See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/242221986

Development and application of the dislocation model for analysis of the

microstructure evolution and deformation behavior of metals subjected to
severe plastic deformation

Article · December 2007

CITATIONS READS

20 461

2 authors, including:

I.V. Alexandrov
Ufa State Aviation Technical University
207 PUBLICATIONS 13,758 CITATIONS

SEE PROFILE

All content following this page was uploaded by I.V. Alexandrov on 23 June 2016.

The user has requested enhancement of the downloaded file.

Development and application
Rev.Adv.Mater.Sci. 16(2007) 51-72
of the dislocation model for analysis of the microstructure... 51

DEVELOPMENT AND APPLICATION OF THE

DISLOCATION MODEL FOR ANALYSIS OF THE
MICROSTRUCTURE EVOLUTION AND DEFORMATION
BEHAVIOR OF METALS SUBJECTED TO SEVERE
PLASTIC DEFORMATION

I. V. Alexandrov and R. G. Chembarisova

Department of Physics, Ufa State Aviation Technical University, 12 K. Marx St., Ufa, 450000, Russia
Received: April 17, 2007

Abstract. The upgraded Estrin-Tóth model is presented to describe the deformation behavior of
the materials subjected to severe plastic deformation. The model has been applied toward study
of tension of the as-received Cu samples and the samples after the 1st, 4th, and 8th equal channel
angular pressing passes. The dislocation density evolution in the two phases, namely in the bound-
aries of fragments/cell blocks and interiors, is studied. The evolution of the concentration of vacan-
cies that formed during deformation, annihilation of dislocations during their non-conservative
motion and generation of dislocations during multiple cross slip are considered. As a result, pecu-
liarities of the microstructure evolution, causes of achieving high values of the vacancy concentra-
tions and dislocation densities are revealed. Features of the deformation behavior of the consid-
ered samples are explained.

1. INTRODUCTION strain hardening (stages IV and V of strain harden-

ing) [7].
It is well known that considerable microstructure
High pressure torsion (HPT) and equal-chan-
refinement and, as a consequence, change of
nel angular pressing (ECAP) are main techniques
mechanical properties of metals takes place as a
for receiving bulk UFG and nanostructured states
result of large plastic strains [1-3]. Low-angle
in different metallic materials [4]. The conducted
boundaries that separate cells and blocks of cells
experiments showed that the yield stress of the
are typical for such structures; they are also char-
UFG Cu processed by SPD is several times higher
acterized by enhanced strength and low ductility.
than that of the conventional coarse-grained Cu
The investigations of recent years have shown
before SPD. The strain hardening is insignificant
that the severe plastic deformation (SPD) leads to
at the plastic flow stage, as compared to the an-
formation of ultrafine-grained (UFG) structures
nealed coarse-grained state [8]. It is stated that not
(grain size d less than 1 µm) and even
only the grain size but also the defect structure of
nanostructured (d <100 nm) states with mainly high-
grain boundaries influence the strength properties
angle grain boundaries [4]. SPD implies large plas-
of Cu and other metals [8]. SPD Cu may have a
tic deformations (e>1) realized under high applied
relatively low [9] or high ductility [5] during tensile
pressures of several GPa [4]. These structures may
testing. This depends on the predominance of low-
be characterized not only by high strength but also
angle or high-angle grain boundaries in the struc-
high ductility (SPD paradox) [5], low-temperature
tures formed at definite ECAP modes.
and/or high-rate superplasticity [6], absence of
Corresponding author: I. V. Alexandrov, e-mail : [email protected]

© 2007 Advanced Study Center Co. Ltd.

52 I. V. Alexandrov and R. G. Chembarisova

As follows from the experiments [10], the aver- the peculiarities of the acting deformation mecha-
age dislocation density in grains increases with the nisms. Various models are used to analyze the
strain up to ~ 1015 m-2 after SPD of Cu. Meanwhile, strain hardening during plastic deformation. The
the dislocation density is much higher in grain simplest of them are the well-known Kocks-
boundaries. At the same time, the average dislo- Mecking-Estrin models, which involve just one in-
cation density can change non-monotonously, with ner variable - the total dislocation density [18,19].
the strain degree rising [11]. At that, the bound- These models give a very good explanation to stage
aries preserve their non-equilibrium state [10]. The III of the strain hardening. S. Kok, A.J. Beaudoin,
intragranular slip makes a large contribution into and D.A. Tortorelly [20] proposed a two-parameter
the total strain. Similar results are observed in case model of hardening to describe stage IV of the strain
of SPD of other metals and alloys [12,13]. hardening. The model is a further development of
Unusual mechanical properties are conditioned the Kocks-Mecking theory [19]. The latter consists
by the peculiarities of structure and processes, in consideration of the density of geometrically
which accompany the deformation of UFG and necessary dislocations (GND).
nanostructured materials. Though intragranular slip Most of the models describing stage IV of the
is the main deformation mechanism for these ma- strain hardening assume that a sample may be
terials, the dislocation accumulation observed in considered as a two-phase composite material
the grain body is not significant [4]; dislocations are which consists of cell walls with a high dislocation
presumably captured by grain boundaries. In- density and interiors (cell bodies) with a low dislo-
creased diffusion is typical for grain boundaries in cation density. A three-parameter model proposed
the UFG Cu [14]. It contributes to the recovery pro- by Y. Estrin, L.S. Tóth, and others [21] relates to
cess during deformation of the UFG Cu at room such models. The model enables describing stages
temperature [4]. The activation energy of the grain III, IV, and V of the strain hardening within one single
boundary diffusion of ~ 79.2 kJ/mole in the UFG formalism. A three-parameter approximation was
Cu is much lower than the value 107 kJ/mole typi- proposed by E. Nes and K. Marthinsen [22,23]
cal for diffusion along the stationary grain bound- during the simulation of the deformation behavior
aries in the coarse-grained Cu [4]. Absence of the of metals. Their composite model describes the
dislocation accumulation may be also caused by microstructure evolution. It is supposed that dur-
activation of the recovery processes, which take ing small strains (stage II of the strain hardening)
place partially by means of the dislocation cross dislocations form a cell structure, which is charac-
slip [15]. terized by the cell size δ and the wall thickness h,
A high value of the yield stress, at which the the dislocation density in the walls ρb and that in
condition the onset of plastic deformation is ful- the interiors ρi. During large strains (stage IV of the
filled, may be explained by an increase in the dis- strain hardening) misorientations between
location density during ECAP. Since any significant subboundaries are accounted for.
dislocation accumulation is not observed in the The models by M. Zehetbauer [24] and M.
grain body, the origin of a high value of the yield Zehetbauer and P. Les [25], which are based on
stress is explained by the model based on the dis- Mughrabi’s concept about the two-phase structure
location curvature mechanism [16]. Shortening of of dislocation cells [26], refer to the models which
dislocation spacing in the grain boundaries leads describe stages IV and V of the strain hardening.
to the stress growth. A high value of the stress is At that, the cell interior size L1 and the cell wall thick-
necessary for a Frank-Read source to start work- ness L2 are used as structural parameters. Accord-
ing in the boundary. ing to the model by M. Zehetbauer [24], a cell body
Equilibrium between the strain hardening and has screw dislocations which are located in cell
recovery in the grain boundaries takes place at the interiors. The density of these dislocations is equal
steady deformation stage [10]. The flow stress is to ρ1. Edge dislocations with the density ρ2 are con-
controlled by nucleation of new dislocations at this centrated in the cell walls. The model accounts for
stage. The results obtained in [6,17] show that the the role of vacancies which create conditions for
mechanical behavior of the UFG metals and al- climb and annihilation of edge dislocations. The
loys is determined by the dislocation substructure correspondence between the modeling and experi-
that formed during SPD. mental data was shown in [27].
The study of the mechanisms of unusual defor- The presented models form a hierarchy of mod-
mation behavior of SPD UFG and especially els depending on the number of the parameters
nanomaterials is of great scientific interest due to used. The first three stages of the strain hardening
Development and application of the dislocation model for analysis of the microstructure... 53

