Research Paper
A statistical approach to understand the role of inclusions on the fatigue resistance of superelastic Nitinol wire and tubing

https://doi.org/10.1016/j.jmbbm.2015.07.003Get rights and content

Abstract

Superelastic wires and diamond-shaped stent surrogates were manufactured from Nitinol rods and tubing, respectively, from five different mill product suppliers – Standard VAR, Standard VIM, Standard VIM+VAR, Process-Optimized VIM+VAR, and High-Purity VAR. High-cycle fatigue tests up to 107 cycles were conducted under tension–tension conditions for wires and bending conditions for diamonds. These materials were compared under both testing methods at 37 °C with 6% prestrain and 3% mean strain (unloading plateau) with a range of alternating strains. The High-Purity VAR material outperformed all alloys tested with a measured 107-fatigue alternating strain limit of 0.32% for wire and 1.75% for diamonds. Process-Optimized VIM+VAR material was only slightly inferior to the High Purity VAR with a diamond alternating bending strain limit of 1.5%. These two “second generation” Nitinol alloys demonstrated approximately a 2× increase in 107-cycle fatigue strain limit compared to all of the Standard-grade Nitinol alloys (VAR, VIM, and VIM+VAR) that demonstrated virtually indistinguishable fatigue performance. This statistically-significant increase in fatigue resistance in the contemporary alloys is ascribed to smaller inclusions in the Process-Optimized VIM+VAR material, and both smaller and fewer inclusions in the High-Purity VAR Nitinol.

Introduction

The detriment of chemical impurities on the fatigue resistance of biomedical materials is well documented. For example, the high-purity composition of extra low interstitials (ELI) Ti–4Al–6V ELI covered by ASTM F136 has become the preferred version for fatigue- and fracture-critical medical implant applications (ASTM F136-13, 2013). Similarly, in recent years, the effects of impurities, primarily C and O, that may form non-metallic inclusions (NMIs), in biomedical-grade NiTi (Nitinol) have been studied. In particular, investigations have demonstrated fatigue crack nucleation in Nitinol from NMIs (and/or voids) in the simple raw material form (Wick et al., 2004, Schaffer and Plumley, 2009, Rahim et al., 2013), surrogate medical device implant fatigue specimens (Tolomeo et al., 2000, Pelton, 2011, Robertson and Ritchie, 2008, Gall et al., 2008, Lin et al., 2011, Lin et al., 2012, Pike et al., 2011, Robertson et al., 2012, Pelton et al., 2008), and even finished commercial devices (Pelton et al., 2008, Hull and Robertson, 2009). On a commercial scale, Nitinol is melted by vacuum arc remelting (VAR), vacuum induction melting (VIM) or a combination of the two. The purity of the raw materials, combined with these different melting techniques leads to variations in the size, abundance, and chemical composition of NMI׳s. Consequently, it is difficult to ascertain the most promising route for improvements in fatigue behavior of Nitinol medical devices without extensive studies.

The industry standard for acceptable NMI limits in Nitinol materials governs the maximum NMI length and area fraction (ASTM F2063-12, 2012). While, these are important parameters that influences fatigue resistance, there are other variables with potentially equal or greater importance. Specifically, to fully evaluate the fatigue resistance of Nitinol we subdivided the possible contributing variables into three distinct categories:

  • 1)

    Stress-concentrations: microscale phenomena that locally elevate the stress levels, thereby promoting the nucleation of a fatigue crack.

    • Lmax=maximum inclusion length in the drawing direction

    • Lmedian=median inclusion length in the drawing direction

    • Lmean=mean inclusion length in the drawing direction

    • G=grain size

    • a=surface defect size, e.g. scratch or gouge depth

  • 2)

    Probabilistic considerations: variables that increase the likelihood of stress concentration residing in a susceptible region of the device that would prematurely nucleate a fatigue crack.

    • A=total area fraction of inclusions measured in percent

    • ρ=density of inclusions measured in number of inclusions per unit area or unit volume

  • 3)

    Macro-mechanics: variables governing the overall stress or strain state of the specimen

    • Af=austenite finish temperature

    • εa=strain amplitude

    • εm=mean strain

    • σU=upper plateau stress

    • σL=lower plateau stress

    • σH=stress hysteresis (σUσL)

    • σY=yield stress, as defined by a 0.2% offset line measured from the martensitic elastic loading portion of the stress–strain diagram

Manufacturing the specimens to have similar mechanical properties (σU, σL, σH, σY), transformation temperature (Af), surface smoothness (a), and grain sizes (G) allowed us to investigate in isolation the influence of the inclusion size (Lmax, Lmedian, Lmean) and population distribution (A, ρ) on the fatigue resistance of Nitinol. However, it is difficult, if not impossible, to isolate the role of each of these variables to determine which, if any, is the dominant variable in improving fatigue resistance. Herein we present data comparing Nitinol with five different NMI-profiles to gain a deeper understanding of the role of each of these variables. Furthermore, this paper presents fatigue data on two forms of testing methodology: wire tension–tension fatigue and bending fatigue on laser-cut “diamond” surrogates from tubing. As such, this research provides the most comprehensive fatigue data on Nitinol with respect to composition and specimen form.

