ARTICLE IN PRESS

International Journal of Adhesion & Adhesives 30 (2010) 408–417

Contents lists available at ScienceDirect

International Journal of Adhesion & Adhesives

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

Evaluation of mechanical interlock effect on adhesion strength of

polymer–metal interfaces using micro-patterned surface topography
Won-Seock Kim a, Il-Han Yun b, Jung-Ju Lee c,n, Hee-Tae Jung b
a
Satellite Structure Department, Korea Aerospace Research Institute, 45 Eoeun-dong, Yuseong-gu, Daejeon 305-333, South Korea
b
Department of Chemical and Biomolecular Engineering, KAIST, Science Town, Daejeon 305-701, South Korea
c
Department of Mechanical Engineering, KAIST, Science Town, Daejeon 305-701, South Korea

a r t i c l e in f o a b s t r a c t

Article history: This study concerns with the explanation of the wide range of adhesion strengths observed depending
Accepted 5 February 2010 on the nature of substrate surface topography by linking macroscopic adhesion strength to microscopic
Available online 27 May 2010 energy-expenditure mechanisms during fracture. The dominant factors to which the adhesion strength
Keywords: of polymer–metal interfaces is attributed are investigated theoretically and experimentally. In an
Surface roughness/morphology attempt to elucidate the effect of mechanical interlock on adhesion strength, micro-patterns were
Fracture Mechanics fabricated on metal surfaces as a designed surface topography. It was found that the molecular
Contact angles dissipation of the polymer in the vicinity of the interface is the major cause of the practical energy of
Adhesion by mechanical interlocking separation. Furthermore, it is shown that loading mode controls the mechanical interlock effect, which
Micro-pattern
is attributed to the fact that the stress distribution at the interface controls the deformation and failure
characteristics of the polymer resin near the interface. Therefore, mechanical interlock promoted by
adsorption provokes energy dissipation processes during fracture, which practically constitute the
adhesion strength of a polymer–metal interface. The contribution of mechanical interlock to adhesion
strength is systematically assessed by varying micro-pattern dimensions. The influence of the work of
adhesion, cohesion and other dissipation energy on adhesion strength is examined by measuring each
contribution to the total work of fracture.
& 2010 Elsevier Ltd. All rights reserved.

1. Introduction Among a number of mechanisms that account for the adhesion

Adhesive bonding is increasingly widely applied as a practical
method for the joining of dissimilar materials in the automotive,
aircraft, microelectronic and medical devices industries, replacing
conventional joining techniques such as welding, bolting and
riveting [1–5]. Various types of polymers are used as adhesives,
and they are selected according to the substrates to be bonded
and the operating conditions of the adhesive joint [6]. Metals,
such as steel, aluminum, copper and gold are used in almost every
industry, and it is often necessary to join these metals with other
materials. Therefore, the polymer–metal interface becomes an
important design consideration in many structures that use
adhesive bonding. One of the predominant concerns for engineers
in using adhesive bonding is to optimize the adhesion strength.
Thus, it is essential to understand the adhesion mechanisms
involved and to investigate the influence of each mechanism on
adhesion strength since this understanding will allow the
formation of strong and reliable adhesive joints.
n
Corresponding author. Tel.: + 82 42 350 3033; fax: + 82 42 350 3210.
E-mail address: leejungju@kaist.ac.kr (J.-J. Lee).
phenomenon, it is well-known that adsorption and mechanical
interlock are the dominant mechanisms that contribute to the
adhesion strength of polymer–metal interfaces [1–4]. Adhesive
forces induced by intermolecular attractions exist between a
polymer and a metal provided molecular-scale intimate contact is
achieved, and the penetration of the polymer into the craters and
pores in rough metal substrates enhances the bond strength
through mechanical interlock. Chemical bonding can also be
introduced as a dominant adhesion strength enhancement factor
by embedding a coupling agent at the interface and forming a
metal/coupling agent/polymer system [7,8].
In previous work, researchers have tried to correlate the
measured joint strength with the theoretical work of adhesion
[9–17]. It is well-known that the thermodynamic work of
adhesion is far less than the practical bond strength; however,
the work of adhesion is regarded as playing a vital role in
adhesion so that small changes in its value can cause large
changes in the practical adhesion strength. Many studies devoted
to measuring adhesion suggest that the greater values of the
practical work of separation compared to the work of adhesion
can be explained by taking the plastic or viscoelastic dissipation of
the adhesive layer or substrates into consideration [9–17]. The
roughness of the interface has been presumed to be the major

