Sie sind auf Seite 1von 38

Accepted Manuscript

Title: Applying infrared thermography to analyse martensitic


microstructures in a Cu-Al-Be shape-memory alloy subjected
to a cyclic loading
Authors: D. Delpueyo, X. Balandraud, M. Gr ediac
PII: S0921-5093(11)00829-X
DOI: doi:10.1016/j.msea.2011.07.050
Reference: MSA 27597
To appear in: Materials Science and Engineering A
Received date: 21-4-2011
Revised date: 20-7-2011
Accepted date: 21-7-2011
Please cite this article as: D. Delpueyo, X. Balandraud, M. Gr ediac, Applying infrared
thermography to analyse martensitic microstructures in a Cu-Al-Be shape-memory
alloy subjected to a cyclic loading, Materials Science & Engineering A (2010),
doi:10.1016/j.msea.2011.07.050
This is a PDF le of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and reviewof the resulting proof
before it is published in its nal form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Page 1 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
> Phase transformation in a SMA single crystal is studied by infrared
thermography.

> Under small cyclic loading, heat source fields are deduced from temperature
fields.

> Fields of latent heat source enabled us to localize the phase transformation zones.

> The thermoelastic coupling effect is also revealed in austenite and martensite.

> An interpretation in terms of martensitic microstructure is deduced.

*Highlights
Page 2 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
1
Applying infrared thermography to analyse martensitic microstructures
in a Cu-Al-Be shape-memory alloy subjected to a cyclic loading

D. Delpueyo, X. Balandraud
*
and M. Grdiac

Laboratoire de Mcanique et Ingnieries (LaMI), Clermont Universit, Universit Blaise Pascal & IFMA,
BP 10448, F-63000 Clermont-Ferrand, France

Abstract. This study aims at revealing martensitic microstructures that exist in a shape-memory alloy.
Infrared thermography was used for this purpose. Experiments were performed on a Cu-Al-Be single
crystal which features a pseudoelastic response at ambient temperature. The specimen was first partially
transformed to martensite by mechanical loading. Then a small cyclic loading was applied while the
temperature evolution on the specimen surface was captured by an infrared camera. Thermal images
obtained were then processed to extract two types of quantities: the maps of heat sources produced by the
material as well as the maps of temperature oscillation amplitudes. Two thermomechanical couplings are
revealed: the thermoelastic coupling and the latent heat due to the small cyclic movement of austenite-
martensite interfaces, thus highlighting the martensitic microstructure distribution in the specimen.

Keywords: shape memory alloys, phase transformation, martensitic microstructures, thermomechanical
couplings, infrared thermography.


*
Corresponding author. Fax: +33 473 28 81 00.
E-mail address: xavier.balandraud@ifma.fr (X. Balandraud).
*Manuscript
Page 3 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
2
1. Introduction

This study aims at revealing martensitic microstructures that exist in a shape-memory alloy (SMA) by
analysing temperature variations measured by infrared (IR) thermography. Experiments were performed
on a copper-aluminium-beryllium (Cu-Al-Be) single crystal featuring a pseudoelastic behaviour at
ambient temperature.

SMAs exhibit peculiar macroscopic properties due to solid-solid phase transformations triggered by both
the stress and the temperature. The parent phase is austenite. It has cubic symmetry in all known SMAs
whereas the symmetry of martensite, which is the product phase, mainly depends on the alloy
composition. The phase change occurring in SMAs is of first order. It is characterized by the co-existence
of austenite and martensite during the transformation of a specimen from one phase to another. The
spatial organization of austenite and martensite within a specimen is often referred to as martensitic
microstructure [1]. The present study aims at proposing a new experimental procedure for studying these
microstructures, which is based on temperature measurements. Tests were performed on a single crystal
because martensitic microstructures grow at a macroscopic scale in this case, contrary to polycrystalline
specimens where microstructures would be confined within microscopic grains. Consequently, capturing
and interpreting thermal maps with considerations directly related to the microstructures is much easier in
the first case.

For a stress-induced phase change at constant room temperature, thermomechanical couplings play an
important role in the mechanical response of the material. Indeed a transformation from austenite to
martensite produces latent heat. This heat causes a temperature increase, which leads to a change in the
transformation kinetics since the phase transformation is partially triggered by temperature. This strong
coupling between temperature change and phase change for a stress-induced transformation has been
studied in depth in the literature (see for instance Refs. [2-8]). Another thermomechanical effect, which
exists in all materials, is related to the thermoelastic coupling (also named isentropic coupling), which is
the coupling between the temperature and the elastic part of the strain through the thermal expansion
appearing in the free energy expression. In case of cyclic loading, an important property is inherited from
Page 4 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
3
this coupling: the greater the amplitude of the stress oscillations, the greater the amplitude of the
temperature oscillations. This phenomenon is used in the so-called thermoelastic stress analysis [9],
which enables one to measure the stress level distribution in structural components. Although this well-
known non-destructive technique is widely applied to many materials, very few studies present
applications to SMAs [10], to the best knowledge of the authors.

In this study, all tests were performed at constant ambient temperature on a Cu-Al-Be SMA single crystal.
The specimen was austenitic at the unstressed state. The mechanical loading was applied by using a
uniaxial testing machine. First, the specimen was stretched in order to partially transform it to martensite.
Second, a small (but rapid) cyclic displacement of the actuator of the testing machine was superimposed
to the mean displacement reached in the first part of the testing procedure. The analysis of the results
obtained consists in processing the temperature fields measured by IR thermography during the cyclic
loading in order to reveal the microstructures in the specimen. Although the objective here is not to study
damping properties, it can be noted that the mechanical loading is quite similar to the loading usually
employed for a dynamic mechanical analysis. SMAs are indeed well-known for their peculiar damping
properties, which are usually explained by an internal friction due to the movements of the microstructure
interfaces (see for instance Ref. [11] and the references which are included). In the present study, the
mechanical loading is chosen to reveal some martensitic microstructures existing in a SMA specimen
through a specific processing of the temperature maps which are captured. It must be noted that the mere
observation of the temperatures does not provide clear information about the martensitic microstructure
evolution because of the thermal diffusion phenomenon, as noted in Ref. [12]. Recently, IR thermal
imaging of a SMA single crystal revealed an inhomogeneous temperature distribution, whose evolution
gave relevant information about the kinetics of the interface motion [13]. In the present paper, a specific
processing of the thermal signal is performed to reveal the microstructures.

