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

net/publication/303916576

Estimation of the elastic modulus of the

alumina coated AA1050 aluminum: Modeling
and Experiments

Article in Materials Science and Engineering A · June 2016

DOI: 10.1016/j.msea.2016.06.012

CITATIONS READS

0 66

4 authors, including:

Bahri Amir Noamen Guermazi

Ecole Nationale d'Ingénieurs de Sfax Ecole Nationale d'Ingénieurs de Sfax
9 PUBLICATIONS 6 CITATIONS 29 PUBLICATIONS 181 CITATIONS

SEE PROFILE SEE PROFILE

Khaled Elleuch
University of Sfax
78 PUBLICATIONS 643 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

in-situ wear sensing View project

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Noamen Guermazi
letting you access and read them immediately. Retrieved on: 03 October 2016
 Materials Science & Engineering A 670 (2016) 188–195

Contents lists available at ScienceDirect

Materials Science & Engineering A

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

Estimation of the elastic modulus of the alumina coated AA1050

aluminum: Modeling and experiments
Amir Bahri, Noamen Guermazi n, Mansour Bargui, Khaled Elleuch
Laboratoire Génie des Matériaux et Environnement (LGME), Ecole Nationale d’Ingénieurs de Sfax (ENIS), Université de Sfax, B.P.1173-3038, Sfax, Tunisia

art ic l e i nf o a b s t r a c t

Article history: This paper deals with the prediction of the elastic modulus of alumina coating deposited on an aluminum
Received 13 May 2016 substrate. Three- and four-point bending tests were used for the mechanical characterization. Three
Received in revised form configurations were retained: a sandwich model, a model with coating subjected to tensile and model
4 June 2016
with coating placed in the state of compression. For the prediction of the elastic modulus of anodic oxide
Accepted 6 June 2016
Available online 9 June 2016
layer, an inverse method based on a numerical/experimental comparison was used. From the main re-
sults, it was found that: Finite element (FE) simulations of three- and four-point bending tests have been
Keywords: successfully conducted. In particular, both numerical 2D and 3D models were considered to be compared
Alumina coating with the experimental results. The predicted elastic modulus was found to largely depend on the coating
Sandwich plates
thickness. The elastic modulus was found to be around 15 GPa for a thickness of 20–50 mm and 20 GPa for
Flexural test
an alumina coating of 80 mm. The numerical/experimental correlation allowed to conclude that when the
Elastic modulus
Modeling alumina coating was subjected to compression it was more resistant than in tensile.
& 2016 Elsevier B.V. All rights reserved.

1. Introduction context, Scanning Electron Microscopy (SEM) has been used by

several authors to evaluate the oxide layer performances, but it
Today, aluminum and aluminum alloys are widely used in could not be sufficient when it is not associated to mechanical
several fields. In particular, the anodizing of aluminum and its tests [10]. Therefore, three-point bending [11–19] and four-point
alloys has diverse applications in transport, abrasion-resistant bending tests [20–22] have been widely used not only to evaluate
coatings, decoration and dielectric and insulating films in the the damage of the metallic coatings, but also to determine their
electronics industry [1,2]. The use of these materials in the auto- mechanical performance.
motive field is very important in order to lighten their structures. Whatever the bending mode, there is always a contact between
However, surface deterioration occurs generally as a result of the punches and alumina coating, which may cause a local da-
mechanical problems, such as friction and deformation [3,4]. An- mage. It is well known that sandwich beams coated from both
odization is, therefore, useful in various applications to enhance sides allow good stiffness and wear resistance when compared to
the mechanical and tribological properties such as micro-hard- an uncoated substrate [23]. It is worth noting that since there are
ness, abrasion resistance and corrosion protection [5]. It is worth no new added materials during the anodizing process, and as the
reminding that this treatment obtained by the electrochemical anodic oxides are supposed to be adherent enough, research ef-
process, consists of converting aluminum into its oxide by the forts have not focused much on the adhesion phenomenon. Many
appropriate selection of the electrolyte and the anodizing condi- experimental studies have assessed the mechanical properties and
tions, such as current density, voltage, temperature etc.. [6,7]. failure behavior of alumina oxide coatings [24–27]. However, to
Anodic oxide layers which are deposited using electrochemical our knowledge, there are few reported numerical works that were
process are generally characterized by very important thickness, a carried out to characterize such alumina coating. The results pre-
strong adhesion and a high hardness [8]. Nevertheless, when these sented in this paper may provide further insight into the problem.
oxide layers are submitted to a concentrated force, cracks and local In particular, this investigation outlines the prediction of the
damages appear in the coating surface [9]. The use of the tensile elastic modulus for alumina coating formed on aluminum sub-
test to characterize the anodized aluminum can be inappropriate strate using both three and four-point bending tests. Besides, re-
for some cracks that may appear in the fixtures. In the same lying on some previous experimental results reported by Bargui
et al. [23], a comparison between the numerical modeling and
n
Corresponding author. experimental approaches was conducted to extract the elastic
E-mail address: [email protected] (N. Guermazi). modulus of the alumina coating.

