Você está na página 1de 122

Lehrstuhl f

ur Werkstoffkunde und Werkstoffmechanik mit


Materialpr
ufamt f
ur den Maschinenbau
Technische Universitat M
unchen

Mechanical properties of Dual-Phase steels


Prodromos Tsipouridis

Vollstandiger Abdruck der von der Fakultat f


ur Maschinenwesen
der Technischen Universitat M
unchen
zur Erlangung des akademischen Grades eines

Doktor-Ingenieurs (Dr.-Ing.)

genehmigten Dissertation.

Vorsitzender:
Pr
ufer der Dissertation:

Univ.-Prof. Dr.-Ing. Horst Baier


1. Univ.-Prof. Dr. mont. habil. Ewald Werner
2. Hon.-Prof. Dr.-Ing, Dr. Eng. (Japan) Hans-Harald Bolt

Die Dissertation wurde am 14.03.2006 bei der Technischen Universitat M


unchen
eingereicht und durch die Fakultat f
ur Maschinenwesen
am 19.06.2006 angenommen.

Acknowledgements

This study was carried out during my employment as research assistant at the Institute for
Materials Science and Mechanics of Materials of TU-Munich (Lehrstuhl f
ur Werkstoffkunde
und Werkstoffmechanik).
I am deeply indebted to my direct advisor Prof. Dr. mont. habil. E. A. Werner for his unfailing
support all these years, for his encouragement to proceed with new ideas and for being daily
available for uncountable scientific discussions. Thank you very much!
My special thanks to my co-advisor Prof. Dr.-Ing., Dr. Eng. H.-H. Bolt (Head of group Materials Synthesis and Materials Characterization of Max-Planck-Institut f
ur Plasmaphysik, IPP
Garching), as well as to the chairman of my PhD examination, Prof. Dr.-Ing. H. Baier (Head
of the Institute for Lightweight Structures, TU-Munich).
My sincere thanks to the project partner voestalpine Stahl Linz GmbH for supplying the
testing material and making possible to conduct the annealing simulations as well as the tensile
and hole expansion tests. Special thanks to Dr. A. Pichler for the support and to Dipl.-Ing.
E. Tragl for the close and continuous collaboration.
I would also like to thank the Christian Doppler Research Association (CDG) for sponsoring
and supporting this study during the years 2002-2005 (as a project/module of the ChristianDoppler-Laboratory for Modern Multiphase Steels).
I should not forget to thank Dr. G. Triantafyllides (Dep. of Chemical Engineering of Aristotle
University of Thessaloniki) for motivating me to choose the field of steel research.
Last, but not least, i would like to thank my colleagues and the technical staff of the Chair,
without whose help this work would be never completed. But most of all, thank you for
creating a pleasant and friendly working environment and for helping me to become an active
member of our group.

Munich, June 2006

Prodromos Tsipouridis

Contents
1 Introduction
1
1.1 Low-alloyed dual-phase steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Grain refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2 Material production and experimental procedure
2.1 Production of the material . . . . . . . . . . . . . . . . . . .
2.2 Thermodynamical calculations . . . . . . . . . . . . . . . . .
2.3 Pre-processing . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4 Cold-Rolling trials . . . . . . . . . . . . . . . . . . . . . . .
2.5 Dilatometric investigations . . . . . . . . . . . . . . . . . . .
2.6 Annealing simulations and austenitization kinetics . . . . . .
2.7 Microstructure investigations . . . . . . . . . . . . . . . . . .
2.8 Mechanical testing . . . . . . . . . . . . . . . . . . . . . . .
2.8.1 Tensile testing . . . . . . . . . . . . . . . . . . . . . .
2.8.2 Ultramicrohardness and microhardness testing . . . .
2.8.3 Hole expansion measurements (Stretch flangeability)
3 Results
3.1 Microstructure investigations . . . . . . . . . . . . . . .
3.1.1 Recrystallization/austenitization kinetics . . . .
3.1.2 Dilatometric investigations . . . . . . . . . . . .
3.1.3 Annealing simulations . . . . . . . . . . . . . .
3.1.4 Quantitative analysis- Grain size measurements
3.2 Mechanical properties . . . . . . . . . . . . . . . . . . .
3.2.1 Tensile testing . . . . . . . . . . . . . . . . . . .
3.2.2 Hardness measurements . . . . . . . . . . . . .
3.2.3 Hole expansion . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

15
15
16
16
17
18
19
20
20
20
21
21

.
.
.
.
.
.
.
.
.

23
23
23
26
33
44
50
50
65
71

4 Discussion
77
4.1 Grain refinement and microstructure investigations . . . . . . . . . . . . . . . . 77
4.2 Mechanical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5 Summary

102

II

Chapter 1
Introduction
The increasing automobile market demands for reduced fuel consumption as well as the need
to comply with the international environmental regulations regarding greenhouse gas emissions
(GHG), resource reduction and recyclability have motivated and/or even forced the automotive
industry to produce more fuel-efficient vehicles by reducing their weight. In order to provide a
steel-based structural platform that fulfills the auto-makers requirements and takes advantage
of the new high strength steels, a new vehicle architecture based on novel design concepts has
been developed. The application of advanced high strength sheet steels exhibiting both high
strength and excellent formability offers the unique option of combining weight reduction (by
using thinner gauges of sheet material) with improved passive safety, optimized environmental
performance and manufacturing feasibility at affordable cost.
The most common steels used presently in the automobile industry are mild steels, which
are low-carbon steels characterized by a yield strength level of 140 MPa and an excellent deep
draw ability. Despite their forming and cost advantages over high strength steels, the ultimate
strength level of mild steels remains at relatively low levels, so that the crash performance is
mainly dependent on the sheet thickness. By consistent controlling of the alloy chemistry, considering the presence of interstitial carbon in ferrite, interstitial-free grades (IF) possessing an
ultra-low carbon content were produced. Traditional strengthening mechanisms such as solid
solution hardening (with phosphorous to be the most common hardening element), precipitation hardening and grain refinement by carbides and/or nitrides were employed to increase the
strength of IF-steels, while maintaining their excellent formability. Micro-alloying with vanadium, niobium or titanium accompanied with fine carbide precipitation and grain refinement
leads to even higher strength levels and increases the ratio of yield to tensile strength. Bakehardening steels (BH) offer a combination of good formability during stamping and provide
an increased yield strength after the paint-baking process. To take advantage of the bakehardening effect a certain content of solute carbon and an appropriate aging heat treatment
are required, aiming at the supersaturation of carbon in the ferrite [1]. Due to their sharp
1

upper yield point, BH-steels exhibit a high dent resistance, which makes them good candidate
materials for outer body panel applications. For the steel grades mentioned above, widely
designated as conventional steels, a reduced formability is an unavoidable consequence when
selecting steels with higher strength levels.

60

Elongation to fracture

(%)

50

IF

40

Mild

AHSS

High Strength IF

IS

30

BH
C-Mn

20

HSLA

DP

TRIP

10
0

Multiphase Steels

200

400

600

Tensile Strength

Rm

800

1000

1200

(MPa)

Figure 1.1: Strength-Formability relationship of thin sheet steels

To overcome this problem new types of high strength steels, the so-called Advanced High
Strength Steels (AHSS), were developed from the sheet steel suppliers in cooperation with the
automakers and design engineers. These grades exhibit higher rates of work hardening than
conventional steels as a result of their low yield strength to tensile strength ratio, show good
press formability and reach higher ultimate tensile strengths, therefore they have the potential
for significantly improved crash performance [2]. AHSS steels are multiphase steels consist
of hard islands of martensite, bainite and/or retained austenite dispersed in a ductile ferritic
matrix, in quantities and combinations sufficient to produce desired mechanical properties.
The multiphase AHSS family includes Dual-Phase (DP, ferritic-martensitic), TRansformation
Induced Plasticity (TRIP) and complex multiphase steels. Ferritic-bainitic steels, also known as
stretch-flangeable (SF) steels, are considered as a subgrade of the DP products. The mechanical
properties of conventional and AHSS thin sheet steels with respect to ductility and ultimate
2

tensile strength are shown in Figure 1.1. As can be observed, a partial overlap between the
strength levels of different steel grades is possible.
According to the results of the ULSAB-AVC Program (Ultra Light Steel Auto Body-Advanced
Vehicle Concepts), an automotive body could be constructed by utilizing approximately 85 %
of AHSS, achieving a weight reduction of 25 % compared with a bench-marked average
base model and without any increase of the manufacturing costs. Figure 1.2 shows that the
clear majority of autobody components is designed using dual-phase steels [3, 4]. Different
criteria such as formability, weldability, spring-back behavior and of course static and dynamic
properties play a significant role in the material selection, even though for some parts more
than one steel grades fulfill the mechanical property standards and hence are also applicable.
In particular, for a number of components that both DP and TRIP steels are viable candidates,
cost-effective DP grades were preferable.

DP
Misc
HSLA

DP
BH
TRIP
MART
CP
IF
HSLA
Misc

IF

CP
MART

TRIP

BH

Figure 1.2: ULSAB autobody structure steel grade distribution

Due to their special microstructural characteristics, ferritic-martensitic dual-phase (DP) steels


provide an attractive combination of strength and ductility and furthermore exhibit continuous
yielding behavior accompanied with a high work hardening rate. In order to meet the
different design requirements of individual components, various DP grades regarding tensile
strength and formability are produced industrially. This variation of mechanical properties
is mainly achieved by controlling the carbon content of the steel. Addition of other alloying
elements such as manganese, chromium, vanadium, molybdenum and nickel, individually or in
3

combination, can also increase hardenability. Another method to increase the tensile strength
of dual-phase steels is to increase the martensite fraction by applying appropriate annealing
schedules, even though this procedure is followed by an expected loss in formability. In the
present study, grain refinement is proposed as an alternative way/solution to improve the mechanical properties of a low-alloyed dual-phase steel. To achieve this, severe plastic deformation
was applied to a pre-processed dual-phase steel by means of conventional cold-rolling, followed
by an appropriate final heat treatment with the aim to produce a homogeneous fine-grained
dual-phase ferritic-martensitic microstructure. The impact of grain refinement on the mechanical properties of the dual-phase steel, regarding tensile properties, ultramicrohardness and hole
expansion behavior, is then investigated in this work.

1.1

Low-alloyed dual-phase steels

Dual-phase (DP) steels were developed in the mid-seventies in order to satisfy the increasing
needs of automotive industry for new high strength steels which combine significant weight
reduction and improved crash performance, while keeping the manufacturing costs at affordable
levels. The high commercial potential of the newly developed alloy has motivated extensive
research in numerous laboratories, resulting in DP-grades having a wide range of chemical
compositions and being produced with various processing routes.
Dual-phase steels are characterized by a microstructure consisting of a fine dispersion of
hard martensite particles in a continuous, soft, ductile ferrite matrix. The term dual-phase,
firstly reported by Hayami and Furukawa [5] and thereafter adopted by the steel research
community, refers to the presence of essentially two phases, ferrite and martensite, in the
microstructure, although small amounts of bainite, pearlite and retained austenite may also be
present. Irrespective of the chemical composition of the alloy, the simplest way to obtain a dualphase ferritic-martensitic steel is intercritical annealing of a ferritic-pearlitic microstructure in
the + two-phase field, followed by a sufficiently rapid cooling to enable the austenite to
martensite transformation.
Three basic approaches exist for the commercial production of dual-phase steels: (a) the
as-hot-rolled approach, where the dual-phase microstructure is developed during the conventional hot-rolling cycle by careful control of chemistry and processing conditions [6
12], (b) the continuous annealing approach, where hot- or cold-rolled steel strip is uncoiled and annealed intercritically to produce the desired microstructure [13, 14] and (c) the
batch-annealing, where hot- or cold-rolled material is annealed in the coiled condition
[1519].
Box- or batch-annealing was mainly considered for economical and practical reasons by steelmakers where continuous facilities were not available. Dual-phase steels could be obtained by
means of batch-annealing in the intercritical region for approximately 3 h to ensure homogeneity, followed by very slow cooling. Due to the extremely low cooling rates, much higher alloying
contents were necessary to achieve the desired hardenability (i.e. 2.5 % Mn, 1.5 % Si, 1.0 % Cr).
On the other hand, dual-phase steel production in the hot strip mill demands precise control of
the transformation, because the transformation starts from single phase austenite. The
determination of an accurate CCT- diagram via dilatometry, where the influences of alloying
elements, heat treatment conditions and desired properties are integrated, is of great importance for the success of the process. The critical point is the correction of the CCT- diagram
to include the presence of strain in austenite (to simulate the real process conditions).
The use of continuous annealing lines (CAL) offers the advantages of high production rates,
better uniformity of the steel properties and, furthermore, the possibility of using lower alloyed
steels (having a lean chemistry). In continuous annealing lines three types of cooling are utilized
5

[20]: (a) water-quenching, (b) gas-jet cooling and (c) air-cooling. The use of CAL equipped
with water quenching facilities makes possible an easy and economical production of dualphase high-strength steel. The basic heat cycle involves intercritical annealing and subsequent
water quenching to form the ferritic-martensitic microstructure. If necessary (according to
application), a tempering stage follows [13, 2123].
From the above short review it becomes clear, that dual-phase steels can be produced by
cooling from the annealing temperature (intercritical or austenitic) by any cooling rate in the
range between batch-annealing and water-quenching. Since every steel producer has different
melting, rolling and cooling facilities, the choice of the alloying elements best suited to the
existing production capabilities is mandatory. Thus, a single widely accepted alloy composition
for each grade of the dual-phase steel family is out of consideration. There are many combinations of alloying elements such as Mn, Si, Cr, Mo and V that can be added to low carbon
(0.1 wt. % C) iron to obtain the desired ferritic-martensitic microstructure. A basis with less
than 0.15 wt. % carbon (for weldability reasons) and 1.5 wt. % Mn is generally acceptable. To
achieve good ductility and toughness, an initial carbon content of about 0.1 wt. % C is ideal, so
that the carbon content and the morphology of martensite can be controlled. For strong and
tough martensite this value is approx. 0.4 wt. % C, depending on the total alloy composition.
For martensite carbon contents higher than 0.4 wt. % C twinned martensite may be formed.
Concerning the other alloying elements, manganese (Mn), chromium (Cr), molybdenum (Mo),
silicon (Si) and vanadium (V) are commonly used to increase the hardenability of austenite
[11, 14, 19, 2428]. The role of Si and Mn is more complex, since they may also contribute to
solid solution strengthening of ferrite and hence to the strength level of the steel. The addition
of elements such as Cr, Mo and V that promote carbide formation demands a careful process
control.
For each applied alloy chemistry there exists a critical value of the cooling rate which sets the
lower limit for the formation of a ferritic-martensitic microstructure. Applying cooling rates
lower than this value results in a ferritic-pearlitic microstructure while higher cooling rates are
capable of producing a microstructure consisting of martensite islands in a ferritic matrix. The
overcooling degree is responsible for the martensite fraction in the final product. Equivalently,
the critical cooling rate (which as previously shown is prompted/necessitated by the production
line capabilities) defines a minimum of alloying content for the dual-phase steels.
Based on experimental results and accepting a similar alloying behavior of Cr and Mo as Mn
(mainly due to the hardenability effects), Tanaka et al. [14] expressed the influence of alloying
elements on the critical cooling rate in terms of a manganese equivalent, Mneq [%]:
log CR [K/s] = 1.73 (Mn)eq [%] + 3.95

(1.1)

where
(Mn)eq = (Mn) + 1.3 (Cr) + 2.67 (Mo) [%].

(1.2)

Though alloying elements may look versatile via equation eq. 1.1, their effects on microstructure
and deformation behavior are not necessarily the same. It is clearly shown that the higher the
amount of alloying elements in the steel the lower is the critical cooling rate necessary to form
a dual-phase microstructure. This explains the reason why batch-annealing processes require
much higher alloying additions. Practically, the critical cooling rate is easily determined with
a series of simple cooling experiments for each individual alloy.
To understand the formation of the ferritic-martensitic microstructure and to be able to interpret the products of the heat treatments it is essential to consider the phase transformations
taking place during heating, intercritical annealing and quenching. The formation of austenite in low carbon C-Mn steels was studied by a number of investigators [2933]. In the case
of cold-rolled ferritic-pearlitic steels, the recrystallization of the cold-worked microstructure is
completed already before reaching the annealing temperature, even during the rapid heating
rates applied on most continuous annealing lines. By entering the intercritical two-phase region, austenite nucleates rapidly at pearlite or in the vicinity of cementite particles and grows
rapidly until the carbides are dissolved. A slower growth of austenite into ferrite is continued
at a rate initially controlled by carbon diffusion in austenite and finally by manganese diffusion
in austenite, until the system reaches the equilibrium state. Practically, due to the very short
annealing times, only carbon redistribution between the phases takes place, because substitutional manganese diffuses much more slowly than interstitial carbon. This effect is referred to
as paraequilibrium.
Assuming that no manganese redistribution occurs, then a vertical section corresponding to
paraequilibrium conditions can be constructed for constant Mn content. Bearing that remark
in mind, Figure 1.3 demonstrates the production concept/scheme of a dual-phase steel by
intercritical annealing. According to the lever rule, for any given carbon content (C0 ) the
amount of austenite increases with increasing the intercritical temperature, becoming equal
to 100 % at the Ac3 -temperature while, as a consequence, the carbon content of austenite
decreases, reaching its minimum value (C =C0 ). In an analogous manner, for any given intercritical annealing temperature the amount of austenite increases with increasing alloy carbon
content, becoming equal to 100 % at a carbon content corresponding to the / + boundary
(C0 =C ). Additional alloying elements may cause some changes in the austenite formation
process and/or even widen or tighten the intercritical temperature field. Such alloying effects
can be roughly estimated by thermodynamical codes (e.g. ThermoCalc) but remain beyond
the scope of this work.

Figure 1.3: Production of dual-phase steels by intercritical annealing. The equilibrium fractions
of austenite and ferrite as well as their carbon content at the annealing temperature can be easily
estimated by applying the lever rule.

The importance of intercritical annealing becomes apparent, since for a given alloy composition and for a pre-selected annealing temperature the maximum amount of austenite that
can be (ideally) transformed to martensite as well as its carbon content -i.e. its hardenabilityare already pre-determined. Therefore, high intercritical annealing temperatures result in high
austenite fractions of decreased hardenability while low annealing temperatures result in low
austenite fractions with increased hardenability.
Even though the products of the austenitic transformation are strongly dependent on the
intercritical annealing parameters, the cooling rate is the final decisive step for the production
of dual-phase steels. The combined influence of cooling rate and intercritical annealing on
the formed microstructures was studied individually by many investigators [26, 28, 3436]. In
each case, a very important parameter that should not be forgotten is the effect of alloying
elements on the stability of austenite, which can be qualitatively measured by means of the
martensite-start temperature (MS ). For low carbon steels it was proposed by Andrews [37]
that:
MS [ C] = 539 423 C 30.4 Mn 17.7 Ni 12.1 Cr 7.5 Mo

(1.3)

or similarly by Eldis [38] for dual-phase steel compositions (also in wt. %):
MS [ C] = 531 391.2 C 43.3 Mn 21.8 Ni 16.2 Cr.

(1.4)

Quenching with high cooling rates from low intercritical temperatures ensures a complete
transformation of austenite into martensite because of the austenite stability; by applying
relatively low cooling rates the intercritically formed austenite partly transforms into ferrite,
enriching further the remaining austenite with carbon, increasing its hardenability and lower
even more the MS -temperature. In this case, isolated retained austenite particles may be
detected in the microstructure in the form of interlath films. However, the formation of
ferrite-carbide aggregates is usually unavoidable. Rapid quenching from high intercritical
temperatures produces even more complex effects: the decreased austenite hardenability
and the absence of time for the necessary diffusions during cooling, both reflected in a high
MS -temperature, result in the formation of autotempered martensite.
Dual-phase steels exhibit a number of superior mechanical properties, such as continuous
yielding behavior, low 0.2 % offset yield strength, high work hardening rate, high tensile strength
and remarkably high uniform and tensile elongations. The mechanical properties of dual-phase
steels arise from structural features, that is the fine dispersion of hard martensite particles
in a ductile ferritic matrix and all the related phenomena that accompany this coexistence.
The yielding and the work hardening behavior have been interpreted in terms of the high
dislocation densities and residual stresses arising in ferrite, as a consequence of the volume
expansion associated with the austenite to martensite transformation. The strength of dualphase steels was found to be dependent primarily upon the volume fraction and the carbon
content of martensite; solid solution strengthening of ferrite may also contribute to strength.
The excellent ductility reported for most of the dual-phase steels is the combined result of
many factors. Among them are the volume fraction and the carbon content of martensite, the
ductility of martensite, topological parameters such as the martensite grain distribution in the
ferritic matrix, the alloy content of ferrite, the dislocation density in ferrite, the presence of
carbides and/or retained austenite. Additionally, lattice imaging from Koo and Thomas [24]
has revealed a good coherency at the ferrite/martensite interface, which prevents decohesive
interface failure during deformation and thus enables the full toughness of ferrite to be realized.
Tempering may be applied as part of the process in some continuous annealing lines, after
water-quenching from the intercritical temperature, to regulate the properties of the dual-phase
steel. It may also be an unavoidable side-effect of an operation, e.g. in a hot dip galvanizing
line. Finally, tempering may be useful in some production practices such as bake hardening
(paint baking). The change in yield strength upon tempering is complex because of the relief
of residual stresses, carbon segregation to dislocations and the return of discontinuous yielding.
After tempering at low temperatures the yield strength increases but discontinuous yielding
returns only to the steels containing low volume fractions of martensite. When tempering at
high temperatures the yield strength decreases but discontinuous yielding appears in all steels.

The tensile strength decreases, while post-uniform and uniform elongations increase due to the
change in hardness of martensite.
There has been a long discussion on the construction of a model correlating the mechanical
properties of dual-phase steels with their microstructural characteristics. Simple empirical rules
of mixtures [2628, 3944] (usually linear regressions between the constituent phases properties, based on Mileikos theory of composites [45]) as well as more complicated/sophisticated
micromechanical models [4651] (introducing the importance of the effective in-situ properties and topological microstructural parameters) have been developed, most (if not all)
of them based on individual sets of experimental data. In some cases, there was a good
agreement between the predicted properties and the experimental results -even for martensite
fractions in the order of 80 %, in other cases some fair and well-established modifications have
to be made. Since each study refers to a specific alloy composition and a different kind of
heat treatment, a comparison between the obtained results contributes to the disagreements
reflected in literature over the past 25 years.
In order to meet the different design requirements of individual automobile-body components
for strength, crashworthiness, energy absorption, part complexity and dent resistance, a variety
of dual-phase grades exhibiting different strength and ductility levels is currently industrially
produced. Despite the numerous studies on the relationship between the mechanical properties
and the microstructural characteristics of dual-phase steels over the last decades, the challenge of increasing their formability at a constant strength level (or equivalently increasing the
strength while maintaining a high ductility) remains still unanswered.
The statement of many researchers that for the improvement of properties of dual-phase
steels the ferrite should be fine-grained, free of ultrafine carbide precipitates and strengthened solely by alloying additions which have minimum effects on ductility is still not fully
explained/affirmed.

10

1.2

Grain refinement

The formation of ultrafine-grained ferritic microstructures in low-alloyed and plain carbon steels
has been intensively investigated during the last decade, since grain refinement is expected to
have a beneficial effect on the mechanical properties of the steel. Though the term ultrafine
is somewhat vague, it reflects the objective of the Japanese super metal project of producing
a strip having a ferrite grain size of 1 m throughout a minimum thickness of 1 mm [52].
According to the Hall-Petch relation (which is applicable to a variety of polycrystalline singleand dual-phase metals), a decrease in grain size (d) results in an increase in yield strength (y ):
y = 0 + ky d 1/2 .

