EP0187452B1

EP0187452B1 - A method of processing a nickel/titanium-based shape memory alloy and article produced therefrom

Info

Publication number: EP0187452B1
Application number: EP85308080A
Authority: EP
Inventors: John A. Simpson; Tom Duerig; Keith Melton
Original assignee: Raychem Corp
Current assignee: Raychem Corp
Priority date: 1984-11-06
Filing date: 1985-11-06
Publication date: 1991-02-06
Anticipated expiration: 2005-11-06
Also published as: CA1269915A; DE3581721D1; EP0187452A1

Description

This invention relates to a method of treating a nickel/titanium-based shape memory alloy and to a method of making composite coupling which comprises a nickel-titanium based shape memory alloy driver member.
Materials, both organic and metallic, capable of possessing shape memory are well known. An article made of such materials can be deformed from an original, heat-stable configuration to a second, heat-unstable configuration. The article is said to have shape memory for the reason that, upon the application of the heat alone, it can be caused to revert or attempt to revert from its heat-unstable configuration to its original heat-stable configuration, i.e., it "remembers" its original shape.
Among metallic alloys the ability to possess shape memory is a result of the fact that the alloy undergoes a reversible transformation from an austenitic state to a martensitic state with a change of temperature. Also, the alloy is considerably stronger in its austenitic state than in its martensitic state. This transformation is sometimes referred to as a thermoelastic martensitic transformation. An article made from such an alloy, for example, a hollow sleeve, is easily deformed from its original configuration to a new configuration when cooled below the temperature at which the alloy is transformed from the austenitic state to the martensitic state. The temperature at which this transformation begins is usually referred to as M_s and the temperature at which it finishes M_f. When an article thus deformed is warmed to the temperature at which the alloy starts to revert back to austenite, referred to as As (A_f being the temperature at which the reversion is complete), the deformed object will begin to return to its original configuration.
Commercially viable alloys of nickel and titanium have been demonstrated to have shape-memory properties which render them highly useful in a variety of applications.
Shape-memory alloys have found use in recent years in, for example, pipe couplings (such as are described in U.S.P. 4,035,007 and 4,198,081 to Harrison and Jervis), electrical connectors (such as are described in U.S.P. 3,740,839 to Otte and Fischer), switches (such as are described in U.S.P. 4,205,293 to Melton and Mercier), etc.
It is, of course, advantageous to have the alloy austenitic at the service temperature which is often but not necessarily near room temperature, since the austenite phase is stronger than the martensite phase. In fact, it would be desirable to have the alloy remain austenitic over a wide range of service temperatures, for example from substantially below room temperature to substantially above room temperature, so that the alloy has practical utility.
As an illustration, Military Specification MIL-F-85421 requires a product that is functional to about -55°C. If the product comprises a shape memory alloy, then for convenience in shipping the product in the heat-unstable configuration, the product should not recover prior to about 50°C. It is a matter of commercial reality, within and without the military, that the product satisfy these requirements.
It is also desirable that the alloy be martensitic in the vicinity of room temperature so that the article can be fabricated, stored, and shipped at or near room temperature. The reason for this is that in the case of an article made from the alloy, a coupling, for example, the article would not recover prematurely.
US-3,948,688 discloses a process for conditioning a nickel-titanium alloy which comprises applying a load to the alloy and then, while the load is applied, thermally cycling the alloy between a temperature less than AS and a temperature greater than A_f. As a result of the termal cycling, the service life, and the elongation activity under stress, are improved.
Conceptually, one way to obtain an alloy that is martensitic near room temperature and which is also austenitic over a large range of temperatures including room temperature, is to have an alloy which exhibits a sufficiently wide tranformation hysteresis, say, greater than about 125°C. If the hysteresis were sufficiently wide and room temperature could be located near the middle of the hysteresis, then the alloy could be fabricated and conveniently stored while in the martensitic condition. Since the hysteresis is sufficiently wide, the alloy would not transform to austenite until heated substantially above room temperature. This heating would not be applied until the alloy (in the form of a coupling, for example) was installed in its intended environment. The alloy, which would then be in the austenitic condition, would remain in the austenitic condition after cooling down since the service temperature (which may be above or below room temperature) would be substantially above the martensite transformation temperature. Thus, the above-noted desirable results could be achieved.
Unfortunately, there is believed to be no commercially viable nickel/titanium-based alloy that has a hysteresis sufficiently wide to achieve these desirable results.
For example, the commercially viable near equiatomic binary nickel-titanium alloys can have a hysteresis width of about 30°C. The location of the hysteresis for this alloy is also extremely composition sensitive so that while the hysteresis can be shifted from sub-zero temperatures to above-zero temperatures, the width of the hysteresis does not appreciably change. Thus, if the alloy were martensitic at room temperature, the service temperature must be above room temperature. Similarly, if the service temperature was at room temperature, the alloy would be martensitic below room temperature so that the alloy would require special cold-temperature equipment for fabrication, shipping, and storage. Ideally, as discussed above, room temperature should be located near the middle of the transformation hysteresis. However, since the width of the hysteresis in the binary alloy is so narrow, the range of service temperatures for any particular alloy is necessarily limited. As a practical matter, the alloy would have to be changed to accommodate any change in service temperatures.
It can be appreciated that the relative lack of commercialization of shape memory alloys must be due, at least in part, to their extreme sensitivity to temperatures as discussed above. Alloying and processing have not solved the problem.
Nickel/titanium/iron alloys, e.g., those in Harrison et al., U.S.P. 3,753,700, while having a wide hysteresis, up to about 70°C, are the typical cryogenic alloys which always undergo the martensite/ austenite transformation at sub-zero temperatures. It should be noted that in general, the colder shape-memory alloys such as the cryogenic alloys have a wider transformation hysteresis than the warmer shape memory alloys. In the case of the cryogenic alloys, the alloys must be kept very cold, usually in liquid nitrogen, to avoid the transformation from martensite to austenite. This makes the use of shape memory alloys inconvenient, if not uneconomical.
The nickel/titanium/copper alloys disclosed in U.S. Patent Application No. 537316 which served to establish a priority date for EP-A-88604 and the nickel/titanium/vanadium alloys disclosed in U.S. Patent Application No. 541844 which served to establish a priority date for EP-A-140621, are not cryogenic but their hysteresis may be extremely narrow (10―20°C) such that their utility is limited for couplings and similar articles.
