- Email: [email protected]
Fabrication of cellular shape memory alloy materials by reactive eutectic brazing using niobium
Materials Science and Engineering A 438–440 (2006) 1113–1118
Fabrication of cellular shape memory alloy materials by reactive eutectic brazing using niobium David S. Grummon a,∗ , John A. Shaw b , John Foltz a a
Department of Chemical Engineering and Materials Science, Michigan State University, E. Lansing, MI 48823, USA b Department of Aerospace Engineering, University of Michigan, Ann Arbor, MI 48104, USA Received 23 March 2005; received in revised form 3 March 2006; accepted 30 March 2006
Abstract A novel metal-joining process is described that exploits interfacial reactions between nitinol and pure niobium to produce a reactive eutectic liquid that readily creates a robust metallurgical bond between nitinol and itself. With this new reactive-brazing process we have been able to create prototypes of superelastic cellular honeycomb topologies from conventional nitinol precursor materials such as tubes and corrugated sheets. These have been found to display excellent isothermal shape-recovery and have high specific strength, low relative density, and extremely high resilience. The method will allow realization of complex space-frames, honeycomb sandwich panels, and other sparse built-up thermally active multifunctional nitinol structures. The braze material is corrosion-resistant, machinable, biocompatible, and has good osteoconductivity, potentially enabling a variety of compliant superelastic implant and bone-replacement materials unobtainable by any other means. © 2006 Published by Elsevier B.V. Keywords: Nitinol; NiTi; Niobium; Joining; Brazing; Honeycombs
1. Introduction Cellular structures that combine ultra-low density with the constitutive behavior of shape memory alloys are of considerable technical interest. The free space kinematics of cellular architectures can amplify thermoelastic shape-recovery beyond monolithic strain limits, and the large surface to volume ratio radically improves thermal response times. Nitinol in porous form has been studied [1–3] and commercialized [4], and open cell forms derived from powder processing having relative densities below 5% have been reported [5,6]. However, metallurgically joining nitinol to itself has historically been problematic, discouraging the creation of built-up cellular structures from wrought NiTi-based shape memory alloys and superelastic materials. The development of such materials will be possible only if means are available to robustly join individual corrugated or dimpled sheets, strips or wires. The joining method must not only provide a robust metallurgical bond containing only biocompatible phases, but must also be a simple, clean, and cost-effective batch process. While a few specialized techniques for soldering and
∗
Corresponding author. Tel.: +1 517 353 4688; fax: +1 517 432 1105. E-mail address: [email protected] (D.S. Grummon).
0921-5093/$ – see front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.msea.2006.03.113
welding nitinol have been developed over the years, until now no low-cost joining method has been available that meets these requirements. We have recently discovered that when niobium is brought into contact with conventional wrought nitinol at elevated temperature, interdiffusion between NiTi and pure Nb quickly leads to the formation of a liquid phase that aggressively wets both the pure niobium and NiTi. The liquid that forms is a quasi-binary eutectic with composition near Ni38 Ti36 Nb26 (at.%). Rich in highly reactive titanium, it readily dissolves oxide scales without the use of fluxes. It flows easily into capillary spaces, and subsequently solidifies into a braze joint having good strength and ductility. Interfacial melting reactions of this sort usually lead to the formation of brittle intermetallic phases after solidification. The present process, however, gives a braze microstructure containing only austenitic Nitinol and body-centered cubic-niobium, both of which are well understood phases that are tough, corrosion-resistant [7,8] and thermally stable. In this paper we report some initial results on open-cell honeycomb structures made by brazing together corrugated strips of nitinol that had been formulated for superelastic response at room temperature. We show that despite the high processing temperatures involved, very robust superelastic response can be
1114
D.S. Grummon et al. / Materials Science and Engineering A 438–440 (2006) 1113–1118
obtained in a material with only 5% relative density. Finally, we discuss some emerging details of the quasi-binary metallurgical system that forms the basis of the new joining process. 2. Methods and materials Honeycomb specimens were fabricated from 5.2 mm wide × 0.2 mm thick rolled nickel-rich nitinol strip obtained from Memry Corporation, Brookfield, CT. The strips were first shape-set into corrugated forms at 500 ◦ C using stainless steel dies. The corrugated strips were then etched in a solution containing 50 mL HNO3 , 38 mL HCl, and 17 mL HF to remove surface oxides. The honeycomb layup was assembled with small 50.8 m thick square pieces (∼3 mm × 3 mm) of 99.7% pure niobium foil placed between the contacting faces. The assembly was lightly secured in TZ-molybdenum support fixtures and washed in acetone and ethanol. No fluxing agents were used. Brazing was conducted in a Centorr M60 tungsten-element diffusion-pumped vacuum furnace evacuated to a base pressure of about 10−5 mbar. Once the base pressure was attained, the specimens were heated quickly to between 1175 and 1200 ◦ C, held for a brief period, and then furnace cooled. After removal from the vacuum furnace, the as-brazed honeycombs were given an aging treatment at between 350 and 515 ◦ C prior to compression testing. In-plane crushing tests were performed using an Instron 5585 electro-mechanical testing machine in room temperature air under quasistatic displacement control at a global strain rate of 0.0033 s−1 . Digital optical photographs were taken periodically using a Nikon D100 camera fitted with a 105 mm Micro Nikkor lens. The microstructure of the braze joints was studied by scanning electron microscopy and energy-dispersive X-ray fluorescence spectroscopy. The latter study was conducted on a braze joint made between a pair of 3.2 mm diameter superelastic nitinol tubes prepared in the same manner as the honeycomb specimens, but using a 500 m diameter pure niobium wire placed adjacent to the tube contact line as the braze metal.
