Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Materials Science and Engineering A, 101 (1988) 13-23 13

Creep Characteristics of an AI-2wt.%Cu Alloy in the Solid Solution Range

PRAB1R K. CHAUDHURY and FARGHALLI A. MOHAMED


Department of Mechanical Engineering, Universityof( "alijbrnia, #vine, CA 92717 (U.S.A.)
(ReceivedJuly 24, 1987; in revised form October 15, 1987)

Abstract strengthening effect is expected to arise from segre-


gation of copper atoms to moving dislocations [7].
"lJw creep characteristics of an Al-2wt, %Cu alloy
Second, recent creep experiments on the high
including the stress exponent, the activation energy
temperature deformation of AI-Mg alloys (1-7.7
for creep and the shape of the creep curve were in-
wt.% Mg) [8-13] revealed the presence of two tran-
vestigated in the temperature range of 783-843 K,
sitions in the stress dependence of the steady state
where copper is in solid solution. The experimental
creep rate 9. The first transition [8-101 occurs at
data of the alloy, when plotted as creep rate against
intermediate stresses and is manifested by a change
stress on a logarithmic scale, show the presence of
in the stress exponent n={O(ln 9)/i3(1n r)}~ for
two transitions in the stress dependence of creep rate;
creep (where r is the applied shear stress and T is
at constant temperature, the stress exponent n fi>r
the absolute temperature) from a value close to 4.5
creep changes from a value o]'4,5 at low stresses (low
at low stresses to a value of about 3 at intermediate
stress region) to a value of about 3 at intermediate
stresses. The second transition I I 1-13] is observed
stresses (intermediate stress region) and then
at high stresses and is characterized by a change in
increases again to a value o]'4.5 at high stresses (high
the stress exponent from about 3 to higher values
stress region). The activation energies for creep at
(n > 4). Theoretical analyses of these two transitions
low, intermediate and high stresses are 155 kJ tool ~,
[11, 14, 15] indicate that the atom misfit parameter
151 kJ mol - i and 180 kJ tool ~ ~ respectively. While
e, which represents the size difference between the
creep curves obtained in the entire stress range of the
solute and solvent atoms, plays an important role in
investigation exhibit a normal primary stage, the
predicting the values of the transition stresses and
extent of the stage is less pronounced at intermediate
that dilute solid solution alloys with intermediate
stresses (n =3.2) than at both low and high stresses
values of this parameter (0.08-0.12), when crept
(n =4.5). The above creep characteristics obtained
under suitable experimental conditions, may exhibit
for Al-2wt. %Cu are discussed in the light of recent
the above two transitions. The AI-Cu system has an
suggestions and predictions regarding deformation
atom misfit parameter of about 0.14 [16] which is
mechanisms in solid solution alloys.
comparable with that of the AI-Mg system (about
0.12) and it is therefore expected that creep tran-
1. Introduction sitions in the stress dependence of steady state
creep rate, similar to those reported for dilute
The work reported here on the high temperature
deformation of an AI-2wt.%Cu alloy under solid AI-Mg alloys, would occur in the plot of the creep
solution conditions was motivated by two consider- data of dilute AI-Cu alloys. At present, experi-
mental evidence in support of this expectation is not
ations. First, the deformation behavior of dilute
available.
A1-Cu alloys (less than 5 wt.% Cu) was extensively
studied in the low temperature range [1-3] where
copper precipitates on aging and where the streng-
2. Experimental procedure
thening effect in the alloys is due to the interaction
between precipitates and dislocations (age or pre- A1-2wt.%Cu (Al-0.86at.%Cu) was supplied by
cipitation hardening) [4-6]. By contrast, no investi- the Kaiser Aluminum and Chemical Corporation.
gation has yet been conducted to study the The alloy was prepared from aluminum of 99.99%
deformation behavior of dilute AI-Cu alloys in the purity and copper of 99.99% purity. The final
high temperature range where copper exists in solid material contained the following elements: 1.99
solution (the alloy is a single phase) and where the wt.% Cu, 0.003 wt.% Si, 0.001 wt.% Fe, 0.004 wt.%

0921-5093/88/$3.50 © Elsevier Sequoia/Printed in The Netherlands


14

Mg, 0.003 wt.% Zn, less than 0.001 wt.% Mn, less
than 0.001 wt.% Ti and the remainder is aluminum. 10-'
Double-shear specimens of shape and dimen-
sions described elsewhere [8, 17] were machined
10.2
from the material. Prior to testing, all specimens
were annealed in situ for 15 h at 883 K to remove
the effects of machining and to produce a stable
uniform grain size. Because of the phase transfor- 16 3

mation which occurred as the samples were cooled,


difficulties were encountered in the measurements
of grain size using standard metallographic pro-
cedures that were applied successfully for A1-Mg
alloys [18]. Accordingly, an alternative procedure
that is based on deformation by grain boundary 10 -~ /
sliding and that was adopted recently for A1-Zn
alloys [19, 20], was used. This procedure, when
repeated, resulted in a consistent grain size of 3
mm.
Double-shear experiments were conducted
10 .6
~ I
4.5

