Materials Transactions, Vol. 50, No. 1 (2009) pp.

64 to 69
Special Issue on Severe Plastic Deformation for Production of Ultrane Structures and Unusual Mechanical Properties: Understanding Mechanisms
#2009 The Japan Institute of Metals

Temperature and Strain Rate Dependence of Flow Stress

in Severely Deformed Copper by Accumulative Roll Bonding
Takahiro Kunimine1; * , Naoki Takata2 , Nobuhiro Tsuji3 ,
Toshiyuki Fujii4 , Masaharu Kato1 and Susumu Onaka1
1

Department
Department
3
Department
4
Department
2

of
of
of
of

Materials Science and Engineering, Tokyo Institute of Technology, Yokohama 226-8502, Japan
Metallurgy and Ceramics Science, Tokyo Institute of Technology, Tokyo 152-8552, Japan
Adaptive Machine Systems, Osaka University, Suita 565-0871, Japan
Innovative and Engineered Materials, Tokyo Institute of Technology, Yokohama 226-8502, Japan

Tensile tests and strain rate jump tests have been carried out at low temperatures (77 K
room temperature (RT)) using pure Cu
specimens that were severely deformed by accumulative roll bonding (ARB). The dependence of the ow stress on the temperature and the strain
rate has been investigated and the strain rate sensitivity m and its variation caused by the change in the ARB cycle N are discussed. At RT, the
strain rate sensitivity for N  4 stays at about 0.005. However, for N  5, m increases with increasing N to become
0:018 when N 8. The
deformation mechanisms of the ARB processed Cu are discussed with the activation volume V  . The temperature dependence of V  and its
variation with increasing N are also discussed. [doi:10.2320/matertrans.MD200809]
(Received July 22, 2008; Accepted October 6, 2008; Published November 19, 2008)
Keywords: copper, ultra-ne grained materials, accumulative roll bonding, strain rate sensitivity, activation volume, deformation mechanism

1.

Introduction

The strain rate dependence of ow stress in ultra-ne

grained (UFG) and nanocrystalline (NC) materials has been
investigated extensively.14) The higher strain rate dependence of the ow stress of UFG and NC materials compared
with that of conventional materials has been reported at room
temperature (RT). However, little is known about the
temperature dependence of the ow stress in UFG materials
prepared by severe plastic deformation (SPD) processes.1)
Although models to describe characteristic mechanical
properties of UFG materials have been proposed,37) experimental studies are still necessary to discuss the deformation
mechanisms.
The accumulative roll bonding (ARB) is one of the SPD
processes.8) This process provides homogeneous nanostructures and large bulky metallic materials.9) In the present
study, tensile tests and strain rate jump tests at low temperatures are made for ARB processed pure Cu. The strain rate
sensitivity as well as activation volume obtained by the strain
rate jump tests are the indexes of the strain rate dependence
of the ow stress. The variation of these strain rate
parameters with the ARB cycles (i.e., the plastic strain
applied by ARB) together with the temperature dependence
of strength are discussed.
2.

Experimental

Polycrystalline copper of 99.99% purity was used as a

starting material. After annealing at 873 K for 2 h, copper
sheets with an initial grain size of about 100 mm were
subjected to the ARB process at room temperature (RT) with
lubrication.10) The ARB process was carried out up to N 8
where N is the number of ARB cycles. The equivalent strain
for N 8 is about 6.4 in terms of the logarithmic strain.11)
*Graduate

Student, Tokyo Institute of Technology

The ARB process initially induces microstructure with

elongated grains to the rolling direction. The microstructure
of copper at N 4 possesses the so-called lamellar boundary
structure.12,13) However, the microstructure for N  6 consists of a mixture of relatively equiaxed and slightly
elongated grains.12,13) At N 8, the spacing of high-angle
lamellar boundaries of elongated grains was about 370 nm in
the present study.
Using an electro-discharging machine, tensile specimens
with 10 mm gauge length and 3 mm gauge width were cut
from the ARB processed sheets with 1 mm in thickness.
Although the tensile axis is parallel to the ARB rolling
direction for the 6 and 8-cycle specimens, other ARB
processed specimens have the tensile axis parallel to the
transverse direction of the sheets. Tensile specimens cut from
the starting Cu polycrystal (the annealed Cu before ARB)
were also prepared. Tensile tests were made with an Instrontype testing machine at 77, 175 K (the melting point of
methanol), 273 and 298 K within the accuracy of 2 K.
Strain rate jump tests were conducted between the strain
rates of 8:3  105 s1 and 8:3  104 s1 . The strain rate
sensitivity m dened as
m

