You are on page 1of 13

Materials Science and Technology

ISSN: 0267-0836 (Print) 1743-2847 (Online) Journal homepage: http://www.tandfonline.com/loi/ymst20

Microstructural evolution and mechanical


behaviour of Fe-30Mn-C steels with various carbon
contents

M. Ghasri-Khouzani & J. R. McDermid

To cite this article: M. Ghasri-Khouzani & J. R. McDermid (2017): Microstructural evolution and
mechanical behaviour of Fe-30Mn-C steels with various carbon contents, Materials Science and
Technology, DOI: 10.1080/02670836.2016.1268662

To link to this article: http://dx.doi.org/10.1080/02670836.2016.1268662

Published online: 16 Jan 2017.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=ymst20

Download by: [FU Berlin] Date: 17 January 2017, At: 19:10


MATERIALS SCIENCE AND TECHNOLOGY, 2017
http://dx.doi.org/10.1080/02670836.2016.1268662

Microstructural evolution and mechanical behaviour of Fe-30Mn-C steels with


various carbon contents
M. Ghasri-Khouzani and J. R. McDermid
McMaster Steel Research Centre, McMaster University, 1280 Main Street West, Hamilton, Ontario, Canada L8S 4L7

ABSTRACT ARTICLE HISTORY


A Fe-30Mn-0.6C sheet steel was decarburised and/or annealed to obtain four Fe-30Mn-C alloys Received 19 August 2016
with carbon contents of 0.06, 0.2, 0.4 and 0.6 wt-%. The primary deformation products were found Accepted 28 November 2016
to be mechanical twins for the 0.2C, 0.4C and 0.6C alloys and a combination of mechanical twins KEYWORDS
and ε-martensite for the 0.06C alloy. Both the ε-martensite and mechanical twin formation kinet- High manganese steel;
ics increased sigmoidally with true strain such that the final twin volume fraction increased with Deformation; Twinning;
increasing alloy SFE and C content, where the latter finding disagrees with some of the accepted Martensitic transformation;
models for high-Mn twinning induced plasticity steels. Moreover, the activation stress for twin EBSD; X-ray diffraction;
formation was found to increase linearly with alloy SFE, per a model previously proposed by the Mechanical behaviour;
present authors. Stacking fault energy

Introduction
Hall–Petch effect. However, other authors [18,19] have
Efforts to meet government-mandated vehicle fuel effi- attributed the Portevin–Le Chatelier (PLC) or dynamic
ciency targets through body-in-white weight reduc- strain ageing (DSA) effect as being responsible for the
tions, while maintaining or enhancing passenger safety, high work hardening rates observed for these steels.
have increased the demand for advanced high-strength The value of the stacking fault energy (SFE) plays
steels for automotive structural components [1]. To ful- a critical role in determining whether the deforma-
fil this objective, high manganese (Mn) steels with Mn tion products in high-Mn austenitic steels are either
contents of 12–33 wt-% and carbon content of 0–1.2 ε-martensite or twins, where the SFE is a strong func-
wt-% – often with additions of aluminium and/or sil- tion of alloy chemistry and temperature [4,9,20,21].
icon – have been developed [2–8]. These steels can Several models have been proposed for computing the
exhibit either the transformation induced plasticity SFE and predicting the deformation products in high-
(TRIP) and/or the twinning induced plasticity (TWIP) Mn steels [4,22–27]. The model of Allain et al. [22]
effects, where the latter leads to significantly higher predicts that the γ → ε mechanical martensitic trans-
ductility owing to the high sustained work hardening formation would take place if the SFE was lower than
rates observed [4,9]. 18 mJ/m2 , whereas mechanical twins would form when
For high-Mn TRIP steels, the austenitic matrix par- the SFE lay between 12 and 35 mJ/m2 . The more recent
tially transforms to ε-martensite during plastic defor- extension of the Allain et al. model by Dumay et al.
mation. The high work hardening rates observed for [23] predicts that a lower SFE (i.e. < 18 mJ/m2 ) favours
these steels are largely explained by the dynamic forma- mechanical ε-martensite formation, while a higher SFE
tion of the highly dislocated HCP ε-martensite phase, ( ≥ 18 mJ/m2 ) favours mechanical twinning. In addi-
which is also accompanied by the creation of numerous tion, according to the SFE model proposed by Saeed-
geometrically necessary dislocations at the austenite/ε- Akbari et al. [26], the γ → ε mechanical martensitic
martensite interface [10,11]. In the case of the high-Mn transformation would take place when the SFE was
TWIP steels, a significant population of mechanical lower than 20 mJ/m2 . However, this model did not pro-
twins develops within the austenite matrix during plas- vide an SFE value range for the formation of mechanical
tic deformation. The mechanism by which these alloys twins. In spite of some inconsistencies in the SFE ranges
achieve their sustained high work hardening rates con- predicted by the available models for the formation
tinues to be a subject of some debate in the literature, of the ε-martensite and mechanical twin deformation
where numerous authors [7,9,12–17] have attributed products, there is general agreement that mechani-
the sustained high work hardening rates observed to cal ε-martensite formation was replaced by mechan-
dynamic twin formation, decreasing the mean free path ical twinning when the SFE value reaches a critical
for dislocation motion – i.e. the so-called dynamic level.

