Engineering Fracture Mechanics 172 (2017) 139–151

Contents lists available at ScienceDirect

Engineering Fracture Mechanics

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

Effects of aluminum surface treatments on the interfacial

fracture toughness of carbon-fiber aluminum laminates
Afshin Zamani Zakaria a,⇑, Karim Shelesh-nezhad a, Tajbakhsh Navid Chakherlou a, Ali Olad b
a
Department of Mechanical Engineering, University of Tabriz, Tabriz, Iran
b
Department of Applied Chemistry, University of Tabriz, Tabriz, Iran

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

Article history: The influence of aluminum surface treatment on the interfacial delamination performance
Received 22 September 2016 of carbon/epoxy composite and aluminum substrate was studied. Aluminum substrates
Received in revised form 30 November 2016 were treated chemically with acid, alkaline and both individually. AFM images suggest
Accepted 6 January 2017
high roughness for acid treatment and porous oxide layer for alkaline treatment, respec-
Available online 11 January 2017
tively. The plate theory was employed in order to obtain the energy release rate of asym-
metric double cantilever beam (ADCB). The results indicated improvements of interfacial
Keywords:
fracture toughness for acid then alkaline treatment of aluminum substrates. Finite element
Fiber metal laminate
Cohesive zone modelling
study of crack growth proved more reliability of energy release rates obtained by plate the-
Delamination ory analysis rather than compliance calibration method, for ADCB specimen.
Interface fracture Ó 2017 Elsevier Ltd. All rights reserved.
Fiber bridging

1. Introduction

Fiber metal laminates (FMLs) are a class of hybrid laminates formed by stacking light weight metal sheets as aluminum or
titanium and polymer composites. Aramid reinforced aluminum laminates and glass aluminum reinforced epoxy were the
primary standardized FMLs which attracted great attention in research and industry over past decades [1]. Utilization of stiff
carbon fibers in FMLs was suggested by Lin et al. [2] for the first time to develop a more fatigue resistant material. The stiff
carbon fibers may bridge among fatigue cracked aluminum sheets and prevent the crack development (Fiber bridging is not a
dominant factor for FMLs with less stiff glass fibers). Principally, fiber bridging become a crucial task when metal layers are
discreetly cracked in FMLs [3]. Also fiber bridging may occur in two perpendicular directions and deviate the crack path in
FMLs [4]. Lin et al. [2] consumed thin glass mates among aluminum and carbon fiber reinforced polymer (CFRP) layers to
avert galvanic corrosion of laminates which may occur by ion exchanging between carbon and aluminum, in the hygrother-
mal conditions. But this may not solve the corrosion problem as the moisture may move in the composite-metal interface
[5,6]. A major solution to the galvanic corrosion of carbon reinforced aluminum laminates (CARALL) is to cover the aluminum
sheets with an oxide layer [7,8].
In FMLs, two kinds of tests were employed to assess their fabrication reliability in static and fatigue modes. The first type
of test is for evaluation of overall fracture toughness of FMLs, which consist of middle cracked tension (MT) [9–11] and com-
pact tension (CT) [10,12,13] specimens. The second type of test focused on the interfacial and delamination characteristics to
evaluate the reliability of FMLs.
Delamination of composite and metal layers is a concern for FMLs [5,14]. A stable oxide layer may strengthen the inter-
facial bonding of metal and epoxy adhesive [15] and prevent delamination. The interface fracture toughness may also rise by

⇑ Corresponding author.
E-mail address: afshin.zamani89@gmail.com (A. Zamani Zakaria).

http://dx.doi.org/10.1016/j.engfracmech.2017.01.004
0013-7944/Ó 2017 Elsevier Ltd. All rights reserved.
140 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151

Nomenclature

a crack length (mm)

A extensional stiffness tensor (N/mm)
A0 extensional compliance tensor (mm/N)
B coupling stiffness tensor (N)
B0 coupling compliance tensor (N1)
C load line compliance (mm/N)
D bending stiffness tensor (Nmm)
D0 bending compliance tensor (N1mm1)
E aluminum Young’s modulus (MPa)
E11 longitudinal modulus of CFRP (MPa)
E22 transverse modulus of CFRP (MPa)
G strain energy release rate (N/mm)
G12 in-plane shear modulus of composite (MPa)
P external force per unit width of laminate (N/mm)
N crack tip load per unit width of a laminate (N/mm)
M moment per unit width of a laminate (N)
X1 upper crack arm length (mm)
w total crack opening displacement (mm)
t total laminate thickness (mm)
t1 upper cracked arm thickness (mm)
t2 lower cracked arm thickness (mm)
d load line displacement (mm)
e midplane strain in Fiber direction (mm/mm)
h0 phase angle of ADCB specimen (deg)
j midplane curvature (mm1)
t Poisson’s ratio of aluminum
t12 major Poisson’s ratio of CFRP
t21 minor Poisson’s ratio of CFRP
r11 CFRP strength in fiber direction (MPa)
r22 CFRP strength in transverse direction (MPa)
ry aluminum yield strength (MPa)
s12 CFRP shear strength (MPa)

