You are on page 1of 11

Theoretical and Applied Fracture Mechanics 112 (2021) 102923

Contents lists available at ScienceDirect

Theoretical and Applied Fracture Mechanics


journal homepage: www.elsevier.com/locate/tafmec

Effect of anisotropy on the ductile fracture in metal reinforcements of


brittle matrix composites
Madhu Kiran Karanam , Gopi Gulivindala , Viswanath R. Chinthapenta *
Micromechanics Lab, IIT Hyderabad, Sangareddy, Kandi, Telangana 502285, India

A R T I C L E I N F O A B S T R A C T

Keywords: Mechanisms of ductile failure in brittle matrix composites with metal reinforcements are investigated using a
Ductile fracture mechanistic approach to incorporate material anisotropy. Rapid void expansion with increasing remote strain
Plastic anisotropy and at nearly constant remote strain (cavitation) is modeled. Fracture energy, total energy absorbed, and
Fracture energy
cavitation limit is obtained as a function of material anisotropy. It is observed that for initial void volume fraction
Material instability
Void expansion
less than or equal to 10− 4 cavitation phenomenon is seen. The critical stress at cavitation is found to increase
with decreasing void size until it reaches the cavitation limit. Further, it is observed that the ductile fracture of
metal reinforcements in the brittle matrix composite is sensitive to the material anisotropy.

1. Introduction through internal void expansion in metal reinforcement, brittle matrix


cracking, and decohesion of the metal reinforcement from matrix [5].
Ductile failure of the metal reinforcements and their influence on the Earlier experimental studies on the toughening of the brittle matrix
overall mechanical properties of brittle matrix composites is an active composites with metal particle reinforcements provide fundamental
area of research [1 2 3 4]. The metal particle reinforcements act as a insights into the failure mechanisms [5 8 9 10]. Krstic et al., 1981 [8]
bridge for the propagation of a macroscopic crack. It gets stretched as studied crack propagation on the fractured surface of glass-reinforced
crack opens, absorbing energy. In the process contributing to the frac­ with Ni and Al particles. They observed that for Ni particle re­
ture energy [5 6]. The increase in fracture energy depends upon the inforcements with weak interface and high residual stresses, the crack
bonding between the matrix and metal particle reinforcement. The was observed to bow away from the particle. Whereas for Al particle
metal particle reinforcement contributes to fracture toughness of com­ reinforcements with strong interface and less residual stress, dimples
posite through the process of crack bridging [7]. In crack bridging, a were observed on the surface of Al particles indicating ductile failure.
zone is formed near the crack tip called a bridged zone, where metal Fracture surface observations discussed by Sigl et al., 1988 [9] on
reinforcements get trapped and deform plastically as the crack opens. Al2 O3 /Al & WC/Co systems revealed necking in Al indicating crack
The contribution to the toughness of composite from the particle is bridging and fractured surface of Co particles showed ductile failure by
dependent on elastic & thermoelastic mismatch between matrix & par­ void growth. Flinn et al., 1989 [10] studied the microstructure of the Al
ticle reinforcement, and residual stresses in the composite. The focus of reinforcement and interface between Al and Al2 O3 in Lanxide (Al2 O3
the current study is limited to systems where there exists good reinforced with Al) using TEM. They observed that the Al reinforcement
compatibility between the matrix and particle in terms of elastic and had large and small precipitates which acted as site of initiation of void
thermoelastic properties. The amount of toughening depends on the nucleation and dislocation pinning, respectively. Further, they per­
toughness of the individual particle and crack bridge length (i.e., num­ formed an SEM investigation on the fracture surface and observed that
ber of participating particles). The matrix surrounding the particle ex­ the dominant mode of failure was the rapid expansion of a single void up
erts material constraints depending on the scenarios discussed above, to failure. Another mode of failure was through debonding at the
and the contribution to fracture energy, therefore, is influenced by the interface, but the relation between the debonding and fracture energy
degree of constraint [5]. The level of constraint determines the mecha­ was not established. The modes of failure observed in their study were
nism of failure. The brittle matrix composites failure is majorly due to analogs to the modes of failure observed by Ashby et al., 1989 [5] in lead
the combination of three different mechanisms, i.e., ductile failure wire reinforcement in the glass matrix. Ashby et al., 1989 [5] observed

* Corresponding author.
E-mail address: viswanath@mae.iith.ac.in (V.R. Chinthapenta).

https://doi.org/10.1016/j.tafmec.2021.102923
Received 3 October 2020; Received in revised form 31 December 2020; Accepted 31 January 2021
Available online 2 February 2021
0167-8442/© 2021 Elsevier Ltd. All rights reserved.
M.K. Karanam et al. Theoretical and Applied Fracture Mechanics 112 (2021) 102923

