Vous êtes sur la page 1sur 24

Available online at www.sciencedirect.

com

International Journal of Plasticity 24 (2008) 702–725


www.elsevier.com/locate/ijplas

A finite-deformation, gradient theory of


single-crystal plasticity with free energy
dependent on densities of geometrically necessary
dislocations
Morton E. Gurtin *
Department of Mathematical Sciences, Carnegie Mellon University, Pittsburgh, PA 15213, USA

Received 6 January 2007


Available online 8 August 2007

Abstract

This paper develops a finite-deformation, gradient theory of single crystal plasticity. The theory is
based on a system of microscopic force balances, one balance for each slip system, derived from the
principle of virtual power, and a mechanical version of the second law that includes, via the micro-
scopic forces, work performed during plastic flow. When combined with thermodynamically consis-
tent constitutive relations the microscopic force balances become flow rules for the individual slip
systems. Because these flow rules are in the form of partial differential equations requiring boundary
conditions, they are nonlocal. The chief new ingredient in the theory is a free energy dependent on
(geometrically necessary) edge and screw dislocation-densities as introduced in Gurtin [Gurtin, 2006.
The Burgers vector and the flow of screw and edge dislocations in finite-deformation plasticity. Jour-
nal of Mechanics and Physics of Solids 54, 1882].
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: A. Crystal plasticity; B. Burgers vector; C. Dislocation densities

*
Tel.: +1 412 681 5348; fax: +1 412 268 6380.
E-mail address: mg0c@andrew.cmu.edu

0749-6419/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijplas.2007.07.014
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 703

1. Introduction

As reviewed by Hutchinson (2000), a number of experimental results including those


from nano/micro-indentation, torsion of micron-dimensioned wires, and bending of
micron-dimensioned thin-films, all show that in the approximate size-range 500 nm–
50 lm, the strength of metallic components undergoing inhomogeneous plastic flow is
inherently size-dependent, with smaller being stronger.1
A generalization of classical single-crystal plasticity that accounts for size effects within
the context of finite deformations was proposed by Gurtin (2002 (G-2002)).2 The resulting
theory is based on:

(i) the decomposition F = FeFp of the deformation gradient F into elastic and plastic
distortions Fe and Fp;
(ii) microscopic stresses consistent with a system of microscopic force balances, one bal-
ance for each slip system, derived via the principal of virtual power;
(iii) a mechanical version of the second law that includes, via the microscopic stresses,
work performed during plastic flow.

Within the framework of (G-2002) the central ingredient leading to size effects is a con-
stitutive equation
b
W ¼ WðGÞ ð1:1Þ
giving the free energy W as a function of the Burgers tensor
G ¼ Fp Curl Fp ;
a field that characterizes the macroscopic Burger vector. This not withstanding, Nicola
et al. (2005) argue that a dependence of free energy on G may not be optimal: based on
a simple problem involving symmetric double-slip of a plane layer, they show that com-
parison with discrete dislocation theory is better achieved using a simple energy due to
Nicola and Van der Geissen and referred to by them as a pile-up energy.
Here working within the basic framework of (G-2002) we reformulate the constitutive
theory to account for continuous distributions of geometrically necessary dislocations as
introduced within the context of finite deformations in (G-2006); there it is shown that
temporal changes in G – as characterized by its plastically convected time derivative –
may be decomposed into temporal changes in distributions of (geometrically necessary)
screw and edge dislocations on the individual slip systems; cf. (2.15). For small deforma-
tions such a theory has already been developed by Gurtin et al. (2007). There, for each slip
system a (=1, 2, . . . , N), the dislocation densities for edge and screw dislocations have the
respective forms (Arsenlis and Parks, 1999)3
qa‘ ¼ sa  grad ca and qa ¼ la  grad ca ; ð1:2Þ

1
Cf., e.g., Stelmashenko et al. (1993), Ma and Clark (1995) and Fleck et al. (1994), Stolken and Evans (1998).
2
Henceforth referred to as (G-2002); Similarly, Gurtin (2006 (G-2006)) is referred to as (G-2006).Other studies
involving single-crystal gradient theories are: Acharya and Bassani (2000) and Acharya and Beaudoin (2000)
(which are local – additional boundary conditions are not required); Menzel and Steinman (2000), Cermelli and
Gurtin (2001), Arsenlis et al. (2004), Evers et al. (2004a,b), Han et al. (2005), Gurtin and Needleman (2002),
Bardella (2007), Ohno and Okumura (2007) and Okumura et al. (2006).
3
See also Fleck et al. (1994).
704 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

where ca denotes the slip on system a, while sa and la are constant (orthonormal) lattice
vectors4 with sa the slip direction, la = ma  sa, and ma the slip plane normal. Here we note
that qa‘ and qa have units of inverse length and may be positive or negative.
On the other hand, within the large-deformation framework of (G-2006) the notion of
slip is not well defined, but the notion of slip-rate is and the counterpart of (1.2) shows the
(geometrically necessary) dislocation densities to be internal-state variables that evolve
according to the differential equations
q_ a‘ ¼ sa  grad ma and q_ a ¼ la  grad ma ; ð1:3Þ
a a a a a
where m denotes the slip-rate on a, while s and l denote s and l pushed forward from
the lattice to the observed space (deformed configuration).
Some of the results of the finite theory mirror results of the small-deformation theory,
but the difference between the relations (1.2) and (1.3) renders the finite theory more dif-
ficult to construct: while (1.2) leads easily to a flow rule given as a partial differential equa-
tion for the slips ca, the flow rule in the finite theory – because it cannot be expressed in
terms of slips – is more complicated.
In comparing the free energy (1.1) to an energy
W ¼ Wð~ b qÞ ð1:4Þ
dependent on the list
q ¼ ðq1‘ ; q2‘ ; . . . ; qN‘ ; q1 ; q2 ; . . . ; qN Þ
~
of dislocation densities, it is worth mentioning that:

(i) while the dislocation densities are derived from the Burgers tensor G, a dependence
of W on G is not equivalent to a dependence of W on ~ q;
(ii) an energetic constitutive equation in the form (1.4) is automatically consistent with
the symmetry of the underlying crystal;
(iii) the pile-up energy of Nicola and Van der Geissen is a special case of (1.4) in which ~
q
is a single edge density (given as the derivative of slip in the direction of slip and
hence not meaningful when the deformation is finite).

2. Single-crystal kinematics

2.1. Basic kinematics

We denote the motion by v so that v ¼ v_ represents the velocity and


F ¼ rv; det F > 0;
represents the deformation gradient.5 Within the framework of finite deformations the
kinematics of single crystals takes as its starting point the Kröner–Lee decomposition (Krö-
ner, 1960; Lee, 1969)

4
We consistently use the term ‘‘lattice vector” for any vector which, when discussed within the context of finite
deformations, is associated with the lattice space (Fig. 1).
5
Notation: $, Div, and Curl denote the gradient, divergence, and curl with respect to the material point X in the
reference space (reference configuration); grad, div, and curl denote these operators with respect to the point
x = v(X,t) in the observed space (deformed configuration); a superposed dot denotes the material time-derivative;
Curl A is defined by (Curl A)ij = eirsoAjs/oXr, with eirs the alternating symbol; curl A is defined similarly. We write
Fe1 = (Fe)1, Fp> = (Fp)>, etc.
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 705

Fig. 1. Schematic of the Kröner–Lee decomposition showing the position of the lattice (space).

