lable at ScienceDirect

Intermetallics 85 (2017) 163e169

Contents lists avai

Intermetallics

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

Multiple-stage transformation behavior of Ti49.2Ni50.8 alloy withdifferent initial microstructure processed by equal channel angularpressing

Y.X. Tong a, *, K.P. Hu a, F. Chen a, B. Tian a, L. Li a, Y.F. Zheng b

a Institute of Materials Processing and Intelligent Manufacturing, College of Materials Science and Chemical Engineering, Harbin Engineering University,Harbin, 150001, Chinab Department of Materials Science and Engineering, College of Engineering, Peking University, Beijing, 100871, China

a r t i c l e i n f o

Article history:Received 23 May 2016Received in revised form13 September 2016Accepted 23 January 2017Available online 24 February 2017

Keywords:Shape-memory alloysEqual channel angular pressingMartensitic transformationMicrostructure

* Corresponding author.E-mail address: tongyx@hrbeu.edu.cn (Y.X. Tong).

http://dx.doi.org/10.1016/j.intermet.2017.01.0170966-9795/© 2017 Elsevier Ltd. All rights reserved.

a b s t r a c t

Multiple-stage transformation of Ti49.2Ni50.8 alloy processed by equal channel angular pressing (ECAP)was investigated as a function of pass number and aging treatment before ECAP. When the pass numberis no more than four passes, three stage transformation, namely A/R, R1/M1 and R2/M2, occurs in theas-ECAP processed alloy initially aged at 450 �C for 60 min. Only the A/R/M forward transformationoccurs provided that the aging duration was decreased/increased to 10/600min. The transformationsequence was discussed based on the microstructure evolution of as-ECAP processed alloy with differentinitial microstructure and pass number.

© 2017 Elsevier Ltd. All rights reserved.

1. Introduction

TiNi-based shape memory alloys (SMAs) have attracted muchattention in many engineering and biomedical applications due totheir superior functional properties, including large work outputforce per unit volume, large displacement generated during shaperecovery and the excellent biocompatibility [1]. In order to furtherenhance the properties of TiNi-based SMAs, equal channel angularpressing (ECAP), a several plastic deformation technique, has beenemployed to refine the microstructure of TiNi-based SMAs [2e4].For example, the ultrafine grained microstructure, with a grain sizeof 0.2e0.3 mm, can be obtained in bulk samples of Ti49.4Ni50.6 andTi49.8Ni50.2 as a result of ECAP [2]. As compared to the coarse-grained counterpart, the ultrafine grained TiNi-based alloys showseveral advantages, including improved shape recovery stress andstrain [5], enhanced cycling stability [6e9], and biocompatibility[10].

Microstructure of the as-ECAP processed TiNi-based alloy can benaturally affected by processing conditions, including processingtemperature [7] and pass number [11] etc. Very recently, our works

[8,12] indicate that the second phase also greatly influences themicrostructure of as-ECAP processed TiNi-based SMAs. To be morespecific, Ti3Ni4 precipitate is beneficial to refining the final micro-structure of as-ECAP processed Ti49.2Ni50.8 alloy when processed byECAP [8], but b-Nb particles retard the grain refinement ofTi44Ni47Nb9 alloy [12]. During ECAP, shear plastic deformation isalso applied on the Ti3Ni4 phase besides the matrix which maycause the re-dissolution of second phase [8,13,14]. In the coarse-grained Ni-rich TiNi alloys, the Ti3Ni4 phase as a result of agingcan cause the multiple-stage transformation [15e17]. It is generallyaccepted that the as-ECAP processed TiNi alloys usually show atwo-stage A(B2 parent phase)/R/M(B190 martensite) trans-formation upon cooling resulting from the refined microstructureand dislocations [8,11,18]. However, as yet, a comprehensive un-derstanding of the effect of initial microstructure on martensitictransformation of as-ECAP processed TiNi SMAs is still missing;knowledge of how the re-dissolution of precipitates influence thetransformation behavior is lacking.

In the present work, the initial microstructure with Ti3Ni4 pre-cipitates having different sizes was obtained in Ti49.2Ni50.8 alloy byaging at 450 �C for different durations. The samples were processedby ECAP for different passes to obtain the various states of Ti3Ni4precipitates. The multiple-stage transformationwas paid particularattention and correlated with the microstructure evolution. This

Y.X. Tong et al. / Intermetallics 85 (2017) 163e169164

understanding may allow optimization of transformation behaviorthrough control of the initial microstructure and the processingcondition.

