Sie sind auf Seite 1von 40

International Journal of Plasticity 20 (2004) 363402

www.elsevier.com/locate/ijplas

Continuum physics of phase and defect


microstructures: bridging the gap between
physical metallurgy and plasticity of
aluminum alloys
Michael V. Glazo a,b,*, Frederic Barlata,c, Hasso Weilanda,d
a
Alcoa Technical Center, Alcoa Center, PA 15069, USA
Vanderbilt University, Department of Physics and Astronomy, Nashville, TN 37235, USA
c
Centre for Mechanical Technology and Automation, Department of Mechanical Engineering,
University of Aveiro, 3810-193 Aveiro, Portugal
d
University of Pittsburgh, Department of Materials Science, Pittsburgh, PA 15261, USA
b

Received in final revised form 2 April 2003


This paper is dedicated to the memory of Dr. Owen Richmond

Abstract
The problem of how alloy composition and stress, phases and defects interact is of key
importance for understanding and successful development of new alloys and tempers
Richmond, 1986; Staley, J.T., 1992. Metallurgical aspects aecting strength of heat-treatable
alloy products used in the aerospace industry. In: Proceedings IIIrd Int. Conf. on Aluminum
Alloys, pp. 107143]. In the area of phase transformations and microstructures, considerable
progress has been achieved using lattice and continuum models with sharp or diuse interfaces. In the area of defect microstructures and plastic instabilities, there is an improved
understanding of material plasticity, dierent aspects of dislocation patterning, which dene
the performance of aluminum alloys. While the techniques to describe phase and defect
transformations are very dierent, both types of changes can occur simultaneously when
processing real-life aluminum alloys. The issue of concentration-sensitive plasticity models
and phase microstructure models incorporating defects (such as dislocations) has been a topic
of the materials community interest during the last several years. One should mention such
events as the Workshop on Concentration-Sensitive Plasticity Models (Rockport, MA, Fall
2000), Materials Research Society (MRS) symposia Interaction of Phase and Defect Microstructures in Metallic Alloys; Inuences of Interface and Dislocation Behavior on Micro-

* Corresponding author. Tel.: +1-724-337-2847; fax +1-724-337-2044.


E-mail address: michael.glazo@alcoa.com (M.V. Glazo).
0749-6419/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0749-6419(03)00093-7

364

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

structure Evolution (Boston, 1998, 2000). In this paper similar conceptual problems arising
both in plasticity theory and in classical thermodynamics are analyzed, which can be traced to
the absence of characteristic length scales in the corresponding phenomenological formalisms.
It is further demonstrated that the concepts behind strain-gradient plasticity and thermodynamics of non-uniform equilibrium systems have common grounds in the same gradient
representation of the corresponding free energy functionals. This similarity gives hope that a
unied theory of plasticity and microstructure is indeed possible and can be developed. A
survey of recent developments in this direction is presented.
# 2003 Elsevier Ltd. All rights reserved.
Keywords: Microstructures of phases and defects; Thermodynamics of alloys; Plasticity theory; Phaseeld approach; Characteristic length scales

1. Introduction
In the present paper an eort was made to propose a roadmap toward the union
of physical metallurgy and material mechanics. We begin by reviewing thermodynamic approaches to microstructure and plasticity and demonstrate that the
classical Gibbsian thermodynamics does not set the goal of incorporating microstructure (even though so-called sharp-interface models can be used in conjunction with front tracking mechanisms in phase transformation studies). The concepts
of thermodynamics of concentrationally non-uniform equilibrium system are further
introduced, and the modied diusion equation derived by Cahn and Hilliard for
the description of spinodal decomposition is represented. It is further demonstrated
how this approach resulted in the appearance of phase-eld modeling of microstructure and its evolution.
The problem of interaction of alloy composition and stress has been one of the
most fundamental in the theory of alloys during the last years (see Richmond, 1986;
Staley, 1992; Cahn, 1994). To what extent can we predict the change in elastic
plastic response of a given material element under dierent processing operations
when the concentrations of alloying elements change, and how will it depend on
phase composition? Traditionally these important issues have been treated separately by the two branches of modern metallurgyphysical and mechanical, by
materials science and solid mechanics, although it is obvious that a holistic approach
to aluminum alloy design and also to fabrication of aluminum alloys, would require
a profound understanding of the underlying physics and mechanics of the both
metallurgy branches mentioned above.
One of the most important features of aluminum alloys is their microstructure.
Dislocations and grain boundaries to a signicant extent aect, and consequently
control, their mechanical properties. On the other hand, physical properties are
strongly inuenced by alloy chemical and phase composition. The formation of
microstructure is accompanied by symmetry-breaking transitions, and typically one
or more length (and time) scales are required for the description of its formation and
evolution. Following the classication proposed by Staley (1992), one can describe
the relevant metallurgical features and the processing steps leading to their forma-

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

365

tion as follows: constituent particles (solidication), dispersoids (ingot thermal


treatment), grain structure (bulk deformation) and its evolution, dislocations (cold
work after quenching), low-temperature clustering reactions, coherent, semi-coherent and non-coherent precipitates of several types, including GuinierPreston zones,
metastable and thermodynamically stable (nal heat treatment) and also such features as twins, stacking faults and dierent kinds of phase interfaces (both diuse
and sharp). On the other hand, in mechanical metallurgy of aluminum alloys
considerable eorts have been made to understand such plastic instabilities as the
PortevinLe Chatelier eect, the Luders bands, coarse slip bands, and microscopic
shear bands. Recent achievements in this area are reported by Ananthakrishna et al.
(2001) and Ziegenbein et al. (2000).
On ner scales, considerable progress was achieved in understanding and modeling the behavior of dierent dislocation structures (assemblies) in both monotonically and cyclically deformed alloys (veins, cells, mazes, persistent slip bands and
their ladder structure in fcc-metals etc.). Recent publications and reviews on this
subject include: Nabarro (2001), Needleman and Van der Giessen (2001); Avlonitis
et al. (2000), Hahner and Zaiser (1999), Swaminarayan and Lesar (2001), Wang et
al. (2000) and Zaiser et al. (1999).
While the presence of defects (dislocations, grains boundaries) aects the
mechanisms and rates of precipitation reactions very strongly (Cahn, 1998), the
inverse statement is also true. Dierent particles present in heterogeneous aluminum
alloys can exert signicant inuence on the plastic properties of materials. This
important inverse relation has attracted less attention from the materials community, which was more successful in incorporating the eects of elastic conjugation
and coherency into phase microstructure models. However, it becomes critical for
aerospace and automotive alloy design, (Barlat et al., 1996, 1997, 2002, 2003; Weiland, 1994).
Consequently, one can expect that as a reection of this situation on the microscale, a continuum formalism should exist on the mesoscopic scale that would provide a unied description of meso-structure(s) of both types. In a sense, such a
theory would involve the union of solid mechanics and materials science to provide
a quantitative basis for the design of deformation processes to achieve controlled
properties as well as shapes (Richmond, 1986).1
The range of applicability of such a continuum theory is illustrated by Fig. 1,
which demonstrates the dependence of mass density on the volume of measurement,
(Eringen, 1968). The mass density of a homogeneous material may be calculated
approximately by weighing a large number of pieces of matter having dierent
volume and calculating the ratio:  Dm=DV, where m is the total mass contained
in V. Fig. 1 demonstrates that thus dened mass density is nearly constant when
V is greater than a certain critical volume V* and starts to exhibit dependence on

Owen Richmonds many contributions to material mechanics have been described by Smelser et al.
(2001).

366

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

V when V* < V. As V approaches zero, oscillations of density become more
and more pronounced which can be explained by the atomistic nature of materials.
The classical continuum theory may therefore be not a good mathematical approximation of a real physical situation in the range V* < V (Eringen, 1968). In this
paper the region of intermediate, mesoscopic length scales will be mostly discussed.
What follows below are some thoughts on further development of understanding
of phase and defect microstructures in aluminum alloys, and on its present status.
These thoughts are based on interesting similarities between internal-variable
approach to plasticity of aluminum alloys with spatial coupling, strain-gradient
plasticity on the one hand, and thermodynamics with order parameters as applied
to the problems of grain growth, solidication, precipitation, spinodal decomposition etc., on the other hand.
The underlying mechanisms of microstructure formation (or material patterning) can be both kinetically and thermodynamically-driven and thus mathematics
somewhat dierent in these two cases: in the rst case one needs to analyze dissipative nonlinear systems of ordinary or partial dierential equations (e.g., for
homogeneous and non-homogeneous yielding, respectively) and to associate symmetry-breaking transitions and spatio-temporal patterning with the corresponding
bifurcation points; while in the latter case one typically deals with variational problems of minimization of a certain free energy functional, which depends on order
parameters (the phase-eld approach). However, understanding some general
mathematical and physical features of patterns and their evolution might provide a
common methodological basis for the two branches of modern metallurgyphysical and mechanical, as applied to understanding of microstructure.
This paper mostly deals with the phenomenological continuum approach to
modeling aluminum alloy systems with microstructure called phase-eld theory, and
with certain approaches in modern plasticity theory, which are based on the notion
of internal-variables. It does not cover the vast area of classical phenomenological
thermodynamics, Gibbs (1906), although it is clear that such models and databases
must be integrated into the framework of microstructure calculations.

Fig. 1. Mass density versus volume of measurementan illustration of the applicability range of a continuum mesoscopic approach (Eringen, 1968).

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

367

Furthermore, the paper does not deal with atomistic, statistical-mechanical and
quantum-mechanical aspects of phase equilibria, although a variety of promising
approaches emerged here recently, (Vitek, 1995; Bulatov et al., 1999).
In Section 2 a brief historical overview of the problem is given, and dierent
schools of thought in modern equilibrium and non-equilibrium thermodynamics are
analyzed. Without such analysis it is impossible to introduce the concept of internal-variables, which had played the key role in the development of modern plasticity theory, and to understand the phase-eld approach. This analysis is carried
out mostly following the fundamental works of Cahn, Rice, Coleman, Noll, and
Truesdell. It is demonstrated that although Rices modication of the internal-variable theory was very useful in answering the question on how microscopic rearrangements of structure may result in the normality ow rule and in existence of plastic
potentials, further work, both on the fundamental (Acharya and Bassani, 2000;
Bassani, 2001), and applied levels was necessary to customize the theory and to
make it applicable to concrete problems of material fabrication and processing.
Additionally, a critical role of such researchers as Asaro (1983), Asaro and Rice
(1977), Fleck and Hutchinson (1997) and Gurtin (2000, 2002, 2003) in the key
developments of modern plasticity theory needs to be emphasized. In terms of using
modern computational tools to solve important industrial problems such as forming
etc., the works of Anand and coworkers need to be highlighted; (see e.g. Anand and
Kothari, 1996; Gearing et al. 2001; Gu et al., 2001; Balasubramanian and Anand, 1998).
A detailed analysis of complex mechanical behaviors at nite strains (mostly in application to polymers) coupled to modern thermodynamics can be found in the works of
Drozdov and Dortmann (2003); also see Khaleel et al. (2001) and Khaleel (2001).
With regard to aluminum alloys this analysis is performed following Richmond
(1986).2 Then a transition is made to strain-gradient plasticity as an attempt to
introduce material length scales and to overcome the well-known limitations of the
classical theory, and to spatially coupled non-local variants of theory in general.
An interesting mathematical analogy between strain-gradient plasticity and concentration-gradient thermodynamics is further explored. This analogy has lead us to
assume that some general scenarios of symmetry-breaking transitions resulting in
the formation of microstructure, both thermodynamically- and kinetically-driven,
may exist in dierent physico-chemical and mechanical systems; see Section 4. In
Section 5 an outline of phase-eld theory is presented, with special emphasis on
the problem of simultaneous modeling of phase and defect microstructures, precipitates and dislocations. Several examples of microstructure evolution are discussed, including recent developments (spinodal decomposition, grain growth in
one-component alloys, precipitation of metastable Al3Li in AlLi system in the
presence of dislocations, etc.) (Wen et al., 2002; Vaithyanathan and Chen, 2000; Hu
and Chen, 2001a,b; Rodney and Finel, 2001; Wang et al., 2001a,b). A detailed analysis
of the interaction of phase microstructures with elastic and superelastic stresses and
2
The pioneering works by G.I. Taylor and his classical model for plastic ow under the conditions
of multiple slip, particularly in metallic alloys should be mentioned in this context (see e.g. Taylor,
1938a,b).