F ¼ Fe Fp ; det Fp ¼ 1; ð2:1Þ
in which F , the elastic distortion, characterizes stretch and rotation of the lattice, while Fp,
e

the plastic distortion, represents distortion of the lattice due to the formation of disloca-
tions. Reckoned pointwise as linear transformations: Fp maps vectors in the reference space
to lattice vectors, while Fe maps lattice vectors to vectors in the observed space (cf. Fig. 1).
A direct consequence of the Kröner–Lee decomposition is the following decomposition of
the velocity gradient, gradv, into elastic and plastic distortion-rate tensors Le and Lp:
)
grad v ¼ Le þ Fe Lp Fe1 ;
ð2:2Þ
Le ¼ F_ e Fe1 ; Lp ¼ F_ p Fp1 :
Single-crystal plasticity is based on a kinematical hypothesis that makes precise an ide-
alized view of the motion of dislocations in crystalline materials. In this view the motion of
dislocations in single crystals takes place on prescribed slip systems a = 1, 2, . . . , N in the
lattice, with each system a defined by a slip direction sa and a slip-plane normal ma, where sa
and ma are constant lattice vectors consistent with
sa  ma ¼ 0; jsa j ¼ jma j ¼ 1: ð2:3Þ
The presumption that plastic flow take place through slip then manifests itself in the
requirement that the distortion-rate tensor Lp be governed by slip rates6 ma on the individ-
ual slip systems via the relation
X
Lp ¼ ma s a  m a : ð2:4Þ
a

The tensor
Sa ¼ sa  ma ; ð2:5Þ

6
It is more common to denote the slip rates by c_ a . We refrain from using this notation: in the finite theory the
‘‘dot” has a precise meaning as a material time-derivative, and ma is not the material time-derivative of a physically
meaningful quantity.
706 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

is referred to as the Schmid tensor for system a; Sa is a mapping of lattice vectors to lattice
vectors and hence a lattice tensor. By (2.3) and (2.4), tr Lp = 0, consistent with (2.1)2. Here
and in what follows: lower case Greek superscripts a, b, . . . denote slip-system labels and as
such range over 1, 2, . . . , N; we do not use summation convention for Greek superscripts;
we use the shorthand
X X N
¼ :
a a¼1

In view of (2.2)3,
X
grad v ¼ Le þ ma ðFe Sa Fe1 Þ ð2:6Þ
a

with Fe Sa Fe1 the Schmid tensor Sa pushed forward from the lattice to the observed space.
Alternatively, if we let
sa ¼ Fe sa and  a ¼ Fe> ma
m ð2:7Þ
denote the slip direction and the slip-plane normal pushed forward to the observed space,
then
 a def
S ¼ Fe Sa Fe1 ¼ sa  m
 a; ð2:8Þ
represents the Schmid tensor pushed forward similarly, and (2.6) takes the form
X
grad v ¼ Le þ ma Sa : ð2:9Þ
a

Since, for any tensor A, tr(FeAFe1) = tr A, a consequence of (2.3)2, (2.5), and (2.8) is
that, for each a,
trSa ¼ trSa ¼ 0 and sa  m
 a ¼ 0: ð2:10Þ

2.2. The Burgers vector and the flow of geometrically necessary screw and edge dislocations

Unlike the deformation gradient F, the elastic and plastic distortions Fp and Fe are not
gradients of mappings, and the Burgers vector may be characterized by the closure failure
of referential circuits as mapped by Fp or of deformed circuits as mapped by Fe1. Based
on this, Cermelli and Gurtin (2001) derive, as a means of characterizing the Burgers vec-
tor, a tensor field G that may be described using the field Fp – or equivalently using the
field Fe – through the relations
G ¼ Fp Curl Fp ¼ ðdet Fe ÞFe1 curl Fe1 : ð2:11Þ
#
The field G is referred to as the Burgers tensor: given any oriented plane P in the lattice,
with n# its unit normal, G>n# represents the Burgers vector – as a vector in the lattice mea-
sured per unit lattice area – for infinitesimal circuits on P#. In terms more suggestive than
precise, G>n# represents the local Burgers vector per unit area for those dislocation lines
piercing P#.
The canonical dislocations for slip on the ath system are:

(i) edge dislocations with Burgers direction sa and line direction


l a ¼ m a  sa ; ð2:12Þ
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 707

(ii) screw dislocations with both Burgers direction and line direction equal to sa.7

Distributions of dislocations (if they exist) would therefore have the form
qa‘ la  sa and qa sa  sa ð2:13Þ
|fflfflfflfflffl{zfflfflfflfflffl} |fflfflfflfflffl{zfflfflfflfflffl}
edge distribution screw distribution

with qa‘ ðx; tÞ


and qa ðx; tÞ
corresponding densities (which may be positive or negative).
A central result of (G-2006) is that the plastically convected rate of G defined by

_  Lp G  GLp>
G¼G ð2:14Þ
may be decomposed into temporal changes in distributions of geometrically necessary
screw and edge dislocations on the individual slip systems via the relation (G-2006, Eq.
(5.6))
 X
G¼ ðq_ a‘ la  sa þ q_ a sa  sa Þ; ð2:15Þ
a

where the dislocation densities qa‘ and qa are internal-state variables with evolution gov-
erned by the differential equations8
q_ a‘ ¼ sa  grad ma ; q_ a ¼ la  grad ma ; ð2:16Þ

with
la ¼ Fe la and sa ¼ Fe sa ð2:17Þ

the lattice vectors la and sa pushed forward to the observed space. (G is the Oldroyd or
contravariant rate of G based on Lp; i.e., the rate following the flow of dislocations
through the lattice.)
Next, the glide directions for dislocations of a given type lie in the slip plane and are
orthogonal to the line direction; thus, by (2.12):

(i) sa = la  ma is the glide direction for edge dislocations on a;


(ii) la = sa  ma is the glide direction for screw dislocations on a.

Each of the density-rates (2.16) is therefore given, modulo sign, as a directional deriv-
ative of the slip rate in the glide direction. Thus, in accord with experience, a large value of
jq_ a‘ j ¼ jsa  grad ma j is indicative of a piling-up of edge dislocations, and similarly for screw
dislocations. (We could equally well have taken sa and la as the glide directions, but then
the signs of the dislocation fluxes and supplies need be changed to ensure that the balances
(2.21) remain valid.)
Since slip results from the flow of dislocations, the vector fields
def def
qa‘ ¼ ma sa ; qa ¼ ma la ð2:18Þ

7
Cf. (Kubin et al., 1992; Sun et al., 1998; Sun et al., 2000; Arsenlis and Parks, 1999).
8
Subject to initial conditions at, say, t = 0. The initial conditions qa‘ ðX ; 0Þ ¼ qa ðX ; 0Þ ¼ 0 (and Fp(X,0) = 1)
would seem appropriate to a body in a virgin state at t = 0.
708 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

represent respective fluxes of screw and edge dislocations. Using (2.17) these fluxes may be
pushed forward to the observed space; the results are
qa‘ ¼ masa ;
 qa ¼ mala
 ð2:19Þ

and introducing spatial screw and edge supplies defined by


a‘ ¼ ma divsa ;
r a ¼ ma divla
r ð2:20Þ

we have the spatial dislocation balances


q_ a‘ ¼ div 
qa‘ þ r
a‘ ; q_ a ¼ div 
qa þ r
a : ð2:21Þ

a‘ represents a volumetric supply of edge dislocations due to distortion of the


Here, e.g., r
lattice as characterized by divla , and similarly for the screw supplies. The local balances
(2.21) represent kinematical relations for the flow of dislocations through the body. As
such they are independent of constitutive relations.