2. Experimental procedure

A commercial TiNi alloy with a nominal composition ofTi49.2Ni50.8 (at.%) was studied. Before processing, the alloy wassolution-treated in a vacuum furnace at 900 �C for 1 h, and thenquenched intowater. The as-quenched alloy has the microstructurewith an average grain size of 70 mm. In order to obtain the pre-cipitate with different sizes, the solution-treated alloys were agedat 450 �C for 10min, 60 min and 600min, respectively. The sampleswere sealed in the evacuated quartz tubes for heat-treatments.Hereafter, the aged samples were represented by A10, A60 andA600 for simplicity, respectively. After aging, the grain size keepsalmost constant, as confirmed by optical microscopy observation.The samples, in the form of 10mm in diameter and 60mm in lengthrods, were processed by ECAP at a temperature of 450 �C for up to 8passes using a die with a channel-intersection angle of Ф ¼ 120�.The rod was kept at 450 �C for 10 min in a furnace prior to eachpass, transferred to the pre-heated ECAP die as quickly as possibleand then extruded at a rate of 15 mm/s. The pressing route Bc wasused in which the sample was rotated by 90� in the same sense,since it is the optimum one for producing the ultra-fine structure[19].

Thermal cycling was performed on a Perkin-Elmer Diamonddifferential scanning calorimeter (DSC) at a constant heating/

Fig. 1. TEM bright field images of the aged samples, (a) A10, (b) A60 and (c) A600. Diffractiothe solution-treated and aged samples are shown in (d). The transformation sequences are

cooling rate of 20 �C/min. DSC samples have amass between 10 and35 mg. Microstructure was carefully observed on a JEOL 2010transmission electron microscopy (TEM) which was operated at200 kV with a double-tilt sample stage. The samples were cut by alow-speed diamond saw to avoid any change of microstructure. TheTEM foils were prepared bymechanical grinding, followed by twin-jet electropolishing using an electrolyte solution consisting of 95%acetic acid and 5% perchloric acid by volume.

3. Results and discussion

The microstructure of aged samples were observed by TEM atroom temperature. Fig. 1 (a), (b) and (c) show the bright field im-ages of the microstructure of A10, A60 and A600, respectively. Asexpected, numerous thin precipitates with a lens shape present inthe samples, as indicated by the arrows. The selected area electrondiffraction (SAED) patterns is characterized by the 1/7 superlatticespot in the 213 direction, indicating that the precipitates are Ti3Ni4phase. The diffraction spots corresponding to R-phase are alsorevealed. For the A10 sample, the Ti3Ni4 precipitates are sur-rounded by strong stress-field, which means that the precipitatesare coherent with the matrix. With increasing aging duration from10 to 60 and 600 min, the average size of Ti3Ni4 precipitate in-creases from about 20 to 38 and 108 nm, respectively. The obser-vation area is in the grain interior. It should be mentioned that thedistribution of Ti3Ni4 precipitate is uniform across the sample. Fig. 1(d) shows the DSC curves of the solution-treated and aged samples.It is seen that upon cooling, all of the three samples show a two-

n patterns corresponding to the [111]B2 were also inserted, respectively. DSC curves ofalso indicated.

Y.X. Tong et al. / Intermetallics 85 (2017) 163e169 165

stage A/R/M forward martensitic transformation. Upon heating,the A10 and A60 samples show the two-stage M/R/A reversetransformation, and the A600 sample shows the single-stage M/Areverse transformation. The transformation sequence was deter-mined by the partial DSC cycling method [15]. The above results areconsistent with the reported results on the effect of aging onmartensitic transformation of Ti49.2Ni50.8 alloy [16,20,21]. It hasbeen reported by Karbakhsh et al. that Ti49.2Ni50.8 alloy was aged at450 �C after sealing into an evaluated quartz tube, then the agedsample shows normal two-stage transformation upon cooling [16].According to their work [16], the two-stage transformation ispossibly related to the higher degree of Ni supersaturation in the

Fig. 2. TEM bright field images of the initially aged samples processed for different passes, (aA600, 8 passes. Diffraction patterns corresponding to the [111]B2 were also inserted, respec

sample aged at 450 �C whichmay result in the uniform distributionof Ti3Ni4 phase. Therefore, the normal two-stage transformationare expected.

