Materials Science and Engineering A 552 (2012) 530–539

Contents lists available at SciVerse ScienceDirect

Materials Science and Engineering A

journal homepage: www.elsevier.com/locate/msea

Study of hot deformation of an Al–Cu–Mg alloy using processing maps and

microstructural characterization
C.M. Cepeda-Jiménez ∗ , O.A. Ruano, M. Carsí, F. Carreño
Department of Physical Metallurgy, CENIM, CSIC, Av. Gregorio del Amo 8, 28040 Madrid, Spain

a r t i c l e i n f o a b s t r a c t

Article history: The forming behaviour of an Al–Cu–Mg alloy (Al 2024-T351) has been studied by processing maps and
Received 10 April 2012 microstructural characterization. Torsion tests were conducted in the range 278–467 ◦ C, between 2.1 and
Received in revised form 24 May 2012 25.6 s−1 . Stress–strain curves obtained from the experiment data were fitted using the Garofalo equation
Accepted 27 May 2012
to obtain the constitutive parameters, obtaining a stress exponent of 6.1 and an activation energy of
Available online 5 June 2012
180 kJ/mol. Electron backscatter diffraction (EBSD) was employed to characterize the microtexture and
microstructure, before and after torsion testing, to evaluate the microstructural changes and instability
Keywords:
phenomena. A peak ductility of the Al 2024 alloy was found at about 400 ◦ C at all strain rates consid-
Electron backscatter diffraction-EBSD
Mechanical characterization
ered. According to the processing maps and microstructure observation, the optimum hot deformation
Aluminium alloys condition for the Al 2024 alloy is in the range 360–410 ◦ C and 2.1–4.5 s−1 . Under these favourable condi-
Hot deformation tions a uniform and fine grain size is obtained by extended dynamic recovery (DRV), which leads to the
Recrystallization formation of subgrain boundaries that progressively transform at large strains into new high angle grain
Processing maps boundaries.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction particular deformation mechanism is dominant [11]. In this study,

the analysis of the deformation behaviour of the alloy was con-
Al 2024 is an age-hardenable aluminium alloy, which is widely ducted by means of the Garofalo equation. This equation gives the
used in automobile and aerospace industry due to its good proper- envelope of all possible power law equations along the strain rate
ties like low density and good damage tolerance [1]. In addition, in range investigated. The Garofalo equation has the form [12]:
the last years investigations have been performed in order to extend  Q
its use in airplanes by introducing it together with other materi- ε̇ = A exp − [sinh ˛]n (1)
als (metallic and non-metallic) in multilayer laminate composite RT
materials with outstanding damage tolerance [2–4].
These metallic materials are frequently processed by hot rolling where A, ˛ and n are material constants, Q is the activation energy,
or forging, generally at elevated temperatures [5], being necessary R is the gas constant and T is the temperature.
to optimize processing parameters for maximum efficiency and The parameters of the Garofalo equation, A, Q, n, ˛, can be
workability. Workability or easiness of hot deformation is usually determined by a non-linear method, that allows an automatic
defined as the amount of deformation that a material can undergo calculation, involving an algorithm specifically developed for the
without cracking or reaching undesirable conditions at a given tem- treatment of this equation. The adjustment and statistical treat-
perature and strain rate [6]. ment of this equation was described elsewhere [13,14].
The processing-map technique has been widely used to under- In this work, torsion tests were conducted at various tempera-
stand the hot deformation of many materials [7–9], in terms of the tures and strain rates. The torsion test has been widely used for
various microstructural changes occurring at the different defor- evaluating the forming behaviour of materials at elevated tem-
mation conditions [10]. peratures [15]. Torsion testing offers the possibility of obtaining
To construct the maps, it is necessary to start from a large deformations without the occurrence of plastic instability,
constitutive equation that relates the experimental variables, under conditions simulating those encountered in industrial form-
stress-temperature-strain rate, obtaining the fields where a ing processes, such as hot rolling [16]. Accordingly, using torsion
data, the parameters of the Garofalo equation were calculated and
simulation of the forming process was conducted using processing
maps.
∗ Corresponding author. Tel.: +34 91 5538900x217; fax: +34 91 5347425. Processing maps were plotted based on the dynamic materials
E-mail address: [email protected] (C.M. Cepeda-Jiménez). modelling (DMM) [17], which considers the workpiece as a power

0921-5093/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.msea.2012.05.082
 C.M. Cepeda-Jiménez et al. / Materials Science and Engineering A 552 (2012) 530–539 531

