Fatigue Resistance Improvement in Cold-Drawn NiTi Wires Treated with ALD: A Preliminary Investigation

Abstract

NiTi shape memory alloys (SMAs) are widely studied for their potential applications, and atomic layer deposition (ALD) is an effective technique for coating them due to its precise control over coating thickness. This study investigates the impact of Al₂O₃ coating on the fatigue behavior of cold-drawn NiTi wires with a 0.125 mm diameter. The wires were coated using atomic layer deposition (ALD) with 100 ALD cycles. Fatigue tests were conducted in tensile mode at room temperature, applying cyclic loading between 0–50, and 700 MPa (700 MPa is almost 40% of ultimate tensile strength). The results show that the cold-drawn NiTi wires failed after an average of 7500 tensile loading cycles, while the lifetime of the coated and stretched NiTi wires with a preload of 1.7–2.8 kg significantly improved, with an average of 293,000 cycles before failure.

1. Introduction

1.1. NiTi Shape Memory Alloys and Their Applications

It has been more than 60 years since the discovery of an approximately equiatomic NiTi alloy called Nitinol. The last three letters in the alloy’s name stand for Naval Ordnance Laboratory, the place of the discovery [1]. During these six decades, NiTi shape memory alloys (SMAs), well known for the shape memory effect and superelasticity due to the reversible solid phase martensitic transformation [2,3,4], have found many applications in industry and medicine. These unique properties allow NiTi alloys to recover their original shape after deformation and to exhibit high elasticity, making them invaluable in various engineering and biomedical applications.

Some examples of NiTi applications include NiTi frames of reading glasses [5], an NiTi switch-off element in electric tea kettles [6], and NiTi pipe couplers used, among other things, to connect titanium hydraulic tubing in the Grumman F-14 aircraft [7]. Additionally, NiTi is used in various actuators and sensors [8,9], fasteners such as Velcro-type NiTi hook fasteners [10], vibration and seismic applications [11], NiTi torque tubes installed in smart wings for wing morphing [12], NiTi self-expandable stents [13], and NiTi elements used in endoscopes [14]. The versatility of NiTi SMAs arises from their ability to perform reliably under cyclic loading conditions, which is critical for dynamic applications.

However, some of the mentioned applications, and not only those, are limited by the fatigue life of cyclically loaded NiTi elements (e.g., Velcro-type NiTi hook fasteners or elastocaloric active regenerators [15]). The degradation of mechanical properties due to repeated stress cycles can lead to premature failure, posing safety risks and reducing the operational lifespan of components. Hence, fatigue-life studies of cyclically loaded NiTi parts have received much attention [16,17]. Understanding the mechanisms of fatigue failure and identifying methods to enhance fatigue resistance are essential for extending the durability and reliability of NiTi-based components. Not long ago, some techniques, such as laser shock peening and shot peening [18,19,20], were proposed to improve the fatigue strength of alloy parts, which are also applicable to NiTi wires by generating compressive stresses on the surface.

1.2. Cold-Drawn NiTi Wires

The application potential of cold-drawn NiTi wires is lower than that of superelastic and shape-memory NiTi wires. Thin (<1 mm diameter) cold-drawn NiTi wires, typically consisting of a mixture of austenite (B2), martensite (B19′), and amorphous phase, have high tensile strength at room temperature, of the order of 1.5 GPa (depending on the amount of cold work) and maximum elongation of 3–4% [21]. These mechanical properties make cold-drawn NiTi wires suitable for applications requiring high strength and moderate flexibility, though their limited superelastic behavior restricts broader use.

To our knowledge, there are only a few studies on the fatigue strength of cold-drawn NiTi wires. Hua et al. presented high fatigue resistance of NiTi micro-pillars composed of alternating amorphous and nanocrystalline layers under compressive loading [22]. Schaffer and Plumley, using a rotary bend fatigue tester, studied the fatigue life of superelastic wires, thus not directly the fatigue life of as-drawn wires; however, they paid attention to the amount of prior cold work [23]. This gap in research highlights the need for comprehensive studies on the fatigue performance of cold-drawn NiTi wires, especially under tensile and cyclic loading conditions relevant to practical applications. Cold drawn NiTi wires, characterized by their high tensile strength, find direct applications in environments where mechanical robustness and corrosion resistance are critical.

