TTWKoopmansThesis_final.

TTWKoopmansThesis_final.
Thermal stability of retained austenite
in Quenching & Partitioning steels
Master of Science Thesis
For the degree of Master of Science in Mechanical Engineering at Delft University of Technology
T.T.W. Koopmans
July 06, 2015
Faculty of Mechanical, Maritime and Materials Engineering (3mE) / Delft University of Technology
This page intentionally left blank
Table of contents
Table of contents
i
Glossary
ii
Acknowledgements
iv
Chapter 1
Introduction
1
Chapter 2
Background
2
2.1
Key concepts in this thesis
2
2.2
Background on Quenching & Partitioning steels
4
2.3
Thermal stability of retained austenite
5
2.4
Approach
9
Chapter 3
Methods, Equipment and Procedures
10
3.1
General overview
10
3.2
Dilatometry
12
3.3
Magnetic measurements
21
3.4
X-ray Diffraction
25
3.5
Microscopy
26
Chapter 4
Results and Discussion
29
4.1
Microstructures after Quenching & Partitioning
29
4.2
Retained austenite decomposition: reheating with 5 °C s-1 to 700 °C
37
4.3
Influence of carbon content in retained austenite on its decomposition
45
4.4
Effect of partitioning time on thermal stability of retained austenite
51
4.5
Discussion and theory of observed RA decomposition mechanisms
55
Chapter 5
Conclusions
60
5.1
Retained austenite decomposition in Quenching & Partitioning steel
60
5.2
About experimental techniques
61
5.3
About Quenching & Partitioning steel
61
Chapter 6
Recommendations
62
Chapter 7
References
63
Appendix A
Comparison between retained austenite fractions as determined by VSM and XRD 68
Appendix B
Stress relief during annealing in a dilatometer
i
69
Glossary
List of symbols
A
Area (m2)
V
Volume (m3)
E
Modulus of elasticity (GPa)
L
Length change (m)
L0
Original length (m)
Tc
Curie Temperature
Bs
Bainite start temperature (°C)
Mf
Martensite finish temperature (°C)
Ms
Martensite start temperature (°C)
A1
Lower critical temperature for austenite (°C)
A3
Upper critical temperature for austenite (°C)
QT
Quenching temperature (°C)
PT
Partitioning temperature (°C)
tp
Partitioning time (s)
Ms
Saturation Magnetization (A m2 kg-1)
M
Magnetization (A m2 kg-1)
G
Gibbs Free Energy (J mol-1)
G
Gibbs Free Energy difference or Driving force (J mol-1)
𝑓
Phase fraction
Greek symbols
Ξ±
Phase: ferrite
α’
Phase: martensite
Ξ³
Phase: austenite
ο₯
Phase: epsilon carbide

Phase: eta carbide
ΞΈ
Phase: cementite
Ξ±
Thermal expansion coefficient (K-1)

Heating rate (°C min-1)
ii
List of acronyms
Nital
Nitric acid in Alcohol
VSM
Vibrating Sample Magnetometer
XRD
X-Ray Diffraction
SEM
Scanning Electron Microscope
(L)OM
(Light) Optical Microscope
DSC
Differential Scanning Calorimeter
DTA
Differential Thermal Analysis
EBSD
Electron Backscatter Diffraction
EDS
Energy Dispersive X-ray Spectroscopy
M1
Martensite from the first quench
M2
Martensite from the second quench
RA
Retained Austenite
Q&P
Quenching & Partitioning
FCC
Face Centered Cubic
BCC
Body Centered Cubic
BCT
Body Centered Tetragonal
TTT
Time-Temperature-Transformation
CCT
Continuous Cooling Transformation
TRIP
Transformation Induced Plasticity
iii
Acknowledgements
I would like to thank Maria Santofimia Navarro for being my daily supervisor. Thank you for all the
interesting discussions we had, your patience and for showing me new points of view whenever I was
stuck. I would like to express my gratitude to my professor Jilt Sietsma for always being available for
interesting discussions and for sharing an impressive amount of knowledge on how to approach the
study of steels. Nico Geerlofs played an integral part in making this thesis possible. Nico, thanks for
all the help with the dilatometer and the discussions on how to do lab work in general.
I would also like to thank Lie Zhao and Ron van Tol for pointing me to the master materials science,
and for getting me started.
In no particular order, I would also like to express my gratitude to Peter van Liempt, Kees
Kwakernaak, Sander van Asperen, Richard Huizinga, Pina Mecozzi, Hans Brouwer and Farideh Haji
Akbari for help with experiments and valuable discussions.
I would like to express my gratitude to Stefan van Bohemen for agreeing to be a committee member
and valuable discussions, and Marcel Sluiter for agreeing to be a committee member as well.
Finishing my master thesis would have been impossible were it not for the Ph. D.’s and postdocs in
the department. Thanks Ashwath, Ankit, Alfonso, Bij-Na, Constantinos, Javier and Zaloa for the good
times.
Furthermore, I would like to thank Frans Bosman for the good training in the departmental gym, and
Lourdes Gallastegui Pujana for valuable personal feedback.
Last but not least, I would like to thank my friends and family for just being there, and especially to
Pieter and Julia for proofreading my thesis and just being good friends in general.
iv
This page intentionally left blank
Chapter 1 Introduction
The drive for ever more safety and fuel efficiency in the automotive industry led the industry to
search for steels with enhanced strength and ductility. Promising candidates to satisfy these
demands are steels with a microstructure consisting of martensite and significant fractions of
retained austenite. One class of steels with such a microstructure are Quenching & Partitioning
(Q&P) steels.
The enhanced strength and ductility of Quenching & Partitioning steels is largely due to the presence
of retained austenite. At elevated temperatures encountered during processes such as welding, hot
dip galvanizing and paint baking, the possibility of retained austenite decomposition into
thermodynamically more stable ferrite and carbides exists. Decomposition of retained austenite
would have detrimental effects on the mechanical properties of Quenching & Partitioning steels. A
requirement for practically applicable Quenching & Partitioning steels is therefore knowledge of the
thermal stability of retained austenite against decomposition.
In this work, multiple microstructures have been created in one particular steel alloy using
Quenching & Partitioning processing. These microstructures have been characterized using X-Ray
Diffraction, Scanning Electron Microscopy, Optical Microscopy and Electron Backscatter Diffraction.
Special attention was paid to the morphology and carbon content of retained austenite. The
response of retained austenite to isothermal and isochronal annealing has been investigated using
dilatometry and thermomagnetic methods.
The main finding of this work is that retained austenite in essence behaves as austenite which is
higher in carbon content compared to the base alloy. Furthermore, the decomposition mechanisms
of retained austenite have been successfully related to existing theory about austenite
decomposition and mapped to a TTT-like diagram. Low-carbon retained austenite decomposes
significantly quicker than high-carbon retained austenite.
A general background on Q&P steel and its microstructural components is presented in Chapter 2,
followed by a literature review regarding the thermal stability of retained austenite in chemically
comparable steels. Chapter 3 describes the methods, equipment and procedures used in this work.
Chapter 4 presents results of the characterization of the created Q&P microstructures. In addition,
the response of retained austenite to isothermal and isochronal annealing is discussed. Chapter 5
presents the conclusions of this work and Chapter 6 contains some recommendations for future
work.
1
Chapter 2 Background
New global standards for vehicle safety and fuel-efficiency are becoming more demanding every year
[1]. The automotive industry keeps searching for new low-cost steels, which should achieve
significant increases in strength whilst offering a reduction in vehicle mass. Collectively, these steels
are known as Advanced High-Strength Steels (AHSS).
These AHSS grades are defined by having both high strength and relatively high formability
capabilities. Figure 2.1 shows the formability capabilities of different steels, expressed as elongation,
plotted against the tensile strength. Conventional steels achieve relatively high ductility, but have
low tensile strength, while AHSS grades generally have high tensile strength and lower ductility. The
diagram shows that when strength is increased for conventional and AHSS grades, formability
generally decreases.
Figure 2.1: Formability diagram of different steels showing the general trade-off between elongation and strength.
Reproduced from [1].
The 3rd generation of AHSS grades however, seeks to achieve both formability and high strength. As
can be seen in the diagram, the current 3rd generation AHSS achieves ductility well above previous
AHSS grades at the same strength level.
One class of steels from the current 3rd generation of AHSS grades are Quenching & Partitioning
steels. These steels can achieve high both high formability and high strength levels due to their
microstructure consisting of martensite and retained austenite. Martensite and retained austenite
will be introduced in §2.1, followed by an introduction on Q&P steels in §2.2.
2.1 Key concepts in this thesis
2.1.1
Martensite
In steels, martensite is a Body Centered Cubic (BCC) / Body Centered Tetragonal (BCT) phase formed
by a diffusionless shear transformation directly from Face Centered Cubic (FCC) austenite. This is
schematically illustrated in Figure 2.2. The martensitic transformation will only happen if the material
is cooled fast enough from the austenitic temperature region. The necessary cooling rate is called the
critical cooling rate. Due to the high cooling rate, carbon interstitials in FCC have little time available
to diffuse. Upon transformation from FCC to BCC the carbon atoms are essentially frozen into place.
2
Since there is less interstitial space available in BCC structure, the frozen carbon atoms stretch the
lattice. This stretching effect results in a BCT structure.
Fe
C
Fe
Fe
C
Figure 2.2: The FCC, BCC and BCT unit cells. Adapted from [2]. Iron and carbon atoms are marked Fe and C respectively.
The temperature at which the martensite starts to form upon quenching is called the Martensite
start temperature, or Ms. This temperature is dependent on the chemical composition of the alloy.
The Koistinen-Marburger (K-M) equation [3] allows the estimation of the fraction martensite
π‘“π‘š during a quench to a given temperature 𝑇:
π‘“π‘š = 1 βˆ’ exp[βˆ’βˆ (π‘‡π‘˜π‘š βˆ’ 𝑇)]
Equation 2.1
where ∝ is a rate parameter and π‘‡π‘˜π‘š the theoretical martensite start temperature.
In low carbon steels martensite has a lath morphology as shown in Figure 2.3a and Figure 2.3b. The
martensite laths have orientation differences of angles of about 10 degrees [4]. The laths are
confined within well parallel developed blocks, and packets contain a number of these blocks.
Martensite has excellent tensile strength and hardness, but is very brittle having low ductility and
toughness. This is due to carbon supersaturation of the martensite, large strains resulting from the
quench and the lath morphology of the martensite.
(a)
(b)
Figure 2.3: (a) Morphology of martensite in low carbon steel. Adapted from [4] (b) Martensitic microstructure as seen in the
SEM.
3
Tempering of martensite will improve the ductility and toughness of the material, at the expense of
tensile strength and hardness. During tempering, the excess carbon from the BCT structure will form
carbides, leaving a microstructure of ferrite and fine carbides. This microstructure is called tempered
martensite.
2.1.2
Retained austenite
Austenite that does not transform to martensite during the quench to room temperature is called
retained austenite. Retained austenite is metastable at room temperature. The stability of the
retained austenite against the martensitic transformation depends on its grain size, local chemical
composition of the retained austenite grain, its carbon content, dislocation density, its surrounding
environment, etc. [5]. It is important to know that higher carbon content increases retained
austenite stability.
In martensitic steels, retained austenite can enhance the ductility, toughness, fatigue life and strain
hardening rate of the steel. Retained austenite will progressively transform to martensite with
increasing strain, thereby enhancing the work hardening rate. This is called the Transformation
Induced Plasticity (TRIP) effect.
Retained austenite can have two morphologies: blocky and film-like. The blocky type originates from
geometrical partitioning of the original austenite grain [6]. The film-like type is retained between
laths of martensite. In Q&P steels, film-like retained austenite is known to have lower carbon content
compared to larger retained austenite grains [7], [8].
2.2 Background on Quenching & Partitioning steels
Quenching & Partitioning (Q&P) steels were first conceived by Speer et al. [9], [10]. The key concept
in these steels is that carbon partitioning from martensite to austenite is possible if alloying elements
are added that suppress carbide formation, such as silicon. The partitioning of carbon into austenite
can stabilize the austenite against martensitic transformation, leaving retained austenite at room
temperature. These steels are created via the Quenching & Partitioning process, as schematically
illustrated in Figure 2.4.
The Q&P process starts with partial formation of martensite from a fully austenitized condition by
quenching to a selected temperature (called Quenching Temperature, QT) between Ms and room
temperature. This is followed by an annealing treatment at a higher temperature (called Partitioning
Temperature, PT) in which the carbon from the supersaturated martensite can partition to the
remaining austenite. This creates carbon-depleted tempered martensite and carbon-enriched
austenite, promoting the stability of austenite against the martensitic transformation in a further
quench. After giving sufficient time for carbon partitioning to occur, the material is quenched to
room temperature. Austenite that was not sufficiently carbon-enriched transforms to martensite
during this quench. The final microstructure consists of carbon-depleted tempered martensite from
the first quench, fresh martensite from the second quench which has a (slightly) higher carbon
content, and carbon-enriched retained austenite. In this work, the carbon-depleted tempered
martensite will be called M1, the fresh martensite from the second quench will be called M2 and the
carbon-enriched retained austenite will be called RA.
4
M1
M2
Figure 2.4: Schematic illustration of Q&P heat treatment, and the microstructure at each step of the treatment. QT and PT
are the quenching temperature and partitioning temperature respectively, while Ci, CΞ³ and Cm represent the carbon
concentrations in the initial alloy, austenite and martensite respectively. [9]
The microstructures resulting from the Q&P process are complex and vary depending on processing
parameters. The composition of the alloy and the selection of the quenching temperature,
partitioning temperature and partitioning time all have an effect on the resulting microstructure.
2.3 Thermal stability of retained austenite
At elevated temperatures, metastable retained austenite can decompose into thermodynamically
more stable ferrite and carbides. These elevated are encountered during processes such as welding,
paint baking or galvanizing. In Quenching & Partitioning steels, the decomposition of retained
austenite would lead to a loss of its enhanced formability properties. It is therefore necessary to
understand the factors influencing the thermal stability of retained austenite for usable Q&P steels.
In this work, the thermal stability of retained austenite is defined as: Resistance of the retained
austenite against decomposition into more stable phases such as ferrite and cementite at elevated
temperatures. In literature, a few studies have been found mentioning the decomposition of
retained austenite in Q&P steel. None however, were devoted to it. These studies will be
summarized, followed by an overview of literature which studied the decomposition of retained
austenite in Transformation Induced Plasticity (TRIP) and Bainitic steels. TRIP and Bainitic steels
contain significant fractions of retained austenite. They are useful for a general idea of the
mechanisms that cause retained austenite decomposition.
2.3.1
Retained austenite decomposition in Q&P steels
In a conference paper by De Moor et al. [11], the effect of Si (0.24C-1.61Mn-1.45Si) and Al (0.18C1.56Mn-1.73Al) additions on tempering in Q&P steels by means of dilatometry and Differential
Scanning Calorimetry (DSC) was studied. These grades were reheated after Q&P processing to 600 °C
with heating rates of 10 °C min-1, 20 °C min-1 and 30 °C min-1. Activation energies for retained
austenite decomposition of 125 kJ mol-1 for the CMnAl grade and 202 kJ mol-1 for the CMnSi grade
were measured. With a heating rate of 20 °C min-1, retained austenite decomposition was thought to
happen between 350 °C and 420 °C for the CMnAl grade.
5
Bigg et al. [12] have studied the dynamics of a Q&P steel (0.64C-4.57Mn-1.30Si) which has a Ms - Mf
range spanning room temperature. The partitioning step at 500 °C was then studied by reheating in a
powder diffractometer. The austenite lattice parameter reached a maximum at 480 °C, and starts
declining thereafter. After the specimens reached 500 °C, the retained austenite fraction began to
decline with increasing partitioning time. From these two observations it was concluded that carbide
formation was occurring in order to consume the carbon released from decomposing austenite.
Carbide precipitation was confirmed by the presence of carbide peaks at room temperature. Not
mentioned was the type of carbide formed. Another observation was that silicon alloying additions
delay, rather than prevent, equilibrium carbide formation at 500 °C.
Another paper by Bigg et al. [13] on the same alloy (0.64C-4.57Mn-1.30Si) shows no decomposition
of retained austenite during a 90 minutes isothermal holding at a lower partitioning temperature of
300 °C. This was in contrast to other papers (not mentioned which ones), which measured a decrease
in retained austenite fraction during partitioning. An increase in carbon content of the retained
austenite during partitioning was measured.
A conference paper by Fawad [14] on a Q&P alloy (0.37C-0.85Mn-1.25Si-1.18Cr) investigated the
tempering of this Q&P steel during reheating and isothermal holding. Blocky and film type retained
austenite was detected, a total of 6 volume %. After 2 hour isothermal tempering at 250 °C, the
blocky type retained austenite was not detected in SEM micrographs. The thin film type retained
austenite was thermally more stable and transformed after 2 hours isothermal tempering at 300 °C.
Still, 4% of retained austenite was observed with XRD after tempering. Observed was that β€œThe
stability of RA depends on its carbon content, size and location within the structure”. The
concentration of carbon in the retained austenite was determined to be about 1 wt. %. After
isothermal tempering for 2 hours at 500 °C, no retained austenite was detected. Upon reheating to
600 °C with a heating rate of 10 °C min-1 in a dilatometer, decomposition of retained austenite was
detected between 250 °C and 360 °C.
2.3.2
Retained austenite decomposition in TRIP steels
Jun et al. [15] studied the decomposition behavior of residual austenite in a TRIP (0.2C-1.51Mn1.96Si) steel during a coiling simulation. After a hot rolling simulation, isothermal holding was
performed using a salt bath. The isothermal holding temperatures were 350 °C, 400 °C and 450 °C,
and the holding times were 5, 20, 60, 120 and 480 minutes. Furthermore, after 20 min holding at 400
°C and quenching to room temperature, a sample was reheated from RT to 500 °C in a
diffractometer.
Table 2.1: Summary of phases transformed from retained austenite during isothermal holding [15].
Isothermal holding
temperature
350 °C
20 min
60 min
480 min
Stable
Stable
Stable
400 °C
Stable
Cementite
Cementite
450 °C
Cementite
Cementite/pearlite
Cementite/pearlite
In-situ XRD heating revealed that retained austenite is thermally stable up to 350 °C. At temperatures
higher than 370 °C the retained austenite can decompose into cementite, ferrite and pearlite. A
6
summary of phases formed from retained austenite during isothermal holding is given in [15], which
is reproduced here as Table 2.1.
The different decomposition behavior of retained austenite at the different temperatures was
thought to be correlated with the diffusivity of carbon. Because of low diffusion of carbon at 350 °C,
the retained austenite was stable, while a much higher diffusion rate at 450°C made the austenite
very unstable. At 400 °C, initial carbon enrichment of retained austenite occurred, but after 1 hour
the retained austenite began to decompose into ferrite and cementite due to solute carbon
redistribution.
In a paper by Shi et al. [16], decomposition of retained austenite in a TRIP steel (0.12C-1.5Mn-0.7Si)
was investigated by means of differential scanning calorimetry (DSC). It was concluded that the
thermal decomposition temperature of retained austenite occurs in the range of 300 °C to 550 °C,
with an activation energy of 212 kJ mol-1.
Amirthalingam et al. [17] studied the decomposition behavior of austenite in a TRIP alloy (0.19C1.63Mn-0.35Si) by thermomagnetic methods. A welded fusion zone and the base metal were both
examined. The samples were reheated in a VSM to 600 °C with a reheating rate of 0.03 °C s-1. In the
base metal, once the temperature reached 295 °C, formation of ο₯ (Fe2,4C) carbide from retained
austenite occurred. Between 390 °C and 400 °C, possibly some  (Fe2C) carbide formation occurred.