Understanding the organization of the martensitic microstructures is of wide interest in mechanics of
SMAs because these microstructures represent a key element influencing the macroscopic behaviour of
these materials [1]. In particular, the fatigue response is mainly related to the hysteresis of the phase
transition [14], which is itself related to the martensitic microstructures that the material is able to create
Page 5 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
4
[15-16]. Recent studies have highlighted the role of the local elastic strain and stress levels in the
microstructures due to incompatibility of transformation strains [16-20]. These results provide interesting
information about the relationship between memory effect and martensitic microstructures. The
experimental procedure which is proposed in the present study to analyse martensitic microstructures in
SMAs is in fact complementary to other techniques or experimental procedures (see Ref. [21] for an
overview). Micrographs provide important information about the morphologies of the microstructures:
see for instance Ref. [22] for the detailed analysis of some peculiar microstructures in a single crystal.
Recently, full-field measurement techniques were used to measure displacement or strain distributions in
SMA single crystals. The advantage of these techniques is to provide maps where the strain contrasts
reveal the microstructures. Digital image correlation and the grid method were recently used for this
purpose (see Refs. [23-24] and [25], respectively). Another type of full-field measurement technique is
employed in present study: infrared thermography. The post-processing performed in the present study
has already been used for studying polycrystalline SMAs (see Ref. [8]). It consists in fact in collecting
temperature maps on the surface of the specimen and extracting heat sources from these maps. The first
originality of the present study is to combine heat source calculation and specific dynamic mechanical
loading to reveal small movements of phase transformation fronts between austenite and martensite. The
second originality is to perform the analysis on a single-crystal specimen, thus allowing the
microstructures to grow at a macroscopic scale, which is accessible to infrared cameras.

This paper is organized as follows. Sections 2 and 3 describe the specimen under study and the
experimental setup. Section 4 presents the post-processing of the temperature maps provided by the IR
camera. Experimental results are discussed in Section 5. The following notation is used for the symbols
throughout the manuscript. Italic letters are scalars. Bold letters are vectors in the 3D space. Underlined
letters are second-order tensors such as strain or stress tensors. They are described by 33 matrices.
Letters featuring a superimposed tilde symbol are fourth-order tensors such as stiffnesses.

Page 6 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
5
2. Specimen

2.1. Composition and dimensions
The specimen under study was a single crystal of Cu Al
11,4
Be
0,5
(wt.%) SMA supplied by Nimesis
Technology (Metz, France). It consisted of a 0.94 17.78 72 mm
3
rectangular sheet along the x-, y- and
z-directions, respectively. Aluminium tabs were bonded at the two ends of the specimen in order to
protect it once placed in the jaws of the uniaxial testing machine. The gauge length along the z-direction
between the tabs was equal to 53 mm.

2.2. Crystallographic properties and phase transformation
Austenite phase - At the unstressed state, the specimen was austenitic at ambient temperature
0
T = 22C
because the Martensite Start temperature (
s
M ) was equal to -2C. The orientation of the cubic crystal was
measured by X-ray diffraction. Measurement revealed a cubic crystal structure with a lattice parameter
0
a
= 0.5824 0.0009 nm, which is in fact the size of the DO
3
cell (see Strukturbericht classification [26]).
The rotation matrix R defined from the cubic axes ([100], [010], [001]) to the specimen axes (x, y, z) was
as follows:
|
|
|
.
|

\
|
=
0.103 - 0.769 0.631 -
0.995 0.079 0.066 -
0.000 0.664 - 0.773 -
R
(1)
Martensite phase - The alloy under study is characterized by a M18R monoclinic martensite (see Refs.
[27-28] for alloys with similar compositions). It is generally merely named ' | . The martensite can exist
through different variants. These martensite variants correspond to the same crystal structure, but with
different orientations in space with respect to the austenite crystal. Twelve distinct martensite variants can
appear from an austenite crystal for the present cubic-to-monoclinic phase transformation (cube-edge
variant type [29]).
Habit planes For the phase transformation which is here considered, any martensite variant is
crystallographically compatible with austenite along two plane interfaces named habit planes [29]. From
the twelve martensite variants, twenty-four distinct configurations of habit plane can potentially be
obtained. Two main criteria exist in the literature for the variant selection when stress-induced
Page 7 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
6
transformation occurs in SMAs [30]: namely the maximum resolved shear stress criterion and the
maximum transformation work criterion. The following quantities enabled us to determine which of the
martensite variants is activated by the uniaxial loading along the z-direction:
(i) the orientation of the austenite crystal (see Eq. (1));
(ii) the transformation parameters characterizing the stretches between the austenite and martensite
crystals [25].
Calculations are not detailed here, but the conclusion is that both criteria lead to the same variant which is
expected to appear preferentially in the specimen (see Ref. [25] for the calculations). The angle of the
trace of the corresponding habit plane in the specimen plane (y, z) is predicted to be equal to +57 deg with
respect to the load axis z.

2.3. Mechanical properties at ambient temperature
Some mechanical properties were experimentally obtained from tensile tests performed along the
longitudinal z-direction at ambient temperature
0
T = 22C. Fig. 1 presents a strain-stress curve obtained
under isothermal condition. The tests can be considered as performed under isothermal conditions thanks
to the very low strain rate that was applied: 210
-5
s
-1
. In practice, the whole loading-unloading cycle
lasted nearly more than two hours. This led the latent heat produced by the phase transformation to be
dissipated in both the ambient air and the grips, thus resulting to very negligible temperature variations in
the specimen itself. The Young modulus along the longitudinal z-direction is deduced from the curve
shown in Fig. 1. It is equal to 14 GPa for the austenite and to 15 GPa for the martensite. The total phase
transformation strain
tr
c is equal to about 7%. No residual macroscopic strain was observed at the end of
the test but a classic mechanical hysteresis loop clearly appeared, as expected. This hysteresis loop in the
strain-stress plane is expected to be minimal under isothermal conditions. Indeed it is purely due to the
mechanical dissipation (also named intrinsic dissipation) in this case [4]. The hysteresis width
hyst
o
measured between the two stress plateaus is equal to about 7 MPa; it is greater at both the beginning and
the end of the phase transformation. However, it was observed that slightly different curves were obtained
when several tests were performed under similar conditions.

