Lawrence Berkeley National Laboratory
Recent Work
Title
Transient-Liquid-Phase and Liquid-Film-Assisted Joining of Ceramics
Permalink
https://escholarship.org/uc/item/5dr9h3c1
Journal
Journal of the European Ceramic Society, 26(4/5/2008)
Authors
Sugar, Joshua D.
McKeown, Joseph T.
Akashi, Takaya
et al.
Publication Date
2005-02-09
eScholarship.org
Powered by the California Digital Library
University of California
Accepted for publication in the Journal of the European Ceramic Society
LBNL Report 57042
Transient-Liquid-Phase and Liquid-Film-Assisted
Joining of Ceramics
Joshua D. Sugar,1 Joseph T. McKeown,1 Takaya Akashi,1 Sung M. Hong,1
Kunihiko Nakashima2 and Andreas M. Glaeser1
1
Department of Materials Science and Engineering
319 Hearst Mining Building, MC1760
University of California, Berkeley, CA 94720-1760
&
Ceramic Science Program
Materials Sciences Division
Lawrence Berkeley National Laboratory, Berkeley, CA 94720-1760
USA
2
Department of Materials Science and Engineering
Faculty of Engineering, Kyushu University 36
Hakozaki 6-10-1, Higashi-ku, Fukuoka 812-81
Japan
Abstract
Two joining methods, transient-liquid-phase (TLP) joining and liquid-film-assisted joining
(LFAJ), have been used to bond alumina ceramics. Both methods rely on multilayer metallic
interlayers designed to form thin liquid films at reduced temperatures. The liquid films either
disappear by interdiffusion (TLP) or promote ceramic/metal interface formation and
concurrent dewetting of the liquid film (LFAJ). Progress on extending the TLP method to
lower temperatures by combining low-melting-point (<450°C) liquids and commercial
reactive-metal brazes is described. Recent LFAJ work on joining alumina to niobium using
copper films is presented.
Keywords: Joining, interfaces, fracture, strength, Al2O3
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
Introduction
Joining is a critical enabling technology, essential to widespread use of ceramics in many
applications. Specifically, it allows the fabrication of large, complex, multimaterial, multifunctional
assemblies through the controlled integration of smaller, less complex, more easily manufactured
parts. Additionally, it can provide an avenue for repair of damaged structures through the
replacement of defective components. This can extend the lifetimes of assemblies, and permit the
reuse of components that are not readily recycled, e.g., fiber-reinforced materials.
Material degradation during joining and interfacial reactions that produce undesirable and
structurally defective reaction layers can limit the properties and reliability of joined assemblies. The
extent of degradation or reaction often increases with increasing joining temperature. Thus, when
materials that are nanostructured and prone to coarsening, are amorphous and may crystallize, or
otherwise have temperature-sensitive properties are part of a joined assembly, it becomes increasingly
important to reduce the joining temperature below some critical threshold temperature to mitigate
such problems. Concurrently, it may be necessary to maintain the potential for service at
temperatures that approach this critical threshold temperature.
Our research focuses on developing joining methods that yield reliably strong interfaces at
“low” joining temperatures, but exploit multilayer interlayer designs that preserve the potential for
use at temperatures that equal or exceed the joining temperature. An illustrative example of an
assembly is shown in Figure 1. The cladding layers are designed to form thin liquid layers at “low”
temperatures. The core layer remains solid during joining. In transient-liquid-phase (TLP) joining,
the overall composition of the interlayer lies in a solid solution phase field. Thus, the liquid is not
stable and disappears due to interdiffusion. While the (transient) liquid is present, it fills interfacial
gaps and facilitates joint formation. The remelt temperature of the homogenized interlayer exceeds
the original joining temperature. Alternatively, the liquid film can be chemically stable but
morphologically unstable. In this case, a thin liquid film interspersed between a ceramic and a solid
!1!
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
core layer can promote rapid diffusion and the development of intimate contact between the ceramic
and core layer metal. Growth of ceramic–core layer contact progressively decreases the area fraction
of liquid along the interface. This approach is referred to as liquid-film-assisted joining (LFAJ).
In this paper, we summarize prior research using TLP joining and describe ongoing efforts
to extend the TLP method to lower temperatures by combining low-melting-point (<450°C)
cladding materials and commercial reactive-metal brazes. We also describe recent LFAJ work on
joining alumina to niobium core layers using copper films. Effects of joining temperature, time, film
thickness, and surface roughness on joint properties and interfacial microstructure evolution are
presented.
Background
Several methods for ceramic-to-ceramic, metal-to-metal and ceramic-to-metal joining are
available, and have been reviewed in the literature [1-5]. A brief overview follows.
Conventional Joining Approaches
One class of joining processes, exemplified by diffusion bonding, involves purely solid-state
processing [1, 3-12]. The components to be joined can be brought into direct contact or, as is often
the case for ceramic-ceramic joining, a metallic foil can be inserted between the ceramic components.
Application of a pressure at elevated temperature promotes the formation of a bonded interface
between the materials to be joined. The temperatures required for joining are often a high fraction of
the melting temperature of the least refractory component due to the need to activate solid-state
diffusion [1-9, 11, 13-15]. Component deformation and microstructural changes such as grain
growth or precipitate coarsening within the components can degrade properties.
A broader range of processes involves melting either the material(s) to be joined, or some
other material introduced into the joint region [1, 2, 4, 5, 16]. Examples include soldering, brazing,
and welding. Solders, by definition, melt at <840°F (<ª450°C). As a result, the joints have limited
!2!
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
temperature capability, and can be mechanically inferior to the bulk materials that have been joined.
Brazes require higher processing temperatures (>840°F). The higher melting temperatures of brazes
can lead to higher service temperatures; however, the higher processing temperatures can overlap
with the aging temperatures of some metallic alloys, resulting in a loss of peak hardness. Other forms
of microstructural degradation are also possible. Welding involves localized heating, melting, and
subsequent solidification. A major concern in welding of metals is the development of a heat-affected
zone. Although metal-metal welding is common, and ceramic-ceramic welding has been explored,
examples of ceramic-metal bonding via welding are sparse.
Nonconventional Joining Approaches
Transient-Liquid-Phase (TLP) Joining
TLP joining has been applied to a range of structural metals [17-22], notably nickel-base
superalloys, and more recently the method has been extended to intermetallics [23-25]. Reviews of
this process are available in the literature [21, 22]. When applied to metal-metal joining, an interlayer
containing a melting point depressant (MPD) is placed between the two objects to be joined. Boron
serves as an effective MPD for nickel, and is thus a common interlayer component when nickel-base
superalloys are joined [17-20]. At the joining temperature, rapid (interstitial) diffusion of boron into
the adjoining (boron-free) nickel-base superalloys leads to a progressive decrease in the amount of
liquid. Ultimately, the liquid disappears. Counter-diffusion of alloying elements in the nickel-base
superalloys into the interlayer region leads to joint chemistries and properties that approach those of
the base material, and such joints are compatible with use in structural applications at elevated
temperature.
When the method is extended to facilitate ceramic joining by metallic interlayers, the
disappearance of the liquid generally requires diffusion of a low-melting-point metal that acts as an
MPD into an adjoining solid phase. For the systems explored in our work, incorporation of the
!3!
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
MPD into the ceramic is slow in comparison to diffusion into the solid core layer of the multilayer
interlayer, and hence this latter diffusion path controls the rate of liquid disappearance.
Formation of successful joints by this approach requires that the liquid can flow along the
interface to fill gaps, and that sufficient liquid is available for complete gap filling. Gaps along the
interface are likely to arise due to roughness and waviness of the substrate surfaces, local depressions
or asperities on the surfaces, and incomplete coating of the substrate (or core layer) with the MPDcontaining layer. In conventional brazing and soldering, if two ceramic components are to be joined,
then it is the contact angle of the liquid braze or solder on the ceramic that will determine whether
the liquid will recede from (enlarge) or advance into (fill) an interfacial gap. In the case of multilayer
metallic interlayers, the liquid film is sandwiched between two dissimilar materials, the metal core
and the ceramic. Thus, two contact angles, and more specifically their sum [26], will dictate the
(short-time) response of the liquid. The surface topography will modify the energetic considerations,
and also impact the liquid film thickness required to fill interfacial voids.
In Figure 2a, a film is shown between two dissimilar but parallel substrates. The contact
angle on the core layer, q1 , is shown to be acute, as would normally be the case for a metal on a
metal, while the contact angle on the ceramic, q 2 , is shown as obtuse, as would normally be the case
for nonreactive metals on ceramics. The liquid film will fill voids if q1 +q 2 <180° . If a typical liquid
metal (with q > 90° ) were sandwiched between two ceramic substrates, the liquid would “dewet” the
interface, introduce significant porosity, and lead to nonhermetic low-strength joints. Thus, one of
the advantages of a multilayer interlayer approach is that a high q 2 is permissible, if q1 is sufficiently
low. When q1 +q 2 <180° , it implies that g Core/Liq + g Liq/Ceramic < g Core/Vapor + g Ceramic/Vapor , where gi/j is the
specific surface or interfacial energy of the i/j interface.
When the bonding surfaces are rough, a more stringent condition emerges. If the contact
angles of liquid on the core layer and the ceramic are again denoted q1 and q 2 , respectively, but local
depressions on the opposing core layer and ceramic surfaces cause angular deviations of a1 and a2 ,
!4!
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
respectively, from a parallel surface geometry, then flow of liquid into voids will only occur if the
condition (q1 + a1 )+(q 2 + a2 )<180° is met. Since q1 and q 2 can vary as the surface orientations and
surface energies of the metal and ceramic grains vary, and a1 and a2 will vary with location along
the interface, the potential exists for regions of the interface with diverging surfaces ( a1 + a2 > 0 ) to
have unfavorable wetting conditions. A rougher surface with locally larger values of a1 and a2
would be more likely to contain voids that persist or develop due to liquid film redistribution. In
addition, spatial variations in (q1 + a1 )+(q 2 + a2 ) could establish conditions that redistribute the
liquid from filled regions where the sum is higher into unfilled regions where the sum is lower,
thereby generating interfacial flaws.
When properly implemented, TLP joining methods are capable of producing joined
assemblies with reproducible and robust joint properties. When incomplete wetting occurs, regions
of the interface remain or become liquid-free, and a triple-junction ridge develops where the liquid
metal, ceramic, and vapor phases form mutual contact. Fractography indicates that these regions are
more prevalent in samples with lower fracture strength [26-29]. The wetting characteristics of the
liquid film can be improved by precoating the ceramic surface with a metal, or by altering the liquid
film chemistry [30-34]. The liquid film chemistry can be adjusted by adding directly to the cladding
layer a second component that improves wetting. Alternatively, since some dissolution of the core
layer is inevitable, the addition of elements that enhance the wetting can be achieved by their
incorporation in the core layer. In some of the systems examined, the implementation of such
approaches has yielded assemblies in which fracture occurs primarily within the ceramic, indicating
that the ceramic-metal interface has higher strength than the ceramic [34].
Liquid-Film-Assisted Joining (LFAJ)
In LFAJ, as in TLP bonding, a thin liquid film is sandwiched between a solid metal core
layer and the ceramic to be joined. Gap filling is again an issue. Thus, the same energetic issues that
apply in TLP bonding, illustrated in Figure 2a, are relevant. When (q1 + a1 )+(q 2 + a2 )<180° , the
liquid will fill gaps. However, in contrast to TLP joining, in LFAJ the liquid former does not have
!5!
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
significant solubility in the core layer, and thus, the liquid phase does not disappear with time.
However, at longer times, other microstructural changes (that may require more sluggish solid-state
diffusion) can occur. The continuous liquid metal film is initially bounded by two interfaces, one
with the core layer and the other with the ceramic. It follows that if g Core/Ceramic < g Core/Liq + g Liq/Ceramic it
would be energetically favorable to break up the liquid film and form discrete particles. “Dewetting”
of the film does not involve the generation of interfacial voids. Instead, in LFAJ the retraction and
instability of the liquid film is accompanied by the growth of ceramic–core layer interface. In the
presence of the liquid film, grain boundaries in both the core layer and the ceramic are etched, and
for diffusional growth of the grain boundary grooves, ridges can form that extend above the original
surface. When the ridge heights become sufficiently large, the ridge crests contact the adjoining
phase. Grain boundary grooving can initiate ceramic–core layer contact, producing microstructures
like those shown in Figure 3a. With increasing anneal time, the area fraction of ceramic–core layer
contact increases and can ultimately exceed 90%, as illustrated in Figures 3b and 3c, respectively.
Results of TLP and LFAJ Approaches
A large number of multilayer interlayer systems have been developed and used to join
alumina and silicon-based ceramics. In general, the low-melting-point component of the interlayer
has been copper and core layers with melting points several hundred degrees higher have been used.
Examples of multilayer interlayers used to join alumina include: Cu/Pt/Cu [27], Cu/Ni/Cu [28],
Cr/Cu/Ni/Cu/Cr, and Cu/80Ni20Cr/Cu [32, 35]. Similar strategies have been employed in
bonding silicon-based ceramics [30, 31, 33, 36]. Ceccone et al. [33] have explored the use of
Au/80Ni20Cr/Au interlayers for TLP bonding of silicon nitride. Silicon nitride [30, 31] and silicon
carbide [30] were also joined using a Cu-Au/Ni/Cu-Au based interlayer designed to form a liquid
phase at <950°C. Changes in processing conditions, specifically the processing temperature, were
found to have a strong effect on silicon nitride joint properties [31]. A plot summarizing strength
distributions for various interlayer and ceramic combinations is provided in Figure 4. Data for
silicon carbide is omitted for clarity, but can be found in reference [30]. A key point is that reliably
!6!
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
strong joints can be produced with interlayer chemistries compatible with higher service
temperatures than those typical of many commercial reactive-metal brazes.
Experimental Procedures
The sample preparation procedures in our current work are generally similar to those in our
prior work, and detailed descriptions are available in the literature [26-32, 34-39]. The summary
provided here emphasizes those procedures or materials that differ from those used previously.
Materials and Sample Preparation
Commercially available 99.5% pure (AD995, Coors Technical Ceramic Co., Oak Ridge,
TN) or 99.9% pure (SSA-999W, Nikkato Corp., Osaka, Japan) aluminum oxide in the form of
19.5 mm ¥ 19.5 mm ¥ 22.5 mm blocks was used for assemblies intended for mechanical testing. The
finer grain size 99.9% alumina has a higher fracture strength, but its properties can be affected by the
thermal cycle during joining [29]. The joining surfaces of the blocks were ground flat on a surface
grinder using a 400-grit diamond wheel. Joints processed with unpolished alumina substrates were
then cleaned while those processed with polished alumina substrates were polished with
progressively finer grit size diamond suspensions (South Bay Technologies, San Clemente, CA)
before cleaning. After polishing with a 1-µm diamond suspension, either a final chemical-mechanical
polish was performed using colloidal silica (Struers, Westlake, OH), or a final mechanical polish
using 0.25-µm grit diamond paste was performed. Samples to investigate interfacial microstructure
evolution were fabricated using ª0.5-mm-thick, high-purity, optical finish, c-axis or m-axis sapphire
substrates (Crystal Systems Inc., Salem, MA) that required no additional polishing.
For brazing and TLP joining, 75-µm-thick, 99.95% pure silver foils (Alfa Aesar, Ward
Hill, MA) and silver-based reactive-metal braze foils, Silver ABA (Morgan Advanced Ceramics,
Belmont, CA), were used. In the TLP bonding experiments, a 99.998+% pure indium source (Alfa
Aesar, Ward Hill, MA) was used to develop cladding layers. For the LFAJ experiments, a flattened
!7!
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
and cleaned, 99.99%-pure, 127-µm-thick niobium foil (Goodfellow Corp., Malvern, PA) provided
the core layer and a commercial grade copper (Consolidated Companies Wire and
Associated, Chicago, IL) served as the basis for the cladding film. Both the indium and the copper
were deposited directly onto the alumina surfaces by melting the source material and allowing it to
evaporate in a high-vacuum chamber containing the ceramic blocks. Film thicknesses were measured
using profilometry (Tencor Instruments, Inc., San Jose, CA) and weight-gain measurements
[27]. The combined thickness of the indium film and a very thin capping layer of 99.9% pure silver
(designed to prevent indium oxidation) was ª2.2 µm. The multilayer interlayer has an overall
composition (in wt%) of 89.1% Ag, 4.8% Cu, 3.9% In, 1.2% Ti, and 1.0% Al. For bonds prepared
with niobium core layers, samples with 1.4-, 3.0-, and 5.5-µm-thick copper films were prepared.
All bonding was performed in a vacuum hot press. Brazing with pure silver and with Silver
ABA was performed at 1000°C and 960°C for 10 min, respectively; silver melts at 960°C, while the
liquidus temperature of Silver ABA is 912°C. TLP bonding with an indium cladding was performed
at 700°C and 800°C, below the Silver ABA solidus temperature of 860°C with holding times varying
from as little as 20 min up to 24 h. Typical heating rates and cooling rates were 10°C/min and
8°C/min, respectively, with a typical vacuum of <10-7 atm and an applied load of ª4.6 MPa.
For polycrystalline alumina joined with Cu/Nb/Cu interlayers, assemblies were processed
with a vacuum pressure below 10–7 atm, an applied load of ª2.2 MPa, a heating rate of 4°C/min, a
6 h soak at 1400°C, and cooling at 2°C/min. To study microstructural evolution at the interface,
sapphire was used as an optically transparent surrogate for polycrystalline alumina, and bonds were
made at 1150°C using the same heating and cooling rates, dwell time, and load. Most sapphire
samples were annealed at 1150°C rather than 1400°C to decrease the rate of interface evolution and
permit a more detailed study of the kinetics.
!8!
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
Characterization
To allow a comparison between the surface roughness of the ceramic substrates and the
cladding film thickness the surface roughness of the as-ground and polished alumina blocks was
determined by atomic force microscopy (AFM).✝ The AFM scans typically covered a 50 µm ¥ 50 µm
area, and provided accurate measurements of the local fine-scale roughness. To assess longer
wavelength (and larger amplitude) variations in the surface topography, profilometer scans spanning
ª1 cm lengths were also conducted on polished and as-ground alumina. For the ground alumina,
scans were taken both parallel and perpendicular to the grinding direction.
After bonding, the assemblies were machined into beams ª3 mm ¥ ª3 mm in cross section
and ª4 cm in length, with the metal interlayer at the center of the beam. The tensile surfaces of the
beams were polished to a 1-µm finish and the beam edges were beveled to remove machining flaws
that could initiate failure. This allowed for a more meaningful measurement of the fracture strength
of the joined assembly, and the observed fracture path provided insight on the relative strengths of
the ceramic-metal interface and the bulk ceramic. Beams were tested at room temperature using
four-point bending.
The interlayer region of intact beams and the fracture surfaces of failed beams were
examined using optical microscopy to assess the interlayer microstructure and the area fraction of
ceramic-interlayer contact. Studies of interfacial microstructure evolution during LFAJ also utilized
optical microscopy. Fiducial marks produced on the outer surfaces provided reference points
allowing the same interfacial region to be identified after varying periods of annealing, thereby
providing a time sequence. Image analysis of the digital images allowed a determination of the area
fraction of copper at the interface for given annealing conditions. Numerous samples and regions
were analyzed.
✝
Measurements were also performed on the niobium foil, but the surface finish of the niobium was constant in the
experiments, and since some dissolution of niobium occurs, the niobium surface roughness will change during bonding.
!9!
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
Results and Discussion
Transient-Liquid-Phase Joining
Silver dissolves a significant amount of indium over a wide range of temperature [40]. It was
thus of interest to assess whether silver-rich interlayers could be produced in situ and used to bond
alumina when indium serves as the low-melting-point cladding layer. Brazing experiments using pure
silver foils, and TLP experiments with pure silver core layers and indium cladding layers were
performed. Neither interlayer produced useful joints. Silver forms an obtuse contact angle on
alumina [41] and was therefore expected to dewet the interface. Indium reportedly forms a high
contact angle on alumina [42], and thus, it was expected that the silver-indium combination would
also be problematic. In practice, assemblies were not sufficiently robust to survive machining into
plates and beams.
It had been anticipated that the wetting of the liquid film on alumina would need
improvement. In prior work by Nakashima et al. [32, 34] alumina joints prepared with Cu/Ni/Cu
interlayers failed exclusively along the alumina-interlayer interface, and the joint strengths varied
considerably. Examination of fracture surfaces indicated that large unbonded regions persisted along
the alumina-interlayer interface. The results suggested that these flaws were involved in failure
initiation, and that the statistical variations in these flaw sizes contributed to the wide strength
distribution. Chromium additions were shown to reduce the contact angle of molten copper on
alumina. By replacing a pure nickel core layer with an 80Ni20Cr core layer, dissolution of the core
layer during joining added chromium to the liquid film. The significant improvement in joint
characteristics achieved with a 80Ni20Cr core layer encouraged a parallel approach for silver-indium
interlayers.
Numerous researchers have used copper-silver eutectic brazes with reactive-metal additions
(i.e., Cusil ABA) to successfully join alumina (see for example reference [2]). Key to the success of
these brazes is the addition of titanium, which promotes wetting of an otherwise nonwetting eutectic
liquid. The copper-silver eutectic temperature is 780°C. Incusil ABA is an interesting derivative of
! 10 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
these brazes. Incusil ABA contains 12.5% indium, which lowers the liquidus temperature (715°C),
and 1.25% titanium, which promotes wetting. Incusil ABA has also been used to successfully join
alumina. This suggests that the copper-rich and silver-rich phases in this alloy, which contain indium
and titanium, form strong interfaces with alumina.
Joining experiments using thin indium cladding layers with Silver ABA have produced
successful joints, and results are summarized in Figure 5. To provide a basis for comparison,
samples were brazed using Silver ABA and Incusil ABA. For Silver ABA, the average four-point
bend strength was 330 MPa, with a standard deviation of 60 MPa; for Incusil ABA, the
corresponding values were 260 MPa and 35 MPa. The as-received alumina had an average fracture
strength of 320 MPa with a standard deviation of 30 MPa. Although most brazed samples failed in
the ceramic, some samples failed along the alumina-interlayer interface, while others showed mixed
ceramic and interfacial fracture paths. In samples brazed using In/Silver ABA/In interlayers, at
elevated bonding temperatures, indium will melt and incorporate both silver and titanium from the
Silver ABA core layer. Since the liquid film is silver-rich, it will be substantially thicker than the
original indium cladding layer. For bonds formed at 800°C, with hold times of 20 min, the average
fracture strength for samples that failed in the ceramic (270 MPa ±35 MPa) was comparable to
those of samples brazed with Incusil ABA. However, low-stress interfacial failures were also
observed. An examination of fracture surfaces of the weak beams suggested incomplete contact
between the interlayer and the ceramic.
Varying the bonding time (1.5, 6, and 24 h) and temperature influenced the strength
distributions. For samples bonded at 700°C, maximum average strength and minimum standard
deviation was attained after a 24 h hold. For samples bonded at 800°C, the best results were obtained
after a 1.5 h hold. In contrast to brazing, where all the titanium in the interlayer is available to form
reaction layers, in TLP bonding, the total amount of titanium in each liquid film is smaller. It is
possible that solid-state diffusion of titanium to the interface plays a role in the variations in
! 11 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
strength. Further study will be required to assess this possibility. However, considering that the core
layer compositions are optimized for brazing rather than TLP bonding, the results are very
encouraging, and studies involving Copper ABA and Cusil ABA core layers are in progress. Results
of these studies and a more complete account of work involving Silver ABA interlayers will be
published elsewhere.
Liquid-Film-Assisted Joining
Multilayer interlayers based on the copper-niobium system have been used to implement
LFAJ, and the use of Cu/Nb/Cu interlayers for joining alumina has been studied extensively [26, 29,
39, 43, 44]. A thin copper-rich film has a desirably low melting temperature (relative to niobium).
The phase diagram of the copper-niobium binary system has only one reaction isotherm, a shallow
eutectic near the copper-rich side of the diagram [40]. The solubility of copper in solid niobium is
low. In view of the high melting temperature of niobium, and the fact that copper is a substitutional
impurity, the rate of copper diffusion into niobium is also expected to be low at modest
temperatures. Thus, the copper-rich liquid that will coexist with niobium at temperatures above the
eutectic temperature will be a “permanent” feature of the interlayer microstructure. At 1150°C, the
niobium-saturated liquid contains ª0.9 at% niobium while at 1400°C it contains ª1.7 at% niobium.
The liquid film provides a high diffusivity pathway for niobium that allows the formation of
extensive alumina-niobium contact (an area fraction >ª90%) along the alumina-interlayer interface
[26, 29, 39, 43], as illustrated in Figure 3.
The energetics of the process lead to a breakup of the continuous film (described as
dewetting) and the formation of discrete copper-rich droplets along the alumina-interlayer interface
[26, 29, 39, 43] that enhance interfacial toughness by undergoing ductile tearing when the failure
propagates along the interface [44]. When properly implemented the average fracture strength of the
bonded assemblies is ª90% of the average fracture strength of the alumina that was bonded, and the
standard deviations in strength for the joined and reference assemblies are comparable [29, 39, 43]. It
! 12 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
is important to note that even at elevated temperatures, above the melting point of the dispersed
phase (ª1080°C), the joints retain a significant fraction of their room temperature strength; loss of
strength reflects in large part the strength loss of the alumina due to softening of an intergranular
glassy film [29], and partly the drop in the yield strength of the copper particles.
Experimental studies of film breakup at polycrystalline alumina-interlayer interfaces are
difficult because the interface cannot be examined nondestructively. Copper distributions in samples
that undergo alumina-interlayer interfacial failure are not necessarily representative. Using sapphire
allows the interfacial microstructure to be observed and recorded, and by examining the same region
of the interface after varying anneal times, a time sequence can be constructed. Optical micrographs
were taken at various regions of sapphire-interlayer interfaces, and image analysis was used to assess
the area fraction of copper at the interface as a function of anneal time at 1150°C. A limited number
of samples were (bonded and) annealed at 1400°C. Multiple contact initiation sites and multiple
mechanisms (asperity contact and growth, grain boundary grooving, surface instability) can
contribute to contact formation and liquid redistribution. The scaling behavior of competing
mechanisms can differ, and thus variations in the dominant mechanism will manifest themselves by
different scaling law exponents.
Efforts were made to fit the evolution behavior to a Johnson-Mehl-Avrami equation of the
form:
( )
f = 1 - exp -kt n
where f is the volume fraction of transformed material, k is a kinetic parameter, t is time, and n is an
exponent typically between 1 and 4. In this case, sapphire-copper contact transforming into
sapphire-niobium contact is interpreted as the phase transformation and f is the area fraction of
sapphire-niobium contact. The curves generated from fitting the data to this model are shown in
Figure 6a and 6b for samples with ª3-µm-thick films annealed at 1150°C and 1400°C, respectively.
! 13 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
The effect of temperature on the kinetics is evident. Samples bonded and then annealed at 1150°C
had values of k and n ranging from .045254–1.0102 and 0.253385–0.78828, respectively. The data
was divided into three subsets representing rapid, “average”, and slow transformation. For the
“average” interface, values of k and n were 0.4062 and 0.417399, respectively, which yields the middle
curve in Figure 6a, and represents the majority of the overall data. At 1400°C values of k and n
ranged from 1.3645–2.3082 and 0.47525–0.71082, respectively. The variable n in the Johnson-MehlAvrami equation is sensitive to the dimensionality of the transformation being modeled, and it
largely affects the shape of the model curve. The large variation in n for data at one temperature is
consistent with the interpretation that the mechanism dominating film breakup can vary with
position at the interface.
The roughness of the bonding surfaces is another factor that plays an important role in
liquid redistribution, film breakup, and ultimate joint properties. One measure of roughness is the
average roughness, R a , defined as:
L
Ra =(1/ L) Ú f (x)dx
0
where L is the length of the measured distance and f (x) is the distance to the surface measured
from the centerline. When surfaces are highly polished and R a is small, thinner films with more
rapid breakup kinetics may be adequate to fill interfacial gaps, but toughening contributions may be
decreased. When R a is large, a thicker film may be required for gap filling. Thick films may leave
larger patches of relatively weaker alumina-copper contact along the interface.
Prior work, conducted using highly polished sapphire wafers or highly polished
polycrystalline alumina substrates, showed that the introduction of a thin copper film improved the
strength characteristics relative those obtained using solid-state diffusion bonding with the same
time-temperature-load cycle. Results of these studies are summarized in Figure 7a for samples
bonded at 1400°C (6 h, 2.2 MPa). Experiments were repeated using as-ground alumina blocks from
! 14 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
the same source, but ground with a 400-grit-wheel. AFM measurements over 50 µm ¥ 50 µm areas
yielded R a values of 0.027 µm and 0.154 µm for the polished and as-ground surfaces, respectively.
Both values are significantly less than the copper film thicknesses examined. Profilometry line scans
over 50 µm lengths provided an R a of 0.011 µm for the polished surfaces and R a values of 0.213 µm
and 0.247 µm for the ground surfaces for scans parallel and perpendicular to the grinding direction,
respectively. However, as the length scale of the measurement was increased to dimensions more
comparable to the sample dimensions, 1 cm, significantly larger differences in the R a became
apparent. For the as-ground samples, scans parallel and perpendicular to the grinding direction
increased to 0.299 and 2.72 µm, respectively. Over the dimensional scale of the bonding surface, large
asperities and deep depressions, greater than 10 µm in height or depth relative to the mean surface
position, exist on the surface. The largest asperities will dictate the initial contact points and the
initial separation distances between the ceramic and the core layer. This, in turn, will influence the
pattern of liquid redistribution and film breakup. Although copper patches were noticeably larger
and occupied a larger area fraction of the interface when ground substrates were used, the resulting
mechanical properties, summarized in Figure 7b, are encouraging. There appears to be a slight
benefit to thicker films, perhaps reflecting more complete filling of the more severe interfacial gaps,
but the overall properties are not significantly degraded in comparison to those of comparable joints
made with polished substrates. The results suggest that the process is relatively forgiving, and useful
for joining as-machined parts.
Summary and Conclusions
TLP bonding provides an opportunity to join materials at reduced temperatures, which can
be essential to preserving the performance of materials with temperature-sensitive microstructures.
The results shown suggest that commercially available reactive-metal brazes coupled with lowmelting-point cladding layers could be used to form joints at temperatures that are more commonly
! 15 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
associated with soldering. TLP bonding relies on a substantial solid solubility of the low-meltingpoint component in a higher-melting-point host material, such as copper in platinum. This limits
the candidate alloy systems. If the interdiffusion process leading to liquid disappearance involves
substitutional diffusion, rather than the much more rapid interstitial diffusion (as for boron in
nickel), joining times can be substantial. To reduce the time required for isothermal solidification,
one must increase the bonding temperature (and diffusivity), or use thinner liquid films to reduce the
amount of liquid former that must diffuse into the adjoining bulk material or core layer. Increasing
the temperature counters the general desire to reduce processing temperature and protect
temperature-sensitive components. Reducing the liquid thickness imposes increasingly stringent
surface preparation demands since there is likely to be insufficient liquid to fill large gaps at the
interface. (Alternative methods of large gap filling rely on powder blends/slurries that more readily
conform to the local topography [23-25].)
These factors and considerations prompted efforts to devise and explore methods of joining
that would exploit a higher diffusivity transport path at lower temperature. What emerged was a
process that would exploit a liquid film, and rapid diffusion of the higher-melting-point metal within
the liquid to facilitate joint formation. Disappearance of the liquid would not be required, merely
rearrangement of the liquid phase so that the joint region no longer contained a continuous layer of
the liquid former. By dispersing the liquid former, melting would have a minimal effect on the hightemperature properties of the joint. This alternative approach opened up a much wider range of alloy
systems as interlayer candidates, ones in which the liquid former has minimal solubility in the host
material.
The Cu/Nb/Cu interlayer system has served as a vehicle for numerous studies of liquidfilm-assisted joining. The process can produce joints with reliably good properties, and hightemperature performance of the joints has been demonstrated. Assessments of phase diagrams
suggest that there are many other candidate interlayer systems, some which provide much greater
! 16 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
solubility of the core metal in the liquid. This could significantly increase the rate of ceramic–core
layer growth provided that the interfacial energetics are appropriate. In addition, hybrids of TLP and
LFAJ may also emerge.
Acknowledgements
This research was supported by the Director, Office of Science, Office of Basic Energy Sciences,
Division of Materials Science and Engineering, of the U.S. Department of Energy under Contract
No. DE-AC03-76SF00098. Takaya Akashi acknowledges support from a one-year fellowship from
the Japanese Ministry of Education, Culture, Sports, Science and Technology. Sung Hong
acknowledges support in the form of a Jane Lewis Fellowship. A. M. Glaeser wishes to thank the
Japan Society for the Promotion of Science for an Invitation Fellowship for Research in Japan
during which some of the ideas in this paper were formulated.
! 17 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
M. G. Nicholas and D. A. Mortimer, “Ceramic/metal joining for structural
applications,” Mater. Sci. Technol. (UK), 1, [9], 657-65 (1985).
R. E. Loehman and A. P. Tomsia, “Joining of ceramics,” Am. Ceram. Soc. Bull. (USA), 67,
[2], 375-80 (1988).
K. Suganuma, Y. Miyamoto, and M. Koizumi, “Joining of ceramics and metals,” Ann.
Rev. Mater. Sci., 18, 47-73 (1988).
G. Elssner and G. Petzow, “Metal/Ceramic Joining,” ISIJ International, 30, [12], 10111032 (1990).
D. Brandon and W. D. Kaplan, Joining Processes: An Introduction. New York:
John Wiley and Sons, 1997.
B. Derby and E. R. Wallach, “Theoretical model for diffusion bonding,” Met. Sci. (UK),
16, [1], 49-56 (1982).
B. Derby and E. R. Wallach, “Diffusion bonding: development of theoretical model,”
Met. Sci. (UK), 18, [9], 427-31 (1984).
C. A. M. Mulder and J. T. Klomp, “On the internal structure of Cu- and Pt-sapphire
interfaces,” Journal de Physique, 46, [Supplement C4], 111-16 (1985).
A. Hill and E. R. Wallach, “Modelling solid-state diffusion bonding,” Acta Metall., 37,
[9], 2425-37 (1989).
I. E. Reimanis, “Pore removal during diffusion bonding of Nb-Al2O3 interfaces,” Acta
Metall. et Mater., 40, [Supplement], S67-74 (1992).
B. J. Dalgleish, E. Saiz, A. P. Tomsia, R. M. Cannon, and R. O. Ritchie, “Interface
formation and strength in ceramic-metal systems,” Scr. Metall. et Mater., 31, [8], 1109-14
(1994).
H. K. D. H. Bhadeshia, “Joining of Commercial Aluminum Alloys,” in Proceedings of
International Conference on Aluminium (INCAL '03), S. Subramanian and D. H.
Sastry, Eds., 2003, pp. 195-204.
J. T. Klomp, “Bonding of metals to ceramics and glasses,” Am. Ceram. Soc. Bull. (USA), 51,
[9], 683-8 (1972).
J. T. Klomp and A. J. C. Van De Ven, “Parameters in solid-state bonding of metals to
oxide materials and the adherence of bonds,” J. Mater. Sci., 15, [10], 2483-8 (1980).
B. J. Dalgleish, M. C. Lu, and A. G. Evans, “The strength of ceramics bonded with
metals,” Acta Metall., 36, [8], 2029-35 (1988).
K. Suganuma, “New process for brazing ceramics utilizing squeeze casting,” J. Mater. Sci.,
26, [22], 6144-50 (1991).
D. S. Duvall, W. A. Owczarski, and D. F. Paulonis, “TLP Bonding: a New Method
for Joining Heat Resistant Alloys,” Welding Journal, 53, [4], 203-214 (1974).
H. Ikawa and Y. Nakao, “Transient Liquid Phase (T.L.P.) Diffusion Bonding of NickelBase Heat Resisting Alloys,” Transactions of the Japan Welding Society, 8, [1], 3-8 (1977).
H. Ikawa, Y. Nakao, and T. Isai, “Theoretical Considerations on the Metallurgical
Process in T.L.P. Bonding of Nickel-Base Superalloys,” Transactions of the Japan Welding
Society, 10, [1], 24-29 (1979).
! 18 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
J. D. Sugar et al.
Y. Nakao, K. Nishimoto, K. Shinozaki, and C. Kang, “Theoretical Research on
Transient Liquid Insert Metal Diffusion Bonding of Nickel Base Alloys,” Transactions of the
Japan Welding Society, 20, [1], 60-65 (1989).
W. D. MacDonald and T. W. Eagar, “Transient Liquid Phase Bonding,” Ann. Rev.
Mater. Sci., 22, 23-46 (1992).
Y. Zhou, W. F. Gale, and T. H. North, “Modelling of transient liquid phase bonding,”
International Materials Reviews, 40, [5], 181-96 (1995).
W. F. Gale, X. Wen, T. Zhou, and Y. Shen, “Microstructural development and
mechanical properties of wide gap and conventional transient liquid phase bonds between
Ti-48Al-2Cr-2Nb substrates,” Mater. Sci. Technol. (UK), 17, [11], 1423-1433 (2001).
W. F. Gale and X. Wen, “Ni3Al based composite interlayers for wide gap transient liquid
phase bonding of NiAl-Hf single crystals to a nickel base superalloy,” Mater. Sci. Technol.
(UK), 17, [4], 459-464 (2001).
W. F. Gale, D. A. Butts, M. Di Ruscio, and T. Zhou, “Microstructure and mechanical
properties of titanium aluminide wide-gap, transient liquid-phase bonds prepared using a
slurry-deposited composite interlayer,” Metallurgical and Materials Transactions a-Physical
Metallurgy and Materials Science, 33, [10], 3205-3214 (2002).
M. L. Shalz, B. J. Dalgleish, A. P. Tomsia, R. M. Cannon, and A. M. Glaeser,
“Ceramic joining III. Bonding of alumina via Cu/Nb/Cu interlayers,” J. Mater. Sci., 29, [14],
3678-90 (1994).
M. L. Shalz, B. J. Dalgleish, A. P. Tomsia, and A. M. Glaeser, “Ceramic joining. I.
Partial transient liquid-phase bonding of alumina via Cu/Pt interlayers,” J. Mater. Sci., 28,
[6], 1673-84 (1993).
M. L. Shalz, B. J. Dalgleish, A. P. Tomsia, and A. M. Glaeser, “Ceramic joining II.
Partial transient liquid-phase bonding of alumina via Cu/Ni/Cu multilayer interlayers,” J.
Mater. Sci., 29, [12], 3200-8 (1994).
R. A. Marks, J. D. Sugar, and A. M. Glaeser, “Ceramic joining. IV. Effects of
processing conditions on the properties of alumina joined via Cu/Nb/Cu interlayers,” J.
Mater. Sci., 36, [23], 5609-24 (2001).
B. J. Dalgleish, A. P. Tomsia, and A. M. Glaeser, “Transient Liquid Phase Bonding of
Silicon Ceramics via Microdesigned Ni-Based Interlayers,” Ceramic Transactions, 46, 555-566
(1994).
B. J. Dalgleish, A. P. Tomsia, and A. M. Glaeser, “A Transient FGM Approach to
Joining Si-Based Ceramics,” in Proceedings of FGM-94, N. Cherradi and B. Ilschner,
Eds. Lausanne: Presses polytechnique et universitaires romandes, 1995, pp. 503-508.
M. R. Locatelli, A. P. Tomsia, K. Nakashima, B. J. Dalgleish, and A. M. Glaeser,
“New strategies for joining ceramics for high-temperature applications,” Key Engin. Mater.,
111-112, 157-90 (1995).
G. Ceccone, M. G. Nicholas, S. D. Peteves, A. P. Tomsia, B. J. Dalgleish, and A.
M. Glaeser, “An evaluation of the partial transient liquid phase bonding of Si3N4 using Au
coated Ni-22Cr foils,” Acta Mater., 44, [2], 657-67 (1996).
K. Nakashima, K. Mori, and A. M. Glaeser, “Wettability of Cu-based Alloys on
Alumina and Joining of Alumina with Microdesigned Nickel-Chromium Alloy Interlayers,”
in Ceramic Microstructures: Control at the Atomic Level, A. P. Tomsia and A.
M. Glaeser, Eds. New York: Plenum Press, 1998, pp. 407-14.
! 19 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
J. D. Sugar et al.
K. Nakashima, T. Makino, K. Mori, and A. M. Glaeser, “Wettability and interfacial
reaction between Si3N4 and Cu-based alloys,” Key Engin. Mater., 159-160, 287-92 (1999).
B. J. Dalgleish, A. P. Tomsia, K. Nakashima, M. R. Locatelli, and A. M. Glaeser,
“Low temperature routes to joining ceramics for high-temperature applications,” Scr. Metall.
et Mater., 31, [8], 1043-8 (1994).
B. J. Dalgleish, K. Nakashima, M. R. Locatelli, A. P. Tomsia, and A. M. Glaeser,
“New Approaches to Joining Ceramics for High-Temperature Applications,” Ceramics
International, 23, [4], 313-22 (1997).
M. R. Locatelli, R. A. Marks, D. R. Chapman, K. Nakashima, and A. M. Glaeser,
“Transient Liquid Phase (TLP) Routes to the Joining of Structural Ceramics,” in
Interfacial Science in Ceramic Joining, A. Bellosi, T. Kosmac, and A. P. Tomsia,
Eds.: Kluwer Academic Publishers, Dordrecht, The Netherlands, 1998, pp. 111-34.
J. D. Sugar, J. T. McKeown, R. A. Marks, and A. M. Glaeser, “Liquid-film-assisted
formation of alumina/niobium interfaces,” J. Am. Ceram. Soc. (USA), 85, [10], 2523-30
(2002).
T. B. Massalski, H. Okamoto, and ASM International, Binary alloy phase
diagrams, 2nd ed. Materials Park, Ohio: ASM International, 1990.
K. Nogi, K. Oishi, and K. Ogino, “Wettability of solid oxides by liquid pure metals,”
Mater. Trans. JIM (Japan), 30, [2], 137-45 (1989).
P. Nikolopoulos, S. Agathopoulos, and A. Tsoga, “A method for the calculation of
interfacial energies in Al2O3 and ZrO2 liquid-metal and liquid-alloy systems,” J. Mater. Sci.,
29, [16], 4393-8 (1994).
R. A. Marks, D. R. Chapman, D. T. Danielson, and A. M. Glaeser, “Joining of
alumina via copper/niobium/copper interlayers,” Acta Mater., 48, [18-19], 4425-38 (2000).
J. J. Kruzic, R. A. Marks, M. Yoshiya, A. M. Glaeser, R. M. Cannon, and R. O.
Ritchie, “Fracture and fatigue behavior at ambient and elevated temperatures of alumina
bonded with copper/niobium/copper interlayers,” J. Am. Ceram. Soc. (USA), 85, [10], 253141 (2002).
! 20 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
Figure Captions
Figure 1
Schematic illustration of an assembly with a multilayer interlayer. Two
relatively thin cladding layers that form a liquid at low temperature flank a
thicker and higher melting point core layer that dominates the composition and
ultimate physical properties of the interlayer.
Figure 2
a) Schematic illustration of a thin liquid film sandwiched between a solid
metallic core layer and a ceramic on which the liquid has contact angles q1 and
q 2 , respectively. For parallel core layer and ceramic surfaces, void filling requires
that q1 +q 2 <180° . b) In LFAJ, at later stages, liquid-film-assisted growth of
ceramic-core contact produces a “dewetting” of the ceramic–core layer interface
and results in isolated droplets of liquid.
Figure 3
Optical micrographs taken of a sapphire/copper/niobium interface showing a)
initial, b) intermediate, and c) final stages of sapphire-niobium contact growth.
Note that in the final stage, sapphire-niobium contact dominates the interface
microstructure (ª90% area fraction), and therefore the thermomechanical
characteristics of the interface.
Figure 4
Room-temperature strength distributions for different interlayer designs used
to join alumina and silicon nitride. Cu/Pt, Cu/Ni, and Cu/80Ni20Cr
interlayers were used to bond alumina. Note the beneficial effect of Cr
additions. The Cu-Au-Ti/Ni interlayer was used to bond silicon nitride; the
two lines correspond to different joining temperatures.
Figure 5
Plots of fracture probability versus fracture strength for alumina joined using
In/Silver ABA/In interlayers. a) 20 min bonding cycle at 800°C, and
comparison to conventional reactive-metal brazing. b) Effect of bonding time
and temperature on strength and failure characteristics of TLP bonds.
Figure 6
Assessment of the “dewetting” kinetics for the copper film in LFAJ. At 1150°C,
the area fraction of sapphire-niobium contact increases rapidly and typically
exceeds 80% after 10-20 hours. At 1400°C, the area fraction of sapphireniobium contact exceeds 90% after just a few hours.
Figure 7
Plots of fracture probability versus fracture strength for alumina joined using
Cu/Nb/Cu interlayers. Results are for bonding cycles of 6 h at 1400°C using a)
polished substrates, and b) as-ground (400 grit) substrates. It is evident that the
liquid copper film is beneficial, especially for ground substrates.
! 21 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
Thin liquid-forming
cladding layer or film
J. D. Sugar et al.
Ceramic
material
Thick, high Tm
core layer
Thin liquid-forming
cladding layer or film
Ceramic
material
Figure 1
Schematic illustration of an assembly with a multilayer interlayer. Two
relatively thin cladding layers that form a liquid at low temperature flank a
thicker and higher melting point core layer that dominates the composition and
ultimate physical properties of the interlayer.
! 22 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
(a)
Figure 2
J. D. Sugar et al.
(b)
a) Schematic illustration of a thin liquid film sandwiched between a solid
metallic core layer and a ceramic on which the liquid has contact angles q1 and
q 2 , respectively. For parallel core layer and ceramic surfaces, void filling requires
that q1 +q 2 <180° . b) In LFAJ, at later stages, liquid-film-assisted growth of
ceramic-core contact produces a “dewetting” of the ceramic–core layer interface
and results in isolated droplets of liquid.
! 23 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
(a)
(b)
(c)
Figure 3
Optical micrographs taken of a sapphire/copper/niobium interface showing a)
initial, b) intermediate, and c) final stages of sapphire-niobium contact growth.
Note that in the final stage, sapphire-niobium contact dominates the interface
microstructure (ª90% area fraction), and therefore the thermomechanical
characteristics of the interface.
! 24 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
100
80
High-Strength
Alumina
Cu/80Ni20Cr
60
Failure Probability (%)
50
40
Cu/Pt
30
Cu/Ni
Cu-Au-Ti/Ni
Cu/80Ni20Cr
Low-Strength
Alumina
Silicon Nitride
20
10
40
Figure 4
50
60
80
100
200
300
Fracture Strength (MPa)
400
500 600
800 1000
Room-temperature strength distributions for different interlayer designs used
to join alumina and silicon nitride. Cu/Pt, Cu/Ni, and Cu/80Ni20Cr
interlayers were used to bond alumina. Note the beneficial effect of Cr
additions. The Cu-Au-Ti/Ni interlayer was used to bond silicon nitride; the
two lines correspond to different joining temperatures.
! 25 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
100
90
80
70
60
a)
Failure Probability (%)
50
J. D. Sugar et al.
TLP: In/Ag ABA/In, 20 min, 800°C
Brazing: Incusil ABA, 1000°C
Brazing: Silver ABA, 960°C
Reference: Polished 99.9% Alumina
40
30
20
Interfacial failures
10
10
100
90
80
70
60
b)
Failure Probability (%)
50
40
20
30
50
100
Strength (MPa)
200 300
500
200 300
500
In/Ag ABA, 1.5 h,700°C
In/Ag ABA, 6 h,700°C
In/Ag ABA, 24 h,700°C
In/Ag ABA, 1.5 h, 800°C
In/Ag ABA, 6 h, 800°C
In/Ag ABA, 24 h, 800°C
Incusil ABA,1000°C
Silver ABA, 960°C
Polished 99.9% Alumina
30
20
10
10
Figure 5
20
30
50
100
Strength (MPa)
Plots of fracture probability versus fracture strength for alumina joined using
In/Silver ABA/In interlayers. a) 20 min bonding cycle at 800°C, and
comparison to conventional reactive-metal brazing. b) Effect of bonding time
and temperature on strength and failure characteristics of TLP bonds.
! 26 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
a)
Area Fraction of Sapphire-Niobium Interface (%)
100
80
A=100(1-exp(-.045254t
60
40
0.78829
0.417899
A=100(1-exp(-.4062t
))
))
20
0.253385
A=100(1-exp(-1.0102t
))
0
0
50
100
150
200
Time (h)
Area Fraction of Sapphire-Niobium Contact (%)
100
b)
80
0.47525
A=100(1-exp(-1.3645t
))
60
0.71082
A=100(1-exp(-2.3082t
40
))
20
0
0
2
4
6
8
10
Time (h)
Figure 6
Assessment of the “dewetting” kinetics for the copper film in LFAJ. a) At
1150°C, the area fraction of sapphire-niobium contact increases rapidly and
typically exceeds 80% after 10-20 hours. b) At 1400°C, the area fraction of
sapphire-niobium contact exceeds 90% after just a few hours.
! 27 !
Transient Liquid Phase & Liquid-Film-Assisted Joining of Ceramics
J. D. Sugar et al.
100
80
a)
60
Failure Probability (%)
40
20
10
8
6
4
2
Reference 99.5% Polished Alumina
Diffusion Bond (No Cu)
1.4 µm Cu
3.0 µm Cu
5.5 µm Cu
1
70 80 90 100
200
Fracture Strength (MPa)
300
400
100
80
b)
60
Failure Probability (%)
40
20
10
8
6
4
2
1
40
Figure 7
Reference 99.5% Polished Alumina
Diffusion Bond (No Cu)
1.4 µm Cu
3 µm Cu
5.5 µm Cu
50 60 70 80 100
200
Fracture Strength (MPa)
300
400
Plots of fracture probability versus fracture strength for alumina joined using
Cu/Nb/Cu interlayers. Results are for bonding cycles of 6 h at 1400°C using a)
polished substrates, and b) as-ground (400 grit) substrates. It is evident that the
liquid copper film is beneficial, especially for ground substrates.
! 28 !