Materials Science and Engineering A 491 (2008) 88102

Plastic deformation of Al and AA5754 between 4.2 K and 295 K

Dong-Yeob Park, Marek Niewczas

Materials Science and Engineering, McMaster University, JHE-357, 1280 Main St. West, Hamilton, ON L8S 4L7, Canada
Received 2 April 2007; accepted 22 January 2008
Abstract
Plastic deformation and work-hardening behaviour of high-purity Al and AA5754 have been studied by a combination of measurements of
mechanical response, electrical resistivity and TEM. The results show that materials deformed at very low temperatures exhibit an unprecedented
level of strength and unusual work-hardening behaviour. Strain rate sensitivity measurements suggest that deformation of high-purity Al is governed
by dislocationdislocation interactions in a broad range of temperatures, whereas Al alloys exhibit a larger thermal component of ow stress due
to the presence of solute atoms in the matrix. The electrical resistivity data provide information about the evolution of both the dislocation density
and the dislocation mean free path and establish the contributions of dislocation storage and dynamic recovery during plastic ow of materials at
4.2 K. The results suggest that fracture is initiated by the collapse of the dislocation network at places where dislocations develop a critical spacing
for spontaneous annihilation. This spacing is estimated as 8 nm at 4.2 K and 12 nm at 78 K.
2008 Elsevier B.V. All rights reserved.
Keywords: Aluminum; AA5754 Al alloys; Plastic deformation; Electrical resistivity; Work hardening; Strain rate sensitivity; Dislocation density; Fracture
1. Introduction
Aluminum alloys of the 5000 (AlMg) series have been con-
sidered for structural applications in the automotive industry
[1]. The non-heat-treatable AA5754 (Al3 wt.% Mg) have been
receiving considerable scientic and technological attention
because they showlittle or no progressive damage accumulation
prior to fracture [2].
Previous studies have shown that the formability and ductil-
ity of these materials are signicantly inuenced by impurities,
especially iron-rich intermetallic particles whereas the shear
localization is the major mechanism responsible for failure
[24]. Deformed 5xxx AlMg alloys are characterized by a
relatively uniform distribution of dislocations because of the
solute drag effect which prevents dislocations from undergo-
ing further rearrangement to lower energy structures [5,6].
The solute drag also increases the work-hardening rate in Al
alloys. In case of AlMg systems it has been shown that
Mg atoms exert a larger effect on work hardening than on
the solution strengthening [7]. Decreasing the temperature
has a similar effect; it has been observed that the work-

Corresponding author. Tel.: +1 905 525 9140; fax: +1 905 521 2773.
E-mail address: niewczas@mcmaster.ca (M. Niewczas).
hardening behaviour of low solute alloys is more sensitive to
the temperature changes than the work-hardening behaviour of
highly alloyed systems. This reects the competition between
the dynamic recovery and the solute drag processes taking
place during plastic ow of these materials [7]. In general,
the ductility of AlMg alloys decreases with increasing the
Mg content [8]. The torsion experiments reveal that Al1%
Mg alloy exhibits a well-dened stage IV of work harden-
ing in a broad temperature range between 78 K and 373 K
[9].
In contrary to Al alloys, deformed Al develops a three-
dimensional dislocation cell structure, the scale of which
depends upon the amount of deformation and the temperature.
Barker et al. [10] reported that the size of the cell structure
is reduced from 2 m to 1.3 m when Al is strained from
10% to 30% at room temperature. Further it was shown that
with increasing strain, the homogeneous rotation of the individ-
ual cells within cell blocks becomes gradually more difcult,
and thus the size of cell blocks decreases. Chu and Morris
[11] studied the inuence of the microstructure in pure Al
on the work-hardening behaviour at 77 K and concluded that
pre-existing subgrains may hinder the formation of disloca-
tion cells. As a consequence, a high rate of work hardening is
retained at high stresses leading to the improved combination of
strength and elongation. Generally, it is known that the ductility
0921-5093/$ see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2008.01.065
D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102 89
Table 1
Composition of the AA5754 Al alloy (in wt.%)
Mg Mn Si Fe Cr Cu Ni V Zn Al
SC 3.21 0.24 0.05 0.09 Balance
DC 3.11 0.25 0.06 0.21 0.04 0.01 0.01 0.01 0.02 Balance
of Al increases as temperature decreases due to the enhanced
work-hardening capacity of this metal at a lower temperature
[12].
Most studies of the plastic deformation of Al-based alloy sys-
tems available in the literature have been carried out at room or
elevated temperatures. However, there is an interest in under-
standing work-hardening behaviours of commercial alloys at
lower temperatures not only because of the fundamental nature
of such studies but also from the practical side as many of these
systems are restricted from being used at elevated temperatures
[1315].
The objective of this paper is to study mechanical proper-
ties and deformation behaviour of commercial strip cast (SC),
direct chill cast (DC) AA5754 Al alloys and of pure Al between
4.2 K and 295 K, a region of temperatures rarely explored in
the literature. Combined measurements of mechanical response,
changes in electrical resistivity during plastic deformation and
TEM studies of the dislocation substructure were employed to
gain a better understanding of the processes which control plas-
tic deformation and processes which lead to failure in these
materials.
2. Experimental procedure
The experiments were carried out on SC and DC AA5754
Al alloys with a composition given in Table 1 and on high-
purity (99.9995%) aluminum. The as-received material was in
the form of cold-rolled sheets, 1 mm thick SC alloy and 7 mm
thick pure Al. Al sheets were further cold-rolled in a laboratory
to a thickness of 3.7 mm. The as-received DC alloy was in a
Fig. 1. True stress vs. true strain characteristics at the initial strain rates of (a) 1.6 10
4
and (b) 1.6 10
5
for SC alloy (SC) and pure Al (Al) and (c) 1.6 10
4
for DC alloy (DC) at temperatures 4.2 K, 78 K and 295 K.
90 D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102
1 mm thick sheet form in an annealed state with the grain size
of 23 m.
Tensile test samples with gauge dimensions of
1 mm3.5 mm60 mm were machined from the as-received
alloys parallel to the rolling direction. Samples made of Al
had dimensions of 2 mm3.5 mm60 mm. Machined Al and
SC specimens were annealed at 350

C for 30 min and 2 h,

respectively, followed by air cooling, to produce recrystallized
materials. The average grain size of annealed samples was
600 m and 14 m for pure Al and SC alloy. Some SC samples
were annealed at 450

C for 2 h to study the effect of grain size

and annealing temperature on mechanical properties; the nal
grain size of these samples was 18 m.
The specimens were deformed in a screw-driven tensile-
testing machine at temperatures of 4.2 K, 78 K, and 295 K at
a constant cross-head velocity corresponding to an initial strain
rate of 1.6 10
4
s
1
and 1.6 10
5
s
1
. The strain rate sen-
sitivity (SRS) measurements were carried at 4.2 K, 78 K and
295 K for pure Al and at 78 K for the alloys by repeating the
instantaneous rate change between 10
4
s
1
and 10
5
s
1
dur-
ing deformation.
To investigate the evolution of deformation-induced defects,
the electrical resistivity measurements were carried out in situ
during deformation of the samples at 4.2 K. Employed was the
four-point method for the measurement of potential drop across
the sample with reversing the current owing through the circuit.
The resistivity ratio, RR=r
295 K
/r
4.2 K
, i.e. ratio of the resistance
at 295 K over the resistance at 4.2 K of pure Al was about 2000,
indicating a very low residual content of impurities and lattice
defects in the material. In contrast, the alloy had a resistivity
ratio of about 2. Measurements of resistivity change due to
deformation-induced defects in such material presents a chal-
lenge; the apparatus must be able to measure potential changes
of the sample smaller than 10
9
V. Other experimental details
are available elsewhere [16].
TEM studies of the dislocation substructure were performed
on selected samples with a Philips CM12 STEMelectron micro-
scope operating at 120 kV. Thin foils were cut in the section
parallel to the wider face of the samples and were prepared by
a standard thinning procedure. The nal polishing was carried
out in a solution of 30% nitric acid in high-purity methanol
(HPLC) at 35