http://dx.doi.org/10.1016/j.msea.2016.06.012
0921-5093/& 2016 Elsevier B.V. All rights reserved.
 A. Bahri et al. / Materials Science & Engineering A 670 (2016) 188–195 189

2. Materials and experimental details the anodic oxide layer was measured using ELCOMETER 355 Top
thickness gauge equipped with eddy current prob. For twenty (20)
2.1. Materials and procedures measurement points on both sides, the average thickness was
considered.
The studied material in this work consists of an AA1050 alu- Three and four-point bending tests were performed on an
minum plate with dimensions of 100
20
3 mm3, a chemical universal machine LLOYD instrument LR 50 kN at a loading speed
composition (wt%): Si 0.11%, Mn 0.005%, Cu 0.005%, Ti 0.014, Zn of 2 mm min  1. The distance between the fixed indenters is
0.009%, Fe 0.37%, Pb 0.006%, Mg 0.005% and aluminum balance. l¼ 60 mm for the two bending modes with a diameter d ¼5 mm,
Alumina coating was prepared by the anodic oxidation of the while the calibrated distance between the loading ones in the
aluminum substrate. Before anodizing, the Al specimens were four-point bending was S¼ 30 mm with a diameter D¼ 10 mm. A
mechanically polished to P1000 grade paper followed by a che- displacement of 8 mm was applied on these punches (Fig. 1a–d).
mical polishing in a 15:85 (V/V) mixture of concentrated HNO3 and For both three-point and four-point bending tests, thirty (30)
H3PO4 at 85 °C for 2 min, etching in 1 M NaOH solution at room measurements were made and the average value was considered.
temperature for 1 min and chemical pickling in 30% (V/V) HNO3 Fig. 1a–d shows all the possible configurations in the case of three
solution at room temperature for 30 s. Water rinsing was used and four-point bending tests. The anodic oxide layer to be formed
after each step as previously reported in the work of Bensalah et al. on aluminum substrate will be subjected to tensile or compression
[28]. Afterwards, aluminum samples were anodized in a vigorously or both tensile and compression.
stirred sulphuric acid solution (146 g/L) maintained within 71 °C
of the temperature set, using a cryothermostat Lab-companion
RW-2025G. After the anodizing step, the samples were washed in 3. Numerical analysis
deionised water and then dried.
In this work, the numerical analysis was carried out using the
2.2. Characterization methods commercial nonlinear FE code ABAQUS. The alumina oxide layer
was assumed to be perfectly bonded to the core, eliminating the
In order to characterize the anodic film formed on the alumi- delamination failure mode. The core was completely constrained
num substrate, various tests were carried out. According to the by the oxide layer. Moreover, the thickness of the tested sheets can
methodology described in previous works [25,27], the thickness of be neglected when compared to the sheet surface, thus a plane-