Fig. 1. Strain-stress curve at ambient temperature
0
T = 22C and under isothermal condition.
Page 8 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
7

2.4. Other physical properties
Table 1 gives some physical properties used in the present work for the post-processing of the thermal
data:
- the density was obtained from the volume and mass of the specimen measured at ambient
temperature.
- the specific heat capacity C of the austenite phase was deduced from the experimental and
theoretical data of Cu-Al-Be alloys given in Ref. [31]. Since the objective of the present study is
to observe martensitic microstructures, it would have been relevant to distinguish the values of C
for austenite and for martensite, but the specific heat capacity of the martensite is not available in
the literature, to the best knowledge of the authors.
- the coefficients of thermal expansion of Cu-Al-Be SMAs are not explicitly given in the literature.
The value which is reported in Table 1 (1710
-6
K
-1
) can be considered as typical of Cu-based
alloys [32]. For the austenite phase, which has a cubic symmetry, the second-order thermal
expansion tensor can be written as follows:
1 6
K 10
17 0 0
0 17 0
0 0 17

|
|
|
.
|

\
|
~
aust
o
(2)
The anisotropic properties of the thermal expansion for the martensite crystal are not available in
the literature, to the best knowledge of the authors.
- the thermal diffusivity was measured during the present study. For the sake of simplicity, only the
thermal diffusivity D of the austenite crystal along the longitudinal z-direction was determined.
This measurement was deduced by processing the temperature map evolution measured during a
mere return to room temperature. Indeed, the temperature evolution in this case is governed by
the heat diffusion equation with no heat source ( 0 = s , see Eqs. (3) and (5) below); so D could be
easily deduced by identification.

Table 1. Physical properties of the austenite phase.

Page 9 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
8
3. Experimental setup

3.1. Mechanical loading
All tests were performed at ambient temperature
0
T = 22 2 C with a uniaxial MTS +/-15kN testing
machine. The same procedure was applied for all these tests (see Fig. 2):
- Step 1: a displacement was imposed to the moving grip of the testing machine in order to partially
transform the specimen to martensite. The macroscopic strain rate was the same for all the tests:
210
-3
s
-1
. The final macroscopic strain denoted
macro
c changed from one test to another.
- Step 2: a waiting time at a fixed displacement was imposed to the specimen. This procedure
enabled the specimen to return to room temperature since the latent heat production during Step 1
led the temperature of the specimen to increase. For all the tests performed in this study, three
minutes were used in practice to ensure that thermal equilibrium was reached.
- Step 3: a cyclic displacement was imposed to the moving grip, around the position reached during
the preceding step. The loading frequency and the strain amplitude are denoted
L
f and
macro
c ,
respectively. The duration of this step was short for all the tests in order to limit fatigue damage:
10 seconds in practice.

Fig. 2. Schematic view of the loading procedure.

The strain amplitude
macro
c must be chosen as small as possible to keep only one configuration of
martensitic microstructures in the specimen during the cyclic loading, without significant change during
the cycles themselves. However,
macro
c must not be too low to be sure that the effect of the thermoelastic
coupling is observed. Indeed, the greater the amplitude of the elastic strain oscillations, the greater the
temperature variations due to the thermoelastic coupling (see Section 4.2 below). The loading frequency
L
f must be sufficiently high to tend toward adiabatic conditions. Besides, the choice of
macro
c and
L
f
also results from the physical limitations of the testing machine. The numerical values of these two
parameters will be given below for each test.

Page 10 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
9
3.2. Temperature measurement
A Cedip Jade III-MWIR infrared camera was used to capture the temperature fields on the specimen
surface during Step 3 of the loading procedure. This camera features IR detectors characterized by a
wavelength range of 3.5-5 m. The integration time used for the measurements was 1500 s. The thermal
resolution at ambient temperature was about 0.02C. The acquisition frequency
a
f was equal to 436 Hz.
This acquisition frequency must not be a multiple of the load frequency
L
f for a better data processing by
discrete Fourier transform (see Section 4.2 below). The maps provided by the camera at this acquisition
frequency are digitized with a 80160 array of pixels. The spatial resolution in temperature,
corresponding here to the pixel size on the specimen, is equal to 0.36 mm.
A thin, opaque and uniform black paint was sprayed on the specimen surface to obtain a thermal
emissivity close to one. This paint spray was applied only a few minutes before testing the specimen to
avoid small cracks to appear in the paint layer during the test.

4. Data post-processing

4.1. Thermomechanical framework
The heat diffusion equation which governs the temperature evolution in the specimen can be written as
follows [4-5]:
( )
C
r s
T D
t
T
ext

+
=
c
c
grad div
(3)
where T is the temperature, the density, C the specific heat capacity, D the second-order thermal
diffusivity tensor,
ext
r the external heat source by radiation and s the heat source produced by the
material itself. In the case of SMAs, the heat source s can be split into different terms:
- the thermoelastic coupling source denoted
the
s . It corresponds to the coupling between
temperature T and the elastic part of the strain
el
c caused by the thermal expansion in the free
energy expression [9]. This term can be written as follows:
t
A T s
el
the
c
c
=
c
o :
~
(4)
Page 11 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
10
where A
~
is the fourth-order stiffness tensor and o the second-order thermal expansion tensor.
N M : is the trace of the matrix product N M . In the case of an isotropic material, Eq. (4)
becomes o o T s
the
= , where o is the isotropic coefficient of thermal expansion and o the rate
of the stress tensor trace (sum of the normal stresses) [9];
- the latent heat source. It is positive for a transformation from austenite to martensite, and negative
for the inverse transformation;
- the mechanical dissipation
1
d , also referred to as the intrinsic dissipation. It is always positive. It
is related to some irreversible phenomena such as fatigue damage or plasticity. Note that the
hysteresis loop observed during a load-unload test is partially due to
1
d (the other part is due to
the thermomechanical couplings for non-isothermal evolutions [4]).