altering the surface roughness. According to Park et al. GLARE interface shear strength, rises by increasing the aluminum sur-
face roughness [16]. However for enormous rough aluminum surface, the peaks and valleys may trap the air during process-
ing and reduce the interface strength.
Discrepancy in coefficient of thermal expansion between metal and composite layers produces a thermal residual stress
at their interface which also may facilitate delamination in this region. Kim et al. presented smart cure cycle in autoclave to
reduce the interfacial residual stress of CARALLs [17,18]. Xue et al. designed a thermal clamp to deteriorate the interfacial
residual stress of CARALLs by constraining the layers during the curing process of FMLs [19]. Existence of interfacial residual
stresses may lead FMLs to distort. Distortion of fiber metal laminates were declined by Abouhamzeh et al. by reducing the
interfacial residual stresses [20]. In the current research we urge to fabricate carbon fiber metal laminates at ambient tem-
perature to reduce the interfacial residual stresses and hence to increase the interfacial adhesion.
Measurement of interfacial fracture toughness of bi-material interfaces was performed previously [21–24]. Ning et al.
applied compliance calibration method (CCM) for mode I fracture toughness calculation of asymmetric double cantilever
beam (ADCB) specimen [23]. However this may ignore the mode II enrolment in ADCB specimen’s fracture toughness assess-
ment. The pure mode I and mode II fracture toughness values were achieved for sandwich panel interface by Davidson et al.
[25]. They equalized the moments of two cracked arms of ADCB specimen and obtained a specific thickness for each arm.
This may solve the interfacial adhesion measurement at the early stages of crack advancement. Nevertheless, when the inter-
face crack length grows further, the upper and lower crack arms may deflect differently and deviate from the pure mode
fracture. Measurement of bi-material interfacial fracture toughness was also performed previously by reduction of mode
II component of fracture for Aluminum/Epoxy interface [26] and Aliminium/Composite interface [27].
In this research the effect of different aluminum surface treatments on the interfacial fracture toughness of CARALLs will
be experimentally and numerically studied. The proper condition of aluminum surface treatment will be discovered by
evaluating the interfacial adhesion behavior of laminates by plate theory analyses along with CCM. The reliability of the esti-
mated fracture energy values obtained by the aforementioned methods, was evaluated by finite element analysis of the crack
growth. In this regard cohesive traction-separation laws (TSL) were achieved by direct method and the crack growth in ADCB
specimens were simulated.
 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151 141

2. Experimental

2.1. Materials

An aluminum alloy (7075-T6) sheet of 2 mm thickness was utilized as the metallic part of fiber metal laminates. 2 mm
thick aluminum sheet was selected to perform delamination tests at the linear elastic region. Aluminums with lower thick-
ness may yield during delamination tests and invalidate the acquired data. The Structural KumhoÒ 828 Epoxy resin with
TETA hardener was incorporated as matrix to wet up 12 K- T300 unidirectional carbon fabrics.

2.2. FML fabrication

Schematic representation of FML fabrication mold is shown in Fig. 1. The mold consists of two ground steel plates
connected by a set of bolts and nuts. Linear springs were placed between nuts and mold plate to pressurize the FML during
processing. Totally 1.2 bar pressure was exerted to the laminate by precisely tightening the nuts which squeezed out the
extra resin. The volume fraction of fibers inside the composite layer was determined to be 40 ± 1% according to ASTM
D3171 standard test method [28]. The system maintained under pressure for 24 h and let to fully cure at ambient temper-
ature for 7 days.
The FMLs consist of two aluminum sheets placed in upper and lower ends having four layers of Carbon/Epoxy composite
layers in core. Each composite layer has a thickness of 0.25 mm which four layers form a 1 mm thick CFRP inside the FML.
Each specimen was constrained by three locating pins to avoid slippage of layers during the pressurizing stage of FMLs.
Laminates were sealed with a bag and vacuumed from the bottom side of the mold to properly consolidate different layers.
Fabricated fiber metal laminates are shown in Fig. 2.
Dimensions and condition of ADCB specimens are represented in Fig. 3. A 25 lm thick release film was placed at one side
of the laminate to make an initial crack with a length of 50 mm. The nominal widths of specimens were 20 ± 0.5 mm. A set of
piano hinges was glued precisely on two sides of laminates to load the ADCB specimen from opposite sides. Testing was per-
formed with a displacement speed of 3 mm/min and the crack advancement was registered by a digital camera.
Delaminated surface of composite at the location near to the end of crack advancement was evaluated by scanning elec-
tron microscopy (SEM-TESCAN).

