Sunteți pe pagina 1din 21

Numerical simulations of necking during

tensile deformation of aluminum single crystals


F. Zhang
a,
*
, A.F. Bower
a
, R.K. Mishra
b
, K.P. Boyle
c
a
Division of Engineering, Brown University, Providence, RI 02912, USA
b
Materials and Processes Laboratory, GM R&D Center, Warren, MI 48090, USA
c
CANMET Materials Technology Laboratory, Ottawa, ON, Canada K1A 0G1
Received 16 October 2007; received in nal revised form 13 December 2007
Available online 31 December 2007
Abstract
Finite element simulations are used to study strain localization during uniaxial tensile straining of
a single crystal with properties representative of pure Al. The crystal is modeled using a constitutive
equation incorporating self- and latent-hardening. The simulations are used to investigate the inu-
ence of the initial orientation of the loading axis relative to the crystal, as well as the hardening and
strain rate sensitivity of the crystal on the strain to localization. We nd that (i) the specimen fails by
diuse necking for strain rate exponents m < 100, and a sharp neck for m > 100. (ii) The strain to
localization is a decreasing function of m for m < 100, and is relatively insensitive to m for
m > 100. (iii) The strain to localization is a minimum when the tensile axis is close to (but not exactly
parallel to) a high symmetry direction such as [100] or [111] and the variation of the strain to local-
ization with orientation is highly sensitive to the strain rate exponent and latent-hardening behavior
of the crystal. This behavior can be explained in terms of changes in the active slip systems as the
initial orientation of the crystal is varied.
2007 Elsevier Ltd. All rights reserved.
Keywords: Localization; B. Crystal plasticity; C. Finite element
0749-6419/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijplas.2007.12.006
*
Corresponding author. Tel.: +1 401 863 2674; fax: +1 401 863 9009.
E-mail address: Fan_Zhang_1@brown.edu (F. Zhang).
Available online at www.sciencedirect.com
International Journal of Plasticity 25 (2009) 4969
www.elsevier.com/locate/ijplas
1. Introduction
Strain localization limits the formability of many ductile materials in processes such as
punching, drawing, and hemming. This is a particular concern to the automotive industry,
where eorts to replace steel with lightweight Al and Mg alloys are impeded by the lower
formability of the replacement materials (Taub et al., 2007). Complex parts such as door
or lift-gate inner panels either need to be redesigned to accommodate the lower formability
of Al and Mg alloys at room temperature, or else must be manufactured using more
expensive and slower processes such as elevated temperature Quick Plastic Forming (Taub
et al., 2007). These demands are inspiring a renewed interest in the problem of strain local-
ization during plastic ow. A principal goal of this eort is to determine the roles of alloy
composition and microstructure in controlling ow localization, with a view to improving
formability through alloy and process design.
Forming limits depend on the part geometry, loading conditions (especially deforma-
tion rate, temperature, and the presence and type of lubrication), and the material prop-
erties. Nevertheless, tensile tests are frequently used as a rapid means to estimate the
ability of a material to resist shear localization. The roles of material properties, specimen
geometry, and loading conditions in controlling localization during a tensile test have been
extensively studied. Early work (Conside`re, 1888; Hill and Hutchinson, 1975; Hutchinson
and Neale, 1977) based on classical plasticity theory, focused on the inuence of strain
hardening and strain rate sensitivity of the specimen on the conditions necessary to initiate
localization. Although these models oversimplify the constitutive behavior of the solid,
their main conclusions remain broadly applicable even in more sophisticated analyses:
namely, (i) materials with a high rate of strain hardening compared with their ow
strength will resist strain localization; (ii) plastic anisotropy can lead to the formation
of vertices in the yield surface, which tend to make the material more prone to shear local-
ization; and (iii) high strain rate sensitivity leads to the formation of diuse necking and
also delays localization. More recent work has exploited the development of constitutive
laws for single crystals, which have been used to investigate the inuence of plastic anisot-
ropy resulting from texture formation in polycrystals on localization (Petryk, 1997; Shen,
1993; Tvergaard and Needleman, 1993; Inal et al., 2002a,b), as well the inuence of geo-
metric softening resulting from crystallographic rotation during tensile deformation of sin-
gle crystals (Asaro, 1979; Peirce et al., 1982, 1983; Balke and Estrin, 1994; Yang and Rey,
1995; Stu we and To th, 2003).
While these studies have provided insight into the mechanics of strain localization, we
still lack the detailed connection between localization and alloy composition or micro-
structure that is required for alloy design. This is partly because models of localization
are based on phenomenological constitutive laws, and it is not easy to relate constitutive
parameters in these models to the underlying processes that control deformation. Current
research is attempting to address this issue: studies range from eorts to establish the role
of alloy chemistry at the atomic scale on plastic ow (Curtin and Olmsted, 2006), to
improving the description of defects and microstructural features in constitutive models
(Ma et al., 2006). As part of this eort, Boyle and Mishra (2005) and Boyle (2005, submit-
ted for publication) have recently developed a hardening law for FCC crystals, with a par-
ticular focus on pure Al and Cu, which includes a more direct description of the
microscopic processes that control hardening. This model is based on the standard
description of rate dependent slip in single crystals, but extends prior models by incorpo-
50 F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969
rating a more detailed description of latent and self-hardening processes. In particular, it
accounts for the three stages of hardening, including the temperature dependence of Stage
III; and also contains a particularly detailed model of the evolution of latent-hardening
with plastic straining in the crystal. Although a large number of phenomenological param-
eters remain in the constitutive equations, which are currently calibrated using experimen-
tal data, many of these (such as the strengths of junctions between dislocations on the
various active slip systems) can in principle be computed from atomic scale studies,
thereby providing a more direct link between macroscopic ow and localization behavior
and microscopic deformation mechanisms.
In this paper, we briey review the main features of the rened model and the hardening
laws and show that an appropriately calibrated constitutive law captures the principal
trends observed in experimental measurements of the tensile behavior of pure Al single
crystals, including the inuence of crystal orientation on the hardening response and the
strain to localization. We use the model to explore the inuence of changes in the proper-
ties of the crystal, with a particular focus on its strain rate sensitivity, orientation, and
latent-hardening behavior.
2. Model
The problem to be solved is illustrated in Fig. 1. A tensile bar with length L and circular
cross-section is subjected to prescribed displacements so as to extend the specimen in uni-
axial tension at constant nominal strain rate. Specically, we enforce u
3
= 0 on the plane at
x
3
= 0 throughout straining and constrain u
1
= u
2
= 0 at the origin. Axial rotation of the
solid is constrained by enforcing u
2
= 0 at x
1
= R, x
3
= 0. The end of the specimen at
x
3
= L in the initial conguration is subjected to a constant velocity parallel to the x
3
direction. The loading is assumed to be quasi-static. Following Hutchinson and Neale
(1977) we introduce a small geometric imperfection in the specimen, which serves to initi-
ate localization in a controlled manner. This geometric imperfection takes the form of a
variation in the cross-sectional area of the bar along its length given by
Fig. 1. Schematic representation of a single-crystal tensile specimen, including the nite element mesh and
geometric parameters used in the simulation (symbols are dened in the text).
F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969 51
A = A
0
1 g cos
2px
3
L
_ _ _ _
(1)
where the A
0
is the mean cross-sectional area of the bar and g is the amplitude of the
variation in cross-sectional area.
The model uses the standard description of the kinematics of slip in a crystal. Let x
i
be
the position of a material particle in the undeformed crystal. The solid is subjected to a
displacement eld u
i
(x
k
), such that the point at x
i
moves to y
i
= x
i
+ u
i
after deformation,
as illustrated in Fig. 2. The deformation gradient and its Jacobian are given by
F
ij
= d
ij