0143-7496/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijadhadh.2010.05.004
 ARTICLE IN PRESS
W.-S. Kim et al. / International Journal of Adhesion & Adhesives 30 (2010) 408–417 409

factor activating the energy dissipation processes in polymeric

adhesives during crack propagation. Various surface topographies
and characteristic surface parameters have been proposed to
improve adhesion strength and to find an influential parameter of
surface topography on adhesion strength [18–25]. However,
quantitative evaluation of mechanical interlock effect on adhesion
strength has not been reported since previous analysis was
performed based on randomly distributed surface roughness. The
topography of a surface is a kind of random data, which requires a
statistical analysis for the determination of its characteristic
parameters. Nevertheless, a surface can be completely character-
ized with several parameters if a surface topography is designed
and fabricated into a patterned structure. Designed surface
topography facilitates the investigation of how the morphological
change of the surface may influence adhesion strength.
The purpose of the present work is to systematically
investigate the influence of adhesion mechanisms on the adhesion
strength of a polymer–metal bond by incorporating periodic
surface patterns on the metal substrate. In order to link
microscopic adhesion phenomena to the macroscopic adhesion
strength of a bonded joint, the fracture mechanics concept has
been utilized measuring adhesion strength as the required energy
for fracture. The interfacial fracture toughness of a polymer–metal
bond was measured using the bi-material end-notched flexure
(ENF) test for mode-II and the single-leg bend (SLB) test for
mixed-mode. The cause of adhesion strength, rather than
adhesion strength enhancement, was targeted in this study by
elucidating the energy-expending process during fracture. Quan-
titative assessment of fracture energetics was achieved by
controlling the area fraction of adhesive and cohesive failure via
systematic variation of surface topography.
2. Experimental
2.1. Materials and sample preparation
In order to investigate the fracture energetics of the polymer–
metal interface, composite/steel bonded joints were fabricated.
Steel (AISI 1045) as a conventional metal and carbon fiber-
reinforced polymer (CFRP) as an advanced structural material
were chosen as the substrates to be bonded. The steel plates were
cut into 35  5  1 mm3 samples and were finely polished using a
rotating disk-type polisher with diamond suspensions. Diamond
suspensions with particle sizes of 9, 6, 3, 1 and 0.25 mm were used
sequentially to attain smooth and flat surfaces. Afterwards, the
polished steel surfaces were coated with a commercial photo-
resist (SU-8 from MicroChem, USA) and exposed to UV through a
transparency-printed mask. The UV-exposed photoresist film was
developed to have a series of line posts, whose width varied from
10 to 90 mm using different masks. After the conventional
photolithography, the steel surfaces not covered by the photo-
resist films were chemically etched in an etchant called Nital
solution (nitric acid and ethanol mixed solution, v/v ¼1/9) to
fabricate microline-patterns on the steel surface (Fig. 1(a)). After
the surface patterning process, CFRP prepregs (USN-150 from SK
Chemical, Korea) were laminated on the steel plate ([09/steel])
and bonded through the standard curing process of the CFRP
prepreg. In other words, composite/metal joint specimens were
fabricated using a co-cure bonding method.
The micro-scale line-patterns, which were fabricated on the
steel plate prior to bonding, were created perpendicular to the
direction of crack propagation. The effect of metal surface
morphology on adhesion strength can be assessed by varying
the width (w1), depth (R) and spacing (w2) of microline-patterns,
since the line-patterned surface morphology can be completely
Fig. 1. (a) Schematic diagram of a microline-patterned steel specimen. (b)
Schematic diagram of microline-patterned metal–polymer interfaces; R: max-
imum etching depth, which corresponds to conventional definition of surface
roughness, w1: excavated line width, w2: preserved line width. Congruent pattern
design was utilized to investigate the roughness effect only. Other surface
parameters (w1/R, w2/R, Areal/Anom) could be maintained as a constant value.
characterized by these three independent parameters (Fig. 1(b)).
The pattern parameters were varied by using different masks
in the process of photolithography and by controlling etching
time. All of the pattern parameters (R, w1, w2) were in the range of
10–90 mm, which covers the minimum pattern sizes that can be
controlled using the isotropic wet-etching process. First, the three
surface design parameters were chosen to vary in a congruent
manner as shown in Fig. 1(b) in order to investigate the roughness
effect only. Using a set of surfaces with congruent geometry, the
aspect ratios (w1/R, w2/R) and roughness factor (the ratio of real
contact area to nominal area, Areal/Anom), which have been
reported to have strong influence on adhesion strength as surface
describing parameters [23–25], can be maintained as constants.