@ ln

1 @

@ ln "_

@ ln "_

is an index to evaluate the dependence of the ow stress
on

the strain rate "_. When a strain rate jump from "_1 to "_2 causes
the change in the ow stress from
1 to
2 during tensile tests,
the strain rate sensitivity m is given by
m

2 
1

:


ln "_
1
2 =2 ln"_2 ="_1

For comparison, tensile tests of 99.99% copper single

crystals were also performed. The single crystals were grown
by the Bridgman method. Tensile specimens with 10 mm
gauge length and 3 mm width were cut from the single
crystals. The tensile axis of the specimens was chosen so that

400

True Stress,

6 Cycles
300
200
8 Cycles

100
0
0

ARB 5 ~ 8 Cycles

High

Annealed Cu before ARB ~ 4 Cycles

Single Crystal
ARB Cycles

RT

65

The number of

4 Cycles

500

Strain Rate Sensitivity, m

/ MPa

Temperature and Strain Rate Dependence of Flow Stress in Severely Deformed Copper by Accumulative Roll Bonding

Low
RT

0.01 0.02 0.03 0.04 0.05 0.06 0.07

True Shear Stress,

True Plastic Strain,

/ MPa

Fig. 1 The true stress-true strain (
") curves of the specimens ARB
processed by 4, 6 and 8 cycles at RT. The steps on the
" curves are the
changes in the ow stress caused by the strain rate jump tests between
8:3  105 s1 and 8:3  104 s1 .

Fig. 2 The relationship between the strain rate sensitivity m and the true
shear stress  at RT for the conventional polycrystals (the annealed Cu
before ARB), the specimens ARB processed by various cycles and the
single crystal specimens.

the initial plastic deformation occurs by single slip. The strain

rate jump tests were carried out at 77, 175 and 298 K. The
change in the strain rate sensitivity m caused by work
hardening was measured during tensile deformation.

using the Schmid factor of 0.5 for (3) the single crystal
specimens.
In Fig. 2, the data points shown by circles are the results
for the single crystals, which indicate the variation of m due
to the increase in the true shear stress  (work hardening
during tensile deformation). Although  increases from about
10 to 100 MPa during the work hardening, the strain rate
sensitivity m does not change signicantly and stays between
0.003 and 0.005. These values are in agreement with
previously reported values for work hardened Cu single
crystals.17)
On the other hand, the data points shown by open and
closed diamonds in Fig. 2 are the results for the starting
polycrystals and ARB processed specimens. The m
relations for the starting polycrystals and ARB processed
specimens with N  4 are similar to that for the single
crystals. However, the higher-cycle ARB processes with N 
5 markedly increase the m value as shown by closed
diamonds and m reaches about 0.018 when N 8. The
tensile strength of the ARB processed specimens with N  4
was nearly the same as shown in Fig. 1, i.e., the large
increase in m and the almost constant  are found for the
higher-cycle ARB processed specimens.
The Cottrell-Stokes (CS) law states that the ratio (

=
) is
constant for various
when
ln "_ is xed.1719) Hence, from
eq. (2), the strain rate sensitivity m should be unchanged
during plastic deformation if the CS law is obeyed. The
validity of the CS law in f.c.c. single crystals during work
hardening has been discussed in previous studies.17,20,21) In
contrast, the ARB processed Cu at N 8 shows the value of
m reaching 0.018. Such high values of m about 0.02 have
been also reported for UFG Cu processed by ECAP.14) In
fact, it is known that nanocrystalline Cu and Ni show high
values of m such as 0.03622) and 0.019.6)

3.