Section snippets

Materials

Five unique materials were examined in this study: Standard VAR, Standard VIM+VAR, Standard VIM, Process-Optimized VIM+VAR, and High-Purity VAR. The designation “Standard” indicates that the material underwent conventional melting techniques with normal purity input raw materials. Two new grades of materials have recently become commercially available that we designated “Process-Optimized” and “High-Purity”. The Process-Optimized material is manufactured using raw materials with purity levels

Wire

To isolate the role of inclusions, the round wire samples were tested in tension–tension fatigue under simple mode-I loading conditions. This method of loading imparts a uniform stress across the entire cross-section, thereby involving all inclusions within the gauge length rather than isolating only the near-surface inclusions or surface processing conditions that dominate rotary bend or stent-surrogate fatigue tests. Fatigue testing was performed using a 5-station Instron Model E3000 (each

Microstructural analysis

Statistical microstructural analysis was done on all three versions of the wire and five versions of the tubing. Here we concentrate on the microstructural analysis of the tubing, although similar trends were observed in the finished wire characterization. Fig. 5 shows representative micrographs of the finished tubes.

There exist several key differences among the five material types. Of the two Standard alloys, the VIM+VAR alloy has a smaller maximum inclusion length and greater inclusion area

Discussion

The trends in fatigue limit at 107 cycles for the various Nitinol compositions are comparable between the two testing methods, albeit with greater strain amplitude ranges for the diamond fatigue testing in bending. The resultant 107-cycle fatigue strain amplitudes for each material and test method are summarized in Table 3. The difference between uniaxial and bending fatigue behavior is well known and speaks to the difference in volume of stressed material with each cycle. Macherauch (2002)

Limitations

Herein we presented fundamental material data and drew comparisons to finished products manufactured from those differing Nitinol melting techniques. However, there are a few key limitations that must be considered before adopting the information.

  • 1)

    While the fatigue strain-life curves and consequent probability plots draw upon hundreds of individual data points, each set of specimens was manufactured from a single tubing lot (diamonds) or wire lot from a single melting campaign. Variations during

Conclusions

Each of the alloys investigated in this study has a different inclusion-profile. Therefore, when that knowledge is combined with the probabilistic fatigue evaluation it provides us valuable insight on the role of the inclusions on fatigue resistance of Nitinol.

  • 1)

    Both a low-inclusion-density, larger inclusion size (Std. VAR) and a high-inclusion-density, smaller inclusion size (Std. VIM+VAR, and Std. VIM) material have statistically indifferent medium- and high-cycle (>105) fatigue performance.

References (23)

  • ASTM F2082-06, Standard test method for determination of transformation temperature of nickel–titanium shape memory...
  • Cited by (56)

    • Effects of martensitic phase transformation on fatigue indicator parameters determined by a crystal plasticity model

      2023, International Journal of Fatigue
      Citation Excerpt :

      These alloys obtain superelasticity (i.e., large recoverable strain) via thermoelastic, stress-assisted, austenite-to-martensite transformations [17–19]. This behavior is particularly important for biomedical stents that are crimped to insert in an angioplasty tube (resulting in a large strain) and as the crimp is released the strain reduces and the material is cyclically loaded via the pulsing of blood vessels [12,21,25,38]. While this transformation is ideally recoverable and absent of any dislocation formation [19], this ideal transformation is rarely achieved.

    • Effect of microstructure on fatigue of superelastic NiTi wires

      2021, International Journal of Fatigue
      Citation Excerpt :

      These fractographic features are common for each sample, regardless of both the average grain size ranging from 20 nm to 250 nm and aging. As the stress-induced martensitic transformation in superelastically cycled NiTi is accompanied by plastic deformation processes [33,38–42], lattice defects generated upon cycling accumulate and gradually change the virgin microstructure. Whether it takes place and how much the microstructure changes depends on the virgin microstructure and test temperature [54].

    • Design and development of advanced SMA actuators

      2021, Shape Memory Alloy Engineering: For Aerospace, Structural, and Biomedical Applications
    View all citing articles on Scopus
    View full text