Effect of Cold Drawing and Annealing in Thermomechanical Treatment Route on the Microstructure and Functional Properties of Superelastic Ti-Zr-Nb Alloy

In this study, a superelastic Ti-18Zr-15Nb (at. %) alloy was subjected to thermomechanical treatment, including cold rotary forging, intermediate annealing, cold drawing, post-deformation annealing, and additional low-temperature aging. As a result of intermediate annealing, two structures of β-phase were obtained: a fine-grained structure (d ≈ 3 µm) and a coarse-grained structure (d ≈ 11 µm). Cold drawing promotes grain elongation in the drawing direction; in a fine-grained state, grains form with a size of 4 × 2 µm, and in a coarse-grained state, they grow with a size of 16 × 6 µm. Post-deformation annealing (PDA) at 550 °C for 30 min leads to grain sizes of 5 µm and 3 µm, respectively. After PDA at 550 °C (30 min) in the fine-grained state, the wire exhibits high tensile strength (UTS = 624 MPa), highest elongation to failure (δ ≥ 8%), and maximum difference between the dislocation and transformation yield stresses, as well as the highest superelastic recovery strain (εrSE ≥ 3.3%) and total elastic + superelastic recovery strain (εrel+SE ≥ 5.4%). Additional low-temperature aging at 300 °C for 30–180 min leads to ω-phase formation, alloy hardening, embrittlement, and a significant decrease in superelastic recovery strain.


Investigated Alloy and Its Processing
The object of this study is a Ti-18Zr-15Nb (at. %) SMA, obtained by vacuum arc melting and isostatic pressing (900 • C, 100 MPa, 2 h). After hot isostatic pressing, the ingot was cooled in air and subjected to radial-shear rolling (RSR) at 950 • C and rotary forging (RF) at 700 • C using an RKM-2 forging machine to obtain 5 mm diameter rods. Then, cold rotary forging (CRF) at RT with an accumulated strain e = 1.5 was carried out. To form a wide range of structural states after cold rotary forging, annealing was carried out at temperatures from 600 to 750 • C (5-120 min) in an air furnace. The evolution of grain structure after all mentioned TMTs is presented in Figure 1, and the measured sizes of the structural elements after annealing are presented in Figure 2.

Investigated Alloy and Its Processing
The object of this study is a Ti-18Zr-15Nb (at. %) SMA, obtained by vacuum arc melting and isostatic pressing (900 °C, 100 MPa, 2 h). After hot isostatic pressing, the ingot was cooled in air and subjected to radial-shear rolling (RSR) at 950 °C and rotary forging (RF) at 700 °C using an RKM-2 forging machine to obtain 5 mm diameter rods. Then, cold rotary forging (CRF) at RT with an accumulated strain e = 1.5 was carried out. To form a wide range of structural states after cold rotary forging, annealing was carried out at temperatures from 600 to 750 °C (5-120 min) in an air furnace. The evolution of grain structure after all mentioned TMTs is presented in Figure 1, and the measured sizes of the structural elements after annealing are presented in Figure 2.  In the initial state, after hot RF, a homogeneous grain structure is formed, which is a mixture of equiaxed grains with an average size of 30 µm. Subsequent CRF leads to grain elongation in the material extraction direction (ED). PDAs at 600 °C (5-120 min) form a grain size in the range of 3 to 11 µm. After PDA at 700 °C (30-60 min), grain size increases up to 19-32 µm, depending to the annealing time. At an annealing temperature of 750 °C for 30 min, a grain size of ≈31 µm was obtained, and for 60 min, a grain size of ≈ 38 µm was obtained. In all cases, equiaxed grains are observed and a recrystallized structure is obtained. To assess the effect of initial grain size on the properties of the Ti-Zr-Nb alloy, three different structural states were chosen: 1. 600 °C, 5 min (small grain size of ≈3 µm); 2. 600 °C, 120 min (intermediate grain size of ≈11 µm); 3. 750 °C, 30 min (large grain size of ≈31 µm).
