Materials Characterization 183 (2022) 111629

Contents lists available at ScienceDirect

Materials Characterization
journal homepage: www.elsevier.com/locate/matchar

Dynamic recrystallization and precipitation behavior of a novel Sc, Zr

alloyed Al-Zn-Mg-Cu alloy during hot deformation
Chenglong Xu a, Jiwu Huang a, b, *, Fuqing Jiang c, Yingge Jiang a, d
a
School of Material Science and Engineering, Central South University, Changsha 410083, PR China
b
Key Laboratory of Nonferrous Metallic Materials Science and Engineering, Ministry of Education, Central South University, Changsha 410083, PR China
c
Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, PR China
d
Hunan Provincial Enterprise Technology Center, Hunan Aerospace Magnetoelectric Co., Ltd., Changsha, 410200, PR China

A R T I C L E I N F O A B S T R A C T

Keywords: The hot deformation behavior of a novel Sc, Zr alloyed Al-Zn-Mg-Cu alloy was explored in the isothermal hot
Al-Zn-Mg-Cu-Sc-Zr alloy compression test with a temperature range of 300–500 ◦ C and a strain rate of 0.001–10 s− 1. The stress-strain
hot compression curve, constitutive equation, and processing map were integrated to evaluate the flow behavior and hot work­
flow behavior
ability of the studied alloy. Besides, the relationship between dynamic recrystallization and precipitation
dynamic recrystallization mechanism
precipitation behavior
behavior and the Zener-Hollomon (Z) parameter during hot deformation was deeply explored. As lnZ decreased,
the dislocation density decreased, while the misorientation angle increased. Meanwhile, with the decrease of lnZ,
the amount of low-angle grain boundaries and medium-low-angle grain boundaries decreased. Moreover, dy­
namic recovery is the dominant dynamic softening mechanism, while three different dynamic recrystallization
(DRX) mechanisms were the main DRX mechanisms at different lnZ values. Discontinuous DRX dominated at
high lnZ value (lnZ ≥33.33); Continuous DRX (CDRX) dominated at middle lnZ value (23.4 < lnZ < 33.33);
Geometric DRX dominated at low lnZ value (lnZ ≤23.44). Meanwhile, increasing lnZ value is conducive to
precipitation behavior. The intragranular precipitates and grain boundary precipitates are identified as the phase
containing Al, Fe, Mn, and Si, the equilibrium η phase containing Cu and Al3(Sc, Zr) particles.

1. Introduction describing the relationship between flow stress and hot deformation
parameters. Besides, the constitutive equation based on lattice diffusion
Al-Zn-Mg-Cu alloy is widely used in transportation, aerospace, key was constructed to describe the hot deformation behavior [9,10]. ANN
components, and other fields due to its lightweight, high strength, high models can learn from examples and recognizes patterns in a series of
toughness, and high fracture toughness [1,2]. With the widespread inputs and outputs. More significantly, it could predict the flow stress
application of Al-Zn-Mg-Cu alloys, the requirements for the alloys are across the deformation mechanism domain, which cannot be achieved
becoming more and more severe, especially the hot workability, which by traditional methods. The Arrhenius equation is used to explore the
directly determines the service performance of the alloy components hot deformation behavior of the experimental alloy during the hot
[3,4]. The addition of trace Sc and Zr has a huge effect on the perfor­ compression process. Combined with hot compression test data, the
mance improvement of Al-Zn-Mg-Cu alloy [5,6]. Even though the hot Arrhenius equation with the Zener-Hollomon parameter (Z) can eval­
deformation parameters (including temperature, hot deformation rate, uate the relationship between deformation temperature, strain rate, and
etc.) of the alloy are changing in actual production, the compression flow stress. Dynamic material model (DMM)-based processing maps are
process at isothermal or equal strain rate can still help to reveal the flow usually established to evaluate the workability of alloys [11], and the
stress behavior and recrystallization mechanism of the alloy. Several processing map can describe efficiency contour lines under different hot
constitutive models have been built to explore the relationship between deformation temperatures and strain rates, which represent power
flow stress and hot deformation parameters, such as phenomenological, dissipation efficiencies during microstructural evolution. In addition,
physically based, and artificial neural network (ANN) [7,8]. The the processing map also reveals microstructural instability according to
Arrhenius model is the most universal phenomenological model the instability value applied to DMM.

* Corresponding author at: School of Material Science and Engineering, Central South University, Changsha 410083, PR China.
E-mail address: huangjw@csu.edu.cn (J. Huang).

https://doi.org/10.1016/j.matchar.2021.111629
Received 6 October 2021; Received in revised form 21 November 2021; Accepted 25 November 2021
Available online 27 November 2021
1044-5803/© 2021 Elsevier Inc. All rights reserved.
C. Xu et al. Materials Characterization 183 (2022) 111629

Table 1
Chemical composition of the novel Sc, Zr alloyed Al-Zn-Mg-Cu alloy alloy (wt%).
Zn Mg Cu Sc Zr Fe Mn Cr Ti Al

5.61 1.87 0.29 0.086 0.087 0.12 0.315 0.10 0.042 Bal.

dislocation-inducing accelerated nucleation; (b) the dislocation-

inducing accelerated growth; (c)accelerated nucleation and growth
caused by non-equilibrium vacancy concentration; and (d)the dissolu­
tion caused by dislocation shearing. Meanwhile, different precipitation
behaviors are related to different Z values in Al-Zn-Mg-Cu alloy.
In this paper, the Arrhenius equation of the novel Sc, Zr alloyed Al-
Zn-Mg-Cu alloy with high-strength was established and the processing
map was constructed to discuss the hot deformation behavior in-depth,
which are helpful to improve the hot workability of the experimental
alloy. In spite that numerous studies have studied the relationship be­
tween Z and the softening mechanism in Al-Zn-Mg-Cu alloys, the effect
of Z on the DRX mechanism has rarely been studied. Based on the
electron backscatter diffraction (EBSD) images, inverse pole figures
(IPFs), and transmission electron microscopy (TEM) images, the DRX
mechanism has been explored. In addition, the relationship between the
dominant DRX mechanism and lnZ, and the formation process of
different DRX mechanisms were investigated. The hot deformation
behavior of the alloy was explained from the perspective of precipitation
behavior. Therefore, the research has guiding significance for the actual
production process of high-strength aluminum alloy.
Fig. 1. Schematic diagram of hot compression experiment process.
2. Experimental procedures