either on a suitably designed creep-testing machine


at a constant load or on an Instron machine operat- l o 0.1 1 10
ing at a constant cross-head speed. Because of the T, MPa
shear configuration of the specimens, the creep tests
Fig. 1. Shear strain rate vs. shear stress (logarithmic scale)
are equivalent to conditions of constant stress and for A1-2wt.%Cu at 7 8 3 K (o), 8 1 3 K (Or) and 8 4 3 K ( v )
the Instron tests are equivalent to conditions of con- using the creep machine and at 813 K (*) using the Instron
stant strain rate. The procedures of testing in the machine.
present investigation were the same as those
described in detail elsewhere [8, 17].
On the creep machine, steady state creep rates each temperature reveal the presence of three
were measured at stresses ranging from 0.2 to 6.5 regions of deformation: region I at low stresses,
MPa. This range of stresses was scanned at three region II at intermediate stresses and region III at
different temperatures, 843, 813 and 783 K, which high stresses. The values of the stress exponent n in
fall in the solid solution range; the solid solution regions I, II and III are 4.5, 3.2 and 4.5 respectively.
range for A1-2wt.%Cu is narrow and extends from As seen in the figure, region II covers an increasing
698 to 890 K. The Instron machine was used only range of measurable shear strain rate with decreas-
at T = 813 K to determine the steady state stresses ing temperature; at T= 843, region II covers one
at high strain rates ranging from 6 x 10 -4 to and a half orders of magnitude of strain rate
8 X 10 -2 S-1. This range ofstrain rates was selected whereas, at T= 773 K, it extends over more than
to provide an overlap with the creep data and t o two orders of magnitude of strain rate. Also, Fig. 1
cover the very high stress range. shows that, at T= 813 K, there is excellent agree-
The activation energy for creep was determined ment between the experimental data using creep
from the analysis of creep rate-stress data obtained (applying external stresses and measuring steady
at the three test temperatures and from the results state creep rates) and those using the Instron
of the temperature-cycling procedure [21 ]. method (imposing strain rates and measuring the
steady state stresses) in regions II and III.
Figure 2 provides a comparison between the
3. Experimental results
creep behavior of AI-2wt.%Cu and that of pure alu-
3.1. Stress dependence of the steady state creep rate minum at the same homologous temperature TITm
Experimental data showing the stress depend- of 0.91; this homologous temperature is equiva-
ence of steady state creep rate are presented in Fig. lent to 813 K for the alloy and to 852 K for the
1, where the steady state shear strain rate ~ is plot- metal. The full line representing aluminum in the
ted as a function of the applied shear stress r on a figure has a slope of 4.5 (which equals the stress
logarithmic scale for three different temperatures, exponent) and its position was determined from the
783, 813 and 843 K. The creep data obtained at equation governing the creep of aluminum under
15

10 2

0.4
101
p.

l O0
0.2
~ g i o n I
"1"=783 K
1"=0.35 M Pa

O'l
10' I I I I
40 80 120 160
t,hr
102

103
-

Region II
T=783 K
T = 1.25 M Pa
10'

10 5
0"i~ 1 I
1 I
40 t,migO 120 160
16'
0.8
10' I / /o I I l i I I
0.1 1 10
'T, M P a 0.6
Fig. 2. Comparison between the creep behavior of
AI-2wt.%Cu at 813 K using the creep machine (o) and using
p. Region III
the Instron machine (e) and that of pure aluminum [22] T = 783 K
( ) at 0.91 Tin,where T,, is the melting point. 0.4 T=5.0 MPa

climb conditions [22-24]. Examination of Fig. 2 0.2


reveals three observations. First, over the entire
experimental stress range at 813 K, the creep
strength of A l - 2 w t . % C u is higher than that of alu- O 1 1 I I
minum. Second, the stress dependence of creep rate 0 20 40 60 80
t,s
for the alloy, unlike that for the metal, exhibits two
transitions; for aluminum, the change in the stress Fig. 3. Typical plots of shear strain vs. time (creep curves) of
Al-2wt.%Cu at 783 K for region I (low stress, r = 0.35 MPa),
exponent from 4.5 to higher values (the advent of region II (intermediate stress, r = 1.25 MPa) and region 11I
the region of the breakdown of the creep power (high stress, r = 5.0 MPa).
law) occurs, according to present calculations, at 7
MPa which is nearly equal to the highest stress
applied to A l - 2 w t . % C u at 813 K. Third, the stress
exponent for creep in the alloy in the low stress in all creep curves at all stress levels and that the
region (region I) is equal to that for creep in alumi- extent of the primary stage is more p r o n o u n c e d in
num. regions I and III than in region II.

3.2. Creep curves 3.3. Activation energy for creep


Typical examples of creep curves obtained in T h e results of activation energy measurements
regions l, II and III are presented in Fig. 3, where by the temperature-cycling method [21] for regions
creep strain is plotted as a function of time. Exami- I and II are shown in Fig. 4. In this method, a speci-
nation of these curves, together with others, indi- men was subjected to a n u m b e r of rapid changes A T
cates that a normal primary stage (d ~,/dt < 0) exists in temperature (about 17 K) while under a constant
16