are controlled by one parameter. The number of process of emission of a second partial disloca-
the parameters should be increased up to three to tion, miscalculation of the system energy, and the
describe stages IV and V of the strain hardening. artificially introduced grain size. In [35] a correct
The models which consider the occurrence of three-dimensional model of emission of half-loops
stages IV and V of the strain hardening as a result of split slip dislocations by grain boundaries in NS
of the structural changes and the evolution of crys- Al is proposed. The model accounts for the real
tallographic texture are of great interest in case of geometry of dislocation slip and is based on pre-
SPD. cise calculation of the energy of elastically inter-
The analysis shows that the Estrin-Tóth model acting dislocation loops. An anomalous width of
is one of the most developed strain hardening stacking faults in NS Al is conditioned by the activ-
models [21,28]. Up to the present, the strain hard- ity of external stress. At the same time, similarly to
ening curves which correspond to different defor- the approach suggested in [34], the anomalous
mation modes have been analyzed with the help width of stacking faults is also a grain size function
of this model [28]. A visco-plastic self-consistent (with the external stress absent). As follows from
(VPSC) model [29] of polycrystal ductility is com- the model presented in [35], generally the width of
bined with the model to consider the crystallo- stacking faults reduces when the grain size de-
graphic texture evolution. The dislocation densi- creases. This result opposed the conclusions made
ties in the cell walls ρw and interiors ρc are used as in [34], which are a consequence of the artificially
modeling parameters. The relation between the introduced grain size.
average dislocation density and the cell size is de- The conducted analysis demonstrates certain
fined. The law of variation of the cell wall volume success in the description and analysis of the de-
fraction against the strain degree is made. Stages formation behavior mechanisms of UFG and
IV and V of the strain hardening in fcc polycrystals nanostructured metallic materials. At the same
are adequately modeled. The evolution character time, the mentioned models possess a number of
of the average dislocation density is predicted. The weak points. For example, a cell structure is con-
obtained results agree well with the experimental sidered without concretizing the boundary struc-
data. The model was also used to analyze the de- ture and cell interiors in the models described in
formation behavior of the materials processed by [21,24,25,28]. Unique properties of materials un-
severe plastic deformation [30]. der SPD are a result of high volume fraction of grain
In recent years special attention has been paid boundaries, which play an important role during
to revelation of the causes of formation of unusual deformation of such materials. Specific properties
mechanical properties in SPD nanostructured (NS) of the structure and processes occurring during
materials with a grain size less than 100 nm. Some deformation keep in the background. This compli-
deformation mechanisms such as emission of par- cates studying the role of these or those deforma-
tial dislocations from grain boundaries and forma- tion mechanisms. The misorientation evolution
tion of stacking fault and twins were predicted on between neighboring cells is not considered. In the
the basis of the molecular-dynamic modeling and present paper an attempt has been made to de-
observed experimentally [31,32]. velop a dislocation model, that could help to carry
A mechanical model of the deformation behav- out a thorough analysis of the microstructure and
ior of NS metals based on the assumption about deformation behavior evolution of pure metals (on
the emission of partial dislocations from grain the example of Cu) subjected to SPD.
boundaries was developed by Asaro et al. [33]. It
was shown that the stress required for the emis- 2. BACKGROUND FOR THE
sion of partial dislocations is much lower than the UPGRADED DISLOCATION MODEL
stress required for the emission of total disloca- DEVELOPMENT
tions when the grain size becomes smaller than
some critical value. Stress is a function of stacking The development of the strain hardening models
fault energy and shear modulus. In [34] a model should be based on modern ideas about the char-
estimating the critical values of the stress and grain acter of microstructure evolution and mechanisms
size, at which the emission of partial dislocations of SPD realization.
from grain boundaries and the start of the twinning According to the experimental data, the collec-
process in nanostructured materials are possible, tive behavior of dislocations complies with general
is proposed. The model’s drawbacks are supposed rules when the strain degree increases [1]. Cha-
to be immobility of a first partial dislocation in the otic distribution of dislocations in a polycrystal at
54 I. V. Alexandrov and R. G. Chembarisova

Fig. 1. Scheme of the deformation microstructure à) during small and mean strains by von Mises (eνm=0.06-
0.80); b) during large strains (eνm≥1) [36].

the initial stages of deformation in a polycrystal

becomes energetically unfavourable. As a result, f =
d − d −w
3
! 3
".
3

(1)
a cell structure and then a fragmented one form. d
The latter one is characterized by high It is assumed that the cell walls have also a cell
misorientation angles of fragments. Meanwhile, a structure. Then the Frank-Read sources, which are
cell structure sort “freezes” during plastic deforma- located on the wall surface, emit dislocations with
tion. Fragmentation occurs on the background of the length d-w into the cell interiors. This leads to
the “frozen” cell structure. Only refinement of frag- increase in the dislocation density in a cell body
ments takes place during further deformation and with the rate:
their relative half-turns increase (Fig. 1).

!ρ! "
Processing of UFG and NS materials by the α* ρw
+
= γ! w ,
r
SPD technique is realized by incessant fragmen- c (2)
tation of grains [3,36]; a specimen becomes highly 3 b
fragmented. The fragment sizes remain practically where α* is the coefficient of the effectiveness of
unchanged in the process of the following plastic the Frank-Read sources. The dislocation slip rate
deformation [1]; perfection of interfragmentary in the cell walls νw is expressed in terms of the
boundaries takes place and misorientations be- resolved shear rate γ! wr according to Orowan’s for-
tween fragments increase. Mainly equi-axed mi- mula: γ! wr = ρw bνw , b - the value of the Burgers vec-
crostructure forms with high-angle non-equilibrium tor.
grain boundaries as a result of SPD. The dislocations leaving the cell interiors are
integrated with the cell wall structure. Accounting
3. MAIN PROVISIONS OF THE ESTRIN for this process, the dislocation density in the cell
– TÓTH MODEL body decreases with the rate
We shall first consider main provisions of the model
r
[28] worked out by Estrin and Tóth to understand 6 γ!
− =−
(ρ! c )w c
the details of the upgrading process. The conven- β* , (3)
bd (1 − f )
1/ 3
tional Estrin-Tóth model considers a composite
structure of a cubic shape with a number of dislo-
where β* - the coefficient of the effectiveness of
cations which form a cell structure as its structural
sink of dislocations in the cell walls. When deriving
element (Fig. 2). The neighboring cubic cells with
the edge length d are separated by the walls with the formula (3), Orowan’s ratio between the dislo-
the thickness w, the volume fraction of which is cation slip rate in the cell body νc and resolved strain
rate γ! c was used: γ! c = ρc bνc.
r r
Development and application of the dislocation model for analysis of the microstructure... 55

Fig. 2. An element of the fragmented structure.

With account of mutual annihilation of disloca- The increase in the dislocation density results in
tions of opposite signs by a cross slip, the disloca- reduction of the length of dislocation segments in
tion density in the cell body reduces according to the boundaries and increase of shearing stresses
the law: necessary for the activity of Frank-Read sources.
The growth of the applied stress allows keeping

! " = −k #%$ γγ!! &('

r −1/ n
−
the activity of the existing sources and also leads
ρ! c γ! c ρc .
c r
cs 0 (4) to activation of new ones. As a result the disloca-
0
tion density will grow with the rate
The parameter k0 characterizes the intensity of
annihilation processes at T=0K, and n character-
! "
ρ! w
+ ( FR )
=
! "
3β * γ! c 1 − f
r
ρw
. (7)
izes their sensitivity to strain rate during cross slip. fb
With the stacking fault energy rising, the value of n
grows and is inversely proportional to T, as the sen- Due to the annihilation of dislocations at the ex-
sitivity of flow stress to the strain rate m [21]. pense of cross slip in the cell walls, the dislocation
As a result of the mentioned processes, the dis- density decreases with the rate

!ρ! " = −k #%$ γγ!! &('

location density inside the cells changes accord- −1/ n
r
ing to −
γ! w ρw .
w r
w cs 0 (8)
0

1 ρw 6 γ! c
r

ρ! c = α * γ! w − β * −
! "
r
The total contribution of all the considered pro-
bd 1 − f
1/ 3
3 b cesses can be stated by the equation:
# γ! &
k % (
r
−1/ n
(5)
! " ! "
$ γ! ' γ! c ρc .
c r
6β * γ! c 1 − f 3β * γ! c 1 − f ρw
r 2/3 r
0
0 ρ! w = + −
bdf fb
The dislocations leaving the cell interiors are
# γ! &
k % (
r −1/ n
(9)
$ γ! '
integrated into the cell wall structure causing growth w
γ! w ρw .
r
0
of the dislocation density in the walls with the rate 0

