- Email: [email protected]
Annealing behavior and shape memory effect in NiTi alloy processed by equal-channel angular pressing at room temperature
Materials Science & Engineering A 629 (2015) 16–22
Contents lists available at ScienceDirect
Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea
Annealing behavior and shape memory effect in NiTi alloy processed by equal-channel angular pressing at room temperature Hamed Shahmir a,n, Mahmoud Nili-Ahmadabadi a, Chuan Ting Wang b,c, Jai Myun Jung d, Hyoung Seop Kim d, Terence G. Langdon b,e a
School of Metallurgy and Materials, College of Engineering, University of Tehran, Tehran, Iran Departments of Aerospace & Mechanical Engineering and Materials Science, University of Southern California, Los Angeles, CA 90089-1453, USA c School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, PR China d Department of Materials Science and Engineering, Pohang University of Science and Technology, Pohang, Republic of Korea e Materials Research Group, Faculty of Engineering and the Environment, University of Southampton, Southampton SO17 1BJ, UK b
art ic l e i nf o
a b s t r a c t
Article history: Received 1 January 2015 Accepted 28 January 2015 Available online 4 February 2015
A martensitic NiTi shape memory alloy was processed successfully by equal-channel angular pressing (ECAP) for one pass at room temperature using a core–sheath billet design. The annealing behavior and shape memory effect of the ECAP specimens were studied followed by post-deformation annealing (PDA) at 673 K for various times. The recrystallization and structural evolution during annealing were investigated by differential scanning calorimetry, dilatometry, X-ray diffraction, transmission electron microscopy and microhardness measurements. The results indicate that the shape memory effect improves by PDA after ECAP processing. Annealing for 10 min gives a good shape memory effect which leads to a maximum in recoverable strain of 6.9 pct upon heating where this is more than a 25 pct improvement compared with the initial state. & 2015 Elsevier B.V. All rights reserved.
Keywords: Shape memory effect Equal-channel angular pressing NiTi shape memory alloys Ultrafine-grained materials
1. Introduction NiTi shape memory alloys have unique characteristics by exhibiting shape memory and superelasticity based on a thermoelastic martensitic transformation which has attracted much attention in engineering applications [1]. The thermoelastic transformation exhibits a crystallographically reversible transformation. Nevertheless, plastic deformation such as slip or deformation twinning is irreversible and these strains cannot be restored even upon heating. Thus, it is important to increase the critical stress for slip by work hardening and/or grain refinement in order to realize good shape memory and superelastic characteristics for these shape memory alloys [1–3]. It has been shown that cold-working followed by annealing leads to good superelasticity as well as a shape memory effect of up to 6 pct and no plastic strains are observed [1]. It is well known that grain refinement by severe plastic deformation (SPD) can improve the physical and mechanical properties of metals and alloys. Recent studies have shown that SPD processing at relatively low temperatures may be used effectively to synthesize bulk nanostructured NiTi alloys with
n
Corresponding author. Tel.: þ 98 2182084078; fax: þ98 2188006076. E-mail address: [email protected] (H. Shahmir).
http://dx.doi.org/10.1016/j.msea.2015.01.072 0921-5093/& 2015 Elsevier B.V. All rights reserved.
enhanced shape memory and superelasticity [4,5]. Processing by equal-channel angular pressing (ECAP) is generally considered superior to most other SPD techniques because it uses relatively large bulk samples and has other advantages such as simplicity in operation [6,7]. However, due to their low deformability it has proven almost impossible to successfully process NiTi alloys by ECAP at room temperature and therefore the processing is generally conducted at elevated temperatures [8–12]. It was reported that the processing of a martensitic NiTi alloy for one pass at room temperature leads to the formation of macro-shear bands due to austenite formation since this is the low formability phase. Furthermore, a sample deformed at room temperature followed by low temperature annealing gave the most promising strength and shape memory characteristics under compression, such that a 5.3 pct recovered strain was achieved at a strength level of 2200 MPa although the recovered strain decreased slightly by comparison with the as-received condition [8]. However, there are no reports to date of the shape memory characteristics under tension of the NiTi alloy after successfully processing by ECAP at room temperature followed by annealing. Very recently, a new billet design was introduced which permitted the successful processing by ECAP of NiTi alloys for up to two passes at room temperature using a conventional die design [13,14]. Accordingly, the present research was initiated in order to obtain a comprehensive description of the annealing
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
behavior and the shape memory effect of NiTi after equal-channel angular pressing at room temperature.
2. Experimental materials and procedures A NiTi alloy was used in these experiments having a nominal compositions of Ni-50.2 at% Ti. The experimental procedure for preparing the alloy was described earlier and involved a solution annealing at 1123 K for 60 min followed by ice-water quenching [13]. Samples were cut from the solution annealed material with wire electro-discharge machining to give rods having lengths of 40 mm and diameters of 3 mm. These samples were the core which fitted within Fe sheaths having diameters of 30 mm and lengths of 50 mm. The processing by ECAP was described in detail in earlier reports [13,14]. After one-pass of ECAP processing, postdeformation annealing (PDA) was performed at 673 K for various times from 5 to 300 min in a vacuum furnace. The heating rate of the specimens was 120 K min 1. The transformation temperatures were measured using differential scanning calorimetry (DSC) with a liquid nitrogen cooling accessory having cooling and heating rates of 10 K min 1 during the thermal cycling. The DSC analyses were performed using non-isothermal (scanning) experiments upon heating at a scanning rate of 10 K min 1. Cylindrical samples of 10 mm length and 2 mm diameter were cut and inserted in an Adamel DT1000 dilatometer to provide an isothermal study of the ECAP and the solution annealed samples. The samples were heated to 673 K in vacuum using a heating rate of 120 K min 1 and then maintained at this temperature for 60 min. X-ray diffraction (XRD) was used to study the phases with Cu Kα radiation at 40 kV and a tube current of 30 mA at room temperature. The XRD measurements were carried out over a 2θ range from 301 to 501 using a step size of 0.021 with a counting time of 9.6 s at each step. Measurements of the Vickers microhardness, Hv, were taken at the centers of the longitudinal sections of the billets parallel to the pressing direction, equivalent to the X direction. A load of 100 gf was applied for a dwell time of 10 s. Every point in the reported values of Hv was taken as the average of five separate hardness values. The ECAP core was also used to prepare foils for transmission electron microscopy (TEM) using focused ion beam (FIB). The different phases were analyzed by selected area diffraction applying different beam directions in a JEOL-2100 TEM operating at 200 kV. Stress–strain curves were recorded for studying the shape memory effect using tensile testing and gauge lengths of 8 mm measured parallel to the pressing direction as shown in Fig. 1. A Santam universal testing machine was used for the tensile testing with a load capacity of 2 kN and operating with a crosshead speed of 0.1 mm min 1 which is equivalent to an initial strain rate of 7.4 10 4 s 1. The strain recovery of the specimens was measured after loading to 6 and 8 pct strain followed by
Fig. 1. Longitudinal section of a billet after ECAP: Fe sheath 30 mm in diameter 50 mm in length and NiTi core 3 mm in diameter 40 mm in length. The white arrow shows the area for the tensile test specimen and also the area of DSC, dilatometric measurements, microhardness measurements and X-ray analysis.
17
unloading and heating up to 423 K by dipping in hot oil followed by ice-water quenching.