CONTACT J. R. McDermid mcdermid@mcmaster.ca

© 2017 Institute of Materials, Minerals and Mining. Published by Taylor & Francis on behalf of the Institute.
2 M. GHASRI-KHOUZANI AND J. R. MCDERMID

Carbon plays a variety of roles in high-Mn Fe-Mn-C average grain size of the as heat-treated alloys, where
steels. In addition to the role of C in the solid solution seven fields within the optical micrographs were anal-
strengthening of austenitic steels and its role in chemi- ysed for each alloy. A LECO HF-400 carbon combus-
cally stabilising austenite, it is generally agreed that the tion analyser was used to determine the bulk carbon
SFE value of high-Mn steels increases with increasing content of the as heat-treated alloys. To construct car-
alloy carbon content [22–24,26,28], although the rate of bon concentration profiles in the through thickness
increase was not consistent within these contributions. direction to determine the homogeneity of the C distri-
Increased alloy carbon content has also been reported bution, successive sample surface layers were removed
to be linked to the appearance of serrations in the ten- by grinding and carbon analyses performed as a func-
sile curve for high-Mn TWIP steels with more than 0.4 tion of the sample through thickness using a Jobin
wt-% C [9,13–15,29–33]. Yvon Glow Discharge Optical Emission Spectrograph
Although numerous contributions on the micro- (GDOES), where the GDOES was calibrated using cer-
structure, deformation products and mechanical tified standards. In this way, it was determined that the
behaviour of various high-Mn Fe-Mn-C alloys can C distribution was homogeneous through the sample
be found in the literature [2,6,12–15,29,34,35], only a thickness for all samples.
limited number have focused on the effect of carbon Room temperature uniaxial tensile testing was car-
content on the microstructural and mechanical prop- ried out at a crosshead speed of 1 mm/min (strain
erty evolution while keeping other compositional and rate of approximately 10−3 s−1 ) using a 10 kN Instron
microstructural parameters constant [9,36,37]. Thus, 5566 tensile frame. Sub-size tensile samples were cut
the main objective of the current study is to determine from the heat-treated coupons using electric discharge
the influence of alloy carbon content on the microstruc- machining (EDM) with dimensions in accordance with
tural evolution and mechanical properties of a series of ASTM E8M [39]. Tensile tests were conducted until
Fe-30Mn-C alloys as a function of applied deformation, fracture to determine the overall sample tensile charac-
where the carbon contents ranged from approximately teristics or were interrupted at defined true strains for
0.06 to 0.6 wt-%. An additional objective of this study microstructural evolution analysis. In all cases, a 25 mm
is to extend and enhance some of the general models of gauge length extensometer was employed to monitor
the mechanical behaviour of high-Mn steels previously sample elongation. The sample uniform elongation was
reported by the present authors [9] with the data from determined from the true stress-true strain plots using
this contribution. the Considère criterion:

Experimental procedure dσ
= σ, (1)

The starting material was cold-rolled Fe-30Mn-0.6C
steel sheet with a thickness of 1.6 mm. A series of Fe- where σ and ε denote true stress and true strain, respec-
30Mn-C steels with carbon contents of approximately tively.
0.06, 0.2 and 0.4 wt-% were obtained through decar- Since ε-martensite and twins are microstructurally
burisation heat treatments of the as-received steels, similar and cannot be differentiated by conventional
where wet hydrogen was used as the decarburising SEM or optical microscopy, Electron Back-Scattered
agent. To achieve homogenous alloys with respect to Diffraction (EBSD) was carried out using a JEOL JSM-
carbon distribution and grain size, annealing heat treat- 7000 FEG–SEM to analyse the microstructural evolu-
ments under an argon atmosphere were also applied. tion of the sample cross-sections before, during and
Both decarburisation and annealing heat treatments after deformation. An acceleration voltage of 20 keV
were performed in a horizontal tube furnace, after and a working distance of 18.4 mm were used for all
which samples were quenched in oil. Heat treat- SEM analyses. The specimen was tilted 70° towards
ment specimens comprised 102 mm × 36 mm coupons. the EBSD detector to increase the backscattered elec-
Table 1 presents the heat treatment parameters for all tron yield. The magnification was 400 × in all cases.
experimental alloys. Optical microscopy and the ASTM For all EBSD phase maps, the step size was 0.4 μm
linear intercept method [38] were used to measure the to obtain sufficient resolution within a relatively large
microstructural region. HKL Channel 5 software was
used for the operation and control of all EBSD analy-
Table 1. Heat treatment parameters for all experimental alloys ses as well as analysis of the EBSD results. In prepar-
Decarburisation Annealing Decarburisation ing the EBSD specimens, sample cross-sections were
time (by wet H2 time (by Ar and annealing initially prepared using standard metallographic pro-
Alloy name (5.2%)+Ar (94.8%)) (99.999%)) temperature
cedures followed by Ar ion milling with a JEOL IB-
0.06C 7h 0 1000°C
0.2C 2 h + 50 min 4 h + 10 min 1000°C 09010 Cross-section Polisher (CP) to eliminate any
0.4C 50 min 6 h + 10 min 1000°C residual surface stresses which may adversely affect the
0.6C 0 7h 1000°C
local transformation products. To determine the twin
MATERIALS SCIENCE AND TECHNOLOGY 3

volume fraction from the EBSD phase maps, the point the models of Dumay et al. [23], Nakano and Jacques
counting method proposed by Renard and Jacques [7] [24] and Saeed-Akbari et al. [26], as summarised in
was used. Table 2. For all the alloys except the 0.06C alloy, the
The volume fraction of ε-martensite in the as heat- measured carbon contents were very close to the tar-
treated and deformed samples was quantitatively deter- get levels. In the case of the SFE values computed using
mined via X-ray diffraction using a Bruker D8 Discover the Dumay et al. [23] model, the deformation prod-
diffractometer. The X-ray beam was generated from a ucts were predicted to be mechanical twins for the 0.2C,
Co source with a Kα line wavelength of 1.79026 Å using 0.4C and 0.6C alloys and a combination of mechani-
a voltage and current of 35 keV and 45 mA, respectively. cal twins and ε-martensite for the 0.06C alloy. The SFE
Spectra were taken in the 2θ range from 36° to 120° with model of Nakano and Jacques [24] predicted mechan-
a 0.03° step size and a dwell time of 1.3 s per step. Sample ical ε-martensite formation for the 0.06C alloy with-
cross-sections were analysed using a 1.0 mm diameter out providing any predictions for mechanical twinning
aperture in all cases. The sample was moved back and for the remaining alloys. The SFE model proposed by
forth along its length and also rotated continuously to Saeed-Akbari et al. [26] did not predict mechanical ε-
mitigate any preferred orientation effects on the analy- martensite formation for any of the experimental alloys.
sis. Each analysis was run in triplicate for each sample It should be noted that the results listed in Table 2 show
and the average of the results presented. that the Dumay et al. model [23] showed a stronger
effect of alloy carbon content on the SFE value com-
Results pared to that of other models.
The average grain size of the as heat-treated samples
As heat-treated alloy carbon content and was 104 ± 4 μm, 101 ± 2 μm, 110 ± 3 and 108 ± 4 μm
microstructure for the 0.06C, 0.2C, 0.4C and 0.6C alloys, respectively.
The carbon contents of the as heat-treated experimen- From these data, it can be concluded that there were no
tal alloys were used to compute their SFE values using significant differences in alloy grain size. Typical EBSD
phase maps from the cross-sections of the as heat-
treated alloys are shown in Figure 1. For all EBSD phase
Table 2. Carbon content and calculated SFE values for all exper-
imental alloys at room temperature maps presented, it should be noted that FCC austenite is
denoted in red, HCP ε-martensite in blue, twin bound-
Carbon
Alloy content SFE (mJ/m2 ) SFE (mJ/m2 ) SFE (mJ/m2 ) aries in yellow and un-indexed areas black. Moreover,
Name (wt-%) by Ref. [23] by Ref. [24] by Ref. [26] grain boundaries are shown as black lines. As can be
0.06C 0.06 17.9 35.3 28.4 seen, the as heat-treated alloy microstructures com-
0.2C 0.20 23.5 37.8 31.1 prised an equiaxed, fully austenitic matrix with some
0.4C 0.41 29.6 40.6 35.1
0.6C 0.59 33.2 43.4 38.5 annealing twins and no detectable ε-martensite.

