materials

Article
Prediction of the Work-Hardening Exponent for 3104
Aluminum Sheets with Different Grain Sizes
Ni Tian 1,2, *, Fei Yuan 1 , Ceheng Duan 1 , Kun Liu 3, * , Guangdong Wang 1 , Gang Zhao 1,2
and Liang Zuo 1
1 Key Laboratory for Anisotropy and Texture of Materials, School of Materials Science &Engineering,
Northeastern University, Shenyang 110819, China
2 Research Center for Metallic Wires, Northeastern University, Shenyang 110819, China
3 Department of Applied Science, University of Quebec at Chicoutimi, Chicoutimi, QC G7H2B1, Canada
* Correspondence: tiann@atm.neu.edu.cn (N.T.); kun.liu@uqac.ca (K.L.);
Tel.: +86-024-83691571 (N.T.); +1-418-545-5011 (ext. 7112) (K.L.)




Received: 28 June 2019; Accepted: 23 July 2019; Published: 25 July 2019


Abstract: A practical approach to predict the yield strength and work-hardening exponent (n value)
to evaluate the deep-drawing performance of annealed 3104 aluminum sheets is presented in the
present work by only measuring and analyzing the grain size of the sheet. The various grain sizes
were obtained through the different annealing treatment and then the evolution of the n value
under different strains and the yield strength of annealed 3104 aluminum sheet were evaluated.
Results showed that the n value and yield strength vary greatly with the grain size. A mathematical
model relating grain size d, work-hardening exponent n, target strain ε, and yield strength Rp0.2 was
developed in the present work. Within the studied grain size range d (12–29 µm), the n value generally
increased with d in a strain-dependent manner, such that n = 0.1875 − 85.03 × exp[−d/1.94] when
the ε was less than 0.5%, but n = 0.3 − 0.15d−1/2 when the ε was greater than 2%. On the other hand,
the n value was found to depend on the target strain ε as n = 0.276 − A1 × exp[−e/1.0435], where A1
varies with d and its value is in the range of 0.132–0.364. In addition, the relationship between Rp0.2
and d followed the Hall-Petch equation (Rp0.2 = 36.273 + 139.8 × d−1/2 ).

Keywords: annealed 3104 aluminum sheet; grain size; n value; target strain; yield strength

1. Introduction
Both the work-hardening exponent (n value) and the yield strength are the essential parameters
governing the deep-drawing properties of sheets. In general, for most metallic material sheets,
the higher the n value is, the lower the yield strength is, and the better the deep-drawing property
is. The n value and yield strength are strongly dependent on the microstructure of the sheet, such as
grain size, second-phase particles, dislocation density, etc. Researchers have attempted to reveal the
relationship between microstructures and n value as well as yield strength in order to predict the
deep-drawing properties of sheets. It is well-known that the relationship between yield stress and
grain size of most metals and alloys can be expressed by the Hall-Petch equation though with different
Hall-Petch constants from the different materials or microstructures [1,2]. However, there are few
studies about the relationship between grain size d, calculated strain point, and the work-hardening
exponent n [3–5]. Previously [6], it was observed that the n value for Al-Mg-Si-Cu alloy sheets
initially increased before stabilizing with increasing strain in a solid solution and in T4 states; on
the other hand, after annealing, it initially increased and then decreased under the same conditions.
These different behaviors can be attributed to the various microstructures resulting from the different
conditions, especially the annealing condition. The n value was also reported to differ in different alloys

Materials 2019, 12, 2368; doi:10.3390/ma12152368 www.mdpi.com/journal/materials

Materials 2019, 12, 2368 2 of 12

subjected to the same strain conditions. Liu et al. [7] found that n was greatest for AA5056, followed
by AA2024, AA2014, and AA7178 under identical strain conditions, although n consistently increased
with strain in all cases. Mo et al. [8] reported that n for A319-T6 aluminum alloy is quite sensitive
to the secondary dendrite arm spacing (SDAS) and Si particles (e.g., their aspect ratio and volume
fraction). The n value decreases with increasing of SDAS and aspect ratio of Si particles. Tsuchida [9]
reported that n of industrial pure aluminum is also related to dislocation dynamics. Many studies
have also shown that the distributions of intermetallics and dislocations significantly affect n in metals
and alloys [10–13]. The n value is thus clearly influenced by distribution of intermetallics and particles,
as well as by dislocations. However, there is scant literature on the impact of grain size d on n value in
aluminum alloys.
Some researchers have worked out the relationship between n value and the grain size d for
steel [14–16], which is convenient to predict the deep-drawing ability of steel. Morrison [14] found
that n increases with d in low-carbon steel, according to the empirical formula n = 10+d5−1/2 (d in mm).
After fully considering the comprehensive interactions between d, the size and volume of precipitates,
and the dislocation density in Ti-IF steel, Antoine [15] proposed a model for predicting n as
 
1/2
 
f
!
 6.6 X V r TiX −2

n = 0.450 − 0.001σeff + √ + TiX
ρ
 p
0.00065  × ln + 1.92 × 10 (1)
  
rTiX 2.48 × 10−7 F0.2 
d TiX
  

where, σeff is a short distance interaction stress or effective stress, d is the grain size (mm), fVTiX and rTiX
are the volume fraction and average radius of the TiX-type precipitates, respectively, and ρF0.2 is the
dislocation density at the yield strength. It can be found that the models proposed by Morrison and
Antoine for calculating n in steel are significantly different, despite low-carbon steel and Ti-IF steel
both displaying a body-centered cubic (BCC) structure. Qiu [16] found that n linearly increases with
decreasing ferrite grain size d−1/2 , and then proposed a model for predicting n as n = 0.3301 − 0.2401d−1/2
(d in the range of 1.30~37 µm) in ultrafine-grained steels. Gashti [17] found that AA1050 aluminum
alloy sheets show lower n due to grain refinement.
On the other hand, there are presently no established models for aluminum alloys and therefore
a strong motivation in industry for developing a model to predict n for a deep-drawn aluminum
sheet. Therefore, the non-age-hardening 3104 aluminum alloy, which is widely used in industry for
fabricating deep-drawing aluminum sheets, was investigated to determine its relationship between n
value, yield strength, and grain size obtained from various annealing processes. The mathematical
model was established based on the experimental data, aiming to predict the n value for annealed
3104 aluminum sheets and then provide experimental data for the stamping process of 3104 aluminum
alloy sheets or other aluminum alloy sheets.