Results

3.1 Stress-strain curves at room temperature

Figure 1 shows the true stress-true strain (
") curves of
specimens ARB processed by 4, 6 and 8 cycles at RT, which
were simply converted from the nominal stress-nominal
strain curves. Since uniform deformation is assumed to
obtain true stress and true strain, the curves after macroscopic
necking are not correct, strictly speaking. Although the
tensile strength of the starting polycrystal before ARB was
about 220 MPa, it increased with increasing the number of
ARB cycles and nearly saturated (450 MPa) even at N 4.
This saturation stress is slightly higher than 400 MPa reported
in a previous study.13) Another change in mechanical
properties caused by ARB is the increase in ductility after
4 cycles: nearly twice larger at N 8 compared with that at
N 4. A similar increase in ductility with increasing the
number of ARB cycles or ECAP passes has also been found
in previous studies.1316)
The changes in the ow stress

associated with the
sudden change of strain rate by a factor of ten amounted to
9 MPa at N 4 and 18 MPa at N 8. Generally speaking,


increased with increasing the number of ARB cycles.
3.2 Strain rate sensitivity at room temperature
The results of the strain rate jump tests for (1) the starting
polycrystals (the annealed Cu before ARB), (2) the ARB
processed specimens and (3) the single crystal specimens
are summarized in Fig. 2 as the relationship between the
strain rate sensitivity m and the ow stress that were
converted into the true shear stress . We have evaluated
the true shear stress  from the true tensile stress
by
using the Taylor factor MT 3:06 for (1) the starting
polycrystals and (2) the ARB processed specimens, or by

3.3

Stress-strain curves and strain rate sensitivity at low

temperatures
Figure 3 shows the temperature dependence of the true
stress-true strain (
") curves of ARB processed specimens

T. Kunimine et al.

Strain Rate Sensitivity, m

66

True Stress,

/ MPa

800

ARB 8 Cycles
77K

600

175K
400

273K

200

RT
0
0

0.05

0.1

ARB 8 Cycles
ARB 6 Cycles
ARB 4 Cycles
Annealed Cu before ARB
Single Crystal

0.15

Temperature, T / K

True Plastic Strain,

Strain Rate Sensitivity, m

Fig. 3 The true stress-true strain (
") curves of the 8-cycle ARB
specimens tested at 77, 175, 273 K and RT.

Change in m with increasing

the number of ARB cycles

RT

Single Crystal

273K

175K
RT

77K

175K
77K

Fig. 5 The relationship between m and T. The curves in this gure giving
the T dependence of m are obtained by considering m at certain strains, i.e.,
" 0:01 for the specimens ARB processed by 4, 6 and 8 cycles, " 0:15
for the annealed Cu before ARB and " 0:02 for the single crystals.

4.

Discussion

4.1 Activation volume at room temperature

The rate equation to describe thermally activated plastic
ow is given by
_ _0 expG =kT;
3
where _ is the shear strain rate, _0 the pre-exponential
constant, k the Boltzmann constant and G the activation
energy.24,25) The activation volume V  is given by24)
V  @G =@ kT@ ln _ =@;

True Shear Stress, / MPa

Fig. 4 The relationship between the strain rate sensitivity m and the true
shear stress  for the usual polycrystals (the annealed Cu before ARB), the
specimens ARB processed by various cycles and the single crystal
specimens tensile-tested at 77, 175, 273 K and RT.

by 8 cycles. With decreasing the test temperature T from RT

to 77 K, both strength and ductility increase. However, with
decreasing T,

caused by the strain rate jump test
decreases. As a result, the strain rate sensitivity m given by
eq. (2) decreases with decreasing T. Figure 4 shows the m
relations at various temperatures. As expected, the m curve
moves towards lower m and higher  with decreasing T, as
observed earlier in Cu single crystals.17) The higher-cycle
ARB processes with N  5 increase m at 273 and 175 K.
However, this eect of N on m is less pronounced at 77 K.
Figure 5 shows the relationships between m and T. The
data in this gure were obtained by plotting m measured
at a certain strain: " 0:01 for the ARB processed specimens, " 0:15 for the annealed Cu before ARB and "
0:02 for the single crystals. We nd that m decreases with
decreasing T for all the specimens. We can see that the
dependence of the m value on the number N of the ARB
cycles decreases as temperature lowers. A similar decrease
in m with decreasing temperature between 273 and 523 K
has also been found for aluminium processed by severe
plastic deformation.16,23)

where  is the applied shear stress. Then, from eqs. (1) and
(4), V  is written as a function of m:
V  MT

kT@ ln "_
kT
MT
:
@

m

Although the strain rate sensitivity m itself is a phenomenological parameter obtained by mechanical testing, m can be
related to the activation volume V  by eq. (5).
The activation volume V  is also a value connected with
the geometry of the activation process. Let us suppose that
many short-range obstacles to the motion of dislocations are
dispersed in materials and the activation process of plastic
deformation is the interaction between a dislocation and a
short-range obstacle. Then, the activation volume V  is
written as
V  Ld b;

