You are on page 1of 12

Materials"Science and Engineering, A 110 (1989) 49-60 49

The Cavitation and Fracture Characteristics of a Superplastic


AI-Cu-Li-Zr Alloy

A. H. CHOKSHI and A. K. MUKHERJEE


Division of Materials Science and Engineering, Department of Mechanical Engineering, Universi~ of California, Davis, CA 95616
(U.S.A.)
(Received September 25, 1987; in revised form June 24, 1988)

Abstract Two different thermomechanical processing


routes may be employed to develop fine grain
An A l - C u - L i - Z r alloy processed to recrystal-
sizes conducive to superplasticity in quasi-single-
lize statically exhibits superplastic characteristics at
phase A1-Li alloys [1]. In the first case the fine
a testing temperature of 723 K, with an optimum
grain sizes are developed by solution treating,
elongation to failure of 560% at a strain rate of
overaging, warm working and statically recrystal-
3 x 10-4 s-1. Under optimum conditions the alloy
lizing the alloys prior to superplastic deformation
exhibits diffuse necking, whereas at strain rates
[2-7]. In the second case the alloy composition
higher and lower than the optimum value there is
and experimental conditions are chosen to facili-
an increasing tendency for sharp necking. It is
tate the occurrence of dynamic recrystallization
shown that the optimum superplastic elongation is
during superplastic deformation [8-11]. Both of
obtained by a suitable combination of the strain
these procedures have been used with varying
rate sensitivity and the strain-hardening coefficient.
degrees of success, and it is now possible to
The alloy exhibits considerable cavitation under all
obtain elongations to failure in the range
experimental conditions, with cavitation being
500%-1000% in such AI-Li alloys.
restricted to regions near the fracture tip in speci-
There have been numerous studies dealing
mens which failed with a sharp neck, and cavita-
with superplasticity in A1-Li alloys [1-24], but
tion being more uniformly distributed in specimens
relatively little attention has been devoted to
with a diffuse neck. Cavities are nucleated during
studying cavitation and fracture in these alloys
superplastic deformation at coarse Al-Fe-Cu parti-
[10-13, 15-19, 22]. It has been demonstrated
cles, and cavity growth occurs predominantly by a
previously that even small levels of cavitation
power-law mechanism. It is shown that surface dif-
developed during superplastic deformation may
fusion at elevated temperatures may lead to the
have a deleterious effect on subsequent room
observation of rounded cavities, even when cavity
temperature properties [25-27], and that in some
growth has occurred by a power-law growth
cases excessive cavitation may lead to premature
mechanism.
failure in superplastic alloys. The present study
was therefore undertaken with the objective of
1. Introduction characterizing cavitation and fracture in a super-
plastic AI-Li alloy.
It is now well known that additions of lithium
lead to significant increases in the stiffness and to
2. Experimental material and procedure
concomitant decreases in the density of alumi-
num-based alloys. Therefore AI-Li alloys have The alloy chosen for this study had the
considerable potential for applications in the following nominal composition in wt.%:
aerospace industry where weight savings are of A l - 2.6 % C u - 2.4%Li-0.2%Zr-0.07%Fe. The
paramount importance. The use of superplastic alloy was obtained in the overaged and warm-
forming to manufacture components from A1-Li worked condition and had been thermomechani-
alloys may lead to additional benefits in cost and cally processed to recrystallize statically prior to
weight savings and, consequently there is a con- superplastic deformation.
siderable interest in developing superplasticity in Tensile specimens with a gauge length of 6.4
AI-Li alloys. mm were machined from the as-received material

0921-5093/89/$3.50 © Elsevier Sequoia/Printed in The Netherlands


50

with the tensile axes parallel to the rolling direc- peratures, and this led to a microstructure con-
tion. The specimens were annealed in air at sisting of uniform fine grains with dimensions
773 K for 11 h to allow for static recrystallization similar to that in the fine-grained core [21, 23].
and the stabilization of microstructure. Tensile Prior to tensile testing, the fine-grained core had
tests were carried out in an Instron machine an initial spatial grain size d = 5.4 pro, where
modified to operate at constant true strain rates. d = 1.74L with /~ the average linear intercept
The specimens were tested at a temperature of grain size.
723 K and at strain rates between about 10 -5
and 10-l s-~ to determine the variation with
3. Experimental results
strain rate in both the flow stresses and the elon-
gations to failure. A few additional experiments 3.1. The variation in flow stress with strain rate
were performed in which tensile tests were inter- Figure 2 illustrates the variation in flow stress
rupted prior to specimen failure. a with strain rate g for the experimental condi-
The cross-sectional areas of fractured speci- tions used in this study. It is clear that the A1-Li
mens were measured along the gauge length. As alloy exhibits a sigmoidal relationship between
illustrated in the schematic Fig. 1, the cross- the flow stress and the strain rate, which is typical
sectional areas were measured from one end of of superplastic alloys. The data shown in Fig. 2
the gauge length, /, and these data were normal- may be divided into three regions: a region I at
ized with the total gauge length at fracture, Lf, to f < 10 -4 s-l, a superplastic region II at strain
facilitate a direct comparison of data from speci- rates between about 10 -4 and 3 x 10 -3 s -1 and a
mens with varying gauge lengths at fracture. region III at g > 3 x 10 -3 s -J. The strain rate
The deformed specimens were sectioned sensitivity m, given by the slope of the curve in
parallel to the tensile axis and polished metallo- Fig. 2, is about 0.45 at strain rates between about
graphically. The final stage of polishing was con- 10 -4 and 10 -3 s-l, and the value of m decreases
ducted with MgO powder using very light both at g< 10 -4 s- 1 and at g> 3 × 10 -3 s- ~. The
pressure. The development of internal cavitation vertical arrows in Fig. 2 show the strain rates used
during deformation was studied by examining the in the following study on flow localization and
polished specimens by both optical and scanning cavitation.
electron microscopy (SEM). The cavity nuclea-
tion sites were identified qualitatively using 3.Z The variation in elongation to failure with
energy-dispersive X-ray analysis in an SEM. The strain rate
polished sections were then etched for about 20 s Figure 3 illustrates the variation in elongation
in a solution of 1.5 ml HC1, 2.5 ml HNO3, 2 g to failure with strain rate. The maximum elonga-
NaF and 195 ml H 2 0 to determine the variation tion to failure of 560% is obtained at a strain rate
in grain size with experimental conditions. The of 3× 10 -4 S-l. A comparison of Figs. 2 and 3
fracture surfaces of selected specimens were also
characterized by SEM. 60
Inspection of the alloy after annealing revealed I I ' ' ' ' ~'"1
the presence of a bimodal grain size distribution T=7~K
at the surfaces and a fine-grained core. Prelimi-
nary experiments indicated that the coarse grains
at the surfaces recrystallized dynamically during 10 ~ s ~ , , j " ~ l 0.45
the initial stages of deformation at elevated tern-