2. Materials and Methods

Semicontinuous DC (direct chill) casting of 3104 aluminum alloy was performed in the lab with
the dimensions of 200 mm (width) × 80 mm (thickness) × 600 mm (length) with the composition of
(wt%) 1.21% Mn, 1.04% Mg, 0.25% Fe, 0.30% Si, 0.18% Cu, 0.11% Ti, and Al balance. The specimen
for the chemical composition analysis was cut from the casting ingot and analyzed by an Oxford
Foundry-Master Pro Optical Emission Spectrometer (Oxford Instruments PLC, Uedem, Germany).
The ingot was homogenized at 580 ◦ C for 24 h, followed by hot-rolling at 480 ◦ C to 6 mm, and then
cold-rolled to 1.0 mm sheet. After rolling, sheets were annealed at 400 ◦ C/1 h, 500 ◦ C/1 h, 550 ◦ C/1 h,
4 h, 8 h, and 24 h to obtain different grain sizes. After each annealing process, three specimens were cut
from the sheet along the rolling direction and then machined to the dimensions in Figure 1 according
to ASTM (American Society for Testing Material) E8/E8M-2013a for tensile testing. Tensile tests were
conducted using a Shimadzu AG-X100KN electronic universal testing machine (Shimadzu Coporation,
Kyoto, Japan) at room temperature with a strain rate of 0.08 min−1 . The true stress vs. true strain
curve was drawn by determining the real-time load and the real-time geometry size of specimen (the
Materials 2019, 12, x FOR PEER REVIEW 3 of 13

Materials 2019, 12, 2368 3 of 12

stress vs. true strain curve was drawn by determining the real-time load and the real-time geometry
size of specimen (the real-time length/width), which were measured by a load sensor, a length
extensometer with 50 mm
real-time length/width), whichgauge
were length,
measuredand abywidtha loadextensometer
sensor, a length with 12.5 mm gauge
extensometer with length,
50 mm
respectively.
gauge length,Specimens
and a width forextensometer
metallography withwere
12.5 cut
mmfromgaugethe tensile
length, specimens Specimens
respectively. before tensile
for
deformation and then ground with sandpaper till #5000 and finalized with
metallography were cut from the tensile specimens before tensile deformation and then ground with the fine polishing on the
polishing
sandpapermachine.
till #5000The
andmorphology
finalized with andthedistribution
fine polishing of the
on intermetallics and dispersoids
the polishing machine. after each
The morphology
annealing treatment
and distribution of thewere observed using
intermetallics an Olympus
and dispersoids GX71
after eachoptical microscope
annealing treatment (OM)
were(Olympus
observed
Coporation, Tokyo, Japan) and a Tecnai G2 transmission electron microscope
using an Olympus GX71 optical microscope (OM) (Olympus Coporation, Tokyo, Japan) and a Tecnai (TEM) (FEI Company,
Hillsboro, OR, electron
G2 transmission USA). TEM samples
microscope (TEM)were (FEIcut from the
Company, tensile OR,
Hillsboro, specimen
USA). TEMbeforesamples
the tensile
were
deformation, mechanical-thinned to approximately 90 μm, and then electropolished
cut from the tensile specimen before the tensile deformation, mechanical-thinned to approximately by twin-jet in a
solution of 25% HNO + 75% CH OH at −25 °C and under a voltage of
90 µm, and then electropolished by twin-jet in a solution of 25% HNO3 + 75% CH3 OH at −25 C and
3 3 15 V (MTP-1A magnetic
◦
twin-jet electropolishing,
under a voltage of 15 V (MTP-1AShanghai Jiaoda
magnetic electromechanical
twin-jet electropolishing, technology development
Shanghai Jiaoda Co. Ltd.,
electromechanical
Shanghai, China). The grain size was revealed by electropolishing in a solution
technology development Co. Ltd., Shanghai, China). The grain size was revealed by electropolishing of 25% HNO 3 + 75%

CH
in a3OH at −5of°C
solution andHNO
25% under a voltage of 20 V and characterized by electron back-scatter diffraction
3 + 75% CH3 OH at −5 C and under a voltage of 20 V and characterized by
◦
(EBSD)
electronwith a step size
back-scatter of 2 μm(EBSD)
diffraction and 40withkV acceleration
a step size ofvoltage
2 µm and by 40
a JMS-7001F scanning
kV acceleration electron
voltage by a
microscope (SEM) (JEOL Ltd., Tokyo, Japan).
JMS-7001F scanning electron microscope (SEM) (JEOL Ltd., Tokyo, Japan).

Figure 1. Dimensions of the tensile specimens.

Figure 1. Dimensions of the tensile specimens.