Figure 1. Typical EBSD phase maps from the cross-section microstructure of the as heat-treated alloys; (a) 0.06C, (b) 0.2C, (c) 0.4C
and (d) 0.6C.
4 M. GHASRI-KHOUZANI AND J. R. MCDERMID

serrations, which can be attributed to the PLC or DSA


effect [31,33].
To assess the effect of increasing alloy Mn content
on alloy yield stress, the data for the present 0.4C and
0.6C alloys (Table 3) were compared to the yield stresses
of the Fe-22Mn-0.4C (223 MPa) and Fe-22Mn-0.6C
(265 MPa) alloys previously documented by the present
authors [9], where the grain sizes, phase make-up and
carbon contents were insignificantly different across the
two sets of alloys. From this comparison, it can be seen
that the yield stress increased with increasing alloy Mn
level. This result disagrees with the previously discussed
model of Bouaziz et al. [29] for the effect of compo-
sition on the yield stress of high-Mn Fe-Mn-C steels,
where the yield strength was predicted to decrease with
increasing alloy Mn content. For the present as heat-
treated 0.06C and 0.2C alloys, it should be noted that
their properties cannot be directly compared to those of
their Fe-22Mn-C counterparts, as the latter contained
significant volume fractions of ε-martensite [9].
It can also be seen from Figure 2(a) and Table 3 that
the ultimate tensile strength (UTS) of the experimental
alloys increased with increasing alloy carbon content,
which can be attributed to the sustained high work
hardening rates exhibited by all the alloys (Figure 2(b))
as well as the solid solution strengthening effect of car-
bon. From Figure 2(a) and Table 3, it can be seen that
the uniform elongation of the experimental alloys did
not significantly change with the alloy C level except in
the case of the 0.6C alloy, which had a slightly higher
Figure 2. (a) Representative room temperature true elongation. Macroscopically, this occurred because the
stress–strain plots of the experimental alloys. (b) Corresponding 0.6C alloy exhibited the highest sustained work hard-
work hardening rate versus true stress. ening rates (i.e. dσ /dε) of the experimental alloys
(Figure 2(b)) owing to the dynamic grain refinement
Mechanical properties effect of the mechanical twins (i.e. reduction of dislo-
cation mean free path). As will be discussed below, this
Representative room temperature true stress–true
was also likely due to the 0.6C alloy having a higher
strain curves for all experimental alloys are provided
volume fraction of twins in the microstructure versus
in Figure 2(a) and the derived tensile characteristics
the other alloys. However, the apparent lack of benefit
for all alloys listed in Table 3. All tensile curves dis-
from the increased volume fraction of twins can likely
played continuous yielding. Furthermore, it can be seen
be attributed to the critical damage process in the lower
that alloy yield stress (YS) increased with increasing
C alloys not being dominated by either ε-martensite or
alloy carbon content (Figure 2(a), Table 3). This can be
twin boundary void nucleation, but being controlled by
explained by the solid solution strengthening effect of
the normal processes found in austenitic alloys – e.g.
carbon in austenite [29,40] and is in qualitative agree-
void nucleation at grain boundaries.
ment with the model proposed by Bouaziz et al. on
Inspection of Figure 2 will show that the experimen-
the effect of composition on the yield strength of high-
tal alloys displayed distinctly different work harden-
Mn steels [29]. As can be seen in Figure 2(a), the flow
ing histories, where the 0.06C and 0.2C alloys showed
curves of the 0.4C and 0.6C alloys showed significant
relatively low work hardening rates (i.e. dσ /dε) ver-
sus the 0.4C and 0.6C alloys and which continuously
Table 3. Tensile characteristics of the experimental alloys at declined with increasing stress. These work harden-
room temperature
ing histories are consistent with that observed by Liang
Yield stress Uniform et al. [6] for a Fe-30Mn-0.02C alloy (wt-%), in which
Alloy name (MPa) UTS (MPa) elongation
dislocation glide was determined to be the dominant
0.06C 153 ± 3 775 ± 7 0.48 ± 0.03
0.2C 174 ± 4 892 ± 8 0.47 ± 0.02 deformation mode. However, the 0.4C and 0.6C alloys
0.4C 251 ± 6 1165 ± 10 0.50 ± 0.03 exhibited several distinct changes in their dσ /dε values
286 ± 5 1457 ± 9 0.61 ± 0.04
0.6C
typical of true TWIP alloys, where the dσ /dε curve
MATERIALS SCIENCE AND TECHNOLOGY 5