Fig. 1. Schematic of the geometry of the sandwich beam and the loading rollers: (a) threepoint bending, (b) four-point bending: sandwich model, (c) four-point bending:
oxide layer placed in state of tensile, and (d) four-point bending: oxide layer placed in state of compression.
190 A. Bahri et al. / Materials Science & Engineering A 670 (2016) 188–195

Width(mm) 4. Results and discussion

0 5 10 15 20 25 30 35
0 4.1. Numerical model applied to the aluminum plate

-0,04 Theoretical deflection The optimized numerical model was firstly tested on an alu-
Deflection (mm)

minum plate in both three and four-point bending configurations.

Numerical deflection The experimental load–deflection curves for the aluminum
-0,08
plate were compared with the simulation results for both three
-0,12 and four-point bending experiments as shown in Fig. 5. It was
observed that the applied load increases monotonically with dis-
-0,16 placement in a linear elastic manner. The simulated curves follow
the experimental ones. It is worthy to note that the little dis-
-0,2
crepancy can be ascribed to the difference in terms of the di-
mensions between the real and modeled aluminum plates. Such
Width (mm) phenomenon can affect directly the flexural response of the ma-
terial. In fact, the aluminum plate was obtained after cutting then
0 5 10 15 20 25 30 35
polishing, hence the thickness may be less accurate than the
0
theoretical value, which is about 2.8 mm. A good agreement was
-0,05 obtained between the experimental load–deflection curves and
Theoretical deflection those of the 2D and 3D numerical models. This result is in con-
Deflection (mm)

-0,1
Numerical deflection
cordance with the findings of Mohana et al. [9] about the failure of
-0,15 sandwich beams consisting of alumina face sheet and alumina face
core in bending. Plane strain 2D model was validated by com-
-0,2
paring the numerical results to the experimental and analytical
-0,25 ones.
-0,3
4.2. Determination of the mechanical properties of the alumina
-0,35 coating with inverse method: four-point bending test

1000
Based on some similar published works [30,31] focusing on the
determination of the mechanical properties of aluminum oxide
800 layers, we have retained:
Load (N)

600  The density (ρ) of the aluminum oxide varying between

Element size (1x1x0.005 mm) 3.4 g cm  3 for the amorphous alumina and 3.96 g cm  3 for the
400 Element size (2x2x0.01 mm hexagonal crystal lattice alumina.
Element size (5x5x0.04 mm)  The Poisson ratio (υ) being equal to 0.25.
200 Optimum size (4x4x0.02 mm)  The Young's modulus, estimated by the indentation technique
Experimental varying between 35 GPa and 185 GPa as the indenter dwell
0 time.
0 0,5 1 1,5 2 2,5 3 3,5
Deflection (mm) In the following section, we propose an inverse method to es-
Fig. 2. Theoretical and numerical deflections: (a) three-point bending, (b) four-
timate the Young's modulus of the alumina coating using a nu-
point bending, and (c) four-point bending; numerical and experimental load- merical model. A comparison with the experimental result was
deflections. systematically achieved to find the best value of the elastic mod-
ulus that allows a good fit to the experimental curve. Such method
was repeated for different thickness values of the anodic oxide
strain analysis was justified in the sandwich beams for the 2D layer (Fig. 6). As can be observed in Fig. 6, all the numerical load–
model. In order to optimize the mesh size of the aluminum sheet, deflection curves are not in agreement with the experimental re-
we have varied the mesh element size and therefore, a compara- sults previously reported in the literature. This is true for the dif-
tive study between the theoretical and numerical load-deflections ferent thickness values of the oxide layer 20 mm, 50 mm and 80 mm.
was conducted as shown in Fig. 2. This proves that elastic modulus value is much lower than 35 GPa,
Four nodded bi-linear plane-strain quadrilateral elements with which is in perfect agreement with the work of Datcheva. et al.
reduced integration (CPE4R) were used to discretize the sandwich [30], as well as the findings of Dukhyun et al. [32] about anodic
geometry for the 2D model [29]; as for the 3D model hexahedral aluminum oxide films. These authors have investigated the elastic
(C3D8R) elements were adopted as shown in Fig. 3. modulus of the oxide layer using bending test, and therefore they
For the sake of an improved accuracy and according to the have found that the mean modulus varies between 35 and 40 GPa
results obtained in Fig. 3, a fine mesh was incorporated in the when using tensile testing.
regions of high stress or strain gradient. The load was applied in The numerical curves were found to be in good agreement with
terms of displacement d ¼8 mm through the two upper loading the experimental results for the thicknesses values of 20 mm and
indenters while two lower indenters were completely constrained 50 mm when the elastic modulus was equal to 15 GPa. However,
(Fig. 4). It should be noted that the same conditions have already for a thickness value of 80 mm, the elastic modulus was about
been used for the 2D model (Fig. 3). The loading and supporting 20 GPa. A difference was observed when comparing the elastic
rollers were modeled as rigid bodies. The aluminum plate was modulus values from the literature to those predicted in our re-
considered to have an elastic-plastic behavior, which was char- search study. Such difference can be related to the fact that we
acterized experimentally using tensile testing. tried to estimate the global elastic modulus for the oxide layer. On
 A. Bahri et al. / Materials Science & Engineering A 670 (2016) 188–195 191