2.3. Characterization of forest dislocations

We write Pa for the ath slip plane. The vector field

Fe> ma
a ¼
m ð2:22Þ
jFe> ma j

represents ma pushed forward as a unit normal from the lattice to the observed space. The
plane Pa in the observed space with unit normal m a represents Pa pushed forward to the
observed space. Thus, succinctly,

(à) Pa with unit normal m


 a represents the deformed ath slip plane.

It is commonly held that dislocations impinging transversely on a slip plane – tradition-


ally called forest dislocations – give rise to hardening (cf. Kuhlmann-Wilsdorf, 1989).
Within the macroscopic theory under consideration the normal component of the Burgers
vector corresponding to dislocations piercing the ath slip plane Pa is characterized by the
field
def
/a ¼ jma  Gma j: ð2:23Þ
a
Consistent with this, Cermelli and Gurtin (2001) show that / characterizes the distortion
of Pa, and, based on this, suggest that (2.23) might be useful as a constitutive quantity re-
lated to forest hardening.9 We refer to the field /a as the forest measure for system a.
Cermelli and Gurtin then show that if elastic strains are negligible, so that Fe is a rotation,
then /a has the simple form

/a ¼ m
 a  curl m
 a: ð2:24Þ

9
Cf. Acharya and Beaudoin (2000), who propose jG>majas appropriate measures of ‘‘forest dislocations”.
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 709

3. Virtual-power formulation of the standard and microscopic force balances

For a single crystal the basic ‘‘rate-like” descriptors for a single-crystal undergoing
finite deformations are the velocity v, the elastic distortion-rate Le, and the slip-rates
m1,m2, . . . , mN, and these fields are constrained by (2.9).

3.1. Internal and external expenditures of power

Let Pt denote an arbitrary spatial region convecting with the body and let n denote the
outward unit normal on oPt . The formulation of the principle of virtual power for a sin-
gle-crystal is based on a balance between the external power WðPt Þ expended on Pt and
the internal power IðPt Þ expended within Pt . Specifically, we allow for power expended
internally by a stress T power-conjugate to Le and, for each slip system a, a scalar internal
microscopic force pa power-conjugate to ma and a vector microscopic stress na power-con-
jugate to gradma. We therefore write the internal power in the form
Z XZ
IðPt Þ ¼ T : Le dv þ ðpa ma þ na  grad ma Þdv: ð3:1Þ
Pt a Pt

Regarding the external power, we allow for the standard power expenditures t(n)  v by
surface tractions and b  v by (inertial and noninertial) body forces. Further, we expect that
the gradient terms na  grad ma should give rise to traction-terms associated with the micro-
scopic stresses na. The following integral identity helps us to choose an appropriate form
for this traction:
Z Z Z
na  grad ma dv ¼ ðna  nÞma da  ma div na dv: ð3:2Þ
Pt oPt Pt

Guided by the term (na  n)ma in (3.2), we assume that power is expended externally by a
scalar microscopic traction Na(n) conjugate to ma; we therefore assume that the external
power has the form
Z Z XZ
WðPt Þ ¼ tðnÞ  v da þ b  v dv þ Na ðnÞma da: ð3:3Þ
oPt Pt a oPt

3.2. Principle of virtual power

Assume that at some arbitrarily chosen time, t0 say, the fields v, Fe, and m1, m2, . . . , mN are
known, and consider the velocity v, the elastic distortion-rate Le, and the slip rates
m1,m2, . . . , mN as virtual velocities to be specified independently in a manner consistent with
(2.6); that is, denoting the virtual fields by ~v, L e e , and ~m1 ; ~m2 ; . . . ; ~mN , we require that
X
grad~v ¼ L ee þ  a:
~ma S ð3:4Þ
a

We denote by Pð¼ Pt0 Þ an arbitrary subregion of the deformed body at the fixed time
t0. If we define a (generalized) virtual velocity to be a list
e e ; ~m1 ; ~m2 ; . . . ; ~mN Þ
V ¼ ð~v; L ð3:5Þ
710 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

and write IðP; VÞ and WðP; VÞ for the corresponding internal and external expendi-
tures of virtual power, then the principle of virtual power is the requirement that the virtual
power balance
Z Z X Z
tðnÞ  ~v da þ b  ~v dv þ a
Na ðnÞ~ma da
oP P oP
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
WðP;VÞ
Z X Z
¼ TL ~ e dv þ ðpa~ma þ na  grad~ma Þdv ð3:6Þ
a
P P
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
IðP;VÞ

be satisfied for any subregion P of the deformed body and any virtual velocity V.

3.3. Consequences of frame-indifference

We assume that under a change in frame the internal power IðP; VÞ is invariant, and
that the virtual fields transform in a manner identical to their nonvirtual counterparts.
Then under a change in frame\
e e ¼ Q L
L e e Q> þ X
with Q the frame-rotation and X the frame-spin. Further, for each a,
pa and ~ma are invariant ð3:7Þ
(because they are scalar fields), but because ‘‘grad” represents the gradient in the deformed
body, the transformation law for grad~ma has the form grad ~ma ¼ Q grad~ma : Next, by (3.7),
writing P and I ðP Þ for the region and the internal power in the new frame, we see that
Z XZ
 
I ðP Þ ¼  e
T :L þe
ðpa~ma þ na  grad ~ma Þdv; ð3:8Þ
P a P

where T* and n are the stresses T and na in the new frame. Since P is P transformed
a*

rigidly, we may replace the region of integration P in (3.8) by P. Thus the requirement
that I ðP Þ ¼ IðPÞ implies that
Z " X
# Z " X
#
e
T:L þe a a
n  grad~m dv ¼  e
T :L þ e a  a
n  grad ~m dv
P a P a
Z " X
#
¼ ~ e Q> þ XÞ þ
T : ðQL na  ðQ grad~ma Þ dv
P a

or, equivalently, since the region P is arbitrary, that


X X
T:Lee þ e e Q> þ XÞ þ
na  grad~ma ¼ T : ðQ L na  ðQ grad~ma Þ: ð3:9Þ
a a

Since the virtual slip rates ~ma are arbitrary we may set them to zero; thus
e e ¼ T : ðQ L
T:L e e Q> þ XÞ
~ e are arbi-
and since the orthogonal tensor Q, the skew tensor X, and the virtual field L
trary, we find that
T is symmetric and frame-indifferent: ð3:10Þ
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 711

Thus (3.9) becomes


X X X
na  grad~ma ¼ na  ðQ grad~ma Þ ¼ ðQ> na Þ  grad~ma : ð3:11Þ
a a a

Assume that ~ma is the sole nonzero virtual slip-rate, let / be an arbitrary spatial vector, and
let
~ma ðxÞ ¼ /  ðx  oÞ;
then grad~ma ¼ / and (3.11) implies that
ðna  Q> na Þ  / ¼ 0:
Thus, since both / and a are arbitrary, na* = Qna for all a and na is frame-indifferent.

