Process Capability of Wire-EDM of NiTi Shape Memory Alloy at Main

Procedia Manufacturing
Volume XXX, 2015, Pages 1–11
43rd Proceedings of the North American Manufacturing Research
Institution of SME http://www.sme.org/namrc
Process Capability of Wire-EDM of NiTi Shape
Memory Alloy at Main Cut and Trim Cut Modes
J.F. Liu, Y.B. Guo∗
Dept. of Mechanical Engineering, The University of Alabama, Tuscaloosa, AL 35487, USA
Tel.: +1 205 348 2615; E-mail address: [email protected]
Abstract
Nitinol is widely used shape memory alloy (SMA) in manufacturing medical devices, actuators,
mechanical couplings, etc. However, mechanical cutting of Nitinol is exceedingly difficult to machine.
Machining induced surface integrity of SMA has significant impacts on device performance. In this
study, Nitinol was machined by wire electric discharge machining (Wire-EDM) in both CH-oil and
deionized water (DI-water). Surface characteristic evolution was examined from main cut (MC), first
trim cut (TC), to finish trim cut (FC) and compared with the traditional mechanical cutting.
Keywords: Nitinol, EDM, Surface integrity, Normal distribution
1 Introduction
Nitinol is a nearly equiatomic nickel-titanium shape memory alloy which has wide applications in
medical and aerospace industries due to its superelasticity and shape memory properties (Henderson &
Buis, 2011; Bansiddhi et al., 2008; Duerig et al., 1999; Calkins et al., 2008). Good biocompatibility,
high corrosion resistance, and superior fatigue performance are important for Nitinol components.
Toxic substances free and long-term functionality are necessary for medical implants. Therefore,
surface integrity of the machined component is critical for Nitinol device performance. However, the
significant challenge in machining Nitinol results from the high ductility, strong strain-hardening, and
complex phase transformation. High tool wear and large burrs are typical pressing issues for
mechanical cutting such as milling (Weinert et al., 2004; Guo et al., 2013).
Compared to mechanical cutting, wire electric discharge machining (Wire-EDM) is an alternative
potential process to machine Nitinol regardless of its hardness and strength. Fig. 1(a) shows the
schematic of main cut and trim cut modes in Wire-EDM process, which allows manufacturing of
complex geometry and high aspect ratio components (König & Klocke, 2007). Fig. 1(b) shows the
material removal mechanism of Wire-EDM (König & Klocke, 2007). The contact free and low force
∗
Corresponding author
Tel.: 1-205-348-2615; fax: +1-205-348-6419. E-mail address: [email protected]
Selection and peer-review under responsibility of the Scientific Programme Committee of NAMRI/SME
c The Authors. Published by Elsevier B.V.
1
Process Capability of Wire-EDM of Nitinol at Main Cut and Trim Cut Modes J.F. Liu and Y.B. Guo
between the workpiece and the electrode wire avoid severe tool wear and other issues inherent in
mechanical cutting.
(b)
(a)
Tool
Pulse generator
Dielectric
Heat flux
(Gaussian distribution)
Workpiece
Fig. 1 (a) Schematic of main cut and trim cut in Wire-EDM; (b) Wire-EDM mechanism
The recent development of “CleanCut” generator enables Wire-EDM to minimize thermal damage.
Fig. 2 shows the typical pulse profiles for modern Wire-EDM at main cut and trim cut modes. The
main cut mode with trapezoidal pulse shape and long discharge duration is used to obtain the basic
dimension and geometry, the subsequent trim cut mode with shorter discharge duration is used to
modify the main cut surface, and the finish trim cut mode with ultrahigh frequency is to obtain the
required surface finish and integrity with minimal thermal damage. (Klink et al., 2011), (Aspinwall et
al., 2008) and (Li et al., 2014) studied Wire-EDM of hardened steel, titanium, and Inconel alloys. The
results showed that isotropic surfaces with Ra as low as 0.2 µm can be produced at finish trim cut
mode. Also, the EDMed surface with a very thin white layer is free of microcracks and microvoids.
