JOM, Vol. 65, No. 5, 2013
DOI: 10.1007/s11837-013-0582-7 Ó 2013 TMS
Latent Cracking of Tantalum-Titanium Welds Due to Hydrogen Embrittlement JOACHIM HOSSICK-SCHOTT,1,4 MARKUS REITERER,2 JASON HEFFELFINGER,1 MIKE HINTZ,1 MIKE RINGLE,1 IRYNA LEVINA,3 and KEVIN GAFFNEY1 1.—Medtronic Energy and Component Center, Medtronic, Inc, Brooklyn Center, MN 55430, USA. 2.—Corporate Core Technologies Department, Medtronic, Inc, Brooklyn Center, MN 55430, USA. 3.—Medtronic Cardiac Rhythm and Disease Management, Medtronic, Inc, Brooklyn Center, MN 55430, USA. 4.—e-mail:
[email protected]
Establishing electrical interconnects in implantable electronic medical devices frequently requires joining of dissimilar materials. A weld between a tantalum wire and titanium sheet metal on a contact module is presented as an example for dissimilar joining. Latent, brittle cracking was observed in the proximity of the weld upon pull testing. The weld cracking occurs by the mechanism known as hydrogen stress cracking (HSC) and is due to titanium hydride formation. Diffusion facilitated hydrogen transport into the weld area. Diffusing hydrogen accumulates preferably in regions of high stress, causing latent titanium hydride formation and embrittlement of the weld. A broad array of analytical tools such as scanning electron microscopy (SEM), transmission electron microscopy, electron backscattered diffraction, dynamic secondary ion mass spectroscopy, and nanoindentation were utilized to identify the root cause for HSC.
INTRODUCTION
MATERIALS BACKGROUND
Implantable medical device technologies often require the joining of dissimilar materials to electrically connect different sections of the therapy delivery circuitry. A robust interconnect design is a must in the implantable medical device field of use, specifically with respect to mechanical stresses that are invariably present. Reliable interconnect technology is ensured by extensive characterization and reliability testing. During reliability testing of a resistance-weld between a tantalum wire and titanium sheet metal on an interconnect module, latent, hydrogen-induced embrittlement of the weld was discovered. A redesign of the joint employing alternative materials resolved the potential issue and ensured that the observed embrittlement will not occur in the released Medtronic product (Minneapolis, MN). This article presents observations highlighting the sensitivity of Ta-Ti welds to dissolved hydrogen in the refractory metal.
The processing of tantalum can increase the concentration of hydrogen in the metal: For example, hydrogen may be created upon dissociative adsorption of ambient water molecules on the surface of the wire following thermal processing, or hydrogen may be absorbed during etching processes in acidic solutions. Once hydrogen has entered the metal, it will diffuse through the metal. Hydrogen diffusion in tantalum metal is documented to be very fast, even at room temperature, allowing hydrogen to travel several millimeters in a matter of days.1 Therefore, hydrogen can readily migrate to the junction between the tantalum wire and titanium sheet metal. The specific gravity of a-titanium is with 4.5 g/cm3, which is about three times lower than that of tantalum (16.6 g/cm3). Therefore, a low hydrogen concentration of, say, 10 ppm (lg/g) hydrogen by weight in tantalum translates into a concentration of 37 ppm (lg/g) by weight in the titanium joined to the tantalum. The arrangement
(Published online March 2, 2013)
625
626
Hossick-Schott, Reiterer, Heffelfinger, Hintz, Ringle, Levina, and Gaffney
of tantalum wire and titanium sheet metal is sketched in Fig. 1. Titanium is known to form brittle hydrides if hydrogen concentrations above the solubility limit are present, specifically under the influence of mechanical stress.2 Generally, exothermic absorbers tend to form hydrides, and both titanium and tantalum absorb hydrogen exothermically. It should be noted that post-weld annealing was not an option for this joint. Hence, the presence of mechanical stress must be assumed. Table I summarizes the materials properties of the tantalum wire and titanium sheet metal as used in the resistance weld. WELD FAILURE PHENOMENOLOGY Pull testing of the tantalum wire-titanium sheet metal joint performed immediately after welding resulted in very few failures of the weld interface, and the welds had acceptable strengths. However, many parts failed at the weld interface when pull testing was performed after the welds had been aged at ambient temperature for 10+ days. In the subsequent analysis, cross sections of the welds were prepared. Figure 2 shows scanning electron micrographs of such cross sections: The image in Fig. 2a was taken shortly after welding. The image in Fig. 2b was obtained from the same cross section about 10 days after the first image, without any further sample preparation. These tantalum to
Fig. 1. Sketch of the tantalum wire and titanium sheet arrangement. Ta wire, U = 0.4 mm is resistance welded onto titanium sheet with U = 0.18 mm in the location indicated. Sketch is not to scale.