10-5 10-4 10-3 10-2 I0-I


(s -1 )
i-. Lf --I
Fig. 2. The experimentally determined variation in the flow
Fig. 1. A schematic illustration of the procedure used to stress with strain rate for the AI-Li alloy. The vertical arrows
normalize the gauge lengths of specimens with varying indicate the strain rates used to study flow localization and
lengths at fracture. cavitation.
51

800
I I ........ I
AI - 2.6% Cu - 2.4% Li - 0 . 2 0 Zr IL
T = 723 K
600

400
!1

200
¢
o I 1 , , ...... I
10-5 10-4 10-3 10-2 10-1

(s -1) II
Fig. 3. The variation in elongation to failure with initial
strain rate, where AL is the change in gauge length and L , is
the initial gauge length.

indicates that the optimum elongation to failure is


obtained at strain rates corresponding to the
lower end of region II. Similar observations have
been reported in some previous studies on super- F
plasticity [28, 29]. However, it is important to
note that the present study was performed under
conditions of constant true strain rates, whereas ¢111
I ' I
most of the previous studies were conducted
0 5 10
under conditions of constant engineering strain
rates in which the true strain rate decreased con- Fig. 4. Photograph of specimens pulled to failure at a tem-
perature of 723 K. Specimen A was not tested, while the
tinuously as the specimen elongated. other specimens were tested at the following strain rates: B.
1.3×10 = s i; C, 1 . 3 x 1 0 ~ S ]; D, 3 x 1 0 4 s t; E,
3.3. The localization offlow 1 . 3 x 1 0 ~s ~;F, 5 x l 0 ~s ~ .
Figure 4 is a photograph of some of the speci-
mens pulled to failure in this study. Specimen A
was untested and the strain rate used to pull the uniform deformation. Inspection of the data in
specimens shown ranges from a high value of Fig. 5 reveals clearly that deformation is quite
1.3 x 10 -2 s -l, specimen B, to a low value of uniform under optimum conditions and that
5 × 10- s s- l, specimen F. It is clear that under the departure from the optimum strain rate leads to a
optimum conditions depicted by specimen D, greater tendency towards flow localization.
there is not much flow localization and the speci-
men exhibits a diffuse neck. However, at high 3.4. The occurrence of cavitation at
strain rates, specimen B, and at low strains rates, ~= 1.3x 10 : s -j
specimen F, there is evidence for an increasing Figure 6 is a scanning electron micrograph of
tendency towards flow localization and necking. the fracture surface of the specimen tested at the
Figure 5 presents in a quantitative manner the highest strain rate used in this study. Fracture
above qualitative description regarding flow appears to have occurred in a mixed inter-
localization. In Fig. 5 the percentage reduction in granular-ductile rupture mode. Cavities with
area is plotted against the normalized gauge dimensions of up to about 50 ~m are clearly
length at failure for the specimen tested under visible on the fracture surface.
optimum conditions and also for the specimens Figure 7 is an optical micrograph of a region
tested at the highest and the lowest strain rates near the fracture tip of the specimen pulled to
used in this study. It is to be noted that a plateau failure at a strain rate of 1.3 x 10 -z s ~. Elon-
in these curves is indicative of the occurrence of gated cavities with dimensions of about 50 /~m
52

100

80

/
6o/
<co

40

10
0.2
I
~:(S-1)
• . 5 x 10"5
0 3 x 10`4

I
0.4
T-723 K

n l . 3 x 10-2
t~L / LO (%)

I
0.6
360
560
180

I
0.8
l\
1.0
m,-

50 p m
t-----q
.Q/Lf
Fig. 7. Optical micrograph of a region near the fracture tip of
Fig. 5. The variation in the reduction in area with the the specimen pulled to failure at g= 1.3x 10 -2 s-L The
normalized gauge length at failure for the specimens tested at tensile axis is horizontal.
the highest, optimum and lowest strain rates used in this
study; A~ and AA are the original cross-sectional area and
the changes in cross-sectional areas respectively. The vertical
arrows indicate the locations of the fracture tips.

