quenched b) Tempered at 200oC
c) Tempered at 300oC d) Tempered at 400oC
e) Tempered at 500oC f) Tempered at 600oC
36
Figure 3.12 SEM micrographs of quenched and tempered SAE 5140 specimens
a) As quenched b) Tempered at 200oC
c) Tempered at 300oC d) Tempered at 400oC
e) Tempered at 500oC f) Tempered at 600oC
37
Figure 3.13 Continuous cooling transformation (CCT) diagram and hardenability curve of
SAE 4140 steel [31]
38
Figure 3.14 Continuous cooling transformation (CCT) diagram and hardenability curve of
SAE 5140 steel [31]
39
in SAE 5140 whose critical cooling rate is higher. Another condition of full hardening is
cooling below Mf temperature. Austenite stabilizing elements like carbon, manganese, nickel,
chromium and molybdenum will lower both the Ms and Mf temperatures. The alloying
additions, especially manganese and molybdenum, of SAE 4140 are slightly higher than
SAE 5140 (Table 2.2). Since quenching mediums were same the lower hardness of SAE
4140 may be due to its lower Ms and Mf temperatures.
A fully martensitic component cannot be put to engineering use since it lacks toughness and
ductility. The brittleness of martensitic microstructures is due to a number of factors that may
include lattice distortion caused by carbon atoms trapped in the octahedral sites of the
martensite, impurity atom segregation at austenite grain boundaries, carbide formation
during quenching and residual stresses produced during quenching. Hence tempering is
necessary and it aims to:
(i) relieve internal stresses associated with the lattice shear due to the formation of martensite,
(ii) restore ductility and toughness at sacrifice of strength and hardness,
(iii) improve the dimensional stability through the breakdown of any retained austenite.
It is generally considered that tempered structures possess superior mechanical properties
than lamellar aggregates of equivalent hardness.
The structure of a steel quenched to form martensite is highly unstable. Reasons for the
instability include the supersaturation of carbon atoms in the body-centered tetragonal
crystal lattice of martensite, the strain energy associated with the fine dislocation or twin
structure of martensite, the interfacial energy associated with the high density of lath or plate
boundaries, and the retained austenite that is invariably present even in low-carbon steels.
The supersaturation of carbon atoms provides the driving force for carbide formation; the
high strain energy the driving force for recovery; the high interfacial energy the driving force
for grain growth or coarsening of the ferrite matrix; and the unstable austenite the driving
force for transformation to mixtures of ferrite and cementite on tempering.
In order to differentiate different stages of tempering the quenched specimens of both steel
types are tempered at various temperatures between 200oC - 600oC. The SEM micrographs
and hardness values indicate expected structural changes occur upon tempering of both
steel types.
Tempering up to 250oC does not change the microstructure and hence the hardness
significantly. In this first stage ε-carbides precipitate and martensite partially loses its
tetragonality due to the fact that carbon atoms can diffuse in the tetragonal lattice.
40
Experimental observations indicate that during this stage of tempering, apart from the initial bct
martensite of original carbon content, another bct martensite with c/a = 1,012 – 1,013
corresponding to ~0,3% carbon in solution appears in the structure. Carbon is precipitated
as ε-carbide which is approximately Fe2,4C and has hexagonal closed packed crystal
structure. This carbide precipitates as narrow laths or rodlets on cube planes of the matrix
and the carbon atoms are at octahedral interstities being as far away from each other as
possible. The epsilon carbide is usually nucleated first not because it is more stable but
because it has a better lattice matching with the matrix and hence greater probability for
nucleation as coherent nucleation can occur without much strain energy [27].
During tempering between 300oC and 400oC cementite replaces ε-carbides and martensite
loses tetragonality. Cementite first appears in the microstructure as plate like particles of
200nm long and ~15nm thickness. Once formed, the cementite particles agglomerate and
grow until a typical spheroidized structure is obtained. During this second tempering stage
the most likely sites for the nucleation of the cementite are the ε-carbide interfaces with
matrix and as the Fe3C particles grow, the ε-carbide particles gradually disappear. Other
possible nucleation sites for cementite are the interlath boundaries of the martensite and the
original austenite grain boundaries. The tetragonality of the matrix disappears and it is then,
essentially, ferrite, not supersaturated with respect to carbon [30]. The dislocation density is
effectively lowered not only by the reduction of dislocations within the laths but also by the
elimination of the low-angle lath boundaries [25].
Cementite to coarsens and spheroidizes; ferrite recrystallizes when tempering temperature is
increased further, up to 500oC and 600oC. At this third stage of tempering cementite particles
undergo a coarsening process and essentially become spheroidized. The spheroidization of
the cementite is encouraged by the resulting decrease in surface energy. The particles which
preferentially grow and spheroidize are located mainly at interlath boundaries and prior
austenite boundaries, although some particles remain in the matrix. The boundary sites are
preferred because of the greater ease of diffusion in these regions. At higher end of this
stage which could be up to 700oC, the martensite lath boundaries are replaced by more
equiaxed ferrite boundaries by a process which is best described as recrystallization [30]. In
addition residual stresses are relieved almost completely when tempering temperature
reaches 500oC [26]. The micro-stress relief is suggested to be due to the diffusion of carbon
from regions of compression to regions of tension [27].