Cwith Tenupol-5 electropolisher. Fracture sur-

face observations in samples deformed to failure were carried
out using a scanning electron microscope (SEM) and an optical
microscope.
3. Results
3.1. Deformation behaviour and fracture
Fig. 1 shows true stresstrue strain characteristics of SC and
DC alloys and pure Al deformed at 4.2 K, 78 K, and 295 K. Ten-
sile curves of the alloys exhibit extensive serrations at 295 K,
associated with the strain aging known as PortevinLeChatelier
(PLC) effect [17,18]. The stress drop of SC alloy, , during
PLC instabilities increases as the ow stress increases, from
approximately 6 MPa at the early stage of deformation to about
Fig. 2. The effect of heat-treatment on mechanical properties of SC speci-
mens heat-treated for 2 h at 350

C and 450

C, respectively. Tensile tests were

conducted at 295 K, with a strain rate of 1.6 10
4
s
1
.
8 MPa at the later stages of deformation. DC alloy show stress
drops in the range 27 MPa (Fig. 1(c)). The PLC effect is sup-
pressed at 78 K and samples exhibit stable plastic ows with
a long onset of uniform deformation followed by the fracture,
as shown in Fig. 1. At 4.2 K, the material shows ow instabili-
ties arising from the adiabatic shear deformation, of a different
nature than the ones observed at room temperature. The adia-
batic deformationis accompaniedbyintensive acoustic emission
occurring with a higher frequency and louder acoustic clicks
than those observed during the PLC deformation. The sample
deformed with a lower strain rate exhibits a lower frequency of
adiabatic instabilities, as shown in Fig. 1(b). Mechanical charac-
teristics reveal that pure Al shows substantially lower owstress
than SC alloys in the entire deformation range. The Al samples
exhibit a stable plastic ow up to the point of fracture, which is
initiated at the stress of approximately 50 MPa, 150 MPa, and
410 MPa at deformation temperatures of 295 K, 78 K and 4.2 K,
respectively. The fracture of the alloys occurs at much higher
stresses, approximately 290 MPa, 500 MPa, and 700 MPa at cor-
responding temperatures of 295 K, 78 K and 4.2 K, respectively
(Fig. 1(a) and (c)). The alloy samples deformed at 78 K show
the highest uniform deformation before the fracture, which is
comparable to the one observed in pure Al. Compared to SC
alloys, DC alloys show very short or no L uders deformation
after yielding, as seen in Fig. 1(c).
Fig. 2 shows the effect of annealing temperature on mechani-
cal properties of SCmaterials. It is seen that specimens annealed
at 350

Cand450

Cexhibit qualitativelysimilar character of the

ow curve. However, they show some differences in the yield
stress, the uniform deformation, and the stress at which frac-
ture occurs. For example, the sample annealed at 350

C has the
yield stress of about 130 MPa compared to 115 MPa for speci-
mens annealed at 450

C. Similarly, the fracture of the sample

annealed at 450

C occurs at the stress level of about 275 MPa

compared to 290 MPa for the sample annealed at 350

C.
Fig. 3 shows fracture surface observations of SC samples
deformed at three temperatures. The micrographs reveal an elon-
D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102 91
Fig. 3. SEM observations of the fracture surface of SC alloys deformed at (a) 295 K, (b) 78 K, and (c) 4.2 K.
gated dimple structure characteristic of ductile failure. Little or
no particles were found on the fracture surface and within the
dimples indicating that the failure is not a particle stimulated
process in SC alloys. One of the characteristic features of the
fracture surface in this material is the temperature dependence of
the dimple size; one observes the formation of smaller dimples
in samples deformed at lower temperatures. Pure Al exhibits
the same trend, as shown in Fig. 4. It is seen that, at room tem-
perature, a few very large dimples dominate its fracture surface
whereas, at 4.2 K, small dimples are evenly distributed across
the fracture surface. Fracture surface observations (Figs. 3 and 4)
indicate that the dimples formed in pure Al are much larger than
these observed in Al alloys at the same deformation temperature,
suggesting the higher ductility of Al, in agreement with tensile
characteristics of these materials.
Fig. 5shows the normalizedwork-hardeningrate, /, plotted
as a function of the normalized effective owstress, (
y
)/,
for both pure Al and Al alloys; the yield stress (
y
) is chosen
at the onset of the L uders deformation for the alloys. The initial
work-hardening rate of the alloy, equal approximately to /10,
is a factor of almost two higher than the work-hardening rate
of pure Al, of the order of /20. This is being observed for all
temperatures and within the broad range of the ow stresses,
indicating that more effective dislocation storage is taking place
during alloy deformation compared to pure Al. For pure Al, the
work hardening decreases very fast at 295 Kand 78 Kindicating
that the intensive dynamic recovery is taking place during plas-
tic ow. At 4.2 K, dynamic recovery is suppressed; the rate of
work hardening decrease is much slower than at higher tempera-
tures, indicating that more effective dislocation storage occurs in
the substructure. Fig. 5 shows that the alloy exhibits the qualita-
tivelysimilar behaviour of the workhardeningwithrespect tothe
deformation temperature, i.e. the work-hardening rate decreases
faster as temperature increases, suggesting that dynamic recov-
ery processes operate more effectively at higher temperatures
also in SCmaterial. At 4.2 Kthe work-hardening rate of the alloy
remains at an unusually high level above /20 for an extended
range of plastic ow. It is seen however that the hardening rate
decreases faster in SC alloy than in pure Al, suggesting that
more intensive dynamic recovery operates in SC material. It has
92 D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102
Fig. 4. SEM observations of the fracture surface of pure Al deformed at (a)
295 K and (b) 4.2 K.
Fig. 5. Normalized by temperature dependent shear modulus work-hardening
characteristics as a function of reduced ow stress. SC and Al denote SC
alloy and pure Al, respectively. NA and AQ denote the natural aged and the
as-quenched AlZnMg alloys, respectively, from the literature [19] for com-
parison.
Fig. 6. Haasen plot characteristics obtained from strain rate sensitivity tests
with rate changes between 10
4
s
1
and 10
5
s
1
. SCand SC4 denote SCalloys
annealed for 2 h at 350

Cand 450

C, respectively; DCdenotes direct cast alloy

and Al denotes pure Al. For comparison with the present data, strain rate sensi-
tivity results for AA6111 are reported fromRef. [21]. A6(1 h) represent AA6111
material aged for 1 h at 180