dissipator [7]. The high temperature forming of metals can be ana- Complementary microstructural characterization thus plays a vital
lyzed by means of the supplied power to the material, P, that can role in understanding the deformation mechanisms associated
be divided in two terms: with a specific regime, and in this study it has been extensively
 ε̇   analyzed by the electron backscatter diffraction technique (EBSD).
P =  ε̇ = dε̇ + ε̇d (2) Therefore, the aim of the present investigation is to study the
0 0 high-temperature behaviour of an Al 2024 alloy using the analy-
or P = G + J, where G, the dissipator content, represents the power sis of hot torsion curves by the Garofalo constitutive equation and
dissipated by plastic work without changing the internal structure, processing maps to determine the optimum hot deformation con-
most of which is converted into viscoplastic heat; and J, the dissipa- ditions. Characterization of the developed microstructure is also an
tor co-content, is related to the power dissipated by metallurgical important aim.
processes, such as recovery, recrystallization, phase transforma-
tion, as well as damage of the material [17]. A relation of efficiency 2. Experimental procedure
factors for G and J, G and J , can be obtained by diving Eq. (2) by
the supplied energy: The aluminium alloy used in the present study was a rolled Al
2024-T351 plate of 12 mm in thickness. The composition in weight
1 = G + J (3)
percentage of the alloy is 4.43% Cu, 1.33 Mg, 0.04 Zn, 0.62 Mn, 0.17
Since the forming process of alloy implies changes in the internal Si, 0.01 Ni, 0.28 Fe, 0.02 Cr, 0.05 Ti, balanced Al.
structure, it is more interesting to study the term J which is defined Hot torsion tests were carried out on a computer-controlled and
by the relation: hydraulically powered torsion machine with the ends of the sam-
  ple axially fixed. The hot torsion machine, SETARAM 7MN, has been
1
J = ε̇d (4) described elsewhere [15]. The torsion samples were machined so
 ε̇ 0 that the gauge length coincided with the plate rolling direction.
The resolution of this equation would be extremely difficult if Samples 17 mm long and 3 mm radius in the gauge zone were tor-
the power law is used. This difficulty is avoided by the use of the sioned to fracture at constant temperature and strain rate. In this
Garofalo equation that is continuous and defined along the entire study, hot torsion temperatures were selected between 278 and
range of working strain rate. The following relation is obtained 467 ◦ C, and four strain-rates (ε̇) between 2.1 and 25.6 s−1 were
substituting Eq. (1) into Eq. (4): considered. Then, Al 2024 samples were solutioned at 467 ◦ C for
   Q 10 min and hold for 15 min for microstructural stabilization, and
1 afterwards, they were cooled in 2 min to testing temperature and
J = A exp − sinh(˛)n d (5)
 ε̇ 0
RT tested to failure. The range of deformation parameters of the tor-
sion test (T and ε̇) covered the conditions used during hot rolling
This parameter and its variation with temperature and strain
of multilayer materials containing this aluminium alloy [2–4]. Fur-
rate form the basis for construction of maps of constant forming
thermore, the recommended hot working temperature range for
efficiency contours.
most aluminium alloys is given between 260–480 ◦ C [6].
In addition, once the maximum efficiency region is determined,
The samples were introduced in a silica tube with an argon inlet,
it is necessary to locate the maximum stability region for the form-
to ensure protection against oxidation, and heated by a high fre-
ing process. Different instability criteria have been formulated to
quency induction furnace. The temperature during the torsion test
identify the regions of flow instabilities and in this work we used
was measured by a two-colour pyrometer. Immediately after frac-
the most common, the Lyapunov criteria [11,18–20].
ture, the samples were water quenched in less than 0.5 s in order
Thermodynamically, the stability is understood as the state
to retain the deformation microstructure, especially the grain size.
where the system evolves continuously diminishing the total
The torsion tests provided directly the curves of torque,  ,
energy. In engineering design, the control of the supplied and the
versus number of turns N. The effective stress (), the effective
dissipated energy is carried out by means of two variables. One is
strain (ε) and the strain rate (ε̇) were calculated by means of the
the entropy, S, whose variation is defined as:
following relationships [29,30]:
 
ıP √
Ṡ = (6)  3 2RN 2RṄ
ıT = (3 + m + ) ε = √ ε̇ = √ (9)
2R3 L 3 L 3
The other variable is the efficiency  = J/G that gives the dissipa-
tion rate. This variable can be substituted under certain conditions where R is the sample radius, L is the gauge length, m is the strain
by the strain rate sensitivity, m [19]. Thus, in this work the two rate sensitivity and  is the work hardening exponent:
 
Lyapunov stability criteria are given as:
∂ln  ∂ln 
  m=  =  (10)
ım ∂lnε̇  ∂lnε 
L1 = <0 (7) ε,T ε̇,T
ılnε̇

For the data analysis the peak stresses were taken into account,
and being  = 0. In addition m was taken equal to 0.13, which corre-
 
ıS sponds to the average value calculated from all torsion data.
L2 = <0 (8) Electron backscatter diffraction (EBSD) was employed in
ılnε̇
 the present investigation to characterize the microtexture and
Typical signatures of instability are shear band formation, microstructure of the Al 2024 alloy tested by hot torsion. The sam-
flow localization, formation of cavities, breaking of particles, etc. ples for EBSD observation were prepared on sections parallel to
[21–27]. the torsion axis at a distance of 0.7R, with R being the radius of
The combination of the efficiency of power dissipation map the gauge length as illustrated in Fig. 1. The EBSD scans were car-
and the instability map is the basis of a processing map, which ried out at the centre of the observation surface. The initial grain
could be used for correlating different regions where individual size of the alloy for each torsion condition was determined in the
microstructural processes occur with processing variables [28]. non-deformed region (head) of the torsion samples.
532 C.M. Cepeda-Jiménez et al. / Materials Science and Engineering A 552 (2012) 530–539

mechanically polished and then electropolished in a 30% nitric acid

solution in methanol at −28 ◦ C and 15 V.

3. Results

3.1. Torsion tests

Fig. 2 shows the results of the torsion tests to fracture carried

out at temperatures ranged between 278 and 467 ◦ C and at strain
rates between 2.1 (Fig. 2a) and 25.6 s−1 (Fig. 2d). Every curve shows
a rapid increase in the stress to a peak value ( p ), followed by dif-
ferent softening degrees until the rupture of the sample at all strain
rates, as a consequence of dynamic microstructural changes. At a
constant strain rate, the strain corresponding to the peak stress
increases slightly with the decrease in deformation temperature.
In addition, the flow stress decreases clearly with temperature at
Fig. 1. Illustration of the flat at a distance of 0.7R of the torsion sample that was a given strain rate, while an increase in the strain rate at a given
used for microstructural analysis. deformation temperature, especially at the lowest test tempera-
tures of 315 and 278 ◦ C, has little influence on the flow stress. At
Orientation maps were performed by EBSD in a scanning elec- the lowest temperatures of 315 and 278 ◦ C and strain rates higher
tron microscope (SEM) JEOL JSM 6500F, with a fully automatic EBSD than 9.6 s−1 (Fig. 2c and d), the stress–strain curves show consid-
attachment, HKL Technology, operating at an accelerating volt- erable softening after reaching the peak. However, at the higher
age and a working distance of 20 kV and 15 mm, respectively. The temperatures, the flow behaviour is clearly different. At tempera-
corresponding data processing was carried out using HKL Chan- tures higher than 360 ◦ C and clearly at low strain rates of 2.1–4.5 s−1
nel 5 software. The area mapping for the torsion tested samples (Fig. 2a and b), the initial work hardening region is reduced and
was 430 × 343 ␮m with a step size of 0.25 ␮m. The low angle nearly flat stress–strain curves are observed after a small initial
grain boundary (LAB) was defined by a misorientation between peak stress, which are indicative of steady-state flow behaviour.
adjacent grains of 2◦ <  < 15◦ , and the high angle grain boundary On the other hand, the curves of the figure clearly show that
(HAB) was defined by  > 15◦ . The HABs and LABs are shown as the strain to failure increases with decreasing strain rate at all test
black and white lines respectively on the maps. Specimens were temperatures. Fig. 3 shows more clearly the relation-ship between