As tempering temperature increases, by the mechanisms explained above, martensite softens
as indicated by hardness values of the steels. Although same tempering treatments are
applied, SAE 4140 steels soften more than SAE 5140. This behavior can be attributed to the
influence of alloying elements on tempering.
41
Tempering is a softening reaction which can be retarded by judicious choice of alloying
elements. The most effective elements in this regard are strong carbide formers such as
chromium, molybdenum, and vanadium. Without these elements, iron-carbon alloys and low
carbon steels soften rapidly with increasing tempering temperature. Alloying elements have
little influence on the first stage of tempering but may raise the second by as much as 100 or
200oC. The second stage, during which cementite precipitates, requires the diffusion of
carbon and the effect of alloying elements, in the absence of the formation of any alloy
carbides, can be appreciated in terms of their effect on diffusion of carbon. Alloying elements
retard softening through retarding the agglomeration of cementite by decreasing the rate of
carbon diffusion or by increasing the stability of cementite by dissolving in it. In carbon steels
the tetragonality of the lattice is observable up to 300oC but in alloy steels containing
chromium, molybdenum, tungsten, vanadium, cobalt and silicon, tetragonality may be
maintained after tempering at 400-500oC. The alloying elements also increase the stability of
low carbon martensite. In contrast manganese and nickel decrease stability.
Several alloying elements, notably silicon, chromium, molybdenum and tungsten, cause the
cementite to retain its fine structure to higher temperatures, either by entering into the
cementite structure or by segregating at the carbide-ferrite interfaces. They significantly
delay the softening process during 2nd and 3rd stages of tempering. This influence on the
cementite dispersion has other effects, in so far as the carbide particles, by remaining finer,
slow down the reorganization of the dislocations inherited from the martensite, with the result
that the dislocation substructures refine more slowly. The cementite particles are also found
on ferrite grain boundaries, where they control the rate at which the ferrite grains grow.
The alloying additions are more effective at first stage of tempering in SAE 5140 whose
chromium concentration is slightly higher. Chromium and manganese replace some iron
atoms from the ε-carbide, thereby increasing its stability and retarding the formation of
cementite [27]. On the other hand, chromium and molybdenum, two strong carbide formers
co-exist in SAE 4140. These elements increase the stability of cementite by dissolving in it
and significantly delay the softening process during higher temperatures. The molybdenum
content of SAE 4140 is 4 times higher than SAE 5140. The coefficient of diffusion of
molybdenum is extremely low as compared with carbon and its retardation effect appears
most prominently after tempering at about 500oC. [32]. The hardness changes reflect the
tempering behavior as well. The softening rate of SAE 5140 is slower at first stages whereas
at higher tempering temperatures like 300oC and 400oC its softening rate becomes faster
than SAE 4140 as attributed to alloying additions.
42
3.2.2 MBN measurements
MBN Profiles
Under the effect of an alternating magnetic field, a representative magnetic hysteresis loop
was induced in the small volume measured due to the energy loss with the irreversible
process of magnetization. This irreversible process is strongly related to the dynamic
behavior of domains, i.e., nucleation, annihilation and growth of domains. Grain/lathe
boundaries, dislocations and precipitates affect this dynamic behavior. Consequently, the
number of domain walls moving at a given instant and the mean free path of the domain wall
displacement decide the MBN peak height.
In a simple model proposed by Saquet et al. [21] change of the local magnetic moment
causing MBN was given by:
).( lSmrrrr δβδ = (3.3)
where β is a coefficient related to the atomic magnetic moment and type of domain wall, S is
the surface of the moving domain wall and δl is the wall displacement between pinning
obstacles. Previously it was mentioned that the principle source mechanism of MBN is the
motion of 180o domain walls. Hence an elementary Barkhausen event δm appears to be
mainly concerned by δl and S. δl is linked to microstructure which provides the pinning
obstacles and S is determined by the magnetic microstructure morphology which is also
related to microstructure. In a quenched steel small martensite needles/lathes determine the
size of domain and hence S. In addition domain wall displacements (δl) are short, so the
expected Barkhausen activity in quenched specimens is low. As tempering temperature
increases, the microstructure coarsens, thus larger domain wall displacements are possible.
Also the surface of a moving domain wall (S) will increase as domains get larger due the
coarsening of microstructure [29]. As a result the MBN activity is expected to increase as
tempering temperature increases.
Figures 3.15 and 3.16 show the graph for MBN signal versus applied field strength (MBN
profiles) for the quenched and tempered SAE 4140 and SAE 5140 specimens. As MBN is
symmetrical with respect to zero magnetic field, only the curves for the increasing applied
magnetic field are plotted. The peak positions were obtained by fitting a parabola to the 15%
of the top of the MBN curve in the fingerprints. The maximum point of this parabola is the
peak position. The as-quenched sample has the weakest MBN peak positioned at the
highest field linked with the high coercivity of martensite. Moreover, the peak position of the
43
signal shifts to the lower values of magnetic field due to tempering. As the tempering
temperature increases, the low amplitude broad peak of as-quenched martensite transforms
into a high amplitude peak situated at low magnetic field. The results show that magnetic
Barkhausen noise is influenced by the tempering which, as a function of the temperature,
causes changes in dislocation density, lattice straining (i.e., micro residual stresses) and the
morphology and size of cementite, and corresponding variations in hardness. The results are
in agreement with those of the previous studies.