1150 Pm I

Fig. 8. Optical micrograph of a region near the middle of the


gauge length of the specimen pulled to an elongation of 300%
at g = 3 x 10 -4 s ~. The tensile axis is horizontal.

Fig. 6. Scanning electron micrograph of the fracture surface


of the specimen tested at g = 1.3 × 10 -2 s-~.
3.5. The development of cavitation at
g= 3 x 1 0 - 4 S -1
Figure 8 shows an optical photomicrograph of
are observed near the fracture tip. An interesting a region near the middle of the gauge length of a
aspect of Fig. 7 is that cavitation is localized in a specimen pulled to an elongation of 300% at a
region near the fracture tip and there is very little strain rate of 3 x 1 0 - 4 S -1. The left side of the
evidence for cavitation in regions well away from micrograph corresponds to the center of the
the fracture tip. Microstructural inspection after gauge length where the true strain was measured
etching revealed that the grains tended to be elon- to be 1.54, and the right side of the micrograph
gated along the tensile axis, and measurements corresponds to a region about 1 mm away from
indicated that the grain size near the fracture tip the middle of the gauge length where the true
was about 8.4/~m. Near the fracture tip, measure- strain was measured to be 1.39. It is to be noted
ment of the local cross-sectional area indicated that cavitation is restricted to a small region at the
that the true strain was 1.37. middle of the gauge length, and regions even
53

~_30 p m I

Fig. 9. Optical micrograph of a region near the fracture tip Fig. 10. A higher magnification optical micrograph of the
of the specimen tested at g = 3 x 10 a s - ~.The tensile axis is specimen pulled to failure at g=3 x 10 -4 S I. The tensile
horizontal. axis is horizontal.

about 1 mm away from the middle of the gauge


length exhibited no cavitation. A considerable
amount of grain growth had occurred during
superplastic deformation, with the grain size
increasing from an initial value of 5.4 /~m to
values of 15 and 12 /~m in regions where the
local true strains were 1.54 and 1.39 respectively.
Figure 9 is an optical photomicrograph of a
region near the fracture tip of the specimen
pulled to failure at a strain rate of 3 × 10-4 S-1.
Inspection of Fig. 9 reveals that there is consider-
able cavitation under these conditions and also
that, in contrast to the specimen deformed at high
strain rate (Fig. 7), cavitation is not restricted to a
small region near the fracture tip. The larger
cavities tend to be elongated along the tensile
axis, whereas the smaller cavities tend to have a
rounded morphology. The true strain and the
Fig. 11. Scanning electron micrograph of the specimen
grain size near the fracture tip were measured as pulled to failure at g=3x 10 -4 s n. The tensile axis is
2.65 and 1 9 / t m respectively. horizontal.
Figure 10 is a higher magnification optical
photomicrograph and shows the occurrence ot
initial stages of deformation but occurred only in
grain boundary cracking* in a region near the
the later stages of deformation. Thus, in the
fracture tip. Figure 11 is a scanning electron
present study, grain boundary cracking was not
micrograph of a region near the fracture tip and
observed in the specimen pulled to an elongation
illustrates the nucleation of cavities at coarse of 300% but was quite apparent in the specimen
particles along with the occurrence of grain boun- pulled to an elongation of 560% (failure). Similar
dary cracking. Energy-dispersive X-ray analysis observations of grain boundary crack networks in
revealed that the cavity nucleation sites were other superplastic quasi-single-phase aluminum
A1-Fe-Cu particles. It is important to note that, alloys were reported earlier by Pilling and Ridley
in a manner similar to an earlier observation in a
[15, 16].
commercial superplastic copper alloy [30], grain
boundary cracking was not observed during the 3.6. The occurrence o f cavitation at
g= 5 x lO-~s -I
*Etching the specimen after obtaining Figs. 10 and 11 Figure 12 is an optical photomicrograph
revealed that the cracks observed were grain boundary
cracks. obtained from a region near the fracture tip of a
54

accordance with the observed decrease in strain


rate sensitivity. Similar trends in the variation in
g -d flow localization with m were reported in earlier
studies by Mohamed and Langdon [35] and
Ahmed and Langdon [36].