3. Results and discussion Fig. 1 shows the stress–strain response of a hexagonal-cell honeycomb fabricated as described above. Twelve load–unload cycles were applied, with the total strain incremented in 5% steps from 5 to 60%, as indicated by circled numbers on the plot. At the right are photographs of the specimen taken before loading, at 60% strain (cycle 12), and after unloading from the last cycle (12a). A limit load is apparent at ∼22% strain on cycle 5 that corresponds to sudden formation of a local band of highly deformed cells. The final cycle, shown by the bold line in Fig. 1, shows that although the closure of the loop is not quite perfect (leaving ∼1.3% residual strain), the wide-hysteresis and non-linear behavior typical of superelastic material undergoing bending deformation is clearly demonstrated. The deformed configuration at maximum load has a nearly symmetric pattern with respect to a vertical plane, with a single relatively undeformed cell at the lower middle location. The final unloaded configuration after 12 strain cycles reflects ∼7.3% unrecovered strain. During the incremental compression tests, no failure of the braze joints was observed. Later, crushing of this honeycomb to 95% compressive strain still did not induce gross bond failure. Following this overload cycle (which crushed the cells to near full density) the specimen recovered to a final gauge height of 11.81 mm, reflecting a residual strain of ∼45%. After warming the specimen past Af , the height recovered by 1.46 mm, giving a final unreversed strain of ∼37%. Fig. 2 shows X-ray fluorescence maps for Nb L-␣, Ni K-␣ and Ti K-␣ radiation taken from a braze joint that had been made between a pair of 3.2 mm diameter nitinol tubes. Approximately half of the joint appears in the field, and the tube curvature, as well as the meniscus shape of the solidified braze, is visible. These images reveal that the solidified braze contains two distinct microconstituents. One appears as a bulbous zone, directly adjacent to the tube surfaces, for which the Nb L-␣ map indicates a small but detectable amount of niobium, together with much larger and approximately equal amounts of Ti and Ni. Between
Fig. 1. Mechanical response of a niobium-brazed open cell honeycomb built-up from corrugated strips of 5.2 mm wide × 0.19 mm thick nitinol strip. The original honeycomb dimensions were 21.5 mm × 5.2 mm × ∼40 mm wide and had a nominal cell diameter of 5.5 mm. The braze joint is located between the horizontal members.
D.S. Grummon et al. / Materials Science and Engineering A 438–440 (2006) 1113–1118
1115
Fig. 2. X-ray fluorescence maps for the three species in the braze metal (the inset schematically indicates the field of the images with respect to the pair of tubes, while white square indicates the field of the magnified view in Fig. 6).
