Annealing effects on the crystallization and shape
memory effect of Ti
50
Ni
25
Cu
25
melt-spun ribbons
S.H. Chang
a, S.K. Wu
a,*
, H. Kimura
ba
Department of Materials Science and Engineering, National Taiwan University, 1 Roosevelt Road, Sec. 4, Taipei 106, Taiwan
bInstitute for Materials Research, Tohoku University, Sendai 980-8577, Japan
Received 5 September 2005; accepted 5 May 2006 Available online 10 July 2006
Abstract
As-spun Ti50Ni25Cu25ribbon is fully amorphous with a lower wavenumberQpthan the amorphous TieNi alloys owing to its high Cu content.
Both crystallization activation energyEaand onset temperatureTxfor Ti50Ni25Cu25ribbon are lower than those for Ti50Ni50ribbon, indicating
that the former has lower thermal stability. When Ti50Ni25Cu25ribbon is annealed at 500C for 3 min, the initial as-crystallized grains contain
a low Cu content and perform a prominent shape memory effect. Through prolonging the annealing time, more grains are crystallized in the ribbon but it becomes more fragile and its recoverable strain decreases. This is due to the increasing Cu content in the crystallized grains.
Crys-tallized Ti50Ni25Cu25ribbon can exhibit a good shape memory effect only under appropriate annealing conditions.
Ó 2006 Elsevier Ltd. All rights reserved.
Keywords: A. Ternary alloy systems; B. Martensitic transformations; B. Shape-memory effects; C. Rapid solidification processing
1. Introduction
TiNi-based alloys are known as the most important shape memory alloys (SMAs) with good shape memory effect (SME), superelasticity (PE) and damping capacity (DC) [1]. Substituting Cu for Ni in binary TieNi SMA has been known to lower the transformation hysteresis, the superelasticity hysteresis, and the flow stress level in the martensite state. The narrow hysteresis of TieNieCu ternary SMAs has the potential for applications which require short response times during thermal cycling [2]. Adding Cu into TieNi binary SMAs also has been reported to reduce the sensitivity of the martensitic transformation start temperature, Ms, to composi-tional changes and to prevent Ti3Ni4 precipitation [2e4].
Moreover, the transformation sequence in TieNieCu SMAs depends on the Cu content. For Cu contents below 10 at.%, the monoclinic B190 martensite is formed on cooling from
the cubic B2 austenite. When the Cu content is approximately in between 10 and 15 at.%, the two-stage transformation behavior of B2 4 B19 4 B190 is obtained where B19 is an orthorhombic martensite[5]. In the case of a higher Cu con-tent, the second step is inhibited and only B2 4 B19 appears
[2,6]. Unfortunately, it was found that Cu additions exceeding
10 at.% embrittle the alloy and seriously reduce the workabil-ity and shape recovery strain[2,7]. This restrains the applica-tion of high Cu content TieNieCu SMAs.
In recent years, melt-spinning techniques have been utilized to fabricate high Cu content TieNieCu ternary SMAs in order to avoid the aforementioned restriction of workability. Ti
50-Ni25Cu25 (at.%) ribbon has been widely studied because of
its small transformation hysteresis, large transformation strain, and one-stage B2 4 B19 transformation compared to con-ventionally fabricated TieNieCu wire [8]. Fabrication of TieNieCu ternary SMA ribbons by means of the melt-spinning technique has been shown to be suitable for producing alloys with controllable amorphous or crystalline structures. Never-theless, the as-spun TieNieCu ribbon is amorphous and shows no SME if the ribbon is fabricated using a high cooling rate in
* Corresponding author. Tel.:þ886 2 2363 7846; fax: þ886 2 2363 4562. E-mail address:[email protected](S.K. Wu).
0966-9795/$ - see front matterÓ 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.intermet.2006.05.014
the melt-spinning procedure. Therefore, a proper thermal an-nealing procedure is required to crystallize the as-spun amor-phous ribbon. Previous studies on Ti50Ni25Cu25 ribbon
focused on the microstructure of the partially or fully crystal-lized alloys, the precipitates generated after thermal treatment, and their effects on the martensitic transformation [8e13]. However, only little works referred to the effect of annealing on the SME and on the mechanical properties of the crystal-lized Ti50Ni25Cu25ribbon[14,15]. Liu[14]stated that a Ti
50-Ni25Cu25 melt-spun ribbon annealed at 500C for 15 min
exhibits a well-defined SME and a good superelastic shape re-covery strain with low hysteresis. No other annealing condition was presented because the ribbon annealed at a temperature higher than 500C shows poorer mechanical properties. Cheng and Xie[15]reported the influence of the annealing tempera-ture on the shape memory properties of a Ti50Ni25Cu25
melt-spun ribbon. They discovered that the recovery strain significantly decreases with increasing annealing temperature and proposed that the decreasing recovery strain may be due to the larger grain size and the more preferentially orientated precipitates which form at higher annealing temperature.
