Evidence of internal Bauschinger test in nanocomposite wires during
in situ macroscopic tensile cycling under synchrotron beam
L. Thilly
a兲
and P. O. Renault
Laboratoire de Métallurgie Physique, University of Poitiers, 86962 Futuroscope, France
S. Van Petegem, S. Brandstetter, B. Schmitt, and H. Van Swygenhoven
Paul Scherrer Institute, CH-5232 Villigen-PSI, Switzerland
V. Vidal and F. Lecouturier
Laboratoire National des Champs Magnétiques Pulsés, UPS-INSA-CNRS, 31400 Toulouse, France
共Received 23 March 2007; accepted 21 May 2007; published online 12 June 2007兲
In situ multiple tensile load-unload cycles under synchrotron radiation are performed on
nanocomposite Cu/Nb wires. The phase specific lattice strains and peak widths demonstrate the
dynamics of the load-sharing mechanism where the fine Cu channels and the Nb nanotubes store
elastic energy, leading to a continuous buildup of internal stress. The in situ technique reveals the
details of the macroscopically observed Bauschinger effect. © 2007 American Institute of Physics.
关DOI:
10.1063/1.2748325兴
In 1881, Bauschinger evidenced that the mechanical re-
sponse of a material can depend not only on the absolute
value of the macroscopic applied stress but also on its direc-
tion, with a reverse 共compressive兲 flow stress often lower
than forward 共tensile兲 flow stress.
1
The observed asymmetry
共so-called “Bauschinger effect”兲 of the flow stress is usually
ascribed to long-range effects 共internal stresses due to
dislocation-microstructure interactions兲 and/or short-range
effects 共directionality of mobile dislocations or annihilation
with reverse strain兲.
2,3
Because of the need to invert the di-
rection of applied load during deformation, true Bauschinger
tests cannot always be performed, as in the case for thin films
or wires. For such sample geometries, special experimental
setups are developed, with simple tensile loading-unloading
tests.
2,4
The strength of 共nano兲composites results from a com-
plex interplay between the mechanical properties of the in-
dividual phases and the presence of interfaces. During defor-
mation, internal stresses develop because of elastic mismatch
between the phases resulting in the manifestation of the
Bauschinger effect.
5–7
When multiple tension-compression
cycles are applied to the composite, the internal stresses will
develop accordingly to the load transfer between the soft and
the hard phase and lead to a rounding of the reverse loading
curves and a decrease of the reverse yield stress, the extreme
situation being that the soft phase yields in compression al-
ready during unloading of the applied tensile load. In this
case, it is foreseen that the soft phase will undergo a true
Bauschinger test during the macroscopic tensile load-unload
cycle, despite it has never been directly observed.
Here we report on such a “built-in” Bauschinger test
performed in situ at the materials beamline of the Swiss
Light Source on a Cu/ Nb nanocomposite wire, where the use
of the microstrip detector allows fast measurements of dif-
fraction patterns over a 2
range of 60°.
8
The real-time re-
solved evolution of the lattice strain and peak width of the
individual phases demonstrate the Bauschinger effect in the
soft Cu phase during multiple load-unload tensile tests,
meanwhile uncovering the details of the load-sharing
mechanism.
The nanocomposite Cu/Nb samples were obtained from
a wire processed via severe plastic deformation to obtain a
structure containing 85
3
Nb nanotubes embedded in a multi-
scale Cu matrix, as shown in Ref.
9. Nb nanotubes 共thickness
t
Nb
=88 nm; volume fraction X
Nb
=20.8%兲 are filled with Cu,
Cu-f 共diameter d
Cu-f
=130 nm兲, separated by the finest Cu-0
channels 共width d
Cu-0
=93 nm兲; groups of 85 Nb nanotubes
are separated by Cu-1 channels 共width d
Cu-1
=360 nm兲; and
groups of 85
2
nanotubes are separated by Cu-2 channels
共width d
Cu-2
=3.9
m兲. Finally, the group of 85
3
nanotubes is
embedded in the external Cu jacket, Cu-3 共d
Cu-3
=21.1
m兲.