368

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

instabilities in such systems are analyzed by Roytburd and Slutsker (2002) and
Slutsker and Roytburd (2002). The paper is concluded by the analysis of possible
future work that would be aimed at understanding phase and defect microstructures
in aluminum alloys.

2. Schools of thought in modern thermodynamics and their relation to the problem


of microstructure
2.1. Equilibrium thermodynamics
The main breakthrough achieved by Gibbs was the understanding that at equilibrium a system of coexisting phases can always be characterized by some appropriate potential functions that achieve an extremum, and the derivation of many
powerful laws (e.g., the Phase Rule, the GibbsDuhem equation) that could be applied
to understanding the behavior of heterogeneous mixtures (Findlay, 1904; Van der
Vaals and Kohnstamm, 1927). A reader interested in the history of thermodynamics is
referred to the excellent texts of Rukeyser (1942), Jae (1958), and Wheeler (1951).
However, the classical Gibbsian thermodynamics does not pursue the goal of
incorporating microstructure of heterogeneous systems into its formalism, i.e. the
spatial distribution of phases. While the theory can predict the concentrations of
components and the relative amounts of the competing phases using the common
tangent construction and the lever rule in a heterogeneous mixture at equilibrium, it
does not pursue the goal to characterize its microstructureone of the key properties for structural materials such as aluminum alloys. The main reason is that Gibbsian formalism does not contain intrinsic length scales, and considers all interfaces as
mathematical surfaces that are innitely thin. Of course, the classical Gibbsian
thermodynamics is internally consistent, it is just that Gibbs purposefully chose not
to deal with interfaces of a certain nite width (e.g., in his theory of absorption),
neither can it predict their characteristic length scale(s). In the asymptotic limit both
the sharp interface and diuse interface formalisms coalesce, and that has lead many
researchers to believe that there is no any fundamental dierence between the two.
However, in the models with sharp interfaces it is impossible to observe the nucleation of a new phase, or its complete disappearance.
In its classical form it is hardly applicable to the description of solids. Signicant
diculties arise when one tries to transfer this formalism developed for uids, to
solid systems. Attempts to apply Gibbsian thermodynamics to this situation may
result in a phase rule which depends on a coordinate system, because the potentials that one must use are not rotationally invariant and, consequently, the governing equations are not frame-indierent (Glazov et al., 1999). Thought
experiments with adding atoms to dierent crystallographic planes of aluminum and
other metals (Cahn, 1994), provide valuable insights into the nature of the problem.
The diculties of adjusting thermodynamics for the description of systems with
microstructure resulted in attempts to introduce some modications. At this point
further development of thermodynamics as related to patterning and microstructure

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

369

has divided into two major branchesthe Prigoginian school of strongly nonequilibrium thermodynamics3 as given by Nicolis and Prigogine (1977), Prigogine
(1947) and Haken (1983), and the CahnHilliard school of equilibrium concentrationally non-uniform systems (Cahn, 1994; also see Cahn and Hilliard, 1959;
Larche and Cahn, 1985; Hilliard, 1970; Wang et al., 1991). It is important to mention several other developmentsRational Thermodynamics, Extended Irreversible
Thermodynamicsand their relation to the problem of microstructure evolution.
2.2. Strongly non-equilibrium thermodynamics
Prigogine started his research, mostly with Defay, which was devoted to understanding the classical thermodynamics and molecular physics of dierent physicochemical systems and laterof strongly nonequilibrium thermodynamics (Thermodynamic Theory of Capillarity and also closely related Dufour and Defays 1963
fascinating Thermodynamics of Clouds, Molecular Theory of Solutions, Chemical Thermodynamics (Defay and Prigogine, 1951; Prigogine, 1957; Prigogine and
Defay, 1950, etc.). The Brussels School of non-equilibrium thermodynamics was
based ideologically on the famous work of De Donder, Thermodynamic Theory of
Anity (De Donder, 1928). De Donder was the rst to understand how time could
be correctly incorporated into thermodynamics, and to have introduced the correct
measure of chemical anity into it. In this sense he was undoubtedly one of the
forefathers of all modern non-equilibrium thermodynamics.
One of Prigogines numerous accomplishments was the fundamental theorem of
the minimum entropy productionhe has demonstrated that for steady-state physico-chemical systems in the linear non-equilibrium regime a generalized potential
(entropy production) exists, which attains a minimum value. However, it was inferred that by virtue of this theorem, patterning is excluded in equilibrium and weakly
non-equilibrium systems. As a result Prigogine went on to propose to analyze the
bifurcation diagrams for the systems of partial dierential equations (that described
mostly uids) and has come to the conclusion that all patterning can happen only far
from equilibrium. An example of a dynamical, kinetically-driven patterning in space
and its comparison to experimental data, is presented in Figs. 2 and 3 which illustrate the formation of the ladder structure of persistent slip bands in a CuAl alloy,
while an example of dynamical patterning in time is given in Fig. 4. Prigogines work
was of the utmost importance in biology (the problem of morphogenesis), chemistry
(the Brusselator, BelousovZhabotinsky oscillatory chemical reactions etc.), turbulence (Benard instability) and related areas, and he was awarded a Nobel Prize in
chemistry for it. However, in spite of some later claims (Nicolis and Prigogine,
1989), it never dealt with solids, and, even less so, with solids with microstructure.
These ideas were later used by other researchers (Aifantis, Kratochvil, Kubin,
3

A somewhat dierent approach was adopted by Coleman, Noll (1964) and Truesdell (1969) in their
rational thermodynamics, which simply views the Second Law of thermodynamics as a powerful tool to
generate dierent types of constitutive equations (the ClausiusDuhem inequality)this approach will
be briey discussed below.

370

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

Fig. 2. An example of a kinetically (dynamically)-driven pattern: patterning in space. Computer simulation of the ladder structure of persistent slip bands in a CuAl single crystal. Dislocation walls comprising the pattern are simulated as sharp peaks in the density of immobilized dislocation dipoles (Glazov
et al., 1995).

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

371

Estrin, Walgraef, Hahner, Zaiser) who applied modern non-equilibrium thermodynamics to spatio-temporal instabilities in metallic alloys (Walgraef and Aifantis,
1985; Aifantis, 1986; Romanov and Aifantis, 1993; Estrin and Kubin, 1988).
2.3. Further development of modern non-equilibrium thermodynamics: Classical,
Rational and Extended variants
The variant developed by Prigogine and his school can be called Classical Irreversible Thermodynamics (CIT) and goes back, in addition to the works of Defay
and De Donder cited above, to the works of De Groot and Mazur (1962), Meixner
(1943), Eckart (1940) and, of course, Onsager (the OnsagerCasimir reciprocal
relations for linear non-equilibrium processes, 1931). CIT was developed under the
following fundamental assumptions: (1). local equilibrium; (2) the existence of a
non-negative rate of entropy production; (3) linear constitutive laws; (4) the OnsagerCasimir reciprocal relations between phenomenological coecients relating

Fig. 3. Comparison of the modeling results to experimental data of Laird (1986. Mater. Sci. Eng., 81,
433). Grain size L=13 mm (according to linear stability analysis, it eectively corresponds to innity and
thus reproduces experimental observations well (Glazov et al., 1995).

372

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

conjugated forces and uxes in the system. Limitations of the CIT are mainly related
to the local equilibrium hypothesis: it may be inapplicable to very fast processes (i.e.
high frequencies), and for too short wave lengths, i.e. when a specic micro-length
scale is needed to describe a non-equilibrium process. It is quite conceivable that not
only conserved equilibrium variables are necessary for a detailed description of
such situations, but also some new independent variables (gradients or uxes).
Moreover, when the Fourier law of heat conduction is introduced into the energy
conservation law, one gets a parabolic equation for the evolution of temperature,
which implies that disturbances in such systems will propagate with innite speed.
Partly as an attempt to overcome these diculties, Coleman and Noll (1963),
Coleman (1964), Noll (1975), and Truesdell (1969) developed a dierent formalism,
which was called Rational Thermodynamics (RT). The main objective of that
approach was to give a general method for deriving constitutive equations. RT is
based on the following fundamental postulates:

Fig. 4. An example of a kinetically (dynamically)- driven pattern: patterning in time. Nonlinear


dynamic modeling of the cyclic analog of the PortevinLe Chatelier eectthe so-called Laird bursts
(Glazov et al., 1997).

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

373

1. Absolute temperature and entropy are considered as irreducible objects,


which dont have a precise interpretation in this phenomenological formalism;
2. Materials are supposed to have memory in the sense that the values of
parameters are determined not instantaneously, but depend on the past history of the system;
3. The Second Law of thermodynamics serves as a restriction on the form of
possible constitutive equations.
These assumptions lead to the set of the basic axioms of RT: (1) the principle of
equipresence; (2) the principle of memory and heredity; (3) the principle of local
action, and (4) the principle of material frame indierence (objectivity). A detailed
description of these principles can be found, e.g., in Truesdell (1985). The last principle means that any constitutive law must be objective, i.e. it should not depend on
observer. In particular, any time-dependent constitutive law should be expressed in
terms of objective time derivatives, e.g. Jaumann, covariant (lower convected), contravariant (upper convected) etc.
This approach also has limitations. First, temperature and entropy are not
objects, which have a clear physical meaning. Second, RT predicts unphysical
behavior for some classes of rheological materials. The requirements of material
frame indierence are not satised in several disciplines: turbulence, rheology, and
molecular hydrodynamics. There is ample experimental evidence that turbulence in
a non-inertial frame is a totally dierent phenomenon in comparison to an inertial
frame (Jou et al., 1993). Analogous doubts have been expressed for visco-elastic
materials. However, it needs to be emphasized that the concept of frame indierence
as specied in rational thermodynamics was not supposed to characterize complicated physical systems with internal inertial forces.
Finally, the third direction in irreversible thermodynamics is connected with the
names of Jou et al. (1993); it is called Extended Irreversible Thermodynamics
(EIT). This approach goes beyond the classical theory of irreversible processes, in
particular, the local equilibrium hypothesis is abandoned and, unlike in Gibbsian
thermodynamics, the gradients of temperature and pressure act as independent
variables. It will be more convenient to give a short description of this theory in
relation to plastic phenomena on the mesoscale later.
2.4. Equilibrium thermodynamics of concentrationally non-uniform systems
In mid-fties Hillert worked at MIT on understanding the thermodynamics and
kinetics of nucleation in some metallic systems. As in an earlier work of Ono (1947),
who most probably was the rst to oer a consistent treatment of the problem of
equilibrium interface width, Hillert realized that in order to describe the problem
adequately, the thermodynamic formalism should incorporate characteristic length
scale(s). However, instead of a rigorous continuum approach, systems of coupled
dierence equations were solved in order to obtain equilibrium interface widtha
procedure which does not allow one to write a close-form solution of the problem.
In addition, the approximation of a regular solution model was used, with its wellknown limitations.

