.
C
:
.
-
-
T
-
-
-
--
I OFL.
ORNL P.
3159

.
'
1
-
.
.
.
.
.
EEEEEEFT
1.25 1.1.4 1.1.6
MICROCOPY RESOLUTION TEST CHART
NATIONAL BUREAU OF STANDARDS – 1963
...
ORNA
p-3.59
a.
Conf.676905--7
: MASTER
JUL 1 9 1967.
Cr21 PRICES
HC $3.00 MN_.65
MECHANICAL BEHAVIOR OF CVD TUNGSTEN AT ELEVATED TEMPERATURES:
tris
H. E. McCoy and J. O. Stiegler
Metals and Ceramics Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee
29 pages
3 tables
21 figures
LEGAL NOTICE
This report was prepared as an account of Government sponsored work. Neither the United
States, nor the Commission, nor any person acting on behalf of the Commission:
A. Makes any warranty or representation, exp or implied, with respect to the accu-
racy, completene88, or usefulness of the informat on contained in this report, or that the use
of any information, apparatus, method, or process disclosed in this report may not infringe
privately owned rights; or
B. Assumes any liabilitiea with respect to the use of, or for damages resulting from the
; use of any information, apparatus, method, 0: process disclosed in this report.
Ao use in the above, "person acting on behalf of the Commission" includes any em-
ployee or contractor of the Commission, or omployee of such contractor, to the extent that
Buch employee or contractor of the Commission, or employee of such contractor prepares,
disseminates, or provides access to, any informatioa pursuant to his employment or contract
with the Commission, or his employment with such contractor,
.
DISTRIBUTION OF THIS DOCUMENT IS UNLIMITED
Mechanical Behavior
-
.
w
ABSTRACT
Many applications of CVD tungsten require that the material be
stressed at elevated temperatures. We have run creep-rupture tests at
1650 and 2200°C to evaluate the mechanical behavior of this material,
and the properties are compared with those of a typical heat of powder
metallurgy tungsten. At 1650°C the CVD product has low fracture strains
(approximately 5%) and a lower minimum creep rate. At high stresses
the rupture life is shorter than that of the PM material; at low stresses
the rupture lives are about equivalent. At 2200°C the minimum creep
rate is higher and the rupture life lower for the CVD product.
Two microstructural features of importance were noted in the CVD
tungsten: (1) the formation and growth of voids and (2) the columnar
nature of the grains. Fractographic techniques were used to study void
nucleation and growth in the material. Nucleation appears to be sponta-
neous as the material is heated to elevated temperatures, indicating the
presence of an impurity having a high vapor pressure. The growth appears
to occur almost entirely by the stress-induced diffusion of vacancies
into the void. At 2200°C under stress the voids reach such a large size
that they comprise about 10% of the test specimen. The columnar grain
structure of the material is important because it is very difficult to
get extensive grain-boundary sliding and rotation in this type of
structure.
We have rationalized the creep behavior of this material on the
basis of the effects that both the void growth and the columnar grain
structure have on the individual deformation processes that seem to
give the overall creep behavior.
INTRODUCTION
Chemical vapor derosition (CVD) is now recognized as an attractive
.......
..
..
..
way of obtaining complex tungsten shapes without subsequent fabrication
as well as a means of obtaining high-purity stock for further fabrica-
tion. The production process generally involves the hydrogen reduction
of WFs to deposit tungsten on a heated substrate (1). It has been
shown that, after suitable heat treatment, the low-temperature mechani-
cal properties of the CVD tungsten are comparable with those of tungsten
produced by conventional methods (2).
The excellent resistance of
tungsten produced by this process to grain growth at elevated terapera-
tures has also been demonstrated (3,4,5). However, voids have been
observed when this material is annealed at elevated temperatures,
particularly when a stress is applied (3,5,6). Taylor and Boone (6)
found that the tensile strength and fracture ductility of the CVD mate-
rial were less than the respective values for powder metallurgy (PM)
tungsten, but the creep properties of this material have not been
evalue.ted.
In the present study we evaluated the creep-rupture properties of
several lots of CVD tungsten at 1650 and 2200°C. We shall compare the
results of these tests with those of similar tests on a heat of PM
tungsten. Since the properties of the two materials differed consider-
ably, we performed extensive optical and electron microscopy in an
effort to ascertain the reasons for the differences. Although numerous
unanswered questions remain, we were able to propose a qualitative
explanation for the creep-rupture behavior of CVD tungsten.
-
vi
EXPERIMENTAL DETAILS
Test Materials
ES
The PM sheet used in this study was obtained under a Bureau of
Naval Weapons Contract and the fabrication details are covered in the
final report on this contract (7). It was fabricated by the PM tech-
nique and was rolled to a final thickness of 0.060 in. The final treat-
ment was a 5-min anneal at 1150°C. The high purity of this material is
illustrated by the results of chemical analyses shown in Tables I and II.
The CVD material was produced by the hydrogen reduction of WF6 on
a heated substrate. The cletails of the deposition process have been
described previously (1). The material was deposited as a box about
2 by 2 in. in cross section and 20 in. long; the thickness varied from
0.050 to 0.060 in. It was deposited on the inside of a mandrel - copper
for the "H" series; molybdenum for the "p" series. The mandrel was
removed by chemical etching, and test specimens were made from the sides
of the box. Chemical data are given for two typical heats in Tables I
and II.
Test Specimens
A small sheet test specimen with a gage section 1.5 by 0.25 in. and
an overall length of 3.5 in. was used. There were small holes near each
end for pinning the specimen to the extension rods. The PM tungsten
specimens were made by grinding. The CVD tungsten was too fragile for
grinding and the specimens were made by electrodischarge macnining. A
tool was made for the Elox machine which utilized simple brass shapes (5).
The tool was inexpensive and could be rebuilt easily. About 0.002 in.
was removed from the machined surfaces by abrasion with diamond paste
to remove small intergranular cracks that were formed by the electro-