C, and A6(SS) represents the supersaturated solid

solution AA6111.
been reported in the literature that AlMgZn alloys deformed at
4.2 K exhibit an abnormally high hardening rate [19]. For com-
parison with present results, Fig. 5 also shows work-hardening
behaviours of naturally aged and solution-treated and quenched
AlMgZn alloys fromRef. [19]. It is seen that SCalloy exhibits
an even higher work-hardening rate than the AlMgZn system
both in aged and quenched states within a broad range of ow
stress.
3.2. Strain rate sensitivity measurements
The experimental data of strain rate sensitivity measurements
are analyzed by plotting (/T ln )|
T,
against (
y
), as
suggested by Haasen [20]. Here is the change in stress due
to the instantaneous change of strain rate, and ,
y
, , T, and
represent the applied stress, yield stress, strain rate, temperature,
and structure, respectively. It is well known that two important
parameters determined from the above analysis are: (i) the ther-
modynamic SRS of a material obtained as a slope of a given
characteristic and (ii) a thermal component of the ow stress
given by the intercept of a plot with the vertical axis. Fig. 6
shows Haasen plots for both pure Al and SC alloys obtained in
this work; for comparison, results for supersaturated solid solu-
tion (SSS) and precipitation-hardened AA6111 alloys available
in the literature [21] are also included in the graph. It is seen that
SCalloys exhibit strain rate sensitivity similar to other classes of
AA6111 materials with the value of about 1.4 10
4
K
1
. The
alloy annealed at 350

C has a slightly higher positive intercept

than the one annealed at 450

C. The intercept of SC alloy is

close to that of SSS AA6111, whereas aged AA6111 exhibits a
higher intercept. The graph also presents data for DC AA5754
alloy tested in this work, with the same nominal composition
as SC alloys but higher contents of iron, i.e. 0.2 wt.% Fe in DC
material versus 0.09 wt.%Fe in SCalloy (Table 1). It is seen that
D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102 93
Fig. 7. Activation volume as a function of the effective owstress on a log scale
for SC alloy and pure Al. Activation are is given on the right axis.
DC data shows the lowest Haasen intercept among all systems
considered, which also has positive value.
Strain rate sensitivity of pure Al increases as tempera-
ture decreases; the SRS values are equal to 2.98 10
4
K
1
,
1.88 10
4
K
1
, and 0.29 10
4
K
1
at 4.2 K, 78 K, and
295 K, respectively. Fig. 6 shows that the intercept of pure Al
data on the Haasen plot is equal to zero at room temperature,
whereas it is slightly negative at lower temperatures. The reasons
behind this behaviour will be discussed later.
During the plastic ow of a material, dislocations overcom-
ing the obstacles move from a stable state (static equilibrium)
to an unstable state (second equilibrium). The apparent activa-
tion area, a, swept by a dislocation segment during the glide
between the unstable and stable equilibrium states at a given
stress is evaluated from the following equation [22,23]:
a =
MkT
b
ln

T,
=
MkT
mb(
y
)
(1)
where k is the Boltzmann constant, M is the Taylor factor, b is
the Burgers vector, and m is the strain rate sensitivity in the
engineering denition.
Fig. 7 shows the activation volume, V

