Sie sind auf Seite 1von 20

Acta Materialia 50 (2002) 42554274

www.actamat-journals.com

Ni4Ti3-precipitation during aging of NiTi shape memory


alloys and its influence on martensitic phase transformations
Jafar Khalil-Allafi a, Antonin Dlouhy b, Gunther Eggeler a,
a

Institut fur Werkstoffe, Ruhr-Universitat Bochum, Universitatsstrasse 150, D-44 780 Bochum, Germany
b
Institute of Physics of Materials, AS CR, Zizkova 22, CZ-616 62 Brno, Czech Republic
Received 22 March 2002; received in revised form 7 June 2002; accepted 10 June 2002

Abstract
The present work studies the microstructure of a Ni-rich NiTi shape memory alloy and its influence on the thermal
characteristics of martensitic transformations. The solution annealed material state is subjected to various isothermal
aging treatments at 773 K; this results in the nucleation and growth of lenticular coherent Ni4Ti3-precipitates, which
were quantitatively characterized using transmission electron microscopy (TEM). Stress free aging for 36 ks results in
a heterogeneous microstructure with precipitates near grain boundaries and precipitate free regions in grain interiors;
this microstructure shows a three step (multiple step) transformation behavior in a differential scanning calorimetry
(DSC) experiment on cooling from the B2 regime, which can neither be rationalized on the basis of a coherency stress
argument (Bataillard et al., 1997) nor on the basis of varying Ni-concentrations between growing precipitates (KhalilAllafi et al., 2002). A new interpretation of evolving DSC chart features is proposed which takes the evolution of
microstructures during stress free and stress-assisted aging into account. Most importantly it is shown that stresses as
small as 2 MPa strongly affect the precipitation process. 2002 Acta Materialia Inc. Published by Elsevier Science
Ltd. All rights reserved.
Keywords: Ni-rich NiTi shape memory alloys; Aging; Ni4Ti3 precipitation; Martensitic transformation; Transmission electron
microscopy (TEM); Differential scanning calorimetry (DSC)

1. Introduction
NiTi shape memory alloys combine good functional and structural properties [14]. There is an
interest in Ni-rich NiTi alloys because phase transition temperatures can be controlled through the
Ni-content [2,5]. Processing of NiTi alloys gener
Corresponding author. Tel.: +49 234 7003022; fax: +49
234 3214235.
E-mail address: gunther.eggeler@ruhr-uni-bochum.de (G.
Eggeler).

ally involves thermo mechanical treatments, which


lead to the precipitation of metastable Ni4Ti3-particles [2]. There is a good understanding of the
crystallographic features of the Ni4Ti3-precipitates
which are coherently precipitated in the matrix [2],
have a lenticular shape [2] and can form eight variants [6] on {111}-planes [2,6]. The precipitates
give rise to coherency stress fields and external
stresses can favor the occurrence of certain precipitate variants [2,6,7]. It is well known that Ni4Ti3particles affect the features of the martensitic transformation in supporting the formation of the R-

1359-6454/02/$22.00 2002 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved.
PII: S 1 3 5 9 - 6 4 5 4 ( 0 2 ) 0 0 2 5 7 - 4

4256

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

phase [8] and affecting the Ni-content of the matrix


[911]. But the role of Ni4Ti3-precipitates on martensitic transformations as observed using differential scanning calorimetry (DSC) has been discussed
controversially in the literature. Thus Bataillard et
al. [12] ascribed multiple step martensitic transformations (one stepone distinct DSC peak, multiple stepsmultiple distinct DSC peaks) to coherency stress fields around precipitates while KhalilAllafi et al. [13] explained two step and multiple
step transformation behavior on the basis of evolving Ni-concentration profiles between particles and
differences in nucleation barriers between R-phase
and B19. Two recent results show that aging of
annealed, defect free materials results in heterogeneous grain boundary precipitation [14] and that
small differences in preload strongly affect the
creep behavior of Ni-rich NiTi alloys [11]; here it
is important to highlight that creep rate is a sensitive probe for microstructural evolution. These two
recent findings suggest that more experimental
work is required to understand the precipitation
processes in Ni-rich NiTi alloys. In order to contribute to a better understanding of precipitation
processes and their influence on martensitic transformations, the present work uses transmission
electron microscopy (TEM) to investigate how
microstructures form during stress free and stress
assisted aging and how these microstructures affect
the thermal characteristics of martensitic transformations.

2. Experiments
A binary nickel-rich NiTi alloy with a nominal
composition of 50.7 at.-% Ni was investigated in
the present study. It was purchased from Memory
Metals, Weil am Rhein in the form of cylindrical
rods of 1 m length and 13 mm diameter. A TEM
micrograph of the microstructure after solution
annealing (1123 K, 900 s) and water quenching is
shown in Fig. 1. Solution annealing results in a
homogeneous equiaxed microstructure with only a
few Ti4Ni2O oxide particles and no Ni4Ti3 precipitates.
Aging of the solution annealed and water
quenched material states was performed at 773 K

Figure 1. TEM micrograph showing a homogeneous equiaxed


microstructure of the Ni-rich NiTi alloy investigated in the
present study (nominal composition: 50.7 at.-% Ni) after solution annealing (1123 K, 900 s) and water quenching. The grain
size of the material as determined by optical microscopy was
35 m. The TEM micrograph shows two grain boundaries (at
the top of the micrograph) and a few Ti4Ni2O oxide particles
(which were chemically analyzed in the TEM using EDAX).

for aging times of 3.6 ks, 36 ks and 360 ks. Experiments to study the effect of superimposed stresses
on aging were carried out in creep machines using
cylindrical specimens with three diameters
(corresponding to stresses of 2, 8 and 20 MPa) to
obtain information on the effect of three stress levels on aging (stress assisted aging); no accumulation of plastic strain was detected during stress
assisted aging. A small piece of material was kept
right next to the cylindrical specimen in the furnace
to provide material, which was aged without stress
(stress free aging). The details of the creep
machines used in the present investigation have

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

been described elsewhere [15] and the geometry of


the specimen used for stress assisted aging will be
published shortly [16]. From all aged materials,
specimens were prepared for DSC measurements
and for TEM investigations. The transformation
behavior during heating and cooling was investigated using a differential scanning calorimeter
DSC 2920 CE from TA Instruments. DSC specimens with masses between 20 and 50 mg were
heated up to 373 K where they were held for three
minutes to establish thermal equilibrium. Then the
DSC measurement started by cooling down to 173
K in 0.17 K/s. At 173 K the specimen was again
held for 3 min and then heated up to 373 K with a
heating rate of 0.17 K/s. Recent neutron diffraction
results confirmed that the matrix of the alloy investigated in the present study was B2 at 373 K and
B19 at 173 K [17]. DSC charts which are obtained
during cooling (from the temperature where B2 is
stable, B2 regime) and during heating (from the
B19 regime) are shown in Figs. 2 and 3.
Transmission electron microscopy (TEM) was
performed using a Philips CM20 analytical microscope operating at 200 kV. TEM specimens were
cut perpendicular to the axis of the as received rod
or to the axis of the cylindrical specimen used for
stress assisted aging. The specimens were carefully
ground to a thickness of 150 m and then electro
polished using a double jet thinning technique
(Tenupol; Struers A8 electrolyte at 20V and intermediate flow rate). TEM was used to obtain micrographs, to identify phases by selected area diffraction (SAD) and for chemical analysis using EDAX.
During the preparation of TEM foils an effort was
made to keep the material from transforming into
martensite by using an electrolyte temperature of
288 K where good thinning conditions were
obtained and where sufficient parts of the matrix
remained either austenitic or R-phase (for short and
intermediate aging times). However, as can be seen
from Fig. 2, the martensite finish temperature after
360 ks of stress free aging is high and therefore no
reliable precipitate information was obtained after
long-term aging.
Precipitate volume fractions were evaluated
from TEM foils where the thickness was measured
using a stereo-pair method [18]. The particle shape
was approximated as an ellipsoidal disk with a