the bulbous regions lies a second zone, with complimentary morphology, that is rich in niobium and contains a minor presence of both Ni and Ti. To understand the metallurgical nature of these distinct zones it is useful to review what is currently known regarding the nature of the Ni–Ti–Nb ternary alloy system. A 900 ◦ C isothermal section of the Ni–Ti–Nb ternary phase diagram, with a liquidus projection and the bounding binary phase diagrams, together with a sketch of the probable liquidus surface, is shown in Fig. 3. This diagram has been constructed as a composite incorporating data from [9,10]. It reveals the presence of five binary compounds, NiTi2 , NiTi, Ni3 Ti, Ni3 Nb and NiNb, together with four ternary intermediate phases labeled 1 , 2 ,
3 , and 4 (a sixth binary compound, NbNi8 , is stable only to about 800 ◦ C.) A critical aspect of the diagram, with respect to the present work, is the complete mutual solubility of Ti and Nb above the Ti ␣– transus at 882 ◦ C. Joining this disordered single-phase field and ordered B2-NiTi compound is a large NiTi + -(Nb, Ni, Ti) two-phase field that extends between approximately 6 and 92 at.% Nb. This strongly suggests the existence of a quasibinary eutectic system between ordered TiNi and the disordered -(Nb, Ni, Ti) solid solution. Note that the tie lines [9] in this field are nearly parallel to the [equiatomic NiTi]–Nb locus. Although the eutectic microconstituent is not usually identified in wide-hysteresis alloys exemplified by Ni47 Ti44 Nb9 , it
Fig. 3. Estimated appearance of the Ni–Ti–Nb ternary equilibrium system based on data from [9,10]. (Left) Probable liquidus surface (schematic) and the Ti–Ni and Ti–Nb binary isopleths. (Right) The 900 ◦ C isothermal section with single-phase regions shown in grey for clarity. The dashed line at the center is the equiatomic TiNi–Nb locus. The bold circle locates the ternary (quasi-binary) eutectic point between NiTi and the -Nb solid solution which occurs at 1170 ◦ C, according to [9].
1116
D.S. Grummon et al. / Materials Science and Engineering A 438–440 (2006) 1113–1118
Fig. 4. The quasi-binary [NiTi]–Nb system. (Left) The quasi-binary isopleth as it relates to the full ternary system; (right) an estimate of the quasi-binary equilibrium phase diagram for [NiTi]–Nb.
has been previously observed [11–13]. Zhao et al. [11] identified the terminal phases and gave the lattice parameter of the disordered cubic phase as 0.32963 nm, and that of the NiTi phase as 0.30182 nm. Fig. 4 shows the ternary system bisected along the NiTi–Nb line to reveal the probable quasi-binary isopleth, which is shown in more detail as a NiTi–Nb quasi-binary phase diagram on the right. The latter diagram is an estimate based on the established temperature and composition of the quasi-binary eutectic (1170 ◦ C, at Ni38 Ti36 Nb26 ), with liquidus lines extrapolated to the component melting points, and solidus lines extrapolated through solidus temperatures cited in [10]. It is this quasi-binary system, having terminal phases with low mutual solubility, a eutectic temperature 140 K below the melting point of Nitinol, and which contains no brittle intermetallic phases, that forms the thermodynamic basis for the new brazing method. The nature of the solidified braze metal may now be appreciated by examination of the secondary electron image shown in Fig. 5. The field of this micrograph is indicated by a white square on Fig. 2. The image reveals that the niobium (the light colored phase) is concentrated in a two-phase region that is recognizable as a eutectic solidification zone, and that the globular features richer in TiNi are homogenous single-phase zones. The latter is therefore identified as a dendritically solidified proeutectic phase whose formation is described further below. It is clear from the quasi-binary phase diagram that an interface between solid NiTi and pure Nb is unstable and will spontaneously melt at temperatures above 1170 ◦ C. The first liquid to appear will have a composition near the eutectic at Ni38 Ti36 Nb24 . Assuming that the system stays close to equilibrium at the solid–liquid interfaces, and that the liquid forms simultaneously at the surfaces of both the solid phases, the liquid zone must contain a composition gradient that would drive transport of (Ti + Ni) toward the Nb-rich solid, and vice versa, causing the size of the melt to increase steadily until the source of niobium is exhausted. This is because (at temperatures above the eutectic) the equilibrium composition of the liquid is different at the L–NiTi interface than at the L–Nb interface. When all sources of pure Nb have been depleted (which might involve cutoff of the liquid from the Nb source as it flows into
capillary spaces), the liquid will still be in contact with solid NiTi. As niobium continues to diffuse from the eutectic liquid into solid NiTi, the NiTi–L interface is forced away from the equilibrium composition. The latter can only be restored by freezing out the TiNi-rich proeutectic phase. When solid proeutectic TiNi (with about 8% dissolved Nb) is deposited on the solid–liquid interface, the Nb content of both the liquid and the interface are restored to the equilibrium levels. Thus, following Nb-exhaustion, proeutectic TiNi will steadily freeze out at the L–NiTi interface. If allowed sufficient time, as dictated by the solid-state diffusivities, the liquid will eventually disappear. The microstructure in Fig. 5, however, reveals that the proeutec-
Fig. 5. Scanning electron micrograph (secondary electron image or ‘SEI’) of a braze joint between two 3.2 mm diameter extruded superelastic NiTi tubes. The joint was formed by heating the pair of tubes in the presence of a 508 m diameter pure niobium wire to a temperature of 1170 ◦ C, holding for 5–10 min, and then furnace cooling. The section has been mechanically polished but has not been etched.