! "
ρ! w
+
=
6β * γ! c 1 − f
r
! " 2/3

. (6)
It is considered that the cell size is proportional to
bdf the dislocation density
56 I. V. Alexandrov and R. G. Chembarisova

The model parameters α*, β*, K, k0 are defined

K
d= , (10)
by approximating the modeling flow curve to the
ρtotal experimental points. The 3D model is used in com-
bination with the VPSC model of polycrystals. This
where K – the proportionality constant, ρtotal - the helps tracing the dependence of hardening curves
total dislocation density, which is calculated as on the deformation modes and texture evolution
! "
ρ total = fρ w + 1 − f ρ c . (11)
[29].
At the same time, the Estrin-Tóth model con-
Eq. (10) cannot be interpreted as a homothety with siders a cell structure of cubic shape without speci-
the coefficient K. In other words, defining the wall fying the structure of cell interiors or boundaries.
thickness w as an image of 1 / ρtotal The authors do not account for the peculiarities of
structure and processes which take place during
K′ deformation of UFG and NS materials mentioned
w= (12)
ρtotal above. In particular, the evolution of misorientations
between neighboring cells and grains is not taken
we get w/d=const and, therefore, the volume frac- into account. The nature of other sources and sinks
tion of cell walls f is also constant. Such a state- of defects in the cell boundaries and interiors, ex-
ment does not agree with the experimental data. cept Frank-Read sources, and annihilation of screw
As a result of data integration, the resolved strain dislocations during double cross slip is not speci-
dependence of the volume fraction of cell walls f fied. The boundary structure and its influence on
was derived in [20] as: the formation of special mechanical properties of

! " )
UFG and NS materials constitute a major problem
f = f∞ + f0 − f∞ exp − γ / ~
γ ,
r r
* (13) in physics of such materials [37] (Fig. 1a), but they
are not considered in this model.
where f0, f∞ - its initial and limited values, ~
γ - the
r
The model does not enable calculating
parameter characterizing the rate of decrease of f misorientations between cells, as it does not con-
depending on the resolved strain γr. sider the density of excess dislocations in the cell
The resolved stresses τ w and τ c in the cell walls
r r
walls. As a result, it may be successfully used for
and interiors are calculated with account of the cor- simulation of large plastic strains for which low-
responding dislocation densities ρw and ρc angle misorientations are typical (Fig. 1b), but the

# γ! &
ρ % (
r 1/ m
model requires upgrading in the case of SPD ma-
terials, since they have a UFG structure with mainly
τ w = αGb
$ γ! '
r w
w
, (14)
0
high-angle misorientations between grains. The aim
of this paper is to upgrade the Estrin-Tóth disloca-
tion model for adequate accounting of peculiari-
# γ! &
ρ % (
r 1/ m
ties of structure evolution and deformation behav-
τ c = αGb
$ γ! '
r c
c
, (15) ior of SPD metals with fcc lattice.
0

where G is the shear modulus, b – the Burgers 4. MAIN PROVISIONS OF THE

vector, γ! 0 - the pre-exponential factor, 1/m – the UPGRADED MODEL
value opposite to the sensitivity of flow stress to
strain rate, α - the constant that reflects interac- According the experimental data [1,4], we shall
tions between dislocations. consider that the structure of an SPD material con-
The deformation behavior of the composite is sists of cell blocks/fragments which are separated
defined by the resolved stress τr connected with by the boundaries containing excess dislocations
the resolved stresses τ w and τ c according to the
r r ‘fluffed’ by non-excess sessile dislocations [3] (Fig.
rule 2). We shall designate such a structure as a frag-
mented one with the fragment size d. Dislocations
r r
! "
τ = fτ w + 1 − f τ c .
r
(16) in the fragment interior form a cell structure with
the total density of immobilized dislocations in the
To fulfill the conditions of strain compatibility cell walls and their interiors equaling to ρc. Account-
along the phase division boundary (the cell interi- ing of the total dislocation density ρw and the den-
ors and walls), it is considered that γ! w = γ! c = γ! .
r r r
sity of non-excess sessile dislocations ρf in the frag-
ment boundaries will enable calculating the den-
Development and application of the dislocation model for analysis of the microstructure... 57

sity of excess dislocations in the boundaries and The increase in the dislocation density of fragment
misorientations between the fragments. The vol- boundaries leads to the increase of misorientations
ume fraction of fragment boundaries is calculated between fragments [4].
according to formula (1). Mobile dislocations can annihilate with non-ex-
Annihilation of dislocations during their non-con- cess sessile dislocations, the density of which is
servative movement conditioned by vacancy diffu- indicated as ρf. The annihilation of edge disloca-
sion and annihilation of edge dislocations are con- tions is discussed in [42], where it is shown that
sidered to explain the absence of hardening on the the edge dislocations of opposite signs annihilate
stage of severe plastic strain. Material flow is con- at distances smaller than the annihilation length
sidered to be supported by dislocation generation which equals to 1.6 nm. As a result, the dislocation
during multiple cross slip [38]. density decreases in the boundaries with the rate:
−
ρ! w = − Rf ρ f γ! / b.
r
(19)
4.1. Evolution of the dislocation
density in the fragment where Rf is the annihilation length.
boundaries The dislocation density in the fragment bound-
aries reduces as a result of the annihilation of non-
Basing on the experimental data, we shall give
excess sessile edge dislocations, which is condi-
mathematical description of physical processes
tioned by their non-conservative motion due to va-
which govern the evolution of the dislocation den-
cancy capture [43]. Climb of dislocations may oc-
sity in the fragment boundaries. As in case of the
cur only on condition that the diffusional mass trans-
conventional Estrin-Tóth model, an increase in the
fer between a dislocation and a surrounding me-
dislocation density in the fragment boundaries is
dium is possible. The diffusional flux on a disloca-
considered to occur at the expense of the disloca-
tion occurs due to oversaturation of point defects
tions coming from the interiors (6) and the activity
in the considered volume. Such a flux is also
of the Frank-Read sources (7). It is supposed that
caused by the force acting on a dislocation in the
emission of the stored deformation energy is car-
direction normal to its slip plane.
ried out by annihilation of screw dislocations as a
According to [44], the dislocation density de-
result of double cross slip, typical for the beginning
creases as a result of the vacancy diffusion through
of stage III of the strain hardening [15]
the tube surrounding the dislocation core accord-
ing to
δ α γ! ρw
r
−
ρ! w = − , (17)
−
b dρ w
= − φ ′ρ f ,
3
(20)
where δα - the coefficient of annihilation of screw dt
dislocations. In [39,40] the following expression was
where φ’ - the function of vacancy concentration.
derived
The hypothesis is that the increment coefficient of
#% &(
γ! 0 the dislocation density ρf is 1/τ. Then the decrease
!"
k bT
δα
= χ α = χα 0 e
A
ln
$ ',
γ!
r

(18) rate of the density of non-excess sessile disloca-

b tions during their non-conservative motion is
where χα(0) is the annihilation coefficient at T=0K, −
dρ w ρf
kb - the Boltzman constant, A – the constant de- =− , (21)
pending on the stacking fault energy. dt τ
The stacking fault energy W of Cu is 0.050 where τ - the annihilation time.
J/m2, see [41]. According to the dependence of Parallel edge dislocations in different but paral-
A/(G*b3) on W/(G*b) [40], the exponent is defined lel slip planes achieve one another only by way of
as n=A/kbT. n=37 for Cu at homologous tempera- climbing, which is possible only during the diffu-
ture T/Tm=0.2 which corresponds to room tempera- sional mass transfer. Let axis 0x is directed along
ture. the Burgers vector (Fig. 3). A pair of dislocations
The defect structure and specific deformation climbing towards each other with the rate υc under
mechanisms of nanomaterials influence the dy- the activity of the corresponding component σxx of
namics of the dislocation density. This fact sug- the stress deviator tensor in a sample which is af-
gests that the fragment boundaries include non- fected by uniaxial tensile stress σ from starting
excess sessile and excess edge dislocations with positions, corresponding to the stable configura-
the Burgers vector perpendicular to the boundary.
58 I. V. Alexandrov and R. G. Chembarisova