Fig. 2. Equivalent stress vs. equivalent strain for the Al 2024 alloy deformed in torsion. The deformation temperatures were in the range 278–467 ◦ C and the strain rates
were (a) ε̇ = 2.1 s−1 , (b) ε̇ = 4.5 s−1 , (c) ε̇ = 9.6 s−1 and (d) ε̇ = 25.6 s−1 .
 C.M. Cepeda-Jiménez et al. / Materials Science and Engineering A 552 (2012) 530–539 533

Fig. 3. Plastic strain to failure (εF ) for the Al 2024 alloy plotted as a function of
temperature at several strain rates.
Fig. 4. Fitting representation of Zener parameter, Z = ε̇ exp(Q/RT ), as a function of
sinh(˛) for the Al 2024 alloy. Lines correspond to interval of confidence for fitting
elongation to failure, εF , and temperature at various strain rates. and upper and lower limits have probability of 95%.

It can be seen that both strain rate and temperature influence the
ductility. For a given temperature, the torsional ductility generally dissipation, J , which characterize the rate of microstructure evo-
decreases with an increase in strain rate. In addition, the general lution in the hot working process [28]. The values of J are between
trend is that, for a given strain rate, a maximum in ductility is 6 and 13% in the figure. It is shown that the efficiency increases for
observed at about 400 ◦ C, especially at low strain rates. It is worth decreasing strain rates at constant temperature. An efficiency range
noting the decrease in ductility at the highest test temperature of of 12–13% is spread from 360 ◦ C to 467 ◦ C at low strain rates, while
467 ◦ C at the lowest strain rates. it decreases to 9% at high strain rates over the same temperature
range, and 6–7% at lower temperatures. It is worth noting that the
3.2. Processing maps flow curves corresponding to the maximum efficiency exhibited
steady state behaviour (Fig. 2a and b).
The Garofalo Eq. (1) has been used to fit the experimental tor- Once the maximum efficiency region is determined, it is neces-
sion data and determine the constitutive parameters. Accordingly, sary to locate the maximum stability region for the forming process.
the optimal solution of the parameters of the Garofalo equation The representation of the 1st and 2nd Lyapunov criterion is given in
obtained in this study is the following: Figs. 6 and 7, respectively. These figures are two dimensional rep-
resentations of the stability criteria as a function of strain rate and
ε̇ exp(179.7 k J/mol RT ) = 4.16876 × 1013 (sinh(0.0103))6.1 (11) temperature. The most stable region in both figures corresponds to
the most negative values of the Lyapunov function. Fig. 6 shows a
For this alloy, the apparent stress exponent is nap = 6.1 and
stability band between 300 and 450 ◦ C, being optimum at approx-
the activation energy for deformation is Qap = 180 kJ/mol. The nap
imately 400 ◦ C for a strain rate of ∼5 s−1 (ln5 = 1.6), between the
value is lower than that consistent with a slip creep mecha-
level lines for −0.007. Similarly, Fig. 7 shows that the most stable
nism controlled by dislocation climb at constant substructure,
n = 8 [31], and the activation energy is higher than the value of
QL = 142 kJ/mol measured for lattice self-diffusion of aluminium
[32]. This is attributed to microstructural coarsening as will be
discussed below.
A representation based on the Zener-Hollomon parameter, Z =
ε̇ exp(Q/RT ), as a function of sinh(˛) is usually performed to prove
the goodness of the fit by the non-linear method used to obtain the
parameters of the Garofalo equation. This representation, given in
Fig. 4, involves combination of original variables, ε̇ and T, into Z.
The points given in the figure are obtained by means of the exper-
imental data using the parameters A, Q, n, ˛, obtained from the
non-linear method. These points fall close to the central line given
in this figure. This line corresponds to the best fit and the two adja-
cent lines defining the 95% confidence band for the best fit. Thus, as
the points fall between the 95% confidence band, and the correla-
tion coefficient of the final solution is r = 0.996, the fitting has good
accuracy and can be considered satisfactory [33].
In addition, the flow stress data have been analyzed using the
processing map technique described in the introduction to evaluate
the behaviour of the Al 2024 alloy during hot deformation.
Fig. 5 shows a two-dimensional map of constant forming effi-
ciency contours, which correspond to a projection on a ε̇, T plane. Fig. 5. Two-dimensional map of constant forming efficiency contours correspond-
The numbers against each contour represent efficiency of power ing to a projection on the ε̇, T plane.
534 C.M. Cepeda-Jiménez et al. / Materials Science and Engineering A 552 (2012) 530–539

