Ti-Ni Shape Memory Alloys I 1035

I Ti-Ni Shape Memory Alloys T.W. Duerig and AR. Pelton, Nitinol Davelopnwnt Corporslion

This datasheet describes some of the key prop erties of equiatomic and near-equiatomic tita- nium-nickel alloys with compositions yielding shape memory and superelastic properties. Shape memory and superelasticity per se will not be re- viewed; readers are referred to Ref 1 to 3 for basic information on these subjects. These alloys are commonly referred to as nickel-titanium, tita- nium-nickel, ‘Ibe-nee, Memo&e”, Nitinol, ‘Duel”, and FlexonTU. These terms do not refer to single al- loys or alloy compositions, but to a family of alloys with properties that greatly depend on exact com- positional make-up, processing history, and small ternary additions. Each manufacturer has its own series of alloy designations and specifications within the “‘D-N? range.

A second complication that readers must ac- knowledge is that all properties change sign& cantly at the transformation temperatures M,, Mf, A,,, and Af (see figure on the right and the section “Tensile Properties”). Moreover, these tempera- tures depend on applied stress. Thus, any given property depends on temperature, stress, and his- brY.

Effect of phase transformation

Schematic illustration of the effects on a phase transform* on lha physkal pmpartlaa of X-Ni. All physkal pmpehs exhibit a dis- continuity. charactetfzed by the transformatlal lenlparatures shcual. Source: CM. Wayman and T.W. Duetfg, En@n&ng Aspsds of Shape Emory Alloys, TW. Dumlg, ef al., Ed., Buttenwnlh-Heine maM.1ggo,p10

Product Forms Titanium-nickel is most commonly used in the and form of cold drawn wire (down to 0.02 mm) or as

Applications barstock. Other commercially available forms not yet sold as standard product would include tubing (down to 0.3 mm OD), strip (down to 0.04 mm in thickness), and sheet (widths to 500 mm and thick- nesses down to 0.5 mm). Castings (Ref 41, forgings and powder metallurgy (Ref 5) products have not yet been brought from the research laboratory.

‘Apical Conditions. !IQanium-nickel is most commonly used in a cold worked and partially an- nealed condition. This partial anneal does not re- crystallize the material, but does bring about the onset of recovery processes. The extent of the post- cold worked recovery depends on many aspects of the application, such as the desired sti&ess, fa- tigue life, ductility, recovery stress, etc. Fully an- nealed conditions are used almost exclusively when a maximum M, is needed. Although the cold worked condition does not transform and does not exhibit shape memory, it is highly elastic and has been considered for many applications (Ref 6).

Response to Heat ‘Ikeatment. Recovery processes begin at temperatures as low as 275 “C (525 “F). Recrystsllisation begins between 500 and 800 “C (930 and 1470 “F), depending on alloy com- position and the degree of cold work.

Aging of unstable (nickel-rich) compositions begins at 250 “C (525 “F), causing the precipitation of a complex sequence of nickel-rich precipitates (Ref 71, as these products leach nickel from the ma- trix, their general effect is to increase the M, tem- perature. The solvus temperature is about 550 “C

(1020 “F). Applications for titanium-nickel alloys can be

conveniently divided into four categories (Ref 8): . Free recovery h&ion) applications are those in

which a shape memory component is allowed to freely recover its original shape during heatr ing, thus generating a recovery strain (Ref 9).

l Constmined recovery (fame) applications are those in which the recovery is prevented, con- straining the material in its msrtensitic, or cold, form while recovering (Ref 9). Although no strain is recovered, large recovery stresses are developed These applications include fas- teners and pipe couplings and are the oldest and most widespread type of practical use.

l Actuators (work) applications are those in which there is both a recovered strain and stress during heating, such as in the case of a titanium-nickel spring being wsrmed to lift a ball (Ref 10). In these cases, work is being done. Such applications are oRen forther catego- rized according to their actuation mode, e.g., electrical or thermal.

l Superelasticity (energy stomge) refers to the highly exaggerated elasticity, or springback, observed in many Ti-Ni alloys deformed above & and below & (Ref 11). The function of the material in such cases is to store mechanicel energy. Although limited to a rather small tem- perature range, these alloys can deliver over 15 times the elastic motion of a spring steel.

Special Many shape memory-related properties are temperatures, superelasticity, etc.). Some proper- Properties discussed in subsequent sections (transformation ties, however, are strictly peculiar to shape mem-

1036 / Advanced Materials

ory alloys and cannot be conveniently categorized in standard outline forms. The more important of these properties are discussed below.

Free-recoverable strain in polycrystalline titanium-nickel can reach 8%, but is limited to a maximum of 6% if complete recovery is expected.

Applied stresses opposing recovery reduce recoverable strain. Clearly, stronger alloys will be af.Tected less by opposing stresses. Work output is maximized at intermediate stresses and strains.

