Surface preparation for high purity alumina ceramics enabling direct brazing in hydrogen atmospheres

[0003] This application is related to co-pending application, “Coating System to Permit Direct Brazing of Ceramics”, by Charles H. Cadden and F. Michael Hosking, both of Albuquerque, N.Mex.; commonly assigned to Sandia Corporation, Albuquerque, N.Mex.

[0004] This invention relates generally to the field of brazing, and more specifically to a method for preparing the surface of a high purity alumina ceramic or sapphire specimen that enables direct brazing in a hydrogen atmosphere using an active braze alloy. The present invention also relates to a method for directly brazing a high purity alumina ceramic or sapphire specimen to a ceramic or metal member using this method of surface preparation, and to articles produced by using this brazing method. In this context “brazing” is defined as the joining of two or more members using a filler metal that is heated above it's liquidus temperature (e.g. melted) in a reduced-oxygen atmosphere furnace at a temperature greater than 450 C.

[0005] In U.S. Pat. No. 4,542,073, Tanaka discloses a method for bonding a ceramic first member to a second member made of ceramic or metal, where the method comprises two sequential thermal treatments. The first thermal treatment is performed by heating the ceramic first member in an oxidizing atmosphere from 1200 C to 1400 C, whereby an oxide surface layer is formed, only if the bulk ceramic contains an oxide-forming element, such as silicon (e.g. Silicon Carbide, Silicon Nitride, or Sialon). Next, a metal sheet (e.g. copper) is placed in-between the two members and a second thermal treatment is performed in a nitrogen atmosphere, whereby the oxidized ceramic surface reacts with the metal sheet to form a chemical bond (e.g. copper oxide). Tanaka teaches that during the second thermal treatment step the peak temperature is not higher than the melting point of the metal filler sheet. Tanaka's limitation precludes the possibility of brazing the two members together, because brazing requires melting of the filler metal, as is practiced by the present invention.

[0006] Traditionally, the brazing of ceramic components requires costly surface preparation steps to ensure adequate bonding to the non-metallic substrate, especially when joining a ceramic to a metal (e.g. Kovar, copper, or steel). The conventional preparation process involves formation of a “metallization layer”, typically a mixture of molybdenum and manganese powders fired at a high temperature (>1400 C). Sintering of this mixture produces a well-bonded, glassy coating having a thickness of about 10-40 microns. This is then followed by a nickel-plating step (e.g. electroplating, electroless coating, or nickel oxide painting), following optionally by a sintering step. This provides a continuous metallic surface of about 2-10 microns thick that can be wet by conventional, non-active brazes. Commonly referred to as the “Mo—Mn—Ni” or the “moly-manganese” metallization process, these costly and environmentally unfriendly manufacturing steps require close process control and inspection.

[0007] Active braze alloys (ABA's) were developed as an alternative means to directly wet a ceramic, without the necessity for prior metallization of its surface. The ABA's ability to perform successfully as a direct braze depends upon the chemical reaction(s) that occur between the active element(s) in the braze and the ceramic surface. Early active alloy systems primarily utilized titanium (Ti) as the active component, although Zr, Cr, V, Hf, Ta, and Nb have also been used. Small additions of titanium to common face centered cubic (FCC) filler metals (e.g. Cu, Ni, Ag, Au, and their alloys) have been shown to facilitate wetting on alumina through the formation of a titanium-oxide chemical reaction layer (e.g. TiO and Ti₂O₃). Unfortunately, Ti-ABA's (e.g. TiCuSil or CuSil-ABA made by Wesgo, Inc., San Carlos, Calif.) are limited to brazing in either vacuum or inert atmospheres, since titanium forms undesirable hydrides in a reactive hydrogen atmosphere.

[0008] Recently, a new family of braze alloys has been developed that are more compatible with brazing in a hydrogen atmosphere. These so-called “second-generation”ABA's employ vanadium (V) as the active constituent. They expand the application of active brazing technology to areas where the preferred, or required, furnace atmosphere is dry hydrogen. One example of a second generation V-ABA braze alloy is based on modifying the eutectic binary AWS BAu-4 alloy (82 Au-18Ni) with additions of Mo and V to form a quarternary alloy (75-98% gold, 0.5-20% nickel, 0.5-6% vanadium, 0.25-6% molybdenum, and 0-5% chromium, wt. %). These ductile braze alloys have a liquidus temperature in the range of approximately 960 C to 1100 C. See U.S. Pat. Nos. 5,385,791, 5,301,861, and 5,273,832 by Mizuhara and Huebel. Such braze joints have yielded generally acceptable hermeticity and tensile strength, but the reaction product(s) responsible for adhesion, a necessary condition for fully realizing the optimum joint properties, is not well understood.