374

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

Some time later Cahn and Hilliard (1958) published their papers on equilibrium
thermodynamics of concentrationally non-uniform systems. Their main idea was
that of a diuse interface, i.e. that in order to understand systems with interfaces
of some nite width, it is necessary to employ the concentration gradient (or the
gradient of an order parameter), rc, as an independent variable. As they pointed
out, the treatment. . . is analogous in some respects to those used for the evaluation
of energy of magnetic and ferroelectric domain walls, of the interface between a
metal in its normal and superconducting states. The idea of the double-well free
energy functional belongs to Ginzburg and Landau, and its detailed description can
be found in Langer (1992) and Landau and Lifshits (1961).
Later a similar approach has been accepted, e.g., by the mechanical engineering
communitythe strain-gradient theories of plasticity by Fleck et al. (1994), and
recently Bassani (private communication), also see footnote 6. In terms of strength
gradient as a more direct surrogate for structural gradient, this idea has been developed by Li and Richmond (1996) in their work on intrinsic instability and nonuniformity of plastic deformation, and further developed by Xue et al. (2002).
The idea of Cahn and Hilliard was that for concentrationally non-uniform systems the local free energy of a molecule should depend not only on concentration
itself, but also on its gradient(s).4 The area of validity of this approach (see Fig. 1) is
micro-continuum. In the continuum approximation it is necessary to assume that
the composition gradient is small compared to the reciprocal distance between the
particles. Following Cahn and Hilliard (1958):
X  @c  X 1  @2 c  1X 2  @c 


2
f c; rc; r c; . . . f0 c
Li
kij
k 

@xi
@xi @xj
2 i;j ij
@xi
i
i;j
 
@c

...
@xj
Applying symmetry considerations for the derivation of the phenomenological
coecients for
 cubic lattice Cahn and Hilliard (1958) arrived at the following
expression: f c; rc; r2 c; . . . f0 c k1 r2 c k2 rc2 . . . Integrating over the
volume V and using the divergence theorem for the second term of this decomposition,5 one can get the key expression for the free energy functional of a concentrationally non-homogeneous system:
4

In the qualitative form this idea was expressed by Van der Vaals in his doctoral thesis as early as in
the end of the XIX century (1983). Perhaps one could call his theory a gradient theory, but not a phase
eld approach inasmuch as an additional phase-eld parameter would still need to be introduced (the
authors are grateful to one of the Referees for this valuable comment).
5
Applying the divergence theorem, one can get the following expression:


V


k1 r2 c dV  dk1 =dcrc2 dV k1 rc  ndS
V

Since Cahn and Hilliard (1958) were not concerned with eects at the external surface, a boundary of
integration was chosen in such a manner that rc  n was set to zero at the boundary. Therefore, the surface integral vanishes, and it could be employed to eliminate the term r2 c in Eq. (1).

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

Fc


f0 krc2 dV

375

where f0 is the free energy density for a concentrationally uniform system. The coefcient k is necessary on the dimensional grounds, and it is via this coecient that a
length scale comes into the constitutive equation(s) naturally.
This is the key expression because it provides one with an ability to study equilibrium systems with microstructure. Specically, the dynamics of spinodal decomposition can be considered within the framework of partial dierential equations
that describe such gradient systems with the potential function F[c] as a Lyapunov
functional. An impressive demonstration of this statement was the successful modeling of kinetics of spinodal decomposition by Cahn and Hilliard by Hilliard (1970)
in his extensive review paper. The authors employed the modied diusion equation
for that goal:
@c
@2 c
@4 c
k1 2  k2 4
@t
@x
@x

It should be noted that Eq. (2) is the linearized CahnHilliard equation if k1 and
k2 are constants. A more complex non-linear equation can be obtained employing
directly the GinzburgLandau double-well potential (Langer, 1986). However, for
the purposes of the present paper the analysis of a simpler expression (2) is sucient.
Phase patterning as described by Eq. (2) arises as a result of competition of the second and the fourth gradient terms Hilliard, 1970).

3. Inelastic constitutive relations for solids: internal-variable theory and its


application to metal plasticity6
3.1. Preliminary remarks
The goal of this paper is to establish the common basis for considering dierent
patterning phenomena (microstructure) in physical and mechanical metallurgy of
aluminum alloys. To advance toward achieving this goal, it is important to understand the thermodynamic structure of the internal-variable theories of plasticity.
The concept of internal variables which would represent dierent aspects of microstructure turned out to be a very powerful one, and it is yet another possibility for
taking microstructural evolution into account. Below we give mostly an outline of
Rices rigorous formalism because it is based on the laws of thermodynamics and
thus has a very solid foundation (Rice 1970, 1973, 1975).
The approach with internal variables views inelastic deformation of a given
sample of material of the type considered under macroscopically homogeneous
strain and temperature as a sequence of constrained equilibrium states: The state of
6

Dr. Braginskys contributions in preparing Section 3 of the present paper are gratefully acknowledged.

376

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

the material sample at any given time in the deformation history is taken to be fully
characterized by corresponding values of the strain and temperature and the collection of internal variables which mark the extent of microstructural rearrangement
within the sample (Rice, 1973). This means that a material is considered simultaneously on two dierent scales- macroscopic, where the process of straining takes
place, and microscopic, where structural rearrangements occur.
The general formalism is developed in terms of a nite number of discrete scalar
internal-variables i i 1; n, and each of them is supposed to characterize the extent
of some local structural rearrangement. This approach goes within the framework of
other internal-variable treatments of inelasticity (e.g. De Groot and Mazur, 1962;
Coleman and Gurtin, 1967), at least with respect to homogeneous deformation.
However, the dierence is in the interpretation of the internal variables. In Rice
(1973) each variable is thought of as describing a specic structural rearrangement
occurring at a local site within the material sample. In a more traditional approach
internal variables are interpreted as some average measures of the structural rearrangements taking place at the many operative sites within the material sample.
The importance of structural variables is in the general normality structure of
constitutive laws. This can be obtained if one infers that the rate, at which a microstructural change proceeds, is governed by its associated thermodynamic force. Once
the normality structure is established, it is no longer necessary to consider the
dependence of the ow potential on the specic structural-variables, and they can be
replaced by more conventional internal variables of the averaging type.
3.2. Characterization of constrained equilibrium states
Consider constrained equilibrium states based on the assumption that internal
variables could be held at any denite set of values by imposition of appropriate
constraints, with the material sample attaining an equilibrium state corresponding
to a prescribed stress S or strain E and temperature  (Rice, 1973). Introducing the
specic free energy and its Legendre transform with respect to strain, complementary
energy, , one may obtain:
E; ;  u  
;

S; ;  E :

@

@E

with u standing the internal energy density of the sample and


its entropy density.
For the variation of the complementary energy one gets: E: S V10 f 

where V0 is the volume of a material sample and f ( =1,n) denote the thermodynamic forces acting on the internal variables.
Then one can dene a plastic portion ( E)p, and elastic (or thermo-elastic) part
( E)e as the part, which would result from the change in stress and temperature, if
the internal variables were held constant. As always, E=( E)e+( E)p. It is further
inferred that a macroscopically homogeneous deformation processes may be
approximated as a sequence of constrained equilibrium states, each fully characterized by the values of E, ,  at the corresponding moment of time (Rice, 1973). Thus
one can write down the following expression:

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

:
e @2
:
@2 :
:S
;
E
@S@S
@S@

:
p 1 @ f S; ;  :

E 0

V
@S

377

3.3. A ow potential for the inelastic strain-rate


The most important conclusion that can be drawn from the internal-variable theory of plasticity is that a normality structure emerges in macroscopic constitutive
relations for a special class
of
:
:  kinetic relations described above (Rice, 1975). This
statement implies that     f ; ;  for  1; 2; . . . ; n. The stress derivatives of
 give an expression for the plastic strain rate. Indeed,
:
p
@OS; ; 
1 @f S; ;  :
0
 f; ;  E
@S
V
@S

which proves the existence of the ow potential.


3.4. Structural variables and internal variables of the averaging type
Although the concept of internal variables as introduced in Rice (1970, 1973,
1975), Kestin and Rice (1969), is very useful in understanding the microstructural
basis of normality structure for plastic ow, it is hardly applicable in practice. There
is a need to reduce a large number of initial internal-variables, to a smaller number of averaged variableshistory-dependent parameters representing microstructure Richmond (1986).
Suppose that a set of variables 1 ; 2 ; . . . ; n exists which, together with S or E and
, completely determines the energy and entropy densities u; ; ;
of the material.
Now a set of generalized forces conjugate to the averaging variables (g1, . . ., gk) can
be introduced, and one can write for the variation of the internal energy density
S: E  g  
u. In analogy to the previous consideration in terms of
specic structural variables, one may write:
@ S; ; z
;
@ S
@ S; ; z
@ ES; ; z
@ g S; ; z
g
; Ep 

@ 
@ 
@ S

3.5. Microstructure-based modeling of deformation processes


Rices approach was very successful in explaining how the concept of internal
variables could explain the underlying normality structure of ow, but hardly could
be used in practice because of the large number of internal-variables, (Rice, 1973). A
transition to internal-variables of the averaging type was necessary which would
give a possibility to describe dierent deformation phenomena from a practical
point of view.

378

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

In many cases models of deformation processing employ primarily ideal materials


like perfectly plastic solids and nonlinear viscous solids whose current response to
applied stresses is completely independent of prior thermo-mechanical history. The
primary concern of such models is to predict the changes in product geometry
resulting from the application of specied forces or alternatively, to predict tool
forces required to achieve certain specied changes in geometry (Richmond, 1986).
However, most shape changes are accompanied by changes in material structure,
and a more holistic approach must include the development of quantitative constitutive equations describing the relationships of processing and product histories to
changes in material structure.
In a brief description of microstructure-based modeling approach that follows
below emphasis will be made on an ideal case of homogeneous plastic deformation. A
transition to non-homogeneous plastic ow can be made if one combines the constitutive equations with the laws of mass, momentum and energy conservation to
form a set of partial dierential equations, and then to subject these to boundary
conditions which describe the geometry and the heat and force transfer at the free
surfaces and tool/workpiece interfaces of a specic process [1]. In other words, it
becomes necessary to introduce spatial coupling into the constitutive equations and
the corresponding boundary value problem.
The properties Pi of a product resulting from a homogeneous process may be
assumed to be functions of a small list of microstructural parameters, which should
be understood in the sense of internal-variables of the averaging type. Following
Richmond (1986, 1987), one can write down: Pi fi S1 ; S2 ; . . ., where the properties might be, for example, tensile strength or tensile ductility, while microstructural
parameters might be porosity and/or dislocation density. These microstructural
parameters, in their turn, are supposed to be functions of prior history of strain rate
:
and temperature, i.e. Sj Fj "; T
It may be more practical to assume that the current stress, , is a function of the
current values of the microstructural parameters as well as of the current strain rate
and temperature. Consequently, one gets an equation of state, which describes such
:
constitutive behavior (Richmond, 1986):  g"; T; S1 ; S2 ; . . .. Finally, to complete
the description of the: material behavior, it is also necessary to assume that the current rate of change, Sj , of the microstructural parameters can be described by a set
of constitutive equations called equations of evolution:
:
:
S1 h1 ; "; T; S1 ; S2 ; . . .
:
:
S2 h2 ; "; T; S1 ; S2 ; . . .
:::::::::::::::::::::::::::::::::::::
:
:
SN hN ; "; T; S1 ; S2 ; . . .
7
Constitutive equations can be integrated for arbitrary processing histories to
determine resulting changes in strain and microstructure (Richmond, 1986, 1987),
and considered a generic variant of the theory featuring microstructure of a
mechanically deformed material element.

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

379

Several applications of this general formalism include: prediction of time for


solution heat treat (an approach based on coupled evolutionary equations for
volume fraction and number density of each soluble phase) Shuey et al. (1996); the
evolution of damage and fracture in iron compacts with various initial porosities
(Spitzig et al., 1988). Fig. 5 illustrates a good agreement between the results of prediction of ow curve and the experimental results for commercially pure aluminum
(Sample et al., 1990).
The formalism of the internal-variable plasticity theory, in analogy with Gibbsian
thermodynamics, does not contain intrinsic length scales. Since all the changes in
material microstructure that accompany deformation processes are modeled using
systems of ordinary (linear and non-linear) dierential equations, it is not possible to
describe localization phenomena and the formation of patterns realistically. Consider, for example, a well established approach to strain localization phenomena.
According to Rudnicki and Rice (1973), Rice (1977) and Asaro and Rice (1977),
these are viewed as a shear band bifurcation that results from a constitutive and
geometric instability. When normality structure governs ow (Schmid behavior, see
above) bifurcations are precluded at positive values of the instantaneous slip-system

Fig. 5. Example of ow curve predictions and corresponding data for deformation at a strain rate of
0.001/s and various temperatures using a single internal-variable model (Sample et al., 1990).

