Вы находитесь на странице: 1из 12

Available online at www.sciencedirect.

com

Acta Materialia 57 (2009) 761772


www.elsevier.com/locate/actamat

Eect of the ZenerHollomon parameter on the microstructures


and mechanical properties of Cu subjected to plastic deformation
Y.S. Li, Y. Zhang, N.R. Tao, K. Lu *
Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, PR China
Received 18 July 2008; received in revised form 13 October 2008; accepted 13 October 2008
Available online 10 December 2008

Abstract
Pure Cu was deformed at dierent strain rates and temperatures, i.e. with dierent ZenerHollomon parameters (Z) ranging within
ln Z = 2266, to investigate the eect of Z on its microstructures and mechanical properties. It was found that deformation twinning
occurs when ln Z exceeds 30, and the number of twins increases at higher Z. The average twin/matrix lamellar thickness is independent
of Z, being around 50 nm. Deformation-induced grain renement is enhanced at higher Z, and the mean transverse grain size drops from
320 to 66 nm when ln Z increases from 22 to 66. The grain renement is dominated by dislocation activities in low-Z processes, while
deformation twinning plays a dominant role in high-Z deformation. An obvious increment in yield strength from 390 to 610 MPa
was found in deformed Cu with increasing Z, owing to the signicant grain renement as well as the strengthening from nanoscale deformation twins.
2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Copper; Plastic deformation; Nanostructures; ZenerHollomon parameter; Nanoscale deformation twinning

1. Introduction
Plastic deformation with high strains (or so-called severe
plastic deformation, SPD) is one of the most important
methods for producing ultrane-grained (UFG; grain size
d in the submicrometer regime) metallic materials, which
exhibit enhanced strength and hardness relative to their
coarse-grained (CG) counterparts [1]. The microstructures
and mechanical properties of metals and alloys subjected
to SPD, such as by equal channel angular pressing (ECAP)
[2] and high-pressure torsion (HPT) [3], have been much
investigated over the years. Grain renement in SPD metals is in fact not fundamentally dierent from that in samples deformed by conventional cold-rolling or drawing
[4,5], and their mechanical properties are very similar.
For example, both the SPD processes and the cold-rolling
can eventually rene grains of pure Cu down to a few

Corresponding author.
E-mail address: lu@imr.ac.cn (K. Lu).

hundred nanometers upon extending straining, with a


saturation yield strength of about 400 MPa [6]. Further
increasing the plastic strain or changing the deformation
approach does not seem to be eective in further reducing
the grain size or increasing the strength [6,7]. Therefore, the
further reduction in the grain size of metals by plastic
deformation, or producing bulk nanostructured (NS;
d < 100 nm) metals and alloys via plastic deformation,
has become a challenging subject.
Plastic strain, strain rate and deformation temperature
are known to be the three primary factors in deformation
processes. Most SPD processes are performed at an ordinary strain rate (<100 s1) and at ambient temperature
[13,57]. However, previous investigations have shown
that very ne grains (far below 100 nm) have been obtained
in various metals and alloys by means of special plastic
deformation processes, such as ball milling [8], surface
mechanical attrition treatment [9,10] and cryogenic rolling
[11,12]. In these processes, either high strain rates or low
temperatures (or both) are applied. Very recently, we demonstrated that bulk NS Cu specimens with much higher

1359-6454/$34.00 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actamat.2008.10.021

762

Y.S. Li et al. / Acta Materialia 57 (2009) 761772

strength can be produced by dynamic plastic deformation


(DPD, i.e. plastic deformation with high strain rates) at
cryogenic temperatures [1315], even though the applied
strain is only 2.0, which is much smaller than that used
in ECAP processes. These observations indicate that strain
rate and deformation temperature are two dominating factors in grain renement. Systematic investigations into the
eect of these two parameters on grain renement mechanism are vital for the development of synthesis techniques
for bulk NS materials.
In the present work, we will systematically study the
microstructure characteristics and mechanical properties
of bulk pure copper samples deformed at dierent strain
rates (103103 s1) and dierent deformation temperatures (77293 K). It is known that an increase in strain rate
during deformation is thought to have an equivalent eect
to that of a decrease in deformation temperature and vice
versa. The combined eects of strain rate (_e) and deformation temperature (T) are often represented by a single
parameter, the ZenerHollomon parameter (Z), as dened
by [16]
Z e_ expQ=RT

where R is the gas constant and Q is the related activation


energy for deformation. Hence, in this work, we use the
ZenerHollomon parameter in a large range (ln Z = 22
66) to demonstrate the combined eects of strain rate
and temperature on the grain renement process in Cu subjected to plastic deformation, as well as on the mechanical
properties of the deformed Cu samples.
2. Experimental

and 101 s1 (QSC2), respectively. The samples were


deformed at three dierent temperatures: liquid nitrogen
temperature (LNT, 77 K), dry-ice temperature (DIT,
195 K) and room temperature (RT, 293 K). The Cu cylinders were immersed into a liquid nitrogen bath or dry ice to
maintain the cryogenic temperature during deformation.
The deformation strain is dened as e = ln (L0/Lf),
where L0 and Lf are the initial and nal thickness of the
deformed sample, respectively. Multiple impacts were
applied to deform the Cu samples to an eventual strain
of e = 2 in the DPD treatments in order to maintain the
high strain rates, while a single compression was applied
to deform the sample to the given strain in the QSC
treatments.
2.3. Microstructural characterization
The microstructures of the deformed copper samples
were characterized by means of transmission electron
microscopy (TEM) on a JEOL 2010 high-resolution transmission electron microscope operated at 200 kV. Cross-sectional thin foils for TEM observations were prepared by
means of double-jet electrolytic polishing in an electrolyte
consisting of 25 vol.% alcohol, 25 vol.% phosphorus acid
and 50 vol.% deionized water at about 10 C.
X-ray diraction (XRD) analysis of the as-deformed Cu
samples was performed on a Rigaku DMAX/2400 X-ray
diractometer with the sample plane perpendicular to the
loading direction of deformation. Microstrain of the samples was determined by means of XRD peak broadening
analysis using the ScherrerWilson method [17]. Each
datum point was averaged from three independent
measurements.