In this work, at first it was necessary to understand how CD affects the Ti-18Zr-15Nb alloy. It was decided to draw the wire until it starts to break. In the fine-grained (FG) state, after PDA at 600 °C (5 min), the wire broke at strain e = 0.8. In the coarse-grained (CG) state, after PDA at 600 °C (120 min), the wire broke at strain e = 1.0. The surface of the rod had the highest degree of defect after PDA at 750 °C (30 min). High-temperature annealing led to the formation of oxide layers, cracks formed on the wire, the die 2.5 mm did not shrink, and so the drawing did not start. Such an effect of RF at 800 °C was previously reported in [12].
The obtained wire was annealed at 550 °C (30 min) in a protective argon atmosphere to form a recrystallized fine-grained structure, which provides an optimal combination of properties [12]. Additional low-temperature aging at 300 °C for 30, 60, and 180 min was applied to the wire to estimate the effect of the ω-phase on the mechanical and functional properties of the Ti-Zr-Nb alloy [38,39]. The schedule of thermomechanical treatment is shown in Figure 3. In the initial state, after hot RF, a homogeneous grain structure is formed, which is a mixture of equiaxed grains with an average size of 30 µm. Subsequent CRF leads to grain elongation in the material extraction direction (ED). PDAs at 600 • C (5-120 min) form a grain size in the range of 3 to 11 µm. After PDA at 700 • C (30-60 min), grain size increases up to 19-32 µm, depending to the annealing time. At an annealing temperature of 750 • C for 30 min, a grain size of ≈31 µm was obtained, and for 60 min, a grain size of ≈38 µm was obtained. In all cases, equiaxed grains are observed and a recrystallized structure is obtained. To assess the effect of initial grain size on the properties of the Ti-Zr-Nb alloy, three different structural states were chosen: In this work, at first it was necessary to understand how CD affects the Ti-18Zr-15Nb alloy. It was decided to draw the wire until it starts to break. In the fine-grained (FG) state, after PDA at 600 • C (5 min), the wire broke at strain e = 0.8. In the coarse-grained (CG) state, after PDA at 600 • C (120 min), the wire broke at strain e = 1.0. The surface of the rod had the highest degree of defect after PDA at 750 • C (30 min). High-temperature annealing led to the formation of oxide layers, cracks formed on the wire, the die 2.5 mm did not shrink, and so the drawing did not start. Such an effect of RF at 800 • C was previously reported in [12].
The obtained wire was annealed at 550 • C (30 min) in a protective argon atmosphere to form a recrystallized fine-grained structure, which provides an optimal combination of properties [12]. Additional low-temperature aging at 300 • C for 30, 60, and 180 min was applied to the wire to estimate the effect of the ω-phase on the mechanical and functional properties of the Ti-Zr-Nb alloy [38,39]. The schedule of thermomechanical treatment is shown in Figure 3.

Experimental Procedure
For microstructure and phase composition study specimens were grinded and polished using a "SAPHIR 560". A longitudinal section of the surface of the specimens from a 10 mm long wire was subjected to grinding on abrasive paper with a suspension of Eposil F based on silicon oxide. Ammonia solutions, hydrogen peroxide, and liquid soap were added to the suspension during polishing. Then, the specimens were cleaned in an ultrasonic bath with isopropyl alcohol. The surface was etched in a 1HF:3HNO 3

Experimental Procedure
For microstructure and phase composition study specimens were grinded and polished using a "SAPHIR 560". A longitudinal section of the surface of the specimens from a 10 mm long wire was subjected to grinding on abrasive paper with a suspension of Eposil F based on silicon oxide. Ammonia solutions, hydrogen peroxide, and liquid soap were added to the suspension during polishing. Then, the specimens were cleaned in an ultrasonic bath with isopropyl alcohol. The surface was etched in a 1HF:3HNO3:6H2O solution for 20-60 s.