4.2. Application to the cyclic loading
The thermal signal recorded during Step 3 is expected to give some useful information concerning the
martensitic microstructures because of the heat produced by the cyclic movement of the austenite-
martensite interfaces. The following considerations have been taken into account to analyse the
temperature fields:
1. About the adiabiaticity of the local thermomechanical response, two cases must be distinguished:
- case 1: no phase transformation occurs. For a purely thermoelastic response, the adiabaticity is
considered as reached if the so-called thermal diffusion length ( )
L
f D t / = is much lower
than the spatial variations of stress (which can be defined by ( ) o o A / 2 where A is the
Laplacian operator) [33]. For example, is equal to 1 mm for the thermal diffusivity D
given in Table 1 and for a load frequency equal to
L
f = 13 Hz. In the present case of uniaxial
tests performed on a specimen exhibiting a constant cross-section, this length can be assumed
to be much lower than the spatial variations of the stress. As a consequence, the diffusion term
in Eq. (3) ( ( ) T Dgrad div ) can be considered as negligible for the zones of the specimen
which are not subjected to phase transformation during a cyclic loading at 13 Hz;
Page 12 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
11
- case 2: phase transformations occur. It must be noted that, contrary to the thermoelastic
coupling [33], no similar theoretical result exists about the adiabaticity in the case of other
thermomechanical couplings, such as those related to phase transformations. As a
consequence, the diffusion term in Eq. (3) was not neglected in the thermal signal post-
processing performed here (see Section 4.3);
2. the thermoelastic coupling source
the
s leads to a temperature which oscillates at the same
frequency as the elastic strain (see Eq. (4)), thus at the same frequency as the load, but in opposite
phase;
3. if a cyclic austenitemartensite transformation occurs somewhere in the specimen because of the
cyclic loading, the corresponding production/absorption of latent heat logically leads to a
temperature oscillation at the same frequency as the load. It can be noted that a transformation
from a martensite variant to another martensite variant (so-called martensite reorientation
phenomenon) does not produce any latent heat;
4. the maximum stress level reached during the tests was chosen in such a way that no plasticity
occurred, at least at the macroscopic scale. Moreover, the fatigue damage was not expected to
produce a significant temperature increase because of the short period of the cyclic loading. From
these two points, it can be assumed that the mechanical dissipation
1
d does not create significant
temperature variations;
5. temperature T in Eq. (4) is in Kelvin. Since the temperature variations did not exceed some
degrees during the cyclic loading, temperature T in Eq. (4) can be assumed to be equal to the
ambient temperature denoted
0
T ;
6. the external source by radiation
ext
r can be assumed to be low and constant vs. time. This
hypothesis is reasonable for two reasons. First, temperature variations within the specimen
remained small. Second, careful protection of the environment was placed around the specimen.
So it can be assumed that the external source by radiation did not significantly modify the
amplitude of the temperature oscillations during the tests.

Page 13 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
12
4.3. Heat source calculation
If the field of external heat source by radiation
ext
r is assumed to be constant vs. time (see item 6 in
Section 4.2 above), Eq. (3) can be rewritten as follows:
( )
C
s
D
t
u
u
=
c
c
grad div
(5)
where u is the temperature variation with respect to a reference temperature
ref
T measured in practice at
the end of Step 2 (see Fig. 2), when the specimen returned to thermal equilibrium. Thus
( ) ( ) ( ) z y T z y T z y
ref
, , , = u . It can be noted that ( )
0
, T z y T
ref
~ . The small difference between ( ) z y T
ref
, and
the ambient temperature
0
T is due to the fact that one of the two grips was slightly warmer than the other
because of the working oil of the machine.
In practice, the heat sources can be calculated by discretizing Eq. (5), using the following finite difference
scheme:
( ) ( )
|
|
.
|

\
| +
+
+

=
+ + +
2
, 1 , , , , 1 ,
2
, , 1 , , , , 1 1 , , 1 , , , ,
2
2
2
2
2 p p
D
t C
s
k j i k j i k j i k j i k j i k j i k j i k j i k j i
u u u u u u u u

(6)
where i and j are the line and column index in the matrix of IR detectors, respectively, p the size of the
square seen by a pixel of the IR camera on the specimen (the dimension are the same along the y and z-
directions), k the time index and t A the time step. It is considered in Eq. (6) that the heat exchange in the
x-direction by convection with ambient air is negligible. Besides, it can be seen in Eq. (6) that the heat
diffusion in the (y, z) plane of the specimen is considered as isotropic, because some of the anisotropic
properties of the alloy remain unknown (see Section 2.4). After some preliminary tests, which are not
detailed here, it was decided not to filter the temperature variation fields before processing them with Eq.
(6), mainly because the amplitude of the thermal signal was sufficiently high compared to the
measurement noise.

Let us conclude with some considerations on units. The heat source s is expressed in W.m
-3
. It is
generally useful to divide this quantity by C , as in Eqs. (3), (5) and (6). In this case, the advantage is
that only one thermophysical material quantity is needed for the calculation: the thermal diffusivity of the
material. Hence, the heat source ) /( C s is expressed in C.s
-1
.

Page 14 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
13
4.4. Amplitude of the temperature oscillations
From items 2 and 3 in Section 4.2, it appears to be relevant to measure the amplitude of the temperature
oscillations. A Fourier analysis was therefore performed because of the periodic nature of the signal (as
expected, experimental measurements exhibited temperature oscillations at the same frequency as the
load during Step 3). In practice, the amplitude T of the temperature oscillations was calculated at each
pixel on the specimen surface and a field denoted ( ) z y T , was obtained, where y and z are the
coordinates of the pixels.
The amplitude of the temperature oscillations T is expected to be representative of the zones of the
specimen which are not subjected to phase transformation during the cyclic loading. In this simple case of
a purely thermoelastic behaviour in adiabatic conditions, integrating Eq. (3) or (5) over one half cycle and
considering items 1, 5 and 6 in Section 4.2 lead to:
( )
el
A
C
T
T c o

:
~
0
=
(7)
where ( )
el
A c o :
~
is the amplitude of
el
A c o :
~
. Since o c
1
~

= A
el
, one can write:
( ) o o

1 0
~
:
~

= A A
C
T
T
(8)
When considering a uniaxial stress state along the z-direction, the ratio of the temperature amplitude T
over the longitudinal stress amplitude
macro
o can be written as follows:
( ) z z =
1 0
~
:
~
A A
C
T

T
macro
o
o
(9)
where the symbol is the tensorial product, such that
j i j i
b a = ] [ b a . This ratio depends on the
material parameters ( , C , A
~
and o ) and, in particular, on the anisotropic properties of the material.

Using data for the austenite phase given in Table 1,
macro
T o / is equal to 0.0016 C/MPa. Even if the
calculation cannot be carried out for the martensite (because of a lack of information), it can be expected
that the order of magnitude is similar.