=b a, for SC alloy
and pure Al as a function of the effective ow stress on a log
scale, where b is the Burgers vector (0.28 nm) and a is the
activation area. It is seen that pure Al exhibits an activation
area with a strong temperature dependence. For example, at the
effective ow stress of 25 MPa, the activation areas have been
determined to be approximately 10 nm
2
, 30 nm
2
, and 175 nm
2
at 4.2 K, 78 K, and 295 K, respectively. For a given ow stress,
the activation area increases as temperature increases due to
the higher contribution from the thermal energy. At a constant
temperature, the activation area decreases as the ow stress
increases (Fig. 7). The presence of solute atoms reduces the
activation area of a material; the activation area of SC alloy at
78 K decreases from approximately 9 nm
2
to 2.5 nm
2
, close to
the values observed for pure Al at 4.2 K (Fig. 7).
3.3. Electrical resistivity
Fig. 8(a) shows raw data of deformation-induced resistivity
over the residual resistivity value as a function of the true strain
for Al, SC and DC alloys deformed at 4.2 K. The super-pure
Al used in this work enabled us to measure change in electrical
resistivity in this material during deformation at 78 K, which is
included in the graph. It is seen that, in every case, the resistivity
increases parabolically with the strain, indicating that deforma-
tion is homogeneous on the macroscopic scale. Compared to
pure Al, the alloys show higher scattering in resistivity data due
to the contribution from solute and impurity scattering. Assum-
ing that all deformation-induced defects are dislocations, the
electrical resistivity is recalculated into dislocation densities,
using the value of 1.8 10
25
m
3
for specic resistivity per
unit length of dislocation in Al [24]. The data in Fig. 8(a) can
thus be interpreted as the evolution of dislocation density dur-
ing plastic deformation for both pure Al and Al alloys. It is
seen that pure Al accumulates in average about 7 10
15
m
2
dislocation density before the fracture. The total dislocation
density stored in SC alloy before the fracture is higher and
is of the order of 1.51.6 10
16
m
2
. DC material accumu-
Fig. 8. Evolution of defect-induced resistivity change during deformation as a
function of (a) true strain and (b) the effective ow stress for SC alloy and pure
Al.
94 D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102
lates in average 1.1 10
16
m
2
dislocation density before the
fracture.
Fig. 8(b) shows deformation-induced resistivity as a function
of effective ow stress for all materials studied. For pure Al,
resistivity increases roughly parabolically with the ow stress
up to the fracture. The resistivity characteristics for alloys show
appreciable scattering of data points; the scattering increases as
the ow stress increases. This reects the unstable deformation
due to adiabatic shearing of the lattice, which contributes to
the scattering of the ow stress at the same dislocation content.
Fig. 8(b) shows that the resistivity evolution as a function of
stress for the SC alloy follows closely the characteristic of pure
Al. It is seen that, for a given ow stress level, resistivity of DC
alloy is lower than that of pure Al or SC material in a broad
range of stresses. As previously shown, the resistivity data can
be recalculated into the dislocation densities. This data shows
that, at the same ow stress level, Al deformed at 78 K produces
the highest dislocation densities among all materials studied.
3.4. TEM observations
Fig. 9 shows the microstructure of SC alloy before
and after plastic deformation at different temperatures. The
annealed SC material contains non-homogeneously distributed
ne particles smaller than 1 m in diameter, produced dur-
ing strip casting of the alloy [25]. These particles are
observed both in the form of strings aligned along rolling
direction as shown in Fig. 9(a) and also in the form of ran-
dom clusters distributed non-homogeneously in the Al-rich
matrix.
Fig. 9(b) shows TEM observations of the microstructure in
the SC material deformed 10% at room temperature. Disloca-
tions are stored in the entire volume of the material and one
can recognize a formation of characteristic cell structure with
the higher dislocation density accumulated in the cell wall and
somewhat less dense in the cell interior. The average cell size
of this structure is of the order of 0.50.7 m. It is seen that
the formation of these features is not associated with the second
phase particles as no particles are found in this area. The alloy
deformed 17%at 4.2 Kexhibits on average much higher disloca-
tiondensities. Insome areas, as is showninFig. 9(c), dislocations
form insulated cells with a size of the order of 0.20.5 m. In
highly deformed samples (Fig. 9(d)) the dislocations are stored
quite homogeneously in the volume of the specimen and there
is no clear separation into higher and lower dislocation density
areas as it was seen in the sample deformed at roomtemperature.
Fig. 10 shows examples of the dislocation substructure in
pure Al deformed till fracture, i.e. 35%, 30%, and 43% of strain
at 295 K, 78 K, and 4.2 K, respectively. In comparison with the
SCalloy, the substructure of pure Al is markedly different. In all
cases, the micrographs show well-developed substructures with
dislocation sub-boundaries separating areas of recovered cell
interiors. There are substantially less defects found inside the
subgrains than was observed in SC alloy. Smaller cells are pro-
duced at lower temperatures. The scale of the subgrains ranges
between 1 m and 3 m in material deformed at 295 K, 0.6 m
and 1.7 m in Al deformed at 78 K, and 0.2 m and 0.7 m in
Al deformed at 4.2 K.
TEM observations (Figs. 9 and 10) show that the disloca-
tion content and the nature of dislocations stored in SC material
are different from these found in pure Al. Pure Al exhibits a
strong tendency towards grain subdivision and development of
low misoriented sub-boundaries within larger grains. This has
not been observed in SC alloy, which tends to form a rela-
tively homogeneous network of dislocations, particularly at low
temperatures.
4. Discussion
4.1. Deformation and work-hardening behaviour
Strip cast aluminum alloys deformed at 4.2 K and 295 K
exhibit plastic ow instabilities associated with two different
processes operating during tensile deformation of the material.
In the subsequent section, we discuss briey these processes,
to develop a better understanding of the hardening behaviour in
materials studied in this work.
At room temperature, PLC deformation is responsible for
the serrated plastic ow of SC alloys. The magnitude of the
serrations increases as strain increases. The PLCinstabilities are
classied into ve different types AE [18]. The SCalloy shows
B-type serrations through most of the plastic owat 295 K, with
a small onset of type Ainstabilities observed at the early stage of
deformation (Figs. 1 and 2). As the strain rate decreases, more
solute atoms diffuse to dislocations and the magnitude of stress
drops (oscillation) increases (Fig. 1). Present results agree with
other studies of the PortevinLeChatelier effect in Al alloys,
published in the literature, e.g. [17,26].
Load instabilities at 4.2 K arise from the thermomechanical
instability in the substructure occurring during adiabatic shear-
ing of the lattice. These processes have been studied in detail for
various systems in the 1950s and 1960s including pure metals
[27,28] and Al alloys [2931]. When shear deformation initiates
at low temperatures due to the low thermal conductivity of the
material, the heat generated during this process is localized in
a small volume of the deformed specimen. Basinski [31] mea-
sured that during adiabatic deformation the temperature of pure
copper increases locally to 60 K. One can expect that the local
increase of the temperature in the SC sample during adiabatic
shearing can be at least of this order of magnitude or higher. The
localized heating produces softening of the substructure along
the deformation path and leads to the shear localization on the
length scale crossing a number of grain boundaries. The adi-
abatic deformation produces load drop instabilities associated
with characteristic sound clicks and affects the work-hardening
behaviour of the material at 4.2 K. The adiabatic shear localiza-
tion can also be initiated at a very high strain rate where the time
for the thermal diffusivity is short and the generated heat cannot
be dissipated in the volume of the material [32,33]. In samples
deformed at a lowstrain rate, the heat per unit time generated by
moving dislocations is lower and the local temperature increases
slowly. Hence, the alloy deformed at 4.2 K with a lower strain
rate exhibits less frequent stress drops, whereas it shows exten-
D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102 95
Fig. 9. TEM observations of the microstructure of SC alloys: (a) prior to deformation, (b) the substructure developed after 10% of deformation at 295 K at the
ow stress of 130 MPa, (c) dislocation substructure developed after 17% of deformation at 4.2 K corresponding to the ow stress of 320 MPa, and (d) dislocation
substructure developed in the SC sample deformed till fracture, i.e. after 33% of deformation at 4.2 K corresponding to the ow stress of 520 MPa (TA denotes the
tensile direction).
sive serrations with a high frequency of load drop at 295 K due
to PLC deformation.
In comparison with Al alloys, pure Al shows a stable defor-
mation at 4.2 Kbecause of the better thermal conductivity, which
allows fast dissipation of the heat produced during the slip event.
This is also reected in work-hardening characteristics of both
materials (Fig. 5), showing that the work-hardening rate of SC
alloy at 4.2 K decreases faster than that of pure Al, suggesting
96 D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102
Fig. 10. TEM observations of the microstructure of pure Al deformed till fracture at different temperatures, i.e. (a) 35% of deformation at 295 K corresponding to a
ow stress of 50 MPa, (b) 30% of deformation at 78 K at the ow stress of 150 MPa, (c) 43% of deformation at 4.2 K corresponding to a ow stress of 415 MPa.
that the recovery rate is higher in alloy than pure metal. Such
behaviour of the work-hardeningresults fromthe adiabatic local-
ized heating causing softening of the substructure and inducing
shear localization, which increases the apparent recovery rate in
SCalloy. In the absence of adiabatic deformation one can expect
that the recovery rate in SC material should be lower.
Present results reveal that SCalloy deformed at 4.2 Kexhibits
a remarkably high strength of the order of 700 MPa just before
the fracture (Fig. 1). This is comparable to the strength level
achieved in precipitation-hardened materials [19,21]. Compared
to those age-hardenable AlMg and AlZnMg alloys, the hard-
ening rate of SC alloy at 4.2 K is much higher in the wide
D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102 97
Fig. 11. The effect of dislocation storage on the strain hardening of SC alloy
and pure Al, respectively.
o
is dened as the slope of a straight line though the
origin in the region of linearity (see text for details).
range of ow stresses (e.g. Fig. 5). It has been suggested that,
in non-heat-treatable alloys, solute atoms lower stacking fault
energy, therefore reducing dynamic recovery and enhancing
work-hardening capacity [34]. The strong hardening of SC
alloy must be attributed to the inuence of alloying elements
present in the solid solution and it is evident, from the results
shown in Fig. 5, that this inuence is particularly effective at
low temperatures. Lloyd [7] studied mechanical properties and
work-hardening behaviours of a 5xxx series of AlMg alloys
down to 85 K and concluded that out of all elements present in
these systems Mg inserts the strongest effect on the hardening
behaviour through the solute drag contribution, which inhibits
dislocation rearrangement and increases hardening rate.
With the exception of 4.2 K, which will be discussed later,
the work hardening decreases approximately at the same rate
for both materials at 78 K and slightly faster in Al than SC at
298 K(Fig. 5), indicating that the solute atoms and impurities do
not change substantially the kinetics of dynamic recovery pro-
cesses at this temperature range. Thus, these processes must be
governed by dislocationdislocation interactions. As discussed
previously, the higher apparent recovery rate in SC material
at 4.2 K compared to pure Al is a direct consequence of the
adiabatic shear deformation, which produces substantial soften-
ing of the microstructure on the macroscopic scale and affects
the work-hardening behaviour of the alloy. To provide a more
detailed picture of the storage mechanisms operating during
deformation of SC alloy in comparison with pure Al, Fig. 11
shows characteristic of (
y
) d/d plotted as a function of
effective ow stress,
y
. In the KocksMecking descrip-
tion [35], the strain-hardening rate is expressed in terms of two
competing mechanisms: athermal storage of dislocations and
dynamic recovery:
=
o

r
( , T) =
o
k (2)
where is the macroscopic-hardening rate,
o
is the athermal
storage termof the strain-hardeningrate,
r
is the dynamic recov-
ery term, is the shear rate, T is the temperature, k is the constant
and is the shear stress. The rate of the work hardening decrease
is proportional to k representing the rate of dynamic recovery.
In the case of deformation of polycrystalline materials, one can
write by application of the Taylor factor that [35]:
(
y
) = (
y
)[
o

