3.1. Macro appearance and comparison to traditional crimps
VFAW produces a NiTi wire/Brass sheet weld with little plastic deformation on the brass side except the wire/sheet welding area (Fig. 3b). After being machined to a similar ring shape (Fig. 3c), the NiTi/Brass weld can be used to simulate or replace a traditional ring type mechanical crimp (Fig. 3a). Current work is underway to weld directly to a crimp. The traditional ring type mechanical crimp presents a maximum pull-out force as low as 140 N [1, 18]. This is much lower than the peak load of the NiTi/Brass impact weld (640 N) which was shown in Fig. 7. This proves that VFAW can be used to produce NiTi actuators with higher strength connection between the NiTi wire and a structural part compared to mechanical crimping method.
3.2. Interfacial microstructures
To reveal the interfacial bonding quality and mechanism, detailed microstructure characterizations were conducted on a typical NiTi/Brass weld cross-section (Fig. 4). Wavy structures with little macro-defects were observed in the NiTi/Brass weld interface (Fig. 4a), which is characteristic of VFAW [19]. The hook-shape gaps on both sides of the NiTi/Brass weld are caused by the high strain rate plastic deformation during the VFAW process. This high strain rate plastic deformation drives the brass alloy to flow upward along the circumferential direction of the NiTi wire. A magnified view of the wavy interface exhibited a solid-state defect-free metallurgical bonding between the NiTi and brass alloy (Fig. 4b). This was verified by the EDS mapping analyses shown in Fig. 4c, d, e, and f which exhibited sharp transitions between the major elements in NiTi (Ni and Ti) and those in brass alloy (Cu and Zn).
3.3. Phase transformation characteristics
Phase transformation behaviors of NiTi SMA subjected to impact welding are good indicators of the functional properties of NiTi dissimilar welds. Fig. 5 compares the DSC results of the NiTi BM and a typical NiTi/Brass weld. Transformation temperatures, thermal hysteresis and enthalpies of both types of samples are summarized in Table 1. Phase transformation temperatures such as Ms, Mf, Rs, Rf, As and Af were determined through the intersection of the baseline with the line of maximum inclination of the transformation peaks based on the ASTM F2004-17 standard. Thermal hysteresis () was determined by the difference between the peak temperatures. Enthalpies were calculated by integrating the peak areas over the baseline (time).
Thermal hysteresis and enthalpies can be used to examine the extent of phase transformation. It was reported that a complete reversible B2 to B19’ transformation generally exhibits a thermal hysteresis ranging from 35 to 50 °C and enthalpies ranging from 10 to 25 J g−1, respectively [20, 21]. In the current work, the calculated thermal hysteresis and enthalpies for NiTi BM are lower than 10 J g−1 and 10 °C, respectively. This shows that only intermediate B2-R reversible transformation occurred in the NiTi base metal. This intermediate phase transformation is possibly due to the presence of precipitates and nanocrystalline grains in the NiTi BM caused by the thermo-mechanical processing. The transformation temperatures, thermal hysteresis and enthalpies for the NiTi/Brass welds cannot be determined due to the lack of apparent heat flow peaks. It was also shown that the phase transformation temperature ranges of the NiTi/Brass welds are widened compared to that in the NiTi BM.
The martensitic transformation was shown to be significantly suppressed in both NiTi BM and the NiTi/Brass weld, while the latter is more severe. For the NiTi BM, the nanocrystalline grains and precipitates restrict the structure and make the martensitic transformation difficult. The large heterogeneity in internal stresses increases the barrier to complete phase transformation. For the NiTi/Brass weld, apart from the microstructure constriction brought by the NiTi BM, the high strain rate plastic deformation in the VFAW process will further increase the barrier to transformation [22, 23]. The thermal mass of the inactive brass alloy likely impedes the phase transformation to a large extent. There may have been limited phase transformation in the NiTi but it was below the detection limits of the DSC system [24]. These all explain why there are no heat flow peaks in the NiTi/Brass weld.
The readings could also have been affected by the thermal mass of brass in the DSC test. Suppression of the phase transformation was limited to the immediate joint area, and is posited as a potential benefit. A gradient in transformation properties may serve to reduce plastic strain buildup as opposed to a step-wise change at the interface between NiTi and brass.