(1.5)

For a low-alloyed steel, for example, a decrease of the ferrite grain size from 5 m to 1 m would
ideally produce an increase of the yield strength by approximately 300 MPa. Additionally, it
has been reported that grain refinement improves the fatigue resistance of steels [53] and can
lead to superplastic behavior at high temperatures and appropriate strain rates [54].
The currently available techniques to obtain ultrafine grains are rapid solidification directly
from the melt, vapor deposition, cryogenic metal-forming, mechanical alloying and severe
plastic straining. Very small grains (with sizes in the nano-scale range) may be produced under
extreme conditions, often leading to impressive physical and mechanical properties. However,
due to the the limited production quantities and the very small grains individually formed
(powder-like), further consolidation and processing is required to produce a bulk material suitable for structural use. Concerning the refinement methods, there exists a permanent conflict
between the achievable grain size in a material, the amount or the dimensions of the material
that can be processed in this way and -the most important- the cost of processing. In the
case of steel applications, an optimum compromise between these factors would be required [55].
Grain refinement in steels can be realized by microalloying. The addition of elements such as
Al, B, P, Sm or Ti in the microstructure can suppress the grain growth of ferrite, utilizing the
pinning effect of secondary phase particles and/or the dragging effect induced by solute atoms
[56]. However, the idea of controlling the grain size and hence the mechanical properties by
thermomechanical processing instead of the classical way of alloying is far more attractive, as
this would result in the production of steels with simpler chemistries and improved recyclability
and would lead to economic benefits as well.
To achieve steel grain refinement, substantial efforts involving severe plastic deformation
(SPD) by using conventional rolling equipment have been made. According to the dynamic
Strain Induced TRansformation method (SITR), proposed by [5761], grain refinement is pro11

moted by the continuous intragranular nucleation of numerous small ferrite grains during hotor warm-rolling of austenite. The process involves a single-pass rolling of the steel strip at
a temperature just above the Ar3 (i.e. immediately above the temperature at which grain
boundary pro-eutectoid ferrite would begin to form) but below the Ae3 , followed by air-cooling
or accelerated quenching depending on the desired microstructure. The required rolling reduction ranges between 35 and 40 %. Ultrafine ferrite (UFF) grains with dF < 2 m form on the
surface layers of the strip (reaching a depth of 1/4- 1/3 of its thickness). In the core of the
strip forms either a mixture of carbides, bainite and coarse ferrite grains (air-cooling) or tempered martensite (dual-phase steel formation caused by accelerated cooling). The Hot Torsion
method is based on the similar SITR concept, with the difference that the deformation mode
involves torsion at elevated temperatures instead of simple rolling [62, 63]. In each case, a very
precise temperature control is required, thus allowing a relatively narrow thermomechanical
production window. The inhomogeneity of the microstructure in thickness remains one of the
main drawbacks of the procedure. Tensile testing has revealed that UFF steels obtained from
this production route exhibit very high lower yield stress to tensile strength ratios, approaching
0.90-0.95. The complete lack of strain hardening inhibits their application in forming processes.
An alternative suggestion was made by Ueji et al. [64, 65], introducing conventional coldrolling and subsequent annealing of a martensitic start microstructure. The method was applied
to plain low carbon steels, undergoing a 50 % cold-rolling reduction and then was warmannealed at 823 K (550 ) to produce a multiphase ultrafine microstructure consisting of UFF
grains and uniformly precipitated fine carbides. Blocks of tempered martensite were occasionally observed. The initial martensitic microstructure and the formation of fine carbides during
warm annealing were identified as the keys to success of the process. Annealing at higher temperatures (700 ) proved to be detrimental for the mechanical performance of the material, by
producing properties similar to a ferritic-pearlitic steel.
The production of fine-grained alloys on the micro-meter scale by conventional thermomechanical processing is limited by the achievable equivalent strains, which are typically in the
order of 3-4, if the final product has to exhibit a minimum thickness of 1 mm. Several novel
processing routes that allow for larger strains and are based on the concept of accumulating unlimited strain in the material are currently under development on a laboratory scale.
Figure 1.4-a describes the procedure followed in Torsion under Hydrostatic Pressure (or High
Pressure Torsion, HPT). A thin disc is deformed in torsion using the friction provided by the
application of a large hydrostatic pressure. By imposing complex strains in the specimen, very
large equivalent strains in the order of 7 can be induced in the material, resulting in grain sizes
of 2 m produced at room temperature [55, 67].
Equal Channel Angular Pressing (ECAP) can be considered as a rapidly developing tech-

12

(a) HPT

(b) ECAP

(c) ARB

Figure 1.4: Schematic illustration of the recently developed methods for severe plastic deformation
(HPT and ECAP from [55], ARB from [66]).

nique for obtaining ultrafine-grained microstructures [6871]. According to the method, the
material is pressed through a die, where two channels form an L-shaped configuration with an
angle of 2 (Figure 1.4-b). The process imposes a severe strain on the sample by means of
shearing. Since there is no concomitant change in the cross-sectional dimensions of the samples,
repetitive pressings can produce very high effective strains (in the order of 10), achieving homogeneous grain refinement with grain sizes on the micro-meter scale. Moreover, the sample is
constrained so also less ductile materials can be processed. The factors influencing the method
are the pressing route by which the sample is rotated during the successive pressings, the die
angle which determines not only the strain per pass but also the geometry of deformation,
the die cross-section geometry, the speed and the temperature associated with pressing. Very
recent investigations revealed that ultrafine-grained ferritic-martensitic dual-phase steels can
be fabricated by ECAP, by applying an effective strain of 4 at 500 and subsequent intercritical annealing at 730 for 10 min. UFF grains with uniformly distributed blocky martensite
islands of 1 m were produced [72]. Despite the scientific interest and the partial success, the
application of the ECAP method is still restricted in terms of commercial capability while even
the future prospects for steel sheet production are questionable. The potential of up-scaling
ECAP is being currently investigated.
Accumulative Roll Bonding (ARB) is also a newly developed technique to realize intense
plastic straining [66, 73, 74]. Following the illustration of Figure 1.4-c, one strip is neatly
placed on top of another strip and the two layers of material are joined together by warm
rolling. The rolled sample is then cut in two halves, which are again stacked together after
an appropriate surface preparation and are roll-bonded. The whole process can be repeated
without any change in the samples thickness and geometry, given that the reduction in thickness

13

after each rolling pass is maintained/controlled to 50 %. This process can theoretically introduce
unlimited amounts of strain in the material. Effective strains of 8 have been reached for
aluminium and steels, resulting in grain sizes below 1 m. Critical factors for the success of the
method are surface preparation and cleaning, the deformation temperature and the amount of
induced strain. Although rolling at elevated temperatures is advantageous for joinability, too
high temperature would cause dynamic recrystallization and cancel the effect of accumulated
strain.
All the above proposed techniques are very attractive but seem to encounter several difficulties
in engineering applications. In each case, very high levels of strain or unrealistic inter-pass
times are required, complex processes and special equipment are involved and the production
capacities are still very low to cover the steel market demands. As a combined result of these
factors, the application of the novel refining methods in a continuous industrial process does
not seem possible in the near future.

14

Chapter 2
Material production and experimental
procedure
2.1

Production of the material

The dual-phase steel investigated in this work was industrially produced by voestalpine Stahl
Linz. Its chemical composition is given in Table 2.1. Slabs with a thickness of 210 mm were
produced on a continuous casting machine. The slabs were reheated in a pusher-type furnace
to a temperature of 1250 and hot-rolled to a final thickness of 2.40 mm. The finishing
temperature was approximately 900 while the coiling temperature was about 600 . Part
of the hot-rolled strip was milled to remove the surface scale and then cold-rolled to 1.00 mm
(cold-reduction of 58 %).
Table 2.1: Chemical composition of the investigated DP-steel.

DP steel
wt. %

C
Si
0.1 0.15

Mn Cr+Mo
1.5
0.8

Fe
Bal.

In the main bulk of this work strips from the hot-rolled ferritic-pearlitic material were used,
cut to specimens of 250 mm in length and 50 mm in width in order to satisfy the requirements
of the laboratory annealing and cold-rolling equipment. The industrially cold-rolled strip was
mainly used as reference material (denoted as R in the following) with regard to its response
to the same heat treatment schedules applied to the pre-processed and laboratory cold-rolled
material.

15

2.2

Thermodynamical calculations

The phase diagram of the alloy as well as the critical temperatures where phase transformations
occur were initially calculated with the help of the program ThermoCalc, by using the database
TCFE3 and considering ferrite, austenite and cementite as the only phases present in the steel.
The intercritical + two-phase temperature range was calculated between 710 and 815 .
Additionally, the fraction of austenite at different intercritical annealing temperature steps was
determined, in a first attempt to set the maximum martensite fraction after cooling down from
the annealing temperature (Figures 2.1-a and 2.1-b). It should be taken into account that
all estimated temperatures and phase fractions represent thermodynamic equilibrium.

(a)

(b)

Figure 2.1: Thermodynamic calculations: (a) concentration section through the phase diagram of
the investigated dual-phase steel (the dashed line indicates the carbon content) and (b) equilibrium
fraction of austenite during intercritical annealing.

2.3

Pre-processing

The first critical stage of the experimental procedure, described with the term pre-processing,
involves annealing of the as hot-rolled material in a Carbolite Three Zone Tube furnace (TZF)
with an adapted inert gas (Ar) supply to avoid prolonged oxidation. The specimens were
annealed at three different temperatures (760, 800 and 900 ) with a holding time of 5 min in
16

an Ar atmosphere, followed by water-quenching to form a ferritic-martensitic (in the first two


cases) or a completely martensitic microstructure (Figure 2.2). The annealing temperatures
were selected by taking into account the thermodynamical calculations and the holding time
was set long enough to assure thermodynamic equilibrium.

(a) DP I

(b) DP II

(c) M

Figure 2.2: Micrographs of the pre-processed materials before laboratory cold-rolling, etched with
LePera.

In the following chapters, the pre-processed materials-specimens will be denoted as DP I,


DP II and M, according to the annealing temperature during pre-processing (Table 2.2).
Table 2.2: Denotation of pre-processed material.

Grade
DP I
DP II
M

2.4

Pre-processing conditions
Tan = 760 , tan = 5 min, WQ
Tan = 800 , tan = 5 min, WQ
Tan = 900 , tan = 5 min, WQ

Martensite fraction (%)


23 %
40 %
100 %

Cold-Rolling trials

The ferritic-martensitic or completely martensitic microstructures produced during preprocessing were cold-rolled in a Carl-Wezel laboratory mill with a roll diameter of 220 mm
at a roll peripheral speed of 16 mmin1 to a final thickness of 1.00 or 0.80 mm (cold reduction
of 58 % or 67 %, respectively). To avoid cracking of such work hardened microstructures, rolling
was performed in multiple passes.
The microstructures of the as hot-rolled ferritic-pearlitic starting material after industrial
cold-rolling as well as that of the pre-processed material after laboratory cold-rolling are shown
in Figure 2.3. There, the microstructures differ not only qualitatively but also quantitatively
17

from each other. Micrograph (a) shows a rather coarse-grained ferritic-pearlitic microstructure,
while in micrographs (b) and (c) dual-phase ferritic-martensitic microstructures are shown,
possessing different martensite fractions due to the initial annealing conditions. The higher
the martensite fraction after pre-processing the finer is the obtained microstructure after
cold-rolling. This is also supported by the last micrograph (d), where a very fine martensitic
microstructure is shown.

(a)

(b)

(c)

(d)

Figure 2.3: (a) Industrially cold-rolled ferritic-pearlitic microstructure of the as hot-rolled starting
material (R), (b) Cold-rolled ferritic-martensitic microstructure, initially annealed at 760 (DP I),
(c) Cold-rolled ferritic-martensitic microstructure, initially annealed at 800 (DP II), (d) Cold-rolled
completely martensitic microstructure (M), etched with Nital.

2.5

Dilatometric investigations

The influence of various annealing cycles on the microstructural evolution of the cold-rolled
material (with respect to grain refinement) was studied via dilatometry, with the samples cut
18

in the rolling direction (10 x 4.0 x 1.0 or 0.80 mm, depending on the sheets thickness). The
dilatometric investigations were conducted on a Bahr dilatometer DIL 805 A/D. The selection
and the appropriate combination of the annealing parameters within a cycle (heating rate, annealing temperature, holding time in the intercritical zone as well as the cooling rate) were made
with regard to the special features of the cold-rolled material, that is the ferritic-martensitic or
the pure martensitic microstructure and the deformation degree. The basic annealing schemes
are presented in Figure 2.4.
900
T

; t

an

temperature (C)

temperature (C)

900
an

600

300

25 K/s

C : 5-80 K/s

600

300

0
time (s)

(a)

time (s)

(b)

Figure 2.4: (a) Intercritical annealing, (b) Repeated heating cycles oscillating between the + and
phase fields.

2.6

Annealing simulations and austenitization kinetics

All annealing simulations were conducted in the laboratory with the Multi-Purpose Annealing
Simulator (MULTIPAS) at voestalpine Stahl Linz on the cold-rolled pre-processed ferriticmartensitic and martensitic as well as on the reference material. Special attention was paid
to the intercritical annealing heat treatments due to their applicability on an industrial scale.
Specimens with a thickness of 1.00 mm were preferred for this purpose, because they offer
the possibility of direct comparison with the standard industrially cold-rolled material (R,
reference).
Although the phase transformation temperatures were calculated with ThermoCalc and the
recrystallization kinetics was investigated via dilatometry, additional annealing simulations were
conducted on the reference material to determine the austenitization kinetics of the alloy in
non-equilibrium conditions on a larger scale. Specimens were annealed in a temperature range
that covers both + two-phase and single-phase regions for different representative holding
times and then were water-quenched.
19

2.7

Microstructure investigations

A light microscope (Olympus AX70) was used for the microstructural investigations of the
heat treated samples. To reveal the microstructure of the fine-grained materials the specimens
were conventionally prepared and then etched with LePeras etchant, which is a mixture of 1 %
sodium metabisulfite in distilled water and 4 % picric acid in ethanol in a 1:1 volume ratio.
This tint etching technique allows the distinction of phases by coloring, staining ferrite brown
and/or blue, bainite dark brown to black while martensite and retained austenite (hardly any
present in the materials of this work) remain white. Quantitative characteristics of the microstructure such as phase fractions and mean grain sizes of the constituents were determined
by line intercept measurements, considering ferrite as the dominating matrix phase and characterizing martensite, bainite and/or retained austenite as a second phase. For severely deformed
microstructures after cold-rolling and before heat treatment or for annealed specimens where
LePeras etching agent could not produce the desirable effects, a 2 % Nital etchant (2 % nitric
acid in ethanol) turned out to be an acceptable alternative.
In cases that a higher magnification analysis beyond the resolution capacity of light microscope was required, e.g. for the investigation of ultrafine-grained microstructures or for the
identification of any third phase present (like bainite), Scanning Electron Microscopy (SEM)
was employed. The SEM observations were conducted on a LEO 1450 SEM and/or on a TOPCON SM-520 Field Emission Gun (FEG)-SEM on demand.
Microstructures of selected specimens representing critical heat treatments were further investigated by means of Transmission Electron Microscopy (TEM). TEM-specimens (thin foils)
were prepared by gentle mechanical thinning down to 80 m followed by electrolytic thinning
in 5 % perchloric acid in acetic acid. The thin foils were analyzed in a Philips CM20 STEM
transmission electron microscope, applying an accelerating voltage of 200 kV. For a first rough
overview of the microstructure, secondary electron (SE) images were taken from the thin foils.
The detailed analysis of the phases was carried out using bright and dark field techniques, while
the identification of the phases was performed using electron diffraction patterns.

2.8
2.8.1

Mechanical testing
Tensile testing

The mechanical properties of the steels were measured on a Roell-Korthaus RKM 250 testing
machine, according to European standard EN 10 002. All tensile specimens were machined
with their tensile axis parallel to the rolling direction and with a gage length of 80 mm. The
laboratory produced grades (DP I, DP II and M) were tensile tested in the as-annealed condition
while the industrially cold-rolled steel (R) was submitted to skin pass rolling of 0.5-1.0 % to
20

improve roughness.

2.8.2

Ultramicrohardness and microhardness testing

The ultramicrohardness of the main constituents of the dual-phase microstructure (ferrite


and martensite) was measured with a Kammrath & Weiss ultramicrohardness tester UMHT-3
equipped with a Vickers diamond square pyramid. The device was mounted in a LEO 1450
Scanning Electron Microscope (SEM). To identify the phases, the specimens were etched with
Nital etchant. The applied indentation load (15 mN) as well as other critical measuring parameters like indentation time and speed were selected to be the same for all measurements, so that
a comparison of the hardness between different phases and various heat treatments becomes
possible. Furthermore, to eliminate any possible errors derived from the optical measurement
of the indentation diagonals, all indentations and measurements in SEM were performed at the
same magnification (12000). Finally, the hardness was calculated from eq. 2.1:
HV =

189 F
,
d2

(2.1)

where d [m] stands for the mean length of the indentation diagonals and F [mN] for the
indentation load.
Additionally, the Vickers microhardness of the investigated dual-phase steels was determined
with a Micro-Duromat 4000 E microhardness tester from Reichert-Jung mounted in a light microscope (Reichert Metaplan 2, Leica AG). The indentation load was set to 150 p (approximately
1471 mN), producing Vickers impressions which are bigger by one order of magnitude compared
to ultramicrohardness measurements.

2.8.3

Hole expansion measurements (Stretch flangeability)

Hole flanging is a process widely applied in thin-sheet forming operations, which employs a
punch for producing structural parts with short necks that are subsequently used for assembly
with other components. During stretch flanging, the deformation mode at the edge of the hole
is a combination of bending and stretching which in some cases causes splitting failure and
therefore cannot be grasped by the conventional uniaxial tensile test.
The hole expansion behavior provides a way to measure the tendency of steels to split as a
hole is expanded under external forces and is characterized by the percentage increase in the
size of the hole at the moment that a crack forms [7579]. A schematic diagram of the hole
expansion equipment is shown in Figure 2.5. Selected dual-phase steel grades were cut to
125 mm 125 mm square test pieces and before testing a 12 mm (0.15 mm) hole was punched
into the centre of each sample. The hole expansion test is conducted by expanding the initial
21

punched hole using a 50 mm diameter punch; the punch driving is immediately stopped when
any crack (which extends all through the samples thickness) is observed at the edge of the hole.
The final diameter of the hole is measured by averaging two readings taken perpendicularly to
each other. The property is expressed as the ratio of the expanded hole size to the original hole
size, as defined by the following equation:
=

df d0
100,
d0

(2.2)

where [%] is the hole expansion ratio, df [mm] the average final hole diameter (after rupture),
and d0 [mm] is the initial hole diameter.

Figure 2.5: Schematic illustration of hole expansion testing equipment of voestalpine Stahl Linz

22

Chapter 3
Results
3.1

Microstructure investigations

3.1.1

Recrystallization/austenitization kinetics

In order to study the austenitization/recrystallization kinetics regarding cementite dissolution


and austenite formation, specimens from the industrially cold-rolled material (R) were subjected
to heat treatments according to the annealing plan of Figure 3.1 which involves annealing not
only in the intercritical two-phase + ferritic-austenitic field but also in the pure austenitic
region. Five annealing temperatures covering a range of 100 in temperature intervals of 25
for holding times of 0 s, 10 s, and 100 s were applied, while the heating rate was set to 25 K/s
to reproduce the industrial conditions. After water-quenching the microstructure of the steel
was investigated.

1.0
0.9

850

0.8

Martensite fraction

Annealing temperature (C)

900

800
750
WQ

WQ

WQ
750 C

700

775 C
800 C
Heating rate: 20 K/s

650

750 C

0.6

775 C

0.5

800 C
825 C

0.4

850 C

0.3
0.2

825 C
850 C

0.1

WQ: water-quenching

600

0.7

20

40

60

80

100

120

140

0.0

160

20

40

60

80

100

120

Annealing time (s)

Annealing time (s)

Figure 3.1: Annealing plan to determine the


austenitization kinetics of the as cold-rolled
dual-phase steel.

Figure 3.2: Influence of annealing temperature and annealing holding time on the fraction of martensite.

23

The fraction of martensite formed during water-quenching was determined by line intercept
measurements, providing the amount of the former austenite prior to the martensitic transformation. Figure 3.2 shows the influence of the annealing temperature for different holding
times on the martensite fraction. Although the martensite fraction increases dramatically in
the first 10 s of annealing, a further increase of the holding time has no significant effect except
for the annealing temperature of 775 . Figure 3.3 shows the microstructure of the steel (R)
as a function of annealing temperature for a holding time of 100 s. It is remarkable that for
annealing temperatures over 800 (even though the calculated + transformation
temperature is 815 ) the martensite fraction approaches the maximum value of 1.0.

(a) Tan = 750

(b) Tan = 775

(d) Tan = 825

(c) Tan = 800

(e) Tan = 850

Figure 3.3: Influence of the annealing temperature on the martensite fraction (former austenite
fraction before water-quenching) for a holding time of 100 s (LePera).

Figure 3.4 shows the microstructure of the steel after water-quenching from the annealing
temperature without a holding stage. For the lower annealing temperature (Figure 3.4-a), the
recrystallization is not completed and undissolved carbides are still detected in the microstructure. At higher annealing temperatures no carbides could be detected. It should be also noticed
that even after water-quenching from high annealing temperatures (825 ) a significant amount
of ferrite is present in the microstructure (Figure 3.4-d).
The zero holding time (0 s), which eliminates the possibility of equilibrium, provides valuable
data/information for the case that heat treatments schedules are applied involving flashing
24

heating and cooling cycles (see Figure 2.4).

(a) Tan = 750

(b) Tan = 775

(d) Tan = 825

(c) Tan = 800

(e) Tan = 850

Figure 3.4: Influence of the annealing temperature on the martensite fraction (former austenite
fraction before water-quenching) for a zero holding time (LePera).

The martensite fraction data are appended to the ThermoCalc diagram in order to compare
the experimental results with the numerical calculation. It was assumed that the martensite
fractions measured for a holding time of 100 s approach equilibrium status and additionally
that the amount of martensite represents the former austenite fraction during annealing. As
can be seen in Figure 3.5, the results are in good agreement with each other.

25

Figure 3.5: Experimental data from martensite fraction measurements (square symbols) and equilibrium fraction of austenite (line) as calculated with the program ThermoCalc, both given as a function
of annealing temperature.

3.1.2

Dilatometric investigations

The laboratory developed materials (DP I, DP II and M) were not submitted to the above waterquenching process, since their recrystallization and austenitization kinetics are expected from
their design concept to be even faster. Nevertheless, systematic dilatometric investigations were
done in order to study the influence of the critical annealing parameters on the microstructure
of these grades, aiming to clarify the recovery-recrystallization-grain growth mechanism and
finally to determine the optimum annealing conditions which lead to grain refinement.
Conventional annealing
Conventional heat treatment schedules following the annealing scheme of Figure 2.4-a were
applied to the laboratory cold-rolled microstructures. A relatively high heating rate of 25 K/s
from room to annealing temperature was applied, in order to simulate the industrial conditions
and additionally to minimize the time available for grain growth after recovery and recrystallization. Three different annealing temperatures (Tan = 750 , 800 and 840 ) were chosen
for the dilatometric investigations. According to the thermodynamical calculations, the first
two temperatures are located in the intercritical + while the third one in the austenite
phase field. However, the ferrite to austenite phase transformation as indicated by the dilatation curves during the heating stage seems to be completed only at temperatures above 840
(see Figure 3.6). This important observation which represents the in-situ measurement of
the + transformation temperature does not actually contradict the results from the
26

austenitization experiments, since in the measurement of the dilatation curve no holding time
is taken into account. This simply means that a temperature of 840 refers to intercritical
annealing conditions for a zero holding time and to austenitic annealing in case that a holding
time stage is applied. To avoid any possible misinterpretations in the discussion of the results,
Tan of 840 will be considered as austenitic annealing in the following pages.

Change in length (m)

120

110

100

90

T
650

750

850

950

Temperature (C)

Figure 3.6: Dilatation curve during heating. T indicates the temperature at which the ferrite to
austenite transformation is completed.

Holding times in the range between 5 s and 120 s were applied during intercritical annealing.
Recovery and recrystallization of the severely deformed microstructures were completed within
the first 30 s of holding stage for all materials investigated. The cold-rolled martensitic grade
(M) is already recrystallized at an even shorter holding time of 10 s. Microstructural observations revealed that holding times longer than 30 s are rather detrimental for the formation of an
ultrafine dual-phase microstructure, by leading to an undesirable grain growth (mainly of the
ferrite grains). Figure 3.7 demonstrates the influence of holding time on the microstructure of
the grade M, intercritically annealed at 800 and quenched with 40 K/s to room temperature.
The specimens are treated with Nital etchant, which stains martensite brown or dark brown
and ferrite light brown to light cream/beige.
Recrystallization is completed even for the shortest annealing time applied. The design and
the production history of the grade M as well as the high cooling rate after annealing eliminate
the possibility of presence of any carbide phases in the final microstructure. Holding time does
not seem to affect the fraction of martensite maintained in the steel after quenching but it is
proved to be a dominating parameter for the achievement of ultrafine-grained microstructures.
A holding time of 30 s provided the optimum solution with respect to grain refinement, as shown
in Figure 3.7-d. Further increase of the annealing time, for example by only 30 s, resulted to
27

(a) tan = 5 s

(b) tan = 10 s

(c) tan = 20 s

(d) tan = 30 s

(e) tan = 50 s

(f) tan = 60 s

Figure 3.7: Microstructures of the martensitic grade M after annealing at 800 for the subscribed
holding times and quenching with a cooling rate of 40 K/s, etched with Nital.

28

rapid grain coarsening of ferrite almost by an order of magnitude as shown in Figure 3.7-f.
Analogous observations regarding the time-dependence of grain refinement at certain annealing
conditions were also made for all laboratory grades investigated.