The wire has a diameter of 0.5 mm. Transmission electron
microscope 共TEM兲 investigations revealed that the cross sec-
tion of the thickness of Nb nanotubes is composed of one
grain. These parallel grains are elongated along the wire axis
with a 具110典 axial texture. The Cu-f, Cu-0, and Cu-1 chan-
nels 共referred as “fine” Cu in the following兲 contain also one
grain in cross section; these parallel grains are elongated
along the wire axis with low dislocation density. The larger
Cu-2 and Cu-3 channels 共referred as “large” Cu兲 are com-
posed of grains from 200 nm to micrometer range, with high
dislocation density. The fine nanochannels 共Cu-f, Cu-0, and
Cu-1兲 correspond to 40% of the Cu matrix. The Cu matrix
presents a double axial texture with 具111典 and 具200典
components.
9
The wires were locally thinned to obtain a reduced
gauge section over 10 mm 共see inset of Fig.
1兲 allowing to
probe directly the nanocomposite interior of the wire that
would be otherwise screened by the external Cu-3 jacket.
Multiple tensile loading-unloading tests up to fracture were
conducted at room temperature with a strain rate of 10
−5
s
−1
and a 24.2 keV x-ray beam scattering on crystallographic
planes parallel to the tensile axis. Si powder was fixed on the
sample surface for angular calibration. Peak profile analysis
was performed using symmetric Pearson VII functions. As
demonstrated in previous in situ experiments,
10
the 共220兲
Cu
reflection, arising from the Cu grains with 具111典 or 具200典
axial textures, is the superposition from two peaks, the first
a兲
Electronic mail: ludovic.thilly@univ-poitiers.fr
APPLIED PHYSICS LETTERS 90, 241907 共2007兲
0003-6951/2007/90共24兲/241907/3/$23.00 © 2007 American Institute of Physics90, 241907-1
coming from the large-Cu channels 共“large Cu peak”兲, the
second from the fine-Cu channels 共“fine Cu peak”兲. The
共220兲
Cu
peak was fitted using two symmetric functions, with
an integrated intensity ratio I
2
/I
1
of 0.8, in good agreement
with the fine to large Cu volume fractions in the probed
region, where most of the Cu-3 jacket was removed.
Figure
1 presents the macroscopic true-stress–true-strain
curve of one of the tested samples: the increasing hysteresis
during tensile load-unload is the signature of large internal
stresses that are built up. Figure
2共a兲 shows the applied stress
versus run number 共i.e., versus time since one run corre-
sponds to a 30 s collection of x rays兲 with indication of the
holding time in the loaded/unloaded states. It must be noted
that during holding time the strain is kept constant. Figure
2共b兲 demonstrates the evolution of the 共220兲 diffraction peak
position of the large and fine Cu peaks versus run numbers
and Fig.
2共c兲 the evolution of the 共110兲 reflection in the Nb
nanotubes. Already in the as-prepared state, the two Cu
phases are in a different stress state: the fine Cu channels
being in larger axial elastic compression than the large Cu
channels since 2
220
共fine Cu兲⬍ 2
220
共large Cu兲
⬍2
220
共annealed Cu兲= 23.12°. The Nb nanotubes are in
axial elastic tension 关2
110
共Nb兲 ⬎ 2
110
共annealed Nb兲
=12.60° 兴. Note that both equilibrium 2
values were deter-
mined by neutron scattering on annealed Cu and Nb samples.
Applying tensile load, the two Cu phases respond differently:
the shift of the peak position of the large Cu stabilizes upon
loading 关Fig.
2共b兲兴, evidencing a pronounced plasticity re-
gime whereas the fine-Cu peak exhibits only a slight devia-
tion from linear behavior at highest stress. The Nb nanotubes
remain in the elastic regime, as demonstrated by the linear
behavior of the peak position versus time 关Fig.
2共c兲兴 and
versus stress 共not shown here兲. The material is therefore
composed of three phases with distinct elastic-plastic behav-
ior, a situation that favors the development of internal
stresses during codeformation, as evidenced by the gradual
change of the peak positions at each unloaded state: both the
axial compression of Cu channels and the axial tension of Nb
nanotubes increase after each cycle.