In view of the applications of Ti50Ni25Cu25 melt-spun
rib-bons, it is important to control appropriate annealing condi-tions to crystallize the as-spun amorphous Ti50Ni25Cu25
ribbons without deteriorating their SME properties. However, up to now, the annealing effects on the crystallization behavior and SME properties of Ti50Ni25Cu25 melt-spun ribbon have
not been elucidated in detail. In this study, the amorphous-crystalline characteristics of Ti50Ni25Cu25 melt-spun ribbons
annealed at 500C for different time intervals are investigated by means of X-ray diffraction (XRD), differential scanning calorimetry (DSC), dynamic mechanical analysis (DMA) and scanning electron microscopy (SEM) with energy disper-sive spectrometry (EDS). The crystallization behavior and shape memory property of annealed Ti50Ni25Cu25 melt-spun
ribbons are also discussed. 2. Experimental procedures
A Ti50Ni25Cu25ingot was prepared by conventional vacuum
arc remelting (VAR). The ingot was re-melted six times in an argon atmosphere for homogenization. The as-melted Ti
50-Ni25Cu25ingot was cut into an appropriate size and then
in-duction-melted in an argon atmosphere in a quartz crucible at 1150C, and subsequently ejected with pressurized argon gas onto a copper roller with a surface velocity of 42 m/s. The final ribbons were about 20 mm in thickness and 1.1 mm in width. Thereafter, the as-spun ribbons were cut into test specimens, sealed in evacuated quartz tubes and annealed at 500C in a salt bath for different time intervals.
The crystallographic features of the ribbons during crystal-lization annealing were determined with a Philips PW 1830 XRD instrument using Cu Ka radiation. The crystallization and transformation temperatures of each specimen were deter-mined by DSC using a TA Q10 DSC equipment. Specimen weights for DSC investigations were 5e6 mg and the heating and cooling rates were 10C/min. The transformation strain
during thermal cycling was measured by tensile testing under constant stress (90 MPa) using a TA 2980 DMA equipment with a tension clamp. The gauge length of each specimen was set as 10 mm and the heating and cooling rates were 3C/min. Microstructure observations and composition analy-sis for the ribbons annealed at 500C for different time inter-vals were performed with a Philips XL 30 SEM instrument equipped with an EDS unit.
3. Experimental results
3.1. Crystallization of Ti50Ni25Cu25melt-spun
amorphous ribbon
Fig. 1 displays the XRD diffraction pattern of as-spun
Ti50Ni25Cu25 ribbon. As illustrated in Fig. 1, only a broad
amorphous peak appears near 2q¼ 41.8. It reveals that the
as-spun Ti50Ni25Cu25 ribbon is completely amorphous. The
position of this broad amorphous peak can be represented by the wavenumber QP¼ 4psin q/l which is inversely
propor-tional to the mean nearest-neighbor distance of the local-ordering clusters of amorphous alloys [16]. The QP value
calculated fromFig. 1 is plotted in Fig. 2 and is compared with the results from the rapid quenching melt-spun TieNi and TieNieCu SMA ribbons[17], mechanical alloyed pow-ders[16]and sputtered thin films[18,19].
Fig. 3displays the DSC curves for Ti50Ni25Cu25melt-spun
ribbons with different heating rates to determine the crystalli-zation activation energy. The peak temperature,Tp, and the
on-set temperature, Tx, of crystallization are also denoted in
Fig. 3. There is only one-stage of crystallization obtained in
Ti50Ni25Cu25 melt-spun amorphous ribbon. The two-stage
crystallization behavior of rapid quenched Ti70Ni30 and
Ti60Ni40 ribbons [20] is not observed. As clearly shown in
Fig. 3, the crystallization temperatures,TpandTx, rise with
in-creasing heating rate. According to the results obtained in
Fig. 3, the crystallization activation energy,Ea, can be
deter-mined from Kissinger’s relation[21]:
30 40 50 60 70 80 90
Intensity (a.u.)