D.S. Grummon et al. / Materials Science and Engineering A 438–440 (2006) 1113–1118
1117
Fig. 6. (Left) Fracture surface through the brazed joint made manually at room temperature showing a fracture path through the braze metal and no tendency for the braze to tear away from the NiTi tube surfaces. (Right) Detail from the fracture surface shown in Fig. 6. The dimpled morphology indicates the ductile nature of the braze alloy fracture.
tic phase is dendritic, and must therefore have solidified under conditions of melt supercooling during cool down from the processing temperature. The resulting phases and microconstituents (indicated by white boxes in Fig. 5) are keyed to the quasi-binary phase diagram in the inset. At the far left of the micrograph is nitinol that has never been melted. Contiguous with this zone, but containing significantly more niobium, is the dendritic proeutectic zone that terminates in bulbous projections. Finally, the eutectic solidification zone is apparent on the right. No unreacted Nb phase has been detected in any of these specimens, indicating that the contact melting reaction proceeds very quickly. Fig. 6 shows a fracture surface from a braze joint like the one shown in Fig. 2. The two tubes were pulled apart manually at
room temperature. As is apparent in the micrograph, the fracture path ran through the braze metal which did not pull away from the adjacent NiTi tube surfaces. A high magnification detail from this fracture (right) surface shows a dimpled surface typical of fully ductile fracture. 4. Conclusions A new brazing method has been discovered that is able to metallurgically bond nitinol to itself in a process that allows the development of strong superelastic response in the nitinol. The method can be exploited to fabricate open cell honeycombs, spaceframes, or other built-up topologies as suggested by Fig. 7 below. It is based on the existence of a quasi-binary eutectic sys-
Fig. 7. Topological forms enabled by the new joining method. Note that form (e) is auxetic.
1118
D.S. Grummon et al. / Materials Science and Engineering A 438–440 (2006) 1113–1118
tem between NiTi and niobium for which the terminal phases are both ductile, tough, and biocompatible. The method requires no fluxes and can be carried out in normal industrial vacuum furnaces. Further work to optimize process parameters and quantify the mechanical properties of the braze is ongoing. References [1] D. Lagoudas, P. Entchev, E. Vandygriff, M. Qidwai, V. De-Giorgi, Proceedings of the Adaptive Structures and Material Systems Symposium, ASME Int’l Mechanical Engineering Congress and Exposition, 2001. [2] K. Thangaraj, Y.C. Chen, K. Salama, Proceedings of the Adaptive Structures and Material Systems Symposium, ASME International Mechanical Engineering Congress and Exposition, vol. 60, 2000, pp. 59–63. [3] B. Li, L. Rong, Y. Li, V.E. Gjunter, Acta Mater. 48 (15) (2000) 3895–3904. [4] A. Chernyshov, M. Leroux, M. Assad, A. Dujovne, E. Garcia-Belenguer, In: D. Mantovani (Ed.), Advanced Biomaterials for Biomedical Applications, Proceedings of the Conference Metallic Society Canadian Institute of Mining, Metallurgy and Petroleum (CIM), 2002, pp.109–119.
[5] J.A. Shaw, A. Gremillet, D.S. Grummon, Proceedings of the ASME Adaptive Structures and Materials Systems Symposium, ASME winter meeting, New Orleans, LA, November 2002. [6] D.S. Grummon, J.A. Shaw, A. Gremillet, J. Appl. Phys. Lett. 82 (2003) 2727. [7] I. Cherghescu, V. Constantin, Chem. Mater. Sci. 60 (3–4) (1998) 137–141. [8] Z. Dong, W. Liu, D. Jia, Z. Tang, D. Wang, Rare Met. Mater. Eng. 29 (3) (2000) 182–184. [9] K.P. Gupta, Phase Diagrams of Ternary Nickel Alloys. Part 2. Section D.2. The Nb–Ni–Ti (Niobium–Nickel–Titanium) System, Indian Institute of Metals, Calcutta, 1991, pp. 163–176. [10] S.B. Prima, L.A. Tret’yachenko, V.M. Petyukk, Powder Metall. Met. Ceram. 34 (1995) 155. [11] L.C. Zhao, T.W. Duerig, C.M. Wayman, Proceedings of the MRS International Meeting on Advanced Materials, vol.9, 1989, Shape Memory Materials, Tokyo, Japan, 31 May–3 June, 1988, pp. 171–176. [12] T. Onda, M. Piao, Y. Bando, H. Ichinose, K. Otsuka, Mater. Trans. JIM 36 (1995) 23–29. [13] M. Piao, S. Miyazaki, K. Otsuka, N. Nishida, Mater. Trans. JIM 33 (1992) 337–345.
