Materials Science & Engineering A 582 (2013) 235–244
Contents lists available at SciVerse ScienceDirect
Materials Science & Engineering A
journal homepage: www.elsevier.com/locate/msea
Bulk combinatorial design of ductile martensitic stainless steels
through confined martensite-to-austenite reversion
H. Springer n, M. Belde, D. Raabe
Max-Planck-Institut für Eisenforschung GmbH, 40237 Düsseldorf, Germany
a r t ic l e i nf o
a b s t r a c t
Article history:
Received 21 March 2013
Received in revised form
5 June 2013
Accepted 8 June 2013
Available online 20 June 2013
The effect of local martensite-to-austenite reversion on microstructure and mechanical properties was studied
with the aim of designing ductile martensitic steels. Following a combinatorial screening with tensile and
hardness testing on a matrix of six alloys (0–5 wt% Mn, 0–2 wt% Si, constant 13.5 wt% Cr and 0.45 wt% C) and
seven martensite tempering conditions (300–500 1C, 0–30 min), investigations were focussed on martensiteto-austenite reversion during tempering as function of chemical composition and its correlation with the
mechanical properties. While Mn additions promoted austenite formation (up to 35 vol%) leading to a
martensitic–austenitic TRIP steel with optimum mechanical properties (1.5 GPa ultimate tensile strength and
18% elongation), Si led to brittle behaviour despite even larger austenite contents. Combined additions of Mn
and Si broadened the temperature range of austenite reversion, but also significantly lowered hardness and
yield strength at limited ductility. These drastically diverging mechanical properties of the probed steels are
discussed in light of microstructure morphology, dispersion and transformation kinetics of the austenite, as a
result of the composition effects on austenite retention and reversion.
& 2013 Elsevier B.V. All rights reserved.
Keywords:
Steels
Mechanical properties
Microstructure
Transformation induced plasticity
Combinatorial alloy design
1. Introduction
Steels are the most versatile structural materials owing to their
multiple equilibrium and non-equilibrium phase transformations.
These enable a huge variety of easy to manipulate kinetic pathways that lead to an unprecedented spectrum of mechanical
properties through the intrinsic nano-structuring of the bulk
materials. Examples are soft interstitial free steels with less than
200 MPa flow stress and more than 60% elongation [1], twinning
induced plasticity (TWIP) steels offering superior ductility above
70% elongation [2–5], weight reduced “Triplex” steels with up to
1.5 GPa strength and yet more than 40% elongation [6–10], or
pearlitic wires that can be drawn to a strength beyond 6 GPa,
stronger than any other bulk material [11–13]. On the other hand,
steels – like all other structural materials – suffer from the fact that
an increase in strength via traditional hardening mechanisms is
associated with a decrease in ductility. This inverse stress–strain
relationship is currently the strongest limitation to the further
development of advanced alloys. It limits particularly the application of carbon-based (C) martensite, the strongest single phase
attainable in steel, despite its very high potential for enabling a
cost-efficient novel generation of ductile high strength steels.
n
Corresponding author. Tel.: +49 211 6792 796; fax: +49 211 6792 333.
E-mail address: h.springer@mpie.de (H. Springer).
0921-5093/$ - see front matter & 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.msea.2013.06.036
In order to overcome this inverse relationship between strength
and ductility, the hard and strong martensitic matrix can be blended e.
g. with ductile ferrite as in dual-phase steel microstructures [14,15] or
with instable and compliant austenite. Utilisation of the latter
approach offers the additional advantage that when exposed to
mechanical load, it can back-transform to martensite, acting as
additional strain hardening effect through the transformation induced
plasticity (TRIP) mechanism [16–19]. The austenite may be incorporated as a retained (i.e. present in the as-quenched state) and/or
reverted phase (formed and stabilised during subsequent tempering)
[20–26], and designed close or even beyond its mechanical or
thermodynamic stability limit [27–31]. Stabilisation of austenite for
retention after cooling to room temperature may be achieved by
substitutional alloying, e.g. with nickel (Ni) or manganese (Mn).
Interstitial austenite formers such as nitrogen (N) or C can furthermore
be readily partitioned during intercritical annealing [32], or, following
a more recent approach, after quenching just below martensite start
temperature MS and immediate tempering [33,34].
Reverted austenite typically offers greater microstructural synergy
with martensite than retained austenite, as it creates thin, compliant
interlayers at the martensite grain boundaries upon reversion treatment [22,23,35–37]. These layers have been shown to form a sort of
composite structure, inhibiting crack propagation otherwise prevalent
along the {100} planes of the martensite laths, and have thus been
linked to increased toughness and decreased ductile–brittle transition
temperatures [37,38]. By choosing appropriate chemical compositions
and thermo-mechanical treatments (TMT), the austenite formed via
236
H. Springer et al. / Materials Science & Engineering A 582 (2013) 235–244
reversion can be adjusted in terms of volume fraction, dispersion and
morphology in order to tailor the properties of the microstructure by
self-organization via grain-scale diffusion, grain boundary segregation
and selective phase transformation [22,23,31,35].
These advantageous effects of reverted austenite incorporated
within a martensitic matrix have thus prompted exploitation in
works on various steel designs such as supermartensitic steels
[31,39,40], austenitic [24,26] and martensitic stainless steels [28–
30,35]; mainly focussing on reversion kinetics, influence of heating
rate, martensite history and texture memory/orientation relationships. Yuan et al. [35] studied the mechanisms governing the
austenite reversion with atom probe tomography (APT) measurements of a martensitic stainless steel (1.4034, 13.5 wt% chromium
(Cr), 0.44 wt% C). They found that during tempering at 400 1C
subsequent to quenching from 1150 1C, C rapidly left the martensite
and segregated at lattice defects as well as in retained austenite
grains (‘kinetic freezing’, by the combination of high C solubility but
low diffusivity). By tailoring the nano-scaled austenite with respect
to content and size, ultra-high strength of up to 2 GPa was achieved
while maintaining good ductility (about 20% tensile elongation) [35].
In light of these numerous investigations regarding the favourable
impact of reversed-austenite regions on the mechanical properties of
high strength martensitic steels, it is of interest to systematically study
the influence of alloying elements. Of special relevance are Mn and
silicon (Si), as they are not only common accompanying elements in
commercially available Cr–C alloyed steels, but also expected to exert
strong effects while being cost effective [41–47]. Despite ongoing
developments in theoretical thermodynamics calculation methods
[48,49], however, the related microstructural phenomena which determine the mechanical properties of such steels are far from equilibrium
and thus still difficult to comprehensively predict. This is further
complicated by the large possible variety of chemical compositions in
the Fe–Cr–C–Si/Mn systems and, accordingly, large possible variations
in associated TMT setups (e.g. tempering time and temperature). As an
experimental alternative, combinatorial techniques can effectively
probe optimal synthesis and processing cycles of a multitude of
chemical concentrations [50,51] and microstructural variations
[52,53]. The recently introduced RAP approach, which is based on
semi-continuous high-throughput bulk casting, rolling, heat treatment
and sample preparation techniques, has demonstrated its effectiveness
in performing an accelerated screening of the tensile, hardness and
microstructural properties as a function of chemical and TMT parameters both systematically and efficiently [6]. It thus enables to
effectively identify compositional and microstructural trends that are
relevant for the bulk mechanical behaviour of the investigated steels.
Table 1
Target and actual chemical composition in wt% of the investigated alloys.
2. Objective
4. Results
The specific objective of this work is to systematically study the
influence of additions of Mn and Si on the microstructure and
mechanical properties of 13.5Cr–0.45C (wt%) based steels dependent on tempering time and temperature. The focus of this
combinatorial study of the compositional and thermomechanical variations lies in identifying ideal conditions for the
interface reversion mechanism and austenite stability with respect
to enhancing martensite ductility.
The mechanical data of the six alloys (Table 1) in their various
tempering conditions are plotted as an overview in Fig. 1. Yield
strength (YS; Fig. 1a), ultimate tensile strength (UTS; Fig. 1b), total
elongation (TE; Fig. 1c) and macro-hardness (Fig. 1d) are shown as
function of alloying additions to the 13.5Cr–0.45C (wt%) based
steels and colour-coded corresponding to the respective tempering treatments. The reference material without any alloying
additions, i.e. Fe–13.5Cr–0.45C, exhibits its highest hardness in
the as-quenched state (∼52 HRC) and, as expected, corresponding
brittle behaviour during tensile testing (premature failure at low
strains despite high inherent strength). Upon tempering the
reference alloy becomes more ductile – increasing with temperature – with a maximum TE value of about 15% after tempering at
500 1C for 10 min. This increase in ductility allows for better
utilising the materials strength, which results in higher UTS values
after tempering, despite lower respective hardness values than
in the as-quenched state. The same tendency applies for the YS
3. Materials and methods
3.1. Combinatorial sample synthesis procedure
Six Fe–13.5Cr–0.45C based alloys with different additions of
Mn, Si as well as both Mn and Si combined (Table 1) were
produced and processed by multiple runs of the RAP method
Target values (wt%)
Actual values (wt%)
Fe–13.5Cr–0.45C
Fe–13.5Cr–0.45C–2.5Mn
Fe–13.5Cr–0.45C–5Mn
Fe–13.5Cr–0.45C–1Si
Fe–13.5Cr–0.45C–2Si
Fe–13.5Cr–0.45C–2.5Mn–2Si
Fe–13.4Cr–0.46C
Fe–13.3Cr–0.45C–2.57Mn
Fe–12.8Cr–0.41C–5.03Mn
Fe–13.8Cr–0.47C–1.08Si
Fe–13.4Cr–0.46C–2.08Si
Fe–13.3Cr–0.48C–2.44Mn–2.05Si
detailed elsewhere [6]. After melting in a vacuum induction
furnace and casting into copper moulds, 80% thickness reduction
was achieved by rolling the ingots at 1000 1C to 2 mm thick
stripes, followed by air cooling to room temperature. Homogenisation was conducted by annealing at 1150 1C for 1 h under argon
atmosphere and water quenching. Tempering was performed at
300, 400 and 500 1C for 0, 10 and 30 min each (in air, oil quenching
to room temperature) for all investigated alloy compositions
(0 min representing the as-quenched state). The different alloys
are referred to by their target alloy composition in wt% (Table 1)
throughout this work.
3.2. Mechanical testing and characterisation
After preparing flat sheet tensile specimens (17.89 mm gauge
length) by spark erosion, testing was conducted at room temperature with an initial strain rate of 10–3 s–1 using a hydraulic testing
machine Instron 8511. Macro-hardness (Rockwell C) was measured
with a Wolpert DIA Testor 2RC on the outer surface (rolling plane)
of the segments after grinding them to 1000 grit. All values of both
tensile and macro-hardness testing represent average data of three
measurements for every alloy/tempering combination.