(a) Main cut mode
(b) First trim cut mode
(c) Finish trim cut mode
Fig. 2 Pulse profiles in Wire-EDM cutting of ASP 23 in CH-oil dielectric
Compared to hardened tool steels, titanium alloys, and Inconel alloys, Nitinol imposes a great
challenge such as thermal induced phase transformation. The previous work on EDM of Nitinol is
only limited to the effect of discharging energy on material removal rate and microstructures (Lin et
al., 2001; Huang et al., 2005; Theisen & Schuermann, 2004; Hsieh et al., 2009; Alidoosti et al., 2013).
Very little research has been done to explore the process capability of Wire-EDM, in particular
relative to traditional mechanical cutting. In this study, the process capability of state-of-the-art
Wire-EDM (in both CH-oil and DI-water) in machining Nitinol was explored at main cut and trim cut
modes. Surface finish by Wire-EDM was also compared to that by mechanical cutting.
2
Process Capability of Wire-EDM of Nitinol at Main Cut and Trim Cut Modes J.F. Liu and Y.B. Guo
2 Experiment Procedure
Wire-EDM experiments were conducted with two state-of-the-art Wire-EDM machines using
CH-oil and deionized water (DI-water) based dielectrics and standard 250 µm brass wires (CuZn36).
Nitinol SE508 (50.8 at.% Ni-49.2 at.% Ti) sheets with dimension of 50 mm (L) × 25 mm (W) × 1.4
mm (H) were chosen as the work material (Table 1). The brass wire and work material properties are
shown in Table 2. Since the commercial Wire-EDM machines are designed to meet specific materials,
thickness and productivity, only the discharge mode could be chosen from the built-in database instead
of specific machining conditions such as discharge voltage, discharge current and discharge pulse
duration. The material was machined using an appropriate sequence of main cut followed by
subsequent trim cuts at reduced discharge energy according to the standard machining technologies
provided by the machine tool manufacturer. In CH-oil based dielectric, one main cut was followed by
9 subsequent trim cuts and in DI-water by 6 cuts. The trend of discharge intensity and frequency from
main cut to finish trim cut is similar to the one shown in Fig. 2. Table 3 shows the discharge energy
levels at different discharge modes.
Table 1 Compositions of Nitinol SE508
Ni (nominal)
55.8 wt.%
Ti
Balance
O (max)
0.05 wt.%
C (max)
0.02 wt.%
Table 2 Material properties of electrode and workpiece
Material property
Brass wire
Nitinol SE508
Yield stress (MPa)
950
2000
Elastic modulus (GPa)
97
42.3
Poisson ratio
0.34
0.3
Density (Kg/m3)
8500
6450
Melting point (°C)
930
1310
Table 3 Process modes of Wire-EDM
Wire-EDM Mode
Dielectric
Discharge Energy
Main cut (MC)
CH-oil
High
st
1 trim cut (TC)
CH-oil
Lower
Finish trim cut (FC)
CH-oil
Lowest
Main cut (MC)
DI-water
High
1st trim cut (TC)
DI-water
Lower
Finish trim cut (FC)
DI-water
Lowest
The Wire-EDMed surfaces from main cut (MC) mode, first trim cut (TC) mode and finish trim cut
(FC) mode were examined to investigate the evolution of surface characteristics during the Wire-EDM
process. The surface characteristics were investigated by stylus profiler, scanning electron microscopy
(SEM), and energy-dispersive X-ray spectroscopy (EDS).