titanium interface images were obtained using a backscattered electron detector. The differences in contrast reflect changes in density: Within a matter of days, and without applying external mechanical stress, the ‘‘dark zone’’ at the interface between the tantalum wire and the titanium sheet had grown by 15 lm–20 lm. In addition, cracks, that were evident but small immediately after cross sectioning the sample had grown appreciably. A hypothesis was developed: The growing ‘‘dark zone’’ at the interface between tantalum and titanium in Fig. 2 and the increase in pull test failures upon simply putting the welded parts on the shelf are related. It was suspected that hydrogen from the tantalum wire was accumulating in the weld zone, causing the apparent contrast in scanning electron microscopy (SEM) imaging, embrittlement, and ultimately brittle pull test failure. Consistent with this hypothesis is the fact that backscattering images reflect the density changes upon formation of titanium hydride. In what follows, the hypothesis will be systematically proven out. Designed Experiments: Turning the Effects of Hydrogen On and Off To further test the hydrogen hypothesis, tantalum wires were deliberately loaded with hydrogen by immersion in 25% hydrofluoric acid for 10 min. To ensure similar surface conditions for high- and low-hydrogen-content wires, low-hydrogen-content wires were produced from hydrogen charged wires by vacuum annealing at 800°C. This annealing process was demonstrated to drive the hydrogen out of tantalum: After annealing, the H concentration was as low as 2 ppm (lg/g). The hydrogen loading for tantalum wires in HF solution was found to be approximately linear with time and therefore readily controllable. In this manner, several couples of high and low H content tantalum wires were produced. Welds performed with hydrogen charged tantalum wires (high H) showed a much enlarged
Table I. Properties of the tantalum lead wire and the titanium sheet metal used in the resistance weld Property
Ti sheet (alpha titanium)
Ta wire
ASTM specification number Hydrogen content as measured in LECO gas analysis, ppm (lg/g) Hydrogen permitted as per ASTM standard, ppm (lg/g) Hydrogen diffusion coefficient(s), room temperature, T = 300 K, cm2/s Maximum concentration, H/metal atom ratio Maximum Concentration, at.% Maximum concentration, ppm Hydrogen heat of solution, DH kJ/mol, (eV)
ASTM B265-11e1 10–20 150 3 9 10 5, Malyshev et al.3à 2 9 10 7, Brauer et al.4 1.75 (hcp), Kolachev6 63.6 35176 52 ( 0.55), Wenzl7
B365-98 BO5400 2–100 15 4 9 10 6, Qi et al.5 0.76, Kolachev6 43.2 4181 38 ( 0.39), Wenzl7
The maximum hydrogen concentrations are stated as bulk concentrations reached under 1 atm of H2 gas pressure at room temperature Deliberate H loadingàThe hydrogen diffusion coefficient in titanium measured by Malyshev et al.3 is corrected for higher H concentrations including titanium hydride formation which tend to slow down diffusion. The value published by Brauer et al.4 is uncorrected.
Latent Cracking of Tantalum-Titanium Welds Due to Hydrogen Embrittlement
(a) Day 1
627
(b) Day 10
Ti
“dark zone”
“dark zone”
Ti
Ta 50 µm
50 µm
Ta
Fig. 2. Signature of a weld failure. Both scanning electron microscopy (SEM) images show a cross section of the same area of a tantalum-wireto-titanium-sheet weld. The Ta wire cross section appears white and the titanium sheet is gray. (b) Recorded 10 days after the image in (a). Within a matter of days, and without applying external mechanical stress, the ‘‘dark zone’’ at the interface between the tantalum wire and the titanium sheet grew by 15 lm to 20 lm. In addition, the cracks grew appreciably.