Recoverable stresses generally reach 80 to 90% of yield stress. In fact, ahoy behavior depends on numerous factors, including the compliance of the resisting force and the constraining strain (Ref 9 and 12). Typical values are as follows:

cmdttion Recoverv-.ma

Anoealed-k Coldwrked~kanwated

at5oo”C(93o”F) Ccldwakedwimanncaledat

4aw(75ov)

400 700

Iwo

Effects of opposing stres8es on recovery strain Work output of a TI-Ni alloy

TcNI-Febarstodcwith5oat%Niand%FeMlyan~,tested inunhxtalteneion. SDUIW: J.L Pmft and T.W. Duet@. En@eedngAspeds of- Mmncxy AJoyq TW. Duartg et al., Ed., htmwrhHeinemann. LorKwl, 1990,p115

Ti-Ni-Fe barstock with 50 at.% Ni end 3% Fe Mty annealed. tested in uniaxhl tension. After defuming libNi to various total strains (x- axis), the material sprtngs back to the plastic stratn levels shown by the opan drdes. After heating above A,, most of the strain is recov- ered,butsomeamnesiapeiststs.Thedffbmw -tf=+pltls- llcstrsinandtheamnaelalsthe mcowable atraln (cksad olrdes). 0ourw: J.L. Proft and TW. Duerig, En@nwiq Asjxcfs of shepe Memory Atbys T.W. Duerig et el.. Ed., Buttewotth-!-leinemann. Lc+ldcn,199o,p115

TG4iiFe barsbzk with 50 at.% Ni and 3% Fe in a work-ha&n& wndrbon, teatad in uniaxial tension. 0ame: J.L. Pmft and T.W. Duerig. E@esfQ AspecLF ofshepe z,zp;lF Dudg et ai.9 Ed.. 0 -Hainemann,

9 I

Density, 6.45 tc 6.5 s/cm3

Titanium-nickel is extremely sensitive to the memory alloys in the range of49.7 to 50.7 at.%. Bi- precise titanium/nickel ratio (see figure below). nary alloys with less than 49.4 at.% titanium are Generally, alloys with 49.0 to 50.7 at.% titanium generally unstable. Ductility drops rapidly as are commercially common, with superelsstic al- nickel is increased. loys in the range of 49.0 to 49.4 at.% and shape Binary alloys are commonly available with &

OfTlNiPhase Diasram

temperatures between -50” and +lOO “C (-58 to 212 “F). Commercially available ternary alloys are available with M, temperatures down to -200 “C (-330 “F). Titanium-nickel is also quite sensitive to alloying additions.

Oxygen forms a Ti,NizO, inclusion (Ref 13), tending to deplete the matrix in titanium, lower M,, retard grain growth, and increase strength. Levels usually are controlled to ~500 ppm. Nitro- gen forms the same compound and has an additive effect to oxygen.

Fe, AI, Cr. Co, and V tend to substitute for nickel, but sharply depress I& (Ref 14 to 161, with V and Co being the weakest suppressants and Cr the strongest. These elements are added to sup- press M, while maintaining stability and ductility. Their practical effect is to stiffen a superelastic al- loy, to create a cryogenic shape memory alloy, or to increase the separation of the R-phase from martensite.

Pt and Pd tend to decrease M, in small quan- tities (-5 to lo%), then tend to increase M,, eventu- ally achieving temperatures as high as 350 “C (660 “F) (Ref 17).

Zr and Hf occasionally have been reported to increase M,, but are generally neutral when sub- stituted for titanium on an atomic basis.

Nh and Cu are used to control hysteresis and

Ti-Ni Shape Memory Alloys / 1037

Effect of composition on Ms

M, temperatures in nickeManium alloys are exbwdy sensbe to ccrnpositionel vadatlcn, particulady at hi@er nickel anMnts. Bourn:: KN. Melton, E@~AspedsdShepeMcmayAC lays, TW. Duertg et&., Ed., Buttenw+ Heinemann,1990,plO

martensitic strength. Nb is added to increase hys- teresis (desirable for coupling and fastener appli- cations), and copper (Ref 19) is added to reduce hysteresis (for actuator applications).

ClpttS/ The high-temperature austenitic phase @)-has a is worth noting that there is ah a Yransition” stnlctllre B2, or CsCl ordered structure with a, = 3.015 A. The structure that preceded the martensite, called the

most common martensitic structure (BlY) &m a Rphase with a rhombohedral structure (Ref 21). complqx monoclinic etructurewitha=2.889qb= Although this R phase exhibits a number of inter- 4.120 A, c = 4.622 A, and p = 96.8” (Ref 20). The M, can range Corn e-200 to +lOO “C (-328 to 212 “F). It

e$$ ymmrties, it will not be reviewed exten-

1038 / Advanced Materials

Time-temperature-tmnsformation curve

nll?etenlperature- tionwrvetorTI-51NI,whichshowspredpitetian~sasafunctionoftemperetureandtime. Bourw: M. Nishida. CM. Wayman, andT. Hcwna, M&M. Trans. 4 Vd 17.1986, p 1506

Transformation The T-T-T diagram shows the aging reactions time increases at a constant temperature. These Products in unstable (>50.6% Nil titanium-nickel alloys precipitation reactions can be readily monitored

(Ref 7). In general, TiNi + TillNil --B TizNia + via transformation temperature or mechanical- TiNis as the aging temperature increases or as property measurements.

Damping Characteristics

Elastic Constants

Internal friction and damping of titanium- nickel alloys are dramatically affected by tem- perature changes (see figure on left). Cooling (or heating) produces peaks, which correspond to the transformation temperatures. At higher tempera- tures, a very sharp increase is observed during

Dynamically measured moduli (R.ef 23 and 24) change markedly with the martensitic transfor- mation and premartensitic effects (see figure on

Ti-Ni-Cu alloy damping charactewistics

lntemal f?tdton d 44.~2g.3Nii26 Cu (WI%) dudng cooling nith muaamnmt fmqumcy d -1 Hz. Swrce:O.MerderendE.~J.Phys.,VdC4(No.43),1~,~ C-4270

cooling through the &. These usually high damp ing characteristics (Ref 22) have been studied for some time, but have not been used on a commer- cial basis due to their limited temperature range and rapid fatigue degradation.

right) .l$icalvaluesofelasticmoduliare40GPa (5.8 x 10 psi) for martensite and 75 GPa (10.8 x lo6 psi) for au&mite. From a practical point of

Dynamic Young’s modulus vs tempwatura

A T-SNi (wt%). 8: 44.Tri.~ (w%). c: 44STb51.7Nt- 3.4t=u (wwo). Bwrce:O.rutdw,KN.~,R~mdAKdlk,PlocM Ccnf tickMoW- Te H.I. m D.E L&J@ Ih,R.F.~andC.M.~,Ed,AIME,1982,p1258

Electrical Resistivity

Magnetic Characteristics

view, however, modulus is of little value; apparent elasticity is more controlled by the transformation and by mechanical twinning. Poisson’s ratio, p, is 0.33 (Ref 25).

