This article appeared in a journal published by Elsevier.

The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elseviers archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright

Author's personal copy

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 36573664

journal homepage: www.elsevier.com/locate/jmatprotec

Fabrication of shape memory alloys using the plasma skull pushpull process
C.J. de Arajo a, , A.A.C. Gomes a , J.A. Silva a , A.J.T. Cavalcanti a , R.P.B. Reis a , C.H. Gonzalez b
Multidisciplinary Laboratory on Active Materials and Structures, Department of Mechanical Engineering, Universidade Federal de Campina Grande, Caixa Postal: 10069, Cep: 58109-970, Campina Grande, PB, Brazil b Department of Mechanical Engineering, Universidade Federal de Pernambuco, Av. Acadmico Hlio Ramos, s/n, Cidade Universitria, Cep: 50740-530, Recife, PE, Brazil
a

a r t i c l e
Article history:

i n f o

a b s t r a c t
In this work, the fabrication of shape memory alloys (SMAs) through the use of the plasma skull pushpull (PSPP) process was studied. In this process, small quantities of metallic elements can be quickly melted on a copper crucible using a rotating plasma torch and injected into a metallic mold. In order to validate the utilization of the PSPP process to produce SMA, several NiTi-based and CuAl-based binary, ternary and quaternary alloys were tested. It was veried that ve melts and re-melts of the SMA before injection are enough to obtain a homogeneous product. The obtained SMAs were characterized by optical microscopy and

Received 14 January 2008 Received in revised form 13 August 2008 Accepted 28 August 2008

Keywords: Shape memory alloys Plasma skull NiTi alloys Metal injection molding

microhardness measurements. The thermoelastic martensitic transformation at the origin of the shape memory effect (SME) was studied via electrical resistance as a function of temperature and differential scanning calorimetry (DSC). All studied SMAs were shown to undergo a thermoelastic transformation, thus demonstrating the viability of the PSPP process to fabricate a great variety of SMAs. 2008 Elsevier B.V. All rights reserved.

1.

Introduction

Shape memory alloys (SMAs) are functional metallic materials that can respond to a temperature change by producing a macroscopic deformation. These special metallic alloys are, for the most part, of two families: copper-based alloys (CuZnAl, CuAlNi, CuAlMn, etc.) and NiTi-based alloys (NiTi, NiTiCu, NiTiNb, etc.). The industrial and laboratory scale production of these materials is carried out using different processes, such as air induction melting (AIM), vacuum induction melting (VIM), vacuum consumable and non-consumable arc melting (VAM), electron beam melting (EBM), powder metallurgy (PM), rapid solidication (mainly melt-spinning) and mechanical alloying (MA). The inuence of

some melting techniques to obtain NiTi SMA was discussed by Russel (2001). Recently, Zhang et al. (2005) studied the reaction between NiTi melts and crucible graphite during VIM of NiTi SMA while Rigo et al. (2005) have used ceramic lters during pouring and analyzed the efciency of the ltering process to avoid contamination by carbon in the NiTi SMA production by VIM. Wu (2001) remarks that in the VAM process, consumable electrodes of NiTi are melted and solidify in a water-cooled copper mold. Because there is no contamination from the crucible, cleaner materials with carbon content less than 200 ppm are produced. Multiple re-melts are required in the VAM process to achieve acceptable homogeneity. Otubo et al. (2004) have compared production of NiTi by VIM and EBM, establishing that EBM is a viable process resulting in

Corresponding author. Tel.: +55 83 3310 1058; fax: +55 83 3310 1272. E-mail address: carlos@dem.ufcg.edu.br (C.J. de Arajo). 0924-0136/$ see front matter 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2008.08.025