The problems experienced with the nickel/titanium-based shape memory alloys have been somewhat overcome by processing in the copper-based shape memory alloys. It is now known that the hysteresis in copper-based shape memory alloys can be temporarily expanded by mechanical preconditioning, austenitic aging and heat treating. In this regard, see Brook et al., U.S.P. 4,036,669; 4,067,752; and 4,095,999.
The methods of the Brook et al. patents have been applied to nickel/titanium-based alloys; however, it has been found that these methods have no beneficial effect on nickel/titanium-based alloys.
It is known that under certain conditions the hysteresis of nickel/titanium-based alloys can be shifted as opposed to expanded. It should be understood that shifting of the hysteresis means that the M_s, M,, A₅, and A_f temperatures have all been translated to M_s', M/, A_s' and A,' such that there is substantially no change in the width of the hysteresis. It should be noted that the translated transformation temperatures may be higher or lower than the normal transformation temperatures. On the other hand, expansion of the hysteresis should generally be understood to mean that As and A_f have been elevated to A_s' and A/ while at least M_s and usually also M, remain essentially constant. Aging, heat treatment, composition, and cold work can all effectively shift the hysteresis. For example, if the stress is applied to the shape memory alloy at room temperature the hysteresis may be shifted so that the martensite phase can exist at a temperature at which there would normally be austenite. Upon removal of the stress, the alloy would isothermally (or nearly isothermally) transform from martensite to austenite.
Miyazaki et al., ("Transformation Pseudoelasticity and Deformation Behavior in a Ti-50.6 at % Ni Alloy", Scripta Metallurgica, vol. 15, no. 3, pp. 287-292, (1981)) have studied the deformation behavior of binary nickel-titanium alloys. As implied in Figure 3 of this reference, the austenite transformation temperatures can be elevated when nonrecoverable strain is imparted to the alloy. That is, when the alloy was strained to 8% or higher and the stress then removed was there, some component of the strain which remained at the deformation temperature of -30°C (compared to an A, of -52°C). This component recovered when heated to 100°C (see dotted lines on Figure 3) although the precise recovery temperature was never measured. It is not clear from this reference whether the hysteresis was shifted or expanded since the binary nickel-rich alloy tested is extremely unstable when rapidly quenched as was done in this reference. In fact, one skilled in the art would have concluded that the hysteresis was shifted and not expanded due to the unstable alloy tested. There is no illustration of the transformation hysteresis to contradict this conclusion.
In the Melton et al. patent previously mentioned, a nickel/titanium/copper alloy was deformed beyond a critical strain so as to impart nonrecoverable strain. However, no expansion of the transformation hysteresis was observed.
While it can be appreciated that it would be desirable to have a nickel/titanium-based shape memory alloy and article with a sufficiently wide transformation hysteresis, the prior art has thus far remained silent on a way to achieve it.
As mentioned earlier, shape-memory alloys have found use in pipe couplings. The pipe coupling may be a monolithic pipe coupling as described in the earlier-mentioned Harrison and Jervis patents.
Alternatively, the pipe coupling may be a composite coupling as described in the earlier-mentioned Clabburn patent and in U.S. Patent Nos. 4,379,575; 4,455,041; and 4,469,357 to Martin. As noted in Martin, the composite coupling comprises a driver member and a sleeve member.
Composite couplings present the problem of how best to assemble them. In the Martin patents, there are noted several ways to assemble the couplings. In one way, the sleeve may be assembled with the driver just after the expansion of the driver so as to take advantage of the elastic springback of the material. The driver and sleeve members are then stored in a cryogenic fluid until ready for installation.
Alternatively, the driver alone may be stored in a cryogenic fluid and then joined with the sleeve at the time of installation. Once joined with the sleeve, the driver is allowed to fully recover.
In practice, the driver may be expanded and, after springback has occurred, joined with the sleeve while both are immersed in a cryogenic fluid. Since no recovery of the driver has occurred, the sleeve is only loosely joined and would, in fact, become separated from the driver if means were not provided to prevent this separation. The means to prevent this separation is usually provided in the form of a flaring of one end of the sleeve which makes for a slight interference fit between the sleeve and the driver.
All of these methods sufferfrom the disadvantage that the driver must be stored in a cryogenic or other cold fluid prior to installation. The second method suffers from the additional disadvantage that the driver may recover prior to joining with the sleeve, thus rendering useless the composite coupling. The last method disadvantageously required the additional step of flaring the sleeve to prevent diengagement of the driver and sleeve.
In Clabburn, a keeper is utilized to apply a stress sufficient to temporarily raise the austenite transformation temperature. The shape-memory alloy remains in the martensitic state while the stress is applied. This method is known as constrained storage.
It can be appreciated that it would be desirable to have the driver and sleeve preassembled such that one could merely remove the preassembled coupling from a carton on a shelf and then proceed to install the coupling without the need to worry about cold storage of the coupling. Thus far, the prior art has remained silent on a way to achieve this desirable result.
Thus, it is an object of the invention to have a nickel/titanium-based shape memory alloy and article with a wide transformation hysteresis.
It is another object of the invention to process a nickel/titanium-based shape memory alloy and article so as to temporarily enlarge the transformation hysteresis of the alloy and article.
It is also an object of the invention to have a method of preassembling a composite coupling without the need for a cryogenic or other cold fluid.
It is another object of the invention to have a method of preassembling a composite coupling wherein the preassembled coupling may be stored without the need for a cryogenic or other cold fluid.
It is a further object of the invention to have a composite coupling preassembled by the method of the invention so that cryogenic or other cold fluid is not necessary.
In one aspect, this invention provides a method of temporarily expanding the transformation hysteresis, defined by M_s, M,, As and A_f temperatures, of a nickeltitanium based shape memory alloy, so that the As and A_f temperatures are increased temporarily to A_s' and A_f' where the difference between A_s' and M_s is greater than the difference between As and M_s, the method comprising:

(a) deforming the article that is formed from the alloy by applying to it a load that is sufficient to provide at least 1 % strain that would not be recovered if, after the deformation, the article is heated under no load to at least the A_f' temperature, the deforming load being so applied to the article that, at all times while the strain is being imparted to the article, the temperature of the article is less than:
- (i) the temperature at which the alloy begins to transform from the martensitic phase to the austenitic phase, and
- (ii) the maximum temperature at which the alloy can be transformed from the austenitic phase to the martensitic phase by the application of stress; and
(b) removing the deforming load from the article while ensuring that the temperature of the alloy does not exceed A_s'.

