+Model

ARTICLE IN PRESS

MSA-23061; No. of Pages 4

Materials Science and Engineering A xxx (2007) xxx–xxx

The influence of carbon and oxygen content on the martensitic transformation temperatures and enthalpies of NiTi shape memory alloy J. Otubo a,∗ , O.D. Rigo b , A.A. Coelho b , C.M. Neto a , P.R. Mei b a b

Instituto Tecnologico de Aeron´autica-ITA, 12228-900 S.J. dos Campos, SP, Brazil Universidade Estadual de Campinas-UNICAMP, 13083-970 Campinas, SP, Brazil

Received 22 May 2006; received in revised form 9 January 2007; accepted 9 February 2007

Abstract The martensitic transformation temperatures and enthalpies of NiTi shape memory alloy strongly depend on the content of nickel and on carbon and oxygen impurities. Nickel stabilizes the high-temperature phase while carbon and oxygen enrich the surrounding matrix with nickel. In this work it is shown that, as a consequence, the martensitic transformation temperatures and enthalpy changes are lower than specially prepared lowcontamination reference samples. Furthermore, the enthalpy changes increase linearly with increasing peak martensitic transformation temperature. This means that the lower the enthalpies, the lower is the peak martensitic transformation temperature suggesting that at some lower temperature no phase change should occur. © 2007 Elsevier B.V. All rights reserved. Keywords: NiTi SMA; Martensitic transformation temperature and enthalpy; Impurities carbon and oxygen

1. Introduction Our research group is working on the NiTi shape memory alloy (SMA) processing since 1997 using two melting processes, electron beam melting (EBM) and vacuum induction melting (VIM) to produce ingots with low-contamination by carbon and oxygen on a pilot scale [1–5]. The contamination by carbon is inherent to VIM due to melting in a graphite crucible and highly depends on the quality of the graphite as well as the size of the crucible used. The larger the crucible the lower is the contamination by carbon due to the smaller area of contact between the melt and the crucible [4]. In EBM, no contamination by carbon occurs since a water cooled copper crucible is used. In this melting route the carbon content in the final product comes from the initial raw material, mainly titanium. In both processes, the contamination by oxygen depends on the raw materials since VIM and EBM operate in high vacuum (better than 10−2 Pa). The contamination by carbon and oxygen renders the mechanical processing of the ingots more difficult due to embrittlement. Precipitates containing carbon and oxygen modify the original matrix composition

and thus influence the martensitic transformation temperatures [2,3]. Depending on the degree of contamination, a decrease as large as 100 K of the direct peak martensitic transformation temperatures MP (as well as for AP ) can be observed in comparison with specially prepared reference samples with low impurities [4,6]. The influence of contamination is highlighted mainly for nickel contents above equiatomic composition. Nickel stabilizes the high-temperature phase while carbon precipitates TiC particles and oxygen forms Ti4 Ni2 O complex oxide, that is, these two compounds withdraw more titanium than nickel thus enriching the surrounding matrix with nickel and explaining the lowering of MP temperatures in comparison with the reference curve [4,7,8]. Frenzel et al. [9] showed a similar behavior for MS (direct martensitic start temperature instead of MP ) which had been initially compiled by Sawaguchi et al. [10]. This work shows that, in addition to the lowering of the martensitic transformation temperatures, the contamination by carbon and oxygen affects the enthalpy changes and their relationship to peak martensitic transformations temperatures. 2. Experimental procedures

∗

Corresponding author. Tel.: +55 12 39475905; fax: +55 12 39475887. E-mail address: [email protected] (J. Otubo).

The NiTi SMA ingots were produced by two processes: VIM and EBM. Through VIM, the ingots were melted in a graphite

0921-5093/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2007.02.137

Please cite this article in press as: J. Otubo et al., Mater. Sci. Eng. A (2007), doi:10.1016/j.msea.2007.02.137

+Model MSA-23061;

ARTICLE IN PRESS

No. of Pages 4

2

J. Otubo et al. / Materials Science and Engineering A xxx (2007) xxx–xxx

Table 1 Ingots chemical compositions and respective martensitic transformation parameters MP , AP , AP –MP , HM , HA , HM and MP INGOT 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 a

Ni (at.%) 48.62 49.50 49.60 49.63 49.66 49.68 49.74 50.01 50.13 50.17 50.18 50.21 50.42 50.49 50.58

C (ppm)a

O (ppm)a

240 70 1880 240 130 660 140 543 1000 1100 600 2440 520 580 150

1536 1050 329 1482 640 1113 640 924 390 621 564 358 570 837 1055

MP (K)

AP (K)

AP − MP (K)

HM (J/g)

HA (J/g)

HM (J/g)

MP (K)

328.1 331.3 246.5 286.9 333.4 342.2 328.0 295.0 227.0 205.5 294.2 255.0 268.1 271.1 243.1

