Acta Materialia 97 (2015) 305–315

Contents lists available at ScienceDirect

Acta Materialia
journal homepage: www.elsevier.com/locate/actamat

Suppression of twinning and phase transformation in an ultrafine

grained 2 GPa strong metastable austenitic steel: Experiment and
simulation
Y.F. Shen a, N. Jia a,⇑, Y.D. Wang a, X. Sun b, L. Zuo a, D. Raabe c,⇑
a
Key Laboratory for Anisotropy and Texture of Materials (MOE), Northeastern University, Shenyang 110004, P.R. China
b
Computational Science and Mathematics Division, Pacific Northwest National Laboratory, Richland, WA 99352, USA
c
Max-Planck-Institut für Eisenforschung, D-40237 Düsseldorf, Germany

a r t i c l e i n f o a b s t r a c t

Article history: An ultrafine-grained 304 austenitic 18 wt.% Cr–8 wt.% Ni stainless steel with a grain size of 270 nm was
Received 18 May 2015 synthesized by accumulative rolling (67% total reduction) and annealing (550 °C, 150 s). Uniaxial tensile
Accepted 24 June 2015 testing at room temperature reveals an extremely high yield strength of 1890 ± 50 MPa and a tensile
strength of 2050 ± 30 MPa, while the elongation reaches 6 ± 1%. Experimental characterization on sam-
ples with different grain sizes between 270 nm and 35 lm indicates that both, deformation twinning
Keywords: and martensitic phase transformation are significantly retarded with increasing grain refinement. A crys-
Ultrahigh strength
tal plasticity finite element model incorporating a constitutive law reflecting the grain size-controlled
Ultrafine grain
Deformation
dislocation slip and deformation twinning captures the micromechanical behavior of the steels with dif-
Twin ferent grain sizes. Comparison of simulation and experiment shows that the deformation of
ultrafine-grained 304 steels is dominated by the slip of partial dislocations, whereas for coarse-grained
steels dislocation slip, twinning and martensite formation jointly contribute to the shape change.
Ó 2015 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction (CG) counterparts. The drop in ductility of UFG metals is attributed

to inhomogeneous deformation at microstructural level, leading to
Non-alloyed metallic materials can be equipped with ultrahigh early plastic localization and the premature onset of damage [2].
strength by employing heavy plastic deformation and/or by creat- For example, a nano-grained (NG) 316L austenitic stainless steel
ing ultrafine-grained (UFG) microstructures. Here UFG materials (with an average grain size of 40 nm) prepared via surface mechan-
are defined as polycrystalline solids with property-controlling ical attrition treatments exhibits an extremely high yield strength
grain structures below the scale of 1 lm. Typical processing routes of 1450 MPa, whereas the ductility is limited to a few percent [2].
include severe plastic deformation (SPD) [1,2], which decomposes Interesting strategies have been proposed for simultaneously
grains through dislocation accumulation and reorganization (poly- increasing the ductility and strength of UFG pure copper and
gonization). Several different SPD processes for manufacturing nanostructured alloys [13–17]. In pure copper, high ductility can
bulk materials typically include equal-channel angular pressing be achieved without sacrificing strength by introducing a high den-
(ECAP) [3], high-pressure torsion (HPT) [4] and accumulative sity of pre-existing twins and a large fraction of high-angle grain
roll-bonding (ARB) [5]. In case of steels, UFG structures can also boundaries. These encouraging results motivated the current study
be obtained via thermomechanical processing, exploiting the inter- in an effort to obtain both, high strength and good ductility from a
play between plastic deformation, recrystallization and phase bulk alloy.
transformation [6–11]. One of the main motivations for obtaining The 304 stainless steel (SS), a meta-stable austenite alloy, has
very fine grains is motivated by the increase in mechanical been widely used in many industrial areas due to its excellent
strength according to the Hall–Petch relationship [12]. However, combination of strength and formability, high corrosion resistance
such enabled UFG structured ultra-high strength metals often and weldability. 304SS contains approximately 18 wt.% Cr and
loose ductility compared to their conventional coarse-grained 8 wt.% Ni [18]. According to the carbon concentration, the 304SS
can be divided into three groups, i.e. 304L SS with less than
⇑ Corresponding authors. 0.03 wt.% carbon content, 304SS with 0.03–0.06 wt.% carbon, and
E-mail addresses: [email protected] (N. Jia), [email protected] (D. Raabe). 304H SS with a carbon content above 0.06 wt.%. The addition of

http://dx.doi.org/10.1016/j.actamat.2015.06.053
1359-6454/Ó 2015 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
306 Y.F. Shen et al. / Acta Materialia 97 (2015) 305–315