In another aspect, the invention provides a method of making a composite coupling which comprises a driver member formed from a nickel-titanium based shape memory alloy that has a transformation hysteresis defined by M_s, M_f, As and A_f temperatures, and a sleeve member positioned in the direction of recovery of the driver member, the method comprising:

(a) expanding the transformation hysteresis of the alloy, so that the As and A_f temperatures are increased temporarily to A_s' and A_f' where the differences between A_s' and M_s is greater than the difference between As and M_s, by deforming the driver member by applying to it a load that is sufficient to provide at least 1 % strain that would not be recovered if, after the deformation, the article is heated under no load to at least the A_f' temperature, the deforming load being so applied to the driver member that, at all times while the strain is being imparted to the driver member, the temperature of the driver member is less than:
- (i) the temperature at which the alloy begins to transform from the martensitic phase to the austenitic phase, and
- (ii) the maximum temperature at which the alloy can be transformed from the austenitic phase to the martensitic phase by the application of stress;
(b) removing the deforming load from the driver member while ensuring that the temperature of the driver member does not exceed A_s';
(c) positioning the sleeve member in the direction of recovery of the driver member; and
(d) increasing the temperature of the driver member to a temperature which is greater than true A_s' but less than A_s' so as to cause the driver member to recover to engage the sleeve member.

We have found that by taking advantage of the expansion of the hysteresis caused by overdeformation of the driver member the composite coupling may be preassembled simply and efficiently.
Embodiments of the invention will now be described, by way of example, with reference to the accompanying drawings, wherein:

Figure 1 is a schematical illustration of the shifting of the shape memory alloy transformation hysteresis.
Figure 2 is a schematical illustration of the expansion of the shape memory alloy transformation hysteresis according to the invention.
Figure 3 is a schematical stress/strain curve for a binary nickel/ titanium-based shape memory alloy.
Figure 4 schematically illustrates the binary alloy strained as in Figure 3 in the unrecovered and recovered state.
Figure 5 is a schematical transformation hysteresis curve for a nickel/titanium/vanadium alloy after recovery of a 5% deformation and illustrating the presence of the R phase.
Figure 6 is a schematical transformation hysteresis curve for a nickel/titanium/vanadium alloy after recovery of a 16% deformation and illustrating the absence of the R phase.