Author's personal copy

3658
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 36573664

ingots with carbon content as low as 70 ppm. For both processes, the nal oxygen content depends upon the initial raw materials (nickel and titanium). According to Bram et al. (2002), as promising PM routes, hot isostatic pressing (HIP) and metal injection molding (MIM) can be used for the fabrication of NiTi compacts. In addition, experiments realized by Shaw et al. (2002) using a precursor, sacricial foam and a PM approach have shown that a low density open-cell NiTi foam can be fabricated. In the case of rapid solidication, Lin and Wu (2006) have used melt-spinning technique to produce grain-size mixed Ti49 at.% Ni melt-spun ribbons while Kim et al. (2006) have studied the effect of the melt-spinning processing parameters on the solidication structures in Ti30 at.% Ni20 at.% Cu SMA. On the other hand, Li et al. (2006) have demonstrated that MA can be successfully applied to prepare Cu, Al, Ni and Mn pre-alloyed powders used to fabricate CuAlNiMn SMA by vacuum hot pressing and hot extrusion. However, little information exists about the application of processes that use the plasma melting technology to produce SMAs. The technique of argon non-consumable electrode arc melting (AAM) is preferred in laboratories because it is applicable to many kinds of alloys (Otsuka and Wayman, 1998). In this method, raw metallic elements are installed on a copper crucible and irradiated by the argon arc from a tungsten electrode. The obtained button is turned over and re-melted repeatedly to increase the homogeneity of the composition. The plasma arc melting (PAM) method uses a low velocity electron beam that is discharged from a plasma cathode. According to Otsuka and Wayman (1998), electron irradiation from the plasma cathode is milder than that from high voltage EBM or AAM. Therefore, the PAM method results in low loss of the alloy elements. Bhat (1972) considers the basic advantages of plasma melting furnaces as: (a) fast melting rates, (b) negligible or no carbon contamination, (c) vacuum quality melts, (d) stable powerless voltage uctuation, (e) a lower cost of furnace installation compared to VIM and VAM, and (f) no loss of high vapor pressure metals in plasma furnaces operated with inert gases, as well as others. Additionally, in the plasma skull method, melting is accomplished on a ne layer of the same metal, thus reducing the possibility of contamination by the crucible. In this work, fabrication of binary, ternary and quaternary SMAs by plasma melting is studied. The employed process involved melting and casting by plasma skull pushpull (PSSP) where the alloy is melted on a copper crucible and, later, injected into a cylindrical aluminum mold. Then, PSPP can be considered to be a combination of PAM and MIM. Even considering that PSPP is an efcient and commercially available method to produce dental alloy melts, this is the rst time that the technique is carefully employed to develop different types of SMAs, mainly the Cu-based SMAs which are generally produced by other processes (AIM and PM). Only after validation of the process for production of high quality SMAs, it must be used to study fundamental mechanism of these special alloys. To perform this validation with the focus only on the process results, NiTi- and CuAl-based SMAs produced by PSPP were characterized by optical microscopy, microhardness, and electrical resistance as a function of temperature.

2.

Experimental procedure

The raw materials used to produce the CuAl-based SMA were commercial, pure metals (>99.9%). In the case of NiTi SMA, ASTM F67-00 (grade 2) biomedical titanium was used. In Table 1, the nominal chemical composition and main characteristics of each investigated SMA are given. A small amount, between 20 and 30 g, of each SMA was produced by the PSPP process. Fig. 1 shows a typical PSPP sequence for a NiTi SMA. Ni and Ti are initially pilled up on a copper crucible (Fig. 1a) and placed under a protective argon atmosphere. Melting of the raw elements is rapidly performed with a plasma rotation torch created by a tungsten electrode (Fig. 1b), creating a small bottom (Fig. 1c). In order to assure good homogeneity of the SMA, it was veried that ve melts and re-melts must take place. The nal product is obtained when the re-melted bottom (Fig. 1d) is injected into a metallic mold (Fig. 1e) creating a tablet (Fig. 1f) or other designed product. CuAl-based SMAs are produced following the sequence shown in Fig. 1. The PSPP process described in Fig. 1 is extremely fast. Fig. 2 shows the temperature evolution of the copper crucible (Fig. 1c) during the melting of a stainless steel sample of about 22 g. For this measurement, a micro-thermocouple (K-type, 80 m in diameter) was installed in the copper crucible, at about 4 mm from the melt. It can be seen in Fig. 2 that melting is nalized after about 15 s. Specimens that were cut off from the SMA tablets (Fig. 1f) were prepared for metallographic observation using optical microscopy (Olympus, BX51M) and microhardness measurements (Future-Tech, FM-700). All samples were betatized at 850 C for 900 s and water quenched. The occurrence of the thermoelastic martensitic transformation (TMT) was veried by electrical resistance as a function of temperature using an apparatus that was specially designed for this task by Reis et al. (2006). For some SMA with low transformation temperatures (TT < 10 C), thermal analysis by DSC (Mettler, TA-3000) was employed to detect the TMT.