4257

Figure 2. DSC charts on cooling from the B2 regime. The four


DSC charts show the influence of aging time. The DSC chart
of the solution annealed (1123 K, 900 s) and water quenched
material is shown at the bottom. Three additional DSC charts
were measured after 3.6, 36 and 360 ks of stress free aging at
773 K. The small vertical bars after 3.6 and 36 ks of aging (and
the corresponding horizontal error bars) represent calculated
peak temperatures based on TEM Ni4Ti3 precipitate volume
fraction measurements and available data on the influence of
Ni on phase transition temperatures (see text). For the 36 ks
stress free aging conditions a small vertical arrow indicates the
temperature where the transformation starts.

diameter D and a thickness t. Tilting experiments were performed in order to determine D and
t values. Fig. 4a and b show two TEM micrographs
from a material which was aged at 773 K for 36
ks under 2 MPa. The two micrographs show the
same region in the TEM foil for two different tilt
positions. Fig. 4a and b demonstrate that TEM foils
can be tilted into positions that allow measuring
the precipitate diameter D (Fig. 4a) and in other
positions where the precipitate thickness t can be
obtained (Fig. 4b). For the material state shown in

4258

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

average to give a resulting volume fraction Vf of


6.7 3.5% for the material state shown in the
TEM micrographs of Fig. 4 (stress assisted aging:
773 K, 36 ks, 2 MPa). All Vf results presented in
this paper are based on the evaluation of more than
100 particles. Details on the TEM procedure for
determining precipitate volume fractions will be
reported elsewhere [19]. In several cases the cube
projection method was used to transfer crystallographic information (on directions and planes)
from convergent beam electron diffraction patterns
to TEM micrographs [20].

3. Results
3.1. Stress free aging

Figure 3. DSC charts on heating from the B19 regime. The


four DSC charts show the influence of aging time. The DSC
chart of the solution annealed (1123 K, 900 s) and water
quenched material is shown at the bottom. Three additional
DSC charts were obtained after 3.6, 36 and 360 ks of stress
free aging at 773 K. The small vertical bars after 3.6 and 36 ks
of aging (and the corresponding horizontal error bars) represent
calculated peak temperatures based on TEM Ni4Ti3 precipitate
volume fraction measurements and available data on the influence of Ni on phase transition temperatures (see text).

Fig. 4, one TEM foil was evaluated. Measurements


were performed at four locations of the TEM-foil.
The corresponding foil thickness, numbers of precipitates, average precipitate diameters D and average particle thickness t are summarized in Table 1.
The evaluation of the measurements presented in
Table 1 yields weighted averages for 137 particles;
average D and t values were obtained as
335 120 nm and 32 7 nm, respectively. Four
foil volumes were calculated from the four foil
thickness given in Table 1 and the areas of the corresponding micrographs (not listed in Table 1).
Four volume fractions were thus obtained which

In this section we report the effect of stress free


aging on martensitic transformations when cooling
from the B2-regime to low temperatures and when
heating B19-microstructures. The corresponding
DSC charts are presented in Fig. 2 (cooling from
B2) and 3 (heating from B19). We first consider
the DSC charts in Fig. 2 where DSC charts evolve
with aging time. There are three characteristic features that change with aging: (1) the type of transformation changes from one step (after solution
annealing) through two steps (after 3.6 ks aging)
and three steps (after 36 ks aging) back to one step
(after 360 ks aging). (2) The width of the overall
transformation temperature range changes from
small (after solution annealing) through large (for
short and intermediate aging times) back to narrow
for (long aging times). (3) There are shifts in peak
positions, and of temperatures where transformations begin and end. Table 2 lists all temperatures
of distinct DSC peaks (peak temperatures)
observed in Fig. 2 (cooling from B2-regime). Fig.
3 shows the DSC results that were obtained for
heating from low temperatures. Again there is an
evolution of DSC chart features with aging time.
The type of transformation changes from one step
(after solution annealing) through two steps (after
short and intermediate aging times) back to one
step (after long term aging). Three step transformations were not detected for the back transform-

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

4259

Figure 4. TEM micrographs of a microstructure after stress assisted aging at 773 K for 36 ks under 2 MPa. (a) TEM foil in tilt
position to evaluate precipitate diameters D. (b) TEM foil in tilt position to evaluate the thickness t of precipitates.
Table 1
Microstructural parameters from a material which was aged at 773 K for 36 ks under 2 MPa. Data for four locations in one TEMfoil are listed and average values of the local foil thickness, numbers of precipitates, the local mean precipitate diameter D and the
local mean precipitate thickness t are reported
Location

1
2
3
4

Foil thickness [nm]

Number of precipitates

78
111
131
87

32
29
34
42

ations. The peak positions observed in Fig. 3 are


listed in Table 2.
On cooling from the B2-regime, DSC charts of
the Ni-rich NiTi material that was aged without
stress show the following features: (1) the temperature where the transformation starts increases with
aging time. (2) The position of the first distinct
DSC peak on cooling shifts towards higher temperatures with increasing aging time. (3) The solution annealed material and the materials which
were aged for short (3.6 ks) and intermediate (36
ks) aging periods have two common characteristics: after aging the positions of the last peaks on
cooling correspond to the position of the single
DSC peak obtained for the solution annealed

Mean precipitate
diameters D [nm]
334
428
256
335

Mean precipitate
thickness t [nm]
30
39
33
27

material; and in all three material states the transformations end at the same temperature (which
generally is referred to as the martensite finish temperature, Mf). On heating from the B19-regime the
DSC charts of the material which was aged without
stress can be described as follows: (1) The back
transformations of the solution annealed material
and the materials which were aged for short (3.6
ks) and intermediate (36 ks) aging times start at
the same temperature. These three material states
also show a distinct DSC-peak at the same temperature (single peak of the solution annealed
material and the first peaks on heating of the aged
materials). (2) The positions of the last distinct
peaks shift to higher temperatures with increasing

4260

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

Table 2
Overview summary of peak positions in the DSC charts of Figs. 2 and 3. The Table also shows results for peak positions which
were calculated using Eqs. (1) and (3). Solution annealing was performed at 1123 K for 900 s. All subsequent aging treatments were
performed at 773 K in the absence of an external stress
Material/peak position:

1st peak on cooling


[K]

2nd peak on cooling


[K]

3rd peak on cooling


[K]

predicted Mp [K]

solution annealed
1 h aging
10 h aging
100 h aging

255.8
278.1
289.4
291.4

254.1
280.1

256.3

278.29
290.231
-

Material/peak position:

1st peak on heating


[K]

2nd peak on heating


[K]

3rd peak on heating


[K]

predicted Ap [K]

Solution annealed
1 h aging
10 h aging
100 h aging

287.1
286
282.4
323.9

300.8
312.3

309.29
321.231

aging times. During both heating and cooling the


difference in peak temperatures between the single
distinct peak observed for the solution annealed
material and the single distinct peak observed after
360 ks aging is 35 K.
The microstructures of the material states that
were aged at 773 K (without stress) for 3.6 ks and
for 36 ks are shown in the TEM micrographs of
Figs. 5 and 6. Fig. 5 shows that heterogeneous
nucleation of Ni4Ti3-precipitates is an important
factor in the early stages of aging. Precipitation of
small lenticular Ni4Ti3 precipitates only occurs on
and near grain boundaries and near oxide particles
(large globular particles stemming from processing
and present in the as received material, see Fig. 1).
The major part of the grain shown in Fig. 5 is free
of precipitates. Increasing the aging time to 36 ks
results in the microstructure shown in Fig. 6.
Clearly, precipitates are coarser than after shorter
aging times. The presence of particles in a belt
trimming the grain boundaries is still a prominent
feature; and in addition precipitates form irregular
networks in the interior of the grain. But even after
36 ks of aging, large parts of the grain interior are
free of precipitates.
We now show the precipitate microstructures
near the grain boundaries after 3.6 and 36 ks of
aging at a higher magnification, Fig. 7a and b. The
micrographs give a good impression on how pre-

Figure 5. TEM micrograph of a grain in a material which was


subjected to stress free aging (773 K, 3.6 ks). There are two
types of particles: (i) large globular Ti-rich Ti4Ni2O oxides can
be found in the grain interior and near grain boundaries. (ii) fine
lenticular Ni-rich Ni4Ti3 precipitates nucleate heterogeneously
close to the grain boundaries and near the oxide particles.

cipitates grow during aging. Both micrographs are


part of stereo pairs from which the spatial orientation of precipitates was determined; for this purpose, cubes with an inscribed tetraedron showing

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

Figure 6. TEM micrograph of a grain in a material which was


subjected to stress free aging (773 K, 36 ks). Ni4Ti3-precipitates
are coarser than after shorter aging times (see Fig. 5). An
important portion of the precipitates is located near grain
boundaries. All other particles form irregular networks in the
grain interior. There still are large areas in the grain interior
which are precipitate free.

4261

the 100 - and 110 -directions of the


cubic B2 matrix [20] are superimposed on the
images of the upper grain in Fig. 7a and the lower
grain in Fig. 7b. The cube projection method [20]
allows differentiating between four {111} precipitate plane groups. After 3.6 ks aging the precipitates near the grain boundary (upper grain) belong
mainly to the plane groups (11 1 ) and (1 1 1). After
36 ks aging the lenticular particles close to the
grain boundary (lower grain) are parallel to (11 1 )-,
(1 1 1)- and (111)-planes of the B2-matrix. No
further effort was made in the present study to
quantitatively account for the distribution of
Ni4Ti3-precipitates into 111 -plane groups.
In Table 3 we summarize the results on mean
particle diameter D, mean particle thickness t, the
interparticle spacing l and Vf- the volume fraction
of precipitates which were obtained for the
material subjected to stress free aging at 773 K for
3.6 and 36 ks. After 3.6 ks aging, precipitate volume fractions were determined for the grain
boundary regions. For the 36 ks aging condition,
precipitate volume fractions were determined from
all regions containing precipitates; the precipitate
free regions were not considered. Therefore the

Figure 7. TEM micrographs of particles near grain boundaries after stress free aging of a Ni-rich NiTi alloy at 773 K for (a) 3.6
ks and (b) 36 ks. (a) After 3.6 ks aging the precipitates near the grain boundary (upper grain) belong mainly to the plane groups
(11 1 ) and (1 1 1). (b) After 36 ks aging the lenticular particles close to the grain boundary (lower grain) are parallel to (11 1 )-,
(1 1 1)- and (111)-planes of the B2-matrix.

4262

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

Table 3
Microstructural data characterizing the Ni4Ti3-precipitate population after stress free aging for 3.6 and 36 ks at 773 K. Listed are the
microstructural parameters Dthe mean diameter of the main circle of the lenticular precipitates, tthe mean thickness of the
lenticular precipitates, lthe interparticle spacing and Vfthe volume fraction of precipitates. After 3.6 ks aging the particle volume
fraction was determined from grain boundary regions. After 36 ks aging the particle volume fraction was determined from grain
boundary regions and from the irregular particle network regions in the grain interior (particle free areas were not considered!)
Material state (stress free aging at 773 K)

D [nm]

t [nm]

l [nm]

Vf [%]

3.6 ks
36 ks

230150
900340

236
6821

28520
785180

2.81.0
5.24.5

precipitate volume fractions reported here over


predict the overall precipitate volume fraction for
the 3.6 and 36 ks aging conditions. When taking
a closer look at the Ni4Ti3 precipitates in the grain
boundary regions (Figs. 5 and 7a, material state
after stress free aging at 773 K for 3.6 ks) it can
be seen that the precipitate size increases with
increasing distance of the precipitates from the
grain boundary; a quantitative evaluation is
presented in Fig. 8.
Fig. 5 to 7 clearly show that during aging pre-

Figure 8. Diameter D of lenticular Ni4Ti3 precipitates as a


function of the distance from the grain boundary after stress
free aging at 773 K for 3.6 ks. Two horizontal lines indicate
mean particle diameters observed in microstructures with homogeneously distributed precipitates after stress assisted aging
(dotted line: 773 K/2 MPa/3.6 ks; dashed line: 773 K/20
MPa/3.6 ks).

cipitates continue to nucleate and grow. However,


precipitate growth is not simply characterized by
an Ostwald ripening type of process [21,22],
because (i) the microstructure has not yet reached
global thermodynamic equilibrium and (ii) there
are additional crystallographic aspects that need to
be considered. Thus Fig. 9 shows diffraction patterns from small (3.6 ks aging, Fig. 9a) and large
Ni4Ti3-precipitates (36 ks aging, Fig. 9b). Both diffraction patterns show the characteristic precipitate
spots that subdivide the reciprocal distance associated with 321 B2 into seven intervals [6].
In the case of small precipitates (like those shown
in Fig. 7a) the diffraction spots are weak and are
only associated with one type of precipitate variant,
Fig. 9a. Large precipitates (which are shown in the
TEM-micrograph of Fig. 7b) create stronger spots
and can consist of two precipitate variants as can
be seen from the twin related rows of corresponding spots, Fig. 9b. Another TEM observation after
stress free aging at 773 K for 3.6 ks is shown in
Fig. 10; here precipitates form two parallel chains.
These chains of precipitates can be rationalized on
the basis of a scenario where one precipitate
nucleates in the stress field of another; this type of
auto catalytic heterogeneous precipitation has been
described for other systems [23] and contributes to
the formation of the irregular networks observed
after longer term aging (see Fig. 6).
3.2. Stress assisted aging
We now report the thermal and microstructural
results which were obtained for the material states
subjected to stress assisted aging; it is important to
underline that small stresses of 2, 8 and 20 MPa

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

4263

Figure 10. TEM micrograph after stress free aging at 773 K


for 3.6 ks. Precipitates form two parallel chains. One precipitate
nucleates in the stress field of another.