3
Process Capability of Wire-EDM of Nitinol at Main Cut and Trim Cut Modes J.F. Liu and Y.B. Guo
3 Results and Analysis
3.1 Surface topography
The Wire-EDMed surfaces (Fig. 3) at MC mode are characterized by the “coral reef”
microstructure in both dielectrics. The spherical debris was formed by the deposition and
re-solidification of vaporized material and splashed molten material due to the quenching effect of the
dielectric. The subsequent TC mode significantly reduced the number of spherical debris, and FC
mode produced isotropic surfaces with uniformly shallow craters free of debris. Similar phenomenon
was also observed in previous researches in EDM of ASP23 mold steel and Inconel 718 alloy (Klink
et al., 2011; Li et al., 2014). At MC mode with relative high discharge energy, most of the molten
materials have been splashed by the high plasma pressure into the dielectric. A certain amount of
molten materials was rapidly quenched, re-solidified, and deposited as fine spherical debris on the
machined surfaces. At FC mode with very low discharge energy, the plasma pressure may be not high
enough to completely eject the molten material into the dielectric, thus, very few spherical debris
could be found on the machined surface. The effect of dielectric on surface microstructure is
manifested by the finer coral reef on the MC surface in DI-water than that in the CH-oil based
dielectric. This phenomenon could be attributed to the higher quenching rate of DI-water than CH-oil
based dielectric.
(a) MC/CH-oil
(c) FC/CH-oil
(b) TC/CH-oil
Microcrack
20 µm
(d) MC/DI-water
(e) TC/ DI-water
Microcrack
20 µm
20 µm
20 µm
(f) FC/ DI-water
Microcrack
20 µm
20 µm
Fig. 3 Surface topography of Wire-EDMed surfaces in CH-oil and DI-water.
Surface microcracks are observed on the MC surfaces in both dielectrics. The formation of
microcracks is induced by the high tensile residual stresses on the surface as the high temperature of
the molten material is rapidly quenched by the dielectrics. The TC surfaces in CH-oil are free of
microcracks, which would be explained by the reduction of tensile residual stress magnitude and depth
in the subsurface as the Wire-EDM temperature would drop at relative low discharge energy. In
contrast, the TC surface in DI-water still has microcracks, which would result from the relative higher
residual stress by the higher quenching rate in DI-water than CH-oil. On FC surfaces in both
dielectrics, no microcracks were formed as the low discharge energy produces low tensile residual
stress which may not fracture the surface materials.
4
Process Capability of Wire-EDM of Nitinol at Main Cut and Trim Cut Modes J.F. Liu and Y.B. Guo
Fig. 4 shows the feed marks on FC surface in DI-water, while it is almost not observed on the FC
surface in CH-oil. The formation of feed mark could be due to the vibration of the brass wire. Since
the viscosity of CH-oil is much higher than water, the brass wire would has been significantly damped
by the CH-oil and resulted in much less vibration. The mechanism for the presence of random micro
pits on FC surface in CH-oil is still not clear, which needs further study in the future.
(b) FC/DI-water
(a) FC/CH-oil
Feed mark
200 µm
200 µm
Fig. 4 Feed marks on the surfaces at FC mode.
3.2 Uncertainty of surface roughness
To demonstrate the process capability in terms of surface roughness over a broad range of
machining conditions, a normal distribution of Ra was computed in this study. According to central
limit theorem in probability theory, a sufficiently large number of independent random variables in
certain condition will be approximately normally distributed (Liu et al., 2014). Thus, a large set of
roughness (Ra) data (thirty six conditions) of the EDMed surfaces was used. After calculating the
mean value and deviation of the measured data (thirty six conditions), Ra distributions from the MC
surfaces to the FC surfaces are given in Figs. 5–7.
Fig. 5 shows the average Ra and distributions for the surfaces at MC, TC, and FC modes in CHoil. The average Ra significantly reduces (3.65 µm vs. 0.22 µm) from MC mode to FC mode. And the
average Ra of TC surfaces (2.64 µm) is also less than the MC surfaces (3.65 µm). By comparing the
width of Ra distributions, the MC and TC surfaces have similar standard deviations (MC/0.23 µm vs.
TC/0.24 µm). It means that Ra uncertainty for the MC and TC surfaces is similar. In contrast, the
average Ra (0.22 µm) and standard deviation of (0.02 µm) for the FC surfaces are much smaller than
the MC and TC surfaces. It implies that FC mode is robust to produce a smooth surface with a narrow
distribution.
