Aluminium Oxide Ceramics is considered as an attractive material in designing of various components when “thermal management issues” are of concern, and Carbon Dioxide is used as a coolant.
In our particular case, a specification of the industrial problem can be outlined as following. A system of rectangular grooves (with dimensions of an order of 1 mm and distance in between of about 1 mm) are machined at the surface of the alumina plate, and then two of such “surface patterned” ceramic plates is bonded (“face-to-face”) together. As a result, a system of “channels” inside a (bulk) ceramic substrate will be created.
At this stage of the project, the use of active brazing as a bonding technique in manufacturing “CO2-cooled substrates” was explored. In the present MEMO we will discuss results of the brazing experiment involved flat Al2O3-ceramics plates and CB4 (AgCu26.5Ti3) “active” filler metal.
“Active” brazing of Al2O3-ceramics (plates 20 ´ 20 ´ 2 mm) was carried out in vacuum furnace using 200 mm foil of the CB4 (AgCu26.5Ti3) filler alloy. A “small load” was applied upon the brazing assembly as shown in Fig. 1a.
Cross-section of the product joint was made by direct grinding of the brazement, and after standard metallographic preparation sample was examined using optical microscopy.
To improve electrical conductivity of the sample and make it suitable for electron-microscopic investigations, a thin layer of carbon was deposited in vacuum on the metallographically prepared surface. Further analysis of the sample was conducted with Scanning Electron Microscopy (SEM) and Electron Probe Microanalysis (EPMA).
Results and Discussion
As expected, Al2O3-ceramics can be directly joined (without any prior metallization) by means of the “active brazing” (Fig. 1). In this work, “Titanium-activated” filler metal CB4 (AgCu26.5Ti3) was employed. Titanium is particularly effective at wetting various oxides, including alumina, and such behaviour is attributed to the high thermodynamic stability of titanium oxides.
Fig. 1: Optical Images of the Al2O3-ceramicparts after brazing using CB4 (AgCu26.5Ti3) “active” filler metal (Sample MT 12-181). Note some “dark coloration” of the filler metal at the “edges” of the brazing seam.
A visual inspection of the brazed sample gave every reason to believe that the joint is sound. Also, the formation of the “fillet” during the brazing process was observed, which underlines good wetting of the ceramic surfaces by brazing alloy. Further examination of the brazement cross-sections by optical microscopy indeed confirmed good quality (“soundness”) of the brazing seam (Fig. 2). The puzzling question that still remains at this moment is a clear visible “dark coloration” of the filler metal at the “edges” of the brazing seam.
Fig. 2: General view of the cross-section (optical image) of the Al2O3– ceramics joint brazed in vacuum using CB4 (AgCu26.5Ti3) “active” filler metal. Sample MT 12-181. Note the formation of the “fillet” during the brazing process.
In Fig. 3 a representative microstructure of the brazing seam is given. These images were taken somewhere in the central part of the joint. No apparent defects (like voids, cracks, etc.) were detected in the transition zone. Instead, continues product phase layer is clear visible along whole Alumina/metal interface.
Fig. 3: Typical microstructure of the brazing seam in the Al2O3 / Al2O3 joint shown in Fig. 2: a) optical micrograph (bright-field image) and b) Back-scattered Electron Image (BEI). The image is taken from the central part of the brazement cross-section. Note that metal/ceramics interfaces are free of any apparent defects.
Somewhat similar morphology was also developed in the “fillet” that formed upon the brazing cycle around the “triple point”. This can be appreciated from the back-scattered electron image shown in Fig. 4a.
Electron microprobe analysis of the product layer adjacent to the ceramic interface indicates that this phase contains Ti, Cu and Al, with Titanium and Copper being the main constituents (Fig. 5). It is to be emphasized that Oxygen was not detected by microanalysis of the Ti-Cu(Al)-based layer because the EDX detector used was a Be-window type. However, earlier research on active brazing of Al2O3-ceramics, found in the literature, provides clear-cut evidence that this phase also contains some Oxygen. It was reported that this phase has similar crystal structure as Ti3Cu3O which is the same as that of many other transition-metal borides, carbides, nitrides, and oxides that, collectively, are known as h-phases.