(IPF) shown in the inset, and the colours represent the crystallo-
graphic orientations parallel to the normal direction (ND) for the
rolled as-received Al 2024 alloy (Fig. 8b), and to orientations par-
allel to the characterization plane normal, shown in Fig. 1, for the
torsion tested samples (Fig. 8c–d).
The microstructure of the as-received Al 2024-T351 alloy (Fig. 8a
and b) consists of recrystallized grains, with spacing between high-
angle grain boundaries (HABs) in the normal to rolling direction of
about 7.4 ␮m. The fraction of high-angle grain boundaries (fHAB )
was 92% and the {1 1 1} pole figure of Fig. 8b shows that the as-
received material was only weakly textured, containing grains of
a wide range of orientations. In addition, large insoluble iron-rich
intermetallic particles and partially soluble constituent particles
were observed to be randomly distributed, and they are ranged in
size from 0.5 to 5 ␮m (Fig. 8a). Previous works on this alloy [34,35]
have reported three types of intermetallic particles, Al7 Cu2 Fe,
(Al,Cu)6 Fe and Mg2 Si.
Fig. 8c and d shows the EBSD maps of the undeformed head of
torsion samples tested under different conditions. The shear direc-
tion in the EBSD maps is vertical and the radial direction is normal to
Fig. 6. Two-dimensional representation of the 1st Lyapunov criterion as a function the page. Both EBSD maps show a very large recrystallized grain size
of strain rate and temperature. elongated in the original rolling direction (∼40 ␮m in thickness).
The large grain growth observed in the undeformed head of the
region in this figure is that comprised between the level lines for torsion samples is produced during static annealing at 467 ◦ C [36],
−0.25 to −0.23. This region is strain rate dependent and optimum and the grain size does not change during the subsequent cooling
stability values between 350 and 400 ◦ C are obtained for about to other test temperatures (Fig. 8d).
4.5 s−1 . It presents thus the lower chance of crack appearance and Fig. 9 shows EBSD maps and {1 1 1} pole figures of tested sam-
is recommended for a stable forming process. ples at 4.5 s−1 and different temperatures. This strain rate has
been selected because the flow curves at the various temperatures,
3.3. Microstructure of the deformed samples Fig. 2b, show the different deformation stages of this material,
i.e., samples tested at this ε̇ and high temperature present simi-
In order to understand the effect of torsion deformation on the lar steady-state flow curves than those tested at lower ε̇ (Fig. 2a);
structure and correlate with the processing maps, the microstruc- and on the contrary, samples tested at 4.5 s−1 and low temperature
ture of the as-received and torsion tested Al 2024 alloy was presents similar low ductility, peak stress and flow softening than
evaluated. samples tested at higher ε̇ (Fig. 2c and d).
Fig. 8 shows a backscattered electron micrograph of the as- In general, the EBSD maps for all torsion test conditions of Fig. 9
received Al 2024-T351 alloy (Fig. 8a) and its corresponding EBSD have a common microstructure constituted by the recrystallized
map (Fig. 8b) in the LT orientation (longitudinal-transversal). In grains, observed in the undeformed head, elongated around the tor-
addition, Fig. 8c and d show the microstructures corresponding to sion axis in a spiral. In addition, a substructure is observed inside
the undeformed part (head) of a torsion sample soaked at 467 ◦ C for the elongated grains, which is characteristic of dynamic recov-
15 min and torsion tested at 467 ◦ C plus water quenched (Fig. 8c), ery (DRV) during deformation. The microstructure corresponding
or soaked at 467 ◦ C and torsion tested at 278 ◦ C (Fig. 8d). The EBSD to the Al 2024 tested at 467 ◦ C (Fig. 9a) consists of chains of
maps have been colour coded according to the inverse pole figure equiaxed or slightly elongated subgrains developed in the vicini-
ties of initial grain boundaries. The average subgrain size developed
is very heterogeneous, and the fraction of high-angle boundaries,
fHABs , and the mean correlated misorientation, ¯ c , is 44% and 19◦
respectively. These values are relatively low as a consequence of a
microstructure characterized by a high density of strain-induced
LABs.
With the decrease of test temperature to 408 ◦ C (Fig. 9b) and
360 ◦ C (Fig. 9c), the original HABs and the new deformation-
induced low-angle boundaries rotate towards the shear direction,
as a consequence of higher εF (Fig. 2b and Fig. 3). Furthermore, new
finer equiaxed (sub) grains surrounded by high-angle grain bound-
aries are developed inside the initial grains, as a consequence of the
higher stress during the torsion test with respect to 467 ◦ C. There-
fore, since it is widely accepted that the (sub) grain size (L) depends
on the stress () [37] such that L ∝ 1/␴, the mechanical behavior and
the lower (sub) grain size obtained at lower test temperature is the
expected. In addition, while the subgrain size is relatively insensi-
tive to strain and is determined by the stress, the misorientation
across subgrain walls increase with increasing deformation degree
[38]. Accordingly, at 408 ◦ C and 4.5 s−1 (Fig. 9b) the microstruc-

ture is still poorly misoriented with a fHAB = 19% and ¯ c = 10◦ .
Fig. 7. Two-dimensional representation of the 2nd Lyapunov criterion as a function However, at 360 ◦ C (Fig. 9c) the average (sub) grain size is finer
of strain rate and temperature. with respect to higher test temperature due to higher stress, and
 C.M. Cepeda-Jiménez et al. / Materials Science and Engineering A 552 (2012) 530–539 535

Fig. 8. (a) and (b) Backscattered electron micrograph and EBSD map, respectively, showing the microstructure of the as-received Al 2024-T351 alloy. (c) and (d) correspond
to EBSD maps of the un-deformed head of the torsioned samples at 467 and 278 ◦ C, respectively.


fHAB and ¯ c increase to 41% and 19◦ , respectively. It is clearly The corresponding textures as {1 1 1} pole figures are also given
seen in Fig. 9c that the formation of new grains takes place along in Fig. 9. FCC metals like aluminium, with high stacking fault energy,
the original grain boundaries from deformation-induced LABs with present stable torsion textures consisting of a strong {h k l}1 1 0 B
increasing misorientations [39]. fibre and a less strong {1 1 1}u v w A fibre [40] (Fig. 9e), i.e., there
Finally, the torsion sample tested at 278 ◦ C (Fig. 9d) exhibits the is a strong tendency for the slip direction to become aligned with
large initial grains, with an apparent work hardened microstructure the shear direction. In aluminium, the ideal {1 0 0}1 1 0 (C orien-
and poorly developed deformation-induced LABs. Furthermore, the tation) and {1 1 2}1 1 0 (B orientation) components are normally
fHAB for this torsion test condition is drastically reduced to fHAB = 9% positions of high orientation density on the B fibre [41].
and ¯ c = 7◦ which is due to the high amount of low angle bound- In general, the textures corresponding to all temperatures
aries and dislocation structures within the grains. (Fig. 9) show the development of the commonly observed shear