Page 15 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
14
5. Experimental results

5.1. Mechanical response
Some preliminary comments must be given concerning the mechanical response of the specimen before
presenting its thermal response caused by the loading defined in Section 3.1 above. Fig. 3 shows the
strain-stress curves obtained with two types of tests: under isothermal conditions (thus performed at a
very low strain rate, as explained above) and under non-isothermal conditions. These tests differ first in
terms of strain rate of the quasi-static part of the test: 210
-5
s
-1
and 210
-3
s
-1
, respectively. They also
differ in terms of loading frequency used in the cyclic part of the tests (Step 3):
L
f = 0.008 and 13 Hz
respectively. The two tests are performed with nearly the same imposed strain amplitude
macro
c ~ 0.2%.
Two main comments can be drawn about the mechanical response:
- in Fig. 3-a, it can be seen that the return to ambient temperature approximately leads to a return to
the curve of the isothermal test (Ref. [34] provides a modelling of this phenomenon);
- Fig. 3-b shows an enlargement of the internal loops during the cyclic part of both tests, during
which the stress amplitude
macro
o was measured. It can be seen that the mean
macro macro
c o /
ratios are quite similar for the two used frequencies (0.008 and 13 Hz): about 3.5 GPa in both
cases. This low value of the stiffness is typical of the peculiar damping properties of SMAs.

Fig. 3. Mechanical response of the specimen. a) comparison between isothermal and non-isothermal tests.
b) enlargement of the cyclic part of the tests.

5.2. Thermal response
The thermal response measured during the cyclic step (Step 3) of the non-isothermal tests is analysed
below.

5.2.1. Preliminary remark about reproducibility
Fig. 4-a presents three different maps of temperature amplitude T obtained under the same loading
conditions. The specimen was removed from the jaws from one test to another and carefully replaced.
Page 16 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
15
Even if these maps correspond to the same macroscopic state (same value of
macro
c ), it clearly appears that
the local states are different. The differences between the local responses may be partially explained by
the positioning of the specimen in the jaws of the testing machine. For example, Fig. 4-b shows the effect
of a misalignment of the specimen which is deliberately introduced with respect to the loading direction.
It can be seen that this misalignment strongly modifies the temperature variation patterns. These remarks
about reproducibility are important because they enable one to understand why the different cases
detailed below sometimes lead to different temperature patterns.

Fig. 4. (color online) Reproducibility of the thermal response.

5.2.2. Evolution of the heat sources
Fig. 5 presents the evolution of the heat source ) /( C s during the cycling loading (Step 3) for
macro
c =
3.85%,
macro
c = 0.192 % and
L
f = 13 Hz.
----
Electronic version only:
Details on heat source evolution can be found in the video presented at the end of the paper. The maps
discussed in the current section are extracted from this video.
----
The heat source maps are calculated using the procedure described above. The time interval between two
successive maps is equal to 1/436 seconds (the acquisition frequency is
a
f = 436 Hz). So a complete
mechanical cycle corresponds to 33 successive maps. The maps presented in Fig. 5 span more than one
mechanical cycle to give a better idea of the phenomena occurring in the specimen. Some parasitic
boundary effects are clearly visible: for instance, the red line at the left-hand side of the specimen must
not be taken into account for the analysis. The fields are displayed in green (almost null heat sources),
except in certain localized zones in red (positive heat source) and blue (negative heat source). Positive
heat sources correspond to latent heat production: an austenite-to-martensite transformation occurs.
Negative heat sources correspond to latent heat absorption: a martensite-to-austenite transformation
occurs. During the cyclic loading (Step 3), positive and negative heat sources successively appear at
Page 17 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
16
nearly the same locations: so a cyclic austenitemartensite transformation occurs at the same frequency
as the load. The transformation zones are thin bands. The triangular zone denoted T in Fig. 5 is also
clearly composed of thin bands (needles), which are close to each other. The spatial resolution is however
sometimes not sufficient to clearly distinguish the needles in certain maps. On the top left-hand part of the
specimen, heat sources also exist, with a lower intensity compared to the other zones of transformation.

Fig. 5. (color online) Evolution of the heat sources s for
macro
c = 3.85%,
macro
c = 0.192 % and
L
f = 13
Hz. The time interval between two successive maps is 1/436 s.

The heat source patterns are mainly composed of inclined bands featuring all nearly the same orientation
with respect to the loading axis z (61 2 deg), except at the bottom left-hand part of the specimen, where
bands oriented by different angles are clearly visible. This angle can be compared with the angle
predicted in Section 2.2 (57 deg): this result shows that the bands are clearly related to habit planes, thus
to plane austenite-martensite interfaces. The different angles at the bottom left-hand part of the specimen
may be explained by a more complex state of stress that takes place at the bottom part of the specimen
because of the bottom jaw, which caused different activated variants of martensite to appear.

Some bands, such as those located at the bottom part of the specimen or in the triangular zone T, do not
cross the whole width of the specimen. On the contrary, the maps clearly exhibit two parallel lines
crossing the specimen width: one is the bottom boundary of zone T; the other one is located almost at the
middle of the specimen. The zone between these two lines is denoted Z in the following (see Fig. 5).
An interesting observation can be made about this zone. Since the heat source fields are quite noisy, the
reader is invited to focus on some zones indicated by arrows in Fig. 5 and also to consider the curves
plotted in Fig. 6:
- the inner part of zone Z is slightly blue (low negative heat source) when its two boundaries are
red (high positive heat source);
- the inner part of zone Z is slightly red (low positive heat source) when its two boundaries are blue
(high positive heat source).

Page 18 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
17
Fig. 6. (color online) Evolution of the heat source s in two zones of the specimen (for
macro
c = 3.85%,
macro
c = 0.192 % and
L
f = 13 Hz): along the boundary of zone Z and in the inner part of zone Z.

The curves plotted in Fig. 6 confirm these observations. They show the heat source vs. time for two
zones: at the boundary of zone Z and in the inner part of this zone. This result confirms that the heat
sources oscillate at the same frequency as the load: 13 cycles per second for a loading frequency
L
f = 13
Hz. Moreover, the heat source in the inner part of Z oscillates nearly in opposite phase with the heat
source along the boundary of the zone Z.
These results enable us to distinguish two phenomena which occur in the specimen: phase transformation
and thermoelastic coupling. Indeed, the minus sign in Eq. (4) indicates that the thermoelastic coupling
term is expected to oscillate in opposite phase with the elastic strain (thus in opposite phase with the
load), contrary to the latent heat.

A final identification of zone Z can be done from the two basic additional two considerations:
(i) the triangular zone T located just above zone Z is composed of non-crossing needles of martensite
inside austenite;
(ii) the two lines bordering zone Z can be interpreted as two habit planes (whose cyclic movements
lead to production and absorption of latent heat).
It can be logically deduced from all these considerations that zone Z is entirely composed of martensite.
The heat sources in the inner part of Z are then related to the thermoelastic coupling of the martensite
phase. Some additional comments about the triangular zone T are given in the following section, where
the influence of the stress amplitude is discussed.