Control joint specimens were also produced using steel plates
with polished surfaces and no patterns.
2.2. Contact angle measurement
The work of adhesion of a polymer–steel interface was
determined by contact angle measurements. In general, contact
angles are formed by the sessile drop method or annealing thin
films. In this study, a kind of annealing method was utilized to
obtain epoxy particles on a steel substrate. The epoxy resin which
composes the CFRP prepreg is in solid state at room temperature
and becomes a liquid as the temperature rises to the curing point
(125 1C), and then irreversible solidification is attained by the
chemical reaction between the epoxide and the premixed cross-
linking agent. The bisphenol-A based epoxy resin was extracted
from the CFRP prepreg and shaped in the form of pellets with a
diameter of about 2 mm, weighing around 5 mg. The pellets were
then placed on the polished steel surface and annealed using the
same temperature cycle as that of the curing process used for
bonding. During the annealing process, the epoxy pellets on the
steel plate wet the substrate, forming small contact angles. Since
there was no significant change of the shape of the wetted
particles after solidification, contact angles were measured at
room temperature. The contact angles of the epoxy particles were
 ARTICLE IN PRESS
410 W.-S. Kim et al. / International Journal of Adhesion & Adhesives 30 (2010) 408–417
observed using a contact angle analyzer (Phoenix 300 from S.E.O.
Corporation, Korea).
2.3. Fracture testing
The interfacial fracture toughness Gc, which is the released
energy per unit interfacial area at which fracture occurs, was
measured as the adhesion strength of the composite/steel inter-
face. Mechanical testing methods for interfacial toughness
measurement include the double-cantilever beam (DCB) test for
mode-I, end-notched flexure (ENF) test for mode-II and single-leg
bend (SLB) test for mixed-mode. Among various test methods, the
aforementioned beam-type specimen geometry has definite
advantages. Beam-type specimens are simple to fabricate, amen-
able to miniaturization and easy to test. Most previous works
[9–15,22,26–29] performed pure mode-I tests using a peel test or
asymmetric DCB configuration to measure adhesion energy while
structural adhesive joints are designed to transfer loads by shear
stress distribution along the bond line. In this study, a bi-material
ENF test, which provides a pure mode-II loading condition [30],
and a bi-material SLB test, which provides various mixed-mode
loading conditions [31], were employed. Fig. 2(a) and (b) show
schematic diagrams of the tests. The dimensions of the specimens
Fig. 2. Adhesion fracture tests schematics and photographs: (a) ENF test schematics
(dimensions in mm); (b) SLB test schematics (dimensions in mm); (c) miniaturized
composite/metal ENF test and (d) miniaturized composite/metal SLB test.
were miniaturized due to the technical difficulty of micro-
patterning on a large area. Micro-patterns were located at the
uncracked ligament of 5 mm (from the initial crack tip to the
loading point). The thickness ratio of ENF specimens was held
constant at (composite thickness)/(steel thickness)¼1.25 to
locate the neutral axis of the bonded beam at the interface. The
thickness ratio of SLB specimens was held constant at (composite
thickness)/(steel thickness)¼ 0.5 to induce a mode-I dominant
loading condition. An interfacial pre-crack was introduced by
smearing mold release agent (ER-650 from Nabakem, Korea) on
the steel surface prior to bonding. A miniaturized three-point
bend fixture was used as shown in Fig. 2(c) and (d), and the
testing was performed under displacement control at a ramp rate
of 10 mm/s using a computer-controlled stepper motor.
Another factor that should be stated in adopting a test method
for adhesion measurement is whether the measured adhesion
strength is intrinsic to the interface. Engineering tests such as the
peel and blister tests are often sensitive to the global inelastic
deformation of the polymer occurring away from the interface.
Thus, the measured interfacial toughness is influenced by the
thickness of a specimen and by the bulk properties of the
polymer. Therefore, inelastic deformations of the polymer should
be confined at the crack tip and in the vicinity of the interface to
propose the measured strength as a property of the interface. The
localization of inelastic deformation along the interface of the
specimen will be demonstrated in the results section.
3. Numerical modeling
A two-dimensional finite element (FE) analysis was conducted
in order to calculate the critical energy release rate, Gc using the
experimentally measured crack onset load, Pc. The numerical ENF
and SLB models consist of an upper CFRP layer and a lower steel
layer, and the interface was modeled as a straight line. Micro-
patterned interfacial roughness was not considered in the FE
model. For this reason, in this study, the measured interfacial
toughness is a macroscopic interfacial property which is a
function of micro- and nano-scale interfacial roughness. Second-
order eight-node quadrilateral elements with full integration