r
(, , T)] (3)
where is the true tensile stress,
y
is the yield stress, and is
the strain rate. The initial slope of the relationship (3) represents
the athermal storage of dislocations, whereas deviation fromlin-
earity occurring at higher stresses indicates the onset of dynamic
recovery process. In the present experiments
o
for Al is equal
to about 1.3 GPa, whereas for SCalloy
o
is found to be equal to
approximately 2.2 GPa.
o
is thus almost exactly equal to E/50
for pure Al and E/32 for SC alloy. The athermal storage term,

o
, is larger for SC alloy than pure Al. This indicates that SC
alloy more effectively accumulates dislocations in the structure.
The athermal-storage contribution
o
is related to the dislo-
cation mean free path, l, and dislocation density, , through the
following equation [35]:

o
=

2l

(4)
where is the shear modulus and is the constant. Refer-
ring back to Fig. 11 and data at 4.2 K, SC characteristic deviate
from the linearity at the effective ow stress of approximately
7080 MPa whereas pure Al at about 80100 MPa, correspond-
ing to average stored dislocation densities of approximately
5 10
13
m
2
and 3 10
14
m
2
, respectively (Fig. 8(b)). TEM
observations reveal independently that at this deformation stage
both materials develop a relatively homogeneous dislocation
substructure. One can estimate that the average dislocation mean
free path calculated based on Eq. (4) is equal approximately
to 0.3 m for pure Al and about 0.6 m for SC alloy. This
is much smaller than the initial grain size of these materials
(600 m Al and 14 m SC alloy) and much larger than the aver-
age distance between solute atoms in the glide plane in SC alloy
1.5 nm. In case of pure Al, the dislocation mean free path
corresponds rather closely to the size of the cell structure pro-
duced after large deformation. More detailed discussion of this
aspect of the present work will be provided later in Section 4.3.
However, based on the above estimate one can conclude that
the dislocationdislocation interactions (not dislocationgrain
boundaries or dislocationparticle or dislocationsolute interac-
tions) are important in the dislocation storage in these materials.
The dislocation mean free path corresponds relatively closely to
the size of the observed cell structure in Al, which indicates that
dislocation storage occurs on cell walls and therefore processes
occurring within the cell walls determine the rate of the dynamic
recovery in this metal at least at the low temperature regime.
After a certain amount of plastic deformation, (
y
) d/d
curves (Fig. 11) develop a negative curvature. At 4.2 K, the
downward curvature occurs at approximately the same ow
stress of about 350 MPa for both pure Al and SC alloy, cor-
responding to average dislocation density of approximately
5 10
15
m
2
(Fig. 8). Although not conrmed, it is unlikely that
the point at which these curves turn down is associated with the
development of any form of damage process because it occurs
98 D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102
relatively far below the ultimate tensile strength of these materi-
als. The alternate possibility is that some accelerated recovery is
taking place in the substructure, initiated at a dislocation density
of the order of 4 10
15
m
2
. It is noted that at higher tempera-
tures (e.g. 78 K) dynamic recoveryoccurs muchearlier inpure Al
than SC alloy. This implies that SC material is more resistant to
this process, which intensies after more advanced deformation
than in pure Al.
TEM observations reveal that, at the ow stress of
300350 MPa, Al develops a well-dened cell structure whereas
SC produces a relatively homogeneous, Taylor-like dislocation
arrangement. We estimate that the relative volume fraction of
the cell walls to the cell interiors in pure Al is approximately
3070%, whereas the density of dislocations stored in the walls
relative to the dislocation density inside the cell is in the ratio
10:1 (Fig. 10(c)). Assuming that the simple rule of mixture
applies, i.e.:

T
= f
w
+(1 f)
i
(5)
where
T
is the total dislocation density stored in the mate-
rial,
w
is the dislocation density stored in the wall,
i
is
the dislocation density stored in the cell interior and f is the
volume fraction occupied by the cell walls, one can estimate
that the downward curvature of (
y
) d/d curves or the
accelerated dynamic recovery occurs in Al when the disloca-
tion density in the wall and the cell interior are approximately:

w
=1.1 10
16
m
2
and
i
=1.1 10
15
m
2
. As mentioned
previously for SCalloys this process occurs at the average dislo-
cation densities 4 10
15
m
2
. These gures correspond to the
average dislocation spacing in SCalloy of the order of 16 nmand
approximately 10 nmand 30 nmin the wall and inside the cell of
pure Al, respectively. It is seen that dynamic recovery intensies
when the spacing between dislocations approaches 1015 nm,
indicating that similar fundamental processes responsible for
dislocation annihilation operate in Al and SC alloy.
It will be of interest to apply the same approach to estimate
the average dislocation spacing in both materials at the point of
fracture. In this case, the dislocation densities are approximately

w
=1.9 10
16
m
2
and
i
=1.9 10
15
m
2
in the cell wall and
the cell interior of pure Al and 1.5 10
16
m
2
in SC alloy.
This represents average dislocation spacing in the SC alloy of
the order of 8 nm and approximately 7 nm and 23 nm in the wall
and inside the cell of pure Al, respectively. It is striking that the
average dislocation spacing in SC alloy and in highly dislocated
walls of Al correspond very well to each other. This may indicate
that, independent of the composition of the system, the fracture
initiates by some form of dislocation collapse occurring in the
areas where the dislocation densities approach a critical level
or a critical spacing for spontaneous annihilation. We estimate
this spacing for Al and SC alloy to be between 7 nm and 8 nm
at 4.2 K.
In pure Al deformed at 78 K the total dislocation density
at the fracture is 1.8 10
15
m
2
(Fig. 8). The volume frac-
tion of the cell walls to the cell interiors is approximately
2080% and we estimate the similar 10:1 ratio for the den-
sity of dislocations stored in the wall relative to cell interior
in Al deformed at 78 K (Fig. 10(b)). The dislocation densi-
ties,
w
and
i
, are thus approximately:
w
=6.4 10
15
m
2
and
i
=6.4 10
14
m
2
, which represents average dislocation
spacing of 12 nm and 39 nm in the wall and inside the cell of
pure Al at 78 K. We do not attempt to evaluate these gures for
Al deformed at 295 K or for alloys, as they would be subjected
to a large error.
4.2. Mechanism of plastic deformation
Haasen characteristics show that one mechanism of plas-
tic deformation operates during deformation of pure metal
and also in Al alloys studied in this work (Fig. 6). Haasen
plot of pure Al deformed at room temperature passes
through the origin, suggesting that the ow stress is deter-
mined by dislocationdislocation interactions and obeys the
CottrellStokes law, which is also conrmed in Fig. 7. A good
linear relationship between the activation volume and the effec-
tive owstress (Fig. 7) indicates that CottrellStokes lawholds at
lowtemperatures as well as roomtemperature because it implies
that
e
V