ou
i
ox
j
; J = det(F) (2)
The velocity gradient L, stretch rate D and spin W are given by
L
ij
=
o_ u
i
oy
j
=
_
F
ik
F
1
kj
; D
ij
= (L
ij
L
ji
)=2; W
ij
= (L
ij
L
ji
)=2 (3)
The total deformation gradient is decomposed into elastic and plastic parts by assuming
that deformation takes place in two stages. The plastic strain is assumed to shear the lat-
tice, without stretching or rotating it. The elastic deformation rotates and stretches the lat-
tice. We think of these two events occurring in sequence, as illustrated in Fig. 2, with the
plastic deformation rst, and the stretch and rotation second, giving
F
ij
= F
e
ik
F
p
kj
(4)
The velocity gradient may then be decomposed into elastic and plastic parts by noting that
L
ij
=
_
F
ik
F
1
kj
=
_
F
e
ik
F
p
kl
F
e
ik
_
F
p
kl
_ _
F
p1
lm
F
e1
mj
_ _
=
_
F
e
ik
F
e1
kj
F
e
ik
_
F
p
kl
F
p1
lm
F
e1
mj
(5)
The velocity gradient contains two terms, one of which involves only measures of elastic
deformation, while the other contains measures of plastic deformation. We use this to
decompose L into elastic and plastic parts as
L
ij
= L
e
ij
L
p
ij
L
e
ij
=
_
F
e
ik
F
e1
kj
L
p
ij
= F
e
ik
_
F
p
kl
F
p1
lm
F
e1
mj
(6)
Fig. 2. Sketch showing the decomposition of plastic strain into elastic and plastic parts.
52 F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969
Plastic ow in the crystal occurs by shearing a set of N slip systems. An FCC crystal has
N = 12 slip systems, which are listed in Table 1 and labeled according to the Schmidt and
Boas convention. For computational purposes the slip systems are characterized by unit
vectors parallel to slip directions s
a
i
and slip plane normals m
a
i
in the undeformed solid.
The rate of shear on the ath system is denoted by _ c
a
. The velocity gradient due to this
shearing is
_
F
p
ik
F
p1
kj
=

N
a=1
_ c
a
s
a
i
m
a
j
(7)
The lattice shearing does not alter the slip plane normal or the slip direction. Conse-
quently, s
a
i
and m
a
i
are stretched and rotated only by the elastic part of the deformation
gradient, so that in the deformed conguration
s
+a
i
= F
e
ik
s
a
k
m
+a
i
= m
a
k
F
e1
ki
(8)
The plastic part of the velocity gradient can be expressed in terms of the shearing rates and
the deformed slip vectors as
L
p
ij
=

N
a=1
_ c
a
s
+a
i
m
+a
j
(9)
Finally, the elastic and plastic parts of the velocity gradient can be decomposed into sym-
metric and skew symmetric parts, representing stretching and spin, respectively, as
D
e
ij
= (L
e
ij
L
e
ji
)=2; W
e
ij
= (L
e
ij
L
e
ji
)=2
D
p
ij
= (L
p
ij
L
p
ji
)=2; W
p
ij
= (L
p
ij
L
p
ji
)=2
(10)
In particular, the plastic stretching and spin can be expressed in terms of the lattice
shearing as
D
p
ij
=

N
a=1
_ c
a
(s
+a
i
m
+a
j
s
+a
j
m
+a
i
)=2; W
p
ij
=

N
a=1
_ c
a
(s
+a
i
m
+a
j
s
+a
j
m
+a
i
)=2 (11)
Table 1
Slip systems for FCC single crystals
Slip plane Slip direction
(111) [0