Figure 9. Diffraction patterns from Ni4Ti3 precipitates after


stress free aging at 773 K. (a) a small particle (3.6 ks aging,
Fig. 7a) and (b) a large particle (36 ks aging, Fig. 7b). Both
diffraction patterns show the characteristic precipitate spots [6]
which subdivide the reciprocal distance associated with
321 B2 into seven intervals. (a) For a small precipitate the
particle spots (highlighted in the diffraction pattern) are weak
and are associated with one precipitate variant. (b) One large
particle can consist of two precipitate variants from the same
plane group as can be seen from the twin related rows of corresponding spots (highlighted in diffraction pattern).

were considered which at 773 K did not result in


any detectable plastic deformation. Fig. 11 shows
the effect of stress-assisted aging on the DSC chart
features on cooling. The DSC chart of the solution
annealed (1123 K, 900 s) and water quenched
material is shown at the bottom. Nine additional
DSC charts were measured after 3.6, 36 and 360
ks of stress-assisted aging (2 MPasolid line, 8
MPadashed line, 20 MPadotted line) at 773
K. The DSC charts (on cooling from the B2
regime) after stress assisted aging in Fig. 11 significantly differ from the DSC charts after stress
free aging presented in Fig. 2; the actual stress
level during stress assisted aging does not strongly
affect the DSC chart features after 36 and 360 ks
of aging. After 3.6 ks aging, increasing stresses
promote a three-step transformation: a third dis-

4264

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

Figure 11. DSC charts of the Ni-rich NiTi alloy investigated


in the present study on cooling from the B2 regime. The four
graphs show the influence of stress assisted aging on the thermal
effects associated with the martensitic transformations. The
DSC chart of the solution annealed (1123 K, 900 s) and water
quenched material is shown at the bottom. Nine additional DSC
charts were measured after 3.6, 36 and 360 ks of stress assisted
aging (2 MPasolid line, 8 MPadashed line, 20 MPadotted line) at 773 K. For the 36 ks stress assisted aging condition
a small vertical arrow indicates the temperature where the transformations start.

tinct DSC peak (which cannot be clearly detected


in the DSC chart observed after stress assisted
aging under 2 MPa and which is absent after 3.6
ks of stress free aging in Fig. 2) becomes clearly
visible when aging under stresses of 8 and 20 MPa.
After 360 ks, all materials that were aged under
stress show a broad transformation peak that has
two distinct maxima (while there was only one
sharp peak after stress free aging, Fig. 2). It is also
interesting to note that after 36 ks of aging at 773
K the transformations on cooling start significantly

earlier after stress assisted aging as compared to


stress free aging (see small vertical arrows pointing
to the start of transformations in Figs. 2 and 11).
In Table 4 we summarize the DSC peak temperatures that characterize stress assisted aging on cooling from the B2-regime.
Fig. 12 shows the effect of stress assisted aging
on the transformation behavior during heating from
the B19 regime. The DSC chart of the solution
annealed (1123 K, 900 s) and water quenched
material is shown at the bottom. Nine additional
DSC charts were measured after 3.6, 36 and 360
ks of stress-assisted aging (2 MPasolid line, 8
MPadashed line, 20 MPadotted line) at 773
K. As a striking difference between the DSC charts
after stress free aging (Fig. 3) and the DSC charts
obtained after stress assisted aging we note that the
pronounced two step transformation which is
observed for 3.6 and 36 ks of stress free aging can
no longer be detected. It should not be overlooked,
however, that 3.6 ks aging under 20 MPa results
in a peak which has a pronounced shoulder on its
high temperature side; the same microstructure
produces three distinct DSC peaks on cooling, Fig.
11. In Table 5 we summarize the DSC peak temperatures which characterize stress assisted aging
on heating from the B19 regime.
The differences in DSC transformation behavior
between the material states subjected to stress free
and stress assisted aging is due to the difference in
corresponding microstructures. Four characteristic
TEM micrographs that document the microstructures after stress-assisted aging at 773 K are shown
in Fig. 13. The top and bottom rows of Fig. 13
show microstructures after 3.6 and 36 ks of stressassisted aging, respectively. The left and right columns of Fig. 13 show microstructures that were
aged at 773 K in the presence of stresses of 2 and
20 MPa, respectively. As can be seen from Fig.
13, stress assisted aging results in a homogeneous
distribution of precipitates for all levels of applied
stress; in this respect it does not seem to matter
whether stress assisted aging was performed under
2 or 20 MPa. However, higher stress levels are
associated with slightly larger precipitates. Fig. 13
also clearly shows that precipitates coarsen with
time. The results of the quantitative TEM evaluation for the microstructures shown in Fig. 13 are

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

4265

Table 4
Overview summary of peak positions in the DSC charts of Fig. 11. The table also shows results for peak positions which were
calculated using Eqs. (1) and (3). All stress assisted aging treatments were performed at 773 K under stresses of 2, 8 and 20 MPa
Material/peak position

1 h aging, 2 MPa
1 h aging, 8 Mpa
1 h aging, 20 MPa
10 h aging, 2 MPa
10 h aging, 8 MPa
10 h aging, 20 MPa
100 h aging, 2 MPa
100 h aging, 8 MPa
100 h aging, 20 MPa
a

1st peak on cooling


[K]

2nd peak on cooling


[K]

289.3
289
289
293.7
293.6
293.7
293.7
295.9
296.2

3rd peak on cooling


[K]

281.2a
274.9
272.9
277.3
277.7
279.1
285.1
288.3
288.7

259.9
260.9
260.6

Predicted Mp [K]

270.29

274.210
295.226

296.217

no distinct peak

summarized in Table 6. The mean values of the


diameters D of the lenticular precipitates after 3.6
ks of stress-assisted aging (at 2 and 20 MPa) are
included in Fig. 8.
Stress assisted aging (even under stresses as
small as 2 MPa) results in homogeneous precipitation throughout the microstructure; and with
increasing aging time precipitate volume fractions
increase and precipitates coarsen, Fig. 13. Fig. 14
shows a constant number density of Ni4Ti3 precipitates after stress-assisted aging (773 K/20 MPa/3.6
ks). A two-beam condition was adjusted for the
grain on the right side of the grain boundary that
traverses the micrograph vertically (g: (1 10)). The
area marked A on the left side of the grain
boundary is shown in the TEM micrograph of Fig.
15 under a different contrast condition. The image
contrast in Fig. 15 was obtained using a g of
(101 ). It can be clearly seen that the precipitates
near the grain boundary (region A) differ in contrast from the majority of precipitates that fill the
interior of the grain. Both types of precipitates
have similar number densities and size distributions. However, they belong to different Ni4Ti3
variant groups. Most importantly, the precipitates
that show the strongest contrast (precipitate plane
group(1 1 1)) are absent near the grain boundary
(region A).