Fig. 2(a) shows the TEM bright field image of the microstructureof A10 sample processed for one pass. After the shear deformationimposed by the one pass ECAP, the initially equiaxed and coarsegrains were greatly elongated and refined. A large number of dis-locations can be observed in the grains having widths of about1 mm. The SAED pattern does not show any superlattice spotsrelated to Ti3Ni4 precipitates, implying that the precipitates havere-dissolved into the matrix due to the plastic deformation. This isconsistent with the reported results [8,13,14]. The brightly spherical

) A10, 1 pass, (b) A10, 8 passes, (c) A60, 1 pass, (d) A60, 8 passes, (e) A600, 1 pass and (f)tively.

Y.X. Tong et al. / Intermetallics 85 (2017) 163e169166

phase is Ti4Ni2Owhich formed duringmelting [22]. This phase doesnot dissolve into thematrix [14] andwill not be discussed hereafter.After processed for eight passes, the elongated grains turned intothe equiaxed shape with an ultrafine grain size of about 0.4 mm, asshown in Fig. 2(b). The evolution of grain morphology with passnumber is a general tendency observed in the as-ECAP processedTiNi alloys [11,14] and other alloys [19,23]. The Ti3Ni4 precipitateswere not found in the matrix.

Fig. 2(c) and (d) show the bright field images of the micro-structure of the A60 samples processed for one pass and eightpasses, respectively. After eight passes, the grain size was reducedto about 0.3 mm. From the inserted SAED pattern, it is seen that theTi3Ni4 precipitates exist after one pass, but re-dissolve into thematrix after eight passes. Fig. 2(e) and (f) show the bright fieldimages of the microstructure of the A600 sample processed for oneand eight passes, respectively. It is seen that the Ti3Ni4 precipitatesexist even after eight passes, as confirmed by the SAED pattern.According to the results shown in Fig. 2, it is proposed that thedegree of redissolution of Ti3Ni4 phase is controlled by the plasticdeformation strain and initial size of precipitate.

Fig. 3(a) and (b) show the DSC curves of the A10 samples pro-cessed for different passes. Upon cooling, as compared to the results

Fig. 3. DSC curves of the initially aged samples processed for different passes, (a) A10, 1 papasses.

shown in Fig. 1(d), the transformation sequence does not change,the transformation temperatures are reduced and the trans-formation peaks become diffused. The reduced transformationtemperatures can be attributed to the suppression of refined grainand dislocations introduced during ECAP [11,12,18]. The diffusedtransformation peak is possibly related to the large number ofdislocations [24]. Upon heating, after one pass, the as-ECAP pro-cessed A10 sample shows the single-stage M/A reverse trans-formation, which is different from that of the A10 sample.

Fig. 3(c) and (d) show the DSC curves of the A60 samples pro-cessed for one and eight passes, respectively. Upon cooling, thesamples processed for less than 4 passes show three-stage forwardtransformation. The sample processed for eight pass shows theA/R/M transformation, same as that of A10 sample. Uponheating, after one pass, the as-ECAP processed sample shows three-stage reverse transformation. After two or four passes, the DSCcurves of as-ECAP processed samples are characterized by the two-stage reverse transformation. The sample processed for eightpasses show the two-stage M/R/A reverse transformation.Fig. 3(e) and (f) show the DSC curves of the A600 samples processedfor different passes. The transformation sequence is the same asthat of the aged one, irrespective of pass number.

ss, (b) A10, 8 passes, (c) A60, 1 pass, (d) A60, 8 passes, (e) A600, 1 pass and (f) A600, 8

Fig. 4. Partial DSC curves of the A60 sample processed for one pass. The original fullDSC curves of this sample has been shown in (a).