l I I
' 1 ' I ' I
16: ~
b

Region II -- 109

, , -

,o
T= 784 K
i I---~
"r
t'L

,rl-" 10~ I0 a

10~

RegiOn I
1~- -- IO T
O. kJ/mole =150 129 1415 1512 1138

,g_ -~ , , , , -

I I I
" i I i ~ , 1.1 1.2 1.3
•',1"=17 17 16 18 17 K
I O 0 0 / T , K -1

2X10 ? [ l i ] I 1 Fig. 5. Determination of the true activation energy of


0 0.2 0.4 0.6
,y AI-2wt.%Cu for creep in regions I and II by plotting
log(~G" iT) vs. 1/T: line a, region l, r=0.35 MPa,
Fig. 4. Shear strain rate vs. shear strain of Al-2wt.%Cu, Q=155_+4 kJ mol '; line b, region II, r=l.5 MPa,
showing the determination of the apparent activation energy Q= 151 ___5 kJ tool-~. The experimental data were taken
for creep in region I (r=0.35 MPa) and region II (r= 1.5 from Figs. 1 and 4.
MPa) from the temperature-cyclingprocedure.

region III because of the very fast rates associated


stress. The apparent activation energy was then with this region. This value of the activation energy
determined after each temperature change from the in region III (180 kJ mo1-1) is larger than those of
relation activation energies in region I (155 kJ mol-1) and
region II ( 151 kJ mol- 1).
R ln(~2/~,)
Q (1)
(T~ ~)/~
-

4. Discussion

where ~] and 92 are the instantaneous creep rates It is well established that a low temperature
immediately before and after the change in temper- strengthening effect can be produced in the A1-Cu
ature from T~ to T2. The average activation energies system by a precipitation reaction (precipitation
estimated for regions I and II, using the data of Fig. hardening) [1-3]. In this case, a dilute A1-Cu alloy
4, together with those obtained under different is solution treated at a high temperature in the
experimental conditions, were 142 kJ mo1-1 and single-phase region, quenched to room temperature
139 kJ mol- l respectively. so that the alloy contains copper in supersaturated
The results of the temperature-cycling method solution and then aged at some intermediate tem-
together with the creep data in Fig. 1 were plotted perature. Such aging treatment results in the forma-
in Fig. 5 as l o g ( ~ G ' - l T ) ( n = 4 . 5 for region I and tion of precipitates which change in size,
n = 3.2 for region II) vs. 1 / T at constant stress; for morphology and composition with time and which
region I, r = 0.35 MPa and, for region II, r = 1.5 act as barriers to dislocation motion during the
MPa. The activation energies inferred from the plot deformation of the alloy. Detailed studies on the
in Fig. 5 were 1 5 5 + 4 kJ mo1-1 and 151_+5 kJ low temperature deformation of age-hardening
tool- ] for regions I and II respectively. AI-Cu alloys showed that both the strength, as rep-
For region III, a value of 180 kJ mo1-1 was resented by the yield stress, and the rate of work
inferred from the creep data in Fig. 1; it was not hardening of the alloys [2] are, in general, higher
possible to use the temperature-cycling method in than those of pure aluminum.
17

The present results provide evidence that a of solute atoms to moving dislocations [7], or the so-
strengthening effect is also produced in dilute called Cottrell-Jaswon interaction, is responsible
AI-Cu alloys when tested at elevated temperatures for the viscous drag process that interacts sequen-
under the solid solution condition. This strengthen- tially with dislocation climb during the creep of an
ing effect+ according to Fig. 2, is manifested by two alloy. In a very recent analysis [28], the glide-climb
experimental trends: (a) for constant creep stress, criterion was modified to incorporate the contribu-
AI-2wt.%Cu creeps at a much slower rate than tions of other viscous drag processes such as those
pure aluminum does and (b) the creep rate-stress based on deviation from random distribution in
data of the alloy, unlike those of the metal, exhibit solid solution alloys [29] and segregation of solute
three deformation regions with two transitions in atoms to stacking faults [30]. The modified criterion
the stress dependence of the steady state creep rate. is represented by [28]
The origin of such a strengthening effect is
examined below in the light of recent advances (;b (3>
made in rationalizing the high temperature defor-
mation of solid solution alloys. where YA is the sum of the values of the interactkm
The three deformation regions observed in parameters for various concurrently active drag
A1-2wt.%Cu resemble those previously reported processes.
for aluminum alloys [8-13, 20], especially A1-Mg Consideration of the experimental data on
alloys [8-13]. The occurrence of these regions in AI-2wt.%Cu indicates the presence of the following
A1-Mg alloys has been explained in terms of (a) trends which are qualitatively compatible with the
transitions in creep mechanisms that control the prediction of the above deformation criterion for
creep behavior of solid solution alloys [11, 14, 15] creep in solid solution.
or (b) dynamic strain aging [25[. (a) The creep characteristics of AI-2wt.%Cu at
low stresses (region I) including the shape of the
creep curve (extensive primary), and the value of
4.1. Transitions in creep mechanisms
the stress exponent (n=4.5) resemble those
4.1. l. Transition frorn region 1 to region H reported for aluminum [22]. This resemblance
It has been suggested [14, 22, 26, 27] that the implies that the creep behavior of AI-2wt.%Cu at
creep deformation of a solid solution alloy may be low stresses, like that of aluminum, is controlled by
controlled by the sequential process of viscous glide some form of dislocation climb process [22, 26,
and dislocation climb (the slower process controls 31-33].
the creep behavior). On the basis of this suggestion, (b) The stress exponent of 3.2 at intermediate
it was predicted [14, 22, 26] that, under a favorable stresses (region 1I) is in good agreement with the
combination of materials parameters (such as the prediction of a creep model for viscous glide formu-
atom misfit parameter) and experimental variables lated originally by Weertman ~33, 34]; the very small
(such as stress), the creep behavior of an alloy difference in the value of the stress exponent
would change from climb-controlled behavior at between theory ( n = 3 ) and experiment (n=3.2)
low stresses to viscous-glide-controlled behavior at may be the result of small contributions from other
intermediate stresses. A deformation criterion concurrently active creep processes, especially dis-
which accounts for such a change in creep behavior location climb, to viscous glide and/or the presence
was developed and is given by [14] of an internal stress [35] during viscous glide in
solid solution alloys.
(c) The extent of the primary stage at intermedi-
eci/.Gb3 } = B ~ Dg (2) ate stresses In = 3.21 is less pronounced than that at
low stresses (n=4.5). This finding is similar to
where k is Boltzmann's constant, e is the atom misfit earlier observations reported for A1-Mg alloys [8,
parameter, c is the solute concentration, G is the 10] whose creep behavior also exhibited a change in
shear modulus, b is the Burgers vector, B the stress exponent from 4.6 at low stresses to about
( = 4 x 1013) is a dimensionless constant, F is the 3.2 at intermediate stresses and in which the change
stacking fault energy of the alloy, D¢ is the diffusion in the stress exponent was accompanied by a corre-
coefficient for the climb process and Dg is the diffu- sponding change in creep substructure [9, 18] from
sion coefficient for the glide process. The develop- that typical of dislocation climb (significant ten-
ment of the above criterion, as given by eqn. (2), is dency to form well-developed subgrains) at low
based in part on the assumption that the segregation stresses to that typical of viscous glide (random dis-
18

