IMP - He2016 PDF

Article
International Journal of Damage

Mechanics
0(0) 1–28
Opening-mode micro-cracking ! The Author(s) 2016
Reprints and permissions:
in brittle coatings on ductile sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1056789516659330
wires/rods ijd.sagepub.com
Xin He, Fangliang Chen and Huiming Yin
Abstract
A simplified two-dimensional model is presented to simulate periodically distributed micro-cracking in a
thin coating fully bonded to an elastoplastic aluminum wire/rod. The alumina coating which is generated by
anodic oxidation is treated as an elastic material, while the ductile aluminum wire substrate is character-
ized by a bi-linear elastoplastic hardening model. An elastoplastic shear lag model is applied to transfer the
shearing stress from the substrate layer to the thin coating. When the coated structure is subjected to
different applied load, the system will undergo different stress levels and exhibit different cracking stages.
Accordingly, explicit solutions corresponding to different loading stages are presented based on a general-
ized axisymmetric formulation. Finite element simulation is employed to verify the present elastic solution
when the applied load is relatively small and the whole system is in the elastic state. Experimental
characterization is conducted to validate the present elastoplastic solution when the substrate or inter-
layer undergoes large deformation. A versatile high-fidelity optical microscope is utilized to check the
micro structure of the coating and continuously monitor the fracture development in the coating layer,
through which valuable detailed micro-cracking information, such as critical applied load corresponding to
crack initiation, crack pattern and crack spacing, is obtained. It shows that the presented fracture model is
able to accurately capture the stress and strain distribution in the coated structure and predict the
fracture initiation, infilling, and saturation in the thin coating layer.
Keywords
Coating/thin film, micro-cracking, opening-mode fractures, energy release rate, crack spacing, shear-lag model
Introduction
The application of a thin ceramic coating is an effective approach to improve the wear resistance and
mechanical properties of metallic or polymeric materials. This technique has been widely used in the
Department of Civil Engineering and Engineering Mechanics, Columbia University, USA

Corresponding author:
Fangliang Chen, Department of Civil Engineering and Engineering Mechanics, Columbia University, 610 Seeley W. Mudd 500 West
120th Street, New York, NY 10027, USA.
Email: fangliang.chen@columbia.edu
Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

2 International Journal of Damage Mechanics 0(0)
aerospace and automobile industries (Wang et al., 1995) for corrosion and abrasion resistance
(Bele et al., 2011; Sreevatsa, 2009), as a thermal barrier (Chen et al., 2012), and so on. Typically,
a substrate-coating system is composed of a relatively brittle coating layer and a ductile substrate.
Under externally applied load, common failure mechanisms such as cracking and delamination in
the coating layer have often been observed (Chen et al., 2015). These negatively affect the perform-
ance of the system and decrease its protection efficacy. Therefore, it is critical to know the failure
mechanism in the coating/substrate system.
A uniaxial tension test is one common method to evaluate the coating fracture and interfacial
strength of the coating-substrate system, owing to its relatively simple testing configuration and a
well-controlled crack pattern (Agrawal and Raj, 1989; Chai, 2011; Xia and Hutchinson, 2000).
Under tensile loading, it is often observed that an opening-mode fracture initiates perpendicularly
to the applied loading. Nakamura and Kamath (1992) presented a three-dimensional finite element
analysis to study the mechanisms of crack growth and decohesion in a highly compliant thin film
bonded to a rigid substrate. The simulation results showed that, in the absence of decohesion, the
stress intensity factor along the leading edge of the crack reaches a steady state value when the crack
length is about twice No need the film thickness. Because of the existence of the singular point along
the leading edge of the crack, many studies have been conducted near the neighborhood of that
point based on the local elastic solution. Beuth (1992) introduced a dimensionless plane strain stress
intensity factor to study the cracking of thin film bonded to an elastic substrate. In order to study the
cracking behavior in a coating layer bonded to an elastoplastic substrate when it is undergoing large-
scale yielding; Beuth and Klingbeil (1996) further developed a numerical model to extend the elastic
work to an elastic-plastic substrate. Although the above mentioned local solutions are useful in the
study of fracture propagation, they cannot be directly used to predict fracture initiation, fracture
spacing, or to study the interaction between fractures. The overall elastic field is essential to provide
a complete picture of cracking in layered materials (Yin, 2010b). In this context, Yin and his cow-
orkers (Yin, 2010a, Yin, 2010b; Yin and Prieto-Muoz, 2013; Yin et al., 2008) developed a two-
dimensional (2-D) elastic model and established analytical solutions for the elastic behavior of a
coating/substrate system. Those analytical solutions are able to predict fracture initiation, infilling,
and saturation in the coating layer when the layered structure remains in the elastic stage, which
makes such solutions only applicable to elastic brittle materials where no or very small plastic
deformation will be developed when it is subjected to external tension or temperature loads.
Obviously, the analytical solutions based on existing elastic models cannot accurately evaluate the
more realistic case of a coating/substrate system undergoing large-scale yielding. However, the
majority of substrates used for these types of ‘‘systems’’ are metallic or polymeric materials,
where considerable plastic deformations have already occurred before cracks or damages to the
coating begin to develop. Therefore, the large plastic deformations that develop in ductile substrates
have to be accounted for in order to achieve better predictions and interpretations of the structural
response and failure mechanisms of the coating/substrate system. To fulfill this purpose, an elasto-
plastic fracture model has been developed recently to study the opening-mode fracture in coating
layered structures undergoing large-scale yielding (Chen et al., 2015).
It is noted that most of the existing studies in literature have only focused on planar structures
with a thin layer of coating or film, while very limited studies can be found on wire or pipe type
structures. Ochiai et al. (2005) found that there is a transition in the failure mechanism of coating/
substrate systems from compressive fracture to inter facial debonding as the coating thickness is
gradually increased. Shiao and Shieu (2000); Shieu and Shiao (1997) concluded that the fracture
morphology depends on the bonding strength of the interface between the coating and substrate.
Traditional mechanical properties of metallic wires with different coatings were experimentally

He et al. 3
measured by (Partridge et al., 1997; Tseng et al., 2015; Wang et al., 2014) without considering
fracture behavior. Since none of the above mentioned studies considered the plastic deformation
in the substrate, there is a high demand for an elastoplastic fracture model that can study the
fracture behavior of the coating layer in a coating/substrate system where the substrate is a ductile
rod or wire. To fill this need, an elastoplastic fracture analysis of a brittle coating bonded to a ductile
wire is conducted in this study.
Consider a thin brittle coating layer with thickness b a, length 2, Young’s modulus E f,
Poisson’s ratio n f that is bonded to a wire substrate with radius a, Young’s modulus E s,
Poisson’s ratio ns. The substrate is subjected to a uniform tensile load as denoted in Figure 1 (left
side). It has been observed that no slip or debonding occurs between the substrate and the coating
layer. Therefore, it is reasonable to assume that the thin film is fully bonded to the substrate
(Chen et al., 2015). Once the opening-mode crack initiates in the coating layer, it will propagate
toward the interface between the coating layer and the substrate. Considering a section between two
adjacent opening-mode cracks, as the geometry and loading are axially symmetric, a simplified 2-D
shear lag model is set up in the polar coordinate as shown in Figure 1 (right side). In this model, the
thin layer of the alumina coating generated by anodization is treated as an elastic media, while the
ductile aluminum substrate is considered an elastoplastic material. An elastoplastic shear-lag model
is applied to transfer the shearing stress from the substrate layer to the thin coating layer (Budiansky
et al., 1986; Chen and Qiao, 2012; Mahesh et al., 2004).
Figure 2 shows the strain–stress curves under tensile loading for an aluminum rod with
and without coating. It is seen that, since the brittle coating is very thin compared to the substrate
Figure 1. Schematic illustration of thin coating/substrate system with multiple parallel cracks (left); and a simplified
2-D shear-lag model with opening-mode fractures (right).