The true stress and strain of the aluminum sheet are assumed to obey σ = Kεn or, equivalently,
The true stress and strain of the aluminum sheet are assumed to obey σ = K ε n or,
log σ = log K + n log ε. The n value was therefore calculated for different sheet strains by a
equivalently, log σ = log K + n log ε . The n value was therefore calculated for different sheet
least-squares approach:
strains by a least-squares approach: P N N
P
N log εi · log σi − log εi · log σi
N i=1 i=N1
nN= ε i ⋅ log σ i - ∑P (2)
∑ logP N N log ε i 2⋅ log σ i
2
n= i =1 N
Ni=1
(log εi ) − (i =1 log εi )
i=N1
(2)
2 2
N ∑ in(log
where N = 20 is the number of strain points ε i ) -strain
the target ( ∑ log ε i )and ε and σ are the instantaneous
range, i i
i =1 i =1
true strain and stress corresponding to the calculated strain point, respectively. The n value was
where N = by
determined 20 increasing
is the number of strain
the strain within points in the plastic
the uniform target deformation
strain range,range, i and σi from
and εstarting are the
yield point ε =true
instantaneous 0.2%.strain and stress corresponding to the calculated strain point, respectively. The n
value was determined by increasing the strain within the uniform plastic deformation range,
3. Results
starting andthe
from Discussion
yield point ε = 0.2%.
3.1. Microstructures of 3104 Sheets after Various Annealing Treatments
3. Results and Discussion
OM and TEM observations of 3104 sheets after various annealing treatments are displayed in
3.1. Microstructures
Figures of 3104 Sheets
2 and 3, respectively. afterare
There Various
manyAnnealing Treatments intermetallics in Figure 2, while
broken micron-scale
someOM fineand
particles (~200 nm) areofobserved
TEM observations in Figure
3104 sheets 3. According
after various to the
annealing SEM-EDSare
treatments results and the
displayed in
literature [18], the broken intermetallics consist of Al 6 (MnFe), which is formed during
Figures 2 and 3, respectively. There are many broken micron-scale intermetallics in Figure 2, while solidification
and then
some finefragmented intonm)
particles (~200 small particles
are observedduring the rolling
in Figure process. to
3. According Thethehigh meltingresults
SEM-EDS point of these
and the
intermetallics prevents their dissolution into the matrix during annealing, so they remain
literature [18], the broken intermetallics consist of Al6(MnFe), which is formed during solidification embedded
within
and thenthefragmented
matrix [18]. into
As shown
small in Figure 2,
particles the size
during and
the volume
rolling of the Al
process. The6 (MnFe) particlespoint
high melting are relatively
of these
independent of the applied annealing protocol, as they are closely related to the Mn
intermetallics prevents their dissolution into the matrix during annealing, so they remain embedded and Fe contents in
the alloy [18]. Some fine particles, reportedly α-Al(MnFe)Si dispersoids, are also observed in the TEM
 Materials 2019, 12, x FOR PEER REVIEW 4 of 13
Materials 2019, 12, x FOR PEER REVIEW 4 of 13
within the matrix [18]. As shown in Figure 2, the size and volume of the Al6(MnFe) particles are
within
relatively
Materials
the12,matrix
2019, independent
2368
[18].ofAsthe
shown in annealing
applied Figure 2, protocol,
the size and volume
as they of the related
are closely Al6(MnFe)
to theparticles
Mn andare4Feof 12
relatively independent of the applied annealing protocol, as they are closely related
contents in the alloy [18]. Some fine particles, reportedly α-Al(MnFe)Si dispersoids, are also to the Mn and Fe
contents in the
observed theTEMalloyimage
[18]. ofSome fine
Figure particles,
3 [18]. reportedly
Their volume α-Al(MnFe)Si
fraction is relativelydispersoids,
low and their aresize
alsois
observed
image
consistently in the
of Figure TEMthe
3 [18].
within image
Their of Figure
volume
range 3 [18].
fraction
of 100–200 nmTheir
is volumelow
relatively
regardless fraction
of is relatively
andannealing
the their size is low and their
consistently
protocol. size in
within
Therefore, isthe
consistently
summary,
range of 100–200 within
the the rangeand
intermetallic
nm regardless of
of 100–200
dispersoid
the nmcharacteristics
annealing regardless
protocol.ofTherefore,
the annealing
(volume, size,
in protocol. Therefore,
and distribution)
summary, the in
are
intermetallic
summary,
relatively
and dispersoid the intermetallic
independent and dispersoid
of the(volume,
characteristics precise size, andcharacteristics
annealing protocol. (volume,
distribution) are size,
relatively and distribution)
independent are
of the precise
relatively independent
annealing protocol. of the precise annealing protocol.

Figure 2. Microstructure of 3104 alloy sheet after various annealing treatments: (a) 400 °C/1 h; (b) 500
Figure 2. Microstructure of 3104 alloy sheet after various annealing treatments: (a) 400 ◦ C/1 h;
Figure
°C/1 h; 2.
(c)Microstructure
550 °C/1 h; (d) of
5503104
°C/4alloy
h; (e)sheet after h;
550 °C/8 various
(f) 550annealing
°C/24 h. treatments: (a) 400 °C/1 h; (b) 500
(b) 500 ◦ C/1 h; (c) 550 ◦ C/1 h; (d) 550 ◦ C/4 h; (e) 550 ◦ C/8 h; (f) 550 ◦ C/24 h.
°C/1 h; (c) 550 °C/1 h; (d) 550 °C/4 h; (e) 550 °C/8 h; (f) 550 °C/24 h.