General values for electrical resistivity of the two primary phases are as follows (Ref 26):

p (martensite) = 76 x lo-6 R. cm p(austenite)=82x WeR.cm Variations in resistivity with temperature are

complex functions of composition and ther- momechanical processing (see figure). Note also the pronounced effect of the R-phase on resistivity.

Magnetic susceptibility also undergoes a dis- continuity during phase transition (Ref 26). ‘&pi- cal values are:

x (martensite) = 2.4 x lO$ emu/g x (austenite) = 3.7 x lO& emu/g

Titanium-nickel generally forms a passive TiOz (x-utile) surface layer (Ref 27). Like titanium alloys, there is a transition temperature of about 500 “C (930 “F), above which the oxide layer will be dissolved and absorbed into the material. Unlike titanium alloys, however, no a case is formed. Tita- nium-nickel will also react with nitrogen during heat treatments, forming a TiN layer.

Ti-Ni Shape Memory Alloys I 1939

Electrical rssistance vs tempersturs

traTl1000~C(1830”F).B:Quenchedfrom10000C(1830”F),eged at4000C(750”F).C:Direcllyagedat400”C(750”F).T,isthebarr sithtemperatumhom austanitebtha -ralRphase.rR is tha shiftad bansition tamperatura tram prowssing effacts. Arbi- balyunitsfctrekcmcalresistance. !Soutw: S. Miyazaki and K. Otsuka. Me&U. Trans. A.Vd 17,1986. Pm

General The rest potential of titanium-nickel in a dilute Corrosion sodium chloride solution is around 0.23 V (SCE),

which compares with 0.38 V for type 304 stainless steel. This puts titanium-nickel on the noble or protected side of stainless steel in the galvanic se- ries. A passive oxide/nitride surface film is the ba- sis of the corrosion resistance of titanium-nickel alloys, similar to stainless steels. Speci.& environ- ments can cause the passive 6lm to break down, thus subjecting the base material to attack. Asum- mar-y of titanium-nickel reactions in various envi- ronments follows (Ref 28).

Seawater. Titanium-nickel is not sf&xted when immersed in flowing seawater; however, in stagnant seawater, such as found in crevices, the protective film can break down, which results in pitting corrosion.

Acetic acid (CH&OOH) attacks titanium- nickel at a modest rate of 2.5 x 10m2 to 7.6 x 1W2 mm/year (mpy) over the temperature range 30 “C (86 “F) to the boiling point and over the concentra- tion range 50 to 99.5%.

Methanol (CHsOH) appears to attack tita- mum nickel only when diluted with low concentra- tions of water and halides. This impure methanol solution leads to pitting and tunneling corrosion similar tc that found in titanium alloys.

Cupric chloride (CuC12) at 70 “C (160 “F) at-

tacks titanium-nickel at 5.5 mpy. Ferric chloride (FeCl$ at 70 “C (160 “F) and

8% concentration attacks titanium-nickel at 8.9 mpy. Titanium-nickel is attacked at a rate of 2.8 mpy in a solution of 1.5% FeCls with 2.5% HCl.

Hydrochloric acid (HCl) has a variety of ef- fects on the corrosion of titanium-nickel alloys de- pending on temperature, acid concentration, and specific alloy composition. With 3% HCl at 100 “C (212 “F) and a range of ahoy compositions, the rate of attack was as low as 0.36 mpy and as high as 3.3 mpy. At 25 “C (77 “F) and 7M solution, titanium- nickel-iron alloys can lose up to 457 mpy.

Nitric acid (HN03) is more aggressive toward titanium-nickel than type 304 stainless steel. At 30 “C (86 “F), 10% HNO3 attacks at a rate of 2.5 x 1W2 mpy; 60% solution attacks at 0.25 mpy; 5% HNO3 at its boiling point attacks at 2 mpy.

Biocompatibility studies have been con- ducted in various media chosen to simulate the conditions of the mouth and the human body In general, no corrosion of titanium-nickel alloys has been reported. For example, in tests where cou- pons of titanium-nickel were sealed at 37 “C (97 “F) for 72 h, the mass corrosion rate was on the order of 10-5 mpy for such media as synthetic saliva, syn- thetic sweat, 1% NaCl solution, 1% lactic acid, end 0.1% HNaSOd acid (Ref 29); see also Ref 30.

HycrroSen The interaction between hydrogen and tita- in excess of 20 ppm by weight can be considered Damage mum-nickel is sensitive to both concentration and detrimental to ductility, with levels in excess of 200

temperature (Ref 31). In general, hydrogen levels ppm severely impairing. Under certain conditions,

1040 / Advanced Materials

hydrogen can be absorbed during pickling, plat- absorbed in hydrogenated water at elevated tem- ing, and caustic cleaning. The exact conditions re- peratures and pressures, such as would be found quired for hydrogen absorption are not well de- in pressurized water reactor primary water sys- fined, so it is advisable to exercise care when terns. Relatively short exposure times have been performing any of these operations. shown to produce hydrogen levels well in excess of

Substantial amounts of hydrogen also can be 1000 ppm (Ref 32).