11[ a
1
[10

1[ a
2
[

110[ a
3
(

11) [011] b
1
[

10

1[ b
2
[1

10[ b
3
(

111) [0

11[ c
1
[

10

1[ c
2
[110] c
3
(1

11) [011] d
1
[10

1[ d
2
[

10[ d
3
F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969 53
Internal forces in the crystal are characterized by the Cauchy (true) stress r
ij
, repre-
senting the force per unit deformed area in the solid. The Kircho, Nominal, and Material
stress tensors as follows:
s
ij
= Jr
ij
S
ij
= JF
1
ik
r
kj
; R
ij
= JF
1
ik
r
kl
F
1
jl
(12)
respectively. Plastic shearing on the slip systems is driven by the resolved shear stress, de-
ned as
s
a
= Jm
+a
i
r
ij
s
+a
j
(13)
The constitutive equations for the crystal must relate the elastic and plastic parts of the
deformation gradient (or strain rates derived from them) to the stresses acting on the crys-
tal. The elastic constitutive equation relates the elastic stretch rate to the Kirchho stress
as
s
ij
\
= C
e
ijkl
D
e
kl
(14)
where s
ij
\
is the Jaumann rate of Kirchho stress,
s
ij
\
=
ds
ij
dt
W
e
ik
s
kj
s
ik
W
e
kj
(15)
and
C
e
ijkl
= F
e
in
F
e
jm
C
nmpq
F
e
kp
F
e
lq
(16)
where C
ijkl
are the components of the elastic stiness tensor for the material with orienta-
tion in the undeformed conguration. For an FCC crystal the elastic modulus tensor can
be characterized by the tensile and shear moduli c
11
= C
1111
, c
12
= C
1122
and c
44
= C
1212
with the remaining terms determined by symmetry. The moduli c
11
,c
12
,c
44
are temperature
dependent, so we set
c
ij
= c
0
ij
c
T
ij
T (17)
where c
0
ij
is the modulus at T = 0, c
T
ij
is the rate of change of modulus with temperature,
and T is absolute temperature. The values used for all the material constants are listed
in Table 2. Elastic constants were t to the data collected by Simmons and Wang (1971).
The plastic constitutive equations specify the relationship between the stress on the
crystal and slip rates _ c
a
on each slip system. We adopt a power-law rate-dependent ow
for this purpose, with the form
_ c
a
= _ c
0
sign(s
a
)
[s
a
[
g
a
_ _
m
(18)
where s
a
= Jm
+a
i
r
ij
s
+a
j
is the resolved shear stress on the slip system, g
a
is its current
strength (which evolves with plastic straining as dened below) and _ c
0
, m are material
properties.
The constitutive law for describing the slip system hardening behavior of the bar is
based on a model for fcc metals developed by Boyle and Mishra (2005) and Boyle
(2005, submitted for publication). The details of the constitutive model and its calibration
are discussed in detail elsewhere (Boyle, submitted for publication). Here, we will merely
summarize the governing equations and give values for the relevant material parameters.
54 F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969
The hardening rule species the relationship between the slip system strengths g
a
and
the plastic strain. At time t = 0 each slip system has the same initial strength g
0
. Thereaf-
ter, the slip systems evolve as a result of the plastic shearing. The ath system may harden as
a result of shearing on the ath system itself or a coplanar system (self-hardening). It may
also harden as a result of shearing on a non-coplanar system (latent-hardening). The two
modes of hardening are distinguished by setting
_ g
a
=

N
b=1
~q
ab
h
b
[ _ c
b
[ (19)
where h
b
is a self-hardening modulus, and ~q
ab
are a set of latent-hardening moduli. The
latent-hardening moduli satisfy ~q
ab
= 1 for a = b. For a,=b, ~q
ab
evolve with plastic strain-
ing, as discussed below. It is convenient to discuss the constitutive equations for self- and
latent-hardening in turn.
2.1. Self-hardening
Strain hardening in crystals is conventionally divided loosely into three (or more) sep-
arate stages, which correspond to characteristic features of the uniaxial tensile stressstrain
curve, and are associated with dierent microscopic processes that contribute to strain
Table 2
List of values used for material parameters for Al
c
0
11
180395.51 MPa
c
T
11
46.71 MPa K
1
c
0
12
63255.16 MPa
c
T
12
9.52 MPa K
1
c
0
44
32156.62 MPa
c
T
44
13.85 MPa K
1
l
0
29720.70 MPa
l
T
14.492 MPa K
1
h
0
60.0 MPa
h
I
15.0 MPa
s
0
0.8 MPa
s
I
1.0 MPa
s
v0
115.5 MPa
A/l 0.2 10
19
m
3
_ c
1
10
7
s
1
c
0
0.005
c
II
0.0
v
N
m
=l 1.0
v
C
m
=l 0.0
v
G
m
=l 1.0
v
H
m
=l 0.75
v
S
m
=l 2.15
q
N
m
2.0
q
C
m
1.1
q
G
m
2.0
q
H
/
m
1.6
q
S
m
2.0
q
s
1.1
F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969 55
hardening (Kocks and Mecking, 2003). Stage I hardening (or easy glide) is observed in
crystals that are oriented to activate a single slip system. It is associated with a low, tem-
perature independent hardening rate. Stage II hardening initiates when a second slip sys-
tem is activated. It is associated with a high rate of hardening, with a roughly linear
relationship between stress and strain. The hardening rate in Stage II is also insensitive
to temperature. Stage III hardening occurs for large strains, and corresponds to a progres-
sively decreasing hardening rate. The rate of hardening in stage III is very sensitive to the
temperature and rate of deformation.
To describe this behavior, the self-hardening modulus h
b
for the three stages (I, II, and
III) of hardening is expressed as
h
b
= h
III
b
(h
I
b
h
II
b
) (20)
The stage I modulus is expressed as
h
I
b
= (h
0
h
I
)sech
2
[(h
0
h
I
)c
b
=(s
I
s
0
)[ h
I
(21)
where h
0
and h
I
are the initial and nal hardening rates during stage I, and
c
b
=
_
t
0
[ _ c
b
[dt
/
(22)
is the accumulated plastic slip on the bth system.
Stage II hardening is assumed to be activated by secondary slip with the stage II mod-
ulus given by
h
II
b
=