, where
e
is the effective ow stress and V

is the
activation volume, is independent of strain, which is known as
the modied CottrellStokes law [36]. The deviation from the
linearity for the alloy in Fig. 7 is usually attributed to the impu-
rity effect on the short-range stress which gives rise to a high
solutedislocation component [37]. Al shows a slightly nega-
tive intercept at lower temperatures in the Haasen plot (Fig. 6),
which again suggests that its substructure provides a more strain
rate sensitive component for the ow stress than forest dislo-
cation interactions. This athermal component may arise from
the nature of the dislocation substructure produced at low tem-
peratures characterized by the higher volume fraction of the
cell walls comprising the higher dislocation content and larger
misorientations across the walls.
The y-axis intercepts of the Haasen plot for alloys are always
higher than for pure Al, due to the presence of solute atoms in
the aluminum matrix, which provide a larger thermal compo-
nent to the ow stress. If more solutes precipitate as in the case
of DC alloy, the thermal contribution decreases and the value of
y intercept is lower. This will also suppress or eliminate L uders
deformation as observed experimentally in Fig. 1(c). Therefore,
it is not surprising that SC alloy, containing more alloying con-
tents in the solid solution than DC alloy due to the effect of
the thermomechanical process, shows the higher positive inter-
cept on the Haasen plot. At the same time, SC alloy annealed
at 450

C, indicated as SC4 in Fig. 6, shows a lower thermal

component than our reference SC material annealed at 350

C.
It also shows the lower yield stress, but the work-hardening
behaviour of these two specimens is the same (Fig. 2). This
is attributed to two combined factors: one related to the grain
size difference and the second to the composition of the matrix.
The strain rate sensitivity of the material is very sensitive to the
compositional change; small changes in the solid solution are
immediately visible on the Haasen characteristic of the mate-
rial [22]. Optical observations reveal that, with the exception of
the grain size, there is no substantial difference in the nature,
the size or distribution of precipitates present in SC samples
D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102 99
annealed at 450

C and 350

C. This suggests that the larger

grain size of the sample annealed at a higher temperature affects
the yield stress and should lower the thermal component of the
ow stress, as observed experimentally. It is also interesting to
note that the y-axis intercept of SC alloys is close to the one of
solution treated AA6111 [21]. This indicates that both materials
are characterized by a similar strength of obstacles, which pro-
vides the similar thermal component to the ow stress during
thermal activation. When ne precipitates are formed in aged
AA6111, the thermal component is even larger than observed in
solution treated material [21].
4.3. Electrical resistivity and TEM
Analysis of the electrical resistivity data for pure Al and Al
alloys showthat the following relationship between the effective
ow stress and the total dislocation density holds up to large
stresses:

y
= MEb
n
(6)
where E is the Young modulus, b is the Burgers vector, M is the
Taylor factor 3.01, is a constant and n is another constant.
Fig. 12 plots log of the effective ow stress as a function of
the log of the dislocation density, i.e. f(log(
y
)) =log()
for different materials. The slope of this gure gives a value of
the exponent, n, in Eq. (6), which varies between pure Al, Al
alloys anddeformationtemperature (Fig. 12). The nvalues deter-
mined from the linear regression analysis are given in Table 2.
Al deformed at 4.2 Kshows that the owstress is proportional to
the square root of the total dislocation density, i.e.
1/2
(e.g.
[38]). SC and DC alloys exhibit substantial deviations from this
relationship with the n exponent equal to 0.29 and 0.34 for DC
and SCalloy, respectively (Table 2). It is important to emphasize
that Eq. (6) reects the relationship between the total disloca-
Fig. 12. Flow stress as a function of the total dislocation density for Al, SC and
DC alloy. Slopes of these relationships give values of the exponent n in rela-
tionship given in Eq. (6). The n values are 0.54, 0.5, 0.34, 0.29 for Al deformed
at 78 K and 4.2 K, SC alloy deformed at 4.2 K and DC alloy deformed at 4.2 K,
respectively (see the text for details). Linear regression lines are also included
in the gure.
Table 2
Values of the tting parameters n, k
1
, and k
2
to Eqs. (6) and (7)
Materials
Al (78 K) Al (4.2 K) SC (4.2 K) DC (4.2 K)
n values 0.54 0.5 0.34 0.29
k
1
(10
8
) 2.52 4.22 8.12 7.07
k
2
2.24 1.10 0.40 0.10
tion density stored in the materials during deformation at 4.2 K
and the ow stress. It has been shown on copper that total dis-
location density produced during deformation at 4.2 K has two
components: one contributes to the ow stress and resistivity
(unrecoverable defects) and the other contributes only to the
electrical resistivity (recoverable defects) [16]. Our preliminary
studies of this effect suggest that it is similarly true for the case
of low temperature deformation of pure Al. One can thus expect
that, after accounting for these two dislocation density compo-
nents, n values corresponding to the relationship between the
density of unrecoverable defects and the ow stress will be dif-
ferent from those given above. Basinski and Basinski [39], after
compiling various data available in the literature, reported for
pure copper the relationship:
0.43
between the ow stress
and the dislocation density, which is attributed to the effect of
other obstacles such as impurities on the ow stress.
Let us now discuss in more detail the evolution of electrical
resistivity (total dislocation density) in pure Al. For a given ow
stress, the resistivity change of Al at 4.2 K is lower than at 78 K
(Fig. 8(b)). This indicates that, inorder tosupport a givenstress, a
higher dislocationdensitymust be storedinthe material at higher
temperatures (see also Fig. 12). This reects the strength of the
obstacle structure that a given dislocation distribution exerts on
the mobile dislocations during plastic ow. It is clear that the
dislocation network produced at 78 Kprovides weaker obstacles
than the network developed at 4.2 K, so that more defects are
required to resist the glide of mobile dislocations at the same
stress level.
The strength of the obstacle structure is reected in the values
of the activation area shown in Fig. 7. It is seen that the activation
area at 78 K is larger than at 4.2 K indicating that dislocations
can more easily overcome the obstacles. On the other hand the
activation area of SC alloy at 78 K resides close to the range of
the one of pure Al at 4.2 K, which suggests that, at the same tem-
perature, SC alloy develops stronger and more stable obstacle
structure than Al, in agreement with mechanical characteristics.
The evolution of the dislocation mean free path in Al and Al
alloys is obtained from the variation of dislocation density with
strain (Fig. 8(a)) according to the expression

l = M/b(d/d),
where M is the Taylor factor 3.01, b is the Burgers vector, is
the density of dislocations and is the true strain [35]. Fig. 13
shows the dislocation mean free path as a function of effective
ow stress. For materials deformed at 4.2 K the mean free path
decreases exponentially and for DC and SC alloy it stabilizes
at the value of 0.1 m at the fracture. In case of Al deformed
at 4.2 K and 78 K the mean free path at the point of fracture is
equal to 0.4 mand 1.5 m, respectively. It is interesting to note
100 D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102
Fig. 13. Evolution of the dislocation mean free path for SC and DC alloys
deformed at 4.2 K and Al deformed at 4.2 K and 78 K.
that at 78 K the mean free path decreases roughly linearly with
the stress, what suggest that plastic deformation of aluminum at
78 K is controlled by different kinetics of storage and recovery
than at 4.2 K.
In all above cases, the mean free path of dislocations deter-
mined from electrical resistivity measurements shows a very
good agreement with the scale of the dislocation substructure
producedduringdeformationas seenbyTEM. Inthe case of pure
Al deformed at 4.2 K, the mean free path of 0.4 mat the fracture
corresponds to the size of the cell structure, which indicates that
cell boundaries provide signicant rates of dislocation storage.
In Al alloys, on the other hand the dislocations are stored quasi-
homogeneously in the volume and the mean free path reects
the spacing between obstacles, i.e. forest dislocations within the
dislocation network.
As the mean free path decreases and the density of dislocation
increases, dynamic recovery plays a more important role during
plastic deformation. This process has been discussed in Section
4.1 based on the behaviour of the work-hardening rate and it was
concluded that at temperatures of 78 K and 298 K the kinetics
of dynamic recovery of Al and Al alloy is very similar and it
seems to be controlled by the dislocation annihilation occurring
within highly dislocated areas of the substructure. The electrical
resistivity results can provide insight into these processes occur-
ring at 4.2 K. As known, the evolution of the dislocation density
with strain has commonly been assumed to be represented by
the sum of two contributions. One contribution is proportional
to the square root of the dislocation density associated with the
hardening storage rate and another contribution is proportional
to the dislocation density associated with the recovery rate. The
rate of dislocation density change with strain is thus expressed
as
d
d
= k
1

