The use of tungsten as a chronically implanted material

The electrochemical behaviour of W has a complicated pH dependence. Some authors have divided its electrochemical behaviour into five distinct pH regimes with different reaction mechanisms, few of which are relevant to the implant designer [21].

W forms a stable oxide layer of WO₃ in acidic solutions (pH  <  4) [22–31]—this is well outside the operating pH range of typical medical implants (except gastric implants).

In solutions of neutral and alkaline pH values, the W_xO_y surface species form soluble tungstate ions (often represented as 'orthotungstate' ${\rm WO}_{4}^{2-}$ [32]), which results in the continual dissolution of surface material; these are the products of W oxide dissolution as well as active W metal dissolution at higher pHs [23, 33–35]. All authors agree that in neutral to alkaline aqueous environments, the oxide phase on the metal surface hydrates, and then passes into solution as these hexavalent tungstate ions [18, 33, 34, 36–42]. Weidman's chronoamperometry measurements at 1.0 V sweeping from pH 0.5 to 13.0 presented sharp increases in current densities between pH 4–6, indicating that there is no passivation occurring via oxidation at higher pH values [22]. Heumann and Stolica [36] used weight loss measurements and coulometric methods to determine that the valence of dissolved W was indeed 6.0, agreeing with earlier work regarding W solubility.

According to Pourbaix, the borderline conditions of corrosion/passivation occur between pH 4–6 (figure 1). Above pH 6, general corrosion will occur, with hydrogen evolution in the absence of oxidising species, and without hydrogen evolution in their presence. The tendency for corrosion rises with increasing potential [19], as may occur when W is used as stimulating electrodes. Weidman reports that at current densities exceeding 0.1 mA cm⁻² there is a breakdown in the passivated surface regardless of pH, even in acidic zones [22]—a possible alteration to Pourbaix's initial diagram. This acid-soluble ${\rm WO}_{2}^{2-}$ ion has been proposed by others [43, 44].

Zoom In Zoom Out Reset image size

The original Pourbaix diagram applies to pure water at 25 °C, not the physiological environment, and thus should be used with caution. His analysis also ignores the numerous complexes W forms—these include hydrochloric complexes of trivalent W, hydrofluoric and oxalic complexes of tetravalent W, and cyanide complexes of pentavalent W. Some tungstates of alkali metals are soluble, while others are not [45].

A useful improvement to the diagram was generated computationally by Patrick for W in phosphate-buffered saline (PBS) using the CorrosionAnalyzer 1.3 software—approximately 60 solid, aqueous, and gaseous species were considered [13]. It is redrawn here for clarity with dominant species labelled (figure 2). Similar conclusions can be made: W is clearly a base metal, with a stable domain in the physiological pH range (x–y) (5.6 to 9.0 [14]) only outside of the water window (a–b).

Zoom In Zoom Out Reset image size

Heumann et al [18] still refer to tungsten's 'protective film' of WO₂—what may be termed a protective or passive oxide film may continually form at all pH values, inhibiting dissolution to some degree [46]. However, the long-term behaviour remains as continual dissolution into soluble hexavalent W⁶⁺. With this continual dissolution process in neutral solutions, W is unsuitable for use as a chronic implant material.

Lassner and Schubert [3] report corrosion of metallic W at the rate of 3.8 µg m⁻² h⁻¹ at 38 °C in distilled water. Gmelin [5, 7] reports corrosion rates of 0.83 µg m⁻² h⁻¹ in 3% NaCl solutions and 89.2 µg m⁻² h⁻¹ in 10% KOH solutions, both aerated and agitated at 20 °C. Of more significance, W exhibits a corrosive weakness to hydrogen peroxide [46]. Peroxide solutions dissolve W without inhibition, and the rate of dissolution increases linearly with H₂O₂ concentration [7, 47]. Hydrogen peroxide etches W along crystal planes (1 0 0), (1 1 1), and (1 1 2), and further at subgrains and wide-angle boundaries. It rapidly dissolves powders [3], as well as bulk billets of compact W, with dissolution rates of 1.6 mg l⁻¹ min⁻¹ reported in aqueous solutions of peroxide at a concentration of 14 g H₂O₂/l over 180 min at 20  ±  1 °C [48]. This peroxide effect has also been noted by Patrick et al [13] in a study on W microelectrodes, where an increase in corrosion rates on the range of two orders of magnitude in benchtop studies with 30 mM H₂O₂ in PBS versus no peroxide was reported.