Heat Capacity A typical plot of specific heat (C ) versus tem- perature for a 50.2% Ti ahoy (see if gure) shows a discontinuity at the M, temperature of 90 “C (195 “F) (Ref 3). The peak and onset temperatures for the peaks are often used to characterize the trans- formation temperatures of an alloy Care must be taken however, (Ref 33), because the presence of an R-phase prior thermal cycling, and residual stresses f?om sample cutting can tend to compli- cate the curves and introduce spurious peaks.

Latent The latent heat of the martensitic transforma- Heats tion strongly depends on the transformation tem-

perature and stress rate (do/dT) through the for- mula

dold’l’ = AHI(AE?“) ‘ljpical values for AEZ are 4 to 12 Cal/g and val-

ues for doId? range from 3 to 10 MPa/“C. The latent heat of fusion can be expressed as:

AH = -34,000 J/m01 (Ref 34).

Thermal The thermal coefficient of linear expansion can Expansion be expressed as (Ref 23):

a (martensite) = 6.6 x 10-W a (austenite) = 11 x 10-W The volume change on phase transformation

(Av) (from austenite to martens&e) is -0.16% (Ref 35).

Specific heat (C,)

Speclflc heat of li-49.9Ni (at.%), with a sharp peak in the spedlc ~tatso~(l950F)correspondingtolheM,temperature. Source: CM Jackson, H.J. Wagwr, and R.J. Wasilewsld, NASA Report, NASASP 5110.1972

Melting Point T,,, = 1310 “C (2390 “F)

Martensitic Characteristic transformation temperatures Transformation depend strongly on composition (see table on

Temperatures next page and the previous section on chemis- try). ‘l)pical hysteresis widths range from 10 “C (18 “F) for certain titanium-nickel-copper alloys, to 40 to 60 “C (72 to 108 “F) for binary alloys, to 100 “C (180 “F) for titanium-nickel-niobium alloys.

Wmsformation temperatures are measured by a number of techniques, including electrical re- sistivity, latent heat of transformation by differen- tial scanning calorimetry, elastic modulus, yield strength, and strain. However, the most useful measurement technique is to monitor the strain on cooling under a constant load and the recovery on heating.

Other important relationships of transforma-

tion temperatures are as follows. Applied stresses increase transformation temperatures according to the stress rate (see the next section on tensile properties). Martensitic deformations increase the stress-free A, temperatures, particularly in al- loys with low yield stresses. The increase is tempo- rary, returning to the previous value after the 5rst heating cycle. Increasing cold work tends to reduce transformation temperatures. The R-phase trans- formation temperature is much more constant than those for martensite, typically 20 to 40 “C (68 to 105 “F) in binary alloys.

fir which is defined as the temperature above which martensite cannot be stress-induced, may be about 25 to 50 “C (50 to 100 “F) higher than AC

Strain of a Ti-NMlb specimen MB temperature as a function of cold working

Strain after deforming and unkxding, measured M the first and sewndheatingcydes.Notethechangsin~andthereawty strain. Scum: K.N. Melton. J.L. P&t, and TW. Duerig. MRS ht. Meeting on Advanced Materials, Vd 9. K. Otauka and K. Shimizu, Ed., Mab ltalsResearchSocwy,1989,p165

Ti-Ni Shape Memory Alloya I 1041

ChangeintheM,temperatumofaT~50.BNiallqrcoklworked92 1040%andsubsaquefltly~at500”c(930”F)for30min. Soutm: G.R. Za&o andTW. Duerig, vlpublii research

Ti-Ni shape memory transformation temparatures M,,, which is defined as the temperature above which martensite cannot be stress induced, may be about from 25 to 50 “C (50 to 100 “F) higherthan I+.

[71-l 46.6 47.6 49.6 50.2 51 51.5 52.8 49.4 49.7 50.4 49.7

WWW

[79&I

50 44 So.1 IO 50.5 -9. -29 51 2oto25

-29

48.1 100 60 48.6 IO1 74 49.0 66 16 49.5 47 19 50.5 5 -31 51.0 -52 -85

57 37 33

-51 -136

4 28 57 20 -30 45

12 I8 13 30

-178 -38 -14 5

-20 -53

81 79 75 33 0

-12 44 63 39

-12 67

117 134 I14 32 -94 46 278 I06 77 0

I20 52

21 60

I23 I40 178 153 56 93 53 80 8 44

-39 -34

DTA(aweceived -)

pmperdei Electrical nxistivity

Compilation hm Phase Diagrams ofBinary 1IfaniumAbys, (J.L. Murray, Ed), ASM International, 1987. p 203. (a) Cited references are aa foUows: 71 Kor: 1.1. Komihv, Ye. V. Kachur, and 0.K Beloww, TXlatation Analysis of’hnsformation in the Compound lJNi,“Fiz I&t. ikfelollooed., 32(2), 42042 (1971) in Ruwiaq TR: Phys. Met. Mew., 32(2), HO-193 (1971). 81 Mel: KN. Melton and 0. Memier, ‘The Mechanical Properties of NiTi-Based Shape Memory Alloys.” Acta Metal., 29.393-398 ( 1981). 80 MiI: RV. hUigan, %t.ermhation of Phase Transformation ~mperatmw of ‘I’iNi Using Differential Thermal h&h,” Titanium ‘80, ‘IF Sci. ‘I&h., Pnx ht. fhnf. Kyotq Ja- pm, May 1822, T. Kimuzi, Ed., 1461- 1467 ( 1980). 68 Wan: EE. Wang, B.E De&wage. and W.J. Buehler, The Imsereible Critical Range in the TiNi lkansition.“J. Appl. Phys.. 396). 21662175 (1968). 79 Che: D.B. Chernov, Yu.1. Pa&aI, VE. Gyunter, LA MO naswich. and E.M. Savitakii, The Multiplicity of Structural ‘kansitions in Alloys Based on ?wi,” Lkkl. Akad. Nash SSSR, 247.864867 (1979) in Russian, TRZ sov. Phye. Dokl., 24(E), 664466 (1979)