[0009] In previous research, the authors fabricated ASTM F-19 tensile specimens of alumina brazed to alumina using Au-16Ni-0.7Mo-1.7V filler metal, that were brazed at 1000 C in slightly positive dry hydrogen. All of these samples were air-fired (e.g. re-sintered) at 1575° C. in air prior to brazing, in order to heal surface flaws from grinding. See Mizuhara, H. and Mally, K. 1985, “Ceramic-to-Metal Joining with Active Brazing Filler Metal”, Welding Journal 64(10): 27-32. The air-fired/brazed 94% alumina tensile specimens had an average tensile strength of 100 MPa, compared to only 40 MPa average strength for the air-fired/brazed high purity 99.8% alumina samples. The increased joint strengths can be attributed to favorable reactions or synergisms with the 6% balance of other constituents in 94% alumina.

[0010] It is well known that direct brazing to alumina using titanium as the active element results in a well-defined titanium-oxide chemical reaction layer. However, the authors could not find a comparable well-defined reaction layer in alumina ceramic joints brazed with a V-containing, gold-based filler alloy. Rather, electron microscopy of those braze joints revealed the presence of a discontinuous spinel-type (Al-V-O) compound of limited thickness, typically less than 30 nm at the metal/ceramic interface. However, machined alumina specimens that were air-fired at 1575° C. prior to brazing exhibited small (<50 angstroms thick) quantities of a silicon-rich constituent on external surfaces. This was particularly evident on 94% alumina specimens, which contain 6% of a deliberately added glassy binder phase consisting of a large fraction of silica (SiO₂). Silicon was also detected, albeit in reduced quantities, on high purity 99.8% alumina samples after the air-firing step.

[0011] It is possible that the presence of silicon on the surface of the 94% alumina samples during brazing may have contributed to the superior bond strength (100 MPa), as compared to the 99.8% alumina samples strength (40 MPa). This speculation is supported by wetting trials, where surface analysis of an Au-16Ni-0.7Mo-1.7V sessile drop on 94% alumina (melted at 1000 C in dry hydrogen) revealed solidified braze material concentrated along the exposed glassy phase between the alumina grain boundaries. Thermodynamic calculations of the Gibbs free energy values for simple binary reactions between alumina (or silica) and V (or Mo) indicate that the combination of silica and V produced the lowest free energy value, which is consistent with the observed wetting behavior of the glass grain boundary phase (mostly silica) in the 94% alumina binder. These calculations indicated that silica has only marginal thermodynamic stability in the presence of vanadium at the brazing temperature. Furthermore, SiO₂is reduced at 1000 C in a hydrogen atmosphere having a dewpoint of approximately −80 C, only slightly drier than the environment used for these experiments (approximately −60 C). Finally, small quantities of silicon were detected in the braze joints of the 94% alumina specimens; providing evidence of some reduction of SiO₂and diffusion of silicon during brazing.

[0012] The presence of the glassy grain boundary phase in the 94% specimens could improve the joint strength of by several potential mechanisms. Since the glassy phase represents approximately 8% (by volume) of the specimen, the surface area of Al₂O₃exposed to the molten braze alloy is reduced accordingly. Secondly, following the air-firing step, a thin (e.g. angstroms) layer of silicon-rich material covers or partially covers the newly exposed (e.g. machined) surfaces. This coverage, in addition to the original 8% surface component, may facilitate wetting and possibly reduce the time necessary for nucleation of the reaction layer.

[0013] Finally, the reduction of SiO₂by vanadium, or by the hydrogen braze atmosphere, could lead to the presence of either free silicon or oxygen dissolved in the molten braze filler metal. These compositional changes may locally influence the thermodynamic activity of vanadium in the braze alloy, thus changing the growth kinetics of the interface reaction product. Similarly, Si or O dissolved in the molten filler metal may also act as a catalyst for the reaction.