The grain structure was studied using a "Versamet-2 Union" optical microscope. The true average grain size of the β-phase was measured using the random linear intercept method [40]. The phase composition was studied by X-ray diffraction analysis using a "Rigaku Ultima IV" (Tokyo, Japan) diffractometer at RT using Cu-Kα radiation, as well as a parallel beam and graphite monochromator in the 30 to 90 deg 2θ range. For X-ray diffraction analysis, 10 mm long specimens were used in an amount of 6-7 pieces for each process so that when the specimens were located next to each other, the total thickness of the specimens was ≈10 mm.
For microstructure and crystallographic texture investigations, a "TESCAN VEGA LMH" scanning electron microscope (SEM) equipped with an electron backscatter diffraction (EBSD) unit, namely the "NordlysMax2" detector (Oxford Instruments Advanced AZ-tecEnergy), was used. Samples were prepared by grinding and mirror-smooth-finish polishing using a "SAPHIR 560". Specimens were tilted by 70° and scanned at 20 kV with a 0.5 µm step.
Static tensile testing to failure was carried out at RT and at a strain rate of 0.02 s −1 using an "Instron 5966" universal testing machine on 30 mm working length wire The grain structure was studied using a "Versamet-2 Union" optical microscope. The true average grain size of the β-phase was measured using the random linear intercept method [40]. The phase composition was studied by X-ray diffraction analysis using a "Rigaku Ultima IV" (Tokyo, Japan) diffractometer at RT using Cu-K α radiation, as well as a parallel beam and graphite monochromator in the 30 to 90 deg 2θ range. For X-ray diffraction analysis, 10 mm long specimens were used in an amount of 6-7 pieces for each process so that when the specimens were located next to each other, the total thickness of the specimens was ≈10 mm.
For microstructure and crystallographic texture investigations, a "TESCAN VEGA LMH" scanning electron microscope (SEM) equipped with an electron backscatter diffraction (EBSD) unit, namely the "NordlysMax2" detector (Oxford Instruments Advanced AZtecEnergy), was used. Samples were prepared by grinding and mirror-smooth-finish polishing using a "SAPHIR 560". Specimens were tilted by 70 • and scanned at 20 kV with a 0.5 µm step.
Static tensile testing to failure was carried out at RT and at a strain rate of 0.02 s −1 using an "Instron 5966" universal testing machine on 30 mm working length wire specimens. All the measurements were carried out using at least three specimens. From the stress-strain diagrams obtained from static tensile tests, relative elongation to failure (δ), yield stress (σ 0.2 ), transformation yield stress (σ tr ), dislocation yield stress (σ dis ), and ultimate tensile strength (UTS) were determined as in [17]. Superelastic cyclic testing was carried out at RT on the same type of specimens according to the scheme "loading the specimen to 1% deformation in the first cycle with a strain increase by 1% in each subsequent cycle for a total strain of 18%". The superelastic strain recovered upon unloading due to reverse β→α transformation (ε r SE ); the elastic strain recovered upon unloading (ε el ), and their combination (ε r el+SE ) was measured from the cyclic stress-strain diagrams, obtained from superelastic cyclic testing. Moreover, according to these diagrams, the accumulated residual strains ε acc , σ tr , and σ dis were determined ( Figure 4). ultimate tensile strength (UTS) were determined as in [17]. Superelastic cycl carried out at RT on the same type of specimens according to the scheme specimen to 1% deformation in the first cycle with a strain increase by 1% i quent cycle for a total strain of 18%". The superelastic strain recovered up due to reverse β→α" transformation (εr SE ); the elastic strain recovered upon u and their combination (εr el+SE ) was measured from the cyclic stress-strain d tained from superelastic cyclic testing. Moreover, according to these diagra mulated residual strains εacc, σtr, and σdis were determined ( Figure 4). The martensitic transformation start temperature (Ms) of the Ti-18Zr-15 obtained using the electrical resistivity, measured upon cooling from RT to − on the change in the electrical resistivity during the β→α" transformation. ohmmeter-voltmeter method with a four-point connection scheme was use meter was replaced by voltmeter and a reference resistor of 0.1 Ω and 100 W tions in situ measuring and recording the voltage of the specimen, reference K-type thermocouple were performed using a "PRIST V7-78/1" millivoltme channel extension board connected to a PC. A laboratory "GW Instek SPS supply was used as the current source with I = 0.5 A. The specimen workin of 70 mm with a diameter of 1.5 and 1.7 mm. The samples were cooled in ni at a rate of 5 °C/min using a custom thermal chamber with programm TRM151" controller.