four different mechanisms of failure in their fractographic study. That is instability and the deformed void shape. Stolken and Evans, 1998 [28],
the failure by growth of single void in metal reinforcement, failure by Fleck & Hutchinson, 2001 [29], and Niordson and Tvergaard, 2006 [30]
decohesion plus the growth of internal void, failure by decohesion plus studied the influence of size effects on cavitation instabilities for smaller
growth of multiple voids, and failure by matrix cracking. voids. A common observation from these studies was that size effects
In this work, the effect of the anisotropy on fracture energy of metal were important when the void radius is in the order of characteristic
particle reinforcement is studied, emphasizing the failure through an in­ length scale of metals [31]. Niordson and Tvergaard, 2006 [30] found
ternal void expansion, which is observed to be a dominant cause of ductile that the cavitation instability was also observed using strain gradient
fracture [5]. Further, a systematic approach is followed to arrive at the plasticity similar to the conventional plasticity theory but at higher
bounds for cavitation instability and estimate the cavitation limit in metal strains (delayed). However, the delayed initiation of cavitation insta­
reinforcement. The cavitation limit is the stress at which unstable growth bility does not influence the critical stress at which cavitation instability
is initiated for a benign void. To the author’s knowledge, only limited occurs. In general, cavitation instability is observed in the presence of
studies have been performed on the ductile failure of metal reinforcements very high-stress triaxiality (pure hydrostatic stress). For such remote
[5 11 12 13 14]. An earlier study from Ashby et al., 1989 [5] used 2D plane stress states the elastic energy in the bulk material is large enough to
strain models employing isotropic material constitutive law to correlate drive the expansion of void in an uncontrolled manner even for smaller
their experiments. Later works from Tvergaard and co-authors [11 12 13 void sizes. However, this is not true for low and moderate stress triaxi­
14] used 3D models and introduced anisotropy through phenomenolog­ ality, where the size effects are predominant. Further, as the void grows
ical material models. The metal reinforcements are anisotropic in nature, due to deformation, the length scale reduces with respect to void size,
and they strongly influence the ductile failure [15]. The phenomenolog­ and therefore the effects are less significant. As instability initiates at
ical material models used in the studies described above did not account very small applied strain values, the length scale effects are short-lived.
for the plastic spin, which is an important part of problems involving Tvergaard, 2004 [12] carried out an equivalent numerical calculation
plastic anisotropy [16]. While crystal plasticity based constitutive law in for the experimental study on constrained metal wires by Ashby et al.,
the current study provides the plastic spin, as shown in Section 2.2. 1989 [5] to understand the effect of residual stress on cavitation insta­
The phenomenon of cavitation instability in metal reinforcements of bility. They observed that the fracture toughness decreased with an in­
brittle matrix composites is triggered under high-stress conditions, such crease in residual stress. Legarth and Tvergaard, 2010 [24] performed
that the energy available from the strained material surrounding the quasi-static transient analysis on a 3D finite element model and found
incipient void is sufficient to drive its continuous expansion at constant a strong effect of plastic anisotropy on critical stress for cavitation
remote stress [11 17 18 19 20]. However, it differs from void growth instability. Cohen and Durban, 2013 [32], studied plastic instabilities in
observed during the ductile failure process, which occurs proportion­ porous cylinders under triaxial loading and observed that the cavitation
ately to the imposed strain on the solid [21 22]. Ductile failure by instability occurred at very low initial porosity values. They, however,
cavitation instability is observed in an experimental study on metal re­ did not model the void explicitly but used Gurson’s porous plasticity
inforcements in the brittle matrix by Ashby et al., 1989 [5] for single and model. Tvergaard and Legarth [13 14], recently studied the effect of
polycrystalline lead wires. Brittle matrix does not deform plastically, plastic anisotropy and initial void shapes on cavitation instability and
and it exerts very high-stress levels on the embedded metal re­ void shape evolution using quadratic and non-quadratic anisotropic
inforcements. Such high-stress triaxiality conditions drive cavitation yield criterion, respectively. They observed that plastic anisotropy had a
instability. Flinn et al., 1989 [10] observed a single dominant void on the significant effect on cavitation instability and deformed void shape.
fracture surface of Al2 O3 reinforced by Al particles indicating rapid Further, deformed void shapes were spheroidal when plastic anisotropy
growth of single void. Dalgleish et al., 1989 [23] observed that in thin was considered, unlike with isotropic material where nearly spherical
ductile metal sheet bonded by ceramic solid, a strong constraint on the voids were observed even with initial spheroidal void shapes.
material leads to high-stress triaxiality promoting rapid void growth. In this study, the cavitation instability in Cu crystal is analyzed for a
Ashby et al., 1989 [5] used the finite element method, slip line field range of void sizes represented by void volume fraction (f0 ), i.e., f0 =
theory, and Bridgeman method to investigate metal reinforcement’s 10− 3 to 10− 7 . The motivation behind performing this study on the
behavior under high constraint conditions. The slip line theory was used metal crystal is that they are the fundamental building blocks of poly­
to capture the deformation mechanism. A natural extension to slip line crystalline solids. And they represent a single continuous grain enabling
theory is crystal plasticity, which accurately captures the deformation the study of the plastic deformation mechanism of the constituent grains
mechanisms for crystalline solids. Previous works [24 13 14] on cavi­ of the polycrystal. This study is applicable under the assumption that
tation instability, discussed next, studied the problem using the elasto- void is contained inside a grain and is of sufficiently small size compared
plastic material model. However, the plastic spin, which is an impor­ to grain size itself. The above assumption is valid considering that earlier
tant parameter in the study, is usually not incorporated in their material studies have shown the possibility of both ductile ruptures in materials
models. To capture the plastic anisotropy, the plastic spin needs to be that do not contain hard particles or pre-existing voids [33] and void
accounted for [16]. Therefore, it is necessary to estimate the critical nucleation in metal single crystals under multi-slip [34]. A simplistic
stress at which the instability is initiated using the mechanistic approach approach to estimate fracture energy contribution from metal rein­
of modeling material behavior such as crystal plasticity. Using crystal forcement to brittle matrix composites is presented. The novelty of this
plasticity, the current study extends the envelope of critical stress for work is to capture the anisotropic effects on fracture energy and cavi­
different void volume fraction tending to zero. This helps to estimate the tation limit. Furthermore, void volume fraction at incipient cavitation
critical stress to initiate the cavitation instability at material in­ instability is determined, which was not addressed in the earlier works.
homogeneities, where benign voids are present. The rest of the paper is organized in the following order. The problem
Bishop et al., 1945 [25], identified cavitation instability limits in formulation describing the representative material volume (RMV),
elasto-plastic material under axisymmetric stress conditions. Hill, 1948 constitutive framework, and fracture energy estimation is discussed in
[26] studied this phenomenon for cavities subjected to internal pressure. Section 2. The results on the role of anisotropy on ductile failure
Ball, 1982 [27] obtained material instability limits for non-linear elas­ mechanisms in metal reinforced brittle matrix composites are discussed
ticity. Huang et al., 1991 [11] observed that the criterion for cavitation in Section 3, followed by conclusions in Section 4.
instability in elasto-plastic solids under axisymmetric loads depends on
the critical value of mean stress. Tvergaard et al., 1992 [18], tested the 2. Methodology
significance of initially spheroidal void shapes using a unit cell con­
taining a single void. They found that the initial void shape did not The brittle behavior of conventional ceramic materials limits its
significantly affect the critical stress at the initiation of cavitation

2
M.K. Karanam et al. Theoretical and Applied Fracture Mechanics 112 (2021) 102923

energy absorption. Therefore, ceramic materials with a brittle matrix are 2.1. Problem formulation
reinforced with ductile metal particles due to their high strength and
toughness [5]. Several such studies showed better toughness and The ductile failure of metal reinforcement occurs primarily by a
strength in composites due to the addition of metal reinforcements to the single void growing rapidly until the ligament shears off under high
ceramic matrix [8 9 10 5 35 36]. The toughening of the ceramic matrix constraint conditions [5]. This rapid growth of void is initiated at critical
composites is due to the metal reinforcement’s resistance to crack stress, after which the stress-strain response of material shows a steep
growth due to the crack bridging phenomenon, which leads to higher fall. As the void (or cavity) grows rapidly under constant stress, it con­
energy absorption. The advancing crack, when encounters the ductile stitutes a state of material instability triggered by cavity expansion,
metal particle, it stretches as the crack opens until it eventually fails or therefore, termed as cavitation instability [11]. An essential condition
decoheres from the matrix [5]. Fig. 1 represents the schematic of the for initiation of the cavitation instability is due to the presence of high
failure mechanism by the ductile failure of metal reinforcement. constraint [11 5 24 14 12]. A representative material volume (RMV) is
The fracture toughness improvement is observed as a consequence of shown in Fig. 2. The RMV consists of ductile metal reinforcement with a
energy absorbed by the ductile failure of reinforcement, decohesion of single void embedded in it. The matrix and the macroscopic crack in the
reinforcement from the matrix, and cracking of the matrix material. All composite are not modeled explicitly; only metal reinforcement is
three failure mechanisms are studied in this work by incorporating considered in the simulation as the scope of work is a ductile fracture in
appropriate boundary conditions. A detailed problem description (Sec­ metal reinforcement.
tion 2.1), the constitutive framework (Section 2.2), and estimation of the The displacement of the material point in the RMV is given by u =
energy release rate (Section 2.3) are presented in this section. x − X, where X and x are positions in the initial configuration and
deformed configuration, respectively. The displacement and traction

Fig. 1. A schematic of fracture in brittle matrix composite with metal reinforcement.

Fig. 2. One eighth symmetry model of the Representative Material Volume, showing boundary conditions, representative mesh, and initial crystallographic
orientation used in the study.

3
M.K. Karanam et al. Theoretical and Applied Fracture Mechanics 112 (2021) 102923

Table 1
Designation of slip systems in the face-centered cubic single crystal [16].
Slip plane (111) (111) (111) (111)
Slip direction [011] [101] [110] [011] [101] [110] [011] [101] [110] [011] [101] [110]