3. Experimental results Fig. 1 shows a longitudinal section of the billet including the NiTi core (3 mm in diameter and 40 mm in length) contained within the Fe sheath (30 mm in diameter and 50 mm in length) after one-pass by ECAP. It is readily apparent that the NiTi alloy was successfully processed by ECAP at room temperature by containing the NiTi sample within the Fe sheath. Processing was performed successfully for up to two passes of ECAP by controlling the processing variables as described in detail in an earlier report [13]. Non-isothermal DSC measurements of the Ni50.2Ti alloy after solution annealing and one-pass of ECAP are illustrated in Fig. 2. The exothermic peak is visible for the ECAP processed sample and this contrasts with the solution annealed sample and demonstrates energy storage during the severe plastic deformation. The appearance of the exothermic peak in the non-isothermal DSC measurements is related to the occurrence of recovery and recrystallization phenomena after cold working. This result shows that the recovery and recrystallization start at 598 K and the integral of the exothermic peak as a stored energy (Estored) is 540 J mol 1. The isothermal dilatometric measurements of Ni50.2Ti at 673 K after solution annealing and ECAP processing are given in Fig. 3. These measurements reveal that, after expansion of the cylindrical samples up to 673 K after 3 min, there is significant contraction as a result of recovery and recrystallization. Based on this evidence, it appears that the recrystallization is completed after 11 min. The occurrence of a small contraction for the solution annealed sample is probably a consequence of thermal shock during the high heating rate. The kinetics of isothermal recrystallization are usually expressed by the well-known Johnson–Mehl–Avrami–Kolmogorov (JMAK) equation [15,16] and through the isothermal study. Thus, it is possible to qualitatively assess the nature of the recrystallization and grain-growth phenomena [17,18]. Assuming the volume fraction of the transformed material is proportional to the dilatation measured under the dilatometric peak, as shown in Fig. 3a, the volume fractions (x) of the transformed material may be plotted against time, t, as shown in Fig. 3b. The Avrami plot of ln[ln (1 x) 1] versus ln(t) (with the time in s) then yields a straight
Fig. 2. Non-Isothermal (scanning) DSC measurement of Ni50.2Ti after solution annealing and ECAP processing: the heating rate is 10 K min 1.
18
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
Table 1 Microhardness measurements under different conditions. Condition SA
ECAP (one pass)
PDA at 673 K for various time (min) 5
Hv
210 75 3067 11
10
30
60
300
281 7 6 264 7 6 256 7 4 239 7 4 2117 5
Fig. 4. DSC curves of the Ni50.2Ti after solution annealing and ECAP followed by PDA at 673 K for 60 min.
Fig. 3. (a) Isothermal dilatometric measurement of Ni50.2Ti after solution annealing and ECAP processing at 673 K; (b) transformed volume fraction as a function of time; (c) the JMA plot from which the Avrami coefficient n is calculated.
line from which the slope, n, can be calculated as shown in Fig. 3c. Thus, the Avrami exponent of n ¼1.2 indicates diffusion-controlled grain growth with grain boundary nucleation [19]. The microhardness measurements are summarized in Table 1 at the centers on longitudinal sections of the processed NiTi cores after one pass of ECAP processing followed by annealing at 673 K for times from 5 to 300 min. Close inspection shows there is a significant increase in the hardness value with reference to the SA condition which is shown in the second column in Table 1. Thus, the hardness in the center of the core increases from the SA value
of HvE210 to a value of HvE306 after one pass of ECAP processing. The results also show that the hardness values decrease with increasing time of annealing and finally reach a hardness value similar to the SA condition of HvE 211 after 300 min. The DSC curves of the SA condition and the one pass ECAP condition followed by annealing at 673 K for 60 min are shown in Fig. 4. The DSC curve for the SA condition reveals one peak upon cooling and one peak upon heating that represent the austenite (B2)-to-martensite (B190 ) and martensite-to-austenite transformations, respectively. After solution annealing and ice-water quenching, the martensitic transformation start (Ms) and finish (Mf) temperatures were measured as 316 and 299 K, respectively, upon cooling. Conversely, upon heating the austenitic transformation start (As) and finish (Af) temperatures were 331 and 348 K, respectively. This shows the specimen after SA and subsequent ice-water quenching is fully martensitic. By comparison, the DSC curves for the one pass ECAP sample followed by annealing at 673 K for 60 min shows two broad and merged peaks corresponding to the transformation of austenite to the R phase and the R phase to martensite upon cooling and the two merged peaks correspond to the transformation of martensite to the R phase and the R phase to austenite upon heating. Accordingly, the Rs and Mf temperatures were 323 and 287 K, respectively, and the Rs and Af temperatures were 325 and 345 K, respectively. This result confirms that the structure after annealing and ice-water quenching, as in the SA specimen, is fully martensitic. The X-ray diffraction patterns of samples after SA and ECAP processing followed by annealing at 673 K for 0 to 60 min are represented in Fig. 5. The microstructure after SA and annealing after the severe plastic deformation is fully martensitic which supports the DSC results. The broad austenite phase peak is clearly revealed after processing by ECAP, thereby confirming the occurrence of the martensite-to-austenite transformation during ECAP processing which was identified and expressed in the earlier report [13]. Also, the peak broadening after ECAP with reference to the SA condition indicates energy storage that is in the direction of the non-isothermal DSC measurements. The R-phase characteristic
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
specimen shows a poor shape memory effect by comparison with the PDA conditions as shown in Table 2. In the latter conditions, the results demonstrate a fully recovered strain after annealing which characterizes a good shape memory effect after loading up to 6 pct strain (Fig. 7a). The characteristic detwinning plateau after the elastic region is revealed in the PDA condition which contrasts with the one pass ECAP condition. It appears that a martensite reorientation does not take place after the heavy deformation and the stress level decreases by increasing the annealing time from 0 to 30 min. These results indicate that the shape memory effect is improved significantly after PDA by comparison with the SA and ECAP condition. Specifically, the fully recovered strain at SA is 5.5 pct while for the PDA conditions this value is enhanced more than 6 pct. This increase is attributed to the strengthening of the martensite matrix and the grain refinement inherent in ECAP followed by PDA. The results demonstrate that annealing for 10 min is the optimum procedure with a maximum recovered strain of 6.9 pct.
t(min) R(112)
60
R(112)
cps
30
R(112)
10
A(110)
5
SA 37
38
39
40
41
M (111)
M (020)
M (111)
M (002)
M (110)
0
36
42
43
44
45
46
19
47
48
2-Theata Fig. 5. X-ray patterns of Ni50.2Ti after solution annealing, one-pass ECAP and PDA at 673 K for 5–60 min.