380

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

hardening modulus, Drucker and Li (1992). As noted by Qin and Bassani (1992),
such non-normality in the constitutive description can signicantly ease the condition at which the shear bands develop (the issues of non-normality in plastic ow of
crystals have been studied extensively by Drucker and Li (1992). Alternatively, it
becomes necessary to introduce small secondary slips in order to make strain localization possible at positive values of hardening moduli. Thus, under single-slip
conditions, non-Schmid eects become necessary to trigger shear bifurcations at
positive hardening (Qin and Bassani, 1992). Recently an interesting description of
how characteristic length scales could be incorporated into the classical plasticity
theory was proposed by Bassani (see footnote 6).
The approach described below views localization phenomena and size eects in
plasticity from a somewhat dierent point of view (Fleck et al., 1994). In discussing
the FleckHutchinson theory, one should mention the couple-stress theory of Toupin (1962) and Mindlin (1964), which was used in their approach. Instead of
exploring the consequences of geometric instabilities on a yield surface, it directly
introduces a characteristic length scale via a modied energy measure. Since the
same approach to patterning in materials science has been adopted somewhat earlier
in thermodynamics of equilibrium concentrationally non-uniform systems, we will
describe it in some detail.
3.6. Strain-gradient plasticity
Cahn and Hilliard (1958) pointed out that the treatment. . .is analogous in some
respects to those used for the evaluation of energy of magnetic and ferroelectric
domain walls, and of the interface between a metal in its normal and superconducting states, thus pointing to the deep physical roots of their approach.
Expression (1) is widely known in physics as the GinzburgLandau functional,
and it was also used, for example, to describe the so-called Abrikosov vortices
of magnetic ux in type-2 superconductors and domain structures of dierent
physico-chemical nature. The reason for this remarkable similarity is that all the
phenomena described above belong to one universality class (Goudreche, 1993). As
a matter of fact, Van der Vaals (1893) was the rst to introduce the gradient
approach to materials science in his thesis Thermodynamic Theory of Capillarity
under the Hypothesis of a Continuous Variation of Density. Below a description of
the strain-gradient plasticity formalism is given, mostly following the work by
Fleck et al. (1994).
3.6.1. Discussion of the constitutive relations of the couple-stress plasticity theory
Only the deformation theory version of plasticity was considered following mostly
the work by Fleck et al. (1994), i.e. no formal distinction between elastic and plastic
parts of strain was introduced. Following this approach, the whole exposition will
be limited to the case of small deformations and small rotations. In many cases of
practical interest rotations are large, as are the plastic strains. A more detailed
exposition of these important issues can be found in the work by Gurtin (2002). The
main assumptions adopted by Fleck et al. (1994) were as follows:

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

381

1. The existence of the strain energy,i.e. the existence of the potential such that
ij @w=@"ij and mji @w= l1 @ij , where mji is the couple-stress tensor,  ij is
the Cauchy stress tensor; w is strain energy, and l is the length scale;
2. Strain energy depends only upon the single scalar strain measure ", which is
the function of the eective strain and eective curvature: "2 "2e l 2 2e
One of the approaches to description of micro-continuum is the use of the Cosserat
type theories; it will be shown specically how the constitutive equations described
above can be derived using the assumption of some special features of continuum
points, i.e. their ability to be deformed and rotated. However, now we will be mostly
concerned with a connection between the models of strain-gradient type, and the
theories describing material behavior using the notion of dislocations (Sedov and
Berdichevsky, 1967; Nye, 1953; Kroner, 1962).
3.6.2. Length scale and geometrically necessary dislocations
Incompatibility of the plastic part of the deformation gradients may result in the
storage of geometrically necessary dislocations (Fleck et al., 1994), and also Kroner
(1962), Ashby (1970) and Basinski and Basinski (1966).7 The magnitude of the
plastic strain gradient is of the order of the average shear strain in the crystal divided
by the local length scale l of the deformation eld. For compatibility of the deformations these strain gradients require the presence of geometrically necessary dislocations: G 4
value
bl where  is the macroscopic plastic shear strain and b ispthe

of the Burgers vector. For the macroscopic shear yield stress:  Gb G S ,


where G is the shear modulus, G is the density of geometrically necessary dislocations and S is the density of statistical dislocations.
Length scale l is some characteristic length of material and loading. For instance,
in polycrystals it may be related to grain size, in metals containing non-deformable
particlesto particle separation. The smaller the length scale of the gradients, the
more important these eects become.
3.6.3. Single crystal formulation for geometrically necessary dislocations
Consider a crystal lattice embedded within a solid. We assume that the material
ows through the crystal lattice by dislocation motion and that the lattice undergoes
rotation and elastic stretching. Then the considerations of incompatibility may justify the existence of geometrically necessary dislocations.
Consider, for example, a relative displacement dui of two material points separated by dxi (Fig. 6) (Fleck et al., 1994). This displacement can be decomposed into:
E
S
dui duSi duR
i dui , where the relative displacement due to slip dui ij dxj ;
R
relative displacement due to lattice rotation dui ij dxj ; relative displacement due
to elastic stretching duEi "elij dxj .
A slip system ( ) can be dened by the vectors of the slip direction (s( )) and the
direction of the slip plane normal (m( )). The slip tensor associated with the amount
7

The authors are grateful to the Referee of their paper for this important correction. As the Referee
pointed out, in simple shear one can have huge strain gradients but no geometrically necessary dislocations.

382

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

Fig. 6. The elasticplastic deformation of a single crystal (one active slip system) (Fleck et al., 1994).

P
of slip  ( ) on each of the active slip systems is: ij  si mj . The density of
geometrically necessary dislocations is related to the net Burgers vector Bi associated
with crystallographic slip. Make a cut in the crystal
in order to produce a surface S

of outward normal n. As the net displacement dui along the closed curve within the
material (i.e. complete displacement eld must be compatible), incompatibility
in the

E
relative displacements due to elastic stretching and lattice rotation G duR

du
i
i is
matched by the incompatibility in the relative displacement due to the slip Bi, which
can be obtained as a resultant displacement due to slip after completion
of the cir
S
cuit around the curve ,
the
boundary
of
S.
Thus,
B

du


dx
.
Using the
i
ij
j
i
G
G

Stokes theorem, Bj S jk nk ds, where jk is the Nyes dislocation density tensor that
gives a direct measure of the density of geometrically necessary dislocations. It is
related to the distribution of dislocations within a crystal (Nye, 1953).
3.6.4. Continuum measure of geometrically necessary dislocations
Dislocation density tensor is not very useful for the purposes of developing a
phenomenological continuum theory because ik can only be dened with reference
to specied slip systems. An alternative measure of dislocation density exists, which
is a well known curvature tensor dened by rotations due to displacement eld.8
This curvature tensor is dened in the simplest model of the Cosserat type couple
stress theory (see below). In this theory particles from which continuum consists are
considered rigid (non-deformable), they can only rotate due to the displacement
of

their centers. Consider the linear deformation tensor "ij 0:5 ui;j uj;i and introduce the curvature tensor: ij i;j 0:5eikl uk;lj 0:5eikl "kl;j , where i 0:5eikl uk;l is
the rotation vector associated with a displacement eld ui. Since the divergence of i
vanishes (i is a rotation) kk=0 and, hence, ij is deviatoric. One can decompose
the curvature tensor, as well as the deformation tensor, into elastic and plastic parts

It should be noted, however, that in the original development by Kroener, the dislocation density
tensor with the curvature tensor were used as a measure of incompatibility and represented geometric
objects that could be used to nd the (mean) long-range stress eld. The authors are grateful to one of
the Referees for this valuable comment.

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

383

=el+pl. Then a connection can be established between the components of this


tensor and the densities of the geometrically necessary and statistical dislocations,
both for stage I and stage II processes, when slip takes place on just one slip system
or when slips are induced on secondary systems, (Fleck et al., 1994). The most recent
developments in multiscale plasticity, and strain gradient theories can be found (e.g.,
in Zbib and de la Rubia, 2002).
After a review of the concept of strain-gradient plasticity, we are nally in a
position to give the comparison of the dierent approaches to patterning (symmetry-breaking transitions with emerging spatio-temporal scales) in material science
systems.
3.7. Patterning in materials science, equilibrium and strongly non-equilibrium
patterns: dierences and similarities
The consequence of the introduction of free energy functionals which depend on
order parameter gradients squared in materials science have been very signicant
and resulted in such ideas as motion by curvature and phase-eld approach.
Below we will consider briey how specically this idea has been implemented in
strain-gradient plasticity, in the thermodynamics of concentrationally non-uniform
systems and in other systems of dierent physico-chemical nature that exhibit a
tendency to patterning.
3.7.1. Strain-gradient plasticity and thermodynamics of concentrationally
non-uniform systems
In the classical plasticity theory with internal variables one faces the same problem
as in the Gibbsian thermodynamicsit is impossible to represent realistically areas
of slip localization, which are supposed to be innitely thin. However, strain-gradient plasticity formalism does give a possibility to study systems with microstructure. The underlying reason for that is that for the measure of plastic strain
(energy) one can write down the following expression: "2 "2e l2 2e , where e is the
strain gradient and l is a material length scale which is required on the dimensional
grounds. If one introduces the measure of strain this way (which can be justied on
the basis of more microscopic considerations leading to the concept of Cosserat-type
continuum), it becomes possible to describe a certain class of plastic phenomena that
involve size dependence (e.g. the formation of macroscopic shear bands, indentation
hardness of metals on the nanoscale etc.). The analogy with the measure of free
energy density in the Cahns approach to systems with interfaces of nite width f
f0 krc2 is complete. In plasticity theory this approach is called strain-gradient
plasticity, in materials sciencephase-eld theory. While internal-variable theories of plasticity which originated with the work of Rice (1970, 1973, 1975), Argon
(1975) and Richmond (1986) and later were developed in application to Al-alloys by
Shuey et al. (1996) and Sample et al. (1990) are well-known, it is interesting to see
what degree of generality can be achieved with the phase-eld approach applied
to the problems of phase and defect microstructures in Al-alloys. Several examples
will be given below, following mostly the works by Cahn and Hilliard (1958), Larche

384

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

and Cahn (1985), Chen (2002), Chen and Wang (1996), Wang and Chen (1991),
Khachaturyan (1983) and Becker (1995, 1996); also see footnote 8.
3.7.2. Patterning in mechanical systems of the Cosserat type and in other systems of
dierent physico-chemical nature including dislocation assemblies
In order to obtain patterning in equilibrium or non-equilibrium systems, one
possibility is to use formalisms that contain intrinsic length scales. There is no any
unique way how a material length scale could be incorporated into constitutive
equations: typical this can include the formation of both geometrical and material
instabilities. For example, a geometric instability can be illustrated by means of
incorporation of a vertex on the yield surface; a vertex is present naturally for crystalline yield surfaces, and including these eects can lead to predictions of strain
localization (Becker, 1996). The second case could be illustrated by localization of
plastic slip due to crystal boundary energy (Becker, 1995), or reaction+diusion
formalism for the description of ladder structure of persistent slip bands, etc. In all
these cases it is assumed, within the framework of continuum phenomenological
approach, that material points do not possess any special properties. However, a
dierent approach to patterning arises if this assumption were abandoned. Below patterning in mechanical and physico-chemical systems will be discussed on the basis of
a fundamental assumption that material points in such systems possess certain
special properties.
3.7.3. Mechanical systems of the Cosserat type
This continuum approach was originated by Duhem; however, its rst systematic
development belongs to Cosserat and Cosserat (1909), and for this reason the theory
is usually referred to as Cosserat type formalism.
Opposite to the classical continuum mechanics where a material point is considered to be dimensionless, in the Cosserat type continuum each material point is
an innitesimal particle, which can be deformed and rotated. Models of this type
were used in mechanics mostly for the description of rods and shells because some of
their geometrical sizes are very small relatively to the other(s). Stresses acting on this
small scale in this limit give a zero force and a non-zero moment, which makes it
impossible to consider these structures without body moments. Models considering
body moments must consider rotations of material points as work conjugate to
moments. It can be shown from thermodynamics (Sedov and Berdichevsky, 1967)
that these considerations also lead to the concept of Cosserat type-continuum.
A rigorous approach to the derivation of constitutive equations of the Cosserattype is based on Truesdells rational thermodynamics (Truesdell, 1969). Below a
more mechanical approach will be described (Mindlin, 1964). In the Cosserat-type
continuum each material point represents a microvolume that can be deformed and
rotated. Displacement of the particle center can be dened in a standard way
(Mindlin, 1964): u=xX, where x and X are position vectors in the deformed and
reference congurations, respectively. For micro-displacements inside the particle:
u0 =x0 X0 , where x0 , X0 are position vectors inside the particle in the deformed and
reference congurations, respectively. Assuming homogeneous deformation for each

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

385

particle,
it is possible

 to dene the strain tensor in the macro-continuum:

 "ij
0:5 @ui =@xj @uj =@xi For macro-rotation tensor: !ij 0:5 @uj =@xi  @ui =@xj . The
components
of the micro-strain and

 micro-rotation tensors are given by: ij
0:5 @u0i =@x0j @u0j =@x0i and ij
0:5 @u0j =@x0i  @u0i =@x0j . The tensor of relative deformation is given by ij @uj =@xi  ij , where ij @u0j =@x0 , and the micro-deformation gradient is given by ijk @ ij =@xi . The compatibility conditions (to ensure
single-valued displacements) are given below:
emik enlj

@2 "kl
0;
@xi @xj

emij

@ikl
0;
@xi


@ 
"jk !jk  jk ijk
@xi

It is assumed that particle rotations are independent of their center displacements.