While the relation between these experimental concentrations and physiological concentrations is debatable, it is worth mentioning given that peroxide is readily produced in the body's foreign-body immune response near implants. In addition to peroxides, reactive oxygen species such as the superoxide anion and the hydroxyl radical are rapidly released by immune cells including microglia in the central nervous system (CNS) [13, 49] and neutrophils in the blood [50] in events termed 'oxidative bursts' [51–53]. In non-acute animal studies, Prasad reports the presence of activated microglia near W electrode tracks in all cases, indicating a neuroinflammatory response regardless of post-implantation survival times and electrode performance [54]. These species would no doubt accelerate the corrosion of W through the oxidation to its soluble hexavalent form.

2.2.1. Chemical reactions.

The overall six-electron stoichiometric equation for neutral to basic solutions is as follows [20, 34]:

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {\rm W}+4{{{\rm H}}_{2}}{\rm O}\to {\rm WO}_{4\left( {\rm aq} \right)}^{2-}+~8{{{\rm H}}^{+}}+6{{e}^{-}}.\nonumber \end{align}$$

This overall anodic equation occurs step-wise as below, elucidating the rate-determining step, which is first-order with respect to OH⁻ [3]:

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {{{\rm W}}_{({\rm s})}}+2{\rm O}{{{\rm H}}^{-}}\to {\rm WO}_{\left( {\rm s} \right)}^{+}+3{{e}^{-}}+{{{\rm H}}_{2}}{\rm O}\nonumber \end{align} \tag{ 1 }$$

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {\rm WO}_{({\rm s})}^{+}+2{\rm O}{{{\rm H}}^{-}}\to {\rm W}{{{\rm O}}_{2\left( {\rm s} \right)}}+{{e}^{-}}+{{{\rm H}}_{2}}{\rm O}\nonumber \end{align} \tag{ 2 }$$

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {\rm W}{{{\rm O}}_{2\left( {\rm s} \right)}}+{\rm O}{{{\rm H}}^{-}}\to {\rm W}{{{\rm O}}_{3}}{{{\rm H}}_{\left( {\rm s} \right)}}+{{e}^{-}} ~ \left( {\rm rate}{\text -}{\rm determining}\,{\rm step} \right)\nonumber \end{align} \tag{ 3 }$$

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {\rm W}{{{\rm O}}_{3}}{{{\rm H}}_{\left( {\rm s} \right)}}+{\rm O}{{{\rm H}}^{-}}\to {\rm W}{{{\rm O}}_{3\left( {\rm s} \right)}}+{{e}^{-}}+{{{\rm H}}_{2}}{\rm O} ~ ({\rm fast})\nonumber \end{align} \tag{ 4 }$$

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {\rm W}{{{\rm O}}_{3\left( {\rm s} \right)}}+{\rm O}{{{\rm H}}^{-}}\to {\rm HWO}_{4\left( {\rm aq} \right)}^{-}\nonumber \end{align} \tag{ 5 }$$

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {\rm HWO}_{4\left( {\rm aq} \right)}^{-}\to {\rm WO}_{4\left( {\rm aq} \right)}^{2-}+{{{\rm H}}_{2}}{\rm O}{\rm .}\nonumber \end{align} \tag{ 6 }$$

The reduction of water/oxygen and of peroxide are the likely in vivo cathodic reactions [13]:

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {{{\rm O}}_{2}}+2{{{\rm H}}_{2}}{\rm O}+4{{e}^{-}}\to 4{\rm O}{{{\rm H}}^{-}}\nonumber \end{align} \tag{ 7 }$$