1042 / Advanced Materials

In general, a superelastic curve is charac- terized hy regions of nearly constant stress upon loading (referred to the loading plateau atrees) and unloading (unloading plateau stress). These pla- teau stress values are better indicators of me- chanical strength than the traditional yield stress. Typical values are shown (see table).

Ti-Ni shape memory: Typical loading and unloading characteristics

450 to 700 MPa Upto250MFQ 11% 6%

40-50 J/cm’

Source: TW. Duering and G.R. Zadno, Engineering Aspects of Shape MemoryAuqys. T.W. Dumiget al., Ed, But&worth-Heine- maon, London, 1990, p 369

AboveMt a is defined as the temperature above which martensite cannot be stress-induced. Conse- quently, titanium-nickel remains au&mite throughout an entire tensile test above M,-J. Tensile strengths depend strongly on alloy condition, and the ultimate tensile strength, yield strength, and ductility of cold worked titanium-nickel wire de- pend on final annealing temperatures (see figure).

Ductility drops sharply as compositions be- come nickel-rich. A review of other factors control- ling ductility can be found in R.ef 36.

Mechanical properties vs anneal temperature

Theinhenceofanr&ingtemperature0nme&ankApmperties0t 0.5 mm (0.02 in.) Ti50.6Ni wire with 40% aM work and annealed 30 min at temperature. Source: G.R. Zactno and TW. Duettg, unpublished research

Schematic of superelastlcity descriptors

Schematic diagram showing key desaipturs of superelsstidtr =u (unloadingpla~umeasuredaslheinfiectionpoint),q(loadkrg~ @au meaeured as the irhcthcm point), 4 (tote deformation sbain). Ed (pemmnmt set, or amnesia) and the stomd energy (shaded =W Same: TW. Duet@ and G.R. i%ho, En@aMn~ AqasUs d ShqmMemtny Allays, TW. Due@ et&.LI., Ed., Buth%w&w mann,Londcn,1990,p369

Aging of nickel-rich alloys increases austenitic strength to a typical peak strength of 800 MPa (116 ksi). Surprisingly, ductility is also increased dur- ing the aging process.

Yield strength vs anneal temperature

Theinilumceofarwahg~onmech&tl poperliesorn- 50.6Nibvinl40%culdwJtlta30minBttemperature. Source: G.R. Zadno andTW. Dumig, unpublished reseamh

Titanium-nickel yield stresses are controlled by the “friction” of the martensite twin interfaces. Typical yields stresses are 120 to 160 MPa (17 to 23 ksi) for binary alloys and as low as 60 to 90 MPa (9

to 13 ksi) for titanium-nickel-copper alloys. Ulti- mate tensile strengths and ductilities are similar to austenitic values.

Ti-Ni Shape Memory Alloys I 1043

Effect of Between M, and M,-J, the material transforms Temperature from au&e&e to martensite during tensile test-

ing. Yield strengths vary continuously from M, to & (see figure). The rate of stress increase is called the stress rate, varying from 3 to 20 MPaPC, with rates generally increasing with w.

Superelasticity, or pseudoelasticity, is an en- hanced elasticity when unloading between A, and M,+ The Md transition is generahy defined as the temperature above which stress-induced marten- site can no longer be formed. On a streee-tempera- ture graph, Md is the temperature where the stress begins to level off.

Superelasticity is also highly temperature de- pendent (see figures). Changing alloy composition and heat treatment can shift the temperature range of superelastic behavior from -100 to +lOO “C (-148 to 212 “F).

‘his datasheet describes some of the key prop- erties of equiatomic and near-equiatomic tita- nium-nickel alloys with compositions yielding

Permanent se4 of supewelastic wire

Pemwlentsetofsupsralasticbinary~um-nidcelwtredefolmed 6.3% and unloaded at vattous temptatures. T-Ni wire with 50.6 at.%Nicddworked40%endannealedBt5M)OC(930n. Thesupemlaskwindwisrcu@y40”C(70°F)inMdth. Source: T.W. Dututg and G.R. Zadno. Ef@eehg Aspeds of Shape A4muryAUoys, TW. Duettg eta.., Ed., BuWwor&Heine mann,Londun,1990,p369

50.6at.%Nicddworked40%andanneeledat500”C(930’F)for 2min.Stressrate=5.7MPaPC. Source: T.W. Duerig and G.R. zadno. IZ@eehg aspactp of Shape Mmtny Alloys, TW. lhettg eta/.. Ed., ButteMlorlh-Haine -,London,199o,p~

shape memory and superelastic properties. Shape memory and super-elasticity per sc will not be re- viewed; readers are referred to Bef 1 to 3 for basic information on titanium-nickel, lbe-nee, MemoriteTM, Nitinol, Tinem, and FlexonTM. These terms do not refer to single alloys or alloy compositions, but to a family of alloys with proper- ties that greatly depend on exact compositional make-up, processing history, and smah ternary ad- ditions. Each manufacturer has its own series of day designations and specifications within the “‘B-N? range.