Fabrication of cellular shape memory alloy materials by reactive eutectic brazing using niobium David S. Grummon a,∗ , John A. Shaw b , John Foltz a a
Department of Chemical Engineering and Materials Science, Michigan State University, E. Lansing, MI 48823, USA b Department of Aerospace Engineering, University of Michigan, Ann Arbor, MI 48104, USA Received 23 March 2005; received in revised form 3 March 2006; accepted 30 March 2006
Abstract A novel metal-joining process is described that exploits interfacial reactions between nitinol and pure niobium to produce a reactive eutectic liquid that readily creates a robust metallurgical bond between nitinol and itself. With this new reactive-brazing process we have been able to create prototypes of superelastic cellular honeycomb topologies from conventional nitinol precursor materials such as tubes and corrugated sheets. These have been found to display excellent isothermal shape-recovery and have high specific strength, low relative density, and extremely high resilience. The method will allow realization of complex space-frames, honeycomb sandwich panels, and other sparse built-up thermally active multifunctional nitinol structures. The braze material is corrosion-resistant, machinable, biocompatible, and has good osteoconductivity, potentially enabling a variety of compliant superelastic implant and bone-replacement materials unobtainable by any other means. © 2006 Published by Elsevier B.V. Keywords: Nitinol; NiTi; Niobium; Joining; Brazing; Honeycombs
1. Introduction Cellular structures that combine ultra-low density with the constitutive behavior of shape memory alloys are of considerable technical interest. The free space kinematics of cellular architectures can amplify thermoelastic shape-recovery beyond monolithic strain limits, and the large surface to volume ratio radically improves thermal response times. Nitinol in porous form has been studied [1–3] and commercialized [4], and open cell forms derived from powder processing having relative densities below 5% have been reported [5,6]. However, metallurgically joining nitinol to itself has historically been problematic, discouraging the creation of built-up cellular structures from wrought NiTi-based shape memory alloys and superelastic materials. The development of such materials will be possible only if means are available to robustly join individual corrugated or dimpled sheets, strips or wires. The joining method must not only provide a robust metallurgical bond containing only biocompatible phases, but must also be a simple, clean, and cost-effective batch process. While a few specialized techniques for soldering and
∗
Corresponding author. Tel.: +1 517 353 4688; fax: +1 517 432 1105. E-mail address: [email protected] (D.S. Grummon).
0921-5093/$ – see front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.msea.2006.03.113
welding nitinol have been developed over the years, until now no low-cost joining method has been available that meets these requirements. We have recently discovered that when niobium is brought into contact with conventional wrought nitinol at elevated temperature, interdiffusion between NiTi and pure Nb quickly leads to the formation of a liquid phase that aggressively wets both the pure niobium and NiTi. The liquid that forms is a quasi-binary eutectic with composition near Ni38 Ti36 Nb26 (at.%). Rich in highly reactive titanium, it readily dissolves oxide scales without the use of fluxes. It flows easily into capillary spaces, and subsequently solidifies into a braze joint having good strength and ductility. Interfacial melting reactions of this sort usually lead to the formation of brittle intermetallic phases after solidification. The present process, however, gives a braze microstructure containing only austenitic Nitinol and body-centered cubic-niobium, both of which are well understood phases that are tough, corrosion-resistant [7,8] and thermally stable. In this paper we report some initial results on open-cell honeycomb structures made by brazing together corrugated strips of nitinol that had been formulated for superelastic response at room temperature. We show that despite the high processing temperatures involved, very robust superelastic response can be
1114
D.S. Grummon et al. / Materials Science and Engineering A 438–440 (2006) 1113–1118
obtained in a material with only 5% relative density. Finally, we discuss some emerging details of the quasi-binary metallurgical system that forms the basis of the new joining process. 2. Methods and materials Honeycomb specimens were fabricated from 5.2 mm wide × 0.2 mm thick rolled nickel-rich nitinol strip obtained from Memry Corporation, Brookfield, CT. The strips were first shape-set into corrugated forms at 500 ◦ C using stainless steel dies. The corrugated strips were then etched in a solution containing 50 mL HNO3 , 38 mL HCl, and 17 mL HF to remove surface oxides. The honeycomb layup was assembled with small 50.8 m thick square pieces (∼3 mm × 3 mm) of 99.7% pure niobium foil placed between the contacting faces. The assembly was lightly secured in TZ-molybdenum support fixtures and washed in acetone and ethanol. No fluxing agents were used. Brazing was conducted in a Centorr M60 tungsten-element diffusion-pumped vacuum furnace evacuated to a base pressure of about 10−5 mbar. Once the base pressure was attained, the specimens were heated quickly to between 1175 and 1200 ◦ C, held for a brief period, and then furnace cooled. After removal from the vacuum furnace, the as-brazed honeycombs were given an aging treatment at between 350 and 515 ◦ C prior to compression testing. In-plane crushing tests were performed using an Instron 5585 electro-mechanical testing machine in room temperature air under quasistatic displacement control at a global strain rate of 0.0033 s−1 . Digital optical photographs were taken periodically using a Nikon D100 camera fitted with a 105 mm Micro Nikkor lens. The microstructure of the braze joints was studied by scanning electron microscopy and energy-dispersive X-ray fluorescence spectroscopy. The latter study was conducted on a braze joint made between a pair of 3.2 mm diameter superelastic nitinol tubes prepared in the same manner as the honeycomb specimens, but using a 500 m diameter pure niobium wire placed adjacent to the tube contact line as the braze metal.