Above 400 °C, formation of cementite occurred. The temperature range of retained austenite
decomposition was 290 °C to 440 °C.
In the welded sample, a lower fraction of retained austenite was detected, but the carbon content in
the retained austenite was higher. Precipitation of ο₯ carbide was detected at 240 °C, and the
temperature range of retained austenite decomposition was 240 °C to 460 °C. The decomposition
kinetics of higher carbon retained austenite (welded sample) is therefore slower compared to the
lower carbon retained austenite (base metal).
2.3.3
Retained austenite decomposition in Bainitic steels
Luzginova et al. [18] studied the thermal stability of retained austenite by thermomagnetic methods
with a bainitic SAE 52100 steel (1.01C-1.36Cr-0.32Mn-0.25Si). Multiple samples with different bainitic
holding times and temperatures, resulting in different volume fractions of retained austenite, were
reheated to 800°C in a VSM with a heating rate of 5 °C min-1. The temperature at which retained
austenite started to decompose was between 180 °C and 240 °C. During reheating, formation of ο₯
carbide during retained austenite decomposition around 280 °C was detected. The decomposition
was finished at 300 °C.
Saha Podder et al. [19] investigated a (0.22C-3Mn-2.03Si) bainitic steel. After bainitic holding at 390
°C for 2 hours, the sample was cooled to room temperature. Tempering was carried out at 450 °C,
and fresh martensite was detected after tempering. Tempering was found to destabilize the retained
austenite due to local reduction in carbon concentration following the precipitation of minute
quantities of cementite. This theory was proposed in 1956 by Cameron [20]. Only after one hour of
tempering and further destabilization does ferrite formation occur from retained austenite.
Tempering at 250 °C for 5 hours was found to have no significant effect on the retained austenite.
7
Saha Podder et al. [21] investigated a (0.39C-4.09Ni-2.05Si) bainitic steel. After bainitic holding at 380
°C for 2 hours, the sample was cooled to room temperature, and then tempered at 400 °C for varying
duration. During the tempering in the synchrotron at the ESRF, diffraction spectra were collected.
From these spectra the retained austenite fraction and carbon concentration were determined. It
was observed that both blocky and film type retained austenite existed in the material. Film type
retained austenite was observed to have higher carbon content, and also to decompose earlier. The
reason for this was thought to be the greater driving force for cementite precipitation due to the
higher carbon content.
2.3.4
Retained austenite decomposition in other steels
Special attention is called to a paper by Morra et al. [22], where the decomposition of retained
austenite in multiple 1 wt. % C, low silicon alloys is studied. The temperature range in which retained
austenite decomposed was 200 °C to 350 °C, with activation energies are in the range of 135-156 kJ
mol-1. This could suggest carbon diffusion in retained austenite as the rate controlling mechanism
(ο‚»140 kJ mol-1), but alternatively, interface movement controlled growth (ο‚»140 kJ mol-1) for the Ξ³ οƒ 
Ξ± transformation is suggested as the rate controlling mechanism.
Waterschoot et al. [23] studied the tempering kinetics of the martensitic phase in DP steel (0.72C1.53Mn-0.11Si). Decomposition of retained austenite was found to occur in the temperature range
250 °C to 350 °C. The activation energy was found to be 154.7 kJ mol-1. Furthermore, multiple
carbides (Fe2C, Fe2C5) were pointed out as capable of forming from retained austenite
decomposition.
Primig et al. [24] studied the activation energy of decomposition of retained austenite in a DSC with
quenched SAE1040 steel (0.67C-0.75Mn-0.22Si). This activation energy was determined to be 166185 kJ mol-1.
2.3.5
Discussion based on literature
Literature seems to focus on the decomposition of retained austenite in the temperature range
between 200 and 400 °C. High (> 1 wt. %) Mn and/or high (> 1.5 wt. %) Si alloys show slow
decomposition of retained austenite in this temperature range, especially after heat treatments
which allow the enrichment of residual austenite with carbon such as TRIP or bainitic treatments. In
literature it is established that silicon significantly slows the precipitation of cementite from austenite
in this temperature range [25], indicating that Si and C play significant roles preventing the
decomposition of retained austenite. At longer isothermal holding times in this temperature region
in steels with high Si content, eventually carbide or cementite formation occurs [15], indicating that
retained austenite is metastable at these temperatures.
Contradicting results in literature seem to suggest that both higher and lower carbon retained
austenite decompose first. The decomposition mechanism of low-carbon retained austenite is not
studied in the literature in this review. High carbon retained austenite in steels with Si forms
cementite first, eventually followed by pearlite formation around 450 °C. This indicates that high
carbon retained austenite is hypereutectoid austenite, which will decompose with cementite as the
leading phase, followed by pearlite formation at temperatures above 400 °C.
8
The activation energy of decomposition is 120 kJ mol-1 to 220 kJ mol-1, depending on composition.
Typically, in literature the rate controlling event of retained austenite decomposition is attributed to
be diffusion of carbon in austenite.
2.4 Approach
Literature offers a fragmented overview of retained austenite decomposition, and a complete
approach seems to be missing with respect to studying retained austenite decomposition in all
temperature regions between room temperature and below A1. Furthermore, no coherent theory of
retained austenite decomposition was found in the studied literature. In Q&P steels, relatively little is
known about retained austenite decomposition. This work therefore focuses on the following:
ο‚·
ο‚·
ο‚·
ο‚·
Creation of Q&P microstructures and characterization of the created microstructures.
Integral approach: study decomposition in a wide range of temperatures below A1.
A study of the response of retained austenite to both isothermal and isochronal annealing
If possible, formulation of a coherent theory of retained austenite decomposition.
A dilatometer will be used to create Q&P microstructures. To study the decomposition behavior of
retained austenite during reheating, dilatometry and thermomagnetic methods will be used.
However, in literature [24], [26], the decomposition of retained austenite between 250 and 350 °C
and the transformation of transition carbides and clustered carbon into cementite often occurs
simultaneously, complicating the study of retained austenite decomposition in a dilatometer.
However, Q&P steels are ideal candidates for the study of retained austenite decomposition in a
dilatometer, since a low fraction of carbon is available for clustering of carbon and formation of
transition carbides.
9
Chapter 3 Methods, Equipment and Procedures
3.1 General overview
In this chapter, an overview of the experimental procedures is given. The alloy used is introduced,
and selected theoretical properties of this alloy are calculated. An overview of the Quenching &
Partitioning treatments used in this work and the post-processing of the resulting samples is
presented. Furthermore, experimental details on the equipment used are given. Dilatometry,
magnetic methods and Electron Backscatter Diffraction are extensively treated. The reader is
assumed familiar with X-ray Diffraction, Scanning Electron Microscopy and Optical Microscropy, and
these methods are not treated in extensive detail.
3.1.1
Materials
The material used was an alloy specially made for research in Q&P steel, and will be referred to as
QP-G. The composition in weight percentage can be seen in Table 3.1. The steel was produced using
a laboratory vacuum induction furnace. After casting, the steel was hot rolled to a final thickness of 4
mm and then air cooled. Cylindrical specimens of 3.5 mm in diameter and 10 mm length were then
machined parallel to the rolling direction for dilatometry, comparable to the procedure in [27].
Table 3.1: Composition of QP-G (wt. %)
Steel
QP-G
C
0.20
Mn
3.51
Si
1.525
Mo
0.509
Al
0.03
S
0.0079
P
0.006
Fe
Balance
At Mn concentrations higher than 3 wt. %, the possibility of Mn segregation exists. QP-G is therefore
susceptible to Mn segregation, and segregation was observed as microstructural banding in some
specimens. A comparison can be made between regions where the Mn concentration is slightly
higher than the base alloy, and regions where it is slightly lower than the base alloy. In regions where
the Mn concentration is slightly higher, the A1 temperature will be slightly lower. The undercooling
during quenching will therefore be smaller. Likewise, regions in which the Mn concentration is
slightly lower will experience a higher undercooling. This difference in undercooling leads to a visibly
different reaction to martensite formation during quenching or bainite formation during isothermal
holding, and is seen as microstructural banding. Figure 3.1 shows typical microstructural banding
encountered in specimens in this work.
Figure 3.1: Optical Micrograph of a sample with visible microstructural banding.
10
The Martensite start and Bainite start temperatures of QP-G were determined using empirical
equations by van Bohemen [28]. The determined Ms temperature was 325 °C, while the determined
Bs temperature was 401 °C.
According to Kop [29], the Curie temperature is determined using Equation 3.1:
𝑇𝑐 = 1042 𝐾 βˆ’ π‘₯𝑀𝑛 × 1500𝐾
Equation 3.1
where 𝑇𝑐 is the Curie temperature and π‘₯𝑀𝑛 is the atomic fraction of Mn in the material. For QP-G,
this gives 𝑇𝑐 = 989 K (716 °C). Another estimate is possible by interpolating data on the effect of Mn,
Si and Mo by Arajs [30], which gives 𝑇𝑐 = 993 K (720 °C). The experimentally determined Curie
temperature was determined from dilatometry data as 𝑇𝑐 = 988 ±3 K (715 ±3 °C).
3.1.2
Design of Quenching & Partitioning treatments
The Q&P treatments applied in this work consisted of full austenitization for 180s at 900 °C, followed
by cooling with 50 °C s-1 to quench temperatures ranging from 340 °C to 140 °C. This was followed by
isothermal holding for 3s at the quench temperature and heating with 10 °C s-1 to the partitioning
temperature of 400 °C. An isothermal holding for 50 s at the partitioning temperature was followed
by a quench to room temperature with 50 °C s-1. All Q&P treatments were performed in a
dilatometer (see § 3.2). The Q&P heat treatments are schematically shown in Figure 3.2a.
Selected samples were reheated directly following the Q&P treatments. The reheating treatments
are schematically shown in Figure 3.2b, Figure 3.2c and Figure 3.2d. Details about which samples
were reheated are found in the main text. A few samples were also directly quenched, and details
about the quenching treatment are found in Figure 3.2e.
(a)
(b)
(c)
(d)
(e)
Figure 3.2: Schematic overview of the heat treatments applied in this work. (a) Q&P heat treatments. (b) (c) (d) Annealing
treatments applied to selected specimens directly following the Q&P heat treatments. (e) Quenching heat treatments.
11
In this work, samples are referred to with a combination of 2 different identifiers:
1. Quenching temperature xxx (e.g. QTxxx)
2. If applicable, reheating temperature yyy and reheating rate zz (e.g. Hyyy_zzCmin)
Selected examples:
1. Quenched and partitioned sample with a quenching temperature of 260 °C : QT260
2. Quenched and partitioned sample with a quenching temperature of 320 °C, reheated to 600
°C with a heating rate of 10 °C min-1: QT320H600_10Cmin
Some exceptions to this general rule exist in § 4.4, and will be clarified in the accompanying text.
3.1.3
Preparation of samples for microstructural investigation
After the Q&P treatment in the dilatometer, the specimens were cut using a Struers Minitom with a
diamond grinding disc (Struers M1D10) into 3 parts as schematically shown in Figure 3.3. Due to
water cooling during cutting, no retained austenite is assumed to have decomposed thermally during
the cutting. The cut side of the 6mm long cylinder was used for X-Ray Diffraction, Optical
Microscopy, Scanning Electron Microscopy and Electron Backscatter Diffraction after metallographic
preparation, while the end disc was used for magnetic measurements.
Figure 3.3: Schematic of cutting of sample and uses for the cut parts.
The metallographic preparation was conventional grinding using SiC papers up to 2000 grit, and
polishing using 3 ΞΌ and 1 ΞΌ diamond suspension. The specimen was then etched with 2% nital for
subsequent use in SEM and OM. The XRD measurements on the specimens were performed after
grinding and polishing up to 3 ΞΌ diamond suspension. EBSD measurements on the specimens were
performed after grinding and polishing with a finishing step of 0.020 ΞΌ alumina particles in a neutral
solution, OP-AN (Struers).
3.2 Dilatometry
3.2.1
Equipment
A dilatometer is a device that measures length change of a specimen during the application of a heat
treatment. In this work, a dilatometer is used to apply heat treatments to samples and record the
resulting length changes due to thermal expansion or contraction and microstructural processes,
such as the formation of martensite and austenite.
12
In Figure 3.4, a schematic representation of a push rod dilatometer is shown. The sample is clamped
between two quartz push rods, and a Linear Variable Differential Transformer is used to record
length changes in the sample and the push rods. A high-frequency induction coil is then used to heat
the sample, and using small holes in the induction coil a cooling gas can be applied to evenly cool the
sample. The thermocouple is used to record the temperature and for control of the temperature of
the sample.
Figure 3.4: Schematic representation of a dilatometer [29].
The length change of the sample was measured in this work using a Bähr 805 DIL A/D dilatometer.
The length resolution of this machine is 50 nm, and the temperature resolution is 0.05 °C. Heating
rates of up to 4000 K s-1 and cooling speeds of 2500 K s-1 can be achieved.
Solid cylindrical samples with a length of 10 mm and a diameter of 3.5 mm were used, with a type S
thermocouple spot-welded on the surface. A vacuum on the order of 10-4 mbar was used during
heating or isothermal segments. Helium was used as the cooling gas. The original length of the
sample 𝐿0 was measured using a Mitutoyo digital caliper, model 500-181U, with a readout error of
±0.02 mm.
3.2.2
Processing of dilatometry data
In this work, the dilatometer was used to create microstructures and study phase transformations.
Phase transformations are accompanied by a change in volume, but a dilatometer can only measure
length change. The essential assumptions which relate volume change and length change and allow
for the study of phase transformations with a dilatometer, are that the material behaves isotropically
and that |
Δ𝑉
|
𝑉
<<1.
These assumptions give rise to the relationship between length change and volume change as given
by Equation 3.2:
3Δ𝑙
𝑙
where
Δ𝑙
𝑙
β‰…
is the length change of the sample, and
Δ𝑉
,
𝑉
Δ𝑉
𝑉
13
the volume change of the sample.
Equation 3.2
3.2.3
Thermal expansion
To study phase transformations in a dilatometer, the thermal expansion behavior of the material
needs to be modeled. The thermal expansion behavior of QP-G was found to best match the nonlinear methods as proposed by Van Bohemen [31]. The calculated thermal expansion of austenite,
ferrite and cementite are plotted in Figure 3.5a, while the calculated coefficients of thermal
expansion of austenite, ferrite and cementite are plotted in Figure 3.5b.
The non-linear lattice expansion of austenite was described by
Δ𝐿𝛾
𝛾
= 𝐡𝛾 𝑇 + 𝐡𝛾 Θ𝐷 [exp (βˆ’
𝛾
𝐿0
where
Δ𝐿𝛾
𝛾
𝐿0
𝑇
𝛾) βˆ’
Θ𝐷
1]
Equation 3.3
is the relative length change of an austenite lattice, 𝐡𝛾 is the thermal expansion coefficient
𝛾
of austenite, 𝑇 the temperature and Θ𝐷 the Debye temperature of austenite. The thermal expansion
coefficient of austenite is calculated by deriving Equation 3.3 as
𝑑 Δ𝐿𝛾
𝑇
( 𝛾 ) = 𝐡𝛾 [1 + exp (βˆ’ 𝛾 )]
𝑑𝑇 𝐿0
Θ𝐷
𝛼𝛾 =
Equation 3.4
Similarly, the non-linear lattice expansion of ferrite was described as
Δ𝐿𝛼
𝛾
𝐿0
where
Δ𝐿𝛼
𝛾
𝐿0
𝛾
=
𝐿𝛼0 βˆ’ 𝐿0
𝛾
𝐿0
+ 𝐡𝛼 𝑇 + 𝐡𝛼 Ξ˜π›Όπ· [exp (βˆ’
𝑇
) βˆ’ 1]
Ξ˜π›Όπ·
Equation 3.5
𝛾
is the relative length change of a ferrite lattice,
𝐿𝛼
0 βˆ’πΏ0
𝛾
𝐿0
is the relative length change between
the BCC and FCC lattices at 0 K, 𝐡𝛼 is the thermal expansion coefficient of ferrite, and Ξ˜π›Όπ· the Debye
temperature of ferrite. The thermal expansion coefficient of ferrite is derived as
𝛼𝛼 =
𝑑 Δ𝐿𝛼
𝑇
( 𝛾 ) = 𝐡𝛼 [1 + exp (βˆ’ 𝛼 )]
𝑑𝑇 𝐿0
Θ𝐷
Equation 3.6
The thermal expansion behavior of martensite does not differ significantly from ferrite [32], so the
thermal expansion behavior of ferrite has been assumed for martensite.
𝛾
The values are taken as 𝐡𝛾 = 24.8 x 10-6 K-1 and Θ𝐷 = 280 K, 𝐡𝛼 =18.3 x 10-6 K-1, Ξ˜π›Όπ· = 320 K and
𝛾
𝐿𝛼
0 βˆ’πΏ0
𝛾
𝐿0
=
103.9 x 10-4, all as determined in [31].
Analogous to the thermal expansion behavior of ferrite and austenite, the thermal expansion
behavior of cementite has been modeled as described in [32] with
βˆ’1
Δ𝐿𝐹𝑒3 𝐢
𝑇
𝑇 βˆ’ 𝑇𝑐 2
𝐹𝑒3 𝐢
=
𝐡
𝑇
+
𝐡
Θ
(βˆ’
)
βˆ’
1]
+
𝛿
[1
+
(
) ]
[exp
𝐹𝑒3 𝐢
𝐹𝑒3 𝐢 𝐷
𝐹𝑒 𝐢
𝐿𝐹𝑒3 𝐢
𝑇𝑀
Θ 3
Equation 3.7
𝐷
where
Δ𝐿𝐹𝑒3𝐢
𝐿 𝐹𝑒3𝐢
is the relative length change of a cementite lattice, 𝐡𝐹𝑒3 𝐢 the thermal expansion
𝐹𝑒3 𝐢
coefficient of cementite, and Θ𝐷
βˆ’1
π‘‡βˆ’π‘‡π‘ 2
)
is
]
𝑇𝑀
the Debye temperature of cementite. 𝛿 [1 + (
14
a
correction factor to account for the magnetic transition of cementite around the Curie temperature
𝑇𝑐 , which has a significant effect on the thermal expansion coefficient of cementite. The thermal
expansion coefficient of cementite is derived as
𝛼𝐹𝑒3 𝐢
𝑑 Δ𝐿𝐹𝑒3 𝐢
𝑇
2𝛿(𝑇 βˆ’ 𝑇𝑐 )
=
( 𝐹𝑒 𝐢 ) = 𝐡𝐹𝑒3 𝐢 [1 + exp (βˆ’ 𝐹𝑒3 𝐢 )] βˆ’
2
𝑑𝑇 𝐿 3
(𝑇 βˆ’ 𝑇𝑐 )2
Θ𝐷
+ 1] βˆ— 𝑇𝑀2
[
2
𝑇𝑀
Equation 3.8
𝐹𝑒 𝐢
The values are taken as 𝐡𝐹𝑒3 𝐢 =17.2 x 10-6 K-1, Θ𝐷 3 = 440 K, 𝛿 = -15*10-4, 𝑇𝑐 = 512 K and 𝑇𝑀 = 110 K,
after a careful analysis of the data in [32] and a discussion with the author.
Ferrite
Cementite
Coefficient of thermal expansion x 10-6 (K-1)
Calculated thermal expansion L/L0 x 10-2
Austenite
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0
200
400
600
800
Temperature (°C)
Austenite
Ferrite
Cementite
30
25
20
15
10
5
0
0
200
400
600
800
Temperature (°C)
(a)
(b)
Figure 3.5: (a) Relative length changes due to thermal expansion as used in this work. (b) Coefficients of thermal expansion
as used in this work.