boundary conditions used in the simulations are specified in terms of metal crystal is discussed in Section 2.2. Full RMV increases the
displacements, ui , and tractions on the faces of the RMV, Ti , as in Eq. (1). computational expense by eight times and more for simulating cavita­
tion instability where void grows rapidly, leading to multiple rezoning
u1 = 0; T2 = T3 = 0 at X1 = 0
and restarts during simulations.
In Fig. 2, two coordinate systems (black and red) are shown; black
u2 = 0; T1 = T3 = 0 at X2 = 0
represents the specimen coordinate system, and red represents the
u3 = 0; T1 = T2 = 0 at X3 = 0 crystal coordinate system. The orientation of the crystal axes w.r.t. the
main loading direction, X2 , represents the initial crystallographic
u 1 = U1 ; T2 = T3 = 0 at X1 = L10 orientation (ICO). In the current study, ICOs [1 0 0], [1 1 0], & [1 1 1] are
used, refer to Section 2.2 for details. The initial void volume fraction of
u 2 = U2 ; T1 = T3 = 0 at X2 = L20 the spherical void considered in this study is calculated through the
( )3
r0
u 3 = U3 ; T1 = T2 = 0 at X3 = L30 (1) relation, f0 = 43 π 2a 0
. Where r0 is the initial radius of the void, and a0
is the length of half side of the cube, see Fig. 2. The f0 values of
Where L10 , L20 , and L30 are initial dimensions of the RMV in X1 , X2 ,
10− 3 to 10− 7 are used in the study. The evolution of void volume (V) is
and X3 directions, respectively. The displacements in transverse di­
calculated by subtracting the current volume of the matrix from the
rections, U1 &U3 , are applied proportional to the displacement in the
current volume of the cube. The current volume of the matrix is obtained
loading direction U2 . The relation between displacements and loga­
by integrating the volume of all the elements. A typical finite element
rithmic strains are as given below,
model used in the study is shown in the inset of Fig. 2, where the mesh is
( )
2U1 generated out of C3D8 elements in ABAQUS. The constitutive model for
E1 = ln 1 + ,
L10 the crystal plasticity discussed in Section 2.2 is applied through a user
(
2U2
) subroutine UMAT in the ABAQUS standard [38 39].
E2 = ln 1 + , (2)
L20
( ) 2.2. Constitutive framework
2U3
E3 = ln 1 +
L30 The general framework for finite strain crystal plasticity proposed by
The work conjugates quantities of imposed macroscopic strains E1 , Asaro, 1983 [40] is implemented in this work to model FCC crystals. The
E2 , and E3 are the macroscopic true (Cauchy) stresses Σ1 , Σ2 , and Σ3 . multiplicative decomposition of the deformation gradient, F, which
The macroscopic stresses on RMV faces are obtained by the integration maps the current state of deformation to a reference state results in, F =
of surface tractions. F* ⋅FP . Where, FP is the plastic part of deformation gradient solely due to
∫ L2 ∫ L3 slip and F* is the elastic part of F. The crystalline material flows through
1
Σ1 = [T1 ]x1 =L1 dx2 dx3 (3) the lattice via dislocation motion such that the lattice undergoes elastic
L2 L3 0 0
deformations and rotations [41]. As the crystal deforms and rotates,
lattice vectors such as the slip direction s, the normal to slip plane m
Where, L1 , L2 , and L3 are the current dimensions of the RMV in X1 , X2 ,
convect with the lattice. The lattice vectors in the current state on a
and X3 directions, respectively. T1 , T2 , and T3 are the tractions on the
faces of RMV with normal along X1 , X2 , and X3 directions, respectively. given slip system α can be represented as s’(α) = F* ⋅s(α) ;m’(α) =
Σ2 and Σ3 are obtained similarly by cyclic permutation of Eq. (3). Note m(α) ⋅F*− 1 . Slip system designation for FCC single crystal is shown in
that there are no shear tractions applied to the RMV. Table 1. The velocity gradient in the current state is given by L =
RMV is subjected to a proportional macroscopic triaxial strain field Ḟ⋅F− 1 = D + Ω, where D and Ω are the total rate of stretching and spin
imposed using logarithmic strains such that different material con­ tensors, respectively. Specified by
straints are achieved. The logarithmic strains E1 , E2 , and E3 satisfies
D = D* + DP
E1 E3
Γ∊ = = (4)
E2 E2 Ω = Ω* + ΩP (5)

at all times. Where Γ∊ is the strain triaxiality ratio, which is an indicator Where, D* and Ω* are the elastic part of stretch and spin tensors repre­
of the constraint from the surrounding matrix on ductile reinforcement. senting the lattice deformations and rigid body rotations and the plastic
The values of Γ∊ range from − 1.0 to 1.0. Γ∊ = 1 corresponds to hydro­ shearing contributions are represented by the plastic part of stretch and
static loading leading to the case of the highest constraint. Γ∊ = 0 cor­
spin tensors DP and ΩP , given by
responds to uniaxial tensile loading and Γ∊ →− 1.0 corresponds to the
∑ ∑
load state at a blunt crack tip representing the least constraint [37]. In DP = P(α) γ̇(α) , ΩP = W (α) γ̇(α) (6)
the current study, − 0.45 ≤ Γ∊ ≤ 1.0 is considered. A one-eight sym­ α α

metric model is considered in this study due to symmetry in the loading


and geometry of the RMV. The symmetry due to material anisotropy Where,
cannot be generalized for all initial crystallographic orientations. 1
Therefore, ICOs [1 0 0] & [1 1 0] with three four-fold axes symmetry are P(α) = (s(α)’ m(α)’ + m(α)’ s(α)’ )
2
considered in this study. And for generic material anisotropy arising due
to various initial crystallographic orientations, a full RMV should be 1
W (α) = (s(α)’ m(α)’ − m(α)’ s(α)’ ) (7)
considered [12], however, for comparison purposes ICO [1 1 1] is also 2
considered with a one-eighth model. The material anisotropy in the It is assumed that the elastic properties are not affected by slip and

4
M.K. Karanam et al. Theoretical and Applied Fracture Mechanics 112 (2021) 102923

Table 2
Material parameters [45 43].
C11 C12 C44 n k̇ h0 τ0

168 GPa 121.4 GPa 75.4 GPa 10 0.001 541.48 MPa 60.84 MPa
τs q
109.51 MPa 1

are characterized by a strain energy function ϕ [42]. The rate of work Where, h0 is the modulus of hardening at reference state, τ0 is the yield
per unit reference volume is specified by stress at the reference stage, τs is stress at the easy-glide stage, and γ is

τ : D = τ : D* + τ : Dp = τ : D* + τ α γ̇α (8) total shear strain over all the slip systems defined as
α ∑∫ t
γ= |γ̇ (α) |dt (17)
Where τ is the Kirchoff stress and τ α = τ : P(α) is the resolved shear stress α 0

on the slip system α. Eq. (17) provides a generalized definition of γ suitable for single and
Introducing the lattice Green strain, E* multiple slips. A good correlation has been shown for this hardening law
∂ϕ * T 1 with the experimental hardening behavior of Cu crystal [43 44]. Table 2
τ = F* . .F and E* = (F* .F* − I) (9)
T

∂E* 2 lists the material parameters used in this study for Cu crystal [43 45].
Where C11 , C12 , and C44 are the elastic constants for cubic material, n is
Where F is the deformation gradient. The Jaumann stress rate is given the rate sensitivity exponent, k̇ is the reference strain rate, h0 is the initial
by hardening modulus, τ0 is the yield stress at the reference stage, τs is stress
at the easy-glide stage and qis the latent hardening parameter.
∇*
τ ≈ L : D* (10) The plastic deformation in FCC metal crystals is attributed to the
mechanism of glide on preferential crystallographic slip planes {1 1 1}
Eq. (10) is a simplified constitutive relation after ignoring the terms
and slip directions 〈110〉. Schmid and Baos, 1938 [46] observed that a
∇*
of O(τ /L). Where τ is the Jaumann rate of Kirchhoff stress defined as critical value of shear stress resolved onto a given slip system (τ = P/
∇* Acosϕcosθ) is required for slip initiation and termed it as critical resolved
τ = τ̇ − Ω* ⋅τ + τ .Ω* , τ̇ is the material rate of Kirchhoff stress, i.e., τ =
shear stress or Schmid stress. Where A is the current area of the cross-
det(F)T, T is the Cauchy stress and L is the elastic moduli tensor such
section, P uniaxial tensile load, and θ & ϕ are the angles between the
that Lijkl = Ljikl = Lijkl = Lklij , the symmetry is possible if we consider L to
tensile axis & the slip direction and slip plane normal. Schmid’s obser­
be derivable from a potential function. The Jaumann stress rate w.r.t
vation, in general, holds good for the pure FCC crystals. However, de­
material axes are required for the formulation of the constitutive law,
viations are observed in FCC crystals containing alloying elements and
and it is defined as τ = τ̇ − Ω⋅τ + τ .Ω. Using the two stress rates

loading conditions resulting in the activation of multiple slip systems
dependent on the lattice and material rotation, respectively, we obtain. [47 48]. In an FCC crystal, due to its material symmetry, the influence of
material anisotropy on uniaxial yielding can be quantified by analyzing
∇* ∑n
τ − τ = (Ω − Ω* ).τ + τ .(Ω − Ω* ) = (11)