[0014] For these reasons, we concluded that the joint strength and hermeticity of a high purity alumina specimen brazed in hydrogen with a gold-based V-ABA filler metal would be greatly improved by the application of a thin coating of pure silicon to the specimen's surface, followed by firing the Si-coated specimen in air at a high temperature (>1400 C) for a time sufficient to oxidize at least some of the silicon coating into silicon dioxide. This stable surface preparation technique could allow successful direct brazing to a second member of ceramic or metal using an vanadium active braze alloy in a hydrogen atmosphere, without needing to perform a conventional Mo—Mn—Ni metallization step.

[0015] FIG. 1 shows a first example of a method of preparing a ceramic surface for brazing, according to the present invention.

[0016] FIG. 2A shows a specimen of high purity alumina.

[0017] FIG. 2B shows a coating of silicon applied by PVD on to the specimen of FIG. 2A, according to the present invention.

[0018] FIG. 2C shows a semi-continuous coating of silicon dioxide, after exposing the specimen of FIG. 2B to air at 1575 C for 2 hours, according to the present invention.

[0019] FIG. 3 shows a third example of a method of preparing a ceramic surface for brazing, according to the present invention.

[0020] FIG. 4A shows a fourth example of the method of the present invention, prior to brazing together two ceramic members.

[0021] FIG. 4B shows a fourth example of the method of the present invention, after brazing together two ceramic members.

[0022] FIG. 5A shows a fifth example of the method of the present invention, prior to brazing together a metal and a ceramic member.

[0023] FIG. 5B shows a fifth example of the method of the present invention, after brazing together a metal and a ceramic member.

[0024] FIG. 6 shows an ASTM F19 tensile specimen, used for testing the ultimate strength of a ceramic-to-ceramic braze joint.

[0025] FIG. 7 shows results of ASTM F19 tensile tests for various grades of alumina specimens direct brazed in dry hydrogen at 1000 C with an V-ABA alloy, with and without a thin coating of PVD silicon, followed by air-firing at 1575 C for 2 hours, according to the present invention.

[0026] The present invention relates to a method for preparing the surface of a high purity alumina ceramic or sapphire specimen that enables direct brazing in a hydrogen atmosphere using an active braze alloy. The present invention also relates to a method for directly brazing a high purity alumina ceramic or sapphire specimen to a ceramic or metal member using this method of surface preparation, and to articles produced by this brazing method.

[0027] FIG. 1 illustrates a first example of a method of preparing the surface for direct brazing, according to the present invention. In step 100, a ceramic component (e.g. high purity alumina) is provided. Next, in step 110, a thin coating of silicon is applied to at least the surface to be bonded. The silicon layer can be applied by a variety of methods well known in the art. Because the silicon layer can provide an effective preparatory coating with a thickness in the range of 25 to 2500 Angstroms, the method of physical vapor deposition (PVD) is ideally suited for this application. Details of the PVD method will be provided later.

[0028] Other methods of deposition, also well known in the art, can be used to deposit an effective coating of silicon. Some examples include physical sputtering, chemical vapor deposition (CVD), plasma-assisted chemical vapor deposition (PA-CVD), thermal evaporation, thermal spray, plasma spray, cold spray, and ion-beam deposition, or combinations thereof. Alternatively, a mixture of silicon powder and a volatile binder can be applied by painting, screen-printing, or spraying, and subsequently fired to volatize and remove the binder.

[0029] In the experimental results presented later, two different thicknesses of silicon were applied: 25 Angstroms and 2500 Angstroms. However, it is believed that silicon thicknesses as small as 10 Angstroms, and as large as 1 micron, would also be effective as a preparatory coating for direct brazing.

[0030] Referring still to FIG. 1, the silicon coating is then converted, in step 120, to form at least a semi-continuous coating of silicon dioxide (SiO₂), also known as silica. Conversion to silica can be accomplished by oxidation, for example, by heating the silicon-coated ceramic component(s) to a high temperature in air for sufficient time to oxidize at least some of the coating. In the experiments described later, the samples were heated at 1575 C for 2 hours in air, which successfully converted the silicon coating to silica. However, it is believed that heating in air at a temperature as low as 500 C, for a time as short as 1 hour, would be sufficient to convert at least some of the silicon coating to silica. Other methods of converting the silicon coating can be used, including chemical treatments, exposure to an oxygen plasma, and energetic oxygen ion bombardment, or combinations thereof.