for the 0.4C alloy initially declined with increasing Figure 4 presents cross-sectional EBSD phase maps
stress and strain until σ ≈ 570 MPa (ε ≈ 0.15) before as a function of true strain for the 0.06C and 0.2C alloys.
attaining a maximum at σ ≈ 890 MPa (ε ≈ 0.33), after For both alloys, the microstructures at low strains
which the alloy work hardening rate declined to frac- (ε ≤ 0.1, Figure 4(a) and (e) for the 0.06C and 0.2C
ture (Figure 2). Similarly, the dσ /dε values for the alloys, respectively) comprised an equiaxed austenite
0.6C alloy initially declined with increasing stress until matrix with some annealing twins. These microstruc-
σ ≈ 660 MPa (ε ≈ 0.18), after which they increased tures were insignificantly different from those of the
until σ ≈ 1050 MPa (ε ≈ 0.38), followed by declining as heat-treated 0.06C and 0.2C alloys, as illustrated in
work hardening rates to fracture. This behaviour is con- Figure 1(a) and (b), respectively, and are consistent
sistent with stages A through C of the work hardening of with the XRD results shown in Figure 3 for the 0.06C
a Fe-20Mn-1.2C TWIP steel described by Renard and alloy. The first deformation products were observed as
Jacques [7] and is consistent with the work hardening the true strain approached ε = 0.2 (Figure 4(b) and
behaviour generally observed for high-Mn TWIP steels (f)). For the 0.06C alloy, the deformation products
[9,13,15,17,31–33,37,41–43]. (Figure 4(b–d)) were small volume fractions of both
mechanical twins and ε-martensite, where the latter
was consistent with the XRD data shown in Figure 3. In
Microstructural evolution with tensile deformation
the case of 0.2C alloy, however, mechanical twins were
To determine the deformation products in the micro- the only significant deformation products observed
structures, EBSD and XRD analyses were performed on (Figure 4(f–h)).
the experimental alloys as a function of plastic strain. EBSD phase maps from the cross-sections of the
Twins were observed in all of the alloys in varying vol- 0.4C and 0.6C alloys as a function of applied strain are
ume fractions, whereas significant volume fractions of presented in Figure 5. For true strains of up to ε ≈ 0.1,
ε-martensite were only observed in the 0.06C alloy. A the microstructures contained no significant deforma-
plot of ε-martensite volume fraction versus true strain tion products and comprised an austenitic matrix with
for the 0.06C alloy is shown in Figure 3, where the some annealing twins, as illustrated in Figure 5(a) and
plot error bars represent the 95% confidence intervals (e) for the 0.4C and 0.6C alloys, respectively. For both
for the average ε-martensite volume fraction. For the alloys, mechanical twins appeared in the microstruc-
un-deformed 0.06C alloy microstructure (at ε = 0), tures as the true strain approached 0.2 (Figure 5(b) and
the ε-martensite volume fraction was approximately (f)). The twin population then increased sharply with
0.025. This observation is reasonably consistent with increasing strain, as can be seen in Figure 5(c) and (d)
the microstructural observations of the as heat-treated for the 0.4C alloy and Figure 5(g) and (h) for the 0.6C
0.06C alloy illustrated in Figure 1. From Figure 3, it can alloy. It should be noted that no significant ε-martensite
be seen that the ε-martensite volume fraction increased was observed in the 0.4C and 0.6C alloys for any strain.
with increasing true strain in a sigmoidal fashion. It can Furthermore, comparison of Figs. 4 and 5 will show that
also be seen that the rate of γ → ε martensitic trans- the twin volume fractions at higher strains were signif-
formation declined for ε ≥ 0.3. These observations are icantly greater for the 0.4C and 0.6C alloys versus that
qualitatively consistent with the Fe-22Mn-0.06C and for the 0.06C and 0.2C alloys.
Fe-22Mn-0.2C alloy ε-martensite formation kinetics Further examination of Figs. 4(e–h) and 5 will reveal
previously reported by the present authors [9]. that the twin population for different grains was not
the same, which is consistent with general findings in
the literature concerning the effect of the local Schmid
factor on twin activation [8,13,44,45]. Based on the
EBSD phase maps in Figs. 4(e–h) and 5, the dominant
deformation product for the 0.2C, 0.4C and 0.6C alloys
was determined to be mechanical twins. The latter two
results are in agreement with the deformation products
reported by Wang et al. for a Fe-30Mn-0.5C alloy ten-
sile tested at room temperature [46]. It should be noted
that for all experimental alloys, the observed deforma-
tion products generally confirmed the predictions of
the Dumay et al. SFE model [23].
The point counting method of Renard and Jacques
[7] was used with the EBSD phase maps to deter-
mine the volume fraction of twins as a function of
alloy C content and applied strain for all experimental
Figure 3. Evolution of ε-martensite volume fraction with ten- alloys, the results of which are shown in Figure 6. From
sile strain for the 0.06C alloy, measured via XRD. Figure 6, it can be seen that the twin volume fraction
6 M. GHASRI-KHOUZANI AND J. R. MCDERMID

Figure 4. EBSD phase maps of cross-sections for the 0.06C and 0.2C alloys as a function of true strain: (a) 0.06C, ε = 0.1; (b) 0.06C,
ε = 0.2; (c) 0.06C, ε = 0.3; (d) 0.06C, ε = 0.44; (e) 0.2C, ε = 0.1; (f) 0.2C, ε = 0.2; (g) 0.2C, ε = 0.3; (h) 0.2C, ε = 0.43.

for all alloys was very low ( ≈ 0.03) for strains of less that the volume fraction of mechanical twins at fracture
than 0.1. This can be explained by invoking a criti- increased with increasing C content. All of these obser-
cal stress for twin nucleation, which has been shown vations are consistent with the microstructural trends
by several authors to be proportional to the alloy SFE observed in Figs. 4 and 5.
[9,43,44]. This correlation, within the context of the Comparison of the twinning kinetics in Figure 6 to
present alloys and previous literature, will be discussed the mechanical properties and work hardening rates
below. For all of the alloys, the twinning kinetics fol- shown in Figure 2 and Table 3 will show that the work
lowed a sigmoidal rate, with the rate of mechanical twin hardening rates (and consequent mechanical prop-
formation being relatively high for strains of 0.1 to 0.3, erties) were directly proportional to the trends seen
after which the rate of twin formation decreased as the in the relative volume fractions of twins and rate of
alloys approached twin saturation. It can also be seen twin formation, as would be expected for the case
MATERIALS SCIENCE AND TECHNOLOGY 7