3.

Results and discussion

As can be observed in Fig. 1d, the nal product of the PSPP process is a SMA tablet. However, Fig. 3 shows that adaptation of some stainless steel components, like plates (Fig. 3a) and tubes and pins (Fig. 3c), in the metallic mold (Fig. 3c) allows fabrication of other semi-nished products like prismatic bars (Fig. 3b), small rods (Fig. 3d) and tubes (Fig. 3e).

3.1.

NiTi-based SMA

Fig. 4 shows the typical microstructure of the NiTi-based SMA summarized in Table 1 (NiTi-1 and NiTi-3). It is well known that NiTi SMA presents very ne grains that are difcult to visualize via optical microscopy (Otsuka and Wayman, 1998). Then, as expected, it can be also observed from Fig. 4 that there is no trace of martensite plates. The microstructure of Fig. 4a is similar to the one observed by Chang et al. (2005).

Author's personal copy

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 36573664

3659

Fig. 1 Melting and injection molding sequence of the PSPP process: (a) pilled up raw materials, (b) rotating plasma torch, (c) SMA bottom, (d) SMA layer on the crucible after mold injection, (e) metallic mold, and (f) SMA tablet.

Table 1 Nominal chemical composition of the studied SMA SMA code

NiTi-1 NiTi-2 NiTi-3 NiTi-4 CuAl-1 CuAl-2 CuAl-3

Nominal composition (wt%)

55Ni45Ti 55.16Ni44.84Ti 49.4Ni44.7Ti5.9Cu 48Ni38Ti14Nb 82.5Cu13.5Al4.0Ni 82.5Cu13.2Al4.0Ni0.3Mn 82.5Cu13.2Al4.0Ni0.3Nb

Main expected characteristics

Classical NiTi SMA presenting R-phase (Otsuka and Ren, 2005) Classical NiTi SMA presenting R-phase (Otsuka and Ren, 2005) Remove R-phase and reduce thermal hyteresis (Nam et al., 1990) High hysteresis NiTi SMA (Zhao et al., 2006) CuAlNi SMA with TT < 100 C (Otsuka and Wayman, 1998) Improve ductility of CuAlNi SMA (Otsuka and Wayman, 1998) Increase TT of CuAlNi SMA (Lelatko and Morawiec, 2003)

TT, transformation temperatures.

Comparing Fig. 4a and b, it can be veried that Cu, substituted in place of Ni, changes the morphology of the binary NiTi SMA. On the other hand, it was observed that the morphology of the 48Ni38Ti14Nb (NiTi-4), and NiTi binary SMA (NiTi-2), are very similar to the one of 55Ni45Ti (NiTi-1, Fig. 4a). These microstructures present the hardness values summarized in Table 2. Fig. 5 shows the transformation behaviors of the NiTi binary SMA (NiTi-1 and NiTi-2), as veried by electrical resistance as a function of temperature (ERT).

Table 2 Vickers microhardness (HV) values for the NiTi-based SMA SMA code
NiTi-1 NiTi-2 NiTi-3 NiTi-4

Microhardnesswater quenched (HV)

229.3 12.8 208.4 16.0 260.5 27.8

Fig. 2 Time of melting and molding of a stainless steel sample by the PSPP process.