N
j=1
v
bj
f (n
a
c
j
=c
0
); f (n
a
c
j
=c
0
) =
0 c
j
< c
II
tanh
2
(n
a
c
j
=c
0
) c
j
> c
II
_
(23)
Here, c
II
is a material constant that represents the amount of secondary slip necessary for
the initial activation of stage II, n
a
is the number of active slip systems, and c
0
is a material
constant, dened so that c
0
/n
a
+ c
II
is the amount of slip after which the interaction
reaches ~60% of peak strength. In this expression, a slip system is considered active if
the fractional shearing rate is greater than a critical amount, i.e.
n
a
=

n
a=1
n
(a)
a
; n
(a)
a
=
1 for _ c
a
[ [=

_ c
b
[ [ P0:1
0 for _ c
a
[ [=

_ c
b
[ [ < 0:1
_
(24)
The coecients v
bj
in Eq. (23) are a set of material constants that control the magnitude of
the dislocation interaction strengthening between slip systems b and j. The value of v
bj
for
a pair of slip systems depends on the type of dislocation reaction that occurs between the
two systems. We consider the following types of interaction:
(1) Systems with parallel slip directions, with [s
b
s
j
[ = 1, form no junctions and have
v
bj
= v
N
;
(2) Coplanar systems with non-parallel slip directions ([s
b
s
j
[ < 1, [m
b
m
j
[ = 1) form
coplanar junctions and have v
bj
= v
C
;
(3) Systems with mutually perpendicular slip planes and slip directions (m
b
s
j
= 0 or
m
j
s
b
= 0) form a glissile lock, and have v
bj
= v
G
;
56 F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969
(4) The remaining systems, which must have s
j
m
b
,= 0 and s
b
m
j
,= 0 form a Hirth
lock, with v
bj
= v
H
if the slip directions combine to produce a vector of [100] type
(s
j
+ s
b
= [001]), and form a sessile lock, with v
bj
= v
S
, otherwise.
The stage III modulus is sensitive to temperature and strain rate and is given by
h
III
b
= 1 tanh
4
2
g
b
s
v
_ _
(25)
where, following Kocks and Mecking (2003), we take the saturation stress s
v
to be given by
s
v
l
=
s
v0
l
0
_ c
b
_ c
1
_ _kT
A
(26)
where l = l
0
+ l
T
T is a temperature dependent average shear modulus for the crystal, l
0
and s
v0
are the shear modulus and staturation stress at 0 K, _ c
1
is a representative strain
rate, k is Boltzmanns constant, T is the absolute temperature, and A is a material param-
eter which depends on the stacking fault energy of the crystal. In our computations, A is
taken to be a xed fraction of the shear modulus, as listed in Table 2.
2.2. Latent-hardening
The latent-hardening ratios ~q
ab
specify the rate of strain hardening due to shearing on
non-co-planar slip systems. Following Boyle (submitted for publication), the latent-hard-
ening coecient ~q
ab
are expressed in terms of the slip system strengths as
~q
ab
=
2
s
I
s
0
[s
I
(q
ab
m
q
s
) s
0
(q
ab
s
1)[sech
2
2
s
I
s
0
(g
b
s
0
)
_ _
q
s
(27)
where b = 1, . . . ,N, and no summation over b is implied. Among the internal variables, s
0
and s
I
are the initial and back-extrapolated critical yield stresses in shear for stage I, while
q
ab
m
and q
s
are maximum and saturation latent-hardening ratios. The values of q
ab
m
depend
on the types of dislocation reaction that occur between slip systems a and b, and are clas-
sied using the same convention that was used to assign values to v
ab
above. For example,
systems which form no junction have q
ab
m
= q
N
m
; systems which form coplanar junctions
have q
ab
m
= q
C
m
; systems which form glissile locks have q
ab
m
= q
G
m
; systems which form Hirth
locks have q
ab
m
= q
H
m
; and systems which form sessile locks have q
ab
m
= q
S
m
.
Parameter values for material constants representing pure Al are listed in Table 2. The
calibration of the model is described in more detail in Boyle (submitted for publication).
3. Finite element implementation
The boundary value problem was solved using the commercial nite element code
ABAQUS, augmented by a user-subroutine that implements the constitutive equations
discussed above. Fig. 1 illustrates a representative nite element mesh, which consists of
8-noded reduced integration brick elements. Mesh sensitivity was tested by repeating
selected simulations with four times the number of elements: our results indicate that local-
ization strains are changed by less than 2.5% with the rened mesh.
F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969 57
4. Results and discussion
4.1. Orientation dependence of limit strain
Fig. 3 shows a set of representative nominal stressnominal strain curves for uniaxial
tensile deformation of model single crystals with properties intended to represent pure
Al at a temperature of 273 K. Table 2 lists the values used for parameters in the constitu-
tive model. The parameter g characterizing the variation of the cross-sectional area of the
specimen (Eq. (1)) was g = 0.005 and the solid was deformed at a constant nominal strain
rate of 0.05 s
1
. Results are shown for several initial orientations of the tensile axis with
respect to the crystal, including loading along [123], [112], [100], and [111] directions.
The behavior of crystals loaded near a high symmetry orientation such as [100] and
[111] is extremely sensitive to small misorientations. Consequently, results are shown both
for perfect [100] and [111] orientated crystals, and also for specimens in which the loading
axis was given a small (0.05) random misorientation away from the high-symmetry
direction.
In our simulations, crystals that were loaded along or near the [100] and [111] direc-
tions failed by strain localization. For these cases, Fig. 3 shows the necking strain, which
is dened as the strain at the instant that the strain rate near the ends of the specimen
drops to zero.
The orientation dependence of the localization strain is simply a consequence of the
high initial rate of hardening that occurs when the crystal is loaded along a symmetry axis,
due to slip on multiple systems. This results in a higher ow stress at strains exceeding
10%; at the same time the hardening rate decreases with continued straining which makes
