Fabrication of Ti-Ni-Zr Shape Memory Alloy by P/M Process

Akira Terayama1, Koji Nagai2;* and Hideki Kyogoku2

1Hiroshima Prefectural Technology Research Institute, Kure 737-0004, Japan2Department of Mechanical Engineering, Faculty of Engineering, Kinki University, Higashihiroshima 739-2116, Japan

This work focuses on the fabrication of Ti-Ni-Zr high-temperature shape memory alloy by powder metallurgy (P/M) process. The effectsof fabrication conditions on the microstructure and shape memory characteristics of Ti-50.2mol%Ni-5mol%Zr alloy were investigated. In thisresearch, elemental Ti, Ni and Zr powders were used. These powders were mixed by a planetary ball mill at a rotational speed of 500 rpm formilling times of 0.6 ks (mixed powder) and 720 ks (MAed powder). The mixtures were sintered by a pulse-current pressure sintering equipmentat 1153K for sintering times of 1.8 ks and 1.2 ks. The solution treatment was carried out at various temperatures between 1073K and 1273K tohomogenize the microstructure of the as-sintered alloy. The microstructure of the alloy became more homogeneous with an increase in solution-treatment temperature. In the case of the mixed powder, however, Zr-rich phases were observed in the microstructure of the solution-treatedalloy. The alloy solution-treated at 1173K showed a yielding behavior in the stress-strain curve, and the tensile strength and elongation of thealloy were more than 350MPa and 2.5%, respectively. On the other hand, in the case of the MAed powder, the microstructure of the as-sinteredalloy was homogeneous. The P/M alloy showed higher transformation temperatures than those of the wrought alloy. But, the alloy showed noshape memory effect and poor tensile property due to contamination of the MAed powder. [doi:10.2320/matertrans.M2009213]

(Received June 23, 2009; Accepted July 30, 2009; Published September 25, 2009)

Keywords: titanium-nickel-zirconium shape memory alloy, mechanical alloying, pulse-current pressure sintering, tensile properties, shape

memory characteristics

1. Introduction

Ti-Ni shape memory alloy (SMA) has been applied tovarious industrial fields because of unique shape memoryeffect and superelasticity. Ternary alloying addition canimprove the shape memory characteristics of the alloy.Especially, substitution of zirconium for titanium makes themartensitic transformation temperatures more than 373K.Thus Ti-Ni-Zr alloy is promising to apply to high-temper-ature SMA. Hsieh et al.1) reported that the transformationtemperatures of Ti-Ni-Zr alloys with various Zr contentfabricated by vacuum arc melting increase with increasing Zrcontent and the alloys with more than 15mol%Zr have themartensitic transformation finished point (Mf ) of more than373K. Laves phase crystallized out at grain boundaries of B2phase during solidification leads to brittleness of the alloy.Therefore, the application of powder metallurgy (P/M)process that is a solid-state process has been tried to fabricateTi-Ni-Zr SMA. Monastyrsky et al.2,3) reported that the P/Mprocessed Ti-Ni-Zr alloys have many types of secondaryphases such as Ti-Ni alloy (Ti2Ni, TiNi3) and Ni-Zr alloy(NiZr2, Ni5Zr). Bertheville4) reported that almost single-phase ternary Ti-Ni-Zr alloys can be obtained using finerTiH2 and ZrH2 powders, although a few Zr-rich secondaryphases exist in the alloy. There are no reports that thewrought Ti-Ni-Zr SMA showed sufficient tensile strengthand shape memory characteristics. It has been reported,however, that sputtered Ti-Ni-Zr thin film shows superiortensile properties and shape memory characteristics becausesputtering process can be obtained submicron-order grainsizes. Sawaguchi et al.5) reported that the sputtered Ni-poor Ti-Ni-Zr thin films have higher critical stress for slipdeformation of more than 320MPa and reducing in volumefraction of Laves phase is quite significant for improvementof recoverable strain.

The authors reported that the P/M Ti-Ni and Ti-Ni-CuSMAs fabricated by a pulse-current sintering using Ti, Ni andCu elementally powders have superior shape memorycharacteristics comparable to those of the wrought alloy.6–8)

Using a mechanically alloyed (MAed) powder has anadvantage in homogeneous microstructure of the sinteredalloy,9) moreover, MA is a favorable process for refinementof grain size in sintered Ti-Ni-Zr alloy.

In this research, Ti-Ni-Zr high-temperature shape memoryalloy was fabricated by P/M process. The effects offabrication conditions on the microstructure and shapememory properties of Ti-50.2mol%Ni-5mol%Zr alloy wereinvestigated.