4. Discussion
4.1. Precipitation in Ni-rich NiTi alloys during
aging
The results of the present study confirm many
of the previous experimental and analytical results
reported in the literature on nucleation and growth
of Ni4Ti3 precipitates [6,7,2428]. Most
importantly they confirm the recent findings of
Filip and Mazanec [14] who reported that Ni4Ti3
precipitates form mainly at grain boundaries when
the over saturated NiTi matrix is defect free. The
results of our study show that heterogeneous grain
boundary precipitation is the dominant feature of
Ni-rich NiTi microstructures after solution
annealing and subsequent aging at 773 K for times
up to 36 ks. The present investigation shows that
this heterogeneous grain boundary precipitation is
no longer observed, when the aging is performed
in the presence of stresses as small as 2 MPa. Such
stress assisted aging results in microstructures with
a homogeneous distribution of precipitates in terms
of number density; however, there is a difference
between grain interiors and regions near grain
boundaries in terms of the precipitate variants that
are observed, Figs. 14 and 15.
Grain boundaries are well known as representing

4266

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

Figure 12. DSC charts of the Ni-rich NiTi alloy investigated


in the present study on heating from the B19 regime. The four
graphs show the influence of stress assisted aging on the thermal
effects associated with the martensitic transformations. The
DSC chart of the solution annealed (1123 K, 900 s) and water
quenched material is shown at the bottom. Nine additional DSC
charts were measured after 3.6, 36 and 360 ks of stress assisted
aging (2 MPasolid line, 8 MPadashed line, 20 MPadotted line) at 773 K.

locations for heterogeneous nucleation in solidstate precipitation processes for a number of


reasons [29]. In Ni-rich NiTi alloys there may well
be a higher Ni-concentration at the grain boundaries; moreover grain boundaries energetically
favor nucleation by decreasing the interfacial
energy between the precipitate and the parent
phase [29]. The results shown in Figs. 5, 7 and 8
can be interpreted as follows: initially many precipitates form at the grain boundary; their sizes
remain small because there is competitive growth.
The coherency stress fields of the early precipitates
assist in the nucleation of other precipitates which
form less frequently in some distance from the

grain boundary and therefore can grow to larger


sizes. It is interesting to note that the sizes of grain
boundary precipitates after stress free aging at 773
K for 1 h are comparable to the dimensions of precipitates in the grain interior after stress assisted
aging at 773 K for one hour in the presence of an
external stress of 2 and 20 MPa, Fig. 8.
Small grain boundary stresses may therefore
well be important in triggering grain boundary precipitation; simple formulas for the stress distributions near grain boundaries (modeled by dislocation arrays) are given in [30]. Using formula (1978) from [30] and assuming a dislocation spacing
of 10 nm in the boundary one can estimate stresses
of the order of 1 MPa in distances of the order of
20 nm from the boundary; therefore short range
stress fields associated with grain boundaries also
need to be considered when discussing grain
boundary precipitation of Ni4Ti3 in solution
annealed Ni-rich NiTi alloys during stress free
aging as presented in Figs. 5, 7 and 8. However,
this does not explain the microstructural heterogeneity shown in Fig. 15 that occurs under conditions
of aging at 773 K for 1 h in the presence of a
stress of 20 MPa. In a zone of 2 m near the grain
boundary (region A in Fig. 15) one precipitate variant that shows good contrast in the grain interior
is absent. This microstructural heterogeneity which
evolves under stress assisted aging may well be
associated with a non homogenous stress distribution in the microstructure due to different orientations of two neighboring B2 grains with respect
to the loading axis; further work is required to clarify this point. For the present work it is important
to highlight that Ni4Ti3 precipitation during short
term aging at 773 K can be heterogeneous no matter whether the material is aged in the presence or
in the absence of stress; under conditions of stress
free aging, a microstructure forms which is characterized by precipitate belts which trim the grain
boundaries. In the presence of small stresses the
environments of grain boundaries contain other
precipitate variants than the grain interiors. From
the results reported in the present study it can also
be concluded that the nucleation is much more
sensitive to the presence/absence of an external
stress than the subsequent precipitate coarsening
(see mean values in Tables 3 and 6); Fig. 13 clearly

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

4267

Table 5
Overview summary of peak positions in the DSC charts of Fig. 12. The table also shows results for peak positions which were
calculated using Eqs. (1) and (3). All stress assisted aging treatments were performed at 773 K under stresses of 2, 8 and 20 MPa
Material/peak position

1st peak on heating


[K]

2nd peak on heating


[K]

3rd peak on heating


[K]

1 h aging, 2 MPa
1 h aging, 8 MPa
1 h aging, 20 MPa
10 h aging, 2 MPa
10 h aging, 8 MPa
10 h aging, 20 MPa
100 h aging, 2 MPa
100 h aging, 8 MPa
100 h aging, 20 MPa

311.1
307.1
306.1
319.6
318.2
317.5
327.7
323.2
321.9

shows that an increase of one order of magnitude


in aging time has a much stronger effect on precipitate size than a one order increase in the level
of the superimposed external stress. Ti4Ni2O
oxides that form during material processing [2]
also act as preferential nucleation sites for Ni4Ti3
precipitates, Figs. 5 and 6. Since the density of
oxides is small for the present material this contribution to microstructural heterogeneity is not considered further.
4.2. Precipitation and Ni-concentration of the
matrix
In the previous section we have discussed Ni4Ti3
precipitation (Ni-concentration cP of precipitates:
57.1 at.-%) in a supersaturated Ni-matrix with a
Ni-concentration c0 of 50.7 at.-%. TEM was used
to determine precipitate volume fractions Vf as precisely as possible. Growth of Ni-rich precipitates
results in a decrease of Ni-concentration in the
matrix where the average Ni-concentration cNi for
a given volume fraction of precipitates Vf is given
by [9,11]
cNi

c0VfcP
1Vf

(1)

In a recent literature review Tang et al. [5] report


the dependence of the martensite start temperature
Ms in solution annealed NiTi alloys on Ni-concentration cNi; some of the data from [5] are shown in
Fig. 16 (original sources referenced in [5]: [31

predicted Ap [K]

301.29

305.210
326.226

327.217

35]). A polynomial fit of fifth degree rationalizes


all Ms-data in Fig. 16 (literature data [5,3135] and
the experimental result of the solution annealed
material of the present study):
Ms(cNi) (a0 a1cNi a2c2Ni a3c3Ni

(2)

a4c4Ni a5c5Ni)K
The dashed line which rationalizes all Ms-data
in Fig. 16 is obtained using the coefficients a0 to
a5 listed in Table 7. In the present study we assume
that we can represent Mp-data by shifting the fit
obtained for the Ms-data in Fig. 16 (Eq. (2) and
coefficients in Table 7) down to our experimental
Mp-value. These Mp-data are then represented by:
Mp(cNi) (b0 b1cNi b2c2Ni b3c3Ni

(3)

b4c4Ni b5c5Ni)K
The solid line that shows the Mp-behavior, which
we assume in the present study, is obtained using
the coefficients b0 to b5 listed in Table 7.
We now estimate Mp-values based on our volume fraction measurements (Tables 3 and 6). From
Eq. (1) we obtain corresponding cNi-values that we
then input in Eq. (3) to obtain Mp-estimates. The
results are listed in Tables 2 and 4 and are also
presented as small vertical bars (with horizontal
error bars reflecting the inherently large scatter in
TEM volume fraction measurements) in the corresponding DSC-charts shown in Figs. 2 and 11. This
approach does not attempt to fully account for all
DSC chart features. But the results confirm that

4268

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

Figure 13. TEM micrographs showing representative microstructures after stress assisted aging at 773 K. Top row: 3.6 ks
aging. Bottom row: 36 ks aging. Left column: 2 MPa. Right
column: 20 MPa. (a) 2 MPa, 3.6 ks (b) 20 MPa, 3.6 ks (c) 2
MPa, 36 ks (d) 20 MPa, 36 ks.