Figure 2. Tensile stress strain cure of the ductile substrate with/without coating.
(b-a a), the effect of the coating layer on the global stress distribution in the substrate can be
neglected. Therefore, it is reasonable to assume the tensile stress in the substrate is uniformly
distributed along the radial direction and thus a one-dimensional analysis is conducted in the sub-
strate layer (Chen et al., 2015). As shown in Figure 2, the axial strain experienced by both the coated
and uncoated wires increase linearly with respect to the applied load until the yielding point is
reached. After the yielding point, the strains again increase proportionally to the applied load,
but at a much lower rate. Thus a bi-linear model will be employed to simulate the elastoplastic
behavior by fitting the two sections of the stress-strain curve with two straight lines.
By following our previous work (Chen et al., 2015), as the applied tensile load continuously
increases, the coating/substrate system will undergo three different stress phases:
(I) When the tensile load is small, the thin film, the interlayer, and the substrate are all in a linear
elastic range. Thus an elastic analysis is conducted.
(II) When the interfacial shear stress in the interlayer or the tensile stress in the substrate reaches
their corresponding yielding strength, plastic behavior will be induced in the ductile interlayer
or in the substrate.
(III) As the tensile load is further increased, a higher proportion of the load will be transfered to
the thin film. When the strain energy developed in the thin film reaches its fracture energy,
opening-mode fractures (OMFs) will initiate at the center of the thin film. The OMFs break
the film that has an original length of 2 into two fragments. The newly formed fragmen-
tations will subsequently go through the same three stress stages and more OMFs will fill in
between the previously formed OMFs. However, this fracture infilling will reach a saturate
point once the ratio of cracking spacing to coating thickness approaches to a critical value,
upon which the stress between two adjacent cracks changes from tensile to compressive
(Bai and Pollard, 2000). In this stage, OMF initiation, infilling, and saturation will be
analyzed.

He et al. 5
The next section (on Phase I) develops general formulations and provides explicit solutions for
the elastic field in the coating/substrate system, which is verified by finite element analysis. The
section on Phase II conducts an elastoplastic analysis of the coating/substrate system. A fracture
analysis and further experimental validation is conducted in Phase III, where crack initi-
ation, infilling, and saturation in the coating is discussed. The final section discusses our
conclusions.
Phase I – elastic analysis and numerical validation

Derivation and formulas
When the applied load is relatively small, the coating, interlayer and substrate are all in elastic stage.
The corresponding elastic solutions are provided in this section
Elastic field in the thin film (a < r b).

The equilibrium equation in the z-direction without body force can be written as
frz
fz,z þ frz,r þ ¼0 ð1Þ
r
where fz , and frz are the normal and shearing stress in the thin film, respectively. The superscript
‘‘f ’’ denotes the coating or thin film.
By following the same principle as used in the planar problem (Chen et al., 2015; Yin and Prieto-
Muoz, 2013; Yin et al., 2008), we assume that all of the points on the same cylindrical face parallel to
the loading axis will remain on the same face, namely
u fr,z ¼ 0 ð2Þ
Note that u fr is the radial displacement. A plane strain formulation is developed by using the fol-
lowing constitutive law
Ef h i
fz ¼ ð1 f f
Þu z,z þ f f
u r,r ð3Þ
ð1 þ f Þð1 2 f Þ
Ef
frz ¼ uf ð4Þ
2ð1 þ f Þ z,r
u fz is the displacement along z direction. By substituting equations (3) and (4) into equation (1) and
invoking equation (2), one can get

2 1 f f 1
u þ u f þ u fz,rr ¼ 0 ð5Þ
1 2 f z,zz r z,r
Equation (5) is a decoupled partial differential equation of u fz . By applying separation of vari-

ables, the general solution can be written as
u fz ðr, zÞ ¼ f ðrÞ gðzÞ þ "uz1 z ð6Þ

The unknown constant "uz1 is introduced as a uniform strain caused by the uniaxial tension in the
substrate. By substituting this general solution into equation (5) produces

2 1 f 00 1
g f þ f 0 g þ f 00 g ¼ 0 ð7Þ
1 2 f r
f, f 0 , and f 00 are functions of r, whereas g, g0 and g00 are functions of z, these two independent
equations can be linked by an independent constant d21 as
00
f þ 1r f 0 2ð1 f Þ g 00
¼ ¼ d21 ð8Þ
f 1 2 f g
From the above equation, the general solution of u fz can be written as
u fz ðr, zÞ ¼ ½c1 sinhðe1 zÞ þ c2 coshðe1 zÞ½c3 J0 ðd1 rÞ þ c4 Y0 ðd1 rÞ þ "uz1 z ð9Þ
where
e1 ¼ d1 ð10Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 2 f
¼ ð11Þ
2ð1 f Þ
and ci(i ¼ 1, . . ., 4) are coefficients to be determined.

By considering the symmetry of the displacement along the z direction, one additional boundary
condition can be written as
u fz jz¼0 ¼ 0 ð12Þ
the free surface of the thin film reads
frz jr¼b ¼ 0 ð13Þ
By substituting equations (12) and (13) into equation (9). The displacement along z direction is
reduced to
u fz ðr, zÞ ¼ C1 sinhðe1 zÞ½J0 ðd1 rÞY1 ðd1 bÞ J1 ðd1 bÞY0 ðd1 rÞ þ "uz1 z ð14Þ
Substituting equation (14) into the constitutive equation and neglecting the normal stress along
radial direction, the stress field in the coating layer can be written as
h i
fz ðr, zÞ ¼ f E f C1 e1 coshðe1 zÞ½J0 ðd1 rÞY1 ðd1 bÞ J1 ðd1 bÞY0 ðd1 rÞ þ "uz1 ð15Þ
frz ðr, zÞ ¼ Gf C1 d1 sinhðe1 zÞ½J1 ðd1 rÞY1 ðd1 bÞ J1 ðd1 bÞY1 ðd1 rÞ ð16Þ
f f
where f ¼ ð1þ1 f
f Þð12 f Þ and G ¼
E
.
2 ð1þ f Þ

He et al. 7
Elastic field in the shear interlayer (R r a).

Referring to our previous study (Chen et al., 2015), the introduced shear interlayer with a thickness
of (a R) is mainly applied to transfer the tensile load in the substrate to the coating layer. We also
assume the shear stress and corresponding shear strain are uniformly distributed along the radial
direction. The constitutive equation of the shear interlayer in the elastic stage reads
0rz ðzÞ ¼ G01 rz

0
ð zÞ ð17Þ
where 0rz ðzÞ and rz

0
ðzÞ are the shear stress and strain in the interlayer and G01 is the elastic shear
modulus of the interlayer.
Considering the continuity of shear stress between the substrate and the interlayer, we define
A1 ¼ J0 ðd1 aÞY1 ðd1 bÞ J1 ðd1 bÞY0 ðd1 aÞ ð18Þ
and
the shear stress and shear strain in the interlayer can be written as
0rz ðzÞ ¼ G f C1 d1 A2 sinhðe1 zÞ ð20Þ

f
0 G
rz ðzÞ ¼ C1 d1 A2 sinhðe1 zÞ ð21Þ
G01
Based on the assumption that there is no slip between the substrate and the coating, and that the
shear strain is uniform along the radial direction in the interlayer, the interface continuity can be
expressed as
0
u sz ðzÞ ¼ u fz ðr ¼ a, zÞ ða RÞrz ðzÞ ð22Þ
where u sz ðzÞ is displacement along the z-direction in substrate. By substituting equations (14) and
(21) into equation (22), the displacement of the substrate can be expressed as
u sz ðzÞ ¼ C1 K1 sinhðe1 zÞ þ "uz1 z ð23Þ
G
where K1 ¼ A1 þ G0f ða RÞd1 A2 .
1
Elastic field in the substrate (0 r < R).