(a) Tan = 750

(b) Tan = 800

(c) Tan = 840

Figure 3.8: Microstructures of the martensitic grade M after annealing at different annealing temperatures for 30 s and quenching with a cooling rate of 40 K/s, etched with Nital.

Based on the experimental determination of an optimum holding time for grain refinement
(30 s in our case), the influence of annealing temperature on the microstructure was additionally studied in the dilatometer before moving to a larger experimental scale. As shown in
Figure 3.8, annealing at higher temperatures leads to higher martensite fractions in the microstructure, consistent with the increased austenite fraction at these conditions. The difference
in martensite fraction between Figure 3.8-b and Figure 3.8-c, referring to annealing temperatures of 800 and 840 , respectively, is not clearly distinguishable, since for the holding time
applied the austenite fraction in both cases approximates the maximum. It should be noticed
that this temperature dependent increase of the martensite fraction is accompanied by a pronounced alteration of the morphology and subsequently of the etching behavior of martensite
grains. Annealing at lower intercritical temperatures (e.g. at 750 , Figure 3.8-a) produces
clear unstructured brown martensite, while higher annealing temperatures (Figures 3.8-b and
3.8-c) lead to the formation of dark-brown structured and generally coarser martensite (the
tints refer to the etching effect of Nital agent).
In all dilatometric investigations described above, a moderate cooling rate of 40 K/s was
applied in order to select the most favorable conditions/parameters for grain refinement with
respect to annealing temperature and annealing time. Once this became clear, the influence
of cooling rate on the microstructural evolution was thoroughly examined. Greater attention
was paid to the grade M because of its starting microstructure peculiarity in comparison with
the other grades. In order to produce a wider possible range of dual-phase microstructures
depending on the cooling rate parameter, six cooling rates, that is 5, 10, 20, 40, 60 and 80 K/s,
were applied. Higher cooling rates were not achievable along the entire cooling stage, due
29

to the restricted cooling capacity that the dilatometer -associated with the cooling gas used
(nitrogen)- could provide.
Figure 3.9 demonstrates the influence of cooling rate on the microstructure of the martensitic
grade M, after annealing at 840 for 30 s. By applying cooling rates below 10 K/s a rather
coarse multi-phase microstructure, containing ferrite, martensite and a third phase indicated
by the black tinted regions located on the ferrite grain boundaries (most probably bainite)
is formed. Grain refinement is achieved only at cooling rates higher than 40 K/s, which is
sufficient to avoid the presence of the third phase. Further increasing of the cooling rate results
also in ultrafine ferritic-martensitic microstructures with increased martensite fractions.
These observations are only qualitatively common to all grades (DP I, DP II and M). More
details about the impact of conventional heat treatments on the microstructure, with regard
to the individual annealing behavior of each laboratory grade investigated, will be described
and discussed in the section of larger scale and more comprehensive annealing simulations (see
3.1.3).
Non-conventional annealing
Non-conventional heat treatment schedules following the concept described in Figure 2.4-b
were applied to the laboratory cold-rolled material, particularly to the thinner grades (thickness of 0.80 mm, corresponding to 67 % cold reduction), where recrystallization kinetics is expected to be extremely fast due to the higher cold deformation degree. The aim of these
heating-cooling cycles around the + transformation temperature was to provoke the
continuous formation of new grains, so to impede grain growth. By this repeated nucleation and
preventing the system from reaching an equilibrium state by rapidly changing the temperature,
an ultrafine-grained microstructure is finally obtained.
The selection of annealing parameters, i.e. the initial heating and final cooling rates, temperature range and heating/cooling rates for the + / - cycling had to be done for each
laboratory grade separately, thereby taking into account its production history and its peculiar
microstructural characteristics. The special case in which only one cycle is applied actually
represents a limiting condition toward conventional annealing with 0 s holding time. It was
observed that more than four cycles have no further beneficial effect towards grain refinement;
on the contrary, this could even lead to grain coarsening.

30

(a) CR = 5 K/s

(b) CR = 10 K/s

(c) CR = 20 K/s

(d) CR = 40 K/s

(e) CR = 60 K/s

(f) CR = 80 K/s

Figure 3.9: Microstructures of the martensitic grade M after annealing at 840 for 30 s and quenching with different cooling rates, etched with Nital.

31

(a)

(b)

(c)

(d)

Figure 3.10: Ultrafine ferritic-martensitic microstructures produced from the following laboratory cold-rolled grades: (a) DP II (0.80 mm), (b) DP I (0.80 mm), (c) M (1.00 mm) and (d) DP II
(1.00 mm). The applied heat treatments are given in the attached illustrations. Nital etchant.

Ultrafine homogeneous dual-phase steels produced by the thermal cycling technique are shown
in Figure 3.10. The specimens were etched with Nital: martensite appears dark brown to black
while ferrite remains light colored. Microstructures with a ferrite mean grain size between 2 and
3 m and with martensite grains generally finer than 1 m could be obtained. Thinner grades,
corresponding to a higher cold deformation degree, react more sensitive than the thicker grades
to the dramatic changes of annealing conditions, in such a way that one or two flashing cycles
were enough to achieve grain refinement (Figure 3.10-a and 3.10-b). The small number of
heating-cooling cycles is simply translated into a shorter total holding time in the intercritical
or in the pure austenitic region and, finally, results in the formation of fine microstructures
possessing a relatively low martensite fraction. This agrees well with the recrystallization
32

experiments described in 3.1.1, according to which for very short (or for zero) holding times
austenitization is not yet completed.
The in-situ variation of annealing parameters during heat treatment was successfully combined with the rapid recovery and recrystallization kinetics of the laboratory cold-rolled preprocessed grades, producing ultrafine dual-phase microstructures. Nevertheless, up-scaling the
steel production by this method faces serious difficulties, mainly due to the rapid temperature
changes necessary between the annealing segments and the complexity of controlling them with
the currently available industrial equipment.

3.1.3

Annealing simulations

Laboratory annealing simulations were conducted in order to reproduce the ultrafine microstructures on a larger scale and so being able to investigate mechanical properties. For
the final selection of annealing conditions the data collected from the dilatometric investigations were taken into account.
The influence of the annealing temperature and of the cooling rate on the microstructure
was studied for all laboratory and industrial grades, since these parameters seem to have the
strongest impact on the microstructure evolution. To avoid grain coarsening, the heating rate
as well as the annealing holding time were held constant for all conventional heat treatment
schedules at 25 K/s and 30 s respectively. Two annealing temperatures were selected, the one
in the intercritical region (800 , more or less standard in continuous annealing lines) and the
other in the austenitic region (840 ). Lower annealing temperatures (750 ) were excluded
from the simulations, because even though they provide with a variety of low martensitefraction steels they have a rather detrimental effect on grain refinement. For each annealing
temperature, a set of six cooling rates -the same as in the dilatometric investigations- was
applied, covering a wide range of microstructures regarding the formation of phases, the fraction
and the morphology of martensite and the mean grain size of ferrite.
Figure 3.11 shows the influence of the cooling rate on the microstructure of the laboratory
grade DP II after intercritical annealing at 800 for 30 s. The specimens are etched with
LePera. At cooling rates lower than 10 K/s a third dark/black colored phase (probably bainite)
is present in the microstructure, located on the grain boundaries and at the grain triple points
(Figures 3.11-a and 3.11-b). As the cooling rate increases, the third phase disappears while
the martensite fraction increases and, simultaneously, the ferrite mean grain size decreases
(Figures 3.11-c to 3.11-f ). Martensite grains remain fine for all cooling rates up to 40 K/s. In
the microstructures quenched with 60 K/s and 80 K/s also coarse martensite grains are found; in
the interior of these coarse grains a brown tinted substructure can be detected (Figures 3.11-e
and 3.11-f ).
An analogous investigation for the martensitic grade M (by keeping the same annealing
33

(a) CR = 5 K/s

(b) CR = 10 K/s

(c) CR = 20 K/s

(d) CR = 40 K/s

(e) CR = 60 K/s

(f) CR = 80 K/s

Figure 3.11: Microstructures of the grade DP II annealed at 800 for 30 s and quenched with
different cooling rates, etched with LePera.

34

parameters constant) revealed a more pronounced influence of the cooling rate on the formed
microstructures, with respect to both grain refinement and martensite fraction. As shown in
Figure 3.12 and in reference to Figure 3.11, a third phase could not be identified even for the
lowest cooling rates applied. Furthermore, coarse structured martensitic grains appear already
after cooling with 10 K/s, increasing in fraction/number with increasing cooling rate and finally
reaching grain sizes of the order of ferrite.
The cooling rate of 40 K/s proves to be a critical point in the direction of grain refinement. Comparing the microstructure in Figure 3.12-c with that in Figure 3.12-d, representing quenching with 20 K/s and 40 K/s respectively, the increase in martensite fraction as well
as the change in the martensite morphology - as indicated by its etching behavior - is distinguishable. This effect is followed by significant grain refinement of ferrite approximately by
an order of magnitude. Applying cooling rates higher than 40 K/s leads to higher martensite
fractions which are not accompanied with a proportional decrease in the grain size of ferrite.
The influence of the annealing temperature on the microstructure of the laboratory as well
as of the reference material is shown in Figure 3.13. For this purpose, a moderate cooling
rate of 20 K/s was applied in order to minimize the impact of the cooling rate on the formed
microstructures. At this condition, the industrial grade R reacts more sensitive to the annealing
temperature than the laboratory grades DP I and M. Increasing Tan from 800 to 840
results in a finer microstructure of R, possessing a higher martensite fraction (Figures 3.13-a
and 3.13-b). Nevertheless, small amounts of a dark-colored third phase are present in the
microstructure for both annealing temperatures.
On the other hand, the annealing temperature has no significant influence on the microstructural characteristics of the grade DP I. Martensite fraction remains on the same level without
any morphological change while no third phase could be observed.
In the case of grade M, the higher annealing temperature has a major influence on the
martensite morphology by increasing the number of structured martensite grains, although
the total martensite fraction does not increase. A third phase could not be detected in the
microstructure.
A more representative comparison between all the grades investigated, regarding their annealing behavior, is given in Figure 3.14. In order to emphasize the main differences, which
are of great importance for the interpretation of the mechanical properties, two extreme cooling
rates of 5 K/s and 80 K/s were applied after annealing in the austenitic region (840 ) for 30 s.
Comparing the microstructures of Figure 3.14 column-wise, it becomes obvious that the
cooling rate has a very strong influence on all materials with respect to grain refinement of
ferrite. This impact is more pronounced for the grades R and M than for grades DP I and DP II.
A cooling rate of 5 K/s is sufficient for a preliminary grain refinement of the dual-phase grades,
as shown in Figures 3.14-c and 3.14-e, partly explained from their already homogeneous

35

(a) CR = 5 K/s

(b) CR = 10 K/s

(c) CR = 20 K/s

(d) CR = 40 K/s

(e) CR = 60 K/s

(f) CR = 80 K/s

Figure 3.12: Microstructures of the grade M annealed at 800 for 30 s and quenched with different
cooling rates, etched with LePera.

36

(a) R, Tan = 800

(b) R, Tan = 840

(c) DP I, Tan = 800

(d) DP I, Tan = 840

(e) M, Tan = 800

(f) M, Tan = 840

Figure 3.13: Microstructures of the grades R, DP I and M annealed at 800 (left column) and
840 (right column) for 30 s after quenching with 20 K/s, etched with LePera.

37

ferritic-martensitic starting microstructure. Irrespective of the grade, all microstructures in the


left column (referring to the lowest cooling rate) possess a small fraction of a third phase. For the
grade M this fraction is negligible (Figure 3.14-g). It should be underlined that the industrial
grade R preserves/retains this third phase (bainite and/or pearlite) even after quenching with
80 K/s (Figure 3.14-b), a fact that differentiates this grade from the laboratory ones.
Additionally, the morphology of martensite is strongly dependent on the cooling rate. Low
cooling rates enhance the formation of white, clear, unstructured martensite grains. LePeras
etchant stains white also grains of retained austenite. However, magnetic volumetric measurements showed that no retained austenite is present in the microstructures investigated. When
a dramatically different/higher cooling rate is applied (microstructures of the right column of
Figure 3.14), part of the martensite grains appear dark brown, which is generally true for the
coarser grains with a substructure in their interior.

38

(a) R, CR = 5 K/s

(b) R, CR = 80 K/s

(c) DP I, CR = 5 K/s

(d) DP I, CR = 80 K/s

(e) DP II, CR = 5 K/s

(f) DP II, CR = 80 K/s

(g) M, CR = 5 K/s
(h) M, CR = 80 K/s
Figure 3.14: Microstructures of the grades R, DP I, DP II and M after annealing at 840 for 30 s
and quenched with the lower and the higher cooling rates of 5 K/s (left column) and 80 K/s (right
column) respectively, etched with LePera.

39

Further investigations were carried out by means of electron microscopy (involving SEM,
FEG-SEM and TEM) in order to identify all the phases present in the fine and ultrafine
microstructures. As the resolution scale increases, the microstructure characteristics become
location-dependent and no representative conclusions can be drawn about the overall fractions
and/or about the mean grain sizes of the phases. However, higher magnifications enable a more
detailed qualitative analysis.
To make the distinction between the phases possible, the creation of a topographic instead of
a color contrast is required, so that the grain boundaries between the phases are distinguishable.
For this reason all specimens for scanning electron microscopy were etched with Nital, which
preferentially etches ferrite, bainite and cementite and outlines their grain boundaries while
leaving martensite intact/undissolved.
Micrographs of the industrial grade R taken with a Field Emission Gun-Scanning Electron
Microscope at a magnification of 4000 are shown in Figure 3.15. The microstructures were
produced by annealing at 800 for 30 s and quenching with 60 K/s and were taken at different
locations of the same specimen. SEM micrographs reveal the presence of a third bainitic phase
in the microstructure of the reference grade R (Figure 3.15-b), supporting the observations
made in the light microscope. This third phase is mainly detected at the ferrite-martensite
grain boundaries and its fraction does not exceed 2 %.

(a)

(b)

Figure 3.15: Micrographs of the industrial grade R annealed at 800 for 30 s and quenched with
60 K/s. The symbols F, M and B marked on the grains stand for ferrite, martensite and bainite,
respectively.

Quenching with cooling rates equal to or higher than 40 K/s (in the given example 60 K/s)
results in the formation of coarse martensite blocks, consisted of strong orientated laths even
within the same block (Figure 3.15-a). Since no grain boundaries can be identified between
the laths, these blocks are considered and evaluated as discrete martensite grains.
40

Analogous difficulties, but this time regarding the identification/observation of grain boundaries between ferrite grains, are presented in Figure 3.15-b. There exist regions that resemble
martensite, which are present in the rim of ferrite grains without any distinct grain boundaries.
Light microscopy does not even allow the identification of these subgrains, irrespective of the
etchant used. To simplify the quantitative part of the evaluation and, moreover, to be on the
safe side in the grain size measurements, these grains are also considered as individual ferrite
grains.
The phenomenon described above becomes more pronounced and hence more critical for
the evaluation of the ultrafine microstructures produced from the laboratory grades, especially
after annealing at high temperatures and quenching with high cooling rates. Figure 3.16
shows FEG-SEM micrographs of the DP I grade, annealed in the austenitic region (840 ) and
quenched with 60 K/s. Although the existence of martensite subgrains partly surrounding the
coarser ferrite grains is evident, no grain boundaries between them could be detected. The
transition from ferrite to martensite is very smooth and the martensite subgrains appear
free of substructures, making the phase identification more difficult. The only indicative characteristic of the martensite presence is a difference in contrast observed between the phases,
where martensite appears brighter due its higher concentration in carbon and in other alloying
elements (Cr, Mn, etc.). Ultramicrohardness measurements have confirmed these assumptions.

(a)

(b)

Figure 3.16: Micrographs of the laboratory ferritic-martensitic grade DP I annealed at 840 for 30 s
and quenched with 60 K/s, (magn. 2000 and 8000 for (a) and (b) respectively).

Martensite grains exhibiting an unusual morphology are also shown in Figure 3.16. The
substructure in the interior of these grains, indicated by the white arrows (Figure 3.16-a), is
definitely different from the lath martensite structure described in Figure 3.15. Such complex
martensitic morphologies were already observed in the light microscope in the form of brown
41

tinted substructures within the coarser white martensite grains (as stained with LePera). According to bibliographic references of related works [80, 81], the emergence of the martensite
substructure can be interpreted as a result of carbide precipitation during tempering - which
in the case investigated refers to autotempering during quenching, since no tempering stage is
applied after annealing.

(a)

(b)

(c)

Figure 3.17: TEM micrographs of grade DP I demonstrating: (a) ferrite and martensite grains,
(b) martensite lath structure and (c) cementite precipitates within a grain of tempered martensite.

To clarify the substructure formation within martensite grains and to confirm the presence
of carbide precipitates, transmission electron microscopy (TEM) investigations were performed
on selected specimens, covering a representative range of grades and annealing conditions.
TEM micrographs of the ferritic-martensitic grade DP I are shown in Figure 3.17. The
investigated samples were annealed at 840 (- phase field) for 30 s and quenched with 5 K/s
and 80 K/s, represented by the images (a) and (b, c) respectively. For a direct comparison,
microstructures of the grade DP I subjected to the same heat treatment were earlier discussed
in Figures 3.14-c and 3.14-d. The microstructure of the samples cooled down with 5 K/s
consists mainly of martensite and ferrite. Sometimes, bainite and/or cementite precipitates
could be found close to martensite grains. No tempered martensite could be detected, which is
in agreement with the light microscopy observations.
In the samples quenched with 80 K/s, martensite grains exhibiting a lath microstructure
could be observed (Figure 3.17-b). Additionally, significant amounts of martensite grains
with a substructure in the interior are found (Figure 3.17-c). The substructure is characterized
42

by the presence of cementite precipitates, aligned along more than one habit plane variants.
These grains are identified as tempered (or to be more specific as autotempered) martensite
based on the observations of Bramfitt et al. [82], according to which the existence of multiple
carbide habit-variants is indicative of autotempered martensite [82]. In the following, the term
autotempered martensite will be used to describe such phase constituents.
Figure 3.18 provides a more detailed image of autotempered martensite formation, taken
from a sample of the martensitic grade M intercritically annealed at 800 and quenched with
80 K/s. As can been observed, autotempered martensite (marked on the micrograph) is located
in the middle of the grain and exhibits a different morphology from martensite located near
the martensite-ferrite interface. Higher resolution analysis confirms the presence of cementite
precipitates, revealing their multi-directional arrangement as well.

Figure 3.18: High resolution TEM micrographs of grade M, highlighting the formation of carbide
precipitates in autotempered martensite.

Additional TEM micrographs of the grade M are presented in Figure 3.19, corresponding
to dramatically different cooling rates after intercritical annealing. The microstructure of the
samples cooled down with 5 K/s (Figure 3.19-a) consists of fine martensite grains homogeneously dispersed in the ferritic matrix, without any tempered martensite present. Despite
the low cooling rate applied, no bainite could be detected. What should be also noticed, is
the presence of relatively coarse cementite precipitates located near martensite grains or in the
ferrite grain boundaries.
The samples quenched with 80 K/s (Figures 3.19-b and 3.19-c) show an ultrafine ferriticmartensitic microstructure. The presence of autotempered martensite, existing in considerable
43

fractions in this grade, was previously discussed in detail. Furthermore, negligible amounts
of retained austenite were identified, trapped in the form of thin plates between the laths of
martensite grains. However, the amount of retained austenite found is too small to be taken
into account in the quantitative analysis of the microstructure.

(a)

(b)

(c)

Figure 3.19: TEM micrographs of grade M intercritically annealed (Tan = 800 ) and cooled with
5 K/s and 80 K/s, for (a) and (b, c) respectively. The micrographs show: (a) coarse cementite precipitates located around a martensite grain, (b) ultrafine ferrite and martensite grains and (c) thin films
of retained austenite between the martensite laths.

3.1.4

Quantitative analysis- Grain size measurements

The determination of the quantitative characteristics of the industrially as well as of the laboratory produced materials was necessary in order to study the influence of the pre-processing
route and of the final annealing parameters on the microstructure evolution, regarding volume
fractions and mean grain sizes of the existent phases. To carry out the quantitative evaluation
of such fine grained microstructures, ferrite was considered as the dominant phase while the sum
of all other phases (that is martensite, autotempered martensite, retained austenite, bainite
and cementite) was characterized as a single second phase. The line intercept measurements
were made on light microscopy images (etched with LePera) using a magnification of 1500 or
2000 on purpose. Since grain refinement of ferrite was a main goal of this work, only ferrite
was measured in detail assuming that the rest represents the fraction of the second phase. This

44

60

Martensite fraction (%)

55
50
45
40
35
30

25

DP I

DP II

20

30

60

90

Cooling rate (K/s)

(a) Tan = 800

60

Martensite fraction (%)

55
50
45
40
35

30

DP II
DP I

25
20

30

60

90

Cooling rate (K/s)

(b) Tan = 840


Figure 3.20: Martensite fraction as a function of cooling rate, for two annealing temperatures.

45

rough classification into ferrite and non-ferrite constituents diminishes the human error sources
stemming from the complexity of phase distinction due to the tint etching effects.
However, for the evaluation of mechanical properties and their correlation with the microstructural characteristics it is essential to estimate the martensite fraction of the steels.
Taking into account that the fractions of bainite and cementite in the materials investigated
are extremely low (esp. for the laboratory grades) and that retained austenite hardly exists,
and, furthermore, accepting an unavoidable methodical error due to the increased surface area
of the grain boundaries - which becomes significant in ultrafine-grained microstructures, the
second phase fraction provides an acceptable approximation of the martensite fraction (without distinguishing between martensite and autotempered martensite). This roughly means that
the martensite fraction shown in the following diagrams is slightly overestimated, particularly
for cooling rates lower than 10 K/s, and for this reason a cautious interpretation of the results
is demanded.
Given that all grades are produced from the same hot-rolled material and subsequently have
identical chemical compositions, the observed differences in martensite fraction result exclusively from the applied heat treatment schedules and their impact on the particular characteristics of each pre-processed grade. The influence of the cooling rate on the volume fraction of
martensite is shown in Figure 3.20. Irrespective of the annealing temperature, an increase in
cooling rate is generally accompanied by an increase in the martensite fraction for all grades.
The maximum martensite fraction achieved by intercritical annealing was 48 % after cooling
with 80 K/s (Figure 3.20-a, M curve). The martensitic grade M seems to react more sensitive
to the changes of cooling rate. Unlikely, the martensite fraction of the grades R, DP I, and
DP II remains relatively stable or slightly increases, ranging between 35 and 40 % for cooling
rates higher than 10 K/s.
Austenitic annealing produces dual-phase microstructures with higher martensite fractions
than intercritical annealing (Figure 3.20-b), as predicted from the thermodynamic equilibrium. At higher cooling rates two subgroups of curves are observed, showing a different
behavior regarding their martensite fraction. Grades M and R possess almost 10 % more
martensite than the grades DP I and DP II at the same cooling rates, reaching fractions in
the order of 55 %.
Figure 3.21 shows the dependence of the ferrite grain size on the applied cooling rate.
Increasing the cooling rate results in a decrease of the mean grain size of ferrite (reaching
1.5 m) and leads to the formation of ultrafine-grained microstructures. The measurements
reflect the microstructural observations and, moreover, reveal the refinement degree in
cases that the changes in grain size are beyond the discriminability of the human eye.
Analogous to Figure 3.20-a, grade M responds positively to the changes of cooling rate,
throughout the whole range. The other three grades maintain a mean ferrite grain size of

46

3.5

Ferrite mean grain size (m)

M
DP II

3.0

DP I
R

2.5

2.0

1.5

1.0
0

30

60

90

Cooling rate (K/s)

(a) Tan = 800


3.5

Ferrite mean grain size (m)

M
DP II

3.0

DP I
R

2.5

2.0

1.5

1.0
0

30

60

90

Cooling rate (K/s)

(b) Tan = 840


Figure 3.21: Influence of the cooling rate on the mean grain size of ferrite.

47

Ferrite mean grain size (

m)

3.5

M
DP II

3.0

DP I
R

2.5

2.0

1.5

1.0
20

25

30

35

40

45

50

Martensite fraction (%)

(a) Tan = 800

Ferrite mean grain size (m)

3.5

M
DP II

3.0

DP I
R

2.5

2.0

1.5

1.0
20

25

30

35

40

45

50

55

60

Martensite fraction (%)

(b) Tan = 840


Figure 3.22: Relationship between ferrite grain size and martensite fraction.