Figure 14. Low magnification TEM micrograph showing a


constant number density of Ni4Ti3 precipitates after stress
assisted aging (773 K/20 MPa/3.6 ks) everywhere in the microstructure. A two beam condition was adjusted for the grain on
the right side of the grain boundary which traverses the micrograph vertically (g: (1 10)). The area marked A on the left side
of the grain boundary is shown in Fig. 15 under a different
contrast condition.

Table 6
Microstructural data characterizing the Ni4Ti3-precipitate population after stress assisted aging for 3.6 and 36 ks at 773 K. Listed are
the microstructural parameters Dthe mean diameter of the main circle of the lenticular precipitates, tthe mean thickness of the
lenticular precipitates, lthe interparticle spacing and Vfthe volume fraction of precipitates
Material state (stress assisted aging at 773 K)

D [nm]

t [nm]

l [nm]

Vf [%]

3.6 ks, 2 MPa


3.6 ks, 20 MPa
36 ks, 2 MPa
36 ks, 20 MPa

11530
14550
335120
405110

7.73.0
8.92.9
327
306

14520
16020
30020
33020

1.71.0
2.21.1
6.73.5
7.12.1

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

Figure 15. TEM micrograph of the grain on the left side of the
grain boundary shown in Fig. 14 (material state: stress assisted
aging773 K/20 MPa/3.6 ks). The position A from Fig. 14 is
highlighted. Image contrast: g(101 ). It can be clearly seen
that the precipitates near the grain boundary (region A) differ
in contrast from the majority of precipitates which fill the
interior of the grain. Both types of precipitates have similar
number densities and size distributions. However, they belong
to different variant groups (see text). Most importantly, the precipitates which show the strongest contrast (precipitate plane
group (1 1 1)) are absent near the grain boundary (region A).

Mp temperatures are expected to increase as the


precipitate volume fractions increase. Also Apestimates can be made on the basis of the
expected Mp-data. First the shift T(cNi) between
the estimated Mp(cNi) and the temperature for the
single peak of the solution annealed material (on
can
be
calculated
cooling)
Mp(50.7)
(T(cNi) Mp(cNi)Mp(50.7)). Then Ap(cNi)
simply is obtained by adding T(cNi) to the
temperature obtained for the single peak of the
solution annealed material (on heating)

4269

Figure 16. Dependence of Ms-temperatures on the Ni-concentration of solution annealed NiTi alloys. The data were reported
in the review of Tang et al. [5] and stem from five different
sources [3135]. The figure also contains two experimental data
points (Ms and Mp from our solution annealed and water
quenched material) obtained in the present work. A polynomial
fit (dashed line) represents all Ms-data (literature data and our
result). A Mp data line (solid line) was obtained by shifting the
polynomial fit for the Ms-data down to the experimental Mp of
the solution annealed material of the present study.

Ap(50.7): Ap(cNi) Ap(50.7) T(cNi).


The
predicted Ap(cNi) temperatures are included in Figs.
3 and 12 (small vertical bars with corresponding
horizontal error bars) and in Tables 2 and 5. The
results show that we expect DSC peaks to shift to
higher temperatures. In cases where the transformations keep their single step character after the
thermomechanical treatments, there is reasonable
agreement between predicted and measured peak
positions; this is the case for the back transformation of the materials which were subjected to
stress assisted aging at 773 K for 3.6 and 36 ks in
the presences of stresses of 2 and 20 MPa, Fig. 12.
For all other cases the direction of the shift also is
predicted but other factors need to be considered
to fully rationalize all DSC chart features.

4270

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

Table 7
Coefficients a0 to a5 and b0 to b5 used in Eqs. (2) and (3) to obtain the fit of the Msdata (dashed line) and the assumed Mp
dependence (solid line) in Fig. 16
Fit coefficients for Eq. (2)
a0
a1
a2
a3
a4
a5

Fit coefficients for Eq. (3)


2027077615
205968221.1
8370845.646
170093.5974
1728.050544
7.022069029

4.3. DSC chart features


Before we now discuss the features of the DSC
charts for all solution annealed and aged material
states we briefly summarize four important microstructural results described in the previous sections: (1) no Ni4Ti3 precipitates are detected after
solution annealing, (2) short term stress free aging
results in heterogeneous microstructures with precipitates in grain boundary areas and remaining
precipitate free regions in the grain interiors, (3)
stress assisted aging results in a homogeneous distribution of precipitates throughout the microstructure in terms of numbers of precipitates per volume
(but there are regions near grain boundaries and in
the interior of grains which differ in terms of the
types of Ni4Ti3 variants present) and (4) higher
stresses during stress-assisted aging result in higher
growth rates of precipitates. We also keep in mind
that precipitation affects the Ni-content of the
supersaturated matrix that in turn affects martensite
transformation temperatures.
It is well known from the literature that DSC
chart features can evolve during aging. Morawiecz
and co-workers [36,37] and Huang and Liu [38]
investigated Ni-rich NiTi materials which were
predeformed (50. 6 at.-% Ni-cold rolling: [36,37],
50.85 at.-% Ni-wire drawing: [38]). These authors
[3638] subjected their materials to aging treatments at different temperatures and observed 1-23-1 [36,37] and 1-2-1 [38] step transformations in
their DSC charts on cooling and 1-2-1 [36,37] and
2-1 [38] step transformations on heating. Bataillard
[39] also found an evolution in DSC chart features
with aging temperature; but he did not attempt to
rationalize this evolution and instead focused his

b0
b1
b2
b3
b4
b5

2027077588.5
205968221.1
8370845.646
170093.5974
1728.050544
7.022069029

attention on an explanation for multiple step transformation behavior for one particular aging condition (solution annealing at 1173 K for 1.8 ks followed by aging at 793 K for 1.8 ks) [12,39].
Recently Khalil-Allafi et al. [13] (who only considered the transformation on cooling) showed that
DSC chart features evolve with aging time at a
constant aging temperature: for stress free aging at
673 and 723 K they observed a 2-3-2 transformation behavior. Khalil-Allafi et al. [13] pointed out
that coherency stress arguments [12,35,36] are not
sufficient to explain such type of DSC chart evolutions; one problem is, that the formation of Rphase represents a stress relaxation process, and
secondly, the coherency stress argument of Bataillard et al. [12] only rationalizes the three step transformation behavior but not the very clear evolution
of DSC charts with aging time. Khalil-Allafi et al.
[13] propose a new explanation for the 2-3-2-transformation behavior that is based on two basic
elements: (1) the composition inhomogeneity that
evolves during aging as Ni4Ti3 precipitates grow.
(2) The difference between nucleation barriers for
R-phase (small) and B19 (large). Based on these
two elements Khalil-Allafi et al. [13] were able to
rationalize the evolution of DSC charts; but they
did not provide microstructural evidence and they
did not consider the features of the back transformation on heating.
We first discuss the evolution of DSC charts
after stress free aging, Figs. 2 and 3. On cooling
Fig. 2 documents a 1-2-3-1-transformation
behavior while the back transformation in Fig. 3
shows a 1-2-2-1 characteristic. It can be clearly
seen in Fig. 2 that the last peaks on cooling after
3.6 and 36 ks aging occur at the same temperature