(a):400℃/1h (b):500℃/1h (c):550℃/1h

(a):400℃/1h (b):500℃/1h (c):550℃/1h

(d):550℃/4h (e):550℃/8h (f):550℃/24h

(d):550℃/4h (e):550℃/8h (f):550℃/24h

Figure 3. TEM images of 3104 alloy sheets after various annealing treatments: (a) 400 ◦ C/1 h;
Figure 3. TEM images of 3104 alloy sheets after various annealing treatments: (a) 400 °C/1 h; (b) 500
(b)°C
500 ◦
Figure
/1 h; /1TEM
C3.(c) h;
550(c) 550h;◦ C/1
images
°C/1 h; (d)
of 3104
(d) 550 alloy
°C/4
◦ C/4 h; (e) 550 ◦ C/8 h; (f) 550 ◦ C/24 h.
550h;sheets
(e) 550after
°C/8various annealing
h; (f) 550 °C/24 h. treatments: (a) 400 °C/1 h; (b) 500
°C /1 h; (c) 550 °C/1 h; (d) 550 °C/4 h; (e) 550 °C/8 h; (f) 550 °C/24 h.
Figure
Figure4 4shows
showsthe theEBSD
EBSDresults
resultsafter
aftervarious
various annealing
annealing processes. Fullrecrystallization
processes. Full recrystallizationis is
achieved
achievedunder
Figure all annealing
4 shows
under the EBSD
all annealing conditions, as
results after
conditions, evidenced by the
various annealing
as evidenced uniform equixed
processes.
by the uniform grains
Fullgrains
equixed with random
recrystallization
with random is
orientations, and
achieved under
orientations, andwithout
allwithout any
anytexture.
annealing texture. However,
conditions, the measured
as evidenced
However, the measured grainsize
by the uniform
grain size varies
equixed
varies with
grains
with the
with
the annealing
random
annealing
process, increasing ◦ C/1 h to 13.80 µm after 500 ◦ C/1 h, 19.46 µm after 550 ◦ C/1 h,
orientations,
process, andfrom
increasing from12.65
without any
12.65 after
µmtexture.
μm 400
afterHowever,
400 °C/1 the
h tomeasured
13.80 μm grain size°C/1
after 500 varies
h, with
19.46the
μmannealing
after 550
21.68 µm after ◦
550 C/4 h,12.65
28.11μm after400
µmafter ◦ C/8 h, and 29.19 µm after 550 ◦ C/24 h. The increase
550°C/1
process, increasing from h to 13.80 μm after 500 °C/1 h, 19.46 μm after 550
in grain size can be attributed to the increase either in the annealing temperature or the holding
time. As shown in Figure 4, full recrystallization is already achieved after only 400 ◦ C/1 h annealing.
The grain therefore begins to grow with increasing annealing temperature and holding time [19,20].
Hence, the specimens with different grain sizes and without any texture but similar distributions
Materials 2019, 12, x FOR PEER REVIEW 5 of 13

°C/1 h, 21.68 μm after 550 °C/4 h, 28.11 μm after 550°C/8 h, and 29.19 μm after 550 °C/24 h. The
increase in grain size can be attributed to the increase either in the annealing temperature or the
holding time. As shown in Figure 4, full recrystallization is already achieved after only 400 °C/1 h
annealing. The
Materials 2019, grain therefore begins to grow with increasing annealing temperature and holding5 of 12
12, 2368
time [19,20]. Hence, the specimens with different grain sizes and without any texture but similar
distributions of intermetallics and dispersoids in the matrix were obtained by controlling the
of intermetallics
annealing process,and dispersoids
providing in the to
the premise matrix
studywere obtained
the effect of theby controlling
grain the properties
size on their annealingand
process,
providing
the n value. the premise to study the effect of the grain size on their properties and the n value.

Materials 2019, 12, x FOR PEER REVIEW 6 of 13

(g)

Figure 4. EBSD mapping of 3104 sheets after various annealing treatments: (a) 400 ◦ C/1 h; (b) 500 ◦ C/1 h;
Figure ◦4. EBSD mapping of 3104 sheets after various annealing treatments: (a) 400 °C/1 h; (b) 500
(c) 550 C/1 h; (d) 550 ◦ C/4 h; (e) 550 ◦ C/8 h; (f) 550 ◦ C/24 h; (g) orientation triangle for EBSD mapping
°C/1 h; (c) 550 °C/1 h; (d) 550 °C/4 h; (e) 550 °C/8 h; (f) 550 °C/24 h; (g) orientation triangle for EBSD
mapping

3.2. Dependence of the Yield Strength on Grain Size

Figure 5 plots the typical engineering strain versus engineering stress and log (true strain)
versus log (true stress) for 3104 alloy sheets after various annealing treatments during the tensile
test. Serrated flow is apparent in Figure 5 for all the annealed sheets, a phenomenon referred to as
 Figure 4. EBSD mapping of 3104 sheets after various annealing treatments: (a) 400 °C/1 h; (b) 500
°C/1 h; (c) 550 °C/1 h; (d) 550 °C/4 h; (e) 550 °C/8 h; (f) 550 °C/24 h; (g) orientation triangle for EBSD
Materials 2019, 12, 2368 6 of 12
mapping

3.2.Dependence
3.2. Dependenceofofthe
theYield
YieldStrength
Strengthon
onGrain
GrainSize
Size
Figure 55plots
Figure plotsthethetypical
typicalengineering
engineering strain
strain versus
versus engineering
engineering stress
stress and
and log
log (true
(true strain)
strain)
versuslog
versus log(true
(truestress)
stress)for for3104
3104alloy
alloysheets
sheetsafter
aftervarious
variousannealing
annealingtreatments
treatmentsduring
duringthe thetensile
tensile
test. Serrated
test. Serrated flow
flow isisapparent
apparent in inFigure
Figure55forforall
allthe
theannealed
annealedsheets,
sheets,aaphenomenon
phenomenonreferredreferredto toas
as
thePortevin-LeChatelier
the Portevin-LeChatelier(PLC) (PLC)effect
effect[21,22],
[21,22],which
whichmay maybe beattributed
attributedtotothethedynamic
dynamicstrain
strainaging
aging
(DSA)that
(DSA) thatoccurs
occursduring
duringtensile
tensiledeformation
deformation[23,24].
[23,24].During
Duringthetherolling
rollingprocess
processandandthethesubsequent
subsequent
annealing,the
annealing, theintermetallics
intermetallics areare fragmented
fragmented into
into small
small particles
particles andand dispersoids
dispersoids are precipitated
are precipitated in thein
the matrix,
matrix, bothboth of which
of which can hinder
can hinder the movement
the movement of dislocations.
of dislocations. Meanwhile,
Meanwhile, as shown
as shown in Figures
in Figures 2
2 and
and 3, the3,spacings
the spacings
between between
particlesparticles and dispersoids
and dispersoids are large
are relatively relatively large and
and therefore therefore
dislocations
dislocations
can slip between can slip
them between
withoutthem without
hindrance hindrance
until until they
they become become
pinned by an pinned by an encountered
encountered particle or
particle or dispersoid.
dispersoid. This effect
This effect gives rise togives rise to the
the serrated serrated
flow stress flow stressinobserved
observed in the stress-strain
the stress-strain curves of
curves of Figure 5. This phenomenon was also reported in annealed 5082
Figure 5. This phenomenon was also reported in annealed 5082 aluminum alloy by Abbadi et al. aluminum alloy by Abbadi
[25].
et al. [25]. PLC
However, However, PLC characteristics
characteristics (e.g.,
(e.g., critical critical
strain andstrain and the
the strain strain decrease)
decrease) areunder
are similar similarallunder
the
all the annealing
annealing conditions,
conditions, reflecting reflecting
the weak theinfluence
weak influence of the
of the grain grain
size size on
on PLC PLC behavior.
behavior. This canThisbe
can be explained mainly in terms of similarities in the microstructure, in particular,
explained mainly in terms of similarities in the microstructure, in particular, the distribution of particles the distribution
of particles
and dispersoidsandindispersoids
annealed 3104 in annealed 3104Asalloy
alloy sheets. shown sheets. As shown
in Figures 2 andin3, Figures
the size,2volume
and 3, fraction,
the size,
volume
and fraction,ofand
distribution distribution
particles of particles
and dispersoids and dispersoids
are similar areannealing
after various similar after various Therefore,
treatments. annealing
treatments.
this gives riseTherefore, this gives
to the similar rise to the
PLC behavior similarthe
despite PLC behavior
different despite
grain sizes.the different grain sizes.