5.2.3. Influence of the stress amplitude
Fig. 7-a shows the amplitude of the temperature oscillation T as a function of the stress amplitude
macro
o for several material points of the specimen (see Fig. 7-b). In practice, the different values of the
stress amplitude
macro
o are obtained by changing the strain amplitude
macro
c in Step 2. The map of
macro
T o / presented in Fig. 7-b has a spatial resolution which is worst than the spatial resolution of the
Page 19 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
18
heat source maps presented in Fig. 5. This is clearly visible by comparing for instance the bottom part of
the specimen between Figs. 5 and 7-b. The data in Fig. 7 have a resolution which is sufficient for the
analysis. Indeed, the
macro
T o / ratio was theoretically estimated to 0.0016 C/MPa for the austenite in
Section 4.2 above. With a thermal resolution of a few hundredths of one degree (see Section 3.2) and
thanks to the data processing carried out by discrete Fourier transform described in Section 4.2 above, the
resolution of the temperature amplitude T is estimated to be equal to some thousandths of one degree.
So, by dividing T by a stress amplitude
macro
o of some MPa, the resolution on the
macro
T o / ratio is
expected to be sufficient to observe the thermoelastic coupling effect in the specimen.

Fig. 7. (color online) Influence of the stress amplitude
macro
o on the temperature amplitude T for
macro
c
= 3.85% and
L
f = 13 Hz.

The following comments can be drawn from Fig. 7-a:
- the thermal responses at points 1 and 7 are similar. The corresponding curves are nearly linear,
with a mean
macro
T o / ratio equal to 0.0014 C/MPa. This value is in agreement with the
theoretical ratio corresponding to the thermoelastic coupling in austenite phase (0.0016 C/MPa,
see Section 4.2);
- at point 3, which belongs to the inner part of zone Z, the
macro
T o / ratio which is measured is
equal to 0.055 C/MPa. This value is greater than the theoretical ratio of austenite: about forty
times greater. In Section 5.2.2, zone Z was considered as entirely composed of martensite. Even
if the thermoelastic coupling in martensite cannot be theoretically estimated (see Section 4.2), it
would be expected to have the same order of magnitude as for the austenite. From Eq. (7), the
difference can be explained by different potential reasons: some effects due to the elastic strain
concentrations in zone Z, or some anisotropic effects combined with a non-purely uniaxial stress,
or some differences between the thermophysical properties of the two phases which could not be
determined or estimated here;
- the other material points exhibit higher
macro
T o / ratios, which cannot solely be attributed to
the thermoelastic coupling. Moreover, the
macro
o - T curves are not linear (the mean trends do
Page 20 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
19
not cross the origin of the graph). The latent heat is clearly involved in these zones. This means
that austenite-martensite interfaces exist in these zones: during the cyclic loading, their
movements lead to a cyclic latent heat production/absorption. Point 5, which belongs to the inner
part of the triangular zone T, exhibits a lower ratio than points 4 and 6, which belong to the
boundaries of this zone.
An interpretation of these observations in terms of microstructures is proposed in the following section.

5.2.4. Proposition of interpretation in terms of microstructures
This section proposes an interpretation of the thermal response in the triangular zone T in terms of
microstructures. The idea is that the thermal activity in this zone is mainly related to cyclic
production/absorption of latent heat, thus to the cyclic movement of austenite/martensite (A/M)
interfaces. Fig. 8-a distinguishes different zones in the specimen, which are deduced from the levels of
thermal activity. An arbitrary four-colour scale is used: white colour for the zones without movement of
A/M interfaces (for example for purely austenitic or purely martensitic zones), and three gray levels to
distinguish different intensities in the thermal activity related to cyclic movements of A/M interfaces. It
can be seen that a moderate, but non-null thermal activity exists in the inner part of zone T.

Fig. 8. (color online) Interpretation in terms of microstructures during the cyclic loading for
macro
c = 3.85%,
macro
c = 0.192% and
L
f = 13 Hz.

Fig. 8-b proposes a more detailed interpretation of the thermal activity in zone T in terms of
microstructure behaviour. In Section 5.2.2, it was proposed that the inner part of zone T was composed of
martensite needles inside austenite, which were close to each other. These needles have an inclination
with respect to the z-direction which is consistent with the habit-plane orientation predicted in Section
2.2. If one considers that these needles may have two modes of growth (longitudinal and lateral), some
differences of thermal activity in the triangular zone can be explained. The following interpretation of the
two-regime thermal activity in the enlarged zone in Fig. 8-b can be proposed. During the cyclic loading,
longitudinal movements of the A/M interfaces occur at the needle tips (for instance at point 6 in Fig. 7-b):
these movements create a significant thermal activity. The lateral movements of the needle interfaces (for
Page 21 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
20
instance at point 5 in Fig. 7-b) create a thermal activity which is lower. This result provides some
information about the mode of growth of the martensite in the austenite. First, this growth occurs along
both directions (longitudinal and lateral). Second, the movements of interfaces are greater along the
longitudinal direction of the needles than along the lateral direction. This interpretation justifies the
difference of colours (green and yellow) observed in the enlargement shown in Fig. 8-b.

5.2.5. Evolution of the thermal response as a function of the macroscopic strain
The objective is now to observe the microstructure evolution when the macroscopic mean strain
macro
c
changes. Several Step 1 Step 2 Step 3 sequences are successively applied to the specimen. The
same loading frequency
L
f = 13 Hz and strain amplitude
macro
c = 0.192% are used for all Steps 3. The
test is stopped when the specimen is quasi entirely martensitic: this can be observed when the slope of the
stress-strain curve during Step 1 becomes close to the value of the Youngs modulus of the martensite. A
return to zero stress is finally applied.
Fig. 9 presents several maps of the
macro
T o / ratio for different values of the macroscopic strain
macro
c .
The evolution of the pattern is clearly visible. Fig. 10 shows some comparisons with heat source maps for
three values of macroscopic strains. As already mentioned above, the heat source maps are clearer than
the
macro
T o / maps.

Fig. 9. (color online) Maps of the
macro
T o / ratio vs. macroscopic strain
macro
c (
macro
c = 0.192% and
L
f
= 13 Hz).

Fig. 10. (color online) Comparison between the map of
macro
T o / and the map of heat source s for three
values of the macroscopic strain
macro
c .