2. Experimental Procedure

In this research, elementally Ti, Ni and Zr powders wereused. The chemical composition and mean particle diameterwere shown in Table 1. The powder was milled using aplanetary ball mill at a composition of Ti-50.2mol%Ni-5mol%Zr under two conditions; at a rotation speed of500 rpm for 0.6 ks (mixed powder) and 720 ks (MAedpowder). Figure 1 shows X-ray diffraction patterns of thesepowders. In the case of the mixed powder, peaks of the Ti, Niand Zr are observed on X-ray diffraction pattern. The powderis just mixed and not alloyed. On the other hand, the Ti, Niand Zr peaks vanishes completely in the case of the MAedpowder. These powders were filled into graphite die andsintered in vacuum using a pulse-current sintering equip-ment. The mixed powder was sintering at a temperature of1153K for 1.8 ks. To homogenize the microstructure of thealloy, the solution treatment was performed in argon atmo-sphere at various temperatures between 1073K and 1273Kbecause the powder was inhomogeneous as already report-ed.6) The microstructure of the MAed powder may consist offiner grains, and a large amount of dislocations stored duringMA. It is important not only to release dislocations but also*Graduate Student, Kinki University

Materials Transactions, Vol. 50, No. 10 (2009) pp. 2446 to 2450#2009 The Japan Institute of Metals

to restrain the formation and grain growth of Laves phaseduring sintering to obtain sufficient shape memory character-istics of the sintered alloy. Therefore, sintering time of theMAed powder is selected a shorter time of 1.2 ks. X-raydiffraction, Energy dispersive spectrometry (EDS) analysisand tensile test were performed to examine the propertiesof the as-sintered alloy and the solution-treated alloy. Thetensile test specimens were prepared by cutting and grindingthe alloys into the dimensions of 55mm in length, 5.5mmin width and 1.5mm in thickness. The gauge length of thespecimens was 25mm.

3. Results and Discussions

3.1 Effect of mechanical alloying on the microstructureof the alloys

Figure 2 shows the secondary electron image (SEI), Ti-K�

image, Ni-K� image and Zr-L� image of (a) the as-sinteredalloy by the mixed powder, (b) the solution-treated (1173Kfor 172.8 ks) alloy by the mixed powder and (c) the as-sintered alloy by the MAed powder. In the case of the as-

Table 1 Powder characteristics and chemical compositions of Ti, Ni and Zr powders.

Mean particle Chemical compositions (mass%)

diameter (mm) Fe O N H Ca Mg C Cu

Ti 24 0.053 0.13 0.005 0.007 — — 0.009 —

Ni 6.2 0.003 0.069 — — — — 0.078 —

Zr 100 0.01 — — — 0.08 0.07 — 0.003

20 30 40 50 60 70 80

Inte

nsi

ty (

a.u

.)

2 ( )

Ti-50.2mol%Ni-5mol%Zr

MAed powder (milled for 720 ks)

Mixed powder(milled for 0.6 ks)

Fig. 1 X-ray diffraction patterns of the mixed and MAed powders.

(a) As-sintered alloy by the mixed powder

(b) Solution-treated (1173 K for 172.8 ks) alloy by the mixed powder

(c) As-sintered alloy by the MAed powder

SEI Ti-Kα Ni-Kα Zr-Lα

Fig. 2 EDS analysis of the alloys fabricated under various fabrication conditions.

Fabrication of Ti-Ni-Zr Shape Memory Alloy by P/M Process 2447

sintered alloy by the mixed powder shown in Fig. 2(a),dispersed white zone whose diameters are approximately30 mm in the SEI corresponds to Zr. It is difficult for the as-sintered alloy to archive homogeneous structure. In the caseof the solution-treated (1173K for 172.8 ks) alloy by themixed powder shown in Fig. 2(b), the interface between Zrand matrix is difficult to observe clearly. This means thatdissolving of Zr into the matrix was proceeded by thesolution-treatment. In all images, Ni distributes uniformlybecause Ni atoms diffuse into Ti and Zr particles during theearly stage of sintering.2) In the as-sintered alloy by theMAed powder shown in Fig. 2(c), uniform distribution ofthree elements is observed. This is different from that of thewrought alloy. It was reported that, in the microstructure ofthe wrought alloy, Laves phase located around the (Ti,Zr)Nigrain boundaries and distribution of elements was notuniform.1) Thus, the MA process is effective to obtainhomogeneous microstructure of the sintered alloy.