2.3. Aluminum surface treatment

The surface treatments those applied to aluminum substrates are summarized in Table 1. First, aluminum substrates were
rinsed with acetone and abraded smoothly with #400 grit sandpaper and then rinsed with distilled water and finally dried in
oven. The aforementioned surface treatment called reference treatment. The other surface treatments were applied to the
aluminum substrates undergone a reference treatment. Second, the reference substrates were acid washed with HCl having
11% volumetric concentration to elevate the aluminum surface roughness. The acid etching was performed at ambient tem-
perature for 30 min. Finally the aluminums were rinsed with distilled water and dried out. The third kind of surface treat-
ment was performed by alkaline etchant. Reference aluminum samples have been immersed in 5 wt% NaOH solvent for
5 min at 70 °C temperature. Afterward, the oxidized aluminum sheets were dried in oven to stabilize the oxide layer and
washed with tap water to remove the oxide residue. Finally samples rinsed with distilled water and dried out. The fourth
kind of surface treatment was performed by HCl etching forwarded by alkaline etching described previously.

Vacuum Breather Aluminum Composite

Vacuum Locating Sealant
bag sheets Layer
port pin

Fig. 1. Schematic representation of FML mold.

142 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151

Fig. 2. Image of fabricated FMLs.

Piano
hinge 25 thick
Release film

50 mm
13 mm
200 mm

Fig. 3. Representation of ADCB specimen for evaluation of interfacial fracture toughness.

Table 1
Illustration of different surface treatments of aluminum substrates.

Surface treatment Description

Reference Substrates were primarily degreased with acetone and then abraded with #400 grit sandpaper and rinsed with distilled water
Acid The reference specimens were immersed for 30 min in 11% v/v HCl acid and then rinsed with distilled water
Alkaline The reference specimens were immersed for 5 min in 70 °C solution of 5 wt% NaOH and then rinsed with distilled water
Acid-alkaline Reference specimens were treated with acid solution and then alkaline solution respectively

Fig. 4. Sample with fibers orientation of 45 degree for determination of CFRP shear modulus. (All tests were replicated for three times).

In order to evaluate the effect of various surface treatments on the surface morphology and roughness of aluminum sub-
strates, micro-profiles of samples were obtained by atomic force microscopy.

2.4. Determination of CFRP elastic properties

Tensile modulus of CFRP in fiber direction and perpendicular to fiber direction was evaluated based on standard test
method ASTM D 3039/3039M [29]. During the test procedure, major Poisson’s ratio was determined by recording the lateral
strain by a digital camera. Also in order to attain the in-plane shear modulus of CFRP, Hennessey method was utilized [30]. In
this respect, samples that have fibers direction of 45 degree with respect to the tension direction were tested (Fig. 4) and the
shear modulus acquired according to Eq. (1).
 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151 143

1
G12 ¼ ð1Þ
4
E45
 E122  E111 ð1  2m12 Þ

3. Calculation of energy release rate

In this research the energy release rate (ERR) of asymmetric double cantilever beam was acquired according to the plate
theory analyses proposed by Davidson et al. [31,32]. According to the plate theory, loads and moments per unit width of a
laminate (N and M respectively) could be related to the midplane strain in x direction (e) and curvature (k) based on Eq. (2)
for un-cracked part of the laminate (Fig. 5).
N ¼ Ae þ Bj
M ¼ Be þ Dj
or ð2Þ
e ¼ A0 N þ B0 M
j ¼ B0 N þ D0 M
where A, B and D are extensional, coupling and bending stiffness tensors of the laminate, respectively. For specimen arms
above and below the crack plane (Fig. 5), N and M are defined as Eq. (3).
Ni ¼ Ai ei þ Bi ji i ¼ 1; 2
Mi ¼ Bi ei þ Di ji i ¼ 1; 2
or ð3Þ
ei ¼ A0i Ni þ B0i Mi i ¼ 1; 2
ji ¼ B0i Ni þ D0i Mi i ¼ 1; 2
The energy release rate could be obtained according to Eq. (4) [31]
1 pffiffiffiffiffiffiffiffiffi 
G¼ c1 N2c þ c2 M2c þ 2 c1 c2 Nc Mc sin C ð4Þ
2
where
sin C ¼ pcffiffiffiffiffiffi
12 ffi
c1 c2
;
c1 ¼ A01 þ A02 þ B01 t 1  B02 t 2 þ D01 t 21 =4 þ D02 t22 =4
ð5Þ
c2 ¼ D01 þ D02
c12 ¼ D02 t 2 =2  D01 t 1 =2  B01  B02
t1 and t2 are the thickness of upper and lower crack arms, respectively. The Nc and Mc are the concentrated crack tip force and
moment which could be obtained as Eq. (6).
Nc ¼ N1 þ a11 N þ a12 M
ð6Þ
Mc ¼ M 1  N1 t1 =2 þ ða11 t1 =2  a21 ÞN þ ða12 t 1 =2  a22 ÞM
where
a11 ¼ A1 A0 þ ðB1  A1 t 2 =2ÞB0
a12 ¼ A1 B0 þ ðB1  A1 t 2 =2ÞD0
ð7Þ
a21 ¼ B1 A0 þ ðD1  B1 t 2 =2ÞB0
a22 ¼ B1 B0 þ ðD1  B1 t2 =2ÞD0