48

approximately 2 m. Quenching from the austenite field has a more pronounced impact on the
grain size of ferrite; all grades exhibit grain sizes below 2 m even for cooling rates lower than
40 K/s.
The interdependency/interaction between martensite fraction and ferrite grain size is demonstrated in Figure 3.22. A reciprocal relationship (approaching a linear form) between them is
noticed irrespective of the grade and the annealing temperature, with the exception of grades
DP I and DP II after intercritical annealing (Figure 3.22-a). The finest microstructures are
obtained in steels with increased martensite fractions.
The higher the cooling rate the shorter is the time available for the austenite-to-ferrite transformation, suppressing in that way the further grain growth of ferrite. It is also presumed that
the increase of martensite fraction in the steel is mainly due to the number of martensite (prior
austenite) grains and not due to their size/volume.

49

3.2
3.2.1

Mechanical properties
Tensile testing

The mechanical (tensile) properties of the investigated dual-phase steel are presented in
Table 3.1, valid for a thickness of 1.0 mm. These values, which stem from the industrially
cold-rolled material laboratory annealed with a Hot-Dip Galvanizing cycle (Tan = 815 , THDG =
475 ) and skin-pass rolled for roughness improvement, will be used as a reference basis.
Table 3.1: Mechanical properties of the HDG-treated industrially cold-rolled DP-steel (DP600). The
quantities R p0.2 and R m stand for the yield and the ultimate tensile strength, while Ag and A denote
the uniform and total elongation, respectively.

Mechanical properties
HDG - DP600

R p0.2 (MPa) R m (MPa) Ag (%)


380
660
15

A (%)
22

To study the mechanical properties of the materials subjected to modified processing (grades
DP I, DP II and M), tensile tests were conducted with specimens of the same thickness as
the industrially cold-rolled material (grade R) so that the results are comparable. Taking into
account the observations/results of annealing simulations (see 3.1.3), the laboratory production
of tensile samples included two annealing temperatures (800 and 840 ), one annealing time
(30 s) and six different cooling rates, that is 5, 10, 20, 40, 60 and 80 K/s. The production of
tensile samples by applying non-conventional heat treatment schedules was also attempted, but
due to the insufficient temperature control and the unavoidable deviations from the annealing
schedule, no ultrafine ferritic-martensitic microstructures could be (re-)produced. For this
reason, such samples were not used in the investigations. The reference material (grade R)
was also tensile tested, after being submitted to the same heat treatment with the laboratory
grades. It should be noticed that all samples were tensile tested in the as-annealed condition.
The range of the applied annealing conditions provides with a wide variety of microstructures,
regarding the characteristics directly controlled by the annealing temperature and the cooling
rate such as the fraction of martensite and the presence of additional phases (e.g. bainite).
Moreover, and according to the results of grain size measurements, the study of the cooling
rate impact indirectly incorporates into the analysis the critical parameter grain size.
The existence of a third phase (bainite) makes the interpretation of the results even more
difficult, because in addition to the martensite fraction and the grain size a third parameter is
introduced, which is different for each grade. The influence of each of these parameters on the
mechanical properties can not be isolated and studied separately. Hence, all results are given in
dependence on the applied cooling rate, reflecting in this way the combined effects of all three
parameters (martensite fraction, grain size, presence of a third phase).
50

460

380
340

p0.2

(MPa)

420

DP II

300

DP I
R

260

30

60

90

Cooling rate (K/s)

(a) Tan = 800

520
480

400
360

p0.2

(MPa)

440

320

DP II
DP I

280
240

30

60

90

Cooling rate (K/s)

(b) Tan = 840


Figure 3.23: Influence of the cooling rate on the yield strength (R p0.2 ) of the dual-phase steels.

51

Figure 3.23 shows the influence of the cooling rate on the yield strength of the investigated
grades for both intercritical and austenitic annealing temperatures. An increase of the cooling
rate and subsequently of the martensite fraction results in an increase of the yield strength.
A simple qualitative comparison shows that the yield strength curves follow the form of the
martensite fraction curves in Figure 3.20. In the present case, a second contribution to the
yield strength is added by the grain refinement that occurs at high cooling rates. Since the grain
size of ferrite and the martensite fraction are not independent from each other, the quantitative
contribution of the two effects to the yield strength seems to be rather complicated.
In Figure 3.23 two groups of curves can be observed. The grades DP I and DP II exhibit
a remarkably lower yield strength than the grade M. This difference becomes more pronounced
as the cooling rate increases and reaches approximately 80 MPa. The behavior of the reference
grade (R) is strongly influenced by the annealing temperature. Intercritical annealing keeps
the yield strength of R in the lower strength group (Figure 3.23-a), while austenitic annealing
increases its yield strength to the higher levels of grade M (Figure 3.23-b).
A similar behavior is observed for the ultimate tensile strength of the grades in Figure 3.24.
The tensile strength increases as the cooling rate and the martensite fraction increase. The
two types of curves according to the above classification are still observed, maintaining also
the same form. In comparison with the tensile strength of the HDG-treated dual-phase steel
(Table 3.1) a significant increase reaching up to 200 MPa is achieved. It is noteworthy that
grades DP I and DP II maintain their yield strength at the same level as the reference hot-dip
galvanized material (that is between 360 and 380 MPa) for a wide range of cooling rates while
achieving higher tensile strengths by an amount of 140-150 MPa.
In Figure 3.25 the uniform and total elongations of all grades are plotted as a function of
the cooling rate. If the increase in strength would exclusively originate from the increase in
martensite fraction, then a decrease in total elongation should be expected. In the case of
intercritical annealing (Figure 3.25-a) this effect is not pronounced for any of the investigated
grades, since no dramatic decrease either of the uniform or of the total elongation occurs as
the cooling rate increases. Grades DP I, DP II and R show rather constant elongation even at
high cooling rates, slightly ranging between 13-14 % and 18-19 % for Ag and A, respectively.
Though grade M keeps also its elongation at a constant level, the exhibited values of Ag and A
are lower by approximately 2 % compared to DP I, DP II and R.
On the contrary, austenitic annealing seems to have a rather beneficial effect on grade
M and a detrimental effect on grade R. For cooling rates above 20 K/s, the decrease in
elongation as the martensite fraction increases is more pronounced for the grade R (in
comparison with grade M), reaching a gap of 5 % in total elongation at a cooling rate of
80 K/s. Hence, quenching with 80 K/s from an annealing temperature of 840 results in a

52

860
820

740

(MPa)

780

700

M
DP II

660

DP I
R

620

30

60

90

Cooling rate (K/s)

(a) Tan = 800

900
860

780
740

(MPa)

820

700

DP II
DP I

660
620

30

60

90

Cooling rate (K/s)

(b) Tan = 840


Figure 3.24: Influence of the cooling rate on the tensile strength (R m ) of the dual-phase steels.

53

24
M

22

DP II
DP I

A , A (%)

20

18
16
14
12
10

30

60

90

Cooling rate (K/s)

(a) Tan = 800

21
19

A , A (%)

17
15
13
11
9

DP I

DP II

30

60

90

Cooling rate (K/s)

(b) Tan = 840


Figure 3.25: Influence of the cooling rate on the uniform and total elongations (denoted as Ag and
A respectively) of the dual-phase steels. The hollow symbols in the diagrams stand for the uniform
elongation.

54

dual-phase steel with a tensile strength of 860 MPa and a total elongation of 10 %, in case the
reference grade (R) is used. If, however, the same heat treatment is applied to the martensitic
grade M, then a dual-phase steel with the same tensile strength level (850 MPa) but with a 5 %
higher total elongation is produced. In the same example, it should not be overlooked that the
total elongation exhibited by the grade R is lower than the uniform elongation of the grade M.
The ferritic-martensitic grades DP I and DP II show even higher uniform and total elongations
up to 14 % and 19 %, respectively, for the entire range of the applied cooling rates (Figure 3.25b). Combining the results of Figures 3.20-b to 3.25-b referring to austenitic annealing, it is
unexpectedly observed that the increase in martensite fraction (from 30 to 45 % on average)
accompanied by refinement of the ferrite grains does not affect the formability behavior of the
steels. Consequently, the grades DP I and DP II can provide a wide assortment of dual-phase
steels exhibiting a yield strength between 270 and 370 MPa and an ultimate tensile strength
between 660 and 810 MPa, while maintaining the same high ductility.
Two quantities that are of great importance for the characterization of dual-phase steels are
the yield to tensile strength (R p0.2 /R m ) and the uniform to total elongation (Ag /A) ratios,
providing a critical relationship between the strength and the ductility properties respectively.
The optimum performance of the material regarding both strength and formability is achieved
at low yield to tensile strength and high uniform to total elongation ratios. Figure 3.26 shows
the impact of the annealing conditions on the yield to tensile strength and on the uniform
to total elongation ratios of the investigated dual-phase grades. Intercritically annealed grades
exhibit a yield to tensile strength ratio ranging between 0.43 and 0.50, increasing with increasing
cooling rate. All the grades show the same behavior, with the only exception being grade M
when quenched with 80 K/s. The uniform to total elongation ratio lies above 0.68 and does
not exceed 0.78 for all grades. This ratio is slightly higher by 0.05 for the pre-processed grades
than that of reference grade R (Figure 3.26-a).
On the other hand, austenitic annealing results in the formation of two groups of R p0.2 /R m
curves. The lower group formed by the dual-phase grades DP I and DP II exhibits yield
to tensile strength ratios between 0.40 and 0.50 while the upper group (M and R grades)
exhibits ratios between 0.50 and 0.58. In both cases, the higher the cooling rate the higher is
the R p0.2 /R m value. However, such a tendency cannot describe the behavior of the Ag /A ratio.
The uniform to total elongation ratio ranges between 0.68 and 0.80 and does not show any
systematic dependence on the cooling rate. The grade R is only slightly affected by the cooling
rate and, furthermore, reaches higher values than the laboratory grades. The unexpected drop
in Ag /A of the grade M that occurs between cooling rates of 40 and 60 K/s cannot be explained
satisfactorily on the basis of the microstructural analysis.

55

0.9

0.8

0.8

0.7

0.7

DP I

0.6

DP II

0.6

p0.2

/ A

0.9

0.5

0.5

0.4

0.4

0.3

0.3
0

30

60

90

Cooling rate (K/s)

(a) Tan = 800


0.9

0.9

0.8

0.8

0.7

0.7

/ A

0.6

DP I

DP II

0.6

p0.2

0.5

0.5

0.4

0.4

0.3

0.3
0

30

60

90

Cooling rate (K/s)

(b) Tan = 840


Figure 3.26: Dependence of yield to tensile strength (R p0.2 /R m ) and of uniform to total elongation
(Ag /A) ratios on the cooling rate. The black dashed arrows point out the corresponding y-axis.

56

Strain hardening

900

900

800

800

700

700

True stress (MPa)

True stress (MPa)

In Figure 3.27 some true stress-logarithmic strain (widely referred as true strain) curves of the
investigated materials are shown.

600

500

400

300

M
DP II

200

600

500

400
M

300

DP II
DP I

200

DP I

100

100

0.00

0.04

0.08

0.12

0.16

0.20

0.00

0.04

Logarithmic strain

1000

1000

900

900

800

800

700

700

600
500
400
M
DP II

200

0.20

600
500
400
M

300

DP II
DP I
R

100

0
0.00

0.16

200

DP I

100

0.12

(b) Tan = 840 , CR = 5 K/s

True stress (MPa)

True stress (MPa)

(a) Tan = 800 , CR = 5 K/s

300

0.08

Logarithmic strain

0
0.04

0.08

0.12

0.16

0.20

Logarithmic strain

0.00

0.04

0.08

0.12

0.16

0.20

Logarithmic strain

(c) Tan = 800 , CR = 80 K/s

(d) Tan = 840 , CR = 80 K/s

Figure 3.27: Strain hardening curves of the dual-phase steels from selected annealing conditions.
Diagrams (a) and (b) represent the lowest cooling rate of 5 K/s while (c) and (d) the highest cooling
rate of 80 K/s.

Again, the curves are, as expected from the results given earlier in this chapter, separated into
groups, following the classification observed in Figures 3.23 and 3.24. All grades exhibit
a continuous yielding behavior and no yield point elongation is observed even for the lowest
cooling rate applied. The sudden change of the curve slope which occurs at logarithmic strains
close to 0.015 is due to a slight increase of the strain rate during the tensile test, though not
affecting the mechanical properties of the material (according to EN).
57

For most of the steels, the region of the true stress () versus logarithmic strain () curve
between the onset of plastic deformation (corresponding to the yield strength in the engineering
stress vs. engineering strain curve) and the onset of necking (corresponding to the ultimate
tensile strength in the engineering stress vs. engineering strain curve) can be approximated by
the Hollomon equation [83]:
= K n .

(3.1)

The constant K is a strength coefficient depending on the thermomechanical history of the


steel, while the parameter n is defined as the strain hardening exponent, also known as the
n-value. By taking logarithms of both sides of the eq. 3.1, the Hollomon equation can be
linearized:
log = log K + n log .

(3.2)

If the experimental data satisfy the Hollomon equation and, additionally, the curve is plotted
on logarithmic coordinates, then a linear regression line with slope n can be determined.
The strain hardening behavior can be evaluated by determining the strain hardening exponent of the steel (n-value). Physically, n gives a measure of the ability of the steel to distribute
strain along the gage length of the tensile specimen. For low-carbon steels used to form complex shapes, the value of n is 0.22. The higher the n-value the more uniform is the strain
distribution and, therefore, the greater is the resistance of the steel to necking and the better
is its formability [84].
The strain hardening behavior of dual-phase steels and the determination of their n-value have
been the subject of numerous investigations. Since the experimental data have shown that nvalue is not constant over the entire range of uniform strain, the empirical strain hardening laws,
such as the Hollomon equation, are unable to describe in detail the work hardening behavior of
the steel. For this reason new methods of analysis have been developed, in which the uniform
elongation stage is divided into several segments and individual n-values are calculated within
specified limits of logarithmic strain. Within the scope of this work and using the experimental
data as digitally recorded by the tensile testing machine, the deformation region/curve was
divided in two or three intervals (depending on the achieved uniform elongation Ag ), yielding
segmental n-values at logarithmic strains between 2.0 and 4.0, 4.0 and 6.0 and, where possible,
between 6.0 and 10.0, denoted as n 24 , n 46 and n 610 , respectively. Although in most cases the
uniform elongation of the materials exceeds 12 %, sometimes reaching even 15 %, the n-values
at higher logarithmic intervals are not available since the next strain segment automatically
provided by the tensile testing machine is n 1018 .
58

0.30
M

0.28

n
n
n

where:

2-4

DP II
DP I

0.26

4-6

6-10

n - value

0.24

0.22

0.20

0.18

0.16

0.14

0.12
0

10

20

30

40

50

60

70

80

90

Cooling rate (K/s)

(a) Tan = 800


0.30
M

0.28

n
n
n

where:

2-4

DP II
DP I

0.26

4-6

6-10

n - value

0.24

0.22

0.20

0.18

0.16

0.14

0.12
0

10

20

30

40

50

60

70

80

90

Cooling rate (K/s)

(b) Tan = 840


Figure 3.28: Dependence of the segmental n-values of the investigated grades on the cooling rate.
Each deformation segment is represented by an individual curve.

59

Figure 3.28 shows the dependence of the n-values of all grades on the cooling rate for both
annealing temperatures applied. As can be observed, the n-value decreases with increasing
strain, that is n 24 > n 46 > n 610 , irrespective of the grade and of the annealing conditions.
Furthermore, intercritical annealing produces grades with higher n-values than austenitic annealing for the same cooling rates. Though not clearly distinguished in the strain hardening
curves of Figure 3.27, the grades DP I and DP II exhibit remarkably higher n-values than the
grades M and R throughout the whole range of cooling rates, reaching values in the order of
0.28 at low cooling rates. This is obviously the reason for their increased uniform elongation
as also of the higher total elongation at fracture (see Figure 3.25).
Cooling rate proves to have a significant influence on the strain hardening exponent. Increasing the cooling rate and thus the martensite fraction of the steel results in a decrease of
the corresponding n-values. The form of the n-value vs. cooling rate curve remains the same
irrespective not only of the logarithmic strain segment but also of the grade. The impact is
more pronounced at cooling rates below 40 K/s. For higher cooling rates the n-value is roughly
independent of the cooling rate. The data for n 610 of the grades M and R for Tan = 840
(Figure 3.28-b) were not available. For this reason the strain hardening behavior of these
materials is graphically represented by using only two n curves (n 24 and n 46 ). A more detailed analysis of the strain hardening behavior of the investigated dual-phase steels is given in
chapter 4.
Microstructure at high strains
The deformed microstructures of the tensile specimens, more pronounced in the neighborhood of
fracture (region where necking takes place), were thoroughly examined by means of light (LM)
and scanning electron microscopy (both SEM and FEG-SEM). Figure 3.29 shows the fracture
profile of the ferritic-martensitic grade DP II, annealed at 840 and quenched with 10 K/s.
In the region of fracture the number of the voids formed is high, decreasing as the distance
from the fracture location increases. Furthermore, a higher density of voids is present in the
middle of the specimens thickness. Generally, the voids are formed at the ferrite-martensite
interface when a certain strain is exceeded. In case that bands of martensite are present in the
specimen, formed commonly in the middle of the sheet where the former pearlitic segregation
bands were located, it is observed that the voids are preferably formed around and along these
martensite bands, either by decohesion between martensite grains in contact or by fracture of
the martensite islands. This phenomenon is observed more often in the tensile specimens of
the industrial grade R, which has not undergone any pre-processing.
The ferrite grains located close to fracture are elongated, with the long axis parallel to the
direction of deformation. The hard martensitic islands do not seem to take part in the deformation process. Moving away from the fracture surface, not only the ferrite grains appear
60

less deformed, but also the number of voids decreases (Figure 3.29-b). Nevertheless, in some
cases voids were found even at distances in the order of [mm] away from the fracture location,
suggesting that the deformation mechanism cannot be not strictly determined by the/an accumulated local strain at the fracture point. Additionally, no significant observation regarding
the size distribution of voids over the region/area of their appearance was made.

(a) DP II, location of fracture

(b) DP II, few mm away from fracture

Figure 3.29: Light microscope micrographs of the dual-phase grade DP II, annealed at 840 and
quenched with 10 K/s. The specimen is etched with LePera (magn. 500 and 1500 for (a) and (b)
respectively).

The deformation mechanism as indicated by the fracture behavior is, however, not independent of the individual microstructural characteristics of the investigated dual-phase grades.
Figure 3.30 shows the region of fracture of the grade M, annealed at 840 and quenched
with two different cooling rates, 5 and 80 K/s. As was already discussed in a former section
of this work, the microstructures produced by applying these extreme cooling rates differ not
only in their martensite fraction but also in their mean ferrite grain size. Figure 3.30-a shows
the fracture profile of the specimen quenched with 5 K/s, possessing a relatively low martensite
fraction. Similar to the fracture area shown in Figure 3.29, a high number of voids is formed,
distributed all over the fracture region and penetrating into the specimen along the tensile axis,
indicating that the fracture process is not simply one of void nucleation and growth. The ferrite
grains carry the plastic deformation alone, while the hard martensite grains remain undeformed.
On the contrary, the fracture profile of the specimen with a high martensite fraction, see
Figure 3.30-b, reveals a different deformation mode. The number of the formed voids is
significantly lower. These few voids are located very close to the fracture surface and, practically,
no voids can be found at distances more than 200 - 300 m away from the fracture. The grains
of the ferrite within the highly strained necked region are severely elongated. Unlikely to the
61

fracture of Figure 3.30-a, the harder martensite grains appear also severely deformed. This
observation suggests that not only the soft ferritic matrix but also the harder second martensitic
phase takes active part in the deformation process.

(a) M, CR = 5 K/s

(b) M, CR = 80 K/s

Figure 3.30: Fracture profile of grade M annealed at 840 , etched with LePera.

FEG-SEM micrographs corresponding to the fracture samples of grade M (Figure 3.30-a)


are shown in Figure 3.31. The higher magnification (3000) and the absence of tint-etching
effects enable the observation of deformation bands within the ferrite grains and in some cases
the identification of the strain transfer paths along the ferritic matrix. The hard martensitic
islands remain undeformed. The sites where crack initiation and void growth occur are usually
found either at the ferrite-martensite interface or between martensite grains.
Additionally, fracture profiles of a high martensite fraction sample (analogous to the deformed
microstructure of Figure 3.30-b) are shown in Figure 3.32. The grains of martensite, most of
which emphasize a substructure due to autotempering or to their lower carbon content, appear
heavily deformed and elongated in the tensile direction. No cracks or deformation bands are
observed in the ferrite grains. A small number of voids is formed at the ferrite-martensite
interface, located very close to the fracture surface.
A quantitative analysis or even a rough estimation with respect to the formation of voids, their
size and area distribution and their preferential nucleation mechanism (either by decohesion of
the ferrite-martensite interface or by fracture of martensite islands) was not possible. Since the
severely deformed regions and the small size of voids do not allow a reliable quantification of
the microscopic observations, any analysis could lead to misinterpretations.

62

(a) M, CR = 5 K/s

(b) M, CR = 5 K/s, fracture

Figure 3.31: FEG-SEM micrographs of the fracture profile of grade M annealed at 840 , etched
with Nital.

(a) M, CR = 80 K/s

(b) M, CR = 80 K/s, fracture

Figure 3.32: FEG-SEM micrographs of the fracture profile of grade M annealed at 840 , etched
with Nital. The tensile axis is indicated.

The correlation of the mechanical properties of the specimens of grade M (examined above)
with their fracture behavior appears to be of great interest. Although the higher martensite fraction of the steel quenched with 80 K/s results in a remarkably higher ultimate tensile
strength by 170-180 MPa, the uniform and the total elongations of both steels lie on the same
level (10-11 % and 14-15 % for Ag and A, respectively). Moreover, according to Figure 3.28,
the steel with the lower martensite fraction exhibits consistently higher n-values by approx.
0.05 at all deformation stages (n 24 , n 46 ). Hence, neither the strain hardening behavior as
expressed by the n-values nor the difference in martensite fraction (which exceeds 25 % for the

63

applied annealing and cooling conditions) can provide a reasonable explanation for the similar
elongation properties of the two steels.
Analogous observations regarding the relationship between the mechanical properties and
the deformation mechanisms were made for the other pre-processed grades (DP I and DP II),
irrespective of the annealing temperature. Again, the unexpected ductility of the dual-phase
steels with high martensite fractions was accompanied by deformation of the martensite grains
in the location of fracture (necking area). This phenomenon motivates further investigations.
Especially the fact, that the second (martensitic) phase undergoes severe deformation during
tensile testing raises critical questions about the hardness of martensite grains.

64

3.2.2

Hardness measurements

Ultramicrohardness (UMH) measurements


In order to find out whether or not there is a relation between the hardness of the phase
constituents and the deformation behavior of the investigated dual-phase steels, the Vickers
ultramicrohardness of ferrite and of martensite was determined with an ultramicrohardness
testing machine (UMHT-3), mounted in a scanning electron microscope. After many trial tests,
the optimum indentation load was found to be 15 mN, which was applied to all measurements
irrespective of the phase and the sample. The higher the indentation load the larger the
indentation trace and consequently the more reliable the measurement. In this work, however,
the ultrafine ferrite and martensite grains of the pre-processed steels were a limiting factor in
the load selection, since the indentation trace/impression should definitely be smaller than the
size of the grain. The positioning precision of the testing equipment was approximately 1 m.
Selected examples of the Vickers indentations are presented in Figures 3.33-3.35. Although
not clearly demonstrated due to the various sizes of the figures, the magnification of all SEM
micrographs is identical (12000).

(a) DP II, CR = 20 K/s

(b) DP II, CR = 10 K/s

Figure 3.33: Indentations in coarse ferrite grains: (a) precise indentation, relatively unaffected from
the neighboring grains, (b) the presence of grain boundaries influences the measurement.

Figure 3.33 shows indentations in ferrite grains. By comparing the two micrographs it can
be noticed that the two diagonals of the pyramid trace are not always equal; a difference in
length up to 5 % between them was often unavoidable. In cases where the presence of grain
boundaries/interphases and/or the existence of a different than the indented phase beneath
the targeted grain strongly influenced the measurement, this difference in length could reach
10 % (Figure 3.33-b). Measurements with higher length deviations were taken as invalid
65

and were discarded.


To enable the martensite indentation, coarse grains with a minimum size of 2.5 m were
selected. The micrographs of Figure 3.34 show indentation traces in clear (unstructured)
martensite grains, taken from the samples quenched with 5 K/s. Even though both measurements are from a first point of view acceptable, the microhardness values calculated from the
mean length of the diagonals differ roughly by a factor of three (904 HV and 297 HV for (a)
and (b), respectively). A more careful examination of the Figure 3.34-b shows the existence
of a microcrack (marked with an ellipse) on the left side of the trace. Although the size and
the form of the crack cannot sufficiently explain the calculated hardness, which approaches the
ferrite hardness, this may be an indication of the existence of a softer ferrite substrate right
beneath the indented martensite grain.