It is well known that the coupled effect of temperature and strain rate
A semi-continuous ingot of Al-Zn-Mg-Cu alloy was utilized as raw
on hot deformation behavior can be reflected by the Z [12] and the Z
materials in this study, and its chemical composition was displayed in
value can also determine the dynamic softening mechanism of the alloy
Table 1. As shown in Fig. 1, cylindrical specimens were cut from the
[13,14]. Although dynamic recovery (DRV) is usually the dominant
initial billets using wire cutting technology and their diameter and
dynamic softening mechanism due to the high-stacking fault energy in
height were 10 mm and 15 mm, respectively. These cylindrical samples
aluminum alloy, dynamic recrystallization (DRX) sometimes occurs,
were first homogenized at 470 ◦ C for 24 h, then air-cooled to room
such as discontinuous dynamic recrystallization (DDRX) [15,16],
temperature. Subsequently, the deformation temperature from 300 ◦ C to
continuous dynamic recrystallization (CDRX) [17,18], and geometric
500 ◦ C(50 ◦ C interval)and the strain rate ranged from 0.001 to 10 s− 1(10
dynamic recrystallization (GDRX). Under different hot deformation
times interval) were selected to simulate the hot deformation utilizing a
conditions, a specific DRX mechanism dominates. DDRX exists in Al-Mg-
Gleeble 3500-GTC thermo-mechanical testing system, and the designed
Mn alloy and it would be suppressed by high-density fine particles and
true strain was 1.2. Samples were preheated to the hot deformation
precipitates [17,19]. At the bulging of the grain boundary, the DDRX
temperature at a heating rate of 10 ◦ C/s and kept at the temperature for
grains nucleate and grow through the migration of the grain boundaries
3 min to ensure the uniform structure. To reduce the friction between
(GBs) [20]. It is more commonly observed that CDRX occurs in the su­
the press head and the sample during the hot deformation process, it is
perplastic tensile test and friction stir processing of Al-Zn-Mg-Cu alloy
imperative to coat the contact surface with graphite lubricant. During
[21,22]. CDRX is a process in which subgrains rotation and growth re­
the hot deformation process, specimens are compressed to a specified
sults in the transformation from low-angle grain boundaries (LAGBs:
strain at a constant strain rate, and the flow stress is accurately
from 2◦ to 10◦ ) into medium-low-angle grain boundaries (MLGBs: from
measured. Upon completing the hot compression process, the sample
10◦ to 15◦ ) and high-angle grain boundaries (HAGBs: above 15◦ )
was quenched with water immediately to maintain its specific
[23,24]. In addition, numerous studies have proposed that GDRX occurs
microstructure.
in the hot deformation process of aluminum alloys [25,26]. In fact,
Samples for microscopic characterization were taken from the mid­
dislocations and subgrain boundaries (SGBs) would be involved in any
dle cross-section of the processed cylindrical specimens (Fig. 1). X-ray
kind of DRX mechanism. Increasing temperature and decreasing strain
diffraction (XRD), EBSD, and TEM were used to characterize the
rate can promote dislocation annihilation and rearrangement of dislo­
microstructure of hot compressed samples. The XRD measurement was
cations into cellular structures, resulting in a decrease in dislocation
obtained using a Rigaku, SmartLab instrument (40 KV, 30 mA), with a
density [27]. Although there are obvious differences among the three
scanning range from 30◦ to 140◦ and a scanning rate of 2◦ /min. The
mechanisms, it is difficult to distinguish the three in practice.
dislocation density was calculated with the XRD result according to the
In the Al-Zn-Mg-Cu alloy system, precipitation strengthening is one
previous investigation [33]. Additionally, the specimens for EBSD and
of the main contributors to high strength, and the general precipitation
TEM observation were obtained by electropolishing in an electrolyte of
order in the artificial aging is GP zone→ η′ phase →η phase (MgZn2)
25% nitric acid and 75% methanol solution at 10–15 V and − 25 ◦ C, and
[28–30]. As above mentioned, the hot deformation behavior of alloy is
their thickness were approximately 120 μm and 80 μm, respectively. A
confined by the precipitates and second phase particles. Zhang et al.
scanning electron microscope (FEI SIRION 200) equipped with an EBSD
found that high-pressure torsion promoted the precipitation of η-phase
system was utilized to obtain the EBSD results at 25 kV. OIM.7 software
[31]. Besides, the η-phase precipitation was synchronized with the strain
was used for analysis. TEM observation, high-angle annular dark-field
and was also affected by the hot deformation parameters. The compli­
scanning transmission electron microscopy (HAADF-STEM) observa­
cated balance between the four kinds following precipitation mecha­
tions and related energy-dispersive X-ray spectroscopy (EDS) mappings
nisms may be affected by hot deformation parameters [32]: (a)the
were carried out on an FEI Tecnai F20 transmission electron microscope

2
C. Xu et al. Materials Characterization 183 (2022) 111629

Fig. 2. The true stress-strain curves of the experimental alloy at 300–500 ◦ C: (a)0.001 s− 1; (b)0.01 s− 1; (c)0.1 s− 1; (d)1 s− 1; (e)10 s− 1.

operating at 200 kV. samples at higher temperature ranges (400–500 ◦ C) remains the above
law, but the stress at lower temperature ranges (<400 ◦ C) first increases
3. Results and discussion rapidly to the approximate equilibrium value and maintain for a period
of time, and then rapidly decreases (Fig. 2(c)). At 1 s− 1, the stress first
3.1. Flow behavior increases rapidly, slowly increases to the peak value, and then decreases
rapidly (Fig. 2(d)). As shown in Fig. 2(e), the stress first rapidly increases
3.1.1. The true stress-strain behavior to the peak value, decreases slowly, and then decreases rapidly at 10 s− 1.
The evolution in flow stress results from the competition between In the primary stage of hot compression, the internal stress of the
work hardening and dynamic softening. Moreover, an increasing tem­ alloy increases rapidly with the increasing strain as a result of the work
perature or decreasing strain rate can boost the dynamic softening hardening mechanism. As the softening effect becomes more significant,
process including DRV and DRX. True stress-true strain curves are dis­ and the stress decreases slowly or maintains at the maximum value after
played in Fig. 2. When strain rates are 0.001 s− 1 and 0.01 s− 1, the stress reaching the peak. Under the condition of low strain rate (0.001 s− 1), the
first increases rapidly, then slowly decreases to the equilibrium value stress decreases slowly after reaching the peak, indicating that the
(Fig. 2(a)-(b)). While the strain rate is 0.1 s− 1, the stress-strain curve of softening effect can completely offset the work hardening. Theoretically,

3
C. Xu et al. Materials Characterization 183 (2022) 111629

1
Fig. 3. Fitted lines between (a) lnσp- Inε̇, (b) σp- lnε̇, (c) ln[sinh(ασp)]- lnε̇, and (d) ln[sinh(ασp)]-1000 T− at different strain rates.