-11
tribution of dislocations) at intermediate stresses. lO
I i I l
Recent work on A1-Mg alloys [36] showed the pres-
ence of subgrains in the viscous glide region, where
n = 3. However, these subgrains are not the result of
deformation because they were also found in the /
unstrained sections of the specimens [36]. -12 /
Although the activation energy for creep at inter- 10 __

mediate stresses is close to that describing the diffu-


// -

sion of copper in aluminum ( 138 kJ mol- 1) [37], it is


difficult, because of the similarity between the Q
value in region I (155 kJ mo1-1) and the Q value in
region II (151 kJ mol-l), to identify deformation ,.D 1() 13
mechanisms in A1-2wt.%Cu unambiguously on the
o
/
basis of activation energy measurements.
~- _
In order to examine quantitatively the correspon-
dence between the glided-climb criterion for solid
solution alloys and the experimental data for
AI-2wt.%Cu, it is necessary to identify the origin of 1614 - - / _
viscous glide creep in A1-2wt.%Cu and to make an //
estimate for the stacking fault energy F of the alloy.
Calculations reported elsewhere [28] show that
//
the major force retarding the glide of dislocations in
dilute A1-Cu alloys arises from the Cottrell-Jaswon
-15
[7] interaction. In this case, it is appropriate to use lO
-6
I
-5
j i 1
eqn. (2), in lieu of eqn. (3), in the analysis of the 5X10 lO 5X10 s
"r/G
present data on A1-2wt.%Cu.
The value of F for A1-2wt.%Cu is not available, Fig. 6. Normalized creep rate vs. normalized shear stress
but it is possible to make an estimate by using the (logarithmic scale) for AI-2wt.%Cu at 843 K (V), 813 K (~r)
and 783 K (<>)in region I (low stresses): . . . . , normalized
normalized creep data of the alloy and by applying creep data of pure aluminum [22].
the finding of earlier analyses on the creep data of
various f.c.c, metals and solid solution alloys [14,
38] which show that, under the condition of climb Fig. 6 is 180 mJ m -2. This value of F is comparable
control, the normalized creep rate 9N( = ~kT/DGb) with those estimated for other dilute aluminum
is related to (F/Gb)3; this cubic dependence of 9N alloys, including Al-lwt.%Mg (190 mJ m -2) [41]
on the normalized stacking fault energy F / G b was and AI-2wt.% Zn (200 mJ m -2) [42].
recently rationalized in terms of a second-power In Fig. 7, eqn. (2) is depicted graphically by plott-
dependence of the climb velocity on F/Gb [39] and ing (kT/ecl/2Gb3) 2 vs. (F/Gb)3(Oc/Og) (r/G) 2 on a
a linear dependence of the mobile dislocation den- logarithmic scale and is represented by a full line at
sity on F/Gb [40]. In Fig. 6, the experimental data 45°; the position of this line was previously deter-
obtained for A1-2wt.%Cu at various stresses and mined using the creep data of A1-3wt.%Mg [8]
different temperatures in region I, where n = 4 . 5 , which exhibited a transition from climb control at
were normalized in terms of the equation of Dorn low stresses to viscous glide control at intermediate
and coworkers [22] by plotting ~N( = 9kT/DGb) stresses. The stress range covering both regions I
against r/G on a logarithmic scale; D = D o and I1 for the alloy together with the experimental
e x p ( - Q / R T ) , D O= 1.86 cm 2 s -~ [16], Q = 155 kJ transition point for each testing temperature was
mol -~ and b =2.86 × 10 -s cm. Also included in the superimposed, as a horizontal line, on Fig. 7 using
figure is a full line representing the normalized data the following values: e = 0 . 1 4 2 [16]; c=0.0086;
of aluminum [22] which were obtained under a F=180 rrd m-2; Dc=1.86 exp(-155/RT);
similar normalized temperature range. As seen in Dg = 0.647 exp( - 151/RT).
the figure, the normalized creep data of The experimental transition points, as shown in
A1-2wt.%Cu fit a single straight line which falls Fig. 7, not only exhibit a trend that is consistent
very close to the line representing aluminum. By with the direction of the boundary between the two
using the relation ( 9N)AI/(9~ )alloy = (rAl/Fanoy )3, the types of creep behavior but also are separated by a
stacking fault energy of A1-2wt.%Cu inferred from factor of only 2 from the position of the boundary.
19