(a) DP II, CR = 5 K/s

(b) DP II, CR = 5 K/s

Figure 3.34: Indentations in unstructured (clear) martensite grains.

Increasing the cooling rate from the annealing temperature results in the formation of coarse
martensite blocks showing a substructure in their interior. These grains were identified as
autotempered martensite. Provided that in the samples quenched with cooling rates higher than
40 K/s the fraction of tempered martensite becomes significant, the majority of the martensite
indentations in these specimens were conducted in tempered martensite. Impressions of the
Vickers indenter in tempered martensite grains are shown in Figure 3.35. Tempered martensite
appears to be apparently softer than untempered martensite and harder than ferrite.
Figure 3.36 shows the influence of the cooling rate on the microhardness of ferrite and
martensite of the grade DP II, after quenching from the austenitic annealing temperature
(840 ). The cooling rate was preferred over the martensite fraction as the independent variable, because it additionally includes the grain size effects and the martensite morphology. To
determine the microhardness of martensite, a minimum number of twenty measurements was
66

(a) DP II, CR = 60 K/s

(b) DP II, CR = 60 K/s

Figure 3.35: Indentations in tempered martensite.

necessary. Particularly in the samples where hard martensite islands were present (see Figure 3.34-a), a distribution of the hardness values within a range of 100 HV was observed.
On the other hand, the microhardness of ferrite remained relatively stable and was determined
as the mean value of ten measurements. In order to minimize the measurement errors, extreme
hardness values that do not make sense (either by being far away from the mean value or by
not corresponding to the indented microstructure) were excluded from the evaluation.
The ultramicrohardness of ferrite remains nearly independent of the cooling rate and, consequently, of the martensite fraction of the specimen. In absolute values it ranges between
230 HV and 250 HV, apparently higher than the ferrite hardness reported in literature. In order to understand this increase, it should be taken into account that the fine grains of ferrite
do not enable indentations unaffected by the surrounding (incl. also the underlying) F/F and
F/M grain boundaries. Additionally, portion of this increase could be interpreted as the result
of the work-hardened ferritic matrix due to the martensitic transformation, see chapter 4.
On the other hand, the ultramicrohardness of martensite grains appears strongly dependent
on the quenching conditions. Increasing the cooling rate results in a well-defined decrease of the
measured hardness, even for the smallest rate steps (e.g. from 5 to 10 K/s). As indicated by the
blue dashed line (fitting curve of the martensite hardness data) in Figure 3.36, the decrease
in hardness is stronger in the initial part of the curve, representing cooling rates lower than
40 K/s. This drop reflects the pronounced decrease in the martensite carbon content which
accompanies the significant increase in its fraction occurring in this cooling rate interval. At
higher cooling rates not only the martensite fraction increases further but also its morphology
changes, since autotempered martensite is formed (see Figure 3.35). The martensite hardness
keeps on decreasing but with a reduced slope. Quenching with 80 K/s (leading to 50 %
martensite) results in the minimum hardness of 390 HV.
67

800

HV
HV
k

0.8

Martensite

700

0.7

500

0.5

400

0.4

300

0.3

mart

0.6

ferr

HV

600

k HV

Vickers hardness (HV)

Ferrite

200

0.2
0

30

60

90

Cooling rate (K/s)

Figure 3.36: Vickers ultramicrohardness of ferrite and martensite grains of grade DP II, annealed at
840 . The triangle-curve (green) represents the ratio k of ferrite to martensite hardness at a given
cooling rate while the dashed lines represent the fitting curves of the experimental data. Indentation
load= 15 mN.

By extrapolating the fitting curve it turns out that a further increase in cooling rate does
only slightly affect the martensite hardness. If this behavior were solely controlled by the
martensite carbon content, then this assumption would contradict the expected results (that is
a further decrease of the martensite hardness). For this reason, fully martensitic samples were
produced from the same grade by water-quenching from the austenitic annealing temperature.
The hardness of martensite ranged between 390 and 400 HV, confirming the prediction of the
extrapolation.
Due to the limitations of the method set by the ultrafine-grained microstructures, the obtained
ultramicrohardness values cannot be used as absolute hardness of the individual phases. For
this reason a new variable k was defined, expressing the ratio of ferrite to martensite hardness:

k = HVF /HVM .

(3.3)

The trend of hardness ratio with increasing cooling rate is also plotted in the diagram of
Figure 3.36. The curve of k is actually generated by using the data of the fitting curves of
68

ferrite and martensite hardnesses. The decrease in the hardness of martensite with increasing
cooling rate has a dramatical effect on the hardness ratio, increasing its value from 0.33 at
5 K/s to 0.60 at 80 K/s. The relative hardness of the phases of the steel given by the hardness
ratio can provide a useful tool for the description of their deformation behavior.
Summarizing, the ultramicrohardness of martensite is between 730 HV and 390 HV in the
range of the cooling rates applied. The measured values correspond with the microstructural
observations of the fracture behavior, supporting the hypothesis that the deformation of martensite in the samples quenched with high cooling rates can be attributed to its substantially lower
hardness.
However, a physically-based interpretation of the drop in martensite hardness in correlation
with the deformation mechanism should take into account the martensite fraction that each
cooling rate represents, the carbon content of this specific martensite and the presence or
absence of tempered martensite as a phase constituent.

Microhardness (MH) measurements


Additionally to the ultramicrohardness measurements, the Vickers microhardness of the investigated dual-phase steels was determined. The indentation load was 150 p, which converted
to SI units is equal to 1471 mN (so actually HV0.15 was measured). Compared to the results
of the ultramicrohardness investigations, the approximately one hundred times higher indentation load produces impressions which are bigger by one order of magnitude. Micrographs of
the impressions in different heat treated samples of the DP II grade are shown in Figure 3.37.
The length of the impression diagonals varied from 30 to 35 m depending on the heat treatment and, hence, on the microstructural characteristics of the sample (martensite fraction,
grain size). The indented area/volume includes a few tens of plastically deformed ferrite and
martensite grains and thereby a significant amount of grain boundaries.

69

(a) DP II, CR = 10 K/s

(b) DP II, CR = 80 K/s

Figure 3.37: Vickers impressions in grade DP II, annealed at 840 and quenched with different
cooling rates (1000, Nital). The average diagonal lengths listed correspond to hardnesses of 240 HV
and 275 HV for the micrographs (a) and (b), respectively. Indentation load= 150 p.

280

Vickers Hardness (HV)

270
260

HV

250

DP

= 164.4+2.21 *

240
230

HV

DPII

220

Linear fit

210
200

25

30

35

40

45

50

Martensite fraction (%)

Figure 3.38: Dependence of the Vickers microhardness on the martensite fraction of the grade DP II,
annealed at 840 . The dashed line represents a linear fit to the experimental data. Indentation
load= 150 p.

The results of the microhardness measurements are presented in Figure 3.38. The data are
plotted as a function of the martensite fraction instead of the applied cooling rate. The micro70

hardness of the dual-phase steel increases nearly linearly with increasing martensite fraction.
According to this linear relationship, the microhardness of a completely martensitic material
(produced from the same alloy) should be in the order of 385 HV. The measurements performed using a fully martensitic material gave a mean microhardness value of 400 HV, very
close to the calculated value. The small deviation is within the acceptable error limits of the
measurement.

3.2.3

Hole expansion

To extend the knowledge about the formability behavior of the dual-phase steels investigated
in this work, the hole expansion property (also known as hole flangeability) of selected specimens was determined using the testing equipment described in section 2.8.3. Due to limited
availability of the pre-processed laboratory material, only the samples of the industrial grade
R were investigated in the entire range of cooling rates and annealing temperatures. A small
number of specimens of the grade M annealed at selected extreme conditions was also tested.
Images of a sample after hole expansion are shown in Figure 3.39. The specimen was
produced from the grade R, annealed at 800 and quenched with 60 K/s. According to the
method, the punch stops when the first crack formed in the lip of the hole propagates through
the whole thickness of the material. Since the judgement of the complete crack is made
visually, a further - though slight - hole expansion may take place during the time that intervenes
between the crack observation and the stop of the punch. It is therefore possible that a second
crack also occurs, as pointed out in Figure 3.39.
It was observed that the crack occurrence is somehow associated to the rolling direction
(RD); in the majority of the examined samples the first tear-crack occurred parallel to the
rolling direction. In case a second crack was additionally formed, this was oriented either
parallel or diagonal to the rolling direction. No cracks in the transverse direction were observed.
To determine the average final diameter of the ruptured hole, the measured perpendicular
diameters were selected to be diagonally oriented to the rolling direction, as illustrated in
Figure 3.39-b. The hole expansion ratio was then calculated using the equation 2.2, where
df = (d1 + d2 )/2. Two specimens were tested for each annealing condition and the average value
of was calculated.
The results of the hole expansion experiments are presented in Figure 3.40. Independently
of the heat treatment parameters, the experimental data can be classified into two groups
according to the achieved -value. The hole expansion ratios of grade R are at least two times
higher than the ratios of grade M, for both annealing temperatures and for all cooling rates.
The -values of R lie between 27 and 35 %, exhibiting a general tendency to decrease as the
cooling rate increases from 5 to 80 K/s. This tendency is also present in the samples of grade
M, for which the -values do not exceed 17 %.
71

(a) overview of the specimen

(b) zoom in on the cracks

Figure 3.39: Specimen of the grade R after the hole expansion test. The formed cracks, the evaluated
perpendicular diameters and the rolling direction are also specified.

The analysis and the interpretation of the hole expansion results was not feasible on the
basis of mechanical (tensile) properties; both the strength and elongation characteristics of
the investigated grades, that is the absolute values as well as the yield and elongation ratios
(R p0.2 /R m and Ag /A), are lying on comparable levels, with the properties of M to be the most
favorable. In an attempt to explain the remarkable difference in -values, the microstructures
of selected data points of the diagram of Figure 3.40 are shown in Figure 3.41.
It is of great interest to compare the microstructures between the positions P1 P4, P2 P5
and P3 P6, corresponding to pairs of samples quenched with the same cooling rate (5, 20
and 80 K/s for each pair, respectively). The microstructures of the samples P1 and P4 are
quite similar regarding the grain sizes and the martensite fractions. The slightly higher fraction of a third (bainitic) phase in P1 and the more homogeneous microstructure of P4 could
be considered as differences. However, the hole expansion property of P1 reaches 34 % instead of 17 % of P4. The same microstructural approach can be made for the positions P3
and P6; for almost identical microstructures, the -values exhibit a two-to-one relationship
(R ' 2 M ), differing by approx. 16 % in absolute values. The last pair involves the samples
P2 and P5, both selected from the grade R. The only observed differences in microstructure are
the more pronounced appearance of martensite bands and the existence of a small amount of
tempered martensite in P2, resulting in a 5 % higher hole expansion ratio. It should be pointed
out that the images of P1 and P2 show the microstructures with the highest -values achieved.
The microstructural investigations conducted in the light microscope could not provide adequate information for the correlation of the hole expansion property with the microstructural
72

40
800 C
P2

840 C

30
P5

P3

25

800C

= (

d -d

)/

x 100 (%)

35

P1

20

P4

15
10

P6

30

60

90

Cooling rate (K/s)

Figure 3.40: Dependence of the hole expansion ratio on the cooling rate, for each annealing
temperature applied. The microstructures of the numbered points are presented in Figure 3.41.

(a) point P1

(b) point P2

(c) point P3

(d) point P4

(e) point P5

(f) point P6

Figure 3.41: Light microscope images corresponding to the hole expansion samples. The specimens
are etched with LePera.

73

characteristics of the samples. The detectable but slight microstructure variations are not sufficient to justify the remarkable difference in hole expansion performance. Further investigations
in FEG-SEM were focused on the study of fracture surface as indicated in Figure 3.42.

Figure 3.42: Overview of the fracture surface of both sides of the crack (magn. 100). The macroscopical changes in the specimens shape/morphology along the cross-sectional area of the hole in the
direction of the punch can be observed.

Macroscopically, the material exhibits ductile fracture. Since overload is the principal cause
of the fracture, the material fails by a process known as void coalescence. The voids nucleate
at regions of localized strain discontinuity, such as inclusions, grain boundaries, second phase
particles and dislocation pile-ups. As the plastic strain increases, the voids grow, coalesce and
finally form a continuous fracture surface. As a result of this fracture mode, numerous cup-like
depressions are formed, referred as dimples [85, 86]. Figure 3.43 shows the fracture surface
of a hole expansion sample of grade R. The sample was annealed at 800 and quenched with
40 K/s, exhibiting a hole expansion ratio of 29 %. The relatively small size of the dimples reflects
the homogeneous distribution of the voids and the big number of nucleation sites. Large dimples
(corresponding to large voids) were sporadically detected, often associated with the presence of
large non-metallic inclusions. Along with the dimples, the formation of mutually parallel other
microcracks ( Figure 3.43-a) was observed, resembling the preferable void nucleation in the
martensite bands. These microcracks propagate along the hole expansion crack in a direction
independent of the loading direction. It was also noticed that the dimples at locations away
from the hole lip appear equiaxed, as if they were formed under conditions of uniaxial plastic
strain in a direction perpendicular to the fracture surface.
As approaching the lip of the hole, the fracture characteristics become consistent with the
complex deformation mode applied during the hole expansion test ( Figure 3.43-b). The dim74

ples slightly change their shape by becoming shallower, an effect which may be attributed to
the joining of voids by shear along slip bands. On the edge of the crack where the impact of
shearing becomes stronger, the dimples appear elongated with the one end being open, that is
the dimples are not completely surrounded by a rim. Cleavage fracture (transgranular fracture
mode) of coarse martensite grains which undergo very little plastic deformation was sparsely
observed.

(a) microcrack propagation along the crack

(b) fracture surface on the lip of the hole

Figure 3.43: FEG-SEM micrographs of the fracture surface of the HE sample. The initials PD stand
for the punching direction.

(a) cleavage fracture among the dimples

(b) elongated dimples and shearing effects

Figure 3.44: Fracture surface of grade M, focused on the edge of the crack (on the lip of the hole).
The arrow is pointing in the punching direction.

The FEG-SEM images of Figure 3.44 present fracture surfaces of the crack, focused on
75

the vicinity of the hole lip. The investigated M grade sample was annealed at 840 and
quenched with 80 K/s, exhibiting an -value of 11.4 %. The formation of very small dimples
(approximately two times smaller compared to the dimples shown in Figure 3.43) confirms
the existence of a great number of nucleation sites, representing faithfully the ultrafine-grained
microstructure of the as-annealed material. Cleavage fracture of the martensite grains was
frequently observed in the edge location of the crack. The well-orientated features/substructures
revealed on the cleavage surface (cleavage steps or river patterns) are formed in the direction
of loading. Furthermore, effects such as elongated dimples and shear fracture are observed due
to the complex load state (non-uniaxial loading).

76

Chapter 4
Discussion
4.1

Grain refinement and microstructure investigations

The main concept of the applied grain refinement process involved severe cold-rolling of preprocessed (i.e. previously heat-treated) dual- or single-phase microstructures, followed by appropriate intercritical or austenitic annealing. The obtained ultrafine microstructures were
qualitatively and quantitatively described in section 3.1. A very wide variety of dual- and
multi-phase microstructures was realized, offering the possibility of a detailed discussion of the
influence of the pre-processing routes and of the annealing conditions on the microstructural
characteristics, such as the formed phases, their phase fractions and grain sizes.
An important step in grain refinement is the subdivision of existing coarse grains in the as
hot-rolled material into finer ones before the final annealing process. Pre-processing before
cold-rolling seems to promote this subdivision. By water-quenching after the initial annealing procedure, the austenite to martensite transformation, which is associated with a volume
expansion of the transforming grain, introduces a large number of dislocations into the surrounding ferritic matrix. This feature as well as the presence of the hard martensitic islands
enhance grain subdivision during severe plastic deformation [64, 87].
Annealing of the pre-processed cold-worked microstructures offers significant advantages compared to the conventional production method of dual-phase steels by intercritical annealing of
cold-rolled ferritic-pearlitic material. Firstly, the number of the existing sites where nucleation
of new grains can occur (grain boundaries and grain triple points) is considerably higher, increasing in that way the possibility to generate new grains. Secondly, the number of dislocations
introduced by deformation is high, providing the necessary driving force for rapid recovery and
recrystallization. By careful selection and matching of the annealing conditions with the individual characteristics of the pre-processed grades, the formation of an ultrafine microstructure
can be achieved.
The ferrite grain sizes shown in Figures 3.21-3.22 resemble mean values estimated from
77

the grain size measurements. The actual distribution of grain sizes follows the scheme of
Figure 4.1, irrespective of the grade and of the annealing conditions. In the given example the
response of all investigated grades (M, DP I, DP II and R) to intercritical annealing followed
by rapid quenching is illustrated. The results are plotted as relative frequency over the range
of ferrite grain sizes present in the microstructure. Each set of bars represents the relative
frequency within a size interval (class) of 0.75 m, i.e. between 0.0-0.75 m (class 1), 0.751.5 m (class 2), etc. It can be safely concluded that the finest microstructure is formed by the
martensitic grade M, where the amount/number of ferrite grains which are finer than 2.25 m
exceeds 80 %.

40

Relative frequency (%)

35
M

30

DP II
DP I

25

20
15
10
5
0

10

11

12

Size class of ferrite grains

Figure 4.1: Ferrite grain size distribution for all grades, annealed at 800 and quenched with 80 K/s.

Indicative of the size homogeneity of the ultrafine microstructure is the size range of the
formed grains, that is the actual spread of the observed values as defined by the difference
between the coarsest and the finest measured grains (here zero m is assumed to be the
smallest/finest value, though without physical meaning). The narrower the size range of the
formed grains the more homogeneous is the microstructure. The size range of grade M is lim78

ited/restricted to 6 m, with the percentage of grains coarser than 3 m being less than 10 %.
On the contrary, the grain size ranges of the laboratory grades DP II and DP I are slightly wider,
reaching 7.5 m and 8.25 m, respectively, and hence increasing not only the mean ferrite grain
size but also the size inhomogeneity of the material. This effect is even stronger in the reference
grade R, where grains as large as 9 m are observed.
What should not be overlooked is that an important side-effect of pre-processing is the complete dissolution of pearlite contained in the as hot-rolled steel. The absence of pearlite in the
cold-rolled material minimizes the required annealing time, since no extra time for dissolution of
cementite is necessary and allows the application of short holding times during final annealing.
Additionally, pre-processing turned out to be an effective way to minimize the presence of
martensite bands in the dual-phase microstructure, which are formed in the positions where the
former pearlitic bands were located in the as cold-rolled state. Pearlite banding is closely associated with the thermomechanical history of the steel, starting already from the solidification
process and taking its final form in the hot-rolled sheet. The necessary but not individually
sufficient conditions that provoke banding are the microsegregation of certain alloying elements
like Mn and Cr, which effectively attract carbon (compared to other elements that reject carbon,
e.g. Si)[88], the slow cooling rates that allow the further carbon enrichment of the manganeserich regions by diffusion and, finally, the fineness of austenite grains by providing an adequate
number of nucleation sites. The technological importance of banding is focused on its influence
on the mechanical properties of the sheet, especially in the case of dual-phase steels where the
pearlite bands are replaced/substituted after annealing by martensite bands. Though no significant influence of banding on the materials tensile properties has been reported up to now,
a remarkable reduction of the impact toughness due to increased anisotropy was found [8991].
According to several experimental and modelling studies [88, 9194] pearlite banding can
be prevented if one of the above conditions is not any longer fulfilled, with the cooling rate
to be the most effective way to achieve it. Since the aim of this work was to produce ultrafine microstructures, the option/solution of austenite coarsening would lead to undesired
coarse-grained martensite. By means of pre-processing, the steel is initially heat-treated in
the intercritical or in the austenitic field so that the pearlite bands are dissolved and transformed into austenite. Though the holding time in the pre-processing stage is relatively short,
a partial redistribution of the alloying elements (mainly of carbon due to its higher diffusion
coefficient) is achieved. This effect is much more pronounced in the case of grade M (Tan =
900 , austenitic field). Water-quenching (very high cooling rates) from the pre-processing
temperature minimizes the number of martensite bands and also prevents a possible formation
of new pearlite bands. As a result, the starting microstructures of grades DP I, DP II and M
(before and after cold-rolling) possess a significantly lower number of bands in comparison with
the ferritic-pearlitic grade R.

79

4.2

Mechanical properties

Yielding behavior
The influence of grain refinement together with the other microstructural characteristics on
the mechanical properties of the dual-phase steels was studied thoroughly because of its great
technological interest. Special attention was paid to the yielding behavior of the material and
its dependence on the grain size of the matrix phase (ferrite) as well as on the volume fraction
of the second phase (martensite). The influence of the cooling rate on the yield strength (R p0.2 )
was shown in Figure 3.23, where an increase of the yield strength with increasing cooling
rate was observed. In case that this increase is attributed through a simplistic approach only
to one of the both strengthening mechanisms (either grain refinement of ferrite or increase of
the martensite fraction), then two different paths of argumentation can be followed for the
interpretation of the results, as being expressed by Figures 4.2 and 4.3.

520

460

480

420

DP II

DP II

DP I

DP I

440

(MPa)

p0.2

380

400

360

340

p0.2

(MPa)

320
300
280

260
1.0

240
1.5

2.0

2.5

3.0

3.5

Ferrite mean grain size (m)

1.0

1.5

2.0

2.5

3.0

3.5

Ferrite mean grain size (m)

(a) Tan = 800

(b) Tan = 840

Figure 4.2: Yield strength R p0.2 as a single function of ferrite grain size.

As can be seen in Figure 4.2, the behavior of all investigated grades is similar, irrespective
of the annealing conditions: the yield strength increases by decreasing the mean grain size of
ferrite. This effect, firstly described by Hall and Petch, is well known and has been investigated
in-depth over the last decades. A linear dependence between the yield strength (actually
between the lower yield stress) and the reciprocal square root of grain size was established
from the experimental data, as stated by the Hall-Petch equation (eq. 1.5). The interpretation
of this relationship is based on the dislocation pile-up theory. When dislocations encounter
obstacles their motion is impeded, causing the stress required to continue the deformation
process to increase. Grain boundaries act as such obstacles. As dislocations pile-up at the
80

grain boundary the stress concentration in the vicinity increases, causing eventually the grain
boundary to yield, or new dislocation sources to become active in the neighboring grain.
A very rough comparison between the different grades with regard to their yield strength
sensitivity to the grain size could be possible by plotting the yield strengths over the reciprocal
1/2
grain size of ferrite (dF ). The slope of each curve represents a ky -parameter of the Hall-Petch
relationship, referring to the specific grade annealed at a certain temperature (Table 4.1). It
becomes obvious from the range of the obtained ky values that no consistent conclusions can
be drawn. Additionally, the calculated 0 friction stresses exhibit negative -without physical
meaning- values for most of the curves, indicating the weak points of this analysis.

Table 4.1: Estimation of the experimental constant ky for the applied annealing conditions. The ky
value of 81.33 is obviously overestimated.

Grade
ky [N mm
], Tan = 800
3/2
ky [N mm
], Tan = 840

M
19.68
24.31

3/2

DP II
20.03
16.80

DP I
81.33 (?)
26.77

R
28.05
22.45

520

460

480

420

p0.2

(MPa)

380
340
M

360
320

DP II

300

400

p0.2

(MPa)

440

M
DP II

DP I

280

DP I
R

260

20

25

30

35

40

45

240

50

Martensite fraction (%)

20

25

30

35

40

45

50

55

60

Martensite fraction (%)

(a) Tan = 800

(b) Tan = 840

Figure 4.3: Yield strength R p0.2 as a single function of martensite fraction.

On the other hand, Figure 4.3 shows the influence of the martensite fraction on the yield
strength of the dual-phase steels without taking into consideration the grain size effect. It is
generally observed that the yield strength of the steel increases almost linearly with increasing
martensite fraction. Based on the grain size measurements, it can be safely assumed that
the increase in the volume fraction of martensite is solely expressed by an increase of the
81

number of martensitic islands finely distributed in the ferritic matrix and is not accompanied
by grain growth of a certain number of martensite grains. The higher number of martensite
grains results in a larger area of the ferrite/martensite phase boundaries, increasing in that
way (proportionally) the possibility for dislocation pile-ups to occur at the ferrite/martensite
interface and, hence, the required stress concentration for these boundaries to yield [48, 95].
According to Fischmeister and Karlsson [48] the initiation of plastic deformation of a coarsegrained ductile matrix with finely dispersed hard inclusions is governed by the yield strength
of the soft matrix. This statement is valid as long as the elastic properties (Youngs moduli) of
the two constituents are equal, which is roughly the case in dual-phase steels. Therefore, any
variation of the yield strength with increasing cooling rate -i.e. martensite volume fractionshould be attributed to an alteration of the in-situ ferrite yield strength in the two-phase
mixture, assuming that ferrite is always the continuous matrix and martensite remains isolated.
To provide a reasonable explanation for the alteration of ferrite yield strength with varying martensite fraction, the thermomechanical history of the steel has to be taken into consideration. The austenite to martensite phase transformation taking place upon quenching
from the annealing temperature is accompanied by a remarkable change in the volume of
the transforming region. The specific volume of body centered tetragonal (b.c.t.) martensite
(0.1298 cm3 /g) is significantly higher than the specific volume of face centered cubic (f.c.c.)
austenite (0.1265 cm3 /g) [49, 96], so that theoretically a volume increase of approx. 2.6 % occurs. In order to accommodate the volume expansion and the shear deformation accompanying
the martensitic transformation, a large number of geometrically necessary dislocations is introduced/produced in the surrounding ferrite grains. The higher the martensite fraction (in terms
of larger number of grains) the higher the number of locations at which ferrite has to accommodate the volume and shape effects of the transformation and hence the more homogeneous
the distribution of the mobile dislocations in the microstructure [35].