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {{{\rm H}}_{2}}{{{\rm O}}_{2}}+2{{{\rm H}}^{+}}+2{{e}^{-}}\to 2{{{\rm H}}_{2}}{\rm O}.\nonumber \end{align} \tag{ 8 }$$

Alternative anodic pathways are reported [22]. Older publications state W₂O₅OH_(s)  +  W₂O_5(s)  →  W₂O₅–OH–W₂O_5(s) as the rate-determining step [33], but this was superseded by newer data consistent with reaction (3) followed by the rapid OH⁻ ion-assisted dissolution of WO₃H_(s) to the ${\rm WO}_{4\left( {\rm aq} \right)}^{2-}$ species [46]. Works quoted by Gmelin state that the W₂O₅ referred to in older literature does not exist and in fact corresponds to W₂₀O₅₈ (or WO_2.9) [19]; this is supported by more recent literature. The above series of equations thus corresponds to the oxidative series of W: W  →  W³⁺  →  W⁴⁺  →  W⁵⁺  →  W⁶⁺.

Anodic polarisation of W in acidic media (or reduction from WO₃) will present distinct stages through the formation of all stable stoichiometric oxides, each with a distinct colour regime:

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle &{\rm W}\left( {\rm grey} \right)\to {\rm W}{{{\rm O}}_{2}}\left( {\rm brown} \right)\to {{{\rm W}}_{18}}{{{\rm O}}_{49}}\left( {\rm violet} \right)\nonumber \\ &\to {{{\rm W}}_{20}}{{{\rm O}}_{58}}\left( {\rm blue} \right)\to {\rm W}{{{\rm O}}_{3}}({\rm yellow}).\nonumber \end{align}$$

Given tungsten's particular weakness to peroxide dissolution, this chemical reaction pathway will be mentioned:

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {{{\rm W}}_{\left( {\rm s} \right)}}+2{{{\rm H}}_{2}}{{{\rm O}}_{2\left( {\rm aq} \right)}}\to {\rm W}{{{\rm O}}_{2\left( {\rm s} \right)}}+2{{{\rm H}}_{2}}{\rm O}.\nonumber \end{align} \tag{ 9 }$$

Two pathways have then been proposed, resulting in various compounds referred to as peroxotungstates, pertungstates, or pertungstic acid [3]:

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {\rm W}{{{\rm O}}_{2\left( {\rm s} \right)}}+2{{{\rm H}}_{2}}{{{\rm O}}_{2\left( {\rm aq} \right)}}\to {{{\rm H}}_{2}}{\rm W}{{{\rm O}}_{4}}\nonumber \end{align} \tag{ 10 }$$

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle {{{\rm H}}_{2}}{\rm W}{{{\rm O}}_{4}}+2{{{\rm H}}_{2}}{{{\rm O}}_{2\left( {\rm aq} \right)}}\to {{{\rm H}}_{2}}{{{\rm W}}_{3}}{{{\rm O}}_{12}}+4{{{\rm H}}_{2}}{\rm O,}\nonumber \end{align} \tag{ 11 }$$

and

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle 2{\rm W}{{{\rm O}}_{2\left( {\rm s} \right)}}+6{{{\rm H}}_{2}}{{{\rm O}}_{2\left( {\rm aq} \right)}}\to {{{\rm H}}_{2}}{{{\rm W}}_{2}}{{{\rm O}}_{11}}+5{{{\rm H}}_{2}}{\rm O}\nonumber \end{align} \tag{ 12 }$$

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle 3{{{\rm H}}_{2}}{{{\rm W}}_{2}}{{{\rm O}}_{11}}\to 2{{{\rm H}}_{2}}{{{\rm W}}_{3}}{{{\rm O}}_{12}}+{{{\rm H}}_{2}}{\rm O}.\nonumber \end{align} \tag{ 13 }$$

Pourbaix represents these reaction products as ${\rm WO}_{5}^{2-}$ ions; this is an uncommon usage [20].