Table 1. Transformation temperatures (°C), thermal hysteresis (), and enthalpies () of the NiTi base metal and NiTi/Brass weld where and indicate the austenite start and finish temperatures, respectively; and indicate the R phase start and finish temperatures, respectively; and indicate the martensite start and finish temperatures, respectively.
|
Transformation temperatures (°C)
|
|
(J g−1)
|
Samples
|
|
|
|
|
|
|
(°C)
|
Cooling
|
Heating
|
NiTi BM
|
-1.95
|
24.6
|
23
|
-11.7
|
-
|
-
|
6.44
|
5.61
|
4.62
|
NiTi/Brass weld
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
3.4. Mechanical properties
3.4.1. Microhardness distributions
To reveal the mechanical behavior of the NiTi/Brass weld, microhardness distributions across the welding interface were examined. Three different progressions (top, middle and bottom) from the brass alloy to the NiTi side were carried out perpendicular to the welding interface (Fig. 6a). It is worthwhile to mention that the hardness measurements on the NiTi side might not be as quantitatively accurate as those on the brass alloy side due to the partial recovery caused by stress induced martensitic transformation.
The results show that the area near the NiTi/Brass interface gained 15.6% increase of hardness compared to the brass alloy base metal while the area near the NiTi side exhibited comparable hardness distributions as the NiTi BM. These measurement results confirm that little or no heat affected zones were formed in the NiTi/Brass impact welds. This has been observed in VFAW of other metal combinations such as NiTi/Stainless steel (SS) [24, 25] and Ti/SS [12]. The lack of HAZ formation is an advantage over other traditional fusion-based and solid-state welding technologies normally involving thermal softening. The improved strength of the welding interface is likely caused by the grain refinement in the adjoining areas induced by the high strain rate plastic deformation in the VFAW process [26].
3.4.2. Lap shear test results and joint efficiency
Lap shear tests were performed on the NiTi BM and NiTi/Brass welds to compare the mechanical strength and ductility of samples before and after welding. Fig. 7 shows the load-displacement curves of the NiTi BM and a typical NiTi/Brass weld. Note that the nominal stress is calculated through dividing the load by the cross-section area of the NiTi BM wire, and the nominal strain is the ratio of crosshead displacement to the gauge length. It was observed that the stress-strain curve of the NiTi/Brass weld exhibited a relatively larger elastic slope compared to that in the NiTi BM. This difference in elastic slope is due to the lap joint morphology of the NiTi/Brass weld causing rotation of the sample. The stress plateau of the NiTi/Brass weld is narrower and more inclined in comparison with that of the NiTi BM due to half of the gauge length being brass alloy. Note that the stress plateau is caused by the transformation of austenite to a variant of martensite predetermined by the applied stress direction [27]. During the plateau stage a certain amount of strain can be accommodated while exceeding a threshold value, plastic deformation will occur. The plastic deformation of brass alloy would cause more rapid shifting of stress and strain towards the elastic-plastic stage. The peak stress of the NiTi/Brass weld is around 97% of the ultimate tensile strength (UTS) of the NiTi BM.
In this work, the ratio of peak stress of the weld to the UTS of the NiTi BM is used as a reference to compare the joint efficiencies among varied joint technologies. Table 2 compares the joint efficiencies and main issues/microstructures for failure of the NiTi/Cu dissimilar welds made by laser welding (LSW) and ultrasonic spot welding (USW), as well as NiTi/Brass welds made by VFAW in the current work. Detailed welding design data including material type, weld format, and base metal dimensions were also provided for good comparisons.
It was observed that NiTi/Cu welds made by LSW exhibited joint efficiency lower than 60 % due to grain growth, microstructure softening as well as the formation of Cu-based or NixTiyCuz IMC [2, 3, 28, 29]. NiTi/Cu/NiTi welds made by USW entail higher joint efficiencies ranging from 60 to 86% [10, 30]. The relatively higher joint efficiencies compared to those made by LSW are due to the solid-state welding nature of USW which relies on friction and plastic deformation. The friction and plastic deformation break the oxide layers and form mechanical interlocking and direct metallurgical bonding between NiTi and Cu. It should be noted that this process is limited to use with very thin foils (e.g. <150μm). The NiTi/Brass welds made by VFAW in the current work exhibited the highest joint efficiency. One of the reasons is that impact welding is a solid-state welding method which minimizes or avoids large scale melting of the base metals. No thermally induced defects including grain growth or brittle NixTiyCuz IMC formed in these welds. The second reason is that during the VFAW process, the high velocity jetting scours away the possible oxides and achieves a nascent surface for metallurgical bonding [24]; compared to USA which will retain oxides in the interface of welds [10]. The lack of HAZ formation, the strengthened interface due to grain refinement (Fig. 6), and no oxide segregation on the interface all contribute to the high strength NiTi/Brass impact welds.
Table 2. Comparison of joint efficiency (weld strength/UTS of the NiTi base metal) and main issues/microstures for failure of NiTi/Cu dissimilar welds made by laser welding and ultrasonic spot welding, as well as NiTi/Brass welds made by VFAW in this work.