Close inspection of the brazing seam in the vicinity of the filler metal/ceramics interface revealed that the product morphology apart from the main constituent of the eutectic microstructure, viz. Silver- and Copper-based (Agss and Cuss) solid solutions, a numerous (often faceted) inclusions of the h-phase (indicated by arrows in Fig. 4b) are present inside the filler metal. One can notice (Fig. 5) that the Al-content in the h-phase is relatively high, which
Fig. 4: Back-scattered Electron Images of the Al2O3-ceramics joint (sample MT 12-181) showing: a) morphology of the “fillet” formed around the “triple point” and b) microstructure of the reaction zone developed between Alumina and “active” filler metal CB4 (AgCu26.5Ti3) during brazing in vacuum. (The Ag- (“white” contrast) and Cu-(“grey” contrast) based solid solutions present within the microstructure are denoted as Agss and Cuss, respectively.) Note that numerous inclusions of the h-phase (indicated by arrows) are also present inside the brazing seam far away from the ceramics/metal interface.
confirms that reduction of the Al2O3 surface occurred during the brazing process because the braze alloy originally contained no Al.
It is important to mentioned that in the early TEM studies of the alumina joints brazed with active Ag-Cu-Ti alloys a very thin (~0.1 mm) Ti-rich layer was detected between Al2O3 and h-phase. This phase was identified as the high-temperature (FCC) modification of the
Titanium monoxide, g-TiO. (Obviously, in our investigation the presents of the g-TiO-interlayer cannot be confirmed by the experimental technique available.)
Fig. 5: a) Characteristic X-ray spectrum taken from the continuous product layer (h-phase) adjacent to the Alumina substrate (Fig. 4b) together with b) the results of the (standardless) quantification. Note that Oxygen was not detected in this layer because the EDS detector used has a Be-window.
It is interesting that under equilibrium conditions, the g-TiO structure is unstable below 1250 °C, undergoing an ordering transition below that temperature. The presence of g-TiO in the Al2O3 brazed joints suggests that its stability may have been influenced by the complexity of the chemical or mechanical “environment” in the reaction zone, and that it exists in metastable equilibrium in the microstructure.
1). Though wetting is essential for bonding, it does not ensure that the adhesion of metal alloy to ceramic surfaces will be good or that strong braze joints will be produced. An illustration of this point can be found in a seminal work of Prof. Yu. Naidich. He has shown that Al2O3 can be wet by Ti-containing melts of both Cu and Au, but that adhesion of the Cu-Ti alloys to the oxide is much better than that for the Au-Ti alloys. The difference in adhesion properties has been attributed to a TiO reaction layer that formed in contact with Al2O3 in the case of the Cu-Ti melts. The oxide Ti2O3 forms for the Au-Ti melts under the same conditions. It has been concluded that adhesion is better for the Cu-Ti alloys because the metallic nature of the TiO reaction layer results in “a more metal-like” transition between the Al2O3 and alloys.
This suggestion about the importance of metallic oxide phase similar to TiO undoubtedly has merit, but is an oversimplification because it does not account for other important factors such as thermal expansion mismatch strains, which can also affect adhesion.
If, however, these arguments have any validity, then the formation of reaction layer of the Ti3Cu3O-type (h-phase) is equally important for high adhesion and strong bonding in the Al2O3/AgCuTi-brazing alloy system. Little is known about the properties of the h-phases. However, because they are characterized by continuous metal lattices containing interstitial oxygen, these phases are likely to be metallic in nature. The formation of h-phase reaction product in conjunction with a TiO-layer then can provide a more gradual transition in chemical bonding between the Al2O3 and metallic phases of the Ag-Cu based filler alloy than TiO will alone. The layer of the h-phase may also provide a more gradual transition in physical properties and help to minimize the effect local strains that develop from thermal expansion coefficient mismatches can have on adhesion.
2). The use of the CB4 (AgCu26.5Ti3) brazing alloy might be, in principle, a solution, although other AgCuTi-based filler metals, like for example, those containing Indium and/or Tin should also be taken into further consideration. However, it is vitally important that any active filler metals selected for the brazing of alumina is based of the (near-) eutectic Ag-Cu alloy.
From a practical point, the following question will certainly cause uneasiness here. “In what way should an “active filler metal” be (pre-) placed in the brazing assembly in order to keep a system of “channels” inside a (bulk) ceramic substrate open?” It is important to realize that upon brazing “active” filler metal will wet (indiscriminately) all ceramic surfaces including those of the rectangular grooves machined at the surface of the alumina substrates.