Fig. 3. Stable configuration of dislocations of opposite signs.

tion (Fig. 3) [41], annihilates in the time τ which where

needs to be estimated. The rate of the relative dis-
location motion is 2νc here. B′ = ! " !
2 / π 1 − ν Dc q c GΩ / kT , "
= D C exp) − H / kT *.
o
If the dislocations are separated by a = r cos45 = m (25)
r 2 / 2 , then da/dτ = 2νc, dτ = da/2νc, where r is Dc co ν c

the running value of the distance between the dis-

Cν - the concentration of vacancies which form dur-
locations. Integrating the last equation, we receive
ing deformation, H cm - the enthalpy of the activa-

+
r0 tion of the vacancy migration along the dislocation
1 dr
τ= , (22) nuclei which is equal to 0.2 eV according to [24], Ω
2 2 0
νc - the atomic volume in Cu, Ω = 1.1⋅10-29 m3, qc - the
square of the cross section of the dislocation core,
where r0 is the initial distance between the disloca-
qc = b2, G – the shear modulus, Dc - the diffusion
tions.
coefficient along the dislocation core, the multiplier
The flux of vacancies emerges in a sample, if
Dc0 is accepted in accordance with [24], Dc0=10-4
the energy of vacancy formation at the expense of
m2/s, ν - the Poisson ratio, ν =0.343 [27].
the applied external tensile stress along x axis is
Using Eq. (20), the function φ = φ‘/ γ! against
r

different in different points of a sample. This mass-

dρw/dt is constructed by going over from t to γr , φ is
transfer results in the plastic deformation, the rate
proportional to the vacancy concentration Cv here.
of which depends on the rate of dislocation climb-
At the first approximation, we consider that the va-
ing υc. σxx - the corresponding stress deviator ten-
cancy concentration (the function φ) depends lin-
sor component, arising as a result of interaction of
early on the deformation γr with the proportionality
the dislocation pair, is the reason of the climb of
constant C, that’s why we can write
dislocations with the rate υc=B’/2 2 r r0 . Then, af-
2

ter integrating Eq. (22), we get

#% dρ &( = φ#% dρ &( + dρ
φ
$ dγ ' $ dγ ' dγ Cdγ .
w w w r

4
r r r (26)
r0 t +1 t t
τ= . (23)
2B ' The density of non-excess sessile dislocations
Taking into account ρf = 1/ r0 , we receive the
2 ρf, which is part of the dislocation density in the
following fragment boundaries ρw, changes according to the
law (20). Comparing the corresponding annihila-
dρf tion constituent in Eq. (24) with expression (20) ,
= −2B ′ρf ,
3
(24) we receive φ‘ = 2BCν, where B = B’/Cν. In the non-
dt
dimensional view, after introducing non-dimen-
Development and application of the dislocation model for analysis of the microstructure... 59
~ ~r ~ ~
sional magnitudes B , γ! , and φ ′ , Bρ 20 w t 0 → B , where Pf ′ is the fraction of the “forest” dislocations
~ r ~ r ~′ ~~! r
φρ 0 w t 0 → φ , γ! t 0 → γ! , φ = φγ , where t0 = 1/ γ! 0 ,
2
among all the dislocations, which came into the
~ ~
the last equation is φ ′ = 2BCν . Depending on de- fragment boundaries.
formation, the vacancy concentration is
~ ~
C ν = φ ′ / 2B . 4.2. Evolution of dislocation density
While including noncomplanar slip systems in the fragment interiors
during deformation, dislocation generation on the
An increase in the dislocation density in the frag-
“forest” dislocations crossing the given slip plane
ment interiors is considered to take place on ac-
becomes predominant [38]:
count of influence of the Frank-Read sources (2).
δ f ρ f γ!
1/ 2 r The reduction of the dislocation density is condi-
ρ! w = , (27) tioned by annihilation during cross slip (4) and sink
b of dislocations from interiors into the fragment
where δf is the coefficient of dislocation generation boundaries (3). Describing the evolution of the dis-
on the “forest” dislocations, the density of which is location density in the fragment interiors, one
ρw. It is stated in [38] that δf ≈ 10-2 at high disloca- should note that the dislocation generation should
tion densities. The dislocation densities in the pri- be additionally taken into account after multiple
mary and secondary slip systems are comparable cross slip [38].
at the beginning of stage II of the strain hardening
δ f ρc γ!
1/ 2
r
and increase with the deformation. That results in
ρ! c = . (31)
the annihilation of screw components of disloca- b
tion loops at low and moderate temperatures. Due
to this, stage III emerges on the flow curve [45]. Taking into account the contribution of all the
The length of dislocations moving from the in- above-mentioned processes, we can state that the
teriors to the fragment boundaries is β*ρcνcdt(d- dislocation density inside the fragment changes
w)2 [28]. The fraction of immobilized dislocations according to
is designated as Pf. Then it can be expected that
the dislocation density will grow as 1 ρw δ f ρc γ!
1/ 2 r

ρ! c = α * γ! + −
r

+
ρ! f =
6Pf β * γ! 1 − f
r
! " 2/ 3

. (28) 6 γ!
r
3 b b
(32)
β* − χ 0 γ! ρc .
! "
r
bdf
bd 1 − f
1/ 3

Thus, summing up Eqs. (7), (17), (19), (24),

(27), and (28), the evolution of the total dislocation
density can be described with the help of the ex-
pression:
4.3. Formalism of the upgraded model
It is supposed that within the upgraded model the

ρ! w =
6 Pf β * γ! 1 − f
r
! " 2/3

+
r
! "
3β * γ! 1 − f ρw
+
resolved shear stress is calculated according to
formulas (14), (15), (16). Introducing non-dimen-
bdf fb sional variables:
δ f ρf γ! δ α ρw γ! Rf ρ f γ! (29)
1/ 2 r r r

− − − 2B ′ρ f .
3

b b b ρw ~ ρ ~ , ρc → ρ
t
→~
ρw , →t, f →ρ ~,
ρ 0w ρ0 w ρ0 c
f c
t0
The density of non-excess sessile dislocations
~
changes with time as follows b ρ0 w → b,

# 6P β * γ! !1 − f " + 3β * γ! !1 − f " ~ τ τ
r r
~ G
ρw d ρ0 w → d , r → G, wr → ~
τw , cr → ~
τc ,
r 2/3 r r r

ρ! = % + τ0 τ0 τ0
$ bdf
f
f
(33)
fb
γ!
r

δ ρ γ! &
~r R ~
= γ! t 0 → γ! , f → Rf ,
r

R ρ γ! (30)
(
1/ 2 r r
!γ 0 b
P′ −
b '
− 2BC ρ ,
f f f f 3
f ν f
b
where τ 0 - the yield stress.
r
60 I. V. Alexandrov and R. G. Chembarisova

Bringing together Eqs. (14), (15), (28), (29), and

b
(32), the obtained system of equations can be writ- θ= = bdρexc . (37)
ten in the dimensionless form: h

,.d~ρ = 6P β*~γ! !1−f " + 3β*~γ! !1−f " ρ~ +

w f
r 2/3 r
w
4.5. Account of the evolution of
crystallographic texture
.. dt~~ ~! bdf ~~ ~
fb
To consider the influence of the crystallographic

.. δ ρb~ γ − χ ~γ! ~ρ − R~ ~ρ ~γ! − ~φρ~

1/ 2 r
f f r r 3 texture, the values of the reduced rates γ! rs in each
a w f f f
slip system of a grain (fragment) were obtained
..dρ~ =# 6P β*~γ! !1−f " + 3β*~γ! !1−f " ρ~ +
r 2/3 r with the help of the VPSC model. The shear rate in

.. dt %$~ bdf
f f w
a grain at rather high rate sensitivities was taken
~~ ~
fb equaling

.- δ ρ~~ γ! &( P ′ − R~ ~ρ ~γ! − ~φρ~

1/ 2 r

.. b ' ~
k

∑ γ!
f f r 3
γ! ≈
r r
f f f f
s
, (38)
s =1

..ddt~ρ~ = α * ρ~~ γ! + δ b~~ρ γ! −

~ r r
c w f c where k is the number of active slip systems. The
(34)
correctness of the last equation is shown in [28].
.. 6β * ~γ! 3b
r
~
A polycrystal was presented as the values of

.. bd !1 − f " −χ ~ ρ γ! weighted orientations. Orientations were connected

r
~~ 1/ 3 a c
with grains and their weights were defined by vol-
r
..~τ = αGb~~ ρ~ )~γ! * , ~τ = αGb~~ ~ρ )~γ! * r 1/ m r r 1/ m ume fractions. Crystallographic texture of a sample
w

/ w c c was described with the help of these values.