4.2. The relationship between strain rate sensitivity


and ductility
The stress-strain rate characteristics of super-
plastic alloys may be divided conveniently into
three regions: a superplastic region II with
m >-0.5 at intermediate strain rates and the non-
superplastic regions I and III with m ~ 0.3 at low
.2L
t 150 Pm~l and high strain rates respectively. The elong-
ations to failure obtained in superplastic alloys
Fig. 12. O p t i c a l m i c r o g r a p h of a r e g i o n n e a r the f r a c t u r e tip
of the s p e c i m e n t e s t e d at g = 5 × 10 - s s - ~. generally follow the variation in m with strain rate
[37], so that maximum elongations to failure are
obtained in region II and the elongations to
specimen pulled to failure at a strain rate of failure decrease in regions I and III. However, as
5 x 10-s s-1 and provides clear evidence for the noted earlier by Langdon [38], a careful examina-
occurrence of extensive cavitation under these tion of the available data suggests that a high m
conditions. The large cavities with dimensions of value is not the only parameter governing the
up to about 200 /am are elongated along the elongation to failure. Thus, for example, although
tensile axis and there appear to be a few smaller a well-defined region II with a constant m extends
cavities that have a rounded appearance. Mea- over two orders of magnitude of strain rate in the
surements indicate that the local strain near the superplastic Zn-22%A1, the elongation to failure
fracture tip is 2.92 and the grain size at this varies from a maximum value of about 2900% to
location is 24/am. a low value of about 1400% in region II [28]. In
addition, a close inspection of data reported in
several superplastic alloys indicates that, fre-
4. Discussion
quently, the optimum elongations are obtained at
4.1. The influence of strain rate sensitivity on flow strain rates in the lower end of region II. Obser-
localization vations of this nature are apparent in the earlier
It has been recognized for a long time that studies of Ahmed and coworkers [28] and Shei
superplastic materials exhibit extremely large and Langdon [29]. One of the difficulties in inter-
elongations to failure by virtue of their high strain preting the previous results is that the experi-
rate sensitivities m. It is known that a high value ments were performed under conditions of
of m enhances ductility, primarily by conferring constant cross-head speed so that the true strain
on superplastic alloys a high degree of resistance rate declined continuously as the specimen elon-
to flow localization and necking. Following the gated. The present results shown in Fig. 3 were
early work of Hart [31 ], several theoretical [32, 33] obtained under conditions of constant true strain
and experimental [34-36] studies have been rate testing and they show the occurrence of opti-
undertaken to examine the influence of m on flow mum elongation to failure at the lower end of
localization. region II.
The experimental data obtained on flow locali- There are three possible causes for a lack of
zation in the present study are in broad agree- observation of maximum ductility under condi-
ment with the trend in the variation in m with tions of maximum m: (a) the value of m changes
strain rate, Fig. 2. Thus under optimum condi- during a tensile test, (b) more cavitation occurs in
tions there is very little evidence of flow localiza- region II than in region I, and (c) more strain
tion, as illustrated by specimen D in Fig. 4 and hardening occurs under other conditions and this
the data in Fig. 5. At strain rates higher and lower makes an additional contribution to the resist-
than the optimum value there is a greater ten- ance to necking. These three factors are now con-
dency towards flow localization and this is in sidered in more detail.
55

There is a possibility that the value of m typically equal to about unity in region II. The
changes during a tensile test so that m determined experimental measurements indicate that the
from the initial stages of deformation (Fig. 2) may grain growth per unit strain, Ad/Ae, is equal to
not reflect the true resistance to necking during 6.4, 5.1 and 2.2/~m for the specimens pulled to
an entire tensile test. Following the earlier sugges- failure at strain rates of 5 x 10-5, 3 x 10 -4 and
tion of Ghosh and Ayres [39], Lian and Baudelet 1.3x 10 -' s -I respectively. These results are
[40] have shown analytically that ductility is consistent with previous detailed analyses of data
governed largely by the value of m just prior to on other superplastic alloys [47] and they imply
failure. However, a recent detailed study by that the rate of strain hardening increases with a
Caceres and Wilkinson [41] revealed that the decrease in strain rate, since the flow stress
value of m remains essentially constant during increases with grain growth.
superplastic deformation. Therefore it is sug- Based on the above discussion, it is concluded
gested that while the value of rn may change that the optimum elongations are obtained at the
during a tensile test, the occurrence of maximum lower end of region II, because under such condi-
ductility at strain rates towards the lower end of tions an appropriate combination of strain rate
region II is not caused by a change in m during sensitivity and strain hardening provides the
tensile tests. maximum resistance to flow localization. At
It is now recognized that most superplastic strain rates in the middle of region II the strain
materials cavitate during deformation and that rate sensitivity is high but the contribution of
failure is caused frequently by the process of cavi- strain hardening to necking resistance is low; in a
tation [42-44]. Therefore a difference in levels of similar manner, at lower strain rates in region I,
cavitation with strain rate may lead to a lack of a although the contribution of strain hardening is
correlation between maximum ductility and high, the strain rate sensitivity is low.
maximum m. However, a detailed inspection of
micrographs such as Figs. 9 and 12 suggests that 4.3. The nucleation of cavities
in the present study there is very little difference The nucleation of cavities is an important first
in the levels of cavitation in regions I and II. An step in the process leading to cavitation failure in
essentially identical observation was reported many superplastic alloys. In the present study,
recently by Caceres and Wilkinson [45]. It is metallographic inspection of deformed speci-
therefore concluded that the occurrence of maxi- mens reveals that cavities nucleate preferentially
mum ductility at the lower end of region II is not at coarse AI-Fe-Cu particles. In addition, both
caused by a difference in levels of cavitation. the increase in the number density of cavities and
As shown by Hart [31], the resistance to flow the distribution of cavity sizes observed at any
localization is generally provided by a combina- given strain, as noted from Figs. 8 and 9, suggest
tion of two factors: (a) a high value of m and (b) a that cavities nucleate continuously during super-
high value of the strain-hardening coefficient 7 plastic deformation. This observation is similar to
(= 6l~0/6e). In early work it was assumed that a recent report on cavitation in a superplastic
superplastic materials exhibit negligible strain commercial copper alloy [48].
hardening and therefore the resistance to flow Stowell [42] has suggested that, because of the
localization and ductility were correlated directly extensive thermomechanical treatment applied to
with rn [37]. Hamilton [46] has recently examined induce a fine grain size, cavities may exist prior to
the combined effect of m and 7 on flow localiza- elevated temperature deformation in many super-
tion during superplastic flow. In the present study plastic alloys. There is some experimental evi-
it is anticipated that the considerable grain dence indicating that coarse particles may crack
growth observed during deformation will lead to during ambient temperature deformation in
strain hardening* because, in superplastic A1-Li alloys [49, 50]. However, a detailed inspec-
materials tested at a constant strain rate, o oc d', tion of the alloy used in the present study failed to
where d is the grain size and the exponent c is reveal the existence of cavities prior to super-
plastic deformation. In addition, there was no
evidence of particle cracking in specimens either
*As noted previously [21], the occurrence of strain hard- before or after superplastic deformation, Fig. 11,
ening may not be apparent in stress-strain curves, because
the flow stresses may be underestimated in the presence of
and cavities appeared to nucleate during super-
diffuse necking. plastic deformation. In this context it is important
56