Fig. 3. Typical FE mesh showing the geometry: (a) three-point bending, (b) four-point bending and (c) Area of mesh refinement.

XSYMM
4.3. Evaluation of the sandwich model with different thickness
U1==UR1=UR3=0 values

For all the retained oxide layer thickness values (20 mm, 50 mm
and 80 mm), the sandwich model gives similar results for both 2D
Displacement and 3D models. Until the deflection value of 5 mm, a good corre-
lation between the numerical and the experimental results was
found. Besides, a sudden decrease of the maximum load magni-
Encastrement tude in the experimental curve was observed indicating a me-
U1=U2=U3=UR1=UR2=UR3=0
chanical failure of the oxide layer, as shown in Fig. 9a. As a result of
the inter-facial shear stress, the applied force seems to be trans-
ferred to the oxide layer inducing the tensile and compression
stresses. As the deformation increases, the stress increases and
then exceeds the fracture strength of the alumina coating. Fur-
thermore, parallel long cracks show up on the aluminum oxide
Fig. 4. Loading and boundary conditions used in the FE simulations.
layers perpendicularly to the normal stress as shown in Fig. 9b and
c. Once the first array takes place, multiple channel cracks are
the other hand, it can be attributed to the difference in the
structure of the oxide layer such as the size and density of the formed throughout the aluminum oxide layer, as observed in al-
pores. It should be noticed that the numerical curves were ob- most all the anodized plates. Under uni-axial tensile, the formation
tained by 3D sandwich model. Fig. 7 shows the variation of the of micro-cracks occurred in two cases: i) when the accumulated
elastic modulus for the oxide layer as a function of the film shear stress is transferred by the interface from the aluminum
thickness extracted from the previous result. (ductile) to the oxide layer (brittle) and ii) when the tensile stress
It can be seen in Fig. 7, the value of the elastic modulus ob- in the coating reaches the fracture strength level. On the tensile
viously depends on the alumina coating thickness. This result can side, the micro-cracks take place on the top and back of the anodic
be probably attributed to the presence and distribution of porosity. oxide layer. It addition, some macroscopic fractures were observed
This accounts for the high-magnification SEM images taken as in the oxide layer. These results agree with the findings related to
given in Fig. 8. Therefore, the microstructure examination reveals a the anodized Al-5% Mg substrate [32–34], as well as the anodized
large number of pores as shown in Fig. 8a and b. Fig. 8c shows AA1050 substrate [23].
parallel fractures which occur after conducting four-point bending
test. The fracture initiation could be the results of pores observed 4.4. Evaluation of anodized layer for both tensile and compression
in the alumina coating. models
In the following section, elastic modulus values of 15 GPa for
the coating thickness of (20 mm, 50 mm) and 20 GPa for a coating In this section, one aluminum oxide layer was applied, a
thickness of 80 mm will be retained. thickness of 30 mm was used on the upper side of the compression
192 A. Bahri et al. / Materials Science & Engineering A 670 (2016) 188–195