3.4. Macroscopic and microscopic force balances

Consider a virtual velocity V for which ~v is arbitrary, L~ e ¼ grad~v, and ma = 0 for each
a, so that the constraint (3.4) is satisfied. Then (3.6) reduces to the classical virtual power
balance for the Cauchy stress. We may conclude that T represents this stress and, further,
that the classical local force balance
div T þ b ¼ 0 ð3:12Þ
and traction condition t(n) = Tn are satisfied.
To derive the microscopic force balances we introduce, for each slip system a, the
resolved shear sa defined by
def  a
sa ¼ S : T ¼ sa  Tm
a ð3:13Þ
a
(so that s represents the Cauchy stress T resolved on the deformed ath slip system). As-
sume that the macroscopic virtual velocity vanishes, ~v  0, so that by (3.4)
X
ee ¼ 
L a
~ma S
a

with ~m1 ; ~m2 ; . . . ; ~mN arbitrary. Then (3.6) and (3.13) yield the microscopic virtual-power
relation
XZ XZ
Na ðnÞ~ma da ¼ ½ðpa  sa Þ~ma þ na  grad~ma dv: ð3:14Þ
a oP a P

Thus appealing to the virtual counterpart of the identity (3.2) we find that
X Z Z 
a a a a a a a
ðN ðnÞ  n  nÞ~m da þ ðdiv n þ s  p Þ~m dv ¼ 0: ð3:15Þ
a oP P

Since the virtual slip rates ~m are arbitrary, (3.15) must be satisfied for all virtual slip rates
and all P; thus a standard argument yields the microscopic force balance10
div na þ sa  pa ¼ 0 ð3:16Þ
a a
and the microscopic traction condition N (n) = n  n for each slip system a.

10
Gurtin (2000, Eq. (48), 2002, Eq. (5.14)).
712 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

4. Free-energy imbalance

Consider an arbitrary spatial region Pt convecting with the body. Let w denote the free
energy and d P 0 the dissipation, with w measured per unit volume in the lattice space, but
with d measured per unit volume in the observed space, so that for
def
J ¼ det F ¼ det Fe

(cf. (2.1)2), the integrals


Z Z
wJ 1 dv and d dv;
Pt Pt

respectively, represent the free energy of – and the dissipation within – Pt . The free-energy
imbalance for Pt is then the assertion that the rate at which the free energy of Pt is chang-
ing is not greater than the external power expended on Pt ,
Z _
wJ 1 dv 6 WðPt Þ; ð4:1Þ
Pt

and since, by (3.6), WðPt Þ ¼ IðPt Þ, (3.1) implies that


Z _ Z XZ
1
wJ dv  e
T  L dv  ðpa ma þ na  grad ma Þdv 6 0: ð4:2Þ
Pt Pt a Pt

Further, a standard result of continuum mechanics implies that, since Pt convects with the
body,
Z _ Z
wJ 1 dv ¼ _ 1 dv;
wJ ð4:3Þ
Pt Pt

thus (4.2) yields


Z " X
#
1 _ e a a a a
J w  T  L dv  ðp m þ n  grad m Þ dv 6 0
Pt a

and, since Pt was arbitrarily chosen,


X
J 1 w_  T  Le  ðna  grad ma þ pa ma Þ 6 0: ð4:4Þ
a

The term T: Le, which represents the elastic stress-power, is most conveniently expressed in
terms of the elastic strain
1
Ee ¼ ðFe> Fe  1Þ: ð4:5Þ
2
To establish such an expression we note that since T is symmetric so also is Fe1 TFe>
and

T : Le ¼ T : ðF_ e Fe1 Þ ¼ ðFe1 TFe> Þ : ðFe> F_ e Þ ¼ J 1 Te : E_ e


M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 713

with Te the lattice stress defined by

Te ¼ J Fe1 TFe> : ð4:6Þ

We may therefore rewrite (4.4) in the form (G-2002, Eq. (6.6))


def
X
d ¼ J 1 w_ þ J 1 Te : E_ e þ ðna  grad ma þ pa ma Þ P 0; ð4:7Þ
a

where d represents the dissipation per unit volume. This local free-energy imbalance is cen-
tral to the development of a suitable constitutive theory. Pointwise – as a linear transfor-
mation – Te maps lattice vectors to lattice vectors and Te : E_ e represents the stress-power
measured per unit volume in the lattice.

5. Constitutive equations

Our goal is a constitutive theory that allows for dislocation densities and slip-rate gra-
dients as independent constitutive variables, but that does not otherwise depart drastically
from more conventional theories.

5.1. Defect forces. Energetic microscopic stresses

Writing

q ¼ ðq1‘ ; q2‘ ; . . . ; qN‘ ; q1 ; q2 ; . . . ; qN Þ


~ ð5:1Þ

for the list of dislocation densities and restricting attention to situations in which the elas-
tic strains are small, we assume that the free energy is given by a standard elastic strain-
energy augmented by a defect energy Wð~ qÞ:
1
w ¼ Ee : CEe þ Wð~
qÞ; ð5:2Þ
2
with elasticity tensor C symmetric and positive definite. We assume further that the lattice
stress Te is given by the standard stress–strain relation
Te ¼ CEe ; ð5:3Þ
then
1 e _ e
E : CE ¼ Te : E_ e ð5:4Þ
2
and the local free-energy imbalance (4.7) becomes
_ X
d ¼ J 1 Wð~ qÞ þ ðna  grad ma þ pa ma Þ P 0: ð5:5Þ
a

Central to the theory are the energetic defect forces defined by


oWð~qÞ oWð~qÞ
qÞ ¼ J 1
f‘a ð~ ; qÞ ¼ J 1
fa ð~ : ð5:6Þ
oqa‘ oqa
714 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

By (2.16)
_ X
J 1 Wð~
qÞ ¼ ðf a q_ a þ fa q_ a Þ
a ‘ ‘
ð5:7Þ
X
¼ a
ðf‘asa þ fala Þ  grad ma ð5:8Þ
with sa and la the lattice vectors sa and la pushed forward to the observed space; cf. (2.17).
We refer to
def
naen ¼ f‘asa þ fala ð5:9Þ
as the energetic microscopic stress for slip system a. By (5.9), naen
is tangent to the deformed
ath slip plane.
With a view toward showing that the energetic microscopic stress naen may be viewed as
a combination of distributed Peach–Koehler forces,11 we use (2.17) to pull naen back to the
lattice:
def
na#
en ¼ F
e1 a
nen ¼ f‘a sa þ fa la : ð5:10Þ
For a distribution of dislocations with line direction l (a lattice vector) evolving on the ath
slip plane, a distributed Peach–Koehler force should lie in the ath slip plane and be per-
pendicular to the line direction l. Thus, since the vectors sa and la, respectively, are perpen-
dicular to the line directions la and sa of edge and screw distributions, and since the scalar
defect forces f‘a and fa are associated with the defect energy, it would seem reasonable to
endow the defect forces f‘a sa and fa la that comprise na# en with the following physical
interpretation:
f‘a sa and fa la :
|fflffl{zfflffl} |{z}
distributed Peach–Koehler force on edge dislocations distributed Peach–Koehler force on screw dislocations

ð5:11Þ
A similar result is established in (G-2002, p. 22) for a defect energy dependent on G, rather
than on dislocation densities, but the argument used is convoluted. In contrast, the argu-
ment leading to (5.11) is closer to the underlying physics – and simpler.