to note that, because of the low flow stresses increased considerably, the rate of grain growth
associated with superplastic flow, only cavities decreases and this reduces the level of accommo-
with fairly large dimensions may remain stable dation to grain boundary sliding [52]. However,
against surface tension forces, and cavities with recent careful studies of concurrent grain growth
radii less than r 0 (= 2v/a, where v is the surface in several superplastic alloys [47], including a
energy) will tend to be sintered at elevated tem- Zn-A1 alloy, indicate that the rate of grain growth
peratures. For the present study, using typical remains constant throughout a tensile test.* Con-
values of o = 10 MPa and v = 1 J m-2, the value sequently, a decrease in the rate of grain growth
of r 0 is determined to be about 0.5/~m. Cavities does not appear to be a plausible explanation for
with these dimensions should be clearly resolved the nucleation of cavities during the later stages of
by SEM, so that the failure to observe any pre- deformation. The present experimental results
existing cavities is not related to the lack of reso- from a specimen pulled to an elongation of 300%
lution. under optimum conditions appear to indicate that
Bampton and Edington [51] and Pandey et al. a critical grain size exceeding about 12 /~m is
[17] have shown that hydrogen gas bubbles may necessary to nucleate cavities. However, an
form in aluminum-based alloys during exposure inspection of the specimen pulled to failure at
to elevated temperatures. In the present study, 1.3 x 10 -2 s- ~ reveals that near the fracture tip
microstructural examination of annealed speci- where cavitation occurs, the grain size is about
mens prior to tensile testing did not reveal the 8.4 ,um. These results indicate therefore that the
presence of hydrogen gas bubbles. However, it is critical grain size to nucleate cavities depends
important to note that in the presence of gas pres- upon the experimental conditions. It is suggested
sure, cavities with dimensions less than r 0 may that the occurrence of grain growth is not related
remain stable, so that it is not possible to rule out directly to the nucleation of cavities, but that con-
completely the existence of very small gas current grain growth assists indirectly in nucleat-
bubbles. ing cavities by increasing the flow stress during
An interesting aspect of this study is the obser- deformation. A similar suggestion was incorpor-
vation that cavitation does not occur during the ated by Ghosh [53] in a theoretical model for con-
initial stages of deformation. Thus, for example, tinuous cavity nucleation in a superplastic
cavitation was not observed under optimum con- aluminum-based alloy.
ditions in a specimen pulled to an elongation of
100%, and even in a specimen pulled to an elon-
gation of 300%, cavitation was observed only in
4.4. The rate-controlling cavity growth
the highly deformed central region of the gauge
mechanisms
It is now recognized that in superplastic
length. Microstructural inspection also revealed
materials cavities, upon nucleation, may grow
the occurrence of concurrent grain growth during
predominantly by one of three independent
deformation, such that in the specimen deformed
mechanisms [54-56]: (a) diffusion, (b) super-
under optimum conditions to an elongation of
plastic diffusion and (c) power-law. The cavity
300%, the grain size had increased from an initial
value of 5.4/~m to values of 15 and 12 ~m in the growth rates per unit strain, dr/de, for these three
highly deformed region with cavities and in a mechanisms are given by the following expres-
region with no cavities about 1 mm away from the sions:
highly deformed region respectively. In the speci- (a) diffusion cavity growth in superplasticity
men tested at a strain rate of 1.3 x 1 0 - 2 S-1 in [55, 57]
region III, cavitation was observed near the frac-
ture tip, and the grain size in the cavitated region d__[r= ffa6D~b( O-- 2 v/r)
of the specimen was measured as 8.4/~m. de 5r2kTg (1)
Livesey and Ridley [52] studied cavitation in
superplastic Zn-22% A1 eutectoid alloys and con- where ff2 is the atomic volume, 6 is the grain
cluded that cavities are nucleated only after the boundary width, Dg b is the grain boundary diffu-
initial fine grain size has increased by concurrent
grain growth to a critical value of about 8 /.tm.
*It is important to note that in most of these studies, con-
These results were rationalized by noting that current gram growth was correlated with the true local strain
after high strains, when the grain size has along the gauge length.
57

sivity, k is Boltzmann's constant and T is the Dgb=5 × 10 14 e x p ( - 8 4 0 0 0 / 8 . 3 T ) m 3 s -~ [59],