Cr enables corrosion resistance while Ni and C stabilize the auste- roller. The workability can be promoted by rolling at elevated tem-
nitic phase and tune its thermodynamic stability. During room peratures at an expense of accumulated strains. This process was
temperature torsion testing, 304L SS can withstand severe shear repeated until a total reduction of 67% was reached. The final
deformation (2.55 true strain), resulting in a remarkable volume dimensions of the steel sheets were 300  100  1.5 mm3. The
fraction of transformed bcc martensite of 63.5% [19]. However, sheets were air-cooled after rolling. Subsequently, a series of heat
synchrotron X-ray diffraction study on the phase transformation treatments were further conducted to obtain samples with various
kinetics and texture evolution of a 304L SS steel subjected to con- grain sizes. The annealing temperatures varied from 550 °C to
ventional compression testing [20] revealed phase transformation 950 °C with an increment of 100 °C. The holding time was 150 s
only at cryogenic temperatures (70 °C). An in-situ X-ray for the sample annealed at 550 °C and the other samples were held
micro-diffraction study in conjunction with uniaxial tensile load- for 180 s at the designed temperatures, respectively.
ing conducted at 63 °C substantiated the occurrence of a The as-prepared steel sheets were cut along the rolling direc-
phase-transformation-enhanced strain/stress accommodation tion (RD) into dog-bone shaped specimens, with a gauge length
mechanism in this material [21]. of 30 mm, width of 3 mm, and a final thickness of 1.5 mm after pol-
Metastable variants of such austenitic SSs have attracted great ishing. Uniaxial tensile tests were performed at an initial strain
interest because of their high strength and ductility when rate of 103 s1 using a SANS micro-force testing system under
equipped with an ultrafine grain size. [22]. In wet air, even the room temperature. A contactless MTS LX300 laser extensometer
resistance to corrosion of such UFG SS increases with decreasing was used to calibrate and measure the strain upon loading. Hard-
grain size, which is attributed to the effect that the many abundant ness measurements were performed with a MVD-401 Vickers
grain boundaries facilitate diffusion of Cr towards the surface dur- hardness testing machine, with a load of 5 kg and a loading time
ing oxidation [23,24]. Heavy cold-rolling in conjunction with sub- of 10 s.
sequent short-time annealing of 304SS is an efficient approach to Cross sections were cut from the gauge of the deformed sam-
refine the grain size. The basis for this processing is due to the ples for evaluating the martensite volume fraction, using X-ray
effect that the initial heavy cold rolling of a metastable austenitic diffraction (XRD) analysis. Samples were chemically cleaned first
SS leads to very fine lath- and dislocation cell-type transformation using a mixture of HCl and HNO3 in 2:1 ratio and then with
martensite which then transforms into ultrafine austenite grains C2H5OH to improve the precision of the analysis. The XRD mea-
upon annealing [22]. In addition, it is reported that warm rolling surement was performed on an X’Pert PRO diffractometer in con-
and thermal cycling [25] can lead to significant grain refinement junction with a Co anode operated at a Ka wavelength of
in a 304SS based on the mechanisms of continuous dynamic k = 0.1789 nm. The penetration depth of this hard X-ray into the
recrystallization [26–31]. steel samples amounts to a few micrometers. The quantitative esti-
Several works have shown that twinning and martensitic trans- mation of the a0 -martensite phase content was based on the prin-
formation play key roles for the deformation of 304SS and that the ciple that the volume fraction of one constitutive phase is
volume fraction of martensite formed during severe deformation is proportional to the integrated intensity of all diffraction peaks
strongly dependent on the pre-strain, stress and strain magnitude, for that phase. Then the volume fraction of martensite can be cal-
strain rate, and deformation temperature [21,32–38]. In that con- culated by measuring the integrated intensity of each reflecting
text earlier papers reported nanoscale embryos of martensite plane of the phase from a single XRD scan. Details of the calcula-
within the microstructure of a tensile deformed austenitic steel tion method can be found in [23]. To reveal the crystallographic
[21,36]. These various observations and phenomena encouraged textures of the steels after annealing, the {1 1 1} and {2 2 0} pole fig-
us to explore the feasibility of obtaining bulk UFG steels through ures of the annealed samples were measured by XRD using Co-Ka
severe plastic deformation, particularly via accumulative rolling. radiation in reflection geometry.
Therefore, in the present study, a series of austenitic stainless The microstructure of the deformed specimens was studied by
steels with different grain sizes (ranging from 270 nm to 35 lm) TEM. Specimens for TEM observation were cut from the deformed
were prepared by means of accumulative rolling and annealing. samples and mechanically polished to a thickness of about 60 lm.
Uniaxial tensile testing shows that the UFG steel with a grain size Subsequently the samples were polished with diamond paste, and
of 270 nm exhibits an extremely high yield strength of then the foils were thinned using a double jet electrolytic polisher
1890 ± 50 MPa and a tensile strength as high as 2050 ± 30 MPa, at a voltage of 32 V and a temperature between 10 and 5 °C.
while the elongation reaches 6 ± 1%. The conventionally synthe- TEM observations of the samples were performed in a Tecnai G2
sized reference sample with 35 lm grain size showed 220 MPa 20 microscope operated at an accelerated voltage of 200 kV.
yield strength, 1640 MPa tensile strength and 59% elongation to
fracture. To explore the micromechanical behavior of the UFG aus- 2.2. Microstructure characterizations
tenitic steels, transmission electron microscopy (TEM) observa-
tions of the deformed microstructure as well as crystal plasticity Fig. 1(a) shows the X-ray diffraction patterns of the 304SS sam-
finite element (CPFE) modeling of the tensile deformation for the ples at different conditions. The calculated results indicate that the
steels with various grain sizes are performed. martensite (a0 ) decreases with increasing annealing temperature
(Fig. 1(b)). For example, the martensite volume fraction is as high
as 73% for the as-cold-rolled sample; however, it decreases to
2. Experiments 10% after annealing at 850 °C for 180 s. The phenomenon indicates
that the martensite induced by cold-rolling has been reversed to
2.1. Experimental method austenite (c) during the annealing.
Our previous work, conducted using electron backscatter
The material used in this study is a commercial 304SS with diffraction (EBSD) for characterizing the microstructures of the
chemical composition of (wt.%): Cr 18.37, Ni 8.51, Mn 1.68, Si steels [39] has shown that for the samples annealed at 650 °C,
0.72, P 0.03, C 0.02, N 0.06, S < 0.01, balanced by Fe. The fully 850 °C and 950 °C for 180 s, the average grain size increases from
annealed sheets with an initial thickness of 6 mm were held for 800 nm to 10 lm and 35 lm with increasing annealing tempera-
600 s at 400 °C in an electrical furnace under air atmosphere, and ture. However, for the sample annealed at 550 °C for 150 s the
immediately rolled in a single-pass to 17% thickness reduction. grain size is too small to be discerned by EBSD. Therefore, in this
The rolling was performed at a roll speed of 300 mm s1 by a twin work TEM observations are performed to measure the grain size.
 Y.F. Shen et al. / Acta Materialia 97 (2015) 305–315 307

volume fractions of both martensite and deformation twinning

(a)
/
increase as the annealing temperature is elevated (Fig. 1(b)). This
α (101)
/
α (211) γ(311) surprising observation is attributed to the fact that a large amount
/ γ(220)
γ(111) γ(200) α (200) o of martensite and twins tend to be formed in the high
CR+950 C/180s
temperature-annealed samples during the subsequent uniaxial
CR+850oC/180s
tensile testing. The texture measurements reveal that after
Intensity

o
CR+750 C/180s 550 °C annealing a minor Goss orientation ðf1 1 0gh0 0 1iÞ with
o the maximum intensity of f(g)  3.9 exists (Fig. 3). With increasing
CR+650 C/180s
annealing temperature, this texture component weakens and the
o
CR+550 C/150s steel annealed at 950 °C shows a random crystallographic texture.

2.3. Mechanical properties determined by tensile tests

CR-67%
Fig. 4(a) shows the tensile true stress vs. strain curves for the
304 SS with various grain sizes at a strain rate of 103 s1 at room
40 50 60 70 80 90 temperature. In comparison with the reference coarse-grained
o
2θ ( ) steel (grain size 35 lm), the UFG samples exhibit significantly
higher strength. One can see that the tensile yield strength ry of
(b) 90 90 the sample with 270 nm grain size (at 0.2% offset) is
Before tension 1890 ± 50 MPa, i.e. about 8 times higher than that of the CG sample
Martensite volume fraction (%)

75 After tension 75 (200 MPa), and the ultimate tensile strength (rUTS) is
2050 ± 30 MPa. The Vickers hardness of the sample is 5.6 GPa. This
Twin volume fraction (%)

60 60 means that the strength of the sample is the same in tension and
compression, i.e. the material is fully dense and free of porosity.
For the sample with an average grain size of 800 nm, the values
45 45
for ry and rUTS are 1170 ± 30 MPa and 1350 ± 40 MPa, respectively.
The data show that both ry and rUTS increase with the refinement
30 30 of grain size while the failure strain is reduced with decreasing
150s
grain size. The elongation-to-failure (ef) is 6 ± 1% (0.06 true
180s
15 15 strain), 12 ± 2% (0.12 true strain) and 59 ± 2% (0.46 true strain)
180s for the samples with a grain size of 270 nm, 800 nm and 35 lm,
180s 180s
0 0 respectively. The reduced plasticity of the UFG samples is attribu-
ted to the reduced dislocation activity as will be discussed in
RT 550 650 750 850 950 Section 4.
o
Annealing temperature ( C)
3. Simulation of uniaxial tensile deformation
Fig. 1. (a) XRD patterns of the cold-rolled Fe–18.37Cr–8.51Ni–1.68Mn–0.72Si–
0.02C–0.06N steel along with those annealed at different temperature and time. (b)
The calculated volume fractions of martensite and deformation twin before and
3.1. Constitutive formulations
after tension, the annealing duration has been indicated for each data.
3.1.1. Flow rule
We use the finite strain kinematic framework in which the
deformation gradient, F, is multiplicatively decomposed according
Typical microstructures obtained at different areas of the sample
to:
are shown in Figs. 2(a)–(c). Many grains are equiaxed and the cor-
responding selected-area-electron-diffraction (SAED) pattern in F ¼ Fe Fp ð1Þ
Fig. 2(a) demonstrates that the mutual misorientation distribution
among the grains is random with many high angle grain bound- where Fe is the elastic part comprising the stretch, Ue, and the lat-
aries. The statistical result indicates that the average grain size of tice rotation, Re; Fp is the plastic part of the deformation gradient.
the sample is 270 nm. The initial plastic deformation gradient Fp0 is set to the inverse of
For the subsequently tensile deformed steels, as they exhibit the local crystal orientation, T1
0 , and evolves at a rate governed
quite different maximum elongations, the volume fraction of by the plastic velocity gradient Lp:
martensite and deformation twin obtained at various elongations
for the samples after 550 °C, 650 °C and 950 °C annealing are listed F_ p ¼ Lp Fp ð2Þ
in Table 1. At a strain of 6%, all steels show very low volume frac- The evolution of the crystal orientation with strain then follows
tions of the martensitic phase transformation and twinning, indi- from the polar decomposition Fe = Re Ue as T = Re. As suggested by
cating that dislocation slip is the main deformation mechanism. Kalidindi [40], the plastic velocity gradient, Lp, contains shear con-
At 12% strain, the significant increase of martensite (15 vol.%) tributions from dislocation slip systems (indexed by a) and twin-
as well as the occurrence of twins (8 vol.%) is identified in the ning systems (indexed by b):
650 °C annealed steel. However, in the 950 °C annealed steel, the
increase of deformation twins (13 vol.%) is more significant than X
12 X
12