3. Results and discussion Fig. 1 shows the stress–strain response of a hexagonal-cell honeycomb fabricated as described above. Twelve load–unload cycles were applied, with the total strain incremented in 5% steps from 5 to 60%, as indicated by circled numbers on the plot. At the right are photographs of the specimen taken before loading, at 60% strain (cycle 12), and after unloading from the last cycle (12a). A limit load is apparent at ∼22% strain on cycle 5 that corresponds to sudden formation of a local band of highly deformed cells. The final cycle, shown by the bold line in Fig. 1, shows that although the closure of the loop is not quite perfect (leaving ∼1.3% residual strain), the wide-hysteresis and non-linear behavior typical of superelastic material undergoing bending deformation is clearly demonstrated. The deformed configuration at maximum load has a nearly symmetric pattern with respect to a vertical plane, with a single relatively undeformed cell at the lower middle location. The final unloaded configuration after 12 strain cycles reflects ∼7.3% unrecovered strain. During the incremental compression tests, no failure of the braze joints was observed. Later, crushing of this honeycomb to 95% compressive strain still did not induce gross bond failure. Following this overload cycle (which crushed the cells to near full density) the specimen recovered to a final gauge height of 11.81 mm, reflecting a residual strain of ∼45%. After warming the specimen past Af , the height recovered by 1.46 mm, giving a final unreversed strain of ∼37%. Fig. 2 shows X-ray fluorescence maps for Nb L-␣, Ni K-␣ and Ti K-␣ radiation taken from a braze joint that had been made between a pair of 3.2 mm diameter nitinol tubes. Approximately half of the joint appears in the field, and the tube curvature, as well as the meniscus shape of the solidified braze, is visible. These images reveal that the solidified braze contains two distinct microconstituents. One appears as a bulbous zone, directly adjacent to the tube surfaces, for which the Nb L-␣ map indicates a small but detectable amount of niobium, together with much larger and approximately equal amounts of Ti and Ni. Between
Fig. 1. Mechanical response of a niobium-brazed open cell honeycomb built-up from corrugated strips of 5.2 mm wide × 0.19 mm thick nitinol strip. The original honeycomb dimensions were 21.5 mm × 5.2 mm × ∼40 mm wide and had a nominal cell diameter of 5.5 mm. The braze joint is located between the horizontal members.
D.S. Grummon et al. / Materials Science and Engineering A 438–440 (2006) 1113–1118
1115
Fig. 2. X-ray fluorescence maps for the three species in the braze metal (the inset schematically indicates the field of the images with respect to the pair of tubes, while white square indicates the field of the magnified view in Fig. 6).
the bulbous regions lies a second zone, with complimentary morphology, that is rich in niobium and contains a minor presence of both Ni and Ti. To understand the metallurgical nature of these distinct zones it is useful to review what is currently known regarding the nature of the Ni–Ti–Nb ternary alloy system. A 900 ◦ C isothermal section of the Ni–Ti–Nb ternary phase diagram, with a liquidus projection and the bounding binary phase diagrams, together with a sketch of the probable liquidus surface, is shown in Fig. 3. This diagram has been constructed as a composite incorporating data from [9,10]. It reveals the presence of five binary compounds, NiTi2 , NiTi, Ni3 Ti, Ni3 Nb and NiNb, together with four ternary intermediate phases labeled 1 , 2 ,
3 , and 4 (a sixth binary compound, NbNi8 , is stable only to about 800 ◦ C.) A critical aspect of the diagram, with respect to the present work, is the complete mutual solubility of Ti and Nb above the Ti ␣– transus at 882 ◦ C. Joining this disordered single-phase field and ordered B2-NiTi compound is a large NiTi + -(Nb, Ni, Ti) two-phase field that extends between approximately 6 and 92 at.% Nb. This strongly suggests the existence of a quasibinary eutectic system between ordered TiNi and the disordered -(Nb, Ni, Ti) solid solution. Note that the tie lines [9] in this field are nearly parallel to the [equiatomic NiTi]–Nb locus. Although the eutectic microconstituent is not usually identified in wide-hysteresis alloys exemplified by Ni47 Ti44 Nb9 , it
Fig. 3. Estimated appearance of the Ni–Ti–Nb ternary equilibrium system based on data from [9,10]. (Left) Probable liquidus surface (schematic) and the Ti–Ni and Ti–Nb binary isopleths. (Right) The 900 ◦ C isothermal section with single-phase regions shown in grey for clarity. The dashed line at the center is the equiatomic TiNi–Nb locus. The bold circle locates the ternary (quasi-binary) eutectic point between NiTi and the -Nb solid solution which occurs at 1170 ◦ C, according to [9].