This separation is not unreasonable considering the 4.1.2. Transition f r o m region II to region I l l
assumptions made in the analysis, the approxima- Analyses of the creep data of solid solution alloys
tion introduced in developing equations and the [15, 20], which exhibit a stress exponent of about 3
errors encountered in measuring various para- (alloy type or class I of Sherby and Burke) and
meters such as the shear modulus and diffusion whose creep behavior is therefore attributable to
coefficients. the operation of viscous glide processes, show that
the transition stress at which the stress exponent
100 starts to deviate from the above value (n = 3) agrees
I ' ' ;'1 I
/ reasonably well with the critical stress required for
the breakaway of dislocations from solute atom
atmospheres; this critical stress is given by ll 5j
,.,, 10 _ ~ __ T + Gb 3

+%,: G
- 0.05 ce- - -
kT
(4)

To examine the validity of eqn. (4) in A1-


2wt.%Cu, the experimental normalized transition
stresses of the alloy are plotted against c e 2 G b ~ / k T
on a logarithmic scale in Fig. 8, which was re-
ported earlier [15, 20] and in which the prediction
of eqn. (4) is represented by a full line having a
O.1 I ' ,J,I I slope of 1. As seen in the figure, the data of
1015 -14 -13 -12
lO lO lO A1-2wt.%Cu, like those of other alloys [13, 20],
scatter about the line describing eqn. (4) and agree
with the prediction of the equation; the agreement is
Fig. 7. The glide-climb criterion for Al-2wt.% Cu repre- within a factor of 2. In addition to this good agree-
sented by plotting (kT/ecl/2Gb3) 2 vs. (F/Gb) 3 (Dc/Dg) (r/G) 2 ment between experiment and eqn. (4), the observa-
on a logarithmic scale. The experimental datum points repre-
senting the transition from region I to region II are shown for tion, noted in Fig. 1, that the transition stress
AI-2wt.%Cu at three different temperatures. between regions I1 and III increases with decreasing

10 -1

AI-5.6MglTI I T I ~ M
AI3"3MgI~] rTI 7 AI'4"2M
AI-2.2 Mg....['r~ ~"~1-2.7 Mg
AI-1.8~ +
10-2 ~ A1-0.86C~ J ~AI*IIMg _-

-- In-3Cd r ~ ] j -
/
--0 In-2Cd ~ fl~ (~ (~) (I) _

1°-3 ,o-7 so ® A,-lOZ°

In-2.9 Sn

1o- t 1 l I zt
-5 -4 -3
10 10 10
"i'/13
Fig. 8. Correlation between the condition of the breakaway of dislocations from the solute atom atmosphere (eqn. (4)) and ex-
perimental data of solid solution alloys including Al-2wt.%Cu (Al-0.86at.%Cu). The compositions of the alloys in this figure are
given in atomic per cent.
20