2.1. Sample
2.4. Thermal analysis
Copper cylinders (9 mm in diameter and 12 mm in thickness) with a purity of 99.995 wt.% were used as raw materials for plastic deformation. Prior to deformation, the Cu
cylinders were annealed in vacuum at 973 K for 2 h to
obtain homogeneous CG structures. Grain sizes of the
as-annealed sample were within a range of 100250 lm
and annealing twins were found in some grains.
2.2. Plastic deformation treatments
All samples were plastically deformed by uniaxial compression, but with dierent strain rates (103103 s1) and
at dierent temperatures (77293 K). The high strain rate
deformation was conducted with a DPD facility [1315],
in which a strain rate of up to 103 s1 was applied to the
sample. Deformation at low strain rates was performed
on an MTS servo-hydraulic test machine. An oil-based
molybdenum disulde lubricant was used to reduce friction
between the specimen and the compression platens to
obtain the uniform deformation of compressed sample.
The copper samples were quasi-statically compressed with
nominal strain rates of 103 s1 (referred to as QSC1)

Dierential scanning calorimetry (DSC) analysis was


carried out using a Perkin-Elmer DSC-7 system in a temperature range of 50300 C at a constant heating rate of
20 C min1. Aluminum pans were used for both samples
and the reference, and high-purity Ar gas was used to protect the sample from oxidation. Every sample was reheated
at the same conditions for determining the baseline, and
the nal DSC curve is the dierence between the second
and the rst run.
2.5. Mechanical property tests
Uniaxial tensile tests were performed on an Instron 5848
MicroTester system with a strain rate of 6  103 s1 at
room temperature. A contactless MTS LX300 laser extensometer was used to measure the sample strain upon loading. The gauge section of the dogbone-shaped tensile
1-mm-thick specimens was 5 mm in length and 1 mm in
width. More than four tensile tests were performed on each
sample. Microhardness measurements were carried out on
a MVK-H300 Vickers hardness testing machine with a load

Y.S. Li et al. / Acta Materialia 57 (2009) 761772

of 50 g and a loading time of 10 s. The hardness value


obtained was averaged from at least 20 indentations for
each sample.
3. Results
In total, eight Cu samples were treated under uniaxial
compressive deformation with dierent combinations of
strain rate and temperature, as listed in Table 1, which
includes deformation variables such as the strain rate, the
nominal temperature (TN), the transient temperature rise
during deformation (DT, which is estimated as described
in Section 3.4) and the corresponding Z values determined
from Eq. (1) by taking the activation energy for grain
boundary (GB) self-diusion in copper (72.5 kJ mol1
[18]).
3.1. TEM characterization of deformed Cu samples with
dierent Z values
It is known from previous investigations that in copper
specimens deformed via either drawing [4] or ECAP [6]
e = 2.0 is a critical strain beyond which a saturation
hardness or strength is achieved corresponding to a stable
characteristic structure size. Hence, microstructural observations in the present work focused on the deformed Cu
samples with e = 2.0.
Cross-sectional TEM images showed that microstructures of RT-QSC2, RT-DPD and DIT-DPD samples are
characterized mainly by parallel lamellar boundaries (LBs)
with elongated grains or subgrains, as in Fig. 1a1a3.
High-density dislocations were found inside these
samples.
For sample RT-QSC2, statistical measurement of the
spacing between neighboring LBs from plenty of TEM
images indicated that it exhibits a normal logarithmic distribution in the range from 100 to 600 nm, with a mean
value of about 290 nm (Fig. 1b1). Misorientations across
the most LBs range from a few up to 20, similar to that
observed in the SPD and the cold-rolled Cu samples. The
average aspect ratio of grains or subgrains is about 510.
For sample RT-DPD, the mean LB thickness was determined to be about 233 nm (Fig. 1a2 and 1b2), smaller than

Table 1
Summary of deformation variables (strain rate, temperature and temperature rise during deformation) and the calculated ln Z values for dierent
Cu samples used in the present work.
Sample ID
RT-QSC1 Cu (ln Z = 22)
RT-QSC2 Cu (ln Z = 25)
RT-DPD Cu (ln Z = 33)
DIT-QSC1 Cu (ln Z = 35)
DIT-QSC2 Cu (ln Z = 38)
DIT-DPD Cu (ln Z = 43)
100K-DPD Cu (ln Z = 59)
LNT-DPD Cu (ln Z = 66)

e_ (s1)
3

10
101
103
103
101
103
103
103

TN (K)

DT (K)