Table II. Hydrogen amounts as detected by dSIMS
Location of dSIMS analysis
Measured relative H concentration (counts/s)
Approximate absolute h concentration, based on absolute values in Table I (ppm, lg/g)
200,000 300
(200,000/300) 9 10 6500 10
Dark zone, Ta-Ti interface Background, bulk Ti
Fig. 3. (a) Thick dark zone formed at the interface of a hydrogen-enriched tantalum wire welded to titanium sheet. The Ta wire was loaded with approximately 70 ppm to 100 ppm of H. (b) Very thin dark zone formed in a weld using a wire first loaded with approximately 70 ppm to 100 ppm of H, then vacuum annealed, leaving less than 5 ppm of H loading. For both conditions, the cross sectioning and imaging was performed 5 days after the weld.
dark zone compared to welds performed with charged and subsequently vacuum-annealed wires (low H), thus further corroborating the above hypothesis. Transmission Electron Microscopy Scanning transmission electron microscopy (STEM) was used as a tool to confirm the titanium hydride crystal lattice parameters. The work was performed by EVANS Analytical Group (Sunnyvale, CA).9 The samples were prepared using the in situ focused ion beam (FIB) lift out technique. The samples were imaged with a FEI Tecnai TF-20 fieldemission gun/transmission electron microscopy
(TEM) (FEI Company, Hillsboro, OR) operated at 200 kV in bright-field mode, high-resolution mode, and high-angle annular dark-field STEM mode. The key finding was that the selected-area diffraction (SAD) patterns in the dark zone at the interface of the weld between tantalum wire and titanium sheet are consistent with titanium hydride. Lattice spac˚ and 1.59 A ˚ were identified, which are ings of 1.36 A deemed typical for titanium hydride. Hardness Measurements Hardness mapping also supports the hypothesis that hydrogen embrittlement causes the tantalum wire to titanium sheet weld failures: Clearly, the
628
Hossick-Schott, Reiterer, Heffelfinger, Hintz, Ringle, Levina, and Gaffney
dark zone represents the hardest area of the weld with hardness values measured between 4 GPa and 5 GPa. These values are in agreement with the measurements made on titanium hydride by Xua et al.8 This increase in hardness at the weld interface alone could be due to a number of other reasons, for example the formation of a titaniumoxygen phase known as alpha-case. However, in the context of the results documented above and below, alpha-case formation does not explain the growth of the dark zone over a few days because at room temperature, the oxygen diffusion coefficients in titanium are orders of magnitude below the hydrogen diffusion coefficients. Oxygen diffusion and growth of the alpha case phase therefore do not agree with the observed growth rates of the dark zone. However, the existing alpha case quite likely promotes hydrogen embrittlement as it may promote segregation of high hydrogen concentrations and hydride formation.
Fig. 4. Hardness map of the dark zone. The dark zone represents the hardest area of the weld with values around 4 GPa–5 GPa in agreement with measurements made on titanium hydride by Xua et al.8
Dynamic Secondary Ion Mass Spectroscopy (dSIMS) dSIMS measurements on tantalum wire to titanium sheet welds were performed at EVAS Analytical.9 Table II shows the results of the dSIMS measurements. In the dark zone at the Ti-Ta interface, the hydrogen concentration increases dramatically. The approximate, absolute hydrogen concentration in the dark zone can be estimated from the relative dSIMS counts per second: The background concentration in the titanium sheet is 10 ppm. In the dSIMS measurement, this value corresponds to 300 counts/s. In the dark zone, 200,000 counts/s of H signal were measured. Hence, the hydrogen concentration in the dark zone of the weld is increased by orders of magnitude. A factor of 650 over the corresponding background concentration maybe calculated assuming linearity. Based on this value, the stoichiometry of the TiHx compound is calculated to be x = 0.3 which may appear somewhat low as stoichiometry values as high as 1.756 and higher10 have been reported (see also Table I). However,
Fig. 5. SEM image of a weld cross section showing the salient signatures of TiHx crystals in the ‘‘dark zone’’ of the tantalum-titanium interface.