absolute temperature; T = 723 K, o/{ = 1.2 x 105 MPa s, the average
(b) superplastic diffusion growth [54-56] grain size d a = ( d i + d f ) / 2 = 15 Mm, where di and
dr are the initial and final grain sizes respectively.
dr_45QODgbo (2) Inspection of Fig. 13 reveals that the super-
de k Td-g plastic diffusion mechanism is not important
under the present experimental conditions. In
(c) power-law growth [58] addition, Fig. 13 indicates that cavities with
dimensions greater than about 1 ,urn should grow
dr 3v by the power-law mechanism and that the diffu-
-r (3)
de 20 sion mechanism is not very important. Since the
diffusional cavity growth rate is inversely propor-
Since these three mechanisms operate inde- tional to the imposed strain rate and the power-
pendently, cavity growth is controlled by the law growth rate is independent of the strain rate,
mechanism giving rise to the highest value of it is anticipated on the basis of Fig. 13 and the
dr~de. Equations (1)-(3) were evaluated for appropriate cavity growth maps [55] that at strain
g = 5 x 10-5 s ~, corresponding to deformation rates higher than 5 x 10 5 s-1 (used in Fig. 13)
in region I, and the results are shown in Fig. 13 as cavity growth will occur largely by the power-law
a logarithmic plot of dr~de vs. r. The follow- mechanism. Therefore it is concluded that cavi-
ing values were used for the parameters in ties grow largely by the power-law mechanism for
eqns. (1)-(3): q2= 1.7 x 10 -29 m 3, v = l . 1 J m -2, all of the experimental conditions used in the
present study. Similar conclusions were reported
in previous studies on cavitation in superplastic
102 J I l i I IIIIJ A1-Li alloys [15-17, 60].
~l "2.6*/. Cu - 2.4% li - 0.2% Zr In general, cavities growing by a diffusion
T-723 K mechanism will tend to be rounded, whereas
G / ~ = 1.2x 1 0 5 M P a s / those growing by a power-law mechanism will
/ tend to be elongated along the tensile axis. It has
10
/ been demonstrated that the trend in the change in
/
cavity morphologies is consistent with the
Power-Law~ theoretical predictions of a transition from diffu-
Growth /
/ . . . . sion to power-law growth in several superplastic
1.0 alloys such as Zn-22%A1 [56, 61], copper-based
E alloys [54-56, 61-63], stainless steel [61] and
c~ some aluminum-based alloys [60, 64]. However,
"o / Growth , in the present study, while the theoretical predic-
tions imply that cavity growth occurs largely by a
10-1 power-law mechanism, the experimental observ-
ations reveal the presence of several small
rounded cavities. A similar inconsistency

10"2
I
I
I
I
' i between cavity shapes and the dominant cavity
growth mechanism was reported earlier by Ridley
et al. [65]. A possible cause for this discrepancy is
the effect of surface diffusion at elevated temper-
atures.
I Hancock [58] developed the power-law growth
mechanism for high temperature creep by modi-
10.3 [ l , , ,,,,,,I
10-1 1.0 10 10-2
fying the low temperature plasticity-induced
cavity growth model of McClintock [66]. Plastic-
r (lain)
ity-controlled cavity growth at low temperatures
Fig. 13. The variation in cavity growth rate with cavity leads to cavities that are elongated along the ten-
radius for a specimen tested at ~ 5 × 10 -5 s t. The theoreti-
cal curves are shown for the diffusion, superplastic diffusion
sile axis. Figure 14 is a schematic illustration for
and power-law growth mechanisms. the development of rounded cavities from
58

or 5. Summary and conclusions


f 1. An A1-Cu-Li-Zr alloy processed to re-
A crystallize statically exhibits superplastic charac-
teristics at a temperature of 723 K. The alloy
exhibits a maximum elongation to failure of
560% at a strain rate of 3 x 10 -4 s -~, and the
elongation to failure decreases both at lower and
at higher strain rates.
2. Under optimum conditions the fractured
la) (b) [cl specimen exhibits diffuse necking. At strain rates
higher and lower than the optimum value there is
an increasing tendency towards flow localization
or and necking. The maximum superplastic elonga-
tion to failure is obtained at an optimum strain
Fig. 14. A schematic illustration of the development of
rounded cavities during elevated temperature testing. rate through a suitable combination of the strain
Initially, a rounded cavity (a) becomes elongated along the rate sensitivity and the strain-hardening coeffi-
tensile axis due to cavity growth by a power-law mechanism cient.
(b). The difference in the radii of curvature at the points
marked A and B then leads to the spheroidization of the 3. The alloy cavitates considerably under all
cavity (c). The tensile axis is vertical. of the conditions used in this study. Cavitation is
restricted to a region close to the fracture tip at
high strain rates, where the specimens fail by the
development of a sharp neck, and cavitation is
elongated cavity shapes during deformation at
more uniformly distributed at lower strain rates,
elevated temperatures. Initially, diffusion growth
where specimens fail by diffuse necking.
may lead to the development of a rounded cavity
4. Cavities nucleate during superplastic defor-
as shown in Fig 14(a). Subsequent growth of this
mation at coarse AI-Fe-Cu particles prese&t in
cavity by a power-law mechanism will tend to
the as-received material. There is no microstruc-
elongate the cavity along the tensile axis as shown
tural evidence for pre-existing cavities or cracked
in Fig. 14(b). In experiments performed at low
particles. The observed concurrent grain growth
temperatures, the plasticity-induced elongated
leads to an increase in the flow stress and this
cavity shapes will remain elongated. However, at
facilitates cavity nucleation during the later stages
high testing temperatures the difference in the
of superplastic deformation.
radii of curvature at the points marked A and B in
5. Cavity growth occurs predominantly by a
Fig. 14(b) will lead to a redistribution of matter
power-law mechanism. It is shown that elongated
by surface diffusion, so that the cavities will tend
cavities growing by a power-law mechanism may
to become spherical as shown in Fig. 14(c). This
develop a rounded morphology because of sur-
possibility was first recognized and suggested by
face diffusion at elevated temperatures.
Rice [67, 68]. The overall rate of spheroidization
will tend to depend on relative rates of surface
diffusion, cavity elongation by the power-law Acknowledgments
mechanism and grain boundary sliding and grain
This work was supported by the Air Force
rotation, which will also tend to distort the shape
Office of Scientific Research under Grant no.
of a cavity. This is quite a complicated pheno-
86091 monitored by Dr. Alan Rosenstein. The
menon and has not been modelled theoretically.
authors are grateful to Dr. J. Wadsworth of Lock-
However, it is reasonable to assert that, owing to
spheroidization, the experimentally observed heed Research and Development Laboratory,
transition from rounded to elongated cavities will Palo Alto, for providing the material used in this
study.
tend to occur at higher cavity radii than predicted
by the theoretical calculations. A mechanism of
the type shown in Fig. 14 can thus rationalize the References
observation of small rounded cavities even when
1 J. Wadsworth, in S. P. Agrawal (ed.), Superplastic Form-
the cavities have grown by a power-law ing, American Society for Metals, Metals Park, OH,
mechanism. 1985, p. 43.
59