peak appears after annealing in the X-ray diffraction patterns and the results indicate that the intensity of the R phase peak decreases by increasing the annealing time and ultimately disappears after 60 min. Accordingly, the annealed specimens are fully martensitic and these include the B190 phase and a small amount of the R phase: in this respect it should be noted that the R-phase martensite shows shape memory behavior similar to B190 phase martensite [1]. It is important to note that the DSC result of PDA for 60 min given in Fig. 4 shows that a fully martensitic microstructure can be obtained by rapidly quenching to 273 K because this temperature is lower that Mf (287 K) and upon heating it follows that Rs (325 K) is higher than room temperature. Fig. 6 shows TEM images in (a) bright-field and (b) dark-field of a specimen after one pass of ECAP processing at room temperature and the inset shows the appropriate diffraction pattern. There are several spot sets which correspond to the diffraction patterns of the martensitic phase. Therefore, the structure is B190 martensite which is consistent with the dilatometric analysis and confirms the stabilization of martensite after SPD. The martensitic plates and contrast change because of the presence of dislocations and the thinner plates are probably formed by reorientation and distortion of the martensite variants. As can be seen from Fig. 6 (a), the martensite is generally very thin (o30 nm) and boundaries are visible but they are not sharp. Fig. 6(c–f) shows TEM images of samples annealed at 673 K for 10 and 30 min and it apparent that there are significant microstructural changes by comparison with the structure after processing by ECAP. The bright-field images and the corresponding diffraction patterns from these samples in Fig. 6(c) and (e) reveal the recrystallized structure, including martensite, where the boundaries of the martensitic plates are clearly visible. The recrystallization is also confirmed by the diffraction patterns which are characterized by the presence of various grains in different orientations. The stress–strain curves of the Ni50.2Ti alloy in the SA condition and after one pass of ECAP processing followed by annealing at 673 K for 0–30 min are represented in Fig. 7 and the characteristic parameters of the stress–strain curves are summarized in Table 2. The results indicate that the solution annealed, one pass ECAP and annealed specimens show strain recovery after loading and unloading followed by heating up to 423 K by dipping in hot oil and ice-water quenching. The plateau (detwinning) and recovered strain of the one pass ECAP specimen are negligible. The SA
4. Discussion For the shape memory effect, plastic deformation processes such as slip or deformation twinning are irreversible and these strains cannot be restored even upon heating. Thus, it is important to increase the critical stress for slip so that shape memory alloys may exhibit good shape memory and superelastic characteristics. Basically, an identification of the deformation mechanisms of the martensitic NiTi alloy using stress–strain curves leads to three distinct stages. The first stage is characterized by an initial linear region which is due to the elastic deformation of the martensite and detwinning or twinning of the martensite with a reorientation of the martensite variants. The second stage corresponds to detwinning/twinning and slip and the third stage has slip and twinning with the twinning in these stages characterized by irreversible twins [20]. In order to achieve a high critical stress for slip, work hardening and grain refinement may be useful. The increase of the critical stress is especially important for NiTi alloys because of the ease of introducing slip [21]. In fact, it is well known that no superelasticity appears and the shape memory effect is very poor in an ideal solution annealed condition, whereas a good shape memory effect and superelasticity effects are realized after the alloy has been subjected to a thermomechanical treatment because slip is easily introduced in the solution annealed condition [1]. Therefore, it is reasonable to anticipate that severe plastic deformation followed by optimum PDA will lead to the appropriate microstructure for a good shape memory effect. Accordingly, the annealing behavior and identifying the optimum PDA condition are important requirements in achieving a good shape memory condition. 4.1. Annealing behavior The non-isothermal analysis proposes that recovery and recrystallization start and finish at 598 and 753 K, respectively, and 673 K is the middle temperature. An isothermal analysis at 673 K indicates that recrystallization starts and finishes after 3.5 and 11 min, respectively, and grain growth starts after 11 min. The value obtained for the Avrami exponent is n¼ 1.2 for one pass of ECAP processing. Assuming random nucleation and constant and isotropic nucleus growth, the theoretical value of the exponent is n ¼4 if the nucleation rate is constant but it becomes n ¼3 for sitesaturated nucleation where all nuclei are generated at t ¼0. In most investigations on metals, the Avrami exponent n deviates from the theoretical predictions [22,23]. Heterogeneous nucleation,
20
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
012M
-1-11M
11-1M 10-1M
11-1M 001M -131M
140 nm
Fig. 6. TEM images (bright and dark filed) of the sample deformed by one pass ECAP at room temperature (a, b) and annealed at 673 K for 10 min (c, d) and 60 min (e, f); the sample deformed by one pass ECAP processing showing a complex and distorted microstructure including martensitic plates and contrast changes because of dislocations; the images (c–f) show a recrystallized microstructure with corresponding SAD patterns. The dark-field images correspond to (11 1)M.
anisotropic nucleus growth and varying growth rates contribute to deviations from n¼3. The unusually low values of the Avrami exponent in the present experiments derive mainly from the decreasing growth rate during recrystallization and the nonisotropic nucleus growth because of the inhomogeneous deformation structure [24]. Inhomogeneous deformation was reported earlier for NiTi after one pass of ECAP processing [14] and TEM results show inhomogeneities in the microstructure in Fig. 6(c). As shown in Fig. 3(a), the incubation time is almost zero and this behavior
suggests that the nucleation can be characterized as site-saturated nucleation [24,25]. It is important to note that recrystallization occurs at an annealing temperature as low as 673 K since the boundaries act as heterogeneous nucleation sites giving a high nucleation rate although the growth rate is low because the ratio of Ta/Tm is only 0.4 where Ta and Tm are the annealing and melting absolute temperatures, respectively. The TEM image after 10 min annealing in Fig. 6(c) reveals significant microstructural changes compared to the ECAP structure in Fig. 6 (a). The growth rate of
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
21
Fig. 7. Stress–strain curves of Ni50.2Ti after ECAP processing followed by various PDA at 673 K for 0–30 min: (a) elongation up to 6 pct and (b) elongation up to 8 pct after loading and unloading: the arrows show the strain recovery upon heating after unloading. The specimens after PDA and loading up to 6 pct are fully recovered upon heating.
Table 2 The characteristic parameters of the stress–strain curves after loading and unloading followed by heating to 423 K. Condition
Recovered strain (pct)
Stress at 6 pct (MPa)
SA ECAP PDA for 5 min PDA for 10 min PDA for 30 min
5.5 5.1 6.1 6.9 6.4
230 440 260 235 229
nuclei decreases with annealing time and this is primarily responsible for the low Avrami exponents. This decrease in the growth rate is due to a reduction in the driving force essentially because of recovery. Close inspection of Fig. 6(e) supports this conclusion and shows also that the microstructure after an anneal of 30 min has not changed significantly compared to the sample annealed for 30 min so that the widths of the martensitic plates are consistently 130 nm. In addition, the decrease in growth rate is due to a reduction in the driving force because of recovery. It is well known that severe cold working, such as one pass of ECAP which imposes a strain of εeq ¼1 [27], introduces defects such as dislocations and vacancies. The defects introduced by processing inhibit the martensitic transformation and lead to a stabilization of martensite in the martensitic state [28–32] and an increase in the thermal hysteresis (e.g. Af Mf). The PDA results in the martensitic transformation are able to regenerate and decrease the thermal hysteresis [26,33,34]. As shown elsewhere, this occurs with some changes, in particular the emergence of a two step transformation on cooling for low temperature annealings: B2R-B190 [35]. It has been established that such a transformation behavior of the cold-worked specimen, followed by annealing as in Fig. 4, corresponds to the recrystallization and growth of very small grains [26]. The transition from two peaks to one peak corresponds to the disappearance of the internal stresses in the recrystallized grains and to the decrease of the microhardness. In addition, the austenite-to-R-phase transformation is inhibited by a strong defect density and this transformation does not occur after cold work but after annealing [26] as depicted by the X-ray patterns after ECAP and PDA in Fig. 5. Therefore, the observations of R phase in the X-ray patterns and DSC results after PDA are consistent with the dilatometric results which demonstrate the occurrence of recrystallization and grain growth after annealing at 673 K.
