Brazing and Diffusion Bonding TZM to Graphite Using Cu, Ti, Pd and Cr Filler Metals
M.H. Biglari, A. Miles, H. Schoonderwaldt and A.A. Kodentsov
Mat-Tech BV, Ekkersrijt 4302, 5692 DR, Son, The Netherlands
Abstract
It is demonstrated that joining of Graphite to Mo and TZM alloy can be accomplished below recrystallization temperature of the parent alloy by active brazing or diffusion bonding in vacuum. In the latter, joining in the solid state was conducted in the present work using well-defined interlayer materials.
Selection of the suitable filler metals and interlayer materials was based on several considerations including suitable melting range, mechanical properties, and availability in various wrought forms. Rationalization of the interfacial reactions occurring during joining was performed using available phase diagram data and information on diffusion mobility of species in relevant material systems.
Introduction
In many advanced applications, Graphite is to be joined to Molybdenum and its alloys (e.g., TZM). When such joints are intended for operation in harsh environment, like for example, a continuous operation at elevated temperature up to 900 ⁰C, an active brazing can still be an option.
In some applications, however, this type of bonded structures must also be able to withstand short exposures to significantly higher (up to 1400 ⁰C) temperatures, which makes selection of filler alloys (especially, those available in the form of foils and wires) for successful active brazing very problematic.
.
For example, it was proposed [1] to use for bonding Graphite to Molybdenum several Ti-based ternary alloys such as Ti-25Cr-21V, Ti-42Zr-15Ge and Ti-48Zr-5Nb with the brazing temperature ranges 1550-1650 ⁰C, 1300-1600 ⁰C and 1600-1700 ⁰C, respectively. Unfortunately, all these alloys are available only in a powder form, and for most of the brazing applications the use of powder fillers is not an option.
Another issue that might cause uneasiness here is that the brazing temperature of the suggested filler metals is well-above the recrystallization temperature of Molybdenum (and, in some cases, even of TZM-alloy) [2].
Perhaps, in this situation, utilization of solid-state (diffusion) bonding techniques is worthy of serious consideration. However, from the Mo-C phase diagram (Fig. 1) [3], it follows that direct diffusion bonding of Graphite to pure Mo (or TZM alloy) in the temperature range below re-crystallization temperature of the parent metal will, invariably, lead to the formation of the Molybdenum Carbide (α-Mo2C) at the Graphite contact interface. This orthorhombic carbide is very brittle, and because of strong diffusion anisotropy in the crystal lattice, it grows in the transition zone between Graphite and Molybdenum in a rather irregular fashion. Therefore, the quality of such joints is not expected to be high.
 |
Fig. 1: The binary phase diagram Mo-C.
There is, however, a way to obviate this problem. By judicious selection of an interlayer between Graphite and Molybdenum (or TZM) end-members, it is possible to create a situation when the brittle carbide α-Mo2C will not form at the Graphite contact surface at all. In this case, bonding across the Graphite contact surface will be provided by solid-state solubility (and diffusion) of Carbon in the interlayer material.
There is another option as well. It can be envisaged to use for the solid-state (diffusion) bonding of Graphite to TZM an interlayer material (transition metal) which possesses very high affinity to Carbon and forms at the Graphite interface metal carbide(s) with cubic, i.e., isotropic structure. This will result in a regular growth of the transition metal carbide, which can be beneficial to the structural integrity of the product joint.
The objective of the present work is to explore the above-mentioned options. To this end, brazing and diffusion experiments were conducted using different interlayer materials and strength of joints was assessed. Rationalization of the interfacial reactions occurring during joining was conducted using available phase diagram data and information on diffusion mobility of species in the relevant material systems.
Experimental Procedure
High-purity, high-density (< 10 % porosity) Graphite with grain size ~ 3 µm and pure Molybdenum (99.9) and TZM alloy (0.5. wt. % Ti; 0. 08 wt. % Zr; 0.02 wt. % C; Bal. Mo) were used as initial materials.