Structure-Phase State
The images of the grain structure of the wire after CD and subsequent P perature of 550 °C are shown in Figure 5. CD contributes to the elongation of drawing direction (DD). The average grain size in two dimensions (in the DD dicular to the DD) is 4 × 2 µm after FG+CD and 16 × 6 µm after CG+CD (Figu PDA at 550 °C for 30 min, the grain size decreases, and the average size i FG+CD+550, 30 ( Figure 5c) and ~3 µm after CG+CD+550, 30 (Figure 5d). I partial recrystallization occurs, while the dislocation substructure is also served. Since there was a finer grain structure after FG+CD, and thus accum The martensitic transformation start temperature (M s ) of the Ti-18Zr-15Nb alloy was obtained using the electrical resistivity, measured upon cooling from RT to −150 • C based on the change in the electrical resistivity during the β→α transformation. The modified ohmmeter-voltmeter method with a four-point connection scheme was used, where ammeter was replaced by voltmeter and a reference resistor of 0.1 Ω and 100 W. The simulations in situ measuring and recording the voltage of the specimen, reference resistor, and K-type thermocouple were performed using a "PRIST V7-78/1" millivoltmeter with a 20-channel extension board connected to a PC. A laboratory "GW Instek SPS-3610" power supply was used as the current source with I = 0.5 A. The specimen working length was of 70 mm with a diameter of 1.5 and 1.7 mm. The samples were cooled in nitrogen vapor at a rate of 5 • C/min using a custom thermal chamber with programmable "OVEN TRM151" controller.

Structure-Phase State
The images of the grain structure of the wire after CD and subsequent PDA at a temperature of 550 • C are shown in Figure 5. CD contributes to the elongation of grains in the drawing direction (DD). The average grain size in two dimensions (in the DD and perpendicular to the DD) is 4 × 2 µm after FG+CD and 16 × 6 µm after CG+CD (Figure 5a,b). After PDA at 550 • C for 30 min, the grain size decreases, and the average size is~5 µm after FG+CD+550, 30 (Figure 5c) and~3 µm after CG+CD+550, 30 ( Figure 5d). In both cases, partial recrystallization occurs, while the dislocation substructure is also partially preserved. Since there was a finer grain structure after FG+CD, and thus accumulated strain energy and grain-boundary energy was higher, the recrystallization process in this case after annealing is faster. Furthermore, after FG+CD+550, 30, a more rapid growth of recrystallized grains occurs, so the grain size in this route is larger [41]. energy and grain-boundary energy was higher, the recrystallization process in this case after annealing is faster. Furthermore, after FG+CD+550, 30, a more rapid growth of recrystallized grains occurs, so the grain size in this route is larger [41].  Figure 6a shows X-ray diffractograms of the Ti-Zr-Nb alloy after CRF followed by intermediate annealing, CD, and post-deformation annealing at 550 °C (30 min). As a result of X-ray diffraction analysis, it was revealed that in all cases, the main phase component is the BCC β-phase. After CD, some distinct X-ray lines of stress-induced α"-martensite are also present. After post-deformation annealing, only β-phase X-ray lines are observed. After low-temperature aging at 300 °C for 30, 60, and 180 min, the main phase is also the β-phase while it is accompanied by a ω-phase (Figure 6b). With an increase in the duration of low-temperature aging, the amount of the ω-phase increases too, which is consistent with the results of [42]. Aging for 180 min leads to appearance of a small amount of the α˗phase.  Figure 6a shows X-ray diffractograms of the Ti-Zr-Nb alloy after CRF followed by intermediate annealing, CD, and post-deformation annealing at 550 • C (30 min). As a result of X-ray diffraction analysis, it was revealed that in all cases, the main phase component is the BCC β-phase. After CD, some distinct X-ray lines of stress-induced α -martensite are also present. After post-deformation annealing, only β-phase X-ray lines are observed. After low-temperature aging at 300 • C for 30, 60, and 180 min, the main phase is also the β-phase while it is accompanied by a ω-phase (Figure 6b). With an increase in the duration of low-temperature aging, the amount of the ω-phase increases too, which is consistent with the results of [42]. Aging for 180 min leads to appearance of a small amount of the α-phase.