3.2.4
Quantification of phase fractions during quenching
3.2.4.1 Lever rule
For the determination of phase fractions, the lever rule can be used. The process is demonstrated
using a martensitic transformation in QP-G steel, which is shown in Figure 3.6. Let phase A be FCC,
Δ𝐿
Δ𝐿
and phase be BCC. Let π‘₯ be = |( 𝐿 ) βˆ’ ( 𝐿 )
0
B (BCC) at temperature 𝑇 and
Δ𝐿
Δ𝐿
0
0
𝑦 be = |( 𝐿 ) βˆ’ ( 𝐿 )
𝐴
𝑒π‘₯𝑝
0
𝐡
Δ𝐿
( 𝐿 ) the
0 𝑒π‘₯𝑝
𝑒π‘₯𝑝
Δ𝐿
| , with ( 𝐿 ) the theoretical length change of Phase
0
𝐡
experimentally measured length change of phase A + B. Let
Δ𝐿
| , with ( 𝐿 ) the theoretical length change of Phase A (FCC) at temperature
0
𝐴
𝑇. The fraction of Phase B (ferrite) at temperature 𝑇 is then
(austenite) at temperature 𝑇 is then
π‘₯
.
π‘₯+𝑦
𝑦
,
π‘₯+𝑦
while the fraction of Phase A
During isothermal transformations the lever rule can also
be applied.
15
Dilatometry data
0.8
Phase A
Phase B
BCC
1.0
FCC
K-M
0.6
x
Phase fraction
Length change Ξ”L/L0 (%)
0.8
0.4
0.2
y
0.0
-0.2
-0.4
0.6
0.4
0.2
-0.6
0.0
50
100
150
200
250
300
350
400
0
Temperature (°C)
50
100
150
200
250
300
350
Temperature (°C)
(a)
(b)
Figure 3.6: (a) Schematic illustration of lever rule. (b) The phase fractions as calculated by the lever rule of Phase A (FCC) and
Phase B (BCC). The fraction of phase B (BCC) is fitted with the Koistinen-Marburger Equation.
3.2.4.2 Koistinen-Marburger equation
The Koistinen-Marburger equation (Equation 2.1) was fitted to data generated during quenching with
50 °C s-1 to room temperature after the same austenization treatment used for Q&P samples, using
the lever rule and the equations and methods proposed by van Bohemen [28], [31] and given in
§3.2.3. The experimentally found fit parameters were π‘‡π‘˜π‘š = 313 °C and Ξ± = 0.0218 K-1, differing less
than 5% and showing good agreement compared to the empirical equations proposed in [28], which
give Ms = 325 °C and Ξ± = 0.0211 K-1. When using the definition that Ms is the temperature where 1%
of martensite has been formed, the experimentally found Ms was 331±3°C. As described by van
Bohemen [28], the discrepancy between the experimentally found Ms and π‘‡π‘˜π‘š found for the fit of
the K-M equation can be explained by an initial gradual transformation during the start of martensite
transformation. A better fit to the experimental data is found using the theoretical martensite start
temperature π‘‡π‘˜π‘š . Typically π‘‡π‘˜π‘š is 5-20 C °C lower than Ms [28].
3.2.5
Detecting phase transformations
Using methodology adapted from [26], the volume effects of the decomposition of austenite and the
precipitation of carbides from martensite can be determined. Apart from length change due to
thermal expansion, a relative increase in length of the sample can be correlated to the
decomposition of retained austenite, 𝛾 β†’ 𝛼 + πœƒ. A relative contraction in length, again apart from
length change due to thermal expansion, can be correlated to the precipitation of carbides from
martensite, 𝛼 β€² β†’ 𝛼 + πœƒ . The relative volume change
β€²
Δ𝑉
𝑉
of the precipitation of carbides from
martensite 𝛼 β†’ 𝛼 + πœƒ can be predicted using
100 βˆ’ 4𝑋
3𝑋
100 βˆ’ 𝑋
πœˆπ›Ό + 12 πœˆπœƒ βˆ’
πœˆπ›Όβ€²
Equation 3.9
2
2
100 βˆ’ 𝑋
πœˆπ›Όβ€²
2
where 𝑋 is the atomic percentage of carbon in the original phase, πœˆπ›Ό the volume of the unit cell for
ferrite, πœˆπœƒ the volume of the unit cell for cementite and πœˆπ›Όβ€² the volume of the unit cell for
Δ𝑉
=
𝑉
martensite. The relative volume change
Δ𝑉
𝑉
of the decomposition of retained austenite
𝛾 β†’ 𝛼 + πœƒ can be predicted using
16
100 βˆ’ 4𝑋
3𝑋
100 βˆ’ 𝑋
πœˆπ›Ό +
𝜈 βˆ’
πœˆπ›Ύ
2
12 πœƒ
4
Equation 3.10
100 βˆ’ 𝑋
πœˆπ›Ύ
4
where πœˆπ›Ύ is the volume of the unit cell for austenite The respective volumes of the unit cells have
been calculated using the data in Table 3.2. The length change can then be calculated by applying
Equation 3.2.
Δ𝑉
=
𝑉
Table 3.2: Crystallographic data as used in this work. Reproduced from [26]
Phase
Structure
Lattice Parameters (Å)
Martensite
BCT
Ferrite
Austenite
ο₯-carbide
BCC
FCC
Hex.
ο₯-carbide
Hex.
-carbide
Orthorhombic
Orthorhombic
a = 2.8664 – 0.013 wt. % C
c = 2.8664 + 0.116 wt. % C
a = 2.8664
a = 3.555 + 0.044 wt. % C
a = 2.752
c = 4.353
a = 2.753
c = 4.335
a = 4.704; b = 4.318
c = 2.830
a = 4.5234; b = 5.0883
c = 6.7426
Cementite (ΞΈ)
Number of
Fe Atoms per
Unit-Cell
2
Volume per Fe Atom
(Å3)
2
4
2
11.78
11.697 (1.1 wt. % C)
14.275
2
14.041
4
14.371
12
12.933
12.188 (1.1 wt. % C)
Since the volume of the unit cell and lattice parameters given in Table 3.2 are valid at room
temperature, the volume of the unit cells has to be corrected to account for thermal expansion
effects at higher temperatures. This has been done using
𝑉𝑒 (𝑇) = V0 [1 + Ο΅(𝑇)]3
Equation 3.11
where 𝑉𝑒 (𝑇) is the volume of a unit cell at temperature 𝑇, V0 the volume of the unit cell at room
temperature and Ο΅(𝑇) the strain for the unit cell calculated at temperature 𝑇 using the non-linear
thermal expansion coefficients as described in § 3.2.3. The corrected volumes of the unit cells are
then used in Equation 3.9 and Equation 3.10.
In Figure 3.7a, the calculated relative length change of decomposition of a 1 wt. % C austenite into
ferrite and cementite has been plotted at different temperatures, while in Figure 3.7b the calculated
relative length change of austenite decomposition into ferrite and cementite has been plotted at
room temperatures for varying carbon content in the austenite. This carbon concentration of the
retained austenite was selected based on experimentally observed carbon concentrations in retained
austenite after Q&P treatments in this work.
Figure 3.8a shows the calculated relative length change of a 0.06 wt. % C martensite precipitating
carbides at different temperatures, while in Figure 3.8b the calculated relative length change of
precipitation of carbides from martensite has been plotted at room temperature for varying carbon
content in the martensite. This carbon concentration of the martensite was selected as the mass
balance of carbon in a sample with the highest observed retained austenite fraction of 0.15 in this
work. The determined carbon content in retained austenite is 1 wt. % C, and a mass balance of
carbon gives a carbon content of 0.06 % wt. C in the martensite.
17
Austenite decomposition
2.0
0.8
1.6
Relative length change (%)
Relative length change (%)
Austenite decomposition
1.0
0.6
0.4
0.2
0.0
1.2
0.8
0.4
0.0
0
200
400
600
0
Temperature (°C)
1
2
Carbon concentration (Wt. %)
(a)
(b)
Figure 3.7: (a) Length change of the decomposition of austenite with 1 wt. % C at different temperatures. (b) Length change
of the decomposition of austenite with different carbon content at room temperature.
Precipitation of cementite from martensite
Precipitation of cementite from martensite
0.00
-0.030
Relative length change (%)
Relative length change (%)
-0.05
-0.035
200
400
-0.15
-0.20
-0.25
-0.30
-0.35
0.0
-0.040
0
-0.10
600
0.1
0.2
0.3
0.4
0.5
Carbon concentration (Wt. %)
Temperature (°C)
(a)
(b)
Figure 3.8: (a) Length change of precipitation of carbides from martensite with 0.06 wt. % C at different temperatures. (b)
Length change of precipitation of carbides from martensite with different carbon content at room temperature.
Having calculated the estimated effects of decomposition of retained austenite and precipitation of
carbides from martensite in a dilatometer, a description of the quantification of these effects is
given.
In earlier literature [26] these quantifications have been performed by an analysis of the length
Δ𝐿
𝑑
Δ𝐿
change 𝐿 . However, an analysis of the derivative 𝑑𝑇 ( 𝐿 ) of the length change allows a clearer
0
0
illustration of the processes occurring. To illustrate this, compare Figure 3.9a, where the length
Δ𝐿
change 𝐿 of a sample during a reheating with 5 °C min-1 is plotted, to Figure 3.9b where the
0
𝑑
Δ𝐿
corresponding derivative 𝑑𝑇 ( 𝐿 ) is plotted. Any length changes, whether due to relative expansion
0
or relative contraction, other than thermal expansion are clearly more readily identified and
β€œmagnified” to the human eye in Figure 3.9b. To quantify the decomposition of retained austenite
and the precipitation of carbides from martensite, in this thesis an analysis of
𝑑 Δ𝐿
( ) will
𝑑𝑇 𝐿0
therefore
be used. This derivative is comparable to the thermal expansion coefficient of the sample.
In this work, the derivative was numerically derived from a 20 data point simple moving average of
both length change and temperature. The number of data points recorded during the reheating was
2 K-1.
18
Derivative of length change
25
0.8
20
d/dT(L/L0) x10-6 (K-1)
Length Change (%)
Length Change
1.0
0.6
0.4
0.2
15
10
5
0
0.0
0
100
200
300
400
500
0
600
100
200
300
400
500
600
Temperature (°C)
Temperature (°C)
(a)
(b)
πœŸπ‘³
Figure 3.9: (a) Experimentally measured length change of a sample during reheating.
π‘³πŸŽ
(b) Corresponding derivative
𝒅
𝒅𝑻
πœŸπ‘³
( ) of (a).
π‘³πŸŽ
As schematically illustrated in Figure 3.10, a relative increase in length, e.g. due to decomposition of
retained austenite, will show in a graphical plot of the derivative as a peak. Likewise, a relative
decrease in length, e.g. due to precipitation of carbides from martensite, will show up as a negative
peak. The area of the peak in the derivative curve, measured relative to a baseline which is defined
here as the derivative curve as if there was no length change, is the length change due to processes
other than thermal expansion. This is expressed as
𝑇2
𝑇2
Δ𝐿
( ) 𝑝 = ∫ 𝛼𝑒π‘₯𝑝 𝑑𝑇 βˆ’ ∫ 𝛼𝑏 𝑑𝑇
𝐿0
𝑇1
Equation 3.12
𝑇1
Δ𝐿
𝐿0
where ( ) 𝑝 is the length change due to decomposition of retained austenite or precipitation of
carbides from martensite, 𝑇1 the temperature at which these transformations start, 𝑇2 the
temperature at which these transformations end, 𝛼𝑒π‘₯𝑝 the experimental derivative
the derivative
𝑑 Δ𝐿
( ) as
𝑑𝑇 𝐿0
if there was no decomposition or precipitation.
19
𝑑 Δ𝐿
( ) and 𝛼𝑏
𝑑𝑇 𝐿0
Figure 3.10: Schematic illustration of relative length increases and relative length decreases as they are seen in curves of
derivatives of the length change in a dilatometer. Also shown is the schematic location of inflection points used to determine
activation energies.
In general, the baseline in this work was taken as a straight line between π›Όπ‘π‘’π‘Ÿπ‘£π‘’ at 𝑇1 and π›Όπ‘π‘’π‘Ÿπ‘£π‘’ at
𝑇2 . The error in calculating the area of the peak is estimated to be around 10%-15% of peak area.
This error is due to inherent variability in (manually) determining 𝑇1 and 𝑇2 , the resulting differences
in the baseline, and the uncertainty in estimating the value of the baseline after the process.
3.2.6
Determination of activation energies
The activation energies of physical processes such as retained austenite decomposition and
precipitation of carbides from martensite can be determined using a Kissinger type analysis [33]–[36]
based on dilatometry data. This analysis enables the determination of activation energies based on a
method of finding the temperature where an inflection points occurs in a dilatometry curve as a
function of heating rate [36]. When neglecting residuals from the derivation leading to Equation 3.13
[36] the general analysis is based on
ln
𝑇𝑖2
𝐸
𝐸
=
+ ln (
)
Ξ¦ 𝑅𝑇𝑖
𝑅𝐾0
Equation 3.13
where 𝑇𝑖 is the temperature of an inflection point, Ξ¦ the heating rate in K/min, 𝐸 the activation
energy, 𝑅 the gas constant and 𝐾0 a pre-exponential factor[34]. The temperature 𝑇𝑖 is taken as the
temperature at which the derivative of the length change versus temperature shows a maximum or
minimum, c.f. Figure 3.10. These maxima and minima occur at the temperature where the reaction
𝑇2
1
rate is highest [37]. Plotting ln Φ𝑖 vs. 𝑇 in a Kissinger plot for multiple values of Ξ¦ allows applying a
𝑖
20
𝐸
linear fit. The slope of this fit is 𝑅, as can be seen in Figure 3.11. From this slope the activation energy
can then be determined. In this work, the maxima and minima were determined using a Gaussian fit
around the peak position. The position of the center of the fit was taken as the inflection point.
Due to a limited amount of data in this work (3 different heating rates, 1 experiment per heating
rate) and inherent variability in the determination of the inflection points, the error of determining 𝐸
is estimated to be about 25% of the value of 𝐸. In literature, lower errors are calculated due to more
heating rates being applied [23].
Compared to other dilatometric studies with tempering or annealing on Q&P steel grades in
literature ([11]: figure 3, [14]: figure 7), fewer inflection points are found in QP-G after the
treatments specific to this thesis than in literature. Other inflection points not found in this work are
related in these studies to carbon clustering and segregation, precipitation of transition carbides ο₯
and  [26][11], and possibly precipitation of alloy carbides [14]. The absence of these inflections
points is possibly related to the averaging used to calculate the derivative. Another possibility is that
the 200 °C higher partitioning temperatures and longer partitioning times used in this work
compared to [14] prevent these processes from occurring.
Figure 3.11: Schematic Kissinger-type plot. The slope of the linear fit to the plotted inflection points is E/R.
3.3 Magnetic measurements
3.3.1
Vibrating Sample Magnetometer (VSM)
A Vibrating Sample Magnetometer [38], [39] or VSM is a device that can measure the magnetic
properties of a sample. In this work a VSM is used to determine the retained austenite fraction from
the magnetic properties. The VSM works by moving a sample inside a uniform applied magnetic field.
The voltage induced by this movement can be detected by pickup coils and is proportional to the
magnetization (see § 3.3.2) of the sample. A schematic overview of a VSM is presented in Figure
3.12a, while Figure 3.12b shows the VSM setup used in this work.
21
Sample
holder
Electromagnet
Pickup
Coils
(a)
(b)
Figure 3.12: (a) Schematic overview of a VSM [40]. (b) LakeShore 7307 series VSM as used in this work. As a reference for
scale, the white sample holder is about 6 cm long.
A LakeShore 7307 series VSM was used to measure the saturation magnetization (see § 3.3.2). The
magnetization during annealing of the sample was measured with the same VSM with a LakeShore
oven type 73034 installed. The software controlling the VSM was LakeShore IDEAS-VSM 3.4.0
software. The samples were disk-shaped with a diameter of 3.5 mm and a height of approximately 2
mm, and measured in plane with respect to the magnetic field. A VSM is sensitive to the positioning
of the sample within the applied magnetic field [41]–[43]. To ensure repeatable positioning within
the applied magnetic field, a specially made sample holder for disc specimens was therefore used.
3.3.2
Background on magnetization
For a detailed background on magnetization, the reader is referred to books by Jiles [44] or Callister
[45].
Moving electric charges result in magnetism. The magnetic moment of an atom π’Ž is the sum of the
magnetic moments of the electrons, which at the atomic level are the moving charges. The
βƒ—βƒ—βƒ— is the vector sum of all atom magnetic moments, or the total
magnetization of a material 𝑴
magnetic moment divided by its volume:
⃗𝑴
βƒ—βƒ— = βˆ‘
π’Š
π’Žπ‘–
𝑉
Equation 3.14
The saturation magnetization 𝑀𝑠 is the value of maximum magnetization in a material, which occurs
when a β€œlarge enough” field is applied. The material only has one domain remaining in which all
magnetic moments are aligned parallel to the magnetic field.
Any difference in saturation magnetization between a sample containing some retained austenite
and a sample containing no retained austenite is directly related to the volume fraction of retained
austenite, when neglecting small volume fractions of cementite. This is because ferrite and
martensite are ferromagnetic, while austenite is paramagnetic [46]. The saturation magnetization for
22
pure ferromagnetic iron in this work is taken as 218 Am2 kg-1 at room temperature [47], while for
pure paramagnetic iron at room temperature the magnetization is taken as <0.5 Am2 kg-1 ([47] by
graphical extrapolation of figure 1, field of 1.5 T). The difference in saturation magnetization
between ferromagnetic and paramagnetic phases is large enough that the influence of the saturation
magnetization of austenite on the saturation magnetization of the sample can be neglected.
3.3.3
Influence of temperature on saturation magnetization
The Curie temperature 𝑇𝑐 is defined here as the temperature below which a material is
ferromagnetic, while above 𝑇𝑐 the material is paramagnetic. This is caused by thermal precession of
the individual atomic moments about the field direction. When the temperature increases, the
thermal precession also increases, which causes the measured (spontaneous) magnetization to be
smaller than the saturation magnetization. When 𝑇𝑐 is reached, this thermal precession is strong
enough to overcome the magnetic moment coupling associated with ferromagnetic phases, and the
material loses the coupling and the magnetic moments become randomly ordered, i.e. paramagnetic.
The dependence on temperature of the magnetization of iron for ferromagnetic phases can be
described by an equation proposed by Arrott and Heinrich [48], and given in a modified form used by
Bojack [49]:
𝑀𝑠 (𝑠) = 𝑀0
(1 βˆ’ 𝑠)𝛽
3
7
(1 βˆ’ 𝛽𝑠 + 𝐴𝑠 2 βˆ’ 𝐢𝑠 2 )
Equation 3.15
𝑇
𝑇𝑐
where 𝑀𝑠 (𝑠) is the saturation magnetization at a dimensionless temperature 𝑠 = , 𝑀0 is the
saturation magnetization at 0 K, and 𝛽, 𝐴 and 𝐢 are material dependent constants.
3.3.4
Experimental procedures
The procedure to determine the retained austenite content in steels with magnetic measurements
was derived from the method described in [46]. A standard NIST nickel specimen was used to
calibrate the VSM. Magnetization curves were made with a stepwise applied magnetic field from 1.7
T to -1.7 T as can be seen in Figure 3.13. The step size was 0.1 T. The value of interest, saturation
magnetization 𝑀𝑠 , was taken as the average of the absolute magnetization values at the maximum
and minimum (i.e. 1.7 T and -1.7 T) applied magnetic field. The unit of 𝑀𝑠 in this work is Am2 kg-1,
which is the total magnetic moment as measured by the VSM, divided by the mass of the sample.
