Research Summary The Active Metal Brazing of TZM·Mo and SiN Ceramics J.J. Stephens, P.T. Vianco, and F.M. Hosking Author's Note: Unless otherwise specified, all compositions are given in weight percent. The active metal brazing of ceramics holds the opportunity to design metal-ceramic braz- ing processes without conventional metalli- zation and nickel-plating steps. Because of their intermediate thermal coefficient of ex- pansion (TCE) properties, molybdenum and the TZM-Mo alloy (Mo-0.5Ti-O.08Zr) are attractive candidates for joining to silicon- nitride ceramics. The recently developed Fe- Ni-Co-based Thermo-Span™ alloy, with re- duced TCE from room temperature up to -400OC, is also an attractive candidate ma- terial for 7000C joint service applications. This article discusses the wetting, solid-state aging, and mechanical behavior of unalloyed molybdenum, TZM-Moalloy, Thermo-Span alloy, and silicon-nitride ceramic brazes made with three different active-metal braze al- loys. • Mo&TZM(Ar) 25 Mo& TZM (Vac) [] TZM (H2) !;] Si3N4(Ar) • SI3N4{Vac) SI3N4(H2) 'Au-Ni- ,V-Mo 830 850 1000 1033 1053 1000 1020 Temperature (0C) Figure 1. Wetting or contact angle as a func- tion of brazing temperature, atmosphere, sub- strate, and ABA filler metal. Note that argon atmosphere refers to a partial pressure of 750 mtorr argon. 54 INTRODUCTION Advanced turbine designs for the aerospace industry are greatly challenging existing joining technologies-the ability to fabricate these next-generation systems will require higher temperature joining capabilities. Since brazing is the most common method used to join metal and ceramic engine components, there is a critical need to develop enabling brazing materials and processes that will yield the kind of joints that can survive the anticipated service conditions (Le., exposure to higher joint tempera- tures for long periods of time in relatively corrosive and oxidizing environments). Currently, brazed engine hardware are typically exposed to peak joint temperatures of 400-S00°c. These temperatures are expected to rise to 700°C or greater; therefore, brazing alloys with higher melting points will be needed to satisfy the more extreme service conditions. Complicating the design and manufacturing problems is the general inability of standard filler metals to directly wet ceramic materials. Metallization of the ceramic is generally necessary prior to brazing to facilitate a chemical reaction between the ceramic and filler metal. More recently, active brazing alloys (ABAs) have been developed that directly bond to a ceramic. 1 This new class of alloys contains active constituents such as titanium, vanadium, or zirconium, which diffuse to the ceramic layer during brazing and form a strong metallurgical bond along the ceramic-braze interface. The most popular ABAs are based on silver or gold compositions. Another important feature of the brazed hardware is the thermal expansion mis- match between the metal and ceramic parts. This mismatch produces residual stresses in the braze joints that can lead to premature and potentially catastrophic failures. 2 • 3 The stress magnitude can be mitigated by compliant or graded interlayers. An intermediate thermal expansion material such as unalloyed molybdenum or TZM-Mo (Mo-O.5Ti-O.lZr) has been commonly used in the latter case, especially when elevated service temperatures are expected and minimum strength levels must be maintained. The ceramic-metal brazing process must be consequently compatible with the molyb- denum interlayer. This is true if the filler metal is a standard composition or an ABA. The study described here was driven by a need to develop brazed joints between silicon nitride and nickel-based alloys for eventual use in turbine engine applications. In particular, we examined the suitability of using a new controlled-expansion known as Thermo-Span™ (nominal composition: Fe-29Co-2SNi-S.5Cr- 4.8Nb-0.8Ti-0.5Al-0.3Si). Because of the large difference in the thermal coefficient of expansion (TCE) between nickel-based alloys and silicon nitride, it was deemed appropriate to examine the use of interlayers to help mitigate the stresses exerted on the ceramic. Due to its relatively low TCE and good availability as well as its resistance to recrystallization, the TZM-Mo alloy was identified as a potential candidate material. In actual practice, TZM-Mo would also have to be protected from the oxidizing turbine environment. Such oxidation protection, although extremely important, was deemed beyond the scope of this exploratory study. WETTING RESULTS The wetting results for the materials described in the sidebar are summarized in Figure 1. Contact angle data for Thermo-Span are not included in the plot, but they were excellent with contact angles of SO or less for the Ag-Cu-Ti and Au-Ni-Ti alloys . The Ag-36.2Cu-1.6Ti wetting results were generally very good with calculated angles being between SO and 10° on molybdenum and near 20° on the ground Si3N4 samples. With polishing, the GS-44 wetting angle was lowered to 10°. This moderate improve- ment in wetting does not necessarily justify the additional processing step, although the removal of surface cracks and chips could significantly impact braze joint strength. The Au-17.5Ni-l.5Ti and Au-lS.3Ni-2V-0.5Mo samples also exhibited good wet- ting. Contact angles were typically less than 2° on unalloyed molybdenum and TZM- Mo with the Au-Ni-Ti alloy and did not appear to be sensitive to the brazing temperature. The wetting angles were slightly higher for Au-Ni-V-Mo on TZM-Mo in dry hydrogen, although still relatively low-less than 10°. The GN-IO specimens yielded the highest contact angles for the Au-ABAs, especially when processed in vacuum- or partial-pressure argon. Lower Au-Ni-V-Mo/GN-IO contact angles com- parable to the unalloyed molybdenum, TZM-Mo, and Thermo-Span results were JOM • January 1996