The crystallographic parameters of the phases detected were calculated as follows. The average value of the BCC β-phase lattice parameter after all treatments is the same within the error limits and amounts to a β = 0.3337 ± 0.0003 nm. The hexagonal ω-phase lattice parameter after aging for 60 and 180 min at 300 • C amounts to a ω = 0.4724 ± 0.0005 nm, c ω = 0.2886 ± 0.0003 nm, and characteristic ratio c/a = 0.611 ± 0.007. The orthorhombic lattice parameters of the stress-induced α -phase (martensite) after cold drawing amount to a α = 0.3207 ±0.0005 nm, b α = 0.5029 ± 0.0006 nm, and c α = 0.4685 ± 0.0018. EBSD images shown in Figure 7a,b demonstrate the microstructure of the wire after PDA at 550 • C (30 min). In Figure 7a,b, the black lines are high-angle boundaries (misorientation angle > 15 • ). White lines correspond to low-angle boundaries (misorientation angle in the range from 3 to 15 • ). Grain-size distribution graphs are shown in Figure 7c,d.
Analysis of the EBSD images shows that after these TMT routes, a partially polygonised dislocation substructure and partially statically recrystallized structure are formed. After both treatments, large areas of the initially deformed structure are visualized, which are represented as elongated internally polygonized grains containing many low-angle sub-boundaries. Moreover, after CG+CD+550, 30, there are more such deformed areas in which polygonization occurs than after FG+CD+550, 30, which can be explained by the fact that the recrystallization process after FG+CD+550, 30 is faster due to the initially finer grain structure. The crystallographic parameters of the phases detected were calculated as foll The average value of the BCC β-phase lattice parameter after all treatments is the s within the error limits and amounts to aβ = 0.3337 ± 0.0003 nm. The hexagonal ω-p lattice parameter after aging for 60 and 180 min at 300 °C amounts to aω = 0.4724 ± 0. dislocation substructure and partially statically recrystallized structure are formed. After both treatments, large areas of the initially deformed structure are visualized, which are represented as elongated internally polygonized grains containing many low-angle subboundaries. Moreover, after CG+CD+550, 30, there are more such deformed areas in which polygonization occurs than after FG+CD+550, 30, which can be explained by the fact that the recrystallization process after FG+CD+550, 30 is faster due to the initially finer grain structure. Inverse pole figure analysis shows the crystallographic textures of the alloy after FG+CD+550, 30 and CG+CD+550, 30 (Figure 8a,b). For a more complete and clear presentation of the obtained results, these crystallographic textures were supplemented with orientation dependence of the recovery strain limit calculated for the Ti-18Zr-14Nb alloy [43] ( Figure 8c). During the β↔α" transformations, the <100>β and <111>β orientations correspond to comparatively low theoretical recovery strain limits (~2-3%), while the <110>β orientation corresponds to a maximum recovery strain limit of 5.7%. The texture with a maximum intensity close to the <102>β direction parallel to the DD obtained in both cases is of about the same strength. According to Figure 8c, the theoretical limit of the recovery strain can be estimated as ~4.3-5% for these orientations. Inverse pole figure analysis shows the crystallographic textures of the alloy after FG+CD+550, 30 and CG+CD+550, 30 (Figure 8a,b). For a more complete and clear presentation of the obtained results, these crystallographic textures were supplemented with orientation dependence of the recovery strain limit calculated for the Ti-18Zr-14Nb alloy [43] (Figure 8c). During the β↔α transformations, the <100> β and <111> β orientations correspond to comparatively low theoretical recovery strain limits (~2-3%), while the <110> β orientation corresponds to a maximum recovery strain limit of 5.7%. The texture with a maximum intensity close to the <102> β direction parallel to the DD obtained in both cases is of about the same strength. According to Figure 8c, the theoretical limit of the recovery strain can be estimated as~4.3-5% for these orientations.