1116
D.S. Grummon et al. / Materials Science and Engineering A 438–440 (2006) 1113–1118
Fig. 4. The quasi-binary [NiTi]–Nb system. (Left) The quasi-binary isopleth as it relates to the full ternary system; (right) an estimate of the quasi-binary equilibrium phase diagram for [NiTi]–Nb.
has been previously observed [11–13]. Zhao et al. [11] identified the terminal phases and gave the lattice parameter of the disordered cubic phase as 0.32963 nm, and that of the NiTi phase as 0.30182 nm. Fig. 4 shows the ternary system bisected along the NiTi–Nb line to reveal the probable quasi-binary isopleth, which is shown in more detail as a NiTi–Nb quasi-binary phase diagram on the right. The latter diagram is an estimate based on the established temperature and composition of the quasi-binary eutectic (1170 ◦ C, at Ni38 Ti36 Nb26 ), with liquidus lines extrapolated to the component melting points, and solidus lines extrapolated through solidus temperatures cited in [10]. It is this quasi-binary system, having terminal phases with low mutual solubility, a eutectic temperature 140 K below the melting point of Nitinol, and which contains no brittle intermetallic phases, that forms the thermodynamic basis for the new brazing method. The nature of the solidified braze metal may now be appreciated by examination of the secondary electron image shown in Fig. 5. The field of this micrograph is indicated by a white square on Fig. 2. The image reveals that the niobium (the light colored phase) is concentrated in a two-phase region that is recognizable as a eutectic solidification zone, and that the globular features richer in TiNi are homogenous single-phase zones. The latter is therefore identified as a dendritically solidified proeutectic phase whose formation is described further below. It is clear from the quasi-binary phase diagram that an interface between solid NiTi and pure Nb is unstable and will spontaneously melt at temperatures above 1170 ◦ C. The first liquid to appear will have a composition near the eutectic at Ni38 Ti36 Nb24 . Assuming that the system stays close to equilibrium at the solid–liquid interfaces, and that the liquid forms simultaneously at the surfaces of both the solid phases, the liquid zone must contain a composition gradient that would drive transport of (Ti + Ni) toward the Nb-rich solid, and vice versa, causing the size of the melt to increase steadily until the source of niobium is exhausted. This is because (at temperatures above the eutectic) the equilibrium composition of the liquid is different at the L–NiTi interface than at the L–Nb interface. When all sources of pure Nb have been depleted (which might involve cutoff of the liquid from the Nb source as it flows into
capillary spaces), the liquid will still be in contact with solid NiTi. As niobium continues to diffuse from the eutectic liquid into solid NiTi, the NiTi–L interface is forced away from the equilibrium composition. The latter can only be restored by freezing out the TiNi-rich proeutectic phase. When solid proeutectic TiNi (with about 8% dissolved Nb) is deposited on the solid–liquid interface, the Nb content of both the liquid and the interface are restored to the equilibrium levels. Thus, following Nb-exhaustion, proeutectic TiNi will steadily freeze out at the L–NiTi interface. If allowed sufficient time, as dictated by the solid-state diffusivities, the liquid will eventually disappear. The microstructure in Fig. 5, however, reveals that the proeutec-
Fig. 5. Scanning electron micrograph (secondary electron image or ‘SEI’) of a braze joint between two 3.2 mm diameter extruded superelastic NiTi tubes. The joint was formed by heating the pair of tubes in the presence of a 508 m diameter pure niobium wire to a temperature of 1170 ◦ C, holding for 5–10 min, and then furnace cooling. The section has been mechanically polished but has not been etched.