350 Simulation E=180 GPa

4000
Simulation E=140 GPa
300 3500 Simulation E=100 GPa
Simulation E=60 GPa
3000
Simulation E=20 GPa
250

Load (N)
2500 Simulation E=15 GPa
Load (N)

2000 Experimental
200
Experimental 1500
150 2D simulation model 1000

3D simulation model 500

100
0
0 2 4 6 8
50
Deflection (mm)
7000
0 Simulation E=180 GPa
0 2 4 6 8 10
6000 Simulation E=140 GPa
Deflection (mm) Simulation E=100 GPa
5000
Simulation E=60 GPa
450

Load (N)
Simulation E=20 GPa
4000
400 Simulation E=15 GPa
3000 Experimental
350
2000
300
Load (N)

1000
250
0
200 0 2 4 6 8
Experimental
150 Deflection (mm)
2D simulation model
100 10000 Simulation E=180 GPa
3D simulation model
9000 Simulation E=140 GPa
50 Simulation E=100 GPa
8000
Simulation E=60 GPa
0 7000
0 2 4 6 8 10 Simulation E=20 GPa
Load (N)

6000 Simulation E=15 GPa

Deflection (mm)
5000 Experimental
Fig. 5. Load–deflection curves for aluminum plates, (a) three-point bending, and 4000
(b) four-point bending configurations.
3000
2000
model, and on the down side for the tensile one. For both con- 1000
figurations, a value of 15 GPa was used as an elastic modulus in the 0
0 2 4 6 8
numerical simulation. A good correlation was found between the
numerical and experimental results concerning load–deflection Deflection (mm)
curves in the four-point bending test. As far as the tensile-sub- Fig. 6. Variation of the elastic modulus for different thickness values of oxide layers
jected oxide layer model is concerned, all the obtained curves for a sandwich model: (a) e¼ 20 mm, (b) e¼ 50 mm, and (c) e ¼80 mm.
display a good fit until a five-millimeter displacement. At this le-
vel, cracks of the oxide layer begin as shown in Fig. 10. The same 25
phenomenon was noted for the compression-subjected oxide layer
Young's modulus (GPa)

model [3]. In fact, Styles et al., [3] have shown that the uni-axial
20
straining for a brittle film in a ductile substrate induces the for-
mation of an array of parallel cracks whose density increases with
the increase of the strain. This proves the robustness of the nu- 15
merical model, including the choice of the elastic modulus values
as previously estimated for the different coatings.
10

4.5. Contributions of the alumina coating

5
Fig. 11 shows the numerical curves obtained for the sandwich
model, which takes into account the oxide layer placed in tensile
0
and compression as well as the aluminum plate itself. The analysis 0 20 40 60 80 100
of these curves reveals that the sandwich beam presents a better
mechanical behavior in four-point bending tests. In fact, it is
Oxide layer thickness (μm)
known that the four-point bending method allows for uniform Fig. 7. Variation of the elastic modulus as function of the oxide layer thickness.
distribution of stress between the two loading rollers and not lo-
cated under the roller in the case of three-point bending method.
Consequently, the four-point bending configuration can tolerate
the eventual damage and reduce any disastrous failure of the
 A. Bahri et al. / Materials Science & Engineering A 670 (2016) 188–195 193

Fig. 8. SEM photos: (a) and (b) alumina coating before bending test and (c) alumina coating after four-point bending test of the face in tensile.