Cross sectional areas of the specimens were prepared in the
plane perpendicular to the rolling direction by grinding and
polishing with standard metallographic techniques. The cross
sections were investigated via optical microscopy (OM; Zeiss
Axiophot 1) and scanning electron microscopy (SEM; Jeol JSM
6490). Phase identification was performed via electron backscatter
diffraction analysis (EBSD; EDAX OIM software v.6.2.0), using
representative mappings of about 0.2 mm2 and a step size of
1 mm. Nano-indentation was carried out on the OM and SEM
observation plane, using a Hysitron TriboIndenter equipped with a
Ti 39-1 Bercovich tip at a constant load of 1000 mN with a step size
of 4 mm.
237
H. Springer et al. / Materials Science & Engineering A 582 (2013) 235–244
1200
1600
1100
1400
1200
900
UTS / MPa
YS / MPa
1000
800
700
600
1000
800
600
500
400
400
300
-
2.5Mn
5Mn
1Si
2Si
2.5Mn–
2Si
Alloy (wt.%) Fe–13.5Cr–0.45C–
20
-
2.5Mn
5Mn
1Si
2Si
2.5Mn–
2Si
Alloy (wt.%) Fe–13.5Cr–0.45C–
-
2.5Mn
5Mn
1Si
2Si
56
18
52
Hardness / HRC
16
TE / %
14
12
10
8
6
48
44
40
36
4
2
32
-
2.5Mn
5Mn
1Si
2Si
2.5Mn–
2Si
Alloy (wt.%) Fe–13.5Cr–0.45C–
Tempering conditions:
as-quenched
2.5Mn–
2Si
Alloy (wt.%) Fe–13.5Cr–0.45C–
300°C, 10min
400°C, 10min
500°C, 10min
300°C, 30min
400°C, 30min
500°C, 30min
Fig. 1. Overview of the mechanical properties as a function of chemical composition (Table 1) and the applied tempering treatments: (a) yield stress (YS); (b) ultimate tensile
stress (UTS); (c) total elongation (TE); (d) macro-hardness.
(1.1–1.2 GPa) and UTS (1.4–1.6 GPa), while the macro-hardness of
Fe–13.5Cr–0.45C remains constant during tempering at about 45
HRC. Adding Mn gradually reduces the brittleness in the asquenched state with increasing Mn concentrations. After tempering, the Mn alloyed steels have lower YS, UTS and macro-hardness
values than the respective reference material conditions (decreasing with Mn content), while the ductility is increased to up to 21%
for the Fe–13.5Cr–0.45C–5Mn material. The highest respective
values of YS, UTS and TE are observed for tempering the Mn
alloyed materials at 400 1C for 30 min. Additions of 1 and 2 wt% Si
result in the highest macro-hardness of all the alloys studied in
this work (56 HRC in the as-quenched state), which is slightly
reduced to values between 50 and 54 HRC by tempering. YS and
UTS increase with the tempering temperature. The ductility, on the
other hand, is with TE values between 2% and 4% lower than for all
the other investigated alloys and no necking of the Si alloyed
samples could be observed. Opposed to the aforementioned trends
observed when alloying with Mn (increase in ductility and drop in
strength) and Si (increase in hardness and higher embrittlement),
the simultaneous addition of Mn and Si leads to a drastically
different mechanical behaviour: Almost independent of the tempering conditions, the Fe–13.5Cr–0.45C–2.5Mn–2Si material is
softer (31–35 HRC) and has both, lower YS (400–450 MPa) and
UTS (480–600 MPa) when compared to the other alloys studied
here. The ductility of the Mn–Si alloyed steel is about 4% in the
as-quenched state, and it increases more strongly upon tempering
than that of the Si alloyed steels. However, despite the materials'
comparatively low strength the ductility remains far below the
respective values reached by solely Mn alloyed materials with a
maximum TE of only 11%.
The above mentioned trends in terms of chemistry-related
mechanical properties are illustrated in Fig. 2a by four selected
tensile curves after tempering at 400 1C for 30 min. In spite of the
drastically differing tensile testing behaviour of the four alloys,
corresponding SEM fractography analysis (Fig. 2b–e) revealed
inter- and trans-granular cleavage as the dominant failure
mechanism of all samples, with small ridges resembling ductile
failure (highlighted by white arrows in Fig. 2b–e). No clear
microstructural evidence linked to the varying mechanical data
could be observed in the fractured surfaces. The low apparent
elastic moduli of the alloys are an artefact of the applied high
throughput testing procedure
OM and SEM investigations (Fig. 3) revealed a predominantly
martensitic microstructure for the samples in the as-quenched
238
H. Springer et al. / Materials Science & Engineering A 582 (2013) 235–244
1600
Stress / MPa
1400
1200
Fe 13.5Cr 0.45C
1000
Fe 13.5Cr 0.45C 2.5Mn
Fe 13.5Cr 0.45C 2Si
800
Fe 13.5Cr 0.45C 2.5Mn 2Si
600
400
200
0
0
2
4
6
50 μm
10 μm
8
10
Strain / %
12
14
16
18
20 μm
10 μm
Fig. 2. (a) Exemplary engineering stress/strain curves from selected alloys after tempering at 400 1C for 30 min; (b–e) corresponding SEM fractographs.
state (former austenite grain size roughly about 100 μm; Fig. 3a),
along with few coarse carbides of about 1 μm diameter (most
probably non-dissolved primary carbides; Fig. 3b). In the tempered
state, varying amounts of finely dispersed nano-scaled carbide
platelets were observed, most notably in the Si-free samples. Aside
from these observations, no pronounced variations between the
different alloys were found except for the amount of austenite,
which was also observed to change drastically with tempering
conditions. The temperature dependence (tempering time 30 min)
of the austenite fraction is shown in Fig. 4 in blue together with
corresponding mechanical data; i.e. UTS (black) and TE data (red).
Maximum and minimum values of the mechanical properties are
shown as error bars to indicate the experimental scatter. The data
for the reference material Fe–13.5Cr–0.45C is plotted in Fig. 4a,
while the effects of alloying additions of Mn, Si and both Mn plus
Si combined are shown in Fig. 4(b–d), respectively. For all four
alloys the austenite content increases with temperature and
exhibits a peak after tempering at 400 1C for 30 min. Both absolute
values and their relative changes with temperature, however,
show strong variations as a function of the respective alloying
elements: adding Mn increases the respective austenite fractions
compared to the equivalent reference material conditions, especially for the peak value ( factor 3). The ductility of the Fe–
13.5Cr–0.45C–2.5Mn alloy follows the same trend as the austenite
fraction, resulting in a similar strength as the reference material
(UTS about 1500 MPa) but with a higher TE (18%). The austenite
fractions of the Fe–13.5Cr–0.45C–2Si alloy are very similar to those
of the reference material except for the peak value, which is more
than four times larger. Nevertheless, no pronounced changes in
ductility can be observed. The alloy Fe–13.5Cr–0.45C–2.5Mn–2Si
exhibits the highest initial austenite content, but a much more
gradual slope as the temperature for austenite reversion treatment
increases, and a much lower peak value than the Mn and Si alloyed
steels. Only a very modest correlation between austenite fraction
and ductility as well as strength can be observed.
More details concerning the changes in austenite fraction,
distribution and morphology of four selected alloys during tempering are shown in Fig. 5. The colour-coded EBSD phase maps
(martensite: red, austenite: green) shown on the left hand side
of the respective image pairs represent the as-quenched state;
images on the right hand side correspond to the same alloys
after quenching plus tempering at 400 1C for 30 min, respectively.
H. Springer et al. / Materials Science & Engineering A 582 (2013) 235–244
50 μm
2 μm
239
tempering results in less detectable austenite (25.1 vol%) than the
materials containing only Mn or Si, respectively. The austenite
reversion of the Mn–Si alloyed steel seems to proceed essentially
via growth of the retained austenite grains and to a lesser extent
by the formation of reverted austenite grains, which appear to
form clusters along the former austenite grain boundaries.
Nano-indentation experiments (Fig. 6) were performed to
obtain insight into the corresponding local mechanical properties
of the different multiphase microstructures shown in Fig. 5.
Colour-coded nano-hardness maps of the alloys containing
2.5 wt% Mn, 2 wt% Si and 2.5 wt% Mn plus 2 wt% Si, all after
tempering at 400 1C for 30 min, are shown in Fig. 6(a–c), respectively. The nano-hardness distributions for the maps of Fig. 6(a–c)
are plotted in Fig. 5d together with their respective colour code.
The higher hardness values, compared to the corresponding
macroscopic Rockwell data (Fig. 1d), are related to the much lower
load of the nano-indentation setup. For the alloy Fe–13.5Cr–0.45C–
2.5Mn a quasi-Gaussian coherent hardness distribution can be
observed (Fig. 6a and d) with an average (peak) value of about 750
HV. Alloying additions of 2 wt% Si result in a similar distribution
(Fig. 6b and d), only slightly shifted to a higher average value of
about 810 HV. This increase in hardness correlates with the
macroscopic observations (43.3 HRC for the Mn alloyed steel,
50.7 HRC for the Si alloyed material), despite the higher austenite
fraction for the alloy Fe–13.5Cr–0.45C–2Si (48.4 compared to
35.4 vol%, Fig. 5b and c). In case of the combined addition of Mn
and Si (Fig. 6c and d, macro-hardness 32.3 HRC), however, the
nano-hardness values span a broader range and a more bimodal
distribution can be observed: while large areas exhibit only about
600 HV, considerable fractions of the investigated cross section
yield about 1000 HV and more.
5. Discussion
1 μm
Fig. 3. Exemplary optical and scanning electron micrographs showing:
(a) martensitic microstructure in the alloy 13.5Cr–0.45C (as quenched), (b) coarse
carbides in the alloy 13.5Cr–0.45C–2.5Mn (as quenched), (c) fine platelets of nanosized carbides in the alloy 13.5Cr–0.45C–2.5Mn after tempering at 400 1C for
30 min.
The image quality data is superimposed in grey scale; epsilon
martensite is not taken into account due to its negligible volume
fraction (maximum 2 vol%) and low confidence index. After
quenching, 0.6 vol% retained austenite, with grain sizes ranging
from about 5 down to 0.5 mm (scan resolution) was detected in the
reference material (Fig. 5a). After tempering, the austenite fraction
of the reference alloy increases to 11.6 vol% (reverted austenite),
via growth of already existing (retained) austenite grains located
at former austenite grain boundaries, as well as by the formation
of new micro-sized austenite grains within the matrix. Addition
of 2.5 wt% Mn (Fig. 5b) results in a slightly increased amount of
retained austenite after quenching (3.2 vol%) and also in a larger
retained austenite grain size (∼10 mm), compared to the reference
material. Tempering the 13.5Cr–0.45C–2.5Mn steel considerably
increases the total amount of austenite to 35.4 vol%, both by more
pronounced growth of the retained austenite grains and an
increased formation of fine and more evenly dispersed austenite
compared to the reference material. Adding 2 wt% Si (Fig. 5c)
effectively suppresses retained austenite (0.2 vol% in the asquenched state), but greatly promotes the formation of fine
grained reverted austenite to a total amount of 48.4 vol% after
tempering at 400 1C for 30 min. While the alloy Fe–13.5Cr–0.45C–
2.5Mn–2Si (Fig. 5d) exhibits the largest amount of retained
austenite of all alloys (5.9 vol% and up to 20 mm large grains),
5.1. Rapid alloy prototyping of martensitic stainless steels
By utilising a novel bulk RAP approach [6], the effect of Mn and
Si additions on microstructure and mechanical properties (Fig. 1)
of martensitic stainless steels could be systematically investigated.