Where L is the dislocation length which takes part in a local

activation event, d the average activation distance and b the
magnitude of the Burgers vector. The product Ld  is the
activation area swept by a dislocation during the activation
event.
Figure 6 shows the relationships between the true shear
stress  and the activation volume V  (calculated from
eq. (5)) for the specimens tested at RT. Since m was almost
constant during work hardening of the single crystals, it is
natural that the calculated V  of the single crystals is
inversely proportional to the true shear stress .26) Similar

Temperature and Strain Rate Dependence of Flow Stress in Severely Deformed Copper by Accumulative Roll Bonding

67

Activation Volume, V* / b3

ECAP 1 and 2 Passes

ARB Cycles

Low
The number of

Activation Volume, V* / b 3

RT

ECAP 4, 8, 12 and 16 Passes

Single Crystal
Annealed Cu before ARB ~ 4 Cycles

High

ARB 5 ~ 8 Cycles

True Shear Stress,

/ MPa

Fig. 6 The activation volume V  as a function of the true shear stress  at

RT. The previous results given by Dalla Torre et al.3,4) for ECAPed Cu are
also shown by the hatched regions in this gure.

Annealed Cu before ARB

ARB 4 Cycles
ARB 6 Cycles
ARB 8 Cycles
Vapor Deposited Cu (d = 500nm)

Temperature, T / K
V   relation is also found for the starting polycrystals and
the ARB processed specimens with N  4. As shown in the
log-log plot of Fig. 6, these data points lie on a single line
with a slope of 1.
To understand the V  / 1= relation, let us consider the
Bailey-Hirsch equation
p
 b ;
7
where  is the density of forest dislocations,  a constant
(0.30.6) and  the shear modulus.27) When the forest
dislocations are the thermal obstacles to the motion of
primary dislocations, the spacing between the obstacles L can
be considered as the dislocation length assigned to one
activation event. Therefore, we have
p
L  1= :
8
From eqs. (6), (7) and (8) and assuming d  b, we indeed
have the V  / 1= relation of the form
V   b3 =:

On the other hand, the obvious deviation from the V  /

1= relation with increasing number of ARB cycles must be
explained in a dierent manner. Dalla Torre et al. have also
reported such a decrease in V  for Cu severely processed by
ECAP.3,4) For comparison, their data are included in Fig. 6.
The anomalous decrease in V  appears to be common to Cu
specimens that were severely deformed by ARB or by ECAP.
This point will be discussed below.
4.2

The temperature dependence of the activation

volume
Using the results shown in Fig. 5 and eq. (5), we have
obtained the temperature T dependence of the activation
volume V  . Figure 7 shows the V  T relations for various
specimens. For the starting polycrystals and ARB processed
specimens at N 4, V  decreases with decreasing T, as in
the case of coarse-grained polycrystals.5) Generally speaking,
ow stress is lower and work hardening is less signicant at

Fig. 7 The activation volume V  as a function of the temperature T. The

previous results for vapor deposited Cu (d 500 nm) shown in the
literature5) are also indicated.