Mechanical Properties
Tensile stress-strain diagrams of the Ti-18Zr-15Nb alloy after CRF and subsequent thermomechanical treatments are shown in Figure 9a. The mechanical properties of the Ti-Zr-Nb alloy after different TMT routes, obtained from tensile stress-strain diagrams, are presented in Figure 9b and Table 1.

Mechanical Properties
Tensile stress-strain diagrams of the Ti-18Zr-15Nb alloy after CRF and subsequent thermomechanical treatments are shown in Figure 9a. The mechanical properties of the Ti-Zr-Nb alloy after different TMT routes, obtained from tensile stress-strain diagrams, are presented in Figure 9b and Table 1.

Mechanical Properties
Tensile stress-strain diagrams of the Ti-18Zr-15Nb alloy after CRF and subsequent thermomechanical treatments are shown in Figure 9a. The mechanical properties of the Ti-Zr-Nb alloy after different TMT routes, obtained from tensile stress-strain diagrams, are presented in Figure 9b and Table 1. Static tensile tests to failure showed that after CRF, FG+CD, and CG+CD, a premature failure occurs. Additional CD leads to an increase in tensile strength (UTS = 740-750 MPa) and an obvious deterioration in ductility (δ = 1%). After FG+CD+550, 30, a high tensile strength (UTS = 624 MPa), the highest plasticity (δ = 8%), and the maximum difference between the dislocation and transformation yield stress are observed, which later determines the development of irreversible plastic deformation by the dislocation mechanism [12]. Subsequent low-temperature aging at a temperature of 300 • C (30-180 min) showed that with increasing aging time, ultimate tensile strength increases and the alloy becomes less ductile. The aged alloy shows the best combination of ultimate tensile strength (UTS ≈ 714 MPa) and ductility (δ ≈ 5%) after FG+CD+550, 30+300, 30. After FG+CD+550, 30+300, 60 and FG+CD+550,30+300, 180, specimens were fractured brittlely, which is associated with the development of the precipitation of the embrittling ω-phase [39].  Figure 10 shows cyclic loading-unloading stress-strain diagrams of Ti-Zr-Nb after different TMTs. The superelastic strain recovered due to reverse β→α transformation (ε r SE ), the elastic strain recovered upon unloading (ε el ), and the total recovery strain (ε r el+SE ) are among the main quantitative characteristics of the functional behavior of the SMA. These characteristics were measured from the cyclic stress-strain diagrams, obtained from superelastic cyclic testing, as shown in Figure 10.

Functional Properties
In fine-grained and coarse-grained states, the alloy exhibits a high value of superelastic strain (3.2% and 2.7%, respectively). After subsequent CD, the alloy in both states does not result in superelastic behavior, and specimens are broken after the second cycle (FG+CD case) and after the third cycle (CG+CD case). Subsequent PDA at 550 • C leads to a significant increase in ε r el+SE values. It should be noted that after PDA of specimens with a fine-grained structure, the alloy exhibits two times greater superelastic recovery strain than in the case of specimens with a coarse-grained structure. After FG+CD+550, 30, the alloy also exhibits the highest superelastic recovery strain (more than 3%) and total recovery strain of 5.4%, which can be explained by the presence of a stronger favorable <102> texture after this treatment. Additional low-temperature aging leads to a decrease in superelastic recovery strain; in all cases, it does not reach 1%. The largest superelastic recovery strain (0.8%) and total recovery strain (3.3%) are maximum after FG+CD+550, 30+300,30, then ε r el+SE decreases with increasing aging time.