Therefore strain gradient appears as an independent variable in the functions of
state. The model also requires that the work conjugate to the strain gradient be the
double stress tensor. The equilibrium equations and boundary conditions can be
written out as follows:

1
r    f  u;
r    F  r r  
3
t n    ;
Tn
9
where  and  are the symmetric and anti-symmetric parts of the stress tensor,  is
the double stress tensor,  is the double force per unit volume, T is the double force
traction per unit area, t is the surface force per unit area, and f is the body force.
The general formalism described above can be reduced to important particular
cases: (1) couple stress theory; and (2) modied deformation theory. In the rst case it
is assumed that a particle is rotated only due to the displacement eld of Cosserat
continuum. A surface element dS of the body can transmit both a force vector (TdS)
and a torque, (q dS). The surface forces are in equilibrium with the non-symmetric
Cauchy stress. This stress tensor can be decomposed into the symmetric part  and
anti-symmetric part, : Tj=( ij+ ij)ni. Following Koiter (1967), one can also introduce the couple stress tensor m the hydrostatic part of which is assumed to be zero,
qj=mijni. The equilibrium equations are:
ji;j ji;j fi 0;

mpi;p eijk jk Li 0

10

In the absence of body forces and body moments:  ij,j+ ji,j=0;  jk=0.5eijkmpi,p.
In the second case, i.e. for modied deformation theory, in the absence of couple
stresses, the deviatoric part of the symmetric Cauchy stress tensor s may be represented by a ve-dimensional vector. When couple stresses are present, s is replaced
by a 13-dimensional vector S=( s, l1, mT ), where mT is the transpose of m,
and S consists of the ve symmetric components of s and the eight components of
asymmetric deviatoric tensor l1 mT. When couple stresses are absent, the vedimensional deviatoric strain measure "0 "  1=3Itr" is replaced by the
13-dimensional vector E "0 ; l.

386

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

In general, the solid is assumed to be compressible in which case "kk60. When


couple stresses are incorporated, the strain energy density ! of the hyper-elastic
solid is taken to depend only upon the volumetric strain "m="kk and the scalar
q
invariant E 23 E  E. On the other hand, the scalar invariant E is related to the
invariants of " and  by the relation: E 2 "2e l 2 2e . In this equation, as well as in
the 13-dimensional vectors =(s, l1, mT) or in E "0 ; l,9 the length scale l is
necessary on the dimensional grounds. In a similar fashion, an expression for the
eective stress can be obtained: 2 e2 l2 m2e an intuitively expected result, (Li
and Richmond, 1996).
An important consequence is that the initial assumption about material points as
small particles, rather than dimensionless points, automatically resulted in the
appearance of a characteristic length scale in the constitutive equations which comes
via strain- (or stress-)gradient in the measure of energy for such systems. In this sense
the Cosserat-type theory serves as a microscopic justication of phenomenological
gradient formalisms on the mesoscopic scale.
3.7.4. Patterning in dipolar and/or multipolar systems of dierent physico-chemical
nature
Naturally, one might expect that the same basic trick, using the words of
Lavenda (1978), could work in other situations when continuum points possess
some special properties. Obvious candidates include: electric and magnetic dipolar
and multipolar systems, dislocations and their assemblies, or multiphase systems
with coherent conjugation of crystals. In all these cases the continuum points will
display long-range potentials of interaction with strong angular dependence.10 It is
because of this fundamental property that the Gibbs phase rule is not applicable to
such systems in the most general case.
For example, the electric eld of such systems is highly asymmetric (angular
dependence), and since the forces acting in the system are of the long-range nature,
in certain situations electromagnetic continua of this type may exhibit a tendency to
patterning. This is true for such dipolar and multi-polar systems as ferrouids,
magnetic bubble system, which can form bubbles, stripes and other types of superstructures (Langer et al., 1992; Halsey, 1993; Keller et al., 1987). An example of a
thermodynamically controlled pattern in lipid monolayers, a micrograph of the lipid
substance DPPC in the two-phase region, is presented in Fig. 7 (Keller et al.,
1987).
Another example is patterning in electro-rheological, magneto-rheological uids
and ferrouids (Langer et al., 1992; Halsey, 1993). Electro-rheological uids consist
of suspensions of electrically polarizable particles of size 0.1100 mm in insulating
solvents. Such uids develop brillated and columnar structures parallel to strong
electric eld.
9

In the latter case couple stresses are absent.


Of course, dislocations or electric charges do not self-organize at all times, and long-range interactions do not necessarily always result in systems with internal structure.
10

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

387

Fig. 7. Photomicrograph of a monolayer of DPPC in the two-phase region at high solid fraction
(McConnell and Moy, 1988).

Thus, if continuum points possess some special properties (e.g. represent Cosserattype micro-volumes in mechanical systems, or admit description in terms of multipolar expansions) length scales which are necessary for patterning both in equilibrium and non-equilibrium systems, come into the constitutive equations naturally,
and in all cases result in the measure of energy which contains such intrinsic length
scales.

4. Brief description of several problems of micro-structural evolution of aluminum


alloys that can be solved using the phase-eld approach
We are now in a position to discuss briey the main principles of the phase-eld
approach. Following Chen (2002), and also Chen and Wang (1996), it is useful to
outline the main steps in the description of phase microstructures with this
approach. Microstructures are described by a set of space- and time-dependent
variables (elds). These variables are called order parameters; the latter can be
conserved and non-conserved. An example of a conserved order parameter
(internal variable) is concentration in the problem of spinodal decomposition, or
solidication of a two-component alloy. The word conserved used in the description of such parameters implies that a conservation law must be obeyed Chen and
Wang (1996): @c=@t rJ, where the ux J can be determined as a gradient of

388

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

chemical potential, which, in its turn, is dened as variational derivative of the


Landau free energy functional.
If a eld variable
is not conserved (examples: ordering and anti-phase domain
coarsening, solidication of one-component alloys (Chen and Fan, 1995; Oono and
Puri, 1987; Wheeler et al., 1993; Karma and Rappel, 1996), then its evolution can be
described by a relaxation-type time-dependent GinzburgLandau equation, sometimes also called the AllenCahn equations (Hohenberg and Halperin, 1977):
@
=@t L  F=
, where F stands for the Lyapounov functional, and L is the
relaxation constant. The derivative in the right-hand side of this expression
should be understood as variational, or functional.
With this approach it becomes possible to treat on a common methodological
basis dierent phenomena in materials science systems associated with symmetrybreaking transitions (patterning) and the evolution of emerging patterns: spinodal
decomposition, ordering and anti-phase domain coarsening, solidication, martensitic transformations, grain growth, ordered precipitate morphology under stress
and precipitation in the presence of dislocations. Inclusion of defects (dislocations)
into the phase-eld description of phase transformations is one of the most exciting
recent developments of the theory. Several examples are discussed below.
4.1. Example 1: isostructural spinodal decomposition and Holts early work on
dislocation patterning
The only order parameter in this case is concentration, a conserved value, for
which the conservation law (4.1) is valid. Experimental data indicates that this
morphology is nontrivial, its most important feature being the periodic modulation
of composition inside the miscibility gap (Hilliard, 1970). For example, for the
ternary system CuNiFe the wavelength of the compositional modulation was of
the order of 100A (Hilliard, 1970).
Although the rst quantitative explanation of the periodicity was given by Hillert
(1956), below a more exible continuum model developed by Cahn (who was
apparently the rst to coin the term spinodal decomposition) will be considered
(Hilliard, 1970; Cahn, 1961, 1962). It is obvious that a standard diusion equation is insucient for proper description of spinodal decomposition, because diusion always tends to wipe out uctuations on all length scales. On the other hand,
for a concentration modulation to occur and develop, one must necessarily assume
that the eective diusion coecient is negative. In order to achieve this goal one
must introduce an additional term, which would inhibit the growth of very short
wavelength modulations. The next major step was to take into account the interface
energy. Since the concentration gradients are quite high within the layer undergoing
a spinodal decomposition, it is insucient to work in the approximation of Gibbsian
thermodynamics, because the Gibbs formalism does not contain intrinsic length
scales which are necessary to account for the nite, albeit very small, width of a
phase reaction layer. Cahns major breakthrough, which resulted in the development
of equilibrium thermodynamics of concentrationally non-uniform systems, was to
assume that instead of the classical free energy one needs to employ the concept of

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

389

the free energy functional depending on the concentration gradient squared. The
corresponding equation is given above as Eq. (2.1).
Once expression (2.1) is established, it becomes straightforward to derive the
modied diusion equation. (For the purposes of brevity, we limit ourselves only to
the case of linearized CahnHilliard equation). For a traditional ux problem the
potential for diusion is determined by the derivative df=dc; in the case when the
free energy of the system is dened, one needs to solve a variational problem of
minimization of the functional (1) subject to condition
that the average composition
of the system as a whole remains constant, i.e. c  c0 dx 0. This gives the
modied diusion equation (Hilliard, 1970):


@c
@J
@2 c
@4 c
1=NV M=NV f 00 2  2K 4
11
@t
@x
@x
@x
00

where parameters M and f are independent of composition.