Referring to the figures in more detail and particularly referring to Figures 1 and 2, there is graphically illustrated the transformation hysteresis for a shape memory alloy. Figure 1 illustrates the shifting of the transformation hysteresis as would occur if, for example, a stress was applied. The hysteresis has moved upwardly in temperature from position 2 to position 4, shown in dotted lines. While the entire hysteresis has moved upwardly in temperature it can be seen that the width of the hysteresis, indicated generally by 6 has remained approximately constant. In other words, M_s, M_f, A_s, and A_f have all been translated to higher temperatures and are now denoted as M_s', M_f', A_s', and A,'. Of course, as stated earlier, there are circumstances where the transformation temperatures may be translated to lower temperatures.
In contrast to the shifting of the hysteresis as illustrated in Figure 1, Figure 2 now illustrates in general the expansion of the hysteresis. It can be seen that the martensite transformation temperatures remain constant but the austenite transition temperatures have been translated upwardly so that the width of the hysteresis indicated generally by 6 has now been expanded as indicated generally by 8. That is, M_s and M, remain constant or nearly constant while As and A_f have been translated to higher temperatures and are now denoted as A_s' and A,'.
The advantages of temporarily expanding the hysteresis versus shifting the hysteresis can be explained as follows. Referring again to Figure 1, a coupling may be expanded and held in the expanded condition so as temporarily to raise, i.e., temporarily shift, the hysteresis. As long as the stress is applied, the hysteresis will be shifted. If it is desired, for example, to use this coupling in ambient temperature, indicated by T_A, the coupling will not transform to austenite as long as temperature T_A is below A_s'. Upon the removal of the stress, the coupling will isothermally (or nearly isothermally) transform into austenite. In other words, the coupling will be at T_A when the stress is removed but the hysteresis will have shifted from position 4 back to position 2. The coupling being martensitic before the shift from position 4 to position 2 must necessarily be austenitic after the shift. This method may be used for constrained storage (see, e.g., Clabburn, U.S.P. 4,149,911) wherein a coupling is expanded and then held on a mandrel in the expanded condition until it is ready to be used, at which time it is cooled to below the M_s temperature so that it may be released from the mandrel and then installed. The problem with this method is that while the coupling is held (during shipping, for example) in the expanded position which is necessary to shift the hysteresis, the coupling may relax so that a certain, perhaps very substantial, amount of recovery motion will be permanently lost.
Referring now to Figure 2 it can be seen that by temporarily widening the hysteresis, as long as the coupling is held at a T_A less than A_s' there will be no transformation. Since no stress need be continually applied to the coupling to widen the hysteresis, relaxation is not a problem. Upon use, the coupling would simply be heated above A_s', transformation from the martensite to the austenite would occur, and the hysteresis would then shrink back down to its former position.
According to the invention there is disclosed a method of processing a nickel/titanium-based shape memory alloy having a transformation hysteresis defined by M_s, M,, A_s, and A_f temperatures. In general, the method comprises temporarily expanding the transformation hysteresis by elevating the As and A_f temperatures to A_s' and A_f', respectively, so that the temperature difference between A_s' and M_s is greater than the temperature difference between As and M_s. The means for expanding the transformation hysteresis may be removed and then the alloy is stored at a temperature less than A_s'.
Usually, according to the invention, both the M_s and M, temperatures will remain essentially constant during the expansion of the hysteresis. However, in certain alloys, as will become apparent hereafter, either or both of the M_s and M_f temperatures may permanently change. This change may result from the varying of the slope or even movement of the martensitic part of the transformation hysteresis curve due to the interaction of certain metallurgical conditions. However, the important point to emphasize here is that there will always be a net increase of the width of the transformation hysteresis according to the method of the invention.
The means for expanding the transformation hysteresis comprises overdeforming the alloy by applying a stress sufficient to cause nonrecoverable strain in the alloy. It should be understood that nonrecoverable strain means strain which is not recovered after deformation and subsequent no-load heating to at least the A_f' temperature.
It is important to understand and appreciate that the current practice in forming shape memory alloys as is well known to one skilled in the art is to avoid any nonrecoverable strain. The reason for avoiding any nonrecoverable strain is that the presence of nonrecoverable strain tends to reduce the amount of motion upon recovery. It has been found, however, that the amount of lost motion is relatively small when compared to the enhanced utility of shape memory alloys having an expanded transformation hysteresis according to the present invention.
Usually (but not necessarily) the stress will be removed after the alloy is overdeformed.
It is necessary to the invention that the overdeforming takes place at a temperature which is less than about the maximum temperature at which martensite can be stress-induced. To those skilled in the art this temperature is commonly known as M_d. It is preferred however that the overdeforming temperature be above M_s.
Once the hysteresis has been expanded at least partial recovery of the alloy article can occur when the alloy is heated to a temperature greater than about A_s'. By heating to at least A_s' the transformation of the martensite to the austenite can effectively begin. It is preferred however that the heating temperature be greater than A_f' so as to effect full recovery of the alloy.
It has been found that the nickel/titanium-based shape memory alloy may be a binary or it can be at least a ternary. If it is a ternary nickel/titanium-based shape memory alloy the ternary consists essentially of nickel, titanium and at least one other element selected from the group consisting of iron, cobalt, vanadium, aluminum, and niobium. The most preferred ternary, for reasons which will become apparent hereafter, consists essentially of nickel, titanium, and niobium.
It has also been found that those shape memory alloys having an M_s less than about 0°C are preferred since these alloys have the most utility and best performance.
The benefits of expansion of the shape memory alloy transformation hysteresis are discussed above. We have found that if the driver member is preassembled with the sleeve in conjunction with the expansion of the hysteresis of the driver member, then the preassembly is greatly facilitated.
Referring again to the figures Figure 3 schematically illustrates a stress-strain curve for a typical shape memory alloy which was overdeformed. The load was then removed. With overdeformation there is by definition a substantial amount of non-recoverable strain imparted to the alloy. Nonrecoverable strain will occur when the alloy, generally speaking, is strained past its second yield point indicated approximately by reference numerical 10. After removal of the stress, the alloy was heated.
In Figure4 curve 12 illustrates the heating afterthe removal of the stress. When the transformation was complete the alloy was cooled down as illustrated by curve 14. During the cooling down under a small load and M_s and M_f temperatures were measured. The alloy was then reheated (curve 16) to measure the recovered austenitic transition temperatures As and A_f.
There is more than one way to locate on a transformation hysteresis curve the martensitic and austenitic transformation temperatures. Referring again to figure 4 the literal starting and ending of the austenitic transformation may be indicated for example by points 18 and 20 respectively on curve 12. However, the austenitic transformation effectively begins at about point 24 (denoted as A_s') and the austenitic transformation effectively ends at about point 26 (denoted as A_f'). Thus it can be said that the bulk of the transformation occurs between A_s' and A_f'. The same is true for the other transformations as illustrated by curves 14 and 16. The effective austenitic and martensitic transformation temperatures may be conveniently determined by the intersection of tangents to the transformation hysteresis curves. For example, tangents 22 on curve 12 locate As' and A_f'. The mid-point of the transformation, for example A₅₀' on curve 12, is vertically equidistant from the literal starting and ending points, for example 18 and 20 on curve 12, of the transformation.
Whenever the austenitic and martensitic transformation temperatures are mentioned in this specification it should be understood that these temperatures refer to the austentic and martensitic transformation temperatures determined by the above noted method of intersecting tangents. Whenever the literal starting and ending points of the martensitic and austentic transformations are indicated these temperatures will be referred to as the true martensitic and austenitic transformation temperatures. Thus, the literal starting and ending points of the austenitic transformation after expansion of the hysteresis are referred to as true A_s' and true A,'.
Curves 14 and 16 represent the shape memory alloy transformation hysteresis in the recovered state while curves 12 and 14 represent the shape memory alloy transformation hysteresis in the unrecovered state. Thus it can be seen that the overdeformation of the alloy according to the patent application above has substantially and temporarily widened the hysteresis.
As mentioned above a second aspect of the invention provides a method of preassembling a composite coupling having at least one heat recoverable driver member and at least one metallic insert. The driver member is made from a nickel/titanium-based shape memory alloy having a transformation hysteresis defined by M_s, M_f, As and A_f temperatures. The method comprises overdeforming the driver member by applying a stress sufficient to cause nonrecoverable strain in the driver member so that the As and A_f temperature are temporarily raised to A_s' and A,', respectively. The method further comprises removing the stress; engaging the driver member and insert; and then warming the driver and insert to a temperature less than A_s'.
According to the invention, there must be at least one driver member; however, there may be more than one such as when ring drivers are used. Similarly, there must be at least one insert but there may be more than one such as when multi-piece inserts were utilized.
It should be understood that while the driver and insert preferably need to be warmed to a temperature which is less than A_s', they in any case need to be raised to a temperature above the true A_s'. The reason for this is that below true A_s' there will not be any recovery of the shape memory alloy. Referring again to figure 4 it can be seen that between true A_s' and A_s' there will be a small amount of recovery. After A_s' is passed the bulk of the recovery will effectively occur. From figure 4 then it is apparent that to get to any amount of recovery the material has to be heated above true A_s'. However, since the amount of recovery occurring between true A_s' and A_s' is much less than the recovery occurring between A_s' and true A_f' little shape memory recovery will actually be lost by allowing the driver member to partially recover according to the invention. This partial recovery is not so great as to crush the insert but only so great as to be able to hold the insert and driver snugly engaged.
It should be understood that the metallic insert may take many forms. For example, the insert may be tubular, tapered or slotted, all of which are disclosed in the above Martin patents. Additionally, the insert may be single or multipiece. Finally, the insert may have an irregular shape such as to be x-shaped, y-shaped or t-shaped.
The insert may also have sealing means as also disclosed in the above Martin patents. The sealing means may comprise, for example, teeth or gall-prone materials.
It should also be understood that the driver member may take many forms. It is preferred, however, that the driver member be a tubular driver or a ring driver.
In the step of overdeforming the driver member, a stress is applied sufficient to cause at least one percent of nonrecoverable strain in the driver member. Of course the nonrecoverable strain may be much more than one percent which is usually the case.
The overdeformation should take place at a temperature which is less than about the maximum temperature at which martensite can be stress-induced. The temperature is also known as the M_d temperature. The reason for this is that when the material has been deformed at a temperature greater than M_d the amount of strain recoverable upon subsequent heating is drastically and dramatically reduced. Generally, the more the deformation temperature is raised above M_d, the greater will be the reduction in recoverable strain. It is most preferred that the overdeforming temperature be between M_s and A_s.
It is desirable that the nickel/titanium-based shape memory alloy has an M_s temperature less than about 0°C. However, it is preferred that the nickel/titanium-based shape memory alloy is stable, does not contain an R phase and has an M_s temperature less than about 0°C. To those skilled in the art the R phase is known as a transitional phase between the austentite and martensite and has a structure different from both. The effect of the R phase is to depress the austenitic and martensitic transformation temperatures. Alloys that are stable (i.e. exhibit temper stability) have an M_s that does not change more than about 20°C after annealing and water quenching and subsequent aging between 300 and 500°C.
It is believed that the teaching of this invention will have most application to couplings processed by the method of the invention. However it should be understood that the teaching of the invention applies to other articles and devices processed by the method of the invention.
Embodiments of the invention are now described, by way of example, with reference to the following examples.