βƒ—βƒ—βƒ— is defined as the total magnetic moment
Compare this to Equation 3.14, where the magnetization 𝑴
divided by the volume. With the practical facilities available, accurate measurement of mass is more
easily achieved than accurate measurement of volume, and since volume and mass are directly
related by density, the use of Am2 kg-1 is allowed as unit for magnetization.
Figure 3.13: VSM program used for determining the saturation magnetization. The values for ±1.7 T are nominal, in practice
±1.62 T was the maximum reachable applied magnetic field.
23
The volume fraction retained austenite of each sample 𝑓𝛾 was then determined by a comparison of
the average 𝑀𝑠 value of at least 3 measurements in the austenite containing sample to a directly
quenched specimen which was confirmed by XRD to have almost no (<1%) retained austenite:
𝑓𝛾 = 1 βˆ’
𝑀𝑠 (𝑐)
𝑀𝑠 (π‘Ÿ)
Equation 3.16
where 𝑀𝑠 (𝑐) is the saturation magnetization of the austenite containing sample and 𝑀𝑠 (π‘Ÿ) is the
saturation magnetization of the austenite free quenched sample. The standard deviation in
determining the fraction of retained austenite by this method is <1% retained austenite. The
standard deviation of the fraction of retained austenite was determined by
2
1 2 2
𝑀𝑠 (𝑐)
2
√
𝜎= (
) πœŽπ‘€π‘  (𝑐) + (
) πœŽπ‘€
𝑠 (π‘Ÿ)
𝑀𝑠 (π‘Ÿ)
𝑀𝑠 (π‘Ÿ)2
Equation 3.17
where πœŽπ‘€π‘  (𝑐) is the standard deviation of measurements of 𝑀𝑠 (𝑐), and πœŽπ‘€π‘  (π‘Ÿ) the standard deviation
of measurements of 𝑀𝑠 (π‘Ÿ).
3.3.4.1 Measuring volume fractions of retained austenite during isochronal annealing
To measure the amount of retained austenite during a heating, the following procedure as
schematically shown in Figure 3.14 was performed in the VSM:
1) First heating with 5 °C min-1 to 600 °C under constant applied magnetic field of 1.5 T
2) Isothermal holding at 600 °C for 20 minutes
3) A cool down to room temperature by natural cooling of the VSM oven (see [49] for a
description of the cooling rate)
4) Second heating with 5 °C min-1 to 600 °C under constant applied magnetic field of 1.5 T
This procedure is partly modelled after [17]. Some fraction of retained austenite decomposes during
the first heating to 600 °C. More retained austenite decomposes during the isothermal holding and
the cooling down to room temperature. If the microstructure is assumed to be fully ferritic after the
cooling (i.e. fully ferromagnetic) the second heating occurs without any phase transformations. This
allows the calculation of the dependence of the 𝑀𝑠 of ferrite on temperature in this material using
Equation 3.15.
The fit of the second heating with Equation 3.15 allows the establishment of a multiplication factor
to calculate the equivalent room temperature magnetization of the sample from the magnetization
values recorded during the reheating. This is done by calculating the saturation magnetization at
room temperature according the fit to Equation 3.15, and dividing the fitted saturation
magnetization at each recorded temperature by this number:
𝑀𝐹(𝑇) =
𝑀𝑓𝑠 (π‘‡π‘Ÿ )
𝑀𝑓𝑠 (𝑇)
Equation 3.18
where 𝑀𝐹 is the multiplication factor at each measured temperature 𝑇, 𝑀𝑓𝑠 (π‘‡π‘Ÿ ) is the saturation
magnetization at room temperature according to the fit and 𝑀𝑓𝑠 (𝑇) is the saturation magnetization
at each measured temperature according to the fit.
24
Multiplying the recorded magnetization values during the first heating by the corresponding
multiplication factor then gives the equivalent room temperature magnetization at each
temperature. This allows a direct calculation of the retained austenite fraction using Equation 3.16.
In Figure 3.14a, a plot of the magnetization curves of such an experiment is shown, and a fit to the
second heating using Equation 3.15 fits well (R2 = 0,9999). This indicates that no further phase
transformations are occurring. The decomposition of retained austenite can be deduced from the
changes in slope of the first heating, very clearly seen in the temperature range 300 °C to 400 °C.
In Figure 3.14b, the resulting corrected equivalent room temperature magnetization of the same
experiment is shown. The increase in magnetization clearly shows retained austenite decomposition.
First Heating
Second Heating
Fit to Eq. 16
180
Mfs(T1)
160
Mfs(T2)
MF(T1) = Mfs(Tr) / Mfs(T1)
140
Corrected equivalent RT magnetization
200
Mfs(Tr)
Magnetization (Am2/kg)
Magnetization (Am2/kg)
200
First Heating
MF(T2) = Mfs(Tr) / Mfs(T2)
120
180
x MF(T1)
x MF(T2)
160
140
120
0
100
200
300
400
500
600
0
Temperature (°C)
100
200
300
400
500
600
Temperature (°C)
(a)
(b)
Figure 3.14: (a) Results of a typical annealing experiment in the VSM. For the second heating, only 1 out of every 4 data
points is shown for better visibility of the fit. (b) Corrected equivalent RT magnetization.
3.4 X-ray Diffraction
X-Ray Diffraction experiments were used to determine the volume fraction of retained austenite and
determine the lattice parameter of retained austenite at room temperature. A Bruker D8 Advance
Diffractometer equipped with a Vantec position sensitive detector was used, using Co KΞ±1 radiation
( = 1.78897 Å), an acceleration voltage of 45 kV and current of 35 mA, while the sample was
spinning at 30 rpm. The measurements were performed in the 2ΞΈ range of 40°-130°, using a step size
of 0,035° 2ΞΈ, with a counting time per step of 4 s. The volume fraction of retained austenite and the
errors in determing the retained austenite fraction were calculated using the Jatczak model [50] as
described in [51], [52].
The lattice parameter of retained austenite was determined by stripping the Co KΞ±2 intensity, fitting
the peaks with a pseudo Voigt function, correcting the peak position for sample displacement and
goniometer errors which were determined with LaB6 calibration measurements, and from the peak
position of the {111}, {200}, {220} and {300} peaks the lattice parameter π‘Ž was calculated using
π‘Ž = π‘‘βˆšβ„Ž2 + π‘˜ 2 + 𝑙 2
Equation 3.19
πœ†
Equation 3.20
2π‘ π‘–π‘›πœƒ
where 𝑑 is the lattice spacing of the diffracting {β„Žπ‘˜π‘™} plane, πœ† the wavelength used and πœƒ the
measured 2πœƒ peak position divided by two. The lattice parameter used in this work is the average of
the 4 lattice parameters found, and the error the standard deviation of those 4 lattice parameters.
𝑑=
25
Using the lattice parameter, the amount of carbon in retained austenite was calculated by Equation
3.21:
π‘Ž = 0.355 π‘›π‘š + 0.044
π‘›π‘š
𝛾
× π‘‹π‘
𝑀𝑑. %
Equation 3.21
𝛾
where π‘Ž is the lattice parameter in nm and 𝑋𝑐 the weight percentage of carbon in retained
austenite. The error in determining the carbon percentage is estimated as 0.05 wt. % C
3.5 Microscopy
3.5.1
Light Optical microscopy
Due to its limited resolution, Light Optical Microscopy (LOM) is not the best technique to analyse the
fine martensitic microstructures. Larger microstructural features however, are best seen with LOM.
In this work, microstructural banding was analysed with the LOM (e.g. Figure 3.1).
Light Optical Microscopy was performed using a Keyence VHX-5000 series digital microscope. The
lenses used were the VH-Z100R (100x-1000x) and the VH-Z250R (250x-2500x).
3.5.2
Scanning Electron Microscopy
Scanning Electron Microscopy (SEM) was used to resolve and image the fine martensitic
microstructures. Furthermore, EBSD measurements (see §3.5.3) were performed in a SEM.
Scanning Electron Microscopy was performed using a JEOL JSM-6500F series field emission gun SEM,
using a Secondary Electron Imaging detector. The acceleration voltage was 15 kV and the nominal
working distance was 10 mm.
3.5.3
Electron Backscatter Diffraction
(b)
(a)
Figure 3.15: (a) Working principles of EBSD [53]. (b) Example of an EBSP made in austenitic steel [54].
In this work, EBSD is used to determine the morphology of austenite and ferrite grains and their
positions. Figure 3.15a shows a schematic of the working principles of EBSD. In the SEM, the incident
electron beam is held stationary at a certain position. Primary electrons which satisfy Bragg’s law
given in Equation 3.20 can backscatter. These electrons are then recorded on a phosphor screen. The
resulting image is called an Electron Back-Scattered Pattern (EBSP) or Kikuchi pattern, and an
example is shown in Figure 3.15b. The visible lines are called Kikuchi bands. In steel, FCC and BCC
phases can be distinguished because of their differing characteristic Kikuchi patterns. Identification of
26
the phases and determination of their crystallographic orientation is performed by comparing the
recorded Kikuchi patterns to reference patterns. For a detailed background on Electron Backscatter
Diffraction (EBSD) and the underlying principles, the reader is referred to [53].
EBSD measurements were performed in a JEOL 6500F series SEM, using an Oxford Instruments
Nordlys II detector. The acceleration voltage was 20 kV, the beam current 1.2 nA, the working
distance 25 mm and the tilt angle 70°. 4x4 binning of the detector was used. A square grid of
300x300 pixels was scanned with a step size of 50 nm, resulting in a scanned area of 15x15 µm2.
Acquisition and post processing of Kikuchi patterns was performed with Oxford Instruments Channel
5 software.
A different setting was used for a single experiment, sample QT260H425. The acceleration voltage
was 15 kV. A square grid of 360x360 pixels was scanned with a step size of 30 nm, resulting in a
scanned area of 10.8x10.8 µm2.
The post-processing involved the phase identification of the created EBSPs. Based on previously
performed calibration measurements in this setup, no misidentification is expected using 9 Kikuchi
bands. Phase identification was therefore carried out using 9 Kikuchi bands. After the phase
identification, cleaning of the resulting maps was performed. Pixels that were unable to be identified
due to insufficient band contrast in the EBSP were identified by automated iterative nearest
identified neighbor filling.
3.5.3.1 Limitations of EBSD
In this work, three problems were encountered which influenced the analysis of EBSD maps. First,
retained austenite films in Q&P steels are expected to be 20 to 100 nm in size [55]. The employed
step sizes of 30 and 50 nm are on the same order of magnitude. This means that film like retained
austenite will be in regions of low band contrast close to grain boundaries, and therefore not easily
identified. An example of a region with low band contrast is indicated in Figure 3.16c. The lack of
indexing will lead to an underestimation of the retained austenite fraction as determined by EBSD. In
some specimens in this work, <0.1 % retained austenite was detected by EBSD while 9 % retained
austenite was determined with magnetic methods.
Second, automated iterative nearest identified neighbor filling can possibly lead to an incorrect
determination of the size and morphology of retained austenite. Figure 3.16a shows a cleaned EBSD
map containing seemingly large retained austenite grains. Figure 3.16b shows the same EBSD map,
but before cleaning. Two RA grains are marked, showing the effect of the cleanup process. The RA
grains are clearly larger after filling, and filling might therefore influence the characterization of RA.
Third, Figure 3.16c and Figure 3.16d show a region of low band contrast filled in as retained austenite
in the marked grain. Regions of low band contrast are likely to be M2 [56]. This region is therefore
incorrectly identified as retained austenite due to the filling procedure.
The combination of these effects means that significant care has to be taken in analyzing
automatically cleaned EBSD maps. In this work, cleaned EBSD maps will therefore be accompanied by
the corresponding uncleaned versions. While the problem of non-indexed pixels is an experimental
limitation and will remain, the effect of the cleaning procedure can be determined by analyzing both
cleaned and uncleaned EBSD maps.
27
Incorrect determination of RA size
and morphology due to filling
Incorrect
identification
due to filling
(b)
(a)
Limited indexing due
to low band contrast
Incorrect
identification
due to filling
(c)
(d)
Figure 3.16: Example of EBSD maps. (a) Combined phase and band contrast map. Blue is identified as FCC, red identified as
BCC. This map has been cleaned. (b) Phase map. Blue is identified as FCC, red identified as BCC. This map has not been
cleaned. White regions are non-indexed regions. (c) Combined phase and band contrast map. Blue is identified as FCC, red
identified as BCC. This map has not been cleaned. (d) Band contrast map.
Of further note is that in some experiments, drift of the electron beam was observed. While the EBSD
maps are shown square, in reality they could be distorted. An example of drift distortion is shown in
Figure 3.17. The approximate boundary of the actual scanned region is marked with solid lines, while
the region supposed to be scanned is marked with dotted lines.
Figure 3.17: Drift of the electron beam.
28
Chapter 4 Results and Discussion
4.1 Microstructures after Quenching & Partitioning
As mentioned in § 3.1.2, Q&P treatments with quenching temperatures varying from 140°C to 340 °C
were performed on QP-G. The resulting microstructures are quantified in § 4.1.1 and § 4.1.2. The
microstructure of a representative selection of samples is characterized in § 4.1.3.
4.1.1
Quantification of phases
Since the aim of this work is to investigate the thermal stability of retained austenite, knowledge of
the fraction of retained austenite fΞ³ in the sample is essential. Using a combination of magnetic
methods and XRD1, the volume fraction of retained austenite was determined and is shown in Figure
4.1a. The maximum fraction of retained austenite was 0.15 and was found in the sample with a
Quenching Temperature of 260 °C, QT260. The fraction of retained austenite as a function of
quenching temperature follows the general prediction made by Speer et al. [9].
0.18
1.0
0.15
0.8
Phase fraction
0.12
f
0.09
0.06
M2
B
RA
M1
0.4
0.2
0.03
0.00
120
0.6
0.0
160
200
240
280
320
360
140 160 180 200 220 240 260 280 300 320 340
Quenching Temperature (°C)
Quenching Temperature (°C)
Ms
(a)
(b)
Figure 4.1: (a) fΞ³ in Q&P specimens as determined by magnetic methods and XRD. (b) Measured phase fractions of retained
austenite and bainite, fractions of M1 as determined with the K-M equation and fractions of M2 as determined by a
microstructural balance using 𝑓𝑀1 + 𝑓𝑅𝐴 + 𝑓𝐡 + 𝑓𝑀2 = 1.
The volume fractions of M1 (Martensite from first quench) were determined using the KoistinenMarburger equation with the parameters as given in §3.2.4.2. The phase fractions are shown in
Figure 4.1b. These estimations were compared with phase fractions determined from dilatometry
data. The agreement was found to be within 3%, except for sample QT320. A QT of 320 °C is above
the theoretical martensite start temperature π‘‡π‘˜π‘š of 313.1 °C used in the K-M equation. The fraction
of M1 in a sample with a quenching temperature of 320 °C is therefore not predictable using the K-M
equation. The phase fraction of M1 in sample QT320 as determined from dilatometry data is shown
in Figure 4.1b instead.
In some specimens, bainite formation is detected during partitioning. Figure 4.2 shows the length
change during partitioning. The maximum length change of 0.07% is seen in sample QT300. This can
be correlated to a maximum bainite fraction of about 0.1. Samples with a QT higher than 240 °C
show some length change, with the exception of sample QT340, which has a QT above Ms. Samples
with a QT lower than 240 °C show very little length change. Taken together, this suggests that bainite
formation is occurring due to martensite nuclei promoting the bainite reaction [57]. AT QT’s lower
1
See Appendix A for details on this calculation and a comparison of RA fractions determined by VSM and XRD.
29
than 240 °C, partitioning of carbon from martensite will enrich the retained austenite with carbon so
no bainite formation can occur. Sample QT340 has a quenching temperature above Ms. No
martensite nuclei will be present in sample QT340 to promote bainite formation, and hence bainite
formation will be limited compared to samples with martensite nuclei.
Volume fractions of M2 (Martensite from second quench) were calculated by balancing the volume
fractions of the microstructural constituents using 𝑓𝑀1 + 𝑓𝑅𝐴 + 𝑓𝐡 + 𝑓𝑀2 = 1.
0.08
0.07
Length change (%)
0.06
Length change (%)
0.007
0.05
0.006
0.005
QT340
QT320
QT300
QT280
QT260
QT240
QT220
QT200
QT180
QT160
QT140
0.004
0.003
0.002
0.001
0.000
40
42
44
46
48
50
Time during partitioning (s)
0.04
0.03
0.02
0.01
See inset
0.00
0
10
20
30
40
50
Time during partitioning (s)
Figure 4.2: Length change (%) during partitioning. The inset is shown for more clarity, as some dilatometry curves are hard
to visually separate.
4.1.2
Lattice parameter of retained austenite
The lattice parameter and carbon content of retained austenite as determined by XRD is shown in
Figure 4.3a. The lattice parameter for sample QT340 could not be determined, because the retained
austenite fraction is below the detection limit of XRD. The smallest lattice parameter is found in
samples QT240 and QT260. Samples QT140-QT220 show a trend of increasing lattice parameter at
decreasing quenching temperatures. This can be correlated with the fraction of M1 formed during
the quench. More martensite formed means more carbon is available to partition into austenite,
thereby increasing the lattice parameter.
Changing the quenching temperature between 280 °C and 320 °C seems to slightly increase the
lattice parameter at increasing quenching temperatures. This is regarded as an effect of bainite
formation during partitioning.
30
Figure 4.3b shows the carbon concentration in retained austenite plotted against the retained
austenite fraction. The general trend is that decreasing amounts of retained austenite contain
increased carbon concentrations.
Increased Bainite
formation
3.608
1.2
3.604
1.1
3.600
1.0
3.596
120
160
200
240
280
320
Carbon concentration in RA (%)
Lattice parameter (Å)
Increased Carbon
partitioning
Carbon concentration in RA (Wt. %)
3.612
1.2
1.1
1.0
0.00
360
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
Phase fraction retained austenite
Quenching Temperature (°C)
(a)
(b)
Figure 4.3: (a) Lattice parameter of retained austenite and carbon content in retained austenite of Q&P specimens.
(b) Carbon concentration in retained austenite plotted against the fraction of retained austenite.
4.1.3
Characterization of microstructure
The microstructures of samples QT160, QT260 and QT320 were characterized in further detail. This
characterization was performed with SEM, OM, EBSD and XRD. Based on SEM micrographs, retained
austenite fractions and lattice parameters, the microstructures of the selected samples are thought
representative for the whole range of Q&P treatments made in this work. The basis of this selection
is discussed in further detail in §4.2.1.
4.1.3.1 Analysis of sample QT160
According to the analysis in § 4.1.1, the microstructure of sample QT160 consists of about 4% RA and
96% M1. Figure 4.4a and Figure 4.4b show uncleaned and cleaned versions respectively of EBSD
phase maps of sample QT160. Cleaning did not significantly affect the resulting retained austenite
grain size and morphology. Approximately 1.5% RA is detected in the EBSD. The detected retained
austenite grains are mostly retained austenite grains with a width > 100 nm and a maximum length
of approximately 1 µm. Film-like retained austenite in Q&P steels has a film width of 20-100 nm [58].
This film width is in the same order of magnitude as the step size of 50 nm of the EBSD
measurement. Film-like RA is therefore unlikely to be resolved reliably, due to insufficient band
contrast. Some scattered single pixels of RA are seen, and these are most likely part of film-like
retained austenite.
Figure 4.4c and Figure 4.4d show SEM micrographs of this sample, at 2000x and 10000x
magnification respectively. Figure 4.4c shows an overview of the microstructure, which is consistent
with tempered martensite. In Figure 4.4d, some carbides can be seen in the tempered martensite.