The presented synthesis, processing and mechanical testing cycles
for all six different alloys, including seven different heat treatments for each composition, was performed within ten days, so
that chemistry- and TMT-related trends could be identified
(Figs. 2–4). Those samples with the most interesting property
profiles were selected for a more thorough microstructure characterisation using SEM and nano-indentation probing (Figs. 5 and
6). Special emphasis was placed on the formation of austenite as a
function of chemical composition and tempering conditions, as
well as on its effectiveness in improving martensite ductility
(Fig. 4).
5.2. Influence of chemical composition on austenite formation
Two types of austenite could be detected in the quenched plus
tempered alloys (Fig. 4), namely, retained and reverted austenite.
For both types the relative fraction, grain size, morphology and
dispersion is strongly linked to the respective alloying additions, as
are the kinetics of martensite-to-austenite reversion (Fig. 5). For
the reference alloy of this study (Fe–13.5Cr–0.45C), 0.6 vol% of
retained austenite was detected. This finding is attributed to the
high austenitisation temperature of 1150 1C. It is much higher than
the austenitisation temperature used in typical heat treatments for
tooling applications. The high temperature promotes the complete
dissolution of the carbides and hence leads to a high solute C
content, which in turn stabilises the austenite. For a 13.5 wt% Cr
240
H. Springer et al. / Materials Science & Engineering A 582 (2013) 235–244
Fe–13.5Cr–0.45C
30
800
20
600
10
400
0
200
0
40
1200
1000
30
800
20
600
10
400
0
200
100
200
300
400
500
Tempering temperature / °C
0
Fe–13.5Cr–0.45C–2Si
800
20
600
10
400
200
0
0
100
200
300
400
500
Tempering temperature / °C
UTS / MPa
30
50
1400
40
1200
1000
30
800
20
600
10
400
200
0
0
austenite / vol.%
1000
austenite / vol.%
UTS / MPa
40
1200
1600
TE / %;
1400
100
200
300
400
500
Tempering temperature / °C
Fe–13.5Cr–0.45C–2.5Mn–2Si
50
TE / %;
1600
austenite / vol.%
1000
50
1400
UTS / MPa
40
1200
1600
TE / %;
1400
UTS / MPa
Fe–13.5Cr–0.45C–2.5Mn
50
TE / %; austenite / vol.%
1600
100
200
300
400
500
Tempering temperature / °C
Fig. 4. Correlation between austenite fraction, ultimate tensile strength (UTS) and total elongation (TE) as function of tempering temperature (duration: 30 min) of four
selected alloys. (For interpretation of the references to colour in this figure, the reader is referred to the web version of this article.)
Fe–13.5Cr–0.45C
as-quenched
Fe–13.5Cr–0.45C–2.5Mn
400 °C, 30 min
Fe–13.5Cr–0.45C–2Si
Fe–13.5Cr–0.45C–2.5Mn–2Si
Normal direction
Transverse direction
60 μm
Fig. 5. Colour-coded EBSD phase maps (austenite: green; martensite: red) of four selected alloys in the as-quenched state (left images) and after tempering at 400 1C for
30 min (right images), respectively. The image quality data is superimposed in grey scale. (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
and 0.44 wt% C steel a MS temperature of 118 1C was reported for
similar experimental conditions [35]. While both, Mn and Si
further decrease MS according to the literature, Mn shows a much
stronger effect [41,44,54]. Consequently, minor additions of Si left
the retained austenite fraction virtually unaffected (0.2 vol% for
the alloy Fe–13.5Cr–0.45C–2Si), while increasing Mn concentrations effectively stabilised the austenite, with 3.2 vol% for the 2.5Mn alloy and up to 6.8 vol% for the material containing 5 wt% Mn
(Figs. 4 and 5) in the as-quenched state. The local formation of
retained austenite is further promoted by the tendency of Mn to
segregate during steel synthesis and processing [55–59]. In this
context it has to be noted that while the relatively large EBSD
maps used in this study enable a sufficiently reliable statistical
analysis, the correspondingly large step size used in the EBSD
mapping (1 mm step size) limits the detection of grains with a size
below this value. The amount of retained austenite of the alloy
H. Springer et al. / Materials Science & Engineering A 582 (2013) 235–244
241
Fe–13.5Cr–0.45C–2Si
Fe–13.5Cr–0.45C–2.5Mn
Normal direction
Transverse direction
10 μm
Fe–13.5Cr–0.45C–2.5Mn–2Si
Number of indents
60
Fe–13.5Cr–0.45C–2.5Mn
Fe–13.5Cr–0.45C–2Si
Fe–13.5Cr–0.45C–2.5Mn–2Si
50
40
30
20
10
0
500 600 700 800 900 1000 1100
Nano-hardness / HV
Fig. 6. Nano-indentation results of three selected alloys after tempering at 400 1C for 30 min: (a–c) colour coded nano-hardness maps; (d) corresponding nano-hardness
distribution and colour-scale of the three maps. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
containing both Mn and Si is with 5.9 vol% higher than the
corresponding values observed for the alloys containing only
either Si or Mn (Fig. 4). This non-linearity is expected in view of
the reportedly complex and even equivocal role of Mn on austenite
formation/retention in rapidly cooled steels [60–63].
According to Yuan et al. [35], austenite reversion during
tempering of martensitic stainless steels is controlled by partitioning, segregation and kinetic freezing of C, as the mobility of
substitutional elements such as Cr is too low at the chosen
tempering temperatures [35]. Following this scenario, formation
and growth of austenite can only occur if the C-concentration is
high enough to provide a sufficient thermodynamic driving force
during the martensite-reversion heat treatment, given the nucleation barrier is overcome. In the latter context it must be considered that the nucleation of austenite reversion in a martensite
matrix is usually not only promoted by the segregation of
elements that stabilise austenite (for instance on the lath grain
boundaries) but also by the gain in elastic distortion energy that is
released when the martensite is transformed into austenite [64–
66]. The actual local C-concentration during quenching and subsequent reversion tempering is governed by two competing
reactions [22,35]. These are the equilibrium segregation of C at
martensite lattice defects, C-enrichment inside the retained austenite through equilibrium partitioning, and local C-depletion in
the martensite via the formation of various carbides upon tempering (i.e. Fe2C, M3C, M7C3, M23C6 [34,42,67]). As a result, the amount
of reverted austenite (i.e. total fraction minus retained austenite)
follows a characteristic trend as a function of the tempering
temperature, as observed in Fig. 6a, with a maximum value at
400 1C. Additionally, all three factors (required C-concentration,
C-enrichment, carbide formation) also strongly depend on the
materials' chemical composition, which can explain the different
effects of Mn and Si additions [41,42] (Fig. 7a).
Of the factors governing the extent of austenite reversion, the
required C-concentration to stabilise the fcc over the bcc lattice
structure can be estimated by CALPHAD-predictions (Thermo-Calc
version S, database SSOL2) as a function of tempering temperature. As
can be seen in Fig. 7b, at the bulk C-concentration of 0.45 wt%, bcc is
the stable phase for all alloys. The (locally) required C-concentration to
reach equilibrium with fcc (critical concentration, Fig. 7b), systematically drops with temperature and characteristically varies with
substitutional alloying (Fig. 7c). The addition of 2.5 wt% Mn effectively
lowers the critical C-concentration by about 0.3 wt% compared to the
reference alloy. This strong effect of Mn is in good agreement with
matching observations of increased austenite reversion in the Mnalloyed material (Figs. 5 and 7a). This is likely to be further enhanced
by possible micro-segregations of Mn at martensite laths [41,46,55,57],
which may also account for the occurrence of austenite clustering. Si,
on the other hand, is thermodynamically not as effective in lowering
the critical C-concentration, but in practice it strongly affects the
formation of carbides in terms of type and temperature range, and in
turn increases the activity of C [43–45]. The resultant austenite
reversion of the alloy containing 2 wt% Si is much stronger and
ubiquitous (almost 50 vol% after tempering at 400 1C for 30 min)
but limited to a more narrow temperature range (Fig. 7a). Combined
additions of 2.5 wt% Mn and 2 wt% Si give the lowest critical
242
H. Springer et al. / Materials Science & Engineering A 582 (2013) 235–244
Also included into Fig. 7a and c are calculations by Yuan et al.
[35] for the industrially produced steel 1.4034 containing 13.5 wt%
Cr and 0.44 wt% C. Despite the comparatively small amount of
additional elements being present as impurities in this material
(0.53Mn, 0.28Si, 0.16Ni, 0.02N; all in wt% [35]), the retained and
reverted austenite fractions are significantly increased and the
critical C concentration reduced in comparison to the reference
alloy of this study. This highlights the complexity and interdependence of alloying elements on austenite retention and reversion
phenomena.
Fe 13.5Cr 0.45C 2.5Mn 2Si
Fe 13.5Cr 0.45C 2.5Mn
Fe 13.5Cr 0.45C 2Si
50
Reverted austenite
fraction / vol.%
Fe 13.5Cr 0.45C
40
1.4034, Yuan et al. 2012
30
Tempering time
30 min
20
5.3. Influence of retained and reverted austenite on the mechanical
properties
10
0
∆G α γ at 400 °C / kJ mol–1
0
100
200
300
400
500
Tempering temperature / °C
Nominal concentration: 0.45 wt.%
1.2
0.8
Critical concentrations
0.4
0
-0.4
-0.8
Critical C-concentration /
wt.%
0.2
0.4
0.6 0.8 1.0 1.2 1.4
C-concentration / wt.%
1.6
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
300
400
500
Tempering temperature / °C
Fig. 7. Austenite reversion influenced by chemical composition, in comparison to
literature data [35]: (a) reverted austenite fraction measured as function of
tempering temperature; (b) CALPHAD-based determination of the critical carbon
content required for austenite reversion; (c) critical carbon content as function of
tempering temperature.
C-concentrations for austenite stability, which is supported by the
microstructural observations of the alloy Fe–13.5Cr–0.45C–2.5Mn–2Si
for tempering at 300 1C and 500 1C. The corresponding peak value
at 400 1C, however, is much lower than for sole additions of Mn and Si.
This finding, together with the austenite reversion appearing mainly
as growth of the retained austenite grains rather than the formation
of fine grains at martensite laths (Fig. 5c), clearly requires more
detailed work. It is thereby a good example for future ‘up-scaling’
options, enabling the production and processing of large quantities of
material by conventional metallurgical methods. Also it is a field for
more detailed investigations using transmission electron microscopy
or APT, guided by insights gained through a high-throughput RAPscreening, thus allowing for an efficient and knowledge based alloy
design process.