k
2
(7)
where k
1
is the constant representing the work-hardening rate,
whereas k
2
is another constant proportional to the rate of
dynamic recovery. In the present work these contributions have
been determined by tting the relationship of d/d =f() with
k
1

k
2
. The experimental data points and ts are shown
in Fig. 14, and constants k
1
and k
2
are given in Table 2. For
pure Al, the constant k
1
increases as the temperature decreases
from 78 K to 4.2 K and is higher for SC and DC alloys than pure
Al. These results, in addition to the hardening characteristics
discussed in Section 4.1, give direct evidence that the work-
hardening component of the ow stress is larger in Al alloys
than in pure Al. On the other hand, Table 2 shows that for pure
Al, the k
2
value is larger at 78 K than at 4.2 K and k
2
in Al alloys
is lower than in pure metal. This indicates that annihilation of
dislocations occurs with a higher rate at higher temperatures
where the thermal energy is available, whereas these processes
are slowed down in Al alloys. This is attributed to the effect of
solute atoms, which increases the dislocation storage capacity
and suppresses dynamic recovery possibly through their inu-
ence on the stacking fault energy, affecting the nature of the
dislocation structure produced in these materials. It should be
emphasized here that the above analysis reects changes occur-
ring entirely within the dislocation substructure without regard
to the macroscopic deformation behaviour of the materials. As
discussed in Section 4.1, the work-hardening characteristic of
SC alloy at 4.2 K (Fig. 5) gives a higher apparent recovery rate
than in pure Al, as a result of adiabatic deformation. One can see
that electrical resistivity data are insensitive to this effect, as the
signal arises from the integrated defect content in the sample,
which does not decrease during adiabatic shearing.
4.4. Fracture
Let us nally discuss the way our materials approach frac-
ture. In all materials studied, fracture occurs reproducibly within
a narrow range of strength, work-hardening rate, dislocation
density and the dislocation mean free path characteristic for
a given material and deformation temperature. The fracture
Fig. 14. Rate of the dislocation density change with the strain d/d as a function
of the dislocation density . The characteristics are tted to the rate equa-
tion: d/d = k
1