α=1
βα γ̇α ICOs [1 0 0], [1 1 0], & [1 1 1]. These ICOs form the vertices of the
standard stereographic triangle. A generic direction [hkl] in the standard
Where, βα = W α .τ − τ .W α . By combining Eqs (7), (10), and (11), the stereographic triangle spans within 0 ≤ ϕ ≤ 4π and 0 ≤ θ ≤
constitutive equation is obtained as arctan(1/cosϕ). The orientations [1 0 0], [1 1 0], and [1 1 1] are sufficient
∑n to understand the effect of material anisotropy as they constitute the
(12)

τ = L : D− α=1
[L : Pα + βα ]γ̇ α points of maximum and minimum yield stress ratio [49]. Further, the
orientations [1 0 0], [1 1 0], and [1 1 1] are multi-slip orientation, which
The plastic slip rate γ̇ as a function of resolved shear stress (τ α ) and
α
results in 8, 4, and 6 active slip systems [50], respectively. The initial
the current strength of each slip system (g α ) is given as
yield stress of the material, σ Y , is determined from the macroscopic
⎛ ⎞⃒ ⃒n− 1
⃒ ⃒
α ⃒ α⃒
stress (Σ2 ) vs. applied strain (E2 ) curve at yield strain. The yield strain is
α τ τ
γ̇ = k̇ ⎝ α ⎠⃒⃒ α ⃒⃒
α
(13) taken as the applied strain at which the value of total cumulative shear
g ̲
⃒g ⃒ ̲ strain (γ) is greater than zero. Total cumulative shear strain (γ) is the
sum of shear strains on all slip systems, see Eq. (17). In the crystal
α
Where n is the rate sensitivity exponent, k̇ defines the reference strain plasticity framework used in the current study, the total cumulative
rate on slip system α. Resolved shear stress (τ α ), in the Eq. (14) is the strain is the only contribution to the plastic strain of the material.
driving force for the plastic strain γα .
2.3. Fracture energy contribution from metal reinforcement
τ α = τ : Pα = τ : (s’α ⊗ m’α ) (14)
Plastic flow is initiated on a slip system,α, if τα ≥ ταcr . Where ταcr is the The interest in the current work is to estimate the increase in the fracture
critical resolved shear stress. The strain hardening is defined by: energy of the composite due to the metal reinforcement (ΔGMR ), i.e., the
∑ energy released during an infinitesimal void expansion in the metal rein­
g α= hαβ γ̇ β (15) forcement. The numerical estimation of ΔGMR can be obtained using the M
integral [51] or virtual void extension approach. In this study, virtual void
β

Where hαβ = qhαα (α ∕ = β) is the latent hardening modulus, q is the latent extension is used for estimation of energy release rate (ΔGMR ), i.e., the void
hardening parameter, β is the number of all slip systems and hαα is the surface is extended, and the increase in the strain energy stored is calculated
self-hardening modulus. The hardening modulus hαα is proposed by with respect to the increased area of the void surface. The energy release
Pierce, Asaro, and Needleman [43], rate for a strain-controlled approach (constant δ) is given by [52]
⃒ ⃒ ( )
ΔU
⃒ h0 γ ⃒
hαα = h(γ) = h0 sech2 ⃒⃒ ⃒ (no sum on α) (16) ΔGMR = lim − (18)
(τs − τ0 ) ⃒
ΔAv →0 ΔAv δ

5
M.K. Karanam et al. Theoretical and Applied Fracture Mechanics 112 (2021) 102923

studied as a function of material anisotropy. Further, the effect of ma­


terial anisotropy on the cavitation limit is discussed in Section 3.4. The
rapid expansion of the void observed during the cavitation instability
demands mesh rezoning as significant errors are observed in response
due to large mesh distortions. The method used for reducing mesh dis­
tortions is detailed in Section 3.1.

3.1. Reducing mesh distortion errors

The simulations of RMV with void were continued until a large void
volume increase more than 80 times the initial void volume in most of
our simulations. The mesh distortions were seen due to the large
expansion of the void. Due to these mesh distortions, the solution pro­
cedure is abruptly terminated. To achieve large VV0 the mesh distortions
are corrected using mesh rezoning. In this approach, the solution is
stopped when the mesh gets distorted, and it is corrected manually. The
solution at which the simulation was stopped is mapped onto the cor­
Fig. 3. Mesh-to-mesh solution mapping to avoid element distortion errors and rected mesh, and the simulation is restarted again. The solution equi­
achieve a higher void volume ratio. For initial void volume fraction (f0 ) = librium was achieved over a few increments before resuming the
10− 5 , Initial crystallographic orientation (ICO) [1 1 0], & strain triaxiality fac­ analysis after every restart. This procedure is referred to as a mesh-to-
tor (Γ∊ ) = 1.0. (a) Initial mesh, (b) restart-1, and (c) restart-2. mesh solution mapping [53] and is shown in Fig. 3 for a typical simu­
lation (f0 = 10− 5 , Γ∊ = 1.0, & ICO [110]) from our study. Mesh-to-
mesh solution mapping approach is used in the current study among
Where,ΔAv is the increase in the void area by virtual displacement of
the available approaches in ABAQUS due to trade-off between accuracy,
void surface; ΔU is the increase in strain energy stored in the body, i.e.,
feasibility, and solution time. The amount of diffusion in the mapped
the difference in the strain energy for an RMV with a void surface area Av
solution due to mesh-to-mesh solution mapping is moderated in our
and RMV with void surface area (Av + ΔAv ).
simulations by performing the mesh rezoning ten increments prior to the
increment where significant mesh distortions are observed. Ensuring
3. Results and discussions
that the computer simulations are of acceptable quality. The mesh dis­
tortions were measured using element quality criteria in-built in ABA­
The study on cavitation instability was carried out for FCC Cu crystal
QUS [53]. The mesh rezoning is performed, keeping the void shape
for initial void volume fraction, f0 , ranging from 10− 3 to 10− 7 . The
unchanged before and after every restart. Multiple restart simulations
selection of the void volume fraction was made such that f0 →0, repre­
were required to achieve a large increase in the void volume.
senting a minute benign void. The material anisotropy is accounted for ( )
by using CPFEM framework for three different initial crystallographic Fig. 3 shows the normalized macroscopic stress ΣσY2 vs. void volume
orientations (ICO = [1 0 0], [1 1 0], & [1 1 1]). A range of material con­ ( )
straints are imposed by varying Γ∊ from − 0.45 to 1.0. The stress-strain ratio VV0 for f0 = 10− 5 , Γ∊ = 1.0, & ICO [110]. Where, V is the
curve of RMV with a void subjected to triaxial loading is governed by
deformed void volume and V0 is the initial void volume. This simulation
two scenarios in the void deformation process. Scenario-1 is peak stress;
required two restarts; the restart points are highlighted by the red open
the incipient stress from which void grows rapidly with increasing
remote strain. And the scenario-2 is the critical stress at which the void circle on the curve. The first restart was performed at VV0 = 15 and second
V
growth continues rapidly at nearly constant remote strain or stress. The restart was performed at V0 = 50. Insets in the Fig. 3 demonstrates the
results for scenario-1 are detailed in Section 3.2, along with the results of process of mesh rezoning and solution mapping. Inset (a) shows the
the investigation on fracture energy and total energy absorbed by metal initial mesh of the RMV zoomed in to show mesh flow near the void. The
reinforcement. In Section 3.3, the conditions leading to scenario-2 is element aspect near the void is intentionally kept large to allow for a
higher void volume increase. Inset (b) shows the steps followed for
restart-1; the distorted mesh is first imported into Altair Hypermesh at a
given increment. Altair Hypermesh is a specialized tool effective for
mesh rezoning on orphan-mesh. The distorted mesh is then corrected,
such that the shape of the deformed void is unchanged. The last
converged solution before restart-1 is then mapped on to the corrected
mesh. Equilibrium for the mapped solution is achieved over the first few
increments in the restart run. Inset (c) shows similar steps followed for
the restart-2. Using this approach, a very large void volume increase of
more than 80 times was possible.