In Z

293
293
293
195
195
195
100
77

020
050
3060
025
050
3565
5575
6080

21.022.9
23.127.5
30.934.8
32.737.8
33.342.4
40.444.8
56.063.2
61.770.5

763

that in sample RT-QSC2. Misorientations across the LBs


and the average aspect ratio of grains are consistent with
those in the sample RT-QSC2. Deformation twins were
identied in this sample and formed bundles in the
deformed structure, constituting about 5 vol.%. This is
understandable as deformation twinning is favored at high
strain rates, as previously addressed [19,20]. Twin/matrix
(T/M) lamellar thickness was determined from TEM
images, varying from 10 to 150 nm, with an average of
47 nm. The length of deformation twins is in the range of
a few hundred nanometers to several micrometers.
In the DIT-DPD Cu sample (Fig. 1a3 and 1b3), the
mean LB thickness is 165 nm, clearly smaller than those
in RT-QSC2 and RT-DPD samples. The volume fraction
of deformation twin bundles is about 10%, with a similar
average T/M lamellar thickness to the RT-DPD sample.
The microstructure of LNT-DPD Cu (Fig. 1a4) is composed of about 35 vol.% nanoscale deformation twin bundles embedded in nano-sized grains. Most nanograins are
elongated, with an aspect ratio of 23. The crystallographic
orientations of the nanograins are nearly random. The
transverse grain size ranges from a few up to 200 nm, with
a mean size of about 66 nm (Fig. 1b4). The average T/M
lamellar thickness determined from TEM images is also
about 46 nm.
Morphologies of deformation twins in dierent Cu samples are quite similar, as shown in Fig. 2a. Plenty of dislocations were identied at the TBs, typical of deformation
twins. Fig. 2bd shows the statistical T/M lamellar thickness distribution in these samples, respectively, which varies from several to about 150 nm in these three samples.
The mean T/M lamellar thickness is rather consistent,
being about 50 nm in each sample.
With systematical TEM characterization of other
deformed samples (RT-QSC1, DIT-QSC1, DIT-QSC2
and 100K-DPD [15]) with a strain e = 2.0, several microstructural characteristic features of the deformed Cu were
summarized as a function of ln Z, as in Fig. 3. One can
see an obvious decreasing transverse grain size (D) at
higher Z values. With ln Z increasing from 22 to 66, D
decreases from 320 to 66 nm. The datum points of conventional cold-drawn [4], cold-rolled [11], HPT [21] and ECAP
Cu samples [6,22] are also included for comparison, which
seem to be consistent with the present results.
No deformation twin was found in Cu samples for
ln Z < 30. As shown in Fig. 3b, for ln Z > 30, the volume
fraction of deformation twin bundles in the deformed Cu
samples increases monotonically with Z. It is worth noting
that the mean T/M lamellar thickness, k, being about
50 nm in each deformed Cu sample, is independent of Z
within the measurement regime (Fig. 3c). As indicated in
our previous work [15], k seems to be independent on deformation strain during the DPD process. In addition, similar k
values were reported in pure Cu samples deformed by shock
loading [19] or deformed at liquid helium temperature
(4.2 K) [23]. These results indicate that the T/M lamellar
thickness is insensitive to the deformation parameters,

764

Y.S. Li et al. / Acta Materialia 57 (2009) 761772

16

b1

12
8
4
0
16

b2

Volume fraction (%)

12
8
4
0
16

b3

12
8
4
0
16

b4
12
8
4
0
0

100

200

300

400

500

600

Transverse grain size (nm)


Fig. 1. Cross-sectional bright eld TEM images for the microstructure morphology and statistic grain size distributions for dierent deformed Cu samples
with a strain e = 2.0: (a1, b1) RT-QSC2 (ln Z = 25); (a2 and b2) RT-DPD (ln Z = 33); (a3 and b3) DIT-DPD (ln Z = 43); and (a4 and b4) LNT-DPD
(ln Z = 66).

Y.S. Li et al. / Acta Materialia 57 (2009) 761772

765

20

16

12
8

Volume fraction (%)

100
100 nm
nm

0
20

16
12
8
4
0
20

15
10
5
0
0

20

40

60

80

100

120

140

T/M lamellar thickness (nm)


Fig. 2. (a) A typical bright eld TEM image for deformation twins in the deformed Cu samples. Statistical distributions for the T/M lamellar thickness in
(b) RT-DPD (ln Z = 33), (c) DIT-DPD (ln Z = 43) and (d) LNT-DPD (ln Z = 66) Cu samples, respectively.

including strain, strain rate and deformation temperature.


More in-depth investigations on this phenomenon are
needed to understand the intrinsic twinning mechanism.
With the above-mentioned structure parameters, the GB
and twin boundary (TB) densities (area per unit volume)
were calculated for each deformed sample, as shown in
Fig. 4. With ln Z increasing from 22 to 66, the GB density
(SGB) increases by a factor of about 4, from 4.01  106 m1
(RT-QSC1) to 17.2  106 m1 (LNT-DPD), while the TB
density (STB) increases from 2.0  106 m1 (RT-DPD) to
7.6  106 m1 (LNT-DPD). Clearly, the total boundary
density (sum of SGB and STB) is increased by one order
of magnitude from 4.01  106 m1 (RT-QSC1) to
2.48  107 m1 (LNT-DPD).
3.2. Microstrain of deformed Cu samples with dierent Z
values
Microstrain determined from XRD peak broadening represents the concentration of lattice defects (mainly dislocations) in the samples. With increasing Z, the measured
microstrain increases monotonically from about 0.13%
(RT-QSC1 sample) to 0.25% (LNT-DPD sample), as shown
in Fig. 5. Previously published data show that the microstrain in Cu samples processed via dierent approaches varies across a wide range, such as from about 0.1% in ECAP Cu
[6], to 0.14% in HPT Cu [24] and 0.2% in ball-milled Cu [25].
Our measurements are close to these data. This indicates that
many more defects (dislocations) are introduced into the
samples plastic-deformed with higher Z parameters, which

is in agreement with the observed decreasing grain sizes at


higher Z values (Fig. 3a) and the stored energy measurements, as will be shown in next section.
3.3. Stored energies of deformed Cu samples with dierent Z
values
Upon heating the as-deformed Cu samples, an exothermic peak corresponding to recrystallization was detected in
the DSC curves (Fig. 6a). For RT-QSC2 Cu, a single broad
exothermic peak (onsets at about 205 C) is seen with a
stored energy of about 0.8 J g1. At higher Z values, the
exothermic peak shifts to lower temperatures while the heat
release is larger. For the LNT-DPD Cu, the onset of the
exothermic peak is at 120 C with a stored energy of about
1.7 J g1. As shown in Fig. 6b, with increasing Z, the characteristic temperatures (onset and peak) drop obviously
and the stored energy increases monotonically. The measured stored energy results are reasonably consistent with
the published data for deformed Cu samples [4,11,2628].
Some minor dierences can be attributed to the impurity
of the processed Cu and the range of measurements in
the DSC experiments.
3.4. Mechanical properties of deformed Cu with dierent Z
values
Variations in the microhardness of the deformed Cu
samples (RT-QSC2, RT-DPD and LNT-DPD) with
imposed strain are shown in Fig. 7. The early deformation