Figure 5. EBSD phase maps of cross-sections for the 0.4C and 0.6C alloys as a function of true strain: (a) 0.4C, ε = 0.1; (b) 0.4C,
ε = 0.2; (c) 0.4C, ε = 0.3; (d) 0.4C, ε = 0.49; (e) 0.6C, ε = 0.1; (f) 0.6C, ε = 0.2; (g) 0.6C, ε = 0.3; (h) 0.6C, ε = 0.45.

of twin boundaries decreasing the dislocation mean (Figure 6) will show that the total volume fraction of
free path through dynamic refinement of the grain deformation products in the microstructure was less
structure. This finding is consistent with the so-called than 0.3 for all strains, indicating that the dominant
dynamic Hall–Petch effect explanation for the mechan- deformation mechanism for the 0.06C alloy was dis-
ical behaviour of high-Mn TWIP steels advocated by a location glide. These findings are in general agreement
number of authors [7,9,12–17,22,29]. with those reported by Liang et al. for a Fe-30Mn-0.02C
alloy subjected to uniaxial tensile deformation at 293
K [6]. A mixture of ε-martensite and mechanical twin
Discussion
deformation products has been previously reported
In the case of the 0.06C alloy, an examination of the for a Fe-22Mn-0.4C alloy by the present authors [9],
EBSD phase maps (Figure 4(a–d)) in addition to the where the SFE for the Fe-22Mn-0.4C alloy (15.0 mJ/m2
ε-martensite (Figure 3) and twinning kinetics data by the Dumay et al. model [23], to be used for all
8 M. GHASRI-KHOUZANI AND J. R. MCDERMID

and 0.6C alloys, with no significant ε-martensite being


detected within the microstructures. From Figure 6,
it can be seen that the onset of mechanical twinning
occurred between ε = 0.1 and ε = 0.2 for both the
alloys. This onset strain was slightly higher than that
reported for the Fe-20Mn-1.2C [7], Fe-18Mn-0.6C-
1.5Al [17], Fe-22Mn-0.4C and Fe-22Mn-0.6C [9] alloys
within the literature. As previously reported [7,9], the
activation stress for mechanical twinning can be deter-
mined using the inflection point associated with the
first increase in the alloy dσ /dε value from the work
hardening curves. From Figure 2(b), this was deter-
mined to be at ε = 0.15 (σ = 570 MPa) and ε = 0.18
(σ = 660 MPa) for the present 0.4C and 0.6C alloys,
respectively.
Figure 6. Evolution of twin volume fraction with true strain for The activation stress for the onset of mechanical
all experimental alloys. twinning for high-Mn Fe-Mn-C steels, where twinning
is the dominant deformation product, has been shown
subsequent SFE computations unless otherwise noted) by the present authors to increase linearly with increas-
was slightly lower than that of the present Fe-30Mn- ing alloy SFE [9] (computed using the model of Allain
0.06C alloy (17.9 mJ/m2 ). For both the Fe-22Mn-0.4C et al. [22]), as proposed by Byun [47] for austenitic
and Fe-30Mn-0.06C alloys, the final ε-martensite vol- stainless steels. These data are re-plotted in Figure 7(a)
ume fraction was approximately 0.12. However, a larger using the more recent multicomponent SFE model of
final volume fraction of mechanical twins ( ≈ 0.25) was Dumay et al. [23]. This relationship is further explored
observed for the Fe-22Mn-0.4C alloy versus that for the and expanded in Figure 7(b), which includes the acti-
present 0.06C alloy ( ≈ 0.14). vation stress for mechanical twinning for the present
For the 0.2C alloy, the twin volume fraction 0.4C and 0.6C TWIP steels. It should be noted that
increased sigmoidally with increasing deformation the revised plot includes the previously presented data
with a trend similar to that observed for the 0.06C alloy derived from Renard and Jacques [7], Ghasri-Khouzani
(Figure 6). It should be noted, however, that the twin and McDermid [9], Jin and Lee [17] and Yang [48]. It
volume fraction was uniformly higher versus that of the was found that the data in Figure 7 obeyed the linear
0.06C alloy for all strains greater than 0.1, with the final relationship:
twin volume fraction in the 0.2C alloy being approxi-
mately 0.20. This explains the higher work hardening SFE
rate observed for the 0.2C alloy compared to that for σt = (2.95 ± 0.07) r2 = 0.99
bp
the 0.06C alloy (Figure 2(b)), which led to the higher
observed UTS for the 0.2C alloy (Figure 2(a) and Table SFE ≥ 13.6 mJ/m2 , (2)
3). The higher twin volume fraction of this alloy com-
pared to that of the 0.06C alloy can be explained by con- where σ t is the activation stress for the onset of
sidering their relative SFE values, where this increased mechanical twin formation (Pa),  SFE (J/m2 ) the SFE
from 17.9 mJ/m2 for the 0.06C alloy to 23.5 mJ/m2 and bp (m) the magnitude of the Burgers vector for the
for the 0.2C alloy. However, the increased twin vol- Shockley partial dislocation, calculated to be 0.144 nm
ume fraction of the 0.2C alloy was not sufficient to based on the previously reported magnitude of the
prevent a continuous decease in the work hardening Burgers vector in TWIP steels of 0.25 nm [14]. Compar-
rate during the deformation of this alloy, as observed ison of Figure 7(a) to 7(b) will show that the addition of
in Figure 2(b). No ε-martensite was observed in the the present data did not significantly change the previ-
microstructure of this alloy for all applied strains, as ously derived linear relationship. Thus, it can be said
shown in (Figure 4(e–h)). Thus, it can be concluded that the onset stress for mechanical twinning under
that dislocation glide was also the dominant deforma- uniaxial tension, and the consequent increase in work
tion mode for this alloy and that, furthermore, the twin- hardening rate for high-Mn TWIP steels, is linearly
ning kinetics and saturated twin volume fraction were dependent on the alloy SFE as described using eq. (2).
insufficient to produce the dynamic Hall–Petch effect However, it should be noted that eq. (2) is applicable
typical of that desired in true TWIP alloys. This resulted only to the case of high-Mn TWIP steels where the alloy
in the tensile behaviour of this alloy being similar to that SFE values are computed using the model of Dumay
of the 0.06C alloy (see Figire 2). et al. [23]. When using other SFE models, the constant
From Figure 5, it can be seen that mechanical twins in eq. (2) may be different as was the case in the previous
were the dominant deformation products for the 0.4C analysis of this relationship by the present authors [9].
MATERIALS SCIENCE AND TECHNOLOGY 9