In the as-quenched specimens, the body-centered tetragonal structure of martensite
determines the domain structure. Since the magnetic structure consists of very small
domains due to small needles\lathes, relative volume occupied by a domain wall is larger
(Figure 3.17). Also the easy direction of magnetization is restricted to the c-axis of tetragonal
structure only. Besides, high dislocation density in the martensite laths acts as a barrier to
the movement of domain walls. A strong field is required for the reversal of magnetization
because of low domain wall displacements and difficulty in nucleating domain walls.
Presence of micro residual stresses in the martensite needles has an additional effect on
reduction of the MBN response.
Tempering at 200oC changes the microstructure very slightly. Although ε-carbides form, the
microstructure is still needle shaped. Therefore, the height and position of the MBN peak do
not change significantly. When tempering temperature reaches 300oC and 400oC, cementite
replaces ε-carbides, the crystal structure of martensite loses its tetragonality, and dislocation
density reduces further. Corresponding magnetization orientation is no longer favored and
reverse domain nucleation and subsequent domain wall motions take place at lower
magnetic fields. All these factors make the domain wall motion easier, and therefore, the
amplitude of the MBN peak increases.
In tempering at 500oC and 600oC, carbides start spheroidizing and recrystallization of ferrite
begins. In parallel to the progressive coarsening of the microstructure, the average size of
the domain walls increases (Figure 3.18) These morphological changes and almost
complete relaxation of residual stresses result in a drastic increase in the MBN peak and a
clear shift to lower external magnetic field in the peak position by reducing the resistance to
the nucleation and movement of domains.
44
0
5
10
15
20
25
30
35
40
45
50
-100 -80 -60 -40 -20 0 20 40 60 80 100
Magnetic Field Strength (% of max.)
Avg
. MB
N le
vel (
% o
f 2V)
As QT 200T 300T 400T 500T 600
Figure 3.15 MBN Profiles of SAE 4140 specimens
0
2
4
6
8
10
12
14
16
-100 -80 -60 -40 -20 0 20 40 60 80 100
Magnetic Field Strength (% of max.)
Avg
. MB
N le
vel (
% o
f 2V)
As - QT 200T 300T 400T 500T 600
Figure 3.16 MBN profiles of SAE 5140 specimens
45
Figure 3.17 The SEM micrographs of as quenched SAE 4140(a); SAE 5140(b) steels and schematic representation of magnetic microstructure [33] (c) in martensite.
(c) Schematic representation
(b) As-quenched SAE 5140
(a) As-quenched SAE 4140
46
Figure 3.18 The SEM micrographs of SAE 4140 (a); SAE 5140 (b) steels tempered at 600oC and schematic representation of magnetic microstructure [33] (c) in tempered martensite.
(c) Schematic representation
(b) SAE 5140 tempered at 600oC
(a) SAE 4140 tempered at 600oC
47
The effects of alloying elements on tempering, retardation of softening during early stages
can easily be seen from the MBN profiles of the SAE 5140 steel. The profiles of as
quenched, and 200-300oC tempered specimens are very close and a clear separation exists
between the profiles of 300oC and 400oC tempered specimens, which indicates that the
structure continues to resist domain wall motion up to 300oC tempering. The profiles of SAE
4140 steel do not show such a clear separation between 300oC and 400oC; also between
500oC and 600oC. Thus, the differences between MBN profiles of SAE 4140 and 5140 are
consistent with microstructure and hardness variations upon tempering.
Table 3.2 Hardness values and MBN parameters of SAE 4140 specimens
Specimen Hardness(HV)
RMS (mV)
Peak Height(% of 2V)
Peak Position (%of max. field)
As Quenched 556 1.781 3.541 37.25 T 200 504 3.702 8.671 35.87 T 300 488 6.289 14.900 29.50 T 400 467 8.850 21.260 30.20 T 500 299 15.050 36.310 20.72 T 600 206 16.680 40.370 14.15
Table 3.3 Hardness values and MBN parameters of SAE 5140 specimens
Specimen Hardness(HV)
RMS (mV)
Peak Height(% of 2V)
Peak Position (%of max. field)
As Quenched 648 1.165 1.705 34.98 T 200 604 1.173 1.9 34.42 T 300 564 1.437 2.605 34.55 T 400 500 3.654 8.01 31.62 T 500 411 4.828 10.61 25.83 T 600 319 6.009 13.31 24.63
Frequency spectrum The raw MBN data is obtained with respect to time and signals in time domain can be
transformed to the frequency domain by application of Fourier transform in order to
determine the spectral content of time domain signals [34]. The frequency spectrum shows
the intensity of the noise given by the square of voltage (V2) as a function of the frequency.
The frequency spectrum is calculated in the given analyzing frequency range (0,1 – 1000
kHz) using “Fast Fourier Transform” (FFT) algorithm. Figure 3.19 and 3.20 show the
frequency spectrums of SAE 4140 and SAE 5140 specimens.
48
0
0.1
0.2
0.3
0.4
0.5
0.6
0.1 1 10 100 1000
Frequency (kHz)
Ampl
itude
(a.u
.)