200
12.56μm 19.46μm
180
(a)
Engineering stress/MPa

160 13.80μm
140
28.11μm 21.68μm
120
29.19μm
100 12.56μm
80 13.80μm
19.46μm
60
21.68μm
40 28.11μm
20
29.19μm

0
Materials 2019, 12, x FOR PEER REVIEW
0 2 4 6 8 10 12 14 16 18 20 22 24 7 of 13
Engineering strain/%

2.5
(b)

19.46μm
Log (True stress)

2.0 13.80μm 12.56μm

28.11μm 29.19μm
21.68μm
1.5
12.56μm
13.80μm
1.0 19.46μm
21.68μm
0.5 28.11μm
29.19μm
0.0
-3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0
Log (True strain)

Figure 5. Complete typical engineering strain-engineering stress (a) and uniform plastic deformation
Figure
part 5. Complete
of log typical(true
(true strain)-log engineering strain-engineering
stress) (b) stresswith
curves of 3104 sheets (a) and uniform
different plastic
grain size. deformation
part of log (true strain)-log (true stress) (b) curves of 3104 sheets with different grain size.

Figure 5 also shows the significant dependence of the flow stress on the grain size. Figure 6
summarizes the relationship between grain size and yield strength and its fitting results. It can be
seen that the yield strengths decrease slowly for grain sizes up to 29.19 μm. The fit of the grain size
data to the yield strength is consistent with the Hall-Petch law as Rp0.2 = 36.273 + 139.8 × d −1 2 for studied
 0.0
-3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0
Log (True strain)

Figure 5. Complete typical engineering strain-engineering stress (a) and uniform plastic deformation
Materials 2019, 12, 2368 7 of 12
part of log (true strain)-log (true stress) (b) curves of 3104 sheets with different grain size.

Figure 55 also
Figure also shows
shows thethe significant
significant dependence
dependence of of the
the flow
flow stress
stress on on the
the grain
grain size.
size. Figure
Figure 66
summarizes the relationship between grain size and yield strength and its fitting results.
summarizes the relationship between grain size and yield strength and its fitting results. It can be It can be
seen
seenthe
that thatyield
the yield strengths
strengths decrease
decrease slowlyslowly for sizes
for grain grainup
sizes up toµm.
to 29.19 29.19 Theμm.
fit The fitgrain
of the of thesizegrain size
data to
−1 2
data to the yield strength is consistent with the Hall-Petch
the yield strength is consistent with the Hall-Petch law as R law as R
= 36.273
p0.2
= 36.273 + 139.8
+ 139.8 × d × d
−1/2 for studied
for studied
p0.2
12.65–29.19 µm.
grain sizes in the range of 12.65–29.19 μm.

80 Measured values
Fitting curve
70

60
Rp0.2 / MPa

50 R p 0.2 =36.273+139.8d - 1/2

40

30

10 15 20 25 30

Grain size/μ m

Figure 6. Relationship between grain size and yield strength and its fitting result.
result.

3.3. Dependence of the Strengthen-Hardening Exponent (n Value) on Grain Size

3.3. Dependence of the Strengthen-Hardening Exponent (n Value) on Grain Size
Figure 7 shows the dependence of n on the target strains for different grain sizes. An initial rise in
Figure 7 shows the dependence of n on the target strains for different grain sizes. An initial rise
n is generally observed for strains under 2%, followed by a slow increase between 2% and 4%, and
in n is generally observed for strains under 2%, followed by a slow increase between 2% and 4%, and
then a slight
Materials decrease
2019, 12, beyond
x FOR PEER 4%.
REVIEW 8 of 13
then a slight decrease beyond 4%.
0.35

0.30

0.25
n value

0.20

0.15

0.10 12.65μm 13.80μm

19.46μm 21.68μm
28.11μm 29.19μm
0.05

0 2 4 6 8 10 12 14 16 18

ε/%
Figure 7. Evolution of n value at different target
target strains
strains and
and the
the fitting
fitting results.
results.