Every grain is a viscoplastic ellipsoid placed in
the effective viscoplastic medium averaged in all
The total dislocation density, the fragment size the grains. Deformation is carried out by slipping
(in non-dimensional variables), and the evolution in all slip systems activated by the resolved shear
of the volume fraction of fragment boundaries are stresses. It is considered that the dependence of
calculated here using expressions (11), (10), (13). the resolved strain rate tensor γ! r on the stress ten-
sor σc is correct for any separate grain
4.4. Calculation of misorientations
between neighboring fragments
γ! ij = γ! 0 ∑
g ij g kl g s : σ
s s
#% &( m −1

$τ ' σ kl = M ijkl σ kl ,
r c c
(39)
τ
s s
One of the upgrading points for the Estrin-Tóth dis- s

location model was the consideration of the

misorientation angle between the neighboring frag- where s is a number of the slip system, gs -
ments. The excess dislocation density in the bound- Schmid’s tensor, τs - the threshold shear stress,
aries can be defined as the difference between the and σ – the applied external stress.
total dislocation density ρw and the density of non- The equation of interaction between the inclu-
excess sessile dislocations ρf in the fragment sion and effective medium, on which the self-con-
boundaries: sistent model is based, was used in the form

ρ exc = ρ w − δ f . (35) )γ! r

− γ!
r
* = n !1 − E " : E:M )σ − σ *,
eff −1 sec c
(40)

Basing on the calculated excess dislocation where γ! r , σ are the overall rate of the resolved
density, the distance between the neighboring dis- strain and overall stress, respectively, E – Eshelby’s
locations in the cell walls h is tensor, which is a homogeneous function of the
overall compliance and the ratio of ellipsoid axes,
1 sec
M - overall secant viscoplastic compliance de-
h= . (36)
dρexc pending on the stress in a sample. The parameter
neff makes it possible to receive the Taylor approxi-
Misorientations between the neighboring cells were mation (neff = 0), tangent (neff = m), and secant (neff =
calculated according to the approach proposed in 1) approximations, Sach’s approximation (neff = ∞).
[46]: The case at which 1< neff < m considers the differ-
Development and application of the dislocation model for analysis of the microstructure... 61

where Vi is the volume fraction of grains with the

given orientation, N - the number of grains, L – the
number of slip systems. The generalized Taylor
factor which corresponds to tension of a polycrys-
tal material was calculated as

γ!
r

M tention =
*
, (42)
ε!
where ε! is the macroscopic strain rate. At that, the
following formula was taken into account

+ #$ ∑ )V γ! *&'dt.
t N

γ =
r r
i i (43)
0 i =1

5. PROCEDURE OF SIMULATION AND

EXPERIMENTS
ECAP is a cyclic process which consists in mul-
tiple pushings of a sample through the two chan-
nels with equal sections intersecting at an angle
o o
90 ≤ φ ≤ 157.5 [47,48] along different routes. When
modeling a regular ECAP pass, the data obtained
during the previous deformation stage were used
as input parameters. The route Bc, considered in
this paper, is performed by turning the morphologic
and crystallographic textures clockwise about axis
3 of the global coordinate system 1, 2, 3 (Fig. 4a).
At that, axes 1’, 2’, 3’ of the coordinate system which
Fig. 4. 1, 2, 3 – axes of the global coordinate sys- is connected with the sample coincide with the axes
tem (à), 1', 2', 3' - axes of the coordinate system, 2, 1, and 3 of the global coordinate system respec-
connected with the sample (b). tively (Fig. 4b). (When coming out of the die chan-
nel, axes 1’, 2’, 3’ of the connected coordinate sys-
tem coincide with axes 1, 2, 3 of the global coordi-
nate system). The second rotation is carried out
clockwise about axis 2 of the global coordinate
system (axis 1’ of the connected system) after each
pass.
At each pass of ECAP along the route Bc, the
ence of interaction between each separate grain specimen is subjected to simple shear along the
and effective medium [47]. In the present paper m plane that is oriented at an angle ϕ = φ/2. The shear
is taken equal to 185, that is 5 times larger than planes intersect at every following pass. The shear
the exponent n in Eq. (17) (i.e. n >> 1, but m > n) at the 3rd pass is reverse in respect to the shear at
and tangential approximation corresponds to the the 1st pass. The shear at the 4th pass is reverse to
Estrin-Tóth model. the shear at the 2nd pass. The pattern of the defor-
The Taylor factor M was defined as the ratio of mation path is repeated in every 4th pass.
the average sum of the resolved strain rates γ! rs in When modeling tension of a sample, the data
all the slip systems to the given strain rate ε! [21]: obtained at the end of simulation of the succeed-
ing ECAP pass were used as input parameters.
N

∑ $ ∑ γ!
# L
&V The Taylor factor was calculated at every defor-
' ,
r
s i mation step with the help of the obtained volume
M= i =1 s =1 (41) fractions Vi and resolved rates γ! rs in each slip sys-
ε! tem according to the Eq. (41).
62 I. V. Alexandrov and R. G. Chembarisova

Fig. 5. Experimental (!) and modeling curves (-) of the flow stress dependence of the tension degree of
Cu (evolution of the Taylor factor is taken into account): initial state (a), after 1(b), 4(c), and 8(d) passes of
ECA-pressing along the route Bc.

The angle of channel’s intersection during ECAP applied stress to the cross section of a sample cal-
was 90°. The punch displacement velocity was 50 culated with the help of the condition of deforma-
mm/min, the test was carried out at room tempera- tion uniformity and stability of the sample volume.
ture. The samples for ECAP had the size of 60×8×8 The true strain was counted as ε = ln(F0/FT), where
mm3 . FT is the current value of the cross section area
Mechanical tensile tests of the pure Cu sample and F0 is the initial value of the cross section area.
annealed at 550 °C during one hour (in the initial The samples were subjected to grinding and pol-
state) and Cu after the 1st, 4th, and 8th passes of ishing.
ECAP along the route Bc were carried out on the The plastic flow of samples was heterogeneous
universal dynamometer INSTRON 1185 at room along the sample length; this led to localization of
temperature with the initial strain rate 5.5.10-4 s-1. plastic strain. Elongation to failure of the initial
The diameter of the working part of a sample for sample and the samples after the 1st, 4th, and 8th
mechanical testing was 3 mm, and the length was ECAP passes was 45%, 15%, 10% and 11%; the
15 mm. True stress was defined as the ratio of the yield stress was 227 MPa, 376 MPa, 436 MPa, and
Development and application of the dislocation model for analysis of the microstructure... 63

Table 1. Initial values of the dislocation densities in the samples taken in different structural states.

Parameters As-received 1 pass 4 passes 8 passes

route Bc route Bc route Bc

ρw(γr = 0.0), m-2 5.0*1014 4.9*1015 1.4*1016 1.8*1016

ρc(γr = 0.0), m-2 2.5*1014 5.7*1014 6.5*1014 9.0*1014

Table 2. The values of the parameters used for modeling.