As QT 200T 300T 400T 500T 600
Figure 3.19 Frequency spectrums of SAE 4140 specimens
0
0,05
0,1
0,15
0,2
0,25
0,1 1 10 100 1000
Frequency (kHz)
Am
plitu
de (a
.u.)
As QT 200 T 300T 400T 500T 600
Figure 3.20 Frequency spectrums of SAE 5140 specimens
49
The intensity of low frequency MBN signals increases as the tempering temperature
increases (Figure 3.19 and 3.20). Since frequency is inversely proportional to time, low
frequency content of MBN signal indicates larger domain wall displacements if the average
wall velocity is constant.
During MBN measurements the specimens are magnetized locally by irreversible
movements of domain boundaries. The applied field in this case produces reorientation of
the electron spins only within the width of the boundary walls as these pass across the
domains. Again the spin axes can reorient themselves only by some mechanism operating at
a finite speed [35]. Considering this, the average wall velocity can be assumed constant and
frequency content is directly related to distance between domain pinning sites.
In the as quenched martensite domain wall displacements are short due to small needles,
which is consistent with the absence of low frequency MBN signals. The frequency content
does not change significantly for tempering up to 200oC due to the fact that the
microstructure is still fine and needle shaped. As tempering temperature increases further
the amplitude of low frequency MBN signals increases, indicating larger domain wall
displacements. Such large displacements are expected due to progressive coarsening of the
microstructure.
Alloying elements that retard softening also affect frequency spectrums like in the case of
MBN profiles. This effect is clearly seen in the spectrums of SAE 5140 steel which was
explained in the previous section.
Pulse Height Distributions Pulse height distribution is a rarely used parameter and can be used to understand the
nature of MBN. This distribution, which is the number of events (pulses) against pulse
amplitude, depends on the number density and nature of pinning sites within the material [4].
The number of pulses detected in the MBN signal depends on the number of pinning sites
[36]. The pulse height distributions of the as-quenched and 600oC tempered specimens, the
most distinct cases, are given in Figure 3.21 and 3.22. As quenched specimens have about
180-200 thousand pulses detected whereas tempered ones have about 120000. In the as
quenched specimen each martensite needle/plate will act as a pinning site so the number of
pulses detected is very high. The quenched specimen has a pulse height up to only 0.4 V
indicating very small domain wall displacements.
50
020000400006000080000
100000120000140000160000180000200000
0 0,5 1 1,5
Pulse Height (V)
Num
ber o
f Pul
ses
As-QT 600
Figure 3.21 Pulse height distributions of SAE 4140 specimens
020000400006000080000
100000120000140000160000180000200000
0 0.5 1 1.5
Pulse Height (V)
Num
ber o
f Pul
ses
As QT 600
Figure 3.22 Pulse height distributions of SAE 5140 specimens
51
For the specimen tempered at 600oC the number of pulses decreases whereas height of the
pulses increases. The decrease in number of pulses is a consequence of softening of
microstructure which decreases the number of pinning sites respectively. Reduced dislocation
density and recrystallized ferrite enhance domain wall movement, thus allowing the domain walls to move longer distances or give larger jumps. These changes upon tempering are
reflected as higher amplitude pulses with heights more than 1.5 V, which are not present in the
as quenched specimen.
Hysteresis Loop Parameters
During MBN measurements a representative magnetic hysteresis loop is induced in the
small volume of specimen due to the energy lost in the form of a kind of internal friction. The noise signals obtained are proportional to rate of change in internal magnetization. Thus the
integration of noise signals along the whole applied magnetic field gives a representative
hysteresis loop from which parameters such as coercivity, remanence and permeability
could be obtained. These parameters may be used then in order to characterize samples.
Figure 3.23 (a) and 3.24 (a) show the coercivity values of SAE 4140 and 5140 steels. The
differences between quenched and tempered samples are not significant and the differences
between these values are small. It would be reasonable to assume that as quenched
samples have high coercivities due to the pinning of domain walls in the presence of small
martensite laths, needles and high dislocation density [37]. However during MBN measurements the actual level of magnetization reached locally is too small when compared
with the saturation magnetization of specimens. Also all specimens experience the same
applied field strength which makes it very difficult to distinguish coercivities. More reliable
results could be obtained from these specimens by obtaining true hysteresis curves form
magnetic pendulums and magnetometers.
Figure 3.23 (b) and 3.24 (b) show the remanence values of SAE 4140 and 5140 steels. The
remanences increase with increasing tempering temperature for both types of steels. The
quenched structures contain large amounts of crystal defects and it is possible that the elimination of part of these defects promoting an atomic rearrangement during the tempering
leads to an increase in the remanence of the steel [38].
Figure 3.23 (c) and 3.24 (c) show the permeability values of SAE 4140 and 5140 steels. For
both types of steels the permeabilities increase with increasing tempering temperature. As
tempering temperature increases magnetic softening of structure occurs parallel to
mechanical softening of the microstructure. The less efficient pinning of domain walls causes
magnetic softening, which also results in an easier domain wall motion [39]. It is expected
that enhanced domain wall motion increases the permeability of the tempered specimens.