that of the martensitic phase transformation (5 vol.%). This sug- Lp ¼ c_ a ma  na þ c_ b mbtwin  nbtwin ð3Þ
a¼1 b¼1
gests that with increasing straining mechanical twinning starts to
play a large role in the high temperature-annealed steel. Still in The vectors m and n denote the normalized directions and
this steel, when the maximum strain of 59% is reached, the marten- plane normals of the deformation systems on which shear occurs
site and twin volume fractions are 53% and 36%, respectively. Then at a rate of c_ . By omitting the volume fraction of the
it is obvious that at the ultimate strains for the studied steels, the non-twinned crystal portions in the contribution of dislocation slip
308 Y.F. Shen et al. / Acta Materialia 97 (2015) 305–315

Fig. 2. Typical TEM micrographs showing the morphology of grains in the cold-rolled Fe–18.37Cr–8.51Ni–1.68Mn–0.72Si–0.02C–0.06N steel after annealing at 550 °C for
150 s. An inset in (a) shows the corresponding selected-area-electron-diffraction (SAED) pattern.

Table 1
Volume fractions of martensite and deformation twin in the steels at different elongations. The statistics for the martensite and twin are obtained from the XRD analyses and TEM
observations, respectively.

Annealing temperature, average grain size Volume fraction 0% elongation 6% elongation 12% elongation 59% elongation
550 °C, 270 nm Martensite (%) 32 34 – –
Twin (%) 0 3 – –
650 °C, 800 nm Martensite (%) 23 26 38 –
Twin (%) 0 5 8 –
950 °C, 35 lm Martensite (%) 0 3 5 53
Twin (%) 0 5 13 36

in Eq. (3) (in contrast to [40]), we (i) assume that twins can be 3.1.2. Dislocation slip
sheared by dislocation slip in a compatible manner to the sur- Conventional flow rules use a reference shear rate and a rate
rounding matrix, and (ii) ignore any potentially different evolution sensitivity exponent which are typically constant. Here, following
of slip resistance within them. For the austenitic phase, 12 Blum and Eisenlohr [41], the evolution of the dislocation densities
f1 1 1gh1 1 0i dislocation slip systems and 12 f1 1 1gh1 1 2i twinning is related to dislocation multiplication, dipole formation as well as
systems are considered. For the martensitic phase, two types of slip dislocation annihilation. The flow rule describes thermally acti-
systems with a common h1 1 1i direction, i.e. f1 10gh1 1 1i and vated dislocation motion through forest dislocations. The shear
f1 1 2gh1 1 1i, are considered, but no bcc twinning is included. rate of the slip system a is:
 Y.F. Shen et al. / Acta Materialia 97 (2015) 305–315 309

Fig. 3. {1 1 1} and {2 2 0} pole figures for the cold-rolled steel after annealing at 550 °C for 150 s.

"  p q #
Q jsa j different slip systems a and a0 . For fcc metals, the possible interac-
c_ a ¼ qasgl bv 0 exp  0 1 a signðsa Þ ð4Þ tion types include self, coplanar, collinear, orthogonal, glissile and
kB T s^
sessile [43]. For bcc metals, only the self-interaction and the inter-
where sa is the current resolved shear stress; s ^a is the slip resis- action between different slip systems are considered.
tance governed by the dislocation population; qasgl is the unipolar
dislocation density; b is the length of the Burgers vector; v 0 is the 3.1.3. Mechanical twinning
dislocation velocity of the slip system when subjected to a stress The model accounts for the evolution of the twinned volume
^a ; Q0 is the activation energy for dislo-
equal to the slip resistance s fraction. Following the twin nucleation model of Mahajan et al.
cation slip; kB and T denote the Boltzmann constant and tempera- [44], we assume that the critical twin nucleation event is the col-
ture, respectively; p and q are numerical parameters to adjust the lective bow-out of three partial dislocations between pinning
obstacle profile [42]. The slip resistance s^a depends on the local dis- points that are separated by a critical length L0. The associated crit-
location densities as: ical stress for twin formation then reads
!1=2 csf 3Gbtwin
X
Nslip
s^twin ¼ þ ð6Þ
^a a0 a0
s ¼ ssolute þ Gb naa0 ðqsgl þ qdip Þ ð5Þ 3btwin L0
a0 ¼1
where btwin is the magnitude of the Burgers vector of the moving
With ssolute being a constant glide resistance stemming from partials and csf is the stacking fault energy. Using this critical stress
0
solid solution effects, G the shear modulus, qadip the dipolar disloca- formulation the mechanically driven twin nucleation rate is
tion density, and naa0 characterizes the interaction strength among expressed as
310 Y.F. Shen et al. / Acta Materialia 97 (2015) 305–315

(a) 3000 c_ b ¼ ð1  f Þctwin V b N_ b ð11Þ

Experiment
where ctwin is the characteristic twin shear. Details of the disloca-
2500 Simulation
270nm tion and twinning models are described in [41,42]. For dislocation
slip the rate equation of the dislocation density evolution is in the
True stress (MPa)

2000 a
form dq ¼ jdbkca j, where ka is the effective (harmonic mean) distance
between obstacles. With the presence of twins in the microstruc-
1500 ture, the evaluation of ka then has to account for twin boundaries
800nm as additional obstacles against dislocation motion. The slip-twin
1000 interaction parameter nab is introduced for the slip system a and
twin system b. If planes a and b are coplanar or b is a cross slip plane
35 um
500 for a, nab assumes a value of 0. In all other cases nab is 1. The mean
free path of dislocations is finally calculated as
0 X
Ntwin
0.0 0.1 0.2 0.3 0.4 0.5 0.6
1 1 1
a ¼ þ nab b ð12Þ
k dgrain b¼1 dtwin
True strain

10000
3.2. Model set-up and simulation runs
(b) Experiment
8000 Simulation Based on the constitutive laws outlined above, the CPFE simula-
tions of the steels with their specific grain sizes and phase compo-
270nm sitions (as listed in Table 1) under tensile loading were carried out
dσ//dε (MPa)