5.2. Dissipative constitutive equations that account for slip-rate gradients

5.2.1. Reduced dissipation inequality


By (5.8) and (5.9)
_ X
J 1 Wð~
qÞ ¼ naen  grad ma ð5:12Þ
a

and (5.5) becomes


X
d¼ ½ðna  naen Þ  grad ma þ pa ma P 0: ð5:13Þ
a

Thus if we define dissipative microscopic stresses nadis through the relations


nadis ¼ na  naen ; ð5:14Þ

11
The classical Peach–Koehler force is a configurational force – energetic in nature – on a dislocation line in a
linear elastic body; cf., e.g., (Teodosiu, 1982, p. 191).
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 715

then (5.13) takes the form


X
d¼ ½pa ma þ nadis  grad ma P 0: ð5:15Þ
a

Next, let grada denote the tangential gradient of u on the deformed ath slip plane, so
that, given any scalar field u,
grada u ¼ grad u  ðm
 a  grad uÞm
 a: ð5:16Þ
Underlying the present theory is the tacit assumption that

the microscopic stresses nadis characterize dissipative microscopic forces associated with the
motion of dislocations on the ath slip plane Pa;

because such dislocations migrate tangentially on Pa, we require that nadis be tangential
to the deformed ath slip plane.12 Granted this, we may without loss in generality replace
the slip-rate gradients gradma in (5.15) by the corresponding tangential gradients gradama;
the result is a reduced dissipation inequality
X
d¼ fpa ma þ nadis  grada ma g P 0 ð5:17Þ
a

basic to our discussion of dissipative constitutive relations.

5.2.2. Conventional single-crystal plasticity revisited


The success of conventional single-crystal plasticity leads us to seek dissipative consti-
tutive equations that allow for slip-rate gradients as independent constitutive variables,
but that do not otherwise depart drastically from conventional theories based on consti-
tutive relations of the form
ma
sa ¼ S a Rðjma jÞ ; ð5:18Þ
jma j
where R(jmaj) characterizes the rate-sensitivity of the material, while Sa > 0a represents a
a
resistance to slip. (More common is the inverted form of (5.18): ma ¼ g1 jsS aj Ss a .) Within
the present framework the conventional theory has vanishing microscopic stresses na and
the microscopic force balance (3.16) implies that pa = sa; (5.18) is therefore equivalent to
the constitutive equation

ma
pa ¼ S a Rðjma jÞ : ð5:19Þ
jma j

Conventional theories are generally presumed consistent with hardening equations


X
S_ a ¼ hab ð~
SÞd b ; ~
S ¼ ðS 1 ; S 2 ; . . . ; S N Þ; ð5:20Þ
a

12
Some (unrelated) remarks: (i) by (3.1), na and (hence) nadis are vectors in the observed space (deformed
configuration); (ii) we have tacitly neglected cross-slip; (iii) note that (5.9) renders the energetic microstress naenj
also tangential to the deformed ath slip plane.
716 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

with hardening moduli hab often of the specific form

hab ð~
SÞ ¼ vab hðS b Þ þ ð1  vab ÞqhðS b Þ; ð5:21Þ
|fflfflfflfflffl{zfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflffl}
self-hardening latent-hardening

where h()P0 is a self-hardening function, q > 0 is an interaction constant (the ratio of the
self-hardening rate to the latent-hardening rate), and vab is defined by vab = 1 if mb = ± ma
and vab = 0 otherwise; Asaro and Needleman (1985) and Kalidindi et al. (1992).

5.2.3. Constitutive relations for pa and na


In laying down a constitutive relation for pa we are guided by the relation (5.19) and by
the success of the conventional theory. Specifically, we consider a constitutive relation for
pa that differs from (5.19) only through the replacement of jmajby an effective flow rate
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
def 2 2
d a ¼ jma j þ ‘2 jgrada ma j ; ð5:22Þ

with ‘ a dissipative length scale. This replacement leads us to introduce a (dimensionless)


rate-sensitivity function13 R() such that
Rð0Þ ¼ 0; Rðd a Þ > 0 for d a 6¼ 0;
and to a constitutive equation for pa of a form
ma
pa ¼ S a Rðd a Þ ð5:23Þ
da
that bears comparison with the conventional relation (5.19).
We next lay down constitutive relations for the dissipative microscopic stresses nadis .
Here the reduced dissipation inequality (5.17) and a desire for mathematical simplicity
suggest a constitutive equation of the form (5.23), but with ma replaced by ‘2 grada ma; viz.14
r a ma
nadis ¼ S a Rðd a Þ‘2 : ð5:24Þ
da
This relation, (5.9), and (5.14) combine to form a constitutive equation for the microscopic
stress na:
grada ma
na ¼ f‘asa þ fala þ S a gðd a Þ‘2 : ð5:25Þ
da
Since
2
grad ma  grada ma ¼ jgrada ma j ; ð5:26Þ
the dissipation (5.17) has a simple form
X
d¼ S a Rðd a Þd a ð5:27Þ
a

 a m
13
For example, one might take Rðd a Þ ¼ dd 0 .
14
The Eqs. (5.23) and (5.24) were proposed by Gurtin (2000, Section15). The structure of these equations bears
some comparison with equations introduced to characterize strengthening in isotropic plastic materials; cf.
Fredriksson and Gudmundson (2005), Gurtin and Anand (2005a,b).
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 717

that is strictly analogous to the conventional dissipation


X X
sa ma ¼ S a Rðjma jÞjma j
a a

associated with (5.18).


Finally, we assume that the slip resistances Sa > 0 are consistent with hardening
equations
X
S_ a ¼ S; ~
hab ð~ /Þd b ; ~
/ ¼ ð/1 ; /2 ; . . . ; /N Þ ð5:28Þ
a

that allow for dependence on the forest measures /a defined in Section 2.3. A simple gen-
eralization of (5.21) might then replace h(Sb) in (5.21) with h(Sb,/b); viz.
S; ~
hab ð~ /Þ ¼ vab hðS b ; /b Þ þ ð1  vab ÞqhðS b ; /b Þ: ð5:29Þ

6. Viscoplastic flow rule

6.1. General form

The decomposition na ¼ naen þ nadis allows us to write the microscopic force balance
(3.16) in the form
sa þ div naen ¼ pa  div nadis ; ð6:1Þ
where we have written the term div naen on the left, since, being energetic, its negative rep-
resents a backstress. When augmented by the constitutive Eqs. (5.9) and (5.24) the balance
(6.1) becomes the flow rule for slip system a:15
 
ma grada ma
sa  div ½f‘a ð~ qÞsa  fa ð~ qÞla ¼ S a gðd a Þ a  ‘2 div S a gðd a Þ : ð6:2Þ
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} d da
energetic backstress |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
dissipative hardening
a
By (2.16) the fields S , qa‘ ,
and qa
should be viewed as internal-state variables with evo-
lution governed by the differential equations
9
q_ a ¼ la  grad ma ; q_ a‘ ¼ sa  grad ma ; =
P ab ~ b ð6:3Þ
S_ a ¼ h ð~S; /Þd : ;
b

As such these equations should be supplemented by initial conditions for Sa, qa‘ , and qa . If
we assume that the body is initially in a virgin state, then appropriate initial conditions
would be
S a ðx; 0Þ ¼ S 0 ; qa‘ ðx; 0Þ ¼ qa ðx; 0Þ ¼ 0 ð6:4Þ
for each slip system a, with S0 > 0, a constant, the initial slip resistance.