As mentioned before, the tensile load is approximately uniform along the radial direction in the
substrate layer, and the substrate can be simplified as a one dimensional problem. The equilibrium
equation can be expressed as
d s
R2 ðzÞ þ 2 R 0rz ðzÞ ¼ 0 ð24Þ
dz z
where sz ðzÞ is the normal stress in the substrate.

The constitutive equation of the substrate under a plane strain assumption in the elastic stage
reads
sz ðzÞ ¼ E1s "sz ðzÞ ð25Þ
where Es1 is the stiffness of the substrate, and "sz ðzÞ is the tensile strain in the substrate. Under the
assumption of small displacements, it can be written as
du sz ðzÞ
"sz ðzÞ ¼ ð26Þ
dz
Combining equations (26), (25) and (20), the equilibrium equation becomes
R s s
E u ðzÞ Gf C1 d1 A2 sinhðe1 zÞ ¼ 0 ð27Þ
2 1 z,zz
by substituting equation (23) into equation (27) yields

s
Es1 E1 2
A1 þ ða RÞd 1 A2 ¼ 0 ð28Þ
Gf G01 Rd1 2
By solving equation (28), d1 can be theoretically determined, and then the tensile stress in the
substrate can be obtained by equations (25) and (23) as
h i
sz ðzÞ ¼ Es1 C1 e1 K1 coshðe1 zÞ þ "uz1 ð29Þ
The boundary conditions are given as
sz ðzÞjz¼ ¼ 01 ð30Þ
Z b
2 r fz ðr, zÞjz¼ dr ¼ 0 ð31Þ
a
where 01 is the applied tensile load. At the surface of the OMF, the stress is supposed to be 0.
However, this condition can not be satisfied exactly due to the assumption in equation (2), and as a
result a relaxed boundary conditions is used in equation (31).
From equations (30) and (31), the coefficients C1 and ex can be finally determined as
01 A3
"uz1 ¼ ð32Þ
Es1 A3 K1
"uz1
C1 ¼ ð33Þ
e1 A3 coshðe1 Þ
2
Rb
where A3 ¼ b2 a 2 a r½J0 ðd1 rÞY1 ðd1 bÞ J1 ðd1 bÞY0 ðd1 rÞdr

He et al. 9
Elastic solutions and numerical verification

It’s worthy to note that three basic assumptions have been applied in deriving the present
theoretical model described in ‘Derivation and formulas’. First, as the brittle coating is very
thin compared to the substrate, it was observed from experimental tests that during the process
of crack infilling, the variation of the tensile stress in the substrate is negligible, thus the sub-
strate is modeled as a one-dimensional (1-D) elastoplastic media. Secondly, the stress in the
substrate subjected to uniaxial tensile loading is transferred to the coating layer through a very
thin interlayer under the shear lag-model scheme. Thirdly, there is no slip or debonding among the
coating, interlayer and substrate, which was indicated by the experimental observation when
the saturated OMFs were developed (Chen et al., 2015). In order to validate those assumptions,
the elastic solutions obtained in the subsection on derivation and formulas will first be verified by
comparing them with finite element solutions since experimental measurement is hard to conduct
at the elastic stage.
The material and geometric constants of the coating/substrate system used in this finite element
analysis (FEA) are provided in Table 1, where Es1 , Es2 and sY are determined from Figure (2). The
material properties of the alumina coating layer, which depend on the generation conditions, were
determined by considering the current anodization procedures and comparisons with existing studies
(Accuratus, 2013; Bradhurst and Leach, 1966; Gong et al., 2009; Jen et al., 2011; Wan et al., 2008).
For the interlayer, which is assumed to be a thin layer of transition zone between aluminum and
alumina, currently there is no straightforward way to measure it since it is sandwiched between the
coating layer and the substrate. In reality, the mechanical properties of interlayer might be continu-
ally varied along the thickness direction. However, in an effort to establish a theoretical model, it is
assumed to be uniform and homogeneous. Our previous study(Chen et al., 2015), assumes that the
shear modulus of the interlayer is the average of the coating and substrate.
The commercial software package ABAQUS 6.11-3 is employed to conduct the numerical simu-
lation. In the FE model, four-node bilinear axisymmetric quadrilateral elements (CAX4) are used to
model the coating layer and interlayer; two-node linear axisymmetric membrane elements (MAX1)
are used to simulate the 1-D substrate. The coating layer, interlayer and the substrate are glued
together to simulate the perfectly bonded interface among the three individual layers assumed in the
theoretical model. In order to better characterize the element optimization and capture the local
stress field more accurately, gradient meshing has been used throughout the FE model and refined
elements are applied at the interface and free surfaces of the film/substrate system as shown in
Table 1. Material and geometric properties of the alumina film/Aluminum substrate

system.
Modulus (GPa) Strength (MPa) Geometry

f
E 200.0 80 mm
Gf 80.0 a 2.2 mm
Es1 64.5 sY 300 b 2.22 mm
Es2 0.99
G10 56.8 Y 173.2 R a ba
100 ¼ 2:1998 mm
G20 4.06

Figure 3. FE model of the OMF film/substrate system: (a) FEA model of mesh; and (b) the deformed shape.
Figure 3(a). In total 12,402 elements were used in the coating and interlayer while 318 elements were
used in the substrate. A constraint is applied to tie the coating layer, interlayer and substrate
together. Therefore, no slipping or debonding is allowed. The deformed shape of the system is
shown in Figure 3(b).
The distributions of the axial displacement across the radial direction at the two ends of coating
layer as shown in Figure 4(a) are symmetric and reflect the real crack shape as indicated in Figure 1.
Figure 4(b) shows the axial displacement distributions in the interlayer and substrate. Recalling the
interface continuity condition shown in equation (22), since the coating thickness is very small, the
displacement of the coating layer at its inner surface is almost overlapped with that of the substrate,
which agrees well with the result from FEA.
The distributions of stress in the substrate and interlayer are provided in Figure 5. It shows
that the normal and shear stress from both the present model and FEA gradually increase from
the midpoint between the cracks toward the free end. The normal stress in the substrate only
varies in a very small range and can be approximately treated as being uniformly distributed.
For the shear stress in the interlayer, a high stress concentration is captured by the present
model at the free end, while free stress is predicted by the FEA which satisfies the free traction
condition. Nevertheless, at a very small distance from the free end, good agreement between the
present solution and the FE simulation is observed, which demonstrates the accuracy of the
present elastic solutions.