Figure 7. Activation stress for mechanical twinning versus the SFE for various high-Mn steels: (a) data of Ghasri-Khouzani and
McDermid [9] re-plotted using the SFE equation of Dumay et al. [23], (b) Figure 7(a) with addition of data from present study.

For both the 0.4C and 0.6C alloys, the rate of twin
formation was highest for the 0.2–0.3 true strain range,
while the rate of twin formation decreased significantly
for higher strains (Figure 6). This decrease in the rate
of twin formation corresponded to a drop in the work
hardening rate for these alloys, as shown in Figure 2(b).
The decrease in twin formation rate (and correspond-
ing drop in work hardening rate) was reported to occur
at ε ≥ 0.3 for Fe-20Mn-1.2C [7], Fe-22Mn-0.4C and
Fe-22Mn-0.6C steels [9] and at ε ≥ 0.37 for a Fe-18Mn-
0.6C-1.5Al alloy [17]. The correspondence between the
decline in the rate of twin formation and the evolution
of the work hardening rate in the present contribution
and similar studies within the authors’ research group
[9,37,48,49] agree with the studies of other authors
[7,12–17,22,29], who attributed the high work harden- Figure 8. Final twin volume fraction as a function of SFE for
ing rates of high-Mn TWIP steels to the role of mechan- various high-Mn alloys.
ical twins in decreasing the mean free path for disloca-
tion motion – also widely referred to as the dynamic
Hall–Petch effect. It should be noted that these find- in Figure 8. It can be seen that the final volume frac-
ings disagree with those of authors who attribute the tion of twins generally increased with increasing SFE;
observed high work hardening rates in high-Mn TWIP however, the increase was not significant for higher
steels to the DSA or PLC effect [18,19]. SFE values. This can be explained by the suppression
From Figure 6, it can be seen that the final volume of mechanical twinning for high-Mn steels with rela-
fraction of mechanical twins increased with increas- tively high SFE values ( > 33 mJ/m2 ) and the increasing
ing alloy carbon level. Furthermore, the final volume dominance of dislocation glide as the primary deforma-
fraction of twins for the 0.4C and 0.6C alloys was tion mode [50]. The 0.06C and 0.2C alloys did not obey
higher than that of their previously studied Fe-22Mn- the trend presented in Figure 8 because they deformed
C counterparts [9], indicating that the final population mainly through dislocation glide despite their relatively
of mechanical twins also increased with increasing alloy low SFE.
manganese level. Furthermore, using the SFE model of However, it must be noted that the trend observed
Dumay et al. [23], the SFE of the previously studied Fe- for the twinning kinetics of the present alloy system
22Mn-C series of TWIP alloys was systematically lower – i.e. the increase in twinning kinetics with increasing
than those of their present 30Mn counterparts, indi- alloy SFE documented in Figure 6 – is not in agree-
cating that the final twin volume fraction may be pro- ment with the model proposed by Bouaziz and Guelton
portional to the alloy SFE. This relationship is explored [15], where the twin formation rate was predicted to
for the present system and others from the literature decrease with increasing alloy SFE. However, it should
10 M. GHASRI-KHOUZANI AND J. R. MCDERMID