6000
using the finite element solver MSC.Marc2013 in conjunction with
the open source user defined material subroutine coded in the
4000 DAMASK package [42,45]. The FE mesh includes 8000
800nm
eight-noded, isoparametric, three-dimensional brick elements.
35 um For the steels consisting of two phases, the elements that are given
2000 specific mechanical properties of austenite and martensite are
mixed homogeneously. By assigning one grain orientation to the
eight integration points pertaining to each finite element, the total
0
number of the initial orientations representing the random starting
0.0 0.1 0.2 0.3 0.4 0.5 0.6
texture of the steel, as observed after the heat treatment, amounts
True strain to 64,000 (8000  8). For modeling uniaxial tensile testing, peri-
odic boundary conditions are adopted and a strain rate of
Fig. 4. (a) Measured and simulated true stress–strain curves for the steels with
ultrafine grains (the average grain size is 800 nm and 270 nm, respectively) during
103 s1 is imposed [46]. The constitutive parameters characteriz-
uniaxial tensile tests, in comparison with the coarse-grained material (with an ing the austenite are determined by fitting the macroscopic stress
average grain size of 35 lm). Tensile loading is imposed parallel to the rolling vs. strain curve as well as the twin volume fraction vs. strain curves
direction of the as-prepared steel sheets. (b) Measured and simulated strain- obtained from the tensile test on the reference sample with an
hardening responses for the steels with different grain sizes.
average initial grain size of 35 lm. It should be mentioned that
as the slip resistance is related to the strength in a solid solution,
   the critical resolved shear stress for activating dislocation slip var-
s^twin r
N_ b ¼ N_ 0 exp  ð7Þ ies among the steels with different grain sizes. Then, based on the
b s
Hall–Petch law that is attributed to dislocation pile-ups against
where sb is the resolved stress in the twin system, and N_ 0 and r are grain boundaries, for the austenitic phase the ssolute values of
fitting parameters. It is further assumed that the so formed twin 0.03, 0.37 and 0.65 GPa are used for steels with a grain size of
expands instantaneously until encountering an obstacle, such as a 35 lm, 800 nm and 270 nm, respectively. For the UFG steels, as
grain or twin interface. The volume of the new deformation twin is the austenitic and the martensitic grains collectively contribute
to the overall yield strength, the initial slip resistance of the
p 2
Vb ¼ skb ð8Þ martensite is determined by fitting the corresponding macroscopic
6
stress vs. strain curves. The ssolute values of 1.59 and 2.50 GPa are
Here, s is a constant twin thickness, and kb is the effective distance used for the martensite in steels with a grain size of 800 nm and
between obstacles against twin growth, 270 nm, respectively. For both phases, all other constitutive
parameters are identical for the differently grain-sized steels, as
1 1 X
12
1
¼ þ n 0 0 ð9Þ listed in Table 2.
kb
dgrain b0 ¼1 bb db
twin
3.3. Simulation results
dgrain is the constant grain size, and the twin spacing dtwin naturally
evolves with the twin volume fraction as according to
Fig. 4(a) shows the measured and the predicted true stress–
b
dtwin ¼ sð1  f Þ=f
b
ð10Þ strain curves for the steels with different grain sizes, respectively.
One could note that some discrepancies exist between experi-
with f being the total volume fraction of the mechanical twins and ments and predictions for the studied steels. For the steel with a
b
f the volume fraction for the twin system b. The twin–twin inter- grain size of 35 lm, as the current simulation does not incorporate
action parameter nbb0 is 0 for coplanar twin systems b and b’ and 1 the martensitic phase transformation at large strains, at a true
for non-coplanar systems. The resulting shear rate for each twin strain above 0.38 the predicted strength is lower than that
system is calculated as obtained by the measurements. For the UFG steels, the overestima-
 Y.F. Shen et al. / Acta Materialia 97 (2015) 305–315 311

Table 2
Constitutive material parameters of the austenitic phase and the martensitic phase used for the crystal plasticity simulations. C11, C12 and C44 are the single crystal elastic
constants. Regarding the thermal activated annihilation of dislocations, the following parameters are used: qsgl,0 and qdip,0 are the initial dislocation density and dipole density,
respectively; v0 is the initial glide velocity; D0 is the bulk diffusion coefficient; and QSD is the activation energy for dislocation climb. The asterisk (*) indicates fitting parameters.

Phase C11 C12 C44 qsgl;0 qdip;0 b

Austenite 204.6 GPa 137.7 GPa 126.2 GPa 1.0  1012 m2 1.0 m2 2.58  1010 m
Martensite 237.0 GPa 141.0 GPa 116.0 GPa 1.0  1012 m2 1.0 m2 2.58  1010 m
v0 D0 Q⁄SD Q⁄0 p⁄ q⁄
Austenite 1.0  104 m s1 4.0  105 m2 s1 3.4  1019 J 1.5  1019 J 1.0 1.0
Martensite 1.0  104 m s1 20.0  105 m2 s1 3.4  1019 J 1.5  1019 J 1.0 2.55
btwin csf s L0 ⁄ N_ 0 ⁄ r⁄
10 2 15 1
Austenite 1.45  10 m 0.021 J m 0.1 lm 560  b 2.0  10 s 3.0
Martensite – – – – – 2.0

tion of both the strength after yielding and the maximum elonga- 0.45
tion is attributed to the fact that the microstructure of the (a) 35 μm (Experiment in Ref [29])
short-time-annealed steels (150 s for the 270 nm grained steel, 0.40
35 μm
180 s for the 800 nm grained steel) may not be homogeneous. 0.35 800 nm

Twin volume fraction

Some dark grains indicate the accumulation of defects whereas 270 nm
the other grains appear less distorted indicating a low defect con- 0.30
tent (Fig. 2). Therefore, during the tensile tests conducted on the
0.25
UFG materials the deformation may be inhomogeneous, deterio-
rating ductility. 0.20
Fig. 4(b) shows the strain-hardening responses (strain-
0.15
hardening rates dr/de) of the steels. For the steel with a grain size
of 35 lm, below 0.1 true strain the decrease of both, the mea- 0.10
sured and simulated strain-hardening rates indicates that disloca-
0.05
tion slip is the main deformation mechanism. At a strain level
above 0.1, the increasing strain-hardening related to mechanical 0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6
twinning is clearly identified, which is typical of metals and alloys
with low stacking fault energy (SFE) such as silver [47], Cu–Zn True strain
alloys [48,49] and TWIP steels [50–52] where twin interfaces play
a large role in the hardening of the materials. After 0.33 true 1.4
strain the experiment shows that the strain-hardening rate contin- (b) 35 μm
ues to increase with deformation, whereas the simulation show a 1.2 800 nm
dislocation slip
significant drop which corresponds to the saturation of the twin 270 nm
volume fraction. Therefore, the deformation stage (0.33 true
Shear contribution