Parameters Values Parameters Values

f∞ 0.060 G, GPa 42.1

~
γ
r
3.2 m 185
γ! 0 , s-1 1 n 37
δf 0.013 b, nm 0.256

460 MPa, and the failure stress was 415 MPa, 600 the values of the volume fraction f and resolved
MPa, 680 Mpa, and 740 MPa, respectively. stress τr obtained via Eq. (16) using the values of
An algorithm and a software (Fortran 77) were the dislocation densities ρw and ρc taken from [28].
developed with to perform calculations in accor-
dance the above-mentioned equations. Calcula- 6. MODELING RESULTS AND
tions were carried out for the wide range of vari- DISCUSSION
able parameters, which were chosen regarding for
the available experimental data and physical sig- Fig. 5 shows the experimental data obtained dur-
nificance. ing tensile testing of Cu samples in the as-received
A polycrystalline sample was assumed to con- state and after the 1st, 4th, and 8th passes of ECAP
sist of 830 weighted orientations. It was supposed along the route Bc.
that up to 12 octahedral slip systems {111}<110> Stage III of the strain hardening is clearly ob-
typical of fcc lattice could operate. The threshold served on the flow curve of the initial state. The
shear stress was the same for all the slip systems. character of the flow curve after the 1st, 4th, and 8th
The parameters used during modeling of tension ECAP passes is more complicated. Stages III and
of Cu in the initial state and after the 1st, 4th, and 8th IV of the strain hardening are distinguishable. The
ECAP passes along the route Bc are presented in modeling curves obtained for the given states cor-
Tables 1 and 2. The values of the dislocation den- respond to the experimental data. The above-men-
sities ρw and ρc were chosen according to the data tioned peculiarities of the deformation behavior of
from [49-51]. It has been taken into account that Cu are also observed on the modeling of harden-
the modeling results are insensitive to the initial ing rate - resolved strain curves (Fig. 6). Stage III
values of variables in the considered interval of of the strain hardening for Cu subjected to ECAP
deformation as stated in [30]. is characterized by a much higher hardening rate
~ as compared to stage III of hardening of the as-
The parameters α*, β*, K’, C, Pf, R f , χa(0) were
obtained approximating the modeling curves to the received Cu. This is due to much higher degree of
experimental points by the least-square method. the accumulated strain.
The simulation was carried out with the true strain The experimental points were presented as the
of 0.51, at which the section in the neck decreased dependence “resolved stress - resolved strain” to
almost 1.6 times. Accounting of texture, the calcu- be compared with the modeling curves (Fig. 5).
lation of the resolved stresses was carried out by Basing on this aim, the integral Taylor factor M*
the technique described in [28]. The constant of was counted for a polycrystalline. The Taylor fac-
the dislocation interaction α was estimated from tor values corresponded to the tension of the
64 I. V. Alexandrov and R. G. Chembarisova

Fig. 6. Experimental (!) and modeling (-) hardening rate dependence of the resolved strain (evolution of
the Taylor factor is taken into account): (a) initial state, after 1st (b), 4th (c), 8th (d) ECAP passes along Bc.

Table 3. The parameters, obtained during tensile strain of Cu in different structured states.
Parameters As-received 1st pass 4th pass 8th pass
route Bc route Bc route Bc

α* 0.059 0.092 0.151 0.154

β* 0.010 0.019 0.022 0.022
χa(0) 4.2 10.7 15.7 15.3
K 10.00 8.00 6.0 6.0
C 4.0⋅10-6 1.1⋅10-2 70 300
~
Rf 6.8 3.2 3.0 3.0
Pf 0.11 0.77 0.92 0.92
Pf ′ 0.35 0.17 0.050 0.001
M* 2.47 2.48 2.38 2.39
f0 0.250 0.145 0.066 0.060
α 0.36 0.46 0.42 0.40
(0≤ γ ≤3.57⋅10-2 ) (0 ≤ γ ≤ 3.58⋅10-2)
0.11 0.38
-2
(3.57⋅10 (3.57⋅10-2
<γ<0.11) <γ<0.11)
γ ≥ 0.11 γ ≥ 0.11
Development and application of the dislocation model for analysis of the microstructure... 65

Fig. 7. Dislocation densities (1 – in the fragment boundaries, 2 – total, 3 – in the fragment interiors, 4 – of
the non-excess sessile dislocations in the fragment boundaries) in Cu, depending on the resolved strain
(evolution of the Taylor factor is considered) during tensile test from the initial state (a), after 1st (b), 4th (c),
8th (d) passes of ECA pressing along the route 3.57⋅10-2.

samples in the above-mentioned states, their val- experimental values of the dislocation density from
ues are presented in Table 3. The values of the [24].
true stresses on the experimental curves were di- As seen from Table 3, the tension after the 1st,
vided by the integral Taylor factor. The values of 4 , and 8th ECAP passes is accompanied by the
th

the true strain were multiplied by the integral Tay- increase of activity of the sources in the fragment
lor factor. boundaries, which is characterized by the param-
The evolutions of the dislocation densities, eter α*, in comparison with the sample deformed
namely, the total dislocation density ρtotal, the dislo- in the as-received state. It may be explained by
cation density in the fragment boundaries ρw, the the fact that the piling-up of dislocations has al-
dislocation density in the interiors ρc, the density of ready occurred during ECAP and that results in
non-excess sessile dislocations ρj in the fragment the growth of internal stresses necessary for the
boundaries, were predicted with the help of the dis- subsequent deformation; therefore, the number of
location model of a composite sample. Fig. 7 shows active sources increases. For example, simulations
the dislocation densities - averaged sum of crys- show that the total dislocation density is almost 3.8
tallographic shears graphs. Fig. 8 presents the times higher during tension after the 8th ECAP pass
66 I. V. Alexandrov and R. G. Chembarisova

Fig. 8. Evolution of dislocation density N during high

Fig. 9. Evolution of misorientations during tension
pressure torsion (+), rolling (!), upsetting (!) with
of Cu depending on the resolved shear stress: 1 –
the rate γ! =10-2 .s-1 [24].
as-received, 2, 3, 4 - after 1st, 4th, and 8th ECAP
passes along the route Bc, respectively.

than during tension of the as-received Cu for the

strain degree γr = 0,8. Activation of Frank-Read
sources in the boundaries results in growth of the
coefficient α*. As a result, the flux of dislocations,
which come into the interiors from the boundaries,
also grows. The increase of the coefficient β* char-
acterizes the activation of dislocation absorption
by the fragment boundaries.
The dislocation density grows in the boundaries
with the strain degree rising in all considered cases.
The growth of the dislocation density in the frag-
ment boundaries leads to some qualitative changes Fig. 10. Evolution of the fragment sizes during ten-
in their structure [50]; we also observed this phe- sion of Cu: 1 – as-received, 2, 3, 4 - after 1st, 4th,
nomenon after the conducted simulation. At that, and 8th ECAP passes along the route Bc.
the increase of misorientations is 6 and 9 times
higher (with respect to the as-received state dur-
ing the deformation after the 4th and 8th passing
along the route Bc) than during the deformation of
the as-received Cu. Misorientations during the de-
formation of Cu after the 1st pass along the route
Bc are almost the same as during deformation of dislocations is accompanied by the growth of the
the as-received Cu (Fig. 9). excess dislocation density and, therefore, by the
The density of non-excess sessile dislocations density of misorientations between fragments (Fig.
in the boundaries decreases with the strain degree 8).
rising during the tension of samples after 4th and The non-excess sessile dislocation density
8th passes. This testifies to the activation of diffu- grows with the strain rising in case of Cu at the
sional processes in the boundaries, annihilation of initial state and after the 1st ECAP pass. This is
dislocations of opposite signs during capture, and connected with the slightly-expressed diffusional
decrease of the fraction of non-excess sessile dis- processes and low rates of dislocation climbing,
locations in the boundaries, which is reflected in which are conditioned by smaller values of the cor-
the reduction of the parameter Pf ′ values. As the responding stress deviator tensor component and,
total dislocation density in the boundaries grows, therefore, by the slight annihilation during capture.
the decrease of the density of non-excess sessile In the given case, the value of the parameter Pf ′
Development and application of the dislocation model for analysis of the microstructure... 67

Fig. 11. Concentration of vacancies in Cu as-received (à), after 1st (b), 4th (c), and 8th (d) ECAP passes
along the route Bc.