Example 1

Commercially pure titanium and carbonyl nickel were weighed in proportions so as to give a composition of 50.7 atomic percent nickel and 49.3 atomic percent titanium. The total mass for test ingots was about 330 grams. These metals were placed in a water-cooled, copper hearth in the chamber of an electron beam melting furnace. The chamber was evacuated to 1.3 x 10-³ Pa (10-⁵ Torr) and the charges were melted and alloyed by use of the electron beam.
The resulting ingots were hot swaged and hot rolled in air at approximately 850°C to produce a strip of approximately 6.35 x 10-⁴ m (0.025-in). thickness. Samples were cut from the strip, descaled and vacuum annealed at 850°C for 30 minutes and furnace cooled.
The strip was then elongated. After elongation the stress was removed and the strip was heated unrestrained so as to effect recovery of the shape memory alloy. The recovery was monitored and plotted as a function of temperature. When the transformation was complete, the sample was cooled and then reheated so as to complete the measurement of the martensite and austenite transformation temperatures before recovery and after recovery. The results are tabulated below in Table 1.
The measure A_s' minus M_s is very useful since M_s is directly indicative of the lower functional limit of the alloy and the A_s' is directly indicative of the highest temperature which may be encountered (e.g. during storing and shipping) before the austenite transformation will effectively begin. Thus, A_s' minus M_s defines the operating range of the alloy when processed according to the invention. This measure should be compared to As minus M_s which defines the operating range of the alloy after the temporary expansion of the hysteresis has been recovered. As minus M_s is also indicative of the operating range of the alloy if it were never processed according to the invention. Thus, comparing A_s' minus M_s to As minus M_s provides useful indicia of the expansion of the hysteresis as well as the advantages of the invention.
Referring now to Table 1, A_s' minus M_s and As minus M_s are about the same at 5% elongation; however, at 16% elongation, the difference becomes substantial. It is useful to note that A_s' after 16% elongation is above normal room temperature so that the alloy may now be handled at room temperature so that the alloy may now be handled at room temperature without the necessity of providing a cold environment.
Another useful measurement for indicating the expansion of the hysteresis are the M_so, A_so. and A₅₀' values. These are the martensite and austenite transformation temperatures at which the transformation is 50% complete. Thus, referring to Table 1 below it can be seen that the the sample was cooled and then reheated so as to complete the measurement of the martensite and austenite transformation temperatures before recovery and after recovery.
In referring to Table 1 it can be seen that the difference between M₅₀ and A_so, the permanent width of the hysteresis, is about 60°C. However, the width of the hysteresis may be temporarily enlarged, i.e., A_so' minus M_so, from 64°C at 5% elongation (at which there is no nonrecoverable strain) to 91°C at 16% elongation (at which there is substantial nonrecoverable strain). The M_so, A_so, and A_so' values are also useful because they are the most easily determined as will become apparent hereafter.