1.0
strain) corresponding to the deviation of the modeling from the
experiments is attributed to the deformation-induced onset of 0.8
the martensitic phase transformation.
The simulated twin volume fraction for the same steel as a func- 0.6
tion of the true strain is presented in Fig. 5(a), showing that
approximately 16 vol.% and 27 vol.% of the steel is twinned at true 0.4
strains of 0.22 and 0.36, respectively. This agrees well with the pre- twinning
viously published experimental observations on the same CG steel 0.2
[36]. For the steels with a grain size of either 800 nm or 270 nm,
the strain hardening dominated by twinning is not predicted by 0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
the modeling, even though experiments show that the
strain-hardening rates of the steels decrease mildly at a true strain True strain
above 0.03 (Fig. 4(b)). To see this more clearly, the shear contri-
butions of the competing deformation systems (dislocation slip Fig. 5. Predicted (a) twin volume fraction and (b) relative shear contributed by the
dislocation slip and twinning systems for the CG and UFG steels as a function of true
and twinning) obtained from the different simulations are plotted
strain along the tensile axis.
as a function of the true strain in Fig. 5(b). The individual contribu-
tions are given as summations over the absolute values of the shear
rates on the different systems for each type of mechanism, normal-
ized by the overall shear in each time step. The simulation for the 270 nm grain-sized steels, the shear rate on the twin systems that
35 lm grain-sized steel shows that an increase in the twinning remains on a constant low level during the whole period of plastic
shear rate occurs at the initial stage of deformation, and the max- straining is predicted. This explains the very minor strain harden-
imum occurs at a strain of 0.25. With further deformation, the ing as well as the negligible twin volume fraction as revealed by
shear rate carried by dislocations increases, indicating that measurements. Moreover, the experimental results show that only
the material is deformed both by slip and by twinning. The differ- very limited martensite transformation occurs in these steels
ence in strain-hardening response between the experiment and the during tensile testing, which leads to a better prediction of the
prediction at this deformation stage is attributed to the absence stress–strain behavior for the UFG steels compared to the CG
the martensite transformation in the model. For the 800 nm and material.
312 Y.F. Shen et al. / Acta Materialia 97 (2015) 305–315

4. Discussion after tensile deformation. Fig. 6 shows sections close to the fracture
surfaces. For the steels with average grain sizes of 35 lm and
4.1. Origin of the ultrahigh strength—Hall–Petch effect 800 nm, a large density of deformation twins with lamella thick-
nesses of 5–30 nm is identified (Figs. 6(a)–(d)). These twins pene-
The ultrahigh strength in these UFG 304L SS samples originates trate the entire grains, whereas some of them have no clearly
mainly from the strengthening due to the Hall–Petch effect [53]: defined boundaries with one end ending within the grains. More-
over, dense dislocation arrangements accumulate at the twin
ry ¼ r0 þ k d1=2 ð13Þ boundaries as well as at the high-angle grain boundaries. For the
here r0 is the friction stress, k is a constant, referred to as grain steel with an average grain size of 270 nm, we find that multiple
boundary resistance, and d is the mean grain size. According to dislocations are blocked by grain boundaries, while some
Mayers et al. [54], r0 is 167 MPa, and k = 19.05 MN mm1/2 for a micro-twins are also seen in grain interiors (Fig. 6(e)). To study this
304 SS. Hence, the tensile yield strength as a function of d1/2 for more closely, high resolution observations are made on the areas
the austenitic phase of the investigated UFG samples can be extrap- indicated by the rectangles, as shown in Figs. 6(f) and (g), respec-
olated from the Hall–Petch line using the literature data for 304 SS tively. We find that the lamella thicknesses of the twins are below
[54] to a mean grain size of 800 nm, yielding a strength of about 10 nm. One partial dislocation is emitted from a twin boundary
840 MPa. This value is lower than but close to the measured yield and stopped at the other twin boundary, leaving a stacking fault
strength of 1170 MPa for the sample cold-rolled to 67% and behind (Fig. 6(f)). In Fig. 6(g), many stacking faults are identified,
annealed at 650 °C. For the sample with an average grain size of though some of them are recognized as e-martensite plates [55].
270 nm, the estimated yield strength of the austenite is This nanostructure is very different from the one observed in the
1325 MPa, which is also lower than the measured value of CG steels [36], in which most of the plastic deformation occurs in
1890 MPa but still captures the strengthening effect with the refine- the form of dislocation slip, twinning, and martensitic phase trans-
ment of grains. Therefore, the Hall–Petch law provides a reasonable formation. The statistical analysis of the TEM observations on sev-
prediction of the yield strength for the UFG steels. The discrepancy eral areas for each of the deformed steel samples incorporating
between the so predicted strength and the measurement is attribu- more than 200 individual grains reveals that the twinned area frac-
ted to the pre-existence of martensite in those steels prior to tion for the 270 nm, 800 nm and 35 lm grain-sized samples is 3%,
loading. 8%, and 36%, respectively. According to the finite element simula-
tion results, twinning is mostly activated in the steel with an aver-
4.2. Origin of the ultrahigh strength—mechanical twinning effect age grain size of 35 lm but suppressed in steels with grain sizes of
800 nm and 270 nm, due to the reduced distance between the
To explore the fundamental mechanisms leading to the ultra- obstacles acting against twin growth with decreasing grain size.
high strength of the UFG austenitic stainless steel, systematic This picture generally agrees with the above-mentioned TEM
TEM observations have been conducted on the CG and UFG steels observations.

Fig. 6. TEM micrographs showing morphologies of the deformed CG and UFG steels with the average grain size of (a),(b) 35 lm; (c),(d) 800 nm and (e)–(g) 270 nm. (f) and (g)
are close observations of the areas indicated by the rectangles in (e). SF: stacking fault.
 Y.F. Shen et al. / Acta Materialia 97 (2015) 305–315 313