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

as the single peak of the solution annealed


material. And it is clear from Fig. 3 that the first
peaks on heating after 3.6 and 36 ks aging appear
in the same temperature range as the corresponding
single peak of the solution annealed material. Our
microstructural results suggest that the last peaks
on cooling after 3.6 and 36 ks of aging are due to
single step transformations from B2 to B19 in the
precipitate free regions of the microstructure; their
intensity decreases with aging time because the
precipitate free volume gradually decreases with
aging time as precipitation proceeds, Figs. 5 and
6. And the first peaks on heating (after 3.6 and
36 ks aging) correspond to the single step back
transformation from B19 to B2 in the precipitate
free regions.
In contrast, the first peak on cooling after 3.6 ks
aging and the first two peaks on cooling after 36
ks of aging is associated with microstructural
regions with precipitates. So far it was generally
assumed that two distinct DSC peaks on cooling
always result from the formation of R-phase
(B2R-phase, first distinct DSC peak) followed
by a transformation from R-phase to B19 (second
distinct peak). Only recently it was shown that
there are cases where already after the first peak
on cooling (in an overall two step transformation)
both R-phase and B19 coexist [17]. Therefore we
do not attribute the first peak on cooling after 3.6
ks of stress free aging to the formation of R-phase
alone; instead we expect that during this first
exothermal event a mixture of R-phase (major
constituent) and B19 (minor constituent) forms.
After 36 ks of stress free aging the two first peaks
on cooling represent the classical two step transformation in precipitate microstructures where precipitates favor the R-phase because of its smaller
transformation strain as compared to B19 (for a
detailed discussion see [8]). It seems reasonable to
assume that precipitate regions only give rise to
one endothermic peak on heating from B19: the
back transformation from B19 to B2 does not have
an energetical advantage from passing through an
intermediate R-phase stage; the transformation
strains for a transformation from B19 to R-phase
and from B19 to B2 are almost the same. Therefore we suggest that the two step back transformations on heating from B19 after 3.6 and 36 ks of

4271

aging are due to a first transformation peak associated with the back transformation in the precipitate free regions and a second transformation peak
in the remainder of the microstructure. Further
work is required to support this interpretation.
After 360 ks of aging there only appears one peak
on cooling and one peak on heating, Figs. 2 and
3. After 360 ks of aging Ni4Ti3 precipitates have
consumed all excess Ni in the matrix; they are very
big and widely spaced and no longer affect the
martensitic transformations on cooling. The shift
of the peaks after 360 ks of aging when comparing
to the solution-annealed material can be explained
with the decrease of matrix Nickel concentration.
It can be noted that the widths of the peaks after
100 h of aging are narrower than after solution
annealing; we suggest that this is due to a more
uniform distribution of Nickel after 360 ks of
aging.
Now we discuss the DSC charts that were
obtained after stress assisted aging, Figs. 11 and
12. After stress assisted aging under stresses of 2,
8 and 20 MPa microstructures are homogenous in
terms of precipitate density. Therefore there is no
reason for two step back transformations to occur
on heating from the B19 regime, Fig. 12. We
always observe one step back transformations on
heating from the B2 regime; we note, however, that
there is a small shoulder in the DSC chart of the
material that was aged for 1 h in the presence of
a stress of 20 MPa. It is interesting to see that this
material state produces the only three-step transformation that is observed on cooling from the B2regime, Fig. 11. As can also be seen in Fig. 11,
this three step type of transformation after 3.6 ks
aging is favored by high aging stresses: it can
hardly be seen for an aging stress of 2 MPa; it is
clearly more pronounced when aging in the presence of a stress of 8 MPa and it is fully developed
for the material which was subjected to 3.6 ks
aging at 773 K under 20 MPa. We attribute these
small effects to the microstructural features
reported in Fig. 15, where microstructures near
grain boundaries and in the grain interior differ
with respect to Ni4Ti3 variants. We expect that
micro structural regions with all types of possible
Ni4Ti3 variants are more effective in assisting martensite nucleation and the subsequent formation of

4272

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

self accommodating martensitic microstructures


than regions where only some of the eight possible
Ni4Ti3 variants are present; therefore we expect
that the second peak on cooling corresponds to the
formation of B19 in regions where all precipitate
variants are present (like the grain interiors in Fig.
15) while the third peak on cooling corresponds to
regions like the area A in Fig. 15; further work is
required to clarify this point. All other DSC charts
after stress assisted aging show a two-step transformation behavior on cooling from the B2 regime
that can be rationalized in terms of precipitates favoring the occurrence of R-phase [8].
Another significant difference between DSC
charts obtained for materials subjected to stress
free (Fig. 2) and stress assisted aging (Fig. 11) is
that the transformations on cooling start at significantly higher temperatures as highlighted by the
small arrows pointing to the start of transformation
on cooling after 36 ks of aging. Since the volume
fractions of precipitates were shown to be more or
less the same, this effect cannot be attributed to
the average Ni concentration. We suggest that it is
the small interparticle spacing (characterizing the
microstuctures after stress assisted aging) which
allows the coherency stress fields of the small precipitates to favor the occurrence of R-phase; this
is the reason for significantly higher Rs-temperatures after stress assisted aging (Fig. 11) as compared to stress free aging (Fig. 2). The competitive
growth of the higher number of precipitates after
stress assisted aging keeps the precipitate size
smaller than after stress free aging. Therefore after
360 ks of stress-assisted aging interparticle spacings are still not large enough to allow for an
unconstrained martensitic transformation of the
matrix. And if Rs were only governed by the local
thermodynamic equilibrium between precipitate
and matrix (as was implicitly assumed in the model
of Khalil-Allafi et al. [13]) then one would expect
the Rs-temperatures after stress free and stress
assisted to be the same; this is not the case when
we compare Figs. 2 and 11 and therefore precipitate sizes and spacings also have to be considered.
Further work is required to clarify this point.
Finally it is clear from what has been outlined
so far that the three step transformation after 36 ks
of stress free aging (Fig. 2) cannot be rationalized

on the basis of coherency stress arguments


[12,36,37]; neither can it be explained on the basis
of a varying Ni-concentration between growing
precipitates [13]. It is the heterogeneous precipitate
microstructure (regions with and without
precipitates) that is responsible for this behavior.