2θ (deg.)
41.8º
lna=Tp2¼ C Ea=RTp ð1Þ
whereC is a constant, a is the heating rate, Eais the
crystal-lization activation energy, and R is the gas constant. Fig. 4
plots lna=T2 p
versus 1/Tpbased on the DSC results shown
inFig. 3for Ti50Ni25Cu25 melt-spun ribbon. The
crystalliza-tion activacrystalliza-tion energy for Ti50Ni25Cu25 ribbon is calculated
as 341 kJ/mol. As shown inFig. 3,Txfor Ti50Ni25Cu25
melt-spun ribbon with the cooling rate of 30C/min is detected as 477C. This crystallization temperature, Tx, is plotted
versus the Ti content of the TixNi1x alloys in Fig. 5 with
the results obtained from other studies[16e19].
3.2. Annealing effect in Ti50Ni25Cu25melt-spun ribbon
3.2.1. XRD measurements
Fig. 6(a) shows XRD results of Ti50Ni25Cu25 melt-spun
ribbon annealed at 500C for different time intervals. As
shown in Fig. 6(a), the as-spun ribbon and the ribbon annealed at 500C for 60 s are completely amorphous. When the ribbon is annealed at 500C for 90 s, a small (200)B2 peak appears at around 2q¼ 61. After annealing
at 500C for 100 s, the intensity of the (200)B2 peak
in-creases. When the ribbon is annealed for 110 s, besides the (200)B2 peak, the other two peaks (110)B2 and (211)B2
develop at 2q¼ 42 and 76, respectively. When the
anneal-ing time is extended to 3 min, the (110)B2 peak becomes
dominant and has a greater intensity than the others. After annealing for 60 min, the diffraction pattern is similar to that of the ribbon annealed at 500C for 3 min. However, now an extra peak (002)B19 of B19 martensite appears at
2q¼ 41 because a partial B2 / B19 transformation occurs
at room temperature. Fig. 6(b) plots the evolution of the lat-tice parameter of B2 parent phase as calculated from the XRD results of (200)B2 and (110)B2 shown in Fig. 6(a).
As indicated in Fig. 6(b), the lattice parameter of the B2
27.5 28.0 28.5 29.0 29.5 30.0 30.5 31.0 31.5 32.0 32.5 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Rapid quenched melt-spun ribbon by Buschow (Ref. 17)
Mechanically alloyed powders by Eckert et al. (Ref. 16)
Linear Fit of Buschow and Eckert et al. results Ti45.60Ni54.40 amorphous films (Ref. 18) Ti49.93Ni50.07 amorphous films (Ref. 19) Ti49.96Ni40.09Cu9.95 amorphous films (Ref. 19) Ti50Ni25Cu25 melt-spun ribbon (this study)
Wavenumber Q
P
(nm
-1)
X in TiXNi1-X (in at.%)
Fig. 2. WavenumberQpof amorphous Ti50Ni25Cu25melt-spun ribbon versus
Ti content in comparison with the results of rapid quenched samples[17], me-chanically alloyed powders[16]and sputtered films[18,19].
466.8°C 464.4°C 475.9°C 472.9°C 481.3°C 477.4°C 485.9°C 481.9°C 489.9°C 485.4°C 459.6°C 457.5°C -10 -5 0 5 10 Heat Flow (W/g) 440 460 480 500 520 Temperature (ºC) 50 °C/min 40 °C/min 30 °C/min 20 °C/min 10 °C/min 5 °C/min heating Tx Tp Exo Up Universal V3.5B TA Instruments
Fig. 3. DSC curves with different heating rates for Ti50Ni25Cu25melt-spun
ribbon. 0.00130 0.00132 0.00134 0.00136 0.00138 0.00140 -12 -11 -10 -9 -8 ln( α /Tp 2) 1/Tp (1/K) Ti50Ni25Cu25 ribbon Ea=341 kJ/mol
Fig. 4. Kissinger’s plots for the data obtained from DSC curves inFig. 3.