temperature is compatible with the prediction of (b) The frictional stress oi~ has the form of a
eqn. (4): r o c G2/T. statistical distribution which is given by
While the transition in the stress dependence of
creep rate in A1-2wt.% Cu at high stresses agrees oD = oD° exp{ ( TBT)2 } (5)
with the concept of the breakaway of dislocations
from solute atom atmospheres [15], the creep
characteristics of the alloy in region IlI are not where aDo is the maximum value of o D and B rep-
entirely consistent with any of the deformation pro- resents the width of the distribution about T.
cesses suggested previously [11, 43-45] to explain (c) The frictional stress a D due to dynamic strain
the high stress creep behavior of metals and solid aging also depends on strain rate; in eqn. (5), ;c is
solution alloys. For example, the value of the stress predicted to be a function of strain rate which is
exponent (n=4.5) and the type of creep curve expressed by the following Arrhenius-type equa-
(extensive primary) in region III appear to be com- tion:
patible with some form of dislocation climb process
controlled by lattice diffusion [11, 43], but the high
value of the activation energy for creep in this
region (higher than that expected for diffusion of
aluminum in A1-2wt.%Cu) precludes the operation where g is the tensile strain rate ( = .~9), A is a con-
of such a process. Furthermore, this high value of stant and (2 is the activation energy for diffusion of
the activation energy rules out the possibility of the solute atoms in the alloy. According to eqns. (5) and
operation of any deformation process that is con- (6), % is negligibly small at very low and high strain
trolled by pipe diffusion [45], whether climb or vis- rates but exhibits a maximum value at some inter-
cous glide. In this context, it is worth mentioning mediate strain rate.
that the presence of a high activation energy for (d) The creep behavior of A1-Mg alloys over the
creep at high stresses (region III) is not confined to entire range of stresses can be described by a single
A1-2wt.%Cu. Recent experimental evidence on the rate equation of the form
creep behaviour of A1-21wt.%Zn [20] shows that
the activation energy for the alloy ( 165 kJ mol- l ) at
g = K(crapp - ORB)"*exp ( - RQ----~) (7)
high stresses, which are comparable with those
applied to A1-2wt.%Cu in region III, is higher than
that estimated for diffusion of aluminum in where K is a constant (for constant temperature),
A1-21wt.%Zn (115 kJ mol-J). This suggests that (Yapp ( = 2 Tapp) is the applied tensile stress, n* ( = 5) is
additional experimental work on other solid solu- the stress exponent at low and high stresses and Q*
tion alloys is needed to establish whether the high is the activation energy in the absence of dynamic
activation energy observed in both A1-2wt.%Cu strain aging. Equation (7) predicts on the basis of
and A1-21wt.%Zn at high stresses is a general the characteristics assumed for oi) that, at both low
feature of the high stress creep behavior of solid and high stresses, n approaches 4.76 (for A1-Mg
solution alloys of the alloy class (n = 3). alloys) while at intermediate stresses it is close to a
value of 3.
In order to examine the correspondence between
4.2. Dynamic strain aging eqn. (7) and the creep behavior for A1-2wt.%Cu,
Very recently, Hong [25J explained the transi- the data of the alloy obtained in regions I and II
tions in the stress dependence of creep rate in were used to determine, K, OD°, B and A in eqns.
A1-Mg alloys at 600 K in terms of dynamic strain (5)-(7). The present calculations give the following
aging; the stress exponent for creep in those alloys, values for these parameters: K = 1.35 x 1022TG3.5;
like that for creep in AI-2wt.%Cu, exhibits a mini- aDO= 13.2 MPa; B = 1.2x 104 K2; A = 108 s -~. In
mum value of about 3 at intermediate stresses. The addition, the values for n*, Q* and Q were taken as
explanation of Hong [25] is based on the following 4.5, 155 kJ mol -~ (present investigation)and 138
assumptions. kJ tool- 1 [37] respectively.
(a) Dynamic strain aging occurs during the creep In Fig. 9, the prediction of eqn. (7) is plotted as a
of A1-Mg alloys, and this leads to a strengthening full curve and the experimental data of
effect which is reflected in the presence of a fric- AI-2wt.%Cu at 813 K were included for the pur-
tional stress oD that opposes the movement of dis- pose of comparison. The figure shows the presence
locations. of excellent agreement between prediction and the
21

A1-2wt.%Cu for two reasons. First, the values of


10 2
Ill /
strain rate ( 1 0 - : - 1 0 : s ~) involved at high tem-
peratures such as 813 K are too high Io be accu-
10 ~ _
rately, and reliably, measured with the existing
K 573
creep equipment. Second, while calculations on the
lo ~ -
basis of eqn. (7), as presented in Fig. 9, show that
the strain rate domain of inlerest in region Ill
10 ~ _ (10 7_ 10-i s 1) becomes feasibly measurable in
AI-2wt,%Cu at 573 K, a creep investigation at this
10 2 _
temperature is impractical because the alloy is no
longer a single-phase solid solution.

10 3
///i The preceding discussion suggests that, in order
to examine the validity of the model proposed by
Hong [25] quantitatively at very high stresses, it is
10 4 __
// / necessary to test, under creep conditions, an alloy in
which solute atoms exist in solid solution over a
/Y / wide range of temperatures and whose creep data
10 5 __
J / show evidence of a transition from n = 3 to n = 5 at
/ / high stresses. Consideration of the information
10 6 __ • / available on the phase diagrams and the creep data
of a number of alloys indicates that Al-2wt.%Mg is
1() z
I I/ i II I an appropriate choice; the alloy meets the above
0.1 1 10 100
T, MPa two conditions.
Tests were conducted on A1-2wt.%Mg at 573 K.
Fig. 9. Comparison between the experimental data of
AI-2wt.%Cu obtained at 813 K using the creep machine (o) In conducting these tests, both the creep machine
and using the Instron machine (el and the prediction of the and the instron machine were used to cover a range
model of dynamic strain aging [25] ( ), together with the of strain rates extending from 6 x 10 ~ to 1()~) s-
prediction of the model [25] for 573 K ( - - ) : - - - , the
extrapolations of the low stress region which coincides, in the The experimental data obtained are plotted as
limit, with the high stress region. shear strain rate against shear stress on a logarith-
mic scale in Fig. 10. Also included in the figure is
the prediction of the model of dynamic strain aging,
using the same equation proposed by Hong [25] for
experimental data for the alloy in regions I and II; AI-Mg alloys. The figure reveals three features. First,
however, this agreement is not surprising since the there is excellent agreement between the data of
experimental results on the alloy in these two creep (applying external stresses and measuring
regions were used to determine the values of steady state creep rates) and those obtained by
various adjustable parameters such as K, A and B. lnstron testing (imposing strain rates and measuring
By contrast, there is poor agreement between the the steady state stresses) in the region of interest:
prediction of eqn. (7) and the data of AI-2wt.%Cu the high stress region. Second, good correspon-
at high stresses (high strain rates) where region IIi dence exists, as expected, between the prediction of
exists (n = 4.5); for example, the deviation in strain the model of dynamic strain aging and the experi-
rate between experiment and prediction is a factor mental data of AI-2wt.%Mg at both low and inter-
of about 4 at a shear stress of 6.5 MPa. mediate stresses. Third, at high stresses, the
A distinctive feature of the prediction of eqn. (7), prediction of the model breaks down and is not
as shown in Fig. 9, is that at very high stresses consistent with the trend of the experimental results
( r > 15 MPa) the curve representing the depen- in two ways: (a) a significant discrepancy in position
dence of creep rate on stress, when plotted on a exists between experimental datum points and the
logarithmic scale, approaches and, in the limit, coin- predicted curve (at r = 3(1 MPa, for example, the
cides with the extrapolation of the line representing discrepancy between experiment and the model is
region 1 (the low stress region); this feature requires more than one order of magnitude of strain rate)
that, at high stresses (high strain rates), i3:(ln g)/ and (b) the experimental data, unlike the prediction
i~(ln 0) 2 changes its sign (point of inflection). This of eqn. (7), exhibit a continuous increase in the
feature, which reflects the characteristics assumed stress exponent for creep with no evidence of a
for oD, could not be tested experimentally in point of inflection.
22