He et al. 11
Figure 4. Displacement distributions of the film/substrate system: (a) distributions for the axial displacement of the
coating along the radial direction at the crack ends; (b) distributions for the axial displacement of the coating and
substrate along the z-direction.
Figure 5. Stress distributions of the film/substrate system along the z-direction: (a) normal stress distribution in the
substrate; (b) shear stress distribution in the interlayer.
Phase II – elastoplastic behavior and solutions

From the elastic solutions presented in the subsection Elastic solutions and numerical verification,
both the tensile stress in the substrate and the shear stress in the interlayer increase from the center to
the end. If the shear stress at the end of interlayer reaches its yielding stress, while the tensile stress in
substrate is still lower than its corresponding yielding stress, yielding will initiate at the two ends
of the interlayer and propagate towards the center. When the tensile load increases to a specific
value, the tensile stress at the ends of the substrate will also reach the yielding stress. As shown in

Figure 6. Schematic diagram of the plastic region: (a) elastoplastic interlayer; (b) plastic substrate and elastoplastic
interlayer; (c) plastic substrate and interlayer.
Figure 5(a), the normal stress in the substrate varies in a very small range and can be approximated
as uniform along the whole cross-section. For simplicity, we assume that the substrate will yield once
the loading stress reaches its yielding stress (Chen et al., 2015). As the applied tensile load increases
further, the yielding in the interlayer will propagate all the way to the center. Therefore, all of the
interlayer and substrate reach the plastic stage.
In the following subsection we provide the formulation when the substrate is in its elastic stage,
while part of interlayer is in the plastic stage. The next subsection (Plastic substrate and elastoplastic
interlayer) will present the analysis when a majority of the interlayer is in the plastic stage and all of
the substrate has yielded. Finally, the solution for the case when both the interlayer and substrate are
fully in the plastic stage will be conducted in the subsection Plastic substrate and plastic interlayer.
Elastic substrate and elastoplastic interlayer

Assuming that under certain applied load ( 0 ¼ 02 ), the developed yielding zone in the shear inter-
layer will extend from distance z1 to (with symmetry) and the substrate will still remain in the
elastic stage. In Figure 6(a), the gray area represents the plastic region in the interlayer. In this case,
the film/substrate system needs to be divided into two continuous parts.
(1) Elastic zone (0 z z1).

In this zone, both the shear interlayer and substrate are in the elastic stage. Thus all the solutions in
the subsection Derivation and formulas are applicable to this elastic zone. Considering the slightly
changed boundary conditions, all the elastic solutions for the thin film/substrate system are rewritten
as in the following expressions
h i
frz ðr, zÞ ¼ G f C2 d2 sinhðe2 zÞ½J1 ðd2 rÞY1 ðd2 bÞ J1 ðd2 bÞY1 ðd2 rÞ ð36Þ

He et al. 13
h i
sz ðzÞ ¼ E s1 C2 e2 K2 coshðe2 zÞ þ "uz2 ð38Þ
Similarly
Gf
K2 ¼ A4 þ ða RÞd2 A5 ð39Þ
G01
where C2 is a constant to be determined, d2 ¼ d1, e2 ¼ e1 and it still hods that e22 ¼ 2 d 22 .
(2) Elastic-plastic zone (z1 z ).

As the thin film is still in the elastic stage, the general expression of the displacement in the thin film
from equation (9) is still applicable. By applying the boundary condition shown in equation (13), the
displacement of the thin film can be reduced to
u fz ðr, zÞ ¼ ½C3 sinhðe3 zÞ þ C4 coshðe3 zÞ½J0 ðd3 rÞY1 ðd3 bÞ J1 ðd3 bÞY0 ðd3 rÞ þ "uz3 z ð42Þ
where C3 and C4 are constants to be determined. Considering the constitutive equations in equations
(3) and (4), the normal and shear stresses in the thin film can be simplified as
h i
fz ðr, zÞ ¼ f E f e3 ½C3 coshðe3 zÞ þ C4 sinhðe3 zÞ½J0 ðd3 rÞY1 ðd3 bÞ J1 ðd3 bÞY0 ðd3 rÞ þ "uz3 ð43Þ
frz ðr, zÞ ¼ Gf d3 ½C3 sinhðe3 zÞ þ C4 coshðe3 zÞ½J1 ðd3 rÞY1 ðd3 bÞ J1 ðd3 bÞY1 ðd3 rÞ ð44Þ
As the shear interlayer enters into the plastic stage in this zone, the shear stress–strain curve for
the interlayer becomes

0rz ðzÞ ¼ G02 rz
0
ðzÞ þ G01 G02 Y ð45Þ
where G02 is the linear hardening rate of the shear interlayer under uniform shear; Y ¼ GY0 is the yield
1
shear strain and Y is yield stress under uniform shear.
By following the similar procedures as in ‘Derivations and formulas’, the displacement and
normal stress of the substrate can be determined by

1 1
u sz ðzÞ ¼ K3 ½C3 sinhðe3 zÞ þ C4 coshðe3 zÞ þ "uz3 z þ ða RÞ 0 0 Y ð46Þ
G2 G1
sz ðzÞ ¼ Es1 e3 K3 ½C3 coshðe3 zÞ þ C4 sinhðe3 zÞ þ Es1 "uz3 ð47Þ
where
Gf
K3 ¼ A6 þ d3 ða RÞA7 ð48Þ
G02

Substituting equations (46) and (47) into the equilibrium equation equation (24) and considering the
shear stress continuity between the film and interlayer yields
s
Es1 E1 2
A6 þ 0 ða RÞd3 A7 ¼ 0 ð51Þ
Gf G2 Rd3 2
from which the coefficent d3 can be theoretically obtained.

The boundary conditions in this case read
Z b
2 r fz ðzÞjz¼ dr ¼ 0 ð52Þ
a

0rz ðzÞ z¼z1 ¼ Y ð53Þ
The continuity equations read

0rz ðzÞ z¼z1 ¼ 0rz ðzÞ z¼zþ1 ð55Þ

u sz ðzÞ z¼z1 ¼ u sz ðzÞ z¼zþ1 ð56Þ

sz ðzÞ z¼z1 ¼ sz ðzÞ z¼zþ1 ð57Þ
From the boundary and continuity equations listed above, the five constants C2, C3, C4, and
"uzi ði ¼ 2, 3Þ can be determined
02 A8
"uz3 ¼ ð58Þ
Es1 A8 K3
2
Rb
where A8 ¼ b2 a 2 a r½J0 ðd3 rÞY1 ðd3 bÞ J1 ðd3 bÞY0 ðd3 rÞdr
Y 1
C2 ¼ ð59Þ
G f A5 d2 sinhðe2 z1 Þ
0
1 2 coshðe3 z1 Þ Y sinhðe3 Þ
C3 ¼ ð60Þ
coshðe3 e3 z1 Þ Es1 e3 ðK2 A8 Þ Gf d3 A7

He et al. 15
Y
C4 ¼ C3 tanhðe3 z1 Þ ð61Þ
Gf d3 A7 coshðe3 z1 Þ

aR 1 1 Y K3 K2
"uz2 ¼ "uz3 þ Y þ ð62Þ
z1 G02 G01 Gf z1 d3 A7 d2 z1 A5
The yielding initiation positon z1 can be determined by substituting equations (58) to (62) into
equations (57), which is simplified as
C2 e2 K2 coshðe2 z1 Þ þ "uz2 ¼ e3 K3 ½C3 coshðe3 z1 Þ þ C4 sinhðe3 z1 Þ þ "uz3 ð63Þ
Plastic substrate and elastoplastic interlayer

Referring to the assumption that the substrate will yield once the loading stress reaches its yielding
stress, when the applied load ( 0 ¼ 03 ) is larger than the yielding strength of substrate, the whole
substrate will get into plastic stage and the developed yielding zone in the shear interlayer will extend
from distance z2 to (with symmetry), which is shown in Figure 6(b). In this case, the film/substrate
system needs to be divided into the following two continuous parts
(1) Elastic zone (0 z z2).