0.4 0.6 0.3 0.5 0.7 0.8 400 450 500 550 600 650 700 750 crystallization temperature T X (°C) X in TiXNi1-X (in at. %)
Melt-spun Ribbons by Buschow (α=50°C/min) (Ref.17 )
Mechanically alloyed powders by Eckert et al. (α=50°C/min) (Ref. 16)
Ti45.60Ni54.40 films (α=30 °C/min) (Ref. 18) Ti49.93Ni50.07 films (α=30 °C/min) (Ref. 19) Ti49.96Ni40.09Cu9.95 films (α=30 °C/min) (Ref. 19) Ti50Ni50 melt-spun ribbons (α=30 °C/min) (Ref. 22) Ti50Ni25Cu25 melt-spun ribbons (α=30 °C/min) (this study)
Fig. 5. Crystallization temperature,Tx, of Ti50Ni25Cu25melt-spun ribbon
ver-sus Ti content in comparison with the results of rapid quenched samples[17], mechanically alloyed powders[16]and sputtered films[18,19].
phase of Ti50Ni25Cu25 ribbon gradually decreases from
0.305 nm to about 0.304 nm with an increased annealing time.
3.2.2. DSC results
Fig. 7 shows the DSC results of Ti50Ni25Cu25 melt-spun
amorphous ribbon annealed at 500C from 3 min to 3 h. The measured DSC cooling and heating curves for each spec-imen are approximately the same, and thus only the cooling curves are discussed. As shown in Fig. 7, the specimen for each annealing time exhibits a single-stage B2 / B19 trans-formation in cooling. The Ms temperatures of B2 / B19 mar-tensitic transformation and their transformation enthalpies DH shown inFig. 7are plotted in Fig. 8(a) and (b), respectively. Upon increasing the annealing time, as illustrated in Figs. 7
and 8, the transformation temperatures of the crystallized
Ti50Ni25Cu25 melt-spun ribbon drastically shift to a higher
temperatures and the corresponding DH value gradually increases with the prolonged annealing time. The DH value
80 100 120 140 160 180 5000 10000 0.300 0.302 0.304 0.306 0.308 0.310 Lattice Parameter (nm) Time (sec) (200)B2 (110)B2
(b)
40 50 60 70 80 90 Intensity (a.u.) 2θ (deg.)(a)
as-spun60 sec 90 sec 100 sec 110 sec 3 min 60 min (211) B2 (200) B2 (110) B2 (002) B19Fig. 6. (a) XRD patterns for Ti50Ni25Cu25melt-spun ribbon annealed at 500C
for different time intervals, and (b) change of the lattice parameter calculated form (200)B2and (110)B2shown in (a) as a function of the annealing time.
-24.8°C -10.0°C 6.7J/g -0.4°C 6.9J/g 7.8°C 7.2J/g 7.4J/g 16.6°C -0.6 -80 -60 -40 -20 0 20 40 60 80 -0.4 -0.2 0.2 0.4 0.6 0.0 0.8 3 min 5 min 15 min 60 min 180 min cooling Ms 6.5J/g ΔH Heat Flow (W/g) Temperature (°C) Exo Up Universal V3.5B TA Instruments
Fig. 7. DSC curves for Ti50Ni25Cu25melt-spun ribbon annealed at 500C for
different time intervals.
0 20 40 60 80 100 120 140 160 180 200 -30 -20 -10 0 10 20 Ms ( °C)
Annealing Time (min)
Ti50Ni25Cu25 500 °C annealing 6.0 6.5 7.0 7.5 8.0 ΔH (J/g) Ti50Ni25Cu25 500 °C annealing
(b)
(a)
0 20 40 60 80 100 120 140 160 180 200Annealing Time (min)
Fig. 8. Change of (a) the Ms transformation temperature and (b) the transfor-mation enthalpy as a function of the annealing time. The data are taken from DSC curves shown inFig. 7.
eventually reaches about 7.5 J/g, which is lower than the normal TieNieCu bulk alloys, say, 17.1 J/g[5].
3.2.3. Tensile test results
Fig. 9(a) shows the selected strainetemperature curves
un-der a constant stress (90 MPa) for the Ti50Ni25Cu25melt-spun
ribbon annealed at 500C for different time intervals. In order to compare the shape recovery effect of each specimen, only a low stress (90 MPa) was applied to ensure that no residual deformation would be retained after the test.Fig. 9(b) plots the evolution of the measured recoverable strain from
Fig. 9(a). As illustrated inFig. 9(a) and (b), each Ti50Ni25Cu25
melt-spun ribbon exhibits a well-defined shape recovery effect when the specimen is annealed for more than 3 min. Among them, the specimen annealed at 500C for 3 min has the max-imum recoverable strain of about 2%, as indicated by the dou-ble arrows inFig. 9(a). When the annealing time is prolonged, however, the recoverable strain gradually decreases and the specimen becomes more brittle. When the Ti50Ni25Cu25
rib-bon is annealed at 500C for 3 h, the specimen becomes very fragile and fractures during the tensile test.