Austenite is the only microstructure constituent of the investigated steels to undergo appreciable plastic deformation and lend
ductility to the strong martensitic matrix. Hence, as the fraction of
(retained) austenite is increased by adding Mn to the reference
state, the ductility – and thus usable strength – of the steels in the
as-quenched state becomes higher (Fig. 1).
Tempering, as discussed in the previous section, generates
additional reverted austenite, but no straightforward link between
the austenite fraction and the respective materials strength and
ductility could be observed in the investigated alloys (Fig. 4).
Apparently, the extremely differing material properties as function
of chemical composition (exemplified in Fig. 2) are the result of
not only the austenite volume fraction, but also its respective grain
size (dispersion), distribution, and stability (Fig. 5). Of specific
interest in this context is the range of the austenite stability
against martensitic transformation upon mechanical loading
[18,19,37]. Here the mechanical mapping results of the nanoindentation experiments (Fig. 6) and the phase maps (Fig. 5)
provide some insight: only a favourable combination of all those
factors leads to the desired simultaneous achievement of both
strength and ductility. In case of the reference alloy, the comparatively little amount of small to medium sized reverted austenite
grains (Fig. 5a) and their reasonable dispersion enables to effectively exploit the materials' high strength of almost 1600 MPa (the
maximum UTS value of all investigated alloys). Mn additions lead
to well dispersed, very fine grained reverted austenite in addition
to small amounts of medium sized retained (and subsequently
grown) austenite. The resulting range of austenite stability and
ductility is probed qualitatively by the EBSD image quality (shown
jointly with phase map Fig. 5b) and the nano-hardness (Fig.6a),
respectively: the lower EBSD image quality of the newly formed
reverted austenite indicates a larger dislocation density compared
to the retained austenite [68]. The variety in dislocation content
and size [26,32,69] translates to a corresponding variety in the
critical strains for martensitic transformation. Such microstructure
heterogeneity promotes an integral mechanical response that is
characterised by gradually stimulated and hence, more permanent
strain-hardening as shown by the blue curve in Fig. 2. The alloy
containing 2.5 wt% Mn hence reveals the most attractive combination of mechanical properties of all the investigated steels by the
utilisation of this gradually activated and heterogeneous TRIP
effect [18,35].
Si additions, on the other hand, led to massive formation of
interconnected, fine grained reverted austenite (Fig. 5c), but the
materials remained hard (Fig. 6b) and brittle. It seems that the
Si-induced reverted austenite is either so stable that it does not
transform to martensite during deformation, or that it transforms
simultaneously at low elongations to brittle martensite containing
high amounts of C (large critical C-concentration, Fig. 7b). Austenite stabilisation may result from C-supersaturation [45], austenite
morphology [37,70] or the hydrostatic pressure of the surrounding
martensite [71,72].
H. Springer et al. / Materials Science & Engineering A 582 (2013) 235–244
The drastically lowered YS and still limited ductility of the alloy
containing both Mn and Si seems to be the result of its disadvantageous, bimodal microstructure, which consists of a network of
large and instable austenite grains embedded in a hard martensitic
matrix containing almost no fine-grained, newly formed austenite
(Figs. 5d and 6c) [70,72].
It should be emphasised that the results of this study were
obtained using the RAP approach, and that consequently some
specific effects of this efficient high-throughput screening procedure need to be taken into account [6]. This means that conventional processing might provide slightly different mechanical and
microstructure results. This applies particularly for the streamlined TMT setup (coarse grained microstructure due to homogenisation and austenitisation in one step), tensile testing procedure
(decreased ductility of high-throughput flat sheet specimens with
remaining scales and surface roughness from processing) and
phase identification experiments (only EBSD measurements, large
maps with good statistics but limited resolution). These factors,
together with the already discussed deviations in chemical composition, may explain the differences compared to the data
reported by Yuan et al. [35] in terms of the respective austenite
contents (see previous section) and mechanical properties of
1.75 GPa UTS and 23% elongation.
6. Summary and conclusions
This study applies a novel bulk high throughput synthesis,
processing and tensile testing sequence with the goal of identifying ductile martensitic compositions and optimal microstructures.
Regarding the composition the effects of both, single and combined alloying additions of Mn (0–5 wt%) and Si (0–2 wt%) on the
microstructure and mechanical properties of a group of 13.5Cr–
0.45C (wt%) based martensitic stainless steels were examined.
Tensile and hardness testing was performed on a matrix of six
alloys and seven tempering conditions (300, 400 and 500 1C for 0,
10 and 30 min each) following the recently introduced rapid alloy
prototyping approach. Based on the results obtained from this
high throughput bulk combinatorial screening, special emphasis
was placed on austenite reversion phenomena and their potential
to enhance martensite ductility while conserving strength as a
function of the chemical composition of the investigated steels.
The following conclusions can be drawn:
(1) Mn additions increase the fraction of austenite retained after
quenching from 1150 1C and thereby improve the ductility and
usable strength of the martensitic matrix. Upon tempering at
400 1C for 30 min, the best combination of strength and
ductility (about 1.5 GPa UTS and 18% tensile elongation) is
observed for the alloy containing 2.5 wt% Mn. Si additions lead
to the highest hardness but also the most brittle material
behaviour, almost unaffected by tempering. Combined additions of 2.5 wt% Mn and 2 wt% Si result in low hardness and
very low YS values (400 MPa), but nevertheless a maximum
elongation of only 12%.
(2) All investigated alloys exhibit austenite reversion during
tempering, which peaks at 400 1C. The respective absolute
and temperature dependencies, however, greatly change with
the chemical composition of the steels. Mn and Si appear to
effectively influence the three partly competing factors for
the mainly C-controlled austenite reversion; i.e. the required
C-concentration for a sufficient thermodynamic driving force,
the locally achievable C-enrichment via partitioning and
segregation at local defects, and the formation of carbides.
(3) The differing properties of the materials were observed to be
caused not only by the differences in austenite fraction, but
243
also by its grain size, dispersion and in turn stability against
martensitic transformation. Mn additions apparently impart
optimum mechanical properties via a large number of fine
austenite grains with a variety of critical strains. Si leads to
high contents of austenite (up to 50 vol%) but of either too
high stability or non-favourable transformation behaviour.
Large amounts of Mn and Si together form a bimodal microstructure of large, net-like distributed austenite grains which
is believed transform easily and rapidly, thereby not promoting ductility.
(4) While its potential of enhancing martensite ductility was
clearly demonstrated, additional work is required to gain more
insight into the mechanisms of austenite reversion as a
function of chemical composition as well as its respective
transformation behaviour during deformation. Especially the
role of carbide formation, or the effect of other alloying
elements such as nitrogen, need to be investigated using
higher resolution characterisation techniques such as TEM
and APT.
(5) The RAP approach has been successfully applied to rapidly
investigate the chosen matrix of alloy systems and tempering
conditions, as well as to identify the corresponding microstructural trends and their relation to the mechanical properties. This allows for efficient and knowledge based alloy
design, as the aforementioned issues can now be readily
addressed in a mechanism-targeted and efficient ‘up-scaling’
based on the insights gathered from this high throughput
screening using conventional metallurgical, processing and
testing methods.
Acknowledgements
C. Baron is acknowledged for support with mechanical testing,
J. Wang for providing data obtained during her master thesis work.
References
[1] A.K. Gupta, D.R. Kumar, J. Mater. Process. Technol. 172 (2006) 225–237.
[2] O. Bouaziza, S. Allaina, C.P. Scotta, P. Cugya, D. Barbier, Curr. Opin. Solid State
Mater. Sci. 15 (2011) 141–168.
[3] I. Gutierrez-Urrutia, J.A. del Valle, S. Zaefferer, D. Raabe, J. Mater. Sci. 45 (2010)
6604–6610.
[4] I. Gutierrez-Urrutia, D. Raabe, Acta Mater. 59 (2011) 6449–6462.
[5] D.R. Steinmetz, T. Jäpel, B. Wietbrock, P. Eisenlohr, I. Gutierrez-Urrutia,
A. Saeed–Akbaric, T. Hickel, F. Roters, D. Raabe, Acta Mater. 61 (2013) 494–510.
[6] H. Springer, D. Raabe, Acta Mater. 60 (2012) 4950–4959.
[7] I. Gutierrez-Urrutia, D. Raabe, Acta Mater. 60 (2012) 5791–5802.
[8] G. Frommeyer, U. Brüx, Steel Res. Int. 77 (2006) 627–633.
[9] K.M. Chang, C.G. Chao, T.F. Liu, Scr. Mater. 63 (2010) 162–165.
[10] W.K. Choo, J.H. Kim, J.C. Yoon, Acta Mater. 45 (1997) 4877–4885.
[11] Y.J. Li, P. Choi, S. Goto, C. Borchers, D. Raabe, R. Kirchheim, Acta Mater. 60
(2012) 4005–4016.
[12] D. Raabe, et al., MRS Bull. 35 (2010) 982.
[13] Y.J. Li, P. Choi, C. Borchers, S. Westerkamp, S. Goto, D. Raabe, R. Kirchheim, Acta
Mater. 59 (2011) 3965–3977.
[14] Y.I. Son, K.L. Young, K.T. Park, C.S. Lee, D.H. Shin, Acta Mater. 53 (2005)
3125–3134.
[15] M. Calcagnotto, D. Ponge, D. Raabe, Mater. Sci. Eng. A 527 (2010) 7832–7840.
[16] G.B. Olson, M. Cohen, J. Less Common Met. 28 (1972) 107–118.
[17] G.B. Olson, M. Azrin, Metall. Trans. A 9A (1978) 713–721.
[18] J. Wang, S. van der Zwaag, Metall. Mater. Trans. A 32A (2001) 1527–1539.
[19] M.Y. Sherif, Mater. Sci. Technol. 20 (2004) 319.
[20] J.G. Krauss, M. Cohen, Trans. Am. Inst. Metall. Eng. 224 (1962) 1212.
[21] J.F. Breedis, Trans. Am. Inst. Eng. 236 (1966) 218.
[22] O. Dmitrieva, D. Ponge, G. Inden, J. Millán, P. Choi, J. Sietsma, D. Raabe, Acta
Mater. 59 (2011) 364–374.
[23] Y.K. Lee, H.C. Shin, D.S. Leem, J.Y. Choi, W. Jin, C.S. Choi, Mater. Sci. Technol. 19
(2003) 393.
[24] R Montaneri, Z. Metallk. 81-H2 (1990) 114.
[25] Y. Ma, J.-E. Jin, Y.-K. Lee, Scr. Mater. 52 (2005) 1311.
[26] S. Rajasekhara, L.P. Karjalainen, A. Kyröläinen, P.J. Ferreira, Mater. Sci. Eng. A
527 (2010) 1986–1996.