Foils (120 µm thick) of active Cu-based brazing alloy containing 2 wt. % Titanium were supplied by Degussa GmbH (at present, Evonik Degussa GmbH), Germany, and 50 µm thick foils of Palladium (99.95) and Titanium (99.99) were provided by Safina (Czech Republic) and Goodfellow (UK), respectively. Chromium-coated Pd pre-forms were also used as an interlayer between Graphite and TZM end-members. Electrolytic plating was performed at CZL Surface Treatments (The Netherlands).
Mating surfaces of the plan-parallel pieces of Graphite and Mo or TZM alloy were ground (to 4000 grid), cleaned in isopropanol (IPA) using ultrasonic agitation and dry in compressed Nitrogen.
Brazing pre-forms with the same dimensions were cut from the interlayer foils, ultrasonically clean in IPA, dry in compressed N2-gas and then, carefully positioned between the Graphite plate and pure Mo or TZM alloy end-members of the test sample. To prevent misalignment of the brazing assembly during processing a small external load was applied on the top of the test sample. In the case of diffusion bonding, application of bonding pressure was accomplished using a compression-transfer system constructed from two TZM-alloy plates with a set of two clamping screws and nuts made of TZM alloy.
Bonding experiments were conducted in vacuum. After bonding, test samples were subjected to visual inspection, and then, some joints were selected for mechanical testing. Assessment of the bonding strength was performed in a specially designed jig allowing for the test sample to be fractured under a well-defined value of torque. Other samples were cross-sections and after standard metallographic preparation were examined by optical microscopy, Scanning Electron Microscopy (SEM) and Electron Probe Microanalysis (EPMA).
Results and Discussion
Active Brazing of Graphite to Mo and TZM alloy Using Ti-activated Cu-based filler metal
Active brazing with Cu-based filler metal, containing 2 wt. % of Ti [3], can be a good option since the use of this solid-solution alloy allows for the Graphite/Mo-alloy joints to be operational up to 1000 ⁰C without risk of liquation in the brazing seam. (Note. Hereafter, this brazing filler will be referred to as CuTi2).
In Fig. 1 microstructure of the central area of the transition zone developed during brazing Graphite to Molybdenum at 1120 ⁰C for 30 min using CuTi2 filler alloy is given as an example. One can see that bonding across the reaction interfaces is microstructurally rather sound; no notable porosity can be found in the transition zone. It is important to notice that thickness of the brazing seam in the central part of the transition zone is significantly smaller that the thickness of the initial brazing CuTi2 alloy foil. Most likely, during vacuum brazing, some molten filler was squeezed to the sample edges.
 |
Fig. 2: Backscattered Electron Image (BEI) of the transition zone developed in central part of the bonded sample during vacuum brazing of Graphite to Mo using CuTi2-alloy foil (1120 ⁰C; 30 min). Note. The Graphite/TiC-interface is somewhat wavy.
It can also be seen that a continuous layer of the Titanium Carbide is formed during exposure to the brazing cycle along the entire Graphite-contact surface. The reaction product layer is somewhat wavy, which can be attributed to the preferential wetting of the Graphite by molten filler alloy along various defect (e.g., grain boundaries), that might be present in the initial material, but absolute proof for this phenomenon is still lacking.
Another interesting finding is to be mentioned here. Some Molybdenum (up to 1 at. %) was detected by EPMA in the Titanium Carbide layer, which implies a very fast transport of Mo across the layer of liquid Cu-based filler alloy at the brazing temperature.
It was repeatedly demonstrated [5] that morphology of the transition zone developed during interfacial reactions can be rationalized just using information about pertinent phase diagrams. According to the binary phase diagram shown in Fig. 3a [3], the CuTi2 alloy composition used in the present investigation is laying in the region of the Cu-based solid-solution. No intermediate phases exist in the binary Cu-Mo as well as in the Mo-Ti systems [3]. However, in the former, the terminal solid-state solubility of the components at brazing temperature is negligible (Fig. 3b), whereas, in the latter (Fig. 3c), the components are completely miscible at this temperature.