Evolutions of the accumulated residual strain, superelastic recovery strain, the transformation, and dislocation yield stresses determined from the cyclic stress-strain diagrams are presented in Figure 11. After all treatments, ε r SE increases; only after FG+CD+550, 30 does it increase to its maximum value before decreasing (Figure 11b). In this latter state, the smallest ε acc is also observed in the first cycles (Figure 11a). The maximum superelastic recovery strain is observed after FG+CD+550, 30 in the eighth cycle (Figure 11b). The improvement in the superelastic behavior with an increase in the number of cycles occurs due to the difference between the dislocation and transformation yield stresses [17]. With an increase in the number of cycles, a decrease in σ tr occurs due to the accumulation of oriented internal micro-stresses ( Figure 11c) and an increase in the σ dis due to hardening accumulation (Figure 11d) in accordance with [11]. Apparent critical stress visible during the first two cycles as an only inflection point in the stress-strain diagrams (Figure 11c) cannot be reliably referred to as the transformation or dislocation yield stresses. After CRF, the FG+CD, CG+CD, FG+CD+550, 30+300, 180 alloy exhibits low-ductility (0.5-1%) and low-superelastic-recovery strain (0.1-0.3%). Additional low-temperature aging for 30 and 60 min also leads to low-superelastic-recovery strain: 0.2% and 0.8%, respectively. Thus, based on the results of the mechanical and functional tests, TMTs exhibiting the most favorable combinations of properties can be ranged from the best to worst as follows: FG+CD+550, 30, FG, CG, CG+CD+550, 30. In fine-grained and coarse-grained states, the alloy exhibits a high value of superelastic strain (3.2% and 2.7%, respectively). After subsequent CD, the alloy in both states does not result in superelastic behavior, and specimens are broken after the second cycle (FG+CD case) and after the third cycle (CG+CD case). Subsequent PDA at 550 °C leads to a significant increase in εr el+SE values. It should be noted that after PDA of specimens with a fine-grained structure, the alloy exhibits two times greater superelastic recovery strain than in the case of specimens with a coarse-grained structure. After FG+CD+550, 30, the alloy also exhibits the highest superelastic recovery strain (more than 3%) and total recovery strain of 5.4%, which can be explained by the presence of a stronger favorable <102> texture after this treatment. Additional low-temperature aging leads to a decrease in superelastic recovery strain; in all cases, it does not reach 1%. The largest superelastic recovery strain (0.8%) and total recovery strain (3.3%) are maximum after FG+CD+550, 30+300,30, then εr el+SE decreases with increasing aging time.