Example 2

Commercially pure titanium, carbonyl nickel and iron were weighed in proportions so as to give a composition of 47 atomic percent nickel, 50 atomic percent titanium and 3 atomic percent iron. The total mass for test ingots was about 330 grams. These alloys were melted in an electron beam furnace in the same manner as the nickel-titanium binary. The resulting ingots were hot swaged at approximately 850°C. Round, tensile bars (6.35 ×10^-3 m 1/4") in diameter were then machined from the hot swaged ingot, vacuum annealed at 850°C for 30 minutes, and then furnace cooled.
The tensile bars were then elongated. After elongation the stress was removed and the bars were heated so as to effect recovery of the ternary shape memory alloy in the same manner as the binary alloy. Due to the extreme low temperatures involved, some of the values had to be extrapolated as noted. The results are tabulated below in Table 2.
The discrepancy in the martensite and austenite transformation temperatures (between 5 and 16% elongation) can be explained in part by the interference of the R phase, to be discussed in more detail later.
As it can be appreciated, the width of the hysteresis and the operating range have been enlarged as a result of the 16% elongation of the alloy. The import of this is that after elongation of the alloy, the alloy no longer has to be stored in liquid nitrogen to prevent it from transforming into austenite. Since A_s' has been raised to -88°C other forms of cold storage may now be used to store and ship the nickel/titanium/iron alloy prior to its final use. It is believed that this will result in greater utility of the alloy.

Example 3

Commercially pure titanium, carbonyl nickel and niobium were weighed in proportions so as to give a composition of 47 atomic percent nickel, 44 atomic percent titanium, and 9 atomic percent niobium. The total mass for test ingots was about 330 grams. The composition was melted in an electron beam furnace as was the case with the alloys in Examples 1 and 2. The resulting ingots were hot swaged in air at approximately 850°C. The resulting bar was machined into rings which were vacuum annealed in 850°C for 30 minutes and then furnace cooled. The rings were then enlarged, unstressed and subsequently heated so as to measure the free recovery of the alloy._The results are tabulated below in Table 3.
It can be seen from Table 3 that the hysteresis width (A₅₀―M₅₀) in the fully recovered state is about 55°C with As being -56°C. With the austenite temperature in this range it is still necessary forthe alloy to be cold stored in order to prevent transformation of the martensite into the austenite. However, if the ring is now enlarged about 5%, the A_s temperature has been temporarily raised to -14°C which would still require cold storage. By enlarging the ring 12.1 % at which point there is now substantial nonrecoverable strain, the As has been temporarily increased to 27°C. Thus, at this temperature the alloy may be stored and shipped at room temperature. No cold storage provisions are required. It also can be seen that the width of the hysteresis has now been increased to 124°C from 55°C and the operating range (A_s'―M_s) has been increased to 117°C. By enlarging the ring 16.2% As has now been temporarily raised to 41°C with the width of the hysteresis now being 140° and the operating range now being 131°C.
It is believed that to have the most commercially practical alloy it is necessary to have an hysteresis width of greater than about 125°C with ambient or room temperature somewhere in the middle of that hysteresis so as to allow a substantial leeway on either side of room temperature for temperature excursions. Strictly speaking, it would be most preferred if the A_s' could be raised to about 50°C.
The first three samples enlarged at 5.2, 12.1, and 16.2% were enlarged in liquid nitrogen which is substantially below M_s. If the samples were now enlarged in -90°C alcohol, which is at the M_s temperature, it can be seen that the austenite transition temperatures have been raised to higher values than when enlarged in liquid nitrogen. By comparison, the A_s' temperatures have been raised from 41 to 50°C. While this increase is not of great magnitude it is nevertheless important.
It is most preferred that the temperature of deformation be above M_s. The importance of this limitation is illustrated in the next sample which was deformed at -70°C (compared to an M_s of -90°C). It can be seen that A_s', and A₅₀'―M₅₀ and A_s'―M_s have all been increased more than any of the previous samples.
The next sample was enlarged at 0°C. While it can be seen that the hysteresis has been expanded, the effect of the expansion of the hysteresis has not been as great as when it was enlarged in -90°C alcohol or -70°C alcohol since A_s' has only been raised to 34°C.
The previously stated results have been obtained by expanding the hysteresis through overdeforming of the alloy so as to impart nonrecoverable strain, removing the stress and then storing the alloy at a temperature less than A_s'.
The process may be varied somewhat so as to give equally dramatic results. Thus a sample may be overdeformed at low temperatures such as -90°C in alcohol to stabilize the martensite at or near room temperature. When the stress is removed there will be an elastic springback of about 4%. Now if the alloy is redeformed at 20°C to the same amount of overdeformation, 16.2%, and the stress removed, it can be seen in the last column of Table 3 that the austenite transition temperatures have been raised to even higher values when compared to a single expansion in -90°C alcohol. Thus, A_s' has been moved from 50°C to 55°C. Again, while this increase in A_s' may appear to be a small amount of temperature increase it is nevertheless of great importance. One easy way to accomplish this process is to deform the ring on a mandrel and then let the ring and mandrel warm to room temperature.
The nickel/titanium/niobium ternary alloys are preferred alloys due to their ready susceptibility to expansion of the transformation hysteresis as illustrated above. Of all the ternary niobium alloys, those that are stable, have an M_s greater than 0°C and do not have an R phase are the most preferred. The R phase, as further discussed below, is a transitional phase between austenite and martensite. Since the R phase is not present, there is substantial uniformity in the martensite and austenite transformation temperatures from sample to sample. Alloys that are stable (i.e., exhibit temper stability) have an M_s that does not change more than about 20°C after annealing and water quenching and subsequent aging between 300 and 500°C.