X1
w

0)
0)

Fig. 5. Asymmetric double cantilever beam specimen for interface fracture toughness evaluation.
144 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151

Supposing that the angle of un-cracked region in ADCB specimen with respect to the horizon to be h0 (Fig. 5), the x com-
ponent of force and moment for the upper arm will be defined as Eq. (8).
N1 ¼ P sin ðh0 Þ
ð8Þ
M1 ¼ PX 1
where X1 is the upper crack arm length (Fig. 5), that was obtained by image processing analyses for a desired crack length.

4. Finite element validation

In order to validate the estimated energy release rates, 2D plane strain finite element simulations were carried out by
ANSYS software. The Aluminum-composite interface behavior was defined by cohesive zone elements. According to Hiller-
borg et al. [33] the interface traction could be obtained as

dG
r¼ ð9Þ
dw
where w is the crack opening displacement at initial crack tip [34] (shown in Fig. 5). In present research, w was acquired by
image processing analyses of ADCB specimens at different crack lengths. For sake of simplicity, the total energy release rate
values (consist of Mode I and Mode II) acquired by plate theory analyses and CCM, differentiated against crack opening dis-
placement and assumed as mode I traction-separation law (TSL). The area under TSL was calculated and defined as mode I
critical separation energy to software along with the maximum traction appeared in TSLs. Mode II critical separation energy
was defined near zero.

5. Results and discussion

The averages of CFRP and aluminum sheet tensile and shear properties are summarized in Table 2. The elasticity con-
stants appeared in Table 2, were utilized in plate theory analysis to determine interfacial energy release rates.
Surface morphology of aluminum substrates treated as reference, acid, alkaline and acid-alkaline are presented in
Fig. 6a–d, respectively. It is clear from Fig. 6a that the reference surface treatment produces quite smooth valleys and peaks
in the substrate. Acid washing of reference substrate in Fig. 6b adds fine roughness on the previous peaks and valleys and
increases the average surface roughness (Ra and Sa increased double-fold as compared to the reference substrate). This
may be beneficial to improve the bonding behavior by enhancing the fiber bridging mechanism. Topography of alkaline
treatment of the reference substrate is shown in Fig. 6c. The figure portrays that the surface roughness values do not affected
by alkaline treatment, hence sharp ridges are formed on surface which could be due to the high surface oxide layer. Acid-
alkaline treatment of substrates (Fig. 6d) subsumed both high roughness generated by acid washing and porous oxide layer
created on the surface by alkaline treatment. This may improve bonding with two mechanisms of interlocking and chemical
bonding, respectively.
Load-displacement curves of ADCB specimens for various surface treatments of aluminum substrates are shown in Fig. 7.
All curves are represented up to 50 mm of crack advancement. It is clear that the acid-alkaline surface treatment displays
relatively the highest peak load and load line displacement, as compared to the other samples.
A frame of interface delamination for the reference sample is exhibited in Fig. 8. It is clear that the interface fracture is gen-
erally cohesive failure, including fiber bridging between aluminum substrate and composite layer. According to Sun et al. [35]
when the failure of composite is cohesive, the total energy release rate consist of matrix and fibers bridging fracture energies.
Bridging of carbon fibers, their pull out and rupture, dissipate considerable amount of energy in cohesive failure during delam-
ination of layers. As stated by Xiao et al. in the bi-material interfaces, the crack tend to deflect into the more compliant mate-
rial which cause a cohesive failure and fiber bridging [36]. The stable cohesive failure of samples which include bridging led to
the formation of resistance curve (R-curve) for delamination rather than a single value energy release rate [37].
It was observed that the overall angle of ADCB specimen (h0) during testing was altered. These angles were obtained by
image processing analyses and are depicted as curves in Fig. 9. It is clear that the angle curves of the acid treated and alkaline
treated samples do not differ significantly with the reference treated sample. However, when the aluminum substrates
treated with both acid etching and alkaline etching, the angle curve rises up. Referring to Fig. 5, by increasing the h0 angle,
the mode II component of the specimen increases, which may be due to the higher interface bonding behavior.