higher temperatures. This means that some of the obstacles to

the motion of dislocations are not eective and L is larger at
higher temperatures. Moreover, the activation distance d is
usually an increasing function of temperature (decreasing
function of stress). Since both L and d become larger as
temperature becomes higher, the experimental result that V 
is an increasing function of T can be understood naturally.
However, Fig. 7 shows that the T dependence of V  starts
to change with increasing the number of ARB cycles N. For
the ARB processed specimens at N 6, V  is almost
independent of T and, then, V  certainly increases with
decreasing T in the ARB processed specimens at N 8. The
same T dependence of V  also has been observed in vapor
deposited Cu (d 500 nm),5) as shown in Fig. 7. Judging
from the fact that Wang et al. reported the same temperature
dependence of V  for NC Ni (d 30 nm),6) it appears that
the change in the temperature dependence of V  with
decrease in the grain size is a universal phenomenon in
f.c.c. UFG metals. As shown in Fig. 7, the transition in the T
dependence of V  occurs in the present ARB processed Cu
at N 6. This critical number N 6 of the ARB cycles
coincides reasonably with the onset of the anomalous
increase of m in Fig. 2 and with the start of the anomalous
decrease in V  in Fig. 6.
To discuss this transition, it is necessary to consider
changes in microstructure with increasing the ARB cycle.
Grain renement occurs markedly by the initial ARB process
up to N 2.28) Afterward, the spacing of high-angle lamellar
boundaries slightly decreases with increasing N and approaches to a constant size, about 350 nm when N  4.28) In
addition to the formation of lamellar boundary structure, the
ARB process initially induces deformation microstructure
consisting of low-angle boundaries.28) However, the fraction

68

T. Kunimine et al.

be regarded as an attempt to nd the transition of the

deformation mechanism. Further experimental and theoretical studies will reveal whether or not the present discussion,
analysis and assumptions are indeed eective and reasonable.

(a) Low Temperature

Boundary Pinning Point

5.

LGB

Grain Boundary

(b) Room Temperature

LGB
Fig. 8 The schematic illustration showing the shapes of bowing-out
dislocations emitted from grain-boundary sources between pinning
obstacles (a) at low temperature, (b) at room temperature.

Conclusions

The most important ndings of this study on copper after

severe plastic deformation by ARB are:
(1) At RT, the values of the strain rate sensitivity m for the
single crystals and the polycrystals before ARB are about
0.005. The low-cycle ARB processes with N  4 do not
aect m. However, the high-cycle ARB processes with N  5
increase the values of m to be 0.018 when N 8.
(2) The strain rate sensitivity m monotonically decreases
with decreasing the temperature from room temperature.
Although the increase in m caused by the increase in the ARB
cycle is clear at RT when N  5, this eect of N on m is less
pronounced at 77 K.
(3) The ow stress and temperature dependence of the
activation volume V  suggests that the transition of deformation mechanisms in the ARB processed specimens occurs
at the ARB cycle N 6.
Acknowledgements

fHAGB of high-angle boundaries in the microstructure

increases when N  2. The N dependence of fHAGB shown
in previous studies for Cu is fHAGB  30% at N 2,28)
fHAGB  55% at N 628) and fHAGB  60% at N 8.29)
When N  6, both the ultra-ne grains and the large fraction
of high-angle boundaries are developed in the ARB processed Cu.28,29) An explanation of the transition in the T
dependence of V  is possible if we assume that the ratecontrolling deformation mechanism changes, as will be
discussed below, when such characteristic microstructures
develop in Cu.
Instead of the interaction between dislocations and the
intra-grain obstacles, Kato et al. have proposed a dislocation
bow-out model to explain mechanical properties of UFG
materials.7) In this model, a glide dislocation is considered to
be nucleated at grain-boundary sources and bow out between
two pinning points on the boundary. Plastic deformation
occurs when a dislocation takes an unstable shape under the
applied stress. The dislocation bow-out between pinning
points on a grain boundary is illustrated in Fig. 8. The
boundary pinning points are most probably short-range
obstacles for the detachment of dislocations from the
boundary, as pointed out by Wang et al.6) Then, a thermal
activation process can operate and a smaller dislocation bowout is sucient to thermally overcome the obstacles as
temperature becomes higher. As a result, as shown in Fig. 8,
the associated activation area (shaded area) becomes smaller
at a higher temperature. Therefore, when the thermally
activated escape of a dislocation from boundary pinning
points is rate-controlling, the anomalous V  dependence on T
can be explained at least qualitatively.
It is now generally believed that for plastic deformation of
UFG and NC materials, grain-boundary dislocation sources
becomes more important than in-grain sources as grain size
becomes smaller.30,31) In this respect, the present study may