The X-ray patterns of the PDA condition exhibit a minor R-phase peak after 5–30 min annealing at room temperature but no R-phase peak after 60 min of PDA. Also, the intensity of the R phase peak decreases by increasing the annealing time and disappears after 60 min. It is important to note that these specimens were cooled to 273 K by quenching in ice-water after PDA. This quenching leads to the formation of fully B190 martensite in the case of PDA for 60 min according to the DSC results in Fig. 4. Thus, the existence of a minor R-phase after 5–30 min annealing in the microstructure, and decreasing amounts of R phase by increasing the annealing time, indicate that Mf in these conditions is slightly lower than 273 K and this temperature is increased by increasing the annealing time. This argument is consistent with earlier research which showed a decrease in the thermal hysteresis and consequently an increase in Ms and Mf by increasing the annealing temperature or time after cold working [26,33,34]. 4.2. Improvement of the shape memory effect In this research the SA and PDA conditions are almost fully martensite and exhibit shape memory behavior. The best shape memory characteristic was for a PDA at 673 K for 10 min immediately after ECAP. It was reported earlier that annealing at 673 K after cold-working leads to the optimum superelasticity and shape memory characteristics up to 6 pct [1] and also a sample deformed by ECAP at room temperature followed by low temperature annealing gave a 5.3 pct recovered strain under compression [8]. Based on the dilatometric and XRD results, it appears that partial recrystallization occurs after 5 min annealing and this condition exhibits a smooth martensitic transformation upon cooling but still retains high strength due to the presence of rearranged dislocations. The high stress level in the stress–strain curve and the broad peak in the X-ray pattern for this condition are consistent with this interpretation. In fact, the dislocations may prevent twin boundary motion, which is equivalent to a conversion of martensite variants for showing a good shape memory effect. In addition, a fully martensitic microstructure is important in attaining the optimum shape memory effect. The existence of the R-phase together with the martensitic phase produces a diminution in the shape memory effect because the available shape memory strain for the R-phase transformation is very small with a maximum of 1% [1]. According to XRD results, it seems the amount of R-phase after 5 min annealing is maximum and it decreases by increasing the annealing time. By increasing the annealing time up to 10 min after ECAP processing, the TEM
22
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
images demonstrate that the specimens are fully recrystallized which leads to the formation of ultra-fine plate martensite (Fig. 6). The results show that annealing at 673 K for 10 min leads to a much superior shape memory effect compared to the annealed condition at 673 K for 5 and 30 min and probably also for longer times. However, the shape memory characteristics of PDA for 10 and 30 min are similar which means the microstructure does not change significantly after 30 min as confirmed by the TEM images, X-ray patterns and hardness measurements. It is expected there will be growth of grains by recrystallization after 10 min annealing and this produces a decrease in the recovered strain as a result of mechanical strength deterioration. This provides an explanation for the poor shape memory effect in the solution-treated specimens as recorded in Table 2. Finally, ECAP processing followed by the optimum PDA leads to an appropriate microstructure for the occurrence of a superior shape memory effect which is attributed to the strengthening of the martensitic matrix and grain refinement. Thus, the optimum procedure is annealing for 10 min after the ECAP processing to produce a maximum recovered strain of 6.8% with an improvement compared with the SA condition of more than 25%. 5. Summary and conclusions
demonstrate that all specimens recover 6 pct strain and the maximum recovered strain of 6.9 pct is related to a specimen annealed for 10 min.
Acknowledgments The authors thank Mr. Mojtaba Mansouri-Arani for kind help during the ECAP processing of the samples. This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea Government (MSIP) (No. 2014R1A2A1A10051322). The work of one of us was supported by the European Research Council under ERC Grant Agreement no. 267464-SPDMETALS (TGL). References [1] [2] [3] [4] [5] [6] [7] [8] [9]
1. A martensitic Ni50.28Ti alloy was successfully processed by ECAP at room temperature and the subsequent annealing behavior was studied using non-isothermal and isothermal analysis. The stored energy after one pass of ECAP was 540 J mol 1 and recovery/recrystallization start after 3.5 min and recrystallization finishes after 11 min by isothermal analysis at 673 K. 2. The hardness increases significantly from the SA value of HvE210 in Ni50.2Ti to a value of Hv E306 after one pass of ECAP processing. The hardness after annealing at 673 K indicates a significant decrease after 10 min and decreases gradually with increasing annealing time to ultimately reach the hardness of the solution annealed condition. 3. The DSC of one pass ECAP followed by annealing at 673 K for 60 min shows two broad merged peaks corresponding to the transformation of B2-R phase and R-B190 upon cooling and two merged peaks corresponding to a reverse transformation. The characteristic peak of the R phase appears after annealing in the X-ray diffraction patterns. The results indicate that the intensity of the R phase peak decreases by increasing the annealing time and disappears after 60 min annealing at 673 K. Accordingly, the specimens are fully martensitic after annealing and include a B190 phase and a minor amount of R phase. 4. The shape memory effect is improved significantly after PDA by comparison with the SA and ECAP condition. This is attributed to the strengthening of the martensite matrix and grain refinement using ECAP followed by PDA. The results
[10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35]
K. Otsuka, X. Ren, Prog. Mater. Sci. 50 (2005) 511–678. D. Wurzel, Mater. Sci. Eng. A 273–275 (1999) 634–638. E. Hornbogen, V. Mertinger, D. Wurzel, Scr. Mater. 44 (2001) 171–178. R.Z. Valiev, Nat. Mater. 3 (2004) 511–516. V.G. Pushin, V.V. Stolyarov, R.Z. Valiev, T.C. Lowe, Y.T. Zhu, Mater. Sci. Eng. A 410 (2005) 386–389. R.Z. Valiev, T.G. Langdon, Prog. Mater. Sci. 51 (2006) 881–981. T.G. Langdon, Acta Mater. 61 (2013) 7035–7059. I. Karaman, H.E. Karaca, H.J. Maier, Z.P. Lu, Metall. Mater. Trans. A 34A (2003) 2527–2539. V.G. Pushin, V.V. Stolyarov, R.Z. Valiev, T.C. Lowe, Y.T. Zhu, Mater. Sci. Eng. A 410–411 (2005) 386–389. Z. Fan, C. Xie, Mater. Lett. 62 (2008) 800–803. X. Zhang, B. Xia, J. Song, B. Chen, X. Tian, Y. Hao, C. Xie, J. Alloy. Compd. 509 (2011) 6296–6301. Y.X. Tong, B. Guo, F. Chen, B. Tian, L. Li, Y.F. Zheng, E.A. Prokofiev, D.V. Gunderov, R.Z. Valiev, Scr. Mater. 67 (2012) 1–4. H. Shahmir, M. Nili-Ahmadabadi, M. Mansouri-Arani, T.G. Langdon, Mater. Sci. Eng. A 576 (2013) 178–184. H. Shahmir, M. Nili-Ahmadabadi, M. Mansouri-Arani, A. Khajezade, T.G. Langdon, Adv. Eng. Mater. (2015), http://dx.doi.org/10.1002/adem.201400248. W.A. Johnson, R.F. Mehl, Trans. AIME 135 (1939) 416–458. J. Avrami, Chem. Phys. 9 (1941) 177–184. A.P. Zhilyaev, G.V. Nurislamova, S. Surinach, M.D. Baró, T.G. Langdon, Mater. Phys. Mech. 5 (2002) 23–30. L.C. Chen, F. Spaepen, J. Appl. Phys. 69 (1991) 679–688. R.C. Sharma, Phase Transformation in Materials, 1st edition, CBS, New Delhi, India, 2002. S. Miyazaki, K. Otsuka, Y. Suzuki., Scr. Metall. 15 (1981) 287–292. S. Miyazaki, Y. Ohmi, K. Otsuka, Y. Suzuki, J. Phys. 43 (1982) (C4-255–C4-260). R.A. Vandermeer, D. Juul Jensen, Acta Mater. 49 (2001) 2083–2094. M. Oyarzábal, A. Martínez-de-Guerenu, I. Gutiérrez, Mater. Sci. Eng. A 485 (2008) 200–209. Y. Lu, D.A. Molodov, G. Gottstein, Acta Mater. 59 (2011) 3229–3243. F.J. Humphreys, M. Hatherly, Recrystallization and Related Annealing Phenomena, Elsevier, Amsterdam, 2004. F. Khelfaoui, G. Guénin, Mater. Sci. Eng. A 355 (2003) 292–298. Y. Iwahashi, J. Wang, Z. Horita, M. Nemoto, T.G. Langdon, Scr. Mater. 35 (1996) 143–146. Y. Liu, Mater. Sci. Eng. A 273–275 (1999) 668–672. H.C. Lin, S.K. Wu, Acta Metall. Mater. 42 (1994) 1623–1630. Y. Liu, D. Favier, Acta Mater. 48 (2000) 3489–3499. H. Nakayama, K. Tsuchiya, Z.G. Liu, M. Umemoto, K. Morii, T. Shimizu, Mater. Trans. JIM 42 (2001) 1987–1993. Y. Liu, S.P. Galvin, Acta Mater. 45 (1997) 4431–4439. H. Shahmir, M. Nili-Ahmadabadi, F. Naghdi, Iran. J. Mater. Sci. Eng. 5 (2008) 25–31. H. Shahmir, M. Nili-Ahmadabadi, F. Naghdi, Mater. Des. 32 (2011) 365–370. H.C. Lin, S.K. Wu, Metall. Trans. A 24A (1993) 293–299.