Fig. 9. EBSD maps and corresponding {1 1 1} pole figures of the Al 2024 alloy torsion tested at 4.5 s−1 and different temperatures: (a) 467 ◦ C; (b) 408 ◦ C; (c) 360 ◦ C and (d)
278 ◦ C; (e) main torsion components on the {1 1 1} pole figure.
536 C.M. Cepeda-Jiménez et al. / Materials Science and Engineering A 552 (2012) 530–539

Fig. 10. Backscattered electron micrographs of the Al 2024 alloy torsion tested at 4.5 s−1 and different temperatures: (a) 467 ◦ C; (b) 408 ◦ C; (c) 360 ◦ C and (d) 278 ◦ C.

texture components, i.e., A {1 1 1}1 1 0, B {1 1 2}1 1 0 and C The plastic elongation to failure (εF ) measured at two differ-
{0 0 1}1 1 0. At 467 ◦ C (Fig. 9a) and 278 ◦ C (Fig. 9d), which showed ent strain rates, 4.5 and 25.6 s−1 as a function of temperature is
lower deformation (Fig. 2b), the main texture component is B ori- depicted in Fig. 11a and b, respectively. Also, the efficiency of power
entation ({1 1 2}1 1 0), while at 408 (Fig. 9b) and 360 ◦ C (Fig. 9c) dissipation (J ) and the instability parameter according to the 2nd
the C texture component ({0 0 1}1 1 0) is clearly developed. These Lyapunov criterion have been included in Fig. 11. The intersection
results are in agreement with previous works on torsion tests of region for maximum stability defined by the two Lyapunov criteria
aluminium alloys [42–45], where the A component {1 1 1}1 1 0 (Figs. 6 and 7) together with the maximum efficiency (Fig. 5) and
was found to decrease rapidly with strain, being replaced by the ductility (Fig. 3) should give the best conditions for the forming
C {0 0 1}1 1 0 and the B {1 1 2}1 1 0 components at large strains. process [46].
In these works, the C component was found to be weak at small At low strain rate (4.5 s−1 , Fig. 11a), the ductility shows
strains, and to increase continuously up to strains of about 5, where a clear peak of maximum ductility at about 400 ◦ C. In addi-
the intensity of this component was seen to have values of 9–10 tion, the efficiency increases progressively with temperature,
times random. On the contrary, the B orientation decreased to although above 400 ◦ C this enhancement is more moderate.
a value of 3 times random. On the other hand, also it has been According to Prasad and Srinivasan [7,47], the domains of high
reported [43] a faster decrease in the intensity of the C component efficiency and ductility can be associated with dynamic restora-
and increase of the B component with increasing stress. This also tion mechanisms, namely dynamic recrystallization (DRX) and
can explain the higher intensity of the B component {1 1 2}1 1 0 dynamic recovery (DRV), which improve the workability of the
for the samples tested at lower test temperatures in the present material.
study. As commented, the efficiency variation in Fig. 11a exhibits a
Finally, Fig. 10 shows the corresponding backscattered electron monotonic increase with temperature, even though the ductility is
micrographs to the EBSD maps of Fig. 9. At 408 ◦ C (Fig. 10b) incip- lower at the higher temperature of 467 ◦ C. High efficiency with less
ient precipitation occurs during cooling to test temperature and ductility and high instability at high temperature can be associated
a precipitate distribution fine and homogenous can be observed. with the presence of solid solution. Previous DSC studies on the pre-
The decrease in test temperature to 360 ◦ C (Fig. 10c) and 278 ◦ C cipitation phenomena in the Al 2024 alloy [48–50] demonstrated
(Fig. 10d) leads to large amount of precipitation, especially at the that precipitation occurs by cooling below 430 ◦ C. Accordingly,
grain boundaries. In contrast, at the highest test temperature of we can assume that the following microstructural changes occur
467 ◦ C (Fig. 10a) only the initial large insoluble intermetallic par- during the homogenization treatment and during the hot-torsion
ticles are observed on the aluminium matrix, being the alloying tests: (i) a prolonged homogenization for 30 min at 467 ◦ C, as per-
elements in solid solution. formed in this study, largely removes  (CuAl2 ) and S (CuMgAl2 )
precipitates, bringing copper and magnesium atoms into solid solu-
4. Discussion tion [49]; (ii) the subsequent cooling to torsion test temperatures
together with deformation produces that very fine particles start
In this work, the aeronautical Al 2024 alloy was deformed by to precipitate dynamically at high torsion temperatures (408 and
hot torsion after solution treatment, in a temperature range of 360 ◦ C) from solid solution; (iii) at lower testing temperatures (315
278–467 ◦ C and 2.1–25.6 s−1 strain rate, and processing maps were and 278 ◦ C), a considerable amount of precipitation must occur as
obtained. The microstructures of the torsioned samples were char- a result of the cooling to the test temperature and deformation
acterized by EBSD in order to correlate with the processing maps. (dynamic precipitation) (Fig. 10).
 C.M. Cepeda-Jiménez et al. / Materials Science and Engineering A 552 (2012) 530–539 537