discharge nuachining. The' as-deposited surfaces were lapped in some
instances, but this did not seem to have any effect on the high-
temperature properties.
Testing Methods
TUU
All of the tests were run in a Brew Creep Testing Apparatus,
Model :. 1062., having a tungsten mesh heating element. The vacuum
system was cold trapped and was capable of maintaining a pressure of
1 x 100? torr at a test, temperature cí: 2200°C. The temperature was
automatically controlled using a total radiation pyrometer which was
sighted on the test specimen and a controller which adjusted the power
to the furnace. This control system was stated to have a control
accuracy of +3°C. The temperature could also be read by an optical
pyrometer through a sight port in the front of the test chamber. The
strain was measured by a dial gage which indicated the motion of the
pull rod.
Test specimens were normally built into the test equipment, pumped
for 2 hr, heated to about 600°C and pumped for 12 hr, heated to the test
temperature in about 4 hr, and loaded.
Several specimens were analyzed for interstitials after testing.
No evidence of contamination was noted except in cases where the pressure
was known to have increased during the test. Data from tests where
leaks were known to have developed have not been considered.
At the time the specimens fractured, the lower pull rod dropped
and caused a temporary pressure rise that caused the power to the furnace
-
to be cut off. Since the specimens were heated by radiation, the
cooling rate was high enough that the structures observed subsequently
were typical of the deformed state.
Fractographic Techniques
Tungsten fractures in a brittle, intergranular manner at low temper-
atures, and voids formed at elevated temperatures can be exposed without
distortion by snapping the specimens at room temperature. The fracture
surfaces were replicated directly with carbon and shadowed with platinum.
Replicas were taken from the creep-fracture surfaces and froin surfaces
below the original fracture surface that were exposed by fracturing at
room temperature.
The replicas were examined in the electron microscope
to determine details about the voids such as size, number, geometry, and
location.
EXPERIMENTAL OBSERVATIONS
The creep-rupture properties of several lots of CVD turigsten are
compared with those for PM tungsten in Fig. 1. At 1650°C the rupture
life of the CVD tungsten is generally less at higher stresses (e.g.,
6000 psi) and equivalent at lower stresses. Lot, PW-20 is somewhat
anomalous in that its rupture life is greater than that of PM tungsten.
At 2200°C, the CVD material has a shorter rupture life except for the
PW-3 specimen tested at the lowest stress.
Figure 2 compares the minimum creep rates of PM and CV tungsten.
At 1650°C the CVD material exhibits a lower creep rate and at 2200°C
the opposite is noted.
*HARE
h itet,-!
S
The fracture strains for the CVD and PM material are compared in
Fig. 3. At 1650°C the CVD material exhibited strains of about 5% with
only two notable exceptions – lot PW-18 with 13.2% and H-16 with 21.0%.
The two exceptions are quite important and will be discussed later. At
2200°C grain growth occurs quite rapidly in the PM tungsten and this
probably accounts for the decrease in ductility with increasing rupture
life. The CVD tungsten generally exhibits fracture strains in the range
of 15 to 25% at 2200°C. Again there are two exceptions – lot PW-19 at
4.2% and PW-3 at 29.2%.
A detailed study has been made of the intergranular void formation
in PM tungsten and these findings have been reported previously (8).
.
However, we shall show some typical micrographs for the purpose of
making comparisons between the PM and CVD materials. Figure 4 shows
the fracture of a PM specimen tested at 6000 psi and 1650°C. The fail-
ure is obviously due to the linking of intergranular cracks. The fracto-
graph shown in Fig. 5, from a region 10 mm below the high-temperature
fracture surface, shows these cracks in an early stage of development.
These cracks originate as intergranular voids that are located quite
randomly and are irregular in shape. There is considerable evidence of
preferential growth and linking in particular directions. The larger
irregularly shaped voids would appear as intergranular cracks when
viewed perpendicularly to the plane of Fig. 5.
Figure 6 shows the microstructure of PM tungsten after testing at
1500 psi and 2200°C. Features of importance are the extensive grain
growth and the formation of intergranular voids. The fractograph in
Fig. 7 defines the nature of the intergranular voids more accurately.
The voids have well-defined shapes and the details of the crystallography
are discussed elsewhere (9).
The CVD tungsten is characterized by large columnar grains which
are oriented in the thickness dimersion of the deposited sheet. A layer
of small grains forms at the substrate surface where the deposit begins.
A cross-sectional view of a CVD tungsten sheet is shown in Fig. 8a.
This particular sheet is atypical in that the gas supply was interrupted
aná a laminate formed near the center of the sheet. A view of the sheet
parallel to the surface is shown in Fig. 86.
The microstructure in this
plane gives the appearance of an equiaxial grain structure. The test
specimens were oriented so that the stress was applied perpendicularly
to the columnar grain boundaries.
Figure 9 shows a microstructure typical of lot PW-3 after creep
obviously made up of interconnected voids.2 The density of voids is
r'easonably high near the fracture and diminishes rapidly away from the
fracture. Figure 10 is a fractograph of a surface about 10 mm from the
fracture. Note that the voids present are quite small. Figure ll is
a fractograph made at a similar location in a specimen tested at 1650°C
and 6000 psi. By comparison with Fig. 10 it can be seen that the void
size is greater at the higher stress, even though this specimen was at
temperature a small fraction of the time of the specimen held at the
lower stress. The voids in Fig. 11 have very well-defined geometrical
shapes as compared with those formed in the PM material under similar
conditions (Fig. 5).
10
Lot H-16 exhibited a somewhat different microstructure, primarily
with respect to a greater propensity for void formation. At 1650°C and
6000 psi numerous voids were formed, but the total void fraction was
small. At 4000 psi (Fig. 12), the voids were quite large and the void-
