Brazing & Soldering
Brazing & Soldering
C. A. WALKER (cawalke@sandia.gov) and V. C. HODGES are with Sandia National Laboratories, Albuquerque, N.Mex.
       Sandia is a multiprogram laboratory operated by Sandia Corp., a Lockheed Martin Company, for the U.S. Dept. of Energy under contract
       DE-AC04-94AL85000.
       Based on a paper presented at the International Brazing & Soldering Symposium held during the 2007 FABTECH International & AWS
       Welding Show, Nov. 11–14, in Chicago, Ill.
                                                                                                                    WELDING JOURNAL            43
Walker Feature October 2008:Layout 1                      9/8/08   3:10 PM    Page 44
       Fig. 2 — Ceramic-to-metal braze using molybdenum-                                  Fig. 3 — Au/Cu brazed metal-to-ceramic sample made using thin-
       manganese/nickel plate metallization.                                              film metallization.
Filler Metal Nonmetal Substrate Metal Substrate Brazing Temperature/Time Furnace Atmosphere Average Tensile Strength(a)
      65 Cu/35 Au               94% Alumina               Fe-29Ni-17Co             1040°C/3 min                Dry Hydrogen           14.5 ksi/100 MPa
      50 Au/50 Cu               94% Alumina               Fe-29Ni-17Co             1000°C/3 min                Dry Hydrogen           17 ksi/118 MPa
      72 Ag-28 Cu               94% Alumina               Fe-29Ni-17Co             810°C/3 min                 Dry Hydrogen           14.3 ksi/99 MPa
      77 Au-13Ag-10 Ge          94% Alumina               Fe-29Ni-17Co             495°C/3 min                 Dry Hydrogen           15.6 ksi/108 MPa
      77 Au-13 Ag-10 Ge         94% Alumina               Fe-29Ni-17Co             455°C/5 min                 Dry Hydrogen           16.1 ksi/111 MPa
          (2x thk)
      The application of the coating may be                     to provide the necessary metallization           controlled expansion alloy often used
      hand-painted, sprayed, or robotically                     component materials or metallization             when brazing to ceramics. All of the
      applied. After air drying, the coating is                 services.                                        brazed samples shown in Table 1, as well as
      fired in a wet hydrogen environment                           The molybdenum-manganese/nickel              those shown in the subsequent tables
      (15°–30°C dew point) at 1450°–1600°C                      plating method also has several disadvan-        (Tables 2–4) passed a helium mass-spec-
      leaving a “glassy” metallic coating                       tages. Included in these are the following:      trometer leak detection test (leak rate <
      300–500 micro-inches (7.6–12.7 microns)                       1) Expense. Specialized high-tempera-        2.0–9 atm-cc/s) prior to being tensile test-
      thick. The fired coating is subsequently                  ture furnaces and plating equipment are          ed. The crosshead speed used for the ten-
      plated with a 0.001–0.003 in. (25.4–76.2                  necessary — Fig. 1.                              sile tests was 3.3–4 in./s (8.38–6 m/s). The
      microns) layer of nickel. The nickel plat-                    2) Lengthy time requirements.                tensile strengths shown in the tables are
      ing is sinter-fired at 850°–950°C in a dry                Multiple high-temperature furnace oper-          averages of samples tested. Variations of
      hydrogen (–50°C dew point or less)                        ations are required as well as the care and      ± 2 ksi (14 MPa) from the average tensile
      atmosphere leaving a finished metallic                    maintenance of plating baths.                    strengths were observed. Formulations of
      surface that can be readily brazed using                      3) Rework limitations. Excessive nick-       brazing filler metals are displayed in wt-%.