363.0 364.0 269.2 319.7 366.7 354.9 359.3 325.7 251.1 233.9 320.2 277.1 290.0 290.8 262.2

34.9 32.7 22.7 32.8 33.3 30.7 31.3 30.7 24.1 28.4 26.0 22.1 21.9 19.7 19.7

28.5 30.2 16.6 21.5 – 31.5 23.9 26.6 14.6 9.1 26.5 16.2 22.2 22.4 16.9

−27.0 −30.1 −18.3 −21.6 – −31.7 −22.9 −27.0 −14.9 −9.1 −26.7 −16.5 −23.3 −23.9 −18.8

−2.2 5.3 19.4 14.0 – 4.2 11.4 7.5 19.0 24.9 6.2 15.7 7.8 6.4 10.1

−4.2 1.2 85.5 44.9 2.3 6.5 1.4 25.0 89.0 107.5 18.5 55.5 31.0 22.5 43.5

ppm are per weight.

crucible and then cast also in a graphite mould. The melting power ranged from 15 to 17 kW and the melting chamber internal pressure from 1 to 10−2 Pa. The EBM ingots were prepared in a 80 kW EB furnace with EB power of up to 13.5 kW and melting chamber internal pressure down to 10−3 Pa. The ingots were melted in a water cooled copper crucible. A grade 1 titanium plate and electrolytic nickel were used as raw materials for both processes. The fabrication procedures were described elsewhere [1–5]. The martensitic transformation temperatures and respective enthalpy changes were taken from differential scanning calorimetry (DSC) data. A STA 409C DSC Netzsch equipment was used with cooling and heating rate of 5 K/min. The temperature range was set between 130 and 423 K and the samples were protected in inert helium gas. Further details are given in reference [6]. 3. Results and discussions Table 1 presents the chemical composition of 15 ingots produced by EBM and VIM processes. The nickel content varied from 48.62 to 50.58 at.%. The lowest value of carbon content was 70 wt. ppm presented by the sample taken from EBM ingot while the maximum carbon content was 2440 wt. ppm presented by VIM ingot. It should be emphasized that the carbon content in the commercial product is about 600 wt. ppm [11]. The oxygen content varied from a minimum of 358 wt. ppm to a maximum of 1536 wt. ppm irrespective of the processing technique and depending only upon the initial raw materials nickel and titanium. The relationship between processing techniques and contamination by impurities can be seen in [4]. Table 1 also shows the direct and reverse peak martensitic transformations temperatures MP and AP , the respective hysteresis AP –MP and the enthalpy changes HM and HA (M for martensite and A for austenite, respectively). Furthermore the variation of the enthalpy changes, HM (=HR − HM ) and the variation of the peak martensitic transformation temperatures MP (=MR − MP ) are shown. HR and MR are the reference

enthalpy changes and reference peak martensitic transformation temperatures respectively, taken from specially prepared samples with carbon content below its solubility limit. The effects of oxygen content above its solubility limit were also taken into account [4,6]. The above mentioned reference samples were also used in this work to plot the enthalpy changes, HM (direct martensitic transformation enthalpy change) as a function of nickel content as shown in Fig. 1. The enthalpy changes of the reference samples, HR presented by black triangles and a fitting curve in Fig. 1, show a maximum of 35 J/g at nearly equiatomic composition. Except for one point (ingot 1), all the data from VIM and EBM ingots (open triangles) presented smaller values of enthalpy changes than the reference curve and they varied from a minimum of 9.1 (ingot 10) to a maximum of 31.5 J/g (ingot 6). Consequently, excepting the ingot 1, which presented a larger value than the reference curve, the HM varied from a minimum of 4.2 (ingot 6) to maximum of 24.9 J/g (ingot 10). The smaller

Fig. 1. Enthalpy changes as a function of nickel concentration. Carbon and oxygen contents represent parametric values in this plot.

Please cite this article in press as: J. Otubo et al., Mater. Sci. Eng. A (2007), doi:10.1016/j.msea.2007.02.137

+Model MSA-23061; No. of Pages 4

ARTICLE IN PRESS J. Otubo et al. / Materials Science and Engineering A xxx (2007) xxx–xxx

Fig. 2. Variation in enthalpy change HM (=HR − HM ) as a function of MP (=MR − MP ).

variations or deviations to lower values in relation to reference curve were presented by ingots with low level of contamination such as ingots 8, 11, 13 and 14. From Table 1, one can see that MP is closely related to HM , decreasing or increasing as the last parameter increases or decreases. This behavior can be clearly visualized in Fig. 2 where the variations in enthalpy changes, HM , are plotted as a function of MP . This shows that the relationship between both parameters is linear with positive slope. Emphasizing this aspect, Fig. 3 shows a corresponding curve for the absolute values of HM plotted as a function of MP (open triangles) adding data also from reference samples (black triangle). Their relationship is linear with positive slope irrespective of sample source or degree of contamination. The enthalpy changes decrease as the MP temperatures decrease indicating that the lower the martensitic transformation temperature