244
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
H. Springer et al. / Materials Science & Engineering A 582 (2013) 235–244
G. Krauss, Acta Metall. 11 (1963) 499.
S. Jana, C.M. Wayman, Trans. Am. Inst. Eng. 239 (1967) 1187.
H. Kessler, W. Pitsch, Acta Metall. 15 (1967) 401.
K.B. Guy, E.P. Butler, D.R.F. West Met, Science 17 (1983) 167–176.
K. Tomimura, S. Takaki, Y. Tokunuga, ISIJ Int. 31 (1991) 1431–1437.
O. Thuillier, F. Danoix, M. Gouné, D. Blavette, Scr. Mater. 55 (2006) 1071.
J. Speer, D.K. Matlock, B.C De Cooman, J.G. Schroth, Acta Mater. 51 (2003)
2611–2622.
D.V. Edmonds, K. He, F.C. Rizzo, B.C. De Cooman, D.K. Matlock, J.G. Speer,
Mater. Sci. Eng. A 25 (2006) 438–440.
L. Yuan, D. Ponge, J. Wittig, P. Choi, J.A. Jiménez, D. Raabe, Acta Mater. 60
(2012) 2790–2804.
N. Nakada, T. Tsuchiyama, S. Takaki, N. Miyano, ISIJ Int. 51 (2011) 299–304.
A.S. Podder, I. Lonardelli, A. Molinari, H.D.H. Bhadeshia, Proc. R. Soc. A 467
(2011) 3141.
J.W. Morris, Z. Guo, C.R. Krenn, Y.H. Kim, ISIJ Int. 41 (2011) 599.
D. Zou, Y. Han, W. Zhang, X. Fang, J. Iron Steel Res. Int. 17 (2010) 17.
H. Smith, D.R.F. West, J. Mater. Sci. 8 (1973) 1413.
E. De Moor, D.K. Matlock, J.G. Speer, M.J. Merwin, Scr. Mater. 64 (2011)
185–188.
D. Edmonds, D. Matlock, J. Speer, La Metall. Ital. 1 (2011) 41–48.
R. Bernst, G. Inden, A. Schneider, Comput. Coup. Phase Dia. Thermochem. 32
(2008) 207–216.
S.S. Nayaka, R. Anumolu, R.D.K. Misra, K.H. Kimc, D.L. Lee, Mater. Sci. Eng. A
498 (2008) 442–456.
B.C. De Cooman, Mater. Sci. 8 (2004) 285–303.
J. Lis, J. Morgiel, A. Lis, Mater. Chem. Phys. 81 (2003) 466–468.
S.S. Babu, K. Hono, T. Sakurai, Appl. Surf. Sci. 67 (1993) 321–327.
Y. Takahama, M.J. Santofimia, M.G. Mecozzi, L. Zhao, J. Sietsma, Acta Mater. 60
(2012) 2916–2926.
M.G. Michal, F. Ernst, A.H. Heuer, Metall. Mater. Trans. A 37 (2006) 1819.
T. Gebhardt, D. Music, M. Ekholm, I.A. Abrikosov, L. Vitos, A. Dick, T. Hickel,
J. Neugebauer, J.M. Schneider, J. Phys.: Condens. Matter 23 (2011) 246003.
[51] J. Cui, Y.S. Chu, O.O. Famodu, Y. Furuya, J. Hattrick-Simpers, R.D. James,
A. Ludwig, S. Thienhaus, M. Wuttig, Z. Zhang, I. Takeuchi, Nat Mater. 5
(2006) 286–290.
[52] X. Zheng, D.G. Cahill, J.C. Zhao, Acta Mater. 58 (2011) 1236–1241.
[53] J.C. Zhao, X. Zheng, D.G. Cahill, JOM 63 (2011) 40.
[54] G. Krauss Steels, Processing, Structure and Performance, ASM International,
Materials Park, Ohio, USA, 2005.
[55] D. Senk, H. Emmerich, J. Rezende, R. Siquieri, Adv. Eng. Mater. 9 (2007) 695.
[56] T.F. Majka, D.K. Matlock, G. Krauss., Metall. Mater. Trans. A 33 (2002) 1627.
[57] S.H. Mun, M. Watanabe, X. Li, K. Oh, D.B. Williams, H.C. Lee, Metall. Mater.
Trans. A 33 (2002) 1057.
[58] E.A. Wilson, Metall. Mater. Trans. A 35 (2004) 352.
[59] M. Nasim, B.C. Edwards, E.A. Wilson, Mater. Sci. Eng. A 281 (2000) 56–67.
[60] A.L. Schaeffler, The Iron Age72.
[61] N. Suutula, Metall. Trans. A 13 (1982) 2121.
[62] S.R. Chen, H.A Davies, W.M. Rainforth, Acta Mater. 47 (1999) 4555–4569.
[63] F.C. Hull, Weld. Res. J. (52) (1973) 193.
[64] S. Rajasekhara, P.J. Ferreira, Acta Mater. 59 (2011) 738–748.
[65] N. Nakada, T. Tsuchiyama, S. Takaki, S. Hashizume, ISIJ Int. 47 (2007)
1527–1532.
[66] P.C. Clemm, J.C. Fisher., Acta Metall. 3 (1955) 70.
[67] L.S. Malinov, A.P. Cheilyakh, O.D. Cheilyakh, Metalloved. Term. Obrab. Met. 8
(1988) 12–15.
[68] V.G. Gorbach, I.V. Sidoruk, E.A. Izmailov, Metalloved. Term. Obrab. Met. 8
(1988) 9–12.
[69] M.C. Somani, P. Juntunen, L.P. Karjalainen, R.D.K Misra, A. Kyröläinen, Metall.
Mater. Trans. A 40 (2009) 729.
[70] E. Jimenez-Melero, N.H. van Dijk, L. Zhao, J. Sietsma, S.E. Offerman, J.P. Wright,
S. van der Zwaag, Acta Mater. 55 (2007) 6713–6723.
[71] H. Berns, W. Theisen, Eisenwerkstoffe Stahl und Gusseisen, Springer Verlag,
Berlin, Heidelberg, 2006.
[72] D.Q. Bai, A.D. Chiro, S. Yue, Mater. Sci. Forum 284 (1998) 253–262.
Available online at www.sciencedirect.com
Acta Materialia 60 (2012) 4950–4959
www.elsevier.com/locate/actamat
Rapid alloy prototyping: Compositional and thermo-mechanical
high throughput bulk combinatorial design of structural materials
based on the example of 30Mn–1.2C–xAl triplex steels
H. Springer ⇑, D. Raabe
Max-Planck-Institut für Eisenforschung GmbH, 40237 Düsseldorf, Germany
Received 20 December 2011; received in revised form 7 May 2012; accepted 15 May 2012
Abstract
We introduce a new experimental approach to the compositional and thermo-mechanical design and rapid maturation of bulk structural materials. This method, termed rapid alloy prototyping (RAP), is based on semi-continuous high throughput bulk casting, rolling,
heat treatment and sample preparation techniques. 45 Material conditions, i.e. 5 alloys with systematically varied compositions, each
modified by 9 different ageing treatments, were produced and investigated within 35 h. This accelerated screening of the tensile, hardness
and microstructural properties as a function of chemical and thermo-mechanical parameters allows the highly efficient and knowledgebased design of bulk structural alloys. The efficiency of the approach was demonstrated on a group of Fe–30Mn–1.2C–xAl steels which
exhibit a wide spectrum of structural and mechanical characteristics, depending on the respective Al concentration. High amounts of Al
addition (>8 wt.%) resulted in pronounced strengthening, while low concentrations (<2 wt.%) led to embrittlement of the material during
ageing.
Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Steels; Mechanical properties; Structural alloys; Combinatorial alloy design; Fe–Mn–Al–C steel
1. Introduction
Increased efficiency in the development of novel metallic
structural materials is a key to providing fundamental and
applied alloy solutions in the fields of energy, mobility,
health, infrastructure and safety. Examples of the associated basic metallurgical questions are alloy- and processing-sensitive changes in complex strain hardening
phenomena. In high strength steels these are often associated with displacive transformations, such as twinning
and transformation-induced plasticity (TWIP and TRIP),
which can be tuned through variations in the stacking fault
energy and microstructure. Associated engineering issues
are systematic analysis of the corresponding trends in
⇑ Corresponding author. Tel.: +49 211 6792 796; fax: +49 211 6792 333.
E-mail address: h.springer@mpie.de (H. Springer).
texture evolution, sheet forming and damage parameters,
joining behaviour, hydrogen susceptibility or fatigue.
In thin film systems, in which combinatorial high
throughput methods were invented and established, substantial progress has been made in a number of materials
design fields, such as shape memory alloys, MAX phases
and shape change materials [1–6]. Such combinatorial
methods have mostly been applied to the design of functional materials, where, typically, the intrinsic composition-dependent properties dominate over the relevance of
the microstructure. In structural alloys, on the other hand,
the relevant length scales that determine the mechanical
behaviour are parameters such as grain size, texture, precipitate dispersion and topology and the dislocation cell
structure, to name but a few. The correlation lengths associated with such lattice defects are usually of the order of
several nanometres to multiple micrometres, hence, they
usually exceed the dimensions accessible in thin films. This
1359-6454/$36.00 Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.actamat.2012.05.017
H. Springer, D. Raabe / Acta Materialia 60 (2012) 4950–4959
means that the mechanical properties of structural materials are not only determined by their chemical composition,
but to a large extent by the microstructure, which in turn is
greatly influenced by the thermo-mechanical treatments
(TMTs) applied after synthesis [7–9].
Consequently, the current conventional approach for
the experimentally guided development of metallic structural materials typically consists of a number of iterative
loops that include the following basic steps: Bulk casting
of a single composition charge, hot and/or cold deformation (e.g. conventional or thermo-mechanical rolling), heat
treatment (such as quenching, partitioning or ageing),
machining of tensile specimens and mechanical testing.
Coupled with the wide range of possible chemical compositions and different TMT setups, this kind of incremental
trial and error approach typically provides robust mechanical data on novel alloys [10,11], however, it is often too
time consuming for a more efficient investigation and the
maturation of complex alloy systems as a function of composition, TMT processing and microstructure.
Thus it becomes clear that the development of novel and
more efficient high throughput methods for the bulk synthesis, TMT processing and following investigations of
structural materials is of great interest. It is aimed at drastically increasing both the scientific and engineering efficiency by reducing the time between an alloy design idea
and final evaluation of the material’s mechanical and
microstructural properties from several weeks or even
months down to hours. Materials exhibiting desirable
properties (“hits”) can be readily selected from the investigated composition/TMT matrix and then scaled up via
conventional methods for more detailed investigations
based on the information collected. Rapid and systematic
screening of prototype alloys and TMT setups, ideally
employed in concert with theory-based, alloy-sensitive simulation methods [12–15], thereby allows the efficient and
knowledge-based design of structural materials.