To measure the structure of matrix and the transformationtemperatures of the alloys, X-ray diffraction analysis andDSC measurement were performed. Figures 3 and 4 demon-strates the X-ray diffraction patterns and the DSC coolingcurves of the solution-treated (1073, 1173 and 1273K) alloyby the mixed powder and the as-sintered alloy by MAedpowder, respectively. In X-ray diffraction patterns of thealloys solution-treated at 1073K and 1173K the peaks ofmany secondary phases in Ti-Ni system (Ti2Ni, TiNi3) andNi-Zr system (NiZr2) are observed. Since the Zr phase isobserved, solid-solution of Zr into the matrix is insufficient.The microstructure becomes homogeneous and the secon-dary phases vanish with elevating the solution-treatmenttemperature. A peak of the B2 austenite phase becomeshigher at a solution-treatment temperature of 1273K. TheB190 martensite phase and the B2 austenite phase areobserved in the as-sintered alloy by the MAed powder.Moreover, although the peaks of the secondary phases are notobserved, the peak of TiNi5O7 is observed. This phase maybe formed during MA because surface of the powder wasactivated by MA process and easily reacted with oxygen.M� indicates the peaks of temperature of the martensitictransformation in Fig. 4. Comparison of M� in the solution-treated alloys, the transformation temperature M� falls with

elevating the solution-treatment temperature. This may berelated to dissolving of Zr into the matrix. The as-sinteredalloy by the MAed powder has the highest transformationtemperatureM� of 364K. But the transformation temperatureM� of the wrought alloy at the same composition wasapproximately 323K as already reported.1) Thus the trans-formation temperature M� of the as-sintered alloy by theMAed powder is much higher than that of the wroughtalloy. The authors reported that the alloy fabricated by theMA powder tends to have higher transformation temper-atures than those of the wrought alloy.9) The reason why thisphenomenon may be caused by presence of internal stressfield formed by storage of dislocation during MA process ordecreasing of Ni content in the alloy by forming Ni-richoxidative product (TiNi5O7). The alloy has the narrowesttransformation temperature width of 7K and the smallestheat change of transformation in the DSC curve as shown inFig. 4. Sandu et al.10) reported that Ti-52mol%Ni-6mol%Zrwrought alloy has a narrow transformation width and asimilar DSC curve to that of our research. This may closelybe related to dissolving Zr into TiNi matrix, but, furtherresearch is needed. The oxidative product also may interruptthe martensitic transformation and lower the heat changeof transformation.

3.2 Tensile properties and shape memory character-istics of the alloys

To measure the tensile properties of the alloys, tensile testswere carried out at room temperature. The stress-straincurves of the alloys are shown in Fig. 5. In the solution-treated alloys by the mixed powder, the deformationresistance increases with an increase in the solution-treat-ment temperature. The tensile strength and elongation of thealloy solution-treated at 1173K were more than 350MPa and2.5%, respectively. The deformation resistance of the alloysolution-treated at 1273K is the highest among three alloys,but the elongation is the lowest. This may be due toappearance of Laves phase, which has the solid-liquidtransition temperature at 1203K1) at a solution-treatmenttemperature of 1273K. The stress-strain curve of the as-sintered alloy by the MAed powder is not given in this figurebecause the alloy fractured at lower stress of approximately125MPa. To improve the tensile property of the as-sintered

30 40 50 60 70

Inte

nsi

ty (

a.u

.)

2 ( )

: TiNi3: NiZr2 : cubic(B2): monoclinic(B19')

: Ti2Ni

Solution-treated at 1073 K

As-sintered (MAed powder)

Solution-treated at 1173 K

Solution-treated at 1273 K

: TiNi5O7

: Zr

Fig. 3 X-ray diffraction patterns of the alloys.

200 250 300 350 400 450

Temperature, T/K

Hea

t F

low E

nd

oth

erm

1073 K

1173 K

1273 K

As-sintered (MAed powder)

M*: 324 K

M*: 322 K

M*: 280 K

M*: 364 K

Fig. 4 DSC cooling curves of the alloys.

2448 A. Terayama, K. Nagai and H. Kyogoku

alloy, the solution-treatment was performed at 1173K for43.2 ks. But the solution-treatment did not improve thetensile property of the alloy by MAed powder. Figure 6shows the fracture surface of the as-sintered alloy by theMAed powder. Ductile dimple cannot be observed. Thisbrittle fracture of the alloy is due to excess oxidation of theMAed powder.

The as-sintered alloy by MAed powder shows homo-geneous microstructure and higher transformation temper-atures, but has poor tensile property. The shape memoryproperties of the alloy by the mixed powder solution-treatedat 1073K, which has highest transformation temperaturesamong the solution-treated alloys, were measured. It isimportant to investigate stabilizing behavior of shapememory properties of the alloy. The alloy was loaded up to1.5% strain and unloaded completely at 300K and thenheated up to austenite finished point (Af ), this cycle wererepeated. Figure 7 shows the stress-strain curves of thesolution-treated (1073K) alloy by the mixed powder. Thenumber in this figure indicates the number of cycles. Theresidual strain is observed after unloaded completely becausethe alloy is consisted of martensite phase at 300K. Figure 8demonstrates the strain-temperature curves of the alloys. As

and A� indicates the reverse transformation start temperature

and the reverse transformation peak temperature, respective-ly. In heating process, the strain gradually recovers due toshape memory effect of (Ti,Zr)Ni phase with insufficientdissolving of Zr. This is because of inhomogeneous micro-structure of the alloy. The residual strain is 0.24% in firstcycle, but it recovers after second cycle.