52
As QT 200
T 300 T 400
T 500T 600
-
0.050
0.100
0.150
0.200
0.250
0.300
0.350
0.400
Frac
tion
of M
ax. A
pplie
d Fi
eld
Stre
ngth
(a)
As Q
T 200
T 300
T 400
T 500 T 600
0
10
20
30
40
50
60
Perc
ent o
f 2V
(b)
As Q
T 200
T 300
T 400
T 500 T 600
0
50
100
150
200
250
300
Slop
e at
Coe
rciv
e Po
int
(c)
Figure 3.23 Coercivity (a), remanence (b), permeability (c) values of SAE 4140
53
As QT 200
T 300
T 400T 500 T 600
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Frac
tion
of M
ax. A
pplie
d Fi
eld
Stre
ngth
(a)
As Q T 200T 300
T 400
T 500
T 600
0
5
10
15
20
25
Perc
ent o
f 2V
(b)
As Q T 200 T 300
T 400
T 500
T 600
0
20
40
60
80
100
120
Slop
e at
Coe
rciv
e Po
int
(c)
Figure 3.24 Coercivity (a), remanence (b), permeability (c) values of SAE 5140
54
Hardness – MBN Correlation Generally a linear relation is found between mechanical hardness and the peak position of
the MBN. The linear relation between magnetic coercive force and hardness explains such
behavior. The microstructural features that impedes dislocation movement, also makes
domain wall movement harder [5]. Thus harder the material higher the position and lower the
height of the peak would be. Residual stresses have an additional effect on both hardness
and MBN in a similar manner. MBN is sensitive to both hardness and stresses which also
influence hardness. Regarding this a good correlation between MBN and hardness is
expected. Generally correlation between the peak height of MBN profile and hardness was
studied [40]. Another parameter that reflects the MBN behavior and is influenced in the same
way as MBN peak height upon tempering, is the root mean square of the Barkhausen signal
(RMS) as seen on Table 3.2 and 3.3.
The raw magnetic noise data consists of a series of voltage pulses and associated magnetic
field values. RMS of all signal amplitudes were sampled for the specified analyzing
frequency range according to the formula:
∑−
=
=1
0
21 n
iixn
RMS (3.4)
It is seen in Table 3.2 and 3.3 that the as-quenched specimens (the hardest ones) have the
lowest RMS values. As tempering temperature increases, in contrast to the decrease in
hardness, RMS value increases. Pinned domain walls due to high dislocation density and
small martensite needles cause lower RMS values. As tempering temperature increases
dislocation density decreases, micro residual stresses diminish and the magnetic structure
comes close to those of a ferrite. Thus, RMS value increases due to the enhancement of
domain wall displacement with softening of martensite. Figure 3.25 and 3.26 show the
correlation graph between the RMS values and peak heights of the MBN signal with the
hardness of specimens. The regression analysis shows an excellent correlation between the
RMS values and peak heights with hardness. The correlation between RMS and hardness is
better than the correlation with peak height. Thus RMS values can also be used for hardness
correlations.
55
Peak Height = -0.1057(HV) + 65.232R2 = 0.9396
RMS = -0.043(HV) + 26.796R2 = 0.9406
05
101520253035404550
0 100 200 300 400 500 600 700Hardness (HV)
Peak
Hei
ght o
r RM
S
Peak HeightRMS
Figure 3.25 Hardness correlations for SAE 4140 specimens
Peak Height = -0.0393(HV) + 26.29R2 = 0.9534
RMS = -0.0165(HV) + 11.408R2 = 0.9535
0
2
4
6
8
10
12
14
16
0 100 200 300 400 500 600 700Hardness (HV)
Peak
Hei
ght o
r RM
S
Peak HeightRMS
Figure 3.26 Hardness correlations for SAE 5140 specimens
56
3.2.3 Detection of Faulty Quenching and Tempering Treatment by MBN
When cooling rate shows significant variations as going from surface to interior, the phase
content and the residual stress state along the thickness of the specimen may differ. In such
cases, the hardness and the microstructure of the surface may not represent the whole
structure. Microstructural investigations showed that the thickness of the samples of SAE
4140 and SAE 5140 used in this study allowed the formation of desired microstructure
uniformly along a penetration depth of the MBN activity, which usually varies between 0.01
and 1.5 mm depending on the analyzing frequency. To demonstrate the effect of low
hardenability on microstructure and MBN, SAE 1040 steel was used. This steel type
practically has very low hardenability when compared with SAE 4140 and SAE 5140 as can
be seen from its continuous cooling transformation given in Figure 3.27.
Figure 3.27 Continuous cooling transformation diagram of SAE 1040 steel [31]
57
Figure 3.28, 3.29, 3.30, 3.31 show the SEM micrographs, MBN profiles and corresponding
frequency spectrums of quenched and tempered SAE 1040 specimens. In all specimens at
the edge regions where cooling rate is higher, fully martensitic or tempered martensitic
structures are observed. However, when going from edge to interior regions cooling rate is
lowered. The effect of low cooling rate combined with low hardenability is best observed at
centre portions of specimens. At the centre sections of all SAE 1040 specimens proeutectoid
ferrite is observed which influences the MBN signals as well.