Table 2
Tensile properties of CFRP and aluminum.

CFRP Aluminum
E11 (MPa) E22 (MPa) G12 (MPa) m12 m21 E (MPa) m
93,200 3748 1433 0.26 0.0104 73,650 0.33
r11 (MPa) r22 (MPa) s12 (MPa) ry (MPa)
954.6 15.53 26.3 473.2
 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151 145

Fig. 6. AFM topography of aluminum substrates. (a) reference, (b) acid, (c) alkaline, (d) acid-alkaline.

Fig. 7. Load-displacement curves of ADCB specimens.

Fig. 10 portrays R-curves of a reference sample obtained by methods of the plate theory and CCM introduced in ASTM D
5528 [38], respectively. It is clear that the energy release rates are quite close in two methods, when the crack length is near
to its initial value (50 mm). This is due to the fact that the h0 angle retains a low value at the initial crack lengths, which
146 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151

Fig. 8. ADCB test of CARALL (Scale bars are 5 mm each).

4
Theta0 (deg)

2 HCl+NaOH
NaOH
1 HCl
Reference
0
50 55 60 65 70 75 80 85 90 95 100
Crack length (mm)

Fig. 9. Variations of h0 angle for ADCB specimens during testing.

1
G (plate theory)
0.829 0.815
G (CCM) 0.759
0.8

0.628
0.6
G (N/mm)

0.507
0.473 0.481
0.543 0.546
0.517
0.4 0.359
0.424
0.266 0.369 0.352 0.334
0.175 0.293
0.2
0.096 0.226
0.172
0.099
0
50 55 60 65 70 75 80 85 90 95 100
Crack length (mm)

Fig. 10. R-curves of a reference sample acquired by the plate theory and compliance calibration method.

causes ADCB specimen to exhibit near mode I fracture. By crack advancement to the higher lengths, plate theory estimates
higher fracture energy values than CCM, which is due to the higher h0 angle and its noticeable effect on the energy release
rate obtained by plate theory. Higher h0 angle cause contribution of mode II on the fracture and increases the overall fracture
toughness.
The R-curves of CARALLs with various surface treatments of aluminum substrates are portrayed in Fig. 11. It is shown in
Fig. 11 that the reference and alkaline surface treatments possess approximately the same fracture energy up to 80 mm crack
length. However, the interface energy release rate of the FML with alkaline treatment drops the corresponding value of the
reference sample at higher crack lengths. As the mode II component of the G dominates at higher crack lengths, the oxidized
layer of alkaline surface may prone to slippage and lower the G value.
The acid treatment of aluminum substrates increases the initiation and propagation interfacial fracture energy of FMLs as
shown in Fig. 11. The fine peaks and valleys created on the aluminum surface after acid washing, increase the interfacial area
and the mechanical interlocking of the composite and metal layer. An acid treatment of aluminum substrates may not con-
tain enough surface oxide to attain sufficient bond durability [39]. The lack of oxide layer on the aluminum may be resolved
by alkaline etching of acid treated substrates. An elevated improvement is seen when aluminum substrates are etched with
 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151 147

Fig. 11. R-curves of FMLs with various aluminum surface treatments.

Reference Alkaline Acid Acid-Alkaline

Fig. 12. Delaminated surface of aluminum substrates treated by different methods.

both acid and alkaline etchants. As suggested by AFM images, in this case, aluminum surfaces possess sufficient surface
roughness and porosity which led to higher interfacial bonding. The FML with acid-alkaline surface treatment have 70%
and 35% higher crack initiation and propagation energy release rates respectively with respect to the reference sample.
In order to better illustrate the R-curves presented in Fig. 11, delaminated surfaces of aluminum substrates are shown in
Fig. 12. Different R-curves for various surface treatments originate from change in the adhesive/cohesive failure contribution
on the fracture energy. It is clear from Fig. 12 that the delaminated surface of the reference sample has low fiber evidence on
its surface and most of its delamination is adhesive failure. In the alkaline treated sample, the cohesive failure contribution
increased, but still some pitted regions exist that are shown by arrows. The evidence of clear aluminum surface on the acid
treated samples could rarely be seen and the failure is cohesive generally. However some minor narrow aluminum surfaces
could be seen which are indicated by arrows. At last the delaminated surface of acid-alkaline surface treated sample is shown
that is dark completely. Synergistic effects of surface roughness and oxide layer led to completely cohesive failure and
increase of R-curve for this sample.
The most prominent factor that promotes the interfacial fracture energy, seems to be fiber bridging between aluminum
layer and composite laminate. Surface treatment of aluminum layers certainly affect the fiber bridging length. The resultant
loads carried by bridged fibers at the interface could change by aluminum surface treatment and this will affect the obtained
fracture energy. The fiber bridging lengths for various surface treatments of aluminum substrates are further discussed at the
cohesive law section.
The SEM images of delaminated surface of composite layer at the stable crack region (near to 100 mm crack length) is
shown in Fig. 13(a) and (b) for acid-alkaline sample. It can be seen from Fig. 13a that the carbon fibers are embedded inside
polymer matrix and some fibers detached from the polymer that show a darker field. From the low magnification image pre-
sented in Fig. 13b, it is clear that some carbon fibers are ruptured during cohesive failure during fiber bridging which may
absorb substantial energy of delamination.
148 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151