Figure
3 focuses on cycles 7 and 8. The evolution of
peak position and full width at half maximum 共FWHM兲 ver-
sus run numbers for the large and fine 共220兲 Cu peaks is
given, respectively, in Figs.
3共b兲 and 3共c兲 and the 共110兲 re-
flection in the Nb nanotubes is provided in Figs. 3共d兲 and
3共e兲. Having in mind that the broadening of diffraction peaks
is induced by the presence of inhomogeneous strains such as
the ones introduced by dislocations, the evolution of FWHM
can be related to the peak position evolution. In Fig. 3共c兲,
FIG. 1. 共Color online兲 Macroscopic true-stress–true-strain curve of one
tested wire. Inset: Schematic of one sample half with reduced section and
beam scattering at crystallographic planes parallel to the tensile axis.
FIG. 2. 共Color online兲 Evolution vs run numbers 共i.e., time兲 and loading-
unloading cycles of 共a兲 true stress, with various holding times at loaded/
unloaded states; 共b兲 position of 共220兲 large-Cu and fine-Cu peaks; and 共c兲
position of 共110兲 Nb peak.
FIG. 3. 共Color online兲 Evolution vs run numbers during cycles 7 and 8 of
共a兲 true stress; 共b兲 and 共c兲 position and FWHM of 共220兲 large-Cu and fine-
Cu peaks; and 共d兲 and 共e兲 position and FWHM of 共110兲 Nb peak.
241907-2 Thilly et al. Appl. Phys. Lett. 90, 241907 共2007兲
tensile load is applied from A 共respectively
␣
兲 to C 共respec-
tively
兲, where the starain is kept constant during 30 s, fol-
lowing an unloading with the same strain rate arriving in the
macroscopic unloaded state E 共respectively 兲. When apply-
ing tensile load at A, the FWHM of the large Cu peak de-
creases first, reaching a local minimum 共B兲, followed by an
increase until the maximum load is attained in C. During
macroscopic unloading, the FWHM first decreases from C to
D followed by an increasing phase from D to E. Note that the
large Cu phase is unloaded at A
⬘
C
⬘
. The evolution of
FWHM can be interpreted as the following: during macro-
scopic loading from A to A
⬘
, the large Cu channels, initially
in axial compression, are unloaded and FWHM decreases.
Then from A
⬘
to C, the large Cu channels are put into ten-
sion but owing to a low yield stress they soon enter in a
plastic regime and FWHM increases at B. When the speci-
men is macroscopically unloaded from C to C
⬘
, the large Cu
channels are unloaded and FWHM decreases. Since the mac-
roscopic stress is higher in the whole sample compared to the
large Cu channels, when the specimen is further unloaded
from C
⬘
to E, these channels are put into compression and
enter in a plastic regime because of low yield stress: FWHM
increases again at D.
The FWHM for the 共220兲 fine Cu peak behaves differ-
ently: upon loading from macroscopic unloaded state
␣
,it
first increases linearly until the unloaded state is reached,
indicated by
␣
⬘
where it suddenly levels off. As long as the
peak position of the fine Cu phase increases linearly, the
FWHM stays approximately constant. When the peak posi-
tion starts deviating from linear behavior 共indicated by

兲 the
FWHM starts increasing again until the maximum macro-
scopic tensile load is reached at
. During unloading,
FWHM decreases very slowly until the unloaded state is
reached 共
⬘
兲, then rapidly until macroscopic unloaded state
共兲. The above mentioned trends are confirmed in cycle 8,
with the difference that here the holding time was 30 min.
Previous in situ TEM deformation studies have shown that
the fine Cu deforms by nucleation of noncorrelated single
dislocation loops nucleated at the Cu/ Nb interfaces and ex-
panding until interaction with a neighboring interface.
11
Such
a mechanism can explain the particular behavior of the
FWHM of the fine Cu channels 关Fig.