were used in the analysis. The interface between the two
adherends was modeled by sharing identical nodes, and the
interfacial crack was modeled by detaching nodes along some
portion of the interface. A frictionless contact option was applied
onto the crack surfaces. Near the interfacial crack tip, the FE mesh
was refined and quarter point singular elements were used to
describe the singular stress field efficiently. The models were
implemented in ABAQUS by performing a plane-strain analysis,
and the energy release rate was calculated through the domain
integration-based J-integral technique. Residual stresses caused
from high-temperature bonding of materials having disparate
thermal expansion coefficients were considered. Interfacial strain
Table 1
Material properties used in analyses.
Material properties of carbon fiber reinforced epoxy (USN 150)
Fiber direction modulus E1 (GPa) 131.6
Transverse direction modulus E2 ¼E3 (GPa) 8.2
In-plane shear modulus G12 (GPa) 4.5
Poisson’s ratio n12 0.28
Fiber direction thermal expansion coefficient a1 (10  6/1C)  0.9
Transverse direction thermal expansion coefficient a2 ¼ a3 (10  6/1C) 27
Material properties of steel (AISI 1045)
Young’s modulus E (GPa) 200
Poisson’s ratio n 0.3
Thermal expansion coefficient a (10  6/1C) 12
 ARTICLE IN PRESS
W.-S. Kim et al. / International Journal of Adhesion & Adhesives 30 (2010) 408–417
energy release rates in the layered specimens have non-zero
initial values before mechanical loading is applied due to the
residual thermal stresses, which indicates that evaluation of
interfacial toughness ignoring residual thermal stresses will be an
underestimation compared to the true interfacial toughness. The
material properties used for calculations are listed in Table 1.
4. Results and discussion
4.1. Surface and interface observation
Surfaces and cross-sections of interfaces were observed using
both optical and scanning electron microscopes (SEM). The SEM
photographs, presented in Fig. 3, show the micro-scale line-
patterned steel surface. Optical microscope cross-sectional
Fig. 3. Scanning electron micrographs of a microline-patterned steel surface. Line
widths were maintained as a constant ratio of w1/w2 ¼ 1.
Fig. 4. Optical micrographs of the polished cross-section of a steel/composite bonded interface. Patterned surface parameters were w1 ¼w2 ¼30 mm and R¼ 15 mm.
411
images, presented in Fig. 4, show the steel/CFRP bonded
interface. The cross-section of the periodic grooves became
semicircular due to the wet-etching process applied to the steel,
an isotropic material. Due to the isotropic wet-etching process,
surface parameters maintain a relationship of constant aspect
ratio: R/w1 ¼0.5. The width of etched line w1 was varied from 10
to 90 mm with its maximum depth, R, ranging from 5 to 45 mm.
The epoxy resin, which was extracted from the CFRP prepreg
during its curing process, completely filled the micro-grooves on
the steel surface. Therefore, mechanical keying, or interlocking, of
the polymer into the steel surface cavities was completely
attained at the microline-patterned steel/CFRP interface.
The submicroscopic morphology of the etched metal surface
also has effect on adhesion strength though it was not under
control in this study [32]. Exactly speaking, micro-scale surface
roughness was designed, but the overspread nano-scale surface
roughness was naturally formed during the micro-patterning
process. Consequently, the naturally formed nano-morphology
was uniformly applied to all etched steel surfaces.
4.2. Work of adhesion
From the thermodynamic point of view, the work of adhesion
of a bi-material bond is the amount of energy required to create
free surfaces by separating the interface. The work of adhesion is
calculated by the Dupre equation:
Wa ¼ g1 þ g2 g12 ð1Þ
where Wa is the reversible energy of adhesion, g1 and g2 are the
surface free energies of the contacting solids denoted by 1 and 2,
and g12 is the interface free energy between the two solids.
Therefore, the thermodynamic work of adhesion, Wa, is a constant
for a specific pair of solids as an intrinsic property of the interface,
and it does not depend on loading mode. In the thermal
equilibrium condition, the interface free energy gi between a
solid particle and a solid substrate is given by the Young equation:
gi ¼ gS gP cos y ð2Þ
where gS and gP are the surface free energies of the substrate and
particle, respectively, and y is the contact angle (Fig. 5(a)).
According to the Young–Dupre equation, the adhesion energy
between the particle and substrate is calculated as follows:
Wa ¼ gS þ gP gi ¼ gP ð1 þ cos yÞ ð3Þ
Therefore, using the surface free energy of the epoxy and the
measured contact angle between the epoxy particle and the steel
substrate, the adhesion energy of the epoxy/steel bond can be
calculated. It is necessary to remember that the surface free
energy, g, refers to the energy required to create new surfaces of
the material in a vacuum. The simplification of zero spreading
pressure underlies using Eq. (3) for calculating the work of
adhesion in the presence of ambient air.
 ARTICLE IN PRESS
412 W.-S. Kim et al. / International Journal of Adhesion & Adhesives 30 (2010) 408–417