Contents lists available at ScienceDirect
Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea
Annealing behavior and shape memory effect in NiTi alloy processed by equal-channel angular pressing at room temperature Hamed Shahmir a,n, Mahmoud Nili-Ahmadabadi a, Chuan Ting Wang b,c, Jai Myun Jung d, Hyoung Seop Kim d, Terence G. Langdon b,e a
School of Metallurgy and Materials, College of Engineering, University of Tehran, Tehran, Iran Departments of Aerospace & Mechanical Engineering and Materials Science, University of Southern California, Los Angeles, CA 90089-1453, USA c School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, PR China d Department of Materials Science and Engineering, Pohang University of Science and Technology, Pohang, Republic of Korea e Materials Research Group, Faculty of Engineering and the Environment, University of Southampton, Southampton SO17 1BJ, UK b
art ic l e i nf o
a b s t r a c t
Article history: Received 1 January 2015 Accepted 28 January 2015 Available online 4 February 2015
A martensitic NiTi shape memory alloy was processed successfully by equal-channel angular pressing (ECAP) for one pass at room temperature using a core–sheath billet design. The annealing behavior and shape memory effect of the ECAP specimens were studied followed by post-deformation annealing (PDA) at 673 K for various times. The recrystallization and structural evolution during annealing were investigated by differential scanning calorimetry, dilatometry, X-ray diffraction, transmission electron microscopy and microhardness measurements. The results indicate that the shape memory effect improves by PDA after ECAP processing. Annealing for 10 min gives a good shape memory effect which leads to a maximum in recoverable strain of 6.9 pct upon heating where this is more than a 25 pct improvement compared with the initial state. & 2015 Elsevier B.V. All rights reserved.
Keywords: Shape memory effect Equal-channel angular pressing NiTi shape memory alloys Ultrafine-grained materials
1. Introduction NiTi shape memory alloys have unique characteristics by exhibiting shape memory and superelasticity based on a thermoelastic martensitic transformation which has attracted much attention in engineering applications [1]. The thermoelastic transformation exhibits a crystallographically reversible transformation. Nevertheless, plastic deformation such as slip or deformation twinning is irreversible and these strains cannot be restored even upon heating. Thus, it is important to increase the critical stress for slip by work hardening and/or grain refinement in order to realize good shape memory and superelastic characteristics for these shape memory alloys [1–3]. It has been shown that cold-working followed by annealing leads to good superelasticity as well as a shape memory effect of up to 6 pct and no plastic strains are observed [1]. It is well known that grain refinement by severe plastic deformation (SPD) can improve the physical and mechanical properties of metals and alloys. Recent studies have shown that SPD processing at relatively low temperatures may be used effectively to synthesize bulk nanostructured NiTi alloys with
n
Corresponding author. Tel.: þ 98 2182084078; fax: þ98 2188006076. E-mail address: [email protected] (H. Shahmir).
http://dx.doi.org/10.1016/j.msea.2015.01.072 0921-5093/& 2015 Elsevier B.V. All rights reserved.
enhanced shape memory and superelasticity [4,5]. Processing by equal-channel angular pressing (ECAP) is generally considered superior to most other SPD techniques because it uses relatively large bulk samples and has other advantages such as simplicity in operation [6,7]. However, due to their low deformability it has proven almost impossible to successfully process NiTi alloys by ECAP at room temperature and therefore the processing is generally conducted at elevated temperatures [8–12]. It was reported that the processing of a martensitic NiTi alloy for one pass at room temperature leads to the formation of macro-shear bands due to austenite formation since this is the low formability phase. Furthermore, a sample deformed at room temperature followed by low temperature annealing gave the most promising strength and shape memory characteristics under compression, such that a 5.3 pct recovered strain was achieved at a strength level of 2200 MPa although the recovered strain decreased slightly by comparison with the as-received condition [8]. However, there are no reports to date of the shape memory characteristics under tension of the NiTi alloy after successfully processing by ECAP at room temperature followed by annealing. Very recently, a new billet design was introduced which permitted the successful processing by ECAP of NiTi alloys for up to two passes at room temperature using a conventional die design [13,14]. Accordingly, the present research was initiated in order to obtain a comprehensive description of the annealing
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
behavior and the shape memory effect of NiTi after equal-channel angular pressing at room temperature.
2. Experimental materials and procedures A NiTi alloy was used in these experiments having a nominal compositions of Ni-50.2 at% Ti. The experimental procedure for preparing the alloy was described earlier and involved a solution annealing at 1123 K for 60 min followed by ice-water quenching [13]. Samples were cut from the solution annealed material with wire electro-discharge machining to give rods having lengths of 40 mm and diameters of 3 mm. These samples were the core which fitted within Fe sheaths having diameters of 30 mm and lengths of 50 mm. The processing by ECAP was described in detail in earlier reports [13,14]. After one-pass of ECAP processing, postdeformation annealing (PDA) was performed at 673 K for various times from 5 to 300 min in a vacuum furnace. The heating rate of the specimens was 120 K min 1. The transformation temperatures were measured using differential scanning calorimetry (DSC) with a liquid nitrogen cooling accessory having cooling and heating rates of 10 K min 1 during the thermal cycling. The DSC analyses were performed using non-isothermal (scanning) experiments upon heating at a scanning rate of 10 K min 1. Cylindrical samples of 10 mm length and 2 mm diameter were cut and inserted in an Adamel DT1000 dilatometer to provide an isothermal study of the ECAP and the solution annealed samples. The samples were heated to 673 K in vacuum using a heating rate of 120 K min 1 and then maintained at this temperature for 60 min. X-ray diffraction (XRD) was used to study the phases with Cu Kα radiation at 40 kV and a tube current of 30 mA at room temperature. The XRD measurements were carried out over a 2θ range from 301 to 501 using a step size of 0.021 with a counting time of 9.6 s at each step. Measurements of the Vickers microhardness, Hv, were taken at the centers of the longitudinal sections of the billets parallel to the pressing direction, equivalent to the X direction. A load of 100 gf was applied for a dwell time of 10 s. Every point in the reported values of Hv was taken as the average of five separate hardness values. The ECAP core was also used to prepare foils for transmission electron microscopy (TEM) using focused ion beam (FIB). The different phases were analyzed by selected area diffraction applying different beam directions in a JEOL-2100 TEM operating at 200 kV. Stress–strain curves were recorded for studying the shape memory effect using tensile testing and gauge lengths of 8 mm measured parallel to the pressing direction as shown in Fig. 1. A Santam universal testing machine was used for the tensile testing with a load capacity of 2 kN and operating with a crosshead speed of 0.1 mm min 1 which is equivalent to an initial strain rate of 7.4 10 4 s 1. The strain recovery of the specimens was measured after loading to 6 and 8 pct strain followed by
Fig. 1. Longitudinal section of a billet after ECAP: Fe sheath 30 mm in diameter 50 mm in length and NiTi core 3 mm in diameter 40 mm in length. The white arrow shows the area for the tensile test specimen and also the area of DSC, dilatometric measurements, microhardness measurements and X-ray analysis.
17
unloading and heating up to 423 K by dipping in hot oil followed by ice-water quenching.