3共c兲兴. At the highest
compressive state 共
␣
兲, most of the dislocations 共nucleated in
previous cycles兲 are pushed back into the interfaces. During
macroscopic tensile loading, the fine Cu first unloads until
␣
⬘
and the dislocations return to an equilibrium configuration,
resulting in an increase of the FWHM. Then when the fine
Cu phase is elastically loaded, the FWHM stays approxi-
mately constant 共
␣
⬘
to

兲 and only when the plastic regime is
reached previous and fresh dislocations expand further be-
tween the Cu–Nb interfaces 共from

to
兲. That the FWHM
does not decrease substantially upon tensile unloading 共
to
⬘
兲 suggests the necessity of a force to unpin the dislocations
from Cu–Nb interfaces. Note that a similar “delay” is ob-
served at the beginning of cycle 8 and could not be distin-
guished in cycle 7.
Finally, for the Nb nanotubes which remain in the elastic
regime in axial tension, a slight increase of FWHM after
each loading-unloading cycle is observed 关Fig.
3共e兲兴 suggest-
ing a gradual increase of root mean square strain induced by
the load transfer from the plastifying Cu channels onto the
elastic Nb nanotubes.
10
Figure 4 shows the true stress 关Fig. 4共a兲兴, the lattice
strains 关Fig. 4共b兲兴, and FWHM derived of the 共220兲 large and
fine Cu diffraction now as function of the macroscopic ten-
sile strain during cycle 7. The plots show that 共1兲 during one
tensile loading-unloading cycle, the Cu channels are sub-
jected to a true Bauschinger test, i.e., a tension-compression
cycle, and 共2兲 the “Bauschinger” hysteresis visible in the
macroscopic stress-strain curve results from the reverse
yielding of the large-Cu phase evidenced in the strong round-
ing upon unloading 关Fig.
4共b兲兴 and the increase in FWHM at
D 关Fig.
3共c兲兴. The unloading of the fine Cu channels is, how-
ever, purely elastic, which can be explained by the higher
value of the yield stress related to the size of the fine Cu
channels and the particular dislocation mechanism associ-
ated.
In summary, the above results demonstrate the power of
in situ techniques in revealing the dynamics of the phase
specific lattice strains and deformation mechanisms reflected
in peak broadening, shedding light on the origin of the mac-
roscopically observed Bauschinger effect. The technique al-
lows exploring the physical origins of the Bauschinger effect
by performing simple tensile load-unload experiments mean-
while making advantage of the naturally built-in Bauschinger
test in a nanocomposite.
1
J. Bauschinger, Civiling N.F. 27, 289 共1881兲.
2
Y. Xiang and J. J. Vlassak, Acta Mater. 54, 5449 共2006兲.
3
Y. Brechet and P. Jarry, J. Phys. III 1, 1985 共1991兲.
4
C. Sinclair, G. Saada, and J. Embury, Philos. Mag. 86, 4081 共2006兲.
5
M. Ashby, Philos. Mag. 21, 399 共1970兲.
6
R. Asaro, Acta Metall. 23, 271 共1975兲.
7
J. Eshelby, Proc. R. Soc. London, Ser. A 241, 376 共1957兲.
8
H. Van Swygenhoven, B. Schmitt, P. M. Derlet, S. Van Petegem, A.
Cervellino, Z. Budrovic, S. Brandstetter, A. Bollhalder, and M. Schild,
Rev. Sci. Instrum. 77, 013902 共2006兲.
9
V. Vidal, L. Thilly, F. Lecouturier, and P. O. Renault, Scr. Mater. 57,245
共2007兲.
10
L. Thilly, V. Vidal, S. Van Petegem, U. Stuhr, F. Lecouturier, P. O.
Renault, and H. Van Swygenhoven, Appl. Phys. Lett. 88, 191906 共2006兲.
11
L. Thilly, O. Ludwig, M. Véron, F. Lecouturier, J. P. Peyrade, and S.
Askénazy, Philos. Mag. A 82, 925 共2002兲.
FIG. 4. 共Color online兲 Evolution vs true strain during cycle 7 of 共a兲 true
stress; 共b兲 position of 共220兲 large-Cu and fine-Cu peaks. The Cu channels
are subjected to an internal Bauschinger test during each cycle 共negative
longitudinal lattice strain during tension and positive during compression兲.
241907-3 Thilly et al. Appl. Phys. Lett. 90, 241907 共2007兲