Author's personal copy

3660
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 36573664

Fig. 3 Some SMA parts obtained from adaptation of the metallic mold in the PSPP process: (a and b) CuAlNi prismatic bar, (c and d) metallic accessories and NiTi mini rods, and (e) NiTi mini tubes.

For these binary SMAs it was observed that the transformation occurs in two steps: from cubic austenite (B2) to an intermediary rhombohedric phase (R), and then to the monoclinic martensite (B19 ). The appearance of the R-phase creates an important peak in the ERT plot during cooling. As reported in the literature (Otsuka and Wayman, 1998; Otsuka and Ren, 2005), reversion of the B19 RB2 transformation during heating is not visible. Fig. 5 also demonstrates that, by increasing the Ni content by about 0.15 wt%, the transformation temperatures are reduced by approximately 20 C. According to Wu (2001), for alloys having greater than 55.0 wt% Ni, a 1wt% deviation in the Ni (or Ti) concentration will result in approximately a 100 C shift in transformation temperatures. This extreme sensitivity requires that a strict requirement

be placed on any melting practice in order to tightly control the Ni and Ti ratio to meet the required tolerance in terms of transformation temperatures. The results presented in Fig. 5 conrm that the PSPP process can fulll these requirements. Some compositions of ternary NiTi-based SMA, like NiTiCu and NiTiNb, present a single transformation (from B2 to B19 ). Fig. 6 shows the ERT and DSC results for the ternary NiTiCu and NiTiNb SMA (NiTi-3 and NiTi-4) produced by the PSPP process. The addition of a small amount of Cu (<7.5 at.%) in place of Ni eliminates the R-phase, thereby creating a classical ERT loop corresponding to a single B2 to B19 transformation

Fig. 4 Microstructures of (a) NiTi-1 and (b) NiTi-3 SMA.

Author's personal copy

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 36573664

3661

Fig. 5 ERT behavior of (a) NiTi-1 and (b) NiTi-2.

(Nam et al., 1990), as conrmed in Fig. 6a. In general, ternary NiTiNb SMA also presents a single B2 to B19 transformation, which is characterized by a large thermal hysteresis (Zhao et al., 2006). The DSC curve of Fig. 6b shows that thermal hysteresis of the NiTiNb produced by PSPP was on the order of 50 C. Table 3 shows all transformation temperatures obtained directly from Figs. 5 and 6.

Fig. 6 Transformation behavior of the NiTiCu and NiTiNb SMA: (a) ERT characteristic of NiTi-3 and (b) DSC of NiTi-4.

3.2.

CuAl-based SMA

Fig. 7 shows all martensitic microstructures and corresponding ERT curves for the CuAl-based SMAs dened in Table 1. Contrary to the NiTi-based SMA, it is possible to identify large grains with martensite plates (Fig. 7a, c and e), which are commonly observed in CuAl-based SMAs (Lin et al.,

Table 3 Transformation temperatures for the studied NiTi-based SMA SMA code Mf
NiTi-1 NiTi-2 NiTi-3 NiTi-4 18.6 6.1 6.9 48.0