A second complication that readers must ac- knowledge is that all properties change signifi- cantly at the transformation temperatures Mg, Mf, &, and Af (see 6gure). Moreover, these tempera- tures depend on applied stress. Thus any given property depends on temperature, stress, and hie-

tory. Superelasticity is an enhanced elasticity oc- curring when unloading between & and M,J (see the section “‘lbneile Properties” in this datasheet.)

Loading and unloading plateau heights in Ti-Ni wire

1044 / Advanced Materials

Stress relaxation

Creep. Very few creep measurements have been made on titanium-nickel, although creep mechanisms have been proposed (Ref 37).

Stress relaxationmis mcritical in many con- strained recovery applications. Several measure- ments have been made (Ref 9,12,38). ‘lb summa- rize, relaxation of stable titanium-nickel alloys is extremely slow below 350 “C (660 “F) and becomes very rapid by 425 “C (795 “F).

Stressrel~~ofaNi,,Ti,F~alloyat375OC (705°F)measured in a dynamically controlled seep machine. Round spedmens with gags length of 6 mm, fully ann&ed. Source: J. Proft and TW. Duetig, Er@needng Aspeds of shepe Memory Alloys, T.W. Duerig et al., Ed., BuMworth-Heinemann, 199o,p115

Stress Controlled

Strain Controlled

Thermal Fatigue

The fatigue behavior of titanium-nickel alloys is extremely complex and encompasses many dif- ferent topics. ‘lb summarize, titanium-nickel ex- cels in low-cycle, strain-controlled environments, and does relatively poorly in high-cycle, strese-

controlled environments. The superelastic mecha- nisms accommodate high strains without exces- sive stresses, but cannot accommodate high stresses without excessively high strains.

Isotherms stress-controlled testing is impor- sively characterized by the coupling and fastener tant in many constrained recovery applications in industries, much of the data remains largely un- which the shape memory effect is only used as a published or highly application specific. Typical S- mode of installation, e.g., fasteners and couplings, N behavior for various alloy compositions tested in which the alloys are used exclusively above the well above their M,j temperatures is shown below M,j temperature. Although fatigue has been exten- (see figure on next page).

Isothermal strain-controlled behavior is highly dependent on the testing temperature relative to the alloy transformation temperature. Fracture follows the Co&-Manson relationship:

N@At+=C where C and 8 are constants; N is the number of cy- cles to failure; and AQ, iB the applied strain ample- tude (see figure). One specific example of particu-

lar interest is superelastic cycling (strain-control- led testing between A, and MJ. There are several modes of degradation that occur under these cir- cumstances (see figure). Again, these depend strongly on specific alloy conditions and the needs of the application. Further data are provided in Ref 39.

Thermal cycling both with and without applied loads such as found in actuators or heat engines, is certainly the most complex mode to analyze. Al- though it has been studied extensively, it remains diEcult to predict failure, largely because failure can consist of fracture, ratchet&g, migration of transformation temperatures, changes in reset force, etc. Moreover, damage accumulation de- pends on B~XBB, strain, temperature change, heat-

ing methods, heating and cooling rates, and even orientation (horizontal or vertical).

Failure too is nebulous. Various degradation modes can occur, including a shift in transforma- tion temperature, a reduction in the available strain, walking (or ratcheting), and fracture itself. Thermal cycling with no applied load can even re- sult in some degradation. See Ref 40 to 43 for fur- ther information.

TI-Ni Shape Memory Alloys I 1045

Typical S-N curves

T~ium-nickel alloys of vartous compo&lkms (at.%) prepared in veriousfashiarsisisothermallytestedinsheetform.Testingisdone well abmm M,. Sheet spedrnens loaded awl unloaded (R= 0). Cy- dedat6O”C(UO”F). source: s. M@zakl, Y. sugaya, and K otsuka. MRS ht. Meeting on Advancd Materials, Vd 9, K Otsuka and K. Shimizu, Ed., Mate rtals Research sodety, 1969, p 257

Effects of isothermal superelastic cycling

(a)

Low-cycle fatigue behavior

Lowcyde fat@s behavior of tttanium-nicksl alloys at mom tern-- pemture tested in tensioncompression(f?=-l).TheM,tempera- tures of the alloys am indicated. Source:K.N. MeltanandO. Merck. Sfnm@o~IbMIbandW p. Haasen et&.., Ed.. Peqamcn Press, 1979, p 1246

W

(d Tests of a superelasttc titanium-nickel tire at three temperatures. Alloy contains 50.6 at.% nickel. Three mcxles of degadatton occur simuitane ously duttng the isothemA superelastic cyciing of tttanium-nickel alloys: wakng. or an accumulation of pemwwnt set (top), a charge in yletd stress(mk!de)andarecMtbninthehysteresiswidth(tottom).

1046 / Advanced Materials

Fatigue Crack Stress-inducing martensite at the tip of a slow crack growth (F&f 441. Other sources of data Propagation propagating crack has been shown to signitIcantly include Ref 45.

Toughness Although Charpy impact testing on titanium- sharp toughness minimum exists at h& (see 6g- nickel has been conducted (Ref 3,461, very little KI, ure). or & data exist. Indications are strong that a

Fatigue crack propagation

Fa~uecraclcpropegationratesinfourtitaniumni~~lkys,~steblemartensJtewithM,>roomtemperahrre,~eusftnideHlithM, c-z roan twlpemtum. krewnwe sbess-induoedmerdensitewlth M, c morn temperaturn 4 4. and mwrsiUe s~n3&w&fewith 4 < rccmtemperatumcM,.TestedwithR=0.1 onlOmmthickCTspecimensat50Hzundercondtlonsofdecreasing~ Source: R.H. Dauskardt.T.W. Duerig. andR.0. Ritohii. MRSInt. Meetingon Advancd Materials, Vd 9. K Otsuka and K. Shimizu, Ed., Mater& Research society, 1989, p 243

Bulk Working Plow Stress. Upset forging tests conducted on seven different alloys at strain rates ranging ti-om 10d to 102 indicate that the material has a very high hot ductility and is not highly strain rate de- pendent. The flow stresses can be characterized by the relation:

o = kim exp(-QlRT)

where (r is the flow stress; i is the strain rate; Q is an activation energy (50,000 cal/mole); and m and k are caMants. Values form were found to range from 0.1 t.0 0.17 with increasmg temperature (see Egure for typical data of an equiatomic binary alloy).