Table 2
lnZ values of the alloy under different deformation conditions.
1
ε̇/s− 10 1 0.1 0.01 0.001
T/◦ C

300 37.94 35.64 33.33 31.03 28.73

350 35.08 32.78 30.48 28.17 25.87
400 32.64 30.34 28.04 25.74 23.44
450 30.55 28.24 25.94 23.64 21.34
500 28.72 26.42 24.12 21.81 19.51

Table 3
Material parameters of the experimental alloy during hot compression.
1 1 1
n1 β n α/10− MPa− Q/KJ⋅mol− A/1013 s− 1

7.08 0.11 5.17 0.16 169.81 1.11

the dynamic softening effect is predicted to be inversely proportional to

the strain rate due to longer dynamic softening effect time at a lower
strain rate. When the strain rate is 0.1 s− 1, the stress remains at an
equilibrium value after reaching the peak value, and the two mecha­
nisms are roughly in dynamic equilibrium. Nevertheless, the internal Fig. 4. The DMM processing map of the alloy at the true strain of 1.2.
stress of the alloy increases rapidly and then slowly at a high strain rate,
indicating that the dynamic softening mechanism is not enough to selected to evaluate the relationship between alloy deformation tem­
completely offset the work hardening effect. perature T(K), strain rate ε̇(s− 1), and flow stress σ(MPa) [34]:

3.1.2. Constitutive modeling with Z parameters Z = ε̇exp( − Q/RT) (1)

The Arrhenius equation with Zener-Hollomon parameters was
ε̇ = AF(σ)exp( − Q/RT) (2)

4
C. Xu et al. Materials Characterization 183 (2022) 111629

Fig. 5. Comparison chart between predicted flow stress (hollow symbols) and experimental data (solid lines) at different strain rates: (a) 0.001 s− 1; (b)0.01 s− 1; (c)
0.1 s− 1; (d)1 s− 1; (e)10s− 1.

where F(σ) is a function related to flow stress and its form changes constant (8.314 J⋅mol− 1⋅K− 1) and Q is the deformation activation en­
with the stress evolution. Moreover, its specific expression form is as ergy (KJ⋅mol− 1). Eqs. (3) and (4) are used to describe the flow stress
follows [35]: behavior at low stress and high stress respectively, yet Eq. (5) applies to
⎧ ( )
⎪ − Q both situations. Notably, the definitive of the material constant α is α =
⎪
⎪
⎪
⎪ A 1 σ n1
exp , if ασ < 0.8 (3) β/n1.Taking natural logarithm on both sides of Eqs. (3), (4), and (5),
⎪
⎪ RT
⎪
⎪
⎨ ( ) they can be denoted as follows:
− Q ⎧
ε̇= A2 exp(βσ)exp , if ασ > 1.2 (4) ⎪ Q
⎪
⎪
⎪
RT ⎪
⎪
⎪ lnA1 + n1 lnσ − , if ασ < 0.8 (6)
⎪
⎪ ( ) ⎪
⎪ RT
⎪ ⎪
⎨
⎪ A[sinh(ασ ) ]n exp − Q , for values of all ασ (5).
⎪ Q
⎪
⎩
RT lnε̇= lnA2 + βσ − , if ασ > 1.2 (7)
⎪
⎪ RT
⎪
Where A1, A2, A, n1, n, α, β are material constants, R is the gas ⎪
⎪
⎪ Q
⎪
⎩ lnA + nln[sinh(ασ ) ] − , for values of all ασ (8)
RT
5
C. Xu et al. Materials Characterization 183 (2022) 111629

Fig. 6. TEM images of dislocation distribution under different hot deformation conditions: (a) 300 ◦ C, 0. 1 s− 1(lnZ = 33.33); (b) 400 ◦ C, 10s− 1(lnZ = 32.64); (c)
350 ◦ C, 0.1 s− 1(lnZ = 30.48); (d) 400 ◦ C, 1 s− 1(lnZ = 30.34); (e) 400 ◦ C, 0.1 s− 1(lnZ = 28.04); (f) 450 ◦ C, 0.1 s− 1(lnZ = 25.94)); (g) 400 ◦ C, 0.01 s− 1(lnZ = 25.74); (h)
500 ◦ C, 0.1 s− 1(lnZ = 24.12); (i) 400 ◦ C, 0.001 s− 1(lnZ = 23.44).

Fig. 7. Evolution of grain boundaries and dislocations of the experimental alloy during hot deformation: (a)fractions of grain boundaries with various misorientation
angles and dislocation density at different lnZ values; (b)distribution map of misorientation angles at different lnZ values (θsub and θav denote the misorientation
angle of subgrains and grains, respectively).

6
C. Xu et al. Materials Characterization 183 (2022) 111629

Fig. 8. EBSD images of samples with different hot deformation parameters: (a)300 ◦ C, 0.1 s− 1(lnZ = 33.33); (b)400 ◦ C, 0.1 s− 1(lnZ = 28.04); (c)450 ◦ C, 0.1 s− 1(lnZ
= 25.94); (d)400 ◦ C, 0.001 s− 1(lnZ = 23.44).