Evolutions of the accumulated residual strain, superelastic recovery strain, the trans- To explain the above-described differences in the functional behavior of the alloy, electrical resistance measurements were carried out to determine the forward martensitic transformation start temperature M s . Figure 12 shows the temperature dependences of the electrical resistivity upon cooling of the Ti-Zr-Nb alloy. Electrical resistivity is represented in an arbitrary unit (R/R 0 ), where R 0 is the electrical resistivity at 10 • C. The M s temperature of the martensitic transformations minus 112 • C for FG+CD+550, 30 (Figure 12a) is somewhat lower than that of minus 96 • C for CG+CD+550, 30 (Figure 12b), hence the functional properties after the latter TMT should be somewhat worse [1]. However, ε r SE is two times higher in the case of FG+CD+550, 30, while the change in electrical resistivity below M s temperature is almost the same. yield stresses. After CRF, the FG+CD, CG+CD, FG+CD+550, 30+300, 180 alloy exhibits low-ductility (0.5-1%) and low-superelastic-recovery strain (0.1-0.3%). Additional low-temperature aging for 30 and 60 min also leads to low-superelastic-recovery strain: 0.2% and 0.8%, respectively. Thus, based on the results of the mechanical and functional tests, TMTs exhibiting the most favorable combinations of properties can be ranged from the best to worst as follows: FG+CD+550, 30, FG, CG, CG+CD+550, 30. To explain the above-described differences in the functional behavior of the alloy, electrical resistance measurements were carried out to determine the forward martensitic transformation start temperature Ms. Figure 12 shows the temperature dependences of the electrical resistivity upon cooling of the Ti-Zr-Nb alloy. Electrical resistivity is represented in an arbitrary unit (R/R0), where R0 is the electrical resistivity at 10 °C. The Ms temperature of the martensitic transformations minus 112 °C for FG+CD+550, 30 ( Figure 12a) is somewhat lower than that of minus 96 °C for CG+CD+550, 30 (Figure 12b), hence the functional properties after the latter TMT should be somewhat worse [1]. However, εr SE is two times higher in the case of FG+CD+550, 30, while the change in electrical resistivity below Ms temperature is almost the same.

General Discussion
To sum up the results of this preliminary study, a comparison of the mechanical properties and the superelastic recovery strain of the Ti-Zr-Nb SMA subjected to conventional high-and low-temperature TMT [12], as well as for nickel-free superelastic alloys subjected to drawing [34,35], is presented in Figure 13. The superelastic recovery strain εr SE was measured after the total induced strain of 4% for all cases.

General Discussion
To sum up the results of this preliminary study, a comparison of the mechanical properties and the superelastic recovery strain of the Ti-Zr-Nb SMA subjected to conventional high-and low-temperature TMT [12], as well as for nickel-free superelastic alloys subjected to drawing [34,35], is presented in Figure 13. The superelastic recovery strain ε r SE was measured after the total induced strain of 4% for all cases.
To sum up the results of this preliminary study, a comparison of the mechanical properties and the superelastic recovery strain of the Ti-Zr-Nb SMA subjected to conventional high-and low-temperature TMT [12], as well as for nickel-free superelastic alloys subjected to drawing [34,35], is presented in Figure 13. The superelastic recovery strain εr SE was measured after the total induced strain of 4% for all cases. The set of mechanical and superelastic properties of the Ti-Zr-Nb bar stock after conventional TMT, including rotary forging, is comparable with the results of this work obtained from wire. However, in this, work low-temperature TMT is applied, the alloy exhibits high recovery strain in the first test cycles comparable to that for the alloy after hightemperature TMT. Given that the alloy after hot deformation exhibits a high functional fatigue life [12,43], such a TMT scheme using drawing looks very promising for future works. Moreover, hot deformation contributes to an increase in the plasticity of the alloy and the ability to be deformed to a greater extent. The ability to apply severe plastic deformation by drawing reveals the likelihood of increasing the strength properties of the  The set of mechanical and superelastic properties of the Ti-Zr-Nb bar stock after conventional TMT, including rotary forging, is comparable with the results of this work obtained from wire. However, in this, work low-temperature TMT is applied, the alloy exhibits high recovery strain in the first test cycles comparable to that for the alloy after high-temperature TMT. Given that the alloy after hot deformation exhibits a high functional fatigue life [12,43], such a TMT scheme using drawing looks very promising for future works. Moreover, hot deformation contributes to an increase in the plasticity of the alloy and the ability to be deformed to a greater extent. The ability to apply severe plastic deformation by drawing reveals the likelihood of increasing the strength properties of the wire, along with maintaining excellent superelasticity as was shown for the nickel-free Ti-25Hf-21Nb SMA [34].
Obviously, the potential of drawing has not been exhausted for Ti-Zr-Nb SMAs. This work is primary for further systematic comprehensive research, as well as the study of the mechanisms of the formation and correlation of microstructure, texture, mechanical behavior, and functional characteristics of Ti-Zr-Nb SMAs after drawing.