Fig. 6 shows the average Ra and distributions for the MC, TC, and FC surfaces in DI-water.
Contrary to the EDMed surfaces in CH-oil, the average Ra of the TC surfaces is even larger (3.69 µm
vs. 2.45 µm) than the MC surfaces. This may be contributed to the removal of partial porous “coral
reef” microstructure while others were not removed at TC mode, which leads to higher Ra data.
However, the standard deviation of TC surface (0.17 µm) is much smaller than the MC surface
(0.47 µm), which indicates a higher uncertainty of Ra distribution at MC mode. The average Ra (0.38
µm) and standard deviation (0.06 µm) for the FC surface are much smaller than the MC and TC
surfaces, which shows that FC mode is also robust to produce a smooth surface with less uncertainty
in DI-water.
5
Process Capability of Wire-EDM of Nitinol at Main Cut and Trim Cut Modes J.F. Liu and Y.B. Guo
Probability density (1/µm)
25
MC
TC
FC
20
15
10
5
0
0
1
2
3
4
Surface roughness Ra ( µm)
5
Fig. 5 Ra distribution of EDMed surfaces in CH-oil.
Probability density (1/µm)
25
MC
TC
FC
20
15
10
5
0
0
1
2
3
4
Surface roughness Ra (µm)
5
Fig. 6 Ra distribution of EDMed surfaces in DI-water.
By combining all the roughness data over the broad range of machining conditions regardless of
dielectric, the average Ra and standard deviation for the MC, TC, and FC surfaces is shown in Fig. 7.
It shows that the MC and TC surfaces have similar average Ra (MC/3.05 µm vs. TC/3.16 µm) and
standard deviation (MC/0.72 µm vs. TC/0.58 µm), but the FC surface has much smaller average Ra
(0.30 µm) and standard deviation (0.09 µm) than the MC and TC surfaces. By comparing the data in
Figs. 5 and 6, it is found that dielectrics result in very different average Ra and distribution for the MC
and TC surfaces but not the FC surfaces. In general, FC mode is very necessary to produce a finishing
surface regardless of dielectric types.
To demonstrate the process capability of Wire-EDM Nitinol, surface roughness by typical
mechanical cutting such as milling and waterjet machining is compared. Fig. 8 shows the range of
surface roughness Ra of the machined surfaces. Wire-EDM is capable of producing smoother surface
with less variation than waterjet machining (Kong et al., 2011) and electrochemical machining (ECM)
(Lee & Shin, 2011). At least, Wire-EDM is comparable to milling in terms of surface roughness (Guo
et al., 2013).
6
Process Capability of Wire-EDM of Nitinol at Main Cut and Trim Cut Modes J.F. Liu and Y.B. Guo
Probability density (1/µm)
5
MC
TC
FC
4
3
2
1
0
0
1
2
3
4
5
Surface roughness Ra (µm)
6
Fig. 7 Ra distribution of EDMed surfaces.
Surface roughness Ra (µm)
7
max: 5.91
min: 5.38
6
5
4
3
2
max: 0.49
min: 0.20
1
max: 0.40
min: 0.19
max: 0.98
min: 0.31
0
[email protected]
Milling
ECM
Waterjet
machining
Fig. 8 Surface roughness comparison by Wire-EDM vs. mechanical cutting.
3.3 Microstructure
The subsurface microstructures of the MC and FC surfaces in DI-water are shown in Fig. 9. It can
be seen that a discontinuous and non-uniform porous white layer (1~10 µm) occurred in MC
subsurface. Microvoids of different size were confined in the white layer, which may result from the
bubble trapped in the material during the re-solidification process. In addition, microcracks in the
white layer do not propagate into the subsurface. For FC subsurface, a very thin sporadic white layer is
barely seen in the subsurface. The absence of microvoids indicates that the majority of molten material
was expelled by the plasma pressure. Microstructure appeared consistent or unchanged in the bulk
material. The high discharge energy in MC mode produces high temperature penetrating deeper into
the subsurface, which melts more material and ultimately results in a thick white layer. On the other
hand, the minimal discharge energy at FC mode produces minimal thermal damage. Therefore,
discharge energy is the critical factor to minimize thermal damage into the subsurface. The subsurface
microstructures of the MC and FC surfaces in CH-oil have similar characteristics to those in DI-water.