Fig. 5. (a) Schematic diagram of contact angle measurement. An epoxy particle on

the steel substrate: (b) before annealing and (c) contact angle formed after
annealing.

Fig. 6. A typical load–displacement curve from the bi-material ENF test.

As the sample was heated near the curing temperature of the

epoxy resin (125 1C), the epoxy pellets on the steel plate wetted
the substrate by forming small contact angles (Fig. 5(b) and (c)).
The temperature cycle for annealing was identical to the
manufacturer’s recommended curing cycle of the prepreg, which
was also used as a bonding process in this study. The measured
contact angle of the epoxy particles on the steel substrate is
24.21 72.31 (the average of ten individual determinations). The
room-temperature surface energy of typical amine-cured epoxy
resin is reported to be 46.2 mJ/m2 ðgS ¼ gD P
S þ gS ¼ 41:2mJ=m þ
2

5:0 mJ=m2 Þ [1,4]. Using the surface free energy value of a typical
epoxy resin (46.2 mJ/m2) and the measured value of the contact
angle (24.21), the work of adhesion between the epoxy resin and
steel was calculated to be 88.3 mJ/m2 via Eq. (3).
The adhesion energy, Wa, between two materials influences
how a liquid (adhesive) will wet a solid (adherend). Thus, the
adhesion energy plays a significant role in the spreading of hot-
melt adhesives. Penetration of a liquid into the irregularities of a
substrate surface is a prerequisite for the functioning of the
mechanical interlock mechanism. Therefore, adsorption promotes
mechanical interlock. The work of adhesion is an important
parameter as a promotion factor for other bond strength
enhancement mechanisms though its value is far below that of
the practical work of separation. In the epoxy/steel bond, the
measured value of the adhesion energy, Wa ¼88.3 mJ/m2, was
enough to form a mechanical interlock-activating interface by
completely filling the epoxy resin into the micro-scale roughened
steel surfaces. The strengthening mechanism of the polymer/
metal bond caused by mechanical interlock and the further role of
the work of adhesion during the fracture process will be discussed
in the following sections.
4.3. Interfacial fracture toughness and failure locus
A typical load–displacement curve obtained from an ENF test is
presented in Fig. 6. Although the epoxy resin in the fabricated
epoxy/steel bond is brittle, there is a gradual transition from
linear to non-linear behavior due to the stable crack propagation,
and the point at which crack propagation began is hard to
determine. The critical point at which the compliance has
increased by 5% was interpreted as the onset of crack
propagation and used for the calculation of the critical energy
release rate, Gc [33,34]. The load at the critical point Pc always fell
between the proportional limit, PNL, and maximum load, Pmax
(Fig. 6), which validates the employment of the 95% secant
Fig. 7. Interfacial fracture toughness, GIIc for different pattern depths, R. The
patterns with different pattern widths and depth maintained congruent shapes.
line-based critical load, Pc, as the crack onset load. The critical
load, Pc was transformed into the critical energy release rate Gc,
the energy dissipated during fracture per unit fracture area by
using the numerical method explained in Section 3. The deter-
mined values of interfacial fracture toughness from ENF test, GIIc
were plotted against the pattern depth, R, in Fig. 7. The width and
depth of the patterns were varied in a congruent manner by
maintaining the width ratio as w1:w2 ¼1:1 to isolate the influence
of the surface roughness (Fig. 1(b)). The lack of correlation in Fig. 7
suggests that the surface roughness, R, does not control adhesion
strength. In order to investigate the locus of failure, fracture
surfaces were examined using a scanning electron microscope.
As seen in Fig. 8, the excavated region on the steel surfaces shows
pure cohesive failure and the preserved region on the steel
surfaces shows pure interfacial failure regardless of the variation
of pattern dimensions. As long as the width ratio is maintained as
w1:w2 ¼ 1:1, the total areas of cohesive and adhesive failure
regions are the same. This observation reveals that the total area
fraction of the cohesive and adhesive failure regions is directly
related to adhesion strength while the surface roughness, R, is an
indirect parameter that contributes to failure mode transition
from adhesive to cohesive or vice versa.
In order to increase adhesion strength by increasing the area
fraction of cohesive failure, the width ratio of line patterns was
 ARTICLE IN PRESS
W.-S. Kim et al. / International Journal of Adhesion & Adhesives 30 (2010) 408–417
Fig. 8. Fracture surfaces of ENF specimens on the patterned steel side. Preserved steel surface region failed adhesively showing the clear steel adherend, and excavated
steel surface region failed cohesively showing the epoxy resin.
Fig. 9. Typical load–displacement curves of bi-material ENF specimens with
different pattern width ratios (w1/w2).
varied as w1:w2 ¼ 1:1; 2:1; 4:1; 6:1; 8:1. Typical load–displace-
ment curves obtained from ENF tests with different pattern width
ratios are presented in Fig. 9. The failure load of adhesive joints
increases according to the increment of the ratio of the excavated
width, w1, to the preserved width, w2, of the line-patterns. As seen
in Fig. 10(a) and (b), the excavated region on the steel surfaces
always shows pure cohesive failure and the preserved region on
the steel surfaces always shows pure interfacial failure regardless
of the variation of the pattern width ratio (w1/w2). The cause of
the high interfacial toughness of specimens having a large pattern
width ratio (w1/w2) can be explained by the increase of the area
fraction of cohesive failure. Cohesive failure, caused by the
molecular decohesion of the polymer resin, expends larger
energy for crack propagation than interfacial failure, caused by
the debonding of secondary bonds across the interface. In the case
of mirror-polished surfaces without any pattern, the load–
displacement curve of an ENF test shows that the initial slope
starts to decrease earlier than those of patterned specimens
(Fig. 9). This indicates that the crack started to propagate at the
moment when the initial slope of the load–displacement curve
started to decrease, though the maximum load is much higher
than the crack onset load. Since there is no mechanical interlock
effect on mirror-polished steel surfaces, the epoxy resin attached