Steels containing high amounts of manganese (Mn), aluminium (Al) and carbon (C) were here chosen as an example material system to utilize the novel approach outlined
above. Also referred to as “Triplex” steels, these materials
exhibit a low mass density, high strength and excellent ductility in comparison with established steels for structural
engineering applications [10,16,17]. The authors derived
this attractive property profile from an austenitic or austenitic/ferritic matrix, both stabilized and strengthened by
alloying iron (Fe) with Mn (18–28 wt.%) and C (0.7–1.2
wt.%), together with Al addition (3–12 wt.%) for low specific weight and improved corrosion resistance [10,18–20].
Profound changes in the mechanical properties can be
achieved by ageing the material after solution annealing
and quenching. More specifically, j-Al(Fe,Mn)3C carbides
were found to precipitate from the matrix during ageing,
growing from C-enriched areas, most probably formed
via spinodal decomposition during quenching [17,21,22].
The further development of Fe–Mn–Al–C alloys must
involve simultaneous investigation of the respective micro-
4951
structural phenomena by high resolution characterisation
techniques such as atom probe tomography (APT) and
transmission electron microscopy (TEM). Such techniques,
however, require increased efforts in sample preparation
and experimental procedures [23–26] and are, therefore,
preferably applied only to those compositions and microstructures that reveal the most interesting characteristics
or properties. The wide range of possible chemical compositions in the Fe–Mn–Al–C quaternary system, combined
with a large matrix of possible TMT routes (especially ageing time and temperature), justifies the high level of motivation for the development and use of rapid alloy
prototyping (RAP) techniques as an efficient tool for structural materials development.
2. Objective
The objective of this work was to present a new
approach to the rapid investigation of bulk metallic structural materials, whereby the mechanical properties of a
group of alloys systematically varied in terms of chemical
composition and the imposed TMT treatments can be evaluated simultaneously and thus with a higher throughput
than with conventional methods and step by step iteration
of these parameters. The approach is referred to as RAP.
As an exemple here we study a novel group of 30Mn–
1.2C steels in terms of the effects of varying Al content
and different ageing conditions on the microstructure and
properties. Selected results are directly compared with data
obtained from conventionally synthesised and processed
samples to validate the approach.
3. Materials and methods
3.1. Production of samples
Fig. 1a schematically illustrates the experimental process
of our RAP approach, which is detailed below. The primary synthesis of 30Mn–1.2C–xAl (wt.%) samples was
conducted via melting and casting in a vacuum induction
melting (VIM) furnace (heating power 60 kW, 4 kHz,
argon (Ar) atmosphere of 400 mbar). The chemical compositions of the alloys in this study (target/actual values by
wet chemical analysis) are listed in Table 1. Fig. 1b illustrates the device installed inside the furnace to create the
five different alloys in one casting operation. Five copper
(Cu) moulds with respective internal volumes of
10 50 150 mm3 was moved step by step on a linear
stage by an electric drive and successively filled with melt
from an ingot of 4 kg basic capacity. After each cast, the
composition of the melt remaining in the ingot was precisely altered by adding precalculated amounts of the alloying elements (in this case Al) through the furnace air-lock.
Each mould was formed into a 20 mm high 45° funnel at
the top for ease of casting and placed on a water-cooled
Cu plate. Thus the shrinkage during solidification occurred
at the top of the enlarged funnel and not as cavities within
4952
H. Springer, D. Raabe / Acta Materialia 60 (2012) 4950–4959
(a)
(b)
Hot rolling,
cutting
Multiple
casting
wt.%
Cu moulds
stage movement
(c)
clamped segments
...
σ
ε
...
Testing and
metallography
Sample
preparation
wt.%
Heat treatments
(matrix)
°C, h
(d)
tensile specimen
rolling direction
Fig. 1. Illustration of the sample production process: (a) schematic sketch of the RAP approach with the differently shaded colours referring to the
different chemical compositions; (b) device for casting five alloys into separate Cu moulds placed on a linear stage; (c) five segments after completion of the
heat treatment clamped together for simultaneous spark erosion; (d) one segment after completion of the spark erosion preparation of three tensile
specimens. RAP, rapid alloy prototyping.
Table 1
Target (left) and actual values (right) for the chemical compositions in
wt.% in the RAP experiments obtained by wet chemical analysis.
Target values (wt.%)
Actual values (wt.%)
Fe–30Mn–1.2C
Fe–30Mn–1.2C–2Al
Fe–30Mn–1.2C–4Al
Fe–30Mn–1.2C–6Al
Fe–30Mn–1.2C–8Al
Fe–30.5Mn–1.23C
Fe–31.0Mn–1.20C–2.12Al
Fe–31.4Mn–1.19C–4.30Al
Fe–31.3Mn–1.20C–6.42Al
Fe–31.6Mn–1.17C–9.04Al
RAP, rapid alloy prototyping.
the bulk. Removing the top (remove the funnel) and bottom (for chemical analysis) from the cast blocks after cooling to room temperature left five 10 50 130 mm3
rectangular blocks.
The 10 mm thick blocks were then hot rolled (air atmosphere) at 1100 °C to 2 ± 0.1 mm thick and 500 mm long
sheets. After the last rolling pass the sheets were reheated
to 1100 °C, quenched in water and cut perpendicular to
the rolling direction into 55 mm long segments. This gave
nine segments with the dimensions 2 60 55 mm for
each alloy (45 in total). Punch marking of the segments
with codes for both alloy composition and the applied heat
treatment ensured reliable identification of the specimens
throughout the following processing steps. Homogenization of the segments was performed by annealing at
1100 °C for 2 h under an Ar atmosphere, followed by water
quenching in clamps to minimise distortion. Ageing of the
segments was performed in air at 450, 500, 550, and 600 °C
for 0, 1 and 24 h at each temperature (0 h representing the
as-homogenised state), followed by oil quenching. This
resulted in a matrix of 45 different alloy/TMT conditions.
Scales were removed from the surfaces by low pressure, fine
grit sand-blasting after completion of the heat treatments.
Samples for mechanical testing and microstructure
investigation were prepared from the segments by spark
erosion. In order to increase the efficiency of this gentle
but rather slow technique (cutting speed 6 mm s 1) five
H. Springer, D. Raabe / Acta Materialia 60 (2012) 4950–4959
segments were clamped on top of each other and simultaneously cut, as shown in Fig. 1c. Tensile samples were prepared with a width of 5 mm (cross-sectional area 10 mm2),
gauge length of 17.89 mm (equivalent to A5 proportional
samples) and a longitudinal axis parallel to the rolling
direction. A thin 0.2 mm bridge was left to keep the cut
specimens attached to their respective segments, so that
they could be identified and separated from the marked
segment just before tensile testing. Three samples were prepared for each material and heat treatment condition, as
shown in Fig. 1d; leftover material was used for hardness
testing and metallographic preparation. RAP sample identification throughout this work follows the respective target alloy compositions in wt.% (Table 1).
For reference and validation of the RAP results tensile
samples of two alloys were produced by conventional metallurgy. Single charge melts of Fe–30Mn–1.2C alloys with
additions of 2 and 8 wt.% Al, respectively, were cast into
a 40 60 200 mm3 Cu mould using the same VIM furnace as for the RAP experiments. The surfaces of the cast
blocks were then milled and the alloys hot rolled at 1100 °C
(air atmosphere) to 10 mm thickness and quenched in
water. After homogenisation at 1100 °C for 2 h (Ar atmosphere) and water quenching the plates were cut by spark
erosion into strips (10 10 mm cross-section), from which
cylindrical tensile samples were lathed (DIN 50125 form B,
6 mm diameter, longitudinal axis parallel to the rolling
direction). Ageing was performed in air at 450 °C for 1 h,
followed by oil quenching.
3.2. Mechanical testing and characterization
Tensile testing of the RAP samples was conducted at
room temperature with an initial strain rate of 10 3 s 1
using a hydraulic Instron 8511 testing machine. Hardness
(Brinell HBW, ball diameter 2.5 mm) was measured with
a Wolpert DIA Testor 2RC on the outer surface of the segments after grinding them with 1000 grit. All values
obtained from both tensile and hardness tests represent
the averages of three individual measurements for each
alloy/TMT combination. Cross-sectional areas of selected
samples were prepared in the plane perpendicular to the
rolling direction by grinding and polishing using standard
metallographic techniques. The cross-sections were investigated by optical microscopy (OM) with a Zeiss Axiophot 1.
A Zwick/Roell Z100 machine was used for tensile testing of the conventionally fabricated samples at room temperature at a starting strain rate of 10 4 s 1. Two tests were
carried out for each alloy.
4953
evaluated within 35 h. The mechanical properties of these
45 different material conditions (i.e. in total 135 tensile tests
and hardness measurements) are shown as an overview in
Fig. 2 in terms of the yield strength (YS) (Fig. 2a), ultimate
tensile strength (UTS) (Fig. 2b), total elongation (TE)
(Fig. 2c) and hardness (Fig. 2d). The results are plotted
according to the systematically varied Al content (Table 1)
and colour coded according to the individual ageing conditions. Pronounced effects associated with the changes in
chemical composition and ageing parameters on the
mechanical behaviour of the materials can be clearly distinguished. For the reference material (no Al addition, i.e. Fe–
30Mn–1.2C) the best mechanical behaviour was observed
for the as-homogenised, non-aged state. This alloy was
characterised by a YS of 360 MPa, strong work hardening
(UTS = 830 MPa) and high ductility (TE = 77%). Ageing
of the Fe–30Mn–1.2C alloy showed that the YS was virtually unchanged and the hardness slightly increased. However, ageing greatly reduces both the UTS and TE. This
embrittlement becomes most apparent for long ageing
times (24 h) and higher temperatures (>500 °C). For the
alloy Fe–30Mn–1.2C–8Al, i.e. the material with the highest
Al concentration, the opposite trend applied. Without ageing the mechanical data are similar to those of the Al-free
alloy, with only a slight change in YS (increase), UTS and
TE (decrease). Ageing treatments for 1 h, however, led to a
simultaneous improvement in YS, UTS and hardness
(increasing with temperature) and to only a slight drop in
TE. Ageing of alloy Fe–30Mn–1.2C–8Al for 24 h further
increased YS, UTS and hardness to values almost twice
as high as in the as-homogenized state, but also drastically
reduced the ductility. The mechanical data for the alloys
with intermediate Al contents appear as the superimposition of the two different behaviours described above: as a
general rule, the values obtained for the alloys with 2, 4
and 6 wt.% Al lie between the respective data from alloys
without and with 8 wt.% Al. The alloys with Al additions
of 4 and 6 wt.% especially are only very weakly affected
by the applied ageing treatments in terms of their mechanical data compared with the alloys Fe–30Mn–1.2C (weakening/embrittlement)
and
Fe–30Mn–1.2C–8Al
(strengthening).
In order to better visualise the above described trends in
the mechanical behaviour for the investigated composition/
TMT matrix selected results are presented in Fig. 3.