2 J. Wadsworth and A. R. Pelton, Scr. Metall., 18 (1984) Metall. Trans. A, 8(1977) 933.
387. 29 S.-A. Shei and T. G. Langdon, J. Mater. Sci., 13 (1978)
3 J. Wadsworth, I. G. Palmer, D. D. Crooks and R. E. 1084.
Lewis, in E. A. Starke and T. H. Sanders (eds.), Alumi- 30 A.H. Chokshi and T. G. Langdon, to be published.
num-Lithium Alloys H, The Metallurgical Society of 31 E. Hart, Acta Metall., 15(1967) 351.
AIME, Warrendale, PA, 1984, p. 111. 32 J. J. Jonas, R. A. Holt and C. E. Coleman, Acta Memll.,
4 R.J. Lederich and S. M. L. Sastry, in E. A. Starke and T. 24(1976) 911.
H. Sanders (eds.), Aluminum-Lithium Alloys 11, The 33 U. F. Kocks, J. J. Jonas and H. Mecking, Acta Metall., 27
Metallurgical Society of AIME, Warrendale, PA, 1984, (1979)419.
p. 137. 34 S. Sagat and D. M. R. Taplin, Met. Sci. J., 10 (1976) 94.
5 A. K. Ghosh and C. Gandhi, in H. J. McQueen, J.-E 35 IL A. Mohamed and T. G. Langdon, Acre Metall., 29
Bailon. J. I. Dickson, J. J. Jonas and M. G. Akben (eds.), (1981)911.
Strength of Metals" and Alloys (ICSMA 7), Vol. 3, 36 M. M. I. Ahmed and T. G. Langdon, J. Mater. Sci., 18
Pergamon Press, Oxford, 1986, p. 2065. (1983) 2407.
6 J. Wadsworth, C. A. Henshall, A. R. Pelton and B. Ward, 37 D.A. Woodford, Trans. ASM, 62 (1969) 291.
J. Mater. Sci. Lett., 4(1985) 674. 38 T.G. Langdon, Scr. Metall., 11 (1977) 997.
7 J. Wadsworth, A. R. Pelton and R. E. Lewis. Metall. 39 A. K. Ghosh and R. A. Ayres, Memll. Trans. A, 7(1976)
Trans. A, 16(1985) 2319. 1589.
8 J. Wadsworth, 1. G. Palmer and D. D. Crooks, Scr. 40 J. Lian and B. Baudelet, Scr. Metall., 21 (1987) 331.
MetalL, 17(1983) 347. 41 C. H. Caceres and D. S. Wilkinson, Acta Metall., 32
9 R. Grimes and W. S. Miller, in E. A. Stari~e and T. H. 1984) 415.
Sanders (eds.), Aluminum-Lithium Alloys" IL The Metal- 42 M.J. Stowell, in N. E. Paton and C. H. Hamilton (eds.),
lurgical Society of AIME, Warrendale, PA, 1984, p. 153. Superplastic Forming of Structural Alloys', The Metal-
10 R. J. Lederich, P. J. Meschter and S. M. L. Sastry, in R. lurgical Society of AIME, Warrendale, PA, 1982,
Pearce and L. Kelly (eds.), Superplasticity in Aerospace p. 321.
Aluminium, Ashford Press, Southampton, 1985, p. 1(/5. 43 N. Ridley and J. Pilling, in B. Baudelet and M. Suery
11 A. J. Shakesheff and P. G. Partridge, J. Mater. Sci., 21 (eds.), Superplasticity, Centre Nationale de la Recherche
(1986) 1368. Scientifique, Paris, 1985, p. 8.1.
12 Y. Zhang and N. J. Grant, Mater. Sci. Eng., 68 (1984) 44 B. P. Kashyap and A. K. Mukherjee, Res Mech., 17
119. 1986) 293.
13 R. J. Lederich, S. M. L. Sastry and P. J. Meschter, Scr. 45 C. H. Caceres and D. S. Wilkinson, Acta Metall., 32
Metall., 19(1985) 177. ( 1984) 423.
14 M. C. Pandey, J. Wadsworth and A. K. Mukherjee, Scr. 46 C. H. Hamilton, in H. J. McQueen, J.-R Bailon, J. T.
Metall., 19(1985) 1229. Dickson, J. J. Jonas and M. G. Akben (eds.), Strength of
15 J. Pilling and N. Ridley, in C. Baker, P. J. Gregson, S. J. Metals and Alloys (ICSMA 7), Vol. 3, Pergamon Press,
Harris and C. J. Peel (eds.), Aluminium-Lithium Alloys Oxford, 1986, p. 1831.