tested sample as reported in Ref. [23]. In addition, the oxide layer for both compression, tensile and the sandwich model.
placed in the state of compression has the capacity to endure the
load when it is subjected to four-point bending rather than the
layer placed in tensile. This phenomenon can be attributed to two 5. Conclusion
reasons:
In this work, FE simulations of three- and four-point bending
i) the alumina coating is porous, and when these pores are tests have been adequately conducted and then compared to ex-
subjected to compression they tend to close down. This fact perimental results. The main objective was to predict the elastic
can enhance the coating resistance to compression stresses. modulus of the alumina coating deposited on an aluminum sub-
However, when the pores are subjected to the tensile, they strate. Three configurations were retained: a sandwich model, a
tend to open up. Consequently, the coating resistance to ten- model with coating subjected to tensile and model with coating
sile stresses may be affected. placed in the state of compression. The following conclusions can
ii) The damage occurs under the loading punch which more da- be drawn from the modeling-experimental comparison:
maging than the crack initiation in the layer under tensile.
Meanwhile, damage propagation seems to be gentle and  Both numerical 2D and 3D models have shown a good fit to the
progressive in the layer submitted to the compression as experimental results, thus the elastic modulus of the alumina
published in similar work [23]. This result puts away the coating was successfully predicted. For the thickness values
assumption about the influence of the local contact loading varying from 20–50 mm, the elastic modulus was found to be
under the punch. Despite the damage of the alumina coating around 15 GPa, whereas it was about 20 GPa for an alumina
by indentation when it is subjected to compression, its rigidity coating with thickness of 80 mm.
or stiffness increases. This agrees well with the findings of the  A sudden decrease of the load was observed on the load–de-
crack interaction and crack density in the strain induced flection curves indicating the oxide layer failure. In addition, the
cracking of a brittle film [35]. In fact, in the previous Section micro-graphs of the transversal face of the contact area rollers/
4.4, it was reported that the stress-strain curve of a compres- oxide layer have shown a crack initiated under the punch and
sion test reached higher stress when compared to the tensile propagated through the thickness.
test. The contribution of the oxide layer was clearly observed,  Two damage modes have been identified: indentation damage
from the load–deflection curves, to resist the bending forces caused by the contact punch/oxide layers and another one
194 A. Bahri et al. / Materials Science & Engineering A 670 (2016) 188–195

(a) 2000 1800

Sandwich model
1800 1600
Experimental Model: layer in compression
1600 1400
2D simulation model Model: layer in tensile
1400 Aluminium plate
1200
3D simulation model

Load (N)
Load (N)

1200
1000
1000
800
800
600
600
400
400
200
200
0
0 0 1 2 3 4 5 6 7 8 9
0 1 2 3 4 5 6 7 8 9
Deflection (mm)
Deflection (mm)
Fig. 11. Numerical load–deflection curves of four-point bending test (Oxide layer of
thickness 30 mm).

between the punches caused by reaching the limit of fracture of

the alumina coating.
 The numerical/experimental correlation has highlighted the
robustness of the sandwich model. It allowed to conclude that
when the alumina coating is subjected to compression it is
more resistant than in tensile. The deflection behavior of the
Fig. 9. (a) Load–deflection curves for four bending test of sandwich model (Oxide
material was found to depend essentially on the response of the
layer of thickness 50 mm, (b) Lateral side of the contact area of the face in com- alumina coating which was subjected to compression stresses.
pression, and (c) Longitudinal side of the face in tensile.
Finally, taking into account the initiation of cracks under flex-
1200 ural tests, it will be very interesting, in a future work, to introduce
a damage criteria in the used numerical code.
1000

Acknowledgement
800
Load (N)

This work was supported by the Ministry of higher Education

600 and Scientific Research, Tunisia.