Fig. 3. Comparison of predicted nominal stressstrain curves for pure Al deformed in uniaxial tension with the
tensile axis initially parallel to [100], [110], [111] and [123] orientations. Necking strains are marked by a dot on
curves. FEM results are shown as solid lines while experimental results are from Hosford et al. (1960). The FEM
results for perfect [100] and [111] orientations are shown as dotted lines, while solid lines for these orientations
correspond to a deviation of 0.05 from the ideal direction.
58 F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969
the material more prone to localization, as predicted by rst approximation calculations
(Conside`re, 1888) over a century ago.
For comparison, Fig. 3 also shows the experimental stressstrain curves for Al single
crystals obtained by Hosford et al. (1960). Although our simulations do not give a perfect
match to the experimental stressstrain curves, the model clearly reproduces the principal
features of experimental data, including (i) the qualitative dependence of the ow stress of
the crystal on the direction of straining: the [111] orientation has the highest strength; fol-
lowed by [100], [112], and [123] oriented crystals as well as the crossover of the stress
strain behavior at high strains for [100] and [112] orientations; (ii) the model correctly
predicts that crystals with [100], [111], and [112] orientations localize, whereas that with
[123] orientation show no evidence of localization up to 50% nominal strain. In addition,
the localization strains predicted by the simulations are close to those observed experimen-
tally. The experiments and computations show detailed features that are not in perfect
agreement, however. For example, the model predicts a strong increase in hardening rate
at approximately 40% strain for the [123] oriented crystal: this is associated with the stage
Istage II transition at the onset of secondary slip in the crystal. Hosford et al. (1960)
found no evidence of a stage III transition in their experiments.
Our main objective in the remainder of this paper is to study in more detail the inuence
of geometric and material parameters in controlling the localization strain for these model
crystals.
4.2. Eects of specimen geometry
It is important to note that critical localization strains predicted by the simulations are
inuenced by specimen geometry, especially geometric imperfections. For example, Fig. 4
shows the critical localization strain as a function of the dimensionless defect parameter g,
which characterizes the initial geometric imperfection in the specimen as dened in Eq. (1).
The results are shown for [100] loading of an Al crystal at 273 K, with two values of the
Fig. 4. Necking strains (dened to be the critical nominal strain where the strain rate outside the localized region
of the specimen drops to zero; see text for details) as a function of the dimensionless geometric defect g dened in
Eq. (1). Results are shown for [100] (with deviation) loading of Al at 273 K.
F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969 59
slip rate exponent m = 20 and m = 150 in Eq. (18). With a low rate exponent m = 20, the
localization strain is evidently proportional to 1/log(g). For the high rate exponent
m = 150, the localization strain still decreases with the increasing value of g, but is not
described by a simple linear relationship with log(g). The sensitivity of localization strain
to specimen geometry makes it dicult to compare quantitatively the predicted localiza-
tion strains with experiment. However, the qualitative inuences of crystal orientation,
temperature and material properties on localization strain are independent of geometry.
In the remaining simulations reported in this paper, a small, xed geometric defect is
used to initiate localization in all simulations. This is not strictly necessary: the inhomoge-
neity associated with the nite element discretization and rounding errors during the com-
putation will cause even a geometrically perfect specimen to localize. In this case,
localization strains would, however, be sensitive to mesh design and machine precision,
so it is preferable to introduce a controlled defect. In actual experiments, localization
may be triggered by a variety of processes, but the qualitative variation of localization
strains with material and loading parameters is not sensitive to the details of the initiation
process.
4.3. Eects of strain rate sensitivity
Fig. 5 shows the results of simulations in which the strain rate exponent m in Eq. (18)
was systematically varied. Results are shown for [100] loading of Al at 273 K, with a geo-
metric defect g = 0.005. We observe two regimes of behavior: for m < 100 the specimen
fails by forming a relatively broad neck near the center of the specimen, and in this regime
the localization is highly sensitive to the strain rate exponent m. For m > 100, the specimen
forms a diuse neck during the early stages of deformation, but then localizes near the
edges of this necked region. In this regime both the stressstrain curves and the critical
strain to localization show a reduced sensitivity to m. Similar trends have been reported
by earlier studies, e.g. Hutchinson and Neale (1977), Peirce et al. (1983).
Fig. 5. Necking strains as a function of strain rate exponent m as dened in Eq. (18). Results are shown for [100]
(with deviation) loading of Al at 273 K, with a geometric defect g = 0.005.
60 F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969
The localization modes for perfectly oriented [100] crystals are dierent in these two
regimes. When m < 100, localization occurs by the formation of a single neck, as shown
in Fig. 6a. In contrast, for m > 100, we observe two necking regions, as shown in
Fig. 6b. However, the double neck is observed only for a crystal with perfect [100] orien-
tation: a very small deviation of the loading axis can change the deformation mode back to
the standard, single neck, as shown in Fig. 6c. In contrast, specimens with lower rate sen-