3.2. Rapid void growth with increasing remote strain

The remote strain on the RMV is defined using the strain triaxiality
ratio (Γ∊ ). The range of Γ∊ is − 1.0 ≤ Γ∊ ≤ 1.0. Where Γ∊ = − 1.0 rep­
resents the state of least constraint on the metal reinforcement from the
surrounding matrix, Γ∊ = 0.0 represents uniaxial straining, and Γ∊ = 1.0
represents hydrostatic loading, which is the state of maximum
Fig. 4. Macroscopic stress vs. applied strain plot for initial void volume fraction constraint. Similar values of Γ∊ were used by Faleskog et al. [37] for a 2D
(f0 ) = 10− 3 , Initial crystallographic orientation [1 1 0] & strain triaxiality factor plane strain RMV.
(Γ∊ ) = − 0.45, − 0.25, 0.0, &1.0. In this section, the variation of normalized macroscopic stress with

6
M.K. Karanam et al. Theoretical and Applied Fracture Mechanics 112 (2021) 102923

Fig. 5. (a) Fracture energy (ΔG) for different material constraints represented by strain triaxiality (Γ∊ ); (b) Energy absorbed (ΔU)for different material constraints
represented by strain triaxiality.(Γ∊ )

respect to applied remote strain (%) is shown for the four values of Γ∊ = strain curve (See Fig. 8b&c in [5]). Our results followed a similar
− 0.45, − 0.25, 0.0, &1.0. For Γ∊ > − 0.5, the void expands & Γ∊ < − 0.5 behavior without a double peak. Due to the absence of explicit modeling
void shrinks and at Γ∊ = − 0.5, the state of incompressibility is reached. of brittle matrix and interface, a double peak is not captured in the
At this remote strain value, numerical simulations are difficult to current simulation. Further, it is observed that the ductility and energy
perform, hence, Γ∊ = − 0.45 (nearly incompressible) is considered in our absorbed in failure due to partial cavitation with brittle matrix cracking
analysis. The selection of the zero and positive values of Γ∊ is straight or decohesion is higher than the ductile failure of metal reinforcement
forward as they represent positive limit and central value, respectively. with void expansion, similar to the experimental results of Ashby et al.,
Further, an additional negative value of Γ∊ = − 0.25 is included to un­ 1989 [5].
derstand the trend of peak stress. Fig. 5a illustrates the effect of the orientation of metal reinforcement
The rapid void growth with increasing remote strain is observed at on fracture energy. Virtual void expansion approach described in Sec­
an initial void volume fraction of f0 = 10− 3 and higher in our simula­ tion 2.3 is followed to obtain fracture energy. In the figure, fracture
tions (scenario-1). A typical stress-strain curve with a higher initial void energy (ΔGΓ∊ ) is normalized with respect to fracture energy of un­
[hkl]

volume fraction (f 0 ≥ 10− 3 ) is identified by peak stress, which repre­


constrained metal reinforcement for [1 0 0] orientation (ΔGΓ∊ =0 ). It is
[100]
sents the initiation of rapid void growth with increasing remote strain.
observed that in both the failure mechanisms, the contribution to the
Hence, in Fig. 4, the results are presented for a void volume fraction of
fracture energy is higher compared to that of the unconstrained metal
10− 3 , ICO [1 1 0], and varying values of Γ∊ (− 0.45 to 1.0). For higher
reinforcement. However, the maximum contribution to fracture energy
material constraint (Γ∊ = 1.0) steepest drop in the stress-strain curve is
is seen for ductile failure with void expansion. The normalized fracture
observed post-peak. With the decrease in the material constraint
energy for ductile failure with void expansion (Γ∊ = 1.0) for ICO’s
(decrease in Γ∊ ), the stress-strain curve post-peak plateaus gradually (see
[1 1 1], [1 0 0] & [1 1 0] is 3.97, 3.25, & 3.28, respectively. Similarly, the
Fig. 4). The dotted line in Fig. 4 represents the envelope of the peak
normalized fracture energy for failure due to partial cavitation with
stress for varying remote strain. For Γ∊ (1.0 to 0.0) the peak stress re­
brittle matrix cracking or decohesion (Γ∊ = − 0.45) for ICO’s [1 1 1],
mains nearly constant, as for Γ∊ > 0, the transversely applied remote
[1 0 0] & [1 1 0] is 3.37, 3.15, & 2.23, respectively.
strain (E1 , E3 ) constraints the RMV in the same sense as the longitudinal
Fig. 5b illustrates the effect of orientation on the energy absorbed by
applied remote strain (E2 ). Whereas, the peaks stress falls non-linearly
metal reinforcement(ΔU). The energy absorbed due to resistance offered
with decreasing Γ∊ (0.0 to − 0.45), as for Γ∊ < 0, the transversely
by constrained metal reinforcement is calculated up to the peak stress
applied remote strain partially relieves the constraints on the RMV as its
value. In the figure, energy absorbed(ΔUΓ∊ ) is normalized with respect
[hkl]
sense is opposite to the longitudinal applied remote strain. Faleskog et al.
1997 [37], observed similar dependence of the peak stress on Γ∊ . to energy absorbed by unconstrained metal reinforcement for [1 0 0]
In Fig. 4, the stress-strain curve for Γ∊ = 0.0 represents uncon­ orientation (ΔUΓ∊ =0 ).
[100]

strained condition of metal reinforcement. While Γ∊ = 1.0 (high The normalized energy absorbed for ductile failure with void
constraint) is analogues to ductile failure through void expansion, expansion (Γ∊ = 1.0) for ICO’s [1 1 1], [1 0 0] & [1 1 0] is 0.44, 0.37, &
similarly, Γ∊ = − 0.45 (low constraint) is analogues to failure through 0.37, respectively. Similarly, the normalized energy absorbed for failure
partial cavitation with brittle matrix cracking or decohesion [5 54 37]. due to partial cavitation with brittle matrix cracking or decohesion (Γ∊ =
Current simulation results are qualitatively compared with Ashby et al., − 0.45) for ICO’s [1 1 1], [1 0 0] & [1 1 0] is 2.43, 2.27, & 1.50, respec­
1989 [5]; the quantitative comparison is not possible due to the differ­ tively. The energy absorbed for failure through the partial cavitation
ence in the material used in both the studies. It is observed that the with brittle matrix cracking or decohesion is 5.59, 6.11 & 4.09 times the
results from our simulations show a good correlation with experimental energy absorbed for failure through void expansion for ICO’s [1 1 1],
results from Ashby et al., 1989 [5]. In the case of ductile failure with void [1 0 0] & [1 1 0], respectively. The higher energy absorption was
expansion (Γ∊ = 1.0) a steep rise to the peak followed by a steep fall in observed for the failure through cavitation with brittle matrix cracking
the stress-strain curve is observed similar to Ashby et al., 1989 [5] (See or decohesion, similar to Ashby et al., 1989 [5]. From these results, it is
Fig. 8a in [5]). While in case of failure due to partial cavitation with observed that fracture energy and the energy absorbed by metal rein­
brittle matrix cracking or decohesion (Γ∊ = − 0.45), Ashby et al., 1989 forcement vary significantly with the material anisotropy. In general,
[5] observed a double peak and significant work hardening in the stress- the contribution to the fracture toughness is higher for the ICO’s [1 1 1]

7
M.K. Karanam et al. Theoretical and Applied Fracture Mechanics 112 (2021) 102923

( )
Fig. 6. Cavitation instability study for the initial void volume fraction f0 = 10− 3 to10− 7 , Initial crystallographic orientation [1 1 0] & strain triaxiality factor (Γ∊ )
= 1.0.

Fig. 7. Plastic strain plots for void growth and cavitation instability. The conditions used in each of the figures (a)-(d) for initial void volume fraction (f0 ), Applied
strain (E2 ), void volume ratio (VV0 ) are mentioned below the respective figures.