3

the more difficult is the transformation and at sufficiently low temperature no transformation should occurs at all. All the discussions above are valid also for the reverse enthalpy changes HA and the reverse peak martensitic transformation temperatures AP on heating. The only difference is that the values are shifted to higher values as can be seen in Table 1. The last aspect that should be pointed out is the small hysteresis AP –MP presented by the NiTi samples as shown in Table 1. Its values varied from a minimum of 19.7 K to a maximum of 34.9 K, typical for thermoelastic martensitic transformation of shape memory alloys [12]. Although not completely clear up to now, the data from Table 1 indicate that the hysteresis values decrease as the MP (or HM ) decrease. Usually, when one compares different SMA, the small hysteresis is attributed to the easiness of direct and reverse martensitic transformation. This is not the case here since we are dealing with only one type of SMA, that is, NiTi. Therefore the decrease in hysteresis as the MP temperature decreases could be attributed to thermodynamic instability of martensitic phase which serves as a driving force for backward movement to high-temperature phase. This aspect is also corroborated by the decrease in enthalpy changes with the decrease of the peak martensitic transformation temperature presenting linear relationship. Studies are underway to better understand this issue. 4. Conclusions The enthalpy changes of reference samples presented a maximum of 35 J/g around equiatomic composition. Except for ingot 1, all the ingots presented enthalpy changes lower than those on the reference curve and varying from a minimum of 9.1 (ingot 10) to a maximum of 31.5 J/g (ingot 6). The enthalpy changes HM presented values from a minimum of 4.2 J/g to a maximum of 24.9 J/g being lower for lower contamination. The relationship between HM (HM ) and MP (MP ) is linear with positive slope irrespectively of sample source or degree of contamination. The hysteresis AP –MP presented by the NiTi samples varied from a minimum of 19.7 K to a maximum of 34.9 K with tendency to decrease when the peak martensitic transformation temperature decreases. Acknowledgments To FAPESP grant 00/09730-1 for financial support. To FINEP grant CT-INFRA 03/2003/30. References

Fig. 3. Enthalpy change HM as a function of peak direct martensitic transformation temperature MP .

[1] J. Otubo, P.R. Mei, S. Koshimizu, L.G. Martinez, In: The Third Pacific Rim International Conference on Advanced Materials and Processing (PRICM 3), M.A. Imam, R. DeNale, S. Hanada, Z. Zhong, D.N. Lee (Eds.), The Minerals, Metals & Materials Society, TMS, vol. 1, 1998, pp. 1063–1068. [2] J. Otubo, O.D. Rigo, C. Moura Neto, M.J. Kaufman, P.R. Mei, J. Physique IV (France) 112 (2003) 873–876.

Please cite this article in press as: J. Otubo et al., Mater. Sci. Eng. A (2007), doi:10.1016/j.msea.2007.02.137

+Model MSA-23061; 4

No. of Pages 4

ARTICLE IN PRESS J. Otubo et al. / Materials Science and Engineering A xxx (2007) xxx–xxx

[3] J. Otubo, O.D. Rigo, C. Moura Neto, M.J. Kaufman, P.R. Mei, Mater. Res. 7 (2004) 263–267. [4] J. Otubo, O.D. Rigo, C. Moura Neto, P.R. Mei, Mater. Sci. Eng. A 438–440 (2006) 679–682. [5] O.D. Rigo, J. Otubo, C. Moura Neto, P.R. Mei, J. Mater. Process. Technol. 162–163 (2005) 116–120. [6] O.D. Rigo, J. Otubo, P.R. Mei, Proceedings of the 60◦ Congresso Anual da ABM, Belo Horizonte, MG, Brazil, 25–28 July, 2005, 2412–2420. [7] Y. Shugo, K. Yamauchi, R. Miyagawa, T. Honma, Bull. Res. Inst. Miner. Dress. Metall. (Tohoku Univ.) 38 (1982) 11–20.

[8] P. Olier, F. Barcelo, J.L. Bechade, J.C. Brachet, E. Lefevre, G. Guenin, J. Physique IV (France) 7 (C5) (1997) 143–148. [9] J. Frenzel, Z. Zhang, K. Neuking, G. Eggeler, J. Alloys Compd. 385 (2004) 214–223. [10] T. Sawaguchi, G. Kaustr¨ater, A. Yawny, M. Wagner, G. Eggeler, Met. Mater. Trans. 34A (2003) 2847–2859. [11] ASTM F-2063-00, Standard specification for wrought nickel-titanium shape memory alloys for medical devices and surgical implants, 2000. [12] J.D. Verhoeven, Fundamental of Physical Metallurgy, John Wiley & Sons, 1975, pp. 490–491.

Please cite this article in press as: J. Otubo et al., Mater. Sci. Eng. A (2007), doi:10.1016/j.msea.2007.02.137