sitivity (m < 100) always fail by formation of a single neck, regardless of orientation. The
neck becomes progressively more diuse as m is reduced.
The results show that strain rate sensitivity clearly plays a critical role in governing
material behavior, and from the standpoint of alloy design, modifying alloy chemistry
and microstructure to increase strain rate sensitivity may be a more eective way to
improve formability than attempting to modify hardening behavior. Assuming that the
above trend for a single crystal holds true for polycrystalline alloys where m values
between 10 and 250 in AlMg alloys (Yao et al., 2000) is achievable, this result can be
taken as a guide to designing high ductility Al alloys.
4.4. Eects of initial orientation of the crystal
In this section, we study in more detail the inuence of the initial orientation of the sin-
gle crystal on its strain to localization. Throughout this section, the orientation of the ten-
sile axis with respect to the crystal will be displayed graphically on an inverse pole gure,
such as that shown in Fig. 7. A detailed description of the construction of an inverse pole
Fig. 6. Localization modes for Al loaded along [100] at 273 K: (a) m = 5; (b) m = 150; and (c) m = 150 but with a
small deviation from the perfect [100] orientation.
F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969 61
gure can be found in Hosford, 1993). Briey, the gure displays stereographic projections
of characteristic crystallographic directions in the convention adopted here, the projec-
tions of the [010] and [001] directions form a Cartesian coordinate system in the plane of
the gure, while the [100] direction points out of the plane. The orientation of the tensile
axis with respect to the crystal can then be specied by the projection of a unit vector par-
allel to the tensile axis, which corresponds to a. The full inverse pole gure for an fcc crys-
tal can be divided into 24 triangular regions, each of which (100), (011), and (111)
directions at the three vertices. Fig. 7 shows two such regions. Each point within these tri-
angles corresponds to a particular loading direction for example, the vertices indicate
loading parallel to [100], [110], or [111] directions. Due to the symmetry of the crystal,
each of the 24 triangular regions in the pole gure are crystallographically identical,
and it is only necessary to explore behavior with the tensile axis in a single representative
triangle. Accordingly, Fig. 7 shows a contour map of the strain to localization as a func-
tion of the initial orientation of the tensile axis relative to the [100], [110], or [111] direc-
tions. The results are for Al deformed at 273 K with strain rate sensitivity m = 150 and a
specimen defect parameter value of 0.005. The necking strain is smallest for initial orien-
tations near the two high symmetry orientations [111] and [100].
The inuence of initial orientation on necking strain is intimately connected to the evo-
lution of the crystals orientation with straining. Fig. 8 shows the progressive orientation
change of the crystal as a function of straining for crystals with a range of initial orienta-
tion plotted on an inverse-pole gure. The initial orientations correspond to points that lie
along lines marked a, b and c in Fig. 7. Each line in Fig. 8 is color-coded to indicate the
nominal strain variation with the orientation of the loading axis.
Fig. 8 shows that the crystal rotation may follow one of three general trends depending
on the initial orientation:
Fig. 7. A contour map showing the strain to localization as a function of orientation of the initial loading axis in
the standard triangle. The results are for Al, deformed at 273 K with strain rate sensitivity m = 150 and a
specimen defect parameter g = 0.005. Lines marked a, b and c refer to deformation paths referred to in Figs. 8
and 9. The simulations were terminated when the nominal strain reached 60%, so no localization occurred in the
red regions of the map.
62 F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969
(1) If the crystal is loaded with the tensile axis perfectly parallel to the [100] or [111]
axis, it does not rotate at all during deformation.
(2) If the initial loading axis is far from the [100] and [111] vertices, the crystal initially
deforms in single slip, and consequently the loading axis rotates along a great circle
towards the [101] orientation. Shortly after the tensile axis leaves the standard trian-
gle, a second slip system is activated: as a result, the crystal then begins to rotate so
as to bring the tensile axis towards the stable [112] orientation. If the crystal could
continue to deform indenitely without localization, the loading axis would eventu-
ally converge to the [112] direction. However, our simulations show that in all cases
localization sets in shortly after the onset of double slip. We note in passing that the
critical strain to activate double slip depends strongly on the latent-hardening behav-
ior of the crystal. This behavior is explored in more detail in the next section.
(3) If the tensile axis lies very close to, but not exactly on, the [100] or [111] loading
direction, our simulations predict extremely complex behavior, as illustrated in
Fig. 8. In this case, multiple slip systems are activated, and the loading axis shows
a tendency to spiral around the high-symmetry orientation. In most cases, the crystal
eventually settles to a state of double slip, but the choice of slip systems in this state is
highly sensitive to the initial misorientation of the crystal, as well as its latent-hard-
ening behavior.
Fig. 9 shows the inuence of initial orientation for crystals loaded near [100] and [111]
orientations in more detail. Fig. 9a and b corresponds to crystal orientation variations
along lines a and b in Fig. 7, respectively. Fig. 9a can be thought of as representing
the eects of misaligning the loading axis from the [100] direction by rotating the crystal
about the [0