fraction was high. Fractographs showed the presence of large voids with
regular, geometrical shapes.
Lot PW-20 also exhibited extensive void formation at 1650°C.
Figure 13 shows the fracture of a specimen tested at 6000 psi.
The
small, almost continuous intergranular network of voids extended
throughout the specimen.
We made an effort to increase the fracture ductility at 1650°C by
creating a more equiaxial grain structure. This was done by (1.) warm
warm
work and recrystallization and (2) deposition of a product with a finer
grain size. The warm-worked material, lot PW-69, was deformed (by
rolling) 80% at 500°C and recrystallized by annealing 1 hr at 2200°C.
The specimen, which was then stressed at 6000 psi, failed at a strain
of 3.7%. The fracture is shown in Fig. 14,
The grain size is large,
but the structure appears equiaxial. The inhomogeneous distribution of
the voids is very cbvious. The impurity responsible for the nucleation
of the voids was probably swept up by the first boundary that moved
through each area, thus accounting for the segregation. Figure 15 shows
a fractograph of the specimen from lot PW-69. The structure is dominated
by the large, crystallographic voids. The material deposited with a
finer grain size was lot PW-18. Figure 16 shows the fracture of a speci-
men from this material that was tested at 1650°C and 6000 psi. A
fracture strain of 13.2% was obtained. Although fracture appears to
ill
have resulted from the connecting of intergranular voids, there is
evidence of considerable plastic deformation. A fractograph of this
specimen made on a face about 10 mm from the fracture is shown in
Fig. 1% The voids in this specimen resemble more closely those of
the PM tungsten (Fig. 5) than those of the other CVD specimens. The
voids are linking together as well as showing growth in a particular
direction.
At 2200°C the various lots of material did not exhibit such a wide
variation in microstructural appearance. Figure 18 shows the type of
microstructure typically observed for the CVD material that failed in
relatively short times. As the stress was lowered and the length of
the test increased, the voids were larger and fewer in number (Fig. 19.1
The grain structure also became more equiaxial, indicating the freedom
of the boundaries to migrate under stress. The data in Table III indi-
cate that the void volume in the specimens from lot PW-3 that were
tested at 2200°C did not vary greatly with rupture life. Thus, the
larger voids (Fig. 3) were probably formed by the coalescence of the
smaller voids (Fig. 18) observed in tests of a shorter duration.
The voids were so large in the specimens tested at 2200°C that they
were difficult to study by fractographic techniques. Figure 20 shows an
area where the bubbles were relatively small. Note their regular crystal-
lographic shapes. The retention of this regular shape along the bounda-
ries (where three grain surfaces intersect) indicates that very little
grain-boundary sliding occurred.
Although the effect of stress on the growth of voids in the CVD
material has been implied in several instances, a comparison of Figs. 18
12
and 21 clearly demonstrates the effect. These specimens were in the
test furnace at the same time, the only difference being that one was
unstressed.
The void fraction data in Table III are again useful. The
average void fraction of the stressed specimen was 6.1%, whereas that
of the unstressed specimen was only 0.20%..
Ir
DISCUSSION OF RESULTS
Awr
i
'
In creep tests at elevated temperatures, void formation or "cavita-
T
ATT TYS
tion" has been observed to lead to intergranular cracking and failure
in numerous materials - copper (10), alpha-brass (10), magnesium (10),
types 304 and 304L stainless steels (11), nickel (12), Inconel 600 (12),
iron (12), Nimonic 90 (13), and others. In this work we observed that
cracks developed in both PM anå CVD materials from small grain-boundary
voids or bubbles although the appearance and behavior of the voids
differed in the two types of tungsten. We will now consider how differ-
ences in nucleation and growth of such cracks can lead to the differences
in structure and properties that were observed.
The several methods that have been proposed for the nucleation of
voids have been reviewed by Davies and Dennison (14). Most mechanisms
involve a grain-boundary discontinuity of some type and a shear stress
a ɔng the boundary. Detailed observations of the PM material suggest
that voids were nucleated in this manner as a result of considerable
Ver
plastic deformation (8). However, in the CVD material, plastic deforma-
tion was not a prerequisite for void formation. Bubbles formed simply
by heating the material to temperatures above about half the absolute
melting point were able to act as void nuclei. The voids (or more
LE4
m
a
nen
"*
.-
-
.
13
more properly bubbles) at this stage were likely caused by a locally
high concentration of some impurity which is gaseous at the test temper-
ature. The voids (bubbles) are generally attributed to fluorine
impurities (3,6,15), but the present observations do not show a correla-
tion between the fluorine content and the volume of voids formed under
stress. For example, heats H-16 (Fig. 12) and PW-20 (Fig. 13) exhibited
a greater propensity for void formation than heat PW-3 (Fig. 9) although
the former heats had lower fluorine contents (Table I). Thus, it
appears that, at least under stress, variables in addition to the bulk
fluorine content must be of importance in void formation.
Once nucleated, voids may grow by plastic deformation (dislocation
· motion and grain-boundary sliding) or by stress-induced vacancy migra-
ticor and condensation. Under some conditions void growth occurs in the
PM material by plastic deformation (8). However, the limited ductilities
of the CVD specimens suggest that plastic processes are not important
in determining the rate of growth of the voids. Accordingly, we will
consider void growth in the CVD material in terms of the vacancy diffu-
sion process. Any growth by plastic deformation will simply augment
the growth by vacancy diffusion.
A void in a solid is thermodynamically unstable and reduction of
the surface area of the void by sintering reduces the total energy of
balance the surface tension of the void such that (16),
..
0 = ricos 0)2
(1)
.
...
..........
...
formatore, ma
* -prom, * ***********
where
o = applied stress,
y = surface energy of the void,
r = radius of the void