Extrusion. Both solid bar and large tube (50 mm diameter with 8 mm wall thickness) have been extruded from titanium-nickel. Parameters re- main proprietary.

Forging. Titanium-nickel has been succes- fully forged into large cups. Parameters remain proprietary.

Rdling. Titanium-nickel can be hot rolled with relative ease, but is diEcult to cold roll, espe- cially in thin, wide sections.

Fracture energy VI temperature

Vacuuminducticnrns4tedalloyhotwwkedandanneakdfori hat !35O”Candairwoled.Star&rd charWv-mpedspedmens were machined. The infiuenca of temperature on fmc3um to@+ ness of 44Twl9NGCu-2Fe (WI%). Note that the mlniium in ths fradumenecgyoccutsjustb&wM,,asdeWmMdhwnthecure ~JOII& 02% yield stmngth and ultimate tensile strength rneas

Source: IiN. Melton and 0. Merder, Act9 AMal. Vd 291962, p 393

Fabrication Casting. Although some unreported experi- ments have been conducted, casting has not been done on a commercial level.

Powder Metallurgy. Although a great deal of experimentation has taken place with both ele- mental and prealloyed powders, nothing has reached near-production levels. Reference 5 pro- vides a review of methods.

Forming. Titanium-nickel sheet has been suc- cessfully formed into a range of complex shapes, both in the martensite and austenitic phases. Springback is high, as is die wear and h-i&ion. Pa- rameters remain proprietary.

Machining. Titanium-nickel is very difRcult to machine. Very low speeds and a great deal of coolant is required, and tool wear is very rapid. Milling and drilling are particularly difIicult. Pro- ducers of couplings have demonstrated that large- scale machining production is possible.

Heat Treatment. Titanium-nickel can be heat treated in air up to -500 “C (930 “F). No a case is formed, but a surface oxide of r-utile develops quickly. Above 500 “C (930 “F), the olride layer be- gins to flake (depending on time). Nitrogen and hy- drogen atmospheres are not recommended. Argon, helium, and vacuum heat treatments are com- monly used to preserve bright finishes.

Recrystallixation is extremely rapid above 700 “C (1290 “F). Solution treatment requires tem- peratures of at least 550 “C (1020 “F). Stress relief is usually accomplished at temperatures as low as 300 “C (570 “F). A ‘ITT diagram is shown in the previous section “Phases and Structures.”

Fully annealed bar stock has a typical hard- ness of 60 HRA. Vicker numbers range &om 190 I-IV in the annealed condition to 240 HV after 15% cold work (Ref 3).

Ti-Ni Shape Memory Alloys / 1047

Flow strese measurements

Natural log (In) of Row strws is shovm to very hearty with the in- w.rseofabsolutetempemture.12mmx6mmdiameWsp&nens testad in compmssion under isothermal oonMons (heated dies) at tha tampsratursa and strain ratea shorm. Source: T.W. Duettg, unpublished data

Joining. Titanium-nickel is di&ult to join be- cause most mating materials cannot tolerate the large strains experienced by the alloy. Most appli- cations rely on crimped bonds. It can be welded to itself with relative ease by resistance and TIG methods. Welding to other materials is extremely di&ult, although proprietary methods do exist and are practiced in large volumes in the produc- tion of eyeglass frames.

Brazing can only be accomplished &r re-en- forcement and plating. Again, methods are pro- prietary; large-scale production is practiced by the eyeglass frame industry.

1. T.W. Duerig et al., Ed., Engineering Aspects of Shape Memory Alloys, Butterworth-Heine- mann, London, 1990

2. J. Perkins, Ed, Shape Memory Eficts in Al- loys, Plenum Press, 1975

3. C.M. Jackson, H.J. Wagner, and RJ. Wasilewski. NASA Rewrt. NASA-SP 5110. 1972 ’

. T

4. J. Takaha&, M. Gkaxaki, K. HiroBhi, and Y. Furuta, private communication, 1984

5. T.W. Duerig, in Advanced Synthesis of Mat&- ads, J. Moore, Ed., 1993, to be published

6. G.R. Zadno and T.W. Duerig, Engineering As- pects of Shape Mevpy Alloys, T.W. Duerig et ~i4Butt.envorth-Hememann, London, 1990, p

7. M. Nishida. CM. Wavman and ‘I! Honma. Met- all. Z~WZS. A, V0117, i986, p 1505

8. T.W. Duerig and R.N. Melton, Proc. of SMA’ 86, C. Youyi et al., Ed., China Academic Publish- ers, 1986, p 397

9. J.L. PI& and T.W. Duerig, Engineering As- pects of Shape Menpry Alloys, T.W. Duerig et &.5 Butterworth-Hememann, London, 1990, p

10. A. Reeley, D. Stackel, and T.W. Duerig, Engi- neering Aspects of Shape Memory Alloys, T.W. Duerig et al., Ed, Butter-worth-Heinemann,