The open-circuit (or E_corr corrosion potential) of W was found to be approximately  −100 mV at pH 7 with respect to the standard hydrogen electrode [42], and decreases by 43 mV per pH increase [18]; as the pH increases, the corrosion potential shifts to less noble potentials [23, 46]. In these studies, the anions of the buffer solutions (${\rm CO}_{3}^{2-}$, ${\rm HCO}_{3}^{-}$, ${\rm PO}_{4}^{3-}$, etc) were found to influence the electrochemistry. Kelsey [34] found the anodic Tafel slope at all concentrations of hydroxide to be 0.14 V decade⁻¹.

Readers with deeper interest in the electrochemical behaviour of W electrodes at anodic potentials, at wider pH ranges, and with results from cyclic voltammetry, linear sweep voltammetry, potentiodynamic polarization experiments, and analyses of the AC impedance spectra of W oxide films should consult the excellent papers by El-Wakkad [42], El-Basiouny [39, 42, 43, 55], Heumann and Stolica [18, 36], Wiedman [22], and Anik et al [21, 23, 30, 56, 57], amongst others.

Tungsten coils have been used to perform transcatheter embolisations of pathological blood vessels due to their high radiopacity (W behaving as a very effective x-ray absorber due to its high density [3]) as well as its thrombogenicity [58, 59]. The W 'Mechanically Detachable System' (MDS) coils (Balt Extrusion, Montmorency, France) are an example of a once commonly used system [16, 60–62].

Rabbit studies of W coils implanted for 3–6 months for embolisation exhibited corrosion, with black particles of metal spectroscopically determined to contain W and Fe found in the adventitia of the aneurysm and outside the lumen, distant from the W coils in phagocytic cells, and also in fibroblasts and spindle cells [63]. The iron is hypothesised by Peuster [62] to be not from coil manufacturing contamination but agglutinated blood on explantation. A similar rabbit study by Peuster et al [64] demonstrated a 3.5-fold increase in serum W concentration as early as 15 min post-implantation.

In 1998, a study by Weill et al [16] first presented the 'disturbing' finding of in vivo W corrosion in five patients with decreasing levels of radiopacity in skull plain x-ray films of MDS embolisation spirals 33–45 months post-implantation for intracranial aneurysm or dural fistula by venous approach. W blood serum levels in patients were 10–50 times the population baseline, proportional to the length of coil implanted. Additionally, elevated blood serum levels in one patient three months post-implantation (4.95 µg l⁻¹ versus baseline  <0.1 µg l⁻¹ from pooled population samples [16, 65]) without a demonstrably noticeable decrease in radiopacity indicates that corrosion occurs earlier than is radiographically detectable.

Barrett et al [66] demonstrated similar results for W MDS coils used for endovascular embolisation of varicoceles in the scrotum, presenting radiographic evidence of W resorption in 4 out of 18 patients (figure 3), and elevated W blood levels in 18 out of 19 patients (mean 40 months post-implantation). Bachthaler et al [60] reported W coil corrosion for intracerebral aneurysms and dural fistulae 26 months post-implantation in 9 out of 14 patients also by plain radiography, with two patients displaying complete coil resorption. W blood levels were elevated in 6 out of 14 patients, and urine levels were elevated in all. Others have reported similar radiographic and blood serum results with the use of W embolisation coils in intracranial aneurysms, abdominal aortic stent grafts, spermatic vein varioceles, oesophageal and gastric varices, and other pathological blood vessels (figures 4 and 5) [61, 67–70].

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In a metallographic analysis, Peuster's SEM examination of the W coils explanted from the rabbit subclavian arteries after four months demonstrated a clear change from a smooth surface of the control with a uniformly circular cross-section to surface roughening, pitting, and crevice corrosion in the implanted coils [64].

Various causes for this corrosion have been proposed in the literature. Weill et al [16, 63] blame the metallic impurity of the MDS coils. This was disproven by studies that spectroscopically verified the purity of the W pre-implantation (>99.9% by mass) [60–62, 71]. Barrett et al [66] attribute the degradation due to mechanical effects of percussive systolic–diastolic blood flow on the spirals, but this explanation was nullified when corrosion was observed by radiography in venous fistula drainage pathways with no percussive flow [16], as well as in cannulae of the spermatic vein [66]. Other cases have been reported [72].