In as-quenched specimen high dislocation density, small martensite needles/plates and
presence of micro residual stresses cause low MBN activity. As tempering temperature
increases MBN activity increases due to the softening of martensitic microstructure. The
edge sections of specimens where fully martensitic structure is present, the MBN activity is
as expected. On the other hand MBN activities of centre sections of specimens are
unexpectedly high. This high activity can be attributed to the presence of ferrite which has
practically no resistance to domain wall motion. Enhanced domain wall motion in ferrite
increases the MBN activity which can be directly observed from MBN profiles and frequency
spectrums. In the presence of ferrite higher amplitude profiles and spectrums are observed
in all specimens.
The difference between edge and centre portions is very significant for the as quenched
specimen whereas for specimens tempered at high temperatures, 500oC and 600oC, MBN
activities of fully martensitic and ferrite containing regions come closer. Such a behavior is
expected since at high tempering temperatures the magnetic microstructure developed is
very close to that of ferrite. It should also be mentioned that the volume fractions of
martensite and ferrite may also influence the MBN activity.
Figure 3.33 shows the MBN profiles of edge regions of SAE 1040 where fully martensitic
structures are present. The amplitudes of tempered specimens are far beyond the as
quenched specimen. This indicates that SAE 1040 shows practically no resistance to temper
softening. Since it contains no alloying elements upon tempering SAE 1040 softens rapidly,
where as in case of low alloy steels of SAE 4140 and especially 5140 the softening is
retarded.
58
As Quenched
05
1015202530354045
-100 -80 -60 -40 -20 0 20 40 60 80 100
Magnetic Field Strength (% of max.)
Avg
. MB
N le
vel (
% o
f 2V)
EdgeCentre
0
0.1
0.2
0.3
0.4
0.5
0.6
1 10 100 1000
Frequency (kHz)
Am
plitu
de (a
.u.)
EdgeCentre
Figure 3.28 SEM micrographs taken from edge(a) and centre(b); MBN profiles(c); frequency spectrums(d) of as-quenched SAE 1040 specimen.
(b) Centre (a) Edge
(d)
(c)
59
Tempered at 300oC
05
1015202530354045
-100 -80 -60 -40 -20 0 20 40 60 80 100
Magnetic Field Strength (% of max.)
Avg
. MB
N le
vel (
% o
f 2V) Edge
Centre
0
0.1
0.2
0.3
0.4
0.5
0.6
1 10 100 1000
Frequency (kHz)
Am
plitu
de (a
.u.)
EdgeCentre
Figure 3.29 SEM micrographs taken from edge(a) and centre(b); MBN profiles(c); frequency spectrums(d) of SAE 1040 specimen tempered at 300oC
(b) Centre
(d)
(a) Edge
(c)
60
Tempered at 500oC
05
1015202530354045
-100 -80 -60 -40 -20 0 20 40 60 80 100
Magnetic Field Strength (% of max.)
Avg
. MB
N le
vel (
% o
f 2V) Edge
Centre
0
0.1
0.2
0.3
0.4
0.5
0.6
1 10 100 1000
Frequency (kHz)
Am
plitu
de (a
.u.)
EdgeCentre
Figure 3.30 SEM micrographs taken from edge(a) and centre(b); MBN profiles(c); frequency spectrums(d) of SAE 1040 specimen tempered at 500oC
(b) Centre
(d)
(a) Edge
(c)
61
Tempered at 600oC
05
1015202530354045
-100 -80 -60 -40 -20 0 20 40 60 80 100
Magnetic Field Strength (% of max.)
Avg
. MB
N le
vel (
% o
f 2V) Edge
Centre
0
0.1
0.2
0.3
0.4
0.5
0.6
1 10 100 1000
Frequency (kHz)
Am
plitu
de (a
.u.)
EdgeCentre
Figure 3.31 SEM micrographs taken from edge(a) and centre(b); MBN profiles(c); frequency spectrums(d) of SAE 1040 specimen tempered at 600oC
(b) Centre
(d)
(a) Edge
(c)
62
0
5
10
15
20
25
30
35
-100 -80 -60 -40 -20 0 20 40 60 80 100Magnetic Field Strength (% of max.)
Avg
. MB
N le
vel (
% o
f 2V)
As - QT 300T 400T 500T 600
Figure 3.33 MBN profiles of edge regions of SAE 1040 specimens
Edge region
Centre region
Figure 3.32 Schematic drawing of edge and centre regions of SAE 1040 specimens
63
CHAPTER 4
CONCLUSIONS For the purpose of characterizing steel microstructures non-destructively the steel
specimens were heat treated using two common procedures. In the first part of this study the
effect of average ferrite grain size was investigated in the annealed low carbon steels. The
influence of tempering induced changes on MBN was studied in the second part.
Various average ferrite grain sizes were obtained by annealing the SAE 1010 specimens at
different time and temperature variations. The effects of pearlite and hardness on MBN were
neglected due to the fact that ferrite volume fraction in all specimens was about 90% and the
hardness differences were so small. Effect of residual stresses was eliminated by cooling the
specimens very slowly in the furnace. Thus only the differences in average ferrite grain size
influenced the MBN response of the specimens.
As grains become coarser, domain size increases and domain density decreases.