4. Discussion
4.1. Amorphous characteristics of Ti50Ni25Cu25
melt-spun ribbon
As shown in Fig. 2, the QP values of sputtered TixNi1x
amorphous films are very close to the line fitted for those of melt-spun TixNi1x amorphous ribbons and mechanically
alloyed TixNi1x powders. However, the QP value of Ti
50-Ni25Cu25melt-spun ribbons calculated in this study, and also
that of sputtered Ti49.96Ni40.09Cu9.95films[19], is lower than
those of the amorphous TixNi1xSMAs deviating from the
fit-ted line. The wavenumber QP is inversely proportional to the
mean nearest-neighbor distance of the local-ordering clusters of the amorphous alloys. This indicates that the addition of Cu to replace Ni in TieNi binary SMAs changes the short range ordering of the amorphous phase, though the size of Cu atom (rCu¼ 0.1413 nm) is close to that of Ni atom
(rNi¼ 0.1377 nm). In other words, the atomic arrangement
of the local cluster both in the Ti50Ni25Cu25melt-spun ribbon
and in the Ti49.96Ni40.09Cu9.95 sputtered film is looser than
that in TixNi1xSMAs. Therefore, the average bonding strength
for the local cluster in TieNieCu ribbon/film is weaker than that in TixNi1x. Furthermore, as illustrated in Fig. 2, the
Ti50Ni25Cu25 ribbon exhibits a lower QP value than the
Ti49.96Ni40.09Cu9.95 film. This feature indicates that a Ti
50-Ni25Cu25 ribbon with a higher Cu content than a Ti
49.96-Ni40.09Cu9.95film can exhibit looser bonding in the amorphous
phase.
As shown inFig. 4, theEavalue for Ti50Ni25Cu25melt-spun
ribbon is calculated as 341 kJ/mol. The Ea andTx values for
Ti50Ni50 melt-spun ribbon are also measured in this study as
430 kJ/mol and 522C (for 30C/min heating rate), respec-tively[22].Fig. 5indicates that Txfor a Ti50Ni25Cu25 ribbon
is about 40C lower than that for a Ti50Ni50ribbon. Both Tx
and Ea are important indicators of the thermal stability of
amorphous material, i.e., the lower values ofTxandEa
repre-sent the lower stability of the amorphous phase[23]. As shown
inFig. 5,Txof the mechanically alloyed powder is the lowest
among those of sputtered thin films and melt-spun ribbons. The lowest level of amorphous stability of the mechanically alloyed powder may have been due to hydrogen contamination
[16] and the large number of defects introduced during the mechanical alloying process.
As is also apparent fromFigs. 4 and 5, all TieNieCu spec-imens have lowerTxandEavalues than TieNi binary SMAs
in both sputtered films and melt-spun ribbons. This feature indicates that the replacement of Ni by Cu indeed lowers the thermal stability of the amorphous phase and assists in crystal-lization occurring. Chen and Park [24]demonstrated that the stability of an amorphous alloy is dominated by the strength of the interactions between the constituent atoms. The relative strength of the interactions between different constituent atoms can be determined by comparing their mixing en-thalpies. Different constituents have a stronger tendency to form an amorphous alloy when they have a larger negative mixing enthalpy [25]. Previous studies have revealed that
0 10 20 30 40 50 60 1.0 1.2 1.4 1.6 1.8 2.0 2.2 Recoverable Strain (%) -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Strain (%)
Annealing Time (min)
(b)
(a)
-60 40 -20 0 20 40 60 80
Temperature (°C) Universal V3.7A TA Instruments ––––––– 500ºC 3 min · · · · 500ºC 5 min –– –– – 500ºC 15 min ––––– 500ºC 60 min cooling heating
Fig. 9. (a) Thermally induced shape recovery curve under a constant stress (90 MPa) for Ti50Ni25Cu25melt-spun ribbon annealed at 500C for different
time intervals and (b) change of the recoverable strain value as a function of the annealing time.
