Kirkendall pore evolution during interdiffusion and homogenization of titanium-coated nickel microwires

Highlights

• NiTi microtubes can be synthesized by vapor deposition of Ti on Ni wires.

• A core/shell structure including intermetallic layers of Ni₃Ti, NiTi, NiTi₂ forms.

• Kirkendall pores merge near the center to form a hollow channel upon homogenization.

• Pore evolution was investigated during homogenization via in-situ X-ray tomography.

• Volume fraction and morphology of pores were studied as a function of wire diameter.

Abstract

In-situ and ex-situ X-ray 3D-tomography is used to characterize the microstructure of Ni microwires, with wire diameters spanning 25–100 μm, (i) after vapor-phase deposition of Ti onto their surface and (ii) after subsequent homogenization to achieve the near-equiatomic NiTi composition desired for shape-memory or superelastic behavior. After Ti deposition at 925 °C, wires are partially homogenized, exhibiting a pure Ni core surrounded by concentric shells of Ni₃Ti, NiTi and NiTi₂ intermetallic phases. Because of the imbalanced Ni and Ti diffusive fluxes, Kirkendall porosity is formed near the center of the wire, which often merges into a single pore in cross-sections, due to spatial confinement of the wire geometry. During subsequent homogenization at 925 °C, these Kirkendall pores grow due to further Ni-Ti interdiffusion, and they coalesce into a single, hollow channel near the central axis of the wire, thus forming a NiTi microtube. In some cases, off-center pores form in addition to the central pore, but these off-center pores do not form continuous channels.

Introduction

Nickel-titanium alloys with near equiatomic compositions display shape memory and superelastic behavior and are used in various applications ranging from biomedical implants to thermal actuators [[1], [2], [3], [4]]. Biomedical use of NiTi in the fabrication of endovascular stents takes advantage of the much lower stiffness in comparison to stainless steel in addition to the superelastic and shape memory behavior [5]. NiTi also exhibits high corrosion resistance by forming a passive oxide layer of titanium oxide, which decreases the risk of toxicity by the release of Ni ions in biomedical implants [6]. Beyond the unique properties of NiTi itself, the introduction of porosity provides further benefit for a variety of reasons. Porosity is desirable for implants as it decreases stiffness mismatch with the surrounding bone and for actuator applications as it increases the surface area, which enhances actuation time in response to a thermal stimulus [7]. Porous NiTi bone implants also facilitate osseointegration due to their internal pore structure similar to that of bones [[7], [8], [9], [10]].

To introduce porosity in NiTi, powder-metallurgy techniques incorporating space holders have been used [[11], [12], [13], [14], [15]]. Laser-based 3D printing techniques, such as selective laser melting, have also successfully produced porous NiTi parts [16]. However, most powder metallurgy techniques suffer from contamination from residual atmospheric oxygen and oxide layers, especially since these processes need to be performed at high temperatures, close to the NiTi melting point [[17], [18], [19], [20], [21]]. Additionally, the composition of pre-alloyed powder and the powder size distribution may influence the final properties of the alloy. An alternative method for creating 3D porous NiTi structures is to weave NiTi wires using a non-crimp 3D orthogonal weaving technique [22]. The feasibility of this technique was recently demonstrated for weaving Cu and Ni-20Cr wires [[23], [24], [25], [26], [27]]. However, it is difficult to use this technique for NiTi wires with <100 μm diameter, due to their very high yield stress (in the as-drawn state) and strong elastic and superelastic spring-back.

To address this problem, structures could be made of pure Ni wires, which are highly ductile and have low spring-back, followed by Ti vapor-phase deposition and diffusion/homogenization, to produce near-equiatomic NiTi. Titanium deposition can be conducted using pack cementation, which is a form of chemical vapor deposition. Pack cementation has been commercially used for several decades to provide protective diffusion coatings on material surfaces including, most notably, on turbine components [28]. As shown in Ref. [29], during pack titanization of Ni wires, a core/tri-shell structure forms, with Ni at the core and intermetallic layers of Ni₃Ti, NiTi and NiTi₂ as the shells, which, following homogenization, form a single-phase NiTi wire. However, as this process involves the imbalanced interdiffusion of Ni and Ti, Kirkendall pores form via the migration of vacancies in the direction opposite that of the faster diffusing species [30]. Because the diffusivity of Ni is higher than that of Ti [31,32], Kirkendall pores nucleate and grow in the central, Ni-rich regions of a Ni-Ti wire during interdiffusion. Due to the radial symmetry and spatial confinement of wires with <100 μm diameters, Kirkendall pores coalesce near the wire center, thus forming a hollow channel, i.e., a NiTi tube. This Kirkendall-based hollowing route has also been successfully demonstrated for the fabrication of hollow nano-spheres and cages [[33], [34], [35], [36]] and for Ni-Cr-Al wires, upon Al diffusion into Ni-Cr wires [27,37].

In our previous work, the formation of microtubes was observed in titanized and homogenized pure Ni wires with 50 μm initial diameter [29]. Shape memory and superelastic properties in the fully homogenized wires were identified using differential scanning calorimetry and dynamic mechanical analysis [38]. An investigation of the deposition kinetics of Ti on 25, 50 and 100 μm diameter Ni wires confirmed the presence of Kirkendall pores in the as-deposited, partially-interdiffused state, but with variations in pore structure in the different wire sizes [39]. The time required for Ti deposition to achieve near-equiatomic NiTi for these three diameters was measured as 0.5, 2 and 8 h, respectively. In the present investigation, we characterize the evolution of the Kirkendall porosity for wires with 25, 50, 75 and 100 μm diameters using in-situ X-ray tomography during homogenization to further elucidate the effect of diffusion distance on the pore morphology and volume fraction. Synchrotron X-ray tomography is a powerful non-destructive evaluation technique that allows for real-time observation of microstructural evolution within a specific region of a material, which is otherwise difficult, if not impossible, using other characterization techniques [[40], [41], [42]]. Results obtained in this study, along with our previous work [27,36,38], provide fundamental knowledge required to extend this work to 3D wire-woven structures, where the individual wires studied here represent the struts within such lattice geometries.