London, 1990, p 181 11. TW. Duerig and G.R. Zadno, Engineering As-

pects of Shape Memory Alloys, T.W. Duerig et al.. Ed. Butter-worth-Heinemann, London. 19&I, p 369

E. C.R. Such. “The Characterixation ofthe Rever- sion Stress in NiTi.” M.S. Thesis. Naval Post Graduate School, Monterrey, CA, i974

13. M.V Nevitt. nans. Met. Sot. AIME. Vol 218. 196O,p327’

14. D.M. Goldstein W.J. Buehler, and RX. Wiley, Effects of Alloymg upon Certain Properties of 55.1 Nitinol, NOLTR 64-235,1964

15. H.K. Eckelmeyer, Sandia Labs Report 74 0418,1974

16. D.B. Chernov et al.. Dokl. Akad. Nauk. SSSR. Vol245 (No. 2),1979, p 360

17. PG. Lindquist and C.M. Wayman, Engineer- ing Aspects of Shape Memory Allqs, T.W. Duerig et al., Butterworth-Hemema, Lon- don. 1990. D 58

18. ‘I!W: Due& end R.N. Melton. l!he Marten&e l+am$orm&n in Science and l’bchmlogy, E. Hornbogen and H. Jost, Ed, Deutache Gesell- achaR fib Metallkunde, Germany, 1988, p 191

19. W. Moberly and R. Melton, Engineenng As- pects of Shape Memory Allwys, T.W. Duerig et

1048 / Advanced Materials

al.. Ed.. Butterworth-Heinemann. London. i9hO,p46

20.0. Matsumoto, S. Miyazaki, K Otsuka, and H. ‘lhmura,Acta Metall., Vol35,1987, p 2137

21. H.C. Ling and R Kaplow, MetaU. lluns. A, Vol 12,1981, p 2101

22. L. Kaukxm. S.A. Kulin. P. Neshe. and R. Salzbrenner,‘Shu Me&r-y Effects h Alloys, J. Perkins, Ed., P r enum Press, 1975, p 547

23. S. Spinner and AG. Rozner. J. Acoust Sot. Am.;v0140,1966, p 1009 ’

24. R.J. Wasilewski, lkan-s. AIME, Vol233,1965, p 1691

25. O.K. Belousov, Russ. Met&., Vol2,1981, p 204 26. J.E. Hanlon. S.R. Butler, and R.J. ~Wa&vski,

lkms. AIM& Vol239,1967, p 1323 27. C-M. Chan. S. ‘lkitzwell. and T.W. Duerk. Surf:

Inte a~eAk,Va 15, isso, p 349 -’ ’ % 28. J.D. arrison, Ray&em Report, 1991

29. S. Lu, Engineering Aspects o Shape Memory Alloys, ‘I!W. Dueriget al., Ed., L utterworth-He- inemann, London, 1990, p 445

30. L.S. Castleman, S.M. Mot&in, El? Alicandri, V.L. Bonawit, and AA Johnson, J. B&d MateE Z&x, Vol 10,1976, p 695

31. R. Burch and N.B. Mason, J.C.S. Famday Z, Vol 75,1979, p 561; R.B. Burch and N.B. Ma- son, J.C.S. Famday Z, Vol75,1979, p 578

32. A.R. Pelton and T.W. Duerig, “An Analysis of Crofit Cou lings AtIer One-Year Service at Seabrook, “kychem Proprietary Report, 1991

33. G. Airoldi and B. Rivolta, Phys. Scrip& Vo137, 1988, p 891

~4. J.C. Gachon and J. Hertz, CALPHALI, Vo17, 1983, p 1

35. K Otsuka, ‘I! Sawamura, K Shimizu, end

gi& Wayman, Met&l. l?-ans., Vo12, 1971, p

36. S. Miyamki et al., Matel: Sci. Forum, Vol5668,

37. k%%%5eriee. J. ADDS. Phvs., Vo139 (No. 5). 1968,p2201- ’ -- -

36. A. l&ui et al.. Nio. Kikai Gakkai Ronbunshu. A. Hen, Vol5i&. 462), 1985, p 488 ’

39. G.R. Zadno, W. Yu, and T.W. Duerig, Materials Science Forum. Vol 56-58. B.C. Muddle. Ed. 199O,p771 ’

, ,

40. H. ‘Ibmura, Y Suzuki and T lbdoroki, Proc. Id. Cod Martensitic Transformations. Janan Institute of Metals, 1986, p 736 ’ *

41. Y. Furuya et al., MRS Int. Meeting on Ad- vanced Materials, Vol 9, K Otsuka and K Shimizu, Ed, Materials Reeearch Society, 1989, p 269

42. Y. Suzuki and H. lkmura, E Shape Memory A&w, T.W.

‘neerin.gAspects o B

%I erigetaZ.,Ed, utterworth-Heinemarm, London, 1990, p 256

43. J.L. Pmft, KN. Melton, and T.W. Duerig, MRS lit. Mee ’

9 on Advanced Materi~s, Vol9, K

Otauka an K Sbimizu, Ed., Materials Re- Bear& Society, 1989, p 159

41. R.H. Dauskardt, TW. Duerig, and RO. Richie, MRS Int. Meeting on Advanced Materials, Vol 9, K Otsuka and K Shimizu, Ed, Materials Research Society, 1989, p 251

45. S. Miyazaki, Y. Sugaya, and K Otauka, MRS Int. Meeting on Advanced Materials, Vol9, K Otmka and K Sbimizu, Ed., Materials Re- search Society, 1989, p 251

46. KN. Melton and 0. Mercier, Actu Metall., Vol 29,1982, p 393