As
As shown
shown in in Figures
Figures 22 and
and 3,3, the
the matrix
matrix contains
contains aa large
large volume
volume ofof fragmented
fragmented particles
particles and
and
dispersoids that can pin the dislocations. However, these particles and dispersoids are
dispersoids that can pin the dislocations. However, these particles and dispersoids are relatively relatively large
(in the(in
large order
theoforder
microns) and dislocations
of microns) can readily
and dislocations canpass them pass
readily via the Orowan
them bypass
via the Orowanmechanism
bypass
and then leave one dislocation loop, thereby reducing the effective particle spacing.
mechanism and then leave one dislocation loop, thereby reducing the effective particle spacing. It is therefore
It is
increasingly difficult for
therefore increasingly subsequent
difficult dislocations
for subsequent to pass in the
dislocations same
to pass inmanner,
the samewhich leads
manner, to a rapid
which leads
increase
to a rapid n at lower
inincrease in nstains in the
at lower stainsinitial stage
in the of tensile
initial stage ofdeformation (~2% in (~2%
tensile deformation the present study).
in the present
study). A similar phenomenon was reported in annealed Al-Mg-Si sheets [6]. For strains beyond 2%,
a stronger resistance to dislocation motion is expected because of the activated multi-slipping
system and the intersection between dislocations form jogs and Lomer-Cottrell dislocation locks.
However, flow stress increases rapidly because of the large n value, making it possible for edge
dislocations to climb around and for screw dislocations to cross-slip obstacles. Hence, n displays a
Materials 2019, 12, 2368 8 of 12

A similar phenomenon was reported in annealed Al-Mg-Si sheets [6]. For strains beyond 2%, a stronger
resistance to dislocation motion is expected because of the activated multi-slipping system and the
intersection between dislocations form jogs and Lomer-Cottrell dislocation locks. However, flow stress
increases rapidly because of the large n value, making it possible for edge dislocations to climb
around and for screw dislocations to cross-slip obstacles. Hence, n displays a downward trend.
The dependence of n on the target strain ε during deformation obeys the following equation:

n = A0 − A1 × exp[−ε/t] (3)

where A0 , A1 , and t are calculated and summarized in Table 1. It can be found that A0 and t are in the
range of 0.268~0.288 and 0.820~1.192, respectively, while A1 is changing from 0.132–0.364. It can be
approximately calculated that both A0 and t are constant, and the weighted average value of A0 and t
are 0.276 and 1.0356 with variances of 0.0000472 and 0.01639, respectively. Therefore, the root mean
square values of A0 and t are 0.276 and 1.0435, respectively, and the dependence of n on the target
strain ε can be expressed as
n = 0.276 − A1 × exp[−ε/1.0435] (4)

Table 1. Parameters in Equation (3) under various grain sizes.

Parameters in Grain Size/µm

Equation (3) 12.65 13.8 16.46 21.68 28.11 29.19
A0 0.26818 0.27149 0.28778 0.27042 0.28061 0.27934
A1 0.36385 0.23346 0.20038 0.14191 0.13157 0.14068
t 0.91024 1.11051 0.82007 1.19239 1.08674 1.09368

Figure 8 plots the dependence of n on the grain size for different target strains. It can be seen that
the n value increases with the increasing of d, which is consistent with the result of AA1050 aluminum
alloy sheet proposed by Gashti [17]. The gradient is generally positive, but the exact dependence varies
with the target strain. As shown in Figure 8a, when the target strain is less than 0.5% (i.e., in the range
0.2% to 0.5%), the dependence of n on the grain size d can be expressed as:

n = 0.1875 − 85.03 × exp[−d/1.94] (5)

With increasing target strain (the target strain is greater than or equal to 2%), the relationship
between n and d changes as below (see fitting results in Figure 8b,c):

n = M − K × d−1/2 (6)

where both M and K are constant values summarized in Table 2. M remains approximately 0.3 for all
the target strains, while K is approximately 0.15, although the target strain is 2%. Therefore, when the
target strain is greater than 2%, the dependence of n on the grain size d can be expressed as:

n = 0.3 − 0.15d−1/2 (7)

 n = M − K  d −1 2 (6)
where both M and K are constant values summarized in Table 2. M remains approximately 0.3 for all
the target strains, while K is approximately 0.15, although the target strain is 2%. Therefore, when
the target strain is greater than 2%, the dependence of n on the grain size d can be expressed as:
Materials 2019, 12, 2368 9 of 12
n = 0.3 − 0.15d−1 2 (7)

0.24

(a)
0.20

0.16

n value
0.12
n=0.1875-85.03×exp(-d/1.94)
0.08

0.04 =
Fitting curve
0.00
12 14 16 18 20 22 24 26 28 30

Grain size/m

0.4

(b)

0.3
n value

0.2

=2% =4%
=6% =8%

0.1
12
Materials 2019, 12, x FOR PEER REVIEW 14 16 18 20 22 24 26 28 30 10 of 13

Grain size /m

0.4

(c)

0.3
n value

0.2
ε=10% ε=12%
ε=14% ε=16%

0.1
12 14 16 18 20 22 24 26 28 30

Grain size/μm
Figure 8. Relationship between grain size and n value at different target strains: (a) ε = 0.5%; (b) ε = 2%,
4%, 6%, and
Figure 8%; (c) ε = 10%,
8. Relationship 12%, grain
between 14%, and
size 16%.
and n value at different target strains: (a) ε = 0.5%; (b) ε =
2%, 4%, 6%, and 8%; (c) ε = 10%, 12%, 14%, and 16%.

Table 2. Parameters in Equation (6) for the n value at different target strains.

Target Strain
2% 4% 6% 8% 10% 12% 14% 16%
Parameter in Equation (6)
M 0.3160 0.3081 0.3151 0.3148 0.3057 0.2976 0.3100 0.3097
K 0.3091 0.1623 0.1624 0.1613 0.1291 0.1062 0.1634 0.1811

According to Equation (7), the relationship between n and the grain size in 3104 alloy sheets
differs from the model established by Morrison for low-carbon steel [14], namely 5 , but is
Materials 2019, 12, 2368 10 of 12

Table 2. Parameters in Equation (6) for the n value at different target strains.