Consequently MBN peak heights and RMS values decrease. The peak positions of MBN
signals indicate that the MBN activity in the annealed specimens occurs during the early
stages of magnetization. At this stage domain nucleation is the predominant mechanism and
grain boundaries are the preferential sites for nucleation. As grain size increases, grain
boundary area per unit volume decreases, causing difficulty in nucleation of new domains.
An increase in average ferrite grain size cause reduced Barkhausen jumps and difficulty in
creating new domains; which decrease MBN activity. However grain boundary segregation in
some specimens alters the magnetic structure and increase the MBN activity. The results
obtained are consistent with the previous studies and theoretical expectations. It could be
concluded that MBN is sensitive to grain size differences.
Martensitic microstructures were obtained by quenching and tempering the low alloy steels
at different temperatures. MBN method is a powerful tool for evaluating different stages of
tempering. In the as quenched samples, pinned domain walls due to high dislocation density
and small martensite needles cause low MBN activity, and MBN peak is at higher magnetic
64
fields due to small domain wall displacements and difficulty in domain nucleation. As
tempering temperature increases dislocation density decreases, micro residual stresses
diminish and the magnetic structure comes close to those of a ferrite. Thus, MBN activity
gets higher due to the enhancement of domain wall displacement with softening of
martensite. Change in the number density of pinning sites and relative domain wall
displacements upon tempering influences the frequency spectrums, pulse height
distributions as well. RMS values can also be used instead of peak height, considering the
better correlation between RMS values and hardness, for hardness correlation. Via
establishing the quantitative relationships between MBN parameters and the microstructural
parameters, this method can be utilized efficiently and effectively for evaluating the hardness
and the microstructure of the steel components.
The alloying additions influence tempering behavior of steels. Alloying elements retard
softening of martensite structure either by stabilizing the carbides or by decreasing the rate
of carbon diffusion. MBN is sensitive to this softening retardation. In the absence of alloying
elements steels soften more rapidly upon tempering and hardenability decreases. Lower
hardenability causes pro-eutectic phases to form in some portions of specimens. Steel with
lower hardenability develop such structures and MBN can be used to detect such
treatments.
The results show that the MBN parameters are sensitive to tempering induced changes,
faulty treatments and grain size differences. The sensitivity of this phenomenon to
microstructural changes gives a wide range of potential applications to the technique.
For further investigations, ferromagnetic materials having nano-sized grains which should
have single domain magnetic structures may be a good candidate for MBN studies. Besides
components produced by powder metallurgy techniques can be investigated by MBN. As a
final suggestion domains and domain wall motion can be observed using Kerr effect and
transmission electron microscopy (TEM) in order to understand the domain mechanisms like
domain nucleation, domain annihilation and irreversible domain wall motion.
65
REFERENCES 1 Chikazumi, S. Physics of Magnetism. (Wiley, New York, 1964).
2 Barrett, C.R., Nix, W.D. and Tetelman, A.S. The Principles of Engineering
Materials (Prentice-Hall, Englewood Cliffs, N.J., 1973).
3 Shilling, J.W. and G. L. Houze, J. Magnetic properties and domain structure in
grain-oriented 3% Si-Fe. IEEE Transactions on Magnetics, 1974(MAG-10), 195-221.
4 Jiles, D. Introduction to Magnetism and Magnetic Materials (Chapman and Hall,
London; New York, 1991).
5 Cullity, B.D. Introduction to Magnetic Materials. (Addison-Wesley, Reading, 1972).
6 Carey, R. and Isaac, E.D. Magnetic domains and techniques for their observation.
(Academic Press, New York, 1966).
7 Neelkanta, P.S. Handbook of Electromagnetic Materials: Monolithic and Composite
Versions and Their Applications. (CRC Press, Boca Raton, Florida, 1995).
8 Schroeder, K. CRC Critical Reviews of Solid State Science, 1976(6), 45-47.
9 ASM Metals Handbook Volume 17 Nondestructive Evaluation and Quality Control.
(ASM, 1997).
10 Ranjan, R. Microstructural characterization of ferromagnetic materials using
magnetic NDE techniques. Materials Science and Engineering, pp. 115, PhD thesis (Iowa
State University, Ames, Iowa, 1987).
11 Blaow, M., Evans, J.T. and Shaw, B.A. Magnetic Barkhausen noise: the influence
of microstructure and deformation in bending. Acta Materialia, 2005(53), 279-287.
12 DeBlois, R.W. Magnetization processes: Reversals and Losses. Journal of Applied
Physics, 1958(29), 459-467.
13 Dobmann, G., Kern, R., Altpeter, I. and Theiner, W. Quantitative hardening-depth-
measurements up to 4mm by means of micromagnetic microstructure multiparameter
analysis. Review of Progress in Quantitative Nondestructive Evaluation, 1988, 1471-1475.
14 Titto, S., Otala, M. and Saynajakangas, S. Non-destructive magnetic measurement
of steel grain size. Non-destructive Testing, 1976(9), 117-120.
15 Ranjan, R., Jiles, D.C. and Rastogi, P.K. Magnetic properties of decarburized
steels: an investigation of the effects of grain size and carbon content. IEEE Transactions on
Magnetics, 1987(MAG-23), 1869-1876.