According to the equilibrium phase diagram, Titanium Carbide with a Face-Cantered Cubic (FCC) NaCl-type structure is the only thermodynamically stable compound (Fig. 3d) [6]. Clearly, the formation of the Titanium Carbide layer at the contact surface between the Graphite and molten CuTi2 alloy is a key factor responsible for bonding across the reaction interface.
|  |
| a | b |
 |  |
| c | d |
Fig. 3: Binary phase diagrams: a) Ti-Cu; b) Mo-Cu; c) Ti-Mo and d) Ti-C.
Mechanical testing of the selected Graphite/TZM joints fabricated by active brazing with CuTi2 filler showed that the product joint can withstand before fracture a maximum torque of about 4.3 Nm.
Diffusion Bonding of Graphite to Mo and TZM Alloy Using Pd-interlayer
Non-carbide-forming metal Palladium attracts our vivid attention due, primarily, to the notable solid-state solubility of Carbon in this “4f-transition metal”, which at the eutectic temperature amounts ~ 7.85 at. % (Fig. 4a) [3]. It is also important to notice that solid-state solubility of Mo in Pd is rather high (35-40 at. % in the temperature range above the stability of the MoPd3 phase), and terminal Mo-based solid-solution contains up to 5 at. % of Pd (Fig. 4b) [3].
 |  |
| a | b |
Fig. 4: Binary phase diagrams: a) Pd-c and b) Pd-Mo.
In Fig. 5, Back-scattered Electron Images (BEIs) of the reaction zone microstructure developed during diffusion bonding of Graphite to pure Mo via 50 µm Pd-foil (1180 ⁰C; 60 min) are shown.
 |  |
| a | b |
Fig. 5: Back-scattered Electron Images (BEIs) of the interaction zone developed in central part of the joint during diffusion bonding of Graphite to Mo using 50 µm Pd-interlayer at peak temperature 1180 ⁰C for 60 min: a) general view and b) magnified images showing morphology of the Molybdenum Carbide formed in the transition zone as a result of internal solid-state reaction.
One can see that no brittle product phases was found at the Graphite contact surface. Instead, the Molybdenum Carbide is formed inside the Pd-based interlayer. It can also be seen that some inclusions of Molybdenum Carbide are also present inside the layer of Pd-Mo solid-solution developed through interdiffusion during exposure of the test sample to the bonding cycle.
At this point, it is necessary to make some general comments about possible reaction mechanism governed diffusion bonding in this system. We recall that formation of a product AB from the reaction of solid A with solid B can, in principle, occurs in two ways [7]. The first way is for compound AB to nucleate at the A/B interface and form a product layer whereby A and B get separated. This occurs in bulk (semi-infinite) reaction couples. The second way is for the AB to precipitate inside a solvent (or matrix) and grow by diffusional addition of A and B to the precipitate. The latter type of reaction is called internal reaction. A change in temperature or pressure with subsequent oversaturation with respect to one of the components may cause this type of reaction inside an originally homogeneous phase.
Yet another type of internal reaction occurs when two components (in our case, Molybdenum and Carbon) counter diffuse in a “non-reactive” solvent phase (in our case, Palladium-interlayer), and compound AB (in our case, Molybdenum Carbide) will nucleate when the maximum solubility product is exceeded [8]. The position of first precipitation, growth of the nuclei and morphological evolution of a band of precipitates is considered in terms of parameters like interdiffusion coefficients, molar volumes, and activity products. Eventually, the array (band) of internal precipitates can agglomerate (through lateral growth) into a “dense” layer of the product phase.