Target Strain
2% 4% 6% 8% 10% 12% 14% 16%
Parameter in Equation (6)
M 0.3160 0.3081 0.3151 0.3148 0.3057 0.2976 0.3100 0.3097
K 0.3091 0.1623 0.1624 0.1613 0.1291 0.1062 0.1634 0.1811

According to Equation (7), the relationship between n and the grain size in 3104 alloy sheets differs
from the model established by Morrison for low-carbon steel [14], namely n = 10+d5−1/2 , but is similar
to the model established by Qiu for ultrafine-grained steels [16], namely n = 0.3301 − 0.2401d−1/2 (d in
the range of 1.30~37 µm). On the other hand, if we remove explicit references to quantities other than
the grain size, Equation (1) simplifies to:

6.6
n = 0.450 − 0.001 × [ √ + A] = C1 − C2 × d−1/2 (8)
d

where C1 and C2 are constants. Equation (7) then resembles Equation (8), in so far as n decreases
linearly with d−1/2 . However, the applicable target strain is in the range 0.1%~0.2% in Antoine’s model
(Equation (1)) for Ti-IF steel, whereas, in the present work, it exceeds or equals to 2% and the grain
size should be less than 29.19 µm for 3104 alloy sheets to obey Equation (6). This indicates that the
relationship between n and the grain size differs in low-carbon steel, ultrafine-grained steels, Ti-IF steel,
and 3104 alloy sheets, though n consistently increases with increasing grain size in all four materials.
The differences are mainly attributed to the differences in crystal structure and the distributions of
secondary particles in steel and aluminum alloy.
In the early stage of deformation (strain below 0.5%), as shown in Figure 8a, the n values for
sheets with different grain sizes are relatively small, indicating the weak effect from strain hardening.
With increasing strain, n increases rapidly with increasing grain size, followed by a slower increase or
a saturation (Figure 7). Because the dislocation density is initially relatively low, grain boundaries and
secondary particles are the dominant obstacles for dislocations. Although the total grain boundary
area decreases with increasing grain size, flow stress also decreases (Figure 5), leading to an increase
in obstacles to dislocation motion. Thus, for a given strain, the rate of strain increase is higher as a
result of stronger strain hardening. This is the likely reason for the rapid increase in n with grain size.
With further increase of strain, n increases more slowly, with a positive grain size dependence. This can
be attributed to additional sources of obstacles to dislocation motion, e.g., the activation of multislip
systems, and the formation of jogs and Lomer-Cottrell dislocation locks (arising at the intersection of
dislocations under strains in excess of 2%). The proportion of obstacles associated specifically with
grain boundaries is thus reduced. Therefore, the rate of increase of n with grain size is much lower for
strains higher than 4%.
Generally, deep-drawing properties will be improved with increasing n value [26]. Therefore,
it seems that the deep-drawing properties of studied annealed 3104 aluminum sheet can benefit
from the coarser grain since the n increases with the grain size in the studied range (~29 um).
However, an orange-peel microstructure is often observed in specimens with larger grains after deep
drawing [27,28], leading to an inferior surface finish of the final product. Further study is therefore
necessary to find the best balance between n and the grain size, in order to improve the deep-drawing
properties and the quality of the surface finish.

4. Conclusions
(1) When increasing grain size, similar serrated flow phenomena (also known as the Portevin-Le
Chatelier (PLC) effect) were observed under all conditions in studied annealed 3104 aluminum
sheet. However, the yield strength (Rp0.2 ) decreased significantly, which followed the Hall-Petch
equation (Rp0.2 = 36.273 + 139.8d−1/2 ) in the studied range of grain size (12.56–29.19 µm).
Materials 2019, 12, 2368 11 of 12

(2) The relationship between n and the target strain ε obeyed n = 0.276 − A1 × exp[−ε/1.0435], where
A1 is varying with d and its value is in the range of 0.132–0.364.
(3) Generally, n increases with grain size d but it follows different tendencies with d during the various
range of strain. When the strain is 0.2–0.5%, the relationship is: n = 0.1875 − 85.03 × exp [−d/1.94] ,
while it changes to n = 0.3 − 0.15d−1/2 when the strain is greater than 2%.

