2. S. Bugat et al. / Materials Science and Engineering A317 (2001) 32–36
33
Table 1
Chemical composition of the investigated steel (wt.%)
C
S
P
N
Si
Mn
Ni
Cr
Mo
Cu
Co
Nb+Ta
Al
Fe
0.036
0.008
0.021
0.051
1.06
0.89
9.70
21.25
2.50
0.16
0.05
0.1
0.02
Bal
2.1. Use of EBSD
The EBSD technique [5,6] was used to determine the
local crystallographic properties of both phases. It allowed us to acquire orientation maps for a plane sample, to identify grains and sub-grains, and to study
morphology, orientation, and in the case of a multiphase material, the phase geometry. In this last case,
specific correlation techniques are used to obtain the
grains for one given phase.
The back scatter device was mounted on a DSM982
Gemini SEM. More details can be found in [7]. Data
were acquired every 55 mm in both directions. To
rebuild the grains from raw data, a specific procedure
was designed, so that, for each phase, the set of acquired data points could be dilated to fill the whole
sample. This procedure was applied to two plane and
three notched samples.
An example of data processing is given in Fig. 2 and
Fig. 3. A light micrograph of the ferrite and austenite
laths (Fig. 2) is used to compare the local phase morphology with the EBSD results. Fig. 3a shows the
acquired data points: austenite is represented by black
dots and ferrite by white dots. Fig. 3b shows the
reconstructed austenite grains; Fig. 3c shows the reconstructed ferrite grains of the same area.
the corresponding
micrograph.
austenite
layer
in
the
light
2.3. Results for austenite
In the case of austenite, the step size of the EBSD
mapping (55 mm) is not negligible compared with the
austenite grain size, at least in the plane of the sample
of Fig. 3. Therefore, the grain definition parameters for
the austenitic phase may have a strong influence on the
final orientation and morphology obtained with data
processing. The selected values correspond to those
giving a stable analysis.
The Kurdjumov–Sachs relationships were first
checked using the inverse pole figure ([001] axis) of the
austenite in the standard triangle of ferrite (Fig. 5a and
b). Experimental results are compared with the theoretical pole figure. Each data point corresponds to a zone
where one ferrite and one austenite grain overlap.
These zones will be referred to as ‘bicrystals’ in the
following paras. About 41 bicrystals were determined
for the microstructure shown in Fig. 3. For each bicrystal, the misorientation angle between the nearest theoretical orientation of the g-grain with respect to the
d-grain and the experimental one was determined. Results are shown in Fig. 5c. About 54% of the zones have
a misorientation less than 5° and 78% a misorientation
2.2. Results for ferrite
About 12 ferrite grains were obtained in the case of
the sample shown in Fig. 2, so that the average grain
area was determined to be about 2 mm2. The grains are
highly textured, their [001] direction corresponds to the
radial direction (R) of the pipe and the other axes ([100]
and [010]) are weakly misoriented with respect to the
longitudinal (L) and tangential (T) axes. This is due to
the centrifugation process. However, two grains
(marked by * in Fig. 3c) had a Ž111 direction parallel
to the R direction of the tube. Within all grains, the
misorientation did not exceed a few degrees.
Comparison between light micrograph and EBSD
map of reconstructed ferritic grains shows that the
ferritic grains are surrounded by a thin continuous
layer of austenite. This can be explained because ferrite
boundaries are preferential sites for austenite nucleation
during solid-phase transformation. This allows a correlation between the phase morphology and the ferritic
grains. An example is given in Fig. 4, where a ferritic
grain boundary obtained by EBSD is compared with
Fig. 1. (a) Austenite (light gray) and ferrite (dark gray) laths, (b)
Cleavage crack cluster in a highly damaged zone.
Fig. 2. Optical micrograph of the sample.
3. 34
S. Bugat et al. / Materials Science and Engineering A317 (2001) 32–36
3b and c, it can be seen that the austenitic grain
boundaries do not systematically correspond to the
ferritic grain boundaries. This is consistent with the fact
that ferritic grains are only slightly misoriented: a given
austenitic grain can easily grow in two different ferritic
grains and keep the K–S relationship with both.