Fig. 13. SEM images of delaminated HCl + NaOH sample. (a) 2500, (b) 100.

Fig. 14. Traction separation laws (TSL) for interfacial behaviors of (a) reference, (b) HCl, (c) NaOH, (d) HCl + NaOH surface treatments of aluminums.

Table 3
Maximum tractions and critical delamination energies obtained by plate theory and CCM method for different FML samples.

No. Sample Plate theory CCM

Tmaxa (MPa) GI-crb (N/mm) Tmax (MPa) GI-cr (N/mm)
1 Reference 1.084 0.5585 0.854 0.4344
2 HCl 1.421 0.8672 1.155 0.5826
3 NaOH 1.222 0.5382 0.998 0.4427
4 HCl + NaOH 1.757 0.8011 1.474 0.6051
a
Maximum traction.
b
Critical mode I delamination energy.
 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151 149

Fig. 15. Equivalent stress distribution after 20 mm displacement of load line for the reference FML. Cohesive zone elements properties were defined
according to (a) plate theory analyses, (b) Compliance calibration method.

(b)

Fig. 16. Comparison of different methods in obtaining delamination load-displacement curves of (a) reference, (b) HCl, (c) NaOH, (d) HCl + NaOH surface
treatments of aluminum.

The cohesive laws for different aluminum surface treatments including reference, acid, alkaline and acid-alkaline are pre-
sented in Fig. 14a–d, respectively. It is seen that all TSLs exhibit softening behavior [40]. The TSLs predicted by plate theory
are beyond that of predicted by CCM, especially at initial separation stages. It is evident from Fig. 14a–d, the tractions are
150 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151

approximately vanished at 1 mm of separation. This indicates that beyond 1 mm of COD, fibers are slightly bridged at the
interface region, hence negligible load was transferred after 1 mm separation at the interface due to fiber bridging.
The areas under TSLs of Fig. 14 were calculated as critical delamination energies and are presented in Table 3 along with
maximum tractions at initiation of TSLs. It is clear that plate theory estimations are higher than CCM. By comparing the max-
imum traction and critical delamination energy for different aluminum surface treatments, it is observant that the alkaline
treatment does not have significant influence on critical separation energy, while it increases maximum traction. On the
other hand, acid treatment improves both critical energy and maximum traction at separation by surface roughness
mechanism.
Stress plots of simulated reference ADCB samples with cohesive law of plate theory and CCM are shown in Fig. 15a and b,
respectively. In both figures maximum stresses occurred at the upper aluminum arm. Also the maximum stress of sample
with plate theory cohesive law is higher than CCM cohesive law. The later fact is due to higher energy and maximum traction
estimated by plate theory.
The load-displacement curves extracted from finite element simulations are compared to the experimental counterparts
for different aluminum surface treatments in Fig. 16. The slopes of curves at initial stages are identical, which suggest reli-
ability of obtained energy release rates at initial crack lengths by two methods. Also at the end of separation curves, the plate
theory curves are closer to the experimental curves. This shows the fact that plate theory estimations of interfacial fracture
energy at higher crack lengths are more authentic than CCM counterparts.
It was found in this research that the acid treatment and alkaline treatment of aluminum substrates promote the bonding
of aluminum to the composite by interlocking and chemical bonding mechanisms. The surface roughness and oxide layer
plays dominant role on interfacial fracture toughness.