      standard braze filler metals.                             el depletion into the braze filler metal can         A scanning electron microscope
          Some of the advantages of the molyb-                  lead to poor braze joint performance.            (SEM) image of a cross-sectioned brazed
      denum-manganese/nickel plating method                         4) Geometric constraints. Large sizes        metal-ceramic assembly, utilizing moly-
      are as follows:                                           and thick cross sections are difficult to        manganese metallization and nickel plat-
          1) Having been developed in the 1930s                 process.                                         ing is shown in Fig. 2. The ceramic is 94%
      (Ref. 2), moly-manganese metallization is                     5) Batch size. Process development for       alumina, and the metal member is Fe-
      a mature technology with a proven histo-                  small quantities is often cost prohibitive.      29Ni-17Co. Notice the 25–35-μm-thick
      ry of success;                                                Table 1 shows the average strengths          reaction zone where the moly-manganese
          2) Postmetallization, ceramic materi-                 typically obtained [14–17 ksi (99–117            metallization diffuses and reacts with the
      als can be easily brazed using standard                   MPa)] when using various gold- and silver-       glassy phases of the alumina ceramic. The
      braze filler metals; and                                  based brazing filler metals to braze 94%         clearly defined nickel plating layer shown
          3) Commercial suppliers are available                 alumina ceramic to Fe-29Ni-17Co alloy, a         has been sufficiently wetted by the braz-
        44    OCTOBER 2008
Walker Feature October 2008:Layout 1                      9/8/08   3:11 PM    Page 45
       Fig. 4 — Ag/Cu brazed metal-to-ceramic sample made using thin-                     Fig. 5 — Ti/Au thin-film deposition layer on Ag/Cu brazed metal-
       film metallization.                                                                ceramic sample.
       Filler Metal             Substrates                Thin Films         Brazing Temperature/Time     Furnace Atmosphere     Average Tensile Strength(a)
      50 Au/50 Cu             94% Alumina                 0.25 μm Ti/             1000°C/3 min               Dry Hydrogen             15.1 ksi/102 MPa
                              Fe-29Ni-17Co                0.5 μm Au
      50 Au/50 Cu             94% Alumina                 0.25 μm Ti/             1020°C/10 min              Dry Hydrogen             12.9 ksi/89 MPa
                              Fe-29Ni-17Co                0.5 μm Au
      50 Au/50 Cu             94% Alumina                 0.25 μm Ti/             1000°C/3 min               Dry Hydrogen             16.1 ksi/111 MPa
                              Fe-29Ni-17Co                0.02 μm Pd/
                                                          0.5 μm Au
      50 Au/50 Cu             94% Alumina                 0.25 μm Ti/             1020°C/10 min              Dry Hydrogen             11.8 ksi/81 MPa
                              Fe-29Ni-17Co                0.02 μm Pd/
                                                          0.5 μm Au
      72 Ag-28 Cu             94% Alumina                 0.25 μm Ti              810°C/3 min                Dry Hydrogen             13.0 ksi/90 MPa
                              Fe-29Ni-17Co                0.5 μm Au
      63 Ag-27 Cu-10 In       94 % Alumina                0.25 μm Ti/             755°C/2 min                UHP Argon                13.2 ksi/91 MPa
                              Fe-29Ni-17Co                0.5 μm Pt
      63 Ag-27 Cu-10 In       951 LTCC                    0.25 μm Ti/             755°C/2 min                UHP Argon                8.2 ksi/57 MPa
                              Fe-29Ni-17Co                0.5 μm Pt
      63 Ag-27 Cu-10 In       951 LTCC                    0.5 μm Ti/              755°C/5 min                UHP Argon                6.5 ksi/45 MPa
                              Fe-29Ni-17Co                0.5 μm Au
      63 Ag-27 Cu-10 In       951 LTCC                    0.5 μm Ti/              755°C/5 min                UHP Argon                3.8 ksi/26 MPa
                              Fe-29Ni-17Co                0.5 μm Pd
      ing filler metal to provide high joint                    to a ceramic substrate so that it may be       nium, chromium, niobium, etc. may be
      strength and hermeticity. The light and                   joined using conventional braze filler         chosen depending on the application and
      dark areas within the brazed joint are the                metals. A combination of materials, usu-       service temperature. Occasionally, an
      silver-rich and copper-rich regions.                      ally two or three, are deposited onto the      intermediate layer or layers are deposited
                                                                nonmetallic surface using a physical           to prevent unwanted metallurgical reac-
      Thin-Film Deposition                                      vapor deposition (PVD) method such as          tions between the initial metal layer and
                                                                evaporation or sputtering. The first layer     the braze filler metal. The top, or outer,
         Depicted in Fig. 1B, thin-film deposi-                 deposited, often titanium, is typically        layer is normally a noble metal such as
      tion is another commonly used (Refs. 2,                   0.05–0.25 μm thick. Other strong oxide-        gold, platinum, or palladium that is
      3) method to apply a metallization layer                  forming elements such as hafnium, zirco-       0.25–1.0 μm thick. A noble metal is cho-
                                                                                                                            WELDING JOURNAL              45
Walker Feature October 2008:Layout 1           9/10/08       1:05 PM   Page 46
       Fig. 7 — SEM backscattered image (BSE) with energy-dispersive         Fig. 8 — Energy-dispersive spectroscopy maps showing postbrazed
       spectroscopy (EDS) maps showing postbrazed zirconium, oxygen,         silver, iron, nickel, and copper concentrations (counterclockwise
       and aluminum (counterclockwise from upper left) concentrations        from upper left) in an active brazed specimen.