1.3. Atomic Layer Deposition

Atomic layer deposition (ALD) is an attractive CVD (Chemical Vapor Deposition) method [24] that is well applicable to NiTi SMAs [25,26,27,28,29] due to the well-controllable coating thickness and outstanding conformality of the ALD coatings. No other thin-film technique can approach the conformality achieved by ALD on high-aspect-ratio structures, such as trench capacitors for DRAM (Dynamic Random-Access Memory) [30] or microcracks in an alloy. The ALD coating has the potential to improve the biocompatibility and corrosion resistance of NiTi [31], which is particularly important for medical implants and devices exposed to harsh physiological environments.

The idea of using ALD coating on the NiTi surface was first realized 10 years ago [25]. However, the drawback of the ALD coatings applied to NiTi might be delamination due to large surface strains in NiTi (up to 10%, locally) [26], especially if the coating is made of ceramic material. Despite these challenges, ALD remains a promising technique for enhancing the surface properties of NiTi components. In the present study, the ALD method is used, for the first time, on a stretched NiTi thin wire to improve fatigue resistance. This innovative approach aims to explore the potential of ALD coatings to mitigate fatigue-related failures by providing a protective barrier that can withstand cyclic mechanical stresses.

2. Experimental

2.1. NiTi Wire Samples

A NiTi as-drawn wire with a diameter of 0.125 mm was purchased from Smatec (Belgium). The length of the prepared wire samples ranged from 60 to 70 mm. Steel capillaries 20–30 mm long, with an inner diameter of 0.180 mm, were attached to both ends of the wire samples to facilitate the stretching of these samples. Special care was taken in attaching the capillaries. The attachment was achieved by inserting the NiTi wire and compressing the capillaries together with the wire. However, the applied compressive force was carefully controlled to prevent damage to the wire at the critical point where it exits the capillary. If breakage occurred at this critical point during fatigue testing, the corresponding test results were excluded from the analysis.

The ultimate elongation and ultimate tensile strength of the as-drawn NiTi wires were 4–5% and 1700–1850 MPa, respectively. After annealing at 400–450 °C for 5 min, the wires exhibited superelasticity at room temperature.

Wire samples intended for the ALD were cleaned according to the following procedure:

(i)

Cleaning in IPA (isopropyl alcohol) by sonication for 5 min,

(ii)

cleaning in DI (deionized) water by sonication for 5 min,

(iii)

cleaning in Ethyl alcohol by sonication for 5 min,

(iv)

cleaning in DI water by sonication for 5 min.

The procedure was repeated twice, and finally, the samples were dried out in nitrogen gas flow.

2.2. Fatigue Tests

The fatigue tension tests were carried out using a custom-made tensile tester MITTER (Miniature Tensile Tester) equipped with an environmental chamber, electrically conductive grips, a load cell, a linear stepping motor, and a position sensor. The testing device enabled us to control temperature and stress and measure strain and electric resistance of the wire samples. More details of MITTER can be found in Ref. [32]. The fatigue tests were conducted at room temperature, with cyclic loading in the stress range of 0 to 700 MPa, where the minimum stress varied between 0 and 50 MPa, until the wire specimen fractured.