Fig. 6. EBSD map of the dark zone at the tantalum-titanium interface. Bright, needle-shaped regions indicate presence of fcc crystal structure indicative of titanium hydride.
Latent Cracking of Tantalum-Titanium Welds Due to Hydrogen Embrittlement
remembering the fact that the dark zone grows in the form of needles that extend out into the bulk material (see Figs. 3 and 4), it is easy to imagine that TiHx needles with higher x, i.e., (1.5 < x < 2), alternate with comparatively pure Ti material. The dSIMS measurement reports an average of the H-rich and Hpoor Ti areas. Hence, it is not surprising that the dSIMS measurements do not show a nominal stoichiometry (e.g., TiH or TiH2), but it is clear that the dark zone is H-enriched by several orders of magnitude over the background concentration. Weld Microstructure in ElectronBackscattered Diffraction (EBSD) Higher-resolution SEM images of weld cross sections were performed of the ‘‘dark zone’’ at the interface of the tantalum and the titanium sheet. Figure 5 reveals the needle-like appearance of TiHx crystals. Furthermore, EBSD analysis performed on the cross-sections of welds revealed the presence of face-centered cubic (fcc) crystals (Fig. 6). At room temperature, titanium hydride forms a phase known as d-phase, which has an fcc lattice.11 Thus, microstructure analysis provides further support for the identification of titanium hydrides in the welds. SUMMARY Hydrogen embrittlement in the form of hydrogenassisted cracking caused by titanium hydride formation was found to be responsible for weld failures observed in tantalum-wire-to-titanium-sheet welds. The hypothesis of hydride formation in tantalumwire-to-titanium-welds is supported as follows: – SEM identified a growing, crack-bearing zone in cross sections of the tantalum wire to titanium sheet welds at the tantalum-titanium interface.
629
– Designed experiments could turn the instability on and off in dependence of the hydrogen loading of the tantalum wire. – TEM identified lattice parameters characteristic for titanium hydride formation. – Microhardness measurements identified an increase in hardness expected for titanium hydride formation. – dSIMS determined the hydrogen concentrations in the dark zone of the tantalum wire-titanium weld interface to be orders of magnitude greater than in the surrounding titanium sheet metal. Such high concentrations are indicative of hydride formation. – EBSD identified the salient fcc lattice features of titanium hydride formation in the dark zone of the tantalum wire-titanium interface in the weld zone. REFERENCES 1. Y. Fukai, The Metal-Hydrogen System (New York: Springer, 2005), p. 303. 2. J.L. Waisman, R. Toosky, and G. Sines, Metall. Trans. A 8A, 1249 (1977). 3. L.G. Malyshev, S.A. Lebedev, R.A. Ryabov, P.V. Gel’d, and S.A. Kuznetsov, Russ. Phys. J. 25, 232 (1982). 4. E. Brauer, R. Doerr, and H. Zuechner, Z Phys. Chem. Neue. Fol., 109 (1976). 5. Z. Qi, J. Voelkl, R. Laesser, and H. Wenzl, J. Phys. F: Met. Phys. 13, 2053 (1983). 6. B.A. Kolachev, Hydrogen Embrittlement of Non-Ferrous Metals (Jerusalem: Israel Program for Scientific Translations, 1968). 7. H. Wenzl, Int. Met. Rev 27, 140 (1982). 8. J.J. Xua, H.Y. Cheung, and S.Q. Shi, J. Alloy Compd. 436, 82 (2007). 9. Evans Analytical, www.eaglabs.com/mc. 10. B. Kappesser and H. Wipf, J. Phys. IV 111, 73 (1996). 11. J. Zhao, H. Ding, W. Zhao, X. Tian, H. Hou, and Y. Wang, Trans. Nonferr. Met. Soc. 18, 506 (2008).