Method
|
Joint efficiency
|
Material type
|
Weld format
|
Base metal dimensions
|
Microstructures for failure
|
Failure locations
|
References
|
LSW-1
|
17.1%
|
NiTi/Cu wires
|
Butt
|
D=400 µm
|
Grain growth and softening
|
Cu BM
|
[28]
|
LSW-2
|
53.6%
|
NiTi wire/Cu sheet
|
Lap
|
DNiTi= 700 µm;
2.0 × 25.0 × 0.5 mm
|
Cu-based IMC
|
FZ
|
[2]
|
LSW-3
|
38.3%
|
NiTi wire/Cu sheet
|
Lap
|
DNiTi= 700 µm;
2.0 × 25.0 × 0.5 mm
|
NiTiCu IMC; equiaxed and lamellar regions; Islands of copper
|
Cross-section of the joint
|
[3]
|
LSW-4
|
18.7%
|
NiTi/Cu wires
|
Butt
|
D=400 µm
|
Columnar dendritic
Microstructures; NixTiyCuz IMC
|
Cu HAZ
|
[29]
|
USW-1
|
64.2%
|
NiTi sheet/Cu foil/NiTi sheet
|
Lap
|
tNiTi= 150 µm; tCu=20 µm
|
Mechanical interlocking and metallurgical adhesion
|
Weld interface
|
[30]
|
USW-2
|
85.6%
|
NiTi sheet/Cu foil/NiTi sheet
|
Lap
|
NiTi (60 × 15 × 0.2 mm);
tCu=100 µm
|
Nano-scale transition layer composed of NiTiCu phase
|
NiTi BM
|
[10]
|
VFAW
|
97%
|
NiTi wire/Brass sheet
|
Lap
|
DNiTi=0.762 mm;
40 × 60 × 1 mm
|
Mechanical interlocking and metallurgical adhesion
|
NiTi BM
|
Current work
|
|
|
|
|
|
|
|
|
|
3.4.3. Cycling tests
The typical applications of NiTi alloys such as actuators generally involves many loading and unloading cycles during the motion control process. Apart from the high joint strengths presented by the lap shear tests, the functional fatigue property of the samples is also important. 500 stress-strain cycles were conducted on both the NiTi BM and NiTi/Brass welds to determine the superelastic behavior and functional stability (Fig. 8). The 1st, 100th, 200th, 300th, 400th, and 500th cycles were singled out to better reflect the stabilization of pseudoelastic plateaus and the accumulation of irrecoverable strains during the cycling tests. Future work is focused on characterizing the response of the welds to thermal martensitic cycling.
The superelastic plateaus on the first cycle for the NiTi BM and NiTi/Brass weld occurred at 240 N and 226 N, respectively. The forward superlelastic plateaus on the last cycle of these two samples occurred at 170 N and 150 N, respectively. This means that after 500 cycles, the superelastic plateaus were lowered down by 70 N for all the samples. From 100th to 500th cycles, the superelastic plateaus only decrease by 15 N which means both samples start to stabilize after 100 cycles. The difference in the plateau stress for the NiTi BM and the NiTi/Brass weld is possibly due to the half gauge length of NiTi/Brass weld being brass alloy which does not exhibit superelastic behavior.
A small stress kink is located ahead of the stress plateau of the NiTi base metal, and such a stress kink usually indicates that a higher stress is required to overcome the barrier at the onset of the stress-induced martensitic transformation (SIMT) [31]. Such stress kink was not observed in the NiTi/Brass weld. This indicates that a lower stress is needed to induce martensitic transformation in the NiTi/Brass welds compared to NiTi BM which corresponds to the results shown in Fig. 8.
As the number of cycles continues, the accumulated irrecoverable strain increases up to a point where the stabilization plateau is reached due to the pile-up of dislocations and other lattice defects induced during the stress induced transformation [32]. NiTi/Brass weld exhibited a high irrecoverable strain (0.42 %) than that of the NiTi BM (0.37 %). The pseudoelastic curve of NiTi BM tends to converge faster than that of the welded joint to the stabilized response, accumulating less irrecoverable strain on each successive cycle. There are three sources for the residual strain: (a) the stabilized R phase; (b) the stabilized martensite in SIMT and (c) plastic deformation [33,34]. These irrecoverable strains are significantly lower than those observed in fusion-based welding methods [33–36]. The melting and solidification that occurred in fusion-based welding methods can form intermetallics and large grains in the fusion zone and HAZ, which contribute to the degradation of functional properties.
It was reported that a lower stress for inducing martensite transformation indicates a better fatigue resistance [37]. A lower barrier to induce the martensite could contribute to a higher fatigue resistance by diluting the stress through local SIMT. Following this logic it is predicted that the fatigue resistance of the NiTi/Brass welds will not be degraded by the VFAW process. Future work is underway to investigate the fatigue resistance of these joints in both mechanical and thermomechanical fatigue.