= angle between the normal to the plane in which the void lies
and the direction of the applied stress.
If this equality is not met, an elastic strain field surrounding the
cavity will either cause it to grow by absorbing vacancies or to
shrink by emitting them. If a gas is present in the voiü, the applied
stress necessary to stabilize a void of a given radius will be reduced.
The problem in calculating the vacancy flux into a void lies in
estimating the elastic strain field surrounding the void. This deter-
mines the chemical potential of a vacancy in the vicinity of the void.
Hull and Rimmer (17) simplified the problem by assuming that the
potential varied linearly from a value of – 2X12 on the surface of the
void to a value o N at a point midway between voids, where N is the
atomic volume and o is the tensile stress normal to the void. On this
'basis they showed that the vacancy flux into the void, j, was given
approximately by:
..so a [042 - ?]
where
D = grain-boundary diffusion coefficient, .
00
= Boltzmann's constant,
E
= absolute temperature,
a = void spacing,
R
P = gas pressure inside void.
-
t
This expression shows that the growth rate is proportional to the differ-
ence between (1) the sum of the applied stress and the pressure in the
bubble and (2) a term proportional to the surface tension of the bubble.
Temperature enters mainly through the exponential dependence of the D.
on temperature.
When CVD material is heated to temperatures above about 1400°C, a
gaseous phase precipitates, forming bubbles having a range of sizes (15).
If a stress is applied, the vacancy flux into the bubble is approximated
by Eq. (2). Small bubbles will grow until they reach an equilibrium
size at which time j = 0. For bubbles larger than some critical size
ra, equilibrium cannot be established and these bubbles will continue
to grow indefinitely. In terms of Eq. (2), ir must decrease to reach
equilibrium, but the sign of j is such to cause r to increase.
We can now understand the differences in creep behavior between
the CVD and PM materials in terms of nucleation and growth of voids.
In the PM material no void nuclei were present initially; voids were
nucleated by plastic deformation. In the CVD material the grain bound-
aries were covered with bubbles which acted as void nuclei. At high
stresses, many of these bubbles were larger than the critical size.
They grew and linked up, causing failure in a short time. At lower
stresses, fewer nuclei of the critical size were present, their growth
rate was lower, and longer rupture times resulted. *This explanation
accounts for the decreased rupture times at the higher stresses for
CVD specimens and for the decreased slope of the stress-rupture time
curve. It also suggests that, for sufficiently low stresses, bubble
growth cannot occur and that the stress-rupture curves of the CVD and
16
PM materials will intersect. For reasons to be discussed below, we
feel that extensive grain-boundary sliding does not occur in CVD mate-
rial. Since cavities in the PM material are believed to result from
sliding, it is possible that the creep-rupture curves cross and at
sufficiently low stresses the ÇVD material may have longer rupture lives.
The question of fracture ductility is more complex, for we must
consider not only growth of voids and the elongation they provide, but
also the concomitant processes of grain-boundary sliding and bulk defor-
mation. The total strain, Em, is the sum of the contributions from bulk
deformation, eg, grain-boundary sliding, fp, and void formation, Evi
The bulk strain arises mainly from dislocation motion. Since the puri-
ties of the PM and CVD materials were comparable, they should show
roughly equivalent bulk strain contributions.
Garofalo (18) has recently reviewed the data on grain-boundary
sliúing. In general, the strain contribution from grain-boundary slid-
********...-... wat moon
ing increases as the stress decreases and as the temperature increases,
Grain-boundary sliding requires a shear stress along the grain boundary.
Such a stress was probably present in the PM specimens as evidenced by
the elongated shapes of the voids. Such shapes could result from plastic
deformation or from stresses that would produce vacancy flux gradients
favorable for this type of growth. However, the magnitude of the strain
in PM tungsten due to grain-boundary shearing at 2200°C probably decreased
rapidly with time due to the excessive grain growth.
The CVD tungsten
had two factors that combined to reduce en to negligible values. The
first was the void formation that occurred. Shearing the voids would
produce some new surface area and thus would require a greater stress
than if the voids were not present.
There may also be a strain field
associated with the voids since some localized plastic deformation
could have occurred during the nucleation and early growth of the void.
Only lot PW-3.8 showed any evidence of grain-boundary deformation (Fig. 17).
A second and probably more important factor was the columnar grain struc-
ture of the CVD material. The microstructure of the deposits approaches
that of long rods packed together with their axes perpendicular to the
applied stress.
It would seem difficult to develop shear stresses of
sufficient magnitude to cause grain-boundary sliding since no shearing
stress could be developed parallel to the axis of the rod. Although
there would be a shearing stress tending to rotate the rods, such a .
movement would involve much material and would require very high stresses.
Thus some sembl.ance of an equiaxial grain structure would appear to be
necessary for extensive grain-boundary sliding to occur. This was only
obtained (1) after long periods of time at 2200°C where some grain-
boundary migration could occur (Fig. 19), (2) in low PW-69 which was
worked and recrystallized (Fig. 14), and (3) in lot PW-18 which was
deposited with a fine-grain structure (Fig. 16).
It is significant that lot PW-18 exhibited good ductility at 1650°C.
The fine-grain size of this material may be due to the cadmium impurity
present. Although the fluorine content was lower than that of the
other heats, there was still profuse bubble formation. Hence, we feel
that the improved ductility of this material was due primarily to the
change in the grain structure. However, we must point out that these
two factors, grain structure and fluorine content, may not be entirely
independent in the CVD material. For example, previous studies have