Furthermore, some retained austenite grains with a width of approximately 100 to 200 nm and
length of 1 µm are marked in Figure 4.4d. Film-like retained austenite grains are not detected in
these figures. It is assumed that the morphology of the RA is about 50% film, and 50% RA grains with
a width > 100 nm and a maximum length of approximately 1 µm.
31
(a)
(b)
(c)
(d)
Figure 4.4: Microstructure of sample QT160. (a) Phase map before cleaning. BCC is shown in red, FCC in blue. (b) Combined
EBSD band contrast and phase map. Cleaned version. (c) (d) SEM micrograph.
4.1.3.2 Analysis of sample QT260
According to the analysis in § 4.1.1, the microstructure of sample QT260 consists of about 15% RA,
70% M1, 3% bainite and 12% M2. Figure 4.5a and Figure 4.5b show SEM micrographs of this sample,
at 2000x and 10000x magnification respectively. Figure 4.5a clearly shows well-etched and lessetched regions. The less-etched regions correspond to RA and M2 [27]. However, retained austenite
etches much less than M2 while retained austenite remains featureless. This allows visual separation
of RA and M2. The feature marked β€œΞ³ (+M2)” in Figure 4.5b consists of M2 in its center and RA at its
left and upper edges. This is consistent with a blocky type austenite grain in which the center was not
sufficiently enriched with carbon to stabilize against the martensitic transformation. Features like this
are also predicted with phase field models [59], and shown in more detail elsewhere in this work (c.f.
Figure 4.18b).
Bainite is hard to distinguish from martensite in SEM images, and therefore no identification of the
small bainite fraction was attempted.
A smaller fraction of carbides compared to sample QT160 can be seen in Figure 4.5b, consistent with
a higher fraction of carbon in solid solution in retained austenite.
32
(a)
(b)
(c)
(d)
Figure 4.5: Microstructure of sample QT260. (a) (b) SEM micrograph (c) Phase map before cleaning. BCC is shown in red, FCC
in blue. (d) Combined EBSD band contrast and phase map. Cleaned version of (c).
Figure 4.5c and Figure 4.5d show cleaned and uncleaned versions respectively of EBSD phase maps of
sample QT260. Some RA grains are significantly inflated due to the cleaning process, as shown in
Figure 3.16. About 2.5% of RA was detected with EBSD. This means a large fraction of RA was
undetectable with EBSD, suggesting a small RA grain size and film-like morphology. A large number of
regions with low band contrast exist in this EBSD measurement. These regions likely contain the filmlike RA grains.
4.1.3.3 Analysis of sample QT320
According to the analysis in § 4.1.1, the microstructure of sample QT320 consists of about 15% M1
and bainite, approximately 4.5% RA and 80% M2. Figure 4.6a Figure 4.6b show EBSD phase maps of
sample QT320. Cleaning did not significantly affect size and morphology of the detected RA grains.
About 0.1% RA was detected with EBSD. This suggests that a large majority of the RA has a film-like
morphology. SEM micrographs, shown in Figure 4.6c and Figure 4.6d, show some features that might
be identified as film-like RA. Blocky type RA is not observed in any available SEM pictures of this
sample.
33
(d)
(a)
(b)
(c)
Figure 4.6: Microstructure of sample QT320. (a) Phase map before cleaning. BCC is shown in red, FCC in blue. (b) Combined
EBSD band contrast and phase map. (c) (d) SEM micrograph
An overview of the microstructure at 2000x magnification is shown in Figure 4.6c. Regions of
tempered martensite (M1) or bainite, and regions of untempered martensite M2 are visible. Carbides
are only present in the tempered regions.
34
4.1.3.4 X-ray Diffraction results
QT320
QT260
QT160
60000
Offset Y values (counts)
12000
Offset Y values (counts)
50000
40000
30000
10000
8000
6000
4000
2000
50
52
54
56
58
60
62
64
2Theta
20000
{111}
{200}
{220}
{311}
10000
0
40
50
60
70
80
90
100
110
120
130
2Theta
Figure 4.7: XRD spectrum of samples QT160, QT260 and QT320. The applied offset values are +4000 for sample QT260 and
+8000 for sample QT320. The inset shows the {111} and {200} peaks of austenite in further detail.
Figure 4.7 shows the XRD spectrum of samples QT160, QT260 and QT320. The retained austenite
peaks in sample QT260 are stronger relative to the other samples. This confirms that the largest
fraction of RA is present in sample QT260.
Some texture might be present in all these materials. The {200} peak of austenite and the {110} peak
of ferrite are stronger than expected. This could indicate the Pitsch orientation relationship:
{010}Ξ³//{101}Ξ± <101>Ξ³//<111>Ξ± [60][61].
4.1.4
Summary: The microstructure of Q&P samples in this work
The microstructural constituents of samples with a QT of 160, 260 and 320 °C are summarized in
Table 4.1.
35
Table 4.1: Summary of the microstructural constituents of samples analyzed in §4.1.3.
Sample
QT160
QT260
QT320
Volume fraction of
retained austenite 𝑓𝛾
0.04 ±0.006
0.15 ±0.006
0.045 ±0.006
Carbon content in
retained austenite
(wt. %)
1.15 ± 0.05
1.02 ± 0.05
1.07 ± 0.05
Morphology of
retained austenite
50%/50% film-like RA
and RA with width
>100 nm, maximum
length of 1 µm
Majority (>80%) of
film-like RA, some
larger grains >100 nm
Large majority (>95%)
of film-like RA
Volume fraction of
martensite from the
first quench 𝑓𝑀1
0.96 ± 0.02
0.7 ± 0.03
0.07 ± 0.03
Volume fraction of
bainite 𝑓𝐡
0
0.03 ± 0.01
0.08 ± 0.02
Volume fraction of
martensite from the
second quench 𝑓𝑀2
0
0.12 ± 0.04
0.8 ± 0.05
Carbides
Yes, located in
tempered martensite
M1
Yes, fewer than QT160,
located in tempered
martensite M1
Yes, fewer than QT160
located in M1, none in
M2
36
4.2 Retained austenite decomposition: reheating with 5 °C s-1 to 700 °C
An initial analysis of the thermal stability of retained austenite was performed by heating Q&P
treated samples to 700 °C as shown in Figure 3.2b. The reheating rate was 5 °C s-1. Upon reaching 700
°C, the samples were directly quenched to room temperature. An analysis of the resulting
microstructure was performed, and this analysis is discussed in §4.2.1 and §4.2.2. A comparison
between the Q&P microstructures before reheating (as analyzed in §4.1) and after reheating to 700
°C will be made in §4.2.3.
4.2.1
Quantification of phases
Figure 4.8 shows the fraction of retained austenite as determined by magnetic methods and XRD2.
Samples with a QT between 140 and 180 °C and between 320 and 340 °C show an increase in the
fraction of retained austenite fΞ³ after reheating. Samples with a QT between 200 and 300 °C show a
decrease in the fraction of retained austenite after reheating. Based on this observation, the samples
were divided into three groups. Region I contains samples with a QT between 140 and 180 °C, region
II contains samples with a QT between 200 and 300 °C and region III contains samples with a QT
between 320 and 340 °C.
Q&P
0.18
Q&P reheated 700 °C
I
II
III
0.15
0.12
f
0.09
0.06
0.03
0.00
120
160
200
240
280
320
360
Quenching Temperature (°C)
Figure 4.8: Fraction of RA as determined by VSM/XRD. For the samples in region I and III, a higher fraction of RA was
detected after the reheating, while in region II a lower fraction of RA was observed.
The lattice parameter of retained austenite as determined by XRD is shown in red in Figure 4.9a.
Shown in black as a comparison are the corresponding samples without reheating. A decrease in the
lattice parameter after reheating is seen in samples with a QT between 140 and 220 °C. For samples
with a QT between 240 and 320 °C, an increase in lattice parameter after reheating is observed.
The lattice parameter and carbon content of retained austenite as determined by XRD is shown in
Figure 4.9a. The increase in fΞ³ in region I is coupled with a decrease in the lattice parameter. In
contrast, in region III, the increase in fΞ³ in region III is coupled with an increase in the lattice
parameter. The decrease in fΞ³ in region II does not show significant effects on the lattice parameter.
2
See Appendix A for details on this calculation and a comparison of RA fractions determined by VSM and XRD.
37
I
II
Q&P
III
3.608
1.2
3.604
1.1
3.600
1.0
3.596
120
160
200
240
280
320
Carbon concentration in RA (wt. %)
Lattice parameter (Å)
Q&P reheated to 700 °C
Carbon concentration in RA (Wt. %)
Q&P
3.612
360
1.2
1.1
1.0
0.00
Quenching Temperature (°C)
Q&P reheated to 700 °C
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
Phase fraction retained austenite
(a)
(b)
Figure 4.9: (a) Lattice parameter of retained austenite and carbon content in retained austenite of Q&P specimens, before
and after reheating. (b) Carbon concentration in retained austenite plotted against the fraction of retained austenite.
Figure 4.9b shows the carbon concentration of retained austenite plotted against the phase fraction
of retained austenite. In contrast to the Q&P samples as discussed in §4.1.2, the reheated Q&P
samples show no clear correlation between the phase fraction and the carbon concentration.
4.2.2
Characterization of microstructure after reheating to 700 °C
In §4.1.3, the microstructure of samples with a QT of 160, 260 and 320 °C was analyzed. A
microstructural characterization of the same selection of samples after reheating to 700 °C will be
discussed in this paragraph.
4.2.2.1 Analysis of sample QT160 H700
Figure 4.10a and Figure 4.10a show EBSD phase maps of sample QT160H700 after reheating. Almost
no RA is observed, even though the fraction of retained austenite fΞ³ has increased compared to
sample QT160 (c.f. Figure 4.4a). Approximately 50% of the pixels could be indexed. Even regions with
seemingly high band contrast were not indexed. This suggests that the quality of the generated
Kikuchi patterns was low, indicating a bad measurement. Due to practical limitations, this
measurement was not redone.
Figure 4.10c and Figure 4.10d show SEM micrographs of sample QT160H700. Carbides and some
film-type RA can be observed in Figure 4.10d. Compared to sample QT160, the carbides have
coarsened (c.f. Figure 4.4). Furthermore, as later shown in Figure 4.14, some precipitation of carbides
from martensite has occurred.
38
(a)
(b)
(c)
(d)
Figure 4.10: Microstructure of sample QT160H700. (a) Phase map before cleaning. BCC is shown in red, FCC in blue. (b)
Combined EBSD band contrast and phase map. Cleaned version. (c) (d) SEM micrograph.
4.2.2.2 Analysis of sample QT260H700
Figure 4.11a and Figure 4.11b show EBSD phase maps of sample QT260H700 after reheating. Similar
to sample QT160H700 in Figure 4.10, almost no retained austenite (<0.1%) is observed. However,
approximately 70% of the pixels could be indexed, indicating good Kikuchi patterns. According to the
analysis in §4.2.1, this sample contains a phase fraction of retained of approximately 0.09. This
suggests that the retained austenite exists in regions which could not be indexed. This could be due
to either a small retained austenite grain size, or the effect of the reheating on the grain boundaries
and surroundings of retained austenite. For example, some cementite could have precipitated in the
grain boundaries of retained austenite, giving low band contrast.
Figure 4.11c and Figure 4.11d show SEM micrographs of sample QT260H700. Coarse carbides and
some film-type RA can be observed in Figure 4.11d. The martensite from the second quench (M2)
will be carbon depleted due to precipitation of carbides. Features which do not etch in nital are
therefore either carbides or austenite. Large unetched features should be large retained austenite
grains. However, due to lack of EBSD indexing, this cannot be said with complete certainty. The
features indicated with Ξ³? In Figure 4.11d could be large RA grains.
39
(a)
(b)
(c)
(d)
Figure 4.11: Microstructure of sample QT260H700. (a) Phase map before cleaning. BCC is shown in red, FCC in blue. (b)
Combined EBSD band contrast and phase map. Cleaned version. (c) (d) SEM micrograph.
4.2.2.3 Analysis of sample QT320H700
Figure 4.12 a and Figure 4.12b show EBSD phase maps of sample QT320H700 after reheating. Similar
to Figure 4.10a and Figure 4.11a, almost no RA is observed. Approximately 70% of the pixels could
be indexed, again indicating good quality of the Kikuchi patterns.
Figure 4.12c and Figure 4.12d show typical SEM micrographs of sample QT320H700. Coarse carbides
and some film-like RA can be seen.
40
(a)
(b)
(c)
(d)
Figure 4.12: Microstructure of sample QT320H700. (a) Phase map before cleaning. BCC is shown in red, FCC in blue. (b)
Combined EBSD band contrast and phase map. Cleaned version. (c) (d) SEM micrograph.
41
4.2.3
Short summary of quantification and identification of reheated samples
Table 4.2: Summary of properties of samples after reheating
Sample
QT160H700
QT260H700
QT320H700
Volume fraction of
retained austenite 𝑓𝛾
0.06 ± 0.01
0.09 ± 0.005
0.05 ± 0.005
Increase or decrease in
𝑓𝛾 after reheating
Increased by 0.02
Decreased by 0.06
Increased by 0.01
Carbon content in
retained austenite
(wt. %)
1.11 ± 0.05
1.035 ± 0.05
1.15 ± 0.05
Carbides
Yes, coarse carbides
Yes, coarse carbides
Yes, coarse carbides
Table 4.2 summarizes selected properties of the analyzed reheated samples. In the EBSD scans, no or
very little A observed. As observed in SEM micrographs, retained austenite exists in large enough
grain sizes to be retained austenite detectable by EBSD after reheating. It is speculated that the
absence of retained austenite in EBSD scans is due to low band contrast due to the presence of
carbides close to retained austenite.
Figure 4.13 shows the XRD spectrum of the selected samples reheated to 700°C. Compared with
Figure 4.7, RA peaks have sharpened, and ferrite peaks are sharper as well. This is most likely due to
reduction of stresses due to annealing of the samples. Furthermore, the distribution of carbon in the
austenite will change, also causing some peak sharpening.
42
QT320H700
QT260H700
QT160H700
60000
12000
10000
Counts
Offset Y values (counts)
50000
40000
8000
6000
4000
30000
2000
50
52
54
56
58
60
62
64
2Theta
20000
10000
0
40
50
60
70
80
90
100
110
120
130
2Theta
Figure 4.13: XRD spectrum of reheated samples. The applied offset values are +4000 for sample QT260H700 and +8000 for
sample QT320H700. The inset shows the {111} and {200} peaks of austenite in further detail.
4.2.4
In which temperature region does retained austenite decompose during reheating
with 5 °C s-1?
Figure 4.14 shows the derivative of dilatometry curves of selected samples during reheating to 700
°C. Furthermore, a sample was quenched to room temperature with 200 °C s-1 after identical
austenitization and reheated to 700 °C. It is shown as a comparison in Figure 4.14. A decrease in the
derivative around 680 to 690°C means that growth of austenite is occurring in samples QT160H700
and QT320H700. Growth of austenite is possible at temperatures above A1, which in QP-G is around
625-650 °C. A decrease in the derivative is not observed in the quenched sample and sample
QT260H700. Significant growth of austenite is therefore not occurring in these samples.
Around 500 °C, the quenched sample and samples QT320H700 and QT160H700 show a relative
contraction, which can be related to precipitation of carbides from martensite. Sample QT260H700
however, does not show the same degree of contraction. Precipitation of carbides from martensite
has an opposite effect on length change compared to austenite decomposition. It is hypothesized
that around 500 °C, RA starts to decompose in sample QT260H700. The increase in length due to RA
decomposition will then partially compensate the length decrease due to precipitation of carbides
from martensite.
43
QT320H700
QT260H700
QT160H700
QH700
Offset Y values d/dT(L/L0) x 10-6 (K-1)
50
40
30
20
10
0
100
200
300
400
500
600
700
Temperature (°C)
Figure 4.14: Derivative of dilatometry curves of selected samples during reheating. Also included is a quenched sample,
-1
reheated to 700 °C with a heating rate of 5 °C s as well.
To test this hypothesis, samples with a QT of 260 °C and 160 °C were reheated to 300, 400, 500 and
600 °C with 5 °C s-1 and directly quenched to room temperature upon reaching these temperatures.
Figure 4.15 shows fΞ³ as measured with the VSM of these samples, with the original samples shown as
a comparison. Two trend lines are drawn to indicate the possible trends of fΞ³ versus temperature.
Figure 4.15 shows that only samples QT260H600 and QT260H700 show significant decomposition of
RA. With a heating rate 5 °C s-1, decomposition of RA in samples with a QT of 260 °C is therefore
occurring above 500 °C, but not occurring between 25 and 500 °C. Samples with a QT of 160 °C show
no significant decomposition up to 600 °C, indicating that RA is more stable in these samples. During
reheating with 5 °C s-1, decomposition of retained austenite therefore starts around 500 °C in
samples with a QT of 260 °C, confirming the hypothesis.
QT260
QT160
0.18
0.15
0.12
f
0.09
0.06
0.03
0.00
0
100
200
300
400
500
600
700
Reheating temperature (°C)
Figure 4.15: Phase fractions as measured with VSM of samples with QT160 and QT260 reheated to indicated temperatures
-1
with 5 °C s .
44
4.3 Influence of carbon content in retained austenite on its decomposition
As discussed in §4.2, no retained austenite decomposition was detected in sample QT160 during
reheating with 5 °C s-1 to 700 °C. However, significant retained austenite decomposition was detected
in sample QT260. As discussed in §4.1, the carbon content of RA in sample QT160 was about 1.15 wt.
% and the carbon content in sample QT260 was about 1.02% wt. It is hypothesized that the
difference in carbon content could cause the different observed decomposition behavior.
To study the influence of carbon content, samples with quenching temperatures of 160 °, 260 °C and
320 °C were heated to 600 °C instead of 700 °C. This temperature was chosen to prevent the increase
in retained austenite fraction observed in §4.2.4. The reheating rate was set at 5 °C min-1, 10 °C min-1
and 15 °C min-1 in the dilatometer. These reheating rates were selected based on literature [11], [26],
to establish activation energies of RA decomposition.
Furthermore, a sample with a QT of 260 °C was reheated to 600 °C with 5 °C min-1 in the VSM. This
QT was selected based on its highest observed retained austenite fraction. The VSM was used to
accurately determine the phase fraction retained austenite during heating. The resulting phase
fraction was then correlated with the dilatometry curve of the corresponding reheated sample.
4.3.1
Dilatometric and magnetic measurements
In Figure 4.16, the retained austenite fraction of a sample with a QT of 260 °C as determined with
VSM during reheating is shown in red. The thermal expansion coefficient measured with the
dilatometer is shown in black. As indicated in Figure 4.16, four distinct regions of decomposition
behavior are observed based on the VSM results:
A) No decomposition of RA (RT – 250 ° C): Negligible change in phase fraction of retained
austenite.
B) Decomposition of RA (250 °C – 370 °C): The phase fraction of retained austenite decreases by
0.04 according to VSM. The relative length change due to decomposition of RA was
determined to be 0.025% using Equation 3.12. A relative length change of 0.025% correlates
to decomposition of a phase fraction of about 0.04 RA3. The decomposed RA phase fractions
as determined with VSM and dilatometer are in agreement.
C) No decomposition of RA (370 °C – 500 °C): Negligible change in the phase fraction of retained
austenite. In the dilatometry curve precipitation of carbides from martensite is observed.
D) Decomposition of RA (500 °C – 600 °C): The phase fraction of retained austenite decreases by
0.04 as determined with the VSM. No corresponding increase in volume could be
determined. The expected length change due to decomposition of a fraction of 0.04 RA is
0.016%. It is likely that the volume decrease due to precipitation of carbides from martensite
is partially compensating the volume increase due to RA decomposition. Furthermore, relief
of stresses originating from the Q&P treatment could also occur. These explanations are
explored in further detail in Appendix B.