Changes in YS, UTS, TE and hardness over the ageing
time at 550 °C of the alloys Fe–30Mn–1.2C, Fe–30Mn–
1.2C–4Al and Fe–30Mn–1.2C–8Al are displayed in
Fig. 3a–c, respectively. The experimental scatter is indicated by the error bars in Fig. 3a.
4. Results
4.2. Microstructure of the RAP samples
4.1. Mechanical testing of the RAP samples
Following the high throughput procedure outlined
above five alloy compositions, each exposed to nine
respective heat treatments, were produced, processed and
Fig. 4 shows optical micrographs after the different processing steps for the example alloy Fe–30Mn–1.2C at different magnifications. No cracks, pores or macro-segregations
can be observed after casting in the overview image on the
4954
H. Springer, D. Raabe / Acta Materialia 60 (2012) 4950–4959
(b)
900
1000
800
900
UTS
S / MPa
M
YS / MPa
(a)
700
600
500
800
700
600
400
500
300
0Al
2Al
4Al
6Al
8Al
0Al
(c)
2Al
4Al
6Al
8Al
Alloy (wt.%) Fe–30Mn–1.2C–
Alloy (wt.%) Fe–30Mn–1.2C–
(d)
80
400
H dness / H
Hard
HBW
70
TE / %
T
60
50
40
30
350
300
250
200
20
150
10
0Al
2Al
4Al
6Al
8Al
0Al
Alloy (wt.%) Fe–30Mn–1.2C–
Aging treatments:
500°C, 24h
2Al
4Al
6Al
8Al
Alloy (wt.%) Fe–30Mn–1.2C–
as-homogenised
450°C, 1h
450°C, 24h
500°C, 1h
550°C, 1h
550°C, 24h
600°C, 1h
600°C, 24h
Fig. 2. Overview of the mechanical properties obtained from the RAP experiments as a function of alloy composition (Table 1) and the applied ageing
treatment: (a) yield stress (YS); (b) ultimate tensile stress (UTS); (c) total elongation (TE); (d) hardness. RAP, rapid alloy prototyping.
left-hand side of Fig. 4a. The high magnification image on
the right-hand side reveals a coarse dendritic microstructure that is typical of as-cast alloy microstructures. Hot
rolling and water quenching (Fig. 4b) result in a fully
recrystallized microstructure (grain size 20 lm) with some
retained micro-segregations (dark lines in the micrographs). Within this context it is important to underline
that chemical homogenization can usually be better
obtained via TMT involving static, post-dynamic or
dynamic recrystallization, rather than by a static heat treatment alone. This is due to the fact that recrystallisation
involves the motion of high angle grain boundaries, which
provide much higher diffusion coefficients compared with
bulk diffusion.
Optical micrographs showing the microstructure of
selected samples after complete processing and heat treatment are presented in Fig. 5. Fig. 5a–c corresponds to
the Fe–30Mn–1.2C-based alloys with the addition of 0, 4
and 8 wt.% Al, respectively. Microstructures in the ashomogenised state (unaged) are shown on the left-hand
side of Fig. 5, those after ageing at 550 °C for 24 h on
the right-hand side, respectively. In the as-homogenized
state the three alloys exhibit almost identical austenitic
microstructures with an average grain size of about
80 lm, few twins (increasing with higher Al content) and
no apparent micro-segregations. After ageing, however,
pronounced differences between the three alloys could be
observed. Coarse particles with a diameter of 10 lm
appear at the grain boundaries of the Al-free alloy (Fe–
30Mn–1.2C), most probably consisting of a pearlitic ferrite/(Fe,Mn)3C microstructure. The addition of Al to the
alloy Fe–30Mn–1.2C–4Al apparently constrained the formation of those phases during ageing, as the number density and size of the particles was now significantly lower
than in the Al-free material and only thin films appeared
on the grain boundaries of the alloy with 4 wt.% Al. A further increase in Al content (alloy Fe–30Mn–1.2C–8Al)
resulted in the complete absence of grain boundary particles during ageing, but a large number of unevenly distributed small particles appeared within the grains, giving them
a darker contrast.
4.3. Tensile testing of the conventionally produced samples
For reference and comparison Fig. 6 shows exemplar
engineering stress–strain curves for the alloys Fe–30Mn–
1.2C–2Al (red) and Fe–30Mn–1.2C–8Al (blue) after ageing
at 450 °C for 1 h, obtained by following the conventional
H. Springer, D. Raabe / Acta Materialia 60 (2012) 4950–4959
80
900
70
800
60
700
50
600
40
500
30
400
20
300
10
0
1
TE / % , hardness
YS , UTS / MPa
1000
/ 1/10 HBW
Fe—30Mn—1.2C
(a)
24
5.1. Microstructure and mechanical properties of the 30Mn–
1.2C–xAl steels
900
70
800
60
700
50
600
40
500
30
400
20
300
10
1
/ 1/10 HBW
80
TE / % , hardness
YS , UTS / MPa
Fe—30Mn—1.2C— 4Al
1000
0
24
Aging time at 550ºC / h
80
900
70
800
60
700
50
600
40
500
30
400
20
300
10
0
1
TE / % , hardness
YS , UTS / MPa
1000
/ 1/10 HBW
Fe— 30Mn— 1.2C—8Al
(c)
UTS (810 to 890 MPa) and almost identical ductility (76
to 73%) of the RAP samples. While this trend (slightly
higher YS and lower UTS) is the same for the alloy Fe–
30Mn–1.2–2Al, the RAP samples now exhibited significantly less ductility than the conventional specimens (49
compared with 84%) under this specific material condition.
The unusually large elastic strain that is apparent in Fig. 6b
can be attributed to slight deformation (bending) of the
RAP samples, which could not be completely avoided
despite the clamping procedures applied during quenching
of the thin segments.
5. Discussion
Aging time at 550 ºC / h
(b)
4955
24
Aging time at 550ºC / h
Fig. 3. Selected mechanical test results from the RAP experiments: yield
strength (YS), ultimate tensile strength (UTS), total elongation (TE) and
hardness as functions of ageing time at 550 °C for alloys: (a) Fe–30Mn–
1.2C; (b) Fe–30Mn–1.2C–4Al; (c) Fe–30Mn–1.2C–8Al.
synthesis and processing (Fig. 6a) and RAP (Fig. 6b)
approaches. For the alloy Fe–30Mn–1.2C–8Al both the
RAP and conventional data are in the same range, with
an only lightly increased YS (610 to 540 MPa), lower
The novel bulk RAP approach introduced in this work
provides, for the first time, a systematic evaluation of the
compositional and thermo-mechanical trends associated
with a change in the Al content of a group of Triplex steels
with high Mn and C concentrations. We observed that
without the addition of Al to the 30Mn–1.2C steels the
most favourable mechanical properties were obtained for
the as-homogenised state (Figs. 2 and 3a). The observed
properties are in reasonable agreement with data reported
for Mn–C alloyed TWIP steels of a similar chemical composition [27,28]. The observed embrittlement during ageing
can be related to the formation of the coarse, pearlitic particles on the grain boundaries [27] (Fig. 5a).
High amounts of Al (8 wt.%), on the other hand, result
in pronounced strengthening during ageing, depending on
the time and temperature (Figs. 2 and 3c), and no coarse
particles could be observed in this case (Fig. 5c). In the
light of previous results this typical precipitation hardening
behaviour, which allows tuning of the strength and ductility, can be explained by the formation and growth of j carbides during ageing [17,18,21]. Due to their reportedly
small size, which is of the order of several nanometres,
the j carbides could not be reliably detected or identified
in the high throughput RAP OM observations conducted
in this study. The darker particles visible in Fig. 5c might
be linked to j carbides.
Alloys with intermediate Al concentrations (about 2–6
wt.%) do not offer mechanical properties on the same level
compared with the aforementioned extreme cases under
their respective optimal conditions (i.e. after the respective
most suitable ageing treatments). On the other hand, a
much smaller influence of the ageing parameters on tensile
behaviour can be observed in these cases (Figs. 2 and 3b).
Within the limitations of this study (confined range of
applied heat treatments, OM investigations, etc.) this
improved stability of the mechanical properties during
thermal exposure can be attributed to a concerted formation of j carbides and grain boundary pearlite, balancing
the strengthening and embrittlement effects of intermediate
amounts of Al.
4956
H. Springer, D. Raabe / Acta Materialia 60 (2012) 4950–4959
(a)
Normal direction
Transverse direction
2 mm
µ
µ
µ
(b)
Fig. 4. Optical micrographs of the alloy Fe–30Mn–1.2C from different stages of the RAP sample production process at different magnifications: (a) after
casting; (b) after hot rolling and water quenching. RAP, rapid alloy prototyping.
In general it should be underlined that detailed investigations of the role of j carbides and pearlite particles on
the deformation mechanisms, as well as the precipitation
type and the structural nature of the j carbides (i.e. spinodal vs. nucleation/growth), require higher resolution techniques, such as TEM or APT. Nonetheless, the
mechanical data on both RAP (Figs. 2 and 3) and conventionally synthesised and processed alloys (Fig. 6a) are in
good agreement with previously reported values for Fe–
Mn–Al–C steels [17,18]. The RAP results suggest that
future efforts regarding more detailed nanostructural investigations should focus on such high-C triplex steels with
high Al concentrations (>8 wt.%), as they offer the possibility of covering the widest range of mechanical properties
via ageing treatments (scalability, Fig. 2) and exhibit the
lowest possible specific weight of all such steels.
5.2. Evaluation of the RAP approach
Following the RAP approach introduced in this work,
45 different material conditions were synthesised, processed
and tested within 35 h (Fig. 1a). Investigations of the same
matrix, i.e. five different alloy compositions combined with
nine different heat treatments, would require a minimum of
about 6 weeks when using conventional metallurgical, processing, and testing methods. For both cases the effect of
the TMT parameters on the duration of such systematic
combinatorial experiments, especially the ageing time,
needs to be taken into account. The main key ingredients
to exploiting the full potential of an accelerated investigation of bulk structural alloys via RAP are the multiple
casting operation (five alloys instead of one, Fig. 1b),
simultaneous sample preparation by spark erosion (15
samples at the same time, Fig. 1c and d) and, most importantly, the efficient sequential adaptation of all processes
throughout the entire chain of experiments. It should be
noted that our approach is a combination of efficiently
adapted but already established processing steps. This
results in robust processing, i.e. less calibrating or interchecking between steps is needed, thus ensuring fast
processing and reliable data. Furthermore, the modular
concept offers the possibility of replacing individual processing steps with additional or different techniques, for
example exchanging synthesis via casting by powder-based
processes, while maintaining the fast sequential TMT and
testing procedures. The RAP synthesis via bulk casting
and hot rolling applied here, however, approximates processes and microstructures which are very similar to those
of structural materials produced on an industrial scale.