shown that greater grain-boundary mobility is obtained with decreasing
fluorine content (3). Thus, the lower fluorine content would allow
grain-boundary migration to occur with the resultant formation of a
more equiaxial structure that would be more conducive to grain-boundary
shearing and rotation. However, it remains to be demonstrated whether
under stress the threshold fluorine (or other impurity) levels are high
enough to be attained in practice. Before stressing, lot PW-69 was
28
heated to 2200°C for recrystallization with attendant large void forma-
we
tion. There was no evidence of grain-boundary sliding in this material.
Thus we would conclude that the ey would be very small at 1650°C, except
in lot PW-18, and would probably be small at 2200°C except after long
test periods.
The grain-boundary sliding process is important not only because
of its contribution to the overail strain but because it acts as a
.!
L
stress-relieving mechanism. The deformation of individual grains leads
to large local stresses that can be relieved by sliding. Hence, in the
absence of sliding, these stresses may lead to cracking between the
.
A
I
FOL
.
L
.
intergranular voids with resultant premature failure.
The strain due to the voids is most significant at 2200°C. Table III
D
lists the results of some void fraction measurements made on the test
specimens. At 1650°C, the strain due to void formation in lot PW-3 was
a maximum of 0.35%. However, the strain due to void formation was higher
in lots PW-20 (Fig. 13) and H-16 (Fig. 12). At 2200°C, lot PW-3 had
void fractions of 7 to 9%. There was no measurable reduction in width
of the specimen, and the total void fraction can be assumed to be axial
strain. The other lots of material exhibited similar void growth at
a
2200°C and we can conclude that up to 10% of the strain in this material
can be attributed to void growth.
The creep properties shown in Figs. 1., 2, and 3 can be rationalized
by the way that the various terms contributing to the total strain are
affected – €. En, and ex. At 1650°C, en and €,, are both generally low
at high stresses. The creep rate is lower, but the inability for bound-
ary deformation to occur leads to reduced ductility and slightly reduced
rupture life. At low stresses (approx 4000 psi) the stress is not high
enough to stabilize large voids, and extensive void growth does not
occur (Fig. 10). Thus, the rupture time more closely approaches that
of the PM material. However, the material still lacks the ability to
undergo 'extensive grain-boundary sliding with resultant low minimum
creep rate and low fracture strain.
The high strain of sample H-16 at 4000 psi was due to the extensive
void formation and a large value of Ex
Lot PW-18 also differed in that
the more equiaxial structure resulted in a relatively large value for en
The result was a higher creep rate and rupture strain. At 2200°C, the
En term is probably initially small and Er is large. This results in
higher creep rates, high apparent ductilities, and generally lower
rupture lives.
CONCLUSIONS
We have compared the creep-rupture results of several lots of CVD
tungsten with those of PM tungsten. The failed PM tungsten specimens
contained intergranular voids that linked together to form intergranular
cracks. The CVD tungsten formed voids simply on heating to elevated
temperatures. These voids grew under the influence of an imposed
es
stress. The presence of these intergranular voids and the columnar
grain structure of the CVD material combined to reduce the amount of
grain-boundary shearing that occurred. At 1650°C, the minimum creep
rate and the rupture ductility of the CVD tungsten were less than those
of the PM tungsten. The rupture life of the CVD tungsten was less at
high stresses and equivalent at lower stresses. The stabilization of
more voids and their subsequent growth at the higher stress are thought
to be responsible for the rupture behavior. The lower creep rate and
ductility were due to the inability for grain-boundary shearing to
occur. One lot of material that had a fine-grain size exhibited much
better fracture ductility.
At 2200°C the voids were so large that they contributed signifi-
cantly to the measured strain. The minimum creep rate was higher and
the rupture life was lower for this material than for PM tungsten.
ете
ACKNOWLEDGMENTS
The authors are indebted to several persons and groups who
contributed to this study: E. Bolling and B. McNabb, for running the
creep-rupture tests; H. R. Tinch, for the optical metallography;
Metals and Ceramics Division Reports Office, for the preparation of the
manuscript; D. G. Gates, for the drawings; ORNL Anal.ytical Chemistry
Division, for the chemical analyses; E. E.. Bloom and A. C. Schaffhauser,
for the technical review of the manuscript.
FOOTNOTES
1. Research sponsored by the U.S. Atomic Energy Commission under
contract with the Union Carbide Corporation.
2. We have used the term "void" throughout this paper in referring
to the cavities that are formed. In the PM materişl these cavities are
indeed voids. During nucleation and the early stages of growth, the
cavities in the CVD material probably contain a gas and are more properly
called bubbles. However, under stress these bubbles grow quite large
by the condensation of vacancies. The gas pressure likely becomes
quite small and the term void is again more descriptive.
ev
REFERENCES
1. R. L. Heestand, J. I. Federer, and C. F. Leitten, Jr.,
Preparation and Evaluation of Vapor Deposited Tungsten, ORNL-3662
(August 1964).
2. A. C. Schaffhauser, "Low-Temperature Ductility and Strength
of Thermochemically Deposited Tungsten and Efl'ects of Heat Treatment,"
pp. 261–276 in Summary of the Eleventh Refractory Composites Working
Group Meeting, AFML-TR-66-179 (July 1966).
3. A. C. Schaffhauser and R. L. Heestand, "Effect of Fluorine
Impurities on the Grain Stability of Thermochemically Deposited Tungsten,"
pp. 204–211 in 1966 IEEE Conference Record of the Thermionic Conversion
Specialist Conference, Institute of Electrical and Electronics Engineers,
New York, 1966.
4. A. F. Weinberg, J. R. Lindgren, N. B. Elsner, and R. G. Mills,
Grain Growth Characteristics of Vapor-Deposited Tungsten at Temperatures
up to 2500°C, GA-6231 (1965).
5. H. E. McCoy, Creep-Rupture Properties of Tungsten and Tungsten-
Base Alloys, ORNL-3992 (1966).
6. J. L.. Taylor and D. H. Boone, J. Less-Common Metals 6, 157 (1964).
7. G. C. Bodine, Tungsten Sheet Rolling Program – Final Report