Another important step was the introduction of the coherency strain energy into
consideration. There is a large body of literature related to the CahnHilliard
equation coupled to elasticity; a detailed review of these approaches can be found
e.g. in Gurtin (1996). Phase eld microstructure simulations performed for systems
described by the GinzburgLandau free energy functional with elasticity eects are
well represented by Onuki (1989) and Onuki and Nishimori (1991). Below the results
of Cahns approach are represented using mostly Eq. (4.1) for simplicity. First of all,
on the basis of Fourier analysis one comes to the conclusion that the second term in
square brackets of (11) prohibits the growth of very short wavelength modulationsa feature that makes this equation distinctly dierent from the standard diffusion problems (Hilliard, 1970). The resulting equation as a whole acts as a
nonlinear lter that eectively selects and stabilizes those length scales, which are
responsible for the formation of distinct morphology of systems undergoing spinodal decomposition.
In Fig. 8 a solution of the Cahns equation for an isotropic material is presented.
This pattern was obtained by adding 100 random sine waves of wavelength l (Cahn,
1965). The points belong to the regions where the concentration is greater than the
average. The pattern is characterized by a high degree of connectivity and the
absence of periodicity.
Later stages of decomposition are described by the nonlinearized diusion equation which was solved numerically by de Fontaine and Hilliard for a set of experimental data representing the system Al22.5% Zn. Fig. 9 shows composition
proles obtained from computer simulations (Hilliard, 1970). Broken horizontal
lines correspond to equilibrium.
It is instructive to compare Fig. 8 to the results of computer modeling using
reaction-diusion approach given in Figs. 1 and 2 (Glazov et al., 1995). These are
the manifestations of nonlinearity in materials science systems, certain fundamental
scenarios that illustrate the pathways from chaos to order through uctuations
(Nicolis and Prigogine, 1977). These generic routes should be studied, compared
and analyzed from an interdisciplinary point of view. In terms of physics, there is the

390

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

Fig. 8. Cross-section of a spinodal decomposition structure in an isotropic material. See explanations in


the text (Cahn, 1965).

need of understanding the achievements and basic tools used in the modern theory
of nonlinear oscillations and wavesqualitative theory of stability, linear stability
analysis, GinzburgLandau weakly non-linear analysis, numerical construction of
bifurcation diagrams, and the incorporation of these tools into modern metallurgy.
It is useful to remind the reader of an early work by Holt (1970) who treated the
problem of dislocation patterning (cell structures) in a fashion similar to Cahns
treatment of spinodal decomposition (Kubin, 1993). A population of parallel screw
dislocations was introduced into consideration, with uniform initial distribution,
and the driving force for the evolution of the system was the reduction in total
elastic interaction energy. Holt assumed that the correlation in the spatial distribution of dislocations disappeared at suciently large distances which was equivalent
to the assumption about a cut-o distance, Rc, for dislocation interactions.
The modication of the total energy of interaction, E, due to small periodic
spatial perturbations  could be further calculated using a Taylor expansion (Holt,
1970): E F1   F2 r2 . Assuming that dislocation velocities are proportional to the interaction forces which, in their turn, can be dened as gradients of the
energy uctuations,  r E, for the dislocation ux one can write down:
J  v Dr E, where D stands for dislocation mobility, v is the velocity of
gliding dislocations. Coupling this expression with the continuity law (which
assumes that there is no overall dislocation density change in the process of deformation), @ =@t divJ 0, one can nally get the following governing equation:

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

391

Fig. 9. Composition proles obtained from computer solution to the nonlinearized diusion equation for
an Al22.5% Zn alloy aged at 100  C for (a) 10, (b) 13.3, (c) 60, (d) 400, and (e) 416 min. The broken
horizontal lines indicate equilibrium compositions (Hilliard, 1970).


@ 
D F1 D  F2 r4 
@t

12

This is obviously the modied diusion equation (with a conserved order parameter)
derived originally by Cahn and Hilliard (1958)Eq. (2) in this paper.
This early result was criticized in the literature (Kubin, 1993) on the basis of the
following shortcomings: (1) there is no applied stress in the system; (2) no annihilation and/or multiplication of dislocations takes place; (3) the constant initial density
only evolves under the inuence of mutual interaction stresses.
Nevertheless, this was the rst attempt to treat defect instabilities from the phaseeld theory point of view, or in other words, to come up with a general formalism

392

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

which would give a possibility to treat both phase and defect microstructures from
the same point of view. Modern computational approaches to 3D-dislocation
dynamics and patterning are described by Zbib et al. (2000).
4.2. Example 2: microstructural evolution accompanied by grain growth in single
phase and two phase systems
This example is taken from an insightful work by Chen and Fan (1995). At rst
the authors dened a set of space- and time-dependent order parameters (elds) that
would provide the necessary time and length scales relevant to microstructural evolution. Two sets of parameters are necessary in order to represent the two-phase
system discussed in this example (Chen and Fan, 1995): Phase a:
1 r;
2 r; . . . ;

p rnon-conserved orientation eld variables; Phase b:


1 r;
2 r; . . . ;
p r
non-conserved orientation eld variables; concentration C(r,t)-conserved eld
variable.
Non-conserved order parameters (which are also called orientation eld variables) represent grains of a given crystallographic orientation of a given phase. The
next step in the analysis is diuse-interface description of the energetics of a twophase system. For the free energy functional one may write down:
"

F f0 Cr;
1 r;
2 r; . . .
p r;
2 r;
2 r; . . .
q r
#
p
q

2
X
X
 2



kc =2 rCr 1=2 ki r
i r 1=2 ki r
i r
d3 r
2

i1

13

i1

In this equation f0 stands for the local free energy density, kc, k i and ki are gradient energy coecients for the composition eld and orientation variables, while p
and q represent the number of orientation eld variables in the two phases. For the
energy of a planar grain boundary,  aa, between an a-grain of orientation 1 and
another a-grain of orientation 2,
1
aa



2 

2
Df kc =2dC=dx2 k 1 =2 d
1 =dx k 2 =2 d
2 =dx

14

1

where





Df f0
1 ;
2 ; C  f0
1;e ;
2;e ; Ca  C  Ca @f0 =@C
i ;Ca

15



In Eq. (15) f0
1 ;
2 ; C represent the free energy density minimized with respect
to orientation parameters at a xed composition Ca. Similar expression can be
derived for the interface boundary energy between a a-grain with orientation 1 and a
b-grain with orientation 1.
Once the diuse-interface energetics of the microstructure is accounted for, the
kinetics of microstructural evolution of a two-phase system can be represented by a

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

393

coupled set of AllenCahn (1979) and CahnHilliard equations (Chen and Wang,
1996; Chen and Fan, 1995; Chen, 2002):


d
i r; t=dt Li F=
i r; t ; i 1; 2; . . . ; p

d
i r; t=dt Li F=
i r; t ; i 1; 2; . . . ; q


16
dCr; t=dt r LC r F= r; t

In Eqs. (16) Li ; Li ; LC are kinetic coecients related to grain boundary mobilities
and diusion coecients of species 1 and 2.
Representing a relatively simple free energy density function as the sum of polynomials with respect to order parameters and numerically solving the set of Eqs.
(16), it is possible to obtain quite a realistic picture of grain growth and ripening.
For a simple one-phase system the demonstration of microstructural evolution
becomes even more convincing (Fig. 10; Chen and Fan, 1995).
4.3. Precipitation of ordered intermetallics, martensitic transformations
Below two more examples of phase microstructure modeling are given, following
Chen and Wang (1996) and Chen (2002). Fig. 11 represents microstructural evolution in the course of precipitation of an ordered intermetallic phase from a disordered matrix and its comparison to experimental observations by Fahrmann for a
NiAlMo alloy (unpublished work).

Fig. 10. Microstructural evolution and grain growth in a single-phase system. 36 non-conserved eld
variables were used: (a) t=1000; (b) t=3000; (c) t=5000; (d) t=8000 (Chen and Wang, 1996).

394

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

Fig. 11. Microstructural evolution during precipitation of an ordered intermetallic phase from a disordered matrix: the white and black particles are in antiphase relation to each other: (a) t=10; (b) t=20;
(c) t=300; (d) t=5000. (e) A comparison between the results of simulation and experimental observation
by Farhman for a NiAlMo alloy aged 2330 h at 775  C (see Chen and Wang, 1996).

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

395

4.4. Precipitation in the presence of dislocations


This is one of the most recent developments in the area of phase-eld simulations.
During the last couple of years Chen, Khachaturian and Finel independently discovered how elastic stresses from crystalline lattice defects, notably dislocations,
could be incorporated into the formalism of phase-eld computations. The nal
eort on the way to a unied formalism of treating phase and defect microstructures
simultaneously, would be the incorporation of plastic eects into consideration.
However, plasticity is an essentially irreversible dissipative process, for which the
procedures of global energy minimization are not applicable (although minimization
can be performed in increments).

5. Conclusions and recommendations for future work on microstructure of


aluminum alloys
In this paper an eort was made to consider, on the same methodological basis,
internal-variables models in plasticity and phase-eld approach in physical metallurgy. A remarkable similarity was uncovered between these two seemingly very
dierent approaches. This suggests that many of the general techniques developed in
this area of modern physical metallurgy, could be transferred to plasticity theory,
and thus produce a synergistic eect on the synthesis of mechanical and physical
metallurgy of aluminum. Recent incorporation of dislocations into phase eld
methodology indeed gives hope that the program is realistic and should be pursued.
Of course, for this to come to fruition a number of important steps need to be
undertaken:
 The length scale(s) and the gradient(s) in the structural (continuum
mechanics) part of a model must have well-dened physical basis;
 The length scale(s) and gradient(s) in microstructure representation must
have well-dened physical underpinnings, and most importantly,
 The gradients from the two must represent the same unifying mechanism(s)11
However, even now one can say that several impressive successes have been
achieved following this direction. The phase-eld technique has been successfully
applied to the following problems:
 solidication from one-component and two-component systems (Wheeler et
al., 1993; Warren and Boettinger, 1995);
 ferroelectric domain formation (Yang and Chen, 1995; Nambu and Sagala, 1994);
 ordered precipitate morphology under stress, tetragonal precipitates in a
cubic matrix Chen and Wang (1996);
 precipitation in the presence of dislocations (see above).
11

Again, the authors would like to thank one of the Referees for this valuable comment.

396

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

This demonstration of generality and eectiveness of the phase-eld approach


indicates that further exploration of such important processes as precipitation in the
presence of dislocations, grain growth, solidication etc. will become more and more
easily modeled within the framework of this approach. Modeling work should be
carried out in close conjunction to image analysis techniques, which give a possibility to determine the material length scales characterizing microstructure. Such
quantitative information, in turn, could be useful in subsequent modeling work. In
Fig. 12 we give an example of computer image enhancement and computational
Hartley transform as applied to the problem of characterization of a square twist
network on the {001} planes in pure Al after a 10% pre-strain and recovery. Work is
underway on the application of such techniques as morphological analysis and
Karhunen Loeve decompositions, to represent microstructures in terms of eigenstructures and link them to properties of aluminum alloys.

Fig. 12. Computer enhancement of a microstructure image and its computational Hartley transform as
applied to the problem of characterization of a square twist network on the {001} planes in pure Al after a
10% pre-strain and recovery for 300 h at 200  C (Glazov et al., 1998).

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

397

One of the very rst works on the subject of interaction of dislocations and precipitates was published by Cahn (1957). In that work the author demonstrated that
the nucleation of a cylindrical precipitate on a dislocation core might be a more
favorable process energetically than in the bulk. Today it is well established that the
atomistic nature of serrated ow and dynamic strain aging in aluminum and other
metallic alloys lies in the formation of the Cottrell atmospheres of point defects on
dislocations (Cottrell, 1948). Moreover, recent developments in experimental techniques like 3DAP actually give a possibility to visualize dislocations surrounded by
such atmospheres (Cadel et al., 2001)
Similar advances have been made in the theory of phase transformations and
plasticity of metallic materials, including the phase-eld approach, modern plasticity
theory with intrinsic length scales and many others. All these developments leave no
doubt that the internal-variable theory of plasticity (in its spatially-coupled form)
and phase-eld approach will bring a better understanding of the fundamental laws
governing the formation and evolution of phase and defect microstructures in alloys.

Acknowledgements
The authors would like to express their sincere gratitude to the many colleagues
who shaped their understanding of the problems discussed in the paper, including
the late Dr. Owen Richmond. A special gratitude is extended to the participants of
the Rockport, MA Symposium in the fall of 2000: Drs. Hughes, Baskes, Baudoin,
Embury, Kubin, Neuhauser, Vitek, Hahner, Zaiser, Picu, Wolf, El-Azab, and others. We are also very grateful to the four Referees of our paper and to Professor
Khan, Editor-in-Chief of the International Journal of Plasticity, whose critical comments and suggestions have helped to improve the manuscript. One of us (MG) would
like to express his sincere gratitude to Dr. Michael V. Braginsky, of the Sandia National
Laboratories (Albuquerque, NM). Section 3 of this paper would have never been written without many hours of our discussions at Penn. A sincere gratitude is also extended
to Professors Laird, Aravas, Bassani, Bonnell, and Vitek, also of the University of
Pennsylvania, and to the late Dr. Owen Richmond for helping MG understand and
appreciate the beauty and complexity of modern mechanics of materials.