15
Cf. Gurtin (2000, eq. (164a)), which has no backstress, and (2002, eq. (7.18)) in which W = W(G) and nadis  0.
718 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

6.2. Simple defect energies dependent on net dislocation densities

The field
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
def
qanet ¼ jqa‘ j2 þ jqa j2 ð6:5Þ
a
represents the net dislocation density on the slip plane P ; cf., e.g., Ohno and Okumura
(2007).

6.2.1. Uncoupled quadratic defect energy


A simple defect energy has Wð~ qÞ uncoupled and quadratic in the net dislocation densi-
ties and hence the form
1 X
Wð~qÞ ¼ S 0 L2 ðqanet Þ2 ; ð6:6Þ
2 a

with L an energetic length scale and S0 the initial slip resistance; Gurtin et al. (2007). In
this case if we write
a‘ ¼ J 1 qa‘ ; q
q a ¼ J 1 qa ð6:7Þ
for the dislocation densities measured per unit volume in the deformed body, then the de-
fect forces (5.6) and energetic microscopic stress (5.9) are given by
)
f‘a ¼ S 0 L2 q
a‘ ; fa ¼ S 0 L2 q a ;
ð6:8Þ
qala  q
naen ¼ S 0 L2 ½ a‘sa ;
and the flow rule for the ath slip system becomes
 
ma grada ma
sa  S 0 L2 div ½ ala ¼ S a gðd a Þ a  ‘2 div S a gðd a Þ
qa‘sa  q : ð6:9Þ
d da

6.2.2. Coupled quadratic defect energy


Consider the defect energy
2 3
61 X a 2 X 7
qÞ ¼ S 0 L2 6
Wð~ 42 ðqnet Þ þ k qanet qbnet 7
5 ð6:10Þ
a a;b
a6¼b

which couples the individual slip systems. In this case, consistent with (6.7), we let
anet ¼ J 1 qanet
q ð6:11Þ
and write ca‘ and ca for the dislocation concentrations defined by
qa‘ a
q qa b‘
q
ca‘ ¼ ¼ a‘ ; ca ¼ ¼ : ð6:12Þ
a
qnet q
net qanet qbnet
The energetic defect forces (5.6) are then given by
2 3 2 3
6 a X 7 6 a X 7
f‘a ¼ S 0 L2 6
4q‘ þ kca‘ bnet 7
q 5; fa ¼ S 0 L2 6
4q þ kca bnet 7
q 5; ð6:13Þ
b b
b6¼a b6¼a
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 719

where k is a constant modulus associated with the energetic coupling of slip systems. Thus,
a‘ ¼ ca‘ q
since q b ¼ ca q
anet and q bnet , the resulting flow rule for system a takes the form
8  2 39
>
> >
>
< 6 X 7=
a 2 a a a a 6 a b 7
s  S 0 L div ðc‘s  c l Þ4q net þ k q
net 5
>
> >
>
: b ;
b6¼a
 
ma grada ma
¼ S a gðd a Þ a  ‘2 div S a gðd a Þ ð6:14Þ
d da
and exhibits energetic coupling between system a and all other systems (via the term with
coefficient k).

6.2.3. Self-energy of Ohno and Okumura


The defect energy
1
X 1þr
Wð~qÞ ¼ S 0 L2 ð1 þ rÞ ðqanet Þ ð6:15Þ
a

with r = 0 was introduced by Ohno and Okumura (2007) to account for the self energy of
geometrically necessary dislocations.16 Since
1þr r a
oðqanet Þ qa‘ oðqanet Þ a r q
ð1 þ rÞ1 ¼ ðqanet Þr ; ð1 þ rÞ1 ¼ ðq net Þ ;
oqa‘ qanet oqa qanet
it follows that
r qa‘ r qa
f‘a ¼ J 1 S 0 L2 ðqanet Þ ; fa ¼ J 1 S 0 L2 ðqanet Þ : ð6:16Þ
qanet qanet
Thus using the concentrations (6.12) we find that the flow rule for slip system a has the
form
a
 a a
a 2 1 a r a a a a a a m 2 a a grad m
s  S 0 L div ½J ðqnet Þ ðc l  c‘s Þ ¼ S gðd Þ a  ‘ div S gðd Þ : ð6:17Þ
d da
Ohno and Okumura (2007) show that the self energy (6.15) leads to strengthening; that
is, to an increase in the initial yield stress. As is clear from the studies of Fredriksson and
Gudmundson (2005), Anand et al. (2005) and Gurtin et al. (2007), such an increase is also
a consequence of dissipative hardening when slip-rate gradients enter the constitutive rela-
tions for the microscopic stresses na, as in (5.24). Since the strengthening introduced by
Ohno and Okumura most certainly leads to kinematic hardening, while dissipative hard-
ening involves no backstress, it might be possible to experimentally ascertain whether one
or both of the hardening mechanisms is the root cause of strengthening.

7. Microscopically simple boundary conditions associated with the flow rules

Each of the flow rules discussed in Section 6 involves second slip-rate gradients and
hence represent a system of partial differential equations for the slip-rates, given the

16
The term involving r > 0, r small, represents a regularization introduced to ensure that the defect forces (5.6)
are defined when qanet ¼ 0; cf. Ohno and Okumura (2007, eq. (52)).
720 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

resolved shears sa. The flow rules are therefore nonlocal and hence in need of concomitant
boundary conditions.
Let B denote the deformed body at an arbitrarily chosen time. The presence of micro-
scopic stresses results in an expenditure of power
Z
ðna  nÞma da ð7:1Þ
oB

by the material in contact with the body, and this necessitates a consideration of boundary
conditions on oB involving the microscopic tractions na  n and the slip-rates ma, where n
denotes the outward unit normal to oB. We restrict attention to boundary conditions that
result in a null expenditure of microscopic power in the sense that

ðna  nÞma ¼ 0 on oB ð7:2Þ

for all a.17 Specifically, we consider microscopically simple boundary conditions asserting
that
ma ¼ 0 on Shard and na  n ¼ 0 on Sfree ð7:3Þ
for a = 1, 2, . . ., N, where Shard and Sfree are complementary subsurfaces of oB respectively
referred to as the microscopically hard and the microscopically free portions of oB.18
The microscopically hard condition corresponds to a boundary surface that cannot
pass dislocations (for example, a boundary surface that abuts a hard material); the micro-
scopically free condition corresponds to a boundary across which dislocations can flow
freely from the body.

8. Plastic free-energy balance

As a result of the constitutive relations, the global free-energy imbalance (4.2) has a
plastic counterpart, which we now derive. By (5.12)
Z _ XZ
J 1 Wð~
qÞdv ¼ naen  grad ma dv; ð8:1Þ
Pt a Pt

on the other hand, using (3.2), (3.16), and (5.14) we rewrite the right side of (8.2) (for each
a) as follows
Z Z
naen  grad ma dv ¼ ½naen  grad ma þ ðsa  pa þ div na Þ ma dv
Pt Pt |fflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
¼0
Z Z
¼ a a
½ðs  p Þm a
nadis a
 grad m dv þ ðna  nÞma da: ð8:2Þ
Pt oPt

17
Cf. Gurtin (2000, Eq. (137), 2002, Eqs. (9.1) and (9.4)).
18
As Gurtin and Needleman (2002) show, the issue of boundary conditions is delicate when: (i) the defect energy
depends on the Burgers tensor G; (ii) the theory does not include constitutive dependencies on slip-rate gradients.
Here neither (i) or (ii) is applicable. Even so we conjecture that, granted (ii), the dependence of the defect energy
on dislocation densities precludes the problems encountered by Gurtin and Needleman; and that a similar
conjecture applies to the small deformation theory of Gurtin et al. (2007).
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 721

The identities (8.1) and (8.2) yield the plastic free-energy balance
Z _ X Z X Z X Z
a
J 1 Wð~
qÞdv ¼ a
sa a
m dv þ a
ðn  nÞm a
da  a
ðpa ma þ nadis  gradma Þdv :
Pt oPt
|fflfflfflfflfflfflfflfflfflfflfflP{zfflfflfflfflfflfflfflfflfflffl
t
ffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} Pt
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl
ffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
cold working microscopic power dissipationP0

ð8:3Þ
If we apply this balance with Pt ¼ Bt to the special case in which the microscopically sim-
ple boundary conditions (7.3) are satisfied, we arrive at the following conclusions:

(i) the temporal increase in defect energy can never exceed the cold working;
(ii) if the defect energy vanishes, then – as in conventional theories of plasticity – the
cold working coincides with the dissipation.