3. Experimental results Fig. 1 shows a longitudinal section of the billet including the NiTi core (3 mm in diameter and 40 mm in length) contained within the Fe sheath (30 mm in diameter and 50 mm in length) after one-pass by ECAP. It is readily apparent that the NiTi alloy was successfully processed by ECAP at room temperature by containing the NiTi sample within the Fe sheath. Processing was performed successfully for up to two passes of ECAP by controlling the processing variables as described in detail in an earlier report [13]. Non-isothermal DSC measurements of the Ni50.2Ti alloy after solution annealing and one-pass of ECAP are illustrated in Fig. 2. The exothermic peak is visible for the ECAP processed sample and this contrasts with the solution annealed sample and demonstrates energy storage during the severe plastic deformation. The appearance of the exothermic peak in the non-isothermal DSC measurements is related to the occurrence of recovery and recrystallization phenomena after cold working. This result shows that the recovery and recrystallization start at 598 K and the integral of the exothermic peak as a stored energy (Estored) is 540 J mol 1. The isothermal dilatometric measurements of Ni50.2Ti at 673 K after solution annealing and ECAP processing are given in Fig. 3. These measurements reveal that, after expansion of the cylindrical samples up to 673 K after 3 min, there is significant contraction as a result of recovery and recrystallization. Based on this evidence, it appears that the recrystallization is completed after 11 min. The occurrence of a small contraction for the solution annealed sample is probably a consequence of thermal shock during the high heating rate. The kinetics of isothermal recrystallization are usually expressed by the well-known Johnson–Mehl–Avrami–Kolmogorov (JMAK) equation [15,16] and through the isothermal study. Thus, it is possible to qualitatively assess the nature of the recrystallization and grain-growth phenomena [17,18]. Assuming the volume fraction of the transformed material is proportional to the dilatation measured under the dilatometric peak, as shown in Fig. 3a, the volume fractions (x) of the transformed material may be plotted against time, t, as shown in Fig. 3b. The Avrami plot of ln[ln (1 x) 1] versus ln(t) (with the time in s) then yields a straight
Fig. 2. Non-Isothermal (scanning) DSC measurement of Ni50.2Ti after solution annealing and ECAP processing: the heating rate is 10 K min 1.
18
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
Table 1 Microhardness measurements under different conditions. Condition SA
ECAP (one pass)
PDA at 673 K for various time (min) 5
Hv
210 75 3067 11
10
30
60
300
281 7 6 264 7 6 256 7 4 239 7 4 2117 5
Fig. 4. DSC curves of the Ni50.2Ti after solution annealing and ECAP followed by PDA at 673 K for 60 min.
Fig. 3. (a) Isothermal dilatometric measurement of Ni50.2Ti after solution annealing and ECAP processing at 673 K; (b) transformed volume fraction as a function of time; (c) the JMA plot from which the Avrami coefficient n is calculated.
line from which the slope, n, can be calculated as shown in Fig. 3c. Thus, the Avrami exponent of n ¼1.2 indicates diffusion-controlled grain growth with grain boundary nucleation [19]. The microhardness measurements are summarized in Table 1 at the centers on longitudinal sections of the processed NiTi cores after one pass of ECAP processing followed by annealing at 673 K for times from 5 to 300 min. Close inspection shows there is a significant increase in the hardness value with reference to the SA condition which is shown in the second column in Table 1. Thus, the hardness in the center of the core increases from the SA value
of HvE210 to a value of HvE306 after one pass of ECAP processing. The results also show that the hardness values decrease with increasing time of annealing and finally reach a hardness value similar to the SA condition of HvE 211 after 300 min. The DSC curves of the SA condition and the one pass ECAP condition followed by annealing at 673 K for 60 min are shown in Fig. 4. The DSC curve for the SA condition reveals one peak upon cooling and one peak upon heating that represent the austenite (B2)-to-martensite (B190 ) and martensite-to-austenite transformations, respectively. After solution annealing and ice-water quenching, the martensitic transformation start (Ms) and finish (Mf) temperatures were measured as 316 and 299 K, respectively, upon cooling. Conversely, upon heating the austenitic transformation start (As) and finish (Af) temperatures were 331 and 348 K, respectively. This shows the specimen after SA and subsequent ice-water quenching is fully martensitic. By comparison, the DSC curves for the one pass ECAP sample followed by annealing at 673 K for 60 min shows two broad and merged peaks corresponding to the transformation of austenite to the R phase and the R phase to martensite upon cooling and the two merged peaks correspond to the transformation of martensite to the R phase and the R phase to austenite upon heating. Accordingly, the Rs and Mf temperatures were 323 and 287 K, respectively, and the Rs and Af temperatures were 325 and 345 K, respectively. This result confirms that the structure after annealing and ice-water quenching, as in the SA specimen, is fully martensitic. The X-ray diffraction patterns of samples after SA and ECAP processing followed by annealing at 673 K for 0 to 60 min are represented in Fig. 5. The microstructure after SA and annealing after the severe plastic deformation is fully martensitic which supports the DSC results. The broad austenite phase peak is clearly revealed after processing by ECAP, thereby confirming the occurrence of the martensite-to-austenite transformation during ECAP processing which was identified and expressed in the earlier report [13]. Also, the peak broadening after ECAP with reference to the SA condition indicates energy storage that is in the direction of the non-isothermal DSC measurements. The R-phase characteristic
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
specimen shows a poor shape memory effect by comparison with the PDA conditions as shown in Table 2. In the latter conditions, the results demonstrate a fully recovered strain after annealing which characterizes a good shape memory effect after loading up to 6 pct strain (Fig. 7a). The characteristic detwinning plateau after the elastic region is revealed in the PDA condition which contrasts with the one pass ECAP condition. It appears that a martensite reorientation does not take place after the heavy deformation and the stress level decreases by increasing the annealing time from 0 to 30 min. These results indicate that the shape memory effect is improved significantly after PDA by comparison with the SA and ECAP condition. Specifically, the fully recovered strain at SA is 5.5 pct while for the PDA conditions this value is enhanced more than 6 pct. This increase is attributed to the strengthening of the martensite matrix and the grain refinement inherent in ECAP followed by PDA. The results demonstrate that annealing for 10 min is the optimum procedure with a maximum recovered strain of 6.9 pct.
t(min) R(112)
60
R(112)
cps
30
R(112)
10
A(110)
5
SA 37
38
39
40
41
M (111)
M (020)
M (111)
M (002)
M (110)
0
36
42
43
44
45
46
19
47
48
2-Theata Fig. 5. X-ray patterns of Ni50.2Ti after solution annealing, one-pass ECAP and PDA at 673 K for 5–60 min.