2.3. The ALD Process

The ALD was realized with a custom-made deposition system at TIRI (Hsinchu, Taiwan). In the ALD process, trimethylaluminum (TMA) was used as a precursor of Al, whereas water vapor served as an oxygen source. One typical ALD cycle consisted of 0.2 s TMA precursor pulse followed by 10 s N₂ purge (2 Torr) to remove excess precursor and the by-products. After the N₂ purge, a 0.1 s water pulse was applied followed by another 10 s N₂ purge to complete the ALD cycle. The pumping time after each purge was 2 s. Each deposition consisted of just 100 ALD cycles (100 Al₂O₃ ALD cycles on NiTi surface corresponds to Al₂O₃ thickness of about 10 nm [25]). The temperature of NiTi wire samples in the ALD chamber was influenced by preheated gasses and gas tubes; hence, temperature might fluctuate during the deposition; however, it did not exceed 100 °C. The wire samples were stretched in the ALD chamber and held in a stretched state during the deposition, with different loads of 1.7, 2.2, and 2.8 kg applied, which remained constant within a single deposition (Figure 1). The tensile force acting on the wire was influenced by the applied load (1.7, 2.2, and 2.8 kg) and the chamber pressure (2 Torr, during N₂ purging). A decrease in chamber pressure resulted in a corresponding decrease in tensile force.

2.4. Analysis of NiTi Wire Samples Using Electron Microscopy

The surfaces of the as-purchased NiTi wire and NiTi wire with ALD coating were examined using scanning electron microscope (SEM) (Tescan, FERA 3 (TESCAN Brno, s.r.o., Brno, Czechia)). Samples for transmission electron microscopy (TEM) of the mechanically cycled wires (until the fracture) were prepared using the Focused Ion Beam (FIB) method using FEI Quanta 3D FIB-SEM electron microscope (FEI Quanta: Thermo Fisher Scientific, Waltham, WA, USA). The FIB system, equipped with a local ion source (LIS), employed gallium ions for precise material removal. Prior to FIB machining, a tungsten layer was locally deposited on the wire surface within the chamber to protect it during the FIB process. TEM lamellas were extracted from the mechanically cycled wire sample to see cross-sectional area adjacent to the wire surface. The following accelerating voltages and FIB currents were applied: 30 kV and 15–30 nA for coarse milling, 30 kV and 0.5–3 nA for moderate milling, and 5 kV and 48 pA for fine milling. The HAADF (high-angle annular dark field) images were obtained to identify the individual layers adjacent to the surface using TEM (FEI Tecnai TF20 X-twin field emission gun (FEI Tecnai: Thermo Fisher Scientific, WA, USA)).

3. Results and Discussion

The idea of potentially increased fatigue life of a wire consists of depositing of ALD coating on the surface of the stretched wire (Figure 2). During stretching, it is assumed that many microcracks on the surface become open. (Multiple mechanical load cycles can be applied in advance to increase the microcrack density on the surface). ALD is a promising method that enables uniform coated layer thickness even in microcracks. After the deposition and unloading, it is assumed that compressive stress (σ < 0) forms in the vicinity of each microcrack, which has a positive effect for suppressing further growth of the microcracks. Al₂O₃ coating was chosen because of the possibility of depositing Al₂O₃ layers at a relatively low temperature of the sample [33], the biocompatibility of Al₂O₃ [34], wear resistance, and high Young’s modulus [35,36,37]. However, in our study, it was not possible to reduce the deposition temperature to room temperature. Hence, the potential influence of unintended heat treatment on fatigue resistance cannot be excluded in this study. To assess the effect of this unintended heat treatment during the ALD process, loading-to-rupture experiments were conducted. Given that the deposition temperature exceeds room temperature, we chose to work with as-drawn NiTi wires rather than annealed wires, as the thermomechanical properties of the latter are more sensitive to temperature variations.

The wire surfaces of the as-purchased and Al₂O₃-coated NiTi wires are shown in Figure 3. Due to the nanometric thickness of the coating, there is no clear indication of an Al₂O₃ layer from Figure 3. Even Figure 3d, which is the back-scatter electron image showing a more compositional than surface relief contrast, provides no evidence of an Al₂O₃ layer composed of the light elements. However, Figure 3 presents a basic characterization of the wire surface. Cracks and signs of delamination in the naturally formed TiO₂ layer are visible. The white arrows in the SEM images indicate microcracks in the exposed alloy region where the TiO₂ layer has detached. The large plastic deformations induced during the cold-drawing process conducted by the wire producer are well accommodated by the NiTi matrix; however, they are detrimental to the brittle, naturally grown TiO₂ ceramic layer.