In this zone, the displacement and stress in the thin film will have similar formulations as presented
in the elastic zone of Phase II i.e.
h i
As the shear interlayer is in the elastic stage in this zone, the shear stress-strain curve for the
interlayer is still determined by equation (17). The substrate is assumed to be a linear hardening
material with a uniaxial stress-strain that can be expressed as

sz ðzÞ ¼ Es2 "sz ðzÞ þ Es1 Es2 "Y ð67Þ
where Es2 is the linear hardening rate of the substrate under uniaxial tension, "Y ¼ EYs is the yield
1
strain, and sY is the yield stress under uniaxial tension.
By following the similar procedures in the subsection ‘Elastic field’ in the shear interlayer
(R r a) and the subsection ‘Elastic field’ in the substrate (0 r < R), the displacement and
normal stress of the substrate can be determined by

Es
sz ðzÞ ¼ Es2 C5 e4 K4 coshðe4 zÞ þ Es2 "uz4 þ 1 2s Y ð69Þ
E1

where
Gf
G01
d4 can be similarly determined by the following formation

s
Es2 E2 2
A 9 þ ða RÞd 4 A10 ¼ 0 ð73Þ
Gf G01 Rd4 2
(2) Plastic zone (z2 z )

In this zone, the displacement and stress in the thin film will have similar formulations as presented
in the plastic zone of section ‘Elastic field’ in the thin film (a < r b), i.e.
u fz ðr, zÞ ¼ ½C6 sinhðe5 zÞ þ C7 coshðe5 zÞ½J0 ðd5 rÞY1 ðd5 bÞ J1 ðd5 bÞY0 ðd5 rÞ þ "uz5 z ð74Þ
h i
fz ðr, zÞ ¼ f E f e5 ½C6 coshðe5 zÞ þ C7 sinhðe5 zÞ½J0 ðd5 rÞY1 ðd5 bÞ J1 ðd5 bÞY0 ðd5 rÞ þ "uz5 ð75Þ
frz ðr, zÞ ¼ Gf d5 ½C6 sinhðe5 zÞ þ C7 coshðe5 zÞ½J1 ðd5 rÞY1 ðd5 bÞ J1 ðd5 bÞY1 ðd5 rÞ ð76Þ
As both the interlayer and the substrate are in the plastic stage, the constitutive equations for
them are provided by equations (45) and (67), respectively. By following similar procedures in the
subsections on the shear interlayer and the substrate, the displacement and normal stress of the
substrate can be determined by

1 1
usz ðzÞ ¼ K5 ½C6 sinhðe5 zÞ þ C7 coshðe5 zÞ þ "uz5 z þ ða RÞ 0 0 Y ð77Þ
G2 G1

s s s u Es2
z ðzÞ ¼ E2 e5 K5 ½C6 coshðe5 zÞ þ C7 sinhðe5 zÞ þ E2 "z5 þ 1 s Y ð78Þ
E1
f
where K5 ¼ A11 þ GG 0 d5 ða RÞA12 , A11 ¼ J0(d5a)Y1(d5b) J1(d5b)Y0(d5a) A12 ¼ J1(d5a)Y1(d5b)
2
J1(d5b)Y1(d5a). and d5 can be determined by
s
Es2 E2 2
A 11 þ ða RÞd 5 A12 ¼ 0 ð79Þ
Gf G02 Rd5 2
The boundary conditions in this case read

Zb
2 r fz ðzÞjz¼ dr ¼ 0 ð80Þ
a

He et al. 17

0rz ðzÞ z¼z2 ¼ Y ð81Þ
The continuity equations read

0rz ðzÞ z¼z2 ¼ 0rz ðzÞ z¼zþ2 ð83Þ

u sz ðzÞ z¼z2 ¼ u sz ðzÞ z¼zþ2 ð84Þ

sz ðzÞ z¼z2 ¼ sz ðzÞ z¼zþ2 ð85Þ
From the boundary and continuity equations listed above, the five constants C5, C6, C7 and
"uzi ði ¼ 4, 5Þ can be determined as

03 A13 1 1 Y A13
"uz5 ¼ þ ð86Þ
Es2 ðA13 K5 Þ Es2 Es1 K5 A13
where
Z b
2
A13 ¼ r½J0 ðd5 rÞY1 ðd5 bÞ J1 ðd5 bÞY0 ðd5 rÞdr ð87Þ
b2 a2 a
Y
C5 ¼ ð88Þ
Gf A10 d4 sinhðe4 z2 Þ
u
1 "z5 coshðe5 z2 Þ Y sinhðe5 Þ
C6 ¼ þ ð89Þ
coshðe5 e5 z2 Þ A13 e5 Gf d5 A12
Y
C7 ¼ C6 tanhðe5 z2 Þ ð90Þ
Gf d5 A12 coshðe5 z2 Þ

aR 1 1 Y K5 K4
"uz4 ¼ "uz5 þ Y þ ð91Þ
z2 G02 G01 Gf z2 d5 A12 d4 A10
z2 can be determined by substituting equations (86) to (91) into equation (85), which is simplified as
C5 e4 K4 coshðe4 z2 Þ þ "uz4 ¼ e5 K5 ½C6 coshðe5 z2 Þ þ C7 sinhðe5 z2 Þ þ "uz5 ð92Þ
Plastic substrate and plastic interlayer

As shown in Figure 6(c), as the applied load further increases, the interlayer will eventually yield.
In this case, both the interlayer and substrate will get into the plastic stage. Following similar

procedures in the derivation and formulas subsection, by replacing the constitutive equations for the
interlayer and substrate with equations (45) and (67), the displacement and stress field in the film/
substrate system can be determined by the following formulas
h i

1 1
u sz ðzÞ ¼ C8 K6 sinhðe6 zÞ þ "uz6 z þ ða RÞ Y ð96Þ
G02 G01
h i Es

sz ðzÞ ¼ Es2 C8 e6 K6 coshðe6 zÞ þ "uz6 þ 1 2s Y ð97Þ
E1
where
Gf
G02
C8 and "uz6 are determined by

0
A16 4 1 1
"uz6 ¼ Y ð101Þ
A16 K6 Es2 Es2 Es1
2
Rb
where A16 ¼ b2 a2 a r½J0 ðd6 rÞY1 ðd6 bÞ J1 ðd6 bÞY0 ðd6 rÞdr
"uz6
C8 ¼ ð102Þ
e6 A16 coshðe6 Þ
where
Z b
2
A16 ¼ r½J0 ðd6 rÞY1 ðd6 bÞ J1 ðd6 bÞY0 ðd6 rÞdr ð103Þ
b2 a2 a
By solving the following equation, d6 can be determined theoretically

s
Es2 E2 2
A14 þ 0 ða RÞd6 A15 ¼ 0 ð104Þ
Gf G2 Rd6 2

He et al. 19
Elastoplastic solutions
The distributions of axial displacement in the thin film across its radial direction when s0 ¼ 0.8 sY
(elastic substrate/elastoplastic interlayer) and s0 ¼ 1.2 sY (plastic substrate/elastoplastic interlayer) are
shown in Figure 7. Both are compared with the elastic solution for the axial displacement distribution
corresponding to a much smaller load in the elastic stage (s0 ¼ 0.2 sY). The symmetric displacement
distribution determined for both the elastic and plastic stages reveals that the presented elastoplastic
model is able to capture the crack evolution as the applied load increases. It is observed from
Figure 7(b) that the displacement in the thin film significantly increases from 5 mm to 350 mm as the
applied stress slightly increases from 0.2 sY to 1.2 sY, which indicates that significant plastic deform-
ation has been developed in the interlayer/substrate. Such elastoplastic behavior in the coating/
substrate system and crack evolution cannot be revealed by any elastic model in the current literature.
Phase III – fracture analysis and experimental validation