As to the bonding strength, the best result in diffusion bonding of Graphite to TZM via 50 µm Pd-interlayer was achieved in the joint fabricated at peak temperature 1280 ⁰C for 60 min. In this case, the product joint was disintegrated when a torque of 7.4 Nm was applied. When peak bonding temperature was decreased to 1180 ⁰C, the product joints fabricated under these conditions were able to withstand before fracture a maximum torque of about 3.4 Nm.
Diffusion Bonding of Graphite to TZM Alloy Using Cr-plated Pd-interlayer
It is appeared that despite the unusually high among the transition metals [9], the solid-state solubility of Carbon in Palladium (interlayer) alone is not enough to provide a sufficient “affinity” to Carbon to facilitate adhesion across the Graphite/Pd-interlayer contact interface during fabrication (diffusion bonding) of the Graphite/Pd-(50 μm) foil /TZM-alloy joints. In contrast, interface between TZM alloy and Palladium (interlayer) in the test sample was found to be very sound, which should, in fact, be expected based on the available phase diagram information (Fig. 4b): the solid-state solubility of Mo in Pd is very high. To enhance affinity of the interlayer material towards Carbon (Graphite), a thin layer of Chromium was electrolytically deposited on the surface of the Pd-foil used as an interlayer for diffusion bonding of Graphite to TZM alloy.
According to the phase diagram shown in Fig. 5a [3], Chromium forms several stable carbides. It is also to be noted that in the binary Cr-Mo system there is a mutual solubility of the component in the solid-state above 880 ºC (Fig. 6b [3]), and terminal solubility of Chromium in Palladium is very high (Fig. 6c [3]). Furthermore, in the binary Pd-Cr system, no intermetallic phases are stable above 570 ºC.
 |  |
| a | b |
 | |
| c | |
Fig. 6: Binary phase diagrams: a) Cr-C; b) Pd-Cr and c) Cr-Mo.
It was found that solid-state bonding of Graphite to TZM alloy can be accomplished at 1180 ⁰C (for 4 hours) using single-side Cr-plated interlayer and these joints can withstand maximum applied torque up to 17.4 Nm. However, when peak bonding temperature was increased to 1280 ⁰C, the product joints fabricated under these conditions were able to withstand (before disintegration) a maximum torque up to 7.5 Nm.
It seems that the use of the Cr-plated, instead of pure, Pd-foil (50 µm) as an interlayer for diffusion bonding of Graphite to TZM alloy at 1180 ⁰C, allows to fabricate significantly stronger joints; the maximum torque required to disintegrate these bonded samples was found to be 17.4 Nm and 3.4 Nm, respectively.
On the contrary, the test samples fabricated at higher temperature, namely, 1280 ⁰C using Cr-plated Pd and/or pure Pd-foil as an interlayer were found to be fractured upon application of practically the same torque, i.e., 7.5 Nm and 7.4. Nm.
Diffusion Bonding of Graphite to Mo and TZM Alloy Using Ti-interlayer
Experimental results described in the previous sections gave indication that chemical affinity to Carbon provided by pure Palladium and Cr-plated Pd used as an interlayer in diffusion bonding of Graphite to TZM-alloy is still not sufficient for achieving good adhesion at the Graphite-contact surface. When, on the other hand, Titanium (50 µm foil), which possesses much high affinity to Carbon was used as interlayer material, the strength of the Graphite / TZM joints fabricated at 1180 ⁰C as well as at 1280 ⁰C turned out to be substantially higher, i.e., 22.8 Nm and 28 Nm, respectively.
In both cases, appearance of the fracture surfaces in the disintegrated samples is rather uniform (Fig. 7), and some Ti was found on the Graphite-side, while on the TZM-side, Carbon was detected by EPMA. These observations point out that, most likely, these joints were fractured across the Titanium Carbide layer formed at the Graphite-contact surface during diffusion bonding.
 |
Fig. 7: Fracture surfaces formed after mechanical testing in the Graphite/TZM joint fabricated by diffusion bonding in vacuum at peak temperature 1180 ⁰C for 240 min using 50 µm Ti-interlayer.