400 Experimental
References
2D simulation model
200
3D simulation model [1] S. Mezlini, K. Elleuch, Ph. Kapsa, The effect of sulphuric anodisation of alu-
minium alloys on contact problems, Surf. Coat. Technol. 200 (2006)
2852–2856.
0
0 2 4 6 8 10 [2] J. Baumeister, J. Banhart, M. Weber, Aluminium foams for transport industry,
Mater. Des. 18 (1997) 217–220.
Deflection (mm) [3] M. Styles, P. Compston, S. Kalyanasundaram, The effect of core thickness on the
flexural behaviour of aluminium foam sandwich structures, Compos. Struct. 80
1400 (2007) 532–538.
[4] J.H. Jeon, P.S. Lee, K.H. Lee, H.C. Park, W. Hwang, Bending fatigue characteristics
of nanohoneycomb structures, Compos. Struct. 82 (2008) 28–35.
1200
[5] H.C. Yoon, F.D. Owen, Mechanical failure of anodic films on Aluminum and
Tantalum, J. Electrochem. Soc. 123 (1976) 1868–1875.
1000 [6] F. Li, L. Zhang, M. Metzger Robert, On the growth of highly ordered pores in
anodized Aluminum Oxide, Chem. Mater. 9 (1998) 2470–2480.
Load (N)

[7] W. Bensalah, K. Elleuch, M. Feki, M. Wery, M.P. Gigandet, F. Ayedi, Optimiza-

800 tion of mechanical and chemical properties of sulfuric anodized aluminum
using statistical experimental methods, Mater. Chem. Phys. 108 (2008)
600 296–305.
[8] T. Aerts, Dimogerontakis Th, I. De Graeve, J. Fransaer, H. Terryn, Influence of
Experimental
the anodizing temperature on the porosity and the mechanical properties of
400 the porous anodic oxide film, Surf. Coat. Technol. 201 (2007) 7310–7317.
2D simulation model
[9] K. Mohana, Y. Tick Hona, S. Idapalapati, H. Pheow Seowa, Failure of sandwich
200 3D simulation model beams consisting of alumina face sheet and aluminum foam core in bending,
Mater. Sci. Eng. A 409 (2005) 292–301.
[10] C. Edeleanu, T.J. Law, M. Phil, Studies on the structure of anodic oxide films on
0 Aluminum, J. Electrochem. Soc. 110 (1963) 262–267.
0 2 4 6 8 10 [11] T. Namazu, Y. Isono, Plastic deformation of nanometric single crystal silicon
Deflection (mm) wire in AFM bending test at intermediate temperatures, Microelectromech.
Syst. 11 (2002) 125–135.
Fig. 10. Load–deflection curves: (a) model one side tensile position, and (b) model [12] T. Namazu, Y. Isono, T. Tanaka, Evaluation of size effect on mechanical prop-
one side. erties of single crystal Silicon by nanoscale bending test using AFM, Micro-
compression position. electromech. Syst. 9 (2000) 450–459.
 A. Bahri et al. / Materials Science & Engineering A 670 (2016) 188–195 195