Transformation temperatures ( C) Ms
29.2 7.6 14.8 23.1

Rs
35.2 16.6

As
11.2 7.2

Af
52.6 25.7 39.2 23.6

2000) and single step transformations (Fig. 7b, d and f) characterized by a narrow thermal hysteresis (1015 C). Fig. 7b shows that the Ms temperature for the Cu13.5Al4.0Ni (CuAl1) is on the order of 83 C. According to Recarte et al. (1999), Ms temperatures between 30 and 120 C are measured in CuAlNi SMA with a Ni content of 4 wt% and Al in the range of 13.013.8 wt%. As demonstrated in Fig. 7d, the addition of 0.3 wt% Mn in place of Al in the ternary CuAlNi SMA does not affect the transformation. However, Fig. 7f shows that the same substitution using 0.3 wt% Nb increases the transformation temperatures by about 50 C. Ternary CuAlNb SMA are characterized by transformation temperatures that are higher than 200 C (Lelatko and Morawiec, 2003). The experiments presented in Figs. 57 were repeated several times to conrm the reproducibility of the PSPP process. As well known from the literature (Otsuka and Wayman, 1998; Otsuka and Ren, 2005), Fig. 8 shows the phase transformation and consequently transition temperatures derived from electrical resistance measurements, demonstrating that the repetition of some thermal cycles

Author's personal copy

3662
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 36573664

Fig. 7 Microstructure and transformation behavior of the CuAl-based SMA: (a and b) CuAl-1, (c and d) CuAl-2, and (e and f) CuAl-3.

with the same specimen affect the ERT plot of equiatomic NiTi (NiTi-1) presenting R-phase while NiTiCu SMA (NiTi-3) exhibiting a single transformation is thermally stable. Table 4 Vickers microhardness (HV) values for the CuAl-based SMA SMA code Microhardness (HV) As-produced
CuAl-1 CuAl-2 CuAl-3 308.1 20.4 335.2 17.5 304.7 12.4

Water quenched
236.1 12.3 209.7 19.5 241.9 13.2

It is known that the ductility of CuAlNi SMA decreases signicantly with a small increase of the Al content. However, substitution of Al with Mn improves the ductility substantially without making any substantial modications of the transformation temperatures (Otsuka and Wayman, 1998). Table 4 summarizes the hardness of each studied CuAl-based SMA. These results conrm that the martensitic microstructure of the CuAlNiMn (CuAl2) was the least hard. It was also veried that the microstructure of the as-produced CuAl SMA is partially martensitic and has hardness values that are higher than those of the fully martensitic water quenched microstructure.

Author's personal copy

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 36573664

3663

Acknowledgements
The authors thank the Conselho Nacional de Desenvolvimento Cientco e Tecnolgico (CNPq), Brazilian ofce for sponsoring the research (CT-Energ grant 550325/2005-0, CTPetro grant 504365/2004-5 and Universal grant 478651/2007-4) during the course of these investigations.

references

Fig. 8 ERT plot evolution during ve thermal cycles: (a) NiTi-1 and (b) NiTi-3.

4.

Conclusions

In this work, it was demonstrated that several types of SMAs can be produced by the PSPP validating this process before performing other studies concerning fundamental mechanism of these special alloys. Working on the design of molds for the PSPP process makes the fabrication of different, small, seminished products, like tablets, tubes, rod bars, prismatic bars and others possible. The pushpull step imposes several limits on the possibility of improving the amount of alloy in onetime production (2030 g). However, the technique is ideal for producing base materials for milling mini- and micro-shaped memory actuators. The occurrence of the thermoelastic martensitic transformation at the origin of the shape memory phenomena for all studied SMAs was demonstrated mainly by electrical resistance as a function of the temperature. The study of binary NiTi SMA with a difference of 0.15 wt% in the Ni content as well as a modication of this SMA by adding Cu and Nb, have conrmed the capacity of the PSPP method to control the chemical composition and produce high quality alloys with predictable transformation temperatures. The PSPP process was also effective for the production of CuAl-based SMAs, thereby clearly demonstrating the inuence of small contents of Mn and Nb in place of Al in ternary CuAlNi SMA.