5. Summary and conclusions


The present paper studies precipitation processes
in a polycrystalline Ni-rich NiTi shape memory
alloy (nominal composition 50.7 at.-% Ni). Transmission electron microscopy (TEM) was used to
study the effects of different aging treatments
(stress free and stress assisted) on microstructures
(B2-matrix and Ni4Ti3-precipitates). The behavior
of different precipitate/matrix-systems during martensitic transformations was characterized using
differential scanning calorimetry (DSC). The following results were obtained:
1. Precipitation processes in solution annealed
NiTi alloys with 50.7 at.-% Ni during aging at
773 K are strongly affected by the presence of
external and internal stresses in the nucleation
stage. Stresses of the order of 2 MPa are sufficient to completely change the precipitation
behavior in solution annealed Ni-rich NiTi
alloys during aging from heterogeneous to
homogeneous.
2. The present study shows how microstructures
evolve during stress free and stress assisted
aging in terms of precipitate size, volume fraction, interparticle spacing and distribution of
precipitates in the microstructure. These parameters were found to evolve during aging treatments. Accordingly, the related characteristics
of DSC charts during subsequent martensitic
transformations depend on aging time.
3. Short term and intermediate term stress free
aging results in heterogeneous microstructures
where precipitates are mainly found within 2
m wide regions around grain boundaries while
grain interiors exhibit precipitate-free zones.
These microstructures show three step transformations on cooling (interpretation: 1. distinct
peakformation of R-phase in the regions con-

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

4.

5.

6.

7.

8.

taining precipitates, 2. distinct peakformation


of B19 in the regions containing precipitates,
3. distinct peaktransformation of B2 to B19
in precipitate free regions). The three step transformations observed in the present work cannot
be rationalized on the basis of coherency stress
fields [12] and they cannot be explained by
varying Ni-concentrations between precipitates
during aging [13,16]. It is the heterogeneous
microstructure (with regions with and without
precipitates) that is responsible for this behavior.
On heating, microstructures which form during
short term stress free aging result in two step
transformations (interpretation: 1. distinct
peakback transformation of precipitate free
regions into B2, 2. distinct peakback transformation of the regions containing precipitates
into B2).
As precipitates grow to large sizes during long
term stress free aging (773 K, 360 ks) the interparticle spacings become so large that the martensitic transformation of the matrix is no longer
affected by the presence of precipitates. Large
precipitates may moreover well loose coherency
and therefore loose their potential to affect the
nucleation of martensitic phases.
It was found that increasing the aging time from
3.6 to 36 ks has a much larger effect on precipitate sizes than increasing the level of applied
stress from 2 to 20 MPa. The level of superimposed stresses does not affect nucleation rate
(which is high for all stress levels) but there is
a small increase of coarsening rates with
increasing superimposed stress.
After 20 MPa stress assisted aging for 3.6 ks,
microstructures are homogenous in terms of
number density and volume fraction of precipitates. However, Ni4Ti3 variants differ between
regions in the grain interior (all variants) and
near grain boundaries (only some variants). This
type of precipitate variant heterogeneity can also
give rise to a three-step transformation on cooling from the B2-regime.
From precipitate volume fractions measured
using TEM, the Ni-depletion of the supersaturated B2 matrix can be calculated. It is then
possible to rationalize the increase of phase

4273

transition temperatures observed in the present


study on the basis of data reported on the influence of Ni on Ms-temperature in the literature.

Acknowledgements
The authors would like to acknowledge funding
by the Deutsche Forschungsgemeinschaft (DFG) in
the framework of the shape memory center SFB
459 (Formgeda chtnistechnik). AD acknowledges
traveling support from the Grant Agency of the
Czech Republic (contract no.: 106/99/1172).

References
[1] Hornbogen E. In: Bunk WGJ, editor. Advanced structural
and functional materials. Heidelberg: Springer-Verlag;
1991. p. 13363.
[2] Saburi T. In: Otsuka K, Wayman CM, editors. Shape
memory materials. Cambridge: Cambridge Unversity
Press; 1998. p. 4996.
[3] Van Humbeeck J. Mat. Sci. Eng 1999;A273-275:134.
[4] Duerig T, Pelton A, Stoeckel D. Mat. Sci. Eng
1999;A273-275:149.
[5] Tang W, Sundmann B, Sandstro m R, Quiu C. Acta
Mater 1999;47:3457.
[6] Tadaki T, Nakata Y, Shimizu K, Otsuka K. Trans. JIM
1986;27:731.
[7] Li DY, Chen LQ. Acta Mater 1997;45:471.
[8] Ren X, Miura N, Zhang J, Otsuka K, Tanaka K, Koiwa M,
Suzuki S, Chumlyakov YI. Mat. Sci. Eng 2001;A312:196.
[9] Gall K, Sehitoglu H, Chumlyakov Yu I, Kireeva IV, Maier
HJ. Trans. ASME 1999;121:28.
[10] Hornbogen E, Mertinger V, Wurzel D. Scripta Mater
2001;44:171.
[11] Khalil-Allafi J, Dlouhy A, Neuking K, Eggeler G. J. Phys.
IV France 2001;11:Pr8529.
[12] Bataillard L, Bidaux J-E, Gotthardt R. Phil Mag
1998;78:327.
[13] Khalil-Allafi J, Ren X, Eggeler G. Acta Mater
2002;50:793.
[14] Filip P, Mazanec K. Scripta Mater 2001;45:701.
[15] Skrotzki B, Rudolf T, Eggeler G. Z. Metallkde
1999;90:393.
[16] Khalil-Allafi J., Dr.-Ing. Thesis, Ruhr-Universita t
Bochum, to be published 2002.
[17] Sitepu H, Schmahl WW, Khalil-Allafi J, Eggeler G,
Dlouhy A, To bbens DM, Tovar M. Scripta Mater
2002;46:543.
[18] Dlouhy A, Pesicka J. Czech J. Phys 1990;B40:539.
[19] Dlouhy A., Khalil-Allafi J., Eggeler G. To be published
in Practical Metallography.

4274

[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]

J. Khalil-Allafi et al. / Acta Materialia 50 (2002) 42554274

Dlouhy A, Eggeler G. Prakt. Met 1996;33:1.


Wagner CZ. Elektrochem. 1961;65:581.
Lifshitz IM, Slyozov VV. I. Phys. Chem. Solids 1961;19:35.
Perovic V, Purdy GR, Brown LM. Acta Metall 1981;29:889.
Nishida M, Honma T. Scripta Metall 1984;18:1293.
Nishida M, Honma T. Scripta Metall 1984;18:1299.
Li DY, Chen LQ. Acta Mater 1997;45:2435.
Li DY, Chen LQ. Acta Mater 1998;46:639.
Li DY. Phil. Mag 1999;A79:2603.
Porter DA, Easterling KE. London: Chapman and Hall,
1997 reprinted 2nd ed.
[30] Hirth JP, Lothe J. Theory of dislocations. Malabar: Krieger Publishing Company, 1992.
[31] Hanlon JE, Butler SR, Wasilewski RJ. Trans. Am. Inst.
Min. Engrs 1967;239:1323.

[32] Wasilewski RJ, Butler SR, Hanlon JE. Metal Sci. J.


1967;1:104.
[33] Kornilov II, Kachur VY, Belousov OK. Fiz. Metall
1971;32:420.
[34] Wasilewski RJ, Butler SR, Hanlon JE, Worden D. Metall.
Trans 1971;2:229.
[35] Saburi T, Tatsumi T, Nemo S. J. Physique
1982;43:C4.
[36] Morawiec H, Stroz D, Chrobak D. J. Physique III
1995;5:C2205.
[37] Morawiec H, Stroz D, Goryczka T, Chrobak D. Scripta
Mater 1996;35:485.
[38] Huang X, Liu Y. Scripta Mater 2001;45:153.
[39] Bataillard L. Ecole Polytechnique Fe de rale de Lausanne,
The`se No. 1518, 1996.

Das könnte Ihnen auch gefallen