Effects of heat treatment and welding process on superelastic behaviour and microstructure of micro electron beam welded NiTi

1 Introduction

Since the discovery of NiTi and its extraordinary properties in 1963 [1], especially superelastic (SE) alloys containing about 56 wt.% Ni became the material of choice for medical devices such as stents and orthodontic devices [2]. Next to the rubber-like deformation behaviour, SE NiTi is characterised by good biocompatibility and high corrosion resistance [2], [3], [4]. The disadvantage in using NiTi as construction material is its difficult processing. Above all joining by welding is still challenging, as SE properties are impacted by heat input. In the literature are different microstructural modifications discussed that may affect the stoichiometric ratio of NiTi matrix. From this a shift of phase transformation temperatures up to the loss of SE characteristics may occurs, if no austenitic structure at ambient temperature is present.

Heat treatment is necessary for adjustment of superelasticity and shape setting of NiTi components. High temperatures above 550°C lead to Ni₃Ti₂-precipitations, which bind high amounts of Ni from the matrix determining a sharply increased austenite finish temperature (A_f). On the contrary, heat treatment of <10 min at 500–550°C reduces the A_f due to the dissociation of Ni-rich Ni₁₄Ti₁₁-precipitations [5].

Longer heat treatment causes the formation of a titanium oxide layer (TiO₂) and a second nickel-rich layer (Ni₃Ti) beneath [6], [7]. Another possibility to reduce the A_f is to generate fine grains as no precipitations appear in a structure with grain sizes <150 nm, accompanied by higher tensile strengths and decreased residual elongation [8].

During the last 20 years, studies were focused on laser welding for joining NiTi which was found to be suited because of a minimal heat affected zone and high cooling rates [9], [10], [11]. Additionally to these advantages micro welding with electron beam (MEBW) features material independent energy absorption and contactless processing under high vacuum, which is necessary particularly with regard to very reactive elements, like titanium. Also due to nearly inertia free and precise beam deflection in combination with in-situ quality control using the high-resolution SEM mode MEBW is well suited for welding medical applications in micro-range. However, scientific results about EBW of NiTi are rare. Yang et al. [12] worked out that nickel evaporates earlier in the weld pool of an electron beam process due to the lower saturation vapour pressure compared to titanium causing a substantially higher A_f. The majority of scientists examining laser welding of NiTi observed also an increase of transformation temperature in the weld metal and explained that with nickel-rich precipitations [10], [13]. Heat treatment and welding is generally a combined process of NiTi applications, but the outcome of both is still insufficiently investigated. Therefore, the aim of this study is to analyse the changing of microstructural and SE properties depending on pre heat treatment at different temperatures and various energy input on thin NiTi foils using MEBW particularly because this method of micro joining had not been considered until now.

2 Experimental details

Ti-55.8 wt.% Ni foils with 200 μm thickness were used throughout this investigation. The as-received flat annealed and oxide free material was heat-treated under atmosphere for 10 min at 350°C, 450°C or 550°C (sample abbreviations: HT – heat treated, NHT – not heat treated) to determine the influence of the heat treatment and oxide layer on SE behaviour of the welding joint. All experiments were done with a micro electron beam welding machine “MEBW-60” of the FOCUS company. The used cathode type for investigations was a LaB₆ single crystal, which allows a beam diameter of <25 μm and thus a constant beam characteristic and an increased power density compared to a tungsten hairpin filament. The maximum beam power is 2 kW with an accelerating voltage of 60 kV. Bead-on-plate welds with different heat input per unit length “Q” (welding speed: 20, 40, 60, 80 mm/s) were produced on various heat-treated foil samples.

Micrographs were made along cross direction of the welded joints and further treated with a colour etching process with Beraha III stock solution. The microstructure of the base material (BM), heat-affected zone (HAZ) and welding zone (WZ) were studied in polarized light with metallographic microscope Zeiss “Axio Imager M2m“. According to the ASTM F2004 standards, the thermal transformation behaviour was determined by differential scanning analysis (DSC), which was operated using a “DSC 204 Phoenix” of the Netzsch company within a temperature range from 193 K to 373 K with an controlled cooling/heating rate of 10 K/min. Samples of 13—22 mg were cut from base material and from a field of parallel welded seams (Figure 1).

Figure 1:

Field of parallel electron beam welded seams for DSC-samples of the weld.

X-ray Photoelectron Spectroscopy (XPS) was used to obtain the elemental composition and binding states (oxide state) with an information depth of about 6 nm on selected samples. In this study the XPS measurements were performed with an “Axis Ultra” system (Kratos) of the employing monochromatised Al Kα X-Ray radiation at pass energy of 20 eV (high resolution scan) and an off angle of 45°. Additionally, energy dispersive X-ray spectroscopy (EDX) with “Leo 1530” (Zeiss) at an accelerating voltage of 15 kV was used to examine chemical compositions of sample surface with an information depth of a few microns.