[0031] Heating alumina specimens at 1575 C for 2 hours can provide an additional benefit of healing any small surface cracks (e.g. re-sintering) that may have been introduced during the machining of the alumina surfaces. It is well-known that grinding can introduce extensive surface fractures, up to 50 microns deep. See Mizuhara, H. and Mally, K. 1985, “Ceramic-to-Metal Joining with Active Brazing Filler Metal”, Welding Journal 64(10): 27-32. Subsequent lapping of the ground surfaces can remove these surface flaws. Alternatively, re-sintering by heating in hydrogen for 30 minutes at 1500 C, or by heating in air for 30-60 minutes at 1600-1650 C, can heal these cracks by liquid phase flow of the glassy phases present in less-than-100% pure alumina ceramics. Re-sintering at 1575 C for 2 hours in air is currently recommended by a manufacturer of alumina products to heal these cracks.

[0032] FIGS. 2A-C illustrate a second example of a method of preparing the surface for direct brazing, according to the present invention. In FIG. 2A, a ceramic component 10 (e.g. high purity alumina) has been provided. Next, in FIG. 2B, a thin coating of silicon 20 has been applied to at least the surface to be bonded of specimen 10. Finally, in FIG. 2C, silicon coating 20 has been converted into at least a semi-continuous layer of silicon dioxide 30.

[0033] FIG. 3 illustrates a third example of a method of preparing the surface for direct brazing, according to the present invention. In step 100, a ceramic component 10 (e.g. high purity alumina) is provided. Next, in step 110, a thin coating of silicon is applied to at least the surface to be bonded. Next, in step 120, the silicon coating is converted into at least a semi-continuous coating of silicon dioxide. Next, in step 130, an active braze filler metal is provided. As explained earlier, the active element(s) can include strong oxide-formers, including Ti, Zr, Cr, V, Hf, Nb, and Ta, or combinations thereof.

[0034] For direct brazing in a hydrogen atmosphere, vanadium is preferred over titanium. One example of a suitable V-ABA braze alloy is 75-98% gold, 0.5-20% nickel, 0.5-6% vanadium, 0.25-6% molybdenum, and 0-5% chromium (wt. %). A preferred V-ABA braze is Au-16Ni-0.7Mo-1.7V, which can be brazed at 1000 C in a dry hydrogen atmosphere. The active braze filler metal can be in the form of a foil, wire, paste, powder, or thin film coating (such as might be deposited by PVD, CVD, sputter-plating, electroplating, etc.). A preferred form is as a foil.

[0035] Next, in step 140, the filler metal is placed in-between the silicon-coated/air-fired ceramic component and a second member to be joined, and the parts are assembled in their proper orientation. A braze fixture, well-known in the art, can be used to provide proper alignment of the parts during brazing. Next, in step 150, the fixtured assembly is heated in a non-oxidizing atmosphere to a temperature above the liquidus temperature of the active braze filler metal. Any method of heating, well known in the art, can be used. Examples include resistance furnace, induction furnace, direct resistive (e.g. Joule) heating, laser heating, IR lamp heating, and electron-beam heating. Preferably, a resistance furnace is used. The furnace atmosphere can be an inert gas, such as dry, high-purity argon, helium or nitrogen. Alternatively, a vacuum can be used. Preferably, the furnace has a dry hydrogen atmosphere with a slightly positive pressure, and a dew point of about −60 C. Finally, in step 160, the brazed assembly is cooled back down to room temperature, preferably slowly.

[0036] The heating step 150 can include holds at various intermediate temperatures (e.g. to provide for adequate thermal equalization, or to accomplish desired metallurgical transformations).

[0037] A preferred multi-step heating cycle for the Au-16Ni-0.7Mo-1.7V active braze alloy can be:

[0038] Heat from ambient to 925° C. at 10° C./min;

[0039] Soak at 925° C. for 10 min;

[0040] Heat from 925 to 1000° C. at 5° C./min;

[0041] Soak at 1000° C. for 5 min;

[0042] Furnace cool to 800° C. at approximately 15° C./min;

[0043] Continue furnace cool from 800° C. to 400° C. at 10° C./min; and

[0044] Complete remaining cool to ambient at 5° C./min.