Fracture analysis
Consider the section with two opening-mode cracks at each end in Figure 1. When the external
tensile loading reaches a critical value, a steady-state channeling straight crack will initiate at the
middle of the section. After cracking, the section is broken into two halves, and the displacement
field in each half can also be solved by replacing with /2 in the new local coordinate system.
Therefore, the crack opening displacement is solved. To recover this OMF, the normal stress along
the central line just before the fracture initiates needs to be applied along the cracking surface.
Therefore, the energy release rate (ERR) can be obtained as the work done to close the crack
opening displacement (Beuth, 1992) is shown in equation (105)
Z b
1
G¼ fz ðr, 0Þu fz r, dr ð105Þ
ðb2 a2 Þ a 2
Figure 7. Displacement distributions of the film/substrate system at: (a) elastoplastic stage I (elastic substrate with
elastoplastic interlayer); and (b) elastoplastic stage II (plastic substrate and elastoplastic interlayer).

Figure 8. Variance of ERR with respect to: (a) fracture spacing, and (b) coating thickness.
The ERR in the coating layer will increase as the applied load increases. Once the ERR surpasses
the fracture toughness, a new fracture will nucleate. Therefore, the following criterion is used to
predict if new fractures will form or not
G cr ð106Þ
where cr is the fracture toughness of the thin film.

Based on the stress and displacement presented in the sections on analysis and behavior, the ERR
can be obtained through equation (105). Figure 8 (a) shows that when the coating thickness is
constant (h1 ¼ b a=20 mm), the ERR increases rapidly as the cracking spacing increases from 20
to 140 mm. Beyond this point, although the cracking spacing increases, the ERR does not change any
more. It indicates that, under the applied load (s0 ¼ 1.2 sY), potential crack spacing of the given thin
film will be in the range of 20 mm to 140 mm. The relationship between the ERR and the coating
thickness at a constant cracking spacing of = 80
m and under the same applied load
(s0 ¼ 1.2 sY) is shown in Figure 8 (b). It reveals that the ERR increases linearly with an increase
of the coating thickness until the ERR reaches a convex point, while after this point, the ERR
rapidly decreases with the increasing coating thickness. Note that fracture saturation will be reached
at this convex point (/h1 . 3.75). Therefore, no more new crack will be developed when
/h1 . 3.75 under this applied load.
Recalling equation (106), when the applied load reaches a critical value, a new OMF will
nucleate. Figure 9 illustrates the applied load that is required to form a new crack in the thin
film with respect to the cracking spacing/coating thickness ratio /h1. The fracture toughness of
the thin film applied in this study is 3.5 MPa m1/2 (Accuratus, 2013; Jen et al., 2011). Figure 9
clearly shows that the required external applied to form new cracks load increases as the crack
spacing decreases. When /(b a) approaches to 1.2, the external load turns to infinity.
Therefore, no new fractures will infill and the fracture saturation will be reached. For this
specified film thickness of h1 ¼ 20 mm, the saturated crack spacing can be approximated to
cr ¼ 1.2 h1 ¼ 24 mm.

He et al. 21
Figure 9. Fracture spacing development with respect to the external tensile loading.
Experimental validation
The elastic solution was verified in the subsection on elastic solutions and numerical verification
through a finite element analysis. However, elastoplastic opening-model fracture in coating/thin-film
problem is quite complex. Therefore, experimental tests will be conducted in this section to verify the
elastoplastic solutions. In this experiment, four aluminum alloy (A6061) rods with a brittle alumi-
num oxide (Al2O3) coating (also called alumina), grown through the process of anodization, were
subjected to a uniform tensile load. During the tensile test, a microscope was used to monitor crack
development, through which the critical applied load that initiated cracking in the thin film can be
obtained. After reaching fracture saturation, the sample was released from the tensile machine and
the crack spacing was observed under a microscopic. The test results agree well with the theoretical
result.
Sample preparation
Four aluminum rods with a radius of 7/16 inch and length of one foot were prepared. The prepared
aluminum samples were polished with sandpaper first, then acetone and distilled water were used to
clean the sample surfaces. The samples were covered with acid resistant tape such that three inches
from each end of the sample are covered as the grip for tension test. According to our previous study
(Chen et al., 2015), under the same anodization procedures, the coating thickness is proportional to
the anodizing time, which was validated by the measurement on the alumina coating on a flat plate.
Since in this study we employed the same anodization procedures, the coating thickness is thus
determined by the anodizing time. Therefore, by keeping these four samples in the anodizing bath
for four different anodization duration times (t1 ¼ 15 min, t2 ¼ 30 min, t3¼ 60 min, t4 ¼ 120 min),
different coating thickness were obtained.

Figure 10. Testing sample preparations: (a) anodization devices; (b) prepared samples.
By following the anodization procedures of Hyun-Chae et al. (2009) and Chen et al. (2015), a
sulfuric electrolyte solution composed of 20% sulfuric acid and 80% distilled water by weight was
placed inside the test beaker followed by the cylindrical lead cathode. A Teflon support was used
to fix the aluminum anode at the center of the lead cathode as shown as Figure 10(a), whereby a
uniform alumina layer was created on the aluminum rod surface. The power supply was adjusted
to 15 volts. The electrolyte was kept at a temperature of about 22 C. From the experimental
observations, the alumina growth rate was found to be independent of the sample’s radius (Chen
et al., 2015). Therefore, by varying the anodization time, the coating thickness could be
roughly controlled. All the samples were labeled according to the anodization time shown in
Figure 10(b).
Uniaxial tensile testing

An Instron 5984 34k universal testing machine was used to apply a uniform tensile load on each
sample. The load was applied at a strain rate of 0.01 inch/min, and the ultimate applied strain for all
samples were set to be 2%. During the tensile test, a Keyence VHX-5000 Digital Microscopic was set
up close to the sample and focused near the center of the sample (see Figure 11). The Keyence VHX-
5000 Digital Microscopic is a versatile high-fidelity optical microscope designed for high depth of
field measurements. The VHX can recognize the focus information automatically when the field of
view is moved and then create a depth composition image quickly. The purpose of this microscope is
to continuously monitor the fracture development in the coating, through which valuable fracture
information, such as critical applied load corresponding to crack initiation, crack pattern and crack
spacing, can be obtained. Once OMFs were observed in the coating, the corresponding applied load
and strain were recorded. After the predefined strain was reached, the applied load was released and
the sample was removed from the testing equipment with all geometric properties (such as coating
thickness, apparent cracks, discontinuities) recorded.
Test results and analysis

The microstructures as well as the fracture configurations of the alumina coatings after tensile tests
were examined by the Keyence VHX-5000 Digital Microscope. Figure 12 (a) shows typical micro-
scope images revealing the microstructures of the brittle alumina coating. The image of a fractured

He et al. 23
Figure 11. Uniaxial tensile configuration.
Figure 12. SEM images on fractured samples: (a) typical porous nature of the brittle alumina coating; and (b) parallel
fractures perpendicularly spaced to the applied load.
specimen with parallel cracks periodically spaced and perpendicular to the direction of loading is
shown in Figure 12(b). Note that, if the applied load is within its elastic stage of the substrate, the
OMF will close once the externally applied load is removed. Considering this, a relatively large
plastic strain was applied to make the cracking separation visible after unloading.
The saturated crack spacings observed under the digital microscopic for samples with different
coating thicknesses are shown in Figure 13 and the corresponding crack spacings are summarized in
Table 2. Results from both Table 2 and Figure 13 show that the crack spacing proportionally
decreases as the thin film thickness decreases. In addition, the applied loads which cause fracture
initiation are listed in Table 2 which show that the crack initiation loads increase with the decrease of