of experimental evidences, it is our contention that the decrease in

ductility at high temperature is due to the presence of solid solution.
On the other hand, generally, the efficiency values associated
with mechanisms such as DRX are about 35–45% [7,47,58] and
20–30% for DRV [7]. The lesser efficiency values for DRV are due
to the less effective energy dissipation through dislocation annihi-
lation compared to DRX in which the migration of interfaces also
contribute to annihilation of a large number of dislocations [10]. In
this study the maximum efficiency value is about 14%, being close
to values reported for DRV.
On the other hand, the deformation conditions of minimum hot
ductility (i.e. at 467 ◦ C and 278–315 ◦ C) in Fig. 11a lie in the maxi-
mum instability regions, taking into account that the most negative
value of the instability parameter corresponds to safer forming
regions. As commented before, the minimum ductility observed
at 467 ◦ C is attributed to the presence of solid solution. However,
at low test temperature the instability increase is mainly due to
the monotonic increment of precipitates with diminishing test-
ing temperature. Lowering the temperature to 278 ◦ C increases
precipitation potential (Fig. 10d), changing dramatically recovery
processes. Microstructure examination of the sample deformed
at 4.5 s−1 and 278 ◦ C (Figs. 9d and 10d) reveals that the soften-
ing observed in the flow stress curves (Fig. 2) at high ε̇ and low
T is due to flow instability, as observed in the processing maps,
Figs. 6, 7 and 11, caused by severe shear flow localization and
dynamic precipitation and coarsening (Fig. 10d). The work hard-
ened microstructure observed at 278 ◦ C (Fig. 9d) is attributed to
the presence of undeformed coarsened precipitates (Fig. 10d), pro-
ducing large local strain gradients around precipitates [59], that
lead to high dislocation density and fine sub-grain size during hot
deformation. Accordingly, this work-hardened microstructure with
Fig. 11. Plastic strain to failure (εF ), efficiency and instability parameter according incipient formation of subgrains showed very low mean misori-
to the 2nd Lyapunov criterion as a function of temperature at two strain rates: (a)
entation (7◦ ). In addition, the energy required to nucleate coarse
4.5 s−1 and (b) 25.6 s−1 .
voids in the grain interior is significantly reduced due to coarse
precipitate formation in the period lasting between the solution
treatment and the torsion testing, together with additional coars-
According to this precipitation sequence, the presence of solid ening by deformation, thus decreasing its ductility (Fig. 3) [60].
solution strengthening at 467 ◦ C adversely affect the climb con- Thus, during the hot forging process, the reinforcements play an
trolled DRV process [51] and shift the optimum deformation important role inducing plastic instability at low temperatures.
condition for DRV to lower temperatures (Fig. 11a). It has been At 25.6 s−1 (Fig. 11b) the ductility is very low and remains
widely reported by different authors that any dislocation moving almost constant with temperature while the efficiency exhibits
through a solid solution will encounter friction drag, raising the considerable increase with temperature. In addition, an increase
energy required for movement [51]. In this way, Charit et al. [52] in instability and a decrease in J is observed at low temperatures
reported that alloying elements in solid solution can diffuse to dis- at this high strain rate. It is our contention that at high strain rate
location cores forming a saturated atmosphere of solute, imposing at all temperatures, strain incompatibility between the aluminium
a drag force. Furthermore, Jin et al. [51] demonstrated that recov- matrix and dynamic coarsened precipitates is a possible source for
ery processes are also hindered by a higher solute atom vacancy void formation due to a strain concentration factor near precipitates
binding energy which effectively reduces the number of vacancies [61,62], diminishing drastically the ductility. Therefore, according
available for dislocation climb whilst the misfit strain effectively to Fig. 11a and b, lower temperatures than 350 ◦ C and high strain
raises the dislocation density, and both processes contributing to rates at any temperature should be avoided in the material pro-
an increase in activation energy for hot deformation. Furthermore, cessing.
it also has been reported that copper (present in the alloy consid- Fig. 12 shows EBSD maps corresponding to the sample deformed
ered in this study) is more effective than magnesium in inhibiting at 408 ◦ C and 2.1 s−1 , having the highest ductility (Fig. 3), high effi-
dynamic recovery [53], taking into account that magnesium has ciency (Fig. 5) and low instability (Figs. 6 and 7). This microstructure
long been known to inhibit dynamic recovery in aluminium alloys shows typical features of dynamic recovery. As commented, high
[54,55]. For Mg, the decrease in dynamic recovery has often been DRV degree is a beneficial process in hot deformation as it gives
associated with a reduction in the stacking fault energy (SFE) [53]. good intrinsic workability by simultaneous softening and develop-
However, although copper increases the SFE [56], when compared ment of a fine and misoriented microstructure [63]. The average
at the same atomic concentration of Mg, it has been found to have grain size of the sample deformed at 408 ◦ C decreases to 1.9 ␮m,
a greater influence on both solid solution strengthening and reduc- and the fHAB and ¯ c are considerably increased to 54% and 23◦ ,
ing the rate of recovery, at least after deformation to conventional respectively. In Fig. 12b, which corresponds to a Kikuchi band con-
strains [57]. In addition, Huang et al. [53] reported that between the trast map, where the high-angle boundaries are painted in black
methods available for limiting grain coarsening, such as the intro- and the low-angle ones are green, it can be seen the very dense
duction of second-phase particles to pin boundary migration, or the HABs network. The high strain experimented at this test condition
use of solute to inhibit recovery, the use of solute is arguably the (ε ∼ 8.2, Fig. 3) leads to the gradual build-up of higher misorien-
most effective approach in aluminium alloys. Therefore, at the sight tation between the neighbouring subgrains. The uniform and fine
538 C.M. Cepeda-Jiménez et al. / Materials Science and Engineering A 552 (2012) 530–539

Fig. 12. EBSD maps and corresponding {1 1 1} pole figure of the Al 2024 alloy torsion tested at 2.1 s−1 and 408 ◦ C. (a) Inverse pole figure (IPF) map which has been colour
coded according to the IPF shown in the inset of the Fig. 8b; (b) Kikuchi band contrast (BC) map with high-angle boundaries (>15◦ ) in black and low-angle boundaries (2–15◦ )
in green. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