Recently, Misra et al. [56,57] investigated the deformation an average size of 270 nm. However, it is worth noting that in UFG
behavior of a nano-grained/ultrafine-grained austenitic stainless steels the length of partial dislocations required for forming a twin
steel with high strength and high ductility. Their results demon- nucleus becomes quite limited compared with the CG samples.
strated that mechanical twinning was an active deformation mech- Then, the required correlated bow-out of three partial dislocations
anism in the steel, while in a CG steel the nucleation of may not be significant enough to trigger the formation of a twin
strain-induced martensite occurred at the shear bands. In order nucleus, which explains the prevalence of dislocation slip in the
to elucidate the transition from dislocation slip to twinning in a 800 nm and 270 nm grained steels as predicted by the simulation.
NG aluminium, Chen et al. [58] proposed a simple model based On the other hand, during plastic deformation the activation of
on the shear stresses required to initiate a full dislocation and a partials is not required due to the pile-up and dissociation of dislo-
Shockley partial emitted from grain boundaries, in which the rela- cations. Under this condition, Zhu et al. [64] gave an analytical
tionship between shear stresses for a full dislocation activation (sf ) description on nucleation of deformation twins in nanocrystalline
0
and the grain size (d) is expressed as: fcc metals in terms of critical grain size, dc :
0
2aGb 2:089  t a2 G d 3:727 þ t a2 G
sf ¼ ð14Þ 6 pffiffiffic 0  6 ð16Þ
d 54:985ð1  tÞ c ln 2d =a 97:053ð1  tÞ c
c
In the case of the localized stress concentration, partial disloca-
tion glide on successive planes can form a large dislocation pile-up Here, t is Poisson’s ratio, 0.27, a is the lattice parameter
0
at the twin front. Thus, stress multiplication at the first dislocation (a  0.361 for fcc metals) [65]). The corresponding dc for the UFG
by pile-up (i.e. ns, where n is the number of partial dislocations in 304SS is in the range of 210–300 nm. The stress data obtained from
the pile-up and s is the applied shear stress) is an effective way to Eqs. (14) and (16) appears to be low, when we consider the yield
penetrate or remove obstacles in front of the twin tip [59]. Conse- strength of 1890 MPa for the UFG 304SS (i.e. shear stress
quently, the shear stress required to initiate a Shockley partial (sp ) 500 MPa, Taylor factor M = 3.61). Consequently, it is conceivable
is: that for the ultrafine-grained steel with an average grain size of
270 nm plastic deformation takes place prior to the onset of
2aGb c twinning.
n  sp ¼ þ ð15Þ Actually, both the twinning process and the formed deforma-
d bp
tion twins can improve the work-hardening ability of UFG materi-
Here a is a parameter with a value of 0.5 for edge dislocations als. First, a stacking fault, precursor of twinning, renders cross-slip
and 1.5 for screw dislocations. It contains the scaling factor more difficult because the stacking fault has to be constricted
between the length of the dislocation source and the grain size before the dislocation can cross slip. Therefore, gliding dislocations
[60]. G is the shear modulus, d is the grain size, c is the stacking are confined to a thin slip band that forms a large pile-up against
fault energy, b and bp are the Burgers vectors of the full dislocation an obstacle, forming the region that is analogous to a grain bound-
and of the Shockley partial dislocation, respectively. ary [9]. Large pile-ups can also generate a large long range back
Evidently, the number of partial dislocations increases with the stress and result in a high strain-hardening rate. Second, it was
decrease of the grain size, since sp becomes lower than sf when the well established that strengthening associated with twin bound-
grain size is reduced to a certain scale due to the enhanced stress aries is similar to that caused by less coherent grain boundaries
concentration under tension. Then it is possible that the deforma- [66]. Molecular dynamics simulations have also demonstrated that
tion mode will switch from slip to slip plus twinning for the mate- once nanoscale twins are formed they can repel certain types of
rial with UFG structures, depending on the emission of partial gliding dislocations and give rise to pile-ups, leading to the strain
dislocations from grain boundaries. In this study, because of the hardening of materials [67].
low SFE of the austenitic stainless steel, partial dislocations acting
as glide dislocation are more widely separated than in high SFE
4.3. Origin of the ultrahigh strength—martensitic transformation effect
materials. Before a dislocation can leave its primary slip plane
and move onto a cross slip plane, the stacking fault between the
The high work-hardening rate observed for the austenitic steels
partials must firstly be compressed, which requires a high stress
is also associated with the deformation-induced martensitic trans-
to overcome the repulsive forces between the partial dislocations.
formation in these materials [35,64]. As documented in Table 1, for
Therefore, cross slip is suppressed and deformation tends to be
the 800 nm grain-sized sample, after tensile deformation to 12%
much more planar in the austenitic steel than in materials with
strain the volume fraction of martensite shows a significant
higher SFE. These faults are precursors of deformation twins found
increase from 23% to 38%. For the 270 nm grain-sized sample at
in the microstructures. Since the faults and twins are formed pro-
6% strain, the volume increment of martensite is only 2%. How-
gressively with straining, twins in the steels increase gradually via
ever, for the 35 lm grain-sized steel, a large amount of martensite
partial dislocation slip. The activated stacking faults and the subse-
(53 vol.%) induced by 59% tensile deformation is formed and this
quently formed twin boundaries can effectively refine the
leads to the significant hardening of the material. Our experimen-
microstructure to a scale similar to UFG or NG. The importance
tal results hence show that the sample with a grain size of 270 nm
of partial dislocations, stacking faults and twins in the plastic
is surprisingly stable against the transformation from austenite to
deformation of fcc metals has recently been addressed by several
martensite, thus the strain-hardening contributed by phase trans-
authors [22,61,62].
formation is negligible during the tensile testing. Assuming that
The shear stress for dislocation emission in UFG austenitic
the austenite transforms to the martensite via a single variant
steels can be determined with the values in a 304SS:
mode, the increase in the elastic-strain energy can be estimated
b = 2.58  1010 m, bp = 1.45  1010 m, G = 76 GPa, a = 1, and
according to [57,68]:
c = 0.021Jm2 [63,64]. Then, the shear stress for partial dislocation
emission (n = 3, based on the twin nucleation scheme proposed by 2
DEt ¼ 0:5E1 e21 ðL=dÞ þ ð0:5E2 e22 þ 0:5E3 e23 ÞðL=dÞ ð17Þ
Mahajan et al. [44]) is 96 MPa, while the emission of a perfect or
full dislocation (sf) reaches 145 MPa, according to Eqs. (14) and where E and e are the Young’s modulus and elastic strain in each
(15). This means that it is easier to activate a partial dislocation lattice plane, respectively. L is the thickness of martensite lath
than a full dislocation in the ultrafine-grained microstructure with and d is the grain size of austenite. This means that the value of
314 Y.F. Shen et al. / Acta Materialia 97 (2015) 305–315