To figure out the material parameters, the relationship curve be­ where S represents the slope of ln[sinh(ασp)]/1000/T at various
tween lnσp (σp represents the peak stress), σp, ln[sinh(ασp)], lnε̇ and strain rates and its average value is 3.951(Fig. 3(d)). Subsequently, the
1000/T is depicted in Fig. 3(a)-(d) based on Eqs. (6)–(8). Given that the deformation activation energy of the alloy can be calculated by Eq. (11),
material parameters meet the following definitions: and its average value is 169.81 KJ⋅mol− 1. The following formula can be
obtained as follows with natural logarithm processing on both sides of
dlnε̇
n1 = (9) Eqs. (1) and (2):
dlnσ
lnZ = lnε̇ + Q/RT (13)
dlnε̇
β= (10) [ ( )]
dσ lnZ = lnA + nln sinh ασ p (14)
⃒
∂lnε̇ ⃒ where lnA and n are respectively the intercept and slope of the curve
n= ⃒ (11)
∂ln[sinh(ασ) ] ⃒T between lnZ and In[sinh(ασ)]. By substituting the values of Q, ε̇, and T
into Eq. (13), the value of lnZ can be obtained in Table 2. In addition, the
The values of n1, β, and n are 7.08, 0.11, and 5.17, respectively and
value of A is 1.110 × 1013 s− 1. The material constants of the alloy are
equal to the reciprocal slope of Fig. 3 (a)-(c). Furthermore, the material
exhibited in Table 3. The constitutive equation of the experimental alloy
parameter α can be calculated as 0.016 through the definition α = β/n1.
is constructed as follows:
In addition, Q can be written as follows:
{ } { [ ( )]} ε̇ = 1.110 × 1013 [sinh(0.0160σ) ]5.170 exp(16981/RT) (15)
∂lnε̇ ∂ln sinh ασp
Q = 1000R [ ( ) ] T⋅ = 1000RnS (12)
∂ln sinh ασp ∂(1000/T) As an effective method to obtain the best processing parameters, the
hot processing map of the experimental alloy is shown in Fig. 4. In the

7
C. Xu et al. Materials Characterization 183 (2022) 111629

Fig. 9. IPFs of samples with different hot deformation parameters: (a)300 ◦ C, 0.1 s− 1(lnZ = 33.33); (b) 400 ◦ C, 0.1 s− 1(lnZ = 28.04); (c) 500 ◦ C, 0.1 s− 1(lnZ = 24.12);
(d) 400 ◦ C, 0.001 s− 1(lnZ = 23.44).

hot processing map, the red shaded section represents the unstable small amount of dislocations are discretely distributed inside grain. The
working area, and the contour line represents the percentage of the existence of substructures also affects the distribution of dislocations.
power dissipation efficiency under different hot deformation parame­ When fine and dispersed particles are present, such as Al3(Sc, Zr)
ters. The unstable area is large and mainly concentrated in the high dispersed particles, the resistance for the dynamic recrystallization
strain rate area, and the power dissipation factor increases with the in­ process would increase (Fig. 6(g)). It has been proved that in the
crease of temperature. Based on the processing map, the optimum hot experimental alloy, Al3(Sc, Zr) dispersed particles can inhibit recrys­
deformation conditions are 400–500 ◦ C and 0.01–0.32 s− 1. Fig. 5 tallization by hindering the migration of dislocations and GBs [36,37]
manifests the comparison curve between the experimental value and the (Fig. 6(h)). Moreover, GBs, SGBs, and dislocation walls can also hinder
predicted value of the flow stress under different hot deformation tem­ the recrystallization process. As the lnZ value decreases to 23.44, dis­
peratures and strain rates. The constitutive equation with the strain locations are almost invisible (Fig. 6(i)).
compensation can predict the flow stress of the alloy very well, except Based on the result of XRD, the law is obtained that the dislocation
for the low temperature (300 ◦ C) and high strain rate conditions (10 density is positively correlated with lnZ (Fig. 7(a)). The strong thermal
s− 1). In fact, the predicted curve based on the constitutive equation does activation effect and long dislocation recovery time (low lnZ value)
not fit the experimental curve well at low temperature or high strain rate cause the dislocation to overcome the pinning effect and to achieve
(Fig. 5), which results from the uneven internal stress distribution due to cross-slipping and climbing, leading to low dislocation density and flow
the poor thermal activation effect in these cases. stress. As a result, the dislocations are absorbed and transformed into
LAGBs, which is also a major process in CDRX. However, the activation
3.2. Microstructural evolution energy of LAGBs is usually low and the slip speed is usually slow, thus it
would be difficult to significantly reduce the dislocation density (stress)
3.2.1. Effect of lnZ on dislocation density in the CDRX process. The high lnZ value corresponds to the DDRX
Fig. 6 manifests the TEM images of the alloy under different hot process in which a weak thermal activation effect makes the dislocations
deformation conditions. At high lnZ values, due to poor thermal acti­ difficult to get rid of the pinning effect. Therefore, dislocations incline to
vation and recovery conditions, the dislocation density is relatively high accumulate and multiply in the substructure and produce greater stress.
(Fig. 6(a)-(b)), and dislocation entanglement occurs (Fig. 6(c)). As can In addition, dislocations are absorbed near HAGBs, causing subgrains to
be seen from Fig. 6(a), numerous dislocations result in the formation of spin and grow. Considering that the activation energy and the fast slip
the dislocation wall. In addition, the high lnZ value will also bring about speed of HAGBs, and the number of dislocations accumulated in HAGBs
the heterogeneous substructure, and dislocations are not uniformly are large, a large number of dislocations are consumed in the DDRX
distributed so that a driving force is formed to promote DRV and DRX process, and the flow stress drops significantly.
processes. From Fig. 6(d)-(f), it can be seen that the dislocation density According to the improved model describing dislocation evolution
decreases, and the area of recovery also increases, indicating a stronger [38], the following equation is established:
recovery effect at the medium lnZ values. When the lnZ value is 25.94, a

8
C. Xu et al. Materials Characterization 183 (2022) 111629

Fig. 10. Cumulative misorientation angles along the lines of recrystallized grains remarked in Fig. 8: (a-d) lines 1–3 refer to black lines 1–3 in Fig. 8(a-d),
respectively.