grain size thus can be attributed to the occurrence of dynamic 2. The working regime for the forming of the alloy has been estab-
recovery, which leads to the formation of subgrain boundaries lished by determination of maximum efficiency and stability
which progressively transform at large strains into new high angle zones. A forming temperature range of 360–410 ◦ C and strain
grain boundaries. Furthermore, at this temperature (408 ◦ C) based rates of 2.1–4.5 s−1 is recommended having high values of effi-
in DSC results [64] still dissolving alloying elements and incipient ciency (J = 13%). In such conditions of T and ε̇ the material
fine precipitates are present, which also contribute by a pinning exhibits dynamic extended recovery, as revealed by EBSD obser-
effect, together with the high temperature, to this transformation vations.
of LABs into HABs. 3. The decrease of test temperatures to 278–315 ◦ C, especially at
On the other hand, for this test condition the main torsion tex- high strain rates, increases the risk of flow instability, due to flow
ture components (B-{1 1 2}1 1 0 and C-{0 0 1}1 1 0) are clearly localization by dynamic precipitation, coalescence and particle
developed, due to the nature of the crystallographic rotation field coarsening. Coarsened precipitates favour crack nucleation, thus
in simple shear and the high ductility. decreasing the ductility.
Finally, the fitting of the experimental data by the Garofalo 4. The increase of test temperature to 467 ◦ C at all strain rates leads
equation gives an apparent activation energy value of 180 kJ/mol. to limited ductility due to the presence of alloying elements in
This value is higher than that for self-diffusion in pure aluminium solid solution, which hinder dislocation movement and dynamic
(QL = 142 kJ/mol [32]). On the other hand, the nap value is lower than recovery processes.
that consistent with a slip creep mechanism controlled by disloca-
tion climb at constant substructure, n = 8 [31]. Thus, although the Acknowledgements
deformation behaviour during hot forming of the Al 2024 alloy is
consistent with this mechanism, the high activation energy and Financial support from CICYT (Project MAT2009-14452) is
low nap values obtained are attributed to microstructural changes gratefully acknowledged. We also thank F.F. González-Rodríguez
occurring at the various torsion test conditions. Accordingly, at high for assistance with the hot torsion tests.
temperature the flow stress is decreased by precipitate dissolu-
tion while at lower test temperatures the stress is increased by
References
dynamic precipitation, coalescence and particle coarsening, which,
additionally, favour void and crack nucleation. [1] J.C. Malas, S. Venugopal, T. Seshacharyulu, Mater. Sci. Eng. A 368 (2004) 41–47.
In summary, the hot deformation of the Al 2024 alloy estimated [2] C.M. Cepeda-Jiménez, R.C. Alderliesten, O.A. Ruano, F. Carreño, Compos. Sci.
by the efficiency, ductility and instability parameters is determined Technol. 69 (2009) 343–348.
[3] C.M. Cepeda-Jiménez, M. Pozuelo, J.M. Garcia-Infanta, O.A. Ruano, F. Carreño,
by the microstructure evolution during deformation, being opti- Metall. Mater. Trans. A 40A (2009) 69–79.
mum when high dynamic recovery degree is obtained. A forming [4] C.M. Cepeda-Jiménez, P. Hidalgo, M. Pozuelo, O.A. Ruano, F. Carreño, Mater. Sci.
temperature interval of 360–410 ◦ C at 2.1–4.5 s−1 is recommended. Eng. A A527 (2010) 2579–2587.
[5] N.S. Babu, S.B. Tiwari, B.N. Rao, Mater. Sci. Technol. 21 (2005) 976–984.
[6] M. Gavgali, B. Aksakal, Mater. Sci. Eng. A 254 (1998) 189–199.
[7] Y.V.R.K. Prasad, S. Sasidhara, Hot Working Guide: A Compendium on Processing
5. Conclusions Maps, ASM International, Metals Park, OH, 1997.
[8] M. Rajamuthamilselvan, S. Ramanathan, J. Alloys Compd. 509 (2011) 948–952.
[9] M. Mu, Z. Zhi-min, Z. Bao-hong, D. Jin, J. Alloys Compd. 513 (2012) 112–117.
The hot deformation behaviour of the Al 2024 alloy was analyzed [10] N. Srinivasan, Y.V.R.K. Prasad, P.R. Rao, Mater. Sci. Eng. A 476 (2008) 146–156.
by torsion testing, constitutive equations and processing maps. [11] M. Carsí, F. Peñalba, I. Rieiro, F. Zapiráin, O.A. Ruano, Z. Metallkd. 91 (2000)
EBSD was employed to characterize the microstructure and its cor- 1057–1062.
[12] F. Garofalo, Fundamentals of Creep and Creeps Rupture in Metals, MacMillan,
relation with processing maps. The main conclusions of this study New York, 1965.
are as follows: [13] I. Rieiro, M. Carsí, F. Peñalba, Rev. Metal. (Madrid) 32 (1996) 321–328.
[14] I. Rieiro, M. Carsí, O.A. Ruano, Mater. Sci. Technol. 25 (2009) 995–1002.
[15] M. Carsí, F. Peñalba, O.A. Ruano, O.D. Sherby, Metall. Mater. Trans. A 28 (1997)
1. The high temperature data can be predicted by means of a Garo- 1913–1920.
[16] M. Ueki, S. Horie, T. Nakamura, J. Mater. Process. Technol. 11 (1985) 365–376.
falo equation with a stress exponent of 6.1 and an activation [17] Y.V.R.K. Prasad, H.L. Gegel, S.M. Doraivelu, J.C. Malas, J.T. Morgan, K.A. Lark, D.R.
energy for deformation of 180 kJ/mol (being QL = 142 kJ/mol), Barker, Metall. Trans. 15A (1984) 1883–1892.
corresponding to a mechanism of constant-substructure slip [18] H.L. Gegel, J.C. Malas, S.M. Doraivelu, V.A. Shende, Metals Handbook, vol. 14,
ASM International, Metals Park, OH, 1988, pp. 417–438.
creep (n = 8), but experiencing microstructural changes due to
[19] J.M. Alexander, in: J.G. Lenard (Ed.), Modelling of Hot Deformation of Steels,
the effect of precipitates at different temperatures. Springer-Verlag, Berlin, 1989, pp. 105–115.
 C.M. Cepeda-Jiménez et al. / Materials Science and Engineering A 552 (2012) 530–539 539