111, The Institute of Metals, London, 1986, p. 184. 47 D. S. Wilkinson and C. H. Caceres, J. Mater. Sci. Lett., 3
16 J. Pilling and N. Ridley, Acta Metall., 34 (1986) 669. ~.1984) 395.
17 M. C. Pandey, J. Wadsworth and A. K. Mukherjee, Mater. 48 A.H. Chokshi, Metall. Trans. A, 18 (1987) 63.
Sci. Eng., 78(1986) 115. 49 E. P. Butler, N. J. Owen and D. J. Field, Mater. Sci.
18 R J. Meschter and R. J. Lederich, Mater. Sci. Eng.. 89 Technol., 1 (1985) 531.
(1987) 169. 50 N. J. Owen, D. J. Field and E. R Butler, Mater. Sci.
19 M.C. Pandey, J. Wadsworth and A. K. Mukherjee, Mater. lechnol., 2(1986) 1217.
Sci. Eng., 89(1987) 171. 51 C. C. Bampton and J. W. Edington, Metall. Trans. A, 13
2(1 A. H. Chokshi, J. E. Franklin and A. K. Mukherjee, in (1982) 1721.
M. G. Yan and S. H. Zheng (eds.), Mechanical Behaviour 52 D. W. Livesey and N. Ridley, J. Mater. Sci., 17 (1982)
of Materials-- V, Vol. 1, Pergamon, Oxford, 1987, p. 461. 2257.
21 A.H. Chokshi and A. K. Mukherjee, in S. I. Anderson et 53 A. K. Ghosh, in N. Hansen, A. Horsewell, T. Leffers and
al. (eds.), Proc. Conf. on Constitutive Relations and Their H. Lilholt (eds.), Deformation of Polycrystals: Mechanisms
Physical Basis, Riso National Laborato~, Roskilde, 1987, and Microstructures, Riso National Laboratory,
p. 265. Roskilde, 1981, p. 277.
22 Y. Ma and T. G. Langdon, Proc. 9th Inter-American ('onf. 54 A. H. Chokshi and T. G. Langdon, in B. Baudelet and M.
on Materials Technology, Santiago, 1987, p. 253. Suery (eds.), Superplasticity, Centre Nationale de la
23 A. H. Chokshi, J. Wadsworth and A. K. Mukherjee, Scr. Recherche Scientifique, Paris, 1985, p. 2.1.
Memll., 21 (1987) 1347. 55 A.H. Chokshi, J. Mater. Sci., 21 (1986) 2(173.
24 R J. Meschter, R. J. Lederich and S. M. L. Sastry, Metall. 56 A. H. Chokshi and T. G. Langdon. Acre Metall., 35
Trans. A, 18 (1987) 1333. (1987) 1089.
25 C. C. Bampton and J. W. Edington, J. Eng. Mater. 57 M. V. Speight and W. Beere, Met. Sci. J., 9 (1967) 190.
Technol., 105 (1983) 55. 58 J.W. Hancock, Met. Sci. J., 10 (1976) 319.
26 N. Ridley, D. W. Livesey and A. K. Mukherjee, Metall. 59 H. J. Frost and M. F. Ashby, Deformation Mechanism
Trans. A. 15(1984) 1443. Maps, Pergamon Press, Oxford, 1982, p. 21.
27 A. J. Shakesheff and R G. Partridge, J. Mater. Sci., 20 60 A.H. Chokshi, J. Mater. Sci. Lett., 5(1986) 144.
(1985) 2408. 61 D. A. Miller and T. G. Langdon. Metall. Trans. A, 10
28 F. A. Mohamed, M. M. I. Ahmed and T. G. Langdon, (1979) 1869.
60

62 N. Ridley and D. W. Livesey, Res Mech. Lett., 1 (1981) Sci., 19(1984) 1321.
73. 66 E A. McClintock, J. Appl. Phys., 4 (1968) 363.
63 D. W. Livesey and N. Ridley, Metall. Trans. A, 13(1982) 67 J. R. Rice, in S. Wolf (ed.), Time Dependent Failure of
1619. Materials, U.S. Department of Energy, Washington, DC,
64 D.A. Miller and T. G. Langdon, Trans. Jpn. Inst. Met., 21 1979, p. 130.
(1980) 123. 68 A. Needleman and J. Rice, Acta Metall., 28 (1980) 1315.
65 N. Ridley, D. W. Livesey and A. K. Mukherjee, J. Mater.

You might also like