The following remarks can be drawn from Figs. 9 and 10:
- for the austenitic or almost austenitic states, say for
macro
c lower than 0.4%, the
macro
T o / field
is nearly equal to 0.0014 C/MPa. This value is related to the thermoelastic coupling in the
austenite phase. Some heterogeneities exist in the specimen, mainly in its bottom part (for
macro
c =
Page 22 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
21
0.15% and 0.19%), and then at the middle of the specimen (for
macro
c = 0.38%). Some of these
heterogeneities are localized along bands, so they can be related to needles of martensite, which
take place in the austenite. They probably appear first at the ends of the specimen because of the
jaws;
- on the strain-stress plateau, say from
macro
c greater than 0.4%, strong concentrations of the
macro
T o / ratio exist in the specimen. They are clearly localized along some bands. Most of
these bands are parallel. Fig. 10 provides some additional information by comparing these bands
with some heat source fields. For
macro
c = 0.96%, non-crossing needles of martensite can be
distinguished in the bottom right-hand region of the specimen. For
macro
c = 3.85%, numerous,
close and thin bands of heat source concentrations appear in most of the specimen (mainly in the
upper half region). This shows that the specimen is covered by numerous thin martensite zones;
- when the specimen approaches the purely martensite state (
macro
c = 7.70%), the heterogeneity of
the
macro
T o / ratio decreases but the upper part of the specimen still exhibits a strong
concentration. The heat source map presented in Fig. 10 shows that a few bands of heat source
concentration (latent heat) still exist in the specimen. The
macro
T o / ratio measured in the
martensite zone is equal to 0.004 C/MPa, thus about three times greater than in the austenite
phase. It can be noted that this value is however much lower than the value measured in the
martensitic zone Z discussed in Section 5.2.3 above: 0.055 C/MPa. This difference may be
explained by a more homogeneous stress distribution when the specimen reaches the purely
martensite state.

5.2.6. Complementary study: influence of the loading frequency
Some final comments can be made about the influence of the loading frequency
L
f on the
thermal response of the specimen. The lower the loading frequency, the greater the influence of
the thermal diffusion. Indeed, the latent heat produced by the phase transition diffuses inside the
specimen if the duration of the loading is long enough. For low values of loading frequency, the
heat diffusion penalizes the interpretation in terms of phase transition localization (note that a
Page 23 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
22
very low loading frequency leads to an isothermal response because of the heat exchanges with
the outside of the specimen). As the objective of the present study is to reveal martensitic
microstructures, the use of low loading frequencies is not relevant. Some complementary tests,
not detailed here, were performed to see the influence of higher loading frequencies. They were
performed for a given macroscopic strain
macro
c = 3.79% and a given strain amplitude
macro
c =
0.038%, with a frequency successively equal to 13 Hz, 17 Hz, 23 Hz, 27 Hz, 33 Hz and 51 Hz.
Two comments can be drawn from this complementary study. First, all the maps of temperature
amplitude T have the same appearance (the same pattern, which consists of inclined bands).
Second, the temperature amplitudes T slightly decrease as the loading frequency increases
from 13 to 51 Hz. Hence the situation seems to differ from the case of a purely thermoelastic
behaviour. Contrary to the case of the thermoelastic coupling [33], no theoretical result exists
about the cyclic adiabatic behaviour with local latent heat production/absorption, to the best
knowledge of the authors.


6. Conclusion

Understanding martensitic microstructures presents a great interest in mechanics of SMAs because they
influence the macroscopic mechanical behaviour of these materials. A new approach was proposed for
studying these microstructures. It is based on the processing of the thermal response of a SMA specimen
measured by infrared thermography. The tests discussed in this paper were performed on a single crystal
of Cu-Al-Be alloy. Two types of quantities were calculated from measured thermal fields: heat source and
amplitude of temperature oscillations. The maps showing these quantities enabled us to reveal martensitic
microstructures by highlighting two thermomechanical couplings:
- the thermoelastic coupling: in the austenite phase, the measured ratio between the temperature
amplitude T and the stress amplitude
macro
o is consistent with the corresponding theoretical
value (0.0014 and 0.0016 C/MPa, respectively). For the martensite phase, there is unfortunately a
Page 24 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
23
lack of information in the literature concerning the anisotropy of the thermal expansion tensor in
SMA martensite crystals. Such data would have helped to theoretically estimate the effect of the
thermoelastic coupling in the martensite phase. It is shown in this work that the effects of the
thermoelastic coupling are different in the two phases;
- the coupling related to the phase transition: during the small cyclic loading, latent heat was
repeatedly produced and absorbed. The strong thermal activity in the specimen was due to the
resulting cyclic movement of the austenite-martensite interfaces. The map of heat sources enabled
us to localize martensitic microstructures. By analysing the thermal activities, it was possible to
propose an interpretation in terms of microstructural phenomena (growth of martensite needles).

The experimental approach proposed in the present study is complementary to other full-field techniques
to analyse martensitic microstructures. Coupled measurements involving both IR and optical
measurements are expected to provide interesting results about martensitic microstructures in SMAs (see
Ref. [25] for instance).


Supplementary material

The online version of this article contains a video showing the evolution of the heat source ) /( C s
during the cycling loading (Step 3) for
macro
c = 3.85%,
macro
c = 0.192 % and
L
f = 13 Hz.

Please insert the video heat_sources.avi.

Caption of the video: Evolution of the heat source ) /( C s during the cycling loading (Step 3) for
macro
c =
3.85%,
macro
c = 0.192 % and
L
f = 13 Hz.