3 Results and analysis

3.1 Microstructure and hardness

The samples welded by 60 mm/s showed the best compromise between a equiaxed grain structure in the weld seam and a low transformation temperature, so this parameter was chosen for further investigations. Considering the microstructure as it is shown in Figure 2, dendritic growth of large grains perpendicular to the weld centreline can be seen in the micrograph of the WZ. As Q decreases, those dendrites become less oriented and more compact. In contrast, fine recrystallised equiaxed grains are located in the HAZ. Grains in the BM are smaller than in the WZ and additional the existence of dislocations in the BM may cause the highest hardness is measured. These dislocations annihilate through the heat input in WZ and HAZ, which implies a decrease in hardness. Furthermore, it must be considered that the SE of austenitic microstructure in BM leads to lightly biased micro hardness measurements with increased values.

Figure 2:

Colour etched micrograph of a cross-section of the welded joint (60 mm/s) and corresponding microhardness profile.

3.2 Composition of surface layer

Surface composition (presented in at.%) and the corresponding Ti/Ni measured by XPS surface scan to a depth of 6 nm for BM and WZ with a welding speed of 60 mm/s of NHT samples are summarised in Table 1. The presence of a high C content and small amounts of N, Ca and Si is due to environmental contaminations.

Table 1:

Surface composition (at.%) of BM NHT and WZ NHT (60 mm/s).

Sample	Ni	Ti	Ti/Ni ratio	O	C	Ca, N, Si
BM NHT	1.71	11.47	6.71	41.09	41.68	4.05 (Ca, N)
WZ NHT 60	1.84	7.83	4.26	26.85	60.44	3.04 (Si)

The surface Ti/Ni ratio of the WZ sample without HT is 4.26 and smaller than measured in the BM sample, which is 6.71. That results mainly from an increase of Ti which is caused by chemical affinity for oxygen. The high power density of the electron beam leads to destruction of oxide layer and the vacuum prohibits fast regeneration, which can be seen at the low O content on the surface of the weld seam.

As depicted in Figure 3, three oxidation peaks in accordance with [14] were measured in the WZ sample: TiO (454.6 eV), Ti₂O₃ (456.7 eV) and TiO₂ (458.7 eV), in which the majority of titanium is in the form of TiO₂ (65.7 at.% to 100 at.%).

Figure 3:

High resolution scan of Ti2p for BM NHT and WZ NHT (60 mm/s).

For both samples, the high resolution scan of Ni shows the existence of metal-bonded Ni (852.6 eV). In addition, EDX measurements in BM show an increase of oxygen content on surface with increasing heat treatment temperature (Figure 4). The formation of Ti oxides on the surface of BM leads to a lower Ni content on the surface and might cause the rise of Ni content in NiTi matrix.

Figure 4:

Plots of oxygen and nickel content measured by EDX against heat treatment.

3.3 Superelastic properties

As shown in Table 2, there is a strong correlation between A_f and Q. With higher energy input the transformation temperature rises.

Table 2:

Dependence of austenite finish temperature (A_f) on heat input per unit length (Q).

Q [Ws/mm] v [mm/s]	1.05 80 mm/s	1.15 60 mm/s	1.35 40 mm/s	1.8 20 mm/s
Af (WZ NHT)	30.0°C	30.5°C	47.4°C	51.5°C

However, the A_f is compared to the as-received material (A_f = 10.1°C) substantially increased and reaches values above room temperature, resulting from the MEBW process. So, no matter which weld speed and beam current is chosen, mainly martensite is present in the melted area.

If the SE NiTi-alloy is heat treated under atmosphere before the welding process, the A_f results as seen in Figure 5. While the SE properties of the BM seem not to be influenced by HT at 350°C, temperatures at 450°C cause a significant increase of A_f. Ni-rich precipitations [5] mainly affect the stoichiometric ratio of NiTi matrix and cause a rise of A_f at 450°C and dissociate above 500°C.

Figure 5:

A_f of BM and WZ as a function of heat treatment.

Next to that the formation of oxide layer must have a positive influence on SE behaviour. Interesting is the effect by heat treatment at 550°C upon the BM. The large drop of A_f at −5.5°C, which is below A_f of the as-received material, probably results as well from the formation of Ti oxides on the surface and therefore causes the rise of Ni content in NiTi matrix. This thesis is confirmed by further DSC measurements on equal heat treated BM (but under vacuum), which show in contrast an approximately constant curve linearity of A_f (∼45°C ± 5°C).

The A_f of the melted material in the WZ decreases slightly with higher previous heat input, but is without exception clearly above room temperature. The loss of SE properties in the WZ corresponds with foregoing studies [10], [13].

4 Conclusion

The effects of micro electron beam welding and preheating on the changes of microstructural and superelastic properties were studied. The key findings are:

1. The microstructure consists of columnar dendritic structure in weld zone and a recrystallised structure in the HAZ. As the heat input decreases, also the A_f of the WZ decreases, but still remains above room temperature (martensitic microstructure).

2. Pre-heating at temperatures of 550°C (10 min) favoures SE of the base material by a large reduction of A_f, but getting lost through welding process.

3. Next to Ni-rich precipitation the formation of Ti-rich oxide layer affects the stoichiometric ratio of the NiTi matrix and therefore may be responsible for lower austenite finish temperatures.