TieNi has a larger negative enthalpy than TieCu, and NieCu even has a positive enthalpy of mixing[26,27]. Therefore, the interaction strength between Ti and Ni atoms is stronger than that of the interaction strength between Ti and Cu, and the interaction between Ni and Cu is the weakest. As a result, the amorphous alloy becomes less stable when Cu atoms are partially substituted for the Ni in TieNi binary SMAs because the average bonding strength between the atoms is decreased. Consequently, the TieNieCu ribbon has lower Tx and Ea
values than those of Ti50Ni50 ribbon because the latter has
a stronger tendency to form an amorphous structure.
4.2. Crystallization of annealed Ti50Ni25Cu25
melt-spun ribbon
4.2.1. Annealing effect on the crystallization
As demonstrated inFig. 1, the as-spun Ti50Ni25Cu25ribbon
is completely amorphous.Fig. 6 reveals that the crystallized degree of the Ti50Ni25Cu25ribbon can be controlled by the
an-nealing time. As can be seen fromFigs. 7e9, DSC and tensile test results also show that the annealing condition significantly influences the shape memory behavior of the crystallized ribbon. Both the transformation temperature and enthalpy detected in the DSC test increase with a prolonged annealing time due to an increase in the crystallization volume. However, as illustrated inFig. 9, with an increase in the annealing time, the crystallized ribbon loses recoverable strain and simulta-neously becomes brittle. In order to disclose the characteristics
of these annealing effects, SEM observation with EDS analy-sis for the crystallized Ti50Ni25Cu25 ribbon was conducted.
Fig. 10(a)e(d) shows SEM micrographs of Ti50Ni25Cu25
ribbon annealed at 500C for 1 min, 3 min, 15 min and 3 h, respectively. As shown inFig. 10(a), there are some spherical particles with a diameter of about 10 mm embedded in the amorphous matrix. The specimen annealed at 500C for 1 min does not show any crystallized peak for the XRD measurement inFig. 6. This means that these spherical parti-cles are not crystallized grains but are rather small bubbles produced during the melt-spinning technique retained in the amorphous matrix following the rapid solidification process
[12]. The boundaries between these spherical particles and the amorphous matrix can act as nucleation sites of the crys-tallized grains during annealing. When the specimen is annealed at 500C for 3 min, as illustrated in Fig. 10(b), some small crystallized grains with diameters of 1e3 mm ap-pear in these spherical particles. As shown inFig. 10(c) and (d), there are more crystallized grains that developed within a prolonged annealing time interval and, therefore, the area of the amorphous matrix is reduced. Table 1 lists the EDS analyses for the chemical composition of points 1e8 denoted
inFig. 10. As depicted in point 1 ofTable 1, the Cu content in
the small bubble is 22.9 at.% which is slightly less than 25 at.% when the ribbon is annealed at 500C for 1 min. After annealing for 3 min, however, the Cu percent of the as-crystal-lized grains drops to only about 16.6 at.%, as shown in point 3
of Table 1. This phenomenon reveals that the as-crystallized
grains contain a much lower Cu content than the amorphous
• 5 • 6 • 7 8 • • 1 • 2 • 4 • 3
(a)
(c)
(b)
(d)
matrix. Through prolonging the annealing time to 15 min, the Cu content in the crystallized grains raises to about 23.4 at.%, as indicated in point 5 ofTable 1. After annealing at 500C for 3 h, as shown in points 7 and 8 ofTable 1, the composi-tions of the crystallized grains and the amorphous matrix are almost the same. This reveals that sufficient annealing can re-sult in a homogeneously crystallized Ti50Ni25Cu25ribbon.
FromFig. 6, the only XRD peak obtained in the
as-crystal-lized ribbon annealed at 500C for 90 s is (200)B2 and the
lattice parameter of B2 parent phase is 0.305 nm. This charac-teristic indicates that the as-crystallized grains have a fiber textureh100i along the normal direction of the ribbon [28]. The highh100i texture of the crystallized grains perpendicular to the ribbon surface is due to the heat flow associated with ribbon processing which is normal to the spin wheel during the rapid solidification process. This result is similar to that of the orientation distribution function (ODF) obtained in the Ti50Ni25Cu25melt-spun ribbon[28,29]. Through
prolong-ing the annealprolong-ing time, however, the intensity of the (110)B2
peak which is widely obtained in most TieNieCu bulk alloys
[5]drastically increases and becomes the dominant peak. After sufficient annealing, i.e., 500C for 3 h, the XRD pattern and the lattice parameter of the B2 parent phase (a¼ 0.304 nm) are similar to those reported by Santamarta et al. [30]. Nam et al.[31]pointed out that the Cu content in bulk TieNieCu SMAs does indeed affect the lattice parameter of the parent phase and therefore changes their transformation elongation. Accordingly, it is reasonable to propose that the alteration of the XRD pattern and the decrease in the B2 lattice parameter shown inFig. 6are due to the variation of Cu content in crys-tallized grains caused by annealing.