3
Calculated using methods described in §3.2.5 under the assumption that RA contains 1 wt. % C. The average
length change of RA decomposition between 250 °C – 370 °C was used (c.f. Figure 3.7a).
45
Dilatometry
VSM
0.16
d/dT(L/L0) x10-6 (K-1)
30
0.12
25
A
B
C
D
0.08
20
0.04
15
10
0
200
400
Phase fraction retained austenite
35
0.00
600
T (°C)
Figure 4.16: Thermal expansion as measured with dilatometer and phase fraction of RA measured with the VSM during
-1
reheating to 600 °C. The heating rate was 5 °C min in both cases. Four distinct regions of RA decomposition behavior are
indicated.
Based on these observations, it was concluded that the dilatometer does not allow observation of RA
decomposition during precipitation of carbides from martensite. In this work, precipitation of
carbides from martensite was found to start between 400 °C and 420 °C. The dilatometer allows
observation of the length change from RA decomposition below that temperature. Keeping this in
mind, the dilatometry curves of samples with QT160, QT260 and QT320 during reheating to 600 °C
with varying heating rates were analyzed. The thermal expansion during reheating is shown in Figure
4.17a for QT260, Figure 4.17b for QT160 and Figure 4.17c for QT320.
For samples with a QT of 260 °C, the length changes due to RA decomposition are observed in Figure
4.17a. The relative length change observed, i.e. the peak area, decreases with increasing heating
rate. The peak position shifts to higher temperatures, i.e. decomposition starts at higher
temperatures with increasing heating rate. Based on this peak position shift, the activation energy
for RA decomposition into ferrite and carbides in QT260 samples is determined4 as 126 ± 15 kJ mol-1,
as shown in the Kissinger plot in Figure 4.17d. This value is comparable to values found in literature
for retained austenite decomposition (132 kJ mol-1 in [26], 113-174 kJ mol-1 in [62]. These activation
energies are typical for either carbon diffusion in austenite or pipe diffusion of iron in ferrite [26].
For samples with a QT of 160 °C, length changes due to RA decomposition are not observed in Figure
4.17b. Decomposition of RA is not occurring.
4
Based on methods described in §3.2.6.
46
For samples with a QT of 320 °C, small length changes due to retained austenite decomposition are
observed in Figure 4.17c. Decomposition of RA again starts at higher temperature with increasing
heating rate. Based on the position of the peaks, the activation energy is determined to be 128 ± 15
kJ mol-1, as shown in the Kissinger plot in Figure 4.17d. This activation energy is similar to the
activation energy of retained austenite decomposition observed for QT260 samples.
QT260:
15 °C min-1
10 °C min-1
5 °C min-1
QT160:
30
20
10
15 °C min-1
10 °C min-1
Heating rate 15°C min -1
Heating rate 10 °C min -1
Heating rate 5 °C min -1
Volume increases due
to RA decomposition
0
100
200
300
400
500
-1
30 Heating rate 15°C min
20
10
Heating rate 10 °C min -1
Heating rate 5 °C min -1
No volume increases:
no RA decomposition
0
600
100
Temperature (°C)
200
300
QT320:
15 °C min
-1
500
600
(b)
10 °C min-1
QT260
5 °C min-1
40
QT320
11.30
30
Heating rate 15°C min -1
20
Heating rate 10 °C min -1
11.04
10.78
Ln(T2i/)
Offset Y values d/dT(L/L0) x 10-6 (K-1)
400
Temperature (°C)
(a)
10
5 °C min-1
40
Offset Y values d/dT(L/L0) x 10-6 (K-1)
Offset Y values d/dT(L/L0) x 10-6 (K-1)
40
E = 126 kJ mol-1
E = 128 kJ mol-1
10.52
Heating rate 5 °C min -1
10.26
Small volume increases:
some RA decomposition
10.00
0.00163
0
100
200
300
400
500
600
0.00166
0.00169
0.00172
0.00175
1/Ti
Temperature (°C)
(d)
(c)
Figure 4.17: Derivative curves of samples with different heating rates. (a) Samples with QT260 (b) Samples with QT160 (c)
-6
samples with QT320 with different heating rates. The applied offset in all of these figures is +10x10 for a heating rate of 10
-1
-6
-1
°C min and +20x10 for a heating rate of 15 °C min .(d) Kissinger plot for the determination of activation energies for the
decomposition of retained austenite in samples with the indicated quenching temperatures.
In summary, retained austenite decomposition below 400 °C was absent in QT160 samples, while it
was observed in QT260 and QT320 samples. The decomposition of RA is heating rate dependent, and
its activation energy in QP-G is approximately 127 kJ mol-1. Carbon diffusion in austenite or pipe
diffusion of iron in ferrite are possible rate controlling factors for the decomposition of retained
austenite.
The main difference between QT160, QT260 and QT320 samples is the carbon concentration in the
RA. In QT160 the carbon content in RA is 1.15 wt. % C, in QT260 it is 1.02 wt. % C and in QT320 it is
1.07 wt.% C. This is an indication that higher carbon content in RA could stabilize retained austenite
against decomposition below 400 °C.
4.3.2
Characterization of decomposition products of RA
If higher carbon content stabilizes the retained austenite against decomposition, the retained
austenite that decomposes earliest should have relatively low carbon content. As found in §4.3.1, a
47
phase fraction retained austenite of 0.04 decomposed up to 500 °C during 5 °C min-1 reheating of a
QT260 sample. This decomposed phase fraction is hypothesized to have lower carbon concentration
compared to the non-decomposed phase fraction.
To characterize the decomposed phase fraction, a sample with a QT of 260 °C was reheated to 425°C
at 5 °C min-1. A phase fraction of 0.04 is assumed to decompose during the annealing. As observed in
Figure 4.16, 425 °C is a temperature slightly below the start of precipitation of carbides from
martensite. The martensitic microstructure will therefore not be strongly tempered. This will limit
the fraction of carbides precipitated from martensite and their growth. Microstructures containing
carbides formed during the decomposition of RA will therefore be more easily detectable [19].
After annealing to 425 °C and quenching, the phase fraction of RA was found to be 0.11. The
decomposition of a phase fraction of 0.04 has occurred, in line with expectations. Figure 4.18a shows
an EBSD phase map and of the microstructure of this sample (QT260H425). Note that the EBSD scan
in Figure 4.18a has been performed with a smaller step size of 30 nm and smaller area compared to
previously shown EBSD scans. An identification rate of 80% was achieved using EBSD, and RA grains
were not significantly affected due to the cleaning. The morphology of the RA has a clear distribution
of grain sizes, ranging from just detectable at 30 nm to approximately 500 nm wide grains.
(b)
(a)
Figure 4.18: (a) EBSD Phase map of sample QT260H425. BCC is shown in red, FCC in blue. The black scale bar represents 2
µm. (b) Tempered M2 encircled in white.
Figure 4.18b is a SEM micrograph of sample QT260H425. The micrograph shows retained austenite
films approximately 100 nm wide and a few µm long. Figure 4.18b shows larger RA grains of 0.5 µm
thick as well. Furthermore, Figure 4.18b shows a large tempered M2 grain, with retained austenite at
its edges.
A search for indications of decomposition of retained austenite using all available SEM micrographs
of sample QT260 and sample QT260H425 yielded no clear results. Although regions of M2 were
clearly tempered, no regions were found which could only be the results of retained austenite
decomposition. In the author’s opinion, other methods than SEM and EBSD are needed for
characterization of retained austenite decomposition products in this sample.
48
4.3.3
Study at longer annealing times
Four different temperature regions of decomposition behavior were observed in §4.3.1. In
temperature regions B, C and D the following was observed:
B) 250 – 370 °C: decomposition of retained austenite
C) 370 – 500 °C: stagnant stage
D) 500 – 600 °C: decomposition of retained austenite
However, the mechanisms of decomposition in these regions are unknown. To study the mechanisms
of decomposition in each of these temperature regions 30 minutes of isothermal annealing was
performed in a dilatometer. Samples with a QT of 260 °C were used. The selected isothermal
annealing temperatures were 350 °C for region B, 450 °C for region C and 550 °C for region D.
The heating rate to the isothermal holding temperature was 5 °C s-1. Due to the relatively high
heating rate, a thermal gradient will exist in the sample, with the center being warmer than the
edges. Upon reaching the holding temperature, this thermal gradient will gradually decrease, causing
an increase in length in the sample. The length change due to reduction of the thermal gradient is
determined as approximately +0.01% to +0.015%, and is unchanging after approximately 250 s of
holding.
The normalized length change during the annealing is shown in Figure 4.19a. The same figure also
shows the retained austenite fraction as determined with magnetic methods after quenching to
room temperature. Figure 4.19b shows the normalized length change during the quench after
annealing.
350 °C
450 °C
550 °C
0.030
350 °C
450 °C
550 °C
0.4
0.135
0.015
0.15
0.010
0.005
0.000
Length change (%)
Length change (%)
0.020
f at RT after annealing
0.025
0.2
ο‚»10% Martensite
formation
0.025
-0.005
-0.010
0.0
0
300
600
900
1200
1500
1800
0
Time during annealing step (s)
100
200
300
Temperature (°C)
(a)
(b)
Figure 4.19: (a) Length change during 1800s isothermal holding at the indicated temperatures for samples with a QT of 260
°C. The phase fraction of RA after quenching to room temperature is indicated. (b) Length change during the quench after
isothermal holding. The arrow indicates martensite formation during the quench from an isothermal holding temperature of
550 °C. The Ms is approximately 125 °C, indicating a carbon content of around 0.74% in the transforming austenite using
equations by van Bohemen [28].
For an annealing and holding temperature of 350 °C, a length change of 0.025% is observed. After
subtraction of the determined length change due to the reduction in thermal gradient, a net length
change of approximately 0.010% is determined. This length change can be correlated to
decomposition of an austenite fraction of approximately 0.015. No significant amount of carbides is
thought to have precipitated from martensite. After quenching to RT, the phase fraction of retained
austenite fΞ³ is measured by VSM as 0.135. Since fΞ³ in Q&P samples with a QT of 260 °C was 0.15, the
49
decomposed fraction as measured by VSM after quenching and the decomposed fraction of fΞ³
measured by dilatometry match.
The length change curve of isothermal holding at 350 °C gives some indications on the kinetics of
decomposition. The rate of length increase is seen to gradually slow during the annealing. This
suggests initial fast decomposition of retained austenite upon reaching 350 °C, followed by a gradual
slowing down of decomposition.
For an isothermal holding temperature of 450 °C, a net length change of -0.005% is determined
during holding. This net reduction of the length is due to precipitation of carbides from martensite.
No significant increase in length can be detected. This means that no significant decomposition of
retained austenite is occurring during the isothermal holding. Another possibility is that the negative
length change of precipitation of carbides from martensite is compensating the positive length
change of decomposition of retained austenite. After quenching to RT, fΞ³ was measured as 0.15,
which is the same fraction as the untreated sample QT260. This confirms that no detectable
decomposition has occurred.
For an isothermal holding temperature of 550 °C, a net length change of -0.02% is determined during
holding. This net length decrease most likely consists of two parts:
ο‚·
ο‚·
Length decrease due to stress relief, as explored in further detail in Appendix B.
Length decrease due to precipitation of carbides from martensite. This is most likely less
significant than length decrease due to stress relief. Figure 4.14 shows that the process of
carbide precipitation has almost finished at 550 °C.
No length increases due to decomposition of RA are observed during the holding. However, during
the quench to room temperature from 550 °C after isothermal holding, formation of martensite is
occurring. A martensitic phase fraction of ο‚»0.1 is formed, with an Ms of ο‚»125 °C. The carbon content
in the transforming austenite was determined as 0.74 wt. % C, using equations by van Bohemen [28].
Samples with a QT of 260 °C without annealing have an average carbon concentration in RA of 1.02
wt. % C. This means that the carbon concentration in RA reduces by approximately 0.3 wt. % C in 30
minutes. This carbon is most likely precipitated into cementite or alloy carbides, but this was not
investigated in further detail. Podder et al. [19] encountered this phenomenon as well at lower
isothermal holding temperatures of 450 °C. They observed minute quantities of cementite
precipitating around austenite. The local reduction in carbon concentration destabilized the retained
austenite against the martensitic transformation upon quenching.
50
4.4 Effect of partitioning time on thermal stability of retained austenite
The effect of partitioning time on the thermal stability of retained austenite was studied by varying
the partitioning time of Q&P samples with a QT of 260 °C. This QT was selected based on its highest
observed retained austenite fraction. The partitioning time was increased from 50 seconds to 300
and 600 seconds, while the rest of the Q&P treatment was unchanged. A sample with a partitioning
time 𝑑𝑝 of 300 s was reheated to 600 °C with a heating rate of 5 °C min-1 in the VSM. The dilatometer
was used to reheat the sample with a 𝑑𝑝 of 600 s to 600 °C with a heating rate of 5 °C min-1.
Figure 4.20 shows the fraction of retained austenite as determined using the VSM. A slight increase
of fΞ³ with increasing partitioning time up to 600 s was measured.
Figure 4.20b shows the thermal expansion coefficient during reheating to 600 °C in the dilatometer.
The standard sample with a 𝑑𝑝 of 50 s is treated in further detail in §4.3. The absence of a clear peak
in the sample with a 𝑑𝑝 of 600 s means that no detectable retained austenite decomposition is
occurring in the temperature range 250 °C to 370 °C. The longer partitioning time has stabilized the
retained austenite.
QT260
0.20
tp 50 s
30
No peak
25
d/dT(L/L0) x10-6 (K-1)
0.16
f
0.12
0.08
0.04
0.00
tp 600 s
20
Peak
15
10
5
0
0
100
200
300
400
500
600
0
Partitioning time (s)
100
200
300
400
500
600
Temperature (°C)
(a)
(b)
Figure 4.20: (a) Fraction of RA as a function of partitioning time in samples with a QT of 260 °C. (b) Thermal expansion
-5
coefficient during reheating to 600 °C in the dilatometer. The offset applied is +10 for the sample with a tp of 600 s.
Confirmation of the absence of retained austenite decomposition is found by thermomagnetic
methods. The retained austenite fraction during reheating to 600 °C as measured in the VSM is
plotted in Figure 4.21. The standard sample with a 𝑑𝑝 of 50 s is treated in further detail in §4.3.
The sample with a 𝑑𝑝 of 300 s shows an approximately constant fraction of retained austenite up to
500 °C. A slight decrease (0.003) in the phase fraction can be seen in the temperature range from 400
°C to 500°C. In the temperature range of 500 °C to 600 °C, the phase fraction shows a stronger
decrease of about 0.04.
51
tp 50 s
tp 300 s
Phase fraction retained austenite
0.20
Stagnant
Stagnant
RA decomp.
RA decomp.
Stagnant
RA decomp.
0.16
0.12
0.08
0.04
0.00
0
100
200
300
400
500
600
Temperature (°C)
Figure 4.21: Phase fraction of retained austenite during reheating to 600 °C in the VSM.
As shown in Figure 4.20 and Figure 4.21, in samples with increased partitioning times of 300 to 600 s,
retained austenite is more stable against decomposition into ferrites and carbides. This statement is
especially true in the temperature region below 400 °C. A theory based on carbon homogenization
during partitioning will now be proposed to explain this increase in stability.
4.4.1
Carbon homogenization during partitioning
Mecozzi et al. [59] simulated the carbon concentration in residual austenite during partitioning. A
phase field model was used to model a (0.25C-3Mn-1.5Si) steel alloy. Figure 4.22 shows the carbon
concentration in retained austenite at partitioning times of 50 s (Figure 4.22a) and 300 s (Figure
4.22b). Figure 4.22c shows a phase map, with ferrite yellow, and austenite all other colors. These
figures are intermediate results of that paper, courtesy of Pina Mecozzi.
As shown in Figure 4.22a, after a partitioning time of 50 s the carbon is mostly concentrated at the
edges of residual austenite grains. In contrast, after a partitioning time of 300 s the carbon has
diffused into these grains. This has led to a homogenization of the carbon concentration gradient.
Figure 4.22d shows a carbon profile along the line A-B of the selected grain for different partitioning
times, illustrating the effect of homogenization. A partitioning time of 300s decreases the carbon
concentration at the Ξ±/Ξ³ interface, while the carbon concentration in the grain is somewhat higher.
Schematic profiles of the carbon concentration in a retained austenite grain are shown in Figure 4.23.
The carbon profile after a Pt of 50 s is shown in Figure 4.23a, while Figure 4.23b shows the carbon
profile after a Pt of 300 s. The carbon content necessary to prevent martensitic transformation
during quenching is indicated with Ms. As discussed in §4.3, retained austenite with relatively low
carbon content decomposes during annealing up to 400 °C. Retained austenite with relatively high
carbon content does not decompose. Indicated with Cbf is the hypothesized carbon content
necessary to prevent retained austenite decomposition during annealing up to 400 °C.
As shown in Figure 4.23a, after 50 s partitioning part of the retained austenite will have a carbon
concentration below Cbf. Part of the RA is able to decompose during annealing. A partitioning time of
300 s however, homogenizes the entire grain to a carbon concentration above Cbf. No retained
austenite will therefore be able to decompose during annealing.
52
Wt % C
2.0
1.5
1.0
0.5
(a)
0
(b)
50 s
300 s
Carbon concentration (wt. %)
2.5
2.0
1.5
1.0
0.5
0.0
0
2
4
6
8
10
Distance along line A-B (m)
(d)
(c)
2
Figure 4.22: (a) Carbon concentration in RA grains, Pt = 50s, QT = 243 °C, size 40x40 µm . (b) Carbon concentration in RA
2
grains, Pt = 300s, QT = 243 °C, size 40x40 µm . (c) Phase map. Ferrite shown in light yellow. (d) Carbon profile along the line
A-B of the selected grain with different partitioning times. All the data in these figures are courtesy of Pina Mecozzi
(a)
(b)
Figure 4.23: Schematic carbon profile in a retained austenite grain after different partitioning times. (a) 𝑑𝑝 50 s. (b) 𝑑𝑝 300 s.
53
During reheating with 5 °C min-1 to temperatures above 500 °C, no effect of homogenization on the
kinetics of RA decomposition is detected as seen in Figure 4.21. However, increasing the partitioning
time up to 600 s still stabilizes the RA below 500 °C, leading to a higher RA fraction at this
temperature during reheating.
It is possible that carbon diffusion during reheating has homogenized the carbon profile in the
sample with 50 s partitioning time as well. This homogenization could nullify any effects of longer
partitioning time on RA decomposition. A simple model has been applied to calculate the total
diffusion distance of carbon during reheating at 5 °C min-1. Its results are shown in Figure 4.24.
The model consists of numerical integration of the equation π‘₯ = βˆšπ·π‘‘, with π‘₯ the diffusion distance,
𝐷 the diffusion coefficient and 𝑑 the time. The total diffusion distance is then βˆ‘ π‘₯𝑖 , where π‘₯𝑖 is the
diffusion distance at temperature 𝑖. The temperature step was 1 °C. For a heating with 5 °C s-1, π‘₯ was
𝑄𝑑
calculated by setting 𝑑 = 12 s and using 𝐷 = 𝐷0 𝑒π‘₯𝑝(βˆ’π‘…π‘‡ ). For austenite, 𝐷0 = 1.5 βˆ— 10βˆ’5 m2 s-1 and
𝑄𝑑 = 142.1 kJ mol-1 were used [59]. For ferrite, 𝐷0 = 1.1 βˆ— 10βˆ’6 m2 s-1 and 𝑄𝑑 = 87.4 kJ mol-1 were
used [45].