[0045] FIGS. 4A-B illustrate a fourth example of a method of direct brazing, according to the present invention. In FIG. 4A, a piece of active braze alloy 40 is placed in-between two ceramic components 10 and 10′. Both ceramic components have been coated with silicon, and subsequently converted into a semi-continuous coating of silicon dioxide 30 and 30′. In FIG. 4B the assembly of FIG. 4A has been brazed in a non-oxidizing atmosphere to form a useful article comprising a braze joint 50 joining together the two ceramic components 10 and 10′.

[0046] FIGS. 5A-B illustrate a fifth example of a method of direct brazing, according to the present invention. In FIG. 5A, a piece of active braze alloy 40 is placed in-between a ceramic component 10 and a metal member 60. Only the ceramic component 10 has been coated with silicon, and subsequently converted into a semi-continuous coating of silicon dioxide 30 and 30′. The metal member 60 does not need to be silicon coated, since the active braze alloy can directly wet most metals of commercial interest. In FIG. 5B the assembly of FIG. 5A has been brazed in a non-oxidizing atmosphere to form a useful article comprising a braze joint 50 joining together the ceramic component 10 and the metal member 60.

[0047] The following text provides a detailed description of a series of experiments that were performed by the authors, which serve as relevant examples of the practice of the present invention.

[0048] The base material consisted of a commercial grade of 99.8 weight % alumina ceramic procured from Coors Ceramics, Golden, Colo. Specimens were machined to a geometry based on the ASTM F19-64 tensile button specimen, as shown in FIG. 6, for evaluating the tensile strength of brazed ceramics. The filler composition used was a commercially available gold-based alloy containing 16% Ni, 0.75% Mo and 1.75% V (wt. %). Braze performs were fabricated from 0.076 mm (0.003 in) thick foil stock obtained from Wesgo, Inc., San Carlos, Calif.

[0049] The annular bonding surface of each as-fabricated (i.e., machined) ceramic tensile button was coated with a thin layer of pure silicon. Plasma vapor deposition (PVD) was used to deposit silicon in two thicknesses: 25 angstroms (0.1 μin.) or 2500 angstroms (10 μin.). The ceramic samples were given a standard cleaning treatment. Samples were then mounted 12 inches above a silicon source. The silicon source was heated to about 1400 K using a 10 KV electron beam gun. The system was pumped down to less than 5×10⁻⁶Torr. The parts were coated at ambient temperature. The 25 angstrom thick silicon coating was deposited at a rate of 1-2 angstroms/second. The 2500 angstrom thick coating was deposited at a rate of 10 angstroms/second. Once the silicon source was cool, the chamber was brought up to atmosphere and the parts removed.

[0050] Following the coating step, the alumina buttons were placed in 99.8% alumina crucibles and fired in a resistance heated air furnace at 1575° C. (2867° F.) for 2 hours. The refractory lining of the furnace consisted of bricks of approximate composition 50% SiO₂and 47% Al₂O₃, with small amounts of iron and titanium oxides.

[0051] Surface analyses of tensile buttons (one of each Si thickness and an uncoated 99.8% and 94% specimen) was conducted after air firing. Secondary Ion Mass Spectroscopy (SIMS) was performed using a Physical Electronics Model TRIFT-1 instrument to analyze surface compositions. Specimens were subsequently carbon-coated and examined using a JEOL 6400 Field Emission Scanning Electron Microscope (FESEM) equipped with Energy Dispersive X-ray Spectroscopy (EDS).

[0052] Eight tensile specimens, four representing each of the two Si coating thicknesses, were brazed at 1000° C. (1832° F.) in a slightly positive dry hydrogen atmosphere with a dewpoint of approximately −60° C. The test specimens were heated in a fixture designed to maintain axial alignment. A small weight was placed on top of the ceramic/braze foil assembly to produce a total bonding pressure of 1.0 psi (7 kPa).

[0053] The programmed furnace thermal profile was as follows:

[0054] Heat from ambient to 925° C. at 10° C./min

[0055] Soak at 925° C. for 10 min.

[0056] Heat from 925 to 1000° C. at 5° C./min.

[0057] Soak at 1000° C. for 5 min.

[0058] Furnace cool to 800° C. at approximately 15° C./min.

[0059] Continue furnace cool from 800° C. to 400° C. at 10° C./min.

[0060] Complete remaining cool to ambient at 5° C./min.

[0061] The furnace temperature was controlled with thermocouples located on the fixtured tensile specimens. Orientation in the furnace (top vs. bottom) was maintained for each tensile sample.