       in an active brazed specimen.
        46    OCTOBER 2008
Walker Feature October 2008:Layout 1            9/10/08   1:05 PM    Page 47
                                                                                B
       Fig. 9 — Active brazed molybdenum to 94% alumina ceramic sam-
       ple (A and B) and active brazed Fe-29Ni-17Co to 94% alumina ce-
       ramic (C and D).
                                                                                                                WELDING JOURNAL          47
Walker Feature October 2008:Layout 1                      9/8/08   3:12 PM    Page 48
Filler Metal Nonmetal Substrate Metal Substrate Brazing Temperature/Time Furnace Atmosphere Average Tensile Strength(a)
      62 Cu-35 Au-            94% Alumina                 Fe-29Ni-17Co            1006°–1026°C                Vacuum/Partial             11–14 ksi/76–97 MPa
      2Ti-1Ni                                                                     6–8 min                     pressure Ar
97 Ag-1Cu-2Zr 94% Alumina Fe-29Ni-17Co 990°C/5 min UHV/Dry Hydrogen 15.4 ksi/106 MPa
      97 Ag-1Cu-2Zr           94% Alumina                 Fe-29Ni-17Co            963°C, 3 min                Partial pressure Ar        21.3 ksi/147 MPa
                                                                                  above liquidus
      63.00 Ag-35.24Cu-       94% Alumina                 Fe-29Ni-17Co            1040°C/2 min                Dry Hydrogen               14.5 ksi/100 MPa
      1.75Ti
      63.00 Ag-35.25Cu-       94% Alumina                 Fe-29Ni-17Co            825°–1040°C/2–10 min        Partial Pressure Ar        11–14 ksi/76–97 MPa
      1.75Ti
      63.00 Ag-35.25Cu-       94% Alumina                 Fe-29Ni-17Co            825°–1040°C/2–10 min        Vacuum                     11–16 ksi/76–110 MPa
      1.75Ti
      59.00 Ag-27.25Cu-       94% Alumina                 Fe-29Ni-17Co            755°C/5 min                 Vacuum                     14.5 ksi/99 MPa
      12.5In-1.25 Ti
      59.00 Ag- 27.25Cu-      DuPont 951 LTCC             Fe-29Ni-17Co            755°C/5 min                 Vacuum                     8 ksi/55 MPa
      12.5In-1.25 Ti
      (a) Tensile strength averages are ± 2 ksi/14 MPa.
      shows the migration of the elemental zir-                 minimal amount of the active element,           active braze filler metal. This relatively
      conium to the ceramic surface where it                    titanium, has reacted with the molybde-         new (Refs. 17, 18) silver-based filler
      reacts with available oxygen and forms                    num allowing for the majority of the tita-      metal uses zirconium as the active ele-
      the layer that the primary filler metal ele-              nium metal to react with the ceramic sub-       ment, but currently has very limited com-
      ment, silver, will wet and adhere to. A                   strate. Figure 9C reveals that a substan-       mercial availability.