8
M.K. Karanam et al. Theoretical and Applied Fracture Mechanics 112 (2021) 102923

& [1 0 0] compared to ICO [1 1 0].

3.3. Determination of cavitation limit (critical stress as f 0 →0)

The critical stress at cavitation represents material instability. This


instability is due to rapid void growth at almost a constant stress and for
an infinitesimally small increment in the remote strain (scenario-2).
While peak stress discussed in Section 3.2 does not represent material
instability as the rapid void growth is observed with increasing remote
strain. In this section, the analysis is carried out for six different f0 values
ranging from 10− 3 to 10− 7 with the objective to understand the cavi­
tation instability; to arrive at the f0 below which cavitation occurs, and
to determine cavitation limit (Σc ).
The simulation with smaller void sizes (f0 < 10− 7 ) are computa­
tionally expensive and sensitive to the precision; therefore, the current
study is performed up to f0 = 10− 7 . And also for the reason that
Fig. 8. Effect of material anisotropy on the cavitation limit (Σc ).
[hkl]

f0 of 10− 7 is sufficiently small to be considered as an approximation for


f0 →0 [37]. Tvergaard and Legarth, 2019 [13] and Legarth and Tvergaard,
portion of the RMV is elastic. Fig. 7b represents the plastic strain con­
2019 [14] in their work on cavitation instability, used f0 = 5⋅10− 7 .
tours at a remote strain of 0.9%, at which the VV0 = 2.98 was observed.
Later in this section, it is shown that f0 = 5⋅10− 7 is sufficient to capture
For the corresponding deformation the mean void radius is 1.44 times
cavitation limit and smaller voids (f0 < 5⋅10− 7 ) leads to precision errors.
the initial void radius. In Fig. 7b, approximately 80% of the RMV is
Fig. 6a shows the normalized macroscopic stress vs. applied strain,
deformed plastically around the void. The material far away from the
E2 (%) for varying f0 = 10− 3 to 10− 7 , ICO [1 1 0], & Γ∊ = 1.0. It is
void is elastic, and a high concentration of plastic strain value of ~2.0 is
observed that the stress-strain curve for f0 = 10− 3 shows a post peak observed on the periphery of the void.
behavior (see Section 3.2) representing a rapid void growth with
Fig. 7c&d represents the plastic strain plots for f0 = 10− 5 before and
increasing strain. Whereas the stress-strain curve for f0 = 10− 4 to 10− 7 at the critical stress (see Fig. 6), respectively. Similar to Fig. 7a, the void
increases and ends abruptly. The rapid void growth at nearly constant
volume ratio, VV0 = 1.03 is observed at a remote strain value of 0.2% in
strain is seen for these f0 values (see together Fig. 6b & a), indicating the
Fig. 7c, hence there is no visible difference between the deformed and
cavitation instability. However, for f0 = 10− 3 , there is no cavitation
un-deformed void shape. The plastic strains less than 0.05 are seen in the
observed as after the peak stress is reached, the material starts to soften
vicinity of the void, while a major portion of the RMV is elastic. Fig. 7d
gradually. This behavior can be attributed to the dominance of material
represents the plastic strain contours at a remote strain of 0.9%, at which
softening due to void expansion over the strain hardening from the
the VV0 = 80 was observed. For the corresponding deformation the mean
material. Hence, in Fig. 6b discussed below, the emphasis is laid on the
f0 values ≤ 10− 4 . Fig. 6b shows the results for the normalized macro­ void radius is 4.31 times the initial void radius. In Fig. 7d, around 80%
scopic stress versus void volume ratio (V/V0 ) for varying f0 in the range of the RMV is deformed plastically. And the material far away from the
void is elastic, and a high concentration of plastic strain value of >3.0 is
of 10− 3 to 10− 7 , Γ∊ = 1.0, and ICO[110]. It was observed that for all f0
values the peak stress is reached for an approximate void volume ratio observed near the void. The smaller void (f0 = 10− 5 ) grows rapidly
V compared to the larger void (f0 = 10− 3 ) for same increment in the
V0 ≤ 30. With the decrease in f0 the critical stress at cavitation converges
remote strain. That is for a remote strain of 0.9% smaller void grows by
to a cavitation limit. That is for f0 = 10− 4 , 10− 5 ,10− 6 , 5⋅10− 7 and 10− 7 ,
80 times whereas larger void grows by 3 times. In the case of a smaller
the critical stress at cavitation observed are 1.82σ Y , 2.71σY ,2.79σY ,
void, rapid void growth is observed at almost constant strain while a
2.85σY , and 3.00σ Y , respectively. The critical stress converges up to f0 =
larger void grows proportional to the remote strain.
5⋅10− 7 on further decreases of f0 to 10− 7 a jump in the critical stress
value is observed (the reasons are discussed later). The difference be­
3.4. Effect of material anisotropy on cavitation limit
tween the critical stress at cavitation for f0 = 10− 6 and f0 = 5⋅10− 7 is
<2.0%. Hence, 5⋅10− 7 can be considered as f0 →0 and cavitation limit as
The variation of critical stress for cavitation instability as a function
Σc ≈ 2.85σ Y . In the earlier studies on the cavitation instabilities from
of material anisotropy is presented in this section. ICO [1 0 0], [1 1 0], &
Tvergaard and co-authors [24 13 14] focus was only on the cavitation
[1 1 1] are considered for studying material anisotropy for 10− 3 ≤ f0 ≤
limit. However, this work provides additional insights such as minimum
5⋅10− 7 and Γ∊ = 1.0 (maximum constraint).
f0 below which the cavitation phenomenon is observed.
Determining the cavitation stress for a very small void helps deter­
Note that for f0 = 10− 7 , even though the results showed a cavitation
mine the stress required for the initiation of the cavitation instability for
phenomenon (Fig. 6b), but the slope of the stress-strain curve (Fig. 6a)
a benign void. The strong effect of anisotropy on cavitation stress was
and the critical stress (Fig. 6b) deviates from that observed for other f0
observed in earlier studies by Tvergaard and co-authors [24 13 14]. But
values. This deviation can be attributed due to the usage of single pre­
information on convergence of the cavitation stress to a constant limit is
cision in all our simulations. For very small voids (f0 ≤ 10− 7 ) precision
not explored in earlier studies, except for [37]. Faleskog et al., 1997 [37]
errors significantly influence the results.
determined the cavitation stress for f0 →0 using 2D plane strain
Fig. 7 presents the plastic strain in the RMV for f0 = 10− 3 &10− 5 , ICO assumption and with the elasto-plastic material model. The convergence
[1 1 0], and Γ∊ = 1.0. Where, f0 = 10− 3 represents a case of rapid void was observed for f0 = 10− 7 . Further, one of the limitations of the vis­
growth with increasing remote strain and f0 = 10− 5 represents a case of coplastic constitutive models used in the earlier studies [13 14 37] is
rapid void growth at a nearly constant remote strain. Fig. 7a&b shows that the plastic spin effects are not considered. To capture the plastic
the plastic strain plots for f0 = 10− 3 before and after peak stress (see anisotropy, also the plastic spin needs to be specified. The crystal plas­
Fig. 4), respectively. The void volume ratio, VV0 = 1.04 is observed at a ticity framework used in this work accounts for a plastic spin as shown in
remote strain value of 0.2% in Fig. 7a, hence there is no visible differ­ Eq. (6). The sensitivity of the material to plastic spin depends on the
ence between the deformed and un-deformed void shape. The plastic strain hardening and the tendency to develop texture [16].
strains less than 0.1 are seen in the vicinity of the void, while the major In Fig. 8, critical stress increases with decreasing f0 for all ICOs until

9
M.K. Karanam et al. Theoretical and Applied Fracture Mechanics 112 (2021) 102923

Fig. 9. Deformed void shape for initial crystallographic orientation (ICO) [1 0 0], [1 1 0], & [1 1 1], strain triaxiliaty factor (Γ∊ ) = 1.0 and initial void volume
fraction (f0 ).