[0062] Prior to mechanical testing, the ASTM F19 tensile specimens were checked for hermeticity using an Alcatel ASM-10 Helium Leak Detector. The open central region of the test specimen was evacuated and helium was introduced around the outer fillet. If no leaks were detected, the test scale was changed to increasing helium sensitivity levels until the machine limit was reached. If a leak was detected, the leak range was quantified. All eight of the tensile specimen assemblies (four brazed with each of the two silicon coating thicknesses) exhibited hermetic joints, i.e. no detectable leaks were found.

[0063] Room temperature tensile tests were performed with a hydraulic load frame in displacement control mode at a test rate of 0.0076 mm/s (0.0003 in/s). Universal joints were used to align the test specimens in the grips during tension testing. The “top” half of each specimen, as determined by its orientation in the brazing furnace, was loaded in the upper grip of the test machine.

[0064] Finally, one brazed tension specimen from each processing group was selected for microstructural analyses. A sample was cross-sectioned, mounted, and metallographically polished for examination using both optical and scanning electron microscopy. Additionally, several fracture surfaces of failed tensile specimens were examined.

[0065] SIMS analysis of the air-fired surfaces indicated the presence of Ca, Mg, Na and K, in addition to Al and Si on all samples. By using the Al peaks, the Si peak energies were corrected and found to be consistent with SiO₂. Silicon levels were found to be approximately 7% on an air-fired 94% alumina sample, and 4%, 7% and 11% respectively on uncoated (bare), 25 angstrom and 2500 angstrom silicon coated 99.8% alumina respectively. The K and Ca values were approximately constant for the three 99.8% alumina samples. Surfaces of the 94% alumna contained substantially higher K and a reduced Ca content.

[0066] SEM images of the four air fired surfaces were analyzed. The 94% alumina specimen appears to have a majority of grains that possess fine, slightly roughened surfaces. Only a small amount, approximately 10% of the surface area, was composed of smooth, sharply cleaved facets. Conversely, the 99.8% alumina uncoated surface shows a slight majority (≈60%) of clean faceted grains. Surfaces of the Si-coated high purity alumina had increased amounts of the “rough” features, approximately 60 and 80 area % for the 25 and 2500 angstrom coated samples, respectively. Surprisingly, the surface coverage of the air-fired 2500 angstrom coating was not continuous.

[0067] EDS analysis of the smooth facets, in all specimens, indicated low silicon content at these areas of less than 0.5 weight %. Silicon content of the rough areas varied; measurements of between 3% and 7% were obtained. Note that whereas EDS utilizes a higher energy electron beam, compositional values are more likely to reflect combined surface and substrate chemistry and are consequently less reliable than the above-mentioned mass spectrometer (SIMS) results which measured average surface composition. However, the EDS data clearly indicate that the silicon content is not uniformly distributed across the ceramic surfaces, but rather concentrated at certain locations.

[0068] A typical braze joint microstructure was observed to have a thickness of 60-80 μm (0.002-0.003 in); is porosity-free; and has several phases distributed throughout the microstructure. Large globular Ni-rich features were observed in the upper half of each joint; the remainder of the microstructure consisted of large, two phase grains separated by finer particles. Generally, the solidified braze filler metal maintained intimate contact with the alumina substrate along the bond interface, with only isolated locations where slight separations were found between the metal and alumina. No significant microstructural differences were found in these specimens compared to previously brazed, uncoated alumina samples.

[0069] Tensile test results are shown in FIG. 7. They are compared with previously obtained values for the two uncoated grades of alumina (94% and 99.8%). The silicon-coated/air-fired 99.8% high purity alumina specimens had improved room temperature tensile strength compared to their uncoated/air-fired 99.8% counterparts, more than double that without any silicon coating. However, none of silicon-coated specimens yielded greater tensile values than the uncoated/air-fired 94% alumina grade ceramic (possibly because the silicon was already present in the 6% glassy phases of the 94% alumina grade).

[0070] Although the thinner silicon coating (25 A) was not as effective as thicker coatings (2500 A), these results indicate that the improvement in performance is not sensitive to minor variations in coating thickness. Consequently, the coating thickness need not be carefully controlled.