Author Contributions: N.T. designed the experiments and wrote this manuscript. F.Y. and C.D. helped with the
experiments and the analysis of the experimental data. K.L. reviewed and edited the manuscript. G.W., G.Z. and
L.Z. gave some constructive suggestions.
Funding: This work was financially supported by the National Key Research and Development Program of China
(2016YFB0300801), the National Natural Science Foundation of China (51871043) and the Fundamental Research
Funds for the Central Universities (N180212010), China.
Acknowledgments: The authors additionally acknowledge the work of Xin Fan for the initial work on performing
the fundamental experiments.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Abbadi, M.; Hähner, P.; Zeghloul, A. On the characteristics of Portevin–Le Chatelier bands in aluminum
alloy 5182 under stress-controlled and strain-controlled tensile testing. Mater. Sci. Eng. A 2002, 337, 194–201.
[CrossRef]
2. Hanse, N. The effect of grain size and strain on the tensile flow stress of aluminum at room temperature.
J. Acta Metal. 1977, 25, 863–869. [CrossRef]
3. Horita, Z.; Fujinami, T.; Nemoto, M.; Langdon, T.G. Equal-channel angular pressing of commercial aluminum
alloys: Grain refinement, thermal stability and tensile properties. Met. Mater. Trans. A 2000, 31, 691–701.
[CrossRef]
4. Tang, C.G.; Zhu, J.H.; Zhang, Y.H.; Zhou, H.J. Effect of strain rate on strain hardening exponent n of some
metallic materials. J. Acta Metall. Sin. 1994, 7, 183–186.
5. Song, Y.Q.; Suo, Z.G.; Guan, Z.P. Mechanical analysis of tensile instability due to material parameter. J. Acta
Metal. Sinica. 2006, 42, 337–340.
6. Song, Y.Q.; Guan, Z.P.; Ma, P.K.; Song, J.W. Theoretical and experimental standardization of strain hardening
index in tensile deformation. Acta Metall. Sin. Chin. Ed. 2006, 42, 673–680. [CrossRef]
7. Tian, N.; Zhao, G.; Zuo, L.; Liu, C. Study on the strain hardening behavior of Al–Mg–Si–Cu alloy sheet for
automotive body. Acta Metall. Sin. Chin. Ed. 2010, 46, 613–617. [CrossRef]
8. Liu, J.; Wei, Q.L.; Bi, Y.M. Deformation hardening behaviors of deformed aluminum alloys. J. Phys. Test.
Chem. Anal. 2008, 44, 4–6. [CrossRef]
9. Mo, D.F.; He, G.Q.; Hu, Z.F.; Liu, X.S. Effect of microstructures on strain hardening exponent prediction of
cast aluminum alloy. Acta Metall. Sin. Chin. Ed. 2010, 46, 184–188. [CrossRef]
10. Tsuchida, S.; Yokoi, H.; Kanetake, N. Measurement of strain hardening exponent n-value by additional
rolling strain for aluminum alloy sheets. J. Jpn. Inst. Light Met. 2009, 59, 195–200. [CrossRef]
11. Caceres, C.; Griffiths, J.; Reiner, P. The influence of microstructure on the Bauschinger effect in an Al-Si-Mg
casting alloy. Acta Mater. 1996, 44, 15–23. [CrossRef]
12. Bergström, Y.; Aronsson, B. The application of a dislocation model to the strain and temperature dependence
of the strain hardening exponentn in the Ludwik-Hollomon relation between stress and strain in mild steels.
Met. Mater. Trans. A 1972, 3, 1951–1957. [CrossRef]
13. Song, Y.Q.; Cheng, Y.C.; Wang, X.W. Experimental measurement and elaborate analysis of strain hardening
exponent in tensile deformation. J. Sci. China Ser. E 2001, 44, 365–376. [CrossRef]
14. Wang, Q.G. Plastic deformation behavior of aluminum casting alloys A356/357. Met. Mater. Trans. A 2004,
35, 2707–2718. [CrossRef]
15. Morrison, W.B. The effect of grain size on the stress-strain relationship in low carbon steel. J. Metall. Mater.
Trans. B 1966, 59, 824–846.
16. Antoine, P.; Vandeputte, S.; Vogt, J.-B. Empirical model predicting the value of the strain-hardening exponent
of a Ti-IF steel grade. Mater. Sci. Eng. A 2006, 433, 55–63. [CrossRef]
Materials 2019, 12, 2368 12 of 12

17. Qiu, H.; Wang, L.; Hanamura, T.; Torizuka, S. Prediction of the work-hardening exponent for ultrafine-grained
steels. Mater. Sci. Eng. A 2012, 536, 269–272. [CrossRef]
18. Gashti, S.; Fattah-Alhosseini, A.; Mazaheri, Y.; Keshavarz, M. Effects of grain size and dislocation density
on strain hardening behavior of ultrafine grained AA1050 processed by accumulative roll bonding.
J. Alloy Compd. 2016, 658, 854–861. [CrossRef]
19. Liu, K.; Chen, X.G. Evolution of Intermetallics, Dispersoids, and elevated temperature properties at various
Fe vontents in Al-Mn-Mg 3004 alloys. J. Metall. Mater. Trans. B 2016, 47, 3291–3300. [CrossRef]
20. Johnson, W.A. Reaction Kinetics in Processes of Nucleation and Growth. J. Trans. Am. Inst. Min. Metall. Eng.
1939, 135, 416–458.
21. Tu, Y.Y.; Qian, H.; Zhou, X.F.; Jiang, J.Q. Effect of Scandium on the Interaction of Concurrent Precipitation
and Recrystallization in Commercial AA3003 Aluminum Alloy. J. Metall. Mater. Trans. A 2014, 45, 1883–1891.
[CrossRef]
22. Cottrell, A. LXXXVI. A note on the Portevin-Le Chatelier effect. Lond. Edinb. Dublin Philos. Mag. J. Sci. 1953,
44, 829–832. [CrossRef]
23. Krishna, K.; Sekhar, K.C.; Tejas, R.; Krishna, N.N.; Sivaprasad, K.; Narayanasamy, R.; Venkateswarlu, K. Effect
of cryorolling on the mechanical properties of AA5083 alloy and the Portevin–Le Chatelier phenomenon.
Mater. Des. 2015, 67, 107–117. [CrossRef]
24. Huskins, E.; Cao, B.; Ramesh, K. Strengthening mechanisms in an Al–Mg alloy. Mater. Sci. Eng. A 2010, 527,
1292–1298. [CrossRef]
25. Klose, F.; Ziegenbein, A.; Weidenmüller, J.; Neuhäuser, H.; Hähner, P. Portevin–LeChatelier effect in strain
and stress controlled tensile tests. Comput. Mater. Sci. 2003, 26, 80–86. [CrossRef]
26. Li, S.L.; Wu, R.; Wei, Y.Z.; Wu, Z.F.; Zhou, X.J.; Zhang, Y.F.; Xu, G. Correlation of morphological anisotropy
and Lankford value of deep drawing sheets hot-rolled by compact strip production. Int. J. Miner. Met. Mater.
2012, 19, 934–938. [CrossRef]
27. Stoudt, M.R.; Ricker, R.E. The relationship between grain size and the surface roughening behavior of Al-Mg
alloys. Met. Mater. Trans. A 2002, 33, 2883–2889. [CrossRef]
28. Stoudt, M.R.; Hubbard, J.B.; Leigh, S.D. On the relationship between deformation-induced surface roughness
and plastic strain in AA5052—Is it really linear? Met. Mater. Trans. A 2011, 42, 2668–2679. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).