The term in (8.3) labelled ‘‘microscopic power” represents the power expended on Pt by
the microscopic tractions na  n acting over oPt . By (5.14) the microscopic stress admits a
decomposition na ¼ naen þ nadis into energetic and dissipative parts, and, using (5.9), we can
write Zthe energetic part of
Z the microscopic power in the form
ðnaen  nÞma da ¼ ½f‘a ma ðsa  nÞ þ fa ma ðla  nÞ da: ð8:4Þ
oPt oPt
The right side of (8.4) represents power expended across oPt by the normal components of
the Peach–Koehler forces f‘a sa and fa la associated with edge and screw dislocations; cf.
the paragraph containing (5.11).
A more interesting interpretation of (8.4) pertains to the transport of dislocations as
described by their fluxes  qa‘ ¼ masa and qa ¼ mala ; cf. (2.19). Bearing in mind that the
a a
defect forces f‘ and f defined in (5.6) would – when interpreted within the framework
of dislocation transport – be considered as chemical potentials19
oW oW
la‘ ¼ J 1 a ð¼ f‘a Þ; la ¼ J 1 a ð¼ fa Þ
oq‘ oq
corresponding to energetic changes resulting from changes in edge and screw densities.
With this interpretation la‘ qa‘ and la 
qa represent fluxes of energy associated with flows
of edge and screw dislocations as described by the fluxes qa‘ and qa , thus using (2.19)
we mayZ rewrite (8.4) as follows:
Z
ðna  nÞma da ¼  ½la‘ ð
qa‘  nÞ þ la ð
qa  nÞ da; ð8:5Þ
oPt oPt
the right side of this relation represents energy carried into Pt across oPt by the flow of
dislocations.

9. Concluding remarks

9.1. The system of flow rules

The theory results in a system of flow rules for the individual slip systems in the form of
a coupled system of partial differential equations (which needs concomitant boundary con-
ditions). Associated with each flow rule are:

19
Cf. Fried and Gurtin (2004, Section 4A).
722 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

(i) an energetic backstress of the form


div ½f a ð~
‘ qÞla
qÞsa  f a ð~


(cf. (6.2)) and hence dependent on gradients of dislocation densities; because the dis-
location densities are internal-state variables with evolution governed by the differen-
tial equations
q_ a ¼ la  grad ma ; q_ a ¼ sa  grad ma
 ‘ ð9:1Þ
(and suitable initial conditions), the backstress depends on histories of second slip-
rate gradients; this – the essential feature of the theory20 – should be compared with
the small-deformation theory of Gurtin et al. (2007) in which the backstress depends
on current values of second slip gradients (not slip-rate gradients!);
(ii) dissipative hardening involving second slip-rate gradients;21this hardening should
lead to strengthening; cf. Footnote 14.

9.2. Connections with dislocation theory

An essential feature of the theory is the manner in which it qualitatively mimics certain
characteristics of dislocations:22

The plastically convected rate of the Burgers tensor G is found to be the sum of rates of
continuous distributions of screw and edge dislocations on the individual slip systems,
distributions that have the canonical forms for slip in a single crystal. These distribu-
tions are automatically consistent with balances, (2.21), that are in a form
fdensity rateg ¼ fdivergence of a fluxg þ fsupplyg
standard in theories of transport. Moreover: the fluxes are in glide directions; each of
the density-rates is given, modulo sign, as a directional derivative of the corresponding
slip rate in the glide direction.

A free energy dependent on dislocation densities that leads to thermodynamically con-


jugate microscopic forces (stresses) parallel to glide directions of the corresponding edge
and screw distributions, which is a central feature of Peach–Koehler forces.

A consequence of the basic equations of the theory is a global free-energy balance (8.3)
asserting that temporal changes of the free energy of any convecting subregion Pt of the
body is equal to the cold working (of the macroscopic stress) plus the energy dissipated
plus(#)]the rate at which the microscopic stresses na expend power over oPt .

For the special case in which the microscopic stresses na are purely energetic, if we
define chemical potentials for the individual dislocation densities in the standard manner
as partial derivatives of the defect energy with respect to corresponding dislocation den-
sities, then(à) the power expenditure (#) gives the net energy flow into Pt across oPt due
to the flow of dislocations, a flow characterized by the individual dislocation fluxes.9.3.

20
The author is unaware of other single-crystal theories with this property.
21
Cf. Gurtin (2000) and Gurtin et al. (2007).
22
The results of the first two bullets, taken from Gurtin (2006 (G-2006)), are essential to the complete picture
given below:
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 723

Some remarks

We do not accompany this presentation with numerical examples and comparisons with
experimental data. But in this regard the small deformation counterpart of the present the-
ory compares very well to:

(i) discrete dislocation calculations (Nicola et al., 2005);


(ii) a large class of experiments on Al, Ni, Cu, Fe, and Steel at submicron to several
micron length scales, granted the theory is endowed with the Ohno–Okumura energy
(6.15) (Ohno and Okumura, 2007).

We see no reason why the present theory should not have property (ii), at least for suf-
ficiently constrained flow.23 Moreover for such flows the behavior of the theory should be
close to the small deformation theory discussed by Gurtin et al. (forthcoming) and, as
explicitly displayed by them, should lead to size-dependent phenomena such as strength-
ening and boundary layers at hard boundaries.
Gradient theories of the type considered here extend naturally to situations involving
grain boundaries. Indeed, for S a subsurface of oB, the presence of microscopic stresses
na, a = 1, 2, . . . , N, results in a power expenditure
XZ
ðna  nÞma da ð9:2Þ
a S

by the material in contact with the body; when this material is that of another grain, then
(9.2) and its analog for the other grain leads to an extension of the principle of virtual
power that can be used to develop force balances and a free-energy imbalance for the grain
boundary. For small deformations such a procedure was used a basis for a treatment of
grain boundaries by Cermelli and Gurtin (2002), Gurtin and Needleman (2002) and Gur-
tin (2007). For finite deformations such a procedure is highly nontrivial and beyond the
scope of this study; it is the subject of a current investigation.

Acknowledgements

I thank Lallit Anand and Paolo Cermelli for valuable discussions. The support of this
work by the Department of Energy and the National Science Foundation is gratefully
acknowledged.