Y.S. Li et al. / Acta Materialia 57 (2009) 761772

Transverse grain size (nm)

766

Present work
ECAP Cu [6,22]

300

Drawing Cu [4]
Rolling Cu [11]
HPT Cu [21]

200

100

Twinbundle vol (%)

40

30
20

10

(nm)

75

50

25

0
10

20

30

40

50

60

70

80

lnZ
Fig. 3. Plots of several characteristic parameters vs. ln Z in Cu samples (e = 2.0) processed via dierent approaches: (a) average transverse grain size; (b)
volume fraction of deformation twin bundles; and (c) average T/M lamellar thickness (k, nm). Data from the literature on grain sizes are included in (a).

0.28

SGB
STB

20

Microstrain (%)

Density of GB/TB (106m-1)

25

15
10

0.20
0.16
0.12

5
0
10

0.24

Present work
ECAP Cu [6]
Rolling Cu [26]
HPT Cu [24]

20

30

40

50

60

70

80

lnZ

0.08
10

20

30

40

50

60

70

lnZ

Fig. 4. Variations in GB and TB densities with ln Z for the deformed Cu


samples with e = 2.0.

Fig. 5. Variation of measured microstrain from XRD analysis with ln Z


for the deformed Cu samples, in comparison with results of SPD Cu
samples from the literature.

stage of all the samples shows obvious strain hardening,


with a rapid increase in hardness with strain. The hardness
tends to level o at a saturation value during subsequent
deformation when the strain exceeds 2.0. The saturation
hardness values are about 1.6, 1.3 and 1.2 GPa for the
LNT-DPD, RT-QSC2 and RT-DPD samples, respectively.

Similar behavior has also been observed in metals and


alloys upon plastic deformation [11,29], which indicates a
dynamic equilibrium between generation and annihilation
of strain-induced defects (dislocations).
Uniaxial tensile tests showed an obvious dierence in
strength and ductility among the deformed samples with

Y.S. Li et al. / Acta Materialia 57 (2009) 761772


0.10
LNT-DPD Cu (lnZ=66)

Heat flow (W/g)

0.08
DIT-DPD Cu (lnZ=43)

0.06

RT-DPD Cu (lnZ=33)

0.04

RT-QSC2 Cu (lnZ=25)

0.02

0.00
100

150

200

250

300

Stored energy (J/g) Temperature (C)

Temperature (C)

240

Peak temperature
Onset temperature

200
160
120
2.0

Present work
Rolling Cu [11]
Drawing Cu [4]
ECAP Cu [28]

1.6
1.2
0.8
10

20

30

40

50

60

70

lnZ
Fig. 6. (a) Typical DSC traces for the LNT-DPD (ln Z = 66), DIT-DPD
(ln Z = 43), RT-DPD (ln Z = 33) and RT-QSC2 (ln Z = 25) Cu samples.
(b) Variations in the onset temperature, peak temperature and stored
energy determined from the DSC experiments with ln Z for the deformed
Cu samples with e = 2.0. Previously published data on stored energy are
included in (b).

767

very little uniform elongation (about 1%), analogous to the


Cu samples deformed via other approaches [4,6,26]. This is
due to an insucient ability to strain harden in the
deformed metals, causing the onset of early necking [30].
However, no obvious change in elongation-to-failure was
noticed with dierent strains.
The tensile stressstrain curves clearly demonstrate that
both yield strength and ultimate tensile strength of the
deformed samples increase with Z value at an equivalent
amount of strain (either at e = 1.0 or e = 2.0). The yield
strength increases from about 390 6 MPa (RT-QSC1)
to about 610 14 MPa (LNT-DPD). The elongation-tofailure drops with increasing Z from about 17% (RTQSC1) to 8% (LNT-DPD). The strength and ductility of
the RT-QSC1 sample are consistent with those of the
ECAP and cold-rolled Cu samples reported in the literature [6,26]. As shown in Fig. 9, both hardness and yield
strength increase almost linearly with ln Z. The Z dependence of yield strength determined in the present work
agrees well with the literature data from Cu samples
deformed via dierent techniques [4,6,12].
The saturation stress of deformed samples with
increased strains can be described by Voce equation [31]


rs  r
e
2
exp 
rs  r0
ec
where r0 is the initial yield strength, rs is the saturation
strength and ec is a characteristic strain that depends upon
the material and the deformation variables. In terms of tensile results, one can calculate the corresponding values of rs
and ec. As shown in Fig. 10, the measured yield strength ts
the Voce equation very well for various samples. The calcu-

1.6

600

1.4

500

LNT-DPD Cu (lnZ=66)
DIT-DPD Cu (lnZ=43)
RT-DPD Cu (lnZ=33)

400

1.2
1.0
0.8

LNT-DPD Cu (lnZ = 66)


RT-DPD Cu (lnZ = 33)
RT-QSC2 Cu (lnZ = 25)

0.6
0.0

0.5

1.0

1.5

2.0

2.5

Strain
Fig. 7. Microhardness vs. strain for the Cu samples deformed via dierent
approaches and dierent Z parameters.