dρ √̅̅̅ dislocations are transformed into LAGBs, MLGBs, or even HAGBs

= k1 ρ − k2 ρ(Z)− m
− k3 ρSre νε̇− 1
(16)
dε through subgrain rotation and growth. A higher lnZ value introduces
where, k1, k2, k3 are material constants, Sre and ν are the area and more dislocations accumulated in HAGBs, and the energy gradient on
speed of moving grain boundaries. As shown in Fig. 2, when the true both sides of the grain boundary is extremely large, resulting in an
increased proportion of LAGBs. Due to the larger storage energy of
stress reaches the dynamic equilibrium (ddρε=0), the relationship between
MLGBs and HAGBs, they are seldom formed when the lnZ is high.
the dislocation density and lnZ is obtained as follows:
Therefore, their proportion would decrease.
√̅̅̅
k1 ρ − k2 ρ(Z)− m − k3 ρSre νε̇− 1 = 0 (17) The formation process of recrystallized grains is inseparable from the
evolution of grain misorientation angle during hot deformation. The
3.2.2. Effect of lnZ on (sub-)grain boundaries distribution map of misorientation angles at different lnZ values is
The evolution of the grain boundary with lnZ with different misori­ manifested in Fig. 7(b). After the reduction of 90% at 300 ◦ C and 0.1 s− 1,
entation angles is shown in Fig. 7(a). As the value of lnZ decreases (that the average grain boundary orientation angle and the misorientation
is, the temperature increases or the strain rate decreases), the proportion angle of the SGBs are 12.39◦ and 3.65◦ , respectively. When the tem­
of LAGBs decreases, that of MLGBs increases slightly, and that of the perature is 500 ◦ C, those two values increase to 15.69◦ and 3.85◦
HAGBs increases. The proportion of HAGBs has never exceeded 50%, respectively. Obviously, the average grain boundary orientation angle
indicating that DRV is the main softening mechanism. The EBSD dia­ and the misorientation angle of the SGBs also increases as the lnZ value
grams under different hot deformation conditions(lnZ) are shown in decreases. Fig. 9 shows the evolution of the micro-texture under
Fig. 8. Under the hot deformation conditions of 300 ◦ C, 0.1 s− 1, LAGBs, different hot deformation conditions. At 300 ◦ C and 0.1 s− 1, the grain
MLGBs and HAGBs account for 76.16%, 2.22%, and 21.62% (Fig. 8), orientation is generally concentrated in <111> (Fig. 9(a)). As the lnZ
respectively. As the temperature increases, a significant decrease in value gradually decreases, the grain orientation becomes dispersed and
LAGBs and an increase in HAGBs can be clearly observed. In addition, evolves into the <001> direction (Fig. 9 (b)-(c)). At 400 ◦ C and 0.001
LAGBs, MLGBs and HAGBs at 400 ◦ C and 0.001 s− 1account for 52.41%, s− 1, the grain orientation is mainly concentrated in <001> and < 111>
5.45%, and 42.14%, respectively. (Fig. 9(d)). Common orientations include deformation texture orienta­
Subgrain boundaries become intermediate substructures between tion and recrystallization texture orientation. The <111> and <001>
dislocations and recrystallized grains in the DRX process. GB bulging is a belong to deformation texture and recrystallization texture respectively.
potential location for DDRX nucleation due to a high local orientation In the case of low lnZ, part of the <111> orientation transforms to the
gradient or strain gradient [39]. In the CDRX process, numerous <001> orientation, indicating that the occurrence of the

9
C. Xu et al. Materials Characterization 183 (2022) 111629

Fig. 11. Schematic diagrams of different DRX mechanisms: (a1-a4) DDRX; (b1-b4) CDRX; (c1-c4) GDRX; (d) the meaning of the legends.

Fig. 12. Distribution of recrystallized grains under different hot compression conditions: (a) 300 ◦ C, 0.1 s− 1, lnZ = 33.33; (b) 400 ◦ C, 0.1 s− 1, lnZ = 28.04; (c) 400 ◦ C,
0.001 s− 1, lnZ = 23.44.

recrystallization process induces an increase in grain orientation. center of the deformed grain, pass through the recrystallized grain, end
at the grain boundary, and all grains are randomly selected). As shown
3.2.3. Effect of lnZ on DRX mechanism in Fig. 10 (a), when the temperature is 300 ◦ C and the strain rate is 0.1
At a high lnZ value (≥33.33), the grains show typical deformation s− 1, only HAGBs and LAGBs are present, while MLGBs are invisible. The
characteristics, and a large number of SGBs are observed, indicating that dominant DRX mechanism is a process that LAGBs are directly trans­
DRV is the dominant DS mechanism at a high lnZ value (Fig. 8(a)). As formed into HAGBs rather than MLGBs. Conclusively, the DDRX process
shown in Fig. 8(a), the small DRX grains are distributed along the is the main DRX mechanism at a high lnZ value.
original grain direction because the shear bands (marked by the black Fig. 8(b)-(c) shows the EBSD images of samples with moderate lnZ
rectangles A and B) are favorable sites for DDRX nucleation under the values (23.44 < lnZ < 33.33). Most of the subgrain and SGBs are visible,
condition of high dislocation density. The area selected by the black but there are only a few recrystallized grains, indicating that DRV is still
rectangle A is enlarged as shown in the inset in Fig. 8(a). Indeed, the main softening mechanism. Evidently, LAGBs are greatly reduced at
numerous fine recrystallization grains gather along the shear band. The 450 ◦ C compared with that at 400 ◦ C, which indicates that a lower lnZ
misorientation angle distribution along with the black arrows 1, 2, and 3 value can promote DRV and DRX processes. As shown in Fig. 10(b)–(c),
in Fig. 8 is shown in Fig. 10 (The black arrows in Fig. 8 start from the the trend of line 1 and line 3 is the transition from LAGBs into MLGBs

10
C. Xu et al. Materials Characterization 183 (2022) 111629

Fig. 13. TEM images of Al-Zn-Mg-Sc-Zr alloy under different hot compression conditions: (a) 300 ◦ C, 0.1 s− 1, lnZ = 33.33(DDRX); (b) 400 ◦ C, 0.1 s− 1, lnZ = 28.04
(CDRX); (c) 400 ◦ C, 0.001 s− 1, lnZ = 23.44(GDRX).