82

1.8

3.62
3.60
3.58
3.56

1.7

3.08

of austenite ()

3.64

c/a of martensite

3.04
1.08

3.00

1.06
2.96

c/a

2.92

1.04

2.88

Axial ratio

a and c of martensite ()

1.02

2.84

2.80
0.0

1.00
0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

C (wt. %)

Figure 4.4: Lattice constants of tetragonal martensite


and austenite in quenched carbon steels (After Honda
and Nishiyama [97]).

However, it should always be kept


in mind that the intensity/impact
of the martensitic transformation regarding volume expansion, shear deformation and changes in crystal
structure is strongly dependent on the
carbon content of austenite prior to
the transformation. A body centered
tetragonal structure is favored in carbon steels. As reported by Honda and
Nishiyama [97], the lattice parameters
change as follows: the a axis slightly
shrinks and the c axis markedly grows
with an increase in carbon content
above 0.25 %. The degree of tetragonality, as measured by the c/a ratio,
increases linearly with carbon content
according to the the following empirical relationship [98]:

c/a = 1.005 + 0.045 (wt. % C). (4.1)


As a consequence of the eq. 4.1, the
volume of the unit cell increases also
roughly linearly with increasing carbon content.

In the present study, the wide variety of investigated microstructures was produced from
a low-alloyed dual-phase steel having a fixed chemical composition. The carbon content of
the alloy is Calloy = 0.1 wt. % (Table 2.1). If it is assumed that this carbon is completely
partitioned between the ferrite and the austenite in the intercritical region, ferrite is carbonfree after quenching and no carbides are present in the dual-phase microstructure, then the
final carbon content of the martensite (CM ) would be inversely proportional to the martensite
volume fraction (VM ):
(wt.% CM ) = (wt.% Calloy )/ VM .

(4.2)

For example, for the lower and the higher martensite fraction of grade DP II annealed at 840
83

(Figure 3.20-b), it can be calculated that CM, 0.28 = 0.357 wt. % C and CM, 0.49 = 0.204 wt. % C
(the numerical indices refer to martensite fractions). Substituting these values into the eq. 4.1
yields (c/a)M, 0.28 = 1.021, while (c/a)M, 0.49 = 1.0142. This difference in the axial ratios corresponds to a higher volume of the lower martensite fraction unit cell by V = 0.65-0.70 %.
As a result, the impact of the martensitic transformation on the surrounding ferritic matrix
should be definitely stronger for the high carbon content grains (i.e. low martensite fraction microstructures). However, in a macro-scale approach to the example, the total volume expansion
effect is controlled by the much higher martensite fraction.
Liedl et al. [99, 100] studied the impact of the austenite to martensite phase transformation on the stress-strain behavior of a dual-phase steel by constructing a micro-mechanical
model employing fully three-dimensional finite element calculations. The simulations revealed
that the ferritic matrix (skeleton) is work-hardened during quenching, due to the local internal
stresses caused by the volume expansion which accompanies the martensitic phase transformation (Figure 4.5). The extent of this work hardening was found to depend on the martensite
content. The calculations were supported by ultramicrohardness measurements in ferrite grains:
it was clearly determined that the hardness of ferrite is higher close to ferrite/martensite phase
boundaries than close to ferrite/ferrite grain boundaries.

Figure 4.5: Plastic equivalent strain in a center section through a unit cell, after the simulated
cooling process. The dark (red) areas adjacent to the martensite grain correspond to high plastic
equivalent strains [99].

Transmission electron microscopy was employed to provide evidence for the microstructural
effects of the martensitic transformation. Figure 4.6 shows high resolution TEM images of the
grades DP I and M in the as-annealed condition. Irrespective of the annealing conditions, it
84

can be observed that the dislocation density in ferrite (white grain) is remarkably higher in the
vicinity of martensite (dark grain) and decreases as moving away from the ferrite/martensite
grain boundary towards the interior of the grain. Analogous TEM observations were reported
by Sherman [101]. According to his measurements, the total average ferrite dislocation density
is a linearly increasing function of the martensite fraction, taking values in the order of 109 /cm2 .
Furthermore, the ferrite dislocation density in grade M (Figure 4.6-b) is much higher than in
grade DP I (Figure 4.6-a).

(a) DP I, Tan = 840 , CR = 5 K/s

(b) M, Tan = 800 , CR = 80 K/s

Figure 4.6: Bright field TEM images of the dual-phase grades DP I and M in the as-quenched
condition.

Each of the two approaches discussed above throws light on the influence of the microstructural characteristics on the yield strength of the dual-phase steel from a different point of
view. However, these arguments cannot hold independently from each other within the scope
of this study, since an interaction between martensite fraction and ferrite grain size is unavoidable (Figure 3.22). Several experimental and theoretical works on the dependence of the
yield strength (or of the flow stress) on the grain size in dual-phase steels have been published
[14, 95, 102, 103], most of them trying to incorporate the effect of martensite fraction. All these
studies showed that a linear relationship between the yield strength and the reciprocal square
root of grain size was satisfied and could be expressed by appropriate modifications of the
85

Hall-Petch relation. Nevertheless, no general agreement on the conclusions of yield strength


dependence on grain size has been reached, mainly because of the different approaches concerning the definition of the controlling grain size. Balliger and Gladman [103], for example,
defined the grain size as the average martensite (second phase) island diameter, Reuben and
Baker [102] used a weighted average of the phase specific grain sizes (by applying a law of mix1/2
1/2
tures, dc1/2 = (1 VM )dF + VM dM ), while Chang and Preban [104] have accepted/adopted
the average ferrite grain diameter as equivalent to the mean free path in ferrite. In each case,
the investigators have emphasized the importance of the grain size of one constituent phase
(ferrite or martensite).
It should always be kept in mind that the existence of any kind of interdependencies among
microstructural parameters/variables makes multi-variable quantitative correlations extremely
risky, so that the requirement for a single-variable data analysis becomes mandatory. To enable
a fundamental Hall-Petch-type approach to the net impact of grain refinement on the yield
strength, coarse-grained dual-phase specimens were produced from the industrial as hot-rolled
strip. The coarsening procedure involved annealing in the austenitic region (Tan = 1200 ) for
a relatively long holding time to ensure homogenization of the microstructure (tan = 10 min),
followed by very slow furnace cooling (CR = 1 K/min) down to the quenching temperature
(TQ = 720 ) and finally oil quenching to room temperature. The quenching temperature was
selected in the low zone of the intercritical region, but high enough to avoid any precipitation
of carbides. The obtained microstructures are shown in Figure 4.7, exhibiting a martensite
fraction of approx. 50-55 % and a mean ferrite grain size of dF = 12-13 m, six to seven times
coarser than the ultrafine-grained material of analogous martensite fraction (e.g. grade M).
The mechanical properties of the coarse-grained material were determined by tensile testing.

(a)

(b)

Figure 4.7: Coarse-grained ferritic-martensitic microstructures (LePera).

86

Figure 4.8 shows the strength dependence of dual-phase microstructures containing 50 %


martensite on the grain size of ferrite, at different levels of plastic strain. Since the plotted
data refer to similar martensite fractions ( 50-55 %), it was assumed that grain refinement is
the dominant strengthening mechanism. Accepting a priori a linear relationship between the
strength and the reciprocal square root of the mean ferrite grain size (Hall-Petch relation), the
experimental data were fitted by linear regressions. The fitting curves (trendlines) and their
mathematical expressions are also shown in the diagram. At a first qualitative glance, the
approach seems reasonable: the yield strength of the steel increases as the grain size of ferrite
decreases and the estimated 0 friction stresses take reasonable values.
900
u

800

(MPa)

700

-1/2

(MPa)

=1.0%

600

= 401 + 10.9*

p1.0

-1/2

(MPa)

=0.2%

500

400

= 254.2 + 9.88*

p0.2

-1/2
F

(MPa)

p0.2

= 531.3 + 13.4*

Extrapolation of theor. values to

300

smaller ferrite grain sizes according


to published exp. data after:
Hodgson

200

Hickson

100

p0.2

= 113.4 + 17.4*

-1/2
F

Han

Pure Ferrite (theor.)


Linear fit, theor.
Fine DP (exp.)

(MPa)

Coarse DP (exp.)
Linear fit, exp.

12

17
-1/2

22

27

32

-1/2

(mm

Figure 4.8: Grain size dependence of the strength of a dual-phase steel containing roughly 50 %
martensite, approached by a Hall-Petch relationship. Literature data referring to pure ferrite enable
a representative comparison.

To allow the evaluation of the experimental results, theoretical data for the pure ferrite yield
strength dependence on the grain size are also included in Figure 4.8. The values are estimated
from the empirical relationship of Gladman and Pickering [105, 106], taking into account the
alloy content (weight %) in Mn, Si and free nitrogen (Nf ) and the grain size of ferrite (dF in
mm):
1/2

y [MPa] = 53.9 + 32.3 %Mn + 83.2 %Si + 354 Nf + 17.4 dF


87

(4.3)

For the chemical composition of the investigated steel -and for zero free (soluble) nitrogeneq. 4.3 takes the form:
1/2

y [MPa] = 113.4 + 17.4 dF

(4.4)

Furthermore, the estimated ferrite yield strengths from the Gladman and Pickering equation
were extrapolated to smaller ferrite grain sizes. Experimental data of ultrafine ferritic microstructures (dF = 1.5-2m) produced from low-alloyed steels verify the validity of this extrapolation [57, 59, 69, 107, 108].
The most significant observation is that the grain size dependence of the dual-phase steel, as
expressed by the slope of the curve, is remarkably smaller compared to that of the ferritic steel,
taking values ky,DP = 9.88 Nmm3/2 and ky,F = 17.4 Nmm3/2 respectively. Analogous results
were produced by Liu [109] for ultrafine-grained mild steels, where the slope of the Hall-Petch
linear fit expressed in Nmm3/2 was found equal to 7.48. Priestner et al. [52] have reported a
Hall-Petch slope of 9.75 Nmm3/2 for ultrafine ferrite grain structures in Nb microalloyed low
carbon steels. According to Tanaka [14], who has also observed very low ky values, a possible
explanation could be that in as-quenched dual-phase steels the stress required to unpin the
dislocations is relatively low due to the existence of unlocked dislocation sources.
For a better established comparison, more data points concerning the grain size variation
at a roughly constant martensite fraction should be acquired, especially for lower martensite
fractions. This was not possible within the framework of this study due to the limited flexibility
in simultaneous control of grain size and martensite fraction. Of equal importance becomes
the consideration of martensite grain size; it is evident that the production of coarse ferrite
grains was inevitably accompanied by the formation of coarse martensite grains (Figure 4.7).
Consequently, the impact of the martensitic transformation on ferrite should be different in the
two cases (coarse and fine microstructure) despite the similar martensite fraction.
Although the analysis is focused on the yield strength, interesting observations arise from the
comparison of the Hall-Petch slope of the different strength levels (that is Rp0.2 , Rp1.0 and
Rm ). At higher plastic strains the grain size dependence of strength becomes more pronounced,
so that the slope of the lines gradually increases from 9.88 Nmm3/2 to 13.4 Nmm3/2 . This
effect has been observed by other authors, too [14, 48, 50, 110], and was attributed to the
rapid increase in the work hardening of the material taking place in the early stages of the
deformation. The inverse effect takes place in materials exhibiting discontinuous yielding, e.g.
ferritic-pearlitic steels [14].

88

Strain hardening
The key factor that differentiates dual-phase steels from the variety of the ferritic steels (e.g.
mild steels, IF-steels, conventional HSLA, etc.) with respect to the deformation characteristics,
is the absence of yield point elongation, i.e. the ferritic-martensitic steels exhibit continuous
yielding. This feature together with the very high initial strain hardening rates account for the
excellent formability properties of dual-phase steels.
It is widely accepted that the main reason for the continuous yielding behavior as well as
for the high work hardening rate of dual-phase steels is the existence of mobile dislocations,
introduced into the ferritic matrix as a result of the austenite to martensite transformation. The
movement of these mobile dislocations results in the elimination of yield point elongation and in
the observed low yield strength. The interaction of the dislocations with each other and with the
finely dispersed martensite grains results in the high strain hardening exponent. Hahns model
[111] for discontinuous yielding predicts that mobile dislocation densities of 102 -104 /cm2 lead
to discontinuous yielding, while mobile dislocation densities of 106 -108 /cm2 result in continuous
yielding behavior. Although dislocation densities of 106 -108 /cm2 are typical in hot-rolled or
annealed steels, the majority of these dislocations are immobile, i.e. pinned by interstitial
atoms [36, 111]. As previously referred and cited, the total average ferrite dislocation density
in dual-phase steels is martensite fraction dependent and ranges between 1-1.5109 /cm2 , hence
satisfying Hahns criterion.
However, an important prerequisite for continuous yielding is that the dislocations produced
in ferrite during the martensitic transformation remain mobile at room temperature. These
dislocations can be pinned by interstitial atoms if a sufficient time at a certain temperature is
available for the necessary diffusion. Since these dislocations are generated below the martensite
start temperature (Ms ), the heat treatment steps following the transformation can be critical,
e.g. in the case that an overaging stage is required. In the annealing cycles applied in this work
this effect has no practical importance, because the time that passes between the Ms and room
temperature is too short for any diffusion to happen (even for the higher observed Ms ).
According to Speich and Miller [42], the rapid increase in the work hardening of dual-phase
steels at low strains is the outcome of three interacting parameters. Firstly, the residual stresses
generated during the austenite to ferrite transformation upon cooling are relieved already by
small amounts of plastic deformation. Secondly, the dislocation density in ferrite is increased
by generation of both statistically stored and geometrically necessary dislocations [112]. The
statistically stored dislocations are those resulting from simple work hardening of ferrite while
the geometrically necessary dislocations are created by the need to maintain contact between
the two phases during plastic deformation (compatibility). Thirdly, because the plastic incompatibility of the two phases cannot be accommodated completely by plastic deformation,
stresses are created within the martensite phase, which are compensated by back stresses in
89

ferrite. These back stresses impede dislocation movement in ferrite.


An accurate description of the plastic behavior of dual-phase steels is of great technological interest, since it is essential to understand the forming characteristics of the material. As
described in section 3.2.1, the determination of the n-value is the most common way to characterize the strain hardening behavior of the steels. Several idealized mathematical descriptions
-known as strain hardening laws- have been developed, with the commonly used [113]:
= KH nH

(Hollomon, see eq. 3.1),

= 0 + KL nL
= KS (1 + )nS

(4.5)

(Ludwik),

(4.6)

(Swift),

(4.7)

where and stand for the true stress and the logarithmic (or true) strain while K, 1 and 0
are constants. The n-values are usually determined from a double logarithmic plot of true stress
vs. logarithmic strain by linear regression. Correlations between the n-values and mechanical
properties of dual-phase steels have been reported [35, 41, 114], however no general conclusions
have been reached. From the mathematical point of view all the above equations (eq. 4.5 to
4.7) are well-founded and, furthermore, almost every stress-strain curve can be described under
the appropriate assumptions by nearly every law. Nevertheless, the fact that the n-value can
be strongly strain-dependent can lead to great deviations and misinterpretations, even between
the results of the different approaches.
Ratke and Welch [115] have defined a differential n-value, which is independent of strain
hardening laws and physically expresses the instantaneous slope of the stress-strain curve in its
double logarithmic form:
n() =

dln
| ,T
=constant .
dln

(4.8)

An analysis according to this method is shown in Figures 4.9 and 4.10. The differential nvalues for grades DP I and M are calculated from the strain-stress raw data, without making
any prior assumptions about the validity of an arbitrary strain hardening law.
The two curves of Figure 4.9 refer to the same heat treatment schedules, involving austenitic
annealing at Tan = 840 and quenching with a high cooling rate (CR = 80 K/s). Though the
mean ferrite grain size of the two specimens at the specific conditions is similar (dF,DP I = 1.59 m
and dF,M = 1.57 m), their martensite fractions differ by approximately 8.5 % (VM,DP I = 45.6 %
90

1.0
(change in strain rate

R
R
A
A

during the tensile test)

p0.2

0.8

MPa

12.8

10.6

18.5

14.7

2
5
.1

.1

0
=

DP I

,M

P
a

,D
v

n
n
a

.1

.1

=
I

P
v

n
n

,M

,D

847.5

= 0.212

4-6

= 0.178

0.2

2-4

YS

MPa

809.1

7
.1
0

,M

,D
v

2-4

= d(ln

0.4

M
483.6

1
.2
0
=

0.6

n
n

) / d(ln )

DP I
395.7

6-10

= 0.152

= 0.166

4-6

= 0.138

0.0
0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

Logarithmic strain ( )

Figure 4.9: Differential n-value of the grades DP I and M, annealed at Tan = 840 and quenched
with CR = 80 K/s.

and VM,M = 54.1 %). Both curves follow the same shape: the n-value increases steeply at very
small strains reaching a maximum value, followed by a sharp drop until it reaches a plateau
and then a smooth decrease takes place at higher strains. The last portion of the curve is
dependent on the uniform elongation; at strains higher than the necking point, the decrease
of n-value becomes more rapid until it equals zero. The peak observed at logarithmic strains
around 0.015 (higher than the strain where yield strength is recorded) is due to a slight change
in the strain rate, as previously discussed (see 3.2.1). Unfortunately, at very low strains the
low density of the recorded stress-strain data does not allow reliable assessments.
The n-values as demonstrated in Figure 3.28 are fairly reproduced from the curves by estimating the mean n-values between the same logarithmic strain intervals. Practically, the
example shows that among two ultrafine-grained ferritic-martensitic microstructures, the one
with the lower martensite fraction exhibits a more pronounced strain hardening for the whole
range of logarithmic strains, which finally leads to higher values of uniform and total elongations. It is worth mentioning that in the strain interval of interest, that is 0.02 < < 0.10, both
91

curves are decreasing with almost equal rates. From one point of view, this is a clear indication
that the initial stages of strain hardening are very crucial for the formability characteristics of
the steel. However, they do not provide any indication of the sudden drop of n-value occurring
at strains beyond necking.

1.0
(change in strain rate

R
R
A
A

during the tensile test)

p0.2

11.5

20.7

15.4

5
.1
0

DP I

,M

,D
v

n
n
2-4

14.6

8
.1
0
=
P

728.7 MPa

= 0.251

4-6

= 0.216

0.2

YS

718.9

.1
0

=
I

,M

,D
v

n
n
2-4

MPa

.2

1
.2

M
371

.2
0
0

=
,M

,D
v

n
0.4

= d(ln

0.6

) / d(ln )

0.8

DP I
320.2

= 0.215

4-6

6-10

= 0.182

= 0.184

0.0
0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

Logarithmic strain ( )

Figure 4.10: Differential n-value of the grades DP I and M, annealed at Tan = 840 and quenched
with CR = 10 K/s.

Qualitatively, these results are in good agreement with the works of Davies [40] and Hayami
[34] -though here the results refer to much finer microstructures- where higher martensite
volume fractions are accompanied by a decrease in the n-value. On the contrary, Karlsson and
Sundstrom [47] have proposed, that the strain hardening rate increases with increasing volume
fraction of the second phase (here martensite) and/or by increasing the difference in hardness
between the two phases. While the findings of this study do not agree with the first half of
his proposal, the second part turns out to be of great interest, as will be discussed later in
this chapter. Obviously, since the investigations involve only one chemical composition, it is
92

not possible to succeed in fulfilling both Karlssons requirements: the formation of either a
high martensitic grade with soft martensite or a low martensitic grade with hard martensite is
feasible.
Additionally, Ashby [112] stated that the compatible deformation of a soft matrix containing
hard, less deformable particles requires the generation of plastic strain gradients (i.e. geometrically necessary dislocations) within the more deformable phase. These plastic strain gradients,
which contribute to work-hardening by impeding the motion of other moving dislocations, are
evidently related to the type and the characteristics of the second phase, e.g. the hardness of
martensite. Speich and Miller [42] verified this statement in dual-phase steels, by investigating
microstructures with different carbon contents of martensite; steels with a high carbon martensite (CM = 0.4 wt. %) exhibited higher work hardening rates than the steels with low carbon
martensite (CM = 0.2 wt. %). The phenomenon was attributed to the greater plastic incompatibility of the harder martensite grains with the ferritic matrix and to the higher residual stresses
associated with their lower transformation temperatures.
On the other hand, Figure 4.10 provides the differential n-values of the same grades as before
(DP I and M) after annealing at Tan = 840 and quenching with CR = 10 K/s. This complementary example involves dual-phase microstructures containing identical martensite fractions
(VM
= 35.7 %) which correspond to different grain sizes (dF,DP I = 2.09 m and dF,M = 2.54 m).
Here, the finer-grained grade exhibits higher n-values and, consequently, better elongation properties. Again, the strain hardening behavior at high logarithmic strains cannot be anticipated.
At a constant martensite volume fraction it can be assumed that the in-situ properties of
ferrite and martensite remain constant, so that the hardness ratio of the two phases should also
remain constant. An increase in grain size results in a decrease of the number of ferrite grains
adjacent to martensite and influenced by the martensitic transformation. Hence, the coarsegrained material would exhibit lower initial dislocation densities distributed inhomogeneously
compared with the fine-grained material, leading to lower strain hardening rates. Analogous
results where reported by Matlock [35].
Deformation behavior at high strains
A metallographic study of the tensile deformation and fracture of the investigated dual-phase
steels has been already presented in the section Microstructure at high strains. Light and
electron microscope images of the fracture profiles have revealed two distinct modes of deformation, based on the type (clear or autotempered) and the strength-hardness of martensite
present in the microstructure. These characteristics are, however, dependent on the heat treatment schedules and especially on the cooling rate after annealing; low cooling rates produce
hard martensite enriched in carbon while high cooling rates result in low carbon, thus softer,

93

martensite.
The deformation mechanism can be comprehensively described in the strain region between
the initial stages of plastic deformation and the point where necking occurs. According to
Rashid [84], deformation occurs first in the ferrite which is the continuous soft phase. When
reaching its maximum capacity for being strained, ferrite matrix transfers strain across the
ferrite-martensite interface, leading to the deformation of the martensite islands. The nature
(and the properties) of martensite determines the extend of its plastic deformation and defines
the occurrence of the two deformation modes.
As shown in Figures 3.29, 3.30-a and 3.31 for low martensite fraction grades, no pronounced
deformation of the hard martensite grains takes place, even in the necking region. In some cases,
a slight rotation of the martensite grains towards the tensile direction could be identified. On
the contrary (Figures 3.30-b and 3.32), the softer martensite of the high martensite fraction
grades undergoes extensive plastic deformation, occurring mainly in the neck of the specimen.
Irrespective of the deformation type, all grades exhibit macroscopically a dimple-like ductile
fracture. Generally, ductile fracture begins with the formation of voids (mostly after necking),
which nucleate at sites of localized strain discontinuity such as non-metallic inclusions, grain
boundaries, phase boundaries, second phase particles and dislocation pile-ups. With increasing
plastic strain these voids grow further, coalesce into cracks and finally form a continuous fracture
surface.
However, it was found that the void formation process is different in the grades examined and
is affected by the in-situ martensite properties. Lower fractions of harder martensite promote
the formation of a high number of voids, distributed not only close to the fracture but over the
whole necking area. In that case, the ferrite-martensite interfaces act as the preferable void
nucleation sites. If martensite bands are present, then voids are formed around and along these
bands, either by decohesion of martensite grains in contact or by fracture of martensite grains.
Higher fractions of softer martensite result in a limited number of voids, strictly located in the
fracture region. The few voids are formed at the ferrite-martensite interface while no cracking
of martensite was observed.
Since cracking or decohesion of the martensite grains is involved in the void nucleation and
void growth process, the post-uniform elongation (A-Ag ) is expected to depend both on the
amount of martensite and on the carbon content of martensite. A number of simultaneous
opposing effects caused by the variations in quantity (fraction) and in quality (composition,
properties) of martensite should be taken into account. This fact is critical for the evaluation
of results in the present work, since only one carbon content is used and therefore CM is
inversely proportional to the martensite fraction. Higher fractions of martensite decrease the
spacing between the voids, so that the plastic strain required to join the voids is lower and,
consequently, should result in a lower ductility. On the contrary, the softer martensite grains

94

yield at lower values of plastic strain (in locations where the martensite yield stress is exceeded)
and so playing an active role in the deformation process, do not easily decohere and do not
crack. In an equivalent manner, martensite grains with higher carbon contents crack or decohere
more easily, so that the ductility should also decrease at high values of CM , which in this study
represents low martensite fractions.
It becomes obvious that a correlation of the microstructural characteristics and the deformation mechanisms with the elongation properties is not a simple task. An optimization of the
desired microstructures should involve all the possible parameters, esp. the contradicting ones
(martensite fraction, carbon content of martensite, morphology of martensite).
Micro- and ultramicrohardness measurements
Ultramicrohardness measurements have revealed that the in-situ hardness of martensite does
not remain constant over the range of the applied cooling rates. The highest measured value
(HVM = 730 HV) corresponded to a cooling rate of 5 K/s while the lowest (HVM
= 390 HV) to
a cooling rate of 80 K/s. This variation is intimately connected with the carbon content and
the morphology of martensite. Additional measurements in fully martensitic material of the
same chemical composition showed that an increase of the cooling rate over 80 K/s does not
further influence the martensite hardness. The hardness of ferrite remained roughly constant at
240 HV, indicating that no severe change regarding the partition of alloying elements in ferrite
takes place during cooling.
The influence of grain refinement on the ferrite hardness was not possible to be incorporated
into the results. It would be expected that as the microstructure becomes finer, the effect of
the increased area of grain boundaries should become greater. However, due to the equipment
limitations, ultrafine ferrite grains (dF < 4 m) could not be indented. To enable the indentation, ferrite grains coarser than 4 m were selected. Analogous size-selective sampling was done
for the martensite grains; only grains coarser than 2.5 m were indented.
To express the microhardness of dual-phase steels as a function of the martensite fraction, a
couple of critical assumptions had to be made. It was primarily assumed that all the grains
of each phase exhibit the same hardness. Even though such a statement could be acceptable
for ferrite grains, its validity in the case of martensite -where a wide scattering of hardness
values is observed due to the changes in its morphology and its carbon content- is questionable.
Moreover, the effect of the grain refinement of ferrite was not taken into account.
There has been an attempt to cast the experimental ultramicrohardness data taken from
specimens of the grade DP II into a linear rule of mixtures:

HVDP = VF HVF + VM HVM ,


95

(4.9)

where HVDP , HVF and HVM stand for the microhardnesses of dual-phase steel, of ferrite and of
martensite, respectively, while VF and VM express the volume fractions of the phases. The only
independent variables of eq. 4.9 are the volume fraction of martensite VM and the hardness of
ferrite HVF . Once VM is defined, the following two variables are accordingly directly or indirectly
defined: the volume fraction of ferrite equals VF = (1- VM ) while the hardness of martensite HVM
depends on its carbon content set by VM . Though HVF remains nearly constant, for the needs
of the example it will be considered as variable. After the necessary substitutions, eq. 4.9 takes
the form:

HVDP = (1 VM ) HVF + VM HVM ,

(4.10)

HVDP = HVF + VM (HVM HVF ).