DEt increases with decreasing grain size of the initial austenite. [9] M. Calcagnotto, Y. Adachi, D. Ponge, D. Raabe, Deformation and fracture
mechanisms in fine- and ultrafine-grained ferrite/martensite dual-phase steels
Consequently, the nucleation ability of the martensite is reduced
and the effect of aging, Acta Mater. 59 (2011) 658–670.
with the refinement of grains, and the lattice displacement must [10] M. Calcagnotto, D. Ponge, D. Raabe, Effect of grain refinement to 1 mu m on
be accommodated by dislocation slip and twinning in the UFG strength and toughness of dual-phase steels, Mater. Sci. Eng., A 527 (2010)
steels. 7832–7840.
[11] M. Calcagnotto, D. Ponge, D. Raabe, Microstructure control during fabrication
of ultrafine grained dual-phase steel: characterization and effect of
intercritical annealing parameters, ISIJ Inter. 52 (2012) 874–883.
5. Conclusions [12] S. Cheng, J.A. Spenser, W.W. Milligan, Strength and tension/compression
asymmetry in nanostructured and ultrafine-grain metals, Acta Mater. 51
(1) An UFG austenitic stainless steel with an average grain size (2003) 4505–4518.
[13] Y.H. Zhao, J.F. Bingert, X.Z. Liao, B.Z. Cui, K. Han, A.V. Serhueeva, A.K.
of 270 nm has been synthesized by accumulative rolling Muhkerjee, R.Z. Valiev, T.G. Langdon, Y.T. Zhu, Simultaneously increasing the
and annealing. The material shows a yield strength of ductility and strength of ultra-fine-grained pure copper, Adv. Mater. 18 (2006)
1890 MPa, an ultimate-tensile stress of 2050 MPa, and an 2949–2953.
[14] Y.H. Zhao, X.Z. Liao, S. Cheng, E. Ma, Y.T. Zhu, Simultaneously increasing the
elongation of 6 ± 1%. The high strength originates mainly ductility and strength of nanostructured alloys, Adv. Mater. 18 (2006) 2280–
from the strengthening effect associated with grain bound- 2283.
aries, according to the Hall–Petch effect. [15] Y.F. Shen, L. Lu, Q.H. Lu, Z.H. Jin, K. Lu, Tensile properties of copper with nano-
scale twins, Scr. Mater. 52 (2005) 989–994.
(2) For the austenitic steels with different grain sizes, when sub- [16] L. Lu, Y.F. Shen, X.H. Chen, L.H. Qian, K. Lu, Ultrahigh strength and high
jected to a uniaxial tensile load, the twinned volume fraction electrical conductivity in copper, Science 304 (2004) 422–426.
amounts to 3% for the 270 nm grain-sized steel after 6% [17] L. Lu, X.H. Chen, X.X. Huang, K. Lu, Revealing the maximum strength in
nanotwinned copper, Science 323 (2009) 607–610.
elongation, 8% for the 800 nm grain-sized steel after 12%
[18] V. Shrinivas, S.K. Varma, L.E. Murr, Deformation-induced martensitic
elongation, and 36% for the 35 lm grain-sized steel after characteristics in 304-stainless and 316-stainless steels during room-
59% elongation, respectively. The volume fraction of the temperature rolling, Metall. Mater. Trans. A 26 (1995) 661–671.
martensite is 34%, 38% and 53% for the 270 nm, 800 nm [19] E. Cakmak, H. Choo, K. An, Y. Ren, A synchrotron X-ray diffraction study on the
phase transformation kinetics and texture evolution of a TRIP steel subjected
and 35 lm grain-sized steels, respectively. By TEM analysis to torsional loading, Acta Mater. 60 (2012) 6703–6713.
we find that both, deformation twinning and martensitic [20] K.X. Tao, D.W. Brown, S.C. Vogel, H. Choo, Texture evolution during strain-
phase transformation have been significantly suppressed induced martensitic phase transformation in 304L stainless steel at a
cryogenic temperature, Metall. Mater. Trans. A 37 (2006) 3469–3475.
with the refinement of the initial austenitic grain size. [21] N. Li, Y.D. Wang, W.J. Liu, Z.N. An, J.P. Liu, R. Su, J. Li, P.K. Liaw, In situ X-ray
(3) A crystal plasticity finite element model incorporating the microdiffraction study of deformation-induced phase transformation in 304
grain size-controlled activation of dislocation slip and defor- austenitic stainless steel, Acta Mater. 64 (2014) 12–23.
[22] S. Rajasekhara, P.J. Ferreira, Martensite?austenite phase transformation
mation twinning captures some of the micromechanical kinetics in an ultrafine-grained metastable austenitic stainless steel, Acta
trends observed for the steels with different grain sizes. In Mater. 59 (2011) 738–748.
combination with the theoretical analysis, we suggest that [23] X. Peng, J. Yan, Y. Zhou, F. Wang, Effect of grain refinement on the resistance of
304 stainless steel to breakaway oxidation in wet air, Acta Mater. 53 (2005)
deformation of instable austenitic UFG steels is dominated 5079–5088.
by the slip of partial dislocations, whereas for CG steels dis- [24] J.-H. Kim, B.K. Kim, D.-I. Kim, P.-P. Choi, D. Raabe, K.-W. Yi, The role of grain
location slip, twinning and martensitic phase transformation boundaries in the initial oxidation behavior of austenitic stainless steel
containing alloyed Cu at 700 °C for advanced thermal power plant
jointly contribute to the plastic straining.
applications, Corros. Sci. 96 (2015) 52–66.
[25] B.R. Kumar, D. Raabe, Tensile deformation characteristics of bulk ultrafine-
grained austenitic stainless steel produced by thermal cycling, Scr. Mater. 66
(2012) 634–637.
Acknowledgements [26] Z. Yanushkevich, A. Belyakov, R. Kaibyshev, Microstructural evolution of a 304-
type austenitic stainless steel during rolling at temperatures of 773–1273 K,
Acta Mater. 82 (2015) 244–254.
The present research is supported by NSAF (Grant No. [27] T. Sakai, A. Belyakov, R. Kaibyshev, H. Miura, J.J. Jonas, Dynamic and post-
U1430132), the National Natural Science Foundation of China dynamic recrystallization under hot, cold and severe plastic deformation
(Grant No. 51231002), the Fundamental Research Funds for the conditions, Prog. Mater Sci. 60 (2014) 130–207.
[28] Y. Estrin, A. Vinogradov, Extreme grain refinement by severe plastic
Central Universities (Nos. N130402005, N130510001), and the Pro-
deformation: a wealth of challenging science, Acta Mater. 61 (2013) 782–
gram for New Century Excellent Talents in University 817.
(NCET-13-0104). [29] N. Dudova, A. Belyakov, T. Sakai, R. Kaibyshev, Dynamic recrystallization
mechanisms operating in a Ni–20%Cr alloy under hot-to-warm working, Acta
Mater. 58 (2010) 3624–3632.
References [30] Y. Huang, J.D. Robson, P.B. Prangnell, The formation of nanograin structures
and accelerated room-temperature theta precipitation in a severely deformed
Al-4 wt.% Cu alloy, Acta Mater. 58 (2010) 1643–1657.
[1] R.Z. Valiev, Nanostructuring of metals by severe plastic deformation for
[31] D.A. Hughes, N. Hansen, High angle boundaries formed by grain subdivision
advanced properties, Nature Mater. 3 (2004) 511–516.
mechanisms, Acta Mater. 45 (1997) 3871–3886.
[2] X.H. Chen, J. Lu, L. Lu, K. Lu, Tensile properties of a nanocrystalline 316L
[32] W.S. Lee, C.F. Lin, Comparative study of the impact response and
austenitic stainless steel, Scr. Mater. 52 (2005) 1039–1044.
microstructure of 304L stainless steel with and without prestrain, Metall.
[3] Y. Iwahashi, J. Wang, Z. Horita, M. Nemoto, T.G. Langdon, Principle of equal-
Mater. Trans. A 33 (2002) 2801–2810.
channel angular pressing for the processing of ultra-fine grained materials, Scr.
[33] W.S. Lee, C.F. Lin, T.H. Chen, M.C. Yang, High temperature microstructural
Mater. 35 (1996) 143–146.
evolution of 304L stainless steel as function of pre-strain and strain rate,
[4] Y. Ivanisenko, R.K. Wunderlich, R.Z. Valiev, H.J. Fecht, Annealing behaviour of
Mater. Sci. Eng., A 527 (2010) 3127–3137.
nanostructured carbon steel produced by severe plastic deformation, Scr.
[34] T. Suzuki, H. Kojima, K. Suzuki, T. Hashimoto, M. Ichihara, An experimental
Mater. 49 (2003) 947–952.
study of the martensite nucleation and growth in 18/8 stainless steel, Acta
[5] Y. Saito, H. Utsunomiya, N. Tsuji, T. Sakai, Novel ultra-high straining process for
Metall. 25 (1977) 1151–1162.
bulk materials-development of the accumulative roll-bonding (ARB) process,
[35] G.B. Olson, M. Cohen, Kinetics of strain-induced martensitic nucleation, Metall.
Acta Mater. 47 (1999) 579–583.
Trans. A 6 (1975) 791–795.
[6] R. Song, D. Ponge, D. Raabe, J.G. Speer, D.K. Matlock, Overview of processing,
[36] Y.F. Shen, X.X. Li, X. Sun, Y.D. Wang, L. Zuo, Twinning and martensite in a 304
microstructure and mechanical properties of ultrafine grained bcc steels,
austenitic stainless steel, Mater. Sci. Eng., A 552 (2012) 514–522.
Mater. Sci. Eng., A 441 (2006) 1–17.
[37] D. Raabe, Texture and microstructure evolution during cold rolling of a strip
[7] R. Song, D. Ponge, D. Raabe, Mechanical properties of an ultrafine grained C–
cast and of a hot rolled austenitic stainless steel, Acta Mater. 45 (1997) 1137–
Mn steel processed by warm deformation and annealing, Acta Mater. 53 (2005)
1151.
4881–4892.
[38] C. Herrera, D. Ponge, D. Raabe, Design of a novel Mn-based 1 GPa duplex
[8] R. Song, D. Ponge, D. Raabe, R. Kaspar, Microstructure and crystallographic
stainless TRIP steel with 60% ductility by a reduction of austenite stability,
texture of an ultrafine grained C–Mn steel and their evolution during warm
Acta Mater. 59 (2011) 4653–4664.
deformation and annealing, Acta Mater. 53 (2005) 845–858.
 Y.F. Shen et al. / Acta Materialia 97 (2015) 305–315 315