be noted that the Bouaziz and Guelton model was based This was likely due to either ε-martensite or twin
on the relative stacking fault energies of an AISI 304L boundaries not being the dominant damage for-
stainless steel and a Fe-22Mn-0.6C TWIP steel and not mation site during the tensile deformation of the
on variations in SFE within the Fe-Mn-C TWIP family lower C alloys
of alloys itself [15]. Furthermore, this trend in the rate (v) The evolution of twin volume fraction with true
of mechanical twin formation increasing with increas- strain followed a sigmoidal kinetic for all the exper-
ing Fe-Mn-C TWIP alloy SFE had previously been imental alloys, where the twinning kinetics and
observed for the family of Fe-22Mn-C alloys previously twin volume fraction for a given stress or strain
studied by the present authors [9]. Reasons for this dis- increased with increasing alloy SFE and C con-
agreement with predictions of the Bouaziz and Guelton tent. Reasons for disagreement of this trend with
model [15] are not known at this time and should be the the predictions of the Bouaziz and Guelton model
subject of further study, but a possible hypothesis for should be the subject of further investigations, but
this disagreement may be the extensive secondary twin- may have arisen from the effects of secondary twin
ning observed in both the 22Mn and 30Mn TWIP alloy formation
systems which may not be accounted for in the Bouaziz (vi) For the higher SFE 0.4C and 0.6C TWIP alloys,
and Guelton model. For example, it has been observed the activation stress for the start of mechani-
by Lü et al. [49], that secondary twinning is a significant cal twinning increased linearly with alloy SFE.
component of mechanical twinning in a Fe-22Mn-0.6C This relationship was consistent with the empiri-
TWIP steel where the secondary twinning kinetics are cal equation previously derived for other high-Mn
a strong function of strain path. These secondary twins TWIP alloys taken from the open literature.
would contribute fresh obstacles for dislocation move-
ment and result in increased work hardening rates until Acknowledgements
the twin fraction approached saturation. The Resource for the Innovation of Engineered Materi-
als (RIEM) programme of CANMETMaterials is gratefully
acknowledged for fabrication of the Fe-30Mn-0.6C sheet
Conclusions
steels used in the experiments. All electron microscopy for
By investigating the microstructural evolution as a this contribution was carried out in the Canadian Centre
for Electron Microscopy (CCEM) and all X-ray Diffrac-
function of carbon content and strain for a series of
tion was carried out within the McMaster Analytical X-Ray
Fe-30Mn-C alloys, the following can be concluded: Diffraction Facility (MAX) within the Brockhouse Institute
for Materials Research.
(i) The deformation products for the 0.06C alloy con-
sisted of minor volume fractions of both mechan- Funding
ical ε-martensite and twins, leading to an alloy
This work was supported by Natural Sciences and Engineer-
exhibiting a lower work hardening rate likely dom- ing Research Council of Canada (NSERC Discovery).
inated by dislocation glide. However, the observed
work hardening rates were high compared to those ORCID
of higher-SFE materials, such as Al and its alloys
J. R. McDermid http://orcid.org/0000-0002-0361-7829
(ii) The deformation products for the 0.2C alloy were
determined to be a relatively small volume fraction
of mechanical twins, with dislocation glide being References
the dominant deformation mode for this alloy [1] European Commission. The Strategic Research Agenda
(iii) The deformed microstructures of both the 0.4C of the European Steel Technology Platform (ESTEP),
Belgium, 2005.
and 0.6C alloys revealed significant volume frac-
[2] Barbier D, Favier V, Bolle B. Modeling the deformation
tions of mechanical twins as their dominant textures and microstructural evolutions of a Fe–Mn–C
deformation products, leading to alloys displaying TWIP steel during tensile and shear testing. Mater Sci
higher sustained work hardening rates and giv- Eng A. 2012;540:212–225.
ing rise to the conclusion that mechanical twin- [3] Bayraktar E, Khalid FA, Levaillant C. Deformation and
ning was the critical deformation mode leading fracture behaviour of high manganese austenitic steel. J
Mater Process Tech. 2004;147:145–154.
to true TWIP behaviour for these alloys with the [4] Grässel O, Frommeyer G. Effect of martensitic phase
overall deformation behaviour continuing to be transformation and deformation twinning on mechan-
dominated by dislocation glide ical properties of Fe–Mn–Si–AI steels. Mater Sci Tech.
(iv) The yield stress and tensile strength of the exper- 1998;14:1213–1217.
imental alloys increased with increasing the alloy [5] Jeong K, Jin J-E, Jung Y-S, et al. The effects of Si on the
mechanical twinning and strain hardening of Fe-18Mn-
carbon content and alloy SFE, whereas the uni-
0.6 C twinning-induced plasticity steel. Acta Mater.
form elongation was not significantly affected by 2013;61:3399–3410.
the alloy C content except for the 0.6C alloy, which [6] Liang X, McDermid JR, Bouaziz O, et al. Microstruc-
displayed a slightly higher uniform elongation. tural evolution and strain hardening of Fe–24Mn and
MATERIALS SCIENCE AND TECHNOLOGY 11