Engineering stress (MPa)

Hardness (GPa)

1.8

a
= 1.0

RT-QSC2 Cu (lnZ=25)

300

RT-QSC1 Cu (lnZ=22)

200
100
0
600

LNT-DPD Cu
100K-DPD Cu (lnZ=59)
DIT-DPD Cu

500

b
= 2.0

RT-DPD Cu

400
300
RT-QSC2 Cu

200

dierent Z values. As in Fig. 8, much enhanced strength


and decreased plasticity (elongation-to-failure) were found
in each deformed sample at strains of e = 1.0 and 2.0, compared with the as-annealed CG Cu. Both yield strength and
ultimate tensile strength of the deformed samples with
e = 2.0 are higher than those with e = 1.0, consistent with
the hardness measurements. The deformed samples exhibit

CG Cu

100
0
0

10

15

20

Engineering strain (%)


Fig. 8. Typical engineering stressstrain tensile curves for the Cu samples
subjected to plastic deformation via dierent approaches with strains of
(a) e = 1.0 and (b) e = 2.0.

768

Y.S. Li et al. / Acta Materialia 57 (2009) 761772


1800

Hardness (MPa)

1600

= 2.0
= 1.0

1400
1200
1000
800

Yield strength (MPa)

= 2.0

700
600
500

Present work
ECAP Cu [6]
Rolling Cu [12]
Drawing Cu[4]
= 1.0
Present work
ECAP Cu [6]

400

Elongation-to-failure (%)

300
20

c
15
10
= 2.0

5
0

10

Present work
ECAP Cu [6]
= 1.0
Present work
ECAP Cu [6]
20

30

40

50

60

70

80

lnZ
Fig. 9. Plots of several mechanical properties vs. ln Z for the deformed Cu samples with e = 1.0 (open symbols) and e = 2.0 (solid symbols), respectively:
(a) microhardness, (b) tensile yield strength and (c) elongation-to-failure. Previously published data on yield strength and elongation-to-failure are
included in (b) and (c), respectively.

T T N DT

where TN is the nominal deformation temperature and DT


is the strain-induced temperature rise. For adiabatic deformation (DPD processes can be regarded as adiabatic deformation due to the high strain rate and very short
deformation duration), the temperature rise of deformation
sample can be calculated according to [18]
Z e2
b
rde
4
DT
qC e1
where b = 0.9 (assuming 90% deformation work was converted to heat), q is the density of the sample and C is

600

Yield strength (MPa)

lated saturation strengths, rs, from the Voce equation are


399, 413, 507, 525 and 652 MPa for RT-QSC1, RTQSC2, RT-DPD, DIT-DPD and LNT-DPD sample,
respectively. The characteristic strain, ec, increases at higher Z values from 0.55 (RT-QSC1) to 0.76 (LNT-DPD).
With these data, the transient temperature rise induced
by plastic deformation can be estimated.
The actual sample temperature (T) can be expressed as

500
400
LNT-DPD Cu (lnZ=66)
Voce equ (c=0.76)

300

DIT-DPD Cu (lnZ=43)
Voce equ (c=0.64)
RT-DPD Cu (lnZ=33)
Voce equ (c=0.62)

200

RT-QSC2 Cu (lnZ=25)
Voce equ (c=0.56)

100
0
0.0

RT-QSC1Cu (lnZ=22)
Voce equ (c=0.55)

0.5

1.0

1.5

2.0

2.5

3.0

Strain
Fig. 10. Variation of measured yield strength and the tted yield strength
using the Voce equation with deformation strain for several deformed Cu
samples with dierent Z parameters.

the specic heat capacity. The mean strain increment 4e


is about 0.4 for each impact in the DPD process. For deformation at low strains, such as in QSC1, a minor tempera-

Y.S. Li et al. / Acta Materialia 57 (2009) 761772

ture rise is induced (<20 K). With an increasing strain rate,


higher temperature rises are produced. Table 1 lists the calculated temperature rises under dierent deformation conditions. In Ref. [18] the maximum temperature increment
was estimated to be 50 K in the ECAP process, which is
consistent with the present results. The highest temperature
rise was estimated to be about 6080 K for the LNT-DPD
samples.
4. Analysis and discussion
4.1. Z eect on grain renement mechanism
An obvious Z eect on the microstructure of deformed
Cu has been demonstrated: ner grains are formed in
deformation with higher Z values, which can be understood from the grain renement mechanism induced by
plastic deformation.
In plastic deformation of Cu at low strain rates and at
ambient temperature or above (i.e. with a low Z), renement of coarse grains is dominated by dislocation activities
by forming various dislocation congurations, including
dislocation cells, walls, geometry necessary boundaries
and incidental dislocation boundaries [20,32]. These dislocation cells gradually transform into subgrains separated
by boundaries of small misorientations (from dislocation
cell walls). With increasing strain, misorientations of the
subgrain boundaries increase, forming high-angle GBs so
that rened grains are randomly oriented. With this mechanism, the eventual grain size is determined by the dislocation cell size. Dislocation cell sizes (Ddc) are dependent
upon the shear stress applied (s) according to [33]
Ddc