peak appears after annealing in the X-ray diffraction patterns and the results indicate that the intensity of the R phase peak decreases by increasing the annealing time and ultimately disappears after 60 min. Accordingly, the annealed specimens are fully martensitic and these include the B190 phase and a small amount of the R phase: in this respect it should be noted that the R-phase martensite shows shape memory behavior similar to B190 phase martensite [1]. It is important to note that the DSC result of PDA for 60 min given in Fig. 4 shows that a fully martensitic microstructure can be obtained by rapidly quenching to 273 K because this temperature is lower that Mf (287 K) and upon heating it follows that Rs (325 K) is higher than room temperature. Fig. 6 shows TEM images in (a) bright-field and (b) dark-field of a specimen after one pass of ECAP processing at room temperature and the inset shows the appropriate diffraction pattern. There are several spot sets which correspond to the diffraction patterns of the martensitic phase. Therefore, the structure is B190 martensite which is consistent with the dilatometric analysis and confirms the stabilization of martensite after SPD. The martensitic plates and contrast change because of the presence of dislocations and the thinner plates are probably formed by reorientation and distortion of the martensite variants. As can be seen from Fig. 6 (a), the martensite is generally very thin (o30 nm) and boundaries are visible but they are not sharp. Fig. 6(c–f) shows TEM images of samples annealed at 673 K for 10 and 30 min and it apparent that there are significant microstructural changes by comparison with the structure after processing by ECAP. The bright-field images and the corresponding diffraction patterns from these samples in Fig. 6(c) and (e) reveal the recrystallized structure, including martensite, where the boundaries of the martensitic plates are clearly visible. The recrystallization is also confirmed by the diffraction patterns which are characterized by the presence of various grains in different orientations. The stress–strain curves of the Ni50.2Ti alloy in the SA condition and after one pass of ECAP processing followed by annealing at 673 K for 0–30 min are represented in Fig. 7 and the characteristic parameters of the stress–strain curves are summarized in Table 2. The results indicate that the solution annealed, one pass ECAP and annealed specimens show strain recovery after loading and unloading followed by heating up to 423 K by dipping in hot oil and ice-water quenching. The plateau (detwinning) and recovered strain of the one pass ECAP specimen are negligible. The SA
4. Discussion For the shape memory effect, plastic deformation processes such as slip or deformation twinning are irreversible and these strains cannot be restored even upon heating. Thus, it is important to increase the critical stress for slip so that shape memory alloys may exhibit good shape memory and superelastic characteristics. Basically, an identification of the deformation mechanisms of the martensitic NiTi alloy using stress–strain curves leads to three distinct stages. The first stage is characterized by an initial linear region which is due to the elastic deformation of the martensite and detwinning or twinning of the martensite with a reorientation of the martensite variants. The second stage corresponds to detwinning/twinning and slip and the third stage has slip and twinning with the twinning in these stages characterized by irreversible twins [20]. In order to achieve a high critical stress for slip, work hardening and grain refinement may be useful. The increase of the critical stress is especially important for NiTi alloys because of the ease of introducing slip [21]. In fact, it is well known that no superelasticity appears and the shape memory effect is very poor in an ideal solution annealed condition, whereas a good shape memory effect and superelasticity effects are realized after the alloy has been subjected to a thermomechanical treatment because slip is easily introduced in the solution annealed condition [1]. Therefore, it is reasonable to anticipate that severe plastic deformation followed by optimum PDA will lead to the appropriate microstructure for a good shape memory effect. Accordingly, the annealing behavior and identifying the optimum PDA condition are important requirements in achieving a good shape memory condition. 4.1. Annealing behavior The non-isothermal analysis proposes that recovery and recrystallization start and finish at 598 and 753 K, respectively, and 673 K is the middle temperature. An isothermal analysis at 673 K indicates that recrystallization starts and finishes after 3.5 and 11 min, respectively, and grain growth starts after 11 min. The value obtained for the Avrami exponent is n¼ 1.2 for one pass of ECAP processing. Assuming random nucleation and constant and isotropic nucleus growth, the theoretical value of the exponent is n ¼4 if the nucleation rate is constant but it becomes n ¼3 for sitesaturated nucleation where all nuclei are generated at t ¼0. In most investigations on metals, the Avrami exponent n deviates from the theoretical predictions [22,23]. Heterogeneous nucleation,
20
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
012M
-1-11M
11-1M 10-1M
11-1M 001M -131M
140 nm
Fig. 6. TEM images (bright and dark filed) of the sample deformed by one pass ECAP at room temperature (a, b) and annealed at 673 K for 10 min (c, d) and 60 min (e, f); the sample deformed by one pass ECAP processing showing a complex and distorted microstructure including martensitic plates and contrast changes because of dislocations; the images (c–f) show a recrystallized microstructure with corresponding SAD patterns. The dark-field images correspond to (11 1)M.
anisotropic nucleus growth and varying growth rates contribute to deviations from n¼3. The unusually low values of the Avrami exponent in the present experiments derive mainly from the decreasing growth rate during recrystallization and the nonisotropic nucleus growth because of the inhomogeneous deformation structure [24]. Inhomogeneous deformation was reported earlier for NiTi after one pass of ECAP processing [14] and TEM results show inhomogeneities in the microstructure in Fig. 6(c). As shown in Fig. 3(a), the incubation time is almost zero and this behavior
suggests that the nucleation can be characterized as site-saturated nucleation [24,25]. It is important to note that recrystallization occurs at an annealing temperature as low as 673 K since the boundaries act as heterogeneous nucleation sites giving a high nucleation rate although the growth rate is low because the ratio of Ta/Tm is only 0.4 where Ta and Tm are the annealing and melting absolute temperatures, respectively. The TEM image after 10 min annealing in Fig. 6(c) reveals significant microstructural changes compared to the ECAP structure in Fig. 6 (a). The growth rate of
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
21
Fig. 7. Stress–strain curves of Ni50.2Ti after ECAP processing followed by various PDA at 673 K for 0–30 min: (a) elongation up to 6 pct and (b) elongation up to 8 pct after loading and unloading: the arrows show the strain recovery upon heating after unloading. The specimens after PDA and loading up to 6 pct are fully recovered upon heating.
Table 2 The characteristic parameters of the stress–strain curves after loading and unloading followed by heating to 423 K. Condition
Recovered strain (pct)
Stress at 6 pct (MPa)
SA ECAP PDA for 5 min PDA for 10 min PDA for 30 min
5.5 5.1 6.1 6.9 6.4
230 440 260 235 229
nuclei decreases with annealing time and this is primarily responsible for the low Avrami exponents. This decrease in the growth rate is due to a reduction in the driving force essentially because of recovery. Close inspection of Fig. 6(e) supports this conclusion and shows also that the microstructure after an anneal of 30 min has not changed significantly compared to the sample annealed for 30 min so that the widths of the martensitic plates are consistently 130 nm. In addition, the decrease in growth rate is due to a reduction in the driving force because of recovery. It is well known that severe cold working, such as one pass of ECAP which imposes a strain of εeq ¼1 [27], introduces defects such as dislocations and vacancies. The defects introduced by processing inhibit the martensitic transformation and lead to a stabilization of martensite in the martensitic state [28–32] and an increase in the thermal hysteresis (e.g. Af Mf). The PDA results in the martensitic transformation are able to regenerate and decrease the thermal hysteresis [26,33,34]. As shown elsewhere, this occurs with some changes, in particular the emergence of a two step transformation on cooling for low temperature annealings: B2R-B190 [35]. It has been established that such a transformation behavior of the cold-worked specimen, followed by annealing as in Fig. 4, corresponds to the recrystallization and growth of very small grains [26]. The transition from two peaks to one peak corresponds to the disappearance of the internal stresses in the recrystallized grains and to the decrease of the microhardness. In addition, the austenite-to-R-phase transformation is inhibited by a strong defect density and this transformation does not occur after cold work but after annealing [26] as depicted by the X-ray patterns after ECAP and PDA in Fig. 5. Therefore, the observations of R phase in the X-ray patterns and DSC results after PDA are consistent with the dilatometric results which demonstrate the occurrence of recrystallization and grain growth after annealing at 673 K.