References
Acharya, A., Bassani, J.L., 2000. Lattice incompatibility and a gradient theory of crystal plasticity.
J. Mech. Phys. Solids 48, 15651595.
Aifantis, E., 1986. On the dynamical origin of dislocation patterns. Mater. Sci. Eng. 81, 563574.
Allen, S.M., Cahn, J.W., 1979. A microscopic theory for antiphase boundary motion and its application
to antiphase domain coarsening. Acta Metall. 27, 1085.
Anand, L., Kothari, M., 1996. A computational procedure for rate-independent crystal plasticity.
J. Mech. Phys. Solids 44 (4), 525558.
Ananthakrishna, G., Noronha, S.J., Fressengeas, C., Kubin, L.P., 2001. Crossover in the Dynamics of
Poretevin- Le Chatelier Eect from Chaos to SOC. Mater. Sci. Eng. A 309310, 316319.
Argon, A.S. (Ed.), 1975. Constitutive Equations in Plasticity. MIT Press, Cambridge, MA.

398

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

Asaro, R.J., 1983. Micromechanics of crystals and polycrystals. Adv. Appl. Mech. 23, 1.
Asaro, R.J., Rice, J.R., 1977. Strain localization in ductile single crystals. J. Mech. Phys. Solids 25, 309
338.
Ashby, M.F., 1970. Deformation of plastically non-homogeneous materials. Philos. Mag. 21, 399.
Avlonitis, M., Zaiser, M., Aifantis, E., 2000. Some exactly solvable models for the statistical evolution of
internal variables during plastic deformation. Probabilist. Eng. Mech. 15, 131.
Balasubramanian, S., Anand, L., 1998. Polycrystalline plasticity: application to earing in cup drawing of
AA2008-T4 sheet. J. Appl. Mech. T. ASME 65 (1), 268271.
Barlat, F., Ferreira Duarte, J.M., Gracio, J.J., Lopes, A.B., Rauch, E.F., 2003. Plastic ow for
non-monotonic loading conditions of an aluminum alloy sheet sample. Int. J. Plasticity 19, 1215
1244.
Barlat, F., Glazov, M.V., Brem, J.C., Lege, D.J., 2002. A simple model for dislocation behavior, strain
and strain rate hardening evolution in deforming aluminum alloys. Int. J. Plasticity 18, 919939.
Barlat, F., Becker, R.C., Hayashida, Y., Maeda, Y., Yanagawa, M., Chung, K., Brem, J.C., Lege, D.J.,
Matsui, K., Murtha, S.J., 1997. Hattori, Yielding description for solution strengthened aluminum
alloys. Int. J. Plasticity 13, 385401.
Barlat, F., Weiland, H., Liu, J., 1996. Modeling the Inuence of Particles on the Plastic Properties of
Aluminum Alloys. Alcoa Technical Center Report No. 96-481-005.
Basinski, S.J., Basinski, Z.S., 1966. Recrystallization, Grain Growth and Textures. Am. Soc. Metals,
Metals Park, OH, p. 26.
Bassani, J.L., 2001. Incompatibility and a simple gradient theory of plasticity. J. Mech. Phys. Solids 49,
19831996.
Becker, R.C., 1995. A material size scale through crystal boundary energy. Mod. Sim. Mat. Sci. Eng. 3,
417435.
Becker, R.C., 1996. Factors Aecting Surface Roughness in Sheet Forming. Alcoa Technical Center
Internal Report No. 96-12-002.
Bulatov, V., Richmond, O., Glazov, M.V., 1999. An atomistic dislocation mechanism of pressure-dependent plastic ow in aluminum. Acta Mater. 47 (12), 3507.
Cadel, A., Fraczkiewicz, A., Blayette, D., 2001. Atomic scale observation of Cottrell atmospheres in
b-doped FeAl (B2) by 3D atom probe eld ion microscopy. Mat. Sci. Eng. A 309-310, 32.
Cahn, J.W., 1998. The kinetics of grain boundary nucleated reactions. In: Carter, W.C., Johnson, W.C.
(Eds.), The Selected Works of John W. Cahn. TMS, Warrendale, pp. 1323.
Cahn, J.W., 1957. Nucleation on dislocations. Acta Metall. 5, 169.
Cahn, J.W., 1961. On spinodal decomposition. Acta Metall. 9, 795.
Cahn, J.W., 1962. On spinodal decomposition in cubic crystals. Acta Metall. 10, 179.
Cahn, J.W., 1965. Phase separation by spinodal decomposition in isotropic systems. J. Chem. Phys. 42,
9399.
Cahn, J.W., 1994. Adapting thermodynamics to materials science problems. J. Phase Equilibria 15 (4),
373379.
Cahn, J.W., Hilliard, J.E., 1958. Free energy of a Nonuniform systems. I. Interfacial free energy. J. Chem.
Phys. 28, 258.
Cahn, J.W., Hilliard, J.E., 1959. Free energy of a Nonuniform system. III. Nucleation in a two-component incompressible uid. J. Chem. Phys. 31, 688.
Chen, L.Q., Fan, D.N., 1995. Possibility of spinodal decomposition in ZrO2Y2O3 alloysa theoretical
investigation. J. Am. Ceram. Soc. 78, 769.
Chen, L.-Q., 2002. Phase-eld models of microstructure evolution. Annual Review of Materials Research
32, 113.
Chen, L.-Q., Wang, Y., 1996. The continuum eld approach to modeling micro-structural evolution.
J. Metals 48 (12), 1318.
Coleman, B.D., 1964. Arch. Rat. Mech. Anal. 17, 1.
Coleman, B.D., Gurtin, M.E., 1967. Thermodynamics with internal state variables. J. Chem. Phys. 47, 597.
Coleman, B.D., Noll, W., 1963. Arch. Rat. Mech. Anal. 13, 167.
Cosserat, E., Cosserat, F., 1909. Theorie des Corps Deformables. Hermann, Paris.

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

399

Cottrell, A.H., 1948. In: Mott, N.F. (Ed.), Report of a Conference on Strength of Solids. The Physical
Society, London, p. 30.
De Groot, R., Mazur, P., 1962. Nonequilibrium Thermodynamics. North Holland, Amsterdam.
De Donder, Th., 1928. LAnite. Gauthier-Villars, Paris.
Defay, R., Prigogine, I.R., 1951. Tension Supercielle et Adsoprtion. Liege.
Drozdov, A.D., 2003. Dorfmann, A, A micro-mechanical model for the response of lled elastomers at
nite strains. Int. J. Plasticity 19, 1037.
Drucker, D.C., Li, M., 1992. In: Beskos, D.E., Ziegler, F. (Eds.) Acta Mechanica, Suppl. 3, Advances in
Dynamic Systems and Stability. Springer-Verlag, pp. 161171.
Dufour, L., Defay, R., 1963. Thermodynamics of Clouds. Academic Press, New York.
Eckart, C., 1940. Phys. Rev. 58, 267, 269, 919.
Eringen, A. Cemal, 1968. In: H. Liebowitz (Ed.), Fracture, an Advanced Treatise, Vol. III. Academic
Press, New York, pp. 621729.
Estrin, Yu., Kubin, L., 1988. Plastic instabilities: classication and physical mechanisms. Res. Mechanica
23, 197221.
Farhmann, M., unpublished work. Cited after Chen, L.-Q., Wang, Y., 1996. The continuum eld
approach to modeling micro-structural evolution. J. Metals 48(12), 1318.
Findlay, A., 1904. The Phase Rule and its Applications (Together with an Introduction to the Study of
Physical Chemistry by Sir William Ramsay). Longmans, Green and Co, New York (the last, 9th edition
of the book was published in 1951).
Fleck, N.A., Muller, G., Ashby, M., Hutchinson, J.W., 1994. Strain-gradient plasticity: theory and
experiment. Acta Mater. 42, 475.
Fleck, N.A., Hutchinson, J.W., 1997. Strain gradient plasticity. Advances in Applied Mechanics 33, 295.
Gearing, B.P., Moon, H.S., Anand, L., 2001. A plasticity model for interface friction: application to sheet
metal forming. Int. J. Plasticity 17 (2), 237271.
Gibbs, J.W., 1906. The Collected Works in Two Volumes. Yale University Press.
Glazov, M.V., Braginsky, M., Laird, C., Bassani, J., 1999. On increased variance of coherent solids.
Mater. Sci. Eng. A 271, 291.
Glazov, M.V., Hector, L.G., Laird, C., 1998. On the quantitative evaluation of characteristic length scales
(correlation lengths) of interior and surface structure. Acta Mater. 46 (6), 2237.
Glazov, M.V., Llanes, L., Laird, C., 1995. Self-organized dislocation structures in fatigued metals. Phys.
Stat. Sol. (A) 149, 297.
Glazov, M.V., Williams, D.R., Laird, C., 1997. Temporal dissipative structures in cyclically deformed
metallic alloys. Appl. Phys. A 64, 373381.
Glazo, M.V., Weiland, H., 2002. Morphological Analysis of Microstructures: A Tool for Understanding
Aluminum Alloy Textures. Alcoa Technical Center, Pittsburgh, PA.
Gu, C., Kim, M., Anand, L., 2001. Constitutive equations for metal powders: application to powder
forming processes. Int J. Plasticity 17, 147209.
Gurtin, M.E., 1996. Physica D 92, 178.
Gurtin, M.E., 2003. On a framework for small-deformation viscoplasticity: free energy, microforces,
strain gradients. Int. J. Plasticity 19, 4790.
Gurtin, M.E., 2002. A gradient theory of single-crystal viscoplasticity that accounts for geometrically
necessary dislocations. J. Mech.Phys. Solids 50, 5.
Gurtin, M.E., 2000. On the plasticity of single crystals: free energy, microforces, plastic-strain gradients.
J. Mech. Phys. Solids 48, 9891036.
Hahner, P., Zaiser, M., 1999. Dislocation dynamics and work hardening of fractal dislocation cell structures. Mater. Sci. Eng. A 272, 443454.
Haken, H., 1983. Advanced Synergetics. Springer-Verlag, Berlin.
Halsey, T.C., 1993. Bubbles and stripes in dipolar uids. Phys. Rev. (E) 48 (2), R673.
Hillert, M., 1956. A Theory of Nucleation for Solid Metallic Solutions. DSc Thesis, Massachusetts Institute
of Technology, Cambridge, MA.
Hilliard, J.E. 1970. Spinodal decomposition. In: Phase Transformations. American Society for Metals,
Metals Park, OH (Chapter 12).