      trace amount of zirconium can also be                     tial portion of the titanium has reacted
      seen in the same image bound to the sur-                  with the Fe-29Ni-17Co substrate to the          Direct-Brazing Method
      face of the Fe-29Ni-17Co. Figure 7C                       point of causing some base metal erosion
      shows a small concentration of oxygen                     to occur and hindering the ability to make          The direct-brazing method is the last
      that has dissolved into the zirconium-rich                a hermetic seal. This scavenging of the         method for joining metals to ceramics to
      region of the solidified braze filler metal.              titanium element can be prevented by            be considered. As the name implies, the
      Notice in Fig. 7D that a slight amount of                 coating the Fe-29Ni-17Co member with a          direct-brazing method allows metals to be
      aluminum from the ceramic material,                       barrier layer (Refs. 12–14). While some         directly brazed to ceramics without the
      having been replaced by zirconium, has                    scavenging of titanium does occur, there        need for metallization coatings. Unlike
      diffused through the molten braze filler                  is sufficient titanium in commercially          active filler metal brazing, however, the
      metal toward the Fe-29Ni-17Co surface.                    available active brazing filler metals to       direct-brazing method utilizes standard
      Figure 8 A–D are companion energy dis-                    make hermetic braze joints to Fe-29Ni-          brazing filler metals to accomplish the
      persive spectroscopy (EDS) maps that                      17Co substrates when careful attention is       metal-to-ceramic braze. The direct-braz-
      show the silver- and copper-rich phases                   given to surface preparation, fixturing,        ing process is illustrated on the right-
      of the resolidified brazing filler metal                  atmosphere, and the brazing thermal             hand side of Fig. 6. Comparisons of the
      along with limited dissolved Fe-29Ni-                     cycle (Refs. 15, 16).                           two brazing methods portrayed in Fig. 6
      17Co base metal.                                              Tensile test results of tensile button      illustrate how similar these processes are.
          The choice of the base metal substrate                samples made with gold- and silver-based        Similar to the active brazing process, a
      and active filler metal element can have a                active braze filler metals are displayed in     direct-braze is made by cleaning the
      substantial impact on the end product as                  Table 3. A comparison of the sample             ceramic and metal materials, fixturing the
      reported by Stephens et al. (Ref. 11), and                strengths in Tables 1 and 2 to those in         assembly with the braze filler metal pre-
      shown in Fig. 9. A and B show a molyb-                    Table 3 reveals that the results are very       placed between the metal and ceramic
      denum substrate brazed to a 94% alumi-                    similar for tensile samples brazed with         substrates and then brazing the entire
      na ceramic using a gold-based active                      similar composition filler metal families       assembly, usually in an inert or UHV
      braze filler metal, 62Cu-35Au-2Ti-1Ni. C                  and temperatures. Of particular interest        brazing atmosphere. During the direct-
      and D show the results when the molyb-                    in Table 3 are the high tensile strengths       braze process, specific metal substrates
      denum is replaced with Fe-29Ni-17Co. 9B                   obtained when using the 97Ag-1Cu-2Zr            and braze filler metal combinations inter-
      and 9C are EDS maps showing the result-                   active braze filler metal. Samples averag-      act to form an adherent metallic oxide
      ing titanium concentrations in the brazed                 ing more than 21 ksi (147 MPa) were             layer on the oxide ceramic faying surface.
      samples. Figure 9B demonstrates that a                    obtained using this recently developed              The dissolution, migration, and inter-
        48    OCTOBER 2008
Walker Feature October 2008:Layout 1            9/8/08     3:13 PM    Page 49
Filler Metal Nonmetal Substrate Metal Substrate Brazing Temperature/Time Furnace Atmosphere Average Tensile Strength
      92 Au-8Pd             94% Alumina               Niobium               1270°C/4 min                Vacuum               12.8 ksi ± 1.8 ksi/
                                                                                                                             88 MPa ± 12 MPa
      62 Cu-35Au-3Ni        94% Alumina               Niobium               1058°C/3 min                Vacuum               8.8 ksi ± 1.0 ksi/
                                                                                                                             61 MPa ± 7 MPa
      50 Au-50Cu            94% Alumina               Niobium               1000°C/3 min                Vacuum               8.8 ksi ± 1.0 ksi/
                                                                                                                             61 MPa ± 7 MPa
      action of the base metal with the filler             2) Good atmospheric control, while         The designer, engineer, or user can
      metal and ceramic surface are shown in            not as stringent as that required when        choose from a traditional metallization
      Fig. 10 (Ref. 19). Electron microprobe            active brazing, is also necessary when        method such as moly-manganese/nickel
      analysis (EMPA) of a niobium-94% alu-             using the direct-brazing method.              plating or from a variety of thin-film coat-