6. Conclusions

The effects of acid and alkaline surface treatment of aluminum substrates on the interfacial fracture toughness of carbon
fiber metal laminates was investigated. In order to precisely fabricate fiber metal laminates at room temperature, a special
mold containing vacuum port and linear springs had been prepared. Aluminum substrates were treated chemically with acid
washing, alkaline solution and both respectively. Atomic force microscopy images show high roughness for acid treatment
and porous oxide layer for alkaline treatment, correspondingly. Mutual influences of acid and alkaline etching preserve both
surface roughness and oxide layer on the aluminum substrate. To evaluate the interfacial fracture toughness, asymmetric
double cantilever beam specimen was utilized. The energy release rates were evaluated by the plate theory analyses and
compliance calibration method. It was observed that the energy release rates acquired by abovementioned methods are
identical in the initial crack lengths due to lower mode II contribution. Nonetheless, by increasing the crack length, the frac-
ture energies estimated by the plate theory surpasses the CCM counterpart. Comparison of interfacial fracture toughness of
FMLs containing different aluminum surface treatments shows that FMLs with both acid and alkaline aluminum treatment
cause higher energy release rate as compared to other counterparts. SEM images show sufficient wetting of carbon fibers in
FML and some fibers were ruptured as a consequence of bridging at the interface. The finite element analysis based on direct
cohesive zone method was performed to evaluate the reliability of estimated fracture energies by two methods. The cohesive
laws extracted from the plate theory showed closer behavior to experiments as compared to CCM counterparts suggested by
FEM study.

References

[1] Vlot A. GLARE: history of the development of a new aircraft material. Netherlands: Kluwer Academic Publishers; 2001.
[2] Lin CT, Kao PW, Yang FS. Fatigue behaviour of carbon fibre-reinforced aluminium laminates. Composites 1991;22:135–41.
[3] Wang W, Rans C, Alderliesten RC, Benedictus R. Predicting the influence of discretely notched layers on fatigue crack growth in fibre metal laminates.
Engng Fract Mech 2015;145:1–14.
[4] Gupta M, Alderliesten RC, Benedictus R. Crack paths in fibre metal laminates: role of fibre bridging. Engng Fract Mech 2013;108:183–94.
[5] Sinmazcelik T, Avcu E, Özgür Bora M, Çoban O. A review: fibre metal laminates, background, bonding types and applied test methods. Mater Des
2011;32:3671–85.
[6] Poodts E, Ghelli D, Brugo T, Panciroli R, Minak G. Experimental characterization of a fiber metal laminate for underwater applications. Compos Struct
2015;129:36–46.
[7] Varma PCR, Colreavy J, Cassidy J, Oubaha M, McDonagh C, Duffy B. Corrosion protection of AA 2024-T3 aluminium alloys using 3,4-diaminobenzoic
acid chelated zirconium–silane hybrid sol–gels. Thin Solid Films 2010;518:5753–61.
[8] Park SY, Choi WJ, Choi HS. The effects of void contents on the long-term hygrothermal behaviors of glass/epoxy and GLARE laminates. Compos Struct
2010;92:18–24.
[9] Rodi R, Alderliesten R, Benedictus R. Experimental characterization of the crack-tip-opening angle in fibre metal laminates. Engng Fract Mech
2010;77:1012–24.
[10] de Almeida CFG, Bastian FL, Castrodeza EM. Comparison of J-R curves and JC values of C(T) and M(T) specimens of bidirectional GLARE 3 5/4 0.3 fiber-
metal laminates. Engng Fract Mech 2016;159:79–89.
[11] Khan SU, Alderliesten RC, Rans CD, Benedictus R. Application of a modified wheeler model to predict fatigue crack growth in fibre metal laminates
under variable amplitude loading. Engng Fract Mech 2010;77:1400–16.
[12] Castrodeza EM, Schneider Abdala MRW, Bastian FL. Crack resistance curves of GLARE laminates by elastic compliance. Engng Fract Mech
2006;73:2292–303.
[13] Castrodeza EM, Bastian FL, Perez Ipiña JE. Fracture toughness of unidirectional fiber–metal laminates: crack orientation effect. Engng Fract Mech
2005;72:2268–79.
 A. Zamani Zakaria et al. / Engineering Fracture Mechanics 172 (2017) 139–151 151