3. Characterization of strains and damage nucleation
3.1. In situ mechanical tests
In order to study the behavior and the damage
nucleation process, SEM in situ tensile tests were performed on plane samples. Two kinds of geometries
were used; smooth and notched samples. A gold grid
was vacuum deposited on one side of the specimens by
means of a microelectrolithographic technique [8]. The
grid step size was 38 mm. It allows to compute the local
strain using image analysis. The macroscopic stress and
strain were also monitored.
3.2. Quantification of strain
Fig. 3. (a) Raw data indicating data points corresponding to the
ferrite (white) and the austenite (black), (b) reconstructed austenite
grains, (c) reconstructed ferrite grains. Thick lines represent grain
boundaries (misorientation angle (q) 15°, and thin lines represent
subgrains (5°B q B15°)
Nine grids were deposited on the plane sample shown
in Fig. 3. They did not cover the whole surface. Table
2 compares the stress–strain curve obtained for the in
situ tensile test with results obtained on a standard
tensile specimen (i.e. a much larger specimen). Both sets
of data differ for average strains larger than 2.5%; this
can be attributed to a scale effect as the ferrite grain
size is of the same order of magnitude as the dimensions of the in situ sample.
The 1024× 1024 pixels images were used to analyze
the grid deformation. By comparison with the initial
Fig. 4. Ferritic grain boundary obtained by EBSD (left) versus
corresponding ferrite grain which is surrounded by an austenite layer
(light micrograph) (right).
less than 10°, showing a satisfactory agreement with the
K –S relationship. This low misorientation suggests that
the austenitic lattice is rotated due to plastic straining
during cooling down after solid-phase transformation.
This is consistent with the fact that within each
austenitic grain some sub-grain boundaries can be
found.
The austenite grains have a complex, non-convex
morphology and are highly intricate. Comparing Fig.
Fig. 5. (a) Theoretical inverse pole figure of the austenite ([001] axis)
in the reference axis of the ferrite, (b) experimental inverse pole figure
of the austenite ([001] axis) in the reference axis of the ferrite, (c)
histogram of the misorientation angle between the nearest theoretical
orientation of the g-grain with respect to the d-grain and the experimental measurement.
4. S. Bugat et al. / Materials Science and Engineering A317 (2001) 32–36
Table 2
Comparison for a given strain of the engineering stress obtained
during in situ testing and the stress obtained on a macroscopic tensile
bar
E (%)
0.556
1.581
2.706
3.806
5.850
8.140
10.20
12.20
in situ
(MPa)
348
427
473
497
537
570
593
613
macro
(MPa)
350
436
490
531
575
610
640
660
35
The laths are obviously too small to be accounted for.
A homogenization procedure for the bicrystal, adjusted
on unit cell calculations, was developed, in which the
mechanical behavior of each phase was modeled using
single crystal plasticity [9]. In that case, the size of the
representative volume element is (50 mm)3. This allows
the computation of the local values of stresses and
strains for each phase.
Calculations were performed using 3D meshes and
constitutive equations for FCC and BCC crystals [9]. In
addition, the test geometry is neither plane stress nor
plane strain.
Preliminary results are presented here in the case of a
notched specimen. A localization band, located at the
interface between two bicrystals (Fig. 7a), was experimentally observed (Fig. 7b and c). This band appeared
at the onset of plastic deformation. The FE simulation
correctly represented this effect (Fig. 7d). It is limited to
the early stages of deformation as small deformation
behaviors are used.
3.4. Damage initiation
Sites of damage initiation are shown (circles) in Fig.
7 for a notched sample and in Fig. 8 for a smooth
tensile sample. Fig. 7 shows that cleavage cracks did
Fig. 6. Average local Green –Lagrange strains and standard deviation
versus macroscopic Green –Lagrange strain (tensile direction 1).
grid, the local displacement field is evaluated. This field
is then derived in order to obtain the local strains. The
average value and the standard deviation (S.D.) of
strains are also calculated. As the initial undeformed
grid is not fully regular, an initial S.D. is measured
( 92.5%). The actual S.D. was corrected assuming that
the initial fluctuations and the actual displacement field
are uncorrelated. Results are shown in Fig. 6 as a
function of the macroscopic strain.
The average local strain in the tensile direction (E11)
does not exactly match the macroscopic strain: this is
due to strain heterogeneity, as the grids did not cover
the whole sample. This is confirmed by the increase in
the S.D. with strain. Due to the grid step size (38 mm),
which is larger than the lath size, the S.D. is representative of the strain heterogeneity between the different
bicrystals.