611[ direction; similarly, Fig. 9b represents the inuence of misaligning


the loading axis from the [111] direction by rotating the crystal about its [

15

4[ direction.
Fig. 8. Inverse pole gures showing the evolution of the tensile axis for crystals with various initial orientations in
the standard triangle. The gure shows results for initial orientations along the lines marked a, b and c in
Fig. 7. The curves are color coded to indicate the variation of nominal strain along the orientation path. Most
initial orientations deform initially in single slip, then transition to double slip after the loading axis leaves the
standard triangle, and approach the stable orientation with loading axis parallel to the [112] direction. If the
loading axis is initially close to a high symmetry orientation, however, it may spiral around the high symmetry
direction.
F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969 63
Several features of these results are worth noting. Firstly, the strain to localization is a
monotonically increasing function of the misorientation angle for angles exceeding 4.
For smaller misorientations, however, the variation of localization strain with misorienta-
tion is complex. A very small misorientation away from the perfect [100] or [111] orien-
tation causes a rapid decrease in localization strain: the eect is particularly noticeable
near the [111] orientation in Fig. 9b, where the localization strain drops from 0.38 to
0.22 as a result of a misorientation of only 0.04. The drop is less pronounced for crystals
loaded near [100], but in this case, the necking strain is a uctuating function of the mis-
orientation. These uctuations are not numerical noise: they are associated with changes in
the history of rotation of the crystal (see Fig. 8). This behavior occurs because the active
slip systems are highly sensitive to small changes in orientation when the crystal is loaded
near a high symmetry orientation. The hardening behavior of the crystal and its rate sen-
sitivity play a critical role in controlling the active slip systems, as discussed in more detail
below.
4.5. Eects of rate sensitivity and latent-hardening on variation of localization strain
The behavior of crystals loaded near high symmetry orientations (e.g. with loading axes
close to [100] or [111] directions) is determined by the number of active slip systems and
the relative shear rates on each active system. These, in turn, are strongly sensitive to the
rate sensitivity and strain hardening behavior of the crystal.
The inuence of strain rate sensitivity and latent-hardening on the orientation depen-
dence of localization strain is illustrated in Fig. 10, which shows the variation of localiza-
tion strain with the misorientation of the tensile axis for a crystal loaded near the [111]
direction. (Fig. 10b shows the same data as (a), but re-plotted to show the behavior at
small misorientations more clearly.) Results are shown for crystals with isotropic harden-
ing and m = 150 as well as with latent-hardening with rate exponent m = 20 and m = 150.
To understand the implications of this comparison, it is helpful to recall that Al has
Fig. 9. Variation of necking strain with misorientation of the tensile axis from a high symmetry direction: (a)
results for tensile axis near [100], following the path marked a in Fig. 7 and (b) results for the tensile axis near
[111], along the path b in Fig. 7.
64 F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969
~q
ab
> 1 (the value of q evolves with straining), which implies that slip on the ath system
causes inactive systems (b ,= a) to harden more rapidly than the active system. In case
of isotropic hardening, with ~q
ab
= 1, all the systems harden at the same rate.
Reducing the strain rate exponent m retards localization for all orientations, as
expected from the results shown in Fig. 5. It also changes the sensitivity of the necking
strain to misorientation. The underlying reason is straightforward: with high strain rate
sensitivity (m = 150), the active slip systems change very rapidly as the loading axis is
rotated away from the perfect [111] orientation. With a lower strain rate exponent
(m = 20), the slip activity changes more gradually with misorientation. The inuence of
this behavior on the rotation of the crystal is shown in Fig. 11, which should be contrasted
with the results shown in Fig. 8 for a crystal with m = 150. In Fig. 8, the loading axis spi-
rals around the high symmetry orientation only if the tensile axis is very close to the [111]
Fig. 11. An inverse pole gure plot showing the rotation of the tensile axis for a crystal with various initial
orientations near the [111] direction. The results are for the model including latent-hardening with strain rate
sensitivity m = 20 and a specimen defect parameter g = 0.005.
Fig. 10. (a) The variation of strain to localization with misorientation for crystals loaded with the tensile axis near
the [111] direction with isotropic hardening and latent-hardening and (b) an enlarged view of the results for small
misorientations.
F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969 65
direction. In contrast, with m = 20, this behavior is observed for larger misorientations.
Consequently, with m = 20 we nd that the localization strain is minimized for a misori-
entation around 2.
Latent-hardening has little inuence on the strain to localization if the crystal is loaded
along a high symmetry orientation. In this case, latent-hardening merely increases the
overall strain hardening behavior of the crystal, and its eects can be understood using
the classical Considere analysis of necking (Conside`re, 1888). For a misoriented crystal,
however, latent-hardening has a much more signicant eect. For small misorientations,
the material with Boyles model of latent-hardening localizes earlier than an isotropically
hardened material as shown in Fig. 12. This behavior is again a consequence of the inu-
ence of hardening on the active slip systems: latent-hardening tends to prevent multiple
slip in crystals that are loaded close to a high symmetry orientation. In contrast, when
the initial loading is not close to a high symmetry orientation, latent-hardening increases
the overshoot prior to the onset of secondary slip, and consequently increases the strain to
localization.
While the above results are for aluminum single crystal, they provide important clues
about designing aluminum alloys to increase the localization strain and enhance ductility.
Reducing geometric imperfections has implications for control of second phase intermetal-
lic distribution. Manipulating orientation is related to controlling texture. Changing strain
rate exponent can be accomplished through solid solution strengthening and precipitation
hardening, which also alter latent-hardening. Assuming that the above trends for single
crystal also holds for aluminum alloys, a combination of these microstructural features
can lead to a high ductility material.
5. Summary
Finite element simulations have been used to study localization in single crystals with
properties representing Al under uniaxial tensile straining. A rened constitutive law
which includes a detailed description of latent and self-hardening in the crystal (Boyle
and Mishra, 2005; Boyle, 2005, submitted for publication) was used in the simulation.
Fig. 12. An inverse pole gure plot showing the rotation of the tensile axis for a crystal with various initial
orientations near the [111] direction. The results are for an isotropically hardening material, deformed at 273 K
with strain rate sensitivity m = 150 and a specimen defect parameter g = 0.005.
66 F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969
The simulations were used to investigate the inuence of specimen geometry, the orienta-
tion of the loading axis with respect to crystallographic directions, and the properties of