k
2
with k
1
and k
2
constants given in Table 2. Both
the experimental data points and the tting curves are shown in the graph.
D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102 101
surface observations reveal that all materials undergo ductile
fracture by the nucleation and growth of voids. In the case of
Al alloys failure is not particle-stimulated and particles seem
to have weak inuence on ductility. From the work-hardening
characteristics (Fig. 5), it can be deduced that, at 4.2 K, fracture
in Al alloys occurs far before the work-hardening capacity is
exhausted, i.e. before Considere criterion is meet. Therefore,
there must be some mechanism that triggers the nucleation
of voids and leads to the material failure. In Section 3.1, it
was shown that SC samples exhibit the largest elongation at
78 K where both PortevinLeChatelier and adiabatic deforma-
tion are suppressed. This suggests that these processes facilitate
the premature necking and fracture of samples. The litera-
ture provides a considerable amount of data suggesting that
PLC deformation has a detrimental effect on ductility, e.g.
[4042].
There is much less data available regarding the effect of
adiabatic deformation on deformation behaviour and frac-
ture. Present results reveal that the adiabatic deformation
affects the macroscopic work-hardening rate of the alloys and,
although it may aid in nucleating strain localization, adiabatic
shear is not a primary cause of the fracture. This is con-
cluded from the fact that the homogeneous elongation of the
sample deformed with a higher strain rate, where the adia-
batic deformation operates very intensively, is not lower than
the elongation of samples deformed at a lower strain rate
(Fig. 1).
TEM observations show that there are signicant differences
in the nature of the dislocation substructure developed in Al
and Al alloys. The intriguing feature is that the average spacing
between dislocations within highly dislocated areas at the point
of fracture is represented by the same gure in both Al and Al
alloys. Thus, it seems that the failure occurs when a certain dis-
location density limit or a critical dislocation spacing is achieved
in the substructure to allow for spontaneous annihilation of dis-
locations and global collapse of the dislocation network in these
areas. We estimate a critical dislocation spacing of about 8 nmin
Al and Al alloys at 4.2 K and 12 nm in pure Al at 78 K. These
gures correspond well to Browns theoretical estimations of
critical height for the athermal collapse of dislocation dipoles in
copper [43].
Destabilization of the dislocation substructure at stresses,
as these developed before the fracture, will cause persistent
ow localization in softened volumes of the material and will
lead to nucleation of voids, their growth and eventually neck-
ing and failure. Assuming that voids form in highly dislocated
areas (e.g. cell walls in Al), this should produce a fracture
surface with a high density of small dimples at low temper-
atures as there are many places where voids will nucleate.
Contrarily, the coarser scale of the substructure developed at
higher temperatures provides less potential sites for the void
nucleation. Thus, there will be fewer voids formed; conse-
quently, a few voids with the large size eventually dominate
in the fracture surface. This is reected in the nature of
the fracture surface produced in Al and Al alloys at 295 K
which is lled up predominantly with a few large dimples
(Figs. 3 and 4).
5. Summary and conclusions
Deformation behaviours during tensile tests of strip and
direct chill cast AA5754 Al alloys and of high-purity Al have
been studied. The focus is on the low temperature regime
where mechanical properties of these materials have not been
sufciently explored. Al alloys show two kinds of ow insta-
bilities, at 295 K resulting from dynamic strain aging known
as PortevinLeChatelier effect and at 4.2 K from adiabatic
deformation. At 78 K, the ow instability is suppressed and
these materials exhibit a homogeneous deformation. Plas-
tic deformation of high-purity Al is governed mainly by
dislocationdislocation interactions, and a stronger athermal
component of the ow stress observed at lower temperatures
arises from the nature of the dislocation cell structure produced
in this material. Al alloys exhibit a larger thermal component of
the ow stress due to the presence of solute atoms in the matrix.
Electrical resistivity measurements reveal that all materials
exhibit different kinetics of dislocation storage and recovery,
which is reected both in the constitutive relationship between
ow stress and the total dislocation density and in the way these
materials approach fracture. The results reveal that SC alloy
deformed at 4.2 K exhibits a remarkably high work-hardening
capacity, with the work-hardening rate of the order of /20 for
most of the plastic ow, attaining eventually the strength of
700 MPa just before the fracture. The work-hardening capacity
of SC alloy at 4.2 K is more effective than in precipitation-
hardened materials probably due to a higher dynamic recovery
component in heat-treatable alloys.
The evolution of the dislocation mean free path shows a good
agreement with TEM observations of the substructure and sug-
gests that dislocations are accumulated within the cell walls in
pure Al and quasi-homogeneously in the volume of Al alloys.
Particles have small or no inuence both in the process of dis-
location storage and during the fracture.
Direct measurements of the evolutionof defect content during
deformation at 4.2 K suggest that the rate of dynamic recovery
occurring within a dislocation network is slower in Al alloys
than in pure Al. However, adiabatic deformation, which oper-
ates during plastic ow, affects the macroscopic work-hardening
behaviour of alloys and gives the higher apparent recovery rate
than observed in pure metal.
Present results show that fracture occurs when dislocation
density approaches a critical level locally in highly dislocated
areas. Dislocation density gures give an estimate for the critical
spacing between dislocations in these areas to trigger sponta-
neous annihilation. At 4.2 Kthis gure is 8 nmfor all materials
whereas at 78 Kit has been estimated only for Al and is 12 nm.
The weakening of the substructure, arising from the collapse of
the dislocation network under the high stresses, forces the ow
localization, the nucleation of voids and fracture.
Acknowledgements
The authors wish to thank the Centre for Automotive Materi-
als and Manufacturing (CAMM) for nancial support. Dr. Olaf
Engler of Hydro Aluminium GmbH is gratefully acknowledged
102 D.-Y. Park, M. Niewczas / Materials Science and Engineering A 491 (2008) 88102
for providing high-purity aluminum used in this work. AA5754
alloys were kindly supplied by Novelis. We wish to thank Dr.
Bjorn Holmedal for comments on the manuscript.
References
[1] W.S. Miller, L. Zhuang, J. Bottema, A.J. Wittebrood, P. De Smet, A. Has-
zler, A. Vieregge, Mater. Sci. Eng. A 280 (2000) 3749.
[2] K. Spencer, S.F. Corbin, D.J. Lloyd, Mater. Sci. Eng. A 332 (2002) 8190.
[3] J. Sarkar, T.R.G. Kutty, K.T. Conlon, D.S. Wilkinson, J.D. Embury, Mater.
Sci. Eng. A 316 (2001) 5259.
[4] K. Spencer, S.F. Corbin, D.J. Lloyd, Mater. Sci. Eng. A 325 (2002)
394404.
[5] M. Verdier, M. Janecek, Y. Brechet, P. Guyot, Mater. Sci. Eng. A248 (1998)
187197.
[6] D. Kuhlmann-Wilsdorf, Mater. Sci. Eng. A 114 (1989) 141.
[7] D.J. Lloyd, Mater. Sci. Forum 519521 (2006) 5561.
[8] A.M. Hammad, K.K. Ramadan, M.A. Nasr, Z. Metallk. 80 (3) (1989)
173177.
[9] A.D. Rollet, U.F. Kocks, J.D. Embury, M.G. Stout, R.D. Doherty, P.O.
Kettunen, in: T.K. Lepisto, M.E. Lehtonen (Eds.), Strength of Metals and
Alloys, vol. 1, Pergamon Press, Oxford, 1988, pp. 433438.
[10] I. Barker, B. Ralph, N. Hansen, in: P.O. Kettnuneu, T. Lepisto, M.E.
Lehtonem (Eds.), Proceedings of the International Conference on the
Strength of Metals and Alloys (ICSMA8), Pergamon Press, Oxford, 1988,
pp. 277282.
[11] D. Chu, W.J. Morris Jr., Acta Mater. 44 (7) (1996) 25992610.
[12] S. Saji, S. Senda, S. Hori, Proceedings of the 7th International Conference
on the Strength of Metals and Alloys (ICSMA7), Montreal, Canada, 1986,
pp. 471476.
[13] M. Kawazoe, T. Shibata, T. Mukai, K. Higashi, Scripta Mater. 36 (6) (1997)
699705.
[14] A.P. Reynolds, Q. Li, Scripta Mater. 34 (11) (1996) 18031808.
[15] S.-S. Kim, M.J. Haynes, R.P. Gangloff, Mater. Sci. Eng. A203 (12) (1995)
256271.
[16] M. Niewczas, Z.S. Basinski, S.J. Basinski, J.D. Embury, Phil. Mag. A 81
(2001) 11211142.
[17] W. Wen, J.G. Morris, Mater. Sci. Eng. A 354 (2003) 279285.
[18] P. Rodriguez, Bull. Mater. Sci. 6 (4) (1984) 653663.
[19] A. Deschamps, M. Niewczas, F. Bley, Y. Brechet, J.D. Embury, L. Le Sinq,
F. Livet, J.P. Simon, Phil. Mag. A 79 (10) (1999) 24852504.
[20] P. Haasen, Phil. Mag. 3 (1958) 384418.
[21] S. Esmaeili, L.M. Cheng, A. Deschamps, D.J. Lloyd, W.J. Poole, Mater.
Sci. Eng. A 319321 (2001) 461465.
[22] B.J. Diak, K.R. Upadhyaya, S. Saimoto, Prog. Mater. Sci. 43 (1998)
223363.
[23] U.F. Kocks, A.S. Argon, M.F. Ashby, Thermodynamics and Kinetics of
Slip, Pergamon Press, 1975.
[24] B.R. Watts, in: F.R.N. Nabarro (Ed.), Dislocations in Solids, vol. 8, North-
Holland, Amsterdam, 1989, pp. 175419.
[25] X.-M. Cheng, J.G. Morris, Mater. Sci. Eng. A 323 (2002) 3241.
[26] G. Horvath, N.Q. Chinh, J. Gubicza, J. Lendvai, Mater. Sci. Eng. A445446
(2007) 186192.
[27] W.F. Hosford, R.L. Fleischer, W.A. Backofen, Acta Metall. 8 (1960)
187199.
[28] G.Y. Chin, W.F. Hosford, W.A. Backofen, Trans. AIME 230 (1964)
437449.
[29] G.Y. Chin, W.F. Hosford, W.A. Backofen, Trans. AIME 230 (1964)
10431048.
[30] Z.S. Basinski, Proc. R. Soc. Lond. A 240 (1957) 229242.
[31] Z.S. Basinski, Aust. J. Phys. 13 (1960) 345358.
[32] T.J. Burns, M.A. Davies, Int. J. Plasticity 18 (4) (2002) 487506.
[33] A. Molinari, C. Musquar, G. Sutter, Int. J. Plasticity 18 (4) (2002) 443
459.
[34] D.A. Hughes, Acta Metal. Mater. 41 (5) (1993) 14211430.
[35] U.F. Kocks, H. Mecking, Prog. Mater. Sci. 48 (2003) 171273.
[36] K. Tanoue, Scripta Metall. Mater. 25 (1991) 565569.
[37] S. Saimoto, H. Sang, Acta Metall. 31 (11) (1983) 18731881.
[38] F.R.N. Nabarro, Z.S. Basinski, D.B. Holt, Adv. Phys. 13 (50) (1964)
193323.
[39] S.J. Basinski, Z.S. Basinski, in: F.R.N. Nabarro (Ed.), Dislocations in
Solids, vol. 4, North-Holland, Amsterdam, 1979, pp. 261362.
[40] J.E. King, C.P. You, J.F. Knott, Acta Metall. 29 (1981) 15531566.
[41] L.P. Kubin, Y. Estrin, Acta Metall. 33 (1985) 397407.
[42] J. Kang, D.S. Wilkinson, M. Jain, J.D. Embury, A.J. Beaudoin, S. Kim, R.
Mishra, A. Sachdev, Acta Mater. 54 (2006) 209218.
[43] L.M. Brown, Phil. Mag. A 82 (9) (2002) 16911711.