Fe–30Mn alloys during tensile deformation. Acta Mater. [25] Remy L, Pineau A. Twinning and strain-induced
2009;57:3978–3988. FCC → HCP transformation in the Fe-Mn-Cr-C sys-
[7] Renard K, Jacques PJ. On the relationship between work tem. Mater Sci Eng. 1977;28:99–107.
hardening and twinning rate in TWIP steels. Mater Sci [26] Saeed-Akbari A, Imlau J, Prahl U, et al. Deriva-
Eng A. 2012;542:8–14. tion and variation in composition-dependent stack-
[8] Yang P, Xie Q, Meng L, et al. Dependence of deformation ing fault energy maps based on subregular solution
twinning on grain orientation in a high manganese steel. model in high-manganese steels. Metall Mater Trans A.
Scripta Mater. 2006;55:629–631. 2009;40:3076–3090.
[9] Ghasri-Khouzani M, McDermid JR. Effect of carbon [27] De Cooman BC, Kwon O, Chin KG. State-of-the-
content on the mechanical properties and microstruc- knowledge on TWIP steel. Mater Sci Tech. 2012;28:
tural evolution of Fe–22Mn–C steels. Mater Sci Eng A. 513–527.
2015;621:118–127. [28] Medvedeva N, Park M, Van Aken DC, et al. First-
[10] Jacques P, Furnémont Q, Mertens A, et al. On the principles study of Mn, Al and C distribution and their
sources of work hardening in multiphase steels assisted effect on stacking fault energies in fcc Fe. J Alloy Compd.
by transformation-induced plasticity. Philos Mag A. 2014;582:475–482.
2001;81:1789–1812. [29] Bouaziz O, Zurob H, Chehab B, et al. Effect of chemi-
[11] Olson GB, Cohen M. Stress-assisted isothermal marten- cal composition on work hardening of Fe-Mn-C TWIP
sitic transformation: application to TRIP steels. Metall steels. Mater Sci Tech. 2011;27:707–709.
Trans A. 1982;13:1907–1914. [30] Allain S, Chateau J-P, Bouaziz O. A physical model of
[12] Allain S, Chateau J-P, Dahmoun D, et al. Modeling the twinning-induced plasticity effect in a high man-
of mechanical twinning in a high manganese con- ganese austenitic steel. Mater Sci Eng A. 2004;387–389:
tent austenitic steel. Mater Sci Eng A. 2004;387–389: 143–147.
272–276. [31] Chen L, Kim H-S, Kim S-K, et al. Localized deformation
[13] Barbier D, Gey N, Allain S, et al. Analysis of the due to Portevin-LeChatelier effect in 18Mn-0.6 C TWIP
tensile behavior of a TWIP steel based on the tex- austenitic steel. ISIJ Int. 2007;47:1804–1812.
ture and microstructure evolutions. Mater Sci Eng A. [32] Lee S, Estrin Y, De Cooman BC. Effect of the strain
2009;500:196–206. rate on the TRIP–TWIP transition in austenitic Fe-12
[14] Bouaziz O, Allain S, Scott C. Effect of grain and twin pct Mn-0.6 pct C TWIP steel. Metall Mater Trans A.
boundaries on the hardening mechanisms of twinning- 2014;45:717–730.
induced plasticity steels. Scripta Mater. 2008;58: [33] Renard K, Ryelandt S, Jacques PJ. Characterisation of
484–487. the Portevin-Le Châtelier effect affecting an austenitic
[15] Bouaziz O, Guelton N. Modelling of TWIP effect TWIP steel based on digital image correlation. Mater Sci
on work-hardening. Mater Sci Eng A. 2001;319–321: Eng A. 2010;527:2969–2977.
246–249. [34] Idrissi H, Renard K, Ryelandt L, et al. On the mecha-
[16] Gutierrez-Urrutia I, Raabe D. Dislocation density mea- nism of twin formation in Fe–Mn–C TWIP steels. Acta
surement by electron channeling contrast imaging Mater. 2010;58:2464–2476.
in a scanning electron microscope. Scripta Mater. [35] Koyama M, Sawaguchi T, Lee T, et al. Work hard-
2012;66:343–346. ening associated with Ïţ-martensitic transformation,
[17] Jin J-E, Lee Y-K. Strain hardening behavior of a deformation twinning and dynamic strain aging in
Fe–18Mn–0.6 C–1.5 Al TWIP steel. Mater Sci Eng A. Fe–17Mn–0.6 C and Fe–17Mn–0.8 C TWIP steels.
2009;527:157–161. Mater Sci Eng A. 2011;528:7310–7316.
[18] Adler PH, Olson GB, Owen WS. Strain hardening [36] Seol J-B, Jung JE, Jang YW, et al. Influence of car-
of Hadfield manganese steel. Metall Mater Trans A. bon content on the microstructure, martensitic trans-
1986;17:1725–1737. formation and mechanical properties in austenite/ε-
[19] Dastur YN, Leslie WC. Mechanism of work hard- martensite dual-phase Fe–Mn–C steels. Acta Mater.
ening in Hadfield manganese steel. Metall Trans A. 2013;61:558–578.
1981;12:749–759. [37] Yang E, Zurob H, McDermid J. Mechanical behavior
[20] Idrissi H, Ryelandt L, Veron M, et al. Is there a rela- and microstructural evolution of Fe-22Mn-C TWIP/
tionship between the stacking fault character and the TRIP steels as a function of C content. Proc. Conf.
activated mode of plasticity of Fe–Mn-based austenitic on ‘Materials Science and Technology’, Houston, USA,
steels?. Scripta Mater. 2009;60:941–944. October 2010, 1914–1925.
[21] Curtze S, Kuokkala VT. Dependence of tensile deforma- [38] ‘Standard test method for determining average grain
tion behavior of TWIP steels on stacking fault energy, size’, E-112-96, ASTM, Philadelphia, PA, USA, 2004.
temperature and strain rate. Acta Mater. 2010;58: [39] ‘Standard test methods for tension testing of metal-
5129–5141. lic materials’, E8M-11, ASTM, Philadelphia, PA, USA,
[22] Allain S, Chateau J-P, Bouaziz O, et al. Correlations 2011.
between the calculated stacking fault energy and the [40] Ohkubo N, Miyakusu K, Uematsu Y, et al. Effect
plasticity mechanisms in Fe–Mn–C alloys. Mater Sci of alloying elements on the mechanical properties
Eng A. 2004;387–389:158–162. of the stable austenitic stainless steel. ISIJ Inter.
[23] Dumay A, Chateau J-P, Allain S, et al. Influence of addi- 1994;34:764–772.
tion elements on the stacking-fault energy and mechan- [41] Haase C, Ingendahl T, Güvenç O, et al. On the applica-
ical properties of an austenitic Fe–Mn–C steel. Mater Sci bility of recovery-annealed twinning-induced plasticity
Eng A. 2008;483–484:184–187. steels: potential and limitations. Mater Sci Eng A.
[24] Nakano J, Jacques PJ. Effects of the thermodynamic 2016;649:74–84.
parameters of the hcp phase on the stacking fault energy [42] Yanushkevich Z, Belyakov A, Kaibyshev R, et al. Effect
calculations in the Fe–Mn and Fe–Mn–C systems. CAL- of cold rolling on recrystallization and tensile behavior
PHAD. 2010;34:167–175. of a high-Mn steel. Mater Charact. 2016;112:180–187.
12 M. GHASRI-KHOUZANI AND J. R. MCDERMID

[43] Steinmetz DR, Jäpel T, Wietbrock B, et al. Revealing the [47] Byun TS. On the stress dependence of partial dis-
strain-hardening behavior of twinning-induced plastic- location separation and deformation microstructure
ity steels: theory, simulations, experiments. Acta Mater. in austenitic stainless steels. Acta Mater. 2003;51:
2013;61:494–510. 3063–3071.
[44] Allain S, Chateau J-P, Bouaziz O. Constitutive model [48] Yang E. The effect of carbon content on the mechanical
of the TWIP effect in a polycrystalline high man- properties and microstructural evolution of Fe-22Mn-C
ganese content austenitic steel. Steel Res. 2002;73: TWIP/TRIP steels. Master’s thesis, McMaster Univer-
299–302. sity, Hamilton, Canada, 2010.
[45] Gutierrez-Urrutia I, Zaefferer S, Raabe D. The effect of [49] Lu Y, Bruhis M, McDermid JR. Effect of strain path
grain size and grain orientation on deformation twin- on the microstructural evolution of a Fe-22Mn-0.6 C
ning in a Fe–22wt.% Mn–0.6 wt-% C TWIP steel. Mater alloy. Proc. 2nd Int. Conf. on ‘High manganese steels’,
Sci Eng A. 2010;527:3552–3560. Aachen, Germany, 2014, 113–118.
[46] Wang X, Zurob HS, Embury JD, et al. Microstructural [50] Remy L, Pineau A. Twinning and strain-induced
features controlling the deformation and recrystalliza- fcc → hcp transformation on the mechanical proper-
tion behaviour Fe–30% Mn and Fe–30% Mn–0.5% C. ties of Co-Ni-Cr-Mo alloys. Mater Sci Eng. 1976;26:
Mater Sci Eng A. 2010;527:3785–3791. 123–132.

You might also like