D.S. Grummon et al. / Materials Science and Engineering A 438–440 (2006) 1113–1118
1117
Fig. 6. (Left) Fracture surface through the brazed joint made manually at room temperature showing a fracture path through the braze metal and no tendency for the braze to tear away from the NiTi tube surfaces. (Right) Detail from the fracture surface shown in Fig. 6. The dimpled morphology indicates the ductile nature of the braze alloy fracture.
tic phase is dendritic, and must therefore have solidified under conditions of melt supercooling during cool down from the processing temperature. The resulting phases and microconstituents (indicated by white boxes in Fig. 5) are keyed to the quasi-binary phase diagram in the inset. At the far left of the micrograph is nitinol that has never been melted. Contiguous with this zone, but containing significantly more niobium, is the dendritic proeutectic zone that terminates in bulbous projections. Finally, the eutectic solidification zone is apparent on the right. No unreacted Nb phase has been detected in any of these specimens, indicating that the contact melting reaction proceeds very quickly. Fig. 6 shows a fracture surface from a braze joint like the one shown in Fig. 2. The two tubes were pulled apart manually at
room temperature. As is apparent in the micrograph, the fracture path ran through the braze metal which did not pull away from the adjacent NiTi tube surfaces. A high magnification detail from this fracture (right) surface shows a dimpled surface typical of fully ductile fracture. 4. Conclusions A new brazing method has been discovered that is able to metallurgically bond nitinol to itself in a process that allows the development of strong superelastic response in the nitinol. The method can be exploited to fabricate open cell honeycombs, spaceframes, or other built-up topologies as suggested by Fig. 7 below. It is based on the existence of a quasi-binary eutectic sys-
Fig. 7. Topological forms enabled by the new joining method. Note that form (e) is auxetic.
1118
D.S. Grummon et al. / Materials Science and Engineering A 438–440 (2006) 1113–1118
tem between NiTi and niobium for which the terminal phases are both ductile, tough, and biocompatible. The method requires no fluxes and can be carried out in normal industrial vacuum furnaces. Further work to optimize process parameters and quantify the mechanical properties of the braze is ongoing. References [1] D. Lagoudas, P. Entchev, E. Vandygriff, M. Qidwai, V. De-Giorgi, Proceedings of the Adaptive Structures and Material Systems Symposium, ASME Int’l Mechanical Engineering Congress and Exposition, 2001. [2] K. Thangaraj, Y.C. Chen, K. Salama, Proceedings of the Adaptive Structures and Material Systems Symposium, ASME International Mechanical Engineering Congress and Exposition, vol. 60, 2000, pp. 59–63. [3] B. Li, L. Rong, Y. Li, V.E. Gjunter, Acta Mater. 48 (15) (2000) 3895–3904. [4] A. Chernyshov, M. Leroux, M. Assad, A. Dujovne, E. Garcia-Belenguer, In: D. Mantovani (Ed.), Advanced Biomaterials for Biomedical Applications, Proceedings of the Conference Metallic Society Canadian Institute of Mining, Metallurgy and Petroleum (CIM), 2002, pp.109–119.
[5] J.A. Shaw, A. Gremillet, D.S. Grummon, Proceedings of the ASME Adaptive Structures and Materials Systems Symposium, ASME winter meeting, New Orleans, LA, November 2002. [6] D.S. Grummon, J.A. Shaw, A. Gremillet, J. Appl. Phys. Lett. 82 (2003) 2727. [7] I. Cherghescu, V. Constantin, Chem. Mater. Sci. 60 (3–4) (1998) 137–141. [8] Z. Dong, W. Liu, D. Jia, Z. Tang, D. Wang, Rare Met. Mater. Eng. 29 (3) (2000) 182–184. [9] K.P. Gupta, Phase Diagrams of Ternary Nickel Alloys. Part 2. Section D.2. The Nb–Ni–Ti (Niobium–Nickel–Titanium) System, Indian Institute of Metals, Calcutta, 1991, pp. 163–176. [10] S.B. Prima, L.A. Tret’yachenko, V.M. Petyukk, Powder Metall. Met. Ceram. 34 (1995) 155. [11] L.C. Zhao, T.W. Duerig, C.M. Wayman, Proceedings of the MRS International Meeting on Advanced Materials, vol.9, 1989, Shape Memory Materials, Tokyo, Japan, 31 May–3 June, 1988, pp. 171–176. [12] T. Onda, M. Piao, Y. Bando, H. Ichinose, K. Otsuka, Mater. Trans. JIM 36 (1995) 23–29. [13] M. Piao, S. Miyazaki, K. Otsuka, N. Nishida, Mater. Trans. JIM 33 (1992) 337–345.