and finally into HAGBs. The presence of MLGBs means that the recrys­ boundary misorientation angle of the brick-like subgrain within the
tallized grains are formed through the subgrain rotation and growth region I is greater than 15◦ , and the other one is less than 15◦ . However,
mechanism. Therefore, CDRX is the dominant recrystallization mecha­ all grain boundary orientation angles of brick-like sub-grains in Region
nism at the middle lnZ value. Owing to the dependence of the CDRX II are greater than 15◦ . At the same time, CDRX grains are not absent at a
process on hot deformation conditions, the tendency of LAGBs and low lnZ value, indicating that other DRX mechanisms also exist. When
MLGBs to transform into HAGBs becomes more obvious with decreasing the temperature is 400 ◦ C and the strain rate is 0.001 s− 1, adjacent grains
the value of lnZ. The transformation from subgrain into recrystallized show similar misorientation angles (Fig. 10(d)), which proves the
grains reduces LAGBs and MLGBs, yet increases HAGBs. In brief, the presence of brick-like subgrain. Based on the above analysis, the
continuous rotation of subgrains is an obvious feature of CDRX, and recrystallized grains are formed by GDRX.
MLABs are the vital symbols for CDRX as well. Besides, the proportion of The process of forming new grains in the CDRX process is shown in
MLGBs under medium lnZ conditions is higher than that under high lnZ Fig. 11(b). A higher lnZ value means a stronger thermal activation effect,
conditions (Fig. 7(a)), indicating that medium lnZ conditions are suit­ which makes the dislocations uniformly distributed and difficult to
able for the occurrence of the CDRX process. produce GB bulging. Under the action of DRV, dislocation walls are
As shown in Fig. 8(d), both greatly-reduced LAGBs and greatly formed (Fig. 11(b1)). Subsequently, dislocations are further consumed
increased HAGBs can be clearly observed at a low lnZ value (≤23.44). and transformed into LAGBs, which are transformed into higher-angle
DRV is the main dynamic softening mechanism, yet several DRX pro­ SGBs through the subgrain rotation and growth mechanism [41]
cesses occur. A large number of brick-like sub-grains can be observed (Fig. 11(b2)-(b3)). At the same time, the grains grow, and the CDRX
(Regions I and II), indicating that GDRX is the dominant DRX mecha­ grains are finally formed through the subgrain rotation and growth
nism at a low lnZ value. The GDRX process can be understood as the mechanism (Fig. 11(b4)). In Fig. 12(b), the grains with similar size are
formation of brick-like SGBs between adjacent parallel HAGBs, predicted to be CDRX grains and the CDRX process can be clearly
including two different intermediate states (Fig. 8(d)). One grain observed in Fig. 13(b). After dislocations are polygonized to form a

11
C. Xu et al. Materials Characterization 183 (2022) 111629

Fig. 14. HAADF-STEM images of the experimental alloy: (a)300 ◦ C, 0.1 s− 1(lnZ = 33.33); (b)400 ◦ C, 0.1 s− 1(lnZ = 28.04); (c)450 ◦ C, 0.1 s− 1(lnZ = 25.94); (d)
400 ◦ C, 0.001 s− 1(lnZ = 23.44).The insert shows XRD patterns of the above four samples.

dislocation wall, then evolving into a dislocation cell (Fig. 13(b1)), the when the jagged HAGBs are close to each. Based on the above descrip­
dislocation cell would be transformed into a subgrain (Fig. 13(b2)). In tion, several brick-like grains with similar orientations are predicted to
addition, the existence of the triple junction of the GBs indicates that the be GDRX grains (Fig. 12(c)). The GDRX process can also be observed in
CDRX process is completed (Fig. 13(b3)). These arguments demonstrate TEM images of samples with low lnZ values (400 ◦ C,0.001 s− 1). It can be
that the dominant recrystallization mechanism at the middle lnZ value is seen from Fig. 13(c1) that the dislocations are arranged in parallel lines
CDRX. to form a geometrically necessary boundary, which is also a prerequisite
The GDRX mechanism is shown in Fig. 11(c). In the process of hot for the occurrence of the GDRX process. This type of boundary is
deformation, the original elongated grains are divided by dislocations conducive to the construction of a three-dimensional network of LAGBs,
into several brick-like dislocation cells with similar misorientation then gradually evolves into HAGBs, which can accelerate the recrys­
orientation angles (Fig. 11(c1)- (c2)). Afterward, the dislocation cells tallization process. In Fig. 13(c2), the elongated brick-like subgrains are
evolved into subgrains of similar size (Fig. 11(c3)). Later, brick-like clearly visible and the orientation of the two SGBs is almost the same.
subgrains are transformed into brick-like grains by absorbing disloca­ Finally, the subgrains are transformed into brick-like GDRX grains by
tions [42,43] (Fig. 11(c3)). Some brick-like subgrains absorb disloca­ absorbing dislocations (Fig. 13(c3)).
tions and are transformed into brick-like recrystallized grains, while
other SGBs intersected each other and evolved into bamboo-like grains 3.2.4. Effect of lnZ on precipitation behavior
(equivalent to the combination of several brick-like grains). Besides, the The precipitates distributions of hot compressed samples with
lateral SGBs would force the original GBs to become jagged and close to different lnZ values are shown in Fig. 14. The number of precipitates in
each other. At the same time, the lateral SGBs are not transformed into the sample at 300 ◦ C and 0.1 s− 1 is far more than that of other samples
HAGBs, but the jagged SGBs are forced to move along the opposite di­ (can be proved in the inset of Fig. 14(a)) due to the strong driving force
rection, resulting in forming coarsening recrystallized grains [44]. of at low temperatures. As the temperature increases, many precipitates
Finally, original grains will be pitched off to form new GDRX grains dissolve into the matrix, and precipitates decrease. Moreover, the

12
C. Xu et al. Materials Characterization 183 (2022) 111629

Fig. 15. Elements EDS mapping analysis: (a)-(b)corresponding to Fig. 14(a), and (c) corresponding to Fig. 14(d).