(4.11)

or equivalently,

The values of HVF and HVM were experimentally determined (Figure 3.36). The estimated
microhardnesses of the dual-phase grade DP II are appended to the microhardness diagram
of Figure 3.38, marked with black diamond symbols as demonstrated in Figure 4.11. Obviously, the dependence between the calculated microhardness and the martensite fraction is not
linear, since eq. 4.11 includes the product of two interdependent parameters (VM HVM ). The
estimated results strongly deviate from the measured microhardness values, even though this
deviation becomes smaller with increasing martensite fraction. This tendency indicates that
the assumption of the same hardness for all martensite grains at a certain volume fraction (i.e.
carbon content) is acceptable only at high martensite fractions, where the hardness scattering
is minimized.
A second approach to the estimation of the dual-phase steel microhardness by a rule of
mixtures was attempted, assuming that both HVM and HVF remain constant. In order to
perform a conservative analysis, the minimum hardness of martensite was selected (395 HV at
VM = 100 %) while the hardness of ferrite was set to a mean value of 240 HV. Then, a linear
relationship between the steels microhardness and the martensite volume fraction is obtained
from eq. 4.11, plotted in Figure 4.11 within a volume fraction range of 20 % to 100 % martensite. The calculated values are consistently overestimating the experimental ones, even for the
smallest hardnesses used. Again, the difference in hardness decreases with increasing martensite fraction. When extrapolated to zero martensite fraction the red fitting curve intersects the
y-axis at HVF = 240 HV, a value which is by 75 HV higher than the hardness of pure ferrite
given by the extrapolation of the experimental curve ( 165 HV). On the contrary, there is a
perfect match between the experimental values (both UMH- and MH-measurements) and the
96

420

HV

DP

= 240+1.55 *

Vickers Hardness (HV)

380
340
300
260

HV
HV

220

DPII

, exp.
, exp.

100% Mart.

180

HV

DP

140
100

HV

DP

20

= 164.4+2.21 *

= 120+2.75 *

30

40

mix. rule,
mix. rule.,

mix. rule,

50

HV
HV
HV f V
HV

Linear fit of

60

70

80

DPII

= 240 HV

= 120 HV

90

100

Martensite fraction (%)

Figure 4.11: Correlation of the experimental microhardness and ultramicrohardness values by means
of a rule of mixtures. The measurements refer to DP II grade annealed at 840 .

calculation of microhardness for the fully martensitic material.


It is quite remarkable that the as-measured microhardness of the dual-phase microstructure
is significantly lower than the microhardness that a rough/simple rule of mixtures predicts,
even though the lowest values obtained from the ultramicrohardness measurements were applied. It is worth mentioning, for example, that a grade containing 28 % martensite exhibited
a microhardness lower than the mean measured ferrite ultramicrohardness.
It has been reported by several authors [116119] that hardness of metals or ceramics in
nanoindentation or even in ultramicrohardness measurements (low-load Vickers tests, [120])
tends to be overestimated due to a strong size-dependence of the tests. According to this
dependence, known as indentation size effect (ISE), the hardness is observed to increase with
decreasing indentation size (or, equivalently, indent load or indent depth). The effect is already
apparent at indentations shallower than 10 m and it becomes even more pronounced in the
sub-micron depth regime. Indentation size effect could not be explained by using a conventional plasticity theory-based analysis, according to which the hardness value should remain
97

independent of the applied load. Based on Ashbys theory of geometrically necessary dislocations, Fleck and Fleck et al. [121123] have proposed and developed a strain plasticity theory
which includes an internal material length scale. The basic concept was that in the sub-micron
scale the material hardening (and yielding) is controlled by the total density of dislocations,
part of which generates and is proportional to the gradient of plastic strain. Plastic strain
gradients appear either because of the loading geometry or because the material itself is plastically inhomogeneous (e.g. containing non-deforming phases, like dual-phase steels). In order
to accommodate these strain gradients surrounding the indent the generation/production of a
certain density of geometrically necessary dislocations is required, which is inversely proportional to a local length scale (in microhardness tests the length scale is directly related to the
indentation size). Concerning the contribution of the dislocations in the strengthening effects,
the model presumes that the flow stress is proportional to the square root of the total density of dislocations (statistically stored plus geometrically necessary dislocations, denoted as S
and G , respectively) and that the density of statistically stored dislocations -which depends
on the uniform strain- remains constant. At large indentation sizes (G S ) the statistically
stored dislocations govern the flow behavior, therefore no size dependence is observed. At
smaller length scales, the contribution of geometrically necessary dislocations to hardening becomes more significant. Finally, after entering a critical size/length regime which depends on
the material, the geometrically necessary dislocations outnumber the statistically stored ones
(G S ) and take control of the hardening effect. The validity of this extended theory was
experimentally verified by Elmustafa and Stone [124, 125] using a combination of microhardness
and nanoindentation measurements.
The indentation sizes involved in the ultramicrohardness measurements of this study lie in
the length scale where indentation size effects take place. The applied load of 15 mN produces
indents with a size of 2-2.5 m in martensite and 3-3.5 m in ferrite grains, corresponding to
maximum indentation depths of 0.35 m and 0.5 m, respectively (note that the depth of a
Vickers micro-indentation is about 1/7 of the diagonal length), that is well inside the sub-micron
range. However, in the case of pure martensite the measured microhardness was unexpectedly
very stable for loads (and indent sizes) differing by two orders of magnitude, namely 15 mN for
UMH and 1500 mN for MH.
Following the concept of the previous paragraph while keeping in mind the deviations of
the rule of mixtures, it turns out that the measured ferrite hardness may be linked to the
indentation size effect or, generally speaking, to any size effect. As already referred, the plastic
strain gradients during the indentation process generate, for compatibility reasons, a number
of geometrically necessary dislocations. Recalling the discussion made previously on the flow
behavior of dual-phase steels and on the impact of martensitic transformation in the ferritic
matrix, it becomes clear that in the ferritic matrix there exists already before indenting a

98

huge number of geometrically necessary dislocations ready to flow. As the indentation load is
applied on ferrite, these dislocations accelerate the strain hardening of ferrite even further so
that eventually a kind of fictitious microhardness is measured. The additional/collateral effects
of the already work-hardened ferritic skeleton, as well as the existence of grain boundaries in the
vicinity of the indent, on the microhardness lead to significantly overestimated measurements.
To support this argumentation and to clarify the observed discrepancy in the results, indentations in coarse ferrite by applying a higher load had to be conducted. Since the production
of pure ferritic material with the same alloying composition was not feasible, specimens of the
coarse-grained dual-phase steel of Figure 4.7 containing 50 % martensite were used. Additionally, even coarser ferrite grains from the same dual-phase grade (though with much lower
martensite fraction) were indented to check whether there is an influence of the ferrite grain
size. Figure 4.12 presents the Vickers pyramid impressions in coarse ferrite grains. Four indentations in each micrograph are shown, the two small ones corresponding to a load of 20 p
(196.1 mN) and the largest two to a load of 50 p (490.3 mN), that is more than an order of
magnitude higher than the ultramicrohardness load. Even though the presence of an ISE effect
was never excluded as possibility, the extent of the size dependence was rather surprising. The
coarse grains of Figure 4.12-a exhibit a hardness of 160 HV while the coarser grains in Figure 4.12-b are even softer with a hardness of only 120 HV. These values can be trusted, since
the partitioning of alloying elements does not change. It should be noticed that the micrographs
below refer to different magnifications.

(a) HVF = 160 HV

(b) HVF = 120 HV

Figure 4.12: Microhardness measurements in coarse ferrite grains of the same grade. In this length
scale, the different loads have only negligible impact on the hardness values (magn. 500 and 200
respectively, Nital). Applied indentation load= 20 or/and 50 p.

As clearly shown, the presence and the density of ready-to-flow mobile dislocations play a
catalytic role in the rapid work hardening of ferrite during the indentation. Actually, the ISE
99

effect is coupled with the grain size effect due to the refinement of ferrite. If an indent into
a ferrite grain is assumed, then the possibility of primary work hardening -to differentiate
it from the work hardening caused by the geometrically necessary dislocations created by the
plastic strain gradients- due to the existing mobile dislocations in the volume influenced by the
indent increases as the size of the ferrite target grain decreases. It was previously mentioned
that Liedl et al. [100] have experimentally shown that the hardness of ferrite is higher close to
ferrite/martensite phase boundaries than close to ferrite/ferrite grain boundaries. However, the
measured hardness difference was not exceeding 40 HV. As the microstructure becomes finer,
the number/amount of grain and phase boundaries influencing the measurement increases and,
hence, a higher microhardness is measured. On the contrary, the absence of grain boundaries in
the vicinity of indent in coarse-grained microstructures leads to measurements that approach the
hardness of pure ferrite. Additionally, a comparison between the example of Figure 4.12-a and
the measurements in fine-grained dual-phase microstructures of the same martensite fractions
verifies the assumption that the finer the distribution (size and topology) of martensite grains
the more pronounced is the impact of martensitic transformation.
The critical question that arises from the above discussion concerns the ferrite hardness value
that somebody could or should use in the simplified rule of mixtures. Apparently, the in-situ
measured HVF,UMH of 240 HV has not produced satisfactory results. The first impression was
that the produced deviations were due to the very hard martensite grains. Considering that
a hardness measurement reflects a flow behavior of the material, then the properties of ferrite
-which is the matrix phase- should be more decisive. On the other hand, if a HVF,MH of 160 HV
is used by keeping the martensite hardness constant at its lowest value of 395 HV, then this
particular rule of mixtures reproduces the linear fit of the experimental results. The reasoning
should be much easier, if no softer ferrite were measured. But since ferrite of the same dualphase steel can take even lower values, then why not using these values in the rule of mixtures.
Following this idea, the green curve (joining the crossed-circles) of Figure 4.11 is plotted. In
this case, the predicted hardness values of the dual-phase grade are underestimated, though this
hardness difference could be partly interpreted for the ultrafine microstructures as additional
strengthening attributed to grain boundary effects (analogous to the observations of [126]). It
is representatively displayed that setting up a model to describe the microhardness of a dualphase steel can be quite difficult, even for the most simple approach of a linear rule of mixtures.
In order to avoid any misinterpretations, the response of dual-phase steels in microhardness and
ultramicrohardness measurements should be holistically examined within the framework of all
their special microstructural characteristics.

100

Hole expansion behavior


A step forward in the direction of understanding the deformation behavior of dual-phase steels
under more complex loading conditions was attempted by means of hole expansion measurements. The results were presented in section 3.2.3. Unfortunately, the limited availability of
pre-processed material did not allow the determination of the hole expansion property for all
the grades involved in the study.
Based on the results, a couple of general observations can be made irrespective of the grades
investigated. The hole expansion ratio was found to decrease with increasing cooling rate
(Figure 3.40). The effect is more pronounced for the specimens annealed in the austenitic
region, probably because their martensite fraction is higher for the same cooling rate. The
higher the martensite fraction of the dual-phase steel the higher is the number of micro-cracks
formed at the surface of the hole during punching (hole-surface damage) and the higher the
strain hardening of the edges of the hole due to the shear deformation. Furthermore, the
presence of a hard second phase in the vicinity of these micro-cracks causes a significant stress
concentration and accelerates the crack propagation during the hole expansion test [127]. No
conclusions could be drawn about the influence of grain refinement and/or of the martensite
morphology on hole expansion, mainly because of the interdependency of these microstructural
parameters and the lack of a complete series of data/experiments.
If the difference in hardness between the phases is somehow reduced, then the surface damage
of the hole edge becomes less pronounced. Of course, this effect does not refer to the present
investigations, where the decrease in the hardness ratio k occurs due to the reduced carbon content of the microstructures with a higher martensite fraction. Several authors have studied the
effect of tempering on the hole expansion properties [77, 78, 127, 128]; by softening martensite,
the ferrite/martensite hardness ratio was drastically reduced. The results were encouraging,
however, in most cases an increase of the hole expansion ratio was accompanied by the return
of discontinuous yielding. Therefore, the application of tempering techniques strongly depends
on the desired properties of the final material. Another way/method proposed for the improvement of hole expansion property was the substitution of one part of martensite with bainite or
retained austenite [75, 76, 127, 129], but such a task was beyond the scope of the present study.
The electron microscopy investigations on the fracture surfaces of the cracks formed during the
hole expansion test provided valuable information about the mechanisms of fracture. However,
since no quantitative fractography was performed, the differentiations in the observed fracture
modes between the different grades and the various annealing conditions can be used only for
a rough interpretation of the results. Since the hole expansion testing methods are currently
under development, and the available up-to-date literature sources are limited, the clarification
of the deformation and fracture mechanisms involved in the test demands further experimental
work.
101

Chapter 5
Summary
In the first section of this work alternative ways of thermomechanical processing were studied,
aiming at the grain refinement of a dual-phase ferritic-martensitic steel by using conventional
industrial facilities/equipment. The main concept involved severe laboratory cold-rolling of
pre-processed microstructures instead of the classical ferritic-pearlitic steel sheet followed by
appropriate annealing. Various classical dual-phase heat treatments (though without an overaging stage) were applied in a dilatometer, until the desired microstructural characteristics
were obtained for each laboratory grade. Non-conventional annealing cycles were additionally
performed, aiming to promote the nucleation of new grains and -at the same time- to impede
excessive grain growth.
Based on the results of dilatometric investigations, laboratory annealing simulations were
conducted in order to reproduce the ultrafine dual-phase microstructures on a bigger scale and
to determine their mechanical performance. Although the results of the novel non-conventional
heat treatments were satisfactory, the production of specimens at a bigger pilot scale was not
feasible mainly due to insufficient control of the rapid temperature changes between the annealing segments with the existent industrial equipment. Therefore, only conventional annealing
cycles were employed, with the objective of studying the influence of the annealing parameters
(annealing temperature and cooling rate) on the microstructure and eventually on the mechanical properties. A wide variety of dual- and multi-phase microstructures was realized, regarding
the phase fractions and the grain sizes. Evidently, the unique response of the pre-processed and
industrial grades to the same heat treatments produced different microstructures. According
to the quantitative microstructural investigations, dual-phase steels with a mean ferrite grain
size between 1.5 m and 3 m and martensite fractions between 25 % and 55 % were obtained.
The mechanical properties of the grades were determined by means of tensile and hole expansion tests. Ultimate tensile strengths in the order of 800-850 MPa were accompanied by
remarkably high uniform and total elongations, only slightly affected by the martensite fraction. Moreover, yield to tensile strength ratios between 0.4 and 0.5 and very high strain hard102

ening rates (n 24 -values > 0.25) were achievable. Of particular interest were the properties of
the grades DP I and DP II, providing a wide assortment of dual-phase steels exhibiting a yield
strength between 270 and 370 MPa and an ultimate tensile strength between 660 and 810 MPa,
while maintaining the same level of high ductility. Though grain refinement proved to have
beneficial effects on the tensile properties of the steels, no safe conclusions on its impact on
hole expansion behavior could be drawn.
The corresponding as-annealed and as-tested microstructures as well as the fracture surfaces
and the fracture profiles were characterized by light and scanning electron microscopy. Intensive studies on the deformation behavior of the steels, focused on the neighborhood of fracture,
helped to identify the different failure mechanisms between the grades. Void formation was
more pronounced in the specimens of low martensitic fraction, consisting of hard martensite
particles embedded in the soft ferritic matrix. The sites where void nucleation and crack initiation occur were usually found either at the ferrite-martensite interface or between martensite
grains. On the other hand, the extraordinary high ductility of the dual-phase steels with high
martensite fractions was associated with severe deformation of the martensite grains in the
location of fracture (necking area). The observations were supported by microhardness and
ultramicrohardness measurements, which revealed the changes in the hardness of martensite
grains (softening) depending on its carbon content and its morphology.
The yield and the strain hardening behavior of the dual-phase steels were extensively discussed
on the basis of the unique material characteristics. The main reason for the continuous yielding
behavior as well as for the high work hardening rate of dual-phase steels is the existence of a
large number of geometrically necessary dislocations, introduced into the ferritic matrix due
to the volume expansion and the shear deformation accompanying the austenite to martensite
transformation during cooling. The movement of these mobile dislocations during the initial
stages of deformation results in the elimination of yield point elongation and in the observed
low yield strength. The interaction of the dislocations with each other and with the finely
dispersed martensite grains results in the high strain hardening exponent.
A Hall-Petch approach to the grain refinement impact on the yield strength was attempted,
assuming a linear relationship between the yield strength and the reciprocal square root of the
mean ferrite grain size. The results of the linear regression were adequately interpreted while
the validity of the assumptions was cross-checked with other published works.

103

Bibliography
[1] Verein Deutscher Eisenhuettenleute, editor. STEEL, A Handbook for Materials Research
and Engineering, Volume 2: Applications, chapter C4, pages 78117. Springer Verlag,
Verlag Stahleisen mbH, 1993.
[2] ULSAB-AVC Consortium. Technical Transfer Dispatch #6, ULSAB-AVC Body Structure
Materials. Technical report, May 2001.
[3] Porsche Engineering Services Inc. Materials and Processes #9, ULSAB-AVCPES Engineering report. Technical report, ULSAB-AVC, October 2001.
[4] n.n. Overview report, Executive Summary. Technical report, ULSAB-AVC.
[5] S. Hayami and T. Furukawa. A family of high strength, cold-rolled steels. In Microalloying
75, Proceedings of the Conference, pages 7887, Vanitec, London, 1975.
[6] A. P. Coldren, G. Tither, A. Cornford, and J. R. Hiam. Development and mill trial of
as-rolled dual-phase steels. In Formable HSLA and Dual-Phase Steels, pages 205227,
Chicago, IL, USA, October 26 1977. AIME.
[7] T. Greday, H. Mathy, and P. Messien. About different ways to obtain multiphase steels.
In J.W. Morris R.A. Kot, editor, Structure and Properties of Dual-Phase Steels, pages
260280, New Orleans, LA, USA, February 1921 1979.
[8] T. Furukawa, H. Morikawa, H. Takechi, and K. Koyama. Process factors for highly ductile
dual-phase sheet steels. In J.W. Morris R.A. Kot, editor, Structure and Properties of
Dual-Phase Steels, pages 281303, New Orleans, LA, USA, February 1921 1979.
[9] J. M. Rigsbee, J. K. Abrahan, A. T. Davenport, J. E. Franklin, and J. W. Pickens.
Structure processing and structure-property relationships in commercially processed dualphase steels. In J.W. Morris R.A. Kot, editor, Structure and Properties of Dual-Phase
Steels, pages 304329, New Orleans, LA, USA, February 1921 1979.

104

[10] J. J. Jonas, R. A. do Nascimento, I. Weiss, and A. B. Rothwell. Effect of deformation


on the transformation in two high silicon dual-phase steels. In Fundamentals of
Dual-Phase Steels, pages 95112, Chicago, IL, USA, February 2324 1981. AIME.
[11] T. Kato, K. Hashiguchi, I. Takahashi, T. Irie, and N. Ohashi. Development of as-hot-rolled
dual-phase steel sheet. In Fundamentals of Dual-Phase Steels, pages 199220, Chicago,
IL, USA, February 2324 1981. AIME.
[12] T. Furukawa and M. Tanino. Structure formation and mechanical properties of
intercritically-annealed or as-hot-rolled dual-phase steels. In Fundamentals of Dual-Phase
Steels, pages 221248, Chicago, IL, USA, February 2324 1981. AIME.
[13] K. Nakaoka, K. Araki, and K. Kurihara. Strength, ductility and aging properties of
continuously-annealed dual-phase high-strength steels. In Formable HSLA and DualPhase Steels, pages 126141, Chicago, IL, USA, October 26 1977. AIME.
[14] T. Tanaka, M. Nishida, K. Hashiguchi, and T. Kato. Formation and properties of ferrite
plus martensite dual-phase structures. In J.W. Morris R.A. Kot, editor, Structure and
Properties of Dual-Phase Steels, pages 221241, New Orleans, LA, USA, February 1921
1979.
[15] T. Matsuoka and K. Yamamori. Metallurgical aspects in cold-rolled high-strength steel
sheets. Metallurgical Transactions, 6A:16131622, 1975.
[16] P. R. Mould and C. C. Skena. Structure and properties of cold-rolled ferrite-martensite
(dual-phase) steel sheets. In Formable HSLA and Dual-Phase Steels, pages 181203,
Chicago, IL, USA, October 26 1977. AIME.
[17] Hsun Hu. Effect of silicon on annealing texture, plastic anisotropy, and mechanical properties of low-carbon phosphorous-containing steels. In Formable HSLA and Dual-Phase
Steels, pages 109125, Chicago, IL, USA, October 26 1977. AIME.
[18] A. Okamoto and M. Takahashi. Control of strength and r-value in box-annealed dual
phase steel sheet. In Fundamentals of Dual-Phase Steels, pages 427445, Chicago, IL,
USA, February 2324 1981. AIME.
[19] A. F. Crawley, M. T. Shehata, N. Pussegoda, C. M. Mitchell, and W. R. Tyson. Processing, properties and modelling of experimental batch-annealed dual-phase steels. In
Fundamentals of Dual-Phase Steels, pages 181197, Chicago, IL, USA, February 2324
1981. AIME.