16 Anglada-Rivera, J., Padovese, L.R. and Capo-Sanchez, J. Magnetic Barkhausen
Noise and hysteresis loop in commercial carbon steel: influence of applied tensile stress and
grain size. Journal of Magnetism and Magnetic Materials, 2001(231), 299-306.
66
17 Gatelier-Rothea, C., Chicois, J., Fougeres, R. and Fleischmann, P. Characterization of pure iron and (130 p.p.m.) carbon-iron binary alloy by Barkhausen Noise
measurements: influence of stress and microstructure. Acta Materialia, 1998(46), 4873-
4882.
18 Yamaura, S., Furuya, Y. and Watanebe, T. The effect of grain boundary
microstructure on Barkhausen Noise in ferromagnetic materials. Acta Materialia, 2001(49),
3019-3027.
19 Sakamoto, H., Okada, M. and Homma, M. Theoretical Analysis of Barkhausen
noise in carbon steels. IEEE Transactions on Magnetics, 1987(MAG-23), 2236-2238.
20 Kameda, J. Characterization of tempered martensite microstructure and
embrittlement by acoustic and magnetic Barkhausen signal measurement. Scripta
Metallurgica, 1988(22), 1487-1492.
21 Saquet, O., Chicois, J. and Vincent, A. Barkhausen noise from plain carbon steels:
analysis of the influence of microstructure. Materials Science and Engineering A, 1999(269),
73-82.
22 Moorthy, V., Vaidyanathan, S., Jayakumar, T. and Raj, B. Microstructural
characterization of quenched and tempered 0.2% carbon steel using magnetic Barkhausen
noise analysis. Journal of Magnetism and Magnetic Materials, 1997(171), 179-189.
23 Moorthy, V., Shaw, B.A. and Evans, J.T. Evaluation of tempering induced changes
in the hardness profile of case-carburised EN36 steel using magnetic Barkhausen noise
analysis. NDT&E International, 2003(36), 43-49.
24 Yi, J.-K., Lee, B.-W. and Kim, H.C. Micromagnetic nondestructive evaluation of
isochronally tempered 12% CrMoV steel. Scripta Metallurgica et Materialia, 1994(30), 313-
318.
25 Krauss, G. Steels: Heat Treatment and Processing Principles. (ASM International,
Materials Park, Ohio, 1989).
26 Avner, S.H. Introduction to Physical Metallurgy. (McGraw-Hill, New York, 1974).
27 Kumar, R. Physical Metallurgy of Iron and Steel. (Asia Publishing House, New York,
1968).
28 Hosford, W.F. Physical Metallurgy. (Taylor & Francis Press, Boca Raton, 2005).
29 Otala, M. and Saynajakangas, S. New electronic grain size analyzer for technical
steel. Journal of Physics E; Sci. Instrum., 1972(5), 669-672.
30 Honeycombe, R. and Bhadeshia, H.K.D.H. Steels: Microstructure and Properties.
(Edward Arnold, London, 1995).
31 Atlas of Isothermal Transformation and Cooling Transformation Diagrams. (ASM
International, Metals Park, Ohio, 1977).
32 Archer, R.S., Briggs, J.Z. and C. M. Loeb, J. Molybdenum - Steels, Irons, Alloys.
(Climax Molybdenum Company, New York, 1965).
67
33 Saquet, O. Caracterisation des aciers par bruit Barkhausen (Interaction
microstructure cristalline / microstructure magnetique). Genie des Materiaux (GEMPPM
INSA Lyon, Lyon, 1997).
34 Libby, H.L. Introduction to Electromagnetic Nondestructive Test Methods. (Wiley-
Interscience, New York, 1971).
35 Brailsford, F. Physical Principles of Magnetism. (Van Nostrand, London; New York,
1966).
36 Martinez-de-Guerenu, A., Galarza, A., Crespo, A., Leunda, I. and Arizti, F. Application of Barkhausen noise techniques to the characterisation of recovery and
recrystallization in cold rolled low carbon steel. In Debarberis, L., Pirfo, S., Wojtas, A. and
Suortti-Suominen, T., eds. 5th International Conferance on Barkhasuen Noise and
Micromagnetic Testing, pp. 59-70 Petten, The Netherlands, 2005).
37 Watanebe, K. Temperature dependence and permanent magnet properties of Fe-
Pt-Ti alloys. IEEE Transactions on Magnetics, 1987(MAG-23), 3196-3198.
38 Tavares, S.S.M., Pedrosa, P.D.S., Teodosio, J.R., Silva, M.R.d., Neto, J.M. and Pairis, S. Magnetic properties of the UNS S39205 duplex stainless steel. Journal of Alloys
and Compounds, 2003(351), 283-288.
39 Ryu, K.S., Nahm, S.H., Park, J.S., Yu, K.M., Kim, Y.B. and Son, D. A new non-
destructive method for estimating the remanent life of a turbine rotor steel by reversible
magnetic permeability. Journal of Magnetism and Magnetic Materials, 2002(251), 196-201.
40 Moorthy, V., Vaidyanathan, S., Jayakumar, T. and Raj, B. On the influence of
tempered microstructures on magnetic Barkhausen emission in ferritic steels. Philosophical
Magazine A, 1998(77), 1499-1514.