[0071] Examination of the fracture surfaces indicated that failures were located at or near the “lower” metal-ceramic interface of each braze joint. This result is in agreement with the previous observations for uncoated specimens. In the uncoated 99.8% specimens, fractures were dominated by clean separation at the metal-ceramic interface. However, in the coated specimens significant portions of the substrate (e.g. ceramic) fractured. Specimens fabricated using the angstrom silicon coating showed nearly equal amounts of ceramic and interface failure; those specimens that were given the thicker silicon coating exhibited approximately 60 to 70% (by area) substrate failure. The uncoated 94% alumina specimens failed almost exclusively in the ceramic substrate.

[0072] Furthermore, the fracture surfaces of uncoated 99.8% alumina specimens were characterized by a clean separation; almost no deformation was evident on the resulting metallic surfaces. Corresponding surfaces of the silicon-coated specimens revealed localized deformation of the metallic side of the interface during fracture.

[0073] The experimental results indicate that a thin coating of silicon on high purity 99.8% alumina more than doubles the tensile strength of the braze joint when the silicon coating is air-fired at a high temperature (e.g. 1575 C for 2 hours) prior to brazing.

[0074] In additional tests, when the step of air-firing the silicon-coated samples was eliminated, no increase in the brazed tensile joint strength could be found. Likewise, no increase in strength could be found if the silicon was added directly to the braze alloy composition, rather than being PVD-coated onto the alumina's surface and then air-fired. This further supports the importance of converting the thin silicon coating to at least a semi-continuous layer of silicon dioxide prior to brazing.

[0075] The beneficial effect of silicon on mechanical strength may be the result of one or more contributing phenomena. The SiO₂-containing film may be partially or completely reduced during brazing in the hydrogen atmosphere, resulting in free silicon dissolved in the molten braze alloy. Since the SiO₂-containing surface layer on the coated samples is very thin compared to the total braze joint width, it seems unlikely that the presence of such limited quantities of silicon would substantially alter bulk properties of the solidified filler metal (lattice parameter, thermal expansion coefficient, modulus, etc.)

[0076] Examination of the Al₂O₃—SiO₂system indicates negligible solubility of SiO₂in Al₂O₃. It is known that at temperatures of 1650° C. and above, semi-infinite diffusion couples formed thin layers of mullite (3Al₂O₃. 2SiO₂) at the interface. At lower temperatures (1550 and 1580° C.), comparable to the air-fired temperature in the current investigation, the mullite phase was absent. However, an amorphous phase of higher SiO₂composition was found. The absence of the mullite phase was attributed to its rapid dissolution into the liquid SiO₂phase. In the current study, the thin surface coating of silicon likely oxidizes to SiO₂during heating to 1575° C. The subsequent layer of molten SiO₂would be very thin. Dissolution of a mullite reaction layer would rapidly saturate the liquid SiO₂with Al₂O₃, thereby stabilizing the growth of a thin mullite film at the Al₂O₃interface. Reduction or dissolution of the predominantly SiO₂surface layer would then expose a thin mullite layer to the molten active filler metal. Mullite may offer superior (compared to Al₂O₃) wetting and adhesion characteristics for the V-containing braze alloy.

[0077] Alternatively, silicon may participate at one or more of the interfaces, specifically: braze alloy/alumina, braze alloy/reaction layer, or reaction layer/alumina. Note that silicon does not appear to be a component of the interfacial reaction product. We identified this compound in both the 99.8% and 94% alumina joints as an Al-V-O constituent, which was void of silicon. Although slight amounts of silicon were found in the bulk solidified filler metal of 94% alumina joints, prior efforts were unable to detect the presence of silicon at the metal-ceramic interface.

[0078] In summary, room temperature tensile strength was found to be proportional to the fraction of air fired surface covered by silicon-containing films. Similarly, the ratio of substrate fracture versus interface separation was also related to the amount of surface silicon present prior to brazing.

[0079] The particular sizes and equipment discussed above are cited merely to illustrate a particular embodiment of this invention. It is contemplated that the use of the invention may involve components or methods having different characteristics. For example, the method may be used for ceramics other than high-purity alumina (e.g. a ceramic with less than 99.8% alumina composition by weight, sapphire, silicon carbide, silicon nitride, sialon, mullite, aluminum nitride, zirconia, partially stabilized zirconia, and beryllium oxide, or combinations thereof).

[0080] It is intended that the scope of the invention be defined by the claims appended hereto.