While it may be possible that mechanical factors such as percussive flows may accelerate corrosion, as discussed at length previously, the underlying reason for W coil corrosion and eventual failure is known, i.e. the unfavourable thermodynamics of metal corrosion within the implant environment of the human body, i.e. the propensity for metallic W to dissolve into tungstate at pH 7.4 and 37 °C [19, 20]. A mass dissolution rate of 29 µg d⁻¹ was reported for W coils [73], but the proper rate relation to exposed surface area is not explored. Peuster [74] reports localised acidosis in occluded vessels due to metabolic processes and lack of oxygen carriers leading to stable passivating oxide layers on the W coil. However, reports of coil dissolution refute this—evidence of a local pH less than 4 (Pourbaix's upper limit for stable oxide formation) was not reported.

The general consensus of authors is that the use of W embolisation coils should cease [16, 60, 61, 63, 66–69, 74]. Recanalization as the failure mode of these W embolisation coils occurred in 58.3% of cases in [68], and was reported by others as proportional to the degree of corrosion [16]—an unacceptable risk for patients.

Long-term corrosion resistance is necessary for extracting stable signals from chronic implanted neural electrodes. Additionally, for single neuron recordings, electrodes must be less than 200 µm from the soma to record the action potential [75]—at such microscopic scales, even the slightest corrosion will cause device failure.

W microwires have been used in animals for this purpose for almost 60 years [54, 76–79]. High-purity wire is readily available, and the metal is stiff and robust enough to be readily fabricated into fine tips capable of performing single neuron electrophysiological recordings [76]. Commercial W microwires of 99.95% purity are available at various diameters (~50–125 µm). These can be sharpened to tips of ultramicroscopic dimensions (0.5–0.05 µm) by electrolytic etching, with oscillatory dipping into saturated KNO₂ or another etching solution while passing a current [80–83]. Tungsten's high melting point and low coefficient of thermal expansion enable it to be easily encapsulated within glass capillaries [84]; other encapsulants such as polyimide [82, 85, 86] and Parylene C have also been used. Exposed tips may be electroplated with gold, platinum, or other metals to decrease the electrode impedance [84, 87].

With the same technology, W has also been used to form recording electrode arrays [85]. These are termed linear microelectrode arrays, microelectrode arrays, microwire arrays, or floating microelectrode arrays. There are many examples of recording sites in commercial and non-commercial research utilising intracortical microelectrode arrays of W microwires [17, 85–91].

Unfortunately, in chronic applications, corrosion is again a limiting factor for their success as implants. Prasad et al [54] note that while W microwires are able to initially provide stable recordings, they fail to do so for years, reporting roughening of the surfaces of their 50 µm diameter microelectrodes two h after implantation, recession and cavitation of surfaces after two weeks to 16 d, and eventual cracking, pitting, hole-formation, and deep recession of the electrodes into the insulation in months following (figure 6). Williams [85] reports W microwire electrodes of 35 µm diameter (unsharpened) having viability of continuous recordings from guinea pig tests of only five weeks or less. Sanchez et al [87] reportes corrosion of W microwires after four weeks of in vivo implantation. Others report similar degradation of recording surfaces of W microwires [87].

Zoom In Zoom Out Reset image size

A very thorough analysis of this corrosion phenomenon was performed by Patrick et al [13] on commercial 50 µm diameter W microwires (California Fine Wire) encased in epoxy (Epotek 302-3M). Reported bench-top corrosion rates were 300–700 µm yr⁻¹ for pure W in 0.9% PBS, and 10 000–20 000 µm yr⁻¹ for Au-plated W (gold deposited circumferentially) in 0.9% PBS  +  30 mM H₂O₂ (the latter to simulate the highly oxidative environment of post-implantation inflammation). Assuming 100% W, ideal cylindrical geometry, full density, and ρ_w  =  19.25 g cm⁻³, this corresponds to 0.45–1.1 µg yr⁻¹ and 15–30 µg yr⁻¹, respectively. No data on pure W in PBS/H₂O₂ was offered.