number of precipitates in the sample at 400 ◦ C and 0.001 s− 1 is signif­ remarkable. As the lnZ value decreases, the number of precipitates is
icantly less than that in the sample at 400 ◦ C and 0.1 s− 1. The lower significantly reduced due to the low dislocation density and poor driving
strain rate means fewer dislocations and fewer favorable nucleation force for precipitation, indicating that the dissolution of precipitates is
sites, inhibiting the heterogeneous nucleation of precipitation. A higher gradually dominant.
lnZ value can promote precipitation behavior during the hot compres­
sion process. Parts of Fig. 14(a) and (d) are selected for elements EDS 4. Conclusions
mapping analysis, and the results are shown in Fig. 15, which shows that
the intragranular precipitates (Fig. 15 (a)-(b)) and grain boundary pre­ • Based on the DDM processing map, the optimum hot deformation
cipitates (Fig. 15 (c)) are identified as the phase containing Al, Fe, Mn conditions are 400–500 ◦ C and 0.01–0.32 s− 1. High temperature and
and Si, the equilibrium η phase containing Cu and Al3(Sc, Zr) second low strain rate mean low flow stress. The constitutive equation of the
phase particles. The number of precipitates is extremely great in the experimental alloy is constructed as follows: ε̇=1.1101013[sinh
sample of 300 ◦ C and 0.1 s− 1 due to the presence of high-density dis­ (0.0160σ)]5.170exp(16,981/RT).
locations and the strong heat-induced diffusion capacity. The formation • As lnZ decreases, the dislocation density decreases, and the misori­
of numerous η phases (Fig. 14(a)) indicates that the effect of dislocations entation angle increases. Besides, LAGBs, and MLGBs decrease with
and vacancies on accelerating the precipitates nucleation and growth is the decreasing lnZ, while HAGBs increase.