KGb
KGb

s  s0
s

where K is a constant (normally taken as 10), G is the shear


modulus, b is Burgers vector and s0 is friction stress. As the
shear stress to initiate dislocation slip increases at higher Z,
Ddc and the eventual grain size become smaller with
increasing Z.
With this grain renement mechanism, it is noted that
the sizes of cells and grains decrease at a larger applied
shear stress. An extreme case is that when the applied shear
stress reaches its upper limit, i.e. the maximum shear stress
of the material (smax), Ddc may reach its minimum value,
i.e. the size limit of dislocation cells. From Eq. (5) and taking smax = G/30, we obtained a limit of Ddc of about 90 nm
for Cu. This means that grains may not be rened further
below 90 nm via the dislocation mechanism, which is
apparently inconsistent with our observations that average
grain sizes below 90 nm have been achieved in the LNTDPD and 100K-DPD samples.
In fact, a dierent grain renement mechanism may
operate at high-Z deformation. During high-Z plastic
deformation when dislocation manipulation and rearrangement are suppressed, the critical shear stress for deformation twinning is lower than that for slip. Hence, twinning

769

becomes the preferred deformation mechanism and deformation twin bundles are formed inside the original coarse
grains. With increasing deformation twinning, the original
three-dimensional coarse grains are rened into twodimensional lamellae by an increasing number of twin
boundaries. The T/M lamellae can be rened into the
nanometer regime.
Further plastic deformation induces transformation of
the nanometer-thick T/M lamellae into nano-sized grains
via two dierent mechanisms. As described in detail in
Ref. [15], nano-sized grains could be formed (i) via fragmentation of nanometer-thick T/M lamellae because of
TBdislocation interactions or, alternatively, (ii) via formation of shear bands within which nano-sized grains
are formed under a large degree of shear deformation.
With mechanism (i), the transverse grain size is very
close to the original T/M lamellae thickness (about
50 nm). In mechanism (ii), grain sizes are slightly larger
than the lamellae thickness (about 75 nm) due to grain
coarsening induced by high stress and the transient temperature rise within shear bands. Statistical measurements showed that in the LNT-DPD Cu sample [15]
about 30% in volume corresponds to nano-sized grains
from fragmentation of nanotwin bundles and about
15% to nanograins formed via shear banding. In contrast, only about 20% is from grain renement via dislocation activities. This means that the majority of
nanograins were rened from nanoscale deformation
twin bundles, from which the average grain size is much
smaller than that in low-Z samples rened via dislocation activities. As shown in Fig. 3b, with increasing Z
value, the tendency of deformation twinning is enlarged
with an increasing number of twin bundles. Hence, an
increasing number of nanograins are rened from nanoscale twin bundles, resulting in a decreased average grain
size.
The transition in grain renement mechanism with
increasing Z can also be identied from the variation of
GB energy, as described in Section 4.2.
Deformation twinning in face-centered cubic (fcc) metals is controlled by deformation conditions (such as strain,
strain rate and temperature) as well as the nature of materials, including grain size and stacking fault energy. A thorough understanding of the twinning behavior and the
subsequent grain renement process might be facilitated
by systematic investigations using molecular dynamic
(MD) simulations. The extremely high strain rates
(107 s1) and very low temperatures usually applied in
MD simulations favor deformation twinning in fcc metals.
In fact, previous MD simulations have demonstrated
deformation twinning in nano-sized grains of Al with a
high stacking fault energy [34]. MD simulations on nucleation and growth of deformation twins, as well as the interaction of twin boundaries with dislocations at the atomistic
level during high-Z deformation, are needed for identifying
the controlling parameters in the twinning-related structural renement mechanism.

770

Y.S. Li et al. / Acta Materialia 57 (2009) 761772

4.2. Z eect on grain boundary energy

0.65

1
 Ed  qg EGB
q

0.60

GB enegy (J/m2)

In terms of the measured stored energy from DSC


experiments, the GB energy in the deformed samples can
be estimated, which is a signature of the structure and thermodynamic states of GBs. In ultrane-grained or nanograined metals, the majority of stored energy released
during recrystallization is attributed to the disappearance
of a high density of GBs and lattice dislocations. The contributions from other defects, such as vacancies and the
associated elastic strain energy, are minor [35]. Consequently, for the deformed samples without deformation
twins (ln Z < 30, e.g. sample RT-QSC2) the stored energy
(E) can be divided into a dislocation term and a GB term
by

0.55
0.50
0.45
0.40
0.35

Present work
ECAP Cu [28]
Rolling Cu [11]

0.30
0.25
10

20

30

40

50

60

70

lnZ
Fig. 11. Variation of GB energy determined from DSC experiments with
ln Z in the deformed Cu samples. Reported results on GB energy in the
literature are also included for comparison.

where Ed is the energy per unit length of a dislocation and


qg is the dislocation density in the grain/cell interiors. The
GB part (EGB) is related to specic GB energy by
cGB = q  EGB/S, in which S is the GB area in the unit volume (S  3/D) and q is the density.
Taking an average value of Ed = 5  109 J m1 [35]
without making any distinction between edge/screw dislocations, complete/partial dislocations and the dislocation
density within the grain/cell interiors of about
2 1  1014 m2 in terms of detailed TEM observations
of the RT-QSC2 sample (which is consistent with data in
the literature [6,7]), we obtained that cGB is 0.54 J m2
for the RT-QSC2 sample. This is reasonably consistent
with the reported high-angle GB energy in Cu
(0.625 J m2 [36]) considering the mixture of characteristics of low- and high-angle GBs in the deformed samples.
Very similar GB energy values were gained in other samples with low Z. GB energies determined in Cu samples
that had undergone dierent plastic deformation modes
in the literature are in the range of 0.50.6 J m2, in agreement with our measurements, as shown in Fig. 11.
For the deformed samples with deformation twin bundles, the energy stored at deformation TBs (including the
energy of dislocations accumulated at TBs) should be taken
into account in calculating the GB energy, as described in
our previous work [26]. The total stored energy can be
described as E = Etwinf nt + Ematrix(1  f nt), where Etwin
1 1
c Ed  qnt [37], f nt is the volume fraction of the
q k TB
nanotwin bundles, cTB is the coherent TB energy (taken
as half of its stacking fault energy, i.e. 0.039 J m2), k is
the average T/M lamellar thickness, qnt is the dislocation
density in the nanoscale twins and Ematrix is the energy
stored in the matrix (submicro- or nano-sized grains) as
expressed in Eq. (6). In terms of TEM measurements of dislocation density in the nanoscale twins in the LNT-DPD
Cu sample, which are about 2 1  1015 m2, the stored
energy associated with deformation TBs was determined.
Then, the average GB energy can be calculated from the
stored energy in the nanograined matrix (Ematrix) in terms