[13] S. Sundararajan, B. Bhushan, Development of AFM-based techniques to mea- surfaces, J. Am. Ceram. Soc. 81 (1998) 2301–2308.
sure mechanical properties of nanoscale structures, Sens. Actuators A: Phys. [25] W. Bensalah, K. Elleuch, M. Feki, M. DepPetris-Wery, H.F. Ayedi, Mechanical
101 (2002) 338–351. failure of anodized films of aluminum in bending, Mater. Des. 30 (2009)
[14] S. Sundararajan, B. Bhushan, T. Namazu, Y. Isono, Mechanical property mea- 3141–3149.
surements of nanoscale structures using an atomic force microscope, Ultra- [26] D. Choi, J. Jeon, P. Lee, W. Hwang, K. Lee, H. Park, Young's modulus measure-
microscopy 91 (2002) 111–118. ments of nanohoneycomb structures by flexure testing in atomic force mi-
[15] K.R. Virwani, A.P. Malshe, W.F. Schmidt, D.K. Sood, Young's modulus mea- croscopy, Compos. Struct. 79 (2007) 548–553.
surements of silicon nanostructures using a scanning probe system: a non- [27] W. Bensalah, K. Elleuch, M. Wery, H.F. Ayedi, Thick and dense anodic oxide
destructive evaluation approach, Smart Mater. Struct. 12 (2003) 1028–1037. layers formed in aluminum in sulfuric acid bath, J. Mater. Sci. Technol. 26
[16] T. Namazu, Y. Isono, Quasi-static bending test of nano-scale SiO2 wire at in- (2010) 113–118.
termediate temperatures using AFM-based technique, Sens. Actuators A: Phys. [28] W. Bensalah, K. Elleuch, M. DepPetris-Wery, H.F. Ayedi, comparative study of
104 (2003) 78–85. mechanical and tribological properties of alumina coatings formed on Alu-
[17] D. Müller, E. Fromm, Mechanical properties and adhesion strength of TiN and minum in various conditions, Mater. Des. 30 (2009) 3731–3737.
Al coatings on HSS, steel, aluminium and copper characterized by four testing [29] Jun Fujiwara, Taku Nakajima, Makoto Ohsaki, and Fumiyoshi Takeda, Shape
Methods, Thin Solid Films 270 (1995) 411–416. Optimization of clamping members of frame-supported membrane structures
[18] Daniel A. Maidenberg, Willi Volksen, Robert D. Miller, Reinhold H. Dauskardt, under stress constraints, in: Proceedings of the 6th China–Japan–Korea Joint
Toughening of nanoporous glasses using porogen residuals, Nat. Mater. 3 Symposium on Optimization of Structural and Mechanical Systems, 2010, pp.
(2004) 464–469. 22–25.
[19] J.C. Grosskerutz, M.B. McNeil, Mechanical properties of metal oxide films, [30] M. Datcheva, S. Cherneva1, M. Stoycheva, R. Iankov, D. Stoychev, Determina-
Electrochem. Soc. 116 (1969) 1232–1237. tion of Anodized aluminum material characteristics by means of na-
[20] Y. Yamazaki, A. Schmidt, A. Scholz, The determination of the delamination noindentation measurements, Mater. Sci. Appl. 2 (2011) 1452–1464.
resistance in thermal barrier coating system by four-point bending tests, Surf. [31] X. Changjin, T. Wei, Cracking and decohesion of a thin Al2O3 film on a ductile
Coat. Technol. 201 (2006) 744–754. Al–5% Mg substrate, Acta Mater. 53 (2005) 477–485.
[21] W.K. Chan, M. Luo, T.Y. Zhang, Molecular dynamics simulations of four-point [32] C. Dukhyun, J. Jihoon, L. Pyungsoo, H. Woonbong, L. Kunhong, Hyunchul Park.
bending tests on SiC nanowires, Scr. Mater. 59 (2008) 692–695. Young's Modulus Measurements of nanohoneycomb Structures by Flexural
[22] Z. Gan, S.G. Mhaisalkar, Z. Chen, S. Zhang, Z. Chen, K. Prasad, Study of inter- testing in Atomic.
facial adhesion energy of multilayered ULSI thin film structures using four- [33] A. Needleman, An analysis of decohesion along an imperfect interface, Int. J.
point bending test, Surf. Coat. Technol. 198 (2005) 85–89. Fract. 42 (1990) 21–40.
[23] M. Bargui, W. Bensalah, K. Elleuch, H.F. Ayedi, Mechanical failure of anodized [34] Y. Goueffon, L. Arurault, C. Mabruc, C. Tonond, P. Guiguea, Black anodic coat-
aluminum under three and four-point bending tests, Mater. Des. 51 (2013) ings for space applications: study of the process parameters, characteristics
67–72. and mechanical properties, J. Mater. Process. Technol. 209 (2009) 5145–5151.
[24] J. Juthamas, Nitin P. Padture, Antonios E. Giannakopoulos, Subra Suresh, [35] C. Chen, A.M. Harte, N.A. Fleck, The plastic collapse of sandwich beams with a
Hertzian-Crack suppression in ceramics with elastic-modulus-graded metallic foam core, Int. J. Mech. Sci. 43 (2001) 1483–1506.