lo 0 5. Conclusions
(a) The creep behavior of A1-2wt.%Cu, when
161 studied in the solid solution range, exhibits two
transitions in the stress dependence of creep rate.
The first transition occurs at intermediate stresses
162 and is manifested by a change in the stress exponent
from a value of 4.5 to a value of 3.2. The second
e transition is observed at high stresses and is charac-
16 3 O

terized by a change in the stress exponent from a


value of 3.2 to a value of 4.5.
10" m
(b) The change in the stress exponent from 4.5
r,,
at low stresses to 3.2 at intermediate stresses
165 appears to be consistent with the prediction of the
glide-climb criterion. This consistency is demon-
strated by three findings: (i) the creep characteristics
16' of A1-2wt.%Cu at low stresses resemble those
reported for aluminum and A1-Mg alloys under the
condition of climb control; (ii) the stress exponent
16' of about 3, together with the extent of the primary
stage in the creep curve, at intermediate stresses is
in agreement with the dominance of viscous glide
1 10 100 creep; (iii) the experimental normalized transition
"/', M ~
stresses correspond well with those predicted from
Fig. 10. Comparison between the creep data of the equation representing the glide-climb criterion
AI-2wt.%Mg obtained at 573 K using the creep machine (e)
and using the Instron machine (o) and the prediction of the for solid solution alloys.
model of dynamic strain aging [25]: the strain rate range (c) The creep data of A1-2wt.%Cu at low
where the discrepancy between the model and experiment stresses, where n = 4.5, when normalized and com-
exists.
pared with those of pure aluminum, suggest that the
stacking fault energy of the alloy is 180 mJ m-2; in
estimating this value of F, the well-established cubic
The above inconsistencies between the experi- dependence of the normalized creep rate on the
mental results on AI-2wt.%Mg at high stresses and normalized stacking fault energy F / G b was used.
the prediction of the model of dynamic strain aging (d) While the values of the transition stresses at
[25] suggest that the model, as represented by eqn. which the stress exponent changes from 3.2 at inter-
(7), cannot explain the creep behavior of solid solu- mediate stresses to 4.5 at high stresses agree
tion alloys over the entire stress range; for the reasonably well with those predicted from the equa-
model to be successful, it must account for the tion describing dislocation breakaway from a solute
creep behavior not only at low stresses, where the atom atmosphere, the origin of the high stress
adjustable parameters incorporated in the model region, because of its high activation energy, is not
can produce good correlations, but also at high known; the value of this energy (180 kJ mol-1), for
stresses. In addition to the problem of poor corre- example, rules out the possibility of the operation of
spondence between the model and experimental some form of dislocation climb controlled by lattice
data at high stresses, there are other considerations diffusion.
[46, 47] which tend to raise questions about the (e) More experimental work on solid solution
validity of a creep model for solid solution alloys alloys of class I (the alloy class in which n = 3) is
that is based on dynamic strain aging. For example, needed to examine whether the high activation
solute locking and unlocking of dislocations, a pro- energy measured in A1-2wt.%Cu at high stresses
cess which is responsible for strain aging and which represents general behavior.
may result in the presence of a frictional stress, is (f) The prediction of the model of dynamic
expected to play a very small role during the defor- strain aging, which was proposed to explain the
mation of alloys at high temperatures [46], creep behavior of A1-Mg alloys at 600 K, breaks
especially above 0.7Tin, because, in this range of down at high stresses. This is demonstrated by two
temperatures, solute atoms will be able to diffuse as observations: (i) there is poor agreement between
fast as the moving dislocations. the creep rates of A1-2wt.%Cu and AI-2wt.%Mg
23