the material, in particular, the slip rate sensitivity exponent m and the latent-hardening
ratio on the strain-to-localization in uniaxial tension. Our simulations showed that:
(1) Two general forms of localization may occur, depending on the value of the strain
rate exponent m. For m < 100, failure occurs by the gradual development of a diuse
neck at the center of the specimen; for m > 100, a sharp neck forms, and if the load-
ing axis is exactly parallel to the [100] direction, two necks may form. In the diuse
necking regime (m < 100) the strain to localization decreases rapidly with increasing
m, while for values of strain rate exponent m > 100 the strain to localization is only
weakly sensitive to m.
(2) The strain to localization is sensitive to specimen geometry: for example, the strain to
localization varies inversely with log(g), where g is a dimensionless measure of the
local thinning at the center of the specimen, dened in Eq. (1).
(3) The inuence of the orientation of the loading axis with respect to the crystal on
strain to localization was explored in detail. Our simulations showed that crystals
localized most readily when the tensile axis is close to (but not exactly parallel to)
a high symmetry orientation such as [100] or [111]. Broadly, this behavior occurs
because a larger number of slip systems are active when the crystal is loaded near
the [100] or [111] direction, which causes a higher rate of hardening. For larger mis-
orientations, the crystal initially deforms by single slip, and rotates until a secondary
slip system is activated, whereupon the crystal tends to rotate so as to align the [112]
direction with the loading axis. In these cases we found that localization invariably
occurred shortly after secondary slip was activated.
(4) Strain rate sensitivity has a strong inuence on the orientation dependence of the
localization strain. Small values of the strain rate exponent m (e.g. m = 20) tend
to promote multiple slip, and the strain-to-localization has a minimum value when
the loading axis is misoriented by approximately 2 from a high symmetry orienta-
tion. For high values of rate sensitivity, the minimum value of strain-to-localization
occurs when the loading axis is misaligned by approximately 0.05 to the loading
axis.
(5) The inuence of latent-hardening on the variation of strain-to-localization with mis-
orientation was probed by contrasting the behavior of the model material with
parameters representing Al with an isotropically hardening solid. In the model Al
material, the hardening is such that slip on a single system tends to harden inactive
systems more rapidly than the active system; whereas in the isotropically hardening
material, all slip systems harden at the same rate. We found that if the loading axis is
perfectly aligned with a high symmetry orientation such as [100] or [111], latent-
hardening has almost no eect on strain to localization. If the loading axis is misa-
ligned with the high symmetry orientation by less than 2, the isotropically hardening
material has a higher strain to localization than the model material with properties
representing Al. In contrast, if the loading axis is misoriented by more then 2, the
trend is reversed. This behavior is a consequence of the inuence of latent-hardening
on the rotation of the crystal. When the loading axis is very near the high-symmetry
orientation, the isotropically hardening solid deforms by slip on more than two sys-
tems, while the material with properties representing Al tends to deform in double
F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969 67
slip, and localizes more readily. When the initial loading axis is misoriented by more
than 2, both materials initially deform in single slip, before eventually transitioning
to double slip just prior to localization. Under these conditions higher latent-hard-
ening tends to delay the onset of double slip, and consequently delays localization.
Acknowledgement
This work was supported by General Motors at Brown University as part of GM-
Brown Collaborative Research Laboratory (CRL).
References
Asaro, R.J., 1979. Geometrical eects in the inhomogeneous deformation of ductile single crystals. Acta Metall.
27 (3), 445453.
Balke, H., Estrin, Y., 1994. Micromechanical modeling of shear banding in single crystals. Int. J. Plasticity 10 (2),
133147.
Boyle, K.P., 2005. Latent hardening in copper and copper alloys. Materials Science Forum, vols. 495497. Trans
Tech Publications, Zurich-Uetikon, Switzerland, pp. 10431048.
Boyle, K.P., submitted for publication. A quantitative crystal plasticity hardening law. Modell. Simul. Mater. Sci.
Eng.
Boyle, K.P., Mishra, R.K., 2005. Latent hardening in crystal plasticity. In: Dislocations, Plasticity, Damage and
Metal Forming: Material Response and Multiscale Modeling. Neat Press, Fulton, pp. 616618.
Conside`re, M., 1888. Die Anwendung von Eisen und Stahl bei Konstruktionen. Gerold-Verlag, Wien.
Curtin, W.A., Olmsted, D.L., 2006. A predictive mechanism for dynamic strain ageing in aluminiummagnesium
alloys. Nat. Mater. 5 (11), 875880.
Hill, R., Hutchinson, J.W., 1975. Bifurcation phenomena in the plane tension test. J. Mech. Phys. Solids 23 (45),
239264.
Hosford, R.L., Fleischer, R.L., Backofen, W.A., 1960. Tensile deformation of aluminum single crystals at low
temperatures. Acta Metall. 8, 187199.
Hosford, W.F., 1993. The Mechanics of Crystals and Textured Polycrystals. Oxford University Press, Oxford.
Hutchinson, J.W., Neale, K.W., 1977. Inuence of strain-rate sensitivity on necking under uniaxial tension. Acta
Metall. 25 (8), 839846.
Inal, K., Wu, P.D., Neale, K.W., 2002a. Instability and localized deformation in polycrystalline solids under
plane-strain tension. Int. J. Solids Struct. 39 (4), 9831002.
Inal, K., Wu, P.D., Neale, K.W., 2002b. Finite element analysis of localization in fcc polycrystalline sheets under
plane stress tension. Int. J. Solids Struct. 39 (1314), 34693486.
Kocks, U.F., Mecking, H., 2003. Physics and phenomenology of strain hardening: the fcc case. Prog. Mater Sci.
48 (3), 171273.
Ma, A., Rosters, F., Raabe, D., 2006. On the consideration of interactions between dislocations and grain
boundaries in crystal plasticity nite element modeling theory, experiments and simulations. Acta Mater. 54
(8), 21812194.
Peirce, D., Asaro, R.J., Needleman, A., 1982. Analysis of nonuniform and localized deformation in ductile single
crystals. Acta Metall. 30 (6), 10871119.
Peirce, D., Asaro, R.J., Needleman, A., 1983. Material rate dependence and localized deformation in crystalline
solids. Acta Metall. 31 (12), 19511976.
Petryk, H., 1997. Plastic instability: criteria and computational approaches. Arch. Comput. Meth. Eng. 4 (2),
111151.
Shen, Y.L., 1993. An analysis of shear band formation in (100)(011) textured fcc metals. Phil. Mag. Lett. 67 (2),
7376.
Simmons, G., Wang, H., 1971. Single Crystal Elastic Constants and Calculated Aggregate Properties. MIT Press,
Cambridge, MA.
68 F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969
Stu we, H.P., To th, L.S., 2003. Plastic instability and Lu ders bands in the tensile test: the role of crystal
orientation. Mater. Sci. Eng. A 358 (12), 1725.
Taub, A.I., Krajewski, P.E., Luo, A.A., Owens, J.N., 2007. Yesterday, today and tomorrow: the evolution of
technology for materials processing over the last 50 years: The automotive example. JOM 59 (2), 4857.
Tvergaard, V., Needleman, A., 1993. Shear-band development in polycrystals. Proc. Roy. Soc. Lond. A 443
(1919), 547562.
Yang, S., Rey, C., 1995. Analysis of shear band development in a viscoplastic double slip, single crystal in tension.
J. Appl. Mech. 62 (4), 827833.
Yao, X.X., Zajac, S., Hutchinson, B., 2000. The strain-rate sensitivity of ow stress and work-hardening rate in a
hot deformed Al1.0Mg alloy. J. Mater. Sci. Lett. 19 (9), 743744.
F. Zhang et al. / International Journal of Plasticity 25 (2009) 4969 69

S-ar putea să vă placă și