400

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

Hohenberg, P.C., Halperin, B.I., 1977. Theory of dynamic critical phenomena. Rev. Mod. Phys. 49, 435.
Holt, D.L., 1970. Dislocation cell formation in metals. J. Appl. Phys. 41, 31973201.
Hu, S.Y., Chen, L.Q., 2001a. Solute segregation and coherent nucleation and growth near a dislocation
a phase-eld model integrating defect and phase microstructures. Acta Mater. 49, 463472.
Hu, S.Y., Chen, L.Q., 2001b. A phase-eld model for evolving microstructures with strong elastic
inhomogeneity. Acta Mater. 49, 18791890.
Jae, B., 1958. Men of Science in America. New York, pp. 307330.
Jou, D., Casa-Vazquez, J., Lebon, G., 1993. Extended Irreversible Thermodynamics. Springer, Berlin.
Karma, A., Rappel, W.-J., 1996. In: Chen, L.-Q., Fultz, B., Cahn, J.W., Manning, J.R., Morral, J.E.,
Simmons, J. (Eds.), Mathematics of Microstructure. pp. 195204.
Keller, D.J., Korb, J.P., McConnell, H.M., 1987. Theory of shape transition in two-dimensional phospholipid domains. J. Phys. Chem. 91, 6417.
Kestin, J., Rice, J.R., 1969. Paradoxes in the application of thermodynamics to strained solids. In: Proc.
Int. Symp. on Critical Review of the Foundations of Relativistic and Classical Thermodynamics,
Pittsburgh, PA.
Khachaturian, A.G., 1983. Theory of Structural Transformations in Solids. Wiley, New York.
Khaleel, M.A., 2001. Special issue on superplasticity and superplastic formingeditorial. Int. J. Plasticity
17, 275.
Khaleel, M.A., Zbib, H.M., Nyberg, E.A., 2001. Constitutive modeling of deformation and damage in
superplastic materials. Int. J. Plasticity 17, 277296.
Koiter, W.T., 1967. Couple stresses in the theory of elasticity, parts I and II. Proc. Ned. Akad. Wet. (B) 67
(1), 1744.
Kroner, E., 1962. Appl. Mech. Rev. 15, 599.
Kubin, L.P., 1993. Dislocation patterning. In: Cahn, W., Haasen, P. (Eds.), Materials Science and Technology (a Comprehensive Treatment), Vol. 6. VCH, Weinheim.
Landau, L.D., Lifshitz, E.M., 1961. The Classical Theory of Fields. Addison-Wesley.
Langer, J.S., 1992. An introduction to the kinetics of rst order phase transitions. In: Goudreche, C.
(Ed.), Solids Far From Equilibrium. Cambridge University Press, Cambridge.
Langer, J.S., 1986. Condensed matter physics, Singapore. In: Grinstein, G., Mazenko, G. (Eds.), World
Scientic, p. 165.
Langer, S.A., Goldstein, R.E., Jackson, D.P., 1992. Dynamics of labyrinthine pattern formation in magnetic uids. Phys. Rev. (A) 46, 4894.
Larche, F.C., Cahn, J.W., 1985. The interactions of composition and stress in crystalline solids. Acta
Metall. 33, 331367.
Lavenda, B.H., 1978. Thermodynamics of Irreversible Processes. Halsted, New York.
Li, M., Richmond, O., 1996. Intrinsic Instability and Nonuniformity of Plastic Deformation. Alcoa
Technical Center Report No. 96-481-019.
McConnell, H.M., Moy, V.T., 1988. Shapes of nite two-dimensional lipid domains. J. Phys. Chem. 92,
45204525.
Meixner, J., 1943. Ann. Phys. (Leipzig) 43, 244.
Mindlin, R.D., 1964. Microstructure in linear elasticity. Arch. Rat. Mech. Anl. 16, 51.
Nabarro, F.R.N., 2001. Sequences of dislocation patterns. Mater. Sci. Eng. A 309310, 1216.
Nambu, S., Sagala, D.A., 1994. Domain formation and elastic long-range interaction in ferroelectric
perovskites. Phys. Rev. (B) 50, 5838.
Needleman, A., Van der Giessen, E., 2001. Discrete dislocation and continuum descriptions of plastic
ow. Mater. Sci. Eng. A 309-310, 113.
Nicolis, G., Prigogine, I., 1977. Self-Organization in Non-equilibrium Systems (from Dissipative Structures to Order through Fluctuations). Wiley, New York.
Nicolis, G., Prigogine, I., 1989. Exploring Complexity: an Introduction. W.H. Freeman, New York.
Noll, W., 1975. The Foundations of Mechanics and Thermodynamics. Springer, Berlin.
Nye, J.F., 1953. Acta Metall. 1, 153.
Ono, S., 1947. In: Mem. Fac. Eng. Kyushu University, Vol. 10, p. 195 (cited after Cahn, J.W., Hilliard,

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

401

J.E., 1958. Free energy of a Nonuniform systems. I. Interfacial free energy. J. Chem. Phys., vol. 28,
p. 258).
Onsager, L., 1931. Phys.Rev. 37, 405. 38, 2265.
Oono, Y., Puri, S., 1987. Computationally ecient modeling of ordering of quenched phases. Phys. Rev.
Lett. 58, 836.
Onuki, A., Nishimori, H., 1991. Anomalously slow domain growth due to a modulus inhomogeneity in
phase-separating alloys. Phys. Rev. B 43 (16), 1364913652.
Onuki, A., 1989. Ginzburg-Landau approach to elastic eects in the phase-separation of solids. J. Phys.
Soc. Jpn. 58, 30653068.
Prigogine, I., 1947. Etude Thermodynamique des Phenomenes Irreversibles. Dunod, Liege.
Prigogine, I., Defay, R., 1950. Thermodynamique Chimique. Desoer, Liege (English translation by Everett, H.D., Longmans-Green, London, 1954).
Prigogine, I.R., 1957. The Molecular Theory of Solutions. Interscience, New York.
Qin, Q., Bassani, J.L., 1992. Non-associated plastic ow in single crystals. J. Mech. Phys. Solids 40, 835
862.
Rice, J., 1973. Inelastic constitutive relations for solids: an internal-variable theory and its application to
metal plasticity. J. Mech. Phys. Solids 19, 433455.
Rice, J., 1970. On the structure of stress-strain relations for time-dependent plastic deformation in metals.
Transactions of the ASME 9, 728737.
Rice, J., 1975. Continuum mechanics and thermodynamics of plasticity in relation to microscale deformation mechanisms. In: Argon, A.S. (Ed.), Constitutive Equations of Plasticity. MIT Press, pp. 2375.
Rice, J.R., 1977. The localization of plastic deformation. In: Koiter, W.T. (Ed.), Theoretical and Applied
Mechanics. North-Holland, Amsterdam, pp. 207220.
Richmond, O., 1986. Microstructure-based modeling of deformation processes. J. Metals (4) 1618.
Richmond, O., 1987. Materials-based deformation process models. In: Proc. TMS Symp., Warrendale,
PA.
Rodney, D., and Finel, A., 2001. Phase-eld methods and dislocations. In: Aindow, M., Asta, M.,
Glazov, M.V. (Eds.), Inuences of Interface and Dislocation Behavior on Microstructure Evolution,
MRS Symposium Proc., Vol. 652, pp.Y491Y496.
Romanov, A.E., Aifantis, E., 1993. On the kinetic and diusional nature of linear defects. Scripta Met.
29, 707712.
Roytburd, A., Slutsker, J., 2002. Coherent phase equilibria in a bending lm. Acta Mater. 50, 1809.
Rudnicki, J.W., Rice, J.R., 1973. Conditions for the localization of deformation in pressure-sensitive
dilatant materials. J. Mech. Phys. Solids 23, 371.
Rukeyser, M., 1942. Josiah Willard Gibbs, N.Y. Sample, V.M., Lalli, L.A., Richmond, O., 1990. A Two
Internal Variable Constitutive Model for p0815 Alloy. Alcoa Technical Center Report No. 12-90-FRZ029-16.
Sample, V.M., Lalli, L.A., Richmond, D., 1990. A two-internal variable constitutive model fo P0815
alloy, Alcoa Technical Center Report #1290FRZ02916.
Sedov, L.I., Berditchevski, V.L., 1967. Proceedings of the IUTAM Symposium on the Generalized Cosserat Continuum and the Continuum Theory of Dislocations with Applications. Springer-Verlag,
Freudenstadt and Stuttgart, Germany. 1968.
Shuey, R., Suni, J., Doherty, R., Chakrabarti, D., 1996. Prediction of time for solution heat treat. Mater.
Sci. Forum 217, 735.
Slutsker, J., Roytburd, A.M., 2002. Control of intrinsic instability of superelastic deformation. Int.
J. Plasticity 18, 1561.
Smelser, R.E., Anand, L., Hector, L.G., 2001. Owen Richmond 19282001in memoriam. J. Appl.
Mech.- T. ASME 68 (6), 825826.
Goudreche, F. (Ed.), 1993. Solids Far from Equilibrium. North-Holland, Amsterdam.
Spitzig, W.A., Smelser, R.E., Richmond, O., 1988. The evolution of damage and fracture in iron compacts with various initial porositities. Acta Metall. 36, 12011211.
Staley, J.T. 1992. Metallurgical aspects aecting strength of heat-treatable alloy products used in the
aerospace industry. In: Proceedings IIIrd Int. Conf. on Aluminum Alloys, pp. 107143.

402

M.V. Glazo et al. / International Journal of Plasticity 20 (2004) 363402

Swaminarayan, S., LeSar, R., 1999. A Monte Carlo method for simulating dislocation microstructures in
three dimensions. Comp. Mater. Sci. 21, 339359.
Taylor, G.I., 1938a. Plastic strain in metals. J. Inst. Metals 62, 307325.
Taylor, G.I., 1938b. Analysis of plastic strain in a cubic crystal. In: Lessels, J.M. (Ed.), Stephen
Timoshenko: Sixtieth Anniversary Volume. MacMillan, New York.
Toupin, R.A., 1962. Arch. Rat. Mech. Anal. 11, 385.
Truesdell, C., 1985. The Elements of Continuum Mechanics. Springer-Verlag.
Truesdell, C., 1969. Rational Thermodynamics. McGraw Hill, New York.
Vaithyanathan, V., Chen, L.Q., 2000. Coarsening kinetics of d0 -Al3Li precipitates: phase-eld simulation
in 2D and 3D. Scripta Mater. 42, 967973.
Van der Vaals, 1893. Thermodynamic theory of capillarity under the hypothesis of a continuous variation
of density (translated into English in 1979 by Rowlinson, J.S.). J. Stat. Physics 20, 197.
Van der Vaals, J.D., Kohnstamm, Ph., 1927. Lehrbuch der Thermostatic. Barth, J.A., Leipzig.
Vitek, V., 1995. Atomistic Modeling in Materials Science. Lecture Notes, Dept. of Materials Science,
Univ. of Pennsylvania.
Walgraef, D., Aifantis, E., 1985. On the formation and stability of dislocation patternsI, II and III. Int.
J. Engng. Sci. 24, 13511372.
Wang, Y., Chen, L.-Q., Khachaturyan, A.G., 1991. Acta Metall. Mater. 41, 279.
Wang, Y., Jin, Y.N., Cuitino, A.M., Khachaturyan, A.G., 2001a. Phase eld microelasticity theory and
modeling of multiple dislocation dynamics. Appl. Phys. Letts. 78, 23242326.
Wang, Y., Jin, Y.N., Cuitino, A.M., Khachaturyan, A.G., 2001b. In: Appl. Phys. Letts, vol. 78, pp. 2324
2326
Wang, Y., Srolovitz, D.J., Rickman, J.M., Lesar, R., 2000. Dislocation motion in the presence of diusing
solutes: a computer simulation study. Acta Mater. 48 (9), 21632175.
Warren, J.A., Boettinger, W.J., 1995. Prediction of dendritic growth and microsegregation patterns in a
binary alloy using the phase eld method. Acta Metall. Mater. 43, 689.
Weiland, H., 1994. Microtexture determination and its application to materials science. J. Metals 46 (9),
3741.
Wen, Y.H., Wang, Y., Chen, L.Q., 2002. Coarsening dynamics of self-accommodating coherent patterns.
Acta Mater. 50, 21131.
Wheeler, P.L., 1951. Josiah Willard Gibbs: The History of a Great Mind, CT.
Wheeler, A.A., Boettinger, W.J., McFadden, G.B., 1993. Phase-eld model of solute trapping during
solidication. Phys. Rev. (E) 47, 18931909.
Xue, Z., Huang, Y., Li, M., 2002. Particle size eect in metallic materials: a study by the theory of
mechanism-based strain gradient plasticity. Acta Mater. 50, 149160.
Yang, W., Chen, L.-Q., 1995. J. Amer. Ceram. Soc. 78, 2554.
Zaiser, M., Glazov, M., Lalli, L., Richmond, O., 1999. On the relations between strain and strain-rate
softening phenomena in some metallic materials: a computational study. Comp. Mater. Sci. 15, 35.
Zbib, H.M., de la Rubia, T.D., 2002. A multiscale model of plasticity. Int. J. Plasticity 18, 11331163.
Zbib, H.M., de la Rubia, T.D., Rhee, M., Hirth, J.P., 2000. 3D-dislocation dynamics: stressstrain
behavior and hardening mechanisms in fcc and bcc metals. J. Nucl. Mater. 276, 154.
Ziegenbein, A., Hahner, P., Neuhauser, H., 2000. Correlation of temporal instabilities and spatial localization during Portvin-Le Chatelier deformation of Cu-10 at.% Al and Cu-15 at.% Al. Comp. Mater.
Sci. 19, 2734.

Das könnte Ihnen auch gefallen