of Eq. (6), being about 0.30 J m2 for the LNT-DPD sample. Similarly, the GB energy in sample 100K-DPD was
determined to be about 0.31 J m2. From the variation in
measured GB energy with ln Z as in Fig. 11, one may see
that GB energy is unchanged at about 0.55 J m2 when
ln Z increases from 22 to 40. As ln Z exceeds 40, an obvious
drop in GB energy appears from 0.55 to 0.3 J m2.
As described earlier, submicro-sized grains in the samples deformed at low Z are evolved from dislocation structures. However, for the high-Z deformation samples
(100K-DPD and LNT-DPD), most nano-sized grains were
derived from nanotwin bundles, either from fragmentation
of T/M lamellae or from shear banding of the nanotwin
bundles. Hence, most GBs in these samples were from
the original low-energy TBs. TBs possess rather low
energy, much smaller than that of high-angle GBs (for
Cu, cTB = 0.020.04 J m2 [36]). The low cGB in these samples (about 0.3 J m2) implies that a large fraction of GBs
are in energy states close to that of the original TBs. Transformation of the original nanotwin bundles into nano-sized
grains will unavoidably generate plenty of GBs that are in
similar low-energy congurations as the original TBs. This
argument is veried by detailed TEM observations on the
nature of GBs in the LNT-DPD samples. Micro-area electron diraction on individual GBs separating nano-sized
grains indicated that a considerable number of boundaries
exhibit twin orientation relationships or slightly deviated
twin relationships in the LNT-DPD samples. The GB
energy drop in Fig. 11 corresponds to an evident increment
in the number of deformation twins (Fig. 3c), verifying the
change in grain renement mechanism.
4.3. Strengthening of grain renement and nanoscale twins
As shown in Fig. 9, an increasing strength corresponds
to higher Z values. This phenomenon can be primarily
attributed to a decreasing grain size accompanied with an
increasing dislocation density with an increasing Z. For
the deformed samples with rened grains in which
strengthening is mainly caused by dislocationdislocation

Y.S. Li et al. / Acta Materialia 57 (2009) 761772

where r0 is the friction stress, a is a constant (taken as 1/3),


M is the Taylor factor and G is the shear modulus. The dislocation density is related to the stored energy as determined
from thermal analysis, as presented in Section 3.3 and Ref.
[26]. The calculated dislocation densities for the deformed
samples with dierent Z values are shown in Fig. 12,
which are consistent with the results in the literature
(1.11.5  1015 m2 in the SPD Cu samples [18]).
The yield strength values of deformed Cu samples with
dierent Z values can then be obtained in terms of Eq.
(7), as shown in Fig. 13. The calculated yield strength is
seen to agree reasonably with the measured data for the
samples with ln Z < 40. For the samples with ln Z > 40,
increasing dierences are seen between the calculated and
the measured data at higher Z. For the LNT-DPD samples, the dierence is as large as 65 MPa.
Such strength dierences originate from the presence of
nanoscale deformation twin bundles in the deformed samples. As described in previous sections, samples with
ln Z > 40 can be recognized as composites of nano-sized
grains with embedded nanotwin bundles, the strength of
which can be treated in terms of the rule-of-mixture:
nt
nt nt
rng
rDPD
y
y 1  f ry f

where f nt is the volume fraction of nanotwin bundles, and


nt
rng
y and ry are the yield strength for the nano-sized grains
and nanotwin bundles, respectively. Following the T/M
thickness dependence of yield strength for the nanoscale
twins in Cu as reported in Ref. [26], the yield strength of
the nanotwin bundles with k = 50 nm is determined, being
about 810 MPa. Therefore, the additional contributions of
nanotwin bundles to the measured yield strength values in
the LNT-DPD, 100K-DPD and DIT-DPD samples were
estimated to be about 90, 85 and 33 MPa, respectively.
These results agree reasonably well with the dierences

Dislocation density (1015 m-2)

3.6
3.2

Present work
ECAP Cu [28]
Rolling Cu [11]

2.8
2.4
2.0
1.6
1.2
10

20

30

40

50

60

70

lnZ
Fig. 12. Dislocation density determined from stored energy as a function
of ln Z for the deformed Cu samples. Previously published data on
measured dislocation density are included for comparison.

600

Yield strength (MPa)

interactions, yield strength (ry) can be correlated to the dislocation density (qd) following the Taylor equation [38]:
p
ry r0 aMGb qd
7

771

Measured
Calculated

550

500

450

400
10

20

30

40

50

60

70

lnZ
Fig. 13. Variations of the measured (solid squares) and the calculated
yield strength (open squares) with ln Z for the deformed Cu samples.