Covering Period 1 June 1960 to 1 March 1963, Fansteel Metallurgical
Corporation, under Contract No. NDW-60-0621-C (March 1, 1963).
8. J. 0. Stiegler, K. Farrell, B.T.M. Loh, and H. E. l.cCoy,
"Nature of Creep Cavities in Tungsten" (to be publ.ished).
9. K. Farrell, B.T.M. Loh, and J. 0. Stiegler, "Morphology of
Bubbles and voids in Tungsten" (to be published).
: ---
23
10.
J. N. Greenwood, D. R. Miller, and J. W. Suiter, Acta Met. 2,
250 (1954).
11. Gunji Shinoda, Tadao Sano, and Tadakazu Sakurai, J. Jepan
Inst. Metals (Serdai) 24, *818 (1960).
12. H. E. McCoy, Jr., Effects of Hydroger on the High-Temperature
Flow and Fracture Characteristics of Metals, ORNL-3600 (1964).
13. D. McLean, J. Inst. Metals 85, 468 (1956-57).
14. P. W. Davies and J. P. Dennison, J. Inst. Metals 87, 119
(1958–59).
15. K. Farrell, J. T. Houston, and A. C. Schaffhauser, "The
Growth of Grain Boundary Gas Bubbles in Chemical Vapor Deposited
Tungsten" (This Conference).
16. R. W. Balluffi and L. L. Seigle, Acta Met. 5, 449 (1957).
17. D. Hull and D. E. Rimmer, Phil. Mag. 4, 673 (1959).
18. Frank Garofalo, p. 142 in Fundamentals of Creep and Creep-
Piupture in Metals, Macmillan, New York, 1965.
24
Table I
Chemical Composition of Test Materials
Chemical Content (wt %)
Lot Number
mēNCH (ppm)
Powder metallurgy 0.0029 < 0.0005 <0.001 0.0004 5
CVD, PW-3.
0.011 < 0.0005 0.003 0.0004 21
CVD, PW-19
0.014 0.0009 0.002 0.0005 23
CVD, PW-20 . < 0.0005 < 0.0005 0.003 0.0001 16
CVD, PW-23
0.0067 < 0.0005 0.001 0.0002
CVD, PW-24
0.022 0.0014 0.004 0.0002 21
CVD, PW-18
0.0010 < 0.0005 < 0.001 0.0002 9,5
CVD, PW-69
25,28
CVD, H-16
0.0064 < 0.0005 0.002 0.0005 14
Table II
Semiquantitative Chemical Analysis
in Parts Per Million (Weight)
Element
PM
PW-3
H-16
PW-18
PW-20
0.4
0.3
0.06
0.7
0.06
2 .
<0.1
.
Ca
ca
0.7
0.2
0.06
0.2
<0.1
0.03
3
1
0.2
<0.1
0.1
0.7
<0.1
6
<0.1
0.3
<0.1
V
< 0.1
V
< 0.1
.
.
0.3
.
.
è m ra
1
0.03
1
< 0.1
V
3
3
K
V
< 0.1
0.5
0.1
0.1
2
<0.1
0.1
6
0.6
0.5
0.03
6
Mn
0.03
<0.1
0.1
0.1
0.3
0.1
30.1
0.1
20
1
3
3
0.1
0.3
0.2
<0.1
Ni
S
Ta
0
10
3
0.03
20
3
0.03
<0.1
<0.1
-
-
Table III
Void Fractions for Various Specimens
Lot
Number
Test Conditions
Rupture Rupture Void Fraction (%) Field Size
Life · Strain
the cow max min av (mm)
PW-3 1650°C, 4000 psi 413
PW-3 200°C, 2000 psi 21.3
PW-3 2200°C, 1500 psi 35.7
PW-3 2200°C, 1250 psi 790
PW-20 2200°C, 2000 psi 1.9.
PW-20 2200°C, o psie
3.2
21.9
16.6
29.2
15.6
0.35 0.07 0.19. 0.43 x 0.31
11.7 4.5 7.5 0.43 x 0.31
8.5 4.9 6.9 0.43 x 0.31
14.9 4.6 9.1 0.43 x 0.31
8.1 4.2 6.1 0.43 x 0.31
0.03 0.50 0.20 0.216 x 0.085
Same thermal history as specimen above.
LIST OF FIGURES
Fig. 1 (ORNL-DWG 67-5384) Comparison of the Creep-Rupture Properties
of CVD and PM Tungsten.
Fig. 2 (ORNL-DWG 67-5383) Comparison of the Minimum Creep Rates of
CVD and PM Tungsten.
Fig. 3 (ORNL-DWG 67-5382) Comparison of the Fracture Ductilities
of CVD and PM Tungsten.
Fig. 4 (Y-60650) Photomicrograph of PM Tungsten Sheet Tested at
6000 psi and 1650°C. Rupture occurred at 135 hr and 31.2% Strain.
As-polished.
Fig. 5 (YE-9293) Fractograph of a PM. Tungsten Specimen Tested at
1650°C and 7000 psi. 11,250%.
Fig. 6 (Y-68655) Photomicrograph of PM Tungsten Tested at 1500 psi
and 2200°C. Rupture occurred at 42.1 hr and 16.6% strain. Etchant:
50 parts NH4OH and 50 parts H202.
Fig. 7 (YE-9306) Fractograph of a PM Tungsten Specimen Tested at
2200°C and 2000 psi. 7500X.
Fig. 8 [(a) Y-67547, (b) Y-67546] Photomicrographs of As-Deposited
CVD Tungsten, Lot PW-3. Etchant: 50 parts NH4OH and 50 parts H202.
(a) Cross section; (b) parallel to surface.
Fig. 9 [(a) Y-63592, (b) Y-67536] Photomicrographs of CVD Tungsten
from Lot PW-3, Tested at 4000 psi and 1650°C. Failed in 413.0 hr and
3.2% strain. (a) Fracture, as-polished; (b) cross section 1/2 in. from
fracture. Etchant: 50 parts NH4OH and 50 parts H202.
Fig. 10 (YE-9304) Fractograph Approximately 10 mm from the Fracture
of a CVD Tungsten Specimen Tested at 4000 psi and 1650°C. Ruptured in
413 hr with 3.2% strain. 11,250%.
Fig. 11 (YE-9301) Fractograph Approximately 10 mm from the Fracture
of a CVD Tungsten Specimen Tested at 6000 psi and 1650°C. Ruptured in
12.8 hr with 3.1% strain. 7500x.
Fig. 12 (Y-67530) Photomicrograph of the Cross Section of a CVD
Tungsten specimen Trom Lot H-16 Tested at 4000 ps1 and 1650°C. failed
in 593.7 hr with 21.0% strain. Etchant: 50 parts NHA OH and 50 parts H202.
Fig. 13 (Y-79912) Fracture of a CVD Tungsten Specimen from Lot PW-20
Tested at 6000 psi and 1650°C. Fractured in 208 hr with 6.3% strain.
Etchant: 50 parts NHA, OH and 50 parts H202. 100x.
Fig. 14 (Y-79236) Fracture of a CVD Tungsten Specimen from Lot PW-69
Tested at 6000 psi and 1650°C. Warm worked 80% and recrystallized at
2200°C prior to testing. Failed in 37 hr with 3.7% strain. Etchant:
50 parts NH4OH and 50 parts H202.
Fig. 15 (YE-9349) Fractograph of a CVD Tungsten Specimen Approxi-
mately 10 mm from Original Fracture. Lot PW-69 tested at 6000 psi and
1650°C. 12,500x.
Fig. 16 (Y-79230) Fractograph of a CVD Tungsten Specimen from
Lot PW-18 Tested at 6000 psi and 1650°C. Failed in 32 hr with 13.2%
strain. Etchant: 50 parts NH4OH and 50 parts H202.
Fig. 17 (YE-9350) Fractograph of a CVD Tungsten Specimen Located .
about 10 mm from Original Fracture Surface. Lot PW-18 tested at 6000 psi.
and 1650°C. Failed in 32 hr with 13.2% strain. 12,500x.
Fig. 18 (Y-68733). Cross Section of CVD Tungsten from Lot PW-20
Tested at 2200°C and 2000 psi. Specimen failed at 1.86 hr and 15.6%
strain. Etchant: 50 parts HN4 OH and 50 parts H202.
---
Fig. 19 (Y-67542) Cross Section of CVD Tungsten from Lot PW-3
Tested at 2200°C and 2000 psi. Failed at 789.9 hr and 29.2% strain.
Etchant: 50 parts NH4OH and 50 parts H202.
Fig. 20 (YE-9348) Fractograph of a CVD Tungsten Specimen from
Lot PW-20 Tested at 2000 ps1 and 2200°C. Located about 10 mm from
original fracture. Failed in 1.9 hr with 15.6% strain. 7500x.
om
Fig. 21 (Y-80260) Photomicrograph of Specimen from Lot PW-20
Annealed 1.9 hr at 2200°C in Vacuum Without Applied Stress. Etchant:
50 parts NH4 OH and 50 parts H202.
i
aziende
mendonin tikintheti
integreeritavitha-
t te er
not the best wis ne pas moins de nomination om de
are
kita memories
mit
tor is a thing I write
minnie
m
i
n
mi
.
40,000
--
-
1
--
ORNL-DWG.67-5384
-P.M.
O P.W.-3
H-16
P.W.-20
A P.W.-24
O P.W.-23
OP.W.-19
O P.W.-18
P.W.-69
10,000
STRESS (PS1)
to-0-
1650°C
1,000
0.1
101
2200°C
1.0
100
1000
RUPTURE LIFE (HRS.)
COMPARISON OF CREEP-RUPTURE PROPERTIES OF CVD AND P.M. TUNGSTEN
10,000