3.3. Modeling of the stress– strain beha6ior
In order to simulate the in situ tests using the Finite
Element (FE) method, the bicrystal had to be modeled.
Fig. 7. Notched specimen; (a) EBSD map of austenite grains (only
one ferrite grain was observed), (b) deformed grid, (c) experimental
strain field, (d) computed strain field. The initial notch width was 1
mm.
5. 36
S. Bugat et al. / Materials Science and Engineering A317 (2001) 32–36
predict the nucleation rates obtained experimentally by
[1,2].
4. Conclusions
Fig. 8. (a) First observed damage zones, (b) damaged area with a low
CSF exhibiting discontinuous slip, (c) undamaged area with a high
CSF exhibiting continuous slip. (CSF, Schmid factor of the slip
system common to g and d phases).
not initiate in the highly deformed area, but close to the
root of the notch. The macro-crack leading to the final
fracture of the sample initiated in this area, where the
elastic stress concentration factor is equal to 1.10. This
seems to indicate that stress plays a more important
role in damage than strain. However, it can be seen that
both sides of the notch are not equivalent in terms of
local deformations and damage initiation. The present
modeling, which accounts for the morphology of the
grains and the anisotropic plasticity of the phases is,
therefore, necessary as an isotropic behavior would not
give this effect.
Two crack clusters were observed on the surface of
the tensile smooth specimen. Ferrite and austenite lattices related by the K–S relationship share one common slip system. The Schmid factor corresponding to
this system (CSF) was computed for different bicrystals. It leads to low values in the damaged zones
(CSFB 0.2), and higher values for the fully undamaged
areas (0.3BCSF B 0.5). This result is corroborated by
the observation of slip traces in damaged and undamaged areas (Fig. 8). Damage is, therefore, initiated
in regions where the strain incompatibility between
both phases is important, thus generating high local
stresses. In the case of the notched sample, the highly
deformed region had a common Schmid factor equal to
0.4, whereas it was equal to 0.05 in the neighboring, less
deformed, zone.
Based on these observations, the origin of heterogeneous damage nucleation can be interpreted. A predictive model for damage nucleation should incorporate a
random distribution of cleavage stresses and the values
of the local stresses in the ferrite (obtained using the
homogenization procedure). The aim of this model is to
.
In this study, the microstructure and damage process
of an austenite/ferrite duplex stainless steel were investigated. Concerning the microstructure, it was shown
that ferrite grains are textured; their average size is
equal to 2 mm, lattice misorientation within a grain
remains limited. Ferrite grains are surrounded by a thin
continuous layer of austenite.
Austenite grains have an irregular morphology. They
contain sub-grain boundaries. In areas corresponding
to constant austenite and ferrite orientations (bicrystals), both lattices are related by the K– S relationship.
A single austenite grain can grow in several ferrite
grains.
In situ tensile tests were used to monitor the local
strain fields and to detect the sites of damage initiation.
Heterogeneity of deformation increases with increasing
average strain. Damage preferentially initiates in areas
where the common slip system of the bicrystal has a
low Schmid factor.
Acknowledgements
This work was supported by Electricite de France.
´
References
[1] P. Joly, A. Pineau, Defect Assessment in Components — Fundamentals and Applications, Mechanical Engineering Publications,
London, 1991, pp. 381 – 414.
[2] P. Joly, Y. Meyzaud, A. Pineau, in: J. Giovanola (Ed.), Advances
in Fracture/Damage Models for the Analysis of Engineering
Problems, ASME, New York, 1992, pp. 151 – 180.
[3] L. Devillers-Guerville, J. Besson, A. Pineau, Nucl. Eng. Design
168 (1997) 211 – 225.
[4] G. Kurdjumov, G. Sachs, Zeitschrift Physik. 64 (1930) 325 –343.
[5] V. Randle, The Institute of Materials, London, 1992.
[6] B.L. Adams, S.I. Wright, K. Kunze, Met. Trans. 24A (1993)
819 – 831.
[7] S. Bugat, Technical report, ENSMP, 1998.
[8] L. Allais, M. Bornert, T. Bretheau, D. Caldemaison, Acta Metall.
Mater. 42 (11) (1994) 3865 – 3880.
[9] S. Bugat, J. Besson, A. Pineau, Computational Mater. Sci. 16
(1999) 158 – 166.