In Fig. 10, the diameter of crystallized grains does not
strongly increase as the annealing time increases. At the same time, the increase in the crystallized volume is not due to the grain growth but is associated with the formation of new grains in the amorphous matrix. This implies that the required energy for new grain nucleation in the amorphous matrix is lower than that for the grain growth of the existing crystallized grains. The supercooling amorphous state pro-vides a high driving force for both nucleation and subsequent grain growth. However, the annealing temperature at 500C conducted in this study may be insufficient to promote sticky atom diffusion. This feature leads to a slow diffusion rate and limits grain growth.
4.2.2. Annealing effect on shape memory properties
When Ti50Ni25Cu25ribbon is mildly annealed, as shown in
Fig. 10(a), no obvious phase transformation and well-defined
shape recovery can be obtained. After annealing at 500C for 3 min, as shown in Fig. 10(b), small crystallized grains ap-pear. At this time, the recoverable strains reach a maximum, as indicated in Fig. 9. Through prolonging the annealing time, more crystallized grains form, as illustrated in
Fig. 10(c) and (d), however, the amount of recoverable strain
decreases, as shown in Fig. 9. We propose that the degener-ated shape recovery effect is mainly due to the composition alteration in the crystallized grains of annealed Ti50Ni25Cu25
ribbon. The as-crystallized grains contain a lower Cu content and perform a prominent shape recovery effect. Through pro-longing the annealing time, on the other hand, the increased Cu content in the crystallized grains results in brittle charac-teristics and a low recoverable strain. This feature is similar to the results obtained in bulk TieNieCu alloys[2,31]in which the variation in Cu content influences the lattice parameters of the phases and changes the recoverable strain during B2 4 B19 transformation. Therefore, the crystallized grains in the longer annealed Ti50Ni25Cu25 ribbon contain a higher
copper content which embrittles the annealed ribbon and reduces the shape recovery effect. Consequently, a Ti50Ni25Cu25
ribbon exhibits a good shape memory effect only under appro-priate annealing conditions, such as annealing at 500C for 3 min in this study.
5. Conclusions
The amorphous Ti50Ni25Cu25melt-spun ribbons which are
partially crystallized by 500C annealing have been investi-gated by means of XRD, DSC, DMA and SEM tests. As shown by the XRD results, the as-spun Ti50Ni25Cu25 ribbon
is fully amorphous with the lowest wavenumber Qp among
the amorphous TieNi and TieNieCu alloys fabricated by me-chanical alloying, rapid quenching and sputter deposition. This indicates that the high Cu content of Ti50Ni25Cu25ribbon can
loosen the atomic bonding in the amorphous phase. The Ea
value of Ti50Ni25Cu25 ribbon is 341 kJ/mol calculated from
the DSC data using Kissinger’s method. The Tx for Ti
50-Ni25Cu25ribbons is about 40C lower than that for Ti50Ni50
ribbons. These characteristics indicate that the as-spun Ti
50-Ni25Cu25 ribbon has a lower thermal stability than Ti50Ni50
ribbon from the viewpoint of the mixing enthalpy. Crystalliza-tion degree in a Ti50Ni25Cu25ribbon can be controlled by
ap-propriate annealing. When a Ti50Ni25Cu25ribbon is annealed
at 500C for less than 1 min, no obvious phase transformation and well-defined shape recovery can be obtained. When a Ti
50-Ni25Cu25 ribbon is annealed at 500C for 3 min, the initial
as-crystallized grains contain a lower Cu content than the amorphous matrix, but they exhibit a prominent shape mem-ory effect. Through prolonging the annealing time, more grains are nucleated and the Cu content in the crystallized grains grad-ually increases. The more the grains are crystallized, the more the annealed ribbons become fragile and, as a consequence, their recoverable strain deteriorates. This feature is related to
Table 1
EDS analyses for the chemical composition of points 1e 8 denoted inFig. 10 Location Composition
Ti (at.%) Ni (at.%) Cu (at.%) Remark
Point 1 52.07 25.01 22.92 Bubble Point 2 48.22 24.86 26.91 Amorphous Point 3 54.51 28.88 16.61 Grain Point 4 49.29 24.97 25.74 Amorphous Point 5 51.72 24.85 23.43 Grain Point 6 48.55 27.38 24.07 Amorphous Point 7 50.04 25.15 24.81 Grain Point 8 49.29 24.97 25.74 Amorphous
the increased Cu content in the crystallized grains which can embrittle the annealed ribbon and influence the lattice param-eters of the phases and affect the recoverable strain during the B2 4 B19 transformation. Consequently, Ti50Ni25Cu25
ribbons can exhibit a good shape memory effect only under appropriate annealing condition.