.
.
.
.
. .
. .
-
..
--
-
-----
-------
40,000
ORNL-DWG. 67-5393
- P.M. .
O P.W.-3
O H-16
A P.W.-20
A P.W.-24
OP.W.-23
2 P.W.-19
O P.W.-18
P.W.-69
1650°C
10,000
STRESS (PSI)
Hooho-
52200°C
1,0000.001
0.1
1.0
100

0.01
Min.CREEP RATE %/HR.
COMPARISON OF MINIMUM CREEP RATES OF C.V.D. AND P.M. TUNGSTEN
.1
.
1...
.
..
.
.
..
.
A
Lat
.
.
.
1
r
ORNL-DWG.67-5382
- P.M.
IC:P.W.-3
H-16
A P.W.-20
A P.W.-24
O P.W.-23
@ P.6.-19
O P.W.-18
P.W.-69
ELONGATION (70)
- 1650°C
220090
L
22ooc | 22ooºco
1650°C
22200°C
| 02200°C
165000
2200
o 1650°C
1650°C 1650°C
2200°C
1650°C
1650°C
10
100
RUPTURE LIFE (HRS)
COMPARISON OF FRACTURE DUCTILITIES OF CVD AND P.M. TUNGSTEN
1,000


XOOL al
N
w

47
common naman
*.
-
painted her
"
media
** seco you."
(4;500X Pro Cha metine) the
.
www
.ccm....