to the metal substrate can easily slide, and the crack can
propagate along the interface without mechanical obstacles.
However, the fracture surface shows a partial cohesive failure
413
mode (Fig. 10(c)). The observed partial cohesive failure mode
implies that the secondary bonds across the interface are strong
enough to cause crack deflection into the polymer resin, resulting
in large energy dissipation during the gradual crack sliding.
In the case of mode-I dominant loading, both mirror-polished
and micro-patterned steel substrates showed poor adhesion
strength compared to the mode-II test result. The failure loads
reached only one-tenth of those under the mode-II loading
condition (Fig. 11). The SLB test of a mirror-polished steel
substrate resulted in almost pure interfacial failure (Fig. 12(a))
while the micro-patterned steel substrates resulted in a partial
cohesive failure (Fig. 12(b)). In a magnified view of the interfacial
failure region on a patterned substrate (Fig. 12(c)), many scraps of
epoxy resin could be found on the fracture surface, which
indicates frictional resistance during pull out of the interlocked
epoxy resins. Therefore, although adhesion strength can be
enhanced under crack opening-mode loading, a mechanically
interlocked steel/epoxy interface under mode-I dominant loading
does not provide sufficiently reliable bond strength to be used for
load-bearing joints.
Scanning electron micrographs of the CFRP/steel interface after
crack propagation show that crack growth under mode-II loading
causes plastic damages in the polymer resin for both adhesive
and cohesive failure (Fig. 13(a)). A long slender damage zone
containing numerous micro-cracks developed during the sliding-
mode crack propagation (Fig. 13(b)). The failure in the interfacial
failure mode region, which showed the clean metal adherend
on the fracture surfaces, was not mere adhesive failure at the
interface; it created a damage layer of about 5 mm in the polymer
neighboring the interface. The damage layer in the cohesive
failure region is also thin and parallel with the interface. Micro-
cracks are oriented approximately 451 from the main crack
propagation direction, which means that the micro-cracks are
oriented perpendicular to the maximum normal stress as is usual
for brittle materials (Fig. 13(b)). As loading increases, these micro-
cracks coalesce with each other, forming the main crack growth.
The high interfacial toughness in mode-II results from energy
dissipation in this damage zone. The experimental observation of
enhanced interfacial toughness by the increment of the area
fraction of cohesive failure region suggests that viscoelastic and
plastic dissipation in the polymer material is more effectively
induced during the crack growth within the polymer resin than
during the crack growth along the interface, though the interfacial
debonding also induces a damage zone in the polymer. The steel/
composite interface after mode-I dominant loading fracture
shows that the epoxy resin that had penetrated into the cavities
of the patterned steel substrates was pulled out without breaking
the internal bonds of the polymer, and plastic damage zones were
not found in the polymer (Fig. 13(c)). The epoxy resin near the
 ARTICLE IN PRESS
414 W.-S. Kim et al. / International Journal of Adhesion & Adhesives 30 (2010) 408–417
Fig. 10. Fracture surface on steel side of ENF specimens: (a) pattern width ratio of
w1:w2 ¼ 2:1; (b) pattern width ratio of w1:w2 ¼8:1 and (c) mirror-polished surface
without patterning.
epoxy/steel interface does not behave in an elastic manner when
loaded to the point of fracture. Chain pullout, microvoid
formation and crazing are common failure mechanisms in
polymers, and the energy dissipation during these damage
formations constitutes the practical fracture energy. The
variation of stress distribution at the polymer–metal interface
caused from the increased crack opening-mode loading changed
the fracture process at the interface from a more to a less energy-
consuming mode. Therefore, mechanical interlock provides the
Fig. 11. Typical load–displacement curves of bi-material SLB specimens with and
without micro-pattern.
main source of the adhesion strength of polymer–metal interfaces
by provoking energy dissipation processes in the polymer resin
during joint failure. Furthermore, different loading modes near
the interfacial crack tip cause different interactions between the
material property and interface geometry, resulting in different
adhesion strengths.
4.4. Energetics of the fracture process
Many researchers have suggested that the adhesion fracture
energy, or interfacial toughness, Gc, comprises two principal
components; one is the intrinsic fracture energy, G0, and the other
is the dissipated energy for the deformation of materials during
fracture, Gdiss:
Gc ¼ G0 þ Gdiss ð4Þ
The intrinsic fracture energy G0 is the energy required to
propagate a crack in the absence of any energy losses in the
material and is only related to thermodynamic surface energy
terms. The energy dissipated in the deformation of the material
during fracture, Gdiss, is usually much larger than G0 and is
responsible for the dependence of Gc on loading mode, tempera-
ture and crack propagation rate. Multiplicative relationships of
the two effects have been sought on the hypothesis that G0 and
Gdiss are coupled
Gc ¼ G0  fðc, T, cÞ ð5Þ
where j (c, T, c) is a mode mixity, c, temperature, T and loading
rate, c dependent energy loss term. The theoretical derivation of
the relationship between G0 and Gdiss was also proposed by
Andrews and Kinlock [10]. The hypothesis that the value of Gdiss is
connected with G0 so that Gdiss is proportional to G0 is now
generally accepted [35–37].
In the present work, quantitative assessment of the role of the
thermodynamic work of adhesion and cohesion in adhesion
strength was facilitated by the adhesive and cohesive failure
mode control using microline-patterned surface design. Inter-
facial toughness values, GIIc obtained from ENF tests using various
pattern dimensions are plotted in Fig. 14 as a function of the line
width ratio, w1/(w1 + w2) on steel surfaces. Since the pattern width
ratios w1/(w1 +w2) and w2/(w1 +w2) were always identical to the
area fractions of the cohesive and interfacial failure regions, the
measured joint failure energy per unit area, Gc, can be
decomposed into cohesive and adhesive failure energies by an
 ARTICLE IN PRESS
W.-S. Kim et al. / International Journal of Adhesion & Adhesives 30 (2010) 408–417 415