The presence (or absence) of Al₂O₃ on the surface of the ALD-coated (or uncoated) sample can primarily be determined from the EDS (energy-dispersive X-ray spectroscopy) analysis. The Al K peak is clearly visible in the energy-dispersive X-ray spectrum of the coated-wire sample with the applied Al₂O₃ (Figure 4b), whereas no aluminum peak is present in the spectrum of the as-purchased NiTi wire (Figure 4a).

Table 1. List of fatigue-tested NiTi wire samples without ALD coating. Nf is the number of mechanical cycles (0 to 700 MPa, with the minimum stress between 0 and 50 MPa) at room temperature until fracture occurred. SD stands for standard deviation.

NiTi Sample ID	Nf	Nf_mean	SD
No_ALD_wire_1	2772	7535	6219
No_ALD_wire_2	12,860
No_ALD_wire_3	1262
No_ALD_wire_4	10,925
No_ALD_wire_5	15,360
No_ALD_wire_6	2032

presents the number of mechanical loading cycles (fatigue or tensile cycles) to fracture (Nf) for the as-purchased samples. The average number of cycles, Nf_mean, is 7535 cycles, with a relatively large standard deviation (SD = 6219 cycles). This variability could potentially be explained by differences in the degree of crystallinity of the samples, as well as the random formation of critical microcracks that may arise during the wire drawing process. These factors are proposed as possible contributors, though further investigation would be required to confirm their impact.

The number of mechanical cycles until the fracture of the coated wires is summarized in

Table 2. List of fatigue-tested NiTi wire samples with ALD coating. Nf denotes the number of mechanical cycles (ranging from 0 to 700 MPa, with the minimum stress between 0 and 50 MPa) at room temperature until fracture occurred.

NiTi Sample ID	Load Applied During ALD (kg)	Stress Corresponding to Applied Load (MPa)	Nf_ald	Nf_ald_mean	SD_ald
ALD_wire_1	1.7	1359	5302	293,875	476,896
ALD_wire_2	2.2	1759	320,024
ALD_wire_3	2.2	1759	30,417
ALD_wire_4	1.7	1359	752
ALD_wire_5	2.8	2238	1,112,882 *

* Sample ALD_wire5 did not fracture, and the experiment was terminated due to time constraints.

. The average value, Nf_ald_mean, is 293,875 mechanical cycles. The standard deviation, SD_ald, is again large (SD_ald is 476,896 mechanical cycles). There is an obvious increase in Nf_ald (Nf_ald/Nf = 39). While the mean value and standard deviation (SD) were calculated for the data in Table 2, it is important to note that these calculations should be interpreted with caution, as the values are derived from measurements under different loading conditions, which may influence the results.

An analysis of the load applied during the ALD process (1.7, 2.2, 2.8 kg) reveals a correlation between Nf_ald and the load level. However, the correlation cannot be evaluated as statistically significant due to the low number of tests. It is noteworthy that annealed NiTi wire exhibiting a superelastic effect typically endures only a few thousand loading cycles when subjected to stresses in the range of 0–700 MPa at room temperature. However, the strain of the superelastic wire at the stress level around 700 MPa is several times greater than that of the as-cold drawn NiTi wire.