-- 0.035 INCHES
XOOL Ni
و . د ۰۰۰
.
.
سمسودها . مسسسسس
مممممممممممممممممممممممممممممسسینه سمسمههسميضمسسيه سمه
ده مهمه .... ... . ... .
.
.
......
...........................
ر
و
من ارادے غم
.
/25X) .../

,
ا
:نن
هه
IN 100x
0.035 INCHES
) / همت
داده
-
:
:
م . مست
. .-ج-


.
تم ایک
- سمسمه-
-
,•.
ماند
شیشه --ست نيه
---
;
سمنتينسننمعنسكستحسن عندنسنيستنشت
-0.035 :NCHES -
N 100x
امرس
-
-
- -
-
سهم ده. .
. مب متن.:.
-
.
ننوت": لا
. . -ه.
. د
اس
بچ
"
. .
میره و .
.
.2
ج
ہ
.و .
د
*
..
...
له
د... مهد ... ...
دد
و
.
.
. .
.
-
.
.
-
، .. .
عم
...
- ۰۲
۰
۰ و
ه
ه . .
-
-
-
-
و
هم

بب
:
....
به x000
SJHONI SEOʻ0-
...
...
.
.
... .
.
.. ..
.. .


(
. .
.
. .
. . .
. - . - -
- -
که م
المسيلمسهممکن، به من بدهیم به مسلم
مار کر۔
In 100x
-0.035 INCHES -
تسمملسننيننسننمنننننلننسس مسمنننننننننننهعهننننننننننننينتينسكسمعه
سے۔
......
سب ہی
.
* من
"حراست
..
سیمسنیم
-
-
-
.
. - رسمن

(
(
.
.... .... .................................
....

，，• • ，
,，
་
ཟ， ་ ，
..་
བ • •
བ
t
་་ ་ཙཁས་
འབམ -
བ•ས་
. .
་
• ་
-
ཀ
ས་



..
....
..
..
...
...
.........
.
.
.
o menor....
m.. niol..
-
4
--
nahon
.....
-
-0.035 INCHES -
.'
*
:
-
*
A
Y
.
-
ins inny...
...
-
-
-
-
-
.

- -
- -
-
-
م حمد -
-
- -
- - - - -
-
.
.
.
.
.
. .. .
.
۱۰۰۰:
.............
م . م .. ..
"
. .
. .. .
.
.. .
.
-م . .
س تمبر
.... .
.جم موه-
....
::
.
۰۰۰
...مهمه. ..
.
-----
د.مص
. . ته
..
ه --
ه م
ت
.. ..
..
.
هه
ده.......................
۰۰۰۰
مهمه ممه
.
....
.
....
.
-0.035 INCHES -
N 100x
. .
. ..
.
.
.
*
*
*
. م .
-
.
- .
-
. -
-
-
- -
-
. - -
. -
-
مد --- - .
- . م .
عيد
.. .
. ..
.
.. هه .
.
"
لا
"
. تصمصممه
....
... اممممممممم
تتنفسشانسحضمحسسنمنتضدعتخشمس
-
5
- 0.03 1:
1
N 0cx
. .
.
. .
. . .
.......

X
.
-
..........
.......


-0.035 INCHES ---
IN 10CX

miwili
4,7
i
n dimento mainan orzamentine content
.
.
is
.........
..
..
دم
من با
.
.
.
.
.





... ons hom
in
10
... you
....
....
................
co sooo • ... :
...
....................
., imo .....
0.035 CHES-
00!
-
!
-
-
-
-
-
-
-
..............
.
.

..'ivi
-
:
.
. -
-
-
-
-
-
-
-
-
-
-
-
-
1
-
.
-
-
-
-
--
-
-
-
-
-
-
-
-
-
IN 100%
0.035 INCHES
..0.000 :
TODO
--...----
--....
...
................
.......
..
....
.
.....
.............


)


་
ལ་
༌ བ བ བ
་ བ
-
-
.
，
་ ་ ，
་་་་
་
བ བ
་བ -
，
ཁག
/7/
/75-
2-Z2_
ཀྱི་་༤
བ་
evvཀ
37::ལ་ བ ， ， པར པ

ای x00
-S3HONI SEO'O-
تمہ.سے
کے
.. مي
به ....
۰۰:".
::: مه
.نمره عرب
نه
.
..
مد
:
. :::
.
. .
...
.. .
.
..
.
.
ه
.
.
. .
. ----
. .
منننننهنسسسسس
. .. .. .. .. .. .. ..... ما . مهم ....... .
سنة: 00
بببببب
*
ق...
.
. . . .
.
. .
. . .
. .
| :::
...
. ..
.
.
..
.
.
م
ن
:
:..
.
.
.
.
مهمتنا
۔ م عه ::
eticے
-
-
-
9 / 8 / 67
DATE FILMED
END

.
.
..
.
.
.:
*
**
.
-
FIL
.
..
.
ET