4. Conclusions

In this research, we investigated the fabrication conditionsof Ti-Ni-Zr high temperature shape memory alloy by P/Mprocess. The effect of fabrication conditions on the micro-structure, transformation temperature, tensile properties andshape memory characteristics were investigated. The resultsobtained are as follows:(1) The microstructure of the as-sintered alloy by the

MAed powder was homogeneous, while the micro-structure of the solution-treated alloy by mixed powderwas inhomogeneous.

(2) The reverse transformation peak temperature of theas-sintered alloy by the MAed powder was higher thanthat of the solution-treated alloys and the wroughtalloy.

(3) The deformation resistance of the solution-treatedalloys by the mixed powder increased with an increase

0

100

200

300

400

500

600

0 0.5 1 1.5 2 2.5 3

Strain (%)

Str

ess,

/MP

aTi-50.2mol%Ni-5mol%Zr

Sintering temp. 1153 K

1073K

Solution-treatedat 1273 K 1173 K

σ

Fig. 5 Stress-strain curves of the alloys by the mixed powder.

Fig. 6 Fracture surface of the as-sintered alloy by the MAed powder.

0

50

100

150

200

250

300

0 0.5 1 1.5 2

Strain (%)

Ti-50.2mol%Ni-5mol%Zr

N=1

N=2N=3

Solution-treatment temp. 1073 KSolution-treatment time 43.2 ks

Str

ess,

σ/M

Pa

Fig. 7 Stress-strain curves of the solution-treated alloy (1073K) by the

mixed powder.

0

0.5

1

1.5

2

280 320 360 400

Temperature, T/K

Str

ain

(%

)

Ti-50.2mol%Ni-5mol%Zr

N=1

N=2N=3

Solution-treatment temp. 1073 KSolution-treatment time 43.2 ks

As A* A

f

Fig. 8 Strain-temperature curves of the solution-treated alloy (1073K) by

the mixed powder.

Fabrication of Ti-Ni-Zr Shape Memory Alloy by P/M Process 2449

in the solution-treatment temperature. The tensilestrength and elongation of the alloy solution-treated at1173K by the mixed powder were more than 350MPaand 2.5%, respectively. The as-sintered alloy by theMAed powder had the lowest tensile strength andelongation because of oxidation of the mixed powder.

(4) The alloy solution-treated at 1073K by the mixedpowder had shape memory effect. The shape recoveryproperty of the alloy was stabilized due to the cycle ofheating after loading.Thus the Ti-Ni-Zr SMA could be fabricated by P/Mprocess using elemental powders.

REFERENCES

1) S. F. Hsieh and S. K. Wu: J. Alloy. Compd. 266 (1998) 276–282.

2) G. E. Monastyrsky, V. Odnosum, J. van Humbeeck, V. I. Kolomytsev

and Yu. N. Koval: Intermetalics 10 (2002) 95–103.

3) G. E. Monastyrsky, J. van Humbeeck, V. I. Kolomytsev and Yu. N.

Koval: Intermetalics 10 (2002) 613–624.

4) B. Bertheville: J. Alloy. Compd. 398 (2005) 94–99.

5) T. Sawaguchi, M. Sato and A. Ishida: Metall. Mater. Trans. A 35A

(2004) 111–119.

6) H. Kyogoku, F. Yoshida and T. Sakuma: Proc. The Fourth Pacific Rim

Int. Conf. on Advanced Materials and Processing (PRICM4, 2001)

pp. 1549–1552.

7) A. Terayama, H. Kyogoku, M. Sakamura and S. Komatsu: Mater.

Trans. 47 (2006) 550–557.

8) H. Kyogoku, T. Tanbo, M. Yokota, S. Komatsu, F. Yoshida, T. Sakuma

and U. Iwata: Proc. Adv. Powder Metall. Particulate Mater. 2001,

MPIF (2001) pp. 7.1–7.9.

9) A. Terayama, H. Kyogoku and S. Komatsu: Mater. Sci. Forum 539–

543 (2007) 3313–3318.

10) A. Sandu, K. Tsuchiya, S. Yamamoto, M. Tabuchi, Y. Todaka and

M. Umemoto: Mater. Sci. Forum 539–543 (2007) 3163–3168.

2450 A. Terayama, K. Nagai and H. Kyogoku