105

[20] K. Nakaoka, Y. Hosoya, M. Ohmura, and A. Nishimoto. Reassesment of the waterquenched process as a means of producing dual-phase formable steel sheets. In J.W. Morris R.A. Kot, editor, Structure and Properties of Dual-Phase Steels, pages 330345, New
Orleans, LA, USA, February 1921 1979.
[21] M. S. Rashid. Tempering characteristics of a vanadium containing dual phase steel. In
Fundamentals of Dual-Phase Steels, pages 249264, Chicago, IL, USA, February 2324
1981. AIME.
[22] R. G. Davies. Tempering of dual-phase steels. In Fundamentals of Dual-Phase Steels,
pages 265277, Chicago, IL, USA, February 2324 1981. AIME.
[23] G. R. Speich and R. L. Miller. Tempering of ferrite-martensite steels. In Fundamentals
of Dual-Phase Steels, pages 279304, Chicago, IL, USA, February 2324 1981. AIME.
[24] J. Y. Koo and G. Thomas. Design of duplex low carbon steels for improved strength:
Weight applications. In Formable HSLA and Dual-Phase Steels, pages 4055, Chicago,
IL, USA, October 26 1977. AIME.
[25] J. W. Morrow, G. Tither, and R. M. Buck. Intercritically-annealed dual-phase steels
for automotive applications. In Formable HSLA and Dual-Phase Steels, pages 151166,
Chicago, IL, USA, October 26 1977. AIME.
[26] A. R. Marder and B. L. Bramfitt. Processing of a molybdenum-bearing dual-phase steel.
In J.W. Morris R.A. Kot, editor, Structure and Properties of Dual-Phase Steels, pages
242259, New Orleans, LA, USA, February 1921 1979.
[27] G. Thomas and J. Y. Koo. Developments in strong, ductile duplex ferritic-martensitic
steels. In J.W. Morris R.A. Kot, editor, Structure and Properties of Dual-Phase Steels,
pages 183201, New Orleans, LA, USA, February 1921 1979.
[28] A. R. Marder. The structure-property relationships in chromium-bearing dual-phase
steels. In Fundamentals of Dual-Phase Steels, pages 145160, Chicago, IL, USA, February
2324 1981. AIME.
[29] C. I. Garcia and A. J. DeArdo. The formation of austenite in low-alloy steels. In J.W. Morris R.A. Kot, editor, Structure and Properties of Dual-Phase Steels, pages 4061, New
Orleans, LA, USA, February 1921 1979.
[30] P. Wycliffe, G. R. Purdy, and J. D. Embury. Austenite growth in the intercritical annealing of ternary and quaternary dual-phase steels. In Fundamentals of Dual-Phase Steels,
pages 5983, Chicago, IL, USA, February 2324 1981. AIME.
106

[31] G. R. Speich, V. A. Demarest, and R. L. Miller. Formation of austenite during intercritical


annealing of dual-phase steels. Metallurgical Transactions A, 12A:14191428, August
1981.
[32] D. Z. Yang, E. L. Brown, D. K. Matlock, and G. Krauss. Ferrite recrystallization and
austenite formation in cold-rolled intercritically annealed steel. Metallurgical Transactions
A, 16A:13851392, August 1985.
[33] S. Sun and M. Pugh. Manganese partitioning in dual-phase steel during annealing. Materials Science and Engineering A, A276:167174, 2000.
[34] S. Hayami, T. Furukawa, H. Gondoh, and H. Takechi. Recent developments in formable
hot- and cold-rolled HSLA including dual-phase sheet steels. In Formable HSLA and
Dual-Phase Steels, pages 167180, Chicago, IL, USA, October 26 1977. AIME.
[35] D. K. Matlock, G. Krauss, L. F. Ramos, and G. S. Huppi. A correlation of processing
variables with deformation behaviour of dual-phase steels. In J.W. Morris R.A. Kot,
editor, Structure and Properties of Dual-Phase Steels, pages 6290, New Orleans, LA,
USA, February 19-21 1979.
[36] S. S. Hansen and R. R. Pradhan. Structure/ property relationships and continuous
yielding behavior in dual-phase steels. In Fundamentals of Dual-Phase Steels, pages 113
144, Chicago, IL, USA, February 2324 1981. AIME.
[37] K. W. Andrews. Empirical formulae for the calculation of some transformation temperatures. Journal of the Iron and Steel Institute, 203(7):721727, July 1965.
[38] G. T. Eldis. Critical review of data sources for isothermal transformation and continuous
cooling transformation diagrams. In D. V. Doane and J. S. Kirkaldy, editors, Hardenability
Concepts with Applications to Steel, pages 126157, Chicago, IL, USA, October 2426
1977. Metallurgical Society AIME.
[39] R. G. Davies. The deformation behaviour of a Vanadium-strengthened dual-phase steel.
Metallurgical Transactions A, 9A:4152, March 1978.
[40] R. G. Davies. The mechanical properties of zero-carbon ferrite-plus-martensite structures.
Metallurgical Transactions A, 9A:451455, March 1978.
[41] W. R. Cribb and J. M. Rigsbee. Work-hardening behaviour and its relationship to thr
microstructure and mechanical properties of dual-phase steels. In J.W. Morris R.A. Kot,
editor, Structure and Properties of Dual-Phase Steels, pages 91117, New Orleans, LA,
USA, February 19-21 1979.
107

[42] G. R. Speich and R. L. Miller. Mechanical properties of ferrite-martensite steels. In


J.W. Morris R.A. Kot, editor, Structure and Properties of Dual-Phase Steels, pages 145
182, New Orleans, LA, USA, February 19-21 1979.
[43] R. D. Lawson, D. K. Matlock, and G. Krauss. The effect of microstructure on the
deformation behavior and mechanical properties of a dual phase steels. In Fundamentals
of Dual-Phase Steels, pages 347381, Chicago, IL, USA, February 2324 1981. AIME.
[44] A. R. Marder. Deformation characteristics of dual-phase steels. Metallurgical Transactions
A, 13A, January 1982.
[45] S. T. Mileiko. The tensile strength and ductility of continuous fiber composites. Journal
of Materials Science, 4:974977, 1969.
[46] H. Fischmeister, J. O. Hjaelmered, B. Karlsson, G. Linden, and B. Sundstroem. Plastic
deformation of two-phase materials with coarse microstructure. In The microstructure
and design of alloys, Proceedings of the 3rd International Conference on the Strength of
Metals and Alloys, volume 1, pages 621625, Cambridge, England, 2025 August 1973.
[47] B. Karlsson and B. Sundstrom. Inhomogeneity in plastic deformation of two-phase steels.
Materials Science and Engineering, 16:161168, 1974.
[48] H. Fischmeister and B. Karlsson. Plastizitaetseigenschaften grob-zweiphasiger werkstoffe.
Z. Metallkde, 68(5):311327, 1977.
[49] P. Uggowitzer and H. P. Stuewe. Plastizitaet von ferritisch-martensitischen zweiphasenstaehlen. Z. Metallkde, 73(5):277285, 1982.
[50] J. Gurland. Some aspects of the plastic deformation of phase mixtures with coarse microstructures. In T. N. Baker, editor, Yield, Flow and Fracture in Polycrystalls, University
of Strathclyde, Glasgow, Scotland, September 1982. Applied Science Publishers.
[51] E. Werner, T. Sigmund, H. Weinhandl, and F. D. Fischer. Properties of random polycrystalline two-phase materials. Applied Mechanics Reviews, 47(1):231240, January 1994.
[52] R. Priestner and A. K. Ibraheem. Processing of steel for ultrafine ferrite grain structures.
Materials Science and Technology, 16:12671272, November-December 2000.
[53] A. Di Schino and J.M. Kenny. Grain size dependence of the fatigue behaviour of a
ultrafine-grained AISI 304 stainless steel. Materials Letters, 57:31823185, 2003.
[54] M. Sagradi, D. Pulino-Sagradi, and R. E. Medrano. The effect of microstructure on the
superplasticity of a duplex stainless steel. Acta Materialia, 46(11):38573862, 1998.
108

[55] F. J. Humphreys, P. B. Prangnell, and R. Priestner. Fine-grained alloys by thermomechanical processing. Current opinion in Solid State and Materials Science, 5:1521,
2001.
[56] J.-H. Park, Y. Tomota, and M.-Y. Wey. Suppression of grain growth in dual-phase steels.
Materials Science and Technology, 18:15171523, December 2002.
[57] P. D. Hodgson and M. R. Hickson R. K. Gibbs. The production and mechanical properties
of ultrafine ferrite. Materials Science Forum, (10):6372, 1998.
[58] P. J. Hurley, P. D. Hodgson, and B. C. Muddle. Analysis and characterisation of ultra-fine
ferrite produced during a new steel strip rolling process. Scripta Materialia, 40(4):433
438, 1999.
[59] M. R. Hickson and P. D. Hodgson. Effect of preroll quenching and post-roll quenching on
production and properties of ultrafine ferrite in steel. Materials Science and Technology,
15:8590, 1999.
[60] P. J. Hurley and P. D. Hodgson. Formation of ultra-fine ferrite in hot rolled strip: potential
mechanisms for grain refinement. Materials Science and Engineering A, A302:206214,
2001.
[61] M. R. Hickson, P. J. Hurley, R. K. Gibbs, G. L. Kelly, and P. D. Hodgson. The production
of ultrafine ferrite in low-carbon steel by strain-induced transformation. Metallurgical and
Materials Transactions A, 33A:10191026, April 2002.
[62] G. L. Kelly, H. Beladi, and P. D. Hodgson. Ultrafine grained ferrite formed by interrupted
hot torsion deformation of plain carbon steel. ISIJ International, 42(12):15851590, 2002.
[63] H. Beladi, G. L. Kelly, A. Shokouhi, and P. D. Hodgson. The evolution of ultrafine
ferrite formation through dynamic strain-induced transformation. Materials Science and
Engineering A, A371:343352, 2004.
[64] R. Ueji, N. Tsuji, Y. Minamino, and Y. Koizumi. Ultragrain refinement of plain low
carbon steel by cold-rolling and annealing of martensite. Acta Materialia, 50:41774189,
2002.
[65] R. Ueji, N. Tsuji, Y. Minamino, and Y. Koizumi. Effect of rolling reduction on ultrafine
grained structure and mechanical properties of low-carbon steel thermomechanically processed from martensite starting structure. Science and Technology of Advanced Materials,
5:153162, 2004.

109

[66] Y. Saito, H. Utsunomiya, N. Tsuji, and T. Sakai. Novel ultra-high straining process
for bulk materials- development of the accumulative roll-bonding (ARB) process. Acta
Materialia, 47(2):579583, 1999.
[67] V. V. Stolyarov, Y. T. Zhu, T. C. Lowe, R. K. Islamgaliev, and R. Z. Valiev. A two-step
SPD processing of ultrafine-grained titanium. NanoStructured Materials, 11(7):947954,
1999.
[68] D. H. Shin, B. C. Kim, Y.-S. Kim, H. Beladi, and K.-T. Park. Microstructural evolution
in a commercial low carbon steel by equal channel angular pressing. Acta Materialia,
48:22472255, 2000.
[69] D.-H. Shin, J.-J. Park, Y.-K. Lee, and K.-T. Park. Ultrafine-grained low carbon steels
fabricated by equal channel angular pressing: Microstructures and tensile properties. ISIJ
International, 42(12):14901496, 2002.
[70] Z. Horita, M. Furukawa, M. Nemoto, and T. G. Langdon. Development of fine grained
structures using severe plastic deformation. Materials Science and Technology, 16:1239
1245, November-December 2000.
[71] K.-T. Park, S. Y. Han, B. D. Shin, Y. K. Lee, and K. K. Um. Ultrafine grained dual phase
steel fabricated by equal channel angular pressing and subsequent intercritical annealing.
Scripta Materialia, 51:909913, 2004.
[72] Y. Son, Y. K. Lee, K.-T. Park, C. S. Lee, and D. H. Shin. Ultrafine grained ferritemartensite dual-phase steels fabricated via equal channel angular pressing: Microstructure and tensile properties. Acta Materialia, 53:31253134, 2005.
[73] N. Tsuji, Y. Saito, H. Utsunomiya, and S. Tanigawa. Ultra-fine grained bulk steel produced by accumulative roll-bonding (ARB) process. Scripta Materialia, 40(7):795800,
1999.
[74] S.-H. Lee, Y. Saito, K.-T. Park, and D. H. Shin. Microstructures and mechanical properties of ultra low carbon IF steel processed by accumulative roll-bonding process. Materials
Transactions, 43(9):23202325, 2002.
[75] K. Sugimoto, J. Sakaguchi, T. Ilda, and T. Kashima. Stretch-flangeability of a highstrength TRIP type bainitic sheet steel. ISIJ International, 40(9):920926, 2000.
[76] D.I. Hyun, S.M. Oak, S.S. Kang, and Y.H. Moon. Estimation of hole flangeability for
high strength steel plates. Journal of Materials Processing Technology, (130131):913,
2002.
110

[77] X. Fang, Z. Fan, B. Ralph, P. Evans, and R. Underhill. The relationships between tensile
properties and hole expansion property of C-Mn steels. Journal of Materials Science,
38:38773882, 2003.
[78] X. Fang, Z. Fan, B. Ralph, P. Evans, and R. Underhill. Effects of tempering temperature
on tensile and hole expansion properties of a C-Mn steel. Journal of Materials Processing
Technology, (132):215218, 2003.
[79] n.n. AHSS Guidelines, Flanging. Technical report, IISI-AutoCo.
[80] S. Traint. Phase transformations and mechanical properties of low alloyed dual-phase and
TRIP-steels. PhD thesis, TUM, 2002.
[81] E. Girault, P. Jacques, Ph. Harlet, K. Mols, J. Van Humbeeck, E. Aernoudt, and F. Delannay. Metallographic methods for revealing the multiphase microstructure of TRIPassisted steels. Materials Characterization, (40):111118, 1998.
[82] B. L. Bramfitt and J. G. Speer. A perspective on the morphology of bainite. Metallurgical
Transactions A, 21A:817829, April 1990.
[83] James. F. Shackelford. Introduction to materials science for engineers, chapter 6.1, page
185ff. Prentice Hall, 5th edition, 2000.
[84] M. S. Rashid. Relationship between steel microstructure and formability. In Formable
HSLA and Dual-Phase Steels, pages 124, Chicago, IL, USA, October 26 1977. AIME.
[85] Victor Kerlins. ASM Handbook: Fractography, volume 12, chapter Modes of Fracture,
page 12ff. ASM International, 9th edition, 1989.
[86] Barbra L. Gabriel. ASM Handbook: Fractography, volume 12, chapter Scanning Electron
Microscopy, page 166ff. ASM International, 9th edition, 1989.
[87] P. Tsipouridis, C. Krempaszky, E. Werner, E. Tragl, S. Traint, and A. Pichler. Influence of
grain refinement on the mechanical properties of a dual-phase steel. In Materials Science
and Technology 2004 Conference Proceedings, volume I, pages 735746, N. Orleans, LA,
USA, September 2629 2004. AIST Proceedings.
[88] S. E. Offerman, N. H. van Dijk, M. Th. Rekveldt, J. Sietsma, and S. van der Zwaag.
Ferrite/pearlite band formation in hot rolled medium carbon steel. Materials Science
and Technology, 18:297303, March 2002.
[89] A. Sakir Bor. Note: Effect of pearlite banding on mechanical properties of hot-rolled steel
plates. ISIJ International, 31(12):14451446, 1991.
111

[90] T. A. Kop, J. Sietsma, and S. van der Zwaag. Anisotropic dilatation behaviour during
transformation of hot rolled steels showing banded microstructure. Materials Science and
Technology, 17:15691574, December 2001.
[91] J.D. Verhoeven. A review of microsegregation induced banding phenomena in steels.
Journal of materials engineering and performance, 9, June 2000.
[92] J.D. Verhoeven. Metallurgy of Steels for Bladesmiths & Others who Heat Treat and Forge
Steel, chapter Solidification, pages 159169. ASM International, March 2005.
[93] W. Xu, P. Rivera-Diaz del Castillo, and S. van der Zwaag. Ferrite/pearlite band prevention in dual-phase and trip steels: Model development. ISIJ International, 45(3):380387,
2005.
[94] R. Groterlinden, R. Kawalla, U. Lotter, and H. Pircher. Formation of pearlitic banded
structures in ferritic-pearlitic steels. Steel Research, 63(8):331336, 1992.
[95] Z. Jiang, J. Liu, and J. Lian. A new relationship between the flow stress and the
microstructural parameters for a dual phase steel. Acta Metallurgica et Materialia,
40(7):15871597, 1992.
[96] Q. Chen, O. Pawelski, and R. Caspar. Contribution to the deformation characteristics of
dual-phase steels. Z. Metallkde, 76(5):348352, 1985.
[97] K. Honda and Z. Nishiyama. Sci. Rep. Tohoku Univ., 21:299, 1932.
[98] Zenji Nishiyama. Martensitic Transformation, chapter 2, pages 1617. Academic Press,
1978.
[99] U. Liedl, S. Traint, and E. A. Werner. An unexpected feature of the stress-strain diagram
of dual-phase steel. Computational Materials Science, 25:122128, 2002.
[100] U. Liedl. Anfangsverformungs- and Alterungsverhalten von Dual-Phasen Stahl. PhD
thesis, TUM, Lehrstuhl fuer Werkstoffkunde unde Werkstoffmechanik, 2003.
[101] A. M. Sherman, R. G. Davies, and W. T. Donlon. Electron microscopic study of deformed
dual-phase steels. In Fundamentals of Dual-Phase Steels, pages 8594, Chicago, IL, USA,
February 2324 1981. AIME.
[102] R. L. Reuben and T. N. Baker. The tensile deformation of a martensitic dual-phase steel.
Materials Science and Engineering, 63:229238, 1984.
[103] N. K. Balliger and T. Gladman. Work hardening of dual-phase steels. Metal Science,
15:95108, 1981.
112

[104] P.-H. Chang and A. G. Preban. The effect of ferrite grain size and martensite volume
fraction on the tensile properties of dual phase steel. Acta Metallurgica, 33(5):897903,
1985.
[105] F. B. Pickering and T. Gladman. Metallurgical developments in carbon steels. ISI Special
Report, (81):1020, 1963.
[106] F. B. Pickering. Physical Metallurgy and the Design of Steels, page 275. Applied Science
Publishers, 1978.
[107] P. D. Hodgson and M. R. Hickson R. K. Gibbs. Ultrafine ferrite in low carbon steel.
Scripta Materialia, 40(10):11791184, 1999.
[108] B. Q. Han and S. Yue. Processing of ultrafine ferrite steels. Journal of Materials Processing and Technology, 136:100104, 2003.
[109] M. Y. Liu, B. Shi, C. Wang, S. K. Ji, X. Cai, and H. W. Song. Normal Hall-Petch
behaviour of mild steel with submicron grains. Materials Letters, 57:27982802, 2003.
[110] R. W. Armstrong. The yield and flow stress dependence on polycrystal grain size. In
T. N. Baker, editor, Yield, Flow and Fracture in Polycrystalls, pages 131, University of
Strathclyde, Glasgow, Scotland, September 1982. Applied Science Publishers.
[111] G. T. Hahn. A model for yielding with special reference to the yield-point phenomena of
iron and related bcc metals. Acta Metallurgica, 10:727739, 1962.
[112] M. F. Ashby. The deformation of plastically non-homogeneous materials. Philosophical
Magazine, 21:399424, 1970.
[113] Ref. 2 (J. H. Hollomon, Trans. AIME 162 (1945) 268), Ref. 1 (P. Ludwik, Elemente der
Technologischen Mechanik, Springer-Verlag, Berlin (1909)) and Ref. 4 (H. W. Swift, J.
Mech. Phys. MSolids 1 (1952) 1) as cited in L. Ratke and P. I. Welch, The questionability
of empirical Work-Hardening laws. Z. Metallkde, 74(4):226232, 1983.
[114] J. Gerbase, J.D. Embury, and R.M. Hobbs. The mechanical behaviour of some dual
phase steels - with emphasis on the initial work hardening rate. In J.W. Morris R.A. Kot,
editor, Structure and Properties of Dual-Phase Steels, pages 118144, New Orleans, LA,
USA, February 1921 1979.
[115] L. Ratke and P. I. Welch. The questionability of empirical work-hardening laws. Z.
Metallkde, 74(4):226232, 1983.

113

[116] W. D. Nix and H. Gao. Indentation size effects in crystalline materials: a law for strain
gradient plasticity. Journal of the Mechanics and Physics of Solids, 46(3):411425, 1998.
[117] W.W. Gerberich, N. I. Tymiak, J. C. Grunlan, M. F. Horstemeyer, and M. I. Baskes.
Interpretation of indentation size effects. Journal of Applied Mechanics, 69:433442, July
2002.
[118] R. Rodrguez and I. Gutierrez. Correlation between nanoindentation and tensile properties. Influence of the indentation size effect. Materials Science and Engineering A,
361:377384, 2003.
[119] Y. Wei and W. Hutchinson. Hardness trends in micron scale indentation. Journal of the
Mechanics and Physics of Solids, 51:20372056, 2003.
[120] K. Miyahara, S. Matsuoka, and T. Hayashi. Nanoidentation as a strength probe- a study
on the hardness dependence of indent size for fine-grained and coarse-grained ferritic steel.
Metallurgical and Materials Transactions A, 32A:761768, March 2001.
[121] N. A. Fleck and J. W. Hutchinson. A phenomenological theory for strain gradient plasticity. Journal of the Mechanics and Physics of Solids, 41(12):18251857, 1993.
[122] N. A. Fleck, G. M. Muller, M. F. Ashby, and J. W. Hutchinson. Strain gradient plasticity:
theory and experiment. Acta Metallurgica et Materialia, 42(2):475487, 1994.
[123] J. Y. Shu and N. A. Fleck. The prediction of a size effect in micro-indentation. International Journal of Solids and Structures, 35(13):13631383, 1998.
[124] A. A. Elmustafa and D. S. Stone. Indentation size effect in polycrystalline fcc metals.
Acta Materialia, 50(14):36413650, 2002.
[125] A. A. Elmustafa and D. S. Stone. Nanoindentation and the indentation size effect: Kinetics of deformation and strain gradient plasticity. Journal of the Mechanics and Physics
of Solids, 51(2):357381, 2003.
[126] Z. Kovacs, N. Q. Chinh, J. Lendvai, Z. Horita, and T. G. Langdon. Effect of indentation
size on plastic deformation processes in an ultrafine-grained Al-3 % Mg alloy. Materials
Science Forum, 396-402:pp. 1073, 2002.
[127] A. Nishimito, Y. Hosoya, and K Nakaoka. Relation between hole expansion formability
and metallurgical factors in dual-phase steel sheet. In Fundamentals of Dual-Phase Steels,
pages 447463, Chicago, IL, USA, February 2324 1981. AIME.

114

[128] K. Hasegawa, K. Kawamura, T. Urabe, and Y. Hosoya. Effects of microstructure on


stretch-flange-formability of 980 MPa grade cold-rolled ultra high strength steel sheets.
ISIJ International, 44(3):603609, 2004.
[129] R. D. K. Misra, S. W. Thompson, T. A. Hylton, and A. J. Boucek. Microstructures of
hot-rolled high-strenth steels with significant differences in edge formability. Metallurgical
and Materials Transactions A, 32A:745760, March 2001.

115

Lebenslauf
Pers
onliche Daten
Name, Vorname:

TSIPOURIDIS PRODROMOS

Geburtsdatum, Geburtsort: 23.03.1978, Drama (Griechenland)


Staatsangehorigkeit: Griechisch
Adresse:

Implerstrasse 4, 81371, M
unchen (Deutschland)

E-Mail Adresse:

tsipouridis@wkm.mw.tum.de, makis.tsipouridis@googlemail.com

Ausbildung
Seit 2002:

Wiss. Mitarbeiter am Lehrstuhl f


ur Werkstoffkunde und Werkstoffmechanik, Fakultat f
ur Maschinenwesen, TU-M
unchen (Pr. Dr. mont.
E. Werner)
Forschungsgebiet I: Dual- und Multi-Phasen Stahle, Promotion zum
Thema Mechanical properties of Dual-Phase steels (June 2006),
Forschungsgebiet II (Aktuell): European RFCS-Program: DP-grades
with improved formability

2001:

Wiss. Mitarbeiter am Lehrstuhl f


ur Technische Chemie II, Fakultat f
ur
Chemie, TU-M
unchen (Pr. Dr. J. Lercher)

2001:

Wiss. Mitarbeiter am Lehrstuhl f


ur Physikalische Chemie, Fakultat f
ur
Chemie-Ingenieurwesen, Aristoteles-Universitat von Saloniki (GR)

1995-2000:

Diplomstudium:
Fakultat f
ur Chemie-Ingenieurwesen, AristotelesUniversitat von Saloniki (GR), November 2000
Diplomarbeit zum Thema: Electrocatalytic Dehydrogenation of SiH4 -CH4
mixture in a Proton Conducting Solid Electrolyte Cell

Stipendien
1998:

Stipendium von I.K.Y. (State Scholarships Foundation, GR)

Sprachkenntnisse
Griechisch (Muttersprache), Englisch (flieend), Deutsch (gute Kenntnisse)

Muenchen, October 30, 2008

Você também pode gostar