Examples 4, 5, and 6

Commercially pure titanium, carbonyl nickel and amounts of vanadium, cobalt, and aluminum were weighed in proportions so as to give compositions of: 46 atomic percent nickel, 49 atomic percent titanium, and 5 atomic percent vanadium; 49 atomic percent nickel, 49 atomic percent titanium, and 2 atomic percent cobalt; and 50 atomic percent nickel, 48.5 atomic percent titanium, and 1.5 atomic percent aluminum. Each of the compositions was melted and 6.35 x 10-⁴ m (0.025-in.) thick strips prepared in the same way as that previously stated with respect to the binary.
After elongation, the stress was removed and the strip was heated unrestrained so as to effect recovery which was monitored and plotted as a function of temperature. When the transformation was complete, the sample was cooled and then reheated so as to complete the measurement of the martensite and austenite transformation temperatures before recovery and after recovery. In the case of the cobalt alloy, the martensite and austenite transformation temperatures were measured with a load of 1.38 x 107 Pa (20 ksi) and then extrapolated to 0 Pa. The results are tabulated below in Tables 4, 5, and 6.
Referring to Table 4, the large discrepancy between the martensite and austenite transformation temperatures at 5 and 16%, respectively, is believed due to the interference of the R-phase. Referring to Figure 5, the presence of the R phase 28 is most noticeable on the austenite leg of the transformation hysteresis for the alloy deformed 5%. As stated previously the R phase is a transitional phase between the austenite and martensite and has a structure different than either. The effect of the R phase is to depress the austenite and martensite transformation temperatures. Figure 6 illustrates the transformation hysteresis curve for the same alloy, but after recovering from 16% deformation. The R phase is noticeably absent. The austenite and martensite transformation temperatures in Figure 6 are also noticeably higher.
Referring again to Table 4, it can be seen that a 5% deformation has little effect on the expansion of the hysteresis. Thus, A_s' minus M_s and As minus M_s are substantially the same. This is not the case after 16% deformation wherein the transformation hysteresis has been noticeably enlarged.
The results in Table 5 are similar to those in Table 4 in that a 5% deformation (no nonrecoverable strain) had little effect on the expansion of the transformation hysteresis whereas a 16% deformation (substantial nonrecoverable strain) had a marked effect on the expansion of the transformation hysteresis.
The change in the recovered martensite and austenite transformation temperatures between the 5% and 16% deformations is again believed due to the interference of the R phase in the sample deformed 5%.
Referring now to Example 6 and Table 6, the sample deformed 16%, and thus having substantial nonrecoverable strain, shows. a marked expansion of the transformation hysteresis (as in the previous two examples) whereas the sample deformed at 5% shows essentially no expansion of the transformation hysteresis.
Again, the interference of the R phase has manifested itself by depressing the martensite and austenite transformation temperatures in the sample deformed 5%.
It can be appreciated that while this invention is most advantageous with respect to those alloys having an enlarged hysteresis with its middle near room temperature, it is within the scope of the invention to apply the teachings of this invention to other alloys as well, as illustrated in the above examples.
It can also be appreciated that the expansion of the transformation hysteresis will be more dramatic in some alloys than in others. This conclusion becomes apparent when comparing the transformation hysteresis expansion of the binary alloy with the transformation hysteresis expansion of the most preferred niobium ternary alloy.

Example 7

A cylindrical driver member was made from an alloy having the composition of 47 atomic percent nickel, 44 atomic percent titanium and 9 atomic percent niobium. The nickel/titanium/niobium alloys, in general, are the most preferred alloys. These alloys are disclosed in U.S. Patent Application No. 668777 which served to establish a priority date for EP-A-185452.
The driver was melted and processed as noted in our patent application above except that a coupling was machined instead of a ring. The driver was machined to have an inside diameter of 2.15 x 10-² m (.847 inches), an outside diameter of 3.34 x 10-² m (1.313 inches) and a length of 5.39 x 10-² m (2.12 inches).
A cylindrical insert was then made to be eventually joined with the driver so as to form a composite coupling. The insert was machined from 316 stainless steel so as to have an inside diameter of 2.16 x 10-² m (.850 inches), an outside diameter of 2.46 x 10-² m (.970 inches) and a length of 5.39 x 10-² m (2.12 inches). It is not necessary to the invention that the insert be made from stainless steel. It is only necessary that the insert be made from a material that is sufficiently soft such that it may be crushed by the driver upon full recovery thereof.
With the particular alloy utilized, the M_s temperature was -90°C, the As temperature was -56°C and the M_d temperature was -10°C. Although not actually measured, such an alloy expanded about 16% at -50°C would be expected to have a true A_s' of -52°C and an A_s' of +52°C. Thus, immediately after expansion, the driver was near the literal starting temperature of the austenitic transformation of the temporarily expanded transformation hysteresis.
After expansion, the driver was removed from the cold fluid and placed on a work bench. The insert was then slipped into the driver. Thereafter, the driver and insert were allowed to warm to room temperature, which it is noted is substantially belowA,'. It was found that the driver and insert were snugly engaged and could only be moved relative to each other with great difficulty. It should be noted that while the driver and insert became snugly engaged, there was no crushing of the insert.
The driver, prepared as described above, would be expected to have about 8% recoverable strain. About 1% of that recoverable strain was utilized in the preassembling of the driver and insert. Thus, about 7% recoverable strain remains for the actual coupling of the substrates.
The composite coupling is now preassembled and ready for storage or use.

Example 8

Commercially pure titanium, carbonyl nickel and niobium were weighed in proportions so as to give a composition of 47 atomic percent nickel, 44 atomic percent titanium, and 9 atomic percent niobium. The total mass for test ingots was about 330 grams. These metals were placed in water-cooled, copper hearth in the chamber of an electron beam melting furnace. The chamber was evacuated to 1.33 x 10-³ Pa (10-⁵ Torr) and the charges were melted and alloyed by use of the electron beam. The resulting ingots were hot swaged in air at approximately 850°C. The resulting bar was machined into rings which were vacuum annealed in 850°C for 30 minutes and then furnace cooled. The rings were then enlarged, unstressed and subsequently heated so as to measure the free recovery of the alloy. The results are tabulated below in Table 7.
While the data relate to the expansion of rings, the date is nevertheless indicative of how the material would perform as a driver. In each case, there is a substantial difference between true A_s' and A_s' indicating that the material will achieve the objects of the invention. The true A_s' for the sample expanded at -70°C is believed to be an anomaly in that the sample may have inadvertently warmed to near room temperature prior to the actual measurement of true A_s' and A_s'.
It is most preferred that expansion takes place between As and M_s. This is because at temperatures higher than As or lower than M_s, elastic springback of the material may be increased. Additionally, the material has somewhat more ductility when expanded between As and M_s.