that controls the fraction of non-excess sessile dis- 10). Fragment sizes reduce more intensively be-
locations is higher. The fraction of excess disloca- fore the 4th pass, while misorientations between the
tions is lower and, therefore, misorientations be- fragments increase (Fig. 9) at insignificant change
tween the fragments are smaller (Fig. 8). The frag- of their sizes after the 4th pass. This agrees with
ment boundaries are less perfect and their struc- the experimental data [30].
ture is less ordered. This is also evidenced by the Severe plastic deformation delivers many lat-
parameter Pf which characterizes permeability of tice defects and vacancies [24,27,51]. Fig. 11 pre-
boundaries. sents the vacancy concentration - strain degree
The fragment sizes reduce during tension of graphs for Cu in the initial state and after the 1st,
samples in the as-received state and after the 1st , 4th, and 8th ECAP passes along the route Bc. The
4th, and 8th ECAP passes. During tension after the concentration of deformation vacancies reveals a
1st pass the fragment sizes are about 200 nm, af- dramatic growth when the number of ECAP passes
ter the 4th pass they are about 100 nm, and after increases. The equi-axed vacancy concentration
the 8th pass they become less than 100 nm (Fig. is about 10-20 at room temperature [24], this value
68 I. V. Alexandrov and R. G. Chembarisova

Fig. 12. Evolution of the dislocation densities, described by the corresponding members of Eq. (29), in the
fragment boundaries as a result of non-conservative motion of dislocations ρw, n-c(a) ,ρw, n-c(b); of annihila-
tion as a result of double cross slip ρw, cs(c); of capture of edge dislocations ρw, capt(d) in Cu during tension
of samples, taken as-received (1), after 1st (2), 4th (3) and 8th (4) ECAP passes along the route Bc.

in Cu increases by 11 orders after the 8th pass. At that, the vacancy concentration increases (Fig.
The concentration of deformation vacancies in Cu 11). This can be explained by obstructivity for va-
after the 8th pass is 4 orders higher than that in the cancy diffusion with the increase in the number of
as-received Cu. These data coincide with the re- passes.
sults received in [43]. Annihilation during the non-conservative motion
Figs. 12a and 12b shows the evolution of the of dislocations contributes significantly to the re-
dislocation density (in the non-dimensional vari- covery process being responsible for the growth of
ables) in the fragment boundaries at the expense misorientations between the fragments.
of the diffusion-conditioned annihilation processes. The density of dislocations which annihilate as
In case of Cu after the 8th ECAP pass, annihilation a result of double cross slip in the fragment bound-
during deformation turns is restricted and the den- aries increases in all considered structure states
sity of annihilating dislocations decreases (Fig. of Cu (Fig. 12c).
12b). As a result, the dislocation density in the The growth in density of the dislocations which
boundaries and the total dislocation density grow. annihilated during capture of edge dislocations of
Development and application of the dislocation model for analysis of the microstructure... 69

Fig. 13. Evolution of the dislocation densities, de-

scribed by the corresponding members of Eq. (32),
in the fragment interiors during the annihilation of
dislocations as a result of double cross slip ρc, cs(a);
coming from the internal areas into the fragment
boundaries ρc, β(b) during tension of Cu samples,
taken (1) as-received, after 1st (2), 4th (3), and 8th
(4) ECAP passes.

Fig. 14. Evolution of the dislocation densities, de-

th
opposite signs is restricted after the 8 ECAP pass. scribed by the corresponding member of Eq. (29),
That results in the increase of the dislocation den- in the fragment boundaries as a result of immobili-
sity in the boundaries and the total dislocation den- zation of dislocations coming from interiors ρw, p(a);
sity (Fig. 12d). of activation of Frank-Read sources ρw, F-R(b); of
The density of the dislocations which annihilated dislocation generation during multiple slip ρw, ms(c)
as a result of double cross slip in the fragment in- during tension of Cu samples, taken as-received
teriors grows with the degree of the accumulated (1), after 1st (2), 4th (3), and 8th (4) ECAP passes.
strain rising (Fig. 13a). The density of dislocations
leaving the interiors also grows with the degree of
the accumulated strain rising (Fig. 13b).
When the strain degree grows, the density of
dislocations in the boundaries increases as a re-
70 I. V. Alexandrov and R. G. Chembarisova

sult of immobilization of the dislocations which

come from interiors (Fig. 14a) and activation of
Frank-Read sources (Fig. 14b). The increase in
the dislocation density during multiple slip is re-
stricted in Cu after the 8th ECAP pass. This can be
explained by a decrease in the density of non-ex-
cess sessile dislocations (Fig. 14c). The growth of
the dislocation density in the interiors is conditioned
by the influence of the Frank-Read sources (Fig.
15a) and dislocation generation during multiple slip
(Fig. 15b).
The results of the conducted modeling exhibit
a significant influence of ECA pressing on the struc-
ture and processes which take place during ten-
sion of Cu taken in different structural states. The
coefficient α* characterizes the effectiveness of the
Frank-Read sources, which are superficial for the
fragment interiors. This coefficient increases 2.6
times after the 8th ECAP pass in comparison with
the as-received Cu. The dislocation sink coefficient
β* increases 2.2 times. The boundaries absorb dis-
locations which come from the interiors. The anni-
hilation coefficient χa(0) in the samples subjected
to SPD is 10.7 after the 1st pass, 15.7, and 15.3
after 4th and 8th passes, respectively; this corre-
sponds to the values typical for the conditions of
hydrostatic pressures [40]. The annihilation coeffi-
Fig. 15. Evolution of the dislocation densities, de-
cient is χa=2÷10 under the usual conditions of fcc
scribed by the corresponding member of Eq. (32),
metals deformation at moderate temperatures,
in the fragment interiors as a result of activation of
while χa = 4.2 for the as-received Cu grade used in
Frank-Read sources ρc, F-R(a); of dislocation gen-
this paper.
eration during multiple slip ρc, ms(b) during tension
The value of the parameter K which character-
of Cu samples, taken as-received (1), after 1st (2),
izes the fragment sizes does not change after the
4th (3), and 8th (4) ECAP passes.
4th ECAP pass. Therefore, the fragment sizes
(Fig.10) after the 4th ECAP pass do not change
significantly. Structure refinement generally takes
place before the 4th pass of ECAP.
The parameter C that controls intensity of the
growth of vacancy concentration increases with the
degree of the accumulated strain rising. The anni-
hilation length decreases before the 4th pass, after
that it remains constant and equals to 3. for the as-received Cu. This testifies to the fact that
The boundaries are practically impermeable after the 4th ECAP pass insignificant structure re-
after the 4th pass. The parameter Pf that character- finement takes place, but the process of boundary
izes the fraction of the dislocations immobilized in improvement occurs, i.e. the boundaries become
the boundaries increases with the degree of the more ordered. Misorientations between the frag-
accumulated strain rising; after the 4th pass it re- ments increase. The fraction of dislocations, which
mains unchanged. The value of the parameter is become non-excess sessile ones, is 50 times lower
about 8.4 times higher after the 8th ECAP pass than after the 8th ECAP pass than after the 4th pass. The
that in the initial state. density of non-excess sessile dislocations de-
The parameter Pf ′ that reflects the fraction of creases, and, therefore, the annihilation of dislo-
non-excess sessile dislocations in the boundaries cations during capture in the boundaries and dis-
decreases 350 times in comparison with its value location generation on the “forest” dislocations re-
duce.
Development and application of the dislocation model for analysis of the microstructure... 71