Fig. 5. Schematic illustration of microstructure evolution and corresponding transformationc represent heating and cooling, respectively. A and B present precipitate-dissolved and pr

Y.X. Tong et al. / Intermetallics 85 (2017) 163e169 167

In order to identify the transformation peaks of the A60 samplesprocessed by ECAP, the partial DSC cycling method was used [15].The results of the sample processed by one pass are shown in Fig. 4,as an example. The full cycle DSC curves were also shown inFig. 4(a) for the comparison purpose. First, the samplewas cooled toa temperature between peaks 1 and 2. It is seen from Fig. 4(b) thatthe peak 1 corresponds to the peak 5, the transformation hysteresisbetween peaks 1 and 5 is about 3.6 �C. Therefore, this pair resultsfrom the R-phase transformation, which is characterized by a smallhysteresis due to its small lattice deformation [25]. When cooled toa temperature between peaks 2 and 3, the heating DSC curve showsthe peak 6 besides the peak 5 which are overlapped. This suggeststhat the peak 2 corresponds to the peak 6. It is found that uponheating during the second partial cycle (Fig. 4(c)), the size of peak 5is reduced as compared to that in the first partial cycle (Fig. 4(b)).This implies that R phase partially transformed to another phase(M1) during the second cooling, which is quite similar to the pre-vious reported results [26]. In addition, the peak 6 locates at thehigher temperature side of peak 5. Therefore, peaks 2 and 6 areidentified to be R1/M1 and M1/A, respectively. When furthercooling from the cool end temperature of second partial cycle, theresidual R phase transformed to martensite (R2/M2), corre-sponding to peak 3. During heating, M2 transformed to R phase,

sequence for the as-ECAP processed samples with different initial microstructure, h andecipitate-rich regions, respectively.

Y.X. Tong et al. / Intermetallics 85 (2017) 163e169168

resulting in the presence of peak 4. The detailed transformationsequence was also marked in Fig. 3(c). It should be mentioned thatthe subscripts 1 and 2 does not mean that there are two kinds of Rphase or B190 martensite. Using the same method, the reversetransformation sequence of the A60 samples processed for two orfour passes was determined to be M1/A and M2/A.

Fig. 5 schematically illustrates the microstructure evolution andthe induced transformation sequence in the as-ECAP processedsamples with different initial microstructure. It is generallyaccepted that the multiple-transformation of the aged Ni-rich TiNialloys results from the inhomogeneous distribution of Ti3Ni4 phase[15]. However, in the case of A10 sample, this mechanism is notapplicable since the Ti3Ni4 phase re-dissolves into the matrix evenafter one pass due to the severe plastic deformation, as shown inFig. 5(a). With increasing the pass number, the microstructure wasgreatly refined and the dislocation density increase. This isresponsible for the two-stage transformation in the as-ECAP pro-cessed A10 sample.

Once the initial aging duration was increased to 60 min, theaverage size of Ti3Ni4 precipitates increases, as shown in Fig. 5(b).Based on the results shown in Fig. 2(a) and (c), it is reasonablysuggested that there is a critical size of Ti3Ni4 precipitates between20 nm and 38 nm for the present ECAP conditions. The precipitateshaving a size below this critical value may completely re-dissolveinto the matrix after one pass ECAP. The precipitates having a sizeabove this value may fracture or partially re-dissolve into the ma-trix, which requires further investigation. The microstructure ofpart region is similar to that the as-ECAP processed A10 sample andcharacterized by the full re-dissolution of Ti3Ni4 phase and highdensity dislocations. Correspondingly, martensitic transformationoccurs in the two-stage A/R(R1)/M1manner. Themicrostructureof other region is characterized by the residual Ti3Ni4 phase andhigh density dislocations. This should be responsible for theA/R(R2)/M2 transformation. After eight passes, the residualTi3Ni4 phase disappeared. The microstructure of as-ECAP processedA60 sample is quite similar to that of A10 sample other than theslightly different grain size, as shown in Fig. 2(b) and (d). Therefore,martensitic transformation occurs in the same manner with that ofas-ECAP processed A10 sample.

In the case of A600 sample, after eight passes, the Ti3Ni4 phase isstill visible, as confirmed by the TEM results shown in Fig. 2 (f),because that the precipitate size is much larger than the proposedcritical value of re-dissolution. The overlarge precipitates alsoretard the grain refinement, as demonstrated by the elongatedgrains of A600 sample. The precipitate size even increases possiblydue to the interpass annealing. This is supported by the increasedtransformation temperatures with the increase of pass number. Themicrostructure of as-ECAP processed A600 sample is still charac-terized by Ti3Ni4 precipitates and high density dislocations intro-duced by ECAP. This gives rise to the two-stage A/R/Mtransformation shown in Fig. 3(e) and (f).