Indicated in Figure 4.24 is the extra diffusion distance achieved in 250 s longer partitioning time,
calculated with the same model. The carbon diffusion distance up to 500 °C during reheating is
significantly larger than that achieved during partitioning.
Homogenization of carbon during reheating with 5 °C min-1 is therefore unlikely to affect the kinetics
of RA decomposition at temperatures higher than 500 °C. Furthermore, homogenization of carbon
during reheating with 5 °C min-1 is hypothesized to play no significant role during reheating up to 300
°C. Homogenization during reheating would stabilize the low carbon RA against bainitic
decomposition, but is not experimentally seen. This is because the diffusion distance is thought too
low for carbon to diffuse from high concentration RA to low concentration RA during reheating.
Total diffusion distance of carbon x 10-6 (m)
In BCC
In FCC
10
8
6
4
Extra diffusion distance
due to 250 s longer Pt
2
0
0
100
200
300
400
500
Temperature (°C)
-1
Figure 4.24: Calculated total diffusion distance of carbon in FCC and BCC during reheating with 5 °C min .
54
4.5 Discussion and theory of observed RA decomposition mechanisms
A summary of the observed decomposition behavior of RA is given. These observations are then
linked to known theory about austenite decomposition.
The following has been observed:
There are 4 different temperature regions of decomposition behavior in QP-G. Each of these regions
shows different decomposition behavior:
A) 25 °C – 250 ° C: No decomposition of retained austenite
B) 250 °C – 370 °C: Decomposition of retained austenite is possible, depending on carbon
content in RA. The mechanism via which the RA decomposes could not be experimentally
observed. Low carbon RA decomposes first. After low carbon RA has decomposed, no
decomposition of high carbon RA is observed.
C) 370 °C – 500 °C: No significant decomposition of retained austenite observed.
D) 500 °C – 600 °C: Decomposition, excess carbon likely precipitates into carbides from RA. The
local reduction in carbon concentration destabilizes the RA, which will transform into
martensite upon quenching.
These regions will be treated one by one in the following sections.
Region A: 25 °C – 250 °C
No decomposition of RA has been observed in this region. The reason for this is straightforward: in
this temperature range, in essence no diffusion of atoms in austenite is occurring. Displacive
decompositions mechanisms are therefore the only kinetically possible transformation mechanisms
in this temperature range. Since the austenite is already retained, i.e. stable against the martensitic
transformation, stability against displacive decomposition mechanisms is implied. Therefore, no
decomposition of RA will occur in practically important timescales.
Region B: 250 °C – 370 °C
In this region, low-carbon RA will decompose. High-carbon RA was not found to decompose in the
investigated timescales. The exact mechanism of decomposition of low-carbon RA has not been
observed. Thermomagnetic measurements (e.g. Figure 4.21) show a decrease in fΞ³ only explainable
by formation of BCC iron. Therefore, the decomposition products of low-carbon RA are assumed as Ξ³
οƒ  Ξ± + ΞΈ (or other carbides). However, the formation of cementite or other carbides resulting from
the decomposition of RA has not been experimentally observed in this work. The lack of any clear
carbide formation due to RA decomposition could also be a result of segregation of carbon to
dislocations, grain boundaries or other defects.
In a classic paper by Thomas [63], he argues that β€œThere is a serious lack of experimental evidence to
document how the transformation of retained austenite is influenced by alloying – whether it
transforms to fresh martensite, or bainite ferrite + M3C or normal ferrite + M3C”. Fresh martensite,
bainitic ferrite and normal ferrite are suggested as decomposition products. In this work, the
formation of fresh martensite is excluded on the basis that RA is already stable against the
martensitic mechanism. This leaves normal ferrite and bainitic ferrite as possible decomposition
products. Based on the temperature region of 250 °C – 370 °C in which the decomposition of low55
carbon RA takes place, bainitic ferrite is the most likely decomposition product. This is suggested as
well by Honeycombe and Bhadeshia [64, p. 186], who state that β€œThe little available evidence
suggests that in the range 230-300°C, retained austenite decomposes to bainitic ferrite and
cementite”. Bhadeshia [65, p. 12] argues that the formation of bainite consists of formation of
supersaturated bainitic ferrite, followed by the precipitation of cementite from the supersaturated
ferrite.
A thermodynamic reasoning based on Bhadeshia’s arguments helps explain the difference in
decomposition behavior between high-carbon and low-carbon RA. Carbon is a strong austenite
stabilizer. If decomposition of RA takes place via (bainitic) ferrite formation, high-carbon RA will have
a higher thermal stability against decomposition into ferrite. Conversely, high-carbon RA should have
lower stability against decomposition into cementite or other carbides, since the driving force for
cementite precipitation will be higher.
The driving force for precipitation from RA under paraequilibrium conditions of both ferrite (Figure
4.25a) and cementite (Figure 4.25b) has been modelled using ThermoCalc in QP-G using varying
carbon concentrations. Figure 4.25a clearly shows that an increase in carbon content decreases the
driving force for the FCC to BCC transformation. Figure 4.25b shows that an increase in carbon
content increases the driving force for the FCC to cementite transformation.
1.6 Wt. % C
5000
B
C
0.6 Wt. % C
1.6 Wt. % C
5000
D
A
4000
Driving Force G (J/mol)
Driving Force G (J/mol)
A
1.0 Wt. % C
3000
2000
1000
0
1.0 Wt. % C
B
C
0.6 Wt. % C
D
4000
3000
2000
1000
0
0
100
200
300
400
500
600
700
0
Temperature (°C)
100
200
300
400
500
600
700
Temperature (°C)
(a)
(b)
Figure 4.25: (a) Driving force for the FCC to BCC transformation under paraequilibrium conditions for QP-G with varying
carbon contents in the RA. A positive driving force means that the transformation is favorable. (b) Chemical driving force for
cementite precipitation under paraequilibrium conditions from austenite with varying carbon content. A positive driving
force means that precipitation is favorable.
The driving forces as shown in Figure 4.25 suggest that the driving force for formation of BCC from
FCC is larger than the driving force for cementite precipitation from FCC. This implies that
decomposition into ferrite is favored in this temperature region.
As is observed in this work, high-carbon RA does not decompose, while low-carbon RA does
decompose into ferrite in this temperature region. This suggests that a critical carbon concentration
Cbf in RA exists allowing decomposition into bainitic ferrite. In RA grains that have a carbon
concentration below Cbf, decomposition of RA into bainitic ferrite is thermodynamically and
kinetically possible. In RA grains that have a carbon concentration above Cbf, decomposition of RA
into bainitic ferrite is thermodynamically and kinetically less favorable. These grains will therefore
not decompose into bainitic ferrite, at least not in the timescales investigated (ο‚»1 hour). Partly, this is
due to the presence of Mn, Mo and Si in QP-G. These elements slow bainite formation [66]–[69].
56
At longer timescales however, high-carbon RA is not expected to be completely stable. The driving
force for cementite precipitation from RA grains is increased when the carbon concentration
increases. In a RA grain which has a carbon concentration above Cbf, bainitic ferrite will not form.
However, it is known that eventually cementite precipitation will take place from metastable RA at
elevated temperatures [15]. This will lead to eventual decomposition of high-carbon RA.
Figure 4.26 shows a schematic extrapolation of the solubility lines of ferrite and cementite in Fe-C
austenite.
Figure 4.26: Schematic extrapolated solubility lines for ferrite and cementite in Fe-C austenite. Adapted from [70]. The
concentration of carbon in a hypothetical low carbon RA grain is denoted with C1, while the concentration of carbon in high
carbon RA is denoted with C2. At equal temperatures, the difference in carbon content C between the extrapolated
solubility lines and C1 and C2 is proportional to the driving force for transformation as indicated.
Region C: 370 °C – 500 °C
In this region, decomposition of RA was not observed in the timescales investigated. At these
temperatures, the driving force for displacive transformations is not high enough for the RA to
transform via displacive mechanisms. Furthermore, diffusion at these temperatures is limited, and no
significant diffusion of iron and substitutional elements will occur. Diffusional transformation
products will therefore show very low growth rates. The decomposition of RA will therefore be slow
because of a combination of (too) low undercooling for displacive mechanisms and low growth rates
for diffusional mechanisms.
Region D: 500 °C – 600 °C
In this region, decomposition of RA was observed. The mechanism was local precipitation of
cementite, which destabilizes RA. Upon quenching from annealing in this region, martensite
formation was detected.
57
Decomposition of RA via this mechanism is comparable to formation of pearlite from hypereutectoid
austenite, and this mechanism will therefore be called the pearlitic mechanism. In pearlite formation
from a hypereutectoid austenite, cementite is the leading phase. Cementite will be formed until the
eutectoid composition is reached. When the eutectoid composition is reached, pearlite formation
will commence.
However, due to the relatively high concentration of Mn in this alloy, pearlite formation is slowed
considerably. Partitioning of Mn can control growth of cementite at these temperatures [71].
This region is expected to extend above 600 °C until an equilibrium is reached, for example in the Ξ±+
Ξ³ region.
TTT diagram applicability
Based on the observed decomposition behavior, separate kinetics of decomposition exists for highcarbon and low-carbon RA. The kinetic decomposition behavior of austenite can be summarized in a
Time-Temperature-Transformation (TTT) diagram. Sarikaya et al. [72] concluded that β€œβ€¦ separate
kinetics must be considered for the retained austenite from that of the bulk alloy”, and proposed
superimposed TTT diagrams containing the bulk alloy and an imaginary alloy corresponding to the
RA, with an appropriate C concentration.
In the author’s opinion, the results in this work validate a TTT-like diagram approach for an
understanding of the decomposition behavior. In essence RA behaves as austenite which is simply
higher in carbon content compared to the bulk alloy. However, the morphology of retained austenite
and its microstructural environment are different compared to bulk austenite with the same carbon
content. The morphology and microstructural environment of retained austenite might influence its
decomposition behavior. Figure 4.27 shows a schematic TTT-like diagram drawn for QP-G, with one
of the performed heat treatments indicated.
58
Figure 4.27: Schematical and Hypothetical TTT-like diagram of QP-G. A very schematic Q&P heat treatment with applied
reheating is shown. Schematical Pearlite and Bainite start lines of the base alloy and on the imaginary RA alloy are shown.
The superimposed reheating curve is shown only to give a rough indication of the behavior of decomposition, since strictly
speaking it should be plotted on a continuous reheating diagrams.
Figure 4.28 summarizes the decomposition mechanisms of retained austenite.
Figure 4.28: Mechanisms of RA decomposition
59
Chapter 5 Conclusions
5.1 Retained austenite decomposition in Quenching & Partitioning steel
In this work, the thermal stability of retained austenite has been studied in a wide range of
temperatures. A coherent theory of retained austenite decomposition could be formulated. The
thermal stability of retained austenite is dependent on the quenching temperature.
The results suggest that during reheating with 5 °C min-1, 4 stages of retained austenite
decomposition exist in the Quenching & Partitioning alloy used in this work:
A) 25 °C to 250 ° C: No decomposition of retained austenite.
B) 250 °C to 370 °C: Decomposition of retained austenite is possible, depending on carbon
content in retained austenite. Low carbon retained austenite decomposes first, and is
hypothesized to decompose into bainitic ferrite and carbides. After low carbon retained
austenite has decomposed, no decomposition of high carbon retained austenite is observed.
C) 370 °C to 500 °C: No significant decomposition of retained austenite observed, because of
(too) low undercooling for displacive decomposition mechanisms and low growth rates for
diffusional decomposition mechanisms.
D) 500 °C to 600 °C: Decomposition of retained austenite. Excess carbon likely precipitates into
carbides from retained austenite. The local reduction in carbon concentration destabilizes
the retained austenite, which will transform into martensite upon quenching. This
mechanism is similar to formation of pearlite from hypereutectoid austenite, which will form
cementite until the eutectoid composition is reached. This mechanism is expected to extend
above 600 °C until an equilibrium is reached in the Ξ±+ Ξ³ region.
The mechanisms of decomposition of retained austenite have been successfully mapped to a TTT
diagram. Carbon-enriched retained austenite shows the same decompositions products as austenite,
and in essence behaves as austenite which is higher in carbon content compared to the bulk alloy.
However, the morphology of retained austenite and its microstructural environment are different
compared to bulk austenite with the same carbon content. The morphology and microstructural
environment of retained austenite might therefore influence its decomposition behavior. However,
carbon content and distribution in retained austenite remain the main factors determining the
thermal stability of retained austenite and its decomposition behavior in Quenching & Partitioning
steels.
Low-carbon retained austenite is less stable than high-carbon retained austenite against
decomposition into ferrite, because carbon stabilizes austenite against the ferrite transformation in
in steel. Conversely, high-carbon retained austenite is less stable than low-carbon retained austenite
against precipitation of cementite from retained austenite. As observed in this work, decomposition
of retained austenite with carbides as the leading phase is a slower process compared to
decomposition with ferrite as the leading phase. During annealing processes, high-carbon retained
austenite will therefore remain stable for longer periods of time compared to low-carbon retained
austenite.
Homogenization of the carbon gradient in retained austenite grains when increasing partitioning
time up to 600 s gives rise to an increase in thermal stability of retained austenite grains. The
absence of low-carbon austenite due to homogenization leads to net slower decomposition.
60
5.2 About experimental techniques
The dilatometer is a useful instrument for tracking the phase fraction of retained austenite before
precipitation of carbides from martensite starts in Quenching & Partitioning steel. When
precipitation starts however, length changes due to retained austenite decomposition are partially
compensated by length changes due to precipitation of carbides from martensite. Furthermore, a
relative decrease in the relative volume change of retained austenite decomposition occurs due to
thermal expansion effects. The combination of these effects leads to limited usefulness of the
dilatometer in tracking the phase fraction of retained austenite at temperatures above 400 °C during
isochronal or isothermal annealing. However, the mechanisms of decomposition of retained
austenite were successfully derived by analysis of dilatometry data.
Thermomagnetic methods have been successfully used to accurately determine the phase fraction of
retained austenite during annealing up to 600 °C. The mechanisms of decomposition of retained
austenite could not be derived by analysis of thermomagnetic data.
The combination of dilatometry and thermomagnetic methods is therefore a powerful instrument to
study the behavior of retained austenite during annealing.
5.3 About Quenching & Partitioning steel
When the quenching temperature is low enough that no martensite from the second quench is
formed, lattice parameters of retained austenite increase with decreasing quench temperature. This
is because more carbon is available from martensite to partition into ever decreasing fractions of
residual austenite.
61
Chapter 6 Recommendations
ο‚·
ο‚·
ο‚·
ο‚·
ο‚·
ο‚·
ο‚·
ο‚·
The mechanical properties after annealing of Q&P steels have not been investigated. A study
on the effect of the decomposition of retained austenite on these properties would be
interesting. In conventional martensitic steels, temper embrittlement occurs due to the
formation of very fine cementite needles due to the decomposition of film-like retained
austenite. This work does not address temper embrittlement, and further study on this topic
is advised.
Film-like retained austenite has lower carbon content in Q&P steels compared to larger
grains. Film-like retained austenite is more stable under stress than larger retained austenite,
and responsible for the enhanced ductility at strain levels exceeding 5%. The retained
austenite which decomposes first during annealing has lower carbon content according to
this work. Hot-TEM studies could be performed to check whether film-like retained austenite
is indeed decomposing first.
In a heat treatment not treated in further detail in this work, QP-G samples with a QT of 160
and 260 °C were reheated with 100 °C s-1 to 700 °C, held there for 20 minutes and quenched.
Retained austenite fractions of 0.2 for QT260 and 0.25 for QT160 were achieved. These are
very high retained austenite fractions, and it would be interesting to see what the
mechanical properties of these samples are.
Heat treatments typical for welding, pre-heating for welding, galvanizing, paint baking and
application of other corrosion resistant coatings should be applied on promising Q&P steels
to check whether they retain their promising mechanical properties after these treatments.
The decomposition behavior at lower temperatures and/or longer timescales has not been
studied.
DTA/DSC studies applying the same isochronal heat treatments as performed in this work
should be performed. A comparison between DTA/DSC and the techniques applied in this
thesis will allow a very complete picture of both retained austenite decomposition, and of
the strengths and weaknesses of the respective techniques in studying retained austenite
decomposition.
Thermomagnetic methods allow accurate observations of the phase fraction of retained
austenite during annealing. In the author’s opinion, it is worthwhile to create a VSM allowing
the application of heat treatments. Current VSM systems have limited heating rates (e.g. a
maximum of 3 °C s-1) and do not allow for quenching or rapid cooling. The ability to perform
magnetic measurements during heat treatments will allow direct and accurate identification
of austenite fractions below the Curie temperature.
In the dilatometer, derivative curves (dL/dT [L/L0]) allow better visual identification of
metallurgical processes during isochronal reheating than purely looking at length change
curves (L/L0). It is therefore recommended that more use should be made of derivative
dilatometry curves during isochronal heating.
62
Chapter 7 References
[1]
S. Keeler and M. Kimchi, β€œAHSS Application Guidelines,” 2014, WorldAutoSteel.
[2]
G. F. V. Voort, Ed., ASM Handbook volume 9, Metallography and Microstructures. 2004.
[3]
D. P. Koistinen and R. E. Marburger, β€œA general equation prescribing the extent of the
austenite-martensite transformation in pure iron-carbon alloys and plain carbon steels,” Acta
Metall., vol. 7, no. 1, pp. 59–60, 1959.
[4]
S. Morito, X. Huang, T. Furuhara, T. Maki, and N. Hansen, β€œThe morphology and
crystallography of lath martensite in alloy steels,” Acta Mater., vol. 54, no. 19, pp. 5323–5331,
2006.
[5]
S. van der Zwaag, L. Zhao, S. O. Kruijver, and J. Sietsma, β€œThermal and Mechanical Stability of
Retained Austenite in Aluminum-containing Multiphase TRIP Steels.,” ISIJ Int., vol. 42, no. 12,
pp. 1565–1570, 2002.
[6]
H. K. D. H. Bhadeshia and D. V. Edmonds, β€œBainite in silicon steels: new composition–property
approach Part 2,” Met. Sci., vol. 17, no. 9, pp. 420–425, 1983.
[7]
X. C. Xiong, B. Chen, M. X. Huang, J. F. Wang, and L. Wang, β€œThe effect of morphology on the
stability of retained austenite in a quenched and partitioned steel,” Scr. Mater., vol. 68, no. 5,
pp. 321–324, Mar. 2013.
[8]
K. S. Choi, Z. Zhu, X. Sun, E. De Moor, M. D. Taylor, J. G. Speer, and D. K. Matlock,
β€œDetermination of carbon distributions in quenched and partitioned microstructures using
nanoscale secondary ion mass spectroscopy,” Scr. Mater., vol. 104, pp. 79–82, 2015.
[9]
J. G. Speer, F. C. Rizzo, D. K. Matlock, and D. V Edmonds, β€œThe β€˜ Quenching and Partitioning ’
Processβ€―: Background and Recent Progress,” vol. 8, no. 4, pp. 417–423, 2005.
[10]
J. Speer, D. K. Matlock, B. C. De Cooman, and J. G. Schroth, β€œCarbon partitioning into austenite
after martensite transformation,” Acta Mater., vol. 51, no. 9, pp. 2611–2622, 2003.
[11]
E. De Moor, S. Lacroix, L. Samek, J. Penning, and J. Speer, β€œDilatometric Study of the Quench
and Partitioning Process,” 3rd Int. Conf. Adv. Struct. Steels, pp. 1–6, 2006.
[12]
T. D. Bigg, D. K. Matlock, J. G. Speer, and D. V. Edmonds, β€œDynamics of the Quenching and
Partitioning (Q&P) Process,” Solid State Phenom., vol. 172–174, pp. 827–832, 2011.