Page 25 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
24
References
[1] K. Bhattacharya. Microstructure of martensite: why it forms and how it gives rise to the shape-
memory effect. Oxford: Oxford University Press, 2003.
[2] J.A. Shaw, S. Kyriakides, J. Mech. Phys. Solids 4 (1995) 12431281.
[3] J.A. Shaw, S. Kyriakides, Acta Mater. 45 (1997) 683700.
[4] R. Peyroux, A. Chrysochoos, C. Licht, M. Lbel, Int. J. Eng. Sci. 36 (1998) 489509.
[5] X. Balandraud, A. Chrysochoos, S. Leclercq, R. Peyroux, C. R. Acad. Sci., Ser. IIb: Mec. 329 (2001)
621626.
[6] B. Vieille, J.F. Michel, M.L. Boubakar, C. Lexcellent, Int. J. Mech. Sci. 49 (2007) 280297.
[7] X.H. Zhang, P. Feng, H.J. He, T.X. Yu, Q.P. Sun, Int. J. Mech. Sci. 52 (2010) 16601670.
[8] D. Favier, H. Louche, P. Schlosser, L. Orgas, P. Vacher, L. Debove, Acta Mater. 55 (2007) 5310
5322.
[9] J.M Dulieu-Barton, P. Stanley, J. Strain Anal. Eng. Design 33 (1998) 93104.
[10] J. Eaton-Evans, J.M. Dulieu-Barton, E.G. Little, I.A. Brown, J. Strain Anal. Eng. Design 41 (2006)
481495.
[11] J. Van Humbeeck, J. Alloys Comp. 355 (2003) 5864.
[12] H. Seiner, M. Landa, P. Sedlak, Euromech Colloquium 478 on Non-Equilibrium Dynamical
Phenomena in Inhomogeneous Solids, June 13-16, 2006, Tallinn Univ. Technol., Tallinn, Estonia, In:
Proc. of the Estonian Acad. of Sci. Physics Mathematics 56 (2007) 218225.
[13] E. Vives, S. Burrows, R.S. Edwards, S. Dixon, L. Manosa, A. Planes, R. Romero, Applied Phys.
Lett. 98 (2011), article number 011902.
[14] Z. Moumni, A. Van Herpen, P. Riberty, Smart Mater. Struct. 14 (2005) S287S292.
[15] C. Chu, Hysteresis and microstructures: A study of biaxial loading in compound twins of copper-
aluminum-nickel single crystals, Ph.D. thesis, University of Minnesota, USA, 1993.
[16] Z. Zhang, R.D. James, S. Mller, Acta Mater. 57 (2009) 43324352.
[17] S. Stupkiewicz, G. Maciejewski, H. Petryk, Acta Mater. 55 (2007) 62926306.
[18] X. Balandraud, G. Zanzotto, J. Mech. Phys. Solids 55 (2007) 194224.
[19] H. Seiner, O. Glatz, M. Landa, Int. J. for Multiscale Computational Eng. 7 (2009) 445456.
[20] X. Balandraud, D. Delpueyo, M. Grdiac, G. Zanzotto, Acta Mater. 58 (2010) 45594577.
Page 26 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t
25
[21] P. Sittner, J. Pilch, P. Lukas, M. Landa, H. Seiner, P. Sedlak, B. Malard, L. Heller, In: P. Vincenzini,
S. Besseghini (Eds.), State-of-art research and application of SMAs technology, Book Series: Advances
in Science and Technology 59, Trans. Tech. Publications Ltd. Switzerland, 2009, pp. 4756.
[22] H. Seiner, M. Landa, Phase Transitions 82 (2009) 793807.
[23] C. Efstathiou, H. Sehitoglu, Scripta Mater. 59 (2008) 12631266.
[24] C. Efstathiou, H. Sehitoglu, J. Carroll, J. Lambros, H.J. Maier, Acta Mater. 56 (2008) 37913799.
[25] D. Delpueyo, M. Grdiac, X. Balandraud, C. Badulescu. Mech. Mater. (2011), submitted.
[26] Strukturbericht, B. II, 19281932. In: Niggli et al. P. (Eds.), Akademische Verlaggesellschaft
M.B.H. Leipzig. (Reproduced by Edwards Brothers, Ann Arbor, MI, 1943) Supplement to Z. Kristallogr.
Kristallgeom. Kristallphys. Krystallchem. See also: http://cst-www.nrl.navy.mil/lattice/struk/.
[27] S.N. Balo, M. Ceylan, M. Aksoy, Mater. Sci. Eng. A 311 (2001) 151156.
[28] C. Hsu, W. Wang, Y. Hsu, W. Rehbach, J. Alloys Comp. 474 (2009) 455462.
[29] R.D. James, H.K. Hane, Acta Mater. 48 (2000) 197222.
[30] X.Y. Zhang, L.C. Brinson, Q.P. Sun, J. Smart Mater. Struct. 9 (2000) 571581.
[31] L. Manosa, J. Zarestky, M. Bullock, C. Stassis, Phys. Rev. B 59 (1999) 92399242.
[32] S. Kustov, M. Morin, E. Cesari, Scripta Mater. 50 (2004) 219224.
[33] S. Offermann, J.L. Beaudoin, C. Bissieux, H. Frick, Exp. Mech. 37 (1997) 409413.
[34] X. Balandraud, E. Ernst, E. Soos, Z. Angew. Math. Phys. 51 (2000) 419448.

Page 27 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t

Fig. 1. Strain-stress curve at ambient temperature
0
T = 22C and under isothermal condition.
Figure(s)
Page 28 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t


Fig. 2. Schematic view of the loading procedure.
Page 29 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t


Fig. 3. Mechanical response of the specimen. a) comparison between isothermal and non-isothermal
tests. b) enlargement of the cyclic part of the tests.
Page 30 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t



Fig. 4. (color online) Reproducibility of the thermal response.
Page 31 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t



Fig. 5. (color online) Evolution of the heat sources s for
macro
= 3.85%,
macro
= 0.192 % and
L
f
= 13 Hz. The time interval between two successive maps is 1/436 s.
Page 32 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t



Fig. 6. Evolution of the heat source s in two zones of the specimen (for
macro
= 3.85%,
macro
=
0.192 % and
L
f = 13 Hz): along the boundary of zone Z and in the inner part of zone Z.
Page 33 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t



Fig. 7. (color online) Influence of the stress amplitude
macro
on the temperature amplitude T
for
macro
= 3.85% and
L
f = 13 Hz.
Page 34 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t


Fig. 8. (color online) Interpretation in terms of microstructures during the cyclic loading for
macro
=
3.85%,
macro
= 0.192% and
L
f = 13 Hz.
Page 35 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t



Fig. 9. (color online) Maps of the
macro
T / ratio vs. macroscopic strain
macro
(
macro
=
0.192% and
L
f = 13 Hz).
Page 36 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t



Fig. 10. (color online) Comparison between the map of
macro
T / and the map of heat source s
for three values of the macroscopic strain
macro
.

Page 37 of 37
A
c
c
e
p
t
e
d

M
a
n
u
s
c
r
i
p
t

Table 1
Physical properties of the austenite phase.
Material property Numerical value
Density 7,300 Kg.m
-3

Specific heat C 430 J. K
-1
.Kg
-1

Thermal expansion coefficient 1710
-6
K
-1

Thermal diffusivity D 4.2310
-5
m
2
.s
-1



Table(s)

Das könnte Ihnen auch gefallen