Simultaneous in vivo studies in adult male Sprague–Dawley rats implanted cranially with Au-plated W reported corrosion rates of 100 µm yr⁻¹ (0.151 µg yr⁻¹) with the difference hypothesised to be due to the formation of an oxygen-blocking biofilm [13].

From purely visual analysis, in the PBS-only samples, the electrode surfaces polished to 1 µm smoothness displayed roughening after 1 d, pitting after 2 d, and full recession from the initial surface after 6 d to a depth of  >4 µm, similar to the reports by Prasad. In the in vivo studies, the W receded within its insulation 24  ±  8 µm after 87 d of implantation (figure 7). Similar results were reported by others (figure 8). No oxide layer was present in either case, contradicting claims of oxide passivation due to localised acidosis.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In cardiovascular medicine, permanently implanted bare-metal stents are used to restore luminal patency in patients with obstructive coronary lesions. A common stent platform alloy used is L605 cobalt–chromium alloy with 15% W wt% [93]. No corrosion or failure has been reported.

In prosthodontics, W is used as a component of cobalt–chromium alloys, e.g. Wirobond^® C (BEGO GmbH, Bremen, Germany) is 5% w/w, containing also Co (61%), Cr (26%), Mo (6%), and small amounts of Fe, Ce, and C [94]. No corrosion or failure has been reported.

In orthopaedics, cobalt-based wrought alloy ASTM F90 is used for total joint replacement; its composition is Co–20Cr–15W–10Ni. W composition is 14–16% (wt%) [95]. It is a popular orthopaedic alloy [96]. No corrosion or failure has been reported.

An exploratory search for 'TUNGSTEN' in the US Food and Drug Administration (FDA) 'Medical Device Recalls' database was hoped to reveal other implant failures. Three cases were found:

• 1.  
  A Class I recall in 2011 (Z-1865-2011) [133] of Axxent FlexiShield Mini, a radiotherapy beam-blocker device made of a circular silicone rubber pad containing tungsten particulate. The pad is 12.7 cm in diameter and 0.1 cm (1 mm) thick. Recall reason: '[The] product may shed particles identified as tungsten'.
• 2.  
  A Class II recall in 2015 (Z-0120-2016) [134] of Medtronic Pipeline Flex Embolisation Device. Made of 75% cobalt–chromium/25% platinum–tungsten. Recall reason: 'The firm is recalling Pipeline and Pipeline Flex Embolisation Devices from US since the devices were shipped with an EU version of the Instructions for Use'.
• 3.  
  A Class II recall in 2010 (Z-1760-2010) [135] of Boston Scientific Back-Up Meier Steerable Guidewires, made of a stainless-steel, PTFE-coated proximal shaft, and a distal tip of radio-opaque gold-plated tungsten spring coil wound around a stainless-steel inner core. Recall reason: 'Through their internal inspection process, they identified that the polytetrafluoroethylene (PTFE) coating on the gold plated distal coil of the Back-up Meier Steerable Guidewires of the identified lots/batches have the potential for PTFE delamination'.

Similarly, a search for 'TUNGSTEN' in the Australian Therapeutic Goods Authority (TGA) 'System for Australian Recall Actions' database reveals one case:

• 1.  
  A Class II recall in 2015 (RC-2015-RN-00263-1) [136] or AngioDynamics Morpheus Smart PICC CT (central venous catheters); tungsten content unclear. Recall reason: 'Based on a review of complaints, AngioDynamics has determined that the power injectable PICC catheters contained in these kits exhibit a higher than anticipated rate of extension tube fracture or leakage'.

Similar searches in the UK The Medicines and Healthcare Products Regulatory Agency (MHRA) 'Alerts and Recalls for Drugs and Medical Devices' database [137] and EU European Medicines Agency (EMA) 'Product defects and recalls' database revealed no similar results.

None were corrosive in nature.