[20] D. Samantaray, S. Mandal, A.K. Bahduri, Mater. Des. 32 (2011) 716–722. [41] G.R. Canova, U.F. Kocks, J.J. Jonas, Acta Metall. 32 (1984) 211–226.
[21] Y.V.R.K. Prasad, T. Seshacharyulu, Mater. Sci. Eng. A 243 (1998) 82–88. [42] F. Montheillet, M. Cohen, J.J. Jonas, Acta Metall. 32 (1984) 2077–2089.
[22] S.L. Semiatin, V. Seetharaman, I. Weiss, Mater. Sci. Eng. A 263 (1999) 257–271. [43] J.K. Solberg, H.J. McQueen, N. Ryum, E. Nes, Philos. Mag. A 60 (1989) 447–471.
[23] T. Seshacharyulu, S.C. Medeiros, J.T. Morgan, J.C. Malas, W.G. Frazier, Y.V.R.K. [44] H.J. McQueen, J.K. Solberg, N. Ryum, E. Nes, Philos. Mag. A 60 (1989) 473–485.
Prasad, Scr. Mater. 41 (1999) 283–288. [45] T. Pettersen, E. Nes, Metall. Mater. Trans. A 34 (2003) 2727–2736.
[24] R.M. Miller, T.R. Bieler, S.L. Semiatin, Scr. Mater. 40 (1999) 1387–1393. [46] M. Carsí, I. Rieiro, J.A. Jiménez, F. Peñalba, O.A. Ruano, J. Mater. Process. Technol.
[25] T. Seshacharayulu, S.C. Medeiros, J.T. Morgan, J.C. Malas, W.G. Frazier, Y.V.R.K. 143–144 (2003) 416–419.
Prasad, Mater. Sci. Eng. A 279 (2000) 289–299. [47] N. Srinivasan, Y.V.R.K. Prasad, Mater. Sci. Technol. 8 (1992) 206–212.
[26] S.L. Semiatin, T.R. Bieler, Acta Mater. 49 (2001) 3565–3573. [48] B. Verlinden, P. Wouters, H.J. McQueen, E. Aernoudt, L. Delaey, S. Cauwenberg,
[27] B.K. Raghunath, K. Raghukandan, R. Karthikeyan, K. Palanikumar, U.T.S. Pillai, Mater. Sci. Eng. A 123 (1990) 229–237.
R. Ashok Gandhi, J. Alloys Compd. 509 (2011) 4992–4998. [49] C. Badini, F. Marino, E. Verné, Mater. Sci. Eng. A 191 (1995) 185–191.
[28] M. Gang, L. Bolong, L. Hongmei, H. Hui, N. Zuoren, Mater. Sci. Eng. A 517 (2009) [50] P. Cavaliere, J. Light Met. 2 (2002) 247–252.
132–137. [51] N. Jin, H. Zhang, Y. Han, W. Wu, J. Chen, Mater. Charact. 60 (2009) 530–536.
[29] D.S. Fields Jr., W.A. Backofen, Proc. Am. Soc. Test. Mater. 57 (1957) 1259–1272. [52] I. Charit, R.S. Mishra, J. Mater. Res. 19 (2004) 3329–3342.
[30] M. Carsí, R. Allende, F. Peñalba, J.A. Jiménez, O.A. Ruano, Steel Res. Int. 75 (2004) [53] Y. Huang, J.D. Robson, P.B. Prangnell, Acta Mater. 58 (2010) 1643–1657.
26–32. [54] D.A. Hughes, Acta Metall. Mater 41 (1993) 1421–1430.
[31] O.D. Sherby, R.H. Klundt, A.K. Miller, Metal. Trans. A 8 (1977) 843–850. [55] F.J. Humphreys, M. Hatherly, Recrystallization and Related Annealing Phe-
[32] H.J. Frost, M.F. Ashby, Deformation-Mechanism Maps, Pergamon Press, Oxford, nomenon, 2nd ed., Elsevier, 2004.
UK, 1982. [56] T.C. Schulthess, P.E.A. Turchia, A. Gonisa, T.-G. Nieh, Acta Mater. 46 (1998)
[33] D.M. Bates, D.G. Watts, Non Linear Regression Analysis and its Applications, 2215–2221.
John Wiley & Sons, New York, 1986. [57] N.Y. Zolotorevsky, A.N. Solonin, A.Y. Churyumov, V.S. Zolotorevsky, Mater. Sci.
[34] C.E. Campbell, L.A. Bendersky, W.J. Boettinger, R. Ivester, Mater. Sci. Eng. A 430 Eng. A 502 (2009) 111–117.
(2006) 15–26. [58] O. Sivakesavam, I.S. Rao, Y.V.R.K. Prasad, Mater. Sci. Technol. 9 (1993) 805–810.
[35] Y. Xue, H. El Kadiri, M.F. Horstemeyer, J.B. Jordon, H. Weiland, Acta Mater. 55 [59] P.J. Apps, J.R. Bowen, P.B. Prangnell, Acta Mater. 51 (2003) 2811–2822.
(2007) 1975–1984. [60] D. Dumont, A. Deschamps, Y. Brechet, Mater. Sci. Eng. A 356 (2003)
[36] J. Dennis, P.S. Bate, F.J. Humphreys, Acta Mater. 57 (2009) 4539–4547. 326–336.
[37] O.D. Sherby, P.M. Burke, Prog. Mater. Sci 13 (1968) 323–390. [61] P. Cavaliere, E. Evangelista, Compos. Sci. Technol. 66 (2006) 357–362.
[38] M.R. Barnett, F. Montheillet, Acta Mater. 50 (2002) 2285–2296. [62] I. Sen, R.S. Kottada, U. Ramamurty, Mater. Sci. Eng. A 527 (2010) 6157–6165.
[39] N. Dudova, A. Belyakov, T. Sakai, R. Kaibyshev, Acta Mater. 58 (2010) [63] H.Z. Li, H.J. Wang, X.P. Liang, Mater. Sci. Eng. A 528 (2011) 1548–1552.
3624–3632. [64] E. Hersent, J.H. Driver, D. Piot, Scr. Mater. 62 (2010) 455–457.
[40] A. Gholinia, P. Bate, P.B. Prangnell, Acta Mater. 50 (2002) 2121–2136.