Fig. 13. SEM micrographs of the composite/steel interface after crack propagation:
Fig. 12. Fracture surfaces on steel side of SLB specimens: (a) mirror-polished
(a) crack growth under mode-II loading; (b) magnified view of the damage zone
surface without patterning; (b) pattern width ratio of w1:w2 ¼ 1:1 and
and (c) crack growth under mode-I dominant loading.
(c) magnified view of patterned fracture surface.

equation of the form are related with thermodynamic surface energy terms as:
w1 w2 Ea ¼ Wa  a ð7Þ
Gc ¼ Ec þ Ea
w1 þ w2 w1 þ w2
w1 Ec ¼ Wc  b ð8Þ
¼ ðEc Ea Þ þ Ea ð6Þ
w1 þ w2
where a and b are factors of viscoelastic and plastic energy
where Ec and Ea are the energies required for unit area separation dissipation for the polymer deformation during joint failure, and
in the pure cohesive and adhesive modes, respectively. Ea and Ec Wa and Wc are the intrinsic work of adhesion and cohesion,
 ARTICLE IN PRESS
416 W.-S. Kim et al. / International Journal of Adhesion & Adhesives 30 (2010) 408–417
Fig. 14. Interfacial fracture toughness, GIIc as a function of pattern width ratio.
respectively. If viscoelastic and plastic energy dissipations are
excluded, interfacial failure requires the work of adhesion,
Wa ¼ g1 + g2  g12, and cohesive failure requires the work of
cohesion, Wc ¼ 2g1 (or 2g2). Energy dissipation factors a and b
caused from mechanical interlock effects can be inferred using the
known intrinsic adhesion strengths and the measured adhesion
strength variation according to surface parameter variation
(Fig. 14). By the interpolation and extrapolation of the
interfacial toughness data in Fig. 14, the quantities Ea and Ec can
be evaluated as Ea E210 and Ec E570 J/m2. From the results
obtained in the previous sections (Wa ¼88.3 mJ/m2 and
Wc ¼46.2 mJ/m2), a E2380 and b E12 340. This means that the
intrinsic energies were increased by the 3rd order of magnitude in
adhesive failure mode and by the 4th order of magnitude in
cohesive failure mode. The dissipation factors, a and b, will be
functions of the mechanical properties of the polymer, which can
be modified by altering its formulation. The upper limit of the
adhesion strength of a polymer–metal bond will be
Gc ¼Ec ¼Wc  b in pure cohesive failure. Therefore, in order to
increase the failure energy of a bonded joint, the interaction
between surface morphology and the mechanical properties of
the polymer resin should be optimized to cause increased energy
expenditure during crack propagation.
5. Conclusions
The surface topography effect on the adhesion strength of an
epoxy/steel bond was investigated. Experimental results show
that the major source of strength enhancement caused by metal
surface topography modification in polymer–metal bonded joints
is the transition from interfacial to cohesive failure, which is also
controlled by loading mode. The adhesion strength is increased by
the increased surface roughness of a substrate only if the
increased roughness causes the transition of failure mode from
adhesive to cohesive. The high fracture toughness of microline-
structured interfaces is attributed to the mechanical interlock
mechanism, which provides the energy expending processes
whereby viscoelastic and plastic energy dissipation and corre-
sponding damage zone development in the polymer are required
for crack propagation. Therefore, once the adhesion between a
specific metal with a certain surface morphology and a specific
polymer is achieved by adsorption, the fracture characteristics
will depend on the interaction between the surface topography of
the metal substrate and the mechanical properties of the polymer by
mechanical interlock. Furthermore, since the stress distribution
at the interface controls the deformation and failure character-
istics of the polymer resin near the interface, strong adhesive
joints can be attained not only through consideration of the
surface morphology of the metal substrate and the mechanical
properties of the polymer, but also by joint design optimization to
transfer loads via shear stresses across the bond line.
This study attempts to explain why the adhesion strength of
polymer–metal bonded joints shows great variation according to
the surface topography of the metal substrates and the loading
mode. It is important to emphasize that mechanical interlock is
not just one aspect governing the adhesion phenomenon.
Mechanical interlock promoted by adsorption provokes energy
expenditure during fracture, which practically constitutes the
adhesion strength.
Acknowledgements
This research was supported by Basic Science Research
Program through the National Research Foundation of Korea
(NRF) funded by the Ministry of Education, Science and Technol-
ogy (No. 2009-0083647).