This research was supported by a Grant-in-Aid for

Scientic Research on Priority Areas Giant straining
process for advanced materials containing ultra-high density
lattice defects (18062002) by the Ministry of Education,
Culture, Sports, Science and Technology of Japan.
REFERENCES
1) Y. M. Wang and E. Ma: Appl. Phys. Lett. 83 (2003) 31653167.
2) Q. Wei, S. Cheng, K. T. Ramesh and E. Ma: Mater. Sci. Eng. A 381
(2004) 7179.
3) F. H. Dalla Torre, E. V. Pereloma and C. H. J. Davies: Scr. Mater. 51
(2004) 367371.
4) F. H. Dalla Torre, E. V. Pereloma and C. H. J. Davies: Acta Mater. 54
(2006) 11351146.
5) H. Conrad: Mater. Sci. Eng. A 341 (2003) 216228.
6) Y. M. Wang, A. V. Hamza and E. Ma: Acta Mater. 54 (2006) 2715
2726.
7) M. Kato, T. Fujii and S. Onaka: Mater. Trans. 49 (2008) 12781283.
8) Y. Saito, N. Tsuji, H. Utsunomiya, T. Sakai and H. G. Hong: Scr.
Mater. 39 (1998) 12211227.
9) B. L. Li, N. Tsuji and N. Kamikawa: Mater. Sci. Eng. A 423 (2006)
331342.
10) N. Kamikawa, T. Sakai and N. Tsuji: Acta Mater. 55 (2007) 5873
5888.
11) B. S. Altan, I. Miskioglu, G. Purcek, R. R. Mulyukov and R. Artan
(eds.): Severe Plastic Deformation, Towards Bulk Production of
Nanostructured Materials, (NOVA Science Publishers, New York,
2006) pp. 543564.
12) N. Takata, K. Yamada, K. Ikeda, F. Yoshida, H. Nakashima and N.
Tsuji: Mater. Trans. 48 (2007) 20432048.
13) N. Takata, S.-H. Lee, C.-Y. Lim, S.-S. Kim and N. Tsuji: J. Nanosci.
Nanotechno. 7 (2007) 15.
14) H.-W. Kim, S.-B. Kang, N. Tsuji and Y. Minamino: Acta Mater. 53
(2005) 17371749.
15) H. W. Hoppel, J. May and M. Goken: Adv. Eng. Mater. 6 (2004)
781784.
16) H. W. Hoppel, J. May, P. Eisenlohr and M. Goken: Z. Metallkd. 96
(2005) 566571.

Temperature and Strain Rate Dependence of Flow Stress in Severely Deformed Copper by Accumulative Roll Bonding
17)
18)
19)
20)
21)
22)

W. Bochniak: Acta Metall. Mater. 43 (1995) 225233.

A. H. Cottrell and R. J. Stokes: Proc. Roy. Soc. A 233 (1955) 1734.
M. A. Adams and A. H. Cottrell: Philos. Mag. 46 (1955) 11871193.
W. Bochniak: Acta Metall. Mater. 41 (1993) 31333140.
S. Saimoto and H. Sang: Acta Metall. 31 (1983) 18731881.
L. Lu, R. Schwaiger, Z. W. Shan, M. Dao, K. Lu and S. Suresh: Acta
Mater. 53 (2005) 21692179.
23) J. May, H. W. Hoppel and M. Goken: Scr. Mater. 53 (2005) 189194.
24) G. B. Gibbs: Phys. Stat. Sol. 10 (1965) 507512.
25) B. de Meester, C. Yin, M. Doner and H. Conrad: Rate processes in
plastic deformation of materials, (ASM, Metals Park, Ohio, 1975)

69

pp. 175226.
26) K. Saitou and T. Hikage: J. Japan Inst. Metals 49 (1985) 344351.
27) H. Mughrabi: Constitutive Equations in Plasticity, (M. I. T. Press,
Cambridge, Mass., 1975) pp. 199250.
28) B. L. Li, N. Shigeiri, N. Tsuji and Y. Minamino: Mater. Sci. Forum
503504 (2006) 615620.
29) N. Shigeiri: Master Thesis, Osaka University (2006).
30) S. Cheng, J. A. Spencer and W. W. Milligan: Acta Mater. 51 (2003)
45054518.
31) Y. J. Li, X. H. Zeng and W. Blum: Acta Mater. 52 (2004) 5009
5018.