it converges to cavitation limit at f0 = 5⋅10− 7 . Note that the critical decrease in the Γ∊ , due to the opposite sense of transverse and longitu­
stress is measured when the stress vs. void volume ratio curve nearly dinal remote strains. The fracture energy and energy absorbed till peak
flattens out. In the current work, the critical stress is measured at VV0 = stress for the high material constraint is 3.97 and 0.44 times of uncon­
30. The cavitation limit of ICOs [1 0 0], [1 1 0], & [1 1 1] are observed to strained material. While the fracture energy and energy absorbed till
be 3.20σ Y , 2.85σ Y , and 3.40σY , respectively. peak stress for the low material constraint is 3.37 and 2.43 times of
From Fig. 8, it is observed that the order of increasing yield stress unconstrained material. The fracture energy for a high material
[49] and the cavitation limit is the same ([1 1 0], [1 0 0], & [1 1 1]). constraint is higher; however, the total energy absorbed is lower than
Earlier studies by Tvergaard and Legarth [24 13] on the cavitation limit that of the low material constraint. Further, fracture energy and energy
using anisotropic Hill’s solid (II, IV) showed that critical stress increases absorbed by metal reinforcement are highly sensitive to the material
proportionally to the yield stress. anisotropy.
Fig. 9 shows the deformed and undeformed void for the three ICOs For initial void volume fraction (f0 ≤ 10− 4 ) cavitation phenomenon
([1 0 0], [1 1 0], & [1 1 1]), f0 = 10− 5 , and Γ∊ = 1.0. The void shape is is seen (scenario-2). In scenario-2, material instability is observed, i.e., a
observed to have a strong dependence on the ICO. Both spheroidal and significant increase in the void size for nearly constant remote strain or
non-spheroidal void shapes are observed during cavitation. For ICO stress. The cavitation instability simulations are computationally
[1 0 0] initially spherical void deformed into a spheroidal void, while for expensive and challenging due to severe mesh distortions induced by
ICO [1 1 0] & ICO [1 1 1] initially spherical void deformed into a non- expanding void at a nearly constant remote strain. At least two mesh
spheroidal void. The non-spheroidal void shapes are attributed to ma­ rezonings are required to achieve a large increase in void volume ratio,
terial anisotropy and void spin. A previous study [21] on void growth as high as 80. The convergence of the critical stress to cavitation limit
with increasing strain captured the void shape evolution for anisotropic (ΣC ) is obtained systematically. The cavitation limit is observed to be at
material, showing spheroidal and non-spheroidal deformed void shapes. the initial void volume fraction f0 = 5⋅10− 7 . This is in line with earlier
However, a detailed analysis of void growth/shrinkage during cavita­ studies on cavitation instability [13 14]. Further, the cavitation limit is
tion for generic material anisotropy taking into account deformed void found to be strongly sensitive to the material anisotropy. The cavitation
morphology is not fully explored, and it will be part of future study. limit of ICOs [1 1 1], [1 0 0], & [1 1 0] are observed to be 3.40σ Y , 3.20σ Y
and 2.85σY , respectively. Further, both spheroidal and non-spheroidal
4. Conclusions void shapes are observed during the cavitation phenomenon.
The novelty of the current work is in capturing the effect of material
The role of material orientation on deformation mechanisms of anisotropy on cavitation, fracture energy, and total energy absorbed
brittle matrix composites with metal reinforcements are studied: ductile using a mechanistic constitutive model that could accurately capture
failure with void expansion, failure due to partial void expansion with crystallographic slip and plastic spin. The void shape evolution during
brittle matrix cracking, or decohesion. The two ductile failure scenarios cavitation for generic material anisotropy requires detailed study. The
due to void expansion are investigated: rapid void growth with symmetry due to material anisotropy cannot be generalized for all initial
increasing remote strain and rapid void growth under constant remote crystallographic orientations. It will be part of future research.
strain. The fracture energy and conditions leading to rapid void
expansion & cavitation instability are obtained as a function of material CRediT authorship contribution statement
anisotropy. Fracture energy contribution to composite due to the energy
release rate from expanding void in metal reinforcement is calculated Madhu Kiran Karanam: Conceptualization, Methodology, Soft­
using the virtual void extension method. And cavitation instability is ware, Writing - original draft. Gopi Gulivindala: Software, Validation, .
explored, offering insights into the cavitation limit for a benign void Viswanath R. Chinthapenta: Conceptualization, Writing - review &
(f0 →0). editing.
At initial void volume fractions, f0 ≥ 10− 3 , rapid void growth with
increasing remote strain is observed. Peak stress, followed by post-peak
Declaration of Competing Interest
softening of the stress-strain curve, is an indicator of rapid void growth
with the increasing strain (scenario-1). Scenario-1 is not a material
The authors declare that they have no known competing financial
instability but occurs due to the void induced softening, overcoming the
interests or personal relationships that could have appeared to influence
material hardening. High material constraint (0.0 ≤ Γ∊ ≤ 1.0) is analo­
the work reported in this paper.
gous to ductile failure with void expansion. In this case, the peak stress is
nearly constant due to the same sense of transverse and longitudinal
remote strains. Low material constraint (− 0.5 ≤ Γ∊ ≤ 0.0) is analogous Acknowledgment
to ductile failure due to partial cavitation with brittle matrix cracking or
decohesion. In this case, the peak stress decreased non-linearly with a The authors acknowledge funding from the Science and Engineering
Research Board (SERB) through the ECR grant ECR/2016/002063.