and those calculated from the model and (ii) the 19 B.S. Chin, W. D. Nix and G. M. Pound, Mefall. Trans. A.
stress dependence of creep rate in A1-2wt.%Mg, 811977) 1523.
20 M. S. Soliman and F. A. Mohamed, Metall. Trans. A, 13
unlike that predicted by the model, does not exhibit
11984) 1893.
a point of inflection. 21 H. I. Huang, O. D. Sherby and J. E. Dorn, lnms. Am.
Soe. Met., 62 (1956) 155.
22 J. E. Bird, A. K. Mukherjee and J. F.. Dorn, in D. G. Bran-
don and A. Rosen (eds.), Quantitative Rehltion Between
Acknowledgments Properties and Microstructures, Israel University Press,
Jerusalem, 1969, p. 255.
This work was supported by the National
23 J. Weertman, J. Mech. Phys. Solids., 2~5 ( 1955 ) 1213.
Science Foundation under Grant DMR 8420615. 24 I. S. Servi and N. J. Grant, Trans. AIME, 1919 (1951190.
Thanks are extended to Janice Johnson for 25 S.I. Hong, Mater. Sci. Eng., 6211986) 175.
typing the manuscript. 26 W. R. Cannon and O. D. Sherby, Metall. I?ans. A, 1
(1970) 1031).
27 J. Weertman, Trans. Am. Inst. Min. Metall. Pet. Eng,, 218
(1960) 21/7.
References 28 F.A. Mohamed, Mater. Sei. Eng., 61 (I 983) 149.
29 J. C. Fisher, Acre Metall., 2 11954) 9.
1 K. Carlsen and R. W. K. Honeycombe, J. Inst. Met., &~ 30 H. Suzuki, Sei. Rep. Res. Inst., Tohoku Utdv., Ser. A, 4
( 1954-1955) 949. 11957) 455; 711955)194.
2 G. Greetham and R. W. K. Honeycombe, J. Inst. Met., 83 31 O. D. Sherby and J. Weertman. Aeta Metall., 27 (1979)
(1960-1961) 83. 779.
3 J. G. Byrne, M. F.. Fine and A. Kelly, Philos. Mag., 6 32 O. D. Sherby and P. M. Burke, l'rog. Mater. Sci., 13
11961)1119. (1968) 1325.
4 N. E Mott and F. R. N. Nabarro, Rep. Conf. on the 33 J. Wcertman, J. Appl. Phys., 26 (1955) 1213.
Strength of Solids, Physical Society, London, 1948, p. 1. 34 J. Weertman, J. Appl. Phys., 28 (195711185.
5 F.. Orowan, Proe. 5),rap. on Internal Stresses in Metals, 35 S. H. Hong and J. Weertman. Acta Metall., 34 (1986)
Institute of Metals, London, 1948, p. 451. 743.
6 J. C. Fisher, F.. W. Hart and R. R. Pry, Acta Metall., 1 36 X. Y. Meng, D. O. Northwood and I. O. Smith, Metallog-
1l 953) 336. raphy, 19 (1986) 285.
7 A. H. Conrell and M. A. Jaswon, Proc. R. Soc. London, 37 N. k. Peterson and S. J. Rothman, Phys. Rev.. 81 (19711:
Ser. A, 19911949) 104. 3264.
8 K. L. Murty, F. A, Mohamed and J. F.. Dorn, Aeta Metall., 38 F. A. Mohamed and T. G. Langdon, .I. Appl. l'h):s'., 45
2011972) 11109. 11974) 1965.
9 F.A. Mohamed, Metall. Trans. A, 911978) 1013. 39 A. S. Argon and W. C. Moffatt, Acta Metall., 29 (1981)
10 R Yavari, F. A. Mohamed and T. G. Langdon, Acta 293.
Melall., 29 ( 1981 ) 1495. 4(1 A. S. Argon and S. Takeuchi, Aeta Metall., 29 (1981)
11 P. Yavari and T. G. Langdon, Aeta Metall., 32 11982) 1877.
2181. 41 V. C. Kannan and G. Thomas, J. Appl. Phys., 37 (I 966)
12 H. Oikawa, K. Honda and S. Ito, Mater. Sci. Eng., 64 2363.
11984) 237. 42 A. Goel, T. J. Ginter and F. A. Mohamed, Metall. Trans.
13 D. O. Northwood, L. Moerner and I. O. Smith, Phys. A, 14 11983) 231/9.
Status' Solidi/t, 54 (1984) 509. 43 K. L. Murty, Scr. Metall., 711973) 899.
14 F. A. Mohamed and T. G. Langdon, Acta Metall., 22 44 H. Oikawa. K. Sugawara and S. Karashima. Trans..Ipn.
(1974) 779. Inst. Met., 19 (1978) 611.
15 F.A. Mohamed, Mater. Sci. Eng., 38 (1979) 73. 45 S. L. Robinson and O. D. Sherby, Acta e14etall., 1711969)
16 H.W. King, J. Mater. Sci., 1 11966) 79. 11)9.
17 F.A. Mohamed, K. L. Murty and J. W. Morris, Jr., Metall. 46 R. W. K. Honeycombe, The Phlstic Defi)rmation Of
7?ans. A, 4 (1973) 935. Metals', Edward Arnold, London, 1984, p. 356.
18 M. S. Soliman and F. A. Mohamed, Mater. Sci. Eng., 55 47 J. Friedel, Dislocations, Pergamon. Oxford, 1964,
11982) 111. Chapter 16.

You might also like