References

Acharya, A., Bassani, J.L., 2000. Incompatibility and crystal plasticity. Journal of the Mechanics and Physics of
Solids 48, 1565–1595.
Acharya, A., Beaudoin, A.J., 2000. Grain size effects in viscoplastic polycrysals at moderate strains. Journal of the
Mechanics and Physics of Solids 48, 2213–2230.
Anand, L., Gurtin, M.E., Lele, S.P., Gething, C., 2005. A one-dimensional theory of strain-gradient plasticity;
Formulation, analysis, numerical results. Journal of the Mechanics and Physics of Solids 53, 1789–1806.
Arsenlis, A., Parks, D.M., 1999. Crystallographic aspects of geometrically-necessary and statistically-stored
dislocation density. Acta Materialia 47, 1597–1611.

23
At present we are not aware of a discrete dislocation theory for general finite deformations.
724 M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725

Arsenlis, A., Parks, D.M., Becker, R., Bulatov, V.V., 2004. On the evolution of crystallographic dislocation
density in non-homogeneously deforming crystals. Journal of the Mechanics and Physics of Solids 52, 1213–
1246.
Asaro, R.J., Needleman, A., 1985. Texture development and strain hardening in rate dependent polycrystals.
Acta Metallurgica 33, 923–953.
Bardella, L., 2007. Some remarks on the strain gradient crystal plasticity modelling, with particular reference to
the material length scales involved. International Journal of Plasticity 23, 296–322.
Cermelli, P., Gurtin, M.E., 2001. On the characterization of geometrically necessary dislocations in finite
plasticity. Journal of the Mechanics and Physics of Solids 49, 1539–1568.
Cermelli, P., Gurtin, M.E., 2002. Geometrically necessary dislocations in viscoplastic single crystals and bicrystals
undergoing small deformations. International Journal of Solids and Structures 39, 6281–6309.
Evers, L.P., Brekelmans, W.A.M., Geers, M.G.D., 2004a. Non-local crystal plasticity model with intrinsic SSD
and GND effects. Journal of the Mechanics and Physics of Solids 52, 2379–2401.
Evers, L.P., Brekelmans, W.A.M., Geers, M.G.D., 2004b. Scale dependent crystal plasticity framework with
dislocation density and grain boundary effects. International Journal of Solids and Structures 41, 5209–5230.
Fleck, N.A., Muller, G.M., Ashby, M.F., Hutchinson, J.W., 1994. Strain gradient plasticity: theory and
experiment. Acta Metallurgica 42, 475–487.
Fried, E., Gurtin, M.E., 2004. A unified treatment of evolving Interfaces accounting for small deformations and
atomic transport with emphasis on grain boundaries and epitaxy. Advances in Applied Mechanics 40, 1–177.
Fredriksson, P., Gudmundson, P., 2005. Size dependent yield strength of thin films. International Journal of
Plasticity 21, 1834–1854.
Gurtin, M.E., 2000. On the plasticity of single crystal: free energy, microforces, plastic strain gradients. Journal of
the Mechanics and Physics of Solids 48, 989–1036.
Gurtin, M.E., G-2002). A gradient theory of single-crystal plasticity that accounts for geometrically necessary
dislocations. Journal of the Mechanics and Physics of Solids 50, 5–32.
Gurtin, M.E., G-2006). The Burgers vector and the flow of screw and edge dislocations in finite-deformation
plasticity. Journal of the Mechanics and Physics of Solids 54, 1882–1898.
Gurtin, M.E., 2007. A theory of grain boundaries that accounts for grain-misorientation and grain-boundary
orientation. Journal of the Mechanics and Physics of Solids, in press.
Gurtin, M.E., Anand, L., Lele, S.P., 2007. Gradient single-crystal plasticity with free energy dependent on
dislocation densities. Journal of the Mechanics and Physics of Solids 55, 1853–1878.
Gurtin, M.E., Anand, L., 2005a. A theory of strain gradient plasticity for isotropic, plastically irrotational
materials. I: Small deformations. Journal of the Mechanics and Physics of Solids 53, 1624–1649.
Gurtin, M.E., Anand, L., 2005b. A theory of strain gradient plasticity for isotropic, plastically irrotational
materials. II: Finite deformations. International Journal of Plasticity 21, 2297–2318.
Gurtin, M.E., Needleman, A., 2002. Boundary conditions for single-crystal plasticity that account for The
Burgers vector. Journal of the Mechanics and Physics of Solids 53, 1–31.
Han, C.S., Gao, H., Huang, Y., Nix, W.D., 2005. Mechanism-based strain gradient crystal plasticity: I Theory, II
Analysis. Journal of the Mechanics and Physics of Solids 53, 1188–1222.
Hutchinson, J.W., 2000. Plasticity at the micron scale. International Journal of Solids and Structures 37,
225–238.
Kalidindi, S.R., Bronkhorst, C.A., Anand, L., 1992. Crystallographic texture evolution in bulk deformation
processing of fcc metals. Journal of the Mechanics and Physics of Solids 40, 537–569.
Kröner, E., 1960. Allgemeine Kontinuumstheorie der Versetzungen und Eigenspannungen. Archive for Rational
Mechanics and Analysis 4, 273–334.
Kubin, L.P., Canova, G., Condat, M., Devincre, B., Pontikis, V., Bréchet, Y., 1992. Dislocation microstructures
and plastic flow: a 3D simulation. Solid State Phenomena 23 (24), 455–472.
Kuhlmann-Wilsdorf, D., 1989. Theory of plastic deformation: properties of low energy dislocation structures.
Materials Science and Engineering 113A, 1–41.
Lee, E.H., 1969. Elastic-plastic deformation at finite strains. Jounal of Applied Mechanics 36, 1–6.
Ma, Q., Clark, D.R., 1995. Size dependent hardness of silver single crystals. Journal Materials Research 10, 853–
863.
Menzel, A., Steinman, P., 2000. On the continuum formulation of higher gradient plasticity for single and
polycrystals. Journal of the Mechanics and Physics of Solids 48, 1777–1796.
Nicola, L., Van der Giessen, E., Gurtin, M.E., 2005. Effect of defect energy on strain-gradient predictions of
confined single-crystal plasticity. Journal of the Mechanics and Physics of Solids 53, 1280–1294.
M.E. Gurtin / International Journal of Plasticity 24 (2008) 702–725 725

Okumura, D., Higashi, Y., Sumida, K., Ohno, N., 2006. Homogenization theory of strain gradient single-crystal
plasticity. Journal of the Mechanics and Physics of Solids.
Ohno, N., Okumura, D., 2007. Higher-order stress and grain size effects due to self energy of geometrically
necessary dislocations. Journal of the Mechanics and Physics of Solids 55, 1879–1898.
Stelmashenko, N.A., Walls, M.G., Brown, M.G., Millman, Y.V., 1993. Microindentations on W and Mo
oriented single crystals: an ATM study. Acta Metallurgica et Materialia 41, 2855–2865.
Stolken, J.S., Evans, A.G., 1998. A microbend test method for measuring the plasticity length scale. Acta
Materiala 46, 5109–5115.
Sun, S., Adams, B.L., Shet, C.Q., Saigal, S., King, W., 1998. Mesoscale investigation of the deformation field of
an aluminum bicrystal. Scripta Materialia 39, 501–508.
Sun, S., Adams, B.L., King, W., 2000. Observations of lattice curvature near the interface of a deformed
aluminium bicrystal. Philosophical Magazine A 80, 9–25.
Teodosiu, C., 1982. Elastic Models of Crystal Defects. Springer, Berlin.

Vous aimerez peut-être aussi