4. Conclusions

The transformation behavior of Ti49.2Ni50.8 alloy with differentinitial microstructure processed by ECAP depends on the passnumber and initial aging treatment. When the alloy was initiallyaged at 450 �C for 10 min/600 min, the as-ECAP processed alloyshows the A/R/M forward transformation. When the alloy wasinitially aged at 450 �C for 60 min, if the pass number is no morethan four, the forward transformation occurs in the three-stagemanner, namely, A/R, R1/M1 and R2/M2. This can be ascribedto the partial re-dissolution of Ti3Ni4 phase. If the pass number iseight, the forward transformation occurs in the two-stageA/R/M sequence.

Acknowledgments

This work was supported by Natural Science Foundation ofChina (51671064), Harbin City Innovative Talents Research SpecialProgram (2015RAXXJ033) and Postdoctoral Funds for ScientificResearch Initiation of Heilongjiang Province (LBH-Q14035).

References

[1] K. Otsuka, T. Kakeshita, Science and technology of shape-memory alloys: newdevelopments, MRS Bull. 27 (2002) 91e100.

[2] V.G. Pushin, V.V. Stolyarov, R.Z. Valiev, N.I. Kourov, N.N. Kuranova,E.A. Prokofiev, L.I. Yurchenko, Features of structure and phase transformationsin shape memory TiNi-based alloys after severe plastic deformation, Ann.Chim. Sci. Mat. 27 (2002) 77e88.

[3] I. Karaman, H.E. Karaca, Z.P. Luo, H.J. Maier, The effect of severe marforming onshape memory characteristics of a Ti-rich NiTi alloy processed using equalchannel angular extrusion, Metall. Mater. Trans. A 34 (2003) 2527e2539.

[4] V.G. Pushin, V.V. Stolyarov, R.Z. Valiev, T.C. Lowe, Y.T. Zhu, NanostructuredTiNi-based shape memory alloys processed by severe plastic deformation,Mater. Sci. Eng. A 410e411 (2005) 386e389.

[5] E. Prokofyev, D. Gunderov, S. Prokoshkin, R.Z. Valiev, Microstructure, me-chanical and functional properties of NiTi alloys processed by ECAP technique,in: 8th European Symposium on Martensitic Transformations, Prague, CzechRepublic, 2009, p. 06028, 9.7-9.11.

[6] B. Kockar, I. Karaman, J.I. Kim, Y. Chumlyakov, A method to enhance cyclicreversibility of NiTiHf high temperature shape memory alloys, Scr. Mater. 54(2006) 2203e2208.

[7] B. Kockar, I. Karaman, J.I. Kim, Y.I. Chumlyakov, J. Sharp, C.J. Yu, Thermo-mechanical cyclic response of an ultrafine-grained NiTi shape memory alloy,Acta Mater. 56 (2008) 3630e3646.

[8] Y.X. Tong, B. Guo, F. Chen, B. Tian, L. Li, Y.F. Zheng, E.A. Prokofiev,D.V. Gunderov, R.Z. Valiev, Thermal cycling stability of ultrafine-grained TiNishape memory alloys processed by equal channel angular pressing, Scr. Mater.67 (2012) 1e4.

[9] Y.X. Tong, F. Chen, B. Guo, B. Tian, L. Li, Y.F. Zheng, D.V. Gunderov, R.Z. Valiev,Superelasticity and its stability of an ultrafine-grained Ti49.2Ni50.8 shapememory alloy processed by equal channel angular pressing, Mater. Sci. Eng. A587 (2013) 61e64.

[10] C.Y. Zheng, F.L. Nie, Y.F. Zheng, Y. Cheng, S.C. Wei, L. Ruan, R.Z. Valiev,Enhanced corrosion resistance and cellular behavior of ultrafine-grainedbiomedical NiTi alloy with a novel SrOeSiO2eTiO2 solegel coating, Appl.Surf. Sci. 257 (2011) 5913e5918.