References
[1] Kinloch AJ. Adhesion and adhesives: science and technology. New York:
Chapman and Hall; 1987.
[2] Packham DE. Handbook of adhesion. UK: Longman; 1992.
[3] Pizzi A, Mittal KL, editors. Handbook of adhesive technology. New York:
Marcel Dekker; 1994.
[4] Pocius AV. Adhesion and adhesives technology: an introduction. New York:
Hanser; 1997.
[5] Yacobia BG, Martin S, Davis K, Hudson A, Hubertb M. Adhesive bonding in
microelectronics and photonics. J Appl Phys 2002;91:6227–62.
[6] Wake WC. Theories of adhesion and uses of adhesives: a review. Polymer
1978;19:291–308.
[7] Plueddemann EP. Silane coupling agents, 2nd ed. New York: Plenum Press;
1991.
[8] Mittal KL, editor. Silanes and other coupling agents. Utrecht: VSP; 1992.
[9] Kaelble DH. Peel adhesion: influence of surface energies and adhesive
rheology. J Adhes 1969;1:102–23.
[10] Andrews EH, Kinloch AJ. Mechanics of adhesive failure. I. Proc Roy Soc A
1973;332:385–99.
[11] Andrews EH, Kinloch AJ. Mechanics of adhesive failure. II. Proc Roy Soc A
1973;332:401–14.
[12] Gent AN, Kinloch AJ. Adhesion of viscoelastic materials to rigid substrates. III.
Energy criterion for failure. J Polym Sci, Part A: Polym Chem 1971;9:
659–68.
[13] Mittal KL. The role of the interface in adhesion phenomena. Polym Eng Sci
1977;17:467–73.
[14] Maugis D, Barquins M. Fracture mechanics and the adherence of viscoelastic
bodies. J Phys D: Appl Phys 1978;11:1989–2023.
[15] Carre A, Schultz J. Polymer–aluminum adhesion II. Role of the adhesive and
cohesive properties of the polymer. J Adhes 1984;17:135–56.
[16] Allen KW. Some reflections on contemporary views of theories of adhesion.
Int J Adhes Adhes 1993;13:67–72.
[17] Seshadri M, Saigal S, Jagota A, Bennison SJ. Scaling of fracture energy in
tensile debonding of viscoelastic films. J Appl Phys 2007;101:093504.
[18] Gent AN, Lin CW. Model studies of the effect of surface roughness and
mechanical interlocking on adhesion. J Adhes 1990;32:113–25.
[19] Chen Y, Kalyon DM, Bayramli E. Effects of surface roughness and the chemical
structure of materials of construction on wall slip behavior of linear low
density polyethylene in capillary flow. J Appl Polym Sci 1993;50:1169–77.
[20] Shanahan MR, Morris AP. Adhesion to sintered substrates. J Adhes
1996;59:51–9.
[21] Janarthanan V, Garrett PD, Stein RS, Srinivasarao M. Adhesion enhancement
in immiscible polymer bilayer using oriented macroscopic roughness.
Polymer 1997;38:105–11.
[22] Arola DD, Yang DT, Stoffel KA. Fatigue of the cement/bone interface: surface
texture and wear debris. J Biomed Mater Res, Part B 2001;58:519–24.
[23] Packham DE. Surface energy, surface topography and adhesion. Int J Adhes
Adhes 2003;23:437–48.
[24] Prolongo SG, Rosario G, Urena A. Study of the effect of substrate roughness on
adhesive joints by SEM image analysis. J Adhes Sci Technol 2006;20:
457–70.
[25] Jiang ZX, Huang YD, Liu L, Long J. Effects of roughness on interfacial
performances of silica glass and non-polar polyarylacetylene resin compo-
sites. Appl Surf Sci 2007;253:9357–64.
 ARTICLE IN PRESS
W.-S. Kim et al. / International Journal of Adhesion & Adhesives 30 (2010) 408–417
[26] Gent AN, Petrich RP. Adhesion of viscoelastic materials to rigid substrates.
Proc Roy Soc A 1969;310:433–48.
[27] Kendall K. Peel adhesion of solid films—the surface and bulk effects. J Adhes
1973;5:179–202.
[28] Cao HC, Evans AG. An experimental study of fracture resistance of bimaterial
interfaces. Mech Mater 1989;7:295–305.
[29] Akisanya AR, Fleck NA. Brittle fracture of adhesive joints. Int J Fract
1992;58:93–114.
[30] Kim WS, Lee JJ. Fracture characterization of interfacial cracks with frictional
contact of the crack surfaces to predict failures in adhesive-bonded joints.
Eng. Fract Mech 2009;76:1785–99.
[31] Davidson BD, Sundararaman V. A single leg bending test for interfacial
fracture toughness determination. Int J Fract 1996;78:193–210.
417
[32] Brockmann W, Geiss P, Klingen J, Schroder K. Adhesive bonding: materials,
applications and technology. Darmstadt: Wiley-VCH; 2009. [Chapter 7].
[33] ASTME 1820-05. Standard test method for measurement of fracture
toughness.
[34] ASTMD 5045-99. Standard test methods for plane-strain fracture toughness
and strain energy release rate of plastic materials.
[35] Jokl ML, Vitek V, McMahon CJ. A microscopic theory of brittle fracture in
deformable solids: a relation between ideal work to fracture and plastic
work. Acta Metall 1980;28:1479–88.
[36] Volinsky AA, Moody NR, Gerberich WW. Interfacial toughness measurements
for thin films on substrates. Acta Mater 2002;50:441–66.
[37] Je H, Was GS, Thouless MD. Measurement of the noibium/sapphire interface
toughness via delamination. Int J Fract 2003;119/120:441–8.