13
C. Xu et al. Materials Characterization 183 (2022) 111629

• DRV is the dominant dynamic softening mechanism, while three [17] M.R. Rokni, A. Zarei-Hanzaki, A.A. Roostaei, H.R. Abedi, An investigation into the
hot deformation characteristics of 7075 aluminum alloy, Mater. Des. 32 (4) (2011)
different DRX mechanisms are the main DRX mechanisms at
2339–2344.
different lnZ values. Moreover, DDRX dominates at a high lnZ value [18] Y. Li, B. Gu, S. Jiang, Y. Liu, J. Lin, A CDRX-based material model for hot
(≥33.33), CDRX dominates at the middle lnZ value(23.4 < lnZ < deformation of aluminium alloys, Int. J. Plast. 102844 (2020).
33.33), and GDRX dominates at a low lnZ value (≤23.44). [19] F.R. Castro-Fernández, C.M. Sellars, Static recrystallisation and recrystallisation
during hot deformation of Al–1Mg–1Mn alloy, Mater. Sci. Technol. (United
• A high lnZ value is conducive to precipitation behavior. The intra­ Kingdom) 4 (7) (1988) 621–627.
granular precipitates and grain boundary precipitates are identified [20] Y.C. Lin, X.-Y. Wu, X.-M. Chen, J. Chen, D.-X. Wen, J.-L. Zhang, L.-T. Li, EBSD
as the phase containing Al, Fe, Mn, and Si, the equilibrium η phase study of a hot deformed nickel-based superalloy, J. Alloys Compd. 640 (2015)
101–113.
containing Cu and Al3(Sc, Zr) particles. [21] X. Yang, H. Miura, T. Sakai, Continuous Dynamic Recrystallization in a
Superplastic 7075 Aluminum Alloy, Mater. Trans. 43 (10) (2002) 2400–2407.
Declaration of Competing interest [22] J.-Q. Su, T.W. Nelson, C.J. Sterling, Microstructure evolution during FSW/FSP of
high strength aluminum alloys, Mater. Sci. Eng. A 405 (1) (2005) 277–286.
[23] S. Liu, Q. Pan, M. Li, X. Wang, X. He, X. Li, Z. Peng, J. Lai, Microstructure evolution
The authors declare that they have no known competing financial and physical-based diffusion constitutive analysis of Al-Mg-Si alloy during hot
interests or personal relationships that could have appeared to influence deformation, Mater. Des. 184 (2019), 108181.
[24] A. Chamanfar, M.T. Alamoudi, N.E. Nanninga, W.Z. Misiolek, Analysis of flow
the work reported in this paper. stress and microstructure during hot compression of 6099 aluminum alloy
(AA6099), Mater. Sci. Eng. A 743 (2019) 684–696.
Acknowledgments [25] W. Blum, Q. Zhu, R. Merkel, H.J. McQueen, Geometric dynamic recrystallization in
hot torsion of Al5Mg0.6Mn (AA5083), Mater. Sci. Eng. A 205 (1) (1996) 23–30.
[26] C. Castan, F. Montheillet, A. Perlade, Dynamic recrystallization mechanisms of an
This work was financially supported by Key-Area Research and Fe–8% Al low density steel under hot rolling conditions, Scr. Mater. 68 (6) (2013)
Development Program of Guangdong Province (grant number: 360–364.
2020B010186002). [27] A. Momeni, K. Dehghani, G.R. Ebrahimi, Modeling the initiation of dynamic
recrystallization using a dynamic recovery model, J. Alloys Compd. 509 (39)
(2011) 9387–9393.
References [28] G. Sha, A. Cerezo, Kinetic Monte Carlo simulation of clustering in an Al–Zn–Mg–Cu
alloy (7050), Acta Mater. 53 (4) (2005) 907–917.
[1] A. Heinz, A. Haszler, C. Keidel, S. Moldenhauer, R. Benedictus, W.S. Miller, Recent [29] A. Deschamps, F. De Geuser, Z. Horita, S. Lee, G. Renou, Precipitation kinetics in a
development in aluminium alloys for aerospace applications, Mater. Sci. Eng. A severely plastically deformed 7075 aluminium alloy, Acta Mater. 66 (2014)
280 (1) (2000) 102–107. 105–117.
[2] J.C. Williams, E.A. Starke, Progress in structural materials for aerospace [30] O.A. Kogtenkova, A.A. Mazilkin, B.B. Straumal, G.E. Abrosimova, P. Zięba,
systems11The Golden Jubilee Issue—Selected topics in Materials Science and T. Czeppe, B. Baretzky, R.Z. Valiev, Phase transformations in Al–Mg–Zn alloys
Engineering: Past, Present and Future, edited by S. Suresh, Acta Mater. 51 (19) during high pressure torsion and subsequent heating, J. Mater. Sci. 48 (13) (2013)
(2003) 5775–5799. 4758–4765.
[3] H. Masuda, H. Tobe, T. Hara, E. Sato, Three-dimensional characterization of [31] Y. Zhang, S. Jin, P.W. Trimby, X. Liao, M.Y. Murashkin, R.Z. Valiev, J. Liu, J.
superplastic grain boundary sliding inside Al–Zn–Mg–Cu alloy sheet, Scr. Mater. M. Cairney, S.P. Ringer, G. Sha, Dynamic precipitation, segregation and
164 (2019) 82–85. strengthening of an Al-Zn-Mg-Cu alloy (AA7075) processed by high-pressure
[4] M. Ebrahimi, M.H. Shaeri, R. Naseri, C. Gode, Equal channel angular extrusion for torsion, Acta Mater. 162 (2019) 19–32.
tube configuration of Al-Zn-Mg-Cu alloy, Mater. Sci. Eng. A 731 (2018) 569–576. [32] J.D.E. A, A.D. B, Y.B. B, The interaction of plasticity and diffusion controlled
[5] B. Li, Q. Pan, X. Huang, Z. Yin, Microstructures and properties of Al–Zn–Mg–Mn precipitation reactions, Scr. Mater. 49 (10) (2003) 927–932.
alloy with trace amounts of Sc and Zr, Mater. Sci. Eng. A 616 (2014) 219–228. [33] F. Jiang, L. Tang, J. Huang, Y. Cai, Z. Yin, Influence of equal channel angular
[6] Y. Deng, Z. Yin, K. Zhao, J. Duan, J. Hu, Z. He, Effects of Sc and Zr microalloying pressing on the evolution of microstructures, aging behavior and mechanical
additions and aging time at 120◦ C on the corrosion behaviour of an Al–Zn–Mg properties of as-quenched Al-6.6Zn-1.25Mg alloy, Mater. Charact. 153 (2019)
alloy, Corros. Sci. 65 (2012) 288–298. 1–13.
[7] Y.C. Lin, X.-M. Chen, A critical review of experimental results and constitutive [34] S. Wang, L.G. Hou, J.R. Luo, J.S. Zhang, L.Z. Zhuang, Characterization of hot
descriptions for metals and alloys in hot working, Mater. Des. 32 (4) (2011) workability in AA 7050 aluminum alloy using activation energy and 3-D processing
1733–1759. map, J. Mater. Process. Technol. 225 (2015) 110–121.
[8] D. Feng, X.M. Zhang, S.D. Liu, Y.L. Deng, Constitutive equation and hot [35] Y.C. Lin, L.-T. Li, Y.-C. Xia, Y.-Q. Jiang, Hot deformation and processing map of a
deformation behavior of homogenized Al–7.68Zn–2.12Mg–1.98Cu–0.12Zr alloy typical Al–Zn–Mg–Cu alloy, J. Alloys Compd. 550 (2013) 438–445.
during compression at elevated temperature, Mater. Sci. Eng. A 608 (2014) 63–72. [36] J.D. Robson, A new model for prediction of dispersoid precipitation in aluminium
[9] J.M. Cabrera, A. Al Omar, J.M. Prado, J.J. Jonas, Modeling the flow behavior of a alloys containing zirconium and scandium, Acta Mater. 52 (6) (2004) 1409–1421.
medium carbon microalloyed steel under hot working conditions, Metall. Mater. [37] C. Shi, X.G. Chen, Effect of Zr addition on hot deformation behavior and
Trans. A 28 (11) (1997) 2233–2244. microstructural evolution of AA7150 aluminum alloy, Mater. Sci. Eng. A 596
[10] J.M. Cabrera, J.J. Jonas, J.M. Prado, Flow behaviour of medium carbon (2014) 183–193.
microalloyed steel under hot working conditions, Mater. Sci. Technol. 12 (7) [38] Z.C. Sun, H.L. Wu, J. Cao, Z.K. Yin, Modeling of continuous dynamic
(1996) 579–585. recrystallization of Al-Zn-Cu-Mg alloy during hot deformation based on the
[11] K. Suresh, K.P. Rao, Y.V.R.K. Prasad, N. Hort, K.U. Kainer, Study of hot forging internal-state-variable (ISV) method, Int. J. Plast. 106 (2018) 73–87.
behavior of as-cast Mg–3Al–1Zn–2Ca alloy towards optimization of its hot [39] P. Poelt, C. Sommitsch, S. Mitsche, M. Walter, Dynamic recrystallization of Ni-base
workability, Mater. Des. 57 (2014) 697–704. alloys—Experimental results and comparisons with simulations, Mater. Sci. Eng. A
[12] C. Zener, J.H. Hollomon, Effect of Strain Rate Upon Plastic Flow of Steel, J. Appl. 420 (1) (2006) 306–314.
Phys. 15 (1) (1944) 22–32. [41] M.J. Wang, C.Y. Sun, M.W. Fu, Z.L. Liu, L.Y. Qian, Study on the dynamic
[13] H.E. Hu, L. Zhen, L. Yang, W.Z. Shao, B.Y. Zhang, Deformation behavior and recrystallization mechanisms of Inconel 740 superalloy during hot deformation,
microstructure evolution of 7050 aluminum alloy during high temperature J. Alloys Compd. 820 (2020), 153325.
deformation, Mater. Sci. Eng. A 488 (1) (2008) 64–71. [42] K. Huang, R.E. Logé, A review of dynamic recrystallization phenomena in metallic
[14] W. Liu, H. Zhao, D. Li, Z. Zhang, G. Huang, Q. Liu, Hot deformation behavior of materials, Mater. Des. 111 (2016) 548–574.
AA7085 aluminum alloy during isothermal compression at elevated temperature, [43] J. Xie, Y. Zhu, F. Bian, C. Liu, Dynamic recovery and recrystallization mechanisms
Mater. Sci. Eng. A 596 (2014) 176–182. during ultrasonic spot welding of Al-Cu-Mg alloy, Mater. Charact. 132 (2017)
[15] D. Ponge, M. Bredehöft, G. Gottstein, Dynamic recrystallization in high purity 145–155.
aluminum, Scr. Mater. 37 (11) (1997) 1769–1775. [44] R. Kaibyshev, S. Malopheyev, Mechanisms of Dynamic Recrystallization in
[16] T. Sheppard, M.G. Tutcher, Development of duplex deformation substructure Aluminum Alloys, Mater. Sci. Forum 794-796 (2014) 784–789.
during extrusion of a commercial Al-5Mg-0.8Mn alloy, Metal, Science 14 (12)
(1980) 579–590.

14