Figure 13. Cracking spacing at different coating thickness.
Table 2. Crack spacing at different coating thicknesses.
Crack initiation
Coating thickness (mm) Crack spacing (mm) Ave. (mm) load (MPa)
9 26.9 26.3 25.5 27.1 20.7 23.4 25.1 309

18 22.0 22.0 29.6 28.8 30.2 29.8 27.1 304
36 36.6 30.9 31.4 37.8 34.5 34.3 34.3 274
71 51.7 52.9 39.9 42.7 49.1 39.1 45.9 218
coating thickness. The difference of crack initiation loads between the case with coating thickness
9 mm and 18 mm is very small as the substrate has yielded. Therefore, a small increase of load would
cause a large increase of displacement and break the thinner coating easily.
Based on the in-situ monitoring from the microscope, it was found that the crack infilling in the
coating is very quick. Once a crack was initiated, two to four OMFs were simultaneously observed in

He et al. 25
Figure 14. Initiated cracks in coating.
Figure 15. Variance of fracture spacing with respect to thin film thickness.

the microscope as shown in Figure 14. Also, since the crack opening displacement is very small at the
initiation stage, high magnification ( >1500 times) was applied in the microscope in order to better
observe the cracking evolution. Thus, only a very small area is monitored under the microscope (the
actual length shown in Figure 14 is only about 0.2 mm). Therefore, once certain OMFs were
observed under the microscope, many cracks might have already developed in the entire coating.
In this respect, the crack initiation loads shown in Table 2 are slightly higher than the actual values.
Nevertheless, it reveals the changing trend of crack initiation load with respect to the coating
thickness.
The comparison of cracking spacing as a function of coating thickness between the experiment
result and theoretical results obtained from the presented model is shown in Figure 15. Both the test
data and the theoretical solution show that the crack spacing proportionally decreases as the thin
film thickness decreases. The theoretical predictions also indicate that there exists a critical coating
thickness, below which the crack spacing quickly increases to infinite and no new cracks will be
developed. Overall, the varying trend of the crack spacing with respect to the coating thickness
revealed by the test results can be effectively captured by theoretical prediction. This further dem-
onstrates that the presented model is not only able to predict the elastoplastic behavior of the
coating layered structure in general, but also disclose more specific details of opening-mode fracture
in the coating, such as micro crack spacing and the critical coating thickness.
Conclusions
In this study, an elastoplastic fracture model has been presented to simulate periodically distributed
micro-cracking in a thin layer coating fully bounded to an elastoplastic aluminum wire/rod.
The stress field in the coating/substrate system was analyzed by studying one section between two
adjacent cracks. The thin layer of alumina coating generated by the anodic oxidation was treated as
an elastic media, while the ductile substrate aluminum wire was characterized by a bi-linear elasto-
plastic hardening model. In order to simulate the stress transfer between the coating layer and
substrate, an elastoplastic interlayer has been employed. As the applied tensile load is increased,
the system undergoes three different stress stages: elastic stage, elastoplastic stage, and opening-
mode fractures. The displacement and stress distributions in the coating/substrate were first solved
for the elastic stages and was verified by a finite element analysis. Elastoplastic solutions were further
provided for the coating/substrate system as large deformation develops with an increase of the
applied load. Based on the elastoplastic solution, energy release rates at different loading stages were
calculated and a fracture analysis was conducted. Simulation results showed that, when the coating
thickness remains constant, an increasing load is required to form a new crack which corresponds to
a decrease in the crack spacing, and fracture saturation is reached when the crack spacing
approaches to the value of 1.2 times the coating thickness.
Experimental characterization was conducted to validate the presented elastoplastic solution
when the substrate or interlayer undergoes large deformation. A versatile high-fidelity optical micro-
scope was used to examine the micro structure of the coating and continuously monitor the fracture
development in the coating layer. This allowed valuable detailed fracture information, such as
critical applied load corresponding to crack initiation, crack pattern and crack spacing, to be
determined. Both the experimental and theoretical results indicate that the crack spacing decreases
linearly as the thin film thickness decreases until the critical film thickness is reached, while below
this critical thickness, no new cracks will develop. Overall, the varying trend of the crack spacing
with respect to the coating thickness revealed by the test results can be effectively captured by the
theoretical prediction, which demonstrates that the presented model is able to accurately capture the

He et al. 27
stress and strain distribution in the coated structure and predict fracture initiation, infilling, and
saturation in the thin coating layer. The present formulation is general and can be applicable to
other types of systems that include ductile substrate with coating.
Acknowledgements
The authors thank Dr. Liming Li for his help on the tests conducted in this study.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publica-
tion of this article: The authors acknowledge financial support from the Department of Homeland Security
under grant number DHS CU09-1155 and the National Science Foundation (NSF) under grant numbers
CMMI 0954717 and CMMI 1301288. The results and opinions presented herein are those of the authors
and do not necessarily reflect those of the sponsoring agency.
References
Accuratus (2013) Aluminum oxide \textbar Al2O3 material properties. Available at: www.accuratus.com/
alumox.html.
Agrawal DC and Raj R (1989) Measurement of the ultimate shear strength of a metal-ceramic interface. Acta
Metallurgica 37(4): 1265–1270.
Bai T and Pollard DD (2000) Fracture spacing in layered rocks: A new explanation based on the stress tran-
sition. Journal of Structural Geology 22(1): 43–57.
Bele E, Bouwhuis B, Codd C, et al. (2011) Structural ceramic coatings in composite microtruss cellular mater-
ials. Acta Materialia 59(15): 6145–6154.
Beuth J (1992) Cracking of thin bonded films in residual tension. International Journal of Solids and Structures
29(13): 1657–1675.
Beuth J and Klingbeil N (1996) Cracking of thin films bonded to elastic-plastic substrates. Journal of the
Mechanics and Physics of Solids 44(9): 1411–1428.
Bradhurst D and Leach JL (1966) The mechanical properties of thin anodic films on aluminum. Journal of The
Electrochemical Society 113(12): 1245–1249.
Budiansky B, Hutchinson JW and Evans AG (1986) Matrix fracture in fiber-reinforced ceramics. Journal of the
Mechanics and Physics of Solids 34(2): 167–189.
Chai H (2011) Channel cracking in inelastic film/substrate systems. International Journal of Solids and
Structures 48(7): 1092–1100.
Chen F, He X, Prieto-Muñoz P, et al. (2015) Opening-mode fractures of a brittle coating bonded to an elasto-
plastic substrate. International Journal of Plasticity 67: 171–191.
Chen F and Qiao P (2012) On the intralaminar and interlaminar stress analysis of adhesive joints in plated
beams. International journal of adhesion and adhesive 36(22): 44–55.
Chen Y, Reed R and Marquis E (2012) As-coated thermal barrier coating: structure and chemistry. Scripta
Materialia 67(9): 779–782.
Gong MF, Qiao SR, Zhang CY, et al. (2009) Strength properties of thin alumina coatings on 4043 aluminum
alloy. Materials Science Forum 610: 668–673.
Hyun-Chae N, Taek-Jin S, Seok-Heon Y, et al. (2009) Formation of unidirectional nanoporous structures in
thickly anodized aluminum oxide layer. Transactions of Nonferrous Metals Society of China 19(4):
1013–1017.
Jen SH, Bertrand JA and George SM (2011) Critical tensile and compressive strains for cracking of Al2O3 films
grown by atomic layer deposition. Journal of Applied Physics 109(8): 084305.
Mahesh S, Hanan J, Üstündag E, et al. (2004) Shear-lag model for a single fiber metal matrix composite with an
elasto-plastic matrix and a slipping interface. International journal of solids and structures 41(15): 4197–4218.