between the measured and the calculated yield strength in


terms of Eq. (7), as shown in Fig. 13.
Consequently, it may be concluded that the deformed
Cu samples with high-Z values are strengthened not only
by the signicant grain renement but also by the presence
of a considerable number of nanoscale twin bundles. The
latter strengthening mechanism does not exist in the Cu
samples deformed at low Z values.
5. Conclusions
By using dierent plastic deformation approaches with
dierent strain rates and temperatures, the eects of the
ZenerHollomon parameter on microstructures and
mechanical properties in pure Cu were investigated. Experimental measurements showed:
(i) Deformation twinning occurs when ln Z is larger than
30, and the number of twins increases with increasing Z. The average T/M lamellar thickness is independent of Z, being around 50 nm in each sample.
(ii) The sizes of grains rened by plastic deformation
decrease at higher Z. With ln Z increasing from 22 to
66, the mean transverse grain size decreases from 320
to 66 nm. For the low-Z processes where grain renement is dominated by dislocation activities, higher
shear stresses for slip at higher Z values lead to smaller
sizes of dislocation cells and of eventual grains. In highZ deformation, deformation twinning plays a dominating role in grain renement. Fragmentation and
shear banding of the nanoscale deformation twin bundles result in even smaller grains than those from dislocation cells. The transition in the grain renement
mechanism was demonstrated by an obvious drop in
GB energy when ln Z exceeds 40.
(iii) With ln Z increasing from 22 to 66, tensile yield
strength of the deformed Cu increases from 390 to
610 MPa. The strength increment can be attributed
to signicant grain renement as well as the strengthening from nanoscale deformation twins.

772

Y.S. Li et al. / Acta Materialia 57 (2009) 761772

Acknowledgements
Financial support from the National Natural Science
Foundation of China (Grants Nos. 50431010, 50671106
and 50071061), the Shenyang Science & Technology Project (Grant No. 1071107-1-00) and the Ministry of Science
and Technology of China (Grants No. 2005CB623604) is
acknowledged.
References
[1] Valiev RZ, Islamgaliev RK, Alexandrov IV. Prog Mater Sci
2000;45:103.
[2] Iwahashi Y, Horita Z, Nemoto M, Langdon TG. Acta Mater
1998;46:3317.
[3] Hebesberger T, Stuwe HP, Vorhauer A, Wetscher F, Pippan R. Acta
Mater 2005;53:393.
[4] Han K, Walsh RP, Ishmaku A, Toplosky V, Brandao L, Embury JD.
Philos Mag 2004;84:3705.
[5] Hughes DA, Hansen N. Acta Mater 2000;48:2985.
[6] Torre FD, Lapovok R, Sandlin J, Thomson PF, Davies CHJ. Acta
Mater 2004;52:4819.
[7] Mishin OV, Gottstein G. Philos Mag A 1998;78:373.
[8] Koch CC. Nanostruct Mater 1993;2:109.
[9] Lu K, Lu J. J Mater Sci Technol 1999;15:193.
[10] Tao NR, Wang ZB, Tong WP, Sui ML, Lu J, Lu K. Acta Mater
2002;50:4603.
[11] Wang YM, Jiao T, Ma E. Mater Trans 2003;44:1926.
[12] Wang YM, Chen MW, Sheng HW, Ma E. J Mater Res 2002;17:
3004.
[13] Zhao WS, Tao NR, Guo JY, Lu QH, Lu K. Scripta Mater
2005;53:745.

[14]
[15]
[16]
[17]

Tao NR, Lu K. J Mater Sci Technol 2007;23:771.


Li YS, Tao NR, Lu K. Acta Mater 2008;56:230.
Zener C, Hollomon JH. J Appl Phys 1944;15:22.
Klug HP, Alexander LE. In: Diraction procedures for polycrystalline and amorphous materials. New York: Wiley; 1974.
[18] Mishra A, Kad BK, Gregori F, Meyers MA. Acta Mater 2007;55:13.
[19] Gray III GT, Follansbee PS, Frantz CE. Mater Sci Eng A 1989;111:9.
[20] Wang K, Tao NR, Liu G, Lu J, Lu K. Acta Mater 2006;54:5281.
[21] Jiang HG, Zhu YT, Butt DP, Alexandrov IV, Lowe TC. Mater Sci
Eng A 2000;290:128.
[22] Zhao YH, Bingert JF, Liao XZ, Cui BZ, Han K, Sergueeva AV, et al.
Adv Mater. doi:10.1002/adma.200601472.
[23] Niewczas M, Basinski ZS, Basinski SJ, Embury JD. Phil Mag A
2001;81:1121.
[24] Zhao YH, Horita Z, Langdon TG, Zhu YT. Mater Sci Eng A 2008;474:342.
[25] Huang JY, Wu YK, Ye HQ. Mater Sci Eng A 1995;199:165.
[26] Zhang Y, Tao NR, Lu K. Acta Mater 2008;56:2429.
[27] Knudsen T, Cao WQ, Godfrey A, Liu Q, Hansen N. Metall Mater
Trans A 2008;39A:430.
[28] Gubicza J, Nam NH, Balogh L, Hellmig RJ, Stolyarov VV, Estrin Y,
et al. J Alloys Compd 2004;378:248.
[29] Belyakov A, Sakai T, Miura H, Tsuzaki K. Phil Mag 2001;A81:2629.
[30] Meyers MA, Mishra A, Benson DJ. Prog Mater Sci 2006;51:427.
[31] Voce E. J Inst Met 1948;74:537.
[32] Hansen N. Mater Sci Technol 1990;6:1039.
[33] Courtney TH. Mechanical behavior of materials. 2nd ed. New
York: McGraw-Hill; 1990.
[34] Yamakov V, Wolf D, Phillpot SR, Mukherjee Ak, Gleiter H. Nat
Mater 2002;1:45.
[35] Rohatgi A, Vecchio KS, Gray GT. Acta Mater 2001;49:427.
[36] Howe JM. Interfaces in materials. New York: Wiley; 1997.
[37] Grace FI. J Appl Phys 1969;40:2649.
[38] Taylor GI. Proc R Soc 1934;145:362.

Вам также может понравиться