The X-ray patterns of the PDA condition exhibit a minor R-phase peak after 5–30 min annealing at room temperature but no R-phase peak after 60 min of PDA. Also, the intensity of the R phase peak decreases by increasing the annealing time and disappears after 60 min. It is important to note that these specimens were cooled to 273 K by quenching in ice-water after PDA. This quenching leads to the formation of fully B190 martensite in the case of PDA for 60 min according to the DSC results in Fig. 4. Thus, the existence of a minor R-phase after 5–30 min annealing in the microstructure, and decreasing amounts of R phase by increasing the annealing time, indicate that Mf in these conditions is slightly lower than 273 K and this temperature is increased by increasing the annealing time. This argument is consistent with earlier research which showed a decrease in the thermal hysteresis and consequently an increase in Ms and Mf by increasing the annealing temperature or time after cold working [26,33,34]. 4.2. Improvement of the shape memory effect In this research the SA and PDA conditions are almost fully martensite and exhibit shape memory behavior. The best shape memory characteristic was for a PDA at 673 K for 10 min immediately after ECAP. It was reported earlier that annealing at 673 K after cold-working leads to the optimum superelasticity and shape memory characteristics up to 6 pct [1] and also a sample deformed by ECAP at room temperature followed by low temperature annealing gave a 5.3 pct recovered strain under compression [8]. Based on the dilatometric and XRD results, it appears that partial recrystallization occurs after 5 min annealing and this condition exhibits a smooth martensitic transformation upon cooling but still retains high strength due to the presence of rearranged dislocations. The high stress level in the stress–strain curve and the broad peak in the X-ray pattern for this condition are consistent with this interpretation. In fact, the dislocations may prevent twin boundary motion, which is equivalent to a conversion of martensite variants for showing a good shape memory effect. In addition, a fully martensitic microstructure is important in attaining the optimum shape memory effect. The existence of the R-phase together with the martensitic phase produces a diminution in the shape memory effect because the available shape memory strain for the R-phase transformation is very small with a maximum of 1% [1]. According to XRD results, it seems the amount of R-phase after 5 min annealing is maximum and it decreases by increasing the annealing time. By increasing the annealing time up to 10 min after ECAP processing, the TEM
22
H. Shahmir et al. / Materials Science & Engineering A 629 (2015) 16–22
images demonstrate that the specimens are fully recrystallized which leads to the formation of ultra-fine plate martensite (Fig. 6). The results show that annealing at 673 K for 10 min leads to a much superior shape memory effect compared to the annealed condition at 673 K for 5 and 30 min and probably also for longer times. However, the shape memory characteristics of PDA for 10 and 30 min are similar which means the microstructure does not change significantly after 30 min as confirmed by the TEM images, X-ray patterns and hardness measurements. It is expected there will be growth of grains by recrystallization after 10 min annealing and this produces a decrease in the recovered strain as a result of mechanical strength deterioration. This provides an explanation for the poor shape memory effect in the solution-treated specimens as recorded in Table 2. Finally, ECAP processing followed by the optimum PDA leads to an appropriate microstructure for the occurrence of a superior shape memory effect which is attributed to the strengthening of the martensitic matrix and grain refinement. Thus, the optimum procedure is annealing for 10 min after the ECAP processing to produce a maximum recovered strain of 6.8% with an improvement compared with the SA condition of more than 25%. 5. Summary and conclusions
demonstrate that all specimens recover 6 pct strain and the maximum recovered strain of 6.9 pct is related to a specimen annealed for 10 min.
Acknowledgments The authors thank Mr. Mojtaba Mansouri-Arani for kind help during the ECAP processing of the samples. This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea Government (MSIP) (No. 2014R1A2A1A10051322). The work of one of us was supported by the European Research Council under ERC Grant Agreement no. 267464-SPDMETALS (TGL). References [1] [2] [3] [4] [5] [6] [7] [8] [9]
1. A martensitic Ni50.28Ti alloy was successfully processed by ECAP at room temperature and the subsequent annealing behavior was studied using non-isothermal and isothermal analysis. The stored energy after one pass of ECAP was 540 J mol 1 and recovery/recrystallization start after 3.5 min and recrystallization finishes after 11 min by isothermal analysis at 673 K. 2. The hardness increases significantly from the SA value of HvE210 in Ni50.2Ti to a value of Hv E306 after one pass of ECAP processing. The hardness after annealing at 673 K indicates a significant decrease after 10 min and decreases gradually with increasing annealing time to ultimately reach the hardness of the solution annealed condition. 3. The DSC of one pass ECAP followed by annealing at 673 K for 60 min shows two broad merged peaks corresponding to the transformation of B2-R phase and R-B190 upon cooling and two merged peaks corresponding to a reverse transformation. The characteristic peak of the R phase appears after annealing in the X-ray diffraction patterns. The results indicate that the intensity of the R phase peak decreases by increasing the annealing time and disappears after 60 min annealing at 673 K. Accordingly, the specimens are fully martensitic after annealing and include a B190 phase and a minor amount of R phase. 4. The shape memory effect is improved significantly after PDA by comparison with the SA and ECAP condition. This is attributed to the strengthening of the martensite matrix and grain refinement using ECAP followed by PDA. The results
[10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35]
K. Otsuka, X. Ren, Prog. Mater. Sci. 50 (2005) 511–678. D. Wurzel, Mater. Sci. Eng. A 273–275 (1999) 634–638. E. Hornbogen, V. Mertinger, D. Wurzel, Scr. Mater. 44 (2001) 171–178. R.Z. Valiev, Nat. Mater. 3 (2004) 511–516. V.G. Pushin, V.V. Stolyarov, R.Z. Valiev, T.C. Lowe, Y.T. Zhu, Mater. Sci. Eng. A 410 (2005) 386–389. R.Z. Valiev, T.G. Langdon, Prog. Mater. Sci. 51 (2006) 881–981. T.G. Langdon, Acta Mater. 61 (2013) 7035–7059. I. Karaman, H.E. Karaca, H.J. Maier, Z.P. Lu, Metall. Mater. Trans. A 34A (2003) 2527–2539. V.G. Pushin, V.V. Stolyarov, R.Z. Valiev, T.C. Lowe, Y.T. Zhu, Mater. Sci. Eng. A 410–411 (2005) 386–389. Z. Fan, C. Xie, Mater. Lett. 62 (2008) 800–803. X. Zhang, B. Xia, J. Song, B. Chen, X. Tian, Y. Hao, C. Xie, J. Alloy. Compd. 509 (2011) 6296–6301. Y.X. Tong, B. Guo, F. Chen, B. Tian, L. Li, Y.F. Zheng, E.A. Prokofiev, D.V. Gunderov, R.Z. Valiev, Scr. Mater. 67 (2012) 1–4. H. Shahmir, M. Nili-Ahmadabadi, M. Mansouri-Arani, T.G. Langdon, Mater. Sci. Eng. A 576 (2013) 178–184. H. Shahmir, M. Nili-Ahmadabadi, M. Mansouri-Arani, A. Khajezade, T.G. Langdon, Adv. Eng. Mater. (2015), http://dx.doi.org/10.1002/adem.201400248. W.A. Johnson, R.F. Mehl, Trans. AIME 135 (1939) 416–458. J. Avrami, Chem. Phys. 9 (1941) 177–184. A.P. Zhilyaev, G.V. Nurislamova, S. Surinach, M.D. Baró, T.G. Langdon, Mater. Phys. Mech. 5 (2002) 23–30. L.C. Chen, F. Spaepen, J. Appl. Phys. 69 (1991) 679–688. R.C. Sharma, Phase Transformation in Materials, 1st edition, CBS, New Delhi, India, 2002. S. Miyazaki, K. Otsuka, Y. Suzuki., Scr. Metall. 15 (1981) 287–292. S. Miyazaki, Y. Ohmi, K. Otsuka, Y. Suzuki, J. Phys. 43 (1982) (C4-255–C4-260). R.A. Vandermeer, D. Juul Jensen, Acta Mater. 49 (2001) 2083–2094. M. Oyarzábal, A. Martínez-de-Guerenu, I. Gutiérrez, Mater. Sci. Eng. A 485 (2008) 200–209. Y. Lu, D.A. Molodov, G. Gottstein, Acta Mater. 59 (2011) 3229–3243. F.J. Humphreys, M. Hatherly, Recrystallization and Related Annealing Phenomena, Elsevier, Amsterdam, 2004. F. Khelfaoui, G. Guénin, Mater. Sci. Eng. A 355 (2003) 292–298. Y. Iwahashi, J. Wang, Z. Horita, M. Nemoto, T.G. Langdon, Scr. Mater. 35 (1996) 143–146. Y. Liu, Mater. Sci. Eng. A 273–275 (1999) 668–672. H.C. Lin, S.K. Wu, Acta Metall. Mater. 42 (1994) 1623–1630. Y. Liu, D. Favier, Acta Mater. 48 (2000) 3489–3499. H. Nakayama, K. Tsuchiya, Z.G. Liu, M. Umemoto, K. Morii, T. Shimizu, Mater. Trans. JIM 42 (2001) 1987–1993. Y. Liu, S.P. Galvin, Acta Mater. 45 (1997) 4431–4439. H. Shahmir, M. Nili-Ahmadabadi, F. Naghdi, Iran. J. Mater. Sci. Eng. 5 (2008) 25–31. H. Shahmir, M. Nili-Ahmadabadi, F. Naghdi, Mater. Des. 32 (2011) 365–370. H.C. Lin, S.K. Wu, Metall. Trans. A 24A (1993) 293–299.