[11] X. Zhang, J. Song, C. Huang, B. Xia, B. Chen, X. Sun, C. Xie, Microstructuresevolution and phase transformation behaviors of Ni-rich TiNi shape memoryalloys after equal channel angular extrusion, J. Alloys Compd. 509 (2011)3006e3012.

[12] Y.X. Tong, P.C. Jiang, F. Chen, B. Tian, L. Li, Y.F. Zheng, D.V. Gunderov,R.Z. Valiev, Microstructure and martensitic transformation of an ultrafine-grained TiNiNb shape memory alloy processed by equal channel angularpressing, Intermetallics 49 (2014) 81e86.

[13] I. Karaman, A.V. Kulkarni, Z.P. Luo, Transformation behaviour and unusualtwinning in a NiTi shape memory alloy ausformed using equal channelangular extrusion, Philos. Mag. 85 (2005) 1729e1745.

[14] J. Song, L.M. Wang, X.N. Zhang, X.G. Sun, H. Jiang, Z.G. Fan, C.Y. Xie, M.H. Wu,Effects of second phases on mechanical properties and martensitic trans-formations of ECAPed TiNi and TieMo based shape memory alloys, Trans.Nonferr. Met. Soc. China 22 (2012) 1839e1848.

[15] G.L. Fan, W. Chen, S. Yang, J.H. Zhu, X.B. Ren, K. Otsuka, Origin of abnormalmulti-stage martensitic transformation behavior in aged Ni-rich Ti-Ni shapememory alloys, Acta Mater. 52 (2004) 4351e4362.

[16] B. Karbakhsh Ravari, M. Nishida, In situ SEM studies of the transformationsequence of multistage martensitic transformation in aged Ti-50.8 at.%Ni al-loys, Philos. Mag. 93 (2013) 2279e2296.

[17] X. Wang, K. Li, D. Schryvers, B. Verlinden, J. Van Humbeeck, R-phase transitionand related mechanical properties controlled by low-temperature agingtreatment in a Tie50.8at.% Ni thin wire, Scr. Mater. 72e73 (2014) 21e24.

[18] Z. Fan, C. Xie, Phase transformation behaviors of Ti-50.9 at.% Ni alloy afterequal channel angular extrusion, Mater. Lett. 62 (2008) 800e803.

[19] R.Z. Valiev, T.G. Langdon, Principles of equal-channel angular pressing as aprocessing tool for grain refinement, Prog. Mater. Sci. 51 (2006) 881e981.

[20] X. Wang, B. Verlinden, J. Van Humbeeck, Effect of aging temperature and timeon the transformation behavior of a Ti-50.8at.%Ni alloy with small grains,Mater. Today Proc. 2S (2015) S565eS568.

[21] K. Otsuka, X. Ren, Martensitic transformations in nonferrous shape memoryalloys, Mater. Sci. Eng. A 273e275 (1999) 89e105.

[22] J. Khalil-Allafi, A. Dlouhy, G. Eggeler, Ni4Ti3-precipitation during aging of NiTishape memory alloys and its influence on martensitic phase transformations,Acta Mater. 50 (2002) 4255e4274.

[23] K.S. Suresh, N.P. Gurao, S. Singh D, S. Suwas, K. Chattopadhyay,S.V. Zherebtsov, G.A. Salishchev, Effect of equal channel angular pressing ongrain refinement and texture evolution in a biomedical alloy Ti13Nb13Zr,

Y.X. Tong et al. / Intermetallics 85 (2017) 163e169 169

Mater. Charact. 82 (2013) 73e85.[24] D.T. Zhang, B. Guo, Y.X. Tong, B. Tian, L. Li, Y.F. Zheng, D. Gunderov, R. Valiev,

Effect of annealing temperature on martensitic transformation of Ti49.2Ni50.8alloy processed by equal channel angular pressing, Trans. Nonferr. Met. Soc.China 26 (2016) 448e455.

[25] S. Miyazaki, C.M. Wayman, The R-phase transition and associated shapememory mechanism in Ti-Ni single crystals, Acta Metall. 36 (1988) 181e192.

[26] Y.N. Liu, J.I. Kim, S. Miyazaki, Thermodynamic analysis of ageing-inducedmultiple-stage transformation behaviour of NiTi, Philos. Mag. 84 (2004)2083e2102.