Bhat, G.K., 1972. New developments in plasma arc melting. Journal of Vacuum Science and Technology 9 (6), 1344 1350. Bram, M., Ahmad-Khanlou, A., Heckmann, A., Fuchs, B., Buchkremer, H.P., Stover, D., 2002. Powder metallurgical fabrication processes for NiTi shape memory alloy parts. Materials Science and Engineering A 337, 254263. Chang, S.H., Wu, S.K., Chang, G.H., 2005. Grain size effect on multiple-stage transformations of a cold-rolled and annealed equiatomic TiNi alloy. Scripta Materialia 52, 13411346. Kim, Y., Yun, Y., Nam, T., 2006. The effect of the melt spinning processing parameters on the solidication structures in Ti30 at.% Ni20 at.% Cu shape memory alloys. Materials Science and Engineering A 438440, 545548. Lelatko, J., Morawiec, H., 2003. The effect of Ni, Co and Cr on the primary particle structure in CuAlNbX shape memory alloys. Materials Chemistry and Physics 81, 472475. Li, Z., Pan, Z.Y., Tang, N., Jiang, Y.B., Liu, N., Fang, M., Zheng, F., 2006. CuAlNiMn shape memory alloy processed by mechanical alloying and powder metallurgy. Materials Science and Engineering A 417, 225229. Lin, K.N., Wu, S.K., 2006. Martensitic transformation of grain-size mixed Ti51Ni49 melt-spun ribbons. Journal of Alloys and Compounds 424, 171175. Lin, Z.C., Yu, W., Zee, R.H., Chin, B.A., 2000. CuAlPd alloys for sensor and actuator applications. Intermetallics 8, 605611. Nam, T.H., Saburi, T., Shimizu, K., 1990. Cu-content dependence of shape memory characteristics in TiNiCu alloys. Materials Transactions JIM 31 (11), 959967. Otsuka, K., Wayman, C.M., 1998. Shape Memory Materials. Cambridge University Press, Cambridge, UK, 284pp. Otsuka, K., Ren, X., 2005. Physical metallurgy of TiNi-based shape memory alloys. Progress in Materials Science 50, 511678. Otubo, J., Rigo, O.D., Moura Neto, C., Kaufman, M.J., Mei, P.R., 2004. Low carbon content NiTi shape memory alloy produced by electron beam melting. Materials Research 7 (2), 263 267. Recarte, V., Perez-Saez, R.B., Bocanegra, E.H., N, M.L., San Juan, J., 1999. Dependence of the martensitic transformation characteristics on concentration in CuAlNi shape memory alloys. Materials Science and Engineering A 273275, 380384. Reis, R.P.B., De Arajo, C.J., Silva, L.A.R., Queiroga, S.L.M., 2006. Development of an electrical resistance measurement system as a function of temperature: application to the characterization of shape memory alloys. In: Proceedings of Forth National Congress of Mechanical Engineering (CONEM 2006), Recife, PE, Brazil, pp. 110 (in Portuguese). Rigo, O.D., Otubo, J., Moura Neto, C., Mei, P.R., 2005. NiTi SMA production using ceramic lter during pouring the melt. Journal of Materials Processing Technology 162163, 116120. Russel, S.M., 2001. Nitinol melting and fabrication. In: Russell, S.M., Pelton, A.R. (Eds.), Proceedings of the International Conference on Shape Memory and Superelastic Technologies. International Organization on SMST, Pacic Grove, CA, pp. 110.

Author's personal copy

3664
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 36573664

Shaw, J.A., Gremillet, A., Grummon, D.S., 2002. The manufacture of NiTi foams. In: Proceedings of 2002 ASME International Mechanical Engineering Congress and Exposition, New Orleans, LA, USA, pp. 110. Wu, M.H., 2001. Fabrication of Nitinol materials and components. In: Proceedings of the International Conference on Shape Memory and Superelastic Technologies, Kunming, China, pp. 285292.

Zhao, X., Yan, X., Yang, Y., Xu, H., 2006. Wide hysteresis NiTi(Nb) shape memory alloys with low Nb content (4.5 at.%). Materials Science and Engineering A 438440, 575578. Zhang, Z., Frenzel, J., Neuking, K., Eggeler, G., 2005. On the reaction between NiTi melts and crucible graphite during vacuum induction melting of NiTi shape memory alloys. Acta Materialia 53, 39713985.