Acknowledgement
The authors gratefully acknowledge the financial support from National Science Council (NSC), Taiwan, Republic of China, under the grant NSC93-2216-E002-003. We also sincerely acknowledge Dr. Ho-Sou Chen, a fellow of the American Physical Society, for his assistance in ribbon prepa-ration and for his guidance in the investigation of the amor-phous state performed in this study.
References
[1] Wayman CM, During TW. Engineering aspects of shape memory alloys. In: During TW, Melton KN, Sto¨ckel D, Wayman CM, editors. London: Butterworth-Heinemam Press; 1990.
[2] Nam TH, Saburi T, Shimizu K. Mater Trans JIM 1990;31:959. [3] Furuya Y, Matsumoto M, Kimura HS, Masumoto T. Mater Sci Eng A
1991;147:7.
[4] Tang W, Sndstrom R, Wei ZG, Miyazaki S. Metall Mater Trans 2000;31A:2423.
[5] Lo YC, Wu SK, Horng HE. Acta Metall Mater 1993;41:747. [6] Bricknell RH, Melton KN, Mercier O. Metall Trans 1979;10A:693.
[7] Miyazaki S, Mizukoshi K, Ueki T, Sakuma T, Liu YN. Mater Sci Eng A 1999;273e275:658.
[8] Xie ZL, Van Humbeek J, Liu Y, Delaey L. Scripta Mater 1997; 37:363.
[9] Ro¨sner H, Shelyakov AV, Glezer AM, Feit K, Schlobmacher P. Mater Sci Eng A 1999;273e275:733.
[10] Satto C, Ledda A, Potapov P, Janssens JF, Schryvers D. Intermetallics 2001;9:395.
[11] Ro¨sner H, Schlobmacher P, Shelyakov AV, Glezer AM. Acta Mater 2001;49:1541.
[12] Santamarta R, Schryvers D. Mater Trans 2003;44:1760.
[13] Santamarta R, Cesari E, Pons J, Goryczka T. Metall Mater Trans 2004;35A:761.
[14] Liu Y. Mater Sci Eng A 2003;354:286.
[15] Cheng GP, Xie ZL. J Alloys Compd 2005;396:128. [16] Eckert J, Schultz L. J Non-Cryst Solids 1991;127:90. [17] Buschow KH. J Phys 1983;F13:563.
[18] Chen JZ, Wu SK. Thin Solid Films 1999;339:194. [19] Chen JZ, Wu SK. J Non-Cryst Solids 2001;288:159. [20] Seeger C, Ryder PL. Mater Sci Eng A 1994;179e180:641. [21] Kissinger HE. Anal Chem 1957;29:1702.
[22] Chang SH, Wu SK, Kimura H. Private communication; 2005. [23] Dong YD, Gergan G, Scott MG. J Non-Cryst Solids 1981;43:403. [24] Chen HS, Park BK. Acta Metall 1973;21:395.
[25] Chen HS. Acta Metall 1974;22:897.
[26] Weeber AW, Bakker H. Physica B 1988;153:93. [27] Miedema AR. Philips Tech Rev 1976;36:217.
[28] Santamarta R, Cesari E, Pons J, Goryczka T. Metall Mater Trans 2004;35A:761.
[29] Khantachawana A, Mizubayashi H, Miyazaki S. Mater Trans 2004;45:214.
[30] Santamarta R, Pons J, Cesari E. J Phys IV France 2001;11:351. [31] Nam TH, Saburi T, Nakata Y, Shimizu K. Mater Trans JIM