The vacancy concentration continues increas- 9. Considerable increase in the yield stress of the
ing. Diffusion processes slow down after the 8th samples after the 1st, 4th, and 8th ECAP passes
ECAP pass, so that annihilation of dislocations in is explained by an increase in the dislocation
the fragment boundaries decreases during their density during ECAP.
non-conservative motion.
Thus, we may conclude that the structure of REFERENCES
fragment boundaries after 8 ECAP passes is of
different quality. They are highly-misoriented. The [1] Ya.M. Vitorskiy, E.Yu. Zubets, S.N. Kaverina,
fraction of non-excess sessile dislocations is neg- V.A. Manilov, V.I. Trefilov and V.I. Firstov //
ligibly small. Misorientations between fragments are Physics of Metals and Metallurgy 33 (1972)
high. 831.
[2] V.M. Bykov, V.A. Likhachev, Yu.A. Nikonov,
L.L. Serbina and L.I. Shibalova // Physics of
7. CONCLUSIONS Metals and Metallurgy 45 (1978) 163.
1. The presented model consistently describes the [3] V.V. Rybin, Large plastic deformations and
observed experimental data. failure of metals (Metallurgia, Moscow, 1986).
2. The presented model allows describing quanti- [4] R.Z. Valiev and I.V. Alexandrov,
tatively the dependency of the dislocation den- Nanostructured materials, processed by
sity on the number of passes along the given severe plastic deformation (Logos, Moscow,
ECAP route. 2000).
3. The study of the evolution of the dislocation den- [5] R.Z. Valiev, I.V. Alexandrov, Y.T. Zhu and T.C.
sity as a result of influence of different mecha- Lowe // J. Mater. Res. 17 (2002) 5.
nisms has shown that the role of annihilation [6] G.T. Gray III, T.C. Lowe, C.M. Cady, R.Z.
processes, conditioned by capture and non- Valiev and I.V. Alexandrov // Nanostr. Mater. 9
conservative motion of edge dislocations, is (1997) 477.
reduced after the 8th ECAP pass; the mecha- [7] R.Z. Valiev, R.K. Islamgaliev and I.V.
nism of dislocation generation is suppressed on Alexandrov // Progr. Mater. Sci. 45 (2000)
the “forest” of dislocations. The mentioned 103.
changes reflect qualitative reconstructions in the [8] V.Yu. Gertsman, R.Z. Valiev, N.A. Akhmadeev
structure of fragment boundaries. The bound- and O. Mishin // Mater. Sci. Forum. 233
aries become practically impermeable and the (1996) 749.
fraction of non-excess sessile dislocations is [9] R.Z. Valiev and R.K. Islamgaliev // Physics of
negligibly small. Metals and Metallurgy 85 (1998) 161.
4. Structure refinement takes place mainly up to [10] R.Z. Valiev, E.V. Kozlov, Yu.F. Ivanov, J. Lian,
the 4th ECAP pass. Upon that, misorientations A.A. Nazarov and B. Baudelet // Acta Metall.
between fragments remain low. Mater. 42 (1994) 2467.
5. Misorientations between the fragments start to [11] T. Ungár, In: Proceedings of the NATO
grow at minor changes of their sizes after the Advanced Research Workshop on
4th ECAP pass. Investigations and Applications of Severe
6. The dramatic growth of vacancy concentration Plastic Deformation, ed. by T.C. Lowe and
takes place with the number of passes increas- R.Z. Valiev (Kluwer Academic Publishers,
ing. At that, the dislocation density in the frag- Dordrecht-Boston-London, 2000), p. 93.
ment boundaries and the total dislocation den- [12] A.A. Popov, R.Z. Valiev, I.Yu. Pyshmintsev,
sity grow. S.L. Demakov and A.G. Illarionov // Physics
7. The model predicts the evolution of vacancy of Metals and Metallurgy 83 (1997) 127.
concentration in the fragment boundaries and [13] R.K. Islamgaliev, D.A. Salimonenko, L.O.
allows estimating their role in strain hardening Shestakova and R.Z. Valiev // Izvestiya
and character of the forming fragment bound- Vuzov. Tsvetnaya Metallurgia 6 (1997) 52.
aries. [14] Yu.R. Kolobov, R.Z. Valiev and G.P.
8. Insignificant strain hardening of the materials Grabovetskaya, Grain boundary diffusion
subjected to ECAP is explained by the fact that and properties of nanostructured materials
growth of the sources’ activity is compensated (Nauka, Novosibirsk, 2001).
by enhancement of annihilation processes. The [15] G.A. Malygin // Solid State Physics. 29
contribution of each process can be estimated. (1987) 2067.
 72 I. V. Alexandrov and R. G. Chembarisova

[16] J. Lian, B. Baudelet and A.A. Nazarov // [36] A.N. Vergazov, V.A. Likhachev and V.V.
Mater. Sci. Engin. A172 (1993) 23. Rybin // Physics of Metals and Metallurgy. 42
[17] V.Y. Gertsman, R. Birringer, R.Z. Valiev and (1976) 1241.
H. Gleiter // Scripta Metall. Mater. 30 (1994) [37] D.A. Hughes and N. Hansen // Acta Mater.
229. 45 (1997) 3871.
[18] Y. Estrin and H. Mecking // Acta Metall. 32 [38] L.E. Popov, V.C. Kobytev and T.A.
(1984) 57. Kovalevskaya, Plastic deformation of the
[19] H. Mecking and U.F. Kocks // Acta Metall. 29 alloys (Metallurgia, Moscow, 1984).
(1981) 1865. [39] G.A. Malygin // Physics of Solid State. 29
[20] S. Kok, A.J. Beaudoin and D.A. Tortorelly // (1987) 2067.
Acta Mater. 50 (2002) 1653. [40] G.A. Malygin // Physics of Solid State. 32
[21] Y. Estrin, L.S. Tóth, A. Molinari and (1992) 3200.
Y. Bréchet // Acta Mater. 46 (1998) 5509. [41] P.I. Polukhin, S.S. Gorelik and V.K.
[22] E. Nes and K. Marthinsen // Mater. Sci. Vorontsov, Principal physics of plastic
Engin. A322 (2002) 176. deformation (Metallurgia, Moscow, 1982).
[23] E. Nes // Progr. Mater. Sci. 41 (1998) 129. [42] U. Essman and H. Mughrabi // Phil. Mag.
[24] M. Zehetbauer // Acta Mater. 41 (1993) 589. A40 (1979) 731.
[25] P. Les and M. Zehetbauer // Key Eng. Mater. [43] M.J. Zehetbauer, H.P. Stüwe, A. Vorhauer,
97-98 (1994) 335. E. Schafler and J. Kohout // Adv. Engin.
[26] H. Mughrabi // Acta Metal. 31 (1983) 1367. Mater. 5 (2003) 330.
[27] M. Zehetbauer and V. Seumer // Acta Metall. [44] F.B. Prinz and A.S. Argon // Acta Metall. 32
Mater. 41 (1993) 577. (1984) 1021.
[28] L.S. Tóth, A. Molinari and Y. Estrin // J. Eng. [45] G.A. Malygin // Physics of Solid State. 34
Mater. Technol. 124 (2002) 71. (1992) 2882.
[29] U.F. Kocks, C.N. Tomé and H.-R. Wenk. [46] W. Pantleon // Acta Mater. 46 (1998) 451.
Texture and Anisotropy (2nd edition, [47] C.N. Tomé // Mod. Sim. Mats. Sci. Eng. 7
Cambridge Univ. Press, Cambridge, 2000). (1999) 723.
[30] S.C. Baik, R.J. Hellmig, Y. Estrin and H.S. [48] V.M. Segal // Mat. Sci. Engin. A197 (1995)
Kim // Z. Metallkd. 94 (2003) 754. 157.
[31] V. Yamakov, D. Wolf, S.R. Phillpot, A.K. [49] T. Ungár, I. Alexandrov and P. Hanák, In:
Mukherjee and H. Gleiter // Nat. Mater. 1 Proceedings of the NATO Advanced
(2002) 45. Research Workshop on Investigations and
[32] X.Z. Liao, Y.H. Zhao, S.G. Srinivasan, Y.T. Applications of Severe Plastic Deformation,
Zhu, R.Z. Valiev and D.V. Gunderov // Appl. ed. by T.C. Lowe and R.Z. Valiev (Kluwer
Phys. Lett. 84 (2004) 592. Academic Publishers, Dordrecht-Boston-
[33] R.J. Asaro, P. Krysl and B. Kad // Phil. Mag. London, 2000), p. 133.
Lett. 83 (2003) 733. [50] M. Zehetbauer, E. Schafler, T. Ungár,
[34] Y.T. Zhu, X.Z. Liao, S.G. Srinivasan and E.J. S. Kopacz and S. Bernstorf // ASME J. Eng.
Lavernia // Journal of Appl. Phys. 98 (2005) Mater. Technol. 124 (2002) 41.
034319. [51] B.B. Straumal, B. Baretzky and A.A. Mazilkin
[35] S.V. Bobylev, M.Yu. Gutkin and I.A. Ovidko // // Acta Mater. 52 (2004) 4469.
Fizika Tverdogo Tela 48 (2006) 1410.

View publication stats