Figure 5a–f present the stress–strain diagrams with cyclically applied stress in the range of 0–700 MPa, up to the fracture of individual non-coated wire samples. Similarly, Figure 6 illustrates the stress–strain diagrams up to fracture at room temperature for individual ALD-coated samples (a)–(e), as well as the electrical resistance versus the number of loading cycles (f) for the NiTi wire sample ALD_wire_3. The cyclic behavior is stable, with minimal accumulated plastic deformation, and the stress–strain curves are nearly linear. A narrow hysteresis is observed in each individual stress–strain curve (not shown in the figure). Figure 6e illustrates an accidental overshoot of the applied tensile stress. Nevertheless, this overshoot did not lead to wire failure. The electrical resistance varies with the number of mechanical cycles applied. After an initial decrease during the first 500 cycles, the electrical resistance increases with the number of cycles. The two sharp drops in electrical resistance in panel (f) (black curve) are artifacts caused by some sudden changes in electrical connection between the sample and the clamping system. The brown curve represents the electrical resistance data after compensation for these two artificial drops. In the compensated curve, the initial electrical resistance value is lower than the value immediately preceding wire fracture. Generally, electrical resistance is proportional to the elongation of the tested wires and can indicate microstructural changes after many tensile loading cycles [38].

Figure 7 shows the HAADF images that were obtained to identify the individual layers adjacent to the surface. The top layer is a tungsten-based layer applied by the STEM (Scanning Transmission Electron Microscopy (FEI Quanta, FEI Tecnai: Thermo Fisher Scientific, WA, USA)) operator to manipulate the lamella. Below the tungsten-based layer, there is a naturally formed titanium dioxide layer. Below the titanium dioxide layer, there is a nickel-rich layer (the chemical composition close to Ni₃Ti) followed by the NiTi core. It is worth noting that TiO₂ is non-stoichiometric in this case, consistent with the findings in [39]. The Al₂O₃ layer is absent in the image. Figure 7c shows the chemical element contents along the path denoted by the orange line in Figure 7b. The only chemical elements found in the sample using the line scan were Ni, Ti, O₂, and W. Several other surface examinations did not reveal any presence of Al₂O₃, most likely due to delamination after many loading cycles. The failure to detect any traces of Al₂O₃ in the coated sample after the fatigue test raises questions regarding the proposed mechanism for the improved fatigue resistance of the coated samples, specifically the formation of compressive stress around the Al₂O₃ layer deposited within microcracks. Another potential mechanism will be proposed in the following subsection.

Figure 8 presents a comparison of the tensile loading curves up to fracture for both the as-purchased wires (four samples) and the ALD-coated wire under a load of 2.8 kg. No significant difference is observed between the stress–strain curve of the ALD-coated sample and those of the as-purchased wires. The incidental heat treatment during the ALD process is not substantial enough to induce a noticeable change in the ultimate elongation or ultimate tensile strength.

Alternative Mechanism for Improved Fatigue Resistance

During the ALD process, the loaded-wire samples, which contain a nonzero fraction of austenite and martensite, are exposed to an elevated temperature. This fulfills the conditions for low-temperature shape setting (LTSS) [40]. LTSS is associated with phenomena observed when a NiTi element containing oriented martensite is heated under external constraints that inhibit both strain recovery and transformation to austenite. Consequently, the reverse transformation occurs alongside dislocation slip [41]. As for our cold-drawn samples, the local plastic deformation due to LTSS that occurs in regions with the highest stress concentrations during the ALD treatment contributes to the alleviation of stress around microcracks [21]. During subsequent fatigue tests, when the wires are loaded, the stress concentrations are lower compared to wires without ALD treatment, which may contribute to the higher fatigue resistance of the ALD-treated wires.

4. Conclusions

Our preliminary study demonstrated that ALD-treated wires exhibit increased fatigue resistance. A comparison of six as-purchased NiTi wire samples with five ALD-coated samples revealed an average increase in fatigue cycles by a factor of 39, indicating superior fatigue resistance in the ALD-coated wires. This improvement in fatigue resistance might be attributed either to compressive stresses induced on the surface during the ALD process or to LTSS process during the ALD treatment. In future research, a thorough study is planned to determine the exact cause of the improved fatigue resistance of the ALD-treated wires. If the increased fatigue resistance is found to result from compressive stresses induced on the surface during the ALD process, this method could potentially be applied to thin wires made of other alloys or pure metals, offering enhanced fatigue resistance.