      mina ceramic sample brazed with 62Cu-                3) The strengths obtained using the        ings applied using PVD methods, which
      35Au-3Ni (BAu-3) braze filler metal               direct-braze method are slightly inferior     are specifically tailored to meet the needs
      shows how the niobium base metal is               to those obtained using the other dis-        of the application. Active braze filler met-
      enriched at the alumina ceramic surface,          cussed metal-ceramic brazing methods, as      als can be used as a replacement system
      where it forms a relatively stable oxide. To      seen when comparing the strength data in      for most metal-to- ceramic brazed assem-
      perform a successful direct-braze, candi-                                                       blies with no loss of mechanical proper-
      date metal substrates must contain an                                                           ties. Whether choosing to use metallized
      element or elements able to form ther-                                                          ceramics or the direct-braze process, con-
      mally stable oxides and have sufficient
                                                           Active braze filler                        ventional braze filler metals can be used
      solubility within the chosen liquid braze                                                       for the brazing operation. The direct-
      filler metal. As shown in Table 4, the
                                                          metals can be used                          braze process has been demonstrated
      direct-braze method was used to produce              as a replacement                           with a limited set of conventional filler
      tensile button assemblies having average                                                        metals to have adequate bond strength
      tensile strengths ranging from 9 to 13 ksi
      (61–88 MPa). For these assemblies, niobi-
                                                            system for most                           when used in conjunction with niobium
                                                                                                      metal substrates. Premetallized sub-
      um base metal provided the active ele-               metal-to-ceramic                           strates may be used without joint geome-
      ment required to react with the alumina                                                         try restrictions; however, active and
      ceramic.                                            brazed assemblies                           direct-brazing techniques work best with
          There is a host of benefits for the                                                         butt or lap-style braze joint geometries
      designer or engineer to use the direct-                with no loss of                          where the brazing filler metal may be pre-
      brazing method. Some of these advan-                                                            placed between the faying surfaces.◆
      tages are                                               mechanical
          1) Ease of use and lower expense,
      compared to other metal-ceramic brazing                 properties.                                         Acknowledgments
      methods;
          2) No metallization equipment or                                                               The authors wish to express their
      associated processes and process devel-           Table 4 to that shown in Tables 1–3.          thanks and appreciation to Mike Hosking
      opment is required;                               Transmission electron microscopy (TEM)        and Paul Vianco for guidance and project
          3) A variety of conventional braze            analysis results on niobium-94% alumina       support; Don Susan for his review of the
      filler metals can be utilized covering a          ceramic direct brazed samples (Ref. 20)       manuscript; and Tom Crenshaw, Alice
      wide range of temperatures;                       showed the niobium bonded with the            Kilgo, and Bonnie McKenzie for their
          4) The direct-brazing method has              glass-phase only. Though not yet evaluat-     mechanical testing capabilities, metallo-
      been successfully used to hermetically            ed, it is anticipated that a metal with the   graphic sample preparation, and image
      join metal-ceramic components used in             ability to form more thermally stable         analysis skills.
      high-reliability long-term applications.          oxides than those of niobium will be
          There are several disadvantages to            required to adequately join high-purity
      using the direct-braze method. Among              alumina ceramics using the direct braze                        References
      these are                                         method.
          1) Not all joint designs are viable.             In conclusion, high-strength, hermeti-        1. AWS C3.2M/3.2:2008, Standard
      Similar to active brazing in this regard, the     cally sealed metal-ceramic assemblies can     Method for Evaluating the Strength of
      braze filler metal must be preplaced              be successfully brazed using a variety of     Brazed Joints. 2008. Miami, Fla.:
      between the faying surfaces because the           methods, some requiring metallization of      American Welding Society. pp. 17–23.
      filler metal is unable to be drawn by capil-      the ceramic member and others allowing           2. Brazing Handbook, 5th Edition.
      lary forces along the bare ceramic surface.       the direct brazing of metals to ceramics.     2007. Miami, Fla.: American Welding
                                                                                                                  WELDING JOURNAL             49
Walker Feature October 2008:Layout 1         9/8/08   3:13 PM    Page 50
50 OCTOBER 2008