Nakamura T and Kamath SM (1992) Three-dimensional effects in thin film fracture mechanics. Mechanics of
Materials 13(1): 67–77.
Ochiai S, Tomida T, Nakamura T, et al. (2005) Fracture behavior of brittle coating layer on metal substrate.
Materials Science Forum 475–479: 929–932.
Partridge PG, Younes CM, Nicholson J, et al. (1997) Microstructure and interfaces in diamond coated steel
wires. Journal of Materials Science 32(16): 4201–4208.
Shiao MH and Shieu FS (2000) Interfacial mechanical properties and fracture morphology of a TiN-coated
steel wire upon tensile loading. Thin Solid Films 358(1–2): 159–165.
Shieu FS and Shiao MH (1997) Measurement of the interfacial mechanical properties of a thin ceramic coating
on ductile substrates. Thin Solid Films 306(1): 124–129.
Sreevatsa S (2009) Carbon nanotube electronic structures as anti-corrosion coatings Ph.D. Dissertation, New
Jersey Institute of Technology and Rutgers, the State university of New Jersey.
Tseng YW, Hung FY and Lui TS (2015) Microstructure, tensile and electrical properties of gold-coated silver
bonding wire. Microelectronics Reliability 55(3–4): 608–612.
Wan D, Zhou Y and Bao Y (2008) Evaluation of the elastic modulus and strength of unsymmetrical Al2O3
coating on Ti3SiC2 substrate by a modified relative methodology. Materials Science and Engineering: A
474(1): 64–70.
Wang D, Peng H, Liu J, et al. (1995) Wear behaviour and microstructural changes of SiCw-Al composite under
unlubricated sliding friction. Wear 184(2): 187–192.
Wang X-d, Wang H, Shen H-x, et al. (2014) Tensile properties and fracture reliability of a glass-coated Co-
based amorphous microwire. International Journal of Minerals, Metallurgy, and Materials 21(6): 583–588.
Xia ZC and Hutchinson JW (2000) Crack patterns in thin films. Journal of the Mechanics and Physics of Solids
48(6): 1107–1131.
Yin H (2010a) Fracture saturation and critical thickness in layered materials. International Journal of Solids and
Structures 47(7): 1007–1015.
Yin HM (2010b) Opening-Mode cracking in asphalt pavements: Crack initiation and saturation. Road
Materials and Pavement Design 11(2): 435–457.
Yin HM and Prieto-Muñoz PA (2013) Stress transfer through fully bonded interface of layered materials.
Mechanics of Materials 62: 69–79.
Yin H, Paulino G and Buttlar W (2008) An explicit elastic solution for a brittle film with periodic cracks.
International Journal of Fracture 153(1): 39–52.

IMP - He2016 PDF

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

IMP - He2016 PDF

Transféré par

Droits d'auteur :

Formats disponibles

Article

International Journal of Damage

Xin He, Fangliang Chen and Huiming Yin

Department of Civil Engineering and Engineering Mechanics, Columbia University, USA

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

Phase I – elastic analysis and numerical validation

Elastic field in the thin film (a < r b).

Equation (5) is a decoupled partial diﬀerential equation of u fz . By applying separation of vari-

u fz ðr, zÞ ¼ f ðrÞ gðzÞ þ "uz1 z ð6Þ

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

From the above equation, the general solution of u fz can be written as

and ci(i ¼ 1, . . ., 4) are coeﬃcients to be determined.

the free surface of the thin ﬁlm reads

frz jr¼b ¼ 0 ð13Þ

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

Elastic field in the shear interlayer (R r a).

0rz ðzÞ ¼ G01 rz

where 0rz ðzÞ and rz

A1 ¼ J0 ðd1 aÞY1 ðd1 bÞ J1 ðd1 bÞY0 ðd1 aÞ ð18Þ

0rz ðzÞ ¼ G f C1 d1 A2 sinhðe1 zÞ ð20Þ

u sz ðzÞ ¼ C1 K1 sinhðe1 zÞ þ "uz1 z ð23Þ

Elastic field in the substrate (0 r < R).

where sz ðzÞ is the normal stress in the substrate.

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

sz ðzÞ ¼ E1s "sz ðzÞ ð25Þ

by substituting equation (23) into equation (27) yields

The boundary conditions are given as

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

Elastic solutions and numerical verification

Table 1. Material and geometric properties of the alumina film/Aluminum substrate

Modulus (GPa) Strength (MPa) Geometry

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

Phase II – elastoplastic behavior and solutions

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

Elastic substrate and elastoplastic interlayer

(1) Elastic zone (0 z z1).

u sz ðzÞ ¼ C2 K2 sinhðe2 zÞ þ "uz2 z ð37Þ

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

A4 ¼ J0 ðd2 aÞY1 ðd2 bÞ J1 ðd2 bÞY0 ðd2 aÞ ð40Þ

A5 ¼ J1 ðd2 aÞY1 ðd2 bÞ J1 ðd2 bÞY1 ðd2 aÞ ð41Þ

where C2 is a constant to be determined, d2 ¼ d1, e2 ¼ e1 and it still hods that e22 ¼ 2 d 22 .

(2) Elastic-plastic zone (z1 z ).

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

A6 ¼ J0 ðd3 aÞY1 ðd3 bÞ J1 ðd3 bÞY0 ðd3 aÞ ð49Þ

A7 ¼ J1 ðd3 aÞY1 ðd3 bÞ J1 ðd3 bÞY1 ðd3 aÞ ð50Þ

from which the coeﬃcent d3 can be theoretically obtained.

The continuity equations read

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

C2 e2 K2 coshðe2 z1 Þ þ "uz2 ¼ e3 K3 ½C3 coshðe3 z1 Þ þ C4 sinhðe3 z1 Þ þ "uz3 ð63Þ

Plastic substrate and elastoplastic interlayer

(1) Elastic zone (0 z z2).

u sz ðzÞ ¼ C5 K4 sinhðe4 zÞ þ "uz4 z ð68Þ

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

A9 ¼ J0 ðd4 aÞY1 ðd4 bÞ J1 ðd4 bÞY0 ðd4 aÞ ð71Þ

A10 ¼ J1 ðd4 aÞY1 ðd4 bÞ J1 ðd4 bÞY1 ðd4 aÞ ð72Þ

d4 can be similarly determined by the following formation

(2) Plastic zone (z2 z )

The boundary conditions in this case read

Downloaded from ijd.sagepub.com at CORNELL UNIV on September 17, 2016

0rz ðzÞ ¼ G01 rz

where 0rz ðzÞ and rz

where C2 is a constant to be determined, d2 ¼ d1, e2 ¼ e1 and it still hods that e22 ¼ 2 d 22 .

C2 e2 K2 coshðe2 z1 Þ þ "uz2 ¼ e3 K3 ½C3 coshðe3 z1 Þ þ C4 sinhðe3 z1 Þ þ "uz3 ð63Þ

C5 e4 K4 coshðe4 z2 Þ þ "uz4 ¼ e5 K5 ½C6 coshðe5 z2 Þ þ C7 sinhðe5 z2 Þ þ "uz5 ð92Þ