10
M.K. Karanam et al. Theoretical and Applied Fracture Mechanics 112 (2021) 102923

References [27] J. Ball, Discontinuous equilibrium solutions and cavitation in nonlinear elasticity,
Philos. Trans. Roy. Soc. B: Biol. Sci. 306 (1982) 557–611.
[28] J. Stölken, A.G. Evans, A microbend test method for measuring the plasticity length
[1] S.R. Pemberton, E.K. Oberg, J. Deana, D. Tsarouchas, A.E. Markaki, L. Marston, T.
scale, Acta Mater. 46 (1998) 5109–5115.
W. Clyne, The fracture energy of metal fibre reinforced ceramic composites
[29] N. Fleck, J.W. Hutchinson, A reformulation of strain gradient plasticity, J. Mech.
(MFCs), Compos. Sci. Technol. 71 (3) (2011) 266–275.
Phys. Solids 49 (2001) 2245–2271.
[2] M. Bernhardt, H. Tellesb, H. Justnes, W. Kjell, Fibre reinforced lightweight
[30] C.F. Niordson, V. Tvergaard, Size-effects in porous metals, Modell. Simul. Mater.
aggregates, J. Eur. Ceram. Soc. 34 (5) (2014) 1341–1351.
Sci. Eng. 15 (1) (2006) S51–S60.
[3] Z. Poniżnik, Z. Nowak, M. Basista, Numerical modeling of deformation and fracture
[31] D. Liu, D.J. Dunstan, Material length scale of strain gradient plasticity: A physical
of reinforcing fibers in ceramic-metal composites, Int. J. Damage Mech. 26 (5)
interpretation, Int. J. Plast. 98 (2017) 156–174.
(2015) 711–734.
[32] T. Cohen, D. Durban, Plastic instabilities in porous cylinders under remote triaxial
[4] J. Maj, M. Basista, W. Węglewski, K. Bochenek, A. Strojny-Nędza, K. Naplocha,
loading, Eur. J. Mech. A. Solids 37 (2013) 193–199.
T. Panzner, M. Tatarková, Effect of microstructure on mechanical properties and
[33] H.G.F. Wilsdorf, The ductile fracture of metals: A microstructural viewpoint,
residual stresses in interpenetrating aluminum-alumina composites fabricated by
Mater. Sci. Eng. 59 (1) (1983) 1–39.
squeeze casting, Mater. Sci. Eng., A 715 (2018) 154–162.
[34] O. Furukimi, C. Kiattisaksri, Y. Takeda, M. Aramaki, S. Oue, S. Munetoh,
[5] M. Ashby, F.J. Blunt, M. Bannister, Flow characteristics of highly constrained metal
M. Tanaka, Void nucleation behavior of single-crystal high-purity iron specimens
wires, Acta Metall. 37 (1989) 1847–1857.
subjected to tensile deformation, Mater. Sci. Eng., A 701 (2017) 221–225.
[6] W. Anzhe, W. Yongzheng, Z. Cong, Z. Tao, H. Liao, On the estimation and modeling
[35] J. Miranda-Hernández, S. Moreno-Guerrero, A.B. Soto-Guzmán, E. Rocha-Rangel,
of fracture toughness in structural ceramics in a simple way, Theor. Appl. Fract.
Production and characterization of Al2O3-Cu composite materials, J. Ceram.
Mech. 103 (2019).
Process. Res. 7 (4) (2006) 311–314.
[7] A.G. Evans, A.H. Heuer, D.L. Porter, The fracture toughness of ceramics, Adv. Res.
[36] K. Se Jin, M. Kyung Ho, K. Young Do, M. In Hyung, A study on the fabrication of
Strength Fract. Mater. (1978) 529–556.
Al2O3/Cu nanocomposite and its mechanical properties, J. Ceram. Process. Res. 3
[8] V.V. Krstic, P.S. Nicholson, R.G. Hoagland, Toughening of glasses by metallic
(3) (2002) 192–194.
particles, J. Am. Ceram. Soc. 64 (9) (1981) 499–504.
[37] J. Faleskog, C.F. Shih, Micromechanics of coalescence - I. Synergistic effects of
[9] L.S. Sigl, P.A. Mataga, B.J. Dalgleish, R.M. Mcmeeking, A.G. Evans, On the
elasticity, plastic yielding and multi-size-scale voids, J. Mech. Phys. Solids 45 (1)
toughness of brittle materials reinforced with a ductile phase, Acta Metal. 36 (4)
(1997) 21–50.
(1988) 945–953.
[38] Y. Huang, A user-material subroutine incorporating single crystal plasticity in the
[10] B. Flinn, M. Rühle, A.G. Evans, Toughening in composites of Al2O3 reinforced with
ABAQUS finite element program, Mech Report 178, Harvard University, 1991.
Al, Acta Metall. 37 (11) (1989) 3001–3006.
[39] J. W. Kysar, P. Hall, Addendum to “A user-material subroutine incorporating single
[11] Y. Huang, J.W. Hutchinson, V. Tvergaard, Cavitation instabilities in elastic-plastic
crystal plasticity in the ABAQUS finite element,” Mech Report 178, Harvard
solids, J. Mech. Phys. Solids 39 (2) (1991) 223–241.
University, 1997.
[12] V. Tvergaard, Effect of residual stress on cavitation instabilities in constrained
[40] R.J. Asaro, Crystal plasticity, J. Appl. Mech. 50 (4b) (1983) 921.
metal wires, J. Appl. Mech. 71 (4) (2004) 560–566.
[41] G.I. Taylor, Plastic strain in metals, J. Inst. Metals 62 (1938) 307–325.
[13] V. Tvergaard, B. Nyvang Legarth, Effects of anisotropy and void shape on
[42] R.J. Asaro, A. Needleman, Overview no. 42 Texture development and strain
cavitation instabilities, Int. J. Mech. Sci. 152 (2019) 81–87.
hardening in rate dependent polycrystals, Acta Metall. 33 (6) (1985) 923–953.
[14] B.N. Legarth, V. Tvergaard, Full three-dimensional cavitation instabilities using a
[43] D. Peirce, R.J. Asaro, A. Needleman, An analysis of nonuniform and localized
non-quadratic anisotropic yield function, J. Appl. Mech. 87 (3) (2019).
deformation in ductile single crystals, Acta Metall. 30 (6) (1982) 1087–1119.
[15] M. Hoffman, B. Fiedler, T. Emmel, H. Prielipp, N. Claussen, D. Gross, J. Rödel,
[44] Y.W. Chang, R.J. Asaro, Lattice rotations and localized shearing in single crystals,
Fracture behaviour in metal fibre reinforced ceramics, Acta Mater. 45 (9) (1997)
Arch. Mech. 32 (3) (1980) 369–388.
3609–3623.
[45] E.H. Jacobsen, Elastic spectrum of copper from temperature-diffuse scattering of X-
[16] A.F. Bower, Applied mechanics of solids, CRC Press, 2009.
rays, Phys. Rev. 97 (3) (1954) 654–659.
[17] I. Demir, H.M. Zbib, M. Khaleel, Microscopic analysis of crack propagation for
[46] E. Schmid, W. Boas, Kristallplastizität, Berlin, Springer, Heidelberg, 1935.
multiple cracks, inclusions and voids, Theor. Appl. Fract. Mech. 36 (2) (2001)
[47] M. Masima, G. Sachs, Mechanische eigenschaften von messingkristallen, Zeitschrift
147–164.
für Physik 50 (3) (1928) 161–186.
[18] V. Tvergaard, Y. Huang, J.W. Hutchinson, Cavitation instabilities in a power
[48] R. Karnop, G. Sachs, Festigkeitseigenschaften von kristallen einer veredelbaren
hardening elastic-plastic solid, Europ. J. Mech., A/Solids 11 (1992) 215–231.
aluminiumlegierung, Zeitschrift für Physik 49 (7) (1928) 480–497.
[19] U. Asim, M.A. Siddiq, M. Demiral, Void growth in high strength aluminium alloy
[49] O. Cazacu, R.B. Benoit, N. Chandola, Solid mechanics and its applications
single crystals: a CPFEM based study, Modell. Simul. Mater. Sci. Eng. 25 (3) (2017),
plasticity-damage couplings: from single crystal to polycrystalline materials,
035010.
Springer, New York, 2019.
[20] V. Tvergaard, Plastic flow localization and ductile fracture, J. Phys. Conf. Ser. 1063
[50] G.P. Potirniche, J.L. Hearndon, M.F. Horstemeyer, X.W. Ling, Lattice orientation
(2018), 012005.
effects on void growth and coalescence in fcc single crystals, Int. J. Plast. 22 (5)
[21] M.K. Karanam, V.R. Chinthapenta, Void growth and morphology evolution during
(2006) 921–942.
ductile failure in an FCC single crystal, Comput. Mech. Thermodyn. (2020).
[51] B. Budiansky, J.R. Rice, Conservation laws and energy-release rates, J. Appl. Mech.
[22] G. Li, S. Cui, A review on theory and application of plastic meso-damage
(1973) 201–203.
mechanics, Theor. Appl. Fract. Mech. 109 (2020).
[52] C.A. Andersson, M.K. Aghajanian, The fracture toughening mechanism of ceramic
[23] B.J. Dalgleish, K.P. Trumble, A.G. Evans, The strength and fracture of alumina
composites containing adherent ductile metal phases, Ceram. Eng. Sci. Proc. 9 (7-
bonded to aluminum alloys, Acta Metall. 37 (1989) 1923–1931.
8) (1988) 621–626.
[24] B. Legarth, V. Tvergaard, 3D analyses of cavitation instabilities accounting for
[53] “Abaqus user manual,” Dassault Systèmes Simulia Corp., Providence, RI, USA.,
plastic anisotropy, ZAMM – J. Appl. Math. Mech./Zeitschrift für Angewandte
2016.
Mathematik und Mechanik 90 (2010) 701–709.
[54] A.G. Evans, The mechanical properties of reinforced ceramic, metal and
[25] R.F. Bishop, H. Rodney, N.F. Mott, The theory of indentation and hardness tests,
intermetallic matrix composites, Mater. Sci. Eng., A 143 (1) (1991) 63–76.
Proc. Phys. Soc. 57 (3) (1945) 147–159.
[26] H. Rodney, A theory of the yielding and plastic flow of anisotropic metals, Proc.
Roy. Soc. London. Ser. A, Math. Phys. Sci. 193 (1033) (1948) 281–297.

11

You might also like