[13]
T. D. Bigg, D. V. Edmonds, and E. S. Eardley, β€œReal-time structural analysis of quenching and
partitioning (Q&P) in an experimental martensitic steel,” J. Alloys Compd., vol. 577, no. SUPPL.
1, pp. S695–S698, 2013.
[14]
F. Tariq, β€œMicrostructural evolution during tempering of Quenched and Partitioned Medium
Carbon Low Alloy Steel,” presented at the Third International Conference on Aerospace
Science and Engineering, Pakistan, 2013
63
[15]
H. J. Jun, S. H. Park, S. D. Choi, and C. G. Park, β€œDecomposition of retained austenite during
coiling process of hot rolled TRIP-aided steels,” Mater. Sci. Eng. A, vol. 379, no. 1–2, pp. 204–
209, Aug. 2004.
[16]
W. Shi, L. Li, B. De Cooman, P. Wollants, and C.-X. Yang, β€œThermal stability of retained
austenite in TRIP steel after different treatments,” J. Iron Steel …, vol. 15, no. 1, pp. 61–64,
2008.
[17]
M. Amirthalingam, M. J. M. Hermans, L. Zhao, and I. M. Richardson, β€œQuantitative analysis of
microstructural constituents in welded transformation-induced-plasticity steels,” Metall.
Mater. Trans. A Phys. Metall. Mater. Sci., vol. 41, no. February, pp. 431–439, 2010.
[18]
N. Luzginova, L. Zhao, and J. Sietsma, β€œEvolution and thermal stability of retained austenite in
SAE 52100 bainitic steel,” Mater. Sci. Eng. A, vol. 448, no. 1–2, pp. 104–110, 2007.
[19]
A. Saha Podder and H. K. D. H. Bhadeshia, β€œThermal stability of austenite retained in bainitic
steels,” Mater. Sci. Eng. A, vol. 527, no. 7–8, pp. 2121–2128, 2010.
[20]
J. A. Cameron, J. Iron Steel Inst., vol. 194, pp. 260–267, 1956.
[21]
A. S. Podder, I. Lonardelli, A. Molinari, and H. K. D. H. Bhadeshia, β€œThermal stability of retained
austenite in bainitic steel: an in situ study,” Proc. R. Soc. A Math. Phys. Eng. Sci., vol. 467, no.
2135, pp. 3141–3156, 2011.
[22]
P. V. Morra, A. J. Böttger, and E. J. Mittemeijer, β€œDecomposition of iron-based martensite: A
kinetic analysis by means of differential scanning calorimetry and dilatometry,” J. Therm. Anal.
Calorim., vol. 64, no. 3, pp. 905–914, 2001.
[23]
T. Waterschoot, K. Verbeken, and B. C. De Cooman, β€œTempering Kinetics of the Martensitic
Phase in DP Steel,” ISIJ Int., vol. 46, no. 1, pp. 138–146, 2006.
[24]
S. Primig and H. Leitner, β€œSeparation of overlapping retained austenite decomposition and
cementite precipitation reactions during tempering of martensitic steel by means of thermal
analysis,” Thermochim. Acta, vol. 526, no. 1–2, pp. 111–117, Nov. 2011.
[25]
E. Kozeschnik and H. K. D. H. Bhadeshia, β€œInfluence of silicon on cementite precipitation in
steels,” Mater. Sci. Technol., vol. 24, no. 3, pp. 343–347, 2008.
[26]
L. Cheng, C. M. Brakman, B. M. Korevaar, and E. J. Mittemeijer, β€œThe Tempering of IronCarbon Martensiteβ€―; Dilatometric and Calorimetric Analysis,” Metall. Trans. A, vol. 19, no. 10,
pp. 2415–2426, 1988.
[27]
M. J. Santofimia, L. Zhao, R. Petrov, C. Kwakernaak, W. G. Sloof, and J. Sietsma,
β€œMicrostructural development during the quenching and partitioning process in a newly
designed low-carbon steel,” Acta Mater., vol. 59, no. 15, pp. 6059–6068, Sep. 2011.
[28]
S. M. C. van Bohemen, β€œBainite and martensite start temperature calculated with exponential
carbon dependence,” Mater. Sci. Technol., vol. 28, no. 4, pp. 487–495, Apr. 2012.
[29]
T. A. Kop, β€œA dilatometric study of the austenite/ferrite interface mobility,” Ph.D. dissertation,
Delft University of Technology, The Netherlands, 2000.
64
[30]
S. Arajs, β€œFerromagnetic Curie Temperatures of Iron Solid Solutions with Germanium, Silicon,
Molybdenum, and Manganese,” Phys. status solidi, vol. 11, p. 121, 1965.
[31]
S. M. C. van Bohemen, β€œThe nonlinear lattice expansion of iron alloys in the range 100–
1600K,” Scr. Mater., vol. 69, no. 4, pp. 315–318, Aug. 2013.
[32]
S. M. C. van Bohemen, β€œAustenite in multiphase microstructures quantified by analysis of
thermal expansion,” Scr. Mater., vol. 75, pp. 22–25, Mar. 2014.
[33]
H. E. Kissinger, β€œReaction kinetics in differential thermal analysis,” Anal. Chem., vol. 29, no.
1956, pp. 1702–1706, 1957.
[34]
E. J. Mittemeijer, L. Cheng, P. J. Schaaf, C. M. Brakman, and B. M. Korevaar, β€œAnalysis of
nonisothermal transformation kinetics; tempering of iron-carbon and iron-nitrogen
martensites,” Metall. Trans. A, vol. 19, no. 4, pp. 925–932, 1988.
[35]
E. De Moor, C. Föjer, J. Penning, A. J. Clarke, and J. G. Speer, β€œCalorimetric study of carbon
partitioning from martensite into austenite steel,” Phys. Rev. B - Condens. Matter Mater.
Phys., vol. 82, pp. 1–5, 2010.
[36]
E. J. Mittemeijer, A. Van Gent, and P. J. Van Der Schaaf, β€œAnalysis of Transformation Kinetics
by Nonisothermal Dilatometry,” Metall. Trans. A, vol. 17A, no. August, pp. 1441–1445, 1986.
[37]
B. Rivolta and R. Gerosa, β€œOn the non-isothermal precipitation of copper-rich phase in 17-4 PH
stainless steel using dilatometric techniques,” J. Therm. Anal. Calorim., vol. 102, no. 3, pp.
857–862, 2010.
[38]
S. Foner, β€œVersatile and Sensitive Vibration-Sample Magnetometer,” Rev. Sci. Instrum., vol. 30,
no. 7, pp. 550–557, 1959.
[39]
S. Foner, β€œThe vibrating sample magnetometer: Experiences of a volunteer (invited),” J. Appl.
Phys., vol. 79, no. 8, p. 4740, 1996.
[40]
β€œDetection setup in a VSM.” [Online]. Available:
http://commons.wikimedia.org/wiki/File:VSM_en.svg. [Accessed: 12-May-2015].
[41]
P. Stamenov and J. M. D. Coey, β€œSample size, position, and structure effects on magnetization
measurements using second-order gradiometer pickup coils,” Rev. Sci. Instrum., vol. 77, no. 1,
p. 015106, 2006.
[42]
A. ZieΜ¨ba, β€œImage and sample geometry effects in SQUID magnetometers,” Rev. Sci. Instrum.,
vol. 64, no. 12, p. 3357, 1993.
[43]
A. Zieba and S. Foner, β€œDetection coil, sensitivity function, and sample geometry effects for
vibrating sample magnetometers,” Rev. Sci. Instrum., vol. 53, no. 9, p. 1344, 1982.
[44]
D. Jiles, Introduction to Magnetism and Magnetic Materials, 2nd ed. Chapman & Hall/CRC,
1998.
[45]
W. D. Callister, Materials Science and Engineering: An Introduction, 6th ed. John Wiley & Sons,
2003.
65
[46]
L. Zhao, N. H. Van Dijk, E. Brück, J. Sietsma, and S. van der Zwaag, β€œMagnetic and X-ray
diffraction measurements for the determination of retained austenite in TRIP steels,” vol. 313,
pp. 145–152, 2001.
[47]
L. Zhao, N. H. Van Dijk, A. J. E. Lefering, and J. Sietsma, β€œMagnetic detection of small fractions
of ferromagnetic martensite within the paramagnetic austenite matrix of TWIP steel,” J.
Mater. Sci., vol. 48, no. 4, pp. 1474–1479, 2013.
[48]
A. S. Arrott and B. Heinrich, β€œApplication of magnetization measurements in iron to high
temperature thermometry,” J. Appl. Phys., vol. 52, no. 3, pp. 2113–2115, 1981.
[49]
A. Bojack, L. Zhao, P. F. Morris, and J. Sietsma, β€œIn-situ determination of austenite and
martensite formation in 13Cr6Ni2Mo supermartensitic stainless steel,” Mater. Charact., vol.
71, pp. 77–86, 2012.
[50]
C. F. Jatczak, J. A. Larson, and S. W. Shin, β€œRetained Austenite and Its Measurements by X-Ray
Diffraction,” Warrendale, PA, 1980.
[51]
R. Delhez, β€œThe volume fraction Austenite in a two-phase steel from XRD measurements and
the standard deviation caused by the counting statistical errors.” TU Delft internal report,
2012.
[52]
R. Hendrikx, β€œSimple improvements in the determination of Austenite content in steel, with Xray diffraction.” TU Delft internal report, 2012.
[53]
A. J. Schwartz, M. Kumar, B. L. Adams, and D. P. Field, Electron Backscatter Diffraction in
Materials Science, 2nd ed. Springer Science + Business Media, 2009.
[54]
β€œElectron
Backscatter
Pattern
FCC
steel.”
[Online].
Available:
http://www.ebsd.com/images/articles/44/figure_4a-10kv_3nA.jpg. [Accessed: 15-Jun-2015].
[55]
D. De Knijf, R. Petrov, C. Föjer, and L. a I. Kestens, β€œEffect of fresh martensite on the stability of
retained austenite in quenching and partitioning steel,” Mater. Sci. Eng. A, vol. 615, pp. 107–
115, 2014.
[56]
M. J. Santofimia, R. H. Petrov, L. Zhao, and J. Sietsma, β€œMicrostructural analysis of martensite
constituents in quenching and partitioning steels,” Mater. Charact., vol. 92, pp. 91–95, Jun.
2014.
[57]
H. Kawata, K. Hayashi, N. Sugiura, N. Yoshinaga, and M. Takahashi, β€œEffect of Martensite in
Initial Structure on Bainite Transformation,” Mater. Sci. Forum, vol. 638–642, pp. 3307–3312,
2010.
[58]
D. De Knijf, E. P. Da Silva, C. Föjer, and R. Petrov, β€œStudy of heat treatment parameters and
kinetics of quenching and partitioning cycles,” Mater. Sci. Technol., vol. 31, no. 7, pp. 817–
828, 2015.
[59]
M. G. Mecozzi, J. Eiken, M. J. Santofimia, and J. Sietsma, β€œPhase field modelling of
microstructural evolution during the quenching and partitioning treatment in low-alloy
steels.” To be submitted.
66
[60]
W. Pitsch, β€œDer Orientierungszusammenhang zwischen Zementit und Ferrit im Perlit,” Acta
Metall., vol. 10, no. 1, pp. 79–80, 1962.
[61]
K. Verbeken, L. Barbé, and D. Raabe, β€œEvaluation of the Crystallographic Orientation
Relationships between FCC and BCC Phases in TRIP Steels,” ISIJ Int., vol. 49, no. 10, pp. 1601–
1609, 2009.
[62]
W. Owen, β€œThe Effect of Silicon on the Kinetics of Tempering,” Trans. Am. Soc. Met., vol. 46,
pp. 812–829, 1954.
[63]
G. Thomas, β€œRetained austenite and tempered martensite embrittlement,” Metall. Trans. A,
vol. 9, no. March, 1978.
[64]
R. Honeycombe and H. K. D. H. Bhadeshia, Steels: Microstructure and properties. ButterworthHeinemann, 2006.
[65]
H. Bhadeshia, Bainite in steels, 2nd ed. Institute of Materials, 2001.
[66]
F. G. Caballero, H. K. D. H. Bhadeshia, K. J. a. Mawella, D. G. Jones, and P. Brown, β€œDesign of
novel high strength bainitic steels: Part 1,” Mater. Sci. Technol., vol. 17, no. 5, pp. 512–516,
2001.
[67]
F. G. Caballero, H. K. D. H. Bhadeshia, K. J. a. Mawella, D. G. Jones, and P. Brown, β€œDesign of
novel high strength bainitic steels: Part 2,” Mater. Sci. Technol., vol. 17, no. 5, pp. 517–522,
2001.
[68]
M. J. Santofimia, F. G. Caballero, C. Capdevila, C. García-Mateo, and C. G. De Andrés, β€œNew
Model for the Overall Transformation Kinetics of Bainite. Part 1: the Model,” Mater. Trans.,
vol. 47, no. 10, pp. 2465–2472, 2006.
[69]
D. Quidort and Y. J. M. Brechet, β€œIsothermal growth kinetics of bainite in 0.5% C steels,” Acta
Mater., vol. 49, no. 20, pp. 4161–4170, 2001.
[70]
A. Stormvinter, β€œLow Temperature Austenite Decomposition in Carbon Steels,” Ph.D.
dissertation, KTH Royal Institute of Technology, Sweden, 2012.
[71]
R. T. Van Tol, β€œMicrostructural evolution in deformed austenitic TWinning Induced Plasticity
steels,” Ph.D. dissertation, Delft University of Technology, The Netherlands, 2014.
[72]
M. Sarikaya, A. K. Jhingan, and G. Thomas, β€œRetained Austenite and Tempered Martensite
Embrittlement in Medium Carbon Steels,” vol. 14, no. June, 1983.
[73]
P. J. Jacques, S. Allain, O. Bouaziz, A. De, A.-F. Gourgues, B. M. Hance, Y. Houbaert, J. Huang,
A. Iza-Mendia, S. E. Kruger, M. Radu, L. Samek, J. Speer, L. Zhao, and S. van der Zwaag, β€œOn
measurement of retained austenite in multiphase TRIP steels β€” results of blind round robin
test involving six different techniques,” Mater. Sci. Technol., vol. 25, no. 5, pp. 567–574, May
2009.
67
Appendices
Appendix A Comparison between retained
determined by VSM and XRD
0.15
0.15
0.12
0.12
0.09
0.09
fractions
as
f
0.18
f
0.18
austenite
0.06
0.06
0.03
0.03
0.00
120
160
200
240
280
320
0.00
120
360
Quenching Temperature (°C)
160
200
240
280
320
360
Quenching Temperature (°C)
(a)
(b)
Figure A-1: Retained austenite fractions of the Q&P specimens as determined by different methods. (a) VSM. (b) XRD.
Figure A-1 shows the retained austenite fractions of Q&P specimens as determined by different
methods. Figure A-1b shows that the standard deviation of the retained austenite fraction is
significantly higher when obtained by XRD compared to magnetic methods in Figure A-1a. However,
the absolute fractions do not differ significantly when comparing both fractions.
Since these methods measure the same quantity, it is possible to average the fractions obtained by
the different methods. The averaging was performed by a weighted average π‘“π‘Ž of the fraction of
retained austenite, with the weight being the standard deviation:
𝑓1
𝑓
β„πœŽ 2 + 2β„πœŽ 2
1
2
π‘“π‘Ž =
πœŽπ‘Ž =
1⁄ + 1⁄
𝜎12
𝜎22
1
√ 1
( ⁄ 2 + 1⁄ 2 )
𝜎1
𝜎2
where 𝑓1 and 𝑓2 are the retained austenite fractions as measured by the different techniques, 𝜎1 and
𝜎2 the respective standard deviations of the measurements, and πœŽπ‘Ž the standard deviation of the
average. The net result is that the fraction and standard deviation are slightly more accurately
measured, as shown for example in Figure 4.1.
However, this method does not take into account any systematic errors between the XRD and VSM
measurements known to exist [73]. Since the differences is measured fraction are small however,
this averaging was judged allowed for specimens that were measured both in the VSM and XRD.
68
Appendix B Stress relief during annealing in a dilatometer
1200
1100 °C - 300 s
Temperature (°C)
1000
800
-1 °C s-1
1 °C s-1
600 °C - 3600 s
600
400
200
0
0
1000
2000
3000
4000
5000
6000
7000
8000
Time (s)
Figure B-1: Heat treatment applied to determine thermal expansion coefficients.
Quenched 1 °C s-1 - Reheat 1 °C s-1
25
Reheated second time
BCC
Area=7.53755E-5
d/dT(L/L0) x10-6 (K-1)
20
15
10
5
Precipitation of carbides
from martensite
0
0
100
200
300
400
500
600
Temperature (°C)
Figure B-2: The thermal expansion of the sample from Figure B-1 during reheating to 600 °C two consecutive times. The
theoretical expansion of BCC iron is plotted as well. Only 1 out of every 5 data points is shown for the two reheatings, for
visual clarity of the fit.
In an attempt to measure the thermal expansion behavior of QP-G, the heat treatment in Figure B-1
was applied. However, eventually the resulting data was used to investigate the difference in thermal
expansion behavior of quenched specimens in this work and the expected theoretical expansion
behavior.
The goal of the heat treatment in Figure B-1 was to measure the thermal expansion behavior of
austenite above 900 °C, to let the formed austenite decompose at 600 °C, and thereafter measure
the expansion behavior of ferrite. No decomposition was detected at 600 °C however, and
69
martensite was formed during the cooling to RT. After in-situ analysis of the generated data, the
sample which was still in place was then reheated multiple times in an attempt to measure the
thermal expansion behavior of ferrite/martensite.
Figure B-2 shows the thermal expansion of this specimen during reheating to 600 °C with 1 °C s-1.
Furthermore, the calculated thermal expansion (see §3.2.3) of pure BCC iron is plotted. During the
first reheating, the thermal expansion deviates significantly from the ideal thermal expansion. Some
retained austenite decomposition is observed around 300 °C. The negative peak around 450 °C
indicates precipitation of carbides from martensite is occurring. This precipitation stops around 525
°C in other quenched samples in this work, but the thermal expansion above 525 °C still deviates
significantly from the calculated thermal expansion indicating further metallurgical processes are
occurring. In the second reheating however, the calculated thermal expansion and the experimental
thermal expansion are nearly identical, indicating no further metallurgical processes are occurring.
The metallurgical processes occurring during the first reheating above 525 °C could be precipitation
of alloy carbides and stress relief, or a combination of both. Precipitation of alloy carbides is certainly
possible in QP-G, since Mn and Mo are present. However, the long soak at 600 °C would have
conceivably formed alloy carbides. This would decrease the likelihood of alloy carbide formation
during subsequent reheating. Therefore, it is hypothesized that stress relief is the main factor causing
the β€œmissing” thermal expansion.
Based on personal conversations with Jilt Sietsma and Peter van Liempt, temperatures above 500 °C
are certainly high enough for dislocation movement and therefore stress relief. Some retained
austenite decomposition is occurring, decreasing the β€œmissing” thermal expansion. The β€œmissing”
thermal expansion is about 7.5 x 10-5 as numerically calculated with Origin and indicated in Figure
B-2. The Young’s modulus of iron between 525 and 600 °C is around 160 GPa. Using Hooke’s law, the
stress relief due to annealing in these particular specimens is 𝜎 = πΈπœ– = 160 × 109 × 7.5 × 10βˆ’5 β‰ˆ
12 MPa. While the sample theoretically should be in complete equilibrium with regards to stress, in
the author’s opinion, this is an approximation of reality. A low stress on the order of 10 MPa could
conceivably have arisen due to martensite formation and the resulting defect and imperfections in
the material.
70
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement