Note: Descriptions are shown in the official language in which they were submitted.
CA 02551256 1996-11-18
- 1 -
Tungsten-Copper Composite Powder
'f,ACK,~ROLTI~,
Tungsten-cor_:per (W-Cu) pseudoalloys are used in' the
manufacture of e:LE:arical contact materials and electrodes,
thermal management devices such as heat sinks and spreaders,
and COnduCtlVe inks and pastes for ceramic metallization_ The
basic methods for the fabrication of articles composed of W-Cu
pSeud~alloys include: infiltration of a porous tungsten
skeleton with liquid copper, hot pressing of blends of
tungsten and copper powders, and various techniques
incorporating liquid phase sintering, repressing, explosive
pressing, and the like. Complex shapes may be made by
injection molding W-Cu composite powders. It is desirable to
be able to manufacture articles made from W-Cu pseudoalloys at
or near the theoretical density of the pseudoalloy_ Besides
having improved mechanical properties, the higher density
pseudoalloys have higher thermal conductivities which are
critical for the application of W-Cu pseudoalloys as heat sink
materials for the. electronics industry.
' , The components in the W-Cu system exhibit only a very
3D small intersolubility. Thus, the integral densification of W-
Cu pseudcalloys occurs above 1063°C in the presence of liquid
copper. The compressive capillary pressure generated by the
CA 02551256 1996-11-18
r
- 2 -
forming and spreading of liquid copper, the lubrication of
tungsten particles by liquid copper and the minute solubility
of tungsten in copper above 1200°C combine to cause the
relative movement of tungsten particles during sintering and
thereby make possible the displacement of tungsten particles.
Local densification and rearrangement of the tungsten
framework causes an inhomogenous distribution of W and Cu
phases in the sintered article and copper bleedout, i.e, the
loss of copper from the sintered article. This leads.to the
degradation of the thermal/mechanical properties of the
sintered article.
Prior art methods directed to improving the homogeneity
of W-Cu composite powders by coating tungsten particles with
copper have not been successful as these copper-coated powders
still exhibit a high tendency towards copper bleedout during
the consolidation of the composite powder into fabricated
shapes.
Thus, it would be advantageous to eliminate copper
bleedout from occurring during the liquid-phase sintering of
W-Cu pseudoalloys while providing a homogeneous distribution
of W and Cu phases in the sintered article.
It is an object of the invention to obviate the
disadvantages of the prior art.
It is another object of the invention to produce a W-Cu
composite powder which can be used to make W-Cu pseudoalloys
having high electrical and thermal conductivities.
CA 02551256 1996-11-18
- 3 -
It is a further object of the invention to produce a W-Cu
composite powder which may be pressed and sintered to near
theoretical density without copper bleedout.
It is still a further object of the invention .to produce
a W-Cu composite powder which may be used to make sintered
articles having a high degree of dimensional control.
In accordance with one object the of invention, there is
provided a tungsten-copper composite powder comprising
individual particles having a tungsten phase and a copper
phase wherein the tungsten phase substantially encapsulates
the copper phase.
In accordance with another object of the invention, there
is provided a W-Cu composite oxide powder comprising
individual particles having a copper tungstate phase and
tungsten trioxide phase wherein the tungsten trioxide phase
exists primarily at the surface of the individual particles.
In accordance with a further object of the invention,
there is provided a method for forming a homogeneous W-Cu
pseudoalloy comprising pressing a tungsten-coated copper
composite powder to form a compact and sintering the compact.
In accordance with a still further object of the
invention, there is provided a W-Cu pseudoalloy having a
microstructural cross-section having tungsten areas and copper
areas, the tungsten areas being less than about 5 ~m in size
and the copper areas being less than about 10 pm in size.
CA 02551256 1996-11-18
- 4 -
SRIEF DESCRIPTIONf OF THE DRAWINGS
Fig. 1 is a photomicrograph of across-section of a
tungsten-coated copper composite powder containing 25 weight
percent copper.
Fig 2 is a photomicrograph of a cross-section of a
tungsten-coated copper composite powder containing 15 weight
percent copper.
Fig. 3 is an illustrative view of a cross-section of a
tungsten-coated copper composite particle wherein the tungsten
phase substantially encapsulates the copper phase.
Fig. 4 is an illustrative view of a cross-section of a
tungsten-coated copper composite particle wherein the tungsten
phase completely encapsulates the copper phase.
Fig. 5 is an illustrative view of a cross-section of a
tungsten-coated copper composite particle having a dendritic
morphology wherein the tungsten phase substantially
encapsulates the copper phase.
Fig. 6 is a graphical illustration of the relationship
between the molar ratio of tungsten trioxide to copper
tungstate and the X-ray Diffraction peak intensity ratio of
tungsten trioxide (3.65 ~) to copper tungstate (2.96 ~) for
synthesized W-Cu composite oxides and mechanical mixtures of
copper tungstate and tungsten trioxide.
Fig. 7 is a graphical representation of the relationship
between the compacting pressure and the electrical
CA 02551256 1996-11-18
- 5 -
conductivity at different sintering temperatures for a W-Cu
pseudoalloy made from a tungsten-coated copper composite
powder having 15 weight percent copper.
Fig. 8 is a graphical representation of the relationship
between the compacting pressure and the electrical
conductivity of W-Cu pseudoalloys made from tungsten-coated
copper composite powders containing varying amounts of copper.
. Fig. 9 is a photomicrograph of a cross-section of a W-Cu
pseudoalloy having 15 wt.% Cu which was made from a milled and
spray dried tungsten-coated copper composite powder.
For a better understanding of the present invention,
together with other and further objects, advantages and
capabilities thereof, reference is made to the following
disclosure and appended claims taken in conjunction with the
above-described drawings.
Several factors influence the solid-state (below 1083°C -
the melting point of copper) and liquid-phase (above the
melting point of copper) sintering behavior of submicron W-Cu
powder systems. Compacted refractory metal powders undergo
considerable microstructural changes and shrinkage during
solid-state sintering (in the absence of liquid phase).
Submicron particle size powders effectively recrystallize and
sinter at temperatures (T) which are much lower than the
melting temperatures (Tm) of refractory metals (T = 0.3 Tm) .
The initial sintering temperature for submicron (0.09-0.16 Vim)
tungsten powder is in the range of 900-1000°C. The spreading of
CA 02551256 1996-11-18
- 6 -
copper and the formation of a monolayer copper coating on
tungsten particles occurs in the temperature range of 1000-
1083°C. By lowering the activation energy for tungsten
diffusion, monolayer copper coatings activate the solid-state
sintering of tungsten. Therefore, a number of complementary
conditions are met for bonding submicron tungsten particles
into a rigid tungsten framework within the composite powder
compact during solid-state sintering (950-1080°C). High
fineness and homogeneity of the starting composite powders are
expected to enhance the sintering of a structurally
homogeneous tungsten framework. Such framework should, in
turn, aid in making a homogeneous pseudoalloy.
Such an idealized mechanism is complicated by local
densification. Due to nonuniform distribution of stresses at
compaction of the starting powder, submicron tungsten
particles may experience rapid densification in local regions.
An unevenly sintered tungsten framework will cause structural
inhomogeneity, localization of integral densification, and
copper bleedout at the liquid-phase sintering stage. Liquid-
phase sintering comprises three stages: particle
rearrangement, grain growth by a dissolution-reprecipitation
mechanism, and the formation and densification of a rigid
skeleton. The operating mechanism for integral densification
in W-Cu systems consists of particle rearrangement with the
rate of densification inversely proportional to particle size
and grain shape accommodation for solid particles aided by
minute solubility of tungsten in copper at 1200°C and above.
Integral densification of W-Cu pseudoalloys strongly
depends on capillary pressure generated by the liquid copper.
This pressure increases with the reduction-of the liquid phase
CA 02551256 1996-11-18
_ 7
wetting angle. The wettability of tungsten by copper improves
with temperature. At 1100°C the wetting angle is already
substantially below 90°, and it steadily diminishes with
temperature to a value close to zero at 1350°C. With
improvement of the wettability, the dihedral angles between
adjacent tungsten particles attain values necessary for the
liquid to penetrate the space between the particles and force
them to slide relative to each other. The rate and extent of
integral densification and rearrangement of the tungsten
framework are also affected by the amount of liquid phase and
the size of tungsten particles.
We have discovered that a homogeneous sintered W-Cu
pseudoalloy article can be formed without copper bleedout by
using a W-Cu composite powder composed of tungsten-coated
copper composite particles. Because the individual particles
have a tungsten outer layer and a copper core, W-W contacts
dominate during compaction of the composite powder. This
results in forming a network of interconnected submicron
tungsten particles within the powder compact immediately after
powder consolidation. Solid-state sintering of the powder
compact results in a structurally homogeneous tungsten
framework which enables uniform internal infiltration of the
framework with copper during liquid-phase sintering and the
formation of a dense pseudoalloy with a continuous
tungsten/copper structure. As used herein, the term "tungsten-
coated" or "W-coated" means that the tungsten phase
_ substantially encapsulates the copper phase. Unless otherwise
indicated, all of the W-Cu composite powders discussed below
are composed of tungsten-coated copper particles.
CA 02551256 1996-11-18
The preferred W-Cu composite powder has a Fisher Sub-
Sieve Sizer (FSSS) particle size range of about 0.5 ~m to
about 2.0 ~m in the copper content range of 2 to 25 wt.%.
Each composite particle has a tungsten coating whose average
thickness varies with the size and shape of the particle and
ranges between about 0.1 to about 0.2 Vim.
Figs. 1 and 2 are Back-scattered Electron Images (BEI) of
cross-sections of tungsten-coated copper powders containing 25
and 15 wt. % copper, respectively. The BEI photomicrographs
clearly show that the tungsten phase which appears as white
has substantially encapsulated the copper phase which appears
as gray. The black areas between the particles are from the
epoxy mounting material. Figa. 3, 4, and 5 are illustrations
of cross-sections of individual tungsten-coated .copper
composite particles. Fig. 3 shows the tungsten phase 10
substantially encapsulating the copper phase 12. Fig. 4 shows
a particle where the W phase 10 has completely encapsulated
the Cu phase 12. Fig. 5 shows a particle having a dendritic
morphology wherein the W phase 10 has substantially
encapsulated the Cu phase 12. At least 50% of the surface of
the Cu phase is encapsulated by the W phase. Preferably, the
tungsten-coated copper composite particles have at least 70%
of the surface of the copper phase encapsulated by the w
phase. More preferably, at least 90% of the surface of the Cu
phase is encapsulated by the W phase.
A preferred method for forming the tungsten-coated copper
composite particles involves the hydrogen reduction of a W-Cu
composite oxide powder containing CuW04 and W03. Such a
composite oxide may be made by reacting an ammonium tungstate,
such as ammonium paratungstate (APT) or ammonium metatungstate
CA 02551256 1996-11-18
_ 9 _
(AMT) , with an oxide cr hydro~~ide of copper. Methods for
forming W-Cu composite oxides are described in U.S, patents
Nos. 5,468,457 and 5,470,549, both from applications filed
X2/22/94.
In order to form an excess of W03 in the composite oxide,
the amount of ammonium tungstate used is greater than the
stoichiometric amount which would be required to form copper
tungstate (CuW04) . 5uch a reaction forms an intimate mixture of
CuW04 and W03 phases having a relative copper content less than.
about 25 percent, i.e. relative to tungsten (CuW04 has a
relative copper content of 25.70). A homogeneous W-Cu
composite oxide powder having a composite-or_ of CuWO~ + ( D _ 035-
15) W03 and a relative copper content- of about 25-°s to about 2%
is the preferred material to make the tungster_-coated copper
composite powder. Examples of the synthesis of W-Cu composite
oxides having variable copper content are given in Table 1.
CA 02551256 1996-11-18
- 10 -
Reaction Relative Copper Content
in W-Cu Composite Oxide
0.547 AMT + 0.5 CuzO + 0.25Oz.-+ 5.0 %
CuW04+5.566 WO~
0.259 AMT + 0.5 Cu~O + 0.25OZ-+ 10.0 %
CuW04 + 2.11 W03
0.163 AMT + 0.5 CuZO + 0.25OZ-+ 15.0 %
CuW04 + 0 . 9 5 8 W03
0.115 AMT + 0.5 CuZO + 0.25OZ-+ 20.0 %
CuW04 + 0 . 3 8 2 W03
0.083 AMT + 0.5 Cu20 --~CuWO,,25.7 %
+ 0.25 OZ
It is believed that the encapsulation of the copper phase
by the tungsten phase starts at the composite oxide synthesis
stage with the formation of a WO~-coated CuWO, core within
every discrete particle of the W-Cu composite oxide powder. It
has been observed that reactions of ammonium tungstatea with
oxides of copper produce composite oxide powders having
particle morphologies which mimic the morphology of the
ammonium tunsgtate particles. This is similar to the effect
observed in the production of tungsten powder by hydrogen
reduction of Wo3. There, the morphology of the tungsten powder
is controlled by the morphology of the W03 which, in turn, is
strongly influenced by the morphology of the ammonium
tungstate powders from which the WO3 is made.
CA 02551256 1996-11-18
- 11 -
I: W-Cu COMPOSITE OXIDE PRECURSOR
W-Cu composite oxide powders having a composition of CuW04
+ 0.958 W03 were synthesized from (1) CuzO + precipitated APT
(angular particle morphology) and (2) Cu20 + spray-dried AMT
(spherical particle morphology). Both APT and AMT powders were
converted to composite oxide particle agglomerates which were
isomorphic with the starting particles of APT and AMT. The
particles of cuprous oxide totally disappeared during
synthesis of the composite oxide. It was not possible to
identify any effect of the starting morphology of Cu20 on the
morphology of the composite oxide powder. The small, mostly
submicron, grains forming each particle of the W-Cu composite
oxide suggest that W03 grains outline the diffusion boundaries
for the solid-phase reaction. It is believed that the
reaction mechanism consists of the diffusion of mobile
molecules of copper oxide into W03 grains. The ability to
control the W-Cu composite oxide morphology by varying the
morphology of ammonium tungstates used in synthesis can be
used to influence the kinetics of the hydrogen reduction of
the oxides and the morphology of co-reduced W-Cu composite
powders.
W-Cu composite oxides having a composition of CuW04 + nW03
(n = 0.035-15) and a relative copper content in the range of
25 to 2% have been synthesized from CuzO and spray-dried AMT.
' A number of these composite oxides and stoichiometric copper
tungstate formed by the same method (25.7 to 5% Cu, n=0-5.566)
were subjected to elemental analysis by Energy Dispersive X-
ray Spectroscopy (EDS) and to phase analysis by X-Ray
Diffraction (XRD). The EDS analysis included measurement of
W-La to Cu-Ka peak intensity ratios from which W/Cu atomic
CA 02551256 1996-11-18
- 12 -
ratios for the different oxide compositions were calculated
for single particles per field and groups of 2 to 5 particles
of different sizes per field. The analytical results for
populations of single particles are given in Table 2. A good
correlation between experimental and theoretical W/Cu atomic
ratios were found for all analyzed particle sizes of the
stoichiometric CuWO, (n=0) composite oxide. However, for the
single-size particle populations corresponding to the CuW04 +
nW03 (n=0.382- 5.566) composite oxides, the experimental W/Cu
atomic ratio deviated significantly from the theoretical W/Cu
atomic ratio. Table 2 shows that the size of the deviation
grows with increasing WO~ content and is especially pronounced
for 1 ~m particles. Tungsten, copper, and oxygen peaks were
present in every analyzed particle which confirmed that the
particles formed in the course of solid-phase synthesis are W-
Cu composite oxides as opposed to mechanical mixtures of CuW04
and W03.
Theoreti cal Data Experimental
Average
W/Cu
RelativeW/CU Atomic
Oxide Copper AtomicRatio
for
Populations
With
a
Particle
Size
of,
~tm
Composition Content,Ratio 1 5 15 30
t
CuWO, 25.7 1.000 0.963 0.957 0.956 0.949
CuWO~ + 0.382 20 1.392 1.553 1.476 1.977 1.502
WO~
CuWO, + 0.958 15 1.958 3.649 2.629 3.351 3.042
WO~
CuWO~ + 2.11 10 3.110 13.9097.203 7.677 3.672
WO~
CuWO~ + 5.566 5 6.566 22.3389.779 8.538 10.738
WO~
CA 02551256 1996-11-18
- 13 -
A better correlation between theoretical and experimental
W/Cu atomic ratios is found when 2 to 5 particles of different
sizes are analyzed. Thus, there is an improvement in material
homogeneity with an increase in the number of analyzed
particles. The homogeneity of populations of particles can be
expressed as a coefficient of variation which is defined as
the ratio of the population standard deviation to its mean
expressed in. percent. Table 3 gives the coefficients of W/Cu
atomic ratio variation for populations of single 1 ~m
particles (Column A) and populations of 2 to 5 particles of
different sizes (Column B).
W-Cu Composite oxideA B
Composition
CuW04 3.0 4.9
CuW04 + 0.382 W03 122.9 12.3
CuW04 + 0.958 W03 141.4 43.2
CuW04 + 2.11 W03 82.6 30.9
CuW04 + 5.566 W03 29.9 17.0
The deviation from CuW04 stoichiometry results in a
reduction of the W-Cu composite oxide homogeneity for particle
sizes of about 1 ~m (Column A). Homogeneity for the single 1
~tm particles improves with increasing amounts of W03. The
effect is less pronounced when larger particles or larger
number of particles are analyzed (Column B). The variations
in homogeneity disappear entirely when thousands of particles
are examined simultaneously as with Sputtered Neutral Mass
Spectrometry (SLAMS) analysis.
CA 02551256 1996-11-18
- 14 -
XRD phase analysis of synthesized composites were
compared to XRD standards made of mechanical blends of CuW04
and W03 powders containing the same proportions of the two
phases. Fig. 6 shows that the W03/CuW04 peak intensity ratio is
consistently greater for the synthesized CuW04 + nW03 (n=0.382-
5.566) composite oxides compared to the corresponding
mechanical blends.
. The EDS and XRD results for the CuW04 + nW03 (n=0.382-
5.566) composite oxides indicate that fine submicron Cuw09
particles are being surrounded by W03 particles within 1 p.m
particle agglomerates. It is believed that this encapsulation
of the CuW04 phase by the W03 phase results in the formation of
finely dispersed tungsten-coated copper particles during
hydrogen reduction of the composite oxide.
Forming the W-Cu composite oxides from ammonium
tungstates and oxides or hydroxides of copper yields
additional benefits. The high surface areas and additional
reactivity generated from the thermal decomposition of these
reactants improves the degree of mixing, shortens the
diffusion distances, and results in sufficient diffusion
activities to promote synthesis reactions in every discrete
grain of W03 which is produced by in-situ thermal decomposition
of the ammonium tungstate. This allows the use of a broad
range of particle sizes for the solid reactants which makes
the process quite forgiving.
To illustrate this, CuW04 + 0.958 W03 composite oxides
(15% Cu) were synthesized from mechanical blends of spray-
dried .AMT and Cu20. Three feedstocks (A, B, C) for hydrogen
CA 02551256 1996-11-18
- 15 -
TCdllCt1D27 WGTE S~JSlt].leEiCeC~ lWi~lTIC~. ~ rWa1'1 j% ~.' t0 1 ZTC'~1'1~t-
G1i 1Ii
~.rlE medlars ~..~~~!'~' pal'tiC~ E S1~E ~rlC1 ~ rtEal~~ ~- 15 tG !.
UG1"_at101i 1r1
Lhe median Cu=D l~a_ tide site . Concomi rant v~ittu tl~e si ce
variation, there was a cGrres=~onding substant-_al ~..%ar~iat~~ ors of
the surface area of solid reactants. The su=dace areas and
particle sizes I9~tr~, Seth, and lDth percentiles as detei-a-nined
by Microtrac*~ are given in Table 4.
* firademarlc
CA 02551256 1996-11-18
- 16 -
U
U
c0 ri N
~ r~ ~ d'
- .
V H .
y O O O,
~ ~
~
b ~
k m
O
o r m ~n
q m m o
o ~ N
ri G
U
OD N I'1
y O
A tt1M d'
r1 ra r~
N N
rd
p o w r
~
, . o
A ' r
q ow
N N N
U
rl
m m O
W l 1 m
~ ~
-- r r
w
~ ~ ~E o o N
a
W
W
U
N
a
0 o r r N
0
A w ,r o
~
W .
i
W
V .,~ m n
~ o
d~
U ~ Ca -1
'
i~
.
PW 01 01 N
rI O
A m ~ y o
A
N N
U
W V ~ d' 01 Of
W N rt a d~
n
E o 0 0
W
w
N
U
H
a
o
_
~,
w
ar
... N
' ~ r~ w a
N o
V
N .-1r-1
rl
H N
p,
d' VI N
p o
~ rr o 0
W c d~ a
W W W in E-~ Et tt1V
A O U aG
CA 02551256 1996-11-18
- 17 -
Within the broad range of reactant particle sizes and
surface areas, there is exhibited a strong trend toward
equalization of these values in the synthesized composite
oxide feedstocks. It is believed that the isomorphism between
the starting AMT powders and the synthesized composite oxide
powders is accountable for this equalization. The synthesized
composite oxides were subjected to XRD phase analysis and EDS
elemental analysis. The measured W03/CuW04 phase ratio and the
W/Cu atomic ratio in all three feedstocks were equivalent to
the corresponding ratios established for the CuW04 + 0.958 Wo3
composite in Tables 2 and 3. Thus, the phase distribution and
homogeneity of W-Cu composite oxides is consistent within
large ranges of reactant particle sizes. W-Cu composite-oxides
having median particle sizes from about 5 um to about 25 ~m
may be produced from mechanical blends of ammonium tungstates
having median particle sizes from about 5 ~m to about 100 ~tm
and oxides/hydroxides of copper having median particle sizes
of about 0.5 ~tm to about 20 Vim.
II. HYDROGEN REDUCTION OF COMPOSITE OXIDES
Co-reduced W-Cu composite powders made from the above
W-Cu composite oxide mixtures, CuW04 + nW03 (n>0), exhibit the
gray color which is characteristic of the color of freshly
reduced tungsten powder. No indication of the presence of
copper is observed. This is consistent with the formation of
W-Cu pseudoalloy particles wherein the copper phase has been
- substantially encapsulated by the tungsten phase.
The reduction of CuW04 begins with the reduction of copper
followed by stepwise reduction of tungsten:
CA 02551256 1996-11-18
- 18 -
CuW09 --> XCu + Cul_xW04-x --~ Cu + (W03 --~ WOZ . g --~ WOz . ~z -~ WOZ -->
W)
The reduction of copper from CuW04 has a very high
thermodynamic probability and can be completed in the
temperature range of 305°C to 488°C.
Substantially higher temperatures are required for
tungsten reduction from W03. For example, it is standard
practice to reduce W03 in the temperature range of 840°C to
900°C. In case of CuW04, the presence of copper lowers the W03
reduction temperature. For example, in the presence of copper,
W03 may be reduced to tungsten via the stepwise reduction of
tungsten suboxides between 661°C to 750°C:
The discrepancy between temperatures needed to reduce
copper from CuW04 (300°C to 400°C) and final stages of tungsten
reduction (750°C to 800°C) is significant and results in
segregation of prematurely reduced copper. Attempts to bring
these temperatures closer by lowering the tungsten reduction
temperatures may trigger another undesirable effect. Stable,
nonpyrophoric ac-W is formed only by the stepwise suboxide
reduction sequence at temperatures well above 500°C to 550°C.
Below these temperatures, a pyrophoric, unstable ~i-W, or a W30
phase, is formed as a result of skipping the suboxide sequence
and going through a W03 -~ W30 -~ W transition. Thus, the
catalytic effect caused by the presence of copper may promote
the formation of pyrophoric (3-W.
In order to circumvent copper segregation and the
formation of pyrophoric ~i-W while achieving a homogeneous
CA 02551256 1996-11-18
- 19 -
distribution of Cu and W phases in the co-reduced powder, it
has been suggested to reduce CuW09 at once at rathe r high
temperatures (about 7oD°C). However, this would be difficult
to implement under industrial hydrogen reduction conditions
where boats containing the composite oxide material ha~.re to
pass through a temperature transition zone before reaching the
final isothermal zone where the reduction would occur.
It has been found that the W03-coated CuW04 particle phase
1D distribution in the W-Cu composite oxides dramatically
influences the reduction kinetics of copper from the composite
oxide compared to its reduction from CuW04. Instead of the
300°C to 400°C temperature range, the appearance of copper is
shifted to a much more favorable range of 55D°C to 700°C. Thzs
i5 leaves less time for copper segregation prior to appearance of
tungsten. ?additionally, the catalytic effect of the presence
of copper metal phase lowers the W03 reduction temperature ~~o
between 700°C to 850°C. The copper particles formed during 'the
reduction of the composite oxide serve as sites f or deposition
20 of tungsten thereby controlling the phase distribution and
size of W-Cu composite powder. The encapsulation of the
copper~articles by tungsten prevents further segregation of
the copper.
25 In the tests on hydrogen reduction of CuW09 + nW03 (n>0)
composite oxides, Inconel* boats with composite oxide boatloads
of BO to 160 grams and bed depths of 3/B" to 3/4" were used.
Hydrogen flow ratES may be from about 20 cm/sec to about 300
cm/sec. The tests were carried out in a laboratory tube
30 furnace having thermal zones for gradual temperature increase,
isothermal hold, and cooling down to ambient temperature. The
* Trademark
CA 02551256 1996-11-18
- 20 -
reduction kinetics of the composite oxides was studied
throughout the stages of temperature increase and isothermal
hold of the reduction cycle. The rate of temperature increase
from ambient temperature to the isothermal hold temperature
varied within a broad range (5°C to 20°C/minute) characteristic
of rates used in industrial reduction furnaces. Reduction
parameters - hydro.gen flow rate and velocity, rate of
temperature increase from ambient to isothermal hold,
reduction temperature (isothermal hold), length of reduction -
could be controlled for close simulation of the industrial
reduction conditions. Table~5 illustrates the hydrogen
reduction kinetics of the CuWO~ + 0.958 W03 composite oxide
(Feedstock A, Table 4). The hydrogen reduction was performed
using bed depths of 3/8" and dry hydrogen with a dew point of
-60°C. XRD, Optical Microscopy (OM) and Back-scattered Electron
Imaging (BEI) were used to analyze the material at different
stages in the reduction process. No phase transition could be
detected by XRD, OM or BEI until the composite oxide was
exposed to an average temperature of 542°C across the boat
length. An increase in surface area of the material and
appearance of the CuW03.19 phase indicated the onset of copper
reduction which was confirmed by OM. BEI photomicrographs
showed the deposition of copper along the grain boundaries of
oxides.
CA 02551256 1996-11-18
- 21 -
U
0
0
0
0
td
N
E~
N
U
G
v
'
U 'i-i d
'
O
~
in ,n ~ ~ r~ t~ o in
-~ v
v
O
o t~ m n ~ o W'
~
U N W o co ov ov
p q
'd
op
E
a~
U
O ~ o
U M
+ -
II
U
~ ao m co.a~ ~ N m rn
U r..l~ r~ o~ dm n c~ o~ ~ ~
~
O V o o r-1.-i,-I o 0 0
NE
O
4 U ~ o
U7 U N
U
w
x v o v
z v ~ b
o ~ , v
U G
~ 3 3 ~
v O ~ p O
'p
3 U 3
U U U
O O h
O O O O O
U U 3 3 ~ 3 3 U
w
v N v N O
dl trl M d~
>N ~I ~1 ~1 ~ v N td
* v N N
v ~n y > > ~ ~ ~ ~, x
* ~ ~ ~
N ~ ~ ~ ~ ' N v N
E !~ FC ~
U t~ ~L
. . . , , , rd
N N O ~ ~ ~
O O O S-1
~ O ~ O v
1.1 Ln lD t~ O O a~
O a0
* O
* lD
CA 02551256 1996-11-18
- 22 -
At an average temperature of 682°C, the reduction of
copper is close to completion. A honeycomb of copper
deposited along the tungsten oxide grain boundaries was
observed by OM. BEI clearly shows a submicron network of
interconnected copper veins and a change in particle
morphology from rounded to dendritic caused by the copper
phase change and partial loss of oxygen. A WOZ,.,2 phase is
observed by XRD showing that the reduction of W03 is proceeding
via the W03->W0z,9~WOz.~z-~WOz->W sequence.
At an average temperature of 768°C, the material's surf ace
area reaches its peak value probably because of the completion
of the W03--~WOZ transition. XRD begins to detect the presence
of tungsten. OM and BEI analysis find signs of both copper
coalescence and tungsten deposition on copper.
The dominating processes throughout the 800°C isothermal
hold in the reduction cycle are: (1) a decrease in the
material's surface area related to an increase in the degree
.of reduction; (2) the coalescence of copper; (3) W02 reduction
and deposition of tungsten on copper surfaces leading to
encapsulation of copper particles; and, (4) control of the
size and morphology of the W-Cu composite particles by the
size and morphology of the copper particles.
The resultant W-Cu composite powder consists of
irregularly shaped particles having a phase distribution
- wherein the tungsten phase substantially encapulates the
copper phase. Based on BEI and SLAMS analyses, the thickness of
tungsten coating varies with the size and shape of the
CA 02551256 1996-11-18
- 23 -
composite particles and is estimated to range between 0.1 and
0.2 ~.m.
Copper plays an important role in controlling the size of
the copper core and the overall size of the W-Cu composite
particle. It was found that the particle size of as-reduced W-
Cu composite powder increases with copper content and ranges
from about 0.5 um to about 2.0 um (FSSS) as the copper content
increases from 2 to 25 wt.%.
Table 6 shows the effect that reduction conditions have
on the particle size of the W-Cu composite powders. CuW04 +
0.958 W03 feedstocks A, B, and C (15% Cu) and a CuW04 + 0.035
W03 feedstock D (25% Cu) were reduced under identical
conditions (800°C for 1 h. and 28 cm/sec H2). The FSSS
particle size ranged from 1.52 to 1.75 ~tm for the unmilled
W-Cu composite powders having 15 wt.% Cu (W-15%Cu). The W-Cu
composite powder having 25 wt.% Cu (W-25%Cu) had a slightly
higher FSSS particle size of 1.8 Vim. BEI analysis of the
W-25%Cu powder shows larger, less interconnected particles
compared to W-15%Cu powder.
Feedstock A was reduced at hydrogen flow velocities of 28
and 80 cm/second. A 2.86 times increase in hydrogen flow
velocity from 28 to 80 cm/sec (800°C, 1 h:) caused a decrease
in the FSSS particle size of the resultant W-Cu composite
powder made from feedstock A from 1.75 ~m to 0.96 Vim. This is
similar to the effect observed in the manufacture of tungsten
powders.
CA 02551256 1996-11-18
- 24 -
Feedstocks A, B, C were reduced at two different
temperatures, 800°C and 750°C. The FSSS particle size of the
reduced unmilled powder was decreased about 1.1 to about 1.4
times by a 50°C decrease in reduction temperature. As a
consequence of size reduction, the powder surface areas and,
correspondingly, the oxygen content are increased. The reduced
powders in Table 6 all had oxygen contents less than about
5000 ppm. For W-Cu composite powders reduced above about
850°C, excessive copper coagulation was observed. For powders
l0 reduced below about 700°C, the powders exhibited pyrophoricity
because they were underreduced, had high surface areas, and
had oxygen contents above 5000 ppm. Thus, it is preferable to
reduce the W-Cu composite oxides between about 700°C to about
850°C
CA 02551256 1996-11-18
- 25 -
0 0 0 0 0 0 0 0 0 0
O O O N ~ O O O O O
b d~ l0 ~D m n m ~ W N M
rW -i '-1 ri r~ f~1r~
Q1
N O n 01 rl N Q1 W t11~D
d~ r1 N 1D n h m ~D ~O V1
r1 rl rl O O O O '-1 M O
U7 N U1 m V) N m i ~ y f1
(n n n n rd O ~-1~ ~ ~ N
W O O O ri '-1 r-1~ ~ ~ e-i
.
b O V1 lf1d~ lD l0 lp 1 ~ ~ OD
~ ~ ~
,..i p o 0 0 0 0 0 ~ ~ ~ o
o m n o ~ ~ m ~ ~ m u
C7 rl ri .-1N ri ~ ~ ~ ~ N
1J
O ,-i l(1lD O m O ~ ~ y f1
JJ d' O ~ ~ 1 ~
p t0 1D N ~ n ~"~~ ~ ~ n
N
p
tl1N n n tf1O N l0 tf1 d' O
N U7 d~ N N n n W 01 m n m
~rl X17
tp !~ e-I rl ri rl .~i r-1O O O rl
N
U
O l0 Q1 d p O ~ m V1 O 01
p rl ri r1 ~ f'1 N N rl M N
LL r-I
~ O e-1
Gi O O N tf1l0 d' ~ N e-1 r~
,.~7 ~ .
p m aw o o ~ ~ n o~
0 0, m ,~ ~ ~ ~ o ,~ m m
~ ~, m ~ . p~ m n
p N N e-IN N ri N N N t1
G) 1d tf1 tf1lf1 O tf1 O
b! ,..~~ ,..~
~ r~1 O .-1 rl ri '-I.-i N
N ~
N
B
ro
w ~ ~ ~ ~
o w m ao m m m m m o 0 o m
p E N N N N N N m m m N
x N U
.b
4!
V o 0 0 0 0 0 0 0 o 0
o u~ ~n u~ 0 0 0 o w o 0
n n n m m m m n n m
Er
v
V .0 -~ ~ CG U ~E C1 U r.Cd ~ p
~ E o
O
U
CA 02551256 1996-11-18
- 26 -
Composite metal powders consisting of two or more metal
phases are more prone to oxidation and pyrophoricity than
powders of a single metal. Passivation of the reduced W-Cu
composite powders with nitrogen immediately after reduction
dramatically decreases the powders tendency to oxidize and
become pyrophoric. As an example, Feedstock A which was
reduced at 700°C using a hydrogen flow velocity of 80 cm/second
had a submicron FSSS (0.74 ~,m), a high surface area and an
oxygen content of 3200 ppm. After having been exposed to air
for about 2 hours, the reduced powder became pyrophoric. When
a similarly reduced powder was placed under nitrogen and
passivated for about 2 hours, the reduced powder did not show
any signs of pyrophoricity after 24 hours of exposure to air.
III. CONSOLIDATION OF W-Cu COMPOSITE POWDERS
The high surface area of the W-Cu composite powders
increases the oxygen content in the powder compact. Surface
oxides on the tungsten-copper interface present a significant
problem in manufacturing a dense pseudoalloy with a continuous
W-Cu structure. It has been established that at 1100-1200°C
oxygen-saturated copper does not wet tungsten. Oxygen-free
copper also does not wet tungsten having surface oxides in the
form of W03 or [WOZ (OH) 2] on tungsten-copper interface . Cleaning
oxygen from W-Cu powder compacts presents a fundamental
problem in manufacturing the pseudoalloys. This problem is
complicated by the "hydrogen disease" of copper and a high
tungsten affinity for oxygen.
The data in Table 7 illustrate the problem of "hydrogen
disease" in copper.
CA 02551256 1996-11-18
- 27 -
Temperature, 1065 1100
C
Form of Cu + 3.5~t Cu~O Cu + 0.54 Cuso or
les s
Oxygen Existence (eutectic) after hydrogen cleaning
Concentration,- 0.39 0.06 or less
t
Effect Swelling in hydrogen.No swelling in
No wetting of hydrogen.
tungsten. Good wetting of
tungsten.
Water molecules formed during hydrogen reduction of
trapped oxygen cause swelling of copper and W-Cu compacts.
The compacts must have 15 to 20% porosity to allow free escape
of water molecules and good hydrogen cleaning of oxygen.
However, this does not completely solve the problem. In the
presence of water vapor, complex iWO2(OH)~] molecules form on
the surface of tungsten. Therefore, forcing oxygen out of the
W-Cu system and maintaining a clean interface between the
metal phases presents a complex metallurgical problem.
Critical applications in electrical/electronic
engineering require W-Cu pseudoalloys with superior structural
homogeneity and performance under increased thermal and
mechanical stresses. For metals, one of the moat fundamental
physical properties is the electrical conductivity. It can
provide information on chemical composition, structural
uniformity, electrical and mechanical properties, and material
response to temperature changes. Of particular interest for W-
Cu pseudoalloys is the excellent correlation between their
CA 02551256 1996-11-18
_ ~g _
electrical and thermal conductivity. The latter property is
of critical importance in developing thermal management
materials for electronics and is difficult to measure.
Modern methods of measuring electrical conductivity are
based on nondestructive eddy current (EC) measurement
principles. The electrical conductivities of the W-Cu
pseudoalloys produced from the W-Cu composite powders were
measured using an EC conductivity meter Type Sigmatest D 2.068
1D manufactured by the Institut Dr. Forster in Germany.
Electrical conductivity measurements were made according to
the International Accepted Conductivity Standard (IACS) issued
by International Electronical Commission in 1914 f or high
conductivity copper as 100% IACS.
Sintering activ'_~ty ;s the basic property which cor_trols
powder consolidation. To eliminate complicating factors,
sintering activity should be tested using deagglomerated
powders having finite particle dimensions. rFrom Table 6, it
2D can be seen that the average median particle size of
deagglomerated (rod-milled) W-15% Cu composite powders reduced
at 800°C from feed-stocks A, B and C is about 2 Vim. These rod-
milled powders were sintered to pxoduce W-Cu pseudoalloy
articles. Consolidation may be carried out in hydrogen,
dissociated ammonia or vacuum.
About 0.5 wt.% of an organic lubricant (Acrawax) was
blended with the W-Cu composite powder to impart pressability.
AcrawaX C is the trade name for ethylene bis-stearamide
manufactured by Lonza Co.,Fair Lawn, IQJ. This white organic
lubricant powder decomposes between 300 to 4OU°C without
*Trademark
CA 02551256 1996-11-18
- 29 -
leaving any harmful residue in the compact. Round samples
(DxH = 14.8 mm x 3 to 4 mm) were die-pressed to a green
density of between about 40 to about 75% of theoretical
density (TD) by applying a compacting pressure in the range of
18.8 to 206.8 ksi. Powder compacts were sintered under
flowing dry hydrogen in a molybdenum tube inside a high
temperature laboratory furnace having automated control of the
heating cycle. The sintering cycle consisted of a combination
of temperature increases and isothermal holds. A temperature
rate increase of 10°C/min was used between 120 minute
isothermal holds at 850, 950, 1050, 1100, 1150, 1200, and
1250°C.
No appreciable linear shrinkage of the compacts was
observed through 950°C. Above this temperature and through the
final sintering temperature, the compacts experienced varying
rates of shrinkage. Table 8 gives the shrinkage of the
compacts as a function of sintering temperature and compacting
pressure.
CA 02551256 1996-11-18
- 30 -
o u~ N ov N o r w w M m m
b u,m o a~ ~ rn~ ~rM ~ m m
N . . . . . . . . ,
riN tf1d~ t0 d'OD ~Dt0 r N M
~ m m m m m m a0m m m m
O
N ~
.-. O
d~ ri
H
m
H
O
O 01 0~N CO rIO~ N r'~lUlr m
W .-1 0
y~ m d' M rl M tf141 t0.~itDN l0
N
N O H N tD~D r r r ~DO o O O
p O H ~ N N N N N N N M M M M
O
O
H
A
M N N ~D M N U1m d'r ri
N m 01 r1 N rl r-)O N1V1 O1
~ N Q1r ~D u1d' M N riO O\
U N H e-1r-Iriri .-1n-W-W~1
N
N
t~
.~ If1M M N M t0 rlN O O rl
o r r ~ o o~ m m avM r
0 N C1r l0 l1M N r1 O O O1
W
N rlr1 rl ~wii-i'irl r1rl
U H a
01
G
U
~
o~ r1 m 01 '-1Q1l0 r c0 d~N m
('~ ~ m d' r~ lpl0 ~0l0 m O V'
rl . ~ r1 m r l0 d~M N rl O O 01
1 ,
~
"3 N H H rl rlri r1H '-Iri
H
E m
o >
N
w ;
d,
N N M 01 M N r M CD10 H
i
M m N O m ~n M d~ m r r
0 m t0V1 d' N N r1O 01m m
r-1r1ri 'i rW~1 rlri
H
O
-- U
cn
4-I
o ro o M m m o~a o w m o sr
O N ~D d' .-~I01 tt110 d~N O
N ~ ~ U7 V1d~ d' d~M M M M M M
a o
x
.r.,
m
~
m
N m tpd~ N O m l0d' N O
~ p
V
U t0 r 10 tt1d'N rlO 01m
1 UI
m x .-iM m r o,,-~M w e m N
e1 rirl riri
O
U
CA 02551256 1996-11-18
- 31 -
In each~sintering run, the total shrinkage progressively
decreased with increasing compacting pressure and green
density. However, the incremental shrinkage of compacts
exhibited an opposite trend: ratios of incremental shrinkage
to total shrinkage at 1250°C were progressively higher with the
increase in compacting pressure and better W-W contacts. The
incremental shrinkage of samples exhibited values of up to
88.67% in the temperature interval of 950-1100°C. Appearance
of liquid copper at 1100°C could not be considered as the only
factor responsible for such high rates of shrinkage since this
temperature is quite low for sintering W-Cu systems.
Therefore, such high sintering rates could only be attributed
to rapid local densification of submicron tungsten particles
at the stage of solid-state sintering. Such densification
rates should adversely affect the sintering process and result
in an unevenly sintered tungsten framework. In turn, this
should cause uneven densification and copper bleedout at the
liquid-phase sintering stage. Indeed, this has been observed
at a sintering temperature of 1250°C.
Referring to the pressing-sintering curves in Fig. 7, a
growth of electrical conductivity caw be observed with the
increase in liquid-phase sintering temperature and shrinkage
of the samples. However, despite the continued shrinkage at
1250°C, there is a drop in electrical conductivity of samples
sintered at this temperature. It is accompanied by appearance
of copper areas at the sample surfaces. Since sintered
densities of 98-98.5% TD have been reached at this
temperature, the bleedout of copper could only be attributed
to nonuniform densification, in particular, to high rates of
CA 02551256 1996-11-18
- 32 -
shrinkage and structurally inhomogeneous tungsten framework
formed at the stage of solid-state sintering.
The complete elimination of copper bleedout at high
sintered densities is achieved by reducing the rate of
temperature increase during the solid-state sintering stage. A
preferred sintering method for the tungsten-coated copper
composite powders includes the steps of: (1) lubricant or
binder removal from the consolidated powder at 300-500°C; (2)
residual oxygen removal at 800-950°C; (3) in situ sintering of
a tungsten framework at very low rates of temperature increase
(about 1°C/minute to about 5°C/minute)in the range of 950-
1080°C; (4) residual oxygen removal from molten copper at 1080-
1130°C; and (5) internal infiltration of the tungsten framework
and densification of the pseudoalloy at 1150-1600°C.
The phase distribution in the composite powder particles
(a submicron tungsten coating over a micron-size copper core)
predetermines the very high densification activity at the
liquid-phase sintering stage. Numerous sintering experiments
verified that the densification temperature should be a
function of the powder copper content as follows:
CA 02551256 1996-11-18
- 33 -
Copper Content, Temperature,
wt . % C
20-25 1150-1200
15-20 1200-1250
10-15 1250-1300
5-10 1300-1350
2-5 1350-1600
Tungsten-copper composite oxides synthesized from spray-
s dried AMT and CuzO and having a relative copper content in the
range of 5 to 25 wt:% Cu were reduced in a laboratory hydrogen
reduction furnace at 800°C using a hydrogen flow velocity of 28
cm/sec. The co-reduced W-Cu composite powders had the
following typical characteristics:
D50 Unmilled - 6.4 - 12.6
~m
D50 Rod-Milled - 1.8 - 2:6 ~m
BET - 0.55 - 0.75 m2/g
OZ - 800 - 1300 ppm
The as-reduced powders without prior deagglomeration were
pressed into round samples as above. The sintering cycle
consisted of one hour duration isothermal holds at 450, 850,
950, 1100°C and one of the alloy densification temperatures
based on its copper content (Table 10):
CA 02551256 1996-11-18
- 34 -
Copper Content, Temperature,
wt.% C
25 1200
20 1250
15 1300
1350
5 1400
The rate of temperature increase between isothermal
5 holds was 4°C/min. Between 950 to 1080°C; the rate of
temperature increase was limited to 1°C/min. After completion
of densification, the parts were cooled down to room
temperature at a rate of .4°C/min.
10 The tests with systematic control of copper content and
sintering temperature confirmed that coreduced W-Cu composite
powders possess a high sintering activity even without prior
deagglomeration. No copper bleedout was observed. High
sintered densities were achieved for W-Cu pseudoalloys with 10
to 25 wt.% of copper. The amount of liquid phase was,
apparently, inadequate for achieving a high sintered density
in the W-5wt.% Cu alloy.
Fig. 8 shows the electrical conductivities for these
samples as a function of compacting pressure. Due to
insufficient copper content, the electrical conductivity in
the W-5wt.%Cu pseudoalloy keeps increasing with the compacting
pressure within the whole range of tested pressures. For the
CA 02551256 1996-11-18
- 35 -
W-lOwt.%Cu samples, the stabilization of electrical
conductivity coincides with the stabilization of sintered
density and occurs at a compacting pressure of about 110-115
ksi. Progressively lower compacting pressures are needed for
stabilization of electrical conductivity and sintered density
for pseudoalloys with higher copper content. They are,
correspondingly, 75-80 ksi for W-l5wt.%Cu, 55-60 ksi for
W-20wt.%Cu, and 40-50 ksi for W-25wt.%Cu.
Table 11 summarizes the analyses of the microstructural
characteristics of the W-Cu pseudoalloy samples. HEI
photomicrographs showing the microstructural cross-sections of
the W-Cu pseudoalloys show uniformly distributed and densely
packed tungsten areas having sizes less than about 5 ~m and
copper areas having sizes less than about 10 Vim.
Somewhat large sintered tungsten areas and isolated
copper areas are characteristic of the W-5wt.% Cu system. The
phase distribution uniformity and interconnection greatly
improves upon increasing the copper content to 10 wt.% and,
especially, to 15 wt.%. However, higher copper contents
result in developing scattered copper areas. Still, the
tungsten and copper phases remain fully interconnected. The
size of tungsten areas in the photomicrographs progressively
decreases with the amount of copper from below 5 ~m at 5 wt.%
Cu to below 1 ~m at 25 wt.% Cu.
A similar analysis of a W-Cu pseudoalloy formed by
infiltrating a sintered tungsten skeleton with copper showed a
much coarser microstructure. Analysis of the microstructural
cross-section showed that the average size of the tungsten
CA 02551256 1996-11-18
- 36 -
areas was 10 to 15 ~tm and the size of the copper areas was 15
to 25 Vim. Thus, the W-Cu pseudoalloys formed by pressing and
sintering the tungsten-coated copper-.composite powders have
much finer microstructures than infiltrated W-Cu pseudoalloys.
Furthermore, W-Cu pseudoalloys having a copper content less
than 10 wt.% cannot be successfully made by infiltration
techniques.
CA 02551256 1996-11-18
- 37 -
F.
w j
0 0
ro ~ ~I w H ~I w
m N
ro
x ~ o
N w G
~ m
x
o z s ar
r
3 E 0 ~
3
2i
~ G '' ~ U
O WN b ~ i
C. tlJ 1
Vro ~ U~ 1 ~ N ~ ~
3 N
1
.
~ (n -'i
l.~ N ~t 1.1 1.1 dl
3
oa~ a~ N a~ N
ro ro
V 3 N~3 N O ~ ~ a
u1 m 3 ~
~ ~ ' ~
m b~ a ~ ?~ m
w a O
N 11 ..ai.J U N ~. rl 5, "
ro Pr t~
.~ ro ro N ~ .~ ~
b ~ .,~ ro
.-~ ~ a~ .~ b
.-~ m a '
~
m ~ ~' o ~
~ o ~ ~
,w a - - o
~ r
o n
o Ha ~n..~p H x H xw ,
v a
N
c~
~. b b
~
-,~ ow ~ ~ a b b b
b ~
w -~
~ b
>
~
x
ro
b ~o
H vl ~ v~ H H
ro
b
~ ~ ~ ~ i
_ 0
~ ~ ~
o a~
~ 0 3
m ~ ; ~ ~ o ~
a o
ti ~ b
,~ N a~ ~ o cn roN o
m o ~
~
a~ ~, m w N v ro w
,~ w o
o
b ..~ a~ ~ ..aro m > -.~ .b
a ..~ -.~
-a
~~ o
o
~, ~ ~ ~ ~
V 4) !I1 UlN ~ r u1 N r
N l N l
N f-1
N
rn roro ~ w w ~ row ro x
ro a~ a~ ~
~n
a ~ N ~ ~ ~p o ~ d ~ ~
~ N N ~ ~
C
a ro. ~ w ~ rop - ~n
ro cn o ~n ro .
ro ~n x
b 'd
w
0
w
o
* V
N ~ O u1 O u1 *
rl e-i N N
U p ~ U
3
V r
3
CA 02551256 1996-11-18
- 38 -
Table 12 compares the electrical conductivities of W-Cu
pseudoalloys made by sintering the co-reduced W-Cu composite
powders with pseudoalloys made by pressing, sintering, and
infiltrating a tungsten skeleton with copper. Appreciably
higher electrical conductivities, and therefore higher thermal
conductivities, are achieved in W-Cu pseudoalloys made from
the W-Cu composite powders having W-coated copper particles.
Copper Electrical
Content Conductivity,
In % IACS
Pseudoalloy, of Pseudoalloy
% Made:
By By From CompositeBy Cu Conductivity
Weight. Volume Powder Infiltration Ratio
of a Tungsten
Skeleton
5 10.2 27.6 --- ---
10 19.3 36.9 35.0 1.054
27.5 41.2 --- ---
35.0 44.2 41.0 1.078
42.D 48.9 45.0 1.086
spray-dried, lubricated, flowable powders which can be
15 formed into various complex shapes are considered an ideal
powder source for manufacturing technologies involving
pressing, molding, rolling, and extrusion. As a rule, powders
are subjected to deagglomeration (milling) prior to spray
drying which significantly improves their uniformity.
Samples of a CuW04 + 0.958 W03 composite oxide (15% Cu)
synthesized from spray-dried AMT and CuzO were reduced in a D-
CA 02551256 1996-11-18
- 39 -
muffle furnace at 800°C (Type 1) and in a tube furnace at '750°C
I,T~~pe 2) under industrial conditions. The co-reduced W-Cu
composite powders were milled for 30 minutes in a water slurry
using Type 440C stainless steel balls as milling media.
Carbowa~:-8000' a high molecular weight polyethylene glycol
manufactured by Union Carbide Corporation, was used as a
binder. The Carbowax-8000 was added to the slurry to give a
2.5 wt.% binder concentration in the spray-dried powder.
Spray drying yielded a spherical powder having excellent
flowability and pressability and a 90% -60 +200 mesh fraction.
Rectangular (16 x 18 x 5 mm) samples weighing 20 g each were
pressed at 70 ksi from both types of spray-dried powder. The
compacts were subjected to the preferred sintering method
described previously. Table 13 gives the average test results.
for pseudoalloys made from Types 1 and 2. No adverse effects
or_ sintering properties were observed at the carbon
concentrations remaining in the compacts after dewa~:ing. Good
sin'~ered density, electrical conductivity, and complete
elimination of any signs of copper bleedout were achieved in
the sintered samples. A representative sample microstructure
is characterized in Table 11 and shown in rig. 9.
Fig. 9 is a BEI photomicrograph of a W-Cu pseudoalloy
made from a milled and spray dried tungsten-coated copper
composite powder having 15 wt.% Cu which was sintered at
1250°C to 98.2% of theoretical density. The tungsten areas in
the microstructural cross-section appear white and the copper
areas appear dark. The photomicrograph clearly shows the
uniform distribution of the tungsten and copper areas as well
as the highly interconnected structure of the pseudoalloy.
* Trademark
CA 02551256 1996-11-18
- 40 -
.~ o m
..~-1
~
H
N
O
U
N
d1
'd
~ ~ vp r~i
O ~ m
W s~ ov ov
N
~
~
~
tn
A
s~
U
td
O
U y~
N
rl
~
U ~ M
~ r1 M
a
'
Id
U
bt
U
0
.a o 0
N o ~1
'
E
p
N
ij
H
-' o
O
O ~ M
~
N
..a
W
p aoo
H O rl
b~
(!~
W
N~
b
3
O
W N N
~
~i ~
W
~~ ~N
~'
~n
CA 02551256 1996-11-18
- 41 -
IV. CERAMIC METALLIZATION
Ceramic metallization is one of the most critical and
precision operations in electronics. It consists of forming a
metal layer on ceramic surfaces with a ceramic-to-metal joint
which is both strong and vacuum-tight. Metallized ceramic
surfaces are used for ceramic-to-metal and ceramic-to-ceramic
connections and for wiring conductor layers (paths) in
electronic technology. Metallized ceramic is produced by
applying a metall~zing paste to sintered or unsintered ceramic
substrates (mostly alumina or beryllia) and firing the paste
to yield an adherent metallized area. Co-firing a paste with
the "green" (unsintered) ceramic is economically advantageous
since it eliminates presintering the ceramic and results in
better quality metallization. A number of paste formulations
and metallization techniques have been disclosed in U.S.
Patent Nos. 3,620,799, 4,493,789 and 4,799,958.
ZO Generally, a paste for ceramic metallization consists of
three basic components: metal powder, frit material, and a
binder. For low-temperature metallization (up to 1200-1250°C)
of presintered ceramics metals like silver, palladium, nickel
and copper may be used. High-temperature metallization or co-
firing with an unsintered ceramic (up to 1600-1700°C) requires
the use of refractory metals (W, Mo) alone, or their mixtures
with Mn or Pt. Frit material comprises oxides and silicates.
Due to their glassy nature, frit components are filling the
pores in the ceramic substrate and metal structure and help
making a vacuum tight ceramic-to-metal joint. Metal powders
and frit materials are held together by binders like ethyl
CA 02551256 1996-11-18
- 42 -
cellulose, nitrocellulose, epoxy resins, and many other
organic binding agents. Usually, the components are milled
and mixed for several hours to decrease their particle size
and make a homogeneous blend. Finally, the consistency of the
metallizing paste is adjusted by a vehicle which is actually a
solvent compatible with the binder. For instance, for ethyl
cellulose binders, ethylene glycol dibutyl ether is preferred.
Consistency adjustment is required according to the method of
application of the paste on ceramic surfaces. Screen printing
technique is usually employed for ceramic wiring boards and
packages; other methods like painting, spraying, dipping, etc.
are also broadly applied.
Refractory metals require high temperatures for
metallization. The electroconductivity of the tungsten
conductor layer is increased in proportion to the sintering
temperature. Higher electrical conductivity makes it possible i
to sinter finer wires (paths or lines) and increase the
packaging density of the wiring boards. However, metallizing
temperatures above 1750°C are very detrimental to the ceramic.
Even at temperatures between 1450 and 1700°C the ceramic
exhibit-~s grain growth which may be detrimental to its
strength, and hot creep to the point where creepage is a
severe problem.
All three objectives - (1) lowering the co-firing
temperature for metallizing an unsintered ceramic; (2)
increasing the electrical conductivity of the metallized
wires; and (3) improving the packaging density of wiring
boards - can be accomplished by using the tungsten-coated
copper composite powder in the metallizing paste formulations.
CA 02551256 1996-11-18
- 43 -
The W-5wt.% Cu powder can be sintered at temperatures as
low as 1400°C to high densities and electrical conductivities.
However, copper content in the range of 2 to 15 wt.% can be
used for broadening the range of metallizing temperatures to
1200-1600°C. By controlling the copper content in the W-Cu
composite powder used in pastes, the electrical conductivity
of the metallized conductive lines can be controlled.
Moreover, due to excellent structural homogeneity of W-Cu
pseudoalloys from the W-coated copper composite powders, the
width of the metallized conductive line can be decreased
thereby improving the packaging density of a semiconductor
assembly. Indeed, x.002 inch line widths have been
successfully metallized using the W-15 wt.% Cu powder.
While there has been shown and described what are at the
present considered the preferred embodiments of the invention,
it will be obvious to those skilled in the art that various
changes and modifications may be made therein without
departing from the scope of the invention as defined by the
appended claims.