[39] W.Y. Xue, Y.F. Shen, D.F. Liu, Z.Y. Liu, Y.D. Wang, Strength and ductility of [55] Y.F. Shen, Y.D. Wang, X.P. Liu, X. Sun, R. Lin Peng, S.Y. Zhang, L. Zuo, P.K. Liaw,
ultrafine grained 304ss prepared by accumulative rolling and annealing, in: Deformation mechanisms of a 20 Mn TWIP steel investigated by in situ
Characterization of Minerals, Metals, and Materials, TMS (The Minerals, Metals neutron diffraction and TEM, Acta Mater. 61 (2013) 6093–6106.
& Materials Society), 2013, pp. 45–52. [56] R.D.K. Misra, Z. Zhang, Z. Jia, L.P. Karjalainen, Probing deformation processes in
[40] S.R. Kalidindi, Modeling anisotropic strain hardening and deformation textures near-defect free volume in high strength high ductility nanograined/ultrafine-
in low stacking fault energy fcc metals, Int. J. Plast. 17 (2001) 837–860. grained (NG/UFG) metastable austenitic stainless steels, Scr. Mater. 63 (2010)
[41] W. Blum, P. Eisenlohr, Dislocation mechanics of creep, Mater. Sci. Eng., A 510– 1057–1060.
511 (2009) 7–13. [57] R.D.K. Misra, V.S.A. Challa, Y.F. Shen, M.C. Somani, L.P. Karjalainen, Interplay
[42] F. Roters, P. Eisenlohr, L. Hantcherli, D.D. Tjahjanto, T.R. Bieler, D. Raabe, between grain structure, deformation mechanisms and austenite stability in
Overview of constitutive laws, kinematics, homogenization and multiscale phase-reversion-induced nanograined/ultrafine-grained austenitic ferrous
methods in crystal plasticity finite-element modeling: Theory, experiments, alloy, Acta Mater. 84 (2015) 339–348.
applications, Acta Mater. 58 (2010) 1152–1211. [58] M.W. Chen, E. Ma, K.J. Hemker, Y.M. Wang, X. Cheng, Deformation twinning in
[43] L. Kubin, B. Devincre, T. Hoc, Modeling dislocation storage rates and mean free nanocrystalline aluminium, Science 300 (2003) 1275–1277.
paths in face-centered cubic crystals, Acta Mater. 56 (2008) 6040–6049. [59] Y.M. Wang, M.W. Chen, F.H. Zhou, E. Ma, High tensile ductility in a
[44] S. Mahajan, G.Y. Chin, Formation of deformation twins in f.c.c. crystals, Acta nanostructured metal, Nature 419 (2002) 912–915.
Metall. 21 (1973) 1353–1363. [60] Y.T. Zhu, X.L. Wu, X.Z. Liao, J. Narayan, L.J. Kecskes, S.N. Mathaudhu,
[45] F. Roters, P. Eisenlohr, C. Kords, D.D. Tjahjanto, M. Diehl, D. Raabe, DAMASK: Dislocation–twin interactions in nanocrystalline fcc metals, Acta Mater. 59
the Düsseldorf advanced material simulation kit for studying crystal plasticity (2011) 812–821.
using an FE based or a spectral numerical solver, Proc. IUTAM 3 (2012) 3–10. [61] W.W. Jian, G.M. Cheng, W.Z. Xu, H. Yuan, M.H. Tsai, Q.D. Wang, C.C. Koch, Y.T.
[46] Z. Zhao, M. Ramesh, D. Raabe, A.M. Cuitino, R. Radovitzky, Investigation of Zhu, S.N. Mathaudhu, Ultrastrong Mg alloy via nano-spaced stacking faults,
three-dimensional aspects of grain-scale plastic surface deformation of an Mater. Res. Lett. 1 (2013) 61–66.
aluminum oligocrystal, Int. J. Plast. 24 (2008) 2278–2297. [62] V. Yamakov, D. Wolf, S.R. Phillpot, H. Gleiter, Dislocation–dislocation and
[47] H. Paul, J.H. Driver, C. Maurice, A. Pia˛tkowski, The role of shear banding on dislocation–twin reactions in nanocrystalline Al by molecular dynamics
deformation texture in low stacking fault energy metals as characterized on simulation, Acta Mater. 51 (2013) 4135–4147.
model Ag crystals, Acta Mater. 55 (2007) 575–588. [63] L.E. Murr, K.P. Staudhammer, S.S. Hecker, Effects of strain state and strain rate
[48] T. Leffers, R.K. Ray, The brass-type texture and its deviation from the copper- on deformation-induced transformation in 304 stainless steel: part II.
type texture, Prog. Mater. Sci. 54 (2009) 351–396. Microstructural study, Metall. Trans. A 13 (1982) 627–635.
[49] N. Jia, F. Roters, P. Eisenlohr, C. Kords, D. Raabe, Non-crystallographic shear [64] Y.T. Zhu, X.Z. Liao, S.G. Srinivasan, E.J. Lavernia, Nucleation of deformation
banding in crystal plasticity FEM simulations: example of texture evolution in twins in nanocrystalline face-centered-cubic metals processed by severe
a-brass, Acta Mater. 60 (2012) 1099–1115. plastic deformation, J. App1. Phys. 98 (2005) 1–7.
[50] D.R. Steinmetz, T. Jäpel, B. Wietbrock, P. Eisenlohr, I. Gutierrez-Urrutia, A. [65] H.W. Zhang, Z.K. Hei, G. Liu, J. Lu, K. Lu, Formation of nanostructured surface
Saeed-Akbari, T. Hickel, F. Roters, D. Raabe, Revealing the strain-hardening layer on AISI 304 stainless steel by means of surface mechanical attrition
behavior of twinning-induced plasticity steels: Theory, simulations, treatment, Acta Mater. 51 (2003) 1871–1881.
experiments, Acta Mater. 61 (2013) 494–510. [66] W.J. Babyak, F.N. Rhines, The relationship between the boundary area and
[51] O. Bouaziz, S. Allain, C. Scott, Effect of grain and twin boundaries on the hardness of recrystallized cartridge brass, Trans. Mettall. Soc. AIME 218 (1960)
hardening mechanisms of twinning-induced plasticity steels, Scr. Mater. 58 21–23.
(2008) 484–487. [67] D. Wolf, V. Yamakov, S.R. Phillpot, A. Mukherjee, H. Gleiter, Deformation of
[52] I. Gutierrez-Urrutia, D. Raabe, Dislocation and twin substructure evolution nanocrystalline materials by molecular-dynamics simulation: relationship to
during strain hardening of an Fe-22 wt.% Mn–0.6 wt.% C TWIP steel observed experiments?, Acta Mater 53 (2005) 1–40.
by electron channeling contrast imaging, Acta Mater. 59 (2011) 6449–6462. [68] S. Takaki, K. Fukunaga, J. Syarif, T. Tsuchiyama, Effect of grain refinement on
[53] N.J. Petch, The cleavage strength of polycrystals, J. Iron. Steel. Inst. 174 (1953) thermal stability of metastable austenitic steel, Mater. Trans. 45 (2004) 2245–
25–28. 2251.
[54] M.A. Mayers, K.K. Chawla, Mechanical Behavior of Materials, Prentice Hall, Inc.,
Upper Saddle River, New Jersey, 1999. 07458, pp. 271–272.