7
Process Capability of Wire-EDM of Nitinol at Main Cut and Trim Cut Modes J.F. Liu and Y.B. Guo
(a) MC/DI-water
Microcracks
(b) FC/DI-water
Free of white layer
and microcraks
Porosity
10 µm
10 µm
Fig. 9 Subsurface Microstructure of Wire-EDMed samples.
3.4 Element analysis
Fig. 10 shows the elements detected by EDS for the MC, TC, and FC surfaces. The very high
temperature of plasma instantly melts and vaporizes the material during the discharging process. Both
the brass electrode and Nitinol workpiece were eroded and complex chemical reactions between the
vaporized gas and the molten pool may form various compounds (such as TiO, Ti2O3, TiC and ZnO).
Element diffusion largely decreased with lower discharge energy. For the MC surfaces, high alloying
effect can be seen. Large amounts of Cu and Zn can be found on the top surface of the recast layer,
which typically diffused from the brass electrode. For FC surfaces, very little amount of Zn and Cu
were detected in both CH-oil and DI-water dielectrics, because the expelled molten material would
give less chance for element diffusion. The high cooling rate of DI-water results in more Cu and Zn
diffusion on the machined surfaces instead of flushed away. Minimal element diffusion is critical to
the functionality of a Nitinol product as its phase transformation temperature is very sensitive to the
added foreign elements.
In the previous studies (Alidoosti et al., 2013; Hsieh et al., 2009), high element diffusion occurred
between the workpiece and dielectric. In this research, no C was detected on the EDMed surface in
CH-oil even at MC mode, and very few O was detected on the EDMed surface in DI-water, which
means that element diffusion from dielectric to workpiece is minimal. The “CleanCut” generator of
the EDM machine can provide ultrahigh repetition frequency of spark to minimize thermal damage, so
that element diffusion between workpiece and dielectrics can be minimized.
8
Process Capability of Wire-EDM of Nitinol at Main Cut and Trim Cut Modes J.F. Liu and Y.B. Guo
(a) MC/CH-oil
(d) MC/DI-water
(b) TC/CH-oil
(e) TC/DI-water
(c) FC/CH-oil
(f) FC/DI-water
Fig. 10 EDS analysis of the Wire-EDMed surfaces.
4 Conclusions
This study focuses on the process capability in Wire-EDM of Nitinol from main cut to finish trim
cut in CH-oil and DI-water. Key findings on surface characteristics may be summarized as follows:
• “Coral reef” surface topography is typical for the EDMed surfaces at main cut and first trim
cut, while finish trim cut produces an isotropic surface.
• Microcracks are formed on surfaces at main cut for both dielectrics. However, the DI-water
produced microcracks even at the first trim cut due to the high tensile residual stress resulted
from the higher quenching rate than CH-oil.
9
Process Capability of Wire-EDM of Nitinol at Main Cut and Trim Cut Modes J.F. Liu and Y.B. Guo
•
•
•
A normal distribution of Ra was provided to characterize the uncertainty of surface roughness
over a broad range of machining conditions. Compared to the main cut and first trim cut,
finish trim cut is robust in terms of a small average Ra and standard deviation.
The relative high discharge energy at main cut results in a discontinuous and porous thick
white layer, which can be minimized at finish trim cut. Microcracks in the subsurface are
confined in the thin white layer.
High element diffusion (Cu, Zn) occurred at main cut and the first trim cut, while alloying
effect is minimal at finish trim cut. In addition, element diffusion mainly happened between
the workpiece and the electrode, while element diffusion from the dielectric was not detected.
Acknowledgement
The work has been supported by the NSF CMMI #1234696.