While alternative methods of mechanical probing can be
used, tensile testing appears to be the most representative
method for the evaluation of the mechanical behaviour of
structural materials, within the specific limitations of our
approach discussed later in the text (surface effects). Alternatively, indentation experiments can, for example, reveal
hardness and even certain strength measures, but they provide only a limited insight into the ductility of the investigated bulk materials. This is clearly illustrated by the
ageing-induced embrittlement of alloy Fe–30Mn–1.2C in
our study. The TE values decreased significantly with
ageing time and temperature, whereas the hardness data
showed a slight increase (Figs. 2a and d and 3a).
Some deviation in chemical composition between the
target and actual values of the RAP samples (Table 1)
H. Springer, D. Raabe / Acta Materialia 60 (2012) 4950–4959
As-homogenised
(a)
4957
Aging treatment 550 ºC, 24h
Fe—30Mn—1.2C
Normal direction
Transverse direction
Fe—30Mn—1.2C—4Al
(b)
Fe—30Mn—1.2C—8Al
(c)
Fig. 5. Optical micrographs of selected RAP samples in the as-homogenized state (left pictures) and after ageing at 550 °C for 24 h (right pictures): (a) Fe–
30Mn–1.2C; (b) Fe–30Mn–1.2C–4Al; (c) Fe–30Mn–1.2C–8Al. RAP, rapid alloy prototyping.
(a)
(b)
Conventional
900
800
800
700
Stress / MPa
700
Stress / MPa
RAP
900
600
500
400
300
600
500
400
300
200
200
100
100
0
0
0 10 20 30 40 50 60 70 80 90
Strain / %
Aging treatment 450°C, 1h:
0 10 20 30 40 50 60 70 80 90
Strain / %
Fe–30Mn–1.2C–2Al
Fe–30Mn–1.2C–8Al
Fig. 6. Engineering stress–strain plots of the alloys Fe–30Mn–1.2C–2Al (red curves) and Fe–30Mn–1.2C–8Al (blue curves) after aging at 450 °C for 1 h:
(a) conventionally synthesized and processed (single charge casting, cylindrical samples); (b) exemplary RAP samples. RAP, rapid alloy prototyping.
occurs as multiple charging is more difficult than conventional single charge casting, and the deviations naturally
become larger with ongoing casting/alloying procedures.
Nonetheless, the achieved chemical concentrations are
accurate enough for the intended screening of the alloy system under investigation. While the deviations in terms of
4958
H. Springer, D. Raabe / Acta Materialia 60 (2012) 4950–4959
tensile properties between conventional and RAP samples
are typically below 10%, the sporadically occurring strong
relative drop in ductility and subsequent lower UTS in the
RAP experiments (Fig. 6) are mainly attributed to surface
effects. The improved surface quality (smoothness,
removed scales and decarburized areas) and the more compliant geometry of the cylindrical tensile samples reduced
the tendency for premature localization and rupture before
the maximum strength of the materials was reached. Other
influencing factors which highlight the dependency between
testing procedures and the obtained values are the higher
speed of tensile testing and, more importantly, the smaller
thickness of the RAP samples (2 mm compared with
6 mm). As thinner specimens reach the set temperature faster than thicker ones the effects of ageing (in this case a
higher YS and lower TE) will thus be more pronounced.
The reasons for the larger discrepancy in ductility for the
alloy Fe–30Mn–1.2C–2Al in contrast to the good agreement in the case of the alloys containing 8 wt.% Al, however, clearly requires more detailed analysis. Future
investigations on a growing number of different alloy systems will result in enhanced statistics, so that the precision
of RAP for screening mechanical properties can be further
improved.
As our approach is designed to identify compositional,
process-related and microstructural trends that are relevant
to the mechanical behaviour (Figs. 2 and 3), it allows reliable pre-selection of material parameters and highlights the
intended role of RAP as a complementary tool in structural
materials development. The surface effects described above
should leave the RAP data generally on the “safe” side, i.e.
slightly improved tensile properties (especially ductility)
can be expected after up-scaling a selected prototype alloy
via conventional methods. It should be stressed that a basic
knowledge of the investigated alloy system, via thermodynamic calculations or existing experience, remains as
important as for conventional synthesis and processing of
structural materials in order to select appropriate TMT
parameters and alloy compositions.
6. Summary and conclusions
In this study we have introduced a new approach to
achieving a higher throughput in the experimentally
guided design of bulk structural materials. We used a
group of 30Mn–1.2C–xAl steels (wt.%) as an example.
Tensile and hardness testing as well as optical microscopy
were performed on a matrix of 45 samples, i.e. five alloy
compositions (0–8 wt.% Al) and nine different ageing
treatments (450–600 °C for 0, 1 and 24 h). Additionally,
two conventionally synthesised and processed alloys of
similar chemical composition were investigated in order
to validate the RAP approach. The following conclusions
can be drawn.
1. Chosen here as an exemplary material system, the Fe–
Mn–Al–C steels investigated exhibited widely differing
characteristics, depending on their respective Al concentration. Ageing of the material without Al addition
resulted in the formation of coarse pearlitic particles at
the grain boundaries and subsequent embrittlement.
On the other hand, pronounced strengthening could be
observed after ageing the alloy with the highest Al concentration, most probably linked to the formation of
small and dispersed j-Al(Fe,Mn)3C carbides. Intermediate Al concentrations led to less favourable mechanical
properties compared with the alloy variants with no or
very high amounts of Al, but exhibited a greater thermal
stability.
2. Future investigations of Triplex steels using high resolution characterisation methods should give a more
detailed insight into the role of j-Al(Fe,Mn)3C carbides
on the deformation behaviour. Our results have shown
that it is particularly pertinent to focus on alloys containing high Al concentrations (>8 wt.%), as they exhibit
a low density and offer the possibility of covering a wide
and well tunable range of mechanical properties via ageing (scalability): YS, 500–940 MPa; UTS, 710–
1020 MPa; TE, 78–8%.
3. Following our RAP approach the 45 different material
conditions were synthesized, processed and tested in
35 h by employing modified and comprehensively
adapted casting, hot rolling and sample preparation
techniques. This represents a time advantage of a factor
of 6–10 compared with conventional metallurgical synthesis. As it allows fast screening of the mechanical
properties dependent on chemical and thermo-mechanical parameters, RAP has significant potential as a novel
tool for efficient alloy design, ideally employed in interactive cooperation with simulation and high resolution
characterisation methods.
4. A direct comparison with results obtained for conventionally synthesised materials reveals that the quality
of both the mechanical data and the obtained chemical
composition precision of the RAP synthesized alloys
was slightly reduced (typically below 10%). This minor
decrease was expected in view of the applied multiple
casting operations and the geometry of the tensile samples (e.g. surface effects). However, we observe that the
data quality obtained by the RAP approach was sufficient to spot relative trends in the bulk material behaviour. The absolute values obtained were in reasonable
agreement with the published data, allowing reliable
pre-selection of material parameters for a more detailed
analysis.
5. The applied synthesis route, via casting and hot rolling,
of the RAP approach is best suited to the investigation
of the most common structural materials, such as steels,
nickel, titanium or aluminium alloys. Future developments aim at a further increase in the processing speed
and applicability for different metallic systems, e.g. by
H. Springer, D. Raabe / Acta Materialia 60 (2012) 4950–4959
the implementation of alternative synthesis processes,
such as strip casting and powder metallurgy techniques,
respectively. The modular concept of RAP facilitates the
replacement of single processing steps (e.g. synthesis)
and expansion of the employed testing procedures (e.g.
fracture toughness) according to specific characterisation needs.
Acknowledgements
H. Springer wishes to thank G. Bialkowski, A. Bobrowski, M. Kulse, F. Rütters, F. Schlüter and J. Wichert for
providing considerable experimental support and experience. Dr I. Gutierrez-Urrutia is acknowledged for valuable
discussions and provision of the tensile testing data for the
conventionally synthesized alloys.
References
[1] Neidhardt J, Mráz S, Schneider JM, Strub E, Bohne W, Liedke B,
et al. J App Phys 2008;6:104.
[2] Mardare AI, Savan A, Ludwig A, Wieck AD, Hassel AW. Electrochem Acta 2009;54:5973.
[3] Mardare AI, Ludwig A, Savan A, Wieck AD, Hassel AW. Electrochem Acta 2010;55:7884.
[4] Abdulkadhim A, Takahashi T, Music D, Munnik F, Schneider JM.
Acta Mater 2011;59:6168.
[5] Gebhardt T, Music D, Ekholm M, Abrikosov IA, Vitos L, Dick A,
et al. J Phys Condensed Matter 2011;23:246003.
[6] Cui J, Chu YS, Famodu OO, Furuya Y, Hattrick-Simpers J, James
RD, et al. Nature Mater 2006;5:286.
4959
[7] Leslie WC, Hornbogen E. In: Cahn RW, Haasen P, editors. Physical
metallurgy. Amsterdam: Elsevier Science; 1996.
[8] Polmear IJ. Metallurgy of the light metals. London: Arnold; 1996.
[9] Gottstein G. Physical foundations of materials science. Berlin: Springer Verlag; 2004.
[10] Frommeyer G, Brüx U. Steel Res Int 2006;77:627.
[11] Raabe D, Ponge D, Dmitrieva O, Sander B. Scripta Mater
2009;60:1141.
[12] Friák M, Hickel T, Körmann F, Udyansky A, Dick A, von Pezold J,
et al. Steel Res Int 2011;82:86.
[13] Roters F, Eisenlohr P, Hantcherli L, Tjahjanto D, Bieler TR, Raabe
D. Acta Mater 2010;58:1152.
[14] Gebhardt T, Music D, Ekholm M, Abrikosov IA, von Appen J,
Dronskowski R, et al. Acta Mater 2011;59:1493.
[15] Music D, Basse FHU, Schneider JM. Crystal Growth Design
2010;10:4531.
[16] Chang KM, Chao CG, Liu TF. Scripta Mater 2010;63:162.
[17] Choo WK, Kim JH, Yoon JC. Acta Mater 1997;45:4877.
[18] Choi K, Seo CH, Lee H, Kim SK, Kwak JH, Chin KG, et al. Scripta
Mater 2010;63:1028.
[19] Banerji SK. Metal Progr 1978;113:59.
[20] Kimura Y, Handa K, Hayashi K, Mishima Y. Intermetallics
2004;12:607.
[21] Tjong SC. Mater Charact 1990;24:275.
[22] Sato K, Tagawa K, Inoue Y. Mater Sci Eng A 1989;111:45.
[23] Williams DB, Carter CB. Transmission electron microscopy: a
textbook for materials science. New York: Plenum Press; 1996.
[24] Browning ND, Chisolm MF, Pennycook SJ. Nature 1993;366:143.
[25] Seidmann DN. Ann Rev Mater Res 2007;37:127.
[26] Dmitrieva O, Ponge D, Inden G, Millán J, Choi P, Sietsma J, et al.
Acta Mater 2011;59:364.
[27] Bouaziz O, Allain S, Scott CP, Barbier D. Curr Opin Solid State
Mater Sci 2011;15:141.
[28] Lai HJ, Wan CM. Scripta Mater 1989;23:179.