It is to be mentioned that in the binary Ti-Mo system, no intermediate phases are present. Instead, there is a complete mutual solubility of the component in the solid state above the α ® β transition in Titanium, i.e., above 882 ºC (Fig. 3c) [3]. On the contrary, in the system Ti-C, there is an intermediate phase, i.e., carbide with NaCl-type structure [6]. In fact, topology of the binary Ti-C phase diagram is dominated by the presence of this cubic compound (Fig. 3d).
The outstanding feature of this carbide is its considerable range of nonstoichiometry on the Carbon-sublattice. The progressive formation of (structural) Carbon-vacancies may affect the distribution of bond types, enhances Carbon diffusion, causes the onset of short-and long-range order, and, most probably, changes the deformation mechanism in the material.
Obviously, in the case at hand, the formation of the Titanium Carbide layer at the Graphite contact surface facilitates solid-state bonding across the reaction interface.
Conclusions
It was shown that Graphite can be join to Molybdenum and TZM alloy by means of active brazing using Ti-bearing solid-solution Cu-alloy CuTi2 or by solid-state (diffusion) bonding using different metallic interlayers. The latter technique can be utilized in the case of manufacturing joints intended for operation at temperatures higher than 1000 ⁰C.
As to the selection of the interlayer material, it should be stated that however elegant ideas in this regard might be, like for example, the utilization of internal solid-state reactions described here, the traditional approach based on the use of carbide-forming transition metal (e.g., Titanium) is appeared to be the most viable option. This can be appreciated from Table 1, where the results of mechanical testing of Graphite/TZM-alloy joints fabricated in the present work are summarized.
Table 1: Results of mechanical testing of Graphite/TZM joints
| Joining Method | Interlayer/Filler | Bonding Temperature, ⁰C | Maximum Torque Before Fracture, Nm |
| Diffusion Bonding | Pure Pd-foil, 50 µm | 1180 | 3.4 |
| 1280 | 7.4 | ||
| Diffusion Bonding | Cr-plated Pd-foil, 50 µm | 1180 | 17.4 |
| 1280 | 7.5 | ||
| Diffusion Bonding | Pure Ti-foil, 50 µm | 1180 | 22.8 |
| 1280 | 28.0 | ||
| Active Brazing | CuTi2 alloy | 1120 | 4.3 |
One might expect that in Graphite/TZM-alloy joints bonded via Ti-interlayer, fracture under applied torque occurs through the Titanium Carbide layer formed at the Graphite-contact surface.
References
[1] D.A. Canonico, N.C. Cole, G.M. Slaughter, Direct Brazing of Ceramics, Graphite, and Refractory Metals, Oak Ridge National Laboratory, ORNL/TM-5195, Contract No. W-7405-eng-26, Tennessee, 1976
[2] Refractory Metals for furnace construction, Metallwerk Plansee GmbH, Austria, 1985
[3] T.B. Massalski, Binary Alloy Phase Diagrams, ASM, Metal Park, Ohio, 1986
[4] Brazing Handbook, 4th Edition, American Welding Society, Florida, 1991
[5] A.A. Kodentsov, A. Paul, Diffusion Couple Technique: A Research Tool in Materials Science, in: A. Paul, S. Divinski (Eds.), Handbook of Solid-State Diffusion-Diffusion Fundamentals and Techniques, Vol 2, Elsevier, 2017, pp. 207-275
[6] J.L. Murray, Phase Diagrams of Binary Titanium Alloys, ASM, 1987
[7] H. Schmalzried, Chemical kinetics of solids, VCH Verlagsgesellschaft, Weinheim, Germany, 1995
[8] H. Schmalzried, M. Backhaus-Ricoult, Internal solid-state reactions, Prog. Sol. St. Chem., 22 (1993) 1-57
[9] E. Fromm, E. Gebhardt, Gase und Kohlenstoff in Metallen, Springer-Verlag, 1976 (in German)