Examples 9 to 13

Commercially pure titanium and carbonyl nickel were weighed in proportions so as to give a composition of 50.7 atomic percent nickel and 49.3 atomic percent titanium. Additionally, commercially pure titanium, carbonyl nickel and amounts of vanadium, cobalt, aluminum and iron were weighed in proportions so as to give compositions of: 46 atomic percent nickel, 49 atomic percent titanium and 5 atomic percent vanadium; 49 atomic percent nickel, 49 atomic percent titanium and 2 atomic percent cobalt; 50 atomic percent nickel, 48.5 atomic percent titanium and 1.5 atomic percent aluminum; and 47 atomic percent nickel, 50 atomic percent titanium and 3 percent iron.
These metals were placed in a water-cooled, copper hearth in the chamber of an electron beam melting furnace. The chamber was evacuated to 1.33 x 10-³ Pa (10-⁵ Torr) and the charges were melted and alloyed by use of the electron beam.
The resulting iron-containing ingots were hot swaged at approximately 850°C. Round, tensile bars (6.35 x 10-³ m ¼ ") in diameter were then machined from the hot swaged ingot, vacuum annealed at 850°C for 30 minutes, and then furnace cooled. The tensile bars were then elongated. After elongation, the stress was removed and the bars were heated unrestrained so as to effect recovery of the shape memory alloy. The recovery was monitored and plotted as a function of temperature. When the transformation was complete, the sample was cooled and then reheated so as to complete the measurement of the martensitic and austenitic transformation temperatures before recovery and after recovery. The results are tabulated in Table 8.
The remaining ingots were hot swaged and hot rolled in air at approximately 850°C to produce a strip of approximately 6.35 x 10-⁴ m (0.025-in.) thickness. Samples were cut from the strip, descaled and vacuum annealed at 850°C for 30 minutes and furnace cooled. The stip was then elongated. After elongation, the stress was removed and the strip was heated unrestrained so as to effect recovery which was monitored and plotted as a function of temperature. When the transformation was complete, the sample was cooled and then reheated so as to complete the measurement of the martensitic and austenitic transformation temperatures before recovery and after recovery. In the case of the cobalt alloy, the martensitic and austenitic transformation temperatures were measured with a load of 1.38 x 10' Pa (20 ksi) and then extrapolated to 0 Pa. The results are tabulated below in Tables 9 to 12.
As stated earlier it is believed that the above data while not derived from drivers per se is nevertheless indicative of how each of these materials will perform as a driver. Thus, for each of these materials, in addition to having an expanded hysteresis, there is a substantial difference between true A_s' and A_s' so that these materials are suitable to achieve the objects of the invention.
Finally, it can be apprecated that while the samples in the above examples were deformed by application of a tensile stress, the objects of the invention can be fully achieved by application of a compressive stress.

Claims

1. A method of temporarily expanding the transformation hysteresis, defined by M_s, M,, A_s and A, temperatures, of a nickel-titanium based shape memory alloy, so that the As and A_f temperatures are increased temporarily to A_s' and A_f' where the difference between A_s' and M_s is greater than the difference between As and M_s, the method comprising:

(a) deforming the article that is formed from the alloy by applying to it a load that is sufficient to provide at least 1% strain that would not be recovered if, after the deformation, the article is heated under no load to at least the A_f' temperature, the deforming load being so applied to the article that, at all times while the strain is being imparted to the article, the temperature of the article is less than:

(i) the temperature at which the alloy begins to transform from the martensitic phase to the austenitic phase, and

(ii) the maximum temperature at which the alloy can be transformed from the austenitic phase to the martensitic phase by the application of stress; and

(b) removing the deforming load from the article while ensuring that the temperature of the alloy does not exceed A_s'.

2. A method as claimed in claim 1, in which the deforming load is so applied to the article that, at all times between first applying the load to the article and the at least 1% non-recoverable strain having been imparted to the article, the rate of increase of stress with increasing strain does not have a value less than zero.

3. A method as claimed in claim 1 or claim 2, in which the temperature at which the article is deformed is greater than M_s.

4. A method as claimed in any one of claims 1 to 3, in which the deformation step takes place at a first temperature, the method including the step of repeating the deformation of the article at a second temperature which is different from the first temperature, preferably greater than the first temperature.

5. A method as claimed in any one of claims 1 to 4, in which the deformation step takes place at a first temperature, and which includes the step of increasing the temperature of the alloy to a second temperature, which is preferably about room temperature, while maintaining the applied strain.

6. A method as claimed in any one of claims 1 to 5, in which about 16% strain is imparted to the article.

7. A method of making a composite coupling which comprises a driver member formed from a nickel- titanium based shape memory alloy that has a transformation hysteresis defined by M_s, M,, As and A, temperatures, and a sleeve member positioned in the direction of recovery of the driver member, the method comprising:

(a) expanding the transformation hysteresis of the alloy, so that the As and A_f temperatures are increased temporarily to A_s' and A_f' where the differences between A_s' and M_s is greater than the difference between As and M_s, by deforming the driver member by applying to it a load that is sufficient to provide at least 1 % strain that would not be recovered if, after the deformation, the article is heated under no load to at least the A_f' temperature, the deforming load being so applied to the driver member that, at all times while the strain is being imparted to the driver member, the temperature of the driver member is less than:

(ii) the maximum temperature at which the alloy can be transformed from the austenitic phase to the martensitic phase by the application of stress;

(b) removing the deforming load from the driver member while ensuring that the temperature of the driver member does not exceed A_s';

(c) positioning the sleeve member in the direction of recovery of the driver member; and

(d) increasing the temperature of the driver member to a temperature which is greater than true A_s' but less than A_s' so as to cause the driver member to recover to engage the sleeve member.

8. A method as claimed in claim 7, in which the temperature at which the driver member is deformed is greater than M_s.

9. A method as claimed in claim 7 or claim 8, in which the deformation of the driver member takes place below room temperature and results in A_s' being elevated to a temperature above room temperature, and in which in step (d) of the method, the temperature of the driver member is increased to about room temperature.

10. A method as claimed in any one of claims 1 to 9, in which the M_s temperature of the alloy is less than about 0°C.

11. A method as claimed in any one of claims 1 to 10, in which the alloy consists essentially of nickel, titanium and niobium.