[14] Marannano GV, Pasta A. An analysis of interface delamination mechanisms in orthotropic and hybrid fiber-metal composite laminates. Engng Fract
Mech 2007;74:612–26.
[15] Salgin B, Özkanat Ö, Mol JMC, Terryn H, Rohwerder M. Role of surface oxide properties on the aluminum/epoxy interfacial bonding. J Phys Chem C
2013;117:4480–7.
[16] Park SY, Choi WJ, Choi HS, Kwon H. Effects of surface pre-treatment and void content on GLARE laminate process characteristics. J Mater Process
Technol 2010;210:1008–16.
[17] Kim HS, Park SW, Hwang HY, Lee DG. Effect of the smart cure cycle on the performance of the co-cured aluminum/composite hybrid shaft. Compos
Struct 2006;75:276–88.
[18] Kim HS, Park SW, Lee DG. Smart cure cycle with cooling and reheating for co-cure bonded steel/carbon epoxy composite hybrid structures for reducing
thermal residual stress. Compos A Appl Sci Manuf 2006;37:1708–21.
[19] Xue J, Wang W-X, Takao Y, Matsubara T. Reduction of thermal residual stress in carbon fiber aluminum laminates using a thermal expansion clamp.
Compos A Appl Sci Manuf 2011;42:986–92.
[20] Abouhamzeh M, Sinke J, Benedictus R. Investigation of curing effects on distortion of fibre metal laminates. Compos Struct 2015;122:546–52.
[21] Shahverdi M, Vassilopoulos AP, Keller T. Modeling effects of asymmetry and fiber bridging on Mode I fracture behavior of bonded pultruded composite
joints. Eng Fract Mech 2013;99:335–48.
[22] Ning H, Iijima T, Hu N, Liu Y, Wu L, Liu F, et al. Investigation on mode-II interface fracture toughness of CFRP/Al laminates toughened by VGCF
interleaves. J Mater Sci 2015;50:1915–23.
[23] Ning H, Li Y, Hu N, Arai M, Takizawa N, Liu Y, et al. Experimental and numerical study on the improvement of interlaminar mechanical properties of Al/
CFRP laminates. J Mater Process Technol 2015;216:79–88.
[24] Patra A, Mitra N. Interface fracture of sandwich composites: Influence of MWCNT sonicated epoxy resin. Compos Sci Technol 2014;101:94–101.
[25] Davidson P, Waas AM, Yerramalli CS. Experimental determination of validated, critical interfacial modes I and II energy release rates in a composite
sandwich panel. Compos Struct 2012;94:477–83.
[26] Liu J, Chaudhury MK, Berry DH, Seebergh JE, Osborne JH, Blohowiak KY. Fracture behavior of an epoxy/aluminum interface reinforced by sol–gel
coatings. J Adhes Sci Technol 2006;20:277–305.
[27] Shimamoto K, Sekiguchi Y, Sato C. Effects of surface treatment on the critical energy release rates of welded joints between glass fiber reinforced
polypropylene and a metal. Int J Adhes Adhes 2016;67:31–7.
[28] ASTM-International. ASTM D3171–15: Standard Test Methods for Constituent Content of Composite Materials. West Conshohocken, PA2015.
[29] ASTM-International. ASTM D3039/D3039M-14: Standard Test Method for Tensile Properties of Polymer Matrix Composite Materials. West
Conshohocken, PA2014.
[30] Hennessey JMW, James M, Riley MB. Experimental methods for determining shear modulus of fiber reinforced composite materials. NASA Technical
Note. Accession Number: AD0623316; 1965.
[31] Davidson BD, Hu H, Schapery RA. An analytical crack-tip element for layered elastic structures. J Appl Mech 1995;62:294–305.
[32] Sundararaman V, Davidson BD. An unsymmetric double cantilever beam test for interfacial fracture toughness determination. Int J Solids Struct
1997;34:799–817.
[33] Hillerborg A, Modéer M, Petersson PE. Analysis of crack formation and crack growth in concrete by means of fracture mechanics and finite elements.
Cem Concr Res 1976;6:773–81.
[34] Khoramishad H, Hamzenejad M, Ashofteh RS. Characterizing cohesive zone model using a mixed-mode direct method. Engng Fract Mech
2016;153:175–89.
[35] Sun Z, Hu X, Chen H. Effects of aramid-fibre toughening on interfacial fracture toughness of epoxy adhesive joint between carbon-fibre face sheet and
aluminium substrate. Int J Adhes Adhes 2014;48:288–94.
[36] Xiao F, Hui C-Y, Kramer EJ. Analysis of a mixed mode fracture specimen: the asymmetric double cantilever beam. J Mater Sci 1993;28:5620–9.
[37] Anderson TL. Fracture mechanics: fundamentals and applications. 3rd ed. CRC Press; 2005.
[38] ASTM-International. ASTM D5528-13, Standard Test Method for Mode I Interlaminar Fracture Toughness of Unidirectional Fiber-Reinforced Polymer
Matrix Composites. West Conshohocken, PA2013.
[39] Critchlow GW, Brewis DM. Review of surface pretreatments for aluminium alloys. Int J Adhes Adhes 1996;16:255–75.
[40] Elices M, Guinea GV, Gomez J, Planas J. The cohesive zone model: advantages, limitations and chalenges. Engng Fract Mech 2002;69:137–63.