References
Alidoosti, A., Ghafari-Nazari, A., Moztarzadeh, F., Jalali, N., Moztarzadeh, S., Mozafari, M. (2013).
Electrical discharge machining characteristics of nickel–titanium shape memory alloy based on
full factorial design. Journal of Intelligent Material Systems and Structures, 24(13), 15461556.
Aspinwall, D. K., Soo, S. L., Berrisford, A. E., Walder, G. (2008). Workpiece surface roughness and
integrity after WEDM of Ti–6Al–4V and Inconel 718 using minimum damage generator
technology. CIRP Annals - Manufacturing Technology, 57(1), 187-190.
Bansiddhi, A., Sargeant, T., Stupp, S., Dunand, D. (2008). Porous NiTi for bone implants: a review.
Acta Biomaterialia, 4(4), 773-782.
Calkins, F. T., Mabe, J. H., Ruggeri, R. T. (2008). Overview of Boeing's shape memory alloy based
morphing aerostructures. ASME 2008 Conference on Smart Materials, Adaptive Structures and
Intelligent Systems, 885-895.
Duerig, T., Pelton, A., Stöckel, D. (1999). An overview of nitinol medical applications. Materials
Science and Engineering: A, 273, 149-160.
Guo, Y., Klink, A., Fu, C., Snyder, J. (2013). Machinability and surface integrity of nitinol shape
memory alloy. CIRP Annals - Manufacturing Technology, 62(1), 83-86.
Henderson, E., Buis, A. (2011). Nitinol for prosthetic and orthotic applications. Journal of Materials
Engineering and Performance, 20(4), 663-665.
Hsieh, S. F., Chen, S. L., Lin, H. C., Lin, M. H., Chiou, S. Y. (2009). The machining characteristics
and shape recovery ability of Ti–Ni–X (X=Zr, Cr) ternary shape memory alloys using the
wire electro-discharge machining. International Journal of Machine Tools and Manufacture,
49(6), 509-514.
Huang, H., Zheng, H., Liu, Y. (2005). Experimental investigations of the machinability of Ni50.6Ti49. 4
alloy. Smart Materials and Structures, 14(5), S297.
Klink, A., Guo, Y. B., Klocke, F. (2011). Surface integrity evolution of powder metallurgical tool
steel by main cut and finishing trim cuts in Wire-EDM. Procedia Engineering, 19(0), 178-183.
Kong, M. C., Axinte, D., Voice, W. (2011). Challenges in using waterjet machining of NiTi shape
memory alloys: an analysis of controlled-depth milling. Journal of Materials Processing
Technology, 211(6), 959-971.
10
Process Capability of Wire-EDM of Nitinol at Main Cut and Trim Cut Modes J.F. Liu and Y.B. Guo
König, W., and Klocke, F., 2007, Manufacturing processes volume 3: removal, generation and
laser material processing. Springer.
Lee, E. S., Shin, T. H. (2011). An evaluation of the machinability of nitinol shape memory alloy by
electrochemical polishing. Journal of Mechanical Science and Technology, 25(4), 963-969.
Li, L., Wei, X., Guo, Y., Li, W., Liu, J. (2014). Surface integrity of Inconel 718 by Wire-EDM at
different energy modes. Journal of Materials Engineering and Performance, 23(8), 3051-3057.
Lin, H., Lin, K., Cheng, I. (2001). The electro-discharge machining characteristics of TiNi shape
memory alloys. Journal of Materials Science, 36(2), 399-404.
Liu, J., Li, L., Guo, Y. (2014). Surface integrity evolution from main cut mode to finish trim cut
mode in W-EDM of shape memory alloy. Applied Surface Science, 308, 253-260.
Theisen, W., Schuermann, A. (2004). Electro discharge machining of nickel–titanium shape
memory alloys. Materials Science and Engineering: A, 378(1–2), 200-204.
Weinert, K., Petzoldt, V., Kötter, D. (2004). Turning and drilling of NiTi shape memory alloys.
CIRP Annals-Manufacturing Technology, 53(1), 65-68.
11
`